Secondary Logo

Journal Logo

Dynamic models of viral replication and latency

Mohammadi, Pejman; Ciuffi, Angela; Beerenwinkel, Niko

Current Opinion in HIV and AIDS: March 2015 - Volume 10 - Issue 2 - p 90–95
doi: 10.1097/COH.0000000000000136
GENOMICS IN HIV INFECTION: Edited by Amalio Telenti

Purpose of review HIV targets primary CD4+ T cells. The virus depends on the physiological state of its target cells for efficient replication, and, in turn, viral infection perturbs the cellular state significantly. Identifying the virus–host interactions that drive these dynamic changes is important for a better understanding of viral pathogenesis and persistence. The present review focuses on experimental and computational approaches to study the dynamics of viral replication and latency.

Recent findings It was recently shown that only a fraction of the inducible latently infected reservoirs are successfully induced upon stimulation in ex-vivo models while additional rounds of stimulation make allowance for reactivation of more latently infected cells. This highlights the potential role of treatment duration and timing as important factors for successful reactivation of latently infected cells. The dynamics of HIV productive infection and latency have been investigated using transcriptome and proteome data. The cellular activation state has shown to be a major determinant of viral reactivation success. Mathematical models of latency have been used to explore the dynamics of the latent viral reservoir decay.

Summary Timing is an important component of biological interactions. Temporal analyses covering aspects of viral life cycle are essential for gathering a comprehensive picture of HIV interaction with the host cell and untangling the complexity of latency. Understanding the dynamic changes tipping the balance between success and failure of HIV particle production might be key to eradicate the viral reservoir.

aInstitute of Microbiology, University Hospital of Lausanne, University of Lausanne, Lausanne

bDepartment of Biosystems Science and Engineering, ETH Zurich

cSwiss Institute of Bioinformatics, Basel, Switzerland

Correspondence to Angela Ciuffi, Institute of Microbiology, CHUV, 1011 Lausanne, Switzerland. Tel: +41 213144099; e-mail:

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License, where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially.

Back to Top | Article Outline


Current antiretroviral therapy (ART) is successful in inhibiting viral replication (defined as undetectable plasma viremia using standard assays, i.e., below 20–50 copies/ml) and transmission, but fails to completely eliminate HIV. Indeed, the presence of continuous low-level viremia (detectable using ultrasensitive assays) or occasional viral blips under ART and the observation that viremia rebounds rapidly upon ART cessation indicate the existence of viral reservoirs [1]. Viral reservoirs are established early, during the first 3 days upon viral exposure [2▪,3,4]. Their exact nature has yet to be completely understood, but includes anatomical sanctuaries and cellular reservoirs. Although still controversial, anatomical sanctuaries are described as sites with incomplete ART penetration, where infected cells may reside with continuous low levels of viral replication and may include lymph nodes, gut-associated lymphoid tissue, and the central nervous system [5–8]. Cellular reservoirs are described as being latently infected cells, i.e. infected cells that do not produce replication-competent virions and include mostly long-lived resting memory CD4+ T cells [8–10]. The current dogma is that the major HIV reservoir originates from activated CD4+ T cells that have been infected and survive while reverting to a resting state, thereby becoming a memory cell. Because of the intrinsic physiology of resting memory cells, no infectious virions are produced, hence the concept of latent infection. Bursts of viral production, however, may occur from time to time as manifested by viral blips. Although the cellular mechanisms contributing to sudden induction of viral blips in otherwise aviremic individuals are yet poorly defined, viral production and viral latency appear to be intertwined and closely linked to cell physiology and activation (Fig. 1).



Viral infection and replication are successful in activated CD4+ T cells, while poorly efficient in resting CD4+ T cells (Fig. 1a). The timing of the infection event might impact on the type of postintegration block in the latent cell (Fig. 1a and b). Therefore, viral reactivation from latency might need different types of stimuli to reverse several types of latency [11▪,12▪▪]. Furthermore, different types of latent cells may need to be exposed for different time lengths to successfully trigger a productive infection (Fig. 1c). Finally, the efficiency of viral production or the type of latency can impact the kinetics of cell death and thus the elimination of the reservoir of latently infected cells (Fig. 1d). An additional layer of complexity may reside in the possibility of having multiple blocks within the same cells, which would require multiple stimuli with specific time schedules of administration to be reactivated. In conclusion, the dynamics of establishment of a successful infection and production, including the type of cell stimulation, the time required for stimulation, the stimulated subset of cells, and the type of block, might be key for understanding viral latency and for the development of appropriate strategies aiming at purging viral reservoirs and eradicating HIV.

Box 1

Box 1

Back to Top | Article Outline


Initial studies measuring the size of the latent viral reservoir aimed at quantifying the number of circulating latently infected cells in the plasma. For this purpose, the gold standard is the viral outgrowth assay: resting CD4+ T cells are collected from patients, serially diluted, and activated with phytohemagglutin/interleukin 2 (IL-2) and 10× excess of irradiated peripheral blood mononuclear cells. CD4+ lymphoblasts (from a healthy HIV donor) are added the next day (day 2) and at days 7 and 14, and viral production is measured by p24 ELISA at day 21 [13▪,14]. Using this assay, 1 of 106 cells was reported to produce infectious particles and thus considered to be latently infected [13▪]. Interestingly, using PCR to detect proviral DNA, 1000 of 106 cells were shown to be infected by HIV [13▪]. Recent findings [15] showed that 88.3% of infected cells carry defective copies of HIV DNA and are thus not able to produce infectious and replication-competent viral particles. The remaining infected cells (11.7%) carry intact viral DNA and are potentially able to be induced (hence referred to as the inducible viral reservoir) and produce infectious particles, suggesting that the size of the viral reservoir is on the order of 60 latently infected per 1 million cells. Only a fraction of these inducible cells, however, is efficiently stimulated in vitro using the standard viral outgrowth assay. An additional round of cell stimulation allows inducing viral production from additional cells, but not all, suggesting that activation conditions, time, and stochasticity play a role in successful induction of viral production from latently infected cells [15]. Similar observations were reported upon stimulation of resting CD4+ T cells using CD3/CD28 antibodies for 7 days in the presence of efavirenz to block viral replication and measuring cell-associated and supernatant viral RNA by reverse transcription followed by quantitative polymerase chain reaction (RT-qPCR) [16▪▪,17▪▪]. Under these conditions, cell-associated viral RNA was detected in 7.5% of latently infected cells, while only 1.5% of cells were able to be induced and produce viral particles (as detected by viral RNA in the supernatant), confirming previous observations [15]. Of note, spontaneous virion production in the absence of activation was detected in 0.041% of latently infected cells.

These recent findings highlight the current gap between the experimentally induced reservoir and the total inducible reservoir, and demonstrate that the type and time of stimulation affect the size of the induced reservoir. This underlines the need for new experiments that investigate the stimulation dynamics leading to maximal viral reactivation, so that the total inducible reservoir is de facto induced.

Back to Top | Article Outline


Numerous modeling approaches have been applied to facilitate understanding of various aspects of HIV biology [18–20]. Here, we consider two major categories of modeling approaches applied to the study of the dynamics of viral replication and latency: mechanistic models that incorporate detailed biological knowledge into systems of differential equations in order to explain the dynamic behavior of biological systems, and descriptive statistical models, on the basis of large datasets obtained from genome-wide transcriptomic and proteomic measurements that explore patterns of similarity.

Back to Top | Article Outline

Mechanistic models

Mathematical analysis of HIV infection has traditionally been performed using mechanistic models. Early pioneering mechanistic models provided insight into the pathogenesis and treatment of HIV infection. These models present a bottom-up approach to mechanistically describe complex kinetic patterns observed in HIV infection. Prominent examples include the complex dynamics of viremia during the course of untreated HIV infection and the multiphasic decline of viral load under effective treatment [18]. Early models of HIV infection estimated the average half-life of productively HIV-infected cells to be approximately 1 day. A recent study by Petravic et al. suggests that this average life span poorly represents reality as individual infected cells may die within a few hours to a few days. The authors [21] observed that the rate of infected cell death decreased over time and was not correlated with viral protein production. This observation may impact the so-called ‘shock and kill’ strategy that aims at reactivating viral protein expression in latently infected cells, thereby mediating virus-induced cytotoxicity to kill the infected cells [22–25].

Mechanistic models have previously been developed to provide insight into the long-term dynamics of viral latency, including degradation of the virus, viral blips in virologically suppressed patients, reactivation of quiescent infection, and time needed to eliminate HIV under ART [19]. Immonen and Leitner used a joint phylogenetic and differential equation approach to model the evolutionary divergence of the virus taken from plasma and the latently infected cells from 26 patients. They [26] observed an overdispersion of evolutionary divergence relative to the molecular clock model, suggesting that a major fraction of infected cells have experienced periods of latency at some point in the past. The stability of the latent reservoir and the emergence of viral blips could be explained by stochastic expression of HIV [27,28▪]. Hill et al.[28▪] used a stochastic model to describe viral rebound after ART interruption and to define a quantitative goal for latency reactivation approaches. The authors conclude that a 10 000-fold decrease in the latent reservoir may be necessary to reach permanent viral remission in half of all individuals. Petravic et al.[29] devised a mathematical model to study the efficacy of antilatency drugs under different strategies and suggested that antilatency treatment should be administered early upon initiation of ART to achieve optimal outcomes. A recent study by Althaus et al. used longitudinal data from five chronically infected HIV patients to model the dynamics of different types of HIV-infected cells defined by the splicing patterns of the viral transcripts. In line with the study of Petravic et al., their model suggests that the reservoir is smaller during acute infection; thus, eradication strategies should be started early on in combination with ART [30].

Back to Top | Article Outline

Statistical models

New transcriptomics and proteomics technologies have enabled the collection of large-scale snapshots of the cellular state that can provide a holistic view of the cellular changes occurring upon viral infection at the cell population level. Repeated high-throughput measurements carried out in longitudinal experiments make allowance for a top-down analysis of the dynamics of the cell. The information provided by such time series experiments facilitates distinguishing causes and consequences observed upon viral infection, viral latency, and reactivation from latency. Permissive T-cell lines have been used to investigate transcriptional reprogramming of the host cell [31–34]. Mohammadi et al.[34] analyzed the joint virus–host transcriptome upon HIV infection with high temporal resolution over a 24 h period. A total of 73% of the expressed genes in the host cells were found to be regulated in concordance with the major viral replication steps, namely, reverse transcription, integration, and a late phase that spans from transcription to the release of new viral particles. The longitudinal design of the experiment and the mathematical analysis of the paired measurements of the viral life cycle intermediates and host transcriptome showed that the early regulated genes were likely due to response to the incoming virus. Mohammadi et al. observed a massive early downregulation probably reflecting a host cell response to viral presence. In contrast, late regulated genes are more likely to be regulated by newly produced viral proteins. These findings are consistent with a recent study that investigated phosphorylation changes upon HIV exposure [35▪]. Using infection with X4-tropic virus in primary resting CD4+ T cells, the authors identified rapid changes of 239 phosphorylation sites from 175 genes, some as early as 1 min after exposure that may prepare the cell for successful viral replication. This finding is also in agreement with another proteomic study that found host-induced early posttranslational modifications of histones in response to the HIV infection occurring as early as 4 h after exposure [36]. Similar patterns of early regulation were observed in microRNAs (miRNAs) [32]. Peng et al.[37] performed a joint analysis of transcriptome by mRNA sequencing and total RNA sequencing at 12 and 24 h after infection in a T-cell model. Their results suggest that a total RNA sequencing assay, quantifying also nascent and nonmature transcripts, may detect transcriptome changes earlier than mRNA sequencing.

Mohammadi et al.[12▪▪] investigated the transcriptome of the infected cells establishing and exiting latency. Resting CD4+ T cells were activated, infected with an HIV-based vector, and allowed to revert to a resting cellular state during 10 weeks using a feeder cell layer. Cells were then stimulated with various latency reactivating agents or with CD3/CD28/IL-2 for 8 or 24 h. This study demonstrated stable persistence of viral transcripts in latently infected cells over time, suggesting that in this system, the latently infected cells failed to produce viral particles because of posttranscriptional rather than transcriptional blocks. The analysis highlighted the biological state of the host cell, i.e. resting versus activated, as a major driver of differences between latent and productive infection [12▪▪,38▪]. Additional time series analyses focusing on the reactivation of cells will help to identify the key determinants driving the cascade of regulatory events leading to successful induction of viral production in latently infected cells.

Back to Top | Article Outline


Mathematical models have proven beneficial to make long-term predictions about the viral and cellular behaviors. They help defining strategies aiming at viral eradication. These predictions include estimation of the ART duration required to eliminate the viral reservoir and the critical size of the viral reservoir for sustained viral remission.

Large-scale temporal analyses have been used to describe HIV replication at the cell population level. In the future, integrative analyses involving collections of paired transcriptomic and proteomic datasets should produce a more comprehensive picture of the sequence of events occurring upon cell exposure to the virus, latency establishment, and reactivation from latency. These analyses may benefit from recent technological developments, such as single-cell technologies to measure the state of multiple individual cells. Single-cell approaches should allow a better assessment of stochastic expression of latent HIV, the impact of latency reactivating agents, and the optimal conditions (treatment type and exposure time) required to induce viral production from different types of latently infected cells. Repeated analysis of the transcriptome or proteome of a single cell over time is currently not feasible. Combination of population and single-cell analysis over time, however, might help identifying and dissecting the multiple types of latent cells.

Back to Top | Article Outline



Back to Top | Article Outline

Financial support and sponsorship

The present work was supported by the Swiss National Science Foundation (grant 31003A_146579;

Back to Top | Article Outline

Conflicts of interest

There are no conflicts of interest.

Back to Top | Article Outline


Papers of particular interest, published within the annual period of review, have been highlighted as:

  • ▪ of special interest
  • ▪▪ of outstanding interest
Back to Top | Article Outline


1. Durand CM, Blankson JN, Siliciano RF. Developing strategies for HIV-1 eradication. Trends in immunology 2012; 33:554–562.
2▪. Whitney JB, Hill AL, Sanisetty S, et al. Rapid seeding of the viral reservoir prior to SIV viraemia in rhesus monkeys. Nature 2014; 512:74–77.

This study shows that the viral reservoir is established very early in an animal model.

3. Deng K, Siliciano RF. HIV: early treatment may not be early enough. Nature 2014; 512:35–36.
4. Siliciano JD, Siliciano RF. AIDS/HIV rekindled HIV infection. Science 2014; 345:1005–1006.
5. Fletcher CV, Staskus K, Wietgrefe SW, et al. Persistent HIV-1 replication is associated with lower antiretroviral drug concentrations in lymphatic tissues. Proc Natl Acad Sci U S A 2014; 111:2307–2312.
6. Cory TJ, Schacker TW, Stevenson M, Fletcher CV. Overcoming pharmacologic sanctuaries. Curr Opin HIV AIDS 2013; 8:190–195.
7. Pomerantz RJ. Reservoirs, sanctuaries, and residual disease: the hiding spots of HIV-1. HIV Clin Trials 2003; 4:137–143.
8. Palmer S, Josefsson L, Coffin JM. HIV reservoirs and the possibility of a cure for HIV infection. J Intern Med 2011; 270:550–560.
9. Chomont N, El-Far M, Ancuta P, et al. HIV reservoir size and persistence are driven by T cell survival and homeostatic proliferation. Nat Med 2009; 15:893–900.
10. Buzon MJ, Sun H, Li C, et al. HIV-1 persistence in CD4+ T cells with stem cell-like properties. Nat Med 2014; 20:139–142.
11▪. Spina CA, Anderson J, Archin NM, et al. An in-depth comparison of latent HIV-1 reactivation in multiple cell model systems and resting CD4+ T cells from aviremic patients. PLoS Pathog 2013; 9:e1003834.

This study reveals the multifactorial nature of latently infected cells, being reactivated differently depending on the latency reactivating agent.

12▪▪. Mohammadi P, di Iulio J, Munoz M, et al. Dynamics of HIV latency and reactivation in a primary CD4+ T cell model. PLoS Pathog 2014; 10:e1004156.

The first temporal analysis of the joint viral and cellular transcriptome in a latency model. This study highlights posttranscriptional block and cellular activation status as components of latency.

13▪. Eriksson S, Graf EH, Dahl V, et al. Comparative analysis of measures of viral reservoirs in HIV-1 eradication studies. PLoS Pathog 2013; 9:e1003174.

This study compares the different methods currently used to assess the size of the reservoir.

14. Laird GM, Eisele EE, Rabi SA, et al. Rapid quantification of the latent reservoir for HIV-1 using a viral outgrowth assay. PLoS Pathog 2013; 9:e1003398.
15. Ho YC, Shan L, Hosmane NN, et al. Replication-competent noninduced proviruses in the latent reservoir increase barrier to HIV-1 cure. Cell 2013; 155:540–551.
16▪▪. Cillo AR, Sobolewski MD, Bosch RJ, et al. Quantification of HIV-1 latency reversal in resting CD4+ T cells from patients on suppressive antiretroviral therapy. Proc Natl Acad Sci U S A 2014; 111:7078–7083.

This study shows that SAHA, a latency reactivating agent, fails to induce viral particle production.

17▪▪. Ho YC, Laird GM, Siliciano RF. Measuring reversal of HIV-1 latency ex vivo using cells from infected individuals. Proc Natl Acad Sci U S A 2014; 111:6860–6861.

This study highlights the discrepancy among the defective viral reservoir, the inducible viral reservoir, and the induced viral reservoir.

18. Perelson AS, Ribeiro RM. Modeling the within-host dynamics of HIV infection. BMC Biol 2013; 11:96.
19. Selinger C, Katze MG. Mathematical models of viral latency. Curr Opin Virol 2013; 3:402–407.
20. Wodarz D. Mathematical models of HIV replication and pathogenesis. Methods Mol Biol 2014; 1184:563–581.
21. Petravic J, Ellenberg P, Chan ML, et al. Intracellular dynamics of HIV infection. J Virol 2014; 88:1113–1124.
22. Katlama C, Deeks SG, Autran B, et al. Barriers to a cure for HIV: new ways to target and eradicate HIV-1 reservoirs. Lancet 2013; 381:2109–2117.
23. Barouch DH, Deeks SG. Immunologic strategies for HIV-1 remission and eradication. Science 2014; 345:169–174.
24. Archin NM, Margolis DM. Emerging strategies to deplete the HIV reservoir. Curr Opin Infect Dis 2014; 27:29–35.
25. Deeks SG. HIV: shock and kill. Nature 2012; 487:439–440.
26. Immonen TT, Leitner T. Reduced evolutionary rates in HIV-1 reveal extensive latency periods among replicating lineages. Retrovirology 2014; 11:81.
27. Wang S, Rong L. Stochastic population switch may explain the latent reservoir stability and intermittent viral blips in HIV patients on suppressive therapy. J Theor Biol 2014; 360:137–148.
28▪. Hill AL, Rosenbloom DI, Fu F, et al. Predicting the outcomes of treatment to eradicate the latent reservoir for HIV-1. Proc Natl Acad Sci U S A 2014; 111:13475–13480.

This study sets a quantitative goal for the ‘shock and kill’ strategies.

29. Petravic J, Martyushev A, Reece JC, et al. Modeling the timing of antilatency drug administration during HIV treatment. J Virol 2014; 88:14050–14056.
30. Althaus CL, Joos B, Perelson AS, Gunthard HF. Quantifying the turnover of transcriptional subclasses of HIV-1-infected cells. PLoS Comput Biol 2014; 10:e1003871.
31. Chang ST, Sova P, Peng X, et al. Next-generation sequencing reveals HIV-1-mediated suppression of T cell activation and RNA processing and regulation of noncoding RNA expression in a CD4+ T cell line. MBio 2011; 2:
32. Chang ST, Thomas MJ, Sova P, et al. Next-generation sequencing of small RNAs from HIV-infected cells identifies phased microrna expression patterns and candidate novel microRNAs differentially expressed upon infection. MBio 2013; 4:e00549–e00612.
33. Lefebvre G, Desfarges S, Uyttebroeck F, et al. Analysis of HIV-1 expression level and sense of transcription by high-throughput sequencing of the infected cell. J Virol 2011; 85:6205–6211.
34. Mohammadi P, Desfarges S, Bartha I, et al. 24 h in the life of HIV-1 in a T cell line. PLoS Pathog 2013; 9:e1003161.
35▪. Wojcechowskyj JA, Didigu CA, Lee JY, et al. Quantitative phosphoproteomics reveals extensive cellular reprogramming during HIV-1 entry. Cell Host Microbe 2013; 13:613–623.

This study is the first phosphoproteomic study in a natural infection design, i.e. X4-tropic HIV on primary CD4+ T cells.

36. Britton LM, Sova P, Belisle S, et al. A proteomic glimpse into the initial global epigenetic changes during HIV infection. Proteomics 2014; 14:2226–2230.
37. Peng X, Sova P, Green RR, et al. Deep sequencing of HIV-infected cells: insights into nascent transcription and host-directed therapy. J Virol 2014; 88:8768–8782.
38▪. Bullen CK, Laird GM, Durand CM, et al. New ex vivo approaches distinguish effective and ineffective single agents for reversing HIV-1 latency in vivo. Nat Med 2014; 20:425–429.

cure; dynamic model; HIV; latency; transcriptome dynamics

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.