mab-31 and the TGF-β pathway act in the ray lineage to pattern C. elegans male sensory rays

Background C. elegans TGF-β-like Sma/Mab signaling pathway regulates both body size and sensory ray patterning. Most of the components in this pathway were initially identified by genetic screens based on the small body phenotype, and many of these mutants display sensory ray patterning defect. At the cellular level, little is known about how and where these components work although ray structural cell has been implicated as one of the targets. Based on the specific ray patterning abnormality, we aim to identify by RNAi approach additional components that function specifically in the ray lineage to elucidate the regulatory role of TGF-β signaling in ray differentiation. Result We report here the characterization of a new member of the Sma/Mab pathway, mab-31, recovered from a genome-wide RNAi screen. mab-31 mutants showed ray cell cluster patterning defect and mis-specification of the ray identity. mab-31 encodes a nuclear protein expressed in descendants of ray precursor cells impacting on the ray cell's clustering properties, orientation of cell division plane, and fusion of structural cells. Genetic experiments also establish its relationship with other Sma/Mab pathway components and transcription factors acting upstream and downstream of the signaling event. Conclusion mab-31 function is indispensable in Sma/Mab signal recipient cells during sensory rays specification. Both mab-31 and sma-6 are required in ray lineage at the late larval stages. They act upstream of C. elegans Pax-6 homolog and repress its function. These findings suggested mab-31 is a key factor that can integrate TFG-β signals in male sensory ray lineage to define organ identity.


Background
Members of the transforming growth factor-beta (TGFβ) family are highly conserved multi-functional cell-cell signaling molecules found in many organisms [1,2]. The basic components of this pathway are secreted ligands, two receptor serine/threonine protein kinases (receptor types I and II), and the Smad (for C. elegans Small and Drosophila mad) proteins [3]. In worms, conventional TGF-β signaling pathways have been identified. They act downstream of two well characterized ligands DAF-7 and DBL-1 [4][5][6], which are referred to as the Dauer pathway and the Sma/Mab pathway, respectively.
The Sma/Mab pathway regulates body size and development of the sensory rays [7]. The secreted ligand encoded by dbl-1 triggers signaling events via sma-6/ daf-4 receptors and Smad transducer molecules sma-2, sma-3 and sma-4 for downstream signaling [8,9]. Mutant animals with defective pathway are small in body size (Small) and mutant males have ray patterning defects in the tail (Male abnormal), i.e. fusion of ray pairs (4-5, 6-7, and 8-9) with different penetrance. While it is well established that the body size is a function of body hypodermis, less is known about the cellular context of the BMP signaling event in male sensory rays. Nonetheless, the wild-type sensory rays would need to express an array of genes to orchestrate developmental decisions that give rise to cells of distinct morphogenetic and functional identities, and dictate cellular positioning and assembly of individual organs.
Most components of the Sma/Mab pathway were identified by forward genetic approaches with a Sma phenotype [10]. More recent works using reverse genetics, DNA microarray analysis and yeast two-hybrid screens continue to identify modifiers, targets, and components of the pathways to uncover hitherto unknown biological roles of existing pathways and novel components [11,12]. Many of the newly identified small mutants, however, have no male tail defects at all [13][14][15]. These results suggest that distinct downstream signaling components may be required for male tail development functions to occur. We report here the characterization of a gene encoding a new TGF-β pathway component, mab-31, from a genome-wide RNAi screen. Animals with attenuated mab-31 activity displayed Mab phenotypes, specifically ray 4,5 fusion and 6,7 fusion. Through subsequent characterization of deletion mutants and genetic analysis, we demonstrate that mab-31 acts in Sma/Mab pathway to pattern specific male sensory rays. Epistatic analysis showed that the Sma/Mab regulatory cascade acts in the ray cell group and more specifically the ray structural cells, to attenuate the function of Pax-6 isoform, mab-18, during sensory ray development. Finally, we provide evidence that mab-31 encodes a nuclear protein working in the TGFβ signaling pathway and may represent a novel modulatory and context-dependent component acting in this sensory organ differentiation process.

Result
Male tail defect in mab-31 mutant resembles phenotypes in Sma/Mab mutants C. elegans male tail is a fan-shaped structure ( Figure 1A) required for copulation. It consists of nine bilateral pairs of peripheral organs known as sensory rays. Each ray consists primarily of a tube like hypodermal layer surrounding two sensory neurons and a glial-like structural cell [16]. Although sensory rays are similar in structure, each has a unique morphology and identity [17][18][19]. Each structural cell interacts with neurons from the same lineage and repels the neighboring structural cells and neurons [20]. Male tail defects in mutants of Sma/ Mab pathway have been characterized as a transformation of ray identity [21], resulting in a high percentage of fusion in rays 4-5, rays 6-7, and rays 8-9 ( Figure 1B). We isolated Mab mutants based on the specific abnormal sensory ray pattern, ray 4,5 and ray 6,7 fusion, through a genome-wide RNAi screen (Yip et al., in preparation). The RNAi bacterial clone I-7J10 from the Ahringer RNAi library could knock-down a nematodespecific gene, Y54E10A.16, on chromosome I. After the knockdown, male animals displayed a ray fusion phenotype and hermaphrodites showed a reduced fecundity. The gene harbored in this fragment has no previously characterized function and yet this knockdown gave a specific ray phenotype in male mutants, resembling that of the dbl/sma mutants.
The Y54E10A.16 gene is embedded in intron 9 of cogc-1, and the two genes are transcribed in opposite directions (Additional file 1). Null allele of cogc-1 did not show any male tail phenotype. The Y54E10A.16 (tm2718) mutation kindly provided by NBRP, Japan, has a 435 bp deletion. This lesion is predicted to result in a truncated product lacking one-third of protein towards the C-terminus. Y54E10A.16(tm2718) mutants exhibit ray defects, with fusion of rays 4-5 (17%), rays 6-7 (60%) and rays 8-9 (31%) ( Figure 1C) ( Table 1, line 2). The ray patterning defects of mab-31 are reminiscent of those of dbl-1, sma-6, daf-4, sma-2, -3, or -4 (existing mutants in the Sma/Mab pathway) and displayed a penetrance matching the strongest ray phenotype of dbl/sma pathway mutations, e.g., null mutation of sma-6. We showed that the frequency of the ray fusion defects could not be enhanced further by feeding tm2718 males with I-7J10 RNAi bacteria. Such a result suggests that tm2718 behaves like a null mutation of Y54E10A. 16. This gene is essential for ray patterning and display of the distinct male tail defect in mutant animals; the gene was designated as mab-31.
Each wild-type ray consists of dendritic processes of two ultra-structurally distinct sensory neurons, RnA and RnB. The dendritic endings of these neurons are held at an opening to the environment by a support cell, i.e. the structural cell (Rnst) ( Figure 1D). Structural examination of the abnormal ray in a ray-fusion mutant by electron microscopy showed the presence of a fused structural cell and separate neuronal processes [21]. We examined the cellular defect of the fused rays in mab-31(tm2718) males using two cell-specific reporters, E1-GFP [22] and ppkd-2::gfp2 [23], which mark Rnst and RnB cells, respectively. In both sma-6(wk7) ( Figure 1E) and mab-31(tm2718) ( Figure 1F), the structural processes were completely fused in 70% of defective rays at a frequency consistent with the observable defects revealed by DIC microscopy. The ppkd-2::gfp2 reporter was active in neuronal B processes from ray 1 to ray 9, but not ray 6 ( Figure 1J). In both sma-6 and mab-31 mutants, the neuronal processes in the fused rays were well-separated ( Figure 1K and Figure 1L). These results suggest that cellular defects in fused rays of both sma-6 and mab-31 mutants are similar. Abnormal fusion was observed only in extended distal processes of structural cells but not the neuronal processes. All existing mutants in Sma/ Mab pathway with ray phenotype also displayed small body phenotype [24]. mab-31(tm2718), however, is not Small and exhibits only a typical Sma/Mab ray fusion phenotype. The average body length of mab-31(tm2718) is 1.25 ± 0.07 mm (n = 100), in contrast to that of wildtype animals (1.23 ± 0.06 mm, n = 100) and sma-6(wk7) mutant (0.81 ± 0.03 mm, n = 100). mab-31(tm2718) mutation impacted other developmental processes also. For example, in mutant males, backward movement in response to an anterior touch is uncoordinated. They have crumpled copulatory spicules, which render the Figure 1 Sensory ray patterns of mutants in Sma/Mab pathway. Wild-type C. elegans male tail contains nine bilateral pairs of sensory rays (A and G). Their structural and neuronal (B type) processes were located by E1-GFP (D) and ppkd-2::gfp2 (J) markers, respectively. sma-6 (B and H) and mab-31 (C and I) mutants displayed rays 4-5 fusion (arrows) and rays 6-7 fusion (arrowheads) defects. In the defective rays, structural cell processes were fused (E and F), while the neuronal processes were separated (K and L). Ventral view. Scale bar = 20 μm. Specification of sensory rays requires both mab-31 and sma-6 at the L3/L4 larval stage Ray identity determination is a dynamic process of ray precursor (Rn) cells differentiation in late larval stages of C. elegans male development. The underlying process of cellular specification of Rn cells is not well understood. Therefore, we examined and compared ray cell cluster profiles of wild-type, sma-6(wk7), and mab-31 (tm2718) animals. To achieve this, we traced the Rn cells lineage with an apical junction marker ajm-1::GFP.
We were particularly interested in differentiation of R6 and R7, since rays 6-7 fusion was the most obvious phenotype in both mab-31 and other sma mutants. When R(6/7).a and R(6/7).p cells were born, there was no detectable difference in wild-type (Figure 2A), sma-6 ( Figure 2E) and mab-31 mutants ( Figure 2I). In the wild-type animal, R7.aa and R7.ap (descendents of R7.a) were produced along the A-P axis and stayed on the dorsal side of R7.p ( Figure 2B). However, in sma-6 and mab-31 mutants, these two cells were present on the ventral side of R7.p after their birth ( Figure 2F and 2J). In these mutants, the abnormal ray cell group derived from R7 sat next to that of R6 ( Figure 2G and 2K) and subsequently, their structural cells were juxtaposed to each other and were fused together ( Figure 2H and 2L). However, no cellular abnormality was observed in differentiation of R6a.a, R6a.p. or R6.p cells, in both mutants, as compared with those in wild-type animals. The defect in rays 6-7 fusion involved mis-localization of ray 7 precursors to ray 6 precursors during ray identity determination window at the L4 larval stage. Our observation suggested that both sma-6 and mab-31 mutants have a wild-type ray lineage with ray precursor cells born at the right time. The defects in these mutants were simply due to transformation of the ray cell identity, which dictates the cell cluster's positioning with respect to other ray cell groups. Indeed, similar defects were subsequently observed in other sma mutants (sma-4 and dbl-1, data not shown) also, inferring that the mab-31 gene probably acts in the same canonical dbl/sma pathway. It was shown that the ray pattern in the adult male tail results from targeting of ray structural cell processes to specific sites, called papillae, in the L4 lateral epidermis [20]. Fused rays arise when multiple papillae are juxtaposed to each other in a cluster. Some of papillae were displaced in mab-31(tm2718) (65%) and sma-6 (wk7) (45%) mutants (N = 50). For example, the ray 7 papillae in mutant animals were not found in their native position but often at an ectopic position just next to the ray 6 papillus. This displacement is also consistent with our observation that ray 6 and 7 cell groups were next to each other during ray development as described above (Figure 2G &2K). The same feature applies to papillae of ray 4 and 5, and to those of ray 8 and 9.
We further used a genetic approach to determine the functional relationship between sma-6 and mab-31 in the specification of sensory rays. Both mab-31(tm2718) and sma-6(wk7) carry recessive null mutations for ray patterning defects. We compared the frequency of rays 6-7 fusion phenotype in each single mutant as well as in double mutants. Worms doubly mutant for mab-31 and sma-6 do not exhibit an enhanced frequency of fusion for rays 6-7 (58%) compared to mab-31 single mutants (60% ; Table 1). Thus, they are unlikely to be working in parallel for the ray 7 specification. The null allele of sma-6(wk7) showed both Sma and Mab phenotypes. However, in a weak loss-of-function allele sma-6(e1482), the male tail defect could not be observed. It was suggested that the residual activity of sma-6(e1482) is sufficient for the male tail development, but its level is below the thresholds required for body size regulation [25]. Interestingly, a reduction of sma-6 activity in e1482 mutant background could enhance mab-31(RNAi) rays 6-7 fusion phenotype by four-fold, as compared with the same treatment in wild-type animals ( Table 1, lines 1 & 6). This synergistic effect strongly argues that these two genes act in the same process and pathway. Furthermore, phenotypes of mab-31(tm2718);sma-6(e1482) double mutants resembled those observed in the mab-31 single mutant (Table 1, lines 2 & 7), suggesting that mab-31 is an indispensable component required for Sma/Mab pathway signaling during male tail development.

MAB-31 may cooperate with SMAD inside nucleus
In the Sma/Mab pathway, SMA-6 receptors transduce the extracellular signal to the effectors, SMA-2, SMA3, and SMA-4, which regulate the expression of downstream targets. Would mab-31 be one of the targets with expression regulated by the SMA-6 receptor and mediated by SMAD transcription factors in rays? To address this question, we first characterized the expression pattern of mab-31 gene. A 2 kb 5' upstream sequence of the mab-31 translational start was utilized for construction of a transcriptional mab-31::gfp reporter. The promoter sequence contained regulatory elements required for ray development, as it could be employed for expressing mab-31 cDNA to rescue both male tail ( Table 1, lines 2 & 8) and other phenotypes in corresponding mab-31(tm2718) mutants. Indeed, the same 2 kb mab-31 5' upstream sequence could be used to drive expression of sma-6 cDNA and rescue sma-6(wk7) mutant ray phenotype (from ray 6,7 fusion drops from 38% to 16%, Table 1, lines 3 & 9). But the significant rescue was not observed in body length (0.82 ± 0.04 mm for sma-6(wk7) of the animals and 0.87 ± 0.06 mm for the same mutant with a transgene). This mab-31 2 Kb 5' franking sequence was active in gut cells of the pretzelstage embryo and throughout the larval stages. In adult males and hermaphrodites, gfp reporter signal was observed in the pharynx, body hypodermis, and intestine. Interestingly, the type I receptor SMA-6 is expressed in all these tissues overlapping with the mab-31 pattern [25]. Yet, while sma-6 mutant animals have a small body phenotype, mab-31 mutants do not. Such a difference is probably due to the fact that mab-31 is not required nor expressed at a significant level at the early larval stage when sma-6 and smad functions are needed [26]. On the other hand, mab-31 expression is also observed in support cells of neuronal sensilla, like amphid socket cells ( Figure 3B) and phasmid socket cells ( Figure 3D), where no expression of sma-6 has been documented. In the wild-type male tail, the mab-31 transcriptional reporter expression was detected in structural cells highlighting the cell bodies and processes of all sensory rays ( Figure  3F). Most importantly, the expression level was not changed in either sma-6 or sma-4 mutant backgrounds (data not shown). Indeed, high copies of mab-31 functional transgenes in ray structural cells cannot by-pass the requirements of either sma-6 or sma-4 in rays (Table 1, lines 3 & 10, 11 & 12). These observations suggest that mab-31 is not acting transcriptionally downstream of the Sma/Mab signaling cascade.
We explored the molecular function of MAB-31 in the Sma/Mab pathway by evaluating its cellular localization. MAB-31 is highly conserved across several Caenorhabditis species (Additional file 1), but it does not resemble any protein other than its worm counterparts. Based on the amino acid sequence, a stretch of polypeptide (RKRREK) was predicted as a nuclear localization signal (NLS) (PredictNLS online) implicating it to function inside the nucleus. To ascertain its in vivo subcellular localization, we generated transgenic animals carrying a MAB-31 translational fusion with GFP tagged at its Cterminus. This MAB-31::GFP protein was a functional equivalent to endogenous MAB-31, since it could fully rescue rays fusion phenotypes in mab-31(tm2718) (Rescue efficiency = 100%, N = 50) ( Table 1, line 8). MAB-31:GFP was found to be localized in nuclei ( Figure  3H), but not in cytoplasm in all developmental stages that we had examined. It was produced at the pretzelstage and then in all expressing cells implicated by the transcriptional reporter. Localization of MAB-31::GFP protein was altered neither in sma-4(e729) nor in sma-6 (wk7) mutants (N = 20 in both cases) throughout the ray differentiation period and in the adult stage. Thus, MAB-31 is likely acting constitutively in the nuclear compartment rather than being shuttled into the nucleus, like the Smad proteins.
Sma/Mab signaling is modulated by epistatic interactions among mab-18, mab-31, and unc-130 in ray patterning Several components in Sma/Mab pathway were shown to negatively regulate the mab-21 gene in ray patterning [5]. In mab-21 mutants, the ray 6 structural cell apparently undergoes a specific transformation into a ray 4 morphogenetic identity and ray 6 is fused with ray 4. Double mutant of mab-21 with either sma-6 or sma-4 resulted in rays 6-4 fusion phenotype, which mimicked the Mab-21 phenotype. We could also establish a similar epistatic relationship between mab-31 and the components in mab-21 pathway. In the double mutant of mab-31(tm2718); mab-21(bx53), only Mab-21 phenotype could be observed ( Figure 4A). Thus the role of mab-31 is similar to Sma/Mab pathway, which negatively regulates mab-21 during the specification of ray 6.
Since the specification of ray 6 by Sma/Mab pathway could be used to dissect genetic function for mab-31, we extended our study to look for the epistatic relationship of mab-31 with other candidates. Pax-6 is a conserved transcriptional factor that plays a variety of roles during metazoan development, primarily in patterning elements of the nervous system [27]. The C. elegans pairbox-less Pax6 isoform, mab-18, acts cell-autonomously in ray 6 [28,29]. Loss-of-function mab-18 also results in transformation of ray 6 to ray 4 identity (Figure 4C). In double mutants of mab-31 and mab-18, only Mab-18 ray fusion phenotype was observed ( Figure 4A and 4E). The same phenotype was also observed in double mutants of mab-18 with either sma-4 or sma-2 (data not shown, N>100). Thus, Sma/Mab pathway is proposed to have acted upstream of mab-18.
The fork-head transcription factor unc-130 functions in both ventral body muscle and male tail developments [30]. unc-130 acts as a negative regulator of unc-129 and works autonomously to repress unc-129 expression in ventral body muscles. In the male tail, unc-130 is required for keeping ray 6 identity independent of unc-129. unc-130(ev505) mutant displayed a rays 6-4 fusion phenotype of 82% of the animals. In the double mutant of mab-31(tm2718);unc-130(ev505), only the rays 6-7 fusion phenotype was observed ( Figure 4A). This finding suggests that unc-130 works further upstream of Sma/ Mab pathway and negatively regulates mab-31 in the ray patterning. Based on all these observations, we propose a novel genetic cascade acting uniquely in the male tail sensory rays specification; mab-18 is the downstream target of mab-31. MAB-31 is localized in the nucleus, and we predict that it acts downstream of the SMA-6 receptor (and also DAF-4 receptor, data not shown), and this TGF-β signaling cascade is negatively regulated by unc-130 acting as the most upstream regulator in this process.

Discussion
Interpretation of the Sma/Mab signal by the ray lineage during male development In this paper, we have reported the characterization of a nuclear factor encoded by mab-31 in the C. elegans for sensory ray patterning. mab-31 mutants present no small body phenotype, but exhibit male tail defects resembling the ray fusion phenotypes in existing Sma mutants of Sma/Mab pathway. We have shown that both MAB-31 nuclear factor and SMA-6 receptor are required to function in the ray lineage and the structural cells during the ray patterning process. They regulate the ray patterning event by altering the position of the R7 derived ray cell cluster and how it associates with the neighboring cluster at the late L4 males. Finally, we establish the genetic relationship of conserved transcriptional regulators in ray patterning, such as unc-130 and mab-18, with Sma/Mab pathway components through epistatic analysis.
In Sma/Mab pathway, the signaling ligand dbl-1 is expressed primarily in motor neurons along the ventral nerve cord in both hermaphrodites and males [5,6]. We have previously showed that ectopic expression of Xenopus BMP4 in these neurons could functionally rescue body size [15] and corrected the ray fusion defect (Wong and Chow, unpublished data) in dbl-1 mutant. Expression of the dbl-1 ligand in neuronal processes along the body is sufficient for its function in the ray patterning, probably through diffusion of this extracellular ligand. Hence, the specificity of Sma/Mab signaling for ray patterning is likely to be controlled by receptors residing in cellular components of the sensory rays but not by the source of the signal acting at a distance. However, expression patterns revealed by transcriptional reporters for TGF-β receptors and Smads did not reveal their expression in any specific cells within the sensory rays [8,25]. For example, hypodermal expression of sma-6, which is governed by a GATA sequence in the sma-6 promoter, was shown to be necessary for maintenance of body length [8]. This expression, however, was not sufficient to rescue the male tail mutant defect (Ho and Chow, unpublished data). In this report, we have shown for the first time that the mab-31 promoter with activity in the ray structural cell ( Figure 3F) could functionally express sma-6 transgene and successfully rescue the ray fusion defect of sma-6 mutant (Table 1, line 9) even when the structural cell-specific sma-6 regulatory element remains undefined. This finding infers that both mab-31 and sma-6 likely have overlapping expressions in the same cells in the ray sublineage including the structural cells as part of the executing machinery of the dbl-1 signaling event. It is likely that their expression is required even earlier in the structural cell precursors R7.ap or R7.a. Without a good structural cell-specific promoter active at the early stage of this cell type or in its precursors, direct experimental verification of this notion would not be possible. Neither would we be able to rule out the possibility that simultaneous expression of sma-6 nor mab-31 in both hypodermis and structural cells is necessary for establishing correct ray identity. In addition, with the cellular defects revealed in mab-31, sma-6 and other sma mutants which converge on having an abnormal division plane of the R7.a, the action of the dbl-1 pathway receptors and relay molecules may indeed act even as early as the R7.a cell is born.
Examination of morphology of ray structural processes in fused rays of mab-31, sma-6, sma-4, and dbl-1 mutants also showed that fusion of structural cells but not fusion of neuronal processes within rays is the primary defect. This observation is similar to the results of the histological examination of the fused ray in mab-21 mutant, which implicates that fused structural cells are the primary cause of ray fusion after the ray identity was altered [21]. Collectively, these evidences support the notion that single ray formation acts through the ray lineage and manifested through assumption of a unique identity which can be manifested by the structural cells. The outcome is abnormal ray fusion, when structural cells of two adjacent rays lose specificitydeterminants required to keep the rays apart as independent entities [16].

Biological role of Sma/Mab signaling in sensory ray patterning
Multiple evidences have shown that Sma/Mab signaling in the structural cells is required in patterning of sensory rays. The attachment of the structural cells to specific positions of the epidermis and cuticle at late L4 stage dictates where the papillae will be formed. In turn, the ultimate position of a ray is determined [20]. When the attachments are altered in different mutants, such as in sma mutants or many mab mutants, ray patterning is changed [21,31]. However, the underlying process governed by Sma/Mab signaling on how structural cells are localized to distinct sites in the epidermis has not been characterized. Ray cell groups tend to reside at the junctions of two Rn.p cells. These hypodermal cells are generated at the first ray precursor cell division before the cell cluster patterning is obvious ( Figure 2C). Ray 5 and 7 cell groups are born above their respective Rn.p cells while the others stay beneath the hypodermal daughter at birth. We have analyzed the arrangement and geometry of developing ray cell groups of the late L3 and L4 stages to examine how these sites are altered in mutants of Sma/Mab pathway. Our results demonstrate that positions of R6 derived cells are essentially normal in these mutants. Sma/Mab signaling is not involved in pre-patterning of ray 6 precursor cells. Similarly, formation of parental cells R7.a and R7.p in either mab-31 or sma-6 mutants did not differ from that in wild-type animals. However, in mutants, R7.a derivatives were abnormally positioned beneath the R7.p cells ( Figure 2G and 2K), which may probably have been caused by the initial skewing of cell divisions leading to the positioning of R7.aa and R7.ap ventral to R7.p ( Figure 2F and 2J). This feature indicates that in wild-type animals, Sma/Mab signaling event is required in the R7.a for proper formation of its daughter cells, R7.aa and R7.ap, possibly by skewing their division plane towards the dorsal side of R7.a. Such organization is maintained in wild-type individuals to separate the daughters of R6.a and R7.a over a distance by R5.p and R6.p. We therefore hypothesize that a novel mechanism of Sma/Mab signaling acts in sensory rays patterning. Receptors SMA-6 and DAF-4 are located at R7.a and are activated upon binding of the DBL-1 ligand, which is secreted from the ventral nerve cord. The signal is then transmitted through intracellular SMAD molecules and nuclear factor MAB-31, to its targets which are potentially required for establishment and maintenance of cell polarity of R7.a and thus identity of its descendants. Disruption of this signaling event may change the cellular fate of R7.a and its subsequent plain of division leading to an abnormal position of the daughter cells with respect to the neighboring cells. As a result, fusion of ray structural cells, a descendent of R7.a, is possible because of the loss of identity and physical proximity between the fusing partners. It ends up with a ray 6-7 fusion defect. Interestingly, in the absence of mab-18, this Sma/Mab signal is dispensable and the identity of ray 7 could be restored. Would the maintenance of ray 7 identity involve deregulation of mab-18 activity through the Sma/Mab signaling? That's certainly a possibility since MAB-18 subcellular localization in the ray 7 cluster is developmentally regulated [29]. Nevertheless, the precise molecular regulation involving MAB-31 and SMAD molecules demands a more detailed characterization. It will be an important prerequisite before the operation of a complete genetic network during sensory ray patterning can be addressed.
Conclusion mab-31 is identified to encode a nuclear protein that acts in the Sma/Mab pathway during C. elegans male sensory ray cell development. We have shown by genetics that mab-31 acts in Sma/Mab signal receiving cells in the ray lineage. Furthermore, we have demonstrated by epistatic analysis that the Sma/Mab pathway acts upstream of mab-21 and mab-18 within the ray lineage including the structural cells, and the entire pathway is negatively regulated by unc-130 upstream of the dbl-1 signals. These data collectively suggest potential crosstalk between ray cells with patterning signal acting via the diffusible DBL-1 and the involvement of important structural cell-specific molecules to mediate cell fate specification and subsequent differentiation.

Strains
Strains were maintained as described by Brenner [10]. To facilitate male tail analysis, each strain, including the mab-31(tm2718) mutant, was crossed with him-5(e1490) V to generate a double mutant (see below). Strain names and genotypes of animals used were:

RNAi
RNAi clone I-7J10 was from the Ahringer RNAi library (MRC Geneservice) [32]. The identity of this clone was determined by in house sequencing. RNAi bacteria cultivation and double-stranded RNA induction were performed as described by Kamath et al. [33]. KC447 worms were synchronized by hypochlorite treatment, and the resulting L1 animals were placed on bacteria spotted plates and incubated at 20°C. Male tail phenotypes were assessed in adult worms after 3 days. In all RNAi assays, E. coli HT115(DE3) carrying the empty RNAi vector L4440 was fed to the same strain as a negative control.

Male tail phenotype characterization and body-length measurement
The male tails were examined by Nomarski microscopy (N > 100

Lineage analysis
Worms were mounted on 2% agarose pads in 2 μl of M9 mixed with a smear of OP50 bacteria. A 12 × 12 mm cover slip was gently lowered onto the worm and divisions of the ray precursor cells marked by the expression of ajm-1::GFP reporter (Strain SU93) were observed and recorded over a period of 4-5 hours.

Transgene construction and rescue experiment
A mab-31 translational GFP fusion construct was made using pPD vectors kindly supplied by the Fire Lab (Stanford University) and verified by sequencing. GFP from pPD95.75 was excised using KpnI and inserted into the KpnI site at the 3' end of mab-31 cDNA amplified from wild-type animals, with stop codon mutated. This GFPtagged 3' mab-31 cDNA was inserted into a vector carrying 2 kb of mab-31 upstream sequence, which is amplified with primers KC 1058 (5' GCTCGAA-TAACCTGG 3') and KC1059 (5' AGGTCCGCC-CACTTGTT CC 3'), to generate a full-length mab-31 cDNA GFP-tagged rescuing construct, MAB-31A::GFP. The transcriptional reporter mab-31::gfp was generated by utilizing the same 5' upstream sequence and cloned into pPD95.75 vector. Germ line transformation with different constructs was carried out as described [34]. All transgene plasmids were co-transformed with either the dominant rol-6 gene or crm-1b::gfp as the selection marker. In general, at least three transgenic lines were generated for each plasmid construct and rescue or expression data were collected and pooled for comparison.

Additional material
Additional file 1: Schematic representation of mab-31 gene structure and protein sequences. mab-31 gene products are embedded in~10.5 kb intron 9 of cogc-1. Mutation tm2718 deleted coding sequence of mab-31a, but it does not affect the transcript level of mab-31b, a transcript irrelevant to the ray fusion phenotype. B) Sequence alignment of nematodes MAB-31 proteins (C. elegans: WP_CE24445, C. briggsae: BP_CBP03618, C. remanei: RP_RP22498, and C. brenneri: CN_CN37332) was shown by ClustalX. The conserved nuclear localization signal (NLS) (PredictNLS online) and deleted region of MAB-31 protein in tm2718 mutant were predicted and labeled.