Presence of brain-derived neurotrophic factor in brain and human and rat but not mouse serum detected by a sensitive and specific immunoassay.

Brain-derived neurotrophic factor (BDNF) is one of several endogenous proteins that play key roles in neuronal development and homeostasis. We describe here the characterization and use of a sensitive and specific enzyme-linked immunoassay (EIA) for BDNF protein. Recombinant BDNF was detected at concentrations as low as 10 pg/ml, whereas the EIA did not detect NT-3, NT-4/5, or NGF at concentrations as high as 100 ng/ml. Because BDNF protein sequences are identical among humans, mice, and rats, we utilized the BDNF EIA to detect BDNF in the circulation or brain regions of these species. High concentrations of BDNF were detected in human and rat serum, and up to 50-fold lower BDNF levels were present in citrated human or rat plasma. The BDNF signal (66-141 pg/ml) in 20% human plasma was completely blocked by pre-exposure of plasma to a monoclonal antibody (Mab) specific for BDNF but not by exposure to 5-fold greater concentrations of an irrelevant Mab of the same isotype (IgG1). There was a significant and positive correlation (r = +0.86) between plasma levels of BDNF and serotonin, an indoleamine that is specifically released from activated platelets. These results are consistent with the view that the BDNF detected in human and rat plasma is derived from platelet degranulation, and that circulating levels of BDNF are negligible. In contrast to human or rat serum, mouse serum contained no detectable BDNF. However, BDNF protein was readily detectable at 108-256 ng/g of tissue in hippocampus, frontal cortex, and neostriatum of mice and rats. Thus, the failure to detect BDNF in murine serum was not due to an assay defect but highlights a significant species difference in the tissue-specific expression of BDNF that may be of biological importance. The presence of BDNF protein in blood and brain regions at quantities which greatly exceed those described for NGF confirm the abundant distribution of this broadly-acting neurotrophic factor.

Modulation of MHC class II expression in human cells by dexamethasone.

The effect of dexamethasone on human MHC class II expression was examined on various cell types including lymphocytes, monocytes, and epithelial cells. Dexamethasone decreased the surface expression of HLA-DR and -DP, but not HLA-DQ, on lymphocytic cell lines that constitutively express these molecules. In addition, dexamethasome down-regulated the mRNA levels of HLA-DRA, but not of HLA-DQB, in Jijoye cells, a human lymphoblastic cell line. Similarly, dexamethasone decreased HLA-DR expression on epithelial and monocytic cell lines that express HLA-DR upon IFN-gamma treatment. In total, these results suggest that dexamethasone inhibits both constitutive and IFN-gamma-inducible MHC class II expression in several cell types. Moreover, these results indicate that the inhibitory effect of dexamethasone on MHC class II expression is selective for HLA-DR and -DP but not HLA-DQ. Possible mechanisms of dexamethasone-mediated regulation of MHC class II expression are discussed.

Purification and identification of brain-derived neurotrophic factor from human serum.

Brain-derived neurotrophic factor (BDNF), a 27-kDa noncovalently linked homodimer with subunits of approximately 13.5 kDa as viewed by SDS-PAGE, is thought to be primarily produced in the central nervous system. We report here the isolation of BDNF from pooled normal human sera, using a two-step purification process followed by SDS-PAGE, transfer to a polyvinylidene difluoride membrane, and subsequent identification of the protein by sequence analysis of the appropriate band(s) from the membrane. The level of BDNF in pooled human sera was estimated to be approximately 15 ng/ml as determined by an enzyme-linked immunosorbant assay. The average for six individuals was 18.9 +/- 5.7 ng/ml. There is an approximately 200-fold increase in the levels of BDNF in serum relative to plasma. Results from experiments using differential centrifugation suggest that the source of this increase is due to release from platelets. The presence of high levels of BDNF in serum suggests a role for this neurotrophin either in nerve repair at sites of injured tissue or in nonneuronal functions.

L cells expressing DQ molecules of the DR3 and DR4 haplotypes: reactivity patterns with mAbs.

cDNAs coding for the HLA class II DR and DQ alpha and beta chains of the diabetogenic haplotypes DR3 and DR4 were introduced into a mammalian expression vector and transfected into L-cell mouse fibroblasts to produce cells expressing individual human class II molecules. Stable L transfectants were generated expressing each of the DR or DQ isotypes of the cis-encoded alpha and beta chains of the DR3 or DR4 haplotypes, as well as the trans-encoded alpha and beta chains of the DQ molecules of the two haplotypes. However, isotype mismatched combinations (DR alpha/DQ beta or DQ alpha/DR beta) did not result in any stable transfectants. The stable DQ L-cell transfectants obtained, along with homozygous B-cell lines expressing the DQ2 and DQ8 specificities, were tested against a large panel of twentyone anti-HLA class II monoclonal antibodies (mAbs). Their unusual reactivity patterns are described including the failure of most "pan-DQ" mAbs to react with all DQ expressing L-cell transfectants. Interestingly, some mAbs react with certain alpha beta heterodimers expressed on B-LCL but fail to recognize the same heterodimers expressed on the transfectants. This is suggestive of minor structural modifications that class II molecules undergo depending on the cells they are expressed on.

Oncostatin-M is an autocrine growth factor in Kaposi's sarcoma.

Oncostatin-M is a cytokine produced by macrophages and activated T lymphocytes that has recently been shown to be a mitogen for AIDS-related Kaposi's sarcoma (KS)-derived spindle cells. The significance of oncostatin-M production in AIDS-related KS in vivo, however, remains unknown. In this study we wanted to determine whether oncostatin-M is expressed in vivo in patients with HIV-I-related KS, define the cell types that express this cytokine, and compared with the control tissues from HIV-I-negative individuals. A second objective of our study was to define the expression of oncostatin-M in AIDS-KS-derived spindle cell isolates cultured in vitro and to determine whether oncostatin-M is an autocrine growth factor for these KS cells. We have determined that oncostatin-M is not expressed in any of the several organs examined in control cases, whereas the tumor tissue obtained from the skin biopsies of HIV-I-infected cases with KS displayed oncostatin-M expression in the spindle cell components of the tumor, as well as the cells lining the vascular structures, smooth muscle cells lining the eccrine sweat glands, and the epidermal layers of the skin. Furthermore, uninvolved skin of patients with HIV-related KS express oncostatin-M in the cells lining normal vessels. The mRNA polymerase chain reaction analysis confirmed findings in the primary tissues and showed expression in all of the AIDS-KS-derived spindle cell isolates examined. We have also shown with the use of oncostatin-M-specific antisense oligodeoxynucleotides that KS cell proliferation is inhibited, which correlated with a more precipitous decline in the production of interleukin-6 by these cells. We conclude that oncostatin-M is only expressed in the skin and KS tumor of HIV-I-infected individuals. Furthermore, we provide evidence that oncostatin-M is an autocrine growth factor for KS.

Relationship of serum levels of oncostatin M to AIDS-related and classic Kaposi's sarcoma.

Serum levels of circulating oncostatin-M (OM) were compared among cases of Kaposi's sarcoma associated with acquired immune deficiency syndrome (AIDS-KS) and multiple controls, including a homosexual man infected with human immunodeficiency virus type 1 (HIV-1), an HIV-1-uninfected homosexual man, and a heterosexual man; and among classic KS cases and heterosexual controls. Cases were selected from abstracts collected by a population-based cancer registry and from local AIDS clinics. Controls for the AIDS-KS cases were matched to the cases by age, sex, and race and were either friends of the cases or residents from the cases' neighborhoods; controls for the classic KS cases were similarly matched, but were obtained solely from neighborhood residents. Blood samples were obtained from participants, serum levels of OM were determined by enzyme-linked immunosorbent assay (ELISA), and CD4 cell counts were obtained by flow cytometry. Geometric mean levels of OM were compared among the risk groups adjusted for age and CD4 cell count. No differences in adjusted OM levels were found between AIDS-KS cases and HIV-1-infected homosexual controls (8.4 pg/ml vs. 10.2) or between classic KS cases and controls (13.3 pg/ml vs. 9.6); however the HIV-1-infected controls (both homosexual and heterosexual) matched to the AIDS-KS cases had higher levels than did the HIV-1-infected cases and controls. Among the HIV-1-infected groups, an inverse correlation between OM and CD4 cell count was observed and was statistically significant for the cases. Among all heterosexual controls (matched to either case group), serum OM was inversely related to age.(ABSTRACT TRUNCATED AT 250 WORDS)

The binding pattern of a neutralizing monoclonal antibody to mutant oncostatin M molecules is correlated with functional activity.

Oncostatin M (OM) is a member of the cytokine family that includes leukaemia inhibitory factor (LIF), granulocyte colony stimulating factor (G-CSF), and interleukin 6 (IL-6). We previously reported the characterization of a monoclonal antibody (MAb) to OM, termed OM2, which neutralizes its functional activity. To gain information about the epitope detected by this MAb, we utilized a sandwich enzyme-linked immunoassay (EIA) to examine OM2 binding on a series of mutant recombinant OM molecules generated by site-directed mutagenesis, encompassing amino acid insertions, deletions, or alterations throughout the molecule. Carboxy-terminal deletions of a putative amphiphilic alpha helix past residue 185 abrogated binding of the OM2 MAb; alteration of a hydrophobic residue in the helix to a neutral one also prevented antibody binding. Analysis of mutants in which cysteines involved in intrachain disulfide binding were changed to serines revealed that one of two disulphide bonds was essential for OM2 binding. Two mutant molecules containing deletions in the amino-terminal one-fourth of the molecule were not bound by OM2, while a third mutant OM molecule with a C-terminal proximal internal deletion outside of the amphiphilic alpha helix was bound. The binding pattern of MAb OM2 to the OM mutant molecules correlated well with their functional activities. The data suggest that residues in both the C-terminal alpha helix and N-terminal one-fourth of the molecule are involved in neutralizing antibody binding and functional activity.(ABSTRACT TRUNCATED AT 250 WORDS)

Detection of oncostatin M in human plasma and serum by a sensitive enzyme immunoassay.

A sensitive and specific enzyme immunoassay was developed for detecting oncostatin M (OM) in human plasma and serum. The assay utilizes three anti-OM monoclonal antibodies that recognize mutually exclusive epitopes, including a neutralizing epitope. A sensitivity of 24 pg/ml was routinely obtainable. The assay showed no cross-reactivity with leukemia inhibitory factor (LIF) or interleukin-6 (IL-6), other members of the cytokine family that includes OM. The utility of the enzyme immunoassay (EIA) was demonstrated by detecting the time-dependent accumulation of OM in plasma from lipopolysaccharide (LPS)-treated human whole blood. The concentration of OM in human sera from normal donors was generally below the detection limits of the assay. However, concentrations of OM greater than 25 pg/ml were found in 17 of 212 serum samples from apparently normal donors. The detection of OM in human plasma and serum demonstrates that the EIA could be a useful tool in examining the role of OM in physiologic and pathologic states.

Correlation of oncostatin M secretion by human retrovirus-infected cells with potent growth stimulation of cultured spindle cells from AIDS-Kaposi's sarcoma.

Oncostatin M (OM), a 30-kDa glycoprotein, recently was identified as a major growth-promoting factor in the conditioned medium (CM) of the 38-0 cell line, a CD4,+ chronically human T lymphotropic virus type (HTLV)-II-infected, transformed T cell line. CM 38-0 induced the proliferation of spindle cells cultured in vitro from AIDS-associated Kaposi's sarcoma (AIDS-KS) cells. To determine how much of the AIDS-KS cell growth activity present in 38-0 CM was because of the presence of OM, we depleted OM by using specific mAb-affinity chromatography. OM purified from this CM stimulated AIDS-KS cell growth in a concentration-dependent fashion. The effluent, completely depleted of OM, failed to induce growth of AIDS-KS cells. To detect the constitutive release of OM by cells acutely or chronically infected with either HTLV-I, HTLV-II, or HIV-1, we utilized an enzyme-linked immunoassay. Whereas the chronically infected cells released significant levels of OM, the acutely infected cells released little or no OM. The presence of OM in HIV-1-infected T-cell CM correlated completely with AIDS-KS cell growth activity. Infrequently, low level AIDS-KS cell growth activity was seen in the absence of OM. This correlated with relatively high levels of IL-6 in the CM. In a CM-containing OM in the absence of detectable IL-6, a neutralizing antibody to OM completely abrogated KS cell growth activity. The presence of specific oncostatin M receptors on the KS cell lines was confirmed by cross-linking experiments. The results shown here suggest that T cells chronically infected with HIV-1 can secrete OM, which may play a role in the initiation or progression of AIDS-KS lesions, either alone, or in concert with IL-6.

Abrogation of the antiproliferative activity of oncostatin M by a monoclonal antibody.

Oncostatin M (OM) is a novel cytokine which exhibits pleiotropic effects on a wide variety of normal and transformed cell lines. To determine some of the physiological functions of OM we have characterized several monoclonal antibodies to the recombinant molecule. Antibodies OM1 and OM2 bound native, but not denatured OM, suggesting they recognize non-contiguous epitopes. A third antibody, OM6, bound predominantly denatured OM. Of the two antibodies which detect discontinuous epitopes, OM2, but not OM1, was identified as a neutralizing antibody based on its ability to abrogate OM activity in the growth inhibition assay (GIA) and to inhibit OM binding in the radioreceptor assay (RRA). OM2 was equally effective in abrogating the functional effects of either natural or recombinant OM, thereby demonstrating that the active sites of these molecules are structurally similar, if not identical.

Oncostatin M as a potent mitogen for AIDS-Kaposi's sarcoma-derived cells.

Oncostatin M, a cytokine produced by activated lymphoid cells, regulates the growth and differentiation of a number of tumor and normal cells. In contrast to its effects on normal endothelial and aortic smooth muscle cell cultures, Oncostatin M was a potent mitogen for cells derived from acquired immunodeficiency syndrome-related Kaposi's sarcoma (AIDS-KS). After exposure to Oncostatin M, AIDS-KS cells assumed a spindle morphology, had an increased ability to proliferate in soft agar, and secreted increased amounts of interleukin-6. Oncostatin M RNA and immunoreactive Oncostatin M protein were found in AIDS-KS-derived cell isolates. These results suggest that Oncostatin M may play a role in the pathogenesis of AIDS-KS.

Macrophage-derived factors increase low density lipoprotein uptake and receptor number in cultured human liver cells.

Recent evidence suggests the possibility that macrophages can influence lipoprotein metabolism. Therefore we investigated the ability of cultured macrophages to alter low density lipoprotein (LDL) uptake in a human liver cell line (HepG2). Conditioned media from phlogogenic-induced mouse peritoneal macrophages or from a human macrophage cell line stimulated with endotoxin increased HepG2 LDL uptake by as much as 60-70%. The increase was due, in part, to a significant macrophage-induced 40% increase in the number of LDL receptors per cell. Although macrophage conditioned media inhibited HepG2 cholesterol synthesis, the LDL receptor up-regulation did not appear to be due to the effects on cholesterol synthesis. The LDL receptor stimulatory activity was sensitive to proteolysis and heat. Its molecular mass was approximately 20 kDa based on gel filtration. Several macrophage secretory proteins were tested in HepG2 cultures for LDL uptake stimulation. Of these, oncostatin M (approximately 18 kDa by gel filtration) gave the strongest response. The rank order for LDL uptake stimulation was oncostatin M much greater than interleukin 6 = interleukin 1 = transforming growth factor-beta 1. A neutralizing antibody directed against oncostatin M inhibited the ability of conditioned media to up-regulate LDL receptors by 85%. Thus, our results indicate that macrophages can secrete several proteins that up-regulate LDL receptors in HepG2 cells and that most of the up-regulatory activity in macrophage conditioned media appears to be due to oncostatin M.

Oncostatin M up-regulates low density lipoprotein receptors in HepG2 cells by a novel mechanism.

Oncostatin M is a growth regulatory protein secreted by macrophages and activated T lymphocytes. In a hepatoma cell line (HepG2) the polypeptide very potently increased low density lipoprotein (LDL) uptake with an EC50 of 0.1-0.2 nM. The stimulation of LDL uptake was detectable by 2 h, was maximal by 8 h, and remained elevated through 20 h of oncostatin M incubation. In a similar fashion, oncostatin M also increased the number of cell surface LDL receptors by a mechanism that was inhibited by cycloheximide or the protein kinase C inhibitor H-7. Oncostatin M stimulation of LDL uptake and receptor protein occurred regardless of the state of cholesterol-dependent regulation of HepG2 LDL receptor (i.e. cells incubated in medium containing lipoproteins responded to the same extent as did cells incubated in the absence of lipoproteins). No significant effects were observed on sterol synthesis over 8 h or on DNA synthesis over 24 h. Oncostatin M induced rapid alterations in HepG2 phospholipid metabolism. Within 5-15 min there was a 20-50% increase in incorporation of 32P into several classes of phospholipids, including the phosphoinositides. Radiolabeled diacylglycerol levels were elevated 20% by 2 min and nearly 50% by 15 min. In addition, the polypeptide induced rapid increased (within 1 min) in phosphorylation of HepG proteins on tyrosine residues. Stimulation of both phosphotyrosine and LDL receptor up-regulation by oncostatin M was decreased by the tyrosine kinase inhibitor genistein. We propose that oncostatin M up-regulates HepG2 LDL receptor expression by a mechanism that includes stimulation of a tyrosine kinase followed by generation of phospholipid-related second messengers.

Mapping of distinct serologic and T cell recognition epitopes on an HLA-DR beta-chain.

A new DR beta-chain allele is defined that is identical to the previously described DR6b molecule except for the first hyperpolymorphic region, where the new allele displays the same polymorphisms found on DR8 and DR12 genes. Two distinct epitopes have been mapped on this new allele. The polymorphism in common with DRw8 and DRw12 is recognized by mAb GS313-9D11. However, alloreactive T cell clones specific for DR6b cells (Dw9) recognize this allele, whereas Dw8-specific T cell clones do not. The mAb determinant maps to the first beta-sheet and probably involves a polymorphic residue lying outside the helix. The binding of mAb 9D11 to this region does not interfere with TCR binding. Alloreactive T cell recognition is associated with polymorphisms located predominantly on the alpha-helical portion of the molecule.

Structural requirements for recognition of the HLA-Dw14 class II epitope: a key HLA determinant associated with rheumatoid arthritis.

Although HLA genes have been shown to be associated with certain diseases, the basis for this association is unknown. Recent studies, however, have documented patterns of nucleotide sequence variation among some HLA genes associated with a particular disease. For rheumatoid arthritis, HLA genes in most patients have a shared nucleotide sequence encoding a key structural element of an HLA class II polypeptide; this sequence element is critical for the interaction of the HLA molecule with antigenic peptides and with responding T cells, suggestive of a direct role for this sequence element in disease susceptibility. We describe the serological and cellular immunologic characteristics encoded by this rheumatoid arthritis-associated sequence element. Site-directed mutagenesis of the DRB1 gene was used to define amino acids critical for antibody and T-cell recognition of this structural element, focusing on residues that distinguish the rheumatoid arthritis-associated alleles Dw4 and Dw14 from a closely related allele, Dw10, not associated with disease. Both the gain and loss of rheumatoid arthritis-associated epitopes were highly dependent on three residues within a discrete domain of the HLA-DR molecule. Recognition was most strongly influenced by the following amino acids (in order): 70 greater than 71 greater than 67. Some alloreactive T-cell clones were also influenced by amino acid variation in portions of the DR molecule lying outside the shared sequence element.

Multiple regions of HLA-DR beta 1 chains determine polymorphic epitopes recognized by monoclonal antibodies.

To investigate the locations of antibody binding epitopes on HLA class II molecules, four DR4/7 beta 1 hybrid cDNA were constructed by exchanging the DNA encoding the NH2-terminal portions (amino acids 1 to 40) or the COOH-terminal portions (amino acids 41 to 94) of the first domains of DR4 beta 1- and DR7 beta 1-chains, in association with DNA encoding either the DR4 beta 1 or DR7 beta 1 second domains. Transfectants expressing a DR alpha cDNA and a wild-type DR4 beta 1 or DR7 beta 1 cDNA or one of four hybrid DR4/7 beta 1 cDNA were produced, and the binding to the transfectants of anticlass II mAb, which detect polymorphic epitopes on either DR4 or DR7 molecules, was analyzed. Four different patterns of mAb binding to the transfectants were observed, indicating that multiple regions of DR beta 1-chains play the predominant roles in the contributions of these chains to polymorphic epitopes recognized by mAb on intact molecules. The relevant regions of these chains and the number of mAb that recognize the associated polymorphic epitopes are: 1) the COOH-terminal portion of the first domain of DR4 beta 1; a DR4-specific mAb, 2) the NH2-terminal portion of the first domain of DR7 beta 1; two mAb, including a DR7-specific mAb, 3) the NH2-terminal portion of the first domain of DR4 beta 1; seven mAb, and 4) the second domain of DR4 beta 1; one mAb.

HLA-DQ alpha polymorphisms: oligonucleotide probes characterize the contribution of first and second domains to electrophoretic variants.

Polymorphism is known to exist within the HLA-DQ alpha locus in the human major histocompatibility complex, although such polymorphism may be "silent" in standard HLA typing. However, DQ alpha polymorphism may be functionally significant, either through DQ alpha epitopes functioning directly in the immune response or by affecting tertiary conformation of Ia molecules through differential alpha/beta pairing. We have previously defined a particular DQ alpha polymorphism through reactivity with a monoclonal antibody and restriction fragment length polymorphism pattern. We now characterize this DQ alpha polymorphism through two-dimensional gel electrophoretic analysis and identify a subset of DQ alpha molecules with unique characteristics. Investigation of these allelic variants using synthetic oligonucleotide probe analysis of genomic DNA suggests a localization of the DNA region encoding the DQ alpha 5 epitope and suggests possible evolutionary mechanisms accounting for these unique patterns.

HLA-DQw3-related determinants: analysis of subunit and spatial relationships.

More polymorphism exists among DQ region gene products than is suggested by present serologic definitions of these class II molecules. DQ beta polymorphism among haplotypes carrying the DQw3 specificity is considerable. The TA10 specificity is present on one allele of at least three different DQ beta alleles that carry the DQw3 specificity. We have examined a series of monoclonal antibodies directed against different DQ beta alleles carrying the DQw3 specificity to determine subunit and spatial relationship among the epitopes detected by these antibodies. The antibodies were examined by Western blotting and for their ability to inhibit the binding of fluoresceinated antibodies on either TA10+ or TA10- DQw3 haplotypes. Our results reveal that (1) multiple DQw3-related epitopes exist; (2) several anti-DQw3-related antibodies generated against TA10- DQw3 molecules are unable to inhibit the binding of a TA10-specific antibody on a TA10+ haplotype while strongly inhibiting binding of an antibody detecting the reciprocal DQ beta polymorphism on a TA10- DQw3 haplotype; and (3) there is a strong requirement for three-dimensional conformation in the formation of the majority of the epitopes examined here. Analysis of previously published amino acid sequences for the haplotypes investigated here suggest that charge changes at amino acids 45, and 57, respectively, may have a significant effect in changing the spatial relationship between the DQw3-related epitope(s) and other polymorphic determinants on DQ beta chains.

Equine class II MHC antigens: identification of two sets of epitopes using anti-human monoclonal antibodies.

Six mouse and 13 rat monoclonal antibodies (mAb) recognizing HLA-DR, DQ and DP antigens were used for the detection of cell surface class II MHC antigens of equine lymphocytes. The monoclonal antibodies were tested against peripheral blood lymphocytes (PBL) from a panel of thoroughbred horses, using two-color fluorescence flow cytometry. Seven of these mAbs reacted with both surface immunoglobulin positive (sIg+) and surface immunoglobulin negative (sIg-) lymphocytes. sIg+ cells stained consistently brighter than sIg- cells. The fluorescence pattern did not vary from donor to donor for each of the mAbs tested, except for SFR1-MI.2, which reacted with a variable intensity with cells from 47 of 53 horses tested. Immunoprecipitation with mAb SFR1-MI.2 and analysis by two-dimensional electrophoresis demonstrated the presence of light and heavy chains equivalent to HLA class II alpha and beta chains. Antibody N297 (DQ specific), previously shown to react with an epitope expressed on human B cells but not on mitogen-induced T cells, reacted only with sIg+ cells in 42 of 53 horses tested. The lack of staining of horses sIg- cells with N297 may be due to a low or lack of expression of this determinant on these cells or to a weak cross-reactivity of this antibody with equine antigens.

Alternative splicing of HLA-DQB transcripts and secretion of HLA-DQ beta-chain proteins: allelic polymorphism in splicing and polyadenylylation sites.

HLA class II antigens are highly polymorphic cell-surface proteins involved in initiation and regulation of the immune response. Allelic sequence variation primarily affects the structure of the first external domains of alpha and beta component chains. Here we provide evidence for other types of allelic polymorphism for the genes encoding these chains. Sequences of two cDNA clones corresponding to HLA-DQB mRNAs from an HLA-homozygous cell line exhibit both alternative splicing and read-through of polyadenylylation. Furthermore, alternative splicing that deletes the transmembrane exon is associated with only a subset of HLA-DQB alleles, while the polyadenylylation-site read-through is found in a larger subset. This suggest that polymorphic cis-acting elements within the HLA-DQB gene control both processing steps. Proteins, presumably encoded by alternatively spliced mRNAs lacking transmembrane exons, are immunoprecipitated with a monomorphic monoclonal antibody directed against HLA-DQ. These proteins are found in supernatants of cultured cell lines for which secretion is predicted, but not in those of cell lines that do not contain alternatively spliced mRNAs.