Enzymatic properties, pathogenic potential, and prevalence of ExfA, a novel Staphylococcus felis exfoliative toxin Authors: Xu GUO, Rokaia Fathi RAGAB, Ahmed ELFADADNY, Keita IYORI, Yoichi TOYODA, Koji NISHIFUJI Published: 2026 ABSTRACT Exfoliative toxins (ETs) produced by staphylococci primarily target and cleave desmoglein 1 (Dsg1), a desmosomal adhesion molecule expressed in stratified squamous epithelia. A new ET gene was recently discovered in Staphylococcus felis, a commensal and opportunistic pathogen responsible for skin infections in Felidae. This study focused on characterizing ExfA, a protein encoded by this novel S. felis ET gene, examining its enzymatic characteristics, pathogenic potential, and prevalence. Recombinant ExfA eliminated immunofluorescence for Dsg1, but not for Dsg3, in the feline epidermis and directly digested the baculovirus-expressed extracellular segment of feline Dsg1. Additionally, recombinant ExfA digested mouse Dsg1α and Dsg1β, and when injected into neonatal mice, caused superficial epidermal splitting. Molecular modeling and docking analyses indicated that ExfA forms a closed oxyanion hole and is predicted to interact with the substrate in a manner similar to that of serine proteases. An epidemiological survey of feline skin isolates in Japan found that 46.7% of S. felis clinical isolates contained exfa, and culture supernatants from 71.4% of exfa-positive isolates digested feline Dsg1. In summary, S. felis ExfA demonstrates enzymatic activity that can degrade feline Dsg1, is frequently found in cat-derived S. felis strains, and may play a role in colonization and invasion of feline skin. INTRODUCTION Staphylococcal exfoliative toxins (ETs) are enzymes that disrupt the epidermal structure. When injected into test mammals, these ETs cause keratinocyte dissociation in the suprabasal epidermis [31]. Research has demonstrated that ETs resemble serine proteases and specifically cleave a peptide bond immediately after glutamate between extracellular domains 3 (EC3) and 4 (EC4) of Dsg1, a crucial desmosomal molecule that maintains the mechanical strength of the superficial epidermis [14]. ETs mainly target host Dsg1: Staphylococcus aureus ETs (ETA, ETB, and ETD) cleave human Dsg1 [2, 3, 44], Staphylococcus hyicus ETs (ExhA, ExhB, ExhC, and ExhD) cleave porcine Dsg1 [10, 34], and Staphylococcus pseudintermedius ETs (ExpA, previously known as EXI, and ExpB) cleave canine Dsg1 [21, 22]. Additionally, S. aureus ETE isolated from an ewe with mastitis degrades ovine Dsg1 [19]. ETs have been identified as virulence factors in human skin conditions, such as bullous impetigo (BI) and staphylococcal scalded skin syndrome (SSSS), as well as in exudative epidermitis in pigs [31]. Furthermore, ETD-producing S. aureus strains have been found in human skin and soft tissue infections (SSTIs) other than BI or SSSS, such as cutaneous abscesses, furuncles, and finger pulp infections [44]. Similarly, S. pseudintermedius strains with expa (exi) or expb genes have been identified in canine pyoderma lesions [11, 21, 22]. Overall, these observations indicate that ETs facilitate the colonization and invasion of the host epidermis by staphylococci, thereby contributing to skin infections. Staphylococcus felis, a coagulase-negative staphylococcal species, is the most common colonizer of feline skin and can act as n opportunistic pathogen, leading to skin infections such as cutaneous abscesses [15, 17, 18, 28, 36]. Previous whole-genome sequencing has identified a novel ET gene in four S. felis strains (GenBank accession numbers CP027770.1, CP065980.1, CP014834.2, and CP094725.1) [29, 42]. We proposed that the ET of S. felis has enzymatic activity that cleaves feline Dsg1, aiding in bacterial colonization and invasion of feline skin. This study examined whether ExfA, encoded by exfa, cleaves the extracellular domains of feline Dsg1 and causes keratinocyte dissociation in the mouse epidermis, as previously demonstrated with other ETs. Bioinformatic analysis was used to predict the molecular interactions between ExfA and feline Dsg1, and epidemiological analysis was conducted to determine the prevalence of ExfA-producing S. felis strains. MATERIALS AND METHODS Bacterial strains This study included 76 S. felis strains. Among these, 75 clinical isolates were collected from cats with skin infections in 27 prefectures across Japan. These samples were sent to 1sec. Co., Ltd. (Chigasaki, Japan) to test their susceptibility to antimicrobial agents between April 2023 and March 2024 (Supplementary Table 1). Detailed clinical diagnoses and lesion types were not available. The individuals who submitted the samples provided written consent for the bacterial samples to be used for research purposes at various institutions, including external facilities associated with 1sec. Co., Ltd. The reference strain S. felis Igimi (ATCC 49168) was obtained from the American Type Culture Collection. Initially, all strains were cultivated on Mannitol Salt Agar (MSA; Shimadzu Co., Kyoto, Japan) for selective isolation, followed by culturing in Luria-Bertani (LB) broth (Kanto Chemical Co., Inc., Tokyo, Japan) under aerobic conditions for 18 hr at 37°C, with shaking at 200 rpm. Comparative analysis of the ExfA amino acid sequence and other ETs The deduced amino acid sequences of exfa (CP027770.1), S. aureus eta (M17347.1), etb (NC013334.1), etd (CP127616.1), and ete (CP025395.1); S. hyicus exha (AF515453.1), exhb (AF515454.1), exhc (AF515455.1), exhd (AF515456.1), and shetb (AB036767.1); and S. pseudintermedius exi/expa (CP031604.1), and expb (AB569087.1) were subjected to sequence alignment and phylogenetic analysis. Sequences were aligned using MAFFT (Multiple Alignment using Fast Fourier Transform; https://www.ebi.ac.uk/jdispatcher/msa/mafft). TrimAl (https://trimal.cgenomics.org) was used to remove low-information regions from the alignments [7], and the resulting trimmed alignments were concatenated into a single alignment. Finally, phylogenetic tree construction and analysis were performed using PhyloSuite (https://dongzhang0725.github.io/documentation) [46]. ModelFinder was used to determine the most suitable nucleotide substitution model [24], and IQ-TREE (https://iqtree.github.io) was used to construct a maximum likelihood (ML) phylogenetic tree [30]. Generation of a recombinant ExfA The complete exfa sequence was amplified from genomic DNA extracted from the S. felis strain Igimi and inserted into the pQE-70 plasmid vector (Qiagen N.V., Venlo, Netherlands), which included an HA-tagged gene fragment, to create the pQE70-ExfA-HA-His construct. The primers used in this study are listed in Supplementary Table 2. The recombinant plasmid was introduced into Escherichia coli (E. coli) XL-10 Gold Ultracompetent cells (Agilent Technologies, Santa Clara, CA, USA) to increase the copy number. The plasmid was then transferred into E. coli BL21 (DE3) competent cells (BioDynamics Laboratory Inc., Tokyo, Japan). The recombinant E. coli was grown at 37°C with agitation at 200 rpm until the OD600 reached 0.6. Protein expression was induced by adding 0.5 mM IPTG and incubating for 18 hr at 20°C. Recombinant ExfA (rExfA) was isolated using a TALON Spin Column (Takara Bio, Kusatsu, Japan) and stored at −80°C until needed. The presence of rExfA in the cell lysate or purified fraction was verified by immunoblotting with a rabbit anti-HA-tag polyclonal antibody (MBL Co., Ltd., Tokyo, Japan) (Supplementary Fig. 1). Feline skin samples Skin samples from a cat’s hindlimb footpads were obtained during amputation of the limb due to a cutaneous mast cell tumor in a pet cat. The cat’s owner was informed of the research objectives via a document and provided written consent. Prior to amputation, the cat received an intravenous injection of butorphanol (0.2 mg/kg, Vetorphal®, Meiji Animal Health, Kumamoto, Japan). General anesthesia was initiated with propofol (10 mg/cat; MSD Animal Health, Rahway, NJ, USA) and maintained with isoflurane (Zoetis, Parsippany, NJ, USA) for the duration of the procedure. The collected samples were embedded in Optimal Cutting Temperature (O.C.T.) compound (Sakura Finetek Japan Co., Ltd., Tokyo, Japan), rapidly frozen in liquid nitrogen, and stored at −80°C until use for immunofluorescence analysis. Additionally, the skin samples were placed in an RNA stabilizer (RNAlater, Qiagen N.V.) and stored at −80°C until total RNA extraction. The Clinical Research Committee of Tokyo University of Agriculture and Technology granted ethical approval for the experiments under approval number #00500015. Generation of baculovirus recombinant feline and mouse Dsg1 Total RNA was isolated from cat skin samples and used as a template to amplify the cDNA encoding the extracellular segment of feline Dsg1. The primers used in this study are listed in Supplementary Table 2. The resulting amplicon was digested with BglII and SalI and subsequently subcloned into pEVmod-cDsg1-EHis [32], which was digested with BglII/XhoI to excise the extracellular segment of canine Dsg1, thereby creating pEVmod-fDsg1-EHis. This recombinant plasmid was co-transfected into Spodoptera frugiperda insect cells (Thermo Fisher Scientific, Inc., Waltham, MA, USA) along with linearized baculovirus (BestBac 2.0 Δv-cath/chiA Linearized Baculovirus DNA, Expression Systems, LLC, Davis, CA, USA) to amplify the recombinant baculovirus. The recombinant baculovirus was then transfected into High FiveTM insect cells (Thermo Fisher Scientific, Inc.) to produce a baculovirus recombinant protein that included the extracellular segment of feline Dsg1, the hinge region of human IgG1, an E-tag, and a His-tag (fDsg1-EHis). Using the same method as for fDsg1-EHis, recombinant baculovirus extracellular segments of mouse Dsg1α (mDsg1α-EHis) and Dsg1β (mDsg1β-EHis) [33] were generated. The presence of fDsg1-EHis, mDsg1α-EHis, and mDsg1β-EHis in the insect culture supernatants was verified by immunoblotting with a rabbit anti-E-tag polyclonal antibody (MBL Co., Ltd.) (Supplementary Fig. 1). In vitro digestion of Dsg1 by a recombinant ExfA Supernatants from High FiveTM insect cultures containing fDsg1-EHis, mDsg1α-EHis, and mDsg1β-EHis were combined with rExfA to reach a final concentration of 10 ng/μL and incubated for 1 hr at 37°C. The digested fragments of recombinant Dsg1s generated in the presence of rExfA were detected by immunoblotting with an anti-E-tag polyclonal antibody. Additionally, insect culture supernatants with fDsg1-EHis were incubated with recombinant toxins from S. aureus ETA, ETB, and ETD; S. hyicus ExhA, ExhB, ExhC, and ExhD; and S. pseudintermedius ExpA and ExpB under identical conditions and analyzed via immunoblotting. Neonatal mouse study A neonatal mouse study was conducted to investigate whether recombinant ExfA (rExfA) could induce suprabasal acantholysis in a living organism, as observed in previously reported ETs. Sixteen neonatal littermates of Slc:ICR mice (<24 hr old) derived from two healthy dams maintained under specific pathogen-free conditions were purchased from Sankyo Labo Service Corp., Inc. (Tokyo, Japan). The dams were acclimated in a standard animal facility at 22–24°C and 20–50% humidity for over 24 hr before giving birth. Neonatal mice were randomly divided into two groups: an experimental group that received a subcutaneous injection of 10 µg purified rExfA into the dorsal skin (n=8) and a control group that received phosphate-buffered saline (PBS) alone at the same sites (n=8). The injections were administered using a 1-mL syringe and a 25-gauge needle. To prevent maternal rejection, the pups were separated from the dam and kept warm under a heat lamp to prevent hypothermia until the samples were collected. Three authors (XG, RFR, and AE) were informed of the group assignment during the study. All pups, whether injected with rExfA or PBS, were euthanized by decapitation, and the skin tissues around the injection sites were collected using scissors and forceps. Sampling was performed at 0, 30, 60, and 120 min post-injection. To ensure consistency, each mouse was assigned to either the experimental or control group at each time point, and the experiment was conducted twice. The experimental pups were examined for the presence of macroscopic blisters. Additionally, we hypothesized that suprabasal acantholysis and loss of the extracellular portion of Dsg1 would be evident in the skin of mice injected with rExfA. To test this hypothesis, we conducted histopathological and immunofluorescence analyses. For histopathological examination, a portion of each skin sample was fixed in formaldehyde, embedded in paraffin, sectioned, and stained with hematoxylin and eosin (H&E). The remaining skin samples were embedded in O.C.T. compound, snap-frozen, and stored at −80°C until immunofluorescence analysis. Animal experiments complied with the ARRIVE 2.0 guidelines (https://arriveguidelines.org/arrive-guidelines). The Animal Research Committee at the Tokyo University of Agriculture and Technology approved the mouse experiments under approval number #R7-75. Statistical analysis was not conducted because the experiments were purely qualitative. Immunofluorescence Feline skin cryosections were treated with either 10 ng/μL rExfA in Tris-buffered saline with 1 mM CaCl2 (TBS-Ca) or TBS-Ca alone for 2 hr at 37°C. The sections were then blocked overnight at 4°C in TBS-Ca containing 1% BSA and 5% goat serum. In contrast, murine skin cryosections were processed without blocking steps. Next, the sections were exposed to human pemphigus foliaceus (PF) serum with anti-Dsg1 IgG (diluted 1:100) [4, 20] and AK15 anti-Dsg3 mouse monoclonal antibody (diluted 1:10) [41] for 1 hr at room temperature. The sections were then incubated with Alexa FluorTM 546 goat anti-human IgG (Thermo Fisher Scientific; 1:200 dilution) and Alexa FluorTM 488 goat anti-mouse IgG (Thermo Fisher Scientific; 1:200 dilution) for 1 hr at room temperature. The sections were then mounted with Fluoro-KEEPER Antifade Reagent (Nakarai Tesque, Kyoto, Japan) and observed under a fluorescence microscope (DP73; Olympus, Tokyo, Japan). Images were captured using cellSens Standard (Olympus). No statistical analysis was conducted because the experiments were purely qualitative. Molecular docking AlphaFold 3 was employed to predict the structure of ExfA and dock it with feline Dsg1 [1], and the confidence of the complex was assessed using both pTM and iPTM metrics. A pTM score exceeding 0.5 indicates a likely valid overall fold of the complex, while the ipTM score assesses the precision of the predicted intersubunit arrangement. ipTM scores above 0.8 indicate a high-confidence prediction of the interface, whereas scores below 0.6 indicate low confidence and a significant chance of an incorrect intersubunit orientation. Models with ipTM scores ranging from 0.6 to 0.8 were cautiously regarded as provisional interface hypotheses. The docking complexes and interaction sites were visualized using the PyMOL Molecular Graphics System (version 3.1.3.1, Schrödinger, LLC, New York, NY, USA) for in-depth structural analysis. Epidemiological survey PCR-RFLP was used to identify exfa in S. felis isolates. Genomic DNA was extracted from each strain using achromopeptidase (Merck, Darmstadt, Germany) and used as a template to amplify a segment of the exfa gene. The primer pairs used in this study are listed in Supplementary Table 2. The resulting PCR products were digested with KpnI (New England Biolabs, Ipswich, MA, USA) and analyzed using agarose gel electrophoresis. If a 642 bp PCR band, which could be cleaved into two fragments of 428 and 214 bp by KpnI, was observed, the isolates were considered positive for exfa. To further assess whether the exfa-positive isolates produced enzymes capable of degrading feline Dsg1, the culture supernatants of these isolates were incubated with fDsg1-EHis for 1 hr at 37°C, followed by immunoblotting with an anti-E-tag polyclonal antibody. Ethics approval and consent to participate The collection of feline skin samples was approved by the Clinical Research Committee of the Tokyo University of Agriculture and Technology (approval number #00500015). The Biosafety Committee at the same university approved the use of E. coli containing the exfa or feline Dsg1 gene and a baculovirus carrying the feline Dsg1 gene under approval numbers #GR5-11 and #GR5-12. The Animal Research Committee at the Tokyo University of Agriculture and Technology approved the mouse experiments (approval number: #R7-75). RESULTS Sequence alignment and phylogenetic analysis of ExfA Amino acid homology searches and phylogenetic tree analyses revealed that ExfA shares a close genetic relationship with S. aureus ETD, S. hyicus SHETB, and S. pseudintermedius ExpB, and shows subsequent similarities to S. aureus ETB, S. pseudintermedius ExpA, and S. aureus ETE (Table 1Table 1.The amino acid identities of ExfA with known exfoliative toxins produced in Staphylococcus aureus, Staphylococcus hyicus and Staphylococcus pseudintermediusSpeciesToxinsIdentity (%)S. aureusETA44.6ETB68.0ETD72.7ETE67.9S. hyicusExhA38.8ExhB39.1ExhC41.0ExhD37.1SHETB70.2S. pseudintermediusExpA70.9ExpB64.5, Fig. 1Fig. 1.Phylogenetic tree comparing the amino acid sequences of ExfA with those of other exfoliative toxins (ETs). Multiple sequence alignments were performed using MAFFT. A maximum likelihood (ML) tree was inferred using PhyloSuite software.). Similar to other ETs, ExfA contains the conserved catalytic triad characteristic of serine proteases: His64, Asp113, and Ser187 (Supplementary Table 3). Table 1.The amino acid identities of ExfA with known exfoliative toxins produced in Staphylococcus aureus, Staphylococcus hyicus and Staphylococcus pseudintermediusSpeciesToxinsIdentity (%)S. aureusETA44.6ETB68.0ETD72.7ETE67.9S. hyicusExhA38.8ExhB39.1ExhC41.0ExhD37.1SHETB70.2S. pseudintermediusExpA70.9ExpB64.5 The amino acid identities of ExfA with known exfoliative toxins produced in Staphylococcus aureus, Staphylococcus hyicus and Staphylococcus pseudintermedius Phylogenetic tree comparing the amino acid sequences of ExfA with those of other exfoliative toxins (ETs). Multiple sequence alignments were performed using MAFFT. A maximum likelihood (ML) tree was inferred using PhyloSuite software. ExfA selectively digests feline Dsg1 Subsequently, we explored whether ExfA specifically targets feline Dsg1. Upon incubation of fDsg1-EHis, which has a molecular weight of approximately 80 kDa, with rExfA, it was cleaved into a smaller fragment weighing approximately 30 kDa (Fig. 2aFig. 2.In vitro digestive activity of rExfA in feline desmoglein 1 (Dsg1). (a) Baculovirus fDsg1-EHis was incubated with either rExfA or phosphate-buffered saline (PBS) and then subjected to immunoblotting with an anti-E-tag antibody. The red arrow indicates a degraded fDsg1-EHis fragment. (b) Immunofluorescence for Dsg1 and Dsg3 in feline footpad skin preincubated with either Tris-buffered saline with 1 mM CaCl2 (TBS-Ca) or rExfA. Bars indicate 100-µm. The basement membrane is indicated by yellow dots.). The molecular weight of this fragment matched that of the cleaved fragments processed by S. aureus ETA, ETB, and ETD; S. hyicus ExhC; and S. pseudintermedius ExpA (Supplementary Fig. 1d), suggesting enzymatic activity similar to that of known ETs. Furthermore, when feline footpad skin cryosections were pre-treated with rExfA, the immunofluorescence for Dsg1 was either completely eliminated or greatly reduced, whereas the fluorescence for Dsg3 remained unchanged (Fig. 2b). In vitro digestive activity of rExfA in feline desmoglein 1 (Dsg1). (a) Baculovirus fDsg1-EHis was incubated with either rExfA or phosphate-buffered saline (PBS) and then subjected to immunoblotting with an anti-E-tag antibody. The red arrow indicates a degraded fDsg1-EHis fragment. (b) Immunofluorescence for Dsg1 and Dsg3 in feline footpad skin preincubated with either Tris-buffered saline with 1 mM CaCl2 (TBS-Ca) or rExfA. Bars indicate 100-µm. The basement membrane is indicated by yellow dots. ExfA exhibits pathogenic activity to dissociate keratinocytes in mouse skin A neonatal mouse study was conducted to assess whether ExfA can induce suprabasal acantholysis in vivo, as observed with other ETs. Previous studies have identified three Dsg1 isoforms in mice: Dsg1α, Dsg1β, and Dsg1γ. Notably, Dsg1α and Dsg1β, unlike Dsg1γ, have a glutamate residue that forms a peptide bond and is cleaved by ETs [5, 33]. We verified that rExfA could degrade mDsg1α-EHis and mDsg1β-EHis in vitro (Fig. 3aFig. 3.ExfA degrades mouse Dsg1 and dissociates epidermal keratinocytes in neonatal mice. (a) In vitro digestion of mouse Dsg1α and Dsg1β by ExfA. Baculovirus recombinant mDsg1α-EHis and mDsg1β-EHis were incubated with either rExfA (+) or PBS (−) and subjected to immunoblotting with an anti-E tag antibody. The red arrows indicate degraded Dsg1 fragments. (b) rExfA induces intraepidermal splitting with suprabasal acantholysis in neonatal mice at 120 min post-injection. (c) ExfA selectively abolishes Dsg1 immunolabeling in mouse skin. Bars indicate 100-µm. The basement membrane is indicated by yellow dots.). Upon subcutaneous injection of rExfA into neonatal mice, blisters appeared at the injection site within 120 min after injection. The skin at these sites could be easily peeled off with physical pressure, demonstrating Nikolsky’s sign. Histopathological examination revealed intraepidermal separation with suprabasal acantholysis at the rExfA-injected sites, while the epidermis remained intact at the PBS-injected sites (Fig. 3b). Immunofluorescence analysis revealed that Dsg1 immunolabeling was either completely lost or significantly reduced, whereas Dsg3 labeling remained unaffected at the edges of the intraepidermal blisters (Fig. 3c). ExfA degrades mouse Dsg1 and dissociates epidermal keratinocytes in neonatal mice. (a) In vitro digestion of mouse Dsg1α and Dsg1β by ExfA. Baculovirus recombinant mDsg1α-EHis and mDsg1β-EHis were incubated with either rExfA (+) or PBS (−) and subjected to immunoblotting with an anti-E tag antibody. The red arrows indicate degraded Dsg1 fragments. (b) rExfA induces intraepidermal splitting with suprabasal acantholysis in neonatal mice at 120 min post-injection. (c) ExfA selectively abolishes Dsg1 immunolabeling in mouse skin. Bars indicate 100-µm. The basement membrane is indicated by yellow dots. Structural modeling predicts a possible interaction between ExfA and the cleavage site of feline Dsg1 AlphaFold3 modeling suggested that ExfA shares structural similarities with other established ETs (Fig. 4aFig. 4.Structural model of ExfA and putative docking with feline Dsg1 cleavage motif. (a) Structural model of ExfA. The structure was predicted using AlphaFold3 in the template-free mode (pTM=0.95). (b) Closed oxyanion hole of ExfA. A hydrogen bond between Ser187 and Pro184 seals this oxyanion hole. Blue: positive charge; red: negative charge. (c) Putative docking complex of a potential cleavage motif (NVIEGAVF) with ExfA, generated using AlphaFold3 (ipTM=0.66, pTM=0.94). Pro184 flips and disrupts the hydrogen bond with Ser187. Instead, Ser187 forms a hydrogen bond with Glu381 in the motif, potentially supporting a spatial arrangement that is compatible with substrate cleavage. Residues labeled in red (Ser187, His64, Asp113, Pro184) are from ExfA, while the residue labeled in purple (Glu381) is from feline Dsg1. and Supplementary Table 5). Furthermore, the AlphaFold3 model predicted that the catalytic residue Ser187 may form a hydrogen bond with the adjacent Pro184, potentially contributing to a closed oxyanion hole, which is crucial for modulating the proteolytic activity of ETs (Fig. 4b). Additionally, in feline Dsg1, a highly conserved VIEG motif recognized by S. aureus ETs as a potential cleavage site [14], was identified. Consequently, we performed docking analysis between ExfA and a motif (NVIEGAVF) corresponding to the proposed cleavage site in feline Dsg1. The top-scoring model produced an ipTM score of 0.66, suggesting an intermediate confidence. In this model, substrate interactions near the catalytic center are linked to a shift in the local hydrogen-bonding pattern, potentially disrupting the hydrogen bond between Ser187 and Pro184. The model also predicted a flip of the carbonyl group of Pro184 (Fig. 4c, with ExfA and Dsg1 sequence positions referenced in Supplementary Tables 3 and 4), which may align with the opening of the oxyanion hole in the active site. Additionally, the free carboxyl group of Ser187 was predicted to form a hydrogen bond with the carbonyl group of Glu381, situated between the EC3 and EC4 domains, possibly supporting the geometry required for nucleophilic attack. Collectively, these findings offer a plausible structural hypothesis for ExfA substrate recognition. Structural model of ExfA and putative docking with feline Dsg1 cleavage motif. (a) Structural model of ExfA. The structure was predicted using AlphaFold3 in the template-free mode (pTM=0.95). (b) Closed oxyanion hole of ExfA. A hydrogen bond between Ser187 and Pro184 seals this oxyanion hole. Blue: positive charge; red: negative charge. (c) Putative docking complex of a potential cleavage motif (NVIEGAVF) with ExfA, generated using AlphaFold3 (ipTM=0.66, pTM=0.94). Pro184 flips and disrupts the hydrogen bond with Ser187. Instead, Ser187 forms a hydrogen bond with Glu381 in the motif, potentially supporting a spatial arrangement that is compatible with substrate cleavage. Residues labeled in red (Ser187, His64, Asp113, Pro184) are from ExfA, while the residue labeled in purple (Glu381) is from feline Dsg1. S. felis strains harboring the exfa were highly prevalent in Japan We obtained 75 S. felis strains from feline skin in Japan and tested them for exfa. Our results showed that 46.7% (35 out of 75) of these strains were exfa-positive (Table 2Table 2.Prevalence of the exfa and degrading activity of fDsg1-EHis in culture supernatants of clinical isolates of Staphylococcus felisexfa+-(n=35)(n=40)fDsg1-EHis digestion+71.40%0%(n=25)(25/35)(0/40)-28.60%100%(n=50)(10/35)(40/40), Supplementary Table 1 and Supplementary Fig. 2). Additionally, 71.4% (25 out of 35) of the exfa-positive strains had culture supernatants that digested fDsg1-EHis (Table 2, Supplementary Fig. 3), indicating the presence of enzymes in these supernatants that can break down feline Dsg1, similar to ExfA. In contrast, none of the culture supernatants from the 40 exfa-negative strains digested fDsg1-EHis. Samples were collected from 27 prefectures across Japan, and exfa-positive strains were found in 22 prefectures (Supplementary Table 1). Table 2.Prevalence of the exfa and degrading activity of fDsg1-EHis in culture supernatants of clinical isolates of Staphylococcus felisexfa+-(n=35)(n=40)fDsg1-EHis digestion+71.40%0%(n=25)(25/35)(0/40)-28.60%100%(n=50)(10/35)(40/40) Prevalence of the exfa and degrading activity of fDsg1-EHis in culture supernatants of clinical isolates of Staphylococcus felis DISCUSSION This study demonstrated that S. felis ExfA exhibits enzymatic activity that degrades host Dsg1, similar to other ETs. ExfA specifically targets the extracellular portion of feline and mouse Dsg1, similar to other ETs [14, 31], potentially through mechanisms resembling serine proteases. Additionally, ExfA induced keratinocyte separation in the suprabasal epidermis of the mice. Dsg1 is a desmosomal cell-cell adhesion molecule predominantly found in stratified squamous epithelia. In mammalian skin, Dsg1 is present throughout the epidermis, whereas Dsg3 is confined to the basal and suprabasal layers. In contrast, within the oral mucous membrane, Dsg1 is located in the suprabasal layer, except for the basal layer, where Dsg3 is expressed [39]. It has been proposed that Dsg1 and Dsg3 may play a compensatory role if both are expressed in the same cell type [12]. Consequently, ETs lead to keratinocyte separation in the superficial epidermal layer, where Dsg3 cannot compensate for the loss of adhesive function. Although ExfA degrades feline Dsg1, its pathogenic role in feline skin conditions has not been thoroughly assessed, as there are no reports linking it to bullous or pustular skin diseases with suprabasal acantholysis, such as human and canine impetigo or porcine exudative epidermitis [10, 22, 45]. However, ETD-producing S. aureus strains have been isolated from bullous or pustular skin diseases and deep staphylococcal infections, suggesting that ETD may facilitate bacterial skin invasion [45]. It has also been noted that S. felis is often isolated from deep skin infections, such as cutaneous abscesses, in cats [23]. Furthermore, it has been suggested that ETs might extend S. aureus colonization on human skin [16]. These observations imply that ExfA could promote S. felis colonization and invasion in the host skin, contributing to the development of deep skin infections in cats. A previous study demonstrated that S. aureus ETA cleaves a single peptide bond immediately after Glu381, located between the EC3 and EC4 domains of human and mouse Dsg1 [14]. Our bioinformatic study suggested that ExfA might exhibit a closed oxyanion hole conformation. In the putative ExfA structure, a hydrogen bond between Ser187 and Pro184 may close the oxyanion hole. Previous crystallographic studies have identified the presence of an oxyanion hole in ETA (PDB ID: 1DUA) [35], ETB (PDB ID: 1DT2) [35], and ETD (PDB ID: 9J9A) [38]. In ETs, the closed conformation of the oxyanion hole is predicted to represent the inactive state of the enzyme before substrate binding [8, 35, 38]. In the hypothetical ExfA-substrate docking complex, substrate binding was predicted to be associated with the opening of the oxyanion hole and the formation of hydrogen bonds with the glutamate residue in the substrate, potentially supporting a geometry compatible with nucleophilic attack and substrate cleavage. This hypothetical mechanism aligns with the previous hypothesis that ETs require substrate activation to open oxyanion holes [8]. Additionally, we found that the molecular weight of fDsg1-EHis digested by rExfA was nearly the same as that of the baculovirus recombinant extracellular segment of human Dsg1 digested by S. aureus ETA [14]. Consequently, we hypothesized that ExfA might cleave a site on feline Dsg1 similar to the ETA cleavage site on human Dsg1. However, this proposed model still requires experimental confirmation to clarify the catalytic mechanism and pinpoint the cleavage sites. In our study, we found that 46.7% of S. felis isolates from cats with skin infections across various regions of Japan carried exfa. The prevalence of expa and expb in S. pseudintermedius isolates from dogs with pyoderma was 23.2% and 23.3%, respectively [11, 22]. Surveys of S. aureus isolates from patients with skin and soft tissue infections (SSTIs) in different countries have reported eta gene prevalence rates ranging from 3.3% to 13.3% and etb prevalence rates ranging from 0% to 6.6% [26, 27, 37, 43]. A study conducted in France documented a 21.8% prevalence of etd among SSTI isolates [45]. Although these previous studies focused on different staphylococcal species from other mammals, our findings suggest that the isolation rate of exfa from S. felis in cats with skin infections is relatively higher than the prevalence of ET genes in other staphylococcal species. However, 28.6% (10 out of 35) of the exfa-positive S. felis isolates did not digest fDsg1-EHis under the laboratory conditions employed in this study. This discrepancy between genotype and phenotype may be attributed to the condition-dependent regulation of exfa expression. Variations in the promoter, upstream regulatory region, or signal peptide can affect transcription, secretion, protein stability, and enzymatic activity [13, 40]. In staphylococci, ET expression is influenced by several regulatory factors, including sarA, sigB, sarS, and agrCA [25]. Additionally, expression of virulence-associated genes in staphylococci is often modulated by environmental factors through global regulatory networks, such as the Agr quorum-sensing system, Fur-mediated responses to iron availability, and the oxygen/redox-responsive SrrAB system [9]. Because these regulatory responses can vary among staphylococcal strains [6], some exfa-positive isolates may not express or secrete sufficient ExfA under the culture conditions used in this study, resulting in undetectable Dsg1-digesting activity. Consequently, these 10 isolates could be considered exfa-positive but lacking detectable functional activity under our in vitro testing conditions or as definitive non-producers of ExfA. Further research is warranted to elucidate the molecular mechanisms underlying exfa gene expression in S. felis. Nonetheless, our findings imply that S. felis carrying exfa is prevalent in Japan and may contribute to the colonization and invasion of feline skin, leading to skin infection in cats. This study had several limitations. Owing to the lack of detailed information about the skin lesions in the cats from which the exfa-harboring strains were isolated, it was not possible to determine the feline diseases in which ExfA might be involved. Additionally, while we demonstrated the pathogenic potential of rExfA to induce suprabasal acantholysis using a neonatal mouse model, this in vivo assay was primarily intended to confirm its ET-like activity in mammalian skin rather than to fully replicate the complex clinical features of naturally occurring feline skin diseases. Further clinical and pathological investigations in cats are required to elucidate the precise role of ExfA in feline infections. Moreover, the exact cleavage site of ExfA on feline Dsg1 was not determined experimentally in this study. Future studies are anticipated to explore the molecular mechanisms associated with ExfA activity, as well as the prevalence and genetic diversity of exf genes in S. felis. Conflicts of Interest KI and YT are the CEO and President of 1sec Co., Ltd., respectively. Supplementary Material