Mutations that affect channel opening of innexin hemichannels in the C. elegans gonad

In C. elegans, gap junctions couple cells of the somatic gonad with the germline to support germ cell proliferation and gametogenesis. We previously characterized a strong loss-of-function mutation (T239I) affecting the second extracellular loop (EL2) of the somatic INX-8 hemichannel subunit. These mutant hemichannels form non-functional gap junctions with germline-expressed innexins. Here we describe the characterization of mutations that restore germ cell proliferation in the T239I EL2 mutant background. We recovered seven intragenic mutations located in diverse domains of INX-8 but not the EL domains. These second-site mutations compensate for the original channel defect to varying degrees, from nearly complete wild-type rescue, to partial rescue of germline proliferation. One suppressor mutation (E350K) supports the innexin cryo-EM structural model that the channel pore opening is surrounded by a cytoplasmic dome. Two suppressor mutations (S9L and I36N) may form leaky hemichannels that support germline proliferation but cause the demise of somatic sheath cells. Phenotypic analyses of three other suppressors reveal an equivalency in the rescue of germline proliferation and comparable delays in gametogenesis but a graded rescue of fertility. These latter mutations may be useful to probe interactions with the biochemical pathways that produce the molecules transiting through soma-germline gap junctions.


Introduction
Gap junctions are nearly ubiquitous in multicellular animals. The molecular constituents of gap junctions differ in chordates (connexins) and non-chordates (innexins), but their properties and biological functions are remarkably similar (recently reviewed in [1]). Although it is still unclear why a different class of gap junction molecule emerged within vertebrates, the ubiquity of gap junctions themselves suggests that being coupled is essential for most multicellular forms of life, and the functions of gap junctions are many. The sizes of connexin and innexin gene families within species further attest to a diversity of function. Coupling may coordinate activity of cells within a tissue, transfer metabolites and signals between cells, or conduct electrical currents. Gap junction coupling and the composition of junctional channels between cells can also dramatically change over developmental time, though the rationale driving these changes is mostly unknown. We are studying the role that gap junctions play in the somatic control of germline development in the C. elegans gonad. Two symmetric gonad arms extend anterior or posterior of a central uterus and vulva ( Figure   1). Initially germ cells proliferate in a single small pool which becomes partitioned by migration of somatic cells from each of the developing gonad arms late in the second larval stage; these somatic cells eventually divide and develop to form the gonadal sheath, spermatheca, and uterus [2]. A somatic distal tip cell (DTC) occupies the leading edge of each expanding gonad arm, and the DTC establishes a stem cell niche supporting germ cell proliferation by producing Delta-class ligands LAG-2 and APX-1, which activate GLP-1/Notch receptors on germline stem cells [3][4][5][6]. At adulthood the soma of a gonad arm includes the DTC, 5 pairs of sheath cells and the spermatheca. The Sh5 pair of sheath cells is the most proximal (closest to the uterus) and is connected to a constriction of the distal portion of the spermatheca. A series of sheath cell contractions, coordinated with the dilation of the distal constriction of the spermatheca, results in the ovulation of a maturing oocyte into the spermatheca, where fertilization and the completion of the meiotic divisions occur. Embryos exit the spermatheca through the spermathecal-uterine valve and enter the uterus, where embryogenesis generally proceeds for a short time before egg laying through the vulval opening ( Figure 1). From the primordial germ cells to the fully developed gonad, the soma and germline are coupled by two classes of gap junction channels [7]. The somatic DTC and sheath express the inx-8 and inx-9 pair of recently duplicated innexins, which constitute an operon (a single innexin is found at this locus in C. briggsae). Either gene can rescue inx-8(0) inx-9(0) null mutants, and we regard the hemichannels that INX-8 and INX-9 subunits form as homomeric (although there may be a difference in phosphokinase recognition sites). Germ cells do not proliferate in inx-8(0) inx-9(0) mutants (~4 per gonad arm). Germ cells express inx-14, inx-21, and inx-22, and their corresponding protein products assemble into two classes of heteromeric hemichannels, composed of INX-14 in conjunction with either INX-21 or INX-22. As in inx-8(0) inx-9(0), germ cells in inx-14(0) or inx-21(0) null mutants fail to proliferate, and animals are sterile. In inx-22(0) null mutant hermaphrodites, the germline appears unaffected; however, feminized inx-22(0) mutants fail to properly inhibit meiotic maturation in the absence of the major sperm protein (MSP) meiotic maturation signal, and unfertilized oocytes are ovulated into the uterus [8]. Clearly the two classes of germline hemichannels enable different processes, and presumably the nature (or quantities) of the molecules traversing the corresponding gap junction channels are different.
The failure of germ cells to proliferate in the absence of soma-germline gap junctions obscures any role that junctions may play during later development of the germline. In an attempt to address this issue, the lag-2 promoter was used to express INX-8 in the DTC, but not sheath cells, in inx-8(0) inx-9(0) mutants using extrachromosomal arrays [7]. In these inx-8(0) inx-9(0); Ex ] animals, germ cell proliferation was restored to ~1/2 wild-type levels, but few progeny (avg. ~2) and dead embryos (~20) are produced compared to the wild-type (brood size ~300). This level of germline proliferation mirrors that observed when sheath cell precursors are ablated from developing gonad arms [9].
To explore specific roles that the soma plays in nurturing the germline, we have focused on manipulating inx-8 somatic hemichannels in an inx-9(0) null background. Under these conditions, all somatic hemichannel functions are dependent on INX-8. Non-null reduction-of-function inx-8 mutations may allow for germ cell proliferation and reveal later phenotypes associated with gap junction functions. Such mutant gap junctions might preferentially restrict the passage of certain molecules or exhibit unusual gating properties.
Here we describe a series of non-null inx-8 mutations derived from a genetic suppressor screen based on restoration of germ cell proliferation in a severe inx-8 loss-of-function background.
The starting mutant INX-8 protein has a T239I change in the second extracellular loop. We previously showed that INX-8(T239I) is capable of supporting formation of gap junction channels with the germline, but these channels are non-functional [7]. Suppressor mutations were isolated at very low frequency, but, surprisingly, were found to be located widely in the molecule. The cryo-EM structure of the C. elegans INX-6 hemichannel was recently determined to be an octamer [10], and it has sufficient amino acid identity with INX-8 to allow for estimation of the positions of suppressor mutations in the INX-8 tertiary structure. Characterization of the phenotypes associated with suppressor mutations are consistent with this cryo-EM model of INX-6. We recently used one of these suppressor mutations to show a requirement for malonyl-CoA transfer from soma to germline to support early and continued embryogenesis [11]. Other suppressor mutations may be similarly useful for investigating genetic interactions with candidate biochemical pathways found in the soma that produce biomolecules that transit through gap junctions to control germline development.

Strains and Genetics
Worms were grown on standard NGM agar plates. All brood counts reported were done at 20 o C. Bristol N2 was used as the wild type, and mIs11 IV was used to balance inx-8 inx-9 mutants. Strain DG3954 inx-8(tn1513) inx-9(ok1502); tnEx195[sur-5::gfp; inx-8(+) inx-9(+)]; tnIs107[inx-8p::mCherry; str-1::gfp] served as the foundation for the suppressor screen. To screen for suppressors, DG3954 was mutagenized with EMS following standard protocols (50 mM, 4 hrs). Mutagenized hermaphrodites were plated singly, grown at 20 o C, and their F2 progeny were screened on a fluorescence dissecting stereomicroscope 5-6 days later for the presence of sur-5::gfp(-) animals with extended and reflexed inx-8p::mCherry gonad arms. Plates identified with such animals were then propaged to identify single hermaphrodites giving rise to candidate suppressors of inx-8(tn1513) inx-9(0). In total, 34 separate mutageneses were performed and 37,300 mutagenized hermaphrodites were plated. In seven mutageneses, a subset of at least 300 plates were scored for sterility, and the overall percentage of sterile or near-sterile mutagenized animals averaged ~35% (range 15-50%). If we assign a very conservative brood size of just 5 F1 progeny produced on average, this screen would represent ~240,00 mutagenized haploid genomes. These suppressors therefore represent rare mutations. Allele designations for suppressor mutations and corresponding amino acid changes follow (tn1789 was used as the representative for E350K in brood counts as this mutation was independently isolated twice).
Photos of phenotypes primarily employed a Zeiss motorized Axioplan 2 microscope with a 63x PlanApo (numerical aperture 1.4) objective lens and an AxioCam MRm camera with AxioVision acquisition software. Germ cell counts were made by photographing dissected, fixed, and DAPIstained gonad arms. Sperm nuclei were not included in the final counts (highest counts were ~60 sperm nuclei).

Results
We previously isolated a strong loss-of-function inx-8(tn1513lf) allele in a null inx-9(0) background [7], producing on average ~21 germ cells per gonad arm (cf. >1000 per wild-type arm [9]). tn1513 encodes a T239I change in the second extracellular loop (EL2). A tagged INX-8(T239I)::GFP construct was introduced as a multi-copy extrachromosomal array into inx-8(0) inx- In this background, lag-2p drives INX-8::GFP expression in the DTC and rescues germ cell proliferation; any expression of INX-8::GFP in the sheath (especially apparent in the proximal arm) can be attributed to INX-8(T239I)::GFP. We showed that this proximal arm expression was sufficient to support localization of germline INX-22 to gap junction plaques, indicating that INX-8(T239I) is capable of assembling into somatic hemichannels and forming gap junctions with germline hemichannels; the resultant channels, however, fail to rescue proximal arm gap junction function [7]. Potentially INX-8(T239I) causes a partial blockage in the channel it forms with the germline (total blockage might be expected to prevent formation of channels with germline hemichannels). inx-8(tn1513) inx-9(0) was therefore an attractive candidate for a suppressor screen to find compensatory mutations that might "unblock" INX-8(T239I) channels.
To carry out the screen, a strain was constructed including the following features: (1) the inx-8(tn1513lf) inx-9(0) mutation to be suppressed; (2) a means of visualizing the gonad at the dissecting microscope level, for which we designed an mCherry construct driven by the inx-8 promoter (inx-8p::mCherry) that was integrated into the genome as a multi-copy array; and (3)   hemichannel functions at nearly wild-type levels ( Table 1). The cryo-EM model for INX-6 predicts

Suppression with reduced germline proliferation, delayed gametogenesis, and reduced brood size
These mutants share approximately equivalent restoration of germline proliferation, exhibit a marked delay in gametogenesis, but display a graded rescue of brood-size numbers. D24N is located in the N-terminus near TM1. It lies close to a second aspartate, D21, that by homology to INX-6, is predicted to contribute to anchoring of the N-terminus to the cytoplasmic dome [10]. We recently extrachromosomal arrays recapitulates these phenotypes, though brood sizes are smaller (Table 1).
Because INX-8(T239I, D24N) behaves as a strong reduction-of-function allele of inx-8 that produces moderate brood sizes, it was used to show genetic interactions with conditional mutants in the fatty acid synthesis pathway; this led to further genetic experiments demonstrating requisite transfer of malonyl-CoA from the somatic sheath to the germline [11].
E350K is located in the C-terminus. INX-8(T239I, E350K) gonad size and germline proliferation is comparable to INX-8(T239I, A288V), but virtually no viable progeny are produced ( Table 1). Lack of progeny prevented assessing the relationship of onset of ovulation to the L4 molt. The position of E350K is coincident with a site (Y356) in INX-6 proposed to contribute to inter-subunit interactions in the cytoplasmic domains of the hemichannel. These interactions support the formation of a dome-like structure that surrounds the hemichannel pore face. Disruption of this cytoplasmic dome by the E350K change may enlarge the pore face and allow increased access to the channel, which may allow for an increased probability of molecular transfer across the T239I-induced site of channel restriction. INX-8(T239I, E350K)::GFP extrachromosomal arrays rescue germline proliferation but not production of embryos in inx-8(0) inx-9(0) ( Table 1).

Figure 5.
Ovulation is delayed in T239I, A288V mutants in relation to the L4 molt. (N2 avg. from [11]). Studies of tryptophan-substitution mutations in the TM1 domain of the Drosophila shakingB(lethal) innexin identified several sites which exhibited increased conductance in relation to wild-type ShakB(L) [13]. These mutants also displayed open-hemichannel activity when expressed in unpaired Xenopus oocytes, which compromised oocyte survival; however, when paired, the survival of oocytes expressing these constructs improved, likely due to reduction of free hemichannels as they assembled into gap junctions. Indeed, that opening of hemichannels may actually drive gap junction formation was originally proposed and demonstrated in a study of connexins expressed in Xenopus oocytes [14]. We speculate that I36N and S9L may enhance germline proliferation in a T239I background by increasing channel conductance in relation to wild-type INX-8; however, this may result in hemichannels that are leaky in the cell membrane when unpaired. Sh5 expresses INX-8 at very high levels, at least partly due to the fact that upon ovulation the oocyte endocytoses gap junctions formed with the soma, thus depleting Sh5 of a significant number of INX-8/9 subunits that must be replaced [7]. Sh5 may therefore be especially susceptible to a defect resulting in open hemichannels. Additionally, because oocytes have not developed and advanced into the proximal arm at the time of Sh5 swelling, it appears that when hemichannels initially arise in Sh5 there are no available pairing partners in the germline with which they might form gap junction channels (sperm do not form gap junctions with sheath cells). This may be exacerbated if there is a delay in gametogenesis as seen with other suppressors, though we have not yet attempted to document this for I36N or S9L. The more severe phenotype of INX-8(T239I, S9L) suggests that other somatic cells that express inx-8 at levels lower than Sh5 are also susceptible to this gain-of-function in the corresponding hemichannels, and S9L may create a greater degree of "openness" than I36N. (D, E) T239I, S9L hemichannels also affect Sh5 but loss of membrane integrity appears to include other cells or organelles. Sp, spermatheca; dis, distal arm; vul, vulva; Sh, sheath.

Expression of suppressor mutations in the absence of T239I
Our interpretation of the nature of the suppressor mutations isolated in this screen is that they might result in increased conduction of the T239I hemichannel. As such, it was of interest to see if these mutations by themselves may confer unusual phenotypes. We expressed each of the inx-8 single suppressor mutations on extrachromosomal arrays to rescue inx-8(0) inx-9(0) mutants (Table 1).
Because expression from such arrays can be variable, we focused on potential qualitative rather than quantitative effects.
As might be predicted, INX-8(M117T)::GFP rescued inx-8(0) inx-9(0) mutants robustly. Tagged versions of A288V, E350K, and I36N also rescued, but to reduced fertility. Multiple independent lines for each of these mutations were generated, and no unusual phenotypes were identified that were seen consistently across each line representing a particular mutation. not detected in embryos. Therefore D24N neither appears to make hemichannels on its own nor associates at significant levels with wild-type INX-8 to contribute to hemichannel formation.

Discussion
Germ cells in the adult gonad progress in an assembly-line fashion, providing a snapshot of continual developmental progression at any single time point. In combination with a detailed understanding of its development from embryo through larval stages, the C. elegans gonad offers an excellent model for examining the gap junction relationships and requirements between cell types in a structure from its origin to its final functional form. Although the composition of hemichannels in soma and germline do not change, we expect that the specific requirements for molecules passing through gap junction channels changes with developmental progression from a mitotic state through the stages of meiosis and gametogenesis. Because the molecules that transit through gap junctions are small (<1000-2500 daltons), it is not easy to tag candidates and follow their intercellular passage in vivo. Genetic tools including mosaic analysis and mutations that perturb channel function have been useful in identifying candidates for traversing innexin gap junctions [11,15]. One question entertained in characterizing these mutants was whether there was any evidence for preferential restriction of a particular class of molecules, e.g. negatively charged or positively charged molecules, or restriction by size. At least at this level of investigation none of the phenotypes associated with a particular suppressor mutant suggested such a strict gating, rather the observed phenotypes seemed to vary only by degree.
Only a limited set of suppressors of INX-8(T239I) was isolated, but the suppressors were widely distributed within INX-8. This may suggest that mutations throughout the innexin subunit may increase conduction, but the nature of our screen selected only those that still allowed germ cell proliferation. Three of these--D24N, A288V, and E350K--in the context of T239I behave as reductionof-function inx-8 alleles. They show similar degrees of support for germline proliferation but vary in their effectiveness of rescuing production of progeny, suggesting that the channel requirements for the latter are more demanding. They also support a gap junction requirement for timely gametogenesis, the nature of which is only a guess at present. It could represent the inability to accumulate a necessary factor, or remove an inhibitory factor. The extent of this delay-almost a full day-is sufficiently long that suppressor screens for the restoration of wild-type timing of egg-laying are feasible and may uncover bypass mutations in the responsible pathway.
Because INX-8(T239I, D24N) produces a moderate brood size it has been useful in establishing genetic interactions with fatty acid synthesis genes [11]. However, it may not be the best choice for investigating interactions with candidates required for germ cell proliferation. We propose that E350K likely disrupts the cytoplasmic dome surrounding the hemichannel pore, which might increase flux through the pore entry but would be unlikely to affect the channel constriction in T239I or the narrowest regions of the channel. As such E350K may be a more sensitive gauge for genetic interactions with other mutations affecting proliferation.
S9L and I36N suppression of T239I seem best explained as hemichannels that have open activity when unpaired. Clearly these hemichannels can form junctions, as they restore germ cell proliferation to some degree. The absence of a deleterious effect on the distal gonad arm suggests that most of the somatic hemichannels there are paired with germline hemichannels, or the levels of somatic hemichannels are sufficiently low that open hemichannel activity has a minimal effect on the plasma membrane. Expression levels of INX-8--graded from low in the distal arm to very high in the most proximal region [7]--are consistent with this being the determining factor in whether or not somatic sheath cell membrane integrity becomes compromised. A contributing factor for the Sh5 pair of cells, however, may be the absence of gap junction pairing partners for open hemichannels when they arise. It is possible such an explanation might apply to other widely expressed gap junction mutants for which defects are restricted to only a subset of expressing cells.
Other than being of structure-function interest, do these proposed open-hemichannel mutants have any utility for use in genetic interaction studies? Possibly. Because INX-8(T239I, I36N) arrays have been successfully isolated, by our interpretation due to lower expression levels, it would be possible to integrate these arrays into a chromosome and establish a reduction-of-function inx-8 allele. Because these mutants are viable but display smaller germlines, they may also be candidates for genetic interaction inquiries with other mutants affecting germline proliferation. Additionally, it may be possible to determine if there are any ramifications to increased conduction across somagermline gap junctions.

Conclusions
A genetic screen for rescue of germ cell proliferation in a strong loss-of-function inx-8 mutant yielded a set of suppressor mutations that likely increases conduction through their corresponding hemichannels. These mutations highlight distinct requirements for soma-germline coupling in the distal and proximal gonad arms and have helped identify a gap junction requirement for timely ovulation and gametogenesis that nevertheless does not inhibit the production of healthy broods. One of these suppressors (D24N) has already proved useful for establishing genetic interactions with the fatty acid synthesis pathway and for providing evidence that malonyl-CoA transits through soma-germline gap junction channels [11]. These suppressor mutations may facilitate the elucidation of the molecules delivered through soma-germline gap junctions and their roles in promoting germline development.