HLA-G is an HLA class Ib gene that is highly expressed in human trophoblast cells. The single HLA-G mRNA is alternatively spliced to generate at least seven transcripts, three of which encode soluble isoforms. Many studies have shown that high levels of soluble antigens are associated with successful implantation and graft acceptance. To study expression, regulation and functions of two of the soluble isoforms, HLA-G5 and HLA-G6, we generated recombinant proteins in eukaryotic cells and developed monoclonal antibodies specific for each of the two proteins. In addition, we investigated the olive baboon Paan-AG gene as a potential functional correlate of HLA-G. Here, we present summaries of the studies that have been conducted in our laboratory using these tools and discuss the results within the context of the research on this topic that is ongoing in ours and other laboratories worldwide. Collectively, the data indicate that soluble HLA-G is a critical contributor to immune privilege in pregnancy and imply that this placenta-derived substance may impact other pathways leading to successful reproduction.

The human major histocompatibility complex (MHC) contains genes encoding the Human Leukocyte Antigens (HLA). Of these antigens, placental immunologists need study only the HLA class I molecules, because HLA class II expression is repressed in the fetal placental cells that are in direct contact with maternal blood and tissues containing maternal immune cells. The class I antigens are subdivided into two general categories. The class Ia antigens are highly polymorphic and are typified by HLA-A, -B, and -C; these are expressed by nearly all somatic cells and stimulate graft rejection when foreign to the host. By contrast, the HLA class Ib antigens, HLA-E, -F, and -G, have restricted expression, few variants, and appear rarely to be immunostimulatory. One class Ia antigen, HLA-C, and the three class Ib antigens are differentially expressed by trophoblast cell subpopulations. In order to understand immune privilege in the pregnant uterus and placenta, it is essential to study the unique structural and functional features of these four genes and their glycoprotein products. In this chapter, we focus on the first class Ib gene identified in human placentas, HLA-G, with emphasis on its two soluble isoforms, HLA-G5 and HLA-G6. We describe methods developed in our laboratory to distinguish mRNAs encoding HLA-G5 and HLA-G6, and antibody-based protocols for identification of the soluble isoforms.

Soluble class Ib HLA-G glycoproteins synthesized in the placenta are abundant in the pregnant uterus and circulate in maternal blood throughout pregnancy. To establish immunogenicity of these proteins, we tested sera from 64 women with at least one successful pregnancy (multigravid), 21 women who had never been pregnant, and 54 males for antibodies to epitopes present on recombinant sHLA-G isoforms (sHLA-G1, sHLA-G2) derived from HLA 6.0 cDNA (HLA-G*0101 allele). By indirect enzyme-linked immunosorbent assay, antibodies to sHLA-G isoforms were identified in six sera, all from multigravid women; all other sera were negative (P = 0.0083). Immunoblots showed that two of the positive sera reacted exclusively with sHLA-G1 and -G2 whereas four reacted to both sHLA-G and pooled HLA class I antigens. To establish potential relationships between anti-sHLA-G and exposure to foreign paternal alleles (*0101, *0103, *0104, *0106), all multigravid women and their partners were genotyped. No relationship between allelic disparity and antibody production was identified. Taken together, these results indicate that (i) tolerance to HLA-G is the usual condition as antibodies to HLA-G were not detected in 91% (58/64) multigravid women, and (ii) pregnancy stimulates loss of tolerance in 9% (6/64) of multigravid women. All six women delivered healthy babies, demonstrating that maternal antibodies to epitopes on sHLA-G do not abrogate pregnancy.

The HLA-G message is alternatively spliced into multiple transcripts, two of which encode soluble isoforms. To initiate studies on the specific functions of the soluble isoforms, we produced soluble rHLA-G1 (rsG1) and rsG2 in human embryonic kidney 293 cells and characterized the proteins. Both isoforms were glycosylated and formed disulfide-bonded oligomers. Recombinant sG1 associated with beta(2)-microglobulin, whereas rsG2 did not. Mouse mAb generated to rsG1 (1-2C3), which identified exclusively sG1, and mAb generated to rsG2 (26-2H11), which identified both soluble and membrane G2 (m/sG2), were used for inummohistochemical isoform mapping studies on placental tissue sections. Soluble G1 protein was abundant in many subpopulations of trophoblast cells, whereas m/sG2 protein was present exclusively in extravillous cytotrophoblast cells. Although both isolated placental villous cytotrophoblast cells and chorion membrane extravillous cytotrophoblast cells contained mRNAs encoding sG1 and sG2, protein expression was as predicted from the immunostains with m/sG2 present only in the invasive trophoblast subpopulation. Analysis of function by Northern and Western blotting demonstrated that both rsG1 and rsG2 inhibit CD8alpha expression on PBMC without changing CD3delta expression or causing apoptotic cell death. Collectively, the studies indicate that: 1) both sG1 and m/sG2 are produced in placentas; 2) transcription and translation are linked for sG1, but not G2; 3) expression of G2 is exclusively associated with the invasive phenotype; and 4) the two isoforms of sG may promote semiallogeneic pregnancy by reducing expression of CD8, a molecule required for functional activation of CTL.

Mouse ovarian surface epithelial cells (MOSEC) were obtained from virgin, mature mice by mild trypsinization and were repeatedly passaged in vitro. Early passage cells (<20 passages) exhibited a cobblestone morphology and contact inhibition of growth. After approximately 20 passages in vitro, cobblestone morphology and contact inhibition of growth was lost. Tumor forming potential was determined by s.c. and i.p. injection of early and late passage cells into athymic and syngeneic C57BL6 mice. Subcutaneous tumors formed in approximately 4 months and were present only at the injection site. Intraperitoneal injection of late passage MOSEC into athymic and syngeneic mice resulted in growth of tumor implants throughout the abdominal cavity, and production of hemorrhagic ascitic fluid. Early passage MOSEC did not form tumors in vivo. Histopathologic analysis of tumors revealed a highly malignant neoplasm containing both carcinomatous and sarcomatous components. Late passage MOSEC expressed cytokeratin and did not produce ovarian steroids in response to gonadotropin stimulation in vitro. Ten clonal lines were established from late passage MOSEC. Each clone formed multiple peritoneal tumors and ascitic fluid after i.p. injection into C57BL6 mice. Three cell lines examined cytogenetically were polyploid with near-tetraploid modal chromosome numbers. Common clonal chromosome gains and losses included +5, +15, +19 and -X, -3, -4. One cell line had a clonal translocation between chromosomes 15 and 18 and another had a small marker chromosome; common structural abnormalities were not observed. These data describe the development of a mouse model for the study of events related to ovarian cancer in humans. The ability of the MOSEC to form extensive tumors within the peritoneal cavity, similar to those seen in women with Stage III and IV cancer, and the ability of the MOSEC to produce tumors in mice with intact immune systems, makes this model unique for investigations of molecular and immune interactions in ovarian cancer development.

Previous studies from our laboratory have shown that prolonged exposure of mouse macrophages to IFN-beta interferes with their subsequent ability to become activated for tumor cell killing. Data reported here show that such inhibition is due to reduced production of NO, resulting from decreased transcription of the gene that encodes inducible NO synthase (iNOS; EC 1.14.13.39). The molecular basis for such suppression was shown to be, at least in part, decreased nuclear accumulation of tyrosine-phosphorylated Stat1 alpha (pStat1 alpha), and a consequent change in the nuclear ratio of pStat1 alpha to non-transactivating pStat1 beta. Reduced phosphorylation was observed despite the fact that time-course studies revealed greater than normal quantities of both Stat1 alpha and Stat1 beta proteins in macrophages that had been pre-exposed to IFN-beta. The decrease in nuclear pStat1 alpha was demonstrated to involve an increase in the rate of turnover of phosphorylated protein. The homodimeric form of pStat1 alpha is essential for the expression of both the iNOS and IFN-regulatory factor-1 genes (the product of the latter is necessary for full expression of the iNOS gene). These results have broad implications, because they suggest that limiting the availability of homodimeric pStat1 alpha is a means by which down-regulation of genes containing promoter-linked IFN-gamma-activated sites might be achieved.

Mechanisms accounting for protection of the fetal semiallograft from maternal immune cells remain incompletely understood. In other contexts, interactions between TRAIL (TNF-related apoptosis-inducing ligand/Apo-2L) and its receptors kill activated lymphocytes. The purpose of this study was therefore to investigate the potential of the TRAIL/TRAIL-R system to protect the placenta against immune cell attack. Analysis by Northern blotting demonstrated mRNAs encoding TRAIL as well as the four TRAIL receptors (DR4, DR5, DcR1/TRID, DcR2/TRUNDD) in human placentas. Immunohistochemical experiments demonstrated that TRAIL protein is prominent in syncytiotrophoblast, an uninterrupted placental cell layer that is continuously exposed to maternal blood, as well as in macrophage-like placental mesenchymal cells (Hofbauer cells). Studies on cell lines representing trophoblasts (Jar, JEG-3 cells) and macrophages (U937, THP-1 cells) showed that both lineages contained TRAIL mRNA and that steady state levels of transcripts were increased 2- to 11-fold by IFN-gamma. By contrast, cell lineage-specific differences were observed in expression of the TRAIL-R genes. Although all four lines contained mRNA encoding the apoptosis-inducing DR5 receptor, only trophoblast cells contained mRNA encoding the DcR1 decoy receptor and only macrophages contained DcR2 decoy receptor transcripts. DR4 mRNA was present only in THP-1 cells and was the only TRAIL-R transcript increased by IFN-gamma. Cytotoxicity assays revealed that the two trophoblast cell lines were resistant, whereas the two macrophage lines were partially susceptible to killing by rTRAIL. Collectively, the results are consistent with a role for the TRAIL/TRAIL-R system in the establishment of placental immune privilege.

In mice and humans, expression of the tumour necrosis factor receptor-1 (TNF-R1) gene in placental trophoblast cells is constitutive whereas expression of the TNF-R2 gene is developmentally programmed. In order to study the individual functions of TNF-R1 and -R2 in this lineage, cell lines were generated from placental explants of homozygous matings of gestation day 10 outbred mice (Swiss–Webster), TNF-R1-deficient (TNF-R1−/−) and TNF-R2−/−transgenic mice as well as the background strain for the TNF-R2−/−mice (WT, C57BL/6×129). All of the cells exhibited trophoblast markers; they contained cytokeratin intermediate filaments, expressed alkaline phosphatase activity and displayed transferrin receptors, but were negative for vimentin filaments and the macrophage marker, F4/80. Analysis of DNA by polymerase chain reaction demonstrated the expected TNF-R genotype in each line. In experiments testing the effects of recombinant mouse TNF-α (rmTNF-α) on viability and proliferation of the cell lines, rmTNF-α modestly but dose-dependently inhibited the growth of WT and TNF-R2−/−cells while having no effect on TNF-R1−/−cells. Actinomycin D-treated WT and, to a lesser extent, TNF-R2−/−cells, were more sensitive to growth inhibition than untreated cells whereas TNF-R1−/−cell responses remained unchanged. These data indicated that rmTNF-α inhibits growth of trophoblastic cells through TNF-R1 and that newly synthesized protein(s) provide partial protection against toxicity. In contrast to the receptor species-specific effects on cell growth exerted by rmTNF-α, both TNF-R mediated inhibition of alkaline phosphatase activity. Collectively, the observations support the postulate that receptor expression is the key factor which determines the nature and extent of TNF-α effects on trophoblast cell growth and function.

Unlike other somatic cells, human placental trophoblast cells do not express the highly polymorphic HLA-A and HLA-B human leukocyte major histocompatibility antigens that would stimulate maternal immunological rejection of the fetus. To investigate mechanisms underlying cell lineage-specific expression, cell lines were generated from homozygous matings of HLA-B27 transgenic mice. Trophoblast cell lines were generated from gestation day 10 placentas and fibroblasts were cultured from gestation day 13/14 embryos. Polymerase chain reaction (PCR) readily identified HLA-B DNA in transgenic trophoblastic cells but specific mRNA was of low abundance, being detectable by reverse transcriptase PCR but not by Northern blot hybridization. HLA-B-specific protein in/on the trophoblast cells was undetectable by cell enzyme-linked immunosorbent assay and the protein was not induced by exposing the trophoblastic cells to interferon-gamma (IFN-gamma). Restricted expression was specific for the HLA-B transgene and its antigen; IFN-gamma-inducible endogenous H-2D(b) class I antigens were detectable on the trophoblast cells. In contrast to the trophoblastic cells, HLA-B27 transgenic fibroblasts expressed IFN-gamma-inducible HLA class I antigens as well as H-2D(b) antigens. Thus, the mechanism(s) regulating expression of the polymorphic HLA-B antigen in trophoblastic cells is gene-specific, IFN-gamma-resistant and operative at the level of transcription or immediate post-transcription.

Nitric oxide (NO), a potent and versatile free radical, is synthesized in macrophages and mast cells as well as in other types of cells by the inducible form of nitric oxide synthase (iNOS). In this study, cells containing iNOS were identified in the uteri of cycling mice by using a rabbit antibody generated to an iNOS-specific peptide. Macrophages were identified in semiserial sections of the same tissues with the monoclonal antibody, F4/80, and mast cells were identified by toluidine blue staining. In tissue sections of uteri obtained from mice in the four stages of the estrous cycle (8 to 11 mice per stage), iNOS immunoreactivity was strongest in diestrus-I uteri and weakest in diestrus-II uteri. Myometrial mast cells and endometrial epithelial cells were prominent locations of iNOS, and specific protein was also present in myometrial smooth muscle and macrophage-like cells in the endometrial stroma. Because cyclic variations suggested regulation of iNOS expression by ovarian steroid hormones, studies were done using ovariectomized mice. Seven days after ovariectomy, immunoreactive iNOS was low but detectable in mast cells and luminal epithelial cells. In the uteri of ovariectomized, estradiol-17 beta (E(2))-treated mice, mast cells were iNOS(+) after 24 h whereas epithelial cells were negative; the reverse was observed in progesterone (P-4)-treated mice. Both mast cells and epithelial cells were iNOS(+) in the uteri of mice that had received a combination of E(2) + P-4. These results indicate that several types of uterine cells produce iNOS and that expression of this enzyme in specific cell lineages is governed by ovarian steroid hormones. The data are consistent with the postulate that NO derived from uterine leukocytes and other types of cells plays a role in uterine cyclicity and preparation for pregnancy.

Endotoxic lipopolysaccharides (LPS) mediate lethality in mice by a complex inflammatory process involving the production of multiple mediators, including tumor necrosis factor-α (TNF-α) and nitric oxide (NO). The present study had two objectives: (i) to determine the extent to which TNF-α contributes to the induction of NO production by mouse macrophages activated with LPS in vitro; and (ii) to assess the contribution of macrophage-derived NO to the pathogenesis of endotoxin shock in mice. The studies reported here show that the synthetic adenyl carbocyclic nucleoside 9-[(1S,3R)- cis-cyclopentan-3-ol]adenine (cPA) inhibited TNF-α, but not NO, production by thioglycollate-elicited peritoneal macrophages that were activated with either LPS alone, LPS + interferon-γ (IFN-γ) or IFN-γ + TNF-α. The expression of cytoplasmic TNF-α in LPS + IFN-γ-activated cells was similarly inhibited by cPA, whereas the appearance of inducible NO synthase was unaffected by the compound. Of significance, pretreatment of mice with a single injection of cPA protected the animals against subsequent LPS challenge in two models of endotoxin lethality. These results suggest that macrophage-derived NO, induced by LPS, may not be an essential mediator of the lethal effects of endotoxin. Further, the results of these studies suggest that TNF-α-induced NO production by tissue macrophages also may not be an essential contributing factor in the pathogenesis of lethality induced by endotoxin in mice.

The role of endogenous tumor necrosis factor alpha (TNF-alpha) and interferon-beta (IFN-beta) in lipopolysaccharide (LPS)-induced activation of the inducible nitric-oxide synthase (i-NOS) gene was investigated. By Northern analysis or reverse-transcription polymerase chain reaction, the mouse macrophage cell line (J774) was found to respond to LPS treatment by increased expression of mRNAs specific for TNF-alpha, IFN-beta, and i-NOS with the kinetics unique for each gene. Bioassay of the culture supernatants showed that TNF-alpha and IFN-beta secreted by J774 cells increased from an undetectable level to about 300 and 340 units/ml, respectively, 3-6 h after LPS stimulation. Nitrite concentration was found to increase from 0 to 7.8 and 28.5 microM by 12 and 24 h, respectively, in the culture supernatant of LPS-treated J774 cells. The presence of a neutralizing dose of antibodies against IFN-beta, but not against TNF-alpha, during treatment with either 10 ng or 1 microgram of LPS/ml significantly, but not completely decreased the level of i-NOS-specific mRNA expression and NO production. The incubation of J774 cells with mouse natural IFN-beta itself (up to the level of 1,200 units/ml) did not induce i-NOS-specific mRNA and therefore did not stimulate J774 cells to produce NO. However, natural IFN-beta synergistically augmented the expression of i-NOS mRNA and the production of NO by J774 cells triggered by suboptimal concentrations of LPS (1 to 5 ng/ml). These data thus suggest that endogenous IFN-beta, but not TNF-alpha, produced by LPS-stimulated J774 cells specifically contributes, probably in an auto/paracrine fashion, to the activation of the i-NOS gene expression by LPS.

Bacterial lipopolysaccharides (LPS) are potent inducers of macrophage activation, leading to the production of a number of proinflammatory mediators. Although several cytokines that prime macrophages for enhanced LPS-triggered responses have been identified, far less is known regarding the role that cytokines play in down-regulating macrophage responses to LPS. This study was designed to determine the effects of recombinant transforming growth factor beta 1 (rTGF-beta 1) on macrophage activation by LPS. Pretreatment of either mouse peritoneal macrophages or cells of the RAW 264.7 macrophage-like cell line with rTGF-beta 1 inhibited their ability to produce both tumor necrosis factor alpha (TNF-alpha) and nitric oxide (NO) in response to LPS. These inhibitory effects were reversed by increasing the concentration of LPS or by priming cells with optimal concentrations of recombinant gamma interferon (rIFN-gamma). Pretreatment of cells with rTGF-beta 1 had only a modest inhibitory effect on the expression of TNF-alpha mRNA. By contrast, the expression of mRNA for the inducible form of nitric oxide synthase (iNOS), which is responsible for NO production in activated macrophages, was significantly inhibited by rTGF-beta 1 pretreatment. Thus, rTGF-beta 1-dependent suppression of macrophage TNF-alpha biosynthesis was manifest at a posttranscriptional level, whereas the inhibition of NO production correlated with a direct effect on iNOS gene expression. Importantly, both of these suppressive effects of rTGF-beta 1 were reversed by exposing the cells to priming concentrations of rIFN-gamma. As with NO production, immunocytochemical analysis of iNOS expression in LPS-stimulated macrophages revealed that rIFN-gamma and rTGF-beta 1 had antagonistic effects, with the former increasing, and the latter reducing, the number of iNOS-expressing cells induced by LPS. These data suggest that a balance between the priming effects of IFN-gamma and the inhibitory effects of TGF-beta 1 can determine the overall level of macrophage activation induced by LPS.

Macrophages and natural killer (NK)-like cells are the major hematopoietic cell populations in the cycling and pregnant mouse uterus and are also found in the embryo. In order to evaluate potential receptivity of these cells to interferon-gamma (IFN-gamma), tissues taken from cycling and pregnant mice were tested for IFN-gamma receptor (IFN-gamma R) mRNA and protein. Macrophages were identified immunohistochemically by using the specific monoclonal antibody F4/80. NK cells were identified by their large size, distinctive intracellular granules, and binding of a monoclonal antibody to the common leukocyte antigen. In cycling uteri, the abundance of IFN-gamma R mRNA relative to an invariant message (glyceraldehyde-3-phosphate dehydrogenase) increased during progression of the hormonally regulated estrous cycle. IPN-gamma R mRNA in situ hybridization signals were slightly higher in macrophage-like than in other types of endometrial stromal cells. In pregnant uteri, the highest proportions of IFN-gamma R mRNB were observed at gestation day (g.d.) 16. Specific message and protein were present in uterine macrophages by p.d. 7 and in NK cells by g.d. 9. IFN-gamma R expression in both lineages remained stable through the balance of pregnancy. In embryos, IFN-gamma R mRNA increased between g.d. 14 and 16. Specific transcripts were present in many cells at g.d. 14, but none were detected in embryonic liver macrophages until g.d. 16. The results of this study; suggest relationships between IFN-gamma R expression and ovarian hormones as well as cell maturation and support the postulate that IFN-gamma receptor-ligand interactions may improve the ability of uterine and embryonic hematopoietic cells to perform specific tasks during gestation.

Abnormal expression of polypeptide growth factors and their receptors is closely associated with tumorigenic transformation. In this study tumor necrosis factor-α (TNF-α) mRNA and protein were analyzed in polyps and proliferative lesions of endometrium as well as in low and high grade endometrial tumors by using in situ hybridization and immunocytochemistry. All samples contained products of the TNF-α gene. Histochemical scores (HS), which reflect the proportion of cells positive for TNF-α message or protein and the intensities of the signals, were higher for epithelial than for stromal cells. Benign lesions (endometrial polyps) contained little TNF-α mRNA or protein, whereas specific message was abundant in proliferative lesions (hyperplasia, adenofibroma). Although neoplastic cells in both low and high grade endometrial tumors contained TNF-α mRNA, two major differences were observed: HS for TNF-α mRNA were significantly less in low grade than in high grade neoplasms, and TNF-α message was restricted to the nucleus in low grade adenocarcinoma cells but was abundant in the cytoplasm of high grade tumor cells. In contrast to cells in benign and proliferative lesions, TNF-α protein scores in endometrial tumor cells were inversely rather than positively correlated with TNF-α mRNA scores. Collectively, the findings in this study are consistent with the postulate that TNF-α is useful to endometrial tumor cells and suggest that production may increase as cells diverge from normal.

The reactive nitrogen intermediate, nitric oxide (NO) is important in host defense against both NO-sensitive microorganisms and tumor cells. Macrophages are one of the chief inflammatory sources, especially when stimulated with the combination of LPS and interferonγ (IFNγ). It is not known, however, whether IFNγ-mediated augmentation of LPS-induced production of NO is the result of greater production by all cells or to the recruitment of more producer macrophages within a given population. This question was addressed, first, by stimulating mouse macrophages (either bone marrow culture-derived, inflammatory peritoneal or those of the cell line, RAW 264.7) with up to 10 U/ml IFNγ for as long as 24 h. Under these conditions, there was little or no production of NO and rare or no cells were immunocytochemically positive for the inducible form of nitric oxide synthase (iNOS), which catalyzes the production of NO. Populations similarly exposed to 1 ng/ml LPS were low producers of NO and contained somewhat more, but still only a few (< 15%), iNOS-positive cells. In contrast, as the concentration of IFNγ was increased (≥ 1 U/ml) in the presence of a constant amount of LPS (1 ng/ml), the principal effect was to increase both the production of NO and the number of iNOS-positive macrophages. The amount of iNOS expressed by some cells also appeared to be increased. Two important conclusions can be drawn from these findings: (1) there is heterogeneity in mouse macrophage populations with respect to the production of iNOS; and (2) increasing concentrations of IFNγ appear to augment LPS-induced secretion of NO by recruiting increasingly greater numbers of macrophages into the production of iNOS. Such results potentially provide important clues as to how IFNγ may be acting at the subcellular level to enhance iNOS synthesis.

Previous studies have shown that the TNF-alpha gene is transcribed and translated in fully differentiated human placental syncytiotrophoblast. In this study, TNF transcripts were identified by in situ hybridization in cytotrophoblastic cells, a progenitor subpopulation that proliferates rapidly in early gestation tissues. To establish molecular and biochemical characteristics of cytotrophoblastic TNF and to evaluate potential utilization, experiments were conducted on two cytotrophoblastic cell lines, Jar and JEG-3. Northern blot hybridization and immunocytochemical tests showed that Jar and JEG-3 cells contained TNF mRNA and specific protein. Enzyme immunoassays demonstrated production of TNF, and immunoprecipitation experiments showed that Jar cell TNF protein was the same molecular mass as macrophage TNF. DNA synthesis in both lines was promoted by rTNF, and experiments employing 17-mer TNF antisense and sense oligonucleotides showed specific inhibition of DNA synthesis by antisense sequences. Both p60 and p80 TNF-R mRNA were present in the choriocarcinoma cell lines, and DNA synthesis was inhibited by antibody to the p60 TNF-R. Although the two lines were similar in many respects, Jar cells produced more TNF and demonstrated a greater reliance on TNF for their growth. Collectively, the results indicate that: 1) the TNF gene is expressed in both normal and malignant cytotrophoblast; 2) certain molecular, immunologic, and biochemical characteristics of trophoblast-derived TNF are similar to macrophage TNF; and 3) the p60 TNF-R facilitates utilization of TNF as an autocrine growth factor by choriocarcinoma cells. Although TNF apparently serves important functions in cytotrophoblast during the course of placental development that might include promotion of proliferation and invasion, constitutive expression of this gene in neoplastic cells could account in part for the remarkable ability of trophoblastic tumors to overcome host defenses.

Enriched fractions of spermatogenic cells were isolated by unit gravity sedimentation and analyzed both for the presence of secreted tumor necrosis factor-alpha (TNF alpha) in vitro by bioassay and for the presence of TNF alpha mRNA by Northern blot analysis. Small quantities of bioactive TNF alpha were consistently detected in medium conditioned by round spermatid fractions. Both pachytene spermatocyte and round spermatid fractions contained RNA that hybridized with murine cDNA probes for TNF alpha, with pachytene spermatocytes containing a normal 1.9-kilobase (kb) transcript, while round spermatids contained principally an approximately 2.8-kb transcript. Both the normal size transcript and the larger haploid-specific transcript were enriched when total RNA from pachytene spermatocyte and round spermatid fractions was passed through an oligo(dT) column. The normal 1.9-kb transcript within pachytene spermatocytes could be induced by exposing the spermatogenic cells to lipopolysaccharides in vitro, yet the approximately 2.8-kb transcript within round spermatids appeared uninduced by LPS treatment. In situ hybridization for the TNF alpha message by using digoxigenin label antisense TNF alpha riboprobe labeled pachytene spermatocytes, round spermatids, and presumptive interstitial macrophages. Spermatogonia and elongating spermatids as well as other interstitial cells were unlabeled or very lightly labeled. Hybridization of 16-day-old prepuberal testis resulted in the labeling of spermatocytes and presumptive interstitial macrophages. RNA from Sertoli cells, but not pachytene spermatocytes or round spermatids, hybridized with human TNF alpha receptor p60 probe in Northern blot analysis. These results are consistent with the working hypothesis that spermatids release TNF alpha, which is detected by Sertoli cells and may serve as a paracrine factor, regulating an as yet unidentified process in spermatogenesis.