In animal development, cells exit the cell cycle after a final round of cell division to differentiate into specialized cell types. However, the impact of failed cytokinesis on the fate specification of postmitotic cells, especially neurons, remains poorly understood. We analyzed the Caenorhabditis elegans mutants lacking the midbody-tethered endonuclease LEM-3/ANKLE1, which resolves chromatin bridges during cytokinesis. Using genetic analysis, fluorescence microscopy, and RNA visualization, we found that unresolved DNA bridges lead to the formation of stable intercellular bridges, creating binucleate syncytia between touch receptor neurons and sister cells, which allows thorough cytoplasmic mixing despite independent nuclear transcription. In some cases, the cytoplasmic linkage between sister cells also led to the suppression of neuronal fates and prevention of apoptotic fates. Thus, LEM-3 prevents syncytium formation to ensure proper cellular differentiation.

The loss of lem-3 results in binucleate syncytia connecting touch receptor neurons (TRNs) and their sister cells through an intercellular bridge

In a forward genetic screen searching for genes that regulate TRN fate specification, we isolated a mutant allele unk14 that caused the expression of TRN markers in extra cells. The wild-type animals have six TRNs, including ALML/R, AVM, PVM, and PLML/R, which were specifically labeled by the mec-17p::TagRFP transgene, whereas unk14 mutants showed extra cells near AVM and PLM cell bodies in ~15% of the animals at 25 °C. Strikingly, the extra cells were physically connected with the AVM and PLM cell body through an stable intercellular bridge, which is also marked by TagRFP (Fig 1A – 1D). Using chromosomal mapping and whole-genome resequencing, we mapped unk14 to a mutation in lem-3, which codes for a midbody-tethered, LEM domain-containing DNA endonuclease [2,12]. The mutation caused a V618M substitution in the LEM-3 protein. To confirm that the loss of lem-3 caused the phenotype of producing extra TRN-like cells, we tested a nonsense allele of lem-3, named mn155(R190*), which caused a similar phenotype and failed to complement unk14. We also generated lem-3 knockout alleles (unk84 and unk85) using CRISPR/Cas9-mediated gene editing and found that these alleles also led to the presence of extra cells labeled by TagRFP. Importantly, the extra TRN phenotype in lem-3(−) mutants was confirmed using other TRN fate markers (S1A and S1B Fig) and can be rescued by a transgene carrying a wild-type copy of lem-3(+), further confirming that the loss of lem-3 resulted in developmental defects in restricting TRN fate in the six canonical TRNs (Fig 1E).

Previous studies found that during the early cell divisions of C. elegans embryos, LEM-3 is recruited to the midbody from anaphase to the late stage of cytokinesis to resolve chromatin bridges caused by various errors in DNA replication, recombination, or chromosome condensation [2]. The presence of chromatin bridges (and possibly ultra-fine DNA bridges) delays abscission through the activation of Aurora B kinase-mediated NoCut pathway [6]. Aurora B activates LEM-3, which cleaves the DNA bridge and allows cytokinesis to proceed. In lem-3(−) mutants, chromatin bridges are not processed and are eventually broken by actomyosin dependent-mechanical force, which results in error-prone repair and chromosomal rearrangement [13]. This forced break of chromatin bridges by cytokinesis was found in continuously dividing cells in culture or in the early embryos. However, the persistence of the cytoplasmic bridges connecting TRNs and presumably their sister cells in adult animals (>3 days after the scheduled cell division) led us to hypothesize that cytokinesis was not only delayed but aborted in response to the chromatin bridges in the last round of cell division that generates postmitotic cells.

We confirmed that the extra cells labeled by the TRN markers are the sister cells of TRNs by examining the cellular features for each of the four TRN subtypes in the lem-3(−) mutants. For the postembryonic AVM neuron, the extra cell connected with it resembled the morphology of its sister cell SDQR neuron (S1C Fig), contained a nucleus labeled by histones (S1D Fig), and expressed SDQ markers lad-2p::GFP, flp-12p::GFP, ast-1::GFP, and ser-2p2::GFP (Figs 1B and S1E). Interestingly, AVM was also labeled by the SDQ marker, suggesting the co-expression of cell fate markers in the binucleate syncytium. Similar results were obtained for PVM-SDQL syncytium, although its occurrence has much lower penetrance (<3%) than the AVM-SDQR syncytium (10%–20%) (Figs 1B and S1F).

The sister cell of ALM is the peptidergic interneuron BDU. We observed the ALM-BDU syncytium in ~5% of the lem-3(−) mutants. In some cases, the ALM cell body is directly connected with the BDU cell body (Fig 1C), while in other cases, a bridge linked the ALM cell body or anterior neurite with the BDU posterior neurite (S1A and S2A–S2C Figs). Despite that ALM and BDU migrate towards the opposite directions after the cell division, their physical connection is maintained throughout development into adulthood and reaches>100 μm in length in adults. BDU markers zig-3p::GFP, nlp-80p::GFP, C15C7.4p::GFP, and flp-10p::GFP were not expressed in the ALM-BDU syncytium (Figs 1C and S2A), suggesting that the BDU fate was suppressed by the linkage with ALM. Nevertheless, we found exceptions in two BDU markers ceh-14::EGFP and ser-2p2::GFP, whose expression persisted in the ALM-BDU syncytium, suggesting that the suppression of BDU fate may be incomplete (S2B Fig). In fact, the morphology of the ALM-connected BDU largely resembled the normal BDU, indicating at least partial differentiation of the BDU fate (S2C Fig).

In the case of PLM, the TRN fate dominated over the fate of its sister cell, ALN neuron, when the two cells were connected. The PLM-ALN syncytium was only labeled by the TRN markers (e.g., mec-17p::TagRFP, mec-4p::YFP, and mec-18p::GFP) and not by the ALN markers (e.g., lad-2p::GFP, nlp-20p::GFP, ast-1::GFP, and ser-2p2::GFP) in lem-3(−) mutants (Figs 1D, S1A, and S2D–S2E). Whenever an extra cell was labeled by the TRN marker, we observed the loss of a GFP-labeled ALN neuron, which indicated that the formation of PLM-ALN connections converted the ALN neuron into PLM-like cells.

The significant variation in the penetrance of the syncytium phenotype among the TRN subtypes is intriguing (Fig 1E). We suspect that the subtypes’ different lineage histories, developmental timing, anatomical positions along the body, and physical distance with their sister cells might contribute to such variation in penetrance.

Ectopic presence of MEC-3 converts TRN sisters into TRN-like fate

The suppression of BDU and ALN fates in the ALM-BDU and PLM-ALN syncytia, respectively, could be explained by the presence of the TRN fate selector MEC-3 [14] in the sister neurons via the diffusion through the intercellular bridge (see Fig 3). To test this idea, we expressed mec-3 in all neurons using the unc-119 promoter and found that the forced expression of mec-3 in BDU and ALN neurons indeed activated the TRN fate marker and suppressed the sister fate markers (S3A, S3B Fig), confirming that ectopic presence of MEC-3 in the sister cells can convert them into a TRN-like state. Interestingly, the converted BDU neuron had an ALM-like morphology (e.g., posteriorly displaced cell body, subdorsal axons, and the loss of ventral dip) and did not express any BDU markers (e.g., zig-3p::GFP and ser-2p2::GFP; S3C Fig), suggesting that the fate conversion caused by MEC-3 overexpression is more extensive than that induced by MEC-3 diffusion from the ALM cell body in the ALM-BDU syncytium. Our results are consistent with previous findings that MEC-3 could outcompete the BDU fate selector PAG-3 for binding to the master regulator UNC-86, which is required for the specification of both ALM and BDU fates [15]. Although the ALN fate selector is unknown, we reason that a similar mechanism might allow ectopic MEC-3 to transform ALN into a TRN-like cell.

To further confirm that MEC-3 is responsible for the fate conversion, we created lem-3(−); mec-3(−) double mutants and found that the ALM-BDU and PLM-ALN syncytia no longer expressed the TRN markers; instead, they showed the expression of the sister cell markers (S4A–S4C Fig). Interestingly, as previously reported [15], ALM neurons adopted a BDU-like fate and morphology (i.e., ventral shift of cell body and axon) in mec-3(−) mutants, but PLM did not adopt the ALN fate.

Independent transcriptional programs of the two nuclei in the syncytium

The co-labeling of AVM-SDQR and PVM-SDQL syncytia by both TRN and the sister cell fate markers could result from either 1) active transcription of both fate reporters in the two nuclei of the syncytia due to the mixing of cell fate determinants or 2) the diffusion of fluorescent proteins produced only in one cell body. To discern the two possibilities, we visualize the transcription of the cell fate-associated genes from their endogenous loci in the binucleate syncytia using single-molecule fluorescence in situ hybridization (smFISH). In the majority of the AVM-SDQR syncytium, only the AVM cell body contained the mRNAs of TRN genes (e.g., mec-17, mec-18, and mec-3), while only the SDQR cell body contained the mRNAs of SDQ genes (lad-2, ast-1, flp-12, and ser-2; Figs 2A and S5), suggesting that the two cells transcribed different genetic programs despite the cytoplasmic connection. This transcriptional independence was not due to the lack of selector diffusion (see below) but to the inability of the selectors to activate their downstream genes in the nucleus of the sister cell. We reason that a restrictive chromatin state, as previously suggested [16,17], may have prevented the activation of alternative fate programs by the ectopic terminal selectors and avoided the mixing of cell fate at the transcriptional level in most AVM-SDQR syncytia.

However, in about 15% of the animals, we did observe both nuclei of the syncytium actively transcribing the TRN genes, suggesting that the TRN and SDQ fates could be co-expressed from the same nucleus in some cases (Fig 2A). This result is consistent with the finding that forced overexpression of mec-3 in SDQR can turn on the TRN fate without repressing the SDQR fate (S3A and S3C Fig), suggesting that a high level of ectopic MEC-3 can enable the co-existence of TRN and SDQ fate in the same cell. Moreover, the activation of the TRN genes in SDQR of the syncytium required MEC-3, since the syncytium no longer expressed the TRN marker and only expressed the SDQ marker in lem-3(−); mec-3(−) mutants (S4D, S4E Fig). Interestingly, two SDQ fate markers flp-12 and ser-2 were activated in ~40% of the AVM nuclei, suggesting that some AVM may adopt a partially mixed fate in the syncytium. The PVM-SDQL syncytium was similar to the AVM-SDQR pair but with a higher frequency of cell fate mixing (S6A Fig).

For the ALM-BDU syncytium, we found that both nuclei transcribed endogenous TRN fate markers in all cases, and the expression of BDU fate markers (e.g., nlp-80, C15C7.4, D2089.3, ser-2) was suppressed to various degrees (Figs 2B, S5, and S6), which supports the conversion of BDU into a TRN-like fate due to the diffusion of MEC-3. A similar fate conversion of ALN neurons was observed in the PLM-ALN syncytium, since both cell bodies showed the mRNAs for TRN genes and did not contain the mRNAs for the ALN genes (e.g., nlp-20, lad-2, and ser-2) (Fig 2C), which confirmed the suppression of ALN fate in the syncytium due to the diffusion of MEC-3 from PLM to the ALN. Interestingly, the ALN gene ast-1 persisted in the ALN nucleus of all the PLM-ALN syncytia, suggesting that the fate conversion is still partial (Fig 2C). It is worth noting that although the otIs358[ser-2p2::GFP] reporter is not expressed in wild-type ALM and PLM neurons, we detected ser-2 mRNAs in ~60% of the TRNs, making it a suboptimal marker for the sister fate (S5C and S5D Fig). Nevertheless, we still observed the suppression of ser-2 mRNA expression in 13% and 38% of ALM-BDU and PLM-ALN syncytia, respectively, compared to their 100% presence in the wild-type BDU and ALN neurons (S6B and S6C Fig). The above results suggest that the transcriptomes of the two nuclei in a syncytium are not necessarily synchronized and can be independently regulated. The fate specification programs of the TRNs and their sister cells appear to have different susceptibilities to perturbation caused by the cytoplasmic linkage between them.

The mixing of cellular contents between two physically connected cells

The AVM-SDQR syncytium provided an opportunity to understand the extent of cytoplasmic mixing between two transcriptionally (and presumably also translationally) independent but physically connected cells. When labeled by the mec-17p::GFP reporter, GFP mRNA was only detected in the AVM cell body of the syncytia (Figs 3A and S6D), but the GFP protein diffused to fill the entire syncytium, including neuronal processes, suggesting that cytoplasmic soluble proteins can pass through the intercellular bridge. To confirm this idea, we expressed the photoconvertible Dendra2 from the mec-17 promoter in the AVM-SDQR syncytium, photoconverted Dendra2 from green to red only in the AVM cell body, and observed the red fluorescent signal in the SDQR cell bodies within minutes (Fig 3B and 3C). Similar results were obtained in the ALM-BDU syncytium, although there is slower kinetics in the increase of Dendra2 signals possibly due to the longer passage way connecting the ALM and BDU cell bodies (S7A and S7B Fig). Moreover, when we severed the intercellular bridge using a UV laser and then photoconverted Dendra2 in AVM or ALM, we could no longer observe the increase of red fluorescence in the sister cells (S7C–S7F Fig), confirming that the bridges mediated the cytoplasmic exchange between TRNs and the sister cells.

Severing of the bridge also led to the gradual disappearance of the TagRFP (expressed from the TRN-specific mec-17 promoter) in the SDQR over time, supporting that the SDQR nucleus did not activate the mec-17 promoter and the fluorescent protein was produced in AVM and diffused into SDQR (Fig 3D and 3E). This result also indicated that cutting the cytoplasmic linkage between AVM and SDQR may help unmix their cell fates and revert them back to their respective fates.

To assess what types of proteins can pass through the intercellular bridge, we expressed fluorescently tagged cytoplasmic proteins (e.g., MEC-18::GFP), nuclear proteins (e.g., GFP::NLS, MEC-3::EGFP), membrane proteins (e.g., PLCdPH::GFP, MEC-4::TagRFP), ER-associated proteins (e.g., MEC-6::GFP), synaptic vesicle proteins (e.g., GFP::RAB-3), and mitochondrial proteins (e.g., mitochondrial matrix tagged GFP) from TRN promoters, which were activated in the AVM nucleus. The presence of these proteins in the SDQR cell body would indicate that they can pass through the bridge. In all cases, we observed the fluorescent signals at comparable intensity in both cell bodies of the syncytium, indicating that diffusion (and/or transport) of the proteins through the bridge led to even distribution in the two connected cells (Fig 3F). Organelle-associated proteins maintained their normal subcellular localization in both cell bodies, suggesting that they could be potentially functional in both cells. Overall, the mixing of diffusible cytoplasmic content in the binucleate syncytium appeared to be thorough despite the largely distinct gene transcription profiles of the two nuclei.

Unresolved chromatin bridges and connected nuclear envelopes lead to the formation of intercellular bridge

Next, we investigated the cause of the intercellular bridge connecting TRNs and their sister cells in lem-3(−) mutants. Since LEM-3 and its human ortholog ANKLE1 are required to enzymatically process chromatin bridges that arise from mitotic nondisjunction [2,5,12], we hypothesized that the intercellular bridges connecting TRNs and their sister cells were induced by unresolved chromatin bridges in the lem-3(−) mutants. Since the penetrance of the TRN-sister syncytia was low (<3% at 20 °C or lower) in lem-3(−) mutants, we reasoned that the frequency of spontaneous chromosome segregation errors that need to be processed by LEM-3 is low. Interestingly, elevating the C. elegans culturing temperature to 25 °C increased the frequency of AVM-SDQR and PLM-ALN syncytia (Fig 4A), potentially due to the more frequent DNA replication and chromosomal segregation errors at higher temperature [18–20]. In fact, disrupting DNA synthesis by knocking down the DNA replication licensing factor mcm-7 [2] or treating the animals with the DNA topoisomerase II inhibitor ICRF-193 [21] significantly increased the penetrance of forming the syncytia in lem-3(−) mutants (Fig 4A). Since ICRF-193 is known to induce chromatin bridges and sister chromatid nondisjunction in anaphase [21,22], the result supports that the intercellular bridge was formed due to unresolved chromosomal linkages. Nevertheless, elevated temperature and the ICRF-193 treatment had only marginal effects on increasing the penetrance of ALM-BDU and PVM-SDQL syncytia (Fig 4B). As controls, treating wild-type animals with mcm-7 RNAi or ICRF-193 never resulted in the syncytium phenotype.

To confirm the presence of chromatin bridges, we found DNA (visualized by DAPI), histones (H2B), and nuclear envelope markers (e.g., LMN-1/Lamin [23] and EMR-1/Emerin [24]) in the intercellular bridge (Fig 4C and 4D). Importantly, these nuclear envelop linkages (e.g., between AVM and SDQR nuclei) were formed immediately after the telophase of the scheduled cell division and can last for several days throughout development and into the adulthood of the animal. When treating the animals with ICRF-193, we observed much stronger signal for nuclear envelope markers in the intercellular bridge (Fig 4C). Nevertheless, only 20~40% of AVM-SDQR syncytia showed persistent nuclear envelope connection 16 hours after the cell division (i.e., 24 hours after recovery from L1 arrest), suggesting that the DNA bridge could break as the two nuclei migrated away from each other during development (Fig 4E).

In fact, we observed the elongation of the AVM-SDQR bridge from ~6 μm in average length in L1/L2 animals to ~27 μm in adults, and the ALM-BDU bridge can extend to >100 μm in length (Fig 4F). The stretching of the chromatin bridge may have led to the mechanical breakage and the loss of nuclear linkage, but the cytoplasmic linkage through the bridge (a membrane structure) was maintained, likely due to the inactivation of cytokinesis machineries (Fig 4G). Compared to the spontaneously formed chromatin bridges, ICRF-193-induced chromatin bridges appeared to be more resistant to mechanical breakage, likely due to the accumulated chromosomal aberrations, since significantly more syncytia maintained the nuclear membrane connection in adults under ICRF-193 treatment compared to untreated lem-3(−) mutants (Fig 4E). When quantifying the distance between the AVM and SDQR nuclei, we found that some syncytia with long nuclear distance could still maintain the nuclear envelope connection into adulthood in the absence of ICRF-193 treatment, suggesting that some naturally occurring DNA linkages might be strong enough to survive the stretching (S8A and S8B Fig). Interestingly, we also found some rare cases where the two nuclei stayed in the same cell body (thus, there was no bridge); we reason that either furrow ingression never occurred, or regression happened due to the presence of chromatin bridges in lem-3(−) mutants (S8B Fig).

The endonuclease activity of LEM-3 is required to cleave the unresolved chromatin bridges during early embryogenesis [2]. As expected, the Y556A and G558A double mutations in the conserved GIY-YIG nuclease motif caused the formation of TRN-sister cell syncytia at similar penetrance as the lem-3(−) deletion allele (Fig 4H). Interestingly, the V618M mutation found in the unk14 allele we isolated from the genetic screen was also mapped to the nuclease domain. Moreover, Aurora B kinase-mediated phosphorylation of LEM-3 is required for its activity in resolving the bridge and preventing the syncytium, because the S192A and S194A double mutation of the putative Aurora B phosphorylation sites [2] caused phenotypes similar to lem-3(−) mutants. Inactivation of air-2, which codes for Aurora B in C. elegans, also led to the formation of the syncytia connecting TRNs and their sister cells, but we only observed the PLM-ALN syncytia using a temperature-sensitive (and likely partial loss-of-function) air-2 allele (Fig 4I).

In addition to LEM-3, we also examined the effects of other endonucleases, including MUS-81/MUS81, SLX-1/SLX1, and XPF-1/ERCC4, which are involved in resolving Holiday junctions, DNA replication errors, and interstrand crosslinks, as well as DNA repair [25–27]. Nevertheless, we did not observe any TRN-sister syncytium in either xpf-1(−) single mutants or xpf-1(−); mus-81(−) double mutants or mus-81(−) slx-1(−) mutants, suggesting that LEM-3 may have specific functions in cutting chromatin bridges connecting sister cells.

The intercellular bridge contains cytoskeleton structures

When a chromatin bridge forms, LEM-3 is recruited to the midbody, which is a transient microtubule structure formed within the intercellular connection between sister cells prior to abscission [2]. Using the centralspindlin protein ZEN-4/MKLP1 as a midbody marker [28], we first confirmed the recruitment of LEM-3 to the midbody to process the chromatin bridge during the cell division that generates AVM and its sister cell SDQR in wild-type animals treated with ICRF-193 (Fig 5A). In lem-3(−) mutants, the midbody formed normally and persisted in the intercellular bridge for hours, likely because the unresolved chromatin bridges aborted cytokinesis (Fig 5B and 5C). However, the ZEN-4 signal eventually disappeared 6 hours after telophase, suggesting that midbody structure cannot last for a long time if abscission does not occur (Fig 5C).

On the other hand, by visualizing the microtubules using a GFP-fused microtubule-associated protein MAPH-1.1 (a homolog of human MAP1 proteins) [29], we found that microtubule structure persisted in the intercellular bridge and likely provided structural support for the formation and maintenance of the bridge (Fig 5D). Interestingly, treatment with both microtubule-destabilizing and -stabilizing drugs (colchicine and paclitaxel, respectively) reduced the penetrance of AVM-SDQR syncytia, indicating that optimal microtubule stability and dynamics are crucial for forming or maintaining the intercellular bridge (Fig 5E). With the structural support provided by microtubules, the bridge could persist into even aged animals (e.g., day 5 adults; S8C Fig).

Using LifeAct::GFP that binds to actin filaments [30], we also found actin filaments in the intercellular bridge (Fig 5D). The presence of both microtubule and actin cytoskeletons suggested that the bridge may be maintained as a regular neuronal process by the syncytium. However, treating the animals with Latrunculin A that blocks actin polymerization did not affect the penetrance of AVM-SDQR syncytia, suggesting that the formation and maintenance of the bridge may not require actin filaments (Fig 5E). On the contrary, inhibiting ROCK-induced myosin phosphorylation using the ROCK inhibitor Y-27632, which is known to delay furrow ingression [31], significantly increased the penetrance of forming the AVM-SDQR bridge in lem-3(−) mutants (Fig 5E). This finding supports that unresolved chromatin bridges and weaker actomyosin contractility would delay cytokinesis and eventually abort abscission, resulting in permanent intercellular bridges between postmitotic sister cells.

Functional consequences of connecting AVM with its sister cell

In adult animals, AVM functions not only as a mechanosensor but also as a signaling hub to integrate anterior signals through its gap junctions with the two ALM neurons and the connections with downstream interneurons [32,33]. AVM’s role in signal integration and relay is thought to be important for the anterior bias in the response of the touch circuit when all TRNs are activated in a plate tapping assay [34]. Using a similar plate dropping assay that activates both anterior and posterior TRNs, we confirmed that 95% of the wild-type animals responded by reversal, supporting that the anterior TRNs-induced backward movement dominated over the posterior TRNs-induced forward movement. However, only 62% of the lem-3(−) mutants with AVM-SDQR syncytia responded to plate dropping by reversal, and 30% of the animals showed no response, suggesting that anterior and posterior stimulation likely canceled out each other (Fig 5F, left). The anterior bias disappeared in these animals. We confirmed that the lack of response was not due to the loss of mechanosensation in lem-3(−) animals, which showed similar touch sensitivity as the wild-type animals (Fig 5F, right). We reason that when AVM is connected with SDQR, the volume of the cell may grow as twice as the normal AVM, therefore diluting the neural signal (in terms of ion concentrations). Since AVM is activated through gap junctions, the AVM-SDQR syncytium may behave like a bigger sink that raises the threshold of activation and signal propagation. The effects of gap junctions in coupling neurons to increase the activation threshold for sensory stimuli was reported in previous studies [35].

Connection between apoptotic and nonapoptotic cells in lem-3(−) mutants

When examining the AVM-SDQR syncytia, we noticed that ~25% of them had a third cell body, which was connected with the other two cell bodies through intercellular bridges (Fig 6B). We referred to these structures as “three-body syncytia”. Since the sister cell (QR.pp) of the precursor (QR.pa) that generates AVM and SDQR undergoes apoptosis (Fig 6A) [36], we reasoned that this additional cell body may be the remains of the apoptotic cell. This third cell body contained a nucleus labeled by the nuclear envelope marker, was generated through cell division (indicated by the presence of the midbody), but did not transcribe TRN genes (Fig 6B – 6D). Interestingly, as the animal develops, the frequency of the three-body syncytia decreased over time (Fig 6E), implying that the third cell body could be removed (likely through engulfment by other cells) without inducing programmed cell death of the entire syncytium. This finding also suggests that the apoptotic fate of the precursor’s sister (QR.pp) may only be delayed but not reversed through the connection with the surviving AVM and SDQR neurons. In fact, when apoptosis was blocked in lem-3(−); ced-3(−) double mutants, the third cell body persisted over time (Fig 6E).

The above observation prompted us to examine cell divisions that generate one terminally differentiated neurons and one apoptotic cell. Among the four cell types we examined, AQR and PQR neurons form syncytia with their apoptotic sisters at detectable penetrance (2%–3%) through oftentimes a long intercellular bridge (especially for AQR) (Fig 6F and 6G), which may be due to the migratory nature of these neurons [36]. We also observed some rare cases where the AQR and PQR neurons were missing (they likely underwent apoptosis or failed to differentiate), suggesting that neuronal survival and differentiation could also be affected by cytoplasmic connection with cells that were scheduled to die. This may be similar to how cytokinetic bridges in mouse embryos allow the sharing of apoptotic factors to coordinate cell elimination [37]. We did not observe such loss of marker expression in the AVM-SDQR syncytia, likely because only their precursor is connected to an apoptotic sister and thus the effect might be weaker.

Similar results were obtained for OLL, ASER, and PVD neurons with the penetrance of syncytia ranging from 1.6% to 4.7% and the penetrance of cell fate marker loss below 1% (Figs 6F–6H, S9A, and S9B). It appears that connecting a postmitotic neuron with its apoptotic sister is more likely to save the sister cell from dying than to induce apoptosis (or failure in differentiation) of the entire syncytium. No statistically significant phenotype was observed for NSM neurons. To further confirm that blocking apoptosis could rescue the sister cells, we crossed some cell fate markers with ced-3(−) mutants (ced-3 codes for caspase 3) and indeed observed extra cells adjacent to AQR, PQR, and OLL neurons, which were presumably the surviving sister cells (Fig 6I).

LEM-3 safeguards the specification of many neuron types

To understand whether LEM-3 has a general function in preventing the formation of syncytia and safeguarding cell fate specification in the nervous system, we next extended our analysis to 11 other cell divisions that generate two daughter cells with distinct neuronal fates (including the CEPD-URX, SMDD-AIY, SMDV-BAG, ASJ-AUA, ADF-AWB, HSN-PHB, AFD-RMD, ADE-ADA, PDE-PVD precursor, FLP-AIZ, and RIS-DB4 sister pairs). We used fluorescent reporters to mark one of the two cell fates in a sister pair in the lem-3(−) mutants. Interestingly, in almost all cases, we observed both the formation of a syncytium with an additional cell body (presumably the sister cell) and the loss of the marker expression (indicating fate suppression likely due to the connection with the sister cell) with one being the dominant phenotype despite the low penetrance (Figs 7A–7F and S9C–S9G). One exception is the RIS neuron, where we did not observe any significant phenotype in the lem-3(−) mutants, but ICRF-193 treatment led to ~20% of syncytium formation (Figs 7F and S9H). We also used the CEPD-URX sister pair as an example to confirm that the appearance of syncytium labeled by the URX marker (gcy-37p::GFP) coincided with the loss of the CEPD marker (dat-1p::mCherry) in the same animal (Fig 7C). In fact, the penetrance of syncytium for the URX marker was similar to the penetrance of fate loss with the CEPD marker.

Lastly, since neurogenesis in C. elegans is largely nonclonal, we also analyzed terminal cell divisions that happen to generate one neuron and one nonneuronal sister. For example, the sister cell of the DVB neuron is the rectal cell, and we found that the expression of two DVB markers (unc-47p::GFP and flp-10p::GFP) was both lost in ~5% of the lem-3(−) animals, suggesting that the DVB fate may potentially be suppressed in the syncytium (Fig 7G and 7I). In the case of PVR neuron, whose sister is the hypodermal cell hyp 7, we found both syncytium and fate loss phenotypes with equal penetrance in lem-3(−) mutants. Interestingly, the extra cell connected to PVR in the syncytium appeared to be neuron-like in some animals and hypodermis-like in others based on the morphology (Fig 7H and 7I). This finding hinted that it may be possible to convert skin cells into neurons through direct cytoplasmic coupling, although this conversion was previously demonstrated by overexpressing certain transcription factors [38] or microRNAs [39]. On the other hand, just like the AVM-SDQR syncytium, two cell fates that are as distinct as neuron and skin might co-exist in the same syncytium that links the two cell bodies through an intercellular bridge.

Strains and constructs

C. elegans strains were maintained at 20 °C on nematode growth medium (NGM) plates seeded with OP50 bacteria according to previous methods [50]. Unless otherwise indicated, the phenotypes of lem-3 mutants were scored at 25 °C, which produced the highest penetrance among the conditions we tested. To generate the lem-3 knockout, we used CRISPR/Cas9-mediated gene editing to make cuts at two sites of the endogenous locus. The target sites are 5′-GTGGTGATTGCTCCGTTTGG-3′ in exon 1 and 5′-AACAATCGACGTGGAGCAGG-3′ in exon 7. We synthesized the single guide RNAs using the EnGen sgRNA Synthesis Kit (E3322V) from NEB and injected them together with 20 pmol recombinant Cas9 (EnGen S. pyogenes Cas9 NLS from NEB, M0646T) into C. elegans using pCFJ104 (myo-3p::mCherry) as a co-injection marker. We genotyped the transformants with red muscles to screen for deletions in lem-3 and obtained unk84 and unk85 alleles. Both alleles deleted the sequence between the two target sites, while unk84 also had a short insertion of 24-bp random sequences (Fig 1A). We used unk84 as the lem-3(−) mutant in most experiments. Similarly, to generate mec-3 null alleles, we targeted the 5′-GTGGACGGCATTTATCACCG-3′ sequence in exon 5 and 5′-TGAGACATTATGAGCAACGT-3′ in exon 7 and obtained unk398, unk400, unk402, and unk403 alleles. Both unk398 and unk403 deletes 2 nucleotides in exon 5 and caused frameshift; unk400 deletes most of exon 5, 6, and 7; unk402 inserts 93 nucleotides in exon 5 and caused premature stops (S4A Fig).

The gt3309(eGFP::Stag::lem-3), gt3310(eGFP::STag::lem-3[S192A S194A]), and gt3311(eGFP::Stag::lem-3[Y556A G558A]) alleles [2] were generously provided by Dr. Anton Gartner and Dr. Ye Hong. To rescue lem-3(−) mutant phenotype, we cloned a 9-kb lem-3 genomic fragment, including 6 kb sequences upstream of the start codon, the coding region, and the 3’UTR, into pUC57 and injected it into the lem-3(unk84) mutants. To create egl-17p::lem-3::GFP, we cloned the lem-3 coding region and fused it to the downstream of a 3.2 kb egl-17 promoter and upstream of GFP coding sequence and an unc-54 3’UTR. The constructs were injected into animals carrying the casIs131[egl-17p::mCherry::zen-4] transgene, which labels the midbody. To label various neuronal fates, we used an array of transgenic fluorescent reporters generated by previous studies and crossed them with the lem-3 mutants. All strains used in this study are listed in S1 Table.

To misexpress mec-3, we cloned a 2,188-bp unc-119 promoter between HindIII and BamHI sites in pPD95.75 and then inserted the mec-3 coding and 3’UTR sequence at the NotI site. The resulting construct was then injected into C. elegans gonad and integrated into the genome to form uIs158[unc-119p::mec-3]. To create transgenes that express proteins with various subcellular localizations under the TRN-specific promoters (e.g., a 1.9-kb mec-17 and a 400-bp mec-18 promoters), we cloned the respective proteins (e.g., dendra2 for photoconvertible fluorescent protein, the PH domain of PLCdelta1 fused with GFP for a membrane protein [51], GFP fusion protein with a nuclear localization signal, the ER chaperone MEC-6 for ER [52], etc.; see S1 Table for details). To label microtubules, we cloned a genomic fragment of maph-1.1 (a homolog for MAP1), including a 3.7-kb promoter, coding region and 1-kb 3’UTR and inserted GFP before the stop codon. To label actin filaments, we fused the LifeAct sequence [30] at the N-terminus of GFP and inserted it downstream of the mec-17 promoter. These constructs were injected into C. elegans gonad to form extrachromosomal arrays, some of which were integrated using gamma irradiation.

smFISH and smiFISH

To monitor transcription of endogenous genes from the two nuclei of the syncytia, we performed single-molecule RNA fluorescent in situ hybridization (smFISH) against TRN-specific genes mec-17, mec-18, and mec-3, and the sister fate-specific genes lad-2, zig-3, ast-1, and D2089.3. To detect transcription of GFP in the syncytia that express the integrated fluorescent fate reporters, we conducted smFISH against GFP. In general, we designed the FISH probes using the Stellaris RNA FISH Probe Designer tool from Biosearch Technologies (Petaluma, CA) and synthesized them with Quasar 670 dye. L1 or L2 stage animals from various strains were fixed, stained with the FISH Probe sets in the hybridization buffer (purchased from Biosearch Technologies), and then washed using the wash buffer based on a previous protocol [53]. The hybridized animals were then mounted using VECTASHIELD Antifade Mounting Medium (Vector Laboratories #H-1000) and imaged using a Leica DM8 microscope with Cy5 filter sets.

We also performed smiFISH (single-molecule inexpensive FISH) [54] against additional TRN sister fate-specific genes, including nlp-20, nlp-80, flp-12, and C15C7.4. smiFISH is a modified version of smFISH in which the primary probe binds to the target mRNA and contains a flap sequence that is bound by a secondary probe which is fluorescently labeled. We designed the primary probes using the Stellaris RNA FISH Probe Designer tool and added the X Flap sequence to the 5′ end of each probe. The X Flap sequence was 5′-CCTCCTAAGTTTCGAGCTGGACTCAGTG-3′. An equimolar mixture of the primary probe set was prepared at 100 μM in IDTE buffer (10 mM Tris-HCl, 0.1 mM EDTA, pH 8.0). Secondary probes were the reverse complement of the X Flap sequence and were labeled with Quasar 670 dye at the 5′ end. Both primary and secondary probes were synthesized by Beijing Tsingke Biotech Company Limited. To hybridize the fluorophore-labeled secondary probes to the 5′-extended primary probe sets, an annealing reaction was performed at a final concentration of 20 μM. The reaction mix consisted of 2 μl of 100 μM primary probe mix (200 pmol in total), 2.5 μl of 100 μM secondary probe (250 pmol in total), 1 μl of 10X T4 DNA ligase buffer (NEB), and 4.5 μl of water. The reaction mix was incubated in a thermocycler using the following programs: 95 °C for 3 min, decrease 0.5 °C per cycle and hold for 30 s (59 cycles), 65 °C for 3 min, decrease 0.5 °C per cycle and hold for 30 s (79 cycles), 95 °C for 5 min. Steps for C. elegans sample preparation, hybridization, washing, and image acquisition were the same as those for smFISH. All FISH probes used in this study are listed in S2 Table.

Pharmacological treatments and mcm-7 RNAi

To treat the animals with various drugs, we added ICRF-193 into the NGM plates at 10 μM, Colchicine at 1 mM, Paclitaxel (Taxol) at 100 nM, Y-27632 at 100 nM, and Latrunculin-A at 1 μM.

For mcm-7 RNAi, we obtained the bacteria expressing dsRNA against mcm-7 from the Ahringer C. elegans RNAi feeding library and grew the bacteria at 37 °C until reaching an OD600 of 0.6 and then seeded the bacteria onto NGM plates supplemented with 1 mM IPTG for overnight induction. L4 hermaphrodites were transferred to the RNAi plates and fed at 25 °C. Treating the animals with 100% mcm-7 RNAi bacteria led to sterility, so we mixed mcm-7 RNAi bacteria and OP50 of equal OD600 at 7:93 ratio to make a 7% mcm-7 RNAi bacteria solution and seeded them on the RNAi plates.

Photoactivation and photoablation

Day-one adult worms were immobilized on 10% agarose pads using polystyrene microspheres (Polysciences #00876–15; diluted 6-fold) and high-precision cover glasses (Paul Marienfeld GmbH and Co. KG #0107052). Worms were raised in the dark to avoid unnecessary photoconversion under natural lighting. As a small amount of Dendra2 is activated during worm handling, a Rhodamine fluorescence filter set was used to locate the TRN syncytium by detecting very low levels of red fluorescence. Using the green (546 nm) light to track the TRN syncytium served two purposes: 1) to avoid further depletion of the inactive Dendra2 pool, and 2) to photobleach activated Dendra2 as much as possible before photoconversion. A 405 nm laser was then used to deliver 2 s of blue light at 5% intensity to the TRN cell bodies via a Leica Infinity Scanner powered by a Widefield Compact Supply Unit on a Leica DMi8 microscope equipped with 63x water objectives. The same 405 nm laser was applied to the TRN cell bodies each time when the red fluorescence in the connected sister cells was imaged using 10% intensity of 546 nm light at various intervals. The fluorescent intensity of activated dendra2 was quantified by using imageJ.

To cut the intercellular bridge using laser, day-one adult animals were similarly immobilized, and 355-nm UV light was delivered to the middle of the bridge through a Pulsed Laser Unit attached to the Infinity Scanner on the Leica DMi8 using 63x water lenses. To minimize injury, the lowest energy that enabled the severing of the bridge was used. Animals were then rescued and placed on NGM plates for recovery and were imaged 24 and 48 hours after the axotomy. The fluorescence in both AVM and SDQR cell bodies was quantified.

Plate dropping assay and gentle touch assay

Both the plate dropping and gentle touch assays were conducted using synchronized day-one adult hermaphrodites. Strains included wild-type animals and lem-3 mutants with AVM-SDQR syncytium. Worms were transferred to the unseeded NGM plate and allowed to acclimate for 2 hours prior to testing. Responses were recorded using a Nikon SMZ745T stereomicroscope equipped with a CTS CWHC-8MB 4K microscope camera at 30 frames per second. The plate dropping assay was adapted from standard protocols to assess mechanosensory responses mediated by TRNs. Plates were subjected to mechanical perturbation by dropping a plate with standardized weight (91–95 g) from a height of 10 cm onto the imaging stage, simulating a vibrational tap to elicit behavioral responses. Each worm was tested individually and experienced 5 drops with a 50-s inter-stimulus interval. Responses were recorded immediately after each drop. Behavioral outcomes were categorized as “Reversal” (backward locomotion), “Acceleration” (forward acceleration), “No response” (continued locomotion without change in velocity), or “Pause” (halted forward movement). A minimum of 8 animals per strain were assayed, with experiments performed in triplicate on independent days.

The gentle-touch assay followed the standard protocol described by Chalfie and colleagues [55]. Gentle touch was applied alternately to the anterior and posterior body regions using eyebrow hair mounted on a glass pipette, with five anterior and five posterior touches per worm, separated by 15-s intervals. Twenty animals were tested and the response of each animal was recorded.

Image acquisition and phenotype scoring

Fluorescent imaging was conducted using a Leica DMi8 inverted microscope outfitted with a Leica K5 monochrome camera. The Leica THUNDER deconvolution system was applied to enhance image clarity and eliminate out-of-focus light in certain images. Image analysis and measurements were carried out using the Leica Application Suite X software. To score the penetrance of syncytium formation, animals were grown at 25 °C for at least two generations and then imaged at day-one adult stage. At least 200 animals from each strain were examined.

Quantification and statistical analysis

All data were plotted as mean ± SD using GraphPad Prism 8.0. For penetrance, three biological replicates were performed and the mean ± SD for the percentage of animals with the phenotype was calculated. For multiple comparisons, we performed one-way ANOVA followed by a Dunnett’s multiple comparisons test (comparison with the wild type) or a Tukey’s honestly significant difference (HSD) test (selected pairwise comparison). Student t test was used to compare only two groups. Single, double, and triple asterisks indicate statistical significance at p < 0.05, < 0.01, and < 0.001, respectively.

S1 Fig. Binucleate syncytia between TRNs and their sister cells in lem-3 mutants.

(A) The TRN-sister cell syncytia in lem-3(unk84) mutants express TRN markers, unkEx309[mec-18p::GFP] and unkEx313[mec-4p::YFP]. Arrows indicate the intercellular bridge of the syncytium. Scale bars = 10 μm. (B) The percentage of TRNs showing the syncytium labeled by various TRN markers, including uIs31[mec-17p::GFP], unkEx309, and unkEx313. (C) Similar morphology of the SDQR neuron in wild-type and lem-3(−) mutants that possess the AVM-SDQR syncytium. (D) The two cell bodies in the AVM-SDQR and PLMR-ALNR syncytia both contain histones labeled by oxTi75[eft-3p::GFP::H2B]. Arrows point to the histone-containing nuclei. “Z1” and “Z2” indicate two focal planes of the same animal. (E) The AVM-SDQR syncytium expresses various SDQR markers, including ynIs25[flp-12p::gfp], ast-1::GFP(vlc19), and otIs358[ser-2p2::GFP]. (F) The PVM-SDQL syncytium is labeled by SDQL markers, flp-12p::GFP and ast-1::GFP. Because ser-2p2p::GFP is expressed in both PVM and SDQL in the wild-type animals, it cannot be used as an SDQL-specific marker for the syncytium. As expected, the PVM-SDQL syncytium shows ser-2p2p::GFP expression. The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3004099.s001

(TIF)

S2 Fig. Repression of sister cell fate markers in ALM-BDU and PLM-ALN syncytia.

(A, B) The lack of expression (dashed circles) of BDU fate markers unkEx914[nlp-80p::GFP], unkEx917[C15C7.4p::GFP], and otIs92[flp-10::GFP] in the ALM-BDU syncytium of lem-3(unk84) mutants. The persistent expression of BDU fate markers, wgIs73[ceh-14::EGFP] (circles) and otIs358[ser-2p2::GFP], in the syncytium. Scale bars = 10 μm. “Z1” and “Z2” indicate two focal planes of the same animal. (C) The comparison of morphology of the BDU neurons labeled by otIs358 and the BDU of the ALM-BDU syncytium labeled by uIs115[mec-17p::TagRFP] in lem-3(unk84) mutants. (D, E) The absence of expression (dashed circles) of ALN fate markers rtEx332[nlp-20p::GFP], ast-1::GFP(vlc19), and otIs358[ser-2p2::GFP] in the PLM-ALN syncytia. For the case of otIs358[ser-2p2::GFP], two scenarios were shown. One shows only the PLMR-ALNR syncytium and wild-type PLML; the other shows both the PLML-ALNL and the PLMR-ALNR syncytia.

https://doi.org/10.1371/journal.pbio.3004099.s002

(TIF)

S3 Fig. Misexpression of mec-3 converts BDU and ALN into TRN-like cells.

(A) BDU neurons express the TRN marker uIs115 and do not express the BDU fate markers otIs358[ser-2p2::GFP] and otIs14[zig-3p::GFP] in the animals carrying the uIs158 transgene. SDQR expresses the TRN marker but also keeps the expression of the SDQ marker otIs358 upon mec-3 misexpression. (B) ALN neurons express the TRN fate marker uIs115[mec-17p::TagRFP] and lose expression of ALN markers otIs358[ser-2p2::GFP] and uIs130[lad-2p::GFP] in animals carrying the uIs158[unc-119p::mec-3] transgene. Wild-type expression pattern of otIs358[ser-2p2::GFP] in the tail can be found in S2B Fig for comparison. Dashed circles indicate the lack of expression. Scale bars = 10 μm. Neuron names with an asterisk mean that the neuron abnormally expresses the TRN marker upon forced expression of mec-3. (C) SDQR expresses both the TRN marker uIs115 and the SDQ marker uIs130[lad-2p::GFP] in animals carrying the uIs158 transgene.

https://doi.org/10.1371/journal.pbio.3004099.s003

(TIF)

S4 Fig. MEC-3 is required for the activation of the TRN fate in the sister cells of the TRN-sister syncytia.

(A) Various mec-3 null alleles generated in the lem-3(unk84); uIs115 background. (B) The loss of TRN fate marker mec-17p::TagRFP in mec-3(unk400) and lem-3(unk84); mec-3(unk398) animals. The ALN-PLM syncytium expressed the ALN marker lad-2p::GFP. (C) The TRN fate marker mec-17p::TagRFP was not expressed in mec-3(unk402) and lem-3(unk84); mec-3(unk403) animals. In both animals, ALM express the BDU marker zig-3p::GFP. (D, E) The TRN marker mec-17p::TagRFP was not expressed in mec-3(unk400) and lem-3(unk84); mec-3(unk398) animals. The AVM-SDQR and PVM-SDQL syncytia expressed the SDQ marker lad-2p::GFP. Scale bars, 10 μm.

https://doi.org/10.1371/journal.pbio.3004099.s004

(TIF)

S5 Fig. Endogenous expression of TRN and sister cell fate markers in wild-type TRNs and sister cells, respectively.

(A–D)smFISH against various cell fate markers was done in wild-type animals carrying the uIs31[mec-17p::GFP] transgene that marks the TRNs. (A) smFISH signals for the mRNAs of TRN genes mec-17, mec-18, and mec-3 in wild-type animals. Dashed orange circles indicate the cell bodies. The uIs31[mec-17p::GFP] transgene was used to label the TRNs. Scale bars = 5 μm. (B) Lack of smFISH signals for SDQ marker genes lad-2, ast-1, and flp-12 in AVM and PVM neurons. SDQR marker ser-2 mRNAs are not found in AVM but are found in 100% (n = 10) of PVM neurons. (C) smFISH signals for BDU fate markers nlp-80, D2089.3, and C15C7.4 are not observed in ALM neurons. ser-2 smFISH signals are found in 62.5% (n = 40) of ALM neurons. (D) smFISH signals for ALN fate markers nlp-20, ast-1, and lad-2 are not found in PLM neurons. ser-2 smFISH signals are found in 60% (n = 40) of PLM neurons. (E–H) smFISH against various cell fate markers was done in wild-type animals carrying the otIs358[ser-2p2::GFP] transgene that marks the sister cells of TRNs. (E) smFISH signals for the TRN fate markers mec-17, mec-18, and mec-3 are not found in the sister cells of TRNs in the wild-type animals. Dashed orange circles indicate the cell bodies. Scale bars = 5 μm. (F) smFISH signals for the SDQ marker genes lad-2, ast-1, and flp-12 in SDQL/R neurons. ser-2 only serves as a SDQR marker, although its mRNAs are found in both SDQL and SDQR. (G) smFISH signals for BDU fate markers nlp-80, D2089.3, C15C7.4, and ser-2 in BDUL/R neurons. (H) smFISH signals for ALN fate markers nlp-20, ast-1, lad-2, and ser-2 in ALNL/R neurons.

https://doi.org/10.1371/journal.pbio.3004099.s005

(TIF)

S6 Fig. Additional evidence for the activation of endogenous fate markers in the two cell bodies of the syncytia.

(A) smFISH signals for the mRNAs of TRN genes (mec-17, mec-18, and mec-3) and SDQL genes (lad-2, ast-1, and flp-12) in the two cell bodies of the PVM-SDQL syncytia in lem-3(unk84) mutants (n ≥ 10). Dashed circles indicate the cell bodies. “Z1” and “Z2” indicate different focal planes of the same animal. Arrows point to the intercellular bridge. Scale bars = 5 μm. (B) ser-2 smFISH signals are absent in the BDU cell body for 13% of the ALM-BDU syncytia (n = 15). (C) ser-2 smFISH signals are absent in the ALN cell body for 38% of the PLM-ALN syncytia (n = 21). (D) smFISH signals for GFP mRNAs in the lem-3(unk84); uIs31[mec-17p::GFP] strain. Percentage of cells having the shown smFISH staining patterns (n ≥ 10).

https://doi.org/10.1371/journal.pbio.3004099.s006

(TIF)

S7 Fig. The intercellular bridge enables the exchange of cytoplasmic content between the TRNs and their sister cells in the syncytia.

(A) Photoconversion of dendra2 in the ALM cell body of the ALM-BDU syncytium led to the increase of red fluorescence in BDU neuron (dashed circle). The times indicate the time points after photoconversion. Scale bars = 5 μm. (B) Quantification of the red fluorescence in both ALM and BDU cell bodies throughout the course of the photoconversion experiment. (C, D) After the connection between ALM and BDU cell bodies was cut using a laser, the photoconversion of dendra2 in ALM no longer caused the increase of red fluorescence in BDU neuron. (E, F) After cutting the intercellular bridge between AVM and SDQR, photoconversion of dendra2 in AVM did not lead to the increase of red fluorescence in SDQR. lem-3(unk84); unkEx271[mec-17p::dendra2] was used in these experiments. The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3004099.s007

(TIF)

S8 Fig. Intercellular bridge and nuclear envelope connection in the TRN-sister cell syncytia are maintained throughout development and aging.

(A) Quantification of the distance between the two nuclei of the AVM-SDQR syncytia at different developmental stages. lem-3(unk84); uIs115[mec-17p::RFP]; ccIs4810[lmn-1p::lmn-1::GFP] was used to label both the cytosol and nuclear envelope. The syncytia with nuclear envelope (NE) links were marked in red in the plot. (B) Representative images of syncytia with a long NE link or with unseparated nuclei at 72 hours post L1 arrest recovery are shown. Scale bars = 10 μm. (C) Intercellular bridge (arrows) of ALM-BDU and AVM-SDQR syncytia at the day 5 adult stage in lem-3(unk84) mutants. The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3004099.s008

(TIF)

S9 Fig. Formation of syncytia between neurons and their apoptotic sisters and between pairs of sister cells in lem-3(−) mutants.

(A) The sister cell of PVDL/R is apoptotic. In the lem-3(unk84); wyIs592[ser-2p3::myr::GFP] animals, which labels the PVD neuron, both syncytia of PVD with an extra cell (“syn”) and the loss of PVD labeling (“loss”) are found. For the former phenotype, arrows point to the bridge and for the latter phenotype, a dashed circle indicates the absence of GFP signal. Scale bars = 5 μm. “Z1” and “Z2” indicate different focal planes of the same animal. (B) The syncytia and the loss of fate marker phenotypes for NSM neurons in the lem-3(−) mutants. (C–G) Representative images of syncytia (“syn”) and fate marker loss (“loss” and dashed circles) phenotypes for URXL/R, ADFL/R, AIYL/R, FLPL/R, and HSNL/R neurons in lem-3(unk84) mutants. In syncytia, sister cells abnormally expressing markers are labeled with yellow names; arrows indicate the intercellular bridge. Quantification of penetrance is shown in Fig 7F. Scale bars = 5 μm. “Z1” and “Z2” mean two focal planes of the same animal. (H) The syncytium phenotype for the RIS neuron in lem-3(unk84) mutants treated with 10 μM ICRF-193.

https://doi.org/10.1371/journal.pbio.3004099.s009

(TIF)