The proprotein convertase BLI-4 promotes collagen secretion prior to assembly of the Caenorhabditis elegans cuticle

Abstract

Some types of collagens, including transmembrane MACIT collagens and C. elegans cuticle collagens, are N-terminally cleaved at a dibasic site that resembles the consensus for furin or other proprotein convertases of the subtilisin/kexin (PCSK) family. Such cleavage may release transmembrane collagens from the plasma membrane and affect extracellular matrix assembly or structure. However, the functional consequences of such cleavage are unclear and evidence for the role of specific PCSKs is lacking. Here, we used endogenous collagen fusions to fluorescent proteins to visualize the secretion and assembly of the first collagen-based cuticle in C. elegans and then tested the role of the PCSK BLI-4 in these processes. Unexpectedly, we found that cuticle collagens SQT-3 and DPY-17 are secreted into the extraembryonic space several hours before cuticle matrix assembly. Furthermore, this early secretion depends on BLI-4/PCSK; in bli-4 and cleavage-site mutants, SQT-3 and DPY-17 are not efficiently secreted and instead form large intracellular puncta. Their later assembly into cuticle matrix is reduced but not entirely blocked. These data reveal a role for collagen N-terminal processing in intracellular trafficking and the control of matrix assembly in vivo. Our observations also prompt a revision of the classic model for C. elegans cuticle matrix assembly and the pre-cuticle-to-cuticle transition, suggesting that cuticle layer assembly proceeds via a series of regulated steps and not simply by sequential secretion and deposition.

Fig 1. Matrix dynamics during assembly of the first pre-cuticle and cuticle.

C. elegans embryos expressing functional pre-cuticle (A-B) or cuticle collagen (C-D) fusion proteins from endogenously-tagged loci. A) noah-1(mc68 [NOAH-1::mCherry(int)]). B) lpr-3(cs250 [ss::SfGFP::LPR-3]). C) dpy-17(syb3685 [DPY-17::mNG]). D) sqt-3(syb3691 [SQT-3::mNG(int)]). Embryos were selected at the 1.5-fold stage and then incubated for the indicated number of hours prior to imaging. Magenta arrowheads indicate pre-cuticle sheath. Green arrowheads indicate cuticle and white arrows indicate extracellular DPY-17::mNG aggregates. Asterisks indicate secreted fusion protein within the extraembryonic space (EES). All images are maximum intensity projections from confocal Z-slices, shown in inverted grayscale for clarity. Images are representative of at least n = 5 embryos per genotype per stage for A and B, and n = 10 embryos per genotype per stage for C and D. Scale bar, 10 microns.

Fig 2. BLI-4 localizes to intracellular and extracellular compartments.

A,B,C) Transcriptional and translational reporters reveal bli-4 expression throughout pre-cuticle and cuticle assembly. Animals also express the epithelial junction marker DLG-1::RFP (mcIs46, magenta) to aid in cell identification and Z-depth assessment. Images are maximum intensity projections from confocal Z-stacks and representative of at least n = 8 animals examined per genotype per stage. A) A bli-4pro::GFP transcriptional reporter (sEx11763, green) [61] is broadly expressed in external epithelia, including in hyp7, seam cells, and in the excretory duct (d) and pore (p). B) An endogenous BLI-4::SfGFP(int) translational fusion (syb5321) marks intracellular puncta within epithelia and is faintly detectable within the extraembryonic space (EES). Single channel images are shown in inverted grayscale for clarity. Asterisk, fusion protein detected in the extraembryonic space. Arrow, intracellular puncta. CRISPR/Cas9 was used to insert SfGFP between the BLI-4 prodomain (Pro) and peptidase domain, as indicated (S2 Table). The schematic shows the short isoform BLI-4f (Genbank NP_001360008.1), but all isoforms should be tagged. ss, signal sequence. P, P domain. C) BLI-4::SfGFP(int) transiently accumulated in the foregut at the 1.5 + 4hr timepoint.

Fig 3. Generation of bli-4 null and isoform-specific alleles.

A) C. elegans BLI-4d (Genbank NP_001021543.1) protein schematic and comparison to human PCSK6 (Genbank BAA21625.1). The % identity at the amino acid level is listed between relevant domains. Like other PCSK family members, both proteins have a signal peptide (SP) and prodomain (Pro) that are removed during trafficking, followed by the peptidase domain and an associated P domain thought to assist with its folding and stability [48]. BLI-4d and PCSK6 also share a cysteine-rich domain (CRD). BLI-4d also has a transmembrane (TM) domain but it lacks the EGF-like (EGFL) domain found in PCSK6. B) bli-4 gene isoforms and mutant alleles. Colors indicate encoded protein domains, as in A. Isoforms are arranged by mutant groups. cs281 and cs283 are 1-2nt deletion/frameshift mutations in exon 2, which is shared among all bli-4 isoforms. e937 is a 3,325 bp deletion that removes intronic sequences and exons associated with isoforms a, e, g, and h [47]. cs302 and cs308 are 4–19 nt indel/frameshift mutations in the first exon unique to isoforms c and d. cs293 and cs295 are identical 5nt deletion/frameshift mutations in the first exon unique to isoform d. See S2 Table for specific allele sequences. C) Predicted sizes of major embryonically-expressed BLI-4::SfGFP(int) fusion proteins before and after removal of the Pro domain. Sizes were estimated based on the isoform sequence using https://www.bioinformatics.org/sms/prot_mw.html. See also S2 Fig for data regarding isoform expression in the embryo. D) Western blot of lysates from BLI-4::SfGFP(int) expressing embryos. Arrowheads indicate four major bands between 80 and 130 kD. Blot is representative of 3 replicates. E) Summary of complementation test results. bli-4(cs281) failed to complement both bli-4(cs302) (n = 59) and bli-4(e937) (n = 120) for the larval lethal (Lvl) and adult Blister (Bli) phenotypes, respectively, while bli-4(cs302) complemented bli-4(e937) (n = 186). Balancer hT2 [bli-4(e937) let-?(q782) qIs48] (I;III) was used as the bli-4(e937)-containing chromosome.

Fig 4. bli-4 null mutants retract and collapse after embryonic elongation.

A) In contrast to a wildtype 3-fold embryo (left), bli-4(-) embryos arrest as misshapen masses with excretory tube dilations (large arrows) and extracellular debris (small arrows). B) bli-4(-) allele cs283 failed to complement bli-4(e937) for the adult Blister (Bli) phenotype (n = 120). Balancer hT2 [bli-4(e937) let-?(q782) qIs48] (I;III) was used as the bli-4(e937)-containing chromosome. Arrowheads point to blistered cuticle. C) Phenotype quantitation and rescue of bli-4(-) mutants. Ex(bli-4+) corresponds to transgene csEx919, which contains fosmid WRM069bE05. ***P<0.0001, Fisher’s Exact Test. D) Stills from time-lapse imaging of bli-4(cs281) and rescued siblings at 12°C. bli-4(-) embryos initially elongated but then began retracting after the 3-hour timepoint (220–330 minutes, n = 5). Arrow indicates excretory tube dilation. Note that the rescued embryo has an elongated eggshell shape, likely caused by pressure from the coverslip; however, this did not interfere with development and hatching.

Fig 5. bli-4 CRD isoform mutants arrest as Dumpy larvae with cuticle defects.

A-D) bli-4(ΔCRD) isoform mutant phenotypes. Only images of bli-4(cs302) and rescued siblings are shown. Phenotype quantitation and rescue data for all alleles are shown below. The bli-4+ rescue transgene is csEx919. *P<0.01, **P<0.001, ***P<0.0001, Fisher’s Exact test. A) Most mutants arrest as Dumpy L2 or L3 larvae. Larvae in image are 48hr after egg lay (AEL). B) Embryos accumulate extracellular debris (small arrows) between the embryo and the eggshell. Embryos are 1.5 fold + 3–4 hours old. C) L1 larvae are slightly Dumpy and entirely lack cuticle alae ridges. Bracket indicates position of alae in the rescued sibling. D) Whereas wild-type L1 larvae have a permeability barrier that excludes Hoechst dye (left), many bli-4(ΔCRD) mutants had strong staining in the gut epithelium. Staining was not apparent in the epidermis or pharynx, suggesting a gut-specific barrier defect. Note that bli-4d is expressed in each tissue (S2 Fig) but likely has different substrates in each location since these tissues vary widely in their aECM composition.

Fig 6. BLI-4 promotes secretion of SQT-3 cuticle collagen.

A) Schematic diagram of the SQT-3::mNG(int) fusion, showing Gly-X-Y collagen domains (gray) and N-terminal CFCS (RTTR, orange). SQT-3 lacks a signal peptide but contains a predicted transmembrane (TM) domain (magenta). A predicted DPY-31-dependent cleavage site (blue) near the C-terminus is also shown. B) SQT-3::mNG(int) is poorly secreted and accumulates in puncta in bli-4 null mutants. Maximum intensity projections of 1.5-fold embryos expressing SQT-3::mNG. Box indicates extra-embryonic region analyzed in C. Images shown are representative of at least n = 10 images per genotype. C,D) Quantitation of SQT-3::mNG(int) signal in the extraembryonic space (EES) (C) or of puncta number (D) in WT vs. bli-4 1.5-fold embryos. ***, P<0.001, ****, P<0.0001, Mann-Whitney U test. E) Maximum intensity projections from confocal Z-stacks of embryos at the 1.5-fold + 4-hour stage. By this stage bli-4 null mutants collapsed with minimal SQT-3::mNG(int) matrix incorporation, whereas bli-4(cs302) mutants showed significant incorporation into the cuticle (green arrowhead). Images shown are representative of at least n = 10 embryos per genotype. F-H) Orthogonal views of WT (F) and bli-4(cs281) (G, H) embryos expressing SQT-3::mNG(int). Embryo in (G) also contains a membrane marker let-653pro::mCherry::PH (csIs98). Aggregates (arrows) appear to be intracellular or to span the plasma membrane. bli-4 embryo in (H) is the same as that shown in (E).

Fig 7. BLI-4 promotes secretion of DPY-17 cuticle collagen.

A) Schematic diagram of the DPY-17::mNG fusion, showing Gly-X-Y collagen domains (gray), signal peptide (SP, magenta) and N-terminal CFCS (RVRR, orange). B) DPY-17::mNG is poorly secreted and accumulates in puncta in bli-4 null mutants. Maximum intensity projections of 1.5-fold embryos expressing DPY-17::mNG. Box indicates extra-embryonic region analyzed in C. Images shown are representative of at least n = 8 images per genotype. C, D) Quantitation of DPY-17::mNG signal in the extraembryonic space (EES) (C) or of puncta number (D) in WT vs. bli-4 1.5-fold embryos. *, P = 0.012, ***, P<0.001,***P<0.0001, Mann-Whitney U test. E) Maximum intensity projections from confocal Z-stacks of embryos at the 1.5-fold + 4 hour stage. By this stage bli-4 null mutants collapsed with minimal DPY-17::mNG matrix incorporation, whereas bli-4(cs302) mutants showed significant incorporation into the cuticle (green arrowhead). Images shown are representative of at least n = 9 embryos per genotype.

Fig 8. SQT-3 and DPY-17 collagens are mutually dependent for secretion.

A) Maximum intensity projections of embryos expressing SQT-3::mNG. dpy-17(e164) mutant embryos retained most SQT-3::mNG intracellularly (n = 3). Similar results were observed in another n = 7 3-fold embryos and match those previously reported by immunostaining [45]. B) Maximum intensity projections of embryos expressing DPY-17::mNG. DPY-17::mNG was barely detectable in sqt-3(e2924ts) mutants, even at the permissive temperature of 15° C (n = 12). Similar results were obtained in another n = 15 3-fold embryos. C) Quantitation of fluorescence intensity in the extraembryonic space. **, P = 0.0044, ****, P<0.0001, Mann-Whitney U test.D) Model summarizing SQT-3 and DPY-17 relationship and possible direct and indirect effects of BLI-4 on each.

Fig 9. CFCS mutants for SQT-3 and DPY-17 form intracellular puncta.

Maximum intensity projections and puncta quantifications of 1.5-fold embryos (A-D) or L1 larvae (E,F) expressing A,B,E) SQT-3::mNG(int) vs. SQT-3(R79A,R82A)::mNG(int)) and (C,D,F) DPY-17::mNG vs. DPY-17(R61A,R63A,R64A)::mNG. Although poorly secreted, the CFCS mutant collagens did eventually incorporate into cuticle. Images shown are representative of at least n = 10 animals per genotype per stage. At the plate level, 100% of mutant L1s showed a Dpy phenotype (n>1000 per strain). ****, p<0.0001, Mann-Whitney U test.

Fig 10. SQT-3 CFCS mutant puncta form independently of DPY-31/astacin.

A-C) Maximum intensity projections of WT vs. dpy-31(e2770) mutants expressing A,B) SQT-3::mNG(int) or C) CFCS mutant SQT-3(R79A,R82A)::mNG(int)). D) Quantification of puncta number and size in embryos from C. DPY-31 promotes timely SQT-3 matrix incorporation (A) and proper cuticle structure (B), but its removal does not suppress the SQT-3 CFCS aggregation defect (C,D). Images shown are representative of at least n = 8 animals per genotype. E) Model for sequential cleavage of SQT-3 collagen by BLI-4/PCSK and DPY-31/astacin. BLI-4 promotes secretion of soluble forms of SQT-3, while DPY-31 and possibly other astacin proteases promote mature matrix assembly. Preventing BLI-4-dependent N-terminal cleavage (X) leads to intracellular retention. This occurs independently of DPY-31, but could reflect premature matrix assembly due to inappropriate modification by other assembly-promoting factors such as other astacin proteases, proline hydroxylases [86], or tyrosine cross-linking enzymes [72,73].