Regulation of Gliogenesis by lin-32/Atoh1 in Caenorhabditis elegans

C. elegans genetics

C. elegans strains were grown on nematode growth media (NGM) plates with E. coli OP50 as their food source. Animals were grown according to standard methods at 20° unless otherwise stated (Brenner 1974). Wild type worms were of the Bristol N2 strain. All transgenes, strains and DNA constructs used are described in Table S1. yadIs46 (Pf16f9.3::GFP) was used to visualize AMsh cells while Pttx-3::RFP was used as a coinjection marker.

The recessive allele lin-32(yad67) was isolated from a visualized EMS mutagenesis screen of over 4000 haploid genomes and was the only allele isolated in the screen with the ectopic AMsh phenotype. During backcrossing, we noticed that yad67 was on the left arm of chromosome X, and whole genome sequencing revealed that lin-32 was the most likely candidate in the region. The mutation was confirmed through rescue experiments.

Cloning and constructs

All DNA expression constructs were generated using Gateway cloning technology (Invitrogen, Carlsbad, CA) and subsequently sequenced. lin-32, ngn-1 and cnd-1 cDNA were all amplified from a homemade genomic DNA pool. Promoters of lin-32 (Forward cgccacccgattagagactag; Reverse ggttggtctgactgaaaacgacgatgtgtgag), ngn-1 and cnd-1 were amplified from genomic DNA using the 2kb sequence upstream of the transcription start site. Murine Atoh1 cDNA was amplified from a home made cDNA pool. In general, plasmid DNAs used in this study were injected at a concentration of 1-50ng/µL with a Pttx-3::RFP co-injection marker injected at a concentration of 50ng/µL.

Microscopy

Representative images were acquired with a Zeiss LSM700 confocal microscope using a Plan-Apochromat 40x/1.4 objective. Worms were immobilized using 1.5% 1-phenoxy-2-propanol (TCI America, Portland, OR) in M9 buffer and mounted on 5% agar slides. 3D reconstructions were done using Zeiss Zen software as maximum intensity projections. A Zeiss Axio Imager 2 microscope equipped with Chroma HQ filters was used to score AMsh number defects. Any animal with more than the wild type AMshL and AMshR glia were scored as having the defect. Each condition represented 3 experiments of at least 50 D1 animals each that were picked at random from the culture plate unless otherwise noted, in accordance with previous literature in C. elegans. Cell numbers were quantified by counting the number of red nuclei labeled by Pf16f9.3::mCherry::H2B and confirmed by referencing the whole cell morphology labeled by Pf16f9.3::GFP.

For tracking of AMsh cell number during larval development, 10 individual L4 worms per genotype were scored under the Zeiss Axio Imager 2 microscope without 1-phenoxy-2-propanol and recovered from the agar slide. They were scored again when they reached the D1 adult stage.

Statistical analysis

Data were analyzed using one-way ANOVA followed by Tukey’s HSD test, Chi-square test, two-tailed Student’s t-test, Spearman’s Rank-Order Correlation and Pearson Correlation in Graphpad Prism (Graphpad Software, La Jolla, CA).

Data availability

Further information and requests for resources should be directed to and will be fulfilled by the Lead Contact, Dong Yan (dong.yan@duke.edu). C. elegans strains and plasmids generated in this study are available from the lead contact without restriction. Supplemental Figures S1 and S2 as well as Strain Table S1 are available at figshare: https://doi.org/10.25387/g3.12650363.

Mutants of Neurog1 and NeuroD1 homologs possess additional glia cells

To study the molecular mechanisms underlying gliogenesis, we focused mainly on the AMsh glial cells, which are a pair of glia that ensheath the dendrites of sensory neurons in the amphid sensilla, the primary chemosensory organ (Oikonomou and Shaham 2011). AMsh glia are critical for the neurons they envelop to function, and they are easy to visualize in vivo (Oikonomou and Shaham 2011). To determine whether C. elegans shares similar mechanisms with mammals during gliogenesis, we tested the functions of the homologs of two well-studied proneural genes Neurog1 and NeuroD1 during AMsh genesis. We found that loss of function mutations of C. elegans Neurog1 and NeuroD1 homologs, ngn-1 and cnd-1, caused approximately 20% of ngn-1(ok2200) and 30% of cnd-1(gk718) animals respectively to possess more than the invariant two AMsh cells observed in wild type animals when examined during the day 1 adult stage (D1). The ngn-1(ok2200);cnd-1(gk718) double mutants had around a 45% mutant phenotype, which is significantly higher than in either of the single mutants (Figures 1A, 1B and 1C). Furthermore, cnd-1 and cnd-1;ngn-1 mutants had higher mean numbers of AMsh cells than wild type animals, while the difference was not statistically significant in ngn-1 animals likely due to the low penetrance of phenotypes (Figure 1D). These results demonstrate that ngn-1 and cnd-1 function in parallel during AMsh formation, and the relatively low penetrance of phenotypes also suggest additional factors are involved in AMsh genesis as well (Figure 1C). Consistent with loss of function, expression of ngn-1 and cnd-1 under their own promoters strongly rescued the ectopic AMsh phenotype in their respective mutants (Figure 1E). Thus cnd-1 and ngn-1 are important in regulating glial specification, consistent with the role of Neurog1 and NeuroD1 in mammals (Sun et al. 2001; Morrow et al. 1999; Tomita et al. 2000), and supports that C. elegans may share common mechanisms with mammals during gliogenesis.

A forward genetic screen reveals that loss-of-function of lin-32 results in ectopic AMsh glia

Finding a function of ngn-1 and cnd-1 in regulating AMsh number, we decided to carry out an unbiased forward genetic screen targeting any mutants that possessed additional AMsh glia. We isolated a mutant, yad67, that possessed more than 2 cells labeled by the AMsh marker Pf16f9.3 (Figures 2A and 2B). The yad67 mutation was identified to affect the proneural gene lin-32, a homolog of Atoh1, consisting of a point mutation in the splice donor of its second intron (Fig. S1A). Rescue experiments confirmed lin-32 to be the gene involved, as expression of lin-32 under its own promoter consisting of its upstream 2kb sequence was able to fully rescue the ectopic AMsh phenotype in yad67 mutants (Figure 2C). Interestingly, expression of murine Atoh1 under the lin-32 promoter was also able to fully rescue the mutant phenotype, suggesting that this regulation of gliogenesis may be conserved (Figure 2C). Further testing of other alleles of lin-32 (tm2044, tm1446 and u282) showed that they all recapitulated the mutant AMsh phenotype observed in yad67 animals (Figures 2D, S1A). In particular, tm2044 is likely a null allele of lin-32 due to it containing a deletion spanning most of the gene, including part of the basic helix-loop-helix (bHLH) domain vital for regulating transcription. Thus, the similar phenotypes and penetrance between yad67 and tm2044 suggest that yad67 is likely a null allele of lin-32 (Figures 2D, S1A). The tm2044 allele will be used for most genetic analyses unless otherwise stated due to its similar phenotype with yad67 and the ease of genotyping. lin-32(tm2044) mutants possessed variable numbers of AMsh glia, with numbers ranging from 2 to 7 cells (Figure 2E). Furthermore, all the cells labeled by the Pf16f9.3 marker were also colabeled by two other AMsh markers Pf53f4.13 and Pt02b11.3 (Figs. S1B-S1D), supporting the conclusion that these additional cells are AMsh glia.

To determine whether this phenotype is limited to only AMsh cells, we also examined the amphid socket (AMso) cells, which are another type of glia that come from a different cell lineage (Figure 2A) (Sulston et al. 1983). A similar mutant phenotype was observed in the AMso cells, where approximately 48% of lin-32(yad67) animals possessed more than the usual pair of AMso cell observed in wild type animals (Figures 2F and G).

Given the important role of lin-32 in gliogenesis, we used AMsh glia to test other genes that have been shown to function together with lin-32 in regulating neuron fate determination and organogenesis, including hlh-2 which can heterodimerize with lin-32 to regulate neuronal specification (Portman and Emmons 2000), the parallel storkhead transcription factor ham-1(n1438) and (tm4595) (Zhu et al. 2014), gain of function (n302) and loss of function (n941) alleles of the potential upstream lin-12/Notch (Greenwald et al. 1983; Sundaram and Greenwald 1993; Zhong and Sternberg 2006), and a Msx homeobox homolog vab-15(u781) found to also regulate hypodermis to neuron transformations similar to lin-32. None were found to have significant ectopic AMsh cell phenotypes, suggesting they may not be involved in the formation of AMsh cells (Fig. S1E). However, hlh-2(tm1768) is not a null allele and vab-15(u781) is not a confirmed null, so it is possible that these genes may still play a role.

lin-32 suppresses a glial fate in different lineages during early embryogenesis

As lin-32 controls cell type determination in different neural lineages (Rojo Romanos et al. 2017; Zhao and Emmons 1995), we hypothesized that lin-32 would likewise function early in development to suppress an AMsh glial fate. As expected, expression of lin-32 in AMsh cells (Pf16f9.3), head neurons (Pdyf-7), AMso and other socket cells (Pglr-2), neurons including the AMsh sister URB neurons (Pflp-3), hypodermis (Pdpy-7), and pharyngeal muscles (Pmyo-2) were not able to rescue the additional AMsh phenotype (Figure 3A), likely due to these promoters turning on after the ectopic glial cells already arise. This is supported by transcriptomic data showing that AMsh, AMso and URB cells show little to no expression of lin-32 (Fig. S1F) (Packer et al. 2019). One earlier promoter is Plin-26, a known regulator of glial and hypodermal cell specification that is also required for proper AMsh cell specification (Labouesse et al. 1996). It is expressed in several cells of the AB lineage starting from around 100 min into embryonic development including the parent cell of AMsh (Packer et al. 2019). However, expression of Plin-26::lin-32 was not able to rescue the additional glia phenotype in lin-32(yad67) animals (Figure 3a), suggesting that lin-32 is required earlier or in a different set of progenitor cells.

Since many types of vertebrate glial cells have the ability to divide during development as well as after injury (Horner et al. 2000; Kornack and Rakic 2001; Fields and Burnstock 2006; Rusznák et al. 2016; Noctor et al. 2001), an alternative explanation is that the ectopic AMsh cells may emerge through additional proliferation. However, we found that the number of AMsh cells did not change when their number was traced in individual mutant animals from late larva to adults (Fig. S2A). Consistent with this finding, when the penetrance of the additional glial phenotype was quantified in L1 larval stage and D1 adult lin-32(yad67) animals, it was found that there was no significant difference in the penetrance between L1 and D1 worms (Figure 3B). Further supporting this, the additional AMsh glia can already be observed late in embryogenesis, shortly after the AMsh reporter Pf16f9.3::GFP turns on (Figure 3C). Use of an integrated rescuing lin-32::GFP reporter (Yi et al. 2000) showed that lin-32 is expressed in different cells before the comma stage, when the AMsh cells begin to develop, and is not detectable in the AMsh cells when the Pf16f9.3 marker turns on (Figure 3D), nor is it present in the AMsh cells in D1 adults (Fig. S2B). These results suggest that lin-32 functions earlier to inhibit a AMsh glial fate rather than preventing glial proliferation.

The proneural genes lin-32, ngn-1, and cnd-1 restrict AMsh formation through independent means

Analysis of the penetrance of the additional AMsh phenotype in all the lin-32, ngn-1, and cnd-1 double and triple mutant combinations show more severe phenotypes, which suggests that they function independently of each other to regulate AMsh cell number (Figure 3E). Furthermore, the mean number of AMsh cells increases from the single mutants to the double or triple mutants (Figure 3F). However, no significant increase in mean AMsh cell number was detected between the lin-32;cnd-1 double mutant and the triple mutant despite the increase in penetrance, which could be due to a saturation of the cells that could be affected by these transcription factors (Figure 3F). On the other hand, loss of function of either cnd-1 or ngn-1 both lead to an increase in the number of cells expressing the Plin-32::GFP transcriptional reporter during the bean stage of embryogenesis, suggesting possible restriction of lin-32 expression by cnd-1 and ngn-1 (Fig. S2C). Loss of function of either lin-32 or cnd-1 did not significantly affect ngn-1 expression though (Fig. S2D). While there may be interactions between these transcription factors, they may affect different cell lineages or function during different time periods to restrict an AMsh glial fate.

Dorsal CEPsh cells mis-differentiate into AMsh cells in lin-32 mutants

AMsh cells and amphid neurons extend their dendritic tips through a process of retrograde extension, where the extracellular proteins DEX-1 and DYF-7 anchor the cell at the anterior end while it migrates posteriorly (Heiman and Shaham 2009). Thus, in dyf-7 loss of function mutants, the amphid neurons and AMsh glia exhibit a process extension defect where it fails to reach the tip of the nose (Heiman and Shaham 2009). We found that all dyf-7 mutants exhibit defects in process extension (Figure 4A). Interestingly in lin-32;dyf-7 double mutants, some animals actually possess AMsh cells that properly extend their processes to the nose and no process extension defects were observed in lin-32 mutant animals (Figure 4A). When they were further divided into animals with 2 AMsh cells and those with more, it was found that the group with more than 2 AMsh cells had over 60% more animals with some AMsh that reach the nose tip (Figure 4B). While no AMsh cells adhere in the dyf-7 single mutants, a small percentage of AMsh cells adhere to the nose tip in lin-32;dyf-7 mutants with only two AMsh cells, which are likely the original AMsh cells. It could be that loss of lin-32 may also affect the cell expression profile of these original AMsh cells as well. It is also possible that at least some of the ectopic AMsh cells may originate from a different cell type or lineage that utilizes a different mechanism for process extension (Cebul et al. 2020).

Each AMsh cell arises from an asymmetric division that yields a URB neuron and an AMsh glia (Fig. S1F). Thus, it is possible that the ectopic AMsh cells in the lin-32 mutants arise from a failure of the parent cell to divide asymmetrically or that the potential URB cell fails to take on a neuronal fate. Next, lin-32(yad67) animals were colabeled with the Pf16f9.3::GFP AMsh marker and the Pflp-3::mCherry marker that labels 3 pairs of neurons including the URB neurons (Li et al. 1999). It was found that while there was a variable reduction of URB neurons in lin-32 mutants, URB neurons alone can not account for all the ectopic AMsh cells, as some animals with more than 2 AMsh cells still have both URB cells (Fig. S2E). Furthermore, many animals possess more than 4 AMsh cells (Figure 2E), suggesting that these ectopic AMsh cells may also come from other lineages. In addition, loss of function of unc-86, a gene important for the terminal differentiation of URB neurons (Finney and Ruvkun 1990; Zhang et al. 2014), resulted in an alteration of cell identity of the neurons including the URB, but the AMsh number remained unchanged (Fig. S2F). This suggests that the inability to adopt a neuronal identity does not necessarily cause the cell to default to a glial fate, and lin-32 likely suppresses a glial fate while independently promoting a neuronal one, similar to Neurog1 (Sun et al. 2001).

To determine whether the ectopic AMsh cells arise from lin-32-expressing cell lineages, cell death was induced in lin-32-expressing cells of wild type and lin-32 mutant animals by overexpressing the apoptotic caspase CED-3 under the lin-32 promoter (Ellis and Horvitz 1986; Shaham and Horvitz 1996). Overexpression of Plin-32::ced-3 did not significantly change the number of AMsh glia in wild type animals, suggesting that these wild type AMsh glia arise from outside of lin-32-expressing lineages (Fig. S1F and S2G). In contrast, there was a significant reduction in AMsh number in lin-32 mutant animals overexpressing Plin-32::ced-3, but did not fully remove all ectopic glia. This suggests that the ectopic glia at least in part arise from cells of the lin-32 lineage, though there may also be up to roughly 50% of ectopic cells that do not arise from lin-32-expressing lineages (Fig. S2G).

In lin-32 mutants, it was observed that roughly 27% of animals had 1-2 missing dorsal CEPsh glial cells (Figure 4C), which are derived from lin-32-expressing lineages (Murray et al. 2012; Packer et al. 2019), but the ventral two CEPsh cells that derive from cell lineages that very weakly express lin-32 are still present in lin-32 mutants (Fig. S1F). Thus, it is possible that ectopic AMsh cells may originate from cells originally fated to become dorsal CEPsh cells. In support of this, coexpression of the CEPsh marker Phlh-17::GFP with the AMsh marker Pf16f9.3::mCherry in lin-32 mutants revealed that certain ectopic AMsh cells also express the CEPsh marker at early developmental stages (Figure 4D). Furthermore, when CEPsh and AMsh cell numbers were quantified in lin-32 mutants, there was a strong negative correlation between CEPsh and AMsh cell number (rho=-0.62, r=-0.65) (Figure 4E). Thus, these ectopic AMsh cells may arise at the expense of the distantly related CEPsh cells. These results suggest that lin-32 can function to specify glial fate among cells that come from distant cell lineages.