Human Immunodeficiency Virus Type 1-Induced Hematopoietic Inhibition Is Independent of Productive Infection of Progenitor Cells In Vivo

Abstract

Human immunodeficiency virus (HIV)-infected individuals exhibit a variety of hematopoietic dysfunctions. The SCID-hu mouse (severe combined immunodeficient mouse transplanted with human fetal thymus and liver tissues) can be used to model the loss of human hematopoietic precursor cell function following HIV infection and has a distinct advantage in that data can be obtained in the absence of confounding factors often seen in infected humans. In this study, we establish that HIV type 1 (HIV-1) bearing a reporter gene inserted into the viral vpr gene is highly aggressive in depleting human myeloid and erythroid colony-forming precursor activity in vivo. Human CD34⁺ progenitor cells can be efficiently recovered from infected implants yet do not express the viral reporter gene, despite severe functional defects. Our results indicate that HIV-1 infection alone leads to hematopoietic inhibition in vivo; however, this effect is due to indirect mechanisms rather than to direct infection of CD34⁺ cells in vivo.

Patients with AIDS often suffer from hematopoietic abnormalities which include thrombocytopenia, anemia, lymphocytopenia, monocytopenia, and neutropenia (24, 26, 35). However, the mechanisms responsible for the hematopoietic dysfunction in human immunodeficiency virus (HIV)-infected patients remain unclear. Hematopoietic abnormalities may be caused by altered stem cell differentiation possibly due to abnormal lineage specific expression of certain cellular genes such as cytokines, receptor tyrosine kinases, and factors involved in embryonic development (3, 10, 17, 20, 31, 33). In general, investigators have failed to detect HIV infection in hematopoietic progenitor cells isolated from infected individuals, suggesting that HIV may have an indirect effect on hematopoiesis (reviewed in reference 26). However, confounding factors such as opportunistic infections, immune system-mediated effects, or the consequences of prolonged physiological stress, which could contribute to decreased hematopoiesis in patients, make the causative role of HIV in vivo uncertain.

Several laboratories have performed in vitro analyses in attempts to elucidate the mechanism of action responsible for altered hematopoiesis during HIV infection. It has been observed that hematopoietic progenitor cell colony growth and differentiation is inhibited in long-term bone marrow cultures of HIV-positive patients (6, 9, 11, 30). In further studies of aborted fetuses from HIV seropositive women, alterations in human fetal hematopoiesis in vitro were associated with maternal HIV infection (5). Early progenitor cells can express CD4 (27), and purified CD34⁺ cells were reported to be susceptible to HIV infection, as shown by PCR analysis for the presence of proviral sequences in the ensuing myeloid and erythroid colonies (8) or by virus production in culture (32). Further, in vitro studies by others suggested that HIV type 1 (HIV-1)-induced inhibition of hematopoiesis was mediated by the viral envelope glycoprotein gp120 and/or by the Nef protein (7, 21). The p24 Gag protein of HIV-1 also was shown to inhibit myeloid colony formation of bone marrow cultures but to have minor effects on erythroid colony formation (29). Thus, the in vitro effects of HIV on hematopoietic progenitors can reveal some of the consequences of HIV infection. However, while these in vitro studies suggest that HIV may have a negative effect on hematopoiesis, they cannot determine how the virus influences complex lymphoid microenvironments in vivo, as these systems lack an appropriate cellular microenvironment that is amenable to efficient HIV infection and which supports long-term pluripotent hematopoietic progenitor cells, which could be relevant to virus-induced indirect effects.

To understand the in vivo role of HIV on hematopoiesis more completely, a suitable animal model is necessary. In this regard, the severe combined immunodeficient (SCID) mouse model coimplanted with human fetal thymus and liver (Thy/Liv) (creating mice referred to as SCID-hu) (23, 28) provides a useful model to study the direct role of HIV on hematopoiesis in vivo (1, 4, 19, 35). The observation that myeloid and erythroid progenitor cells can be detected in these implants (23) makes this model amenable to study which lineages of hematopoietic cells are susceptible to HIV infection and the differentiation stage at which they are infected. This system also allows the controlled introduction of a cloned HIV strain into a functioning hematopoietic organ, in the absence of confounding factors such as opportunistic infections or antiretroviral or recreational drugs. In addition, no host immune response is mounted, thus eliminating immune system-mediated phenomena from the pathogenic profile. Since the mouse itself is not infected, effects of stress on normal murine physiologic functions also should be minimal. Last, the high virus loads seen following infection of thymic implants make this model an extremely stringent tool for assessing the infectability of the various cell types present. Thus, this model allows the causal role of HIV itself on hematopoiesis in vivo to be assessed.

We and others have previously reported that HIV-1 infection inhibits the recovery from Thy/Liv implants of hematopoietic precursor cells capable of giving rise to myeloid and erythroid colonies ex vivo (16, 18). Such inhibition preceded the expected depletion of CD4⁺ thymocytes. Those few colonies that were recoverable following in vivo infection did not harbor HIV provirus. However, it could not be ruled out that any infected precursors failed to grow into colonies. While infection induces depletion of some CD34⁺ cells (16), the number of CD34⁺ cells relative to remaining thymocytes does not decrease following HIV-1 infection. In contrast, our studies found that colony-forming activity (CFA) was severely depleted relative to recoverable progenitors, suggesting that the remaining CD34⁺ precursors were functionally impaired (18). Further, antiretroviral drug therapy of HIV-1-infected SCID-hu mice administered after depletion of hematopoietic CFA caused a transient resurgence of multilineage precursor cell activity, thereby supporting the notion that viable pluripotent stem cells remained functional in the infected microenvironment in vivo (18, 37).

Our previous studies have also provided information regarding the mechanism of depletion of CFA. Viruses bearing a syncytium-inducing CXCR4-tropic phenotype appear more aggressive toward colony-forming precursors than do non-syncytium-inducing, CCR5-tropic strains (18). Our laboratory has also identified a pediatric HIV-1 isolate derived from a child with severe hematopoietic abnormalities, which in SCID-hu mice was preferentially inhibitory to myeloid and erythroid CFA rather than toward CD4⁺ thymocytes (18). These data suggest that specific viral sequences or regions may be responsible for conferring such a preferential tropism or phenotype, towards non-T-cell hematopoiesis.

Our laboratory has previously used several accessory gene (vpr, vpu, nef, and vif) deletion mutants of HIV-1 (12) to infect SCID-hu mice (2, 13). Our earlier studies on the replicative ability of these deletion mutants in thymocytes derived from SCID-hu mice suggested a variability between loss of expression of each of the accessory genes and replication of these mutants (2). In this report, we investigated whether selective deletion of each of these accessory genes allows the virus to retain the ability to inhibit myeloid and erythroid CFA. Our data indicate that deletion of vpr has almost no effect on altering virus-induced inhibition of CFA. This result allowed us to use an HIV-1 reporter virus recently constructed in our laboratory that expresses the murine heat-stable antigen (HSA) by virtue of the insertion of HSA-encoding sequences in the viral vpr region (15). We have previously shown that this virus replicates to high titer, retains and expresses HSA sequences, and is highly pathogenic for human CD4⁺ thymocytes in the SCID-hu mouse. In addition, the extent of productive infection can be assessed in a cell population while maintaining the viability of the cells. We used this recombinant virus in this study to detect potential infection of CD34⁺ cells, as a possible mechanism of HIV-1-induced inhibition of hematopoiesis. Herein we establish that although highly enriched CD34⁺ human progenitor cells are recoverable from HIV-infected Thy/Liv implants, they are functionally reduced in the ability to form myeloid and erythroid colonies ex vivo. Furthermore, we detect no evidence of productive infection of these cells in vivo. These results indicate that the inhibitory effect on immature hematopoietic progenitor cells in vivo is unequivocally mediated by HIV; however, this inhibition is through indirect mechanisms.

MATERIALS AND METHODS

Accessory gene deletion mutants and recombinant HIV-1 expressing HSA.

Mutants with deletion of each of the HIV-1 accessory genes vpr, vpu, nef, and vif were obtained from Ron Desrosiers (12). Construction and replication kinetics both in vitro and in vivo of the HSA reporter virus NL-r-HSAS have been previously described (15). Plasmid DNAs of cloned viruses were electroporated into CEM T cells, and resulting virus production was determined by p24 enzyme-linked immunosorbent assay as previously described (2).

HIV-1 infection of SCID-hu mice.

Thy/Liv implants were infected by direct intraimplant injection as described previously (18) with 200 infectious units of wild-type HIV-1 NL4-3 or the vpr deletion mutant virus or with 2,000 infectious units of either of the vpu, nef, and vif deletion mutant viruses, or the NL-r-HSAS recombinant. Sequential wedge biopsies were performed on virus-infected and mock-infected control mice at regular time points, and approximately 25% or more of the Thy/Liv implant was removed to perform studies described herein. Cell lysates of infected Thy/Liv implants were analyzed by PCR using appropriate primers to detect deletions of the HIV-1 genome to confirm the identity of the virus (not shown).

Quantitative PCR analysis to determine viral loads.

The number of proviral DNA copies per 100,000 cells was determined as previously described, using the HIV-1 R-U5 primer pair (AA55-M667) to estimate the number of proviral sequences and a human β-globin-specific primer pair as an internal standard to assess the number of cell equivalents (2, 18, 38).

Depletion of CD3⁺ cells.

Briefly, total cells derived from Thy/Liv implants were stained with anti-CD3 (OKT3) monoclonal antibody and layered onto the surface of tissue culture flasks precoated with goat anti-mouse antibody as previously described (27). This procedure was repeated a total of three times, and the number of remaining cells was assessed. CD3 depletion was confirmed by flow cytometry, and the CD3⁺ cell fraction did not produce colonies in methylcellulose, indicating the absence of colony-forming progenitor cells.

Hematopoietic CFA.

CFA was determined exactly as described previously (18) by growing colonies in methylcellulose in the presence of erythropoietin (2 U/ml) and stem cell factor, granulocyte-macrophage colony-stimulating factor, interleukin-3, and interleukin-6 (100 ng of each per ml) for a period of 2 weeks (18). Colonies were then enumerated by standard light microscopy at a magnification of ×400.

Flow cytometry to assess CD4 depletion and HSA expression.

A total of 0.5 × 10⁶ cells derived from Thy/Liv implants were costained with labeled antibodies purchased from Becton Dickinson, Mountain View, Calif., for three-color flow cytometric analysis. One set of the mixture of antibodies consisted of anti-CD4-phycoerythrin (PE), anti-CD8-fluorescein isothiocyanate (FITC), and anti-CD3-peridinin chlorophyll protein (PerCP), and the other set included anti-CD34-PE, anti-CD24 (HSA)-FITC, and anti-CD45-perCP antibodies. The staining procedures were identical to those described previously (18). The anti-CD24-FITC (a rat immunoglobulin G2bκ [IgG2bκ] isotype) antibody was purchased from PharMingen, San Diego, Calif.). The cells were fixed in phosphate-buffered saline containing 1% formalin, and data were acquired with a Becton Dickinson FACScan flow cytometer. Single-color and isotype controls (IgG1-PE, IgG1-FITC, IgG1-PerCP, and rat IgG2bκ-FITC) were used to set compensation and gates, respectively, for the analyses. The data was converted by a FACS (fluorescence-activated cell sorting) Convert program for analysis using CellQuest software to calculate the percentage of cells stained by each antibody.

Cell sorting to separate CD34⁺ cells.

The CD3-depleted cells of Thy/Liv implants were costained with anti-CD34-PE, anti-CD24-FITC, and anti-CD38-allophycocyanin monoclonal antibodies and sorted for CD34⁺ cells in a FACStar Plus I70 water-cooled flow cytometer equipped with two 2.5-W lasers (manufactured by Becton Dickinson) at 260 mW, using the 488-nm argon laser. Sorted cells were subsequently analyzed for expression of CD24 and CD38. Backgating was performed on CD24⁺ cells to determine whether they coexpress CD34 and CD38.

RESULTS

Pathogenicity of HIV-1 deletion mutants in thymocytes.

To determine the role of HIV-1 accessory genes in CFA inhibition, we used HIV-1 accessory gene mutants to infect Thy/Liv implants in SCID-hu mice (vif, nef, and vpu mutants at 10× inoculum; the vpr mutant and wild-type HIV-1_NL4-3 at 1× inoculum). The 10-fold increase in inoculum of some of these mutants is to compensate for their decreased ability to infect this model (2). With respect to thymocytes, these mutant viruses behaved in Thy/Liv implants similarly to what we previously reported (2, 13). Specifically, the nef and vif deletion mutants exhibited attenuated replication and CD4⁺ cell depletion (Fig. 1A and B), with the vif mutant displaying the most attenuated phenotype. The low level of replication of the vif mutants suggests that thymocytes are at least somewhat permissive for virus production in the absence of vif. In contrast, under the conditions used here, the loss of either the vpu or the vpr gene resulted only in minimal decreases in virus replication or induction of CD4⁺ cell depletion in the Thy/Liv implants. These effects of the deletion mutants on thymocytes are included as controls for comparison of their effects on cells of the myeloid and erythroid lineages as reported below.

FIG. 1.

(A) Replicative ability of HIV-1 accessory gene mutants in Thy/Liv implants. Each symbol represents viral load (number of copies of HIV DNA per 100,000 cells) from a single Thy/Liv implant, as determined by quantitative PCR. (B) Percentages of total CD4⁺ (combined CD4⁺ CD8⁻ and CD4⁺ CD8⁺) cells present in SCID-hu Thy/Liv implants shown in panel A at different times postinfection as determined by three-color flow cytometry. The implants were directly infected with different HIV-1 accessory gene mutants. The cells were stained with anti-CD4-PE, anti-CD8-FITC, and anti-CD3-PerCP monoclonal antibodies. (C) Myeloid and erythroid CFA of total cells (5 × 10⁶) derived from SCID-hu Thy/Liv implants, either mock infected, HIV-1_NL4-3 infected, or infected with HIV-1 accessory gene mutants as indicated. Zidovudine (1 μg/ml) was also included in the methylcellulose to prevent the possibility of virus spread in the culture medium. The number of animals are as indicated in panel A.

Effects of deletion mutants on myeloid and erythroid CFA.

We investigated the effects of HIV-1 accessory gene mutants on CFA in parallel with the studies described above. All deletion mutants inhibited CFA from Thy/Liv implants to some degree. The degree of inhibition (Fig. 1C) correlated with the replicative ability of these deletion mutants in thymocytes (Fig. 1A). Interestingly, the nef deletion mutant exhibited delayed depletion of CD4⁺ cells yet depleted CFA fairly efficiently (Fig. 1B and C). Implants showing very low levels of vif deletion virus exhibited no CFA inhibition. The vpr deletion mutant was as inhibitory of CFA as was wild-type HIV-1_NL4-3. No effect of the viral accessory genes on CFA was observed beyond the influence of these individual genes on viral replication.

Pathogenic properties of a vpr-deleted HIV-1 recombinant reporter virus.

The experiments described above demonstrated that at identical multiplicities of infection, the vpr deletion mutant did not exhibit apparent differences from wild-type HIV-1_NL4-3 either in thymocyte depletion or in CFA inhibition. To further investigate the mechanism of HIV-1-induced hematopoietic inhibition, we infected Thy/Liv implants with an HIV-1_NL4-3 reporter virus (NL-r-HSAS) containing cDNA encoding the murine HSA in the deleted vpr region. This virus directs surface expression of HSA, facilitating detection of cells actively expressing viral genes (15). Since the SCID-hu mouse contains murine cells which express HSA and could confound results, we used costaining with antibodies specific for human CD45, to identify all human hematopoietic cells.

Our previous studies showed that NL-r-HSAS infects and depletes CD4⁺ thymocytes in SCID-hu mice with kinetics slightly delayed relative to wild-type HIV-1_NL4-3 infection. Whereas HIV-1_NL4-3 depletes the CD4⁺ CD8⁺ thymocytes in about 24 days, NL-r-HSAS was found to induce depletion beginning at 27 days postinfection (15). Further, at the time of thymocyte depletion, the viral loads of NL-r-HSAS were consistently at levels comparable to those of wild-type HIV-1_NL4-3. This reporter virus thus provides a means to assess productive infection of hematopoietic progenitor cells in vivo.

In this study, two mock-infected and two NL-r-HSAS-infected animals were studied in parallel in each of two separate experiments (experiments 1 and 2). Up to 12% of the cells derived from the Thy/Liv implants were productively infected by NL-r-HSAS, as assessed by flow cytometry (Fig. 2A). The depletion of thymocytes by this recombinant virus correlated with its replicative ability (not shown) as well as with the percentage of thymocytes positive for HSA expression (Fig. 2A and B). The depletion of CD4⁺ cells was seen primarily in the CD4⁺ CD8⁺ subset (Fig. 2A and B). All mock-infected implants contained total CD4⁺ thymocytes (CD4⁺ CD8⁺ and CD4⁺ CD8⁻) at greater than 95% of Thy/Liv cells, whereas all infected implants exhibited major decreases with total CD4⁺ cell levels of 59 and 68 (experiment 1) and 52 and 54% (experiment 2) at 4.5 weeks postinfection. NL-r-HSAS also induced a reduction in CFA; both myeloid and erythroid lineage colonies were decreased approximately 20-fold following infection (Fig. 2C; Table 1). Thus, we were able to use this reporter virus to determine if colony-forming precursor cells were productively infected in vivo.

FIG. 2.

(A) Cell surface expression of the HSA (CD24) antigen by flow cytometry. Single-cell suspensions from mock-infected or NL-r-HSAS-infected implants obtained in experiment 1 (4.5 weeks postinfection) were costained with anti-CD24-PE and anti-CD45-FITC, the latter antibody to establish that viral expression was occurring in human cells of the Thy/Liv implants. Cells were also costained in parallel with anti-CD4-PE and anti-CD8-FITC, to determine whether CD4⁺ cell depletion was induced. In the NL-r-HSAS infected implant represented here, we observed a maximum 12% of the human cells expressed virus (CD24⁺ CD45⁺) (top); depletion of CD4⁺ CD8⁺ cells was observed in the same infected animal (bottom). The percentage of each cell subpopulation is denoted in the histogram quadrants. The second animal infected in parallel showed up to 9% cells expressing HSA and similar thymocyte depletion (not shown). (B) Depletion of CD4⁺ CD8⁺ thymocytes after infection with NL-r-HSAS, 4.5 weeks postinfection from experiment 1 (two animals). The percentage of each of the cell populations was determined by flow cytometry as described for Fig. 1A. Data from experiment 2 are similar and hence not shown. (C) Inhibition of myeloid and erythroid CFA of SCID-hu Thy/Liv implants infected with NL-r-HSAS, 4.5 weeks postinfection. Two mock-infected and two virus-infected implants were compared for CFA as indicated (experiment 1). Similar data from experiment 2 are summarized in Table 1.

TABLE 1.

Enhancement of CFA with enrichment of CD34⁺ cells^a

Expt	Wk postinfection	Infection	Cell population	No. of:			Fold increase
				Input cells	Colonies
					Myeloid (mean ± SD)	Erythroid (mean ± SD)
1	3.5	Mock	Total Thy/Liv	5 × 106	194 ± 6	217 ± 6	1
			CD3 depleted	5 × 104	177 ± 10	181 ± 5	87
			CD34 enriched	5 × 102	164 ± 9	180 ± 11	8,370
	3.5	NL-r-HSAS	Total Thy/Liv	5 × 106	23 ± 3	32 ± 2	1
			CD3 depleted	5 × 104	21 ± 2	30 ± 3	93
			CD34 enriched	5 × 102	18 ± 1	23 ± 1	7,453
	4.5	Mock	Total Thy/Liv	5 × 106	190 ± 10	221 ± 11	1
			CD3 depleted	5 × 104	154 ± 6	186 ± 6	83
			CD34 enriched	5 × 102	128 ± 11	154 ± 10	6,861
	4.5	NL-r-HSAS	Total Thy/Liv	5 × 106	10 ± 2	13 ± 2	1
			CD3 depleted	5 × 104	8 ± 1	11 ± 1	83
			CD34 enriched	5 × 102	5 ± 1	8 ± 1	5,652
2	4.5	Mock	Total Thy/Liv	5 × 106	175 ± 15	147 ± 14	1
			CD3 depleted	5 × 104	164 ± 13	154 ± 15	99
			CD34 enriched	5 × 102	150 ± 11	160 ± 9	9,627
	4.5	NL-r-HSAS	Total Thy/Liv	5 × 106	9 ± 3	10 ± 2	1
			CD3 depleted	5 × 104	7 ± 2	9 ± 2	84
			CD34 enriched	5 × 102	6 ± 2	8 ± 2	7,368

^aThe CFA of total cells derived from two mock-infected and two NL-r-HSAS-infected Thy/Liv implants, CD3-depleted cells (postpanning), and CD34⁺ cells (postsorting) in each of two different experiments was determined. After successive enrichment of CD34⁺ cells by panning and sorting, the indicated numbers of cells from each preparation were plated in methylcellulose and assessed for CFA. The fold increase in CFA after each CD34 enrichment process was calculated for the total number of colonies (myeloid plus erythroid) compared to the unfractionated sample. 

Expression of HIV-1 by CD34⁺ cells.

To determine whether CD34⁺ cells are productively infected in this in vivo system, cells derived from two mock-infected and 2 NL-r-HSAS-infected Thy/Liv implants from each of the two experiments were enriched for CD34⁺ cells. Total Thy/Liv cells were first depleted of CD3⁺ thymocytes by panning and thus enriched for hematopoietic progenitor cells. Approximately 1% of Thy/Liv cells were recovered following this procedure; CFA was enriched by over 80-fold in both mock- and virus-infected implants following panning (Table 1). Further enrichment for CD34⁺ cells by FACS resulted in a 0.2% recovery of input CD3-depleted cells, enhancing the CFA by an additional 80- to 100-fold over that of the CD3-depleted cells, such that over 50 to 60% of these highly enriched cells from normal Thy/Liv implants were capable of differentiating into myeloid or erythroid lineages ex vivo. Thus, the CFA segregated with CD34⁺ cells.

Infection of Thy/Liv implants with NL-r-HSAS and depletion of CD3⁺ cells by panning showed that HSA expression primarily resides in the thymocytes, as depletion of CD3⁺ thymocytes reduced HSA expression to background levels (Fig. 3A). Following sorting for CD34⁺ cells, we found that infection with NL-r-HSAS did not result in an apparent global loss of CD34⁺ cells relative to thymocytes in experiment 1 (Table 2; Fig. 3B, top row) and an approximately 50% loss in experiment 2 (Table 2), confirming our previous observation for the wild-type HIV-1_NL4-3 strain (18), and those of Jenkins et al. (16) that HIV infection can result in some loss of CD34⁺ cells. However, this loss is not total and is less dramatic than the nearly complete loss of CFA. The somewhat higher number of CD34⁺ cells in infected implants at 4.5 weeks postinfection in the first experiment may be due to depletion of thymocytes and consequent relative enrichment of CD34⁺ cells. Quantitative PCR analyses showed low levels (7 and 8% in experiment 1; and 6% in experiment 2) of proviral genomes in the CD34-enriched cells (Fig. 3C). In these four animals, totals of 7 and 9% (experiment 1) and 2 and 3% (experiment 2) of the highly enriched CD34⁺ population stained positive for the viral reporter gene (Fig. 3B, middle row), consistent with the PCR analyses (Fig. 3C). However, postsort analyses by backgating on true CD34⁺ cells indicated that these cells were essentially all HSA (CD24) negative (Fig. 3B, bottom row; Table 3). PCR was not performed on true CD34⁺ cells since further separation of the CD34-enriched population was technically unfeasible. Thus, we detected essentially no productive virus infection of the true CD34⁺ population. However, the CFA of these postsorted non-virus-producing cells derived from infected implants was inhibited greater than 20-fold compared to controls (Table 1). Therefore, HIV-1-induced inhibition of CFA appears to be independent of productive infection of CD34⁺ cells, suggesting an indirect role of the virus on the hematopoietic microenvironment supporting the ability of these cells to differentiate.

FIG. 3.

(A) HSA expression in cells derived from NL-r-HSAS-infected SCID-hu Thy/Liv implants compared to mock-infected implants from experiment 1 (4.5 weeks postinfection). Each panel depicts the proportion of infected cells either before (left) or after (right) CD3 depletion. The data suggest that most of the HSA expression lies in the depleted thymocytes. The dark line in each of the two histograms represents the NL-r-HSAS-infected cells, and the dashed line represents the mock-infected cells. (B) Analysis of HSA expression in the CD34-enriched population. The top row represents the CD34-enriched cell populations following sorting from a mock-infected (left) and two NL-r-HSAS-infected (center and right) Thy/Liv implants 4.5 weeks postinfection from experiment 1. Before sorting, the CD3-depleted cells were stained with anti-CD34-PE, anti-CD38-APC, and anti-CD24 (HSA)-FITC monoclonal antibodies. Following sorting, due to the relatively low number of CD34⁺ cells in the implants, approximately 50% of the sorted cells expressed CD34. The entire population shown in the top row is referred to in the text as CD34 enriched, and the true CD34⁺ cells are represented in the upper right quadrants of the upper panels. Totals of 7 and 9% of these CD34-enriched cells from the two infected implants were positive for HSA expression (middle and right) as indicated. Backgating on these CD34⁺ cells revealed that these cells were actually HSA (CD24) negative (bottom row). Numbers in the quadrants (top row) or denoted above the gates (middle row) indicate the percentage of cells in the particular region. The numbers in the lower panels indicate percentages of true CD34⁺ cells expressing HSA, as determined by the backgating analyses. Similar data from experiment 2 are summarized in Tables 2 and 3. (C) Quantitative DNA PCR analyses of NL-r-HSAS infection. Total or CD34-enriched Thy/Liv cells from the two NL-r-HSAS-infected animals from experiment 1 were analyzed at 4.5 weeks postinfection for HIV proviral sequences and for cellular (β-globin) sequences in parallel with control standards. Relative proviral burden, as determined by radioanalytic imaging, is shown below each sample band. PCR data from experiment 2 are stated in the text.

TABLE 2.

Quantitation of CD34⁺ cells in Thy/Liv implants postenrichment^a

Expt	Infection	Mouse no.	CD34+ cells/100 million total cells
			3.5 wk	4.5 wk
1	Mock	173-49	8,000	11,300
		173-50	5,800	6,900
	NL-r-HSAS	173-39	5,600	21,700
		173-30	7,600	8,900
2	Mock	T6-9	NDb	17,000
		T6-12	ND	13,000
	NL-r-HSAS	T6-3	ND	9,000
		T6-7	ND	7,000

^aTotal cells from the implants were first panned to deplete the CD3⁺ thymocytes and then sorted for CD34⁺ cells by FACS. The number of truly CD34⁺ cells was calculated from the cells recovered postenrichment by flow cytometric analysis. 

^bND, not determined. 

TABLE 3.

Levels of HSA (CD24) reporter antigen expression in CD34⁺ cells postenrichment^a

Expt	Infection	Mouse no.	% CD24+ CD34+ cells
1	Mock	173-49	2.5
		173-50	0
	NL-r-HSAS	173-39	0
		173-30	2.5
2	Mock	T6-9	2.8
		T6-12	1.9
	NL-r-HSAS	T6-3	3.8
		T6-7	3.6

^aBackgating analyses were performed on the CD34⁺ cells postenrichment for HSA expression of the true CD34⁺ cells. 

DISCUSSION

We have examined the individual HIV-1 accessory genes for their contributions to virus-induced CFA inhibition. As previously reported, deletion of vpu, vif, and nef affected the replicative ability of HIV and slowed the kinetics of thymocyte depletion (2). Here we report that deletion of these accessory genes also influenced the inhibition of CFA, most likely due to the reduced replicative ability of these mutants. However, we found no evidence for a direct role of viral accessory genes in CFA inhibition.

The loss of the vpr gene failed to attenuate the replication kinetics, CD4⁺ cell depletion, or inhibition of CFA. This result allowed us to use a reporter virus with HSA inserted into the deleted vpr gene, in order to monitor virus expression on the cell surface, thereby allowing detection of target cells actively expressing viral genes. While insertion of this reporter somewhat attenuated depletion of mature CD4⁺ cells, it did not disturb the HIV-1-induced inhibition of CFA under investigation in this study. With this reporter construct, we were therefore able to examine the mechanism of HIV-1-induced CFA inhibition.

We previously reported that there was no significant infection of surviving hematopoietic colonies ex vivo as determined by PCR, and a severe loss of CD34⁺ cells relative to recoverable thymocytes did not occur following HIV-1 infection (18). Recently Jenkins et al. (16) quantitated total CD34⁺ cell levels early following infection and noted a loss of some of these cells; however, the majority of CD34⁺ cells remained in infected implants. In the present study, we highly enriched for progenitor cells surviving HIV infection and found little loss in numbers relative to surviving thymocytes, consistent with our previous observations and those of Jenkins et al. (16). In addition, we observed no differences in surface phenotype of these cells (CD34⁺ CD38⁺ CD45⁺ CD3⁻ CD4⁻ CD8⁻) (18) between infected and control implants. Levels of proviral DNA were very low in these cells, and we could not detect expression of virus-encoded HSA in cells expressing CD34; thus, these progenitors are not productively infected in vivo (Table 3). The 20-fold inhibition of colony-forming potential of these enriched cells seen in Table 1 is thus likely due to an effect of the virus on the microenvironment that supports the differentiation of pluripotent progenitor cells. Taken together, our results (18) coupled with those of Jenkins et al. (16) suggest that HIV infection induces two mechanisms deleterious to early myeloid/erythroid progenitor function: (i) an early loss of CD34⁺ progenitor cells (16) and (ii) a loss of function of surviving CD34⁺ cells due to indirect mechanisms (ref. 18 and data herein). Further supporting these conclusions are our previous data showing that antiviral drug treatment of infected SCID-hu mice, administered following loss of CFA, transiently revives the CFA (18). Thus, cells capable of further differentiation into hematopoietic colonies must remain in the HIV-infected implants. The segregation of CFA with enrichment of CD34⁺ cells is therefore additional evidence for the preservation or viability of these progenitor cells despite exposure to high levels of virus in the surrounding microenvironment and strengthens the notion that the originating stem cell is also not infected. These results suggest that HIV-1 itself inhibits hematopoiesis and further support previous studies suggesting that hematopoietic progenitor cells isolated from HIV-infected patients for gene therapeutic strategies will not be infected by the virus.

These studies evaluated the effects of HIV in vivo in the absence of confounding factors seen in infected patients. Our results present evidence that in vivo, HIV-1 itself strongly inhibits hematopoiesis, although the mechanism involved is indirect since hematopoietic precursors themselves are not greatly reduced in number and are not productively infected. This is in sharp contrast to the loss of CD4⁺ thymocytes, which are killed by direct infection in this model (14). Our results are supported by a recent study which found that despite expression of both CD4 and HIV coreceptors, G₀ stem cells and immature hematopoietic progenitor cells are refractory to infection in vitro (34). The effect on differentiation of hematopoietic progenitor cells seen in our model may thus be due to infection of the surrounding cells in the microenvironment. This type of indirect effect has been suggested by results from several in vitro studies (5, 6, 9, 11, 25, 30), although this is not a universal finding, as HIV-1 infection of bone marrow-derived adherent cells in at least one in vitro study did not alter the levels or activity of hematopoietic progenitor cells (22). The effects of these other cell types on hematopoietic progenitors in SCID-hu mice may occur via secretion of soluble host factors or alteration of cell surface interactions. Further studies to elucidate these indirect HIV-mediated effects on hematopoiesis in vivo may help to develop therapeutic interventions to resolve the loss of bone marrow function seen in HIV-infected individuals.