Page 44 - Read Online

Alekseeva et al. Microbiome Res Rep 2023;2:10 https://dx.doi.org/10.20517/mrr.2023.06 Page 11 of 15

originating from the Lactobacillaceae family also contained peptidases S8 - serine-type endopeptidases,
catalyzing the hydrolysis of internal alpha-peptide bonds in the polypeptide chains and containing one FN3
domain without MCRs.

Analysis of the FN3 domains of these proteins using the SAS software demonstrated that only FN3 domains
containing annotated MCRs (Group 1) displayed sequence homology with FN3 domains of human proteins
[Supplementary Table 4]. Sequences of FN3 domains that lack MCRs showed no homology with FN3
domains of eukaryotic proteins.

DISCUSSION
In our opinion, studying the FN3 domains of the FN3 protein encoded by the PFNA operon of
bifidobacteria is likely to yield important scientific results. Previously, we demonstrated that a fragment of
the FN3 proteins of B. longum subsp. longum GT15 (ΔFN3.1) is capable of selective binding of cytokine
[24]
TNFα in vitro . However, since the ΔFN3.1 protein has a heterogeneous domain structure, it is important
to determine which region accounts for binding cytokines. To answer this question, we assessed in our work
the cytokine-binding ability of two fragments of the ΔFN3.1 protein, which fundamentally differ in their
constituent domains: 2D FN3, containing two FN3 domains, and CD FN3, the C-terminal fragment of the
FN3 protein [Figure 1]. The recombinant 2D FN3 and CD FN3 proteins were isolated and purified in
sufficient quantities to study their binding to TNFα. The structure of one of the protein fragments, namely
2D FN3, was studied using circular dichroism (CD) spectroscopy together with our colleagues from the
Moscow State University (Feofanov A.V. - unpublished data). The CD results showed that the antiparallel β
-structure predominates in the structure of the studied protein, the formation of which involves 57% of
amino acids (including the β-fold). These results are consistent with the data on the organization of typical
fibronectin domains of the “β-sandwich” type. We found that neither 2D FN3 nor CD FN3 fragments were
able to bind to TNFα separately; only the entire ΔFN3.1 protein exhibited such cytokine-binding properties
[Figure 3]. This allowed us to conclude that FN3 domains alone are not enough for binding cytokines.
Obviously, the whole ΔFN3.1 protein is required for the formation of the cytokine-binding region. In our
future projects, we intend to study the binding that occurs between ΔFN3.1 protein and TNFα, as well as
other cytokines using other methods and approaches suited for characterizing protein-protein interaction.

Our hypothesis is also supported by the fact that using the trRosetta servers, we predicted that the ΔFN3.1
protein forms a pocket potentially capable of binding cytokines. The three-dimensional model of the Δ
FN3.1 protein structurally resembles the structure of the complex of a eukaryotic protein containing FN3
[34]
domains with the GM-CSF cytokine receptor (PDB code 4NKQ) . Analysis of the obtained three-
dimensional models of the ΔFN3.1, 2D FN3, and CD FN3 proteins shows that only the ΔFN3.1 protein has a
V-shaped block in which binding to TNFα can occur, which is consistent with the experimental data. Thus,
we experimentally demonstrated that the ΔFN3.1 protein binds TNFα, and by modeling and analyzing the
three-dimensional structures of the protein, we predicted its putative binding region. In further
experimental work, it is planned to obtain crystal structures of ΔFN3.1 proteins, including those forming a
complex with TNFα.

Comparative analysis of amino acid sequences encoding ΔFN3.1 proteins derived from the genomes of
sequenced strains of bifidobacteria belonging to the phylotype B. longum subsp. longum, made it possible to
distinguish between four groups of strains. Using the trRosetta software, we demonstrated that these amino
acid substitutions are capable of altering the conformation of the spatial structure of the ΔFN3.1 protein,
thereby affecting the efficiency and specificity of cytokine binding. Further experimenting and studying of
the 3D crystal complex of the ΔFN3.1 bound to TNFα should confirm or refute this hypothesis.