INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Moloney murine leukemia virus (M-MuLV) is a simple replication-competent retrovirus that induces T-cell lymphomas in susceptible strains of mice. M-MuLV-induced leukemogenesis has been studied extensively (9). It is a multistep process that includes well-characterized molecular events such as insertional activation of proto-oncogenes (5, 11) and the formation of recombinant polytropic (i.e., mink cell focus-forming) viruses (10). End-stage tumors appear with a mean latency of 3 to 4 months after neonatal inoculation.

In our studies of M-MuLV pathogenesis, we have made use of an enhancer variant of M-MuLV, Mo+PyF101 M-MuLV. Mo+PyF101 M-MuLV contains enhancer sequences from the F101 strain of polyomavirus inserted directly downstream of the M-MuLV enhancer sequences in the U3 region of the viral long terminal repeat (LTR) (14). We have previously shown that this M-MuLV variant is poorly leukemogenic when used to inoculate neonatal NIH Swiss mice subcutaneously (s.c.) despite the fact that it replicates well in vivo (7). Mo+PyF101 M-MuLV also does not appear to induce many of the preleukemic changes normally associated with M-MuLV pathogenesis (2). Interestingly, if Mo+PyF101 M-MuLV is injected intraperitoneally (i.p.), mice develop disease with wild-type kinetics (1). There is a delay in the appearance of infectious virus in the bone marrow of mice infected with Mo+PyF101 M-MuLV s.c. compared to mice infected i.p., which suggested that efficient leukemogenesis requires high-level infection in the bone marrow at early times postinfection (1). This work also suggested that there is a cell type(s) present in the skin that is restricted for expression of Mo+PyF101 M-MuLV but not for wild-type M-MuLV. Such a cell would efficiently deliver wild-type M-MuLV (but not Mo+PyF101 M-MuLV) infection from the skin to the bone marrow after a s.c. inoculation.

In this study, immunohistochemistry was used to identify cells that are infected in the skin after an s.c. inoculation with M-MuLV and Mo+PyF101 M-MuLV. The results indicated that cells of developing hair follicles are the predominant site for M-MuLV infection in the skin and that these cells are not efficiently infected by Mo+PyF101 M-MuLV. In addition, cells of the developing hair follicle appeared to be primary targets of infection, since the same cells were infected after s.c. inoculations with a replication-defective M-MuLV based vector.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Cell culture and viral inoculations. The M-MuLV and Mo+PyF101 M-MuLV producer cells lines (43D and 25-3, respectively [7, 12]) and the NIH 3T3 cell line were all grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% calf serum. The Pam212 cell line (23) was grown in DMEM supplemented with 10% fetal bovine serum. For virus stocks, supernatants from 43D and 25-3 cells were harvested from cultures at 70% confluency after 48 h and clarified by low-speed centrifugation (1,200 × g for 10 min). Aliquots of viral supernatants were stored at 70°C until use and thawed only once. The titers of the viral stocks were determined by a focal immunofluorescence assay described below. Viral stocks were used to inoculate neonatal NIH Swiss mice subcutaneously at 1 to 2 days of age. The BAG vector stock was obtained from Psi-2 cells expressing the BAG vector as above and as described previously (17, 19). Neonatal (1 to 2 days old) NIH Swiss mice were inoculated s.c. with 200 µl of BAG vector stock (ca. 2 × 10⁶ to 6 × 10⁶ BAG infectious units [IU]/ml).

Preparation of skin protein extracts. Mice were sacrificed by cervical dislocation or CO₂ asphyxiation and depilated by using a chemical depilatory agent (Nair; Carter Products, New York, N.Y.). Skin from around the site of inoculation was removed, weighed, and placed in a 10-fold excess of a solution containing 2% sodium dodecyl sulfate (SDS), 100 mM dithiothreitol, and 60 mM Tris (pH 6.8). The samples were agitated at room temperature for 2 to 4 h and boiled for 5 min. Large particulate matter was removed by low-speed centrifugation (400 × g for 3 min). Chromosomal DNA was sheared by passing the sample repeatedly through a 20-gauge needle. The sample was then subjected to centrifugation at 10,000 × g, and the supernatant was removed and stored at 20°C until use.

Immunohistochemistry. Skin samples were fixed in 10% neutral buffered formalin (Sigma) overnight. Samples were embedded in paraffin and sectioned by a commercial histology laboratory. These sections were then deparaffinized, and endogenous peroxidase activity was neutralized by incubation in 2% H₂O₂ for 5 min at room temperature. Slides were then incubated in Dako Target Retrieval Solution (Dako, Carpinteria, Calif.) for 30 min at 90 to 95°C to unmask antigens that may have been affected by the paraffin embedding. After the slides were washed for 5 min in phosphate-buffered saline (PBS), they were blocked by incubation with 10% normal goat serum (NGS) in PBS for 2 to 4 h at room temperature. The slides were then incubated with an anti-M-MuLV capsid antigen (CA) rabbit polyclonal antibody (16) at a dilution of 1:10,000 in PBS plus 3% NGS overnight at 4°C. Slides were washed twice with PBS supplemented with 1% NGS for 10 min and incubated with a peroxidase-conjugated goat anti-rabbit antibody (Vector Laboratories, Burlingame, Calif.) at a dilution of 1:200 in PBS supplemented with 3% NGS for 1 h at room temperature. The slides were washed as described above and incubated with a peroxidase substrate (ABC; Vector Laboratories), and infected cells were visualized by light microscopy.

Western blot analysis. Ten to 15 µg of protein extract from skin samples were separated by electrophoresis on a SDS-10% polyacrylamide gel electrophoresis gel and transferred to nitrocellulose by electroblotting. To test if transfer of protein had occurred, the nitrocellulose membranes were stained with Pounceau red (Sigma, St. Louis, Mo.) immediately after transfer. Membranes were blocked for 1 h in 5% nonfat dry milk (Carnation, Glendale, Calif.) and then washed twice for 5 min and once for 15 min in PBS containing 0.05% Tween 20. The nitrocellulose membranes were then incubated for 1 h under constant agitation with a 1:5,000 dilution of the rabbit anti-M-MuLV CA antibody (see above) in PBS containing 0.05% Tween 20. The blots were then washed as before and incubated for 1 h with a 1:20,000 dilution of a peroxidase-conjugated donkey anti-rabbit antibody (Amersham, Arlington Heights, Ill.). The membranes were washed as before and incubated with the SuperSignal Chemiluminescent Substrate (Pierce, Rockford, Ill.) for 2 min before being exposed to X-ray film.

Infectivity assay. Infectivity assays were performed by seeding 10⁵ NIH 3T3 fibroblasts or Pam212 cells onto replicate 5-cm tissue culture dishes followed by infection with 1 ml of serial dilutions of M-MuLV or Mo+PyF101 M-MuLV. The cells were allowed to grow to confluence and were then fixed with 50% methanol for 5 min and washed with PBS for 5 min. Cells were then incubated for 30 min with the supernatant from the 548 hybridoma cell line that produces an anti-gag monoclonal antibody that reacts with M-MuLV-infected cells (4). Cells were washed twice with PBS containing 1% calf serum. A fluorescein isothiocyanate (FITC)-conjugated anti-mouse antibody (1:200 dilution; Pierce) was then overlaid for 30 min. Cells were washed as before, and infected foci were visualized by fluorescence microscopy with an FITC filter.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Identification of M-MuLV-infected cells in the skin. To identify cells in the skin that were infected by M-MuLV after an s.c. inoculation, neonatal NIH Swiss mice were inoculated with ca. 10⁶ IU and sacrificed at various times postinfection (1 to 10 weeks). Skin at the site of inoculation was removed, fixed, embedded, sectioned, and processed for immunohistochemical staining with a rabbit polyclonal antibody to M-MuLV CA protein as described in Materials and Methods. Microscopy of the stained sections was used to visualize the infected cells.

View larger version (127K): [in this window] [in a new window]   FIG. 1.   M-MuLV infection in the skin. Neonatal NIH Swiss mice were inoculated s.c. with ca. 106 fluorescence immunoassay PFU (see Materials and Methods). Animals were sacrificed at various times postinfection, and skin from around the site of inoculation was harvested, paraffin embedded, and sectioned. Immunohistochemistry was then performed on these sections to identify infected cells. The sections were first incubated with an antibody to the M-MuLV CA protein. This was followed by incubation with a horseradish peroxidase-conjugated goat anti-rabbit secondary antibody. The tissue was then incubated with a horseradish peroxidase substrate (that gives a purple color after reaction), and infected cells were visualized by light microscopy. Examples of this are shown here. Panels A and B show skin samples from control uninoculated mice that were stained with the anti-CA antibody. Panels C, E, and G show skin samples from mice that were infected with wild-type M-MuLV and sacrificed 4 weeks postinfection. Longitudinal (E) and transverse (C and G) cross-sections of hair follicles are shown; panel G is at higher magnification. Note the intense signal from cells that make up the outer region of the hair follicles (the ORS), indicating extensive infection. Panels D, F, and H show skin samples from mice that were infected with Mo+PyF101 M-MuLV which were also sacrificed 4 weeks postinfection. The regions that are infected with Mo+PyF101 M-MuLV are the same as those infected with wild-type M-MuLV (cells of the ORS), but the intensity of the signal is far less. Small cells that were not associated with hair follicles were also found to be infected (arrow in panel G) with both M-MuLV and Mo+PyF101 M-MuLV, but these cells were not common. Bar, 10 µm.

Low efficiency of skin infection by the Mo+PyF101 M-MuLV variant. As described in the introduction, the motivation for studying M-MuLV infection in the skin came from experiments with the poorly leukemogenic Mo+PyF101 M-MuLV variant. Figure 2A shows the differences between the wild-type and Mo+PyF101 M-MuLV LTRs. Our previous experiments had indicated that Mo+PyF101 M-MuLV does not efficiently deliver infection from the site of an s.c. inoculation to the bone marrow. This in turn suggested that Mo+PyF101 M-MuLV does not efficiently infect some cells in the skin. Thus, it was of interest to examine the skin from Mo+PyF101 M-MuLVinfected mice. As shown in Fig. 1D, when mice were infected s.c. with equivalent amounts of this virus, the frequency of hair follicle infection was lower than for mice infected with wild-type M-MuLV. When multiple microscopic fields were counted, approximately 50% of the hair follicles of mice inoculated with wild-type M-MuLV were infected compared to 27% of the hair follicles infected in Mo+PyF101 M-MuLV-infected mice. Moreover, the intensity of CA protein staining in the Mo+PyF101 M-MuLV-infected hair follicles was generally less than that observed for infection by wild-type virus (more easily visible at higher magnification [compare Fig. 1E and F and Fig. 1G and H). Individual infected cells outside of the hair follicles could also be detected in Mo+PyF101 M-MuLV-infected mice, and their frequency was equivalent to that of wild-type M-MuLV-infected animals. The lower frequency and intensity of CA antigen-staining in mice inoculated s.c. with Mo+PyF101 M-MuLV supported the notion that this virus is less efficient at infecting and expressing in cells of the skin.

View larger version (24K): [in this window] [in a new window]   FIG. 2.   The M-MuLV and Mo+PyF101 LTRs. (A) M-MuLV LTR. The enhancer sequences of M-MuLV lie within two 75-bp direct repeats in the U3 region of the LTR. Each direct repeat contains consensus binding sites for a number of well-characterized transcription factors. The Mo+PyF101 LTR is shown in the lower half of panel A. The enhancer region of the polyomavirus strain F101 has been inserted directly downstream of the M-MuLV direct repeats. Binding motifs in the PyF101 enhancers include two copies of a BPV-like enhancer core (B core) and three copies of a polyoma enhancer core (C1 and C2) (7). None of the viral structural proteins are altered in Mo+PyF101 M-MuLV. (B) BAG and Mo+PyBAG vectors. The BAG vector has been previously described (19). It contains the wild-type M-MuLV LTR driving the transcription of the bacterial lacZ gene. It also contains the Neor gene as a selectable marker. The Mo+PyBAG vector is similar to the BAG vector except that it has the LTR from Mo+PyF101 M-MuLV regulating lacZ expression. Both vectors were obtained from transfected Psi-2 cells as described in Materials and Methods.

View larger version (34K): [in this window] [in a new window]   FIG. 3.   M-MuLV and Mo+PyF101 M-MuLV infection in the skin. Protein extracts were made from the skin of mice that had been infected s.c. with wild-type M-MuLV and Mo+PyF101 M-MuLV and sacrificed at various times postinfection. Western blot analysis was then performed on these extracts as described in Materials and Methods. Equal amounts of skin extract (10 µg of protein) were analyzed in each lane. With the anti-CA antibody, both the viral p30 and the Pr65 gag proteins were detected. As shown here, protein extracts from mice that had been infected with wild-type M-MuLV had significantly higher levels of both viral proteins than equivalent amounts of protein extract from age-matched mice infected with Mo+PyF101 M-MuLV.

View larger version (26K): [in this window] [in a new window]   FIG. 4.   Quantification of Mo+PyF101 M-MuLV infection in the skin. To quantify the differences in the amount of virus present in the skin, serial dilutions of a wild-type M-MuLV-infected skin extract were compared to an extract from Mo+PyF101 M-MuLV-infected skin on the same Western blot. Panel A shows protein extract from mice sacrificed 4 weeks postinfection. A serial dilution of 10, 7.5, 5.0, 2.5, and 1.0 µg of M-MuLV-infected skin extract was compared to 10 µg of protein extract from an age-matched Mo+PyF101 M-MuLV-infected mouse. The viral p30 protein is shown. The intensity of the signal from 10 µg of the Mo+PyF101 M-MuLV skin extract was comparable to the signal seen from 1.0 µg of M-MuLV skin extract, indicating that there was approximately 10-fold less viral protein. A similar analysis was performed in panel B with mice that had been sacrificed 6 weeks postinfection. In this case, the Mo+PyF101 M-MuLV skin extract contained about fourfold less viral protein.

Mo+PyF101 M-MuLV shows less efficient infection of a keratinocyte line in vitro. The immunohistochemistry indicated that ORS cells of the hair follicle are a predominant site of infection by M-MuLV in the skin. Since the Mo+PyF101 M-MuLV-infected mice showed less infection of these cells, this suggested that cells of the ORS may support less efficient infection of the latter virus. To test this, since the ORS consists predominantly of keratinocytes, in vitro infection of the Pam212 mouse keratinocyte line was carried out. Table 1 shows infection of wild-type and Mo+PyF101 M-MuLV on NIH 3T3 and Pam212 cells, as measured in a focal immunofluorescence assay. The results indicated that wild-type M-MuLV infected the two cell lines with similar efficiencies; in contrast, Mo+PyF101 M-MuLV showed less infection when measured on Pam212 cells. After correction for the relative efficiency of wild-type M-MuLV infection on Pam212 cells versus NIH 3T3 cells, the relative infectivity for Mo+PyF101 M-MuLV on Pam212 cells was approximately fivefold less than for wild-type M-MuLV. This was in general agreement with the estimated decrease in infection in the skin for this virus as seen in the immunohistochemistry and Western blot analyses.

Hair follicle cells are the initial targets of infection in the skin. While the preceding experiments indicated that hair follicle cells were the predominant infected cells in the skin after s.c. inoculation, there were two possible explanations for their infection. Hair follicle cells (or their progenitors) could have been initially infected by M-MuLV during the s.c. inoculation. Alternatively, infection could have spread from other cells to them. To distinguish between these possibilities, we employed s.c. infection with a helper-free replication-defective M-MuLV-based vector expressing the bacterial -galactosidase gene (BAG; Fig. 2B). When the BAG vector is produced by transfection of Psi-2 packaging cells (that express M-MuLV virion proteins [15]), the resulting vector particles will infect the same cells that M-MuLV does, but they will not spread by infection to neighboring cells. We have previously used in vivo infection of BAG vector to identify the first cells infected in the bone marrow of i.p.-inoculated mice (17).

View larger version (139K): [in this window] [in a new window]   FIG. 5.   Infection by BAG vector in the skin. Neonatal NIH Swiss mice were inoculated with ca. 106 BAG vector particles. Mice were sacrificed 1 week postinfection, and skin was removed from around the site of inoculation and prepared as described above. Infected cells were visualized by immunohistochemistry with an anti--galactosidase antibody as described in Materials and Methods. In this protocol, a specific antibody reaction results in deposition of a brown stain. Panels A and B show skin samples from age-matched uninoculated mice stained with the anti--galactosidase antibody. Panels C and D show skin samples from BAG-infected mice. Note that in panel C cells in the ORS of the hair follicle are infected by the BAG vector and in panel D cells of the sebaceous gland are infected. Since the BAG vector is replication defective, these cells must be primary targets for M-MuLV infection. Bar, 10 µm.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

In these experiments, we studied the cell types infected by M-MuLV in the skin after a neonatal subcutaneous inoculation. The results indicated that the great majority of infected cells were in hair follicles. Indeed, by 4 to 6 weeks of age, approximately 50% of the hair follicles showed extensive infection, as measured by immunohistochemistry for M-MuLV CA antigen. The predominant infected cells within the hair follicles were cells of the ORS. The ORS consists primarily of keratinocytes; cells of this region play an important role in the generation and maintenance of hair follicles (13). In addition, cells of the sebaceous gland (which produces sebum for the hair follicle) were also infected; this was plausible since sebaceous gland cells are derived from the ORS during development (22). The finding that hair follicles are extensively infected by M-MuLV in the skin was plausible since cells in the hair follicles have high mitotic activity even in adult mice. Productive infection by simple retroviruses such as M-MuLV requires passage of cells through mitosis (21); moreover, even in cells that have acquired M-MuLV proviruses, virus production is more efficient if the cells are cycling (18).

In addition to the hair follicles, M-MuLV infection was also detected in single cells within the skin of animals inoculated s.c. These cells were much less frequently detected, i.e., the frequency of individually infected cells was approximately 1% of the frequency of infected hair follicles; moreover, if each infected cell in a hair follicle were counted, then the difference would have been even greater. As described in Results, these infected cells did not stain with markers for hematopoietic cells such as Thy-1 and mac-1. Thus, they do not appear to be T lymphocytes or macrophages, two cell types that are known to be readily infectable by M-MuLV later in the course of infection. One cell type that would be of interest is the Langerhans cell, by analogy to infection by lentiviruses (20). However, Langerhans cells are terminally differentiated, so it seems unlikely that they could be efficiently infected with a simple retrovirus such as M-MuLV.

While extensive hair follicle infection was identified after s.c. inoculation, experiments with wild-type M-MuLV could not distinguish between hair follicles being primary targets for infection or alternatively becoming infected by some other initially infected cell in the skin. In order to address this question, infection with the helper-free, replication-defective M-MuLV-derived BAG retroviral vector was carried out. Infection resulted in the appearance of BAG-infected hair follicles, albeit at a lower frequency than in mice infected with replication-competent M-MuLV. Thus, hair follicles are primary (and possibly secondary) targets for M-MuLV infection after s.c. inoculation of neonates. It was very noteworthy that BAG-infected hair follicles often showed the same high frequency of infected cells within an infected hair follicle as observed in wild-type M-MuLV-infected mice. Since the BAG vector is replication defective, this suggests that most of the infected cells in a hair follicle arose by division of an infected cell, as opposed to spread of the infection within the hair follicle. Indeed, it seems likely that the initial cells infected in the neonatal animal after s.c. inoculation are hair follicle progenitors. Interestingly, others have reported that an area of the ORS known as the "Wuste" or bulge region may contain follicular stem cells capable of regenerating an entire hair follicle (6).

As described in the introduction and in Results, the motivation behind these experiments was to identify the cells in the skin responsible for delivering M-MuLV infection to the bone marrow. Since the hair follicles contain the predominant cells infected by M-MuLV in the skin, they are the likely candidates for delivering infection to the bone marrow. Experiments with the Mo+PyF101 M-MuLV variant supported this suggestion. Mo+PyF101 M-MuLV does not efficiently establish early bone marrow infection after s.c. inoculation (1), and we found that the skin from mice inoculated s.c. with this virus showed substantially lower levels of infection. Both the number of infected hair follicles and the intensity of antigen staining in the infected cells was lower. Quantification of the levels of CA antigen in skin of Mo+PyF101 M-MuLV-infected animals also revealed lower levels of infection (ca. 4- to 10-fold) than for animals infected with wild-type virus. Finally, in vitro infection of a mouse keratinocyte line also showed less efficient infection than for wild-type M-MuLV.

If, as the results suggest, hair follicle cells are responsible for delivery of virus to the bone marrow, the mechanism by which this takes place must still be elucidated. One possibility is that the hair follicle cells act as centers of virus production; the resulting virus could then enter the circulation and infect bone marrow cells. Another possibility is that the hair follicle cells might infect other cells that deliver infection to the bone marrow. For instance, the singly infected cells detected in the skin might be mobile cells that traffic to the bone marrow.

The finding of infected hair follicle cells also has potential implications for spread of infection from animal to animal. It seems likely that infectious virus could be present in the skin or on the hair released from infected animals. This virus might be a source of natural spread to an uninfected animal, as a result of grooming, biting, or nursing.

It should be noted that, in these experiments, we examined the skin for infection. However, during s.c. inoculation, tissues under the skin also come into direct contact with the inoculated virus, e.g., the underlying musculature and the membranes separating the muscle and the skin. It is possible that cells of these other tissues are also involved in the delivery of infection from the site of an s.c. inoculation to the bone marrow. In future experiments, it will be important to address this issue.