Impact of Early Antiretroviral Therapy Initiation on HIV-Specific CD4 and CD8 T Cell Function in Perinatally Infected Children

Mother-to-child transmission is the main cause of HIV infection in children. Implementation of guidelines for prevention of mother-to-child transmission with antiretroviral therapy (ART) has greatly reduced the incidence of perinatal HIV transmission, but a sizeable number of infants continue to be infected. World Health Organization guidelines recommend that ART should be initiated within 1 y of age in all children diagnosed with HIV, regardless of clinical staging or CD4 cell count (1).

Timing of early treatment has varied greatly in clinical studies with data for treatment initiation in vertically HIV-infected children spanning from a few hours after birth (2) up to 1 y of age (3). Benefits of early ART in virally controlled HIV-infected individuals have been described in both adults and children and include reduction in HIV reservoir size and prevention of disease progression. It is now well established that the earlier the treatment is initiated postinfection, the greater the reduction in the reservoir in HIV-infected adults (4, 5) and children (6–9). Early treatment is also associated with longer time to viral rebound after ART interruption (2, 10, 11).

The impact of timing of treatment initiation on induction of HIV-specific immune responses or the durability of HIV-specific immunity after virologic suppression is less clear and even controversial. Early treatment, within 6 mo of infection in HIV-infected adults, has been associated with preservation of quantity and quality of both T and B cell immune function (12–14) and enhanced recovery of CD4 T cells (15). Positive effects of early ART initiation have been noted in children as well, with evidence of preservation of memory B cell responses in children treated at age <1 y (3, 16). However, there is no conclusive evidence about the persistence and, more importantly, about the quality of the HIV-specific T cell responses in virally suppressed early-treated (ET) children.

Early treatment can also be associated with loss of HIV-specific immunity. Virus suppression due to early treatment initiation results in shorter exposure of the immune system to HIV Ags with potential negative effects on the Ag-specific immune response. Indeed, it is reported that ∼50% of individuals treated within 3 mo of age lose HIV-specific circulating Abs and become seronegative (8, 17). Poor detection of HIV-specific response in CD8 (18, 19) and CD4 (20) T cells was reported in HIV-infected children under viral control in whom ART had been initiated within 6 mo of age.

A clear understanding of the impact of timing of treatment initiation on the HIV-specific T cell memory responses in situations of controlled viremia is important for developing strategies aimed at permanent HIV remission in HIV-infected children.

In this study, we applied an in vitro stimulation protocol and multiparameter flow cytometry based approach in PBMC of virally suppressed HIV-infected young individuals in whom ART was started early (≤6 mo of age) or late (>1 y) after birth. Our goal was to ascertain quantity and quality of the HIV-specific CD4 and CD8 T cell responses to two different HIV Ags (GAG and ENV). We also evaluated the transcriptional signatures in unstimulated PBMC from these groups to identify persistent transcriptional effects associated with time of treatment initiation.

PBMCs were collected from 20 perinatally HIV-infected children (age range 2.1–15.9 y) with durable viral control (plasma HIV RNA < 50 cp/ml) being followed at Bambino Gesù Children Hospital, Rome, Italy.

Characteristics of the participants are reported in Table I. Briefly, children were divided based on their time of ART initiation into 14 ET, for those who initiated ART at ≤6 mo of age and six late treated (LT) for those who initiated ART at >1 y of age. ET were statistically younger than LT and with more females than LT but without differences in time on suppressive ART that was an average of 5.3 and 5.4 y, respectively, for ET and LT. CD4 absolute count in ET was higher than LT. For one LT, we did not have the exact date of ART initiation or the exact visit date, and this participant was therefore dropped for all the analyses in which these variables were involved. An HIV-negative group (n = 6), similar in age distribution (range 2.5–26.9 y) to the HIV infected, was included only for the RNA sequencing (RNAseq) experiment.

The following fluorochrome-conjugated anti-human mAb were used for flow cytometry studies: PD-1 BV421, CD40L BV605, perforin PE/Dazzle 594, and CD8 PerCP from BioLegend (San Diego, CA); CD3 BUV496, CD4 APC-Cy7, CD69 BV650, IL-2 BV711, CXCR5 Alexa 647, IFN-γ PE-Cy7, TNF-α FITC, granzyme B and CD27 BV480 from BD Biosciences (San Jose, CA); IL-21 PE from eBioscience, (San Diego, CA); and CD45RO PE-Cy5.5 from Beckman Coulter (Fullerton, CA). LIVE/DEAD Fixable Blue Dead Cell Stain Kit from Thermo Fisher Scientific (Boston, MA) was used to detect and exclude dead cells. All the reagents were tested and titrated for optimum concentration before usage.

PBMC were thawed, and rested overnight at 37°C 5% CO₂. Cells were stimulated at the concentration of 5 million per milliliter for 12 h in presence of 2 μg/ml gp140 (a gift from Dr. Kalyanaraman, Advanced BioScience Laboratories, MD), 2 μg/ml GAG PTE peptides (AIDS Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases [NIAID], National Institutes of Health [NIH]: HIV-1 PTE Gag Peptide Pool from NIAID, Division of AIDS), 1 μg/ml of staphylococcal enterotoxins B (SEB; List Biological Laboratories) or medium (negative control). At the beginning of culture, the following reagents were added: 1 μg/ml costimulatory molecule anti-CD28 mAb (BD Biosciences), 0.65 μl/ml protein transport inhibitor monensin (BD GolgiStop), and the degranulation marker (CD107a). After 6 h of stimulation, brefeldin A (10 μg/ml) was added to the stimulation to permit the accumulation of cytokines within the PBMC (21).

Following the culture period, cells were stained for flow cytometry using previously titrated mAb. Briefly, live cells were stained with Fixable Blue LIVE/DEAD and then incubated with appropriate surface Ab mixture (CD3, CD4, CD8, CD27, CD45RO, and CXCR5). Cells were then fixed and permeabilized with Cytofix/Cytoperm Buffer (BD Biosciences) and stained for intracellular molecules CD69, CD40L, IL-2, IFN-γ, TNF-α, IL-21, granzyme B, and perforin.

Total CD4 and CD8 T cells and subsets were analyzed for the expression of activation induced molecule CD40L or CD69 to identify the Ag-specific T cells (Supplemental Fig. 1A). Ag-specific cells were further evaluated for the stimulus-induced expression of intracellular cytokines IFN-γ, IL-21, IL-2, and TNF-α for CD4 T cells and IFN-γ, IL-2, TNF-α, perforin, and granzyme B for CD8 T cells (Figs. 2A, 4A). Stained cells were acquired on a BD LSRFortessa (BD Biosciences) and analysis performed using FlowJo v10.0.8 (Tree Star) software. Polyfunctionality of T cells was defined as the simultaneous detection of two or more of the analyzed cytokines based on Boolean analysis.

Total HIV DNA was quantified from PBMC at the enrollment time point for all patients using an in-house, real-time quantitative PCR assay as previously described (22). Briefly, QIAGEN DNA Mini Blood Kit was used for extraction of DNA. An ABI PRISM 7500 real-time PCR instrument with Invitrogen RT-PCR reagents was used for amplification and detection of HIV-1 DNA. Results are reported as copies of HIV per million cells.

Total RNA was extracted (QIAGEN RNeasy Plus Mini Kit) from cryopreserved PBMC and sequenced (NextSeq 500; 75 bp, paired-end, 40 million reads per sample; Illumina). Raw demultiplexed FASTQ paired-end read files were trimmed of adapters and filtered using the program skewer (23) to remove any with an average Phred quality score of <30 or a length of <36 bp. Trimmed reads were aligned to the Homo sapiens National Center for Biotechnology Information reference genome assembly version GRCh38 using the HISAT2 (24) and sorted using SAMtools (25). Aligned reads were counted and assigned to gene meta-features using the program featureCounts (26) as part of the Subread package. Count files were assessed for quality control, normalized, and analyzed using an in-house pipeline using the LIMMA-trend method (27) for differential gene expression testing and the gene set variation analysis (GSVA) (28) library for GSVA. The accession number for upload to the Gene Expression Omnibus and Sequence Read Archive public databases is GSE138804 (https://www.ncbi.nlm.nih.gov/geo/).

For the flow cytometry data, all the correlations were evaluated using the Spearman test. Comparison between two groups was made using the Mann–Whitney U test. Statistical tests were two sided, and the tests were considered statistically significant with a p value ≤0.05 and performed using GraphPad Prism version 8.0.0 for Windows (GraphPad Software, La Jolla, CA, http://www.graphpad.com).

Boolean analysis and figures were produced using Simplified Presentation of Incredibly Complex Evaluations (distributed by the NIAID, NIH [http://exon.niaid.nih.gov/spice]).

Differentially expressed genes and pathways were determined by moderated t-statistics (p ≤ 0.05), with pathway enrichment being performed by GSVA in R Bioconductor.

Written informed consent was signed by a parent or a legal guardian.

The study was approved by the University of Miami Institutional Review Board and Bambino Gesù Children Hospital ethical committee.

We characterized the CD4 and CD8 T cell compartments in the study population (Table I) by flow cytometry (Fig. 1A). ET and LT had similar frequencies of each of the major maturation subsets for CD4 T cells (Fig. 1B). In CD8 T cells, we observed a trend (p = 0.09) toward lower frequency of total CD8 T cells in ET. The frequency of the maturation CD8 subsets were similar between ET and LT with the exception of central memory CD8 T cells that were lower in ET (Fig. 1C).

After in vitro stimulation of PBMC in ET and LT with HIV peptides, the ENV- and GAG-responsive CD4 T cells were identified by upregulation of CD40L as an Ag-induced activation marker (Supplemental Fig. 1A). The function of CD40L⁺ CD4 T cells was investigated by cytokine production in response to Ag-specific stimulation. We assessed IFN-γ, IL-2, and TNF-α (Fig. 2A) by intracellular flow cytometry based on their key roles in achieving protective immune responses to viral infection, especially in combination (29, 30). We also included IL-21 to monitor the ability of HIV-specific T follicular helper (Tfh)–like cells to improve B cell responses (31–33). Despite similar frequencies of GAG-specific CD4 T cells in both groups, total GAG-specific IL-21⁺ and IFN-γ⁺ CD4 T cells were higher in the ET. Frequencies of IL-2 or TNF-α producing GAG-specific CD4 T cells were also higher in ET but did not reach statistical significance (Fig. 2B).

To get a better understanding of the quality of the HIV-specific CD4 T cells, we evaluated “polyfunctionality,” as defined by the capacity of the cells to produce two or more cytokines simultaneously. Cells that showed no production of cytokines were defined as “paucifunctional.” First, we compared the cumulative frequencies of Ag-specific CD4 T cells producing four, three, two, one, or no cytokines in ET and LT to generate an overview of the proportions of polyfunctional and paucifunctional cells in each group. This analysis showed that GAG-specific polyfunctional CD4 T cells producing three or two cytokines were statistically higher in ET (Supplemental Fig. 1B). Next, we characterized the composition of cytokine coproduction using Boolean analysis to identify whether the differences between the two groups were attributable to the expansion of specific subsets of Ag-specific cells. We observed higher frequencies of the combination of four, three, or two cytokine-producing, GAG-specific CD4 T cells in ET, whereas LT showed higher frequency of paucifunctional GAG-specific CD4 T cells (Fig. 2C, Supplemental Fig. 1B).

This analysis was repeated for ENV-specific CD4 T cells. Similar to the GAG-specific CD4 T cells, frequencies of ENV-specific CD4 T cells were comparable in ET and LT, but ET demonstrated higher frequency only for IL-21–producing cells with no statistical differences noted in the frequencies of IFN-γ–, IL-2–, or TNF-α–producing ENV-specific CD4 T cells (Fig. 3A). In ET, the cumulative analysis showed higher frequencies of two cytokine producers in CD4 T cells responding to ENV stimulation. Boolean analysis showed a trend toward higher frequencies of combinations of four and three cytokine producers, but a single IL-21–producing, ENV-specific CD4 T cell subset was the only one significantly higher in ET, whereas paucifunctional CD4 T cells were again higher in LT (Fig. 3B, Supplemental Fig. 1C).

To understand whether the differences in the HIV-specific CD4 T cell response observed between ET and LT were due to a generalized functional defect of the CD4 T cell compartment of the LT, we analyzed cytokine production following stimulation with the superantigen SEB (34). Frequencies of SEB-specific CD40L⁺ CD4 T cells were similar in ET and LT, but the profile of the cytokines produced was different from that of HIV Ag stimulation (Supplemental Fig. 2A).

Finally, we evaluated the correlation between age, time under ART, and HIV DNA with total ENV- and GAG-specific CD4 T cells, but no significant association were noted (Supplemental Table I).

Overall, these results demonstrated that early treatment preserves function and quality of the ENV- and GAG-specific CD4 T cells and these differences are not due to a generalized impaired function of the CD4 T cells in LT).

The CD8 T cell compartment was also evaluated for HIV-specific cell quantity and quality. After in vitro stimulation, the ENV- and GAG-responsive CD8 T cells were identified by upregulation of CD69 as an Ag-induced activation marker (Supplemental Fig. 1A). For assessment of CD8 T cell function, in addition to antiviral cytokines such as IFN-γ, IL-2, and TNF-α, we also evaluated molecules involved in CD8 T cell cytotoxic functions including the degranulation marker CD107a and the lytic molecules perforin and granzyme B (Fig. 4A). Frequencies of GAG-specific CD8 T cells were higher in LT. However, ET showed statistically higher proportion of CD107a⁺ and granzyme B⁺ cells along with a trend of higher IL-2⁺ (p = 0.08) and IFN-γ⁺ (p = 0.1) GAG-specific CD8 T cells (Fig. 4B). Cumulative frequency analysis showed a statistically higher proportion of three and two cytokine/marker⁺ cells as well as a trend of higher proportion of five (p = 0.1) and four (p = 0.054) producing GAG-specific CD8 T cells in ET (Fig. 4C, Supplemental Fig. 1D). Interestingly, within the polyfunctional subsets that were higher in ET, we found the cytotoxic GAG-specific CD8 T cells identified by the coexpression of CD107a, perforin, and granzyme B (Fig. 4D). LT showed higher frequency of paucifunctional GAG-specific CD8 T cells (Fig. 4C, Supplemental Fig. 1D).

The frequency of ENV-specific CD8 T cells was also higher in LT. However, qualitative analysis showed very different results compared with the GAG specific CD8 T cells. In fact, only a trend toward higher IFN-γ production in ET was observed for the single-cytokine producing ENV-specific CD8 T cells (Fig. 5A). Moreover, analysis of cumulative frequencies showed a trend toward more paucifunctional CD8 T cells despite a higher proportion of five cytokine/marker positive cells in LT (Fig. 5B, Supplemental Fig. 1E). Frequencies of CD69⁺ CD8 T cells after SEB stimulation were not different between ET and LT, despite a trend (p = 0.07) toward higher frequencies in LT. We observed higher frequencies of CD107a⁺CD69⁺ CD8 T cells in ET, but no statistical differences were observed for the remaining five markers investigated (Supplemental Fig. 2B). Total, ENV-, and GAG-specific CD8 T cells showed no association with age, time under ART, and HIV DNA (Supplemental Table I).

Overall, ET show a lower frequency but a better quality of HIV-specific CD8 T cells, especially in the GAG-specific CD8 T cell compartment.

To understand the effect of timing of treatment initiation at transcriptional level, we performed RNAseq in PBMC. Sequencing was performed in 13 ET, six LT, and six HIV-negative participants for comparison.

Pathway enrichment analysis of the differentially expressed genes between the participant groups revealed different pathway clusters that segregated HIV negative from both HIV-positive groups as well as ET from LT (Fig. 6). Only four HIV-infected participants (three ET and one LT) were segregated from their groups, but no peculiarity was seen in these individuals in terms of age distribution (range 62–14.6 y), time under ART (4.8–12.7 y), HIV DNA (0–314 copies per 1 × 10⁶ PBMC), and CD4 absolute count (432–1359 cells/μl). However, we noted that all of them were female.

To have an overview of the differences in the transcriptional profiles for the three groups in analysis, we evaluated the heatmap of the top 50 differentially expressed pathways.

LT exhibited an overall upregulation of almost all of the top 50 differentially modulated pathways. Cluster 1 (HIV up) contains pathways upregulated in ET and LT compared with HIV negative. The pathways included in this cluster are known to be upregulated during viral infection (e.g., TCYTOTOXIC) or by HIV infection (e.g., TCAPOPTOSIS, CTLA-4, and IL-17) (35, 36). Cluster 2 (ET down) is composed of pathways downmodulated in ET individuals compared with LT and HIV negative. Of interest is the presence of pathways associated with metabolism, particularly glycolysis (e.g., leptin, PGC1A, glycolysis, and feeder). Cluster 3 (LT up) is comprised of pathways upregulated only in LT, in which the majority are strongly associated with proliferation (e.g., SHH), calcium signaling (e.g., PLCE and calcineurin), and T cell activation (e.g., NFAT). Several pathways identified in clusters 2 and 3 (e.g., SHH, PLCE, calcineurin, NFAT, glycolysis, and feeder) were also directly associated with the time of treatment initiation (Supplemental Fig. 3A) and negatively with the frequency of the GAG-specific cytotoxic CD8 T cells (Supplemental Fig. 3B).

Overall, these results indicate that despite being under ART-mediated viral control for more than 5 y, vertically HIV-infected children who started ART after 1 y of life maintain an activated transcriptional profile in PBMC with a distinct upregulation of pathways related to proliferation, calcium signaling, and immune activation not seen in ET or HIV-negative individuals.

Early diagnosis and ART initiation can limit the size of the HIV reservoir and preserve immune system in children with perinatal HIV infection (2, 3, 6, 7, 10, 11, 16). However, early treatment also limits viral exposure and can affect the development of an efficient Ag-specific immune response. HIV Ab in particular can be lost with early treatment, but knowledge of cellular HIV-specific immunity is limited and is largely restricted to CD8 T cells (13, 14, 18–20). The objectives of the current study were to characterize the immune and transcriptional profile along with quantitative and qualitative evaluation of ENV- and GAG-specific CD4 and CD8 T cells in ET and LT HIV-infected children. To improve our ability to detect discrete differences, we selected two groups of children who had initiated ART at different ages (≤6 mo or >1 y) and had maintained virus suppression for a period of 1–10 y on ART. The participants were a subset of children from the group recently reported showing relationship of HIV reservoir and seroreversion in ET children (8). Our findings indicate that ET children maintain HIV-specific CD4 and CD8 T cell responses that are qualitatively superior to LT children. Moreover, they exhibit a transcriptomic profile in PBMC that is strikingly different in comparison with the LT group. These results illustrate the long-term impact of timing of ART initiation on the immune system that positions the ET perinatally HIV-infected children favorably for interventions aimed at achieving a functional cure.

Viral suppression with ART has an overall restorative effect on the immune systems of HIV-infected individuals (37, 38) in most instances. Indeed, our analysis showed that, except for higher central memory in CD8 of LT, the ET and LT were similar in the distribution of T cell maturation subsets and functional response to a non-HIV stimulus, SEB, suggesting that both groups have a similar potential for an effective T cell immune response. Recent data in HIV-infected children, however, indicate that despite overall immune restoration after 1 y of ART and virus suppression, the extent of qualitative recovery of HIV-specific T cell function remains uncertain (39). Using in vitro Ag stimulation followed by surface and intracellular staining, we identified CD4 and CD8 T cells that were activated by HIV and further evaluated their quality based on production of cytokines. We used HIV Ag–induced upregulation of CD40L on CD4 T cells and CD69 on CD8 T cells as markers of Ag-specific activation (40, 41). This approach enhanced the dataset because we were able to detect cells that responded to stimulation but did not produce cytokines. Additionally, this approach allowed us to maximize our participant group and overall sample size by avoiding a prescreening for specific HLA types needed for tetramer technology for the identification of Ag-specific cells (41).

Regardless of the Ag used, frequencies of HIV-specific CD4 T cells were similar in both groups, whereas those of CD8 T cells were lower in ET. This result is in agreement with previous observations that the frequency of HIV-specific CD8 T cells is reduced by early ART (13, 14, 18–20). Differences between ET and LT groups became evident in functional analysis of HIV-specific CD4 and CD8 T cells. Different Ags are known to elicit different responses, with ENV response arising earlier then GAG (42), but the latter is known to be more protective (43–45). In ET, both GAG-specific CD4 and CD8 T cells not only showed a higher frequency of the single cytokine producer cells, but also the frequencies of the polyfunctional cells producing combinations of two or more cytokines were higher. Polyfunctionality of HIV-specific CD4 T cells, especially GAG specific, and the frequency of cytotoxic HIV-specific CD8 T cells are characteristics observed in HIV viral controllers and are associated with better protection and slower disease progression (29, 46). In addition to a lower frequency of polyfunctional HIV-specific cells, LT showed a higher frequency of paucifunctional HIV-specific CD4 and CD8 T cells for both GAG and ENV stimulation. The presence of these cells in LT is of interest, because despite their inability to produce any of the cytokines investigated, they were Ag-specific cells based on expression of CD40L or CD69. Although we consider these cells as paucifunctional, a more in-depth investigation is required to understand whether these cells lack functionality or whether they produced a different set of cytokines. The possibility of an exhausted phenotype also needs consideration, because it was not examined. Overall, our data showed that early treatment initiation preserves HIV-specific CD4 and CD8 T cell functions that are important for control of infection.

We investigated potential links between HIV-specific CD4 and CD8 T cells and the viral reservoir in our participants but did not observe any correlation between HIV DNA and the frequency of HIV-specific CD4 or CD8 T cells. Because we also did not find any correlations between age, time of ART initiation, or time on suppressive ART with frequency of HIV-specific CD4 or CD8 T cells (Supplemental Table I), we speculated that other factors could contribute in the preservation of these cells, such as viral characteristics or a specific immune profile at the time of treatment initiation. Altogether, these results suggest that maintenance of the HIV-specific T cell response is not solely related to age, HIV DNA copies, or time under ART but may also be impacted by other factors.

Recently, it was shown in adults that starting ART during hyperacute infection also affects the immune transcriptome (14). To explore this in children, we investigated the transcriptional profile of unstimulated PBMC from both groups and included a healthy HIV-negative control group. Despite using PBMC and the absence of stimulation, we could clearly separate the three groups on the basis of three different pathways clusters. The LT up cluster of pathways showed upregulation of pathways associated with proliferation, calcium signaling, and T cell immune activation in LT, suggesting that early ART initiation preserved not only the HIV-specific T cell response, but the immune transcriptome as well. Interestingly, the ET down cluster showed downmodulation of pathways associated with the potential for immune activation (e.g., glycolysis pathways) compared with LT and HC. Activated cells have a higher energy demand and rely on glycolysis for sustenance (47). Thus, upregulation of pathways associated with glycolysis metabolism is considered a marker of immune activation. Additionally, HIV envelope Ags are reported to induce proliferation and a switch toward glycolysis (48, 49). The reason behind such differences in ET and LT is probably linked to a reduced systemic immune activation in ET because of lower viral replication and/or microbial translocation (50). Reduced immune activation is linked to lower exhaustion and better immune function (51). Accordingly, we observed that the majority of the pathways upregulated in LT compared with ET were also negatively associated with the frequency of the cytotoxic GAG-specific CD8 T cells. Further in-depth analysis of the metabolic and molecular targets identified in this study is needed to elucidate additional relationships.

Our study had two main limitations: 1) the low number of study participants, and 2) the cross-sectional study design that prevented us from gathering information about the kinetics of these cells. In fact, it is not known whether the beneficial effect of early ART initiation is in preventing the death of particular cell subsets or an indirect preservation of their function (e.g., by controlling systemic inflammation).

This study reinforces the concept that perinatally infected children treated early are potentially ideal candidates for evaluation of therapeutic interventions aimed at achieving a functional cure. Studies that further define early treatment in greater detail from birth onwards are needed for understanding the long-term impact of time of treatment initiation on HIV-specific immunity in virally controlled perinatally HIV-infected children.

We acknowledge all patients and guardians who participated in the study. We thank the Onco-Genomics Shared Resource at Sylvester Comprehensive Cancer Center for next generation sequencing services; Maria Pallin, Celeste M. Sanchez, and Margaret Roach for technical help; and Nadia Iavarone and Tamara Di Marco for clinical assistance. We thank Flow Cytometry Core Facility of the University of Miami for instrumentation and for facilitating conduct of the flow cytometry experiments. The following reagent was obtained through the AIDS Reagent Program, Division of AIDS, NIAID, NIH: HIV-1 PTE Gag Peptide Pool.

The authors have no financial conflicts of interest.