Good things come in threes: Distinct C. elegans basement membranes utilize novel collagen IV trimers

In architecture, it is said that form follows function. Yet in biology, structure and function are inseparable. Nowhere is the relationship between structure and function better exemplified than in the architecture of the extracellular matrix, where distinct collagen molecules with different structures have distinct functions. In the stromal matrix of tendons and bones, type I collagens—long and straight—come together to make collagen fibers that provide support along their linear axis, whereas in the basement membrane, type IV collagens—with kinks and bends—form a net-like structure that supports cell sheets, providing support throughout the plane of the tissue. The type IV network is assembled extracellularly from building blocks called protomers, and each protomer is generated within the cell from three distinct collagen peptides. In the article by Srinivasan et al., Sherwood and colleagues report the existence of a new type of collagen IV protomer, and there are hints that this new form may confer slightly different function to the collagen IV network (1).

Collagen IV protomers are trimers of collagen IV protein monomers, and throughout the animal kingdom these monomers come in two types, α1 and α2. Until this report from Sherwood and colleagues, it was understood that each collagen IV protomer is a heterotrimer containing two α1-like and one α2-like monomers. Humans and vertebrates generally have three α1-like and three α2-like monomers, and these six collagen IV subunits assemble to make three distinct types of human heterotrimeric protomers, each following the α1–α1–α2 pattern of protomer assembly. Yet, these different protomers form distinct networks found in different tissues and are believed to have different mechanical properties (2). In contrast, the roundworm Caenorhabditis elegans and the fruit fly Drosophila have only one α1 and one α2 (3, 4, 5). In C. elegans, the model system used by Sherwood and colleagues, the subunits are known as LET-2 and EMB-9, and it was expected they would form only a single collagen IV heterotrimer. The authors use a variety of approaches to upend that assumption, providing plentiful evidence for the existence of collagen IV homotrimers and other combinations.

For several years, the Sherwood lab has been working to generate a complete “basement membrane toolkit” of endogenous fluorescently tagged alleles for every basement membrane protein in worms. However, tagging collagen IV had been challenging, as previous tagged versions were detrimental to animal health (6, 7). To overcome this hurdle, the authors compared many fluorescently tagged let-2 and emb-9 alleles using a variety of fluorophores and protein insertion sites. They found that some insertion sites caused accumulation of the tagged protein in the ER, with reduced basement membrane fluorescence and reduced animal survival. However, tagging LET-2 and EMB-9 at the C terminus gave rise to healthy animals with brightly labeled basement membranes and little intracellular accumulation (Fig. 1). One potentially exciting sequel to this work may be that this new tagging site is generally useful for other model organisms, as previous tagging has utilized a site that works well in Drosophila but not well in C. elegans. Another opportunity afforded by this new tagging site is visualizing mutant protein behavior, specifically from disease-related alleles. As a proof of concept, the Sherwood lab generated tagged mutant C. elegans alleles that mimic Gould syndrome—a human disorder caused by mutations in collagen IV. Evidence points to Gould syndrome mutants having aberrant extracellular function, but the mechanism has been unclear (8, 9, 10). Using their new tagging site, the Sherwood lab generated a photoconvertible-tagged mutant collagen IV and found that these mutant proteins were removed more slowly from the basement membrane than control counterparts, a novel defective function that gives insight into the etiology of Gould syndrome. This experiment highlights how this insertion site can be used to study disease models.

Once the authors had generated fully functional tagged alleles of the two collagen IV genes, they used these tags to quantify levels of each protein in basement membranes. Using liquid chromatography coupled with tandem mass spectrometry on a mixed-age population of whole larvae, they found that the extracellular matrix fraction had a twofold higher level of LET-2 than EMB-9, a result that would seem to confirm the LET-2/LET-2/EMB-9 (LLE) heterotrimer as the exclusive collagen IV protomer. However, the Sherwood lab has a long interest in understanding how basement membranes differ in different organs, so they next analyzed the LET-2 and EMB-9 fluorescence levels in four different basement membranes in L4 larvae: the pharynx (feeding organ), the body wall muscle, the distal tip cell (that leads gonad migration), and the spermatheca. Surprisingly, they identified tissue-specific alterations in the L:E ratio, ranging from 1.64 in the distal tip cell to 2.24 in the spermatheca. This finding suggests two new ideas: first, there are other collagen IV protomers besides the expected LLE heterotrimer, and second, that there is a tissue-specific preference for protomers of distinct composition (Fig. 1). How is a preference determined for specific basement membranes?

Using a translational reporter system, the authors showed that the local level of translation of let-2 and emb-9 directly corresponds to the altered ratios visible in the basement membrane, suggesting that the availability of monomer subunits drives the composition of the resulting protomers. To confirm this hypothesis, the authors manipulated the available levels of emb-9 by RNAi-mediated knockdown and overexpression. Knockdown of emb-9 shifts the ratio of L:E from 1.8 to 3.5; the only way to achieve a 3.5 ratio is to generate some LLL homotrimers. Overexpression of emb-9 gave the opposite effect, shifting the L:E ratio from 1.8 to 1.3, suggesting the generation of alternative protomers, composed of either LEE or EEE. These results confirm that the composition of collagen IV trimeric protomers is determined by the available pool of monomers, and by extension, the tissue-specific differences in basement membranes reflects the differences in cellular gene expression. Interestingly, the LLE heterotrimer does appear to be favored for assembly over the other combinations, because genetically altering the availability of emb-9 generates visible aggregates of collagen IV in the ER. Further, when emb-9 is knocked down, LET-2 levels in the basement membrane are concomitantly decreased. These suggest a slower or more complex mechanism for the assembly of alternative protomers, all of which is yet to be discovered.

This strong genetic and cell biological evidence for the presence of alternative collagen IV protomers in the basement membrane—with tissue specificity!—raises many questions to address going forward. With regard to the tissue specificity, an L:E ratio >2 (implying LLL protomers) is found in the basement membrane of the spermatheca, an organ that undergoes repeated and frequent stretching during egg laying. Do the LLL protomers in the basement membrane confer mechanical properties important for this dynamic environment? If the new protomers make distinct mechanical contributions, these could come from supplying new binding sites for partner proteins that alter mechanics, or from the flexibility of the long collagen triple-helical region, or from the NC1 heads, which are variably cross-linked to each other to make a collagen IV network. A related question is, do the new collagen protomers get integrated into the standard LLE network, or do they form a segregated network potentially with different properties? Perhaps the most important question is, how widespread is the phenomenon of atypical protomers? In most animal genomes, collagen α1 and α2 genes occur as pairs that are positioned head-to-head, allowing coordinate transcriptional regulation, whereas in some protostomes, like C. elegans, the α1 and α2 genes are found on different chromosomes (11). The standard head-to-head arrangement allows for consistency of gene expression within the two subunits, whereas the different chromosomes would provide more opportunities for different expression levels. Supporting the possibility of protomer diversity in vertebrates, however, is the existence of the 3–4–5 protomer: although α3 and α4 are arranged head-to-head, and α5 and α6 are arranged head-to-head, these gene pairs are on different chromosomes. Still a 3–4–5 protomer forms and is a critical component of the glomerular basement membrane in the kidney, and the loss of any of the component monomers results in a defective glomerular basement membrane and Alport syndrome, demonstrating that collagen IV monomer composition is not always defined by coordinated head-to-head transcription in vertebrates. The study by Srinivasan et al. is an example of the critical contributions made by model organisms to the important field of basement membrane biology and highlights the connection of collagen IV structure to basement membrane function by uncovering tissue-specific alternative trimers of collagen IV and connecting pathogenic mutants of collagen IV to extracellular functional defects.