INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Numerous hematopoietic disorders following both in vivo and in vitro murine leukemia virus (MuLV) infection have been described, providing key insights into oncogenesis and hematopoietic differentiation (39). One of the mechanisms by which MuLV infection results in hematopoietic cell transformation involves the presence of a rearranged and constitutively active oncogene of cellular origin in the MuLV genome. Additionally, some endogenous murine retroviral sequences function as oncogenes upon their transduction and rearrangement in an MuLV genome (24). All known MuLVs which harbor oncogenes are replication defective and require replication-competent helper MuLVs for replication and spreading.

A second oncogenic mechanism, first described for replication-competent viruses and designated insertional mutagenesis, depends on the random insertion of MuLV proviral DNA in the vicinity of cellular genes involved in proliferation. This insertion may then result in the aberrant expression of these genes (27, 53, 54). The lag time between retrovirus-mediated insertional mutagenesis and the development of disease is generally longer than that observed following the direct introduction of an MuLV harboring an oncogene. Although insertional mutagenesis has generally been described for replication-competent MuLVs lacking oncogenes, this mechanism has also been shown to participate in transformation following infection with replication-defective MuLV (34, 44).

Inoculation of mice with the replication-competent Moloney MuLV and the related Friend MuLV, neither of which harbors oncogenes, leads to the development of distinct types of leukemia. In both cases, insertional mutagenesis resulting in the rearrangement of cellular proto-oncogenes is associated with transformation. These rearrangements and the cell type involved in the leukemogenic process have been shown to depend on an array of parameters which includes the mouse strains used and the age of the animals at the time of inoculation as well as the conditions of inoculation (1, 8, 11, 21, 42, 45, 52, 56). Specifically, the prototypic virulent strains of Moloney MuLV induce a lymphoma characterized by gross thymic enlargement in all newborn mice from susceptible strains (17). In contrast, inoculation of the related prototypic Friend MuLV under the same conditions leads exclusively to the development of erythroleukemia (24). Rearrangements of several distinct cellular proto-oncogenes have been associated with the two types of mouse leukemia (3, 4, 11, 13, 23, 31, 43, 54). Furthermore, the type of leukemia has in both cases been exclusively mapped within the U3 region of the long terminal repeat (LTR), between the enhancer region and the MuLV promoter (5, 6, 20, 47, 51). Nevertheless, non-LTR sequences of Moloney MuLV and Friend MuLV have been shown to influence the latency of Moloney MuLV-induced promonocytic leukemia in an adult mouse model with inflammation (35).

In an attempt to evaluate the role of non-LTR intragenic sequences in replication and retroviral pathogenesis, we derived a series of Friend and Moloney MuLV mutants with mutations in the viral capsid structural gene which did not alter the potential coding ability of this gene. Strikingly, we found that one such mutant Moloney MuLV with synonymous differences (MSD1), in which the U3 enhancer region remained unchanged, exhibited unexpectedly broader oncogenic properties than those observed with the parental wild-type (WT) Moloney MuLV. The observation that clonal leukemias induced by this Moloney MuLV variant recruited thymic, erythroid, and myelomonocytic lineages revealed an unsuspected role for this region in the leukemic process.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Oligonucleotide-directed mutagenesis and DNA manipulation. Viral DNA samples of the prototypic Moloney MuLV, 8.2 (46), and Friend MuLV 57 (36) were prepared from the p8.2/B and p57/A plasmids containing a permuted version of the MuLV genomic DNA with a single copy of the LTR. These MuLV genomes were inserted at the HindIII or EcoRI sites, respectively, of two previously described pUC19-derived plasmids (47). MSD1, a Moloney MuLV mutant with synonymous differences in the gag-coding region, was obtained after oligonucleotide-directed mutagenesis with the forward oligonucleotide 5'-CAGGCAGGacGcAACCACCTAGTCCACTAT-3', spanning positions 1590 to 1619 in the capsid, and the reverse oligonucleotide 5'-AATGTGGTGGGTCCGTCCtgCgTTGGTGGA-3', spanning positions 1580 to 1609, with the MSD1 mutations indicated in lowercase letters. Mutagenesis was performed on a 2.5-kb SpeI-BclI gag-pol Moloney MuLV subclone. A 174-bp mutated fragment flanked by the viral XhoI and AflII sites was reinserted into a WT Moloney MuLV XhoI-BclI subclone, yielding the corresponding pBS11 plasmid. This plasmid contained the MSD1 point mutations at positions 1598, 1599, and 1601 of Moloney MuLV, with the numbering starting at the first nucleotide of R. This plasmid was derived to facilitate substitution of the minimal mutated fragment originating from the PCR within the previously described pMRU5G plasmid containing the NheI-SalI insert of WT Moloney MuLV in pBR327 (35). The XhoI-BclI insert of pBS11 was then substituted in pMRU5G DNA prepared on strains of Escherichia coli carrying the dam mutation, yielding the pMRU5G(MSD1) subclone. The pMSD1/B plasmid carrying the entire Moloney MuLV sequence with the MSD1 mutations was generated by ligation of the 4.1-kb and 6.8-kb NheI-SalI fragments of pMRU5G(MSD1) and p8.2/B, respectively. Introduction of the desired mutations was verified by sequencing of the entire pBS11 insert, and intermediate and final constructions were monitored by digestion of plasmid DNA with several discriminating restriction enzymes.

Tissue culture, transfection, and viral stocks. Mus dunni fibroblasts (28) used for transfections and preparation of viral stocks were cultivated in Dulbecco's modification of Eagle's medium (DMEM) complemented with glutamine (2 mM), penicillin (50 IU/ml), streptomycin (50 mg/ml), and 10% heat-inactivated fetal calf serum. In order to excise the single-LTR-permuted retroviral genome DNA, the p8.2/B and pMSD1/B plasmid DNAs were digested with HindIII and that of p57/A was digested with EcoRI. Viral insert DNA was purified by direct filtration of the agarose through a 0.45-µm-pore-diameter filter, as described previously (32), and submitted to phenol-chloroform extraction, precipitation, and resuspension in 0.1× TE buffer (10 mM Tris HCl, pH 7.5, and 0.1 mM EDTA) before transfection into M. dunni cells without prior ligation. Briefly, M. dunni fibroblasts were adjusted to a level of 2 × 10⁶ cells per dish in a 100-mm culture dish and transfected with 1 µg of DNA by the lipofectAmine (Gibco BRL) method following the manufacturer's recommendations. Transfections were monitored by focal immunofluorescence assay (FIA) (49) and assay of reverse transcriptase activity (19). Since mutations that hamper replication might have yielded recombinant or pseudorevertant variants (10, 33), transfection experiments that led to production of infectious viruses as rapid as that after transfection of WT virus were selected. Thus, upon transfection of the single-LTR-permuted form, cultures producing either WT Moloney MuLV or MSD1 MuLV within 20 days were selected. For viral stock preparation, fresh supernatants (prepared within 10 to 18 h) were removed from subconfluent infected cells, filtered through 0.45-µm-pore-diameter filters, and stored in aliquots at 80°C. Viral stocks were titrated by FIA as previously described, with monoclonal antibodies H48 for Friend MuLV and H538 or 83A25 (7, 9, 16) for WT Moloney MuLV and mutant MuLV strains. The titers of all MSD1 and WT Moloney MuLV viral stocks used in this study ranged from 2 × 10³ to 4 × 10⁴ and from 10³ to 10⁵ focus-forming units (FFU) per ml, respectively. No variations in the biological effects of either strain were observed within these titer ranges.

Immunoblot analysis for viral protein expression. NIH 3T3 cells were plated at 8 × 10⁵ cells per dish in a 100-mm tissue culture dish and infected with 1 ml of viral supernatant, adjusted to 3.8 × 10⁴ FFU, in 10 ml of complete DMEM. Four days later, cells were washed twice with 10 ml of ice-cold phosphate-buffered saline and collected in 1 ml of phosphate-buffered saline. Cells were then pelleted and resuspended in 1 ml of IPB (20 mM Tris HCl [pH 7.5], 150 mM NaCl, 1 mM EDTA, 0.1% sodium dodecyl sulfate [SDS], 0.5% sodium deoxycholate, 1% Triton X-100, and 0.02% sodium azide) plus 0.1 mM phenylmethylsulfonyl fluoride, followed by microcentrifugation to remove debris. Lysates were homogenized and saved. Cell samples were prepared for loading by boiling for 4 to 5 min following addition of an equal volume of 2× sample buffer (12.5 mM Tris HCl [pH 6.8], 2% SDS, 20% glycerol, 0.25% bromophenol blue, and 5% -mercaptoethanol). A portion (1/10) of each aliquot was subjected to SDS-polyacrylamide gel electrophoresis, and gels were electroblotted onto nitrocellulose filters. Gag proteins were detected by immunoblotting with the R187 rat anti-p30 monoclonal antibody (a kind gift from B. Chesebro), used at a 1:200 dilution, as the primary antibody and a peroxidase-conjugated rabbit anti-rat immunoglobulin antibody (dilution, 1:2,000) (DAKO) as the secondary antibody. Detection was performed by the enhanced chemiluminescence detection method.

In vitro and in vivo viral spreading assays. To evaluate the efficiency of in vitro viral production, 5 × 10⁴ M. dunni fibroblasts were plated in 60-mm tissue culture dishes and incubated overnight. The medium was removed and replaced by 4 ml of complete DMEM containing 8 µg of Polybrene (no. H-9268; Sigma)/ml and 0.5 ml of a previously titrated viral stock adjusted to a concentration of approximately 6 × 10² FFU/ml, in order to obtain approximately 300 infected foci per dish. After overnight incubation at 37°C, the supernatant was replaced with fresh medium, and 2 or 3 days later, when monolayers reached approximately 80 to 90% confluence, the supernatants were collected, filtered through 0.45-µm-pore-diameter filters, and stored in 1-ml aliquots at 80°C. The exact number of infected foci per dish was then determined by FIA, and the titer of the corresponding supernatant aliquot was also determined by FIA on M. dunni cells. Comparison of viral production was based on the number of FFU per milliliter obtained per virus-producing cell focus. Direct comparison of the numbers obtained with WT Moloney MuLV and its MSD1 synonymous mutant was facilitated by the fact that, under these conditions, the average sizes of cell foci following infection with the two strains were equivalent.

Origin, infection, and clinical evaluation of mice. All animal experimentation was conducted on Swiss mice (Centre d'élevage R. Janvier, Le Genest, France). Newborn mice were inoculated intraperitoneally at 1 to 2 days of age with 50 µl of freshly thawed viral stocks. Mice over 30 days of age were regularly examined for organ enlargement by palpation under anesthesia. Animals displaying gross organ enlargement and moribund animals were bled in order to determine hematocrits prior to sacrifice and autopsy. Blood samples (20 µl) were taken under anesthesia at the retro-orbital sinus with heparinized capillary tubes, and hematocrits, the volume of erythrocytes expressed as a percentage of blood volume, were determined after a 5-min centrifugation in a Sigma 112 centrifuge. Splenocyte suspensions were prepared from the enlarged spleens of sacrificed animals either for genomic DNA extraction as indicated below or for staining after fixation as previously described (7, 50). Nonspecific esterase activity was monitored with a commercial kit (no. 181-B; Sigma) based on detection of alpha-naphthyl butyrate esterase in the presence or absence of sodium fluoride. Myelomonocytic cells were detected with a commercial kit for Sudan black B staining (no. 380; Sigma).

DNA amplification and sequence analysis of in vivo spreading viruses. The presence of WT or mutated Moloney MuLV sequences in spleen tumors was assessed by DNA amplification performed on genomic DNA prepared from splenocytes of moribund mice inoculated as newborns. Briefly, 50,000 splenocytes were homogenized in 1 ml of TE buffer, pH 8.0. Ten volumes of total extracting buffer (10 mM Tris HCl [pH 7.4], 100 mM EDTA [pH 8.0], 1% SDS, and 20 mg of RNase A/ml) were added, and the homogenate was incubated for 2 h at 37°C with gentle rocking. Proteinase K was added to a final concentration of 200 mg/ml, and the sample was incubated for 2 h at 56°C. After one phenol extraction and two phenol-chloroform extractions, the DNA was ethanol precipitated in 2 M ammonium acetate and washed twice with 75% ethanol, and the pellet was dissolved in 1 ml of 1× TE buffer, pH 8.0.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Comparison of spreading abilities of WT Moloney MuLV and MSD1, a mutant with synonymous differences in the capsid region. Synonymous mutations were introduced within the gag coding sequences of Moloney MuLV in order to evaluate the cis effects of this region in viral replication and pathogenesis. One of the Moloney MuLV mutants with synonymous differences, MSD1, harbored three nucleotide differences, at positions 1598, 1599, and 1601 in the capsid, such that the corresponding glycine and arginine codons were maintained (Fig. 1). Viral spreading was readily observed upon transfection of cells with either WT or MSD1 mutant DNA, enabling us to generate viral stocks with similar titers. The expression and processing of capsid-containing proteins in cells infected by either WT Moloney MuLV or MSD1 mutant virus were identical in normalized infections, as assessed by immunoblotting of cell extracts with an anticapsid monoclonal antibody (Fig. 1). The presence and processing of capsid proteins in cell-free virions were also identical in Moloney MuLV and MSD1 (data not shown). However, a precise comparison of MSD1 and Moloney MuLV supernatant titers obtained from the same number of infected foci revealed that the infectious MSD1 virions were produced at 2- to 10-fold-lower levels (data not shown). Nevertheless, the kinetics of in vivo spreading of the MSD1 mutant was similar to that observed following inoculation with WT Moloney MuLV. Specifically, 3 weeks following inoculation with either virus, approximately 0.1% of splenocytes were identified as productively infected (10³ infectious centers per 10⁶ splenocytes). Thus, the introduced synonymous mutations did not affect Gag expression, and the WT Moloney MuLV and MSD1 mutant MuLV had indistinguishable in vivo spreading abilities.

View larger version (69K): [in this window] [in a new window]   FIG. 1.   A replication-competent mutant of Moloney MuLV (MSD1) with synonymous differences in the capsid-coding gene. (A) Schematic organization of the MuLV genome and DNA sequence of the mutated capsid region. The potential amino acid coding ability, common to WT Moloney MuLV and MSD1, is indicated in italics below the nucleotide sequences. Nucleotides are numbered starting from the first nucleotide of R in the 5' LTR. (B) NIH 3T3 cell extracts producing WT Moloney MuLV or the MSD1 mutant were immunoblotted with the anticapsid R187 monoclonal antibody (7). Also shown are mock-infected cell extracts, and positions of molecular weight markers are indicated. The band corresponding to the cleaved capsid p30 as well as those corresponding to the Pr65 Gag and Pr75 glyco-Gag precursors was detected at the same position and at similar levels in cells infected with either of the two viruses.

Distinct leukemogenic profiles following inoculation with WT Moloney MuLV and the MSD1 synonymous mutant. Following inoculation with either WT Moloney MuLV or the MSD1 mutant, 50% of animals died or developed leukemia by 3 months of age and over 80% were sick or dead between 4 and 5 months of age (Fig. 2). As expected, all animals inoculated with WT Moloney MuLV developed an enlarged thymus without evidence of severe anemia. However, while a high percentage of MSD1-inoculated mice also demonstrated this phenotype (22 of 39 animals), a significant proportion (14 of 39 animals) had no obvious thymic enlargement. This latter group presented with gross enlargement of the spleen or, in one case, isolated lymph node enlargement (Table 1). Furthermore, a distinctive severe anemia (hematocrit below 35%) was observed in 9 of 39 of these leukemias (23% of all leukemias). This marked anemia-inducing effect was never observed following inoculation with WT Moloney MuLV, which resulted in hematocrits ranging from 36 to 48% (mean ± standard error of the mean [SEM], 41% ± 3%). It was striking to note that some of these MSD1-inoculated animals presented with severe anemia (hematocrit range, 12 to 34%; mean ± SEM, 26% ± 3%) without evidence of thymic involvement, a profile resembling that in mice inoculated with the prototypic erythroleukemia-inducing Friend MuLV (Table 1). Nevertheless, this profile was observed in only 15% (6 of 39 animals) of the MSD1 leukemias versus in 100% of mice inoculated with Friend MuLV (hematocrit range, 19 to 33%; mean ± SEM, 25% ± 4%) (Table 1). Thus, the MSD1 mutant virus caused leukemias which were distinct from the prototypic thymomas and erythroleukemias exclusively induced by parental Moloney MuLV and the related Friend MuLV, respectively.

View larger version (16K): [in this window] [in a new window]   FIG. 2.   Latency of leukemia in mice inoculated as newborns with either WT Moloney MuLV or the MSD1 mutant. Newborn Swiss mice were inoculated intraperitoneally with 50 µl of either WT Moloney MuLV, MSD1 mutant, or Friend MuLV viral stocks, containing between 50 and 500 FFU each. Animals were regularly monitored for gross organ enlargement and hematocrit changes. The number of animals used in each series is indicated in parentheses. Latency is defined as the time at which 50% of the animals were either dead or presented with spleen, thymus, or lymph nodes grossly enlarged.

Phenotype of erythroid and myelomonocytic leukemias induced by the Moloney MuLV synonymous mutant, MSD1. Distinct phenotypic groups of leukemia can be further distinguished by identifying the phenotypes of the cells responsible for the gross spleen enlargement. For this purpose, splenocyte suspensions from control and leukemic animals were stained for nonspecific esterase activity, a marker of the erythroid lineage, and Sudan black-positive cytosolic grains, a marker of the myelomonocytic lineage. Leukemias were then characterized on the basis of esterase activity, Sudan black staining, type of organomegaly, and the presence or absence of severe late anemia. Diagnoses based on these four criteria are summarized in Table 2. The percentages of esterase and Sudan black-positive cells in grossly enlarged spleens obtained from mice with WT Moloney MuLV-induced thymoma were not increased, ranging from 6 to 24% (mean ± SEM, 14% ± 4%) and from <1 to 9% (mean ± SEM, 5% ± 3%), respectively. The corresponding numbers in age- and sex-matched controls were 8 to 24% (mean ± SEM, 17% ± 4%) and 1 to 7% (mean ± SEM, 4% ± 2%), respectively. Thus, as expected, the Moloney MuLV-inoculated leukemic animals, which developed exclusively a T-cell leukemia with thymoma, showed no detectable erythroid or myelomonocytic involvement (Table 3 and Fig. 3A). Additionally, animals inoculated with Friend MuLV demonstrated no increase in the numbers of Sudan black-positive spleen cells (range, <1 to 5%; mean ± SEM, 1.5% ± 1%), whereas esterase activity increased dramatically to 66 to 96% (mean ± SEM, 78% ± 6%). These latter data were in complete agreement with the prototypic erythroleukemogenic characteristics of Friend MuLV (Table 3 and Fig. 3E).

View larger version (45K): [in this window] [in a new window]   FIG. 3.   Staining for erythroid and myelomonocytic cells in smears from leukemic spleens. Suspensions of splenocytes were stained for nonspecific esterase activity (red cells in the upper panels) and Sudan black-positive cytosolic grains (examples shown with white arrowheads in the lower panels). These colorations allowed the identification of abnormal erythroid and myelomonocytic proliferations, respectively. (A) Prototypic WT Moloney MuLV-induced T lymphoma; (B) MSD1-induced pure T lymphoma; (C) MSD1-induced erythroleukemia; (D) MSD1-induced mixed erythroid and myelomonocytic leukemia; and (E) prototypic Friend MuLV-induced erythroleukemia.

Conservation of the capsid mutation and WT U3 enhancer sequence in MSD1-induced erythroid and myelomonocytic leukemias. Because of the broad leukemia phenotypes induced by the MSD1 mutant, it was important to determine whether the synonymous mutations in the gag region of Moloney MuLV were maintained in these various spleen tumors. For this purpose, we amplified and sequenced the Moloney MuLV gag region using tumor samples from unexpected MSD1-induced erythroid and myelomonocytic leukemias (Table 3). Unambiguous sequencing of a fragment of 263 nucleotides (bp 1527 to 1789), encompassing the capsid region, where the synonymous mutations were introduced (Fig. 1), revealed no detectable reversion or new mutations in these MSD1-induced erythroid and myelomonocytic tumors. Additionally, in spleens with WT Moloney MuLV-induced leukemias, no mutations were detected in this region. Finally, since the U3 enhancer region of the LTR is the only sequence known to directly influence the type of leukemia induced by Moloney MuLV in the newborn-mouse model, we also sequenced this region after amplification in tumor samples. We found that both the enhancer and the promoter regions in distinct MSD1 tumors were identical to those of WT Moloney MuLV. Therefore, neither systematic reversions of the synonymous mutations introduced in the capsid gene nor systematic mutations of the U3 enhancer or promoter regions occurred in the atypical tumors induced by the MSD1 mutant.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We describe a new mutation in the MuLV capsid-coding region that extended the leukemogenic properties of Moloney MuLV in mice inoculated as newborns. The mutation consisted of three synonymous nucleotide substitutions which maintained the coding abilities of the two codons of concern in the capsid gene (substitution of GGA CGC for GGT AGG, both Gly-Arg). Although slightly altered in its spreading abiliy in cell cultures, this synonymous mutant, MSD1, and WT Moloney MuLV had indistinguishable in vivo spreading abilities. Additionally, MSD1 caused leukemia in all inoculated animals with a latency which was similar to or shorter than that observed for WT Moloney MuLV.

The U3 region of the LTR has been shown to play a major role in both latency and leukemia cell type determination following inoculation of newborn mice with Moloney MuLV (5, 6, 18, 20, 21, 51). Since MSD1 clearly demonstrated unique leukemogenic properties, we assessed whether WT enhancer and promoter U3 sequences were maintained in the induced tumors. Importantly, we found that only WT U3 sequences were detected in all MSD1 spleen tumors as well as in WT Moloney MuLV-induced spleen tumors. Also, the introduced synonymous substitutions in the gag region were maintained in all types of tumors. The fact that both capsid and U3 sequences of the inoculated virus were conserved in the tumors argues strongly in favor of a direct cis effect of the capsid determinant in extending leukemia development to new cell types. Nevertheless, the Moloney MuLV background in which these mutations were introduced appears to have been essential to the extended leukemogenic properties of the MSD1 mutant, since the same mutations introduced in Friend MuLV did not modify its strict erythroleukemogenic properties (data not shown).

Although there are multiple mechanisms by which replication-competent MuLVs induce leukemias following insertional mutagenesis (27, 39, 54), no mechanism involving the coding regions of MuLV has been reported for leukemias of mice inoculated as newborns. Nevertheless, enhancer or other cis-acting effects involving an intragenic determinant in the MuLV genome cannot be excluded. Interestingly, a systematic scanning of the mutated capsid region for the presence of consensus sequences for potential binding of transcriptional factors (40) revealed that the mutations in MSD1 created a new Ets-1/p54 motif, CAGGACGC, originally absent from the WT Moloney MuLV sequence, CAGGTAGG. Since the Ets-1/p54 transcriptional factor has been described to be preferentially expressed in lymphoid T cells compared to erythroid or myelomonocytic cells (26), an association between this factor and the broader leukemogenic phenotype of the MSD1 mutant remains elusive. However, it remains possible that the interactions of novel transcription factors with this mutated region play a role in the ability of MSD1 to induce leukemia in a broader panel of cell types.

An alternative insertional mutational mechanism by which MSD1 may exert its effects is the posttranscriptional modulation of proto-oncogene mRNA stability, splicing, or transport. This type of modification is more likely to be affected by a cis-acting intragenic determinant that might be provided by the MSD1 point mutations. In this regard, it is interesting to note that an intragenic RNA transport element upstream of the capsid gene in Moloney MuLV has been described (25). Also, the mutation we introduced eliminated the consensus for a potential 5' splice site used in c-myb rearrangements. These c-myb rearrangements precede Moloney MuLV-induced late promonocytic leukemias in adult mice with inflammation (35, 45). Here, we determined that alteration of this cryptic splice site in the MSD1 mutant favored the development of both erythroid and myelomonocytic leukemias in the newborn mouse model, with a greater susceptibility in the former cell type. Therefore, either alternative splicing was not associated with the change in the leukemic phenotype induced by MSD1 or the synonymous mutations introduced in MSD1 allowed additional and/or new modifications of cellular proto-oncogene mRNA. Interestingly, proviral integration sites initially shown to be preferentially associated with MuLV erythroleukemias (2, 34, 37, 43) are also rearranged in Graffi MuLV-induced myelomonocytic leukemias (12). The association between altered expression of a specific oncogene and transformation of a specific cell type remains unclear. Elucidating the precise relationship between the MSD1 gag mutation, the wider leukemia spectrum, and the proviral integration sites involved in the wide range of MSD1-induced tumors will help to elucidate this point. In this regard, it is interesting to note that MSD1-induced myelomonocytic and/or erythroid proliferations which presented without detectable thymic involvement were indeed clonal leukemias, as demonstrated by a discrete pattern of Moloney MuLV provirus integration (data not shown).

One of the major events that occurs upon infection of newborn mice with all known replication-competent MuLVs is the production of new recombinant MuLVs. These recombinants result from the substitution of mouse endogenous retroviral envelope genes for the parental ecotropic MuLV and are designated as polytropic or mink cell focus-inducing (MCF) viruses. MCF viruses recognize a receptor that is distinct from that used by the parental ecotropic MuLV (41). Additionally, recombinations may include substitutions of endogenous gag sequences for that of the parental MuLV (15). Leukemias developing following insertional mutagenesis of replication-competent MuLV can be associated either with the original ecotropic inoculum or with polytropic MCF viruses (39). It is therefore important to note that the prototypic thymoma-inducing Moloney MuLV and the erythroleukemogenic Friend MuLV differ in the endogenous envelope loci that are used to generate in vivo spreading of MCF viruses (14, 30). Genetic mapping of these differences includes gag sequences at the 3' end of the capsid gene, downstream of the mutations introduced in MSD1 (30). Therefore, it remains conceivable that the mutations present in MSD1 modulate the types of MCF recombinant MuLVs that are generated with new pseudotyping envelopes, allowing extended cell lineages to be targeted (29).

Unlike the generation of replication-competent MCF viruses, which is systematic and profuse and occurs very early in susceptible mouse strains models (48), the generation of replication-defective viruses harboring oncogenes is rarely observed. Interestingly, though, in one such case where recombination of Friend MuLV resulted in the generation of a defective myeloproliferative virus (38), the spectrum of leukemias that resulted evokes that observed with MSD1 (55). It is therefore tempting to speculate that the point mutations in MSD1 might be associated with an increase in recombination events yielding new oncogene-harboring MuLVs. Further physiopathological and molecular analyses of viral isolates obtained from the distinct MSD1-induced leukemias should help to elucidate these issues and may lead to the identification of new actors involved in hematopoietic differentiation and oncogenesis.