Downloaded by guest on October 1, 2021 Proc. Nati. Acad. Sci. USA Vol. 88, pp. 2603-2607, March 1991 Neurobiology Evidence for thymopoietin and thymopoietin/a-bungarotoxin/ nicotinic receptors within the brain

M. QUIK*t, U. BABUf, T. AUDHYA*, AND G. GOLDSTEINt *Department of Pharmacology, McGill University, McIntyre Medical Building, 3655 Drummond Street, Montreal, Quebec H3G 1Y6, Canada; and timmunobiology Research Institute, Route 22 East, P.O. Box 999, Annandale, NJ 08801-0999 Communicated by Edward A. Boyse, December 20, 1990

ABSTRACT Thymopoietin, a polypeptide of the These observations suggested that thymopoietin may repre- that has pleiotropic actions on the immune, endocrine, sent an endogenous ligand for the nicotinic a-BGT-binding and nervous systems, potently interacts with the neuromuscu- site in brain; however, the identification of receptor-specific lar nicotinic acetylcholine receptor. Thymopoietin binds to the binding of radiolabeled thymopoietin to a site in brain that nicotinic a-bungarotoxin (a-BGT) receptor in muscle and, like also recognized a-BGT and nicotinic ligands would substan- a-BGT, inhibits cholinergic transmission at this site. Evidence tially support the above hypothesis. The present studies show is given that radiolabeled thymopoietin similarly binds to a that in brain 125I-thymopoietin binds in a specific manner to nicotinic a-BGT-binding site within the brain and does so with a receptor site that also interacts with nicotinic ligands and the characteristics of a specific receptor ligand. Thus specific a-BGT and, furthermore, that a constituent in brain extracts binding to neuronal membranes was saturable, of high affinity has the immunological reactivity and molecular mass of (Kd = 8 nM), linear with increased tissue concentration, and thymopoietin. readily reversible; half-time was -5 min for association and 10 min for dissociation. Binding of '2sI-labeled thymopoietin was MATERIALS AND METHODS displaced not only by unlabeled thymopoietin but also by a-BGT and the nicotinic receptor ligands d-tubocurarine and Materials. Thymopoietin was isolated and purified from nicotine; various other receptor ligands (muscarinic, adrener- thymus as described (1, 2) and radioiodinated (7.8 AGCi/,ug; 1 gic, and dopaminergic) did not affect binding of 'SI-labeled Ci = 37 GBq) according to Audhya and coworkers (2, 11). thymopoietin. Thymopoietin was shown by ELISA to be pres- a-BGT, purified from Bungarus multicinctus venom as de- ent in brain extracts, displacement curves ofthymus and brain scribed (12), was obtained from Miami Serpentarium Lab extracts being parallel to the standard thymopoietin curve, and (Salt Lake City, UT). All other chemicals were obtained from Western (immuno) blot identified in brain and thymus extracts standard commercial sources. a thymopoietin-immunoreactive polypeptide of the same mo- Preparation of the Tissue Extracts. Brain, , thymus, lecular mass as purified thymopoietin polypeptide. We con- and were removed from female Sprague-Dawley rats clude that thymopoietin and thymopoietin-binding sites are weighing =280 g. The tissues were suspended at 20%o (wet present within the brain and that the receptor for thymopoietin wt/vol) in chilled 10 mM Tris-HCl, pH 6.5, containing 1 mM is the previously identified nicotinic a-BGT-binding site of phenylmethylsulfonyl fluoride/i mM ethylenediamine tetra- neuronal tissue. acetate (EDTA)/1 AtM bacitracin/1 ,uM 2-mercaptoethanol/l mM ascorbic acid and were homogenized in an Eberbach Thymopoietin is apolypeptide hormone ofthe thymus (1-3) that con-torque homogenizer on ice for 5 min. The homogenates has pleiotropic actions on the immune, endocrine, and nervous were then centrifuged at 45,000 x g for 20 min at 40C, and the systems (4). Studies of the thymus and myasthenia gravis supernatants were aliquoted and stored at -20'C until used revealed an effect ofthymopoietin on neuromuscular transmis- for ELISA or immunoblot analysis. sion (5), and thymopoietin was subsequently shown to interact ELISA. Rat thymopoietin and tissue extracts were assayed with the nicotinic acetylcholine receptor (nAChR) of the Tor- with a high titer of rabbit anti-human thymopoietin antibody pedo electroplax (6) and of the neuromuscular synapse (7). that also reacts with rat thymopoietin. A limiting dilution of Thymopoietin bound to the muscle-type nAChR with a Kd of antibody was incubated with different concentrations of rat 0.5-3 nM (6, 7) and appeared to interact with a similar region of thymopoietin or different dilutions of the tissue extracts for the receptor as the snake toxin a-bungarotoxin (a-BGT) be- 2 hr at 370C on a rocking platform. The incubation mixture cause thymopoietin prevented binding of '2-I-labeled a-BGT was then added to a rat thymopoietin-coated polyvinyl chlo- (125I-a-BGT), and a-BGT blocked binding of II5I-labeled thy- ride microtiter plate and incubated at 370C for 2 hr. After the mopoietin (125I-thymopoietin) in Torpedo electroplax (6). Fur- plate was washed with phosphate-buffered saline/Tween 20, thermore, thymopoietin inhibited cholinergic transmission at goat anti-rabbit antibody conjugated to horseradish peroxi- the nAChR by a mode ofaction similar to a-BGT (7), although dase was added, and the plate was incubated for 1 hr at 370C. an additional mechanism involving Ca2+-dependent accelera- After three further washes with phosphate-buffered saline/ tion and maintenance ofthe desensitized state ofthe nAChR has Tween 20, a substrate/chromogen system consisting ofH202 also been described (8). and 3,3',5,5'-tetramethylbenzidine dihydrochloride was a-BGT receptors are also present within the brain, and we added, and color was allowed to develop at room temperature have previously shown that thymopoietin effectively com- for 30 min. The reaction was terminated with the addition of petes for these neuronal 125I-a-BGT-binding sites (9). Fur- 2.5 M HCI; optical density was recorded at 450 nm. thermore, thymopoietin was detected in brain extracts and Immunoblot Analysis. The sample was character- supernatants of developing rat spinal cord cultures and of ized by preparing 10 A.l of the extracts in 0.25 A.l of 10% SDS certain neuronal cultures by using an immunoassay (10). and 1 A.l of bromophenol blue and electrophoresing on a The publication costs of this article were defrayed in part by page charge Abbreviations: a-BGT, a-bungarotoxin; nAChR, nicotinic acetyl- payment. This article must therefore be hereby marked "advertisement" choline receptor. in accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed. 2603 2604 Neurobiology: Quik et al. Proc. Natl. Acad. Sci. USA 88 (1991)

Pharmacia PhastSystem SDS electrophoresis system with a 1.5 Ad 20% SDS homogenous polyacrylamide gel. Filter paper and B, -A 0.2-,um nitrocellulose paper were wetted with transfer buffer a (25 mM Tris/192 mM glycine/20% methanol, pH 8.3). A transfer sandwich was assembled as follows: three pieces of filter paper, nitrocellulose paper, the gel, and three pieces of 1.0 filter paper. The protein was then transferred in a Phast- L CO) Transfer system for 15 min at 25 mA per gel. The membrane z was blocked overnight with 3% bovine serum albumin in phosphate-buffered saline/Tween 20, washed three times, In and allowed to react with a horseradish peroxidase- J 0.5- conjugated monoclonal antibody against thymopoietin (37) at 370C for 2 hr. The membrane was washed three times and incubated with the substrate/chromagen mixture (3,3',5,5'- 0 tetramethylbenzidine) for 15-30 min, washed with water, and U-A air dried. 0.1 1 10 100 1000 Preparation of Membranes for the Binding Assay. Male ITPJ (ng/ml) Sprague-Dawley rats were killed by decapitation, and the brains were rapidly removed and dissected on ice. The brain minus the cerebellum was homogenized in 20 vol of 10 mM 2560 640 160 40 10 Tris-HCl, pH 7.4, with a Polytron homogenizer at setting 7 for EXTRACT TITER 10 s. After centrifugation at 45,000 x g for 15 min, the supernatant was discarded, and the membranes were resus- FIG. 1. Immunoreactivity of rat brain (m), thymus (v), liver (o), pended with a Polytron in the appropriate buffer to be used and thyroid (A) extracts as shown by inhibition of ELISA. Parallel in the binding assay. displacement curves were observed with purified rat thymopoietin 125I-Thymopoietin Binding. Membranes were resuspended at (i) and rat brain and thymus extracts. TP, rat thymopoietin. 75 mg/ml (unless otherwise indicated) in 10 mM Tris HCI, pH 7.4. A 160-sl aliquot ofmembranes was preincubated for 15min the same molecular mass position; liver extract provided a at 22°C with the indicated concentration of drug, followed by a negative control with no band detected on the immunoblot. 15- or 20-min incubation at 22°C with 5 nM 125I-thymopoietin '25I-Thymopoietin-Binding to Rat Brain Membranes. Spe- (unless otherwise indicated),. The assay was terminated by cific binding of 125I-thymopoietin to brain membranes was addition of 1.0 ml of 0.2 M NaCl immediately followed by defined as the difference in binding of 125I-thymopoietin centrifugation (5 min) in an Eppendorf microcentrifuge (8800 x without and with 10-7 M thymopoietin. Specific binding was g). The pellet was washed twice, and the radioactivity in the maximal in a 10 mM Tris-HC1 buffer, pH 7.4; other buffer pellet was determined by using a y counter. Specific binding systems tested (50 mM Tris HCI, pH 7.4; 50 mM phosphate was defined as the difference in binding in the absence and buffer, pH 7.4; 50 mM Tris-HCI, pH 7.4, containing 120 mM presence of 10-v M thymopoietin. Specific binding was =40- NaCl, 5 mM KCl, 2 mM CaC12, and 1 mM MgCl2) resulted in 50% of the total binding; this binding could be enhanced by similar or lower specific binding. With regard increasing the number ofwashes ofthe membrane pellets during to incubation the course ofthe centrifugations. However, this procedure also temperature, 220C proved optimal, whereas no specific bind- caused a marked decline in reproducibility within a group of ing was observed at00C. Centrifugation rather than filtration samples, presumably because of the reversibility of 1251 was used to separate free and membrane-bound 1251_ thymopoietin-binding to the membranes. The fraction of radio- thymopoietin. Filter binding, for which a variety of filters labeled thymopoietin bound to membranes was always <10lo preabsorbed with different buffers with or without polyeth- of the total 11I-thymopoietin added to the sample. ylenimine or bovine serum albumin was used, was very Protein Determinations. Protein was determined by the high-from 4% to 44% of total radioactivity in the sample. method of Lowry et al. (13) by using bovine serum albumin Binding of 125I-thymopoietin was saturable (Fig. 3A). Scat- as standard. chard analysis of the saturation curve (Fig. 3B) yielded a Kd of 8.5 nM and a Bm. of 102 fmol/mg of protein. Linearity RESULTS with increasing tissue concentration was also observed. Identification of Thymopoietin-Like Immunoreactivity in Rat Brain. Fig. 1 shows that the rabbit anti-human thymopoi- etin antibody reacted well with rat thymopoietin, giving sensitive displacement in the 5- to 50-ng range with rat thymopoietin by ELISA. , serum albumin, immuno- 6K - globulins, and aprotinin did not give displacements, and control tissue extracts, such as liver and thyroid, also failed C) U to displace specific thymopoietin binding. However, rat ox thymus extracts showed full displacement parallel to the ><) 0- x x 0) 0 displacement curve for purified thymopoietin, an expected Co E a1) finding because thymopoietin is extracted from thymus. Co> CZ Similarly, brain extract gave a displacement curve similar to thymus extract that was also parallel to the curve for purified Co U C- thymopoietin. CZ E: :5 Immunoblot analysis further implied that authentic thy- I mopoietin wasbeing detected by ELISA(Fig. 2). Thymopoi- etin itself yielded a dense band at a position slightly below the FIG. 2. Immunoblot analysis of rat brain, thymus, and liver myoglobinII molecular mass marker (6.380 kDa); rat thymus extracts. The electrophoretic mobility of myoglobin II (molecular and brain extracts produced somewhat less intense bands at mass 6.380 kDa) is indicated by 6K. Neurobiology: Quik et al. Proc. Natl. Acad. Sci. USA 88 (1991) 2605 A 80 A 80

.er Z c az z O z ffi 40 m fix 0 E 0 ob 40 m IN ( Z E LLI E .-I a --

O , 12.5 25.0 [ SlI-THYMOPOErIN] nM 30 60 TIME (min) B 12 B 90

Z0z.c. N~ 6 z O 45

E

a -

0 60 120 BOUND protein) 0 (fmol/mg 30 60 FIG. 3. (A) Saturation curve of 125I-thymopoietin binding to rat TIME (min) brain membranes. Membranes were preincubated for 15 min without or with 10-7 M thymopoietin after which various concentrations of FIG. 4. Association and dissociation of1251-thymopoietin to brain 11I-thymopoietin were added for 20 min. Specific binding was membranes. (A) Membranes were preincubated without or with 10-7 defined as total binding minus binding in the presence of 10-7 M M thymopoietin and subsequently incubated with 5 nM 1251_ thymopoietin. The curve is representative offour experiments, each thymopoietin for the periods indicated. (B) Membranes were prein- done in triplicate. (B) Scatchard analysis of the data; bound/free cubated for 20 min with 5 nM '11I-thymopoietin, and then thymopoi- (B/F) data are expressed as fmol/mg of protein per nM. etin was added to yield a final concentration of 10-7 M; binding was assessed at the times indicated. Each value represents the mean ± Association and dissociation of 15I-thymopoietin to brain SEM ofthree or four experiments. SEM is not depicted where it was membranes are depicted in Fig. 4. Both association and c5% of the mean. dissociation were rapid. The half-time (tl1) for association was --5 min, and binding reached equilibrium after 15-20 min saturable, of high affinity (Kd = 8 nM), and reversible- ofincubation (Fig. 4A). The til2 for dissociation was 410 min, characteristics common to endogenous receptor ligands. and the dissociation curve reached a plateau after =30 min 125I-Thymopoietin binding was displaced by a-BGT with an (Fig. 4B). affinity essentially similar to that seen with thymopoietin The pharmacological characteristics of the '"I-thymopoi- itself, a finding in accord with our previous observation that etin-binding site in brain were investigated by studying the thymopoietin effectively competed with '75I-a-BGT at the inhibition of binding by various drugs (Fig. 5, Table 1). neuronal binding sites (9). Furthermore, nicotinic receptor a-BGT and thymopoietin displaced 1251-thymopoietin binding with IC50 values of 0.4 and 0.5 nM, respectively. Both the 100 nicotinic receptor agonist nicotine and antagonist d-tubo- curarine competed with 1251-thymopoietin for receptor bind- ing but required much higher concentrations. On the other O hand, muscarinic and other receptor agonists and antagonists z z0 had no effect at concentrations as high as 10-4 M. m c0- o 50

DISCUSSION L&J '-- a. The present results suggest that brain has a thymopoietin C', content similar to that of thymus: (i) extracts of both tissues gave similar displacement curves by ELISA that paralleled that of purified thymopoietin; (it) the immunoreactive con- 0 stituent was identical in molecular mass to purified thymopoi- -10 -8 -6 -4 etin by immunoblotting. These findings support and extend [DRUG] the observations (10) of thymopoietin-like material in brain Log extracts and supernatants of spinal cord cultures and certain FIG. 5. Inhibition of 115I-thymopoietin binding by various drugs. neuronal cell lines. Membranes were preincubated for 15 min with the indicated con- Our studies with radiolabeled thymopoietin showed that centrations of drug and then incubated for 20 min with 5 nM the polypeptide bound to neuronal membranes with the 125I-thymopoietin. Each curve represents results from three or four characteristics of a specific receptor ligand. Binding was experiments. 2606 Neurobiology: Quik et al. Proc. Natl. Acad. Sci. USA 88 (1991) Table 1. Inhibition of 1251-thymopoietin binding to brain bind thymopoietin/a-BGT. In contrast, neuronal tissue, such membranes by drugs as adrenal medullary chromaffin cells and the cell line SH- Drug IC50, nM SY5Y, bind thymopoietin/a-BGT, but this binding does not affect acetylcholine-mediated activity (35, 36). Thymopoietin 0.4 ± 0.1 Overall, our findings suggest that thymopoietin is present a-BGT 0.5 ± 0.1 within the brain, where the previously identified nicotinic Nicotine 530 ± 130 a-BGT-binding sites possibly represent the thymopoietin d-Tubocurarine 2500 ± 700 receptors. However, unlike at the peripheral nAChR, where Atropine >100,000 thymopoietin appears to inhibit cholinergic transmission, the Scopolamine >100,000 its Muscarine >100,000 functional significance of thymopoietin interaction with Pilocarpine >100,000 receptors within the brain remains to be elucidated. >100,000 The authors thank J. Philie for excellent technical assistance. This Serotonin >100,000 work was supported, in part, by a grant to M.Q. from the Medical >100,000 Research Council (Canada). Phenylephrine >100,000 Membranes were preincubated for 15 min with the indicated drugs; 1. Audhya, T., Schlesinger, D. H. & Goldstein, G. (1981) Bio- a 20-min incubation was then initiated by addition of 5 nM 1251. chemistry 20, 6195-6200. thymopoietin. The reaction was terminated by adding 1.0 ml of 0.2 2. Audhya, T., Schied, M. P. & Goldstein, G. (1984) Proc. Natl. M NaCl and centrifugation. Each value represents the mean ± SEM Acad. Sci. USA 81, 2847-2849. of three experiments. 3. Audhya, T., Schlesinger, D. H. & Goldstein, G. (1987) Proc. Nati. Acad. Sci. USA 84, 3545-3549. ligands, such as nicotine or d-tubocurarine, inhibited 1251_ 4. Goldstein, G. (1987) in Immune Regulation by Characterized binding at concentrations in the same range as Polypeptides, eds. Goldstein, G., Bach, J. F. & Wigzell, H. thymopoietin (Liss, NY), pp. 51-59. they affected a-BGT binding. These results suggest that 5. Goldstein, G. (1974) Nature (London) 247, 11-14. thymopoietin and a-BGT interact at a similar or the same 6. Venkatsubramanian, K., Audhya, T. & Goldstein, G. (1986) receptor recognition site within the brain, as well as at Proc. Natl. Acad. Sci. USA 83, 3171-3174. neuromuscular nAChR as has been shown (6); in summary, 7. Quik, M., Collier, B., Audhya, T. & Goldstein, G. (1990) J. a-BGT binding appears to define thymopoietin-binding sites. Pharmacol. Exp. Ther. 254, 113-119. This finding is ofinterest because the association ofa-BGT 8. Revah, F., Mulle, C., Pinset, C., Audhya, T., Goldstein, G. & that occurs at Changeux, J.-P. (1987) Proc. Natl. Acad. Sci. USA 84, 3477- binding with high-affinity acetylcholine binding 3481. the neuromuscular nAChR does not appear to apply within 9. Quik, M., Afar, R., Audhya, T. & Goldstein, G. (1989) J. neuronal tissues, where several distinct nicotinic receptors Neurochem. 53, 1320-1323. appear to be present (14-17). One of these is the receptor, 10. Brown, R. H., Schweitzer, T., Audhya, T., Goldstein, G. & hereafter referred to as the neuronal nAChR, that binds Dichter, M. A. (1986) Brain Res. 381, 237-243. monoclonal antibodies against the neuromuscular or electric 11. Audhya, T., Talle, M. A. & Goldstein, G. (1984) Arch. Bio- organ nicotinic receptor; the neuronal nAChR also binds chem. Biophys. 234, 167-177. [3H]nicotine, [3H]acetylcholine, and/or [3H]methyl- 12. Quik, M. & Lamarca, M. V. (1982) Brain Res. 238, 385-399. 13. Lowry, O. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J. carbachol with high affinity but does not bind a-BGT. An- (1951) J. Biol. Chem. 193, 265-275. other receptor is termed the nicotinic a-BGT site, which 14. Lindstrom, J., Schoepfer, R. & Whiting, P. (1987) Mol. Neu- exhibits some characteristics of a nicotinic receptor, as both robiol. 1, 281-337. nicotinic agonists and antagonists prevent binding of the 15. Quik, M. & Geertsen, S. (1988) Can. J. Physiol. Pharmacol. 66, toxin to the receptor. 971-979. This classification or distinction between the nAChR and 16. Berg, D. K. & Halvorsen, S. W. (1988) Nature (London) 334, the nicotinic a-BGT site is based on extensive experimental 384-385. evidence. Functional studies have shown that although 17. Steinbach, J. H. & Ifune, C. (1989) Trends Neurosci. 12, 3-6. 18. Morley, B. J., Kemp, G. E. & Salvaterra, P. M. (1979) Life Sci. a-BGT bound to a site with nicotinic characteristics, the 24, 859-872. toxin did not alter responses mediated by nicotinic receptors 19. Oswald, R. E. & Freeman, J. A. (1981) Neuroscience 6, 1-14. (18, 19). Purification studies also distinguished between the 20. Boulter, J., Evans, K., Goldman, D., Martin, G., Treco, D., neuronal nAChR and the nicotinic a-BGT site. These studies Heinemann, S. & Patrick, J. (1986) Nature (London) 319, demonstrated the existence ofseveral nAChR subtypes ofthe 368-374. general subunit composition a243,82-3, carrying various forms 21. Boulter, J., Connolly, J., Deneris, E., Goldman, D., Heinemann, of the a subunit (a2, a3, a4, and a5) and the (8 subunit (132, S. & Patrick, J. (1987) Proc. Natl. Acad. Sci. USA 84, 7763-7767. (33, and (34) (14, 20-24). On the other hand, the a-BGT- 22. Whiting, P. & Lindstrom, J. (1987) Proc. Natl. Acad. Sci. USA binding protein appeared to comprise more than two different 84, 595-599. 23. Duvoisin, R. M., Deneris, E. S., Patrick, J. & Heinemann, S. subunits, possibly ofa composition similar to muscle nAChR (1989) Neuron 3, 487-4%. (14, 22, 25, 26). Receptor localization studies at light and EM 24. Boulter, J., O'Shea-Greenfield, A., Duvoisin, R. M., Connolly, levels also imply that the binding patterns of high-affinity J. G., Wada, E., Jenson, A., Gardner, P. D., Ballivet, M., neuronal nAChR and nicotinic a-BGT sites are distinct Deneris, E. S., McKinnon, D., Heinemann, S. & Patrick, J. (27-29). Furthermore, studies involving receptor regulation (1990) J. Biol. Chem. 265, 4472-4482. have indicated that the two receptor populations are not 25. Gotti, C., Esparis Ogando, A. & Clementi, F. (1990) Neuro- consistently altered in parallel after various experimental science 32, 759-767. manipulations (30-34). 26. Kemp, G., Bentley, L., McNamee, M. G. & Morley, B. J. Studies in vitro with neuromuscular preparations and/or (1985) Brain Res. 347, 274-283. also the between 27. Jacob, M. & Berg, D. K. (1983) J. Neurosci. 3, 260-271. cells in culture support distinction central 28. Loring, R. H., Dahm, L. M. & Zigmond, R. E. (1985) Neuro- nicotinic a-BGT binding sites and the neuromuscular science 14, 645-660. nAChR. Both thymopoietin and a-BGT inhibit cholinergic 29. Clarke, P. B. S., Schwartz, R. D., Paul, S. M., Pert, C. B. & transmission between phrenic nerve and hemidiaphragm (7) Pert, A. (1985) J. Neurosci. 5, 1307-1315. and inhibit acetylcholine-induced ion flux in TE671 cells (35); 30. Marks, M. J., Burch, J. B. & Collins, A. C. (1983) J. Pharma- both of these preparations have a muscle-type nAChR and col. Exp. Ther. 226, 817-825. Neurobiology: Quik et al. Proc. Nati. Acad. Sci. USA 88 (1991) 2607

31. Mitsuka, M. & Hatanaka, H. (1983) J. Neurosci. 3, 1785-1790. 35. Lukas, R., Audhya, T., Goldstein, G. & Lucero, L. (1990) Mol. 32. Kemp, G. & Edge, M. (1987) Mol. Pharmacol. 32, 356-363. Pharmacol. 38, 887-894. 33. Quik, M., Fournier, S. & Trifaro, J. M. (1986) Brain Res. 372, 36. Quik, M., Afar, R., Geertsen, S., Audhya, T., Goldstein, G. & 11-20. Trifaro, J. M. (1990) Mol. Pharmacol. 37, 90-97. 34. Quik, M., Geertsen, S. & Trifaro, J. M. (1987) Mol. Pharmacol. 37. Talle, M. A., Brown, M. J., Blynn, C. M., Audhya, T. & 31, 385-391. Goldstein, G. (1991) Thymus, in press.