Portal de Periódicos da CAPES

The Cloning and Characterization of a New Stress Response Protein

1999; Elsevier BV; Volume: 274; Issue: 8 Linguagem: Inglês

10.1074/jbc.274.8.5131

1083-351X

Reinhard Kodym, Peter R. Calkins, Michael D. Story,

Redox biology and oxidative stress

Using differential display, a cDNA fragment was identified as being overexpressed in a mouse lymphoma cell line that had gained resistance to cell death after exposure to a variety of agents used in cancer therapy. The full-length cDNA of 1.1 kb that was cloned contained an open reading frame coding for a previously unidentified 28-kDa mammalian protein, p28. p28 showed significant homologies to a large family of stress response proteins that contain a glutathione S-transferase (GST) domain. In correspondence with the sequence homology, p28 was found to bind glutathione; however, GST or glutathione peroxidase activity could not be demonstrated. Northern analysis of the mRNA of this protein showed abundant expression in mouse heart and liver tissues, whereas anti-p28 antibody binding identified p28 expression in mouse 3T3 cells and early passage mouse embryo fibroblasts. Subcellular protein fractionation revealed p28 localization in the cytoplasm, but with thermal stress p28 relocated to the nuclear fraction of cellular proteins. Based on sequence homology and protein activity we conclude that p28 acts as a small stress response protein, likely involved in cellular redox homeostasis, and belongs to a family of GST-like proteins related to class θ GSTs. Using differential display, a cDNA fragment was identified as being overexpressed in a mouse lymphoma cell line that had gained resistance to cell death after exposure to a variety of agents used in cancer therapy. The full-length cDNA of 1.1 kb that was cloned contained an open reading frame coding for a previously unidentified 28-kDa mammalian protein, p28. p28 showed significant homologies to a large family of stress response proteins that contain a glutathione S-transferase (GST) domain. In correspondence with the sequence homology, p28 was found to bind glutathione; however, GST or glutathione peroxidase activity could not be demonstrated. Northern analysis of the mRNA of this protein showed abundant expression in mouse heart and liver tissues, whereas anti-p28 antibody binding identified p28 expression in mouse 3T3 cells and early passage mouse embryo fibroblasts. Subcellular protein fractionation revealed p28 localization in the cytoplasm, but with thermal stress p28 relocated to the nuclear fraction of cellular proteins. Based on sequence homology and protein activity we conclude that p28 acts as a small stress response protein, likely involved in cellular redox homeostasis, and belongs to a family of GST-like proteins related to class θ GSTs. To cope with a variety of adverse environmental influences, including temperature extremes, toxins, nutritional deprivation, and oxidative damage, cells from bacteria to mammals have developed common molecular responses, including alterations in gene expression that up-regulate heat shock or stress response proteins. These stress response proteins can act as molecular chaperones by binding to denatured or misfolded proteins, by regulating the correct folding of proteins, by dissociating protein aggregates, or by facilitating the transfer of proteins to specific cellular locations (for review, see Refs. 1Leppa S. Sistonen L. Ann. Med. 1997; 29: 73-78Crossref PubMed Scopus (90) Google Scholar, 2Schlesinger M.J. Pediatr. Res. 1994; 36: 1-6Crossref PubMed Scopus (79) Google Scholar, 3Burel C. Mezger V. Pinto M. Rallu M. Trigon S. Morange M. Experientia. 1992; 48: 629-634Crossref PubMed Scopus (99) Google Scholar, 4Trautinger F. Kindas-Mugge I. Knobler R.M. Honigsmann H. J. Photochem. Photobiol. B Biol. 1996; 35: 141-148Crossref PubMed Scopus (90) Google Scholar, 5Schirmer E.C. Glover J.R. Singer M.A. Lindquist S. Trends Biochem. Sci. 1996; 21: 289-296Abstract Full Text PDF PubMed Scopus (576) Google Scholar). At least one heat shock protein regulates glutathione levels, resulting in shifts in cellular redox status (6Mehlen P. Kretz-Remy C. Preville X. Arrigo A.-P. EMBO J. 1996; 15: 2695-2706Crossref PubMed Scopus (518) Google Scholar). Stress response mechanisms, like many biological phenomena, are double-edged swords in that the induction of heat shock proteins or detoxification enzymes such as glutathione S-transferases (GSTs) 1The abbreviations GSTglutathioneS-transferasePCRpolymerase chain reactionRACErapid amplification of cDNA endsPBSphosphate-buffered salineTBSTTris-buffered saline containing 0.05% Tween 20 have beneficial effects—increasing normal tissue stress tolerance, and harmful effects—enabling tumor cells to resist cytotoxic therapy, especially by drugs that increase intracellular reactive oxygen intermediates (6Mehlen P. Kretz-Remy C. Preville X. Arrigo A.-P. EMBO J. 1996; 15: 2695-2706Crossref PubMed Scopus (518) Google Scholar, 7Uozaki H. Horiuchi H. Ishida T. Iijima T. Imamura T. Machinami R. Cancer (Phila.). 1997; 79: 2336-2344Crossref PubMed Scopus (122) Google Scholar, 8Huot J. Roy G. Lambert H. Chretien P. Landry J. Cancer Res. 1991; 51: 5245-5252PubMed Google Scholar, 9Hayes J.D. Pulford D.J. Crit. Rev. Biochem. Mol. Biol. 1995; 30: 445-600Crossref PubMed Scopus (3254) Google Scholar, 10Mehlen P. Schulze-Osthoff K. Arrigo A.-P. J. Biol. Chem. 1996; 271: 16510-16514Crossref PubMed Scopus (582) Google Scholar, 11Ban N. Takahashi Y. Takayama T. Kura T. Katahira T. Sakamaki S. Niitsu Y. Cancer Res. 1996; 56: 3577-3582PubMed Google Scholar). glutathioneS-transferase polymerase chain reaction rapid amplification of cDNA ends phosphate-buffered saline Tris-buffered saline containing 0.05% Tween 20 Such an increase in resistance of tumor cells to cytotoxic drugs or ionizing radiation is observed in the murine lymphoma cell model described below. The cell line, named LY-as, was derived from a syngeneic mouse B cell lymphoma designated LY-TH (12Drewinko B. Jurin M. Howes T.A. Proc. Soc. Exp. Biol. Med. 1972; 140: 339-341Crossref PubMed Scopus (4) Google Scholar). This cell line proved to be very susceptible to apoptotic cell death induced by ionizing radiation (13Story M.D. Voehringer D.W. Malone C.G. Hobbs M.L. Meyn R.E. Int. J. Radiat. Biol. 1994; 66: 659-668PubMed Google Scholar) or chemotherapeutic drugs when maintainedin vitro. 2M. D. Story and R. E. Meyn, submitted for publication. However, during in vitro culture this cell line reproducibly loses its high susceptibility to apoptosis and becomes much more resistant to the cytotoxic agents mentioned (13Story M.D. Voehringer D.W. Malone C.G. Hobbs M.L. Meyn R.E. Int. J. Radiat. Biol. 1994; 66: 659-668PubMed Google Scholar).2 This resistant cell type was designated LY-ar (13Story M.D. Voehringer D.W. Malone C.G. Hobbs M.L. Meyn R.E. Int. J. Radiat. Biol. 1994; 66: 659-668PubMed Google Scholar). The gain in resistance is associated with an up-regulation of Bcl-2 oncoprotein and doubling of the cellular content of reduced glutathione (15Mirkovic N. Voehringer D.W. Story M.D. McConkey D.J. McDonnell T.J. Meyn R.E. Oncogene. 1997; 15: 1461-1470Crossref PubMed Scopus (203) Google Scholar). Modulation of glutathione levels by cysteine deprivation or by treatment with diamide or diethyl maleate, agents that deplete cellular thiols, after radiation results in a restoration of radiosensitivity and apoptosis induction characteristic of the parental LY-as cell line (15Mirkovic N. Voehringer D.W. Story M.D. McConkey D.J. McDonnell T.J. Meyn R.E. Oncogene. 1997; 15: 1461-1470Crossref PubMed Scopus (203) Google Scholar).2 The Bcl-2 protein level does not change as a result of cysteine deprivation (15Mirkovic N. Voehringer D.W. Story M.D. McConkey D.J. McDonnell T.J. Meyn R.E. Oncogene. 1997; 15: 1461-1470Crossref PubMed Scopus (203) Google Scholar), but it has been suggested that it is nuclear GSH levels that are critical in determining the extent of apoptosis and that nuclear membrane-associated Bcl-2 is responsible for maintenance of those levels (16Voehringer D.W. McConkey D.J. McDonnell T.J. Brisbay S. Meyn R.E. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 4426-4430Crossref Scopus (255) Google Scholar). To determine how the cellular phenotype changes from sensitive to resistant, we examined differences in the level of gene expression between the two cell lines by using the differential display technique. Apart from the expression differences of several known genes, 3R. Kodym and M. D. Story, submitted for publication. we also found a cDNA fragment exclusively expressed in the resistant cell line LY-ar that showed no homology to mRNAs of known mammalian proteins. Here we describe the cloning of the full-length cDNA of which this fragment is a part. The cDNA codes for a novel protein with a calculated molecular mass of 28 kDa. This protein shows homologies to small stress proteins and GSTs, which all belong to a large ancient protein superfamily of proteins that are related by their sequence and tertiary structure to class θ GSTs (18Koonin E.V. Mushegian A.R. Tatusov R.L. Altschul S.F. Bryant S.H. Bork P. Valencia A. Protein Sci. 1994; 3: 2045-2054Crossref PubMed Scopus (131) Google Scholar). This superfamily, which includes GSTs, small stress proteins, the eukaryotic translation elongation factor 1γ, bacterial dehalogenases, and etherases, as well as several uncharacterized proteins, is characterized by two conserved domains, a glutathione binding domain and a second domain of unknown function. The sequence homology of p28 to the conserved domains and the similar functional characteristics of p28 to other family members, in particular relocalization in response to thermal stress and ability to bind glutathione, argue for the inclusion of p28 as a new mammalian member of this superfamily. LY-as and LY-ar cells were passaged in RPMI 1640 medium supplemented with 10% fetal bovine serum and antibiotics (100 units/ml penicillin, 100 μg/ml streptomycin). The cells were incubated at 37 °C in a humidified atmosphere containing 5% CO2. For thermal stress the cells, in tissue culture flasks, were submerged in a water bath (Haake DC1) at the appropriate temperature (±0.1 °C) for 1 h. Small scale RNA isolations for the differential display were performed using RNAzol B (TEL-TEST Inc.), which is a modification of the method of Chomczynski and Sacchi (19Chomczynski P. Sacchi N. Anal. Biochem. 1986; 162: 156-159Crossref Scopus (63191) Google Scholar). Poly(A)+ RNA was isolated from total RNA using the Poly(A)Tract System III (Promega) according to the instructions given by the manufacturer. Differential display polymerase chain reaction (PCR), the cloning of the differentially expressed fragments, and the sequencing of these fragments was done as described previously.3 The full-length cDNA was cloned by rapid amplification of cDNA ends (RACE) PCR using the Marathon cDNA Amplification Kit (CLONTECH). Briefly, double-stranded cDNA was synthesized from 1 μg of poly(A)+ RNA isolated from LY-ar cells. The cDNA was ligated to adapters, and RACE PCR was performed with the gene-specific primer GGGGAAATCACAGTTTTCAGACATG and the adapter primer AP1 using the enzyme mix and the buffers of the Expand Long Template PCR System (Boehringer Mannheim). The cycling conditions were as follows: 30 cycles of 94 °C for 30 s, 60 °C for 30 s, and 68 °C for 3 min. The PCR products were separated by agarose gel electrophoresis. The major amplification product was recovered from the gel and cloned into the pCR2.1 vector and transfected into competent Eschericia coli from the TA-Cloning Kit (Invitrogen). After colony selection, sequencing was performed with a cycle sequencing kit (Amersham Pharmacia Biotech) using33P-labeled primer according to the protocol given by the manufacturer. Tissue-specific expression levels of p28 mRNA were determined by Northern blot analysis using a commercially available mouse multiple tissue Northern blot membrane (CLONTECH). The 32P-labeled probe was generated from the coding sequence of the cDNA using PCR as described previously.3 Hybridization and washes were performed exactly as proposed by the membrane manufacturer (Amersham Pharmacia Biotech). The largest cDNA product produced by RACE PCR and subsequently subcloned into the pCR2.1 plasmid as described above was cut from pCR2.1 with EspI andEcoRV and then subcloned in frame into the EcoRV site of the pET32(+) vector (Novagen). The expression vector was transformed into BL21(DE3) cells, and protein expression was induced by incubating the cells with 1 mmisopropyl-β-thiogalactopyranoside for 3 h. The vector coded for a fusion protein consisting of a thioredoxin tag, a 6xHis site, and an enterokinase cleavage site fused to the N terminus of p28. This fusion protein was purified using a His-Bind Ni2+ affinity column (Novagen). Antiserum was raised against the purified fusion protein in rabbits using standard immunization protocols. p28 was localized according to the protocol of Dyer and Herzog (20Dyer R.B. Herzog N.K. BioTechniques. 1995; 19: 192-195PubMed Google Scholar), with the exception that dithiothreitol was not included, by producing subcellular fractions of LY-ar cells. A nuclear fraction was produced by lysis of the cells in isotonic buffer containing 0.5% Nonidet P-40 followed by centrifugation for 5 min at 700 × g to pellet the nuclei. A mitochondria-rich fraction was obtained by centrifugation of the supernatant at 10,000 × g for 30 min, and then the microsomal fraction was separated from the cytoplasmic fraction (supernatant) by a 2-h centrifugation at 100,000 ×g. To determine the localization of p28 after heat treatment, the cells were heated to the appropriate temperature for 1 h, immediately washed once in cold PBS, and then lysed for 3 min on ice in a lysis buffer containing 10 mm Tris (pH 7.4), 5 mmMgCl2, 10 mm NaCl, 0.1 mmphenylmethylsulfonyl fluoride, and 0.5% Triton X-100. The particulate and detergent-soluble fractions were separated by a 5-min centrifugation step at 16,000 × g. Further discrimination of the localization of p28 was carried out as in the work of Dyer and Herzog (20Dyer R.B. Herzog N.K. BioTechniques. 1995; 19: 192-195PubMed Google Scholar) on cells heated or not to 44 °C for 1 h before lysis and separation into nuclear, mitochondrial, and microsomal fractions. Proteins were separated on a 12% acrylamide gel using the Laemmli (21Laemmli U.K. Nature. 1970; 227: 680-685Crossref PubMed Scopus (207233) Google Scholar) buffer system and then electrotransfered to a nitrocellulose membrane (Amersham Pharmacia Biotech). After the membrane was blocked with BLOTTO, it was incubated for 1 h in antiserum diluted 1:200 with BLOTTO. After three 15-min washes in Tris-buffered saline containing 0.05% Tween 20 (TBST), the membrane was incubated for 1 h in a 1:1500 dilution of peroxidase-conjugated goat anti-rabbit antibody (Amersham Pharmacia Biotech) in BLOTTO. After three final 15-min washes in TBST, the blot was developed using a chemiluminescence detection kit (ECL, Amersham Pharmacia Biotech). For GSH affinity precipitation, proteins from 5 × 106 LY-as and LY-ar cells were radiolabeled with 50 μCi [35S]methionine (Amersham Pharmacia Biotech) over 3 h. The cells were lysed in the same lysis buffer used as above for p28 localization after thermal stress. A 50-μl volume of lysate representing cytoplasmic proteins and containing ∼1.5 × 106 cpm was diluted with 450 μl of PBS and mixed with 30 μl of amino-linked GSH-Sepharose (Sigma G-9761). The mixture was incubated at 4 °C for 1 h, and then the GSH-Sepharose was washed three times for 15 min in PBS at 4 °C. The bound proteins were solubilized and released from the GSH-Sepharose matrix by boiling the GSH-Sepharose in SDS gel loading buffer (21Laemmli U.K. Nature. 1970; 227: 680-685Crossref PubMed Scopus (207233) Google Scholar), separated on a 12% polyacrylamide gel, and detected by autoradiography. p28 was identified on the same membrane by chemiluminescence as described above. GST activities were determined by monitoring the production of thioethers, as 1-chloro-2,4-nitrobenzene, trans-4-phenyl-3-buten-2-one, ethacrynic acid, and androstene-3,17-dione were conjugated with GSH, by UV absorbance spectrophotometry (22Habig W.H. Jakoby W.B. Methods Enzymol. 1981; 77: 398-405Crossref PubMed Scopus (2081) Google Scholar). Glutathione peroxidase activity was determined with the substrate cumene hydroperoxide in a coupled enzymatic reaction with GSH-reductase and NADPH again by spectrophotometry as described previously (23Bompart G.J. Perevot D.S. Bascands J.L. Clin. Biochem. 1990; 23: 501-504Crossref PubMed Scopus (55) Google Scholar). After using the random primer TCGATACAGG and the anchored primer T12VG for PCR and subsequent reverse transcription of LY-as and LY-ar mRNA, differential display resolved two bands with sizes of 135 and 128 bp that appeared only in the gel lane containing the LY-ar cDNA sample (Fig.1 A). Sequencing revealed that both fragments were similar to one another and only differed slightly in their polyadenylation site (Fig. 1 B). A data base search did not show any homology of these fragments to mRNAs coding for known proteins. Because these cDNA fragments are necessarily 3′ because of the anchored primer used, 5′ RACE PCR was used to obtain the upstream cDNA sequence. The position of the gene-specific primer used to amplify the cDNA is shown in Fig. 1 B. The 5′ RACE yielded a single product of ∼1.1 kb, which was homogeneous as judged directly from the cycle sequencing of this PCR product. Northern analysis of mRNA expression using the cDNA as probe against RNA from both cell lines confirmed that p28 RNA was expressed exclusively in LY-ar cells (Fig. 1 C). The full-length cDNA sequence is given in Fig. 2. The size of the cDNA, 1159 bp, was in good agreement with the size of the hybridization signal seen in the Northern blot, 1.3 kb. The position of the start codon was verified by an in-frame stop codon upstream at position 11.Figure 2Sequence of the full-length p28 cDNA. This sequence was deposited in GenBank under accession number U80819. The position of the start codon (base 100) is verified by an upstream in-frame stop codon (underlined, base 11). Apart from the major polyadenylation site given in this sequence, a minor site was identified at base 891. The major site is preceded by a polyadenylation signal from bases 1129 to 1134 (underlined), which is an imperfect match to the consensus sequence.View Large Image Figure ViewerDownload (PPT) Fig.3 shows the expression levels of p28 mRNA in various mouse tissues as determined by Northern blot analysis. The highest levels of p28 mRNA were found in liver, lung, and heart tissues. Kidney, skeletal muscle, spleen, and brain showed low levels of p28 transcript, whereas it was almost undetectable in testicular tissue. In most of the tissues, a major, larger transcript with a size of ∼1.3 kb and a minor, smaller mRNA of ∼1.1 kb were detected. 3′ RACE amplification and sequencing of the smaller cDNA revealed that the size difference was caused by the alternative polyadenylation site located at base 891. The reason for the intermediate size of the transcript in spleen cells was not determined. The full-length cDNA described above contained one long open reading frame coding for a protein of 240 amino acids with a predicted molecular mass of 27.5 kDa. The predicted protein had significant sequence homology to a diverse family of stress response proteins that contain a GST domain (18Koonin E.V. Mushegian A.R. Tatusov R.L. Altschul S.F. Bryant S.H. Bork P. Valencia A. Protein Sci. 1994; 3: 2045-2054Crossref PubMed Scopus (131) Google Scholar). Protein sequence alignments with some important members of this family are shown in Fig. 4. The closest matching sequence, 33% identity and 52% similarity, was from a hypothetical 28.5-kDa protein of unknown function found by genomic sequencing ofCaenorabditis elegans (24Wilson R. Ainscough R. Anderson K. Baynes C. Berks M. Bonfield J. Burton J. Connell M. Copsey T. Cooper J. Coulson A. Craxton M. Dear S. Du Z. et al.Nature. 1994; 368: 32-38Crossref PubMed Scopus (1439) Google Scholar). p28 also showed 40% identity and 67% similarity with a 92-amino acid-long protein fragment fromAplysia californica named protein 9. Its expression was found to be induced in Aplysia ganglia after depolarization and treatment with serotonin (25Noel F. Koumenis C. Nunez-Regueiro M. Raju U. Byrne J.H. Eskin A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 4150-4154Crossref PubMed Scopus (5) Google Scholar). Because protein 9 is a fragment, the statistical power of the seemingly greater sequence identity and similarity seen is low, which is why the C. elegans protein was considered a closer match to p28. p28 also showed significant homologies to a large number of plant proteins that have an either proven or suspected GST function and are induced as part of the plant stress response. Of these, the alignments for the GST GTX1, also called pathogenesis-related protein 1 (26Taylor J.L. Fritzemeier K.H. Hauser I. Kombrink E. Rohwer F. Schroder M. Strittmatter G. Hahlbrock K. Mol. Plant Microbe Interact. 1990; 3: 72-77PubMed Google Scholar), fromSolanum tuberosum and for the heat shock protein 26A (27Czarnecka E. Nagao R.T. Key J.L. Gurley W.B. Mol. Cell. Biol. 1988; 8: 1113-1122Crossref PubMed Scopus (126) Google Scholar) from soybean are shown in Fig. 4. Both sequences showed an identity of 22 and 24% and a similarity of 42 and 43%, respectively. Further homologies existed with each domain of the TcAc2 protein (28Schoneck R. Plumas-Marty B. Taibi A. Billaut-Mulot O. Loyens M. Gras-Masse H. Capron A. Ouaissi A. Biol. Cell. 1994; 80: 1-10Crossref PubMed Scopus (36) Google Scholar), a Trypanosoma cruzi protein that contains a tandemly repeated domain structure that catalyzes the thiol-disulfide exchange between dihydrotypanothione and glutathione disulfide (29Moutiez M. Aumercier M. Schoneck R. Meziane-Cherif D. Lucas V. Aumercier P. Ouaissi A. Sergheraert C. Tartar A. Biochem. J. 1995; 310: 433-437Crossref PubMed Scopus (53) Google Scholar). Finally, p28 showed a significant, albeit lesser, degree of homology to the sequence of the stringent starvation protein A (30Serizawa H. Fukuda R. Nucleic Acids Res. 1987; 15: 1153-1163Crossref PubMed Scopus (40) Google Scholar) from E. coli, with 23% identity and 39% similarity. The stringent starvation protein A is considered a bacterial regulatory protein (31Williams M.D. Ouyang T.X. Flickinger M.C. Mol. Microbiol. 1994; 11: 1029-1043Crossref PubMed Scopus (101) Google Scholar) that is heavily up-regulated after amino acid deprivation (32Reeh S. Pedersen S. Friesen J.D. Mol. Gen. Genet. 1976; 149: 279-289Crossref PubMed Scopus (80) Google Scholar). An antiserum was raised in rabbits using the recombinant fusion protein described under “Experimental Procedures” that consists of an N-terminal thioredoxin linked to amino acids 29–240 of p28. This antiserum recognized a protein in the Western blot with an apparent molecular mass of 32 kDa (Fig. 5 A) that was not recognized by preimmune serum (Fig. 5 B). The difference between calculated and apparent molecular mass is likely not attributable to post-translational modification of the protein, because recombinant p28 expressed in E. coli also showed the apparent molecular mass of 32 kDa when separated on an SDS-polyacrylamide gel (Fig. 5 C). The antiserum also detected high amounts of p28 in mouse fibroblast cell lines. In both NIH 3T3 and low passage mouse embryonal fibroblasts, the protein could be detected at precisely the same apparent molecular mass (data not shown). Because of the sequence homologies of p28 to proteins of the aforementioned GST superfamily, GSH binding by p28 was examined. Radiolabeled cytoplasmic proteins of LY-ar and LY-as cells were affinity precipitated with GSH linked to Sepharose, eluted from the GSH-Sepharose matrix, and separated on SDS-polyacrylamide gels. A number of protein bands of various molecular masses were apparent in extracts from both LY-ar and LY-as cells, with the exception of a 32-kDa protein that is apparent in only the LY-ar protein preparations (Fig.6 A). The identification of this glutathione-binding protein as p28 was confirmed by its reactivity with p28 antiserum (Fig. 6 B). Using 1-chloro-2,4-nitrobenzene,trans-4-phenyl-3-buten-2-one, ethacrynic acid, and androstene-3,17-dione, which are regarded as representative substrates for many types of GSTs (33Jakoby W.B. Methods Enzymol. 1985; 113: 495-499Crossref PubMed Scopus (70) Google Scholar), we were unable to demonstrate an enzymatic function for the recombinantly derived protein. Nor could we demonstrate glutathione peroxidase activity against the substrate cumene hydroperoxide. Crude nuclear, mitochondrial, microsomal, and cytoplasmic protein extracts of LY-ar cells were prepared and analyzed for their p28 content (Fig.7 A). The greatest amount of p28 was found in the cytoplasmic fraction. Very small amounts of protein were also detected in the mitochondrial and nuclear fraction; however, because these two fractions likely contained small amounts of contaminating cytoplasmic proteins, the primary subcellular localization of p28 was considered cytoplasmic. It is well established that small heat shock proteins from mammalian cells localize into the nucleus or into the detergent-insoluble fraction of cellular protein preparations after exposure of cells to elevated temperatures (34Arrigo A.P. Suhan J.P. Welch W.J. Mol. Cell. Biol. 1988; 8: 5059-5071Crossref PubMed Scopus (300) Google Scholar). p28 was a cytoplasmic protein at 37 °C (Figs. 7 and 8). If LY-ar cells were incubated at elevated temperatures, in this case for 60 min, p28 relocalized from the detergent-soluble fraction,e.g. cytosol, into the particulate fraction (Fig.7 B). This effect could be observed at temperatures ranging from 42 to 44 °C. After 60 min at 44 °C, almost all of the p28 signal was found in the particulate fraction. Further fractionation of cellular proteins revealed that p28 relocalization as a result of heat exposure was predominantly to the nucleus (Fig. 8 A). Although hyperthermic treatment leads to a rapid induction of some small heat shock proteins (1Leppa S. Sistonen L. Ann. Med. 1997; 29: 73-78Crossref PubMed Scopus (90) Google Scholar), we were unable to observe any heat inducibility of p28 in LY-as cells. The sequence homology of p28 suggests that it belongs to a broad family of proteins evolutionarily related to class θ GSTs (35Pemble S.E. Taylor J.B. Biochem. J. 1992; 287: 957-963Crossref PubMed Scopus (180) Google Scholar). Proteins belonging to this class have two defined motifs, the first of which contains a glutathione binding domain, whereas the second is thought to be a core structural domain (18Koonin E.V. Mushegian A.R. Tatusov R.L. Altschul S.F. Bryant S.H. Bork P. Valencia A. Protein Sci. 1994; 3: 2045-2054Crossref PubMed Scopus (131) Google Scholar). Amino acids 69–95 of p28 conform to the first motif, which comprises the glutathione binding domain. By way of comparison of the GST domain of p28 with class θ GSTs, analysis of the crystal structure of a GST θ class enzyme fromLucilla cuprina has identified the amino acids within the first motif that interact with glutathione (36Wilce M.C.J. Board P.G. Feil S.C. Parker M.W. EMBO J. 1995; 14: 2133-2143Crossref PubMed Scopus (218) Google Scholar). Amino acids 64–66 in the Lucilla enzyme, the highly conserved glutamate and serine and a less conserved arginine, interact with the γ-glutamyl residue of GSH. The corresponding amino acids in p28 are at positions 85 and 86, which are conserved, and position 87, which is a less-conserved valine. The crystal structure of the Lucillaenzyme also demonstrated that a serine residue at position 9 is responsible for activation of the thiol group of GSH. There are four serine residues in the N-terminal region of p28, at positions 5, 6, 8, and 13, that could fulfill that function. Finally, two other amino acids interact with the cysteinyl residue of GSH in theLucilla enzyme. They are isoleucine and tyrosine at amino acid positions 52 and 113, respectively, and although valine appears to be substituted for isoleucine in p28, no amino acid corresponding to the tyrosine could be identified on the basis of the protein sequence. The same holds true for the two histidines at positions 38 and 50 of the Lucilla enzyme, which were found to interact with the glycine of glutathione. The amino acid residues 162–189 in the p28 sequence conform to the second conserved motif described by Koonin et al. (18Koonin E.V. Mushegian A.R. Tatusov R.L. Altschul S.F. Bryant S.H. Bork P. Valencia A. Protein Sci. 1994; 3: 2045-2054Crossref PubMed Scopus (131) Google Scholar). This domain is not involved in substrate binding, and its function is less clear. A conserved aspartic acid residue (amino acid 173 in p28) is considered important (18Koonin E.V. Mushegian A.R. Tatusov R.L. Altschul S.F. Bryant S.H. Bork P. Valencia A. Protein Sci. 1994; 3: 2045-2054Crossref PubMed Scopus (131) Google Scholar), and this domain may be a key structural element in the conserved core of θ class GSTs. The presence of these conserved domains in p28 justified the assumption that it would bind glutathione and catalyze the conjugation of glutathione to other substrates. We were able to demonstrate that p28 binds to glutathione, but regardless of the substrate used, no enzymatic activity could be associated with the recombinant p28 protein. The substrates used to identify an enzymatic activity for p28 have been shown to be substrates for a number of GST isoenzymes (9Hayes J.D. Pulford D.J. Crit. Rev. Biochem. Mol. Biol. 1995; 30: 445-600Crossref PubMed Scopus (3254) Google Scholar). However, when compared with more common GSTs, recognition of a specific electrophilic ligand by p28 would be unusual because of the low specificity for electrophilic ligands by more common GSTs (33Jakoby W.B. Methods Enzymol. 1985; 113: 495-499Crossref PubMed Scopus (70) Google Scholar). Furthermore, at least 100 different chemicals are known to activate GSTs, and many also act as the substrate for the up-regulated GST; so it is conceivable that a substrate for p28 is among them. Another reason for the lack of measurable GST activity could be that the recombinant protein, despite its solubility, is incorrectly folded. We have, however, observed a similar lack of GST activity by p28 using the same substrates against mammalian cells that express p28 as part of their normal growth condition (data not shown). Induction of many of the plant proteins to which p28 shares homology is caused by a wide variety of stresses. Fungal infections induce protein GTX1 in S. tuberosum (26Taylor J.L. Fritzemeier K.H. Hauser I. Kombrink E. Rohwer F. Schroder M. Strittmatter G. Hahlbrock K. Mol. Plant Microbe Interact. 1990; 3: 72-77PubMed Google Scholar), auxins induce GTXA inArabidopsis thaliana (37van der Kop D.A.M. Schuyer M. Scheres B. van der Zaal B.I. Hooykaas P.J.J. Plant Mol. Biol. 1996; 30: 839-844Crossref PubMed Scopus (38) Google Scholar), and heavy metal exposure or heat shock induces heat shock protein 26A in soybean (38Czarnecka E. Edelman L. Schoffl F. Key J.L. Plant Mol. Biol. 1984; 3: 45-58Crossref PubMed Scopus (114) Google Scholar). We therefore tested the hypothesis that p28 might respond to stress, in this case heat shock. By fractionating cellular proteins based on detergent solubility, we first showed that with thermal stress p28 localized into the detergent-insoluble fraction of cellular protein isolates. When examined in more detail, p28 was shown to be localized in the nuclear fraction as a result of heat shock, an association that is a well known feature of small heat shock proteins (34Arrigo A.P. Suhan J.P. Welch W.J. Mol. Cell. Biol. 1988; 8: 5059-5071Crossref PubMed Scopus (300) Google Scholar). For example, the constitutively expressed protein hsp27 relocalizes from a detergent-insoluble fraction to a soluble cytoplasmic fraction after stress (39Mehlen P. Mehlen A. Guillet D. Preville X. Arrigo A.-P. J. Cell. Biochem. 1995; 58: 248-259Crossref PubMed Scopus (100) Google Scholar, 40Mehlen P. Arrigo A.-P. Eur. J. Biochem. 1994; 221: 327-334Crossref PubMed Scopus (112) Google Scholar), whereas hsp70 relocalizes to the nucleoli when cells are stressed (41Velazquez J.M. Lindquist S. Cell. 1984; 36: 655-662Abstract Full Text PDF PubMed Scopus (304) Google Scholar). Concomitant with the p28 localization, the inducibility of p28 after thermal stress was examined, but no increase in p28 levels were apparent in LY-ar cells (perhaps p28 is already expressed maximally) or in LY-as cells, in which p28 expression could not be measured. p28 contains a glutathione binding domain and is translocated after thermal stress. p28 mRNA was up-regulated in mouse tissues associated with higher oxygen tensions such as heart, lung, and kidney but was less evident in spleen, skeletal muscle, and testicular tissues. LY-ar cells, isolated from LY-as cells after extensive cell culture, also contain twice the level of cellular reduced glutathione. Note that although LY-ar cells have twice the glutathione concentration, bithionine sulfoximine, the specific inhibitor of glutathione synthesis, does not alter the apoptotic propensity of LY-ar cells (15Mirkovic N. Voehringer D.W. Story M.D. McConkey D.J. McDonnell T.J. Meyn R.E. Oncogene. 1997; 15: 1461-1470Crossref PubMed Scopus (203) Google Scholar), and addition of N-acetyl cysteine in millimolar concentrations reduces apoptosis in LY-as cells only slightly (data not shown). Although generic inhibitors of thiols are effective in modulating apoptosis as stated in the introduction, it may be the specific compartmentalization of glutathione that is critical (16Voehringer D.W. McConkey D.J. McDonnell T.J. Brisbay S. Meyn R.E. Proc. Natl. Acad. Sci. U. S. A. 1998; 95: 4426-4430Crossref Scopus (255) Google Scholar). Because p28 is abundantly expressed in 3T3 and mouse embryo fibroblasts, it argues against a role for p28 in apoptosis, because neither 3T3 nor embryo fibroblasts undergo radiation-induced apoptosis. In addition, altered gene expression in LY-ar cells3 could be attributable to LY-ar cells having become somewhat more differentiated. Admittedly, we have no evidence of that having occurred. However, lymphoid cells such as K562 cells are known to differentiate as a result of environmental stress. In fact, it is conceivable that although Bcl-2 up-regulation in LY-ar cells is responsible for blocking apoptosis, it may not be responsible for the radiation or chemoresistance of LY-ar cells, because sufficient evidence now exists showing that Bcl-2 up-regulation does not predict resistance (14Lock R.B. Stribinskiene L. Cancer Res. 1996; 56: 4006-4012PubMed Google Scholar, 17Elliot M.J. Stribinskiene L. Lock R.B. Cancer Chemother. Pharmacol. 1998; 41: 457-463Crossref PubMed Scopus (22) Google Scholar). We speculate that p28 may be involved in the cellular defense against or in the adaptive response to altered cellular redox conditions. LY-as cells were originally isolated from a mouse B cell lymphoma grownin situ, and the conversion of LY-as cells to LY-ar cells is likely a prime example of the redox-adaptive response as these cells adapt to growth in culture conditions foreign to them, e.g.high oxygen tension. This may also be the case for 3T3 cells and especially low passage mouse embryo fibroblasts, in which p28 was highly expressed. Whether p28 functions directly to alter redox status by glutathione binding or modulates the activity of other proteins via binding through its translocation activity is unknown at this time but is under pursuit. However, the expression and translocation of p28 may provide cells with a greater diversity of mechanisms for coping with insults such as thermal stress or altered redox status that may be responsible for the radiation and chemoresistance of LY-ar cells. We acknowledge the help and expert advice of Dr. Michael Weil in the field of molecular biology and Marvette L. Hobbs in cell culture. We also acknowledge the help of Dr. Zhaohui Pan and Dr. Jian Kuang for critically reading the manuscript. For the production of the antibody used we acknowledge National Institutes of Health Institutional Core Grant CA16672 and thank the staff of the Department of Veterinary Science of the M. D. Anderson Cancer Center for their skillful support.