Investigating the dynamics of proviral silencing in polyclonal HIV-1 infected Jurkat cell populations

1. Problem Statement

HIV-1 persists in a reservoir of transcriptionally silent but replication-competent proviruses, primarily within resting CD4+ T cells, despite antiretroviral therapy (ART). While clonal expansion of infected cells is known to contribute to reservoir stability, the mechanisms driving the long-term silencing of HIV-1 expression in polyclonal populations remain poorly understood. Previous studies on individual Jurkat cell clones (8–9 days) showed stable bimodal expression patterns (mixtures of active and inactive cells), yet polyclonal pools of these cells exhibit a gradual decline in overall HIV-1 expression over time. The central question addressed is whether this population-level silencing is driven by uniform transcriptional repression across all clones or by heterogeneous, clone-specific dynamics involving clonal expansion, attrition, and variable transcriptional stability.

2. Methodology

The authors employed a high-throughput barcoding and sequencing approach to track hundreds of individual proviral clones within a polyclonal population over extended periods.

• Viral Vector Design: A replication-deficient HIV-1 vector (NL4-3 based, lacking env, vpr, and nef) was engineered. It contained:
  • An eGFP reporter driven by the HIV-1 LTR to monitor transcriptional activity.
  • A puromycin resistance gene driven by a constitutive EF1α promoter to select for infected cells regardless of LTR activity.
  • A 20-bp randomized DNA barcode inserted into the U3 region of the 3' LTR. Upon reverse transcription, this barcode is duplicated to both LTRs, serving as a unique, heritable identifier for each integration event.
• Cell Culture: Jurkat T cells were infected at a low Multiplicity of Infection (MOI = 0.0005) to ensure single integration events. Two independent polyclonal pools were established:
  • Pool 22d: Cultured for 22 days.
  • Pool 90d: Cultured for 90 days.
  • Both pools were split into replicate cultures (A and B) 14 days post-infection.
• Experimental Workflow:
  1. Flow Cytometry: Periodic measurement of the percentage of GFP-positive cells (%GFP+) to track population-level silencing.
  2. Cell Sorting: At specific timepoints (t1, t22, t90), cells were sorted into GFP+ and GFP- fractions.
  3. Genomic DNA Isolation & Sequencing: DNA was extracted from sorted fractions and unsorted controls. High-throughput sequencing (Azenta Amplicon EZ) was used to quantify barcode abundance in GFP+ vs. GFP- fractions.
  4. Data Analysis: Clonal %GFP+ was calculated for each barcode. Clones with fractional abundance >0.001 were analyzed. Statistical correlations (Spearman) were used to compare replicates and reconstructed populations.
  5. Resorting Assay: To test the stability of expression states, sorted GFP+ and GFP- subpools were cultured separately for 22 days and resorted to observe phenotypic switching.

3. Key Contributions

• Longitudinal Clonal Tracking: The study provides one of the most detailed longitudinal analyses of HIV-1 expression dynamics in a polyclonal setting, tracking hundreds of clones over 90 days.
• Dissection of Silencing Mechanisms: It distinguishes between population-level silencing caused by the selective expansion of low-activity clones versus intrinsic transcriptional silencing of individual proviruses.
• Quantification of Stability: It quantifies the proportion of clones that maintain stable bimodal expression patterns over extended periods (22 and 90 days) despite stochastic switching at the single-cell level.

4. Key Results

A. Population-Level Silencing Dynamics

• Pool 22d: %GFP+ declined from ~64% (t1) to ~38% (t22).
• Pool 90d: %GFP+ declined from ~60% (t1) to ~18% (t90).
• This decline was not uniform; it resulted from a complex interplay of mechanisms rather than a global shutdown of transcription.

B. Mechanisms Driving Silencing

1. Selective Clonal Expansion (Early Phase): The initial decline in population GFP+ was primarily driven by the disproportionate expansion of a small subset of clones with low LTR activity. In Pool 22d, the top 5 clones by abundance shifted from t1 to t22, with lower-ranked, low-activity clones expanding to dominate the population.
2. Clonal Attrition and Intrinsic Silencing (Late Phase): In the 90-day pool, additional factors emerged:
   • Clonal Loss: ~78% of clones detected at t1 were "dropped" (fell below detection threshold) by t90. These were predominantly high-LTR-active clones.
   • Intrinsic Silencing: A subset of previously active clones showed a gradual reduction in their %GFP+ over time.

C. Stability of Bimodal Expression

• Short-term (22 days): Approximately 52% of detectable clones in Pool 22d maintained stable bimodal expression patterns (log change in %GFP+ < ±0.05).
• Long-term (90 days): Approximately 17% of clones in Pool 90d retained stable bimodal expression patterns throughout the experiment.
• Phenotypic Switching: Despite stable clonal proportions, individual cells within these stable clones continued to oscillate between GFP+ and GFP- states. When sorted GFP+ and GFP- subpools were cultured separately and resorted, the descendants did not remain purely in their parental state, confirming ongoing stochastic switching. However, the weighted average of the resorted populations closely matched the original unsorted population, proving that the clone-specific equilibrium is heritable.

D. Reproducibility

• Results were highly concordant between independent replicates (Pool A and B) and between two independently generated pools (22d and 90d), confirming that these dynamics are robust features of the system.

5. Significance

• Heterogeneity over Uniformity: The study demonstrates that HIV-1 population silencing is not a uniform process but emerges from heterogeneous, clone-specific behaviors. Some clones are inherently "fit" for expansion despite low expression, while others are lost or silenced over time.
• Reservoir Dynamics: The findings mirror in vivo observations where the HIV-1 reservoir is dominated by a few expanding clones. The model suggests that the reservoir's stability is maintained not just by latency, but by the fitness advantage of low-expression clones that expand without triggering immune clearance or cytopathic effects.
• Stochastic vs. Deterministic: The work reinforces that HIV-1 expression is governed by a combination of stochastic noise (single-cell switching) and deterministic, clone-intrinsic factors (integration site, chromatin environment) that set the equilibrium ratio of active to inactive cells.
• Therapeutic Implications: Understanding that the reservoir is dynamic, with clones expanding, contracting, or silencing at different rates, is crucial for "Shock and Kill" or "Block and Lock" strategies. It suggests that targeting only transcriptionally active cells may be insufficient if low-activity clones are the primary drivers of reservoir persistence through clonal expansion.

In conclusion, this paper provides a mechanistic framework for how HIV-1 populations evolve in culture, revealing that long-term silencing is a composite result of selective clonal expansion of low-activity variants, the loss of high-activity variants, and the maintenance of stable, heritable bimodal expression states in a subset of the population.