Pangolin scales may function to trap microorganisms

To gain a better understanding of the structure of scales, we visualized scales obtained from Malayan pangolins (Manis javanica) using Scanning Electron Microscopy (SEM). The scale surface was generally smooth and composed of flakes stacked together to form a hard, protective layer with the particular feature of ‘holes’ distributed across the entire surface area (Fig. 1d-h). The side view showed a porous and honeycomb-like structure with diverse pore sizes, approximately 10-20 µm (Fig. 1g). It is possible that the presence of the pores may have no functions important for the life of pangolins. However, viruses and bacteria typically have sizes less than 10 µm and may enter into the scales through the larger surface holes and become trapped by the complexity of the intra-scale morphology [13].

To investigate whether microorganisms can enter into scales, we examined the presence of microorganisms in three individual scales (CS1-CS3) of Malayan pangolins (Additional file A: Figure S1 and Additional file B: Tables S1-2). Our data revealed the presence of microbial DNA in pangolin scales using a 16S rRNA gene amplification and also a whole-genome shotgun metagenomics approach (Fig. 1i and Additional file A: Figure S2-3) [14,15,16,17,18]. Metagenomics sequencing exposed that the most abundant species found within the microbiome of CS1 was Macrococcus caseolyticus (18.4%), followed by Acinetobacter johnsonii (9.6%). For CS2, the most abundant species were Aeromonas hydrophila (3.2%) and Leclercia adecarboxylata (3%), whereas Macrococcus caseolyticus (16.7%) and Aeromonas caviae (10.5%) were among the known species found within the microbiome of CS3. The scales from these three different individual pangolins showed generally different taxonomic diversity of microbiomes, probably reflecting their exposure to different environments. We also assembled genome sequences of the topmost abundant species using the metagenomics data and obtained near-complete genomes (79.9–97.9%) from these samples, suggesting the presence of microbiota in pangolin scales (Additional file A: Figure S4). Overall, the presence of microbial DNA in pangolin scales supports the view that the scale can function to trap invading microorganisms, providing the first line of defense to protect pangolins.

Scale proteome and the identification of nano-sized exosomes

To examine the protein composition of scales, we analyzed the proteomes of three individual scales (CS4-CS6) of Malayan pangolins (Additional file A: Figure S5 and Additional file B: Table S1) using high-performance liquid chromatography with tandem mass spectrometry (HPLC–MS/MS) technology. Each scale was split into two portions: a proximal portion (PS: the scale area attached to the skin) and a distal portion (DS: the scale area not attached to the skin), which yielded six experimental protein datasets in total (Fig. 1c). Sixty-one prominent proteins that fit our stringent definition of detectability (present in at least five out of six experimental sets with protein identification probability of > 96%) were identified in the scale proteome (Additional file B: Table S3). Notably, scales are composed of flat keratinized cells that are produced when living cells die and are filled with important proteins. Therefore, it is expected that scales would not have many proteins. The majority of these proteins were predicted to be active in cellular components such as exosomes (42 genes), followed by cornified envelope (10 genes), keratin filament/desmosome (17 genes), ficolin-1-rich granule lumen/membrane (11 genes), azurophil granule lumen (4 genes), and blood microparticle (5 genes) (Additional file B: Table S4). To confirm the presence of vesicles, we isolated vesicles from pangolin scale by ultracentrifugation and visualized them using Transmission Electron Microscopy (TEM). TEM analysis revealed that exosomes in pangolin scales possessed rounded and cup-like membrane structures (Fig. 2a). Nano-flow cytometry analysis estimated these exosomes have average sizes of 73.2 ± 14.7 nm and with a concentration of 8.16 × 108 particles/mL (Fig. 2b). We conclude that pangolin scales are rich in nanoscale exosomes supported by evidence from the large number of exosome-related proteins and the exosome-like structure and size.

Fig. 2

Characterization of exosomes, STRING interaction network and comparative analyses of 61 prominent scale proteins. a Analysis of exosomes in pangolin scale by TEM. b The size distribution profile of exosomes was analyzed by nano-flow cytometry. c Overview of the interactome of scale proteins revealed eight prominent clusters, representing several significant functional groupings. Cluster 1 was the largest cluster, composed of 19 proteins mainly involved in immune responses and/or responses to stress. Cluster 2 was largely composed of 14 keratin proteins. Cluster 3 was mainly composed of proteins involved in keratinisation and/or immune response. Clusters 4–6 and 8 were mainly composed of immunity-related proteins and/or those functional in responses to stress. Node colours represent the biological processes proteins are involved in. Red = keratinization; Blue = immune response; Green = response to stress. d Immunity response network. Red = neutrophil degranulation; Blue = antimicrobial humoral response; Green = defense response; Yellow = peptidyl-cysteine S-nitrosylation; Purple = regulation of peptide transport. e Stress response network. Red = defense response to other organism; Blue = secretion by cell; Purple = response to external stimulus; Yellow = response to unfolded protein; Cyan = cellular response to chemical stimulus. f Keratinization network. Cyan = cornification; Yellow = cell–cell adhesion; Purple = peptide cross-linking; Red = intermediate filament organization; Green = desmosome organization; Blue = cell junction assembly. Interactions between nodes are depicted by coloured lines. Different colours represent evidence from different sources such as text mining (yellow), curated database (cyan), experimentally determined (magenta), and coexpression (black). g Overlaps between scale and human hair proteins. Two hair protein sets from Adav et al. (2018) using the urea extraction (left) and the combined methods (right) were used for comparisons. h Overlaps between scale and nail proteins. i Overlaps between scale, hair and nail proteins

Pangolin scales, primarily composed of dead cells, present an intriguing source of exosomes. We hypothesized that these extracellular exosomes might originate from adjacent living cells. To elucidate their origin, we mapped 61 identified scale proteins to ExoCarta [19], an exosome database, revealing unexpected cellular sources. Substantial protein overlap was found with mesenchymal stem cells (47.5%), platelets (42.6%), thymus (34.4%), keratinocytes (27.9%), and B cells (24.6%) (Additional file B: Table S5). This diverse profile suggests contributions from stem cells, hematopoietic lineages, and epithelial cells, with the presence in blood-related sources indicating circulatory and immune system involvement.

To gain better insights into the interactome of scale proteins, we performed a network analysis using STRING [20] (Fig. 2c). The STRING functional enrichment analysis revealed three major functions: immune response (41%), response to stress (41%), and keratinization (34.4%) (Fig. 2d-f & Additional file B: Table S6).

Immune response

Immunity-related proteins were enriched in functions related to neutrophil degranulation (84%), defense response (48%), antimicrobial humoral response, and regulation of peptide transport (32%) (Fig. 2d). Interestingly, the majority of them (e.g., Lysozyme (LYZ), Secretory Leukocyte Peptidase Inhibitor (SLPI), Annexin A2 (ANXA2), S100 Calcium Binding Protein P (S100P), and Macrophage Migration Inhibitory Factor (MIF) are involved in neutrophil degranulation, the regulated exocytosis of secretory granules containing mediators such as proteases and inflammatory proteins [21]. For instance, Lysozyme (LYZ), is a well-known cornerstone of innate immunity and a critical antimicrobial protein for host defense [22]. Lysozymes have a direct antimicrobial action and work in acellular environments [22]. Another secreted protein, SLPI, is an anti-inflammatory mediator and has antimicrobial activity [23]. Two annexins, Annexin A1 (ANXA1) and Annexin A2 (ANXA2), prominent contributors to the innate immune response and anti-inflammation, were also identified [24]. Several S100 proteins (e.g., S100 calcium-binding protein A8 (S100A8) and S100 calcium-binding protein A9 (S100A9)) that are induced after infection or inflammation and exhibit antimicrobial activity were found [25]. We also identified eight proteins (e.g., Peptidylprolyl Isomerase A (PPIA) and Glyceraldehyde-3-Phosphate Dehydrogenase (GAPDH)) involved in the regulation of peptide transport.

Response to stress

We identified twenty-five stress-related proteins enriched in responses to external stimulus (56%), cellular response to chemical stimulus (52%), secretion by cell (48%), defense response to other organisms (36%), and response to unfolded protein (24%) (Fig. 2e). Pangolin scales may respond to external stimuli including external forces and pathogens. Nine proteins (LYG2, LPO, LYZ, SLPI, S100A12, S100A8, S100A9, HIST2H2BE, and GAPDH) are involved in protecting hosts from damage caused by other organisms. For instance, Lysozyme G2 (LYG2) works as a potent antibacterial protein, playing an important role in innate immunity [22]. Interestingly, we identified six proteins (e.g., Heat Shock Protein B1 (HSPB1), Heat Shock Protein A5 (HSPA5) and Heat Shock Protein A8 (HSPA8)) enriched in the responses to unfolded proteins, which may play a role in stress. HSPB1 is known for its antioxidant properties and functions as a chaperone to maintain proteins in a folding state, thus, it is critical in stress resistance [26].

Keratinization

Twenty-one scale proteins are involved in keratinization, an important process that forms the tough scale structure (Fig. 2f). They were divided into two distinct subclusters. One subcluster was mainly keratin proteins, likely playing an important role in the development of pangolin scales with mechanical resistance, structural stability, and water repellence. The second subcluster was composed of seven non-keratin proteins with functions such as cornification, cell adhesion, peptide crosslinking, and desmosome organization.

Together, these results suggest that pangolin scales may possess antimicrobial, anti-stress or anti-inflammation proteins that could enable defense against microorganisms and stress.

To broaden our insights, we also relaxed our protein detection criteria (present in ≥ 3/6 experimental sets, > 96% identification probability), identifying 94 proteins in the pangolin scale proteome (Additional file B: Table S7). While this approach may capture low-abundance proteins missed by stringent criteria, STRING analysis of this expanded dataset yielded similar enriched biological processes to the original 61-protein set (Additional file B: Table S8). Notably, the distribution pattern of exosome-derived cell types remained consistent with our initial findings (Additional file B: Table S9).

Comparative analysis of scale, hair, and nail

Observations of pangolins and their closest relatives (Carnivora, e.g., dogs and cats) suggest that scales might have evolved from hair. To obtain better insights into the differences between scale and hair, we compared the pangolin scale protein set with the two protein sets published for human hair: Set 1 contains 175 hair proteins using the urea extraction method, whereas Set 2 contains 443 proteins identified using three different methods [27]. Of the 61 scale proteins, 31.1% of them had orthologs among human hair proteins (Set 1) (Fig. 2g). When comparing with Set 2 of proteins, this percentage increased to 59%, whereas the remaining proteins were scale-specific. These scale-specific proteins were enriched in specific functions: response to stimulus (80%), response to stress (48%), and immune response (44%), and neutrophil degranulation (36%) (Table 1). Nearly half of them were involved in immune responses, including defense response to bacteria and fungi.

Table 1 Comparative proteomic analyses. Comparisons among pangolin scale, human hair and nail proteomes. Using very stringent criteria, we removed any scale proteins that can be detected in the human hair and nail, yielding a highly confident set of 19 scale-specific proteins. These proteins were significantly enriched in immunity and stress-related processes. √=present; ×=absent. The extracellular exosome prediction was derived from the cellular component enrichment analysis performed using STRING

We next investigated differences in the protein composition between pangolin scales and human nails since both have been suggested to be homologous structures [28]. We compared our scale protein set with the set of 143 proteins identified in human nail [29]. Of the 61 scale proteins, nearly half were scale-specific and enriched in functions related to response to stimulus, response to stress, immune response, and neutrophil degranulation (Fig. 2h and Table 1).

By using more stringent criteria, we discarded any scale protein that can be found in human hair and nail structures, yielding 19 proteins (31.1%) that were scale-specific, which were enriched in response to stimulus (78.9%), followed by response to stress (52.6%), immune response (36.8%) and neutrophil degranulation (31.6%) (Fig. 2i and Table 1). For instance, Lactoperoxidase (LPO), a natural effective antimicrobial enzyme, contributes to host defense against infection [30] and acts synergistically with lysozyme in its antimicrobial capacity [31]. Cystatin A (CSTA) has antimicrobial activity against various bacteria and viruses, and functions in immune modulation [32]. It inhibits the growth of bacteria with its apparent bactericidal activity [33]. Hemoglobin Subunit Delta (HBD) is a component of extracellular hemoglobin that can bind to cell-derived danger-associated molecular pattern (DAMPs) agents, such as heat-shock protein and S100A8, or pathogen-associated molecular patterns (PAMPs) to alert the host innate immunity [34]. S100A8/A9 form a heterodimer called calprotectin that inhibits bacteria by sequestering transition metals [35]. Both S100A9 and S100A12 have antimicrobial activity [36]. S100A9 also acts as an effective inhibitor of replication of coronavirus [37]. The presence of several heat shock proteins in scales suggests that these chaperones may help pangolins to cope with stress-induced protein denaturation [38]. Our analyses reveal the pathogen defense-related proteins in the scale-specific protein set, indicating a specific role for the pangolin scale in protection from diseases compared to human nail and hair structures.

Analysis of the 19 scale-specific proteins revealed that 11 (58%) were predicted to localize in exosomes (Table 1). Extending this analysis to the larger 94 scale protein-set identified 37 scale-specific proteins, of which 22 (59%) were predicted as exosome-associated (Additional file A: Figure S6). This consistent proportion across datasets suggests a potential role for exosome-derived proteins in pangolin scales. We hypothesize that these proteins may contribute to the unique functional properties of pangolin scales, distinguishing them from other keratinized structures in humans. This finding opens new avenues for understanding the molecular basis of pangolin scale formation and function.

Identification and screening of antibacterial metabolites

We next analyzed the scale metabolome using high-throughput mass spectrometry (MS) technology. We identified 78 prominent metabolites (Additional file B: Table S10). To verify the antibacterial capability of pangolin scales, we performed a systematic pre-screening of the anti-bacterial effects of 33 metabolites identified in pangolin scale against two typical bacteria that are proved commonly present in pangolins or their environment [39], Escherichia coli (Gram-negative) and Staphylococcus aureus (Gram-positive). We identified seven metabolites (malic acid, succinic acid, hippuric acid, fumaric acid, L-valine, citric acid, and glycine) that exhibited significant antibacterial activity (e.g., bacterial growth rates < 80% for both types of bacteria; see Additional file B: Table S11). From these, malic acid exhibited the best inhibitory effects with a Minimum Inhibitory Concentration (MIC) of 0.25C (“C” was defined as the dose of metabolites relative to their proportions in the pangolin scales) against both types of bacteria.

We next tested the antibacterial effects of a combination of malic acid and six other effective metabolites. We found that the two-metabolite combinations of malic acid can reduce MIC to 0.2C. Interestingly, the four-metabolite (malic acid, citric acid, glycine, and hippuric acid) and most higher metabolite combinations demonstrated the best inhibitory effects with a MIC of 0.1C for both S. aureus and E. coli (Fig. 3a-b and Additional file B: Table S12). Further examinations also demonstrated that the four-metabolite combination can inhibit the growth of Pseudomonas aeruginosa and Serratia marcescens (Fig. 3c-d). The antibacterial efficacy of the four metabolites was also investigated against all four bacterial species at various times of incubation. A substantial reduction was observed when the duration of the bacterial exposure to the four metabolites occurred across 24 h (Fig. 3e-f & g-h).

Fig. 3

Anti-bacterial activities of the 4-metabolite combination (malic acid, fumaric acid, glycine, and hippuric acid). a-d MIC assays of different concentration gradients of the 4-metabolite combination against E. coli, S. aureus, P. aeruginosa, and S. marcescens treated for 24 h, respectively. e–h Growth curves of E. coli, S. aureus, P. aeruginosa, and S. marcescens across different time points after treatment, respectively. (i) SEM images of E. coli and S. aureus untreated (EC = E. coli control; SC = S. aureus control) and treated (ET = E. coli treatment; SC = S. aureus treatment) for 24 h. All cells were examined using high-resolution Scanning Electron Microscopy (SEM) with magnifications at 10,000X (left panel), 30,000X (middle panel), and 60,000X (right panel), respectively. All experiments were performed using three biological replicates per group

To understand the underlying mechanism of these metabolites, we selected the minimal combination of metabolites with the best inhibitory effect for Minimum Bactericidal Concentration (MBC) and based on their effects on bacterial cell morphology. MBC assay proved that the four-metabolite combination has bactericidal activity (MBC =  ~ 0.4C) against E. coli and S. aureus (Additional file B: Table S13). Furthermore, SEM analyses showed there were apparent alterations in the morphology of bacterial cells after the four-metabolite treatment for 24 h. Compared with untreated bacteria, the outer membrane integrity of E. coli cells was damaged leaving large holes, whereas the surface of S. aureus cells was rough and shrunken (in some cases, the hole was formed on cell surfaces) after treatment (Fig. 3i). Altogether, our results showed that the metabolites identified in scales may kill both bacterial types by causing damage to the integrity of bacterial cells.

Metabolic and proteomic differences in pangolin scale and skin surface

We hypothesized that some active compounds may diffuse out from the scale pores and spread to the surface of pangolin skin, providing protection to the skin. To test this, we examined whether the skin surface of pangolins has metabolites similar to those found in scales. MS analysis identified 70 prominent metabolites that were collected on the skin surface by the swabbing method (Additional file B: Table S14). Of these metabolites, nearly all (98.6%) were found in scales, indicating that the two have near-identical sets of metabolites (Additional file A: Figure S7a). Metabolites were from classes such as amino acids, nucleotides, and peptides, and were enriched in pathways that are important for immunity or pathogen defense, including phenylalanine, tyrosine and tryptophan biosynthesis, arginine biosynthesis, tricarboxylic acid (TCA) cycle and purine metabolism [40,41,42] (Additional file A: Figure S7b). In addition, differential metabolomics analysis revealed twelve metabolites in scales that were more abundant than the metabolites on the skin surface including allantoin (80-fold increase in scales), which exhibits wound and anti-inflammatory properties, and osmolytes such as taurine (14-fold) and betaine (9.3-fold) which have been used to treat infections, inflammation or immune dysfunction in the clinical practices [43] (Additional file A: Figure S8 and Additional file B: Table S15).

To investigate whether scale and skin surfaces have similar protein profiles, we collected specimens and identified 14 prominent proteins on the skin surface using mass spectrometric (MS) technology (Additional file B: Table S16). Of these proteins, 71.4% were common on both scale and skin surfaces (Additional file A: Figure S7c). Constituent proteins were mainly enriched in cornification, defense response, inflammatory response, and response to stress. For instance, Lysozyme (LYZ), S100 calcium-binding protein A9 (A100A9), S100 calcium-binding protein A12 (S100A12), Keratin 1 (KRT1) and Actin Gamma 1 (ACTG1) were enriched in defense and stress responses against a foreign body or injury. Moreover, A100A9, S100A12 and LYZ formed a subcluster enriched in inflammatory response to infection or injury. A100A9 is also known to contribute to wound healing, psoriasis, skin inflammation and other skin diseases field [44]. Expanding our analysis to the larger 94-protein scale set only marginally increased this overlap, resulting in 11 shared proteins (Additional file A: Figure S6). Our analyses suggest that these common proteins may help pangolins to block pathogens from invading through the skin and cope with skin stress and injury.