Steady increase in cellular HIV-1 load during the asymptomatic phase of untreated infection despite stable plasma viremia
17 July 2010
Pasternak, Alexander O; Jurriaans, Suzanne; Bakker, Margreet; Berkhout, Ben; Lukashov, Vladimir V
aLaboratory of Experimental Virology, Department of Medical Microbiology, Center for Infection and Immunity Amsterdam, The Netherlands
bLaboratory of Clinical Virology, Department of Medical Microbiology, Academic Medical Center of the University of Amsterdam, Amsterdam, The Netherlands.
"It has been previously shown that levels of HIV-1 usRNA in nonprogressors, whose CD4+ cell counts are high and stable, are significantly lower than those in progressors [16,17]. In accordance with these earlier observations, we now show that the levels of usRNA and prDNA, but not the plasma RNA levels, inversely correlate with the CD4+ cell count, which is itself a marker of infection progression (Fig. 4). These results indicate that the HIV-1 replication rates in PBMCs during the asymptomatic phase directly correlate with the state of the immune system (which is reflected in CD4+ cell counts).....Therefore, progressive weakening of the antiviral immune response during the asymptomatic phase might be one of the factors defining the temporal increase in the relative numbers of HIV-producing cells, and, therefore, the increase in HIV-1 replication rates in PBMCs that we and others have found.....This observation, as well as our recent finding that levels of HIV-1 usRNA in PBMCs of cART-treated patients with undetectable plasma viremia are significantly higher in patients who subsequently fail therapy than in patients who remain virologically suppressed  positions the HIV-1 usRNA in PBMCs as a unique HIV laboratory marker that may, in several aspects, perform better than the 'traditional' plasma viral RNA load."
Objective: To compare the dynamics of HIV-1 molecular markers in peripheral blood mononuclear cells (PBMCs) and in plasma during the asymptomatic phase of untreated HIV-1 infection.
Design and methods: Using seminested real-time PCR assays, we measured the levels of HIV-1 proviral (pr) DNA, unspliced (us) RNA, and multiply spliced RNA in the PBMCs of 10 untreated HIV-1-infected men at multiple time points during the asymptomatic phase of infection and compared the longitudinal trends of these markers with those of viral RNA in plasma.
Results: Whereas plasma RNA levels did not significantly change in any of the individuals, levels of usRNA significantly increased with time in six out of 10 persons, and levels of prDNA in four. Slopes, changes, and time-weighted changes from baseline of usRNA, prDNA, and CD4+ cell count, but not of plasma RNA, were significantly different from zero (P < 0.01). No significant longitudinal trend of plasma RNA was observed in the study group using linear mixed models, whereas the trends of usRNA, prDNA, and CD4+ cell count were highly significant (P < 0.001). usRNA levels increased significantly faster than those of plasma RNA or prDNA, suggesting a temporal increase in viral replication rates in PBMCs. Finally, CD4+ cell count inversely correlated with levels of usRNA and prDNA, but not with plasma RNA level.
Conclusion: During the asymptomatic phase of untreated HIV-1 infection, when virion production and clearance are balanced, resulting in stable plasma viremia, viral load in PBMCs steadily increases and is a sensitive and direct longitudinal virological marker of infection progression.
The clinical progression of HIV-1 infection in untreated individuals is characterized by the asymptomatic phase of variable length, during which CD4+ cell count shows a consistent decline, whereas plasma viral RNA level in many HIV-1-infected individuals exhibits the 'setpoint' (steady-state) kinetics, starting to significantly rise only before the onset of AIDS [1-5]. Such 'setpoint' kinetics is believed to reflect the balance between the virus production and clearance. This balance, however, may be imperfect in other HIV-infected individuals, and their plasma viremia would, therefore, slowly increase [6,7]. Individuals who have higher plasma HIV-1 RNA setpoint progress to AIDS faster , but a recent large study found no significant correlation between HIV-1 plasma RNA load and the rate of CD4+ T-cell decline . This means that no reliable longitudinal virological markers of the progression of HIV infection in the asymptomatic phase are known.
Several recent studies [10-12] have investigated kinetics of HIV-1 molecular markers (RNA and DNA) in peripheral blood mononuclear cells (PBMCs) of patients on combination antiretroviral therapy (cART). However, it is unclear to what extent the levels of HIV-1 molecular markers in PBMCs reflect the progression of infection in untreated individuals. HIV-1 DNA levels in PBMCs can predict the subsequent disease progression [13-15], and early reports have demonstrated that cellular HIV-1 RNA levels correlate with disease progression [16,17], but studies on the longitudinal trends of HIV-1 markers in PBMCs have yielded conflicting results [18-20].
We have recently developed sensitive seminested real-time PCR methods based on TaqMan chemistry for quantitation of intracellular HIV-1 proviral (pr) DNA and both unspliced (us) and multiply spliced (ms) RNA forms . Using these methods, we have recently shown that levels of HIV-1 usRNA in PBMCs from patients on cART with undetectable plasma viremia are predictive of the therapy outcome . Here, we used these methods to measure the longitudinal trends of the HIV-1 molecular markers in PBMCs of untreated individuals in the asymptomatic phase of the infection.
In this study, we directly compared the longitudinal trends of HIV-1 load in PBMCs and plasma during the asymptomatic phase of the infection in untreated individuals. We found remarkable differences in the dynamics of these molecular markers. Whereas levels of viral RNA in plasma were stable, those of usRNA and prDNA in PBMCs were significantly increasing over time. Levels of usRNA increased in time significantly faster than those of prDNA, as the usRNA/prDNA ratio grew significantly as well (Fig. 3). On the other hand, no significant longitudinal increase was observed in the usRNA/msRNA ratio (Fig. 3), arguing against a temporal shift in the HIV-1 splicing pattern with the progression of infection. As participants were followed for different periods and as the observation periods were not synchronized with respect to time from seroconversion, these longitudinal trends, in a strict sense, should not be interpreted as reflecting the natural history of HIV-1 viremia. Ideally, generalization of our findings requires a larger study with patients followed from seroconversion until the onset of clinical symptoms of AIDS. Design of such a study, however, would require a large sample collection established before the era of cART.
Before the cART era, several studies have investigated the dynamics of HIV-1 molecular markers in PBMCs. Gupta et al.  and Furtado et al.  observed a longitudinal increase in the level of HIV-1 RNA in PBMCs during the asymptomatic phase in five and six typical progressors, respectively. However, these studies did not compare the dynamics of viral markers in PBMCs and plasma. Such comparison was performed by Bagnarelli et al. , who found no significant longitudinal trends in the HIV-1 markers in both PBMCs (usRNA and prDNA) and plasma in 12 untreated patients. The differences between their findings and the results of the present study could relate to the fact that their longitudinal analysis was limited to 0.5-2 years, and such a small time frame may not adequately reflect the progression of HIV-1 infection. An alternative explanation is that the sensitivity and accuracy of the competitive PCR methods they used may be suboptimal compared to the seminested real-time PCR assays based on TaqMan chemistry that we used.
It has been previously shown that levels of HIV-1 usRNA in nonprogressors, whose CD4+ cell counts are high and stable, are significantly lower than those in progressors [16,17]. In accordance with these earlier observations, we now show that the levels of usRNA and prDNA, but not the plasma RNA levels, inversely correlate with the CD4+ cell count, which is itself a marker of infection progression (Fig. 4). These results indicate that the HIV-1 replication rates in PBMCs during the asymptomatic phase directly correlate with the state of the immune system (which is reflected in CD4+ cell counts). Whether the observed increase in viral replication in PBMCs, as the infection progresses, reflects an increase in the relative numbers of HIV-producing cells per HIV DNA+ cell, or an upregulation of viral transcription in PBMCs on the cellular level, remains currently unclear. In agreement with our recent observation that the levels of HIV-1 usRNA in PBMCs of cART-treated patients strongly inversely correlate with their pretherapy CD4+ cell counts , in the present study, we observed significant inverse correlations between the baseline CD4+ cell counts and the subsequent longitudinal slopes of usRNA and prDNA. Interestingly, by quantitative microculture assay, Gupta et al.  demonstrated that the relative numbers of HIV-producing cells in untreated individuals paralleled the levels of usRNA and inversely correlated with the CD4+ cell counts. Therefore, progressive weakening of the antiviral immune response during the asymptomatic phase might be one of the factors defining the temporal increase in the relative numbers of HIV-producing cells, and, therefore, the increase in HIV-1 replication rates in PBMCs that we and others have found. Alternatively, cellular and/or viral factors might influence the rate of viral replication in PBMCs. Several genome-wide studies have shown associations of host genetic variants with control of HIV-1 and with rates of disease progression [23-25], and interesting differences in host genetic loci associated with HIV-1 plasma RNA and cellular DNA levels were reported . However, another recent study  failed to identify a distinct cellular mRNA expression profile associated with effective viral control and concluded that changes in global RNA expression reflect responses to viral replication.
Whatever might be its cause, the temporal increase in the HIV replication levels in PBMCs during the asymptomatic phase should trigger a corresponding increase in the amounts of free virus produced. Yet, we did not measure any significant rise in plasma HIV-1 RNA load. This may be explained by the fact that the plasma HIV-1 RNA load reflects the balance between the virus production and clearance, whereas the cellular HIV-1 RNA load reflects the virus production only. It is possible that the ability of the immune system to clear free virus is compromised during the asymptomatic phase to a lesser extent than the ability to clear HIV-infected cells. During this phase of infection, the weakening immune system may still possess a capacity to adjust the virus clearance rates to the increased virus production; when this capacity is exhausted, the uncontrolled virus replication rapidly leads to the onset of AIDS. An alternative explanation for the discrepancy between the longitudinal trends of the viral load in PBMCs and plasma could be that because most HIV-1 replication takes place in the lymphoid tissue, the virus produced by the infected PBMCs might not substantially contribute to the total amount of free virus found in plasma. However, our recent observation that levels of viral RNA in PBMCs from patients on cART predict the therapy outcome  suggests that the dynamics of HIV-1 molecular markers in PBMCs adequately reflects the virus replication processes in an infected individual.
Whereas plasma HIV-1 load is predictive of the disease progression early after seroconversion, only the CD4+ cell counts remain predictive later in the asymptomatic phase . Furthermore, a large recent study has demonstrated that only a small proportion (4-6%) of the CD4+ T-cell loss variability could be explained by the plasma HIV-1 load . Here, we have shown that HIV-1 load (especially, HIV-1 usRNA level) in PBMCs is a more direct and sensitive marker of the infection progression in the asymptomatic phase than the viral load in plasma. This observation, as well as our recent finding that levels of HIV-1 usRNA in PBMCs of cART-treated patients with undetectable plasma viremia are significantly higher in patients who subsequently fail therapy than in patients who remain virologically suppressed  positions the HIV-1 usRNA in PBMCs as a unique HIV laboratory marker that may, in several aspects, perform better than the 'traditional' plasma viral RNA load.
Quantitation of HIV-1 molecular markers (prDNA, usRNA, and msRNA) was performed in the cryopreserved longitudinal PBMC samples from 10 untreated HIV-1-infected individuals in the asymptomatic phase of infection. usRNA and prDNA were detectable in all 53 PBMC samples, whereas msRNA was undetectable in PBMC samples of two individuals (P9 and P10), presumably due to the mismatches between the real-time PCR primers or probe and the viral templates. In individuals P1-P8, msRNA was detectable in 40 out of 43 PBMC samples. For the three samples, in which msRNA was undetectable, the msRNA values were left censored at the detection limit. As secondary markers, we calculated the ratios usRNA/prDNA and usRNA/msRNA.
The cumulative follow-up data for all participants are shown in Fig. 1. Table 1 shows the longitudinal trends of the analyzed parameters in the individual participants. Whereas plasma RNA levels have not significantly changed over time in any of the participants, statistically significant longitudinal changes, as assessed by linear regression, were observed in the levels of usRNA (six participants, all trends positive), prDNA (four, all positive), msRNA (one positive), usRNA/prDNA (two positive and one negative), usRNA/msRNA (two, both positive), and CD4+ cell count (five, all negative). Levels of usRNA and CD4+ cell count have significantly changed in time in significantly larger proportions of participants than has plasma RNA load (P = 0.01 and P = 0.03, respectively).
Figure 2 shows the longitudinal slopes, changes during the observation period, and TWC of the analyzed parameters in the study group. Slopes, changes, and TWC of usRNA, prDNA, and CD4+ cell count were significantly different from zero (P < 0.01), meaning significant longitudinal trends, whereas no such trends were observed for the levels of msRNA and plasma RNA (P > 0.1). The slope, change, and TWC values of usRNA were significantly larger than the corresponding values of plasma RNA (P = 0.002 for both slope and TWC and P = 0.004 for changes during the observation period). The differences between the slope and TWC values of usRNA and prDNA were also significant (P = 0.05 and P = 0.04, respectively).
To estimate the average longitudinal trends of the analyzed parameters, we fitted the linear mixed models, taking into account correlations of repeated measurements within the individuals (Fig. 3). In this analysis, longitudinal trends of usRNA, prDNA, usRNA/prDNA, and CD4+ cell count were highly statistically significant (P ≤ 0.001), the trend of msRNA was also significant (P = 0.01), but no significant longitudinal trend was observed for plasma RNA levels (P = 0.08) and usRNA/msRNA (P = 0.2). The F values were the highest for the trends of CD4+ cell count and usRNA, followed by prDNA and usRNA/prDNA.
Next, we assessed Spearman's correlations between baseline and slope values of the analyzed parameters (see Table, Supplemental Digital Content 1, http://links.lww.com/QAD/A42). Interestingly, we observed significant inverse correlations between the baseline CD4+ cell counts and the subsequent longitudinal slopes of usRNA and prDNA (rs = -0.68; P = 0.04 for both usRNA and prDNA), but not of plasma RNA (rs = -0.43; P = 0.2). No significant correlation, however, was found between the baseline levels of any of the studied virological markers and subsequent CD4+ cell count slopes. In addition, the slopes of prDNA (rs = -0.82, P = 0.006) and usRNA (rs = -0.72, P = 0.02), but not of plasma RNA (rs = -0.39, P = 0.3), inversely correlated with the observation periods.
Finally, we assessed Spearman's correlations between the pairs of analyzed parameters in all participants combined. Strong inverse correlations were observed between CD4+ cell count and levels of usRNA or prDNA, but not between CD4+ cell count and plasma RNA (Fig. 4). As expected, usRNA levels correlated strongly and in a positive manner with prDNA and msRNA levels, and to the lesser extent with levels of plasma RNA.
Materials and methods
Study participants and samples
We used archival PBMC samples from HIV-1-infected individuals who participate in the Amsterdam Cohort Studies (ACSs) on HIV infection and AIDS. The ACSs have been conducted in accordance with the ethical principles set out in the Declaration of Helsinki, and written informed consent has been obtained prior to sample collection. The study has been approved by the ACS committee. The ACSs have been approved by the Medical Ethical Committee of the Academic Medical Center.
We have selected 10 HIV-1-infected individuals who were in the asymptomatic phase of infection and did not receive ART for the whole study period (average, 4.6 years; range 1.5-10.6 years). Nine individuals were ART-naive at the inclusion date (all of them started ART after the study period), and one (P2) was ART-experienced (stopped the therapy 3 months before the inclusion date). For the ART-naive individuals, the average inclusion year was 1990 (range 1986-1996), and P2 was included in 2000. Seroconversion dates could be estimated for five individuals (P6-P10), based on the date of their last HIV-negative and first HIV-positive sample. The rest (P1-P5) were already seropositive at entry in the ACS. For every individual, we included only PBMC samples from more than 6 months from the first HIV-positive sample. This study was stopped when a participant started ART, or when his CD4+ cell count dropped below 200 cells/µl in two consecutive measurements. All participants were homosexual men infected with HIV-1 subtype B strains, and their ages at inclusion are shown in Table 1.
The number of PBMC samples per study participant ranged from four to eight, with a total of 53 samples. Concurrent CD4+ cell count measurements were available for all PBMC samples used in the study, and concurrent measurements of plasma viral RNA were available for 51 of the 53 samples.
Quantitation of HIV-1 load in peripheral blood mononuclear cell samples
For quantitation of cellular HIV-1 RNA and DNA load, PBMCs were isolated by standard Ficoll-Hypaque density gradient centrifugation and frozen in aliquots in liquid nitrogen. Total cellular nucleic acids were extracted from PBMC samples (-106 PBMC was used for one extraction) according to the isolation method of Boom et al. , eluted in water, and frozen in aliquots at -80°C until further processing.
HIV-1 prDNA and both forms of cellular HIV-1 RNA (usRNA and msRNA) were quantified by seminested real-time PCR, as described earlier . The eluted cellular DNA was directly subjected to two rounds of PCR amplification: a limited-cycle preamplification step and a real-time PCR step, using seminested primers. For RNA quantitation, the eluted RNA samples were first subjected to DNase treatment (DNA-free kit; Ambion Inc., Austin, Texas, USA), to remove HIV-1 prDNA, which could interfere with the quantitation, and subsequently to reverse transcription. For both usRNA and msRNA assays, two rounds of amplification with seminested primers were performed on the resultant cDNA. For all assays, no positive signals have been obtained from the no-template PCR controls, as well as from the PCR controls without the reverse transcription step (for RNA assays), which were included in the quantitation.
The amounts of PBMC-derived HIV-1 DNA and RNA were normalized to total cellular inputs, which were quantified in separate real-time PCR by using the β-actin detection kit (Applied Biosystems Inc., Foster City, California, USA) and expressed as number of copies per 106 PBMCs. To control for the reverse transcription efficiency in the RNA assays, we used the ribosomal RNA detection kit (Applied Biosystems). The sensitivity of all three assays was four copies per reaction, which translated into approximately 40 copies/106 PBMCs for the prDNA assay and 80 copies/106 PBMCs for RNA assays, and the linear range was at least five orders of magnitude. The sensitivity, reproducibility, and accuracy of these assays have been documented earlier [12,21].
For cellular and plasma HIV-1 load, statistical analyses were performed on log10-transformed values. Longitudinal trends in individual participants were assessed by using the linear regression, and in the study group by fitting the linear mixed models. The slopes were calculated by linear regression, and the time-weighted changes from baseline (TWC) by linear trapezoidal integration. Changes during the observation period were calculated by subtracting the baseline value from the value at the last time point. Slopes, changes, and TWC were compared with zero and between each other by using the paired Wilcoxon signed-rank test. Correlations of virological parameters and CD4+ cell count were assessed by using Spearman's tests. Proportions of participants were compared by using Fisher's exact test. Linear mixed model analysis was performed by using SPSS 16.0 (http://www.spss.com/), and all other statistical tests were performed by using GraphPad Prism 5.01 (http://www.graphpad.com). All statistical tests were two-sided. P values less than 0.05 were considered statistically significant.