Surface epithelia play multiple roles in development. They provide protective barriers within which tissues can survive and differentiate, and they also have important morphogenetic roles in the generation and maintenance of body form. This function is particularly important in organisms with exoskeletons, such as arthropods and nematodes. In both of these groups, the external cuticle contains a thick extracellular layer (chitinous in arthropods and collagenous in nematodes) that supplies mechanical strength and defense against injury or invasion by pathogens. The outermost face of the cuticle defines the main interface between the organism and its environment, and must consequently have special chemical and physical properties. Different areas of the surface may also need to have distinct features, which must be generated in the correct spatial arrangements by the underlying cells.
The rectal epithelium of the nematode C. elegans provides a case in point. The lining of the rectal tube in C. elegans is provided by part of the collagenous cuticle. It is continuous with the general external cuticle, which forms the exoskeleton of the worm, and is molted and reformed afresh at each larval molt. However, it is likely that the structure and surface of the rectal epithelia need to have properties distinct from the rest of the nematode epithelia, because the rectum is exposed to different chemicals (in gut effluent) and is subject to different mechanical constraints, as compared to the rest of the surface of the worm.
The rectum is a specialized section of the gut of C. elegans, which is formed by a distinct set of cells, different both from the large intestinal cells and from the general hypodermal cells and syncytia that form most of the surface of the worm. The rectal epithelium in C. elegans L1 larvae is composed of three rings of two cells each: K and K, F and U, B and Y. This structure is largely complete and functional at the end of embryogenesis, and only two changes occur in this epithelium during subsequent larval development: Y differentiates to become the neuron PDA and its position in the rectum is taken by the hypodermal cell P12.pa (hyp12). Also, K divides once giving two daughters, one of which (K.a) takes on the epithelial function of K, while the other (K.p) becomes the neuron DVB. Although F, U, and B do not divide in hermaphrodites, all three are blast cells in males. Together with the Y cell, which is also a male blast cell, these cells divide post-embryonically to give rise to 66 surviving progeny that generate the spicules, post-cloacal sensillae, and proctodeum (Sulston and Horvitz,1977).
Considerable progress has been made in understanding the specification and differentiation of these cells during the development of both hermaphrodites and males. The morphology and function of the hindgut cells can be disrupted by loss of function mutations. which include egl-38, mab-9, and mab-23 (Chisholm and Hodgkin,1989; Chamberlin et al.,1997; Woollard and Hodgkin,2000). In addition, these mutations can also interfere with organ integrity in the egg-laying system in hermaphrodites and with the development of male-specific tail structures. However, few specific differentiation markers have been developed for the rectal epithelial cells, as yet.
In this study, we make use of targeted pathogenesis to reveal a molecular specialization of rectal cells, which involves the local expression of bus-1, a gene encoding a predicted acyltransferase. This gene activity appears likely to modify the surface properties of the rectal cuticle in such a way as to permit local adhesion and infection by a coryneform parasite, Microbacterium nematophilum. In C. elegans, bus-1 is expressed significantly only in rectal epithelial cells. The tightly localized expression of bus-1 has been investigated by examining the effect of transcription factors that affect differentiation of rectal epithelial cells.
The infection by M. nematophilum has also revealed the ability of the rectal epithelium to undergo morphological change, because the rectal region undergoes striking enlargement and deformation in infected wildtype worms. This rectal swelling does not involve cell division, but it is not clear from superficial examination which cells are swelling up, and the cellular mechanism of the swelling has not been established. Some kind of vacuolar enlargement within the cells seems likely to play a part, because extensive vacuolation can be seen in tail cells of strongly infected, severely swollen individuals. A bus-1::GFP reporter construct provides a convenient means of visualizing the infection-induced swelling of particular rectal epithelial cells.
The complete loss of bus-1 function by mutation has no obvious effect on the worm other than to cause resistance to M. nematophilum. Despite its lack of obvious function in C. elegans, the bus-1 gene exhibits strong evolutionary conservation. The C. briggsae ortholog of bus-1, Cbr-bus-1, was identified and shown to fully rescue a C. elegans bus-1 mutant, when provided as a transgene. Rectal expression of bus-1 is conserved in C. briggsae, but in addition Cbr-bus-1 is also expressed in vulval cells, and this expression is correlated with adhesion of M. nematophilum to both vulval and rectal cells in C. briggsae, in contrast to its purely rectal tropism in C. elegans.
Another epithelial acyltransferase gene, bus-18/acl-10, which also affects bacterial infection, was identified during this work. Like bus-1, bus-18 is strongly expressed in rectal cells but in contrast to bus-1, bus-18 is much more widely expressed in hypodermal and other tissues, and the loss of bus-18 is severely deleterious, resulting in general cuticle fragility and poor viability. Both genes are involved in determining surface properties of C. elegans and presumably many other nematode species, but one gene has a very local role, whereas the other appears to be widely expressed.
Bacterial Infection Depends on Rectum-Specific Surface Properties
Infection by the bacterial pathogen M. nematophilum provides a novel means of revealing the specialized properties of the rectal epithelium of C. elegans, because these bacteria are able to adhere tightly to the rectal and peri-anal surface, not being dislodged by extensive washing (Fig. 1A). In contrast, the bacteria show no ability to adhere to the rest of the surface of the worm. This ability to adhere to the rectal cuticle is probably crucial to pathogenicity by M. nematophilum because it allows establishment of a colony in the rectum, which is likely to provide a benign, nutrient-rich environment for a bacterium. Infection by most other pathogens able to infect C. elegans involves colonization of the intestine, which has a microvillar surface entirely different from that of the rectum.
One way to acquire resistance against infection by M. nematophilum is, therefore, to change the development of the rectum such that bacteria are no longer able to colonize and initiate infection (Fig. 1B). In a large screen for mutants that are resistant to infection by M. nematophilum, the majority of those recovered (defining more than 15 genes) were no longer infectable. Some of these, such as mutants of the gene srf-3, bus-8, and bus-17, are detectably altered in general surface properties over the whole worm, as revealed by lectin staining and other properties. Others do not show obvious general abnormalities, and may be more specifically altered in the differentiation of the rectum (Gravato-Nobre et al.,2005).
A strong candidate for such a rectal differentiation gene is bus-1, which was a frequently targeted gene in the chemical and transposon mutageneses. All alleles recovered in these screens (7 EMS, 18 transposon, 1 spontaneous mutant) had similar properties: they appeared normal in behaviour, fertility, and response to other surface-adherent pathogens such as Yersinia spp. (Creg Darby personal communication), but they failed to be infected by M. nematophilum, and never exhibited the characteristic swelling response. No alterations in lectin staining or other general cuticle properties were seen.
bus-1 Encodes a Predicted Integral Membrane O-Acyltransferase
In order to examine its properties and expression pattern, we set out to clone bus-1. This was achieved by a combination of positional cloning, transgene rescue, and transposon insert identification (see Experimental Procedures section; Fig. 2). Attempts to speed up this process by means of Tc1 transposon insertion display (Wicks et al.,2000) were not successful, due to a high background of irrelevant Tc1 transposon inserts even in extensively outcrossed lines. Similarly, RNAi screens using candidate genes in the genomic region of bus-1 were not initially informative, probably as a result of the known weakness of RNAi phenotypes for many postembryonically expressed genes.
The reference bus-1 mutant was convincingly rescued by transformation with cosmid R03H4 and then with a 4.1-kb PCR fragment that includes the promoter and coding region of the gene R03H4.6 (Fig. 2D). Two closely related genes on the same cosmid, R03H4.1 and R03H4.5, were also tested but failed to rescue. RNAi experiments using R03H4.6 yielded a weak and variable Bus phenotype when wildtype worms were injected with dsRNA corresponding to this gene, and a stronger but still incompletely penetrant Bus phenotype when hypersensitive rrf-3 worms were injected. Subsequent sequencing identified alterations in the sequence of R03H4.6 in all 11 alleles, as discussed below, confirming that the correct gene had been identified (Fig. 3 and Supp. Fig. S1, which is available online).
R03H4.6 encodes a predicted membrane acyltransferase and belongs to a large gene family in C. elegans (PFAM: PF01757; Supp. Fig. S2), for which very little functional information is known. Given that some members of this family have an RNAi phenotype (as annotated in Wormbase), we tested a number of family members for which mutants were available, and also tested a further set by RNAi. Of the nine tested by RNAi (see Experimental Procedures section), none showed any detectable alteration in response to infection.
Mutations are available for a few of the more distantly related acyltransferase genes, and some of these were also tested for response to M. nematophilum. The strain MP108, originally isolated for resistance to nordihydroguaretic acid and to fluoxetine and carrying a mutation in the acyltransferase gene ndg-4, was found to exhibit a strong Bus phenotype, with no rectal colonization and no swelling. However, further investigation revealed that this strain carries both the ndg-4 mutation and an unlinked bus mutation, which proved to be another allele of bus-1. A less pleiotropic gene in the same family, nrf-6, is similarly resistant to fluoxetine but shows no resistance to M. nematophilum and no vulval adherence.
bus-18 Encodes a Lysocardiolipin Acyltransferase Homolog
Surveys of other, more distantly related acyltransferase family genes suggested a possible correspondence between one of these, acl-10, and another M. nematophilum-resistance gene, bus-18, because the genetic map position of the bus-18 mutation was found to be very close to the location of acl-10. Sequencing of acl-10 (F55A11.5) in the single characterized bus-18 mutant, e2795, revealed a nonsense mutation, W109amb, in the coding sequence, which would be expected to result in a null allele. Moreover, a deletion allele, tm1045, has been generated for this gene, which results in the same phenotypes as bus-18(e2795): Bus, skiddy movement, fragile cuticle, drug hypersensitivity, abnormal cuticle staining. The two mutations tm1045 and e2795 fail to complement and both tm1045 and e2795 are fully rescued for all phenotypes by a transgene expressing F55A11.5. These results establish that bus-18 and acl-10 are the same gene.
In contrast to bus-1, bus-18/acl-10 shows significant phylogenetic conservation, with substantial homology to lyscocardiolipin acyltransferases genes. The bus-18/acl-10 gene is also closely related to two other C. elegans genes, acl-8 and acl-9, but no phenotype has been reported to result from RNAi knockdowns of either gene, and they are not located close to any known bus loci. Moreover, homozygotes for tm2164, which is a 5′ deletion allele of acl-9, exhibit no obvious phenotype and are non-Bus.
bus-1 Gene Structure and Predicted Function
The predicted bus-1 structure is a protein of 646 amino acids and eight predicted transmembrane regions, with a long C-terminal cytoplasmic domain. As such, it is similar to members of the acyltransferase 3 gene family (integral membrane O-acyltransferases) in other organisms, few of which have any well-defined biochemical properties. BUS-1 is unusual in that its predicted catalytic domains appear partly hydrophobic, in contrast to most members of the family.
The lesions in different mutant alleles of bus-1 provide little information about the biochemical function of its product (Fig. 3 and Supp. Fig. S1). Most of the mutator-induced alleles are, as expected, insertions of either Tc1 or Tc3, which are the most active of C. elegans transposons, and these insertions are likely to result in complete loss of gene activity. The reference allele e2678 is an ochre nonsense mutation in the second exon, again expected to be a null allele. Two missense alleles (e2683 P23L and e2712 R196H) affect conserved invariant residues in the predicted transmembrane helices of BUS-1, and therefore probably encode non-functional proteins. Therefore, all bus-1 alleles appear to be null or strong loss-of-function mutations, consistent with their identical phenotypes.
bus-1 Is Specifically Expressed in Rectal Epithelial Cells
In order to examine where bus-1 acts in the worm, we examined its expression pattern using a series of GFP and RFP transgene constructs, starting with a fusion of 995 bp 5′ plus the first 483 bp of coding region to DNA encoding DsRed2 (Fig. 4A–D). Transgenic animals carrying this reporter showed strong fluorescence in the presumptive rectal epithelial cells K, F, and U (Fig. 4A). This localization was confirmed using a hindgut-specific marker, lin-48p::GFP, which expresses in the same cells (Fig. 4B–D).
The reporter construct exhibited a conspicuous punctate peri-nuclear pattern of fluorescence in these cells, consistent with localization to the endoplasmic reticulum or Golgi apparatus (Fig. 4B,C). It is possible that the reporter contains enough bus-1 sequence to direct it to its normal site of action, and indeed an ER membrane retention signal RDQI has been detected in the sequence of the first exon (between residues 2 and 5), using software PsortII (http://psort.ims.utokyo.ac.jp). A similar reporter construct with a mutated first codon of bus-1 (ATG changed to AGG) was found to be expressed in the same cells, but showed diffuse fluorescence filling the entire cytoplasm of the rectal cells.
The first construct exhibited no bus-1 rescuing activity (Table 1). We therefore constructed a polycistronic construct (Cheung et al.,2004) consisting of an operon containing the bus-1 promoter region upstream of bus-1, the bus-1 gene, an intercistronic sequence, and the GFP coding region (Fig. 4E–I). This operon, when introduced as a transgene, fully restored bacterial adhesion and tail swelling to infected animals (Fig. 4E and Table 1), and exhibited the same expression in cells K, F, and U, with a diffuse cytoplasmic distribution of the GFP reporter (Fig. 4E, G). Since this is a rescuing construct, the pattern of fluorescent cells seen is likely to represent the authentic expression pattern for bus-1. Table 1 summarizes the rescuing data obtained with different constructs.
Table 1. Summary of the Rescuing Analysis of bus-1 (e2678) Using Different Constructsa
No. of lines generated
% of rescue
All tests were performed in bus-1 (e2678) carrying indicated transgene. The table depicts the penetrance of the rescue of the Bus phenotype by picking randomly animals of each genotype that carried the transgenic marker rol-6, sur-5::GFP or unc-119 (+), from a selected line. N = 50 animals for each strain.
eEx530 [R03H4.6; pTG96]
eEx557 [bus-1p::DsRed2; pRF4]
eEx586 [bus-1p::bus-1::GFP; pRF4]
eEx587 [CBG19162; pTG96]
eEx588 [bus-1p::bus-1 cDNA; pTG96]
eEx598 [hsp16.41p::bus-1; DP#MM016B]
A construct carrying a bus-1 cDNA (under the control of the same 995-bp promoter region) rescued bus-1 function at lower efficiency, as assayed by the presence of tail swelling after infection by M. nematophilum (Table 1). To confirm the current gene structure model for bus-1, we next performed RT-PCR using primers complementary to the predicted 5′ and 3′ UTRs (as annotated by Wormbase release WS193). A 5′ gene-specific primer in combination with an oligo(dT) primer generated a polyadenylated mRNA product of 2.0 kb, consistent with expectation. Sequencing confirmed that all predicted introns were absent from this 2.0-kb transcript, thus corroborating the Wormbase model for bus-1. The reduced rescuing activity of the cDNA construct may be due to the absence of regulatory elements in the 5′ and 3′ UTRs and/or introns.
In addition to the conspicuous expression in the rectal epithelial cells (Fig. 4G,H), a few presumptive tail neurons of uncertain identity were seen to express the construct, and also one pair of head neurons (Fig. 4I), which from their location are likely to be one of the bilateral pairs of amphidial chemosensory neurons.
In view of the highly specific location of bus-1 expression, we examined whether related acyltransferase genes might also exhibit comparably restricted expression patterns. The two genes adjacent to bus-1, oac-40/R03H4.1 and oac-41/R03H4.5, were tested by making GFP fusions driven by their presumed promoter regions. One gave low level expression throughout the hypodermis at all developmental stages, while the other R03H4.5, exhibited strong fluorescence in the seam cells of dauer larvae but not at other stages (Supp. Fig. S5).
We also examined expression of bus-18/acl-10, using a polycistronic rescuing construct (Fig. 5). Strong expression of this construct was observed in the rectal epithelial cells (Fig. 5E, F), but in addition widespread expression was observed in other epithelial cells. General expression in the hypodermis and vulva was observed (Fig. 5D), as well as in the labial support and arcade cells in the head, the excretory gland cell, the spermatheca, and the phasmid socket and sheath cells (Fig. 5A–C). The widespread expression in the hypodermis is consistent with the general defects in cuticle integrity and surface properties that are observed in the bus-18 mutant.
bus-1::GFP Reveals Swelling of Rectal Epithelial Cells
The conspicuous tail-swelling response exhibited by worms after infection by M. nematophilum has been hard to interpret in terms of exactly which cells are changing shape and in what way, as a result of the complicated cellular anatomy of the rectal region and the gross structural distortion of the region resulting from infection. The operon construct described above was used to visualize rectal cells K, F, and U, after exposing worms to the pathogen for 72 hr Fig. 6A, B). As illustrated in Figure 6, these cells can be seen to become greatly enlarged after infection, and their enlargement seems sufficient to explain much of the observed distortion of the whole region (Fig. 6C, D). This enlargement can also be seen with a lin-48 reporter, but much less clearly. Using the bus-18 reporter, the other major rectal epithelial cell B was also seen to swell up during the course of the response to infection, particularly at late stages. It is not yet clear what the mechanism of cellular swelling might be.
The level of fluorescence in the swollen cells does not appear to have increased, suggesting that there is no up-regulation of bus-1 expression during infection. RT-PCR experiments (data not shown) confirmed that bus-1 expression levels are not affected by infection. Also, the non-rescuing construct was crossed onto a bus-1 mutant background, and found to give the same expression pattern as in bus-1(+) worms, indicating that bus-1 does not affect its own expression (Table 1).
Correct Expression of bus-1 Is Dependent on Known Hindgut Regulatory Genes
The developmental regulators responsible for normal development of the C. elegans hindgut have been extensively studied, and many mutants affecting this process have been isolated. The specific rectal expression of bus-1 was examined by crossing the bus-1 reporter constructs onto these mutant backgrounds. The majority did not affect bus-1 expression. This set included some genes that might have been expected to have an effect, such as the Hox gene egl-5, which is required for tail swelling (Nicholas and Hodgkin, unpublished data) and has been shown to affect rectal expression of the MAP kinase gene mpk-1, which acts in the signal transduction cascade that drives tail-swelling. Another gene that might be expected to have an effect is lin-48, which encodes a zinc-finger transcription factor that is essential for proper specification of K, F, and U. However, lin-48 mutants continued to express the bus-1 reporter in these cells, and in fact exhibited much brighter fluorescence in these cells, as compared to a wildtype background. lin-48 mutants are also susceptible to infection by M. nematophilum and exhibit a normal swelling response. Listed in Supplementary Table S1 are the responses to M. nematophilum of a number of C. elegans mutants, which show disruptions in genes that express in the hindgut, together with bus-1p::DsRed2 expression in some of these mutant backgrounds.
Mutants of three genes did result in an altered bus-1 expression pattern (Table 2). First, a weak mutation of the PAX domain gene egl-38, n578 (G69E) hypomorphic allele, resulted in worms that failed to express the bus-1 reporter (Fig. 7B) and also exhibited a Bus phenotype after exposure to M. nematophilum, with no rectal colonization by the pathogen and no swelling response (Fig. 7A). In contrast, a stronger (but still hypomorphic non-null) allele, sy294 (G33V), resulted in a different effect, with continued and enhanced expression of a bus-1P::DsRed2 reporter in rectal cells (Fig. 7C,D). Worms mutant for sy294 have distorted and swollen rectal regions even in the absence of infection, so it is hard to determine whether they exhibit a Dar response to infection (Fig. 7C). The enhanced expression of the reporter in egl-38(sy294) animals could be the result of loss of lin-48 expression, because the sy294 allele is defective in transactivation of lin-48, whereas the n578 allele is not (Johnson et al.,2001). Null alleles of egl-38 are lethal and, therefore, could not be tested.
Table 2. Expression of bus-1p::DsRed2 in Hindgut Mutantsa
Two genes were found to have a negative effect on the bus-1 expression pattern (Table 2). The gene mab-23 encodes a DM (DOUBLESEX/MAB-3) class transcription factor, and is required for normal male tail development but has no obvious role in hermaphrodites, although it is expressed in some hermaphrodite tail cells (Lints and Emmons,2002). In mab-23 mutant hermaphrodites, the bus-1P::dsRed2 reporter exhibits an expanded pattern of expression, with fluorescence visible in the hindgut cell B, which suggests that one function of MAB-23 may be to directly repress bus-1 expression in cell B (Fig. 8B, C). Inspection of the bus-1 promoter region revealed one possible binding site for DM factors, and a test of this site's importance for bus-1 expression was carried out by site-directed mutagenesis. However, the result of this test was negative: a bus-1 reporter transgene carrying a mutated version of this site was still expressed only in cells K, F, and U, and not in cell B (data not shown). The expanded expression of bus-1 in mab-23 mutant animals does not have any obvious effect on the swelling response, which appears indistinguishable from wildtype.
A second gene that acts to restrict bus-1 expression is the T-box gene mab-9, which normally functions to control the fates of the dorsal rectal cells F and B. In mab-9 mutants, F and B take on the fates of their ventral neighbours U and Y. This causes minor defects in hermaphrodite rectal development, but results in major abnormalities in male tail development, as a result of the absence of B, which is a major blast cell for male tail formation (Chisholm and Hodgkin,1989; Chamberlin et al.,1999; Woollard and Hodgkin,2000). When the bus-1P::DsRed2 transgene was introduced to a mab-9 mutant background, ectopic expression of the reporter was observed, notably in ventral tail neurons (Fig. 8E, F mab-9 mutant hermaphrodites are hypersensitive to rectal infection by M. nematophilum, so that animals grow much slower than WT (Fig. 8D), and adult hermaphrodites can exhibit a swelling response after exposure to the pathogen, a situation that is not usually observed in wildtype adult hermaphrodites (Supp. Fig. S4E). mab-9 mutant males are even more hypersensitive to infection, experiencing massive invasion by the pathogen and usually dying as larvae (Supp. Fig. S4A–C).
In view of this hypersensitivity, we constructed a mab-9; bus-1 double mutant strain, and tested it for sensitivity to infection by M. nematophilum. In contrast to mab-9(+); bus-1 worms, the double mutant is susceptible to infection.
To examine this unexpected result, we quantified the effect of M. nematophilum on single and double mutants. When exposed to M. nematophilum at 25°C, most wildtype animals can complete development within approximately 96 hr (i.e., one day later than when they would if they grew on OP50 lawns). In contrast, exposed bus-1 mutants can reach adulthood in less than 72 hr. Growth rate provides a less subjective assay than the swelling response, which is hard to quantify. Synchronized wildtype and mutant L1 animals were cultured on OP50 or mixed CBX102/OP50 lawns at 25°C for 70 hr, and the number of larvae that reach adulthood within this period was counted (Fig. 8G). On OP50, all the strains grew to adulthood at approximately the same rate, on mixed lawns. On mixed lawns, 92% of bus-1 were able to reach adulthood, in contrast to mab-9 mutants where no adults could be found on plates. mab-9;bus-1 double-null animals showed significantly more adults (51%) than did either the mab-9 single mutant (0%) or the wild type (34%) (P < 0.05), indicating partial protection by the bus-1 mutation. However, the double mutant also grows much less well than the bus-1 mutant alone, indicating that in a mab-9 background, the absence of bus-1 does not protect the animals and infection occurs, albeit to a lesser extent.
This suggests that bus-1 activity is normally necessary for the initial stages of adhesion and colonization by M. nematophilum, but that the bacterium can still invade and proliferate if the rectal epithelium is compromised and abnormal, which is likely to be the situation in mab-9 mutants. Work by Appleford et al. (2008) suggests that mab-9 is required in the posterior hypodermis for the secretion and/or modification of the cuticle. BUS-1 is not expressed in this tissue.
Overexpression of bus-1
In view of the tight localization of bus-1 gene expression, and its coincidence with the limited zone of cuticle to which M. nematophilum can adhere, we tested to see if ectopic expression of bus-1 might lead to corresponding ectopic adhesion by the bacteria or to other abnormalities. We also wished to test when bus-1 activity was required during development. To this end, we used the promoters from the heat-shock genes hsp-16.2 and hsp-16.41 to drive bus-1 expression, because these promoters are believed sufficient to drive transcription in many different tissues after heat treatment. Transgenes with these promoters fused to bus-1 were constructed and introduced into a bus-1(e2678) mutant background. Heat-shock applied to either of the lines (carrying either hsp-16.2p::bus-1 or hsp-16.41p::bus-1) during L1 or early L2 stages, together with exposure to the pathogen, resulted in rescue of the Bus phenotype, with normal bacterial infection of the rectum and tail-swelling (Fig. 9B and Table 1). However, no abnormalities were seen in the heat-shocked animals in the absence of the pathogen, and no ectopic adhesion of bacteria or other hypersensitivity was observed after exposure to M. nematophilum (Fig. 9A). The rescue of the mutant phenotype shows that bus-1 function can be restored using either of the heat-shock promoters, but overexpression and ectopic expression of this gene, which are expected to result from activating these strong promoters, has no obvious phenotypic consequences.
Heat-shock applied at stages later than early L2 failed to rescue the Bus phenotype, which is surprising because wild-type worms can be infected and exhibit a Dar phenotype at both L3 and L4 larval stages. Possibly, the failure to rescue is because expression levels are not restored to sufficient levels by heat-shocking at these later stages.
Conserved and Divergent Expression and Function of bus-1
Loss of bus-1 has no obvious effect on the development, growth, or viability of mutant worms, except to make them uninfectable by M. nematophilum and, therefore, in one sense, fitter than wildtype worms. We, therefore, wished to see whether this apparently dispensable gene has been conserved in evolution, with respect to its expression pattern and function. The complete genome sequence of the related species C. briggsae has been determined, and was searched for a possible bus-1 ortholog. Initial annotation was ambiguous as to which predicted gene of C. briggsae might correspond to Cel-bus-1, but a BLAST search using the 5′ regulatory region of bus-1 identified CBG19162 as a candidate (Supp. Fig. S3). Correspondingly, the predicted product of this gene, protein CBP19548, appears to be the closest homolog of Cel-bus-1, with an overall amino acid identity of 74%. An inter-species rescue experiment was carried out, introducing CBG19162 as a transgene into bus-1(e2678) mutant worms. This transgene was able fully to rescue the Bus-1 mutant phenotype, confirming that CBG19162 is Cbr-bus-1 (Fig. 10E and Table 1).
We also examined the expression pattern for this ortholog by constructing a DsRed2 reporter driven by 789 bp of 5′ sequence from Cbr-bus-1. When injected into C. elegans, the Cbr-bus-1 exhibited fluorescence in rectal cells K, F, and U (Fig. 10A), and also in a set of vulval epithelial cells, probably the vulA, vulB, and vulC rings (Fig. 10B, C). The corresponding construct from Cel-bus-1 did not express in the vulval cells in C. elegans, as described above; this construct was also injected into C. briggsae AF16 and found to exhibit the same expression as in C. elegans, with strong expression in rectal epithelial cells and no vulval expression (Fig. 10D). Therefore, it appears that the expression pattern of bus-1 has been largely conserved between the two species, with the distinction that the C. briggsae gene contains additional elements conferring vulval expression.
The additional expression of bus-1 in vulval cells of C. briggsae became more significant when the susceptibility of C. briggsae to M. nematophilum was examined in detail. C. briggsae can be infected, like C. elegans, with rectal colonization and consequent tail swelling (Fig. 11A), but the bacterium is also able to adhere to vulval cells of some strains of C. briggsae, in contrast to its inability to adhere to the vulva of C. elegans (Fig. 11B, C). It is, therefore, possible that this difference is the result, or partly the result, of the difference in bus-1 expression pattern. We examined the infectability of C. elegans expressing the rescuing Cbr-bus-1, but found that M. nematophilum still did not adhere to the vulval cells of these worms. Therefore, expressing bus-1 is probably not enough to make the vulval cells adherent to M. nematophilum. We have also observed that the vulval colonization by M. nematophilum is not seen in all races of C. briggsae. For example, the natural isolate strain PB800 exhibits conspicuous rectal colonization and tail swelling, substantially stronger than that seen in C. briggsae AF16, but there is no obvious vulval attachment by the bacteria in strain PB800. As noted elsewhere, C. briggsae exhibits more variability between natural isolates in its response to M. nematophilum than does C. elegans (Akimkina et al.,2006).
In most respects, however, the bus-1 gene seems to be highly conserved between the two species, implying that it has an important role in the life of rhabditid nematodes. Its real physiological function remains enigmatic but its expression provides a novel and informative way of examining differentiation and shape change in the rectal epithelium, and potentially in vulval epithelia also.
We have identified two genes, bus-1 and bus-18/acl-10, which are strongly expressed in the rectal epithelial cells of C. elegans, and render the worm susceptible to infection by the nematode pathogen M. nematophilum. Use of a pathogen with a specific tissue-tropism (in this case, to the rectum) provides an effective way of revealing regional specializations within an epithelium, which might be otherwise undetectable. The two genes have contrasting properties: bus-1 appears to be expressed only in the rectum, and its mutants exhibit no abnormalities other than resistance to M. nematophilum, whereas bus-18/acl-10 is expressed not only in the rectum but also throughout the epithelium and elsewhere. Its mutants exhibit multiple defects including cuticle fragility, drug hypersensitivity, and abnormal locomotion, most of which are probably due to a severe general alteration of surface properties.
The specific rectal expression of bus-1 provides a new tool for examining differentiation in hindgut cells, as discussed further below. It also allows better visualization of the unusual morphological change induced by M. nematophilum infection, which involves gross swelling of cells in the tail region of the animal, without cell division.
Function of bus-1
The absence of any obvious phenotype associated with loss of bus-1 function, other than resistance to M. nematophilum, raises the question of what role this gene may play in the normal development and physiology of the organism. It encodes a conserved transmembrane protein with clear homology to the MBOAT (membrane bound O-acyl transferase) family. At least 58 other members of this gene family can be identified in the C. elegans genome, none of which has been associated with any specific function thus far. Members of the MBOAT family are hydrophobic proteins with eight to ten membrane-spanning regions, which catalyze the transfer of fatty acids onto hydroxyl groups of membrane-embedded substrates. Characterized target substrates include lipids (cholesterol acyltransferases), diacylglycerol (diacylglycerol O-acyltransferases), sterols (Are1/2), alginate (AlgI), and waxes (wax synthase) (Hofmann,2000). It is, therefore, possible that BUS-1 is an enzyme responsible for the secondary modification of a surface-associated secreted molecule or molecules, which might act as the receptor for rectal adhesion by M. nematophilum.
Alternatively, BUS-1 might be acting more indirectly in modulating protein trafficking and secretion in the rectal cells, by affecting membrane properties of these cells. Evidence exists that lipid rafts are involved in infection by a number of pathogens, including bacteria, which may exploit host rafts to mediate their infectious process, (reviewed by Manes et al., 2003). It has been shown that lipid rafts can act as the preferred sites for bacterial infection, providing not only binding sites, but also mediating oligomerization, internalization, or intracellular trafficking of pore-forming bacterial toxins (Fivaz et al.,1999). Membrane rafts are also associated with specific proteins, many of which need to be acylated to provide appropriate anchorage and localization. Rafts may also contribute as sorting devices to distribute proteins to the apical side of the epithelial cells (Simons and Toomre,2000). However, the pathogenic bacteria appear to be adhering to the external surface of the rectal epithelium, probably to the glycocalyx, which is separated from the plasma membrane of the epithelial cells by a thick layer of collagenous cuticle and epicuticle, so a raft-mediated effect would have to be indirect.
In either case, further biochemical analysis will be needed to determine possible substrates for the action of bus-1, but this is unlikely to be straightforward owing to the apparent high hydrophobicity of the protein and its probable membranous environment and substrates.
Loss of bus-1 most probably results in a subtle alteration to the properties of the rectal surface, which may be deleterious under some circumstances, but it does not obviously affect the normal function of worms grown under laboratory conditions. It may act redundantly with other, related acyltransferase genes, but we have not so far observed any obvious synergistic effects caused by RNAi knockdown of related genes in a bus-1 background (unpublished results). Alternatively, the danger of infection by different pathogens may provide a natural function for bus-1. Conceivably, bus-1 null mutants may have increased susceptibility to some other pathogen, while becoming resistant to other pathogens. A precedent for such a situation is provided by srf-3 mutants, which are resistant to the bacteria M. nematophilum and Yersinia spp. (Hoflich et al.,2004), but are hypersensitive to attack by the fungus Duddingtonia flagrans ((Mendoza De Gives et al., 1999).
Although bus-1 itself is conserved between C. elegans and C. briggsae, other members of this gene family appear to be evolving relatively rapidly, because many sub-family members can be identified only in the genome of C. elegans, and not in C. briggsae and C. remanei (see Treefam, http://www.treefam.org/). Our results suggest that bus-1 has a role in subtle surface modification, which can have important consequences for biotic interactions such as bacterial infection. We suggest that other members of the family may have similar functions, which would explain both their rapid evolution and their lack of obvious function by standard RNAi gene knockdown tests. A role in surface modification is implied for at least one other member, R03H4.5, for which we observed stage- and tissue-specific expression in the seam cells of dauer larvae.
In contrast to bus-1, the other acyltransferase examined in this work, bus-18/acl-10, is more strongly conserved, with very significant similarity to mammalian lysocardiolipin acyltransferase. Three nematode paralogs of the mammalian gene (acl-8, acl-9, bus-18/acl-10) can be identified in each of the three fully sequenced Caenorhabditis genomes, but only bus-18/acl-10 seems to have a major function. The predicted homology with lysocardiolipin acyltransferase may offer an easier route to defining the biochemical function of bus-18/acl-10, as compared to bus-1, but the same uncertainty applies as to the mechanism of presumed surface modification. bus-18/acl-10 is also much more widely expressed than bus-1 and has a general effect on surface properties, such that mutants are drug-sensitive and very impaired in movement. The bus-18 null mutants are similar in these phenotypes to mutants of the predicted glycosyltransferases BUS-8 (Partridge et al.,2008) and bus-17 (Yook and Hodgkin,2007), which also affect general surface properties. This suggests that the hypodermal expression of bus-18 may represent its major site of action, and that its neuronal expression is not involved in the uncoordinated phenotype. In terms of effects on rectal properties, it is conspicuous that bus-18, like bus-1, is strongly expressed in rectal cells although it also occurs in other tissues; this observation is again consistent with some kind of local specialization of surface properties in the rectum.
Resistance to M. nematophilum in bus-1 and bus-18 mutants seems most likely to arise by loss or alteration of a recognition or adhesion receptor for the bacterium, because no rectal colonization or adhesion is observed after exposing these mutants to M. nematophilum, but more complex scenarios are still possible. Involvement in the actual swelling response is unlikely because ubiquitous over-expression of bus-1, by means of two different heat-shock promoters, failed to induce any swelling or other change in phenotype, in the absence of infectious challenge. Involvement in specification of a surface receptor is also supported by the observation that no change in phenotype is seen in double mutants between bus-1 and other genes affecting infection, such as srf-3 (data not shown).
Tissue-Specific Expression of bus-1
We have examined the regulation of bus-1 by testing expression of reporter constructs in mutants with alterations in known regulators of hindgut differentiation. The morphology and function of the hindgut cells can be disrupted by loss of function mutations, which include egl-38, mab-9, and mab-23 (Chisholm and Hodgkin,1989; Chamberlin et al.,1997; Woollard and Hodgkin,2000). In addition, these mutations can also interfere with organ integrity in the egg-laying system in hermaphrodites and with the development of male-specific tail structures.
bus-1 mutants did not show any obvious defects in hindgut integrity or in the posterior body region, in either hermaphrodites or males. However, our genetic analysis studies have indicated that activity of Pax 2/5/8 transcription factor EGL-38 controls bus-1 expression. EGL-38 is a paired domain transcription factor, important for the proper organ development including the egg-laying system, where it is necessary for the correct identity of four uterine cells, and the rectum in particular, the male hindgut, where it is required to make the U cell different from Y (Chamberlin et al.,1997). These functions appear to be partially separable since they are preferentially disrupted by different egl-38 hypomorphic mutations. For example egl-38(n578) mutants are defective in egg-laying but retain tail functions including male tail development (only 24% are Mab) as well as lin-48 activity. In contrast, in egl-38(sy294) animals, male tail functions and lin-48 activity are disrupted but they retain a high level of function in egg-laying (Chamberlin et al.,1997). A corresponding difference is seen in the present study, in that in the egl-38(n578) mutant, expression of bus-1 transgenes is ablated and animals fail to swell when exposed to M. nematophilum, whereas in the egl-38 (sy294) animals bus-1 expression was enhanced, and a slight Dar phenotype was observed. This indicates that mutations in egl-38 can have two antagonistic effects on bus-1 activity in K, F, and U. One possibility is that egl-38 affects its targets using two alternative routes, by binding to an intermediate molecule/transcription factor or by repressing a repressor.
We also provide evidence that bus-1 is negatively regulated by two other transcription factors, one a member of the DM (DOUBLESEX/MAB-3) class, mab-23, and the other a member of the T-box family of transcriptional regulators, mab-9. Ectopic expression of bus-1 was detected in the presumptive B cell of the hindgut in mab-23 mutants. Given that egl-38 was found to be a negative regulator of bus-1 expression, this result was not unexpected. There is genetic evidence that egl-38 is a transcriptional activator essential for mab-23 expression in K, F, and U. Furthermore, in mab-9 mutants, expression of mab-23 is increased in the presumptive F cell (Sewell et al.,2003).
Our reporter construct experiments demonstrated an enhanced bus-1 expression in mab-9 mutants, where it gained expression in ventral tail neurons. It remains to be seen which of these effects is mediated by direct binding of transcription factors. Despite the presence of a predicted DM binding site in the promoter region of bus-1, no change in gene expression or function was seen when this region was altered by site-directed mutagenesis. These results and others underline the complexity of gene regulation and differentiation in rectal cells. For example, the Hox gene egl-5 is also involved in specifying rectal cell identity, and is required for expression of the ERK MAP kinase gene mpk-1 in these cells, but surprisingly it does not affect expression of bus-1.
Conserved Expression of bus-1
Using interspecific gene expression assays, we have shown that Cel-bus-1 is expressed in the C. briggsae hindgut in the same pattern as in C. elegans. Conversely, bus-1 transgenes from C. briggsae were able to restore both the adhesion and swelling response in the bus-1 mutants. However, in both C. briggsae and C. elegans, the Cbr reporter gene is expressed in both the rectum and the vulva of adult hermaphrodites. This is correlated with the adhesion of M. nematophilum to the vulva of C. briggsae but not of C. elegans, although this vulval infection does not seem to have any pathogenic consequences.
The conserved rectal expression of bus-1 between C. elegans and C. briggsae is consistent with a significant function for this gene in rhabditid nematodes. The additional vulval expression of bus-1 in C. briggsae is of interest for two reasons. First, it is likely to explain, in part, the observation that M. nematophilum can adhere to the vulval cells of C. briggsae but not of C. elegans. Second, it suggests that there may be common features in the surface properties of rectal and vulval cells. This would not be surprising, because both are required to sustain cuticle at high curvature, frequent mechanical deformation (during defecation and egg-laying, respectively), and unusual chemical conditions (gut or uterine effluent). These stresses make both tissues likely targets for pathogenic attack. It will be interesting to see, as knowledge of the complete repertoire of gene expression patterns in C. elegans expands, whether many further genes prove to have shared rectal and vulval expression.
Culture Conditions and General Methods
General methods for the culture, manipulation, and genetics of C. elegans were as described by Sulston and Hodgkin (1988). All strains were cultured at 20°C unless otherwise stated. For infection assays, Escherichia coli strain OP50 and M. nematophilum strain CBX102 were grown to stationary phase in LB medium at 37°C and mixed at a 9:1 ratio (v/v), respectively. Lawns of this mixture were prepared on NGM plates and C. elegans strains were cultured on these plates at 25°C.
Expression plasmids were constructed by amplifying the respective sequences from wild-type genomic DNA. All cloning junctions were sequenced to confirm that reading frames were intact. The high fidelity Pfu Turbo DNA polymerase (Stratagene), was used in all PCR reactions. PCR primer sequences are available upon request.
pPD plasmid vectors were kindly provided by A. Fire (obtained from FireLab Vector kit). Cosmid clones were obtained from Alan Coulson (The Sanger Center, UK).
Strains and Transgenic Lines
The standard wild type strain was C. elegans Bristol N2. The Hawaiian (HA) strain CB4856 was used for snip-SNP mapping experiments. C. briggsae analysis was carried out in the AF16 genetic background.
Mutations used were: LGII: mab-9(e2410)II; him-5(e1490)V, rrf-3(pk1426)
LG IV: egl-38(n578), mec-3(n3197), egl-38(sy294), srf-3(yj10), him-8 (e1489)
LG V: dpy-11(e224), him-5(e1490), mab-23(e2518), unc-42 (e270), acl-10 (tm1045)
Deficiencies: sDf71, mDf1 V
Chromosomally integrated transgenes:
saIs14[lin-48::GFP; pDP#MM016B] (Sewell et al.,2003) (a kind gift of Helen Chamberlin); jcIs1[ajm-1::GFP; pRF4] (Koppen et al.,2001).
Isolation of bus-1 Alleles
The bus-1 mutations were generated by ethyl methane sulfonate (EMS) mutagenesis or mut-7-induced mutagenesis, as described elsewhere (Gravato-Nobre et al.,2005; unpublished data). Eleven bus-1 alleles were examined: one spontaneous allele (e2712), three EMS alleles (e2678, e2683, e2971), eight mut-7 alleles (e2720, e2738, e2743, e2746, e2750, e2751, e2757, and e2758). Unless otherwise indicated, e2678 was used as the reference mutation.
Transgenic lines were created using standard methods (Mello and Fire,1995). Cotransformation markers were pTG96 [sur-5::nls::GFP] (Yochem et al.,1998), pRF4 [rol-6 (su1006)] (Mello and Fire,1995), or pDP#MM016B[unc-119(+)] (Maduro and Pilgrim,1995).
Test DNA was used at 10–50 ng/μl, and cotransformation markers at 50–80 ng/μl concentration. The following extragenic arrays were generated by injecting the described constructs together with a transformation marker into appropriate hosts, and establishing transmitting lines: eEx525 [R03H4; pTG96]: Cosmid clone R03H4.
eEx530 [R03H4.6; pTG96]: PCR fragment containing the entire bus-1 gene plus 924 bp of 5′ UTR and 786 bp of 3′ UTR (bp 23,438 to 19,344 of R03H4).
eEx557 [bus-1p::DsRed2; pRF4]: The pbus-1::DsRed2 reporter gene construct contains a genomic fragment of 1.478 Kb (995-bp promoter plus 483-bp coding region of bus-1), which was generated by engineering BamHI and AgeI sites upstream and downstream of the predicted ATG (BamHI site at position 23,438 and AgeI site at position 21,975 in exon 3 of bus-1, in R03H4 cosmid). The BamHI- AgeI fragment was subcloned into the reporter plasmid pNK38 containing DsRed2 (a kind gift of Neline Kriek). This construct was used to inject C. elegans wild type and C. briggsae AF16 strain.
eEx572 [CBG19162p::DsRed2; pRF4]: A Cbr-bus-1 reporter construct was generated by amplifying a 1.2-Kb fragment from AF16 genomic DNA (bp 116,309 to 117,510 of c005500768.Contig2), which included BamHI and AgeI sites and consisted of a 798-bp region upstream of the predicted ATG initiation codon, plus 403 bp in the coding region of the predicted CBG19162 in exon 2. The amplicon was subcloned into the BamHI-AgeI sites of the pNK38 vector. This construct was injected into C. elegans wild type.
eEx586 [bus-1p::bus-1::GFP; pRF4]: An operon fusion of bus-1 to GFP was constructed by amplifying a 3.3-kb bus-1 fragment from genomic DNA using gene-specific primers. The DNA fragment encompassed KpnI sites and contained 995 bp upstream of the predicted ATG initiation codon and the full length bus-1 gene. The amplicon was subcloned into the KpnI site of the pDNOR plasmid. This vector was designed to place the gene of interest in an artificial operon with GFP (a kind gift of Mario de Bono).
eEx587 [CBG19162; pTG96]: PCR product amplified from C. briggsae AF16 genomic DNA using CBG19162-specific primers (bp 116,309 to 121,094 of c005500768.Contig2). The resulting 4.8-Kb PCR product includes the predicted coding region CBG19162 plus 5′ and 3′ UTR (720-bp promoter, 2.3-kb coding, 1.6 kb 3′ UTR). The amplicon was injected into bus-1 mutants (e2678).
eEx588 [bus-1p::bus-1 cDNA; pTG96]: A 995-bp BamHI fragment upstream bus-1 (position 23,438 to 22,480 of cosmid R03H4)) was cloned into MCSI of pPD49.26 vector. A 1.9-Kb KpnI-NcoI fragment was amplified from a bus-1 cDNA clone obtained from Open Biosystems (position 22,456 to 20,134). The resulting amplicon was subcloned into MCSII. The cDNA clone used as template did not contain any 3′ UTR sequences.
eEx591: [CBG19162p::DsRed2; pRF4]: Identical to eEx572; crossed onto jcIs1 background.
eEx598[hsp16.41p::bus-1; pDP#MM016B]: An hsp16.41p::bus-1 expression construct was generated by PCR amplification of bus-1 (position 22,439 to 19,962 of cosmid R03H4) and subcloning of the amplicon into the KpnI-NcoI sites of pPD49.83 vector. The PCR fragment contained 18 bp 5′UTR and 187 bp 3′ UTR sequences.
eEx601[R03H4.5p::DsRed2; pRF4]: A genomic fragment of 1 Kb encompassing 580 bp R03H4.5 promoter plus a 400-bp coding region of R03H4.5 with BamHI site and AgeI sites was subcloned into the reporter plasmid pNK38 containing DsRed2.
eEx659 [bus-18p::bus-18::GFP; pDP#MM016B]: A 3.9-Kb PCR fragment corresponding to 1 Kb upstream the predicted ATG initiation codon and the full-length bus-18/acl-10 (2.9 Kb) was fused to the KpnI site of the pDNOR plasmid. This vector was used to place bus-18 and GFP in an artificial operon and used to inject bus-18 mutants (e2795).
Living animals were mounted on 2% agar pads containing 4% (v/v) propylene phenoxytol in M9 and examined under Nomarski differential interference contrast microscopy (Zeiss Axioplan2 microscope with Axiocam). To visualize bacterial colonization in the rectum, worms were stained with the live-cell nucleic acid stain SYTO 13 (Molecular Probes) as described by Hodgkin et al. (2000).
Total RNA from C. elegans wild type was extracted from mixed stages using Trizol extraction and following the manufacturer's specifications (Invitrogen). cDNA from 5 μg RNA samples was synthesized in 20 μl reaction mixtures using gene-specific and/or oligo dT primers and SuperScript III enzyme according to the manufacturer's recommendations (Invitrogen). Two microliters of the cDNA reaction mixture was used as the template in 50-μl PCR reaction mixtures. The PCR amplicon of 2.0 Kb in size was gel purified and the corresponding DNA sequenced.
The templates used for RNA synthesis were amplified by PCR, using primers that were derived from the genomic sequences of the predicted genes. Primers contained a T3 and T7 promoter sequence at their 5′end. Templates for the following genes were prepared: oac-40/R03H4.1, oac-41/R03H4.5, bus-1/R03H4.6, oac-17/F17B5.2, oac-45/T09E11.7, oac-46/T14D7.2, oac-28/F41D3.5, oac-54/W07A12.6, and oac-20/F36G9.12. DsRNA was prepared using an in vitro transcription kit (Promega) as described (Fire et al.,1998). Equal molar ratios of the sense and antisense strands were annealed in the presence of injected buffer to produce dsRNA, and the presence and quantitation of production of dsRNA confirmed by gel electrophoresis (Fire et al.,1998; Fire,1999).
Thirty to forty hermaphrodites of L4 or young adult stage (P0) of N2 or rrf-3(pk1426), were injected and left on OP50- or OP50/CBX02-seeded plates for 12 hr at 20°C. After this period, P0 worms were transferred individually onto OP50- or CBX102-seeded plates and subsequently moved onto new plates at 12–24-hr intervals. Progeny of injected P0 animals were observed at different intervals under both dissecting and optical microscopes.
For heat shock assays, animals carrying hsp-bus-1 transgenes or hsp vector alone, were grown synchronously and collected at each developmental stage. The isolated animals were then transferred to lawns of OP50 alone or mixed lawns with CBX102, and submitted to heat shock treatment for 1 hr at 35°C and subsequent incubation at 20°C for 24 hr. For embryonic assay, gravid hermaphrodites were allowed to lay eggs for 3 hr and the embryos were heat-shocked following the same regime. Rescue of the Bus phenotype was assayed by the presence of post-anal cell swelling.
We thank Alison Woollard for helpful discussions and continuous support throughout this work, Simon Hänni for useful advice on RT-PCRs, Charles Brabin for help with statistical analysis, and Erich Schwarz for consultation on gene names. We are also grateful to Helen Chamberlin and Neline Kriek for reagents. Some strains used in this work were provided by the Caenorhabditis Genetic Center, which is funded by the NIH National Center for Resources (NCRR), and by the C. elegans Gene Knockout Consortium and the National BioResource Project, Tokyo, Japan.