Page 57 - Read Online

Sasaki et al. Microbiome Res Rep 2023;2:12 https://dx.doi.org/10.20517/mrr.2023.08 Page 9 of 15

-1
4.22 mg/mL, 63.2 ± 5.2 s , and 15.5 mL/mg/s, respectively. The optimal pH and temperature for AAfase
activity were 6.5 and 35 °C-45 °C, respectively [Supplementary Figure 2].

Amino acid residues involved in the substrate specificity of AAfase
AAfase and GAfase mainly released β-L-Arap-(1→3)-L-Ara and α-D-Galp-(1→3)-L-Ara from AGPs,
respectively. To identify the amino acids that govern the differentiation of substrate specificity between
AAfase and GAfase, the amino acids involved in the catalytic reaction were selected and compared among
the bifidobacterial GH39 candidates based on the amino acid sequence of GH39 α-L-(β-1,2)-
[24]
arabinofuranobiosidase (NF2152), whose crystal structure had been previously determined
[Figure 4A, Supplementary Figure 3]. Amino acid sequences with 57%-75% identity with AAfase from the
following species were termed “Bifidobacterium GH39s” in this study: homologous genes from B. longum
(60% identity), B. adolescentis (57% identity), B. dentium (75% identity), B. asteroides (65% identity), and B.
reuteri (59% identity) [Figure 4A]. The alignment of amino acid sequences showed that critical residues,
190
315
including the putative acid/base catalytic residue (Glu in AAfase) and nucleophile (Glu in AAfase), were
116
conserved across all GH39s, except for Tyr in AAfase that only differed from Asn in GAfase. The
sequences of GAfase-type (Asn) and AAfase-type (Tyr) Bifidobacterium GH39s were divided into different
groups in the phylogenetic tree. Based on the previously reported docking model of NF2152 and α-L-(β1,2)-
arabinobiose , the residue Glu in NF2152 is a counterpart of Asn /Tyr in GAfase/AAfase and is
119
116
110
[24]
located in the vicinity of C-5 position of L-arabinofuranose at the -2 subsite. Due to the different linkage
patterns (1,2/1,3) and cyclic structures (furanose/pyranose) of substrates between NF2152 and AAfase/
GAfase, the nonreducing end of the substrate was not necessarily accommodated in the same orientation as
previously reported. The amino acid residue was hypothesized to be related to substrate recognition by
GAfase and AAfase. Based on the structural simulation model via AlphaFold2 (ColabFold), the targeted
amino acid residues, Tyr (AAfase)/Asn (GAfase), were predicted to be located on the far side of the
116
119
substrate pocket [Figure 4B]. However, other amino acid residues "Tyr (AAfase)/Tyr (GAfase) and
126
129
Trp (AAfase)/Trp (GAfase)" had different orientations in GAfase and AAfase.
124
127
To confirm whether an amino acid residue is involved in the differentiation of substrate specificity, a
mutation was introduced into GAfase. We obtained a GAfase N119Y mutant and compared the substrate
specificity between the wild-type (WT) and mutant using transglycosylated products
(β-L-Arap-(1→3)-α-L-Araf-OMe and α-D-Galp-(1→3)-α-L-Araf-OMe) as substrates [Table 2]. The β-L-Ara
p-(1→3)-α-L-Ara-releasing activity of AAfase WT was 195-fold higher than its α-D-Galp-(1→3)-α-L-Ara-
releasing activity, but that of GAfase WT was 556-fold lower. The GAfase N119Y mutant had 21.7-fold
higher activity against β-L-Arap-(1→3)-α-L-Araf-OMe than against α-D-Galp-(1→3)-α-L-Araf-OMe, i.e., it
119
converted “GAfase-type” activity into “AAfase-type” activity. Thus, Asn in GAfase and Tyr in AAfase
116
are critical for determining substrate specificity. Moreover, the activity of another type of GAfase mutant,
N119D, was measured using pre-desalting protein because the activity was lost by desalting. The substrate
preference of the GAfase N119D mutant did not shift, which demonstrated its higher activity against α-D-
Galp-(1→3)-α-L-Araf-OMe than against β-L-Arap-(1→3)-α-L-Araf-OMe, suggesting that the mutation to
tyrosine is meaningful [Supplementary Figure 4].

In vitro assimilation test of β-L-Arap-(1→3)-L-Ara and larch AGP in B. pseudocatenulatum
AAfase mainly releases β-L-Arap-(1→3)-L-Ara from gum arabic and larch AGP. The AAfase gene is flanked
by genes encoding putative β-L-arabinopyranosidase and ABC transporter, and this gene cluster was
expected to be involved in assimilating β-L-Arap-(1→3)-L-Ara as a sugar source. Therefore, we performed
the in vitro assimilation test of β-L-Arap-(1→3)-L-Ara and larch AGP using AAfase-carrier ( B.
pseudocatenulatum MCC10289, B. pseudocatenulatum MCC10285, and B. kashiwanohense MCC10250) and
noncarrier (B. pseudocatenulatum MCC10311 and B. pseudocatenulatum JCM1200 ) Bifidobacterium strains
T