INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Moloney murine leukemia virus (M-MuLV) is a simple retrovirus that induces T lymphoma in susceptible mice. Leukemogenesis by M-MuLV has been studied extensively (reviewed in reference 6). It has become clear that it is a multistep process, with several well-defined events taking place in an orderly fashion. Two well-recognized events include the insertional activation of proto-oncogenes and the generation of polytropic envelope recombinants (MCF recombinants [7]) in the infected animal.

We have employed an enhancer variant of M-MuLV, Mo+PyF101 M-MuLV, to study M-MuLV leukemogenesis in mice. This virus contains enhancer sequences from the F101 strain of murine polyoma virus inserted into the M-MuLV long terminal repeat downstream of the M-MuLV enhancers (11). Mo+PyF101 M-MuLV shows substantially attenuated leukemogenicity when inoculated subcutaneously (s.c.) into newborn mice (3, 5). Comparative studies with Mo+PyF101 and wild-type M-MuLV have identified a series of preleukemic events induced by wild-type M-MuLV, notably hematopoietic hyperplasia in the spleen. The splenic hyperplasia appears to result secondarily from stromal defects in the bone marrow (10). The leukemogenic defect of Mo+PyF101 M-MuLV is also dependent on the route of inoculation. When Mo+PyF101 M-MuLV is inoculated s.c. it shows attenuation; when inoculated intraperitoneally (i.p.) it shows leukemogenicity equivalent to that of wild-type M-MuLV (1). Comparative studies of mice inoculated s.c. and i.p. with Mo+PyF101 M-MuLV provided further insight (1). The rate of infection for Mo+PyF101 M-MuLV in the thymus (the ultimate target organ for M-MuLV leukemogenesis) did not differ between s.c. and i.p. inoculation. On the other hand, early infection (1 to 2 weeks) in the bone marrow was substantially reduced in mice infected s.c. with Mo+PyF101 M-MuLV compared to those infected by the i.p. route. This indicated that early bone marrow infection is essential for efficient leukemogenesis by M-MuLV. One possibility is that the bone marrow might seed infection to lymphoid precursors that subsequently migrate to the thymus.

In light of the identification of the bone marrow as a critical target for M-MuLV infection, we were interested in a more detailed characterization of early bone marrow infection. In particular, we were interested in identifying the cell types that become infected and the order in which this occurs. To identify the first cells infected, we employed a replication-defective M-MuLV-based retroviral vector (BAG) that expresses a readily detectable reporter gene, the gene that encodes bacterial -galactosidase. In vivo infection with this vector allows the identification of cells that are directly infected by the injected virus, since the vector cannot spread to other cells. Infection of the bone marrow after i.p. inoculation with an M-MuLV-based retroviral vector is characterized in this report.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Viruses and inoculation of mice. Psi-2 cells (12) were transfected by a plasmid containing the BAG vector (kindly provided by Constance Cepko [15]) and were selected for the presence of the vector by growth in a medium containing G418. G418-resistant cells were grown in Dulbecco modified Eagle's medium (DMEM) supplemented with 10% calf serum as described previously (15). The cell culture supernatant was harvested and concentrated 10- to 20-fold by ultrafiltration with an Amicon Centriprep 50 (Amicon Inc., Beverly, Mass.). To titrate viral supernatants, serial dilutions were used to infect NIH/3T3 cells that had been pretreated for 1 h with 20 µg of Polybrene per ml. Cells were allowed to grow until confluence and were stained for -galactosidase activity as described below, and the number of blue colonies was counted. Viral vector titers of 2 × 10⁶ to 6 × 10⁶ infectious units/ml were routinely obtained.

Cell staining and immunofluorescence. -Galactosidase activity in cells was detected by staining with X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) as described by Price et al. (15). Briefly, after cytocentrifugation onto slides, cells were fixed with 0.5% paraformaldehyde in PBS and then stained with an X-Gal reaction mixture containing 1 mg of X-Gal (Gibco) per ml, 15 mM potassium ferrocyanide, 15 nM potassium ferricyanide, and 1 mM MgCl₂. Staining was performed at 37°C for 6 to 8 h.

Osteoclast cultures. To promote osteoclast outgrowth, cells flushed from mouse femurs were grown on glass coverslips in 24-well plates for 7 to 10 days at 37°C in DMEM containing 10% fetal bovine serum and 10 nM 1,25-dihydroxycholicalciferol as described by Hata et al. (8).

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

To identify the first cells infected by M-MuLV in the bone marrow of neonatal mice, a replication-defective M-MuLV-based retroviral vector was used. The BAG vector developed by Price et al. (15) expresses bacterial -galactosidase as a fusion with the M-MuLV Gag protein (Fig. 1). When the BAG vector plasmid is transfected into Psi-2 packaging cells (15), the resulting virus particles consist of M-MuLV proteins with BAG vector RNA. The BAG vector produced by these cells would infect the same cells that wild-type M-MuLV does, but it would not spread past the initially infected cells, since the vector does not encode all of the M-MuLV proteins necessary for productive infection.

View larger version (6K): [in this window] [in a new window]   FIG. 1.   BAG vector. The organization of the BAG vector (15) in its proviral form is shown. The vector contains the bacterial -galactosidase gene (lac z) and a selectable marker (the bacterial neomycin phosphotransferase gene [neor] under the control of the simian virus 40 [SV40] promoter). For these experiments, the plasmid DNA containing the BAG vector sequences was transfected into Psi-2 packaging cells. LTR, long terminal repeat; pBR ori, origin of replication.

BAG vector was prepared by harvesting the supernatant from stably transfected Psi-2/BAG cells, followed by the clarification and concentration procedures described in Materials and Methods. Infectious agent titers of 2 × 10⁶ to 6 × 10⁶/ml were obtained. The BAG vector stocks were free of replication-competent M-MuLV, as measured by XC syncytial plaque assays.

Neonatal NIH/Swiss mice were inoculated i.p. with BAG vector (ca. 10⁶ infectious units/animal). This route of infection is a standard way of inducing leukemia by M-MuLV and one that leads to early high-level bone marrow infection in M-MuLV-infected mice. At various times postinfection (2 to 12 days), the animals were sacrificed and bone marrow flushes from their femurs were harvested. Single-cell suspensions (2 × 10⁵ cells) were deposited onto microscope slides by cytocentrifugation and then stained with X-Gal to detect infected cells (which were stained blue and were visible under light microscopy). Within 2 days postinfection, blue cells that were either large or small could be detected (Fig. 2A and B). The small BAG-infected cells were morphologically indistinct and resembled the majority of cells in the bone marrow. The large BAG-infected cells were irregular in shape after cytocentrifugation and were either mono- or multinucleate. Based on their size, the large cells could be either large tissue macrophages, megakaryocytes, or osteoclasts. When bone marrow from age-matched uninfected mice was analyzed in parallel, no blue cells were evident after X-Gal staining.

View larger version (157K): [in this window] [in a new window]   FIG. 2.   X-Gal staining of BAG-infected bone marrow cells. Neonatal mice were infected i.p. with 5 × 105 infectious units of BAG vector. Six days postinfection, the animals were sacrificed and single-cell suspensions from the bone marrow were deposited on microscope slides by cytocentrifugation. The slides were then stained with X-Gal with or without secondary histochemical stains. (A and B) X-Gal staining alone, showing large and small BAG-infected cells (the blue stain is indicated by arrows). (C) Staining of a slide for both -galactosidase and NSE (reddish stain). The arrow indicates a cell that is simultaneously stained positive for -galactosidase and NSE. (D) Staining of a slide for both -galactosidase and AChE. In this slide, a hematoxylin-eosin counterstain was used to visualize all cells (purple color). The open arrow shows an AChE-positive megakaryocyte (reddish stain); in the same field, a large X-Gal-stained cell (the blue stain is indicated by the solid arrow) is also seen. None of the X-Gal-stained cells were also positive for AChE. Bar, 10 µm.

The total number of BAG-infected cells in the bone marrow of animals after i.p. inoculation is shown in Fig. 3. Although there was variation between different infected animals, there appeared to be a general increase in infected cells between 2 and 8 days postinfection, with typical maximal numbers ranging from 10² to 10³ infected cells per animal. The increase might have reflected the influx into the bone marrow of BAG-infected cells or alternatively the division of infected cells already in the bone marrow. Likewise, a potential decrease in infected cells between 8 and 12 days postinfection could have reflected the turnover of BAG-infected cells or their emigration from the bone marrow.

View larger version (13K): [in this window] [in a new window]   FIG. 3.   Numbers of BAG-infected cells per animal. The total number of BAG-infected cells in each animal's marrow were calculated as the number of X-Gal-stained cells in 2 × 105 cells (the total number of bone marrow cells recovered). Results for individual animals are plotted as a function of the number of days postinfection.

To further characterize the BAG-infected bone marrow cells, secondary histochemical stains were performed. NSE is a histochemical stain used to identify cells of the myeloid lineage (4). Myeloid cells that would stain positive for NSE include monocytes, macrophages, and immature osteoclasts (preosteoclasts). Mature osteoclasts (from adult mice) have not been reported to stain positive for NSE. Bone marrow cells from BAG-infected mice were stained first with X-Gal and then for NSE and were screened for the presence of cells that stained positive for both (double positive). As shown in Fig. 2C, doubly stained cells could be detected. Quantification of the doubly stained cells is shown in Table 1; a percentage of both large and small BAG-infected cells stained positive for NSE, indicating that they were in the myeloid lineage. A total of 17% of the small BAG-infected cells were double positive, and 8% of the large BAG-infected cells were double positive.

View this table: [in this window] [in a new window]   TABLE 1.   Simultaneous staining of BAG-infected bone marrow for -galactosidase and NSE

Since the large BAG-infected cells were potentially megakaryocytes, double staining with AChE, a marker for megakaryocytes, was carried out. When bone marrow cells from BAG-infected mice were stained with X-Gal and for AChE, no doubly infected cells were found (Fig. 2D). In addition, the morphology of the AChE-positive cells was different from that of the BAG-infected cells. The AChE-positive cells were regular and even after cytocentrifugation, compared to the irregular appearance of the BAG-infected cells. Thus, it appeared that the large BAG-infected cells were not megakaryocytes.

It was also interesting to examine BAG-infected bone marrow by staining for TRAPase. TRAPase is a marker for osteoclasts (20), cells in the bone responsible for bone resorption and derived from cells of the myeloid lineage (16). Indeed, peripheral monocytes or macrophages can be induced to differentiate into osteoclasts when cocultured with bone marrow stromal cell lines (19). Unfortunately, the conditions for X-Gal staining and TRAPase staining were incompatible, so double histochemical staining could not be performed. However, when TRAPase staining by itself was carried out, large irregular multinucleated cells, as well as small cells, stained positive (Fig. 4). The morphology of the large TRAPase-positive cells resembled that of the large BAG-positive cells. This supported the possibility that the large BAG-positive cells were osteoclasts. Another phenotypic marker that has been described on osteoclast precursors is the CD34 antigen. As shown below, a substantial fraction of the large BAG-positive cells were also CD34 positive.

View larger version (104K): [in this window] [in a new window]   FIG. 4.   TRAPase staining of infected bone marrow. A cytocentrifuge preparation of bone marrow from a BAG-infected mouse was stained for TRAPase. The figure shows two darkly stained osteoclasts (one large and one small). Note that the large osteoclast is multinucleate and irregular in shape. Bar, 10 µm.

To further characterize the BAG-infected cells, double immunofluorescent staining with an antibody to -galactosidase, in combination with monoclonal antibodies specific for lineage-specific cell surface markers, was carried out. As shown in Fig. 5, the anti--galactosidase antibody detected BAG-infected cells in the bone marrow preparations. Both large and small BAG-infected cells were detected by immunofluorescent staining (Fig. 5C and E), and the numbers of infected cells were equivalent to those detected by X-Gal staining.

View larger version (87K): [in this window] [in a new window]   FIG. 5.   Immunofluorescent staining of infected bone marrow. Cytocentrifuge preparations were made from BAG-infected mouse bone marrow. Two-color immunofluorescence microscopy was carried out with a rabbit polyclonal antibody for -galactosidase (with an FITC-conjugated secondary antibody), along with lineage-specific monoclonal rat antibodies (with a Texas red-conjugated secondary antibody). (A) Staining with FITC-conjugated secondary antibody alone. (B) Staining with the anti--galactosidase antibody of bone marrow from an uninfected mouse. In some other slides, this antibody produced nonspecific staining at the periphery of all cells; specific staining was evident as homogeneous cytoplasmic staining (see other panels). (C and D) Simultaneous staining of the same field for -galactosidase, photographed through a green filter (C), and for mac-1 antigen, photographed through a red filter (D). Note the large cell that is stained positive for -galactosidase but not mac-1. (E and F) Simultaneous staining of the same field for -galactosidase (E) and mac-1 (F). A small -galactosidase-positive cell did not stain positive for mac-1 (arrows). Other mac-1-positive cells are evident in panels D and F. Insets in panels E and F show a higher magnification of the region around the arrows. (G and H) Simultaneous staining of the same field for -galactosidase (G) and CD34 antigen (H). Note the large -galactosidase-positive cell that stained positive for CD34. Bar, 10 µm.

Double immunofluorescent staining of BAG-infected bone marrow with an anti--galactosidase antibody and a monoclonal antibody for mac-1 antigen was carried out. mac-1 is expressed on cells of the myeloid lineage, including myeloid progenitors and terminally differentiated macrophages. The large BAG-infected cells in the bone marrow did not stain positive for mac-1 (Fig. 5D), although numerous mac-1-positive cells were detected (Fig. 5D and F). This further supported the conclusion that the large BAG-infected cells in the bone marrow were not macrophages. Although osteoclast progenitors belong to the myeloid lineage, mature osteoclasts do not express mac-1 (14). Therefore, the results of the mac-1 staining supported the conclusion that the large BAG-infected cells might be osteoclasts.

Somewhat surprisingly, the small BAG-infected bone marrow cells also did not stain positive for mac-1 (only 1 of 31 examined; see below). Given the fact that some of the small BAG-infected cells were NSE positive (assigning them to the myeloid lineage), it had seemed likely that they would be mac-1 positive as well. One possible explanation is that some of the small infected cells were preosteoclasts that had down-regulated the expression of mac-1 but had not yet lost NSE activity.

Given the indications that the large BAG-infected bone marrow cells were osteoclasts, staining for CD34 and mac-2 antigens was of interest. Previous reports indicated that osteoclasts from freshly isolated bone marrow of adult mice express mac-2 (14). However, when bone marrow cells from uninfected or BAG-infected neonatal mice were tested, no mac-2-positive cells were detected. Thus, neonatal mouse osteoclasts are apparently mac-2 negative. However, when bone marrow from BAG-infected mice was cultured in vitro in a medium containing 1,25-dihydroxycholicalciferol, a vitamin D analog that promotes the outgrowth and differentiation of osteoclasts (19), large BAG-positive cells that also contained mac-2 could be detected readily (Fig. 6C and D). Approximately 70% of the BAG-positive cells in the osteoclast cultures were mac-2 positive. A high percentage of the cells in these in vitro cultures (ca. 50%) also stained positive for TRAPase (Fig. 6G), confirming the outgrowth of osteoclasts. These results indicate that neonatal osteoclasts can be induced to express mac-2, and they further supported the identification of the large BAG-infected cells as osteoclasts.

View larger version (72K): [in this window] [in a new window]   FIG. 6.   Staining of in vitro-differentiated osteoclasts from BAG-infected mice. Bone marrow cells from BAG-infected mice were cultured in vitro on glass slides with 1,25-dihydroxycholicalciferol to induce osteoclast differentiation. Two-color immunofluorescent staining is shown. (A) Staining of the culture with secondary FITC-conjugated antibody alone. (B) Staining of a culture from an uninoculated animal for -galactosidase. (C and D) Two-color staining of the same field for -galactosidase (C) and mac-2 antigen (D). A large double-positive cell is evident. (E and F) Two-color staining of the same field for -galactosidase (E) and CD34 antigen (F). Two double-positive cells are evident. (G) Staining of a slide from the same in vitro culture for TRAPase. Bar, 10 µm.

CD34 has been reported to be present in osteoclast precursors (17). A significant fraction of the large BAG-infected cells were CD34 positive (Fig. 5G and H), further supporting the identification of the large BAG-infected cells as osteoclasts. As expected, a high percentage of the in vitro-differentiated osteoclasts from BAG-infected neonatal mice were CD34 positive (Fig. 6E and F).

Double immunofluorescent staining was used to investigate further the nature of the small BAG-infected cells. Given their size and location in the bone marrow, it seemed likely that they might be hematopoietic precursors. To test for multipotential hematopoietic precursors, staining for sca-1 and CD34 was carried out. sca-1 is expressed on the earliest (self-renewing) hematopoietic precursors (18), while CD34 is expressed on more committed multipotential progenitors as well as stromal cells in the bone marrow (17). As shown in Table 2, no small BAG-infected cells expressed either sca-1 or CD34, indicating that early hematopoietic progenitors were not infected.

Staining with an additional hematopoietic lineage-specific monoclonal antibody was also carried out (Table 2). gr-1 antigen is a marker for committed granulocytic precursors; this was of interest in light of the NSE-positive small BAG-infected cells (presumptively myeloid). However, no BAG-infected cells showed staining for this marker, although other noninfected gr-1-positive cells in the bone were readily detected. Therefore, these experiments did not provide any evidence for the BAG infection of myeloid or multipotential hematopoietic progenitors. In limited preliminary experiments, none of the small BAG-infected bone marrow cells expressed the lymphoid markers Thy-1 and B220, either.

Since all of the experiments described above involved the helper-free BAG vector, it was important to test if bone marrow from mice infected with wild-type M-MuLV showed evidence of infection of the same cell types. Therefore, neonatal mice were infected i.p. with wild-type M-MuLV (1.0 × 10⁶ PFU/animal, as measured by XC assay), and 2 to 14 days postinfection, bone marrow cells were examined by immunofluorescent staining with an antibody specific for M-MuLV CA protein. As with the BAG-infected animals, both large and small M-MuLV-infected cells were detected (Fig. 7C and E). Some of the large M-MuLV-infected cells also stained positive for CD34 (Fig. 7D and Table 2), although the percentage of CD34-positive osteoclasts was lower for M-MuLV-infected animals than for BAG-infected animals. These results support the conclusion that osteoclasts are primary targets of infection by M-MuLV in the bone marrow after i.p. infection of neonatal mice. It was also interesting that in the M-MuLV-infected mice, significant numbers of small infected cells showed staining for the other hematopoietic markers (Fig. 7F and Table 2). This was particularly notable for mac-1, where approximately 24% (5 of 21) of the small M-MuLV-infected bone marrow cells stained positive for mac-1 versus approximately 3% (1 of 31) of the BAG-infected bone marrow cells. In addition, while 0 of 33 small BAG-infected cells stained positive for either gr-1 or CD34, 4 of 55 (7%) small M-MuLV-infected cells stained positive for these markers. These results might reflect the secondary infection of hematopoietic progenitors from initially infected cells.

View larger version (107K): [in this window] [in a new window]   FIG. 7.   Immunofluorescent staining of bone marrow from M-MuLV-infected mice. Cytocentrifuge preparations were made from M-MuLV-infected mice 8 days postinfection. (A) Staining with secondary FITC-conjugated antibody alone. (B) Staining with -galactosidase antibody and FITC-conjugated secondary antibody of cells from uninfected mice. (C and D) Simultaneous staining of the same field for CA (C) and CD34 antigen (D). Note the large double-positive cell. Some nonspecific binding of CA antibody to the periphery of cells is evident at this photographic exposure (see also panel E). (E and F) Simultaneous staining of the same field for CA (E) and mac-1 (F). Note the small mac-1-infected cell (arrows). Insets in panels E and F show a higher magnification of the region around the arrows. Bar, 10 µm.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

In these experiments, the first cells infected in the bone marrow of neonatal mice inoculated i.p. with M-MuLV were characterized. The use of a replication-defective M-MuLV-based vector allowed the identification of direct targets of infectioni.e., cells directly infected by the injected virussince infection could not spread beyond those cells. One cell type in the bone marrow was identified as a direct target of infection: the large infected cells were osteoclasts, cells involved in bone resorption. BAG infection of osteoclasts was deduced from a combination of criteria: (i) the size and morphology of the large BAG-infected cells was consistent with those of TRAPase-positive osteoclasts in primary bone marrow; (ii) the large BAG-infected cells were positive for CD34, which is present on osteoclasts and osteoclast precursors; (iii) in vitro differentiation of infected bone marrow by 1,25-dihydroxycholicalciferol yielded BAG-infected cells that stained positive for CD34 as well as mac-2, antigens characteristic of osteoclasts; and (iv) the large BAG-infected cells lacked markers specific for other large bone marrow cells (AChE and mac-1).

Small directly infected cells in the bone marrow were also detected, but the identities of the small BAG-infected cells were less clear. It is likely that some of them were osteoclast precursors, which would be consistent with the staining of a fraction of them with NSE. However, it was interesting that no evidence for primary infection of known hematopoietic progenitors was obtained. The small BAG-infected cells were negative for markers on multipotential progenitors (sca-1 and CD34) and on committed ones (mac-1 and gr-1). On the other hand, in adult mice infected neonatally with wild-type M-MuLV, there is ample evidence for the infection of hematopoietic progenitors (2). This suggests that the infection of hematopoietic progenitors in the bone marrow may result from secondary spread from osteoclasts or other directly infected cells. Indeed, the data in Fig. 7 and Table 2 demonstrating the presence of hematopoietic markers on significant numbers of small M-MuLV-infected bone marrow cells support this notion. Detailed time course experiments comparing the pattern of infection by BAG and wild-type M-MuLV are in progress. An alternate explanation is that hematopoietic progenitors lacking the markers tested here were the initial targets of M-MuLV infection. It will be interesting to culture BAG-infected bone marrow in vitro under conditions that allow the growth of hematopoietic progenitor colonies and to test if any of these colonies originated from BAG-infected cells.

What is the mechanism by which i.p. inoculation with the BAG vector led to osteoclast infection? One possible explanation is that the inoculated virions entered the circulation from the peritoneum and traveled to the bone marrow, where they infected osteoclasts or osteoclast progenitors. This seems rather unlikely, given the relatively modest input titers of BAG vector and the likelihood that many tissues could adsorb virus from the circulation. A more attractive explanation is that the BAG vector infected an osteoclast progenitor in the peritoneum and this infected progenitor subsequently migrated to the bone marrow. One candidate for such a cell would be a peritoneal macrophage or monocyte. It has been shown that peripheral macrophages and monocytes can differentiate into osteoclasts in vitro when cultured on bone marrow stromal lines (19).

It is unlikely that terminally differentiated osteoclasts are the primary target for BAG and M-MuLV infection. Simple retroviruses such as M-MuLV require the passage of the infected cell through mitosis to allow the breakdown of the nuclear envelope and the entry of the preintegration complex into the nucleus (9). Thus, precursors to osteoclasts that are still cycling are the most likely targets.

These experiments also have implications for gene therapy involving retroviral vectors. It has been previously demonstrated that retroviral vectors can infect early hematopoietic progenitors in vitro and that these vector-infected progenitors can drive hematopoiesis when injected back into animals (21). Thus, it is plausible that the direct injection of retroviral vectors into animals could also efficiently target early hematopoietic progenitors. However the results presented here indicate that the predominant bone marrow cells that are directly infected after i.p. injection are osteoclasts. Moreover, small vector-infected bone marrow cells lack the markers characteristic of early hematopoietic progenitors (sca-1 and CD34).

While the experiments reported here identified osteoclasts as at least one cell type initially infected by M-MuLV in the bone marrow after i.p. inoculation, there were limitations to the conclusions. First, we did not test whether the infected osteoclasts or osteoclast progenitors were productively infected. It was not possible to address this question with the BAG vector, but it will be important to test if osteoclasts or their progenitors infected by wild-type M-MuLV produce infectious virus. Second, the initial motivation for these experiments was to identify infected cells in the bone marrow that are important for leukemogenesis by M-MuLV. The experiments reported here do not address that issue directly, although they do provide the starting point. Future experiments with modified vectors and viruses may provide insight into this question.