First sequencing of ancient coral skeletal proteins

Endogenous biomolecules preserved in fossil biominerals can record past molecular composition relevant to evolutionary and environmental reconstructions. Ancient DNA has been a frequent target but its preservation window is limited (typically up to several hundred thousand years in vertebrates and less than 10,000 years in invertebrates). Proteins, however, can persist for hundreds of thousands to tens of millions of years and have informed identification and phylogenetic relationships of extinct organisms. Invertebrate biominerals, including corals, generally lack embedded cells and have low organic content, but they do incorporate specialized extracellular biomineralization proteins that can become embedded within mineral crystals, potentially protecting them over geological timescales. Prior to this study, the oldest invertebrate skeletome data were from shells several thousand years old, leaving a gap for older invertebrate fossils. The present study asks whether endogenous skeletal proteins can be recovered and sequenced from Pleistocene (Stage 5E) coral skeletons, and if so, which protein types persist and what factors control their preservation. The work aims to validate mineralogical integrity of fossil specimens, assess amino acid racemization to distinguish endogenous proteins from contamination, and apply LC-MS/MS proteomics to sequence fossil proteins for comparison to modern coral skeletomes, thereby extending the temporal range of invertebrate paleoproteomics.

Prior research has established that ancient DNA provides phylogenetic information but degrades relatively quickly in geological terms. Conversely, proteins persist longer and have been recovered from Mesozoic vertebrates, enabling phylogenetic insights. In corals and other invertebrates, skeletal organic matter content is low (~0.01–4% by weight depending on taxon), which challenges biomolecule recovery. Coral biomineralization has been studied for over a century; recent advances include proteomic characterization of modern coral skeletal matrices, revelation that highly acidic proteins (CARPs/SAARPs) can precipitate aragonite from seawater, and that skeletal carbonic anhydrases are active. Some invertebrate fossils have shown preserved organic nitrogen and peptide bonds, and Triassic corals preserve amino acid signatures, suggesting long-term preservation potential. However, confirmed sequence data for invertebrate skeletal proteins older than the Holocene have been lacking, motivating the present investigation.

Specimens and screening: Five modern and three Pleistocene (Stage 5E; 125–138 ka) Caribbean coral specimens (Orbicella annularis and Montastraea cavernosa) from NHMLA collections were evaluated. Mineralogical integrity was assessed using X-ray powder diffraction (Panalytical X’Pert Pro) and quantitative phase analysis (Reference Intensity Ratio method) to determine aragonite versus calcite content. Elemental ratios (Mg/Ca, Sr/Ca, B/Ca) were measured by HR-ICP-MS (Element XR) to verify primary aragonite signatures. Sample preparation and cleaning: Slabbed fragments were chemically cleaned (30% hydrogen peroxide and 3% sodium hypochlorite), ground to 125 µm, and re-cleaned. Clean powders were handled in age-specific glove bags to prevent cross-contamination, with contamination checks by BCA assay and Stain-Free SDS-PAGE. Amino acid racemization and composition: Free and total hydrolysable amino acids (FAA/THAA) were extracted and analyzed in duplicate at the Northern Arizona University Amino Acid Geochronology Laboratory via HPLC with internal standard (L-homo-arginine). D/L ratios for Asx and Glx and relative amino acid compositions were determined. Protein extraction and visualization: Approximately 1 g of cleaned powder per sample was decalcified in 0.5 M glacial acetic acid. Acid-insoluble matrix (AIM) pellets were rinsed; acid-soluble matrix (ASM) was precipitated and rinsed. Subsets were analyzed by SDS-PAGE (4–20% TGX Stain-Free gels) to assess molecular weight distributions. Proteomics (LC-MS/MS): AIM and ASM proteins were solubilized in 2% SDS and processed via FASP or MED-FASP. Primary digestion used trypsin; residual material on filters underwent secondary digestion with GluC. Peptides were analyzed by nanoLC coupled to a QE-Plus Orbitrap (positive ion mode, DDA; MS1 70,000 resolution at m/z 400; MS2 17,500). Mascot searches (v2.4) were performed against coral databases (Montastraea cavernosa, M. faveolata, Platygyra carnosus, Orbicella faveolata, and O. annularis predicted proteins), plus contaminants and human UniProt; additional runs against UniProt bacteria, cyanobacteria, and fungi were used to flag non-coral matches. Fixed modification: carbamidomethyl (C); variable: Met oxidation, N-terminal acetylation, Asn/Gln deamidation. Trypsin specificity with one missed cleavage; precursor tolerance 10 ppm; fragment tolerance 20 mmu; precursor charge 2+, 3+, 4+. A 1% FDR decoy strategy set significance thresholds; error-tolerant searches followed. Proteins were retained if above cutoff with at least two independent significant peptides or one peptide detected significantly multiple times. BLAST (Blast2GO) verified identities; potential human contaminants with identical peptide sequences were removed. SAARP3 phylogeny: CARP4/SAARP1, CARP5/SAARP2, and P27/acidic SOMP/SAARP3 homologs were collected, aligned (T-Coffee), trimmed (TrimAl gappyout), model-selected (ProtTest), and trees inferred by PhyML (WAG+G+I; 1000 bootstraps). Statistics: THAA compositions of fossils were compared to modern corals using Shapiro–Wilk tests and Student’s t-tests in RStudio.

Mineralogy and chemistry: Among fossils, one O. annularis specimen (Mann4) retained 93–100% aragonite with 0–7% calcite and element/Ca ratios consistent with primary aragonite; Mcav1 contained 70–85% aragonite and 15–30% calcite; Mann2 had fully recrystallized to calcite/Mg-calcite. The modern O. annularis was 100% aragonite. Protein presence and degradation: SDS-PAGE of fossil AIM indicated preserved proteins but biased toward low molecular weight peptides (<20 kDa), consistent with degradation; Mann4 showed a smear typical of acid-extracted biomineral proteins with a band near ~60 kDa. Amino acid racemization and composition: Fossil THAA D/L Asx values were 0.380 (Mann4), 0.634 (Mcav1), and 0.611 (Mann2), compared to 0.212 in modern O. annularis; THAA D/L Glx values were also elevated in fossils. Mann4 THAA Asx pool comprised 33.7% FAA, suggesting a larger polymerized fraction than previously reported Pleistocene corals. THAA compositions of fossils broadly resembled modern corals (Asx- and Glx-rich), though fossils showed decreased relative Asx and Ser (Mann4, Mcav1), decreased Glx (Mann2), and generally increased Ala and Val. Proteome sequencing: No coral proteins were identified from fully or largely recrystallized fossils (Mann2, Mcav1). In contrast, Mann4 (>90% aragonite) yielded six coral proteins meeting stringent criteria across AIM and ASM fractions: - Acidic skeletal organic matrix protein-like (SAARP3/CARP4/5 family member) detected in AIM. - Coadhesin-like (extracellular thrombospondin type-1 repeats) detected in AIM. - LanC-like protein 3 isoform X detected in AIM. - Polyamine-modulated factor 1-binding protein 1-like detected in AIM after GluC. - Two uncharacterized proteins (one similar to P16 from Stylophora pistillata; one unique to O. annularis skeletome). Peptide evidence included a SAARP3 peptide lacking BLAST hits to Homo sapiens and showing MS1-consistent deamidation of both asparagines, a known paleoproteomic degradation signature. Comparative yield: Modern O. annularis skeletal proteome (newest growth) yielded 61 proteins across fractions; only three of these overlapped with the six detected in the fossil, highlighting selective preservation. Enzymology: Additional GluC digestion after trypsin increased fossil protein detection (two added proteins), consistent with racemization of Lys/Arg diminishing trypsin efficiency.

The recovery and sequencing of six endogenous proteins from a Pleistocene coral demonstrate that specific skeletal proteins can persist embedded within primary aragonite crystals for over 100,000 years. Absence of sequenced proteins from recrystallized specimens indicates that preservation is strongly tied to retention of primary aragonite and likely to intimate protein–mineral interactions. The identification of SAARP3, a highly acidic member of the CARP/SAARP family known to precipitate calcium carbonate, supports the hypothesis that acidic proteins bind tightly to calcium in aragonite and are preferentially preserved. Detection of coadhesin and other adhesion/scaffolding-related proteins suggests persistent components of the extracellular matrix at the calcifying interface may be incorporated into and protected by the mineral. Multiple lines of evidence argue against contamination: mineralogical screening and trace-element ratios, age-appropriate racemization signatures (elevated D/L Asx and Glx), degraded peptide size distributions, deamidation in fossil peptides, and database searches excluding human and microbial assignments. Despite successful recovery, overall proteome coverage was low relative to modern samples, likely due to protein degradation and amino acid racemization impeding enzymatic digestion. The results extend the temporal range of invertebrate paleoproteomics, provide targets (acidic matrix proteins) most likely to persist, and highlight methodological considerations (specimen screening, dual-enzyme digestion) for future studies.

This study reports the first sequencing of endogenous skeletal proteins from a Pleistocene invertebrate, demonstrating that coral skeletome components, particularly highly acidic proteins and adhesion-related matrix proteins, can be preserved within primary aragonite for >100 ka. Effective recovery required careful selection of well-preserved aragonitic specimens and stringent contamination controls. The findings open avenues for reconstructing ancient biomineralization mechanisms and potentially for phylogenetic and paleoenvironmental inferences from invertebrate fossils. Future work should focus on optimizing peptide recovery from degraded proteins, including non-enzymatic cleavage strategies, increasing sample input where ethically permissible, targeting mineral-bound acidic proteins, and expanding surveys across taxa with varying organic contents to refine preservation models.

Proteins in fossil corals are extensively degraded, yielding predominantly small peptides and low overall sequence coverage. Amino acid racemization, particularly of Lys and Arg, likely reduces trypsin efficacy, even when partially mitigated with GluC. Low initial organic content of coral skeletons limits recoverable material. Recrystallization to calcite severely compromises protein preservation. The destructive nature of protein extraction constrains the amount of material available for analysis. Low peptide yields precluded deeper analyses (e.g., distinguishing isoaspartate/γ-glutamate products). Potential oxidative modifications (e.g., methionine oxidation) may limit applicability of some non-enzymatic cleavage methods (e.g., CNBr).