Reprints Available Directly from the Publisher Photocopying Permitted by License Only Stromal-cell and Cytokine-dependent Lymphocyte Clones Which Span the Pre-b-to B-cell Transition

Five stromal-cell-dependent lymphocyte clones are described that correspond to late pre-B or early B-cell stages of differentiation..They are useful for determining the molecular requirements for pre-B replication, for studying the stromal cells that supply those factors, and for delineating the final sequence of differentiation events as newly formed lymphocytes prepare to exit the bone marrow. The efficiency of lymphocyte growth at limiting dilution varied substantially on different stromal-cell clones and may reflect functional heterogeneity of stromal cells. Most lymphocyte clones were similar to uncloned lymphocytes from Whitlock-Witte cultures in that they responded only transiently to interleukin-7 (IL-7) and then died, unless maintained on a stromal-cell clone. One unusual lymphocyte clone (2E8) was propagated for more than 1 year in IL-7 alone and was selectively responsive to that cytokine. Most of the lymphocyte clones were not tumorigenic in immunodeficient mice. However, one pre-B clone (1A9) grew autonomously in culture when held at high density, responded to conditioned medium from a number of cell lines, and was tumorigenic. Tumors derived from this clone were infiltrated by stromal cells and lymphocytes taken from the tumors' retained characteristics of the original clone. Ly-6 antigens were inducible on 2E8 and 1A9 cells, but the lymphocytes were otherwise arrested in differentiation. The 2E8 cells had rearranged and expressed n: light-chain genes but displayed them on the surface along with surrogate light chains and heavy chains. Thus, expression of authentic Tight chain need not coincide with termination of surrogate light-chain utilization in newly formed B cells. Several glycoproteins have recently been demonstrated to be associated with surface immunoglobulin (Ig) on mature B-lineage cells and plasma-cell tumors. We now show that one member of this family (approximately 33 kD) was associated with the +surrogate light-chain complex on the 1A9 pre-B-cell clone. When compared to mature B lymphomas, fewer bands coprecipitated with the surface-labeled Ig isolated from pre-Band early B-cell lines, suggesting that components of the antigen receptor are sequentially acquired during development. The normal replication and differentiation of pre-B cells is probably regulated by complex interactions with multiple cytokines and matrix components of the marrow microenvironment. Cloned lymphocyte lines that are dependent on stromal cells should continue to be important tools for molecular definition of those interactions.

However, one pre-B clone (1A9) grew autonomously in culture when held at high density, responded to conditioned medium from a number of cell lines, and was tumorigenic. Tumors derived from this clone were infiltrated by stromal cells and lymphocytes taken from the tumors' retained characteristics of the original clone. Ly-6 antigens were inducible on 2E8 and 1A9 cells, but the lymphocytes were otherwise arrested in differentiation. The 2E8 cells had rearranged and expressed n: light-chain genes but displayed them on the surface along with surrogate light chains and heavy chains. Thus, expression of authentic Tight chain need not coincide with termination of surrogate light-chain utilization in newly formed B cells. Several glycoproteins have recently been demonstrated to be associated with surface immunoglobulin (Ig) on mature B-lineage cells and plasma-cell tumors. We now show that one member of this family (approximately 33 kD) was associated with the +surrogate light-chain complex on the 1A9 pre-B-cell clone. When compared to mature B lymphomas, fewer bands coprecipitated with the surface-labeled Ig isolated from pre-Band early B-cell lines, suggesting that components of the antigen receptor are sequentially acquired during development. The normal replication and differentiation of pre-B cells is probably regulated by complex interactions with multiple cytokines and matrix components of the marrow microenvironment. Cloned lymphocyte lines that are dependent on stromal INTRODUCTION Cytokine-dependent cell lines have been extremely valuable as biological indicators for investigating the growth requirements of lymphoid and hemopoietic cells. One lymphocyte clone (2E8) was isolated from murine long-term bone marrow cultures and used for the initial selection of a series of stromal-cell *Corresponding author.
clones . We now describe four additional stromal-cell-dependent lymphocyte clones that have different growth requirements and potential for tumor formation in vivo. We also report that surrogate light chains and associated proteins were detectable on the surface of cells that appear to represent the final stages in B-lymphocyte formation. These long-term cultured lymphocytes may or may not have exact counterparts in normal bone marrow. Hoewever, they reveal functional hetero-

General Characteristics and Patterns of Gene Expression
Four new lymphocyte clones were established by limiting dilution from long-term WhitlockoWitte cultures. Three of our lymphocyte clones were derived from BALB.xid strain mice. These may be classified as immature B cells, expressing both and c, as well as the CD45R epitope detected by the monoclonal 14.8 antibody ( Table 1). The 1A9 lymphocyte clone was isolated from cultures derived from normal BALB/c-strain mice and it is a pre-B-cell line. A detailed characterization was then performed to determine which B-lineage genes are expressed in ttese clones and localize them at particular developmental stages. Two of the clones initially bore the BP-1 antigen, which is preferentially expressed on pre-B cells and newly formed B cells in normal bone marrow (Cooper et al., 1986;Wu et al., 1990; Table 1). Some subcultures of the 2E8 clone subsequently lost this marker. Class II antigen is normally acquired at the same time, or soon after, expression of surface Ig molecules  and this was just detectable on one clone (2H6) and clearly positive on three others (2E8, 3A6 and 2G5). Only the 1A9 clone expressed the integral membrane proteoglycan Syndecan, which has previously been found on pre-B cells (Sanderson et al.,. 1989).
The 70Z/3 pre-B lymphoma line has been extensively studied as a model of inducible n: light-chain expression (Paige et al., 1978). We did not find evide.nce for a similar response in 1A9 cells, whose light-chain genes are in germ-line configuration and the characteristics of 1A9 cells recovered from in vivo tumors were similar to the 1A9 culture line (see what follows). Two of the lines (2E8 and 1A9) were responsive to yIFN and this resulted in strong expression of the Ly-6A antigen (Table 1 and data not shown). This was not seen with the 2H6 and 3A6 clones and in none of the clones did the density of class II antigens change with IFN exposure. These experiments do not rule out the possibility that some or all of the cloned cells have differentiation potential. However, they do indicate that their characteristics are relatively stable with prolonged culture. In contrast to another recently described group of lymphocyte clones (Tominaga et al., 1989), none of our clones had detectable CD5 (Ly-1) antigen. In this respect, they were like the bone marrow precursors from which they derived.
Previous studies demonstrate that expression of certain genes generally corresponds to stages of B-lineage differentiation (Sakaguchi and Melchers, 1986;Zimmerman et al., 1986;Bender and Kuehl, 1987;Park and Osmond, 1987). Selected probes were used to compare transcripts from our lymphocyte clones and several transformed cell lines by Northern blot analysis (summarized in Table 2 and selected examples shown in Fig. 1). B29 is a member of the Ig superfamily that is highly specific for B-lineage cells (Hermanson et al., 1988a) and has sequence similarity to an Ig-associated protein (Reth, 1989). It was expressed at a high level in all of these cells. More variable were levels of transcripts for B37, which is thought to encode a protein with homology to nuclear oncogenes (Hermanson et al., 1988b;Wall, manuscript in preparation). The same was true for the type II membrane protein, Lyb-2 (Nakayama et al., 1989). Only one of the new clones expressed terminal deoxynucleotidyl transferase (TdT), which is thought to be responsible for insertion of nongerm-line nucleotides into rearranging Ig genes (Desiderio et al., 1984). The ,s gene is thought to encode the co surrogate light chain (Sakaguichi and Melchers, 1986;Tsubata and Reth, 1990). Although a trace was detectable in the W279 B lymphoma, levels were extremely high in the pre-B Myb N-myc BP-1 C045 Lambda 5 B29 Actin FIGURE 1. Northern blot analysis of lymphocyte clones compared to pre-B lymphoma (70Z/3), B lymphoma (W279), and myeloma (MPC11) cells. Replicate Northern blots were repeatedly hybridized, stripped, and reprobed with the indicated cDNA probes. These results are representative of a more extensive analysis, which is summarized in Table 2. lymphoma (70Z/3) from which it was originally isolated, and our new lymphocyt6 clones.
The fact that many long-term cultured lymphocytes lack the CD45R epitope  was reflected by the loss of this marker on 1A9 lymphocytes with subculture (Table 1). It is interesting that Northern blots prepared with RNA from CD45R negative 1A9 cells had the short mRNA species typical of thymocytes, rather than the approximately 5-kb band in B cells (Fig. 1). Loss of the CD45R epitope may have resulted from a change in exon utilization (ThomaS et al., 1987) with time in culture. Similarly, an endopeptidase encoded by the BP-1/ 6C3 gene may be lost in transformed cells that contain a shortened transcript (Wu et al., 1990). BP-1 mRNA was most abundant in 2E8 cells that were strongly positive for the BP-1 antigen (Table 2 and Fig. 1).
A lymphocyte tyrosine kinase ,lck) was expressed by all of the lymphocyte clones, albeit in different amounts, whereas transcripts for a B-cell-specific tyrosine kinase (btk) were detectable only in 2E8 cells (Fig. 1). Products of these genes may be important for signal transduction in normal lymphocytes and are expressed at variable levels in established cell lines (Garvin et al., 1988;Dymecki et al., 1990). Altered expression of two protooncogenes has been associated with the transition of pre-B cells to B cells (Zimmerman et al., 1986;Bender and Kuehl, 1987). One of our B-cell clones contained an N-myc transcript, consistent with its retention of most other pre-B characteristics. N-myc mRNA was also detectable in the 1A9 pre-B-cell clone (Fig. 1). All of our clones expressed myb mRNA, whose transcription level drops dramatically in newly formed B cells (Bender and Kuehl, 1987). None of the lymphocyte clones had transcripts detectable by Northern analysis for CD14, ckit, IL-6, or IL-7.
Southern blot analyses were done to determine the rearrangement status of immunoglobulin genes in several of the clones. Both alleles were rearranged with respect to heavy-chain JH segments in 1A9 and 2E8 cells. One c allele was rearranged in 2E8 cells, whereas neither c nor light-chain genes were rearranged in the 1A9 pre-B-cell clone (data not shown).
On the basis of surface markers and the pattern of expression of mRNA species, we conclude that these lymphocyte clones correspond to the late pre-B-(1A9) or newly formed B-cell (2E8, 2H6, 2G5, and 3A6) stages of development. Of some 14 nonimmunoglobulin genes and markers whose expression was tested, only two, Syndecan and class II antigen, consistently distinguished the pre-Band B-cell clones.
Immunoprecipitated with: Light Chains and Ig-Associated Proteins The 1A9 lymphoycte clone had a number of interesting features. Firstly, it expressed substantial amounts of surface / heavy chains, but was negative for conventional c or light chains (Table 1). Rare murine and human pre-B-cell lines have been found with this characteristic and the membranebound / chains were associated with one or more "surrogate" light chains (Sakaguchi and Melchers, 1986;Kudo and Melchers, 1987;Pillai and Baltimore, 1987;Kerr et al., 1989). Cell-surface glycoproteins of 1A9 cells were biotinylated, extracted, and immunoprecipitated with rabbit anti-/. In addition to the 75-85-kD /-chain band, a doublet of approximately 18 and 19 kD bands was visible when this material was subjected to polyacrylamide gel electrophoresis ( Fig. 2). This was observable regardless of whether digitonin or Triton-X detergents were used (Fig. 2). Of particular interest was the finding that the more mature 2E8 cells also had these surface molecules, in Solubilized with Digitonin 1 2 3 4 5 6 FIGURE 2. Immunoprecipitation of biotin-labeled cell-surface antigens. The indicated cell lines were surface-labeled, and solubilized with either Trion X-100 or Digito.nin as indicated. Lysates were precleared with normal rabbit serum (NRb) coupled to protein G beads prior to secondary immunoprecipitation with rabbit antimouse/-chain-coupled beads. Precipitated material was analyzed by 12% (A) or 10% (B) SDS-PAGE under reducing conditions and visualized by exposure to an avidin-peroxidase conjugate. addition to the conventional 27kD c light-chain band. As noted before, the 25 transcript thought to encode the co surrogate light chain was detectable in all of the lymphocyte clones.
A complex of at least three proteins has recently been discovered in association with the surface Ig of mature B cells (Haustein'and Von der Ahe, 1986;Hombach et al., 1988;Campbell and Cambier, 1990;Chen et al., 1990;Hombach et al., 1990;Wienands et al., 1990). Little is known about their utilization on pre-B cells and this question was addressed with two of our new lymphocyte clones. These proteins were detectable only in digitonin (Oettgen et al., 1986) lysates ( Fig. 2A). The complex pattern of approximately 33-, 35-, and 36-kD species obtained with W279 B lymphoma cells (Fig. 1B) probably corresponds to the 32-39-kD proteins previously detected on mature IgM -+ B cells (Campbell and Cambier, 1990;Hombach et al., 1990). Only one clear band of approximately 35 kD was observed with extracts of 1A9 cells (Fig. 2). A slightly more complex pattern (probably reflecting the presence of 33and 35-kD species) was consistently observed in studies of 2E8 cells (Fig. 2). Thus, Ig-associated proteins may be displayed in a sequential pattern during B-lineage differentiation.

Growth on Cloned Stromal-Cell Lines
Cloning of lymphocytes in methyl cellulose on stromal-cell clones permits quantitative assessment of their growth requirements and provides information about the functional heterogeneity of stromal cells . The four new clones differed in two respects from an earlier lymphocyte clone (5A12), which was initially used for classification of our stromal-cell clones. Firstly, none proliferated on a spleen-derived stromal-cell clone (SS1) (Fig. 3). One possible explanation for this may lie in the fact that, in other studies, the spleen-derived stromal cells appeared by Northern blot analysis to elaborate little interleukin-7 (Gimble et al., 1989b on the BMNS1 stromal-cell clone, but did not grow on adherent layers of BMS1. A quite different pattern was observed with the 2E8 and 2H6 lymphocyte clones. Both of them consistently cloned with high efficiency on BMS2or BMSl-adherent cell layers. In other studies, the BMS2 stromal-cell clone has been subjected to detailed characterization because of its universal support of all lymphocyte clones (Gimble et al., 1989a;Gimble et al., 1989b;Gimble et al., 1990). The BMS1 stromal-cell clone appears to be highly selective for growth of 2E8 or 2H6, but not 1A9 lymphocytes. Correlation of the growth of these lymphocyte clones with expression of cytokine genes in the BMS1 and BMS2 cells may be informative with respect to molecular components of the bone marrow microenvironment.

Tumorigenicity
Due to the high proliferative rate of both the lymphocyte and stromal-cell clones used in these studies, we investigated their tumorigenicity by injecting the cells subcutaneously into immunodeficient (nu/nu) or SCID mice (Table 3). None of the stromal-cell clones formed tumors under these conditions and only the 1A9 lymphocytes produced palpable nodules within 6 to 21 days after injection.
BAll 1A9 tumors were always visible 6-21 days after inoculation. Tumors arose from 2E8 cells in this experiment only and were not visible before 7 weeks after inoculation. No tumors were detectable in subsequent experiments done with different freezings and observed for 8 weeks.
A striking finding was that the tumors contained large adherent cells with characteristics of stromal cells. In frozen sections of the 1A9 tumor examined with our KMA1.4 monoclonal antibody to stromal cell s, a reticular staining pattern was obvious (not shown). Furthermore, it was possible to isolate stromal cells from this source, which have the capacity to support the clonal growth of lymphocyte clones in the methylcellulose assay.
The 1A9 cells readily migrated beneath stromal cells in culture, a phenomenon that has been termed pseudoemperipolesis (Nishi et al., 1982). This was also the case with cells recovered in primary cultures of 1A9 tumors. It is noteworthy that the growth of such tumor-derived 1A9 cells was, like the parent 1A9 cells, stromal-dependent. These observations indicate that 1A9 cells may attract and interact with stromal cells in vivo. It would be interesting to learn the origin of these stromal cells and whether they contribute in tumor cell growth.

Responsiveness to Cytokines
Interleukin-7 (IL-7) is a potent short-term stimulus for the growth of normal pre-B cells in culture (Lee et al., 1989). When assessed by thymidine incorporation in 72oh cultures, the growth of all of the lymphocyte clones was augmented by IL-7 (data not shown). The clones are, therefore, similar to freshly isolated bone marrow cells in this respect. Continued addition of IL-7 to freshly isolated bone marrow cultures does not sustain cell division for more than a few weeks (Lee et al., 1989). Similarly, most of our lymphocyte clones survived DISCUSSION for only short periods when removed from stromal cells and cultured with this cytokine (Fig. 4). After The lymphocyte clones described here join many an initial small increase in cell number, cell death previously described hemopoietic cell lines that are appeared to balance cell division until survival could informative with respect to differentiation events no longer be maintained. In contrast, the 2E8 and the regulatory mechanisms that control them.
lymphocyte clone expanded logarithmically when These late pre-B and early B lymphocyte-stage cells IL-7 was the only stimulus. This clone has been have different requirements for survival and growth maintained in subculture with IL-7 for up to 8 in culture. The most representative of normal bone months. The 2E8 cells responded to IL-7 in a dose-marrow cells grow indefinitely on stromal cells and dependent fashion when evaluated with MTT or give a brief proliferative response to IL-7. An addithymidine incorporation assays (not shown). The tional factor or factors are needed for their longlower limit of cytokine concentration required for term propagation. One clone, which should be.
growth was similar to that reported previously for extremely valuable as a biological indicator, can another IL-7-dependent lymphocyte clone (Namen grow continuously in this cytokine alone. Yet Park et al., 1990; and Namen, unpub-another lymphocyte clone, which was the only one lished observations). A survey of other cytokines consistently capable of tumor formation, is stromalindicates that these cells can be exploited in IL-7 cell-dependent but responsive to autocrine signals. bioassays (Fig. 5).
Thorough ch,aracterization of these lymphocyte The 1A9 clone was found to be capable of auto-clones with monoclonai antibodies and cDNA nomous growth above a critical cell concentration probes revealed that quantitative, rather than quali- (Fig. 6). This was not the case with any of our other tative, changes in gene expression may occur as lymphocyte clones, but is similar to one recently maturing cells make the transition from pre-B cells described in another laboratory (Lemoine et al., to B cells. 1988). Those investigators obtained evidence for a The coordinate influence of many microenvironlow-molecular-weight cytokine that augmented mental stimuli provided by cytokines and contact growth at low cell densities. Autocrine growth with the extracellular matrix may be required for stimulation may also occur with 1A9 cells and condifferentiation of multipotential stem cells to funcditioned medium from any of the established clones, tional B lymphocytes. The lymphocyte clones or a pre-B lymphoma cell line was stimulatory (Fig. described here are indicators of growth and survival 7). It is interesting that conditioned medium from factors only. Although two of them responded to BMS1 cells even had this effect, although 1A9 cells interferon by expression of Ly-6A, there was little did not proliferate in direct contact with this indication that the clones could undergo significant stromal-cell clone ( Fig. 3 and  face and cytoplasmic markers was virtually unchanged even after recovery from nude mouse tumors. Random mutations may occur in lymphocytes held in long-term culture to result in a simplification of their growth requirements. The 2E8 clone would seem to represent cells that retain only the need for IL-7, whereas normal lymphocytes and the other cloned lines respond only transiently to this factor (Lee et al., 1989). Although our survey revealed no other cytokines that stimulate 2E8 cells, they do grow on the $17 stromal-cell clone (Collins and Dorshkind, 1987), which is thought not to elaborate IL-7 (K. Dorshkind, C. Paige, and O. Witte, personal communications). Thus, 2E8 cells might be useful for identification of a novel cytokine(s) made by that stromal-cell clone. Similarly, 1A9 cells could serve as indicators for a seemingly ubiquitous stimulus, which may have previously been studied by other investigators (Lemoine et al., 1988;Muirhead and Davis, 1990).
It should be noted that only large pre-B cells, and not the small pre-B or newly formed B cells from bone marrow, can be stimulated by   (Lee et al., 1989). The 2E8, 2H6, and 2G5 clones are all phenotypically B cells, that is, are surface IgM-positive and thus past the stage of differentiation that is normally responsive to this cytokine. These clones also FIGURE 6. Autonomous growth of 1A9 cells when held at high initial cell densities. The 1A9 clone was unique in being able to survive and grow at high density in liquid medium. For thse experiments, cells were removed from stromal cells and passed through G-10 columns before use.  Melchers, 1986;Tsubata and Reth, 1990) gene, for cell-cycle status. The final stages in B lymphopoieses which mRNA was easily detectable on Northern proceed with little or no cell division (Landreth et blots. It is particularly noteworthy that some sural., 1981). Therefore, IL-7 may promote the survival rogate light chain was detectable on 2E8 cells, which and continued proliferation of cells that are placed rearranged, expressed, and displayed conventional in cycle through other means. In this context, it is light chain on the surface. Thus, differential exprespossible that IL-7 prevents programmed cell death, sion of these two genes is not coordinately conas has recently been suggested for other hemo-trolled in this line. In a previous study of Abelsonpoietic growth factors (Williams et al., 1990). virus-transformed lymphocytes, acquisition of sur-There are a number of common characteristics, as face Ig also did not correspond to abrupt termiwell as some heterogeneity, among the lymphocyte nation of pre-B-cell characteristics (Bender and supporting stromal-cell lines that have been Kuehl, 1987). Furthermore, another group has just described (reviewed in Kincade, 1987; found that loss of surrogate light chain did not 1989). In terms of function, all lymphocyte clones accompany pre-B-cell maturation (Takemori et al., grew and could be cloned in methyl cellulose on the 1990). BMS2 stromal-cell clone. However, a more complex It has recently been demonstrated that at least pattern of growth was obtained with other stromal three proteins are associated with surface immunocell lines (Fig. 2) and the molecular basis for this is globulin on mature B cells (Haustein and Von der not yet clear. The most noteworthy difference in the Ahe, 1986;Hombach et al., 1988; Campbell and clones used for this study is their apparent ability to Cambier, 1990;Chen et al., 1990;Hombach et al., make IL-7 (Gimble et al., 1989b). Little or no mRNA 1990; Takemori et al., 1990;Wienands et al., 1990). for IL-7 was detected in preparations of our spleen-These are most easily visualized by use of digitonin derived stroma] cells and the lymphocyte clones did in the membrane-extraction procedure and a comnot grow well on them. However, we caution that plex of 33-36 kD was detectable with our biotinlevels of the transcript for this cytokine are always labeling procedure using the W279 B.lymphoma line at the limit of detection and can fluctuate with time (Fig. 1B). Similar analysis of one pre-Band one following subculture of stromal cells. Furthermore, newly formed B-cell clone indicated that fewer Igat least one stromal-cell subclone can actively sup-associated species are displayed on the surface of press lymphocyte growth . these immature cells. It is possible that this differ-Nishikawa and colleagues found that two of four ence is due to technical limitations in labeling low stromal-cell-dependent lymphocyte clones grew on a numbers of such molecules on the surface of pre-B stromal-cell clone that does not make, and cannot be cells and experiments are underway to address that induced to make, IL-7 (Nishikawa et al., 1988; Sudo possibility. Such Ig-associated proteins are thought et al., 1989). Until experiments with neutralizing to participate in signal transduction via antigen antibodies are performed, .we believe it unwise to receptors on B cells similar to the function of the attribute all proliferative activity to IL-7. CD3 complex on T cells (Oettgen et al., 1986). At Our 1A9 cells share an interesting feature previ-least one of them (MB 1; B34; IgMcz) is required for ously found in only some murine and human pre-B surface display of immunoglobulins on plasma cells lymphoma lines (Paige et al., 1981;Hardy et al., (Hombach et al., 1990). Our findings suggest that 1986; Pillai and Baltimore, 1987;Kerr et al., 1989; only one may be sufficient for surface expression of Park et al., 1990). That is, detectable quantities of/ /-plus surrogate light chain on late stage pre-B heavy chains are present on the cell surface in the cells. absence of conventional light chians. As with the Our analyses of these lymphocyte clones provide 70Z/3 lymphoma line, virtually all of the / chains the most detailed information yet available about were associated with "surrogate" light chains, which cells that are undergoing the final events in B-cell on reducing gels were approximately 18 kD (Pillai formation. It might be expected that expression of and Baltimore, 1987). However, in contrast to 70Z/3 many genes would abruptly change to facilitate cells, which are inducible for expression of c (Paige migration from the bone marrow (adhesion et al., 1978), conventional light-chain genes are molecules) and responsiveness to foreign antigen unrearranged in this new clone. At least one com-(Ig-associated proteins, tyrosine kinases, etc.). Howponent of the surrogate light chains precipitated ever, most of the changes that corresponded to the from 1A9 cells is a product of the 5 (Sakaguchi and display of complete surface immunoglobulin 158 K. ISHIHARA et al. molecules were quantitative. One exception to this pattern was the transmembrane proteoglycan Syndecan, which was detected on pre-B-, but not B-cell clones (Table 1). Further studies might reveal a unique function for Syndecan on maturing pre-B cells (Sanderson et al., 1989).

10
Lymphocytes grown in conventional long-term cultures only rarely form tumors and only after being maintained for longer than 6 months (Whitlock et al., 1984). Of the lymphocyte clones that we tested, only 1A9 was consistently tumorigenic. There is precedent for lymphocyte lines that depend on host-derived cytokines for tumor formation (Lasky and Thorbecke, 1989) and further studies might reveal the origin of stromal cells that are attracted to 1A9 tumors. The clone is also the best example we have found of lymphocytes that actively migrate beneath stromal cells in culture. The basis for this interesting phenomenon, known as "pseudoemperipolesis" (Witte et al., 1987), is unknown but could be explained in terms of chemotactic factors concentrated beneath stromal cells, or a polarization or redistribution of stromal-cell membranes.
We expect that these and other lymphocyte clones isolatecl from long-term bone marrow cultures will be valuable for detecting regulatory cytokines and for investigating stromal-cell functions. They may also be exploited in studies of the molecular events that occur as lymphoid cells progress from a normal to transformed phenotype. Finally, clones that span the pre-Bto B-cell transition stages provide an opportunity to explore mechanisms that control the surface display of surrogate light chains and other -chain-associated proteins.

MATERIALS AND METHODS
Establishment and Maintenance of Lymphocyte Clones Long-term bone marrow cultures from immunodeficient BALB/c.xid or normal BALB/c mice were prepared and maintained as described in our previous publications Hayashi et al., 1989). Lymphocyte clones were derived by limiting dilution on stromal-cell layers prepared from CBA/ HT6T6 marrow. They were routinely maintained throughout this study on stromal-cell clones or with recombinant IL-7. Generally, stromal cells (approximately 105 were plated the night previously in 100mm dishes (Corning) to yield subconfluent layers.
Approximately 2 xl0 lymphocytes were added per dish in 8 ml of Whitlock-Witte medium. These cultures were fed with 75% fresh medium at weekly intervals and subcultured monthly to fresh adherent stromal-cell layers. The IL-7-dependent 2E8 clone was adapted to growth in this cytokine and maintained at an approximate concentration of 5x105 cells/ml with 1,000 units/ml of murine or I ng/ml of human recombinant IL-7 (PeptroTech, Inc., Rocky Hill, NJ). Cells were harvested at least every 3 days, and adjusted to the original cell concentration with fresh medium and IL-7. This cell density was critical to maintainence of satisfactory viability and growth of 2E8 cells, while oxygen tension (5% 02 vs. ambient O2) had no obvious influence. The initial derivation and cloning of 2E8 cells was done with Whitlock-Witte medium, but growth was more vigorous in McCoy's Modified 5A medium (Gibco) with 5% FCS (Hyclone, Logan Utah, Lot # 1111885) 5 xl0 -s M 2-mercaptoethanol, L-glutamine, and Penn-strep. Functional assessment of stromal cells was done by means of the methylcellulose colony assay previously described in detail . Initial numbers of stromal cells were 3x104 per 35-mm dish (Lux) and colonies were scored 7 days later with a Zeiss IDO3 inverted microscope with a 10x objective. Foci containing greater than 40 cells were counted as colonies..

Cell-Surface-Marker Analysis
Immunofluorescence and flow cytometry were used to assess surface-marker expression on lymphocyte clones. Particularly valuable monoclonal antibodies were donated by Drs. J. Kearney (,), M. Cooper (BP-1), M. Bernfield (Syndecan), and U. Hammerling (Ly-6). Anticlass II antibodies (Iad) were purchased from Becton-Dickinson. Responsiveness to inductive stimuli was assessed after 48 h of culture at a concentration of 6.25 xl05 cells per ml in 100-mm tissueculture dishes (Corning). Either 500 units per ml of recombinant rat interferon gamma (Amgen) or 10 g per ml of lipopolysaccharide were added to particular cultures. Expression of surface n: and Ly-6 antigens was then assessed with a Coulter Epics V flow cytometer.

Cell-Surface Labeling and Immunoprecipitation
Cells were surface-labeled by a novel biotinylation procedure (Miyake et al., 1990;Miyake, in preparation). Briefly, after three washes in HBSS, cells were suspended in labeling buffer containing 100mM HEPES, 150 mM NaC1, pH 8.0, at a concentration of I xl0Y/ml. Sulfo-N-hydroxysuccinimidobiotin (sulfo-NHS-biotin; Pierce Chemical Co., Rockford, IL) was added to cell suspensions such that the final concentration of sulfo-NHS-biotin was 0.1 mg/ml. After a 30min incubation at room temperature with occasional shaking, cells were washed once in chilled RPMI 1640 and twice in RPMI 1640 supplemented with 3% FCS. The cells were lysed for 15 min on ice at a concentration of 5x107/ml in Triton X-100 lysis buffer (1% Triton X-100; 10 mM Tris, pH 7.5, I mM phenyl methyl sulfonyl fluoride (PMSF), 10/g/ml soybean trypsin inhibitor, 1/g/ml leupeptin, 1 U/ml aprotinin, 50 mM iodoacetamide, 0.1% NaN 3, and I mM EDTA) or in digitonin lysis buffer (identical to the previous but with 1% digitonin (Sigma)). After centrifugation, the lysates were precleared three times with 40 1 of immobilized protein G agarose (Pierce) conjugated with normal rabbit serum. The precleared lysates were added to 40/1 of rabbit antimouse IgM (Miles) conjugated protein G agarose. The mixture was rotated for 60 min at 4C and then washed three times with a 10mM Tris-HC1 buffer, pH 7.5, containing 150 mM NaC1, 0.1% NaN 3, and 0.2% Triton X-100, or 0.2% digitonin, respectively. The absorbed proteins were released by boiling for 5 min in a sample buffer containing 0.125 Tris-HC1, 2% SDS, 10% glycerol, and 5% 2ME, pH 6.8. Then SDS-PAGE electrophoresis was carried out on Laemmli 10-12% polyacrylamide gels.

Tumorigenicity
For the assessment of stromal-cell tumorigenicity, lx10 cells in 200./1 of PBS without Ca+//Mg ++ were injected subcutaneously into the backs of BALB/c.nu/nu or CB.17.SCID mice. As a control, similar numbers of NIH 3T3 cells were injected and this always resulted in tumor formation. In the case of lymphocyte clones, 1.5-2.0 X10 6. cells in 200 1 of PBS were injected subcutaneously. A PBS control was done for each experiment and the animals were checked for tumors biweekly for a period of 8 weeks.

Tumor Cell Recovery
Animals were euthanized with CO and submerged in 70% ethanol for I min. A 3-cc syringe with an 18G, l-in, needle was inserted into the tumor (the outer periphery) and tumor cells withdrawn. Two milliliters of media (RPMI 1640, 5% FCS) were then drawn into the syringe and the biopsy ejected into a sterile tube. A viable cell count was done using trypan blue dye exclusion and the cells were cultured in media consisting of RPMI 1640, 5% FCS, 5 xl0 -5 M 2 ME, L-glutamine, Penn/Strep at a concentration of 1 X10 cells per ml. Within 7-10 days, adherent cells could be seen in the culture flasks.

Frozen Sections
Small cubes of tumor tissue were immersed in O.C.T. compound (Miles Scientific) in plastic wells floating in 2 Methylbutane (Fisher) in a stainless steel beaker. The beaker was then placed in liquid nitrogen for I min or until the O.C.T. was almost completely frozen. Specimens were then stored in sealed bags at -20C before sectioning on a microtome. Immunoperoxidase staining was performed as in our previous studies (Witte et al., 1987) with monoclonal KMA1.4 antibodies to stromal cells (K. Miyake, manuscript in preparation).

Responsiveness to Cytokines
Survival and proliferation of lymphocytes were assessed by direct counting with Trypan blue due, by MTT assay (Mosmann, 1983), and by 3H thymidine incorporation . A survey was made to determine if the IL-7-dependent 2E8 clone is responsive to other recombinant cytokines.

Preparation of Conditioned Media
Confluent cultures of stromal cells were generated in 100-mm dishes and all medium was aspirated and replaced with 5 ml of Whitlock-Witte medium. After an additional 7 days of culture, conditioned medium was harvested and centrifuged at 5000rpm for 10min to remove cells and debris. Lymphocyte s were harvested and washed four times with medium to remove residual cytokines. Those maintained on stromal-cell feeder layers were also passed through a G10 Sephadex column to remove any contaminating stromal cells. The cells were cultured at 5 xl0 cells per ml, 2 ml per well in Corning 24 well plates and conditioned medium collected after 48 h. The conditioned medium was centrifuged at 5000rpm for 10min to remove dead cells and debris. All conditioned mediums were stored at 4C until assayed at final concentrations of 10-25% total volume.

ACKNOWLEDGMENT.S
This work was supported by grants Ab20069 and AI-19884 from the National Institute of Health. We thank Drs..Linda Thompson and Carol Webb for helpful comments on the manuscript.  (Accepted October 12, 1990)