Relative distances between genomes (calculated from a concatenated ribosomal protein tree): A. lipoferum 4B to Azospirillum sp.510 – 0.01; Rhizobium etli to Rhizobium leguminosarum – 0.02; A. lipoferum 4B to A. brasilense Sp245 – 0.10; Rhizobium etli to S. meliloti – 0.11.

(A) Proportion of ancestral and horizontally transferred genes predicted in three Azospirillum genomes with varying confidence: intensity of color shows high (dark), medium (medium) and low (light) levels of confidence for predictions (see Materials and Methods). Genes that cannot be assigned using this protocol are shown in white. Majority of these genes are unique to each species and have no identifiable homologs; thus, they are likely the result of HGT. (B) Proportion of ancestral and horizontally transferred genes in genomes of Rhodospirillaceae. Only genes that were predicted with high confidence are shown.

Color intensity indicates high (dark), medium (medium) and low (light) confidence levels for prediction (see Materials and Methods for details).

Color intensity indicates high (dark), medium (medium) and low (light) confidence levels for prediction. See Table S9 for details.

All three Azospirillum species are shown on the left panel. Known cellulose degrader (Dickeya dadantii 3937, T+) and non-degrader (Agrobacterium tumefaciens NT1, T-) are shown as positive and negative controls, respectively.

Quantification of biofilm formed by wild type (wt) and a pili mutant (cpaB) on glass using crystal violet staining (left panel) and 3-D-reconstruction of the biofilm formed by wild type (top) and a pili mutant (bottom) by confocal microscopy (right panel).

A maximum-likelihood tree built from 16S rRNA sequences from members of Rhodospirillaceae. Acetobacter acetii, a member of the same order Rhodospirillales, but a different family, Acetobacteriaceae, is shown as an outgroup. Aquatic inhabitants are not highlighted; terrestrial are highlighted in brown and plant-associated Azospirillum is highlighted in green. See Table S1 for details.

See Materials and Methods for details.

Only genes that were predicted with high confidence are shown.

The genomes of Azospirillum encode from 26 to 34 glycoside hydrolases that belong to various CAZy [54] families (Table S12). Total number of glycoside hydrolases in Azospirillum species is similar to that in a soil cellulolytic bacterium Thermobifida fusca [61]. All three species have orthologs of putative cellulases (AZOLI_p10561, AZOLI_p40099; AZOBR_p1110164; AZL_a06890; AZL_d05040) with unique domain architecture: GH_5 – CalX-β. The other two putative cellulases (AZOBR_150049, AZOBR_p470008) are found only in A. brasilense. In addition to putative cellulases, Azospirillum species encode putative extracellular endoglucanases that may be involved in cellulose/hemicellulose degradation. For example, glycoside hydrolases that belong to family GH8 (AZOLI_p30425, AZL_c05150), which are known for a wide range of cellulose-containing substrates [62]–[64] and family GH12 (AZOBR_p440082). All three species are predicted to secrete a number of putative hemicellulases, that belong to glycoside hydrolase families GH1 (β-glycosidases), GH4 (glucuronidase/galactosidase), GH10 (endo-xylanases) and GH16 (licheninases) (Table S12). CAZy families were assigned as described in Materials and Methods.

The trees exemplify ancestral and HGT relationships, respectively, that were predicted with high confidence. Trees were built from aligned sequences of the A. brasilense Sp245 query and twenty most similar sequences determined by BLAST. The thiamine synthetase set contains only representatives of alpha-proteobacteria including Rhodospirillaceae (shown in red). The cellulase set consists of representatives of Actinobacteria, Firmicutes, and Chloroflexi with only one representative of alpha-proteobacteria other than Azospirillum (that are shown in blue, highlighting their HGT origin), Azorhizobium.