Infectious Cellular Load in Human Immunodeficiency Virus Type 1 (HIV-1)-Infected Individuals and Susceptibility of Peripheral Blood Mononuclear Cells from Their Exposed Partners to Non-Syncytium-Inducing HIV-1 as Major Determinants for HIV-1 Transmission in Homosexual Couples

Abstract

To study risk factors for homosexual transmission of human immunodeficiency virus type 1 (HIV-1), we compared 10 monogamous homosexual couples between whom transmission of HIV-1 had occurred with 10 monogamous homosexual couples between whom HIV-1 transmission had not occurred despite high-risk sexual behavior. In the group of individuals who did not transmit virus, peripheral cellular infectious load was lower and the CD4⁺ T-cell counts were higher than in the group of transmitters. HIV-1 RNA levels in serum did not differ between transmitters and nontransmitters. Compared with peripheral blood mononuclear cells (PBMC) from normal healthy blood donors, 8 of 10 nonrecipients and only 3 of 8 recipients had PBMC with reduced susceptibility to in vitro infection with non-syncytium-inducing (NSI) HIV-1 variants isolated from either their respective partners or an unrelated individual. No difference in susceptibility was observed for infection with a syncytium-inducing variant. Among the individuals who had PBMC with reduced susceptibility, five nonrecipients and one recipient had PBMC that were equally or even less susceptible to NSI variants than PBMC that had low susceptibility and that were derived from healthy blood donors that were heterozygous for a 32-bp deletion in the CCR5 gene (CCR5 Δ32). Three of these individuals (all nonrecipients) had a CCR5 Δ32 heterozygous genotype themselves, confirming an association between low susceptibility to NSI variants and CCR5 Δ32 heterozygosity. All three recipients with less susceptible PBMC had partners with a high infectious cellular load; inversely, both nonrecipients with normally susceptible PBMC had partners with a very low infectious cellular load. These results suggest that a combination of susceptibility of target cells and inoculum size upon homosexual exposure largely determines whether HIV-1 infection is established.

Human immunodeficiency virus type 1 (HIV-1) can be transmitted either vertically, parenterally, or sexually. The risk factors involved in these routes of transmission have been the subject of many studies. The presence of neutralizing antibodies (21, 31, 37), high CD4⁺ T-cell counts (36), and low levels of HIV-1 RNA in serum (15, 18, 26) have been associated with reduced vertical transmission rates. The reduction in transmission rates by treatment of mothers with zidovudine during pregnancy (9), thereby reducing the viral load, is in agreement with the latter observation. For heterosexual transmission, the clinical stage of the potential HIV-1 donor and the viral load were reported as important risk factors (14, 33).

In the western world, transmission of HIV-1 via homosexual contact is still a major route of infection (17). Studies of homosexual men that included both HIV-1-infected individuals and frequently exposed yet uninfected individuals revealed that sociobehavioral factors, such as the number of sexual partners and specific sexual techniques, in particular, anal receptive intercourse, are risk factors for homosexual transmission of HIV-1 (reviewed in reference 6). Furthermore, the infectivity of the initially HIV-1-infected individual, which is influenced by factors such as the clinical stage of disease and the presence of additional sexually transmitted diseases, and the susceptibility of the exposed individual, which is determined by factors such as the presence of ulcerative sexually transmitted diseases (6), anti-HIV-1 activity of CD8⁺ T cells, and susceptibility of target cells (28), are associated with homosexual HIV-1 transmission. Indeed, individuals lacking CC-chemokine receptor 5 (CCR5), the coreceptor for HIV-1, are relatively nonsusceptible to HIV-1 infection (11, 20, 24, 30).

Since in most studies performed so far the HIV-1 donors were not known, it has been impossible to study the risk for transmission by combining factors associated with both the initial HIV-1 carrier and the exposed sexual partner. In order to analyze these risk factors, we studied 20 monogamous homosexual couples, all participating in the Amsterdam Cohort Studies on AIDS (ACS). Between 10 of the couples, HIV-1 had been transmitted, which was confirmed by homology in the V3 and gag sequences detected in both partners. In the other 10 couples, no transmission had occurred despite periods of unprotected sexual contact ranging from 6 months to 13 years. In these couples, nontransmitters and transmitters were compared with respect to cellular infectious load, HIV-1 RNA load in serum, and CD4⁺ T-cell counts at the moment of (possible) transmission, and peripheral blood mononuclear cells (PBMC) from nonrecipients and recipients were analyzed for HIV-1 susceptibility.

MATERIALS AND METHODS

Subjects.

Twenty monogamous homosexual couples with a discordant HIV-1 antibody status at the moment of entry in the ACS were selected for this study. In 10 of the couples (concordant couples), the initially seronegative individuals seroconverted during follow-up (recipients) after a mean risk period of 1.3 years (Table 1). Phylogenetic analysis of the gag and V3 region sequences confirmed that HIV-1 had been transmitted between the partners of these couples (data not shown). The moment of transmission and seroconversion was estimated to be the midpoint between the last seronegative visit and the first seropositive visit, which were mostly 3 months apart. PBMC and serum from the HIV-1-infected individuals who transmitted virus to their partners (transmitters) were analyzed from samples obtained close to the moment of seroconversion of the receiving partner when possible (Table 1).

TABLE 1.

Characteristics of study subjects

Couplea	Ageb (yr) of:		Reported length of risk period (yr)c	Time of load determina- tion (mo)d	Virus phenotype at moment of (possible) transmissione	No. of virus clones analyzed
	NR or R	NT or T
D1	26	28	3	6.5	NSI	4
D2	41	43	13	5.9	NSI	5
D3	43	46	4	8.5	NSI	5
D4	24	38	3	2.9	NSI	5
D5	45	53	4	3.0	NSI	5
D6	33	33	7	12.2	NSI	4
D7	34	41	2	6.6	NSI	4
D8	33	40	NK	8.8	NSI	5
D9	40	40	9	3.0	NSI	4
D10	40	34	0.5	3.3	NSI	5
C1	40	49	1	4.3	NSI	5
C2	56	41	0.5	31.1	NSI	3
C3	30	27	1	7.3	NSI	5
C4	34	49	<0.5	4.2	NSI	5
C5	36	28	NK	−3.1	NSI	5
C6	37	34	1	21.4	NSI	5
C7	44	30	NK	6.1	SI	5f
C8	28	39	4	2.7	NSI	5
C9	48	32	NK	17.1	NSI	0
C10	31	24	2	3.5	NSI	0

^aD, discordant; C, concordant. 

^bAge of transmitters (T) and recipients (R) at moment of transmission and age of nontransmitters (NT) and nonrecipients (NR) at the end of the risk period. 

^cIn concordant couples, the period of unsafe sex until the moment of transmission. NK, not known. 

^dMonths after entry for nontransmitters and after the moment of seroconversion of the partner for transmitters. 

^eIn nontransmitters, the phenotype of the first time point at which virus clones were successfully grown; in transmitters, the phenotype close to the time point of seroconversion of the partner. 

^fTwo NSI and three SI variants were analyzed. 

Ten monogamous couples remained discordant, despite unprotected sexual contact for periods ranging from 0.5 to 13 years (Table 1). In this period, the seropositive partners did not transmit (nontransmitters) virus to the specific sexual partner under study. Seven of 10 individuals who did not get infected by their partners (nonrecipients) had had receptive anal intercourse with their seropositive partners. In all discordant couples, the period of unsafe sex (and therefore the period in which transmission could have occurred) was just prior to or partly coincided with entry in the ACS. Analyses of nontransmitter PBMC and serum were performed on the first available samples after entry (Table 1).

With the exception of one transmitter (of concordant couple C7) who had syncytium-inducing (SI) variants, all transmitters and nontransmitters only had non-syncytium-inducing (NSI) variants at the moment of analysis (Table 1).

Virus isolation and determination of cellular infectious virus load.

Cryopreserved PBMC from transmitters and nontransmitters were thawed and cocultivated under limiting diluting conditions in order to obtain biological virus clones as described previously (22, 32). Briefly, the PBMC were cocultivated with healthy donor PBMC that had been stimulated with phytohemagglutinin (PHA) for 2 to 3 days in 96-well microtiter plates. Every week, culture supernatants were tested for p24 antigen by an in-house p24 antigen capture enzyme-linked immunosorbent assay (35). At the same time, one-third of the culture volume was transferred to new 96-well plates, and fresh PHA-stimulated healthy donor PBMC were added to propagate the culture. From the positive wells, virus stocks were grown and stored at −70°C until use. The syncytium-inducing phenotype of the biological virus clones was analyzed by cocultivation with MT2 cells (23).

The proportion of productively infected CD4⁺ T cells was calculated with the formula for Poisson distribution: F = −ln(F₀), in which F₀ is the fraction of negative cultures. In some cases, no (or only a few) virus clones could be grown from the PBMC sample closest to transmission, and (additional) biological virus clones were therefore obtained from PBMC samples 10 to 20 months later.

Analysis of CD4⁺ T-cell counts.

T-lymphocyte immunophenotyping for the CD4⁺ T cells was carried out at 3-month intervals by flow cytofluorometry. PBMC were stained with CD4 monoclonal antibodies according to standard procedures for fluorescence-activated cell sorting analysis. For nontransmitters, the mean CD4⁺ T-cell counts for the first three visits were calculated. For transmitters, the mean CD4⁺ T-cell counts for three visits including and close to the seroconversion date of the partners were calculated. In two cases, the transmitters entered the ACS after seroconversion of their partners (1.8 [C6] and 2.5 [C2] years); therefore, no samples were available close to the seroconversion date. In these cases, the mean CD4⁺ T-cell counts for the first three visits after entry were calculated.

Quantitation of RNA load in serum.

RNA levels were analyzed in cryopreserved serum samples derived from the same (or at most 2 months apart) visit as that from which the PBMC samples were obtained by use of a nucleic acid-based amplification assay (HIV-1 RNA QT; Organon Teknika, Boxtel, The Netherlands [38]).

Susceptibility of recipient and nonrecipient PBMC.

HIV-1 stocks were titrated on PBMC of different origins, and the differences between titers were used as a measure of in vitro susceptibility to HIV-1 infection. Relative susceptibilities to three to five biological virus clones isolated from the respective partners as well as to unrelated NSI and SI biological clones were studied for PHA-stimulated peripheral blood lymphocytes from the corresponding recipients and nonrecipients and five healthy blood donors (bd1 to bd5). The control NSI virus variant was a transmitted variant (based on the V3 sequence) derived from one of the transmitters, and the SI virus variant was derived from an ACS participant who was otherwise not involved in this study. Due to the limited availability of PBMC from the nonrecipients, neither control variant was titrated on PBMC from the nonrecipient of couple D6, and the SI control variant was not titrated on PBMC from the nonrecipient of couple D1.

Two types of blood donors could be distinguished: those having PBMC with normal susceptibilities (bd2, bd4, and bd5) and those having PBMC with low susceptibilities (bd1 and bd3) to HIV-1 infection. For each couple, the titers of the transmitter and nontransmitter viruses (expressed as 50% tissue culture infective doses [TCID₅₀] per milliliter of supernatant) established on both types of blood donor PBMC and on the specific recipient and nonrecipient PBMC were compared and analyzed with the Wilcoxon signed rank test. Thus, the susceptibilities of recipient and nonrecipient PBMC were classified relative to the susceptibilities of both types of blood donor cells (Table 2).

TABLE 2.

Classification of relative susceptibilities of recipient and nonrecipient PBMC to partner HIV-1

Susceptibility relative to PBMC of healthy blood donorsa		Relative susceptibilities of recipient and non- recipient PBMC
bd2, bd4, and bd5	bd1 and bd3
Higher	Higher	+++
Equal	Higher	++
Equalb	Equal	++
Lower	Higher	+
Lowerb	Lower	+
Lower	Equal	+/−
Lower	Lower	−

^aAs analyzed with the Wilcoxon signed rank test. 

^bNo differences in virus titers on the two types of blood donor PBMC were observed. 

Cryopreserved PBMC obtained from the recipients prior to seroconversion (minimally 6 months prior to the first HIV-1-seropositive sample) were used for analysis. Two of the concordant couples (C9 and C10) were excluded from analysis because preseroconversion PBMC were not available.

CCR5 genotyping.

Genomic DNA was isolated from cryopreserved PBMC (Qiagen [Hilden, Germany] blood kit) and analyzed by PCR with primers flanking the 32-bp deletion in CCR5 (CCR5 Δ32) (13).

For sequence analysis, CCR5 DNA was amplified with the full-length primers CCR5FLS (5′-GGTGGAACAAGATGGATTAT-3′; positions 229 to 248; sense) and CCR5FLAS (5′-AGAGTTGTGCACATGGCT-3′; positions 1324 to 1343; antisense) (12). For amplification, DNA was denatured for 5 min at 94°C, followed by 30 cycles of 1 min at 94°C, 1 min at 50°C, and 1 min at 72°C. PCR products were purified with a Qiaquick Spin PCR purification kit (Qiagen). The positive strand was sequenced with CCR5FLS, CCR5Δ32S (5′-GATAGGTACCTGGCTGTCGTCCAT-3′; positions 612 to 635), and CCR5A (5′-GCAGTAGCTCTAACAGGTTGGACC-3′; positions 1045 to 1068). The negative strand was sequenced with CCR5FLAS and CCR5Δ32AS (5′-AGATAGTCATCTTGGGGCTGGT-3′; positions 829 to 850). Radiolabeled terminator cycle sequencing with Thermo Sequenase DNA polymerase (Amersham) was performed with primers CCR5FLS, CCR5Δ32S, and CCR5FLAS. Dye terminator cycle sequencing with Amplitaq DNA polymerase (Perkin-Elmer) was performed with primers CCR5Δ32AS and CCR5A.

Both DNA purification and sequencing procedures were performed according to the instructions of the manufacturers.

Statistical analysis.

Differences in cellular infectious load, RNA load in serum, CD4⁺ T-cell counts, and susceptibilities to the control NSI and SI variants were analyzed with the Mann-Whitney U test. The unpaired Student t test was used to analyze the differences in susceptibilities to all NSI variants between healthy blood donors. For each couple, the differences between virus titers on PBMC from the two types of healthy blood donors and PBMC from the recipients and nonrecipients were analyzed with the Wilcoxon signed rank test.

RESULTS

Transmitter and nontransmitter variables: viral load and CD4⁺ T-cell counts.

The individuals who did or did not transmit HIV-1 to their respective partners were compared with respect to markers of disease progression. Close to the moment of transmission, the median infectious cellular load in peripheral CD4⁺ T cells was significantly higher in the individuals who did transmit HIV-1 than it was at the end of the high-risk period in the nontransmitters (119 versus 8 TCID/10⁶ CD4⁺ T cells; P = 0.005) (Fig. 1a and Table 3). Nevertheless, three individuals who transmitted virus to their partners had a low cellular infectious load at the moment of transmission (10, 11, and 16 TCID/10⁶ CD4⁺ T cells). The median HIV-1 RNA levels in serum, analyzed at the same moment as the infectious cellular load, showed no difference between the transmitters and the nontransmitters, i.e., 45,500 (=4.6 log units) versus 28,500 (=4.5 log units) RNA copies/ml of serum (P = 0.6) (Fig. 1b and Table 3).

FIG. 1.

Peripheral viral load and CD4⁺ T-cell counts in nontransmitters and transmitters. (a) Cellular infectious load, expressed as log TCID per 10⁶ CD4⁺ T cells. (b) RNA load in serum, expressed as log RNA copies per milliliter of serum. (c) Mean CD4⁺ T-cell counts per microliter of blood. For nontransmitters, PBMC and serum samples obtained as close as possible to the moment of entry into the ACS were analyzed. For transmitters, PBMC and serum samples obtained as close as possible to the date of seroconversion of the recipients were analyzed. PBMC and serum samples were mostly derived from the same visit (in 16 cases) or, maximally, 2 months apart. The mean CD4⁺ T-cell counts were calculated from the CD4⁺ T-cell counts on the first three visits after the moment of entry for the nontransmitters and from the CD4⁺ T-cell counts on the three visits closest to the date of seroconversion of the recipients for the transmitters. Bars indicate median values. ns, not significant.

TABLE 3.

Determinants for homosexual HIV-1 transmission

Couplea	Infectious cellular load (TCID/ 106 CD4+ T cells)	RNA load (log copies/ml of serum)	Mean CD4+ cell counts/μl	Relative susceptibility	CCR5 genotypec for:
					NR or R	NT or T
D1	4	3.40	520	++	wt	hz
D2	3	4.52	523	++	wt	wt
D3	3	3.93	443	+	wt	wt
D4	44	4.80	863	+	wt	wt
D5	4	4.38	330	+	wt	hzd
D6	21	4.94	807	+/−	wt	wt
D7	12	4.72	660	+/−	hz	wt
D8	66	4.72	613	−	hz	wt
D9	18	4.20	457	−	wt	wt
D10	4	3.79	630	−	hz	wt
C1	10	3.00	390	+++	wt	wt
C2	115	4.64	560	+++	wt	wt
C3	11	3.77	363	++	wt	wt
C4	16	3.18	367	++	wt	hzd
C5	122	4.46	480	++	wt	wtd
C6	382	4.81	360	+	wt	wt
C7	1,868	5.32	120	+	wt	hzd
C8	213	4.67	467	+/−	wt	wt
C9	79	5.23	503	NTb	wt	hze
C10	395	5.60	520	NT	wt	wt

^aD, discordant; C, concordant. 

^bNT, not tested. 

^cwt, wild type; hz, heterozygous. 

^dIndividuals from whom viruses were isolated that replicated to normal titers on PBMC with reduced susceptibilities from blood donors bd1 and bd3. 

^eViruses were not tested for replication on PBMC of different origins. 

In accordance with the inverse correlation between cellular infectious load and CD4⁺ T-cell counts (10, 22), the median CD4⁺ T-cell counts were lower in the individuals who had transmitted HIV-1 to their partners (429 versus 568 CD4⁺ counts/μl of blood; P = 0.03) (Fig. 1c and Table 3).

For two of the transmitters, the first samples available were already 21.4 and 31.1 months after the seroconversion dates of their partners (couples C6 and C2, respectively) (Table 1). Both individuals had a high load at these times; however, exclusion of these two individuals from viral load and CD4⁺ T-cell count analyses did not change the statistical significance.

Recipient and nonrecipient variables: susceptibility of PBMC to HIV-1 variants of the partners.

In a panel of five healthy blood donors, two types were distinguished with respect to susceptibility to in vitro HIV-1 infection. The mean titer of all 82 NSI variants analyzed in this study was significantly lower on PBMC of healthy blood donors bd1 and bd3 than on PBMC from the other three blood donors, bd2, bd4, and bd5 (P < 0.001). Based on this information, PBMC from bd1 and bd3 were classified as less susceptible (+/−) and PBMC from bd2, bd4, and bd5 were classified as normally susceptible (++).

The susceptibility of both types of blood donor PBMC to the virus variants of each transmitter and nontransmitter was analyzed separately. The HIV-1 susceptibility of the nonrecipient and recipient PBMC was classified relative to the susceptibility of both types of blood donor PBMC to the same variants (Table 2). As a result, PBMC from recipients and nonrecipients were classified as less susceptible (+, +/−, or −) or normally to highly susceptible (++ or +++) to infection with HIV-1 variants derived from the respective partners. Low susceptibility was defined as susceptibility equal to or lower than that of PBMC from bd1 and bd3. The majority of HIV-1 variants derived from one of the nontransmitters (D5) and three of the transmitters (C4, C5, and C7) replicated equally well on both types of blood donor PBMC; in these cases, the recipient and nonrecipient PBMC were classified as less or normally susceptible. Representative patterns are shown in Fig. 2.

FIG. 2.

Analysis of relative susceptibilities of recipient and nonrecipient PBMC to corresponding transmitter and nontransmitter HIV-1 variants. Virus variants isolated from transmitters and nontransmitters were titrated (expressed as TCID₅₀ per milliliter of supernatant) on PHA-stimulated PBMC from the respective recipients and nonrecipients and on PHA-stimulated PBMC from five healthy blood donors. Within each panel, the different symbols represent different virus variants derived from the specific transmitter or nontransmitter. The virus titers on PBMC from blood donors 1 and 3 (bd 1 + 3) and blood donors 2, 4, and 5 (bd 2,4+5) are shown as separate groups and are compared with the virus titers on PBMC from the respective nonrecipients (nr) and recipients (r) of couple D10 (a), couple C1 (b), couple D3 (c), and couple C4 (d). −, +, ++, and +++ indicate the relative susceptibilities of the recipient and nonrecipient PBMC.

Eight of 10 (80%) nonrecipients and 3 of 8 (38%) recipients had PBMC with susceptibility lower than that of normally susceptible healthy blood donor PBMC (Fig. 3a and Table 3). Furthermore, the least susceptible PBMC were only found in nonrecipients, while the most susceptible PBMC were only found in recipients. In general, PBMC of nonrecipients were less susceptible to HIV-1 variants isolated from their sexual partners than were PBMC of recipients.

FIG. 3.

Relative susceptibilities of PBMC from nonrecipients and recipients to partner and unrelated NSI and SI HIV-1 variants. (a) Relative susceptibilities of PBMC from each nonrecipient and recipient to HIV-1 variants isolated from their partners are plotted. Different symbols indicate individuals having PBMC with low susceptibility (white), intermediate susceptibility (gray), and normal to high susceptibility (black). (b and c) PBMC from recipients and nonrecipients were stratified by their susceptibilities to partner HIV-1 variants and analyzed for their susceptibilities (expressed as log TCID₅₀ per milliliter of virus stock) to an unrelated NSI variant (b) and an unrelated SI variant (c). Bars indicate median values. ns, not significant.

In addition to the analysis of susceptibility to HIV-1 variants isolated from the corresponding sexual partners, recipient and nonrecipient PBMC were tested for susceptibility to unrelated primary NSI and SI variants. PBMC derived from individuals who had low susceptibility to their partners’ viruses also had low susceptibility to the control NSI variant (Fig. 3b). In contrast, all PBMC were equally susceptible to infection with the SI variant (Fig. 3c). Similarly, the healthy blood donor cells which were less susceptible to NSI variants were normally susceptible to this SI variant (data not shown).

Susceptibility and CCR5 genotype.

Since NSI and SI variants may use different coreceptors (12, 16, 19) and the observed reduced susceptibility was specific for NSI variants, we analyzed the CCR5 genotype in recipient, nonrecipient, and healthy blood donor PBMC. Both healthy blood donors who had PBMC with low susceptibility (bd1 and bd3) were heterozygous for the 32-bp deletion in the CCR5 gene. Blood donors bd2, bd4, and bd5, with normally susceptible cells, had a wild-type CCR5 genotype. Three of 10 nonrecipients had a CCR5 Δ32 heterozygous genotype (Table 3). All three of these individuals had PBMC with low susceptibility (+/− or −). The other three individuals who had PBMC with low susceptibility (two nonrecipients and one recipient) had a wild-type CCR5 genotype, which was confirmed by sequence analysis of the coding region of the CCR5 gene (data not shown). In addition, the remaining recipients and nonrecipients with normally to highly susceptible cells all displayed the wild-type CCR5 genotype.

Three transmitters and two nontransmitters also had a CCR5 Δ32 heterozygous genotype (Table 3). Interestingly, the majority of biological HIV-1 clones from three of these individuals replicated to normal titers on the (in general) less susceptible PBMC from either or both of the CCR5 Δ32 heterozygous blood donors (Fig. 2d).

DISCUSSION

Biological factors that may influence the risk for homosexual transmission of HIV-1 were studied in 10 discordant and 10 concordant homosexual couples. In general, we observed a higher peripheral cellular infectious load and a lower CD4⁺ T-cell count in the individuals who had transmitted HIV-1 to their partners. This result is in agreement with previous reports on the association between stage of disease and the risk for transmission (reviewed in reference 6). Surprisingly, we found no correlation between HIV-1 RNA load in serum and transmission. This finding might be explained by the absence of a correlation between viral RNA load in serum and that in semen (25), the latter being a more relevant compartment for sexual transmission. Alternatively, transmission might predominantly occur via cell-associated HIV-1. In agreement with the latter idea, an association has been found between cell-associated HIV-1 variants in the semen of HIV-1-infected individuals and the virus variants present in the sexual partners to whom they transmitted the virus (42).

Despite their different associations with transmission, both measures of peripheral viral load correlated with each other within the group of transmitters and the group of nontransmitters (data not shown), in agreement with our previous observations (5).

Eight of 10 nonrecipients had PBMC with reduced susceptibility to infection with virus variants isolated from their partners and with an unrelated NSI HIV-1 variant compared with healthy blood donor PBMC. In contrast, only three of eight recipients had less susceptible cells. Moreover, the most susceptible cells were found only in recipients, whereas the least susceptible cells were found only in nonrecipients. These results are suggestive of a role of susceptibility to HIV-1 in homosexual transmission. In accordance, CD4⁺ T cells of multiply exposed yet uninfected individuals have been shown to be generally less susceptible than CD4⁺ T cells of nonexposed controls (28). Also, in vertical transmission, the susceptibility of the child’s target cells to virus variants from the mother has been shown to play a role (27).

Three nonrecipients had PBMC with normal susceptibility to HIV-1 infection. However, their respective HIV-1-infected partners had a very low cellular infectious load (3 to 4 TCID₅₀/10⁶ CD4⁺ T cells), at least at the end of the period in which transmission could have occurred. Inversely, the three recipients with less susceptible cells all had HIV-1-infected partners with an extremely high cellular load close to the moment of transmission, being 0.7 to 2.8 log units higher than the load in the partners of nonrecipients with similarly susceptible cells. Furthermore, the recipient with the least susceptible cells seroconverted only after 4 years of high-risk sexual behavior (C8; Tables 1 and 3). These data suggest that the susceptibility of target cells in combination with the size of the virus inoculum determines whether infection is established upon exposure via homosexual contact. From this suggestion it would follow that factors influencing either viral load (e.g., antiviral treatment) or susceptibility (e.g., CCR5 genotype) also correlate with HIV-1 transmission.

It was previously shown that CCR5 Δ32 homozygous cells are not susceptible to NSI HIV-1 infection in vitro and in vivo (24, 30). This fact and the fact that CCR5 is the coreceptor for macrophage-tropic HIV-1 variants (12) are in good agreement with the observations that macrophage-tropic variants establish infection upon entry in a new individual (39, 41). Nevertheless, the existence of HIV-1 variants capable of using other receptors might explain the fact that homozygous individuals are not completely protected from infection (4). Heterozygosity for the deletion in the CCR5 gene was shown not to protect from HIV-1 infection (11, 13, 20), although the opposite has also been suggested (30). The absence of protection seems in contrast to the observed lower susceptibility to in vitro infection of CCR5 Δ32 heterozygous cells and the association between transmission and susceptibility. The fact that low susceptibility is associated with protection from infection while CCR5 Δ32 heterozygosity is not might be explained by the fact that low susceptibility has additional causes besides CCR5 dysfunction. Moreover, we argue here that the inoculum size upon sexual exposure in combination with the susceptibility of target cells largely determines whether infection is established. Promiscuous sexual behavior increases the chance of exposure to a high viral inoculum, which might override the relative lack of susceptibility. As a consequence, promiscuous individuals might need cells with lower susceptibility than those of monogamous partners of HIV-1-infected individuals in order to remain uninfected, which would explain the increased frequency of CCR5 Δ32 homozygotes but not CCR5 Δ32 heterozygotes among frequently exposed uninfected individuals (28).

Whether the relative lack of susceptibility of PBMC from CCR5 Δ32 heterozygotes is due to reduced expression of CCR5, to higher expression of β-chemokines, or to a combination of both is currently under investigation.

Compared to the susceptibility of PBMC from CCR5 Δ32 heterozygotes, PBMC from three individuals (two nonrecipients and one recipient) had similar low susceptibility to NSI HIV-1 and normal susceptibility to SI HIV-1 yet had a wild-type CCR5 genotype. The fact that the NSI variants showed impaired replication on the PBMC from CCR5 Δ32 heterozygous blood donors indicated their dependence on CCR5 for efficient replication, excluding the dysfunctioning of other coreceptors as a likely explanation for the reduced susceptibility of the PBMC from individuals with a wild-type CCR5 genotype. Whether there are sequence-independent conformational differences for CCR5 that determine differences in ligand affinity and thus HIV-1 susceptibility will be the subject of future research.

From three transmitters and from one nontransmitter, virus variants that replicated well on the CCR5 Δ32 heterozygous blood donor PBMC were isolated. This finding might be explained by a capacity of these viruses to use other chemokine receptors as cofactors for entry or, alternatively, by an enhanced affinity of these viruses for CCR5. Interestingly, three of these individuals had a CCR5 Δ32 heterozygous genotype themselves, which may have been the driving force behind evolution toward the use of other coreceptors or toward a higher affinity for CCR5. Whether this is indeed the case is currently the subject of our studies.

This study does not exclude a role for immunity in homosexual transmission. The presence of neutralizing antibodies to autologous virus (21, 31) is associated with a lower rate of transmission of HIV-1 from mother to child. Similarly, neutralizing antibodies in semen (3, 40) might reduce the infectious titer of free virus and thereby influence transmission. Some studies have suggested a role for cellular immunity (8) in homosexual transmission. Also, rare HLA subtypes have been associated with protection from infection in frequently exposed Nairobian prostitutes (29). The basis for this protection might be alloreactivity against HLA proteins present in virion envelopes (1). In support of this idea, macaques which were vaccinated with uninfected human cells or HLA class I or class II molecules were subsequently protected from infection with cell-free simian immunodeficiency virus grown in human cells (2, 7, 34). The data presented in the present study, however, suggest that target cell susceptibility and inoculum size are major determinants of homosexual HIV-1 transmission.