Role of the HAP1 protein in repair of oxidative DNA damage ... - NCBI

British Journal of Cancer (1996) 74, (Suppl. XXVII) S145-S150

© 1996 Stockton Press All rights reserved 0007-0920/96 $12.00

Role of the HAP1 protein in repair of oxidative DNA damage and regulation of transcription factors G Barzilay, LJ Walker, DG Rothwell and ID Hickson Imperial Cancer Research Fund Laboratories, Institute of Molecular Medicine, University of Oxford, Oxford OX3 9DU, UK.

John

Radcliffe Hospital,

Keywords: HAPI; DNA damage; DNA repair; apoptosis; transcription factors

All organisms that grow in an aerobic environment generate reactive derivatives of molecular oxygen during the course of their normal cellular metabolism. These reactive oxygen species (ROS) are a threat to cellular viability and must be counteracted by defence mechanisms. The primary cellular target for most ROS generated both endogenously by metabolism, and exogenously by chemical oxidants and radiation, is DNA. Damage to DNA can take various forms, necessitating the expression by cells of a wide variety of different DNA repair enzymes. However, recent evidence indicates that certain key enzymes involved in DNA metabolism, in particular a family of proto-oncogene products which act as transcription factors, are also targets for oxidants, leading to a diminution in their DNA binding activity. Thus, oxidative stress may not only damage cellular macromolecules directly, but also perturb the function of key regulatory proteins. Oxidative stress can be defined as an elevation in the intracellular level of one or more ROS and represents a potentially toxic insult to cells which can be counteracted in a number of different ways. The ROS most frequently implicated in cytotoxicity are the superoxide anion, nitric oxide, hydrogen peroxide and the short-lived, but highly reactive, hydroxyl radical (Sies, 1991; Ames et al., 1993). The consequences of an accumulation of ROS at levels which exceed the capacity of a cell to counteract them are numerous, and include: (1) lipid peroxidation, leading to membrane dysfunction; (2) DNA damage, leading to an acquisition of mutations and/or cell death; (3) oxidation of proteins, leading potentially to loss of function or dysfunction. To prevent a high level of damage to one or more cellular micromolecules, all cells have evolved a network of defence mechanisms which permit the maintenance of viability even in the face of a continuous assault by ROS. Among these defences are enzymes which act directly upon a particular reactive species and render it harmless, such as the catalases, which convert hydrogen peroxide into water. If such detoxification systems are overwhelmed, damage to cellular macromolecules will result. However, DNA damage in particular, can often be repaired by the wide range of DNA repair enzymes expressed in all living organisms. This repair serves to maintain the integrity of the genome and to prevent an accumulation of mutations and/or cell death, unless the level of damage is sufficiently great to preclude complete repair. It appears unlikely that damage to proteins is 'repaired' in an analogous way; instead damaged proteins may be targeted for degradation. However, evidence has emerged that the function of certain proteins inactivated by oxidation can be restored by a reducing function provided by chemical

Correspondence: ID Hickson

reductants be viewed

other cellular proteins. This process can thus form of 'protein repair'. It seems likely, therefore, that all cells have evolved numerous enzymological and regulatory strategies for resisting damage and hence maintaining viability. This paper will review the various strategies adopted by cells to minimise damage to nucleic acid and proteins caused by oxidative stress, and will focus primarily on the action of a human nuclear enzyme, HAP1 (Robson and Hickson, 1991), also called APE, APEX and Ref-I (Demple et al., 1991; Seki et al., 1991; Xanthoudakis and Curran, 1992a), which performs a dual role in protection against cytotoxic stress. HAPI is both a DNA repair enzyme, which participates in the removal of a wide range of DNA lesions, and a 'redox' factor, which can modulate the DNA binding activity of several transcription factors (reviewed by Barzilay and Hickson, 1995). We shall review both of these functions and address the possible role of HAP1 in regulating cellular responses to cytotoxic stresses; in particular, exposure to low levels of oxygen (hypoxia). or

as a

Current status Role of HAP] in DNA base excision repair The DNA base excision repair pathway has evolved to remove both small adducts from DNA (such as that resulting from alkylation), which are generally associated with a single base residue, and oxidative damage involving base modification and/or disruption of the sugar-phosphate backbone of DNA (reviewed by Demple and Harrison, 1994). There are numerous target sites for alkylation and oxidative attack on DNA bases. Among the most significant of these are alkylation at the 06-position of guanine, which is both cytotoxic and mutagenic, alkylation at the N-3 position of adenine and guanine, which is toxic, and finally oxidation at the C-8 position of guanine, which is highly pro-mutagenic (reviewed by Demple and Harrison, 1994). Disruption of the sugar-phosphate backbone is also potentially cytotoxic, particularly when that disruption fragments the deoxyribose moiety and causes a strand break that has atypical termini (i.e. not 3' OH and 5' phosphate) that preclude immediate gap filling repair. Since all DNA repair polymerases use as a priming site an oligonucleotide terminating in a 3' OH group, no simple resynthesis of new DNA to replace the damaged section can occur until repair enzymes restore a 3' OH priming group on the non-template strand. The basic features of the base excision repair pathway are shown in Figure la. Alkylated or oxidised bases are recognised by repair enzymes called DNA glycosylases which effect the hydrolysis of the N-glycosyl bond linking the base to the DNA backbone. A wide variety of glycosylases has been identified in bacteria and eukaryotes. Some are highly selective for one or a small number of modified bases, while others recognise a wide range of modifications, such as several forms of base alkylation. By their action, DNA glycosylases generate apurinic/apyrimidi-

HAP1 and oxidative damage G Barzilay et al

Repair of oxidative DNA damage

(a)

(b)

(I)

(I)

5A91

LKXCS1IVDCG L3ANZGL DWVKZAPD LCXQ3!=Z NKPALQ3L 111 V X.OMWRQL&A XV Q.PDV XGLQTKVH DWPLZVAN 49

xoiI

*

* *

* ***

*

EAP1 PGLSBQTS PSDUGS

-G-G*-Gill III III -C- C-C-

DNA glycosylase

(11)

(111)

* *

GLLSQYCPLK VSYGI .EZ

QK ... .GHYGV ALLTK3!11A VMRODMS LQ . RR *

**

**

**

*

*

** *

*

160 94

** *

E^PZ

ExoZil

- ...DsrV LtV!AVPC

.. VULKY nQwrwu

XXP5LXIWT VIWGYPQ= SRDHPZIA

VLXO&S.

203

YLUTZLKDN 144

Iphosphodiesterase

(11) -G-C-

(111)

-G-C- c-c-

*****

-G-

-C-C-C-

-c-c-cAP endonuclease

*

-G-< III III

G

-G.

E:xorZI VC. ymvT

**

-G-

IAP1

ExoIIX

PLVLCGDLNV ABZIDLRNP K. LRLV Q NL......... PVLIMNDI SPTDLDIGXG 3UNCLR!T GKC8FV MR I .

*

** *

*

* *

* ***

247 194

*

c-cDNA polymerase p DNA ligase

NUP1 ExoXil

PLADMVBY PN!PYAYmFW TsBn JSD V=UX.DY7LL SRSL .... LN LVDTY=M PQThDRISUW DYRSMKDDN DGLRXDLLLA SQPLAZCCVZ

293

243

-G-G-G-

-c-c-c-

WiP

ALCDSKZR8K ALG8DM:PXT LYLAL

ExOilZ TGIDYUIRBM ZKPSDIEAPVW ATrU

318

268

p

phosphodiesterase

(IV)

-G-Gill III -C- C-C-

DNA polymerase ,B DNA ligase

(V)

Figure 2 Alignment of the HAPI and exonuclease III amino acid sequences. Identical residues are indicated by an asterisk. Residue numbers are shown on the right. The -60 residue Nterminal domain of HAP1 not present in exonuclease III has been omitted from the alignment.

-G-G-GIII

III

III

-c-c-cFigure 1 Repair of oxidative DNA damage. (a) The base excision repair pathway. The first stage is removal of an abnormal base (indicated by an asterisk) by a DNA glycosylase. The abnormal base is excised via hydrolytic cleavage of the Nglycosyl bond to leave an AP site. An AP endonuclease then cleaves the phosphodiester backbone 5' to the AP site via a hydrolytic mechanism. This generates a 3'OH priming terminus and 5' deoxyribose phosphate (5' dRp) group. The 5' dRp group is excised by a deoxyribose phosphodiesterase, producing a one residue gap bounded by 3'OH and 5' phosphate groups. This gap is then filled by a DNA polymerase, almost certainly polymerase ,B in human cells, and the repair reaction completed by the sealing of the nick by a DNA ligase. (b) Repair of DNA strand breaks. Oxygen radicals generate strand breaks in which a fragment of the sugar-phosphate backbone is left at the 3' side of the break. This fragment is excised by a phosphodiesterase to leave a one residue gap, which is then filled and sealed as in scheme a.

nic (AP) sites, lesions that are both mutagenic and toxic. Indeed, AP sites can be generated 'spontaneously' as a result of the inherent susceptibility of the N-glycosyl bond in DNA (not RNA) to hydrolysis which is accentuated by certain forms of alkylation damage and by oxidative attack on the DNA backbone (reviewed by Lindahl, 1990, 1993). A comprehensive review of AP sites and their repair has been published recently (Barzilay and Hickson, 1995). All organisms express dedicated enzymes which recognise AP sites and cleave the sugar-phosphate backbone 5' to the site via a hydrolytic mechanism. These enzymes are called AP endonucleases (reviewed by Wallace, 1988; Doetsch and Cunningham, 1990; Demple and Harrison, 1994; Barzilay and Hickson, 1995). While this DNA cleavage step generates a 3' OH priming site for a DNA repair polymerase, the 5' deoxyribose phosphate group left after AP endonuclease action may not be efficiently removed by certain polymerases and requires excision by a deoxyribose phosphodiesterase. Repair can then be completed by the filling in of the one residue gap by a polymerase and the sealing of the break by a DNA ligase. The versatility of DNA repair enzymes One striking feature of the enzymes that participate in the base excision repair process is their versatility. This combined with the fact that many different enzymes can often perform

the same basic task (albeit with different efficiencies), means that a considerable degree of redundancy is built into the system. For example, all known AP endonucleases that have been tested possess a phosphodiesterase activity which can excise fragments of the sugar group left at the 3' side of DNA strand breaks induced by X-rays or other free radical generating agents (Figure lb). Similarly, the glycosylase responsible for removal of most oxidised purines in E. coli, the FPG or MutM protein, is also capable of excising deoxyribose phosphate groups from the 5' side of strand breaks generated by AP endonuclease action (Graves et al., 1992). Moreover, recent evidence indicates that DNA polymerase /, the enzyme implicated in the majority of DNA synthesis associated with base excision repair, is also an efficient deoxyribose phosphodiesterase (Matsumoto and Kim, 1995). Human AP endonuclease I (HAPJ) protein The major thrust of the work in the authors' laboratory concerns the characterisation of the major AP endonuclease from human cells, which is variously termed HAPI, APE, APEX and Ref-I (Robson and Hickson, 1991; Demple et al., 1991; Seki et al., 1991; Xanthoudakis and Curran, 1992). As stated above, HAPI possesses at least two repair activities relevant to base excision repair, AP endonuclease and 3' phosphodiesterase, although the latter is of low specific activity and therefore its relevance in vivo can be questioned. The cloning and sequencing of the cDNA encoding HAPI revealed strong sequence similarity to E. coli exonuclease III (Figure 2). With the exception of an approximately 60 residue N-terminal domain absent from exonuclease III, the two proteins show approximately 30% sequence identity and 50% sequence similarity (if conservative changes are included). Exonuclease III is the major AP endonuclease in E. coli and furthermore possesses a wide range of other enzymatic activities with relevance to DNA repair and/or general nucleic acid metabolism: these include 3'-+5' exonuclease, RNAase H, 3' phosphodiesterase and 3' phosphatase activities (reviewed by Barzilay and Hickson, 1995). The HAPI and exonuclease III proteins are not only structurally related but are true functional homologues in that expression of the HAP1 protein in E. coli xth mutants deficient in exonuclease III leads to restoration of a wild-type phenotype with respect to most abnormalities exhibited (Robson and Hickson, 1991). The structural and functional similarity of the exonuclease III and HAPI proteins has been exploited in an attempt to delineate the key active site residues of the HAPI protein. Mol et al. (1995) have determined the crystal structure of the exonuclease III

HAPI and oxidative damage G Barzilay et al

protein at 1.7 A resolution and based on the structure of the ternary complex of the protein with a bound metal ion and bound dCMP, a catalytic reaction mechanism was proposed. In this, critical acidic and histidine residues are responsible for activation of a water molecule to a nucleophile which attacks the scissile phosphate at the AP site. Using a combination of modelling of the HAP1 structure on that of exonuclease III and site-directed mutagenesis, we have tentatively identified the active site residues of HAPI and verified at least some of their roles in catalysis (Barzilay et al., 1995). The HAP1 active site is proposed to consist of a critical aspartate-histidine pair (Asp-283 and His-309), which are directly involved in activation of the water molecule (Figures 3 and 4). The single metal ion at the active site is probably required to orient the target phosphate correctly and possibly for polarising the P-03' bond making it more susceptible to nucleophilic attack. The metal ion is bound by the carboxylate side chains of Glu-96 and Asp-308. This one-metal ion mediated catalysis mechanism is almost certainly conserved in a number of other hydrolytic nucleases (see Barzilay and Hickson, 1995 for a discussion of this

conservation).

Asp-283

//C\0Ql e

0

.

,?

Figure 3 Proposed reaction mechanism for the hydrolytic cleavage of the sugar-phosphate backbone 5' to an AP site. Histidine-309 abstracts a proton from a water molecule, converting it to a nucleophilic hydroxide ion. This ion then attacks the phosphate group resulting in cleavage of the P-03' bond. The Asp-283 residue forms a hydrogen bond with His-309, stabilising it in a transiently positively charged state. The divalent cation, which is bound by Glu-96 and possibly Asp-308 (not shown), interacts with the negatively charged phosphate, assisting the nucleophilic attack by the hydroxide ion. See Mol et al. (1995) and Barzilay et al. (1995) for further details.

Active site of HAP1

Asp-283

HAPI

as a redox regulator of the DNA binding activity of transcription factors

It is now well established that many transcription factors bind to their specific DNA recognition sequence only when a cysteine residue located in the DNA binding domain is in a reduced state (Abate et al., 1990a, b; Frame et al., 1991). Initially, studies of this process were limited to those using simple chemical reducing agents such as dithiothreitol (DTT). Xanthoudakis et al. (1992) extended these observations by showing that a human nuclear protein, termed Ref-1, was able to substitute for DTT and reductively activate the binding of oxidised c-Fos, c-Jun and other transcription factors to their respective DNA recognition sequences. Subsequent cDNA isolation identified that Ref-I was identical to HAP1 (Xanthoudakis and Curran, 1992). It appears that the HAP1 protein itself requires to be in a reduced state in order to reactivate oxidised Fos or Jun protein, although the functional significance of this is not clear. We and others have investigated whether the DNA repair and 'redox' functions of HAPI/Ref-I are linked biochemically via an analysis of both truncated versions of HAPI and site-specific mutant forms of the protein (Walker et al., 1993; Xanthoudakis et al., 1994). We showed that the N-terminal domain from residue 36-62 was necessary for the reductive activation of oxidised c-Jun protein and that one of the 7 cysteine residues in HAPI (Cys-65) was also required for this activity. In contrast, while the DNA repair active site utilises amino acids located at different positions within the primary HAP1 sequence, many key catalytic residues lie towards the C-terminus of the protein. Moreover, a mutant HAPI protein lacking cysteine-65 is active as an AP endonuclease. Consistent with these data, exonuclease III, which lacks both the N-terminal domain found in HAP1 and a cysteine residue in an analogous position to that of cysteine-65, possesses no redox activity (Walker et al., 1993). Thus, we would conclude that HAPI comprises two structurally and functionally distinct domains, one involved in DNA repair and the other in the reduction of oxidised proteins. The domain structure of HAPI and its homologues from other eukaryotic species and bacteria is outlined in Figure 5.

FNLSIREDOXj 1

36

REPAIR

81

No. of residues l

318

H. sapiens (HAPl)

318

A. thaliana (Arp)

527

D. melanogaster (Rrp1)

679

S. pneumoniae (ExoA) E. coli (exo III)

275

_

265

LQETK

His-309

A-:n /5bp-JUo

Figure 4 Relative disposition of amino acid residues and a bound metal ion (hatched circle) in the active site of the HAPI protein. Data based on the crystal structure of exonuclease III (Mol et al., 1995) and a model for the structure of HAPI (Barzilay et al., 1995).

Figure 5 Schematic representation of the structure of AP endonucleases from a variety of different species (as indicated on the left). The minimal DNA repair domain, indicated at the top, is represented by E. coli exonuclease III and is shown as a solid black box. Within this domain is the highly conserved LQETK motif which is implicated in binding of the active site divalent metal ion. The numbers within the black boxes indicate percentage sequence identity between each homologue and the HAPI protein. Domains of the HAPI protein involved in nuclear localisation (NLS) and 'redox' functions are shown as a stippled box and an open box respectively. The residues numbers above the HAPI protein represent boundaries of domains, as defined by expressing recombinant polypeptides in E. coli. The N-terminal domains of the Arp and Rrpl proteins, which are not homologous to HAPI or to each other, are shown with crosshatching and shading respectively. The total number of amino acid residues in each enzyme is indicated in the right hand column.

HAP1 and oxidative damage G Barzilay et a!

It should be noted that the evidence linking the 'redox' activity of HAPI to a functionally significant alteration in transcription factor activity in vivo is scant at present. Two pieces of circumstantial evidence that the critical cysteine in the DNA binding domain is an important residue for Fos/Jun action (but not necessarily as a target for oxidation/reduction) are: (1) the transforming oncogene v-Jun has a cysteine to serine mutation at this site compared with the non-transforming c-Jun protein. However, this is not the only difference between the c-Jun and v-Jun proteins. (2) a mutant Fos protein lacking the target cysteine residue has enhanced transforming potential (Okuno et al., 1993). Clearly, further work is required to show definitively that the DNA binding activity of transcription factors is regulated in vivo by changes in redox status.

'a

0)

a)C (acn0 4)

Functional roles of HAPN in human cells 0

In the absence of cell lines lacking HAP1 protein, we and others have sought to use depletion of HAP1 protein via expression of antisense HAPI RNA to study the roles of HAPI in DNA repair/protection against cytotoxic stress (Ono et al., 1994; Walker et al., 1994). Using HeLa cells, we have shown that HAPI- depleted cells are hypersensitive to killing by a wide range of cytotoxic agents, including simple alkylating agents and peroxides (Walker et al., 1994), the radiomimetic glycopeptide bleomycin (Figure 6a) and the redox cycling drug menadione (Figure 6b). Chen and Olkowski (1994) also showed that HAPI- depleted cells are sensitive to killing by X-rays. Combined with the observation that HAP1 is not apparently involved in the repair of ultraviolet light-induced damage (Walker et al., 1994), it would appear that HAPI-depleted cells show a phenotype similar to that of E. coli mutant lacking the two major AP endonucleases. This suggests, but does not prove by any means, that the drug/radiation-sensitive phenotype of HAPI antisense RNA-expressing cells is due to DNA repair deficiency. However, it is possible that the hypersensitivity to oxidants is a reflection of a failure to adequately regulate transcription factors. Further work is required to confirm whether this is true or not. One surprising observation is that HAP1 appears to be required for maintenance of viability under conditions of altered oxygen tension (Walker et al., 1994). While hyperoxic conditions (95% oxygen was used in our studies) might be expected to be highly toxic to DNA repair-deficient cells due to the formation of oxidative DNA damage, HAPI-depleted cells are also hypersensitive to killing by hypoxia (1% oxygen). Little is known about the identity of cytotoxic species generated by hypoxia, although depletion of cellular ATP may be a key factor contributing to cell death. The question that obviously arises, therefore, is what role might HAP1 play in resistance to hypoxic stress? It is possible that hypoxia generates one or more DNA lesions requiring the action of HAP1 as a DNA repair enzyme. Alternatively, the putative regulatory role of HAP1 in transcription may be a necessary event for coordinating the cellular response to hypoxic stress. One well-established response to hypoxic stress is an elevation in the expression of a number of cellular factors, including glycolytic enzymes and certain growth factors. We have previously failed to observe a consistent increase in expression of HAPI under a variety of stress conditions, such as following irradiation or exposure to cytotoxic drugs (unpublished results). However, the HAP1 protein is expressed at elevated levels after growth of cells under hypoxic conditions (Figure 7). The mechanism underlying this accumulation of HAP1 protein is not clear at this time, although Yao et al. (1994) have shown that hypoxia can cause an increase in the level of HAPI mRNA. However, unlike the protein expression studies, in which cells were lysed immediately after removal from a hypoxic environment, the elevation in HAP1 mRNA levels may reflect reoxygenation stress following hypoxia and not hypoxia per se. Further work is required to clarify this point.

30

20

10

Bleomycin (gg ml-1)

1n a) 0)

4) 40 a)

cJ

La)

I

0.00

0.01

0.02

0.03

0.04

0.05

Menadione (mM) Figure 6 (a)Clonogenic survival curves of HeLa transfectants after exposure to bleomycin. Points represent the mean of two independent determinations. E, HeLa cells expressing HAPI cDNA in the sense orientation. A, HeLa cells expressing the pcDNAneo vector alone. 0, HeLa transfectants expressing the HAP1 cDNA in the antisense orientation. (b) Clonogenic survival curves of HeLa transfectants after exposure to menadione. Points represent the mean of two independent determinations. Symbols as for Figure 6a.

Time (h) 0

1

2

4

24

kDa

M

-97 -69 HAP I-

-46 -30 -

21.5

Figure 7 Western blot of HeLa cells after growth for different periods of time (as indicated above the lanes) under conditions of low oxygen tension (1% oxygen). The sizes of molecular standards (lane M) run in parallel are shown on the right. The position of the 37kDa HAP1 protein is indicated on the left.

HAP1 and oxidative damage G Barzilay et al

S149

Is HAP1 required for the cellular response to hypoxic stress? It may be significant that a number of factors, in addition to HAPI, that are implicated in control of gene expression and/ or cellular responses to stress, are found to be expressed at an elevated level during growth under hypoxic conditions. The most relevant of these to the theme of this paper would be cFos and members of the Jun family, including c-Jun and Jun D (Yao et al., 1994). The time course of induction of these different factors is not identical, however, in that the Jun proteins are induced strongly and rapidly without a requirement for reoxygenation, while Fos protein is induced more slowly and appears to require reoxygenation stress after hypoxia for maximal expression. The Fos and Jun proteins comprise the subunits of a heterodimeric transcription factor known as activator protein-I (AP-1) which binds to an element in the promoters of regulated genes called the AP-1 site (TGAGTCA, or variants of this sequence; reviewed by Ransome and Verma, 1989). That the hypoxic induction of the Fos and Jun polypeptides is of functional significance is strongly suggested by the observation that AP-1 binding activity, as measured by a gel retardation assay using cell extracts of control vs hypoxically stressed cells, is strongly elevated during an 8 h exposure to hypoxia and remains elevated for at least 24 h (Yao et al., 1994). The link between increased AP-1 binding to DNA and the HAPI protein was provided by the observation that immunodepletion of HAP1 from nuclear extracts reduces the extent of AP-1 binding (Yao et al., 1994). Unfortunately, this does not necessarily mean that AP-1 activity in vivo is regulated by the HAPI protein, since studies in cell extracts merely confirm the wellestablished ability of HAP1 to restore the DNA binding activity of Fos and Jun proteins. Not surprisingly, it is technically very difficult to show conclusively that HAPI plays a regulatory role in the action of AP-1 during hypoxic cell exposure.

Hypoxia and cell death One feature of hypoxic, as well as other stresses, is that they induce cell death via a mechanism that is characteristic of programmed cell death (apoptosis). The features of apoptotic cell death that distinguish it from necrotic death are that cells undergoing apoptosis contain a small number of discrete, highly condensed islands of chromatin with an intact nucleus (apoptotic bodies), and an intact but abnormally structured cell membrane. In addition, apoptotic cells progressively degrade their nuclear DNA in a characteristic fashion (reviewed by Collins and Rivas, 1993; Steller, 1995). Yao et al. (1995) showed that hypoxia induces apoptosis with the morphological features described above being visible as early as 4 h after cells were switched to low-oxygen growth

conditions. Moreover, even after returning hypoxic cells to normoxic conditions, apoptotic cells continued to be generated. In one experiment, HT-29 cells exposed to 8 h of hypoxia showed 30% apoptotic cells 4 h after the return to normoxic conditions. One of the key factors controlling the apoptotic pathway is the transcription factor c-myc (Amati et al., 1993). It has been suggested that the expression of c-myc, which rises during hypoxic stress, is regulated by AP-1 (Yao et al., 1995). Although circumstantial, the available data are consistent with the existence of a cascade of inducible factors that regulate the expression of c-myc and therefore the extent to which apoptosis is induced by hypoxia. In this scheme, elevated expression of c-Fos and Jun proteins leads to increased AP-1 binding activity which in turn leads to an increase in c-myc gene expression. The AP-1 factor is in turn regulated at the level of its DNA binding potential by the availability of reduced HAPI protein, which is also activated by hypoxic stress. While this is an apparently attractive proposal for regulating the response to hypoxia which would presumably be progressive in that extended hypoxia would activate a death response pathway via a progressive increase in expression of the c-myc protein, a direct role for HAPI in this process is currently only proposed. A good deal more work will be required to provide more direct evidence for a role of this protein in the apoptotic death response. Future directions

The genes and their respective proteins required for efficient DNA repair have been identified in a large number of different laboratories in the last decade. As more is learnt from structural studies about the catalytic mechanism of action of DNA repair enzymes, we should know in considerable detail how cells counteract the mutagenic and cytotoxic DNA damage which arises in all cells due to oxidative and chemical stresses. In contrast, our knowledge of how protein function is regulated by oxidation-reduction is still very patchy and a major challenge is to produce definitive evidence that protein function is regulated in vivo by changes in the oxidation state of individual amino acid residues. If this proves to be the case, such a redox regulatory process could turn out to be as important and widespread as regulation of protein function via changes in phosphorylation status. Acknowledgements We thank members of the ICRF Molecular Oncology Laboratory for useful discussions and Elizabeth Clemson for typing the manuscript. Work in the authors' laboratory is supported by the Imperial Cancer Research Fund.

References ABATE C, LUK D, GENTZ R, RAUSCHER III FJ AND CURRAN T.

(1990a). Expression and purification of the leu zipper and DNA binding domains of Fos and Jun: Both Fos and Jun contact DNA directly. Proc. Natl Acad. Sci. USA, 87, 1032- 1036. ABATE C, PATEL L, RAUSCHER III FJ AND CURRAN T. (1990b). Redox regulation of Fos and Jun DNA binding activity in vitro. Science, 249, 2257 - 1161. AMATI B, LITTLEWOOD TD, EVAN GI AND LAND H. (1993). The cMyc protein induces cell cycle progression and apoptotis through dimerization with Max. EMBO J., 12, 5083-5087. AMES BN, SHIGENAGA MK AND HAGEN TM. (1993). Oxidants, antioxidants, and the degenerative diseases of aging. Proc. Natl Acad. Sci. USA, 90, 7915-7922. BARZILAY G AND HICKSON ID. (1995). Structure and function of apurinic/apyrimidinic endonucleases. BioEssays, 17, 713 - 719.

BARZILAY G, MOL CD, ROBSON CN, WALKER LJ, CUNNINGHAM RP, TAINER JA AND HICKSON ID. (1995). Identification of

critical active site residues in the multifunctional human DNA repair enzyme HAP1. Nature Struct. Biol., 2, 561 -568. CHEN DS AND OLKOWSKI ZL. (1994). Biological responses of human apurinic endonuclease to radiation-induced DNA damage. In DNA Damage Effects in DNA Structure and Protein Recognition, Ann. NY. Acad. Sci., Wallace SS, van Houten B and Kow YW. (eds) 726, 306- 308. COLLINS MKL AND RIVAS AL. (1993). The control of apoptosis in mammalian cells. TIBS, 18, 307- 309. DEMPLE B AND HARRISON L. (1994). Repair of oxidative damage to DNA: Enzymology and biology. Annu. Rev. Biochem., 63, 915-948.

HAPI and oxidative damage G Barzilay et al

S150 DEMPLE B, HERMAN T AND CHEN DS. (1991). Cloning and expression of APE, the cDNA encoding the major human apurinic endonuclease: Definition of a family of DNA repair enzymes. Proc. Natl Acad. Sci. USA, 88, 11450- 11454. DOETSCH PW AND CUNNINGHAM RP. (1990). The enzymology of apurinic/apyrimidinic endonucleases. Mutat. Res., 236, 173- 201. FRAME MC, WILKIE NM, DARLING AJ, CHUDLEIGH A, PINTZAS A, LANG JC AND GILLESPIE DAF. (1991). Regulation of AP-1/DNA complex formation in vitro. Oncogene, 6, 205-209. GRAVES RJ, FELZENSZWALB I, LAVAL J AND O'CONNOR TR. (1992). Excision of 5'-terminal deoxyribose phosphate from damaged DNA is catalyzed by the Fpg protein of Escherichia coli. J. Biol. Chem., 267, 14429- 14434. LINDAHL T. (1990). Repair of intrinsic DNA lesions. Mutat. Res., 238, 305-311. LINDAHL T. (1993). Instability and decay of the primary structure of DNA. Nature, 362, 709-715. MATSUMOTO Y AND KIM K. (1995). Excision of deoxyribose phosphate residues by DNA polymerase f, during DNA repair. Science, 269, 699 - 703.

SEKI S, IKDEA S, WATANABE S, HATSUSHIKA M, TSUTSUI K, AKIYAMA K AND ZHANG B. (1991). A mouse DNA repair

enzyme (APEX nuclease), having exonuclease and apurinic/ apyrimidinic endonuclease activities: purification and characterisation. Biochim. Biophys. Acta, 1079, 57-64. SIES H. (1991). In Oxidative Stress: Oxidants and Antioxidants, Sies H. (ed.) Academic Press: New York. STELLER H. (1995). Mechanisms and genes of cellular suicide. Science, 267, 1445- 1463. WALKER LJ, ROBSON CN, BLACK E, GILLESPIE D AND HICKSON

JA. (1995). Structure and function of the multifunctional DNA repair enzyme exonuclease III. Nature, 374, 381 - 386.

ID. (1993). Identification of residues in the human DNA repair enzyme HAPI (Ref-i) that are essential for redox regulation of Jun DNA binding. Mol. Cell. Biol., 13, 5370-5376. WALKER LJ, CRAIG RB, HARRIS AL AND HICKSON ID. (1994). A role for the human DNA repair enzyme HAP 1 in cellular protection against DNA damaging agents and hypoxic stress. Nucleic Acids Res., 22, 4884-4889. WALLACE S. (1988). AP Endonucleases and DNA glycosylases that recognise oxidative DNA damage. Environ. Mol. Mutagen., 12, 431 -477. XANTHOUDAKIS S AND CURRAN T. (1992). Identification and characterization of Ref-i, a nuclear protein that facilitates AP-1 DNA-binding activity. EMBO J., 11, 653-665.

OKUNO H, AKAHORI A, SATO H, XANTHOUDAKIS S, CURRAN T

XANTHOUDAKIS S, MIAO GG, WANG F, PAN Y-CE AND CURRAN

AND IBA H. (1993). Escape from redox regulation enhances the transforming activity of Fos. Oncogene, 8, 695-701. ONO Y, FURUTA T, OHMOTO T, AKIYAMA K AND SEKI S. (1994). Stable expression in rat glioma cells of sense and antisense nucleic acids to a human multifunctional DNA repair enzyme, APEX nuclease. Mutat. Res., 315, 55-63. RANSOME LJ AND VERMA IM. (1989). Association of nuclear oncoproteins fos and jun. Curr. Opin. Cell Biol., 1, 536- 540. ROBSON CN AND HICKSON ID. (1991). Isolation of cDNA clones encoding a human apurinic/apyrimidinic endonuclease that corrects DNA repair and mutagenesis defects in E. Coli xth (exonuclease III) mutants. Nucleic Acids Res., 19, 5519-5523.

T. (1992). Redox activation of Fos-Jun DNA binding activity is mediated by a DNA repair enzyme. EMBO J., 11, 3323-3335. XANTHOUDAKIS S, MIAO GG AND CURRAN T. (1994). The redox and DNA repair activities of Ref-I are encoded by nonoverlapping domains. Proc. Natl Acad. Sci. USA, 91, 23- 27. YAO K-S, XANTHOUDAKIS S, CURRAN T AND O'DWYER PJ. (1994). Activation of AP-1 and of a nuclear redox factor, Ref-i, in the response of HT29 colon cancer cells to hypoxia. Mol. Cell. Biol., 14, 5997-6003. YAO K-S, CLAYTON M AND O'DWYER P. (1995). Apoptosis in human adenocarcinoma HT29 cells induced by exposure to hypoxia. J. Natl Cancer Inst., 87, 117 - 122.

MOL CD, KUO C-F, THAYER MM, CUNNINGHAM RP AND TAINER