The discovery of T-cell growth factor (TCGF) has made it possible to now routinely grow in tissue culture normal and neoplastic human T cells for long periods and in large amounts. TCGF has been recently purified. It is a small protein released by a subset of mature T cells following lectin-antigen activation, which in turn acts upon other T-cell subsets that have developed specific receptors for TCGF after lectin-antigen stimulation. Thus, release of TCGF and development of receptors for it appear to be obligatory for the clonal expansion of all activated T cells. Unlike normal T cells, neoplastic T cells respond directly to TCGF, requiring no prior in vitro lectin-antigen activation. This has led to the development of several new cell lines from patients with T-cell leukemias and lymphomas. In some cases, these cells become independent of exogenous TCGF by producing their own growth factor, implying a role for TCGF in the continuous proliferation of these cells. These developments necessitate a reevaluation of some concepts of immunoregulation of T-cell activities in terms of production and response to TCGF. In addition, this information has clinical implications. Recent results have shown that a major defect of the athymic nude mouse is the inability to produce TCGF and that some immunosuppressive agents, such as glucocorticosteroids and cyclosporin-A, exert their effects on T cells by disrupting the TCGF-T-cell interaction. Some human immune deficiencies might be due to a failure to respond to or to produce TCGF, which in some cases might be corrected by exogenous TCGF.

Anti-Tac monoclonal antibody has been identified as a putative antibody against the receptor for T-cell growth factor (TCGF). We now show that: (i) TCGF blocks 85% of 3H-labeled anti-Tac binding to phytohemagglutinin-activated lymphoblasts and (ii) both anti-Tac and anti-TCGF immunoprecipitate a protein band that appears to represent TCGF crosslinked to its receptor on HUT-102B2 cells. In HUT-102B2 cells, the TCGF receptor is a Mr 50,000 glycoprotein with internal disulfide bond(s) and a pI of 5.5-6.0, and it represents approximately equal to 0.05% of total cellular de novo protein synthesis. It contains a peptide of Mr 33,000 that is processed to a mature form that includes N-linked and O-linked sugars and sialic acid.

Rat spleen cells activated in vitro by concanavalin A produce lymphokine molecules that possess biologic activity in a number of murine lymphocyte response assays. A single class of lymphokine most adequately described as T cell growth factor (TCGF, Interleukin-2) with a m.w. of 15,000 as estimated from gel filtration studies and with an isoelectric range of 5.4 to 5.6 stimulates i) the growth of established T cell lines in culture, ii) the proliferation of thymocytes in the presence of Con A under culture conditions where Con A alone is non-mitogenic, iii) the induction of antibody responses to heterologous erythrocyte antigens in athymic (nude) mouse spleen cell cultures, and iv) the generation of cytotoxic T lymphocytes (CTL) in thymocyte cultures and in nude mouse spleen cell cultures. We suggest that in each of the assay systems tested, this class of rat lymphokine acts directly on activated T cells. Nonactivated T cells must be stimulated by either mitogen or antigen before becoming responsive to lymphokine, but do not require antigen or mitogen for continued lymphokine-dependent proliferation. Similarly, human peripheral blood mononuclear cells activated by phytohemagglutinin (PHA) produce a class of lymphokines of identical size with an isoelectric point of 6.0 to 6.5 that possess the same biologic properties as measured in murine lymphocyte response systems.

Human peripheral blood lymphocytes from an HLA-Dw1,3 individual were primed in vitro with influenza A virus (A/Texas/1-77/x-49) and subsequently cloned by limiting dilution in TCGF. Of the 96 TLCs originally obtained, nine were characterized in detail. TLCs were antigen specific, responding to influenza A virus, not to influenza B, TGAL, GAT, tetanus toxoid, or KLH, and only when antigen was presented by cells unable to form rosettes with AET-treated SRBC. Presentation of antigen by unseparated PBL often resulted in significant "back stimulation," probably via production of growth factors. The MHC requirements for the induction of TLC proliferation were analyzed. Of four representative clones analyzed, three required Dw1;DR1 compatibility for successful presentation of viral antigens by a panel of antigen-presenting cells. In contrast, one TLC showed an unusual pattern of response that could not be correlated to a particular HLA haplotype. Monoclonal anti-T cell antibody analysis of the surface phenotype of two TLCs maintained in continuous culture for 5 mo indicated that they were OKT3+, 4+, and 8-, consistent with an inducer/helper phenotype. To confirm the clonal nature of TLCs, data on the functional properties of TLC subclones are also presented.

In this report, the ontogeny of precursors of T cell growth factor (TCGF)-producing cells in the mouse thymus was investigated using a recently described limiting dilution microculture system. In agreement with previous studies, in the adult thymus TCGF production by cells stimulated by alloantigens was largely the property of the Lyt-2-negative subpopulation. Furthermore, when Lyt-2-negative cells were stained with monoclonal antibody GK-1.5 and sorted according to fluorescence intensity, all precursors of TCGF-producing cells were quantitatively recovered in the GK-1.5-positive subpopulation. During ontogeny, TCGF production by Lyt-2-negative thymocytes was first detectable on the 19th day of embryonic development at which time the precursor frequency was 1/10th that found in the adult thymus. As in the adult thymus, all precursors of TCGF-producing cells had the GK-1.5-positive, Lyt-2-negative phenotype. In parallel to these functional studies, the ontogeny of GK-1.5+, Lyt-2- cells was investigated. In the adult thymus, 80% of cells expressed both GK-1.5 and Lyt-2 antigens, whereas minor subpopulations of 10% and 5% (corresponding to phenotypically mature thymocytes as defined by cortisone-resistant thymocytes [CRT]) expressed GK-1.5 or Lyt-2 exclusively; 3% of cells expressed neither antigen. During ontogeny, thymocytes expressing both GK-1.5 and Lyt-2 first appeared on the 16th day of embryonic development and their proportion increased rapidly thereafter. Interestingly, the GK-1.5+, Lyt-2- subpopulation first appeared in significant numbers on day 19 in parallel with the appearance of functional TCGF activity. Taken together with our previous studies correlating cytolytic T lymphocyte precursor (CTL-P) activity with the Lyt-2+, GK-1.5- subpopulation, these results further emphasize the strict correlation between functional activity and mature surface phenotype of both embryonic and adult thymocytes.

The fate in culture of the T cell growth factor (TCGF), which is required for continued growth of human cultured T cells (CTC) in vitro, was studied. TCGF activity was stable for 7 days at 37 degrees C. However, it was no longer detectable after incubation with actively growing CTC at 37 degrees C for 3 days. This loss of TCGF activity also occurred quite rapidly and was detectable within 1 hr of incubation of 0.3 ml supernatant with 2 to 5 x 10(7) CTC at 23 degrees C. 2 x 10(8) mononuclear peripheral blood leukocytes were not effective in removing TCGF activity, and incubation with similar numbers of cells from B and T cell lines had no effect. Three-day-old concanavalin A and phytohemagglutinin blasts were very reactive with TCGF, so that 10(7) or 2 x 10(7) cells consistently removed TCGF activity. These experiments suggested specific absorption of TCGF by activated T cells, and led us to develop a model of ligand-activated TCGF-induced proliferation of T cells: Ligands induce production of TCGF by T-producer cells and deliver a first signal to the T-responder cells. This causes a receptor for TCGF to appear on T-responder cells. Only then does TCGF deliver the obligatory second signal that is needed to drive the T-responder cells into proliferation.

Continuous lines of murine cytotoxic T lymphocytes (CTL) directed to type A influenza viruses have been generated in vitro by stimulation of individual CTL precursors in the presence of T cell-growth factor TCGF and syngeneic virus-infected stimulator cells. The cloned CTL lines are H-2 restricted in their target cell recognition and exhibit distinct patterns of influenza virus recognition. All CTL lines appear to be restricted in target cell recognition to either the H-2K or the H-2D end of the appropriate H-2 haplotype. Likewise, CTL lines of F1 origin are restricted in recognition exclusively to one of the parental haplotypes. All CTL lines examined express the Thy-1.2 and the Lyt-2-surface antigen markers. 4 of 11 cytotoxic lines examined also expressed detectable levels of the Lyt-1- surface antigen. These findings confirm at the clonal level previous observations on the H-2K/D restriction of virus-specific CTL and also demonstrate heterogeneity among H-2 restricted CTL both from the standpoint of viral antigen recognition and cell surface phenotype.

The immunologic potential of T lymphocytes and antigen-presenting cells (APC) from male and female mice were compared. Lymphocytes from female mice or from male mice that cannot produce and respond to testosterone (Tfm/y) were more reactive than male lymphocytes to alloantigens in MLR. Spleen cells from Tfm/y mice equipped with estrogen implants showed a higher responsiveness than control Tfm/y to alloantigens. The removal of suppressive adherent cells or the addition of T cell growth factor (TCGF) enhanced the proliferative activity of the cells in the MLR. The responsiveness of female cells to alloantigens, however, remained superior to that observed in male cells. Similarly, in the presence of TCGF, thymocytes from female mice react more effectively than male cells in MLR. In addition, Con A-stimulated spleen cells from female mice produce more interleukin 2 (IL 2) than do spleen cells from males or female mice treated with testosterone. Lymphocytes from immunized mice were tested for their ability to respond to soluble antigens (KLH and OVA) in vitro. Again, female immunocompetent cells respond more vigorously than male cells or cells originating in female mice with testosterone implants. APC from female spleen were more efficient than male APC in initiating a secondary response in primed lymphocytes from either males or female mice. Moreover, castration of male mice enhanced, and treatment of female mice with androgen reduced, the efficiency of antigen presentation. In conclusion, these data suggest that female cells are superior to male cells in immunologic functions that are known to be associated with reactions to and recognition of histocompatibility antigens, i.e., antigen presentation and MLR. Furthermore, our present data indicate that the differential reactivity of immunocytes between male and female mice depends on the hormonal balance of the animal.

Radiolabeled molecules from detergent-solubilized human T cell blasts were fractionated on affinity supports coupled with T cell growth factor (TCGF) and anti-Tac antibody. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis indicated that a glycoprotein of approximately 58,000 mol wt was bound in both cases. Sequential binding to the two affinity supports demonstrated that the molecules recognized in each instance were identical. Thus, the Tac antigen contains the cellular binding site for TCGF. Note added in proof: Preliminary binding experiments using high concentrations of [3H]leucine, lysine TCGF with a low specific radioactivity indicate the existence of a sizeable pool of receptor sites with an affinity 2,000-10,000 times lower than that of the high affinity receptors measured in Fig. 1. Such sites may explain the numerical discrepancy between early TCGF binding experiments (5) and the binding of the anti-Tac antibody. The hypothesis that the anti-Tac antibody apparently reacts with both classes of receptor would explain its effect on the physiological response (high affinity binding) and the high level of Tac antigen on the cell surface.

Long-term cultures of human cytotoxic T-cell lines (H-CTLL) were established. H-CTLL cells were strictly dependent on growth upon a T-cell growth factor (TCGF) produced by phytohemagglutinin-stimulated normal human peripheral blood lymphocytes. H-CTLL cells were maintained in TCGF-dependent exponential proliferative culture for over 4 mo during which time they continued to mediate stimulator antigen-specific cytotoxicity as measured by a 4-h 51Cr-release assay. H-CTLL cells recovered from cryopreserved stocks and re-established in long-term culture demonstrated similar high levels of antigen-specific cytotoxicity. H-CTLL cells were 95--100% E-rosette positive and expressed normal T and Ia-like cell surface markers. The ability to sustain differentiated antigen-specific T-effector cells in long-term culture may provide a new means for the study of both the mechanism and regulation of T-cell-mediated immunity.

Murine mast-cell and T-cell growth factor activities, distinct from interleukins 3 and 2 (IL-3 and IL-2), have been identified and partially purified from the supernatant of the activated helper T-cell line Cl.Ly1+2-/9. This mast-cell growth factor (MCGF) activity supports only low levels of proliferation of several IL-3-dependent mast-cell lines and synergistically enhances the growth of mast cells in the presence of IL-3. The T-cell growth factor (TCGF) stimulates the proliferation of several T-cell lines, but to a lesser extent than recombinant IL-2. The MCGF and TCGF activities were not separable despite multiple biochemical fractionations, suggesting that both activities reside in the same protein. The MCGF/TCGF was separated from endogenous IL-3 by cation-exchange chromatography at neutral pH and could be distinguished from IL-2 by unique elution conditions from reverse-phase columns. Two bands of MCGF/TCGF activity were eluted from gels after sodium dodecyl sulfate/PAGE; under nonreducing conditions, the activities corresponded to molecular masses of 20 and 15 kDa, while after reduction, the molecular masses were 21 and 16 kDa. Thus, both activities may correspond to single polypeptide chains. The majority of the MCGF/TCGF activity appears to reside in the 20-kDa species, which displays a pI of 6.2 on chromatofocusing.

A method has been developed for isolating and growing lymphoid cells infiltrating murine solid tumors. When tumor cell suspensions are placed in T cell growth factor (TCGF) lymphoid cells and not tumor cells proliferate. By 7 to 9 days lymphoid cells free of tumor can be harvested and grown continuously in TCGF. The growth of these cells is dependent on the presence of TCGF but not on the presence of concanavalin-A. Tumor cell suspensions that have been freed of T lymphoid cells by treatment with anti-Thy-1.2 serum and complement or by fractionation on Ficoll gradients exhibit no cell out-growth in TCGF. The lymphoid cells growing in TCGF exhibit significant cytotoxicity for syngeneic tumor cells and normal fibroblasts grown in culture but do no lyse normal lymphoid cells.

Activated mature T cells require T-cell growth factor (TCGF) for continuous proliferation. However, many mature T cells infected with human T-cell leukemia-lymphoma virus grow independently of exogenously added TCGF. It is now reported that cells infected with this virus also lack detectable TCGF messenger RNA (less than one copy per cell) and thus do not produce their own growth factor. The results apparently rule out an autostimulation mechanism of growth control.

The mechanism whereby Cyclosporin A (CsA) inhibits secondary mixed lymphocyte responses was assessed. CsA added to secondary MLR cultures inhibited proliferation and induction of cytolytic lymphocyte activity. This inhibition was found to be associated with the inhibition of T lymphocyte stimulating growth factor(s) (TCGF) production in the supernatants of secondary MLR cultures. As little as 1.0 micrograms/ml of CsA added to secondary MLR cultures resulted in no measurable TCGF activity. In contrast, moderate doses of CsA (1.0, 2.5 micrograms/ml), which completely inhibited the secondary MLR response to alloantigen, did not inhibit the proliferative and CML response of alloantigen-primed lymphocytes to these stimulating growth factors. Even at high doses of CsA (20 micrograms/ml), substantial levels of proliferation (50% of control response) and CML induction (60% of control response) were observed when the primed cells were exposed to secondary MLR supernatants containing TCGF activity. It was concluded that inhibition of secondary mixed lymphocyte responses by CsA may be due in part to the inhibition of TCGF production rather than the inhibition of the effect of TCGF on mature cytotoxic T lymphocytes.

Cell-free conditioned media from human T cells transformed by human T-cell leukemia-lymphoma virus (HTLV-I) were tested for the production of soluble biologically active factors, including several known lymphokines. The cell lines used were established from patients with T-cell leukemia-lymphoma and from human umbilical cord blood and bone marrow leukocytes transformed by HTLV-I in vitro. All of the cell lines liberated constitutively one or more of the 12 biological activities assayed. These included macrophage migration inhibitory factor (MIF), leukocyte migration inhibitory factor (LIF), leukocyte migration enhancing factor (MEF), macrophage activating factor (MAF), differentiation inducing factor (DIF), colony stimulating factor (CSF), eosinophil growth and maturation activity (eos. GMA), fibroblast activating factor (FAF), gamma-interferon and, in rare instances, T-cell growth factor (TCGF). Some cell lines produced interleukin 3 (IL-3), platelet-derived growth factor (PDGF), or B-cell growth factors (BCGF). Such cells should prove useful for the production of lymphokines and as sources of specific messenger RNA's for their genetic cloning.

Concanavalin A (Con-A)-induced suppressor T cells were found to respond to T cell growth factor (TCGF) by proliferation. TCGF abrogated the suppressor activity exerted by these cells on phytohemagglutinin (PHA)- and alloantigen- induced lymphocyte proliferation and on pokeweed mitogen (PWM)-driven immunoglobulin secretion. The Con-A-activated suppressor T cells absorbed the TCGF activity, preincubation of these active suppressor cells with TCGF abolished their suppressor activity and addition of increasing numbers of Con-A-activated T cells reverted the abrogator,/ effect of TCGF. Altogether, these findings suggest that Con-A-induced suppressor T cells exert their function by decreasing the available levels of TCGF. Cyclosporin-A (CYA), which is known to inhibit the expression of receptors for TCGF on T cells, also inhibited the suppressor activity as determined in both indicator systems, namely PHA- or alloantigen-induced DNA synthesis and PWM-induced immunoglobulin synthesis. CYA made Con-A-treated T cells unresponsive to TCGF and unable to absorb the growth factor, supporting the notion that CYA inhibits the expression of TCGF receptors on T cells, a mechanism by which this drug seems to abrogate Con-A-induced suppressor T cell function.

Procedures are described for the isolation of a mast cell growth factor (MCGF) from medium conditioned by mitogen-activated splenic leukocytes (CM). Although optimal conditions for the production of MCGF in CM are identical to those for the production of T cell growth factor (TCGF), MCGF can be dissociated from TCGF after the first stage of purification on a DEAE-cellulose column. MCGF elutes from the column in the breakthrough fraction, whereas TCGF binds avidly to DEAE and is eluted only at high salt concentration. MCGF also differs from TCGF with respect to m.w. (as estimated by Sephadex G-150 chromatography) and sensitivity to trypsin. In addition, MCGF is produced by the murine myelomonocytic leukemia WEHI-3 and the radiation induced thymic lymphoma LBRM-33 cells, whereas TCGF is produced only by the latter in the presence of a mitogen. Another hemopoietically active factor, granulocyte colony-stimulating factor (G-CSF) present in media conditioned by WEHI-3 and LBRM-33 cells, however, shares a number of properties with MCGF. Although studies with purified or partially purified MCGF have thus far failed to reveal a correlation between MCGF and G-CSF, further biochemical analyses are necessary to dissociate MCGF from G-CSF.

Using a cloned cDNA copy of T-cell growth factor (TCGF) mRNA from the Jurkat leukemic T-cell line, we have isolated three overlapping TCGF genomic clones from a human DNA library. The entire TCGF gene is contained within two adjacent EcoRI fragments spanning about 8 kilobases. The complete nucleic acid sequence was determined. The gene is divided into four exons. The 5' untranslated region and the first 49 amino acids of the protein, 20 of which constitute a signal polypeptide and are not present in the secreted protein, are encoded by the first exon. Exons 2 and 3, separated from each other by a long intervening sequence, contain coding information for the next 20 and 48 amino acids, respectively. The remaining 36 amino acids and the 3' untranslated region are contained in the fourth exon. A promoter sequence T-A-T-A-A-A is present 77 base pairs (bp) upstream from the translation initiation site, and a CAT homology region occurs 104 bp upstream from the initiation site. A putative site for initiation of mRNA transcription was identified 53 bp 5' of the translation initiation codon. The organization of the gene was shown by Southern blot analysis to be identical in normal peripheral blood lymphocytes and in a variety of malignant lymphoid cell types. Restriction analysis of these cellular DNAs produced results exactly as predicted by the map for the cloned genomic TCGF, indicating that there is only a single copy of the human TCGF gene.

A murine monoclonal antibody directed against human T-cell growth factor (TCGF) from the JURKAT cell line was used for affinity column purification of the factor. Bound TCGF was eluted nearly quantitatively at low pH, and the recovered factor appeared homogeneous by two-dimensional gel electrophoresis. The molecule is markedly hydrophobic, with a high content of leucine. A single NH2-terminal sequence of 36 residues was obtained by automated Edman degradation, further supporting the homogeneity of the material. Thus, significant quantities of purified TCGF have been prepared in a single step, making possible detailed analysis of its molecular structure and biological role.

Mice were examined for the presence of splenocytes specifically cytotoxic for a rat insulinoma cell line (RIN) during the induction of diabetes by streptozotocin (SZ) in multiple low doses (Multi-Strep). Cytotoxicity was quantitated by the release of 51Cr from damaged cells. A low but statistically significant level of cytolysis (5%) by splenocytes was first detectable on day 8 after the first dose of SZ. The cytotoxicity reached a maximum of approximately 9% on day 10 and slowly decreased thereafter, becoming undetectable 42 d after SZ was first given. The time course of the in vitro cytotoxic response correlated with the degree of insulitis demonstrable in the pancreata of the Multi-Strep mice. The degree of cytotoxicity after Multi-Strep was related to the number of effector splenocytes to which the target RIN cells were exposed and was comparable to that detectable after immunization by intraperitoneal injection of RIN cells in normal mice. The cytotoxicity was specific for insulin-producing cells; syngeneic, allogeneic, and xenogeneic lymphocytes and lymphoblasts, 3T3 cells, and a human keratinocyte cell line were not specifically lysed by the splenocytes of the Multi-Strep mice. This phenomenon was limited to the Multi-Strep mice. Splenocytes from mice made diabetic by a single, high dose of SZ exhibited a very low level of cytotoxicity against the RIN cells. The cytotoxic response was also quantitated in splenocytes from control and Multi-Strep mice (10 d after the first dose of SZ) before and after culture with mitomycin-treated RIN cells in the presence of T cell growth factor (TCGF). The cytotoxicity of the Multi-Strep splenocytes was enhanced more than fivefold after such culture, suggesting the proliferation of an effector cell that could be stimulated and supported in vitro by TCGF. These results support the hypothesis that cell-mediated anti-beta cell autoimmunity may play a role in the destruction of the beta cells in this animal model. The stimulation of this response by TCGF may provide a tool by which enough cytotoxic effector cells could be obtained to establish their possible direct pathogenetic role in the induction of insulin-dependent diabetes. In addition, such cells will be a valuable tool to define the specific beta-cell antigens that may direct the highly selective cell-mediated destruction of these cells in experimental models and, perhaps, in human insulin-dependent diabetes mellitus.

The partial amino acid sequences of human T-cell growth factors (TCGFs) isolated from normal peripheral blood lymphocytes and from a leukemia T-cell line (Jurkat) show that the amino-terminal sequences of the two proteins (15 residues) are identical. Oligonucleotides based on the published Jurkat TCGF DNA sequence were used to isolate six cDNA clones of TCGF mRNA from normal lymphocytes. The predicted amino acid sequence of normal lymphocyte TCGF was identical to the sequence of the Jurkat protein, showing that the differences in biochemical properties of the two proteins result from post-translational events. Amino acid and nucleotide sequence data suggest that TCGF is derived from a precursor polypeptide that is cleaved at the amino terminus but not at the carboxyl terminus. Hybridization of the cloned lymphocyte TCGF cDNA to cellular DNA and RNA strongly suggested that the TCGF gene is expressed as a single mRNA species from a single-copy gene. No differences in the organization of the TCGF gene in normal, leukemic, and human T-cell leukemia/lymphoma virus-infected cells was detected regardless of whether they produce TCGF or not.

Adult T cell leukemia (ATL) and Sézary leukemia are malignant proliferations of T lymphocytes that share similar cell morphology and clinical features. ATL is associated with HTLV (human T cell leukemia/lymphoma virus), a unique human type C retrovirus, whereas most patients with the Sézary syndrome do not have antibodies to this virus. Leukemic cells of both groups were of the T3, T4-positive, T8-negative phenotype. Despite the similar phenotype, HTLV-negative Sézary leukemic cells frequently functioned as helper cells, whereas some HTLV-positive ATL and HTLV-positive Sézary cells appeared to function as suppressors of immunoglobulin synthesis. One can distinguish the HTLV-positive from the HTLV-negative leukemias using a monoclonal antibody (anti-Tac) that appears to identify the human receptor for T cell growth factor (TCGF). Resting normal T cells and most HTLV-negative Sézary cells were Tac-negative, whereas all ATL cell populations were Tac-positive. The observation that ATL cells manifest TCGF receptors suggests the possibility that an abnormality of the TCGF-TCGF receptor system may partially explain the uncontrolled growth of these cells.

By using human T-cell growth factor (TCGF), 10 cell lines were established from tissue samples of 10 patients with adult T-cell leukemia (ATL). Three cell lines were adapted to growth in medium lacking TCGF. The surface markers of all cell lines were characteristic of inducer/helper T cells, i.e., OKT3+, OKT4+, OKT6-, OKT8-, OKIa1+, and human Lyt2+ and Lyt3+, except that one cell line was OKT3-. The expression of the viral antigen was examined during establishment of 8 of the 10 cell lines. The viral antigen was not expressed in leukemic cells before cultivation. In 5 lines, the viral antigen was detected by immunofluorescent staining after a short period of cultivation. However, 3 cell lines, ATL-6A, ATL-9Y, and ATL-1K did not express the viral antigen during short-term culture: the ATL-6A and ATL-9Y cell lines became positive for the viral antigen after 5 and 2 months of cultivation, respectively; the ATL-1K cell line remained antigen-negative throughout a culture period of 13 months. Southern blot hybridization assay showed that all of the cell lines, including the viral antigen-negative ATL-1K cell line, contained the viral genome. Thus, the retrovirus was associated with all 10 cell lines established from ATL patients, but there was a heterogeneity in the expression time of the retroviral antigen in leukemic cells maintained in vitro. Our findings suggested that the expression of the viral antigen was not required for maintenance of the leukemic state in vivo and for growth of leukemic cells in vitro.