Nuclear Envelope Retention of LINC Complexes Is Promoted by SUN-1 Oligomerization in the Caenorhabditis elegans Germ Line

SUN (Sad1 and UNC-84) and KASH (Klarsicht, ANC-1, and Syne homology) proteins are constituents of the inner and outer nuclear membranes. They interact in the perinuclear space via C-terminal SUN-KASH domains to form the linker of nucleoskeleton and cytoskeleton (LINC) complex thereby bridging the nuclear envelope. LINC complexes mediate numerous biological processes by connecting chromatin with the cytoplasmic force-generating machinery. Here we show that the coiled-coil domains of SUN-1 are required for oligomerization and retention of the protein in the nuclear envelope, especially at later stages of female gametogenesis. Consistently, deletion of the coiled-coil domain makes SUN-1 sensitive to unilateral force exposure across the nuclear membrane. Premature loss of SUN-1 from the nuclear envelope leads to embryonic death due to loss of centrosome–nuclear envelope attachment. However, in contrast to previous notions we can show that the coiled-coil domain is dispensable for functional LINC complex formation, exemplified by successful chromosome sorting and synapsis in meiotic prophase I in its absence.

biological readout. The germ line comprises a syncytium in which mitotically-proliferating nuclei at the distal portion give rise to meiocytes. These enter meiosis and go through the prolonged prophase of the first meiotic division (Greenstein 2005). At the gonad bend, meiocytes undergo cellularization to form oocytes. Once they pass the spermatheca they are fertilized and immediately undergo two meiotic divisions (Hubbard and Greenstein 2005).
Successful gamete formation requires proper segregation of the parental homologous chromosomes in the first meiotic division. To achieve this with high accuracy, homologous chromosomes must move, pair, and recombine during prophase (Gerton and Hawley 2005). Movement is achieved by transmitting microtubule-mediated forces to chromosome ends via the SUN-1-ZYG-12 (KASH homolog expressed in C. elegans germ line) NE bridge (Baudrimont et al. 2010;Wynne et al. 2012;Labrador et al. 2013;Woglar and Jantsch 2014). Interfering with force transmission leads to a lack of presynaptic chromosome alignment and synaptonemal complex (SC) establishment between nonhomologs (Penkner et al. 2007;Sato et al. 2009;Labrador et al. 2013). In mitotic cells SUN-1 is evenly distributed along the NE. During the chromosome movement stage, corresponding to leptotene/zygotene and also known as the transition zone (TZ) in C. elegans; SUN-1 relocates to form pronounced aggregates around chromosome ends in close proximity to the NE Sato et al. 2009), a feature conserved in vertebrates (Ding et al. 2007;Schmitt et al. 2007), which is mirrored by ZYG-12 aggregates (Labrador et al. 2013;Sato et al. 2009). Aggregate formation and consequent chromosome end mobilization are seemingly controlled by signals from inside the nucleus involving the chk-2 and polo kinases (PLKs 1 and 2) Harper et al. 2011;Labella et al. 2011).
sun-1 null mutants are sterile and gonads degenerate with decreased numbers of irregularly-sized aneuploid nuclei (Fridkin et al. 2004;Penkner et al. 2007). The LINC complex also contributes to gonad architecture by ZYG-12-mediated recruitment of dynein at the NE to build up the tension required for nuclear positioning (K. Zhou et al. 2009). In embryos this complex mediates centrosome attachment to the NE (Malone et al. 2003).
The role of the cc motifs in the oligomeric nature of SUN proteins has been put forward in several studies (Padmakumar et al. 2005;Crisp et al. 2006;Q. Wang et al. 2006Q. Wang et al. , 2012Lu et al. 2008;Sosa et al. 2012;Z. Zhou et al. 2012). The structure of the mammalian SUN-domain protein SUN2 has been elucidated in the context of binding to the KASH partner Nesprin (Sosa et al. 2012;W. Wang et al. 2012;Z. Zhou et al. 2012). SUN2 forms trimers mediated by the region adjacent to the SUN domain, which forms a helical stem to create a clover-like structure. KASH peptides reside between the trimeric SUN domains and multiple hydrophobic interactions mediate KASH-peptide binding to SUN-domain interfaces ( Figure 1A). Measurements have shown that the triple-helical cc fits into the perinuclear space (Sosa et al. 2012), suggesting that the LINC complex, and the SUN-protein cc domain in particular, have a role in maintaining an even spacing. Interestingly, in C. elegans the UNC-84 protein can be mutated without affecting nuclear membrane spacing, except in body wall muscle cells where nuclei are under mechanical stress (Cain and Starr 2015). The structural analysis of SUN-KASH peptides led to the development of models for higher-order assembly of the SUN-KASH module. The SUN trimers may bind to multiple KASH peptides from several different KASH oligomers to build up higherorder clusters of the LINC complex (Sosa et al. 2012).
Previous yeast two-hybrid (Y2H) assays proposed a role for C. elegans SUN-1-predicted cc regions in self-interaction (Minn et al. 2009). To investigate the function of these domains in more detail, we generated several mutants with deletions in the cc regions and assessed the phenotype and cellular readout of the mutations in the germ line and embryos. Here, we show that despite disrupting SUN-1 oligomerization, deletion of the cc domains does not abrogate the formation of a functional LINC complex. Surprisingly, all aspects of SUN-1-mediated chromosome pairing and synapsis do not rely on the SUN-1 oligomerization domains, albeit chromosome movement exerts mechanical strain on the LINC complex. Strikingly, the cc region has a role in efficient protein retention at the NE, which becomes more pronounced prior to oocyte cellularization in the germ line.

Immunoprecipitation analysis
Worms were harvested in homogenization buffer and frozen at 280°. Worms were ground in liquid nitrogen and sonicated three times on ice (30 sec, amplitude of 70-80%). Protein lysates were centrifuged and supernatants were added to GFP-Trap agarose beads (ChromoTek) and incubated overnight at 4°. Beads were washed three times in wash buffer and boiled in SDS sample buffer at 90°for 10 min. Bound proteins were analyzed by SDS-PAGE.

Western blot analysis
Two hundred very young adult worms (containing at the most one to two eggs) were picked into lysis buffer (13 TE; 23 complete protease inhibitor) and snap frozen three times in liquid nitrogen. After the last thawing step, Laemmli buffer was added to a final concentration of 13 and worms were boiled for 10 min. Protein extracts were resolved on an acrylamide gel and transferred for 1 hr at 4°. Anti-SUN-1 ) and anti-actin (Santa Cruz Biotechnology) antibodies were both diluted to 1:3000 in 13 TBST (13 TBS, 0.1% Tween-20) to probe the membranes.
Immunofluorescence analysis L4 hermaphrodites were incubated at 20°for 24 hr. Gonads were dissected in PBS and fixed in 1% formaldehyde for 5 min (Martinez-Perez and Villeneuve 2005). For immunostaining of gonads, nonspecific binding sites were blocked with 3% BSA in PBS for 20 min. All antibodies were diluted in 3% BSA in PBS. Gonads were incubated with primary antibody overnight at 4°and with secondary antibody for 2 hr at room temperature.

Fluorescence microscopy
All microscopy evaluations were done using a DeltaVision microscope with SoftWoRx image analysis deconvolution software (Applied Precision), ImageJ (National Institutes of Health), and Adobe Photoshop software. Intensity measurement of SUN-1::GFP in the nuclear rim of the mutant and wild-type animals was performed on nondeconvolved images. The average signal intensity of the rim and the average background for each nucleus were measured using ImageJ, and the GFP signal intensity was calculated by subtracting the mean fluorescence intensity of both.

Live imaging
Live imaging was performed as described by (Baudrimont et al. 2010) and analysis was done with ImageJ using the stackreg and manual tracking plugins.

Photo-conversion
To perform photo-conversion on sun-1(D158-235)::eos3.2, young adult hermaphrodites were mounted in PBS containing 10 mM levamisole. The experiment was done using a DeltaVision deconvolution microscope (Applied Precision) with a DAPI filter, 100% laser intensity, and a 603 objective. Images of dissected gonads were acquired 4 hr postrecovery. Photo-conversion experiments on wild-type hermaphrodites tagged with Dendra2 were done with an LSM5 microscope (Carl Zeiss, Thornwood, NY) and were performed using a 405 nm filter with 100% laser output. A few cell rows of each zone were photo-converted and images were acquired after 8 hr, when photo-converted nuclei had reached the next meiotic stage.
To study SUN-1 de novo synthesis in embryos, young sun-1(wt)::dendra2 hermaphrodite adults were mounted in PBS containing 10 mM levamisole and 21 diakinesis oocytes were bleached using a DeltaVision deconvolution microscope with a DAPI filter and 100% laser intensity. Images were acquired 90 min postrecovery.

RNA interference
RNA interference (RNAi) feeding was performed with the zyg-12 clone from the Ahringer library as described in (Kamath et al. 2001). L4 worms were incubated at 20°for 48 hr before dissection.

Data availability
Strains are available upon request. Sequence data are available at GEO under the submission number GSE76773. The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article.
To study the importance of the cc motifs in SUN-1 oligomerization, we performed co-immunoprecipitation (coIP) analysis of GFP-tagged single-copy integrated sun-1 wild type [SUN-1(WT)::GFP, a functional transgene described in Woglar et al. (2013)], or mutants with deleted cc motifs SUN-1(WT)::GFP coIP and the subsequent Western blot analysis revealed 2 bands of 100 and 60 kDa, corresponding to the GFP-tagged transgene and coprecipitated endogenous SUN-1. In the absence of the cc motifs, we could not detect a band corresponding to coprecipitated endogenous SUN-1. This result shows that the transgenic-tagged protein cannot interact with wild-type protein in the absence of its luminal cc domain ( Figure 1B). SUN-1 is evenly distributed throughout the NE during late meiotic prophase and mitotic interphase. In contrast, during early meiotic prophase and mitotic metaphase/anaphase SUN-1 forms aggregates at chromosome ends or spindle poles (Penkner et al. 2007;Penkner et al. 2009). We next investigated whether differences in the SUN-1 localization pattern (as aggregates or evenly distributed throughout the NE) correlated with cc-driven SUN-1 oligomerization in the germ line. To address this, we co-immunoprecipitated SUN-1(WT)::GFP/SUN-1(WT) in previously-described mutant backgrounds either enriched for or lacking SUN-1 aggregates. mpk-1(ga111) mutants arrest in leptotene/zygotene at restrictive temperature, and their gonads are therefore highly enriched in SUN-1 aggregates at chromosome end attachments ( Figure 1C and Figure S1B). These mutants do not produce embryos (Lackner and Kim 1998;Leacock and Reinke 2006;Lee et al. 2007;Arur et al. 2009). In contrast, SUN-1 fails to form aggregates in the TZ in plk-2(vv44) gonads, and within the germ line SUN-1 is evenly distributed throughout the NE ( Figure 1C) (Labella et al. 2011). Here we used the fertilization-defective background spe-26(hc138) to prevent the presence of embryos in our extracts (Varkey et al. 1995). In both mutant backgrounds we could detect the coprecipitated endogenous SUN-1 protein as in the wild type ( Figure 1D).
These data suggest that SUN-1 forms oligomers in all populations of SUN-1 (both at the rim and the aggregates at chromosome ends) and that oligomerization is only abrogated in the absence of the cc motifs.
Deletion of cc domains leads to loss of SUN-1 from the NE, particularly in late prophase I To elucidate the function of SUN-1 oligomerization, we analyzed SUN-1 localization in the germ line, brood size, offspring viability, and the occurrence of male offspring (as an indicator of meiotic chromosome nondisjunction) of self-fertilized hermaphrodites expressing a single-copy integration of either untagged or GFP-tagged sun-1(D158-235) constructs in the absence of endogenous SUN-1 [sun-1(ok1282)]. All experiments were performed in both tagged and untagged mutant hermaphrodites, unless mentioned otherwise.
Deletion of SUN-1 cc motifs had a severe impact on offspring viability: the brood size of sun-1(D158-235)::gfp was 72% of the wild type and only 12% of the eggs hatched. Brood size and hatch rate were even more dramatically decreased in the sun-1(D158-235) untagged mutant: the brood size was 50% of wild type and only 0.7% of eggs gave rise to viable progeny ( Figure 2A). These results suggest that the propensity of GFP to dimerize mitigates the effect of deletion of the cc domains. Subsequent analyses were consistent with the GFP-tag mutant being a hypomorphic allele. Analysis of offspring viability in the presence of endogenous wild-type protein revealed that the mutant is fully recessive to the wild type ( Figure 2A). In contrast to hermaphrodites that carry two X chromosomes, males have only one X chromosome and arise by spontaneous meiotic nondisjunction of the X chromosome. An increased number of males (i.e., a high incidence of males, him phenotype) therefore indicates a failure in meiotic pairing or recombination (Hodgkin et al. 1979). Among the surviving progeny of mutant hermaphrodites the ratio of males was not increased, reflecting the undisturbed segregation of the X chromosomes. To elucidate the cause of embryonic death (be it through aneuploidy or developmental defects), we investigated SUN-1 localization in the germ line and the processes of prophase of meiosis I in more detail in both the tagged and untagged mutant.
Abrogation of SUN-1 oligomerization did not impair SUN-1 reorganization into aggregates at the onset of meiosis. Immunofluorescence staining showed SUN-1 aggregates in the TZ and even distribution of the protein from midpachytene onward, as in wild type ( Figure 2B). Against expectation, ZYG-12 was successfully recruited to the ONM when both cc domains were deleted ( Figure 2C). Consistent with this observation, in vivo time-lapse imaging of SUN-1 aggregates in worms expressing either sun-1(D158-235)::gfp or sun-1(WT)::gfp as the sole source of SUN-1 revealed that deletion of these motifs had no effect on chromosome movement. Figure 2D shows the displacement tracks of SUN-1 aggregate in sun-1(WT)::gfp II; sun-1(ok1282)V and sun-1 (D158-235)::gfp II; sun-1(ok1282)V over 3 min. The average velocity of SUN-1 aggregates was 52.73 nm/sec (n = 9 nuclei) in the wild type vs. 65.76 nm/sec (n = 14 nuclei) in the mutant. Movement of the aggregates in the mutant indicated the formation of a functional LINC complex and the successful transduction of cytoskeletal forces to chromosome ends.
When studying SUN-1 distribution we detected an overall decrease in SUN-1 signal intensity in the entire mutant germ line compared to wild type (Figure 3, A-D). This drop in signal intensity was most severe in the later stages of prophase I prior to oocyte cellularization; from this stage onwards, the SUN-1 signal was barely detectable (Figure 3, A, C, and D). We also noticed that the GFP tag slowed down the rate of SUN-1 signal loss slightly and thus represents a milder allele ( Figure 3C).
Concomitant with loss of SUN-1, we could not detect ZYG-12 in the ONM in late prophase ( Figure 4A). In the absence of a functional SUN-KASH bridge and the proper transduction of cytoskeletal forces to meiocytes, the arrangement and positioning of nuclei is disrupted in the C. elegans germ line (K. Zhou et al. 2009). In sun-1(D158-235) mutants, instead of a linear alignment of oocytes after cellularization we observed a perturbed distribution ( Figure 4B).
We also analyzed protein localization in the sun-1(D158-235)::gfp mutants in the presence of endogenous SUN-1. This showed that loss of SUN-1 from the NE in the absence of the cc domains was not rescued in the presence of the endogenous wild-type protein, emphasizing the lack of interaction between the two populations of the protein ( Figure S2).

SUN-1 cc motifs are dispensable for execution of early meiotic prophase I events
The formation of six bivalents (homologs connected by crossovers and cohesion) in oocytes results from successful pairing, synapsis, and recombination processes that depend on LINC complex function. As in wild type, six DAPI signals could be detected in the oocytes of both tagged and untagged mutants lacking the cc domains, thus indicating the proper orchestration of meiosis [DAPI signals in diakinesis: wild type = 5.8 6 0.37, n = 31; sun-1(D158-235) = 6.2 6 1.7, n = 42; and Figure 5A]. Apoptosis eliminates defective meiocytes (Gartner et al. 2008). However, even in the ced-3(n717) apoptosis-deficient background, the cc mutant did not display an increase in achiasmatic chromosomes ( Figure 5A). We also investigated homologous pairing, synapsis, and the dynamics of meiotic double-strand-break (DSB) repair in sun-1 (D158-235) mutants vs. wild type and could not detect any differences ( Figure 5, B-D). Staining against HIM-8 (a marker for the X chromosome subtelomeric region; Phillips et al. 2005Phillips et al. , 2009 revealed that pairing was similar in both wild type and mutant ( Figure 5B). Our previous studies showed a role for SUN-1 N-terminal modifications in SC polymerization kinetics (Woglar et al. 2013). The kinetics of HTP-3 and SYP-1 (chromosome axis and synapsis markers) loading was also similar in both wild type and mutants (Colaiacovo et al. 2003). SC polymerization started in the TZ, and chromosomes were fully synapsed by early pachytene ( Figure 5C). Also, analysis of the appearance and disappearance of RAD-51 over the entire length of the gonad from the distal mitotic zone until diplotene showed that the dynamics of DSB repair were unaltered in the absence of cc motifs ( Figure 5D; Alpi et al. 2003).
Timely progression through prophase I correlates with the accomplishment of meiotic tasks; such as pairing, synapsis, and generation of a crossover intermediate (of an unknown nature) (Rosu et al. 2013;Woglar et al. 2013). Previous studies have shown a link between phosphorylation of the N-terminal nucleoplasmic portion of SUN-1 and the time period in which those tasks are accomplished. In the presence of meiotic errors, for instance incomplete synapsis, the zone of phosphorylated SUN-1 is extended, indicating a progression delay (Woglar et al. 2013). Staining for phosphorylated SUN-1S8 revealed that the dynamics of phosphorylation and dephosphorylation of this residue are identical in the mutant and wild type ( Figure 6, A and B). Staining for two other markers of this region, DSB-2 (DNA-DSB factor; Figure 6, A and B; Rosu et al. 2013) and phosphorylated CHK-1 (pCHK-1; Figure 6C; Jaramillo-Lambert et al. 2010) also confirmed wildtype progression through meiosis in the absence of cc motifs.
Our results show that despite the reduced amount of SUN-1 in the NE, a functional LINC complex is formed and all early processes of meiotic prophase I were accomplished as in wild type.
Deep sequencing analysis of RNA extracted from the mutant and wild-type worms revealed that the expression of SUN-1 and all known meiotic genes, genes required for early embryonic development, or housekeeping was similar between the two genotypes ( Figure S3). The experiment was done in an embryo-deficient background spe-26(hc138) to exclude embryonic messenger RNAs (mRNAs) from the analysis. This analysis revealed 17 genes to be differently  SUN-1/GFP (green) costaining in the mitotic zone of worms expressing tagged and untagged wild-type and mutant SUN-1 (in the absence of endogenous wild-type SUN-1). Note the reduced SUN-1 signal intensity in the rim of mitotic nuclei in the mutants. (C) Late pachytene and diplotene stages of sun-1(wt)::gfp and mutant gonads. Note the pronounced loss of SUN-1 from the NE in the mutants. Bar, 10 mm. (D) SUN-1::GFP signal intensity in the nuclear rim of the mutant and wild-type gonads. **** denotes the significant decreased amount of SUN-1 in the NE throughout the gonad (two-tailed t-test, P , 0.0001 for each zone). For each zone 30 nuclei of 3 independent gonads were analyzed. Error bars represent SD. The experiment was performed in the absence of endogenous wild-type SUN-1. Dia, diakinesis; Dip, diplotene; EP, early pachytene; LP, late pachytene; MP, midpachytene; TZ, transition zone. SUN-1 Nuclear Envelope Retention expressed between wild type and mutant, either up-or downregulated (Table S2). Quantitative PCR, performed with RNA extracted from dissected gonads, did not reproduce different RNA amounts of these ORFs and sun-1 itself in the mutant (our unpublished results). These data show that the absence of cc regions leads to decreased amounts of SUN-1 in the nuclear rim despite normal expression levels of sun-1 itself but also other germ line expressed genes. This phenotype becomes most pronounced as meiocytes enter late pachytene/diplotene. However, lack of these domains does not abrogate the formation of the LINC complex and SUN-1 higher-order structures upon meiotic entry. The ability to form SUN-1 aggregates at chromosome end attachments (which support chromosome end mobilization) in the cc mutant strongly supports the notion that formation and function of the LINC complex is independent of SUN-1 oligomerization and prophase I events proceed as in the wild type. SUN-1 cc motifs are therefore required for efficient protein localization to the INM, especially at the later stages of prophase I, in particular during female oogenesis (see below). cc motifs are required for SUN-1 retention at the NE rather than for targeting The amount of SUN-1 localized to the NE increases strongly during prophase and peaks in the cellularized oocyte, suggesting that continuous incorporation of this protein takes place (Figure 3, A and D). Furthermore, photo-conversion experiments using SUN-1 tagged with the photo-convertible fluorophore Dendra2 (full green-to-red photo-conversion) (Gurskaya et al. 2006) confirmed our cytological observations and revealed that SUN-1 is continuously inserted into the NE in the wild-type germ line as nuclei progress through meiosis. De novo SUN-1 loading to the NE is most prominently seen from midpachytene onwards ( Figure 7A).
To find the reason for the lack of SUN-1 in late pachytene/ diplotene we tested if cc motifs were required for the targeting of the de novo synthesized SUN-1 from midpachytene onwards or whether they were required for efficient retention of SUN-1 in the NE. For this purpose we performed a photo-conversion experiment using a sun-1(D158-235) transgene tagged with the photo-convertible fluorophore Eos 3.2 (M. Zhang et al. 2012). We photo-activated SUN-1 in a few rows of nuclei at the midpachytene stage, i.e., the zone showing the most prominent de novo loading of SUN-1. Nuclei migrate along the germ line at 1 row/hr (Crittenden et al. 2006). Thus, worms were recovered after 4 hr and the distribution of SUN-1 was analyzed. If there were problems in protein targeting, we would be unable to detect newlyincorporated protein at the nuclear rim and the SUN-1 population detected in this zone would consist of only red photo-converted molecules. However, if targeting was normal, we would be able to detect a mixture of activated and nonactivated SUN-1 proteins ( Figure 7B). The photo-conversion experiments showed that newly-synthesized wild-type and  mutant SUN-1 protein was inserted next to the existing pool ( Figure 7C).
When analyzing SUN-1 localization, we noticed blobs of mutant SUN-1 protein in the cytoplasm. This was not observed for wild-type SUN-1 protein and we propose that this is the pool of protein lost from the NE ( Figure S4A). Western blot analysis confirmed that the mutant SUN-1 protein amount is reduced, which suggests that once the protein is lost from the NE it is subjected to degradation ( Figure S4B).
To address whether nuclear membrane anchorage of ccdeleted SUN-1 was less robust, we depleted the KASH partner ZYG-12 by RNAi. zyg-12 depletion was assessed in the zyg-12::gfp line ( Figure S5), GFP signal depletion goes in hand with defects in chromosome pairing monitored by HIM-8 (Phillips et al. 2005. We used the consequent pairing defect as a readout for efficient zyg-12 knockdown since immunostaining against ZYG-12 was challenging. 48 hr post treatment, HIM-8 paired signals were lost in the entire Figure 6 Early meiotic prophase I events progress with wild-type dynamics in the absence of cc motifs. (A) DAPI (blue), , and DSB-2 (red) staining in wild-type and mutant germ lines; with no differences. White bars mark the zone containing phosphorylated SUN-1 and the zone of DSB-2 localization. (B) Graph showing quantification of the zone positive for phospho-SUN-1S8 and DSB-2. The region positive for these markers was normalized to gonad length from meiotic onset until diplotene. There is no significant difference between wild type and mutant (two-tailed t-test, P . 0.581). Bars represent SD Psun-1(wt)::gfp II; sun-1(ok1282) V, n = 11; Psun-1(D158-235)::gfp II; sun-1(ok1282) V, n = 6; sun-1(wt), n = 6; Psun-1(D158-235) II; sun-1(ok1282) V, n = 6. (C) Costaining of GFP (to mark SUN-1) and pCHK-1. White bars mark the zone containing pCHK-1. Note the higher signal intensity and extension of the zone containing pCHK-1 until the end of pachytene in the absence of synapsis compared with other genotypes. Bar, 10 mm. germ line ( Figure 8A and Figure S5). In wild type, SUN-1 localization to the NE was not affected ( Figure 8A). In contrast, sun-1(D158-235) mutant germ lines were severely affected. Loss of ZYG-12 from the ONM led to a failure to retain SUN-1 in the INM ( Figure 8A) and gonads developed a severe sterility phenotype (reminiscent of the sun-1 deletion mutant phenotypes).
To summarize, our findings are consistent with a model in which cc-deleted SUN-1 is targeted successfully to the INM but fails to be retained efficiently and becomes highly sensitive to strain imposed only from one side of the LINC complex. Once the protein is lost from the NE, it is eventually subjected to degradation.
The severe loss of SUN-1 in late pachytene/diplotene is specific for oogenesis We wondered if the prominent loss of SUN-1 in the mutant background was linked to processes prior to oocyte cellularization. To address this we examined nuclei of corresponding stages (late pachytene/diplotene) in the germ line of L4 larvae. Due to the hermaphroditic nature of C. elegans, germ cells undergo spermatogenesis in the L4 stage. Here meiocytes do not arrest in diplotene, but instead proceed through the two divisions to produce sperm. During spermatogenesis SUN-1 is not prominently lost from the NE in late pachytene/ diplotene in sun-1(D158-235) mutants, despite an overall reduced amount of protein ( Figure 8B). We therefore hypothesize that the female germ line-specific loss of SUN-1 prior to diplotene could be linked to processes related to oocyte cellularization.
C. elegans embryos use the maternally-supplied protein pool of SUN-1 While addressing the cause of embryonic death in sun-1 (D158-235), we noticed that SUN-1 was undetectable in these mutants. Costaining of SUN-1 and SPD-5 (Hamill et al. 2002) revealed a centrosome detachment phenotype ( Figure 9A). In 14 out of 16 early embryos (1-4 cell stage) detached centrosomes were detected. This phenotype has also been observed in the sun-1(jf18) mutant that has a dysfunctional SUN-KASH bridge, in which ZYG-12 is not recruited to the ONM (Penkner et al. 2007). Centrosome detachment leads to severe aneuploidy and explains the low embryonic viability observed in the mutant.
Failure of SUN-1 retention in the NE, most prominently in late pachytene/diplotene, and lack of SUN-1 in the embryos of the mutants made us ask how SUN-1 is produced in the embryos. We wondered whether SUN-1 protein was deposited in the embryos rather than sun-1 mRNA, since maternal deposition of SUN-1 has been suggested previously (Fridkin et al. 2004). We therefore performed a photo-conversion experiment using wild-type SUN-1 tagged with Dendra2. We photo-activated the SUN-1 population in the 21 oocyte, allowed the animal to recover, and scored for fluorescence in the embryo after 90 min (i.e., 70-90 min postfertilization). If the embryos used the maternal pool of SUN-1 protein, we would expect photo-activated SUN-1 (in red) in the embryos, and if new protein was synthesized in the embryos, a mixture of activated and nonactivated (green) protein would be expected ( Figure 9B).
Our result showed that newly-translated SUN-1 is not detectable in early embryos and that embryos exclusively rely on the NE pool of SUN-1 protein maternally supplied to the oocyte ( Figure 9C). As a consequence of cc-motif deletion, SUN-1 is lost in oocytes and embryos, and ZYG-12 recruitment to the NE is abrogated. This causes defective centrosome positioning, leading to chromosome mis-segregation in the following mitotic divisions and embryonic lethality.

Discussion
This study provides further insight into the role of the cc motifs of SUN-domain proteins in LINC complex formation and function. We showed that SUN-1 forms oligomers in the Costaining of DAPI (blue) and SUN-1 (red) in L4 gonads (undergoing spermatogenesis) of wild type and Psun-1(D158-235) II; sun-1(ok1282) V. Note the absence of the severe SUN-1 loss in the L4 gonads in the later stages of prophase I in the mutant. Bar, 10 mm.
C. elegans germ line. Oligomerization depended on the cc motifs but was not a prerequisite for SUN-KASH interaction and the formation of higher-order SUN-1 assemblies. In the absence of SUN-1 oligomerization, a functional LINC complex was formed that could execute mitotic and meiotic functions. Nevertheless, SUN-1 cc motifs and oligomerization were important for SUN-1 retention in the NE. In the absence of self-interaction, SUN-1 was not retained at the INM; this was most pronounced prior to oocyte cellularization. Furthermore, the pool of SUN-1 protein present in early embryogenesis was exclusively derived from maternal protein located at the NE in the oocyte.
The SUN-1 cc motifs are required for self-interaction Previous studies showed that SUN proteins form oligomers for which they need an intact SUN domain and adjacent cc motifs (Padmakumar et al. 2005;Crisp et al. 2006;Q. Wang et al. 2006;Lu et al. 2008;Sosa et al. 2012;Z. Zhou et al. 2012;W. Wang et al. 2012). Recent in vitro studies and crystallization analysis of hSUN2 suggest that SUN proteins have a trimeric configuration that enables them to form a clover-like head, which is a prerequisite for building a functional LINC complex. hSUN2 trimers recruit KASH-domain protein trimers, thereby forming a 3:3 hexameric complex (Sosa et al. 2012;W. Wang et al. 2012;Z. Zhou et al. 2012). Our pull-down assays suggested that in the germ line, SUN-1 proteins always form oligomers irrespective of the meiotic stage. Although it has been suggested that the region extending from the transmembrane domain to the SUN domain could form a single continuous cc (Rothballer et al. 2013), our in vitro and in vivo assays confirmed the work of Minn et al. (2009) showing that SUN-1 oligomerization relies on the luminal-predicted cc domains. Surprisingly, and in contrast to published data (Sosa et al. 2012;W. Wang et al. 2012;Z. Zhou et al. 2012), deletion of both cc motifs and thus lack of SUN-1 self-interaction did not impair LINC complex formation and KASH partner recruitment. All early events sustained by SUN-1, such as chromosome movement, synapsis, or DSB repair took place in the deletion mutant with wild-type dynamics.
The SUN-1 cc domain is required for NE retention SUN-1 expression in the nuclear rim throughout the entire gonad is significantly weaker in the absence of both cc motifs. Due to a reduced amount of SUN-1 in the nuclear rim, the aggregates appear more prominent and the small aggregates become more obvious. Nevertheless, the amount of SUN-1 protein present at the NE is still sufficient for functionality during meiosis, even under conditions when force is exerted on the SUN-KASH bridge during chromosome movement. Perhaps, at this stage, coalescence of SUN-1 molecules at pairing center sites compensates for the inability for selfinteraction. The loss of SUN-1 becomes more prominent when nuclei enter diplotene.
Different studies have revealed that NE localization and membrane targeting of SUN proteins relies on multiple signals that contribute redundantly to their proper localization (Padmakumar et al. 2005;Turgay et al. 2010;Tapley et al. 2011). The precise mechanism by which Cel-SUN-1 is targeted to the INM is unknown. Our work supports a model in which the cc domains are required for Cel-SUN-1 retention Figure 9 The embryonic SUN-1 pool is maternally supplied. (A) Centrosome detachment in Psun-1 (D158-235) II; sun-1(ok1282) V mutants.  and GFP (green) staining mark the centrosome and SUN-1 in the wild type and mutant. Arrows indicate centrosomes in the vicinity of the NE in the wild type; arrowheads indicate detached centrosomes in the mutant. Note the decrease in SUN-1 signal intensity in mutant embryos. (B) Diagram of the photo-conversion experiment design. If SUN-1 is synthesized in the embryo, then a mixture of red photo-converted protein and green newly-synthesized SUN-1 is expected (resulting in yellow). If embryos exclusively use maternallysupplied SUN-1, a red photo-converted pool is expected. in the NE. The use of photo-convertible SUN-1 and the spatiotemporal arrangement of meiocytes allowed us to show that the cc-deleted protein was initially targeted to the NE, but was lost soon after. This was most prominent in late pachytene/ diplotene. At this stage of meiocyte development, chromosomes lose contact with the NE and condense to form individual bivalents. The karyosome (the state in which chromatin translocates to the center of the nucleus) is best studied in Drosophila; it was shown in this system that the process depends on NHK-1-dependent phosphorylation of the BAF protein, with BAF connecting chromatin to the NE (Lancaster et al. 2007). We therefore propose that chromosome detachment from the NE at this stage might contribute to SUN-1 destabilization at the membrane, because SUN-1 oligomerization might support robust NE anchorage to counteract the mechanical strain exerted on the LINC complex from the cytoplasm. SUN proteins are targeted to the INM by a lateral diffusion mechanism (Ungricht et al. 2015). Once the protein is targeted, it will be anchored in the INM through binding to the nuclear lamina or other nucleoplasmic proteins on one side and the KASH protein on the other side (Rothballer et al. 2013;Link et al. 2015). ZYG-12 knockdown via RNAi treatment in the cc mutant led to severe loss of SUN-1, whereas in the wild type SUN-1 localization was not affected. We hypothesize that lack of ZYG-12 and consequently the cytoplasmic anchor leads to dramatic loss of SUN-1 in the mutant, whereas in the wild type SUN-1 oligomerization still supports the retention. During the process of cellularization in diplotene, the cellular and nuclear volumes increase. In wild type, SUN-1 oligomers can resist unbalanced cytoplasmic forces originating from changes in the cytoskeletal architecture when oocytes cellularize; however, disruption of the oligomerization leads to loss of SUN-1 from the NE. Upon loss of NE localization the protein is eventually degraded. Consistently, in the same stage in male gametogenesis (late pachytene/diplotene) SUN-1 is not lost from the NE.
The cc-deleted strains allowed us to ask whether gene expression was altered upon loss of protein, as had been observed in SUN-1 mutant mice (Chi et al. 2009). Premature loss of SUN-1 did not alter gene expression, in particular sun-1 expression itself.
Embryogenesis relies on the maternal pool of SUN-1 C. elegans embryos use the maternal supply of SUN-1 protein, thus emphasizing the importance of de novo SUN-1 incorporation into the NE during prophase I, especially from midpachytene onwards. Current models suggest that NE proteins are retracted to the ER during mitosis and that this pool is reused for NE reassembly after mitosis (Wandke and Kutay 2013). The failure of SUN-1 NE retention leads to a lack of SUN-1 in both the oocyte and the developing embryo. Therefore, SUN-1 cc-mutant embryos display defects in centrosome attachment due to the absence of a functional SUN-KASH bridge (Fridkin et al. 2004;Penkner et al. 2007;Meyerzon et al. 2009).
Altogether we show that in C. elegans, formation of a functional SUN-KASH complex in the germ line does not depend on an oligomeric conformation of SUN-1. In the absence of SUN-1 self-interaction, functional LINC complexes and SUN-1 higher-order structures are formed at meiotic entry. Nevertheless, cc domains are required for SUN-1 retention in nuclear membranes. This is of the utmost importance for the development of healthy zygotes because the maternal pool of SUN-1 that is necessary for embryonic development is provided through the nuclear membrane of the oocyte.