The fourth step in the development of toxicity is inappropriate repair and adaptation (Fig. 3-1). As noted previously, many toxicants alter macromolecules, which eventually cause damage at higher levels of the biological hierarchy in the organism. Progression of toxic lesions can be intercepted by repair mechanisms operating at molecular, cellular, and tissue levels (Fig. 3-21). Another strategy whereby the organism can resist the noxious chemical is by increasing its own readiness to cope with it and with its harmful effects. This phenomenon is called adaptation. Because the capacity of the organism to repair itself and adapt to the toxic exposure and effects is so important in determining the outcome of chemical exposure, the mechanisms of repair and adaptation will be discussed below.
Repair mechanisms. Dysfunction of these mechanisms results in dysrepair, the fourth step in the development of numerous toxic injuries. ECM, extracellular matrix.
Damaged molecules may be repaired in different ways. Some chemical alterations, such as oxidation of protein thiols and methylation of DNA, are simply reversed. Hydrolytic removal of the molecule’s damaged unit or units and insertion of a newly synthesized unit or units often occur with chemically altered DNA and peroxidized lipids. In some instances, the damaged molecule is totally degraded and resynthesized. This process is time-consuming but unavoidable in cases such as the regeneration of cholinesterase after organophosphate intoxication.
Thiol groups are essential for the function of numerous proteins, such as receptors, enzymes, cytoskeletal proteins, and TFs. Oxidation of protein thiols (Prot-SHs) to protein disulfides (Prot-SS, Prot1-SS-Prot2), protein–glutathione mixed disulfides (Prot-SSG), and protein sulfenic acids (Prot-SOH) can be reversed by reduction (Watson et al., 2004; Gravina and Mieyal, 1993) (Fig. 3-22). The endogenous reductants are thioredoxins and glutaredoxins, small, ubiquitous proteins with two redox-active cysteines in their active centers (Holmgren et al., 2005). These proteins as well as thioredoxin reductase have two isoenzymes; those labeled 1 are located in the cytosol, whereas those labeled 2 are mitochondrial. Because the catalytic thiol groups in these proteins become oxidized, they are reduced by NADPH, which is generated by NADP+-dependent isocitrate dehydrogenase localized in various cell compartments (cytosol, mitochondria, peroxisomes), as well as by the cytosolic glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase in the pentose phosphate pathway. ROS may oxidize methionines in proteins into sulfoxides (protein-Met-S=O), forming both the S and R epimers, which can be reduced by methionine sulfoxide reductase (Msr) A and B enzymes, respectively (Moskovitz, 2005). Msr enzymes can reverse this modification at the expense of oxidation of their catalytic cysteine to sulfenic acid (Msr-Cys-S-OH). This then reacts with a neighboring thiol (in MsrA), forming an intramolecular disulfide, or with glutathione (in some other Msr proteins), forming a protein–glutathione disulfide. Finally the disulfide enzyme is reduced by thioredoxin or glutaredoxin, respectively, with subsequent steps depicted in Fig. 3-22. Reduction of methionine sulfoxides in lens proteins (eg, α-crystallin) is especially critical for maintenance of the transparency of the eye lens. MsrA knockout mice develop cataract on repeated exposure to hyperbaric oxygen (Kantorow et al., 2010). Repair of oxidized hemoglobin (methemoglobin) occurs by means of electron transfer from cytochrome b5, which is then regenerated by a NADH-dependent cytochrome b5 reductase (also called methemoglobin reductase).
Repair of proteins oxidized at their thiol groups. Protein disulfides (Prot-SS, Prot1-SS-Prot2) and protein sulfenic acids (Prot-SOH) are reduced by thioredoxin (Trx-[SH]2) or thioredoxin-like proteins. Protein–glutathione mixed disulfides (Prot-SSG) are reduced by glutaredoxin (Grx-[SH]2), which is also called thioltransferase. The figure also indicates how Trx-[SH]2 and Grx-[SH]2 are regenerated from their disulfides (Trx-SS and Grx-SS, respectively). In the mitochondria, Trx-SS also can be reduced by the dithiol dihydrolipoic acid, a component of the pyruvate- and α-ketoglutarate dehydrogenase complexes. GSH, glutathione; GSSG, glutathione disulfide; GR-[SH]2 and GR-SS, glutathione reductase (dithiol and disulfide forms, respectively); TrxR-[SH]2 and Trx-SS, thioredoxin reductase (dithiol and disulfide forms, respectively).
Soluble intracellular proteins such as cytosolic enzymes are typically folded into a globular form with their hydrophobic amino acid residues hidden inside, whereas the hydrophilic residues are located externally together with a hydrophobic cleft that constitutes the ligand (substrate) binding site. Physical or chemical insults may evoke an unduly large opening of this cleft that may lead to unfolding of the protein (denaturation) and its aggregation. Molecular chaperones such as the heat-shock proteins (Hsp, eg, Hsp90, Hsp70, and Hsp40) can prevent protein unfolding by “clamping down” onto the exposed hydrophobic region of their client protein, using the energy of ATP hydrolysis to execute their conformational change that entails this maneuver. According to a recent model (Pratt et al., 2010), Hsp90 plays the leading role in stabilizing the cleft and impeding further unfolding, which would eventually lead to degradation of the protein by the ubiquitin–proteasome system (UPS) as described below. Indeed, the benzoquinone ansamycin Hsp90 inhibitors (eg, geldanamycin and herbimycin) promote degradation of Hsp90 client proteins by the UPS.
When unfolding progresses beyond a limit, Hsp90 dissociates. This may occur in response to chemical-induced protein damage, such as that inflicted by mechanism-based enzyme inactivators that are converted into reactive metabolites that bind covalently to the catalytic site of the enzyme. Dissociation of Hsp90 from the unfolding protein allows Hsp70 and its co-chaperone Hsp40 to recruit the Hsp70-dependent E3 ubiquitin ligases (eg, CHIP and Parkin) that in turn direct ubiquitin-charged E2 enzymes to the Hsp70-bound client protein to tag it with a polyubiquitin chain as the final step in a process presented in Fig. 3-23 (Bedford et al., 2011). The protein tagged with a Lys48-linked polyubiquitin chain is then recognized and degraded in the proteasomes. The 26S proteasome is large barrel-shaped multiprotein complex that binds the ubiquitinated protein, deubiquitinates and unfolds it using ATP, and finally hydrolyzes the protein with its threonine-containing protease active sites into small peptides. The UPS controls the cellular level of numerous regulatory proteins (eg, p53, IκB, Nrf2, β-catenin, cyclins) and also has a prominent role in eliminating oxidized or otherwise damaged and unfolded intracellular proteins (Poppek and Grune, 2006). NOS exposed to its mechanism-based inactivator guanabenz that causes loss of tetrahydrobiopterin from the substrate-binding cleft will suffer this fate (Pratt et al., 2010). CYP2E1, in which the heme becomes cross-linked to the protein as a consequence of reductive dechlorination of CCl4 to the trichloromethyl free radical by this enzyme, is also eliminated by proteasomal degradation.
The process of ubiquitination and its possible outcomes. Ubiquitin (Ub) is a small (8.6 kDa) ubiquitous protein that when covalently linked to a target protein (as a monomer or as a polymer of Ub units) alters its fate and/or function. Ubiquitination is a posttranslational protein modification that takes place in an enzymatic cascade composed of 3 types of proteins—the E1 Ub-activating enzyme, the E2 Ub-conjugating enzyme, and the E3 Ub ligase. Humans possess 8 E1, over 30 E2, and hundreds of E3 enzymes. E1 first activates Ub in an ATP-consuming reaction by adenylating the terminal carboxyl group of Ub, thus forming Ub-AMP, and then it transfers Ub from Ub-AMP to its catalytic SH group, forming a high-energy thioester bond between E1 and Ub. In the subsequent transthiolation reaction, Ub is transferred from E1 to the catalytic SH group of E2. Thereafter, E3 binds both the target protein and the Ub-charged E2 and catalyzes the transfer of Ub from E2 to the target protein, forming an isopeptide bond between the terminal carboxyl group of Ub and the ε-amino group of a lysine in the substrate. Most E3 ligases, including the so-called RING finger domain-containing E3 ligases (which constitute 1 of the largest enzyme groups in the cell), ubiquitinate proteins this way. However, the HECT domain-containing E3 ligases, which possess an active site cysteine, take Ub from E2 in a transthiolation reaction, and then they (rather than E2) transfer Ub to the target protein. As Ub contains 7 lysine residues, it can also be ubiquitinated. Multiple rounds of ubiquitination generate polyubiquitin chains. In general, monoubiquitination triggers endocytosis or autophagy of the protein. Substrates that are tagged with Lys48-linked Ub chains are recognized and degraded by the 26S proteasome into small peptides, with the Ubs released for reuse. Proteins polyubiquitinated with Lys63-linked Ub chains are generally not degraded but are essential components of signaling pathways, functioning as scaffolds to assemble signaling complexes. For example, the activation of transcription factor NF-κB on signaling (see Fig. 3-12) involves assembly of a signaling complex by means of Lys63-linked Ub chains. Ubiquitination can be reversed by deubiquitinating enzymes, which are Ub-specific proteases (cysteine proteases or zinc metalloproteases). These may rescue a substrate from degradation by removing a degradative Ub signal or may change or remove a nondegradative Ub signal. Instead of Ub, proteins may also be modified by covalent attachment of Ub-like proteins, such as NEDD8, SUMO, and ISG15, using the enzymatic machinery described above.
Oligomerization and aggregation of damaged and unfolded proteins preclude the proteasome from degrading them; such substrates can even trap proteasomes, rendering them nonfunctional. After ubiquitination, protein aggregates can be eliminated by autophagy, a process described in more detail in the section “Cellular Repair.” Ubiquitin on the modified proteins is recognized and bound by autophagic adaptor proteins, such as p62 or Nbr1, which interact with LC3 to deliver them to autophagosomes, newly formed vesicles that encapsulate their protein cargo. After fusing of autophagosomes with lysosomes, the damaged protein is hydrolyzed by proteases. Lysosomal proteases degrade, for example, the immunogenic trifluoroacetylated proteins that are formed in the liver during halothane anesthesia (Cohen et al., 1997). Erythrocytes have ATP-independent, nonlysosomal proteolytic enzymes that rapidly and selectively degrade proteins denatured by HO• (Davies, 1987). Red blood cells containing protein aggregates (Heinz body) after exposure to methemoglobin-forming chemicals are removed via phagocytosis by macrophages in the spleen.
Phospholipids containing fatty acid hydroperoxides are preferentially hydrolyzed by phospholipase A2, with the peroxidized fatty acids replaced by normal fatty acids (van Kuijk et al., 1987). Peroxidized lipids (eg, fatty acid hydroperoxides and phospholipid-associated hydroperoxides) may be reduced by the glutathione peroxidase–glutathione–glutathione reductase system (see Fig. 3-24) or by the peroxiredoxin–thioredoxin–thioredoxin reductase system (shown in part in Fig. 3-22). Again, NADPH is needed to “repair” the reductants that are oxidized in the process.
Repair of peroxidized lipids. Phospholipid peroxyl radicals (PL-OO•) formed as a result of lipid peroxidation (Fig. 3-9) may abstract hydrogen from α-tocopherol (TOC-OH) and yield phospholipid hydroperoxide (PL-OOH). From the latter, the fatty acid carrying the hydroperoxide group is eliminated via hydrolysis catalyzed by phospholipase (PLase), yielding a fatty acid hydroperoxide (FA-OOH) and a lysophospholipid (LPL). The former is reduced to a hydroxy fatty acid (FA-OH) by glutathione peroxidase (GPX), utilizing glutathione (GSH), or peroxiredoxins (not shown), whereas the latter is reacylated to phospholipid (PL) by lysophosphatide fatty acyl-coenzyme A transferase (LFTF), utilizing long-chain fatty acid–coenzyme A (FA-CoA). The figure also indicates regeneration of TOC-OH by ascorbic acid (HO-ASC-OH), regeneration of ascorbic acid from dehydroascorbic acid (O=ASC=O) by glutaredoxin (Grx-[SH]2), and reduction of the oxidized glutaredoxin (Grx-SS) by GSH. Oxidized glutathione (GSSG) is reduced by glutathione reductase (GR-[SH]2), which is regenerated from its oxidized form (GR-SS) by NADPH, the ultimate reductant. NADPH is produced by NADP+-dependent isocitrate dehydrogenases and during metabolism of glucose via the pentose phosphate shunt. TOC-O•, tocopheroxyl radical; •O-ASC-OH, ascorbyl radical.
Despite its high reactivity with electrophiles and free radicals, nuclear DNA is remarkably stable, in part because it is packaged in chromatin and because several repair mechanisms are available to correct alterations. The mitochondrial DNA, however, lacks histones and efficient repair mechanisms and therefore is more prone to damage. Different types of damages are corrected by specialized mechanisms, each employing a different set of repair proteins (Christmann et al., 2003).
Certain covalent DNA modifications are directly reversed by enzymes such as DNA photolyase, which cleaves adjacent pyrimidines dimerized by UV light. Inasmuch as this chromophore-equipped enzyme uses the energy of visible light to correct damage, its function is restricted to light-exposed cells.
Minor adducts, such as methyl groups, attached to DNA bases by alkylating agents (eg, MMS) may be removed by special enzymes (Christmann et al., 2003). Such groups attached to the O6 position of guanine are cleaved off by O6-methyguanine-DNA-methyltransferase (MGMT). While repairing the DNA, this alkyltransferase sacrifices itself, transferring the adduct onto one of its cysteine residues. This results in its inactivation, ubiquitination, and proteasomal degradation. Thus, like glutathione, which is depleted during detoxication of electrophiles, MGMT is consumed during the repair of DNA. Methyl groups attached to N1 of adenine and guanine, and N3 of thymine and cytosine are removed by oxidative demethylation catalyzed by DNA dioxygenases (ABH2 and ABH3). These peculiar O2, Fe2+, ascorbate, and 2-oxoglutarate-dependent enzymes oxygenate the methyl group adduct, which in turn leaves as formaldehyde, whereas 2-oxoglutarate is oxidatively decarboxylated to succinate.
Base excision and nucleotide excision are two mechanisms for removing damaged bases from DNA (see Chaps. 8 and 9). Lesions that do not cause major distortion of the helix typically are removed by base excision, in which the altered base is recognized by a relatively substrate-specific DNA glycosylase that hydrolyzes the N-glycosidic bond, releasing the modified base and creating an apurinic or apyrimidinic (AP) site in the DNA. For example, 8-OH-Gua, a major mutagenic product of oxidative stress, is removed from the DNA by a specific 8-OH-Gua DNA glycosylase. The AP site is recognized by the AP endonuclease, which hydrolyzes the phosphodiester bond adjacent to the abasic site. After its removal, the abasic sugar is replaced with the correct nucleotide by a DNA polymerase and is sealed in place by a DNA ligase. Interestingly, AP endonuclease is a bifunctional protein and is also called redox factor-1 (Ref-1). In concert with thioredoxin and thioredoxin reductase, it maintains TFs with sensitive thiol groups in their DNA-binding domain (Fos, Jun, NF-κB) in an active reduced state (Hansen et al., 2006).
Bulky lesions such as adducts produced by aflatoxins or aminofluorene derivatives and dimers caused by UV radiation are removed by nucleotide excision repair system, which consists of approximately 30 proteins (Christmann et al., 2003). Lesions in the nontranscribed strands or the nontranscribed regions of the genome are corrected by the global genomic repair system. This involves proteins that recognize the distorted double helix at the lesion, unwind the DNA, and excise a number of intact nucleotides on both sides of the lesion together with the one containing the adduct. The excised section of the strand is restored by insertion of nucleotides into the gap by DNA polymerase and ligase, using the complementary strand as a template. Lesions in the transcribed DNA strand blocking the RNA polymerase in the actively transcribed genes are removed by another variation of nucleotide excision repair, the transcription-coupled repair system. This involves assembly of repair proteins to remove the stalled RNA polymerase before excision of the damage and filling the gap. Resynthesis of the removed section of the strand is designated “unscheduled DNA synthesis” and can be detected by the appearance of altered deoxynucleosides in urine. Excision repair has a remarkably low error rate of less than one mistake in 109 bases repaired.
PARP appears to be an important contributor to excision repair. On base damage or SSB, PARP binds to the injured DNA and becomes activated. The active PARP cleaves NAD+ to use the ADP-ribose moiety of this cofactor for attaching long chains of polymeric ADP-ribose to nuclear proteins, such as histones. Because one ADP-ribose unit contains two negative charges, the poly(ADP-ribosyl)ated proteins accrue negativity and the resultant electrorepulsive force between the negatively charged proteins and DNA causes decondensation of the chromatin structure. It is hypothesized that PARP-mediated opening of the tightly packed chromatin allows the repair enzymes to access the broken DNA and fix it. Thereafter, poly(ADP-ribose) glycohydrolase gains access to the nucleus from its perinuclear localization and reverses the PARP-mediated modification of nuclear proteins (D’Amours et al., 1999). Other features of PARP that are relevant in toxicity—such as destruction of PARP by caspases during apoptosis as well as the significance of NAD+ (and consequently ATP) wasting by PARP in necrosis—have been discussed earlier in this chapter.
Nonhomologous End Joining
This process repairs DSB that may be formed when two SSB occur in close proximity, or when DNA with SSB undergoes replication. This repair system directly ligates the broken ends without the need for a homologous template. DSBs are recognized by the Ku protein (a heterodimer of Ku70 and Ku80) that binds to the DNA end. Ku then binds and activates DNA-dependent protein kinase catalytic subunit (DNA-PKcs), leading to recruitment and activation of end-processing enzymes, DNA polymerases, and DNA ligase. Although the nonhomologous end joining is error-prone, it can operate in any phase of the cell cycle and is the mechanism for DSB repair in nondividing terminally differentiated cells (eg, neurons). By contrast, the other and more faithful mechanism for DSB repair, homologous recombination, can function only after replication (in S and G2 phases), when sister chromatid sequences are available for use as templates (see below), and thus is not an option in nondividing cells.
Recombinational (or Postreplication) Repair
Homologous recombination can be used to fix postreplication gaps and DSBs (Li and Heyer, 2008). The former defect may result when excision of a bulky adduct or an intrastrand pyrimidine dimer fails to occur before DNA replication begins. At replication, such a lesion prevents DNA polymerase from polymerizing a daughter strand along a sizable stretch of the parent strand that carries the damage. Then the replication results in two homologous (“sister”) yet dissimilar DNA duplexes; one has a large postreplication gap in its daughter strand and an intact duplex synthesized at the opposite leg of the replication fork. This intact sister duplex is utilized to complete the postreplication gap in the damaged sister duplex. This is accomplished by recombination (“crossover”) of the appropriate strands of the two homologous duplexes. After separation, the sister duplex that originally contained the gap carries in its daughter strand a section originating from the parent strand of the intact sister, which in turn carries in its parent strand a section originating from the daughter strand of the damaged sister. This strand recombination explains the phenomenon of “sister chromatid exchange,” which is indicative of DNA damage corrected by recombinational repair. After completing the postreplication gap by recombination, the damage still present in the parent strand may be removed by nucleotide excision.
In addition to eliminating postreplication gaps, homologous recombination can also repair DSB in a complex process employing numerous proteins. These include, for example, the MRN complex (the trimer of Mre11, Rad50, and Nbs1 proteins, which binds to the DSB and is involved in end processing by generation of single-stranded DNA), Rad51, Rad54, and the breast cancer susceptibility proteins BRCA1 and BRCA2 (which assist the single-stranded DNA in invading the undamaged template in an ATP-dependent process), as well as polymerases, nucleases, helicases, and other components, which mediate DNA ligation and substrate resolution. A combination of excision and recombinational repairs occurs in restoration of DNA with interstrand cross-links caused by bifunctional electrophiles, such as nitrogen mustard-type drugs and cisplatin.
Autophagic removal of damaged cell organelles may be viewed as a universal mechanism of cellular repair, whereas clearance and regeneration of damaged axons is a mechanism specific for nerve cells.
Autophagy of Damaged Cell Organelles
Cells suffering mild injury may repair themselves by removing and degrading damaged components, such as organelles and protein aggregates, in a process called autophagy, meaning eating of self (Rabinowitz and White, 2010). While autophagy serves the purpose of cell restitution in all cells, it is particularly important in terminally differentiated cells, such as neurons and myocytes, where cell renewal by cell replication is not possible. In addition, autophagy of normal cell constituents is an ultimate means of nutrient generation (see the section “Adaptation to Energy Depletion—The Energy Stress Response”).
In autophagy, a so-called isolation membrane (or phagophore) emerges possibly from the EPR, which engulfs the cytoplasmic material and then encapsulates it in a double-membrane vesicle (called autophagosome). This vesicle moves along microtubules, driven by dynein motors, to the lysosome and on fusing with the lysosome generates an autolysosome. The cargo as well as the inner membrane of the original autophagosome is then degraded by lysosomal hydrolytic enzymes, such as proteases, lipases, nucleases, and glycosidases (which work optimally at the acidic pH in the lysosomes) to amino acids, lipids, nucleosides, and carbohydrates. These breakdown products are then released by lysosomal permeases and transporters into the cytosol, where they may be further metabolized or used for synthesis of new macromolecules.
Autophagic removal of damaged cell organelles is a method of their quality control, disposal and recycling. This is well exemplified by deletion of depolarized mitochondria through selective autophagy or mitophagy in cells exposed to mitochondrial poisons, such as the protonophoric uncoupler CCCP, valinomycin, and paraquat (see Table 3-6). Several proteins mediate this process (Mehrpour et al., 2010; Youle and Narendra, 2011), including (1) Pink1 (a protein kinase) localized in the MOM that accumulates there on dissipation of the mitochondrial membrane potential, (2) the cytosolic Parkin (an E3 ubiquitin ligase) that is recruited to the mitochondria by Pink1 and that then polyubiquitinates itself and other proteins, among them (3) the MOM protein mitofusins (Mfn1, Mfn2) and voltage-dependent anion channel (VDAC), (4) p62 (an autophagic adaptor protein) that binds to both ubiquitinated proteins and to (5) the cytosolic LC3 (microtubule-associated protein light chain 3), which, after cleavage to LC3-I, becomes conjugated with phosphatidylethanolamine by (6) a ubiquitin-like conjugation system that requires Atg7 (E1-like), Atg3 (E2-like), and the Atg12-Atg5:Atg16L complex (E3-like), (7) the phosphatidylinositol 3-kinase vesicular protein sorting 34 (Vps34), which is activated by interacting with (8) Beclin 1 (Bcl-2-interacting protein-1). The lipidated LC-3 (called LC3-II) can integrate into the membrane of autophagosome and can recruit the ubiqutin-tagged mitochondria into this structure. The latter protein (which later decomposes in the lysosome) is a marker for autophagosomes.
Removal of damaged mitochondria by autophagy is important not only for maintaining a functional mitochondrial pool but also for limiting oxidative cell damage and preventing apoptosis because injured mitochondria may be sources of ROS and apoptotic factors, such as cyt c, Smac, and AIF (see Fig. 3-20). Indeed, overexpression of Parkin suppresses cell death (Tanaka, 2010). In contrast, loss-of-function mutations in the genes encoding Pink1 and Parkin cause early onset monogenic forms of Parkinson disease probably through defective mitophagy of damaged mitochondria in the dopaminergic nigrostriatal neurons (Youle and Narendra, 2011). Autophagy also contributes to structural restitution of liver cells. For example, leftover peroxisomes after exposure to peroxisome proliferators (eg, fibrate esters) and lipid droplets in hepatic steatosis are also cleared by autophagy.
Regeneration of Damaged Axons
Peripheral neurons with axonal damage can regenerate their axons with the assistance of macrophages and Schwann cells. Macrophages remove debris by phagocytosis and produce cytokines and growth factors, which activate Schwann cells to proliferate and transdifferentiate from myelinating operation mode into a growth-supporting mode. Distal to the injury, Schwann cells play an indispensable role in promoting axonal regeneration by increasing their synthesis of cell adhesion molecules (eg, N-CAM), by elaborating ECM proteins for base membrane construction, and by producing an array of neurotrophic factors (eg, brain-derived neurotrophic factor, glial cell line–derived neurotrophic factor, and nerve growth factor) and their receptors (Boyd and Gordon, 2003). While comigrating with the regrowing axon, Schwann cells physically guide as well as chemically lure the axon to reinnervate the target cell.
In the mammalian central nervous system, axonal regrowth is prevented partly by the nonpermissive external environment. At the site of injury, the oligodendrocytes produce growth inhibitory myelin-associated proteins (eg, Nogo, NI 35, myelin-associated glycoprotein, oligodendrocyte myelin glycoprotein, ephrin, and semaphorin), which act through specific receptors, whereas the astrocytes produce chondroitin sulfate proteoglycans and fibrotic scar (whose components include laminin, fibronectin, and collagen IV) (Johnson, 1993) that hinder axonal regrowth. Signaling by TGF-β through Smad2 TF (see pathway #8 in Fig. 3-12) plays a leading role in overexpression of these extrinsic growth inhibitory factors. In addition, CNS neurons lose their intrinsic growth capacity postnatally, which occurs simultaneously with increased expression of some neuronal growth inhibitory Kröppel-like TFs, such as KLF4. For these reasons, damage to central neurons is irreversible but is compensated for in part by the large number of reserve nerve cells that can take over the functions of lost neurons. For example, in Parkinson disease, symptoms are not observed until there is at least an 80% loss of nigrostriatal neurons.
In tissues with cells capable of multiplying, damage is reversed by deletion of the injured cells and regeneration of the tissue by proliferation. The damaged cells are eliminated by apoptosis or necrosis.
Apoptosis: An Active Deletion of Damaged Cells
Apoptosis initiated by cell injury can serve as tissue repair for two reasons: first, because it may intercept the process leading to necrosis, as discussed earlier (see Fig. 3-20). Necrosis is a more harmful sequel than apoptosis for the tissue in which the injured cell resides. A cell destined for apoptosis shrinks; its nuclear and cytoplasmic materials condense, and then it breaks into membrane-bound fragments (apoptotic bodies) that are phagocytosed (Bursch et al., 1992). During necrosis, cells and intracellular organelles swell and disintegrate with membrane lysis. Whereas apoptosis is orderly, necrosis is a disorderly process that ends with cell debris in the extracellular environment. The constituents of the necrotic cells attract aggressive inflammatory cells, and the ensuing inflammation amplifies cell injury (see further on). With apoptosis, dead cells are removed without inflammation. Second, apoptosis may intercept the process leading to neoplasia by eliminating the cells with potentially mutagenic DNA damage. This function of apoptosis is discussed in more detail in the final section of this chapter.
It must be emphasized, however, that apoptosis of damaged cells may serve tissue restoration only for tissues that are made up of constantly renewing cells (eg, the bone marrow, the respiratory and GI epithelium, and the epidermis of the skin), or of conditionally dividing cells (eg, hepatic and renal parenchymal cells), because in these tissues the apoptotic cells are readily replaced. The role of apoptosis as a tissue repair strategy is markedly lessened in organs containing nonreplicating and nonreplaceable cells, such as the neurons, cardiac muscle cells, and female germ cells, because deletion of such cells, if extensive, can cause a deficit in the organ’s function. Apoptosis may also be harmful when it occurs at a critical location. For example, in the pulmonary alveolar epithelium, an extremely tight barrier, apoptosis could cause flooding of the alveolar space with interstitial fluid, a potentially lethal outcome.
Proliferation: Regeneration of Tissue
Tissues are composed of various cells and the ECM. Tissue elements are anchored to each other by transmembrane proteins. Cadherins allow adjacent cells to adhere to one another, whereas connexins associate into tubular structures and connect neighboring cells internally (gap junctions). Integrins link cells to the ECM. Therefore, repair of injured tissues involves not only regeneration of lost cells and the ECM but also reintegration of the newly formed elements. In parenchymal organs such as liver, kidney, and lung, various types of cells are involved in the process of tissue restoration. Nonparenchymal cells of mesenchymal origin residing in the tissue, such as resident macrophages and endothelial cells, and those migrating to the site of injury, such as blood monocytes, produce factors that stimulate parenchymal cells to divide and stimulate some specialized cells (eg, the stellate cells in the liver) to synthesize ECM molecules.
Replacement of Lost Cells by Mitosis
Soon after injury, cells adjacent to the damaged area enter the cell division cycle (Fig. 3-25). Enhanced DNA synthesis is detected experimentally as an increase in the labeling index (which is the proportion of cells that incorporate administered 3H-thymidine or bromodeoxyuridine into their nuclear DNA during the S phase of the cycle), or by increased expression of proliferating cell nuclear antigen (PCNA, a trimeric protein that functions as a sliding clamp to hold DNA polymerase-δ to the template DNA strand during DNA replication). Also, mitotic cells can be observed microscopically. As early as 2 to 4 hours after administration of a low dose of carbon tetrachloride to rats, the mitotic index in the liver increases dramatically, indicating that cells already in the G2 phase progress rapidly to the M phase. The mitotic activity of the hepatocytes culminates at 36 to 48 hours, after a full transit through the cycle, indicating that quiescent cells residing in G0 enter and progress to mitosis (M). Peak mitosis of nonparenchymal cells occurs later, after activation and replication of parenchymal cells. In severe toxic liver injury, when hepatocyte replication is impaired (eg, in rats dosed with galactosamine or acetylaminofluorene, and in humans intoxicated with acetaminophen), restoration of the liver may depend on stem cell–derived cells, called oval cells, which are located in the terminal bile ductules. These cells proliferate and differentiate into both hepatocytes and biliary epithelial cells (Fausto et al., 2006; Vessey and Hall, 2001). As oval cells produce α-fetoprotein, the increase in serum α-fetoprotein levels indicates an improved outcome of acetaminophen-induced injury. Hepatic sinusoidal endothelial cells originate from bone marrow progenitors and when they are chemically injured (eg, by monocrotaline) bone marrow–derived precursor cells move into the sinusoids and contribute to their replacement, as demonstrated in a rat model of hepatic sinusoidal obstruction syndrome (Harb et al., 2009). Accordingly, bone marrow infusion ameliorates, and myelosuppression aggravates, such liver injury. In an ozone-exposed lung, the nonciliated Clara cells and type II pneumocytes undergo mitosis and terminal differentiation to replace, respectively, the damaged ciliated bronchial epithelial cells and type I pneumocytes (Mustafa, 1990). In some tissues, such as intestinal mucosa and bone marrow, stem cells first divide to provide self-renewal and then differentiate to replace more mature cells lost through injury.
The cell division cycle and the participating cyclins and cyclin-dependent protein kinases. Areas representing phases of the cycle are meant to be proportional to the number of cells in each phase. Normally, most cells are in G0 phase, a differentiated and quiescent state. After receiving signals to divide, they progress into the G1 phase of the cell division cycle. G0/G1 transition involves activation of immediate early genes so that cells acquire replicative competence. Now increasingly responsive to growth factors, these cells progress to the phase of DNA synthesis (S). If this progression is blocked (eg, by the accumulated p53 protein), the cells may undergo apoptosis (A). After DNA replication, the cells prepare further for mitosis in the G2 phase. Mitosis (M) is the shortest phase of the cell cycle (approximately 40 minutes out of the 40-hour-long cycle of hepatocytes) and most likely requires the largest energy expenditure per unit of time. The daughter cells produced may differentiate and enter into the pool of quiescent cells (G0), substituting for those which had been lost. During the cycle, the levels of various cyclins temporarily surge by synthesis and degradation (see figure). These proteins bind to and activate specific cyclin-dependent protein kinases (Cdk, see figure), which, in turn, phosphorylate and thus activate enzymes and other proteins required for DNA replication and cell division (Johnson and Walker, 1999) (see Fig. 3-32). In addition to cyclines, phosphorylation also regulates the activity of Cdks: phosphorylation by Wee1 protein kinase (not shown) inactivates Cdk1 and Cdk2, whereas dephosphorylation by cdc25 phosphatases activates them. After tissue necrosis, the number of cells entering the cell division cycle markedly increases at areas adjacent to the injury. The proportion of cells that are in S phase in a given period is reflected by the labeling index, whereas the percentage of cells undergoing mitosis is the mitotic index (see text).
Sequential changes in gene expression occur in cells that are destined to divide. In rats subjected to partial hepatectomy to study regeneration of the liver, oligonucleotide microarray analysis revealed that more than 150 genes are involved in the early gene response with upregulation or downregulation (Su et al., 2002). The overexpressed genes include those that code for TFs important in proliferative signaling, such as c-fos, c-jun, Egr1, and c-myc (see Fig. 3-12), the genes of the antiapoptotic protein Bcl-XL (see Fig. 3-19), and that of mdm2 that constrains p53, a proapoptotic and cell cycle inhibitory TF (see Fig. 3-32). Interestingly, some genes whose products decelerate the cell cycle also become temporarily overexpressed (eg, the cyclin-dependent kinase inhibitor p21 and gadd45; see Fig. 3-32), suggesting that this duality keeps tissue regeneration precisely regulated. Nevertheless, the genetic expression is apparently reprogrammed so that DNA synthesis and mitosis gain priority over specialized cellular activities. For example, as a result of dedifferentiation, regenerating hepatocytes underexpress cytochrome P450, N-acetyltransferase-2, as well as PPARα, and hepatic stellate cells cease to accumulate fat and vitamin A.
It has been speculated that the regenerative process is initiated by the release of chemical mediators from damaged cells. The nonparenchymal cells, such as resident macrophages and endothelial cells, are receptive to these chemical signals and produce a host of secondary signaling molecules, cytokines, and growth factors that promote and propagate the regenerative process (Fig. 3-26). In rodents subjected to partial hepatectomy, the initial or priming phase of liver regeneration is controlled by the cytokines TNF and IL-6, whose hepatic mRNA and serum levels increase. TNF originates from the Kupffer cells. This cytokine acts on these macrophages in an autocrine manner, activating its receptor (item 2 in Fig. 3-12) and the coupled signal transducing network. This in turn causes activation of NF-κB, which increases IL-6 expression. The secreted IL-6 then acts on the hepatocytes and through its receptor (item 1 in Fig. 3-12) activates Janus kinase (JAK) and induces TFs (eg, Stat3, C/EBPβ), which activate several target genes. This cytokine network promotes transition of the quiescent liver cells (G0) into cell cycle (G1) and makes them receptive to growth factors (“priming”). Growth factors, especially the hepatocyte growth factor (HGF), transforming growth factor-α (TGF-α), and heparin-binding epidermal growth factor–like growth factor (HB-EGF), initiate the progression of the “primed” cells in the cycle toward mitosis (Costa et al., 2003; Fausto et al., 2006). Despite its name, neither the formation nor the action of HGF is restricted to the liver. It is produced by resident macrophages and endothelial cells of various organs—including liver, lung, and kidney—and in a paracrine manner activates receptors on neighboring parenchymal cells (Fig. 3-26). In rats intoxicated with carbon tetrachloride, the synthesis of HGF in hepatic and renal nonparenchymal cells increases markedly (Noji et al., 1990) and HGF levels in blood rise rapidly (Lindroos et al., 1991). The communication between parenchymal and nonparenchymal cells during tissue repair is mutual. For example, TGF-α, a potent mitogen produced by regenerating hepatocytes, acts both as an autocrine and a paracrine mediator on liver cells as well as on adjacent nonparenchymal cells (Fig. 3-26). By activating their receptors (item 4 in Fig. 3-12), these growth factors initiate signaling through the MAPK pathway and the PI3K–Akt pathway (Fig. 3-12), thereby mediating activation of TFs (c-Jun, c-Fos, c-Myc, FoxM1B, NF-κB, Stat3). These, among others, induce cyclins and the protein phosphatase cdc25, two groups of short-lived regulatory proteins. Then cyclins activate Cdks by associating with them (Fig. 3-25), whereas cdc25 activates Cdk1 and Cdk2 by dephosphorylating their 2 amino acid residues (Thr14 and Tyr15). The activated Cdks accelerate the cell cycle mainly by phosphorylation of pRb. This in turn releases the TF E2F, which induces enzymes and regulatory proteins needed for cell cycle progression (Fig. 3-32). The growth factor signaling also activates mTOR that upregulates mRNA translation, thereby meeting the demand for increased protein synthesis (see Fig. 3-29). Similar signaling appears to mediate regeneration of the S-(1,2-dichlorovinyl)-l-cysteine-injured kidney that exhibits increased expression of the cytokine IL-6, the growth factors TGF-α and HB-EGF, the growth factor receptors EGFR (also receptor for TGF-α) and IGF-1R, and the MAP kinase isoform Erk1 (Vaidya et al., 2003).
Mediators of tissue repair and side reactions to tissue injury in liver: (1) growth factors promoting replacement of cells and the extracellular matrix; (2) mediators of inflammation, acute-phase protein (APP) synthesis, and fever; and (3) cytotoxic mediators of inflammatory cells. HGF, hepatocyte growth factor; PDGF, platelet-derived growth factor; TGF-α, transforming growth factor-alpha; TGF-β, transforming growth factor-beta; NO•, nitric oxide; PGI2, prostacyclin; LTC4, leukotriene C4; IL, interleukin; LTB4, leukotriene B4; PAF, platelet-activating factor; CINC (the rat homologue of IL-8), cytokine-induced neutrophil chemoattractant; MCP-1, monocyte chemotactic protein; TNF, tumor necrosis factor. Cells presented are E, endothelial cells; G, granulocyte; H, hepatocyte; M, macrophage (Kupffer cell); S, stellate cell (also called perisinusoidal, Ito, or fat-storing cell). *Rather than the endothelial cells, other stromal cells are the main sources of chemokines (eg, stellate cells for MCP-1). Solid arrows represent effects of growth factors on cell division, whereas the dashed arrow shows the effect on extracellular matrix formation. When directed to a cell, pointed and blunted arrows indicate stimulation and inhibition, respectively. See text for further details.
Although the cytotokine- and growth factor–controlled mitotic cell replacement is likely an essential part in the repair of most tissues built up of cells with proliferative capacity, there are also tissue-specific features of tissue repair. For example in the liver, bile acids also stimulate hepatocyte proliferation through their nuclear receptor, FXR (Huang et al., 2006). In mice subjected to partial hepatectomy, liver regrowth is hastened by bile acid feeding, but is markedly delayed when FXR is deleted or recirculation of bile acids from the intestine to the liver is prevented. Epithelia composed of a single cell layer form important barriers; therefore, replacement of mortally injured epithelial cells, which become detached from the basement membrane, is an urgent need. This can be achieved more rapidly by cell migration than by mitotic cell replacement. For example, in the damaged mucosa of the GI tract, cells of the residual epithelium rapidly migrate to the site of injury as well as elongate and become thin to reestablish the continuity of the surface even before this could be achieved by cell replication. Cell movement involves orderly dissociation of cadherin-mediated cell–cell contacts (involving β-catenin phosphorylation) and integrin-mediated cell–ECM contacts at focal adhesions (involving FAK phosphorylation), assembly of actin stress fibers, and formation of lamellopodia (cell projection filled with F-actin meshwork). Mucosal repair is dictated not only by growth factors and cytokines operative in tissue repair elsewhere but also by trefoil factors (TFFs). TFFs are small (7–12 kDa) protease-resistant proteins that are abundantly secreted from specific mucosal cells (eg, intestinal goblet cells) and are associated with the mucous layer of the GI tract (Taupin and Podolsky, 2003). TFF expression is rapidly upregulated at the margins of mucosal injury by cytokine and growth factor signaling. Whereas growth factors (eg, HGF and EGF) exert both motogenic (motility-increasing) and mitogenic effects on enterocytes, TFFs are potent motogens, but are not mitogens. Although TFFs do not act alone, they are the only peptides shown to be essential for restitution of the injured intestinal mucosa. Normal mice exposed to dextran sodium sulfate in the drinking water develop diffuse colonic mucosal injury, some exhibiting bloody diarrhea. In contrast, the majority of TFF3-null mice develop frank bloody diarrhea and die in response to dextran sodium sulfate. The immediate target molecule (eg, receptor) for TFFs remains unknown. Migration of the surviving cells also precedes mitotic cell replacement in the tubular epithelium of the injured kidney. It appears that ECM components, such as collagen IV, beneath the tubular epithelial cells aid in the restitution of the injured epithelium and reestablishment of its polarity (Nony and Schnellmann, 2003).
Replacement of the Extracellular Matrix
The ECM is composed of proteins, glycosaminoglycans, and the glycoprotein and proteoglycan glycoconjugates (Gressner, 1992). In liver, these molecules are synthesized by stellate or fat-storing cells located in the space of Disse, between the hepatic sinusoid and the hepatocytes (Fig. 3-26). The stellate cells become activated during liver regeneration, undergoing mitosis and major phenotypic changes. The latter changes include not only increased synthesis and secretion of ECM constituents but also expression of α-smooth muscle actin as well as the myogenic TF MyoD and loss of fat, vitamin A, and PPARγ content. Thus, resting stellate cells become transdifferentiated into myofibroblast-like contractile and secretory cells. Activation of stellate cells is mediated chiefly by 2 growth factors—platelet-derived growth factor (PDGF) and TGF-β (Fig. 3-26). Both may be released from platelets (which accumulate and degranulate at sites of injury) and later from the activated stellate cells themselves. The main sources of TGF-β, however, are the neighboring tissue macrophages residing in the hepatic sinusoids (Gressner, 1992). A dramatic increase in TGF-β mRNA levels in Kupffer cells is observed with in situ hybridization after carbon tetrachloride–induced hepatic necrosis. Proliferation of stellate cells is induced by the potent mitogen PDGF. Through its receptor (like item 4 in Fig. 3-12), PDGF activates the PI3K–Akt pathway, which not only signals for proliferation but also inactivates the FoxO1 TF, which would halt the cell cycle in G1 phase by increasing the expression of p27, a cyclin-dependent kinase inhibitor. TGF-β acts on the stellate cells to induce their transdifferentiation and to stimulate the synthesis of ECM components, including collagens, fibronectin, tenascin, and proteoglycans. TGF-β acts through its Ser/Thr kinase receptor (item 8 in Fig. 3-12), which phosphorylates the TFs Smad2 and 3 (Flanders, 2004). TGF-β also plays a central role in ECM formation in other tissues. In kidney and lung, for example, TGF-β targets the mesangial cells and the septal fibroblasts, respectively. Remodeling of the ECM is aided by matrix metalloproteinases, which hydrolyze specific components of the matrix, as well as by tissue inhibitors of matrix metalloproteinases. The former group of these proteins originates from various types of nonparenchymal cells, including inflammatory cells; however, their inhibitors are mainly produced by stellate cells (Arthur et al., 1999).
The way in which tissue regeneration is terminated after repair is unclear, but the gradual dominance of TGF-β, which is a potent antimitogen and apoptogen, over mitogens is a contributing factor in the termination of cell proliferation. ECM production may be halted by an intracellular negative feedback mechanism in the ECM-producing cells, that is, by induction of Smad7 (an inhibitory Smad), which competitively inhibits phosphorylation of the receptor-activated Smads (Smad2 and 3). In addition, extracellular products of the proliferative response, such as the proteoglycan decorin and the positive acute-phase protein α2-macroglobulin, can bind and inactivate TGF-β; thereby they may contribute to its silencing (Gressner, 1992).
Side Reactions to Tissue Injury
In addition to mediators that aid in the replacement of lost cells and the ECM, resident macrophages and endothelial cells activated by cell injury also produce other mediators that induce ancillary reactions with uncertain benefit or harm to tissues (Fig. 3-26). Such reactions include inflammation, altered production of acute-phase proteins, and generalized reactions such as fever.
Alteration of the microcirculation and accumulation of inflammatory cells are the hallmarks of inflammation. These processes are largely initiated by resident macrophages secreting cytokines, such as TNF and interleukin-1 (IL-1), in response to tissue damage (Baumann and Gauldie, 1994) (Fig. 3-26). These cytokines, in turn, stimulate neighboring stromal cells, such as the endothelial cells and fibroblasts, to release mediators that induce dilation of the local microvasculature and cause permeabilization of capillaries. Activated endothelial cells also facilitate the egress of circulating leukocytes into the injured tissue by releasing chemoattractants and expressing cell adhesion molecules, which are cell surface glycoproteins (Jaeschke, 1997). One group of cell adhesion molecules, called selectins, located on the membrane of endothelial cells, interact with their ligands on the surface of leukocytes, thereby slowing down the flow of these cells and causing them to “roll” on the capillary surface. Subsequently a stronger interaction (adhesion) is established between the endothelial cells and leukocytes with participation of intercellular adhesion molecules (eg, ICAM-1) expressed on the endothelial cell membrane and integrins expressed on the membrane of leukocytes. This interaction is also essential for the subsequent transendothelial migration of leukocytes. This is facilitated by gradients of chemoattractants that induce expression of leukocyte integrins. Chemoattractants originate from various stromal cells and include chemotactic cytokines (or chemokines), such as the MCP-1 and IL-8 (whose rat homologue is the CINC-1), as well as lipid-derived compounds, such as platelet-activating factor (PAF) and leukotriene B4 (LTB4). Ultimately all types of cells in the vicinity of injury express ICAM-1, thus promoting leukocyte invasion; the invading leukocytes also synthesize mediators, thus propagating the inflammatory response. Production of most inflammatory mediators is induced by signaling, turned on by TNF and IL-1, which results in activation of TFs, notably NF-κB and C/EBP (Poli, 1998) (see Fig. 3-12). Genes of many of the proteins mentioned above (eg, selectins, ICAM-1, MCP-1, IL-8) and below (eg, inducible NOS, acute-phase proteins) as well as the genes of TNF and IL-1 themselves contain binding sites for the NF-κB (Lee and Burckart, 1998).
Inflammation—ROS and RNS Production
Macrophages, as well as leukocytes, recruited to the site of injury undergo a respiratory burst, discharging free radicals and enzymes (Weiss and LoBuglio, 1982) (Fig. 3-26). Free radicals are produced in the inflamed tissue in three ways, each of which involves a specific enzyme: Nox, NOS, or myeloperoxidase.
Nox is an electron-transporting protein complex composed of two transmembrane proteins (one of them the FAD- and heme-containing catalytic subunit) and 4 cytoplasmic proteins (including the G-protein Rac) (El-Benna et al., 2005). In resting cells Nox is dormant; however, in activated cells the cytoplasmic subunits become extensively phosphorylated and move to the membrane to assemble the Nox complex. Constituents of microorganisms, such as the bacterial LPS (the active endotoxin component of gram-negative bacteria), acting through the cell surface Toll-like receptors (TLR), and PKC activators, such as phorbol miristate acetate (PMA), may evoke Nox activation. Unlike in nonphagocytic cells, in macrophages and granulocytes activation causes a sudden and rapid electron transfer from NADPH through the FAD and heme in Nox to the molecular oxygen, releasing the thus formed superoxide anion radical (Display Formula) in a burst (“respiratory burst”) at the external membrane surface into the phagocytic vacuole:
The Display Formula can give rise to the hydroxyl radical (HO•) in two sequential steps: the first is spontaneous or is catalyzed by SOD, and the second, the Fenton reaction, is catalyzed by transition metal ions (see also Fig. 3-4):
Macrophages, but not granulocytes, generate another cytotoxic free radical, nitric oxide (•NO). This radical is produced from arginine by NOS (Wang et al., 1993), which is inducible in macrophages by bacterial endotoxin and the cytokines IL-1 and TNF:
l-Arginine + O2 → l-citrulline + •NO.
Subsequently, Display Formula and •NO, both of which are products of activated macrophages, can react with each other, yielding peroxynitrite anion; on reaction with carbon dioxide, this decays into two radicals, nitrogen dioxide and carbonate anion radical (Fig. 3-4):
Granulocytes, but not macrophages, discharge the lysosomal enzyme myeloperoxidase into engulfed extracellular spaces, the phagocytic vacuoles (Wang et al., 1993). Myeloperoxidase catalyzes the formation of HOCl, a powerful oxidizing agent, from HOOH and chloride ion:
HOOH + H+ + Cl− → HOH + HOCl.
Like HOOH, HOCl can form HO• as a result of electron transfer from Display Formula Fe2+ or from to HOCl:
All these reactive chemicals, as well as the discharged lysosomal proteases, are destructive products of inflammatory cells. Although these chemicals exert antimicrobial activity at the site of microbial invasion, at the site of toxic injury they can damage the adjacent healthy tissues and thus contribute to propagation of tissue injury (see the section “Tissue Necrosis”). Moreover, in some chemically induced injuries, inflammation plays the leading role. For example, ANIT, a cholestatic chemical, causes neutrophil-dependent hepatocellular damage. ANIT apparently acts on bile duct epithelial cells, causing them to release chemoattractants for neutrophil cells, which on invading the liver, injure hepatocytes (Hill et al., 1999). Kupffer cell activation, TNF release, and subsequent inflammation are also prominent and causative events in galactosamine-induced liver injury in rats (Stachlewitz et al., 1999).
Whereas it is well recognized that chemical-inflicted tissue injury can induce inflammation as a side reaction of tissue repair, it is becoming clear that inflammation (even if harmless alone) can precipitate an overt tissue injury on chemical exposure that is noninjurious alone. For example, a small harmless quantity of the macrophage activator LPS converts the nontoxic doses of monocrotaline, aflatoxin B1, and allyl alcohol into markedly hepatotoxic doses (Roth et al., 2003). Moreover, in rats pretreated with LPS, unlike in untreated animals, chlorpromazine, ranitidine, and trovafloxacin caused liver injury. These drugs (and many others) when given to patients can induce rare, unexpected, and not obviously dose-related liver injury. Therefore, it is hypothesized that such idiosyncratic drug reactions develop when some endotoxin exposure decreases the threshold for drug toxicity by priming the Kupffer cells that produce ROS and inflammatory mediators discussed above. Manifest or subclinical infection, GI disturbance, or alcohol consumption (which greatly increases the intestinal permeability for endotoxin) may be the source of endotoxin. It is not surprising that most idiosyncratic drug reactions affect the liver, because this organ contains 80% to 90% of the body’s fixed macrophages (ie, Kupffer cells), because the liver is the first organ to be exposed to LPS translocating from the intestinal lumen, and because the Kupffer cells not only are activated by LPS but also remove it from the circulation, thereby protecting other organs from its inflammatory effects (Roth et al., 2003). However, the causative relationship between inflammation and idiosyncrasy needs further substantiation.
Altered Protein Synthesis: Acute-Phase Proteins
Cytokines released from macrophages and endothelial cells of injured tissues also alter protein synthesis, predominantly in the liver (Baumann and Gauldie, 1994) (Fig. 3-21). Mainly IL-6 but also IL-1 and TNF act on cell surface receptors and increase or decrease the transcriptional activity of genes encoding certain proteins called positive and negative acute-phase proteins, respectively, utilizing primarily the TFs NF-κB, C/EBP, and STAT (Poli, 1998; see Fig. 3-15). Many of the hepatic acute-phase proteins, such as C-reactive protein and hepcidin, are secreted into the circulation, and their elevated levels in serum are diagnostic of tissue injury, inflammation, or neoplasm. Increased sedimentation of red blood cells, which is also indicative of these conditions, is due to enrichment of blood plasma with positive acute-phase proteins such as fibrinogen.
Apart from their diagnostic value, positive acute-phase proteins may play roles in minimizing tissue injury and facilitating repair. For example, many of them, such as α2-macroglobulin and α1-antiprotease, inhibit lysosomal proteases released from the injured cells and recruited leukocytes. Haptoglobin binds hemoglobin in blood, MT complexes metals in the cells, heme oxygenase oxidizes heme to biliverdin, and opsonins facilitate phagocytosis. Thus, these positive acute-phase proteins may be involved in the clearance of substances released on tissue injury. Induction of hepcidin, which triggers degradation of the iron exit channel ferroportin (see the section “Mechanisms of Adaptation”) and thus decreases the duodenal Fe2+ absorption and Fe2+ export from macrophages, may be protective by limiting iron availability in the circulation for pathogens and by minimizing the Fe2+-catalyzed Fenton reaction (Fig. 3-4), the source of reactive hydroxyl radicals that may cause tissue injury.
Negative acute-phase proteins include some plasma proteins, such as albumin, transthyretin, and transferrin, as well as hepatic enzymes (eg, several forms of cytochrome P450 and glutathione S-transferase), ligand-activated TFs (eg, PPARα and the bile acid receptor FXR), and transporters, such as bile acid transporters at the sinusoidal and canalicular membrane of hepatocytes (NTCP and bile salt export pump [BSEP], respectively) and the bile canalicular export pump Mrp2. Because the latter enzymes and transporters play important roles in the toxication, detoxication, and excretion of endobiotics and xenobiotics, the disposition and toxicity of bile acids and toxicants may be altered markedly during the acute phase of tissue injury.
Although the acute-phase response is phylogenetically preserved, some of the acute-phase proteins are somewhat species-specific. For example, during the acute phase of tissue injury or inflammation, C-reactive protein and serum amyloid A levels dramatically increase in humans but not in rats, whereas the concentrations of α1-acid glycoprotein and α2-macroglobulin increase markedly in rats but only moderately in humans.
Cytokines released from activated macrophages and endothelial cells at the site of injury also may evoke neurohumoral responses. Thus, IL-1, TNF, and IL-6 alter the temperature set point of the hypothalamus, triggering fever. IL-1 possibly also mediates other generalized reactions to tissue injury, such as hypophagia, sleep, and “sickness behavior” (Rothwell, 1991). In addition, IL-1 and IL-6 act on the pituitary to induce the release of ACTH, which in turn stimulates the secretion of cortisol from the adrenals. This represents a negative feedback loop because corticosteroids inhibit cytokine gene expression.
Adaptation may be defined as a harm-induced capability of the organism for increased tolerance to the harm itself. It involves responses acting to preserve or regain the biological homeostasis in the face of increased harm. Theoretically, adaptation to toxicity may result from biological changes causing (1) diminished delivery of the causative chemical(s) to the target, (2) decreased size or susceptibility of the target, (3) increased capacity of the organism to repair itself, and (4) strengthened mechanisms to compensate the toxicant-inflicted dysfunction. Mechanistically, adaptation involves sensing the noxious chemical and/or the initial damage or dysfunction, and a response that typically occurs through altered gene expression. Such mechanisms will be briefly overviewed below.
Adaptation by Decreasing Delivery to the Target
The first step in the development of toxicity is delivery of the ultimate toxicant (a xenobiotic, its metabolite or xenobiotic-generated ROS and RNS) to the target (Fig. 3-2). Certain chemicals induce adaptive changes that lessen their delivery by diminishing the absorption, increasing their sequestration by intracellular binding proteins, enhancing their detoxication, or promoting their cellular export.
Repression of Iron Absorption
Adaptive mechanisms triggered by iron itself adjust the intestinal absorption of this essential yet potentially harmful metal ion, a catalyst of hydroxyl radical formation by the Fenton reaction (Fig. 3-4), to its demand. Fe2+ is taken up from the intestinal lumen into the enterocyte by the proton-coupled divalent metal transporter 1 (DMT1) localized in the luminal membrane and is exported across the basolateral membrane of these cells through the ferroportin iron channel. High iron intake diminishes the expression of both DMT1 and ferroportin in the enterocytes, whereas low intake has the opposite effect.
The expression of DMT1 is regulated at the translational level by the intracellular iron-regulatory proteins (IRP) that contain a [4Fe-4S] cluster. Iron supply influences the Fe content of this cluster and in turn affects the capacity of IRP to bind to mRNAs containing iron response element (IRE) in their untranslated regions and to alter translation. In iron deficiency, loss of Fe from the [4Fe-4S] cluster allows the thus formed IRP1 apoprotein to bind to and protect DMT1 mRNA from enzymatic degradation. On becoming more abundant, DMT1 mRNA translates into more DMT1 protein. Conversely, iron overload saturates IRP1 with Fe, making it incapable of binding to and stabilizing the DMT1 mRNA. The consequential loss of DMT1 mRNA and its translation into less DMT1 protein then limits Fe2+ absorption. This physiologically important adaptive mechanism may have toxicological corollary. Since DMT1 also mediates the transport of Cd2+, DMT1 overexpression in iron deficiency increases intestinal absorption of this highly toxic metal ion (Park et al., 2002).
Ferroportin allows exit of iron not only from enterocytes but also from macrophages (which obtain iron from phagocytosed senescent or damaged erythrocytes) and hepatocytes (which store iron). The abundance of ferroportin in the plasma membrane of these cells is regulated directly by hepcidin, a peptide hormone secreted by the hepatocytes of the liver (Ganz, 2011). By binding to ferroportin, hepcidin triggers the internalization and lysosomal hydrolysis of ferroportin. Hepatic production of hepcidin is sensitive to iron supply. Iron deficiency decreases hepcidin secretion, precluding ferroportin degradation and thus increasing the abundance of this iron exporter in the basolateral membrane of enterocytes. This (together with the upregulated DMT1 in the apical membrane of these cells) permits increased iron absorption from the duodenum. In contrast, iron overload increases the secretion of hepcidin, which in turn induces degradation of ferroportin, thereby preventing iron export from the enterocytes. Thus, decreased synthesis of DMT1 (see above) and degradation of ferroportin act in concert to shut down iron absorption from the intestine and protect against iron excess from the diet, from lysed red blood cells, or from ferritin. The toxicological significance of this mechanism is exemplified by findings with diquat, a redox cycling hepatotoxic herbicide that (like paraquat, see Fig. 3-3) can accept an electron, forming diquat monocation radical (DQ•+), which in turn transfers the electron to O2, producing superoxide anion radical (Display Formula). Soon after diquat injection to rats, hepatic free iron increased dramatically (purportedly because both DQ•+ and had reductively released Fe2+ from ferritin) followed by a rise in hepatic hepcidin mRNA first, and later by a decrease in plasma iron (Higuchi et al., 2011), a seemingly appropriate protective response. Inappropriately elevated hepcidin, however, may cause iron deficiency. For example, one might speculate that elevated plasma levels of prohepcidin observed in chronic lead intoxication contribute to lead-induced anemia.
Hepcidin production by hepatocytes is transcriptionally regulated by Smad4 and STAT3 TFs through their cognate response elements in the hepcidin gene (Ganz, 2011). Two pathways converge on Smad4: one signals the amount of intracellular iron, whereas the other signals the quantity of plasma iron. Increase in intracellular iron in the liver cells is detected by unidentified sensors that induce secretion of bone morphogenetic protein (BMP6, a TGF-β-like signaling molecule) acting in an autocrine or paracrine manner on its cell surface BMP receptor that activates Smad4 by phosphorylation (like TGF-β does in pathway 8 shown in Fig. 3-12). Increase in plasma iron is detected by the transferrin receptor (TFR) of hepatocytes when activated by transferrin saturated with iron. Then TFR relays the signal through adapter proteins to the BMP receptor, which thus becomes increasingly sensitive to its ligand, BMP6. The pathway descending to STAT3 is initiated by IL-6 (see Fig. 3-12, item 1), a cytokine that triggers the acute-phase response, explaining why hepcidin is also a positive acute-phase protein (see above). In summary, the pathways initiated by transferrin and BMP receptors upregulate the hepatic production of hepcidin that reduces iron absorption in response to iron overload, whereas the pathway triggered by IL-6 has a similar effect in response to inflammation, tissue injury, or cancer, conditions with increased IL-6 production. The latter disorders may cause anemia by this mechanism.
Induction of Ferritin and Metallothionein
Adaptive cellular accumulation of the binding proteins ferritin and MT is protective against their respective ligands, iron and cadmium ions. Interestingly, upregulation of ferritin by iron overload, like downregulation of DMT1, is also translationally mediated by IRP1. However, the apoIRP1 (which is formed in iron deficiency) acts oppositely on ferritin mRNA and blocks its translation. Iron overload relieves this blockade, as it yields IPR1 with the [4Fe-4S] cluster, which does not bind to the IRE in ferritin mRNA, thus causing a surge in ferritin translation. Ferritin is protective as it removes Fe2+ from the Fenton reaction (Fig. 3-4) and incorporates it as Fe3+ through its ferroxidase activity.
MT is greatly induced by cadmium and elevated levels of MT protect the liver by restricting distribution of this toxic metal ion to sensitive intracellular targets (Klaassen et al., 1999). Induction of MT by Cd2+ is likely indirect, mediated by Zn (displaced from intracellular binding sites), which activates the metal-responsive transcription factor 1 (MTF-1) that in turn augments transcription of the MT gene by binding to the metal-responsive elements in its promoter (Lichtlen and Schaffner, 2001).
Induction of Detoxication—The Electrophile Stress Response, Part 1
Adaptive increases in detoxication (ie, elimination of xenobiotics, their reactive metabolites, harmful endobiotics, or ROS and RNS by biotransformation) and cellular export have a major role in limiting toxicity. Such adaptation is typically induced by compounds with thiol reactivity (ie, soft electrophiles, oxidants, and those generating oxidative stress), which are sensed by the cytosolic Keap1–Nrf2 protein complex. The response is initiated by the TF Nrf2, which activates genes with electrophile response element (EpRE) in their regulatory region (Fig. 3-27; Dinkova-Kostova et al., 2005). Normally Nrf2 is retained in the cytoplasm by Keap1, a cysteine-rich homodimeric protein. Keap1 keeps Nrf2 inactive and at low intracellular levels by anchoring it to the cytoskeleton and also by linking it to cullin 3, a component in certain E3 ubiquitin ligase complexes, thereby initiating ubiquitination of Nrf2 for subsequent proteasomal degradation. On disruption of the Keap1–Nrf2 complex, the active Nrf2 escapes rapid degradation and accumulates in the cell. Electrophiles, such as quinones (eg, t-butylquinone), quinoneimines (eg, NAPBQI derived from acetaminophen), quinone methides (eg, metabolite of butylated hydroxytoluene), α,β-unsaturated aldehydes and ketones (eg, the lipid peroxidation products 4-oxonon-2-enal, 4-hydroxynon-2-enal, and 15-A2t-isoprostane), isothiocyanates (eg, ANIT), thiol-reactive metal ions (eg, Cd2+), and trivalent arsenicals, as well as direct and indirect oxidants (eg, HOOH, diquat, quinones) may attack Keap1 at its reactive cysteine thiol groups by binding to them covalently or oxidizing them, thereby forcing Keap1 to release Nrf2. After being released from Keap1, Nrf2 translocates into the nucleus, forms a heterodimer with small Maf proteins, and activates genes through binding to EpREs.
Signaling by Keap1/Nrf2 mediates the electrophile stress response. Normally NF-E2-related factor 2 (Nrf2) is kept inactive and at a low intracellular level by interacting with Keap1 that promotes its proteasomal degradation by ubiquitination. Electrophiles covalently bind to, whereas oxidants oxidize the reactive thiol groups of Keap1, causing Keap1 to release Nrf2. Alternatively, Nrf2 release may follow its phosphorylation by protein kinases. After being released from Keap1, the active Nrf2 accumulates in the cell, translocates into the nucleus, and forms a heterodimer with small Maf proteins to activate genes that contain electrophile response element (EpRE) in their promoter region. These include enzymes, binding proteins, and transporters functioning in detoxication and elimination of xenobiotics, ROS, and endogenous reactive chemicals, as well as some proteins that can repair or eliminate oxidized proteins. Induction of such proteins represents an electrophile stress response that provides protection against a wide range of toxicants. Nrf1, a transcription factor structurally related to Nrf2, also interacts with Keap1 and Maf proteins as well as EpRE and its role is partially overlapping with that of Nrf2. Abbreviations: AR, aldose reductase; CES carboxylesterase; EH1, microsomal epoxide hydrolase; GCL, glutamate–cysteine ligase; GGT, gamma-glutamyl transpeptidase; GPX2, glutathione peroxidase 2; GR, glutathione reductase; GST, glutathione S-transferase; HO-1, heme oxygenase 1; NQO1, NAD(P)H:quinone oxidoreductase; Mrp2, Mrp3, and Mrp4, multidrug resistance protein 2, 3, and 4; SOD1, superoxide dismutase 1; Srx1, sulfiredoxin 1; UGT, UDP-glucuronosyltransferase; Trx, thioredoxin; TrxR, thioredoxin reductase.
There are many genes with EpRE motifs that encode proteins known to be important in detoxication and export (Fig. 3-27). These include genes that code for (1) enzymes that detoxify xenobiotics (eg, NQO1, NQO2, AR, EH-1, CES1e1, CES2a6, GST, and UGT), (2) enzymes that eliminate (ie, SOD1) and HOOH (eg, GPX2, catalase, and Srx1; the latter reduces the overoxidized peroxiredoxins that in turn can reduce not only HOOH [see Fig. 3-5] but also lipid hydroperoxides and peroxinitrite), (3) proteins that detoxify heme (HO-1) and Fe2+ (ferritin), (4) enzymes involved in the synthesis of GSH and its regeneration from GSSG (eg, GCL and the NADPH-forming G6PDH), and (5) transporters that pump xenobiotics and their metabolites out of cells (eg, Mrp2, 3, and 4. Other Nrf2-induced proteins of known toxicological relevance will be mentioned later.
As demonstrated in Fig. 3-28 through the example of t-butylhydroquinone (tBHQ), these Nrf2-mediated adaptive changes facilitate elimination and detoxication of electrophilic chemicals, as well as ROS that may be generated by them, and assist in repairing or removing damaged proteins (to be discussed later); therefore, Nrf2 conveys protection against a wide range of toxicants. Indeed, Nrf2 knockout mice are more sensitive to the hepatotoxicity of acetaminophen, pentachlorophenol, and carbon tetrachloride, the pulmonary toxicity of butylated hydroxytoluene, hyperoxia, or cigarette smoke, the neurotoxicity of 3-nitropropionic acid, and the carcinogenicity of benzo[a]pyrene (Klaassen and Reisman, 2010). Conversely, liver-specific deletion of Keap1 constitutively activates Nrf2 in hepatocytes, causing them to overexpress many detoxifying enzymes and become resistant to acetaminophen-induced hepatotoxicity.
Adaptive changes in response to the t-butylhydroquinone (tBHQ)–induced electrophile stress that influence the metabolic fate and some effects of tBHQ and of superoxide anion radical generated in the course of tBHQ biotransformation. This figure illustrates the numerous proteins (ie, enzymes, exporters, repair proteins) that, when induced in response to the tBHQ-induced electrophile stress, facilitate the detoxication and export of tBHQ and/or its metabolites, the detoxication of ROS formed during the biotransformation of tBHQ, and repair or removal of proteins damaged by the reactive metabolites. Proteins induced by Nrf2 as a result of adaptation to the electrophile stress are marked with an asterisk.
tBHQ (1) can be detoxified by UDP-glucuronosyltransferase to form tBHQ glucuronide (2) and toxified by cytochrome P450–catalyzed dehydrogenation to t-butylquinone (tBQ; 3), which contains electrophilic carbon atoms (+), and is believed to be the actual inducer of electrophile stress response. tBQ can undergo 3 biotransformations. First, it can be detoxified by conjugation with glutathione (GSH) to form tBHQ-SG (4), which, together with tBHQ-glucuronide, is exported from the cell by Mrp2, 3, and 4. Second, tBQ can covalently bind to SH groups in proteins (5), including Keap1. Third, by accepting an electron (e), tBQ can form t-butylsemiquinone radical (6), which can pass the electron to molecular oxygen to form superoxide anion radical (O2•¯), completing a redox cycle. In a process catalyzed by superoxide dismutase (SOD), O2•¯ can form HOOH, which is detoxified by GSH-peroxidase (GPX), using GSH (whose synthesis is rate-limited by GCL), or toxified by Fe2+-catalyzed Fenton reaction to form hydroxyl radical (HO•). HO• can react with proteins to form oxidized proteins (Prot-C=O, Prot-S-S), which can be degraded in the proteasome, or Prot-S-S can also be repaired through reduction by the thioredoxin (Trx)–thioredoxin reductase (TrxR) system. HO• can also react with lipid (L) and form lipid hydroperoxide (LOOH), which can be reduced to lipid alcohol (LOH) by GPX at the expense of GSH. This results in formation of glutathione disulfide (GSSG), which can be reduced back to GSH by glutathione reductase (GR) at the expense of NADPH generated by glucose-6-phosphate dehydrogenase (G6PDH). Virtually all of these processes become more effective after the electrophile stress response.
Nrf2 can be activated by treatment with several chemicals of low toxicity that are thiol-reactive as soft electrophiles or oxidants. These include tBHQ, butylated hydroxyanisole (BHA), sulforaphane (a compound in broccoli with an electrophilic isothiocyanate group), curcumin (a compound in turmeric containing 2 α,β-unsaturated ketone groups), resveratrol (a hydroquinone-type compound in red grapes that may undergo oxidation to a quinone and redox cycling, like shown for tBHQ in Fig. 3-28), derivatives of oleanolic acid, such as 2-cyano-3,12-dioxooleana-1,9(11)-dien-28-oic acid (CDDO, which contains 2 electrophilic enone groups), and oltipraz (a dithiolenethione compound whose metabolites can form mixed disulfides with thiols, probably also with those on Keap1). These chemicals induce Nrf2 target genes (Fig. 3-27) and protect from toxicant-induced tissue injury and cancer. For example, treatment with oleanolic acid or CDDO ameliorates acetaminophen-induced hepatotoxicity and aflatoxin-induced hepatocarcinogenesis (Klaassen and Reisman, 2010). The superpotent Nrf2 agonist CDDO is now tested clinically for use in chronic kidney diseases. Of these chemopreventive agents, some (eg, tBHQ, BHA, and resveratrol) are antioxidants; hence the inappropriate names “antioxidant response” and “antioxidant response element” (ARE) for the Nrf2-mediated adaptive alterations and the cognate DNA-binding site for Nrf2, respectively. It became apparent only later that the electrophilic quinone metabolites of these chemicals are the inducers and not the antioxidants.
Adaptation by Decreasing the Target Density or Responsiveness
Decreasing the density and sensitivity of the xenobiotic target is an adaptation mechanism for several cell surface receptors. Such alterations underlie the tolerance induced by opioids, abused drugs of considerable clinical toxicological interests.
Induction of Opioid Tolerance
The main target of opioids (eg, morphine, heroine, methadone) is the μ-opioid receptor. Stimulation of this Gi protein–coupled inhibitory receptor by an agonist results in adenyl cyclase inhibition (causing decline in cyclic AMP levels and PKA activity) and K+ channel opening (causing hyperpolarization) (Fig. 3-15) in neurons with opioid receptors, such as those in the midbrain periaqueductal gray. Even brief stimulation induces adaptive alterations: the receptor is desensitized by G-protein receptor kinase–mediated phosphorylation and β-arrestin binding, and then becomes uncoupled from the G protein and internalized via a clathrin-dependent pathway. Whereas some receptors are recycled to the cell membrane, others are degraded in the lysosomes, causing receptor downregulation (Bailey and Connor, 2005). On prolonged stimulation, adenyl cyclase signaling undergoes a compensatory increase. Tolerance to opioids, though far from being clarified mechanistically, may result from downregulation of the receptors and upregulation of adenyl cyclase signaling. These changes would require increasing doses of agonist to produce an effect (ie, inhibition of adenyl cyclase signaling) as intensive as after its first application. These adaptive changes could also explain the withdrawal reaction, that is, appearance of clinical symptoms (dysphoria, excitement, pain sensation), contrasting with the pharmacologic effects of opioids (euphoria, sedation, analgesia), on abrupt termination of drug treatment, because withdrawal of the opioid would disinhibit the reinforced signaling it had inhibited. Nevertheless, mechanistic relationships between tolerance and the withdrawal reaction remain controversial (Bailey and Connor, 2005). An important clinical feature of opioid tolerance is that the tolerance to the respiratory depressive effect is short-lived and sensitivity returns after some abstinence. Therefore, abusers often kill themselves with a dose tolerated earlier.
Adaptation by Increasing Repair
There are several repair mechanisms that can be induced after toxicant exposure. Some of these may aid in repairing damaged molecules, proteins, and DNA, others in regenerating the injured tissue.
Induction of Enzymes Repairing Oxidized Proteins—The Electrophile Stress Response, Part 2
After sublethal exposure to chemicals, such as tBHQ, 4-hydroxynon-2-enal, and Cd2+, not only enzymes functioning in xenobiotic detoxication but also some of those mediating protein repair become overexpressed as part of the above-described electrophile response. The induced proteins include thioredoxin 1 (Trx1) and thioredoxin reductase 1 (TR1), which can reduce oxidized proteins (protein disulfides, -sulfenic acids, and -methionine sulfoxides) (Fig. 3-22), and several subunits of the proteasome complex, which hydrolyzes damaged proteins. These repair proteins are transcribed from genes containing EpRE, and their transcription is controlled by Nrf2 (Fig. 3-27). As Trx1 and TR1 are reduction partners for ribonucleotide reductase, they support this enzyme in forming deoxyribonucleotides for DNA synthesis. Thus, induction of Trx1 and TR1 also assists DNA repair.
Induction of Chaperones Repairing Misfolded Proteins—The Heat-Shock Response
The cellular abundance of many molecular chaperones, which can disaggregate and refold denatured proteins, also increases after physical and chemical stresses (eg, heat, ionizing radiation, oxidants, electrophile reactants, and metal ions). Two adaptive reactions involving overexpression of chaperones are known; they are the heat-shock response and the EPR stress response.
Although first observed as a result of hyperthermia, the heat-shock response is an adaptive mechanism also triggered by various pathologic conditions (eg, trauma, tissue ischemia) and by virtually all reactive chemicals and/or their metabolites (eg, electrophiles, oxidants, lipid peroxidation products, metal ions, arsenite) that denature proteins. Thus, it takes place simultaneously with the electrophile response discussed above. This reaction, however, is governed by heat-shock transcription factors (HSF), mainly HSF1, which transactivate genes that encode Hsp through heat-shock response elements (HSE) within the promoter region. HSF1, like Nrf2, normally resides in the cytoplasm, where it associates with Hsp90, Hsp70, and Hsp40. On heat- or chemical-induced protein damage, these Hsps are purportedly sequestered by damaged proteins, allowing HSF1 to be released. HSF1 then migrates into the nucleus, trimerizes, undergoes phosphorylation, and stimulates the transcription of Hsp genes. As described in the section “Repair of Proteins,” the chaperones Hsp90 and Hsp70, together with co-chaperone proteins, are especially important in maintaining the integrity of hundreds of proteins (Pratt et al., 2010). The client proteins include not only those carrying out housekeeping functions but also those involved in signaling and apoptosis. Therefore, induction of Hsps has pleiotropic effects besides increased protection from cytotoxity. However, when the proteotoxic stress induces excessive protein unfolding, Hsp70 and Hsp40 recruit the ubiquitin system (Fig. 3-23) to tag the aberrant proteins for proteasomal degradation (Bedford et al., 2011).
Induction of Chaperones Repairing Misfolded Proteins—The Endoplasmic Reticulum Stress and the Unfolded Protein Response
All proteins that are destined for export or insertion into cellular membranes pass through the EPR. In this Ca2+-rich oxidative environment, they may be subjected to N-glycosylation at asparagine residues by oligosaccharyltransferase, formation of disulfide bonds by protein disulfide isomerases, and folding assisted by EPR-resident chaperones, such as the glucose-regulated proteins Grp78 (also called binding immunoglobulin protein [BiP]) and Grp94 as well as the Ca2+-binding proteins calreticulin and calnexin. Damage of proteins being processed in the EPR by reactive metabolites produced in situ by CYP enzymes (eg, the quinoneimine metabolite of acetaminophen, Cl3C• and Cl3COO• radicals formed from CCl4, and ROS generated by CYP2E1 during oxidation of ethanol) and/or depletion of Ca2+ in the EPR lumen (eg, by inactivation of the EPR Ca2+-ATPase by reactive metabolites) causes accumulation of unfolded or misfolded proteins, a condition known as EPR stress (Cribb et al., 2005; Nagy et al., 2007; Malhi and Kaufman, 2011). Protein folding disorder can also be experimentally induced by some natural compounds that disturb the homeostasis of EPR at specific steps, such as thapsigargin (which blocks SERCA-mediated Ca2+ uptake into EPR; see Table 3-7), tunicamycin (which inhibits N-glycosylation of proteins), castanospermine (which interferes with maturation of nascent glycoproteins by deglucosylation), and brefeldin A (which inhibits transport of proteins from EPR to Golgi).
When the load of unfolded or misfolded proteins in the EPR exceeds the capacity of EPR-resident chaperones, these proteins trigger a complex adaptive response, called unfolded protein response (UPR). This may constrain exacerbation of the disorder (1) by attenuation of mRNA translation to decrease the functional load on the EPR, (2) by increased transcription of EPR chaperones to boost the folding capacity, (3) by initiating the translocation of aberrant proteins via translocon peptide channels from the EPR into the cytosol for proteasomal degradation, and (4) by eliminating the affected cell via apoptosis, if the EPR stress is sustained or massive. The eukaryotic translation initiation factor 2α (eIF2α) and TFs X-box binding protein 1 (XBP1), ATF6, ATF4, and C/EBP homologous protein (CHOP) (or Gadd153) are the major executioners of this adaptive response.
Three transmembrane EPR proteins sense the overload of the EPR lumen with misfolded proteins and transduce this signal to initiate the UPR. These are (1) PERK, (2) inositol-requiring enzyme 1α (IRE1α), a protein with kinase and endoribonuclease (RNase) activities, and (3) the precursor form ATF6 (activating TF 6). Normally, these sensor proteins are turned off by being associated at their luminal domain with the EPR-resident soluble chaperone BiP (or Grp78). However, when EPR stress occurs, the unfolded proteins that accumulate in the lumen of the EPR sequester BiP away from these sensors, thereby turning them on to signal for the UPR. PERK stripped off BiP is activated by dimerization and trans-autophosphorylation. With its kinase activity thus raised, PERK catalyzes the phosphorylation of the eIF2α, with the phosphorylated eIF2α causing a global translation attenuation and decreased synthesis of most (but not all) proteins. The second EPR stress sensor IRE1α also undergoes dimerization and trans-autophosphorylation on dissociation from BiP. These changes boost the RNase activity of IRE1α, enabling it to cleave and process a mRNA into one that translates into XBP1, an active TF, another executioner of UPR. The third sensor ATF6 resides as a 90 kDa precursor protein in the EPR with its Golgi localization sequences masked by BiP. On dissociation from BiP, ATF6 translocates to the Golgi, undergoes a so-called regulated intramembrane proteolysis (by site-1 and site-2 proteases), yielding a 50 kDa fragment, the active TF ATF6. XBP1 and ATF6 bind alone or as heterodimers to cognate DNA sequences, such as EPR response element (ERSE) and unfolded protein response element (UPRE), and activate the transcription of genes coding for chaperone proteins that promote protein folding (eg, Grp78, Grp94, Mdg1/Erdj4, Herp) and for proteins that assist in degradation of misfolded proteins (eg, EPR degradation-enhancing α-mannosidase-like protein [EDEM]).
On excessive EPR stress, the three activated sensors, that is, PERK, ATF6, and IRE1α, may act in concert to initiate apoptosis (Malhi and Kaufman, 2011). While PERK-catalyzed phosphorylation of eIF2α decreases the synthesis of most proteins (as stated above), it increases the synthesis of some, including that of ATF4. This TF (as well as ATF6) promotes the expression of CHOP, yet another TF. CHOP in turn can transcriptionally induce the expression of the proapoptotic Bim protein and the cell surface death receptor TRAIL-2, while inhibiting the expression of the antiapoptotic Bcl-2 protein (see Fig. 3-19). The activated IRE1α may initiate cell death via recruiting the adaptor protein TNF receptor-associated factor-2 (TRAF2), with subsequent activation of apoptosis signal-regulating kinase 1 (ASK1) and JNK. By phosphorylation, JNK can activate the proapoptotic Bim and inactivate the antiapoptotic Bcl-2 proteins (Malhi and Kaufman, 2011). The IRE1α–TRAF2 complex may also recruit and activate procaspase 12. Alternatively, Ca2+ released from the EPR can activate calpains, which cleave the EPR-associated procaspase 12 into active caspase 12. The latter then engages the caspase cascade (see Fig. 3-19) by activating effector caspases therein to sacrifice the cell by apoptosis. Several toxicants have been shown to induce EPR stress in experimental animals and in isolated cells by demonstrating expression of some elements of the UPR on toxicant exposure (Cribb et al., 2005; Nagy et al., 2007; Malhi and Kaufman, 2011).
Induction of Enzymes Repairing DNA—The DNA Damage Response (DDR)
DDR is initiated by detection of the DNA damage. Double-stranded DNA breaks are recognized by the MRN complex (the trimer of Mre11, Rad50, and Nbs1 proteins) or the Ku protein (the dimer of Ku70 and Ku80; see the section “Repair of DNA”) that bind and activate protein kinases ataxia telangiectasia mutated (ATM) and DNA-PKcs, respectively. A third kinase, ataxia telangiectasia and Rad3-related (ATR), is activated in response to persistent single-stranded DNA coated with replication protein A (RPA), an intermediate during nucleotide excision or recombination DNA repair processes. Directly or through checkpoint kinases (Chk1, Chk2), these kinases phosphorylate p53 (Christmann et al., 2003; McGowan and Russell, 2004), a protein that can play the role of a TF regulating gene expression and the role of an associate protein affecting the function of its interacting protein partner. Normally, p53 is kept inactive and at low levels by its binding protein mdm2 (see Fig. 3-33), which ubiquitinates p53, facilitating its proteasomal degradation. On phosphorylation, p53 escapes from mdm2, allowing its activation and stabilization. Indeed, the levels of p53 protein in cells increase dramatically in response to DNA damage caused by UV or gamma irradiation or genotoxic chemicals. p53 then facilitates DNA repair by a number of mechanisms. For example, mainly by transcriptionally upregulating the cyclin-dependent kinase inhibitor protein p21, p53 arrests cells in G1 phase of the cell cycle (see Fig. 3-33), allowing more time for DNA repair. As a TF, p53 also increases expression of proteins directly involved in DNA repair (Harms et al., 2004). Such proteins include (a) growth arrest and DNA damage inducible (gadd45), which interacts with histones and facilitates access of proteins (eg, topoisomerase) to DNA, (b) XPE and XPC, members of the xeroderma pigmentosum group of proteins important in UV-induced DNA damage recognition before nucleotide excision repair, (c) MSH2 operating in mismatch repair, (d) PCNA that holds DNA polymerase-δ to DNA during DNA replication as well as reparative DNA synthesis in the excision and the recombination repair processes, and (e) a form of ribonucleotide reductase that provides deoxyribonucleotides for sealing DNA gaps. As a partner protein, p53 supports the function of several proteins of the nucleotide excision machinery (eg, TFIH, XPB, XPD). (Other roles p53 plays in apoptosis and in carcinogenesis as a tumor suppressor protein are illustrated in Figs. 3-19 and 3-33 and discussed elsewhere in this chapter.) The DNA-damage-activated protein kinases (ATM, ATR, and DNA-PK) also phosphorylate histone H2AX adjacent to the damage, which then becomes ubiquitinated. These markings in turn recruit repair proteins (including BRCA1) and induce chromatin relaxation for better access of repair enzymes to the lesion. The DDR-initiating kinases have numerous other protein substrates, including other kinases; therefore, DDR may extend to diverse cellular functions. It is important to realize that while both the complex DDR and distinct DNA repair mechanisms (eg, direct repair by MGMT) protect normal cells that suffered DNA damage from transformation into cancer cells (a process to be discussed below), these mechanisms, if remain operative in cancer cells, can protect them from mortal DNA damage inflicted by radiotherapy or chemotherapy, and thereby DDR and DNA repair can contribute to resistance of tumor cells to anticancer treatments.
Adaptive Increase in Tissue Repair—A Proliferative Response
Many toxicants potentially injurious to cells, for example, electrophiles, oxidants, and those inducing oxidative stress, can initiate mitogenic signaling as a prelude to tissue repair via cell replacement. It appears that the need for mitogenesis is sensed by PTP (eg, PTP1B) and the lipid phosphatase PTEN, which contains reactive cysteine thiols at their active site (Rhee et al., 2005). These phosphatases serve as brakes on the growth factor receptor–initiated mitogenic signaling, as PTPs dephosphorylate (and inactivate) the receptors themselves (eg, EGFR, PDGFR, IGFR) as well as some protein kinases (eg, Src and JAK), whereas PTEN dephosphorylates PIP3, an important second messenger in the PI3K–Akt–IKK–NF-κB pathway (Fig. 3-12). Electrophiles covalently bind to essential cysteine-SH groups in these phosphatases. HOOH can oxidize the critical -SH group in PTEN to an intramolecular disulfide, whereas it oxidizes the -SH group of PTP1B, through sulfenic acid (-S-OH), to a 5-membered cyclic sulfenyl amide species in which the sulfur atom is covalently linked to the nitrogen of the neighboring serine (Rhee et al., 2005). Inactivation of PTPs and PTEN, which decrease the proliferative signal transduction, amplifies intracellular signaling for mitosis and survival.
It has been known for some time that oxidative stress, if not severe, activates the TF NF-κB (Dalton et al., 1999). For example, silica, which can produce ROS on its surface, activates NF-κB as well as PI3K when added to various cells (Castranova, 2004). In light of new information discussed above, NF-κB activation is now attributed to the fact that this TF is situated downstream of growth factor receptors (which are negatively controlled by the ROS-sensitive PTP) and PIP3 (which is eliminated by ROS-sensitive PTEN) (Fig. 3-12). Furthermore, NF-κB is at the focal point of proliferative and prolife signaling, as it transactivates genes producing cell cycle accelerators (eg, cyclin D1 and c-Myc) and apoptosis inhibitors (eg, antiapoptotic Bcl proteins and the caspase IAPs) (Karin, 2006). In addition, NF-κB also transactivates the genes of ferritin, GST, SOD1, HO-1, a proteasome subunit, and gadd45, facilitating detoxication and molecular repair. All these roles of NF-κB explain its involvement in tolerance to chemically induced tissue injury, resistance against cholestatic liver injury caused by bile acids, adaptation to ionizing radiation, as well as the phenomenon termed preconditioning. This is a tolerance to ischemic tissue injury (eg, myocardial infarction), a tolerance induced by temporarily enhanced ROS formation evoked by hyperoxia or brief periods of ischemia–reperfusion.
In addition to signaling for cell replacement in damaged tissue—in which NF-κB plays a leading role—the growing cells need to boost protein synthesis. This is done under the control of the protein kinase mTOR. As shown in Fig. 3-29, mTOR activation results from signaling through both pathways coupled to growth factor receptors, that is, the MAPK pathway leading to phosphorylation of the MAPK isoform Erk and the PI3K pathway leading to phosphorylation of Akt (see Fig. 3-12). Importantly, these pathways are subject to activation in response to oxidant or electrophile exposure as they are controlled by PTPs and PTEN (Fig. 3-12). Erk and Akt protein kinases activate mTOR through a complex mechanism (Fig. 3-29), and mTOR in turn phosphorylates and regulates effectors of protein synthesis, such as the translation repressor protein 4EBP1 and the protein kinase S6K, which modifies ribosomes increasing their translational efficiency (Shaw and Cantley, 2006). As described in the sections “Adaptation to Hypoxia—The Hypoxia Response” and “Adaptation to Energy Depletion—The Energy Stress Response,” mTOR signaling is switched off in the cell to save energy as a measure to adapt to the energy shortage caused by hypoxia or toxic impairment of ATP synthesis.
Modulation of protein synthesis and autophagy by signaling through the mTOR kinase pathway as a means for cellular adaptation— increased signaling via mTOR promotes protein synthesis at translational level and suppresses autophagy, thereby permitting cell growth and proliferation, whereas attenuated signaling via mTOR permits energy saving (by halting protein synthesis) and a desperate attempt for energy generation (by starting autophagy) when hypoxia, nutrient shortage, or toxic injury causes energy deficit. Growth factor receptors signaling via either the MAPK pathway (leading to phosphorylation of Erk) or the PI3K pathway (leading to phosphorylation of Akt) (see Fig. 3-12) activate the serine/threonine kinase, mammalian target of rapamycin (mTOR; rapamycin is also called sirolimus) by an indirect mechanism. The protein kinases Erk and Akt catalyze inactivating phosphorylation of TSC2, a member of TSC1/2 complex. TSC2 is a GTPase-activating protein whose substrate is Rheb, a small G protein (Ras homologue), which is active in the GTP-bound form and inactive in the GDP-bound form. With its GTPase activating capacity blocked by Erk or Akt, TSC2 cannot convert Rheb-GTP into inactive Rheb-GDP, and thus Rheb-GTP activates mTOR. In turn, mTOR phosphorylates 2 substrates that are necessary to initiate translation of mRNA into proteins, that is: (1) eukaryotic initiation factor 4E-binding protein-1 (4EBP1), which thus releases the translational initiation factor eIF4E, and (2) ribosomal protein S6 kinase-1 (S6K1), which phosphorylates ribosomal protein S6, thereby increasing translational efficiency of mRNAs that encode ribosomal proteins (“ribosomal biogenesis”). Simultaneously, mTOR phosphorylates and inactivates unc-51-like kinase-1 (ULK-1) and Atg13, thereby preventing nonselective or bulk autophagy that is dependent on formation of the stable complex of ULK-1, Atg13, and focal adhesion kinase family-interacting protein of 200 kDa (FIP200).
Protein synthesis for cell growth and proliferation is halted, whereas bulk autophagy is facilitated in times of energy deficit resulting from hypoxia, nutrient shortage, or toxic impairment of ATP production. Then AMP levels increase and AMP binds to the AMP-activated protein kinase (AMPK), facilitating its phosphorylation by protein kinase LKB1. Activated AMPK phosphorylates TSC2 (at a site different from that targeted by Akt and Erk), thereby increasing the GTPase-activating capacity of TSC2. This in turn switches Rheb off, making mTOR inactive. This suspends mRNA translation and starts autophagy. Hypoxia can initiate this process in a more specific way as well, that is, via stabilization of the hypoxia-inducible factor (HIF). This transcription factor induces the synthesis of REDD1 (by a mechanism that is not completely understood), which activates TSC2. See text for further details on cellular responses to hypoxia and energy deficit.
Adaptation by Compensating Dysfunction
Dysfunctions caused by toxicants or drug overdose manifested at the level of organism (eg, hypoxia), organ system (eg, hypotension and hypertension), or organ (eg, renal tubular dysfunction) may evoke compensatory mechanisms.
Adaptation to Hypoxia—The Hypoxia Response
When O2 delivery is impaired and hypoxia persists for more than a few minutes, a response involving gene expression alteration is initiated. This reaction is mainly orchestrated by hypoxia-inducible factor-1α (HIF-1α), a ubiquitous TF whose activity and cellular abundance is greatly increased in response to hypoxia (Maxwell and Salnikow, 2004; Pouyssegur et al., 2006). HIF-1α is maintained at very low intracellular levels because of continuous hydroxylation of its 2 proline residues by HIF-prolyl hydroxylases. This permits a ubiquitin ligase subunit (called von Hippel Lindau protein [VHL]) to capture HIF-1α and initiate its destruction by proteasomal degradation. Indeed, HIF-1α is one of the shortest lived proteins with a half-life of less than 5 minutes. In addition, HIF-1α is kept transcriptionally inactive by hydroxylation at one of its asparagine residues by HIF-asparagine hydroxylases, which prevents interaction of HIF-1α with transcriptional coactivators, such as p300 and CREB-binding protein (CBP). These two types of HIF hydroxylases are O2 sensors: they use O2 as a substrate to carry out proline/asparagine hydroxylations with concomitant oxidative decarboxylation of 2-oxoglutarate to succinate, with the KM of O2 being close to the ambient O2 concentration. As the O2 concentration falls, decreases in the hydroxylation rate of HIF-1α as well as its VHL-mediated ubiquitination and proteasomal degradation occur, and this increases its abundance and transcriptional activity. HIF hydroxylases belong to the Fe2+ and ascorbate-dependent dioxygenases (the largest group of nonheme oxidases); therefore, not only hypoxia but also Fe2+ deficiency impairs their activity. The latter feature explains why iron chelators (eg, deferoxamine) or Fe2+-mimicking metal ions (eg, Co2+ and Ni2+) also induce and activate HIF-1α (Maxwell and Salnikow, 2004). When induced after hypoxic conditions, HIF-1α dimerizes with HIF-1β (also called Arnt, which, coincidentally, is the dimerization partner for the Ah receptor as well). The HIF complex transactivates a vast array of genes with hypoxia response element (HRE) in their promoter. Many of the gene products assist in acclimatization to hypoxia (Pouyssegur et al., 2006). These include (1) erythropoietin (EPO) that is produced largely in kidney and activates erythropoiesis in bone marrow, (2) proteins involved in iron homeostasis (eg, transferrin, TFR, ceruloplasmin, and heme oxygenase) that may increase availability of iron for erythropoiesis, (3) vascular endothelial growth factor (VEGF) and angiopoietin-2 that stimulate blood vessel growth (ie, angiogenesis), (4) proteins facilitating anaerobic ATP synthesis from glucose (ie, glycolysis), such as the glucose transporter GLUT1 and some glycolytic enzymes, (5) proteins that correct acidosis caused by glycolytic overproduction of lactate (eg, a monocarboxylate transporter and a Na+/H+ exchanger for export of lactate and H+, respectively), (6) the REDD1 signal transducer protein that initiates a complex signaling pathway that leads to suspension of the ATP-consuming protein synthesis via inactivation of the protein kinase mTOR (Fig. 3-29), and (7) many other proteins, such as those that promote ECM remodeling (eg, matrix metalloproteinase-2) and cell migration (perhaps to facilitate access of the cells to the blood vessel), as well as BNIP3, a proapoptotic MOM protein that is also involved in autophagy of mitochondria in reticulocytes (perhaps to induce apoptosis of cells subjected to extreme hypoxia and to facilitate the terminal differentiation of red blood cells). Experiments on mice kept in low O2 environment (hypoxic preconditioning) demonstrated that HIF-1α became stabilized in the retina of these animals, hypoxia-responsive genes (EPO, VEGF) were induced, and the retina became resistant to light toxicity (Grimm et al., 2005). Adaptation to hypoxia also occurs, for example, in response to high-altitude hypoxia, chronic cardiorespiratory dysfunction, and ischemic preconditioning, along with other adaptive responses discussed above. The hypoxia response is also expected to develop as a result of toxicities causing hypoxia acutely or subacutely (eg, respiratory muscle weakness after organophosphate intoxication, diquat-induced pulmonary injury) or as a delayed sequel (eg, respiratory surface restriction in hard metal disease).
Adaptation to Energy Depletion—The Energy Stress Response
Cells try to maintain their adenosine nucleotide pool in triphosphorylated, energized state, which is in the form of ATP. When the rephosphorylation rate of AMP and ADP to ATP does not keep up with the rate of ATP use, because, for example, oxidative phosphorylation is impaired or ATP use for muscle contraction or ion pumping is excessive, the ratio of AMP to ATP increases. A cellular mechanism has evolved to sense this menacing energy deficit and, in order to compensate, boosts ATP production and curtails ATP consumption (Hardie et al., 2006). The sensor is a ubiquitous heterotrimeric intracellular protein complex called AMP-activated protein kinase (AMPK). AMP strongly activates AMPK allosterically, and also by making it susceptible for phosphorylation by protein kinase LKB1 (or by the calmodulin-dependent protein kinase kinase [CaMKK] in neurons). The phosphorylated, and thus activated AMPK, targets 2 sets of proteins. One set includes those whose activation facilitates ATP production from catabolism of glucose and fatty acids as well as by promoting the biogenesis of mitochondria. For example, AMPK activation increases (a) glucose uptake (via recruiting to the cell membrane or activating glucose transporters GLUT4 and GLUT1), (b) glycolysis (via phosphorylation and activation of 6-phosphofructo-2-kinase [PFK-2] whose product, fructose-2,6-bisphosphate, is a glycolytic activator), and (c) fatty acid oxidation in mitochondria (via phosphorylation and inactivation of acetyl-CoA-carboxylase, whose product, malonyl-CoA, is an allosteric inhibitor of carnitine palmitoyltransferase-1 [CPT-1], which mediates uptake of long-chain fatty acid CoA esters into mitochondria). Another set of proteins, which are inactivated by AMPK (directly or indirectly), includes those that are involved in biosynthetic ATP-consuming reactions. Thus, AMPK inhibits (a) glycogen synthesis via phosphorylation and inactivation of glycogen synthase, (b) lipid synthesis by phosphorylating and inactivating acetyl-CoA-carboxylase, whose product, malonyl-CoA, is an essential substrate for fatty acid synthesis, (c) cholesterol synthesis by phosphorylating and inactivating HMG-CoA reductase, (d) glucose synthesis via inactivating phosphorylation of a transcriptional coactivator, TORC2, which then decreases expression of key gluconeogenetic enzymes, such as phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase, and (e) protein synthesis, and thus cell growth, by inhibiting the protein kinase mTOR (Fig. 3-29). AMPK-mediated modulation of cellular energy supply and consumption involves mainly kinase reactions rather than new protein synthesis. Therefore, this adaptation is a rapid process. It can be a response to any harmful condition that compromises oxidative phosphorylation, such as hypoxia, hypoglycemia (especially in neurons), and chemically induced mitochondrial toxicity. For example, cells exposed to arsenite exhibit rapid increases in the AMP/ATP ratio and AMPK activity, with concomitant declines in HMG-CoA reductase activity as well as fatty acid and cholesterol synthesis (Corton et al., 1994).
Apart from the above-described rapid AMPK-dependent reprogramming of the cell’s intermediary metabolism from energy-consuming operation to the energy-producing mode, there is an ultimate means for the cell to generate fuel for ATP production in case of nutrient shortage. This slower process involves consumption of the cell’s own constituents (lipid droplets, glycogen particles, proteins, and organelles) by nonselective or bulk autophagy (Rabinowitz and White, 2010) involving their lysosomal hydrolysis in order to gain fuel (amino acids, fatty acids, nucleosides, and carbohydrates). An essential actor in this form of autophagy is a protein complex of unc-51-like kinase-1 (ULK-1), autophagy-related protein 13 (Atg13), and focal adhesion kinase family-interacting protein of 200 kDa (FIP200), which is an initiator of autophagosome formation (Fig. 3-29). When nutrients are abundant, the ULK-1–Atg13–FIP200 complex associates with mTOR complex 1 (mTORC1), with both ULK-1 and mTOR kinase being bound by Raptor, a scaffolding protein in mTORC1. Then mTOR phosphorylates and inactivates ULK-1 and Atg13, thus inhibiting autophagy. In nutrient shortage, signaling pathways, such as those activated by AMP and hypoxia (see Fig. 3-29), inactivate mTOR kinase, thereby forcing mTORC1 to release the complex of ULK-1–Atg13–FIP200, which in turn initiates autophagosome nucleation (Hosokawa et al., 2009).
In summary, mTOR kinase is a key regulator of the cell’s energy homeostasis, permitting or limiting cell growth according to availability of resources, the use of which can also be controlled by mTOR through switching autophagy on and off. In the presence of abundant nutrients and growth factors, mTOR is activated, thereby promoting cell growth and metabolic activity while suppressing nonselective autophagy. In nutrient deprivation or stress (eg, energetic failure, hypoxia), signaling pathways inactivate mTOR kinase activity. This both suppresses cell growth to reduce energy demand and induces autophagy to enable stress adaptation and survival (Fig. 3-29).
After surveying the major cellular adaptation mechanisms to toxicants, it is easy to recognize that one noxious effect may initiate several adaptive responses. For example, cells exposed to a hypoxic environment can rapidly respond with both AMPK-mediated program of energy stabilization and HIF-1α-directed adaptation to oxygen shortage. Theoretically, an electrophile toxicant that can bind covalently to cellular macromolecules and can also generate oxidative stress, such as a redox cycling quinone, would be expected to induce a number of adaptive processes, including the electrophile response, the heat-shock response, the EPR stress response, the DDR, and the proliferative response, and if it compromises ATP synthesis as well, it even induces the energy stress response.
Adaptation by Neurohumoral Mechanisms
There are numerous adaptive responses to dysfunctions of organs or organ systems that are mediated by humoral or neuronal signals between cells located in the same or different organs. For example, the rapid hyperventilation evoked by acute hypoxia or HCN inhalation is mediated by a neural reflex initiated by glomus cells in the carotid body. These chemosensitive cells generate a Ca2+ signal via the above-described AMP sensor, AMPK, which becomes activated by hypoxia or CN− through impairment of oxidative phosphorylation in these cells, causing rise in the AMP/ATP ratio (Evans et al., 2005). Besides CN−, mitochondrial electron transport inhibitors (eg, rotenone, antimycin A, myxothiazole), uncouplers (eg, 2,4-dinitrophenol), or ATP-synthase inhibitors (eg, oligomycin) (see Table 3-6 and Fig. 3-16) as well as AMPK activators mimic the response to hypoxia in these cells. There are numerous other neurohumoral adaptive mechanisms in the body, such as the sympathetic reflex as well as activation of the renin–angiotensin–aldosterone system in response to hypotension, and the feedback systems between endocrine glands and the hypothalamus–hypophysis, which correct abnormal hormone levels. For information on these and other mechanisms, the reader is referred to textbooks of physiology.
When Repair and Adaptation Fail
Although repair mechanisms operate at molecular, cellular, and tissue levels, for various reasons they often fail to provide protection against injury. First, the fidelity of the repair mechanisms is not absolute (eg, the nonhomologous end joining that restores DNA DSB), making it possible for some lesions to be overlooked or erroneously fixed. However, repair fails most typically when the damage overwhelms the repair mechanisms, as when protein thiols are oxidized faster than they can be reduced. In other instances, the capacity of repair may become exhausted when necessary enzymes or cofactors are consumed. For example, alkylation of DNA may lead to consumption of MGMT (a self-sacrificing enzyme), lipid peroxidation can deplete α-tocopherol, and overproduction of oxidized or otherwise damaged proteins can exhaust the pool of ubiquitin. Sometimes the toxicant-induced injury adversely affects the repair process itself. For example, ethanol generates ROS via CYP2E1 that impairs the proteasomal removal of damaged proteins. Autophagic removal of cell constituents may be compromised by lysosomotropic drugs (lipophilic amines) that increase the pH in these organelles, thus decreasing the activity of lysosomal hydrolases. This may underlie the mechanism of chloroquine-induced myopathy in rats. Thiol-reactive chemicals may inactivate ubiquitin-activating enzymes (E1), ubiquitin-conjugating enzymes (E2), and the HECT domain-containing ubiquitin ligases (E3), each of which contains a catalytically active cysteine, as well as the lysosomal cysteine proteases (eg, cathepsins B, H, and L), thereby compromising the clearance of damaged proteins by both the UPS and autophagy. Indeed, diminished proteolysis occurs in hepatocytes exposed to toxic concentrations of acetaminophen. These repair mechanisms are especially important in neurons, as genetic deletion of proteins mediating UPS and autophagy causes accumulation of intraneuronal inclusions, neuronal loss, or neurodegeneration. In fact, human neurodegenerative diseases can be directly attributed to some dysfunction of the UPS and/or autophagy (Bedford et al., 2011). The finding that individual overexpression of some E2 and E3 enzymes confers resistance to methylmercury toxicity in yeast (Hwang et al., 2006) lends some support to the speculation that impairment of the UPS and/or autophagy might contribute to the neurotoxicity of methylmercury. After exposure to necrogenic chemicals, mitosis of surviving cells may be blocked and restoration of the tissue becomes impossible (Mehendale, 2005). Finally, some types of toxic injuries cannot be repaired effectively, as occurs when xenobiotics are covalently bound to proteins or when protein carbonyls are formed. Thus, toxicity is manifested when repair of the initial injury fails because the repair mechanisms become overwhelmed, exhausted, or impaired or are genuinely inefficient.
It is also possible that repair contributes to toxicity. This may occur in a passive manner, for example, if excessive amounts of NAD+ are cleaved by PARP when this enzyme assists in repairing broken DNA strands, or when too much NAD(P)H is consumed for the repair of oxidized proteins and endogenous reductants. Either event can compromise oxidative phosphorylation, which is also dependent on the supply of reduced cofactors (see Fig. 3-16), thus causing or aggravating ATP depletion that contributes to cell injury. Excision repair of DNA and reacylation of lipids also contribute to cellular deenergization and injury by consuming significant amounts of ATP. However, repair also may play an active role in toxicity. This is observed after chronic tissue injury, when the repair process goes astray and leads to uncontrolled proliferation instead of tissue remodeling. Such proliferation of cells may yield neoplasia, whereas overproduction of ECM results in fibrosis. A specific case is when the cellular repair processes that normally degrade damaged proteins, such as the UPS and the autophagy pathway, become inappropriately stimulated by adverse signaling, causing degradation of ordinary intracellular proteins. This occurs in muscle wasting induced by glucocorticoids, such as cortisol and dexamethasone. This condition appears to be secondary to suppression of the growth-promoting IGF-1–PI3K–Akt signaling that leaves FoxO TFs unchecked, which in turn increase the expression of muscle-specific ubiquitin ligases (MAFbx, also called atrogin-1, and MuRF1) as well as several components of the autophagy–lysosome pathway (Fig. 3-30).
A model of glucocorticoid-induced muscle wasting: by attenuating the IGF-1–PI3K–Akt growth signaling pathway, glucocorticoids disinhibit FoxO transcription factors that upregulate the degradation of muscle proteins by both the ubiquitin–proteasome system and the autophagy–lysosome pathway. The trophic condition of the skeletal muscle is positively regulated by IGF-1, a muscle anabolic growth factor. IGF-1 acts in autocrine and paracrine manner on its tyrosine kinase receptor to trigger signaling through insulin receptor substrate 1 (IRS-1), phosphatidylinositol 3-kinase (PI3K), and the lipid mediator phosphatidylinositol (3,4,5)-trisphosphate (PIP3), causing activation of the serine/threonine protein kinase Akt (see also Fig. 3-12). As shown in Fig. 3-29, Akt activation leads to activation of mTOR, which boosts protein translation and halts bulk autophagy. In addition, the active Akt also phosphorylates and inactivates Forkhead box O transcription factors (eg, FoxO1 and FoxO3), thereby preventing them from translocating into the nucleus and activating the expression of genes whose products would mediate degradation of muscle proteins. These mechanisms collectively contribute to muscle hypertrophy.
Glucocorticoids (GC) induce muscle atrophy mainly by suppressing the IGF-1–PI3K–Akt signaling pathway in the skeletal muscle. Under this condition the phosphorylation of FoxO by Akt ceases, and then the nonphosphorylated FoxO translocates into the nucleus and activates the transcription of its target genes. Among these are the genes coding for muscle-specific E3 ubiquitin ligases, that is, muscle atrophy F-box (MAFbx, also called atrogin-1) and muscle ring finger 1 (MuRF1). By ubiquitinating their substrates, which include structural proteins (eg, myosin heavy chain) and regulatory proteins (eg, the transcription factor MyoD and the translation initiation factor eIF3f), these ubiquitin ligases promote their proteasomal degradation. FoxO also turns on the other degradative machinery, the autophagy–lysosome system (Mehrpour et al., 2010), by increasing the expression of a number of its components, such as LC3, Beclin 1, ULK-1, Vps34, BNIP3, and cathepsin L (see more details in the sections “Cellular Repair” and “Adaptation to Energy Depletion—The Energy Stress Response”).
GC suppress the IGF-1–PI3K–Akt signaling by multiple mechanisms. They decrease production of IGF-1 by the muscle, downregulate the expression of IRS-1, and inhibit PI3K activity (apparently by direct interaction of the activated GC receptor with the regulatory subunit of PI3K). In addition, GC upregulate the expression of myostatin in the muscle probably by downregulating the expression of miRNA-27a and b, which target myostatin mRNA (see elsewhere). Myostatin is a TGF-β family member protein, which is a negative regulator of muscle mass. It acts by both endocrine and paracrine fashion on its receptor (similar to item 8 in Fig. 3-12) and activates Smad2, which inhibits Akt (Glass, 2010). Other mechanisms may also contribute to GC-induced muscle wasting (Hasselgren et al., 2010).
Although adaptation mechanisms, such as the Nrf2-mediated electrophile response and the NF-κB-induced proliferative reaction, boost the capacity of the organism to withstand toxicant exposure and damage, excessive exposure can overwhelm these protective responses. Moreover, toxicants may impair the adaptive process. For example, moderate oxidative stress activates NF-κB, AP-1, and Nrf2 to initiate adaptive protection. However, extensive oxidant exposure aborts this program because it leads to oxidation of thiol groups in the DNA-binding domain of these TFs (Hansen et al., 2006). Similarly, Hg2+ can incapacitate NF-κB, thus inhibiting the prolife program activated by this TF. This promotes Hg2+-induced renal tubular cell injury (Dieguez-Acuna et al., 2004).
Some adaptive mechanisms may be harmful under extreme conditions. For example, acute tubular injury, which impairs tubular reabsorption and causes polyuria, triggers a tubuloglomerular feedback mechanism that reduces glomerular blood flow and filtration. Ultimately, this may precipitate anuric renal failure. It is possible that an adaptive mechanism that is beneficial in the short term may become harmful when forced to operate for a prolonged period of time. Bulk autophagy likely contributes to lethal wasting syndrome induced by TCDD in experimental animals. Chronic inflammation, tissue injury, or cancer may lead to iron deficiency and anemia because IL-6 (the acute-phase-triggering cytokine overproduced in these conditions) upregulates hepcidin secretion from the liver, which in turn reduces intestinal iron absorption. As discussed earlier, NF-κB activation is indispensable for repair via proliferation of the acutely injured tissue. However, NF-κB also targets cytokine genes, and the cytokines (eg, TNF, IL-1β) in turn activate NF-κB through their receptors (see Fig. 3-12). This vicious cycle may lead to chronic inflammation and cancer when repetitive tissue injury maintains NF-κB signaling (Karin, 2006). This occurs after occupational exposure to silica (Castranova, 2004). Sustained activation of HIF-1α in tumors facilitates invasiveness, in part by increasing VEGF expression and angiogenesis. In the kidney, HIF-1α may be involved in fibrogenesis, as it targets critical genes, such as tissue inhibitor of metalloproteinase-1 (TIMP-1).
Toxicity Resulting from Inappropriate Repair and Adaptation
Like repair, dysrepair occurs at the molecular, cellular, and tissue levels. Some toxicities involve dysrepair at an isolated level. For example, hypoxemia develops after exposure to methemoglobin-forming chemicals if the amount of methemoglobin produced overwhelms the capacity of methemoglobin reductase. Because this repair enzyme is deficient at early ages, neonates are especially sensitive to chemicals that cause methemoglobinemia. Formation of cataracts purportedly involves inefficiency or impairment of lenticular repair enzymes, such as the endopeptidases and exopeptidases, which normally reduce oxidized crystalline and hydrolyze damaged proteins to their constituent amino acids. Dysrepair also is thought to contribute to the formation of Heinz bodies, which are protein aggregates formed in oxidatively stressed and aged red blood cells. Defective proteolytic degradation of the immunogenic trifluoroacetylated proteins may make halothane-anesthetized patients victims of halothane hepatitis.
Several types of toxicity involve failed and/or derailed repairs at different levels before they become apparent. This is true for the most severe toxic injuries, such as tissue necrosis, fibrosis, and chemical carcinogenesis.
As discussed above, several mechanisms may lead to cell death. Most or all involve molecular damage that is potentially reversible by repair mechanisms. If repair mechanisms operate effectively, they may prevent cell injury or at least retard its progression. For example, prooxidant toxicants cause no lipid fragmentation in microsomal membranes until α-tocopherol is depleted in those membranes. Membrane damage ensues when this endogenous antioxidant, which can repair lipids containing peroxyl radical groups (Fig. 3-24), becomes unavailable (Scheschonka et al., 1990). This suggests that cell injury progresses toward cell necrosis if molecular repair mechanisms are inefficient or the molecular damage is not readily reversible.
Progression of cell injury to tissue necrosis can be intercepted by 2 repair mechanisms working in concert: apoptosis and cell proliferation. As discussed above, injured cells can initiate apoptosis, which counteracts the progression of the toxic injury. Apoptosis does this by preventing necrosis of injured cells and the consequent inflammatory response, which may cause injury by releasing cytotoxic mediators. Indeed, the activation of Kupffer cells, the source of such mediators in the liver, by the administration of bacterial LPS (endotoxin) greatly aggravates the hepatotoxicity of galactosamine. In contrast, when the Kupffer cells are selectively eliminated by pretreatment of rats with gadolinium chloride, the necrotic effect of carbon tetrachloride is markedly alleviated (Edwards et al., 1993). Blockade of Kupffer cell function with glycine (via the inhibitory glycine receptor; see item 4 in Fig. 3-15) also protects the liver from alcohol-induced injury (Yin et al., 1998).
Another important repair process that can halt the propagation of toxic injury is proliferation of cells adjacent to the injured cells. This response is initiated soon after cellular injury. A surge in mitosis in the liver of rats administered a low (non-necrogenic) dose of carbon tetrachloride is detectable within a few hours. This early cell division is thought to be instrumental in the rapid and complete restoration of the injured tissue and the prevention of necrosis. This hypothesis is corroborated by the finding that in rats pretreated with chlordecone, which blocks the early cell proliferation in response to carbon tetrachloride, a normally non-necrogenic dose of carbon tetrachloride causes hepatic necrosis (Mehendale, 2005). The sensitivity of a tissue to injury and the capacity of the tissue for repair are apparently two independent variables, both influencing the final outcome of the effect of injurious chemical—that is, whether tissue restitution ensues with survival or tissue necrosis occurs with death. For example, variations in tissue repair capacity among species and strains of animals appear to be responsible for certain variations in the lethality of hepatotoxicants (Soni and Mehendale, 1998).
It appears that the efficiency of repair is an important determinant of the dose–response relationship for toxicants that cause tissue necrosis. Following chemically induced liver or kidney injury, the intensity of tissue repair increases up to a threshold dose, restraining injury, whereupon it is inhibited, allowing unrestrained progression of injury (Mehendale, 2005). Impaired signaling to mitosis (see Fig. 3-12), caused by high tissue concentrations of toxicants (eg, acetaminophen in the liver or S-(1,2-dichlorovinyl)-l-cysteine in the kidney) and their reactive metabolites may account for lagging tissue repair (Boulares et al., 1999; Vaidya et al., 2003), but maintenance of DNA and protein synthesis, mitotic machinery, and energy supply may also be impaired at high-dose chemical exposures. That is, tissue necrosis is caused by a certain dose of a toxicant not only because that dose ensures sufficient concentration of the ultimate toxicant at the target site to initiate injury but also because that quantity of toxicant causes a degree of damage sufficient to compromise repair, allowing for progression of the injury. Experimental observations with hepatotoxicants indicate that apoptosis and cell proliferation are operative with latent tissue injury caused by low (non-necrogenic) doses of toxicants, but are inhibited with severe injury induced by high (necrogenic) doses. For example, 1,1-dichloroethylene, carbon tetrachloride, and thioacetamide all induce apoptosis in the liver at low doses, but cause hepatic necrosis after high-dose exposure (Corcoran et al., 1994). Similarly, there is an early mitotic response in the liver to low-dose carbon tetrachloride, but this response is absent after administration of the solvent at necrogenic doses (Mehendale, 2005). This suggests that tissue necrosis occurs because the injury overwhelms and disables the repair mechanisms, including (1) repair of damaged molecules, (2) elimination of damaged cells by apoptosis, and (3) replacement of lost cells by cell division.
As in tissues and organs several types of cells are integrated and support the function of each other, toxic injury to different cell types may exacerbate the tissue damage and promote its progression to tissue necrosis. This principle is exemplified by the acetaminophen-induced hemorrhagic hepatic necrosis. Even before causing manifest injury to the parenchymal liver cells, acetaminophen overdose in mice has a deleterious effect on the sinusoidal endothelial cells (McCuskey, 2008). These cells swell and lose normal function (eg, endocytosis); their fenestrae coalesce into gaps that permit red blood cells to penetrate into the space of Disse. The subsequent collapse of sinusoids reduces blood flow, thus impairing oxygen and nutrient supply of hepatocytes that also endure direct damage by the reactive metabolite of acetaminophen.
Fibrosis is a pathologic condition characterized by excessive deposition of an ECM of abnormal composition. Hepatic fibrosis, or cirrhosis, results from chronic consumption of ethanol or high-dose retinol (vitamin A), treatment with methotrexate, and intoxication with hepatic necrogens such as carbon tetrachloride and iron. Pulmonary fibrosis is induced by drugs such as bleomycin and amiodarone and prolonged inhalation of oxygen or mineral particles. Doxorubicin may cause cardiac fibrosis, whereas drugs acting as agonists on 5-HT2B receptors of human valvular interstitial cells, such as bromocriptine, ergotamine, methysergide, and fenfluramine after long-term use as well as 5-HT itself when overproduced in carcinoid syndrome, induce proliferative valve disease with fibrosis. Exposure to high doses of ionizing radiation induces fibrosis in many organs. Most of these agents generate free radicals and cause chronic cell injury.
Fibrosis is a specific manifestation of dysrepair of the chronically injured tissue. As discussed above, cellular injury initiates a surge in cellular proliferation and ECM production, which normally ceases when the injured tissue is remodeled. If increased production of ECM is not halted, fibrosis develops.
The cells that manufacture the ECM during tissue repair (eg, stellate cells and myofibroblasts in liver, mesangial cells in the kidney, fibroblast-like cells in lungs and skin) are the ones that overproduce the matrix in fibrosis. These cells are controlled and phenotypically altered (“activated”) by cytokines and growth factors secreted by nonparenchymal cells, including themselves (see Fig. 3-26). TGF-β appears to be the major mediator of fibrogenesis, although other factors are also involved. These include growth factors, such as connective tissue growth factor (CTGF, a TGF-β-induced growth factor) and PDGF, vasoactive peptides, such as endothelin-1 and angiotensin-II, and the adipocyte-derived hormone leptin (Lotersztajn et al., 2005). The evidence is compelling to indicate that TGF-β, acting through its receptor (item 8 in Fig. 3-12), and receptor-activated TFs (Smad2 and 3), is a highly relevant causative factor of fibrosis. For example, subcutaneous injection of TGF-β induces local fibrosis, whereas overexpression of TGF-β in transgenic mice produces hepatic fibrosis. Smad3-null mice are relatively resistant to radiation-induced cutaneous fibrosis, bleomycin-induced pulmonary fibrosis, and CCl4-induced hepatic fibrosis. TGF-β antagonists, such as anti-TGF-β immunoglobulin and decorin, as well as Smad3 antagonists, such as halofuginone and overexpressed Smad7 protein (which is antagonistic to Smad2 and 3), ameliorate chemically induced fibrogenesis (Flanders, 2004). In several types of experimental fibrosis and in patients with active liver cirrhosis, overexpression of TGF-β in affected tissues has been demonstrated. Specific factors may also be involved in the pathomechanism of chemically induced fibrosis. For example, in alcoholic liver cirrhosis stellate cells may be activated directly by acetaldehyde (formed by alcohol dehydrogenase), by ROS (generated by the ethanol-induced CYP2E1), and by bacterial endotoxin (LPS), which is increasingly absorbed from the gut, the permeability of which is enhanced by chronic alcohol exposure. LPS can stimulate stellate cells both directly and indirectly through Kupffer cells (see Fig. 3-26) because both cells express TLR through which LPS acts.
The increased expression of TGF-β is a common response mediating regeneration of the ECM after an acute injury. However, whereas TGF-β production ceases when repair is complete, this does not occur when tissue injury leads to fibrosis. Failure to halt TGF-β overproduction could be caused by continuous injury or a defect in the regulation of TGF-β. Indeed, after acute CCl4-induced liver injury, hepatic stellate cells exhibit a TGF-β-mediated induction of Smad7 (which purportedly terminates the fibrotic signal by inhibiting activation of Smad2 and Smad3 by TGF-β receptor); however, after chronic injury, Smad7 induction fails to occur (Flanders, 2004).
The fibrotic action of TGF-β is due to increased production and decreased degradation of ECM components. TGF-β stimulates the synthesis of individual ECM components (eg, collagens) by specific target cells via the Smad pathway (see Fig. 3-12) and also by downregulation of the transcription of miR-29 miRNA family members, which inhibit the translation of collagen. Downregulation of miR-29 family members occurs in murine hepatic stellate cells exposed to TGF-β, in the stellate cells of mice with hepatic fibrosis induced by CC14 or bile duct ligation, and in the liver of patients with advanced liver cirrhosis (Roderburg et al., 2011). TGF-β inhibits ECM degradation by disproportionately increasing the expression of inhibitor proteins that antagonize ECM-degrading enzymes, such as TIMP-1 and plasminogen activator inhibitor-1 (PAI-1), compared with the expression of ECM-degrading metalloproteinases (Arthur et al., 1999; Flanders, 2004). Interestingly, TGF-β induces transcription of its own gene in target cells (Flanders, 2004), suggesting that the TGF-β produced by these cells can amplify in an autocrine manner the production of the ECM. This positive feedback (autoinduction) may facilitate fibrogenesis.
Fibrosis involves not only excessive accumulation of the ECM but also changes in its composition. The basement membrane components, such as collagen IV and laminin, as well as the fibrillar-type collagens (collagen I and III), which confer rigidity to tissues, increase disproportionately during fibrogenesis (Gressner, 1992).
Fibrosis is detrimental in a number of ways:
The scar compresses and may ultimately obliterate the parenchymal cells and blood vessels.
Deposition of basement membrane components between the capillary endothelial cells and the parenchymal cells presents a diffusional barrier that contributes to malnutrition of the tissue cells.
An increased amount of ECM and its rigidity unfavorably affect the elasticity and flexibility of the whole tissue, compromising the mechanical function of organs such as the heart and lungs.
Furthermore, the altered extracellular environment is sensed by integrins. Through these transmembrane proteins and the coupled intracellular signal transducing networks (see Fig. 3-12), fibrosis may modulate several aspects of cell behavior, including polarity, motility, and gene expression (Raghow, 1994).
Chemical carcinogenesis involves malfunctions in various repair and adaptive mechanisms. At the molecular level, a crucial feature of carcinogenesis is altered expression of critical proteins, that is, proto-oncogenic proteins and tumor suppressor proteins. This may result (1) from mutation of critical genes due to insufficient adaptive response to DNA damage and missed DNA repair or (2) from inappropriate transcriptional control at the regulatory regions of critical genes that gives rise to overexpression or underexpression of their products. (Note that the genes that we designate here as “critical” include those that are transcribed into mRNAs that in turn translate into proto-oncogenic or tumor suppressor proteins, as well as those that are transcribed into miRNAs that in turn regulate the translation of proto-oncogenic or tumor suppressor proteins. These proteins will be defined below.) At the cellular level, the fundamental feature of tumorigenesis is proliferation of cells, which may result (1) from failure to execute apoptosis and/or (2) from failure to restrain cell division.
As to be described in more detail later, carcinogenesis entails gene expression alterations initiated by two fundamentally distinct types of mechanisms that often work simultaneously and in concert, that is, genetic and epigenetic mechanisms. Genetic mechanisms bring about a qualitative change in gene expression, that is, expression of an altered gene product, a mutant protein, or miRNA, with gain or loss in activity. In contrast, epigenetic mechanisms cause quantitative change in gene expression resulting in more or less gene product. Whereas genetic mechanisms alter the coding sequences of critical genes, epigenetic mechanisms eventually influence the regulatory (promoter) region of genes. Thus, chemical and physical insults may induce neoplastic transformation of cells by affecting critical genes through genotoxic and nongenotoxic (ie, epigenetic) mechanisms. However, either mechanism ultimately induces cancer by causing cellular failures in executing apoptosis and/or restraining cell division, thereby giving rise to an uncontrollably proliferating cell population.
Genotoxic Mechanisms of Carcinogenesis: Chemical Damage and Disrepair in the Coding Region of Critical Genes Leading to Mutation
Chemicals that react with DNA may cause damage such as adduct formation, oxidative alteration, and strand breakage. In most cases, these lesions are repaired or the injured cells are eliminated. If neither event occurs, a lesion in the parental DNA strand may induce a heritable alteration, or mutation, in the daughter strand during replication. The mutation may remain silent if it does not alter the protein encoded by the mutant gene or if the mutation causes an amino acid substitution that does not affect the function of the protein. Alternatively, the genetic alteration may be incompatible with cell survival. The most unfortunate scenario for the organism occurs when the altered genes express mutant proteins that reprogram cells for multiplication and escaping apoptosis (ie, immortalization). When such cells undergo mitosis, their descendants also have a similar propensity for proliferation. Moreover, because enhanced DNA replication and cell division increases the likelihood of mutations (for reasons discussed below), these cells eventually acquire additional mutations that may further augment their growth advantage over their normal counterparts. The final outcome of this process is a nodule, followed by a tumor consisting of rapidly proliferating transformed cells (Fig. 3-31).
The process of carcinogenesis initiated by genotoxic carcinogens. The figure indicates that activating mutation of proto-oncogenes that encode permanently active oncoproteins and inactivating mutation of tumor suppressor genes that encode permanently inactive tumor suppressor proteins can cooperate in neoplastic transformation of cells. It is important to realize that overexpression of normal proto-oncogenes (eg, by hypomethylation of their promoter) and underexpression (silencing) of normal tumor suppressor genes (eg, by hypermethylation of their promoter) may also contribute to such transformation (see text for explanation).
The critical role of DNA repair in preventing carcinogenesis is attested by the human heritable disease xeroderma pigmentosum. Affected individuals lack excision repair proteins of the XP series and exhibit a greatly increased incidence of sunlight-induced skin cancers. Cells from these patients are also hypersensitive to DNA-reactive chemicals, including aflatoxin B1, aromatic amines, polycyclic hydrocarbons, and 4-nitroquinoline-1-oxide (Lehmann and Dean, 1990). Also, mice with ablated PARP gene are extremely sensitive to γ-rays and N-methyl-N-nitrosourea and show genomic instability, as indicated by sister chromatid exchanges and chromatid breaks following genotoxic insult (D’Amours et al., 1999).
A small set of cellular genes is the target for genetic alterations that initiate neoplastic transformations. Included are proto-oncogenes and tumor suppressor genes (Barrett, 1992).
Mutation of Proto-Oncogenes
Proto-oncogenes are highly conserved genes encoding proteins that stimulate the progression of cells through the cell cycle, or oppose apoptosis (Smith et al., 1993). The products of proto-oncogenes that accelerate the cell division cycle include (1) growth factors; (2) growth factor receptors; (3) intracellular signal transducers such as G proteins, protein kinases, cyclins, and cyclin-dependent protein kinases; and (4) nuclear TFs (see Figs. 3-12 and 3-32). A notable proto-oncogene product that inhibits apoptosis is Bcl-2 (see Fig. 3-19).
Key regulatory proteins controlling the cell division cycle with some signaling pathways and xenobiotics affecting them. Proteins on the left, represented by brown symbols, accelerate the cell cycle and are oncogenic if permanently active or expressed at high level. In contrast, proteins on the right, represented by blue symbols, decelerate or arrest the cell cycle and thus suppress oncogenesis, unless they are inactivated (eg, by mutation).
Accumulation of cyclin D (cD) is a crucial event in initiating the cell division cycle. cD activates cyclin-dependent protein kinases 4 and 6 (cdk4/6), which in turn phosphorylate the retinoblastoma protein (pRb) causing dissociation of pRb from transcription factor E2F (Johnson and Walker, 1999). Then the unleashed E2F is able to bind to and transactivate genes whose products are essential for DNA synthesis, such as dihydrofolate reductase (DHFR), thymidine kinase (TK), thymidylate synthetase (TS), and DNA polymerase (POL), or are regulatory proteins, such as cyclin E (cE), cyclin A (cA), and cyclin-dependent protein kinase 1 (cdk1), which promote further progression of the cell cycle. Expression of cD is increased, for example, by growth factors signaling through Ras proteins and the MAPK pathway (see Fig. 3-12) as well as by Wnt and Hedgehog (Hh) ligands that ultimately signal through B-cat and Gli transcription factors, respectively (see Figs. 3-13 and 3-14). Some carcinogens, for example, benzpyrene (BP) and reactive oxygen species (ROS), and diethylnitrosamine (DENA) may cause mutation of the Ras or Raf gene that results in permanently active mutant Ras or Rab protein, but BP as well as TCDD may also induce simple overexpression of normal Ras protein.
Cell cycle progression is counteracted, for example, by pRb (which inhibits the function of E2F), by cyclin-dependent protein kinase inhibitors (such as p15, p16, and p21), by p53 (which transactivates the p21 gene), and by ARF (also called p14 that binds to mdm2, thereby neutralizing the antagonistic effect of mdm2 on p53). Signals evoked by DNA damage and TGF-β will ultimately result in accumulation of p53 and p15 proteins, respectively, and deceleration of the cell cycle. In contrast, mutations that disable the tumor suppressor proteins facilitate cell cycle progression and neoplastic conversion and are common in human tumors. Aflatoxin B1 (ATX), BP, and UV light cause such mutations of the p53 gene (Bennett et al., 1999), whereas pRb mutations occur invariably in methylcholanthrene (MC)–induced transplacental lung tumors in mice (Miller, 1999).
Fig. 3-32 depicts several proto-oncogene products that are closely involved in initiating the cell division cycle. The legend of that figure outlines some important details on the function of these proteins and their interaction with tumor suppressor proteins (to be discussed below). Transient increases in the production or activity of proto-oncogene proteins are required for regulated growth, as during embryogenesis, tissue regeneration, and stimulation of cells by growth factors or hormones. In contrast, permanent activation and/or overexpression of these proteins favor neoplastic transformation. One mechanism whereby genotoxic carcinogens induce neoplastic cell transformation is by producing an activating mutation of a proto-oncogene. Such a mutation is so named because the altered gene (then called an oncogene) encodes a permanently active protein that forces the cell into the division cycle.
An example of mutational activation of an oncogene protein is that of the Ras proteins. Ras proteins are G-proteins with GTP/GDP-binding capacity as well as GTPase activity (Anderson et al., 1992). They are localized on the inner surface of the plasma membrane and function as crucial mediators in the signaling pathways initiated by growth factors (see Figs. 3-12 and 3-32). Ras is located downstream from growth factor receptors and nonreceptor protein tyrosine kinases and upstream from mitogen-activated protein kinase (MAPK) cascade whose activation finally upregulates the expression of cyclin D and initiates the mitotic cycle (Fig. 3-32). In this pathway, Ras serves as a molecular switch, being active in the GTP-bound form and inactive in the GDP-bound form. Some mutations of the Ras gene (eg, a point mutation in codon 12) dramatically lower the GTPase activity of the protein. This in turn locks Ras in the permanently active GTP-bound form. Continual rather than signal-dependent activation of Ras can lead eventually to uncontrolled cell division and transformation. Indeed, microinjection of Ras-neutralizing monoclonal antibodies into cells blocks the mitogenic action of growth factors as well as cell transformation by several oncogenes. Ionizing radiation and carcinogenic chemicals (eg, N-methyl-N-nitrosourea, polycyclic aromatic hydrocarbons, benzidine, aflatoxin B1) induce mutations of Ras proto-oncogenes that lead to constitutive activation of Ras proteins (Anderson et al., 1992). Most of these chemicals induce point mutations by transversion of G35 to T in codon 12.
Another example for activating mutation of a proto-oncogene is B-Raf mutation, although Ras and Raf mutations are mutually exclusive (Shaw and Cantley, 2006). Raf proteins are protein kinases, lying just downstream from Ras and being the first signal transducers in the MAP kinase pathway (see Fig. 3-12). After recruitment by Ras to the cell membrane, Raf is activated by the growth factor receptor (see item 4 in Fig. 3-12) through phosphorylation in its activating segment. B-Raf mutations occur in mouse liver tumors induced by diethylnitrosamine (Jaworski et al., 2005) in 66% of malignant melanomas and a wide range of human cancers. All mutations are within the activation segment of B-Raf, with a single amino acid substitution (V599E) accounting for the majority. The mutant B-Raf protein has elevated kinase activity probably because substitution of the nonpolar valine with the negatively charged glutamate mimics an activating phosphorylation. Thus, the constitutively active B-Raf continually sends Ras-independent proliferative signal down the MAPK pathway. Indeed, transfection of the mutant B-Raf gene into cells induced neoplastic transformation even in the absence of Ras proteins (Davies et al., 2002). Another proto-oncogene product that often undergoes activating mutation in breast and colon tumors is p110-α, the catalytic subunit of PI3K (Shaw and Cantley, 2006). This can cause permanent proliferative signaling via the GF receptor– PI3K–Akt pathway (see Fig. 3-12).
Whereas constitutive activation of oncogene proteins, as a result of point mutation, is a common initiator of chemical carcinogenesis, permanent overexpression of such proteins also can contribute to neoplastic cell transformation. Overexpression of proto-oncogene proteins may result from amplification of the proto-oncogene, that is, the formation of more than one copy (Anderson et al., 1992). Such an event may be initiated by DNA strand breaks, and therefore often observed after exposure to ionizing radiation; however, proto-oncogene amplification also occurs in spontaneous human cancer. An example for a proto-oncogene protein that is overexpressed in response to gene damage is the antiapoptotic Bcl-2 protein (see Fig. 3-19). The aberrantly increased expression of Bcl-2 is caused by a chromosomal translocation and is responsible for B-cell lymphoma, a spontaneous human malignancy (see later). Overexpression of proto-oncogene proteins as a result of nongenotoxic, epigenetic mechanisms will be discussed later.
Mutation of Tumor Suppressor Genes
Tumor suppressor genes encode proteins that inhibit the progression of cells in the division cycle, or promote DNA repair or apoptosis on DNA damage. Fig. 3-32 depicts such proteins, which include, for example, cyclin-dependent protein kinase inhibitors (eg, p15, p16, and p21), TFs (eg, p53 and Smad) that activate genes encoding cyclin-dependent protein kinase inhibitors, proteins (eg, pRb) that block TFs involved in DNA synthesis and cell division, and proteins (eg, ARF) that block inhibitors of tumor suppressor proteins. Other notable tumor suppressor gene products include, for example, the protein kinases (eg, ATM, ATR) that sense the DNA damage and signal for the p53-controlled response shown in Fig. 3-33, proteins involved in DNA repair, such as O6-methyguanine-DNA methyltransferase (which removes alkyl groups adducted to guanine) as well as BRCA1 and BRCA2 proteins (which contribute to recombinational DNA repair), proapoptotic proteins (eg, Bax, Puma, Noxa, and Bim) induced after DNA damage (Fig. 3-33), the suppressor of cytokine signaling (Socs) protein (Fig. 3-12), the phosphatase PTEN (which dephosphorylates the membrane lipid PIP3, an essential intermediate in the PI3K–Akt pathway) that turns off the PI3K–Akt pathway-mediated proliferative signaling (Fig. 3-12), and the tuberous sclerosis complex-2 (TSC2) that prevents activation of mTOR (Fig. 3-29). Uncontrolled proliferation can occur when the mutant tumor suppressor gene encodes a protein that cannot suppress cell division. Inactivating mutations of specific tumor suppressor genes in germ cells are responsible for the inherited predisposition to cancer, as in familial retinoblastoma (pRb; see Fig. 3-32), Wilms tumor (WT1, WTX, and β-catenin; see Fig. 3-13), familial polyposis (Smad4; see Figs. 3-12 and 3-32), and Li–Fraumeni syndrome (p53; see Figs. 3-19, 3-32, and 3-33). Mutations of tumor suppressor genes in somatic cells contribute to nonhereditary cancers. The genes of p16, PTEN, and pRb are frequently mutated in human cancer. The best known tumor suppressor gene involved in both spontaneous and chemically induced carcinogenesis is p53.
The guardian of the genome: p53 tumor suppressor protein—its role and regulation. When activated on DNA damage, the p53 protein may mediate cell cycle arrest, DNA repair, and apoptosis. When inducing these effects, p53 acts chiefly as a transcription factor that can activate the transcription of most target genes, while repressing some of others, such as those marked with (–) in the figure (Liu and Chen, 2006). For example, p53 transactivates p21 and gadd45 genes (whose products are inhibitors of cyclin–cyclin-dependent protein kinase complexes) and arrest the cell cycle in G1 and G2 phases, respectively, but p53 represses the Cdk1 and cyclin B1 genes (whose products are indispensable for the cells to transit from G2 phase to M) (see Fig. 3-25). p53 also induces the expression of miR-34 (a microRNA), which in turn represses the translation of cyclin D, CDK4/6, and E2F, important cell cycle accelerator proteins (see Fig. 3-32) (Chen et al., 2010). p53 also transactivates the genes of some DNA repair proteins and proapoptotic proteins (eg, bax and fas; see Fig. 3-19) and represses the genes of antiapoptotic proteins (eg, Bcl-2 and IGF-1 receptor), whereby it promotes apoptosis. These (and other) p53-induced proapoptotic mechanisms may be cell-specific, that is, all are not necessarily occurring in the same cell at the same time.
The intracellular level and activity of p53 depends primarily on the presence of mdm2 protein, which inactivates p53 by ubiquitinating it; monoubiquitination causes export of p53 from the nucleus, whereas polyubiquitination promotes its proteasomal degradation. The influence of mdm2 on p53 may be disrupted by overexpression of the ARF (or p14) protein (which binds to mdm2 and removes it from p53) or by posttranslational modification of p53 through phosphorylation by protein kinases (see below), acetylation by acetyltransferases (eg, p300 and CBP), methylation by methyltransferases (eg, Set9) and deubiquitination by ubiquitin-specific-processing protease 7 (USP7, also called HAUSP) (Liu and Chen, 2006). These mechanisms release p53 from mdm2 and stabilize the p53 protein, thereby greatly increasing its abundance and activity. Phosphorylation of p53 is induced by DNA damage. This is sensed by kinases, such as ataxia telangiectasia mutated (ATM) and ataxia telangiectasia related (ATR), which directly or through checkpoint kinases (Chk1, Chk2) phosphorylate p53 to induce cell cycle arrest, DNA repair, or apoptosis (McGowan and Russell, 2004).
It is important to emphasize that there is also a p53-independent mechanism to arrest the cells suffering DNA damage before mitosis. Like induction of p53, this is also initiated by the activated Chk1, which phosphorylates and inactivates cdc25A, a protein phosphatase, which normally would dephosphorylate Cdk1 and activate the Cdk1–cyclin B complex (see Fig. 3-25). Thus, when cdc25A is inactivated, Cdk1 stalls and mitosis is delayed. Interestingly, p53 assists in keeping Cdk1, this mitosis-driving molecular motor, off track as it induces 14-3-3σ, a cytoplasmic binding protein, which associates with both cdc25A and the Cdk1–cyclin B complex and sequesters them in the cytoplasm.
By arresting division of cells with potentially mutagenic DNA damage, facilitating the DNA repair or eliminating such cells, p53 protein counteracts neoplastic development. p53-null mice, like ARF-null mice, develop tumors with high incidence. Mutational inactivation of the p53 protein is thought to contribute to the carcinogenic effect of aflatoxin B1, sunlight, and cigarette smoke in humans. Overexpression of mdm2 can lead to constitutive inhibition of p53 and thereby promotes oncogenesis even if the p53 gene is unaltered. See the text for more details.
The p53 tumor suppressor gene encodes a 53 kDa protein with multiple functions (Fig. 3-33). Acting as a transcriptional modulator, the p53 protein (1) activates protein-coding genes whose products arrest the cell cycle (eg, p21 and gadd45), repair damaged DNA (eg, XPE, MSH2), or promote apoptosis (eg, Bax, Puma, and Fas receptor); (2) activates miRNA-coding genes whose products (eg, miR-34a) repress the translation of mitogenic TFs and cell cycle accelerator proteins (eg, Myc, E2F3, cyclin D, and CDK4/6); and (3) represses protein-coding genes that encode cell cycle accelerators (eg, cyclin B1, Cdk1), or antiapoptotic proteins (eg, Bcl-2 and IGF-1 receptor) (Bennett et al., 1999; Liu and Chen, 2006). DNA damage activates protein kinases (ATM, ATR, Chk1, Chk2) to phosphorylate and stabilize the p53 protein, causing its accumulation (Fig. 3-33). The accumulated p53 may induce cell cycle arrest and apoptosis of the affected cells. Thus, p53 eliminates cancer-prone cells from the replicative pool, counteracting neoplastic transformation; therefore, it is commonly designated as guardian of the genome.
Indeed, cells that have no p53 are a million times more likely to permit DNA amplification than are cells with a normal level of this suppressor gene. Furthermore, mice with the p53 gene deleted develop cancer by 6 to 9 months of age, attesting to the crucial role of the p53 tumor suppressor gene in preventing carcinogenesis.
Mutations in the p53 gene are found in 50% of human tumors and in a variety of induced cancers. The majority are “missense mutations” that change an amino acid and result in a faulty or altered protein (Bennett et al., 1999). The faulty p53 protein forms a complex with endogenous wild-type p53 protein and inactivates it. Thus, the mutant p53 not only is unable to function as a tumor suppressor protein but also prevents tumor suppression by the wild-type p53.
Carcinogens may cause characteristic mutations in the p53 tumor suppressor gene. An example is the point mutation in codon 249 from AGG to AGT, which changes amino acid 249 in the p53 protein from arginine to serine. This mutation predominates in hepatocellular carcinomas in individuals living where food is contaminated with aflatoxin B1 (Bennett et al., 1999). Because in human hepatocytes the CYP-activated metabolites of aflatoxin B1 induce the transversion of G to T in codon 249 of the p53 tumor suppressor gene (Aguilar et al., 1993), it appears likely that this mutation in primary human liver cancer is indeed caused by this mycotoxin. Although the incriminated mutation probably contributes to the hepatocarcinogenicity of aflatoxin B1 in humans, it is not involved in aflatoxin B1–induced hepatocarcinogenesis in rats, as the transformed liver cells from the toxin-exposed rats do not show this mutation.
Another example for inactivating mutation of a tumor suppressor gene of environmental origin is that of Patched, which often occurs in basal cell carcinoma of the skin that may be induced by UV and ionizing radiation, as well as arsenic exposure (Tang et al., 2007). Normal Patched, the membrane receptor of the Hh ligand, suppresses the mitogenic Hh pathway in absence of Hh ligand, which would inhibit Patched (Fig. 3-14). Mutant Patched loses its power to restrain this signaling network, which thus promotes cell division even in absence of its inhibitor ligand.
In addition to aberrations in critical protein-coding genes, damage in genes coding for miRNA (which repress the translation of critical proteins; see Fig. 3-11) may also contribute to carcinogenesis. For example, amplification of miR-21 (which is oncogenic by repressing the translation of PTEN; see Fig. 3-12) and deletion of miR-15 and miR-16 (which is tumor suppressive by repressing the translation of cdc25A; see Fig. 3-25) occur in spontaneous human malignancies. Therefore, miRNA gene damage may also underlie the mechanism of chemical carcinogenesis. The role of altered miRNA expression as an epigenetic mechanism of chemical carcinogenesis will be discussed below.
Epigenetic Mechanisms in Carcinogenesis: Inappropriate Activation or Responsiveness of the Regulatory Region of Critical Genes
Whereas some chemicals cause cancer by reacting with DNA and inducing a mutation, others do not damage DNA, yet induce cancer after prolonged exposure (Barrett, 1992). These chemicals are designated nongenotoxic (or epigenetic) carcinogens and include (1) xenobiotic mitogens, that is, chemicals that promote proliferative signaling, such as the PKC activator phorbol esters and fumonisin B1, as well as the protein phosphatase inhibitor okadaic acid (see Fig. 3-12); (2) endogenous mitogens, such as growth factors (eg, TGF-α) and hormones with mitogenic action on specific cells, for example, estrogens on mammary gland or liver cells, TSH on the follicular cells of the thyroid gland, and luteinizing hormone on Leydig cells in testes; (3) toxicants that, when given chronically, cause sustained cell injury (such as chloroform and d-limonene); (4) xenobiotics that are nongenotoxic carcinogens in rodents but not in humans, such as phenobarbital, DDT, TCDD, and peroxisome proliferators (eg, fibrates, WI-14643, dihalogenated and trihalogenated acetic acids); and (5) ethionine and diethanolamine, which interfere with formation of the endogenous methyl donor S-adenosyl methionine (Poirier, 1994). Because several of the listed chemicals promote the development of tumors after neoplastic transformation has been initiated by a genotoxic carcinogen, they are referred to as tumor promoters. Despite the initial belief that promoters are unable to induce tumors by themselves, studies suggest that they can do so after prolonged exposure.
It appears that nongenotoxic chemicals, like the genotoxic ones, eventually also influence the expression of proto-oncogenes and/or tumor suppressor genes, but in a different manner. When continuously present, nongenotoxic carcinogens can permanently induce the synthesis of normal proto-oncogene proteins and/or repress the synthesis of normal tumor suppressor proteins, rather than inducing the synthesis of permanently active mutant proto-oncogene proteins or permanently inactive mutant tumor suppressor proteins, as the mutagenic genotoxic carcinogens do.
Carcinogens may alter the synthesis of proto-oncogene proteins and tumor suppressor proteins at transcriptional and/or translational levels. Chemicals may modify transcription of the genes into mRNAs for proto-oncogene and tumor suppressor proteins by perturbing the signal transduction to the promoter region of these genes and/or by altering the signal receptivity of this gene region by DNA methylation and histone modifications. Translation of proto-oncogene proteins and tumor suppressor proteins from their mRNA may be controlled by specific miRNAs, which repress this process (Fig. 3-11). Carcinogen-induced perturbations in signaling and promoter alterations may not only influence transcription of protein-coding genes but also affect the transcription of miRNA-coding genes. The resultant changes in the abundance in miRNAs in turn alter, in a reciprocal (inverse) manner, the synthesis of proto-oncogene proteins and tumor suppressor proteins, provided the miRNA in question targets the mRNA of such a protein. The forthcoming discussion deals with the importance of signaling, DNA as well as histone modifications, and miRNA expression in carcinogenesis.
Role of Signaling in Carcinogenesis
An important target of nongenotoxic carcinogens is the promoter region of critical genes where they can act via two distinct modes. First, they may alter the abundance and/or activity of TFs, typically by influencing upstream signaling elements ranging from extracellular signaling molecules (eg, TSH) to intracellular transducer proteins (eg, PKC). Obviously both xenobiotic and endogenous mitogens mentioned above can thus activate proliferative pathways (see Fig. 3-12) that descend to TFs (eg, Myc) that act on the promoter of proto-oncogenes coding for mRNAs of proto-oncogene proteins (eg, cyclins), and, as to be discussed below, on the promoter of genes coding for oncogenic miRNAs (eg, miR-17-92). It is easy to recognize that even nongenotoxic carcinogens of the cytotoxic type act in this manner. As described under tissue repair, cell injury evokes the release of mitogenic growth factors such as HGF and TGF-α from tissue macrophages and endothelial cells. Thus, cells in chronically injured tissues are exposed continuously to endogenous mitogens. Although these growth factors are instrumental in tissue repair after acute cell injury, their continuous presence is potentially harmful because they may ultimately transform the affected cells into neoplastic cells. Indeed, transgenic mice that overexpress TGF-α develop hepatomegaly at a young age and tumors by 12 months (Fausto et al., 2006). Mitogenic cytokines secreted by Kupffer cells are apparently involved in hepatocyte proliferation and, possibly, tumor formation induced by peroxisome proliferators in rats (Rose et al., 1999) and in the formation of the endothelial cell–derived hepatic hemangiosarcoma in mice exposed to 2-butoxyethanol (Corthals et al., 2006).
Role of DNA Methylation and Histone Modification in Carcinogenesis
As a second mode of action, nongenotoxic carcinogens may alter expression of critical genes by modifying the responsiveness of the promoter region of these genes to TFs. Promoter responsiveness is typically controlled by DNA methylation. This takes place at C5 of specific cytosine residues located in CpG islands (ie, clusters of CpG dinucleotides) in the promoter and is catalyzed by DNA cytosine methyltransferases (eg, DNMT1, DNMT3a, and DNMT3b) using S-adenosyl methionine as the methyl donor. It is well known that promoter methylation decreases the transcriptional activity of genes. For example, tissue-specific genes (eg, the genes of GI TFFs) are hypermethylated in tissues where they are not expressed, but are typically hypomethylated in tissues where they are expressed.
Promoter methylation can silence genes because it weakens binding of TFs to the promoter and because it triggers secondary alterations in histone proteins that reduce the accessibility of the promoter for TFs (Esteller, 2005). The latter mechanism involves modification of the protruding amino-terminal tails of core histone proteins by deacetylation and methylation of lysine residues. This makes the histone more compact, thereby diminishing access of TFs and other proteins involved in transcription initiation to the gene promoter. Histone deacetylases that remove the acetyl group from histone tails may be recruited by DNMTs directly, or through proteins that recognize and bind to the methylated CpG dinucleotides (eg, MeCP1, MeCP2, MBD2, MBD3). Some of these methyl-CpG-binding domain-containing proteins (eg, MBD2, MBD3) also have histone deacetylase activity. Finally, both DNMTs and MBDs can recruit histone methyltransferases to methylate histone tail lysines.
While methylation of CpG dinucleotides codes for gene silencing via histone deacetylation and methylation (“histone code”), hypomethylated genes are alert. In fact, CpG islands in the 5′-end region (promoter, untranslated region, exon 1) of genes are relatively hypomethylated in normal tissues, except for tissue- or germ-line-specific genes, and genomically imprinted genes (see earlier). Thus, if the appropriate TFs are available for a particular gene, and if the CpG island remains hypomethylated, the histones acetylated and hypomethylated, then the gene will be transcribed (Esteller, 2005). Histones are acetylated by histone acetyltransferases, such as p300 and CBP, which also acetylate other proteins, including p53 (see Fig. 3-33), and are also transcriptional coactivators.
It is well documented that the normal methylation pattern of DNA is disrupted in cancer cells. Such cells are characterized by global (average) hypomethylation, that is, decreased content of genomic 5′-methylcytosine. Paradoxically, DNA hypomethylation occurs in the face of hypermethylation of the CpG islands in the promoter region tumor suppressor genes, which are normally demethylated (Esteller, 2005). The frequently hypermethylated tumor suppressor genes in human cancers are, for example, the Cdk inhibitor p15 and p16, the mdm2-binding protein ARF, pRb that tethers E2F (see Fig. 3-30), the DNA repair enzyme MGMT, and the lipid phosphatase PTEN (see Fig. 3-12). Importantly, both global hypomethylation and tumor suppressor gene hypermethylation intensify with increased malignancy of the tumor. Relevance of hypermethylation-induced silencing of tumor suppressor genes in carcinogenesis is supported by the finding that inhibitors of DNMTs (such as 5-aza-2′-deoxycytidine) can stop the growth of cancer cells and induce their differentiation by demethylating the dormant tumor suppressor genes, thereby restoring their expression (Esteller, 2005). The consequence of global hypomethylation of DNA is less clear. Nevertheless, hypomethylation of proto-oncogenes and increased expression of their products is a plausible mechanism (Goodman and Watson, 2002), although others have also been proposed (eg, chromosomal instability, reactivation of transposable DNA elements, and loss of genomic imprinting). On comparing mouse strains, it has been suggested that their sensitivity to chemical carcinogens may be related inversely to their capacity to maintain normal patterns of DNA methylation. It is worth noting that DNA methylation is more stable in human cells than in rodent cells (Goodman and Watson, 2002).
It appears that some of the nongenotoxic carcinogens alter DNA methylation. Inhibition of DNA methylation is a plausible mechanism that underlies tumorigenesis induced by ethionine, which depletes S-adenosyl-methionine (the methyl donor for DNMTs), and diethanolamine, which inhibits cellular uptake of choline, a dietary methyl group source for methylation of homocysteine to methionine. On prolonged administration, virtually all nongenotoxic rodent liver carcinogens listed above decrease DNA methylation. In addition to global DNA hypomethylation, promoter hypomethylation of the following proto-oncogenes have been observed in mouse or rat liver: Ras or Raf, following phenobarbital treatment, c-Jun, c-Myc, and IGF-2, after treatment with dichloroacetic acid, dibromoacetic acid, or trichloroacetic acid, and c-Myc after WY-14643 dosing. Long-term arsenic exposure of mice also induces hypomethylation of hepatic DNA globally and in the promoter of ERα, a hormone-activated TF. This is associated with an increase of cyclin D, a potentially ERα-linked gene (Chen et al., 2004). In contrast, in human keratinocytes TCDD induces promoter hypermethylation in the tumor suppressor genes p16 and p53 as well as immortalization of these cells (Ray and Swanson, 2004). Although the mechanisms that initiate altered promoter methylation are currently unknown, it is possible that that altered promoter methylation plays a role in tumor promotion by these nongenotoxic carcinogens.
Role of MicroRNAs in Carcinogenesis
As miRNAs almost always repress the translation of proteins from their mRNAs (Fig. 3-11), a miRNA plays an oncogenic role if it represses the translation of tumor suppressor proteins, that is, proteins that counter cell division, repair DNA, or promote apoptosis. For example, the miR-17-92 cluster is considered oncogenic, because its members promote mitosis by repressing the translation of PTEN (an inhibitor of the mitogenic PI3K–Akt signaling; see Fig. 3-12) and p21 (a cyclin-dependent kinase inhibitor; see Fig. 3-32) and because they inhibit apoptosis by repressing the translation of the proapoptotic protein Bim. Other examples for oncogenic miRNAs include miR-29b (which also targets the mRNAs of p21 and Bim), as well as miR-221/222, the targets of which include mRNAs translating into tumor suppressor proteins, such as PTEN, p27, p57, FOXO3 (see Fig. 3-30), TIMP3 (metalloproteinase inhibitor 3), and Bim (Chen et al., 2010; Garofalo and Croce, 2011). Overexpression of such oncogenic miRNAs promotes neoplastic cell transformation, whereas their underexpression opposes it.
Conversely, a miRNA has tumor suppressive role if it represses the translation of proto-oncogene proteins, that is, proteins that facilitate cell division or oppose apoptosis. For example, the let-7 family of miRNAs is regarded tumor suppressive as they silence the translation of Ras and Myc proteins, components of proliferative signaling (see Fig. 3-12). Furthermore, miR-15a and miR-16-1 repress the translation of protein phosphatase cdc25, which dephosphorylates and thus activates CDK1 and 2 that drives the cell cycle (see Fig. 3-25), and of Bcl-2, which inhibits apoptosis (see Fig. 3-19). Another tumor suppressor miRNA is miR-34, whose targets include the mRNAs that translates into the mitogenic TFs Myc and E2F3, as well as the cell cycle accelerator proteins cyclin D and CDK4/6 (see Figs. 3-25 and 3-32) (Chen, 2010; Garofalo and Croce, 2011). Upregulation of such tumor suppressive miRNAs counters neoplastic cell transformation, whereas their downregulation stimulates it.
Expression of miRNAs, like that of mRNAs coding for proteins, can be regulated at the promoter region of their genes by TFs as well as DNA methylation and histone modification. For example, Myc activates the transcription of the oncogenic miR-17-92, whereas p53 activates the transcription of the tumor suppressive miR-34. Therefore, the latter event is part of the adaptive DNA damage stress response discussed earlier and depicted in Fig. 3-33. Indeed, genotoxic carcinogens such as N-ethyl-N-nitrosourea and tamoxifen (which forms a DNA-reactive metabolite in rats, but not in humans) as well as ionizing radiation typically induce overexpression of mi-34.
The role of miRNAs in tumorigenesis induced by nongenotoxic carcinogens is well exemplified by the rodent hepatocarcinogenesis induced by the PPARα-activator WY-14643. This process involves transcriptional downregulation of the gene coding for let-7c and consequential derepression of the translation of c-Myc mRNA into c-Myc protein. Due to its increased abundance, this TF (see Figs. 3-12 and 3-32) increasingly activates the gene coding for cyclin D, a mitogenic protein, as well as the gene coding for miR-17-92, an oncogenic miRNA. miR-17-92 in turn represses the translation of proteins that counter mitosis (eg, PTEN and p21) and promote apoptosis (eg, Bim), thus forcing hepatocytes to proliferate (Gonzalez and Shah, 2008).
Cooperation of Genotoxic and Epigenetic Mechanisms in Carcinogenesis
Genotoxic and epigenetic mechanisms most likely complement and amplify each other in chemical carcinogenesis. One chemical may exert both genotoxic and epigenetic effects. For example, estrogens produce mutagenic free radicals via redox cycling of their quinone and hydroquinone metabolites and induce receptor-mediated proliferative effect (Newbold, 2004). Well-known genotoxic carcinogens can also cause epigenetic alterations. For example, 2-acetylaminofluorene does not only form DNA adducts but also induces DNA methylation in the promoter of Rassf1a and p16 genes as well as histone methylation around the promoter of Rassf1a, p16, Socs1, Cdh1, and Cx26 genes, thereby silencing all these genes that encode tumor suppressor proteins (Pogribny et al., 2011). The p16, ARF, and MGMT tumor suppressor genes are also hypermethylated in aflatoxin B1–induced mouse lung tumor and in 7,12-dimethylbenzanthracene-induced skin tumor. The antiestrogen tamoxifen, a hepatocarcinogen in rats (but not in humans), exerts genotoxic effect and induces global hypomethylation of DNA and histones in rat liver (Tryndyak et al., 2006). Thus, epigenetic alterations evoked by genotoxic carcinogens may drive the process of carcinogenesis after the initiating genotoxic event (Pogribny et al., 2011). Conversely, global DNA hypomethylation and tumor suppressor gene hypermethylation may increase the occurrence of mutations in cells exposed to nongenotoxic carcinogens. In this sense, epigenetic mechanisms may also initiate carcinogenesis (Goodman and Watson, 2002).
Failure to Execute Apoptosis Promotes Mutation and Clonal Growth
As discussed earlier, in cells suffering DNA damage, the stabilized p53 protein may induce cell death by apoptosis (Fig. 3-33). Apoptosis thus eliminates cells with DNA damage, preventing mutations to initiate carcinogenesis. Initiated preneoplastic cells have much higher apoptotic activity than do normal cells (Bursch et al., 1992) and this can counteract their clonal expansion. In fact, facilitation of apoptosis can induce tumor regression. This occurs when hormone-dependent tumors are deprived of the hormone that promotes growth and suppresses apoptosis. This is the rationale for the use of tamoxifen, an antiestrogen, and gonadotropin-releasing hormone analogues to combat hormone-dependent tumors of the mammary gland and the prostate gland, respectively (Bursch et al., 1992).
Thus, inhibition of apoptosis is detrimental because it facilitates both mutations and clonal expansion of preneoplastic cells. Indeed, apoptosis inhibition plays a role in the pathogenesis of human B-cell lymphomas, in which a chromosomal translocation results in aberrantly increased expression of Bcl-2 protein, which overrides programmed cell death after binding to and inactivating the proapoptotic Bax protein (see Fig. 3-19). Increased levels of Bcl-2 are also detected in other types of cancer, and a high Bcl-2/Bax ratio in a tumor is a marker for poor prognosis (Jäättelä, 1999). Other antiapoptotic proteins, such as Hsp27, Hsp70, and Hsp90 (Sreedhar and Csermely, 2004) and IAP family members, may also contribute to progression of neoplasia. Survivin, a member of the IAP family, is expressed in all cancer cells but not in adult differentiated cells (Jäättelä, 1999).
Inhibition of apoptosis is one mechanism by which the rodent tumor promoter phenobarbital promotes clonal expansion of preneoplastic cells. This has been demonstrated in rats given a single dose of N-nitrosomorpholine followed by daily treatments with phenobarbital for 12 months to initiate and promote, respectively, neoplastic transformation in liver (Schulte-Hermann et al., 1990). From 6 months onward, phenobarbital did not increase DNA synthesis and cell division in the preneoplastic foci, yet it accelerated foci enlargement. The foci grow because phenobarbital lowers apoptotic activity, allowing the high cell replicative activity to manifest itself. The peroxisome proliferator nafenopin, a nongenotoxic hepatocarcinogen, also suppresses apoptosis in primary rat hepatocyte cultures (Bayly et al., 1994). Based on a recent study with WY-14643, another peroxisome proliferator (Gonzalez and Shah, 2008), it may be suggested that miR-17-92 upregulation and consequential suppression of Bim translation accounts for apoptosis inhibition.
Failure to Restrain Cell Division Promotes Mutation, Proto-Oncogene Expression, and Clonal Growth
Enhanced mitotic activity, whether it is induced by oncogenes inside the cell or by external factors such as xenobiotic or endogenous mitogens, promotes carcinogenesis for a number of reasons:
First, the enhanced mitotic activity increases the probability of mutations. This is due in part to activation of the cell division cycle, which invokes a substantial shortening of the Gl phase, allowing less time for the repair of injured DNA before replication, thus increasing the chance that the damage will yield a mutation. Although repair still may be feasible after replication, postreplication repair is error-prone. In addition, activation of the cell division cycle increases the proportion of cells that replicate their DNA at any given time. During replication, DNA becomes unpacked and its amount doubles, greatly increasing the effective target size for DNA-reactive mutagenic chemicals, including ROS.
Enhanced mitotic activity may compromise DNA methylation, which occurs in the early postreplication period and is carried out by DNMTs that copy the methylation pattern of the parental DNA strand to the daughter strand (maintenance methylation). Limitation of DNMTs by shortened G2 phase or by presence of other transacting factors might impair methylation and may contribute to overexpression of proto-oncogens, starting a vicious cycle.
During cell division, the cell-to-cell communication through gap junctions (constructed from connexins) and AJ (constructed from cadherins) is temporarily disrupted. Several tumor promoters, such as phenobarbital, phorbol esters, and peroxisome proliferators, decrease gap junctional intercellular communication. It has been hypothesized that this contributes to neoplastic transformation. Increased susceptibility of connexin-knockout mice to spontaneous and chemically induced liver tumors supports this hypothesis (Chipman et al., 2003). Lack of the AJ contributes to the invasiveness of tumor cells (see below).
Another mechanism by which proliferation promotes the carcinogenic process is through clonal expansion of the initiated cells to form nodules (foci) and tumors.
In summary, transformation of normal cells with controlled proliferative activity to malignant cells with uncontrolled proliferative activity is driven by 3 major forces: (1) accumulation of genetic damage in the form of mutant proto-oncogenes (which encode permanently activated cell cycle accelerator and/or antiapoptotic proteins) and mutant tumor suppressor genes (which encode permanently inactivated cell cycle decelerator and/or proapoptotic proteins); (2) increased transcription and/or translation of normal proto-oncogenes (causing expression of more cell cycle accelerator and/or antiapoptotic proteins); and (3) silencing of normal tumor suppressor genes at transcriptional and/or translational level (causing expression of less cell cycle decelerator and/or proapoptotic proteins). Uncontrolled proliferation results from offset of the balance between mitosis and apoptosis (Fig. 3-34).
A model depicting the modes of action of genotoxic and nongenotoxic carcinogens and the cooperation between proto-oncogenes and tumor suppressor genes in transformation of normal cells with controlled proliferation into neoplastic cells with uncontrolled proliferation. When produced in appropriate quantities, the normal proteins encoded by proto-oncogenes  and tumor suppressor genes  reciprocally influence mitosis and apoptosis and thus ensure controlled cell proliferation. However, the balance between the effects of these 2 types of proteins on the fate of cells is offset by carcinogens via genotoxic and nongenotoxic mechanisms resulting in uncontrolled proliferation.
A genotoxic (or DNA-reactive) carcinogen may induce cell proliferation in 2 ways. In the first way, it inflicts DNA damage (eg, by forming DNA adducts) that ultimately brings about an activating mutation  in a proto-oncogene  and the mutant proto-oncogene (then called an oncogene)  in turn encodes a constitutively (ie, permanently) active oncogene protein  that continuously signals for mitosis or against apoptosis , depending on its function. In the second way, the DNA-reactive chemical produces DNA damage that eventually yields an inactivating mutation  in a tumor suppressor gene , with the mutant gene  encoding an inactive tumor suppressor protein  that cannot restrain mitosis or evoke apoptosis (eg, in response to DNA damage). In both instances, the rate of mitosis will exceed the rate of apoptosis  and uncontrolled proliferation of the affected cells will ensue . Such a scenario may underlie the carcinogenicity of aflatoxin B1, which induces mutation sometimes in the Ras proto-oncogene and often in the p53 tumor suppressor gene (see text for details).
Nongenotoxic (epigenetic) carcinogens may also induce cell proliferation by 2 modes of action: first, by causing the overexpression of normal proto-oncogenes , yielding increased quantity of their protein products , which in turn excessively stimulate mitosis or inhibit apoptosis . The second mode involves the underexpression of normal tumor suppressor genes , yielding diminished quantity of their protein products , which thus fail to restrain mitosis or promote apoptosis appropriately . Nongenotoxic carcinogens may induce the synthesis of proto-oncogene proteins  at transcriptional and/or translational levels. They may facilitate the transcription of a proto-oncogene  into its mRNA  either by increasing the abundance of active transcription factors (TFs) at the promoter region of the gene or by facilitating the accessibility of TFs to the promoter (eg, by hypomethylation of this region) . Nongenotoxic carcinogens may promote the translation of the mRNA into proto-oncogene protein  by decreasing the expression of microRNA (miRNA)  that normally represses the translation of this protein . (Such miRNAs have tumor suppressor roles.) Nongenotoxic (also called epigenetic) carcinogens also may reduce the synthesis of tumor suppressor proteins  at transcriptional and/or translational levels. They may diminish the transcription of a tumor suppressor gene  into its mRNA  either by decreasing the abundance of active TFs at the promoter region of the gene or by impeding the accessibility of TFs to the promoter (eg, by hypermethylation of this region) . Nongenotoxic carcinogens may downregulate the translation of the mRNA into a tumor suppressor protein  by increasing the expression of a miRNA  that normally represses the translation of this protein . (Such miRNAs have oncogenic roles.) Eventually, overexpression of proto-oncogene proteins and/or underexpression of tumor suppressor proteins produce mitosis rate that exceeds the rate of apoptosis , thereby leading to uncontrolled proliferation of the affected cells . Examples for nongenotoxic carcinogens acting by these mechanisms are given in the text.
In effect, the modes of action of these 2 types of chemical carcinogens are more complex: genotoxic carcinogens may also exert epigenetic effects and nongenotoxic carcinogens may increase the frequency of spontaneous mutations as well as the division and survival of cells carrying mutations (see the text for details).
Genotoxic carcinogens appear to induce cancer primarily by inducing activating mutations in proto-oncogenes or inactivating mutations in tumor suppressor genes, and secondarily by causing inappropriate expression rate of these critical genes. In contrast, nongenotoxic carcinogens cause cancer primarily by transcriptional or translational upregulation of the synthesis of proto-oncogene proteins or downregulation of the synthesis of tumor suppressor proteins. Secondarily, however, nongenotoxic carcinogens may also increase mutation of critical genes, which is initiated by genotoxic agents or spontaneous DNA damage. In human cells, spontaneous DNA damage may give rise to mutation at a frequency of 1 out of 108 to 1010 base pairs (Barrett, 1992). Genotoxic carcinogens increase this frequency 10- to 1000-fold. Nongenotoxic carcinogens also increase the rate of spontaneous mutations through a mitogenic effect (see above) and increase the number of cells with DNA damage and mutations by facilitating their division and inhibiting their apoptosis. Both enhanced mitotic activity and decreased apoptotic activity expand the population of transformed cells.
According to an emerging theory, cancers may form by genotoxic and/or epigenetic mechanisms in pluripotent stem cell populations. Such cells are characterized by quiescence, self-renewal, and conditional immortality, thus would potentially supply a lifelong, latent neoplastic population after carcinogen attack. Stem cells may be an especially likely target for transplacental carcinogens, such as DES or arsenic (Tokar et al., 2011), because they are abundant during fetal development and have the longevity to carry the molecular lesion to the adult period of life.
Finally, further changes in gene expression may occur in these proliferating cells making them capable of invading the tissue and forming metastasis. These alterations confer the cells capacity for increased mobility by (1) their transdifferentiation into a mobile fibroblastoid phenotype (called epithelial-to-mesenchymal transition [EMT]), (2) disruption of the AJ that anchor cells to their neighbor (by downregulation of E-cadherin expression), and (3) degradation of the surrounding ECM (by upregulation of matrix metalloproteinases). Signaling systems that orchestrate EMT include those triggered by TGF-β (Fig. 3-12), the Wnt ligand (see Fig. 3-13), and the Hh ligand (see Fig. 3-14) (Katoh and Katoh, 2008).