3.5 Genes for a novel opine system in pRi1855
The pRi1855 plasmid has several regions with genes of unknown function. It shares a large region of about 65 kbp (area from 88-153 kbp on the map of figure 2) with the other Ri plasmids. This region contains genes putatively involved in sugar transport, glycerol metabolism and encodes several transcription regulators and two chemoreceptors (Moriguchi et al. 2001). Besides, the pRi1855 plasmid has several unique areas with genes that are found in none of the other types of Ri and Ti plasmids described. These include transposable elements (fig.2) and two larger areas of about 20 kbp (area from 20-40 kbp on the map adjacent to the right border of the TL-region in fig.2) and about 24 kbp (area from 64-88 kbp on the map adjacent to the agropine catabolic genes). The latter area contains mainly metabolic genes and may have been introduced into pRi1855 by transposition as it is surrounded by IS5-like insertion sequences. It may have originated from the chromosome of anotherRhizobium species, as a very similar stretch of DNA was detected in the recently sequenced chromosome of Rhizobium lusitanumstrain 629 (supplementary figure S6). The 20 kbp segment adjacent to the right border of the TL-region may be involved in the transport and catabolism of a new opine. In this area we identified all the three characteristic genes that together code for a putative flavin-containing opine dehydrogenase (supplementary figures S7-S9). Flavin-containing opine dehydrogenases such as octopine, nopaline and succinamopine dehydrogenase consist of three subunits OdhABC that are encoded bynoxABC/ooxABC -like genes arranged in tandem in the genome (Watanabe et al. 2015). The three genes in pRi1855 (F3X89_28345, F3X89_28350, F3X89_28355) encode closely related proteins in which the characteristic binding sites for the FAD and FMN co-factors and the Fe-S cluster have fully been conserved (supplementary figures S7-S9). These three genes are surrounded on both sides by genes for a transport system and a LysR-type regulator. In Ti and Ri plasmids genes encoding an opine dehydrogenase are often accompanied by genes encoding the permease required for uptake of specific opine into the bacterial cell. Besides in the vicinity often genes are present encoding enzymes involved in the catabolism of the specific products released by the opine dehydrogenase. In this area of pRi1855 genes encoding such metabolic proteins are also present including genes encoding a putative saccharopine dehydrogenase and a putative aminoadipate semialdehyde dehydrogenase, which may form part of a catabolic pathway of the amino acid lysine (de Mello Serrano et al. 2012). A gene for an AsnC/Lrp regulator is located at the end of this DNA segment. The Lrp family of transcriptional regulators is known to control amino acid metabolism in bacteria (Brinkman et al. 2003).
If these genes are involved in the catabolism of an opine, a gene for an unknown opine synthase should be present in the T-region of pRi1855. Genes for agrocinopine synthase are located at the extreme left end of the T-region in Ti and Ri plasmids, while genes for nopaline synthase, octopine synthase, and succinamopine synthase are located at the extreme right end of the T-region in Ti plasmids. In cucumopine, mikimopine, and mannopine Ri plasmids the genes for cucumopine, mikimopine, and mannopine synthesis are likewise located immediately next to the right border repeat. We find at the very right end of the TL-region of pRi1855 two related genes (fig. 3): orf15/rolD and orf16 , which share 55% identity (fig. 4, table 1). These genes are not present in the T-regions of any of the other types of Ri plasmids (fig. 3). Theorf15 has been calledrolD ; the encoded RolD protein has weak sequence homology with ornithine cyclodeaminases and indeed can convert ornithine into proline (Trovato et al. 2001). The role of rolD in hairy root formation is marginal, but the gene can influence plant development by its metabolic activity (Trovato et al. 2018). Using BLASTP with the proteins encoded by orf15 and orf16 as a query we picked up the succinamopine synthases encoded by the T-region of chrysopine pTiChry5 (Shao et al. 2018) and the agropine pTiBo542 (Oger et al. 2001) as the most related proteins. The proteins encoded by orf15 andorf16 share 44-47% identity with the two succinamopine synthases, which themselves share 93% identity (table 1). All these proteins (encoded byorf15/rolD, orf16, susL ) are evolutionary related to ornithine cyclodeaminases (encoded by ocd genes). They share, for instance, about 19-21% identity with the ornithine cyclodeaminase encoded by the nopaline Ti plasmid. That ornithine cyclodeaminase (ocd ) genes can evolve novel biochemical functions during evolution is known for some time. For instance its function has been reported to evolve into an alanine dehydrogenase in Archaeoglobus fulgidus and into a tauropine dehydrogenase in Halichondria japonica (Sharma et al. 2013; Watanabe et al. 2014). Apparently, it can evolve also in an opine (succinamopine) synthase. Therefore, it would seem possible that either or both of the ocd -like genes at the right end of pRi1855 (orf15, orf16 ) similarly have evolved a novel opine synthase function, producing an unknown opine that can be degraded by the putative opine dehydrogenase encoded in the area with genes of unknown function located adjacent to the right border of the TL-region.
Discussion
The second complete genomic sequence of a Rhizobium rhizogenesstrain and the first of a virulent strain enabled us to make a comparison with the sequence of the previously sequenced biocontrol strain K84. This revealed a high conservation of the primary chromosome, but showed large differences in the secondary megacircle, the chromid. It has been described that chromids have a plasmid-like RepABC replication system, but have a similar GC content as the primary chromosome and this is also the case in LBA9402. It has been proposed that chromids are plasmids that evolve into secondary chromosomes and over time exchange genes with the primary chromosome (Slater et al. 2009; Harrison et al. 2010). We found that the chromid of strain LBA9402 was much smaller than that of strain K84 due to the absence of a segment of 724 kbp that may have been deleted in LBA9402 or inserted in K84. Also we found that this insertion/deletion was accompanied by a large inversion of a segment of 1.8 Mbp. The presence of such complex rearrangements is in line with their plasmid descent and the genes which they carry being mostly non-essential.
Our genomic sequence includes the complete sequence of the agropine pRi1855 plasmid. Over the years sequences have already been published dealing with specific parts of the closely related agropine Ri plasmid pRiA4 and earlier this year a draft of the completed pRiA4 sequence was published (Thompson et al. 2020). When compared our pRi1855 sequence is indeed closely related, but still differs at numerous areas, both by base substitutions and small insertions/deletions.
Hairy roots formed by agropine strains contain agropine, agropinic acid, mannopinic acid and mannopine (Petit et al. 1983). The agropine Ri plasmid, however, enables host strains to degrade agropine, but not the other mannityl opines. R. rhizogenes strains such as A4 , but not 1855 or HRI, contain a second, catabolic plasmid with genes for catabolism of the other three mannityl opines (Petit et al. 1983). We have now identified the genes for agropine catabolism in pRi1855 adjacent to the right border of the TR-region. The region embraces anagcA gene for the delactonase converting agropine into mannopine,mocC for oxidizing mannopine, and mocD and mocEtogether determining the enzymes that can release the amino acid and a phosphorylated sugar from the conjugate. This pathway would allow the bacterium to degrade both agropine and mannopine. However, it is known that the bacteria carrying pRi1855 cannot degrade mannopine. This may be because mannopine cannot induce the catabolic genes or because the bacterium cannot import mannopine in the cell. Indeed the pRi1855 plasmid contains genes for an agropine permease, but not for a mannopine transport system.
Agrobacteria induce neoplasias in which opines are formed that serve as a nutritional source of the bacteria. All the different types of Ti and Ri plasmids described sofar have a gene coding for an opine synthase at the very right end of the T-region. Our sequence now shows that also the agropine Ri plasmid has one larger orf (orf16 ) at the very right end of the TL-region, which shares 55% identity with the neighboring rolD gene. This gene encodes a protein that is evolutionary related to ornithine cyclodeaminase (ocd) and which still has ornithine cyclodeaminase activity (Trovato et al. 2001). Here we discovered that both rolD and orf16 have also significant identity of 44-47% with the susL genes encoding succinamopine synthase in the agropine and chrysopine Ti plasmids. Ornithine cyclodeaminase encoding (ocd ) genes have been reported to have evolved into genes encoding new enzymatic activities such as alanine dehydrogenase activity in Archaeoglobus fulgidus and tauropine dehydrogenase activity in Halichondria japonica (Sharma et al. 2013; Watanabe et al. 2014). We hypothesize that an ocd -like gene may have evolved in the agropine Ri plasmid into a gene for a new opine synthase. Opine catabolic genes are often located close to the synthase gene, but on the other side of the right border, an arrangement seen in many different Ti and Ri plasmids. Three of the genes located here together indeed have the signature of the trios of genes that are known to encode the octopine and nopaline dehydrogenase (Watanabe et al. 2015), and thus may encode the opine dehydrogenase needed for catabolism of the novel unknown opine.
In view of its frequent application, the available sequence will facilitate the use of R. rhizogenes and especially LBA9402 in both the laboratory and for biotechnological purposes.