Page 13 - Read Online
P. 13

Mueller et al. Microbiome Res Rep 2024;3:33  https://dx.doi.org/10.20517/mrr.2024.09  Page 7 of 18

               Akkermansia species are differentiated by their core genomes and predicted metabolic function
               To further differentiate between A. muciniphila, A. massiliensis, and A. biwaensis, we first identified gene
               clusters that were present in every genome of a single species, but not in the other species. We next
               identified gene clusters that were present in all genomes of two species, but not the remaining, and gene
               clusters present in all genomes of all three species [Figure 2A]. We find that 43% of these gene clusters
               (1,192/2,748) are present in all three species. A. biwaensis contains 664 gene clusters which are not core to
               A. muciniphila or A. massiliensis, A. massiliensis contains 153, and A. muciniphila only contains 85. That is,
               while 95% (1,524/1,609) and 91% (1,568/1,721) of the total A. muciniphila and A. massiliensis core genomes,
               respectively, are shared with the other species, only 73% (1,792/2,456) of the total A. biwaensis core genome
               is shared. This decrease in core genome size from A. biwaensis to A. massiliensis to A. muciniphila correlates
               with the relatedness of these species relative to A. glycaniphila.

               To define the predicted metabolic capabilities of the new Akkermansia species and phylogroups, we
               assigned functions to predicted ORFs based on the KEGG Orthology (KO). The resulting KOs were filtered
               to include only those annotated as present or absent in at least three genomes in each phylogroup
               [Figure 2B]. As previously reported, gene clusters involved in assimilatory sulfate reduction are absent in
               A. massiliensis and A. biwaensis , and gene clusters involved in cobalamin biosynthesis are unique to
                                          [20]
                           [22]
               A. massiliensis . Additionally, A. biwaensis is enriched with genes involved in formaldehyde assimilation
               but lacks lipoic acid biosynthetic genes. A. massiliensis and A. biwaensis are enriched with genes associated
               with chondroitin sulfate degradation, while multidrug resistance pumps present in the AmIa and AmIb
               phylogroups are missing from A. massiliensis, A. biwaensis, or AmV Akkermansia. The predicted metabolic
               capabilities of AmIII are largely the same as those of A. massiliensis, except that AmIII lacks genes for lipoic
               acid biosynthesis. Likewise, AmVI is similar to A. biwaensis; however, the AmVI strains are not enriched
               with genes related to formaldehyde assimilation.

               Phenotypic characterization of Akkermansia species
               Analysis of Akkermansia pangenomes suggests there is significant diversity among Akkermansia species in
               terms of metabolic function [20,22,66] . Additional studies corroborated the phenotypic diversity between isolates
               of these species in vitro [20,23] . Two studies further characterized a few selected isolates using biochemical test
               kits and fatty acid methyl ester (FAME) analysis [29,30] . However, these studies were restricted to a single strain
               of A. muciniphila (Muc ), A. glycaniphila (Pyt , isolated from python), and one isolate of either
                                     T
                                                           T
               A. massiliensis or A. biwaensis. To account for strain differences within species, we performed FAME
               analysis and API 20A biochemical assays on 16 Akkermansia isolates, incorporating multiple isolates from
               each species and AmI phylogroups.


               Fatty acid composition is a recommended means to phenotype microorganisms and has been used to
               differentiate closely related species, such as those of Legionella [67,68] . FAME analysis was performed on six
               representative A. muciniphila isolates (including Muc , all isolated from humans), in addition to two AmV
                                                             T
               (isolated from mice), three A. massiliensis (isolated from humans), three A. biwaensis (isolated from
               humans), and two AmVI (isolated from humans) isolates. To compare the results across phylogroup, we
               then performed a Principal Component Analysis (PCA) [Figure 2C and Supplementary Table 4]. Three
               clusters of isolates were distinguished from PC1 and PC2, which account for 61.61% of the overall variance.
               These clusters primarily consist of (i) A. muciniphila and AmV, (ii) A. massiliensis, and (iii) A. biwaensis
               and AmVI. Of the fatty acids present at an abundance greater than 1%, we found that C14:0 iso and C15:0
               iso distinguished A. massiliensis from the other species [Figure 2D and Supplementary Figure 3].
               Distinguishing between the A. muciniphila and A. biwaensis by individual fatty acids was not possible.
   8   9   10   11   12   13   14   15   16   17   18