Inner nuclear membrane protein LEM-2 is required for correct nuclear separation and morphology in C. elegans

The nuclear envelope is an essential structure in eukaryotic cells. It separates nuclear and cytoplasmic compartments, enabling regulated access of transcription factors to the genome, as well as processing and nuclear export of RNA (Mekhail and Moazed, 2010). The nuclear envelope is also involved in the organization of the nucleus by providing an architectural scaffold for chromatin, and interaction surfaces for transcription factors, signaling molecules and chromatin modifiers. The nuclear envelope is composed of outer nuclear and inner nuclear membranes and, in metazoans, the nuclear lamina, which is a thin meshwork formed by lamin proteins. The nuclear envelope also contains large nuclear pore complexes that constitute transport channels for macromolecules across the nuclear envelope.

A subset of inner nuclear membrane proteins is characterized by a 40–45 amino-acid-long LAP2–emerin–MAN1 (LEM domain), which binds to the chromatin-associated protein barrier-to-autointegration factor 1 (BAF-1 in C. elegans; BANF1 in mammals) (Brachner and Foisner, 2011). The human genome contains seven LEM genes (ANKLE1, ANKLE2, LEMD1, LEMD2, LEMD3, EMD, TMPO), five of which encode inner nuclear membrane proteins. Of those, emerin (EMD) has attracted particular attention because of its causative link to Emery-Dreifuss muscular dystrophy (EDMD) (Bione et al., 1994; Koch and Holaska, 2014). EDMD represented the first example of a class of rare diseases known as laminopathies, which also include neuropathies, pathologies that affect adipose tissues and premature-aging syndromes (Burke and Stewart, 2014). How mutations in widely expressed genes induce malfunction in specific tissues is not fully understood, but plausible explanations include altered cell signaling through nuclear-envelope-associated transcription factors, changes in cell-type-specific chromatin organization and different susceptibility to mechanical stress on the nuclear envelope.

In contrast to the severe symptoms of EDMD patients, mouse and Caenorhabditis elegans mutants for emerin show very mild phenotypes (Barkan et al., 2012; Brachner and Foisner, 2011; Melcon et al., 2006; Ozawa et al., 2006). The C. elegans genome encodes four LEM proteins, i.e. EMR-1, LEM-2, LEM-3 and LEM-4 (EMD, LEMD2, ANKLE1 and ANKLE2 in humans), although LEM-4 has no recognizable LEM domain (Asencio et al., 2012). A lem-2-null mutation in C. elegans causes mild contraction defects in smooth muscles (Barkan et al., 2012). However, simultaneous depletion of EMR-1 and LEM-2 causes embryonic lethality and chromosome segregation defects in C. elegans embryos (Barkan et al., 2012; Liu et al., 2003). This suggests that LEM proteins have overlapping functions, which is supported by the observation that overexpression of LEMD2 can compensate for knockdown of emerin during in vitro differentiation of myoblasts (Huber et al., 2009).

Post-mitotic recruitment of LAP2 (officially known as TMPO), EMD/EMR-1 and LEMD2/LEM-2 depends on interaction with BANF1 and BAF-1, respectively, in HeLa cells and C. elegans embryos (Gorjánácz et al., 2007; Haraguchi et al., 2001). Moreover, LEM proteins interact directly with lamins and depletion of the latter interferes with localization of EMD/EMR-1 and LEMD2/LEM-2 to the nuclear envelope (Brachner et al., 2005; Gruenbaum et al., 2002; Liu et al., 2003; Sullivan et al., 1999). LEM proteins are also important for communication between the nucleus and the cytoskeleton, controlling nuclear position and mechanotransduction. A main structure in these processes is the linker of nucleoskeleton and cytoskeleton (LINC) complex, which consists of KASH (Klarsicht, ANC-1, SYNE/nesprin homology) and SUN (Sad1, UNC-84) domain proteins in the outer and inner nuclear membrane, respectively (Starr and Fridolfsson, 2010). KASH proteins interact with F-actin and microtubules in the cytoplasm, and with SUN proteins in the nuclear envelope lumen, whereas SUN proteins connect to the nuclear lamina lining the inner nuclear membrane. In addition, nesprins bind directly to emerin and lamins (Mislow et al., 2002; Zhang et al., 2005).

EMR-1 and LEM-2 are ubiquitously expressed in C. elegans (Gruenbaum et al., 2002; Liu et al., 2003), but their relative distribution has not been analyzed. Consistent with clinical manifestations of laminopathies, several nuclear envelope proteins have been ascribed tissue-specific functions (Gomez-Cavazos and Hetzer, 2012). To explore the possibility that EMR-1 and LEM-2 are enriched in specific tissues, we generated strains that carry single-copy transgenes integrated into the genome at known intergenic loci (Frøkjær-Jensen et al., 2012). Sequences upstream and downstream of the emr-1- and lem-2-coding regions were included to most closely mimic endogenous expression. To rule out any bias introduced by fluorescent tags or the integration sites, we analyzed different combinations, i.e. lem-2::gfp chrII, emr-1::mCherry chrIV and emr-1::gfp chrII, lem-2::mCherry chrIV. Analyzing both configurations also enabled us to convert ratios of GFP and mCherry fluorescence into EMR-1 and LEM-2 protein ratios, respectively (supplementary material Fig. S1). Importantly, the transgenes completely suppressed the embryonic lethality of doubly depleted embryos, arguing that they encode biologically active fusion proteins (Fig. 1A,B). Consistent with their redundant function in chromosome segregation and nuclear envelope assembly (supplementary material Fig. S2) (Liu et al., 2003) both proteins were found in all tissues (Fig. 1C; supplementary material Fig. S3). Quantifying nuclear envelope signal intensities revealed that EMR-1 is generally more abundant than LEM-2, and is particularly enriched in contractile cells (Fig. 1D). We conclude that the relative abundance of EMR-1 and LEM-2 differs up to 27-fold between tissues, which suggest that cell types have different requirements for these two LEM proteins. The high levels of EMR-1 in muscle cells and neurons are consistent with our observation that EMR-1 is required for proper neuromuscular junction activity (González-Aguilera et al., 2014).

Although the dynamics of EMR-1 and LEM-2 during mitosis have been studied (Gorjánácz et al., 2007; Lee et al., 2000), our transgenic strains provided a unique opportunity to compare the two proteins simultaneously in living embryos. Surprisingly, LEM-2 was recruited much earlier to reforming nuclei than EMR-1. Accumulation of LEM-2 was detected 55–70 sec after anaphase onset, initially at the apical surfaces of chromatin (Fig. 2A-C) from where LEM-2 spreads to the lateral surfaces facing the spindle microtubules (Fig. 2A; 85–110 sec). This recruitment pattern coincides with that of several nucleoporins, such as NPP-19, NPP-8 and NPP-4 (in mammals NUP35, NUP155 and NUPL1, respectively) (Fig. 2D), (Franz et al., 2005), suggesting that assembly of the nuclear pore complex is initiated at association sites of LEM-2-containing membrane. The initial apical recruitment of LEM-2 is intriguing, considering that BAF-1, its best-described interaction partner on chromatin, accumulates first on the lateral surfaces (Gorjánácz et al., 2007; Haraguchi et al., 2008; Haraguchi et al., 2001). Following the initial accumulation of LEM-2 its concentration in the nuclear envelope rapidly decreased due to expansion of the nuclear envelope, until it reached a steady level (Fig. 2A,B; 110–470 sec). In contrast, the concentration of EMR-1 in the nuclear envelope increased gradually from telophase and throughout interphase (Fig. 2A,B). Immunofluorescence analysis of endogenous proteins similarly indicated that LEM-2 is recruited faster than EMR-1 (Fig. 2E). This suggests that EMR-1 and LEM-2 are recruited from different membrane compartments during nuclear envelope assembly.

Because EMR-1 and LEM-2 behave differently in the early embryo, we hypothesized that they make distinct contributions to nuclear structure. Measuring the nuclear contour ratio (4πA/perimeter²) of fully grown interphase nuclei in 4-cell stage embryos of more than ten animals per strain revealed a significant decrease in lem-2-null mutants (Fig. 3A). Moreover, nuclear envelope invaginations were more frequent in the absence of LEM-2 (Fig. 3A). This is consistent with observations from human cancer cell lines, where knockdown of LEMD2 was reported to alter nuclear morphology (Ulbert et al., 2006). In contrast, the nuclei of emr-1-null mutant embryos were slightly more spherical than control nuclei. As an independent measure for the involvement in control of nuclear structure we tested for genetic interactions with a partial loss-of-function npp-19 allele that, at elevated temperatures, causes nuclear assembly defects (Ródenas et al., 2009). At the semi-permissive temperature (20°C), 45% of npp-19 embryos died before hatching, whereas 7% arrested during larval development and 48% reached adulthood (Fig. 3B). Crossing the npp-19 allele into the emr-1-null background did not affect development, whereas significantly more npp-19 lem-2 double mutants arrested during embryogenesis (62%) or larval development (18%) and only 20% reached adulthood. This suggests that LEM-2 is involved in the regulation of nuclear structure, independently of EMR-1.

We next hypothesized that LEM-2 is required for stable association of other proteins of the inner nuclear membrane, such as EMR-1. We first performed FRAP experiments in 4-cell stage embryos to quantify EMR-1 mobility within the inner nuclear membrane. This revealed a recovery half-time of 37±6 s (mean±s.e.m.) in control embryos and 35±7 s in lem-2 embryos, indicating that EMR-1 residence time is independent of LEM-2 (Fig. 3C). We next measured the turnover of EMR-1 protein by using adult nematodes expressing EMR-1 fused to the photoconvertible fluorescent protein Dendra2. We used whole-animal illumination conditions under which ∼50% of EMR-1::Dendra2 was converted to red fluorescent protein (Fig. 3D; t = 0 d) and acquired images of post-mitotic cells the following seven days. Degradation and new synthesis of EMR-1::Dendra2 cause disappearance of red fluorescence and increase of green fluorescence, respectively. We found that the red:green ratio in hypodermal and muscle nuclei (grouped) and neurons of control adults decreased with a half-life of 60 h and 70 h, respectively; in lem-2 mutants, these values were reduced to 41 h (32% decrease) and 51 h (27% decrease), respectively. We conclude that, in several tissue types, EMR-1 turns over ∼30% faster in the absence of LEM-2. Whether this effect is specific to EMR-1 or reflects a general increase in protein turnover in lem-2 mutants remains to be investigated.

During our observations of 4-cell stage embryos, we realized that separation of daughter nuclei in the ABa and ABp cell pair was delayed. To quantify this defect we captured videos of embryos expressing EMR-1::mCherry from before onset of division of the AB cell. In control embryos, rapid separation was observed from ∼300 s after onset of anaphase, whereas separation did not initiate until ∼360 s in lem-2 mutants (Fig. 4A,B). Once separation initiated, the speed was comparable (∼2.4 µm/min in control and ∼2.7 µm/min in lem-2 embryos).

Nuclear separation depends on microtubule-associated motor proteins and centrosome movement. We speculated that LEM-2 is involved in centrosome attachment to the nuclear envelope, as reported for EMD in human cell lines (Salpingidou et al., 2007), and explored this in embryos expressing fluorescent tubulin and histones (Fig. 4C). We found that chromosome segregation in lem-2 embryos was normal during anaphase and telophase (28– ∼180 s), whereas the nuclear separation phenotype was evident from ∼280 s after onset of anaphase (Fig. 4C,D). Surprisingly, whereas a single microtubule-organizing center (MTOC) was found in close association with nuclei during nuclear separation in control embryos (n = 10), two MTOCs were observed in all lem-2 embryos (n = 9); one proximal that remained close to the nucleus and one distal that separated away (Fig. 4C; enlarged regions at 301 s). The nuclear-separation and MTOC-splitting phenotypes were also present during division of the P1 cells, suggesting that LEM-2 depletion affects several cell types (Fig. 4C; enlarged regions at 497 s; Fig. 4E). Moreover, cell cycle timing was delayed in lem-2 embryos: in control embryos P1 divided 111±18 s after AB, whereas P1 division occurred 148±12 s after AB division in the absence of LEM-2 (Fig. 4C; 126 s; P = 0.009). Depletion of LEM-2 by RNA interference (RNAi) similarly inhibited nuclear separation and cell cycle progression (supplementary material Fig. S4) as well as induced MTOC separation (Fig. 4F), demonstrating that the phenotypes are reproducible in different genetic backgrounds.

We conclude that LEM-2 is required for timing of entry to mitosis, nuclear positioning and efficient MTOC anchoring to the nuclear envelope as well as the correct nuclear morphology. LEMD2 has not been linked to human pathologies. However, it is intriguing that depletion of LINC components or interacting proteins, such as lamins, emerin or nesprins, causes similar MTOC-anchoring defects. Specific mutations in any of these genes lead to EDMD but in almost half of EDMD patients the disease-causing mutations are still unknown and many, presumably, map to other genes (Burke and Stewart, 2014). Analysis of the LEMD2 locus in these cases may identify another piece in the puzzle to understand human laminopathies.

Plasmids to express EMR-1 or LEM-2 contain 1568 bp or 1109 bp upstream of the start codon and 301 bp or 618 bp downstream of the stop codon, respectively. Fluorescent tags were inserted immediately before the stop codon. Plasmids are listed in supplementary material Table S1.

Transgenic strains were generated by Mos1-mediated single-copy insertion (Frøkjær-Jensen et al., 2008). Strains were cultured at 20°C using standard C. elegans methods (Stiernagle, 2006), and are listed in supplementary material Table S2.

RNA interference (RNAi) experiments were performed by feeding at 20°C for ∼36 h (Askjaer et al., 2014). The RNAi bacteria expressed double-stranded RNA corresponding to either full-length emr-1 cDNA, positions 1–682 of lem-2 gDNA or, as negative control, 185 bp of unrelated sequence from the pPD129.36 empty vector.

Samples were mounted between a coverslip and a 2% agarose pad; embryos were mounted in 3 µL M9 buffer, whereas larvae and adults were mounted in 3 µL 10 mM levamisole (Sigma-Aldrich, St. Louis, MO; cat. no. L9756). Epifluorescence and DIC images were recorded at 22–24°C with a Nikon A1R confocal microscope through a Plan Apo VC 60×/1.4 objective (Nikon, Tokyo, Japan).

Samples were processed as described (Ródenas et al., 2012). Antibodies are listed in supplementary material Table S3. Confocal images were obtained with a Leica SPE microscope equipped with an ACS APO 63×/1.3 objective (Leica, Wetzlar, Germany).

We are grateful to Cristina Ayuso García for excellent technical assistance and to Anita G. Fernandez for insightful comments on the manuscript.