Anti-CD52 antibody treatment depletes B cell aggregates in the central nervous system in a mouse model of multiple sclerosis

Background Multiple sclerosis (MS) is a chronic autoimmune disease of the central nervous system (CNS) for which several new treatment options were recently introduced. Among them is the monoclonal anti-CD52 antibody alemtuzumab that depletes mainly B cells and T cells in the immune periphery. Considering the ongoing controversy about the involvement of B cells and in particular the formation of B cell aggregates in the brains of progressive MS patients, an in-depth understanding of the effects of anti-CD52 antibody treatment on the B cell compartment in the CNS itself is desirable. Methods We used myelin basic protein (MBP)-proteolipid protein (PLP)-induced experimental autoimmune encephalomyelitis (EAE) in C57BL/6 (B6) mice as B cell-dependent model of MS. Mice were treated intraperitoneally either at the peak of EAE or at 60 days after onset with 200 μg murine anti-CD52 vs. IgG2a isotype control antibody for five consecutive days. Disease was subsequently monitored for 10 days. The antigen-specific B cell/antibody response was measured by ELISPOT and ELISA. Effects on CNS infiltration and B cell aggregation were determined by immunohistochemistry. Neurodegeneration was evaluated by Luxol Fast Blue, SMI-32, and Olig2/APC staining as well as by electron microscopy and phosphorylated heavy neurofilament serum ELISA. Results Treatment with anti-CD52 antibody attenuated EAE only when administered at the peak of disease. While there was no effect on the production of MP4-specific IgG, the treatment almost completely depleted CNS infiltrates and B cell aggregates even when given as late as 60 days after onset. On the ultrastructural level, we observed significantly less axonal damage in the spinal cord and cerebellum in chronic EAE after anti-CD52 treatment. Conclusion Anti-CD52 treatment abrogated B cell infiltration and disrupted existing B cell aggregates in the CNS. Electronic supplementary material The online version of this article (10.1186/s12974-018-1263-9) contains supplementary material, which is available to authorized users.


Background
Multiple sclerosis (MS) is one of the most prevalent autoimmune disorders of the central nervous system (CNS) [1]. The disease is characterized by inflammation, demyelination, and axonal degeneration [2], eventually leading to severe and irreversible neurological deficits. The typical age of MS onset is between 20 and 40 years with a ratio of 2:1 female to male patients and a higher prevalence in the northern hemisphere. While the etiology of MS is not fully understood, a combination of genetic, environmental, and infectiological factors is assumed to be responsible for disease initiation and progression [1].
Traditionally, the autoimmune pathology of MS has been considered to be mainly T cell-driven [3], while more recent clinical observations as well as experimental data suggest a close interaction of T cells and B cells. Along these lines, depositions of immunoglobulins and of complement components have been found in active demyelinating brain lesions of MS patients [4], and oligoclonal bands (OCB) in the cerebrospinal fluid (CSF) of the majority of patients demonstrate local antibody production. Apart from antibody production, B cells function as highly potent antigen-presenting cells [5] particularly in the brain and are furthermore able to regulate ongoing inflammatory processes, e.g., by secreting pro-and/ or anti-inflammatory cytokines. In addition, the overall clinical efficacy of B cell-depleting anti-CD20 antibody treatment and the therapeutic success of plasma exchange in a subset of MS patients with severe corticosteroid-resistant relapses further underlines the involvement of B cells in the immunopathology of MS [6,7].
Alemtuzumab (Campath-1H) is a humanized anti-CD52 antibody [8], which was first approved for the treatment of patients with chronic lymphocytic leukemia. Alemtuzumab was later shown to be highly effective in patients with relapsing-remitting MS (RR-MS), lowering both the annual relapse rate and the sustained accumulation of disabilities in multiple clinical studies [9]. Improvement of pre-existing disabilities has also been reported. Cells of the acquired immune system, such as T cells and B cells, show high expression of CD52 molecules [10], hence being reduced drastically in numbers in the course of treatment, while cells of the innate system, such as natural killer cells and monocytes, are depleted to a lower extent due to lower CD52 expression. Plasma cells and hematopoietic stem cells lack CD52 and therefore remain unaffected. The absence of CD52 on hematological precursor cells enables maturation and repopulation after treatment, which appears to have a long-lasting tolerance-inducing effect in patients with MS, with more than 50% not requiring disease-modifying therapy 3 years after treatment [11]. This may be due to the slow repopulation of T cells with a relative expansion of CD4 + CD25 + CD127 low regulatory T cells [12][13][14] as well as a rapid repopulation of mostly immature B cells transiently exceeding baseline levels [15,16]. While there is no doubt about the value of alemtuzumab for the treatment of RR-MS, further studies are needed to determine if it is equally effective in the case of progressive MS and also to determine whether there are patient subgroups for which individualized treatment would be appropriate.
To further study the immunological mechanisms of action of alemtuzumab, a human CD52 transgenic mouse strain was established [17]. In this mouse strain, both the distribution of CD52 molecules on leukocytes and the depletory capacity of alemtuzumab are comparable to that in humans [18]. However, for the ease of experimentation, a murine anti-CD52 antibody was subsequently established and used in experimental autoimmune encephalomyelitis (EAE) [19], which is the most common animal model of MS. Studies of anti-CD52 treatment in EAE have so far involved the myelin oligodendrocyte glycoprotein (MOG) peptide 35-55 [19][20][21] and the rMOG:1-121 model in C57BL/6 (B6) mice, as well as the relapsing-remitting proteolipid protein (PLP) peptide 13-151 model in SJL mice [19]. In all of these models, mice were treated on five consecutive days in an early disease stage, which resulted in the depletion of T cells and B cells comparable to MS patients treated with alemtuzumab. Anti-CD52-mediated lymphocyte depletion significantly attenuated clinical EAE, altered peripheral cytokine production, and reduced CNS inflammation, demyelination, and axonal damage. In addition, elevated levels of CD39 + regulatory T cells were observed in the gut-associated lymphatic tissue [20]. Since none of the aforementioned studies has used a B cell-and antibodydependent model, the impact of anti-CD52 treatment on the peripheral vs. central B cell compartment in EAE still remains to be investigated. In addition, the antibody was previously given already with the onset of EAE when scores had not yet reached their maximal severity. Thus, it is unclear whether anti-CD52 antibody treatment is equally effective in more severe forms of EAE and in the chronic stage of the disease. To address these questions, we have employed the MP4-induced EAE model, in which mice are immunized with a fusion protein of myelin basic protein (MBP) and proteolipid protein (PLP) [22][23][24], and in which both pathogenic B cells and antibodies are induced. Our data demonstrate differential clinical effects of anti-CD52 treatment in acute and chronic EAE. In addition, this study is the first to show that treatment was highly effective in targeting both peripheral and central B cells, including those B cells that had already formed aggregates in the CNS. The disruption of these B cell aggregates was accompanied by a significant decrease in axonal pathology, even in late stages of the disease.

Mice
Six-week-old female B6 were purchased from Envigo (The Netherlands) and maintained under specific pathogen-free conditions at the animal facility of the Zentrum für Mund-und Kiefergesundheit at the University of Würzburg. The mice had free access to a standard rodent diet (Altromin Spezialfutter GmbH & Co. KG, Lage, Germany) and autoclaved water. Food was softened and kept at ground level for mice that displayed paralytic symptoms to ensure a sufficient diet. All animal experiments were performed according to protocols that were approved by the Regierung von Unterfranken (approval number 91/14) and complied with the German Law on the Protection of Animals and the "Principles of laboratory animal care" (NIH publication no. 86-23, revised 1985).

EAE induction and clinical score
For EAE induction, incomplete Freund's adjuvants was prepared by mixing paraffin oil (EM Science, Gibbstown, NJ, USA) and mannid monooleate (Sigma-Aldrich, St. Louis, USA) at a 9:1 ratio. To obtain complete Freund's adjuvants (CFA), 5 mg/ml Mycobacterium tuberculosis H37 Ra (Difco Laboratories, Franklin Lakes, NJ, USA) was added. Each mouse was immunized subcutaneously into both sides of the flank with a total dose of 200 μg MP4 (Alexion Pharmaceuticals, Cheshire, CT, USA), emulsified in a total volume of 200 μl CFA. In addition, mice received 200 ng pertussis toxin (List Biological Laboratories, Hornby, ONT, Canada) by intraperitoneal injection at the day of immunization and 48 h later. Mice were evaluated daily to record onset and progression of clinical symptoms based on the standard EAE scoring system: (0) no disease, (1) floppy tail, (2) hind limb weakness, (3) full hind limb paralysis, (4) quadriplegia, and (5) death. Increments of 0.5 were used to account for clinical deficits in between the defined hallmarks.

Treatment
Mice were treated either with a 200 μg (10 mg/kg body weight) anti-mCD52 antibody, obtained from Sanofi Genzyme (Cambridge, MA, USA), or with a murine IgG2a isotype control antibody (InVivo, Henningdorf, Germany) for five consecutive days. Treatment was given by intraperitoneal injection, and mice were subsequently monitored daily for at least 10 days to determine the treatment effect. Mice were treated either at the peak of EAE ("acute EAE") or at~60 days after EAE onset ("chronic EAE"). For randomization purposes, each mouse in each cohort was assigned to one of the two treatment groups in an alternating fashion once the mouse had developed EAE. Yet, the extent of CNS inflammation was almost exclusively dependent on the EAE score, rather than on the disease duration after onset. Hence, slight variations of the initial randomization strategy occurred since the two groups were scorematched at the beginning of the treatment (Table 1).

Tissue sampling and preparation
Blood samples for flow cytometry were taken from the tail vein 1 day before sacrifice. Mice were sacrificed with CO 2 before a non-perfused draining inguinal lymph node was collected for ELISPOT and blood was drawn from the inferior vena cava for ELISA analysis. Mice were then perfused with 4% paraformaldehyde (PFA) (Roth, Karlsruhe, Germany) in 0.01 M phosphate-buffered saline (PBS) (pH 7.4) for immunohistochemistry (IHC) or with 4% PFA/4% glutaraldehyde (GA) (Roth, Karlsruhe, Germany) in 0.01 M PBS (pH 7.4) for electron microscopy. The cerebellum, lumbar spinal cord, and the other draining inguinal lymph nodes were carefully dissected and post-fixed overnight at 4°C under slight agitation. While samples for IHC were embedded in paraffin, samples for electron microscopy were embedded in Epon.

IHC for CD3 + T cells and B220 + B cells
In MP4-induced EAE, a focus of B cell infiltration is observed in the cerebellum [25]. To this end, paraffin-embedded cerebella were cut into 4-5-μm-thick serial sections covering the entire tissue. Every seventh (acute treatment group) or ninth (chronic treatment group) section was stained with hematoxylin/eosin in order to screen for the presence of inflammatory infiltrates. To this end, n = 12 sections were analyzed per mouse. In case of infiltration, the adjacent section was stained for CD3 + T cells and B220 + B cells by IHC. The diaminobenzidine (DAB)/Vector Blue staining protocol consisted of the following steps: paraffin-embedded sections were rehydrated and endogenous peroxidase was blocked. After citrate buffer-mediated epitope retrieval and protein blocking, the primary antibodies rabbit anti-mouse CD3 (

Phosphorylated neurofilament heavy (pNF-H) ELISA
pNF-H ELISA was performed with the sera of the same mice that were used for histological analysis. We used a commercially available ELISA kit from Merck according to the manufacturer's instructions.

Electron microscopy
For EM analysis, we obtained a standardized tissue block from the lumbar spinal cord of each mouse. We carefully prepared the whole spine and isolated the part that contained the spinal cord between thorax and the beginning of the cauda equina. Ultra-thin sections were then obtained from the top of each tissue block. For EM analysis of the cerebellum, the tissue of each mouse was cut into six similar blocks. These blocks corresponded to those areas that most likely contained B cell aggregates and thus neurodegeneration according to our previous experience. All six blocks were screened, and images were acquired from those blocks that were affected by neuroinflammation. For embedding, samples were first washed with PBS and subsequently with 0.1 M cacodylate buffer (CACO, sodium dimethyl arsenate in aqua dest) (EMS, München, Germany) (pH 7.35) three times for 10 min each. Afterwards, samples were stored in 0.1 M CACO buffer at 4°C under slight agitation. For epon embedding, samples were washed again with 0.1 M CACO buffer three times on a rotating wheel for 10 min each. Then, they were placed into 0.1 M CACO buffer containing 1% osmium tetroxide (OsO 4 ) (EMS, München, Germany) at room temperature for 4 h. During that time, the tissue was protected from light and deposited on a rotator. The tissue was then washed with aqua bidest twice for 15 min and then for 15 min each in 30%, 50%, 70%, 80%, 90%, and 96% ethanol. After washing twice in 100% ethanol for 15 min each, the tissue was incubated in propylene oxide (EMS, München, Germany) twice for 30 min. The samples were then transferred into a 1:1 mixture of propylene oxide and epon at room temperature overnight and afterwards incubated with pure fresh epon for 2 h. Epon was changed again, and embedded tissue samples were polymerized at 60°C for 48 h. Epon was prepared by mixing 26 g glycid ether (Serva, Heidelberg, Germany), 11 g dodecenyl succinic-acid anhydride (DDSA) (Serva, Heidelberg, Germany), 15 g methylacid anhydride (MNA) (Serva, Heidelberg, Germany), and 0.25 g 2-,4-,6-tris(dimethylaminomethyl)phenole (DMP) (Serva, Heidelberg, Germany). Embedded tissue was cut transversely at 1 μm on an ultra-microtome (Leica, Wetzlar, Germany, serial number 365952) for semi-thin analysis. Sections were dried at 60°C for at least 20 min and rinsed in methylene blue (Merck, Darmstadt, Germany) for approximately 45 s. After sections had been washed and dried on a heating plate, they were covered in DEPEX (Serva, Heidelberg, Germany). Digital images were acquired using a Leica DM LB2 microscope, equipped with a Zeiss camera (AxioCam MRc) and Zeiss software (AxioVision 40 4.7). For EM analysis, tissue was cut into 65-85-nm-thick sections. Sections were stretched with chloroform (Roth, Karlsruhe, Germany) while floating in the water tub of the diamond knife (EMS, München, Germany). Afterwards they were carefully put on 100 mesh formvar-coated nickel grids (EMS, München, Germany). For contrast enhancement, the sections were stained with 2% uranyl acetate (Serva, Heidelberg, Germany) in 70% ethanol for 20 min and Reynold's lead citrate solution (Plano, Wetzlar, Germany) for 7 min. The tissue was analyzed using a Zeiss AB912 LEO transmission electron microscope at 80-kV acceleration voltage. Digital images were acquired with an EM digital camera system (Tröndle sharp eye camera, Moorenweis, Germany) equipped with Zeiss software (Image SP × 64) at × 6300 magnification. At least 10 images per mouse and tissue were analyzed, which covered the areas of the spinal cord and cerebellar lesions.

Statistical analysis
Mann-Whitney and one-way ANOVA tests were used for statistical analysis and computed using GraphPad Prism 6 (La Jolla, CA, USA). Statistical significance was set at p ≤ 0.05. Mean values and standard errors of the mean (SEM) are shown in the graphs.

Results
Anti-mCD52 antibody treatment attenuates acute but not chronic MP4-induced EAE MP4-immunized B6 mice received 200 μg of anti-mCD52 antibody or IgG2a isotype control by intraperitoneal injection on five consecutive days either at the peak of acute EAE or at~60 days after EAE onset. The mean EAE onset of all treated mice was 16.12 ± 0.55 days after immunization. There was no significant difference in disease severity in anti-mCD52-and isotype control-treated mice at the beginning of treatment in both acute (p = 0.83) and chronic (p = 0.59) EAE ( Fig. 1a-d, Additional file 2, Table 1). In acute EAE, anti-mCD52-treated mice displayed a significant decrease in disease severity 11.56 ± 0.18 days post-treatment, while the scores of isotype control-treated mice remained unaffected (Fig. 1a, Table 1; p = 0.02). In chronic EAE, anti-mCD52 treatment did not have any effect on the difference between pre-and post-treatment scores ( Fig. 1b; Table 1; p = 0.47). In order to determine the extent of depletion of circulating lymphocytes after anti-mCD52 treatment, we performed flow cytometry on tail vein blood obtained 10.37 ± 0.18 days post-treatment and we determined the percentages of viable CD4 + T cells and CD19 + B cells. In the chronic cohort, CD4 + T cells showed a significant reduction from 24.02 ± 1.71% to 3.39 ± 0.52% (p < 0.001; Fig. 1e). The percentage of CD19 + B cells was significantly reduced to 6.70 ± 1.23% in anti-mCD52-treated mice compared to 38.28 ± 2.82% in the isotype control group (p < 0.001; Fig. 1f ). In the acute EAE cohort, comparable results were observed for both T cells (compare 33.82 ± 3.68% to 0.92 ± 0.19%; p = 0.016) and B cells (compare 21.51 ± 4.41% to 3.22 ± 1.34%; p = 0.016).

Anti-mCD52 treatment does not affect MP4-specific B cells and autoantibodies
In order to determine the effect of anti-mCD52 treatment on peripheral MP4-specific B cells, we performed ELISPOT on draining lymph node cells (Fig. 2a, b) of both anti-mCD52-and isotype control-treated mice in the acute and chronic EAE cohort. Samples were collected 11.37 ± 0.18 days after treatment onset. As shown in Fig. 2, there was no difference between the numbers of MP4-specific antibody secreting B cells expressed as anti-MP4 IgG/total IgG ratio in anti-mCD52-vs. isotype control-treated animals (p acute = 0.51; p chronic = 0.81). There was also no difference between anti-MP4-specific IgG titers in the sera of both groups (p acute = 0.73; p chronic = 0.75) (Fig. 2c, d).

Anti-mCD52 treatment significantly decreases immune cell infiltration in the cerebella of acute and chronic MP4immunized mice
Since our results showed a reduction in blood lymphocyte counts in FACS analysis, we set out to further determine the impact of this lymphocyte depletion on immune infiltration in the CNS. To this end, we screened the cerebella of mice treated at the peak of EAE and~60 days after EAE onset for CD3 + T cell and B220 + B cell infiltration by IHC. The tissue was collected 11.37 ± 0.18 days after treatment. To further differentiate infiltrate categories in the CNS, we defined three different types of lymphocyte infiltration (Fig. 3a-c). Loose perivascular infiltrates of CD3 + T cells in the absence of B220 + B cells were defined as non-B cell infiltrates, while the simultaneous presence of individual CD3 + T cells and B220 + B cells was characteristic of diffuse B cell infiltrates. B cell aggregates were defined as highly compact perivascular aggregation of stained lymphocytes containing at least one-third B220 + B cells. The percentage of perivascular infiltration per parenchymal area was determined for each category in acute and chronic EAE. As shown in Fig. 3, there was a significant overall reduction in immune cell infiltration in the cerebella of anti-mCD52-treated animals in the acute (p = 0.03) ( Fig. 3d-f) and chronic EAE cohort (p = 0.002) (Fig. 3g-i). Strikingly, B cell aggregates were entirely absent in all of the n = 14 anti-mCD52-treated animals, while only n = 4 of 15 isotype control-treated mice were aggregate-free. Because of this complete depletion of B cell aggregates in the anti-CD52-treated group, we did not perform any further analysis on potential lymphoid neogenesis resulting from B cell aggregation in this study.

Anti-mCD52 therapy does not have neuroprotective effects in MP4-induced EAE
We have demonstrated that anti-mCD52 treatment significantly reduced both peripheral lymphocyte counts and CNS infiltration in MP4-induced EAE. Yet, there was no impact on clinical disease severity in the chronic stage of the disease. This lack of effect on late-stage EAE may be explained by the absence of primarily neuroprotective properties of the drug. We stained sections of the Fig. 1 Disease course in mice treated with anti-mCD52 antibody in the acute or chronic stage of EAE. MP4-immunized mice were treated with 200 μg anti-mCD52 antibody or IgG2a isotype control for five consecutive days (tx) either a at the peak of the disease or b at~60 days after EAE onset. c, d Representative disease course of individual animals treated during c acute or d chronic EAE. Arrows indicate the treatment period. e, f Flow cytometry of CD4 + T cells and CD19 + B cells in the blood after treatment with anti-mCD52 compared to vehicle. ***p < 0.001, Mann-Whitney test. Mean values ± SEM are shown. For EAE scores, statistics were calculated based on differences on a daily basis lumbar spinal cord of anti-mCD52-and isotype controltreated B6 mice of the chronic treatment cohort for demyelination using LFB (Fig. 4a) and for SMI-32 + damaged axons (Fig. 4b). Both the demyelinated area and the number of SMI-32 + axons were set in relation to the total white matter (WM) spinal cord area that was analyzed. We also determined the cell densities of Oli-g2 + APC + oligodendrocytes (Fig. 4c). The data show that the extent of demyelination (p = 0.74) and axonal damage (p = 0.97) as well as the cell densities of Olig2 + APC + oligodendrocytes (p = 0.07) were comparable between anti-mCD52-and isotype control-treated mice. Similar results were obtained when the cerebellum was analyzed (Additional file 3). To further corroborate these results, we performed pNF-H ELISA on serum samples from the same mice that were used for histology (Fig. 5). It has previously been proposed that the detection of pNF-H in CSF and blood of patients with neurodegenerative diseases provides information about the degree of axonal injury [26]. Indeed, pNF-H has been reported to be present in large amounts in CSF and blood following experimental spinal cord and brain injury in rats [27]. In addition, the detection of pNF-H has been associated with amyotrophic lateral sclerosis [28,29], optic neuritis [30], and also with MS [31]. In our study, elevated levels of pNF-H were detectable in the sera of MP4-immunized compared to non-immunized control mice. However, there was no difference between anti-CD52-and isotype control-treated mice (Fig. 5).
While light microscopic and serum analyses provide a first impression on neurodegeneration, ultrastructural analysis is needed for detailed quantification. To this end, we used electron microscopy to assess myelin pathology as well as beginning and irreversible axonal damage in both anti-mCD52-and isotype control-treated mice in the spinal cord (Fig. 6) and cerebellum (Fig. 7). In order to distinguish between de-and remyelinating nerve fibers, the g-ratio was used. This ratio describes the relationship of the myelin sheath to its respective nerve fiber. Analysis of n = 8 naïve mice allowed us to determine the physiological range of the g-ratio in both the spinal cord (0.40-0.86) and cerebellum (0.48-0.82). While beginning axonal damage can be assessed by the so-called nearest neighbor neurofilament distance (NNND) with a decreased NNND indicating pathology, irreversible axonal damage is reflected by axolysis and axonal loss. Figures 6 and 7 demonstrate that both anti-mCD52-and isotype control-treated mice displayed a significantly increased number of pathological nerve fibers both in terms of demyelination and axonal damage, which pertained both to the spinal cord and the cerebellum. Along these lines, the number of axons was significantly reduced in both EAE groups. While there was no difference in the extent of deand remyelination comparing anti-mCD52-and isotype  (Figs. 6b and 7b), anti-mCD52-treated displayed significantly decreased axonal pathology and axonal loss (Figs. 6c and 7c).

Discussion
Alemtuzumab, a humanized anti-CD52 antibody, is a relatively new approved treatment option for MS with promising results in multiple clinical trials [32][33][34][35]. Yet, the mechanistic effects of the drug on different aspects of MS pathogenesis have remained subject to further research, including the impact of alemtuzumab on the B cell component of the disease. The most recent reports of treatment failure in individual patients with subsequent severe and possibly B cell-mediated relapses underline the importance of further research along these lines [36]. While longitudinal studies on the impact of alemtuzumab on peripheral and CNS B cells are difficult to perform in MS patients themselves, the EAE model offers the possibility of investigating long-term effects of anti-CD52 antibody treatment on both the autoimmune response and also on neurodegeneration and nueroregeneration.
In our model, anti-CD52 antibody treatment drastically reduced the numbers of circulating CD4 + T cells and B220 + B cells, while the titers of MP4-specific serum antibodies and the numbers of the corresponding plasmablast population in the blood remained unaffected. This was expected and well in line with previous studies in EAE [19][20][21] and in humans [12,13,15,37] and could be explained Fig. 3 Histological analysis of B cell aggregates in the cerebella of MP4-immunized mice. a-c CD3/B220 double staining of representative cerebellar sections of a mouse treated with isotype control at the peak of disease. The panels demonstrate the different infiltrate categories. Scale bars represent 100 μm. d-f Quantification of the number of infiltrates per parenchymal area and distribution of infiltrate categories in mice treated either with anti-mCD52 antibody or isotype control at the peak of disease. g-i Quantification of the number of infiltrates per parenchymal area and distribution of infiltrate categories in mice treated~60 days after disease onset. Graphs show means ± SEM. *p < 0.05, **p < 0.01, Mann-Whitney test by differences in the expression of CD52 on different lymphocyte subsets. In particular, low levels of CD52 on plasmablasts have been noted before [37].
Unlike previous studies that all relied on T cell-dependent EAE, we immunized mice with the MP4 molecule that induces both encephalitogenic T cells and B cells/antibodies. Using this model, we were able to show that anti-CD52 treatment did not only deplete circulating B cells, but also B cells that had formed aggregates in the CNS. Interestingly, treatment not only prevented the formation of aggregates, but also disrupted already existing aggregates in MP4-induced EAE. The clinical relevance of B cell aggregates in the brains of MS patients is still subject to debate, but an association with an earlier disease onset, progression, and death and with more severe cortical pathology is assumed [38,39]. These results may support recently published data that found alemtuzumab to reduce the relapse rate in MS patients more efficiently than fingolimod, while being comparable to natalizumab [40]. Mechanistically, it is unclear whether the disruption of B cell aggregates in the brain after anti-CD52 antibody treatment is the result of direct B cell depletion or rather due to a lack of replenishment from circulating peripheral B cells. Natalizumab is a potent anti-VLA4-antibody that blocks the migration of lymphocytes through the blood-brain barrier (BBB) by inhibiting their binding to the cell adhesion molecules  [41,42] indicates that constant supply from the periphery is necessary to maintain B cell aggregates if they are considered at least partially responsible for OCB development. We have previously shown that treatment with the S1P 1 receptor agonist fingolimod, which inhibits lymphocyte egress from secondary lymphoid organs, prevented the development of B cell aggregates in the CNS in acute MP4-induced EAE, while it did not affect existing infiltrates in the chronic stage of the disease [43]. This rather suggests the independence of B cell aggregates, at least partially, from peripheral support. Yet, our data have also shown that the effect of fingolimod on B cells was much less pronounced than on T cells. Therefore, it is conceivable that there is still ongoing migration of autoreactive B cells into the CNS in the setting of fingolimod treatment, while natalizumab blocks B cell migration more effectively and alemtuzumab depletes the majority of migrating lymphocytes.
Any efficient depletion of lymphocytes in the CNS relies on the penetration of the antibody through the BBB, on sufficient binding to the target cell, and on the activation of antibody-dependent cellular and/or complementdependent cytotoxicity (ADCC/CDC) in case of the currently available type I antibodies. This has been confirmed for alemtuzumab in vitro [44] and in vivo [17]. While in gadolinium-enhanced magnetic resonance imaging (MRI) studies of lesions in RR-MS patients the BBB has been described to be permeable for antibodies, and complement levels have been reported to be elevated in the CNS, the BBB in secondary progressive (SP-MS) and primary progressive (PP-MS) patients exhibits lower permeability. Indeed, a clinical trial comparing a combination of intrathecal and intravenous application of the anti-CD20 antibody rituximab to placebo in patients with low-inflammatory SP-MS was terminated early since it failed to reach the expected efficiency in the depletion of intrathecal B cells and axonal damage markers in the CSF [45]. Decreased CDC, lack of natural killer cells, and lower CSF flow were made responsible for the failure of the trial. These arguments may also apply to alemtuzumab treatment.
In order to prevent further or even reverse pre-existing neurological disabilities, the search for new treatment options for MS patients focuses not only on immune modulation, but also on neuroprotection [46,47]. A post hoc subgroup analysis of patients in the CAMMS223 cohort with no clinical evidence of immunological disease activity before treatment and no clinical and radiological disease activity on-trial displayed an overall reduction of clinical disabilities [48]. This could not be explained by the suppression of inflammation due to the depletion of lymphocytes since ongoing neuroinflammation was not detectable in this subgroup. These findings raised hope that alemtuzumab may create an environment of "neuroprotective autoimmunity," with reported higher levels of neurotrophins and oligotrophins resulting in disability reduction. Along these lines, Turner et al. [19] described the reversal of EAE scores in the MOG 35-55 , MOG 1-121 , and PLP 139-151 model early after disease onset after anti-CD52 antibody treatment. Furthermore, they demonstrated neuroprotective effects of the drug in the spinal cord of MOG 35-55 -immunized mice during acute EAE, reporting a significant reduction in demyelinated white matter, axonal damage, as well as the preservation of axonal conductance. When we applied the anti-CD52 antibody at the peak of MP4-induced EAE, we observed a similar attenuation of clinical EAE. However, there was no effect on EAE scores when treatment was initiated in the chronic stage of the disease~60 days after EAE onset, which had not been evaluated in the other EAE models before [19][20][21]. Further analysis revealed differential effects of the treatment on CNS degeneration in chronic MP4-immunized mice. In contrast to the results of Turner et al., we did not observe any difference between isotype-and anti-CD52 antibody-treated mice in the light microscopic analysis of demyelination and SMI-32 + axons, indicating axonal damage in the lumbar spinal cord and cerebellum of MP4-immunized mice. These results were underlined by similar levels of pNF-H in the sera obtained from the two mouse groups. Furthermore, we did not observe ongoing remyelination when staining for Olig2 + APC + oligodendrocytes or when performing ultrastructural electron microscopic analysis. Yet, on the ultrastructural level, the numbers of damaged axons were significantly decreased after anti-CD52 treatment. Indeed, electron microscopy allows the precise evaluation of CNS pathology on the level of individual nerve fibers and The assay was performed on serum samples from non-immunized control mice as well as from isotype control-and anti-mCD52-treated MP4immunized EAE mice that had received treatment~60 days after EAE onset. Graphs show means ± SEM. ***p < 0.001, ANOVA is therefore much more sensitive than the light microscopic techniques. These beneficial effects on CNS histopathology are likely to the result of reduced CNS inflammation and potentially the disruption of B cell aggregates, which have been described to secrete soluble toxic molecules into the adjacent tissue inducing CNS pathology [39]. This assumption is supported by a 6-year follow-up clinical trial that displayed sustained reduction of the relapse rate, however without improvement of clinical disabilities in a more selective group of alemtuzumab-treated patients with poor prognostic indicators such as high relapse rate, active disease on cMRI scans, and rapidly accumulating disability [34].

Conclusion
This study underlines the potency of anti-CD52 antibody treatment to deplete T cells and B cells in the context of EAE. The novel aspect of this study is the demonstration of a treatment effect on B cell aggregates in the CNS. It will be subject to future studies to determine whether alemtuzumab is equally effective in depleting B cell aggregates in MS patients and whether such depletion is actually associated with clinical benefit. On the one hand, Fig. 6 Electron microscopic analysis of demyelination and axonal damage in the spinal cord of MP4-immunized mice. a-c Representative electron microscopic images of the spinal cord of non-immunized control and in MP4-immunized mice that had received treatment with either isotype control or anti-mCD52 antibody~60 days after EAE onset. Scale bars denote 1 μm. The arrow indicates a demyelinating axon. The asterisks represent axons with a decreased NNND, and the diamond shows an axolytic axon. b Quantification of the percentage of physiological and of de-and remyelinating nerve fibers in non-immunized control vs. MP4-immunized mice of the two treatment cohorts. c Quantification of axonal damage and loss in nonimmunized control vs. MP4-immunized mice of the two treatment cohorts. Graphs show means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ANOVA suitable imaging tools to detect B cell aggregate formation in the brains of MS patients would help to address this question. Until such tools may eventually be available, the development of blood-and/or CSF-based biomarkers to measure autoantigen-specific B cell activity may be useful to predict and monitor the effect of alemtuzumab on the B cell compartment in patients and to correlate such an effect with clinical outcomes. On the other hand, while B cell aggregate formation is a characteristic feature of the MP4 model, the actual existence and relevance of these structures in the MS brain is still heavily debated.
Eventually, the permeability of alemtuzumab through the BBB and its mode of action in the CNS itself needs to be further investigated. It may also be worthwhile to consider a type II anti-CD52 antibody, which is supposed to be less dependent on CDC and ADCC by inducing apoptosis directly.

Additional files
Additional file 1: Gating strategy for flow cytometric analysis of viable CD19 + B cells and CD4 + T cells in the blood. (TIF 1297 kb) Additional file 2: Comprehensive disease course of mice treated with anti-mCD52 antibody in the acute or chronic stage of EAE, respectively. (TIF 143 kb) Fig. 7 Electron microscopic analysis of demyelination and axonal damage in the cerebellum of MP4-immunized mice. a-c Representative electron microscopic images of the cerebellum of non-immunized control and MP4-immunized mice that had received treatment with either isotype control or anti-mCD52 antibody~60 days after EAE onset. Scale bars denote 1 μm. The arrow indicates a demyelinating axon. The asterisk represents an axon with a decreased NNND, and the diamond shows an axolytic axon. b Quantification of the percentage of physiological and of de-and remyelinating nerve fibers in non-immunized control vs. MP4-immunized mice of the two treatment cohorts. c Quantification of axonal damage and loss in nonimmunized control vs. MP4-immunized mice of the two treatment cohorts. Graphs show means ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ANOVA