Page 91 - MI-2-4

Microbes & Immunity Brachyspira pilosicoli novel outer membrane proteins

identify structural homologs across five protein databases: proteins, to identify OMBB proteins (Figure 1). Ten
PDB100, CATH50, AFDB50, AFDB-SWISSPROT, and computational tools were used for predictions: Pepstats,
AFDB-Proteome. The top hit, based on the template DEG database, SignalP, LipoP, CELLO, PSORTb, OMPdb,
53
modeling score, was selected for functional annotation. MCMBB, TMBETADISC-RBF, and TMbed (Table S2).
Sequence-based functional annotation was performed Prediction outputs from all tools were combined for each
using PANNZER (http://ekhidna.biocenter.helsinki.fi/ protein. Tools that specifically predict OMBB proteins
pannzer; accessed on February 13, 2025) and eggNOG- (e.g., OMPdb, MCMBB, TMBETADISC-RBF, and TMbed)
mapper v2 (https://eggnog-mapper.embl.de/; accessed on were prioritized for OMBB protein prediction (Table S2).
February 2, 2025). 54,55 Through stringent screening criteria and manual
2.4. Amino acid sequence variation among different curation, a total of 42 OMBB proteins were selected and
strains of B. pilosicoli classified into two groups: Group A (13 proteins, predicted
as OMBB by all four tools) and Group B (29 proteins,
Predicted OMBB proteins from the reference genome predicted as OMBB by any three out of four tools) (Table 1).
95/1000 were searched for similar proteins using BLASTP
(E-value < 1×10 ; bit score > 100) across nine completed To gain structural insights into the predicted proteins,
−3
genomes of B. pilosicoli to analyze the amino acid sequence we searched for their structures in the PDB but found
variation in the predicted β-barrel proteins (Table S1). no experimentally determined models available. Using
43
Multiple sequence alignment (MSA) was performed for AlphaFold 3, structural models of the predicted proteins
orthologous sequences of each protein using ClustalW were generated, revealing typical features of OMPs, such
(Thompson JD, Gibson TJ, Higgins DG; EMBL, as β-barrel architectures with central pores, periplasmic
Heidelberg, Germany). Analysis of the MSA revealed loops, and surface-exposed loops.
56
amino acid substitutions among the orthologs. Mapping of Given that most proteins were unannotated, both
these variations onto the structural models was conducted structure- and sequence-based approaches were employed
using PyMOL. 46 to assign putative functions. Top-ranking hits were
In addition, the 16S rRNA gene sequence of B. pilosicoli considered for functional annotation. This analysis
strain P43/6/78 was retrieved from the NCBI database and revealed structural and sequence homologs of well-
used as a query for BLASTn against available B. pilosicoli characterized proteins, including OMP assembly factor
genome sequences. From the BLASTn results, strains BamA, LPS-assembly protein D (LptD), Neisserial surface
previously included in the sequence variation analysis protein NspA, OM porin F (OmpF), OM phospholipase
were identified, and their corresponding 16S rRNA gene A (OMPLA), and vitamin B transporter protein BtuB,
12
sequences were extracted. A multi-FASTA file was created, thereby providing valuable insights into their possible roles
followed by MSA using Multiple Sequence Comparison by (Tables 2 and S3).
Log Expectation (Robert C. Edgar, USA). Subsequently, a OMPs are located on the bacterial surface, serving as
phylogenetic tree was constructed using the Neighbor-Joining the primary interface between host and pathogen. Due
method in MEGA12 software (Kumar S, Tamura K; Temple to exposure to the host environment, these proteins are
University, Philadelphia, USA, and Tokyo Metropolitan subjected to strong selection pressures, making the analysis
University, Japan) to determine the evolutionary relationships of their sequence variability essential for understanding
among the selected strains (Figure S1). 57 pathogen evolution (Table 3). Building on this, we
59
2.5. Structure alignment using the US-align server analyzed the amino acid sequences of the predicted OMBB
proteins for residues exhibiting sequence variation across
Structural models were aligned using the US-align server nine B. pilosicoli strains (Table S4). Mapping these variations
(https://zhanggroup.org/US-align/; accessed on July 23, onto the structural models revealed that many variations
2024) an online web server to assess structural similarities were located on extracellular loops (ECLs), which are more
58
and variations. Structural alignments were visualized, and likely to interact with the host environment (Table 3).
figures were generated using PyMOL.
3.2. Identification of OMPs
3. Results and discussion 3.2.1. Group A
3.1. Prediction of OMBB proteins using a consensus- Group A comprised of 13 proteins, consisting of three
based computational approach proteins with 16 stranded β-barrel domain (BP951000_
A consensus-based computational framework was applied RS05730, BP951000_RS10215, and BP951000_RS04760);
to the B. pilosicoli 95/1000 proteome, consisting of 2,275 seven proteins with eight-stranded β-barrel domain

Volume 2 Issue 4 (2025) 83 doi: 10.36922/MI025230050