NUDC is critical for rod photoreceptor function, maintenance, and survival

NUDC (nuclear distribution protein C) is a mitotic protein involved in nuclear migration and cytokinesis across species. Considered a cytoplasmic dynein (henceforth dynein) cofactor, NUDC was shown to associate with the dynein motor complex during neuronal migration. NUDC is also expressed in postmitotic vertebrate rod photoreceptors where its function is unknown. Here, we examined the role of NUDC in postmitotic rod photoreceptors by studying the consequences of a conditional NUDC knockout in mouse rods (rNudC−/−). Loss of NUDC in rods led to complete photoreceptor cell death at 6 weeks of age. By 3 weeks of age, rNudC−/− function was diminished, and rhodopsin and mitochondria were mislocalized, consistent with dynein inhibition. Levels of outer segment proteins were reduced, but LIS1 (lissencephaly protein 1), a well‐characterized dynein cofactor, was unaffected. Transmission electron microscopy revealed ultrastructural defects within the rods of rNudC−/− by 3 weeks of age. We investigated whether NUDC interacts with the actin modulator cofilin 1 (CFL1) and found that in rods, CFL1 is localized in close proximity to NUDC. In addition to its potential role in dynein trafficking within rods, loss of NUDC also resulted in increased levels of phosphorylated CFL1 (pCFL1), which would purportedly prevent depolymerization of actin. The absence of NUDC also induced an inflammatory response in Müller glia and microglia across the neural retina by 3 weeks of age. Taken together, our data illustrate the critical role of NUDC in actin cytoskeletal maintenance and dynein‐mediated protein trafficking in a postmitotic rod photoreceptor.

of a conditional NUDC knockout in mouse rods (rNudC −/− ).Loss of NUDC in rods led to complete photoreceptor cell death at 6 weeks of age.By 3 weeks of age, rNudC −/− function was diminished, and rhodopsin and mitochondria were mislocalized, consistent with dynein inhibition.Levels of outer segment proteins were reduced, but LIS1 (lissencephaly protein 1), a well-characterized dynein cofactor, was unaffected.Transmission electron microscopy revealed ultrastructural defects within the rods of rNudC −/− by 3 weeks of age.We investigated whether NUDC interacts with the actin modulator cofilin 1 (CFL1) and found that in rods, CFL1 is localized in close proximity to NUDC.In addition to its potential role in dynein trafficking within rods, loss of NUDC also resulted in increased levels of phosphorylated CFL1 (pCFL1), which would purportedly prevent depolymerization of actin.The absence of NUDC also induced an inflammatory response in Müller glia and microglia across the neural retina by 3 weeks of age.Taken together, our data illustrate the critical role of NUDC in actin cytoskeletal maintenance and dynein-mediated protein trafficking in a postmitotic rod photoreceptor.

| INTRODUCTION
The nud (nuclear distribution) genes, including nudA (dynein heavy chain), nudC, nudF (LIS1), nudG (dynein light chain or LC8), and nudK encode elements of the dynein/dynactin complex in Aspergillus nidulans. 13][4][5] Loss of NudC is lethal in Aspergillus 6 and stalls cell division at the one-cell stage in cultured mammalian cells. 73][14][15][16] NUDC is expressed in the nervous system 16 as well as in migrating neurons in cell culture. 17Our group found that NUDC is expressed in postmitotic vertebrate rod photoreceptors and is required for rod maintenance.Upon shRNA knockdown of NudC in the Xenopus laevis retina, rod outer segment (OS) discs overgrew and coincided with retinal degeneration. 18UDC also interacts with rhodopsin and may affect rhodopsin trafficking mediated by dynein through the inner segment to the connecting cilium. 18isc overgrowth accompanying NUDC depletion suggests NUDC involvement in actin dynamics.][27] Additionally, knockdown of the actin polymerizationinducing protein Arp2/3 in rods leads to F-actin loss at the proximal OS and the formation of membrane whorls rather than discs. 28NUDC stabilizes the actin-modulating protein cofilin 1 (CFL1) 29 to modulate actin branching as well as the balance between filamentous and globular actins. 30CFL1 plays a critical role in cytoskeletal remodeling by modulating the quantity and length of F-actin. 31When phosphorylated by Lim kinase, phosphorylated CFL1 (pCFL1) no longer binds F-actin, purportedly preventing the production of branched or G-actin. 32Understanding actin dynamics is essential to determine how rod OS discs form and are maintained, and how constituent proteins are trafficked between the rod OS and inner segment (IS).
The role of NUDC in postmitotic cells is incompletely understood.Here, iCre75-induced homologous recombination was used to generate rod-specific NudC knockout mice.We show that NUDC loss in postmitotic mouse rods leads to photopic and scotopic dysfunction, morphologic changes in the rod OS, glial inflammation, changes in proteins involved in actin cytoskeletal maintenance, and retinal degeneration.The plethora of observed phenotypes highlights the essential role of NUDC in rod photoreceptor function regulating the dynamics of the actin cytoskeletal network and the function of dynein.

| Research animals
Flippase (FLP) and iCre75 transgenic mice were obtained from The Jackson Laboratory (Bar Harbor, ME, USA).Mice were maintained under 12-h cyclic dark/light conditions in standard cages and fed ad libitum.All animal studies were conducted in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of the University of Alabama at Birmingham.

| Generation of rod-specific
NudC −/− mice NudC embryonic stem (ES) cells were purchased from the KOMP (Knockout Mouse Project) and developed at the University of Utah Transgenic Gene-Targeting Mouse Facility.The construct of this clone contained a gene trap (GT) cassette inserted into intron 2 and floxed exons 3 and 7 (see Figure 1A for diagram).Gene targeting of ES cells was confirmed using PCR, including the 3′ and 5′ recombination arm, FRT sites, and LoxP sites.ES cell blastocyst injection and generation of chimeric and heterozygous NudC tm1a (GT/+) mice was performed at the University of Utah transgenic core.Homozygous gene-trapped mice did not show a phenotype indicating that the gene did not block translation of NUDC ("leaky" gene trap).Next, we crossed NudC GT/+ with FLP (ROSA26::FLPe) knockin mice (C57BL/6 background) to yield animals with the NudC tm1c floxed allele (NudC fl/+ ).The presence of the floxed NudC allele was confirmed using PCR with the following primers flanking the third loxP site: (a) 5′GTAAGCAACTTGGGCATCAG 3′ and (b) 5′CTAGCCAAGACAGAACCACG 3′.Homozygous NudC fl/fl mice were crossed with rod photoreceptorspecific Cre recombinase (iCre75) mice to generate NudC tm1d rod-specific rNudC +/− and rNudC −/− mice.iCre75 mice expressed Cre recombinase specifically in rod actin, cofilin1, dynein, NUDC, phosphorylated cofilin, photoreceptor, retinal degeneration photoreceptors from P4-P7 with total excision completed by P18.The presence of iCre75 was confirmed using PCR with the primers 5′GGATGCCACCTCTGATGAAG 3′ and 5′ CACACCATTCTTTCTGACCG 3′, and animals were bred such that there was only one iCre75expressing allele.Excision of exons 3-7 was confirmed by PCR of retina lysates using the following primers: (c) 5′ GCAGTTTCTGAGTCTGTTGAGAG 3′, (d) 5′TCCTGATCCTCTGCATCCTTTT 3′, and (e) 5′GCAGAAGTGGAAAAGTGACCTC.

| Transmission electron microscopy
Following euthanasia, retinal tissues were harvested and fixed in mixed aldehyde solutions (2% paraformaldehyde, 2% glutaraldehyde, 0.05% calcium chloride in 50 mM MOPS, pH 7.4), then osmicated, dehydrated, resin embedded in Epon and sectioned at 70 nm.Sections were placed on copper grids and stained with 2% uranyl acetate and 3.5% lead citrate (19314; Ted Pella).Images were acquired at 2.18 nm/px with a LEOL JEM-1400 TEM equipped with a 16-Mpixel Gatan camera using SerialEM software.Image mosaics were assembled with Nornir software application.Images were also acquired on a Tecnai Spirit T12 Transmission Electron Microscope (Thermo Fisher, formerly FEI) with an AMT (Advance Microscopy Techniques, Corp) Bio Sprint 29-megapixel digital camera.

| Immunofluorescence
Enucleated eyes from 3-week-old mice of either sex were placed in 4% paraformaldehyde in 0.1 M PBS (pH 7.4) overnight at 4°C.Following an additional overnight incubation in 30% sucrose in PBS at 4°C, the eyes were embedded in Tissue-Tek O.C.T. (Sakura Finetek, Torrance, CA, USA), flash frozen in isopentane cooled to near freezing by liquid nitrogen and sectioned into 10-μmthick slices using a cryomicrotome.Following sectioning, the eyes were processed for IHC as previously described. 18The sections were incubated in 10 mM sodium citrate and 0.05% Tween (pH 6.0) for 1 h for heat-induced epitope retrieval and cooled to room temperature for 30 min.Primary antibodies anti-NUDC, Abcam, 1:400; anti-GFAP, Cell Signaling, 1:400; anti-CRALBP (Müller glia marker), Abcam, 1:250; anti-Iba1, Cell Signaling, 1:100; anti-rhodopsin B6-30N, a gift from Drs. Paul Hargrave and W. Clay Smith (University of Florida), 1:500; and wheat germ agglutinin (1:750) Thermo Fisher Scientific, Waltham, MA, USA, conjugated with Alexa Fluor 555 or Alexa Fluor 647 were used to counterstain the photoreceptor OS.For TUNEL labeling, the Click-It Plus TUNEL kit (Thermo Fisher Scientific) was used to label apoptotic nuclei according to the manufacturer's instructions.Images of stained retinal sections were acquired on a Zeiss LSM 800 Airyscan confocal microscope (Carl Zeiss, Oberkochen, Germany) with a 63× oil-immersion objective or using an ECHO Revolution (ECHO, BIO Convergence Company).ImageJ software (NIH) was used to compile maximum projection z-stacks and process acquired images.

| Proximity ligation assay
Cryosections (10 μm) or fixed retinal tissue were incubated in 10 mM sodium citrate and 0.05% Tween (pH 6.0) for 1 h for heat-induced epitope retrieval and cooled to room temperature for 30 min prior to following the protocol for Duolink PLA Fluorescence (Sigma).Briefly, slides were blocked in Duolink blocking solution for 1 h at 37°C in a humidified chamber prior to primary antibody incubation overnight at 4°C in a humidified chamber.Primary antibodies used include anti-NUDC, Abcam, 1:400; anti-NUDC, Santa Cruz, 1:100; anti-CFL1, Abcam, 5 μg/mL; anti-CFL1, Cell Signaling, 1:400.Following two washes in Wash Buffer A (WBA), retinal slices were incubated with Duolink Probes diluted 1:5 for 1 h at 37°C.Retinal slices were incubated in ligation buffer, washed twice with WBA, and exposed to ligase for 30 min at 37°C.Amplification in the presence of polymerase for 100 min at 37°C was followed by a series of washes in Wash Buffer B prior to mounting with Duolink Mounting Medium with DAPI.Slides were then stored at 4°C until imaging using a Zeiss LSM 800 Airyscan confocal microscope (Carl Zeiss, Oberkochen, Germany).

| Spidergrams
Cryosections (10 μm) of fixed retinal tissue were stained with hematoxylin and eosin as previously described, 34 and brightfield images were obtained using an ECHO Revolution (ECHO, BIO Convergence Company) at 20× and joined together to produce an image of the entire retina, including the optic nerve.ImageJ software was used to measure the distance from the optic nerve and count nuclei in each 200 μm segment of the ONL.
Western blots to determine the ratio of G-actin to F-actin were performed on retinal lysates that were processed using a G-actin/F-actin In Vivo Assay Kit (Cytoskeleton, Inc.) following the manufacturer's instructions.Briefly, retinal tissues from both eyes of each genotype, with n = 3 separate animals per genotype, were homogenized in 100 μL of 37°C LAS2 and incubated at 37°C for 10 min before centrifugation at 350 g for 5 min at room temperature to pellet debris.The supernatant was removed and placed in a fresh tube before centrifugation at 100 000 g at 37°C for 1 h.Following centrifugation, the supernatant (G-actin pool) was placed in a fresh tube, and F-actin depolymerization buffer was added to the pellet (F-actin pool) and incubated on ice for 1 h, pipetting every 15 min.SDS sample buffer (5×) was added to the pellet and supernatant fractions and mixed well.The samples were stored at −20°C prior to western blot analysis for actin (Cytoskeleton 1:1000).Standards (10, 20, and 50 ng) were resuspended in F-actin depolymerization buffer, and 12% SDS Page Criterion gels (BioRad) were loaded 10 μL per sample prior to being run as described above.The primary antibody, anti-αactin (Cytoskeleton, Inc., 1:1000) was incubated overnight at 4°C in 3% BSA in TBS-T.AlexaFluor anti-mouse antibody (1:10 000) was incubated in 3% BSA in TBS-T for 1 h at room temperature, and membranes were visualized as described above.Unlike a typical western blot, this assay results in a single sample being separated into fractions to produce pools of F-and G-actin and therefore does not utilize a loading control such as GAPDH.

| Statistical analysis
Data are presented as the mean ± SD, where n represents the number of mice analyzed.Statistical comparisons were performed using one-way anova for all experimental data.A value of p < .05indicates a statistically significant difference.

| Conditional knockout of NudC in the mouse retina
Previously, our work showed NUDC localization in the mouse, tree shrew, rhesus macaque, and X. laevis retina, where NUDC and a mutant form of NUDC, NUDC L280P , were localized to the IS and outer nuclear layer (ONL) of photoreceptors. 18To uncover the role of NUDC in postmitotic rod photoreceptors, we generated a rod-specific NudC conditional knockout mouse by creating a mouse line with loxP sites flanking exons 3-7 of the mouse NudC gene (NudC fl/fl ) (Figure 1A,B, Figure S1).By breeding NudC fl/fl mice with mice expressing Cre under the rod opsin promoter (iCre75), 35 we selectively ablated exons 3-7 of NudC in rods (rNudC −/− ) (Figure 1C).Hereafter, NudC rod-specific conditional knockout mice (NudC fl/fl iCre75 + ) are referred to as rNudC −/− , and heterozygous mice (NudC fl/+ iCre75 + ) are referred to as rNudC +/− .Polymerase chain reaction (PCR) analysis of the five genotypes studied provided predicted amplicons for NudC (Figure 1D).Primers were designed to confirm the presence or absence of floxed exons in retinal DNA.Amplification with primers c and d using NudC +/+ templates will produce one band of approximately 900 bp (Figure 1D, lane 2).In the presence of the floxed allele (rNudC −/− ), the presence of FRT and loxP sites will produce a higher molecular weight band at 1 kb.In the presence of iCre75, exons 3-7 will be deleted producing an amplicon of 576 bp (Figure 1D, lanes 3, 4).
In rNudC −/− at 3 weeks, there was significantly less total NUDC protein detected in mouse retinal lysates (Figure 1E,F).There was still NUDC present in the whole retina western blot lysates, and we have previously shown that NUDC localizes not only to the IS and outer nuclear layer of rod photoreceptors but also to the retinal pigmented epithelium (RPE). 18Here, using immunofluorescence to stain for NUDC, we show that in rNudC −/− , while there was diminished NUDC found in the IS and ONL of rNudC −/− compared with rNudC +/+ , NUDC was still present in the inner nuclear layer (INL), inner and outer plexiform layers (IPL and OPL) and retinal ganglion cell (RGC) layer of the same retina (Figure 1G, middle panels, and Figure S2).Consistent with the western blotting quantification, there was a noticeable decrease in NUDC stain in the rNudC +/− retina compared with rNudC +/+ (Figure 1G, right panels).

| Loss of NUDC in rods triggers retinal degeneration as early as 3 weeks postnatal
We next determined whether any change in retinal thickness occurred via histological staining and counting nuclei within the ONL in 200-μm increments from the optic nerve head in 3-week-old animals.Spidergram (number of nuclei measured at various distances from the optic nerve head) analysis of hematoxylin and eosin-stained retinal cryosections (Figure 1H) revealed a 40% loss of ONL nuclei in rNudC −/− mice compared with that in rNudC +/+ and rNudC +/− mice as well as that in the floxed NudC (rNudC fl/fl iCre75 -) and iCre75-positive (rNudC+/+iCre75 + ) rNudC −/− or rNudC +/− control mice (Figure S3).This difference in retinal thickness across the ONL of the retina can also be seen in the IHC images for NUDC (Figure 1G).

| NUDC deficiency in rod photoreceptors results in ultrastructural malformations and mislocalized mitochondria
We evaluated the ultrastructure of the remaining rods in 3-week-old mice by transmission electron microscopy (TEM).TEM in 3-week-old rNudC −/− mice revealed a thinning retina, and while the remaining OSs appeared relatively normal compared with those of rNudC +/+ mice (Figure 2A), rNudC −/− rod cells frequently had an extended periciliary ridge (Figure 2B, asterisks), and in some cases, the discs in the rNudC −/− retina grew parallel to the axoneme (Figure 2B, arrowhead).TEM of the retina of 3-week-old rNudC +/− mice demonstrated compact nascent discs with morphology comparable to that of rNudC +/+ mice (Figure 2C).Lack of NUDC also resulted in mislocalized mitochondria within the IS.In rNudC +/+ mice, mitochondria were located against the plasma membrane (Figure 2D), and in knockout rNudC −/− mice, mitochondria appeared to be smaller and dispersed throughout the IS (Figure 2E).At 6 weeks of age, the rNudC +/− retina displayed discs similar to those of rNudC +/+ mice (Figure S4C).However, at 6 weeks of age, rod photoreceptors in rNudC −/− mice had completely degenerated, with no detectable OS region (Figure S4B).

NudC
We determined whether the loss of NUDC in rods specifically affected retinal function in 3-week-old and 6-week-old mice using electroretinography (ERG).At 3 weeks of age, rNudC +/− ERG traces were indistinguishable from those of rNudC +/+ controls, while rNudC −/− ERG traces (Figure 3A) displayed diminished a-wave and b-wave amplitudes at 0.25 and 25 cd s/m 2 , with a significant difference in overall light intensities [*p < .05,**p < .01](Figure 3B).The a-wave and b-wave implicit times of each genotype were indistinguishable across all light intensities (Figure 3B).
The a-wave and b-wave trace amplitudes of 3-week-old rNudC −/− and rNudC +/+ mice are displayed under photopic conditions at 2.5 and 25 cd s/m 2 (Figure 3C).There was a significant decrease in both a-wave and b-wave amplitudes (*p < .05,**p < .01,***p < .001);however, no difference in a-wave or b-wave implicit times was observed (Figure 3D).
Under scotopic conditions, the floxed NudC (rNudC fl/f iCre75 − ) and iCre75-positive (rNudC +/+ iCre75 + ) control genotypes showed slight differences, exhibiting a slightly larger a-wave amplitude and longer b-wave implicit time (Figure S5A,B).Under photopic conditions, there was no difference in the a-wave amplitude, but rNudC +/+ iCre75 + mice exhibited a delay in the a-wave implicit time.There was a slight difference in the b-wave amplitude, and differences in both genotypes were observed for the b-wave implicit time compared with that in the rNudC +/+ control (Figure S5C  phenotype expected due to the lack of photoreceptors (Figure 2H).Statistical analysis indicated a significant decrease in the a-wave amplitude and implicit time at all flash intensities for rNudC −/− mice compared with rNudC +/+ mice (*p < .05,**p < .01)(Figure S6D).A marked difference was also observed in the b-wave amplitude and implicit time in rNudC +/− mice compared with rNudC +/+ mice (**p < .01,***p < .0005)(Figure S4D).Functional losses were not observed in rNudC fl/fl iCre75 −/− or rNudC +/+ iCre75 +/− control mice (Figure S7).

| Loss of rod photoreceptors is concurrent with loss of rod OS proteins
Western blotting at 3 weeks of age showed that rod-specific NudC knockout resulted in a statistically significant loss of proteins that are normally trafficked to the OS or between the IS and OS of rod photoreceptors, including rhodopsin (RHO) and transducin (GNAT1), compared with rNudC +/+ mice (Figure 4A,B).Proteins normally located in the IS (LIS1 and RAB11A) remained unchanged in each genotype (Figure 4A,B).

F I G U R E 4 Protein levels are affected by the absence of NudC.
(A) Representative western blots of 3-week-old mouse retina, loaded in triplicate and probed for outer segment (OS) proteins peripherin (PRPH2) and rhodopsin (RHO), inner segment (IS) proteins LIS1 and RAB11A, and proteins that are translocated between the IS and OS, arrestin (SAG), and transducin (GNAT1), depending on light intensity.(B) Quantification of the western blots described in (A) revealed a significant loss of RHO and GNAT1 in rNudC −/− compared with rNudC +/+ .n = 3 for each genotype, *p < .05,**p < .01.
staining revealed mislocalization of RHO in −/− mice compared with the normal distribution of RHO only in the OS in the rNudC +/+ mouse retina (Figure 5A).Interestingly, mild mislocalization (Figure 5A, arrow) of RHO was observed in the rNudC +/− mouse retina at 3 weeks of age, prior to any retinal degeneration.RHO mislocalization also occurs in 2-week-old rNudC −/− mice as well (Figure S8A).By 3 weeks, the loss of NUDC resulted in several apoptotic TUNEL-positive cells (Figure 5B, arrows) in the ONL of the rNudC −/− retina, while the rNudC +/+ and rNudC +/− retina lacked TUNEL-positive staining (Figure 5B).
While the loss of NUDC results in a clear mislocalization of RHO at 2 (Figure S7A) and 3 weeks (Figure 5A), the same mice do not exhibit a mislocalization of PRPH2 by 3 weeks of age (Figure S9).

| Loss of NUDC increases glial reactivity
Immunofluorescence revealed increased microglial reactivity via Iba1 staining in rNudC −/− mice compared with that in rNudC +/+ mice (Figure 5C).While microglia are normally located in the plexiform layers, as they are in rNudC +/+ , in rNudC −/− , microglia were also found in the IS (Figure 5C, asterisk) as well as in the retinal ganglion cell (RGC) layer (Figure 5C, arrows).The rNudC −/− mouse retina exhibited an increase in glial reactivity indicated by GFAP-positive Müller glial processes extending up through the retina (Figure 5D, middle panel).At the 2-week timepoint, there is no evidence for Müller glial reactivity via GFAP expression (Figure S8B).

3.7
| Cofilin 1 is found mouse retina in proximity to NUDC Adult wild-type mouse retina stained for CFL1 showed staining in the IS (Figure 6A), and proximity ligation assay for NUDC and CFL1 resulted in puncta in the IS of adult wild-type retina (Figure 6B) indicating that NUDC and CFL1 were found within 40 nm of one another in the IS.

| Loss of NUDC leads to actin dysregulation
When the F-actin:G-actin ratio was compared, no statistically significant differences were observed among rNudC +/+ , rNudC −/− , and rNudC +/− mice (Figure 7A,B) at 3 weeks of age.Because NUDC interacts with the actinremodeling protein CFL1, 29 we investigated the levels of CFL and its phosphorylated form, pCFL1, in the rodspecific NudC knockout.We found that although CFL1 levels remained unchanged in all genotypes, a statistically significant increase in pCFL1 was observed in rNudC −/− mice compared with that in rNudC +/+ mice, and a significant decrease was observed in the rNudC +/− compared with rNudC +/+ (Figure 7C,D).

| DISCUSSION
Nearly 50 years ago, proteins involved in nuclear migration were identified using the filamentous fungus, Aspergillus nidulans, a well-established genetic model organism for studying dynein function in vivo. 3,36enetic analysis of the nuclear distribution (nud) family of proteins identified a set of genes (nudA, nudC, nudE, nudF, nudG, and nudK) with strong homology to mammalian dynein subunits.NudA is a homolog of the dynein heavy chain, sharing 52% identity with rat brain dynein heavy chain (DYNC1H1).NudC is 78% similar to human NUDC 37 and is required for nuclear movement. 4n HEK293T cells, NUDC interacts with DYNLRB1 (ROBL1) and light intermediate chain 1 (DYNC1LI1 or DLIC1), 38 but not with the motor domain.NudF, a homolog of human LIS1, interacts with NudE/NDEL1 and is involved in switching dynein from the autoinhibited form to the open form to facilitate dynein activation. 39,40udG is a dynein light chain of 94 amino acids with 66% identity to human dynein light chain 75% identity with human dynein light chain LC8-type 2 (DYNLL2), and nudK encodes ARP1 (actin-related protein 1), which is a component of the dynactin complex. 1 Deletion of nudA and nudF affects nuclear migration while deletions of the nudC gene in Aspergillus resulted a severe phenotype, profoundly affecting the morphology and composition of the cell resulting in lethality, presumably by inactivating dynein and vesicle axonemal transport. 6Mutations in NudC have not been associated with human disease to date, most likely due to its well-established role in mitotic cell division where loss of NUDC would cause cell death at an early stage of fetal development.Due to this limitation, the role of NUDC as a cofactor of dynein in postmitotic neurons such as photoreceptors is incompletely understood.
In rod photoreceptors, NUDC localizes to the IS, ONL (where dynein is known to reside), OPL, and INL (Figure 1G). 41Mouse and human NudC genes 37,42 encode proteins carrying coiled-coil, nuclear localization signals (NLS), and CS domains.CS domains are protein-protein interaction domains characteristic of proteins that act as co-chaperones of Hsp90 (Figure S1) and serve to stabilize proteins.To gain insight into the role of NUDC in mouse photoreceptors, we generated a rod-specific NudC conditional knockout facilitated by the Cre-driver iCre75, 35 which expresses Cre under the control of the rhodopsin promoter.Cre expression begins at P7-8 at which point rod photoreceptors are differentiated, and the involvement of dynein in nuclei positioning and establishment of the connecting cilium 43 has already occurred.This knockout strategy allows rods to develop normally until Cre levels are high enough to delete NudC.
By 3 weeks of age, the NUDC knockout was evident (Figure 1E,F), and by 6 weeks, rNudC −/− rod photoreceptors were completely absent (Figure S4).At 3 weeks, deletion of NUDC resulted in the loss of approximately 40% of rod photoreceptors (Figure 1H) and significantly lower levels of rod OS proteins (RHO, GNAT1) (Figure 4).Rhodopsin is synthesized at the rough endoplasmic reticulum surrounding rod nuclei and is normally transported via dynein to the periciliary ridge.In 2-and 3-week-old rNudC −/− , rhodopsin was retained in the ONL and IS (Figure 5A, Figure S8A middle panel, respectively) indicating that in the absence of NUDC, rhodopsin is not trafficked normally to the OS via dynein.Interestingly, peripherin-2, another protein which is localized to the OS, is not mistrafficked in the absence of NUDC (Figure S9).In zebrafish neurons, loss of NUDC leads to failed dynein-cargo attachment for the initiation of retrograde transport, presumably by arresting the Phi state of dynein.This results in an accumulation of not only cargo but also dynein in the axon terminal. 44Mitochondria, which are also normally trafficked via dynein and are normally found within the IS ellipsoid region, appeared unfused and scattered throughout the rNudC −/− IS (Figure 2E).As with rhodopsin mislocalization, mitochondrial mislocalization in the absence of NUDC may be due to a failure of dynein-cargo attachment.
In a rod-specific knockout of DYNC1H1, photoreceptors degenerated after P14, showing decreased scotopic ERG a-wave amplitudes, shortening of OS by P16, and complete photoreceptor degeneration by P30. 41Rod OS proteins (Rho, PDE) are reduced in the mutant OS starting at P16.In comparison, in the rNudC −/− mice, we observed functional loss with diminished a and b waves at 3 weeks of age (P21) and complete loss at 6 weeks of age (P42) (Figure 3).The functional losses seen in rNudC −/− at 3 weeks are similar to those observed in the Rho +/− mouse, which also exhibited a 30%-40% in the tor layer and diminished ERG response 12 weeks. 45ompared to the deletion of the heavy chain, the rod degeneration of the NudC knockout produces a phenotype that is similar to the DYNC1H1 knockout though the rate of rod photoreceptor degeneration is slower in the absence of NUDC.
The mechanism leading to dynein inactivation by loss of NUDC is currently unclear.NUDC is required to maintain the protein levels of NUDF/LIS1 in Aspergillus but nudF and nudC do not co-sediment in sucrose gradients. 39By contrast, the mammalian ortholog of NudC was identified as interacting with LIS1 in a yeast-twohybrid screen, presumably together with other components, 42 and it has been suggested that NUDC may act as a post-transcriptional regulator of NUDF/LIS1. 46NUDC, like LIS1, is required for neuronal migration during neocorticogenesis in radial glial progenitor cells 47 and deletions in LIS1 result in dynein inactivity and severe brain malformations, known as lissencephaly. 48While NudC mutation leads to a near complete loss of LIS1 protein in axon terminals, 44 we do not see changes in LIS1 protein levels in whole retinal lysates when NUDC is absent in rods at 3 weeks of age (Figure 4).The inhibited, weakly processive state (the φ or Phi particle) of dynein 49,50 is typically activated by binding to LIS1 and stabilizing the uninhibited open form of dynein cells, 51 which enables association with dynactin and a cargo-loading effector, such as bicaudal D homolog 2 (BICD2). 52The fact that LIS1 levels remain unchanged at 3 weeks of age in our rNudC −/− mouse retina could be because mouse NUDC acts upstream of LIS1 and its absence can arrest dynein in the Phi inactive form, 49,50 preventing its interaction with dynactin and cargo loading (Figure S10).LIS1 may also potentially be stabilized by the NUDC-like protein 2 (NUDCL2). 53,54NudC −/− rods showed evidence of glial reactivity (Figure 5C,D) and apoptosis (Figure 5B) by 3 weeks of age, and while the OSs are shorter compared to wild type, the surviving rod OS discs appear largely structurally normal.Many remaining rods displayed an extended periciliary ridge and abnormal disc overgrowth (Figure 2B, asterisks and arrowhead).This phenotype was not observed in the rod DYNC1H1 knockout but was previously recognized in a knockdown of NUDC in X. laevis. 18Elongated and overgrowth of discs also occur following cytocholasin D treatment of retinal eye cups 25 and in the rod-specific knockout of ArpC3 −/− in mice by P45. 55Recent results of photoreceptor ARPC3 (a subunit of the ARP2/3 complex) deletion demonstrated that loss of the F-actin network halted the generation of new discs and caused overgrowth of nascent discs. 56Further, it was shown that PCARE (photoreceptor cilium actin regulator) and its downstream interactor, WASF3 (Wiskott-Aldrich syndrome protein family member 3) regulate ciliary F-actin assembly and ciliary membrane expansion during disc morphogenesis. 57ur results show loss of NUDC in rods affected actin cytoskeleton regulation.NUDC is known to stabilize CFL1, which is a potent regulator of actin filament dynamics.The ability of CFL1 to bind and depolymerize actin is abolished by phosphorylation of serine residue 4 by LIMK1. 58,59We found CFL1 to be localized within close proximity (40 nm) to NUDC in the rod IS (Figure 6B), and the loss of NUDC resulted in an increase in the phosphorylated form of CFL1 (pCFL1).This increase in pCFL1 indicates that a significant amount of CFL1 was no longer bound to F-actin, and a portion of actin was no longer being clipped to form G-actin or branched F-actin.We expected to see an increase in the F-actin: G-actin ratio as a result of the loss of NUDC; however, rNudC −/− mice at 3 weeks of age did not reveal a difference compared with rNudC +/+ (Figure 7).This may be because a 40% loss of rods was already observed at this stage in rNudC −/− mice, F I G U R E 8 Schematic diagram of mouse rod photoreceptors under normal conditions (left panel) or in the absence of NUDC in rods (right panel) at 3 weeks of age.In rNudC +/+ mouse rods (left), rhodopsin (RHO, red circles) is properly trafficked to the outer segment (OS), and mitochondria are elongated and located next to the plasma membrane.In rNudC −/− mouse rods (right), the mitochondria appear smaller in ultrathin sections and scattered throughout the inner segment (IS), and some rods exhibit an extended periciliary ridge.Additionally, RHO is mislocalized in the IS and outer nuclear layer (ONL) of the retina.Müller glia (green cells) express GFAP throughout their processes, and microglia (yellow cells) migrate toward the outer retina.While NUDC normally stabilizes CFL1, in the absence of NUDC, there is an increase in the proportion of CFL1 that is phosphorylated (pCFL1).
any change in actin pools not evident in the remaining because we measured the total F-actin: Gratio in whole retinal lysates.There was, however, a slight (though not significant) increase in F-actin: G-actin in rNudC +/− mice at 3 weeks as well as a statistically significant decrease in pCFL.
While well established as necessary for cell division, our results indicate a role for NUDC in dynein-dependent protein and organelle trafficking as well as in actin cytoskeletal dynamics in postmitotic rod photoreceptors (Figure 8).Deletion of NUDC in rods presumably arrests dynein in the inactive Phi form, affecting protein and organelle trafficking in postmitotic rods, and loss of NUDC also affects pCFL1 levels and therefore actin cytoskeletal dynamics.These findings establish a critical role for NUDC in postmitotic rod photoreceptor maintenance and survival.

F I G U R E 6
Cofilin 1 (CFL1) localization and proximity to NUDC.(A) Adult mouse retina stained for CFL1 (green), wheat germ agglutinin (WGA, red), and the nuclear marker DAPI (blue).CFL1 is localized to the inner segment (IS).(B) Proximity ligation assay of adult mouse retina stained for DAPI (blue), WGA (red), and CFL1: NUDC within 40 nm (green puncta) showed that CFL1 and NUDC are in close proximity to one another throughout the IS (arrows).Scale bar = 20 μm.OS, outer segment; ONL, outer nuclear layer.

F I G U R E 7
Cytoskeletal dysregulation resulting from NudC deletion.(A) Western blot of F-actin and G-actin pools in retinal lysates, loaded in triplicate.(B) Quantification of the F-actin: G-actin ratio for each genotype (n = 3 for each genotype) revealed no statistically significant difference among genotypes.(C) Western blot of retinal lysates, loaded in triplicate and probed for cofilin (CFL1) and phosphorylated CFL1 (pCFL1).(D) Quantification of western blots in C revealed no change in CFL1 levels across genotypes, while an increase in pCFL1 was observed in the rNudC −/− retina compared with rNudC +/+ at 3 weeks (n = 3 for each genotype).*p < .05.F-actin, filamentous actin; G-actin, globular actin.