Depletion of CD34 CD4 Cells in Bone Marrow from HIV-1–Infected

Depletion of CD34+CD4+ cells in bone marrow from HIV-1–infected individuals

Nirmal K. Banda,1 George R. Simon,1 Jefrey D. Sipple,1 Kristina L. Terrell,1 Phillip Archer,2 Elizabeth J. Shpall,1 Ramesh K. Akkina,3 Adam. M. Myers,4 Gail Singer Harrison1

From the Departments of 1Medicine and 2Preventive Medicine and Biometrics, University of Colorado Health Sciences Center, Denver, Colorado; 3Department of Pathology, Colorado State University, College of Veterinary Medicine, Fort Collins, Colorado; 4Department of Medicine, Denver Health Medical Center, Denver, Colorado

Offprint requests: Nirmal K. Banda, PhD, Department of Medicine, Box B171, University of Colorado Health Sciences, 4200 East 9th Ave., Denver, CO 80262; [email protected]

(Received 26 August 1998; accepted 8 April 1999)

ABSTRACT Pancytopenia as a consequence of bone marrow abnormalities is commonly seen in HIV-infected individuals. To examine the effect that HIV-1 has on hematopoietic cells, we compared hematopoietic properties of bone marrow samples from HIV+ patients at various stages of disease with bone marrow samples from uninfected donors. While the absolute number of recovered CD34+ cells and the cloning efficiency of these cells did not differ significantly in HIV+ donors, the percentage of CD34+CD4+ cells was significantly depleted in late-stage HIV+ patients. We observed a direct correlation between the numbers of CD34+CD4+ cells in the bone marrow and the peripheral CD4 count. Further characterization of the CD34+CD4+ subpopulation demonstrated that these cells expressed lower levels of HLA-DR on their surface compared with CD34+CD4– cells, suggesting an immature phenotype. We also found evidence for expression of HIV-1 coreceptors CXCR-4 and CKR-5 message and protein in CD34+ bone marrow cells. While this finding suggested that hematopoietic cells might be susceptible to HIV infection at an early stage of maturation, thus affecting different cell lineages as they matured, we did not find any evidence for infection of HIV in these cells. These data suggest that HIV affects early hematopoietic progenitor cells either directly or indirectly, and in particular CD34+CD4+ cells. This finding has important implications for disease pathogenesis and for application of gene therapy approaches that use CD34+ hematopoietic cells.

KEY WORDS: Hematopoietic • HIV • AIDS • Bone marrow • CD34 cells

INTRODUCTION set of progenitor cells is directly infected with HIV. For this Cytopenias are very commonly seen in HIV-infected to be feasible, the subset of cells must express the CD4 individuals, particularly in the late stages of disease. Several receptor required for HIV infection at some point during mechanisms have been postulated, all of which may con- their maturation process. In support of this, several groups tribute to the hematopoietic defects seen in these patients. including our own have demonstrated the existence of The first and most straightforward explanation is that the CD34+CD4+ cells [2–8]. These reports have indicated a antiretroviral drug regimens used to treat infection induce wide range of expression of CD4 on CD34+ cells, from 1 to myelosuppression [1]. This is unlikely to account in entirety, 65%. We and others have shown that the CD4 molecule however, for the wide range of cytopenias observed. A sec- present on CD34+ cells is capable of binding gp120 [8,9], ond explanation for hematopoietic dysfunction is that a sub- suggesting that even in the absence of direct HIV infection, the CD34+CD4+ cell could interact with HIV, resulting in impairment [10]. Such an impairment could have devastat- ing effects on hematopoiesis, since depletion or alteration of Supported by National Institutes of Health grant U0135231 and Pediatric the earliest progenitor cells could affect all differentiated AIDS Foundation grant 50613 to G.S.H. and by the Colorado Institute for progeny cells such as granulocytes, myeloid cells, erythro- Advanced Technology. cytes, and T and B cells. Thus, a third potential explanation

162 BB&MT Depleted CD34 CD4 Cells n AIDS for hematopoietic dysfunction could be indirect effects of indirect mechanisms such as gp120-induced, fas-mediated HIV on early cells in the absence of productive infection. apoptosis [8,10,33,34]. This would have important implica- The presence of proviral DNA in purified bone marrow tions for disease pathogenesis and for use of these cells for CD34+ cells from HIV-infected patients seems very rare or gene therapy and hematopoietic reconstitution. A better absent [11–13]. Some studies have shown active viral repli- understanding of the effects that HIV-1 exerts on CD34+ cation at the bone marrow level [14–18]; these infected cells cell maturation may lead to therapeutic interventions to may be either monocyte precursors [14,17] or megakaryo- improve hematopoiesis in HIV-infected individuals. cytes [15]. It has been reported that bone marrow stromal cells are infected with HIV-1 in vivo; as a consequence, the production of certain stromal cell–derived hematopoietic PATIENTS AND METHODS growth factors is deficient [19]. While these studies suggest Sources of normal and HIV-1–infected bone marrow direct infection of HIV in a subset of hematopoietic cells, Bone marrow (BM) from adult HIV-1–seronegative efforts to detect viral sequences in CD34+ cells have been donors was obtained as residual material from the Universi- largely unsuccessful [12,20–24]. ty of Colorado Bone Marrow Transplantation Unit Susceptibility to HIV infection would require not only (BMTU). Eight BM samples from HIV-1–seropositive expression of CD4 on the cell surface but also coexpression donors were obtained using Institute Review Board (IRB)- of one of the HIV coreceptors. Two such coreceptors have approved protocols of University of Colorado Health Sci- recently been identified, CXCR-4 [25–29] and CKR-5 ences Center from seven patients who gave informed con- [30–32]. They have been referred to in the literature as R4 sent. One patient donated bone marrow twice at a 6-month and R5, respectively. CXCR-4 provides a point of entry for interval (samples 1 and 5). The median age of HIV-1–infect- HIV grown in T cell lines but not for primary isolates [25]. ed patients was 38 years (27–49). The median CD4 count CKR-5 is important for the entry of macrophage tropic was 153 cells/mm3 (27–640). Four HIV+ patients had HIV isolates, and its mRNA has been detected in cell types advanced AIDS with opportunistic infections or malignan- susceptible to infection with these isolates [32]. Our data cies. One patient had AIDS without opportunistic infections here support the findings by Deichmann et al. [7] that or malignancies; the remaining two patients did not have CD34+ cells express message for CXCR-4. Additionally, we AIDS and had relatively high CD4 counts. Patient informa- show that message for CKR-5 and protein for both corecep- tion for the HIV+ BM donors is shown in Table 1. All tors is present in these cells. The cells do not contain HIV patients had received treatment with protease inhibitors in a RNA or DNA at detectable levels, however. triple-therapy regimen except patient 2, who for medical In this study, we analyzed the hematopoietic properties reasons received only AZT. Approximately 60 mL of bone of eight bone marrow CD34+ cell samples derived from marrow was aspirated from the posterior iliac crest of each HIV-infected donors with various CD4 counts, and also of donor and collected in heparinized 60-mL tubes. 11 bone marrow samples from HIV-negative control donors. Our data support the existence of CD34+CD4+ CD34+ selection of bone marrow cells hematopoietic cells, demonstrate that these cells are signifi- BM samples were diluted with column buffer (1ϫ phos- cantly depleted in bone marrow of late-stage HIV-infected phate-buffered saline, 0.2% human serum albumin, 5 mM individuals, and show that these cells are enriched for HLA- EDTA) and then subjected to Ficoll-Hypaque gradient cen- DRlow–expressing cells, suggesting a primitive phenotype. trifugation. Mononuclear cells were collected and washed Although these cells appear to express coreceptor message twice with column buffer, and 2.4ϫ108 cells were incubated and protein, and thus should be susceptible to HIV infec- with 240 µL A1 blocking reagent (human IgG) and 240 µL tion [7 and this study], we did not find any evidence for A2 blocking reagent (QBEND/10, mouse IgG1) together HIV-1 RNA or DNA in these cells. These results raise the for 15 minutes at 4°C. Cells were washed once with column possibility that the CD34+CD4+ cells could be depleted by buffer, followed by incubation with 240 µL colloidal super-

Table 1. Clinical status of HIV + BM donors

AIDS? Current Sample Peripheral blood Viral burden (opportunistic infections/ antiretroviral no. CD4 cells/mm3 (viral copies/mL) associated malignancies) medication

1 27 50,112 Yes (none/NHL) AZT, ddI, saquinavir 2 30 125,000 Yes (none/none) AZT 3 96 <400 Yes (thrush, MAI/none) d4T, 3TC, indianavir 4 41 71,608 Yes (none/none) 3TC, ddI, saquinavir 5* 348 400 Yes (none/NHL in remission) d4T, 3TC, indianavir 6 520 1600 No d4T, 3TC, nelﬁnavir 7 640 600 No d4T, 3TC, indianavir 8 209 6000 No d4T, nelﬁnavir, nevirapine

Viral burden was determined by RT-PCR. 3TC, lamivudine; AZT, zidovudine; d4Y, stavudine; ddI, didanosine; MAI, macobacterium avium intercellular; NHL, non-Hodgkin’s lymphoma. *Sample 5 was taken from the same patient 6 months after sample 1.

BB&MT 163 paramagnetic MACS microbeads (B1) for 15 minutes at 3 (100 U/mL; R&D Systems). Duplicate plates were ana- 4°C, then washed again with 1ϫ column buffer. CD34+ cells lyzed for each sample. Plates were incubated at 37°C with were purified using VS+ separation columns (large columns) 5% CO2 and humidity. Colonies were counted 11–14 days on VarioMACS (Miltenyi Biotec, Auburn, CA). To obtain after plating and scored for granulocytic/monocytic (GM), higher purity, cells were fractionated two times on separate monocytic (M), erythrocytic (BFU-E), and multilineage VS+ columns. Purity of CD34+ was examined via fluores- (GEMM) colonies. Cloning efficiency (CE) was calculated cence-activated cell sorting (FACS) by staining 1ϫ105 cells as follows: with phycoerythrin (PE) (Becton Dickinson, San Jose, CA), fluorescein isothiocyanate (FITC) (Becton Dickinson), or Total number of colonies counted CE= ϫ 100 Tricolor (TC) (Caltag Laboratories, San Francisco, CA) Total number of cells plated conjugated antibodies. Purity of CD34+ cells by FACS analysis averaged 88.8% (median 94.7%) and 91.6% (medi- Polymerase chain reaction (PCR) on bone marrow cells an 93.0%) for BM from HIV + and HIV – donors, respective- for fusin coreceptor (CXCR-4), CKR-5, and HIV reverse ly. Final yields from BM aspirates were 2.0–5.7ϫ106 and transcriptase (HIV-RT) 0.12–6.2ϫ106 CD34+ cells from HIV – and HIV + donors, Nested PCR for HIV-RT has been previously described respectively. Because of limited cell numbers, not all analy- [35]. For RT-PCR, 2ϫ106 MNCs and 1–2ϫ106 CD34+- ses were performed on all samples. purified bone marrow cells from HIV + patients were used to extract total cellular RNA using Trizol reagent (Gibco) as FACS analysis on bone marrow CD34+ cells described in detail by the manufacturer. The RNA pellet CD34+ bone marrow cells from HIV – and HIV + indi- was resuspended in 33 µL RNAse-free MilliQ water and viduals were analyzed for the presence of CD4. Purified stored at –70°C until used for cDNA synthesis. Before CD34+ bone marrow cells (1ϫ105) were stained with a mix- cDNA synthesis, all samples were treated with 1 µL DNase ture of PE- and FITC-conjugated anti-CD34 and anti-CD4 1 (Gibco) to remove the contaminating DNA and 1 µL monoclonal antibodies (Leu 3a; Becton Dickinson). As an RNasin (Promega, Madison, WI) to protect from RNA isotype control, a mixture of matched IgG1 isotype PE- and degradation. cDNA synthesis was performed as previously FITC-conjugated antibodies were used in all FACS analy- described [36]. Primers for CXCR-4 were selected using the ses. Staining procedure was according to our published MacVector program. CXCR-4 primers were: CAAGG studies [8]. Cursors were set based on the matched isotype CAGTCCATGTCATCTACAC (sense) and TGCACAGT controls. To further characterize the phenotype of the GTTCTCAAACTCACACC (antisense), amplifying a CD34+CD4+ and CD34+CD4– cells, expression of HLA-DR sequence of 516 bp. Nested primers used to amplify CKR-5 on the cell surface was measured using tricolor immunofluo- have been published [7], generating a band of 491 bp. PCR rescence FACS with the following antibodies: FITC-conju- analysis was performed as previously described [35]. The gated HLA-DR (Becton Dickinson), PE-conjugated anti- PCR conditions used allowed us to detect one copy in 103 CD4, and TC anti-CD34 (Caltag Laboratories). cells (not shown). As a control for contaminating DNA, all CD34+ bone marrow cells from HIV – donors were also samples were run with and without reverse transcriptase examined for the presence of HIV second coreceptor pro- (RT). In a parallel reaction, ␤-actin message was amplified teins CXCR-4 and CKR-5. Cells (2ϫ105) were incubated to adjust for the amount of RNA in the reaction. Primers to for 20 minutes in the dark at 4°C with anti-CXCR-4 (PE- ␤-actin have been described [37]. HOS-CD4 cells, used as a conjugated) and anti-CKR-5 (FITC-conjugated) monoclo- positive control for CXCR-4 and CKR-5, were grown as nal antibodies (Pharmigen, San Diego, CA). Matched iso- described [31]. In some cases, peripheral blood mononuclear type controls IgG2a␬ (PE) and IgG2a␬ (FITC) were also cells, either freshly isolated or grown with 100 U/mL IL-2 used to set the cursors. In parallel, HOS-CD4 [31] and sta- (kindly provided by Hoffman-LaRoche, Nutley, NJ) and 4 bly transfected 293 cells expressing CKR-5 were used as µg/mL phytohemagglutinin (PHA), were used as controls positive controls. for PCR. Amplification products were resolved on a 1% agarose gel using 1ϫ Tris acetate EDTA (TAE) buffer. A Methylcellulose assay on mononuclear and CD34+-purified 123-bp DNA ladder was used as a molecular weight marker. bone marrow cells After electrophoresis, the agarose gels were stained with Methylcellulose colony (MC) formation assay was used ethidium bromide, visualized, and photographed. to determine the colony-forming capability of CD34+ cells DNA-based PCR for RT and ␤-actin used the same derived from total mononuclear cells (MNCs) and CD34+- conditions as described above for RT-PCR, except that only purified BM cells from HIV + and HIV – donors. For MNCs, external amplification was used to amplify RT. To increase 3ϫ105 cells were used; for CD34+-purified cells, 2ϫ103 cells sensitivity of detection, we used chemiluminescence with were used. Cells were plated in 35-mm2 petri dishes after biotinylated probes using the Phototope-Star Detection Kit resuspending in 200 µL Iscove’s modified Dulbecco’s medi- (New England Biolabs, Beverly, MA), as described by the um (IMDM) containing 1 mL methylcellulose (44%; Stem manufacturer. Cell Technologies, Minneapolis, MN), 30% fetal calf serum (FCS) (not heat-inactivated; Gibco BRL, Grand Island, Statistical analysis NY), bovine serum albumin (1%; Boehringer Mannheim, Comparison of independent samples was carried out Indianapolis, IN), 2-mercaptoethanol (10–4 M), methylpred- using nonparametric (usually randomization test) techniques nisolone (10–6 M), erythropoietin (1 U/mL; R&D Systems, when sample sizes were small or parametric techniques (Stu- Vancouver, British Columbia, Canada), and interleukin (IL)- dent’s t test) when the necessary assumptions were satisfied.

164 BB&MT Depleted CD34 CD4 Cells n AIDS

Figure 2. Cloning efﬁciency of hematopoietic bone marrow pro- Figure 1. Ratio of bone marrow CD34+cells to total MNC in HIV+ genitors in HIV+ and HIV– samples and HIV– samples Each value shown is averaged from duplicate methylcellulose plates. Sample numbers for HIV + patients correspond to Table 1. Each value shown is averaged from two counts for each sample.

patient (as measured by CD4 count, viral burden, and accompanying diseases; see Table 1). The relationship between peripheral CD4 count and CD34+CD4+ percentage in the bone marrow was character- Cloning efficiency of purified CD34+ bone marrow cells ized using standard linear regression techniques. Averages from HIV+ patients are expressed as mean ± standard deviation (SD). The clonogenic potential of purified CD34+ cells derived from HIV + and HIV – donors is shown in Fig. 2. The average cloning efficiencies were 2.2 ± 3.6 and 1.1 ± 1.2 for HIV + and RESULTS HIV – samples, respectively. This difference was not statisti- A total of eight BM samples from seven HIV-infected cally significant (p > 0.05). We noticed a significant and con- patients with various CD4 counts and 11 samples from sistent difference in the size of the colonies from the HIV + vs. uninfected controls were analyzed. HIV + samples are identi- HIV – CD34+ cells. In samples from late-stage AIDS patients fied as numbered samples according to patient number (samples 1–4), the colonies were at least fivefold smaller than (Table 1). Table 1 shows the CD4 count, viral burden, and colonies from the control BM and had a tendency to die after current antiretroviral drug regimens for the donors. Sam- 12–14 days in methylcellulose culture. Thus, while the differ- ples were analyzed for total number of CD34+ cells as a ratio ence between the number of colonies was barely significant, of total MNC cells, cloning efficiency in methylcellulose, there was a substantial reduction in colony size and viability percentage of CD34+CD4+ cells, PCR and p24 assays for in the HIV + samples. We did not observe a difference in the HIV, and PCR for CXCR-4 and CKR-5. types of colonies obtained, which were primarily G/M or M for both HIV + and HIV – samples (not shown). Recovery of CD34+ bone marrow cells from HIV+ patients There was no apparent difference in the purity of Decrease of CD34+CD4+ cells in HIV+ patients with AIDS CD34+ preparations from HIV + vs. HIV – donors (average and correlation with CD4 count 88.8 and 91.6%, respectively). The total number of purified We and others have previously described the existence CD34+ cells as a ratio of total MNC cells is shown in Fig. 1. of a subset of CD34+CD4+ cells. In the analysis of BM sam- The average ratio for HIV + donors was 0.49 ± 0.44%, com- ples from HIV + patients, we observed a very significant pared with 0.91 ± 0.86% for HIV –. This trend toward lower decrease in this population compared with HIV – controls. CD34+ cell recovery in the HIV + samples was not significant The data in Fig. 3 show an average percentage for this pop- (p > 0.05), and appeared to depend on two HIV – samples ulation (expressed per total CD34+ cells) of 10.7 ± 9.2 and (samples 2 and 3) that had higher-than-average ratios. 34.8 ± 9.6 in HIV + and HIV – samples, respectively. This There did not appear to be a consistent relationship difference was very significant by either parametric or non- between the CD34+/MNC ratio and clinical state of the parametric testing (p < 0.005).

BB&MT 165 ᮀ CD34+CD4+

Figure 4. Positive correlation between CD34+CD4+ bone marrow cells and peripheral CD4+ T cells The correlation is shown between percentage of CD34+CD4+ cells (as shown in Fig. 3) and peripheral CD4 count of the patients. Linear regression analysis was used to fit the line (see MATERIALS AND METHODS). Figure 3. Significant depletion of CD34+CD4+ cells in bone marrow from HIV-1–infected patients The percentage of CD34+CD4+ cells per total CD34+ cells is compared in HIV + and HIV – samples. Each value is from a single FACS analysis. CD34+CD4+ population expressed relatively low levels HLA-DR compared with the CD34+CD4– population. The log intensity of HLA-DR positivity peaked at about 21 for CD34+CD4+ (Fig. 5F) and 31 for CD34+CD4– (Fig. 5E) From the data in Fig. 3, there appeared to be a positive cells. There was thus a 1.5-fold difference in these cell pop- correlation between percent CD34+CD4+ cells and CD4 ulations, consistent with the idea that CD34+CD4+ cells are count of the patient (Table 1). We therefore examined this earlier in their hematopoietic maturation than CD34+CD4– correlation, shown in Fig. 4. A statistical analysis showed the cells. Similar results were obtained with the CD38 marker correlation to be highly significant, as the slope of the and with CD34+ cells isolated from cord blood of HIV – regression of double-positive percentage vs. CD4 count was donors (data not shown). significantly greater than zero (p < 0.002). A test of the Consistent with the hypothesis that CD34+CD4+ cells intercept showed that the line went through the origin with are earlier than CD34+CD4– cells, we also observed in these slope 0.035. The r 2 value was 0.87. Thus, the reduction in cells a slightly increased cloning efficiency (4.95 vs. 3.58%, CD34+CD4+ cells in HIV + BM was correlated with CD4 respectively) and a higher percentage of BFU-E colonies in count in the patient. methylcellulose (1.5 vs. 0% of total colonies, respectively; data not shown). Reduced expression of HLA-DR on CD34+CD4+ cells compared with CD34+CD4– cells CD34+ cells from HIV+ donors express the CXCR-4 and Reduction in the percentage of CD34+CD4+ cells in CKR-5 coreceptors HIV + patients could have important implications for disease Two BM samples were analyzed for expression of pathogenesis and might result from either direct (HIV CXCR-4 and CKR-5 HIV coreceptors both before and infection) or indirect (secondary to HIV infection) mecha- after CD34+ cell purification, along with positive controls. nisms. Impairment of an early cell would be expected to RT-PCR with control ␤-actin is shown in Fig. 6. All samples affect all subsequent lineages of that cell. We therefore were also run through PCR without RT as a negative con- examined expression of HLA-DR (expressed at lower levels trol. The positive controls, fresh peripheral blood mononu- on earlier cells and higher levels on more committed cells) clear cells and HOS CD4 cells, were positive for ␤-actin in CD34+CD4+ cells vs. CD34+CD4– cells using triple-stain- (bottom), CXCR-4 (top), and CKR-5 (middle). Similarly, all ing FACS analysis on HIV + BM sample 7. Figure 5A shows BM samples tested (MNC before CD34+ selection and the expression of HLA-DR and CD4 on the starting CD34+ purified CD34+ cells) were positive for ␤-actin, CXCR-4, population used for gating in this experiment (gated popula- and CKR-5. In all cases, the samples without RT were nega- tion shown in Fig. 5D). The CD34+CD4+ subpopulation tive. Thus, CXCR-4 and CKR-5 were expressed in the two shown in Fig. 5B was further analyzed for the expression of HIV + BM samples examined, both before and after CD34+ HLA-DR (Fig. 5E and F). The results show that the selection.

166 Depleted CD34 CD4 Cells n AIDS

A B C

D E F

Figure 5. Low expression of HLA-DR on CD34+CD4+ cells compared with CD34+CD4– cells A. Expression of HLA-DR and CD4+ on the starting CD34+ population. B. Gating on the CD34+CD4+ and CD34+CD4– subpopulations. C. Overall expression of HLA-DR on CD34+ total cells. D. Purity of gated CD34+ population. Mean ﬂuorescence intensity of HLA-DR on CD34+CD4– (E) and CD34+CD4+ (F) cells.

CD34+ cells from HIV– donors express CXCR-5 and CKR-5 CD34+-selected), and 7 (MNC and CD34+-selected). We receptors on their surface did not detect any HIV message in any of these preparations CD34+ cells from HIV – bone marrow samples were ana- (Fig. 8). All samples did, however, exhibit a positive ␤-actin lyzed for the surface expression of CXCR-4 and CKR-5 signal (Fig. 6, bottom panel). Thus, hematopoietic MNC (Fig. 7). Isotype controls are shown in Fig. 7A and D. and CD34+ cell populations from these patients (who were Although both were expressed on a subset of CD34+ cells, relatively healthy; see Table 1) did not express HIV the percentage of cells expressing CXCR-4 (76.2%; Fig. 7B) sequences. was much higher than the percentage of cells expressing We also examined for the presence of proviral sequences CKR-5 (9.7%; Fig. 7E). As positive controls, cell lines using DNA-based PCR in some of the samples. These data HOS-CD4 and transduced 293 cells were used for CXCR-4 are shown in Fig. 9. Although all samples tested clearly and CKR-5, respectively (Fig. 7C and F). showed a ␤-actin band, RT sequences were not ampliﬁed in any of the samples, even using sensitive chemiluminescent PCR analysis for HIV sequences detection. CD34+ and MNC fractions, as well as dozens of RT-PCR using nested primers for HIV-RT was per- MC colonies tested, were also all negative for HIV produc- formed on HIV + BM samples 5 (MNC only), 6 (MNC and tion by the p24 assay (data not shown).

BB&MT 167 genic cells derived from HIV + BM. In one study, the number of CFU-GM and BFU-E colonies derived from BM of six HIV-infected patients was found to be unaltered [11]. In other studies, the numbers of CFU-GM [13,38,39] and megakaryocyte progenitors [39] were found to be significantly reduced. Differences in these studies might be explained by limited patient samples and the disease stage of the patients. A study by Kearns et al. [23] indicated that the percentage of CD34+ cells and the clonogenic potential of these cells were reduced in patients with low CD4 counts but not in patients with higher CD4 counts. Marandin et al. [22] reported that while CD34+CD38+ cells were not reduced in number or potential as judged by long-term culture initiating cell (LTC-IC) assays, the more primitive CD34+CD38– cells were in fact impaired. All of the patients in this study had CD4 counts <300. In a study by Junker et al. [24] with asymptomatic HIV-infected donors, hematopoietic cells from mobilized blood were not impaired. Thus, other recent studies are in agreement with our data present- ed here, suggesting that impairment of hematopoiesis depends on the disease stage of the patient. Since antiretro- Figure 6. CD34+ bone marrow cells expressed HIV-1 second core- viral drugs are known to affect hematopoietic cells, as more ceptor sequences patients receive the now-standard triple combination thera- RT-PCR for expression of CXCR-4 (top), CKR-5 (middle), and ␤-actin (bot- py, more consistency in the data for the different patient tom). Cells were total mononuclear preparations (MNC) or CD34-selected populations should be forthcoming. (CD34). +/– refers to whether reverse transcriptase was included in the reac- To our knowledge, this is the first report showing a + + tion. Control cells were either freshly isolated or PHA + IL-2 blasted peripher- significant depletion of the CD34 CD4 subpopulation in + al blood mononuclear cells, and the HOS-CD4 cell line. MilliQ (MQ) water late-stage HIV patients. We present data here, consistent was included as a negative control. Sizes are indicated by the 123-bp ladder. with the findings of others [2–4,6,7], that this subpopulation displays a phenotype likely to represent a more immature cell than the CD34+CD4– cell, based here on clonogenic assays and low expression of HLA-DR. Louache et al. [3] showed that the CD34+CD4+CD38lowHLA-DRlow subpopulation was DISCUSSION enriched for LTC-IC compared with the CD34+CD4– sub- Our major finding was a highly significant depletion of population. Similarly, Muench et al. [6] and Zauli et al. [2] the subpopulation of CD34+CD4+ cells in late-stage HIV + showed that these cells have an immature phenotype and infection. There was also a statistically nonsignificant trend higher proliferative potential than CD34+CD4– cells [6]. We toward diminished recovery of CD34+ cells and impaired have also confirmed low expression of CD38 on CD34+ cells clonogenic efficiencies in BM of HIV + patients. In the seven derived from cord blood (not shown). BM samples examined for CD34+CD4+, we found the per- HIV effects on these early hematopoietic cells might centage of this subpopulation to correlate positively with the translate into impaired differentiation of multiple lineages CD4 count of the patient. One patient donated two separate that derive from CD34+CD4+ cells. Thus, suppressive BM samples 6 months apart (samples 1 and 5). At the time effects of HIV-1 on hematopoietic stem/progenitor cells sample 1 was taken, the patient was acutely ill with non- have important implications for disease pathogenesis and Hodgkin’s lymphoma (NHL) and had a CD4 count of 27. therapeutic strategies. In addition to CD4 helper cell deple- The percentage of CD34+ cells that were also CD4+ was 7% tion, which is a hallmark of HIV infection, a variety of at that time. Six months later, the patient’s NHL was in cytopenias in other blood cell lineages characterized by ane- remission, and with triple antiretroviral therapy, the CD4 mia, neutropenia, and thrombocytopenia, alone or in vari- count increased to 348. This was accompanied by a slight ous combinations (pancytopenia), are consistently observed increase in the percentage of CD34+CD4+ cells, to 10.5%. in HIV-infected patients. The incidence of cytopenias This increase is interesting, but whether or not it is signifi- increases with the stage of the disease [40], and up to 80% cant is unclear. If an increase in CD4 count is in fact corre- of patients in late-stage disease manifest these conditions lated with an increase in CD34+CD4+ cells, this subpopula- [41]. A plausible explanation for cytopenias of several lin- tion might be renewable even in late-stage HIV + patients. eages may lie in an impaired proliferation/differentiation Continued analysis of multiple samples from this and other capacity of the hematopoietic stem cell/progenitor compart- patients will be necessary to determine whether such a cor- ment. Impaired replenishment of CD4+ cells as well as other relation exists and whether an increase in CD34+CD4+ cells lineages could occur if CD34+ cells are directly or indirectly is correlated with positive clinical outcomes. affected by HIV-1. Other investigators have also described some level of The role of the CD4 marker on CD34+ cells has not yet impairment of CD34+ cells in HIV + patients. Various studies been elucidated. It has been postulated [2] that these cells have reported alteration in the number or type of clono- might be capable of interacting with class II MHC-positive

168 Depleted CD34 CD4 Cells n AIDS

A B C

D E F

Figure 7. Surface expression of CXCR-4 and CKR-5 on CD34+ bone marrow cells Matched isotype controls for CXCR-4 (A) and CKR-5 (D). Expression of CXCR-4 (B) and CKR-5 (E) on CD34+ cells. Positive single-stain controls using HOS- CD4 (C) and transduced 293 (F) cells. Isotype controls for C and F are indicated in the ﬁgures at 2.0 and 1.8%, respectively.

bone marrow accessory cells; this would be consistent with the later stages of disease and often have high viral burdens the role of CD4 on T cells. Expression of CD4 might enable (which would depend on treatment regimen as well). The these cells to interact with other cell types in the bone mar- high viral burden means that more viral particles would be in row microenvironment. Alternatively (or in addition), the the bone marrow environment to affect the CD34+CD4+ CD34+CD4+ cell might be a precursor to mature CD4+ cells cells, either directly or indirectly. that would migrate into the thymus. While this has not been A critical question that remains unanswered is whether demonstrated in in vitro and in vivo systems that mimic thy- early hematopoietic cells are susceptible to HIV infection at mopoiesis, it is possible that the conditions needed for matu- any stage. Our data show that BM cells (albeit from relative- ration of the earliest stem cells have not been met in these ly healthy HIV + donors) did not contain HIV-1 DNA or systems. If CD34+CD4+ cells in the bone marrow do give rise express HIV sequences. Studies from other laboratories to mature CD4+ cells (which is not presently clear), then have addressed this question in two different ways. One way alterations in the CD34+CD4+ subpopulation by HIV could has been to expose CD34+ cells to HIV in vitro and deter- translate into reduced CD4 counts in the periphery. An alter- mine whether methylcellulose colonies from the cells con- native possibility is that patients with low CD4 counts are in tain HIV sequences. Such studies detected low to absent

BB&MT 169 Figure 8. No evidence for HIV sequences in CD34+ and MNC from HIV-infected patients Nested RT-PCR for expression of HIV-RT sequences is shown. Cells were total mononuclear preparations (MNC) or CD34-selected (CD34). +/- refers to whether reverse transcriptase was included in the reaction. Controls corresponded to 1 and 10 copies per genome of HIV-RT, as preciously described [35]. MilliQ (MQ) water was included as a negative control. Sizes are indicated by the 123-bp ladder.

levels of infection in the colonies [21,42]. The second in CD34+ cells. Alternatively, a cellular factor not present in approach has been to directly characterize CD34+ cells from these cells might be required for sustained viral production. patients with HIV infection [20,22–24]. These studies also Thus, the virus could get into the cell but would not repli- pointed to a very low or absent level of infection of CD34+ cate efficiently. This idea is supported by a recent study cells, although very low levels were detected in MNC cells from Shen et al. [43]. before CD34 selection [23]. Nonetheless, it is still possible Finally, the possibility exists that the CD34+CD4+ cells that CD34+CD4+ cells are susceptible to HIV infection at are not infected by HIV but are indirectly impaired or some stage, and that either they do not sustain viral produc- deleted by the virus or viral proteins. Consistent with that tion or they are rapidly killed by the virus and therefore viral sequences are not detected. The hypothesis that a subset of CD34+ cells might be susceptible to HIV infection is supported by the finding (reported here and by others) that these cells express message for the HIV coreceptors CXCR-4 and CKR-5. We also found that the cells expressed coreceptor protein on their surface. Using cells from HIV – donors, Deichmann et al. [7] showed that five of eight and two of eight CD34+CD4+ purified cells expressed CXCR-4 and CKR-5 message, respectively, while seven of eight and one of eight CD34+CD4– cells expressed CXCR-4 and CKR-5 message, respectively. Our results with CXCR-4 are in general agreement with Deichmann et al., showing expression on most samples of CD34+ cells. However, we also found expression of CKR-5 in all MNC as well as CD34+ samples, while Deichmann et al. found expression on only three of 16 total CD34+ samples examined. However, our FACS analysis shown in Fig. 7 confirms that a relatively low percentage of the population actually express the CKR-5 protein on the cell surface. This is consistent with the observation of Deichmann et al. While the CD34+-enriched populations were less than 100% pure, we believe the coreceptor expression was on CD34+ cells and not contaminating cells for several reasons. First, in Fig. 6, CD34+ samples 6 and 7 show a very similar signal intensity even though the CD34+ purities were different (98.5 and 86.4%, respectively; contaminating cells would thus represent 1.5 and 13.6%). Second, the FACS analysis showing expression of the protein on the surface of CD34+ cells is consistent with message expression in these cells. Since these cells express HIV coreceptor along with Figure 9. No evidence of HIV proviral sequences in CD34+ cells and CD4, they should theoretically be targets for HIV infection; MNC from HIV-infected patients however, high levels of HIV have not been reported in these DNA-based PCR for HIV-RT and ␤-actin is shown. Cell samples were a sub- cells. Two possibilities could explain this. First, another set of those shown in Figs. 6 and 8. Controls corresponded to 1 and 10 copies per coreceptor required for HIV entry might not be expressed genome of HIV-RT. MilliQ (MQ) water was included as a negative control.

170 Depleted CD34 CD4 Cells n AIDS notion, we have demonstrated that CD34+ cells can specifi- REFERENCES cally bind gp120 and that the cells undergo apoptosis as a 1 Wainberg MA, Margolese RG: Strategies in the treatment of AIDS result. Apoptosis has been shown in both CD4+ and CD8+ and related diseases: the lessons of cancer chemotherapy. Cancer Invest T lymphocytes from HIV-infected individuals [44]. In 10:143, 1992 addition, crosslinking CD4 with gp120 and anti-gp120 2 Zauli G, Furlini G, Vitale M, Re MC, Gibellini D, Zamai L, Visani G, upregulates expression of fas antigen in normal lympho- Borgatti P, Capitani S, Placa ML: A subset of human CD34+ hemato- cytes [8,45], consistent with the idea that gp120-CD4 poietic progenitors express low levels of CD4, the high-affinity recep- interactions on the surface of CD4+ cells mediate apoptosis tor for human immunodeficiency virus-type 1. Blood 84:1896, 1994 [46]. It has recently been reported that apoptotic, uninfect- 3 Louache F, Debili N, Marandin A, Coulombel L, Vainchenker W: ed CD4+ T cells in HIV-infected lymph nodes carry gp120 Expression of CD4 by human hematopoietic progenitors. Blood on their surface [47]. Another possibility is that infection 84:3344, 1994 of adjacent cells such as stromal cells in the bone marrow 4 Moore TA, Zlotnik A: T-cell lineage commitment and cytokine microenvironment might indirectly cause cell death of responses of thymic progenitors. Blood 86:1850, 1995 progenitor cells by a similar mechanism as reported by 5 Cleveland RP, Liu Y-C: CD4 expression by erythroid precursor Badley et al. [48], where infected macrophage cells express- cells in human bone marrow. Blood 87:2275, 1996 ing fas ligand cause apoptosis of peripheral CD4 cells. An 6 Muench MO, Roncarolo MG, Namikawa R: Phenotypic and func- indirect effect on early hematopoietic progenitors was pos- tional evidence for the expression of CD4 by hematopoietic stem cells tulated in a recent study by Jenkins et al. [49] in the SCID- isolated from human fetal liver. Blood 89:1364, 1997 hu Thy/Liv model. In that study, depletion of progenitors 7 Deichmann M, Kronenwett R, Haas R: Expression of the human was demonstrated in SCID-hu mice grafts infected with immunodeficiency virus type-1 coreceptors CXCR-4 (fusin,LESTR) HIV-1, followed by a decline in thymocyte cell number. and CKR-5 in CD34+ hematopoietic progenitor cells. Blood 89:3522, Since proviral genomes were not detected in hemato- 1997 poietic cells, the authors concluded that HIV acted indi- 8 Banda NK, Tomczak JA, Shpall EJ, Sipple J, Akkina RK, Steimer KS, rectly on hematopoietic cells, resulting in reduced colony- Hami L, Curiel TJ, Harrison GS: HIV-gp120 induced cell death in forming units and, down the road, loss of mature thymo- hematopoietic progenitor CD34+ cells. Apoptosis 2:61, 1997 cytes. These data are entirely consistent with the studies 9 Zauli G, Vitale M, Gibellini D, Capitani S: Inhibition of purified we report here. CD34+ hematopoietic progenitor cells by human immunodeficiency The exact mechanism of depletion of the CD34+CD4+ virus 1 or gp120 mediated by endogenous transforming growth factor subpopulation in late-stage AIDS patients is not yet clear. As ␤1. J Exp Med 183:99, 1996 discussed above, depletion could occur through direct inter- 10 Zauli G, Capitani S: HIV-1-related mechanisms of suppression of action and infection with HIV or indirect mechanisms such CD34+ hematopoietic progenitors. Pathobiology 64:53, 1996 as apoptosis. Impairment of an early progenitor cell would 11 Molina JM, Scadden DT, Sakaguchi M, Fuller B, Woon A, Groopman be manifested in defective thymopoiesis and peripheral JE: Lack of evidence for infection of or effect on growth of hemato- cytopenias such as observed in AIDS patients and would also poietic progenitor cells after in vivo or in vitro exposure to human likely contribute to impaired immune reconstitution after immunodeficiency virus. Blood 76:2476, 1990 effective antiretroviral therapies. Our finding also suggests 12 Davis BR, Schwartz DH, Marx JC, Johnson CE, Berry JM, Lyding J, that gene therapy approaches that use these cells as targets Merigan TC, Zander A: Absent or rare human immunodeficiency virus for gene transfer and hematopoietic reconstitution in infection of bone marrow stem/progenitor cells in vivo. J Virol 65:1985 patients need to be further evaluated. On one hand, the evi- 13 De Luca A, Teofili L, Antinori A, Lovino MS, Mencarini P, Visconti E, dence to date (low HLA-DR expression, BFU-E colony- Tamburrini E, Leone G, Ortona L: Haemopoietic CD34+ progenitor cells forming ability) suggests that these cells are early cells, and are not infected by HIV-1 in vivo but show impaired clonogenesis. Br J so they might be a good choice for gene therapy in terms of Haematol 85:20, 1993 multilineage repopulation. On the other hand, our data 14 Folks TM, Kessler SW, Orenstein JM, Justement JS, Jaffe ES, Fauci show that these cells are either directly or indirectly affected AS: Infection and replication of HIV-1 in purified progenitor cells of by HIV, and so using these cells in late-stage patients might normal human bone marrow. Science 242:919, 1988 not be fruitful. Ultimately, the exact function of the 15 Zucker-Franklin D, Cao Y: Megakaryocytes of human immunodefi- CD34+CD4+ cell and its role in the pathogenesis of AIDS ciency virus-infected individuals express viral RNA. Proc Natl Acad Sci needs to be defined before effective gene therapy approach- U S A 86:5595, 1989 es can be devised. 16 Zucker-Franklin D, Semeretis S, Zheng ZY: Internalization of human immunodeficiency virus type 1 and other retroviruses by megacary- ocytes and platelets. Blood 75:1920, 1990 ACKNOWLEDGMENTS 17 Kitano K, Abbourd CN, Ryan DH, Quan SG, Baldwin GC, Golde DW: The authors gratefully acknowledge Karen Helm and Macrophage-active colony-stimulating factors enhance human immun- the University of Colorado Cancer Center for FACS analy- odeficiency virus type 1 infection in bone marrow stem cells. Blood sis and media preparation, Dr. James Murphy for assistance 77:1699, 1991 with statistical analysis, and Dr. Edward Berger (National 18 Moses AV, Williams S, Heneveld ML, Strussenberg J, Rarick M, Love- Institute of Health) for helpful discussions. We also thank less M, Bagby G, Nelson JA: Human immunodeficiency virus infection of Lila Orr for assistance with obtaining HIV + bone marrow bone marrow endothelium reduces induction of stromal hematopoietic samples. HOS-CD4 cells were obtained through the AIDS growth factors. Blood 87:919, 1996 Research and Reference Reagent Program from Dr. 19 Cen D, Zauli G, Szarnicki R, Davis BR: Differential effect of human Nathaniel Landau. immunodeficiency virus type 1 isolates on bone marrow hematopoiesis.

BB&MT 171 Br J Haematol 85:596, 1993 JP: Role of fas ligand in the mechanism of T-cell depletion in acquired 20 Neal TF, Holland HK, Baum CM, Villinger F, Ansari AA, Saral R, immunodeficiency syndrome: effect on CD4+ lymphocyte depletion Wingard JR, Fleming WH: CD34+ progenitor cells from asymptomatic and human immunodeficiency virus replication. Blood 89:1357, 1997 patients are not a major reservoir for human immunodeficiency virus-1. 35 Dinges MM, Cook DR, King J, Curiel TJ, Zhang XC, Harrison GS: Blood 86:1749, 1996 HIV-regulated diphtheria toxin A chain gene confers long-term protec- 21 Mosca JD, Kaushal S, Davis T, Gartner S, Kessler S, LaRussa V, Hall tion against HIV type 1 infection in the human promonocytic cell line E, Yu Z, Ritchey D, Xu J, Perera P, Lee K, Carroll R, Kim J, Louis DS, U937. Human Gene Ther 6:1437, 1995 June C, Burke D: Pluripotent CD34+ stem cells are not susceptible to 36 Harrison GS, Long CJ, Curiel TJ, Maxwell F, Maxwell IH: Inhibi- HIV infection (Abstract). J Cell Biochem (Suppl. 21A):397, 1995 tion of human immunodeficiency virus-1 production resulting from 22 Marandin A, Katz A, Oksenhendler E, Tulliez M, Picard F, transduction with a retrovirus containing an HIV-regulated diphtheria Vainchenker W, Louache F: Loss of primitive hematopoietic progenitors toxin A chain gene. Human Gene Ther 3:461,1992 in patients with human immunodeficiency virus infection. Blood 37 Sekhar M, Kotani H, Doren S, Agarwal R, McGarrity G, Dunbar CE: 88:4568, 1996 Retroviral transduction of CD34-enriched hematopoietic progenitor 23 Kearns K, Bahner I, Bauer G, Wei S-F, Valdez P, Wheeler S, Woods L, cells under serum-free conditions. Human Gene Ther 7:33, 1996 Miller R, Casciato D, Galpin J, Church J, Kohn DB: Suitability of bone 38 Zauli G, Re MC, Visani G, Furlini G, Mazza P, Vignoli M, Placa marrow from HIV-1 infected donors for retrovirus-mediated gene ML: Evidence for a human immunodeficiency virus type 1-mediated transfer. Human Gene Ther 8: 301:1997 suppression of uninfected hematopoietic (CD34+) cells in AIDS 24 Junker U, Moon JJ, Kalfoglou CS, Sniecinski I, Forman SJ, Zaia JA, patients. J Infect Diseas 166:710, 1992 Kaneshima H, Bohnlein E: Hematopoietic potential and retroviral trans- 39 Zauli G, Re MC, Davis B, Sen L, Visani G, Gugliotta L, Furlini G, duction of CD34+Thy-1+ peripheral blood stem cells from asympto- Placa ML: Impaired in vitro growth of purified (CD34+) hematopoietic matic human immunodeficiency virus type-1-infected individuals mobi- progenitors in human immunodeficiency virus-1 seropositive thrombo- lized with granulocyte colony-stimulating factor. Blood 89:4299, 1997 cytopenic individuals. Blood 79:2680, 1992b 25 Feng Y, Broder CC, Kennedy PE, Berger EA: HIV-1 entry cofactor: 40 Hambleton J: Hematologic complications of HIV infection. Oncol- functional cDNA cloning of a seven-transmembrane, G protein-cou- ogy 10:671, 1992 pled receptor. Science 272:872, 1996 41 Zon LI, Arkin C, Groopman JE: Hematological manifestations of 26 Endres MJ, Clapham PR, Marsh M, Ahuja M, Turner JD, McKnight human immunodeficiency virus (HIV). Br J Hematol 66:251, 1987 A, Thomas JF, Stoebenau-Haggarty B, Choe S, Vance PJ, Wells TN, Powers 42 Kaushal S, LaRussa VF, Gartner S, Kessler S, Perfetto S, Yu Z, Ritchey CA, Sutterwala SS, Doms RW, Landau NR, Hoxie JA: CD4-independent DW, Xu J, Perera P, Kim J, Reid T, Mayers DL, Louis DS, Mosca JD: infection by HIV-2 is mediated by fusin. Cell 87:745, 1996 Exposure of human CD34+ cells to human immunodeficiency virus type 27 Federsppiel B, Melhado IG, Duncan AMV, Delaney A, Schapert K, 1 does not influence their expansion and proliferation of hematopoietic Clark-Lewis I, Jirik FR: Molecular cloning of the cDNA and chromoso- progenitors in vitro. Blood 88:130, 1996 mal localization of the gene for a putative seven-transmembrane seg- 43 Shen H, Cheng T, Preffer FI, Dombkowski D, Tomasson MH, Golan ment (7-TMS) receptor isolated from human spleen. Genomics 16:707, DE, Yang O, Hofmann W, Sodroski JG, Luster AD, Scadden DT: Intrinsic 1993 human immunodeficiency virus type 1 resistance of hematopoietic stem 28 Jazin EE, Yoo H, Blomquist AG, Yee F, Weng G, Walker MW, Salon cells despite coreceptor expression. J Virology 73:728, 1999 J, Larhammer D, Wahlestest C: A proposed bovine neuropeptide Y 44 Meyaard L, Otto SA, Jonker RR, Mijnster MJ, Keet RPM, Miedema (NPY) receptor cDNA clone, or its human homologue, confers neither F: Programmed cell death of T cells in HIV-1 infection. Science NPY binding sites nor NPY responsiveness on transfected cells. Regul 257:217, 1992 Pept 47:247, 1993 45 Oyaizu N, McCloskey TW, Than S, Hu R, Kalyanaraman VS, Pawha 29 Loetscher M, Geiser T, O’Reilly T, Zwahlen R, Baggiolini M, Moser B: S: Cross-linking of CD4 molecules upregulates Fas antigen expression Cloning of a human seven-transmembrance domain receptor, LESTR, in lymphocytes by inducing interferon-gamma and tumor necrosis fac- that is highly expressed in luekocytes. J Biol Chem 289:232, 1994 tor-alpha secretion. Blood 84:2622, 1994 30 Dragic T, Litwin V, Allaway GP, Martin SR, Huang Y, Nagashima K 46 Banda NK, Bernier J, Kurahara DK, Kurrle R, Haigwood N, Sekaly R A, Cayanan C, Maddon PJ, Koup RA, Moore JP, Paxton WA: HIV-1 entry P, Finkel TH: Crosslinking CD4 by human immunodeficiency virus into CD4+ cells is mediated by the chemokine receptor CC-CKR-5. gp120 primes T cells for activation-induced apoptosis. J Exp Med Nature 381:667, 1996 176:1099, 1992 31 Deng R, Liu R, Ellmeier W, Choe S, Unutmaz D, Burkhart M, 47 Sunila I, Vaccarezza M, Pantaleo G, Fauci AS, Orenstein JM: gp120 DiMarzio P, Marmon S, Sutton RE, Hill CM, Davis CB, Peiper SC, Schall is present on the plasma membrane of apoptotic CD4 cells prepared TJ, Littman DR, Landau NR: Identification of a major co-receptor for from lymph nodes of HIV-1-infected individuals: an immunoelectron primary isolates of HIV-1. Nature 381:661, 1996. microscopic study. AIDS 11:27, 1997 32 Alkhatib G, Combadiere C, Broder CC, Feng Y, Kennedy PE, Murphy 48 Badley AD, Dockrell D, Simpson M, Schut R, Lynch DH, Leibson P, PM, Berger EA: CC CKR5: a Rantes, MIP-1␣, MIP-1␤, receptor as a Paya CV: Macrophage-dependent apoptosis of CD4+ T lymphocytes fusion cofactor for macrophage-tropic HIV-1. Science 272:1955, 1996 from HIV-infected individuals is mediated by FasL and tumor necrosis 33 McCloskey TH, Oyaizu N, Kaplan M, Pahwa S: Expression of the fas factor. J Exp Med 185:55, 1997 antigen in patients infected with human immunodeficiency virus. 49 Jenkins M, Hanley MB, Moreno MB, Wieder E, McCune JM: Human Cytometry (Comm Clin Cyto) 22:111, 1995 immunodeficiency virus-1 infection interrupts thymopoiesis and multi- 34 Sloand EM, Young NS, Kumar P, Weichold FF, Sato T, Maciejewski lineage hematopoiesis in vivo. Blood 91:2672, 1998

172