Novel Insights into the Pathogenesis of the Graffi
http://www.100md.com
病菌学杂志 2006年第8期
Laboratoire de Biologie Moleculaire, Departement des Sciences Biologiques, Universite du Quebec à Montreal, Quebec, Canada
Institut de Recherche en Immunologie et Cancer, Universite de Montreal, Quebec, Canada
ABSTRACT
The Graffi murine leukemia virus (MuLV) was isolated in 1954 by Arnold Graffi, who characterized it as a myeloid leukemia-inducing retrovirus. He and his team, however, soon observed the intriguing phenomenon of hematological diversification, which corresponded to a decrease of myeloid leukemias and an increase of other types of leukemias. Recently, we derived two different molecular clones corresponding to ecotropic nondefective genomes that were named GV-1.2 and GV-1.4. The induced leukemias were classified as myeloid based on morphological analysis of blood smears. In this study, we further characterized the two variants of the Graffi murine retrovirus, GV-1.2 and GV-1.4, in three different strains of mice. We show that the Graffi MuLV is a multipotent retrovirus capable of inducing both lymphoid (T- and B-cell) and nonlymphoid (myeloid, erythroid, megakaryocytic) leukemia. Many of these are very complex with concomitant expression of different hematopoietic lineages. Interestingly, a high percentage of megakaryocytic leukemias, a type of leukemia rarely observed with MuLVs, arise in the FVB/n strain of mice. The genetic backgrounds of the different strains of mice influence greatly the results. Furthermore, the enhancer region, different for GV-1.2 and GV-1.4, plays a pivotal role in the disease specificity: GV-1.2 induces more lymphoid leukemias, and GV-1.4 induces more nonlymphoid ones.
INTRODUCTION
About 50 years ago Arnold Graffi, at the Cancer Research Institute in Berlin, Germany, isolated a new retrovirus capable of inducing leukemia in a specific strain of mice, named Agnes-Bluhm, bred in his laboratory. Graffi (18) induced a large proportion of myeloid leukemias with a very high incidence of chloroleukemias (70%), characterized by a greenish coloration of the lymph nodes, and he classified the virus as a myeloid leukemia-inducing retrovirus.
Working on better characterizing the pathogenesis of the virus, Graffi and his team were soon confronted with complex results. They observed the intriguing phenomenon of hematological diversification, which corresponded to a decrease in the percentage of chloroleukemias and an increase in other types of leukemias (11, 12, 19). This virus could induce multiple kinds of myeloid leukemias, from immature to differentiated forms, but also reticular, lymphoid, and erythroid leukemias and mixed forms (based on the classification at this time).
It is very likely that Graffi was working at that time with a viral mixture. In 1993, Ru et al. (48) cloned two nondefective ecotropic retroviral genomes from NIH-3T3 cell lines chronically infected with the original Graffi acellular extract (a gift of Nathalie Teich). These two viral variants were called GV-1.2 and GV-1.4. The genomic restriction map and the sequences of their long terminal repeats (LTRs) (GenBank accession numbers L14415 and L14416 for GV-1.2 and GV-1.4, respectively) showed that they differ in the U3 region, one 60-bp segment being duplicated in GV-1.2 and not in GV-1.4. GV-1.2 induces disease with a shorter latency period. Based mainly on morphological analysis of blood smears, the leukemias induced with GV-1.2 and GV-1.4 were described as mainly myeloid/granulocytic when injected into newborn BALB/c and NFS mice (48). In several tumors, we observed myeloid characteristics (myeloperoxidase [MPO] staining) together with rearrangements of the germinal configuration of the T-cell receptor (TCR) gene and/or of the immunoglobulin heavy chain gene, specific for T-cell and B-cell lineages, respectively. This phenomenon was called lineage infidelity.
In this report, we extended our characterization of the Graffi retrovirus. GV-1.2 and GV-1.4 were inoculated into three strains of mice (BALB/c, NFS, and FVB/n), and more than 100 leukemias were analyzed by flow cytometry with several different hematopoietic markers combined with molecular biology tools.
The complete genome sequences of the two variants were determined and used to complement the immunophenotyping analysis. This study reveals that the Graffi murine leukemia virus (MuLV) is a complex virus capable to induce a large spectrum of leukemias. It replicates efficiently in several cell types, as already suggested in studies of the LTR (3, 4). It is multipotent and induces both lymphoid and nonlymphoid leukemia, including megakaryocytic leukemias, which are quite rare in MuLV-induced pathologies.
MATERIALS AND METHODS
Viruses, inoculation of mice, and tissue collection. Viral stocks of GV-1.2 and GV-1.4 were made from cell culture supernatants derived from chronically infected NIH-3T3 cells. The culture supernatant of the infected cells was collected, centrifuged, aliquoted, and kept at –80°C. Newborn (<24 h-old) mice of strains BALB/c, NFS, and FVB/n were inoculated intraperitoneally with 0.1 ml of filtered virus. The viral titers were 3.106 PFU for GV-1.2 and 1.106 PFU for GV-1.4. The mice were checked routinely for clinical signs of disease (loss of weight, anemia, enlarged spleen). Moribund mice were sacrificed, and several organs (thymus, lymph nodes, spleen, liver, kidneys, brain) were frozen on dry ice and stored at –80°C. A sterile cell suspension from the spleen, thymus, and lymph nodes was prepared by gently teasing the organs apart in cold RPMI cell culture medium and by passing them through nylon mesh. Viable cells were used for flow cytometry analyses or stored at –80°C in 90% fetal calf serum and 10% dimethyl sulfoxide. Bone marrow was recovered by flushing cold RPMI cell culture medium through the femur with a 26G3/8 needle. Blood and spleen samples were incubated in hypotonic ammonium chloride solution to eliminate the erythrocytes.
Immunophenotyping. 106 cells were suspended in 100 μl of standard phosphate-buffered saline buffer. For detection of surface antigens, the antibodies (quantities calculated from a titration experiment) were added, and the cells were incubated for 30 min in the dark. The cells were then washed with cold phosphate-buffered saline and resuspended in 500 μl of a standard isotonic solution (Hematal; Fisher). The antibodies (BD Pharmingen, Mississauga, Canada) used were as follows: fluorescein isothiocyanate (FITC)-conjugated rat anti-mouse CD4, FITC anti-CD3, phycoerythrin (PE)-conjugated anti-CD90, FITC anti-CD45R/B220, FITC anti-CD11b (Mac-1, M1/70), FITC anti-CD71 (transferrin receptor), FITC anti-CD41(glycoprotein IIB), FITC anti-CD34, FITC anti-CD117 (cKit, stem cell factor ligand), FITC anti-Sca1, PE anti-CD8a, PE anti-Gr-1 (Ly6G/Ly6C, RB6-8C5), and PE anti-TER119 (glycophorin A). Isotype controls (BD Pharmingen) were included for each antibody used. Flow cytometric analyses were done with a fluorescence-activated cell sorter scanner (FACScan; Becton Dickinson), and the data were processed with the WinMDI software. CD3 together with CD90, CD4 together with CD8, Mac-1 together with Gr-1, and Ter119 together with CD71 were used in double-staining combinations.
RNA extraction and Northern blotting. Total RNA samples were extracted from the tissues using the TRIzol reagent kit (Invitrogen, Burlington, Canada) and stored at –80°C. Fifteen micrograms of total RNA samples were denatured with formamide and separated on a 1% agarose gel containing formaldehyde. The RNA was transferred to a nylon membrane, UV fixed, and hybridized with DNA probes labeled by the random primer extension method using oligohexamers (GE Healthcare, Baie d'Urfe, Canada), and approximately 2 x 106 cpm/ml was used. The hybridization process was done as described previously (50). The cDNA fragments used as probes were as follows: myeloperoxidase (S. Gisselbrecht), PU.1 (F. Moreau-Gachelin), EpoR (H. Lodish), Fli-1 (A. Bernstein), FOG-1 (S. Orkin), c-mpl, GpIIb, and PF4 (F. Morle). The expression of beta-actin was used as a control for equal RNA loading and to normalize relative expression data.
DNA extraction and Southern blotting. The genomic DNA was extracted from normal and leukemic tissues by standard procedures as described previously (50). The DNA was digested with appropriate restriction enzymes, separated on a 0.8% agarose gel, and transferred to nylon membranes (Ambion) for hybridization. A 600-bp TCR probe (kindly provided by H. Fan) and a 600-bp probe corresponding to the Jh region (provided by F. Alt) were used to detect T-cell receptor and heavy chain immunoglobulin gene rearrangements, respectively.
PCR amplification of the proviral LTR. PCR amplification of the proviral LTR region was performed by using genomic tumor DNAs with the forward primer 5'-CCCCACCATAAGGCTTAGCAAGCTAG-3' and the reverse primer 5'-TAGTTTCAAATGAGGCGCAAG-3' and using the PCR CORE kit (QIAGEN, Mississauga, Canada). Each reaction product was amplified by PCR as follows: 1 cycle at 95°C for 5 min; 30 cycles at 94°C for 1 min, 61°C for 1 min, 72°C for 1 min; and 1 cycle at 72°C for 5 min.
RESULTS
Disease latency. Neonate mice of three different strains (BALB/c, NFS, and FVB/n) were inoculated with similar doses of the two Graffi MuLV variants GV-1.2 and GV-1.4 (see Materials and Methods). Leukemias appeared with an incidence of 99%. About 1% of the mice suffered from hindlimb paralysis and were not included in the study. Figure 1 shows the cumulative survival curves in the three different stains of mice inoculated with GV-1.4 (Fig. 1A) and GV-1.2 (Fig. 1B).
After the inoculation of GV-1.2 (Fig. 1B), 50% of the mice showed signs of disease at day 68 for the FVB/n group, day 79 for the NFS group, and day 81 for the BALB/c group. After the inoculation of GV-1.4 (Fig. 1A), 50% of the leukemias were detected after 139, 158, and 181 days in FVB/n, NFS, and BALB/c mice, respectively. Thus, GV-1.2, which contains two direct repeats in the LTR, induces leukemias with a shorter latency than GV-1.4, which contains only one direct repeat as previously described (48). Uncorrelated with the viral variant inoculated, the latency period is also significantly dependent upon the mouse strain (as measured by a log rank test; P < 0.05), reflecting the effect of the genetic background on the disease development. Leukemias developed more quickly in FVB/n mice than in NFS and BALB/c mice, the latter strain showing the longest latency period.
Gross pathology of leukemic mice. During the life span of the inoculated mice, no sign of pathology was detected until a few days before death. According to the Bethesda proposals for classification of neoplasms (28, 37), two groups of mice were obviously distinguishable, one presenting signs of lymphoid leukemia (thymoma and/or lymph node enlargement, no anemia) and one presenting signs of nonlymphoid leukemia (severe anemia, absence of lymph node or thymoma). The gross pathology of the leukemic mice is presented in Table 1. From these results, two major conclusions can be drawn. First, GV-1.2 induces more lymphoid leukemias than GV-1.4. Second, the genetics of the mice greatly influence the results: BALB/c mice are more susceptible to lymphoid leukemias, whereas FVB/n mice produce preferentially nonlymphoid leukemias marked by severe anemia. Eighteen percent of the mice showed signs of both lymphoid and nonlymphoid leukemias. Their mixed phenotypes were further elucidated by flow cytometry and by the analysis of the expression of some lineage-specific genes (see below). The results reflect the complexity of the leukemias induced by the Graffi retrovirus.
Flow cytometry. The immunophenotype of the leukemic cells present in the different hematopoietic organs of each mouse was analyzed using flow cytometry. The leukemic population was identified according to the enlarged size of the cells and their high proportion compared to that of normal control mice.
The results shown in Table 1 and Fig. 2A reveal that the Graffi murine retrovirus is able to induce leukemias arising from all of the hematopoietic lineages: the lymphoid lineage (B and T cells), the myeloid lineage, and the erythroid and megakaryocytic lineages. Among the 108 mice analyzed, 53 leukemias of lymphoid origin were detected; among these were 35 T-cell and 18 B-cell leukemias. The other 55 leukemias had nonlymphoid characteristics and included 9 myeloid, 15 megakaryocytic, and 31 erythroid leukemias.
In BALB/c mice, GV-1.2 induced almost exclusively T-cell leukemias (91%), whereas GV-1.4 induced also B-cell (28%), myeloid (15%), and erythroid (10%) leukemias. Similar results were obtained with the NFS and FVB/n strains, although the percentage of T-cell leukemias was reduced in these strains for both variants. The broadest spectrum of leukemias is induced with the NFS strain. These results correlate well with the observations of gross pathology in mice. They confirm that the two Graffi variants induced different patterns of leukemia and that the genetics of the mice influence the results.
The immunophenotypic analyses also confirmed the existence of complex phenotypes, either mixed or biphenotypic (28, 37). Mixed leukemias (two different leukemic populations in the organism) were exclusively erythroleukemias combined with T-cell leukemias. Biphenotypic leukemias (a unique leukemic population) were essentially lymphoid, expressing specific markers of the myeloid lineage.
Global results of the induced leukemias. To complete and confirm the gross pathology and the immunophenotypic analyses, molecular biology tools were applied to RNA and DNA isolated from each tumor (Fig. 3 and 4). For each leukemia, the flow cytometry results and the molecular characterization of the tumor RNA and DNA were compared. These global results for each type of leukemia are described below.
T-cell leukemia. The T-cell leukemias induced by the Graffi MuLV are characterized by the presence of an enlarged thymus and spleen, often accompanied by enlarged lymph nodes, with no anemia. They were identified based on four specific markers: CD3, CD90 (Thy1.2), CD4, and CD8a. All these surface molecules are expressed by thymocytes and mature lymphocytes (9, 31). A total of 31 T-cell leukemias were induced and analyzed. In BALB/c mice, GV-1.2 generated more than 90% of T-cell leukemias (Table 1), but this incidence was largely reduced in NFS (48%) and even more in FVB/n (28%) mice. For each mouse strain tested, GV-1.4 corresponded to an incidence of T-cell leukemias lower than that induced with GV-1.2 (Table 1).
Five groups of leukemias could be distinguished based on the expression of CD4 and CD8 markers: CD4– CD8–, CD4+ CD8+, CD4+ CD8lo, CD4+ CD8–, and CD4– CD8+. The two predominant forms were CD4+ CD8– and CD4+ CD8+, representing 12 and 10 cases, respectively, (Fig. 2A).
Rearrangements of the TCR gene were analyzed by Southern blotting (Fig. 3A and Table 2). Fifteen of the 31 T-cell leukemias analyzed had rearrangements of the TCR locus. Rearrangement of the TCR chain is initiated at the double-negative stage, and thus, we expected that the more immature phenotypes, CD4– CD8– (double-negative stage) and CD4+ CD8+ (double-positive stage) would be those that retained a germ line configuration. Three out of four CD4– CD8– and four out of seven CD4+ CD8+ leukemias were indeed unrearranged at the TCR locus. But in fact, 50% of the more-mature-phenotype (CD4+ CD8–) leukemias had also a TCR locus in germinal configuration. Thus, the TCR locus rearrangement and the several phenotypes observed suggest that the retrovirus does not homogeneously block the differentiation pathways of the leukemic cells. No TCR rearrangement was found in myeloid or in nonlymphoid leukemias (Fig. 3A and data not shown).
Recombination of the JH locus, which is specific to the B-cell lineage, was also tested, and atypically, we found that some T-cell leukemias also had rearrangements of this locus (Table 2). No resting B cells (less than 5%) were detected by flow cytometry in the spleen of these animals, excluding the possibility that the rearrangements observed corresponded to normal B cells (Fig. 2B, panels a and b). More curiously, other T-cell leukemias were biphenotypic and expressed the myeloid marker CD11b. This occurred in five cases, two with a CD4+ CD8– phenotype and three with a CD4– CD8– phenotype. Such a leukemia is illustrated on Fig. 2B, panel c. An aberrant abundant expression of MPO was detected in 10 tumors that were not biphenotypic (Fig. 4A, lane 10; Table 2; also data not shown), although no myeloid cells were detected by flow cytometry in the spleen of these animals. Some leukemias (7) expressed the erythroid-specific FOG gene (Fig. 4B, lane 26; also data not shown), but all of them were mixed T-cell/erythroid leukemias arising predominantly in the spleen.
B-cell leukemia. The B-cell leukemias were preferentially induced by GV-1.4 in NFS and FVB/n mice and were particularly recognizable by the presence of enormous lymph nodes and enlarged spleen but normal thymus and no anemia. They were classified as B-cell leukemias based on the presence of the surface antigen B220/CD45R, which is present from the pro-B-cell stage through the mature and activated B-cell stages, and the absence of other restricted specific lineage markers (2, 23). Some leukemias expressed progenitor markers at variable levels, and thus, six leukemias expressed Sca-1 and two expressed c-Kit (CD117). Rearrangements of the JH locus which occur at the pro-B-cell stage (2) were also analyzed. Table 2 summarizes both the immunophenotype and the molecular characteristics of the 15 B-cell leukemias. Figure 3B shows the Southern blot analysis of the heavy chain immunoglobulin (JH) gene rearrangement for the B-cell leukemias in comparison with other types of leukemias and controls. Twelve out of the 15 B-cell leukemias have rearrangements of JH (Fig. 3B, lanes 1 to 6, 9 to 10, 12, and 14 to 16). Thus, the majority of the B220+ leukemias had rearrangements of JH except for those induced in NFS mice by GV-1.2 (Table 2).
A number of these B-cell leukemias showed a complex phenotype. First and surprisingly, 8 of the 15 B-cell leukemias expressed very high levels of MPO RNA (Fig. 4A), although no myeloid cells were detected by flow cytometry as was the case for some T- cell leukemias. Second, two of them were biphenotypic since they were B220+, had rearrangements of JH, and expressed the myeloid marker CD11b, although only one expressed MPO. Surprisingly, none of the B-cell leukemias showed TCR rearrangement. Thus, we presume that the Graffi targeted B-cell progenitors are more committed than the targeted T cells. JH and TCR rearrangements may also be regulated by two separate mechanisms implicating different genes.
Myeloid leukemia. Myeloid leukemia was recognizable by an enlarged spleen, mild anemia, slightly enlarged lymph nodes and no thymus enlargement. To immunophenotype these leukemias, we used two myeloid-specific antibodies: CD11b, which is expressed on both granulocytes and monocytes (25, 29), and Gr-1, more characteristic of the granulocyte lineage (13, 29). Double-positive CD11b+ Gr-1+ cells represent myeloid precursors largely found in the bone marrow of normal mice. Nine cases of myeloid leukemia were observed. Five of these were induced in NFS mice with GV-1.4, and four were induced in BALB/c mice (three with GV-1.4 and one with GV-1.2). Three phenotypes were obtained: CD11b+ Gr-1– (five cases), CD11b– G-1+ (one case), CD11b+ Gr-1+ (three cases).
We also tested the expression of MPO, which is synthesized during the promyelocytic stage of myeloid differentiation (46). Three CD11b+ Gr-1+, one CD11b+ Gr-1–, and the unique CD11b– Gr-1+ tumor expressed very high levels of MPO (Fig. 4A, lanes 14 and 15; also data not shown). The other myeloid leukemias (mainly the Gr-1– phenotype) did not express MPO at detectable levels. Surprisingly, MPO was not only expressed in the spleens of the animals suffering from myeloid leukemias. It was detected abundantly in many other leukemia cases, as described above for the T-cell and B-cell leukemias and shown on Fig. 4A. In total, 10 T-cell, 8 B-cell, 2 erythroid, and 1 megakaryocytic leukemia showed a high expression of this specific myeloid gene even though no CD11b- or Gr-1-positive cells were detected in the spleens of these mice by flow cytometry. Interestingly, in contrast with the reports of Graffi (18), we induced only one chloroleukemia (with GV-1.4 in an NFS mouse) in this study.
Erythroid and megakaryocytic leukemia. These two types of leukemias induce severe anemia (hematocrit between 10% and 30%), an enlarged spleen, and often an infiltrated liver. The Ter119 (glycophorin A) and CD71 (transferrin receptor) cell markers were used to immunophenotype the erythroid leukemias. Ter119 and CD71 are expressed from the proerythroblast through erythrocyte stages (27, 33). Megakaryocytic leukemias were identified with the use of CD41 (glycoprotein IIb), which is specific to platelets, megakaryocytes, and early hematopoietic progenitors (57). The megakaryocytic leukemias analyzed in this study also expressed high levels of cKit (Fig. 2A, panel e) and were negative for Sca-1.
These two types of nonlymphoid leukemias developed primarily in FVB/n mice (Table 1). In this strain, we induced 72% and 36% of Ter119+ leukemias with GV-1.2 and GV-1.4, respectively, and 48% of CD41+ leukemias with GV-1.4. Complex (mixed or biphenotypic) erythromegakaryocytic leukemias were also induced, including two with GV-1.2 and two with GV-1.4. Some mixed erythroid and T-cell leukemias also developed (as described above).
Since megakaryocytic leukemias are rarely observed in murine models of retrovirus-induced leukemias and since CD41 is also expressed at the surface of multipotent stem cells, the expression of several genes highly expressed in the megakaryocytic lineage (43, 56) was analyzed by Northern blotting (Fig. 4B). As expected, the results indicate that in all the erythroid and megakaryocytic leukemias, Fog1 is expressed highly and at a much higher level than in the other types of leukemia and controls. Fli1 also is abundantly expressed in erythroid and megakaryocytic leukemias compared to the other types of leukemias and controls (Fig. 4B). These two genes cannot be used to distinguish the erythroid from the megakaryocytic tumors. Interestingly, the tumors that expressed very high levels of Fli1 also harbored a retroviral integration in the locus (not shown). GpIIb (CD41) was highly expressed in all the megakaryocytic leukemias and in mixed erythroid/megakaryocytic leukemias (Fig. 4B) in good correlation with their CD41+ immunophenotype. A few erythroid leukemias also expressed high levels of GpIIb mRNA although they were CD41–, possibly due to a poorer maturation of the protein in erythroid cells. PF4 and c-mpl are both considered to be specific to the megakaryocytic lineage (21, 62). In our study, c-mpl was found to be highly expressed in the majority of the megakaryocytic leukemias, whereas PF4 was expressed in only one case (Fig. 4B). Surprisingly, few erythroid leukemias expressed these genes (Fig. 4B). The expression of these genes in both types of leukemias was often correlated with a strong expression of GpIIb (Fig. 4B). Thus, this study enables us to conclude that the CD41/cKit leukemic cells derive from the megakaryocytic lineage. Furthermore, it shows that the erythroid and megakaryocytic leukemias induced by Graffi are very closely related.
Integrity of the U3 enhancer region. The U3 region of the proviruses of mature tumors was analyzed to detect possible rearrangements or duplications in the enhancer region, as it was found in previous studies with other MuLVs and especially with SRS-19.6 (20).
We PCR amplified the proviral U3 enhancer region from the genomic DNA of infiltrated spleens of leukemic mice. GV-1.2 has a 60-bp duplication of the enhancer region absent in GV-1.4, and therefore the number of duplication can be deduced from the fragment's size: 351 bp for GV-1.2 and 291 bp for GV-1.4 (Fig. 5). We hybridized the PCR products with a U3-specific probe to confirm the specificity of the amplifications (data not shown). The results indicate that the Graffi MuLV enhancer is not stable and acquired modifications during the replication cycles. As shown on Fig. 5, several fragments of different size are detected for many tumors. For each tumor, the most abundant fragment generally corresponds to the input viral variant (GV-1.2 and GV-1.4), and the additional fragments correspond to altered proviruses. The smallest fragment corresponded to the intact GV-1.4 form, which is therefore the smallest enhancer region necessary for efficient replication.
In the lymphoid leukemias, some fragments were of a size larger than that expected from the input retrovirus (Fig. 5A and B). PCR from GV-1.4 tumors all showed the expected 291-bp fragment but also presented fragments of larger size corresponding to enhancer modification, presumably duplications and triplications. Also, PCRs for GV-1.2 tumors all showed the expected 351-bp fragment and additional larger fragments (Fig. 5A and B). Interestingly, a few GV-1.2 tumors contained the lower-molecular-weight 291-bp fragment typical of GV-1.4 (Fig. 5A, lane 3, and B, lanes 2 to 4).
The phenomenon was more rarely observed in the nonlymphoid leukemias, since they do not show duplication or multimerization of the enhancer region (Fig. 5C and D). All the nonlymphoid tumors induced with GV-1.4 (Fig. 5C, lanes 1 to 13) showed no alteration of the provirus enhancer region; lanes 2 and 8 on Fig. 5C show mixed nonlymphoid and T-cell leukemias. However, GV-1.2-induced erythroid leukemias (Fig. 5D) often presented enhancer alteration with a majority of additional smaller fragments corresponding to the GV-1.4 enhancer (Fig. 5D, lanes 3 and 5 and 6 to 8). Thus, the constitution of the enhancer region can be altered during the viral replication cycles, and this surely has an impact on tumor development.
DISCUSSION
This study on the pathogenesis of the Graffi murine retrovirus indicates that it is an extremely multipotent virus, able to give rise to leukemias from every lineage of the hematopoietic system: T-cell, B-cell, myeloid, erythroid, and even megakaryocytic lineages. However, GV-1.2 and GV-1.4 induce the disease with different latencies, and more importantly, they induce different types of leukemia. GV-1.2 induces a large percentage of T-cell leukemias, although GV-1.4 induces a much wider spectrum of leukemias, including nonlymphoid leukemias.
Graffi is extremely multipotent and induces complex leukemias. The induced leukemias were found to be heterogeneous. The leukemia from each animal harbored its particularities, and several leukemias showed complex phenotypes. This heterogeneity reflects the very high polyvalence of the Graffi MuLV. In the case of T-cell leukemias, some tumors were more complex and did not show the expected phenotypes. The TCR rearrangement was in several cases not correlated with a more mature phenotype, and more curiously, some leukemias harbored an immunoglobulin H rearrangement. This phenomenon was already observed in leukemias induced by Moloney viral recombinants and in thymomas in AKR/J mice (20, 22, 24, 40). Possibly, the leukemic T cells derive from very early lymphoid progenitors that still retain some B-cell potential, and the deregulation induced by the viral integration may activate the machinery of the immunoglobulin rearrangement. Some leukemias showed much more complex phenotypes that could be considered aberrant. First, these are the biphenotypic lymphoid leukemias expressing the CD11b myeloid marker. Corresponding human leukemias are widely reported in the literature (for examples, see references 7, 16, 38, 44, 49, and 51), and expression of CD11b on human leukemic B cells is associated with unfavorable prognosis (26, 45, 55). Some studies showed that these biphenotypic cells do exist in normal mice although they are very rare (1, 5, 17, 34). Second, this concerns also the apparently aberrant expression of MPO mainly in T-cell and B-cell leukemias but also in two erythroid and one megakaryocytic case. Since RNA was extracted from whole spleen extracts, the high MPO level can be explained in two ways. First, it could be due to a large infiltration of adherent myeloid cells (histiocytes) that were not included in the flow cytometry analysis since the cell suspension was obtained by mechanical disruption without collagenase treatment. Second, the nonmyeloid leukemic cells could exhibit promiscuous gene expression of MPO as a consequence of retroviral insertional activation of this gene. We favor the latter possibility because we could not find any signs of myeloid cell infiltration in the tumors by morphological assessment of the samples (spleen imprints). This highly aberrant expression of MPO in many leukemias is probably the main feature responsible for the higher percentage of tumors classified as myeloid leukemias observed by Ru et al. (48), who used the MPO staining as an important criterion.
Thus, the high versatility of Graffi MuLV suggests that the target cells are either very early or early committed progenitors from all the lineages. Indeed, when methylcellulose colony assays were performed with bone marrow cells from infected mice, Graffi proviral DNA could be detected in every type of colony tested (CFU-G, CFU-GM, CFU-M, CFU-E, CFU-GEMM) as early as 4 days postinfection, and by 15 days postinfection, all colonies were highly positive (data not shown).
Importance of mouse genetics. Interestingly, each strain of mice responded differently to Graffi infection. First, the latency was different among the different strains, and second, the distribution of the various leukemias was specific to each strain of mice. BALB/c mice showed the longest latency period and preferentially developed T-cell leukemias. The NFS strain was more susceptible to B-cell leukemias. Finally, the FVB/n strain showed the shortest latency and developed mainly nonlymphoid leukemias. Interestingly, this strain of mice carries an activated form of the K-ras oncogene (52) which could reduce the number of hits required for leukemogenesis.
Similar observations were made in previous studies. For example, a Moloney MuLV with a mutated core motif was shown to induce 60% of erythroleukemia cases in NFS mice but almost 100% of T-cell lymphoma cases in BALB/c mice (61). Similarly, the recombinant virus MOL4070LTR (Moloney with the 4070A LTR) induces 46% of lymphomas in BALB/c mice and 23% in FVB/n mice (60).
Importance of the retroviral LTR. The importance of the retroviral LTR in disease specificity is clearly established. The Graffi retroviruses are capable to infect a very large spectrum of cell types, and the major determinant is very likely due to its LTR. Indeed, we analyzed the Graffi LTR-driven transcriptional activity in several cell lines (4). The U3 region of GV-1.2 was found to be a very strong promoter in all the hematopoietic cell lines tested (erythroid, myeloid, and T-lymphoid). GV-1.4 LTR was globally less powerful than GV-1.2 but showed a similar wide-range specificity. Thus, the Graffi U3 region could bind transcription factors present in different cell lineages and be equally activated in these cells, in contrast to Friend and Moloney viruses, which show enhanced activity only in erythroid and T-cell lines, respectively (4).
Indeed, DNase I footprint analyses using the LTR of GV-1.4 revealed a similar pattern of protected regions in different cell lines (erythroid, myeloid, and T-cell) (4). Some of the LTR's specific transcription factors were identified for Graffi, and we showed that GATA-1, -2 and -3 were able to activate the promoters of both GV-1.2 and GV-1.4 (3). GATA-1 is expressed in all erythroid cell lineages, megakaryocytes, mast cells, and multipotent stem cells and has been involved in the regulation of most erythroid genes (39). GATA-3 is expressed at all stages of T-cell development (14, 39). GATA-2 has a much broader expression pattern, including myeloid and erythroid cell lineages and endothelial cells, and is essential for multilineage hematopoiesis (39). Thus, the importance of the GATA members in the Graffi LTR activation can explain the large spectrum of infected cells. These proteins act in cooperation with other cofactors, such as Tal1/Scl (30). Interestingly, a perfect consensus site for TAL1 is present proximal to the GATA binding site. It would therefore be noteworthy to identify the factors interacting with the GATA members in different cell lines.
An interesting feature that became evident in this study is that the viral variants GV-1.2 and GV-1.4 have some differences in latency and type of induced leukemia. The latency period of GV-1.2 is shorter than that of GV-1.4 (Fig. 1) (48), and GV-1.2 induces a higher percentage of T-cell leukemias (Table 1). We sequenced the two genomes, GV-1.2 and GV-1.4, which originate from the same tumor, and we found that they are very similar except for the duplication in the U3 LTR region (GenBank/EMBL/DDBJ accession numbers: GV-1.4, AB187566; GV-1.2, AB187565) (V. Voisin et al., in preparation). Except for the U3 region, only 16 nucleotide differences were found along their entire genomes.
The differences in latency are not likely to be influenced by viral titer since the two variants showed equivalent titers. These variations in latency are more likely explained by the presence of two direct repeats in the GV-1.2 LTR. Probably, this allows the binding of more transcription factors, making it globally more transcriptionally powerful as shown previously (4).
GV-1.2 induces a higher percentage of T-cell leukemias. The direct repeat encompasses the well-studied region of the LVb/CORE binding sites. The CBF and MCREF bind the CORE of Moloney in T-cells and Friend in erythroid cells, respectively (15, 32, 35, 53). Therefore, the duplicated region contains elements that are known to be more important for activation in T cells and also erythroid cells, which correspond to the major types of leukemia induced by GV-1.2. In FVB/n mice, a high incidence (72%) of erythroleukemia is induced with a very short latency compared to that in NFS and BALB/c mice. This suggests that the duplication has a strong impact on promoter activity in erythroid cells in this particular strain.
Granger et al. (20) hypothesize that the tandemerization of the enhancer region has a favorable impact on the development of T-cell leukemias. Interestingly, they found that the proviruses from lymphoid tumors induced by a SRS-19.6-Moloney recombinant virus had acquired new sequence duplication in the enhancer region. Similar sequence duplications were also observed in wild-type SRS-19.6-induced tumors and were restricted to T-cell tumors. Considering the high homology between GV-1.4 and SRS-19.6 LTRs, we verified the integrity of the proviral enhancers in mature tumors. The analysis revealed that the Graffi MuLV enhancer is not stable and fluctuates between different forms during the replication cycles, showing duplication, multimerization, and even loss of duplication (Fig. 5). One could hypothesize that these alterations are favorable to tumorigenesis. Granger et al. (20) proved that the tandemerization of the enhancer region of SRS-19.6 was associated with T-cell leukemias but not with other leukemias. We can conclude from this study that a multimerization of the enhancer region seems to be indeed related with lymphoid leukemia. Indeed, a much higher percentage of lymphoid leukemias is generated by GV-1.2, which contains an enhancer duplication, and most of the alterations observed with this variant yield fragments of higher molecular weight. Moreover, the minimal enhancer corresponding to the GV-1.4 form seems more favorable to the induction of nonlymphoid leukemias since most leukemias of this kind arose after the inoculation of GV-1.4. In contrast with the lymphoid leukemias, nonlymphoid leukemias do not show duplication or multimerization of the enhancer (Fig. 5C). Furthermore, some GV-1.2-induced erythroid leukemias showed a loss of duplication (Fig. 5D, lanes 3, 5, and 6 to 8).
Since GV-1.2 shows a shorter latency period in addition to inducing more T-cell leukemias, one could speculate that these leukemias develop faster than the nonlymphoid ones. This was not the case, and there was no correlation between the latency period and the type of leukemia observed in each individual litter (data not shown).
Comparison with SRS-19-6 MulV. The sequencing of the Graffi genomes revealed an overall homology of 97% between the Graffi and SRS-19.6 MuLVs, demonstrating that they must originate from a common ancestor. However, there are enough differences to suggest that they represent distinct viruses. The nucleotide changes are dispersed along the genomes even in well-conserved regions. Two segments (the 5'end of the POL gene and the 3'end of the ENV gene) are the most divergent regions between Graffi and SRS-19.6 (Voisin et al., in preparation). The SRS-19.6 virus was cloned from a transmissible system developed in China and involving passaging of the viral mixture in Kunming mice, an inbred strain developed from NIH/Swiss mice (6). This virus is able to induce a large spectrum of leukemias in NIH/Swiss mice (myeloid leukemias, B- and T-cell lymphomas, and erythroid leukemias). This distribution is very similar to our results, in particular those obtained with the GV-1.4 variant inoculated into NFS mice (Table 1). However, different strains of mice were used in these two studies. As the mouse genetic background appears to be very important, the two viruses could still give slightly different phenotypes if inoculated into the same strain of mice, reflecting the differences found in their respective genomes.
A good model for megakaryocytic leukemia. A high percentage of nonlymphoid leukemias were induced with GV-1.4, especially in the FVB/n strain. They consist mostly of erythroid (TER119+ CD71+) and megakaryocytic leukemias (CD41+ cKit+). Some MuLVs, such as Friend and Rauscher, are known to induce erythroid leukemias, but very few studies on MuLVs have reported megakaryocytic leukemia. Thus, Graffi MuLV seems to be a good model to study megakaryocytes. Some specific genes of megakaryocytes and eythroid lineages were analyzed to confirm the true lineage of these leukemias (Fig. 4B). The expression of these genes did not show any erythroid versus megakaryocytic specificity or vice versa. This demonstrates the closed relationship of the two lineages that is noteworthy and widely reported in the literature. Several examples of human erythroid leukemic cell lines from patients (47, 54) or from MuLV models (41, 58) are reported to bear both erythrocyte and megakaryocyte characteristics. Moreover, a common erythromegakaryocytic progenitor, bearing the surface molecules TER119 and CD41, was identified in the bone marrow (36, 42, 59). This suggests that the TER119+ CD41+ leukemias observed mostly in GV-1.4-infected FVB/n mice may correspond to this progenitor and therefore represents an interesting model to study the erythromegakaryocytic lineage.
Graffi: a myeloid leukemia-inducing virus All together, these results show that both Graffi variants, GV-1.2 and GV-1.4, can induce a variety of leukemias in BALB/c, NFS, and FVB/n strains of mice. This exhaustive study shows that Graffi MuLV induces complex leukemias and that several immunological and molecular markers are required to classify them properly. These results are different from our previous study on the characterization of Graffi MuLV (48), where leukemias were considered mainly myeloid. Our earlier study (48) was mainly based on the observations of blood smears, spleen imprints, MPO staining, and some molecular biology analysis. We described these cases as myeloid leukemias with lineage infidelities since they also contained either immunoglobulin or TCR gene rearrangements or both. In the present study, the high levels of MPO expression in the spleen, often in lymphoid leukemias, are probably caused by aberrant gene expression.
Erkeland et al. (10) have described Graffi MuLV as inducing a majority of myeloid leukemias in the FVB/n strain of mice. The virus also induces a fair proportion of other tumor types. This discrepancy in results could be interpreted by a different classification approach and by the use of different antibodies. We followed the Bethesda proposal (28, 37) to classify the leukemias. In comparison with their study, we did not use the surface antigen F4/80, as we found it redundant with CD11b (Mac-1), nor did we use ERMP-58, also specific to the myeloid lineage (8). We did phenotype some of the leukemias with ERMP-58 (gift of P. Leenen) (data not shown): it gave high positivity for all the myeloid leukemias, especially the CD11b+ Gr-1+ phenotype. We tested three additional antibodies: c-Kit, CD34, and CD41. This last antibody allowed the characterization of the 48% megakaryocytic leukemias found in GV-1.4-infected FVB/n mice. Finally, we cannot rule out a possible dissimilarity in the FVB/n mouse strain origin or viral isolate.
The Graffi-induced leukemias are now well characterized and can be used as a model to study hematopoiesis and leukemic progression in mice. It is now established that SRS 19-6 and Graffi are tightly related viruses and have a common ancestor. To date, as far as we know, the high incidence of megakaryocytic leukemias induced with the inoculation of GV-1.4 into the FVB/n strain of mice was never described with any other MuLVs. The results indicate also a tight relationship between the erythroid and megakaryocytic lineages, and further studies would certainly increase the knowledge about these two hematopoietic lineages.
ACKNOWLEDGMENTS
We thank Denis Flipo for his help in flow cytometry. We thank Sonia Do Carmo and Philippe Legault for their help in sequencing the two Graffi genomes.
This work was supported by grant MOP-37994 from the Canadian Institutes of Health Research and by grant 3279-01 from the Natural Sciences Engineering Council of Canada.
REFERENCES
Abrahao, T. B., E. Freymuller, R. A. Mortara, J. D. Lopes, and M. Mariano. 2003. Morphological characterization of mouse B-1 cells. Immunobiology 208:401-411.
Allman, D., J. Li, and R. R. Hardy. 1999. Commitment to the B lymphoid lineage occurs before DH-JH recombination. J. Exp. Med. 189:735-740.
Barat, C., and E. Rassart. 1998. Members of the GATA family of transcription factors bind to the U3 region of Cas-Br-E and Graffi retroviruses and transactivate their expression. J. Virol. 72:5579-5588.
Barat, C., and E. Rassart. 1998. Nuclear factors that bind to the U3 region of two murine myeloid leukemia-inducing retroviruses, Cas-Br-E and Graffi. Virology 252:82-95.
Borrello, M. A., and R. P. Phipps. 1995. Fibroblasts support outgrowth of splenocytes simultaneously expressing B lymphocyte and macrophage characteristics. J. Immunol. 155:4155-4161.
Bundy, L. M., M. Ru, B. F. Zheng, L. Cheng, P. K. Pattengale, J. L. Portis, et al. 1995. Biological characterization and molecular cloning of murine C-type retroviruses derived from the TSZ complex from mainland China. Virology 212:367-382.
Butcher, B. W., K. S. Wilson, S. H. Kroft, R. H. Collins, Jr., and V. Bhushan. 2005. Acute leukemia with B-lymphoid and myeloid differentiation associated with an inv(5)(q13q33) in an adult patient. Cancer Genet. Cytogenet. 157:62-66.
de Bruijn, M. F., R. E. Ploemacher, A. E. Mayen, J. S. Voerman, W. A. Slieker, W. van Ewijk, et al. 1996. High-level expression of the ER-MP58 antigen on mouse bone marrow hematopoietic progenitor cells marks commitment to the myeloid lineage. Eur. J. Immunol. 26:2850-2858.
Dialynas, D. P., Z. S. Quan, K. A. Wall, A. Pierres, J. Quintans, M. R. Loken, et al. 1983. Characterization of the murine T cell surface molecule, designated L3T4, identified by monoclonal antibody GK1.5: similarity of L3T4 to the human Leu-3/T4 molecule. J. Immunol. 131:2445-2451.
Erkeland, S. J., M. Valkhof, C. Heijmans-Antonissen, A. van Hoven-Beijen, R. Delwel, M. H. A. Hermans, and I. P. Touw. 2004. Large-scale identification of disease genes involved in acute myeloid leukemia. J. Virol. 78:1971-1980.
Fey, F. 1969. Studies on the pathogenesis and mechanism of hematologic diversification by re-isolation of the myeloid leukemia virus (Graffi). Acta Haematol. 42:65-75.
Fey, F., and A. Graffi. 1965. Erythroblastic leukemia after the injection of mouse myelogenous leukemia virus. Z. Krebsforsch. 67:145-151. (In German.)
Fleming, T. J., M. L. Fleming, and T. R. Malek. 1993. Selective expression of Ly-6G on myeloid lineage cells in mouse bone marrow. RB6-8C5 mAb to granulocyte-differentiation antigen (Gr-1) detects members of the Ly-6 family. J. Immunol. 151:2399-2408.
Georgopoulos, K. 1997. Transcription factors required for lymphoid lineage commitment. Curr. Opin. Immunol. 9:222-227.
Golemis, E., Y. Li, T. N. Fredrickson, J. W. Hartley, and N. Hopkins. 1989. Distinct segments within the enhancer region collaborate to specify the type of leukemia induced by nondefective Friend and Moloney viruses. J. Virol. 63:328-337.
Gonzalez Garcia, J. R., S. K. Bohlander, M. Gutierrez Angulo, M. A. Esparza Flores, V. J. Picos Cardenas, J. P. Meza Espinoza, et al. 2004. A t(1;9)(q23.3 approximately q25;q34) affecting the ABL1 gene in a biphenotypic leukemia. Cancer Genet. Cytogenet. 152:81-83.
Graf, B. A., D. A. Nazarenko, M. A. Borrello, L. J. Roberts, J. D. Morrow, J. Palis, et al. 1999. Biphenotypic B/macrophage cells express COX-1 and up-regulate COX-2 expression and prostaglandin E(2) production in response to pro-inflammatory signals. Eur. J. Immunol. 29:3793-3803.
Graffi, A. 1957. Chloroleukemia of mice. Ann. N. Y. Acad. Sci. 68:540-558.
Graffi, A., F. Fey, and T. Schramm. 1966. Experiments on the hematologic diversification of viral mouse leukemias. Natl. Cancer Inst. Monogr. 22:21-31.
Granger, S. W., L. M. Bundy, and H. Fan. 1999. Tandemization of a subregion of the enhancer sequences from SRS 19-6 murine leukemia virus associated with T-lymphoid but not other leukemias. J. Virol. 73:7175-7184.
Gurney, A. L., K. Carver-Moore, F. J. de Sauvage, and M. W. Moore. 1994. Thrombocytopenia in c-mpl-deficient mice. Science 265:1445-1447.
Hanecak, R., P. K. Pattengale, and H. Fan. 1988. Addition of substitution of simian virus 40 enhancer sequences into the Moloney murine leukemia virus (M-MuLV) long terminal repeat yields infectious M-MuLV with altered biological properties. J. Virol. 62:2427-2436.
Hardy, R. R., Y. S. Li, D. Allman, M. Asano, M. Gui, and K. Hayakawa. 2000. B-cell commitment, development and selection. Immunol. Rev. 175:23-32.
Herr, W., A. P. Perlmutter, and W. Gilbert. 1983. Monoclonal AKR/J thymic leukemias contain multiple JH immunoglobulin gene rearrangements. Proc. Natl. Acad. Sci. USA 80:7433-7436.
Ho, M. K., and T. A. Springer. 1982. Mac-1 antigen: quantitative expression in macrophage populations and tissues, and immunofluorescent localization in spleen. J. Immunol. 128:2281-2286.
Jensen, G. S., A. R. Belch, F. Kherani, M. J. Mant, B. A. Ruether, and L. M. Pilarski. 1992. Restricted expression of immunoglobulin light chain mRNA and of the adhesion molecule CD11b on circulating monoclonal B lineage cells in peripheral blood of myeloma patients. Scand. J. Immunol. 36:843-853.
Kina, T., K. Ikuta, E. Takayama, K. Wada, A. S. Majumdar, I. L. Weissman, et al. 2000. The monoclonal antibody TER-119 recognizes a molecule associated with glycophorin A and specifically marks the late stages of murine erythroid lineage. Br. J. Haematol. 109:280-287.
Kogan, S. C., J. M. Ward, M. R. Anver, J. J. Berman, C. Brayton, R. D. Cardiff, et al. 2002. Bethesda proposals for classification of nonlymphoid hematopoietic neoplasms in mice. Blood 100:238-245.
Lagasse, E., and I. L. Weissman. 1996. Flow cytometric identification of murine neutrophils and monocytes. J. Immunol. Methods 197:139-150.
Lecuyer, E., and T. Hoang. 2004. SCL: from the origin of hematopoiesis to stem cells and leukemia. Exp. Hematol. 32:11-24.
Ledbetter, J. A., R. V. Rouse, H. S. Micklem, and L. A. Herzenberg. 1980. T cell subsets defined by expression of Lyt-1,2,3 and Thy-1 antigens. Two-parameter immunofluorescence and cytotoxicity analysis with monoclonal antibodies modifies current views. J. Exp. Med. 152:280-295.
Li, Y., E. Golemis, J. W. Hartley, and N. Hopkins. 1987. Disease specificity of nondefective Friend and Moloney murine leukemia viruses is controlled by a small number of nucleotides. J. Virol. 61:693-700.
Lok, C. N., and T. T. Loh. 1998. Regulation of transferrin function and expression: review and update. Biol. Signals Recept. 7:157-178.
Lu, M., H. Kawamoto, Y. Katsube, T. Ikawa, and Y. Katsura. 2002. The common myelolymphoid progenitor: a key intermediate stage in hemopoiesis generating T and B cells. J. Immunol. 169:3519-3525.
Manley, N. R., M. O'Connell, W. Sun, N. A. Speck, and N. Hopkins. 1993. Two factors that bind to highly conserved sequences in mammalian type C retroviral enhancers. J. Virol. 67:1967-1975.
McDonald, T. P., and P. S. Sullivan. 1993. Megakaryocytic and erythrocytic cell lines share a common precursor cell. Exp. Hematol. 21:1316-1320.
Morse, H. C., III, M. R. Anver, T. N. Fredrickson, D. C. Haines, A. W. Harris, N. L. Harris, et al. 2002. Bethesda proposals for classification of lymphoid neoplasms in mice. Blood 100:246-258.
Neame, P. B., P. Soamboonsrup, G. Browman, R. D. Barr, N. Saeed, B. Chan, et al. 1985. Simultaneous or sequential expression of lymphoid and myeloid phenotypes in acute leukemia. Blood 65:142-148.
Orkin, S. H. 1992. GATA-binding transcription factors in hematopoietic cells. Blood 80:575-581.
Ott, D. E., J. Keller, K. Sill, and A. Rein. 1992. Phenotypes of murine leukemia virus-induced tumors: influence of 3' viral coding sequences. J. Virol. 66:6107-6116.
Paoletti, F., A. M. Vannucchi, A. Mocali, R. Caporale, and S. A. Burstein. 1995. Identification and conditions for selective expression of megakaryocytic markers in Friend erythroleukemia cells. Blood 86:2624-2631.
Papayannopoulou, T., M. Brice, D. Farrer, and K. Kaushansky. 1996. Insights into the cellular mechanisms of erythropoietin-thrombopoietin synergy. Exp. Hematol. 24:660-669.
Pereira, R., C. T. Quang, I. Lesault, H. Dolznig, H. Beug, and J. Ghysdael. 1999. FLI-1 inhibits differentiation and induces proliferation of primary erythroblasts. Oncogene 18:1597-1608.
Perentesis, J., N. K. Ramsay, R. Brunning, J. H. Kersey, and A. H. Filipovich. 1983. Biphenotypic leukemia: immunologic and morphologic evidence for a common lymphoid-myeloid progenitor in humans. J. Pediatr. 102:63-67.
Pinto, A., L. Del Vecchio, A. Carbone, M. Roncadin, R. Volpe, and D. Serraino. 1991. Expression of myelomonocytic antigens is associated with unfavourable clinicoprognostic factors in B-cell chronic lymphocytic leukaemia. Ann. Oncol. 2(Suppl. 2):107-113.
Rosmarin, A. G., S. C. Weil, G. L. Rosner, J. D. Griffin, M. A. Arnaout, and D. G. Tenen. 1989. Differential expression of CD11b/CD18 (Mo1) and myeloperoxidase genes during myeloid differentiation. Blood 73:131-136.
Rowley, P. T., B. A. Farley, S. LaBella, R. Giuliano, and J. F. Leary. 1992. Single K562 human leukemia cells express and are inducible for both erythroid and megakaryocytic antigens. Int. J. Cell Cloning 10:232-240.
Ru, M., C. Shustik, and E. Rassart. 1993. Graffi murine leukemia virus: molecular cloning and characterization of the myeloid leukemia-inducing agent. J. Virol. 67:4722-4731.
Rubio, M. T., N. Dhedin, C. Boucheix, J. H. Bourhis, O. Reman, J. M. Boiron, et al. 2003. Adult T-biphenotypic acute leukaemia: clinical and biological features and outcome. Br. J. Haematol. 123:842-849.
Sambrook, J., and D. W. Russell. 2001. Molecular cloning: a laboratory manual, 3rd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
Scamurra, D. O., F. R. Davey, D. A. Nelson, A. S. Kurec, and J. Goldberg. 1983. Acute leukemia presenting with myeloid and lymphoid cell markers. Ann. Clin. Lab. Sci. 13:496-502.
Shafarenko, M., J. Mahler, C. Cochran, A. Kisielewski, E. Golding, R. Wiseman, et al. 1997. Similar incidence of K-ras mutations in lung carcinomas of FVB/N mice and FVB/N mice carrying a mutant p53 transgene. Carcinogenesis 18:1423-1426.
Speck, N. A., and D. Baltimore. 1987. Six distinct nuclear factors interact with the 75-base-pair repeat of the Moloney murine leukemia virus enhancer. Mol. Cell. Biol. 7:1101-1110.
Tani, T., J. Ylanne, and I. Virtanen. 1996. Expression of megakaryocytic and erythroid properties in human leukemic cells. Exp. Hematol. 24:158-168.
Tassies, D., E. Montserrat, J. C. Reverter, N. Villamor, M. Rovira, and C. Rozman. 1995. Myelomonocytic antigens in B-cell chronic lymphocytic leukemia. Leuk. Res. 19:841-848.
Tsang, A. P., J. E. Visvader, C. A. Turner, Y. Fujiwara, C. Yu, M. J. Weiss, et al. 1997. FOG, a multitype zinc finger protein, acts as a cofactor for transcription factor GATA-1 in erythroid and megakaryocytic differentiation. Cell 90:109-119.
Uzan, G., M. Prenant, M. H. Prandini, F. Martin, and G. Marguerie. 1991. Tissue-specific expression of the platelet GPIIb gene. J. Biol. Chem. 266:8932-8939.
Vannucchi, A. M., S. Linari, C. Cellai, M. J. Koury, and F. Paoletti. 1997. Constitutive and inducible expression of megakaryocyte-specific genes in Friend erythroleukaemia cells. Br. J. Haematol. 99:500-508.
Vannucchi, A. M., F. Paoletti, S. Linari, C. Cellai, R. Caporale, P. R. Ferrini, et al. 2000. Identification and characterization of a bipotent (erythroid and megakaryocytic) cell precursor from the spleen of phenylhydrazine-treated mice. Blood 95:2559-2568.
Wolff, L., R. Koller, X. Hu, and M. R. Anver. 2003. A Moloney murine leukemia virus-based retrovirus with 4070A long terminal repeat sequences induces a high incidence of myeloid as well as lymphoid neoplasms. J. Virol. 77:4965-4971.
Yuen, P. H., Y. H. Khang, and C. Matherne. 1992. Virulence in lymphomagenesis is modulated by alterations in proteins binding to CORE and GRE-LVa elements of the MoMuLV enhancer. Leukemia 6(Suppl. 3):76S-82S.
Zhang, C., M. A. Thornton, M. A. Kowalska, B. S. Sachis, M. Feldman, M. Poncz, et al. 2001. Localization of distal regulatory domains in the megakaryocyte-specific platelet basic protein/platelet factor 4 gene locus. Blood 98:610-617.(Veronique Voisin, Corinne)
Institut de Recherche en Immunologie et Cancer, Universite de Montreal, Quebec, Canada
ABSTRACT
The Graffi murine leukemia virus (MuLV) was isolated in 1954 by Arnold Graffi, who characterized it as a myeloid leukemia-inducing retrovirus. He and his team, however, soon observed the intriguing phenomenon of hematological diversification, which corresponded to a decrease of myeloid leukemias and an increase of other types of leukemias. Recently, we derived two different molecular clones corresponding to ecotropic nondefective genomes that were named GV-1.2 and GV-1.4. The induced leukemias were classified as myeloid based on morphological analysis of blood smears. In this study, we further characterized the two variants of the Graffi murine retrovirus, GV-1.2 and GV-1.4, in three different strains of mice. We show that the Graffi MuLV is a multipotent retrovirus capable of inducing both lymphoid (T- and B-cell) and nonlymphoid (myeloid, erythroid, megakaryocytic) leukemia. Many of these are very complex with concomitant expression of different hematopoietic lineages. Interestingly, a high percentage of megakaryocytic leukemias, a type of leukemia rarely observed with MuLVs, arise in the FVB/n strain of mice. The genetic backgrounds of the different strains of mice influence greatly the results. Furthermore, the enhancer region, different for GV-1.2 and GV-1.4, plays a pivotal role in the disease specificity: GV-1.2 induces more lymphoid leukemias, and GV-1.4 induces more nonlymphoid ones.
INTRODUCTION
About 50 years ago Arnold Graffi, at the Cancer Research Institute in Berlin, Germany, isolated a new retrovirus capable of inducing leukemia in a specific strain of mice, named Agnes-Bluhm, bred in his laboratory. Graffi (18) induced a large proportion of myeloid leukemias with a very high incidence of chloroleukemias (70%), characterized by a greenish coloration of the lymph nodes, and he classified the virus as a myeloid leukemia-inducing retrovirus.
Working on better characterizing the pathogenesis of the virus, Graffi and his team were soon confronted with complex results. They observed the intriguing phenomenon of hematological diversification, which corresponded to a decrease in the percentage of chloroleukemias and an increase in other types of leukemias (11, 12, 19). This virus could induce multiple kinds of myeloid leukemias, from immature to differentiated forms, but also reticular, lymphoid, and erythroid leukemias and mixed forms (based on the classification at this time).
It is very likely that Graffi was working at that time with a viral mixture. In 1993, Ru et al. (48) cloned two nondefective ecotropic retroviral genomes from NIH-3T3 cell lines chronically infected with the original Graffi acellular extract (a gift of Nathalie Teich). These two viral variants were called GV-1.2 and GV-1.4. The genomic restriction map and the sequences of their long terminal repeats (LTRs) (GenBank accession numbers L14415 and L14416 for GV-1.2 and GV-1.4, respectively) showed that they differ in the U3 region, one 60-bp segment being duplicated in GV-1.2 and not in GV-1.4. GV-1.2 induces disease with a shorter latency period. Based mainly on morphological analysis of blood smears, the leukemias induced with GV-1.2 and GV-1.4 were described as mainly myeloid/granulocytic when injected into newborn BALB/c and NFS mice (48). In several tumors, we observed myeloid characteristics (myeloperoxidase [MPO] staining) together with rearrangements of the germinal configuration of the T-cell receptor (TCR) gene and/or of the immunoglobulin heavy chain gene, specific for T-cell and B-cell lineages, respectively. This phenomenon was called lineage infidelity.
In this report, we extended our characterization of the Graffi retrovirus. GV-1.2 and GV-1.4 were inoculated into three strains of mice (BALB/c, NFS, and FVB/n), and more than 100 leukemias were analyzed by flow cytometry with several different hematopoietic markers combined with molecular biology tools.
The complete genome sequences of the two variants were determined and used to complement the immunophenotyping analysis. This study reveals that the Graffi murine leukemia virus (MuLV) is a complex virus capable to induce a large spectrum of leukemias. It replicates efficiently in several cell types, as already suggested in studies of the LTR (3, 4). It is multipotent and induces both lymphoid and nonlymphoid leukemia, including megakaryocytic leukemias, which are quite rare in MuLV-induced pathologies.
MATERIALS AND METHODS
Viruses, inoculation of mice, and tissue collection. Viral stocks of GV-1.2 and GV-1.4 were made from cell culture supernatants derived from chronically infected NIH-3T3 cells. The culture supernatant of the infected cells was collected, centrifuged, aliquoted, and kept at –80°C. Newborn (<24 h-old) mice of strains BALB/c, NFS, and FVB/n were inoculated intraperitoneally with 0.1 ml of filtered virus. The viral titers were 3.106 PFU for GV-1.2 and 1.106 PFU for GV-1.4. The mice were checked routinely for clinical signs of disease (loss of weight, anemia, enlarged spleen). Moribund mice were sacrificed, and several organs (thymus, lymph nodes, spleen, liver, kidneys, brain) were frozen on dry ice and stored at –80°C. A sterile cell suspension from the spleen, thymus, and lymph nodes was prepared by gently teasing the organs apart in cold RPMI cell culture medium and by passing them through nylon mesh. Viable cells were used for flow cytometry analyses or stored at –80°C in 90% fetal calf serum and 10% dimethyl sulfoxide. Bone marrow was recovered by flushing cold RPMI cell culture medium through the femur with a 26G3/8 needle. Blood and spleen samples were incubated in hypotonic ammonium chloride solution to eliminate the erythrocytes.
Immunophenotyping. 106 cells were suspended in 100 μl of standard phosphate-buffered saline buffer. For detection of surface antigens, the antibodies (quantities calculated from a titration experiment) were added, and the cells were incubated for 30 min in the dark. The cells were then washed with cold phosphate-buffered saline and resuspended in 500 μl of a standard isotonic solution (Hematal; Fisher). The antibodies (BD Pharmingen, Mississauga, Canada) used were as follows: fluorescein isothiocyanate (FITC)-conjugated rat anti-mouse CD4, FITC anti-CD3, phycoerythrin (PE)-conjugated anti-CD90, FITC anti-CD45R/B220, FITC anti-CD11b (Mac-1, M1/70), FITC anti-CD71 (transferrin receptor), FITC anti-CD41(glycoprotein IIB), FITC anti-CD34, FITC anti-CD117 (cKit, stem cell factor ligand), FITC anti-Sca1, PE anti-CD8a, PE anti-Gr-1 (Ly6G/Ly6C, RB6-8C5), and PE anti-TER119 (glycophorin A). Isotype controls (BD Pharmingen) were included for each antibody used. Flow cytometric analyses were done with a fluorescence-activated cell sorter scanner (FACScan; Becton Dickinson), and the data were processed with the WinMDI software. CD3 together with CD90, CD4 together with CD8, Mac-1 together with Gr-1, and Ter119 together with CD71 were used in double-staining combinations.
RNA extraction and Northern blotting. Total RNA samples were extracted from the tissues using the TRIzol reagent kit (Invitrogen, Burlington, Canada) and stored at –80°C. Fifteen micrograms of total RNA samples were denatured with formamide and separated on a 1% agarose gel containing formaldehyde. The RNA was transferred to a nylon membrane, UV fixed, and hybridized with DNA probes labeled by the random primer extension method using oligohexamers (GE Healthcare, Baie d'Urfe, Canada), and approximately 2 x 106 cpm/ml was used. The hybridization process was done as described previously (50). The cDNA fragments used as probes were as follows: myeloperoxidase (S. Gisselbrecht), PU.1 (F. Moreau-Gachelin), EpoR (H. Lodish), Fli-1 (A. Bernstein), FOG-1 (S. Orkin), c-mpl, GpIIb, and PF4 (F. Morle). The expression of beta-actin was used as a control for equal RNA loading and to normalize relative expression data.
DNA extraction and Southern blotting. The genomic DNA was extracted from normal and leukemic tissues by standard procedures as described previously (50). The DNA was digested with appropriate restriction enzymes, separated on a 0.8% agarose gel, and transferred to nylon membranes (Ambion) for hybridization. A 600-bp TCR probe (kindly provided by H. Fan) and a 600-bp probe corresponding to the Jh region (provided by F. Alt) were used to detect T-cell receptor and heavy chain immunoglobulin gene rearrangements, respectively.
PCR amplification of the proviral LTR. PCR amplification of the proviral LTR region was performed by using genomic tumor DNAs with the forward primer 5'-CCCCACCATAAGGCTTAGCAAGCTAG-3' and the reverse primer 5'-TAGTTTCAAATGAGGCGCAAG-3' and using the PCR CORE kit (QIAGEN, Mississauga, Canada). Each reaction product was amplified by PCR as follows: 1 cycle at 95°C for 5 min; 30 cycles at 94°C for 1 min, 61°C for 1 min, 72°C for 1 min; and 1 cycle at 72°C for 5 min.
RESULTS
Disease latency. Neonate mice of three different strains (BALB/c, NFS, and FVB/n) were inoculated with similar doses of the two Graffi MuLV variants GV-1.2 and GV-1.4 (see Materials and Methods). Leukemias appeared with an incidence of 99%. About 1% of the mice suffered from hindlimb paralysis and were not included in the study. Figure 1 shows the cumulative survival curves in the three different stains of mice inoculated with GV-1.4 (Fig. 1A) and GV-1.2 (Fig. 1B).
After the inoculation of GV-1.2 (Fig. 1B), 50% of the mice showed signs of disease at day 68 for the FVB/n group, day 79 for the NFS group, and day 81 for the BALB/c group. After the inoculation of GV-1.4 (Fig. 1A), 50% of the leukemias were detected after 139, 158, and 181 days in FVB/n, NFS, and BALB/c mice, respectively. Thus, GV-1.2, which contains two direct repeats in the LTR, induces leukemias with a shorter latency than GV-1.4, which contains only one direct repeat as previously described (48). Uncorrelated with the viral variant inoculated, the latency period is also significantly dependent upon the mouse strain (as measured by a log rank test; P < 0.05), reflecting the effect of the genetic background on the disease development. Leukemias developed more quickly in FVB/n mice than in NFS and BALB/c mice, the latter strain showing the longest latency period.
Gross pathology of leukemic mice. During the life span of the inoculated mice, no sign of pathology was detected until a few days before death. According to the Bethesda proposals for classification of neoplasms (28, 37), two groups of mice were obviously distinguishable, one presenting signs of lymphoid leukemia (thymoma and/or lymph node enlargement, no anemia) and one presenting signs of nonlymphoid leukemia (severe anemia, absence of lymph node or thymoma). The gross pathology of the leukemic mice is presented in Table 1. From these results, two major conclusions can be drawn. First, GV-1.2 induces more lymphoid leukemias than GV-1.4. Second, the genetics of the mice greatly influence the results: BALB/c mice are more susceptible to lymphoid leukemias, whereas FVB/n mice produce preferentially nonlymphoid leukemias marked by severe anemia. Eighteen percent of the mice showed signs of both lymphoid and nonlymphoid leukemias. Their mixed phenotypes were further elucidated by flow cytometry and by the analysis of the expression of some lineage-specific genes (see below). The results reflect the complexity of the leukemias induced by the Graffi retrovirus.
Flow cytometry. The immunophenotype of the leukemic cells present in the different hematopoietic organs of each mouse was analyzed using flow cytometry. The leukemic population was identified according to the enlarged size of the cells and their high proportion compared to that of normal control mice.
The results shown in Table 1 and Fig. 2A reveal that the Graffi murine retrovirus is able to induce leukemias arising from all of the hematopoietic lineages: the lymphoid lineage (B and T cells), the myeloid lineage, and the erythroid and megakaryocytic lineages. Among the 108 mice analyzed, 53 leukemias of lymphoid origin were detected; among these were 35 T-cell and 18 B-cell leukemias. The other 55 leukemias had nonlymphoid characteristics and included 9 myeloid, 15 megakaryocytic, and 31 erythroid leukemias.
In BALB/c mice, GV-1.2 induced almost exclusively T-cell leukemias (91%), whereas GV-1.4 induced also B-cell (28%), myeloid (15%), and erythroid (10%) leukemias. Similar results were obtained with the NFS and FVB/n strains, although the percentage of T-cell leukemias was reduced in these strains for both variants. The broadest spectrum of leukemias is induced with the NFS strain. These results correlate well with the observations of gross pathology in mice. They confirm that the two Graffi variants induced different patterns of leukemia and that the genetics of the mice influence the results.
The immunophenotypic analyses also confirmed the existence of complex phenotypes, either mixed or biphenotypic (28, 37). Mixed leukemias (two different leukemic populations in the organism) were exclusively erythroleukemias combined with T-cell leukemias. Biphenotypic leukemias (a unique leukemic population) were essentially lymphoid, expressing specific markers of the myeloid lineage.
Global results of the induced leukemias. To complete and confirm the gross pathology and the immunophenotypic analyses, molecular biology tools were applied to RNA and DNA isolated from each tumor (Fig. 3 and 4). For each leukemia, the flow cytometry results and the molecular characterization of the tumor RNA and DNA were compared. These global results for each type of leukemia are described below.
T-cell leukemia. The T-cell leukemias induced by the Graffi MuLV are characterized by the presence of an enlarged thymus and spleen, often accompanied by enlarged lymph nodes, with no anemia. They were identified based on four specific markers: CD3, CD90 (Thy1.2), CD4, and CD8a. All these surface molecules are expressed by thymocytes and mature lymphocytes (9, 31). A total of 31 T-cell leukemias were induced and analyzed. In BALB/c mice, GV-1.2 generated more than 90% of T-cell leukemias (Table 1), but this incidence was largely reduced in NFS (48%) and even more in FVB/n (28%) mice. For each mouse strain tested, GV-1.4 corresponded to an incidence of T-cell leukemias lower than that induced with GV-1.2 (Table 1).
Five groups of leukemias could be distinguished based on the expression of CD4 and CD8 markers: CD4– CD8–, CD4+ CD8+, CD4+ CD8lo, CD4+ CD8–, and CD4– CD8+. The two predominant forms were CD4+ CD8– and CD4+ CD8+, representing 12 and 10 cases, respectively, (Fig. 2A).
Rearrangements of the TCR gene were analyzed by Southern blotting (Fig. 3A and Table 2). Fifteen of the 31 T-cell leukemias analyzed had rearrangements of the TCR locus. Rearrangement of the TCR chain is initiated at the double-negative stage, and thus, we expected that the more immature phenotypes, CD4– CD8– (double-negative stage) and CD4+ CD8+ (double-positive stage) would be those that retained a germ line configuration. Three out of four CD4– CD8– and four out of seven CD4+ CD8+ leukemias were indeed unrearranged at the TCR locus. But in fact, 50% of the more-mature-phenotype (CD4+ CD8–) leukemias had also a TCR locus in germinal configuration. Thus, the TCR locus rearrangement and the several phenotypes observed suggest that the retrovirus does not homogeneously block the differentiation pathways of the leukemic cells. No TCR rearrangement was found in myeloid or in nonlymphoid leukemias (Fig. 3A and data not shown).
Recombination of the JH locus, which is specific to the B-cell lineage, was also tested, and atypically, we found that some T-cell leukemias also had rearrangements of this locus (Table 2). No resting B cells (less than 5%) were detected by flow cytometry in the spleen of these animals, excluding the possibility that the rearrangements observed corresponded to normal B cells (Fig. 2B, panels a and b). More curiously, other T-cell leukemias were biphenotypic and expressed the myeloid marker CD11b. This occurred in five cases, two with a CD4+ CD8– phenotype and three with a CD4– CD8– phenotype. Such a leukemia is illustrated on Fig. 2B, panel c. An aberrant abundant expression of MPO was detected in 10 tumors that were not biphenotypic (Fig. 4A, lane 10; Table 2; also data not shown), although no myeloid cells were detected by flow cytometry in the spleen of these animals. Some leukemias (7) expressed the erythroid-specific FOG gene (Fig. 4B, lane 26; also data not shown), but all of them were mixed T-cell/erythroid leukemias arising predominantly in the spleen.
B-cell leukemia. The B-cell leukemias were preferentially induced by GV-1.4 in NFS and FVB/n mice and were particularly recognizable by the presence of enormous lymph nodes and enlarged spleen but normal thymus and no anemia. They were classified as B-cell leukemias based on the presence of the surface antigen B220/CD45R, which is present from the pro-B-cell stage through the mature and activated B-cell stages, and the absence of other restricted specific lineage markers (2, 23). Some leukemias expressed progenitor markers at variable levels, and thus, six leukemias expressed Sca-1 and two expressed c-Kit (CD117). Rearrangements of the JH locus which occur at the pro-B-cell stage (2) were also analyzed. Table 2 summarizes both the immunophenotype and the molecular characteristics of the 15 B-cell leukemias. Figure 3B shows the Southern blot analysis of the heavy chain immunoglobulin (JH) gene rearrangement for the B-cell leukemias in comparison with other types of leukemias and controls. Twelve out of the 15 B-cell leukemias have rearrangements of JH (Fig. 3B, lanes 1 to 6, 9 to 10, 12, and 14 to 16). Thus, the majority of the B220+ leukemias had rearrangements of JH except for those induced in NFS mice by GV-1.2 (Table 2).
A number of these B-cell leukemias showed a complex phenotype. First and surprisingly, 8 of the 15 B-cell leukemias expressed very high levels of MPO RNA (Fig. 4A), although no myeloid cells were detected by flow cytometry as was the case for some T- cell leukemias. Second, two of them were biphenotypic since they were B220+, had rearrangements of JH, and expressed the myeloid marker CD11b, although only one expressed MPO. Surprisingly, none of the B-cell leukemias showed TCR rearrangement. Thus, we presume that the Graffi targeted B-cell progenitors are more committed than the targeted T cells. JH and TCR rearrangements may also be regulated by two separate mechanisms implicating different genes.
Myeloid leukemia. Myeloid leukemia was recognizable by an enlarged spleen, mild anemia, slightly enlarged lymph nodes and no thymus enlargement. To immunophenotype these leukemias, we used two myeloid-specific antibodies: CD11b, which is expressed on both granulocytes and monocytes (25, 29), and Gr-1, more characteristic of the granulocyte lineage (13, 29). Double-positive CD11b+ Gr-1+ cells represent myeloid precursors largely found in the bone marrow of normal mice. Nine cases of myeloid leukemia were observed. Five of these were induced in NFS mice with GV-1.4, and four were induced in BALB/c mice (three with GV-1.4 and one with GV-1.2). Three phenotypes were obtained: CD11b+ Gr-1– (five cases), CD11b– G-1+ (one case), CD11b+ Gr-1+ (three cases).
We also tested the expression of MPO, which is synthesized during the promyelocytic stage of myeloid differentiation (46). Three CD11b+ Gr-1+, one CD11b+ Gr-1–, and the unique CD11b– Gr-1+ tumor expressed very high levels of MPO (Fig. 4A, lanes 14 and 15; also data not shown). The other myeloid leukemias (mainly the Gr-1– phenotype) did not express MPO at detectable levels. Surprisingly, MPO was not only expressed in the spleens of the animals suffering from myeloid leukemias. It was detected abundantly in many other leukemia cases, as described above for the T-cell and B-cell leukemias and shown on Fig. 4A. In total, 10 T-cell, 8 B-cell, 2 erythroid, and 1 megakaryocytic leukemia showed a high expression of this specific myeloid gene even though no CD11b- or Gr-1-positive cells were detected in the spleens of these mice by flow cytometry. Interestingly, in contrast with the reports of Graffi (18), we induced only one chloroleukemia (with GV-1.4 in an NFS mouse) in this study.
Erythroid and megakaryocytic leukemia. These two types of leukemias induce severe anemia (hematocrit between 10% and 30%), an enlarged spleen, and often an infiltrated liver. The Ter119 (glycophorin A) and CD71 (transferrin receptor) cell markers were used to immunophenotype the erythroid leukemias. Ter119 and CD71 are expressed from the proerythroblast through erythrocyte stages (27, 33). Megakaryocytic leukemias were identified with the use of CD41 (glycoprotein IIb), which is specific to platelets, megakaryocytes, and early hematopoietic progenitors (57). The megakaryocytic leukemias analyzed in this study also expressed high levels of cKit (Fig. 2A, panel e) and were negative for Sca-1.
These two types of nonlymphoid leukemias developed primarily in FVB/n mice (Table 1). In this strain, we induced 72% and 36% of Ter119+ leukemias with GV-1.2 and GV-1.4, respectively, and 48% of CD41+ leukemias with GV-1.4. Complex (mixed or biphenotypic) erythromegakaryocytic leukemias were also induced, including two with GV-1.2 and two with GV-1.4. Some mixed erythroid and T-cell leukemias also developed (as described above).
Since megakaryocytic leukemias are rarely observed in murine models of retrovirus-induced leukemias and since CD41 is also expressed at the surface of multipotent stem cells, the expression of several genes highly expressed in the megakaryocytic lineage (43, 56) was analyzed by Northern blotting (Fig. 4B). As expected, the results indicate that in all the erythroid and megakaryocytic leukemias, Fog1 is expressed highly and at a much higher level than in the other types of leukemia and controls. Fli1 also is abundantly expressed in erythroid and megakaryocytic leukemias compared to the other types of leukemias and controls (Fig. 4B). These two genes cannot be used to distinguish the erythroid from the megakaryocytic tumors. Interestingly, the tumors that expressed very high levels of Fli1 also harbored a retroviral integration in the locus (not shown). GpIIb (CD41) was highly expressed in all the megakaryocytic leukemias and in mixed erythroid/megakaryocytic leukemias (Fig. 4B) in good correlation with their CD41+ immunophenotype. A few erythroid leukemias also expressed high levels of GpIIb mRNA although they were CD41–, possibly due to a poorer maturation of the protein in erythroid cells. PF4 and c-mpl are both considered to be specific to the megakaryocytic lineage (21, 62). In our study, c-mpl was found to be highly expressed in the majority of the megakaryocytic leukemias, whereas PF4 was expressed in only one case (Fig. 4B). Surprisingly, few erythroid leukemias expressed these genes (Fig. 4B). The expression of these genes in both types of leukemias was often correlated with a strong expression of GpIIb (Fig. 4B). Thus, this study enables us to conclude that the CD41/cKit leukemic cells derive from the megakaryocytic lineage. Furthermore, it shows that the erythroid and megakaryocytic leukemias induced by Graffi are very closely related.
Integrity of the U3 enhancer region. The U3 region of the proviruses of mature tumors was analyzed to detect possible rearrangements or duplications in the enhancer region, as it was found in previous studies with other MuLVs and especially with SRS-19.6 (20).
We PCR amplified the proviral U3 enhancer region from the genomic DNA of infiltrated spleens of leukemic mice. GV-1.2 has a 60-bp duplication of the enhancer region absent in GV-1.4, and therefore the number of duplication can be deduced from the fragment's size: 351 bp for GV-1.2 and 291 bp for GV-1.4 (Fig. 5). We hybridized the PCR products with a U3-specific probe to confirm the specificity of the amplifications (data not shown). The results indicate that the Graffi MuLV enhancer is not stable and acquired modifications during the replication cycles. As shown on Fig. 5, several fragments of different size are detected for many tumors. For each tumor, the most abundant fragment generally corresponds to the input viral variant (GV-1.2 and GV-1.4), and the additional fragments correspond to altered proviruses. The smallest fragment corresponded to the intact GV-1.4 form, which is therefore the smallest enhancer region necessary for efficient replication.
In the lymphoid leukemias, some fragments were of a size larger than that expected from the input retrovirus (Fig. 5A and B). PCR from GV-1.4 tumors all showed the expected 291-bp fragment but also presented fragments of larger size corresponding to enhancer modification, presumably duplications and triplications. Also, PCRs for GV-1.2 tumors all showed the expected 351-bp fragment and additional larger fragments (Fig. 5A and B). Interestingly, a few GV-1.2 tumors contained the lower-molecular-weight 291-bp fragment typical of GV-1.4 (Fig. 5A, lane 3, and B, lanes 2 to 4).
The phenomenon was more rarely observed in the nonlymphoid leukemias, since they do not show duplication or multimerization of the enhancer region (Fig. 5C and D). All the nonlymphoid tumors induced with GV-1.4 (Fig. 5C, lanes 1 to 13) showed no alteration of the provirus enhancer region; lanes 2 and 8 on Fig. 5C show mixed nonlymphoid and T-cell leukemias. However, GV-1.2-induced erythroid leukemias (Fig. 5D) often presented enhancer alteration with a majority of additional smaller fragments corresponding to the GV-1.4 enhancer (Fig. 5D, lanes 3 and 5 and 6 to 8). Thus, the constitution of the enhancer region can be altered during the viral replication cycles, and this surely has an impact on tumor development.
DISCUSSION
This study on the pathogenesis of the Graffi murine retrovirus indicates that it is an extremely multipotent virus, able to give rise to leukemias from every lineage of the hematopoietic system: T-cell, B-cell, myeloid, erythroid, and even megakaryocytic lineages. However, GV-1.2 and GV-1.4 induce the disease with different latencies, and more importantly, they induce different types of leukemia. GV-1.2 induces a large percentage of T-cell leukemias, although GV-1.4 induces a much wider spectrum of leukemias, including nonlymphoid leukemias.
Graffi is extremely multipotent and induces complex leukemias. The induced leukemias were found to be heterogeneous. The leukemia from each animal harbored its particularities, and several leukemias showed complex phenotypes. This heterogeneity reflects the very high polyvalence of the Graffi MuLV. In the case of T-cell leukemias, some tumors were more complex and did not show the expected phenotypes. The TCR rearrangement was in several cases not correlated with a more mature phenotype, and more curiously, some leukemias harbored an immunoglobulin H rearrangement. This phenomenon was already observed in leukemias induced by Moloney viral recombinants and in thymomas in AKR/J mice (20, 22, 24, 40). Possibly, the leukemic T cells derive from very early lymphoid progenitors that still retain some B-cell potential, and the deregulation induced by the viral integration may activate the machinery of the immunoglobulin rearrangement. Some leukemias showed much more complex phenotypes that could be considered aberrant. First, these are the biphenotypic lymphoid leukemias expressing the CD11b myeloid marker. Corresponding human leukemias are widely reported in the literature (for examples, see references 7, 16, 38, 44, 49, and 51), and expression of CD11b on human leukemic B cells is associated with unfavorable prognosis (26, 45, 55). Some studies showed that these biphenotypic cells do exist in normal mice although they are very rare (1, 5, 17, 34). Second, this concerns also the apparently aberrant expression of MPO mainly in T-cell and B-cell leukemias but also in two erythroid and one megakaryocytic case. Since RNA was extracted from whole spleen extracts, the high MPO level can be explained in two ways. First, it could be due to a large infiltration of adherent myeloid cells (histiocytes) that were not included in the flow cytometry analysis since the cell suspension was obtained by mechanical disruption without collagenase treatment. Second, the nonmyeloid leukemic cells could exhibit promiscuous gene expression of MPO as a consequence of retroviral insertional activation of this gene. We favor the latter possibility because we could not find any signs of myeloid cell infiltration in the tumors by morphological assessment of the samples (spleen imprints). This highly aberrant expression of MPO in many leukemias is probably the main feature responsible for the higher percentage of tumors classified as myeloid leukemias observed by Ru et al. (48), who used the MPO staining as an important criterion.
Thus, the high versatility of Graffi MuLV suggests that the target cells are either very early or early committed progenitors from all the lineages. Indeed, when methylcellulose colony assays were performed with bone marrow cells from infected mice, Graffi proviral DNA could be detected in every type of colony tested (CFU-G, CFU-GM, CFU-M, CFU-E, CFU-GEMM) as early as 4 days postinfection, and by 15 days postinfection, all colonies were highly positive (data not shown).
Importance of mouse genetics. Interestingly, each strain of mice responded differently to Graffi infection. First, the latency was different among the different strains, and second, the distribution of the various leukemias was specific to each strain of mice. BALB/c mice showed the longest latency period and preferentially developed T-cell leukemias. The NFS strain was more susceptible to B-cell leukemias. Finally, the FVB/n strain showed the shortest latency and developed mainly nonlymphoid leukemias. Interestingly, this strain of mice carries an activated form of the K-ras oncogene (52) which could reduce the number of hits required for leukemogenesis.
Similar observations were made in previous studies. For example, a Moloney MuLV with a mutated core motif was shown to induce 60% of erythroleukemia cases in NFS mice but almost 100% of T-cell lymphoma cases in BALB/c mice (61). Similarly, the recombinant virus MOL4070LTR (Moloney with the 4070A LTR) induces 46% of lymphomas in BALB/c mice and 23% in FVB/n mice (60).
Importance of the retroviral LTR. The importance of the retroviral LTR in disease specificity is clearly established. The Graffi retroviruses are capable to infect a very large spectrum of cell types, and the major determinant is very likely due to its LTR. Indeed, we analyzed the Graffi LTR-driven transcriptional activity in several cell lines (4). The U3 region of GV-1.2 was found to be a very strong promoter in all the hematopoietic cell lines tested (erythroid, myeloid, and T-lymphoid). GV-1.4 LTR was globally less powerful than GV-1.2 but showed a similar wide-range specificity. Thus, the Graffi U3 region could bind transcription factors present in different cell lineages and be equally activated in these cells, in contrast to Friend and Moloney viruses, which show enhanced activity only in erythroid and T-cell lines, respectively (4).
Indeed, DNase I footprint analyses using the LTR of GV-1.4 revealed a similar pattern of protected regions in different cell lines (erythroid, myeloid, and T-cell) (4). Some of the LTR's specific transcription factors were identified for Graffi, and we showed that GATA-1, -2 and -3 were able to activate the promoters of both GV-1.2 and GV-1.4 (3). GATA-1 is expressed in all erythroid cell lineages, megakaryocytes, mast cells, and multipotent stem cells and has been involved in the regulation of most erythroid genes (39). GATA-3 is expressed at all stages of T-cell development (14, 39). GATA-2 has a much broader expression pattern, including myeloid and erythroid cell lineages and endothelial cells, and is essential for multilineage hematopoiesis (39). Thus, the importance of the GATA members in the Graffi LTR activation can explain the large spectrum of infected cells. These proteins act in cooperation with other cofactors, such as Tal1/Scl (30). Interestingly, a perfect consensus site for TAL1 is present proximal to the GATA binding site. It would therefore be noteworthy to identify the factors interacting with the GATA members in different cell lines.
An interesting feature that became evident in this study is that the viral variants GV-1.2 and GV-1.4 have some differences in latency and type of induced leukemia. The latency period of GV-1.2 is shorter than that of GV-1.4 (Fig. 1) (48), and GV-1.2 induces a higher percentage of T-cell leukemias (Table 1). We sequenced the two genomes, GV-1.2 and GV-1.4, which originate from the same tumor, and we found that they are very similar except for the duplication in the U3 LTR region (GenBank/EMBL/DDBJ accession numbers: GV-1.4, AB187566; GV-1.2, AB187565) (V. Voisin et al., in preparation). Except for the U3 region, only 16 nucleotide differences were found along their entire genomes.
The differences in latency are not likely to be influenced by viral titer since the two variants showed equivalent titers. These variations in latency are more likely explained by the presence of two direct repeats in the GV-1.2 LTR. Probably, this allows the binding of more transcription factors, making it globally more transcriptionally powerful as shown previously (4).
GV-1.2 induces a higher percentage of T-cell leukemias. The direct repeat encompasses the well-studied region of the LVb/CORE binding sites. The CBF and MCREF bind the CORE of Moloney in T-cells and Friend in erythroid cells, respectively (15, 32, 35, 53). Therefore, the duplicated region contains elements that are known to be more important for activation in T cells and also erythroid cells, which correspond to the major types of leukemia induced by GV-1.2. In FVB/n mice, a high incidence (72%) of erythroleukemia is induced with a very short latency compared to that in NFS and BALB/c mice. This suggests that the duplication has a strong impact on promoter activity in erythroid cells in this particular strain.
Granger et al. (20) hypothesize that the tandemerization of the enhancer region has a favorable impact on the development of T-cell leukemias. Interestingly, they found that the proviruses from lymphoid tumors induced by a SRS-19.6-Moloney recombinant virus had acquired new sequence duplication in the enhancer region. Similar sequence duplications were also observed in wild-type SRS-19.6-induced tumors and were restricted to T-cell tumors. Considering the high homology between GV-1.4 and SRS-19.6 LTRs, we verified the integrity of the proviral enhancers in mature tumors. The analysis revealed that the Graffi MuLV enhancer is not stable and fluctuates between different forms during the replication cycles, showing duplication, multimerization, and even loss of duplication (Fig. 5). One could hypothesize that these alterations are favorable to tumorigenesis. Granger et al. (20) proved that the tandemerization of the enhancer region of SRS-19.6 was associated with T-cell leukemias but not with other leukemias. We can conclude from this study that a multimerization of the enhancer region seems to be indeed related with lymphoid leukemia. Indeed, a much higher percentage of lymphoid leukemias is generated by GV-1.2, which contains an enhancer duplication, and most of the alterations observed with this variant yield fragments of higher molecular weight. Moreover, the minimal enhancer corresponding to the GV-1.4 form seems more favorable to the induction of nonlymphoid leukemias since most leukemias of this kind arose after the inoculation of GV-1.4. In contrast with the lymphoid leukemias, nonlymphoid leukemias do not show duplication or multimerization of the enhancer (Fig. 5C). Furthermore, some GV-1.2-induced erythroid leukemias showed a loss of duplication (Fig. 5D, lanes 3, 5, and 6 to 8).
Since GV-1.2 shows a shorter latency period in addition to inducing more T-cell leukemias, one could speculate that these leukemias develop faster than the nonlymphoid ones. This was not the case, and there was no correlation between the latency period and the type of leukemia observed in each individual litter (data not shown).
Comparison with SRS-19-6 MulV. The sequencing of the Graffi genomes revealed an overall homology of 97% between the Graffi and SRS-19.6 MuLVs, demonstrating that they must originate from a common ancestor. However, there are enough differences to suggest that they represent distinct viruses. The nucleotide changes are dispersed along the genomes even in well-conserved regions. Two segments (the 5'end of the POL gene and the 3'end of the ENV gene) are the most divergent regions between Graffi and SRS-19.6 (Voisin et al., in preparation). The SRS-19.6 virus was cloned from a transmissible system developed in China and involving passaging of the viral mixture in Kunming mice, an inbred strain developed from NIH/Swiss mice (6). This virus is able to induce a large spectrum of leukemias in NIH/Swiss mice (myeloid leukemias, B- and T-cell lymphomas, and erythroid leukemias). This distribution is very similar to our results, in particular those obtained with the GV-1.4 variant inoculated into NFS mice (Table 1). However, different strains of mice were used in these two studies. As the mouse genetic background appears to be very important, the two viruses could still give slightly different phenotypes if inoculated into the same strain of mice, reflecting the differences found in their respective genomes.
A good model for megakaryocytic leukemia. A high percentage of nonlymphoid leukemias were induced with GV-1.4, especially in the FVB/n strain. They consist mostly of erythroid (TER119+ CD71+) and megakaryocytic leukemias (CD41+ cKit+). Some MuLVs, such as Friend and Rauscher, are known to induce erythroid leukemias, but very few studies on MuLVs have reported megakaryocytic leukemia. Thus, Graffi MuLV seems to be a good model to study megakaryocytes. Some specific genes of megakaryocytes and eythroid lineages were analyzed to confirm the true lineage of these leukemias (Fig. 4B). The expression of these genes did not show any erythroid versus megakaryocytic specificity or vice versa. This demonstrates the closed relationship of the two lineages that is noteworthy and widely reported in the literature. Several examples of human erythroid leukemic cell lines from patients (47, 54) or from MuLV models (41, 58) are reported to bear both erythrocyte and megakaryocyte characteristics. Moreover, a common erythromegakaryocytic progenitor, bearing the surface molecules TER119 and CD41, was identified in the bone marrow (36, 42, 59). This suggests that the TER119+ CD41+ leukemias observed mostly in GV-1.4-infected FVB/n mice may correspond to this progenitor and therefore represents an interesting model to study the erythromegakaryocytic lineage.
Graffi: a myeloid leukemia-inducing virus All together, these results show that both Graffi variants, GV-1.2 and GV-1.4, can induce a variety of leukemias in BALB/c, NFS, and FVB/n strains of mice. This exhaustive study shows that Graffi MuLV induces complex leukemias and that several immunological and molecular markers are required to classify them properly. These results are different from our previous study on the characterization of Graffi MuLV (48), where leukemias were considered mainly myeloid. Our earlier study (48) was mainly based on the observations of blood smears, spleen imprints, MPO staining, and some molecular biology analysis. We described these cases as myeloid leukemias with lineage infidelities since they also contained either immunoglobulin or TCR gene rearrangements or both. In the present study, the high levels of MPO expression in the spleen, often in lymphoid leukemias, are probably caused by aberrant gene expression.
Erkeland et al. (10) have described Graffi MuLV as inducing a majority of myeloid leukemias in the FVB/n strain of mice. The virus also induces a fair proportion of other tumor types. This discrepancy in results could be interpreted by a different classification approach and by the use of different antibodies. We followed the Bethesda proposal (28, 37) to classify the leukemias. In comparison with their study, we did not use the surface antigen F4/80, as we found it redundant with CD11b (Mac-1), nor did we use ERMP-58, also specific to the myeloid lineage (8). We did phenotype some of the leukemias with ERMP-58 (gift of P. Leenen) (data not shown): it gave high positivity for all the myeloid leukemias, especially the CD11b+ Gr-1+ phenotype. We tested three additional antibodies: c-Kit, CD34, and CD41. This last antibody allowed the characterization of the 48% megakaryocytic leukemias found in GV-1.4-infected FVB/n mice. Finally, we cannot rule out a possible dissimilarity in the FVB/n mouse strain origin or viral isolate.
The Graffi-induced leukemias are now well characterized and can be used as a model to study hematopoiesis and leukemic progression in mice. It is now established that SRS 19-6 and Graffi are tightly related viruses and have a common ancestor. To date, as far as we know, the high incidence of megakaryocytic leukemias induced with the inoculation of GV-1.4 into the FVB/n strain of mice was never described with any other MuLVs. The results indicate also a tight relationship between the erythroid and megakaryocytic lineages, and further studies would certainly increase the knowledge about these two hematopoietic lineages.
ACKNOWLEDGMENTS
We thank Denis Flipo for his help in flow cytometry. We thank Sonia Do Carmo and Philippe Legault for their help in sequencing the two Graffi genomes.
This work was supported by grant MOP-37994 from the Canadian Institutes of Health Research and by grant 3279-01 from the Natural Sciences Engineering Council of Canada.
REFERENCES
Abrahao, T. B., E. Freymuller, R. A. Mortara, J. D. Lopes, and M. Mariano. 2003. Morphological characterization of mouse B-1 cells. Immunobiology 208:401-411.
Allman, D., J. Li, and R. R. Hardy. 1999. Commitment to the B lymphoid lineage occurs before DH-JH recombination. J. Exp. Med. 189:735-740.
Barat, C., and E. Rassart. 1998. Members of the GATA family of transcription factors bind to the U3 region of Cas-Br-E and Graffi retroviruses and transactivate their expression. J. Virol. 72:5579-5588.
Barat, C., and E. Rassart. 1998. Nuclear factors that bind to the U3 region of two murine myeloid leukemia-inducing retroviruses, Cas-Br-E and Graffi. Virology 252:82-95.
Borrello, M. A., and R. P. Phipps. 1995. Fibroblasts support outgrowth of splenocytes simultaneously expressing B lymphocyte and macrophage characteristics. J. Immunol. 155:4155-4161.
Bundy, L. M., M. Ru, B. F. Zheng, L. Cheng, P. K. Pattengale, J. L. Portis, et al. 1995. Biological characterization and molecular cloning of murine C-type retroviruses derived from the TSZ complex from mainland China. Virology 212:367-382.
Butcher, B. W., K. S. Wilson, S. H. Kroft, R. H. Collins, Jr., and V. Bhushan. 2005. Acute leukemia with B-lymphoid and myeloid differentiation associated with an inv(5)(q13q33) in an adult patient. Cancer Genet. Cytogenet. 157:62-66.
de Bruijn, M. F., R. E. Ploemacher, A. E. Mayen, J. S. Voerman, W. A. Slieker, W. van Ewijk, et al. 1996. High-level expression of the ER-MP58 antigen on mouse bone marrow hematopoietic progenitor cells marks commitment to the myeloid lineage. Eur. J. Immunol. 26:2850-2858.
Dialynas, D. P., Z. S. Quan, K. A. Wall, A. Pierres, J. Quintans, M. R. Loken, et al. 1983. Characterization of the murine T cell surface molecule, designated L3T4, identified by monoclonal antibody GK1.5: similarity of L3T4 to the human Leu-3/T4 molecule. J. Immunol. 131:2445-2451.
Erkeland, S. J., M. Valkhof, C. Heijmans-Antonissen, A. van Hoven-Beijen, R. Delwel, M. H. A. Hermans, and I. P. Touw. 2004. Large-scale identification of disease genes involved in acute myeloid leukemia. J. Virol. 78:1971-1980.
Fey, F. 1969. Studies on the pathogenesis and mechanism of hematologic diversification by re-isolation of the myeloid leukemia virus (Graffi). Acta Haematol. 42:65-75.
Fey, F., and A. Graffi. 1965. Erythroblastic leukemia after the injection of mouse myelogenous leukemia virus. Z. Krebsforsch. 67:145-151. (In German.)
Fleming, T. J., M. L. Fleming, and T. R. Malek. 1993. Selective expression of Ly-6G on myeloid lineage cells in mouse bone marrow. RB6-8C5 mAb to granulocyte-differentiation antigen (Gr-1) detects members of the Ly-6 family. J. Immunol. 151:2399-2408.
Georgopoulos, K. 1997. Transcription factors required for lymphoid lineage commitment. Curr. Opin. Immunol. 9:222-227.
Golemis, E., Y. Li, T. N. Fredrickson, J. W. Hartley, and N. Hopkins. 1989. Distinct segments within the enhancer region collaborate to specify the type of leukemia induced by nondefective Friend and Moloney viruses. J. Virol. 63:328-337.
Gonzalez Garcia, J. R., S. K. Bohlander, M. Gutierrez Angulo, M. A. Esparza Flores, V. J. Picos Cardenas, J. P. Meza Espinoza, et al. 2004. A t(1;9)(q23.3 approximately q25;q34) affecting the ABL1 gene in a biphenotypic leukemia. Cancer Genet. Cytogenet. 152:81-83.
Graf, B. A., D. A. Nazarenko, M. A. Borrello, L. J. Roberts, J. D. Morrow, J. Palis, et al. 1999. Biphenotypic B/macrophage cells express COX-1 and up-regulate COX-2 expression and prostaglandin E(2) production in response to pro-inflammatory signals. Eur. J. Immunol. 29:3793-3803.
Graffi, A. 1957. Chloroleukemia of mice. Ann. N. Y. Acad. Sci. 68:540-558.
Graffi, A., F. Fey, and T. Schramm. 1966. Experiments on the hematologic diversification of viral mouse leukemias. Natl. Cancer Inst. Monogr. 22:21-31.
Granger, S. W., L. M. Bundy, and H. Fan. 1999. Tandemization of a subregion of the enhancer sequences from SRS 19-6 murine leukemia virus associated with T-lymphoid but not other leukemias. J. Virol. 73:7175-7184.
Gurney, A. L., K. Carver-Moore, F. J. de Sauvage, and M. W. Moore. 1994. Thrombocytopenia in c-mpl-deficient mice. Science 265:1445-1447.
Hanecak, R., P. K. Pattengale, and H. Fan. 1988. Addition of substitution of simian virus 40 enhancer sequences into the Moloney murine leukemia virus (M-MuLV) long terminal repeat yields infectious M-MuLV with altered biological properties. J. Virol. 62:2427-2436.
Hardy, R. R., Y. S. Li, D. Allman, M. Asano, M. Gui, and K. Hayakawa. 2000. B-cell commitment, development and selection. Immunol. Rev. 175:23-32.
Herr, W., A. P. Perlmutter, and W. Gilbert. 1983. Monoclonal AKR/J thymic leukemias contain multiple JH immunoglobulin gene rearrangements. Proc. Natl. Acad. Sci. USA 80:7433-7436.
Ho, M. K., and T. A. Springer. 1982. Mac-1 antigen: quantitative expression in macrophage populations and tissues, and immunofluorescent localization in spleen. J. Immunol. 128:2281-2286.
Jensen, G. S., A. R. Belch, F. Kherani, M. J. Mant, B. A. Ruether, and L. M. Pilarski. 1992. Restricted expression of immunoglobulin light chain mRNA and of the adhesion molecule CD11b on circulating monoclonal B lineage cells in peripheral blood of myeloma patients. Scand. J. Immunol. 36:843-853.
Kina, T., K. Ikuta, E. Takayama, K. Wada, A. S. Majumdar, I. L. Weissman, et al. 2000. The monoclonal antibody TER-119 recognizes a molecule associated with glycophorin A and specifically marks the late stages of murine erythroid lineage. Br. J. Haematol. 109:280-287.
Kogan, S. C., J. M. Ward, M. R. Anver, J. J. Berman, C. Brayton, R. D. Cardiff, et al. 2002. Bethesda proposals for classification of nonlymphoid hematopoietic neoplasms in mice. Blood 100:238-245.
Lagasse, E., and I. L. Weissman. 1996. Flow cytometric identification of murine neutrophils and monocytes. J. Immunol. Methods 197:139-150.
Lecuyer, E., and T. Hoang. 2004. SCL: from the origin of hematopoiesis to stem cells and leukemia. Exp. Hematol. 32:11-24.
Ledbetter, J. A., R. V. Rouse, H. S. Micklem, and L. A. Herzenberg. 1980. T cell subsets defined by expression of Lyt-1,2,3 and Thy-1 antigens. Two-parameter immunofluorescence and cytotoxicity analysis with monoclonal antibodies modifies current views. J. Exp. Med. 152:280-295.
Li, Y., E. Golemis, J. W. Hartley, and N. Hopkins. 1987. Disease specificity of nondefective Friend and Moloney murine leukemia viruses is controlled by a small number of nucleotides. J. Virol. 61:693-700.
Lok, C. N., and T. T. Loh. 1998. Regulation of transferrin function and expression: review and update. Biol. Signals Recept. 7:157-178.
Lu, M., H. Kawamoto, Y. Katsube, T. Ikawa, and Y. Katsura. 2002. The common myelolymphoid progenitor: a key intermediate stage in hemopoiesis generating T and B cells. J. Immunol. 169:3519-3525.
Manley, N. R., M. O'Connell, W. Sun, N. A. Speck, and N. Hopkins. 1993. Two factors that bind to highly conserved sequences in mammalian type C retroviral enhancers. J. Virol. 67:1967-1975.
McDonald, T. P., and P. S. Sullivan. 1993. Megakaryocytic and erythrocytic cell lines share a common precursor cell. Exp. Hematol. 21:1316-1320.
Morse, H. C., III, M. R. Anver, T. N. Fredrickson, D. C. Haines, A. W. Harris, N. L. Harris, et al. 2002. Bethesda proposals for classification of lymphoid neoplasms in mice. Blood 100:246-258.
Neame, P. B., P. Soamboonsrup, G. Browman, R. D. Barr, N. Saeed, B. Chan, et al. 1985. Simultaneous or sequential expression of lymphoid and myeloid phenotypes in acute leukemia. Blood 65:142-148.
Orkin, S. H. 1992. GATA-binding transcription factors in hematopoietic cells. Blood 80:575-581.
Ott, D. E., J. Keller, K. Sill, and A. Rein. 1992. Phenotypes of murine leukemia virus-induced tumors: influence of 3' viral coding sequences. J. Virol. 66:6107-6116.
Paoletti, F., A. M. Vannucchi, A. Mocali, R. Caporale, and S. A. Burstein. 1995. Identification and conditions for selective expression of megakaryocytic markers in Friend erythroleukemia cells. Blood 86:2624-2631.
Papayannopoulou, T., M. Brice, D. Farrer, and K. Kaushansky. 1996. Insights into the cellular mechanisms of erythropoietin-thrombopoietin synergy. Exp. Hematol. 24:660-669.
Pereira, R., C. T. Quang, I. Lesault, H. Dolznig, H. Beug, and J. Ghysdael. 1999. FLI-1 inhibits differentiation and induces proliferation of primary erythroblasts. Oncogene 18:1597-1608.
Perentesis, J., N. K. Ramsay, R. Brunning, J. H. Kersey, and A. H. Filipovich. 1983. Biphenotypic leukemia: immunologic and morphologic evidence for a common lymphoid-myeloid progenitor in humans. J. Pediatr. 102:63-67.
Pinto, A., L. Del Vecchio, A. Carbone, M. Roncadin, R. Volpe, and D. Serraino. 1991. Expression of myelomonocytic antigens is associated with unfavourable clinicoprognostic factors in B-cell chronic lymphocytic leukaemia. Ann. Oncol. 2(Suppl. 2):107-113.
Rosmarin, A. G., S. C. Weil, G. L. Rosner, J. D. Griffin, M. A. Arnaout, and D. G. Tenen. 1989. Differential expression of CD11b/CD18 (Mo1) and myeloperoxidase genes during myeloid differentiation. Blood 73:131-136.
Rowley, P. T., B. A. Farley, S. LaBella, R. Giuliano, and J. F. Leary. 1992. Single K562 human leukemia cells express and are inducible for both erythroid and megakaryocytic antigens. Int. J. Cell Cloning 10:232-240.
Ru, M., C. Shustik, and E. Rassart. 1993. Graffi murine leukemia virus: molecular cloning and characterization of the myeloid leukemia-inducing agent. J. Virol. 67:4722-4731.
Rubio, M. T., N. Dhedin, C. Boucheix, J. H. Bourhis, O. Reman, J. M. Boiron, et al. 2003. Adult T-biphenotypic acute leukaemia: clinical and biological features and outcome. Br. J. Haematol. 123:842-849.
Sambrook, J., and D. W. Russell. 2001. Molecular cloning: a laboratory manual, 3rd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
Scamurra, D. O., F. R. Davey, D. A. Nelson, A. S. Kurec, and J. Goldberg. 1983. Acute leukemia presenting with myeloid and lymphoid cell markers. Ann. Clin. Lab. Sci. 13:496-502.
Shafarenko, M., J. Mahler, C. Cochran, A. Kisielewski, E. Golding, R. Wiseman, et al. 1997. Similar incidence of K-ras mutations in lung carcinomas of FVB/N mice and FVB/N mice carrying a mutant p53 transgene. Carcinogenesis 18:1423-1426.
Speck, N. A., and D. Baltimore. 1987. Six distinct nuclear factors interact with the 75-base-pair repeat of the Moloney murine leukemia virus enhancer. Mol. Cell. Biol. 7:1101-1110.
Tani, T., J. Ylanne, and I. Virtanen. 1996. Expression of megakaryocytic and erythroid properties in human leukemic cells. Exp. Hematol. 24:158-168.
Tassies, D., E. Montserrat, J. C. Reverter, N. Villamor, M. Rovira, and C. Rozman. 1995. Myelomonocytic antigens in B-cell chronic lymphocytic leukemia. Leuk. Res. 19:841-848.
Tsang, A. P., J. E. Visvader, C. A. Turner, Y. Fujiwara, C. Yu, M. J. Weiss, et al. 1997. FOG, a multitype zinc finger protein, acts as a cofactor for transcription factor GATA-1 in erythroid and megakaryocytic differentiation. Cell 90:109-119.
Uzan, G., M. Prenant, M. H. Prandini, F. Martin, and G. Marguerie. 1991. Tissue-specific expression of the platelet GPIIb gene. J. Biol. Chem. 266:8932-8939.
Vannucchi, A. M., S. Linari, C. Cellai, M. J. Koury, and F. Paoletti. 1997. Constitutive and inducible expression of megakaryocyte-specific genes in Friend erythroleukaemia cells. Br. J. Haematol. 99:500-508.
Vannucchi, A. M., F. Paoletti, S. Linari, C. Cellai, R. Caporale, P. R. Ferrini, et al. 2000. Identification and characterization of a bipotent (erythroid and megakaryocytic) cell precursor from the spleen of phenylhydrazine-treated mice. Blood 95:2559-2568.
Wolff, L., R. Koller, X. Hu, and M. R. Anver. 2003. A Moloney murine leukemia virus-based retrovirus with 4070A long terminal repeat sequences induces a high incidence of myeloid as well as lymphoid neoplasms. J. Virol. 77:4965-4971.
Yuen, P. H., Y. H. Khang, and C. Matherne. 1992. Virulence in lymphomagenesis is modulated by alterations in proteins binding to CORE and GRE-LVa elements of the MoMuLV enhancer. Leukemia 6(Suppl. 3):76S-82S.
Zhang, C., M. A. Thornton, M. A. Kowalska, B. S. Sachis, M. Feldman, M. Poncz, et al. 2001. Localization of distal regulatory domains in the megakaryocyte-specific platelet basic protein/platelet factor 4 gene locus. Blood 98:610-617.(Veronique Voisin, Corinne)