Choi and Kim (PNAS 2006) mapped the evolutionary ages of the common structural ancestors (CSAs) of protein sequence families onto a global map of protein structure space and concluded that new ancestral proteins were 'born' repeatedly throughout evolution (the multiple birth model), not descended from a single ancestral set. The paper explicitly left open three questions: (i) by what molecular/genomic mechanism genes encoding a new CSA arise (e.g., assembly from multiple gene fragments by recombination or phage/virus-mediated events, versus de novo emergence); (ii) whether the global protein structure space is still expanding through continued birth of new structures or has reached an equilibrium in which the destabilizing cost of larger genes/proteins offsets functional diversification; and (iii) whether the observed stage-like changes in the relative abundance of the four major fold classes reflect early-Earth environmental transitions rather than sampling or methodological artifacts. The literature documents substantial progress on each facet — fragment-sharing evidence for the assembly mechanism, near-saturation versus thousands-of-new-folds census debates, independent fold chronologies, and robustness-checked age estimates — but contains no work that establishes the genomic birth mechanism of new CSAs, that empirically distinguishes continuing expansion from equilibrium, or that causally tests the environmental interpretation of the stage-like fold-class abundance shifts. The problem is therefore open.

Background

Proteins are built from independently folding domains, and when all known domain structures are compared directly by their three-dimensional coordinates, they cluster into four elongated, sparsely populated regions of 'protein structure space' that correspond roughly to the four classical structural classes (all-alpha, all-beta, alpha+beta, alpha/beta) used in classification databases such as SCOP (Structural Classification of Proteins). The sparseness of this space suggests that mutations in protein-coding genes have been constrained to produce chains that remain stably folded within one of the 'allowed' regions — an idea called structural selection or designability. Sequence families (groups of proteins with recognizably similar amino acid sequences) typically share one common folded structure; the most ancient detectable progenitor of such a family is its common structural ancestor (CSA). Choi and Kim (2006) estimated a relative evolutionary age for the CSA of each known-structure protein family by mapping the family's member organisms onto a reference phylogenetic tree of life (built from small-subunit ribosomal RNA sequences of 345 taxonomic groups) and assuming the CSA existed in the most recent common ancestral (MRCA) organism of those members; horizontal gene transfer (gene acquisition from other species rather than vertical inheritance) was filtered by a jackknife procedure. They found that CSAs of different families have very different ages — new ancestral proteins were 'born' at many different evolutionary times rather than all descending from one primordial set — the 'multiple birth model' for protein sequence families — and that the relative ranking of the four structural classes changed in a stage-like pattern over three evolutionary stages. The paper then posed the questions that define this problem: how is a gene for a new CSA born (they proposed assembly from multiple gene fragments through recombination or phage/virus-mediated events, without evidence on the actual process); whether the protein structure space is still expanding or has reached an equilibrium, where a newly born protein would grow to larger sizes with more diversified function until the destabilizing effects of large gene/protein size outweigh further functional gains; and whether the stage-like abundance shifts mark major early-Earth environmental transitions or are artifacts of sampling and method. These questions sit at the intersection of structural bioinformatics and evolutionary genomics; answering them requires operationalizing what counts as a genuinely 'new' structural architecture (difficult because remote common descent is hard to exclude), identifying birth events over deep evolutionary time (confounded by horizontal gene transfer, uneven taxonomic sampling, and extinction), and testing the proposed tradeoff between functional diversification and the destabilizing effects of larger gene/protein size against data.

Problem Statement

Resolve the tempo and mode of the birth of new protein structural architectures, as posed by Choi and Kim (2006): (1) establish the molecular/genomic mechanism(s) — for example assembly from multiple gene fragments by recombination, phage- or virus-mediated events, or de novo emergence — by which genes encoding a new common structural ancestor (CSA) of a protein sequence family arise; (2) determine empirically whether the global protein structure space is still expanding through continued birth of new structural architectures or has reached an equilibrium in which functional diversification is offset by the destabilizing and size-related costs of larger genes and proteins; (3) decide whether the stage-like changes in the relative abundance of the four major structural classes over evolutionary time reflect early-Earth environmental transitions or instead sampling biases or methodological artifacts, in view of the limitations of the age-mapping procedure that the authors acknowledged — a single small-subunit rRNA reference tree of 345 taxa, jackknife filtering of horizontal gene transfer, and sampling restricted to globular proteins of known structure. These are the three questions Choi and Kim (2006) themselves left open, and together they concern how genuinely new protein structural architectures have entered, and continue to enter, the biosphere. The three questions are jointly posed, and each is independently attackable.

References

  1. Choi I-G, Kim S-H (2006) Evolution of protein structural classes and protein sequence families. Proc Natl Acad Sci USA 103(38):14056-14061. doi:10.1073/pnas.0606239103, https://doi.org/10.1073/pnas.0606239103 (PMC1560931)
  2. Kim SH, Shin DH, Liu J, Oganesyan V, Chen S, Xu QS, Kim JS, Das D, Schulze-Gahmen U, Holbrook SR, Holbrook EL, Martinez BA, Oganesyan N, DeGiovanni A, Lou Y, Henriquez M, Huang C, Jancarik J, Pufan R, Choi IG, Chandonia JM, Hou J, Gold B, Yokota H, Brenner SE, Adams PD, Kim R (2005) Structural genomics of minimal organisms and protein fold space. J Struct Funct Genomics 6(2-3):63-70. doi:10.1007/s10969-005-2651-9, https://doi.org/10.1007/s10969-005-2651-9
  3. Taylor WR (2007) Evolutionary transitions in protein fold space. Curr Opin Struct Biol 17(3):354-361. doi:10.1016/j.sbi.2007.06.002, https://doi.org/10.1016/j.sbi.2007.06.002
  4. Caetano-Anollés G, Wang M, Caetano-Anollés D, Mittenthal JE (2009) The origin, evolution and structure of the protein world. Biochem J 417(3):621-637. doi:10.1042/BJ20082063, https://doi.org/10.1042/BJ20082063
  5. Edwards H, Abeln S, Deane CM (2013) Exploring fold space preferences of new-born and ancient protein superfamilies. PLoS Comput Biol 9(11):e1003325. doi:10.1371/journal.pcbi.1003325, https://doi.org/10.1371/journal.pcbi.1003325
  6. Kolodny R, Pereyaslavets L, Samson AO, Levitt M (2013) On the universe of protein folds. Annu Rev Biophys 42(1):559-582. doi:10.1146/annurev-biophys-083012-130432, https://doi.org/10.1146/annurev-biophys-083012-130432
  7. Nepomnyachiy S, Ben-Tal N, Kolodny R (2014) Global view of the protein universe. Proc Natl Acad Sci USA 111(32):11691-11696. doi:10.1073/pnas.1403395111, https://doi.org/10.1073/pnas.1403395111
  8. Roche DB, Brüls T (2015) An assessment of the amount of untapped fold level novelty in under-sampled areas of the tree of life. Sci Rep 5:14717. doi:10.1038/srep14717, https://doi.org/10.1038/srep14717
  9. Alva V, Söding J, Lupas AN (2015) A vocabulary of ancient peptides at the origin of folded proteins. eLife 4:e09410. doi:10.7554/eLife.09410, https://doi.org/10.7554/eLife.09410
  10. Tal G, Boca SM, Mittenthal JE, Caetano-Anollés G (2016) A Dynamic Model for the Evolution of Protein Structure. J Mol Evol 82(4-5):230-243. doi:10.1007/s00239-016-9740-1, https://doi.org/10.1007/s00239-016-9740-1
  11. Kolodny R, Nepomnyachiy S, Tawfik DS, Ben-Tal N (2021) Bridging Themes: Short Protein Segments Found in Different Architectures. Mol Biol Evol 38(6):2191-2208. doi:10.1093/molbev/msab017, https://doi.org/10.1093/molbev/msab017
  12. Tang Q-Y, Ren W, Wang J, Kaneko K (2022) The Statistical Trends of Protein Evolution: A Lesson from AlphaFold Database. Mol Biol Evol 39(10):msac197. doi:10.1093/molbev/msac197, https://doi.org/10.1093/molbev/msac197
  13. Porter LL (2023) Fluid protein fold space and its implications. Bioessays 45(9):2300057. doi:10.1002/bies.202300057, https://doi.org/10.1002/bies.202300057
  14. Harteveld Z, Van Hall-Beauvais A, Morozova I, Southern J, Goverde CA, Georgeon S, Rosset S, Defferrard M, Loukas A, Vandergheynst P, Bronstein MM, Correia BE (2024) Exploring 'dark-matter' protein folds using deep learning. Cell Syst 15(10):898-910. doi:10.1016/j.cels.2024.09.006, https://doi.org/10.1016/j.cels.2024.09.006
  15. Lau AM, Bordin N, Kandathil SM, Sillitoe I, Waman VP, Wells J, Orengo CA, Jones DT (2024) Exploring structural diversity across the protein universe with The Encyclopedia of Domains. Science 386(6721):eadq4946. doi:10.1126/science.adq4946, https://doi.org/10.1126/science.adq4946
  16. Pei J, Andreeva A, Grego T, Chuguransky S, Lázaro Pinto B, Morvelli Flores N, Paysan-Lafosse T, Schaeffer RD, Bateman A, Cong Q, Grishin NV (2025) Exploring the Terra incognita of AI-based domain classifications. Protein Sci 34(12):e70392. doi:10.1002/pro.70392, https://doi.org/10.1002/pro.70392
  17. Corlett T, Smith HB, Smith E, Goldford JE, Longo LM (2026) The history of enzyme evolution embedded in metabolism. Proc Natl Acad Sci USA 123(33):e2609531123. doi:10.1073/pnas.2609531123, https://doi.org/10.1073/pnas.2609531123