Potential anti-EBV effects associated with elevated interleukin-21 levels: a case report

Epstein-Barr virus (EBV) infects 90–95% of adults worldwide. In health and in EBV lymphoproliferative disorders, memory B cells and plasma cells, collectively, constitute the largest functional EBV reservoir [1, 2]. Interestingly, the increase in EBV loads, associated with immunosuppression, can be accounted for by an increase of latently infected, germinal center (GC) derived, memory B cells [3]. Due to the immunosuppression, a fraction of (circulating) memory B cells can now complete spontaneous lytic replication and thereby increase the EBV load [3]. Interleukin-21 is a GC derived cytokine which may impact EBV persistence in several ways. Interleukin-21 was shown to induce, in an EBV⁺ cell line, which resembled post-GC B cells, increased latent membrane protein 1 (LMP1) and Epstein-Barr virus nuclear antigen (EBNA1) expression, Ig secretion and plasmablast differentiation [4], suggesting that IL-21 may support lytic EBV replication. Mice studies have documented, that interleukin (IL)-21, contributes to the formation of T follicular helper cells and thereby supports GC derived memory B cell and plasma cell differentiation [5]. However, IL-21 has potentially potent anti-EBV effects too. Firstly, animal models have documented, that effector CD8⁺ T cells, capable of killing virus infected cells, require the presence of IL-21 producing CD4⁺ T cells [6]. Secondly, in-vitro studies have shown, that IL-21 is pro-apoptotic for activated and resting B cells [7].

We provide the first documentation that elevated CD4⁺ IL-21⁺ T cell frequencies accompany human EBV disease. Furthermore, our case indicates several anti-EBV effects of sustained and increased IL-21 levels.

Due to undiagnosed progressive pulmonary infiltrates, a 67-year old lady of Danish ancestry, non-smoker, with no infectious history, was referred to our university clinic. An open lung biopsy revealed a mixed lymphocyte cell infiltrate dominated by CD3⁺ T cells and by areas with centroblastic/ immunoblastic/ Hodgkin-like CD20⁺ B cells. The proliferation marker Ki-67⁺ was extensively expressed (Fig. 1). CD30⁺ immunoblasts were also visible. EBV positive cells (more than 50 per high power field) were found by EBV RNA in-situ hybridization using a commercial (Ventana, Tucson, USA) Epstein-Barr virus small encoded RNA (EBER) probe (Fig. 1). PCR for immunoglobulin heavy-chain gene rearrangements identified a smaller population of B cells with clonal IgH gene-rearrangements placed on a polyclonal background. Clonal rearrangements of the TCR beta-chain were not found. Lung tissue CD3⁺ T cells were dominated by CD4⁺ T cells but perforin and granzyme B positive CD8⁺ T cells were also present. Our hematopathologist’s concluding diagnosis was lymphomatoid granulomatosis grade III. The EasySep Human B cell enrichment Kit (STEMCELL™ technologies), which functions optimally with fresh cells, was used on thawed cryopreserved patient and controls PBMC. Among negatively selected patient CD19⁺ B cells (suspension 1) and among the remaining cells (positively selected patient CD4⁺ T cells and CD16⁺/CD56⁺ NK-cells, suspension 2), we determined EBV copy numbers by DNA amplification using the Lightcycler 480 (Roche Diagnostics). DNA, isolated from 200 μL of B cell enriched suspension 1, which contained 46.000 cells (with 53% B cells), generated 13.500 EBV copies, resulting in an average EBV-copy number per cell = 0.29 copies. Similarly, DNA, isolated from 200 μL of B cell depleted suspension 2, which contained 436.000 cells (with 5% B cells), generated 65.900 EBV copies, resulting in an average EBV-copy number per cell = 0.15 copies. Assuming that the average EBV copy number, among the non-B cells in the B cell enriched suspension 1, was also 0.15 copies, suggested that the B cell EBV copy number was even higher than 0.29 copies. In comparison, both suspension 1 and 2, from a healthy IgG anti-VCA positive control, were EBV copy negative.

A (positron emission tomography) PET scan revealed mediastinal adenopathy with increased fluorodeoxyglucose (FDG) uptake. The hematologists found no indication for chemotherapy. The patient had circulating IgG, but no IgM, specific for (EBNA, concentration: 47.5 U/mL, cut-off: > 20 U/mL) and viral capsid antigen (VCA, concentration: 577 U/mL, cut-off: 20 U/mL). EBV plasma DNA levels were repeatedly (5x over a period of three years) moderately elevated (from < 1000 copies/ml (positive) to 3905 copies/ml). By serology and PCR, she tested negative for human immunodeficiency virus, hepatitis B and C virus, cytomegalovirus, parvovirus, varicella zoster virus, and syphilis. High titer anti-nuclear antibodies (ANA) were present (homogeneous nuclear staining, titer > 1280 (+++), normal range: 0–160) (Table 1). Auto-antibody specificities (anti- Cyclic Citrullinated Peptide (CCP), IgM rheumatoid factor, SS-A Ro52, SS-A Ro60, SS-B, Sm, ribo-nuclear protein (RNP), Scl-70 and Jo-1, Table 1) were negative. Markers for auto-immune hepatitis (soluble liver antigen (SLA) antibodies, liver cytosol (LC1) antibodies, antimitochondrial antibodies (AMA), ACTIN, liver-kidney microsomes (LKM) antibodies) were also negative and clinical signs of autoimmunity were absent. Patient 25-Hydroxy –Vitamin D3 levels were repeatedly (69 and 105 nmol/L) normal. A pan human leukocyte antigen (HLA)-I class specific antibody (e-bioscience clone W6/32) demonstrated that the patient’s leukocytes expressed HLA-class I similar to that of a control (data not shown). HLA-genotyping revealed homozygosity for HLA-A*01, 01; HLA-B*08, 08 and HLA-C*07, 07 but not for HLA-DRB1*03, 13 or HLA-DQB1*02, 06. The patient displayed prolonged elevated lactate dehydrogenase (> 255 U/L), alkaline phosphatase (> 105 U/L) and alanine transaminase (> 45 U/L) levels, consistent with hepatitis. Due to markedly reduced IgA, IgG and IgM levels (Table 1), intravenous immunoglobulin substitution was initiated. Lately, the patient has experienced a 30–40 kg weight loss, her pulmonary function is rapidly deteriorating and she constantly relies on supplemental oxygen (Fig. 2). In agreement with informed written consent and the study protocols (S-20150176 and S-20192000-48), immunologic work-up was initiated. A peripheral blood count revealed monocytosis (Table 1) and flow cytometry revealed reduced CD19⁺ B cell concentrations, reduced frequencies of CD19⁺ CD27⁺ IgD⁻ (GC derived) memory B cells, normal CD4⁺ T cell concentrations but extremely low CD8⁺ T cell concentrations (Table 1). Patient CD3⁻ CD16⁺/CD56⁺ natural killer (NK) cell concentrations (> 130 × 10⁶/L) were normal. Next generation sequencing found no variants in a targeted panel of genes associated with EBV disease: SH2D1A, PRF1, XIAP, CD27, CTPS1, RASGRP1, CD70, RLTPR, ITK, MAGT1, PRKCD, UNC13D, STX11, STXBP2, FAAP24 and CORO1A, with CVID: ICOS, TNFRSF13B (TACI), TNFSF13 (APRIL), TNFRSF13C (BAFF-R),TNFSF12 (TWEAK), CD19, CD81, CR2 (CD21), MS4A1(CD20), LRBA, CTLA4, PLCG2, NFKB1, NFKB2, PIK3CD, PIK3R1, VAV1, RAC2, BLK, IKZF1 (IKAROS) and IRF2BP2 and genes of interest: BLIMP-1 (PRDM1), BCL-6, CD8A, IL21, IL21R and STAT3. Sanger sequencing of the IL-21 gene promoter revealed no mutations. The tridecavalent pneumococcal conjugate vaccine (PCV) was used to vaccinate two age and gender matched controls and the patient. For all three subjects, this was their first PCV vaccination. Three weeks post-vaccination, our patient’s PCV titers were significantly diminished compared to age and gender matched controls (Table 1) and only the patient failed to generate protective antibody levels to some PCV serotypes: 4, 5, 6B and 18C (data not shown). After vaccination, the patient, in contrast to three controls, responded with a 5-fold increase in CD4⁺ IL-21⁺ frequencies but with no induction of CD4⁺ CD45RA⁻ CXCR5⁺ CCR7^low PD-1^high (peripheral) T follicular helper cells ((p)T_FH) (Fig. 3). CD4⁺ IL-21⁺ T cell frequency determination on a patient sample (sampled > 1 year after vaccination) and an additional non-vaccinated control sample, accentuated the difference in peripheral CD4⁺ IL-21⁺ T cell frequencies (patient vs. all controls: Mann-Whitney U -test p < 0.01). Contrary to four control sera (2 are shown in Table 1), only patient sera (2 of 4) were positive for IL-21 (30.1 and 37.8 pg/mL). Peripheral T_FH and CD4⁺ IL-21⁺ frequencies were negatively correlated among vaccinated subjects (pre-, 7 and 21 days post-vaccination for the patient + 4 controls = 15 time points, Spearman’s rho: − 0.86, p < 0.001, Fig. 4). Patient CD4⁺ IL-21⁺ T cells were predominantly CXCR5⁻ (approximately 82%, data not shown). Three days of stimulation with anti-CD3/anti-CD28/IL-2 and subsequent intracellular staining for B-cell lymphoma 6 protein (Bcl-6) revealed that patient peripheral CD4⁺ T cells tended to be less often Bcl-6 positive (n = 2 different time points, 31.5%; 26.6–36.6%) than peripheral controls CD4⁺ T cells (n = 5 controls, 65.0%; 48.4–83.7%, p = 0.05). Among ex-vivo peripheral IL-21⁺ CD4⁺ T cells, only those of the patient were dominated by a chemokine receptor profile (CCR7⁻) consistent with tissue homing (CCR7- / CCR7+ fraction: 1.3 vs. two age and gender matched healthy controls: both 0.8, Table 1). The patient’s CD4⁺ CD25^high FoxP3⁺ T_reg frequencies were reduced (0.9% of CD4⁺T cells) compared to those of 4 healthy adult controls (median: 2.5%; range: 1.7–3.1% of CD4⁺ T cells). The patient’s healthy son (23 years younger than the patient) had normal CD4⁺ and CD8⁺ T cell concentrations, positive IgG anti-EBNA but no detectable EBV copies in his blood (data not shown).

In health and in EBV lymphoproliferative diseases, GC derived memory B cells and plasma cells constitute the primary functional EBV reservoir [2, 3, 9]. Hence, by inhibiting GC derived memory B cell and subsequent plasma cell differentiation, EBV loads may be reduced. We have described the immunological effects associated with an elevated CD4⁺ IL21⁺ profile in an elderly lady suffering from chronic EBV infection where some of the clinical and immunological (severe CD8⁺ lymphopenia, B lymphopenia, hypogammaglobulinemia) findings resembled those of B cell chronic active EBV disease (B cell CAEBV) [10]. However, due to only moderately elevated EBV copy numbers, which is not a typical characteristic of B cell CAEBV [10], we decided to classify her condition, based on the hematopathologist’s diagnosis, as having high grade EBV reactivation. Based on our calculations, we cannot exclude that non-B cells contributed to the patient’s EBV burden. However, according to our calculations, her B cells contained on average at least twice as many EBV-copies than non-B-cells, consistent with the pathological findings. T- and NK- cell tropic EBV infections are rare entities primarily reported among East Asians (our patient was of Danish ancestry). Our histological examination provided no evidence for clonal T cell expansion. T and NK-cell EBV disease also primarily affects children or young adults who develop symptoms as fever, persistent hepatitis, skin symptoms, uveitis, hepatosplenomegaly or pancytopenia [11], none of which characterized our patient. Animal studies have shown, that CD4⁺ T cell produced IL-21 is crucial for anti-viral CD8⁺ T cells [6] and CD8⁺ T cells are critical for controlling EBV [12]. Hence, the elevated CD4⁺ IL-21⁺ T cell frequencies, reported here for the first time in the context of human EBV disease, could be seen as an appropriate response to her severe EBV infection. We found no mutations in the patient’s IL-21, IL-21R, IL-21 promoter or in genes related to EBV disease or CVID. We also excluded auto-immune hepatitis and hepatitis B and C as causes for the elevated patient CD4⁺ IL-21⁺ T cell frequencies. Increased IL-21 levels, likely produced by pT_FH, have been observed in type I autoimmune hepatitis [13] and chronic hepatitis B [14], however our patient’s CD4⁺ IL-21⁺ T cells were primaly non-pT_FH (CXCR5-)_. Collectively, our findings suggest that our patient’s elevated IL-21 profile was indeed secondary to her pulmonary EBV reactivation. Our patient’s IL-21 producing CD4⁺ T cells were dominated by an effector profile (CCR7⁻) and showed phenotypically no signs of exhaustion (similar PD-1 expression as controls, data not shown). Collectively, this phenotypic profile was consistent with the CD4⁺ IL-21⁺ T cells being able to home to and sustain antiviral CD8⁺ T cell function in peripheral tissues. Our patient’s CD8⁺ lymphopenia is a characteristic also commonly found (44%) among patients with B cell CAEBV [10]. We excluded that her CD8⁺ lymphopenia was secondary to HIV, bare lymphocyte syndrome type I (lack of HLA-class I molecules) [15], vitamin D deficiency [16], CD8A variants or other genetic variants associated with primary immunodeficiency. Furthermore, she had a late onset infectious history and her middle aged son had normal CD8⁺ T cell counts and no EBV reactivation. Collectively, this points to a secondary CD8⁺ T-cell penia, however one fully consistent with the effects of IL-21, since IL-21 causes accumulation of virus specific effector CD8⁺ T-cells in peripheral tissues [17]. Actually, IL-21 induces expression of the gut homing receptor integrin α₄β₇ consistent with the intestinal accumulation of CD8⁺ T cells [17]. Accumulation of CD8⁺ T cells in lymph nodes and spleen is also observed in IL-21 transgenic mice [18]. Our patient’s modest EBV copy number elevations also indicated some preserved CD4⁺ and CD8⁺ T-cell functionality in her peripheral tissue. Despite the increased intra- and extracellular IL-21 levels (which likely explained her monocytosis), the patient generated neither pT_FH nor protective protein specific antibodies, notwithstanding the central importance of IL-21 for both (p) T_FH differentiation [19] and GC derived antibody formation [5]. T follicular helper cells (T_FH) are located in the GC and are critical for the formation of GC derived memory B cells [20] and plasma cells [21]. As a proxy marker for GC located T_FH, we focused on their peripheral counterparts: (CD4⁺ CD45RA⁻ CXCR5⁺ CCR7^lo PD-1^hi) pT_FH since there are functional and developmental connections between T_FH and pT_FH, implying that peripheral expansion of pT_FH, one week post-vaccination, correlates with GC derived antibody formation [8]. Furthermore, both the differentiation of T_FH and pT_FH is dependent on the GC transcriptional repressor Bcl-6 [8]. Hence, our patient’s lack of pT_FH was consistent with her reduced peripheral CD4⁺ Bcl-6⁺ T cell frequencies.

Priming and differentiation of CD4⁺ IL-21⁺ T cells takes place in lymph nodes, consistent with the increased CD4⁺ IL-21⁺ T cell frequencies, observed seven days post-vaccination (Fig. 3). IL-21 is released within the GC, or in peripheral tissues, and is bound by local IL-21R⁺ cells (B cells, CD4⁺ and CD8⁺ T cells). This might explain why not all patient sera contained elevated IL-21 levels. While having the most pronounced CD4⁺ IL-21⁺ induction among all vaccinees, the patient had absolutely no Bcl-6 dependent pT_FH induction day 7. Substantiating a potential antagonism between Bcl-6 dependent pT_FH differentiation and the differentiation of CD4⁺IL-21⁺ T cells, we observed a strong inverse correlation between pT_FH and CD4⁺ IL-21⁺ formation, pre- and post-vaccination, in patient and controls. This could indicate that other IL-21 producing CD4⁺ T cells subsets [22], apart from pT_FH /T_FH [8], might either directly compromise the development of the latter or represent alternative GC-derived differentiation pathways. In agreement with this inverse correlation, patient CD4⁺ IL-21⁺ T cells were predominantly CXCR5⁻ and hence of a non-pT_FH phenotype. We have not been able to address whether constantly elevated IL-21 levels per se compromised pT_FH /T_FH formation. Although IL-21 is a Bcl-6 stimulator [5], IL-21 can also suppress intra-nodal Bcl-6 through signal transducer and activator of transcription 3 (STAT3) induced Blimp-1 expression [23]. It can therefore not be excluded, that the patient’s constantly elevated CD4⁺ IL-21⁺ levels could have suppressed intra-nodal, Bcl-6 dependent, pT_FH and T_FH induction [24] and thereby compromised the patient’s memory B cell formation [20] and subsequent plasmablast generation [21].

Interleukin-21 also induces apoptosis in resting and activated B-cells [7] consistent with the widespread B cell lymphopenia and hypogammaglobulinemia observed in this patient and in a large fraction (42%) of B cell CAEBV patients not treated with rituximab [10]. Collectively, our data suggests several mechanisms whereby a prolonged increased IL-21 profile might reduce EBV loads: 1) apoptosis of resting and activated B cells combined with compromised GC derived 2) memory B cell and 3) plasma cell differentiation, secondary to disruption of Bcl-6 dependent pT_FH formation. Due to the very low frequencies of circulating patient memory B cells, the estimation of EBV content in this subset as well as in plasma cells (which are bone marrow resident) is technically not feasible. However, as the B cells constituted the dominant reservoir in our patient, a reduction in memory B cells and plasma cell formation could be a possible mechanism to reduce EBV loads. In addition, our patient’s circulating CD4⁺ CD25^high FoxP3 T_regs frequency was markedly reduced compared to adult controls - a finding potentially attributable to the T_reg inhibitory effect of IL-21 [25]. As T_regs can inhibit CD8⁺ effector T cell function [26], reduced T_regs frequencies could aid CD8⁺ T effector cell efficacy against EBV but could also increase the risk of collateral tissue (lung) damage due to unconstrained CD8⁺ effector T cell activity. The latter might explain her deteriorating lung function. Consistent with the aforementioned anti-viral mechanisms, the patient had only moderately elevated EBV copy numbers. We can only speculate, as to why our patient developed high grade EBV reactivation, but her homozygosity for HLA-A* 01, a well-known risk allele for EBV+ tumors [27] could be implicated as well as clonal EBV escape due to extended HLA-class I homozygosity.

We present the first report of elevated CD4⁺ IL-21⁺ T cell frequencies in human EBV disease, thereby complementing animal models documenting that IL-21 is critical for containing chronic viral infections. Secondly, this case suggests that continuous IL-21 overproduction, apart from driving T cell mediated anti-EBV responses, may disrupt GC derived memory B cell and plasma cell formation and thereby diminish the two B cell compartments so important for EBV persistence.