- Email: [email protected]
Cuticle collagen genes
Reviews
C. elegans cuticle collagen genes
Cuticle collagen genes expression in Caenorhabditis elegans Collagen is a structural protein used in the generation of a wide variety of animal extracellular matrices. The exoskeleton of the free-living nematode, Caenorhabditis elegans, is a complex collagen matrix that is tractable to genetic research. Mutations in individual cuticle collagen genes can cause exoskeletal defects that alter the shape of the animal. The complete sequence of the C. elegans genome indicates upwards of 150 distinct collagen genes that probably contribute to this structure. During the synthesis of this matrix, individual collagen genes are expressed in distinct temporal periods, which might facilitate the formation of specific interactions between distinct collagens. he exoskeleton of the free-living nematode Caenorhabditis elegans consists predominantly of small collagen-like proteins encoded by a multi-gene family of approximately 154 members1,2. These molecules are synthesized by specialized epithelial cells and after secretion, polymerize on the apical surface of the epithelium to form a complex multi-layered structure3,4. The entire process of cuticle synthesis is repeated five times during development, once in the embryo before hatching and then towards the end of each of the four larval stages before moulting5. During each synthesis, different collagen genes are expressed in discrete temporal periods, the reason for which could relate to the mechanism of control of collagen polymerization and formation of different structural components of the cuticle6. The basic temporal pattern is repeated during each synthesis. The developmental process controlling collagen gene expression and exoskeletal formation appears to be reset after embryonic hatching and then after each of the larval moults, the cyclic repeats of the process being terminated by the larval to adult moult. Collagens are structural proteins involved in the synthesis of a variety of extracellular matrices in animals7,8. Collagen occurs as a triple helix, the result of trimerization of three monomeric collagen chains9,10. The amino acid sequence of the monomeric collagens is repetitive, with glycine at every third residue. This Gly-X-Y repeat occurs in blocks, the length and organization of which depends on the type of collagen. The residues at the X and Y positions within the repeats are frequently proline or hydroxyproline. In nematodes, as typified by the freeliving species C. elegans, collagens are involved in the formation of two distinct structures, the basement membrane and the cuticle. It is the cuticular collagens, and the multigene family that encode this class of proteins that are the subject of this review.
T
The nematode cuticle The nematode cuticle is an exoskeleton that is synthesized by the underlying epithelial tissue, called the hypodermis1,2,11. Cuticle functions include provision of a barrier between the animal and its environment, maintenance of post-embryonic body shape12–14 and movement via attach0168-9525/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(99)01857-0
ments to muscle15,16. The cuticle is synthesized five times during development and is shed at each moult, first during embryogenesis when the cuticle of the L1 larva is formed, and then again during each of the four larval stages before each moult. Cuticular material is synthesized and secreted apically from epithelial cells and polymerizes on the external epithelial surface (Fig. 1). During synthesis, submembranous actin filaments form within the hypodermis and are organized circumferentially around the cylindrical body of the worm. These filaments are coincident with furrows that form on the epithelial cell membrane during cuticle synthesis and subsequently with circumferential furrows that delineate the annulae (Fig. 1) on the surface of the polymerized cuticle17. This suggests that the shape of the polymerized cuticle is influenced by the shape of the surface on which it polymerizes. In addition to the annulae, longitudinal ridges, termed alae, exist on the lateral surfaces of the cuticle at certain developmental stages, most notably the adult (Figs 1, 2)3. Although both of these structures can be visualized with a light microscope, the cuticle is transparent giving a deceptive view of its true structural complexity. It consists of possibly six definable layers that are distinct in their ultrastructure, as imaged by electron microscopy4,18. The precise nature of the structure and layering varies at different developmental stages3; the mechanisms by which proteins polymerize to form these ordered structures are currently unknown.
Mutant collagen genes The major protein components of the C. elegans cuticle are small collagen-like polypeptides. These are encoded by a multi-gene family, consisting of approximately 154 members. From random mutagenesis of C. elegans, approximately 45 loci have been identified that affect body shape. Currently, 17 of these loci have been assigned to known gene sequences, and nine of these (bli-1, bli-2, dpy-2, dpy-7, dpy-10, dpy-13, rol-6, sqt-1 and sqt-3) are cuticular collagen genes. Among the remainder are genes encoding enzymes that are thought to be involved in collagen processing and genes involved in X-chromosome dosage compensation. Phenotypes associated with mutations in the collagen genes fall predominantly into TIG January 2000, volume 16, No. 1
Iain L. Johnstone [email protected] The Wellcome Centre for Molecular Parasitology, Anderson College, University of Glasgow, Glasgow, UK G11 6NU. 21
Reviews
C. elegans cuticle collagen genes
FIGURE 1. The cuticle and hypodermis of Caenorhabditis elegans Annulus
(a)
Dorsal hypodermis Lateral seam cells Alae
mutations that cause the roller non-dumpy phenotype are more frequently dominant. Among the recessive alleles are true genetic nulls, but most are glycine substitutions within the collagen domains1,11. Dominant alleles include mutations at the proposed ‘subtilisin-like’ cleavage site, which is believed to be required for processing of the mature collagen20,21. Some of the mutations that cause the discussed phenotypes are recessive and lead to loss of function, indicating that at least some of the collagens have specific roles in the formation of the exoskeleton1.
Cuticular collagen structure Ventral hypodermis
Annular furrows
External cortical layer Cell membrane
(b)
Actin filament bundles Hypodermal cells Cortical layers
(c)
Striated layers
trends in Genetics
(a) A section of the cylindrical body of the nematode. Annulae are present on the surface of the cuticle at all stages, alae are visible only on the L1, the dauer and the adult cuticle3. The bottom two drawings represent cross sections through the cuticle and underlying hypodermal cells. (b) The presence of furrows in the apical surface of the hypodermis and juxtaposed bundles of actin filaments that form during cuticle secretion. These are coincident with the positioning of the annular furrows in the newly formed cuticle. (c) The organization of the hypodermis after cuticle secretion, after dissociation of the actin bundles.
three basic types described as: (1) blister (bli), which is a blistering of cuticular material away from the surface of the animal; (2) dumpy (dpy), which is a shortening in the length of the animal as compared with wild type; and (3) roller (rol), which is a helical twisting of the animal’s body (see Fig. 2). The phenotypes are not necessarily exclusive. The blister phenotype (Fig. 2) is restricted to the adult stage and might identify a cuticular structure that is specific to adults. The dumpy phenotype (see dpy-10; Fig. 2) can affect all post-embryonic developmental stages, although mutations in different dpy genes can cause greater or lesser effects at different stages. The defective exoskeleton in dpy mutants significantly alters body shape. This is also true for the roller phenotype. The alae in the cuticle (see Fig. 1) run parallel to the side of the animal in the wild type (Fig. 2d and e, left panels), but are helically twisted along the length of the animal in roller mutants, as indicated by their angled aspect relative to the side of the animal (Fig. 2d and e, right panels). The helical distortion in roller mutants is not restricted to the exoskeleton itself. In the roller phenotype, the exoskeleton, the underlying ectodermal cells and musculature are all helically twisted along the length of the animal with a similar turn to that seen for the alae19. Mutations that cause the blister, dumpy and the combined dumpy and roller phenotypes are generally recessive; 22
TIG January 2000, volume 16, No. 1
A typical polypeptide structure has been predicted for the cuticular collagen genes (Fig. 3). The nine genes discussed above also fit this basic structure13,14,22–24. They are predicted to encode two main blocks of Gly-X-Y collagenlike polypeptide that are flanked and separated by three clusters of cysteine residues, which are termed Domains I, II and III in this review (Fig. 3). The smaller N-terminal Gly-X-Y block typically contains 8–10 Gly-X-Y collagen repeats, the larger C-terminal block has 40–42 Gly-X-Y repeats. Generally, the C-terminal collagen block has one or two very small disruptions within the Gly-X-Y repeats, the precise position and size of which vary between the products of different genes. In addition to these collagenlike regions, the molecules are predicted to have a noncollagen N-terminal region that varies considerably in size, but that contains conserved regions including a predicted signal peptide and a proposed subtilisin-like cleavage site, which might be involved in the generation of a mature polypeptide from a pro-collagen precursor21. Mutations that disrupt this proposed cleavage site have been shown for some genes to be neomorphic2, generating a phenotype that is distinct from both the wild-type and loss-of-function alleles. This could be explained by its insertion into the cuticle of collagen with its pro-domain still attached. In addition to the neomorphic alleles that affect the proposed subtilisin-like pro-collagen maturation site, various other lesions have been characterized for the group of nine genes discussed above13,14,20,23,24. The most common lesion is substitution of a Gly within the Gly-X-Y collagen regions. For some of the genes, true genetic nulls have been identified. Where data are available, the Gly substitutions seem to behave as loss of function. DPY-7 protein containing a Gly substitution appears to accumulate in the cytoplasm during cuticle synthesis, but very little mutant protein gets incorporated into the cuticle (L. McMahon, pers. commun.). It seems likely that collagen that contains such Gly substitutions is recognized as being aberrant by cellular machinery and is not incorporated into the exoskeletal structure. This phenomenon of removal of Gly substitution mutant collagen has also been described for vertebrate collagens8.
Cuticular collagen gene families The complete sequence of the C. elegans genome permits an accurate indication of all possible collagen-like genes. In the absence of experimental data, the number of predicted cuticular collagen genes must depend on the precise criteria used to define this class. There are approximately 154 predicted cuticular collagen genes (http://www. worms.gla.ac.uk/collagen/cecolgenes.htm); in addition, there are several gene predictions that contain some but not all of these regions (shown in Fig. 3). Similarly sized collagen gene families have been described for other nematode species25.
Reviews
C. elegans cuticle collagen genes
FIGURE 2. Phenotypes associated with mutated collagen genes
The top three images are adult animals of (a) the wild-type strain N2, (b) a bli-2 (e768) mutant and (c) a dpy-10 (e128 ) mutant. Blistering of the cuticle in bli-2 is indicated by arrows. dpy-10 is significantly shorter and fatter than wild type. The bottom panels show high magnification images of (d) a wild-type N2 strain and (e) a mutant rol-6 (su1006 ). In each case, the animal’s body lies longitudinal from top to bottom of the image and the image represents approximately half the width of the nematode. Both left-hand panels show a deep optical section where part of the side of the animal is in focus. The orientation of the animal’s side is emphasized by the black vertical line. Cuticular structures termed alae are indicated (black arrow) in a top optical section for both wild type and rol-6 mutant in the right-hand panels. As can be seen, the alae run parallel to the side of the animal in the wild type, but are at an angle in the rol-6 mutant.
Phylogenetic analysis of this gene family is complicated by the high levels of similarity within collagen-encoding regions. The family can be divided into groups based on the pattern of conserved cysteines of their predicted encoded collagens (Fig. 3). The evolutionary significance of these groupings is supported by homologies within the N- and C- non-collagen regions, which place the genes into groups that agree with the groupings that are based on the cysteine patterns alone. Of the 154 genes, 68 belong to Group 1, 12 to Group 1a, 38 to Group 2 and 31 to Group 3. The dpy-7 and dpy-2 groups have three and two members, respectively. Of the genes defined by mutation, bli-1, sqt-1 and rol-6 belong to Group 1; bli-2 belongs to Group 2; dpy-13 and sqt-3 belong to Group 3; dpy-7 belongs to the dpy-7 Group and dpy-2 and dpy-10 constitute the dpy-2 Group. Group 1a is not represented among the genes that are defined mutationally. The nine mutationally defined genes have been identified from random mutagenic screens and selected on the basis that the mutation caused an easily detectable change in body shape. For these nine genes, multiple alleles of each have been obtained. There are only approximately 20 shape-change mutants not yet assigned to a gene defined by sequence. It therefore seems likely that most of the cuticular collagen genes that are defined by sequence might not be mutated readily to generate detectable phenotypes. Because of the high similarities between many members of this gene family, many of the gene products might have overlapping functions. No single, molecularly definable characteristic unites all of the nine genes so far identified. Only the proposed collagen Group 1a is not represented among them, and this is a very highly
conserved group. Four of the nine might be considered unusual as their encoded proteins, DPY-2, DPY-10, DPY-7 and BLI-1 all have significantly longer than average C-terminal tails. In each case, loss-of-function alleles cause significant phenotypes when they are homozygotic, implying a specific function for these proteins. The remaining five have no obvious distinguishing feature; however, two of these, sqt-1 and rol-6, have very weak or undetectable phenotypes associated with loss-of-function alleles2. All of the original alleles of these two genes are gain-of-function, loss-of-function only being obtained as intragenic suppressors of gain-of-function alleles. These groups are not unique to C. elegans. Genes that are closely related to the different proposed groups in C. elegans have been described for various distantly related nematode species; examples for all of the main proposed groups can be found in the sequence databases, indicating that the different groups must have evolved before the divergence of many (if not all) modern day nematodes. This is supported by the observation that some individual members of groups within C. elegans have closer relatives within other species than within their group in C. elegans. As an example, a gene termed Mjcol-3 from the plant parasitic nematode species Meloidogyne javanica26 is predicted to encode a collagen that has significantly greater similarity with that predicted for dpy-7 of C. elegans than dpy-7 is predicted to have with the other two members of its group within C. elegans.
Regulated expression of cuticular collagen genes Cuticular collagen genes are subject to strict spatial and temporal modes of regulation. As stated earlier, the cuticle TIG January 2000, volume 16, No. 1
23
Reviews
C. elegans cuticle collagen genes
FIGURE 3. Organization of cuticular collagen ccc
cc
c c
Gly-X-Y
N
c50b6.4 zk1193.1 f27c1.8
41–42 aa
AG..QCNCGAQSSGCPA AGFEQCNCGPKSEGCPA RSNSQCSCGLPSQGCPA
f54d1.3 b0222.8 f17c8.2
GLPAWCQCEPTKPTCPP GLPAWCQCEPTKPTCPP GLPAWCQCEPAKPVCPP
k02d7.3 f46c8.2 f14h12.1
SGGGSCCSCGV GGGGGCCGCGV EPAPQCCTCQQ
c53b4.5 f55c10.3 t05a1.2
VGNGQCEGCCLP ASTGGCDACCLP QEHNSCDGCCLP
t14b4.7 t14b4.6
PKEPCEPITPPPCKPCPE PQQPCDPITPPPCQPCPQ PQSPCEPLTPPPCPACPP
GVCPKYCALDGGVFFED GICPKYCAIDGGVFFED GICPKYCAIDGGVFFED
134 aa
VDLEPQEELPCSICPA VPLDPEPAFPCVICPA IQPESEPELPCVICPA
PLETECPGCCIP PQFQECPACCIP
NQTCPLNQVREPPPCRPCPK NASCIPERVFEPPPCLPCPQ
47 aa
GACDHCPPPRTAPGY GGCEHCPPPRTAPGY GDCFHCPTPRTPPGY
~132 aa
GVLHQCSQCTRLHCPQ SDNQQCTSCVQLRCPP MGGAFCKGCFLLSCPQ
dpy-2
AAYCACPPRSAVFLSRH AAYCACPPRSAVFVSRH AAYCPCPPRSAVFVNRFAH
~130 aa
NQPAGPDSF.CFDCPA ATPAPNYDW.CFDCPP APSDGLQSEPCMICPP
42 aa
dpy-7 f56d5.1 f46c8.6 c31h2.2
TYAPINCPQVSFDCIKCPA TFAPLTCAPVSQDCVKCPE TFAPITCPPKDPSCVKCPE
43–44 aa
Group 3
AAYCPCPARSVAVQRS AGYCPCPSRAAYKA AEYCPCPERKRRRV
133 aa
45 aa
Group 2
~133 aa
VGIVSE.GGPCIKCPA VAITHDIPGGCIKCPP IPIPNDFPKECIKCPA 42 aa
Group 1a
Domain III
Domain II
Domain I
Group 1
C
Gly-X-Y
TYCPSDCGVNTILSQFG GHCPSSCGVQEIVAPSV SYCPSDCGVQPILTEMF
131 aa
GVCVCQNVDSILLIN GTCVCQDTEVVMNDE trends in Genetics
The predicted organization of a typical cuticular collagen. The positions of the two blocks of Gly-X-Y sequence are indicated, as are the three cysteine-containing domains. Underneath this, the actual sequence for the three cysteine Domains I, II and III for representative examples of genes for all of the proposed classes. The gene names are those used by the Caenorhabditis elegans genome project that are GENEFINDER predictions39. The number of amino acid residues between conserved cysteines is indicated above the sequence for each class. Three examples from within each proposed group have been selected for comparison avoiding close neighbours within groups, except dpy-7 and dpy-2 groups that contain three and two members. Group 1 has 68 members, Group 1a has 12 members, Group 2 has 38 members and Group 3 has 31 members. The molecules are predicted to have a carboxy non-collagen tail of varying lengths, but frequently very short. For many of the genes, the entire carboxy tail is the region indicated as Domain III in the upper section. However, some genes are predicted to encode significant extensions beyond this domain. This multi-gene family is generally dispersed throughout the genome; however, there are several examples of local clusters. In those instances where genes are adjacent to one another in the genome, they are frequently closely related by sequence and are hence within the same group.
is synthesized by specialized epithelial cells termed the hypodermis5. Reporter gene fusions between the promoters plus regulatory sequences, for several cuticular collagen genes and either lac-Z or green fluorescent protein (GFP) give specific expression of the reporter gene within the hypodermis of transgenic animals6,27. (To date, we have tested eight cuticular collagen genes in this way.) Transgene expression of dpy-7/GFP has been shown within hypodermal cells (Fig. 4). For the gene dpy-7, we have been able to show that only a relatively short region of upstream 24
TIG January 2000, volume 16, No. 1
gene sequence, approximately 125 bp, is sufficient to drive apparently wild-type tissue-specific expression of this gene27. We have recently generated a specific monoclonal antibody against the C-terminal tail of the DPY-7 protein. Preceding the secretion of the cuticle, DPY-7 protein is detected within hypodermal cells; after secretion, it localizes to the annular furrows of the cuticle (L. McMahon and I.L. Johnstone, unpublished). The cellular localization of pre-secreted DPY-7 is similar to the predicted expression pattern indicated by the reporter gene assays.
Reviews
C. elegans cuticle collagen genes
Assembly of collagens Very little is known about the control of cuticular collagen assembly into the multi-protein cuticle structure. If the molecules are true collagens, the primary polymerization must be the formation of a trimer with the collagen triple helix, followed by subsequent higher order interactions between trimers that ultimately form the cuticle structure. Mutant alleles predicted to cause glycine substitutions within the collagen-like Gly-X-Y repeats for some of the mutationally defined genes discussed above have been shown to interfere with function, supporting the involvement of these peptide sequences in trimer formation1,11,13,14. Similar lesions can destabilize the collagen triple helix in vertebrate collagens8,28. The cuticle is a complex, multi-layered structure. The temporal series of cuticular collagen gene expressions described above might exist to promote the formation of distinct substructures (possibly layers) within the cuticle. If this is the case, distinct cuticle collagens would localize to specific regions of the cuticle, and collagens expressed at the same time might localize to the same sub-cuticular structure or region, or at least to a structure created at the same relative time. The proposed primary interaction of cuticular collagen trimer formation is only likely to happen between molecules that are expressed at the same time. For reasons that relate to the behaviour of glycine-substitution mutants, trimerization of cuticular collagens is likely to be restricted either to homo-trimerization or trimerization between very closely related members of cuticle collagen families1. For some vertebrate collagens, the C-terminal non-collagen domain has been shown to be involved in the primary act of trimer formation29 forming a nucleus for
FIGURE 4. Transgenic dpy-7/GFP reporter gene fusion
An animal transgenic for a dpy-7/GFP reporter gene fusion. (a) Differential interference contrast (DIC) image and (b) fluorescence for green fluorescent protein (GFP). The GFP is localized to the nucleus by a nuclear localization signal. This reporter transgene is expressed in most hypodermal cells. The strain was generated by Eric Stewart. Scalebar represents 0.1 mm.
the ‘zipper’-like mechanism that propagates in a C- to Nterminal direction to form the collagen triple helix30. An obvious feature of the C-terminal non-collagen domain of all predicted cuticular collagens is the two cysteine residues in Domain III (Fig. 2) that are positioned immediately C-terminal to the last Gly-X-Y repeats. Extraction of collagens from the C. elegans cuticle requires the presence of reducing agents18; clearly disulphide bonds could be involved in interactions between the three monomers of a trimer or with higher order interactions between trimers. Given the very small size of most of the cuticular collagen C-terminal tails (typically 15–20 residues) and the
FIGURE 5. Temporal changes in mRNA abundance Relative mRNA abundance
In addition to the tissue-specific regulated expression, cuticular collagen genes are under tight temporal control. During the life cycle, a cuticle is synthesized five times, once in the embryo before hatching, and then at the end of each of the four larval stages before moulting (Fig. 5). As might be expected, cuticular collagen genes are expressed in a temporal pattern that reflects this repetitious set of cuticle-synthesis events, once per synthesis. However, genes are not all expressed at the same time during synthesis (Fig. 5). Clear temporal differences can be detected6, for example, dpy-7 mRNA shows a peak of abundance approximately 4 h before each moult. By contrast, col-12 mRNA abundance peaks 4 h later, at each moult. We have analyzed approximately 20 cuticular collagen genes in this way, the data for six of which have been published6. In addition to the early and late expression times represented here by the genes dpy-7 and col-12, we find genes such as dpy-13 and sqt-1 that are expressed in between these two temporal extremes. For those cuticular collagen genes that show such oscillating peaks of mRNA abundance relative to the larval life cycle, time of expression relative to the moult is effectively constant. For example, a gene that is expressed 4 h before the L1 moult is expressed approximately 4 h before each moult. In addition to this repetitious larval expression pattern, we have detected genes that are expressed at a significant level only in the adult after the moulting cycle is complete. Also, for the genes that are expressed in an oscillating pattern during larval development, some are expressed in the adult after the L4 to adult moult and some are not – col-12 is, dpy-7 is not (Fig. 5). The adult animal grows significantly without moulting. It is possible that a limited set of collagen genes is expressed in the adult as part of an adult cuticle growth mechanism.
dpy-7 col-12
12 10 8 6 4 2 0 1
10
20
30
40
Hours post L1 developmental arrest L1
L3 L4 Adult L2 moult moult moult moult trends in Genetics
The graphs show life-cycle temporal changes in mRNA abundance (as measured relative to a control transcript, detailed in Ref. 6) for the two cuticular collagen genes dpy-7 and col-12. Time is indicated on the x-axis as hours after release from an L1 larval developmental arrest that is induced by hatching embryos in the absence of food. Timing of the four moults and the life cycle stages are indicated below.
TIG January 2000, volume 16, No. 1
25
Reviews
C. elegans cuticle collagen genes
close proximity of the conserved cysteine residues within the tail to the Gly-X-Y repeats, it is worth considering the possible involvement of these cysteine residues in interchain disulphide bond formation as part of the generation of a nucleus for triple helix formation or in providing correct register between monomers. The absolute conservation of these residues in all predicted cuticular collagens indicates their functional importance, but does not distinguish between importance for assembly of the trimer and subsequent interactions or function. Ethyl methanesulphonate (EMS)-induced mutations in the sqt-1 gene that cause a substitution of one of these cysteines with either tyrosine or serine are able to generate a recessive left-roller phenotype that is distinct from the almost wild-type phenotype of null mutations in this gene20,21. That the phenotype is distinct from the complete loss of the SQT-1 protein might indicate the presence of the mutant protein within the cuticle, and suggests that the single-cysteine-substitution SQT-1 protein is assembled to some degree. The phenotype of these mutants might therefore be the result of the presence of aberrant protein within the cuticle as opposed to the loss of wild-type protein. However, the recessive nature of these alleles over wild type suggests that either in the presence of wild-type protein, mutant protein fails to produce a phenotype, or wild-type protein is more efficiently assembled into the cuticle than the mutant. In vitro substitution of either C-terminal cysteine in SQT-1 or ROL-6 (a closely related collagen to SQT-1) can generate a similar left-roller phenotype when transgenically expressed in the respective null mutant background20,21. Existing data is not conclusive as to the precise function of these conserved cysteines, especially whether they form bonds between the three members of a trimer or between trimers. Indeed, these two distinct possibilities need not be exclusive. It is at least conceivable that disulphide bonds formed within a trimer during its assembly could be broken and reformed between trimers at a later stage in collagen biosynthesis. The class of enzyme that could perform such an exchange of disulphide bonds is protein disulphide isomerase (PDI). In vertebrates, PDIs have been shown to be involved, along with other chaperones, in the trimerization of some collagens31. C. elegans has two PDI genes, pdi-1 and pdi-2. The gene pdi-1 exists as part of an operon with a gene termed cyp-9. Reporter-gene assays suggest that this operon is transcribed specifically within the same hypodermal cells that synthesize cuticular collagen32. Interestingly, cyp-9 encodes a putative peptidyl prolyl cis–trans isomerase, another enzyme implicated in promoting collagen triple helix formation in vertebrates33. There is, therefore, a possible functional relationship of cuticle collagen biosynthesis for these two co-transcribed C. elegans genes. The PDI, at least in theory, provides the enzymatic function for disulphide bond partner exchanges.
The developmental decision to synthesize a cuticle Moulting in insects is hormonally controlled, and a derivative of cholesterol, 20-hydroxyecdysone, plays a central References 1 Johnstone, I.L. (1994) The cuticle of the nematode Caenorhabditis elegans: a complex collagen structure. BioEssays 16, 171–178 2 Kramer, J.M. (1997) Extracellular Matrix. In C. elegans (Vol. II) (Riddle, D.L. et al., eds), pp. 471–500, Cold Spring Harbor
26
role in moulting in Drosophila melanogaster, its activity being mediated through its interaction with nuclear hormone receptor (NHR) class transcription factors. Evidence in Drosophila suggests a cascade of gene expression in response to the ecdysone hormone, with the transcription of genes, including NHR genes, being induced early in response to the hormone34. The early response genes are thought to be involved in the subsequent transcriptional control of late response genes. C. elegans has a very large family of nuclear-hormone-receptor-like transcription factor genes; approximately 270 are predicted by the C. elegans genome project35. One of these, nhr-23, is expressed in the hypodermis and its activity has been implicated as a requirement for normal moulting36. Recently, analysis of a C. elegans gene, lrp-1, predicted to encode a member of the low density lipoprotein receptor family of proteins, was shown to be required for normal execution of moulting in C. elegans37. C. elegans has a nutritional requirement for sterol. Sterol starvation can mimic the phenotype, including moulting defects, of loss of function of lrp-1 (Ref. 37). LRP-1 is synthesized within and secreted apically from the major hypodermal cells – those cells also responsible for most of the cuticle synthesis. Yochem et al. speculate that the role of LRP-1 might be to endocytose sterols from extracellular fluids37. That lrp-1 mutants and sterol starvation can both generate moulting defects is interesting; however, as yet, it is unclear why this occurs. One possibility is a need for sterols for the synthesis of a signalling molecule like ecdysone used in controlling the moulting cycle. However, other possibilities, including more structural roles relating to the mechanics of moulting or hypodermal membrane behaviour, must also exist. Finally, a phylogenetic analysis of 18S ribosomal DNA sequences suggests a close relationship between arthropods, nematodes and all other moulting phyla38. Although the involvement of a cholesterol derivative such as ecdysone has not as yet been demonstrated in nematodes, there is a body of evidence that might point towards a common evolutionary origin of moulting for these animal groups.
Acknowledgements The C. elegans Genome Project is responsible for the generation of most of the gene sequence data discussed here and the contribution of all those who have worked on this project is acknowledged. The developers and curators of ACeDB are also thanked for the essential provision of information. Thanks to J. Kramer for discussion and data on bli-1 and bli-2 before publication. I would also like to thank members of my lab, J. Muriel and L. McMahon for unpublished data and to Eric Stewart for the transgenic dpy-7/GFP C. elegans strain used for the images in Fig. 4. Tragically and very unexpectedly, Eric died earlier last year. His presence in the lab is sorely missed. Thanks also to T. Page for frequent discussions on worm cuticles and collagen processing enzymes. I.L.J. is the recipient of a Medical Research Council Senior Fellowship in Biomedical Research.
Laboratory Press 3 Cox, G.N. et al. (1981) The cuticle of Caenorhabditis elegans. II. Stage-specific changes in ultrastructure and protein composition during postembryonic development. Dev. Biol. 86, 456–470 4 Peixoto, C.A. and Desouza, W. (1995) Freeze-fracture and deep-etched view of the cuticle of Caenorhabditis elegans.
TIG January 2000, volume 16, No. 1
Tissue Cell 27, 561–568 5 Singh, R.N. and Sulston, J.E. (1978) Some observations on moulting in Caenorhabditis elegans. Nematologica 24, 63–71 6 Johnstone, I.L. and Barry, J.D. (1996) Temporal reiteration of a precise gene expression pattern during nematode development. EMBO J. 15, 3633–3639
Reviews
C. elegans cuticle collagen genes
7 Prockop, D.J. (1998) What holds us together? Why do some of us fall apart? What can we do about it? Matrix Biol. 16, 519–528 8 Prockop, D.J. and Kivirikko, K.I. (1995) Collagens – molecular biology, diseases, and potentials for therapy. Annu. Rev. Biochem. 64, 403–434 9 Ramachandran, G.N. (1967) Structure of collagen at the molecular level. In Treatise on Collagen (Ramachandran, G.N., ed.), pp. 103–183, Academic Press. 10 Ramachandran, G.N. and Ramakrishnan, C. (1976) Molecular structure. In Biochemistry of Collagen (Ramachandran, G.N. and Reddi, A.H., eds), pp. 45–81, Plenum Press 11 Kramer, J.M. (1994) Genetic analysis of extracellular matrix in C. elegans. Annu. Rev. Genet. 28, 95–116 12 Kramer, J.M. et al. (1988) The sqt-1 gene of C. elegans encodes a collagen critical for organismal morphogenesis. Cell 55, 555–565 13 Levy, A.D. et al. (1993) Molecular and genetic analyses of the Caenorhabditis elegans dpy- 2 and dpy-10 collagen genes – a variety of molecular alterations affect organismal morphology. Mol. Biol. Cell 4, 803–817 14 Johnstone, I.L. et al. (1992) Molecular analysis of mutations in the Caenorhabditis elegans collagen gene dpy-7. EMBO J. 11, 3857–3863 15 Francis, R. and Waterston, R.H. (1991) Muscle-cell attachment in Caenorhabditis elegans. J. Cell Biol. 114, 465–479 16 Hresko, M. et al. (1999) Myotactin, a novel hypodermal protein involved in muscle-cell adhesion in Caenorhabditis elegans. J. Cell Biol. 146, 659–672 17 Costa, M. et al. (1997) The role of actin filaments in patterning the Caenorhabditis elegans cuticle. Dev. Biol. 184, 373–384 18 Cox, G.N. et al. (1981) Cuticle of Caenorhabditis elegans: its isolation and partial characterisation. J. Cell Biol. 90, 7–17 19 Higgins, B.J. and Hirsh, D. (1977) Roller mutants of the
nematode Caenorhabditis elegans. Mol. Gen. Genet. 150, 63–72 20 Kramer, J.M. and Johnson, J.J. (1993) Analysis of mutations in the sqt-1 and rol-6 collagen genes of Caenorhabditis elegans. Genetics 135, 1035–1045 21 Yang, J. and Kramer, J.M. (1994) In vitro mutagenesis of Caenorhabditis elegans cuticle collagens identifies a potential subtilisin-like protease cleavage site and demonstrates that carboxyl domain disulfide bonding is required for normal function but not assembly. Mol. Cell. Biol. 14, 2722–2730 22 Kramer, J.M. et al. (1990) The Caenorhabditis elegans rol-6 gene, which interacts with the sqt-1 collagen gene to determine organismal morphology, encodes a collagen. Mol. Cell. Biol. 10, 2081–2089 23 van der Keyl, H. et al. (1994) Caenorhabditis elegans sqt-3 mutants have mutations in the col-1 collagen gene. Dev. Dyn. 201, 86–94 24 von Mende, N. et al. (1988) dpy-13: a nematode collagen gene that affects body shape. Cell 55, 567–576 25 Johnstone, I.L. et al. (1996) Cuticular collagen genes from the parasitic nematode Ostertagia circumcincta. Mol. Biochem. Parasitol. 80, 103–112 26 Koltai, H. et al. (1997) The first isolated collagen gene of the root-knot nematode Meloidogyne javanica is developmentally regulated. Gene 196, 191–199 27 Gilleard, J.S. et al. (1997) cis regulatory requirements for hypodermal cell-specific expression of the Caenorhabditis elegans cuticle collagen gene dpy-7. Mol. Cell. Biol. 17, 2301–2311 28 Byers, P.H. (1990) Brittle bones – fragile molecules: disorders of collagen gene structure and expression. Trends Genet. 6, 293–300 29 Lim, A.L. et al. (1998) Role of the pro-alpha 2(I) COOHterminal region in assembly of type I collagen: Truncation of the last 10 amino acid residues of pro-alpha 2(I) chain
30
31
32
33
34 35
36
37
38
39
prevents assembly of type I collagen heterotrimer. J. Cell. Biochem. 71, 216–232 Engel, J. and Prockop, D.J. (1991) The zipper-like folding of collagen triple helices and the effects of mutations that disrupt the zipper. Annu. Rev. Biophys. Chem. 20, 137–152 Wilson, R. et al. (1998) Protein disulfide isomerase acts as a molecular chaperone during the assembly of procollagen. J. Biol. Chem. 273, 9637–9643 Page, A.P. (1997) Cyclophilin and protein disulfide isomerase genes are co-transcribed in a functionally related manner in Caenorhabditis elegans. DNA Cell Biol. 16, 1335–1343 Steinmann, B. et al. (1991) Cyclosporine-a slows collagen triple-helix formation in vivo – indirect evidence for a physiological-role of peptidyl-prolyl cis- trans-isomerase. J. Biol. Chem. 266, 1299–1303 Russell, S.H. (1996) Nuclear hormone receptors and the Drosophila ecdysone response. Biochem. Soc. Symp. 62, 111–121 Ruvkun, G. and Hobert, O. (1998) The taxonomy of developmental control in Caenorhabditis elegans. Science 282, 2033–2041 Kostrouchova, M. et al. (1998) CHR3: a Caenorhabditis elegans orphan nuclear hormone receptor required for proper epidermal development and molting. Development 125, 1617–1626 Yochem, J. et al. (1999) A gp330/megalin-related protein is required in the major epidermis of Caenorhabditis elegans for completion of molting. Development 126, 597–606 Aguinaldo, A.A. et al. (1997) Evidence for a clade of nematodes, arthropods and other moulting animals. Nature 387, 489–493 C. elegans sequencing consortium (1998) Genome sequence of the nematode C. elegans: A platform for investigating biology. Science 282, 2012–2018
TGFb-related pathways roles in Caenorhabditis elegans development Genetic and molecular analysis in Caenorhabditis elegans has produced new insights into how TGFb-related pathways transduce signals and the developmental processes in which they function. These pathways are essential regulators of dauer formation, body-size determination, male copulatory structures and axonal guidance. Here, we review the insights that have come from standard molecular genetic experiments and discuss how the recently completed genome sequence has contributed to our understanding of these pathways. n recent years, rapid progress has been made in understanding how transforming growth factor-b (TGFb) and related ligands signal, in part because of a wealth of genetic and developmental information previously available on the pathways in which these ligands function. Model genetic systems show us how TGFb-related pathways signal, how they are regulated and what cellular processes they control. As the Caenorhabditis elegans genome is completely sequenced and the tools to analyse these basic cellular processes is expanding, C. elegans will continue to play a major role in elucidating these functions and networks. In this review, we discuss the genetics and developmental biology of C. elegans TGFb signaling. The TGFb superfamily plays critical roles in several important processes, such as cell proliferation, embryonic patterning and cell-type specification1–6. Biochemical iden-
I
0168-9525/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(99)01916-2
tification of serine-threonine kinase receptors as mediators of TGFb signaling was an important advance in the field, as was the identification of cytoplasmic and nuclear effectors belonging to the Smad family (a fusion of sma and Mad gene names). In Drosophila, Mothers against dpp (Mad) was genetically identified as part of the Decapentaplegic (dpp) pathway, and its cDNA sequence indicated it is a cytoplasmic protein, which is consistent with a role as a mediator of receptor signaling7. Work in C. elegans revealed three Smads that function in the same TGFb signaling pathway, suggesting that multiple Smads might be required in other pathways8. Cloning of mammalian homologs demonstrated that these genes are conserved across diverse metazoan phyla8,9. Furthermore, developmental studies in Xenopus led to the identification of Smad2, a potent mesoderm inducer10. These discoveries spurred a flurry of Smad cloning and database harvesting. TIG January 2000, volume 16, No. 1
Garth I. Patterson patterson@waksman. rutgers.edu Richard W. Padgett* padgett@waksman. rutgers.edu Department of Molecular Biology and Biochemistry, 604 Allison Road, Rutgers University, Piscataway, NJ 08854, USA. *Waksman Institute, 190 Frelinghuysen Road, Rutgers University, Piscataway, NJ 08854, USA. 27
C. elegans cuticle collagen genes
Cuticle collagen genes expression in Caenorhabditis elegans Collagen is a structural protein used in the generation of a wide variety of animal extracellular matrices. The exoskeleton of the free-living nematode, Caenorhabditis elegans, is a complex collagen matrix that is tractable to genetic research. Mutations in individual cuticle collagen genes can cause exoskeletal defects that alter the shape of the animal. The complete sequence of the C. elegans genome indicates upwards of 150 distinct collagen genes that probably contribute to this structure. During the synthesis of this matrix, individual collagen genes are expressed in distinct temporal periods, which might facilitate the formation of specific interactions between distinct collagens. he exoskeleton of the free-living nematode Caenorhabditis elegans consists predominantly of small collagen-like proteins encoded by a multi-gene family of approximately 154 members1,2. These molecules are synthesized by specialized epithelial cells and after secretion, polymerize on the apical surface of the epithelium to form a complex multi-layered structure3,4. The entire process of cuticle synthesis is repeated five times during development, once in the embryo before hatching and then towards the end of each of the four larval stages before moulting5. During each synthesis, different collagen genes are expressed in discrete temporal periods, the reason for which could relate to the mechanism of control of collagen polymerization and formation of different structural components of the cuticle6. The basic temporal pattern is repeated during each synthesis. The developmental process controlling collagen gene expression and exoskeletal formation appears to be reset after embryonic hatching and then after each of the larval moults, the cyclic repeats of the process being terminated by the larval to adult moult. Collagens are structural proteins involved in the synthesis of a variety of extracellular matrices in animals7,8. Collagen occurs as a triple helix, the result of trimerization of three monomeric collagen chains9,10. The amino acid sequence of the monomeric collagens is repetitive, with glycine at every third residue. This Gly-X-Y repeat occurs in blocks, the length and organization of which depends on the type of collagen. The residues at the X and Y positions within the repeats are frequently proline or hydroxyproline. In nematodes, as typified by the freeliving species C. elegans, collagens are involved in the formation of two distinct structures, the basement membrane and the cuticle. It is the cuticular collagens, and the multigene family that encode this class of proteins that are the subject of this review.
T
The nematode cuticle The nematode cuticle is an exoskeleton that is synthesized by the underlying epithelial tissue, called the hypodermis1,2,11. Cuticle functions include provision of a barrier between the animal and its environment, maintenance of post-embryonic body shape12–14 and movement via attach0168-9525/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(99)01857-0
ments to muscle15,16. The cuticle is synthesized five times during development and is shed at each moult, first during embryogenesis when the cuticle of the L1 larva is formed, and then again during each of the four larval stages before each moult. Cuticular material is synthesized and secreted apically from epithelial cells and polymerizes on the external epithelial surface (Fig. 1). During synthesis, submembranous actin filaments form within the hypodermis and are organized circumferentially around the cylindrical body of the worm. These filaments are coincident with furrows that form on the epithelial cell membrane during cuticle synthesis and subsequently with circumferential furrows that delineate the annulae (Fig. 1) on the surface of the polymerized cuticle17. This suggests that the shape of the polymerized cuticle is influenced by the shape of the surface on which it polymerizes. In addition to the annulae, longitudinal ridges, termed alae, exist on the lateral surfaces of the cuticle at certain developmental stages, most notably the adult (Figs 1, 2)3. Although both of these structures can be visualized with a light microscope, the cuticle is transparent giving a deceptive view of its true structural complexity. It consists of possibly six definable layers that are distinct in their ultrastructure, as imaged by electron microscopy4,18. The precise nature of the structure and layering varies at different developmental stages3; the mechanisms by which proteins polymerize to form these ordered structures are currently unknown.
Mutant collagen genes The major protein components of the C. elegans cuticle are small collagen-like polypeptides. These are encoded by a multi-gene family, consisting of approximately 154 members. From random mutagenesis of C. elegans, approximately 45 loci have been identified that affect body shape. Currently, 17 of these loci have been assigned to known gene sequences, and nine of these (bli-1, bli-2, dpy-2, dpy-7, dpy-10, dpy-13, rol-6, sqt-1 and sqt-3) are cuticular collagen genes. Among the remainder are genes encoding enzymes that are thought to be involved in collagen processing and genes involved in X-chromosome dosage compensation. Phenotypes associated with mutations in the collagen genes fall predominantly into TIG January 2000, volume 16, No. 1
Iain L. Johnstone [email protected] The Wellcome Centre for Molecular Parasitology, Anderson College, University of Glasgow, Glasgow, UK G11 6NU. 21
Reviews
C. elegans cuticle collagen genes
FIGURE 1. The cuticle and hypodermis of Caenorhabditis elegans Annulus
(a)
Dorsal hypodermis Lateral seam cells Alae
mutations that cause the roller non-dumpy phenotype are more frequently dominant. Among the recessive alleles are true genetic nulls, but most are glycine substitutions within the collagen domains1,11. Dominant alleles include mutations at the proposed ‘subtilisin-like’ cleavage site, which is believed to be required for processing of the mature collagen20,21. Some of the mutations that cause the discussed phenotypes are recessive and lead to loss of function, indicating that at least some of the collagens have specific roles in the formation of the exoskeleton1.
Cuticular collagen structure Ventral hypodermis
Annular furrows
External cortical layer Cell membrane
(b)
Actin filament bundles Hypodermal cells Cortical layers
(c)
Striated layers
trends in Genetics
(a) A section of the cylindrical body of the nematode. Annulae are present on the surface of the cuticle at all stages, alae are visible only on the L1, the dauer and the adult cuticle3. The bottom two drawings represent cross sections through the cuticle and underlying hypodermal cells. (b) The presence of furrows in the apical surface of the hypodermis and juxtaposed bundles of actin filaments that form during cuticle secretion. These are coincident with the positioning of the annular furrows in the newly formed cuticle. (c) The organization of the hypodermis after cuticle secretion, after dissociation of the actin bundles.
three basic types described as: (1) blister (bli), which is a blistering of cuticular material away from the surface of the animal; (2) dumpy (dpy), which is a shortening in the length of the animal as compared with wild type; and (3) roller (rol), which is a helical twisting of the animal’s body (see Fig. 2). The phenotypes are not necessarily exclusive. The blister phenotype (Fig. 2) is restricted to the adult stage and might identify a cuticular structure that is specific to adults. The dumpy phenotype (see dpy-10; Fig. 2) can affect all post-embryonic developmental stages, although mutations in different dpy genes can cause greater or lesser effects at different stages. The defective exoskeleton in dpy mutants significantly alters body shape. This is also true for the roller phenotype. The alae in the cuticle (see Fig. 1) run parallel to the side of the animal in the wild type (Fig. 2d and e, left panels), but are helically twisted along the length of the animal in roller mutants, as indicated by their angled aspect relative to the side of the animal (Fig. 2d and e, right panels). The helical distortion in roller mutants is not restricted to the exoskeleton itself. In the roller phenotype, the exoskeleton, the underlying ectodermal cells and musculature are all helically twisted along the length of the animal with a similar turn to that seen for the alae19. Mutations that cause the blister, dumpy and the combined dumpy and roller phenotypes are generally recessive; 22
TIG January 2000, volume 16, No. 1
A typical polypeptide structure has been predicted for the cuticular collagen genes (Fig. 3). The nine genes discussed above also fit this basic structure13,14,22–24. They are predicted to encode two main blocks of Gly-X-Y collagenlike polypeptide that are flanked and separated by three clusters of cysteine residues, which are termed Domains I, II and III in this review (Fig. 3). The smaller N-terminal Gly-X-Y block typically contains 8–10 Gly-X-Y collagen repeats, the larger C-terminal block has 40–42 Gly-X-Y repeats. Generally, the C-terminal collagen block has one or two very small disruptions within the Gly-X-Y repeats, the precise position and size of which vary between the products of different genes. In addition to these collagenlike regions, the molecules are predicted to have a noncollagen N-terminal region that varies considerably in size, but that contains conserved regions including a predicted signal peptide and a proposed subtilisin-like cleavage site, which might be involved in the generation of a mature polypeptide from a pro-collagen precursor21. Mutations that disrupt this proposed cleavage site have been shown for some genes to be neomorphic2, generating a phenotype that is distinct from both the wild-type and loss-of-function alleles. This could be explained by its insertion into the cuticle of collagen with its pro-domain still attached. In addition to the neomorphic alleles that affect the proposed subtilisin-like pro-collagen maturation site, various other lesions have been characterized for the group of nine genes discussed above13,14,20,23,24. The most common lesion is substitution of a Gly within the Gly-X-Y collagen regions. For some of the genes, true genetic nulls have been identified. Where data are available, the Gly substitutions seem to behave as loss of function. DPY-7 protein containing a Gly substitution appears to accumulate in the cytoplasm during cuticle synthesis, but very little mutant protein gets incorporated into the cuticle (L. McMahon, pers. commun.). It seems likely that collagen that contains such Gly substitutions is recognized as being aberrant by cellular machinery and is not incorporated into the exoskeletal structure. This phenomenon of removal of Gly substitution mutant collagen has also been described for vertebrate collagens8.
Cuticular collagen gene families The complete sequence of the C. elegans genome permits an accurate indication of all possible collagen-like genes. In the absence of experimental data, the number of predicted cuticular collagen genes must depend on the precise criteria used to define this class. There are approximately 154 predicted cuticular collagen genes (http://www. worms.gla.ac.uk/collagen/cecolgenes.htm); in addition, there are several gene predictions that contain some but not all of these regions (shown in Fig. 3). Similarly sized collagen gene families have been described for other nematode species25.
Reviews
C. elegans cuticle collagen genes
FIGURE 2. Phenotypes associated with mutated collagen genes
The top three images are adult animals of (a) the wild-type strain N2, (b) a bli-2 (e768) mutant and (c) a dpy-10 (e128 ) mutant. Blistering of the cuticle in bli-2 is indicated by arrows. dpy-10 is significantly shorter and fatter than wild type. The bottom panels show high magnification images of (d) a wild-type N2 strain and (e) a mutant rol-6 (su1006 ). In each case, the animal’s body lies longitudinal from top to bottom of the image and the image represents approximately half the width of the nematode. Both left-hand panels show a deep optical section where part of the side of the animal is in focus. The orientation of the animal’s side is emphasized by the black vertical line. Cuticular structures termed alae are indicated (black arrow) in a top optical section for both wild type and rol-6 mutant in the right-hand panels. As can be seen, the alae run parallel to the side of the animal in the wild type, but are at an angle in the rol-6 mutant.
Phylogenetic analysis of this gene family is complicated by the high levels of similarity within collagen-encoding regions. The family can be divided into groups based on the pattern of conserved cysteines of their predicted encoded collagens (Fig. 3). The evolutionary significance of these groupings is supported by homologies within the N- and C- non-collagen regions, which place the genes into groups that agree with the groupings that are based on the cysteine patterns alone. Of the 154 genes, 68 belong to Group 1, 12 to Group 1a, 38 to Group 2 and 31 to Group 3. The dpy-7 and dpy-2 groups have three and two members, respectively. Of the genes defined by mutation, bli-1, sqt-1 and rol-6 belong to Group 1; bli-2 belongs to Group 2; dpy-13 and sqt-3 belong to Group 3; dpy-7 belongs to the dpy-7 Group and dpy-2 and dpy-10 constitute the dpy-2 Group. Group 1a is not represented among the genes that are defined mutationally. The nine mutationally defined genes have been identified from random mutagenic screens and selected on the basis that the mutation caused an easily detectable change in body shape. For these nine genes, multiple alleles of each have been obtained. There are only approximately 20 shape-change mutants not yet assigned to a gene defined by sequence. It therefore seems likely that most of the cuticular collagen genes that are defined by sequence might not be mutated readily to generate detectable phenotypes. Because of the high similarities between many members of this gene family, many of the gene products might have overlapping functions. No single, molecularly definable characteristic unites all of the nine genes so far identified. Only the proposed collagen Group 1a is not represented among them, and this is a very highly
conserved group. Four of the nine might be considered unusual as their encoded proteins, DPY-2, DPY-10, DPY-7 and BLI-1 all have significantly longer than average C-terminal tails. In each case, loss-of-function alleles cause significant phenotypes when they are homozygotic, implying a specific function for these proteins. The remaining five have no obvious distinguishing feature; however, two of these, sqt-1 and rol-6, have very weak or undetectable phenotypes associated with loss-of-function alleles2. All of the original alleles of these two genes are gain-of-function, loss-of-function only being obtained as intragenic suppressors of gain-of-function alleles. These groups are not unique to C. elegans. Genes that are closely related to the different proposed groups in C. elegans have been described for various distantly related nematode species; examples for all of the main proposed groups can be found in the sequence databases, indicating that the different groups must have evolved before the divergence of many (if not all) modern day nematodes. This is supported by the observation that some individual members of groups within C. elegans have closer relatives within other species than within their group in C. elegans. As an example, a gene termed Mjcol-3 from the plant parasitic nematode species Meloidogyne javanica26 is predicted to encode a collagen that has significantly greater similarity with that predicted for dpy-7 of C. elegans than dpy-7 is predicted to have with the other two members of its group within C. elegans.
Regulated expression of cuticular collagen genes Cuticular collagen genes are subject to strict spatial and temporal modes of regulation. As stated earlier, the cuticle TIG January 2000, volume 16, No. 1
23
Reviews
C. elegans cuticle collagen genes
FIGURE 3. Organization of cuticular collagen ccc
cc
c c
Gly-X-Y
N
c50b6.4 zk1193.1 f27c1.8
41–42 aa
AG..QCNCGAQSSGCPA AGFEQCNCGPKSEGCPA RSNSQCSCGLPSQGCPA
f54d1.3 b0222.8 f17c8.2
GLPAWCQCEPTKPTCPP GLPAWCQCEPTKPTCPP GLPAWCQCEPAKPVCPP
k02d7.3 f46c8.2 f14h12.1
SGGGSCCSCGV GGGGGCCGCGV EPAPQCCTCQQ
c53b4.5 f55c10.3 t05a1.2
VGNGQCEGCCLP ASTGGCDACCLP QEHNSCDGCCLP
t14b4.7 t14b4.6
PKEPCEPITPPPCKPCPE PQQPCDPITPPPCQPCPQ PQSPCEPLTPPPCPACPP
GVCPKYCALDGGVFFED GICPKYCAIDGGVFFED GICPKYCAIDGGVFFED
134 aa
VDLEPQEELPCSICPA VPLDPEPAFPCVICPA IQPESEPELPCVICPA
PLETECPGCCIP PQFQECPACCIP
NQTCPLNQVREPPPCRPCPK NASCIPERVFEPPPCLPCPQ
47 aa
GACDHCPPPRTAPGY GGCEHCPPPRTAPGY GDCFHCPTPRTPPGY
~132 aa
GVLHQCSQCTRLHCPQ SDNQQCTSCVQLRCPP MGGAFCKGCFLLSCPQ
dpy-2
AAYCACPPRSAVFLSRH AAYCACPPRSAVFVSRH AAYCPCPPRSAVFVNRFAH
~130 aa
NQPAGPDSF.CFDCPA ATPAPNYDW.CFDCPP APSDGLQSEPCMICPP
42 aa
dpy-7 f56d5.1 f46c8.6 c31h2.2
TYAPINCPQVSFDCIKCPA TFAPLTCAPVSQDCVKCPE TFAPITCPPKDPSCVKCPE
43–44 aa
Group 3
AAYCPCPARSVAVQRS AGYCPCPSRAAYKA AEYCPCPERKRRRV
133 aa
45 aa
Group 2
~133 aa
VGIVSE.GGPCIKCPA VAITHDIPGGCIKCPP IPIPNDFPKECIKCPA 42 aa
Group 1a
Domain III
Domain II
Domain I
Group 1
C
Gly-X-Y
TYCPSDCGVNTILSQFG GHCPSSCGVQEIVAPSV SYCPSDCGVQPILTEMF
131 aa
GVCVCQNVDSILLIN GTCVCQDTEVVMNDE trends in Genetics
The predicted organization of a typical cuticular collagen. The positions of the two blocks of Gly-X-Y sequence are indicated, as are the three cysteine-containing domains. Underneath this, the actual sequence for the three cysteine Domains I, II and III for representative examples of genes for all of the proposed classes. The gene names are those used by the Caenorhabditis elegans genome project that are GENEFINDER predictions39. The number of amino acid residues between conserved cysteines is indicated above the sequence for each class. Three examples from within each proposed group have been selected for comparison avoiding close neighbours within groups, except dpy-7 and dpy-2 groups that contain three and two members. Group 1 has 68 members, Group 1a has 12 members, Group 2 has 38 members and Group 3 has 31 members. The molecules are predicted to have a carboxy non-collagen tail of varying lengths, but frequently very short. For many of the genes, the entire carboxy tail is the region indicated as Domain III in the upper section. However, some genes are predicted to encode significant extensions beyond this domain. This multi-gene family is generally dispersed throughout the genome; however, there are several examples of local clusters. In those instances where genes are adjacent to one another in the genome, they are frequently closely related by sequence and are hence within the same group.
is synthesized by specialized epithelial cells termed the hypodermis5. Reporter gene fusions between the promoters plus regulatory sequences, for several cuticular collagen genes and either lac-Z or green fluorescent protein (GFP) give specific expression of the reporter gene within the hypodermis of transgenic animals6,27. (To date, we have tested eight cuticular collagen genes in this way.) Transgene expression of dpy-7/GFP has been shown within hypodermal cells (Fig. 4). For the gene dpy-7, we have been able to show that only a relatively short region of upstream 24
TIG January 2000, volume 16, No. 1
gene sequence, approximately 125 bp, is sufficient to drive apparently wild-type tissue-specific expression of this gene27. We have recently generated a specific monoclonal antibody against the C-terminal tail of the DPY-7 protein. Preceding the secretion of the cuticle, DPY-7 protein is detected within hypodermal cells; after secretion, it localizes to the annular furrows of the cuticle (L. McMahon and I.L. Johnstone, unpublished). The cellular localization of pre-secreted DPY-7 is similar to the predicted expression pattern indicated by the reporter gene assays.
Reviews
C. elegans cuticle collagen genes
Assembly of collagens Very little is known about the control of cuticular collagen assembly into the multi-protein cuticle structure. If the molecules are true collagens, the primary polymerization must be the formation of a trimer with the collagen triple helix, followed by subsequent higher order interactions between trimers that ultimately form the cuticle structure. Mutant alleles predicted to cause glycine substitutions within the collagen-like Gly-X-Y repeats for some of the mutationally defined genes discussed above have been shown to interfere with function, supporting the involvement of these peptide sequences in trimer formation1,11,13,14. Similar lesions can destabilize the collagen triple helix in vertebrate collagens8,28. The cuticle is a complex, multi-layered structure. The temporal series of cuticular collagen gene expressions described above might exist to promote the formation of distinct substructures (possibly layers) within the cuticle. If this is the case, distinct cuticle collagens would localize to specific regions of the cuticle, and collagens expressed at the same time might localize to the same sub-cuticular structure or region, or at least to a structure created at the same relative time. The proposed primary interaction of cuticular collagen trimer formation is only likely to happen between molecules that are expressed at the same time. For reasons that relate to the behaviour of glycine-substitution mutants, trimerization of cuticular collagens is likely to be restricted either to homo-trimerization or trimerization between very closely related members of cuticle collagen families1. For some vertebrate collagens, the C-terminal non-collagen domain has been shown to be involved in the primary act of trimer formation29 forming a nucleus for
FIGURE 4. Transgenic dpy-7/GFP reporter gene fusion
An animal transgenic for a dpy-7/GFP reporter gene fusion. (a) Differential interference contrast (DIC) image and (b) fluorescence for green fluorescent protein (GFP). The GFP is localized to the nucleus by a nuclear localization signal. This reporter transgene is expressed in most hypodermal cells. The strain was generated by Eric Stewart. Scalebar represents 0.1 mm.
the ‘zipper’-like mechanism that propagates in a C- to Nterminal direction to form the collagen triple helix30. An obvious feature of the C-terminal non-collagen domain of all predicted cuticular collagens is the two cysteine residues in Domain III (Fig. 2) that are positioned immediately C-terminal to the last Gly-X-Y repeats. Extraction of collagens from the C. elegans cuticle requires the presence of reducing agents18; clearly disulphide bonds could be involved in interactions between the three monomers of a trimer or with higher order interactions between trimers. Given the very small size of most of the cuticular collagen C-terminal tails (typically 15–20 residues) and the
FIGURE 5. Temporal changes in mRNA abundance Relative mRNA abundance
In addition to the tissue-specific regulated expression, cuticular collagen genes are under tight temporal control. During the life cycle, a cuticle is synthesized five times, once in the embryo before hatching, and then at the end of each of the four larval stages before moulting (Fig. 5). As might be expected, cuticular collagen genes are expressed in a temporal pattern that reflects this repetitious set of cuticle-synthesis events, once per synthesis. However, genes are not all expressed at the same time during synthesis (Fig. 5). Clear temporal differences can be detected6, for example, dpy-7 mRNA shows a peak of abundance approximately 4 h before each moult. By contrast, col-12 mRNA abundance peaks 4 h later, at each moult. We have analyzed approximately 20 cuticular collagen genes in this way, the data for six of which have been published6. In addition to the early and late expression times represented here by the genes dpy-7 and col-12, we find genes such as dpy-13 and sqt-1 that are expressed in between these two temporal extremes. For those cuticular collagen genes that show such oscillating peaks of mRNA abundance relative to the larval life cycle, time of expression relative to the moult is effectively constant. For example, a gene that is expressed 4 h before the L1 moult is expressed approximately 4 h before each moult. In addition to this repetitious larval expression pattern, we have detected genes that are expressed at a significant level only in the adult after the moulting cycle is complete. Also, for the genes that are expressed in an oscillating pattern during larval development, some are expressed in the adult after the L4 to adult moult and some are not – col-12 is, dpy-7 is not (Fig. 5). The adult animal grows significantly without moulting. It is possible that a limited set of collagen genes is expressed in the adult as part of an adult cuticle growth mechanism.
dpy-7 col-12
12 10 8 6 4 2 0 1
10
20
30
40
Hours post L1 developmental arrest L1
L3 L4 Adult L2 moult moult moult moult trends in Genetics
The graphs show life-cycle temporal changes in mRNA abundance (as measured relative to a control transcript, detailed in Ref. 6) for the two cuticular collagen genes dpy-7 and col-12. Time is indicated on the x-axis as hours after release from an L1 larval developmental arrest that is induced by hatching embryos in the absence of food. Timing of the four moults and the life cycle stages are indicated below.
TIG January 2000, volume 16, No. 1
25
Reviews
C. elegans cuticle collagen genes
close proximity of the conserved cysteine residues within the tail to the Gly-X-Y repeats, it is worth considering the possible involvement of these cysteine residues in interchain disulphide bond formation as part of the generation of a nucleus for triple helix formation or in providing correct register between monomers. The absolute conservation of these residues in all predicted cuticular collagens indicates their functional importance, but does not distinguish between importance for assembly of the trimer and subsequent interactions or function. Ethyl methanesulphonate (EMS)-induced mutations in the sqt-1 gene that cause a substitution of one of these cysteines with either tyrosine or serine are able to generate a recessive left-roller phenotype that is distinct from the almost wild-type phenotype of null mutations in this gene20,21. That the phenotype is distinct from the complete loss of the SQT-1 protein might indicate the presence of the mutant protein within the cuticle, and suggests that the single-cysteine-substitution SQT-1 protein is assembled to some degree. The phenotype of these mutants might therefore be the result of the presence of aberrant protein within the cuticle as opposed to the loss of wild-type protein. However, the recessive nature of these alleles over wild type suggests that either in the presence of wild-type protein, mutant protein fails to produce a phenotype, or wild-type protein is more efficiently assembled into the cuticle than the mutant. In vitro substitution of either C-terminal cysteine in SQT-1 or ROL-6 (a closely related collagen to SQT-1) can generate a similar left-roller phenotype when transgenically expressed in the respective null mutant background20,21. Existing data is not conclusive as to the precise function of these conserved cysteines, especially whether they form bonds between the three members of a trimer or between trimers. Indeed, these two distinct possibilities need not be exclusive. It is at least conceivable that disulphide bonds formed within a trimer during its assembly could be broken and reformed between trimers at a later stage in collagen biosynthesis. The class of enzyme that could perform such an exchange of disulphide bonds is protein disulphide isomerase (PDI). In vertebrates, PDIs have been shown to be involved, along with other chaperones, in the trimerization of some collagens31. C. elegans has two PDI genes, pdi-1 and pdi-2. The gene pdi-1 exists as part of an operon with a gene termed cyp-9. Reporter-gene assays suggest that this operon is transcribed specifically within the same hypodermal cells that synthesize cuticular collagen32. Interestingly, cyp-9 encodes a putative peptidyl prolyl cis–trans isomerase, another enzyme implicated in promoting collagen triple helix formation in vertebrates33. There is, therefore, a possible functional relationship of cuticle collagen biosynthesis for these two co-transcribed C. elegans genes. The PDI, at least in theory, provides the enzymatic function for disulphide bond partner exchanges.
The developmental decision to synthesize a cuticle Moulting in insects is hormonally controlled, and a derivative of cholesterol, 20-hydroxyecdysone, plays a central References 1 Johnstone, I.L. (1994) The cuticle of the nematode Caenorhabditis elegans: a complex collagen structure. BioEssays 16, 171–178 2 Kramer, J.M. (1997) Extracellular Matrix. In C. elegans (Vol. II) (Riddle, D.L. et al., eds), pp. 471–500, Cold Spring Harbor
26
role in moulting in Drosophila melanogaster, its activity being mediated through its interaction with nuclear hormone receptor (NHR) class transcription factors. Evidence in Drosophila suggests a cascade of gene expression in response to the ecdysone hormone, with the transcription of genes, including NHR genes, being induced early in response to the hormone34. The early response genes are thought to be involved in the subsequent transcriptional control of late response genes. C. elegans has a very large family of nuclear-hormone-receptor-like transcription factor genes; approximately 270 are predicted by the C. elegans genome project35. One of these, nhr-23, is expressed in the hypodermis and its activity has been implicated as a requirement for normal moulting36. Recently, analysis of a C. elegans gene, lrp-1, predicted to encode a member of the low density lipoprotein receptor family of proteins, was shown to be required for normal execution of moulting in C. elegans37. C. elegans has a nutritional requirement for sterol. Sterol starvation can mimic the phenotype, including moulting defects, of loss of function of lrp-1 (Ref. 37). LRP-1 is synthesized within and secreted apically from the major hypodermal cells – those cells also responsible for most of the cuticle synthesis. Yochem et al. speculate that the role of LRP-1 might be to endocytose sterols from extracellular fluids37. That lrp-1 mutants and sterol starvation can both generate moulting defects is interesting; however, as yet, it is unclear why this occurs. One possibility is a need for sterols for the synthesis of a signalling molecule like ecdysone used in controlling the moulting cycle. However, other possibilities, including more structural roles relating to the mechanics of moulting or hypodermal membrane behaviour, must also exist. Finally, a phylogenetic analysis of 18S ribosomal DNA sequences suggests a close relationship between arthropods, nematodes and all other moulting phyla38. Although the involvement of a cholesterol derivative such as ecdysone has not as yet been demonstrated in nematodes, there is a body of evidence that might point towards a common evolutionary origin of moulting for these animal groups.
Acknowledgements The C. elegans Genome Project is responsible for the generation of most of the gene sequence data discussed here and the contribution of all those who have worked on this project is acknowledged. The developers and curators of ACeDB are also thanked for the essential provision of information. Thanks to J. Kramer for discussion and data on bli-1 and bli-2 before publication. I would also like to thank members of my lab, J. Muriel and L. McMahon for unpublished data and to Eric Stewart for the transgenic dpy-7/GFP C. elegans strain used for the images in Fig. 4. Tragically and very unexpectedly, Eric died earlier last year. His presence in the lab is sorely missed. Thanks also to T. Page for frequent discussions on worm cuticles and collagen processing enzymes. I.L.J. is the recipient of a Medical Research Council Senior Fellowship in Biomedical Research.
Laboratory Press 3 Cox, G.N. et al. (1981) The cuticle of Caenorhabditis elegans. II. Stage-specific changes in ultrastructure and protein composition during postembryonic development. Dev. Biol. 86, 456–470 4 Peixoto, C.A. and Desouza, W. (1995) Freeze-fracture and deep-etched view of the cuticle of Caenorhabditis elegans.
TIG January 2000, volume 16, No. 1
Tissue Cell 27, 561–568 5 Singh, R.N. and Sulston, J.E. (1978) Some observations on moulting in Caenorhabditis elegans. Nematologica 24, 63–71 6 Johnstone, I.L. and Barry, J.D. (1996) Temporal reiteration of a precise gene expression pattern during nematode development. EMBO J. 15, 3633–3639
Reviews
C. elegans cuticle collagen genes
7 Prockop, D.J. (1998) What holds us together? Why do some of us fall apart? What can we do about it? Matrix Biol. 16, 519–528 8 Prockop, D.J. and Kivirikko, K.I. (1995) Collagens – molecular biology, diseases, and potentials for therapy. Annu. Rev. Biochem. 64, 403–434 9 Ramachandran, G.N. (1967) Structure of collagen at the molecular level. In Treatise on Collagen (Ramachandran, G.N., ed.), pp. 103–183, Academic Press. 10 Ramachandran, G.N. and Ramakrishnan, C. (1976) Molecular structure. In Biochemistry of Collagen (Ramachandran, G.N. and Reddi, A.H., eds), pp. 45–81, Plenum Press 11 Kramer, J.M. (1994) Genetic analysis of extracellular matrix in C. elegans. Annu. Rev. Genet. 28, 95–116 12 Kramer, J.M. et al. (1988) The sqt-1 gene of C. elegans encodes a collagen critical for organismal morphogenesis. Cell 55, 555–565 13 Levy, A.D. et al. (1993) Molecular and genetic analyses of the Caenorhabditis elegans dpy- 2 and dpy-10 collagen genes – a variety of molecular alterations affect organismal morphology. Mol. Biol. Cell 4, 803–817 14 Johnstone, I.L. et al. (1992) Molecular analysis of mutations in the Caenorhabditis elegans collagen gene dpy-7. EMBO J. 11, 3857–3863 15 Francis, R. and Waterston, R.H. (1991) Muscle-cell attachment in Caenorhabditis elegans. J. Cell Biol. 114, 465–479 16 Hresko, M. et al. (1999) Myotactin, a novel hypodermal protein involved in muscle-cell adhesion in Caenorhabditis elegans. J. Cell Biol. 146, 659–672 17 Costa, M. et al. (1997) The role of actin filaments in patterning the Caenorhabditis elegans cuticle. Dev. Biol. 184, 373–384 18 Cox, G.N. et al. (1981) Cuticle of Caenorhabditis elegans: its isolation and partial characterisation. J. Cell Biol. 90, 7–17 19 Higgins, B.J. and Hirsh, D. (1977) Roller mutants of the
nematode Caenorhabditis elegans. Mol. Gen. Genet. 150, 63–72 20 Kramer, J.M. and Johnson, J.J. (1993) Analysis of mutations in the sqt-1 and rol-6 collagen genes of Caenorhabditis elegans. Genetics 135, 1035–1045 21 Yang, J. and Kramer, J.M. (1994) In vitro mutagenesis of Caenorhabditis elegans cuticle collagens identifies a potential subtilisin-like protease cleavage site and demonstrates that carboxyl domain disulfide bonding is required for normal function but not assembly. Mol. Cell. Biol. 14, 2722–2730 22 Kramer, J.M. et al. (1990) The Caenorhabditis elegans rol-6 gene, which interacts with the sqt-1 collagen gene to determine organismal morphology, encodes a collagen. Mol. Cell. Biol. 10, 2081–2089 23 van der Keyl, H. et al. (1994) Caenorhabditis elegans sqt-3 mutants have mutations in the col-1 collagen gene. Dev. Dyn. 201, 86–94 24 von Mende, N. et al. (1988) dpy-13: a nematode collagen gene that affects body shape. Cell 55, 567–576 25 Johnstone, I.L. et al. (1996) Cuticular collagen genes from the parasitic nematode Ostertagia circumcincta. Mol. Biochem. Parasitol. 80, 103–112 26 Koltai, H. et al. (1997) The first isolated collagen gene of the root-knot nematode Meloidogyne javanica is developmentally regulated. Gene 196, 191–199 27 Gilleard, J.S. et al. (1997) cis regulatory requirements for hypodermal cell-specific expression of the Caenorhabditis elegans cuticle collagen gene dpy-7. Mol. Cell. Biol. 17, 2301–2311 28 Byers, P.H. (1990) Brittle bones – fragile molecules: disorders of collagen gene structure and expression. Trends Genet. 6, 293–300 29 Lim, A.L. et al. (1998) Role of the pro-alpha 2(I) COOHterminal region in assembly of type I collagen: Truncation of the last 10 amino acid residues of pro-alpha 2(I) chain
30
31
32
33
34 35
36
37
38
39
prevents assembly of type I collagen heterotrimer. J. Cell. Biochem. 71, 216–232 Engel, J. and Prockop, D.J. (1991) The zipper-like folding of collagen triple helices and the effects of mutations that disrupt the zipper. Annu. Rev. Biophys. Chem. 20, 137–152 Wilson, R. et al. (1998) Protein disulfide isomerase acts as a molecular chaperone during the assembly of procollagen. J. Biol. Chem. 273, 9637–9643 Page, A.P. (1997) Cyclophilin and protein disulfide isomerase genes are co-transcribed in a functionally related manner in Caenorhabditis elegans. DNA Cell Biol. 16, 1335–1343 Steinmann, B. et al. (1991) Cyclosporine-a slows collagen triple-helix formation in vivo – indirect evidence for a physiological-role of peptidyl-prolyl cis- trans-isomerase. J. Biol. Chem. 266, 1299–1303 Russell, S.H. (1996) Nuclear hormone receptors and the Drosophila ecdysone response. Biochem. Soc. Symp. 62, 111–121 Ruvkun, G. and Hobert, O. (1998) The taxonomy of developmental control in Caenorhabditis elegans. Science 282, 2033–2041 Kostrouchova, M. et al. (1998) CHR3: a Caenorhabditis elegans orphan nuclear hormone receptor required for proper epidermal development and molting. Development 125, 1617–1626 Yochem, J. et al. (1999) A gp330/megalin-related protein is required in the major epidermis of Caenorhabditis elegans for completion of molting. Development 126, 597–606 Aguinaldo, A.A. et al. (1997) Evidence for a clade of nematodes, arthropods and other moulting animals. Nature 387, 489–493 C. elegans sequencing consortium (1998) Genome sequence of the nematode C. elegans: A platform for investigating biology. Science 282, 2012–2018
TGFb-related pathways roles in Caenorhabditis elegans development Genetic and molecular analysis in Caenorhabditis elegans has produced new insights into how TGFb-related pathways transduce signals and the developmental processes in which they function. These pathways are essential regulators of dauer formation, body-size determination, male copulatory structures and axonal guidance. Here, we review the insights that have come from standard molecular genetic experiments and discuss how the recently completed genome sequence has contributed to our understanding of these pathways. n recent years, rapid progress has been made in understanding how transforming growth factor-b (TGFb) and related ligands signal, in part because of a wealth of genetic and developmental information previously available on the pathways in which these ligands function. Model genetic systems show us how TGFb-related pathways signal, how they are regulated and what cellular processes they control. As the Caenorhabditis elegans genome is completely sequenced and the tools to analyse these basic cellular processes is expanding, C. elegans will continue to play a major role in elucidating these functions and networks. In this review, we discuss the genetics and developmental biology of C. elegans TGFb signaling. The TGFb superfamily plays critical roles in several important processes, such as cell proliferation, embryonic patterning and cell-type specification1–6. Biochemical iden-
I
0168-9525/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. PII: S0168-9525(99)01916-2
tification of serine-threonine kinase receptors as mediators of TGFb signaling was an important advance in the field, as was the identification of cytoplasmic and nuclear effectors belonging to the Smad family (a fusion of sma and Mad gene names). In Drosophila, Mothers against dpp (Mad) was genetically identified as part of the Decapentaplegic (dpp) pathway, and its cDNA sequence indicated it is a cytoplasmic protein, which is consistent with a role as a mediator of receptor signaling7. Work in C. elegans revealed three Smads that function in the same TGFb signaling pathway, suggesting that multiple Smads might be required in other pathways8. Cloning of mammalian homologs demonstrated that these genes are conserved across diverse metazoan phyla8,9. Furthermore, developmental studies in Xenopus led to the identification of Smad2, a potent mesoderm inducer10. These discoveries spurred a flurry of Smad cloning and database harvesting. TIG January 2000, volume 16, No. 1
Garth I. Patterson patterson@waksman. rutgers.edu Richard W. Padgett* padgett@waksman. rutgers.edu Department of Molecular Biology and Biochemistry, 604 Allison Road, Rutgers University, Piscataway, NJ 08854, USA. *Waksman Institute, 190 Frelinghuysen Road, Rutgers University, Piscataway, NJ 08854, USA. 27