Neurons and glia display remarkable morphological plasticity, and remodeling of glia may facilitate neuronal shape changes. The molecular basis and control of glial shape changes is not well understood. In response to environmental stress, the nematode Caenorhabditis elegans enters an alternative developmental state, called dauer, in which glia and neurons of the amphid sensory organ remodel. Here, we describe a genetic screen aimed at identifying genes required for amphid glia remodeling. We previously demonstrated that remodeling requires the Otx-type transcription factor TTX-1 and its direct target, the receptor tyrosine kinase gene ver-1. We now find that the hunchback/Ikaros-like C2H2 zinc-finger factor ztf-16 is also required. We show that ztf-16 mutants exhibit pronounced remodeling defects, which are explained, at least in part, by defects in the expression of ver-1. Expression and cell-specific rescue studies suggest that ztf-16, like ttx-1, functions within glia; however, promoter deletion studies show that ztf-16 acts through a site on the ver-1 promoter that is independent of ttx-1. Our studies identify an important component of glia remodeling and suggest that transcriptional changes may underlie glial morphological plasticity in the sensory organs of C. elegans.

THE shapes of neuronal receptive structures such as dendritic spines and sensory receptive endings are plastic and can be remodeled by developmental, hormonal, and environmental signals. For example, the dendritic spines of Purkinje cells rapidly grow and retract during development of the mouse cerebellum (Dunaevsky et al. 1999), and estrogen levels affect the number and density of dendritic spines in the hippocampus during the estrous cycle in rats (Woolley et al. 1990). In the mouse barrel cortex, changes in somatosensory environmental input increase the turnover of pyramidal neuron dendritic spines (Trachtenberg et al. 2002). Likewise, reduced sensory input affects the shape of the AWB neuron sensory ending in the nematode Caenorhabditis elegans (Mukhopadhyay et al. 2008).

Glia, which are intimately associated with neurons, also exhibit complex shapes and morphological plasticity, and changes in glial shape often correlate with neuronal remodeling (Procko and Shaham 2010). For example, retraction of astrocyte processes in the hypothalamus of lactating rats correlates with synaptic changes in associated supraoptic nucleus neurons (Theodosis and Poulain 1993); and perturbation of the glial ephrin-A3 cell-surface protein affects the shape of dendritic spines in the mouse hippocampus (Carmona et al. 2009). These observations, coupled with the close proximity of glia to neurons and the ability of glia to regulate and perceive their extracellular environment (Meyer-Franke et al. 1995; Porter and Mccarthy 1996), suggest that glia are well positioned to facilitate or, more speculatively, to direct changes in neuronal receptive ending shapes. Although glial shape changes in vertebrate systems have been documented (Theodosis and Poulain 1993; Lippman et al. 2008), the molecular basis for these changes is not well understood.

The nematode C. elegans has a relatively small number of glia and neurons with stereotyped shapes (Ward et al. 1975; White et al. 1986), and, unlike vertebrate glia, C. elegans glia can be ablated without affecting neuronal survival (Bacaj et al. 2008; Yoshimura et al. 2008). In response to environmental stressors, including starvation, crowding, and high temperature, C. elegans enters a developmentally arrested stress-resistant state termed “dauer” (Cassada and Russell 1975; Golden and Riddle 1984), in which the anterior bilateral amphid sensory organs are remodeled (Albert and Riddle 1983). Each amphid consists of sensory neurons that extend dendritic processes to the nose tip and there terminate in specialized sensory endings. Associated with these neurons are single amphid sheath (AMsh) glial cells. Each AMsh glia also extends a process toward the nose tip, and there ensheaths sensory neuron ciliated receptive endings (Figure 1A) (Ward et al. 1975). In dauer animals, the two AMsh glia expand at the nose tip and use the protein AFF-1 to fuse (Albert and Riddle 1983; Procko et al. 2011). Concomitantly, the ciliated sensory endings of the AWC amphid neurons expand within the new compartment defined by the glia, such that the left and right AWC cilia now extensively overlap (Figure 1, B and C). Importantly, AMsh glia remodel even in the absence of the AWC neurons, and blocking glial remodeling perturbs AWC neuron shape, suggesting that glial remodeling facilitates neuronal shape changes (Procko et al. 2011).

Figure 1 

A schematic of amphid sensory organ remodeling in C. elegans dauer animals. (A) A schematic of the head of the animal, showing the two bilateral AMsh glia and AWC sensory neurons. The horizontal line indicates the position of the transverse sections shown in B and C. (B and C) Sections through the nose tip showing the left and right AWC neuron receptive endings (AWCL/R; dark shading) and the ensheathing AMsh glia (AMshL/R; light shading) in wild-type non-dauer (B) and dauer (C) animals. Note AMsh glial fusion and AWC receptive ending expansion and overlap in dauers.

We previously found that glial remodeling depends on the Otx-type transcription factor ttx-1 and its direct target gene, the receptor tyrosine kinase ver-1, both acting within AMsh glia (Procko et al. 2011). Expression of ver-1 is induced in AMsh glia following dauer entry and in non-dauer animals of all stages upon cultivation at high temperature (25°), a dauer stimulus. Here, we report results of a genetic screen aimed at identifying mutants defective in ver-1 expression and identify the gene ztf-16 (zinc-finger putative transcription factor family) as a major regulator of ver-1 expression and AMsh glia remodeling. Our results suggest that transcription factors play important roles in AMsh glia morphological plasticity.

Materials and Methods


Animals were cultivated at 20° using standard methods (Brenner 1974), unless otherwise noted. The wild-type strain used was Bristol (N2). Mutant alleles used were the following: LGIII—daf-7(e1372) and lit-1(ns132, t1512); and LGV—ttx-1(p767, oy26). Mutant alleles isolated in this study were the following: LGV—tam-1(ns167, ns170, ns174, ns234, ns237, ns238, ns241, ns249, ns258, ns268) and ttx-1(ns235, ns252, ns255, ns259, ns260, ns267); and LGX—ztf-16(ns169, ns171, ns178). Alleles not mapped to a chromosome include ns231 and ns257. ver-1 promoter::gfp was nsIs22. Other integrated transgenes and extrachromosomal arrays are as indicated elsewhere in the text.

Plasmid construction and isolation of cDNAs

The initial ver-1 promoter::gfp construct (∼2 kb upstream promoter through +263 of the ver-1 gene fused to gfp) was a gift of R. Roubin and C. Popovici. Vector backbones are derived from the pPD vectors (gift of A. Fire). cDNA template was prepared from mixed-stage animals. The glia::ztf-16a and glia::ztf-16b constructs were made by replacing the heat-shock promoter of vector pPD49.78 with a 2-kb promoter from the F16F9.3 gene (Bacaj et al. 2008) at PstI/BamHI and then inserting either ztf-16a or ztf-16b cDNA at XmaI/NcoI. The glia::ztf-16 construct that was integrated to generate strain nsIs245 was made by inserting 2.5 kb of the T02B11.3 promoter (Bacaj et al. 2008) into pPD49.78 at PstI/BamHI and ztf-16b cDNA at XmaI/SacI. Embryonic glia::ztf-16b was made by inserting a lin-26 glia enhancer region and a myo-2 minimal promoter (Landmann et al. 2004) into pPD49.78 at SphI/XmaI, followed by ztf-16b at XmaI/NcoI. The F16F9.3 promoter was inserted into the PstI site of pPD95.75 followed by ztf-16b cDNA in frame with gfp at XmaI/KpnI to generate glia promoter::ztf-16b::gfp. The ztf-16 promoter::gfp construct that gave gfp expression in the amphid and phasmid glia included the region −4637 to −2536 of the ztf-16 promoter relative to the +1 translation start site inserted into pPD95.75 at SphI/KpnI. To generate the F58F9.10 promoter::gfp construct, 2 kb of upstream promoter was inserted into pPD95.75 at SphI/BamHI. To make the F58F9.6 promoter::gfp, a 4-kb BamHI/XmaI promoter fragment was inserted at BamHI/AgeI.

Germline transformation and transgene integration

Germline transformations were carried out using standard protocols (Mello and Fire 1995). Co-injection markers used were either plasmid pRF4 (Mello et al. 1991) or unc-122 promoter::dsRed. pSL1180 (GE Healthcare) is an empty cloning vector used to increase the DNA concentration of injection mixtures. The glia promoter::ztf-16 rescue transgene (nsIs245) was integrated by treating animals carrying extrachromosomal arrays with UV/psoralen (Mello and Fire 1995). The generated strain was backcrossed to N2 three times.


ver-1 promoter::gfp (nsIs22) expression was assayed using a fluorescence dissecting microscope (Leica). Adult hermaphrodites were scored, except as noted. Compound microscope images were taken on an Axioplan II microscope using an AxioCam CCD camera (Zeiss) and analyzed using Axiovision software (Zeiss). Additional images were taken on a Deltavision Image Restoration Microscope (Applied Precision/Olympus) and analyzed using SoftWoRx software (Applied Precision). Dauer animals for electron microscopy (EM) were grown at 25°. These were prepared and sectioned using standard methods (Lundquist et al. 2001). Imaging was performed with an FEI Tecnai G2 Spirit BioTwin transmission electron microscope equipped with a Gatan 4K × 4K digital camera.

Dauer selection

Animals were starved and dauers selected by treatment with 1% SDS in M9 solution for 20 min. Alternatively, animals carrying the daf-7(e1372) mutation were induced to form dauers by incubation at 25°.

Mutagenesis and mapping

L4 animals carrying the ver-1 promoter::gfp transgene (nsIs22) in the N2 strain background were mutagenized with 30 mM ethyl methanesulfonate (EMS) for 4 hr. Mutagenized animals were picked to separate 9-cm NGM agar plates seeded with Escherichia coli OP50 and cultivated at 25°. F2 animals were screened. Mapping was performed by crossing to the Hawaiian strain (CB4856), picking mutant F2 progeny, and observing linkage to single nucleotide polymorphisms (SNPs) (Wicks et al. 2001).

Cytoplasmic mixing assay to score AMsh glia fusion

Adult animals carrying an nsEx1391 (AMsh glia::gfp) array were picked to plates seeded with OP50 bacteria and cultivated at 25°. From these plates, L1 and L2 progeny carrying the nsEx1391 array in one of the two AMsh glia were picked to freshly seeded plates. These mosaic animals were incubated for 48 hr at 25° before scoring GFP presence in either one or both AMsh glia. Animals carrying a daf-7(e1372) mutation were scored only if they were dauer larvae by morphology at the end of the assay period. See Procko et al. (2011).

Behavioral assays

Thermotaxis and chemotaxis assays were performed as previously described (Procko et al. 2011).


A genetic screen for mutants defective in ver-1 expression

To study how dauer remodeling is initiated, we aimed to identify mutants defective in this process. However, existing methodologies for following the remodeling process—namely electron microscopy and assessment of glial fusion by tracking mosaic animals (Procko et al. 2011)—are low throughput. We turned, therefore, to a more indirect approach. The C. elegans tyrosine-kinase receptor gene ver-1 is expressed in the AMsh and phasmid sheath (PHsh) glia of the amphid and phasmid sensory organs upon dauer entry and at high temperature (25°) and is important for amphid glia remodeling in dauers. ver-1 induction in dauers and at high temperature is, at least in part, similarly regulated as mutations in ttx-1, encoding a direct transcriptional activator of ver-1, blocks induction in both settings (Procko et al. 2011). Furthermore, ttx-1 mutations also block AMsh glia remodeling in dauers. To identify genes involved in the initiation of glia remodeling, we therefore sought mutants with defects in ver-1 expression at high temperature. Wild-type animals carrying a ver-1 promoter::gfp reporter transgene (nsIs22) were mutagenized with EMS, and >35,000 F2 progeny grown at 25° were screened for reduced GFP fluorescence in AMsh glia. A total of 21 independent mutant alleles were identified (Table 1). Two of these alleles, ns235 and ns252, failed to complement the ttx-1(p767) allele when scored for ver-1 promoter::gfp expression (Table 1), and both had the same G-to-A nucleotide change within the ttx-1-coding region, altering amino acid 230 from glutamate to lysine (Figure 2). The isolation of mutations in ttx-1, a known glial remodeling regulator, validated our screen strategy.

View this table:
Table 1  Mutant strains that reduce ver-1 promoter::gfp expression
Figure 2 

Mutations in the Otx transcription factor ttx-1 reduce ver-1 promoter::gfp expression. (A) A schematic of the ttx-1 gene. Exons are represented by boxes; the start codon (+1) is indicated; and regions encoding the DNA-binding homeodomain are shaded. Mutant ttx-1 alleles isolated in our screen are shown, and the corresponding amino acid changes are indicated. The asterisk indicates a premature stop mutation. The region of the ns260 deletion is represented by a solid horizontal line [includes insertion (T)7CA(T)14]. ns255 likely represents a rearrangement of the gene, shown by a horizontal dashed line (precise ends unknown), as PCR products using primers covering exons 2–6 are absent, weak, or consist of multiple-sized fragments (data not shown). In addition, ns255 has the same base substitution as ns235 and ns252, causing amino acid change E230K. The previously described ttx-1 alleles p767 and oy26 are also shown (Hedgecock and Russell 1975; Satterlee et al. 2001). (B) Alignment of the DNA-binding homeodomains of Otx-type factors TTX-1 (A isoform), Drosophila OTD, and murine Otx1 and Otx2. Amino acid positions are shown in parentheses. ns235 and ns252 (and ns255) code for a glutamic acid to lysine change in a conserved residue.

Four of the mutant lines that we identified, ns255, ns259, ns260, and ns267, had a strong dominant reduction in ver-1 promoter::gfp expression and could not be assayed in complementation studies (Table 1; data not shown). As shown in Figure 2 and in Supporting Information, Figure S1, all four dominant alleles also had sequence changes within the ttx-1-coding region. The allele ns260, which is predicted to lack sequences encoding the TTX-1 DNA-binding domain, is likely homozygous embryonic lethal, as viable progeny from ns260/+ parents either were homozygous for the wild-type ttx-1 allele and expressed wild-type levels of ver-1 promoter::gfp or were ns260/+ and had low GFP fluorescence (Figure S2A). Consistent with an essential role for ttx-1, we found that animals heterozygous for the dominant ttx-1 allele ns259 also gave rise to fewer than expected ns259 homozygous progeny. Only 7 of 68 progeny of an ns259/+ parent grown at 25° were ns259 homozygotes (P = 0.02, χ2 test), and all 7 homozygotes were sterile. Similarly, at 15°, 6 of 65 progeny were ns259 homozygotes (P < 0.02), and all were sterile.

ttx-1 is expressed in both AMsh glia and the AFD thermosensory neurons (Satterlee et al. 2001; Procko et al. 2011). However, we were unable to rescue ttx-1(ns260) lethality using either glial or AFD promoters (Figure S2). In addition, sterile ns259 homozygous animals possessed AMsh glia of normal morphology, as assayed by expression of a glial F16F9.3 promoter::dsRed transgene (n = 16). Taken together, these results suggest that ttx-1 is likely to have essential developmental roles in cells other than AMsh glia and the AFD neuron.

Nine of the alleles that we isolated failed to complement the allele ns258, also identified in our screen (Table 1; data not shown). SNP mapping (Wicks et al. 2001) was used to map one of these alleles, ns268, to a ∼160-kb interval on chromosome V (Figure 3A). Cosmid F26G5 within this interval rescued the ver-1 expression defect when injected into ns268 mutants (Figure 3B), and sequencing of coding regions spanned by this cosmid uncovered two mutations within the gene tam-1. Mutations in tam-1 were also identified in each of the other nine alleles of this complementation group (Figure 3C), confirming tam-1 as the relevant affected gene. tam-1 encodes a protein predicted to contain a C3HC4 zinc finger (RING finger) and a B-box motif and has been shown to broadly and nonspecifically regulate gene expression of transgenes from simple DNA arrays (Hsieh et al. 1999). Therefore, the effects of tam-1 mutations on ver-1 promoter::gfp expression may not reflect a role in the control of endogenous ver-1 expression, and we did not pursue further characterization of this gene.

Figure 3 

Mutations in the RING finger and B-box domain factor tam-1 reduce ver-1 promoter::gfp expression. (A) A schematic of the interval on chromosome V to which the mutant allele ns258 was mapped. The flanking SNPs are on cosmid C02E7 (base 9924) and C04F5 (base 1949). The regions spanned by the F26G5 cosmid and tam-1 gene are also shown. (B) Rescue of ver-1 promoter::gfp expression defects of tam-1(ns268) mutants by two independent extrachromosomal arrays containing the F26G5 cosmid (nsEx2169 and nsEx2170). (C) A schematic of the tam-1 gene (WormBase release WS225). Exons are represented by boxes. Mutant tam-1 alleles isolated in our screen are shown, and the corresponding amino acid changes are indicated. Asterisks indicate a premature stop mutation. ns268 carries two putative amino acid changes.

Two additional alleles identified in our screen, ns231 and ns257, did not harbor mutations in ttx-1, tam-1, or ztf-16 (see below), had only weak defects in ver-1 expression, and were not further studied.

C2H2 zinc-finger factor ztf-16 is required for ver-1 expression

The ns169 and ns178 alleles fail to complement ns171, another allele isolated in our screen, and in all three alleles expression of a ver-1 promoter::gfp transgene in the AMsh and PHsh glia of adult animals raised at 25° is strongly attenuated (Table 1; Figure 4, A and B; data not shown). Furthermore, whereas 100% of wild-type dauers expressed ver-1 promoter::gfp in the AMsh glia, 0% of ns169, ns171, or ns178 mutant dauers induced by starvation at 15° expressed the reporter (n = 50 for each allele; Figure 4C). By contrast, mutations in this complementation group had little or no effect on an AMsh glia reporter that is expressed constitutively and independently of dauer entry (Figure 4D). Unlike mutations in ttx-1, which also disrupt AFD sensory neuron-mediated thermotaxis behavior and AFD morphology (Hedgecock and Russell 1975; Satterlee et al. 2001; Procko et al. 2011), ns171 mutants exhibited nearly normal thermotaxis behavior and AFD morphology (Figure 5, A–F; Figure S3). Together, these observations suggest that the gene defined by the ns171 complementation group is specifically required for ver-1 promoter::gfp expression in AMsh glia and does not affect glial cell fate, AFD cell fate, or general aspects of gene expression in these cells.

Figure 4 

Temperature- and dauer-induced expression of ver-1 is reduced in ns171 mutants. (A) Representative fluorescence images and DIC and fluorescence merged images of ver-1 promoter::gfp (nsIs22) expression in one of the two AMsh glial cells of a wild-type adult cultivated at 15° (left) and 25° (right). (B) As in A, except in an ns171 mutant animal. Exposure time for gfp (A and B), 800 msec. (C) Representative fluorescence images of ver-1 promoter::gfp (nsIs22) expression in one of the two AMsh glial cells of a wild-type dauer induced by starvation at 15° (left) and a ns171 mutant dauer animal (right). Exposure time, 200 msec. (D) Representative fluorescence images of vap-1 promoter::dsRed (nsIs53) expression in the AMsh glia of a wild-type adult (left) and a ns171 mutant adult animal (right). Exposure time, 600 msec. Bar in all images, 50 μm. Anterior is up.

Figure 5 

ns171 mutants do not strongly perturb sensory neuron morphology or behavior. (A–C) Representative fluorescence images of the AFD dendrite endings of adult wild-type (A), ns171 mutant (B), and ttx-1(oy26) mutant (C) strains at 25°. The wild-type microvillar projections are absent or reduced in ttx-1 mutants, and instead the dendrite terminates in an aberrant, elongated single cilium (arrow in C). Bars, 5 μm. Anterior is up. A gcy-8 promoter::gfp transgene (oyIs17) was used to visualize the AFD neurons. (D–F) Thermotaxis of wild-type (D), ns171 mutant (E), and ttx-1(p767) mutant (F) strains. Animals were cultivated at 15° (blue), 20° (yellow), or 25° (red) prior to performing each assay. The linear temperature gradient is represented by bins 1–6 on the horizontal axis from cold (∼18°) to hot (∼26°). Wild-type animals migrate to the temperature at which they were cultivated. All values are mean ± SD. All animals also carry the ver-1 promoter::gfp transgene (nsIs22). F is reproduced from Procko et al. (2011). (G and H) Representative fluorescence images of the AWC dendrite endings of L4 wild-type (G) and ns171 mutant (H) strains. Bars, 5 μm. Anterior is up. An odr-1 promoter::yfp transgene (oyIs45) was used to visualize the AWC neurons. (I) Chemotaxis index of wild-type and ns169 and ns171 mutant adult animals to the AWC-sensed odorant benzaldehyde (0.2%). Values are mean ± SD. The difference between ns169 and wild-type animals is not significant (P > 0.05), whereas for ns171 mutants compared to wild type, P = 0.021 (Student’s t-test).

To identify the gene corresponding to the ns171 complementation group, we used SNP mapping (Wicks et al. 2001) to place the ns171 mutation within an interval of ∼370 kb on chromosome X (Figure 6A). Cosmids spanning the 5′ region of this interval were injected into ns171 mutants and scored for rescue of ver-1 promoter::gfp expression in adults raised at 25°. One of these cosmids, R08E3, gave rescue (Figure 6B). Candidate coding regions were sequenced within this interval, and a single C-to-T substitution at codon 236 of the ztf-16 open reading frame was identified. This mutation is predicted to cause a premature stop (Figure 6C). ns178 mutants have the same base alteration as ns171 animals, and ns169 mutants harbor a different C-to-T mutation, at codon 131, which is also predicted to generate a premature stop (Figure 6C). Taken together, these studies demonstrate that ztf-16 is the relevant gene affected in mutants of the ns171 complementation group.

Figure 6 

The C2H2 zinc-finger factor ztf-16 is required for ver-1 expression. (A) A schematic of the interval on chromosome X to which mutant allele ns171 was mapped. The flanking SNPs are on cosmids F55D10 (base 14867) and C42D8 (base 5707). The regions spanned by the cosmids used for the rescue experiments shown in B are indicated, as is the position of the ztf-16 gene. (B) The number of lines carrying extrachromosomal arrays of the indicated cosmid that rescued the reduced ver-1 promoter::gfp (nsIs22) expression defect of ns171 mutant adult animals cultivated at 25°. (C) A schematic of the ztf-16 gene: exons are boxed; and regions encoding C2H2 zinc-finger domains are shaded. We isolated two splice forms of ztf-16 on the basis of EST data available from WormBase (release WS225) and have named these gene models ztf-16a and ztf-16b. The mutant ztf-16 alleles isolated in our screen are shown, and the corresponding amino acid change is indicated. Asterisks indicate a premature stop mutation.

ztf-16 encodes a protein predicted to contain up to eight C2H2 zinc-finger domains. C2H2 zinc-finger proteins are abundant transcriptional regulators in mammals, with >130 expressed in the brain alone (Iuchi 2001). On the basis of the pattern of C2H2 zinc fingers, ztf-16 has been classified as a hunchback- and Ikaros-like transcription factor (Large and Mathies 2010). In vertebrates, the Ikaros family of C2H2 zinc-finger transcription factors have broad roles in the development of the hematopoietic system (Smale and Dorshkind 2006), while hunchback was identified as a factor regulating Drosophila embryo patterning (Tautz et al. 1987). hunchback- and Ikaros-like transcription factors have a unique arrangement of C2H2 zinc fingers: four amino-terminal or middle C2H2 zinc fingers bind DNA (Molnar and Georgopoulos 1994), while two carboxy-terminal C2H2 zinc fingers mediate dimerization (McCarty et al. 2003). In ztf-16, it is likely that zinc fingers 3–6 form the putative DNA-binding domain (Large and Mathies 2010).

On the basis of expressed sequence tag (EST) data available from WormBase (release WS225), we isolated two alternatively spliced cDNAs, ztf-16a and ztf-16b, derived from the ztf-16 locus. ZTF-16a and ZTF-16b proteins are predicted to differ at their carboxy-termini. ZTF-16a lacks zinc fingers 7 and 8, possessing instead a short sequence absent in ZTF-16b (Figure 6C).

ztf-16 functions within glia to regulate ver-1 expression

ztf-16 was previously suggested to play a minor role in somatic gonad development and was shown to be expressed in this tissue (Large and Mathies 2010). Our findings suggest that ztf-16 also has roles in glia that may be independent of its gonadal functions. To test this, we generated animals carrying transgenes containing regions of the ztf-16 promoter fused to gfp. A 2.5-kb region immediately adjacent to the ztf-16 start codon is expressed in hypodermal and other cell types, but not in glia (data not shown). By contrast, a 2-kb region further upstream (Figure 7A) gives strong, specific expression in AMsh and PHsh glia, in AMso and PHso socket glia, and in an unidentified pair of neurons in the head (Figure 7B). Consistent with this expression pattern, cosmid F43C9, which includes all ztf-16-coding fragments but only 300 bp of upstream regulatory sequences, fails to rescue ver-1 promoter::gfp expression in ztf-16(ns171) mutants (Figure 6, A and B). Furthermore, ver-1 expression defects in ztf-16(ns169) and ztf-16(ns171) mutants are rescued by expression of either ztf-16a or ztf-16b cDNAs using the constitutive F16F9.3 glia-specific promoter (Bacaj et al. 2008) (Table 2). Finally, we found that a ZTF-16::GFP fusion protein tightly localizes to AMsh nuclei (Figure 7C; n = 50), where it may be poised to regulate transcription. Taken together, these results demonstrate that ztf-16 functions cell-autonomously to regulate transcription of ver-1 within glia. These results also suggest that the two carboxy-terminal C2H2 zinc fingers, which are absent in ZTF-16a, are dispensable for regulation of ver-1 expression.

Figure 7 

ZTF-16 is expressed in glia and localizes to the nucleus. (A) A schematic of the ztf-16 promoter. The start codon (+1) is shown, as well as ∼4.6 kb of upstream promoter sequence (solid, horizontal line). The enhancer element used in B is between −4637 and −2536 relative to the +1 start site. (B) Fluorescence image (left) and fluorescence and DIC merged image (right) of an adult animal carrying a transgene containing the ztf-16 promoter region shown in A driving gfp expression (nsEx3001). Fluorescence is seen in the two AMsh glia (arrows), the AMso glia (arrowheads), and a pair of unidentified neurons anterior of the AMsh cell bodies. Expression is also observed in the PHsh and PHso glia in the tail (data not shown). (C) Localization of a ZTF-16::GFP fusion protein to the nucleus of the AMsh glia when expressed under a glia-specific promoter (nsEx1347). Arrows indicate the AMsh glia nucleus. Bar, 50 μm. Anterior is up.

View this table:
Table 2  ztf-16 acts in glia to control ver-1 expression

The F16F9.3 promoter is active starting at the threefold stage of embryogenesis, well after AMsh glia are born (Bacaj et al. 2008). That F16F9.3 promoter::ztf-16 cDNA constructs are able to rescue the ver-1 expression defects of ztf-16 mutants suggests that the gene is not required for early aspects of glia generation, morphogenesis, or development. Indeed, we were unable to rescue ztf-16 mutants by expressing ztf-16 cDNA in AMsh glia using an embryonic glia promoter that is not expressed in later larval and adult stages (Table 2). This embryonic promoter is able to rescue other early AMsh glia defects (Perens and Shaham 2005; Oikonomou et al. 2011).

ZTF-16 regulates dauer expression of ver-1 through a site distinct from that bound by TTX-1

We previously showed that robust expression of ver-1 promoter::gfp transgenes requires residues +1 to +263 of the ver-1 gene (relative to the ATG start site). We further described a smaller ∼90-bp interval sufficient for weak expression of the reporter in glia upon dauer entry. Within this interval, we identified a direct TTX-1-binding site with the core-binding residues located between position +176 and +179 (Procko et al. 2011) (Figure 8). Strikingly, we find that the ztf-16(ns171) mutation reduces ver-1 reporter expression only if residues +221 to +263 of the ver-1 gene are present. Specifically, GFP expression in animals carrying a transgene in which residues +221 to +263 of the ver-1 promoter are deleted is not altered in ztf-16 mutants (compare expression in dauers at 25° in wild-type and ztf-16 animals carrying either the top construct or the second construct from the bottom in Figure 8). Within the region of ver-1 regulated by ztf-16, we identified a potential ZTF-16-binding site, CATGAAAAC, at positions +217 to +225 on the basis of homology to Drosophila Hunchback, which binds the consensus sequence (G/C)(C/A)TAAAAAA (Stanojevic et al. 1989). Mutating these residues to GGGCCCAAC resulted in reduced ver-1 promoter::gfp expression (compare expression from top and bottom constructs in wild-type adult animals at 25° in Figure 8), raising the possibility that ZTF-16 may bind directly to the ver-1 gene to regulate its expression. To test for direct binding in vitro, we initially attempted to purify soluble full-length GST::ZTF-16a or GST::ZTF-16b protein induced in E. coli, but were unable to do so. We were able to purify zinc fingers 2–6 of the protein, but these showed only weak, nonspecific binding to a 40-bp biotin-labeled probe from the ver-1 gene (data not shown). Thus, it remains unclear whether ZTF-16 directly binds ver-1.

Figure 8 

ZTF-16 regulates expression from the ver-1 promoter through a site independent of the TTX-1-binding site. The indicated ver-1 promoter fragments (left, boxes) were fused to a gfp fluorescent reporter and tested for expression in the AMsh glia in adults raised at 15° and 25° and in dauers induced by starvation at 15° and 25°. The ver-1 gene fragments used are indicated relative to the +1 start codon. The ver-1 promoter::gfp transgene used for the mutant screen (nsIs22) is shown at the top (∼2 kb of upstream promoter sequence through +263 of the ver-1 gene). The positions of the TTX-1-binding site (Procko et al. 2011) and the Hunchback (Hb)-related binding site are shown. The asterisk indicates a mutated site (see Results). To test if ztf-16 was required for expression of a particular reporter, the reporter was crossed to ztf-16(ns171) mutants. P-values of reporter expression in the ztf-16 mutant dauers at 25° were determined by comparing the mutant strain against wild type using a χ2 test. The integrated transgenes and extrachromosomal arrays used were, from top to bottom, nsIs22, nsEx1136, nsEx2174, nsEx1269, and nsEx3022. Arrays are representative of others. n.d., not determined; n.s., not significant.

Taken together, our promoter studies suggest that ZTF-16 regulates ver-1 expression directly or indirectly through a site in ver-1 that is distinct from that used by TTX-1. Our studies also suggest that ztf-16 does not confer dauer dependence on ver-1 expression: although ztf-16 mutants have reduced ver-1 expression, induction of expression in dauers at 25° is still evident (Figure 8).

Our finding that TTX-1 and ZTF-16 are each required for expression of ver-1 raised the possibility that these transcription factors require each other to promote AMsh glia-specific expression. To test this idea, we scanned the genome for TTX-1-binding sites similar to that found in ver-1 (Procko et al. 2011) and identified a highly similar sequence (GATTATCGGATTCAG) within a cluster of divergently transcribed genes encoding proteins with thrombospondin domains (Figure S4A). Other proteins with such domains had been previously implicated in glial function in C. elegans and in vertebrates (Christopherson et al. 2005; Bacaj et al. 2008). While a promoter::gfp reporter for one of these genes is expressed exclusively in the AFD neurons, which normally express TTX-1, a similar reporter for another of the divergently transcribed genes of the thrombospondin-domain gene cluster is expressed in AMsh glia (Figure S4, B and C). Expression of the reporter is eliminated in ttx-1(p767) mutants (Figure S4D). Importantly, reporter expression was normal in ztf-16(ns171) mutants (n = 53), suggesting that ttx-1 and ztf-16 need not function together to promote AMsh gene expression.

ztf-16 function is required in glia for AMsh glia remodeling

Our screen aimed to identify genes controlling the initiation of glia remodeling by identifying regulators of ver-1 expression. To examine whether ztf-16 is indeed required for glia remodeling, we used an assay that we previously developed to score fusion of the two AMsh glia in dauer animals (Procko et al. 2011). Briefly, young daf-7(e1372ts) mutant larvae cultivated at 25° and harboring an AMsh glia::gfp transgene on an unstable extrachromosomal array (nsEx1391) were selected for mosaic expression of GFP in one of the two AMsh glia. Mosaic animals were allowed to grow for an additional 48 hr, at which point nearly all became dauers as a result of the daf-7 mutation. These dauers were then examined for cytoplasmic mixing of GFP between the two glia, which occurs only if the cells have remodeled and fused (Procko et al. 2011).

Using this cytoplasmic mixing assay, we found that ztf-16(ns169); daf-7(e1372) and ztf-16(ns171); daf-7(e1372) dauers had significantly reduced AMsh glia fusion compared to daf-7(e1372) single-mutant dauers (Figure 9A). Consistent with this result, we found that three of three ztf-16(ns171); daf-7(e1372) dauer animals examined by EM failed to exhibit AMsh glia extension and fusion (Figure 9, B and C; Figure S5). We could rescue the fusion defect by restoring ztf-16 function specifically in glia (Figure 9A). Rescue was more efficient for the ztf-16(ns171) allele, perhaps because it has a weaker defect than ztf-16(ns169). Together, these findings suggest that ztf-16 functions within glia to promote dauer-dependent remodeling.

Figure 9 

ztf-16 function is required for AMsh glial remodeling in dauer animals. (A) Percentage of daf-7(e1372) dauer animals of the indicated genotype with fused AMsh glia as scored by cytoplasmic mixing (see Procko et al. 2011). Number of animals (n) is shown above each column. *P = 0.001 (Fisher’s exact test); **P < 0.001 (χ2 test). The glia::ztf-16 transgene is nsIs245 (T02B11.3 promoter::ztf-16b). (B and C) Representative electron micrographs (left) and schematic outlines (right) of the amphid sensory organs of a daf-7(e1372) dauer (B) and a ztf-16(ns171); daf-7(e1372) dauer (C) cultivated at 25°. Left and right AWC neuron sensory cilia (AWCL/R; dark shading) and AMsh glia (AMshL/R; light shading) are indicated. In C, a section close to the anterior tip of the glial processes is shown where maximum AWC expansion occurs. AMsh glial fusion can occur on either the dorsal or the ventral side of the animal; for simplicity, only the top dorsal half is shown. Bar, 5 μm. Dorsal is up. B is reproduced from Procko et al. (2011).

Our findings that ver-1 mutants have reduced AMsh glia fusion in dauers (Procko et al. 2011) and that ver-1 expression is greatly reduced in ztf-16 mutants suggest that, like ttx-1, ztf-16 acts in part to effect remodeling by controlling ver-1 expression. We also previously demonstrated a role for the gene aff-1 in glia fusion (Procko et al. 2011). AFF-1 protein functions as a fusogen-promoting syncytium formation in C. elegans (Sapir et al. 2007). To test whether ztf-16 might also regulate aff-1 expression, we examined dauer animals carrying an aff-1 promoter::gfp reporter (hyEx167). We found that 95% of wild-type dauers expressed gfp in the AMsh glia (n = 44) and that 87% of ztf-16(ns171) mutants expressed gfp (n = 38). These observations suggest that ztf-16 is not required for aff-1 expression.

ztf-16 may function as a general regulator of AMsh glia morphology or may have specific roles in dauer remodeling. To distinguish between these possibilities, we examined non-dauer ztf-16 mutant adults carrying a glia-specific vap-1 promoter::dsRed reporter transgene (Figure 4D). We found no gross defects in AMsh glia morphology. Similarly, overall glial shape is normal in electron micrographs of ztf-16(ns171) mutants. However, in these micrographs, the amphid sensory channel stains abnormally darkly, as do pockets within the AMsh glia (two of three animals examined; Figure S6). Furthermore, some sensory neurons fail to traverse the amphid channel, instead becoming trapped within the AMsh glial cell (Figure S6). These results suggest that, while ztf-16 may have a general role in proper amphid channel morphology, its function in glial plasticity may be specific to dauer animals, consistent with our findings that ztf-16 appears to act postembryonically to regulate ver-1 expression.

We previously showed that AMsh glia morphological plasticity plays an important role in controlling shape changes of the associated AWC sensory neurons (Procko et al. 2011). Indeed, in the three ztf-16 mutant animals that we examined by EM, we found that the AWC wing-like cilia that are ensheathed by the AMsh glia failed to expand as they normally do in wild-type dauers (Figure 9, B and C). In non-dauer adult or fourth-stage (L4) animals, AWC wing morphology is only mildly affected in ztf-16 mutants as assessed using an odr-1 promoter::yfp reporter (L’Etoile and Bargmann 2000). Specifically, 100% of AWC neurons of wild-type adults had normal AWC wing morphology, while 88 and 83% of ns171 and ns169 mutants had normal AWC wing morphology, respectively (n = 40 for each strain; Figure 5, G and H). Furthermore, AWC-mediated attraction to the odorant benzaldehyde (Bargmann et al. 1993) is only somewhat reduced in ztf-16 mutant adults (Figure 5I). These findings are consistent with the hypothesis that changes in AMsh glia influence shape changes of associated AWC neurons (Procko et al. 2011).


Morphological changes are commonplace for both neurons and glia in the development and homeostasis of the vertebrate nervous system. How these structural changes in glia are controlled, and whether glial and neuronal shape changes are related, has been largely unexplored. We previously demonstrated that dauer-induced morphological remodeling of the two AMsh glial cells of C. elegans influences concomitant changes in the glia-ensheathed AWC sensory neurons (Procko et al. 2011), suggesting that this setting is appropriate for investigating mechanisms and functions of glia remodeling. We showed that glia remodeling depends on the transcription factor ttx-1 and its direct downstream target, the receptor tyrosine kinase ver-1, whose transcription is induced by dauer entry and high temperature, a dauer stimulus (Procko et al. 2011). Here we demonstrate that, in addition to ttx-1, the transcription factor ztf-16 is required for both ver-1 expression and dauer-induced AMsh glia remodeling. Furthermore, EM analysis of dauer animals suggests that the AWC wing-like cilia fail to take on their expanded overlapping morphology in ztf-16 mutants, most likely as a result of a failure in glia remodeling. Our results are consistent with a model whereby the transcriptional regulators TTX-1 and ZTF-16 act independently through distinct binding sites to regulate ver-1 and perhaps other genes required for AMsh glia remodeling.

How might ztf-16 function be regulated? ztf-16 was recently shown to interact with the Nemo-like kinase LIT-1 in a yeast two-hybrid assay (Oikonomou et al. 2011). Intriguingly, lit-1 expression is strongly induced in dauers by the DAF-12 nuclear hormone receptor, which integrates dauer neuroendocrine signals to promote dauer entry (Shostak et al. 2004). Furthermore, ztf-16 mutants possess similar defects in AMsh glial compartment morphology to those of lit-1 mutants (Oikonomou et al. 2011). These observations raise the possibility that LIT-1 kinase may control ZTF-16 function in amphid glia. However, we found that two different alleles of lit-1 had no defects in ver-1 promoter::gfp expression (data not shown), suggesting that LIT-1 is unlikely to control ZTF-16 function in this context. Nonetheless, AMsh glia remodeling requires membrane growth and is therefore likely mediated by extensive changes in the glial cytoskeleton. LIT-1 was proposed to regulate embryonic aspects of AMsh morphogeneis through physical interactions with the Wiskott–Aldrich Syndrome Protein and actin (Oikonomou et al. 2011). Thus, it is possible that LIT-1 and ZTF-16 function together in processes distinct from ver-1 expression to control glial remodeling.

If ZTF-16 does physically interact with other factors, it is possible that these interactions occur via the two amino-terminal or two carboxy-terminal C2H2 zinc-finger domains, which are unlikely to be required for DNA binding (Large and Mathies 2010). Indeed, the carboxy-terminal zinc fingers of the related Ikaros transcription factor enable dimerization of the protein (Sun et al. 1996). However, in our ver-1 expression rescue studies, we found that the carboxy-terminal zinc fingers are dispensable for ztf-16 function. Nonetheless, different ZTF-16 isoforms may fine-tune ZTF-16 activity, as is the case for Ikaros, whose activity can be controlled by dimerization with nonfunctional isoforms of the protein (Sun et al. 1996).

What other genes control glia remodeling? Most of the mutations that we identified in our screen were alleles of one of three different genes, ztf-16, ttx-1, or tam-1, suggesting that the screen was close to saturation. However, our screen selected for genes involved in controlling ver-1 expression at high temperatures, and not for genes specifically required for dauer induction of ver-1. Indeed, it remains unclear how dauer signals that induce ver-1 transcription are perceived by the AMsh glia. These signals may be direct neuroendocrine signals from amphid sensory neurons [e.g., the TGF-β ligand DAF-7 (Ren et al. 1996)], secondary signals induced as a result of dauer entry (e.g., radial shrinkage of the body circumference), or environmental signals perceived directly by the glia. It is possible that mutant screens assessing ver-1 expression specifically in dauer animals, rather than in non-dauer adults, may uncover these signals. Direct assessment of glial remodeling, rather than reliance on ver-1 expression as a proxy, may reveal additional components functioning in parallel to or downstream of ver-1 to promote glial plasticity.


We thank P. Sengupta and B. Podbilewicz for strains. Other strains used in this study were provided by the Caenorhabditis Genetics Center, which is supported by the National Institutes of Health (NIH). A. Fire, C. Popovici, and R. Roubin provided plasmids and A. Singhvi provided the image of ttx-1 mutants. This work was supported by NIH grants R01NS073121, R01HD052677, and R01NS064273 to S.S.


  • Received December 13, 2011.
  • Accepted January 20, 2012.

Available freely online through the author-supported open access option.

Literature Cited

View Abstract