Persistence of HIV infection despite suppressive antiretroviral therapy (ART) is suspected to be due to viral reservoirs which include long-lived infected cells, latent infection in various forms and sanctuary sites [1–4]. The role some of these play in viral production has been estimated previously from the dynamics of plasma HIV RNA levels (pVL) and HIV DNA after commencement of ART . Although pVL quickly decreases at a rate initially indicative of the loss of productively infected CD4+ T cells, second and third phases occur that reflect other sources, such as long-lived infected cells and reactivation of latently infected cells. Ultra-sensitive assays have shown that pVL remains detectable despite prolonged ART .
HIV DNA exhibits significantly less change over time. This cellular component is established early in infection and decreases very slowly following ART [7,8]. Moreover, viral DNA is expressed in a variety of forms including integrated and productive, integrated but latent, an unintegrated linear form, and circular episomal forms of 1-long terminal repeats (LTR) and 2-LTR HIV DNA that represent aborted integration events . The timing of progression through these separate stages of the viral life cycle and their implications for levels of ongoing infection with ART are uncertain and of considerable interest.
Previous investigations of the integrase inhibitor raltegravir (RAL) revealed a marked effect on second phase plasma HIV RNA levels reducing it by 70% in comparison to a regimen that did not contain RAL. Those results were obtained from a study in which pVL was measured with an assay cut-off of 50 RNA copies/ml and did not assess HIV DNA, so provided limited insight into interaction between these viral forms [10,11]. The PINT study was formulated to determine the effects of combination ART including RAL on pVL as well as on the latent viral reservoir, when applied to therapy-naïve patients during primary (PHI) compared to chronic HIV infection (CHI). Identifying the links between pVL phases and different forms of viral DNA was a crucial part of this investigation. We collected data on each of these forms frequently throughout the period of investigation and to low levels of quantification. Comparison between the PHI and CHI patients and their different durations of infection allowed us to investigate how these reservoirs develop over time in the absence of ART; analysis of changes after commencing this ART regimen also provided estimates of its impact on suppressing ongoing viral replication.
Acute PHI was documented by the presence of three bands or less on a Western blot and a positive p24 antigen and/or positive proviral DNA equivalent to Fiebig stage IV or earlier. Early PHI was defined by documented infection diagnosed by a positive BED ELISA result or previously negative serology within 6 months of confirmed positive serology. CHI was defined by documented HIV infection of at least 12 months duration. Patient baseline characteristics are described in Supplementary Material (Supplementary Table 1, http://links.lww.com/QAD/A201).
pVL, total HIV DNA, integrated HIV DNA and episomal HIV DNA (2-LTR) copies per 106 CD4+ T cells were determined as previously described . Briefly HIV RNA copies/ml of plasma were determined by an assay with a quantification limit of 0.3 copies/ml . HIV DNA was extracted from purified peripheral blood CD4+ T cells. Total HIV DNA was quantified by a real-time PCR assay specific for HIV-gag, whereas integrated HIV DNA was quantified using a nested real-time PCR . Quantification of episomal HIV DNA was based on a real-time PCR assay specific for the 2-LTR junction which incorporated a sequence-specific, dual-labelled fluorogenic TaqMan probe .
The mean responses over all patients and for each viral species were determined using non-linear mixed effects (NLME) models in which each parameter was assumed to have a fixed and random effect. A bi-phasic or tri-phasic model of the form
was fitted to the data using the nlmefit method in Matlab R2010b (see Supplementary Methods, http://links.lww.com/QAD/A201).
Comparisons between groups were performed with a Wilcoxon rank-sum test and a significance value of P less than 0.05 was chosen. Since this was an exploratory study no adjustments were made for multiple testing. All numerical calculations were performed within Matlab.
The PINT study recruited eight PHI and eight CHI, ART-naive patients who commenced a combination regimen consisting of tenofovir disoproxil fumarate and emtricitabine (Truvada) and the integrase inhibitor RAL. Patients were followed for 1 year with intensive quantification of pVL and cell-associated HIV DNA virus.
HIV RNA dynamics
Except for the time point halfway through the first week, CHI individuals exhibited significantly higher pVL from commencement of ART to week 12 (Fig. 1a), encompassing the period typically described as including first and second phase viral decay . There was an average 3-week delay between diagnosis of PHI and commencement of ART, and this may have contributed to their lower pVL at early time points. The mean response, obtained using a nonlinear mixed effects model (Methods section), showed three distinct phases including a virtually flat third phase (doubling time: 25 years). pVL was not different between groups during this final phase (P = 0.9) settling at a mean of 9 HIV RNA copies/ml by approximately week 16. Second phase decay rates were significantly faster for PHI patients (t1/2 5.9 days for PHI and 15.2 days for CHI, P = 0.007 comparison of medians; Figure 2a, b, Supplementary Table 2, http://links.lww.com/QAD/A201).
Previous comparisons of ART that contained RAL or the non-nucleoside reverse transcriptase inhibitor efavirenz revealed a 70% lower second phase viral level for the integrase inhibitor (INI) regimen . This calculation was based on linear regression of the identified second phase extrapolated back to time zero and taken as a proportion of baseline HIV RNA levels, termed the M ratio. The median M ratio for the RAL arm in that study was 0.0042 compared to 0.014 for the efavirenz arm. The median M ratios of 0.0034 in both PHI and CHI groups in this study are consistent with previous calculations. The faster second phase decay rate for the PHI group also suggests the source of pVL during second phase is more susceptible to clearance in PHI, at least under the influence of this RAL regimen.
Total HIV DNA
There was less than a 10-fold change over the year in HIV DNA, whereas HIV RNA decreased 10 000-fold. Total HIV DNA levels per CD4+ T cell were significantly lower for PHI compared to CHI at 6 of 11 time points over the course of the year (Fig. 1), exhibiting two phases of decay (Fig. 2c, d), but decay rates did not differ between groups. The mean decay rates of total HIV DNA for the two phases corresponded to half-lives of 31 and 1206 days, respectively.
Integrated HIV DNA
Levels of integrated HIV DNA/106 CD4+ T cells showed a similar degree of variation as total HIV DNA copies and were significantly higher for CHI at all time points (Fig. 1d). Integrated HIV DNA exhibited a biphasic decay (Fig. 2e, f) with a significantly faster first phase in PHI than in CHI (mean half-life of 10 days versus 43 days, P = 0.04), reaching its second phase within 63 and 172 days, respectively. Second phase decay rates were not significantly different and overall individuals showed a slight increase with time (doubling time of 3401 days), although this was not significantly different from no change with time (P = 0.5).
2-long terminal repeat circles
At no time were 2-LTR copies/106 CD4+ T cells different between groups (Fig. 1c). Furthermore, their early dynamics were very different to other components of viral DNA. Rather than decreasing from the outset, median 2-LTR levels increased for a period of 2–4 weeks (Fig. 1c, Fig. 2g, doubling time 11.3 days). After this time they decayed rapidly over the following 4–6 weeks with a half-life of 24.6 days, before settling into a slower decay rate. This profile occurred in both groups. Excluding this transient behaviour, during the period from baseline to day 50, there was a slow decay in 2-LTR numbers with a half-life of 169 days. This rate of loss was faster than the second phases of each of the other viral DNA measurements.
The aborted integration events represented by the spike in 2-LTR numbers were partly reflected in the early loss of integrated HIV DNA copies in the primary patients. Redirection of unintegrated linear HIV DNA away from successful integration towards episomal DNA in the presence of an integrase inhibitor (Fig. 3a) is expected from this class of drug .
How much unintegrated linear HIV DNA is there?
At baseline median levels of 2-LTR circles and integrated HIV DNA were 38% of median levels of total HIV DNA for the primary group and 31% for the chronic group. From week 4 onwards in PHI median total HIV DNA was almost identical to median copies of 2-LTR circles which indicates that 1-LTR circles were a minor component. Integrated DNA was approximately 1/10th as prevalent as total DNA (Fig. 3c, d). We were unable to directly measure linear unintegrated HIV DNA; however, if it is assumed that it represents the remainder after subtracting 2-LTR and integrated HIV DNA from total HIV DNA, then at baseline linear HIV DNA constitutes 62% of total HIV DNA in PHI and 69% in CHI. After week 2 the proportion of estimated linear HIV DNA (total minus 2-LTR minus integrated) was significantly smaller in PHI individuals (P = 0.03; Fig. 3c, d), and although this decreased with time, it comprised a substantial proportion of total HIV DNA at 3 months and out to 1 year for the three PHI and four CHI individuals where all viral DNA components were assessable (Fig. 3b).
Accumulation of integrated HIV DNA with duration of infection
The presence of higher levels of integrated HIV DNA in CHI may be due to its accumulation over the course of infection. The second phase levels of integrated HIV DNA reflect this accumulation of latent and/or defective integrated HIV DNA, and absolute levels significantly correlated with duration of infection (r = 0.82, P = 0.001). This increase in integrated HIV DNA levels with duration of infection has been previously noted .
A simple description of the dynamics of this integrated latent pool, that will comprise replication competent and replication defective forms, is that it is formed at constant rate a but is depleted through cell loss and activation at a proportional rate b (Supplementary Methods, http://links.lww.com/QAD/A201). Nonlinear regression fitting of this model shows that this latent pool accumulates with time achieving 50% of its maximum after 2 years (Fig. 4a). The optimal value for the parameter describing the rate at which this latent pool forms implies that on average there are approximately 100 000 integration events per day in an untreated individual (Supplementary Methods, http://links.lww.com/QAD/A201); furthermore after commencement of ART this pool decreases with a half-life of approximately 2 years.
Modelling progression of linear unintegrated HIV DNA to either integration or episomes
In order to assess how different species of HIV DNA are dynamically related we constructed a mathematical model describing generation of 2-LTR circles and integrated HIV DNA from linear unintegrated HIV DNA. Since it seems that linear and integrated species in CHI also contain forms that may have accumulated with duration of infection, we only compared the model to data from PHI. We used median data since there were only three primary patients who had all species assayed that were above detection limits. We restricted the fitting procedure for integrated HIV DNA to the first phase values (weeks 0, 2, and 4) that are more likely to represent the loss of the productively infected component as new infection is shunted to 2-LTR circles under the impact of RAL. The mathematical model assumed linear HIV DNA is produced at a constant rate subject to the efficacy of the reverse transcriptase inhibitor (RTI). This then flowed on to integration, or to 2-LTR forms at some background level before ART and which is elevated after commencing ART.
Nonlinear regression fitting of the mathematical model (Fig. 3a, Supplementary Methods, http://links.lww.com/QAD/A201) to the median data determined half-lives of 18 days for linear unintegrated HIV DNA, 29 days for 2-LTR circles or the cells containing them, 24 days for cells with integrated HIV DNA in the first phase loss, and 656 days for the half-life of the source of infection, and rates of progression from unintegrated linear to integrated with a half-life of 3.2 days (Fig. 4b). This is consistent with a relatively short half-life of 2-LTR circles or the cells that contain them. Their high levels after 1 year of therapy also indicates substantial new infection of cells with HIV RNA although the vast majority of this new infection will not result in productively infected cells, since the modelling estimated a 97% efficacy for the RTI component of the regimen and those viruses that bypass that inhibition progress mostly to the aborted 2-LTR form. By our estimates 99.9% of the 2-LTR circles present after 1 year had arisen from new rounds of HIV transmission occurring after the commencement of this regimen. Performing individual nonlinear regression fitting to the data for the 3 PHI individuals also indicated 2-LTR or the cells in which they reside were relatively short-lived with half-life estimates of 14.9, 24.6 and 3.7 days (Supplementary Figure 1, http://links.lww.com/QAD/A201).
Median data over a group may not represent individual behaviour well. Hence we also used mixed effects modelling to fit a simpler model to total and episomal HIV DNA data for the seven individuals (three primary and four chronic) with detectable total, episomal and integrated data (Supplementary Methods, http://links.lww.com/QAD/A201). This analysis indicated episomal DNA had a half-life of 16 days. The 95% confidence interval (CI) for the decay rate of 2-LTR circles corresponded to a half-life range of 4.8–54 days, which encompasses the 29-day half-life estimate from the previous method.
All of these methods indicated half-lives for linear unintegrated and episomal HIV DNA in the range of 2–4 weeks.
This RAL-containing regimen quickly reduced pVL to below the limits of quantification of standard assays. By week 16 pVL had reached 9 HIV RNA copies/ml where it stabilized. Unlike individuals on standard regimens who exhibited continued reduction in HIV RNA levels for a further year after reaching viral suppression (less than 50–75 HIV RNA copies/ml) , these individuals exhibited virtually no change in their final phase from week 16. Hence RAL may achieve the lowest levels of viral replication faster than other antiretroviral classes.
If the source of viral production responsible for second phase virus can be represented by a HIV DNA species examined here, then that component should be lower in PHI as this group exhibited lower second phase levels (Fig. 1a). Both total HIV DNA and integrated HIV DNA were lower in PHI. Moreover our calculations suggest there is a smaller pool of unintegrated linear HIV DNA in primary patients (Fig. 3c, d). One hypothesis producing differential effects on second phase assumes a pool of unintegrated linear HIV DNA is available for conversion to productive infection and that RAL reduces this conversion . Our analyses suggest a substantial reservoir and/or renewed generation of this form particularly for CHI (Fig. 3b, d). An excess of total to integrated HIV DNA has also been observed in patients on ART regimens that do not contain an integrase inhibitor , even when accounting for 2-LTR levels . Hence the contribution of the latent pool of linear unintegrated HIV DNA to new rounds of productive infection may be considerably higher than previously supposed.
However, because linear unintegrated DNA is problematical to measure directly, this analysis is indirect based on addition of numbers generated from different PCR results, and needs to be interpreted with caution. Although optimized for each patient there is no guarantee that they perform with equal efficacy across each assay. Moreover we could not directly assess 1-LTR levels; however, it would be difficult to describe why these would accumulate with duration of infection, whereas 2-LTR levels do not. Furthermore the data presented here only assessed reservoirs of HIV infection that circulate through peripheral blood and so could not include components such as infected macrophages.
Our calculations of integrated ‘latent’ HIV DNA did not assess its replication competence and so these estimates will exceed the amount of that true viral reservoir [7,8]. One of the major aims of the PINT study was to better understand the effects of an integrase inhibitor on the interplay between HIV RNA and the various viral DNA forms. We therefore focused on a detailed, longitudinal analysis of the time course of changes rather than replication competence, because of the known in-vitro effects of RAL .
Our calculations of the accumulation of long-term latent infection, as represented by the second phase component of integrated HIV DNA, challenges the viewpoint that this reservoir is completely established early after primary infection. Under our model, which includes integration levels measured over durations of HIV infection from several months to 23 years, this component is generated at a fairly constant rate of approximately 100 000 integrations per day. It appears to accumulate quickly since the constant input rate initially greatly outweighs their proportional loss. By our estimates, it is only when levels are high, at approximately 800 integrated HIV DNA copies per 106 CD4+ T cells, that input matches loss in the absence of therapy, and long-term integrated HIV DNA levels saturate. Although this accumulation drastically slows with time, this reservoir may continue to increase. Approximately half of our estimated eventual level of long-term integrated HIV DNA is achieved 2 years after HIV infection (Fig. 4a). The fast and progressive build-up of this long-lived latent pool of infection indicates that there are diminishing impacts upon this reservoir by commencing ART later in infection with the greatest per year return achieved with very early therapy.
Other studies that have addressed this question have showed that levels of replication competent latent infection did not change significantly in the absence of ART over a time period of up to 4 months after primary infection , and also ranged mostly over a 1 log10 range soon after commencement of ART despite different durations of infection . Our results agree with these findings in that the second phase component of integrated HIV DNA only ranged over a 1.6 log10 interval despite a difference of 22 years in duration of untreated HIV infection (Figs. 2 and 4a). However, this 22-year period with no ART allowed us to determine a statistically significant increase in this estimate of latent infection, and the question is whether replication competent infection also exhibits this change.
There is conflicting evidence for the stability of 2-LTR circles, and/or of the lifespan of cells in which they reside [9,22–24]. The results of the mathematical modelling undertaken in the analysis of 2-LTR circles presented in this study preferentially support the view that they are labile or the population of cells in which they survive is not long-lived. Firstly, there was no difference in 2-LTR levels between the CHI and PHI groups. If 2-LTR circles survived indefinitely within cells and arose by chance during abortive infection events while ART-naive, then they should accumulate to greater levels in CHI than PHI. Differences of this sort were observed for total and integrated forms of HIV, but not for 2-LTR circles. Secondly, the fast increase in 2-LTR copy numbers after commencement of therapy with an integrase inhibitor was followed by a similarly quick decline with a half-life of approximately 25 days. If 2-LTR circles were long-lived then they would decrease from the early elevated level at a slow rate representative of their loss with cell death and dilution by cellular division .
Under our calculations, the long-term persistence of 2-LTR circles was a result of the slow decay of the source of infection, the continued production of linear HIV DNA despite greater than an estimated 90% efficacy of the RTI, and the driving force of the INI that redirects integration to the episomal form. ART that does not contain an integrase inhibitor results in almost all HIV DNA after 1 year being in an integrated form . Our results are consistent with this, given that the integrase inhibitor in our study, which promotes the occurrence of episomal HIV DNA, is replaced in these other studies by a drug that reduces their likelihood. The observed peak in 2-LTR circles after therapy intensification with RAL  is also consistent with our findings. The lower 2-LTR levels in that study are due to reduced levels of unintegrated HIV DNA and 2-LTR since the patients were ART-suppressed before addition of RAL.
Treatment with a RAL regimen quickly reduced pVL to low levels of approximately 9 HIV RNA copies/ml. Although this reduction was slower for chronically infected individuals there was no difference in the final phase pVL; however, there were significantly higher levels of latent integrated HIV DNA. With the slow decay of this latent pool, and continued entry of virus into cells as evidenced by the dynamics of 2-LTR circles, attempts to purge the latent reservoir would be assisted by earlier commencement of ART.
Conflicts of interest
This study is supported in part by a research grant from the Investigator-Initiated Studies Program of Merck and the Kirby Institute, National Health and Medical Research Council (NHMRC) programme grant number 510448, NHMRC project grant (J.Z., J.M., K.S.) number 510325 and a Practitioner Fellowship (A.K.). The Kirby Institute is funded by the Australian Government Department of Health & Ageing and is affiliated with the Faculty of Medicine, The University of New South Wales.
D.B. received travel support and honoraria from Merck Sharp and Dohme (MSD) and Gilead; S.E. has received honoraria, consultancies, research grants or been an investigator in clinical trials sponsored by Gilead Sciences and MSD; D.A.C. has received honoraria and consultancies from MSD; A.D.K. has received travel support from MSD. All other authors declare no conflicts of interest.
1. Chun TW, Fauci AS. Latent reservoirs of HIV: obstacles to the eradication of virus
. Proc Natl Acad Sci USA
2. Pomerantz RJ. Reservoirs of human immunodeficiency virus type 1: the main obstacles to viral eradication
. Clin Infect Dis
3. Shen L, Siliciano RF. Viral reservoirs, residual viremia, and the potential of highly active antiretroviral therapy to eradicate HIV infection
. J Allergy Clin Immunol
4. Trono D, Van Lint C, Rouzioux C, Verdin E, Barre-Sinoussi F, Chun T-W, et al. HIV persistence and the prospect of long-term drug-free remissions for HIV-infected individuals
5. Perelson AS, Essunger P, Cao Y, Vesanen M, Hurley A, Saksela K, et al. Decay characteristics of HIV-1-infected compartments during combination therapy
6. Palmer S, Maldarelli F, Wiegand A, Bernstein B, Hanna GJ, Brun SC, et al. Low-level viremia persists for at least 7 years in patients on suppressive antiretroviral therapy
. Proc Natl Acad Sci USA
7. Chun TW, Engel D, Berrey MM, Shea T, Corey L, Fauci AS. Early establishment of a pool of latently infected, resting CD4(+) T cells during primary HIV-1 infection
. Proc Natl Acad Sci USA
8. Finzi D, Blankson J, Siliciano JD, Margolick JB, Chadwick K, Pierson T, et al. Latent infection of CD4+ T cells provides a mechanism for lifelong persistence of HIV-1, even in patients on effective combination therapy
. Nat Med
9. Sharkey ME, Teo I, Greenough T, Sharova N, Luzuriaga K, Sullivan JL, et al. Persistence of episomal HIV-1 infection intermediates in patients on highly active antiretroviral therapy
. Nat Med
10. Murray JM, Emery S, Kelleher AD, Law M, Chen J, Hazuda DJ, et al. Antiretroviral therapy with the integrase inhibitor raltegravir alters decay kinetics of HIV, significantly reducing the second phase
11. Markowitz M, Nguyen BY, Gotuzzo E, Mendo F, Ratanasuwan W, Kovacs C, et al. Rapid and durable antiretroviral effect of the HIV-1 Integrase inhibitor raltegravir as part of combination therapy in treatment-naive patients with HIV-1 infection: results of a 48-week controlled study
. J Acquir Immune Defic Syndr
12. Koelsch KK, Boesecke C, McBride K, Gelgor L, Fahey P, Natarajan V, et al. Impact of treatment with raltegravir during primary or chronic HIV infection on RNA decay characteristics and the HIV viral reservoir
13. Palmer S, Wiegand AP, Maldarelli F, Bazmi H, Mican JM, Polis M, et al. New real-time reverse transcriptase-initiated PCR assay with single-copy sensitivity for human immunodeficiency virus type 1 RNA in plasma
. J Clin Microbiol
14. Suzuki K, Shijuuku T, Fukamachi T, Zaunders J, Guillemin G, Cooper D, et al. Prolonged transcriptional silencing and CpG methylation induced by siRNAs targeted to the HIV-1 promoter region
. J RNAi Gene Silencing
15. Chomont N, El-Far M, Ancuta P, Trautmann L, Procopio FA, Yassine-Diab B, et al. HIV reservoir size and persistence are driven by T cell survival and homeostatic proliferation
. Nat Med
16. Brussel A, Mathez D, Broche-Pierre S, Lancar R, Calvez T, Sonigo P, et al. Longitudinal monitoring of 2-long terminal repeat circles in peripheral blood mononuclear cells from patients with chronic HIV-1 infection
17. Hazuda DJ, Felock P, Witmer M, Wolfe A, Stillmock K, Grobler JA, et al. Inhibitors of strand transfer that prevent integration and inhibit HIV-1 replication in cells
18. Chomont N, El-Far M, Ancuta P, Trautmann L, Procopio FA, Yassine-Diab B, et al. HIV reservoir size and persistence are driven by T cell survival and homeostatic proliferation
. Nat Med
19. Hatano H, Delwart EL, Norris PJ, Lee TH, Neilands TB, Kelley CF, et al.Evidence of persistent low-level viremia in long-term HAART-suppressed, HIV-infected individuals
20. Agosto LM, Liszewski MK, Mexas A, Graf E, Pace M, Yu JJ, et al. Patients on HAART often have an excess of unintegrated HIV DNA: implications for monitoring reservoirs
21. Graf EH, Mexas AM, Yu JJ, Shaheen F, Liszewski MK, Di Mascio M, et al. Elite suppressors harbor low levels of integrated HIV DNA and high levels of 2-LTR circular HIV DNA compared to HIV+ patients on and off HAART
. PLoS Pathog
22. Sharkey M, Triques K, Kuritzkes DR, Stevenson M. In vivo evidence for instability of episomal human immunodeficiency virus type 1 cDNA
. J Virol
23. Pierson TC, Kieffer TL, Ruff CT, Buck C, Gange SJ, Siliciano RF. Intrinsic stability of episomal circles formed during human immunodeficiency virus type 1 replication
. J Virol
24. Butler SL, Johnson EP, Bushman FD. Human immunodeficiency virus cDNA metabolism: notable stability of two-long terminal repeat circles
. J Virol
25. Koelsch KK, Liu L, Haubrich R, May S, Havlir D, Gunthard HF, et al. Dynamics of total, linear nonintegrated, and integrated HIV-1 DNA in vivo and in vitro
. J Infect Dis
26. Buzon M, Massanella M, Llibre JM, Esteve A, Dahl V, Puertas MC, et al. HIV-1 replication and immune dynamics are affected by raltegravir intensification of HAART-suppressed subjects
. Nat Med