The Cell Tropism of Human Immunodeficiency Virus Type 1 Determines the Kinetics of Plasma Viremia in SCID Mice Reconstituted with Human Peripheral Blood Leukocytes

Abstract

Most individuals infected with human immunodeficiency virus type 1 (HIV-1) initially harbor macrophage-tropic, non-syncytium-inducing (M-tropic, NSI) viruses that may evolve into T-cell-tropic, syncytium-inducing viruses (T-tropic, SI) after several years. The reasons for the more efficient transmission of M-tropic, NSI viruses and the slow evolution of T-tropic, SI viruses remain unclear, although they may be linked to expression of appropriate chemokine coreceptors for virus entry. We have examined plasma viral RNA levels and the extent of CD4⁺ T-cell depletion in SCID mice reconstituted with human peripheral blood leukocytes following infection with M-tropic, dual-tropic, or T-tropic HIV-1 isolates. The cell tropism was found to determine the course of viremia, with M-tropic viruses producing sustained high viral RNA levels and sparing some CD4⁺ T cells, dual-tropic viruses producing a transient and lower viral RNA spike and extremely rapid depletion of CD4⁺ T cells, and T-tropic viruses causing similarly lower viral RNA levels and rapid-intermediate rates of CD4⁺ T-cell depletion. A single amino acid change in the V3 region of gp120 was sufficient to cause one isolate to switch from M-tropic to dual-tropic and acquire the ability to rapidly deplete all CD4⁺ T cells.

The envelope gene of human immunodeficiency virus type 1 (HIV-1) determines the cell tropism of the virus (11, 32, 47, 62), the use of chemokine receptors as cofactors for viral entry (4, 17), and the ability of the virus to induce syncytia in infected cells (55, 60). Cell tropism is closely linked to but probably not exclusively determined by the ability of different HIV-1 envelopes to bind CD4 and the CC or the CXC chemokine receptors and initiate viral fusion with the target cell. Macrophage-tropic (M-tropic) viruses infect primary cultures of macrophages and CD4⁺ T cells and use CCR5 as the preferred coreceptor (2, 5, 15, 23, 26, 31). T-cell-tropic (T-tropic) viruses can infect primary cultures of CD4⁺ T cells and established T-cell lines, but not primary macrophages. T-tropic viruses use CXCR4 as a coreceptor for viral entry (27). Dual-tropic viruses have both of these properties and can use either CCR5 or CXCR4 (and infrequently other chemokine receptors [25]) for viral entry (24, 37, 57). M-tropic viruses are most frequently transmitted during primary infection of humans and persist throughout the duration of the infection (63). Many, but not all, infected individuals show an evolution of virus cell tropism from M-tropic to dual-tropic and finally to T-tropic with increasing time after infection (21, 38, 57). Increases in replicative capacity of viruses from patients with long-term infection have also been noted (22), and the switch to the syncytium-inducing (SI) phenotype in T-tropic or dual-tropic isolates is associated with more rapid disease progression (10, 20, 60). Primary infection with dual-tropic or T-tropic HIV, although infrequent, often leads to rapid disease progression (16, 51). The viral and host factors that determine the higher transmission rate of M-tropic HIV-1 and the slow evolution of dual- or T-tropic variants remain to be elucidated (4).

These observations suggest that infection with T-tropic, SI virus isolates in animal model systems with SCID mice grafted with human lymphoid cells or tissue should lead to a rapid course of disease (1, 8, 44–46). While some studies in SCID mice grafted with fetal thymus and liver are in agreement with this concept (33, 34), our previous studies with the human peripheral blood leukocyte-SCID (hu-PBL-SCID) mouse model have shown that infection with M-tropic isolates (e.g., SF162) causes more rapid CD4⁺ T-cell depletion than infection with T-tropic, SI isolates (e.g., SF33), despite similar proviral copy numbers, and that this property mapped to envelope (28, 41, 43). However, the dual-tropic 89.6 isolate (19) caused extremely rapid CD4⁺ T-cell depletion in infected hu-PBL-SCID mice that was associated with an early and transient increase in HIV-1 plasma viral RNA (29). The relationship between cell tropism of the virus isolate and the pattern of disease in hu-PBL-SCID mice is thus uncertain. We have extended these studies by determining the kinetics of HIV-1 RNA levels in serial plasma samples of hu-PBL-SCID mice infected with primary patient isolates or laboratory stocks that differ in cell tropism and SI properties. The results showed significant differences in the kinetics of HIV-1 replication and CD4⁺ T-cell depletion that are determined by the cell tropism of the virus isolate.

MATERIALS AND METHODS

Virus.

HIV-1_SF2, HIV-1_SF162, HIV-1_JR-FL, HIV-1_JR-CSF, and HIV-1_89.6 have been described previously (13, 19, 35, 36). HIV-1₂₄₁ and HIV-1₂₄₂ are recently described molecular clones that differ only by a glutamine (HIV-1₂₄₁)-to-glutamic acid (HIV-1₂₄₂) change at position 25 of the V3 loop (14). HIV-1₂₄₁ is dual-tropic and SI, while HIV-1₂₄₂ is M-tropic and non-syncytium inducing (NSI) (14, 58). The two virus isolates showed similar kinetics of virus replication in activated peripheral blood mononuclear cell (PBMC) cultures (p24 antigen levels on days 4 and 8 of culture were as follows: for HIV-1₂₄₁, 130 and 18,367 pg/ml and for HIV-1₂₄₂, 90 and 9,390 pg/ml). Primary patient isolates CS93, CD65, and MT82 are M-tropic, T-tropic, and T-tropic, respectively (52), and were used after a single round of in vitro expansion. hu-PBL-SCID mice were infected with 10³ tissue culture infectious doses (TCID) by intraperitoneal injection. TCID was determined by limiting dilution of virus stocks with phytohemagglutinin and interleukin 2 (IL-2)-activated human PBMC.

Mice.

C.B-17 scid/scid mice were bred in a closed, specific-pathogen-free environment at The Scripps Research Institute. Mice were screened for mouse immunoglobulin at 6 to 8 weeks of age, and animals with >10 μg/ml were discarded as “leaky” (9). Human PBMC were prepared by density separation from normal adult donors who were Epstein-Barr virus seronegative and who were demonstrated by PCR to have normal CCR5 coding regions (52). A single normal human donor was used for each of four experiments, and no donor was used twice. The experiment shown in Fig. 5 employed a donor who was heterozygous for the 32-bp deletion in CCR5 (48). A total of 20 × 10⁶ cells were injected intraperitoneally into SCID mice 2 weeks prior to virus exposure, as described previously (28, 41). Each experimental group consisted of 3 to 5 mice, and most virus isolates were compared in at least two experiments.

FIG. 5.

Plasma HIV RNA copy number and recovery of CD4⁺ T cells in hu-PBL-SCID mice generated from a single donor who is heterozygous for the CCR5 Δ32 mutation. Five hu-PBL-SCID mice were infected with either HIV-1₂₄₁ or HIV-1₂₄₂ (as in the experiment shown in Fig. 4). The geometric mean RNA copy number (± relative SE) for each group of mice is shown for 1 to 3 weeks after infection (A). Three mice from each group were used to determine the extent of CD4⁺ T-cell depletion at 2 weeks after infection (B). Samples of peritoneal lavage cells (PC) were assayed from individual mice, and the mean ± SE is shown. Samples from lymph nodes (LN) were pooled prior to analysis, so no SE is displayed.

Virus infection.

Plasma virus RNA copy number was determined by quantitative PCR assay (Amplicor; Roche Molecular Systems, Somerville, N.J.). Plasma samples from multiple time points of infection were frozen and subsequently assayed at the same time to minimize interassay variation. Virus infection of animals was confirmed by isolation of virus in cocultures of activated human PBMC and cells recovered from hu-PBL-SCID mice by peritoneal lavage or preparation of cell suspensions from spleens or local lymph nodes or by amplification of proviral gag sequences by PCR (45).

Flow cytometry.

Recovery of human cells and CD4⁺ T-cell depletion was monitored by flow cytometry. Briefly, recovered cells were stained with fluorescein- or phycoerythrin-labeled monoclonal antibodies to murine H-2K^d and human CD45, CD3, CD4, CD8, CD25, CD69, CD45RA, and CD45RO. Antibodies were obtained either from immunocytometry systems (Becton Dickinson, Mountain View, Calif.) or Pharmingen (San Diego, Calif.). Staining was evaluated for a minimum of 10⁴ cells with a FACscan (Becton Dickinson) flow cytometer, and data were analyzed with Cellquest (Becton Dickinson) software. Data are presented as the mean percentage of CD4⁺ T cells as a fraction of total CD3⁺ T cells, standardizing the result for variable recovery of total human T cells in different sites and different animals. The ratio of CD4⁺ to CD3⁺ T cells is reported separately for human cells recovered by peritoneal lavage and from pooled samples of local (mesenteric and periportal) lymph nodes. These lymph nodes have a higher proportion of CD45RA⁺ T cells (see Fig. 5) and show kinetics of CD4⁺ T-cell depletion that differ from those of human cells recovered by peritoneal lavage (reference 29 and below). In HIV-1-infected mice with CD4⁺ T-cell depletion, the number of CD8⁺ T cells was always close to or equal to the number of CD3⁺ T cells, indicating a loss of T cells and not CD4 modulation (data not shown).

RESULTS

Selection of HIV-1 isolates differing in cell tropism.

The cell tropism and V3 sequences of the HIV-1 isolates used in these experiments are presented in Table 1. The origin and V3 sequences of the primary patient isolates CS93 and CD65 have recently been reported elsewhere (52). The T-tropic MT82 isolate was recovered from a patient with hemophilia who was asymptomatic at the time of isolation (>10 years after infection) but has subsequently progressed to AIDS. Each of these virus isolates replicated well in primary cultures of PBMC or purified CD4⁺ T cells, and none of the M-tropic isolates were able to replicate in PBMC from donors homozygous for the CCR5 32-bp deletion (reference 52 and data not shown). By contrast, T-tropic isolates replicated well in MT-2 cells and PBMC from CCR5-negative donors (52).

TABLE 1.

V3 sequences for primary and laboratory HIV-1 isolates used in these experiments

Virus isolate	Sequence	Tropism
cons	CTRPNNNTRK--SIHIGPGRAFY----TTGE-IIGDIRQAHC
CS93	----------xx-----------xxxxA--Dx---N------	M, NSI
CD65	----------xxG--------V-xxxxA-DRx----------	T, SI
MT82	---LG-----xx--R----PGR-NTVF---DVT---------	T, SI
SF2	----------xx--Y-------Hxxxx---Rx------K---	T, SI
SF162	----------xx--T--------xxxxA--Dx----------	M, NSI
89.6	---------RRL--xx-------xxxxARRNx----------	M/T, SI
JR-FL	----------xx-----------xxxx----x----------	M, NSI
JR-CSF	----S-----xx-----------xxxx----x----------	M, NSI
242	---------Rxx--S-------Rxxxx--x-x----------	M, NSI
241	---------Rxx--S-------Rxxxx--xQx----------	M/T, SI

Kinetics of plasma viremia following HIV-1 infection.

Each HIV-1 isolate was used to infect multiple hu-PBL-SCID mice in one or more of four replicate experiments, depicted in Fig. 1 through 4. Weekly samples of plasma were used for the determination of viral RNA copy number, and results are presented for individual animals over the duration of each experiment. In experiments 1 and 4, the number of human CD4⁺ T cells was also determined and compared to that in uninfected control animals. In the experiment shown in Fig. 1, we infected mice with one of two T-tropic isolates, SF2 or CD65, or one of two M-tropic isolates, SF162 or CS93. Infection with T-tropic isolates leads to transient viral RNA expression between 1 and 4 weeks after infection, and few or no residual CD4⁺ T cells were detectable by 7 to 8 weeks after HIV-1 infection. By contrast, most mice infected with M-tropic viruses showed increasing levels of viral RNA for up to 6 weeks after infection, and residual CD4⁺ T cells were detected in all mice with high viral RNA levels, whereas declining viral loads were associated with reduced numbers of CD4⁺ T cells. The two hu-PBL-SCID mice with the highest viral RNA copy numbers after infection with HIV-1_SF162 died before the termination of the experiment. A second, similar experiment with the M/T-tropic 89.6 isolate as well as the T-tropic CD65 and the M-tropic CS93 HIV-1 isolates is shown in Fig. 2 (A through C, individual mice; D, group means). Infection with 89.6 resulted in a burst of viral RNA detected at 1 week after infection, followed by a decline to undetectable levels by 2 weeks, when depletion of CD4⁺ T cells is complete (29). Infection with the T-tropic CD65 isolate resulted in a peak of plasma viral RNA at 1 or 2 weeks after infection, with a subsequent decline to the limit of detection by 3 to 5 weeks after infection. As observed previously (Fig. 1), infection with the M-tropic CS93 isolate caused a progressive increase in viral RNA levels to a mean value of >10⁶ copies/ml. In the experiment shown in Fig. 3, the studies with 89.6 and SF2 were repeated and the T-tropic MT82 isolate and the M-tropic JR-FL isolate were added. Both the dual-tropic 89.6 and the T-tropic MT82 isolates produced peaks of plasma viral RNA at 1 week after infection, with rapid declines to baseline levels thereafter. Infection with T-tropic SF2 led to a later peak in plasma viral RNA levels, followed by a decline to baseline levels by 4 weeks after infection, when most CD4⁺ T cells have been depleted (41). Infection with the M-tropic JR-FL isolate led to a more sustained high level of virus replication that was still high in all mice after 4 weeks of infection. In the last of the four experiments (Fig. 4), the M-tropic 242 isolate, the M/T-tropic 241 isolate, and the M-tropic JR-CSF isolate were compared. As noted above, 242 and 241 differ in only a single amino acid (14, 58). Human CD4⁺ T-cell survival was measured at 2 and 4 weeks after infection in this experiment. Infection of hu-PBL-SCID mice with the dual-tropic 241 isolate caused a burst of virus production at 1 week after infection and depleted all human CD4⁺ T cells by 2 weeks after infection, a pattern similar to that seen with 89.6 infection. By contrast, infection with the M-tropic 242 isolate caused a sustained increase in viral RNA levels and caused minimal depletion of CD4⁺ T cells at either 2 and 4 weeks after infection in human cells recovered by peritoneal lavage (Fig. 4D) or from local lymph nodes (Fig. 4E). The single amino acid change that alters coreceptor usage and cell tropism also alters the kinetics of virus replication and CD4⁺ T-cell depletion. Infection with the M-tropic JR-CSF isolate caused lower but sustained and increasing levels of viral RNA and substantial depletion of CD4⁺ T cells by 4 weeks after infection.

FIG. 1.

Plasma HIV RNA copy number in hu-PBL-SCID mice infected with the T-tropic HIV-1 isolate SF2 (A), the M-tropic isolate SF162 (B), the primary T-tropic patient isolate CD65 (C), or the primary M-tropic isolate CS93 (D). Each line represents serial measurements on an individual hu-PBL-SCID mouse from 1 to 6 weeks after infection. (E) Percentage of remaining CD4⁺ T cells in the peritoneal cavity (compared to total human CD3⁺ T cells) at 7 to 8 weeks after infection. This number was determined for individual mice (symbol above each column matches symbol for each animal in panels A through D) and compared to the mean (± standard error [SE]) for four uninfected hu-PBL-SCID mice (open column with error bar). All hu-PBL-SCID mice were derived from a single human donor who was Epstein-Barr virus seronegative and homozygous wild type at the CCR5 locus. The detection limit of RNA copy number was 800 in this experiment, so samples with undetectable viral RNA were assigned a value of 800. Serial samples from individual mice were saved and compared in the same Roche HIV Monitor Amplicor assay plate, and the available volume of mouse plasma determined the cut-off value for HIV RNA detection.

FIG. 4.

Plasma HIV RNA copy number and percentage of CD4⁺ T cells in hu-PBL-SCID mice generated from a single human donor. Five hu-PBL-SCID mice were infected either with HIV-1₂₄₁, a dual-tropic, SI virus (A), or HIV-1₂₄₂, an M-tropic, NSI virus that differs only by a glutamine-to-glutamic acid change in position 25 of the V3 loop (14) (B). An additional group of five mice was infected with the M-tropic JR-CSF isolate (C). Three mice in each group were used for determination of CD4⁺ T-cell levels at 2 weeks after infection, and the remaining two mice were examined after 4 weeks of infection. (D) Percentages ± SE of CD4⁺ T cells (of total human CD3⁺ T cells) recovered by peritoneal lavage of the animals compared to mean values for four uninfected control mice. Only one hu-PBL-SCID mouse infected with HIV-1₂₄₂ was available at week 4, because the mouse with >10⁷ HIV RNA copies/ml at week 3 after infection died. CD4⁺ T cells were also enumerated in regional lymph nodes (panel E) pooled from three mice (2 weeks after infection) or two mice (4 weeks after infection). Since these samples were pooled, no SE is shown.

FIG. 2.

Plasma HIV RNA copy number in hu-PBL-SCID mice reconstituted with cells from another human donor (also EBV negative, CCR5 wild-type homozygous) and infected with the dual-tropic 89.6 HIV-1 isolate (A), the primary T-tropic isolate CD65 (B), or the primary M-tropic isolate CS93 (C). Panels A through C show the viral RNA levels in individual hu-PBL-SCID mice, and panel D shows the mean ± SE of each group of mice over the 5-week duration of the experiment. The limit of detection in this experiment was 200 copies/ml, and samples with no detectable viral RNA were assigned this value.

FIG. 3.

Plasma HIV RNA copy number in hu-PBL-SCID mice reconstituted with cells from a third human donor (also EBV negative, CCR5 wild-type homozygous) and infected with the dual-tropic 89.6 isolate (A), the T-tropic SF2 isolate (B), the T-tropic primary patient isolate MT82 (C), or the M-tropic JR-FL isolate (D). The limit of detection in this experiment was 400 copies/ml, and samples with no detectable viral RNA were assigned this value.

We performed a similar experiment with HIV-1₂₄₂ and HIV-1₂₄₁ in hu-PBL-SCID mice derived from a CCR5 Δ32/+ heterozygous donor, since previous experiments had shown delayed kinetics of M-tropic virus replication in such mice (52). The results (Fig. 5A) show a difference in the kinetics of plasma virus RNA levels similar to that observed in the experiment shown in Fig. 4, with HIV-1₂₄₁ producing a more transient viremia than HIV-1₂₄₂. Infection with 241 caused a profound decline in CD4⁺ T cells in both peritoneal cells and lymph nodes by 2 weeks after infection, whereas mice infected with 242 showed no decline in CD4⁺ T cells at this time (Fig. 5B). The reduction in CCR5 expression in this Δ32/+ heterozygous donor does not impact the distinct kinetics of replication and CD4⁺ T-cell depletion caused by the amino acid change between 242 and 241.

Potential sites of HIV-1 replication.

These results imply that infection with M-tropic virus may cause sustained viremia because of slower depletion of CD4⁺ T cells, but our earlier data suggested that M-tropic viruses caused more rapid CD4⁺ T-cell depletion than T-tropic viruses (28, 41). We repeated these earlier experiments with the M-tropic isolate SF162 and the T-tropic isolate SF2, but we evaluated survival of human CD4⁺ T cells in both peritoneal lavage cells (as before) and human cells recovered from lymph nodes draining the peritoneal cavity. The percentage of surviving human CD4⁺ T cells at 2 weeks after infection is shown in Fig. 6A. Although infection with SF162 caused a greater depletion of CD4⁺ T cells in the population of human cells recovered from the peritoneal cavity, in agreement with our earlier results, infection with SF2 caused a significantly (P < 0.05) greater depletion of human CD4⁺ T cells in the local lymph nodes. It is thus possible that the human cells in these lymph nodes are the source of continued virus replication following infection with M-tropic but not T-tropic or dual-tropic isolates. We analyzed the composition of human cells recovered from the two sites, peritoneal cavity and local lymph nodes, of uninfected hu-PBL-SCID mice by two-color flow cytometry (Fig. 6B and C). While CD3⁺ T cells represented >90% of recovered human cells (CD45⁺) in both sites, the lymph nodes had a higher percentage of CD4⁺ T cells and one-third of CD4⁺ T cells had the naive CD45RA⁺ phenotype. By contrast, over 90% of human CD4⁺ T cells recovered from the peritoneal cavity had the activated/memory CD45RO⁺ phenotype (6), and a higher percentage also expressed the CD25 IL-2 receptor, indicating recent activation. These results suggest that human CD4⁺ T cells are more activated and/or selected for memory cells in the peritoneal cavities of hu-PBL-SCID mice than in repopulated lymph nodes.

FIG. 6.

(A) Recovery of human CD4⁺ T cells from peritoneal lavage cells (filled bars) or local lymph nodes (hatched bars) of hu-PBL-SCID mice infected 2 weeks earlier with either HIV-1_SF2 or HIV-1_SF162. The numbers represent the mean ± SE of individual determinations on 4 to 5 mice per group and are expressed as a percentage of recovered total human T cells (CD3⁺). CD3⁺ T cells represented >90% of recovered human cells. Human cells represented 79 to 88% of all cells recovered from the peritoneal lavage of uninfected hu-PBL-SCID mice and from 37 to 89% of cells recovered from local lymph nodes. The numbers of recovered human cells declined in parallel with the loss of CD4⁺ T cells after HIV-1 infection in both sites. (B) Phenotype of human cells recovered from the peritoneal cavity of control, uninfected hu-PBL-SCID mice. Two-color immunofluorescence staining of cells was performed to identify memory/activated CD4 T cells, which express CD45RO, CD25 (IL-2R alpha chain), and/or CD69. (C) Phenotype of human cells recovered from the local lymph nodes draining the peritoneal cavity of hu-PBL-SCID mice and stained as in panel B.

DISCUSSION

These results show that HIV-1 isolates which differ in cell tropism and coreceptor usage give two distinct patterns of virus replication and CD4⁺ T-cell depletion in the hu-PBL-SCID model. Infection with the M-tropic isolates SF162, JR-CSF, JR-FL, 242, and CS93 resulted in high and increasing levels of plasma virus RNA over the first 4 to 6 weeks of infection, and peak levels of viral RNA often exceeded 10⁶ copies/ml (Table 2). Residual human CD4⁺ T cells were usually present in hu-PBL-SCID mice infected with M-tropic HIV-1, particularly in the lymph nodes of reconstituted mice (Fig. 6) and in those mice with the highest viremia (Fig. 1). Declining levels of viral RNA correlated with a complete or near complete loss of CD4⁺ T cells (e.g., Fig. 1D and E). Within the group of M-tropic isolates, infection with SF162 and 242 caused more rapid increases in viral RNA levels and infection with JR-CSF resulted in slower increases (e.g., Fig. 4B versus C), but these differences did not obscure the general pattern of plasma viremia associated with M-tropic viruses. The peak levels of viral RNA differed between HIV isolates (Table 2) but were generally higher than those attained following infection with M/T- or T-tropic isolates and peaked later.

TABLE 2.

Peak viral RNA levels following infection with HIV-1 isolates differing in cell tropism

Virus isolate	Cell tropism	Mean peak viral RNA copy number	Mean week peak level attained	na
CS93	M-tropic	924,790	4.71	7
JR-FL	M-tropic	255,043	4.3	3
SF162	M-tropic	2,214,337	4.25	4
CD65	T-tropic	91,807	1.8	10
SF2	T-tropic	102,557	2.67	6
89.6	M/T-tropic	48,291	1.0	7
241	M/T-tropic	18,959	1.0	5

^an, number of hu-PBL-SCID mice analyzed. 

Infection of hu-PBL-SCID mice with M/T- or T-tropic HIV-1 isolates led to a distinctive and shared pattern of viral replication and CD4⁺ T-cell depletion. Infection with the dual-tropic isolates 89.6 and 241 led to a peak level of viral RNA at 1 week after infection, and no viral RNA was detectable at 2 or more weeks after infection in most animals (Fig. 2 through 5). Infection with M/T-tropic isolates caused the loss of nearly all CD4⁺ T cells in both the peritoneal cavity and lymph nodes of infected mice within 2 weeks (Fig. 4 and 5), and the failure to continue virus replication is probably explained by the elimination of all CD4⁺ target cells for infection. The peak levels of viral RNA ranged from 10⁴ to 10⁵ copies/ml following infection with M/T-tropic isolates (Table 2). Infection of hu-PBL-SCID mice with the T-tropic isolates SF2 and CD65 caused a pattern of plasma viremia intermediate between those of M-tropic and M/T-tropic viruses (Fig. 1 through 3), but infection with the T-tropic primary isolate MT82 caused an earlier and more transient peak of viremia, similar to the pattern seen with M/T-tropic isolates (Fig. 3). Peak levels of viral RNA were observed between 1 and 3 weeks after infection, and levels tended to decline to undetectable by 4 weeks after infection. Peak viral loads varied from 10³ to nearly 10⁶ copies/ml, with the mean peak values being close to 10⁵ copies/ml (Table 2), a level intermediate between the higher values seen with M-tropic virus infection and the lower values seen with M/T-tropic virus infection, although there was considerable overlap in peak viral RNA levels between individual mice infected with either M/T- or T-tropic isolates.

The levels of plasma viral RNA attained 1 week after infection presumably reflect the efficiency of virus transmission (in this case, the result of intraperitoneal injection of 10³ TCID of cell-free infectious virus) and of the initial rounds of virus replication. We have not observed the onset of CD4⁺ T-cell depletion prior to day 9 postinfection in any of a large series of experiments with these HIV-1 isolates. Although some hu-PBL-SCID mice infected with M-tropic isolates showed very high viral RNA levels by 1 week after infection, there was no consistent significant difference in the transmission of virus isolates that segregated with cell tropism (e.g., Fig. 2D). This result suggests that M-tropic viruses have only a small advantage in transmission in this animal model. However, all hu-PBL-SCID mice challenged with M- or M/T-tropic viruses are consistently infected, whereas occasional mice challenged with T-tropic isolates fail to become infected, in agreement with the results of one recent study (39). Another study (59) indicates that SCID-hu mice transplanted with human fetal thymus and liver are more easily infected with M-tropic variants of HXSB-2 than with the parental T-tropic isolate. These studies and our observations support a possible transmission advantage for M-tropic HIV-1 isolates that was not particularly evident in the present series of experiments.

The most striking correlation seen in this series of experiments was that between the levels and duration of plasma viremia and the rate of CD4⁺ T-cell depletion. Viremia was low and transient in hu-PBL-SCID mice infected with M/T-tropic isolates when the pace of CD4⁺ T-cell depletion is most rapid. Viremia was somewhat higher and more persistent following infection with two of three T-tropic isolates, resulting in intermediate rates of CD4⁺ T-cell depletion, although the rate seemed to vary between human T cells in the peritoneal cavity (Fig. 6A, slower, and references 28 and 41) and those found in local lymph nodes (Fig. 6A, faster). Finally, viremia was sustained at high levels following infection with M-tropic isolates, resulting in slower rates of CD4⁺ T-cell depletion, particularly in the lymph nodes repopulated with both naive and memory CD4⁺ T cells (Fig. 6). This interpretation is in general agreement with the many observations that primary M-tropic HIV-1 isolates are less cytopathic and often show lower replication rates than M/T- or T-tropic isolates from late-stage patients (12, 20, 34, 56, 60, 61) and differs from our interpretation of earlier studies that suggested M-tropic HIV-1 isolates were more pathogenic than T-tropic isolates in the hu-PBL-SCID model (41, 42). These data also suggest that the switch from M-tropic to M/T-tropic HIV-1 may be more important for increasing the rate of CD4⁺ T-cell loss than the switch to a T-tropic variant (20, 21, 64). The increased rate of CD4⁺ T-cell depletion caused by the single amino acid change between HIV-1₂₄₂ and HIV-1₂₄₁ was particularly striking (Fig. 4 and 5). These two viruses show little difference in replication rate (see Materials and Methods), so the change in cell tropism and coreceptor usage appears to be the major explanation for the enhanced pathogenicity.

One of the T-tropic, SI isolates we have studied previously was SF33 (41); infection with SF33 appears to cause a more persistent infection, with lower levels of plasma viremia, than infection with other T-tropic isolates studied (data not shown). More extensive analysis of infection with this isolate is under way, but the delayed CD4⁺ T-cell depletion caused by SF33 infection seems to be isolate specific.

Recent work (7) has shown that CXCR4 is expressed primarily on naive CD4⁺ T cells and that CCR5 is expressed mainly on memory T cells. This observation may explain the different kinetics of CD4⁺ T-cell depletion following infection with M- or T-tropic HIV-1 in the two major sites of human cell reconstitution in hu-PBL-SCID mice (Fig. 6). The fraction of naive CD45RA⁺ CD4⁺ T cells in the lymph nodes of the mice would be resistant to M-tropic virus infection because of low expression of the CCR5 coreceptor, yet susceptible to T-tropic virus infection because of higher expression of the CXCR4 coreceptor. Almost all of the human CD4⁺ T cells recovered from the peritoneal cavity are of the memory or activated phenotype (Fig. 6) and thus may express less CXCR4 and more CCR5. Infection with M-tropic HIV-1 isolates may progress faster in this compartment and result in the earlier loss of CD4⁺ T cells (Fig. 6 and references 28 and 41). The sustained high viral RNA levels seen following infection with M-tropic isolates would be predicted to result from a more persistent infection in the lymph nodes of hu-PBL-SCID mice which might be due to an ongoing conversion of naive T cells to an activated phenotype. Some of these cells may migrate to the peritoneal cavity, providing a source of new CD4⁺ T cells to replace those previously depleted by infection. Alternatively, CD4⁺ T cells infected with M-tropic HIV-1 may produce progeny virions for substantially longer than T cells infected with M/T- or T-tropic virus. The additional target cells available to M- and M/T-tropic virus, in this case monocyte-derived macrophages, could also contribute to the sustained virus production following infection with the M-tropic isolates. However, M/T-tropic viruses appear to be unable to establish persistent infection in macrophages or lead to macrophage destruction in this animal model, since no plasma viremia can be detected at 2 or more weeks after infection. Alternatively, infection of a very small number of macrophages may not result in detectable plasma viral RNA.

It is striking that a single amino acid substitution that alters coreceptor usage (58) and syncytium-inducing properties (14) has such a profound effect on viral replication and CD4⁺ T-cell depletion in this animal model for HIV-1 infection. The predominance of M-tropic isolates throughout most of the course of natural infection with HIV-1 suggests that there must be selective factors promoting the maintenance of the M-tropic phenotype, since the high mutation rate of the virus (18) and the small number of amino acid changes necessary to change cell tropism would otherwise allow rapid evolution of M/T- and T-tropic variants. Our results provide one potential explanation for the predominance of M-tropic HIV-1 if viral RNA levels observed during the relatively short course of infection in the hu-PBL-SCID mice can be extrapolated to infectious virions recovered from chronically infected humans. If an individual were infected with a mixture of M-tropic and M/T-tropic viruses, our results suggest that most of the plasma viremia would be contributed by the M-tropic virus. This effect would result from the longer duration of virus production by individual CD4⁺ T cells infected with M-tropic viruses compared to CD4⁺ T cells infected with the more rapidly cytopathic dual-tropic virus. The switch to dual tropism, or acquisition of the ability to use CXCR4 as a coreceptor, would not compensate for this effect, since few naive CD4⁺ T cells would be present as new targets for infection (3, 53). It is thus possible that many M/T-tropic variants arise transiently, are unable to displace the predominant M-tropic virus population, and disappear due to destruction of their target population, without detection in the plasma virus pool. The acquisition of the SI phenotype may also serve to isolate M/T-tropic virus in localized foci of infected cells (40, 54) and limit systemic spread of the virus. It should also be noted that differential half-lives of cells infected with HIV-1 differing in cell tropism would affect current calculations of viral and cell turnover (30, 49, 50).