Isolation of a cloned cell line expressing variant H-2Kk using fluorescence-activated cell sorting

Detection and isolation of rare cells

Cells of biomedical interest are often present in very small numbers (i.e. they are rare), despite their functional significance. The analysis and isolation of previously inaccessible rare cells, such as peripheral hematopoietic stem cells, fetal cells in maternal blood, residual tumor cells or antigen-specific lymphocytes, has now become feasible through the development of new methods. Today, these techniques allow the detection, isolation and analysis of cells less frequent than one in a million. Not many problems in immunology would require higher resolution.

Application of flow cytometric measurement of surface IgG in kinetic analysis of monoclonal antibody synthesis and secretion by murine hybridoma cells

The kinetics of antibody synthesis and secretion from a murine hybridoma cell line were studied using measurements of total cell-associated IgG, surface IgG, and IgG secreted into the medium. Kinetic analysis of IgG secretion demonstrates approximately constant secretion rate per viable cell over the entire batch cultivation. A correlation was observed (r2 = 0.74) between mean surface immunofluorescence and the total cell-associated IgG determined by ELISA of detergent-extracted cell lysates. No correlation was found between specific secretion rate and mean surface IgG level estimated by immunofluorescence flow cytometry measurements. Material balances on cellular IgG demonstrated that about 7% of the antibody which was synthesized during exponential batch growth was not released to the growth medium. Distributions of single-cell surface antibody content showed two subpopulations, one with very low surface IgG. The fraction of the population with low surface IgG increased throughout a batch cultivation.

Lymphocyte antigen Leu-19 as a molecular marker of regeneration in human skeletal muscle

Antigen Leu-19 (Leu19-Ag), a 200- to 220-kDa surface glycoprotein, was originally identified on a subset of human peripheral lymphocytes exhibiting non-major histocompatibility complex-restricted cytotoxicity. Here we report that monoclonal antibody Leu-19 (mAb-Leu19) labels structures in human skeletal muscle: (i) satellite cells, which form the stem cell pool of muscle fiber regeneration, both in normal and diseased muscle; (ii) myotubes and myotube projections in regions of muscle fiber repair; (iii) periodically organized fibrillar structures in areas of regeneration; (iv) the surface of myoblasts and developing myotubes in culture. mAb-Leu19 precipitated a protein of approximately 200 kDa from cultured muscle cells. Our data show that Leu19-Ag is expressed on muscle-specific components of myosegments in repair and thus represents a molecular marker of muscle regeneration. On the basis of this molecular marker and using laser scan microscopy, it is possible to visualize at the light microscopic level hitherto undetectable details of muscle regeneration in routine cryostat sections.

Immunoselection of structural H-2Kb variants: use of cloned cytolytic T cells to select for loss of a CTL-defined allodeterminant

Functional studies concerning the unique interaction between class I H-2 allodeterminants and cytolytic T lymphocyte (CTL) antigen receptors have benefitted from the development of H-2Kb mutant mouse strains and somatic H-2 variants selected with monoclonal antibody. Here, we describe the development of a novel approach to immunoselection of somatic H-2Kb variants employing a Kb-specific CTL clone as the negative selective agent. The rationale for this method is that the use of an alloreactive CTL clone as the selective agent should enable us to detect the emergence of structural Kb variants based on their loss of the relevant CTL-defined allodeterminant. Thus, these structural variants are well suited to an in-depth analysis of the functional relationship between H-2 antigens and receptor recognition by CTL. Using this approach, we successfully isolated two types of structural Kb variants, as well as numerous Kb-loss variants. The functional studies described in this paper indicate that these structural variants exhibit alterations in expression of both CTL-defined and serologically defined H-2Kb allodeterminants. The structural characterization of such variants should enable us to identify the precise amino acid residues responsible for the creation of the relevant CTL-defined Kb allodeterminants.

A structural somatic variant of the Kk antigen is generated by point mutation

We have previously selected structural variants of the Kk antigen from a (C3 X D2)F1 T-cell lymphoma. Those mutants were identified by the loss of certain epitopes defined by monoclonal antibodies. The variant Kk molecule from HK13.S3 cells is no longer recognized by 40% of the trinitrophenyl-specific, Kk-restricted cytotoxic T lymphocytes. Here we report on the primary structure of the altered Kk molecules from the cell line HK13.S3. Comparison with the parental Kk reveals a single base pair exchange, GCG to GTG, that results in an alanine to valine exchange in position 40 of the protein. This observation emphasizes that minor structural alterations in class I molecules may have a strong effect on the H-2-restricted T-cell response.

Inexpensive upgrading of a FACS I and isolation of rare somatic variants by double-fluorescence sorting

We have modified a FACS I by addition of a tunable dye laser and an optical system for fluorescence detection that allows physically independent measurement of green and red immunofluorescence. These modifications are inexpensive and should be applicable to most single-laser systems. By using the modified machine and double-fluorescence activated selection with H-2Kk-specific monoclonal antibodies labelled with FITC or Texas Red, we have isolated from a murine T-cell lymphoma line variant subclones expressing structurally altered histocompatibility class I (H-2Kk) molecules.

Selection of cell lines resistant to anti-transferrin receptor antibody: evidence for a mutation in transferrin receptor

Some anti-murine transferrin receptor monoclonal antibodies block iron uptake in mouse cell lines and inhibit cell growth. We report here the selection and characterization of mutant murine lymphoma cell lines which escape this growth inhibition by anti-transferrin receptor antibody. Growth assays and immunoprecipitation of transferrin receptor in hybrids between independently derived mutants or between mutants and antibody-susceptible parental cell lines indicate that all of the selected lines have a similar genetic alteration that is codominantly expressed in hybrids. Anti-transferrin receptor antibodies and transferrin itself still bind to the mutant lines with saturating levels and Kd values very similar to those of the parental lines. However, reciprocal clearing experiments by immunoprecipitation and reciprocal blocking of binding to the cell surface with two anti-transferrin receptor antibodies indicate that the mutant lines have altered a fraction of their transferrin receptors such that the growth-inhibiting antibody no longer binds, whereas another portion of their transferrin receptors is similar to those of the parental lines and binds both antibodies. These results argue that the antibody-selected mutant cell lines are heterozygous in transferrin receptor expression, probably with a mutation in one of the transferrin receptor structural genes.

Deletion of the IgH enhancer does not reduce immunoglobulin heavy chain production of a hybridoma IgD class switch variant

Immunoglobulin (Ig) gene promotors are active only in cells of the B-lymphocyte lineage. Transfection experiments have shown that this is due in part to tissue specific 'activating' DNA sequences, so called enhancers. It is not entirely clear whether these sequences are necessary for initial activation or also for maintenance of transcription of a gene. We describe here the isolation and characterisation of a mouse hybridoma cell line that has deleted in vitro the 'activating' sequence from the active IgH locus, the only IgH locus it contains. Nevertheless, Ig heavy chain production of the variant cell is not impaired and remains comparable with that of other hybridoma cells. Therefore, a high rate of Ig heavy chain production in antibody-producing cells is either independent of any sequences enhancing transcription or else these can easily be replaced by other DNA sequences with a similar function that have been moved into the vicinity of the V region.

Selection of cell lines resistant to anti-transferrin receptor antibody: evidence for a mutation in transferrin receptor

Some anti-murine transferrin receptor monoclonal antibodies block iron uptake in mouse cell lines and inhibit cell growth. We report here the selection and characterization of mutant murine lymphoma cell lines which escape this growth inhibition by anti-transferrin receptor antibody. Growth assays and immunoprecipitation of transferrin receptor in hybrids between independently derived mutants or between mutants and antibody-susceptible parental cell lines indicate that all of the selected lines have a similar genetic alteration that is codominantly expressed in hybrids. Anti-transferrin receptor antibodies and transferrin itself still bind to the mutant lines with saturating levels and Kd values very similar to those of the parental lines. However, reciprocal clearing experiments by immunoprecipitation and reciprocal blocking of binding to the cell surface with two anti-transferrin receptor antibodies indicate that the mutant lines have altered a fraction of their transferrin receptors such that the growth-inhibiting antibody no longer binds, whereas another portion of their transferrin receptors is similar to those of the parental lines and binds both antibodies. These results argue that the antibody-selected mutant cell lines are heterozygous in transferrin receptor expression, probably with a mutation in one of the transferrin receptor structural genes.

Delineation of serologically distinct monomorphic determinants of human MHC class II antigens: evidence of heterogeneity in their topographical distribution

Three monoclonal antibodies (mAbs), FMC4, FMC14 and FMC15, which react with invariant sites of major histocompatibility complex (MHC) class II (Ia-like) molecules were studied in various serological assays. Sequential immunodepletion experiments show that all three epitopes are present on the same class II molecules. However, a minor subset may exist which does not express epitope 15. In competitive binding assays, using several different B-lymphoblastoid cell lines (B-LCLs), e.g. BRISTOL-8 8392, B-85, RAJI, 8866, CESS-B and LD-B, FMC4 did not block the binding of FMC14, or FMC15, and vice versa. In contrast, mutual inhibition was observed between FMC14 and FMC15. Furthermore, pairwise combinations of saturating amounts of FMC4 + FMC14 and FMC4 + FMC15 gave additive binding whilst FMC14 + FMC15 did not. These results demonstrate that epitopes 4 and 14/15 are spatially distinct; 14 and 15 on the other hand appear to be spatially related. However, contrary to this partial and reciprocal inhibition was consistently observed between FMC4 and FMC14 on two other LCLs, namely DAUDI and BALM-2. Furthermore, on certain cells, FMC14 and FMC15 show markedly disparate binding. Taken together, these observations indicate that the juxtaposition of certain epitopes on class II antigens can vary according to the cell type. This demonstrates a hitherto unreported heterogeneity of antigenic determinants and of their topographical distribution on the class II molecule.

Interaction between MHC-encoded products and cloned T cells. II. Analyses of physiological requirements indicate two different pathways of stimulation by class I alloantigens

The interaction between class I major histocompatibility complex (MHC) products and T cells was studied using H-2Kb-specific alloreactive T-cell lines and clones obtained by repeated in vitro stimulation with allogeneic cells. Induction of proliferation of these T cells appeared to involve two signals: the H-2Kb alloantigen and interleukins. Immunopurified liposome-inserted H-2Kb, which stimulates specific secondary in vitro cytotoxic T lymphocyte (CTL) responses, could not replace cell-associated H-2Kb in the stimulation of these T-cell lines, even in the presence of feeder cells and interleukins. When T-cell lines were initiated in vitro and repeatedly stimulated with H-2Kb liposomes and feeder cells, it was possible to obtain T cells that could proliferate in response to H-2Kb liposomes in the presence of feeder cells and interleukin-2-containing supernatants or on H-2Kb-expressing cells. Only stimulation with cells permitted maintenance of these T cells in culture for more than 12 weeks. Analyses of cell surface markers and of patterns of inhibition of proliferation by monoclonal antibodies (mAb) of T-cell lines induced in vitro with cell- or liposome-associated H-2Kb indicated that T-cell stimulation by class I antigen can occur in at least two ways. In the first, the H-2Kb-induced proliferation of Lyt-1- Lyt-2+ T4- T cells is inhibited by H-2Kb- and by Lyt-2-specific mAb, but not by Ia or T4-specific mAb. In the second, both Lyt-2+ and T4+ T cells are involved and the H-2Kb-induced proliferation is inhibited by H-2Kb- and Lyt-2-specific mAb and by Ia- and T4-specific mAb.

Immunological surveillance of tumors in the context of major histocompatibility complex restriction of T cell function

The immunological surveillance hypothesis was formulated prior to the realization of the fact that an individual's effector T cells generally only see neoantigen if it is appropriately presented in the context of self MHC glycoproteins. The biological consequence of this mechanism is that T lymphocytes are focused onto modified cell-surface rather than onto free antigen. The discovery of MHC-restricted T cell recognition, and the realization that T cell-mediated immunity is of prime importance in promoting recovery from infectious processes, has thus changed the whole emphasis of the surveillance argument. Though the immunological surveillance hypothesis generated considerable discussion and many good experiments, there is no point in continuing the debate in the intellectual context that seemed reasonable in 1970. It is now much more sensible to think of "natural surveillance" and "T cell surveillance," without excluding the probability that these two systems have elements in common. We can now see that T cell surveillance probably operates well in some situations, but is quite ineffective in many others. Part of the reason for this may be that the host response selects tumor clones that are modified so as to be no longer recognized by cytotoxic T cells. The possibility that this reflects changes in MHC phenotype has been investigated, and found to be the case, for some experimental tumors. In this regard, it is worth remembering that many "mutations" in MHC genes that completely change the spectrum of T cell recognition are serologically silent. The availability of molecular probes for investigating the status of MHC genes in tumor cells, together with the capacity to develop cloned T cell lines, monoclonal antibodies to putative tumor antigens, and cell lines transfected with genes coding for these molecules, indicates how T cell surveillance may profitably be explored further in both experimental and human situations.

Somatic H-2Kk variants reveal nonidentity of serological and cytotoxic T cell-defined Kk determinants

The relationship of serologically defined determinants to determinants recognized by cytotoxic T cells on molecules encoded by the Kk gene of the murine major histocompatibility complex (H-2) has been analyzed. For this purpose we used three somatic variants of a Kk-expressing lymphoma line lacking individual determinants of the Kk molecule, as defined by monoclonal antibodies (mAb), as target cells for Kk-specific alloreactive and Kk-restricted cytotoxic T lymphocytes (CTL) cloned by limiting dilution. Neither alloreactive nor fluorescein isothiocyanate, influenza- or Newcastle disease virus-specific Kk-restricted CTL clones were found to distinguish between variants and wild type cells, indicating that the serologically defined determinants lost by the variants were not essential for antigen recognition of CTL with these specificities. On the other hand, two of the variants lacking either one of a pair of serological determinants were discriminated from Kk wild type cells by about 40% of Kk-restricted, trinitrophenol (TNP)-specific CTL clones. The third variant, lacking both of the determinants, however, was lysed by all CTL clones to the same extent as wild type cells. From these results we conclude that the determinants restricting the TNP-specific CTL were also not identical with those defined by mAb. In experiments performed to optimize the conditions for the limiting dilution analysis we found that the specificity of the CTL stimulation was strongly dependent on the concentration of T cell growth factor (interleukin 2) in the cultures during CTL stimulation. High concentrations of IL2 resulted in a drastic increase in the frequency of CTL clones. Part of these clones, however, were found not to be specific for antigens present on the stimulator cells.

Distribution and mobility of murine histocompatibility H-2Kk antigen in the cytoplasmic membrane

The topographical distributions and mobilities of the murine histocompatibility antigen H-2Kk and of concanavalin A (Con A) binding sites have been studied on a murine lymphoma cell line. The spatial distribution of H-2Kk antigens, the average distance between H-2Kk antigens and Con A binding sites, and the separation of different determinants on the H-2Kk antigen itself were determined by using fluorescence resonance energy-transfer measurements with a dual-laser flow sorter. From the lack of energy transfer between bound monoclonal anti-H-2Kk antibodies conjugated with fluorescein (donor) and rhodamine (acceptor), we conclude that the H-2Kk antigen exists without appreciable clustering on the cell surface. Substantial energy transfer between appropriately labeled Con A and antibodies bound to the H-2Kk antigen shows that the two populations are interspersed. Donor/acceptor pairs of monoclonal antibodies binding to different determinants on the same H-2Kk antigen exhibited a degree of energy transfer indicative of a mean separation of 8.6 nm between the sites. Time-resolved phosphorescence anisotropy measurements with anti-H-2Kk antibodies labeled with eosin or erythrosin yielded rotational mobility information for the antigen-antibody complexes on the cell membrane. The rotational correlation time of 10-20 mus and the finite residual anisotropy are compatible with an uniaxial mode of rotation of monomeric antigen around its transmembrane portion and, thus, provide additional evidence for an unclustered distribution. Capping by rabbit anti-mouse IgG immobilized the antigen-antibody complex. Fluorescence recovery after photobleaching was used to calculate an apparent lateral diffusion coefficient of 5 +/- 3 X 10(-10) cm2 . s-1 for the H-2Kk antigen labeled with fluoresceinated IgG or its corresponding Fab fragment.

Loss of serological determinants does not affect recognition of H-2Kk target cells by an influenza-specific cytotoxic T cell clone

The recognition of "self determinants" by an H-2Kk-restricted cytotoxic T (Tc) cell clone (K5) has been investigated as follows: (a) differential inhibition of cytotoxicity by several monoclonal antibodies directed to determinants on the H-2Kk molecule of the target cells and (b) recognition and lysis of target cell variants of cell line LDHB, which have lost the majority or all of the serological determinants defined by the inhibiting antibodies while still expressing an H-2Kk molecule. Such variant cells infected with influenza virus were effectively recognized by Tc cell clone K5, whereas a target cell line, which lacks the Kk molecule, was not lysed. The results suggest; (a) that virus-specific Tc cell see "self" in a manner distinct from the recognition of serological determinants by B cell and (b) that antibody inhibition indicates conformational closeness, but not identity, of the class I determinant seen by T cells.

Selection of membrane IgM- variants from a mIgM+ murine B lymphoma cell: problems and solutions

We have used antibody-mediated complement killing to isolate membrane IgM-negative (mIgM-) variants from the mIgM+ murine B cell lymphoma, WEHI 279.1. This procedure has been used previously to select variants which lack expression of other cell-surface antigens on lymphoid cells. In those experiments, multiple rounds of selection have often been required for selection of the negative variants. We found that many cycles of selection produced very few variants and that those isolated had reduced, but still measurable, levels of mIgM. We were able to select large numbers of stable mIgM- variants by subjecting the populations with reduced levels of mIgM to two rounds of immunoselection within one cell cycle. These variants are stable and exhibit a variety of defects which are all expressed as a failure to display IgM on their external surface. Analysis of these variant clones at the biochemical level will begin to define the requirements for proper display of mIgM on the cell membrane of B lymphoma cells.

Stable transformation of mouse L cells for human membrane T-cell differentiation antigens, HLA and beta 2-microglobulin: selection by fluorescence-activated cell sorting

We isolated stable transformants of mouse L cells expressing human cell surface differentiation antigens by using immunofluorescence with monoclonal antibodies and selection with a fluorescence-activated cell sorter (FACS). Mouse L cells (TK-) were cotransformed with human cellular DNA and the herpes simplex virus thymidine kinase (TK) gene. TK+ transformants were first selected. The TK+ populations were stained with various fluorescent antibodies to membrane antigens, and positive cells were sorted and cloned by using a FACS. Transformants for HLA class I antigens, for beta 2-microglobulin, and for the T-cell differentiation antigens Leu-1 and Leu-2 were isolated. The frequency of antigen transformants among the TK+ transformants was about 0.5 X 10(-3). The sizes of the HLA, Leu-1, and Leu-2 molecules expressed by the transformants were the same as those of the proteins present on DNA-donor cells.

HLA antigen structural gene mutants selected with an allospecific monoclonal antibody

The HLA-A2 antigen-specific monoclonal antibody BB7.2 and complement were used to immunoselect mutants from an ethyl methanesulfonate-mutagenized human B lymphoid cell line, T5-1. Surviving colonies were screened by radioimmune binding with BB7.2 and with a monospecific HLA-A2 alloantiserum, Stewart, and HLA antigens of selected clones were immunoprecipitated and studied by isoelectric focusing. Several classes of mutants could be distinguished: mutants that expressed no HLA-A2 heavy chain; mutants that expressed an HLA-A2 heavy chain that was unable to associate with beta 2-microglobulin (beta 2m) and was not expressed at the cell surface; mutants with reduced HLA-A2 heavy chain-beta 2m association and cell surface expression of HLA-A2 dimer with or without heavy chain charge alterations; mutants with normal HLA-A2 heavy chain-beta 2m association and normal quantitative cell surface HLA-A2 expression but with HLA-A2 heavy chain charge alterations; and mutants with as yet incompletely defined lesions. Mutants with altered cell surface HLA antigens were not found in previous selections with alloantisera and should be useful for epitope mapping and structure-function studies of HLA molecules.

Rapid isolation of cloned isotype switch variants using fluorescence activated cell sorting

We have used highly specific, directly fluorescein-conjugated heterologous (conventional) and monoclonal antibodies directed against mouse immunoglobulin isotypes in conjunction with the fluorescence activated cell sorter (FACS) to enrich and clone hybridoma cells producing new immunoglobulin heavy chain constant regions. Each variant retains the parental heavy chain variable region and the parental immunoglobulin light chain; thereby each variant binds the same dansyl (DNS) hapten. These isotype switch variants occur at frequencies of approximately 10-5 to 10-6. We were able to isolate the variants by first sorting for an approximate 1000-fold enrichment of the desired immunoglobulin-producing cells, growing these cells for five to nine days, followed by a second 1000-fold enrichment and direct cell cloning into 96 well culture trays. Clones were screened only 3-5 weeks after the original selection for secretion of dansyl-binding immunoglobulin of the selected isotype. Judicious combination of existing methods permits improved analytical techniques using the cell sorter. These include: first, "red" fluorescence staining of dead cells with ethidium bromide or propidium iodide and using the red fluorescence measurement to exclude dead cells from the green fluorescence selection; and second, the use logarithmic amplification of fluorescence signals, allowing for more succinct selection of fluorescence parameters for sorting.

Gene dosage and gene expression in the HLA region: evidence from deletion variants

Among variants selected in a human cell line for nonexpression of a single gene product in the HLA complex, most are single-gene variants, but several have been isolated that are cis-acting and abolish expression of a series of closely linked genes. The two most plausible mechanisms by which such variants could arise are mitotic crossing-over and chromosome deletion. In two HLA variants the presence of a visible chromosome deletion, a 50% reduction in activity of glyoxalase I (a closely linked marker), or both provided evidence for deletional origins. In a third variant these changes were not demonstrable. All three variants showed reductions in amount of cell surface HLA antigens: 40% for the Ia antigens (HLA-DR) and 20-25% for HLA-ABC antigens. The reductions in cell surface antigen in deletion variants have an important implication: in the case of the HLA-A, -B, and -C heterodimer, which consists of a subunit coded for within the major histocompatibility complex and another subunit (beta(2)-microglobulin) coded for on a different chromosome, it is the gene of the major histocompatibility complex that is limiting. The nonmutant haplotype includes A2; binding of an A2 monoclonal antibody in two of the mutants was found to be approximately equal to that in the wild-type cells. Thus, loss of one copy of HLA-ABC genes does not lead to gene dosage compensation-i.e., increased activity by the remaining ABC alleles. The results with the two types of antibodies support a deletional mechanism and are inconsistent with mitotic crossing-over. Of interest with respect to the potential use of deletion variants for purposes of mapping is the fact that each of these variants has distinctive breakpoints. The absence of mitotic crossing-over in 1.2 x 10(7) cells selected suggests that the event is rare in this autosomal region, if it occurs at all.

Antigenic structure of influenza virus haemagglutinin defined by hybridoma antibodies

The recurrence of influenza virus infection in man is attributed primarily to changes occurring in the antigenic structure of the viral surface glycoproteins, especially of the haemagglutinin (HA) molecule. Comparative antigenic analysis of epidemic influenza virus strains has allowed the description of 'strain-specific' and 'cross-reactive' antigenic determinants. However, the interpretation of these findings remained ambiguous, because the specificity of the applied antisera was insufficiently defined and because the antigenic differences among the HA molecules of various epidemic virus strains resulted presumably from a large number of amino acid substitutions. Thus, in characterizing the antigenic structure of the HA molecule, our approach has been (1) to generate a panel of monoclonal anti-HA hybridoma antibodies, (2) to use some of these antibodies to select mutants of the influenza A/PR/8/34 (PR8) virus expressing antigenically altered HA molecules, and (3) to construct an operational antigenic map of the HA molecule by comparative antigenic analysis of the mutant viruses with the monoclonal antibodies. As we report here, analysis of the 34 mutant viruses selected has enabled us to define four antigenic sites on the HA molecule. Our observation that these sites have undergone antigenic drift to a different extent in nature implies that the mechanisms responsible for antigenic drift act selectively on distinct structures of the HA molecule.

Anti-H-2 hybridoma antibodies as immunoselective agents. I. Isolation of forward and reverse mutations for reactivity with a monoclonal antibody

Using immunoselection with an H-2Kk-specific monoclonal antibody following mutagenesis on an (H-2k/H-2d) F1 cell line we have obtained variants that do not react with the selecting monoclonal antibody but continue to react with other monoclonal antibodies directed against the same gene product. The mutants fall into two classes based on their serological profile. This phenotype is suggestive of a structural mutation in the selected gene. If the genetic change involved is a point mutation (as opposed to a deletion), one should be able to obtain revertants. Using the fluorescence-activated cell sorter, we have been able to obtain from one of the monoclonal-antibody-nonreactive mutants cells that do bind the selecting antibody. In order to prove that the presumptive revertant is not a contaminant wild-type cell that inadvertently got mixed into the resistance mutant, we first introduced an outside marker, resistance to the purine analogue 2-amino-6-mercaptopurine (6-thioguanine), into the monoclonal-antibody-resistant mutant. The revertants obtained using the cell sorter continue to express the nonselective phenotype of resistance to 6-thioguanine, showing that they are not wild-type cells. In addition, their serological characteristics are different from those of either the wild-type cells or the hybridoma-resistant mutants from which they were derived. Based on the serological analyses, it would seem that we have isolated at least three variant forms of the H-2Kk gene product.