|Home | About | Journals | Submit | Contact Us | Français|
Genetic polymorphisms in the IL-2Rα chain (CD25) locus are associated with several human autoimmune diseases, including multiple sclerosis (MS). Blockade of CD25 by the humanized monoclonal antibody (Ab) daclizumab decreases MS-associated inflammation, but has surprisingly limited direct inhibitory effects on activated T cells. The present study describes unexpected effects of daclizumab therapy on innate lymphoid cells (ILCs). The number of circulating RORγt+ ILCs, which include lymphoid tissue inducer (LTi) cells, was found to be elevated in untreated MS patients compared to healthy subjects. Daclizumab therapy not only decreased numbers of ILCs, but also modified their phenotype away from LTi cells and toward a natural killer (NK) cell lineage. Mechanistic studies indicated that daclizumab inhibited differentiation of LTi cells from CD34+ hematopoietic progenitor cells or c-kit+ ILCs indirectly, steering their differentiation towards immunoregulatory CD56bright NK cells through enhanced intermediate affinity IL-2 signaling. Because adult LTi cells may retain lymphoid tissue inducing capacity or stimulate adaptive immune responses, we indirectly measured intrathecal inflammation in daclizumab-treated MS patients by quantifying the cerebrospinal fluid CXCL13 and immunoglobulin G (IgG) index. Both of these inflammatory biomarkers were inhibited by daclizumab treatment. Our study indicates that innate lymphoid cells are involved in the regulation of adaptive immune responses, and their role in human autoimmunity should be investigated further, including their potential as therapeutic targets.
Daclizumab, a humanized monoclonal antibody against the alpha chain of the IL-2 receptor (IL-2Rα; CD25), promotes development of tolerance in solid organ transplantation (1) and limits target-organ inflammation in inflammatory uveitis (2) and multiple sclerosis (MS) (3–5). Daclizumab selectively blocks the low affinity (Kd = 10 nM) IL-2-binding domain on CD25, a non-signaling chain of the IL-2R. The two remaining chains of IL-2R, the beta (CD122) and gamma (CD132) chains, both have intracellular signaling motifs, and together bind IL-2 with intermediate affinity (Kd = 1 nM). Association of CD25 with the intermediate affinity IL-2R enhances the affinity of IL-2 binding 10–100-fold, resulting in a high affinity IL-2R (Kd = 10 pM). (6)
Because T cells upregulate CD25 during activation, and activated T cells are the main consumers of IL-2 through the high-affinity IL-2R, daclizumab was designed as an immunotherapy with selective inhibitory action towards activated (effector) T cells. However, our previous studies demonstrated that daclizumab has limited direct effects on activated T cells; its immunomodulatory potency resides in unexpected effects on components of the innate immune system (7–10). Interestingly, some of these effects are an indirect result of daclizumab-driven inhibition of IL-2 consumption by activated T cells and FoxP3-expressing regulatory T cells (Tregs), resulting in greater availability of IL-2 for signaling by cells that express high levels of intermediate affinity IL-2R. (11)
Despite the prominent inhibition of Tregs (11, 12), MS-related inflammation is significantly inhibited by daclizumab (3, 4, 13). We have previously described two mechanisms that can explain this apparent paradox: 1) CD56bright NK cells, which are expanded (7) and activated (11) by daclizumab therapy, have overlapping immunoregulatory functions with FoxP3 Tregs (9) and 2) daclizumab also inhibits antigen-specific priming of effector T cells by blocking trans-presentation of IL-2 by mature dendritic cells (mDC) (10). The current paper describes a third mechanism of how daclizumab inhibits MS-related inflammation.
While investigating effects of CD25 blockade on mDCs (10), we observed a significantly decreased population of lineage negative lymphocytic cells lacking expression of typical DC markers in the daclizumab-treated cohort. A subsequent review of the literature indicated that these cells likely represent innate lymphoid cells (ILCs) (14, 15).
Three major categories of ILCs have been identified (14–16): 1. NK cells (also called ILC1 cells), which in humans are phenotypically subdivided into CD56bright and CD56dim subsets; 2. RORγt+ ILCs, which include fetal and adult lymphoid tissue inducer (LTi) cells, IL-22 producing ILCs (ILC22, which express NKp44) and IL-17 producing ILCs (ILC17; it is unclear whether these are distinct from LTi or ILC22 cells); and 3. Type 2 ILCs (ILC2 or nuocytes), which are independent of RORγt and secrete Th2-type cytokines, such as IL-5 and IL-13. Although this categorization is conceptually useful, it remains uncertain whether sub-groups of ILCs represent truly distinct lineages or developmentally related and plastic phenotypes.
Multiple ILC subsets originate from CD34+ hematopoietic precursors in an Id2-dependent manner (15, 16), because Id2-deficient mice have greatly diminished levels of NK cells and LTi cells (17). Despite the fact that multiple studies addressed developmental relationships between different ILC subsets (14, 18–23), a unifying concept is still missing. Although methodological differences, especially the origin and the exact phenotype of LTi cells may explain apparent discrepancies between published studies, we conclude that prevailing evidence implies existence of a common precursor at the hematopoetic stem cell level (24) and probably also of a less differentiated ILC that responds to different environmental cues to generate both LTi cells and NK cells. IL-2/IL-15 represents at least one environmental stimulus that drives differentiation toward NK cell lineage (18–20, 25).
Although fetal LTi cells play a vital role in the development of secondary lymphoid tissues (26), the role of adult LTi cells has been unclear. Although LTi cells retain lymphoid-tissue inducing capacity postnatally (at least in the gastrointestinal tract (26–30)), tertiary lymphoid follicles can form in RORγt-deficient animals, which lack LTi cells (31, 32). Both T cells (33, 34) and B cells (31) acquire lymphoid-tissue inducing capacity in different inflammatory animal models. Nevertheless, adult LTi cells may play an important role in the evolution of T cell memory and CD4-dependent high affinity Ab responses by providing OX40 and CD30 signals (35, 36), raising the possibility that these cells may participate in the development of autoimmunity. However, to our knowledge the role of LTi cells in human autoimmune diseases has not been investigated thus far. Our unforeseen observation that daclizumab therapy affects levels of ILCs prompted us to investigate the role of these cells in the MS disease process.
We used Influenza vaccination (Flulaval™) as a tool to assess daclizumab-induced changes in the activation of the immune system in vivo by standardized antigenic stimulation. Using ex vivo flow cytometry, we compared changes in the phenotype of immune cells induced by influenza vaccination in MS patients under long-term daclizumab therapy with age/gender matched controls. CD56bright NK cells were significantly expanded in daclizumab-treated patients. ILCs – lineage negative (CD3, -11c, -14, -19, -56, -123) lymphocytic cells were observed in significantly lower proportions in daclizumab-treated patients before and further decreased 7 days after Flu vaccination, whereas no vaccination-induced change was observed in controls (Fig. 1A). We confirmed that the decrease in circulating ILCs was daclizumab-driven in two additional cohorts: First, in matched cryopreserved samples from a published Phase II clinical trial of daclizumab in MS (Clinicaltrials.gov identifier NCT00071838 (13)) collected before- and 3- and 8-months after initiation of daclizumab treatment (Fig. 1B); and second, in the ongoing Phase I/II trial of daclizumab high-yield process (DAC-HYP; Clinicaltrials.gov identifier NCT01143441) that focuses on defining the full mechanism of action of daclizumab in MS. In this last cohort, fresh ex vivo samples were directly compared with cryopreserved samples (Fig. 1C). There were no significant differences in the numbers of ILCs, but cryopreservation significantly decreased RORγt and c-kit expression (fig. S1). Consequently, we used fresh blood samples from the ongoing trial to better characterize the effects of daclizumab therapy on ILCs (Fig. 1D).
We confirmed that a large portion of ILCs represent LTi cells in the fresh blood of healthy donors (HD) as determined by constitutive expression of c-kit, RORγt and lymphotoxin-α (LTα; Fig. 1D). However, we also observed expression of NK lineage markers such as CD122, CD161 and CD7 on ILCs. In order to better assess what proportion of ILCs represent true LTi cells, as compared to ILC22 cells, we stained ILCs and NK cells for NKp44 (marker of ILC22), CD25 (expressed on LTi cells) and two TNF-super family receptor ligands, which have been previously shown to be expressed on virtually all (OX40L; (37)) or a proportion (CD30L; (35)) of human LTi cells (Fig. 1E). We observed that 52.1% of c-kit+ lineage negative lymphoid cells in fresh blood samples expressed OX40L (range 26.4–70.4%), whereas NK cells lacked expression of this marker. CD30L was expressed on 26.4% of ILCs (range 3.3–40.1%). CD25 was expressed on 42.2% of ILCs (range 13.8–83.7%) and as reported previously, was also expressed in a proportion of CD56bright NK cells (15.0%; range 1.9–28.9%). In contrast, NKp44 was expressed only on 13.4% of blood ILCs (range 0.0–31.0%) and 11.0% of CD56bright NK cells (range 0.1–23.2%).
When we analyzed the numbers of c-kit+/RORγt+ ILCs (normalized to 1000 T cells), we observed that untreated MS patients had significantly elevated numbers of these proinflammatory ILCs whereas daclizumab-treated MS patients had comparable levels of c-kit+/RORγt+ cells to those observed in HD (Fig. 1F&G).
Next we asked whether daclizumab therapy solely inhibits the absolute number of ILCs, or whether it also affects their phenotype. Daclizumab-treated MS patients compared to matched untreated MS controls had significantly decreased proportion of ILCs that expressed LTα, IL-22 and TNFα (Fig. 2A) and concomitantly increased proportions of ILCs that expressed NK cell markers CD122, CD161 and CD7 (Fig. 2B). In contrast, we observed no significant changes in expression of other surface markers, including OX40L and CD30L (fig. S2).
Furthermore, we observed a significant correlation between contraction of ILCs and expansion of CD56bright NK cells in all daclizumab-treated patients (r = 0.447, p<0.001) and between contraction of c-kit+/RORγt+ LTi cells and expansion of CD56bright NK cells in a smaller DAC-HYP patient cohort (r = 0.520, p =0.046).
Based on the observation that ILCs in daclizumab-treated patients expressed significantly higher levels of CD122, we tested the hypothesis that signaling through the shared IL-2/IL-15 intermediate affinity receptor on ILCs is also enhanced. In comparison to untreated controls, ILCs from daclizumab-treated MS patients had 2–3 fold higher Stat5 phosphorylation to IL-2 and IL-15, but not to IL-7 (Fig. 3A&B).
Though Cupedo et al. (23) used IL-7 as the common γ-chain (γc)-signaling cytokine to support differentiation of ILCs toward an NK cell lineage in vitro, based on our combined ex vivo/in vitro observations summarized above, we asked whether intermediate affinity IL-2/IL-15 signaling could also promote differentiation of lineage-negative, c-kit+ ILCs toward a NK cell lineage (e.g. into CD56bright NK cells). To examine this, we cultured >99% pure lineage negative ILCs with high expression of c-kit (fig. S3; purified from fresh aphaeresis samples by negative selection or cell-sorting), in the presence of stem cell factor (SCF) and FMS-like tyrosine kinase 3 ligand (Flt3L) with either no additional cytokine or with IL-2, IL-7 or IL-15 for 7 days. Both IL-2 and IL-15, but not IL-7, significantly enhanced differentiation of ILCs towards CD56bright NK cells (Fig. 3C&D).
We next sought to determine whether these NK cell populations were functional, as evidenced by their expression of cytolytic enzymes and their ability to kill MHC-I-deficient targets. We observed that IL-2/IL-15-expanded CD56bright NK cells from ILC cultures expressed comparable or higher levels of perforin and granzymes A and B as CD56bright NK cells in peripheral blood (Fig. 3E). Finally, we performed killing assays to determine the functional status of NK cells derived from ILCs and observed almost 3-fold higher cytotoxicity in IL-2- and IL-15-cultured NK cells as compared to IL-7-cultured NK cells (Fig. 3F).
Previous reports by Freud et al. (24) demonstrated that the precursor for human CD56bright NK cells is a subpopulation of CD34+ hematopoietic progenitor cells (HPC) residing in the lymph nodes (or isolated from the blood, but not bone marrow) and that the differentiation process is dependent on IL-2 or IL-15 signaling. Therefore, we studied a unifying hypothesis that LTi cells and CD56bright NK cells share CD34+ HPCs and that intermediate affinity IL-2R signaling is the main driving force pushing differentiation away from LTi cells and toward CD56bright NK cells. To explore this, we isolated CD34+ HPC (>90 % purity; Fig. 4A) and cultured them with SCF and Flt3L in the presence or absence of IL-2. Without IL-2, CD34+ HPC differentiated into LTi cells (indicated by constitutive expression of LTα) but not to CD56bright NK cells (Fig. 4B, upper panels). In contrast, the presence of IL-2 inhibited differentiation to LTi lineage and instead produced high proportions of CD56bright NK cells (Fig. 4B, lower panels).
In order to determine whether inhibition of LTi cell development by daclizumab is clinically meaningful, we indirectly measured the functions of intrathecal inflammation by measuring CSF levels of CXCL13 (the chemokine linked to lymphoid neogenesis (38–41)) and IgG index (validated measure of intrathecal IgG production). After 6.5 months of daclizumab treatment, we observed that CXCL13 levels decreased by 50.4% (from an average of 35.92 pg/ml to 17.83 pg/ml; p = 0.008; Fig. 5A) and the IgG index decreased by 13.5% (from an average of 1.008 to 0.872; p = 0.003; Fig. 5B).
The goal of our study was to investigate in detail effects of daclizumab therapy on ILC subtypes, the mechanisms behind these effects, as well as to explore potential role of ILC subsets in the MS disease process. We observed that although untreated MS patients have significantly elevated levels of pro-inflammatory ILCs, defined as c-kit+/RORγt+ lineage negative cells, which are known to contain LTi cells, daclizumab therapy restores ILC numbers to levels comparable to those observed in healthy subjects. Furthermore, daclizumab therapy skews the phenotype of ILCs away from the LTi lineage and toward the NK cell lineage. We were able to reproduce these in vivo effects by adding IL-2 (or IL-15) to in vitro differentiation assays of CD34+ HPC or c-kit+ ILCs: In both instances, addition of IL-2 drove differentiation of ILCs away from the LTi lineage and toward the NK cell lineage, in agreement with published animal studies (18–20, 25). We conclude that CD25 blockade alters differentiation of ILCs by paradoxically enhancing IL-2 signaling on common precursor(s) that can differentiate to either LTi cells or CD56bright NK cells, depending on environmental cues. By blocking high affinity IL-2 signaling on T cells, daclizumab increases the in vivo availability of IL-2 for cells that are capable of signaling via the intermediate affinity IL-2R (11), such as CD56bright NK cells (7) and subtypes of ILCs.
The crucial question is whether the phenotypic switch of ILCs from LTi to NK cell lineage induced by daclizumab inhibits aberrant adaptive immune responses associated with autoimmunity. Utilizing an open-label cross-over trial methodology, we previously demonstrated that daclizumab, either in combination with IFN-β (3) or as a monotherapy (8, 13) significantly inhibits formation of brain inflammatory lesions in MS and stabilizes neurological disability. These findings have now been fully reproduced in an independent double-blind, placebo-controlled clinical trial (4).
Although the current study is the first direct evidence linking adult LTi cells with autoimmunity in humans, by facilitating OX40 and CD30 signaling on T cells, LTi cells may play an important role in promoting the development of CD4 T cell memory (28, 35) and production of CD4 T cell-dependent high-affinity Ab (36). Both of these advanced functions of adaptive immunity are linked to secondary lymphoid tissues or organized lymphoid aggregates (such as “ectopic” or “tertiary” lymphoid follicles) that have been observed in many autoimmune diseases (26), including MS (42, 43). Indeed, the hallmark of MS is intrathecal production of IgG, measured clinically as IgG index, with characteristics of an antigen-driven affinity maturation process (44, 45). Formation of high affinity IgG against protein antigens is dependent on CD4 T cell help and occurs only in germinal centers of lymphoid organs. Persistent presence of oligoclonal bands (expanded IgG clonotypes detected only in the CSF but not in the serum), presence of all stages of B cell differentiation in the CSF of MS patients (46), and the recent discovery of CXCL13-rich tertiary lymphoid follicles in the meninges of some MS patients (42, 43) collectively imply that the intrathecal production of IgG in MS is facilitated and/or sustained within meningeal lymphoid aggregates.
Hence, we searched for in vivo evidence that would link our observation of daclizumab-driven inhibition of LTi cells to meningeal inflammation in MS. There is an abundance of literature linking ectopic lymphoid follicles in a variety of human autoimmune diseases with CXCL13 expression in the affected tissue (39) and elegant animal studies linked CXCL13 mechanistically to lymphoid neogenesis (40, 41). We observed that CSF CXCL13 levels decreased by 50.4% after 6.5 months of daclizumab therapy. Furthermore, this was associated with 13.5% decrease in the intrathecal production of IgG, measured as IgG index. This effect on IgG production was specific for the intrathecal compartment, as we have previously reported that daclizumab therapy has no effect on systemic production of immunoglobulins (3). To our knowledge, daclizumab is the first immunomodulatory drug that decreases levels of intrathecally produced IgG in MS, and it is doing so without depleting or limiting access of immune cells to the intrathecal compartment (13). Furthermore, because meningeal lymphoid follicles have been associated with neuronal loss in the underlying brain tissue (43, 47), inhibition of meningeal inflammation would be expected to slow down accelerated development of gray matter atrophy observed in MS. Indeed, our pilot data comparing 27 MS patients on long-term daclizumab therapy (total of 982 MRI scans) to a matched cohort of 44 MS patients treated with first line FDA-approved therapies (mainly interferons) indicate that daclizumab inhibits atrophy of brain structures that have direct contact with meninges/CSF by 38–60% in comparison to standard treatments (48). Although these encouraging observations need to be reproduced in large controlled studies, our data suggest that inhibition of LTi cells may be a highly efficacious therapeutic modality for MS.
There are limitations of the current study: Although our in vivo observations and in vitro mechanistic studies reinforce the evidence for a developmental link between LTi cells and NK cells through regulation of a common precursor, we have not performed detailed investigation and single cell cloning experiments that would define the precise phenotype of this precursor. Similarly, because the visualization of meningeal lymphoid follicles or even direct quantification of diffuse meningeal inflammation is currently not possible in living human subjects, our study provides only an indirect, albeit plausible, link between LTi cells and intrathecal inflammation in MS. Clearly, follow-up studies evaluating ILCs in autoimmune diseases amenable to biopsies of the targeted tissue, or clinical trials of novel therapies that target specifically pro-inflammatory ILCs will be necessary to establish a definite pathogenic role of LTi cells in human autoimmunity. We hope the current study will spark the necessary interest in these innate lymphoid effectors, which will lead to a better understanding of their role in human health and disease. Such understanding is the necessary prerequisite for development of targeted therapies.
Daclizumab-treated MS patient data was obtained from patients participating in NINDS IRB-approved clinical protocols, and all patients signed informed consent. Five cohorts of patients and controls were utilized in the current study and described in Supplementary Materials.
The surface and intracellular flourochrome-conjugated antibodies from BD Bioscience, eBioscience or R&D systems were used and described in Supplementary Materials.
Immunophenotyping of ILCs was performed on lysed whole blood within 30 minutes of ex vivo collection and stained according to a previously established protocol (7). Minimum of 1×106 cells were stained in order to acquire a minimum of 1000 gated ILCs. Gating for all intracellular and cell surface markers was based on isotype controls. Data were acquired on an LSR II with HTS delivery system and analyzed with FACSDiva 6.1(BD).
ILCs were isolated from fresh aphaeresis samples. PBMCs were isolated by Ficoll gradient and ILCs for each experiment were isolated from 2×108 PBMCs by negative selection using a combination of the lineage cell depletion kit (Miltenyi Biotec; 130-092-211) and anti-CD34 microbeads (Miltenyi Biotec; 130-046-702). Purity was verified by flow cytometry after isolation (fig. S2A). Purified ILCs were cultured in X-vivo-15 (Lonza) medium in the presence of stem cell factor (SCF) and Flt3L (both 10ng/ml; R&D Systems and Peprotech respectively) for 7 days with IL-2 (100 IU/ml), IL-7 (10ng/ml) or IL-15 (10ng/ml) as indicated (R&D Systems). Cultured ILCs were characterized by surface and intracellular flow cytometry for NK lineage markers after 0–7 days of cell culture.
ILCs isolated from fresh PBMC samples were stimulated for 10 minutes at 37 °C with exogenous IL-2 (100 IU/ml), IL-15 (10ng/ml) or IL-7 (10ng/ml) and immediately formaldehyde fixed and stained cells for phosphorylated STAT5 or corresponding isotype according to the manufacturer protocol (BD Bioscience).
The cytotoxicity of NK cells differentiated from ILC cultures toward MHC-I-deficient targets was assessed by a flow-cytometry based killing assay using GFP-tagged K562 cells. Targets were seeded with or without NK cells differentiated from ILCs at day 7 at a 1:1 effector:target ratio overnight. Cultures were then collected and the absolute number of live K562 cells was proportionally enumerated between conditions by flow cytometry, using equal amounts of fluorescent beads added to each condition before analysis as a reference. Percentage of K562 cell killing was calculated using the following formula: % killing = 100 − ((#K562 with NK/#K562control)*100).
CD34+ HPCs were purified from fresh PBMCs using magnetic-labeled CD34+ beads (Miltenyi Biotec;130-046-702) after lineage cell depletion and immediately cultured in X-vivo supplemented with FBS 20%, SCF and Flt3L (both 10ng/ml) +/− IL-2 (100IU/ml) for 14 days. CD34+ HPC were analyzed by flow cytometry at day 0, 7, 10 and 14 for differentiation into c-kit+ LTα+ LTi or CD56+ NK cells.
CSF supernatant from 16 patients enrolled in the daclizumab clinical trial (NCT00071838) was collected and frozen at −80°C. Undiluted CSF was measured for CXCL13 by a DuoSet ELISA Development Kit (R&D Systems) according to the manufacturer’s instructions. The detection limit of the assay was determined to be 6.25 pg/ml.
IgG index was determined by immunonephelometric method using an automated analyzer (BN II, Siemens Healthcare Diagnostics).
Group differences were analyzed using one-way ordinary or repeated measures analysis of variance (ANOVA) followed by Tukey’s multiple comparisons test. For comparisons of two groups, student’s independent-samples t tests for analysis of independent groups or paired t test for analysis of repeated measurements within identical groups of patients. When data were not normally distributed, nonparametric tests were used. Data are presented as mean ± SEM using the GraphPad Prism 5.0 or SigmaPlot 11.0 program and p < 0.05 was considered significant. In figures, values of p are shown as follows: *p < 0.05; **p < 0.01;***p < 0.001.
We thank Azita Kashani for aphaeresis processing, the NIB nursing staff and clinicians for patient care and coordination of aphaeresis and blood collection, and Jayne Martin for technical assistance. Above all, we are grateful to all patients participating in NIB protocols whose samples contributed to this study.
Funding: This research was supported by the Intramural Research Program of the NINDS and NIH Clinical Center.
Author contributions: B.B. developed the concept of the study and supervised the project. J.S.A.P., S.H. and B.B. designed the experiments. J.S.A.P., S.H., Q.X., M.L.H., L.B.K., G.C., B.B. performed the experiments and analyzed the data. J.S.A.P., S.H. and B.B. wrote the paper.
Competing interests: B.B. is a co-inventor on NIH patents related to daclizumab therapy and as such has received royalty payments based upon license to US patents 7,575,742 and 7,258,859 (and patent applications claiming priority to these patents) and other National Stage patents and patent applications claiming priority to PCT/US2002/038290 or PCT/US2003/020428. The other authors declare no competing interests.