Reprints Available Directly from the Publisher Photocopying Permitted by License Only the Recognition of Hypothalamo-neurohypophysial Functions by Developing T Cells

Neuropeptide signals and specific neuropeptide receptors have been described in the thymus supporting the concept of a close dialogue between the neuroendocrine and the immune systems at the level of early T-cell differentiation. In this paper, we review recent data about neurohypophysial (NHP)-related peptides detected in the thymus from different species. We suggest that we are dealing in fact with other member(s) of the NHP hormone family, which seems to exert its activity locally through a novel model of cell-to-cell signaling, that of cryptocrine communication. This model involves exchange of signals between thymic epithelial cells and developing thymocytes. The NHP-related peptides have been shown to trigger thymocyte proliferation and could induce immune tolerance of this highly conserved neuroendocrine family.

It is assumed that the action of neuropeptides is mediated through specific receptors on the target cells. The presence of such neuropeptide receptors in the thymus also was reported. Thus, binding sites for SP (Shigematsu et al., 1986), ANP (Kurihara et al., 1987), OT (Elands et al., 1988a), and VP (Geenen et al., 1988b) have been reported. Thymic binding sites for SP were found associated with the vasculature in the medulla, where they may control blood flow and vascular permeability (Shigematsu et al., 1986). On the other hand, specific binding sites for ANP were identified on the rat cortical and medullary thymocytes (Kurihara et al., 1987). It was suggested that they could promote the proliferation and the maturation of thymocytes. Furthermore, it is likely that LH-RH and NKA binding sites are localized on thymocytes because they exert a mitogenic effect on these cells, although it is not excluded that this effect could be mediated by other cell types present in cell cultures (Marchetti et al., 1989;SOder and Hellstr6m, 1989).
In this review, we reconsider the observations about neurohypophysial (NHP)-related peptide signals (OT, VP, and associated NPs) as well as their intrathymic receptors in the context of a new concept of intercellular dialogue, the cryptocrine communication, which was recently proposed for testicular Sertoli cells and thymic nurse cells (TNC) (Funder, 1990). Briefly, cryptocrine signaling differs from the paracrine one in that it involves exchange of signals between specialized epithelial cells (Sertoli in the testis, TNC or other TEC in the thymus) and migratory developing elements (spermatids and respectively). thymocytes, THE NEUROHYPOPHYSIAL PEPTIDE FAMILY Ten NHP hormones, naturally occurring in vertebrates, constitute a family of nonapeptides, highly conserved throughout evolution (Acher and Chauvet, 1988). They can be arranged in two lineages corresponding to the OT-like and VPlike hormones. These peptides all consist of nine amino acids with cysteine residues in positions one and six forming a disulfide bridge. Substitutions occur at positions 4 and 8 and, less frequently, at positions 2 and 3 (cf. Table 1). Vasotocin (VT), originally a synthetical hybrid of oxytocin ring and Arg-vasopressin side chain, was later found in the brain of all the adult nonmammalian vertebrates. It appears to be the most primitive of the NHP peptides_ and may be the common ancestor of all other NHP hormones. However, this molecule has also been detected and characterized in fetal mammals (Pavel, 1975;Smith and McIntosh, 1983;Ervin et al., 1985) as well as in the mammalian pineal gland (Pavel et al., 1975;Legros et al., 1976;Nieuwenhuis, 1984).
Most vertebrates usually produce two NHP peptides: an OT-like hormone, involved in reproductive processes (parturition and lactation), and a VP-like peptide, engaged in the regulation of water metabolism. Howev.er, a single nonapeptide, VT, is found in the lamprey, one of the most primitive vertebrates (Acher and Chauvet, 1988). Therefore, it has been assumed that the OT-like and the VP-like lineages arose from the duplication of one ancestral gene. Recent observations have further supported this postulated molecular mechanism (Morley et al., 1990).
In mammals, the neurohormones of OT and VP are under the control of two independent genes expressed in magnocellular neurons of the hypothalamus. They are synthesized as large molecular weight precusors, which are then cleaved during axonal transpor t (Fig. 1). Prooxyphysin (Pro-OT) is a 16-kD protein containing the OT sequence at its N-terminus, which is separated from the OT-associated neurophysin (OT-NP) by a Gly-Lys-Arg spacer sequence, and which contains an extra C-terminal amino acid (Land et al., 1983). Propressophysin (Pro-VP) is a 20-kD glycoprotein that contains the VP sequence at the N-terminus separated from its associated NP by the same tripeptide spacer (Gly-Lys-Arg). In addition, Pro-VP contains a 39 amino acid terminal glycopeptide, the copeptin, which is separated from VP-NP by a single Arg spacer (Land et al., 1982).
Neurophysins are small (93-95 residues) acidic proteins. The central part (residues 10-74), which is encoded by the second exon, is notably conserved during the evolutionary history of these peptides (Ivell and.Richter, 1984). This sequence is nearly identical between the two NPs of a given species and between various animal species. Variations between lineages occur in the N-terminal part (residues 1-9) and in the C-terminal part (residues 76-95) of the NP molecule. Neurophysins exert a role of carrier for VP and OT in the NHP system, but their high degree of conservation throughout evolution suggests other physiological roles that remain to be established.
On the basis of both functional and pharmacological criteria, two types of VP receptors can be distinguished" V1 receptors that mediate pressor and glycogenic responses to VP, by activating  (Michell et al., 1979). Further, pharmacological studies using peptide binding assays have led to distinguish two V1 receptors subtypes named Via and VID (Jard et al., 1986).
OT receptors were described in the uterus, mammary gland, oviduct, and hippocampus. These receptors seem to be very closely related to the V1 type and also are coupled to phosphoinositide metabolism (Flint et al., 1986).

EVIDENCE FOR NEUROHYPOPHYSIAL-LIKE PEPTIDES IN THYMIC EXTRACTS
We have previously demonstrated that ir-OT could be extracted by acetic acid from human thymuses as evidenced by specific radioimmunoassay (RIA) using antiserum AsO2 (Geenen et al., 1986). On isolated rat uterus, thymic extracts induced oxytocic contraction and the quantitative bioactivity was in agreement with the amounts detected by RIA. Ir-NPs also were detected in human thymic extracts using the antiserum A/5/IV. The molar ratio of ir-OT (2.2-18.4 ng/g) to ir-NPs (24-142 ng/g) was similar to that found in the hypothalamus (1:1), suggesting a local synthesis by cleavage from a common precursor. Ir-VP and ir-VP-NP also were detected in human thymic extracts (0.01-0.06 ng/g and 34-90 ng/g, respectively). The further finding of positive dot blot hybridizations of human thymic mRNA with bovine OT and VP cDNA probes constituted another argument for in situ synthesis, although this observation does not establish a conclusive demonstration of intrathymic OT and VP gene expression (Geenen et al., 1987).
Thymic OT was further characterized by gel filtration on G-75 Sephadex: separate peaks of ir-OT and ir-NP were evidenced at the same positions as their respective reference preparations. HPLC analysis of thymus extracts showed a single peak of ir-OT with a similar elution site as synthetic OT (Geenen et al., 1986).
Independently, ir-VP has been described in thymic extracts of rats and mice (Markwick et al., 1986), and OT-, VP-ald VT-like peptides were reported in ovine thymus (Ervin et al., 1988). The presence of ir-OT was recently confirmed in fetal human thymus (Jevremovic et al., 1990) and in rat thymus (Argiolas et al., 1990a). The finding of a modulation of rat thymic ir-OT concentrations after diverse manipulations further supports a local intrathymic synthesis of an OT-like peptide (Argiolas et al., 1990b).

LOCALIZATION OF NEUROHYPOPHYSIAL-LIKE PEPTIDES IN HUMAN THYMUS
By using polyclonal antibodies, we localized ir-OT and ir-VP in the subcapsular cortex (SCC) and in the medulla of human thymus, as well as in mouse lymphoepithelial complexes called thymic nurse cells (TNC) (Wekerle and Ketelsen, 1980;Geenen et al., 1988a). Furthermore, we confirmed the presence of ir-OT in human, rat, and mouse thymuses with 033, a monoclonal antibody (Mab) directed to OT specifically recognizing hypothalamic magnocellular neurons ( Fig.  2A). We only observed a labeling of the stromal component of thymuses; thymocytes and nerve fibers were never visualized. In human thymus, 033 revealed two immunoreactive areas (Fig.  2B): a monolayer of flattened cells in the SCC and a dense reticular network of stellate cells in the medulla, whereas in the inner cortex, very few scattered stellate cells were labeled. In the mouse thymus, this antibody mainly labels a homogeneous network of cortical reticular cells and some medullary stromal cells (Fig. 2C). Several strains of mice (C3H, BALB/c, C57BL/Ka, CBA, SJL/J) as well as Wistar rats (Fig. 2D) exhibited the same pattern of reactivity. Moreover, the thymus of Brattleboro rats, which exhibit a deficit of the hypothalamic VP gene caused by a single deletion (Richter and Schmale, 1984), also was stained by 033 (Fig. 2E).
Using double immunofluorescence staining with A/5/III, a rabbit serum anti-NPs, we found that the same cells express together ir-OT and ir-NP (Robert et al., 1991). Among the Mabs to VP, we found that BER-312 was the brightest though it labeled thymus sections less intensely than 033, in accordance with the lower concentration of ir-VP found in human thymuses. The pattern of reactivity of BER-312 on human thymus was similar to that obtained with 033 (Robert et al., 1991).
Further, in human thymus, we showed that the stromal cells containing ir-neuropeptides corre-spond to a subset of epithelial cells expressing KL1-and KL4-defined cytokeratins (Robert et al., 1991). The epithelial nature of the cells containing NHP-related peptides was confirmed, in rodent thymuses, using an antiserum to human keratins and KL4 (data not shown). However, KL1 only stained a few medullary TEC in rodents (Nicolas et al., 1989), whereas it strongly labels cortical and medullary TEC from humans (Robert et al., 1991), evidencing a certain disparity between rodent and human TEC.
Besides, both in humans and rodents, it FIGURE 2. Immunostaining with a Mab to oxytocin (033  (Geenen et al., 1988a;Robert et al., 1991), which recognizes the thymic neuroendocrine microenvironment producing the so-called thymic hormones (Haynes et al., 1983;Savino and Dardenne, 1984). In human thymuses, we found that these TEC also contain some interleukin-1 (IL-lfl) immunoreactivity that was colocalized with ir-OT (Robert et al., 1991). It is as well possible that they produce other lymphokines (interleukin-6, colony-stimulating factors, leukemia inhibitory factor) that were evidenced in culture supernatants of human TEC (Le et al., 1990). Furthermore, we observed that, in the rat thymus, a subset of 033 + TEC also was labeled by a specific antiserum directed to NKA, kindly provided by E. Brodin (Karolinska Institute, Stockholm, Sweden).
THYMIC NEUROHYPOPHYSIAL PEPTIDE RECEPTORS Specific OT receptors with a high affinity for OT were demonstrated in the rat thymus with the use of a selective, radioiodinated OT antagonist (Elands et al., 1990). These receptors were detected in membrane preparations and on both mature and immature thymocytes, whereas VP receptors of the V1 type were identified on splenic lymphocytes. Using 3H-AVP and 3H-OT, we have found specific binding sites on RL12-NP, an immature T-cell line derived from a murine Xray-induced thymic lymphoma (Geenen et al., 1988b), as well as on murine CTL-L, a cytotoxic T-cell line (Martens et al., in press (Martens et al., in press). The heterogeneity of the receptor expressed by T cells was further evidenced by the inhibition of transduction in RL12-NP cells by a V1 antagonist, while this inhibition was observed in CTL--L by both OT and V1 antagonists. As a consequence, these observations establish the expression and the functionality of NHP peptide receptors on immature and differentiated T cells. Furthermore, we have observed that OT, VP, and VT increase [3H]-TdR incorporation by human thymocytes cultured in serum-free medium (Martens et al., in press).

IDENTITY OF THYMIC ir-OT AND ir-VP
Altogether, these data establish the intrathymic synthesis of peptide signal(s) exhibiting very close analogy with NHP peptides and the expression of functional cognate receptors by T cells. However, the presence of a material in TEC that exhibits immunological similarities with OT and, to a lesser extent with VP, does not substantiate that we are dealing with authentic OT and VP. Most data rely upon antibodies to OT, VP, and NPs that could, as well, detect some related epitopes. Some molecular differences between hypothalamic and thymic nonapeptides are obvious, since our Mabs all recognized OT or VP in brain slices (Robert et al., 1985;Burgeon et al., 1991), whereas human thymus sections were only brightly stained by 033. As O13 does not stain thymus sections, it could be that the tail part of OT differs in the thymus and in the brain. In addition, the coexistence of ir-OT and ir-VP in TEC provides another conflicting debate. Finally, classical molecular biological methods failed to demonstrate the intrathymic expression of known hypothalamic OT and VP genes. Therefore, thymic ir-OT and ir-VP appear like peptide(s) related to, but distinct, from hypothalamic OT and VP. The expression of this OT/VPlike peptide could be under the control of a different gene coding for another natural NHP peptide, like VT, expressed preferentially during fetal development in peripheral organs. A different splicing of known OT and VP mRNA also can be considered and these questions are under current investigation. Furthermore, the pharmacological characterization of thymic OT receptors is not selective enough to assume that the binding peptide is authentic OT, as it has been shown that OT recep-tors from the hippocampus, the uterus and the mammary gland does not discriminate between OT, VP, and VT Elands et al., 1988b).

PHYSIOLOGICAL IMPLICATIONS
The physiological actions of OTand VP-like peptides within the thymus remain to be further defined. They are not secreted in vitro (unpublished observation) and the classical model of neurosecretion is therefore questionable with regard to thymic epithelium-derived NHP signals. Rather, they seem implicated in thymic cell-to-cell interactions according to the novel "cryptocrine" model 1991b). The affinity of T-cell receptors for OT/VP-like peptides (around 10-1M, in best conditions) provides another argument in favor of this concept. The biological significance of a binding to lymphocytes of blood VP and OT neurohormones (concentrations around 10 -12 M) would be rather irrelevant in any case.
These ir-neuropeptides are localized in areas (SCC in human thymus, TNC in murine thymus) where thymocytes divide actively. It was reported that VP/OT exert a mitogenic effect on several cell types including rat (Whitfield et al., 1970) and human (Martens et al., in press) thymocytes, as well as rat bt)ne marrow cells (Hunt et al., 1977). On cultured fibroblasts, VP is considered to deliver an early regulatory signal of mitogenesis (Rozengurt, 1986); therefore, one can reasonably consider that thymic VP/OT-related peptide(s) provides accessory activation signals for thymocyte proliferation.
Ir-OT/VP are also detected in the medullary epithelium of thymus from different species and in cortical epithelium from rodent thymuses. Tolerogenic properties have been reported for murine TEC including TNC (Lorenz and Allen, 1989;Marrack et al., 1989;Webb and Sprent, 1990), at least for some antigens (Ransom et al., 1991). Obviously, the tolerance to OT seems particularly powerful and autoimmune processes have never been described against OT or OT-producing neurones. This is not the case for VP because some idiopathic dia.betes insipidus seem to be secondary to autoimmune "hypothalamitis" (Scherbaum and Bottazzo, 1983). Another indirect argument for the tolerogenic aspect of OT is derived from published work and from our personal experience. Indeed, the frequency as well as titers of antisera are higher for VP than for OT, and their immunodominant epitope is generally located in the cyclic part of these nonapeptides. Their protection from an autoaggression by the immune system presumes the presentation in the thymus of a peptide (or of a selfimmunodominant epitope) representative of this .family. The tolerogenic properties of NHPrelated peptide(s) and the hypothesis of their intrathymic presentation by major histocompatibility complex proteins are actually investigated in our laboratory.
The rabbit serum against arginine-vasopressin used in RIA (AS2) was mainly directed to the C-terminal linear side chain of the VP molecule (Smitz et al., 1988).
Rabbit antisera to bovine NPs (A/5/III, A/5/IV) were previously shown to recognize the central part of the molecule that is common to OT-NP and to VP-NP and is well conserved through mammals (Legros, 1975).
To characterize VP-NP, we used a specific rabbit antiserum that did not recognize OT-NP, neither OT or VP (Legros and Ansseau, 1989).
The characterization of Mabs to oxytocin was published in detail elsewhere (Burgeon et al., 1991). One Mab (O13) is very specific for OT as it discriminates OT from isotocin (IT) or VP either conjugated to ovalbumin or as free peptides. It does not label the suprachiasmatic nucleus (claimed to contain only VP) and is absorbed by OTbut not by VP-coated beads. As its binding to [125I]-OT is inhibited by OT but not by analogs differing at position 8, we conclude that the C-terminal linear chain of the OT molecule is included in the epitope recognized by O13. We also characterized two Mabs (022 and 033) that are not so strictly specific for OT as they recognize not only OT but also IT and VP conjugates. They label the rat suprachiasmatic nucleus and their immunoreactivity is as well absorbed by OTas by VP-coated beads. As they do not bind to [125I]-OT, we assume that they react with the cyclic part of OT and probably recognize determinant(s) shared by several natural peptides. However, some differences between 022 and 033 were observed: 022 recognizes IT better than VP and it labels nerve fibers more strongly than O13 and 033 (Burgeon et al., 1991). O13 and 022 belong to the IgG1 subclass, whereas 033 is an IgM.
The production and characterization of Mabs to vasopressin were previously reported (Robert et al., 1985). Briefly, seven hybridomas (IgG1) reacting with VP-thyroglobulin complexes but not with thyroglobulin alone were established.
All Mabs bound [125I]-VP and stained Long Evans but not Brattleboro (a VP defective mutant) rat neurohypophysis. These antibodies seem to be strictly directed to the cyclic moiety of the VP molecule. However, BER-312 and BER-624, but not CLA-223, were able to detect frog NHP peptides (vasotocin and/or mesotocin), suggesting that they recognize a slightly different epitope.
To further characterize stromal cells, we used commercially available Mabs to human cytokeratins (KL1 and KL4; Immunotech, Marseille, France) as well as a rabbit antiserum directed against human cytokeratins (A-575, Dako, Copenhagen, Denmark). Mouse hybridoma A2B5 (ATCC HB-29) was grown in our lab and used as undiluted supernatants. Antibodies to IL-lfl were produced by Medgenix Diagnostics (Fleurus, Belgium). goat antisera to mouse IgG or IgM (Fc specific) or to rabbit Ig (Nordic, Tilburg, Netherlands) were used as second-step reagent for 30 min at room temperature. The double-staining procedure involved sequential incubations of tissue sections.
No specific staining was observed in the thymus when dilution buffer, normal rabbit serum (NRS; Dako, Copenhagen, Denmark), or anti-BSA or anti-OVA control ascites (IgG1 or IgM) were used in place of the primary antibodies. Preadsorption of 033 and BER-312 monoclonal antibodies with homologous synthetic peptides, OT and VP, respectively, coupled to CNBr-activated sepharose-4B (Pharmacia, Sweden), abolished the immunostaining. We also checked that our Mabs to neuropeptides did not detect a cross-reacting antigen in epidermal epithelium.  (Accepted September 25, 1991) Immunohistochemistry Thymus fragments were obtained from children undergoing corrective cardiovascular surgery or from 1-month-old mice (C3H, BALB / c, C57BL/Ka, CBA, SJL/J) and rats (Wistar and Brattleboro). These fragments were embedded in Tissue-tek, immediately frozen in dry ice and then stored at-70 C. Frozen sections (5 were cut on a cryostat, air dried, and postfixed with picric acid / formaldehyde. Sections were incubated for 30 min, at room temperature, with 10%.normal goat serum. Firststep antibodies were incubated overnight at +4 C (antineuropeptides) or 30 min at room temperature. Fluoresceine-or rhodamine-conjugated