Dynamics of immunoglobulin sequence diversity in HIV-1 infected individuals

Abstract

Advances in immunoglobulin (Ig) sequencing technology are leading to new perspectives on immune system dynamics. Much research in this nascent field has focused on resolving immune responses to viral infection. However, the dynamics of B-cell diversity in early HIV infection, and in response to anti-retroviral therapy, are still poorly understood. Here, we investigate these dynamics through bulk Ig sequencing of samples collected over 2 years from a group of eight HIV-1 infected patients, five of whom received anti-retroviral therapy during the first half of the study period. We applied previously published methods for visualizing and quantifying B-cell sequence diversity, including the Gini index, and compared their efficacy to alternative measures. While we found significantly greater clonal structure in HIV-infected patients versus healthy controls, within HIV patients, we observed no significant relationships between statistics of B-cell clonal expansion and clinical variables such as viral load and CD4(+) count. Although there are many potential explanations for this, we suggest that important factors include poor sampling resolution and complex B-cell dynamics that are difficult to summarize using simple summary statistics. Importantly, we find a significant association between observed Gini indices and sequencing read depth, and we conclude that more robust analytical methods and a closer integration of experimental and theoretical work is needed to further our understanding of B-cell repertoire diversity during viral infection.

Keywords: B-cell receptor; Gini index; diversity.

Figure 1.

Network visualization of the diversity of BCR sequences obtained from untreated patient 3 at week 4 (a) and week 108 (b), and from treated patient 4 at week 4 (c) and week 108 (d). Each vertex represents a unique sequence, and the size of each vertex is proportional to the number of reads identical to that sequence. Edges are drawn between vertices that are one nucleotide substitution apart. See §2(f) for further details.

Figure 2.

BCR diversity statistics and viral load values for patients 1–4 (a–d) over the study period of 108 weeks. Patients 1–3 were untreated, while patient 4 received ART until week 48. Four plots are provided in each panel. The top plot shows viral load, with each point representing a clinical sample. The second plot shows vertex Gini index values of the BCR sequences obtained at each time point. The third plot shows the proportion of reads at each time point that belong to ‘large’ clones, i.e. those that occupy more than 0.1% of reads at any time point. The bottom plot in each panel shows the proportion of all reads at each time point that are occupied by the 20 largest clones observed across all time points. Each of the 20 largest clones is represented by a bar of a different colour, and lines connect bars at adjacent time points that represent the same clone.

Figure 3.

BCR diversity statistics and viral load values for patients 5–8 (a–d) over the study period of 108 weeks. All patients received ART until week 48. See figure 2 legend for details.

Figure 4.

Comparison of three BCR summary statistics among healthy uninfected individuals (blue) and HIV-infected patients. Values from HIV+ patients are calculated separately for weeks 4–24 (HIV+ early; green) and weeks 52–108 (HIV+ late; black). For each period, statistics were averaged using linear interpolation (see §2f for details). Each box and solid line represents the first, third and second quartiles, respectively, while whiskers indicate range. Data points are superimposed on the box and whisker plots.