NIMA-related kinase 1 (NEK1) regulates the localization and phosphorylation of α-Adducin (ADD1) and Myosin X (MYO10) during meiosis

Summary statement: NEK1 kinase regulates the assembly and function of the meiosis I spindle by phosphorylating α-adducin (ADD1) and thereby facilitating its interaction with Myosin X (MYO10) Abstract: NIMA-related kinase 1 (NEK1) is a serine/threonine and tyrosine kinase that is highly expressed in mammalian germ cells. Mutations in Nek1 induce anemia, polycystic kidney and infertility. In this study we evaluated the role of NEK1 in meiotic spindle formation in both male and female gametes. Our results show that the lack of NEK1 provokes an abnormal organization of the meiosis I spindle characterized by elongated and/or multipolar spindles, and abnormal chromosome congression. The aberrant spindle structure is concomitant with the disruption in localization and protein levels of myosin X (MYO10) and α-adducin (ADD1), both of which are implicated in the regulation of spindle formation during mitosis. Interaction of ADD1 with MYO10 is dependent on phosphorylation, whereby phosphorylation of ADD1 enables its binding to MYO10 on mitotic spindles. Reduction in ADD1 protein in NEK1 mutant mice is associated with hyperphosphorylation of ADD1, thereby preventing the interaction with MYO10 during meiotic spindle formation. Our results reveal a novel regulatory role for NEK1 in the regulation of spindle architecture and function during meiosis.


Introduction
Meiosis is a specialized cell division characterized by a single round of DNA replication 55 followed by two rounds of chromosome segregation, resulting in the formation of haploid gametes for sexual reproduction. At the first male meiotic division (MI), homologous (maternal and paternal) chromosomes must segregate equally into two daughter cells, each of which then undergoes a mitosis-like second meiotic division (MII) in which sister chromatids separate into the haploid gametes. In mammals, meiosis in males results in four haploid gametes, while 60 meiosis in female results in one haploid gamete per meiosis, the remaining genetic material being distributed between two polar bodies. Regardless of the sex of the individual, in order to achieve accurate segregation at both divisions, tension must be established on the meiotic spindles and this is achieved by the formation of crossovers between homologous chromosomes in meiosis I, and by cohesion between sister chromatids in both meiosis I and 65 meiosis II . Cohesion is also necessary for sister kinetochore attachment to microtubules during meiotic spindle formation (Duro and Marston 2015).
Several families of kinases such as Polo kinases, Aurora kinases and NEK kinases have been implicated in the regulation of cell cycle events in both mitosis and meiosis (Fry et al. 70 2012, Meirelles et al. 2014). The founding member of the NEK family is the Aspergillus nidulans never-in-mitosis-gene-A (NIMA) protein (Oakley and Morris 1983). NIMA is a serine/threonine kinase involved in the initiation of mitosis and in promoting chromosome condensation. Genetic ablation of NIMA results in cell cycle arrest at G2, while overexpression of NIMA leads to premature entry into mitosis (Fry et al. 2012). In mammals, there are 11 orthologs of NIMA that 75 comprise the NIMA-like (NEK) family of kinases. Nek1 is unique in that it encodes both the serine/threonine kinase activity typical of the NEK family as well as tyrosine kinase activity that is not a feature of other NEKs (Letwin et al. 1992). NEK1 has been implicated in ciliogenesis (White and Quarmby 2008) and DNA damage response (Chen et al. 2011, Hilton et al. 2009, Liu et al. 2013, Patil et al. 2013. Additionally, NEK1 is highly expressed in mouse germ cells 80 (Upadhya et al. 2000), where it appears to play essential roles in meiosis I and possibly also at later stages (Brieño-Enriquez et al. 2016, Holloway et al. 2011. The Kat2J allele of NEK1 is a spontaneous point mutation within the coding sequence of Nek1 that results in a premature stop codon leading to a truncated protein lacking the both 85 kinase domains (Upadhya et al. 2000). Nek1 kat2j/kat2j mice display polycystic kidneys, dwarfism, anemia and male sterility (Janaswami et al. 1997, Upadhya et al. 2000. Our previous studies of meiosis in Nek1 kat2j/kat2j mice showed that the loss of NEK1 activity induces aberrant retention of cohesin subunit Structural Maintenance of Chromosomes protein 3 (SMC3), at the end of meiotic prophase I (Holloway et al. 2011). More recently, we showed NEK1 regulates cohesin 90 removal, in part, through regulation of wings apart-like homolog (WAPL) during meiotic prophase I (Brieño-Enriquez et al. 2016), as part of the so-called Prophase pathway for cohesin removal. This regulation of WAPL by NEK1 is mediated through WAPL interactions with the cohesin-associated protein, PDS5 homolog B (PDS5B), and protein phosphatase 1 gamma (PP1γ), both of which also interact with NEK1 (Brieño-Enriquez et al. 2016).

95
In the current study, and given the conserved function of NEK proteins in spindle assembly dynamics, we investigated the role of NEK1 downstream of prophase I events. Our results demonstrate that loss of NEK1 induces failure of meiotic spindle organization in both males and females, leading to elongated spindles, multipolar spindles, and abnormal 100 chromosome congression. Concurrent with this, we observe altered levels and localization of two proteins known to regulate mitotic spindle dynamics: the unconventional myosin, myosin X (MYO10), and α-adducin (ADD1). Interaction of ADD1 with MYO10 is dependent on phosphorylation of ADD1 which allows for the interaction of these two proteins on mitotic spindles (Chan et al. 2014). Depletion of ADD1 or changes in its phosphorylation status results 105 in abnormal mitotic spindles (Chan et al. 2014). Our studies demonstrate that loss of NEK1 kinase causes changes in MYO10 and ADD1 protein levels, and hyperphosphorylation of ADD1 caused by abnormal function of PP1 during meiotic spindle formation, independently of the role of NEK1 in prophase I events.

Loss of NEK1 induces abnormal meiotic spindle formation and failed chromosome congression at the first meiotic division in males and females
We analyzed the role of NEK1 in spindle formation and chromosome orientation at MI in 115 oocytes and spermatocytes from Nek1 +/+ and Nek1 kat2j/kat2j mice. In male mice, we evaluated the shape, number of poles and misaligned chromosomes in both Nek1 +/+ (n=100) and Nek1 kat2j/kat2j (n=100) spermatocytes. MI spindles from Nek1 +/+ spermatocytes showed a bipolar structure with the chromosomes aligned at metaphase plate (Fig. 1A), while spindles from Nek1 kat2j/kat2j spermatocytes showed defects in the structure of the spindle as well as in chromosome 120 congression. These defects included MI spindles without a pole, with only one pole, with misaligned chromosomes (Fig. 1B), and with multiple poles (Fig. 1C). A total of 55% of spindles from Nek1 kat2j/kat2j males were abnormal, while only 18% of the spindles from Nek1 +/+ male spermatocytes showed abnormalities (Fig. 1D). The percentage of misaligned chromosomes in Nek +/+ spermatocytes was 18%, while in Nek1 kat2j/kat2j mice the percentage reached the 38% 125 (Fig. 1E).
The analysis of MI spindles in oocytes from Nek1 +/+ females showed a bipolar structure of the spindle and appropriate congression of the chromosomes at the metaphase plate ( Fig. 2A).
However, MI spindles from Nek1 kat2j/kat2j females showed a distortion of the structure of the spindle characterized by the presence of multiple spindles, mini spindles and misaligned chromosomes (Fig 2B, 2C). We evaluated the percentage of abnormal spindles in both wildtype oocytes (n=100) and mutant oocytes (n=100). In Nek1 kat2j/kat2j female spindles, 82% of the MI spindles were abnormal, while only 22% of the Nek1 +/+ oocytes showed any abnormality (Fig.   2D). Similarly, 88% of MI spindles from Nek1 kat2j/kat2j oocytes showed at least one misaligned chromosome, compared to only 23% in oocytes from Nek1 +/+ mice (Fig. 2E).

Nek1 ka2tj/kat2j spermatocytes and oocytes have an abnormal spindle length and width
The appropriate length and width of the spindle are critical for establishing tension during segregation, and thus both parameters were evaluated them in male and female spindles. We measured the distance between poles in MI spermatocyte spindles and we observed a 140 significant increase of the length of Nek1 kat2j/kat2j males (27.9 µm ± 7.0 s.d.) compared to that of Nek1 +/+ males (23.1 µm ± 2.0 s.d.; Unpaired t test p=0.0020; Fig. 3A). Increases in MI spindle length in male spermatocytes were also accompanied with increases in the width in MI spindles from Nek1 kat2j/kat2j male mice (26.7 µm, ± 6.4 s.d.) compared to that of Nek1 +/+ male mice (22.9 µm ± 2.1 s.d; Unpaired t test p=0.0097; Fig. 3B). We also analyzed the length and width of the 145 female spindles. Nek1 kat2j/kat2j oocytes showed an increase in the length of the spindle in MI oocytes (31.6µm ± 7.3 s.d.) compared to that found in Nek1 +/+ females (26.0 µm ± 2.3 s.d.; Unpaired t test p=0.0007; Fig. 3C). The lack of NEK1 also resulted in altered width of the MI spindles in female meiosis . Nek1 kat2j/kat2j oocytes showed a decrease in the width of the spindle (15.1 µm ± 4.0 s.d.) compared to the Nek1 +/+ oocyte spindles (17.0 µm ± 1.7 s.d.; Unpaired t 150 test p=0.039; Fig. 3D). Taken together our results show that NEK1 is required for the correct spindle formation and chromosome alignment during meiosis in both male and female gametes.

Lack of NEK1 results in deregulation of myosin X (MYO10) and α-adducin (ADD1)
Given the disrupted spindle apparatus during meiosis I, we investigated possible protein 155 targets of NEK1 to explain this phenotype. In a previous study, we had performed Mass spectrometry (MS) screening using testis protein extracts from Nek1 +/+ and Nek1 kat2j/kat2j mouse (Brieño-Enriquez et al. 2016). Review of these data using GO terms for centrosome and centromeric proteins revealed no significant differences in these proteins levels in Nek1 kat2j/kat2j mouse extracts compared to wildtype mice (Fig. S1). Several proteins associated with spindle 160 formation and regulation were identified and quantified, however no differences between Nek1 +/+ and Nek1 kat2j/kat2j mouse were observed for tubulins, actins and SAC proteins ( Previous reports showed that MYO10 has a direct function in meiotic spindle formation (Sandquist et al. 2016, Weber et al. 2004, Woolner et al. 2008. In contrast to the tubulins, 165 actins, and other spindle proteins that did not appear to be altered in the absence of NEK1, we observed that the levels of the myosins, MYO5, MYO7A, MYO10 and MYO15, were significantly increased in testis protein extracts from Nek1 mutant mice compared to wildtype littermates (Table S1). By contrast, the myosins, MYO1B, MYO1C, MYO1D, MYO7A and MYO9B were unaltered. The function of MYO10 in spindle formation and integrity is related to its binding 170 partner α-adducin (ADD1) (Chan et al. 2014). Our MS results revealed that ADD1 showed lower protein levels in Nek1 kat2j/kat2j mouse compared to wildtype mouse (Table S1). The changes in protein levels observed by MS indicates that the loss of NEK1 results in an imbalance in the ratio MYO10/ADD1, and this could be affecting the proper spindle formation and function.

Nek1 kat2j/kat2j spermatocytes and oocytes show aberrant localization of myosin X (MYO10) on meiotic spindles
We evaluated the localization of MYO10 on MI spindles from male spermatocytes and female oocytes. Immunofluorescence (IF) analysis in Nek1 +/+ spindles revealed that MYO10 localizes along the β-tubulin fibers specifically in the region where tubulin interacts with the 180 chromosome (the kinetochores) in mouse spermatocytes (Fig. 4A). We analyzed the number of MYO10 foci per nucleus in the coronal plane ( Fig. 4B) and found an increase in the number of foci in the Nek1 kat2j/kat2j mice (75.4 ± 6.7 s.d.) compared to Nek1 +/+ animals (66.3 ± 5.5 s.d.; Unpaired t test p=0.0001). Analysis of transverse sections of male spindles revealed that there is a subtle accumulation of MYO10 at the spindle pole in Nek1 +/+ male mice (Fig. 4A), and this 185 was disrupted in Nek1 kat2j/kat2j mice (Fig. 4B). In the transverse plane, we observed an increase in the MYO10 foci number in Nek1 kat2j/kat2j mice (74.7 ± 5.4 s.d.) compared to the Nek1 +/+ mice (61.7 ± 4.4 s.d.; Unpaired t test p=0.0001). The increase in MYO10 focus counts was supported by our mass spectrometry analysis showing increased MYO10 protein in testis extracts from Nek1 kat2j/kat2j males compared to Nek1 +/+ mice (Table S1).

Localization of ADD1 on spermatocytes and oocytes is disrupted in absence of NEK1
Our previous MS results showed that ADD1 proteins levels are reduced in testis extracts from Nek1 kat2/kat2j mouse compared to wildtype littermates (Table S1) (Brieño-Enriquez et al. 2016). Thus, we evaluated the localization and protein levels of ADD1 in both male and female 200 spindles during meiosis I. ADD1 localizes both to the poles of the spermatocyte spindles ( Fig.   5A, asterisk) and with β-tubulin fibers of the spindle (Fig. 5A, arrow heads). We evaluated the ADD1 focus number associated with the spindle structure in the coronal and transverse planes.

210
In oocytes from Nek1 +/+ females, ADD1 also localized to the spindle poles as two large foci on each pole (3.6 ± 0.9 s.d.) (Fig. 5C, 5D). This localization pattern was disrupted in Nek1 mutant females, where we observed multiple foci on the poles (6.08 ± 1.4 s.d.; Unpaired t test p=0.0002) indicating the mislocalization of the protein could be result as a pole fragmentation like in somatic cells (Chan, et al., 2014).

Loss of NEK1 activity induces abnormal phosphorylation of ADD1
During mitotic spindle formation mammalian cell lines, ADD1 binds to the motor domain of MYO10 on the spindle. ADD1 and MYO10 interaction on mitotic spindles is negatively regulation by phosphorylation. Phosphorylation of ADD1 induces the loss of the interaction 220 between both proteins and therefore the complex unloads from the spindle (Chan et al. 2014).
Phosphorylation of ADD1 results in abnormal meiotic spindle morphology (elongation, multipolar spindles and aberrant chromosome alignment) and loss of its interaction with MYO10 (Chan et al. 2014). We evaluated the proteins levels of ADD1 in testis lysates and isolated oocytes. In both male and female germ cells, we observed a double band in Nek1 +/+ lysates (both bands 225 were included for quantitation), however this double band is reduced or not present in protein lysates from Nek1 -/testes and oocytes. Our WB results revealed a decrease in the protein levels in testis from mutant mice compared to wildtype testis and oocyte lysates ( Fig. 6A; Unpaired t test p=0.0031). Decreased levels of ADD1 and the loss of bands in WB analysis of Nek1 kat2j/kat2j mouse compared to wildtype littermates suggested that the lack of NEK1 activity 230 affects ADD1 phosphorylation, leading to a reduced interaction with MYO10. To test this hypothesis, we screened our previous phosphoproteomics data obtained from Nek1 mutant animals, specifically focusing on changes in ADD1. Loss of NEK1 results in a significant increase in the phosphorylation of ADD1 (Fig. 6B) (Brieño-Enriquez et al. 2016). Specifically, loss of NEK1 results in hyperphosphorylation of ADD1 on serine 465 (S465) (Fig. 6C). In 235 oocytes the WB analysis revealed a decrease in the protein levels in Nek1 kat2j/kat2j oocytes compared to Nek1 +/+ oocytes ( Fig. 6D; Unpaired t test p=0.0003). To evaluate the changes in the ADD1 protein levels at various times of oocyte culture, we observed ADD1 protein levels at 6 hours after initial culture. Interestingly, in Nek1 +/+ oocytes we observed the loss of the heavier band indicating a change in ADD1 phosphorylation. However in the Nek1 kat2j/kat2j oocytes we 240 observe that both bands almost disappear indicating that there is a change in the phosphorylation in the mutants but also degradation of the protein in the mutant oocytes.
We previously reported the interaction of PP1γ with NEK1 and its function regulating the prophase pathway during meiosis (Brieño-Enriquez et al. 2016). To test if the hyperphosphorylation of ADD1 depends on PP1 activity, we cultured Nek1 +/+ oocytes for 6h in the 245 presence of the PP1 inhibitor, Calyculin (Fig.6E). The WB analysis of these cultures showed that the inhibition of PP1 induces a statistically significant increase in the ADD1 protein levels, predominantly the heavier band, suggesting that the inhibition of PP1 maintains the phosphorylation status of ADD1. By contrast, loss of NEK1 loss did no affect MYO10 phosphorylation status, suggesting that the effects on MYO10 protein may be indirect, through 250 ADD1. Thus, these results suggest that the lack of NEK1 results in increased ADD1 phosphorylation that is associated with a disruption in ADD1-MYO10 interactions.

Discussion
In eukaryotes, the structure orchestrating chromosome alignment and segregation 255 during cell division is the microtubular spindle (Bennabi et al. 2016). To ensure correct spindle dynamics, the spindle assembly checkpoint (SAC) becomes activated in situations where tension is not appropriately established between the chromosomes and the spindle poles, and this will block the progression of the cell cycle to prevent aberrant segregation. In addition to the canonical SAC, other levels of control have been described, including MYO10 (Sandquist et al. 2016, Weber et al. 2004, Woolner et al. 2008. In oocytes and frog embryos, disruption of MYO10 negatively affects nuclear anchoring and meiotic spindle formation (Weber et al. 2004), while depletion of MYO10 in frog epithelial cells induces abnormal spindle movements, spindle elongation, multi-spindle and pole fragmentation (Kwon et al. 2015, Woolner et al. 2008).
MYO10 has an amino-terminal globular head domain that harbors acting biding and ATPase 265 activities, giving to the protein the capacity to bind F-actin. The tail of MYO10 has the MyTH4 and FERM domains that give to the protein the ability to interact with microtubules (Hirano et al. 2011, Weber et al. 2004. Accordingly, recent studies have demonstrated that loss of Myo10 results in spindle dysfunction, while overexpression of only the MyTH4 domain alone affects the structure and shape of the spindle (Sandquist et al. 2016).

270
In the current study, we present evidence to suggest that loss of NEK1 induces increases in MYO10 protein levels, and that this in turn may result in spindle defects. Our interpretation that overexpression of Myo10 leads to spindle defects is in line with the suggestion that overexpression of the MyTH4 domain of MYO10 results in competition with the wildtype protein, leading to the displacement of the latter from the spindle (Hirano et al. 2011). Furthermore, it 275 has been suggested that over-expression of only the MyTH4 may disrupt the functional link between F-actin, microtubules and MYO10. In our data (Table S1), we observed an increase in the MYO10 protein levels without changes in F-actin or microtubules that could be inducing an imbalance in the protein levels provoking a similar phenotype to the overexpression of the MyTH4 domain.

280
ADD1 is an actin-biding protein that is important for membrane stabilization (Anong et al. 2009) and for cell-cell adhesion (Abdi andBennett 2008, Franco et al. 2016 domain is necessary for its interaction with F-actin, spectrin, and calmodulin (Hirano et al. 2011, Kaiser et al. 1989, Matsuoka et al. 1996, Scaramuzzino and Morrow 1993. Previous studies in somatic cells showed that loss of ADD1 leads to a failure in mitotic spindle formation characterized by distortion, elongation, multipolar spindles and abnormal chromosome alignment (Chan et al. 2014). Here, we show that loss of NEK1 leads to a reduction in ADD1   unclear what kinase is phosphorylating ADD1 during meiosis, since the increased phosphorylation in the absence of NEK1 suggests the up-regulation of a kinase that is itself regulated by NEK1 (either directly or indirectly). Clear candidates could be CDKs, however, our 305 MS results did not show any significant change in the profile of CDK in the absence of NEK1, and specifically no differences in CDK1 indicating that increases in phosphorylation of ADD1 does not depend of CDK1 pathway. Alternatively, the increased phosphorylation of ADD1 could suggest the existence of a phosphatase that is inactivated by the loss of NEK1, allowing the hyperphosphorylation of protein. We previously described that in Nek1 mutants, there is a down 310 regulation and abnormal phosphorylation in whole testis lysate of PP1γ, and our results of oocyte cultures in the presence of a PP1 inhibitor suggest that the increase in ADD1 phosphorylation could be mediated by a similar mechanism.
In summary, we propose that the loss of NEK1 activity induces abnormal spindle formation through a mechanism mediated by the imbalance of MYO10 and ADD1. The 315 decreases in ADD1 and it abnormal localization on the meiotic spindle is mediated by an increase its phosphorylation that leads to loss of interaction of MYO10 with ADD1. Taken together our results show the importance of NEK1 in the control of spindle formation during male and female meiosis.

Mice and genotyping
All mouse studies were conducted with the prior approval of the Cornell Institutional Animal Care and Use Committee (Protocol 2010-0054). The Nek1 kat2j/kat2j line was obtained originally from Jackson Laboratory (Bar Harbor, Maine), and has been established in our mouse colony 325 for more than nine years. Genotyping of this mouse line was performed following the protocol described elsewhere (Holloway et al. 2011). For the purposes of these studies, male mice were 8 weeks old and female where used at 24-28 days old. Homozygous mutant animals of (Nek1 kat2J/kat2J ) were compared with wild type (Nek1 +/+ ) littermates on a C57Bl/6J background.

Male and female spindle preparation
Male spindle analysis was performed in testis from 8 week old mice following the protocol described elsewhere (Kotaja et al. 2004, Page et al. 1998). Briefly, following euthanasia, the testes were removed, decapsulated and placed in PBS. Using fine forceps, tubules were separated and analysed under a stereomicroscope. Differential light absorption of the tubules 335 creates different zones that can be defined by different cell populations (Kotaja et al. 2004). The differentent zones are consequence of the paracrine regualtion by Sertoli cells, spermatogenesis proceeds in synchronized waves along the seminiferus tubules, and every given cross-section contains only certain cell types, that produces a diferent light absorbation creating zones (Kotaja et al. 2004). These zones are the weak spot that correspond to the 340 spermatogenic stages XII-I, the strong spot (stages II-VI), the dark zone (stages VII-VIII) and the pale zone (stage IX-XI) We selected the zone corresponding to the spermatogenetic development XI and XII that are rich in meiotic cells during both meiosis I and meiosis II.
Tubules were placed in fresh PBS and these required zones were excised and placed on polylysine-coated slides (P4981, Thermo Fisher). Tubule sections were fixed for 10 minutes (1% 345 paraformaldehyde and 0.15% Triton-X, pH 9.2) and then squashed under a coverslip. Slides with the squashed tubules were put into liquid nitrogen for 20 seconds and thereafter the coverslip removed and washed with PBS-0.4% Kodak Photo flo (Kodak) (3 times for 5 minutes each). Slides were kept at -80°C or stained immediately.

370
Slides were washed and mounted with Prolong Gold antifade (Molecular Probes). Image acquisition was performed using a Zeiss Imager Z1 microscope under 20X, 40X or 63X magnifying objectives, at room temperature. Images were processed using ZEN 2 (Carl Zeiss).

375
Mass spectrometry was performed in the Cornell University Proteomics and Mass Spectrometry facility. 2D LC-MS/MS raw data files were acquired using Orbitrap Elite (Thermo Scientific). We performed a database search using Mascot searching against the SwissProt mouse database from Uniprot website (http://www.uniprot.org) using Mascot software version 2.3.02 (Matrix Science, UK). The default Mascot search settings were as follows: one missed cleavage site by 380 trypsin allowed with fixed MMTS modification of cysteine, fixed four-plex iTRAQ modifications on Lys and N-terminal amines and variable modifications of methionine oxidation, deamidation of Asn and Gln residues, and 4-plex iTRAQ on Tyr for iTRAQ 4-plex analysis. One or twomissed cleavage site by trypsin allowed with fixed carboxamidomethyl modification of cysteine, fixed six-plex TMT modifications on Lys and N-terminal amines and variable modifications of 385 methionine oxidation, deamidation of Asn and Gln residues, and 6-plex TMT on Tyr for TMT 6plex analysis. The quantitative protein ratios were weighted and normalized by the median ratio with outlier removal set automatic in Mascot for each set of experiments. Only those proteins with ratios equivalents to two-fold increase or two-fold reduction were considered significant. To obtain the specific post-traslational modification of the peptides we perfumed TiO2 enrichment 390 followed by the proteomics analysis as was described above.

Western blotting
Whole testis protein and oocyte protein were extracted by sonication in RIPA buffer. Samples were boiled for 5 min in sample buffer, electrophoresed on SDS-polyacrylamide gels (8%) and 395 transferred to nitrocellulose membranes. Primary antibody incubation was performed overnight at 4°C at 1:1000 dilution (antibodies are the same used in spindle staining). Incubation with secondary antibodies was performed for two hours at room temperature (secondary HRP conjugated antibodies were obtained from Pierce, Life Technologies). Signal-detection was carried out using the superSignal substrate (Thermo Scientific). Loading control was performed 400 using GAPDH-HPR (PA1-987-HRP, from ThermoFisher). Images were captured with BIO RAD Image Lab 5.1and analyzed by ImageJ version 1.49v (http://rsbweb.nih.gov/ij).

Acknowledgements
We are grateful to Peter Borst for assistance in animal care and genotyping. We thank Dr.