THE JOURNAL OF BIOLOGICAL CHEMISTRY B 1993 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 268, No. 4, Issue of February 5, pp. 2571-2576,1993 Printed in U.S.A.

Expression, Characterization, and Tissue Distribution of a New Cellular Selenium-dependent Glutathione Peroxidase, GSHPx-GI*

(Received for publication, March 26, 1992)

Fong-Fong ChuS, James H. Doroshow, and R. Steven Esworthy From the Department of Medical Oncology and Therapeutics Research, City of Hope Natwml Medical Center, Duarte, California 91010

We have characterized a new selenium-dependent glutathione peroxidase, GSHPx-GI, by expressing a GSHPx-GI cDNA isolated from human hepatoma HepG2 cells in human mammary carcinoma MCF-7 cells, which have virtually undetectable expression of either the classical cellular enzyme, GSHPx-1, or GSHPx-GI at the protein level. One of the G418-re- sistant clones, neo-Dl, expresses the transfected GSHPx-GI cDNA. This is based on 1) the presence of an additional GSHPx-GI DNA restriction fragment detected by Southern analysis; 2) the presence of a 1.9- kilobase (kb) GSHPx-GI mRNA in addition to the 1.0- kb endogenous mRNA by Northern analysis; and 3) the appearance of a 22-kDa ‘%e-labeled protein which is absent in parental MCF-7 cells revealed by SDS-poly- acrylamide gel electrophoresis.

GSHPx-GI expressed in neo-Dl is a tetrameric pro- tein localized in cytosol. GSHPx-GI does not cross- react with antisera against human GSHPx-1 or human plasma glutathione peroxidase (GSHPx-P). Similar substrate specificities are found for GSHPx-1 and GSHPx-GI; they both catalyze the reduction of H202, tert-butyl hydroperoxide, cumene hydroperoxide, and linoleic acid hydroperoxide with glutathione, but not of phosphatidylcholine hydroperoxide. GSHPx-GI mRNA was readily detected in human liver and colon, and occasionally in human breast samples, but not other human tissues including kidney, heart, lung, pla- centa, or uterus. In rodent tissues, GSHPx-GI mRNA is only detected in the gastrointestinal tract, and not in other tissues including liver. In fact, GSHPx-GI appears to be the major glutathione-dependent perox- idase activity in rodent GI tract. This finding suggests that GSHPx-GI could play a major role in protecting mammals from the toxicity of ingested lipid hydroper- oxides.

In conclusion, we have demonstrated that GSHPx- GI is the fourth member in the selenium-dependent glutathione peroxidase family, in addition to GSHPx- 1, GSHPx-P, and phospholipid hydroperoxide gluta- thione peroxidase (PHGPX).

* This work was supported by American Heart Association Greater Los Angeles Affiliate, Grant-in-Aid 901 GI-2, and National Cancer Institute Cancer Center Support Grant CA 33572. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduer- tisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequencefs) reported in thispaper has been submitted

X68314. to the GenBankTM/EMBL Data Bank with accession number(s)

$ To whom correspondence should be addressed Dept. of Medical Oncology, City of Hope National Medical Ctr., 1500 E. Duarte Rd., Duarte, CA 91010-0269. Tel.: 818-359-8111 (ext. 3831/2426); Fax: 818-301-8233.

Selenium-dependent glutathione peroxidase activity is due to the expression of multiple isozymes. Three of the isozymes have been previously purified and characterized at the protein level. They are: (a) the classical cellular glutathione peroxi- dase, GSHPx-1’ (l) , ( b ) the phospholipid hydroperoxide glu- tathione peroxidase, PHGPX (2), and (c) the plasma gluta- thione peroxidase, GSHPx-P (3, 4). Both GSHPx-1 and GSHPx-P are tetramers and have similar substrate specifici- ties. They can metabolize Hz02 and fatty acid hydroperoxides effectively but metabolize phospholipid hydroperoxide and cholesterol hydroperoxide poorly (5, 6). On the other hand, PHGPX is a monomer and has different substrate specificities compared to GSHPx-1 and GSHPx-P. PHGPX catalyzes the reduction of phospholipid hydroperoxide, cholesterol hydro- peroxide, and linoleic acid hydroperoxide much more effec- tively than the reduction of H202 and tert-butyl hydroperox- ide.

GSHPx-1 and PHGPX are present in most of the tissues that have been examined. GSHPx-1 is present abundantly in erythrocytes, kidney, and liver (7); and PHGPX is present in testis at a high level (8). On the other hand, GSHPx-P exhibits tissue-specific expression: it is detected in plasma, milk, and lung (3, 4, 9, 10). In humans, GSHPx-P mRNA is expressed in liver, kidney, heart, lung, placenta, and breast tissue; whereas, in the rodent GSHPx-P is not expressed in liver, although rodent kidney has a higher level of expression which may compensate for no GSHPx-P expression in liver (11). The cDNA sequence of these three glutathione peroxidase isozymes have been previously reported; they all contain a UGA codon coding for selenocysteine (12-15).

In addition to these three isozymes, a 24-kDa epididymal secretory protein homologous to glutathione peroxidase has been reported in mouse. This 24-kDa protein is inducible by testosterone, and it is only expressed in epididymis (16, 17). Since the published nucleic acid sequence of the 24-kDa protein does not contain a UGA codon, whether it should be included as an isozyme of the selenium-dependent glutathione peroxidases is debatable.

We isolated a GSHPx-GI clone from human liver and human hepatoma HepG2 cDNA libraries in Xgtll; this cDNA shares sequence homology with GSHPx-1 (18). The cDNA clone, GSHPx-GI, was subsequently reported by Akasaka et al. (19), who also isolated it from a human liver library. GSHPx-GI cDNA contains a UGA codon; this suggests that it may be a fourth selenium-dependent glutathione peroxidase isozyme. Because the protein product of GSHPx-GI had not been identified, it was possible that GSHPx-GI was a pseu- dogene. In this report, we present data on the analysis of

The abbreviations used are: GSHPx, selenium-dependent gluta- thione peroxidase; PHGPX, phospholipid hydroperoxide glutathione peroxidase; DMEM, Dulbecco’s modified Eagle’s medium; PAGE, polyacrylamide gel electrophoresis.

2571

2572 Tissue-specific Expression of GSHPx-GI mRNA

GSHPx-GI transfectants which express a 22-kDa selenopep- tide and elevated enzymatic activity. Our results provide the first evidence establishing GSHPx-GI as the fourth isozyme of the selenium-dependent glutathione peroxidase family. Ad- ditionally, we have detected GSHPx-GI mRNA expression in gastrointestinal tract in human and rodent as well as human liver.

EXPERIMENTAL PROCEDURES

Isolation and Sequencing of GSHPx-GI cDNA-A 50-base oligo- nucleotide was synthesized based on a fragment of the bovine GSHPx-1 amino acid sequence (20). It was used to screen a bovine liver cDNA library (Clontech Lab., Inc., Palo Alto, CA). A 69-mer of bovine cDNA was isolated; its sequence, CTTCGAAAAGTGCG- AGGTGAATGGCGAGAAGGCGCATCCGCTCTTCGCCTTCCTT CGGGAGGTTCTGCC, corresponds to amino acids 107-129 of bo- vine GSHPx-1. This bovine GSHPx-1 cDNA clone was used to screen a human liver cDNA library (Clontech Lab., Inc.). Five positive clones isolated were sequenced and found to have the GSHPx-GI sequence. Unfortunately, none of these clones contained the AUG start codon for translation. Therefore, we used the GSHPx-GI cDNA to screen a HepG2 cDNA library (kindly provided by Dr. Aldon J. Lusis, Uni- versity of California, Los Angeles, CA), and isolated another three partial GSHPx-GI clones and one full-length clone. The full-length HepG2 clone and one of the liver clones were sequenced bidirection- ally. These clones share over 99% sequence identity with the pub- lished sequence (19). The sequencing reactions were performed fol- lowing Sanger's dideoxy sequencing method (21) using a Sequenase kit (United States Biochemical Co.).

Expression of GSHPx-GI cDNA in MCF-7cl Cells-The full-length GSHPx-GI cDNA clone in Xgtll was first subcloned into Bluescript plasmid (pBS, Stratagene, San Diego, CA) at an EcoRI site. To express in mammalian cells, Rous sarcoma virus long terminal repeat was used as the promotor. The construction of pRSV-GSHPx-GI was made by excision of the GSHPx-GI insert from pBS with HindIII and BamHI restriction enzymes, and ligation with a pRSV vector which was excised from pRSV-P-globin (American Type Culture Collection, Rockville, MD) with HindIII and BglII (22, 23). The pRSV-GSHPx-GI DNA was cotransfected with pSV2-ne0 or pSK1- mdr (kindly provided by Susan Kane, City of Hope) using the calcium phosphate-precipitation method (24, 25). MCF-7cl cells, a subclone of the human MCF-7 mammary carcinoma cell line, which exhibits virtually no endogenous GSHPx-1 or GSHPx-GI expression, was used as the recipient of DNA transfection (26). The transfectants were selected with either 400 pg/ml of geneticin for neo expression, or 8 ng/ml of colchicine for mdrl expression.

Cell Culture-Unless specified, all of the cell lines were obtained from the American Type Culture Collection. MCF-7 and HepG2 cells were cultured in DMEM/F-12 supplemented with 5% fetal bovine serum. MCF-7H6 cells, a GSHPx-1 transfectant, were cultured in the same medium supplemented with 400 pg/ml of geneticin (26). DU4475, MCF-IOA, and MDA-MB175 cells were cultured in DMEM/ F-12, supplemented with 5 pg/ml each of insulin and transferrin, 21 ng/ml epidermal growth factor, 60-100 ng/ml cholera enterotoxin, 50 nM hydrocortisone, 0.1 nM triiodothyronine, 10 nM estradiol, and 5% fetal bovine serum.

Southern and Northern Analysis-High molecular weight DNA was isolated from the nuclear fractions of transfected MCF-7cl cells following published procedures (27). Total RNA was isolated using the acid guanidinium thiocyanate-phenol-chloroform extraction method (28). Poly A-containing mRNA was isolated from mouse liver using an mRNA isolation kit (Invitrogen Co., San Diego, CA). The types of tissues analyzed from the indicated number of subjects included as follows: (i) Human tissues: 3 liver, 3 kidney, 3 lung, 1 heart, 6 breast, 2 uterus, 1 placenta, 2 stomach, 2 small intestine, 1 normal and 1 tumor colon samples from the same subject; (ii) mouse tissues: 4 liver, 5 kidney, 1 lung, 6 heart, 1 breast of a pregnant female, 1 uterus, 1 muscle, 1 brain, 1 spleen, 1 testis, 1 epididymis, 1 esophagus; and (iii) rat tissues: 1 liver, 2 kidney, 3 heart and cardiac myocytes, 3 esophagus, 2 stomach, 5 small intestine, 3 colon.

Ten pg RNA was resolved by formaldehyde-containing gels. Ran- dom primed human [32P]GSHPx-1 and [32P]GSHPx-GI cDNA and chicken [/j"32P]actin cDNA were used as probes. The blots were hybridized in a solution containing 20 mM sodium phosphate, pH 6.5, 5 X SSC, 33% formamide, 1 X Denhardt's reagent, 0.1% SDS, and radiolabeled probe with z lo6 cpm/ml at 52 "C for overnight. The

blots were washed in 0.1 x SSC and 0.1% SDS at room temperature for 15 min, followed by at 37 'C for 15 min, and at z 52 "C for 30-60 min with GSHPx-GI probe, or at 2 65 "C for 30-90 min with GSHPx- 1 probe. The exact washing conditions are indicated in the figure legends.

Selenoprotein Analysis by SDS-Polyacrylamide Gel Electrophore- sis-Cells were maintained in DMEM/F-12 (GIBCO) supplemented with 10% fetal bovine serum (Irvine Scientific Inc., Irvine, CA). Selenoproteins were labeled metabolically by growing cells in DMEM/F-12 medium containing 7-14 nM [76Se]selenious acid, 0.5% fetal bovine serum, 5 pg/liter insulin, and transferrin each for 1-2 weeks.

For analysis of selenoproteins and enzyme assays, cells were rinsed once with ice-cold phosphate-buffered saline (PBS), and then har- vested by scraping in PBS. After centrifugation at 1,000 rpm for 3 min, the cell pellet was washed once with homogenization buffer (buffer A) containing 10 mM Hepes, 3 mM M$+, and 5 mM EGTA at pH 7.4. The cell pellet was then resuspended in buffer A which was supplemented with a mixture of protease inhibitors containing 17 pg/ ml each of phenylmethylsulfonyl fluoride and benzamidine, and 5 fig/ ml each of leupeptin and pepstatin A at five times of the pellet volume. Cells were sonicated for 15 bursts at a control output of 6 and 80% of the duty cycle with a sonicator (Branson Sonifier, Cell Disruptor 200). The cytosol fraction was obtained as the supernatant after centrifuging the cell homogenate at 100,000 X g for 1 h. The pellet was washed once by resuspension in buffer A with protease inhibitors and sonication with five bursts at the same setting. After centrifuging at 15,000 X g for 30 min, the pellet was resuspended in buffer B containing 0.2 M each of NaPOl and NaCI, and 0.2% of Triton X-I00 after sonication with five bursts. After centrifugation at 15,000 X g for 30 min, this supernatant (containing 70% of the "Se label) was assayed as the organelle or membrane fraction.

The protein concentration of the cytosolic and organelle fractions was determined by the bicinchoninic acid method (BCA reagents, Pierce Chemical Co.). Bovine serum albumin was used as the stand- ard. During the initial screening for GSHPx-GI transfectants, the same amount of cytosolic and organelle or membrane proteins were analyzed by SDS-PAGE in 12% acrylamide and 0.37% bisacrylamide (29). The electrophoresis was performed as previously described (30).

Gel Filtration-To determine the native molecular weight of GSHPx-GI, the cytosolic fraction was chromatographed through a 45 X 1.5-cm Sephadex G-200 column at 4 "C in 0.1 M Tris (pH 7.2), 5 mM EDTA and 0.5 mM GSH. The eluted fractions were analyzed by SDS-PAGE. The column was calibrated with bovine serum albumin (68 kDa), carbonic anhydrase (29 kDa), and ribonuclease A (14.2 kDa) as the standards.

Subcellular Fractionation-Ten 2-liter roller bottles of confluent 75Se-labeled cells were rinsed in ice-cold PBS three times and then harvested by scraping. Cells were homogenized gently in buffer A with Dounce homogenizers. Cells were disrupted after 90 strokes as monitored by phase contrast microscopy. Cell homogenate was mixed at 1:l in volume with 0.5 M sucrose, 10 mM MgCL, and 5 mM EGTA, and then centrifuged at 500 X g for 10 min. After adding EDTA to the low speed supernatant to 10 mM, it was centrifuged at 10,000 X g for 10 min. The intermediate speed supernatant was centrifuged at 100,000 X g for 1 h to produce a high speed supernatant, which was used as the cytosolic fraction. The high speed pellet was washed once by resuspending in buffer A and pelleted at high speed again. This fraction was used as the microsomal fraction.

The intermediate speed pellet was resuspended in 10 mM Tris, pH 7.4,0.1% bovine serum albumin, and 0.8 M sucrose. This fraction was loaded onto a step sucrose gradient which was composed of 1,1.3,1.6, and 2 M sucrose. It was then centrifuged at 80,000 X g for 2 h. The enriched mitochondrial fraction at 1.2 M sucrose was recovered at the first interphase. The mitochondrial fraction was washed once by adding 2 volumes of 10 mM Tris, pH 7.4, and 1 mM EDTA and pelleted at 20,000 X g to remove bovine serum albumin and sucrose. The pellet was used as the mitochondrial fraction.

The low speed pellet was washed once by resuspending in buffer A and centrifuging at 500 X g for 10 min. This low speed pellet was resuspended in 20% glycerol, 3 mM MgC12, and 10 mM Hepes, pH 7.4, by Dounce homogenization, and then centrifuged at 500 X g for 10 min. The pellet was used as the nuclear fraction.

human RBC GSHPx-1 and human GSHPx-P were raised as described Immunoprecipitation and Enzyme Assays-Rabbit antisera against

previously (26). The IgG fraction was used for immunoprecipitation. Protein A (Staphylococcus aureus; ICN Immunobiologicals, Lisle, IL) was added to the immune complexes to enhance the precipitation as

Tissue-specific Expression of GSHPx-GI mRNA 2573

described. Glutathione peroxidase activity was measured by recording at ODsro the oxidation of NADPH in the presence of GSH, glutathione reductase, and substrates as described previously (30). Substrates tested included H202, tert-butyl hydroperoxide, cumene hydroperox- ide, linoleic acid hydroperoxide, and phosphatidylcholine hydroper- oxide.

RESULTS

The primary nucleic acid sequence of our full length GSHPx-GI cDNA, and its translated 190 amino acids are shown in Fig. 1. It has a UGA codon for selenocysteine at nucleotide positions 152-154. As shown in the figure, sequence discrepancies were found at five locations between our full length HepG2 clone and the sequence published by Akasaka et al., and at four locations between our HepGZ clone and our liver clone. The only place that all three clones have different sequences is at nucleotides 143-145, which results in three different codons.

We have transfected the full length HepG2 GSHPx-GI cDNA in pRSV, a mammalian expression vector, into a hu- man mammary carcinoma cell line, MCF-7 cells. Either pSV2- neo or pSK1-mdr were cotransfected to enable the selection for transfectants. Sixteen transfectants which were resistant to 400 pg/ml G418, and six transfectants which were resistant to 8 ng/ml colchicine were isolated as individual clones. All of these transfectants were analyzed by SDS-PAGE for [7sSe] selenoproteins, Northern and Southern analyses. One G418- resistant neo-Dl clone and one colchicine-resistant mdr-5 clone were found expressing the transgenic GSHPx-GI cDNA.

Because it is crucial to identify the protein product of GSHPx-GI, we have analyzed 7sSe-labeled cytosolic proteins from the transfected cells by SDS-PAGE. Each lane was

~~~~~~CTCTCQ~QC(IETCCCATQ a c ~ TTC ATT acc AM TCC TTC TAT aAc 64 W A I I A K S I Y D

nC ADT QCC ATC MC OAT UiG O M AM OTA OAT TTC M T ACQ TTC Coo OOC 118 L S A I S L D O L K V D I N T I R O

*CQ (R) I-IL)

A f f i QCC OTQ A l T O M M T OTQ GCT TOO CPc TQA WC ACA ACC ACC Coo OAC 172 R A V L I L N V A S L ~ O T T T R D

TTC ACC CM CTC MC OM cm CM nw: coc m ccc ma cac cm a m ax: CTT 226 I T Q L N L L Q C R I P R R L V V L

aac TTC ccr nw: MC CM m QUA CAT CM a m MC TQT CM MT a m om ATC 280 *AIS)

O I P C N Q I ~ H Q L N C Q N E L I

cm MC MT CTC AM TAT am COT ccr aaa QUT OOA TAC ma ccc ACC TTC ACC 334 L N S L K Y V R P O O O Y Q P T I T

CTT 0% CM AM TQT O M OTQ M T Ow CM M C O M CAT CCT QTC TTC QCC TAC 388 L V Q K C E V N O Q N L H P V I A Y

cm MG aAc AM crc ccc TAC CCT TAT a m aAc CCA m TCC CTC ma ACC QAT 442 L K D K L P Y P Y D D P I S L W T D

DX A M Cl'C ATC A m TOQ MC CCT QTO CQC CQC TCA OAT OTO aCC Too M C l T T 496 P K L I I W S P V R R S D V A N N I

om M a TTC CTC ATA aaa cca a m QUA a= ccc TTC CQA ox TAC MC ox ACC 550 L K I L I G P L O L P I R R Y S R T

TTC CCA ACC ATC MC ATT am CCT aAc ATC AM cac crc CTT AM am acc ATA 604 M(W

I P T I N I E P D I K R L L K V A I

=CA TMATQTQM IXWTCMCA CACAGATCTC CTA-TC CAQTCCFOM aaaccrcma ATUCA 669

GCATQ ccTTc"i A C A c m c T o o A-T T C C C T X A T A TC" T C W T W " 734

OECTTOCCTT CTOTlTCfTT MOXTCTG4 TXGTSATTC MCTT 199

OooCF " QUT- 7 MAATSATGO CAOXTCCTA MCCCXATQ 864

c-

*T OGTGOTQTCF awwxcam ~UOOOOCMO AOCCA- c m a m a w a EMT- aaaa 929

I"" .""

TaTaQ AMcaacAM AAA"M "Ma AM" 911

FIG. 1. The primary sequence of full length human GSHPx- GI cDNA clone isolated from HepG2 library. The discrepant sequences from the published sequence are indicated by an asterisk (19), those from our liver clone are indicated by #. The dash indicates deletion. The amino acids encoded bv the discreDant seauences are indicated in the parentheses.

loaded with 50 pg of protein. As shown in Fig. 2, a 22-kDa [7sSe]selenopeptide was found in the cytosols of neo-Dl and mdr-5 cells, but not in other drug resistant transfectants or the parental MCF-7 cells (lanes 1-11). Cytosolic proteins isolated from HepG2 cells which express endogenous GSHPx- GI and GSHPx-1 genes were resolved in lane 13; and cytosolic proteins isolated from MCF-7H6 cells which express a trans- fected human GSHPx-1 cDNA were resolved in lane 14. GSHPx-GI selenopeptide has a mobility similar to GSHPx-1 during electrophoresis. Furthermore, GSHPx-GI was not de- tected in the total particulate fraction of any of the cells tested (data not shown).

To demonstrate the specificity of our GSHPx-GI cDNA probe, and that neo-Dl expresses the transfected GSHPx-GI, we analyzed mRNA isolated from neo-Dl, MCF-7, HepG2, and MCF-7H6 cells in Fig. 3. Shown in A, a weak signal of endogenous 1.0-kb GSHPx-GI mRNA was detected in MCF- 7 and MCF-7H6 cells. The neo-Dl cells expressed a 1.9-kb mRNA and a stronger signal corresponding to the endogenous mRNA hybridized to our GSHPx-GI probe. The same blot was subsequently probed with GSHPx-1 cDNA after being stripped off the GSHPx-GI probe at 80 "C for 1 h. As shown in B, no endogenous GSHPx-1 mRNA was detected in neo- D l and MCF-7 cells.

Southern analysis was performed to illustrate the presence of transfected GSHPx-GI in addition to the endogenous gene. Shown in Fig. 4A, a single 15-kb band hybridized with GSHPx-GI probe in MCF-7 DNA which was digested with EcoRI, an additional 1.0-kb cDNA was detected in EcoRI digested neo-Dl DNA. This 1.0-kb band corresponds to that of the GSHPx-GI cDNA, since its sequence is flanked by the EcoRI site in pRSV-GSHPx-GI. B shows the same blot being reprobed with GSHPx-1 cDNA. Similar to the Northern data,

123456789lO11121314 kDa - 57

-29 - 22 - 18 - 14

FIG. 2. Autoradiography of metabolically labeled ['6Se]pro- teins resolved by SDS-polyacrylamide gel electrophoresis. Cytosolic fractions were obtained from MCF-7 transfectants neo-A1 (lane I ) , neo-B2 (lane 2), neo-Dl (lane 3 ) , neo-E2 (lane 4 ) , neo-E6 (lane 5), mdr-4 (lane 6), mdr-5 (lane 7), mdr-6 (lane 8), nontrans- fected MCF-7 cell (lane 9), neo-7 (lane IO), neo-8 (lane 11) . Other control lanes contain molecular weight markers (lane 12), HepG2 (lane 13), and MCF-7H6 (lane 14) homogenates.

A 1 2 3 4 1 2 3 4 0

kb

4.7 2.5

1.5 - 1.0

J .9

FIG. 3. Northern analysis of mRNA isolated from transfec- tants and wild-type MCF-7 cells. A shows a blot probed with GSHPx-GI cDNA; it was washed at 65 "C for 30 min at 0.1 X SSC. B shows a blot probed with GSHPx-1 cDNA; it was washed at 69 "C for 60 min. Total RNA was isolated from MCF-7H6 (lane 1 ), MCF- 7 (lane 2), neo-Dl (lane 3 ) , and HepG2 (lane 4 ) .

2574 Tissue-specific Expression of GSHPx-GI mRNA

FIG. 4. Southern analysis of EcoRI-digested DNA isolated from MCF-7 and neo-Dl cells. A shows a blot probed with GSHPx-GI cDNA; it was washed at 65 'C for 60 min at 0.1 X SSC. R shows a blot probed with GSHPx-1 cDNA; it was washed at 65 "C for 30 min. Multiple bands hybridized to GSHPx-1 are due to the presence of possible pseudogenes located in chromosome 21 and X (35).

0-

d R

A 1 2 3 4 5 4 2 3 4 5 kDa

-57'

-l18. -22.

FIG. 5. Autoradiography of '%e-labeled proteins from dif- ferent subcellular fractions of transfected MCF-7 cells re- solved by SDS-PAGE. A shows the subcellular fractions of neo-Dl cells, and B shows those of MCF-7H6 cells. Each panel contains total cellular homogenate (lane I ), cytosol (lane 2 ) , mitochondria (lane 3), nuclei (lane 4 ) , and microsomes (lane 5).

GSHPx-1 and GSHPx-GI probes do not cross-hybridize in Southern analysis.

The subcellular localization of GSHPx-GI was analyzed by SDS-PAGE of [75Se]selenopeptides in neo-Dl cells. MCF- 7H6 cells overexpressing GSHPx-1 were also analyzed as a control. Differential centrifugation was used to separate four major subcellular fractions. As shown in Fig. 5, the 75Se- labeled proteins in cytosolic, mitochondrial, nuclear, and mi- crosomal fractions of neo-Dl and MCF-7H6 were resolved by SDS-PAGE. Both GSHPx-1 and GSHPx-GI are mainly cy- tosolic proteins. A 20-kDa selenopeptide shown in lanes 4 and 5 of A of the membrane fractions appears to be an endogenous protein which is present in all human cell lines that we have analyzed. This selenopeptide is often overshadowed by the 22-kDa polypeptide in the membrane fractions as occurred in B.

GSHPx-GI was not recognized by antisera raised against human erythrocyte GSHPx-1 and human GSHPx-P (Fig. 6). We have performed immunoprecipitation of total 75Se-labeled cytosolic proteins isolated from neo-Dl with both antisera, and did not precipitate any of the selenoproteins. When total 75Se-labeled cytosolic proteins isolated from MCF-7H6 cells were studied, a 22-kDa selenoprotein was precipitated by anti- RBC GSHPx-1 antisera. Concomitantly, enzymatic activity was removed from the cytosol. This was a positive control for the immunoprecipitation reactions.

The multimeric nature of GSHPx-GI was studied by gel filtration of cytosolic proteins obtained from neo-Dl cells. Similar results were obtained in three gel filtration studies. Fig. 7 shows the pattern of [75Se]selenopeptide fractionated from a Sephadex G-200 column. A shows the autoradiography of selenoproteins in neo-Dl cells resolved from this column,

* 1 2 3 4 ' 1 2 3 4 kDo

-66

- 29 - 22 - 14

FIG. 6. Immunoprecipitation of GSHPx-GI protein by anti- sera against human RBC GSHPx-1 or GSHPx-P. A shows an autoradiograph of 7sSe-labeled proteins present in the postimmuno- precipitated supernatant resolved by SDS-PAGE. B shows an auto- radiograph of 75Se-Iabeled proteins present in the immunoprecipitates resolved by SDS-PAGE. Lanes 1 and 2 are 75Se-labeled cytosolic proteins isolated from MCF-7H6 cells and lanes 3 and 4 are seleno- proteins obtained from the cytosol of neo-Dl cells. Lanes I and 3 were treated with anti-human GSHPx-P antisera, and lanes 2 and 4 were treated with anti-human RBC GSHPx-1 antisera.

A m ~141618202224262830323436384042444648

kDa

57 -

22- 18- 14-

B $141613 202zz426za302%~3.4404244

kDa

57 -

22 - 18- 14-

FIG. 7. Autoradiography of 76Se-labeled cytosolic proteins eluted from a Sephadex G-200 column resolved by SDS- PAGE. A shows the pattern of 6 mg of neo-Dl cytosolic protein fractionated from the sizing column, and B shows the pattern of 5.6 mg of MCF-7H6 cytosolic protein fractionated from the same sizing column. The pattern of MCF-7 cytosolic fractions from the same column has been shown previously (30).

whereas B shows the autoradiography of selenoproteins in MCF-7H6 cells resolved from the same column. The native apparent molecular weight of GSHPx-GI is estimated to be 75 f 2 kDa, which is smaller than the 88-kDa tetramer of GSHPx-1. GSHPx-GI is most likely also a tetramer. The smaller size of a native GSHPx-GI tetramer is predicted from its cDNA sequence which codes for 190 amino acids, whereas GSHPx-1 cDNA codes for 201 amino acids (13). Although a similar amount of cytosolic protein was loaded on the column, neo-Dl appears to express much less GSHPx-GI than MCF- 7H6 expresses GSHPx-1, as shown in Fig. 2; the autoradiog- raphy in A was exposed three times longer than that in B to compensate for the difference.

The relative substrate specificities of GSHPx-GI are pre- sented in Table I. Enzymatic activity was assayed from the pooled GSHPx-GI fraction of a Sephadex G-200 column. The background activity, determined by the activities obtained from the same fractions of parental MCF-7 cells, was sub- tracted on a per mg of protein basis from the specific activity obtained in both neo-Dl and MCF-7H6 cells. The small peroxidase activity toward lipid hydroperoxides detected in the MCF-7 sample may be derived from an endogenous glu- tathione transferase with peroxidase activity which eluted at fractions 32-35 and would not fully resolve from GSHPx-GI (30). This transferase activity does not react with H202. The

Tissue-specific Expression of GSHPx-GI mRNA 2575

TABLE I Enzymatic activities determined from enriched enzyme preparations expressed as nmole X min" X mg" f S.D.

Two cytosol samples assayed in duplicate were determined for neo-Dl and MCF-7H6 cells. MCF-7 cytosol was done once for this study and agree with the results from prior studies (30). The samples were obtained by pooling 22-kDa 75Se-labeled polypeptide-containing fractions eluted from a Sephadex G-200 sizing column. The specific activities obtained from nontransfected MCF-7 cells were subtracted from those obtained from neo-Dl and MCF-7H6 cells. The substrate concentrations used were: H202 a t 60 p ~ , tert-butyl hydroperoxide a t 60 p M without Triton X-100, cumene hydroperoxide a t 150 PM without Triton X-100, linoleic acid hydroperoxide at 24 PM without Triton X-100, and phosphatidylcholine hydroperoxide a t 25 p~ with 0.1% Triton X-100.

Substrate Enzyme source

HzOz tert-butyl Cumene Linoleic acid Phospholipid hydroperoxide hydroperoxide hydroperoxide hydroperoxide

MCF-7 0.0 1.5 6.7 f 1.0 8.1 f 0.2 0.0 neo-Dl MCF-7H6

302 f 21 201 f 15 231 f 11 225 f 10 1.6 rt 0.7 2084 f 88 638 f 43 908 f 44 1187 f 11 4.0 f 2.0

A 1 2 3 4 5 6

28s rRNA-

18s rRNA-

GSHPx-1- * 9 ' r i F m

' 1 2 3 4 5 6

28s rRNA-

18s rRNA-

GSHPx-GI-

FIG. 8. Northern analysis of GSHPx-GI mRNA in five hu- man tissues and mouse liver. Poly(A)-containing mRNAs were isolated from mouse liver (lane I ) ; total RNAs were isolated from five human tissues: liver (lane 2 ) , heart (lane 3 ) , kidney (lane 4 ) , colon (lane 5), and breast (lane 6). A shows a blot hybridized to human GSHPx-1 cDNA; it was washed a t 69 "C for 60 min. B shows a blot hybridized to human GSHPx-GI cDNA; i t was washed a t 52 "C for 60 min.

A 1 2 3 4 5 6 7 8 9 1 0

28s rRNA -

GSHPx-1 -

' 1 2 3 4 5 6 7 8 9 1 0

* :.x - .

28s rRNA -

GSHPrGI -

FIG. 9. Northern analysis of GSHPx-GI mRNA in human cell lines, human breast tissue, and rat gastrointestinal tis- sues. Total RNAs were isolated from human breast cell lines: MCF- 10A (lane I ) , DU4475 (lane 2) , MDA-MB175 (lune 3 ) ; normal and tumorous human breast tissue (lanes 4 and 5); four rat tissues: esophagus (lane 6), stomach (lane 7), small intestine (lane 8), and colon (lane 9); and neo-Dl cells (lane IO). A shows the blot hybridized t o human GSHPx-1 cDNA, i t was washed at 65 "C for 90 min. B shows the blot hybridized to human GSHPx-GI cDNA; i t was washed at 52 "C for 30 min.

enzyme assays were performed twice from two independent preparations.

We have studied the expression of GSHPx-GI mRNA in normal tissues. Shown in Fig. 8, GSHPx-GI mRNA was detected in human liver and colon, but not in human heart or kidney, and mouse liver. Other human tissues that expressed GSHPx-GI mRNA included stomach, small intestine, and one breast sample out of six studied. Human tissues analyzed that did not have detectable GSHPx-GI mRNA were uterus, placenta, and lung (data not shown). In rodent, GSHPx-GI mRNA was only detectable in gastrointestinal tissues, includ- ing rat esophagus, stomach, small intestine and colon, as well as mouse esophagus (Fig. 9 and data not shown). Other rodent tissues studied without detectable GSHPx-GI mRNA were rat liver, skeletal muscle, heart, kidney, and mouse kidney, lung, heart, spleen, brain, skeletal muscle, testis, epididymis, uterus, and mammary gland (Fig. 9 and data not shown). Although the GSHPx-GI-probed blots were washed at 52 "C in these figures, the same results were obtained from blots washed at the more stringent temperatures of 58 "C with rodent tissues and 65 "C with human tissues. No GSHPx-1 mRNA was detected in rat GI tract as well as human MDA- MB175 and neo-Dl cell lines, even though the blot was washed at a less stringent condition which did not eliminate the nonspecific binding to 28 S rRNA (Fig. 9A).

We have also analyzed several cell lines for GSHPx-GI mRNA expression. It was detected in two human cell lines: HepG2 hepatoma cells and DU4475 breast cancer cells (Fig. 9). MDA-MB175, a human breast cancer cell line, had a weak signal in another Northern blot (data not shown). The nega- tive cell lines included MCF-1OA normal human breast cells, AdfMCF-7, adriamycin-resistant breast cancer cells, Hep3B human hepatoma cells, mouse fibroblast L cells, and HA-1 Chinese hamster ovary fibroblasts (Fig. 9 and data not shown).

DISCUSSION

Although the GSHPx-GI cDNA clones were isolated with a 69-base oligonucleotide of bovine GSHPx-1 as the probe, the sequence homology of human GSHPx-1 and human GSHPx-GI in their coding regions is 66%; there is little homology at the 3' nontranslated region, thus no cross-hy- bridization of human GSHPx-1 and human GSHPx-GI se- quences was observed when full length cDNA probes were used.

Although GSHPx-GI mRNA is expressed in human liver and HepG2 cells, both also express GSHPx-1 which has similar biochemical properties to GSHPx-GI. This makes the analysis of GSHPx-GI protein from these sources difficult. In order to analyze GSHPx-GI protein, we have transfected and expressed a pRSV-GSHPx-GI in human mammary carcinoma

2576 Tissue-specific Expression of GSHPx-GI mRNA

MCF-7 cells. Although these cells have endogenous GSHPx- GI mRNA, it is difficult to detect any 22-kDa selenopeptide or glutathione peroxidase activity (30). These cells express PHGPX, but this can be easily separated from native GSHPx- GI by gel filtration. PHGPX does not reduce H202 efficiently (30). The GSHPx-GI transfectants, neo-Dl and mdr-5, ex- pressed a 1.9-kb GSHPx-GI mRNA which may represent a hybrid mRNA containing GSHPx-GI and SV40 polyadenyl- ation sequences in the pRSV-GSHPx-GI construct (31). This was also observed in MCF-7H6 expressing pRSV-GSHPx-1 plasmid (26). A 22-kDa selenopeptide resolved by SDS-PAGE was detected in neo-Dl and mdr-5 cells. This selenoprotein has glutathione peroxidase activity. These results provide the first evidence suggesting that GSHPx-GI mRNA is not the product of a pseudogene.

The biochemical analyses performed on neo-Dl cells reveal that GSHPx-GI is a tetrameric cytosolic protein composed of 22-kDa monomers, similar to GSHPx-1. None of the partic- ulate fractions including nuclei, mitochondria, and micro- somes have enriched levels of GSHPx-GI selenoprotein. Im- munologically, the human recombinant GSHPx-GI did not react with anti-GSHPx-1 or anti-GSHPx-P antisera. Al- though GSHPx-GI has a similar spectrum of substrates as GSHPx-1, it has relatively higher reactivity toward organic hydroperoxides when their enzyme activities were normalized against H202-metabolizing activity. It may be more important for tissues of the gastrointestinal tract to express GSHPx-GI, because dietary hydroperoxides are predominantly in an or- ganic form (32). Whereas it is more important for erythrocytes t o express GSHPx-1, because H202 is a byproduct of hemo- globin metabolism (33). Patients with trisomy 21 have 50% higher glutathione peroxidase activities than the normal pop- ulation (34). The functional human GSHPx-1 gene has been mapped to chromosome 3 (35, 36). We determined whether chromosome 21 contained the GSHPx-GI gene. Only a mouse GSHPx-GI gene was detected in mouse-human somatic hy- brid cells containing human chromosome 21 (data not shown). This indicates that GSHPx-GI is not localized in chromosome 21, and leaves the observation of elevated enzyme activity unexplained.

Since GSHPx-GI mRNA is only detected in human and rodent GI tract in addition to human liver, we name it GSHPx-GI. Human liver expresses a t least four isozymes of glutathione peroxidase, namely GSHPx-1, GSHPx-GI, GSHPx-P, and PHGPX, whereas rodent liver only expresses GSHPx-1 and PHGPX. Rodent liver has 10 times higher selenium-dependent glutathione peroxidase activity than hu- man liver. Rodent and human have opted for two different modes of GSHPx expression in this tissue. Speculation on the extraordinary level of GSHPx-1 in rodent liver includes the notion that it is a storage form of selenium, which is sacrificed when dietary selenium is lacking (37). The expres- sion of GSHPx-GI and GSHPx-P in human liver may provide extra protection to fit the specific requirements of human metabolism, including possible constraints in selenium utili- zation or responses to selenium deficiency.

Glutathione peroxidase activity has been reported in the rat gastrointestinal tract based on assays with cumene hydro- peroxide, which can also be metabolized by some glutathione transferases (30). We have analyzed glutathione peroxidase activities with H202 in rat tissues, and found the highest

activity in stomach (1.2 pmol X min" X mg") followed by decreasing activities in esophagus (0.8 pmol x min-' x mg-'), colon (0.4 pmol X min-' X mg"), and intestine (0.3 pmol X min" X mg"), whereas rat liver has an activity of 2.3-2.7 pmol X min-' x mg". This activity found in the gastrointes- tinal tract appears to be contributed mainly by GSHPx-GI, since GSHPx-1 mRNA was not detectable in these tissues.

In conclusion, we have demonstrated the presence of a new isozyme of glutathione peroxidase, which was previously only known in the form of a cDNA. Based on the biochemical, immunological, molecular, and enzymatic analyses, we pro- pose the newly characterized GSHPx-GI to be the fourth isozyme of the selenium-dependent glutathione peroxidase family.

REFERENCES 1. Rotruck, J. T., Pope, A. L., Ganther, H. E., Swanson, A. B., Hafeman, D.

2. Maiorino, M., Gregolin, C. and Ursini, F. (1990) Methods Enzymol. 186,

3. Maddipati, K. R., Gasparski, C., and Marnett, L. J. (1987) Arch. Biochem.

4. Takahashi, K., Avissar, N., Whitin, J., and Cohen, H. (1987) Arch. Biochem.

5 . Esworthy, R. S., Chu, F.-F., Paxton, R. J., Akman, S., and Doroshow, J. H.

6. Thomas, J. P., Maiorino, M., Ursini, F., and Girotti, A. W. (1990) J. Biol. (1991) Arch. Biochem. Biophys. 286,330-336

Chem. 265,454-461 7. Chambers, I., and Harrison, P. (1988) in Oxy-Radicals in Molecular Biology

and Pathology, pp. 289-300, Alan R. Liss Inc., New York 8. Roveri, A,, Casasco, A., Maiorino, M., Dalan, P., Calligaro, A,, and Ursini,

F. (1992) J. Bid . Chem. 267,6142-6146 9. Yoshimura, S., Watanabe, K., Suemizu, H., Onozawa, T., Mizoguchi, J.,

Tsuda, K., Hatta, H., and Moriuchi, T. (1991) J. Biochern. (Tokyo) 109 , 918-923

10. Avissar. N.. Slemmon. J. R., Palmer, I. S., and Cohen, H. J. (1991) J. Nutr.

G. and Hoekstra, W. G. (1973) Science 179,588-590

448-457

Biophys. 254,9-17

Bzophys. 256,677-686

11.

12.

13.

14.

15.

121,'1243-1249

(1992) Blood 79,3233-3238

. .

Chu, F.-F., Esworthy, R. S., Doroshow, J. H., Doan, K., and Liu, X.-F.

Chambers. I.. FramDton. J.. Goldfarb. P.. Affara. N.. McBain. W.. and . ~~~ ~~~ ~

Harrisoh, P . R . (1'986) : $ v ~ O J. 5 , 1221-1227 Mullenbach, G. T., Tabnzl, A., Irvine, B. D., Bell, G. I., and Hallewell, R.

A. (1987) Nucleic Aczds Res. 15,5484 Takahashi K. Akasaka, M., Yamamoto Y. Kobayashi, C., Mizoguchi, J.,

and Koiam;, J. (1990) J. Biochem. (TbkGo) 108 , 145-148 Schuckelt, R., Brigelius-FlohB, R., Maiorino, M., Roveri, A., Reumkens, J.,

StraBburcer. W.. Ursini. F.. Wolf. B.. and FlohB, L. (1991) Free Rad. Res.

. . , .

Commun: 14,343-361 . . . .

16. Ghyselinck, N. B., and Dufaure, J.-P. (1990) Nucleic Acids Res. 18 , 7144 17. Faure, E., Ghyselinck, N. B., Jimenez, C., and Dufaure, J. P. (1991) Biol. 18. Chu, F.-F., Akman, S. A., and Doroshow, J. H. (1988) Proc. Am. Assoc.

19. Akasaka. M.. Mizomchi. J., and Takahashi, K. (1990) Nucleic Acids Res.

Reprod. 44; 13-22

Cancer Res. 2 9 , 440 (abstr.) - .

15,4619 20. Giinzler, W. A., Steffens, G. J., Grossmann, A., Kim, S.-M. A., Otting, F.,

Wendel, A,, and FlohB, L. (1984) Hoppe-Seyler's 2. Physiol. Chern. 365 ,

21. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) Proc. Natl. Acad. Sci. 195-212

22. Mulligan, R. C., Howard, B. H., and Berg, P. (1979) Nature 277 , 108-114 (I. S. A. 74,5463-5467

23. Gorman, C. M., Merlino, G. T., Willin ham, M. C., Pastan, I., and Howard, B. H. (1982) Proc. Natl. Acad. Sci. 8. S. A. 79,6777-6781

24. Southern, P. J., and Berg, P. (1982) J. Mol. Appl. Genet. 1,327-341 25. Kane S. E., Reinhard, D. H., Fordis, C. M., Pastan, I., and Gottesman, M.

26. Chu, F.-F., Esworthy, R. S., Akman, S., and Doroshow, J. H. (1990) Nucleic M.'(1989) Gene (Amst.) 84,439-446

27. Hill, R. E., Show, P. H., Barth, R. K., and Hastie, N. D. (1985) Mol. Cell. Acids Res. 18,1531-1539

28. Chomczynski, P., and Sacchi, N. (1987) Anal. Biochem. 162, 156-159 Biol. 5,2144-2122

29. Laemmli, U. K. (1970) Nature 227,680-685 30. Maiorino, M., Chu, F.-F., Ursini, F., Davies, K. J. A., Doroshow, J. H., and

31. Nielsen, D. A. Chou J., MacKrell, A. J., Casadaban, M. J., and Steiner, D.

32. Vilas, N. N., Bell, R. R.! and Draper, H. H. (1976) J. Nutr. 106,589-596 33. Mills, G. C. (1957) J. Bzol. Chem. 229,189-197 34. Sinet, P. M., Michelson, A. M., Bazin, A., Lejeune, J., and Jerome, H.

35. McBride, 0. W., Mitchell, A., Lee, J., Mullenbach, G., and Hatfield, D.

36. Chada, S., Le Beau, M. M., Casey, L., and Newburger, P. E. (1990) Genomics

37. Burk, R. F. (1991) FASEB J. 5,2274-2279

Eswortby, R. S. (1991) J. Biol. Chern. 2 6 6 , 7728-7732

F. (1983) P;oc. Ndtl. Acad. Sci. U. S. A. 80,5198-5202

(1975) Biochem. Blophys. Res. Common. 67,910-915

(1988) BioFactors 1,285-292

6,268-271