Communication Vol. 264, No. 24, Issue of August 25, pp. 13967-13970,1989 THE JOURNAL OF BIOLOGICAL CHEMISTRY

Printed m U, S. A.

Selective Inhibition of Protein Disulfide Isomerase by Estrogens*

(Received for publication, December 29, 1988)

John C. M. TsibrisSP, Lois T. Huntn, Gustavo BallejoS, Winona C. Barkern, Leighton J. ToneyS, and William N. SpellacySP From the $Department of Obstetrics and Gynecology, University of Illinois College of Medicine, Chicago, Illinois 60612 and the llNational Biomedical Research Foundation, Washington, D. C. 20007

Protein disulfide isomerase (PDI) is a multifunc- tional microsomal enzyme that participates in the for- mation of protein disulfide bonds. PDI catalyzes the reduction of protein disulfide bonds in the presence of excess reduced glutathione and has been implicated in the reductive degradation of insulin; E. coli thiore- doxin is homologous to two regions in PDI and can also degrade insulin. PDI activity, measured by lZSI-insulin degradation or reactivation of randomly oxidized RNase in the presence of reduced glutathione, is non- competitively inhibited by estrogens; half-maximal in- hibition was observed at approximately 100 nM estro- gen. Other steroid hormones at 1 PM had little or no effect. PDI segment 120-163 (which corresponds to exon 3 of the PDI gene) and 182-230 have significant similarity with estrogen receptor segments 350-392 and 304-349, respectively, located in the estrogen binding domain but not with the steroid domains of the progesterone and glucocorticoid receptors or with thio- redoxin, which is insensitive to estrogens. We propose the hypothesis that enzymes can acquire sensitivity to a hormone via exon shuffling to the enzyme gene from the DNA region coding for the hormone binding do- main of the hormone’s receptor.

Protein disulfide isomerase (PDI’; EC 5.3.4.1) catalyzes thiol-disulfide exchange reactions in various proteins, e.g. ribonuclease, procollagen, prolactin, and immunoglobulins (1- 3). PDI and thioredoxin can also catalyze the reductive deg- radation of insulin by GSH (4, 5). PDI (6) also acts as the /3 subunit of prolyl4-hydroxylase (7) , as a component of the N - glycosyltransferase complex (8), and as a membrane-bound thyroid hormone binding protein (9, 10) with an anticipated 5”deiodinase activity (11). While measuring insulin binding and degradation by human placenta we titrated a membrane preparation with various steroids and found that only estro-

*This work was supported in part by Grants RR01821 and CA40474 from the National Institutes of Health (to L. T. H. and W. C. B.) and by a Biotechnology Fellowship from the Rockefeller Foundation (to G. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

S Present address: Dept. of Obstetrics and Gynecology, University of South Florida, Harbour Side Medical Tower, 4 Columbia Dr., Tampa, FL 33606.

The abbreviations used are: PDI, protein disulfide isomerase; GSH, reduced glutathione; HPLC, high performance liquid chroma- tography; ER, estrogen receptor.

gens increased the specific binding of insulin’ and decreased insulin degradation. Subsequently, chromatography of placen- tal cytosol on a DEAE-vinyl polymer separated GSH-depend- ent insulin degrading activity from proteolytic (GSH-inde- pendent) activity, and it became apparent that only the former was inhibited by estrogens. In this paper, we characterize the inhibition by estrogens of ‘251-insulin degradation or the reac- tivation of “scrambled” RNase catalyzed by a cytosolic and a membrane PDI preparation from human placenta and by purified PDI solubilized from ox liver (1).

EXPERIMENTAL PROCEDURES

P D I Preparations

Normal human term placentae were used (90 g of rinsed tissue/ 600 ml of 0.3 M sucrose) to prepare the cytosolic (12) and plasma membrane/microsomal (particulate) fractions by differential centrif- ugation, as described for endometrium (13); the 100,000 X g pellets were further fractionated on a 30-60% sucrose gradient. The crude cytosol, 500 ml, was applied on a 2.5 X 15-cm Fractogel TSK DEAE- 650s column and PDI activity (trichloroacetic acid assay) was eluted after the GSHindependent ’251-insulin-degrading activity with a lin- ear 0-0.4 M NaCl gradient. Pooled PDI fractions were lyophilized, dialyzed, and are referred to as cytosol. Purified ox liver PDI was purchased from Genzyme Ltd, Boston, MA.

P D I Assays 125 I-Insulin Degradation-Incubations were carried out in duplicate

in 0.25 ml containing 0.05 M sodium phosphate, pH 8, 2 pl of steroid solution in methanol, particulate (0.15 mg) or cytosol (0.07 mg) PDI, 2 mM EDTA, 1 mM GSH, and 0.2-0.6 ng of A14 (‘251-monoiodinated porcine insulin at tyrosine A14, 380 Ci/g), purified by HPLC. After 5 min at 37 “C and then cooling to 0 “C, 4 mM iodoacetate was added for 15 min. Particulate incubation mixtures were spun at 140,000 X g for 5 min; insulin degradation was measured in the resulting supernatant or in the cytosol incubations by precipitation with 5% trichloroacetic acid. Reverse-phase HPLC on a pBondapak C,, col- umn and a 0.25 M triethylamine formate, pH 6, 50% acetonitrile convex gradient, with the Perkin-Elmer Ti-410 system, was also used to assay A14 degradation and two metabolite peaks were eluted ahead of intact A14; only 3-5% free iodide was observed in all incubations, which excluded tracer deiodination. A 3-6% blank (enzyme, GSH, EDTA, A14 a t 0 “ C ) has been subtracted from all trichloroacetic acid degradation values; net A14 degradation was kept below 20% to assure assay linearity. Trichloroacetic acid assays underestimated A14 deg- radation by a factor of 3.6, compared to reverse-phase HPLC, but gave similar percent inhibition by estrogens.

RNase Reactions-Randomly oxidized RNase A (35 pg, Sigma) (14) was incubated for 30 min at 37 “C in 1 ml containing 0.2 mg of cytosolic PDI, 2 mM GSH, 0.2 mM GSSG, 1 mM EDTA, 2 p1 of estradiol in methanol, and 0.1 M Tris-C1, pH 7.9; RNase was assayed (15) by the hydrolysis of cytidine 2’:3’-cyclic phosphate and a 30% blank (reactivation in the absence of PDI) was substracted from all values.

[3H] 17B-Estradiol Binding

[6,7-3H]Estradioi (Du Pont-New England Nuclear) binding was determined by equilibrium dialysis (16), 0.07-0.1 mg of cytosol incu- bated for 22 h at 23 “C, or by incubation of 0.3 mg of membrane PDI/ ml in 10 mM Tris-C1, pH 7.4, for 1 h at 26 “C, followed by centrifu- gation at 140,000 X g for 5 min and determination of the radioactivity in the pellet (bound [‘H]estradiol).

’ Tsibris, J., Saleem, T., Banuelos, G., Ballejo, G., Sheets, E., and Spellacy, W. (1983) Society for Gynecologic Inuestigation, 30th Annual Meeting, March 17-20, Washington, D. C., Abstr, 319.

13967

13968 Estrogens and Protein Disulfide Isomerase Cross-linking of '251-Insulin

The three preparations (Fig. 3) gave 10-12% A14 degradation by the trichloroacetic acid method. A14 was cross-linked (17) with 0.3 mM disuccinimidyl suberate at 0 "C, in the absence of GSH/EDTA, and with or without 1 p M 17P-estradiol; inclusion of 1 NM insulin inhibited the labeling of all bands (data not shown). Electrophoresis was on a 10% polyacrylamide-sodium dodecyl sulfate gel followed by autoradiography.

RESULTS AND DISCUSSION

The steroid specificity of the inhibition of PDI (Fig. 1) is intriguing. At 1 p~ concentration only the most potent estro- gens showed a 40-60% inhibition of insulin degradation by human placental cytosol and 20-30% by the membrane prep- arations. 17a-Estradiol, a much weaker estrogen, inhibited insulin degradation to a lesser extent than 17P-estradiol. 3- Methoxyestriol had little effect, emphasizing the importance of the free phenol group of estrogens. Two tamoxifen isomers (Fig. 1) showed little inhibition of the placental preparations and had no effect on ox liver PDI. Moreover, trans-tamoxifen did not antagonize and cis-tamoxifen did not enhance the effect of 178-estradiol (Fig. l), as was observed with rat uterine estrogen receptors (18). When the RNase PDI assay (15) was used with a cytosolic placental PDI, similar steroid specificity was observed but 17a-estradiol and truns-tamoxi- fen were totally inactive. Maximal estrogen effect was ob- served within 5-15 min in both PDI assays.

The concentration dependence of the estrogen effect on insulin degradation was determined with 17P-estradiol and estrone (Fig. 2, bottom and middle panels). The inhibition is saturable at concentrations above 1 p~ although an inflection point is apparent (Fig. 2, middle panel) at approximately 200 nM. The binding of [3H]estradiol to PDI also shows a similar inflection point around 200 nM (Fig. 2, toppanel), as had been observed earlier with the binding of estrogens to rat liver microsomes (18); however, at higher concentrations the bind- ing of estradiol was not saturable either in their liver micro- somal (19) or both of our placental PDI preparations. The estrogen concentrations used in this study are physiologic for developing follicles in the human ovary (3-5 pM, Ref. 20) and human term pregnancy; in placenta free estriol, 17P-estradiol and estrone content is 1, 0.4, and 0.2 pmol/kg, respectively, and free estriol in maternal serum is 0.2-1 pM (21).

Western blot analysis of the soluble placental and ox liver

FIG. 1. Inhibition by steroids of GSH-dependent '261-insulin (A14) degrading activity of two PDI preparations from hu- man placenta relative to control (no steroid present, 0% in- hibition). A14 degradation by the trichloroacetic acid assay, de- scribed under "Experimental Procedures," ranged from 10 to 17%, and each steroid was at 1 p ~ ; mean values of 3 experiments are shown t S.E. The abbreviations used are: 17b-E2, 170-estradiol; 17u- E2, 17~-estradiol; 3MeO-E3, 3-methoxyestriol; DES, diethylstilbes- trol; c-Turnox or c-T, cis-tamoxifen; t-Turnox or t-T, trans-tamoxifen; D H T , dihydrotestosterone.

C I I I I I4

I I I 0 75 150 225 300

171-ESTRADIOL , nM

20

40 0 I -u- 1

I I I I I L j 0 1000 2000 3000 4000 5000

ESTROGEN , nM

FIG. 2. Inhibition of PDI activity and binding of ['Hlestra- diol ('H-E2) to PDI preparations as a function of estrogen concentration, Bottom panel, inhibition of '251-insulin degradation with GSH by cytosol in the presence of 170-estradiol (+), estrone (A); by placental membranes in the presence of 170-estradiol (V), estrone (A); or inhibition of the reactivation of RNase with GSH by cytosol in the presence of 17P-estradiol (0). Middle panel, Inhibition of '251-insulin degradation in the presence of 170-estradiol by cytosol (0) or by 5 milliunits/ml ox liver PDI (1) (m), and inhibition by 170- estradiol of scrambled RNase reactivation in the presence of cytosol (A). Toppanel, binding of [3H]estradiol by cytosol (0) or by placental membranes (A). Human placental cytosol, membranes, and ox liver PDI were incubated in duplicate with '251-insulin, RNase was reacti- vated by human placental cytosol, and [3H]estradiol binding was carried out as described under "Experimental Procedures."

PDI (1) with polyclonal antibody p3632 against the triiodo- thyronine-binding protein (10) showed one band at approxi- mately 56 kDa (data not shown). Cross-linking of '251-insulin (17) to PDI and thioredoxin preparations (Fig. 3) showed binding of insulin to a 59-kDa band in the PDI samples (suggestive of cross-linking via the A-chain of insulin) and to 30- and 43-kDa proteins in the thioredoxin sample; the latter, apparently, represent multimers of the 11-kDa thioredoxin. A 43 and 30% decrease in '251-insulin cross-linking to the 59- kDa bands was caused by 1 p~ estradiol in the placental and liver samples, respectively, but no effect was seen with thio- redoxin. Thus, estradiol decreases the cross-linking of insulin to PDI in the absence of GSH. Lineweaver-Burk plots of lz5I-

insulin degradation by the soluble placental or ox liver PDI showed noncompetitive inhibition by 1 p~ estradiol or es- trone; the K,,, for '251-insulin was 0.5 nM ( 4 ) .

Computer-assisted comparisons between PDI (7) and ste-

Estrogens and Protein Disulfide Isomerase 13969 1 2 3 4 5 6 ” ”

200-

116- , 92 - 66- - 45 -

31- -- “W E 2 - + - + - +

FIG. 3. Cross-linking of ‘261-insulin to ox liver PDI (lanes I and a) , Escherichia coli thioredoxin (Calbiochem) (lanes 3 and 4 ) , and cytosolic PDI from human placenta (lanes 5 and 6) was carried out as described under “Experimental Proce- dures”. The 59-kDa bands shown in the autoradiograph were excised and the lZ5I content was measured. Molecular weight markers were: myosin, E. coli 0-galactosidase, phosphorylase b, bovine serum albu- min, ovalbumin, and carbonic anhydrase.

Thioredoxin

\

FIG. 4. Dot matrix graphic plots (24) of human PDI (7), human estrogen receptor (22,23), and E. coli precursor thio- redoxin (25) sequences. A mutation data matrix (minimum score 20, window 20) was used. The arrows show the positions of the vicinal thiol active sites. Diagonal lines represent regions of similarity, in- cluding repeats.

Thioredoxin 100

FIG. 5. Dot matrix graphic plots (24) of E. coli precursor thioredoxin (25) and human PDI (7) with rat phosphoinositide (PI)-specific phospholipase C I (27). The mutation data matrix (minimum score 20, window 20) was used. PDI regions encoded by exons 3, 4, and 5 are indicated. Diagonal lines represent regions of similarity, including repeats.

roid receptor sequences revealed that PDI segments 120-1633 and 182-230 have significant similarity only with the estrogen

Tsibris, J. C. M., Barker, W. C., Ballejo, G., Frank, B. H., and Spellacy, W. N. (1988) Society for Gynecologic Investigation, 35th Annual Meeting, March 17-20, Baltimore, MD, Abstr. 437.

receptor ( E R 22,23), as shown, hPDl 120-1 63 AGREADDIVNYLKKRTCPMTTLPDCASLVESS--EVAVIGFF

hER 350-392 ADRELVHMNUAKAVPCFVDLTiJiDQVH---LLECAVLEILI.IICLV . .. ... :: :. .: . ::.: . :.:.. :. .::.. . ..

hPDl 182-230 DDIPFCITSNSDVFSKYQLDKDGWLFKXFDEGRWNFEGEVTKENLLDF

hER 304-349 N S L ~ S L T A D Q M T S A - - L L D A E P P I L Y S ~ ~ ~ - F S E A . ..... 8 . . : . :: . .: .: : : . : -.

(single dots indicate amino acids with similar physicochemical properties and codons having one base exchange) but not with the progesterone receptor or glucocorticoid receptor. The ALIGN scores (24) for the 120-163 and 182-230 PDI seg- ments compared to those of ER are, respectively, 6.3 and 5.4 S.D. deviations higher than the mean of 100 randomized sequences of the same regions and indicate a high degree of similarity between these protein regions. Supporting evidence for the similarity between the two segments in PDI and ER is given by the dot matrix plots (Fig. 4); the progesterone receptor shows some similarity only with the 120-163 segment of human PDI (not shown). Fig. 4 also shows that thioredoxin (5 , 25), which is not inhibited by estrogens, is highly similar to residues (approximately) 40-120 and 370-445 of PDI but does not have any similarity with the aforementioned PDI regions (residues 120-163 and 182-230). Plots of ER and thioredoxin showed no regions of similarity. We have found (26) that PDI has significant overall similarity with the newly sequenced phosphoinositide-specific phospholipase C I (27), which also contains two regions homologous to thioredoxin, as shown in the dot matrix plots of Fig. 5. It can be seen that both PDI and this phospholipase C are approximately the same length, contain two thioredoxin-like regions in the same positions, and share additional regions of similarity including the two similar to those in the steroid-binding domain of ER (26). PDI and this phospholipase C must derive from a com- mon ancestral gene (26); it is expected that these two proteins would have common catalytic properties because their ho- mology is more pronounced in the PDI’s active site-containing segments.

The genes for human PDI (28) and ER (29) were recently characterized. The two thioredoxin-like regions of PDI, each containing the putative active site sequence Cys-Gly-His-Cys, are encoded by exons 1-2 and 8-9, respectively. No function is known for exon 3, which exactly codes for the 120-163 segment having the high degree of similarity to the 350-392 segment of ER. Half of the 182-230 PDI segment is encoded by exon 4 and the other by exon 5 . In the ER the steroid- binding domain spans residues 301-552 (29) and is encoded by a DNA region that begins in the middle of exon 4 and extends up to the middle of exon 8. A 21-residue consensus sequence was recently characterized that is found (30) near the putative substrate binding site of steroidogenic cyto- chrome P-450s, and within the steroid-binding domains of receptors (e.g. human ER 534-555) and a sex steroid-binding protein; this region, at least in the steroid-binding receptors, contains some, but not all, residues required for steroid bind- ing and in ER is encoded by exon 8. We suggest that regions 350-392 and 304-349 in ER may contain residues that interact specifically with estrogen. Based on the PDI-ER sequence similarities and the inhibition by estrogens of PDI, but not thioredoxin, it appears that the two PDI regions contain enough residues for an estrogen-specific domain that interacts with the thioredoxin-like active site(s) of PDI. We propose the hypothesis that through an exon-shuffling mechanism other enzymes exist that are or could be made (31) hormone- sensitive by the presence, near their active site, of hormone- binding sequences similar to those found in the binding do-

13970 Estrogens and Protein Disulfide Isomerase

mains of hormone receptors. This hypothesis can be tested experimentally and may prove useful in future gene- and chemo-therapy approaches.

Acknowledgments-We thank Dr. B. H. Frank from Lilly Research Laboratories for the porcine '251-insulin, Dr. S. Cheng for the antibody to the triiodothyronine-binding protein, Stuart Pharmaceuticals for the tamoxifen samples, and Nancy C. Kramer and Kathy Donnelly for their excellent technical assistance.

1. 2. 3.

4. 5. 6.

7.

8.

9.

10.

11.

12. 13.

REFERENCES Bulleid, N. J., and Freedman, R. B. (1987) Nature 335, 294-295 Myllyla, R., and Koivu, J. (1987) J. Biol. Chem. 262, 6159-6164 Roth, R. A., and Pierce, S. B. (1987) Biochemistry 26, 4179-

Duckworth, W. B. (1988) Endocr. Reu. 3,319-345 Holmgren, A. (1985) Methods Enzymol. 54,237-271 Edman, J. C., Ellis, L., Blacher, R. W., Roth, R. A., and Rutter,

W. J. (1985) Nature 317, 267-270 Pihlajaniemi, T., Helaakoski, T., Tasanen, K., Myllyla, R., Hun-

tala, M., Koivu, J., and Kivirikko, K. I. (1987) EMBO J. 6,

Geetha-Habib, M., Noiva, R., Kaplan, H. A., and Lennarz, W. J . (1988) Cell 54, 1053-1060.

Yamauchi, K., Yamamoto, T., Hayashi, H., Koya, S., Takikawa, H., Toyoshima, K., and Horiuchi, R. (1987) Biochem. Biophys. Res. Commun. 146, 1485-1492

Cheng, S., Gong, Q., Parkinson, C., Robinson, E. A,, Appela, E., Merlino. G. T.. and Pastan. I. (1987) J. Biol. Chem. 262.

4182.

643-649

11221-11227 Boada, R. J., Campbell, D. A., and Chopra, I. J. (1988) Biochem.

Biophvs. Res. Commun. 155,1297-1304 Posn&,B. I. (1973) Diabetes 22, 552-563 Sheets, E. E., Tsibris, J. C. M., Virgin, S., Cook, N. I., DeMay,

R. M., and Spellacy, W. N. (1985) Am. J . Obstet. Gynecol. 153, 60-65

14. Myllyla, R., and Oikarinen, J. (1983) J. Biochem. Biophys. Meth-

15. Lang, K., and Schmid, F. X. (1988) Nature 331,453-455 16. Pousette, A., Appelgren, L., Carlstrom, K., Skoldefors, H., Wilk-

ing, N., and Theve, N. 0. (1984) Acta Chem. Scand. B38,407- 414

17. Shii, K., Baba, S., Yokono, K., and Roth, R. A. (1985) J. Biol. Chem. 260,6503-6506

18. Robertson, D. W., Katzenellenbogen, J. A., Long, D. J., Rorke, E. A,, and Katzenellenbogen, B. S. (1982) J. Steroid Biochem.

19. Blyth, C. A., Freedman, R. B., and Rabin, B. R. (1971) Nature

20. Erickson, G. F. (1986) Semin. Reprod. Med. 4, 233-254 21. Nachtigall, L., Bassett, M., Hogsander, U., and Levitz, M. (1968)

Am. J . Obstet. Gynecol. 101, 638, 648 22. Greene, G. L., Gilna, P., Waterfield, M., Baker, A., Hort, Y., and

Shine, J. (1986) Science 231, 1150-1154 23. Green, S., Walter, P., Kumar, V., Krust, A., Bornert, J. M., Argos,

P., and Chambon, P. (1986) Nature 320, 134-139 24. Hunt, L. T., George, D. G., Seibel-Ross, E. I., and Barker, W. C.

(1987) in Methods in Protein Sequence Analysis (Walsh, K. A., ed) pp. 555-563, Humana Press, Clifton, NJ

25. Wallace, B. T., and Kushner, S. R. (1984) Gene (Amst . ) 32, 399- 408

26. Tsibris, J., Hunt, L., Barker, W., and Spellacy, W. N. (1988) J. Cell Biol. 107, (Abstr. 3908) 691

27. Bennett, C. F., Balcarek, J. M., Varrichio, A., and Crooke, S. T. (1988) Nature 334, 268-270

28. Tasanen, K., Parkkonen, T., Chow, L. T., Kivirikko, K. I., and Pihlajaniemi, T. (1988) J. Biol. Chem. 263, 16218-16224

29. Ponglikitmongkol, M., Green, S., and Chambon, P. (1988) EMBO

30. Picado-Leonard, J., and Miller, W. L. (1988) Mol. Endocrinol. 2,

31. Picard, D., Salser, S. J., and Yamamoto, K. R. (1988) Cell 54,

ods 7, 115-121

16 , l -13

New Biol. 230,137-139

J. 7,3385-3388

1145-1150

1073-1080