Lower Human Immunodeficiency Virus (HIV) Type 2 Viral Load Reflects the Difference in Pathogenicity of HIV-1 and HIV-2

Abstract

Human immunodeficiency virus type 2 (HIV-2) is less pathogenic than HIV type 1 (HIV-1), but the mechanisms underlying this difference have not been defined. We developed an internally controlled quantitative reverse transcriptase-polymerase chain reaction to measure HIV-2 viral load and determined levels of plasma virus in a cohort of registered commercial sex workers in Dakar, Senegal. The assay has a lower limit of detection of 100 copies/mL and is linear over 4 logs. HIV-2 viral RNA was detectable in 56% of all samples tested; the median load was 141 copies/mL. Levels of viral RNA in the plasma were inversely related to CD4⁺ cell counts. HIV-2 and HIV-1 viral loads were compared among the seroincident women in the cohort; the median viral load was 30X lower in the HIV-2-infected women (P< .001, Wilcoxon rank sum test), irrespective of the length of time infected. This suggests that plasma viremia is linked to the differences in the pathogenicity of the 2 viruses.

Human immunodeficiency virus type 1 (HIV-1) infection is characterized by high rates of virus production and clearance of both infected cells and cell-free virions. The net result of these factors is reflected in the RNA viral load in the plasma of each infected individual [1, 2]. Variations in HIV-1 viral load are associated with differences in clinical status [3, 4], and multiple studies have now shown that the level of HIV-1 plasma viremia early in infection is highly predictive of future clinical course [5–7]. Mellors et al. found that, in an asymptomatic study population, each 3-fold increase in viral load measurement was associated with a 50% decrease in survival over a 10-year period [8]. Treatment with drug combinations that greatly improve clinical status have been shown to drastically lower plasma viremia, and O'Brien et al. calculated that much of the benefit resulting from treatment is due to the reduction in viral load [9]. Conversely, individuals infected with HIV-1 for at least 7 years while maintaining CD4⁺ cell counts >500/μL, defined as long-term nonprogressors, have low viral loads, usually <10⁴ RNA copies/mL [10–13].

Viral replication is also an important determinant of pathogenesis in other lentiviral infections. In macaques inoculated with SIVsm, the level of plasma viremia in the postacute phase of infection was correlated with the rate of AIDS-associated disease [14]. In another study of vaccinated macaques, levels of viremia in primary infection, modified by the ability of the immune system to control replication, determined the outcome of infection with SIVsmE660 [15]. A clone of SIVagm that caused disease in pig-tailed macaques but not in rhesus macaques or African green monkeys replicated to a higher viral load in the former, again supporting the notion that higher viral replication is linked to pathogenesis [16]. Similarly, higher plasma viral loads in feline immunodeficiency virus-infected cats are associated with more rapid disease progression [17]. In contrast, naturally infected African green monkeys and sooty mangabeys support levels of viral replication that would be considered pathogenic in human HIV-1 infection while remaining free of disease [18–20].

HIV-1 and HIV type 2 (HIV-2) are highly related human lentiviruses, yet HIV-2 is less pathogenic than HIV-1. Heterosexual transmission occurs less for HIV-2 than for HIV-1, and vertical transmission is rare [21–23]. Rates of disease development in HIV-2 infection are much lower than for HIV-1, and >95% of infected individuals followed for at least 8 years fit a clinical definition of long-term nonprogession [24, 25]. Given the primacy of viral RNA load in the pathogenesis of HIV-1 and other lentiviral infections, the epidemiology and biology of HIV-2 infection suggest that plasma viremia in HIV-2 infection may be significantly lower than in HIV-1 infection. The frequency of successful virus isolation from HIV-2-infected persons is also lower than for HIV-1 [26], thus supporting this hypothesis. Quantitative assessments of HIV-2 proviral load have demonstrated an inverse correlation with CD4⁺ cell counts and clinical status similar to that found in HIV-1, suggesting that viral replication in HIV-2 bears a similar relationship to disease. Interestingly, the levels of provirus are similar to those seen in HIV-1-infected individuals [27, 28].

We have developed a quantitative reverse transcriptase-polymerase chain reaction (RT-PCR) assay to evaluate viral load in HIV-2 infection. By using this assay, we have measured the levels of plasma viremia in samples obtained from a cohort of HIV-2-infected women. We have also compared the levels of viral RNA in HIV-1- and HIV-2-infected subjects drawn from the same population and evaluated the maintenance of levels of viral RNA with time since infection.

Methods

Study population and sample collection

Samples were obtained from a cohort of registered female sex workers in Dakar, Senegal. The epidemiologic and clinical aspects of HIV-1 and HIV-2 infection in the cohort have been fully described elsewhere [21, 29]. All women had given informed consent prior to enrollment, and none had received antiretroviral therapy. Serostatus was determined by immunoblot on whole virus lysates. The time of infection for women who converted to HIV-seropositive status while in the study was estimated as the midpoint between their last seronegative and first seropositive bleed. Plasma samples for detection of viral RNA were collected beginning in 1996; blood was collected in EDTA-containing Vacutainer tubes, and the plasma stored at −70°C within 6 h of collection, following separation by use of Ficoll-Hypaque (Organon Teknika Cappel, Durham, NC).

HIV-2 sample preparation

Twenty-five microliters of a 0.025% solution of polystyrene beads (Bangs Laboratories, Fishers, IN) was added to 500μL of plasma, and then spun at 21000 g for 1 h at 4°C. The supernatant was removed, and the pellet was resuspended in a lysis buffer containing 68% guanidinium isothiocyanate, 100 mMB-Mercaptoethanol, and 20 mM Tris, pH 7.5. Samples were incubated for 10 min at room temperature, and 2500 copies of the internal control RNA were added. The RNA was precipitated in isopropanol by use of 25 μg of glycogen as a carrier. After washing with ethanol, RNA was resuspended in 50 μL water, and stored at −70°C.

Preparation of HIV-2 assay standards

DNA templates were derived by amplifying a region of the gag gene of the HIV-2 plasmid pGH123 [30] with the primers OG53 and OG106 [31]. The resulting PCR product was cloned into pCRII (Invitrogen, Carlsbad, CA), forming a standard (STD) template. To create an internal control (IC), complementary internal primers carrying extra sequence at the 5′ end were used to amplify the same sequence in 2 parts. The products were ligated after restriction digestion and cloned into pCRII, creating a gag insert 25 bp larger than that of the STD [28]. The inserts were sequenced to confirm the size and identity of the templates. The plasmids were linearized by use of Bam H1 and purified by ethanol precipitation.

RNA transcripts were prepared from linearized templates by use of the Ambion Megascript kit (Ambion, Austin, TX). The transcripts were purified by 2 passages over spin columns, followed by ethanol precipitation, and resuspended in 10 mM Tris, 1mM EDTA. The transcripts were quantified by spectrophotometry at OD260 by use of an extinction coefficient of 40 μg/mL/OD and specific molecular weight values for each base [32]. The OD260/280 ratio of each transcript was at least 1.9, and a parallel transcript incorporating ³²P-dUTP was examined by PEI Chromatography (Schleicher & Schuell, Dassel, Germany) to ensure that the transcripts were full length. Calculations of yield from percentage of incorporation of the ³²P-dUTP were consistent with those from optical density. The RNA was also separated on an acrylamide gel, and the ethidium bromide-stained bands were examined by densitometry. Transcripts were diluted to 10⁶ copies/μL in TE, and aliquots were stored at −80°C. Subsequent dilutions were made in a solution containing 10 μg/mL yeast tRNA (Sigma, St. Louis).

The HIV-2_AL strain was used for the experiments that assessed reproducibility of the assay. This strain was originally isolated from an individual in Dakar, Senegal, and sequencing showed the virus to be HIV-2 subtype A (S. Popper, unpublished data). Culture supernatant was collected after passage through a 0.45μ filter and diluted in plasma obtained from HIV-negative blood donors.

Both the in vitro transcripts and purified nucleic acid from the cell-culture supernatant were amplified by use of Taq polymerase (Perkm-Elmer, Brandenburg, NJ) and the cycling conditions described for the RT-PCR, without a reverse-transcription step, to assure that reaction products were not a result of DNA present in the purified nucleic acid.

RT-PCR and detection

A one-step RT-PCR kit (rTth EZ kit, Perkin-Elmer) was used to detect HIV-2 viral RNA. Primers OG63 and OG81a [31] were included at 20 pmol/50 μL reaction, along with 1.5 mMMn. OG63 was labeled with 6-FAM at the 5′ end (Bethesda Research Laboratories, Gaithersburg, MD). Theseprimers have been used extensively for amplification of HIV-2 proviral DNA [28, 33]. For reverse transcription, 20 μL of purified RNA was heated at 94°C for 1 min, followed by 55°C for 5 min, and 60°C for 25 min, ending with 94°C for 3 min. The PCR then proceeded with 32 cycles at 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s. After 5 min at 70°C, the temperature was brought to 4°C. Dilutions of 1 : 5 and 1 : 25 were prepared in 5 mM EDTA, 10 mg/mL dextran blue, 80% formamide, and 1.5 μL was processed on an ABI 373XL Automated Sequencer (ABI, Foster City, CA). The intensity of the fluorescence from each of the 2 products (sample and IC) was recorded by use of Gene-Scan software (ABI, Foster City, CA). The sample copy number was calculated as the ratio of fluorescence of the 2 products, multiplied by the number of copies of IC per RT-PCR reaction (1000), and adjusted for the volume of sample processed (200 μL).

HIV-1 viral load was determined by use of the Quantiplex HIV RNA 2.0 bDNA assay (Chiron, Emeryville, CA), according to the manufacturer's instructions.

Results

Quantitative HIV-2 viral load assay

Each sample was purified and amplified with an internal control RNA that contained the same primer binding sites as the samples, but was 25nt longer, enabling us to distinguish the IC and sample amplicons by size. The linear range of the detection system was measured by use of serial dilutions of the amplified IC (figure 1A), and samples with a saturated signal were further diluted and rerun.

To evaluate the assay and determine the linear range, we also generated a standard RNA from a gag sequence lacking the insert. Both the standard and IC RNA were carefully quantified, and serial dilutions of the standard were reverse-transcribed and amplified in the presence of 1000 copies of the IC. The linear range of the assay was at least 4 logs, from 10 to 100,000 copies of RNA. The input and calculated copies of standard were very similar with a slope for the standard curve close to 1.0 (figure 1B). The sensitivity of the assay, on the basis of repeated testing of single concentrations of the standard, was 100% at 20 copies and 60% at 10 copies. Two hundred microliters of plasma was tested in each assay, and thus the level of sensitivity of the assay was conservatively set at 100 RNA copies/mL, though samples with copy numbers below this level may be detected.

Virions were concentrated by use of high-speed centrifugation of the plasma prior to extraction. Because this step of the procedure was not internally controlled, we measured the efficiency by comparing the yield of viral RNA from the plasma supernatant before and after centrifugation, by using a QIA-GEN viral RNA kit (Qiagen, Valencia, CA) and measuring the yield using the RT-PCR. Centrifugation at either 17000 g or 23000 g pelleted 85%–90% of the virions. Recovery of the RNA from the lysis solution was equally efficient (data not shown). To assess the reproducibility of the entire procedure, a stock of HIV-2 virus was diluted in HIV-negative plasma and tested in quadruplicate in multiple experiments. At a high concentration of virus, with a mean assay value of 152,711 copies/mL, the intra-assay coefficient of variation was 6%, and the interassay variation was 7%. A more dilute viral stock, with a mean value of 4529 copies/mL, had an intra-assay variation of 10% and an interassay variation of 37%. Reverse transcription and amplification of extracted samples in duplicate resulted in a <2% difference in the coefficient of variation, demonstrating that the high degree of variability occurring in those reactions was well controlled by the inclusion of the IC RNA.

HIV-1 and HIV-2 viral load in the study cohort

We tested the levels of HIV-2 plasma viremia in samples collected from 68 women in the Dakar cohort. RNA was detected in 38 (56%) of 68 of the samples; this included 3 samples for which the calculated viral load was <100 copies/mL. Plasma viral load was in the range of 50 copies/mL to 39,861 copies/mL, with a median of 141 copies/mL. There was a trend toward decreasing viral load across categories of CD4⁺ cell counts; those with >800 CD4⁺ cells had significantly lower levels of plasma viremia than individuals with <400 or 400–800 CD4⁺ cells (P < .05, Wilcoxon rank sum test) (figure 2). The proportion of samples with undetectable levels of RNA increased as CD4⁺ cell counts increased, from 2 (25%) of 8, to 8 (36%) of 22, then to 8 (62%) of 13, though this was not significant. Overall, there was an inverse correlation between HIV-2 viral load and CD4⁺ cell counts (p = −0.36, P< .05, Spearman rank correlation). Levels in seroprevalent and seroincident HIV-2-infected women were similar, and both were significantly lower than the levels in HIV-1-infected women from the same cohort (P< .05, Wilcoxon rank sum test) (figure 3).

Previous analysis has demonstrated a dramatic difference between HIV-1 and HIV-2 disease progression in a cohort of women with known time of infection [24]. We have now evaluated plasma RNA levels in samples from this seroincident subset of the cohort; the characteristics of those sampled are presented in table 1. The length of time infected and age at infection was similar for the 2 groups, but there were significant differences in the quantity of CD4⁺ cells, consistent with differences in the pathogenicity of the 2 viruses. The median viral load was 30 × lower in the HIV-2-infected women. Viral RNA was detectable in only 19 of 30 HIV-2-infected women, and 27 (90%) of 30 women had a viral load <10,000 copies/mL. Conversely, 23 of 26 HIV-1-infected women had levels of viral RNA above the cutoff of 500 copies/mL specified by the bDNA assay, and 11 (42%) of 26 had a viral load >10,000 copies/mL. The distribution of HIV-1 subtypes among the individuals tested was not different from what was recently described for all HIV-1 seroincident women in the cohort (K. Travers, manuscript in preparation) [34]. The samples collected represented a nearly 12-year period postseroconversion. HIV-2 viral load was significantly lower than HIV-1 viral load in years 1–4 and 4–8, and remained low in years 9–12, when there was only 1 HIV-1 sample (figure 4).

Discussion

Levels of viral RNA in plasma have proved to be an important determinant of disease in a number of lentiviral infections. Despite this, a commercial assay for the measurement of HIV-2 plasma viremia is not available. We have developed a nonradioactive quantitative assay on the basis of RT-PCR that uses an internal standard to control for variation in the purification, reverse transcription, amplification, and detection of the viral genome. The reproducibility and sensitivity are equal to that of commercially available HIV-1 assays [35]. We have applied the assay to samples obtained from women infected with HIV-2 and have found viral load to be much lower than that found in HIV-1 infection. Levels of HIV-2 plasma viral RNA are strikingly low; more than one third of the women sampled had a level of virus <100 copies/mL. It is unlikely that these samples or the low viral load overall result from problems in sample collection, storage, or transport, because we followed standard guidelines for such samples and aliquots of samples obtained on the same day, but stored in Boston versus Dakar, Senegal, yielded equivalent results when tested in parallel (data not shown). Furthermore, HIV-1 samples from the same cohort were collected and stored in the same manner, and viral RNA was readily detectable in most of those samples.

HIV-2 viral load was consistently lower than that of HIV-1, regardless of the time since seroconversion. We did not observe any association between length of time infected and either HIV-1 or HIV-2 viral load (P> .05, Spearman rank correlation). This may be associated with the clinical status of this group of individuals; only 3 of the HIV-1-infected women, and none of the HIV-2-infected women, had CD4⁺ cell counts <200. Longitudinal studies of HIV-1 infection have shown that, in the absence of AIDS, viral load may not increase for years after infection [36, 37]. The apparent stability of HIV-2 viral load, even many years after infection, is also consistent with the lack of disease progression that has been found in prospective studies [24] and is similar to that found in HIV-1 long-term nonprogressors [12]. It is unlikely that the relative lack of disease was due to differential sampling of healthy individuals; there was no difference in the clinical status of the women from whom we obtained RNA samples, compared with others in the cohort with the same serostatus and in the same time stratum (P > .05, x²). The larger number of HIV-2-infected women observed at a later time may be due to reduced pathogenicity of HIV-2 and may also reflect the dynamics of the HIV epidemic in this population, where HIV-1 was more recently introduced [21]. Longitudinal studies will be necessary to definitively describe the temporal patterns of HIV-2 viral load.

The observed differences in viral load between HIV-1 and HIV-2 are consistent with differences in the pathogenicity of the 2 viruses. The differences roughly correspond to the HIV-1 viral loads in 2 risk categories identified by Mellors et al. [38]. In that study, individuals with 10,000–30,000 copies/mL were 6× more likely to develop AIDS when compared with those with ⩽500 copies/mL. That difference in risk of disease progression is similar to the differences in progression to AIDS for HIV-1 and HIV-2 found in this cohort [24] and suggests that the absolute levels of viral replication may be an important determinant of pathogenicity. We also observed increases in HIV-2 viral load at lower CD4⁺ cell counts, indicating that viral replication is related to HIV-2 pathogenesis.

Berry et al. recently reported the results of HIV-1 and HIV-2 plasma viral loads from individuals selected to represent a range of CD4⁺ cell counts [39]. They also found a large difference in viral load among asymptomatic individuals, who did not have significantly impaired CD4⁺ cells, and were unable to detect RNA in more than half of those with CD4⁺ total lymphocytes >14%, roughly corresponding to the absolute CD4⁺ cell counts seen in HIV-2 infection in this study. Although they were able to evaluate HIV-1 and HIV-2 viral load among those with lower CD4⁺ cell counts, their study design did not capture the distribution of CD4⁺ cell counts that results from the natural history of HIV-1 and HIV-2 infection; the reduced rate of disease progression after infection with HIV-2 is readily observed in the prospective cohort from which we obtained samples. Our study used an RNA assay with high sensitivity and demonstrated that the majority of HIV-2 infections result in lower plasma viremia than that found in HIV-1 infection. Further studies in this group of individuals will be useful for identifying the reasons for differences in viral load and pathogenicity of HIV-1 and HIV-2, and for clarifying the relationship of HIV-2 viral load and disease.

Acknowledgments

We thank Dr. Phil Kantoff, Krishna Krithivas, Fredrik Vannberg, Dr. Ibrahima Traoré, and Dr. Mamadou C. Dia for their assistance; and Drs. Boris Renjifo and Jean-Louis Sankalé for helpful comments and discussion.

References

Author notes