Chronically Activated T-cells Retain Their Inflammatory Properties in Common Variable Immunodeficiency

Ethical approval for this study for all participants was received from the Medical Ethical Committee of the Erasmus MC University Medical Center in Rotterdam, The Netherlands (METC: 2013–026). Written informed consent was obtained from all patients and controls according to the Declaration of Helsinki.

Patients were diagnosed with CVID according to the European Society for Immunodeficiencies criteria [1] and were included at the outpatient clinics of the UMC Utrecht and the Erasmus MC University Medical Center Rotterdam, The Netherlands. Patients were eligible if they were aged seven or older, and they and/or their legal guardians signed informed consent. Household members of patients were recruited as healthy controls (HC). Medication use up to 3 months prior to sampling was recorded.

Peripheral blood mononuclear cells (PBMC) were isolated from blood by Ficoll density centrifugation (GE Healthcare-Biosciences, AB), and frozen at – 180 °C until use. Cells were subsequently thawed, counted and plated at 1,000,000 live cells per panel per sample. For the panels including intracellular cytokine measurement, cells were first restimulated with 20 ng/mL phorbol 12-myristate 13-acetate (PMA, MilliporeSigma) and 1 μg/mL ionomycin (MilliporeSigma) for 4 h at 37 °C with addition of monensin (Golgistop, BD Biosciences, 1:1500) during the last 3.5 h. Cell death was stained in all panels using Fixable Viability Dye eFluor 506 (eBioscience). Next, cells were incubated with the surface antibodies (Supplementary Table 1) for 20 min at 4 °C and washed. Cells were then permeabilized with fixation/permeabilization reagent (eBioscience) for 30 min at 4 °C, washed, and incubated with the intracellular antibodies (Supplementary Table 1). Cells were stored at 4 °C until the next day, when they were measured on the LSR Fortessa (BD Biosciences).

For flow cytometric data, median fluorescence intensities (MFI) and percentages of positive cells were analyzed in FlowJo. All statistical analyses and graphic representations were done in R 3.2.0 [21]. Continuous variables were compared using the Mann–Whitney rank test, or paired Wilcoxon-rank test for paired samples. Correlation was calculated using Spearman’s correlation. P values below 0.05 were considered statistically significant.

To investigate immune activation and regulation in CVID, PBMC were isolated from 12 healthy controls (HC), 12 CVID patients with infections only (CVIDio), and 20 patients with CVIDid (Table 1). Patients were selected from a cross-sectional Dutch primary immunodeficiency cohort when they received IgRT and did not use any immunomodulatory medication during and the last 3 months prior to sampling. Three of the CVIDid patients had a known CVID-associated monogenetic disease (CTLA4 haploinsufficiency, STAT1 gain of function, and PIK3R1). Flow cytometry was performed as described in the Supplementary Information (see also Supplementary Table 1 and Supplementary Figs. 1–4).

First, pooled flow cytometry data was analyzed in an unsupervised manner using principal component analysis (PCA, Fig. 1A). CVIDio and HC clustered closely together and were distinct from most CVIDid samples. This suggests that most variation in the flow cytometric T-cell data related to immune dysregulation and not to the hypogammaglobulinemia shared by CVID patients. CVIDid patients did not cluster by treatment history or location of autoimmunity (Supplementary Fig. 5), suggesting that peripheral blood T-cell skewing may be generalized among CVIDid patients.

Next, the distribution of CD4 + and CD8 + T-cell subsets in CVIDid was assessed. Frequencies in CD4 + , CD8 + , and naïve/memory subsets were similar to that reported in previous studies [12, 22]: a decreased CD4/CD8 ratio (Fig. 1B), and a trend of decreased naïve (CD45RA + CCR7 +) CD4 + T-cells in CVIDid compared to CVIDio (p = 0.07) (Fig. 1C). In addition, a subset of CVIDid patients showed a high percentage of CD4 + T effector memory cells, and CD4 + T effector memory cells re-expressing CD45RA (TEMRA), which are associated with chronic activation such as observed in viral infections [23] (Fig. 1C). Within the naïve CD4 + T-cells, CD31 + recent thymic emigrants were more abundant in CVIDid than the CD31-central naïve T-cells (Supplementary Fig. 6). No differences in CD8 + T-cell distribution were observed (Fig. 1C).

Within the effector/memory (CD45RO +) subpopulation, proportions of CD4 + T-cells expressing HLA-DR, Ki67, and IFNγ were significantly increased in CVIDid, while IL17a, IL13, and TNF-α expressing T-cells were not different between CVIDid and CVIDio (Fig. 2A). A similar activation pattern was observed in the CD8 + effector memory population for HLA-DR, Ki67, and GzmB (Fig. 2B). We previously described [17] serum cytokine and chemokine levels in an overlapping cohort, allowing comparison between soluble serum markers and T-cell characteristics. Pooling the data from HC, CVIDio, and CVIDid, we observed that the proportion of IFNγ + Th cells correlated with serum levels of interferon-inducible chemokines CXCL9, 10, and 11 (Fig. 2C). While we observed an increase of serum IL17a in CVIDid, there was no corresponding increase of IL17a-producing Th cells, and the frequency of these cells did not correlate with the serum IL17a levels (Supplementary Fig. 7).

Next, we investigated whether this immune activation resulted in immune exhaustion in CVIDid, which is known to happen in the context of chronic inflammation [24]. In CVIDid CD4 + -cells, we observed an increased proportion of cells expressing negative regulators of immune activation PD1, LAG3, CTLA4, ICOS, and TIGIT (Fig. 3A). In addition, CVIDid CD4 + T-cells expressed higher levels of CD95 (FAS-L), showing that they are terminally differentiated and may be more prone to apoptosis (Fig. 3B). In CD8 + T-cells, LAG3, CTLA4, ICOS, and TIGIT, but not PD1 and CD95, were similarly increased in CVIDid (Supplementary Fig. 8).

In order to assess whether the PD1 and LAG3 expressing cells were functionally exhausted, production of IFNγ, GzmB, Ki67, and CD95 was compared between the PD1/LAG3 + and PD1/LAG3 − populations (Fig. 3C and Supplementary Fig. 9). In CD4 + T-cells, the PD1 + and LAG3 + subpopulations showed equal or increased expression of IFNγ, Ki67, and/or GzmB as the PD1 − and LAG3 − subpopulations, indicating retained functional capacity. Expression of PD1 correlated with serum levels of pro-inflammatory CXCL9, 10, and 11 as well as immune regulatory IL10 (Fig. 3D), further illustrating the association between PD1 and chronic inflammation. In CD8 + T-cells, IFNγ and GzmB expression was also intact in PD1 + and LAG3 + cells, but Ki67 + cells were less frequent in the LAG3 + subpopulation, indicating reduced proliferative capacity (Supplementary Fig. 10).

In addition to negative regulators of immune activation, regulatory T (Treg) cells also contribute to limiting inflammation-related pathology. In contrast to previous studies [25], we did not observe a decreased proportion of CD25 + FOXP3 + Tregs in CVIDid, and a subgroup of CVIDid patients had even increased Treg frequencies (Fig. 4A). ICOS expression in CVIDid Tregs also indicated normal-to-increased activation [26] of Tregs (Fig. 4B), but CTLA4 expression was reduced in CVIDid Tregs, both in cell frequency as MFI in the CTLA4-positive population (Fig. 4B). The failure to upregulate CTLA4 in CVIDid Tregs was isolated to the activated and usually highly suppressive [27] CD27 + Treg population (Fig. 4C), and may result in reduced Treg function, as expression of CTLA4 is important for the suppressive function of Tregs [28]. In addition, a subpopulation of CVIDid patients showed low levels of TIGIT-expressing Tregs, which may impair TIGIT-driven selective suppression of Th1 and Th17 effector cells [29].

The immune dysregulation of most CVIDid patients occurs not only in the systemic immune system, but also locally in affected tissues, which are often mucosal sites (lung and/or gut). Expression of the mucosal homing marker CCR9 was increased on naïve CD4 + and CD8 + populations of CVIDid T-cells (Fig. 5A), but not on non-naïve (CD45RA −) CD4 + T-cells (Fig. 5B), suggesting that this upregulation was at least not entirely antigen-driven. CVIDid CD8 + non-naïve (CD45RA −) T-cells (Fig. 5C) did also show increased expression of CCR9 as well as integrin α4β1, which mediates homing to inflamed tissues, including the lung. CVIDid FOXP3 + CD4 + T-cells (Fig. 5D) also showed increased migratory capacity, as they were enriched for cells expressing integrin α4β1 and gut-homing marker integrin α4β7.

Finally, in order to illustrate patterns of T-cell immune dysregulation in individual patients with varying severity of specific immune dysregulation phenotypes, we selected markers most strongly associated with (chronic) immune activation in CD4 + T-cells (HLA-DR, Ki67, IFNγ, and IFNγ in PD1 + and LAG3 + , PD1 + , LAG3 + , TEMRA, and CD95) and immune regulation (Tregs, CTLA4 in Tregs, MFI of CTLA4 in Tregs, TIGIT in Tregs, ICOS in Tregs, and CTLA4 in CD4 + T-cells) in CVIDid. We investigated these in five patients with GLILD (Fig. 6), ranging from stably mild radiographic GLILD without lung function deterioration (patients 1 and 2), to clinical GLILD that required immunosuppressive treatment shortly after sampling (patients 4 and 5). In patients 1–4, immune activation increased with severity of disease. In patient 3, who at the time of sampling had subclinical GLILD that exacerbated 3 years later, as well as patients 4 and 5, the median fluorescence intensity of CTLA4 on Tregs decreased, which was not observed for the two subclinical GLILD patients that remained stable (1 and 2). Patient 4 with STAT1 GoF, who required treatment for GLILD shortly after sampling, showed high levels of immune activation and PD1 expression, and low Treg functional markers. Finally, patient 5 with CTLA4 haploinsufficiency demonstrated low CTLA4 expression on Tregs as expected, but also low T-cell activation (HLA-DR, Ki67, and IFNγ) in peripheral blood, despite progressive GLILD and the need for treatment shortly after sampling.