A eukaryotic nicotinate-inducible gene cluster: convergent evolution in fungi and bacteria

Nicotinate degradation has hitherto been elucidated only in bacteria. In the ascomycete Aspergillus nidulans, six loci, hxnS/AN9178 encoding the molybdenum cofactor-containing nicotinate hydroxylase, AN11197 encoding a Cys2/His2 zinc finger regulator HxnR, together with AN11196/hxnZ, AN11188/hxnY, AN11189/hxnP and AN9177/hxnT, are clustered and stringently co-induced by a nicotinate derivative and subject to nitrogen metabolite repression mediated by the GATA factor AreA. These genes are strictly co-regulated by HxnR. Within the hxnR gene, constitutive mutations map in two discrete regions. Aspergillus nidulans is capable of using nicotinate and its oxidation products 6-hydroxynicotinic acid and 2,5-dihydroxypyridine as sole nitrogen sources in an HxnR-dependent way. HxnS is highly similar to HxA, the canonical xanthine dehydrogenase (XDH), and has originated by gene duplication, preceding the origin of the Pezizomycotina. This cluster is conserved with some variations throughout the Aspergillaceae. Our results imply that a fungal pathway has arisen independently from bacterial ones. Significantly, the neo-functionalization of XDH into nicotinate hydroxylase has occurred independently from analogous events in bacteria. This work describes for the first time a gene cluster involved in nicotinate catabolism in a eukaryote and has relevance for the formation and evolution of co-regulated primary metabolic gene clusters and the microbial degradation of N-heterocyclic compounds.

Thr 1010 (HxA1041, HxnS1066) are shown in an orientation such as to clearly visualise the different positioning of Thr residue in relation to oxypurinol in HxA and HxnS. We note in panel A, but more clearly in panel D the changed orientation of Phe1040 of HxA while Phe1064 of HxnS has exactly the same orientation as Phe1009 of the B. taurus enzyme. If this is a genuine difference it cannot affect the specificity or functionality of the enzyme, which is a completely orthodox XDH, with some differences with the B. taurus XDH in the kinetics of allopurinol and oxypurinol inhibition [3,4]. HxA and HxnS were modelled with I-Tasser (see Materials and Methods), the most probable models were chosen in each case. For HxA Cscore 0.63, Estimated TM-score 0.80±0.09, Estimated RMSD 8.1±4 Å. For HxnS C-score 0.83, Estimated TM-score 0.83±0.08, Estimated RMSD 7.7±4.4 Å. The two modelled proteins and the monomer of the 3BDJ structure were super-imposed with the multi-seq facility included in VMD, which was used to visualise the active site residues (see Materials and Methods).

Comments on the phylogeny of XDH-like enzymes in the fungi.
The putative XDH from Galactomyces (Geotrichum) candidum (http://www.ncbi.nlm.nih.gov/bioproject/247755) clusters, as expected, with other Saccharomycotina. However there is another sequence at the NCBI database of a genome from a different strain of an identically named organism (http://www.ncbi.nlm.nih.gov/bioproject/243259). This strain shows an HxA orthologue clustering with the HxA orthologues of the Leotiomycetes, and an HxnS orthologue in a mixed clade, clustered (alTR 1.00) with the Leotiomycetes Oidiodendron maius, Hymenoscyphus repandus and Glarea lozoyensis (shown in grey). There seem to be no question that these sequences correspond to two different organisms, the former a genuine member of the Saccharomycotina, the latter of the Leotiomycetes. A few other Purine hydroxylase proteins have an unexpected position: three members of the Eurotiales (in green lettering) which include the only Penicillia to have an HxnS orthologue, cluster with the Hypocreales. The proteins of the basal members of the Taphrinomycotina, Saitoella complicata and species of genus Taphrina do not cluster together as would be expected. The Basidiomycota are separated into two discrete clades, one (comprising Ustilago maydis), which appear as an out-group of all the ascomycete sequences and a second, which clusters with the putative orthologues of HxnS. An interesting positioning is that of the Oidiodendron maius (Leotiomycetes, Helotiales) putative XDH. The cognate protein maps within the Pezizomycotina HxnS-like clade. It is one of the two species among all the sequenced Pezizomycotina to have a putative orthologue of HxnS, in the (apparent) absence of an orthologue of HxA. It shows both characteristic sequence insertions in the 2Fe/2S cluster and between the 2Fe/2S domain and the FAD/NAD binding domain. It carries a hydrophobic residue (Val533) where we have Tyr454 and Ile478 in HxA and HxnS, respectively. It has a His (His1124) where HxA has Phe1044 and HxnS has His1069. The ORF is interrupted by four introns, of which the first two are widely conserved among HxnS orthologues, the fourth is conserved in several other HxnS orthologues of the Helotiales, while the third is -within the limits of the genomes available -unique of O. maius. There is no question this protein is phylogenetically related to HxnS rather than to HxA. However, uniquely among all putative HxnS structural orthologues, it does not have the sequence FATALH (HxnS: 1064-1069) in its substrate binding site nor does it have FTALF, near universal among HxA Pezizomycotina orthologues (HxA: 1040-1044) and all biochemically characterised XDHs (with one exception, see below); instead it specifies FGALH, (1120-1124). This sequence change is identical to one occurring in the putative XDH of the four Taphrina species, while the characterised XDH of Blastobotrys (Arxula) adeninivorans [5] has FGATF. Indeed, a Gly residue replaces the Thr in all putative Saccharomycotina HxAs.
While no biochemical work is extant in O. maius, we propose that this enzyme is not a Nicotinate hydroxylase but rather a XDH. O. maius has all the enzymes of purine breakdown (summarised in [6]), including an orthologue of UaY, the pathway specific regulatory gene characterised in A. nidulans [6,7] and N. crassa [8]. On the contrary, it has none of the nicotinate specific clustered genes to be described below. There is no other hxA/hxnS paralogue present in the O. maius genome. A second species has in its genome an HxnS orthologue in the absence of an HxA orthologue. This is Rhytidhysteron rufulum (Dothideomycetes, Hysteriales). At variance with the situation in O. maius, the R. rufulum gene encodes a typical HxnS enzyme, included in a conserved hxn gene cluster.
Other independent, probably genus or even species-specific duplications of XDH-like enzymes occurred in non-dikarya (Mortierella alpina, Conidiobolus incongruus, Basidiobolus meristosporus, Gonapodya prolifera). None of these paralogues carry the diagnostic HxnS sequence FATAL(H). No biochemical work is extant in these fungal species.

Comments on the exon-intron structure of hxnS orthologues.
The intron-exon organization is broadly class specific. Eurotiales and Onygenales (Eurotiomycetes) share multiple intron positions with Helotiales (Leotiomycetes) and Xylonomycetaceae (Xylonomycetes), including those of all three A. nidulans introns. Most Eurotiales (see below for outstanding exceptions in the Penicillum genus) have four conserved introns (in A. nidulans the second intron is absent) but only the most 5' intron is present in species of the early divergent class of Pezizomycetes, whose hxnS genes usually have five introns. Potentially, this is the only intron that survived the re-functionalisation as it is present in the hxA orthologue genes of three of the dozen species of Basidiomycota (Heterobasidion annosum, Microbotryum violaceum, Septobasidium sp. strain PNB30-8B) that occur at the basis of the HxnS branch in the PHI/PHII phylogeny ( figure 4 and figure S3). Nevertheless, none of the four hxnS introns conserved in the Eurotiales are present in Pleosporales or Botryosphaeriales (usually two introns, one conserved across these orders of Dothideomycetes), Hypocreales and Glomerellales (both Sordariomycetes, 8 and 9 introns, respectively) and Symbiotaphrina (recently assigned to Xylonomycetes, 3 introns, none corresponding to the 5 occupied intron positions found in Xylonomycetaceae).
Intriguing is the exon-intron structure of the hxnS gene in some early divergent species in Penicillium, the sister genus of Aspergillus in the family of the Aspergillaceae. The genome sequences of P. paxilli, P. citrinum and the misnamed species Hymenoscyphus varicosporoides all specify 8 introns at exactly the same positions as those in hxnS genes of the Nectriaceae family, and share none with Aspergillus hxnS (the other sequenced Penicillium species have no hxnS orthologue). This exon-intron structure is completely coherent with the phylogeny of fungal Purine Hydroxylase paralogues (figure S4) that shows that the HxnS orthologues in these Penicillium are directly related to those of Hypocreales (Sordariomycetes) rather than to HxnS proteins from other Eurotiales, while the situation for the HxA orthologues is taxonomically completely orthodox: all Penicillium species (including P. citrinum, P. paxilli) clustered with Eurotiales and none with Sordariomycetes. These circumstantial evidence strongly suggests that these exceptional species of Penicillium have (re-)acquired an hxnS gene from a species of Hypocreales by horizontal gene transfer. Loss-offunction mutation changes indicated with red triangles, constitutive mutations indicated with green triangles. The 123 orthologue hxnR loci andwhen extant the accession numbers of the corresponding HxnR orthologues are listed in Table S3.