Immunosenescence of the CD8⁺ T cell compartment is associated with HIV-infection, but only weakly reflects age-related processes of adipose tissue, metabolism, and muscle in antiretroviral therapy-treated HIV-infected patients and controls

Between November 2010 and October 2012, 75 participants were included in this cross-sectional study: 60 patients with HIV-infection (HIV⁺) and 15 healthy men of similar age (Controls). HIV⁺ were recruited from the Department of Infectious Diseases, and Controls were recruited by advertisement at Copenhagen University Hospital, Hvidovre, Denmark. Inclusion criteria were: male sex; white ethnicity; > 18 years old; testing negative for hepatitis B and C; no intravenous drug use; and no current immunomodulating, lipid-lowering, anti-diabetic, or endocrinologic treatment. HIV⁺ were also required to have plasma HIV-RNA <400 copies/mL; ART for at least 12 months; and CD4 ⁺ T cell counts ≥200 cells/μL. The relationships between inflammatory biomarkers, adipose tissue distribution and low muscle mass in this cohort have been investigated [17]. Moreover, microdialysis results from 18 of the study participants have been published in a methodological study [18]. The study was approved by the local ethics committee of the Capital Region of Denmark (H-4-2010-045), the Danish Data protection agency (2010-41-4952), and was conducted in accordance with the Declaration of Helsinki. All participants gave written informed consent.

Plasma levels of the inflammation markers soluble urokinase plasminogen activator receptor (suPAR) and IL-6 were measured by enzyme-linked immunosorbent assays: suPARnostic (ViroGates A/S, Birkerød, Denmark) and Quantikine HS-IL-6 (R&D Systems, Minneapolis, MN). CD4⁺, CD8⁺ T cell counts, and HIV-RNA were routine measurements by the Department of Clinical Biochemistry, Copenhagen University Hospital, Hvidovre, Denmark. Peripheral blood mononuclear cells were isolated by centrifugation using BD Vacutainer cell preparation tubes (BD Biosciences, San Jose, CA) and cryopreserved.

Cryopreserved peripheral blood mononuclear cells from all participants were thawed and washed in complete RPMI 1640 media supplemented with GlutaMAX (Life Technologies, Grand Island, NY), 10 % fetal bovine serum, 100 U/mL penicillin, and 100 μg/mL streptomycin. 1.5 million cells were resuspended in 100 μL PBS and stained with the Zombie NIR Fixable Viability Kit (BioLegend, San Diego, CA) for 30 min. Cells were washed and resuspended in 10 % heat inactivated human serum in PBS for 20 min for Fc receptor blocking. Cells were incubated for 30 min with titrated anti-CD3-FITC, anti-CD8-PerCp-Cy5.5, anti-CD27-APC, anti-CD28-PE-Cy7 or anti-CD45RA-PE-Cy7 (BD Biosciences), and anti-PD-1-PE or anti-KLRG1-PE (BioLegend) antibodies and washed in FACS buffer (2 mM EDTA (Life Technologies) and 0.5 % bovine serum albumin (Miltenyi Biotec, Cologne, Germany) in PBS). Matching isotype and unstained controls were included for each participant. Compensations were generated for each fluorochrome using CompBeads (Anti-Mouse Ig, κ) (BD Biosciences). For each sample, 50 000 to 100 000 CD8⁺ events were recorded on a FACSCanto II flow cytometer (BD Biosciences). HIV- and lipodystrophy-status of the participants were concealed until after data analysis. Data were analysed using Cytobank [19].

CD8 ⁺ T cells could be subtyped into four maturation subsets based on CD27 and CD45RA expression: naïve (T_N: CD27⁺CD45RA⁺); central memory (T_CM: CD27⁺CD45RA⁻); effector memory (T_EM: CD27⁻CD45RA⁻); or effector memory re-expressing CD45RA (T_EMRA: CD27⁻CD45RA⁺) [20, 21]. By evaluating the expression of CD27 and CD28, CD8 ⁺ T cells could also be subtyped into four differentiation subsets: early differentiated (T_ED: CD27⁺CD28⁺); intermediate differentiated (T_ID: CD27⁺CD28⁻); late differentiated (T_LD: CD27⁻CD28⁻); or CD27⁻CD28⁺ [8]. The expression of the exhaustion marker PD-1 and of the senescence marker KLRG1 was evaluated in each of the maturation and differentiation subsets and in the total CD8 ⁺ T cell population [22].

Anthropometric measurements comprised body weight, height, and waist and hip circumferences, measured as in [15]. Abdominal VAT mass at the level of L4 was derived from computed tomography scans (Somatom Sensation 10, Siemens, Germany). Body composition was determined by dual energy X-ray absorptiometry (Norland XR-36, Gammatec A/S, Værløse, Denmark). Fat mass index (FMI) was calculated as: total fat mass/height² (kg/m²). Leg lean mass index (lLMI) was calculated as: leg lean mass/height² (kg/m²). Lean mass was used as a surrogate marker for muscle mass as lean mass is principally made of skeletal muscle. This is especially true for lean mass in the extremities [23].

Fasting glucose levels were routine measurements by the Department of Clinical Biochemistry. Fasting plasma levels of insulin were measured using Immulite 2000 System (Siemens, NY, USA) and insulin resistance was calculated using the homeostasis model assessment (HOMA-IR) [24]. Metabolic syndrome was defined according to the 2009 consensus definition for high risk populations [25]. The assessment of metabolic syndrome was based on the presence of at least three of the five following criteria: elevated waist circumference (≥94 cm, as recommended for high risk populations); elevated triglycerides (≥1.7 mmol/L); reduced HDL-C (<1.0 mmol/L); elevated blood pressure (systolic ≥130 or diastolic ≥85 mm Hg, or anti-hypertensive treatment); elevated fasting glucose (≥100 mg/dL).

Lipodystrophy status and the type of lipodystrophy were evaluated by clinical evaluation of the adipose tissue distribution of several subcutaneous regions (i.e. face, retroauricular and dorsocervical regions, upper arms, thighs, buttocks, and abdomen) and of the visceral region, as in [15]. HIV⁺ were characterized as lipodystrophic or non-lipodystrophic. Patients with lipodystrophy were further characterized as lipohypertrophic, lipoatrophic, or having mixed type lipodystrophy.

Comparisons between HIV⁺ and Controls were performed using unpaired Student’s t-test or Wilcoxon two-sample test where appropriate. For categorical parameters, Chi-squared test or Fisher’s exact test were used. To investigate the effect of HIV-infection on CD8 ⁺ T cell phenotypes, we used unpaired Student’s t-test or Wilcoxon two-sample test and linear regression analyses. The associations between HIV-and ART-related variables and CD8 ⁺ T cell phenotypes were investigated in the HIV⁺ group using multiple linear regression analyses adjusted for age. The associations between age, inflammation, metabolism, body composition, and CD8 ⁺ T cell phenotypes were investigated in the HIV⁺ group using simple and multiple linear regression analyses adjusted for age, HIV-, ART-duration, and current ART. The T_CM, T_EM, and T_EMRA subsets were grouped into one Memory group. Thus CD8 ⁺ T cell maturation was defined as either an antigen-inexperienced naïve (T_N) or an antigen-experienced Memory phenotype. CD8 ⁺ T cell differentiation was defined as either an undifferentiated CD28⁺ phenotype composed of T_ED and CD27⁻CD28⁺ cells, or a highly differentiated CD28⁻ phenotype composed of T_ID and T_LD cells. The loss of CD28 expression is considered to identify a group of cells characterized by functional changes and replicative senescence [9]. By grouping the individual subsets according to relevant biological and functional characteristics, we could reduce the number of analyses and simplify the interpretation. Nonetheless, data on the individual subsets are available as Additional files. Due to the compositional nature of the data, only the estimates of the Memory and CD28⁻ cells are presented in analyses for group sizes. The estimates for T_N cells are the opposite of the estimates for Memory cells since their proportions add up to 100 %, and the same is true for CD28⁺ and CD28⁻ cells. If the residuals were not normally distributed, the parameter was log-transformed to fulfil model requirements using log₂(x). Beta estimates (β) for log-transformed parameters were back transformed using (2^β -1) × 100, and thereby shown as percent change in outcome per unit increase of the covariate.

Lipodystrophy was not associated with changes in CD8 ⁺ T cell phenotype (Additional file 1 and Additional file 2). Therefore, patients with and without lipodystrophy were grouped (HIV⁺) in further analyses. CD8 ⁺ T cell maturation phenotypes: naïve (T_N), central memory (T_CM), effector memory (T_EM), or effector memory re-expressing CD45RA (T_EMRA), were determined based on CD27 and CD45RA expression. HIV⁺ had significantly lower proportions of T_N cells (P < 0.001) and significantly higher proportions of T_EM cells (P = 0.002) than Controls (Fig. 2a). CD8 ⁺ T cell differentiation phenotypes: early (T_ED), intermediate (T_ID), late (T_LD) differentiated, or CD27⁻CD28⁺, were determined based on CD27 and CD28 expression. HIV⁺ had significantly lower proportions of T_ED cells (P < 0.001) and significantly higher proportions of T_LD cells (P < 0.001) than Controls (Fig. 2b). We investigated whether these changes were mediated by differences in age or age-related processes using multiple regression analyses. The differences in CD8⁺ T cell subset distribution between HIV⁺ and Controls were primarily related to HIV or ART. Adjusting for age, inflammation, metabolism, or body composition, hardly affected the associations between HIV-infection and CD8⁺ T cell maturation and differentiation (Additional file 3).

Moreover, we investigated whether HIV-, ART-duration or current ART was associated with differences in CD8⁺ T cell subset distribution in HIV-infected patients using linear regressions adjusted for age. Current use of protease inhibitors was significantly associated with higher proportions of T_EM (P = 0.002) and T_LD (P = 0.002), and lower proportions of T_N (P = 0.007) and T_ED (P = 0.04) CD8⁺ T cells. Current use of non-nucleoside reverse-transcriptase inhibitors was associated with higher proportions of T_ID (P = 0.01). HIV- (P = 0.36–0.99) or ART-duration (P = 0.07–0.94) did not have an effect on CD8⁺ T cell subset sizes.

CD8 ⁺ T cell senescence and exhaustion were assessed on the basis of KLRG1 and PD-1 expression, respectively. HIV-infection did not have a significant effect on PD-1 and KLRG1 expression, neither in the total CD8 ⁺ T cell population or in the subsets (P = 0.06–0.94) (Fig. 3). Instead, PD-1 and KLRG1 expression depended on the maturation and differentiation stages of the cells. PD-1 expression was highest in intermediate subsets (Fig. 3a). KLRG1 expression was highest in highly differentiated subsets (Fig. 3b). Moreover, there was a significant association between proportions of PD-1⁺ and KLRG1⁺ CD8 ⁺ T cells across both groups, the proportion of KLRG1⁺CD8 ⁺ T cells was multiplied by 2.7 % for each 1 % increase in the proportion of PD-1⁺CD8 ⁺ T cells (Estimate = 2.72 %; P = 0.02, 95 % CI = 0.45–5.04). We did not find any association between HIV-, ART- duration, or ART and changes in KLRG1 or PD-1 expression (P = 0.06–0.89).

The aim of the study was to investigate changes within the CD8⁺ T cell compartment in well-treated HIV-infected patients, and further to investigate whether these changes were associated with age-related parameters. In contrast to the virus, which is thought to affect subsets with specific effector functions, we expected the ageing process to have a more general effect on the CD8⁺ T cell population. Therefore, we grouped T_CM, T_EM, and T_EMRA subsets into one Memory group; T_ED and CD27⁻CD28⁺ cells into one CD28⁺ group; and T_ID and T_LD into one CD28⁻ group. When grouping the subsets, HIV-infection remained associated with lower proportions of T_N and CD28⁺, and higher proportions of Memory (P = 0.0002) and CD28⁻ (P = 0.0003) CD8⁺ T cells. We investigated the associations between CD8⁺ T cell distribution, age, and age-related parameters in the HIV⁺ group in simple and adjusted analyses (Fig. 4). Analyses for the individual subsets are available in Additional file 4. HOMA-IR index was significantly associated with higher proportions of Memory (Unadjusted P = 0.02, adjusted P = 0.03), mainly T_EM cells (Unadjusted P = 0.02, adjusted P = 0.003), and lower proportions of T_N cells (Unadjusted P = 0.02, adjusted P = 0.03) (Fig. 4 and Additional file 4). We found no significant associations between memory or CD28⁻ CD8⁺ T cell group sizes with age, inflammation or VAT overall (Fig. 4). However, at the individual subset level, IL-6 levels were associated with changes in the proportions of T_ID, and VAT was associated with lower proportions of T_EMRA CD8⁺ T cells (Additional file 4).

We found PD-1 and KLRG1 expression to vary according to subset differentiation and maturation, but not to HIV-infection (Fig. 3). Thus, we wanted to investigate whether PD-1 and KLRG1 expression in total CD8 ⁺ T cells were associated with age and age-related processes in the HIV⁺ group in simple and adjusted regression analyses. When grouping the subsets, HIV-infection was not associated with changes in KLRG1 or PD-1 expression (P = 0.14–0.87). There were no associations between KLRG1 or PD-1 expression in total CD8 ⁺ T cells and age-related parameters (Fig. 5).

Within each CD8 ⁺ T cell subset, there were still large variations in the number of KLRG1⁺ and PD-1⁺ cells. We therefore also investigated whether age and age-related processes were associated with PD-1 and KLRG1 expression in specific CD8 ⁺ T cell groups: T_N and Memory, and CD28⁺ and CD28⁻ cells. In participants with metabolic syndrome, the proportion of KLRG1⁺CD28⁻ cells increased by 14 % in unadjusted analyses, and by 18 % in adjusted analyses (Unadjusted P = 0.05, adjusted P = 0.04), compared to those without metabolic syndrome (Table 2). lLMI was associated with an increase in the proportions of KLRG1⁺T_N cells (P = 0.02). VAT was associated with an increase in the expression of KLRG1 in CD28⁺ (Unadjusted P = 0.02, adjusted P = 0.03) and CD28⁻ cells (Adjusted P = 0.04). Associations of KLRG1 expression and metabolic syndrome, VAT and lLMI were also observed in the individual subsets (Additional file 5).

We found no associations between PD-1 expression and age-related parameters (Additional file 6 and Additional file 7).