Epilepsy protein Efhc1/myoclonin1 is expressed in cells with motile cilia but not in neurons or mitotic apparatuses in brain

EFHC1 gene encodes the myoclonin1 protein, also known as Rib72-1. Pathogenic variants in EFHC1 have been reported in patients with juvenile myoclonic epilepsy (JME). Although several studies of immunohistological investigations reproducibly showed that the myoclonin1 is expressed in cells with flagella and motile cilia such as sperm, trachea and ependymal cells lining the brain ventricles, whether myoclonin1 is also expressed in neurons still remains controversial. Here we investigated myoclonin1 expression using widely-used polyclonal (mRib72-pAb) and self-made monoclonal (6A3-mAb) anti-myoclonin1 antibodies together with Efhc1 homozygous knock-out (Efhc1−/−) mice. All of the western blot, immunocytochemical, and immunohistochemical analyses showed that mRib72-pAb crossreacts with several mouse proteins besides myoclonin1, while 6A3-mAb specifically recognized myoclonin1 and detected it only in cells with motile cilia but not in neurons. In dividing cells, mRib72-pAb signals were observed at the midbody (intercellular bridge) and mitotic spindle, but 6A3-mAb did not show any signals at these apparatuses. We further found that the complete elimination of myoclonin1 in Efhc1−/− mouse did not critically affect cell division and migration of neurons in cerebral cortex. These results indicate that myoclonin1 is not expressed in neurons, not a regulator of cell division or neuronal migration during cortical development, but expressed in choroid plexus and ependymal cells and suggest that EFHC1 mutation-dependent JME is a motile ciliopathy.

www.nature.com/scientificreports/ cells and choroid plexus in mouse 14 and in cytoplasm, nuclei, centrosome, mitotic spindle and midbody of cultured cells [14][15][16][17] . They further reported that a suppression of Efhc1 by small hairpin RNAs (shRNAs)-mediated RNA interference (RNAi) in cultured cells or rat embryonic brain caused disruption of mitotic spindle structure, impaired M-phase progression, increase of apoptosis, impaired cell cycle exit of cerebrocortical progenitors, defective radial glia scaffold organization, impaired locomotion of postmitotic neurons, and marked disruption of radial migration 17 . With these results, they proposed that myoclonin1 is a regulator of cell division and neuronal migration during cortical development and that disruption of its function leads to JME 17 . In order to further investigate the above-mentioned discrepancy for the distribution of myoclonin1, in the present study we carefully re-examined the histological and cytological distributions of myoclonin1 in mouse brain and cultured cells by using the mRib72-pAb and 6A3-mAb antibodies together with the Efhc1 −/− mouse. Our results show that the mRib72-pAb signals in neurons are non-specific and myoclonin1 is expressed in cells with motile cilia but not in neurons and that Efhc1-deficiency causes no apparent abnormalities in cell division, radial glia scaffold organization and apoptosis in brain.

Results
The 6A3-mAb specifically detects myoclonin1, but mRib72-pAb non-specifically crossreacts with additional proteins besides myoclonin1. To verify the specificity of 6A3-mAb monoclonal and mRib72-pAb polyclonal antibodies, we at first performed western blot analyses. The 6A3-mAb successfully detected a 75 kDa band of myoclonin1 in brain and lung tissue lysates of wild-type (WT) mouse, and these bands well disappeared in Efhc1 −/− mouse (Fig. 1A-left and Supplementary Fig. S1). The 6A3-mAb did not show apparent extra bands. In contrast, although mRib72-pAb was able to detect the myoclonin1 band in lung which disappeared in that of Efhc1 −/− mouse, mRib72-pAb detected additional bands with much higher intensities than that of myoclonin1 and these bands remained in Efhc1 −/− mouse ( Fig. 1A-right). The mRib72-pAb hardly detected the 75 kDa band of myoclonin1 in brain of WT mice, possibly because of low sensitivity (Fig. 1A-right). Previous studies [14][15][16] showed immunosignals of mRib72-pAb in mitotic apparatuses such as mitotic spindle and midbody in cultured cells including mouse neurosphere cells (NSC) and human embryonic kidney (HEK) cells. We re-investigated whether mRib72-pAb and 6A3-mAb can specifically detect myoclonin1 in NSC and HEK cell cultures. In western blot analyses of NSC cell lysates, 6A3-mAb again successfully detected the 75 kDa band in WT mouse, and the band well disappeared in Efhc1 −/− mouse (Fig. 1B-top). In contrast, mRib72-pAb detected multiple sized bands in both WT and Efhc1 −/− mice (Fig. 1B-bottom). In order to investigate the nature of Figure 1. The 6A3-mAb specifically detects myoclonin1 at ~ 75 kDa, while mRib72-pAb detects multiple non-specific signals besides myoclonin1. (A) Western blots of brain and lung lysates probed with 6A3-mAb or mRib72-pAb (two independent experiments, N = 1 WT and 1 Efhc1 −/− ). Myoclonin1 was detected by both antibodies at the expected size (~ 75 kDa, arrows) in the lysates from WT, but not in those of Efhc1 −/− mouse. The 75 kDa band by mRib72-pAb was quite weaker than that of 6A3-mAb in the lung, and it was hardly detectable in the brain. The mRib72-pAb also detected additional bands (asterisks) that are much more intense than that of myoclonin1, and those remained in Efhc1 −/− mouse. Mouse IgG in mouse tissue lysates was detected by anti-mouse IgG secondary antibody (arrow heads in left panel). (B) Western blots of lysates from NSC and HEK cultured cells (two independent experiments, N = 1 WT and 2 Efhc1 −/− ). The 6A3-mAb detected 75 kDa band in cultured mouse neurosphere cells (NSC, arrow) from WT and HEK cells (double arrow), and this band well disappeared in Efhc1 −/− mouse. The molecular size of human myoclonin1 (640 amino acids, a.a.) in HEK cells is a little smaller than that of mouse myoclonin1 (648 a.a., GenBank accession number: ACB20692). In contrast, the mRib72-pAb detected multiple bands those remained in Efhc1 −/− mouse. An antibody to GAPDH was used as a control and shown in the lower panels (A, B).  (Table 1). As a control, Spot 3 (37 kDa) was detected by anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody (Supplementary Fig. S2A-middle), and confirmed to be GAPDH by LC-ESI-MS/MS (Table 1). These results indicate that mRib72-pAb crossreacts with non-myoclonin1 proteins with high affinities.

Myoclonin1 is not expressed in neurons but cells with motile cilia.
Previous studies 14,17 showed that mRib72-pAb immunosignals in neurons. Our immunohistochemistry on pre-and postnatal mouse brain sections also showed mRib72-pAb signals in cortical neurons from WT, but those signals remained in Efhc1 −/− mouse ( Fig. 2A, B). In contrast, 6A3-mAb did not show any signals in cerebral cortex in both mice ( Fig. 2A). Meanwhile, both mRib72-pAb and 6A3-mAb showed intense signals at the ependymal motile cilia in WT and these signals well disappeared in Efhc1 −/− mouse ( Fig. 2A), the observations are consistent to our previous study 10 . These results indicate that both mRib72-pAb and 6A3-mAb well detects myoclonin1 at ependymal motile cilia, but the mRib72-pAb immunosignals in neurons are non-specific.

Myoclonin1 does not localize at mitotic apparatuses.
As reported previously 14 , immunocytochemistry on cultured NSC from WT mouse, those were well positive for Nestin (marker for neural stem cells), showed that the mRib72-pAb surely developed immunosignals at cytoplasm (Fig. 3A) and at mitotic spindles during cellular mitosis (Fig. 3B). However, these signals remained in Efhc1 −/− (Fig. 3A, B). In dividing HEK cells, mRib72-pAb signals were also observed at the midbody ( Fig. 3C) as reported in previous studies [14][15][16] , but 6A3-mAb did not show any signals at these mitotic apparatuses (Fig. 3D). These results indicate that myoclonin1 does not localize at mitotic apparatuses such as mitotic spindle and midbody.

Myoclonin1 deficiency does not critically affect cortical development. The Belgium group
reported that a suppression of Efhc1 by shRNA-mediated RNAi in rat embryonic brain caused disruption of mitotic spindle structure, increased apoptosis, impaired locomotion of postmitotic neurons, and marked disruption of radial migration 17 . We investigated whether Efhc1 −/− mouse has any abnormalities in cerebrocortical progenitors, locomotion of postmitotic neurons, or radial glia scaffold organization by using antibodies for SOX2 (a marker for progenitor cells), phospho-Histone H3 (PH3; a marker for mitotic cells), and brain lipidbinding protein (BLBP; a marker for radial glia) those were used in the previous study 17 . We did not observe any apparent differences in the distribution of immunopositive cells detected by these antibodies between WT and Efhc1 −/− mice (Fig. 4A, B). We also performed TUNEL assay on brain sections, however it revealed no differences between WT and Efhc1 −/− (Fig. 4C). These results indicate that the elimination of myoclonin1 does not largely affect cell cycle exit of cerebral cortical progenitors, radial glia scaffold organization and apoptosis.

Discussion
The present study described a re-evaluation of expression profile of myoclonin1 and the discrepancy in histological phenotypes between the Efhc1 constitutive knock-out in mice 10,12 and acute shRNA-mediated knock-down in rats 17 . We confirmed that myoclonin1 is predominantly expressed in ependymal motile cilia but not in neurons of brain. These present and previous results 10,12 are consistent to the previous in situ hybridization analysis showing that Efhc1 mRNA was predominantly appeared in the ependymal and choroid plexus epithelial cells but not in neurons in mouse and rat brains 13 . We also showed that myoclonin1 is not expressed in mitotic apparatuses such as mitotic spindles and midbody in dividing cells, and that the complete elimination of myoclonin1 did not critically affect cell division and migration of neurons in the cerebral cortex of Efhc1 −/− mouse. These phenotypic discrepancy between the rat with acute knock-down of Efhc1 by shRNA with drastic alterations in cell division and neuronal migration in embryonic cerebral cortex 17 and the Efhc1 −/− mouse without those alterations in present and previous studies 12 is most possibly explained by the off-target effect of shRNA 18 . Our present results therefore deny the proposal of the Belgium group that myoclonin1 is expressed in neurons, radial glia cells, mitotic apparatuses, and plays critical roles in neuronal cell division and migration during cortical development [14][15][16][17] . Our Table 1. mRib72 immunoreactive proteins were not myoclonin1 isoforms. a MASCOT score is defined as − 10 × Log(P), where P is the probability that the observed match is a random event. Scores greater than 67 are significant (p < 0.05). www.nature.com/scientificreports/ www.nature.com/scientificreports/ present and previous studies 10,12,13 may also deny our own previous proposal that the interaction of myoclonin1 and R-type voltage-dependent Ca 2+ channel (Ca v 2.3) in neurons plays a role in JME 1 . Meanwhile, our another proposal that the functional interaction of myoclonin1 and the transient receptor potential M2 channel (TRPM2) plays a role in the pathogenesis of JME 19 may still survive as a possible pathomechanism of JME because of the TRPM2 expression in ependymal cells in addition to neurons.
Recently we identified ciliogenesis associated kinase 1 (CILK1), also known as intestinal-cell kinase (ICK), as another gene responsible for JME 20 . Interestingly, CILK1 is again highly expressed in choroid plexus and ependymal cells 20 . PRICKLE1 and PRICKLE2 gene mutations have been identified in patients with JME and other types of myoclonic epilepsies 21,22 , and disruptions of these genes in zebrafish, D. melanogaster or mouse lead to increased seizure susceptibility 22,23 . Interestingly again, these genes are correlated to ciliogenesis or ciliary functions [24][25][26] . These observations further support the notion that impairments of cells with motile cilia in brain cause JME.
Efhc1 −/− mice developed frequent spontaneous myoclonus, decreased seizure threshold, and reduced ciliary beating frequencies (CBF) of postnatal ependymal motile cilia 12 as well as neonatal choroid plexus epithelial cilia 27 . Although the reduced CBF in Efhc1 −/− mice may not be directly relevant to the pathogenesis of JME which has been inferred from the inconsistency of seizure susceptibility and CBF reduction in heterozygous Efhc1 +/− mice 12 , other possible impairments of motile cilia, ependymal cells or choroid plexus (e.g. sensory antenna, ion exchange, cerebral spinal fluid (CSF) secretion, or pH of CSF, etc.) could be the causes of JME. Further studies are warranted to figure these out. www.nature.com/scientificreports/ In conclusion, our results presented here indicate that myoclonin1 is not expressed in neurons, not a regulator of cell division or neuronal migration during cortical development, but expressed in cells with motile cilia in brain and therefore suggest that EFHC1-dependent JME is a motile ciliopathy. www.nature.com/scientificreports/

Materials and methods
Animal experiments. All animal experimental protocols were approved by the Animal Experiment Committee of Institute of Physical and Chemical Research (RIKEN). All animal breeding and experimental procedures were performed in accordance with the ARRIVE guidelines and the guidelines of the Animal Experiments Committee of RIKEN. Animals were maintained on 12 h light/dark cycle with ad libitum access to food and water at the Research Resources Division (RRD) of the RIKEN Center for Brain Science. Efhc1-deficient mouse was described previously 12 . The heterozygous mice were maintained on the C57BL/6J (B6J) background, and the resultant heterozygous mice were interbred to obtain WT, heterozygous, and homozygous mice. Genotyping was carried out as described previously 12 .

In-gel digestion for LC-ESI-MS/MS and mass analysis.
TUNEL assay. Paraffin (6-μm-thick) sections of mouse brains (N = 2 WT and 2 Efhc1 −/− for E14.5; N = 1 WT and 1 Efhc1 −/− for E16.5) were used in the assay. Apoptotic cells in paraffin sections were detected by using the DeadEnd Colorimetric TUNEL System (Promega). For a negative control, sections were incubated in buffer without the recombinant terminal deoxynucleotidyl transferase (rTdT) enzyme. DNase I-treated sections were used as positive controls. Images were acquired by the Biozero BZ-X710 (KEYENCE) microscope.