Paraneoplastic CDR2 and CDR2L antibodies affect Purkinje cell calcium homeostasis

We used four female patients’ sera that were antibody positive for CDR2 (hCDR2^+(PS1)), CDR2L (hCDR2L^+(PS2)) or both CDR2 and CDR2L (hCDR2/2L^+(PS3/PS4)) [22] and negative for P/Q-type VGCC (RIA; DLD Diagnostika, #RA006/12). Patient data are listed in Table S2. The sera were collected before treatment took place and stored at −80 °C at the Paraneoplastic Neurological Diseases Bio-bank (#484) with the approval of the Regional Committee for Medical and Health Research Ethics in Western Norway, Diagnostic markers of cancer (188.05). As control serum, we pooled samples from 100 healthy blood donors without any known autoimmune disease (non-hCDR).

All procedures were performed according to the National Institutes of Health Guidelines for the Care and Use of Laboratory Animals Norway (FOTS 20135149/20113133). To prepare cOTSC, we used 152 Wistar Hannover GLAST rat pups (in-house breeding colony), age P10–P15. Following anesthesia and decapitation, we transferred the cerebellum into ice-cold EBSS solution (Gibco, #24010043) containing 0.5 % D-glucose (Sigma, #G8769) and 10 mM HEPES (Gibco, #15630056). Four to five cerebellar parasagittal slices (400 μm thick) were cut on NVSLM1 motorized advance vibroslice (WPI) and transferred individually onto 0.4 μm pore size membranes (Millicell, Millipore, # PICMO1250). Slices were maintained in 24-well plates at the air/culture media interface consisting of 30 % advanced DMEM/F12 solution (Gibco, #126340010), 20 % MEM solution (Gibco, #41090028), 25 % EBSS solution, 25 % heat-inactivated horse serum (Sigma, #H1138), 1 mM l-glutamine (Gibco, #35050038), 5 mg/mL d-glucose, and 2 % B-27 serum-free supplement (Gibco, #17504044), and incubated with 5 % CO₂ at 35 °C (Fig. 1a). The culture medium was removed and replaced 24 h post-slicing (100 %) and then every 2nd day (75 %). The slices recovered for 7 days before treatment.

hCDR and rCDR were heat inactivated (56 °C, 30 min) to destroy complement factors prior to treatment to obtain neutral IgG antibody assays and added at various concentrations to the cOTSC media for 2–6 days (Fig. 1b). We used human sera hCDR2⁺, hCDR2L⁺, and hCDR2/2L⁺, non-hCDR at dilutions of 1:250, 1:500, and 1:1000, and affinity-purified polyclonal rabbit antibody rCDR2 (Sigma, #HPA023870), rCDR2L (Sigma, #HPA022015), or both (1:1 mixture) and rIgG (Millipore, #12370) in total concentrations of 20, 40, 125, or 400 ng/mL. Each independent experiment (n ≥ 3) included positive and negative controls to avoid variations in immunoreactivity to the applied hCDR and rCDR between each experiment.

The following drugs were used: MDL28170 a potent, selective inhibitor of calpain (K _i = 10 nM; Tocris, #1146); CNQX an AMPA receptor antagonist (IC₅₀ = 1–2 μM; Tocris, #1045); ω-agatoxin a selective and reversible blocker of Cav2.1 (P/Q-type VGCC) (Alomone, #A-530); K252a a non-selective protein kinase inhibitor that inhibits PKC (IC₅₀ = 32.9 nM; Tocris, #1683); U0126 a selective non-competitive inhibitor of MEK-1 and MEK-2 (IC₅₀ = 60–70 nM; Tocris, #1144).

The antibodies used in immunohistochemical (IHC), Western blot (WB), and immunoprecipitation (IP) analyses are detailed in Table S3.

Multiphoton images were collected with a Ti Sapp laser (Coherent Chameleon Ultra2) and DM6000 CFS-TCS SP5 microscope (Leica) using an HCX PL APO 20 × 1.0 water-immersion objective (with a digital zoom of ×1.7 or ×3.0 [Fig. 3b]). Excitation was performed at 740 nm (6.8–7.2 mW laser power); emission was detected for Alexa Fluor^® 488/594 with the NDD1/NDD2 external detectors, respectively. The fluorescence intensity was adjusted to 75 % of the maximum in untreated controls for each experiment. Z-stack images were taken at 0.5–1 μm intervals. Pictures were superimposed using LASAF software version 2.5.1 (Leica Microsystems CMS GmbH). Additionally, images from Alexa Fluor^® 488/594 Donkey Anti-Mouse IgG staining in Figs. 2b, 3b, c, h were pseudo-colored in gray and inverted for clarity using Fiji. Figure S1b was created as 3D projections with orthogonal section from z-stacks using the Fiji plug-in 3D Viewer. Calbindin D_28K (CB⁺)- and L7/Pcp-2 (L7/Pcp-2⁺)-positive PCs were counted manually and automatically, but blind in three to eight images of 750 × 750 × 100 μm in each slice for each experiment (n _E) and group and projected to mm³ (Fig. 1c). The automatic count was performed with Fiji: (1) image → stack → plug-in: z projection [max intensity]; (2.) image → adjustment → plug-in: auto threshold [yen]; (3) analyze → plug-in: 3D objects counter [threshold: 50; size filter: min: 200 and max: 1000]. The manual and automatic counts produced equivalent numbers.

The treated cOTSC were washed with pre-warmed 1xPBS (pH 7.4) and sonicated in ice-cold lysis buffer (in mM: 50 Tris–HCl (pH 7.5), 150 KCl, 1 DTT, 1 PMFS, 0.1 % NP40, 1 % SDS, and complete protease inhibitor cocktail (Roche, #1187380001)). The lysates (40 μg) were subjected to gradient SDS-PAGE under reducing conditions (15–4 % Tris TGX gels; Bio-Rad, #456-1086), transferred onto PVDF membrane (0.2 μm pore size; Bio-Rad, #162-0177) and incubated overnight (4 °C) with anti-calbindin D_28K, anti-calpain-1, anti-calpain-2, anti-Cav2.1, anti-PKCγ, and anti-β-tubulin. The blots were developed using horseradish peroxidase-conjugated 2nd Ab (polyclonal Swine Anti-Rabbit IgG-HRP, Dako, #P0217; Goat Anti-Mouse IgG-HRP, Santa Cruz Biotechnology Inc., #SC-2500; 90 min, 22 °C) and the Pierce ECL chemiluminescence system (Thermo Fisher Scientific Inc., #3210634075). Semi-quantitative densitometry analysis was performed (Bio-Rad, Quantity One 4.6.6). For each sample, the protein level was normalized to β-tubulin and expressed relative to untreated naive control, set at 1 (arbitrary units).

Co-IP was performed with rat cerebellar tissue. The tissue was homogenized by 10–15 strokes in a Dounce-type glass homogenizer (buffer in mM: 20 Tris–HCl (pH 7.4), 137 NaCl, 10 % glycerol, 1 EDTA, 1 PMSF, 1 DTT, 1 % NP-40, protease inhibitors). Homogenates were centrifuged (13,000g, 4 °C, 10 min) and the supernatants were used for immunoprecipitation after pre-clearing with protein G magnetic beads (1 h, 4 °C; Millipore, # LSKMAGG02). The supernatants were incubated with 2 μg/mL of specific Ab with constant agitation (overnight, 4 °C). A further 40 μL of fresh protein G magnetic beads was added and incubated for 1 h at 4 °C. The beads were washed three times (IP buffer in mM: 50 Tris–HCl (pH 7.4), 150 NaCl, 1 EDTA, 1 PMSF, 1 DTT, 0.1 % NP-40, protease inhibitors) and incubated with 100 mM dithiothreitol and 2× Laemmli sample buffer (5 min, 95 °C). Eluted proteins were resolved by SDS-PAGE. We immunoprecipitated against: anti-CDR2, anti-CDR2L, anti-Cav2.1, and anti-calbindin D_28k (2 μg).

Data analysis and calculations were performed using the software Excel 2003 and Graph Pad Prism 4.0. Data are presented as mean ± SEM, and statistical significance was determined using the non-parametric two-tailed paired Mann–Whitney’s U test. The level of significance is indicated with asterisks: *p < 0.05; **p < 0.01; ***p < 0.001.

We investigated four female patients’ sera that were positive for CDR2, CDR2L, or both CDR2/2L and negative for P/Q-type VGCC (Table S2). Cerebellum cryostat sections incubated with those human patients’ sera (hCDR; 1:2000) confirmed that hCDRs bound to PCs. hCDR2/2L^+(PS3/PS4) showed stronger PC binding than hCDR2^+(PS1) and hCDR2L^+(PS2) (Fig. 2a).

To evaluate the pathological effects of hCDR internalization into PCs in the absence of the lymphatic system and blood–brain barrier, we applied the sera to cOTSC (Fig. 1a). We collected samples at different time points and studied PC survival by counting CB-positive PCs in multiphoton micrographs (Fig. 1b, c) [28, 38]. To exclude Ab diffusion into dead PCs with permeable membranes, we performed control monoclonal anti-CB staining without Triton X-100 in the incubation buffers. We did not find false-positive CB immunoreactivity (data not shown).

hCDR internalization significantly reduced the number of CB-positive PCs (CB⁺-PCs) over time (n _E = 4; Fig. 2c, d; Table 1). The remaining CB⁺-PCs showed reduced dendritic arborizations (Fig. 2b). Within 48 h, CB⁺-PC loss was significantly higher in hCDR2L^+(PS2) than hCDR2^+(PS1) (*p = 0.0373; Fig. 2c), but did not differ between hCDR2/2L^+(PS3) and hCDR2/2L^+(PS1+PS2) and was most strongly pronounced in hCDR2/2L^+(PS4) (Fig. 2d; Table 1). Compared to the non-hCDR control, more than 70 % of CB⁺-PCs were lost at day 6 of hCDR internalization (***p < 0.0001; Fig. 2c, d; Table 1).

To elucidate whether the observed pathology was induced by CDR antibodies, we treated the cOTSC with heat-inactivated affinity-purified polyclonal rabbit Abs against CDR2 (rCDR2⁺) and CDR2L (rCDR2L⁺), or both (1:1 mixture, rCDR2/2L). We found that within 6 days both rCDR2 and rCDR2L were internalized and led to extensive loss of CB staining (400 ng/mL; Fig. 3a, merged). We observed three patterns of CB immunoreactivity modifications after rCDR treatment: (1) whole cell staining with neighboring cells unstained, (2) dendritic staining with only a weak remaining signal in the soma, and (3) soma staining with no dendritic staining (Fig. 3b). The remaining CB⁺-PCs showed morphological alterations in their dendritic arborizations with a loss of tertiary branches (Fig. 3c).

To verify the progression of rCDR-induced PCD-like pathology, samples were collected at different time points (n _E = 4; 400 ng/mL; Fig. 3c, d). We found no effect on the number of CB⁺-PCs in rIgG control. Meanwhile within 48 h, we observed that rCDR treatment caused a 50–60 % CB⁺-PCs loss. The CB⁺-PC loss under rCDR2/2L and rCDR2L was of similar magnitude, but significantly greater than in rCDR2 (n _E = 4, *p = 0.0369; Fig. 3d; Table 1). rCDR reduced CB⁺-PCs by ~80 % within 6 days (rCDR groups compared to rIgG⁺ ***p < 0.0001; Fig. 3d; Table 1). The progression of the antibody-driven loss of CB in the first 48 h was similar in rCDR2 and rCDR2L compared to hCDR2^+(PS1) and hCDR2L^+(PS2) (Table 1).

To study antibody level effects, we tested four rCDR concentrations (ng/mL): 400, 125, 40, and 20 (n _E = 3, 4 days). We found a reduction of CB⁺-PCs by ~80 % (400 and 125 ng/mL), 60 % (40 ng/mL), and 40 % (20 ng/mL) compared to rIgG control (***p < 0.0001; Fig. 3e) with no difference between rCDR2 and rCDR2L.

The Western blot analysis of rCDR-treated cOTSC lysates (6 days, 125 ng/mL) showed no effect on CB protein concentration compared to untreated naive control (n _E = 10; Fig. 3f, g). It is therefore likely that the reduced CB immunoreactivity is caused by structural modifications of the CB Ca²⁺-binding site or misfolding of the protein itself [11, 50]. To investigate this hypothesis, we replaced the rCDR-containing media (125 ng/mL, 6 days) with non-CDR media (Fig. 1b, washout, 7 days). The number of CB⁺-PCs in rIgG control was not affected by washout or by time (Fig. 3h, i). We found that the washout of rCDR improved the CB⁺-PC count in rCDR2L and rCDR2/2L -treated cOTSC within 48 h (D2wo) (rCDR2: p = 0.1383, rCDR2L: **p = 0.0018, rCDR2/2L: **p = 0.0010; Fig. 3i). All rCDR groups showed substantial recovery of CB⁺-PCs at day 7 after washout (***p < 0.0001; Fig. 3h, i; Table 1), but plateaued between days 4 and 7 at ~ 40 % CB⁺-PC loss before reaching the rIgG control count. This supports our hypothesis that the CB Ca²⁺-binding epitope is reversibly altered or blocked, or that the protein might be misfolded.

To analyze whether CDR proteins and CB interact at physiological expression levels, we assayed by reciprocal co-immunoprecipitation (Co-IP) CDR2, CDR2L, and CB proteins from cerebellar lysates. As shown in Fig. 4a, b, CB and CDR2 specifically Co-IPed, indicating endogenous CDR2–CB complex in PCs. Interestingly, CDR2L did not form any complex with CB (anti-CDR2L from three sources), but showed strong CDR2 Co-IP (Fig. 4a, c). These Co-IP results suggest that CDR2 antibody can substantially influence CB function by binding to an interacting protein.

Although CB immunoreactivity was reduced after hCDR and rCDR internalization, the micrographs in Fig. 3a showed no cell loss. HE staining of hCDR2/2L^+(PS3)-treated cOTSC (1 μL/mL, 6 days) showed normal cytoarchitecture and no distinct loss of PCs (Fig. S1a). Furthermore, staining for the apoptotic marker of active cleaved caspase-3 in rCDR2/2L-treated cOTSC (100 ng/mL, 6 days) showed small scattered cleaved caspase-3-positive apoptotic cells which did not co-localize with CB⁺-PCs (Fig. S1b). This caspase-3 staining profile was not different from the rIgG control.

PCs express GoLoco domain protein L7/Pcp-2, which interacts with Gα_o and Gα_i proteins and thereby modulate the P/Q-type VGCC (Cav2.1) functions in a concentration-dependent manner [39]. The P/Q-type is the major VGCC in PCs and linked to ataxia pathogenesis [27, 69]. To evaluate whether L7/Pcp-2 is influenced by CDR internalization, we stained hCDR- and rCDR-treated cOTSC against L7/Pcp-2. The L7/Pcp-2 positive (L7/Pcp-2⁺) PC loss was similar to the observed CB⁺-PC loss (Fig. S2; Table 1). Both hCDR (1 μL/mL) and rCDR (100 ng/mL) induced L7/Pcp-2⁺-PC loss of more than 65 % (6 days, hCDR n _E = 3; rCDR n _E = 4; compared to rIgG control ***p < 0.0001; Fig. S2a–d; Table 1).

This suggests that another important Ca²⁺ homeostasis regulator is affected by CDR antibody internalization.

To confirm the suggested interaction of L7/Pcp-2 and Cav2.1, we performed Co-IPs on rat cerebellum extract and confirmed the protein complex (Fig. 5a). As a consequence of this interaction, the CDR antibody-induced loss of L7/Pcp-2 might influence the VGCC protein expression. To test whether VGCCs are affected by CDR internalization, we performed Western blot analysis of rCDR-treated cOTSC. We found that rCDR internalization increased the protein expression of Cav2.1 by more than twofold (100 ng/mL, 6 days, all groups compared to rIgG controls, ***p < 0.0001; Fig. 5b, c) and co-treatment with VGCC antagonist ω-agatoxin (150 nM) blocked this effect (Fig. 5g, h; ^# p < 0.005; n _E = 6). ω-agatoxin co-treatment averted CB⁺- and L7/Pcp-2⁺-PC loss (Fig. 5d–f). The rescue of CB and L7/Pcp-2 immunoreactivity was most strongly pronounced in rCDR2/2L⁺/ω-agatoxin co-treated cOTSC (***p < 0.0001, 6 days; n _E = 5; Fig. 5e, f; Table 1). ω-Agatoxin at 150 nM did not affect the CB⁺- and L7/Pcp-2⁺-PC count or Cav2.1 protein expression in rIgG and untreated naive controls.

Inactivation of VGCC by blocking AMPA receptor-mediated synaptic depolarization with AMPA receptor antagonist CNQX revealed a similar rescue of CB⁺- and L7/Pcp-2⁺-PC loss (Fig. 6a–e; Table 1). We found that co-treatment of hCDR2/2L^+(PS3) (1 μL/mL) with CNQX minimized the CB⁺-PC loss significantly in a concentration-dependent manner (CNQX: 44 ± 8 % (5 μM), 24 ± 8 % (10 μM), 14 ± 6 % (25 μM); ***p < 0.0001, 6 days; n _E = 6; Fig. 6b). Similar CB⁺- and L7/Pcp-2⁺-PC loss reduction was obtained under 10 μM CNQX/hCDR2/2L^+(PS4) co-treatment (1 μL/mL hCDR2/2L^+(PS4); ***p < 0.0001; 6 days; n _E = 4; Fig. 6b, c; Table 1). The CDR-specific effects of CNQX (10 μM) were also tested with rCDR2 and rCDR2L (100 ng/mL, 6 days). rCDR/CNQX co-treatment partially rescued CB⁺-PCs (***p = 0.0001; n _E = 6; Fig. 6d; Table 1), with better rescue in rCDR2L/CNQX (63 %) than rCDR2/CNQX (38 %) (**p = 0.0054). The L7/Pcp-2⁺-PC count in co-treated rCDR2 or rCDR2L cOTSC with 10 μM CNQX showed reduction of L7/Pcp-2⁺-PC loss to 48 ± 3 % (rCDR2/CNQX) and 27 ± 5 % (rCDR2L/CNQX), respectively (***p < 0.0001; 6 days; n _E = 4; Fig. 6e; Table 1). The CNQX co-treatment was more beneficial for rCDR2L than for rCDR2, as seen with CB staining (L7/Pcp-2⁺-PC rescue: 30 % (CDR2) vs. 60 % (CDR2L), **p = 0.0011). The number of CB⁺- and L7/Pcp-2⁺-PCs in the rIgG control was not affected by the AMPA receptor antagonist.

Taken together, these data show that reducing Ca²⁺ influx through VGCC mitigates the CDR antibody-induced pathology by lowering intracellular Ca²⁺ levels.

The postsynaptic Ca²⁺ transient can be modulated by inhibition of Ca²⁺-activated phospholipid-dependent protein kinase C (PKC) activity to prevent AMPA receptor-mediated synaptic depolarization [19, 73]. PKC is associated with apoptosis in cerebellar ischemia and stroke [8]. Missense mutations in PKCγ gene cause spinocerebellar ataxia type 14 [58, 60]. We found that 6 days of rCDR2 and rCDR2/2L internalization significantly increased PKCγ protein expression (rCDR2: 1.43 ± 0.12, **p = 0.0079; rCDR2/2L: 1.62 ± 0.29, *p = 0.0250; Fig. 7a, b), but not rCDR2L (1.23 ± 0.18, p = 0.1810; Fig. 7b). Therefore, we co-administrated 50 nM K252a to inhibit PKC activity during CDR internalizations. Blocking PKC activity with K252a prevented CDR antibody-induced CB⁺- or L7/Pcp-2⁺ -PC (hCDR2/2L^+(PS3)[1 μL/mL]; 6 days; ***p < 0.0001; n _E = 4; and rCDR [125 ng/mL]; ***p < 0.0001; 6 days; n _E = 4; Fig. 7c–f; Table 1) and blocked the increased PKCγ protein expression caused by rCDR2 and rCDR2/2L internalization (Fig. 7g, h; ^# p < 0.003; n _E = 6). rCDR2L internalization had no effect on PKCγ protein expression, while CB⁺ - and L7/Pcp-2⁺-PC loss was rescued by K252a co-treatment in the rCDR2L group. The antagonist had no effect on the CB⁺- or L7/Pcp-2⁺ count or PKCγ protein expression in the rIgG and untreated naive control.

The results show that the CDR antibody-mediated pathology can be attenuated in the presence of VGCC, AMPA receptor, and PKC signaling inhibition.

Changes in VGCC, AMPA receptor, and PKC signaling can lead to elevated intracellular Ca²⁺ levels and induce Ca²⁺-dependent proteases activity. To examine whether CDR internalization affects Ca²⁺-dependent proteases activity of calpain-1 and calpain-2, we performed Western blot analysis of rCDR-treated cOTSC lysates. Calpain-1 expression was not affected, but calpain-2 expression was significantly increased (rCDR2: 2.05 ± 0.14, ***p < 0.0001; rCDR2L: 1.72 ± 0.14, ***p = 0.0002; rCDR2/2L: 1.84 ± 0.19, **p = 0.0012; 125 ng/mL; 6 days; Fig. 8a–c). We then investigated whether calpain-2 over-activation/-expression could be antagonized by calpain-specific inhibitor MDL28170, a known neuroprotective substrate in ischemia and traumatic injury models [9, 36]. MDL28170 co-application with hCDR2/2L^+(PS3/PS4) (6 days) reduced the CB⁺-PC loss to ~45 %, as well as the CDR-induced pathology on the dendritic arborizations in the remaining PCs (n _E = 6 for PS3 and n _E = 4 for PS4; Fig. 8d, e; Table 1). We found no difference between the 10 (42 ± 9 %) and 20 μM (44 ± 7 %) MDL28170/hCDR2/2L^+(PS3) treatment (p = 0.3841; Fig. 8e; Table 1). The MDL28170-induced rescue of CB⁺-PCs differed between hCDR2/2L^+(PS4) (18 %) and hCDR2/2L^+(PS3) (48 %) (*p = 0.0207; 10 μM; Fig. 8e; Table 1). Furthermore, hCDR2/2L^+(PS4)-induced L7/Pcp-2⁺-PC loss was partially rescued by 10 μM MDL28170 (***p < 0.0001; n _E = 3; Fig. 8f; Table 1). MDL28170 (10 μM) co-treatment of rCDR2, rCDR2L, and rCDR2/2L-treated cOTSC (125 ng/mL, 6 days) showed an improvement of the dendritic morphology and significant reduction of CB⁺-PC loss (***p < 0.0001; n _E = 5; Fig. 8g; Table 1). The increase of calpain-2 expression seen during CDR internalization was blocked by 10 μM MDL28170 (Fig. 8h, i; ^# p < 0.008; n _E = 6).

These data show that CDR internalization leads to an increase in intracellular unbound Ca²⁺ to the millimolar range, which thereby activate the Ca²⁺-dependent protease calpain 2.

Activation of mitogen-activated protein (MAP) kinase signaling cascades are associated with PKC-dependent synaptic depression/declustering of glutamate receptors, inflammation, death receptor activation, apoptosis, and oxidative stress [12, 23, 35]. Co-treatment of cOTSC with hCDR2/2L⁺ (1 μL/mL, 6 days) and MAP kinase inhibitor U0126 (5 μM) did not prevent the CB⁺- or L7/Pcp-2⁺-PC immunoreactivity loss (CB: hCDR2/2L^+(PS3): 70 ± 10 vs. 68 ± 7 % (5 μM U0126); p = 0.4268; L7/Pcp-2: hCDR2/2L^+(PS4): 65 ± 3 vs. 64 ± 2 % (5 μM U0126), p = 0.5249; n _E = 4; Fig. 8j).

OTSC can be used to study neurochemical, structural, and physiological changes linked to diseased in vivo brains [13, 44]. Hence, we created an ex vivo antibody-mediated PCD model by applying CDR-positive patients’ sera and affinity-purified rabbit CDRs to the culture medium of cerebellar rat OTSC. These slices offer unique advantages because the tissue architecture is preserved, synaptic circuitries are maintained, and various treatments can be evaluated without the influence of activated immune cells and the blood–brain barrier (BBB) [13, 21, 42‐44]. The mechanisms of autoimmunity in paraneoplastic neurological diseases are complex and not yet understood. It is suggested that brain epitope-specific serum Abs can access the brain only if the integrity of the BBB is compromised [18, 20]. However, brain epitope-specific cerebrospinal fluid Abs can be produced by intrathecal synthesis due to activated T and B cells that can cross the BBB [51, 62]. Data from anti-Yo/CDR2-Ab positive PCD patients are contradictory in terms of cytotoxic T cell involvement [2, 3, 10, 56, 61, 65]. Therefore, it is of interest that PCs can internalize IgGs and that anti-Yo Abs can lead to cell injury and non-apoptotic death without the influence of activated immune cells [31, 33].

Here, we demonstrate that both hCDR and rCDR can cause similar PC pathology. By using rCDR we exclude cytotoxic T cell involvement, since the cOTSC was not exposed to related peptides which could activate brain-naive resident T cells [42]. Therefore, our data are most likely immune cell independent. We found that both CDR2 and CDR2L are internalized independently and their removal can partly reverse the calpain-2-dependent, but caspase-3-independent, pathology. We found equal effects of CDR2 and CDR2L on PC survival, although only a combination of CDR2/2L is associated with PCD [22]. However, CDR2/2L pathology was stronger, e.g., PKCγ results, and reciprocal Co-IP showed a CDR2–CDR2L complex in PCs, which supports the hypothesis that increased Ab avidity enhances the pathological effects [71]. Our CDR-induced pathology followed a similar time frame as described by Greenlee and colleagues [31], but the CDR internalization mechanisms remain to be explored. A recent study by Congdon et al. [14] demonstrated antibody internalization via IgG-FcγII receptor endocytosis. IgG-FcγII receptors are found in PC and regulate cerebellar function [47].

In Fig. 9 we show the complexity, by which h/rCDR internalization changes PC physiology by targeting postsynaptic signaling factors and thereby modifies Ca²⁺ homeostasis. We described these factors along with related neurological diseases in Table S1. We found that the immunoreactivity of Ca²⁺ homeostasis regulator CB and VGCC modulator L7/Pcp-2 was reduced. These proteins are functionally linked, since CB shapes the post-tetanic potentiation induced by the Ca²⁺ influx through AMPAR and VGCC and can act as an activity-dependent sensor [32, 57]. When endogenous CB levels are reduced, PC dendritic outgrowth and differentiation is inhibited, and motor coordination and sensory integration fail [1, 4, 37, 74]. We found that the total CB protein levels are not affected by CDR internalization. However, Co-IP reveals that CDR2–CB forms a strong protein complex and therefore CDR2 antibody binding to CDR2 protein in the cell may disrupt CB function by sequestering CDR2 and thus hijacking CB’s interacting partner. “CDR antibody washout” restored the CB immunoreactivity only partially, indicating irreversible modifications of the CB Ca²⁺-binding mechanism or protein structure, as seen in other neurodegenerative diseases [11, 50].

CB dysfunction or deletion can result in drastic fast intracellular Ca²⁺-transient elevation which greatly influences Ca²⁺ homeostasis [4]. In PC signaling, intracellular Ca²⁺ microdomain variations due to AMPAR, VGCC, or PKC activity play a crucial role [24, 27, 34, 57, 68]. AMPAR activation by glutamate binding and site-specific PKC and receptor tyrosine kinases’ phosphorylation provides the depolarization necessary for opening VGCCs and therefore increases Ca²⁺ influx [19, 32, 57, 73]. We found that the P/Q-type VGCC modulator L7/Pcp-2 immunoreactivity is reduced and Cav2.1 (P/Q-type VGCC) protein concentration is increased by twofold after CDR internalization. L7/Pcp-2 modulates VGCC function in a concentration-dependent manner and enhances or dampens VGCC kinetics to shape fast Ca²⁺ influx in synaptic Ca²⁺ microdomains [39]. Inhibiting VGCC activity directly (ω-agatoxin) or indirectly (CNQX) prevented or reduced dramatically the CDR-induced CB and L7/Pcp-2 loss, respectively. CDR2L-induced pathology was more effectively reversed than CDR2 during AMPAR inhibition. Data from primary PC culture and synaptoneurosome preparation (data not shown) support that CDR2L is synaptic and could thereby be involved in membrane-associated signaling cascades. CB and L7/Pcp-2 dysfunction and VGCC up-regulation will promote excessive Ca²⁺ influx into PCs, which can activate cell death pathways by enhancing PKC and Ca²⁺-dependent protease activity [41].

PKC is associated with apoptosis in ischemia and stroke [8, 29, 60]. In PCs, enhanced PKC activity will reduce dendritic differentiation, whereas reduced PKC activity increases it [45, 46]. We found that the up-regulation of PKCγ expression under CDR internalization was accompanied by reduced dendritic arborizations and loss of tertiary branches. In retina, L7/Pcp-2 interacts with PKC [64] and we found that the increased PKC activity led to reduced L7/Pcp-2 expression (data not shown). Inhibition of PKCγ activity prevented the CDR-induced morphological changes, the loss of CB as well as L7/Pcp-2 immunoreactivity and blocked the increased PKCγ protein expression. However, the increase in PKCγ expression was only seen for CDR2 and CDR2/2L, but not for CDR2L internalization, although K252a rescued the CB and L7/Pcp-2 immunoreactivity in all rCDR-treated cOTSCs. We therefore hypothesize that CDR internalization enhances PKC activity, which: first, reduces L7/Pcp-2 expression; second, increases Cav2.1 expression; third, increases intracellular Ca²⁺, which activates Ca²⁺-dependent proteases such as calpain-1 and calpain-2 and thereby induces neuronal death.

Calpain-1 and calpain-2 are activated by different cellular Ca²⁺ concentrations. In Alzheimer’s, Huntington’s, and Parkinson’s disease, calpain over-activation or dysregulation triggers neuronal death by truncation of important synaptic substrates [5, 16, 17, 26, 53, 59]. Western blot analysis revealed that CDR internalization significantly increased calpain-2 protein expression, but not calpain-1 which indicates a intracellular Ca²⁺ level above 0.250 mM [75]. Inhibiting calpain activity by calpain antagonist MDL28170 reduced the CDR-induced CB loss and blocked calpain-2 protein enhancement. Interestingly, the CDR2L induced pathology is less affected by calpain inhibition than CDR2. This may be explained by the distribution of CDR2 (cytosol) and CDR2L (membrane) proteins [22]. We hypothesize that CDR internalization affects activity-dependent modifications of synaptic integrity, stability, and function of target proteins, which modulates the synaptic Ca²⁺ transient by mediating over-activation of calpain-2 [5, 17]. As calpain-mediated truncation of substrates is regulated by their phosphorylation state, there is a possibility of cross talk between calpain activation and activation of mitogen-activated protein (MAP) kinase. MAP kinase signaling cascades are associated with PKC-dependent synaptic depression and declustering of receptors, inflammation, death receptors’ activation, apoptosis, and oxidative stress [12, 23, 35]. Althought PKC inhibition prevented CDR-induced CB and L7/Pcp-2 loss, blockage of MAP kinase activity did not show any rescue effects.