FLN-2 functions in parallel to linker of nucleoskeleton and cytoskeleton complexes and CDC-42/actin pathways during P-cell nuclear migration through constricted spaces in Caenorhabditis elegans

Abstract

Nuclear migration through narrow constrictions is important for development, metastasis, and proinflammatory responses. Studies performed in tissue culture cells have implicated linker of nucleoskeleton and cytoskeleton (LINC) complexes, microtubule motors, the actin cytoskeleton, and nuclear envelope repair machinery as important mediators of nuclear movements through constricted spaces. However, little is understood about how these mechanisms operate to move nuclei in vivo. In Caenorhabditis elegans larvae, six pairs of hypodermal P cells migrate from lateral to ventral positions through a constricted space between the body wall muscles and the cuticle. P-cell nuclear migration is mediated in part by LINC complexes using a microtubule-based pathway and by an independent CDC-42/actin-based pathway. However, when both LINC complex and actin-based pathways are knocked out, many nuclei still migrate, suggesting the existence of additional pathways. Here, we show that FLN-2 functions in a third pathway to mediate P-cell nuclear migration. The predicted N-terminal actin-binding domain in FLN-2 that is found in canonical filamins is dispensable for FLN-2 function; this and structural predictions suggest that FLN-2 does not function as a filamin. The immunoglobulin-like repeats 4–8 of FLN-2 were necessary for P-cell nuclear migration. Furthermore, in the absence of the LINC complex component unc-84, fln-2 mutants had an increase in P-cell nuclear rupture. We conclude that FLN-2 functions to maintain the integrity of the nuclear envelope in parallel with the LINC complex and CDC-42/actin-based pathways to move P-cell nuclei through constricted spaces.

Introduction

The positioning of nuclei is important to many developmental processes, including pronuclear migration, muscle development, and neurodevelopment (Bone and Starr 2016). However, moving nuclei to the proper subcellular location, especially if it needs to traverse a constricted space, is poorly understood. Nuclei are the largest and stiffest organelle, making nuclear deformation and migration the rate-limiting step for cell migrations through constricted spaces significantly smaller than the resting diameter of nuclei (Davidson et al. 2014). For example, during hematopoiesis, white blood cell nuclei migrate out of the bone marrow, while the stiffer nuclei of megakaryocytes cannot, leading to the formation of anucleated platelets (Junt et al. 2007; Shin et al. 2013; McGregor et al. 2016; Salvermoser et al. 2018). Metastasis of some cancer cells also relies on the deformation of their nuclei so that the cells can squeeze through small openings in the extracellular matrix (Fu et al. 2012; Seyfried and Huysentruyt 2012; Gensbittel et al. 2021). Thus, understanding how nuclei normally traverse constricted spaces could help our understanding of metastasis. Tissue-cultured cells can be studied migrating through fabricated constrictions as small as 5% the resting diameter of their nuclei (Davidson et al. 2014; Thiam et al. 2016; Renkawitz et al. 2019). Such experiments generated novel mechanistic insights into this process, including how the nuclear envelope and DNA are repaired after damage caused by severe morphological changes (Denais et al. 2016; Raab et al. 2016; Xia et al. 2018). These approaches mostly rely on migrations through fabricated matrices or physical manipulations of cells in tissue culture, and very little is understood about how nuclei traverse confined spaces in vivo.

Caenorhabditis elegans hypodermal P cells are an excellent in vivo model for studying nuclear migration through constricted spaces as a normal part of development (Chang et al. 2013; Bone et al. 2016; Starr 2019). Six P cells are born on each lateral side of the embryo and extend protrusions to form the hypodermis that encloses the ventral surface of the embryo (Williams-Masson et al. 1997). P-cell nuclei remain at their original lateral positions throughout embryonic morphogenesis and into the early L1 larval stage when P cells cover the ventral surface of larvae (Altun and Hall 2024). Throughout early L1, the hypodermal hyp7 syncytium intercalates and replaces P cells to cover the ventral surface. During this process, P cells narrow along their anterior–posterior axis and at mid-L1, P-cell nuclei migrate from their lateral positions to the ventral cord (Bone et al. 2016). P-cell nuclear migration occurs through a constricted space between the body wall muscles and the cuticle where the hypodermis narrows to ∼5% of the resting diameter of a nucleus in early L1. This constriction is small so that the hemidesmosome-like fibrous organelles can form in the hypodermis to mechanically connect body wall muscles to the cuticle (Francis and Waterston 1991; Cox and Hardin 2004; Bone et al. 2016). Although the fibrous organelles are removed just prior to P-cell nuclear migration, P-cell nuclei still undergo severe morphological changes as they squeeze through this narrow opening (Bone et al. 2016). Soon after migrating to the ventral cord, P cells divide and their lineages give rise to the vulval cells and 12 γ-aminobutyric acid (GABA) neurons in the ventral cord (Sulston and Horvitz 1977). Consequently, if P-cell nuclear migration fails, P cells die and the animal develops Egl (egg-laying deficient) and Unc (uncoordinated) phenotypes due to the absence of vulval cells and neurons, respectively (Sulston and Horvitz 1981; Malone et al. 1999; Starr et al. 2001).

Multiple molecular pathways contribute to ensure P-cell nuclear migration. The LINC complex pathway consists of UNC-84 and UNC-83, which function together to span the nuclear envelope (Starr and Fridolfsson 2010; Starr 2019). The SUN (Sad1 and UN C-84) protein UNC-84 is an integral protein of the inner nuclear membrane with a N-terminal nucleoplasmic domain that interacts with lamins (Bone et al. 2014) and a C-terminal luminal domain that extends across the perinuclear space with its conserved SUN domain adjacent to the inner leaflet of the outer nuclear membrane (Cain et al. 2014; Jahed et al. 2019). The klarsicht, ANC-1, and Syne homology (KASH) protein UNC-83 is found in the outer nuclear membrane with a short conserved C-terminal KASH peptide of 17 residues that extend into the perinuclear space and a large N-terminal cytoplasmic domain that interacts with microtubule motors (Starr et al. 2001; Meyerzon et al. 2009; Fridolfsson et al. 2010; Fridolfsson and Starr 2010). The KASH peptide of UNC-83 and the SUN domain of UNC-84 directly interact within the perinuclear space to form a physical bridge across the nuclear envelope, connecting the cytoskeleton to the nuclear lamina (McGee et al. 2006; Cain et al. 2018). During P-cell nuclear migration, the LINC complex works primarily by recruiting the dynein/dynactin complex to the nuclear envelope to move nuclei toward the minus ends of microtubules (Bone et al. 2016; Ho et al. 2018).

Null mutations in unc-84 or unc-83 cause a temperature-sensitive phenotype (Malone et al. 1999; Starr et al. 2001). At 15°C, most P-cell nuclei migrate normally, but at 25°C, about 50% of P-cell nuclei fail to migrate (Sulston and Horvitz 1981; Malone et al. 1999; Starr et al. 2001; Chang et al. 2013). This suggests that at least one additional pathway acts in parallel with the LINC complex pathway during P-cell nuclear migration. We screened for additional players in P-cell nuclear migration by mutagenizing and selecting for strains where P-cell nuclear migration failed in an unc-84 background at the normally permissive temperature of 15°C (Chang et al. 2013). This forward genetic approach identified toca-1, which encodes a F-bar protein that localizes to curved membranes and is thought to promote the nucleation of actin filaments through the small GTPase CDC-42 and the Arp2/3 complex (Ho et al. 2004; Giuliani et al. 2009; Chang et al. 2013) and cgef-1, a predicted GEF (guanine nucleotide exchange factor) for CDC-42 (Chan and Nance 2013; Ho et al. 2023). CDC-42, the Arp2/3 complex member ARX-3, and nonmuscle myosin NMY-2 also function in the actin-based pathway (Ho et al. 2023).

Here, we report that FLN-2 enhances the P-cell nuclear migration defect of unc-84. FLN-2 is related to filamins, which usually crosslink or bundle actin filaments (DeMaso et al. 2011; Kovacevic et al. 2013). However, we showed that the N-terminal actin-binding domain of FLN-2 was not necessary for its function during P-cell nuclear migration through constricted spaces and that the actin network is not severely disrupted in P cells of fln-2 mutant animals. Instead, we identified a stretch of immunoglobulin (Ig) repeats in FLN-2 that are necessary for P-cell nuclear migration and found an increase in nuclear rupture in fln-2 mutant P cells. Furthermore, cgef-1 and ­fln-2 had a synergistic nuclear migration defect, suggesting that FLN-2 is a member of a third mechanism that is necessary to move P-cell nuclei in parallel to the LINC complex and actin-based pathways.

Materials and methods

C. elegans genetics, strains, and gene editing

Animals were maintained on nematode growth media plates spotted with E. coli strain OP50 and maintained at 15°C unless otherwise noted (Brenner 1974; Stiernagle 2006). The strains used in this study are listed in Table 1. FX4687 was a gift from Shohei Mitani (Tokyo Women's Medical University) (Consortium 2012). Some strains were obtained from the Caenorhabditis Genome Center, which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440).

Table 1.

List of worm strains used in this study.

Strain	Genotype	Reference
N2	Bristol, wild-type	Brenner (1974)
FX4687	fln-2(tm4687) X	Consortium (2012)
UD87	unc-84(n369) oxIs12[unc-47::gfp, dpy-20(+)]; ycEx60[odr-1::rfp, WRM0617cH07] X	Chang et al. (2013)
UD363	fln-2(tm4687) unc-84(n369); ycEx202[odr-1::gfp; phlh-3::nls::tdTomato; phlh-3::vab-10-abd::mVenus] X	This study
CZ13799	juIs76[unc-25::gfp] II	Huang et al. (2002)
UD368	unc-84(n369) fln-2(tm4687) X; juIs76[unc-25::gfp] II	This study
UD370	fln-2(tm4687) X; juIs76[unc-25::gfp] II	This study
UD381	ycIs11[odr-1::gfp; phlh-3::vab-10-abd::mVenus; phlh-3::nls::tdTomato]	Chang et al. (2013)
UD407	unc-84(n369) X; ycIs11	Bone et al. (2016)
UD593	ycSi1[phlh-3::lifeact::mKate2 pCFJ150[unc-119(+)]] II	This study
UD614	ycEx283[10 ng/ul phlh-3::nls::gfp::lacZ, 90 ng/ul odr-1::gfp]; ycSi1	This study
UD727	unc-84(n369) fln-2(tm4687) X; oxIs12[unc-47::gfp]; ycEx261[(5 ng/ul WRM0625B05), (95 ng/ul pGH8 (prab-3::mCherry))]	This study
UD749	unc-84(n369) fln-2(yc115[ΔIg9-15]) oxIs12[unc-47::gfp, dpy-20(+)] X	This study
UD758	unc-84(n369) fln-2(yc114[ΔIg4-9]) oxIs12[unc-47::gfp, dpy-20(+)] X	This study
UD771	fln-2(yc117[spGFP11::fln-2c]) X	This study
UD774	fln-2(yc117[spGFP11::fln-2c ycEx265[phlh-3::spGFP1-10, odr-1::gfp])	This study
UD801	unc-84(n369); fln-2(yc116[ΔIg15-23]) oxIs12[unc-47::gfp] X; ycEx60[odr-1::rfp, unc-84(+)]	This study
UD847	ycIs11[odr-1::gfp; phlh-3::vab-10-abd::mVenus; phlh-3::nls::tdTomato]; fln-2(tm4687)	This study
UD923	cgef-1(gk261), unc-84(n369), oxIs12 X; ycEx60	Ho et al. (2023)

Some strains were made by injecting plasmid DNA to make extrachromosomal arrays (Evans 2006). The plasmid encoding the vab-10-abd::mVenus (Kim et al. 2011) was cloned under control of the P-cell-specific hlh-3 promoter (Doonan et al. 2008; Chang et al. 2013). DNA for phlh-3::nls::tdTomato was previously described (Chang et al. 2013). pSL832 [phlh-3::nls::gfp lacZ] was created by amplifying nls::gfp::LacZ from pPD96.04 (a gift from Andrew Fire, Addgene plasmid # 1502) and inserting this amplicon into a pSL780 backbone (phlh-3::lifeact::mKate2), replacing the lifeact::mKate2 with our nls::gfp::lacZ. This plasmid was microinjected into UD593 animals to create UD614 (ycEx283[10 ng/ul phlh3::nls::gfp::lacZ, 90ng/ul odr-1::gfp] ycSi1[phlh3::lifeact::mKate2]II).

The fln-2::spGFP11 strain and the fln-2 domain deletion alleles were produced via CRISPR/Cas-9 gene editing using the crRNA guides and repair templates found in Table 2. We used a dpy-10 co-CRISPR approach, as previously described (Arribere et al. 2014; Paix et al. 2017; Hao et al. 2021). Two crRNA guides and one single-stranded oligodeoxynucleotide (ssODN) repair template were used to generate each of the fln-2 deletion alleles (Table 2). The split-GFP approach (Cabantous et al. 2005) was used, and the 11th beta-strand of GFP (spGFP11) was inserted to the 5′ end of the predicted open reading frame of fln-2c, while the rest of GFP (spGFP1–10) was expressed from an extrachromosomal array under the control of the hlh-3 promoter (Doonan et al. 2008; Chang et al. 2013). CRISPR/Cas9 injection mixes were prepared as follows: 0.2 μl of dpy-10 crRNA (8 μg/μl) (Dharmacon/Horizon Discovery), a total of 1.5 μl of the target gene crRNA (4 μg/μl) (0.75 μl of each guide if there are two guides) (Dharmacon/Horizon Discovery), 1.9 μl of tracrRNA (4 μg/μl) (Dharmacon/Horizon Discovery), 0.5 μl of dpy-10 ssODN, 1.1 μl of the target gene ssODN (1 μg/μl), 4 μl of Cas9 (10 μg/μl) (UC Berkeley QB3), and 0.8 μl of H₂O. The mix was injected into the gonads of young adult hermaphrodite animals (Evans 2006). All ssODN repair templates were manufactured by Integrated DNA Technologies.

Table 2.

Sequences of CRISPR crRNA guides and ssODN repair templates used in this study.

New allele	crRNA	ssODN repair template
fln-2(yc114[ΔIg4-9])	ATGGAGCATCGGGTACACGT	ACCGACGTGCACCCGATGTTCCATCTGCTTGCAATTACGACCCAGCCAAAATTAAAGTCGGAGCAATTCC
	GATCCAGCTAAGATTAAAGT
fln-2(yc115[ΔIg9-15])	TGTCCAACCCTGGAAAGCGA	CCGCCAAAACTACCGTTTCCAAAATTCCATCACTCTTAGATCCATCAGCTGTGAGAGTTATTGGACTGAAAAATGCTCC
	AGCGTGGATTATACTAATAG
fln-2(yc116[ΔIg15-23])	CTCAAGATACCTTCGCGGCG	AGCGCAAAATGCAAACAGTGATTGACCCACTTAATGTCGACACCGAGAAAGAACTCCCAAGACACGTTCG
	GATACACGTACTCCGCCGTC
fln-2(yc117[spGFP11])	AGAGCCAGGCGAGCACAGTG	TTTCTTGATTTGCAGGCGAGCACAGCGTCGATGTCATGAGAGATCACATGGTTCTTCATGAATATGTAAATGCAGCTGGAATTACAGGAGGTTCTGGCGGAATGCTAGCAGACCAACGTGTACCCGATGCTCCATT

The lifeact::mKate2 in P cell strain (UD593) was made using mos1-mediated single-copy insertion (mosSCI) (Frøkjær-Jensen et al. 2014). DNA encoding the hlh-3 promoter (Doonan et al. 2008), lifeact (Riedl et al. 2008), and mKate2 (Shemiakina et al. 2012) were from pSL780 (Bone et al. 2016) and cloned into Gateway entry plasmids (pDONR 221 P4-P1R, pDONR 221, and pDONR 221 P2r-P39 (Thermo Fisher Scientific)). Subsequently, they were cloned using the Gateway LR reaction into the pCFJ150 targeting vector (a gift from Erik Jorgensen; Addgene plasmid # 19329) (Frøkjær-Jensen et al. 2008) and named pSL831. This plasmid was injected into EG6699 (ttTi5605) animals (Frøkjær-Jensen et al. 2012) along with pCFJ601 (Peft-3 Mos1 transposase; 50 ng/μl), pMA122 (peel-1 negative selection; 10 ng/μl), pGH8 (Prab-3::mCherry; 10 ng/μl), pCFJ90 (Pmyo-3::twk-18(cn110); 2.5 ng/μl), and pCFJ104 (Pmyo-3::mCherry) (5 ng/μl). Animals harboring a single copy of integrated pSL381 were selected as unc-119(+) and lacked the red extrachromosomal array markers (Frøkjær-Jensen et al. 2008, 2012).

C. elegans P-cell nuclear migration assays, synchronization, and microscopy

Our assay for counting failed P-cell nuclear migration events using unc-47::gfp to mark GABA neurons in the ventral cord has been previously described (Chang et al. 2013; Fridolfsson et al. 2018). We scored potential nuclear rupture events as follows. In cells scored as normal, the red fluorescence from the P-cell-specific nucleoplasmic tdTomato marker was normally constrained to a bright nucleoplasmic signal. In cells scored as a ruptured nucleus, the red fluorescence was more diffuse and found throughout the entire cytoplasm of the P cell.

For some experiments, C. elegans animals were roughly synchronized in the middle of the L1 larval stage when P-cell nuclear migration takes place (Sulston and Horvitz 1977). Six plates of gravid hermaphrodites were bleached as described (Stiernagle 2006), and the surviving embryos were split between three plates to develop at 15, 18, or 20°C for 18 hours. The plates with the most L1 animals in the middle of P-cell nuclear migration were identified by examining P-cell nuclei with the phlh-3::nls::tdTomato marker using a Leica FLIII fluorescent dissecting stereo microscope. L1 animals were washed off the plate with M9 buffer (Stiernagle 2006) and mounted on a 2% agarose pad. Then, tetramizole was added to a final concentration of 1 μM as a paralytic, and a cover slip was applied and sealed with VALAP (equal parts vaseline, lanolin, and paraffin mixture). L1 larvae were imaged with the Zeiss LSM 980-Airyscan2 with a 63x NA 1.4 objective in the MCB Light Microscopy Imaging Facility at UC Davis made available through an NIH grant S10OD026702.

Alphafold

Alphafold models of FLN-1 and FLN-2 were generated using ColabFold version 1.5.2 (Mirdita et al. 2022) using the “alphafold2_ptm” model. No templates were used, and five models were generated (using default settings). The models shown have the highest predicted local distance difference test (pLDDT) score. Protein diagrams were colored by sequence order from blue (N terminus) to red (C terminus) using the “rainbow” command in UCSF ChimeraX, developed by the Resource for Biocomputing, Visualization, and Informatics at UCSF with support from the National Institutes of Health R01-GM129325 and the Office of Cyber Infrastructure and Computational Biology, National Institute of Allergy and Infectious Diseases (Pettersen et al. 2021). To disentangle the folded Ig domains from disordered loops, low-pLDDT regions were stretched out by assigning each torsion angle drawn from a normal distribution (phi = −165° +/− 5°, psi = 165 +/− 5°, where 5° is the standard deviation).

Results

fln-2 is an enhancer of the nuclear migration defect observed in unc-84 animals

We previously identified enhancers of the nuclear migration defect of unc-84 (i.e. emu genes), which function parallel to LINC complexes during P-cell nuclear migration (Chang et al. 2013). While analyzing mutant alleles from an emu screen, we found that mutants in fln-2 enhance the nuclear migration defect of unc-84 at 15 and 25°C (Fig. 1a–d). We counted the number of GABA neurons present in young adults using an established unc-47::gfp marker (McIntire et al. 1997). Since 12 of the 19 GABA neurons are derived from P cells (Sulston and Horvitz 1977), the absence of GABA neurons can be used as a proxy for scoring the number of P cells that died during nuclear migration in L1 larvae (Fig. 1a–c) (Chang et al. 2013; Fridolfsson et al. 2018). Thus, a completely penetrant P-cell nuclear migration defect would result in 12 missing GABA neurons. Animals harboring fln-2(tm4687), an expected null allele, had a mild P-cell nuclear migration defect with an average of 2.4 missing GABA neurons at 15°C (Fig. 1a–d). Both fln-2(tm4867) and fln-2(RNAi) animals significantly increased the nuclear migration defects observed in unc-84 mutant animals with an average of 6.6 and 4.6 missing GABA neurons, respectively, in the double mutants grown at 15°C (Fig. 1d). fln-2(tm4687) also significantly enhanced unc-84(n369) at 25°C (Fig. 1d). Therefore, fln-2 likely plays a role during P-cell nuclear migration.

Fig. 1.

fln-2 mutants enhance the P-cell nuclear migration defect observed in unc-84 worms. a–c) Representative epifluorescence images of unc-47::gfp-marked GABA neurons in wild-type, unc-84(n369), or unc-84(n369); fln-2(tm4687) animals. d) The numbers of missing GABA neurons are shown from the following genotypes: unc-84(n369) (black), fln-2(tm4687) (magenta), unc-84(n369) fln-2(tm4687) (red), and unc-84(n369) fln-2(RNAi) (pink). **** signifies a P-value < 0.0001 when compared with unc-84(n369) alone. e) Schematic of four of the more than 25 predicted isoforms of fln-2. The purple diamond indicates the ot611 point mutant producing an early stop codon. The pink box indicates the X-4P01 clone used for RNAi. The magenta box shows the location of the tm4687 deletion allele. The fosmids used in these experiments are marked below the isoforms. The red box marks sequences encoding a predicted actin-binding domain of three calponin homology repeats that is only present in the longer isoforms. f–g) Missing GABA neuron counts for the unc-84(n369) fln-2(tm4687) mutants with and without the indicated fosmids. Rescue experiments were performed on animals grown at 15, 20, or 25°C as indicated below the graphs. + indicates the presence of the extrachromosomal array carrying the fosmid(s) and an RFP, while- indicates there was no fosmid(s) present. *, **, ***, and **** indicate a P-value of <0.02, <0.003, 0.0007, and <0.0001, respectively; error bars are 95% confidence intervals.

According to Wormbase (www.wormbase.org), over 25 fln-2 isoforms are predicted to exist, four of which are shown in Fig. 1e (DeMaso et al. 2011). We aimed to determine which fln-2 isoforms were necessary for P-cell nuclear migration. The fln-2(tm4687) allele, which has a 547-bp deletion that is predicted to cause a frameshift and disrupt most fln-2 isoforms, caused a strong nuclear migration defect in the unc-84(null) background (Fig. 1e). Furthermore, fln-2(RNAi) performed using clone X-4P01 (Fraser et al. 2000; Kamath et al. 2003) caused a similar phenotype. In contrast, the fln-2(ot611) mutant, which introduces an early stop codon in the longer fln-2 isoforms, was found in our background strain, UD87, that was used for our enhancer screen (Chang et al. 2013). Therefore, the ot611 allele has no phenotype, suggesting that the longer fln-2 isoforms are dispensable for P-cell nuclear migration (Fig. 1a–e). To confirm that the fln-2(tm4687) P-cell nuclear migration defect was caused by the deletion in fln-2 and not somewhere else in the genome of this strain, we performed fosmid rescue experiments (Fig. 1e–g). A combination of two fosmids (i.e. WRM0625cB05 and WRM0620dC04) that span the largest fln-2 isoforms rescued the fln-2(tm4687) unc-84(n369) P-cell nuclear migration defect (Fig. 1f). Furthermore, a shorter fosmid, WRM0625cB05, that did not cover the predicted actin-binding domain in the longer fln-2 isoforms, was also able to rescue the P-cell nuclear migration defect of fln-2(tm4687) unc-84(n369) (Fig. 1g). The level of rescue observed here was significant but may not have been complete. Nonetheless, these results support the hypothesis that the expression of the smaller fln-2 isoforms that lack the predicted actin-binding domain, fln-2c or fln-2d, is sufficient for P-cell nuclear migration.

Ig-like repeats 4–8 of FLN-2 are necessary for P-cell nuclear migration

Like other filamin family members, the shorter isoforms fln-2c/d are predicted to encode proteins mostly made of tandem Ig-like repeats (DeMaso et al. 2011). We performed CRISPR/Cas9-mediated gene editing to generate in-frame deletions in the endogenous fln-2 locus to better determine which FLN-2 Ig-like repeats were necessary for P-cell nuclear migration. fln-2c consists of 2,776 amino acids and has 20 predicted Ig-like repeats (Fig. 2a). We divided FLN-2c into three overlapping segments, Ig repeats 4–9, 9–15, or 15–23 (DeMaso et al. 2011) and made three different deletion strains (Fig. 2a). Each deletion strain was analyzed for defects in P-cell nuclear migration by counting the number of missing GABA neurons present at 15°C in the presence or absence of unc-84 (Fig. 2). Deletion of Ig-like repeats 4–9 significantly enhanced the unc-84(null) nuclear migration defect (Fig. 2b), whereas deletion of repeats 9–15 or 15–23 had no effect (Fig. 2c–d). These results narrow down a part of FLN-2 that is necessary for P-cell nuclear migration to Ig-like repeats 4–8, which includes the disordered region between repeats 6 and 7.

Fig. 2.

Deletion of Ig-like repeats 4–9 enhances the unc-84 P-cell nuclear migration defect. a) Schematic of the FLN-2c isoform, which contains 20 predicted Ig-like repeats (maroon boxes). Ig-like repeats 1–3 are only found in the longer isoforms of FLMN-2 with interspersed disordered regions indicated by the zigzag lines. The region between repeats 6 and 7 is predicted to be disordered. The deleted portions of FLN-2 in three strains made by CRISPR/Cas9 editing are indicated below the schematic. The scale bar indicates 100 amino acids. b) Missing GABA neurons at 15°C were scored as a proxy for failed P-cell nuclear migrations. See Table 1 for details on the genotypes of the strains used here. Means with 95% confidence intervals are shown. *** indicates significant differences with a P-value of <0.001 when compared with unc-84 null animals, and ns stands for no significant difference from unc-84 null animals. A t-test was performed to generate P-values.

The actin network in P cells is not grossly disrupted in fln-2 mutants

Since filamins are thought to crosslink and bundle actin (Razinia et al. 2012), we examined the organization of the actin cytoskeleton during P-cell nuclear migration in fln-2 mutant L1 larvae. We compared two reporters for actin filaments—the small actin-binding domain from vab-10 bound to a fluorescent protein (vab-10::mVenus) under control of the P-cell-specific hlh-3 promoter expressed from a high-copy number extrachromosomal array (Bone et al. 2016) and a single-copy transgene expressing lifeact::mKate2 (Riedl et al. 2008) under the hlh-3 promoter inserted into a known locus (ttTi5605) using mosSCI (Frøkjær-Jensen et al. 2014). Both the over-expressed vab-10::mVenus (Fig. 3) and the single-copy lifeact::mKate2 markers localized in similar patterns when imaged with Airyscan super-resolution microscopy.

Fig. 3.

Actin networks during P-cell nuclear migration. a) Representative Airyscan super-resolution images of a wild-type L1 larvae during P-cell nuclear migration. P-cell nuclei are marked by nls::tdTomato (magenta), and actin in P cells is tagged with vab-10::mVenus (cyan). The gold box indicates a P-cell mid-nuclear migration that is enlarged in the inset below. The orange arrow points out actin perpendicular to the direction of nuclear migration in the lateral compartment, and the white arrow indicates actin parallel to the direction of nuclear migration in the constriction. A ventral view is shown, and the ventral cord is marked by the dashed line. anterior is left; scale bar is 10 mm. b and c) A ventral view (b) and lateral view (c) of P cells during the narrowing (b and c) and migration (b′ and c′) stages. P cells are in cyan; nuclei are in magenta. The dashed line indicates the ventral cord. The pharynx indicates the anterior end of the worm, and body wall muscles are represented by the transparent gray bar. d) Actin (vab10::mVenus; cyan) and P-cell nuclei (nls::tdTomato; magenta) in wild-type, unc-84(n369), fln-2(tm4687), or fln-2(tm4687) unc-84(n369) double-mutant animals. Maximum projections of z stacks are shown. The scale bar is 10 mm.

First, we characterized the wild-type organization of actin filaments during the narrowing and migration stages of P-cell nuclear migration (Bone et al. 2016). The most anterior pairs of P cells narrow, and their nuclei migrate earlier than the more posterior pairs. During nuclear migration, pairs of P-cell nuclei deform and move through the narrow constriction between the cuticle and body wall muscles to reach the ventral cord (Bone et al. 2016). Actin filaments were present both behind the nucleus and in the constriction during P-cell nuclear migration in 16 out of 16 worms (Fig. 3a). We frequently observed actin bundles forming perpendicular to the direction of nuclear migration in the lateral portion of P cells, positioned behind migrating nuclei. These actin bundles could push nuclei from the rear as they migrate into the constricted space (orange arrows, Fig. 3a). Additionally, we observed actin fibers forming parallel to the direction of nuclear migration within the constriction (white arrows in Fig. 3a). We then compared the organization of actin networks in our mutant animals. In unc-84(n369) or fln-2(tm4687) single mutants and in fln-2(tm4687) unc-84(n369) double mutants, the actin networks did not have any gross changes as compared to wild type (Fig. 3d), suggesting that fln-2 is not required for the presence or organization of actin networks during P-cell nuclear migration.

FLN-2 is not a canonical filamin

If FLN-2 was acting as a canonical filamin, we would expect it to localize to the actin network, as canonical filamins directly interact with actin (Razinia et al. 2012). We therefore explored the extent to which FLN-2 might colocalize with actin networks. We localized FLN-2c in vivo using a split-GFP assay where the beta-strand 11 of GFP was fused to the endogenous locus for FLN-2c and strands 1–10 of GFP were expressed under the control of a P-cell-specific promoter, hlh-3 (Fig. 4a). The spGFP11 tag did not disrupt FLN-2 function; unc-84(n369); gfp11::fln-2c animals had no significant defect in the number of GABA neurons when compared with unc-84(n369) animals (Fig. 4b). spGFP11::FLN-2c localized diffusely throughout the cytoplasm and was slightly enriched in the nucleoplasm (Fig. 4a). This diffuse pattern of FLN-2 localization was quite distinct from the localization pattern of actin observed in these cells (Fig. 3), suggesting that the bulk of FLN-2 does not co-localize to actin networks in P cells.

Fig. 4.

Localization of FLN-2::spGFP in larval P cells. a) Representative Airyscan super-resolution image of fln-2c::GFP expressed in a C. elegans L1 larva. A lateral view is shown; ventral is toward the lower right. Scale bar is 10 mm. b) Graph depicting the number of missing GABA neurons in unc-84(n369), unc-84(n369) + Ex[unc-84(+)], fln-2c::spGFP unc-84(n369), and fln-2c::spGFP unc-84(n369) + Ex[unc-84(+)] worms. Error bars are 95% CI, and nonsignificant P-values > 0.05.

The localization of FLN-2 and the lack of importance of the predicted actin-binding domain present in longer isoforms of FLN-2 during P-cell nuclear migration are not consistent with FLN-2 functioning as a canonical filamin. In further support, the primary amino acid sequence of FLN-2 is divergent from canonical filamins, falling far outside of a phylogenetic tree where human FLNA, FLNB, FLNC, C. elegans  FLN-1, and Drosophila Jbug cluster together (DeMaso et al. 2011). We therefore used Alphafold to predict the structures of FLN-2d and the canonical filamin FLN-1 (Fig. 5). FLN-1 was predicted with high confidence to consist of two N-terminal CH domains (gray in Fig. 5) followed by a chain of Ig repeats, as expected. In contrast, only a few Ig repeats were predicted in FLN-2, and instead, it was predicted to have extensive disordered stretches.

Fig. 5.

Alphafold predictions of the structure of FLN-1 and FLN-2. The highest confidence of five FLN-1a (top) and FLN-2d (bottom) structures predicted by ColabFold version 1.5.2 is shown. On the right, the pLDDT scores (a per-residue confidence metric) for five different predictions are shown. The protein diagrams are colored by sequence order from blue (N terminus) to red (C terminus).

fln-2 mutant P-cell nuclei rupture post-migration

The nuclei of cultured mammalian cells migrating through fabricated constrictions often rupture, releasing soluble nucleoplasmic proteins into the cytoplasm (Denais et al. 2016; Raab et al. 2016). We examined the extent to which P-cell nuclei rupture during migration through constricted spaces in C. elegans L1 larvae. We followed nuclear rupturing with the localization of a soluble nuclear marker (nls::tDtomato) to see whether it leaked into the cytoplasm before, during, or after P-cell nuclear migration. Normally, nls::tDtomato was restrained within the nucleus, but after nuclear rupturing, the fluorescent signal was diluted and detected throughout the cytoplasm in both the lateral and ventral compartments of P cells (arrowheads in Fig. 6a). In wild-type L1 larvae, there were rarely any ruptured nuclei prior to or during nuclear migration. However, a few wild-type nuclei ruptured toward the end of the nuclear migration process (Fig. 6b). The number of nuclear rupture events observed in late migration was significantly increased in both unc-84(n369) and fln-2(tm4687) single-mutant larvae relative to wild type. Furthermore, fln-2(tm4687), unc-84(n369) double-mutant animals had a significant enhancement of the nuclear rupture phenotype (Fig. 6b). These data suggest that unc-84 and fln-2 function synergistically in maintaining the integrity of the nuclear envelope during P-cell nuclear migration.

Fig. 6.

P-cell nuclei in fln-2  unc-84 mutant animals often rupture during migration. a) Representative Airyscan super-resolution images of P cells late in L1 after or during nuclear migration into the ventral cord in wild-type, unc-84(n369), fln-2(tm4687), and fln-2(tm4687) unc-84(n369) animals. Actin is shown to mark the shape of P cells. Arrows point to suspected nuclear rupture events where the nls::tdTomato marker has leaked into the cytoplasm of the P cell. Maximum projections of z stacks are shown. The scale bar is 10 mm. b) The number of P cells with cytoplasmic nls::tdTomato per worm in late L1. Nuclear rupture was assayed at about 20°C, room temperature for the microscope room. Means and 95% confidence intervals are shown. **** denotes a P-value of <0.00001.

fln-2 functions in parallel to the LINC complex and actin-based pathways

The fln-2, unc-84 double-mutant phenotype is not completely penetrant. Even at the restrictive temperature, many P-cell nuclei still complete their migration, suggesting that other pathways are still functioning. Double mutants in the CDC-42/actin-based pathway and the LINC complex pathway have a similar, incompletely penetrant, phenotype (Ho et al. 2023). We therefore tested whether fln-2 is in the same actin-based pathway as the cdc-42 GEF cgef-1. To test this hypothesis, we examined P-cell nuclear migration in fln-2(RNAi), cgef-1(gk261), unc-84(n369) triple-mutant animals. Animals carrying all three mutations had significantly more missing GABA neurons than either fln-2(RNAi), unc-84 or cgef-1, unc-84 double mutants (Fig. 7). Thus, fln-2 and cgef-1 have a synergistic relationship during nuclear migration through confined spaces in P cells. These data are consistent with a model where fln-2, cgef-1, and unc-84 act in three parallel pathways to ensure P-cell nuclear migration through constricted spaces.

Fig. 7.

fln-2 has synthetic interactions with unc-84 and cgef-1. Graph depicting the number of missing GABA neurons in the designated genotypes. The mean is shown with 95% confidence interval error bars. Data were collected at 15°C. n is at least 15. *** denotes P < 0.001 from a paired t-test.

Discussion

We identified fln-2 as part of a new genetic pathway that functions to move P-cell nuclei through constricted spaces in developing C. elegans. Previous studies focused on two other pathways. First, the LINC complex pathway recruits the microtubule motor dynein to the surface of nuclei to move them toward the minus ends of microtubules (Fridolfsson et al. 2010; Bone et al. 2016; Ho et al. 2018). The second pathway is an actin-based process that includes the small GTPase CDC-42, its GEF CGEF-1, TOCA-1, and the Arp2/3 nucleating complex (Chang et al. 2013; Ho et al. 2023). Our genetic analyses showed that mutations in fln-2 synergistically interacted with both unc-84 of the LINC complex pathway and cgef-1 of the actin-based pathway. Thus, there are at least three genetic pathways that function together to move P-cell nuclei. Since P-cell nuclei migrate as a normal part of development through narrow openings (Bone et al. 2016), this system is an excellent model for better understanding how nuclei might migrate through constricted spaces as part of neuronal development, hematopoiesis, inflammation, and metastasis.

One potential model for how FLN-2 could function during nuclear migration through confined spaces in P cells is by organizing the actin cytoskeleton. FLN-2 is a predicted divergent filamin (DeMaso et al. 2011), and mammalian filamins function, at least in part, by crosslinking or bundling actin filaments (Nakamura et al. 2011). However, our data do not support such a role for FLN-2. First, a short isoform, FLN-2c, that is missing the three N-terminal calponin homology repeats previously predicted to be an actin-binding domain, is sufficient for nuclear migration in P cells. Second, the fln-2(ot611) allele, which is predicted to introduce an early stop codon in the longest isoform of FLN-2, is found in many wild-type strains, including our lab's version of N2, suggesting that the longest isoforms with the predicted actin-binding domains are not selected for any significant functions (Sarin et al. 2010). Our own sequencing found that the ot611 mutation was not present in the fln-2(tm4687) background. Furthermore, FLN-1 is the canonical filamin in C. elegans, while FLN-2 is evolutionarily quite distant from other animal filamins (DeMaso et al. 2011), suggesting that it may play divergent functions. Furthermore, Alphafold predictions of FLN-2 did not support the hypothesis that it folds like a canonical filamin, predicting instead that FLN-2 consists of extensive stretches of disordered domains. Third, fln-2 functions parallel to the CDC-42/actin-based pathway. Finally, fln-2 mutant animals had grossly normal actin networks in larval P cells. Together, these data do not support a model where FLN-2 functions in the organization of actin networks during P-cell nuclear migration and that it probably is better if FLN-2 is not thought of as a filamin.

An alternative model that we favor is that FLN-2 helps to maintain and/or repair the mechanical integrity of the nuclear envelope during P-cell nuclear migration. fln-2, unc-84 mutant P-cell nuclei had a significant increase in rupture frequency of the nuclear envelope compared to the single mutants or wild type. Nuclear envelope rupture and subsequent DNA damage regularly occurs when dendritic or cancer cells migrate through fabricated constrictions (Denais et al. 2016; Raab et al. 2016). Nuclear envelope rupture is repaired by the endosomal sorting complex require for transport (ESCRT) complex (Olmos 2022). Since P cells divide soon after nuclear migration to the ventral cord (Sulston and Horvitz 1977; Altun and Hall 2024), the FLN-2 pathway might function to prevent nuclear rupture and therefore prevent DNA damage during a window when the cell is preparing to divide.

FLN-2 also functions in the formation of multivesicular bodies. Three alleles in fln-2 were found in a forward genetic screen for diminished VHA-5::RFP-containing puncta, a marker for multivesicular bodies (Shi et al. 2022). fln-2 mutants also have a reduced number of multivesicular bodies as assayed by electron microscopy (Shi et al. 2022). Interestingly, both multivesicular body formation and nuclear envelope repair have a common molecular mechanism through the ESCRT pathway (Olmos 2022). However, FLN-2 acts at a different step than the ESCRT pathway to regulate multivesicular bodies (Shi et al. 2022). Shi et al (2022) conclude that the N-terminus of the longest FLN-2 isoforms functions by binding to both actin filaments and the VHA-8 subunit of multivesicular bodies. However, they did not investigate whether the fln-2(ot611) allele induces similar phenotypes. The presence of the fln-2(ot611) allele in many wild-type strains predicts that the N-terminal domain of FLN-2 is dispensable for P-cell nuclear migration. It is thought that the fln-2(ot611) allele arose in what the Caenorhabditis Genetics Center calls the N2 male strain; the ot611 allele was likely not in the original N2 strain and is not in other tested wild isolates (Zhao et al. 2019). Nonetheless, the ot611 allele is now found in many laboratory strains, including some of the strains from the Gene Knockout Project and the UD87 strain we used for our previously described enhancer of unc-84 screen (Chang et al. 2013; Zhao et al. 2019). The fln-2(ot611) allele reduces early mortality caused by pharyngal bacterial infections, increasing lifespan ∼11% (Zhao et al. 2019). This defect is subtle compared to the nuclear migration defects described here, which leads to strong Egl and Unc phenotypes. The molecular mechanisms underlying how long isoforms of fln-2 partially protect adults from bacterial pharyngeal infections and how these mechanisms relate to the function of fln-2c in P-cell nuclear migration remain unknown. In conclusion, these data suggest a model where shorter isoforms of FLN-2 function in nuclear envelope and multivesicular body maintenance and/or repair. The molecular mechanism underlying how FLN-2 functions and interplays with the ESCRT pathway requires further studies.

Data availability

Strains and plasmids are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables.

Funding

This research was funded by a grant from the National Institutes of Health, R35GM134859, to D.A.S. and the training grant T32GM00737 supported L.M.