RREB1 regulates neuronal proteostasis and the microtubule network

Transcription factors play vital roles in neuron development; however, little is known about the role of these proteins in maintaining neuronal homeostasis. Here, we show that the transcription factor RREB1 (Ras-responsive element-binding protein 1) is essential for neuron survival in the mammalian brain. A spontaneous mouse mutation causing loss of a nervous system–enriched Rreb1 transcript is associated with progressive loss of cerebellar Purkinje cells and ataxia. Analysis of chromatin immunoprecipitation and sequencing, along with RNA sequencing data revealed dysregulation of RREB1 targets associated with the microtubule cytoskeleton. In agreement with the known role of microtubules in dendritic development, dendritic complexity was disrupted in Rreb1-deficient neurons. Analysis of sequencing data also suggested that RREB1 plays a role in the endomembrane system. Mutant Purkinje cells had fewer numbers of autophagosomes and lysosomes and contained P62- and ubiquitin-positive inclusions. Together, these studies demonstrate that RREB1 functions to maintain the microtubule network and proteostasis in mammalian neurons.


INTRODUCTION
Although great diversity in cell specificity, molecular disruptions, and behavioral alterations exists among neurodegenerative disorders, protein inclusions are a unifying pathological hallmark associated with many of these diseases (1).The process of regulating proteins in the cell, or proteostasis, is maintained via two protein degradation mechanisms: autophagy and proteasome-mediated degradation.Defects in either autophagy or proteasome function have been associated with the formation of protein inclusions.Inclusion formation and subsequent cell death in sporadic forms of neurodegenerative diseases may be linked directly to dysfunction in proteostasis.
The role of transcription factors has long been examined in the context of cellular differentiation and multicellular organism growth and development; however, recent studies have extended the role of these proteins in regulation of cellular homeostasis in terminally differentiated cells.A few studies have highlighted transcription factors that function in neuron homeostasis and, more specifically, neuronal proteostasis.For example, the transcription factor EB (TFEB) (so named for its recognition and binding to E-box motifs) belongs to the family of microphthalmia/transcription factor E (MiT/TFE) genes, has been shown to control expression of genes involved in autophagosome formation and lysosome biogenesis, and has reduced expression in many neurodegenerative diseases (2)(3)(4).Another MiT/TFE transcription factor, transcription factor binding to immunoglobulin heavy constant mu enhancer 3, is reduced in neuronal nuclei of patients with Parkinson's disease and has recently been shown to positively regulate autophagy (5).Although these studies have illuminated a likely role for transcription factors in maintaining proteostasis in a pathological context, causal mutations in genes encoding these transcription factors have rarely been identified.
Our study demonstrates that hypomorphic loss of the transcription factor RREB1 is associated with the formation of protein inclusions in cerebellar Purkinje cells and subsequent neurodegeneration.Chromatin immunoprecipitation and sequencing (ChIP-seq) and RNA sequencing (RNA-seq) experiments suggest that RREB1 positively regulates the endomembrane system.We also find that RREB1 positively regulates many genes associated with the microtubule cytoskeletal network.Disruption of these genes is associated with changes to the microtubule network as well as impaired dendritic branching.Together, our data demonstrate a unique role for RREB1 in neuron homeostasis.

Rreb1 V7 deficiency is associated with Purkinje cell loss
The new mutant 3888 (nm3888) mutation arose spontaneously in nonobese diabetic (NOD).129P2(Cg)-Il10tm1Cgn /DvsJ mice and was backcrossed to NOD mice to remove the Il10 mutation.Mice homozygous for the nm3888 mutation had mild ataxia and tremors beginning at 4 weeks of age (movies S1 and S2).Immunohistochemistry for calbindin D-28, a calcium-binding protein expressed in Purkinje cells, revealed Purkinje cell degeneration beginning at 4 weeks of age (Fig. 1A).Death of these neurons progressed rostrally to caudally with approximately 90% survival at 1 month of age, 35% at 6 months of age, and 16% at 12 months of age (Fig. 1B).Purkinje cells in mice heterozygous for this mutation also degenerated, but cell death had a later onset beginning at 6 months of age, and death of these neurons progressed more slowly with cell survival at 77% at 3 months of age and 72% at 12 months of age.Cleaved caspase 3 was detected in dying neurons but not wild-type neurons, indicating that these cells were undergoing apoptosis (Fig. 1C).
Genome scans, using polymorphic microsatellite markers on F2 nm3888 −/− mice generated from crosses of NOD/ShiLtJ-nm3888 −/− and CAST/Ei mice, revealed the nm3888 mutation was on Chromosome 13.Analysis of additional F2 mice demonstrated that the nm3888 mutation resided in a 2-megabase (Mb) region between D13Mit18 and rs291434 on Chromosome 13, a region containing eight protein-coding genes (Fig. 2A).To examine the expression levels of protein-coding genes in the critical region, we performed quantitative reverse transcriptase polymerase chain reaction (RT-qPCR) using wild-type and nm3888 −/− cerebellar cDNA from 3-week-old mice.While we did not observe significant differences in expression of seven of these genes, the expression of a zinc finger transcription factor, Rreb1, was decreased 34-fold in mutant cerebella (Fig. 2B).Northern blot analysis for total Rreb1 demonstrated a loss of the predominant transcript (>9 kb) in nm3888 −/− brains (Fig. 2C).We performed RT-PCR with primers specific to the four transcripts of Rreb1 (National Center for Biotechnology Information, https://ncbi.nlm.nih.gov/gene/68750) that were 9.0 kb or greater: XM_006516750.2(X2), XM_006516752.2(X3), XM_006516753.3(X4), and OP867049 (variant 7, hereafter referred to as V7).X2, X3, and X4 each have different 5′ untranslated regions (5′UTRs) but contain the same coding sequence that produces a protein with 16 zinc fingers.V7 has a unique 5′UTR, a later ATG start site, and splices around the eighth coding exon found in X2, X3, and X4, producing a protein with 14 zinc fingers (Fig. 2D).We did not detect X3 in wild-type cerebella, and X2 and X4 were expressed at comparable levels to wild-type in the mutant cerebellum (Fig. 2E).However, expression of V7 was markedly lower in the nm3888 −/− cerebellum relative to wild-type cerebellum.Northern blot analysis using a probe specific to V7 demonstrated that it corresponded to the predominant Rreb1 transcript in the wild-type cerebellum and confirmed its loss in the mutant cerebellum (Fig. 2F).We examined Rreb1 in nm3888 −/− mice to identify the molecular changes that resulted in decreased expression.We sequenced all exons and identified no mutations.However, upon sequencing Rreb1 introns, we found a 573-bp insertion of an ERV3 retrotransposon element in intron 4 (Fig. 2, G to I), which cosegregated with the degeneration seen in nm3888 heterozygotes and homozygotes.Given the specific impact of the nm3888 mutation on expression of V7, we hypothesized that disruption of this transcript was due to aberrant splicing from exon 4, which is uniquely found in the V7 transcript, into the retrotransposon (Fig. 2J).PCR using cerebellar cDNA from nm3888 −/− mice and primers corresponding to exon 4 of V7 and the retrotransposon element generated an appropriately sized amplicon (Fig. 2K).Sequencing confirmed that this amplicon was a V7 exon 4/ retrotransposon fusion product.We were unable to amplify an X4/ retrotransposon fusion product when we used primers corresponding to the unique exon in X4, suggesting that the retrotransposon does not disrupt expression of the X4 transcript.Together, our genetic mapping data, identification of the retrotransposon insertion, and altered expression of Rreb1 strongly suggest that loss of Rreb1 is causal for the Purkinje cell degeneration seen in nm3888 mutant mice.

Expression of Rreb1 V7 is nervous system specific and temporally regulated
To evaluate whether additional tissues require Rreb1 V7, we performed RT-qPCR to assess expression of Rreb1 V7 in wild-type tissues throughout the body.Although we broadly detected total Rreb1 using primer pairs that amplify all transcripts of Rreb1, Rreb1 V7 was highly enriched in the brain, spinal cord, and eye (Fig. 3A).To determine which cell types express Rreb1 in the brain, we performed in situ hybridization with a 3′UTR probe that was complementary to all Rreb1 transcripts.We detected Rreb1 in mitral cells of the olfactory bulb, multiple thalamic nuclei, and the retrosplenial area within the cortex (Fig. 3B).In addition, Rreb1 was highly expressed in Purkinje cells (Fig. 3C), dopaminergic neurons in the substantia nigra pars compacta (Fig. 3D), granule cells of the dentate gyrus of the hippocampus (Fig. 3E), and deep cerebellar nuclei (Fig. 3F).We did not detect Rreb1 in the nm3888 −/− brain (Fig. 3G), confirming that V7 is the predominant transcript expressed in the brain.
To determine whether the temporal expression of the V7 transcript correlates with the onset of Purkinje cell degeneration, we performed RT-qPCR on wild-type embryos.Although total Rreb1 was expressed in the developing brain of embryonic day 13.5 (E13.5)embryos, we did not detect V7 at this time but found that it was expressed at low levels in E18.5 cerebellum (Fig. 3H).Expression levels of both total Rreb1 and the V7 transcript were constant in the cerebellum from E18.5 through postnatal day 7 (P7).Between P7 and P14, a period of ongoing cerebellar development, cerebellar levels of V7 significantly increased.However, the levels of total Rreb1 did not increase during this time, suggesting that down-regulation of other Rreb1 isoforms occurs during cerebellar development.Together, these data indicate that both the temporal and spatial expression of the Rreb1 V7 transcript correlate with the postembryonic and nervous systemspecific defects found in Rreb1 V7-deficient mice.
RREB1 regulates the endomembrane system and microtubule network in Purkinje cells RREB1 directly binds DNA and recruits cofactors that alter histone modifications to potentiate or repress transcription of target genes.Studies of Rreb1 have generally centered on its role in tumorigenesis, where an increase in copy number and expression of Rreb1 has been linked to the formation of multiple cancers (6)(7)(8)(9)(10)(11).In addition, total loss of Rreb1 impairs cell migration during early embryogenesis, resulting in embryonic lethality (12,13).To determine a role for RREB1 in Purkinje cells, we sought to identify Purkinje cell-specific targets of Rreb1 via ChIP-seq.To identify targets of Rreb1 V7, we generated transgenic mice expressing V7 with three V5 epitope tags at the C terminus (Fig. 4A).Expression of this transgene was driven by the Purkinje cell-specific promoter, Purkinje cell protein 2 (PCP2) and PCP2 exon 4, which, when placed after the stop codon of a transgene, increases Purkinje cell-specific expression (14).Immunofluorescence performed on cerebellar sections from transgenic mice using an antibody to the V5 epitope tag confirmed that the transgene was exclusively expressed in Purkinje cells, and RREB1-V5 localized to the nucleus (fig.S1A).
We performed ChIP-seq in cerebellar lysates from transgenic mice using an antibody against V5.Analysis of these data revealed 3037 peaks, a third of which occurred in introns, another third in intergenic regions, and a third at known promoter regions/transcriptional start sites (Fig. 4B and data S1, ChIP_seq).We assigned 1932 genes to peaks based on proximity, and we identified a strong enrichment of RREB1 binding in protein-coding genes (85%) (Fig. 4C).We performed ChIP-qPCR on three target genes to validate RREB1 binding (fig.S1B).De novo motif analyses demonstrated a significant enrichment (P = 10 −925 ) for a Purkinje cell RREB1 binding motif that is highly similar to the binding motif of a known RREB1 cofactor, NEUROD1 (match score = 0.98) (15).This de novo motif was enriched in 24% of RREB1 ChIP peaks versus 2.5% of background peaks (fig.S1C).To elucidate cellular processes directly regulated by RREB1, we performed cellular component enrichment analysis on all RREB1 putative target genes.This analysis revealed a significant enrichment for genes associated with the microtubule network, somatodendritic compartment, and the endomembrane system, along with several neurotypical compartments such as the axon and synapse (Fig. 4D and data S1, ChIP_Cell_Comp).Of the top 10 most significantly enriched cellular components, we found the largest number of RREB1 putative targets (396) belonging to the endomembrane system, which encompassed other cellular components, including the synapse and exocytic vesicles.
To determine direct and indirect downstream targets of RREB1 in Purkinje cells, we performed RNA-seq from wild-type and nm3888 −/− Purkinje cells using RiboTag mice (16).The hemagglutinin (HA)-tagged ribosomal protein RPL22 was expressed in Purkinje cells using PCP2-CRE, and we immunoprecipitated epitope-tagged ribosomes from Rpl22 fl/+ : PCP2-Cre mice either homozygous for the nm3888 mutation or wild type at this locus.We enriched nascent transcripts for polyA + RNA and generated cDNA libraries for sequencing.RNA-seq analysis revealed 1598 differentially expressed genes with approximately   (G) violin plot showing differential expression of all cytoskeleton system putative targets (green) compared to all other detected genes (light gray) and differential expression of all endomembrane system putative targets (blue) compared to all other detected genes (dark gray).We determined P values by Wilcoxon rank sum tests.
one-third of these genes up-regulated and two-thirds down-regulated in nm3888 −/− Purkinje cells (Fig. 4E and data S1, RNA_seq_DE).We performed RT-qPCR on three differentially expressed genes to validate expression changes in the nm3888 mutant (fig.S1D).Molecular pathway analysis of differentially expressed genes revealed a strong enrichment of genes that encode tubulin-binding proteins as well as transcription regulator activity, mitogen-activated protein kinase phosphatase activity, soluble N-ethylmaleimide-sensitive factor attachment protein receptor binding, and hydrolase activity (Fig. 4F and data S1, GO_Mol_Path_Upreg and GO_Mol_Path_Downreg).We identified 170 of the differentially expressed genes as RREB1 targets by our ChIP-seq data (fig.S1E and data S1, ChIPseq_RNAseq_Overlap).
We next looked at expression of all putative targets (identified by ChIP-seq) belonging to the endomembrane system and microtubule cytoskeletal network, as both pathways were enriched in our ChIPand RNA-seq analyses.This analysis revealed that these genes were largely down-regulated in nm3888 −/− Purkinje cells relative to wildtype cells (Fig. 4G), suggesting that RREB1 plays a role in positively regulating genes associated with both systems.

Loss of RREB1 alters the microtubule network and dendritic branching
Tubulin-binding proteins govern microtubule dynamics through acetylation, polyglutamylation, tyrosination, and removal of these groups.Loss of tubulin-binding proteins has been associated with alterations in posttranslational modifications of tubulin and subsequent neurodegeneration (17).To determine whether the dysregulation of genes encoding tubulin-binding proteins in nm3888 −/− Purkinje cells led to alterations in posttranslational modification of tubulin and tubulin networks, we performed immunofluorescence with antibodies to tubulin and modified tubulin.We observed a reduction in the intensity of immunofluorescence in nm3888 −/− Purkinje cells relative to wild-type cells using antibodies to beta III tubulin, acetylated alpha tubulin, and polyglutamylated alpha tubulin in the anterior or mid region of the cerebellum (lobules II and IV) (Fig. 5, A to F).We did not observe a change in Purkinje soma immunofluorescence intensity between genotypes for alpha tubulin and tyrosinated alpha tubulin in this region (fig.S2, A to  D).In caudal regions of the cerebellum (lobule IX), we also saw a decrease in beta III tubulin (fig.S2, E and F).However, in contrast to the rostral region of the cerebellum, we saw increased immunofluorescence intensity for alpha tubulin in caudal cerebellum of nm3888 −/− mice (fig.S2, G and H).Additional differences between the caudal cerebellum of wild-type and nm3888 −/− mice include an increase in acetylated alpha tubulin (fig.S2, I and J), no altered fluorescence intensity for polyglutamylated alpha tubulin (fig.S2, K and L), and an increase in tyrosinated alpha tubulin (fig.S2, M and N).These data suggest that there are likely different pathways regulating the microtubule network between rostral and caudal regions of the cerebellum.
Immunofluorescence for tyrosinated alpha tubulin in lobule IV revealed alterations in dendritic branching of nm3888 −/− Purkinje cells.Dendritic arborization of neurons is known to be highly regulated by microtubule-associated proteins, loss of which results in a reduction of dendritic complexity and disorganized arborization (18,19).To further investigate branching of Purkinje cell dendrites in this region of the cerebellum, we performed cell labeling by injecting lobule IV Purkinje cell soma with biocytin and visualized this with Alexa Fluor 555-streptavidin (Fig. 5, G and I).We tranced dendritic branches from z-stacks of confocal images of Purkinje cells and performed Sholl analyses on the traces (Fig. 5, H and J).We observed a clear reduction in both the branching complexity and the total number of branch points in the dendritic arbors of nm3888 −/− Purkinje cells (Fig. 5, K and L).
Loss of RREB1 is associated with alterations in the autophagic system and proteostasis RNA-seq analysis showed that 59 of 396 RREB1 endomembrane system targets identified by ChIP-seq were significantly differentially expressed in nm3888 −/− Purkinje cells.Analysis of these genes based on molecular function revealed that 25% of these genes have roles in autophagy (data S1, Endo_Targets).
Autophagy is integral to neuron homeostasis, and loss of genes essential for autophagy has been shown to result in the accumulation of protein inclusions, a hallmark of many neurodegenerative diseases.Given this and our sequencing data, we examined whether the cellular machinery essential for autophagy was compromised in nm3888 −/− Purkinje cells at 3 weeks of age, before the onset of degeneration.We first examined autophagosomes that engulf cargo bound by the degradation signaling proteins P62 and ubiquitin.As autophagosomes form, microtubule-associated protein 1A/1B-light chain 3 (LC3), a lipidated protein that integrates into the autophagosome membrane, can be visualized as puncta (20).To visualize autophagosomes, we crossed nm3888 −/− mice to mice containing an LC3 transgene conjugated to fluorophores (CAG-RFP-EGFP-LC3 Tg ).We counted LC3positive puncta in Purkinje cells of nm3888 −/− ; CAG-RFP-EGFP-LC3 Tg and CAG-RFP-EGFP-LC3 Tg mice, and we observed a reduction in the number of LC3 puncta in nm3888 −/− Purkinje cells (Fig. 6, A and C), suggesting a reduction in the number of autophagosomes.
Autophagosomes fuse with lysosomes forming autolysosomes in which the sequestered cell material is degraded.To determine whether Rreb1 mutants show altered lysosome numbers in Purkinje cells, we performed immunofluorescence on 3-week-old wild-type and nm3888 −/− cerebella for lysosomal-associated membrane protein 1 (lamp1), a glycoprotein localized to both late endosomes and lysosomes, and cathepsin-D, a protease localized exclusively to lysosomes (21).The number of puncta for both proteins was reduced in nm3888 −/− Purkinje cells, indicating that, like autophagosomes, fewer lysosomes exist in these mutant Purkinje cells (Fig. 6, A to C).To determine whether loss of Rreb1 affected lysosome and autophagosome numbers in the caudal region of the nm3888 −/− cerebellum where Purkinje cells do not degenerate, we examined lysosome and autophagosomes in lobule IX of 3-week-old mutant and wild-type cerebellum.We observed a significant reduction in the number of lamp1 puncta; however, there was no reduction in LC3 or cathepsin-D puncta in lobule IX of the nm3888 −/− cerebellum (Fig. 6, B and D).
Proteostatic disruption in Purkinje cells has been shown to result in the accumulation of protein inclusions (22)(23)(24).Such inclusions that have failed to be degraded by autophagy or the proteasome contain an accumulation of degradation signaling proteins.These include ubiquitin, which labels proteins for degradation by both autophagy and the proteasome, and ubiquitin-binding protein (P62), an autophagyspecific signaling protein.To identify whether nm3888 −/− Purkinje cells develop protein inclusions, we performed immunofluorescence for P62 and ubiquitin.We observed P62-positive and ubiquitin-positive puncta and puncta positive for both proteins in nm3888 −/− Purkinje cells from mice that were 3 to 6 weeks of age (Fig. 6, E to H).Neither nm3888 −/− nor wild-type Purkinje cells from lobule IX displayed P62-positive or ubiquitin-positive puncta between 3 and 6 weeks of age.Together, these findings suggest heterogeneity in the regulation of proteostasis among Purkinje cells.

DISCUSSION
We have demonstrated that loss of V7, an Rreb1 transcript highly expressed in the nervous system, is associated with Purkinje cell degeneration in mice.In addition, we have identified Purkinje cell targets of the zinc finger transcription factor RREB1 and shown that targets of the microtubule network and endomembrane system are positively regulated by RREB1.Loss of Rreb1 affects proteins associated with the microtubule network, lowering levels of beta III tubulin, acetylated alpha tubulin, and polyglutamylated alpha tubulin.Within the endomembrane system, we observe reduced numbers of autophagosomes and lysosomes as well as the formation of protein inclusions, and apoptosis.Many studies have shown a role for microtubules in autophagy, particularly during the formation of autophagosomes, where microtubule-associated proteins, such as the RREB1 target microtubule associated protein 1A (MAP1A), associate with the autophagic membrane component LC3 to facilitate the interaction between LC3 and microtubules (25).The maturation of lysosomes is also highly dependent on the microtubule network, and previous studies have demonstrated that inhibition of microtubule dynamics impairs trafficking of proteases essential for protein degradation, such as cathepsin-D, to lysosomes (26).Microtubules have also been shown to have an essential role in the formation of autolysosomes where depolymerization of the microtubule network impedes colocalization of autophagosomes and lysosomes at the microtubule organizing center (27)(28)(29)(30).These studies suggest that disruption of the microtubule network may affect proteostasis in Rreb1-deficient Purkinje cells.The microtubule cytoskeleton also plays an integral role in nervous system development (31)(32)(33).Studies have also demonstrated that tubulin and microtubule-binding proteins that regulate the microtubule cytoskeleton are critical for neuron migration during development (34,35) and dendritic arborization (36,37), where loss of these proteins results in improper or underdeveloped brain structures and altered dendritic complexity.Among the 23 differentially expressed somatodendritic targets of RREB1, two belonged to the stathmin family of tubulin-binding proteins [Stathmin2 (STMN2) and STMN4], which have been shown to promote microtubule catastrophe and inhibit polymerization, thereby influencing microtubule dynamics (38,39).Previous studies have demonstrated roles for stathmins during Purkinje cell dendritic arborization, where altered activity of these stathmins results in decreased dendritic branching, which is rescued with transgenic expression of the missing stathmin protein (40,41).Thus, loss of Rreb1 may directly impair Purkinje cell dendritic development through dysregulation of STMN2 and STMN4.In addition, STMN2 loss has been shown to drive axon degeneration, motor neuropathy, and neuromuscular junction denervation and has been implicated in amyotrophic lateral sclerosis pathogenesis (42)(43)(44).It is unknown whether reduced levels of STMN2 contribute to Purkinje cell degeneration in nm3888 −/− mice.Future studies will investigate whether transgenic expression of STMN2 in Rreb1-deficient mice can alter the course of Purkinje cell degeneration.
While there are no reported human cases of homozygous loss of hRREB1, heterozygous loss of hRREB1 has been reported in patients with interstitial microdeletions of chromosome 6p (45).All patients presented with intellectual disability, craniofacial dysmorphisms, and cardiac abnormalities, consistent with Noonan-spectrum disorder, and many of these features, including cardiac defects and craniofacial abnormalities, have also been observed in mice heterozygous for Rreb1 loss.These deficits, as well as the early embryogenesis defects in Rreb1 knockouts (12,13), underscore the multiple roles of hRREB1 in development.In contrast, nm3888 mutant mice have a hypomorphic reduction in Rreb1 with loss of a nervous system-specific transcript, V7, resulting in postnatal phenotypes rather than early developmental defects.Mice homozygous and heterozygous for the nm3888 mutation provide models of neurological phenotypes associated with loss of Rreb1 that may prove fruitful in ongoing efforts to understand postnatal phenotypes incurred by patients experiencing heterozygous loss of hRREB1.
Our Northern blot analysis of total Rreb1 in whole mouse brain followed by RT-qPCR suggests the expression of two protein products in the brain.The V7 transcript encodes one protein isoform with a length of 1618 amino acids and 14 zinc fingers, while the remaining transcripts expressed in the brain encode a protein isoform with a length of 1754 amino acids and 16 zinc fingers.The differences in protein structure occur in the N terminus where the V7 isoform contains a later, in-frame ATG start site, resulting in one fewer zinc finger at the N terminus, and in lack of exon 12, which contains another zinc finger.Because these sequences are nearly identical (95%), it is possible that the functional differences in mouse RREB1 and human hRREB1 protein isoforms are predominantly dictated by their unique spatiotemporal expression patterns, and thus cellular context, rather than by structural differences.Alternatively, given the ability of zinc fingers to bind DNA and proteins, it is possible that the zinc finger difference at the N terminus and in exon 12 alters DNA binding, resulting in unique targets for each protein isoform, or in unique interactions with binding partners when recruiting complexes or cofactors to regulate gene expression.
Our study has identified gene targets of RREB1 in Purkinje cells that have previously been associated with the endomembrane system and the regulation of autophagy.In addition, Rreb1-deficient Purkinje cells have reduced numbers of autophagosomes and lysosomes and develop protein inclusions.Future studies are necessary to establish a direct connection between dysregulation of RREB1 target genes, alterations in autophagic and lysosomal machinery, the formation of protein inclusions, and Purkinje cell degeneration.

Mice
All mouse studies were under the guidance of and approved by the Institutional Animal Care and Use Committees of the University of California, San Diego and the Jackson Laboratory.We housed mice in a 12-hour light/dark cycle with free access to food and water.Where possible, we compared littermates of wild-type and nm3888 −/− mice within each experiment.We used mice from both sexes for all experiments.We obtained B6N.129-Rpl22tm1.1Psam/J(stock no.011029), C57BL/6-Tg(CAG-RFP/EGFP/Map1lc3b)1Hill/J (stock no.027139), and B6.129-Tg(Pcp2-cre)2Mpin/J (stock no.004146) mice from the Jackson Laboratory.These mice have been previously described (16,46).
To generate transgenic Tg(PCP2-Rreb1/3xV5)20/J mice, we subcloned the coding sequence of Rreb1 transcript V7 cDNA with triple C-terminal V5 tags inserted before the stop codon into the BamHI site of the fourth exon of the L7/PCP2 minigene (14).We linearized the construct and injected this into the pronucleus of C57BL/6J zygotes.We identified founders by PCR using primers that recognize tandem repeats of the transgene (forward: 5′TCTGACCAAATACCACCACC3′; reverse: 5′CAGATACTGCATGCGTGGTC3′).

In situ hybridization
We used an RNAScope probe for Rreb1 (ACDBio, 542641-C2) for in situ hybridization on two wild-type and two nm3888 −/− mice following the manufacturer's protocol.Briefly, we transcardially perfused mice with 10% neutral-buffered formalin and processed and embedded brains for paraffin sectioning.We baked 7-μm sections at 60°C for 1 hour before deparaffinization.We treated sections with hydrogen peroxide (ACDBio, 322281) before antigen retrieval in target retrieval buffer (ACDBio, 322000) and treatment with protease plus (ACDBio, 322281).We incubated sections with the probe for 30 min and then washed them in 1× wash buffer (ACDBio, no.310091).We stored slides in 5× SSC overnight before a series of amplification steps and incubation with Cy3 (PerkinElmer, FP1170, 1:1500).For immunofluorescence after in situ hybridization, we incubated either calbindin D-28 (rabbit, Swant, cb38, 1:1000) or tyrosine hydroxylase (mouse, Leica, NCL-L-TH, 1:250) on slides overnight, then incubated them with secondary antibody.We stained nuclei with DAPI and slides and stored them in the dark at 4°C before imaging.

Imaging
We generated images for calbindin D-28 immunohistochemistry using an Olympus VS200 SlideScanner.We took in situ images (tiled) and tubulin and aggregate immunofluorescence images using a Zeiss Axiovert 200.We took images for V5 and calbindin immunofluorescence of Tg(PCP2-Rreb1/3xV5)20/J on a Leica SP5 confocal microscope.We imaged dye-filled neurons and images used for quantification of lamp1, cathepsin-D, and LC3 using a Nikon Ti-2 Eclipse and CSU W1 SORA confocal with a variable sized stack custom to each section and 0.3-μm z stepsize.

RNA extraction
We euthanized three wild-type and three nm3888 −/− mice, dissected out organs and flash froze them in liquid nitrogen for storage at −80°C.We homogenized tissues in 1 ml of TRIzol (Life Technologies, 15596018) using a polytron homogenizer and isolated RNA as per the manufacturer's protocol.We determined concentration and A260/ A280 of isolated RNA by Nanodrop.

Reverse transcription and PCR
We performed cDNA synthesis on deoxyribonuclease-treated total RNA from three wild-type and three nm3888 −/− mice using oligo(dT) primers and the SuperScript III First-Strand Synthesis System (Invitrogen, 18080051).We performed quantitative RT-PCR using SYBR Green Supermix (Bio-Rad, 1708880).We performed each quantitative RT-PCR reaction with 40 ng of cerebellar cDNA.For tissue panel and developmental time course of Rreb1 expression, we used the following primers: Total Rreb1, forward: 5′GTCACATGCTGGTGCACTCT3′, reverse: 5′CGTGACTCAGCTTCCTTTTG3′; Rreb1 V7, forward: plotted these on the bar graph in Fig. 1.Statistical analyses are indicated in the legend for Fig. 1.

RT-qPCR analysis
For quantitative RT-PCR experiments, we evaluated expression in cerebella from three wild-type mice and three nm3888 −/− mice.We normalized expression levels of the genes of interest to Etf1 or Gapdh using the 2 −∆∆CT method (47).

Quantification of immunofluorescence intensity
We used one section that was 7 μm thick from each of four wild-type and five nm3888 −/− mice.We deparaffinated and stained sections together.We used five neurons from each section to quantify immunofluorescence intensity.We opened images in ImageJ and drew an ellipse around the cell soma of a given Purkinje cell, measured the integrated density, and plotted this in the bar graph in Fig. 5. Statistical analyses are indicated in the legend for Fig. 5.

Quantification of dendritic branching
To quantify the number of dendritic branch points in biocytin-labeled Purkinje cells, we used the plugin Simple Neurite Tracer in ImageJ to trace dendritic arbors of each cell in three-dimensional space.We used three wild-type and three nm3888 −/− mice for this experiment, and we traced between five and six neurons for each animal.Statistical analyses are indicated in the legend for Fig. 5.

Sholl analysis
We used the Simple Neurite Tracer plugin for ImageJ to perform a Sholl analysis.We used a radius of 20 μm and generated a polynomial trendline using the "Best Fitting Degree" determined by the plugin algorithm.Statistical analysis is indicated in the legend for Fig. 5.

Quantification of LC3, lamp1, and cathepsin-D puncta
For lamp1, LC3, and cathepsin-D puncta quantification, we analyzed a minimum of 30 Purkinje cells in cerebellar lobule IV and lobule IX per section and analyzed one section from each of three wild-type and three nm3888 −/− mice.For quantification, we used the ImageJ plugin puncta analyzer with a rolling ball radius set to 50 and minimum puncta size set to four pixels.For each cell, we manually adjusted the threshold for puncta to detect individual puncta.Statistical analyses are indicated in the legend for Fig. 6.

Quantification of P62 and P62 + ubiquitin-positive puncta
For inclusion quantification, we quantified the total number of P62positive Purkinje cells (as marked by calbindin D-28 positivity), in total cerebella for three 7-μm sections spaced 100 μm apart per mouse.We performed analyses for three nm3888 −/− mice at each time point and three 6-week-old wild-type mice.Statistical analyses are indicated in the legend for Fig. 6.

Analysis of sequencing data
For RNA-seq, we deduplicated reads and trimmed them with fastp (49) before alignment to the mm10 genome with Bowtie2 (50).We used featurecounts (51) to calculate read counts for each gene in the mm10 genome, used DESeq2 (52) for differential expression analyses, and used batch as a covariant.We used ShinyGO (53) for molecular pathway enrichment and input significantly up-regulated and significantly down-regulated genes separately.
For ChIP-seq, we trimmed reads with fastp (49) to remove adapter sequences before alignment to the mm10 genome with Bowtie2 (50).We used Sambamba (54) to filter out multimappers and deduplicate reads and used MACS3 (55) to perform peak calling using the following parameters: -g 1.87e9, -q 0.05.We merged samples using bedtools where only overlapping peaks were merged.For gene ontology enrichment, we used ShinyGO (53).We used Homer (56) for de novo and known motif enrichment analysis with default parameters.

Supplementary Materials
This PDF file includes: Figs.S1 and S2 legend for data S1 legends for movies S1 and S2 References Other Supplementary Material for this manuscript includes the following: data S1 Movies S1 and S2

Fig. 4 .
Fig.4.RREB1 regulates genes associated with the endomembrane system and cytoskeletal network.(A) Schematic representation of Rreb1 variant 7 (V7) cdnA transgene: PcP2 promoter and fourth exon (light blue), Rreb1 V7 cdnA (dark blue), three v5 epitope tags (red), and polyA signal (green).(B) Breakdown of types of genomic regions containing peaks defined by MAcS3 and annotated by hOMeR.transcriptional termination sequence (ttS).(C) Breakdown of types of genes associated with MAcS3 peaks; gene types annotated by hOMeR.(D) cellular component enrichment of putative chiP targets performed using ShinyGO, plotted by FdR with fraction of chiP targets/number of genes in category, and percent enrichment [(number of chiP targets/number of genes in category) × 100%] indicated.(E) volcano plot of gene expression in nm3888 −/− Purkinje cells relative to wildtype neurons.Up-regulated genes are in red, down-regulated genes are in blue, and genes with a P adj ≥ 0.05 are in gray.numbers of genes up-regulated and down-regulated are indicated.(F) Molecular pathway enrichment for significantly differentially expressed genes performed with ShinyGO Molecular Pathway enrichment; arrows indicate directionality of pathway.(G) violin plot showing differential expression of all cytoskeleton system putative targets (green) compared to all other detected genes (light gray) and differential expression of all endomembrane system putative targets (blue) compared to all other detected genes (dark gray).We determined P values by Wilcoxon rank sum tests.

Fig. 5 .Fig. 6 .
Fig. 5. Loss of Rreb1 affects the tubulin network and reduces dendritic branching of Purkinje cells.(A) immunofluorescence for beta iii tubulin in lobule iv and (B) average cell soma fluorescence intensity (a.u., arbitrary units) with SeM represented by error bars.(C) immunofluorescence for acetylated alpha tubulin in lobule ii and (D) average cell soma fluorescence intensity with SeM represented by error bars.(E) immunofluorescence for polyglutamylated alpha tubulin in lobule ii and (F) average cell soma fluorescence intensity with SeM represented by error bars.[(A), (c), and (e)] Molecular layer (Ml), Purkinje cell layer (Pl), and granule cell layer (Gl) are indicated.images are pseudocolored on the basis of intensity; an 8-bit lUt scale ranging from 0 to 255 was used.[(B), (d), and (F)] Multiple t tests were performed ****P ≤ 0.0001.[(A) to (F)] We used 3-week-old mice for images and quantification.(G) Maximal intensity projection and (H) dendritic branch tracing of biocytin-labeled Purkinje cells from a wild-type (Wt) Purkinje cell.(I) Maximal intensity projection and (J) dendritic branch tracing of biocytin-labeled dendrites from an nm3888 −/− Purkinje cell.[(h) and (J)] total branch points for each cell are indicated in the top right corner of tracing.(K) Sholl analysis for wild-type and nm3888 −/− Purkinje cells from lobule iv in 3-week-old mice.A two-way analysis of variance (AnOvA) was performed, which showed a significant effect for genotype; P value is indicated on the graph.(L) Graph of average number of branch points for traced dendrites from wild-type and nm3888 −/− Purkinje cells; SeM is indicated.****P ≤ 0.0001 (unpaired t test).Scale bars, 25 μm [(A), (c), and (e)] and 50 μm [(G) and (i)].