Increased ex vivo cell death of central memory CD4 T cells in treated HIV infected individuals with unsatisfactory immune recovery

Subject characteristics

Subjects recruited in this study were stratified according to the level of recovery
of CD4 T cells after suppressive HAART, as immunodiscordant or immunoconcordant (cutoff
value 350 CD4 T cells/?L). The time course of CD4 T-cell recovery for each group is
shown in Figure 1a. At the time of sample analysis, the median CD4 T-cell count was 220 and 798 cells/?L
for immunodiscordant and immunoconcordant HIV-infected individuals respectively (p  0.0001), reflecting the blunted dynamics of CD4 T-cell recovery of the former group.
Absolute CD8 T-cell counts showed no differences between HIV-infected groups. Median
length of HIV infection (from diagnosis) was 11.8 and 10.1 years (p = ns), while median time on treatment was 11.2 and 5.2 years (p = ns) for immunoconcordant or immunodiscordant subjects, respectively. Overall, immunodiscordant
individuals showed the previously reported higher CD4 T-cell death and activation
with significantly lower nadir CD4 T-cell counts 8], 17]. However, some immunoconcordant individuals showed also low nadir values. To evaluate
the effect of nadir on immune recovery, immunoconcordant individuals were further
divided according to the median nadir CD4 T-cell value (cut-off value 250 CD4 T cells/?L)
into two subgroups with low and high nadir values (n = 17 ad n = 16, respectively).
This allowed the direct comparison of immunoconcordant individuals with immunodiscordant
subjects, avoiding the confounding effect of nadir and enabling us to evaluate the
impact of low nadir values in immunoconcordant individuals. No major differences were
observed among immunoconcordant subgroups, except for a higher presence of PI-based
treatments and HCV co-infection in the low nadir group (Table 1).

Figure 1. Characterization of CD4 T-cell subsets. a shows the evolution of CD4 T-cell counts in the three groups defined in this study
(Immunodiscordant in red (n = 23), immunoconcordant with low and high nadir in dark and light green, m = 17 and n = 16, respectively). Dots correspond to individual determinations of CD4 T-cell counts and lines show non-lineal regression of data plotted for comparative purposes. b The absolute count of circulating and T
_N
, T
_CM
, T
_TM
, T
_EM
and T
_TD
CD4 T cells was analyzed in immunodiscordant individuals (red boxes), immunoconcordant individuals with low or high nadir (LN and HN; dark and light green boxes, respectively) and healthy HIV uninfected individuals (n = 11, blue boxes). c The frequencies of the above-mentioned subsets in the CD4 T-cell compartment were
also analyzed. In panels b and c, data shown correspond to median values (band), IQR (boxes) and 10–90 interquartile values (whiskers). Figure show significant p values (permutation test adjusted by false discovery rate): *p  0.05; **p  0.01 and ***p  0.005.

Table 1. Main characteristics of the different groups

Analysis of the CD4 T-cell maturation

Absolute counts and frequency of different CD4 T-cell subsets were analyzed in immunodiscordant,
immunoconcordant (low and high-nadir subgroups) and 11 uninfected individuals. The
data show that lower CD4 T-cell counts in immunodiscordant subjects (Figure 1a) were the consequence of lower levels of T
_N
, T
_CM
, T
_TM
and T
_EM
cells compared with immunoconcordant individuals, while the absolute numbers of TTD
cells were similar in all groups (Figure 1b). Interestingly, immunoconcordant patients, irrespective of the nadir values showed
similar counts of all subsets that were in turn comparable to uninfected controls
except for the T
_EM
subset, suggesting a proper recovery of the CD4 T-cell subsets in these individuals
(Figure 1b). The frequency of each subset in the CD4 T-cell compartment showed a significant
underrepresentation of TN cells in immunodiscordant subjects (as compared to concordant
or HIV-uninfected individuals) that was compensated by an overrepresentation of T
_TM
cells and a less evident but still significant increase in T
_EM
and T
_TD
cells (Figure 1c). Conversely, T
_CM
cells showed similar values in all groups. Again, both subgroups of immunoconcordant
subjects showed similar values of subset frequencies reaching the levels of HIV-uninfected
controls (Figure 1c).

CD4 T-cell maturation and CD4 T-cell death

In our previous studies we have shown that CD4 T-cell death, in particular intrinsic
apoptosis, is a major determinant of immune recovery 8], 17]. Therefore, we explored the association of unbalanced CD4 T-cell maturation with
the rate of cell death in ex vivo cultures of fresh PBMC. Spontaneous CD4 T-cell death
was unrelated to the frequency of CD4 T
_CM
or T
_TD
cells but showed a significant negative correlation with the frequency of CD4 TN and
positive correlation with T
_TM
and T
_EM
cells (Figure 2a). Since the frequency of CD4 T
_N
and T
_TM
cells were strongly inversely correlated (data not shown), we addressed independent
associations by using a model including data from all subsets. This model (Additional
file 2: Table S1) confirmed the independent positive association of CD4 T-cell death with
the frequency of T
_TM
CD4 T cells, clearly linking the higher presence of these cells with the increased
cell death observed in immunodiscordant individuals.

Figure 2. Association of CD4 T-cell maturation with CD4 T-cell death. a Relationships between the frequencies of the different CD4 T-cell subsets was plotted
against spontaneous intrinsic CD4 T-cell apoptosis. Data from immunodiscordant (n = 23,
red), low nadir immunoconcordant (n = 17, dark green) or high nadir immunoconcordant individuals (n = 16, light green) are shown with color-coded linear regression for each data set. Linear regression
for the global data is shown by black lines. Correlation coefficients and p values of Spearman’s test for the global analysis are shown in each plot. b Spontaneous cell death was assessed in sorted T
_N
, T
_CM
, T
_TM
, T
_EM
/T
_TD
CD4 T cells stained with the potentiometric probe DIOC(6). Dot plots of DIOC(6) and
CD3 staining for a representative individual show the percentage of dead cells in
the left (DIOC low) gate. c The level of spontaneous cell death in sorted T
_N
, T
_CM
, T
_TM
, T
_EM+TD
T cells from immunoconcordant (green, n = 5) or immunodiscordant (red, n = 5) treated HIV infected individuals are shown. Figure shows median values (bands), IQR (boxes) and 10-90 interquartile values (whiskers). Asterisks denote significant differences (non parametric permutation or Mann–Whitney tests).
d Correlations of cell death sorted T
_N
, T
_CM
, T
_TM
and T
_EM
/T
_TD
CD4 T cells with absolute counts of circulating CD4 T cells. Correlation coefficient
and p values (Spearman) are shown in each graph.

However, these data do not discriminate whether TTM cell death rate is similar among
treated HIV infected individuals or is higher in immunodiscordant patients. To experimentally
confirm these possibilities, we addressed the analysis of the lifespan of T
_N
, T
_CM
, T
_TM
and T
_EM+TD
CD4 T-cells from immunoconcordant and immunodiscordant individuals. Since PBMC culture
did not allow for a direct assessment of cell death occurring in different CD4 T-cell
subsets due to phenotypic changes upon cell death 21], we chose a previously reported sorting strategy prior to ex vivo culture to approach
this issue 21]. Although this strategy may induce some additional stress to sorted cells, it allowed
for a consistent measure of subset cell death (Figure 2b). In this set of experiments, we observed that TN cells showed significantly lower
cell death rates than those observed in purified memory cells in both groups (Figure 2c). No significant difference in TN cell death was observed between groups. However,
memory cells in immunodiscordant individuals showed higher rates of death (Figure 2c). This was associated with higher sensitivity to cell death in TCM cells from immunodiscordant
individuals (p = 0.056), and non-significant trends in TTM and TEM
₊
TD cells. Consistently, T
_N
cell death did not correlate with CD4 T-cell counts while a significant negative association
(p = 0.013) was observed for T
_CM
and trends were noticed for T
_TM
and T
_EM+TD
cell death (p = 0.067 and 0.054, respectively; Figure 2d). No significant correlations with senescence markers could be identified (data
not shown).

To evaluate the impact of higher cell death rates on transition from T
_CM
to more advanced stages of CD4 T-cell maturation, we analyzed the ratio between T
_CM
and T
_TM
cells. This parameter was significantly lower in immunodiscordant subjects compared
to immunoconcordant (irrespective of nadir values) or uninfected individuals (Figure 3a). Interestingly, T
_CM
to T
_TM
transition is not completely normalized in immunoconcordant individuals (Figure 3a), suggesting that this parameter could be useful to assess the quality of immune
recovery. Reinforcing the relevance of the T
_CM
/T
_TM
ratio, a strong negative correlation was observed between this parameter and spontaneous
CD4 T-cell death (Figure 3b). Overall, our data suggest that memory, but not T
_N
cells, are major contributors to the increased cell death observed in immunodiscordant
individuals 17]. Higher sensitivity to death of memory cell populations, especially T
_CM
, may help to explain the inability of immunodiscordant individuals to recover proper
T
_CM
/T
_TM
ratios.

Figure 3. Ratio CD4 TCM/TTM and its association with total CD4 T-cell death. a The ratio between T
_CM
and T
_TM
cells was calculated as a measure of CD4 T-cell differentiation. Data shown correspond
to median values (bands), IQR (boxes) and 10–90 interquartile values (whiskers) for immunodiscordant individuals (n = 23, red boxes), immunoconcordant individuals with low or high nadir (n = 17 and n = 16, dark and light green boxes, respectively) and healthy HIV uninfected individuals (n = 11, blue boxes). Figure shows significant p values (permutation test adjusted by false discovery rate). b Correlation between total CD4 T-cell death and the ratio of CD4 T
_CM
/T
_TM
cells. P values for Spearman’s test correlation are shown in black for all data points, and in red or green for immunodiscordant and immunoconcordant individuals respectively.

CD57 expression and immune recovery

Besides T
_CM
or T
_TM
dysfunction, replicative immunosenescence of CD4 T cells has also been related to
immune recovery 14]. In our cohort, analysis of the frequency of CD28
^?
CD57
⁺
CD4 T cells (Additional file 1: Figure S1) showed higher levels in immunodiscordant subjects than in both immunoconcordant
subgroups, which in turn showed similar levels to HIV-uninfected individuals (Figure 4a). The analysis of CD57 expression in the different CD4 T-cell subsets showed that
despite apparent full immune recovery, immunoconcordant individuals displayed higher
CD57 expression than uninfected controls in all CD4 T-cell subsets. In addition, immunodiscordant
individuals showed highly significant differences with concordant subjects particularly
in T
_N
and T
_CM
cells (Figure 4b).

Figure 4. Characterization of CD57 expression in CD4 T cells. a The replicative senescence of the whole CD4 T-cell compartment was defined by the
frequency of CD57
⁺
CD28
^?
cells. b In addition, the expression of CD57
⁺
cells was evaluated in the different CD4 T cell subsets by analyzing the frequency
of CD57
⁺
cells. c The relationship between levels of spontaneous CD4 T-cell death and replicative immunosenescence
is shown. Plots shows data from immunodiscordant (n = 23, red), low nadir immunoconcordant (n = 17, dark green) or high nadir immunoconcordant individuals (n = 16, light green) with color-coded linear regression for each data set. Linear regression for the
global data is shown in black lines. Correlation coefficients and p values of Spearman’s test for the global analysis are shown. a, b also show data from HIV uninfected individuals (n = 11, blue bars).

We also evaluated the relationship of CD57 expression with CD4 T-cell survival. No
significant correlation was found between replicative senescence (CD28
^?
CD57
⁺
 cells) and spontaneous CD4 T-cell death (Figure 4c). These data suggest that different mechanisms regulate the expression of CD57 in
different CD4 T-cell subsets and show that replicative senescence is not coupled with
the spontaneous CD4 T-cell death rate characteristics of immunodiscordant responses
to HAART.

Analysis of the CD8 T-cell compartment

Major comorbidities in HIV infected individuals have been related to CD8 T-cell activation
1], 22] and recently to the CD4/CD8 T-cell ratio. Therefore, we also evaluated differentiation
and immunosenescence in CD8 T cells. Elevated CD8 T-cell counts in HIV infected individuals
analyzed in this study (Table 1) were the consequence of an expansion of all subsets except CD8 T
_CM
cells (Figure 5a). Comparison of immunodiscordant and immunoconcordant subjects exclusively showed
differences in the TN subset, consistent with the widely described thymic dysfunction
of these patients 6]. The percentage of CD8 T cells was not normalized in immunoconcordant individuals,
which showed intermediate values between immunodiscordant and uninfected subjects
(Table 1). Among immunoconcordant and immunodiscordant individuals, T
_TM
, T
_EM
and T
_TD
CD8 T-cell subsets showed similar frequencies. However, a decreased size of the CD8
T
_N
subset was observed in immunodiscordant individuals, while a reduced TCM subset was
only observed in immunoconcordant subjects (Figure 5b). Replicative immunosenescence (CD28
^?
CD57
⁺
cells) was increased in both groups of HIV infected individuals compared to uninfected
controls; however, no differences between immunoconcordant and immunodiscordant individuals
were observed. A deeper analysis of CD57 expression in the different CD8 T-cell subsets
commonly showed a lower expression of CD57 in uninfected donors compared to HIV infected
groups, with no differences between immunodiscordant and immunoconcordant individuals
(Figure 5c). Again, we observed similar CD57 expression when this latter group was subdivided
according to nadir values of CD4 T-cell counts. In summary, individuals displaying
different CD4 T-cell recovery profiles seem to have similar CD8 T-cell compartments,
except for the number of CD8 T
_N
cells.

Figure 5. Characterization of CD8 T-cell subsets. The absolute number of circulating and T
_N
, T
_CM
, T
_TM
, T
_EM
and T
_TD
CD8 T cells (a) was analyzed in immunodiscordant individuals (n = 23, red boxes), immunoconcordant individuals with low or high nadir (n = 17 and n = 16, dark and light green boxes, respectively) and healthy HIV uninfected individuals (n = 11, blue boxes). The frequency of the above-mentioned subsets in the CD8 T cell compartment (b) was also analyzed. The expression of CD57
⁺
cells was further evaluated in the different subsets by the frequency of CD57
⁺
cells (c). In all panels data shown correspond to median values (bands), IQR (boxes) and 10–90 interquartile values (whiskers) for immunodiscordant individuals (red boxes), immunoconcordant individuals with low or high nadir (green boxes) and healthy HIV uninfected individuals (blue boxes). Figure show p values (permutation test adjusted by false discovery rate).