Background
Multiple sclerosis (MS) is a chronic immune-mediated central nervous system (CNS) disorder characterized by inflammatory demyelination, neuro-axonal degeneration and reactive astrogliosis [
1]. MS affects approximately 2.8 million people worldwide, with a ~ 3 times higher prevalence in women compared to men [
2]. The disease is typically diagnosed in adults aged 20–40 years. The diagnosis of MS relies on the integration of clinical, imaging and laboratory findings [
3]. Disease activity and progression are defined by reversible episodes of new or worsening neurological deficits (also known as relapses), lesion activity in magnetic resonance imaging (MRI) and accumulation of disability over time. The spectrum of MS phenotypes is categorized into relapsing–remitting MS (RRMS), primary progressive MS (PPMS) and secondary progressive (SPMS) [
4]. In most patients (~ 85–90%), acute relapses and periods of stability characterize the first years of the disease (RRMS) before a gradual worsening of clinical disability becomes prominent (SPMS). A minority of patients (~ 10–15%) have a progressive disease course from onset (PPMS). The etiology of MS remains unclear, but various genetic, environmental and lifestyle factors are known to contribute to disease development and severity [
5,
6].
No curative treatment is yet available for MS. Disease-modifying therapies (DMTs) for MS, especially for RRMS, can favorably change the quality of life and long-term outlook for many patients [
7]. They act by suppression or modulation of immune function to reduce the rate and severity of relapses, prevent lesion formation and delay the accumulation of permanent disability [
8,
9]. DMTs for MS can be categorized into ongoing or maintenance therapies that are continuously administered (e.g., fingolimod, glatiramer acetate and natalizumab) and newer so-called pulsed immune reconstitution therapies (IRTs) that are administered in short courses and have the potential to induce long-term drug-free disease remission [
10,
11]. The concept of IRTs is to eliminate a pathogenic adaptive immune repertoire through intense short-term immune cell depletion and then allow the immune system to renew itself. Examples of pulsed IRTs for MS are alemtuzumab and cladribine. Alemtuzumab is a humanized monoclonal antibody against CD52, which is highly expressed on B and T cells [
12‐
14]. The therapy with alemtuzumab leads to a rapid and long-lasting depletion of CD52
+ cells, followed by a slow repopulation arising from hematopoietic precursor cells. Alemtuzumab is infused for 5 consecutive days in the first course and for 3 days in the second course 1 year later, though up to 2 additional treatment courses may be considered as needed. Cladribine is a chlorinated analogue of deoxyadenosine that is activated through phosphorylation preferentially in lymphocytes. Activated cladribine interferes with DNA synthesis and repair and triggers apoptosis [
15,
16]. Cladribine tablets are administered over 4–5 consecutive days at months 0 and 1 (first year of treatment) and at months 12 and 13 (second year of treatment). Ocrelizumab, an anti-CD20 monoclonal antibody, is currently the only approved DMT for PPMS [
17,
18]. As the therapeutic options in PPMS are limited, repeated pulse therapy with corticosteroids has occasionally been used [
19,
20]. However, patient selection, risk stratification and therapy guidance are challenging in the context of IRTs. Thus, deeper insights into their mechanisms of action and reliable markers to identify patients with suboptimal treatment response and to inform physicians whether to retreat or to switch therapy are urgently needed.
B cells play a key role in the pathobiology of MS, which is underscored by the high efficacy of therapeutic strategies that target these cells [
7,
8] and the fact that they are a primary target of Epstein–Barr virus (EBV) infection, the leading risk factor for MS [
21‐
23]. B cells are thought to contribute to MS through their antigen-presenting function and the formation of tertiary lymphoid-like structures in the CNS, which are the likely source of an abnormal immunoglobulin production detectable in the cerebrospinal fluid (CSF) [
24,
25]. Circulating B cells from individuals with MS are potent activators of autoreactive T cells as they exhibit an imbalance in the secretion of pro- and anti-inflammatory cytokines and express increased levels of co-stimulatory molecules, such as CD80 [
26,
27]. Within the B-cell population, memory B cells may be a driver subset in MS, as indicated by genetic studies [
28] and studies of cell population shifts in the peripheral blood in response to DMTs [
29]. The therapy with alemtuzumab leads to an effective depletion of circulating CD19
+ B cells by ~ 90%, which is followed by repopulation, with total B-cell counts returning to baseline at 3 months and then rising further to ~ 165% of baseline at 12 months after treatment [
30]. However, the apparent increase in the number of CD19
+ B cells is generated by newly produced immature (transitional) B cells and mature naive B cells, whereas CD27
+ memory B-cell recovery is slow, reaching only ~ 25% of baseline by month 12 [
30,
31]. Similarly, cladribine induces a 70–90% depletion of B cells [
32]. Within 1 year after the first administration of cladribine (i.e., until the second treatment course), the B cells repopulate to 70–80% of baseline, but the number of memory B cells remains persistently low for over 12 months (~ 80% reduction compared to baseline) [
33‐
36]. These changes in the immune reconstitution phase are associated with a decrease in pro-inflammatory responses [
11,
37]. However, previous studies could not detect significant differences in the recovery of any blood cell population between patients with and without recurrent disease activity after treatment [
32,
38‐
41]. Moreover, the effects of these IRTs for MS on the transcriptome profile of B cells have not been investigated so far.
In this study, we analyzed B cells from the peripheral blood of MS patients and healthy controls at the cellular and transcriptome level. We show that the gene expression pattern of B cells is substantially altered in patients receiving alemtuzumab or cladribine, which reflects the shift in B-cell subsets in response to these IRTs and affects a wide variety of biological functions. We also compared the transcriptome data between patients remaining free of relapses and patients who developed a relapse in the year following alemtuzumab administration to find potential biomarkers of treatment outcome. For selected genes, the possible involvement in immune mechanisms related to MS will be discussed.
Discussion
Over the past years, evidence has accumulated that B cells and their interplay with T cells are central in the pathogenesis of MS [
25]. Our understanding of this complex disease has considerably advanced with the success of therapies that mediate the depletion or functional inhibition of immune cells [
29]. Previous studies have characterized the shifts at the cellular level that occur during the treatment with pulsed IRTs and anti-CD20 agents [
59]. Both alemtuzumab and cladribine induce a rapid depletion of lymphocytes, after which memory B cells repopulate only slowly and thus are persistently depleted in the blood of patients with MS [
30,
31,
33‐
36]. This is thought to reduce B-cell trafficking from the periphery to the CNS, antigen presentation to T cells, pro-inflammatory cytokine production and the generation of antibody-secreting cells [
8]. Here, we utilized a transcriptomics approach to obtain more detailed insights on the therapeutic effects at the molecular level. We explored the biological processes that are influenced as a consequence of the gene expression alterations following the administration of IRTs and filtered biomarker candidates of the effectiveness of alemtuzumab treatment in preventing relapses.
Our analysis was based on 121 blood samples that were collected from 91 subjects and divided into 6 study groups. We included a healthy group, a PPMS group and four RRMS subgroups (before IRT, alemtuzumab, cladribine and natalizumab). The patient cohort was typical for this disease in terms of age, sex and degree of disability [
42], but we did not include patients with SPMS, patients who were therapy-naive and patients treated with other DMTs, such as ocrelizumab. Moreover, older individuals were underrepresented in the healthy group, which may have impacted the results. However, age-related changes in the proportions of B-cell subsets are most pronounced in the first 5 years of life [
46], whereas in adults, two reference studies generally found no statistically significant change with age, even though a marked decrease in CD27
+IgD
+ B cells and plasmablasts was observed [
60,
61]. Of note, the patients who received alemtuzumab would have met the inclusion criteria of the respective phase III clinical trials in terms of age, EDSS score, course of MS and number of relapses in the pre-treatment phase. The patients from whom we obtained a B1 sample before starting cladribine therapy, however, would not have met the criterion of having at least one relapse in the previous 12 months. This resembles the finding that the number of relapses before treatment is the most frequent clinical trial criterion that is not fulfilled in routine clinical care [
62]. The therapies with alemtuzumab and cladribine are referred to as pulsed IRTs as they induce a partial immune reset to achieve long-term drug-free remission of disease activity, which is a different concept compared to therapies that need to be given continuously to maintain their therapeutic efficacy [
10]. However, despite the high efficacy of IRTs in the relative reduction in relapse risk [
63], relapses still occur in some patients following treatment with alemtuzumab [
64] or cladribine [
65]. Some patients thus require retreatment with alemtuzumab. In our collection, we had samples from 4 patients who received a 3
rd or 4
th course of alemtuzumab, because they had 1 or 2 relapses in the past year. For practical reasons, we included patients before and after different treatment courses of alemtuzumab and cladribine. A longitudinal blood collection for each patient from the beginning of IRT and across multiple timepoints in the subsequent years would have been more appropriate but difficult to implement.
Our study focused on B cells as they are major contributors to the immune responses involved in MS [
25]. In addition to the transcriptome profiling, we used flow cytometry to characterize the B cells from the peripheral blood. When we compared the healthy controls with the RRMS patients before IRT, we could find no difference in the frequencies of the distinct B-cell subsets (Tukey test
p > 0.05). However, substantial B-cell subpopulation shifts were apparent in the treatment groups. The patients who received a pulsed IRT showed significantly higher proportions of transitional and naive B cells and much lower proportions of memory B cells, which is consistent with earlier studies on the effects of alemtuzumab [
30,
31] and cladribine [
33‐
36]. In contrast, the therapy with natalizumab, an antibody to α4 integrins, leads to a preferential expansion of the memory B-cell pool, which is attributable to a decreased retention of these cells within secondary lymphoid tissues [
66‐
68]. At the same time, the proportion of CD21
−/lowCD38
−/low B cells, which have been shown to be enriched with autoreactive unresponsive clones in some autoimmune diseases [
47,
69], was significantly lower in patients on IRTs but significantly higher in patients on natalizumab therapy. Further research on CNS-resident and antigen-specific B cells may provide deeper insights into the therapeutic mechanisms of action. Besides, IRTs for MS also have effects on T cells and to a lesser extent on circulating cells of the innate immune system [
9,
30‐
34,
70], which also deserve to be explored in more detail at the cellular and transcriptome level.
To our knowledge, the B-cell transcriptomes of MS patients undergoing IRTs were measured for the first time in our study. This was done using Clariom D arrays, which were introduced in 2016 as successor of previous high-density microarray solutions [
71]. These arrays are highly reproducible in estimating gene and exon levels, and they allow to detect even small variations in expression, especially for low-abundant transcripts [
72]. However, as a limitation, they offer a lower dynamic range than RNA sequencing and cannot provide insights into the expression of single cells. In comparison of the 6 study groups, a total of 6,280 DEGs resulted after FDR correction, and we took a closer look at the top 500 DEGs. Remarkably, except for the few genes in cluster 5 and cluster 6, the expression profiles were relatively similar between healthy subjects, PPMS patients (who were treated) and RRMS patients before IRT (who just discontinued another DMT). However, strong and opposite transcriptome alterations were observed for the IRT groups and the natalizumab group. These gene expression differences are essentially a consequence of the treatment-related shifts in B-cell subsets. Following IRT (i.e., after B-cell depletion and B-cell repopulation), the majority of the DEGs were reduced in expression, while cluster 8 genes were expressed at much higher levels and clearly related to the expression signature of naive B cells. Differences between the alemtuzumab group and the cladribine group were rather confined to the expression of cluster 1 and cluster 3 genes. The analysis of the paired samples revealed that the transcriptome changes in response to alemtuzumab primarily occurred after the first treatment course, whereas there were smaller effects on gene expression after the second and third annual course. We suspect that the response to cladribine is also strongest after the first course, but we could not verify this because of the variable timing of blood withdrawals and the small number of samples in this group.
Our data show that the B-cell composition that reconstitutes following IRT is functionally different from that before IRT and from that of MS patients on other therapies. Among the top 500 DEGs, there were several genes that are involved in the activation of lymphocytes. For instance,
CR2 (from cluster 8), which encodes CD21, a cell surface receptor for complement C3 and for EBV on human B cells [
73], was significantly higher expressed in patients treated with alemtuzumab or cladribine. In these patients, we also measured lower mRNA levels of
FCGR2B (cluster 4), which encodes a receptor for the Fc region of immunoglobulin gamma complexes that inhibits B-cell receptor (BCR) signaling and antibody production [
74]. Moreover, in those patients who received an IRT, we observed an increased expression of
CD1A (cluster 8) and a reduced expression of
CD80 (cluster 4), which encode membrane proteins that play a role in T-cell activation by B cells and other immune cells [
75,
76]. Transcripts for the cytokine receptors IL10RA (cluster 2) and IL21R (cluster 8) were also found to be differentially expressed between the study groups. IL10RA, which appeared to be expressed at lower levels in B cells of MS patients compared to healthy controls as previously reported [
77], mediates the immunosuppressive signal of IL10 by inhibiting the expression of pro-inflammatory genes [
78]. IL21R, which is predominantly expressed by naive B cells and was thus increased in expression in patients treated with an IRT, transduces the signal of IL21, which is produced by T-cell subsets and regulates the proliferation and differentiation of B cells and antibody responses [
79]. Serum levels of IL21 have been proposed as biomarker for the risk of developing secondary autoimmunity following alemtuzumab treatment [
80]. However, we could not study this issue on the basis of our data. Other DEGs were found to regulate apoptotic processes. For instance, CALM2 (from cluster 2) is an intracellular calcium-binding protein involved in cell death upon BCR stimulation [
81], RIPK2 (cluster 2) is a serine/threonine protein kinase suppressing apoptosis by regulating nuclear factor κB signaling [
82], and IGF1R (cluster 3) is a receptor with tyrosine kinase activity that is known to mediate anti-apoptotic effects via the PI3K/AKT pathway [
83]. We observed the lowest average mRNA expression of these genes in the cladribine group (
CALM2 and
RIPK2) and the alemtuzumab group (
IGF1R), respectively. Cluster 1, in turn, was significantly associated with the GO term "RNA processing", because it contains protein-coding and non-coding genes that promote the splicing of pre-mRNAs (e.g.,
BCAS2 and
RNU6-1) [
84] and the biogenesis of transfer RNAs and ribosomal RNAs (e.g.,
RMRP,
RPPH1 and small nucleolar RNAs) [
85‐
87]. In addition, genes encoding ribosomal proteins belong to this cluster. In the MS patient subgroups, we also detected an increased expression of genes regulating cell contact and adhesion (e.g.,
PCDH9,
PDLIM1 and
PTPRK from cluster 6) [
88‐
90]. Among the most highly connected genes in the interaction network were
ESR2 and
PHB (cluster 3, low in cladribine group) and
RC3H1 (cluster 8, low in natalizumab group). The estrogen receptor ESR2 and the ubiquitously expressed protein PHB are regulators of transcription [
91,
92]. Furthermore, PHB is involved in CD86 signaling in B cells [
93] and in the correct folding of mitochondrial proteins [
94]. RC3H1 is a post-transcriptional repressor of mRNAs (e.g.,
IL6 mRNA) [
95] and also regulates the decay of microRNAs (e.g., miR-146a) [
96]. Of note, we have focused here on the top 500 DEGs, even though the expression shifts under IRT were much broader. In a recent study, Moser et al. reported reduced proportions of CD19
+ B cells with CD44, ITGA4, ITGAL, ITGB1 and HLA-DR surface expression at 24 months after the initiation of cladribine therapy [
97]. In our analysis, those genes were not among the top 500 DEGs, but they were differentially expressed with FDR < 0.05 and all of them had the lowest average expression in the cladribine group. Thus, our data confirm their results from flow cytometry measurements at the transcript level.
We used the B-cell transcriptome profiles to search biomarkers for identifying patients with active disease following the administration of alemtuzumab. Although no gene remained significant after adjustment for multiple testing, our stringent selection resulted in 17 genes whose expression differed substantially when comparing patients with relapse and patients without relapse in the year after the 1st or 2nd alemtuzumab treatment course. This analysis was limited by the small number of patients per group. Nevertheless, we consider the genes to be reasonable candidates for further confirmatory studies at the RNA or protein level. For instance, in relapse-free patients,
BCL2 was in most cases decreased in expression at the follow-up timepoints, whereas in patients with relapse, its expression was usually increased, which resulted in a more than twofold higher average expression in these patients at F1 and F2, respectively.
BCL2 encodes a key anti-apoptotic protein that controls mitochondrial outer membrane permeability [
98]. The apoptosis pathway that is regulated by the Bcl-2 protein family is critical for lymphocyte development, maintenance of peripheral tolerance and prevention of autoimmunity [
99], and Bcl-2 family antagonism has been demonstrated to be a potential approach for the treatment of autoimmune diseases [
100]. It is thus possible that lower mRNA levels of
BCL2 in response to IRT may correlate with reduced disease activity in patients with MS. Another interesting gene is
IL13RA1, which was also expressed at higher levels in patients who relapsed.
IL13RA1 encodes a receptor subunit that mediates the signaling events induced by IL13 [
101]. Previous studies reported significantly higher percentages of IL13-producing T cells in the blood and CSF of patients in relapse compared to patients in remission [
102,
103]. Similarly, higher levels of the receptor might, therefore, be related to a higher risk of clinical relapse due to a suboptimal disease control. We also observed higher levels of
SLC38A11 in alemtuzumab-treated patients experiencing a relapse, while the average expression decreased after each treatment course. SLC38A11 is a member of the SLC38 family of transmembrane sodium-coupled amino acid transporters, which are particularly expressed in cells that carry out significant amino acid metabolism [
104,
105]. However, its role in B cells and MS is still unclear. In the interpretation of our results, it should be noted that early disease activity after initiation of IRT does not necessarily implicate treatment failure and that it is usually appropriate to continue the therapy. For example, one of our patients had 3 relapses in the year before IRT and another relapse in the first year of alemtuzumab therapy but was relapse-free in the second year. This patient received a pre-treatment with fingolimod, which has been reported to be a risk factor of relapses following alemtuzumab infusion [
106]. Further research is needed to study the relationship of gene expression signatures in the blood and specific treatment sequences with the individual course of disease. This should help to translate potential candidates into clinically useful molecular biomarkers and to guide more personalized therapeutic decisions in the near future.
A hallmark but also a limitation of the present study is the sole focus on B cells. Furthermore, the source of RNA for the transcriptome analysis was a mixture of B-cell subsets. Meanwhile, the recent rise of single-cell multi-omics technologies has enabled researchers not only to study gene expression patterns at the single-cell level but also to obtain information on the (epi)genetics and proteomics of individual cells at the same time [
107]. Others used RNA sequencing to investigate the temporal dynamics in B-cell immunoglobulin heavy chain repertoires during IRT [
108,
109]. Through integration of such different types of data, together with metabolomic profiles, it should be possible to better define perturbations in the immune signature of patients with MS. Further advances in our understanding of the disease processes will ultimately drive the development of even more selective, effective and safe therapeutics for MS. This may bring us closer to the goal of preventing neurological deterioration and inducing long-lasting drug-free disease stability. Another limitation of our study is the rather small number of patients per therapy timepoint. Therefore, the identification of potential gene expression markers of relapse activity in alemtuzumab-treated patients was exploratory in nature. Moreover, we did not analyze other treatment outcomes, such as MRI findings and the development of secondary autoimmune disorders, because the available data were too sparse and heterogeneous. Additional studies are required to confirm that therapeutic efficacy correlates with the expression of genes that we have nominated as biomarker candidates. If they prove to be useful for prognosis and monitoring of disease activity, they may allow to select patients who will benefit most from an IRT and/or patients who need an additional treatment course.