Regulatory architecture controlling terminal differentiation of an interoceptive paraneuron in Caenorhabditis elegans

ABSTRACT

Interoceptive paraneurons are neuron-like cells located within internal epithelial cell surfaces that sense internal stimuli to evoke specific behavioral or physiological responses. The elucidation of terminal differentiation programs of paraneurons is expected to provide insights into how epithelial cells acquire neuron-like features during development and possibly also over evolutionary time. We define here the transcriptional programs that control the terminal differentiation of an interoceptive paraneuron class in the nematode Caenorhabditis elegans, called uv1. We show that, as in canonical neurons, the neuron-like secretory features of uv1 are controlled by a combination of CUT homeobox genes, while the combinatorial terminal gene battery that defines the unique functional features of uv1 is jointly controlled by a combination of at least three transcription factors: a LIM homeodomain (LIN-11), a SoxD (EGL-13) and a Pax (EGL-38) family protein. These factors act in a terminal selector-type manner to jointly co-regulate the many distinct uv1-paraneuron specific molecular features, such as sensory and neuromodulatory receptors, neuropeptides, and tyramine synthesis machinery. Our findings demonstrate notable similarities in the dichotomous architecture of gene regulatory programs of neurons and paraneurons.

Summary: The similarity of the regulatory architecture of neurons and paraneurons revealed using cell identity markers to characterize the neuronal features of Caenorhabditis elegans uv1 paraneurons in more detail.

INTRODUCTION

Paraneurons were defined by Fujita in the 1970s as cells that do not display an obvious neuronal morphology but that act in a neuron-like manner, being depolarized by external or internal signals and relaying these signals to the nervous system to control specific behavioral and physiological responses (Fujita, 1977; Fujita et al., 1988; Hobert, 2025). One prominent type of paraneurons are interoceptors located within internal epithelial surfaces, such as airways, the circulatory system or the intestine, that sense specific internal cues and relay them to the nervous system. The functional similarities of neurons and paraneurons raises intriguing questions about what the definition of a neuron is, how cells acquire neuronal features during development and how neurons came into being in an evolutionary context, with paraneurons perhaps being illustrative of an intermediary step in the evolutionary trajectory of neurons (Hobert, 2025). One way to explore these issues is to (1) more precisely define the molecular composition of paraneurons and (2) ask whether the gene regulatory architecture that controls the paraneuronal expression of such molecular features shares any conceptual similarity with that of neurons.

The nematode Caenorhabditis elegans is particularly well suited to address these types of questions. First, cells that can be classified as paraneurons exist in C. elegans (Hobert, 2025). Second, the differentiation programs of neurons are particularly well defined throughout the C. elegans nervous system, hence lending themselves to a deep comparison with paraneuronal differentiation programs that remain to be elucidated. Through the use of classic genetic loss-of-function analysis, several lessons have been learned about terminal differentiation programs in neuronal cell types of C. elegans: First, rather than being regulated in a piecemeal manner, large cohorts of terminal identity markers of a neuron are co-regulated by cooperating sets of transcription factor(s), called terminal selectors (reviewed by Hobert, 2016). Terminal selectors do not act in isolation but cooperate in neuron type-specific combinations, often acting in obligatory heteromeric complexes (reviewed by Hobert, 2016). Second, many terminal selectors are homeodomain proteins, a possible reflection of their diverse protein-protein interaction surfaces (Hobert, 2021; Reilly et al., 2020). Third, while individual neuron type-specific gene batteries are controlled by terminal selectors, the expression of pan-neuronal identity features, which are shared by all neurons (such as synaptic vesicle proteins or neuropeptide processing machinery), are regulated by pan-neuronally expressed CUT homeodomain proteins that act in parallel with neuron type-specific terminal selector combinations (Leyva-Diaz and Hobert, 2022; Stefanakis et al., 2015). Hence, neuronal differentiation programs are characterized by a dichotomous gene regulatory architecture, in which two distinct gene regulatory routines act in parallel to control generic (pan-neuronal) and neuron type-specific features. We explore here whether the terminal differentiation programs of a paraneuron share a similar gene regulatory architecture.

The cells that fit the paraneuron definition most clearly in C. elegans are the four uv1 cells that derive from an epithelial cell layer in the uterus of the worm (Newman et al., 1996) (Fig. 1A). uv1 cells are interoceptors that deploy TRP channels to sense internal pressure, triggered by the presence of eggs in the uterus, which stretches these epithelial cells (Collins et al., 2016; Jose et al., 2007; Yan et al., 2025) (Fig. 1A). In response to mechanosensory signals, uv1 cells utilize a combination of neuropeptides, as well as the monoamine tyramine, directly synthesized by uv1, to modulate the activity of the HSN neurons, which directly innervate vulval musculature (Banerjee et al., 2017; Collins et al., 2016; Jose et al., 2007; Yan et al., 2025) (Fig. 1A). uv1 activity is likely also modulated by other signals, as inferred by the expression of monoaminergic, GABAergic and cholinergic metabotropic receptors (Fernandez et al., 2020), as well as other orphan GPCRs (Vidal et al., 2018).

Fig. 1.

uv1 displays neuron-like features. (A) Overview of structural elements of the uterus pertinent to the uv1 cells. Adapted from Newman et al. (1996), where it was published under a CC-BY license (https://creativecommons.org/licenses/by/4.0/) and from Collins et al. (2016). (B) Oblique dorsal 3D reconstruction view of uv1 cells from the electron microscopy (EM) dataset. Also see Movies 1 and 2. (C,D) EM from the vulval region of late L4 C. elegans with pseudo-color overlay highlighting the uv1 (violet), the vulval muscles 1 (orange) and vulval muscles 2 (red). Also see Movies 1 and 2. Since archived EM data lack magnification metadata, no scale bar is included for these images. We estimate the total field of view as 32 µm x 17 µm for C and 10 µm x 5 µm for D. (D) Area outlined in C showing the junction of contralateral uv1 cells. Vesicles (arrowheads and pseudo-colored) are seen at various locations along the length of uv1 processes. At least one vesicle is seen in 27/90 sections from which we reconstructed uv1. Also see Movies 1 and 2. (E) Fluorescent reconstruction of sri-5^prom::gfp(otEx6406)+ida-1^prom::rfp(vsIs269) transgenic reporter strain. A, anterior; P, posterior; D, dorsal; V, ventral; L, left; R, right. Scale bar: 10 μm in E.

Previous work has provided substantial insights into the earlier patterning events that generate uterine and vulval cell types. Together with their sister cells, the utse uterine epithelial cells, the four uv1 cells are daughters of the π cell precursors (Gupta et al., 2012; Newman et al., 1996). uv1 fate is induced in a subset of the π cell descendants via an EGF signal emanating from neighboring vulval epithelial cells (VulF) (Chang et al., 1999; Gupta et al., 2012; Newman et al., 1996). Via an EGF receptor and a let-60/Ras-dependent signaling cascade, this EGF signal induces the expression of the Pax transcription factor egl-38 (Pax2/5/8 homolog) in the nascent uv1 cells (Chamberlin et al., 1997; Chang et al., 1999; Rajakumar and Chamberlin, 2007; Webb Chasser et al., 2019). egl-38 is required for the induction of several terminal markers of uv1 identity (neuropeptides nlp-2 and nlp-7, as well as ida-1) (Rajakumar and Chamberlin, 2007; Webb Chasser et al., 2019). To what extent egl-38 affects uv1 differentiation is not presently known, nor is it known whether egl-38 is only transiently required to induce uv1 or is required after initial differentiation to maintain stable uv1 identity.

Two other transcription factors, the lin-11 LIM homeobox gene and the SoxD-type egl-13 gene, are also expressed in uv1 cells (Hanna-Rose and Han, 1999; Newman et al., 1999). Neither of these transcription factors appears to be required for expression of the other and they may each be independent targets of EGFR signaling in uv1 (Hanna-Rose and Han, 1999; Newman et al., 1999; Rajakumar and Chamberlin, 2007). Due to the absence of proper differentiation markers at the time, uv1 differentiation has not been studied in lin-11 mutants. Similarly, while it is known egl-13 (SoxD homolog) affects π precursor cell proliferation and the proper cell fusion of the π-derived utse cells (Cinar et al., 2003), the impact of egl-13/SoxD on the differentiation of π-derived uv1 cells is not known either.

In this paper, we deploy a panel of cell identity markers to characterize the neuronal features of the uv1 paraneurons in more detail. We reconstruct the morphology of these cells and show that, as in neuronal cell types, the pan-neuronal features of uv1 are controlled by CUT homeobox genes. We show that the egl-38, lin-11 and egl-13 transcription factors act akin to terminal selectors of neuronal cell types in that they co-regulate the entire battery of uv1-specific neuron-like features. Hence, our analysis not only provide more insights into the differentiation program of a paraneuron but also illustrates striking similarities between neuronal and paraneuronal differentiation programs.

RESULTS

Morphology of uv1 cells

The morphology and shape of uv1 cells have not been described in detail before. We reconstructed the morphology of the four uv1 cells using available electron micrographical images (Fig. 1B-D; Movies 1 and 2), as well as optical reconstruction of confocal image stacks of a transgenic strain that expresses an uv1-restricted marker, sri-5::gfp (Vidal et al., 2018), and ida-1::rfp, a marker that is expressed in the entire egg-laying circuit, including uv1, VC4/5 and HSN (Banerjee et al., 2017; Zahn et al., 2001) (Fig. 1E). Together, these different visualization approaches reveal a neuron-like morphology of the uv1 cells, extending thin processes along the lining of the uterine opening to the vulva. At this point, it remains unclear whether it may be the stretch of these thin processes or pressure on the uv1 soma that can be made responsible for the mechanosensory function of uv1. The EM data also suggest the presence of vesicles along the length of uv1 processes (Fig. 1D). However, the vesicles do not have the appearance of dense-core vesicles, and they are not clustered at specific release site. Taken together, the interoceptive function of uv1 and its unusual morphology but apparent ability to communicate with other neurons allow the uv1 cells to be classified as ‘paraneurons’, even though their previous categorization as ‘neuroendocrine’ remains equally valid.

uv1 expresses pan-neuronal features in a CUT gene-dependent manner

Previous reporter gene studies, using fosmid-based reporter transgenes had already noted expression of several, pan-neuronally expressed synaptic vesicle and neuropeptide-processing/secretory proteins in the uv1 cells (Stefanakis et al., 2015). We corroborated this observation using both improved reagents (engineered reporter alleles that tag the endogenous gene locus) and additional markers (Table 1). These include two CRISPR/Cas9-engineered reporter alleles that we generated specifically for this study, reporting on expression of unc-13 and unc-31 (see Materials and Methods). We found that markers of the synaptic vesicle cycle [unc-18, unc-13, ric-4 (SNAP25 homolog) and rab-3], vesicular transport [unc-104 (kinesin homolog)] as well as genes involved in neuropeptide processing and signaling [egl-3 (PC2 homolog) and unc-31 (CAPS homolog)] are expressed in uv1 (Fig. 2A). We also found that the voltage-gated calcium channel, unc-2, for which we generated a novel reporter allele, is expressed in uv1. The NeuroPAL reporter transgene, which contains the so-called UPN pan-neuronal driver, a composite of cis-regulatory elements from several pan-neuronal genes (Yemini et al., 2021), is also expressed in uv1 (Fig. 2A).

Table 1.

uv1 cell identity markers used in this study

Expression category	Gene	Protein type	Functional category	Reagent*
Subset of cells in nervous system	ocr-4	TRP channel	Mechanoreceptor	Transgene
	ocr-2	TRP channel	Mechanoreceptor	Transgene
	sri-5	GPCR	Orphan receptor	Transgene
	gar-2	ACh receptor	Neurotransmitter receptor	Transgene
	dop-5	Dopamine receptor	Neurotransmitter receptor	Transgene
	kcc-2	K/Cl co-transporter	Ion channel	Transgene
	ida-1	IA-2/phogrin	Neuropeptide signaling	Transgene
	flp-11	Neuropeptide	Neuropeptide signaling	Transgene
	flp-22	Neuropeptide	Neuropeptide signaling	Transgene
	nlp-2	Neuropeptide	Neuropeptide signaling	Engineered reporter allele
	tdc-1	Tyrosine decarboxylase	Neurotransmitter synthesis	Engineered reporter allele
	lin-11	LIM homeobox	Transcriptional regulator	Engineered reporter allele
	egl-38	Prd type	Transcriptional regulator	Engineered reporter allele
	egl-13	Sox type	Transcriptional regulator	Engineered reporter allele
Pan-neuronal	unc-18	Sec1/Munc18	Synaptic vesicle cycle	Transgene
	unc-13	Munc13	Synaptic vesicle cycle	Engineered reporter allele
	rab-3	Small GTPase	Synaptic vesicle cycle	Transgene
	ric-4	SNAP25	Synaptic vesicle cycle	Engineered reporter allele
	unc-104	Motor protein	Vesicular transport	Transgene
	unc-2	CaV2 calcium channel	Excitability	Engineered reporter allele
	egl-3	Proprotein convertase	Neuropeptide signaling	Engineered reporter allele
	unc-31	CAPS	Neuropeptide signaling	Engineered reporter allele
	ceh-44	CUT homeobox	Transcriptional regulator	Engineered reporter allele
	ceh-48	CUT homeobox	Transcriptional regulator	Engineered reporter allele

*Engineered reporter allele indicates a fluorophore inserted into endogenous locus via CRISPR/Cas9 genome engineering. See Materials and Methods for allele, transgene and strain names.

Fig. 2.

Expression and regulation of pan-neuronal markers in uv1 cells. (A) Expression of pan-neuronal markers in uv1 cells. Reporter genes used are CRISPR/Cas9-engineered reporter alleles for unc-2(ot1614[unc-2::SL2::gfp::h2b]), egl-3(syb4478[egl-3::SL2::gfp::h2b]), unc-13(syb6325[unc-13::SL2::gfp::h2b]), unc-31(syb6138[unc-31::SL2::gfp::h2b]) and ric-4(syb2878[ric-4::SL2::gfp::h2b]), and transgenic reporters for unc-18(otEx5925), unc-104(otEx4555), rab-3(otIs498) and UPN(otIs669). (B) Expression of pan-neuronal CUT homeobox genes ceh-44 and ceh-48 in uv1 cells. Reporter genes used are CRISPR/Cas9-engineered translational gfp reporter fusions for ceh-44(ot1015[ceh-44::gfp]) and ceh-48(ot1126[ceh-48::gfp]). (C) ric-4(syb2828) reporter expression in CUT-site deleted allele (ot1123 ot1179) (top panel). egl-3(syb4478) reporter expression in ceh-38(tm321), ceh-44(tm1028) and ceh-48(tm6112) CUT homeodomain triple mutant (bottom panel). (D) Quantification of ric-4 reporter signal from C; statistical analysis was performed using Fisher's exact test (left) and Kruskal–Wallis test (right), ****P<0.0001. ‘ON’ and ‘OFF’ refer to reporter signal. Animals in all panels were imaged at the young adult stage and uv1 cells were identified by colocalization with a tdc-1 or ida-1 red marker in the background (example shown in Fig. 3). All images are approximate lateral views with anterior to the left and ventral downwards. Dashed lines delineate the ventral outline of worms. Scale bars: 10 μm.

Pan-neuronally expressed CUT homeobox genes act in neurons to control pan-neuronal gene expression (Leyva-Diaz and Hobert, 2022). Using CRISPR/Cas9-engineered reporter alleles, we observe expression of the two otherwise neuron-restricted CUT homeodomain proteins, CEH-44 and CEH-48, in uv1, in addition to the ubiquitously expressed CEH-38 CUT homeodomain protein (Fig. 2B).

We demonstrated the requirement of these CUT homeobox genes for expression of pan-neuronal features in uv1 in two independent manners. First, we assessed expression of the pan-neuronal marker egl-3 in CUT homeobox single, double and triple mutant backgrounds. As reported in the nervous system (Leyva-Diaz and Hobert, 2022), we observed a reduction of egl-3 expression (Fig. 2C,D). Second, we examined expression of a reporter gene-tagged ric-4/SNAP25 locus in which CUT homeodomain binding sites have been deleted through CRISPR/Cas9 genome engineering. This cis-regulatory allele shows a reduction of expression in the nervous system (Leyva-Diaz and Hobert, 2022) and we also observed a reduction of expression in the uv1 cells (Fig. 2C,D).

The transcriptional regulatory signature of uv1

As a first step to understand uv1 cell differentiation, we set out to examine the expression of additional transcription factors in these cells. Previous transgenic reporter gene studies had shown the combinatorial expression of three transcription factors, the LIM homeodomain protein LIN-11, the Sox protein EGL-13 and Pax protein EGL-38 in uv1 (Hanna-Rose and Han, 1999; Newman et al., 1999; Rajakumar and Chamberlin, 2007). Since these observations were made with reporter transgenes that captured only part of the surrounding regions of each locus, we sought to validate these findings using fosmid-based reporters as well as CRISPR/Cas9-engineered reporter alleles. These reagents corroborated continuous co-expression of these three transcription factors in the four uv1 paraneurons throughout the life of these cells (Fig. 3). Since their expression does not appear to overlap elsewhere in the animal, lin-11, egl-38 and egl-13 define a unique regulatory signature within the uv1 cells.

Fig. 3.

A combinatorial transcription factor code for uv1. Expression of cell-specific regulatory factors egl-38, lin-11 and egl-13 in uv1. All strains have ida-1prom::rfp reporter in the background to assess uv1 position. Reporter genes used are CRISPR/Cas9-engineered reporter alleles for lin-11(ot958) and egl-13(dev199) (Ma et al., 2021), and a transgenic fosmid reporter for egl-38(wgIs171). A mScarlet-tagged egl-38 locus (Fig. 5) shows the same expression in uv1 as the fosmid-based reporter. Scale bars: 10 μm.

lin-11 coordinately regulates a large cohort of uv1 identity markers

We first tested the involvement of lin-11 in uv1 differentiation. Previous work examining lin-11 gene function had used either hypomorphic alleles or a presumed lin-11 null allele that eliminated only parts of the locus and also eliminated two adjacent loci (Young et al., 2022). We generated unambiguous molecular null alleles of lin-11 through the elimination of all coding regions of the locus using the CRISPR/Cas9 system (Fernandez et al., 2025). To determine the effect of lin-11 removal, we assessed the expression a large array of uv1 terminal identity markers that have been described in the literature over the past few years (summarized in Table 1). These include a wide range of effector genes that are, in addition to uv1, also expressed selectively in specific neuron classes and therefore attest to the neuron-like molecular identity of uv1. These markers include two mechanosensory TRP channels (ocr-2 and ocr-4) (Jose et al., 2007; Upadhyay et al., 2016), three GPCRs (gar-2, sri-5 and dop-5) (Fernandez et al., 2020; Vidal et al., 2018), two of which are metabotropic neurotransmitter receptors (ACh for gar-2 and monoamine for dop-5), the ligand-gated ion channel lgc-55 (Collins et al., 2016), four neuropeptides (flp-11, flp-22, nlp-2 and nlp-7) (Banerjee et al., 2017; Kim and Li, 2004; Van der Auwera et al., 2020), a dense core vesicle component involved in neuropeptide release (ida-1) (Banerjee et al., 2017; Zahn et al., 2001), the tyramine biosynthetic enzyme tdc-1 (Alkema et al., 2005) and a potassium chloride co-transporter, kcc-2 (Tanis et al., 2009). We found that the expression of every single marker is essentially eradicated upon genetic elimination of lin-11, achieved through CRISPR/Cas9 genome engineering (Fig. 4A).

Fig. 4.

Loss of lin-11 results in an uv1 to utse transformation. (A) Effect of the lin-11 null mutant on uv1 markers. All panels show molecularly identical lin-11 null alleles (ot1867, ot1868, ot1869, ot1870, ot1871, ot1872, ot1873, ot1874, ot1875, ot1876, ot1877, ot1878 and ot1845) generated independently in different reporter backgrounds with identical crRNA and ssODN reagents. Reporter genes used are CRISPR/Cas9-engineered reporter alleles for egl-3(syb4478) and nlp-2(syb5697), and transgenic reporters for ida-1(vsIs269), dop-5(vsIs254), flp-11(ynIs40), flp-22(ynIs50), gar-2(vsEx912), ocr-4(kyEx581), tdc-1(zfIs10) and kcc-2(kvIs136). ‘ON’ and ‘OFF’ refer to reporter signal in wild-type versus lin-11 mutant background. (B,C) lin-11 null mutants have additional utse cells, as seen with nas-22(ot1844[nas-22::SL2::gfp::h2b]) (B) and exc-9(ot1866[exc-9::SL2::gfp::h2b]) CRISPR/Cas9-engineered reporter alleles (C). Animals in all panels were imaged at the young adult stage. All images are approximate lateral views with anterior to the left and ventral downwards. Dashed lines delineate ventral outline of worms. Scale bars: 10 μm. Statistical analysis was performed using Fisher's exact test (A,C) and Welch's t-test (B): **P<0.01, ***P<0.001, ****P<0.0001.

To assess the possibility that, in lin-11 mutants, the uv1 paraneurons adopt the fate of their sister cells, the uterine epithelial utse cells, we developed cellular fate markers for this cell type. Specially, we selected the astacin-like protease NAS-22 and the LIM-only domain protein EXC-9 as candidates, based on pervious transgenic reporter gene studies (McKay et al., 2003; Tong and Buechner, 2008). We inserted an SL2::GFP::H2B tag at the 3′end of the respective coding regions in the genome and indeed observed restricted, nuclear localized expression in the utse cells and no expression in uv1 (Fig. 4B,C). In lin-11 null mutants, we observed additional cells expressing both of these markers, consistent with the possibility of an uv1 to utse cell fate change (Fig. 4B,C).

Temporally controlled lin-11 removal demonstrates protein function in mature uv1 cells

Since lin-11 is not only expressed during and after uv1 differentiation, but already in their precursor cells, the π cells, and in the vulF cells required to induce uv1 fate (Newman et al., 1999), we tested whether we could tie lin-11 function more selectively to the induction of the terminally differentiated state of uv1. To this end we deployed the auxin-dependent degron system. The mScarlet-I3 fluorophore with which we had tagged the lin-11 locus contains two separate mIAA7 degrons at each end of the mScarlet-I3 sequence. Using a ubiquitously expressed TIR1^F79G-driver line, we exposed lin-11::mIAA7::mScarlet-I3::mIAA7 animals at the mid-L4 stage to auxin. Taking into account the time it takes for auxin to induce LIN-11 degradation and the time it takes to then observe an effect on LIN-11 downstream target expression, we infer that the timing of the removal of LIN-11 coincided with uv1 differentiation. Using four different terminal uv1 markers (flp-22, unc-2, tdc-1 and unc-13), we found that LIN-11 removal affects expression of all these markers (Fig. 5A). We conclude that LIN-11 is required for the initiation/maintenance of the terminal uv1 differentiation program.

Fig. 5.

Temporally controlled LIN-11 and EGL-38 removal disrupts uv1 differentiation. (A,B) lin-11(ot1834[lin-11::mIAA7::mScarlet-I3::mIAA]) (A) and egl-38(ot1835[egl-38::mIAA7::mScarlet-I3::mIAA]) (B) were temporally depleted using the AIDv2 system (Hills-Muckey et al., 2022). Worms at the mid L4 stage were transferred onto ethanol- (control) or 5-Ph-IAA-coated plates (depleted) and scored 1 day later. Worms were expressing TIR1^F79G under the ubiquitous eft-3 promoter (osIs182). Reporter genes used are CRISPR/Cas9-engineered reporter alleles for unc-2(ot1614), unc-13(syb6325) and tdc-1(syb7768), and transgenic reporter for flp-22(ynIs50). All images are approximate lateral views with anterior to the left and ventral downwards. Maximum intensity projections of limited image slices are shown, which makes expression in other cells seem more variable than it really is; oftentimes the other cells are not within the image slices used. Dashed lines delineate the ventral outline of worms. Scale bars: 10 μm. Statistical analysis was performed using Welch's t-test: *P<0.05.

The Paired-box EGL-38 protein affects uv1 differentiation

Using three uv1 markers (nlp-2, nlp-7 and ida-1), an effect of a hypomorphic allele of the Pax2/5/8-ortholog egl-38 on uv1 differentiation has been reported in earlier reports (Rajakumar and Chamberlin, 2007; Webb Chasser et al., 2019). An assessment of egl-38 function in uv1 is, however, complicated by the fact that egl-38 is known to be already required in the vulF cells to produce the EGF signal that is essential for uv1 fate specification (Chamberlin et al., 1997; Chang et al., 1999). Constitutive activation of this signal has been used to mask this earlier function of egl-38 and focus on the function of egl-38 in uv1 (Rajakumar and Chamberlin, 2007), but it is difficult to exclude the possibility that the uv1 differentiation defects observed in this genetic background are entirely independent of earlier egl-38 function in vulF. To circumvent this earlier function of egl-38, we used an alternative strategy, again making use of the auxin-inducible degron system. A strain in which mIAA7 degrons were inserted, together with mScarletI3, at the 3′end of the egl-38 locus was subjected to auxin treatment at the L4 stage. We found that such removal affected the expression of three terminal uv1 markers, flp-22, unc-2 and unc-13 (Fig. 5B). As with LIN-11, we interpret these results to mean that EGL-38 acts during the initiation and/or maintenance phase of the terminal uv1 differentiation program.

The SoxD transcription factor egl-13 also affects uv1 differentiation

egl-13 (SoxD homolog) has previously been shown to be expressed in uv1 (Hanna-Rose and Han, 1999) but its function in uv1 differentiation has not previously been examined. To assess a potential role of egl-13, we utilized the previously described ku194 null allele of egl-13 (Hanna-Rose and Han, 1999) and crossed nine of the uv1 cell identity markers described above into this mutant background. We found that expression of each of these markers is lost in egl-13 mutants (Fig. 6A). Similarly, the pan-neuronal uv1 color of the NeuroPAL transgene is lost (Fig. 6A).

Fig. 6.

The SoxD transcription factor egl-13 affects uv1 differentiation. (A) Effect of the egl-13(ku194) loss-of-function allele on uv1 markers. Reporter genes used are CRISPR/Cas9-engineered reporter alleles for lin-11(ot958), unc-13(syb6325) and tcd-1(syb7768), and transgenic reporters for flp-22(ynIs50), ocr-2(vsIs177), ocr-4(kyEx581), ida-1(vsIs269), sri-5(otEx6406) and NeuroPAL(otIs669). ‘ON’ and ‘OFF’ refer to reporter signal in wild-type versus egl-13 mutant background. (B,C) Loss and gain of utse markers in egl-13(ku194) mutants, as seen with exc-9(ot1866) (B) and nas-22(ot1844) (C) CRISPR/Cas9-engineered reporter alleles. Animals in all panels were imaged at the young adult stage. All images are approximate lateral views with anterior to the left and ventral downwards. Maximum intensity projections of limited image slices are shown, which makes expression in other cells seem more variable than it really is; oftentimes the other cells are not within the image slices used. Dashed lines delineate the ventral outline of worms. Scale bars: 10 μm. Statistical analysis was performed using Fisher's exact test: **P<0.01, ***P<0.001, ****P<0.0001.

To assess whether, like in lin-11 mutants, the uv1 cells may adopt an utse identity in egl-13 mutants we used the two utse markers, nas-22 and exc-9, described above. As in lin-11 mutants, exc-9 expression is observed in many additional cells in egl-13 mutants (Fig. 6C). In contrast to lin-11 mutants, however, we observed a complete loss of nas-22 expression in egl-13 mutants (Fig. 6B), indicating that egl-13 is required not only for utse cell fusion to the anchor cell, as described before (Hanna-Rose and Han, 1999), but also for the manifestation of this specific differentiation feature of the utse cells. Taken together, since the uv1/utse precursor cells (the π cells) show additional rounds of cell division in egl-13 mutant animals (Cinar et al., 2003), we surmise that these additional cells may adopt an utse-like fate (as inferred from exc-9 expression), but, as inferred from the lack of nas-22 expression, are not properly/fully differentiated utse cells. We conclude that egl-13 is also required for uv1 differentiation and has additional effects in sister and precursor cells of the uv1 cell.

DISCUSSION

The main purpose of our study was to probe the neuron-like features of uv1 and to characterize transcription factors that control uv1 differentiation in order to draw comparisons between uv1 and canonical neuronal differentiation programs. We observe a dichotomous gene regulatory architecture in uv1 cells that is akin to neuronal cell types: So-called ‘pan-neuronal’ features, i.e. genes involved in synaptic vesicle release and neuropeptide processing, are expressed in uv1 and their proper expression requires pan-neuronally and uv1-expressed CUT homeobox genes. In parallel, neuron- and/or uv1-specific combinations of transcription factors control more cell type-specific aspects of neuron and/or uv1 differentiation. These factors appear to operate in a terminal selector-type manner, i.e. they control multiple and perhaps all cell type-specific features. The uv1 regulatory architecture involves at least one member of a family of transcription factors, homeodomain proteins, that is prominently involved in neuronal cell type specification.

One of two of the transcription factors that we characterize here as a terminal selector of uv1 differentiation through temporally controlled gene removal, egl-38, also has an earlier, upstream role in driving the uv1 differentiation program. egl-38 apparently regulates the ability of the vulF cell to induce uv1 identity via an EGF-mediated signal (Chamberlin et al., 1997; Chang et al., 1999; Webb Chasser et al., 2019). Once received, EGFR signaling induces egl-38 expression, to then drive, together with lin-11 and egl-13, the terminal differentiation program of uv1. The mutant phenotypes of lin-11, and to some extent also egl-13, illustrate another common feature of terminal selectors that has been observed numerous times in the nervous system (Arlotta and Hobert, 2015; Reilly et al., 2022). Rather than just activating a specific differentiation program, terminal selectors often also repress alternative differentiation programs, hence resulting in cell identity changes after terminal selector removal (Arlotta and Hobert, 2015; Reilly et al., 2022). Such a cell identity change is particularly apparent in lin-11 null mutants, in which uv1 now converts to an utse-like identity. lin-11 may act by repressing the expression of a transcription factor that drives utse identity.

How specific combinations of terminal selectors are induced in individual neuron classes in C. elegans is, with a few exceptions, not well understood. uv1 fate specification requires an inductive EGF signal from the neighboring vulF cells which induces EGL-38 expression through EGF receptor (LET-23) signaling (Chang et al., 1999; Webb Chasser et al., 2019). EGL-38-independent functions of EGFR and/or LET-23 signaling in uv1 have been inferred (Webb Chasser et al., 2019) and those may involve the possible regulation of lin-11 and egl-13 expression by let-23 (EGFR homolog). EGF signals are required for neuroblast cell specification in several cellular contexts in C. elegans (Chamberlin and Sternberg, 1994; Jiang and Sternberg, 1998), but whether EGF also triggers identity specification of postmitotic neurons, as it does for the uv1 paraneuron, remains to be investigated.

One unifying theme of paraneuron development across animal phylogeny appears to be the involvement of proneural basic helix-loop-helix (bHLH) transcription factors in their initial specification (Hobert, 2025). We find the C. elegans ASc homolog hlh-3 to be expressed in uv1, but that its removal either in isolation or in combination with removal of the proneural Atonal homolog lin-32, and the NeuroD1 and neurogenin homologs cnd-1 and ngn-1 does not affect uv1 generation. These quadruple mutant animals do, however, display partially penetrant embryonic lethality and it is possible that an absence of uv1 defects in viable ‘escapers’ may be misleading. Another ASc-like C. elegans bHLH gene with proneural function, hlh-14, is an essential gene (Frank et al., 2003) that we could not easily include in our analysis.

Even though the involvement of bHLH factors remains unresolved, the apparent similarity in regulatory architecture of neuronal cell types and the paraneuron that we describe here suggests a potentially shared evolutionary origin of these cell types. Expression of a core set of signaling features – from sensory receptors to secretory machinery – may have been coordinated by a set of transcription factors in primitive proto-neurons. The genomic expansion of transcriptional regulators permitted the assembly of cell type-specific combinations of regulatory factors that diversified differentiation programs of distinct types of ‘proto-neurons’. Some of these cell types may have remained in a morphologically perhaps more primitive secretory type, while others evolved more elaborate morphological and functional features that resemble present-day neurons.

MATERIALS AND METHODS

Strains

A list of strains used in this study is provided in Table S1.

C. elegans genome-engineered strains

lin-11 null alleles

Due to strongly compromised mating efficiency, multiple molecularly identical lin-11 null alleles (in which the entire coding region was removed) were generated by injecting separate stains carrying different uv1 reporters. CRISPR/Cas9 genome engineering was performed as described previously (Eroglu et al., 2023). The following crRNA and ssODN sequences were used: crRNA1, ATTGAGAAGGGAGTAAAAGG; crRNA2, CGTGGAATACTCCTGTATGT; ssODN, TTCGTGGTCGTTCTTCTTCTTCTTCTCCTCCTCCTTACAGGAGTATTCCACGTTCGTGTAGTTTTTCTTC.

lin-11(ot1834) and egl-38(ot1835) AID alleles

lin-11 and egl-38 loci were tagged at the C-terminus with the sequence GSGGSGGTGGSG::mIAA7::wrmScarlet-I3::mIAA7 using CRISPR/Cas9 genome engineering as described previously (Eroglu et al., 2023). The following crRNA sequences were used: lin-11(ot1834) crRNA, agacttgggaaaaccaactc; egl-38(ot1835) crRNA, attctaccacaaaactattg.

For the preparation of the ssODN, since PCR amplification of a fragment containing two identical mIAA7 sequences was problematic, we ordered a synthetic gene fragment with two different synonymous mIAA7 sequences that could be amplified by PCR without problems.

nas-22(ot1844) and exc-9(ot1866) reporter alleles

nas-22 and exc-9 loci were tagged at the C-terminus with the sequence SL2::GFP::H2B using CRISPR/Cas9 genome engineering as described previously (Eroglu et al., 2023). The following crRNA sequences were used: nas-22(ot1844) crRNA, gtcaaactccaacaattgtc; exc-9(ot1866) crRNA, tcaactgcttctccacgttc.

unc-31(syb6138) and unc-13(syb6325) reporter alleles

unc-31 and unc-13 loci were tagged at the C-terminus with a SL2::GFP::H2B tag using CRISPR/Cas9 genome engineering. These strains were generated by SunyBiotech.

Conditional LIN-11 and EGL-38 protein degradation

We used conditional protein depletion with a modified auxin-inducible degradation system, AID2 (Hills-Muckey et al., 2022; Sepers et al., 2022). mIAA7-tagged proteins are conditionally degraded when exposed to 5-phenyl-indole-3-acetic acid (5-Ph-IAA) in the presence of AtTIR1^F79G. To generate the experimental strains, the conditional alleles lin-11(ot1834[lin-11::mIAA7::Scarlet-I3::mIAA7]) and egl-38(ot1835[egl-38::mIAA7::Scarlet-I3::mIAA7]) were crossed with osIs182[eft-3p::TIR1(F79G)], which expresses AtTIR1^F79G ubiquitously. Conditional LIN-11 and EGL-38 protein degradation was performed on NGM plates containing 100 μM 5-Ph-IAA that were prepared using a previously described protocol (Sural et al., 2024). Animals were transferred to 5-Ph-IAA plates at the mid L4 stage and imaged 1 day later. Control plates had the same volume of solvent alone (ethanol). All plates were stored in the dark for the entire duration of the experiment.

Cell identification

Expression of gfp-based reporter genes in uv1 was confirmed with a bona fide uv1 marker, ida-1::mCherry, present in the background. In mutant backgrounds, a loss of gfp-based marker gene expression was usually also accompanied by a loss of ida-1::mCherry; in these cases, we assessed loss of signal based on presumptive cell position of uv1 and other landmarks around the cell.

Fluorescent microscopy

Worms were anesthetized using 100 mM sodium azide (NaN3) and mounted on 5% agarose pads on glass slides. Z-stack images (0.5-1 μm thick) were acquired using a Fluar 40×/1.30 oil objective on a Zeiss compound microscope (Imager Z2) with ZEN Blue software. Maximum intensity projections of 2-30 slices were generated with Fiji/ImageJ software (Schindelin et al., 2012). In many images, the slices are focused around the uv1 cells, leaving other cells either out of focus or not visible at all and therefore giving the false impression that marker expression in other cells may be variable.

Electron microscopy

Archived electron microscopy negatives and prints of a late L4 hermaphrodite, originally collected at the Medical Research Council (MRC) and previously published by Newman et al. (1996) were digitized at high resolution by the Center for C. elegans Anatomy. Negatives were aligned using TrakEM2 (Cardona et al., 2012) and imported into VAST (Berger et al., 2018) for segmentation. Identification of individual cells was made by referencing original print annotations and corresponding MRC notebooks. 3D models .obj files were generated in VAST and imported into Blender for rendering following a remesh modifier and Laplacian smoothing.

Statistical analysis

Data points for all biological replicates and sample sizes (n) are displayed in the figures. Statistical tests used in each figure and the corresponding P-values are listed in the figure legends. P<0.05 was considered to be statistically significant. All statistical tests and plotting were performed on GraphPad Prism 10.

Peer review history

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.205749.reviewer-comments.pdf