Endogenous antibodies contribute to macrophage-mediated demyelination in a mouse model for CMT1B

Mice heterozygously deficient for P0 (P0het; [17]) were crossbred to animals specifically lacking B-lymphocytes (JHD−/−; [18]) according to previously published protocols [13] and investigated at the age of 1 and 6 months with the corresponding littermates of both gender. As additional controls, RAG1-deficient mice, lacking both B- and T-lymphocytes, were analyzed [19]. All mice were on a C57BL/6 N background, and genotypes were identified after purification of genomic DNA from ear biopsies using DNeasy Blood & Tissue Kit (Qiagen, Venlo, The Netherlands) according to the manufacturer’s guidelines. Genotypes of P0 mutants were determined by conventional PCR using oligonucleotides 5′-TCAGTTCCTTGTCCCCCGCTCTC-3′, 5′-GGCTGCAGGGTCGCTCGGTGTTC-3′, and 5′-ACTTGTCTCTTCTGGGTAATCAA-3′ leading to 334 or 500 bp products for the P0 null mutation or wildtype allele; genotypes of JHD mutants with oligonucleotides 5′-GAGGAGACGGTGACCGTGGTCCCTGC-3′, 5′-GGACCAGGGGGCTCAGGTCACTC-AGG-3′, 5′-GCCGCATTGCATCAGCCATGAT-GGA-3′, and 5′-CCTTGCGCAGCTGTGC-TCGACGTTG-3′ leading to 180 or 195 bp products for the JHD null mutation or wildtype allele; genotypes of RAG1 mutants with oligunucleotides 5′-GAGGTTCCGCTACGACTCTG-3′, 5′-CCGGACAAGTTTTTCAT-CGT-3′, and 5′-TGGATGTGGAATTGTTGCGAG-3′ leading to 530 or 474 bp products for the RAG1 null mutation or wildtype allele, respectively. Animals were kept in the animal facility of the Department of Neurology, University clinic of Würzburg, in a 12 h/12 h day (<300 lux)/night rhythm under barrier conditions using individually ventilated cages. All animal experiments were approved by the local authority, the Government of Lower Franconia, Germany.

Wildtype mice were deeply anesthetized by an intraperitoneal injection of a mixture of ketamine and xylazine (10 μl per g body weight). The right sciatic nerve was exposed and crushed at the region of the sciatic notch by using a non-serrated clamp using a constant pressure for 15 s. This resulted in a completely translucent appearance of the crushed area of the nerve. After 3 days, the mice were sacrificed and lesioned nerves were removed for analysis of complement deposition (see below).

For preparation of fresh frozen spleens and femoral nerves, animals were killed by asphyxiation with CO₂ (according to guidelines by the State Office of Health and Social Affairs Berlin), blood was transcardially rinsed with phosphate-buffered saline (PBS) containing heparin, followed by harvesting and embedding nerves in O.C.T. medium (Sakura, Alphen aan den Rijn, The Netherlands) and frozen in liquid nitrogen-cooled methyl butane.

For preparation of fixed femoral nerves, blood of sacrificed mice (see above) was transcardially rinsed with PBS/heparin followed by perfusion with 4% paraformaldehyde (PFA) in PBS for 10 min. Dissected nerves were post-fixed in the same solution for 2 h, rinsed in 30% sucrose/PBS overnight at 4°C, and embedded in O.C.T. medium. Peripheral nerves were cut into 10 μm-thick cross sections on a cryostat (Leica, Solms, Germany) and stored at −20°C.

Single teased-fiber preparations were processed as described elsewhere [20]. Briefly, blood was transcardially rinsed with PBS/heparin followed by perfusion with 2% PFA in PBS for 10 min. Femoral quadriceps nerves were dissected, and single fibers were separated by forceps on glass slides.

For electron microscopy, femoral quadriceps nerves were processed as described previously [11]. In short, mice were transcardially perfused with 4% PFA and 2% glutaraldehyde in 0.1 M cacodylate buffer. Dissected nerves were post-fixed in the same solution overnight at 4°C. After osmification and dehydration, samples were embedded in Spurr’s medium. Ultrathin sections (80 nm) were mounted to copper grids and counterstained with lead citrate.

For identification of endogenous antibodies on nerve cross sections or teased fibers, PFA-fixed samples were blocked with 10% bovine serum albumin (BSA) and 1% normal goat serum (NGS) in 0.1 M PBS, followed by incubation with Cy3-conjugated goat-anti-mouse IgG-Fc antibodies (1:300, 715-166-150, Dianova, Hamburg, Germany) in 1% BSA and 1% NGS in 0.1 M PBS for 1 h at room temperature. Alternatively, samples were incubated overnight with non-coupled rabbit-anti-mouse IgG-Fc (1:1,000, 31194, Thermo Scientific, Waltham, MA, USA) at 4°C, followed by incubation with the corresponding Cy3-conjugated secondary antibodies for 1 h (1:300, 111-165-144, Dianova). To determine whether endogenous nerve antibodies are bound to extra- or intracellular domains, unfixed and non-permeabilized native sciatic nerves were loosely teased and incubated free floating in a 96-well plate. IgG-Fc staining was performed similarly as described above. To control for internal antibody deposition, nerve fibers were permeabilized by repeated cycles of freezing in liquid nitrogen and thawing. Incubation with rabbit-anti-mouse β-III-Tubulin (1:500, ab18207, Abcam, Cambridge, UK) and detection with the corresponding Cy3-conjugated secondary antibodies for 1 h (1:300, 111-165-144, Dianova) were additionally performed and served as positive controls for successful permeabilization and internal antibody binding.

Complement deposition was identified on teased fiber preparations. Briefly, samples were fixed in acetone (10 min, −20°C), blocked with 5% NGS with 0.3% TritonX-100 in 0.1 M PBS, and incubated with FITC-conjugated goat-mouse C3 (1:100, 0855500, MP Biomedicals, Santa Ana, CA, USA) in 1% NGS with 0.3% TritonX-100 in 0.1 M PBS for 1 h at room temperature. As positive controls, teased fibers from distal parts of lesioned sciatic nerves (3 days after crush) were identically treated as uninjured nerves of mutant mice. Diaphragms, incubated ex vivo with anti-ganglioside monoclonal antibodies and normal human serum [21], were investigated for complement deposition as lesioned nerves, whereby postsynaptic terminals were identified with Alexa Fluor 555-conjugated α-Bungarotoxin (1:300, B35451, Molecular Probes, Life Technologies, Carlsbad, CA, USA).

Quantification of endoneurial macrophages was performed on cross sections according to previously published protocols [8,11]. Briefly, fresh-frozen samples were fixed in acetone (10 min, −20°C) and blocked with 5% BSA in 0.1 M PBS, followed by an avidin-biotin blocking step (SP-2001, Vector Laboratories, Burlingame, CA, USA); biotinylated rat-anti-F4/80 (1:300, MCA497B, Serotec, Kidlington, UK) primary antibodies in 1% BSA in 0.1 M PBS were applied for 1 h at room temperature and detected by Cy3-conjugated Streptavidin (1:100, CED-CLCSA1010, Biozol, Eching, Germany). Nuclei were labeled with DAPI (Sigma-Aldrich, St. Louis, MO, USA). Whole nerve cross sections were analyzed, and the mean number of macrophages per section in seven to ten consecutive sections per animal was calculated.

Determination of B-lymphocytes was performed on cross sections of the spleen. In short, fresh-frozen samples were fixed in acetone (10 min, −20°C), blocked with 5% BSA in 0.1 M PBS and incubated with rat-anti-B220/CD45R (1:100, 550286, BD Pharmingen, San Jose, CA, USA) primary antibodies in 1% BSA in 0.1 M PBS overnight at 4°C, and detected by Cy3-conjugated secondary antibodies (1:300, 112-165-167, Dianova).

Determination of protein concentrations and Western blot analysis were performed as previously described [11]. Briefly, animals were killed by asphyxiation with CO₂, blood was rinsed with PBS/heparin, and femoral quadriceps nerves were quickly dissected and snap frozen in liquid nitrogen. After sonification (Sonoplus HD60, Bandelin Electronic, Berlin, Germany) in 100 μL RIPA lysis buffer per 10 mg tissue, protein concentration was determined by a Lowry assay (Sigma-Aldrich). Proteins were resolved by sodium dodecyl sulfate-polyacrylamide.

After gel electrophoresis (SDS-PAGE with 10% Acrylamide), proteins were transferred to a nitrocellulose membrane and visualized with Ponceau S (Roth, Newport Beach, CA, USA). Membranes were blocked with 5% skimmed milk in PBST and incubated with rabbit-anti-ERK1/2 (1:10,000, sc-94, Santa Cruz Biotechnology, Dallas, TX, USA) antibody solution overnight at 4°C. Corresponding horseradish peroxidase (HRP)-conjugated secondary antibodies (1:3,000, NA9340V, GE Healthcare, Little Chalfont, UK) were probed for 1 h at room temperature, and detection of the immune reaction was determined by ECL reagent and ECL hyperfilm (GE Healthcare Bio-Sciences AB). Endogenous antibodies were detected by directly labeled HRP-anti mouse IgG-Fc antibodies (1:5,000, 115-035-008, Dianova) overnight at 4°C or for 2 h at room temperature. Densitometric analyses were performed with Image J (NIH) and depicted as the ratio of IgG to ERK1/2. Results from two independent experiments are shown.

Multiple image alignments were acquired, and abnormally myelinated fibers, consisting of thinly and demyelinated axons, onion bulbs, and foamy macrophages were quantified in relation to the total number of axons in cross sections of the femoral quadriceps nerve. The g-ratio was determined by dividing the diameter of the axon by the diameter of the same axon including its myelin sheath. At least 125 fibers per animal were randomly selected and analyzed. Analysis was performed using a ProScan Slow Scan CCD camera mounted to a Leo 906E electron microscope (Zeiss) with corresponding iTEM software (Olympus Soft Imaging Solutions GmbH, Münster, Germany).

150 μg of mouse IgGs (PMP01, Serotec) or mouse-anti-keyhole limpet hemocyanin (KLH)-antibody (orb11067, Biorbyt, Cambridge, UK) or 50 μg mouse IgG Fc fragments (31205, Pierce, Rockford, IL, USA) were injected into the tail vein (i.v.) once or at four consecutive weeks followed by analysis 1 week later. After dissection and tissue processing, specific binding of antibodies in peripheral nerves was controlled by immunohistochemistry and Western blot analysis.

All experiments were performed in a blinded manner, with the investigators unaware of the genotypes of the analyzed mice. Data sets were controlled for normal distribution by Shapiro-Wilk test. Comparison of more than two groups with normally distributed data (parametric) was tested by one-way ANOVA, followed by Tukey post hoc test or Bonferroni-Holm correction. In the case of non-normally distributed data sets, the nonparametric Kruskal-Wallis test with Bonferroni-Holm correction was applied. Significance levels (*, # P < 0.05; **, ## P < 0.01; ***, ### P < 0.001) are indicated together with the applied statistical tests within the figure legends. Quantifications are shown as mean + SD. Statistical analyses were performed using PASW Statistics 18 (SPSS, IBM, Armonk, NY, USA) software, and diagrams were generated with Microsoft Excel 2007.

In order to detect antibody binding, we performed immunohistochemical stainings for IgGs on femoral nerve cross sections from 6-month-old P0het mutant mice and P0wt littermates. We found a strong antibody deposition in the perineurium of femoral quadriceps nerves of both P0wt and P0het mutants (Figure 1A). However, in the endoneurium of mutant nerves, a higher IgG immunoreactivity was observed compared to wildtype controls which could be confirmed by IgG signal intensity measurements (Figure 1A,B). Of note, in the non-affected sensory branch of the femoral nerve (saphenous nerve), this prominent antibody deposition could not be detected in P0het mutant mice (Figure 1A). We also performed IgG immunocytochemistry on single teased-fiber preparations, in order to detect potentially preferred antibody depositions along myelinated nerve fibers. Indeed, we could confirm a stronger deposition along myelinated fibers in P0het mice compared to wildtype controls, with strongest IgG immunoreactivity at nodes of Ranvier (Figure 1D). To determine whether endogenous nerve antibodies are bound to extra- or intracellular domains, nerve fibers of unfixed and non-permeabilized sciatic nerves of P0het mice were loosely teased and stained free floating for IgG and β-III-Tubulin. Again, IgG immunoreactivity could be detected along myelinated nerve fibers indicating external deposition of endogenous antibodies, whereas β-III-Tubulin immunoreactivity was confined to some axonal profiles at nodes of Ranvier (Figure 1E), likely reflecting the vulnerability of this structure by mechanical teasing. As controls for the accessibility of intracellular domains by permeabilization, nerve fibers were repeatedly frozen in liquid nitrogen followed by incubation with antibodies to IgGs or β-III-Tubulin. As expected, only after permeabilization, β-III-Tubulin staining showed an extended immunoreactivity along the entire axon including internodes, while IgG immunoreactivity was still visible along the outer margins of the myelinated fibers (Figure 1E). Thus, endogenous antibodies bind to extracellular domains of peripheral nerve fibers.

In order to substantiate our immunocytochemical findings, we performed Western blot analysis of peripheral nerve extracts using antibodies to mouse IgG and IgM subtypes. We found increased IgG (Figure 1F,G) and IgM signals (not shown) in P0het mice in comparison to extracts from P0wt mice, corroborating the immunocytochemical observations. Specificity of the respective staining was further controlled by performing immunocytochemistry and Western blot analysis in RAG1-deficient mice, lacking T- and B-lymphocytes (and, therefore, antibodies). As expected, this resulted in a complete lack of decorating antibodies in peripheral nerves (Figure 1). We also investigated antibody deposition in 1-month-old P0het mice and P0wt littermates, an early time point in disease before features typical for demyelination are morphologically visible. Indeed, by immunohistochemistry (Figure 1C) and Western blot analysis (not shown), we detected already at this early age a similarly elevated IgG immunoreactivity in the mutant nerves as at 6 months of age. However, perineurial staining was substantially weaker.

Our data suggest that specific endogenous antibodies bind to extracellular domains of endoneurial tubes in peripheral nerves of myelin mutant mice, a potential early prerequisite for modulating pathogenesis in an animal model for CMT1B.

Deposition of antibodies is often associated with the classical or alternative pathways of complement activation [22,23]. We, therefore, investigated whether in our P0het mice, C3, a central component of both pathways, is deposited on mutant nerve fibers. By immunohistochemistry using a C3-specific antibody, we failed to identify complement deposition, while - as positive controls - C3 deposition was amply detectable on nerve fibers of crushed nerves [24] and on explanted diaphragms incubated with anti-ganglioside antibodies and normal human serum [21] (Figure 2).

In order to identify the role of antibodies in a model for CMT1B, we crossbred P0het mice with JHD-deficient mutants that lack B-lymphocytes and are, thus, incapable to produce antibodies. As expected, JHD−/− mice were devoid of B-lymphocytes in the spleen and antibodies in peripheral nerves as revealed by immunohistochemistry and Western blot analysis (Figure 3).

Next, we investigated the number of macrophages in P0wt and P0het mice either positive or homozygously deficient for JHD (Figure 4). Numbers of macrophages were significantly elevated in femoral quadriceps nerves of 6-month-old P0het mice when compared to age-matched wildtype mice, confirming previous observations [8,9]. Interestingly, P0het mice additionally deficient for JHD and lacking endogenous antibodies showed a significantly attenuated elevation of F4/80-positive macrophages (Figure 4A,B). However, in pathologically non-affected saphenous nerves, the number of macrophages was not altered in single and double mutant mice (Figure 4A,B). We additionally focused on macrophage activation by electron microscopy. We found a substantial number of macrophages only in P0het mice that were JHD positive, whereas all other genotypes, including P0het JHD−/− mice, lacked foamy macrophages (Figure 4C).

As a next step, we investigated whether absence of antibodies in P0het JHD−/− mice leads to an alleviation of demyelination, a typical feature of our CMT1B mouse model. Quantification by electron microscopy of profiles indicative of pathological alterations revealed a significant reduction of abnormally myelinated fibers and of supernumerary Schwann cells (‘onion bulbs’) in P0het JHD−/− mice compared to P0het JHD+/+ mice at 6 months of age (Figure 5). Moreover, quantification of the g-ratio (axon diameter/fiber diameter) showed a significant increase in myelin thickness in P0het JHD−/− compared to P0het JHD+/+ mice (Figure 5D) demonstrating an improved myelin integrity. Thus, endogenous antibodies seem to contribute to early macrophage-mediated demyelination in P0het myelin mutant mice.

In order to clarify the role of endogenous antibodies in the pathogenesis in a model for CMT1B, we performed reconstitution experiments into P0het JHD−/− mice to determine if antibodies can revert the observed beneficial effects. We performed i.v. injections of IgGs either once or on four consecutive weeks and analyzed mice at the age of 6 months. By immunohistochemistry, we found that after passive transfer into P0het JHD−/− mice, IgGs accumulated on endoneurial tubes of peripheral nerves comparable to P0het JHD+/+ mice, although to a weaker degree (Figure 6A). IgG deposition was increased in femoral quadriceps nerves after four weekly injections compared to mice that received only a single IgG injection (Figure 6A). The passive transfer led not only to increased antibody detection in nerve sections but also to elevated numbers of F4/80-positive macrophages and foamy macrophages in P0het JHD−/− mice, comparable to P0het JHD+/+ mutants (Figure 6B,C), whereas antibody reconstitution had no effect on numbers of putative resident macrophages in P0wt JHD−/− mice (not shown). To analyze the pathogenic impact of injected antibodies, we quantified nerve fiber damage by electron microscopy. P0wt JHD−/− showed no pathological alterations after IgG injections (Figure 7A). However, IgG injections reverted the ameliorated phenotype of P0het JHD−/− mice, as determined by an increased number of abnormally myelinated fibers and supernumerary Schwann cells (Figure 7B,C), suggesting that antibodies are sufficient to aggravate nerve damage along with macrophage-mediated demyelination.

Finally, we investigated whether antigen specificity is needed for the detrimental effect of antibodies. We performed passive transfer experiments into P0het JHD−/− mice using mouse antibodies that show specificity for a non-mammalian antigen, like keyhole limpet hemocyanin (KLH), or mouse-IgG Fc fragments. Expectedly, in both cases, we failed to demonstrate antibody binding to peripheral nerves by immunohistochemistry (not shown). Unexpectedly, we could observe a slight, non-significant increase of total macrophage numbers after injection of non-specific IgGs or mouse-IgG Fc fragments in individual animals (Figure 6B) together with an increase of phagocytosing macrophages (Figure 6C). Consistent with this macrophage activation, we could also detect slightly increased nerve fiber damage after reconstitution (Figure 7), suggesting that antibodies not binding to neural tissue can activate macrophages and contribute to demyelination in peripheral nerves of P0het mice.