Search

Anti-alpha 4 integrin mAb coprecipitated CD81 (TAPA-1), a 25-kDa cell surface protein, from various alpha 4 beta1 -positive hemopoietic cell lines, including Molt4, Jurkat, Ramos, and alpha 4-transfected K562 (KX4C4) cells. In reciprocal experiments, the integrin alpha 4 beta 1 (VLA4, CD49d/CD29) could be reprecipitated from CD81 immunoprecipitates. Anti-alpha 4 integrin mAb also coprecipitated CD81 from the alpha 4 beta 7-positive B cell line RPMI 8866. In contrast, no CD81 was identified in alpha 2 beta 1, alpha 5 beta 1, or alpha L beta 2 immunoprecipitates. Abs to other members of the transmembrane-4 superfamily, including CD53, CD63, and CD82, also coprecipitated alpha 4 beta 1. As shown by confocal microscopy, CD81 and CD82 colocalized with alpha 4 beta 1 in cell surface clusters. The cytoplasmic domain of the alpha 4 integrin was not necessary for alpha 4 beta 1/CD81 association, nor was the association influenced by divalent cations, EDTA, integrin-activating mAb, or alpha 4 subunit cleavage. Notably, two independent alpha 4 adhesion-deficient mutants (D346E and D408E) were deficient in their ability to associate with CD81. Thus, CD81 and other transmembrane-4 superfamily members may participate in functionally relevant interactions with alpha 4 beta 1 and other integrins.

The integrin beta 4 subunit often undergoes proteolytic cleavage within its long cytoplasmic tail to yield a characteristic protein pattern of 205, 165, and 125 kDa. The results in this study suggest that beta 4 cleavage often occurs during or after cell lysis, where it was readily inhibitable by calcium chelators (EDTA, EGTA) and inhibitors of cysteine proteases (E64c, leupeptin). The cleavage of beta 4 is catalyzed by a calpain-like enzyme because (i) it requires calcium, (ii) it is mimicked by purified milli-calpain, and (iii) it is inhibited by several calpain inhibitors including the calpain-specific inhibitor calpastatin. Within intact cells, cleavage of beta 4 was cell type-specific and observed only when the cells were made permeable to calcium. Substantial cell viability was retained during beta 4 cleavage induced by ionomycin plus calcium, indicating that cleavage within intact cells was not necessarily a consequence of cell death. However, manipulations of cells including suspension, synchronization, and stimulation with serum, phorbol esters, or other agents all failed to induce cleavage, suggesting that if cleavage is physiologically relevant, it is not easily duplicated in vitro. Analysis of multiple cell types showed a wide variation in beta 4 sensitivity to proteolytic cleavage, suggesting that this process might be differentially regulated depending on the cellular environment.

Upon Ag stimulation, an arsonate-specific murine T cell clone exhibited a rapid but transient increase in cell adhesion to collagen, fibronectin, and laminin. This increase in cell adhesion was not observed when a mutant T cell clone lacking TCR expression was utilized. However, upon stimulation by phorbol esters, both parent and mutant T cell clones exhibited a similar transient increase in adhesion to the three matrix proteins. The observed cell adhesion was extensively inhibited by antibodies to the integrin beta 1 subunit, indicating the involvement of VLA proteins. Despite changes in the adhesive properties, there was essentially no difference in the expression of VLA-1, -3, -4, -5, and -6 between resting and stimulated T cells. Together these results suggest that Ag stimulation transmits signals via the TCR complex resulting in a rapid, but transient, up-regulation of matrix protein binding by VLA proteins already present at the cell surface. Because the appropriate reagents that recognize individual mouse VLA proteins were not available, we used the human T cell line Jurkat to demonstrate that T cell binding to collagen, laminin, and fibronectin is mediated largely by VLA-2, VLA-6, and a combination of VLA-5 and VLA-4, respectively.

Extracellular matrix metalloproteinase inducer (EMMPRIN; CD147) is a heavily glycosylated protein containing two immunoglobulin superfamily domains. It is enriched on the surface of tumor cells and stimulates the production of matrix metalloproteinases (MMPs) by adjacent stromal cells. Here we use CD147 transfectants and immobilized recombinant CD147-Fc fusion protein to show that CD147/FMMPRIN engages in a homophilic interaction, predominantly through the first immunoglobulin domain. Anti-CD147 antibody 8G6 and recombinant CD147-Fc fusion protein markedly inhibited not only homophilic interaction, but also the production of secreted MMP-2 by breast cancer cell line MDA-435 and the MMP-2-dependent invasion of MDA-435 cells through reconstituted basement-membrane Matrigel. Purified native CD147 induced the production of secreted MMP not only by dermal fibroblasts (MMP-1) but also by MDA-435 cells themselves (MMP-2), suggesting homophilic CD147-binding may occur in the context of both heterotypic and homotypic cell-cell interactions. Purified deglycosylated CD147 failed to induce MMP-1 or MMP-2, but instead antagonized the MMP-1-inducing activity of purified native CD147. Our results suggest that homophilic CD147 interactions may play a key role in MMP-2 production and tumor cell invasion, and that perturbation of this molecule may have potential therapeutic uses in the prevention of MMP-2 and MMP-1-dependent cancer metastasis.

In earlier work we established that phosphoinositide 4-kinase (PI 4-kinase) may associate with transmembrane 4 superfamily (TM4SF, tetraspanin) proteins, but critical specificity issues were not addressed. Here we demonstrate that at least five different TM4SF proteins (CD9, CD63, CD81, CD151 and A15/TALLA1) can associate with a similar or identical 55 kDa type II PI 4-kinase. These associations were specific, since we found no evidence for other phosphoinositide kinases (e.g. phosphoinositide 3-kinase and phosphoinositide-4-phosphate 5-kinase) associating with TM4SF proteins, and many other TM4SF proteins (including CD82 and CD53) did not associate with PI 4-kinase. CD63-PI 4-kinase complexes were almost entirely intracellular, and thus are distinct from other TM4SF-PI 4-kinase complexes (e.g. involving CD9), which are largely located in the plasma membrane. These results suggest that a specific subset of TM4SF proteins may recruit PI 4-kinase to specific membrane locations, and thereby influence phosphoinositide-dependent signalling.

The anti-integrin beta1 MAb 15/7 sometimes may be a reporter of integrin activation or ligand occupancy. However, certain beta1 tail deletions eliminate ligand binding despite inducing maximal constitutive 15/7 expression [Puzon-Mclaughlin et al. (1 996): J Biol Chem 271:16580-16585]. Here we describe beta1 tail mutations (e.g., double point mutations [D759L/F763L, F766L/E769L], or replacement of the beta1 tail by the beta5 tail) that prevent rather than induce constitutive appearance of the 15/7 epitope. Despite variable losses of constitutive 15/7 epitope, these mutants all retained a similar inducible 15/7 epitope component as seen upon incubation with GRGDSP peptide ligand. In addition, constitutive 15/7 expression did not correlate with integrin localization into focal adhesions. In conclusion, we show for the first time for a fully functional integrin that specific mutations within the beta1 tail can down-regulate the constitutive appearance of an extracellular conformation defined by MAb 15/7. Because this regulation occurs away from the ligand binding site and does not correlate with responsiveness to integrin ligand, cell adhesion, or localization into focal adhesions, a novel type of conformational regulation is suggested.

The alpha3beta1 integrin forms complexes with other cell-surface proteins, including transmembrane-4 superfamily (TM4SF) proteins (e. g. CD9, CD53, CD63, CD81, and CD82). To identify additional cell-surface proteins associated with alpha3beta1 integrin, a monoclonal antibody selection protocol was developed. Mice were immunized with integrin alpha3beta1-containing complexes isolated from HT1080 fibrosarcoma cells, and then 712 hybridoma clones were produced, and 95 secreted antibodies that recognized the HT1080 cell surface. Among these, 12 antibodies directly recognizing integrin alpha3 or beta1 subunits were eliminated. Of the remaining 83, 16 co-immunoprecipitated proteins that resembled integrins under non-stringent detergent conditions. These 16 included 15 monoclonal antibodies recognizing EMMPRIN/basigin/OX-47/M6, a 45-55-kDa transmembrane protein with two immunoglobulin domains. The EMMPRIN protein associated with alpha3beta1 and alpha6beta1, but not alpha2beta1 or alpha5beta1, as shown by reciprocal immunoprecipitation experiments. Also, association with alpha3beta1 was confirmed by cell-surface cross-linking and immunofluorescence co-localization experiments. Importantly, EMMPRIN-alpha3beta1 complexes appear not to contain TM4SF proteins, suggesting that they are distinct from TM4SF protein-alpha3beta1 complexes.

Anti-alpha 4 integrin mAb coprecipitated CD81 (TAPA-1), a 25-kDa cell surface protein, from various alpha 4 beta1 -positive hemopoietic cell lines, including Molt4, Jurkat, Ramos, and alpha 4-transfected K562 (KX4C4) cells. In reciprocal experiments, the integrin alpha 4 beta 1 (VLA4, CD49d/CD29) could be reprecipitated from CD81 immunoprecipitates. Anti-alpha 4 integrin mAb also coprecipitated CD81 from the alpha 4 beta 7-positive B cell line RPMI 8866. In contrast, no CD81 was identified in alpha 2 beta 1, alpha 5 beta 1, or alpha L beta 2 immunoprecipitates. Abs to other members of the transmembrane-4 superfamily, including CD53, CD63, and CD82, also coprecipitated alpha 4 beta 1. As shown by confocal microscopy, CD81 and CD82 colocalized with alpha 4 beta 1 in cell surface clusters. The cytoplasmic domain of the alpha 4 integrin was not necessary for alpha 4 beta 1/CD81 association, nor was the association influenced by divalent cations, EDTA, integrin-activating mAb, or alpha 4 subunit cleavage. Notably, two independent alpha 4 adhesion-deficient mutants (D346E and D408E) were deficient in their ability to associate with CD81. Thus, CD81 and other transmembrane-4 superfamily members may participate in functionally relevant interactions with alpha 4 beta 1 and other integrins.

Integrin alpha 4 beta 1 is a leukocyte receptor for fibronectin and vascular cell adhesion molecule-1 (VCAM-1). It is important in inflammatory recruitment of leukocytes, lymphopoiesis, and a number of development events. Here we have mapped a panel of functional monoclonal antibodies (mAbs) recognizing the integrin alpha 4 chain, using murine/human chimeric constructs expressed in COS7 cells. We find that: 1) mAbs that induce homotypic aggregation (epitope A mAbs) map to the most N-terminal 100 amino acids of the human alpha 4 chain; 2) mAbs that block adhesion of alpha 4 beta 1 to VCAM-1 and fibronectin (epitope B mAbs) map to a 52-amino-acid region between residues 152 and 203 of human alpha 4; 3) epitope B mAbs that do or do not induce aggregation (epitope B2 and B1 mAbs, respectively) map to the same regions and are therefore indistinguishable by this analysis; 4) mAbs that neither induce homotypic aggregation nor block adhesion (epitope C mAbs) map to a distinct region of the molecule comprising amino acids 422-606. The N-terminal region of the alpha 4 chain identified by functional A and B epitope mAbs does not correspond to ligand binding sites identified in other alpha subunits, such as cation binding sites or the "I-domain," which alpha 4 lacks, and thus represents a novel site for epitope functionality among the integrins.

To investigate the role of the alpha subunit cytoplasmic domain in the regulation of VLA-2 functional activity, we expressed several chimeric and deleted forms of the alpha 2 subunit in two different human cell lines, K562 and RD. Each mutant construct formed surface VLA-2 heterodimers as efficiently as wild type alpha 2 subunit, except for a construct (X2CO1127) truncated just before the consensus GFFKR cytoplasmic domain motif, that was not expressed at the cell surface. Truncation of the alpha 2 cytoplasmic domain just after the GFFKR motif resulted in a complete loss of constitutive activity of VLA-2 in RD cells. If the integrin was already constitutively inactive, as in K562 cells, the cytoplasmic domain deletion had no effect. In both K562 and RD cells, cytoplasmic tail deletion eliminated up-regulation of adhesion in response to the phorbol ester, phorbol 12-myristate 13-acetate (PMA). In comparison, exchange of the alpha 2 cytoplasmic domain with the alpha 4 or alpha 5 cytoplasmic domains had no effect on constitutive activity (in RD cells), or on constitutive inactivity (in K562 cells) and did not eliminate PMA-stimulated activity (in K562 or RD cells). These results clearly demonstrate that the cytoplasmic domain of an alpha chain (not necessarily from alpha 2 itself) is required to maintain VLA-2 constitutive activity and to allow a responsiveness to PMA stimulation. In cases where VLA-2 was either constitutively inactive (as in K562 cells) or inactive due to cytoplasmic domain deletion (e.g. in RD cells), agents such as Mn2+ or the anti-beta 1 monoclonal antibody TS2/16 caused a marked increase in adhesive function, thus proving that the integrins were not irreversibly inactive, and that cellular regulatory constraints could be bypassed by extracellular stimuli.

To assess directly the functional role of the integrin VLA-3 (alpha 3 beta 1), we transfected human alpha 3 cDNA into erythroleukemia (K562) cells and rhabdomyosarcoma (RD) cells. The resulting transfectants (KA3 and RA3) expressed alpha 3 beta 1 on the cell surface as confirmed using a panel of nine anti-alpha 3 monoclonal antibodies. Neither of the transfected cells exhibited increased adhesion to the extracellular matrix proteins fibronectin, laminin, and collagen. However, the KA3 transfectants did bind strongly to the extracellular matrix deposited by epidermal and carcinoma cell lines, allowing the cells to attach and spread. Binding to this cell-deposited ligand, probably containing epiligrin/kalinin, was specific to VLA-3 and could be inhibited by anti-alpha 3 antibodies and by EDTA, but not by RGD peptides. In marked contrast to other integrins (VLA-2 and VLA-4), VLA-3 showed high constitutive activity in K562 cells, but was minimally active in RD cells. Also contrasting with other beta 1 integrins, VLA-3 was minimally stimulated by the anti-beta 1 monoclonal antibody TS/216 under normal conditions. VLA-3-mediated adhesive function was well supported by either Mg2+ or Mn2+, but was almost completely abolished by the presence of 1 mM Ca2+. Surprisingly, this negative Ca2+ effect was completely overcome by the addition of the stimulatory anti-beta 1 monoclonal antibody TS2/16. Together, these results point to markedly distinct regulation for VLA-3 function compared to other beta 1 integrins. Also, all anti-VLA-3 antibodies were able to induce temperature-dependent homotypic cell aggregation of KA3 cells, but not K562 cells. However, this aggregation did not appear to be directly mediated by VLA-3 since it was not inhibited by EDTA. In addition, no enhancement of heterotypic cell-cell adhesion was observed in alpha 3-transfected cells.

We examined the effects of a stimulatory anti-beta 1 mAb (TS2/16) and different divalent cations on VLA-4-mediated cell adhesion to vascular cell adhesion molecule-1 (VCAM-1), to the fibronectin-derived CS1 peptide, and to larger fibronectin fragments. Using optimal binding conditions (in the presence of mAb TS2/16 and 1.0 mM Mn2+), the levels of VLA-4-mediated adhesion to VCAM-1 and to CS1 peptide were virtually indistinguishable, and half-maximal inhibition of adhesion to both ligands was achieved using similar levels of an anti-alpha 4 antibody. However, using suboptimal adhesion conditions, two critical differences between adhesion to CS1 peptide (or larger fibronectin fragments) and VCAM-1 were consistently observed. First, stimulation by added mAb TS2/16 had a substantially greater effect on adhesion to CS1 than to VCAM-1 and second, Ca2+ was much less able to support adhesion to CS1 than to VCAM-1. These two differences between adhesion to CS1 peptide and to VCAM-1 were most obvious among cell lines which synthesized inactive or partly active VLA-4 but were not obvious for fully active VLA-4. Together, these results not only reveal crucial differences in the mechanisms of VLA-4 binding to its two ligands, but also lead to increased understanding of the variable activation states of VLA-4. The differential ability to utilize Ca2+ displayed by VLA-4 in different states of activation and the activation of inactive or partly active VLA-4 by the addition of Mn2+ both point to divalent cation sites playing an essential role in determining VLA-4 regulation and ligand specificity.

Upon Ag stimulation, an arsonate-specific murine T cell clone exhibited a rapid but transient increase in cell adhesion to collagen, fibronectin, and laminin. This increase in cell adhesion was not observed when a mutant T cell clone lacking TCR expression was utilized. However, upon stimulation by phorbol esters, both parent and mutant T cell clones exhibited a similar transient increase in adhesion to the three matrix proteins. The observed cell adhesion was extensively inhibited by antibodies to the integrin beta 1 subunit, indicating the involvement of VLA proteins. Despite changes in the adhesive properties, there was essentially no difference in the expression of VLA-1, -3, -4, -5, and -6 between resting and stimulated T cells. Together these results suggest that Ag stimulation transmits signals via the TCR complex resulting in a rapid, but transient, up-regulation of matrix protein binding by VLA proteins already present at the cell surface. Because the appropriate reagents that recognize individual mouse VLA proteins were not available, we used the human T cell line Jurkat to demonstrate that T cell binding to collagen, laminin, and fibronectin is mediated largely by VLA-2, VLA-6, and a combination of VLA-5 and VLA-4, respectively.

In our studies of VLA-3 we have shown (i) that a single integrin (VLA-3) can bind to multiple ligands by different mechanisms, involving RGD and non-RGD sites, which are regulated differently by divalent cations. Also we showed from the primary sequence of VLA-3 that it is only distantly related to the other cleaved alpha subunits. In our studies of VLA-2 we have shown that a single integrin may have at least three functional forms, depending on the cell type where expressed. In addition, we have expressed functional VLA-2 in RD cells, resulting in both Coll and Lm binding functions in vitro, and increased tumor cell metastasis in vivo in nude mice.

The thymus is thought to be the primary central lymphoid organ in which T cells mature. Although thymic cortical and medullary compartments are distinct histologically, few antigens have been described that are absolutely acquired during the presumed intrathymic maturation pathway from cortical to medullary thymocytes. In this paper, we describe the acquisition during human intrathymic T cell maturation of a novel protein (p80) defined by a monoclonal antibody (A1G3). Although the p80-A1G3 antigen is distributed throughout the body and is not T cell specific, our study demonstrates that expression of p80-A1G3 antigen in normal human thymus is associated with thymocyte functional maturity and location in the thymus medulla. Moreover, in contrast to other markers of mature human T cells, the p80-A1G3 cell surface protein is not expressed on T6+ cortical thymocytes, and, therefore, is absolutely acquired by medullary thymocytes during T cell maturation. Thus, the p80-A1G3 antigen and the A1G3 antibody provide a heretofore unavailable system for the study of molecular events that transpire during the maturation of thymocytes.

By using a single monoclonal antibody, a novel glycoprotein complex composed of at least three distinct bands was defined on the surface of mitogen- or alloantigen-stimulated T cells. These bands (210,000, 165,000, and 130,000 Mr) were not disulfide linked and could be radio-labeled with 125I or 35S-methionine and readily detected by immunoprecipitation with the monoclonal antibody A-1A5. Only approximately 20% of normal T cells (E rosette positive) and approximately 47% non-T cells (E rosette negative) were reactive with A-1A5. However, upon activation of T cells, the amount of A-1A5 binding per cell and the percentage of positive cells significantly increased. This increase was most pronounced in the activated cell subpopulations currently undergoing cell division (S, G2, and M phases), which became 79% A-1A5 positive (after PHA stimulation) and 99% A-1A5 positive (in long-term culture with alloantigen and IL2). Resting lymphocytes contained only the 130,000 Mr band reactive with A-1A5. The two other bands (210,000 and 165,000 Mr) were markers for T cell activation and only appeared several days after T cell stimulation and became especially prominent after the addition of exogenous IL 2. All T lymphoblastoid cell lines tested expressed at least the lower bands (130,000 Mr), and the T cell line HSB also expressed one of the activation-related larger proteins (165,000 Mr). B lymphoblastoid cell lines expressed only a very weak lower band (130,000 Mr), and the cell line U-937 (in the monocyte-macrophage lineage) expressed only a single band (145,000 Mr) not aligned with any of the bands found on lymphoid cells. The estimated number of A-1A5 binding sites per cell was much higher on U-937 (11 X 10(5)), and generally higher on other cell lines of myeloid lineage (1 to 4 X 10(5)) than on lymphoid cell lines (0.2 to 1.3 X 10(5)).

During studies of human hematopoietic cells, two novel anti-mycoplasma monoclonal antibodies reacting in human cell surface binding assays were generated. These monoclonal antibodies define antigens (called MP71 and MP33) that showed variable positive and negative expression in different cultures of the same human cell lines. Furthermore, negative cell cultures could be converted to positive cultures upon incubation with filtered media from a previously positive cell culture. A DNA staining assay was used to demonstrate the presence of mycoplasma in the cultures positive for MP71. The expression of MP71 on a panel of cell lines always correlated with the expression of MP33, suggesting that the latter antigen is also mycoplasma related. The monoclonal antibodies against both antigens (MP71 and MP33) were each able to block the growth of mycoplasma newly transferred to hematopoietic cell cultures. Biosynthesis of these mycoplasma antigens differed from biosynthesis of host cell surface antigens in that the former showed a relative insensitivity to gamma irradiation. Internal labeling with 35S-methionine or host cell surface labeling with 125I and analysis by immunoprecipitation showed that the antigens MP71 and MP33 had sizes of Mr 71,000 and 33,000, respectively. Also the antigen MP71 was almost completely, although transiently, cleared from the host cell surface by trypsin. Growth inhibition analysis of different mycoplasma species indicated that the monoclonal antibody recognizing MP71 was uniquely reactive with Mycoplasma hyorhinis.

A glycoprotein complex of 210,000 and 130,000 m.w., found on mitogen or alloantigen-stimulated human T cells and not on other hematopoietic cells, has been defined by a monoclonal antibody (Mab). The components of this complex are a subset of a larger family of proteins (210,000, 165,000 and 130,000 m.w.) defined by a second Mab. In a panel of hematopoietic cell lines and cell types, only activated T cells (including the cell line HUT-102) express the 210,000/130,000 complex and these cells also express the IL 2 receptor, a characteristic marker for activated T cells. The 210,000/130,000 m.w. complex (reactive with the Mab TS2/7) is present on all long-term activated T cells, including both the OKT4 and OKT8 subsets. The 210,000 m.w. subunit is expressed only on activated T cells. Other lymphoid cells express either the 130,000 m.w. subunit alone (unactivated lymphocytes, thymocytes, HUT-78) or the 130,000 subunit together with a 165,000 subunit (MOLT-4, HSB, and other leukemic T cell lines). The 210,000/130,000 m.w., 165,000/130,000 m.w. and 130,000 m.w. complexes are antigenically related in that all share reactivity with the Mab A- 1A5 . Among non-lymphoid hematopoietic cells and cell lines, none express the 210,000 m.w. chain; adherent cells (monocytes) and myeloid cell lines each express single proteins of 130,000 to 155,000 m.w. Granulocytes and red blood cells are negative and platelets express multiple bands (165,000 and 140,000 m.w.). Immunoperoxidase staining of tissue sections showed that a broad range of tissues and cell types had material cross-reactive with the lymphoid 130,000 m.w. protein. However, only a discrete subset of those tissues and cells including blood vessel walls, connective tissue, smooth muscle, kidney mesangial cells, and some non-cellular matrix tissue, had material cross-reactive with the 210,000 m.w. protein on activated T lymphocytes.

A monoclonal antibody, 12F1, has been produced that specifically immunoprecipitates the human cell surface structure VLA-2 from platelets and long-term activated T cells, as well as from fibroblast and neuroblastoma cell lines. Cross-linking studies indicate that the VLA-2 structure exists on the cell surface as a 165,000 Mr heavy chain (alpha 2) in noncovalent 1:1 association with a 130,000 Mr light chain (beta). The monoclonal antibody A-1A5, which reacts with the beta subunit common to all VLA structures, was able to completely preclear VLA-2, indicating that all of the alpha 2 subunit was associated with VLA beta-chain. The specificity of 12F1 for VLA-2 allowed independent immunoprecipitation and flow cytometry analysis of this alpha 2 beta structure separate from any other VLA structures that may have been present such as VLA-1 or free beta-subunit. Subunit dissociation studies were used to demonstrate that 12F1 recognizes an epitope on the alpha 2 chain on VLA-2, which is consistent with the 12F1 specificity for VLA-2 alone among the VLA proteins. Analysis of activated T cells indicated that VLA-2, like VLA-1, is another "very late" appearing T cell activation antigen that arises concurrently with VLA-1 starting at day 7 and increasing through 2 wk. VLA-2 was found on many of the same cells as VLA-1 (inactivated T cells, T cell leukemia cells, fibroblasts, SK-N-SH neuroblastoma cells), but VLA-1 and VLA-2 can be expressed independently, because VLA-2 was also present on VLA-1-negative cells such as HSB and platelets, and VLA-1 was present on VLA-2-negative C8215 cells.

The mitogens phytohemagglutinin (PHA) and concanavalin A inhibited the appearance of the very late activation antigen (VLA)-1, but did not inhibit VLA-2 expression on cultured activated T cells. In contrast to diminished VLA-1 expression, mitogen treatment caused increased cell surface expression of other activation antigens such as T10, HLA-DR, interleukin 2 (IL 2) receptor, and 4F2, and greater cell proliferation. Conversely, when T cells were not repetitively restimulated with mitogen, these less proliferative "postactivated" T cells had elevated VLA-1 expression. The diminished expression of VLA-1 caused by PHA was reversible since subsequent removal of mitogen was associated with increased VLA-1, paralleled by a decrease in interleukin 2 receptor levels. In addition to preventing or delaying the initial appearance of VLA-1, PHA stimulation also was somewhat effective in causing the disappearance of VLA-1 already present, especially on recently established cultures. However, cultures that had either never seen PHA, not seen PHA for several weeks, or been stimulated regularly with PHA, but were several months old, did not lose VLA-1 in response to PHA stimulation, suggesting that a state of insensitivity to PHA effects could be attained. Unlike PHA-stimulated T cells, T cells repetitively restimulated with alloantigen or the monoclonal antibody T3 did not show a marked absence of VLA-1 but rather showed an increased level of VLA-2 relative to VLA-1. Taken together, results of stimulation by either mitogen, alloantigen, or anti-T3 monoclonal antibody support the conclusion that T cell stimulation in general can cause a decreased VLA-1:VLA-2 ratio, whether by decreased VLA-1 or increased VLA-2. These shifts in VLA-1:VLA-2 ratios are probably not simply the result of shifts in the relative proportions of different subpopulations, because similar growth-related changes in this ratio were observed on the T cell line ANITA, which is a homogeneous population of cells. Because both VLA-1 and VLA-2 are differentially regulated on cultured, long term activated T cells depending on stage of activation and growth conditions, and are members of a family of at least five heterodimers that includes cell matrix adhesion molecules, we suggest that these studies will provide clues to novel aspects of T cell growth regulation, perhaps relating to T cell-matrix adhesion.

The monoclonal antibody 4F2 recognizes a disulfide-linked ricin-binding glycoprotein complex (Mr congruent to 125,000) composed of a sialylated heavy subunit (Mr congruent to 85,000 on T cell lines) and an unsialylated light subunit (Mr congruent to 41,000). The antigen (T85,41) recognized by 4F2 on T cell lines is structurally distinct from the antigen (B93, 41) on B cell lines. The heavy subunits, but not the light subunits, from all T cell lines examined were uniformly smaller in size than the heavy subunits from several B cell lines. This reflects differences in carbohydrate rather than protein represent in B93,41 compared with T85,41, because both heavy subunits have a common unglycosylated form (p65) and a common partially glycosylated precursor form (p68). Among non-T, non-B hematopoietic cell lines, the monocytoid line U-937 expressed an antigen that resembles B93,41, whereas the erythroleukemic line K-562 expressed an antigen more similar to T85,41. 4F2 recognizes a protein determinant on the heavy subunit (with or without N-linked glycosylation) and also the unglycosylated heavy subunit retains the ability to associate with light subunit. The light subunit itself contains no detectable N-linked carbohydrate. Unlike the transferrin receptor, synthesis of the antigen recognized by 4F2 on the promyelocytic cell line HL-60 did not diminish upon dimethylsulfoxide-induced differentiation, and thus is not tightly correlated with cell proliferation.

A monoclonal antibody, 12F1, has been produced that specifically immunoprecipitates the human cell surface structure VLA-2 from platelets and long-term activated T cells, as well as from fibroblast and neuroblastoma cell lines. Cross-linking studies indicate that the VLA-2 structure exists on the cell surface as a 165,000 Mr heavy chain (alpha 2) in noncovalent 1:1 association with a 130,000 Mr light chain (beta). The monoclonal antibody A-1A5, which reacts with the beta subunit common to all VLA structures, was able to completely preclear VLA-2, indicating that all of the alpha 2 subunit was associated with VLA beta-chain. The specificity of 12F1 for VLA-2 allowed independent immunoprecipitation and flow cytometry analysis of this alpha 2 beta structure separate from any other VLA structures that may have been present such as VLA-1 or free beta-subunit. Subunit dissociation studies were used to demonstrate that 12F1 recognizes an epitope on the alpha 2 chain on VLA-2, which is consistent with the 12F1 specificity for VLA-2 alone among the VLA proteins. Analysis of activated T cells indicated that VLA-2, like VLA-1, is another "very late" appearing T cell activation antigen that arises concurrently with VLA-1 starting at day 7 and increasing through 2 wk. VLA-2 was found on many of the same cells as VLA-1 (inactivated T cells, T cell leukemia cells, fibroblasts, SK-N-SH neuroblastoma cells), but VLA-1 and VLA-2 can be expressed independently, because VLA-2 was also present on VLA-1-negative cells such as HSB and platelets, and VLA-1 was present on VLA-2-negative C8215 cells.

On platelets and other cell types, VLA-6 is a typical integrin heterodimer, with alpha 6-beta 1 subunit association. However, on colon carcinoma cell lines and other epithelial cells the alpha 6 subunit associates with a novel protein (called beta 4) rather than the VLA beta 1 subunit. The beta 4 protein differs from beta 1 because (i) it is not recognized by anti-beta 1 antibodies, (ii) it yields different V8 protease cleavage products, (iii) it has a more limited cell distribution, (iv) it has multiple forms, each larger in size than beta 1, and (v) it is susceptible to protease digestion which does not effect beta 1. Although different in many respects, the beta 4 subunit does have partial N-terminal sequence similarity to the already defined integrin beta 1, beta 2, and beta 3 subunits. The presence of alpha 6-beta 4 complexes was demonstrated by coprecipitation of beta 4 with an anti-alpha 6 antibody and by covalent cross-linking experiments. Although alpha 6-beta 4 complexes were present on certain cells, other VLA alpha subunits on those same cells remained associated with the VLA beta 1 subunit to form typical VLA heterodimers (e.g. VLA-1, VLA-2, VLA-3). By the criteria of N-terminal amino acid sequencing, antibody recognition, V8 peptide maps, and reduced/nonreduced gel migration, the alpha 6 subunit which associates with beta 4 appears identical to the alpha 6 associated with the VLA beta 1 subunit on platelets and other cell types. The beta 4 subunit may be of major importance because (i) it is highly abundant on the surface of colon carcinoma cell lines, and (ii) it is highly immunogenic relative to other surface proteins.

A monoclonal antibody (B-5G10) was produced which specifically recognizes the Mr 150,000/130,000 VLA-4 complex on the surface of human cells. Cross-linking studies indicated that the Mr 150,000 alpha 4 subunit of VLA-4 is in noncovalent 1:1 association with the Mr 130,000 VLA beta subunit. In the absence of cross-linking, the VLA-4 alpha 4 beta subunit complex was easily dissociated, especially in Nonidet P-40 detergent, or at elevated pH (above 8.0). Studies of dissociated subunits showed that B-5G10 recognizes an epitope on the Mr 150,000 alpha 4 subunit of VLA-4, whereas the beta subunit is immunologically identical to the Mr 130,000 beta subunit common to all VLA heterodimers. VLA-4 is widely distributed on hematopoietic cells, including thymocytes, peripheral blood lymphocytes, monocytes, activated T cells, T and B lymphoblastoid cell lines, and myeloid cell lines. However, VLA-4 is only weakly expressed on most adherent cell lines tested. Immunoprecipitates of VLA-4 often contain additional proteins of Mr 80,000 and Mr 70,000. These are probably derived from the Mr 150,000 alpha 4 subunit because: 1) they are both recognized by anti-alpha 4 sera, but not anti-beta sera; 2) the sum of their sizes is equal to the size of alpha 4; 3) they are selectively coexpressed with alpha 4 and not other VLA alpha subunits; 4) the Mr 80,000 protein has an identical NH2-terminal sequence to alpha 4; 5) like alpha 4, the Mr 70,000 and 80,000 peptides can variably associate with the VLA beta subunit; and 6) trypsin appears to cleave the Mr 150,000 alpha 4 subunit into products of Mr 70,000 and 80,000.

During studies of human hematopoietic cells, two novel anti-mycoplasma monoclonal antibodies reacting in human cell surface binding assays were generated. These monoclonal antibodies define antigens (called MP71 and MP33) that showed variable positive and negative expression in different cultures of the same human cell lines. Furthermore, negative cell cultures could be converted to positive cultures upon incubation with filtered media from a previously positive cell culture. A DNA staining assay was used to demonstrate the presence of mycoplasma in the cultures positive for MP71. The expression of MP71 on a panel of cell lines always correlated with the expression of MP33, suggesting that the latter antigen is also mycoplasma related. The monoclonal antibodies against both antigens (MP71 and MP33) were each able to block the growth of mycoplasma newly transferred to hematopoietic cell cultures. Biosynthesis of these mycoplasma antigens differed from biosynthesis of host cell surface antigens in that the former showed a relative insensitivity to gamma irradiation. Internal labeling with 35S-methionine or host cell surface labeling with 125I and analysis by immunoprecipitation showed that the antigens MP71 and MP33 had sizes of Mr 71,000 and 33,000, respectively. Also the antigen MP71 was almost completely, although transiently, cleared from the host cell surface by trypsin. Growth inhibition analysis of different mycoplasma species indicated that the monoclonal antibody recognizing MP71 was uniquely reactive with Mycoplasma hyorhinis.

The tumor-associated antigen complex, TSP-180, was previously defined in carcinoma cell lines and found to be expressed in higher amounts in tumor than in normal tissue. Here, the mouse TSP-180 complex is shown to consist of three related proteins (bands 1, 2, and 3) associated with a distinct protein (band 5) that is probably derived from a precursor protein (band 4). All of these proteins are cell surface glycoproteins, and the largest protein (band 1) can be readily labeled with 32PO4. The mouse TSP-180 complex described here strongly resembles the recently described human integrin alpha 6-beta 4 complex. This homology was confirmed using two distinct rat anti-alpha 6 monoclonal antibodies, each of which recognized both human alpha 6-beta 4 and mouse TSP-180 complexes. Furthermore, the TSP-180 band 5 protein (mouse alpha 6) had an N-terminal sequence identical to that of human alpha 6. Finally, two different monoclonal antibodies are described, 346-11A and 439-9B, which directly recognize the multiple forms of mouse and human beta 4 proteins, respectively.

The thymus is thought to be the primary central lymphoid organ in which T cells mature. Although thymic cortical and medullary compartments are distinct histologically, few antigens have been described that are absolutely acquired during the presumed intrathymic maturation pathway from cortical to medullary thymocytes. In this paper, we describe the acquisition during human intrathymic T cell maturation of a novel protein (p80) defined by a monoclonal antibody (A1G3). Although the p80-A1G3 antigen is distributed throughout the body and is not T cell specific, our study demonstrates that expression of p80-A1G3 antigen in normal human thymus is associated with thymocyte functional maturity and location in the thymus medulla. Moreover, in contrast to other markers of mature human T cells, the p80-A1G3 cell surface protein is not expressed on T6+ cortical thymocytes, and, therefore, is absolutely acquired by medullary thymocytes during T cell maturation. Thus, the p80-A1G3 antigen and the A1G3 antibody provide a heretofore unavailable system for the study of molecular events that transpire during the maturation of thymocytes.

The mitogens phytohemagglutinin (PHA) and concanavalin A inhibited the appearance of the very late activation antigen (VLA)-1, but did not inhibit VLA-2 expression on cultured activated T cells. In contrast to diminished VLA-1 expression, mitogen treatment caused increased cell surface expression of other activation antigens such as T10, HLA-DR, interleukin 2 (IL 2) receptor, and 4F2, and greater cell proliferation. Conversely, when T cells were not repetitively restimulated with mitogen, these less proliferative "postactivated" T cells had elevated VLA-1 expression. The diminished expression of VLA-1 caused by PHA was reversible since subsequent removal of mitogen was associated with increased VLA-1, paralleled by a decrease in interleukin 2 receptor levels. In addition to preventing or delaying the initial appearance of VLA-1, PHA stimulation also was somewhat effective in causing the disappearance of VLA-1 already present, especially on recently established cultures. However, cultures that had either never seen PHA, not seen PHA for several weeks, or been stimulated regularly with PHA, but were several months old, did not lose VLA-1 in response to PHA stimulation, suggesting that a state of insensitivity to PHA effects could be attained. Unlike PHA-stimulated T cells, T cells repetitively restimulated with alloantigen or the monoclonal antibody T3 did not show a marked absence of VLA-1 but rather showed an increased level of VLA-2 relative to VLA-1. Taken together, results of stimulation by either mitogen, alloantigen, or anti-T3 monoclonal antibody support the conclusion that T cell stimulation in general can cause a decreased VLA-1:VLA-2 ratio, whether by decreased VLA-1 or increased VLA-2. These shifts in VLA-1:VLA-2 ratios are probably not simply the result of shifts in the relative proportions of different subpopulations, because similar growth-related changes in this ratio were observed on the T cell line ANITA, which is a homogeneous population of cells. Because both VLA-1 and VLA-2 are differentially regulated on cultured, long term activated T cells depending on stage of activation and growth conditions, and are members of a family of at least five heterodimers that includes cell matrix adhesion molecules, we suggest that these studies will provide clues to novel aspects of T cell growth regulation, perhaps relating to T cell-matrix adhesion.

By using a single monoclonal antibody, a novel glycoprotein complex composed of at least three distinct bands was defined on the surface of mitogen- or alloantigen-stimulated T cells. These bands (210,000, 165,000, and 130,000 Mr) were not disulfide linked and could be radio-labeled with 125I or 35S-methionine and readily detected by immunoprecipitation with the monoclonal antibody A-1A5. Only approximately 20% of normal T cells (E rosette positive) and approximately 47% non-T cells (E rosette negative) were reactive with A-1A5. However, upon activation of T cells, the amount of A-1A5 binding per cell and the percentage of positive cells significantly increased. This increase was most pronounced in the activated cell subpopulations currently undergoing cell division (S, G2, and M phases), which became 79% A-1A5 positive (after PHA stimulation) and 99% A-1A5 positive (in long-term culture with alloantigen and IL2). Resting lymphocytes contained only the 130,000 Mr band reactive with A-1A5. The two other bands (210,000 and 165,000 Mr) were markers for T cell activation and only appeared several days after T cell stimulation and became especially prominent after the addition of exogenous IL 2. All T lymphoblastoid cell lines tested expressed at least the lower bands (130,000 Mr), and the T cell line HSB also expressed one of the activation-related larger proteins (165,000 Mr). B lymphoblastoid cell lines expressed only a very weak lower band (130,000 Mr), and the cell line U-937 (in the monocyte-macrophage lineage) expressed only a single band (145,000 Mr) not aligned with any of the bands found on lymphoid cells. The estimated number of A-1A5 binding sites per cell was much higher on U-937 (11 X 10(5)), and generally higher on other cell lines of myeloid lineage (1 to 4 X 10(5)) than on lymphoid cell lines (0.2 to 1.3 X 10(5)).

The family of human cell surface heterodimers which includes VLA-1 and VLA-2 is now shown to include three additional heterodimers, here called VLA-3, VLA-4, and VLA-5. Each of these separate VLA structures is composed of a distinct alpha subunit (Mr 110,000-200,000 nonreduced) noncovalently associated with a common beta subunit (Mr 110,000 nonreduced). Chemical cross-linking experiments provided evidence that each VLA complex exists predominantly as a 1:1 alpha beta heterodimer. The VLA proteins are widely distributed, with one or more of the heterodimers present on nearly all cell types tested. Evidence for five distinct VLA alpha subunits was obtained from differences observed in antibody recognition, cell distribution patterns, two-dimensional gel analyses, and V8 protease cleavage patterns. On the other hand, the beta subunit present in each heterodimer was immunochemically and electrophoretically indistinguishable, and yielded identical V8 cleavage fragments. Immunoblotting experiments revealed that besides the Mr 110,000 beta normally seen, another beta protein was present that is smaller in size (Mr 90,000 nonreduced), altered or deficient in glycosylation, and not available for cell surface radiolabeling.

The monoclonal antibody 4F2 recognizes a disulfide-linked ricin-binding glycoprotein complex (Mr congruent to 125,000) composed of a sialylated heavy subunit (Mr congruent to 85,000 on T cell lines) and an unsialylated light subunit (Mr congruent to 41,000). The antigen (T85,41) recognized by 4F2 on T cell lines is structurally distinct from the antigen (B93, 41) on B cell lines. The heavy subunits, but not the light subunits, from all T cell lines examined were uniformly smaller in size than the heavy subunits from several B cell lines. This reflects differences in carbohydrate rather than protein represent in B93,41 compared with T85,41, because both heavy subunits have a common unglycosylated form (p65) and a common partially glycosylated precursor form (p68). Among non-T, non-B hematopoietic cell lines, the monocytoid line U-937 expressed an antigen that resembles B93,41, whereas the erythroleukemic line K-562 expressed an antigen more similar to T85,41. 4F2 recognizes a protein determinant on the heavy subunit (with or without N-linked glycosylation) and also the unglycosylated heavy subunit retains the ability to associate with light subunit. The light subunit itself contains no detectable N-linked carbohydrate. Unlike the transferrin receptor, synthesis of the antigen recognized by 4F2 on the promyelocytic cell line HL-60 did not diminish upon dimethylsulfoxide-induced differentiation, and thus is not tightly correlated with cell proliferation.

The very late antigen complexes VLA-1 and VLA-2 which appear on long-term activated human T cells have been characterized with respect to 1) subunit arrangement, 2) location of monoclonal antibody (MAb) binding sites, 3) carbohydrate content, and 4) protein homology. Cross-linking experiments showed that the VLA-1 complex is a heterodimer composed of an Mr 210,000 subunit (alpha 1) in acid-labile association with an Mr 130,000 subunit (beta). The VLA-2 complex is a heterodimer with an Mr 165,000 subunit (alpha 2) in base-labile association with the Mr 130,000 beta subunit. The subunits of VLA-1 (alpha 1 beta) and VLA-2 (alpha 2 beta) each appear to be arranged with 1:1 stoichiometry. The MAb A-1A5 has been shown to bind to an epitope on the common beta subunit, consistent with its recognition of both the VLA-1 and VLA-2 heterodimers. On the other hand, MAb TS2/7 bound to an epitope of the alpha 1 subunit, thus explaining the specific recognition of the VLA-1 heterodimer by TS2/7. Digestion of the alpha 1, alpha 2, and beta subunits with neuraminidase and with endoglycosidase F revealed that each subunit contains substantial sialic acid and N-linked carbohydrate. By one-dimensional peptide mapping, the alpha 1, alpha 2, and beta subunits were shown to be highly nonhomologous with respect to each other, although each subunit from different T cell sources appeared highly homologous if not identical.