The sorting receptor Rer1 controls Purkinje cell function via voltage gated sodium channels

Rer1 is a sorting receptor in the early secretory pathway that controls the assembly and the cell surface transport of selected multimeric membrane protein complexes. Mice with a Purkinje cell (PC) specific deletion of Rer1 showed normal polarization and differentiation of PCs and normal development of the cerebellum. However, PC-specific loss of Rer1 led to age-dependent motor deficits in beam walk, ladder climbing and gait. Analysis of brain sections revealed a specific degeneration of PCs in the anterior cerebellar lobe in old animals. Electrophysiological recordings demonstrated severe deficits in spontaneous action potential generation. Measurements of resurgent currents indicated decreased surface densities of voltage-gated sodium channels (Nav), but not changes in individual channels. Analysis of mice with a whole brain Rer1-deletion demonstrated a strong down-regulation of Nav1.6 and 1.1 in the absence of Rer1, whereas protein levels of the related Cav2.1 and of Kv3.3 and 7.2 channels were not affected. The data suggest that Rer1 controls the assembly and transport of Nav1.1 and 1.6, the principal sodium channels responsible for recurrent firing, in PCs.

Scientific RepoRts | 7:41248 | DOI: 10.1038/srep41248 motor phenotype and degeneration of PCs in old mice. The underlying molecular cause is a reduction in Na v 1.1 and 1.6 protein levels, suggesting that Rer1 is involved in the quality control of Na v ion channels.

Rer1-deficiency does not affect normal growth and differentiation of Purkinje cells. Rer1
is expressed throughout the rodent brain. In-situ hybridization and β -Gal activity from a heterozygous gene-trap allele inserted at the Rer1 locus 13 indicated a high level of expression in cerebellar PC (Fig. 1a). Immunofluorescence staining of PC with a Rer1-specific antibody demonstrated an early Golgi-localization pattern (Fig. 1a,c, supplemental Fig. 1), as shown in other cell types 12,13 . Mice homozygous for the gene-trap allele inserted at the Rer1 locus are embryonic lethal at the blastocyst stage 13 . Therefore, to study the function of Rer1 in vivo we established a conditional knock-out mouse line by inserting loxP sites in intron 2 and 4 (Fig. 1b). Crossing these mice with L7/pcp-2:Cre, a PC specific cre-line 21 , resulted in offspring with specific Rer1 deletion in PC (Rer1 ΔPC ), but not cerebellar granule neurons or other cell types (Fig. 1c). Rer1 ΔPC mice are viable, fertile, have normal body weight and normal activity (data not shown). Cre-mediated deletion of Rer1 in PC was clearly observed at P13, and was completed at P16 (Fig. 1d). The cerebellum was normally developed without obvious differences in PC number or dendritic tree arborization (Fig. 1c). An indicator for proper development of PCs is the thickness of the molecular layer, because this is primarily determined by the dendritic tree of PCs 22,23 . No differences in mean thickness were observed in P16-25, P36 and 3 monthss old Rer1 ΔPC and control littermates mice (supplemental Fig. 2). Differentiation of PCs (and the cerebellum) is completed at around P21, after cre-recombination. Therefore, these data suggest that Rer1 is not involved in maturation of PCs.
Rer1-deficiency in Purkinje cells results in severe motor deficits. Phenotypic analysis of young (2-3 months) mice on a Rotarod showed no difference between Rer1 ΔPC mice and their non-cre expressing (a) Left: radioactive in-situ hybridization (ISH) of a wt cerebellar section with a Rer1-specific antisense probe; middle: X-Gal staining of a cerebellar slice of a heterozygous Rer1-genetrap mouse with a β -Geo insertion in intron 3 of Rer1 13 . Arrows indicate PC layer; right: immunostaining of a cerebellar slice with Rer1 antibody (red) and calbindin (calb, green). Asterisks indicate PCs. Scalebar 150 μ m (left, middle) and 10 μ m (right). (b) Targeting strategy for conditional deletion of Rer1. Shown is the genomic structure (top), the position of the loxP sites in the Rer fl/fl allele and the deleted Rer1 ΔPC allele after crossing with the PC-specific L7Pcp2-Cre line. (c) Rer1 is specifically deleted in PCs. Vibratome sections of perfused 4 months old animals with the indicated genotype were processed for immunofluorescence with antibodies against Rer1 and calbindin (Calb) and imaged by confocal microscopy. Three consecutive confocal sections were merged. In the lower panel a magnification of the Rer1 staining of the boxed area is shown. PCs are outlined. Scalebar 10 μ m. (d) Deletion of Rer1 in PCs starts before P13 and is completed at P16. Sagittal sections from cerebella from perfused Rer1 ΔPC of indicated age were stained for Rer1 and images were taken from the anterior lobe at the position indicated in the overview Hoechst-stained image on the right. Scale bar 50 μ m for Rer1 staining, 150 μ m for Hoechst.
littermates, independent of sex (supplemental Fig. 3b and not shown). However, footprint analysis of 14 and 24 months old mice revealed significant gait differences (Fig. 2a-c). When climbing a ladder, 8-12 months old Rer1 ΔPC mice made 2.5 times more mistakes than their littermates (2.6 ± 0.7 lapses in 12 Rer1 ΔPC vs. 0.04 ± 0.04 lapses in 12 control mice, t-test P = 0.006). Finally, challenging the mice on a 1 cm beam walk demonstrated that Rer1 ΔPC mice needed significantly more time to cross the beam, and their difficulties increased with age (Fig. 2d). Motor learning, as judged from the improvement in performing Rotarod and beam walk 24 was not affected in Rer1 ΔPC mice (suppl. Fig. 3).
Taken together, the motor analysis clearly demonstrated that deletion of Rer1 in PCs results in a severe, age-dependent gait disorder.
Rer1-deficiency in Purkinje cells results in their age-dependent loss. The progressive motor deficits prompted us to analyze cerebellar morphology in young and old mice of both genotypes. The overall size of the cerebellum was not different until 3-5 months of age. However, starting at the age of 6-12 months the anterior part of the cerebellum was smaller in Rer1 ΔPC mice. In old mice (≥ 17 months) the difference was statistically significant (Fig. 3a,b). Calbindin staining revealed an age-dependent, progressive loss of PCs in the anterior part starting at 5 months, where also axonal swellings were detected (Fig. 4a). Glial cells were unchanged, as indicated by GFAP co-staining (Fig. 4a). The axonal swellings, suggestive of sick or dying neurons, became visible in 4 months old Rer1 ΔPC mice, and were frequently detected in 9 months old Rer1 ΔPC mice but not in littermates (Fig. 4b). These data show that loss of Rer1 in PCs results in an age-dependent PC degeneration, which in turn leads to a selective volume reduction of the anterior cerebellum. (a-c) Hind paws of mice of indicated age and genotype were stained with ink and the mice were allowed to walk across a paper. Base width and stride length were determined from the footprints. Displayed is the mean from 3 ctrl and 3 Rer1 ΔPC 14 months (mo) old mice, 5 ctrl and 5 Rer1 ΔPC 24 mo old mice ± SEM. (d) Mice of indicated age and genotype were placed on three consecutive days on a 1 cm broad, 1 m long beam and the time needed for crossing was determined. Displayed is the mean time of 4 ctrl and 4 Rer1 ΔPC 4-6 mo, 4 ctrl and 5 Rer1 ΔPC 8 mo, 6 ctrl and 6 Rer1 ΔPC 10-11 mo old mice ± SEM. Ctrl refers to Rer fl/fl /cre −/− littermates.
Scientific RepoRts | 7:41248 | DOI: 10.1038/srep41248 Action potential generation is impaired in Rer1-deficient Purkinje cells. The overall morphology of PCs lacking Rer1 was not changed (Fig. 1c). To test whether their electrophysiological properties were affected, patch clamping of PCs of cerebellar slices of 6-7-weeks old Rer1 ΔPC and Rer1 wt mice was performed in the loose cell-attached mode. High-frequency firing was strongly attenuated in Rer1 ΔPC PCs (Fig. 5a,c; Rer1 wt : 44.4 ± 7.4 Hz; Rer1 ΔPC 8.5 ± 3.1 Hz; Student's t-test: p = 0.000013; n = 22/31). The percentage of non-spiking PCs was 7-fold increased in slices of Rer1 ΔPC compared to preparations of control mice ( Fig. 5b; control: 9.1%; Rer1 ΔPC 64.5%). After exclusion of all non-spiking neurons from both groups, the analysis still revealed a significant diminished firing frequency of Rer1 ΔPC PCs compared with neurons from control slices (Fig. 5a,c; control: 48.8 ± 7.4 Hz; Rer1 ΔPC 24.1 ± 7.7 Hz; Student's t-test: p = 0.042; n = 20/11). Next we measured resurgent currents, a characteristic feature of PCs caused by a special form of sodium channel gating 25 in cerebellar slices of 7-8 week old Rer1 ΔPC mice and control littermates (Fig. 5d). The amplitude of the resurgent current was significantly reduced in Rer1 ΔPC PCs indicating a decrease in density of Na v channels at their surface. Current-voltage relations for peak resurgent currents showed a reduced peak amplitude in Rer1 ΔPC mice (Fig. 5e). Interestingly, the decay constant tau at − 30 mV was unchanged (11.1 ± 1.9 msec (wt) and 11.5 ± 3.03 msec (ko); Student's t-test, p > 0.05), suggesting a decrease in density of Na v channels at the cell surface, not changes in kinetic properties of individual channels. These data suggest that deficits in resurgent currents severely affect spontaneous action potential generation in Rer1 ΔPC PCs because of reduced amounts of Na v ion channels at the AIS.

Rer1-deficient Purkinje cells have normal axonal initial segments with reduced Na v 1.6 levels.
The axonal swellings, the motor phenotype as well as the high-frequency-firing deficits are similar to the phenotype of mice with a PC-specific Na v 1.6-deletion 26 . Action potentials are generated at the AIS, where Na v 1.6 is (b) Anterior and posterior parts of cerebella of indicated age and genotype were marked and their area determined and displayed as mean ± SEM. n = 6 (P16-36); n = 6 (3-5 mo); n = 3 (6-12 mo) and n = 7 (≥ 17 mo); 3-4 sections for each mouse were analyzed. a.u., arbitrary units. enriched 2 . We therefore tested whether the structure of the AIS or its molecular composition is dependent on Rer1 and performed co-immunostainings with ankyrin G and Na v 1.6 ( Fig. 6a, for specificity of Na v 1.6 antibody see supplemental Fig. 4). No gross difference was observed in ankyrin G and Na v 1.6 distribution between Rer1 ΔPC mice and control littermates and the average AIS-length didn't differ between the two genotypes (14.2 ± 0.3 μ m in control, 14.1 ± 0.3 μ m in Rer1 ΔPC mice; n = 243 AIS from 7 ctrl mice and n = 267 AIS from 8 Rer1 ΔPC mice). However, the average intensity of Na v 1.6 at the AIS was significantly diminished in PCs of Rer1 ΔPC mice, sometimes below the detection limit (Fig. 6a, long arrows). From 152 AIS from control littermates (n = 6) 9,95 ± 1,6% had very low or undetectable Nav1.6 immunostaining. In contrast, from 169 AIS from Rer1 ΔPC mice (n = 6) almost three times more AIS (26,3 ± 6%) had very low or undetectable Nav1.6 immunostaining. Consistently, cerebellar lysates of Rer1 ΔPC mice had slightly reduced levels of Na v 1.6 at P23, before overt loss of PCs (not shown). Since PC only constitute a minor fraction of the cerebellum, and because of the specificity of the deletion, large differences in total cerebellar lysates are unlikely. We therefore made use of another mouse line in which Rer1 is deleted in the whole brain using nestin-cre (Rer Δbrain ). These mice die early postnatally. Brain lysates from P3, P10 and P15 Rer Δbrain mice showed almost complete absence of Rer1, demonstrating successful deletion  Fig. 7a). Rer1 deficiency does not change general brain architecture, indicated by the unchanged protein levels of synaptophysin (p38), PSD95, APP and the early Golgi marker GM130 in brain lysates (Fig. 7a) and by histological analysis (not shown). However, Rer Δbrain mice displayed strong reductions in Na v 1.6 expression levels (Fig. 7a). Also the protein levels of the closely related Na v 1.1 are strongly reduced. In contrast, protein levels of an abundant calcium channel from the same ion channel superfamily, Ca v 2.1, and of three tested potassium channels K v 1.1, 3.3 and 7.2, as well as the levels of a K-Cl Cotransporter KCC2, are not changed in Rer Δbrain mice. Protein levels of the Na v β 1 subunit were only slightly reduced in Rer Δbrain mice (Fig. 7a). Na v 1.1 and 1.6 are complex glycosylated. De-glycosylation using endoglycosidase H (endoH, cleaves only immature sugars on ER/early Golgi-resident proteins) and PNGase F (F, removes all N-linked sugars) showed that the higher, prominent band in Na v 1.1 and Na v 1.6 is endoH resistant (Fig. 7b). This fraction corresponds to the mature channel at the PM, while the weaker, faster migrating band is endoH sensitive, corresponding to the immature fraction of the channel in the ER/early Golgi that is not yet fully folded and/or assembled (Fig. 7b). Only the mature Na v 1.1 and 1.6 levels are strongly reduced in the absence of Rer1 and no accumulation of immature channels is observed (Fig. 7a). This suggests that in the absence of Rer1 misfolded and/or unassembled Na v 1.1 and 1.6 escape the ER quality control and are degraded in later compartments, similar to the acetylcholine receptor 13 . Rer1 is not involved in controlling Ca v or K v channels and not or only to a minor degree in β 1.
Taken together, we show that loss of Rer1 in PC elicits a very specific phenotype with a loss of functionality of PCs causing severe motor deficits and age-dependent degeneration. Rer1 is essential for the key property of PCs, their ability to generate high-frequency action potentials, by specifically controlling protein levels of mature Na v ion channels.

Discussion
The PC, and a neuron in general, is an incredibly complex information processing entity. It only functions properly if a plethora of proteins and mechanisms interact like cogs mesh in a gearbox. For example, for generation of the resurgent Na + current that is essential for action potentials in PCs, a specific voltage-dependent Na-channel, the Na v 1.6, with its multiple TM domains needs to be co-translationally inserted into the ER, properly folded, assembled with the correct β -subunit(s), transported through the Golgi, sorted into vesicles destined for the AIS, inserted there, and fulfill its function as an ion channel. We here show that a disturbance at one or a few of these steps has dramatic consequences. Loss of Rer1, and thus disturbance of the quality control system that ensures proper folding and assembly of multimeric complexes, leads to lower amounts of mature Na v 1.1 and 1.6 in PCs. This results in almost complete failure of the constant firing of action potentials, which in turn results in the severe motor phenotypes we observed in older mice. PCs fail to generate action potentials at a very young age, and maybe were never able to do so. We could not determine the earliest time point where motor problems can be detected. Yet, the development of motor phenotypes is clearly age-dependent, similar to other cerebellar mouse mutants 27 , probably because of initial compensation for loss of function. In contrast, the early complete loss of PCs at P15-30 in the Pcd mice results in early ataxia 28,29 and problems on the Rotarod already at P20 30 . In mice with a PC-specific deletion of Na v 1.6 early ataxia and impaired Rotarod performance was observed at 6-8 wks 26 . Why our Rer1 ΔPC do not show motor coordination problems on the Rotarod is unclear, but could be due to the rather insensitive Rotarod 29 or because of residual Na v 1.1 and 1.6 that correctly assemble and make it to the AIS in the absence of Rer1. Interestingly, motor learning as assessed from the learning curves on Rotarod and beam walk was not affected by the deletion of Rer1, similar to the tambaleante mice, where PCs are lost after 4-6 months 24 . However, for a detailed analysis of motor learning eyeblink classical conditioning needs to be performed, as for example described in ref. 31.
PCs without Rer1 are normally developed and maintained to young adulthood. However, they start to degenerate in adult mice, maybe as a consequence of the loss of functionality. This degeneration leads to reduction of cerebellar size in old mice and probably contributes to the severe motor phenotypes. Interestingly, the degeneration affects mainly the anterior lobe, although Rer1 is deleted in the whole cerebellum. The underlying cause is not known, but patterned or area-selective cell death of PC has been frequently observed (reviewed in ref. 32). The most prominent pattern in PCs is the pattern of zebrin-positive and zebrin-negative zones, with the PCs in the respective zone having distinct electrophysiological properties 33 . Expression of Rer1 is similar in PCs from all areas of the cerebellum and is not patterned. Unlike for example in model mouse lines of Niemann-Pick Type C disease 34,35 , in Rer1 ΔPC mice PC degeneration seems uniform and not patterned. However, since in the anterior lobe the zebrin-negative areas dominate (where we see the prominent PC loss), a potential correlation has to be analyzed in more detail in future studies.
The evidence that disturbances in voltage-gated sodium channels Na v 1.1 and 1.6 homeostasis are the main cause for the observed phenotype is several fold. PC-specific deletion of Na v 1.6 26 , loss-of-function-mutation in Na v 1.1 5 and deletion of Rer1 (this study) in mice caused very similar phenotypes. All models displayed severe motor deficits caused by a strongly reduced firing rate of PCs. Both Levin et al. 26 and we observed swellings in axons of PCs and we additionally observed PC degeneration and cerebellar shrinkage in older mice, which was not analyzed by Levin et al. 26 . Expression levels of Na v 1.1 and 1.6 are strongly reduced in the absence of Rer1. The protein levels of β 1 are only slightly reduced, suggesting it is not a direct substrate of Rer1. In addition, it was shown that loss of scn1b (coding for the β 1 subunit) leads to a reduction in Na v 1.6 with a concomitant up-regulation of Na v 1.1 36 , which we did not observe in our mice. This argues against Rer1 acting via β 1 on Na v 1.1 Scientific RepoRts | 7:41248 | DOI: 10.1038/srep41248 and 1.6. The β 2-subunit does not seem to be involved in assembly of Na v . Deletion of its gene, scn2b, causes only mild phenotypes unrelated to the motor deficits observed here 37 . Rer1 seems to act specifically on Na v channels, not on closely related Ca v and on K v channels.
A number of substrates for Rer1 have been identified 13,[16][17][18][19] , but in vivo evidence for Rer1 function in mammals was scarce. We here provide in vivo evidence of the role of Rer1 in a specific cell type, the murine PC. Importantly, the deletion of Rer1 in our model occurs at a stage where the PCs (and the cerebellum) are not yet fully developed. Yet, loss of Rer1 had no observable effect on the thickness of the molecular layer, indicative of a normally developed dendritic tree 22,23 . Altogether, this strongly suggests that major developmental pathways are not affected and that Rer1 is not a general sorting receptor. Instead, Rer1 probably has only a very limited number of substrates. Specifically in PC the most prominent substrate seems to be Na v s, but in the current study we cannot rule out that there are additional substrates of Rer1 in PC. Using morpholinos and siRNA against Rer1 in zebrafish and mammalian cells, respectively, resulted in shorter primary cilia and developmental defects 38 . We find no evidence for a developmental role of Rer1 in PC, but cannot rule out that a deletion at an earlier time point would have an effect. Primary cilia in PCs were difficult to detect in our hands, but primary cilia length was not changed in the hippocampus of brains with forebrain-specific Rer1 deletion (by CamKII-cre; mean ciliary length in dendrate gyrus was 5.5 ± 0.2 pixels in wt, 6.0 ± 0.3 pixels in KO mice, in CA1 region 10.4 ± 0.5 pixels in wt, 11.4 ± 0,4 pixels in KO mice). This suggested Rer1 is not involved in cilia function in mammalian postnatal brain, but might be important for cilia formation in different organs or organisms.
Further work is needed to characterize the exact mechanism of Rer1 involvement in Na v biogenesis. In analogy to the role of Rer1 in acetylcholine receptor assembly 13 , we hypothesize that the assembly rates of Na v are reduced. Free subunits, normally returned by Rer1, escape the ER/cis-Golgi and get degraded, probably in lysosomes, and this results in the reduced total levels of mature Na v .
Mutations in SCN1A (encoding Na v 1.1) are among the most frequent mutations in epilepsy 39,40 and mutations in SCNA8A, the gene encoding Na v 1.6, were found in patients with epileptic encephalopathy and mental disabilities 41 . Mutations leading to increased channel activity might lead to seizures, whereas reduced activity might lead to intellectual disabilities 41 . Not surprisingly, Na v s are major drug targets 42 . Future work has to elucidate to what extent Rer1 controls Na v channels also in other brain areas besides the cerebellum and whether it might be a novel therapeutic target. Increasing Rer1 levels or stabilizing its interaction with Na v might overcome deficits caused by loss-of-function or hypomorphic mutations, similar to other rescue approaches 43,44 . A partial rescue might be sufficient, as has been shown in the case of mutations in Na v 1.1 where a partial rescue by lowering temperature or co-expressing binding proteins resulted in a constitutive gain-of-function 43 . In case of over-active mutations reducing Rer1 levels or Na v 1.6 interactions might be beneficial.
In conclusion, we here present the first in vivo data how a member of the ER/Golgi quality control machinery, the sorting receptor Rer1, controls a specific class of voltage-gated sodium ion channels that are essential for the electrical properties of cerebellar Purkinje cells.

Methods
Antibodies. The following antibodies were used. Anti-Rer1, rabbit polyclonal SA5457 and SPY008 13  Targeted inactivation of murine Rer1. The Rer1 conditional KO mice were generated by TaconicArtemis (now Taconic Biosciences GmbH), Cologne, Germany. Exons 3-4 were flanked by LoxP sites. Positive selection markers were flanked by FRT (NeoR) and F3 (PuroR) sites and inserted into intron 2 and intron 4, respectively. A targeting vector was generated using BAC clones from a C57BL/6J RPCI-23 BAC library and transfected into TaconicArtemis C57BL/6N Tac ES cell line. After isolation of homologous recombinant clones using positive (Neomycin and Puromycin resistances) and negative (Thymidine kinase -Tk) selections, the conditional KO allele was obtained after Flp-mediated removal of selection markers. Chimeric mice were generated by blastocyst injection, germline transmitters selected and kept in C57Bl/6 background. For conditional inactivation of Rer1, Rer1 fl/fl mice were crossed with L7/pcp-2:Cre mice (for specific Purkinje cell deletion 21 ) or with nestin-cre mice (for CNS-specific deletion 46 ) or with CamKII-cre (for forebrain-specific deletion 47 ).
Mice. All animal experiments were conducted according to the German animal welfare legislation and approved by the Thüringer Landesverwaltungsamt (reg. nr 03-008/12). Mice were fed ad libitum with standard laboratory chow and water in ventilated cages under a 12 h light/dark cycle.

Motor coordination tests. Gait analysis.
Female mouse hind paws were painted with non-toxic, water-soluble ink and footprint patterns were analyzed using a runway (80 cm ×10.5 cm wide) with white paper on the bottom. Five consecutive strides were measured from three runs of each animal. Ladder-climbing. Female mice were habituated to climb to their home cage on a 1,1 m long ladder inclined to 22,5° angle with steps 2 cm apart and the average number of lapses was counted from video-recordings from 3 runs per mice. Male mice behaved identically (data not shown).
Beam walking. Female mice were trained to run along a 1 m long beam (3 cm thick) to their home cage. The test was performed on three consecutive days on a 2 cm thick beam followed by three consecutive days on a 1 cm thick beam with two runs each day. The mice were video-taped and the time to cross the beam and the number of foot slips were measured. Male mice behaved identically (data not shown).
Patch clamp recordings. Coronal cerebellar slices were placed in a submerged recording chamber mounted on an upright microscope (BX51WI, Olympus). Slices were continuously superfused with gassed aCSF (2-3 ml/min, 32 °C, pH 7.3). Loose-patch cell attached recordings (seal resistance < 1 GΩ) were performed under visual control with a 40x water-immersion DIC objective. Purkinje-neurons with typical dendritic trees and large cell bodies were selected for recordings using a patch-clamp amplifier (Multiclamp 700B, Molecular Devices). Responses were low-pass filtered at 4 kHz and digitized at 20 kHz (Digidata 1440, Molecular Devices). All data were acquired, stored, and analyzed using pClamp 10.2 and Clampfit 10.2 (Molecular Devices). Patch pipettes with an impedance of 3-4 MΩ were pulled from borosilicate glass (outer diameter 1.5 mm, Science Products GmbH) on a Sutter Instrument micropipette puller (P-97) and filled with aCSF. Spike frequencies were calculated for each cell from recordings with a minimal duration of 1 min. For voltage clamp recordings of resurgent sodium currents from Purkinje neurons of acute cerebellar slices a slightly modified protocol was used as described previously 49 . As extracellular solution served a modified aCSF (in mM): 110 NaCl, 3 KCl, 1.3 MgSO 4 , 1.25 NaH 2 PO 4 , 2.5 CaCl 2 , 10 TEA-chloride (tetraethylammonium chloride), 0.04 CdCl 2 , 10 D-glucose and 25 NaHCO 3 , gassed with 95% O 2 /5%CO 2 , pH7.3. Patch pipettes were filled with (in mM): 135 CSF, 1 NaCl, 3 KCl, 2 Mg-ATP, 0.5 Na-GTP, 10 HEPES, 5 EGTA adjusted to pH 7.3 with CsOH. After whole-cell configuration was established, cells were recorded. From a holding potential of − 90 mV cells were depolarized to + 30 mV for 3 ms followed by steps from − 50 mV to 0 mV with an increment of 5 mV. For determination of the resurgent current amplitudes 1 μ M tetrodotoxin was applied and the protocol was repeated and subtracted from the initial recording. The time constant (τ ) was determined by monoexponential fitting.
Immunohistochemistry. Mice of either sex were anesthetized with Isoflurane and perfused transcardially with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS). The brain tissue was post-fixed in 4% PFA/PBS overnight at 4° C, washed in PBS and embedded in 4% agarose. 50 μ m thick sagittal brain sections were obtained using a vibratom and were stored in PBS with 0.05% NaN 3 at 4° C.
For immunostainings vibratom sections were mounted on Superfrost ® Plus microscope slides (ThermoScientific) and air dried. For antigen retrieval the sections were placed in pre-warmed 10 mM sodium citrate buffer pH 6.0, briefly boiled and incubated for 10 min at sub-boiling temperature (10%) in a microwave followed by 30 min at room temperature. After washing with PBS the sections were incubated for 1 h in blocking solution (5% normal goat serum, 1% BSA and 0,4% Triton X-100 in PBS) and then overnight at 4° C with primary antibodies in blocking solution, washed with PBS, then incubated for 1 hour at RT with Alexa Fluor-conjugated secondary antibodies (1:1000) and Hoechst and after wash with PBS were mounted in Fluoromount (F4680, Sigma).
In-situ hybridization. Radioactive in-situ hybridization using a full-length mouse Rer1 cDNA in pBluescript was performed as described 50 . Immunoblotting. Brain tissues of mice of either sex were lysed in STEN-lysis buffer (50 mM Tris pH 7.6, 150 mM NaCl, 2 mM EDTA, 1% NP40 and protease inhibitor mix) or in Complexolyte 48 buffer (Logopharm) with protease inhibitor mix using Precellys beads (1,4 mm) and the Precellys 24 lysis and homogenization device for 2 × 15 sec at 5000 rpm (PeqLab). Proteins were separated on 6% SDS-PAGE gels (for detection of Na v , Ca v ), 10% for detection of Kv, KCC2 or 12% gels (for detection of Rer1, Na v β 1, PSD-95, APP, p38 and GM130) and transferred to PVDF membranes. Membranes were cut at appropriate positions and blotted with antibodies as indicated.
Microscopy and image quantifications. For higher resolution immunhistochemistry a Zeiss 510 Meta LSM equipped with 25x/0.8 NA and 40x/1.3NA objectives was used. For overviews of cerebellar morphology an Olympus AX70 with 4x/0.10 NA objective was used. For the primary cilia length and molecular layer thickness determination a Zeiss Observer Z1 equipped with 10x/0.3NA and 20x/0.8NA objective and an Apotome slider was used. The thickness of the molecular layer was measured on 2D images of calbindin-immunostainings of the cerebellum along the III-IV and VIII-IX lobes at six positions at each lobe in 2-3 midsagittal sections per mouse (3 wt and 3 Rer1 ΔPC mice). For measurement of cilia length 3 sagittal sections per mouse (3 wt and 3 Rer1 ΔFN mice) were stained with anti-sstr3 antibody 45 . ApoTome sections from CA1 and DG regions of the hippocampus were taken and 3D reconstructions of were made using ZEN software. The length of cilia (in pixels) was measured using the polygon curve tool. At least 200 cilia were measured per region of the hippocampus. The anterior and posterior cerebellar areas were measured on 2D images of Hoechst-stainings of three mid-sagittal sections per mouse (3 ctrl and 3 Rer1 ΔPC mice). Using NIH ImageJ software the anterior area was measured within a line drawn around the lobes I-V 51 that are anterior from the primary fissure and respectively the posterior area was measured within a line drown around the lobes VI-IX that are posterior from the primary fissure. For measurement of the axon initial segment confocal images from midsagittal sections stained with anti-ankyrin G and anti-Nav1.6 antibodies were obtained using identical settings. Image stacks were converted into a maximal intensity Z-axis projection and measurements were performed. The fluorescence intensity of Na v 1.6 was measured using NIH ImageJ software by drawing a segmented line along the length of the Ankyrin G signal at the AIS. The background fluorescence was subtracted based on a 1 μ m segment distal to the AIS. The AIS fluorescence density was calculated as integrated density divided by the AIS length.
Statistical analysis. Data are always presented as mean + /− SEM and the numbers of experiments or mice (n) are indicated. The unpaired Student's t-test was applied to evaluate differences between experimental groups (GraphPad Prism software). P-values ≤ 0.05 were considered statistically significant. One asterisk P ≤ 0.05; two asterisks P ≤ 0.01; three asterisks P ≤ 0.001.