Harwood Academic Publishers imprint, part of The Gordon and Breach Publishing Group. Printed in Malaysia Transformation of Undifferentiated Thy-1 Lymphoid Cells by the Abelson Murine Leukemia Virus

Intrathymic injection of the Abelson murine leukemia virus (A-MuLV) results in transformation of immature T and B lymphoid cells. In this report we demonstrate that the concentration of A-MuLV injected into murine thymi influences the selection of the transformation target. Thus, concentrated A-MuLV gives rise to Thy-1+ B220- thymomas. In contrast, dilute virus induces B220+ thymomas that also express low levels of Thy- 1 (Thy-1lo), a phenotype that is similar to marrow-derived progenitor B-lymphoid cells (pro-B cells) that are highly susceptible to A-MuLV transformation in vitro. However, rare B220+ lymphoid cells isolated from normal adult thymi were not transformed by A-MuLV in vitro, while B220+ cells isolated from bone marrow were highly susceptible to transformation by A-MuLV. The Thy-1lo B220+population in the primary thymomas had not rearranged TCRγ, TCRβ, or Igκ genes, but contained subpopulations that assembled Ig DJH or VDJH genes and were therefore similar to transformed pro- and pre-B cells obtained from A-MuLV infected fetal liver and adult bone marrow, respectively. However, unlike A-MuLV-transformed pro- and pre-B cells, many (40– 70%) of the Thy-1lo B220+ transformed thymoma cells had not rearranged Igh genes, and therefore appear to represent undifferentiated lymphoid cells. We conclude that A-MuLV may transform an undifferentiated lymphoid target in the thymus.


INTRODUCTION
The ability of the v-abl oncogene of the Abelson murine leukemia virus (A-MuLV) to immortalize immature lymphoid cells from bone marrow and fetal liver has been useful for identifying and characteriz-ing immature stages of B-cell development (Rosenberg and Witte, 1988;Alt et al., 1984). Depending on the extent of Igh rearrangement, most transformed cell lines derived from A-MuLV infected bone marrow or fetal liver can be divided into one of three categories. Cell lines in the most common category represent transformed pre-B cells that exhibit complete variable region formation on at least one allele.
However, because of the imprecision of VHDJH recombination, these transformants may or may not express cytoplasmic mu chains (Cg) (Alt et al., 1984). Even if a functional Igh variable region is formed and Ct is expressed, these pre-B cell transformants are usually arrested at this stage of B-cell differentiation by v-abl suppression of RAG gene expression and/or inhibition of NF-KB/rel-mediated intron enhancer activity Klug et al., 1994).
A second marrow-derived A-MuLV-transformed cell type is arrested at the DJ H stage of maturation. These cells fail to rearrange V H to DJ H and are referred to as DJ-fixed (DJ-f) (Ramakrishnan and Rosenberg, 1988;Wang and Rosenberg, 1993). A-MuLV transformed cell lines derived from fetal liver also have rearranged only DJ H segments, but unlike the DJ-f population obtained from marrow, are not blocked at this stage of development and continue assembling Igh genes so that _<10% of the cells express Cg. This third type of A-MuLV transformant is referred to as DJ-rearranging (DJ-r) (Ramakrishnan and Rosenberg, 1988).
Also of interest are the relatively rare A-MuLVinduced thymoma-derived cell lines reported to express low levels of Thy-1 (Thy-ll) (Cook, 1982;Cook, 1985;Cook and Balaton, 1987;Holland et al., 1991) because in murine hematolymphoid populations, Thy-11 often characterizes progenitor and stem cells (Kondo et al., 1997;Lian et al., 1997;McKenna and Morrissey 1998;Morrison and Weissman 1994;Miiller-Sieburg et al., 1986;Soloffet al. 1997). In addition, Thy-11 is also found expressed on early thymic progenitors (Antica et al. 1994;Carlyle et al. 1997;Hattori et al? 1994;Kawamoto et al. 1999;Wu et al. 1991). Unlike the other categories of A-MuLV-transformed thymoma cells that clearly represent distinct stages of T or B-cell development, the few A-MuLV-transformed Thy-11 cell lines that have been examined express an ambiguous phenotype that cannot unequivocally be assigned to a particular stage of T or B lymphocyte development (Cook, 1985;Cook and Balaton, 1987;Risser et al., 1985;D'Andrea et al., 1987;Holland et al., 1991). These thymoma cells may be analogous to cases of acute human leukemia in which malignant lymphoblasts with an ambiguous phenotype are thought to arise from the transformation of a transitory lymphoid progenitor poised at the juncture of commitment to either the T or B-lymphoid lineage (Paietta, 1992;Greaves, 1986). Since progenitor populations capable of giving rise to both T and B-cells can be found in the thymi of adult and fetal mice (Akashi et al. 1998;Hattori et al. 1994;Matsuzaki et al., 1993;Peault et al. 1994;Wu et al., 1991), it is feasible that primitive lymphoid progenitor cells not fully committed to either lymphoid lineage may be targets for A-MuLV transformation in the thymus. FIGURE Coexpression of B220 and Thy-11 cells on immature lymphocytes in primary thymomas induced by A-MuLV. Three representative thymomas in which 35%, 67% and 80% of the cells were B220 were analyzed for coexpression of B220 and Thy-1 by two color flow cytometry as described in the Materials and Methods. The data show that the B220 cells expressed low levels of Thy-1, and that B220cells were Thy-1hi. This is representative of 22 thymic tumors analyzed. B220 cells were also sorted to >95% purity over the anti-B220 magnetic column then stained with (solid histograms) FITC-R6-60.2 (anti-IgM), FITC-S7 (anti-CD43), PE-53-7.3 (anti-CD5), PE-RM2-5 (anti-CD2), or PE-M1/69 (anti-HSA) (data not shown) and analyzed by flow cytofluorometry. Isotype-matched irrelevant antibodies were used to establish background fluorescence (open histograms). These data are representative of nine Thy-11 B220 tumor populations that were examined. Staining profiles for normal thymus and electronically gated apleen and bone marrow B220 cells are included for control. While comparing thymomas induced by intrathymic injection of different retroviruses, we noticed that distinct Thy-ll and Thy-lhi populations often co-existed in the same primary tumor that was induced with either the v-abl or BCR-ABL oncogenes.
In contrast, we only observed Thy-lhi cells in tumors induced with the Moloney murine leukemia virus (Clark et al., 1993). The Thy-11 tumor cells could represent the in vivo counterpart of the rare Thy-11 A-MuLV transformed thymoma derived cell lines that are of an undetermined stage of lymphocyte development (Cook, 1985;Cook and Balaton, 1987;Holland et al., 1991). Since the Thy-1 l category of A-MuLV transformed thymomas may be relevant to early lymphocyte development (Clark et al., 1993;Cook, 1985;Cook and Balaton, 1987;Holland et al., 1991), it is important to better characterize these tumor cells and the normal thymocyte population that gives rise to them. Even though there are certain advantages to characterizing tumor populations by analyzing cell lines derived from them, we initially chose to characterize the Thy-11 cells in the primary thymomas in order to minimize phenotypic changes that often occur during adaptation to in vitro culture.

RESULTS
Coexpression of B220 and Thy-1 l on immature lymphocytes in primary thymomas induced by A-MuLV We previously reported that thymomas induced by either v-abl or BCR-ABL oncogenes usually contain distinct populations of Thy-lhi and Thy-ll cells (Clark et al., 1993). Further examination revealed that B220 + cells also were present at approximately the same frequency as the Thy-1 l cells (Clark et al., 1993). Therefore, we used two-color flow cytometry to determine whether cells in A-MuLV-induced thymomas coexpress Thy-11 and B220+. This analysis revealed the presence of two major thymoma subpopulations that could be distinguished by their pheno-type. One subpopulation expressed high levels of Thy-1 and was B220-. The second subpopulation was B220 + and expressed Thy-1 at about a 10-fold lower peak fluorescence than the Thy-lhi population (Fig. 1). Thus, the Thy-11 subpopulation that we previously identified in primary thymomas, also coexpresses B220. However, because the Thy-11 signal overlaps the isotype controls (data not shown), it is possible that some B220 + thymoma cells may be Thy-1-. In contrast to the thymic B220 + tumor cells, cells from peripheral lymphomas that often appear in thymoma-bearing mice did not express any Thy-1 signal above the isotype background (data not shown) which is typical for A-MuLV-transformed pre-B cells.
To further characterize the B220 + subpopulation, thymoma cells were first enriched over a magnetic column to >95% B220 + cells, then examined by immunofluorescent staining for surface Ig, CD2, CD5, CD43 and heat stable antigen (HSA or llD). The B220 + cells from all thymomas examined expressed a homogenous undifferentiated phenotype (Fig and Table I). Thus, flow cytometry revealed that B220 + thym0ma cells did not express surface IgM or Igc. In addition, Western blot analysis and immunostaining of cytospin preparations both failed to demonstrate expression of Cg in any of the B220 + tumor subpopulations. In contrast, using these methods we did detect Ct in normal spleen cells as well as in A-MuLV transformed bone marrow controls (data not shown). Flow cytometry further showed that B220 + tumor cells were CD5and CD2 neg-l but expressed uniform levels of both CD43 and HSA ( Fig. 1 and data not shown). These phenotypic data are summarized in Table I. Both Thy-1 + B220" and Thy-1 l B220 + thymoma populations arise from A-MuLV-infected cells To determine whether the different thymoma subpopulations arose from A-MuLV-infected cells, we first attempted to use DNA blot hybridization to examine the pattern of proviral integration. Thymoma populations were sorted on the basis of B220 expression over the magnetic column and a small amount of gag.abl (491 bp) FIGURE 2 Semiquantitative PCR measurement of relative proviral content shows that the B220 and B220thymoma subpopulations both arose from A-MuLV infected cells. After magnetically sorting thymomas into B220+and B220fractions, their DNA was amplified using primers that span a 491 bp gag-abl fragment and a 737 bp a-actin fragment which served as an internal amplification control. The relative proviral content was determined by normalizing the gag-abl signal to the actin signal. Lanes 1-5: Amplification standards were prepared by mixing DNA from normal liver and from the v-abl transformed cell line, LPC 5-9, in different proportions. Lanes 5-8: v-abl transformed pre-B cell lines. Lanes 9-14: sorted B220 thymoma cells. Bands were visualized and quantitated by Phosphorimager (Molecular Dynamics) from which the ratio of the gag-abl to the actin signal was calculated.
DNA was extracted from the subpopulations. The DNA was digested witl EcoRI, which does not cut within the A-MuLV genome, and then processed for Southern blot hybridization with a v-abl probe. DNA from the B220 + and B220subpopulations revealed a 30 kb c-abl genomic band along with several unclear bands of submolar hybridization intensity and/or a smear running from about 9-30 kb. These results were repeated on several Southern blots of DNA from column sorted tumor subpopulations (data not shown), which suggests that the subpopulations were oligoclonal with respect to proviral integration, as we previously reported for the unsorted thymomas (Clark et al., 1993). This fact prevented a close comparison of proviral integration in the different subpopulations that coexisted within individual thymomas. On the other hand, DNA obtained from several unsorted peripheral lymphomas revealed single distinct proviral bands for each tumor along with a strong signal from the 30 kb c-abl band. Therefore, unlike the thymomas, peripheral lymphomas were clonal in origin.
Because it was difficult to obtain high resolution on DNA blots of the sorted subpopulations, we used the PCR strategy illustrated in Fig. 2 to quantitate the relative proviral content in order to evaluate the contribution of A-MuLV infected cells to each sorted tumor subpopulation. Bands from the PCR reactions were quantitated by Phosphorimager and the relative proviral content of each subpopulation was calculated as the intensity of the 491 bp proviral gag-abl band normalized to the intensity of the 737 bp actin band. These values were checked against the proviral contents of standards prepared by serially diluting DNA from a v-abl transformed pre-B cell line with DNA from normal liver (Fig. 2, lanes 1-5). In the sorted B220 + (Fig. 3, lanes 9-14) and B220-(data not shown) tumor subpopulations, the mean relative proviral content was 0.7 (range 0.5-1.0), which was somewhat higher than the 0.5 mean relative proviral content of four clonal v-abl transformed bone marrow cell lines (Fig. 3, lanes 5-8, range 0.4-0.5). Nevertheless, this confirmed that both B220 + and B220thymoma subpopulations were infected with A-MuLV. . Thymoma samples were sorted by anti-B220 magnetic column, except for those marked in parenthesis which were not sorted because they were >85% B220 or >85% Thy-1hi. "L" denotes that the sample was a peripheral lymphoma that arose in an intrathymically injected animal. Thymomas are grouped based upon B220 phenotype and the extent of Igh rearrangement as determined by PCR according to the following scheme: (A) B220 thymomas with only DJ H rearrangement. (B) B220 thymomas with less VHDJH than DJ H rearrangement. b. The extent of rearrangement of the Igh gene is presented as the mean percent loss of germline structure as illustrated in Fig. 4. The values represent the mean from 3-6 separate amplifications from each sample. Standard errors of the mean percentage loss were <7%. <10% loss of germline is considered as no rearrangement. Rearrangement of the lg: light chain gene was not detected in any samples tested.
c. GG=germline configuration on both alleles, RR=complete rearrangement of both alleles, rG=minor rearrangement detected with predominant germline band, Rg=significant rearrangement detected with minor germline band. Parentheses indicate that the minor band could be attributed to contaminating cells in unsorted thymomas Significant subpopulations of Thy-1 l B220+A-MuLV transformed thymoma cells do not demonstrate Ig or TCR gene rearrangement In order to assess the extent of antigen receptor gene rearrangement in the Thy-11 B220 + subpopulation, DNA was extracted from thymoma subpopulations that were sorted for B220 expression on the magnetic column. The DNA was analyzed either by PCR amplification for Igh rearrangement or by Southern blot for TCR rearrangement. The PCR strategy for Igh rearrangement allowed DJ H and VHDJH rearrangement to be measured as a loss of germline structure relative to unrearranged liver DNA (Fig. 3A). In order to confirm that this analysis was quantitative, liver DNA was serially diluted with DNA from an A-MuLV transformed pre-B cell line, LPC 5.9, amplified by PCR (Fig. 3A), and the loss of Igh germline was determined by normalizing the unrearranged signal to the amplified actin signal. Figure 3B shows that the measured loss of Igh germline is linear and strongly correlates with the expected Igh rearrangement (correlation coefficient r .994 for loss of VII to DJ H germline and .997 for loss of D to JH germline).
An example of this analysis is shown in Fig. 3C  showed approximately equivalent VII to DJ H and D to JI-I rearrangement (Table IC). Southern blot analysis of. Igh rearrangement also was performed on some tumor populations, the results of which were in close agreement with the PCR data (Table IA and 1B and data not shown). In addition, four unsorted thymomas that were >85% B220 + were analyzed by Southern blot only which revealed predominantly Igh germline structure (Table ID); None of the Thy-1 B220 + thymoma subpopulations showed rearrangement of IgK light chain genes either by DNA blot or by PCR analyses (Tables IA, 1B, 1C). The Thy-1 B220 + thymoma cells also did not demonstrate significant TCR' or TCR rearrangement by Southern blot analysis ( Fig. 4A  indicates V H to DJ H assembly. The 737 bp -actin #agment was amplified along with immunoglobulin #agments to serve as an internal amplification control for quantitation. DNA #om normal liver and a v-abl transformed pre-B cell line (LPC 5-1) were mixed together in the indicated propoions and all three sequences were amplified in the same tube. The bands were visualized and quantitated by Phosphorimager. (B) The reaangement of V H to DJ H and of D to JH sequences was measured by normalizing the intensity of each amplified band to &e amplified actin product, and &is normalized signal was plotted against the expected percent loss of germline structure in ee dilution series.
The co,elation coefficients, h are indicated and show a linear co,elation between the measured loss of germline DNA with the expected degree of Igh reaangement. (C) A representative experiment shows that the magnetically sorted B220 and B220subpopulations #om thymoma 25.7 exhibit different degrees of Igh reaangement. Neither subpopulation showed @preciable loss of the V H to DJ H germ]ine region, but differed in &e degree of reaangement of D to JH. Lane 1: liver DNA, Lanes 2-3: C# and C#maow-derived A-MuLV pre-B cell lines (both subclones of LPC 5-9). Lanes 4-5:B220 and B220subpopulations sorted from thymoma 25.7. The C# and B220 phenotypes, as well as percent loss of V H to DJ H and D to JH germline structure are listed below the figure B220 + thymoma subpopulations that were examined, negligible rearrangement of TCR genes has occurred.
In contrast to the Thy-ll B220 + thymoma subpopulation, Thy-1 hi B220tumor cells ( Fig. 4A and 4B, lanes 2, 4, 6, and 9) and normal thymus (lane 14) showed extensive rearrangement of both TCR' and TCR[3 genes by DNA blot analysis. TCR' rearrangement was evident by 33-68% loss of germline bands ( Fig 4D, lanes 2, 4, 6, and 9) and by the appearance of rearranged bands (Fig 4A, lanes 2, 4, 6, and 9). TCR rearrangement was also apparent by the 54-59% loss of the CI31 germline band ( Fig. 4B and 4D, lanes 2, 4, 6, and 9). Since the TCR3 probe detects the 3' untranslated region of C31 and does not cross-hybridize with C[2, complete loss of the C131 germline sequences without the appearance of rearranged bands in these oligoclonal populations is likely due to rearrangement involving C12 (Clark et al., 1993). In contrast to this extensive rearrangement of TCR genes in the Thy-lhi B220-thymoma subpopulation, PCR analysis of Igh genes revealed minimal (15-36%) loss of D H to JH germline and negligible (<10%) loss of V H to DJ H germline sequences, which was confirmed by the DNA blots (Table IE). In the Thy-1 hi B220-thymomfi cells, this pattern of minimal lgh and extensive TCR/and 3 rearrangement is consistent with a T-cell lineage origin. Like the B220 + thymoma subpopulation, the B220 + peripheral lymphomas showed negligible rearrangement of TCR' and TCR[3 ( fig 4A and B, lanes 11 and 12). However, unlike the thymoma subpopulation, these B220 + lymphomas were entirely Thy-1-and demonstrated complete loss of Igh germline structure on both alleles by Southern blot (Table IG). This pattern is, as expected, consistent with a B-cell lineage origin of the A-MuLV lymphoma population. None of the tumor populations or cell lines that were examined showed rearrangement of Ig light chain genes by DNA blot analysis or by PCR analysis (Table I).
The concentration of A-MuLV can affect the transformation of different thymic targets The Thy-11 B220 + and Thy-1 hi B220tumor subpopulations that co-exist in the thymomas could arise either from the transformation of different thymic lymphocyte targets or by the divergent maturation of a transformed lymphocyte progenitor. We reasoned that if different thymic targets were transformed by A-MuLV, altering the virus titer might affect the selection of different targets for A-MuLV transformation. On the other hand, if both thymoma populations did arise from the same transformed target, then the virus titer would not affect the proportional representation of each population in individual thymomas.
A freshly prepared stock of A-MuLV (105 ffu) was either concentrated 10-fold or diluted 10-fold in culture media prior to intrathymic injection. The median latent period for tumor formation (35-38 days) was similar in mice injected with the different concentra-tiQns of A-MuLV (data not shown). B220 and Thy-1 expression were evaluated by two-color flow cytometry and the data are summarized in Fig. 5. All tumors induced with the diluted (0.1x) virus primarily consisted of Thy-11 B220 / cells and none of the thymomas in this group were predominately Thy-1 hi B220-. In contrast, 70% of the tumors induced with the 10x concentrated virus were almost entirely Thy-1 hi B220-, and an intermediate fraction (20%) of tumors from animals injected with the neat virus developed predominately Thy-1 hi B220thymomas. Although the Thy-1 hi B220thymomas were phenotypically indistinguishable from thymomas induced by Mo-MuLV alone (Clark et al., 1993), all tumors that developed in mice injected with the different A-MuLV concentrations expressed the P160 v-a/'l product (data not shown). In addition, all A-MuLV-induced tumors arose after a significantly shorter latent period (range 22-56 days, data not shown) than expected for thymomas induced by intrathymic injection with Mo-MuLV alone (range 81-157 days) (Clark et al., 1993).

Rare B220 + cells in normal thymi do not contain targets for in vitro transformation
In bone marrow a Thy-11 B220 + B-lymphoid progenitor population is highly enriched for A-MuLV transformation targets (Tidmarsh et al., 1989). Since this phenotype is similar to the A-MuLV-transformed thy- FIGURE 4 DNA blot analysis of TCR rearrangement. HindIII digested genomic DNA from B220 sorted thymomas (lanes 1-9) or from extrathoracic lymphomas (lanes 11,12) were examined for rearrangement of TCRy (A) and TCR# (B). A c-myc probe was used to quantitate the relative amount of DNA loaded in each lane (C). In order to measure the loss of germline structure, bands were quantitated by phosphorimaging and the ratio of the TCRy or TCR germline signal to c-myc in the tumor populations was compared to a similarly normalized control ratio from unrearranged liver DNA (D). The TCRy germline signal was measured by comparing the signal from individual J/1-3 germline bands to the c-myc control. Lanes and 2:B220 sorted fractions from thymoma 25.9. Lanes 3 and 4:B220 sorted fractions from thymoma 25.7. Lanes 5 and 6:B220 sorted fractions from thymoma 26.8. Lane 7:B220 sorted fraction from thymoma 25.1. Lane 8:B220 sorted fraction from thymoma 25.5. Lane 9: B220sorted cells from thymoma 26.7. Lanes 10 and 13: normal liver. Lanes 11 and 12: peripheral lymphomas from mice 21.3 and 22.3, respectively. Lane 14: normal thymus. The arrowheads point to unrearranged germline bands moma population described in this report, we examined normal adult thymi for a similar lymphoid progenitor target susceptible to in vitro transformation by A-MuLV. Flow analysis revealed that only 0.3 % of all thymocytes express B220. Therefore, the B220 + cells were first enriched approximately 100-fold over an anti-B220 magnetic column and then examined for co-expression of other surface antigens by flow cytometry. This revealed that B220 + thymocytes represent a heterogeneous population in which about 50% express surface IgM and IgK. About 40% of the thymic B220 + population were also Thy-1 l which resembled the Thy-ll B220 + thymoma subpopulations as well as the marrow-derived B-cell progeni-tors that are targets for A-MuLV transformation (Tidmarsh et al., 1989). For this reason, we tested the ability of A-MuLV to transform B220 + thymocytes in vitro. As shown in Table II, B220 + bone marrow cells were enriched in targets for in vitro transformation by A-MuLV. Conversely, in three different experiments, B220 + cells from normal thymi never showed in vitro transformation by A-MuLV, even when plated over a bone marrow stroma with exogenous IL-7, conditions that greatly enhanced transformation of bone marrow pre-B cells (data not shown). In order to test whether contaminating thymocytes in the sorted populations could actively inhibit in vitro transformation of a rare pro-B cell, thymocytes and normal bone marrow were admixed prior to infection. No inhibition of in vitro transformation was detected (data not shown). Thus, B220+thymic lymphocytes are distinct from B220 + bone marrow lymphocytes in their ability to be transformed in vitro by A-MuLV.

DISCUSSION B220 + thymoma cells resemble early lymphoid progenitors
We demonstrated that two distinct tumor subpopulations often co-exist in A-MuLV induced primary thymomas. The subpopulation that was B220was also uniformly Thy-1 hi and had extensively rearranged TCR, TCR7 and DJ H regions but not V H. This subpopulation, therefore, consists of A-MuLV transformed T-lymphoid cells, similar to A-MuLV-transformed thymic cell lines that have been described by others (Clark et al., 1993;Cook, 1982;Cook, 1985;Cook and Balaton, 1987;Holland et al., 1991;Risser et al., 1985;Scott et al., 1986;Heuze et al., 1992). The other major thymoma subpopulation was B220 + and expressed low levels of Thy-1. This subpopulation had not rearranged TCR genes, but had undergone variable rearrangement of Igh genes. This population is interesting since it contained a significant proportion of cells that apparently have not rearranged any antigen receptor genes and, therefore, cannot be assigned to either the B or T-lymphoid lineage.
The Thy-11 B220 + thymoma subpopulation exhibited a composite phenotype that also was consistent with immature lymphoid cells (CD43 + HSA + CD5-CD2 neg-l sIg-Ct-) but that was not specifically characteristic of either the B or T-lymphoid lineage.
Co-expression of Thy-ll B220 + CD43 + HSA + and negligible expression of CD2 is consistent with a pro-B cell phenotype (Hardy et al., 1989;Hardy, et al., 1991;Li et al., 1996;Nutt et al., 1997;Rolnick et al., 1996;Sen et al., 1990;Winkler et al., 1995;Yagita et al., 1989) and similar to the phenotype of A-MuLV transformed bone marrow pre-B cells (Rosenberg and Kincade, 1994). However, the fact that a significant proportion of these cells had not rearranged Igh genes is inconsistent with the observation that they express HSA which first appears on B lymphocyte progenitors in the process of rearranging the D to JH region (Allman et al., 1999;Nutt et al., 1997).  The composite phenotype of the Thy-11 B220 + thymoma subpopulation also is not inconsistent with the phenotype of primitive cortical T-cells. Even though the majority of thymocytes are Thy-lhi CD2 / CD5 / HSA 1 (Duplay et al., 1989), an early cortical pro-thymocyte subpopulation expresses a Thy-11 CD2-CD5-HSA hi phenotype (Duplay et al., 1989;Scollay and Shortman, 1985). As the cells mature, expression of CD2 and CD5 increase (Duplay et al., 1989;Sen et al., 1989) while HSA levels decrease (Scollay and Shortman, 1985;Crispe and Bevan, 1987). Additionally, while expression of B220 is more often associated with cells of the B-lymphoid lineage, there are several examples of B220 expression on immature T-cells. For instance, B220 / Thy-1 neg-hi pro-T cell lines transformed with v-abl (Heuze et al., 1992), v-myc (Brightman et al., 1988) and RadLV (Ho and O'Neill, 1993) have been reported. In addition, autoimmune-prone strains of mice often carry B220 / T-lymphocytes (Matsuzaki et al., 1992;Landolfi et al., 1993;Giese et al., 1994) which develop from thymic B220 / progenitor cells (Budd et al., 1985;Asensi et al., 1990). However, while these cells are usually found in autoimmune-prone mice and express CD5, the Thy-1 l B220 + thymoma subpopulation was uniformly CD5-. The B220 antigen was also reported to be expressed on a minor subpopulation of immature CD4-8double negative (DN) thymocytes from normal mice  as well as on DN thymocytes undergoing apoptosis (Mixter et al., 1999). Finally, rare B220 + lymphoid progenitor cell populations from the bone marrow (Onishi et al., 1993) and from fetal liver (Sagara et al., 1997) reportedly are able to repopulate the thymus (but see Morrison et al., 1994). Therefore, the fact that immature T-cells can sometimes express B220 makes the composite phenotype of the Thy-1 l B220 + thymoma population consistent with immature cells in the T as well as B-lymphoid lineages. We conclude that, based on surface antigen phenotype alone, the A-MuLV transformed Thy-11 B220 + thymoma cells are not easily assigned to either the B or T-lymphoid lineage. shown Overall the immature Thy-11 B220 + cells did not appear to have rearranged TCR,, TCR, or Ign genes; however, in some thymomas, minor subpopulations of Thy-11 B220 + cells appeared to have undergone VII to DJ H rearrangement and, therefore, probably represent transformed thymic pre-B cells similar to those described by others (Cook, 1982;Cook, 1985;Cook and Balaton, 1987;Holland et al., 1991;Kimoto et al., 1989;Risser et al., 1985;Scott et al., 1986).
More interesting are those Thy-11 B220 + thymoma populations that, based on the predictable events of Ig gene assembly (Schatz et al., 1992), contained a significant number of cells that had not rearranged Igh genes. For example, <30% of the Thy-1 l B220 + population in thymomas 26.9 and 21.2 showed loss of D to JH germline structure and no loss of V H to DJ H germline sequences (Table IA) and, therefore, appeared to be similar to A-MuLV transformed cell lines that have only rearranged DJ H (Ramakrishnan and Rosenberg, 1988;Wang and Rosenberg, 1993). This also means that either a maximum of 30% of the population has rearranged both DJ H alleles or, that up to 60% of the cells have only rearranged one DJ H allele. Therefore, we conclude that 40-70% of the Thy-1 l B220 + cells in these thymomas have not undergone D H to J rearrangement and may represent transformed undifferentiated lymphocytes. The Thy-1 B220 + cells in thymomas 23.3 and 23.9 showed a somewhat different pattern of Igh rearrangement (Table IC). In both thymomas, these cells apparently lost equal amounts of D to JH and V H to DJ H germline segments. Since V H rearrangement does not occur until D to JH assembly has occurred on both alleles (Schatz et al., 1992), it is reasonable to conclude that the cells that showed loss of Ig germline structure had completed VHDJH rearrangement on both alleles. It would follow that the remaining cells in these tumors maintain Igh genes in germline configuration. For instance, thymoma 23.3 lost about 30% and 40% of D to JH and VII to DJI_ germline sequences, respectively. Therefore, the remaining 70% of the cells in the Thy-l B220 + population in these tumors probably have not undergone Igh rearrangement and are undifferentiated lymphocytes. Thymomas 25.2 and 26.7 showed yet another pattern of Ig rearrangement, in which twice as much D H to JII germline structure was lost compared to V H to DJII germline structure (Table IB). Both thymomas showed about 40% loss of unrearranged D H to JH structure and 20% loss of V H to DJ H germline sequences. Since both DJ H alleles rearrange prior to V H to DJ H rearrangement, it is possible that up to 40% of the Thy-11 B220 + thymoma cells have rearranged D H to JH on both alleles and V H to DJ H on one allele. This would mean that the remaining 60% of the thymoma population have not rearranged Igh genes. However, in these tumors, it is also possible that up to 20% of the tumor population completed Vt to DJ H rearrangement on both alleles. In this case, then, DJII rearrangement would be found on another 20-40% of the population, depending on whether the cells rearranged one or both alleles. This means that a minimum of 40% of the cells in these thymoma populations appear to be undifferentiated lymphocytes that have not undergone Igh rearrangement.
Together, the quantitative PCR data along with the Southern blot data demonstrate that in some tumors as much as 70% of Thy-1 B220 + cells have not rearranged either Ig or TCR genes, and, therefore, resemble the earliest lymphoid progenitor. This progenitor-like subpopulation is distinct from most of the previously described A-MuLV-transformed thymoma cell lines, all of which have undergone significant rearrangement of their Ig and/or TCR genes (Clark et al., 1993;Cook, 1982;Cook, 1985;Cook and Balaton, 1987;Holland et al., 1991;Kimoto et al., 1989;Risser et al., 1985;Scott et al., 1986). Nevertheless, the Thy-l B220 + progenitor-like cells in the primary thymomas do share similarities with some other transformed lymphoid populations. For instance, these progenitor-like lymphocytes closely resemble thymomas that arise in bcl-2/myc transgenic mice (Strasser et al., 1990) and that are Thy-1 B220 + sIgand have not rearranged T or B lymphocyte antigen receptor genes. It is conceivable that A-MuLV and the bcl-2/myc transgenes may transform a similar lymphoid progenitor population. Also, Holland et al. (1991), described an A-MuLV-transformed thymoma cell line that was Thy-1 neg-l B220 + and that only had minimal Igh rearrangement within a subpopulation of the line. These cells were capable of rearranging TCRy genes after intrathymic injection and of continuing Igh rearrangement in vitro. It is an intriguing possibility that the Thy-1 lo B220 + progenitor-like cells found in our primary thymomas represent an A-MuLV-transformed thymic lymphocyte progenitor (Carlyle et al., 1997;Wu et al., 1991) that also may be capable of undergoing differentiation along the T and/or B-cell pathways. However, in light of the recent observation that commitment to the B lymphoid lineage may occur before D to JH recombination (Allman et al., 1999), it is possible that these progenitor-like thymoma cells may already be committed to either the T or B lymphoid lineage. Experiments are underway to attempt to resolve this issue.

Thymic targets for A-MuLV transformation
The progenitor-like thymoma population expressed a Thy-11 B220 / phenotype that is similar to bone marrow derived B-lymphoid progenitors that are enriched for A-MuLV transformation targets (Tidmarsh et al., 1989). This similarity raised the possibility that a B-cell progenitor in the adult thymus is a target for transformation by v-abl. However, unlike the bone marrow B220 / cells, thymic B220 / cells could not be transformed in vitro by A-MuLV. On one hand, it is possible that the thymocyte transformation target that gives rise to the Thy-11 B220 / progenitor-like A-MuLV transformants is not found within the normal B220 / thymocyte subpopulation even though 40% of these cells also expressed Thy-11. However, it is also possible that A-MuLV transformation of a thymic B220 / pro-lymphocyte can only occur within a thymic microenvironment, which would explain our inability to transform these cells in vitro. Either way, there appear to be fundamental differences between the transformation targets in the bone marrow vs the thymus that give rise to B220 / transformed cells.
The co-existence of A-MuLV-transformed pre-B and pre-T cells in primary thymomas raises the question of whether these different populations arise from the transformation of the same or different target cells. The presence of pre-T and pre-B cell transfor-mants in the same thymoma could feasibly arise from A-MuLV infection of separate T and B cell progenitors that may be present in the adult thymus; however, the surprising inability to transform thymic B cell progenitors in vitro is not consistent with this explanation. Alternatively, the coexisting pre-T and pre-B cell thymoma populations could feasibly arise from transformation of an undifferentiated lymphocyte progenitor that may undergo Tand B-lymphopoiesis after transformation by A-MuLV (Whitlock et al., 1983). However, it seems unlikely that the coexisting Thy-1 hi B220-and Thy-11 B220 / thymoma subpopulations arose from the transformation of a single target cell because intrathymic injection of different concentrations of A-MuLV induced thymomas with distinct phenotypes. If both thymoma populations arose from transformation of a single target cell, it is unlikely that the virus titer would affect their proportional representation in the tumors.
In summary, these observations suggest that a subset of the Thy-11 B220 / thymoma population represents an undifferentiated lymphoid progenitor that cannot be unequivocally assigned to either the T or B-lymphoid lineage. It will be of interest to identify the normal thymocyte target that gives rise to this transformed progenitor population and to assess its capacity for differentiation.

Retrovirus
A-MuLV (1-2 x 105 focus forming units per ml) was collected from clonal NIH 3T3 fibroblast cells infected with a helper-free A-MuLV and superinfected with Moloney murine leukemia virus (Mo-MuLV). To make a concentrated preparation of A-MuLV, viral supernatant was centrifuged over a 20% sucrose cushion, and the virion pellet was gently resuspended in 0.01 volumes of media (Wang et al., 1991). The virus was quantitated by slot blot hybridization which showed a 50-fold enrichment of viral RNA in the concentrated stock; however, quantitation of infectious virus by measuring v-abl oncoprotein expression in acutely infected NIH 3T3 fibroblasts revealed a 10-fold concentration of active virus.

Tumor cell preparation
Thymic lymphomas were induced in 4-5-week-old female BALB/c mice (Harlan-Sprague-Dawley, Indianapolis, IN) by intrathymic injection of A-MuLV as previously described (Clark et al., 1993). Thymomas arising in individual mice are identified according to the experiment and animal number. Thus, thymoma 25.2 identifies the second animal to develop thymic lymphoma in experiment number 25. Cell suspensions were made from thymomas or peripheral lymphomas and were either processed immediately for immunophenotype analyses or were cryopreserved. Upon thawing cryopreserved samples, dead cells were removed by agglutination in high phosphate buffered saline or by filtration through glass wool (Mishell et al., 1980).

Preparation of normal thymocytes
3-4 week-old female BALB/c mice were injected intraperitoneally with 0.1 ml India ink diluted 1:5 in PBS. After 20 minutes, mice were sacrificed by CO 2 inhalation and perfused by intracardial injection of 10ml ice cold staining buffer [HBSS (without CaCO 3, MgC12, MgSO 4, NaHCO 3, or phenol red) containing 0.5% charcoal treated, dialyzed BSA (fraction V, Sigma Chemical Company, St. Louis, MO) and 10 mM NAN3]. The thymus was removed and prepared by removing visible blood vessels and India ink stained parathymic lymph nodes.

Magnetic cell separation of thymic lymphocytes
For phenotypic and genotypic analyses of B220 + cells in tumor populations that also contained B220cells, B220 + cells were separated over a magnetic column as described by others (Miltenyi et al., 1990). Basically, cells were incubated with RA3-6B2-coupled superparamagnetic microbeads (anti-B220; Miltenyi Biotec Inc., Sunnyvale, CA) at 10 gl anti-B220 beads per 10 7 thymoma cells, or at 10 gl per 108 normal thymocytes. After 10 min., biotinylated RA3-6B2 (Pharmingen, San Diego, CA) (2.5 gg/ml) was added and the cells were incubated for an additional 20 minutes at 4C. This permitted indirect immunofluorescent detection of B220 expression of column sorted populations. Cells were washed in cold staining buffer and passed over a magnetized A2 separation column according to the manufacturer's protocol (Miltenyi Biotec Inc.). Cells that flowed through were collected as the B220fraction. To collect the B220 + fraction, the column was removed from the magnetic field and the adherent cells were washed from the column with cold staining buffer. The efficiency of separation was assessed by flow cytometry after staining with avidin conjugated fluorescein-isothiocyanate (FITC) (Becton-Dickinson, Mountain View, CA). From thymomas with mixed populations of B220and B220 + cells, this protocol routinely gave >95% B220 + cells from the column and <5% in the flow-through fraction. From normal thymi where B220 + cells represented <0.3% of the population, B220 + cells were enriched 100-fold over the column and further purified to homogeneity by FACS. Immunofluorescent cell staining and flow cytometry Immunofluorescent cell staining was performed as described before (Clark et al., 1993), except that cells were preincubated with FcBlock (Pharmingen) in order to prevent potential Fc receptor binding of monoclonal antibodies. Strategies for simultaneous analysis of multiple surface antigens were essentially as described by Hardy et al. (Hardy et al., 1991). After cells were flow sorted, an aliquot was reanalyzed by flow cytometry in order to confirm the sorting efficiency. PCR analysis was performed on sorted populations that were 95-98% pure.

Cell cycle analysis
Simultaneous staining of cell surface antigens and nuclear DNA was done exactly as described previously (Schmid et al., 1991). DNA blot analysis DNA blotting procedures and probes used to analyze tumor clonality and antigen receptor gene rearrangement have been described previously (Clark et al., 1993). The genomic 4.7 kb c-myc fragment from pmyc-26 contains c-myc exons 1, 2 and 3 and was kindly provided by Dr. Richard Schwartz (Michigan State University).

PCR quantitation of relative proviral content
In order to measure relative v-abl provirus content in different tumor populations, a 491 bp gag-abl fragment was amplified by PCR using primers specific for retroviral gag (5' TCC ACT ACC TCG CAG GCA TTC 3') and abl (5' CCT CCA CCC AAC TTG TGC TTC 3') sequences. Amplification of a 737 bp region of the ot-actin gene (Hardy et al., 1991) was done in the same tube to serve as an internal amplification control. For this PCR analysis, genomic DNA (25 ng) was first digested with EcoRI, then treated with 40 tg/ml RNAase (30 minutes at 37C). The DNA was then added to 100 tl reaction buffer [60 mM Tris-HC1 (pH 9.0), 15 mM (NH4)2SO4, 1.5mM MgC12] containing 1 tM of each primer and 0.5 U Taq polymerase (Promega, Madison, WI). Samples were heated to 80C before 50 tM dNTPs (Promega) were added. The amplification conditions were 95C for 1.0 minute, 63C for 0.5 minute, and 72C for 1.5 minutes for 20 cycles. Five tl of amplification products were electrophoresed through 2.5% agarose and blotted to Gene Screen Plus (NEN Research Products, Boston, MA). PCR amplification products from a v-abl transformed pre-B cell line (LPC 5-1) DNA were gel purified and used as probes. The intensities of the gag-abl and c-actin amplified signals were quantitated by Phosphorimager (Molecular Dynamics, Sunnyvale, CA).

PCR analysis of Ig rearrangement
In order to measure the extent of Ig gene assembly in different tumor populations, quantitative PCR using primers described by Hardy et al. (Hardy et al., 1991) was used to amplify Ig gene regions that would be deleted at different stages of rearrangement. D H to JH rearrangement was measured by loss of a 472 bp germline segment that is immediately 5' of JH 1. V H to DJ H rearrangement was measured by loss of a 1227 bp germline segment 5' of DFL16.1. V: to J: rearrangement was measured by loss of a 435 bp germline segment immediately 5' of J:l. PCR amplification of a 737 bp region of the ot-actin gene was done in the same tube with the Ig primers to serve as an internal amplification control as described above.
The parameters for PCR amplification of Ig genes were identical to those described above, except that 0.1 mg of EcoRI-digested genomic DNA was used as template, the reaction buffer contained 2.0 mM MgC12, and 23 cycles of amplification were performed. Unrearranged IgH and IgL products and the actin fragment amplified from liver DNA were used as probes. The Ig amplification signal from each tumor sample was normalized to the actin signal amplified in the same tube, and this value was compared to a similarly normalized unrearranged Ig amplified signal from liver in order to quantitate the loss of Ig germline structure. Each sample was amplified and analyzed 3-6 times.