Last in first out: SIV proviruses seeded later in infection are harbored in short-lived CD4+ T cells

HIV can persist in a latent form as integrated DNA (provirus) in resting CD4+ T cells of infected individuals and as such is unaffected by antiretroviral therapy (ART). Despite being a major obstacle for eradication efforts, the genetic variation and timing of formation of this latent reservoir remains poorly understood. Previous studies on when virus is deposited in the latent reservoir have come to contradictory conclusions. To reexamine the genetic variation of HIV in CD4+ T cells during ART, we determined the divergence in envelope sequences collected from 10 SIV infected rhesus macaques. We found that the macaques displayed a biphasic decline of the viral divergence over time, where the first phase lasted for an average of 11.6 weeks (range 4–28 weeks). Motivated by recent observations that the HIV-infected CD4+ T cell population is composed of short- and long-lived subsets, we developed a model to study the divergence dynamics. We found that SIV in short-lived cells was on average more diverged, while long-lived cells harbored less diverged virus. This suggests that the long-lived cells harbor virus deposited starting earlier in infection and continuing throughout infection, while short-lived cells predominantly harbor more recent virus. As these cell populations decayed, the overall proviral divergence decline matched that observed in the empirical data. This model explains previous seemingly contradictory results on the timing of virus deposition into the latent reservoir, and should provide guidance for future eradication efforts.

Row 4 shows the divergence dynamics from start of ART to ≈ 150 weeks for fast-decaying cells (orange), slow-decaying cells (black) and total population (blue).Divergence dynamics were computed analytically.The only cases when divergence declined in a biphasic manner was when the mean divergence of the fast-decaying population was higher than that of the slow-decaying population at start of ART.Row 4 shows the divergence dynamics from start of ART to ≈ 150 weeks for fast-decaying cells (orange), slow-decaying cells (black) and total population (blue).Divergence dynamics were computed analytically.The only cases when divergence declined in a biphasic manner was when the mean divergence of the fast-decaying population was higher than that of the slow-decaying population at start of ART.Row 4 shows the divergence dynamics from start of ART to ≈ 150 weeks for fast-decaying cells (orange), slow-decaying cells (black) and total population (blue).Divergence dynamics were computed analytically.The only cases when divergence declined in a biphasic manner was when the mean divergence of the fast-decaying population was higher than that of the slow-decaying population at start of ART.Row 4 shows the divergence dynamics from start of ART to ≈ 150 weeks for fast-decaying cells (orange), slow-decaying cells (black) and total population (blue).Divergence dynamics were computed analytically.The only cases when divergence declined in a biphasic manner was when the mean divergence of the fast-decaying population was higher than that of the slow-decaying population at start of ART.Row 4 shows the divergence dynamics from start of ART to ≈ 150 weeks for fast-decaying cells (orange), slow-decaying cells (black) and total population (blue).Divergence dynamics were computed analytically.The only cases when divergence declined in a biphasic manner was when the mean divergence of the fast-decaying population was higher than that of the slow-decaying population at start of ART.S1 and S2.S1 and S2.dF and dS are the divergence in fast-and slow-decaying cells at start of ART; τF mixed effects and τS mixed effects are the fixed effect half-lives of fast-and slow-decaying cells corrected by the random effect for each macaque, respectively, estimated by pooling the data of all 10 macaques; τF individual fit and τS individual fit are the the half-lives of fast-and slow-decaying cells, respectively, estimated separately for each macaque; f(0) is the frequency of fast-decaying cells at start of ART.

Fig. S1 .
Fig. S1.Divergence dynamics with theoretical distributions.Here the fast-decaying population is dominated by more diverged sequences, with the divergence histogram having a linearly increasing shape.The top 3 rows show the divergence at start of ART of fast-decaying populations (row 1, orange), slow-decaying populations (row 2, black), and the total population (row 3).

Fig. S2 .
Fig. S2.Divergence dynamics with theoretical distributions.Here the fast-decaying population is dominated by less diverged sequences, with the divergence histogram having a linearly decreasing shape.The top 3 rows show the divergence at start of ART of fast-decaying populations (row 1, orange), slow-decaying populations (row 2, black), and the total population (row 3).

Fig. S3 .
Fig. S3.Divergence dynamics with theoretical distributions.Here the fast-decaying population has an equal representation of different divergence values, with the divergence histogram having a uniform shape.The top 3 rows show the divergence at start of ART of fast-decaying populations (row 1, orange), slow-decaying populations (row 2, black), and the total population (row 3).

Fig. S4 .
Fig. S4.Divergence dynamics with theoretical distributions.Here the fast-decaying population is dominated by more diverged sequences, with the divergence histogram having an exponentially increasing shape.The top 3 rows show the divergence at start of ART of fast-decaying populations (row 1, orange), slow-decaying populations (row 2, black), and the total population (row 3).

Fig. S5 .
Fig. S5.Divergence dynamics with theoretical distributions.Here the fast-decaying population is dominated by less diverged sequences, with the divergence histogram having an exponentially decreasing shape.The top 3 rows show the divergence at start of ART of fast-decaying populations (row 1, orange), slow-decaying populations (row 2, black), and the total population (row 3).

Fig. S6 .
Fig. S6.Histograms of divergence of fast-(orange) and slow-decaying (grey) populations using plasma viral RNA sequences at start of ART in 9 SIV-infected rhesus macaques.Plasma viral RNA sequences could not be collected for macaque T624.Orange and grey curves show the smoothed-out histograms.Solid vertical lines correspond to the mean divergence of fast-(orange) and slow-decaying (grey) populations.Vertical green dashed lines show the median divergence of the total population.

Fig. S7 .
Fig. S7.Empirically observed mean divergence dynamics (red lines) compared to model simulated divergence.Divergence dynamics using half-lives estimated by pooling the data of all macaques are shown by blue lines.Divergence dynamics using half-lives learnt separately for each individual macaque are shown by green lines.Corresponding bold lines show means of 10 3 simulations.Thin green/blue lines show individual stochastic simulations of divergence obtained by sampling according to experimental sampling times and number of sequences.Turquoise is the result of the blue and green lines overlapping.For macaque T625, green lines are absent as the data was insufficient to estimate half-life individually.The red tick marks on the x-axis show the times when experimental samples were obtained.The x-axis for the histograms is shown at the top of the figure, while the x-axis for the divergence dynamics plots is shown at the bottom of the figure.Simulations largely track observed change in divergence even when plasma viral RNA is used to compute divergence at start of ART.

Fig. S8 .
Fig. S8.Responsiveness to parameters was estimated by the normalized mathematical elasticity, P , for each individual macaque (a) or human subject (b), shown by separate lines.f(0) is the initial frequency of fast-decaying cells; τF and τS are the the half-lives of fast-and slow-decaying CD4 + T cells, respectively, estimated by first pooling the data of all 10 macaques and then adjusting for individual variations; dF and dS are the initial divergence in fast-and slow-decaying cells, respectively.Details in Supplementary Methods, parameter values in Supplementary TablesS1 and S2.

Fig. S9 .
Fig. S9.Responsiveness to parameters was estimated by the normalized mathematical elasticity, P , for each individual macaque (a) or human (b), shown by separate lines.f(0) is the initial frequency of fast-decaying cells; τF and τS are the the half-lives of fast-and slow-decaying cells, respectively, estimated separately for each macaque (estimation did not converge for macaques T624 and T625); dF and dS are the SIV divergence in fast-and slow-decaying cells.Details in Supplementary Methods, parameter values in Supplementary TablesS1 and S2.

Table S2 . Statistics for divergence distributions at start of ART for 10 macaques using intact proviral DNA.
reported here compares the divergence of the fast-and slow-decaying sequences per monkey at start of ART using the paired Wilcoxon exact test.The null hypothesis is the two samples come from the same distribution.A p-value < 0.05 indicates that the fast-and slow-decaying populations indeed have different distributions.Shift is computed as fast (slow) shift = fast (slow) mean -total median.

Table S3 . Statistics for divergence distributions at start of ART for 10 macaques using plasma viral RNA at timepoint 0, and intact proviral DNA thereafter.
reported here compares the divergence of the fast-and slow-decaying sequences per monkey at start of ART using the paired Wilcoxon exact test.The null hypothesis is the two samples come from the same distribution.A p-value < 0.05 indicates that the fast-and slow-decaying populations indeed have different distributions.Shift is computed as fast (slow) shift = fast (slow) mean -total median.