Evolution of Sequence-Diverse Disordered Regions in a Protein Family: Order within the Chaos

Abstract

Approaches for studying the evolution of globular proteins are now well established yet are unsuitable for disordered sequences. Our understanding of the evolution of proteins containing disordered regions therefore lags that of globular proteins, limiting our capacity to estimate their evolutionary history, classify paralogs, and identify potential sequence–function relationships. Here, we overcome these limitations by using new analytical approaches that project representations of sequence space to dissect the evolution of proteins with both ordered and disordered regions, and the correlated changes between these. We use the fasciclin-like arabinogalactan proteins (FLAs) as a model family, since they contain a variable number of globular fasciclin domains as well as several distinct types of disordered regions: proline (Pro)-rich arabinogalactan (AG) regions and longer Pro-depleted regions. Sequence space projections of fasciclin domains from 2019 FLAs from 78 species identified distinct clusters corresponding to different types of fasciclin domains. Clusters can be similarly identified in the seemingly random Pro-rich AG and Pro-depleted disordered regions. Sequence features of the globular and disordered regions clearly correlate with one another, implying coevolution of these distinct regions, as well as with the N-linked and O-linked glycosylation motifs. We reconstruct the overall evolutionary history of the FLAs, annotated with the changing domain architectures, glycosylation motifs, number and length of AG regions, and disordered region sequence features. Mapping these features onto the functionally characterized FLAs therefore enables their sequence–function relationships to be interrogated. These findings will inform research on the abundant disordered regions in protein families from all kingdoms of life.

Fasciclin-Like Arabinogalactan-Protein 16 (FLA16) Is Required for Stem Development in Arabidopsis

“…Group B members contain fasciclin domain types R-H, with R being one of the most basal fasciclin domain types and present in algal FLAs. Fasciclin type H first appeared in land plants (embryophytes) and is almost exclusively associated with type R (Shafee et al, 2020). In addition to the very strong co-evolution of R and H fasciclin domains, robust association of specific types of AG and non-AG regions also occurs in the Group B FLAs and this is very stable compared to other FLA members (Shafee et al, 2020).…”

“…All group A and C members are predicted to be GPI-anchored whereas all of group B and three members of group D are not predicted to be GPI-anchored. Detailed bioinformatic characterization of FLA sequences has revealed further distinguishing features within these groups, including 19 fasciclin and 10 distinct AG types (Shafee et al, 2020). A strong correlation between sequence features was found, for example, the fasciclin domain type strongly predicts N-glycosylation site position and GPI anchor presence and is also highly predictive (45%) of the type of AG glycomotifs (Shafee et al, 2020).…”

“…FAS1 proteins were initially identified in grasshopper growth cones and fasciculating axons (Bastiani et al, 1987). Since this discovery FAS1 domains have been identified in proteins across all kingdoms from a broad spectrum of taxonomic classifications, including insects (Elkins et al, 1990), humans (Horiuchi et al, 1999;Gonzalez-Gonzalez and Alonso, 2018), bacteria (Ulstrup et al, 1995;Moody and Williamson, 2013), fungi (Miyazaki et al, 2007), algae (Huber and Sumper, 1994), and plants (Johnson et al, 2003;Seifert, 2018;Shafee et al, 2020). Fasciclin domains are associated with cell adhesion functions and occur predominantly in proteins predicted to be located in the extracellular space.…”

“…Detailed bioinformatic characterization of FLA sequences has revealed further distinguishing features within these groups, including 19 fasciclin and 10 distinct AG types (Shafee et al, 2020). A strong correlation between sequence features was found, for example, the fasciclin domain type strongly predicts N-glycosylation site position and GPI anchor presence and is also highly predictive (45%) of the type of AG glycomotifs (Shafee et al, 2020). The group B FLAs in Arabidopsis (FLAs 15,16,17,18) were found to contain fasciclin types R-H which has a strong co-occurrence with specific types of AG and non-AG regions.…”

On the Potential Function of Type II Arabinogalactan O-Glycosylation in Regulating the Fate of Plant Secretory Proteins

“…However, so far, no O- glycan specific sorting receptors nor the identity of the involved cytosolic adaptor complexes are known in plants. Another possibility is that a low degree or absence of O- glycosylation might increasingly expose the intrinsic disorder in proline - rich regions ( Shafee et al., 2020 ). Because intrinsically disordered protein regions are known to undergo a variety of molecular interactions in a fuzzy or plastic way ( van der Lee et al., 2014 ), these regions might be recognized by novel cargo receptors.…”