Orthologous PDR transporters with typical NBF-TMS-NBF-TMS topology are found in all higher fungi such as
S. cerevisiae [
6,
13], the hemiascomycetous human pathogen
C. albicans [
35],
Euascomycetes [
36],
Archiascomycetes [
37,
38],
Basidiomycetes [
39] and all plants [
40] including
Arabidopsis thaliana [
41] and
Nicotiana plumbaginifolia [
42]. In contrast, the PDR topology has not been detected in protozoa [
43] or chordates [
44] where multiple drug resistance results from the overexpression of MDR/MRP efflux transporters of opposite TMS-NBF-TMS-NBF topology.
Our analysis of 56 full-sized Pdrp from 9 different Hemiascomycetes shows that the Génolevures subfamily GL3C0025 [
29] is subdivided into five phylogenetic clusters labeled A to E (Figure ). Each of these clusters can be characterized by its SACE members originally identified by D
ECOTTIGNIES and G
OFFEAU[
13]. Within the phylogenetic
Saccharomyces complex defined by K
URTZMAN[
28], a total of 40 Pdrp homologs sharing significant chromosomal neighborhood were classified in seven SONS, labeled "a" to "g" (Figure ), in which phylogenetic and neighborhood data are combined.
It now appears that the Pdrp sensu stricto phenotype can be allocated only to cluster A (Pdr12p), to cluster B (Snq2p/YNR070wp) and to cluster C (Pdr5p/10p/15p), given that all three share not only the NBF-TMS-NBF-TMS topology but also the two additional Pdrp traits: efflux drug pumping and typical Walker A1 and ABC Signature 1 motifs.
On the other hand, no experimental information is available concerning the unknown substrates of the members of cluster D (YOL075cp), while the SACE members of cluster E (Aus1p/Pdr11p) are clearly sterol influx pumps [
25]. Moreover, as neither cluster D nor cluster E contains members with the typical Pdrp motifs, they may be considered to be Pdrp
sensu lato.
Our combined phylogenetic and neighborhood analyses support the evolutionary pattern illustrated in Figure for the Pdrp
sensu stricto of the
Saccharomyces complex [
28].
Pdr12p is a Pdr
sensu stricto, according to our three criteria. Its origin remains questionable as no Pdr12p neighbors are shared either with Pdr5p/15p or Snq2p. In SACE, the function of Pdr12p, which effluxes food preservatives such as short chain weak acids [
20], is different from that of Snq2p and Pdr5p which share the function to efflux a series of antifungal azoles and other hydrophobic substrates [
17]. Nevertheless, the SACE Pdr12p sequence is closer to SACE Snq2p (46%) than to SACE Pdr5p (38%). This may suggest a common origin of both Pdr12p and Snq2p clusters. The sequence analysis of new yeast species, phylogenetically situated between DEHA and ERGO, should make it possible to disentangle the exact evolution lineage.
The unlinked ancestor of Pdr5p/15p ohnologs of KLTH jumped and triplicated on ZYRO chromosome D. An ancestor of the three copies, already in this neighborhood, was duplicated through WGD and produced Pdr5p and Pdr15p in all Saccharomycetes.
The phylogenetic tree of the Pdr5p/15p clusters shows several subclusters (C3 and C4) which aggregates members belonging to specific species such as DEHA or YALI. These are cases where species-based clustering is suspected of hindering a functional inference based on sequence similarity only. For example, the subcluster C4 contains four YALI members belonging to four different chromosomal fragments in unshared environments. Despite being inside the same subcluster, theses genes seem to have very different functions. One of them, YALI0E14729g, exhibits the function of alkane efflux, unique as yet, while another, YALI0C20265g, seems to control the classic azole resistance function of the SACE Pdr5p/15p pumps.
The evolutionary patterns of the Pdrp
sensu lato are simpler and totally independent from that of the Pdr
sensu stricto. The members of Pdr
sensu lato Aus1p/Pdr11p cluster are present in four different species only: ERGO, SAKL, CAGL and SACE. These genes may be lost in other species. Indeed, sterol influx pumps are essential only under anaerobiosis conditions. However, no clear relation between the presence of homologs of Aus1p or Pdr11p sterol-influx pumps and facultative anaerobic growth could be detected among the species composing the
Saccharomyces complex [
45].
The members of the Pdr
sensu lato YOL075c cluster are of unknown function and belong to species of both ancestral clade (
YALI0D25828g) and recent clades. No homolog of
YOL075c is found in DEHA and ERGO. Extensive neighborhood preservation is observed for all
Kluyveromycetes and
Saccharomycetes species. Even though the substrates of this cluster of putative transporters are unknown, its continued presence in many pre- and post-WGD species from the
Saccharomyces complex indicates that these gene products exert an important physiological function. This function and the subcellular localization of the transporter have not yet been determined. Curiously all members of the
YOL075c SONS have at least 100 less amino acid residues (1247 to 1328) than all other members of the GL3C0025 family. Moreover, the cysteine of SGCT motif of the N-Walker A is replaced by the "classic" lysine residue and the NVEQ motif of the C-Walker Signature is replaced by SGGE (Additional file
1).
In three cases only, we observed a divergence between the neighborhood and phylogenetic analyses. ERGO0B08162g, KLLA0D03476g and SAKL0C11704g are allocated to the C cluster (Pdr5p/15p) by phylogeny and to the SONS b (Snq2p/YNR070wp) by neighborhood. This divergence can be explained by the fact that they belong to a tandem gene array in which a copy has been subjected to neofunctionalization. In another case, ZYRO0D17710g (subcluster C1) is allocated to SONS b (Snq2p/YNR070wp) by neighborhood. Its high amino acid similarity with Pdr5p (70%) and significant difference of sequence with SAKL0C11704g and ZYRO0A04114g suggest an evolution based on gene conversion with one of the member of the ZYRO0D11836g, ZYRO0D11858g and ZYRO0D11880g tandem gene array. Finally, the divergence observed for DEHA2F16478g (subcluster C3 and SONS d) may be linked to a species-based clustering of DEHA genes within the subcluster C3.