PWH (n=25) and healthy donors (n=16) were recruited between March and July 2021 at the Hospital Universitario Severo Ochoa (Madrid, Spain) and the Primary Healthcare Center Doctor Pedro Laín Entralgo (Madrid, Spain), respectively, for this prospective, longitudinal, observational study. Sample size was calculated using the sample size calculator Granmo (29) based on a level of confidence α=0.05 (95%) and power of the analysis β=0.2 (80%). The inclusion criteria were to be over 18 years old, CD4 > 500 cells/μl, not having a previous diagnosis of COVID-19, and to have been vaccinated through the Spanish Vaccination Program with two doses of one vaccine approved at the time for these participants: Comirnaty (BioNTech, Mainz, Germany; Pfizer, New York, NY) or Spikevax (Moderna, Cambridge, MA); or with one dose of Jcovden (Janssen, Titusville, NJ).

Blood samples were obtained before vaccination and 4–6 weeks after receiving the full vaccination schedule. Basal serology to detect IgG against SARS-CoV-2 was performed in plasma from the first blood sample to discard an asymptomatic infection before vaccination. All participants were followed up for 12 months after receiving the full vaccination schedule to record SARS-CoV-2 breakthrough infections.

All individuals gave informed written consent to participate in the study before their inclusion. Current Spanish and European Data Protection Acts guarantees the anonymity and confidentiality of all participants. Protocol for this study (CEI PI 29_2021-v3) was prepared in accordance with the Helsinki Declaration and previously reviewed and approved by the Ethics Committees of Instituto de Salud Carlos III (IRB IORG0006384) and the participating centers.

Peripheral blood mononuclear cells (PMBCs) and plasma were isolated from whole blood collected in EDTA Vacutainer tubes (Becton Dickinson, Madrid, Spain) using Ficoll-Hypaque density gradient (Pharmacia Corporation, North Peapack, NJ) by centrifugation and then they were cryopreserved until analysis. Vero E6 cell line (African green monkey kidney; ECACC 85020206) was kindly provided by Dr. Antonio Alcami (CBM Severo Ochoa, Madrid) and it was cultured in DMEM supplemented with 10% FCS, 2 mM L-glutamine, and 100 units/ml penicillin and streptomycin. HEK-293T (embryonic human kidney; ECACC 85120602) and TZM-bl/JC53BL-13 (human cervix; NIH AIDS Research and Reference Reagent Program, no. 8129) cell lines were obtained from the existing collection of the Instituto de Salud Carlos III (Madrid, Spain) and they were cultured as described above for Vero E6 cells.

Serology against the spike (S) protein of SARS-CoV-2 was performed in plasma of all participants using Euroimmun Anti-SARS-CoV-2 ELISA Assay (EI 2606–9601-10 G; Euroimmun, Lübeck, Germany) that can detect Alpha, Beta, Gamma, and Delta variants. Semi-quantitative results were analyzed by calculating the ratio of extinction of each plasma sample over the calibrator. Results were considered positive with IgG titer > 1.1 and negative with IgG titer < 0.8. Undermined results were located between values 1.1 and 0.8.

Pseudotyped single-cycle pNL4–3Δenv_SARS-CoV-2-SΔ19(G614)_Ren virus was obtained by co-transfection of HEK-293T cells with vector pNL4–3Δenv_Ren that contains HIV-1–1 genome without env and renilla luciferase gene as reporter (30), together with vector pcDNA3.1-SARS-CoV-2-SΔ19 that expresses SARS-CoV-2 S glycoprotein gene without the last 19 amino acids (QHU36824.1) (31). Co-transfection of pcDNA3.1-SARS-CoV-2-SΔ19 with vector pcDNA-VSV-G that expresses spike (S) glycoproteins of vesicular stomatitis virus (VSV) was used as specificity control for neutralizing assays [VSV_SARS-CoV-2-SΔ19(G614)]. The concentration of HIV-1 p24-Gag protein in the cell culture supernatants was measured 48 hours after transfection by Elecsys HIV-1 AG (Roche Diagnostic, Basel, Switzerland).

To measure the neutralization activity without ART interference, IgG were purified from plasma using Protein A HP SpinTrap (Sigma Aldrich-Merck, Darmstadt, Germany) and quantified by NanoDrop spectrophotometer (Thermo Scientific). 2-fold serial dilutions of purified IgG from IgG-positive plasma (1 to 0.008 mg/ml) were pre-incubated for 1 hour with pNL4–3Δenv_SARS-CoV-2-SΔ19(G614)_Ren pseudovirus (10ng p24-Gag per well), as previously described (32). This mixture was then added to a monolayer of Vero E6 cells and incubated for 48 hours. Viral infection in the monolayer was quantified by measuring Renilla luciferase activity (Renilla Luciferase Assay, Promega, Madison, WI) in the luminometer Centro XS3 LB 960 with MikroWin 2010 software (Berthold Technologies, Baden-Württemberg, Germany).

Titers of neutralizing antibodies were calculated as 50% inhibitory dose (NT50) which is the highest dilution of plasma that produced 50% reduction of Renilla luciferase activity in comparison with control by using a non-linear regression in GraphPad Prism Software v10.1.2. (GraphPad, Inc., San Diego, CA).

Direct cellular cytotoxicity (DCC) of PBMCs was determined using Vero E6 cells infected with pNL4–3Δenv_SARS-CoV-2-SΔ19(G614)_Ren virus as target, as previously described (33). Briefly, Vero E6 cells were infected with pNL4–3Δenv_SARS-CoV-2-SΔ19(G614)_Ren (100ng p24-Gag) for 48 hours and then co-cultured for 1 hour with PBMCs from the participants (ratio 1:2). Vero E6 cell monolayer was dissociated with trypsin-EDTA (Sigma Aldrich-Merck, Darmstadt, Germany) and caspase-3 activity in these cells was measured by chemiluminescence using Caspase-Glo 3/7 Assay system (Promega, Madison, WI) and a luminometer Centro XS3 LB 960 with MikroWin 2010 software (Berthold Technologies) as a measure of PBMCs cytotoxic activity against target cells (34). PBMCs were collected from the supernatants and analyzed by flow cytometry to characterize the cytotoxic cells. The following controls were used: target cells alone as negative control, target cells infected with pseudotyped SARS-CoV-2 as basal control for viral replication, and infected target cells co-cultured with NK cell line NKL (CVCL_0466) during the same time that PBMCs from the participants as positive control for viral replication and caspase-3 activity.

TZM-bl cells infected with HIV-1 wild-type strain NL4–3 for 48 hours were used as target to evaluate the DCC of PBMCs against HIV-1. HIV-1-infected TZM-bl cells were co-cultured for 1 hour with PBMCs (ratio 1:2). TZM-bl cell monolayer was detached with Trypsin-EDTA and caspase-3 activity was measured by chemiluminescence as described above. Viral replication was determined by quantification of luciferase activity in the monolayer using the Luciferase Assay System (Promega) in the luminometer Centro XS3 LB 960 with MikroWin 2010 software (Berthold Technologies). PBMCs were collected from the supernatants and analyzed by flow cytometry to characterize the cytotoxic cells. The following controls were used: target cells alone as negative control (mock cells), target cells infected with NL4–3_wt strain as basal control for viral replication, and infected target cells co-cultured with NKL cells during the same time that PBMCs from the participants as controls for stimulation of caspase-3 activity and interference with viral replication.

The characterization of PBMCs that were co-cultured for 1 hour with Vero E6 and TZM-bl cell monolayers was performed by staining with the following conjugated antibodies purchased from BD Biosciences (San Jose, CA): CD3-PE, CD8-APC-H7, CD56-BV605, TCRγδ-FITC, and CD107a-PE-Cy7. CD107a was used as a degranulation marker and its expression at the cell surface peaks within 1 hour of target cell engagement (35), being afterwards actively recycled from the cell surface (36). CD3+CD8- were assumed to be CD4+ T cells, which included those cells with downregulated CD4 expression caused by HIV-1 infection (37). Isotype controls were used to determine the background signal. Data acquisition was performed with BD LSRFortessa X-20 flow cytometer (BD Biosciences) and data analysis with FlowJo software v10.0.7 (Tree Star Inc.). Gating strategy is shown in Supplementary Figure 1 .

All statistical analyses were performed using GraphPad Prism software v10.1.2. (GraphPad Software Inc., San Diego, CA). Data normality was assessed with Kolmogorov-Smirnov test. According to non-normal distribution of data, for comparisons within the same group in two longitudinal samples, we applied two-tailed Wilcoxon matched-pairs signed rank test. For comparisons between two independent groups, we performed unpaired, two-tailed, nonparametric Mann-Whitney U test to compare between ranks. According to data non-normality, nonparametric Spearman’s rank correlation coefficient was calculated to evaluate the correlation between the cytotoxic activity against target cells and viral replication. Simple linear regression analysis was performed using STATA 14.2 software (StataCorp LLC, College Station, TX) to estimate β regression coefficient in the comparisons between all parameters determined in PBMCs from PWH compared to healthy donors, before and after receiving the full vaccination schedule. Differences in breakthrough infections between PWH and healthy donors were evaluated by chi-square test for 2x2 contingency table using Graph Pad Prism v10.2.1. P values (p) < 0.05 were considered statistically significant in all comparisons.

Sociodemographic and clinical characteristics of all participants are summarized in Table 1 and detailed for PWH in Supplementary Table 1 . Median age was 37 years old (Interquartile range (IQR) 33–39) in PWH and 40 years old (IQR 36–44) in healthy donors. 88% of PWH and 50% of healthy donors were male. Most participants (76% PWH and 81% healthy donors) were vaccinated with two doses of Comirnaty; 16% and 12%, respectively, received Spikevax, and 8% and 6%, respectively, received Jcovden. Median time between doses was 23 (IQR 21–33) and 21 days (IQR 21–22) for PWH and healthy donors, respectively. None of the participants in the cohort of PWH had detectable HIV-1 viremia after receiving the full vaccination schedule against SARS-CoV-2.

Median time of HIV-1 infection was 6 years (IQR 3–13) and median age at diagnosis was 29 years old (IQR 24–32). Median CD4 count was 775 (IQR 622–998) cells/µl, and median CD8 count was 858 (IQR 589–1167) for PWH. Median CD4/CD8 ratio was 1.0 (IQR 0.7–1.3) for PWH and 2.0 (IQR 1.5–2.8) for healthy donors. All PWH were on ART and the most common regimen contained three drugs of two nucleoside reverse transcriptase inhibitors (NRTIs) with one integrase inhibitor (INI) (52%), followed by two-drugs combinations of one NRTI and one INI (24%). Most PWH (92%) showed undetectable viral load at sampling.

All participants were followed up for 12 months after receiving the last dose of the full vaccination schedule to record SARS-CoV-2 breakthrough infections ( Table 1 ). Statistical significance in the comparison between PWH and healthy donors was determined by chi-square test ( Supplementary Figure 2 ). Eight (32%) PWH and eight (50%) healthy donors had breakthrough infections with SARS-CoV-2 during this period (p=0.0097; Supplementary Figure 2A ). Clinical data of breakthrough infections are detailed in Table 2 . Only one (12.5%) participant from PWH cohort with CD4/CD8 of 0.68 required hospitalization due to complications of COVID-19 (p=0.0004; Supplementary Figure 2B ). Most common symptoms during SARS-CoV-2 breakthrough infection in PWH were fever (50%), cough (50%), and dyspnea (37.5%) (p<0.001, p=0.0003, and p<0.0001, respectively), while healthy donors only reported cough (25%) as main symptom ( Supplementary Figure 2B ). Median time from the last vaccine dose to SARS-CoV-2 breakthrough infection was 5.6 months (IQR 4.5–7.7) and 5.3 months (IQR 0.9–5.7), respectively. Most individuals with breakthrough infections from both cohorts were vaccinated with Comirnaty (87.5% for both PWH and healthy donors), which was the vaccine most administered to the participants.

The first blood sample was collected before receiving the first vaccine dose. A basal serology against SARS-CoV-2 was performed in this sample to identify and exclude those individuals with previous asymptomatic COVID-19. Median time since the participants received the full vaccination schedule and the collection of the second blood sample was 42 (IQR 27–50) and 33 days (IQR 29–37), respectively, in PWH and healthy donors ( Table 1 ).

The analysis of IgG levels against SARS-CoV-2 revealed an excellent humoral response of PWH after receiving the full vaccination schedule, which was comparable to healthy donors (p<0.0001 and p=0.0002, respectively) ( Figure 1A ).

The neutralizing capacity of plasma isolated from PWH was evaluated by using pseudotyped NL4–3Δenv_SARS-CoV-2-SΔ19(G614)_Ren virus to infect Vero E6 as target cells, as well as pseudotyped VSV_SARS-CoV-2-SΔ19(G614) virus to determine the presence of unspecific reaction, as described previously (33). We observed a high non-specific neutralization activity of VSV_SARS-CoV-2-SΔ19(G614)_Ren virus in plasma from PWH ( Supplementary Figure 3 ) that has been previously related to interference with ART based on integrase inhibitors (18). To avoid false positive neutralization results, the neutralizing capacity against SARS-CoV-2 was analyzed after purification of IgG from plasma of PWH and healthy donors. The neutralizing capacity of IgG purified from plasma of PWH was 4.7-fold higher than healthy donors (p=0.0295) ( Figure 1B ).

DCC of PBMCs from PWH against Vero E6 cells infected with pseudotyped NL4–3Δenv_SARS-CoV-2-SΔ19(G614)_Ren virus was 3.3-fold (p=0.0351) higher in the basal sample than in PBMCs from healthy donors ( Figure 2A ). Accordingly, viral replication in Vero E6 cells was reduced 2.7-fold (p=0.0110) in the presence of PBMCs from PWH previously unexposed to SARS-CoV-2 ( Figure 2B ). This difference was still significant after receiving the full vaccination schedule (-2.6-fold; p=0.0077). Calculation of the Spearman’s rank correlation coefficient showed a significant negative correlation between DCC and viral replication in SARS-CoV-2-infected target cells in the presence of PBMCs from PWH (r=-0.5059; p=0.0479) and healthy donors (r=-0.5294; p=0.0372) after receiving the full vaccination schedule, and before vaccination in PBMCs from PWH (r=-0.5088; p=0.0464), but not from healthy donors ( Supplementary Figure 4 ).

No significant differences between PWH and healthy donors were observed in the levels of CD4+ T cells before and after the administration of COVID-19 vaccines ( Supplementary Figure 5 ), but the levels of CD8+ T cells in PBMCs from PWH were higher than healthy donors before and after receiving the full vaccination schedule (1.5-fold, p=0.0080 and 1.4-fold, p=0.0298, respectively) ( Figure 3A , left graph). The degranulation capacity of these cells, determined by the expression levels of CD107a of the cell surface, was similar between both cohorts but it was significantly higher in PWH after receiving the full vaccination schedule (1.3-fold; p=0.0124) ( Figure 3A , right graph).

The levels of Tγδ cell populations with phenotypes CD3+CD8-TCRγδ+ and CD3+CD8+TCRγδ+ were higher in the basal sample of PBMCs from PWH than in healthy donors (1.8-fold, p=0.0393 and 2.6-fold, p=0.0088, respectively) ( Figures 3B, C , left graphs). The levels of CD8- Tγδ cells were still higher in PWH than in healthy donors after receiving the full vaccination schedule (3.9-fold; p=0.0028). Within cohorts, the levels of CD8- Tγδ cells did not change after receiving the full vaccination schedule in either group, while the levels of CD8+ Tγδ cells were reduced 2.0-fold (p=0.0018) in PWH after receiving vaccination. The degranulation capacity of Tγδ cell populations was significantly higher in PBMCs from PWH than in healthy donors, and these differences achieved statistical significance in the comparison after receiving the full vaccination schedule (1.8-fold, p=0.0115 and 2.0-fold, p=0.0040, respectively) ( Figures 3B, C , right graphs). In PWH, CD8+ Tγδ cells also showed higher degranulation capacity than healthy donors in response to infected cells before vaccination (1.6-fold; p=0.0414) ( Figure 3C , right graph).

We did not find significant differences in the levels of NK and NKT cells in PBMCs from both cohorts, before and after receiving the full vaccination schedule, or in the expression levels of CD107a in these cell populations ( Supplementary Figure 6 ). The effect of COVID-19 vaccination on the levels of CD8+ T cells and Tγδ cells expressing CD56 on the cell surface was also analyzed. After vaccination, the levels of CD3+CD8+CD56+ cells increased 1.8-fold (p=0.0462) in healthy donors, in comparison with PWH ( Supplementary Figure 7A , left graph). No changes in the expression of CD56 in Tγδ CD8+ or CD8- cells were observed between PWH and healthy donors ( Supplementary Figures 7B, C , left graph). In addition, the expression of CD107a increased 2.0-fold (p=0.0142) in Tγδ CD8+CD56+ from PWH, in comparison with healthy donors, after vaccination ( Supplementary Figure 7C , right graph).

Simple linear regression analysis was performed to estimate β regression coefficient in the comparisons between all parameters involved in cellular immunity determined in PBMCs from PWH compared to healthy donors before and after receiving the full vaccination schedule. Before vaccination, the average difference between DCC of PBMCs from PWH in comparison with healthy donors was positive (β=69957.19; p=0.050), while the average difference for SARS-CoV-2 replication from infected target cells was negative (β=-101924.30; p=0.009) ( Supplementary Table 2A ). In addition, the average difference for the levels of CD8+ T cells (β=12.4840; p=0.003), Tγδ CD8+ cells (β=5.7361; p=0.043), and their degranulation capacity (Tγδ CD8+CD107a cells) (β=16.3553; p=0.021) was also positive in the comparison between PWH and healthy donors. These results supported the notion that PWH had cellular immunity with capacity to eliminate SARS-CoV-2-infected cells before being exposed to the virus or vaccine. After vaccination, the average difference between the ability of PBMCs from PWH to eliminate SARS-CoV-2-infected target cells in comparison with healthy donors was still negative (β=-90834.43; p=0.004), while the average difference for the levels of CD8+ T cells (β=10.2614; p=0.016), Tγδ CD8- cells (β=1.4706; p=0.034), and Tγδ CD8+CD107a cells (β=20.8745; p=0.002) was still positive ( Supplementary Table 2B ).

DCC against TZM-bl cells infected with NL4–3 wild-type strain was evaluated in PBMCs from PWH to determine the effect of the full vaccination schedule against SARS-CoV-2 in the immune response against HIV-1. DCC response was higher against target cells infected with HIV-1 than with pseudotyped SARS-CoV-2 (see Figure 2A ), but no significant changes in DCC against HIV-1-infected cells were observed before and after vaccination against SARS-CoV-2 ( Figure 4A ) or on the viral replication in the target infected cells ( Figure 4B ). No changes were observed either in the levels of CD8+ T cells and Tγδ CD8+ or CD8- cells ( Figure 4C , left graphs) or in the expression of the degranulation marker CD107a after vaccination in the response against HIV-1 infected target cells ( Figure 4C , right graphs). Calculation of the Spearman’s rank correlation coefficient showed a significant negative correlation between DCC and viral replication in HIV-1-infected target cells in the presence of PBMCs from PWH before (r=-0.4289; p=0.0412) and after receiving the full vaccination schedule (r=-0.4444; p=0.0383) ( Supplementary Figure 8 ).