The gene products of rhiA–rhiF represent a giant processing
line (a modular PKS–NRPS) that assembles the
carbon backbone of the macrolide (Figure 3). The backbone
chain is released by a thioesterase with cyclization to yield
the macrolide. Tailoring reactions such as epoxidation and
O-methylation furnish the final product [33,35].
It is remarkable that the rhizoxin biosynthesis genes
are most similar to orthologs identified as putative NRPS–
PKS in the genome of Pseudomonas fluorescens Pf-5 (a
plant commensal), where they are similarly organized in a
single gene locus [36]. Although this organism was not
known as a producer of rhizoxin, closer investigations
showed that P. fluorescens Pf-5 is indeed able to produce
a similar blend of rhizoxin derivatives as the fungal endosymbionts
[37]. In B. rhizoxinica the genome region
responsible for rhizoxin biosynthesis is flanked by transposase
genes, indicating that it might be a mobile genetic
element. Although in P. fluorescens Pf-5 the gene cluster is
not described as a putative foreign or mobile element, it is
remarkable that the closely related strain P. fluorescens Pf-
1 does not harbor the rhizoxin gene locus.
Without further genomic studies, the origin and evolution
of rhizoxin biosynthesis genes remains a mystery.
However, the presence of this gene cluster in both bproteobacteria
(Burkholderia) and g-proteobacteria (Pseudomonas)
represents a striking example for horizontal
transfer of a giant genomic island coding for secondary
metabolite biosynthesis. Furthermore, it should be emphasized
that the ecological role of the produced compound
appears to be surprisingly different in the two phyla.
Whereas rhizoxin from fungal endosymbionts serves as
an ‘‘offensive weapon’’ employed by a phytopathogenic