ABSTRACT
Major intrinsic proteins (MIPs), commonly known as aquaporins, transport water and other non-polar solutes across membranes. MIPs are believed to be involved in host-pathogen interactions. Herein, we identified 17, 24, 27, 19, 19, and 22 full-length MIPs, respectively, in the genomes of six Phytophthora species, P. infestans, P. parasitica, P. sojae, P. ramorum, P. capsici, and P. cinnamomi, which are devastating plant pathogens and members of oomycetes, a distinct lineage of fungus-like eukaryotic microbes. Phylogenetic analysis showed that the Phytophthora MIPs (PMIPs) formed a completely distinct clade from their counterparts in other taxa and were clustered into nine subgroups. Sequence and structural analysis of homology models indicated that the primary selectivity-related constrictions, including aromatic arginine (ar/R) selectivity filter and Froger’s positions in PMIPs were distinct from those in other taxa. The substitutions in the conserved Asn-Pro-Ala motifs in loops B and E of many PMIPs were also divergent from those in plants. We further deciphered group-specific consensus sequences/motifs in different loops and transmembrane helices of PMIPs, which were distinct from those in plants, animals, and microbes. The data collectively supported the notion that PMIPs might have novel functions and could be considered attractive anti-Phytophthora targets.
Highlights
Major intrinsic proteins (MIPs) family in Phytophthora species are divergent.
Phytophthora MIPs (PMIPs) were phylogenetically and structurally distinct from their counterparts in other taxonomic domains.
PMIPs might have novel functions.
The MIPs are suggested to be involved in host-pathogen interactions and could be considered attractive anti-Phytophthora targets.
1. Introduction
The super family of major intrinsic proteins (MIPs) possesses channel-forming integral membrane proteins that transport water and other non-polar small solutes, such as ammonia, urea, boron, silicon, carbon dioxide, glycerol, hydrogen peroxide, antimony, and arsenite (Gomes et al., 2009; Ishibashi et al., 2009; Azad et al., 2012; Verkman, 2012; Maurel et al., 2015; Pommerrenig et al., 2015; Azad et al., 2016; Azad et al. 2018). They are found in all living organisms from bacteria to mammals and are abundant in plants (Gomes et al., 2009; Azad et al., 2011; Maurel et al., 2015). Orthodox aquaporins (AQPs), which transport only water and aquaglyceroporins (AQGPs), that can transport other uncharged small solutes in addition to water or without water, are prototype members of MIPs (Mukhopadhyay et al., 2014; Maurel et al., 2015; Pommerrenig et al., 2015).
All AQPs share some structural features, although their amino acid sequences differ substantially. Each monomer of MIP is composed of six transmembrane (TM) α-helices (H1-H6) with five connecting loops (loops LA–LE) and cytoplasmic N-and C-termini. The pore of the channel is characterized by two constrictions that theoretically specify the profile of transport selectivity. The first constriction is formed at the center of the pore by two highly conserved Asn-Pro-Ala (NPA) motifs on loops B and E because of the close opposition of their asparagine residues (Wallace and Roberts, 2004). This constriction is involved in proton exclusion (Tajkhorshid et al., 2002). The second constriction, called the aromatic/arginine (ar/R) constriction or the selectivity filter, is formed toward the luminal side of the membrane by four residues from helix 2 (H2), helix 5 (H5), and loop E (LE1 and LE2) (Fu et al., 2000; Sui et al., 2001). Mutation at this ar/R selectivity filter is thought to determine the broad spectrum of substrate conductance in plant AQPs (Wallace and Roberts, 2004; Gupta and Sankararamakrishnan, 2009; Azad et al., 2012; Azad et al., 2016). Five relatively conserved amino acid residues known as Froger’s positions (FPs) and designed P1–P5 play roles in MIP sub-grouping and substrate selectivity (Froger et al., 1998; Heymann and Engel, 2000; Azad et al., 2016).
Although 13 different MIPs identified in mammals are divided into three major subfamilies, the genomes of plants encode 2–5-fold or more MIPs, which are grouped into 4–7 subfamilies (Maurel et al., 2015; Azad et al., 2016; Potokar et al., 2016). However, fungi genomes have up to five MIPs with diversified subgroups (Pettersson et al., 2005; Verma et al., 2014). Algae have 1–6 MIPs, but are highly divergent and share only limited similarities with land plant MIPs (Anderberg et al., 2011). In humans, MIPs play significant roles in brain-water balance, kidney nephrons, cell migration, cell proliferation, neural activity, pain, epithelial fluid secretion, skin hydration, adipocyte metabolism, and ocular function (Ishibashi et al., 2009; Verkman, 2012). They are associated with many human diseases, such as glaucoma, cancer, epilepsy, nephrogenic diabetes insipidus, and obesity (Ishibashi et al., 2009; Verkman, 2012), and therefore, they have been potential drug targets (Soveral and Casini, 2017). In plants, MIPs are involved in many physiological processes, such as motor cell movement, root and leaf hydraulic conductance, diurnal regulation of leaf movements, rapid internode elongation, responses to numerous biotic and abiotic stresses, temperature-dependent petal movement, and petal development (Azad et al., 2004; Peng et al., 2007; Azad et al., 2008; Uehlein and Kaldenhoff, 2008; Gomes et al., 2009; Gao et al., 2010; Muto et al., 2011; Azad et al., 2013; Maurel et al., 2015; Afzal et al., 2016; Sonah et al., 2017). MIPs also play important roles in host-parasite interactions and numerous MIPs have been reported in several protozoan parasites, such as Plasmodium, Trypanosoma, and Leishmania species (Hansen et al., 2002; Montalvetti et al., 2004; Beitz, 2005; Fadiel et al., 2009; Kun and de Carvalho, 2009; Baker et al., 2012). These MIPs are considered potential drug targets (Beitz, 2005; Fadiel et al., 2009; Kun and de Carvalho, 2009). In mycorrhized plants, both plant and fungal MIPs play significant roles in water and nutrient transport and in the drought resistance of plants (Giovannetti et al., 2012; Li et al., 2013; Maurel et al., 2015). MIPs in pathogenic fungi may act as attractive targets for antifungal drugs (Verma et al., 2014). However, no study has been conducted on the MIPs of oomycetes, a distinct lineage of fungus-like eukaryotes with diverse microorganisms that are related to organisms such as brown algae and diatoms (Beakes et al., 2012; Thines, 2014; Fawke et al., 2015; Wang et al., 2016). Oomycetes are globally distributed and ubiquitous in marine, freshwater, and terrestrial environments (Thines, 2014), and cause devastating diseases to both plants and animals (Derevnina et al., 2016).
Among oomycetes, the Phytophthora genus comprises more than 117 species, which are highly devastating to a wide range of agriculturally and ornamentally important plants, causing severe economic losses (Martin et al., 2012; Kroon et al., 2004; Tyler et al., 2006; Beakes et al., 2012; Thines, 2014; Fawke et al., 2015; Wang et al., 2016). P. infestans, the cause of late blight of potato and tomato, resulted the Irish Potato Famine in the mid-19th century (Haas et al., 2009; Yoshida et al., 2013; Thines, 2014; Wang et al., 2016), and P. sojae costs the soybean industry millions of dollars each year (Tyler et al., 2006; Sahoo et al., 2017). P. ramorum, the cause of death of oak trees in North America, has severe impact on natural ecosystem, and P. cinnamomi is a substantial threat to natural eucalyptus forests in Australia (Tyler et al., 2006; Lamour et al., 2012; Thines, 2014). P. capsici causes foot rot disease in black pepper and Phytophthora blight in vegetable crops, which causes serious effects on the production of cucurbits, peppers, eggplants, and numerous other important vegetables worldwide (Jackson et al., 2012; Hao et al., 2016; Johnson et al., 2016; Liu et al., 2016). P. parasitica infects a broad range of plants, being capable of infecting over 72 plant genera worldwide (Blackman et al., 2015; Meng et al., 2015).
The genome sequences of these six plant pathogenic Phytophthora species have been completed (Tyler et al., 2006; Haas et al., 2009; Lamour et al., 2012; Fawke et al., 2015) and are available in ‘FungiDB’ (http://fungidb.org/fungidb/). In the study reported herein, we identified and characterized MIP genes in the genomes of these six Phytophthora species (PinMIP, PpaMIP, PsoMIP, PraMIP, PcaMIP, and PciMIP genes of P. infestans, P. parasitica, P. sojae, P. ramorum, P. capsici, and P. cinnamomi, respectively) by using bioinformatics tools. To the best of our knowledge, this is the first report on MIPs of any organism of oomycetes. This report showed that the genomes of Phytophthora species have several-fold greater number of MIP homologues than those of algae, fungi, other parasites, and even more than in mammals. The numbers of their MIP homologues are comparable to that of plants. Comprehensive analysis with different bioinformatics tools revealed that the MIPs in Phytophthora species are phylogenetically and structurally distinct from their counterparts in other taxa.
2. Materials and methods
2.1. Identification of PinMIP, PpaMIP, PsoMIP, PrMIP, PcaMIP, and PciMIP genes
The genomes of P. infestans T30-4, P. parasitica INRA-310, P. sojae strain P6497, P. ramorum strain Pr102, P. capsici LT1534, and P. cinnamomi CBS 144.22 are available at FungiDB (http://fungidb.org/fungidb/). In FungiDB, a search with the keyword ‘aquaporin’ revealed that 21, 26, 35, 32, 25, and 32 aquaporin hits were available for each strain of P. infestans, P. parasitica, P. sojae, P. ramorum, P. capsici, and P. cinnamomi, respectively. We retrieved the protein sequences of these available AQPs for every organism and checked the typical features of AQPs in their primary protein structures. All of the AQP protein sequences available in every Phytophthora species were used as queries for MIPs in the genome of a particular Phytophthora species using TBLASTN and BLASTp tools. Furthermore, the genomes of the Phytophthora species were searched for MIPs using TBLASTN and BLASTp tools with the protein sequences of the complete set of 55 MIPs from Populus trichocarpa (PtMIP), 22 MIPs from Physcomitrella patens (PpMIP), and 13 MIPs from humans (MAQPs) as queries so that XIPs (uncharacterized X intrinsic proteins), GIPs (GlpF-like intrinsic proteins), HIPs (Hybrid intrinsic proteins) or superaquaporins could be detected if they were encoded in the genomes of the six Phytophthora species. MIPs in P. infestans (PinMIPs), P. parasitica (PpaMIPs), P. sojae (PsoMIPs), P. ramorum (PraMIPs), P. capsici (PcaMIPs), and P. cinnamomi (PciMIPs) were included until no more MIPs could be found from the corresponding species. Every sequence from each Phytophthora species was individually compared to identify the maximum number of MIPs for further analyses. Some of the MIP sequences might have been partial or might not have had all the features associated with its MIP channel. To investigate this, we used the multiple alignment program Clustal Omega (http://www.ebi.ac.uk/Tools/msa/clustalo/) to align all the sequences in an individual species. The multiple sequence alignment was used to determine the following features specific to the MIP family: (i) presence of two NPA or NPA-like motifs, (ii) presence of six TM α-helices, and (iii) two functionally important loops possessing the features characteristically present in MIP channels. The TM α-helices were predicted by SOSUI (http://bp.nuap.nagoya-u.ac.jp/sosui/),TMpred (http://www.ch.embnet.org/software/TMPRED_form.html), and the tools of ExPASy (http://kr.expasy.org/tools/). The genomic regions containing MIP genes were further used to determine the gene structure using the program GeneMark.hmm ES-3.0 (Lomsadze et al., 2005) (http://exon.gatech.edu/GeneMark), a self-training based algorithm for prediction of genes from novel eukaryotic genomes. When short genes were found, their sequences with 1000 base flanking regions were subjected to Genetyx_SV_RC_version 7 to investigate their protein sequences.
2.2. Phylogenetic and domain analysis of PinMIPs, PpaMIPs, PsoMIPs, PrMIPs, PcaMIPs, and PciMIPs
To understand the diversity and evolution of PMIPs and to compare them with those in plants, animals, fungi, and algae, we performed phylogenetic analysis of all MIPs in six Phytophthora species using Molecular Evolution Genetic Analysis (MEGA) version 5.0 (Tamura et al. 2011). PinMIPs, PpaMIPs, PsoMIPs, PrMIPs, PcaMIPs, and PciMIPs were aligned with all PtMIPs, PpMIPs, and MAQPs, and five representative members from each of 10 subfamilies of MIPs in fungi (Verma et al., 2014) and all MIPs in the genomes of nine algae (Anderberg et al., 2011) using the Clustal Omega program (http://www.ebi.ac.uk/Tools/msa/clustalo/) and a phylogenetic tree was constructed with MEGA. The evolutionary history was inferred using two different clustering algorithms, namely neighbor-joining and maximum parsimony methods which are generally used for phylogentic analysis of MIPs (Abascal et al. 2014; Verma et al. 2014; Danielson and Johanson 2010), and the genetic distance was estimated by the p-distance method. Reliability of individual branches of the tree was estimated by performing bootstrapping with 1000 replicates. To construct the phylogenetic tree with the MIPs in the six Phytophthora species, all of their MIPs were aligned as above. The identified PinMIPs, PpaMIPs, PsoMIPs, PrMIPs, PcaMIPs, and PciMIPs were classified into different subfamilies and groups based on the phylogenetic tree constructed from them.
2.3. Prediction of subcellular localization and computation of Ka/Ks value
The subcellular localizations of PMIPs were predicted in silico by using tools of WoLF PSORT (http://www.genscript.com/wolf-psort.html) and Cello prediction system (http://cello.life.nctu.edu.tw/) as described previously (Azad et al., 2016). The Ka/Ks values of the PMIPs were calculated using an online Ka/Ks calculation tool at http://services.cbu.uib.no/tools/kaks. A Ka/Ks value greater than one implied gene evolution under positive or Darwinian selection; less than one indicated purifying (stabilizing) selection, and a Ka/Ks value of one suggested a lack of selection or possibly a combination of positive and purifying selections at different points that canceled each other out (Zhang et al., 2013).
2.4. Homology modeling
Homology models were constructed using the Molecular Operating Environment software (MOE 2009.10; Chemical Computing Group, Quebec, Canada), based on a segment-matching procedure (Levitt, 1992) and a best-intermediate algorithm with the option to refine each individual structure enabled. The sequence of each MIP homolog was aligned with the open conformation of spinach PIP, SoPIP2;1 (PDB, Protein Data Bank ID: 2B5F) (Törnroth-Horsefield et al., 2006), or other AQP templates as indicated, using the MOE software as previously described (Azad et al., 2012). The alignment of each MIP homolog was based on both sequence and structural homology with the structure of SoPIP2;1. The 3D structure models were formed using the MOE homology program and the stereochemical quality of the templates and models were assessed as previously described (Azad et al., 2011).
2.5. Determination of pore diameter and pore lining residues
To analyze the MIP channels, the poreWalker server (http://www.ebi.ac.uk/thornton-srv/software/PoreWalker/ (Pellegrini-Calace et al., 2009) was used, which is a fully automated method designed to detect and characterize transmembrane protein channels from their 3D structures. The 3D structure of each MIP was uploaded to the server, which generated its specific pore characteristics; in particular, the conformation and the regularity of the channel cavity, the corresponding pore lining residues and atoms, and the location of pore centers along the channel. The pore diameter at the ar/R selectivity filter was determined from the PoreWalker outputs as previously described (Azad et al., 2016). The pore lining residues, which are essential for channel formation, were identified using the PoreWalker server.
3. Results
3.1. Genome-wide identification of PinMIP, PpaMIP, PsoMIP, PraMIP, PcaMIP, and PciMIP genes
The whole genome shotgun sequence (WGS) of P. infestans, P. parasitica, P. sojae, P. ramorum, P. capsici, and P. cinnamomi available at FungiDB was searched for PinMIP, PpaMIP, PsoMIP, PraMIP, PcaMIP, and PciMIP genes using TBLASTN. The query PinMIP, PpaMIP, PsoMIP, PtMIPs, PpMIPs and MAQPs sequences resulted in 21, 27, and 33 hits for PinMIP, PpaMIP, and PsoMIP, respectively. We further analyzed the PinMIP, PpaMIP, and PsoMIP sequences for manual inspection of their amino acid sequences, TM domains, and homology models. After all these analysis, we found that out of 21, 27, and 33 hits for PinMIP, PpaMIP, and PsoMIP, respectively, 4, 3, and 6 were deemed to be pseudo MIP genes in P. infestans, P. parasitica, P. sojae, respectively, and were discarded. Characteristics, such as short sequences, N-or C-termini less, addition or deletion of sequences, and combinations thereof were observed in the discarded MIPs. However, the query PraMIP, PcaMIP, and PciMIP returned no hits in the TBLASTN and BLASTp searches. Therefore, we retrieved the MIP sequences of P. ramorum, P. capsici, and P. cinnamomi from the FungiDB and analyzed them as mentioned above. Out of 32 MIP sequences in each of P. ramorum, P. capsici, and P. cinnamomi, 13 MIPs in the former two and 10 MIPs in the latter were deemed pseudo MIPs for the reasons noted above and were discarded. We ultimately obtained 17, 24, 27, 19, 19, and 22 full-length PinMIP, PpaMIP, PsoMIP, PraMIP, PcaMIP, and PciMIP protein sequences from the WGS of P. infestans, P. parasitica, P. sojae, P. ramorum, P. capsici, and P. cinnamomi, respectively (Tables 1-6). The amino acid lengths of PinMIP, PpaMIP, PsoMIP, PrMIP, PcaMIP, and PciMIP homologues with their maximum sequence identity with MIPs in other Phytophthora, humans (taxid 9606), Arabidopsis thaliana (taxid 3702), fungi (taxid 4751), and algae (taxid 3027) are tabulated in Tables 1–6. Although the MIPs of one Phytophthora species exhibited a maximum 80–95% identity with those of other Phytophthora species (Tables 1-6), the TBLASTN search revealed that their highest identity with MIPs of Arabidopsis thaliana (taxid: 3702), Homo sapiens (taxid: 9606), fungi (taxid: 4751), and algae (taxid: 3027) was within 33, 45, 52, and 32%, respectively (Tables 1-6). This result indicated that PMIPs have higher identity with MIPs of fungi compared to those of animals, plants, and algae. This result further supported the notion that PMIPs might have extensive divergence in their sequence and structural properties compared to those in other taxonomic groups.
The Ka/Ks value was >1 only for PinMIPC1;2 and PsoMIPG1;9, indicating that the evolution of these genes in P. infestans and P. sojae, respectively, was likely under positive or Darwinian selection (Tables 1-6). The Ka/Ks value 0 for PpaMIPC1;3, PsoMIPG1;8, PraMIPG1;2, and PraMIPG1;3 indicated neutral selection. The remaining PMIPs showed Ka/Ks values <1, demonstrating purifying selection. The score of the Cello prediction system showed that all PMIPs might have been localized in the plasma membrane (Supplementary Table S1). However, WoLF PSORT scores suggested that some PMIPs have possibility to localize in other intragranular membranes, such as the cytoskeleton, endoplasmic reticulum, vacuole, golgi, and mitochondria.
3.2. Phylogenetic distribution and nomenclature of PMIPs
The sequence of PMIPs was routinely compared with the subfamilies of those in other kingdoms by constructing multiple and/or pair-wise alignments using Clustal Omega and EMBOSS Needle, respectively. To investigate the structural conservation of PMIPs with other MIPs in plants, animals, and microbes, a 3D structural alignment was constructed with the templates of human AQP1 (PDB ID, 1J4N), spinach plasma membrane intrinsic protein, SoPIP2;1 (PDB ID, 2B5F), and E. coli glycerol facilitator, GlpF (PDB ID, 1FX8). All 3D models constructed with these three templates showed the typical hourglass shaped AQPs with six TM helices and five connecting loops (Figure 1). Superposition of the three models of each PMIP showed that the helices superposed very closely. Deviation was observed only in the loops. This structural alignment was used as a guide for sequence alignment. To classify the PMIPs, their protein sequences were analyzed phylogenetically with representatives of subfamilies of MIPs in two plant genomes, and a human, fungi, and algae genome, as described in Materials and Methods. Although PtMIPs and PpMIPs were divided into five and seven subfamilies, respectively (Danielson and Johanson, 2008; Gupta and Sankararamakrishnan, 2009), and those in fungi and algae were separated into 10 and seven subfamilies, respectively (Anderberg et al., 2011; Verma et al., 2014), PMIPs formed a completely distinct clade (Figure 2). Despite one MIP in P. parasitica, which was associated with Delta AQGPs of fungi (Verma et al., 2014), no other PMIP was associated with any subfamilies of MIPs in other organisms in the three domains of life. We therefore constructed a phylogenetic tree with all 126 MIP homologues in the six Phytophthora species (Figure 3 and supplementary figure S1). The 126 MIPs were clustered into nine subgroups. Because the PMIPs were not phylogenetically clustered with any subfamily or group of MIPs reported in plants, animals, fungi, algae, or bacteria, these groups were arbitrarily named MIPA-MIPI. Each MIP homologue in every Phytophthora species was named by taking the first letter (Upper case) from the genus and the first two letters (lower case) of the species names with the MIP group (A to I), and the number of the homologues in each group was stated consecutively, i.e., PsoMIPA1;1 for MIPA in P. sojae and so on. MIPAs and MIPCs-MIPGs included one to several MIPs from each of the six Phytophthora species. MIPBs contained five homologues of five species, except P. capsici, and MIPHs consisted of four MIPs from three Phytophthora species. The MIPI group had only one homologue, which was Delta AQGP, as previously mentioned. However, although these groups included a small number of MIP homologues, they had a distinct ar/R constriction. The characteristics of each of the nine PMIPs groups are detailed in the next sections.
3.3. Sequence analysis of PMIPs
We calculated the pair-wise sequence identity and similarity among the intragroup and intergroup PMIPs by using EMBOSS Needle (http://www.ebi.ac.uk/Tools/psa/emboss_needle/). The average identity and similarity among the intragroup PMIPs was 76% and 83%, respectively, whereas those in the intergroup PMIPs were 40 and 52%, respectively (Supplementary Table S2). However, the sequence identity and similarity of MIPHs to MIPA-MIPGs was only 34 and 48%, respectively. Nevertheless, the intergroup average sequence identity among MIPAs-MIPDs and MIPEs-MIPGs was 46 and 50% and similarity was 64 and 62%, respectively. In contrast, the intergroup average sequence identity and similarity of PMIPs from MIPAs-MIPDs to MIPEs-MIPGs was 48 and 63%, respectively. These results indicated that PMIPs of MIPAs-MIPDs were closer compared to those of MIPEs-MIPGs and vice versa. The intergroup sequence identity varied from 2 to 30% and similarity varied from 1 to 56%, indicating that each group was divergent from the others.
3.4. Gene structure of MIPs in Phytophthora species
All the full-length MIP sequences found in P. infestans, P. parasitica, P. sojae, P. ramorum, P. capsici, and P. cinnamomi were analyzed for introns and exons (Figure 4). Interestingly, among the 126 PMIPs, 107 homologues had no introns. This characteristic might be common to prokaryotic genes. Thirteen PMIPs had one introns. Two introns were observed in four PMIP genes, namely PciMIPH1;1, PinMIPE1;1, PinMIPG1;2, and PsoMIPH1;1. Three and five introns were found only in PciMIPA1;4 and PinMIPC1;3, respectively. Despite a few disparities, conserved intron positions were not found in PMIPs (Figure 4).
3.5. Analysis of ar/R selectivity filter and FPs of PMIPs
The amino acid residues in the ar/R selectivity filter and FPs are crucial for functional grouping and substrate selectivity of MIPs (Froger et al., 1998; Heymann and Engel, 2000; Hove and Bhave, 2011; Pommerrenig et al., 2015; Azad et al., 2016; Azad et al. 2018). To determine the residues in the ar/R selectivity filter and FPs, we constructed 3D models of all PMIPs. The structure-based alignments and multiple sequence alignments of PMIPs helped us to identify the four amino acid residues at the ar/R selectivity filter, and the five residues in the FPs. The residues at the ar/R selectivity filter and FPs in nine groups are shown in Figure 3. Although MIPs in plants and fungi conserve group-specific ar/R selectivity filters and FPs (Wallace and Roberts, 2004; Gupta and Sankararamakrishnan, 2009; Verma et al., 2014; Pommerrenig et al., 2015; Azad et al., 2016), the ar/R filter and FPs were identical in several groups of PMIPs (Figure 3). Despite a few disparities (PciMIPA1;4, PsoMIPA1;4, PpaMIPA1;3, and PinMIPE1;2), the tetrad in the ar/R selectivity filter in MIPAs and MIPDs-MIPGs was WGYR, which was found in β-AQPGs in fungi (Verma et al., 2014). However, MIPBs, MIPCs, MIPHs, and MIPI contained group-specific ar/R selectivity filters, having the tetrad WGCR, WALR, WSLR, and WTAR, respectively, which are not usually observed in MIPs of other taxonomic groups. Nevertheless, the residue in the H5 position of the ar/R filter in PciMIPH1;1 and PsoMIPH1;1 was deleted. The ar/R selectivity filter, as the name suggests, usually consists of an aromatic residue and R, was found in the H2 and LE2 position, respectively. This was true for all PMIPs, although exceptions have been reported in plants, fungi, and algae (Gupta and Sankararamakrishnan, 2009; Anderberg et al., 2011; Verma et al., 2014; Azad et al., 2016). The H5 position in PMIPs is occupied by small residues (G/A/S), which is generally the case in many MIPs in plants (TIPs, NIPs, and SIPs), fungi, and algae (Anderberg et al., 2011; Verma et al., 2014; Azad et al., 2016). However, all PIPs in plants, most of the AQPs in mammals, and some groups of MIPs in algae conserved H, and some fungal, algal, and plant MIPs conserved L/I/V in the corresponding position ((Anderberg et al., 2011; Azad et al., 2012; Verma et al., 2014; Azad et al., 2016), Supplementary Figure S2). A/C/G/S/T residues are usually found in the LE1 position in MIPs of plants, animals, fungi, and algae (Anderberg et al., 2011; Verma et al., 2014; Azad et al., 2016). Except for some fungal MIPs, an aromatic residue is not usually available at the LE1 position in eukaryotes. Interestingly, most of the PMIPs (MIPAs, MIPDs-MIPGs) conserved the bulky hydrophobic aromatic Y in the LE1 position (Figure 3). The bulky aromatic W and Y in the H2 and LE1 positions in many PMIPs in the five groups collectively might have influenced the channel properties and their transport profile. Hydrophobic L, which is not generally found in MIPs of plants, animals, and algae, but available in some fungal MIPs, is conserved in homologues of MIPC and MIPH. This hydrophobic larger amino acid would have changed the channel property and transport profile compared to MIPs that have A/C/G/S/T in the same position. Superposition of the 3D models of PMIPs with crystal structures of bacterial glycerol facilitator (Glps), bovine aquaporin 1 (AQP1), and spinach PIP, SoPIP2;1, or that of intergroup homologues revealed that the architecture of the ar/R selectivity filter in PMIPs would be influenced by the residue at the H5 and LE1 positions (Figure 5). However, this might be one of the reasons for transport selectivity. The conserved residue WGYR in the ar/R selectivity filter of PMIPAs and MIPDs-MIPGs indicated that their transport selectivity might be caused by another mechanism or they have the same transport profile.
The FPs, P1 from H3, P2 and P3 from LE, and P4 and P5 from H6 as described by Froger et al. (Froger et al., 1998) were YDRFW in MIPAs-MIPDs, excluding two homologues, PsoMIPA1;4 and PsoMIPA1;5, in which the F was substituted by G and W, respectively (Figure 3). Although these FPs are diverse from MIPs in plants, animals, and algae, most of the groups of fungal MIPs conserve them (Supplementary Figure S2). In MIPEs-MIPHs, the P2-P5 positions were conserved with DRFW, except PsoMIPF1;2, PpaMIPF1;3, PsoMIPF1;3, PcaMIPF1;2, and PciMIPF1;3, in which the W in the P5 position was substituted by L. The P1 position in these groups was diverse, Q/D in MIPEs; D/N in MIPFs; Q in MIPGs; and H/N in MIPHs. The FPs in the MIPI (PpaMIPI1;1) was FDRCW.
3.6. PMIPs with substituted NPA motifs
The conserved NPA motifs in LB and LE were found in MIPAs, MIPBs, MIPDs, MIPI, and a subgroup of MIPCs, except PciMIPA1;4, in which the P of the NPA in LE was substituted by S (Figure 3). As reported for the small basic intrinsic proteins (SIPs) in all plants (Gomes et al., 2009; Azad et al., 2016), the homologues of MIPEs-MIPGs conserved a usual NPA motif in the LE, but an unusual NPA motif in the LB, apart from PpaMIPG1;1 that contained the usual NPA in both loops and PinMIPG1;1, in which the N of the NPA in LE was substituted by K. However, unlike SIPs, MIPEs-MIPGs have high molecular weight with longer amino acid sequences. Moreover, although substitution of A by T or L in the NPA motif of LB is observed in SIPs (Verma et al., 2014; Azad et al., 2016), P was substituted by T in all MIPEs and MIPGs. In seven homologues of MIPFs, P was substituted by C or S, and in six homologues of the same group, A was substituted by S. Interestingly, two MIP homologues from each of the six Phytophthora species had unusual NPA motifs both in LB and LE, and they were clustered as a subgroup in MIPCs. More intriguingly, the NPA motifs of LB and LE in MIPHs were substituted with ISV and S(P/V/I)N, respectively. The NIPs with unusual NPA motifs, where A was in LB and that in LE were substituted by S and V, respectively, had characteristic R-rich C-termini (Azad et al., 2016; Azad et al. 2018). However, MIPCs with both unusual NPA motifs had a RSxGPYE(Y/F) C-terminus followed by the group-conserved GYHH motif. Although MIPHs’ C-termini have no such conserved sequences before the ExQH motif, they contain comparatively more charged residues (Supplementary Figure S3).
3.7. Group-specific consensus sequences/motif of PMIPs
We further compared the PMIPs sequences for group-specific consensus sequences. In the TM regions, the intergroup similarities among the PMIPs were very high. The group-specific deviation was particularly observed in the loops and N-and C-termini (Table 7). Most of the homologues in MIPEs-MIPGs had longer N-termini compared to MIPAs-MIPCs (Supplementary Figure S3). Interestingly, MIPDs had short N-and C-termini. It is more intriguing that excluding only three PMIPs in group E (PinMIPE1;1, PcaMIPE1;1, and PpaMIPE1;1), all PMIPs conserved a positively charged histidine residue in the C-terminus. With this conserved H residue, all PMIPs conserved a group-specific motif (Table 7). Except MIPEs and MIPHs, the C-termini in all PMIPs were H-enriched, to a greater extent in MIPAs and MIPBs. The C-termini of MIPEs were enriched with D and E and had di-or tri-acidic residues. Despite the NPA motifs in LB and LE, every loop (LA-LE) had group-conserved sequences (Table 7). However, the sequence shown in LB shared both TM2 and LB. In all PMIPs, the LE was longer, where two group-conserved sequences were observed. The first one composed of eight residues included the P2 position. The second consisted of six residues and was located downstream of the P3 position. We further identified group-specific consensus sequences, with one in each of TM3 and TM5, and two in TM4. In the upstream region of TM3, there was a YxxxQ motif in all PMIPs, in which xxx contained group-specific residues. In the consensus sequence composed of eleven residues (Pxxxxxxxxx(M/S) in TM5, the interior positions were blocked with group-wise preserved residues. Nevertheless, the first group-wise conserved sequence located at the start of TM4 was composed of eight residues; and the second one located before the end of the TM4 was composed of seven residues.
4. Discussion
MIPs have a main role in water and solute transport and aid in homeostasis during plant stress responses (Afzal et al., 2016). Recently, transcriptome data provided important clues about the involvement of MIPs in host-pathogen interactions (Galindo-Gonzàlez and Deyholos, 2016; Reeksting et al., 2016; Guo et al., 2017). Although there has been no study on PMIPs, in the present study, we identified and characterized a total of 126 MIP homologues from the genomes of six Phytophthora species, which cause severe economic losses because of devastating effects on numerous agriculturally and ornamentally important plants (Tyler et al., 2006; Haas et al., 2009; Lamour et al., 2012; Thines, 2014; Fawke et al., 2015; Derevnina et al., 2016; Wang et al., 2016). This study provided comparative information in context of genome-wide number, subclasses or groups, structural insight, and evolution of PMIPs relative to those in taxonomic groups.
4.1. PMIPs are a new paradigm in microbial aquaporins
Although the number of MIP homologues varies from organism to organism, plants comparatively have more homologues than animals and microbes. Although bacterial genomes have 1–2 and fugal and/or algal genomes have only 1–6 MIP genes (Wang et al., 2005; Anderberg et al., 2011; Azad et al., 2011; Verma et al., 2014), this study showed that the genomes of fungi-like Phytophthora species of oomycetes had 17–27 MIP genes (Tables 1-6). The number of MIP genes in the genomes of Phytophthora species was higher than that even in the human genome and almost similar to that in many plants (Maurel et al., 2015; Azad et al., 2016; Potokar et al., 2016). The increase in the number of PMIPs might be have been caused by gene duplication or horizontal gene transfer in addition to the polyploidy nature of Phytophthora spp. (Bancroft, 2001; Moore and Purugganan, 2003; Bertier et al., 2013). More interestingly, systematic searching and phylogenetic and structural analysis revealed that PMIPs were distinct and did not cluster to those in taxonomic groups (Figures 2, 3). Data collectively indicated that despite some fungal MIPs (Verma et al., 2014), PMIPs had distinct ar/R filter and FPs, and substitution in the conserved NPA motifs in comparison with those in plants, animals, and algae were divergent. The large numbers, phylogenetical and structural novelty of PMIPs reflected their wide diversity in function and physiological relevance.
4.2. Group-specific characteristic C-termini and consensus motifs likely to be associated with novel functions
The uneven length of N-and C-termini of PMIPs groups (Table 7, Supplementary Figure 3) might have affected their interaction with other molecules or physical interaction for heteromerization of PMIPs (Fetter et al., 2004; Yaneff et al., 2014) because the protein termini were generally exposed on the surface of protein structures making them available for interaction (Jacob and Unger, 2007; Tanco et al., 2015). The physical interaction through heteromerization is one of the mechanisms for regulation of intrinsic permeability of MIPs (Yaneff et al., 2014). The C-termini of proteins have been associated with diverse biological functions and processes, such as membrane integration of proteins, protein activity, protein sorting, post-translational modification, protein-protein interaction, or formation of protein complexes (Chung et al., 2002; Azad et al., 2008; Tanco et al., 2015). The conserved positively charged H residue in all PMIPs, which was included in the group-specific C-terminal motif (Table 7), is a novel character not observed in other MIPs. Furthermore, despite MIPEs that are enriched with negatively charged D and E residues, the C-termini in most PMIPs were enriched with positively charged H, although its extent is group-specific (Supplementary Figure 3). Therefore, the distinctive C-termini in PMIPs might be associated with protein sorting, protein-protein interaction, post-translation modification, or other novel functions. KDEL, HDEL, or KKXX motifs in the C-termini of proteins are involved in the retention of protein in the endoplasmic reticulum and prevent them entering into the secretory pathway (Chung et al., 2002; Capitani and Sallese, 2009; Tanco et al., 2015). The NIPs with unusual NPA motifs in LB and LE, have a characteristic R-rich C-termini, which are not seen in NIPs with only one unusual NPA motif, and the R-rich C-termini are thought to be involved in structural stabilization of MIPs (Ishibashi, 2006; Worth and Blundell, 2010; Azad et al., 2016). Although a subgroup of MIPCs with both unusual NPA motifs have no such R-rich C-termini, their conserved RSxGPYE(Y/F) sequence would have been involved in the same function in addition to the other aforementioned functions of C-termini because of having charged and phosphorylatable residues. Similarly, MIPHs having both unusual NPA motifs would have different functions because of the presence of comparatively more charged (20–40%) residues in the last 10 amino acids of the C-termini (Supplementary Figure 3). In contrast, SIPs with only one unusual NPA motif have K-rich C termini (Azad et al., 2016), which is a potential endoplasmic retention signal (Ishibashi, 2006; Gomes et al., 2009). However, MIPEs-MIPGs have no such K-rich C-termini. It is usually supposed that MIPs with unusual NPA motifs are involved in non-aqua transport rather than water transport (Pommerrenig et al., 2015; Azad et al., 2016). However, water transport activity has been demonstrated in two AtSIP1s, but not in AtSIP2;1, and the latter is supposed to have non-aqua transport activity (Fetter et al., 2004).
Comparison of PMIPs with MIPs in other taxonomic groups revealed that the group-specific motifs or consensus sequences in PMIPs shown in Table 7 are distinct from the corresponding positions in MIPs of other domains of life (Supplementary Figure S2). However, some of these motifs or consensus sequences are partly similar to some of the fungal MIPs. Interestingly, the YxxxQ motif in TM3 is largely conserved in most MIPs of other taxonomic groups. This might have important structural roles in MIPs as is reported for NPA motifs. Some of the residues in the group-specific motif or consensus sequences are pore-lining (Supplementary Figure 3), which in turn, may influence the transport profiles of PMIPs. Experimentally proven non-aqua transporter MIPs in plants have substrate-specific signature sequences (SSSS) or specificity-determining positions (SDPs) in the NPA regions, ar/R filters, and FPs (Hove and Bhave, 2011; Azad et al., 2016). These SSSS and SDPs have been shown to be used as tools to predict non-aqua transport profiles of plant MIPs (Hove and Bhave, 2011; Azad et al., 2016; Azad et al. 2018). Based on the SSSS or SDPs in individual constrictions, several PMIPs were predicted to be non-aqua transporters (Supplementary Table S3). However, considering only the PMIPs commonly predicted by the SSSS or SDPs for all three constrictions, no PMIP was predicted to be non-aqua transporter. This result indicated that the SSSS and SDPs of PMIPs in the three constrictions might be structurally distinct, which in turn, would have imparted novel functions to PMIPs. It will be interesting to conduct wet-lab experiments to determine the transport activities and elucidate the structure-function relationship of PMIPs.
4.3. PMIPs could be attractive targets for anti-Phytophthora disease
As previously mentioned, Phytophthora spp. have been identified to cause devastating diseases in wide range of plants, including agriculturally and economically important crops, such as rice, potato, tomato, wheat, rye, barley, and fruits (Fawke et al., 2015; Reeksting et al., 2016; Wang et al., 2016). These plant pathogens display resistance to existing fungicides, indicating the necessity of new drugs to eradicate Phytophthora from crops. To develop new drugs and treatments, the physiology of these pathogenic Phytophthora need to be studied in detail, specifically the significance of membrane transporters such as MIPs in plant-Phytophthora interactions. Understanding the plant-Phytophthora interactions and the infection mechanism is tremendously important and will aid in developing anti-Phytophthora drugs. MIPs are known to transport water, glycerol, and other physiologically important small-uncharged solutes (Chaumont and Tyerman, 2014; Afzal et al., 2016; Azad et al., 2016). Our present study revealed that PMIPs are phylogenetically and structurally distinct from their counterpart in other taxonomic domains (Figures 2, 3). The ar/R selectivity filter of PMIPs is unique and hydrophobic in nature and is likely to transport novel hydrophobic solutes. The distinctive FPs and group-specific consensus motifs (Figure 3, Table 7) further support the same notion that PMIPs might have novel transport profiles, which might include essential interactors for plant-Phytophthora interactions. The nature of molecules that are transported through PMIPs and the role of group-specific consensus motifs that might have regulated the channel function must first be experimentally investigated. Hence, PMIP channels in disease causing Phytophthora can be considered as attractive drug target like some AQP modulators that are promising agents for the treatment of human disorders (Huber et al., 2012). However, the revelation of 3D structures and crystallographic information of PMIPs will broaden the understanding of drug design against Phytophthora-mediated plant diseases.
5. Conclusions
After its discovery in the last decade of the twentieth century, the MIP has fascinated scientists with its potential to aid in understanding the molecular physiology of organisms and development of new innovative pharmacological agents for the treatment of human and plant diseases, owing to their potential structure and functions. The host-pathogen interactions at Phytophthora-induced infection require understanding at the molecular level. In this context, understanding the evolution, structure, function, and diversity of PMIP channels might be very significant. In the present study, a genome-wide identification of MIPs has been performed in six Phytophthora species, which are devastating plant pathogens and members of oomycetes, a distinct lineage of fungus-like eukaryotic microbes. Every Phytophthora genome encodes several-fold MIP homologues compared to other eukaryotic microbes, such as fungi and algae, even more than in the human genome, and a similar number of homologues as found in many plants. The PMIPs were phylogenetically and structurally distinct from their counterparts in other taxonomic domains. Sequence analysis and homology modeling studies indicated that the ar/R selectivity filter, Froger’s positions, and group-specific consensus sequences/motifs in different loops and TM helices of PMIPs are distinct from those in other taxonomic domains. The substitutions in the conserved NPA motifs of many PMIPs were also unique. The data collectively support the notion that PMIPs might have novel functions and could be considered attractive anti-Phytophthora targets. Currently, no functional studies are available on PMIPs. The present study has provided a picture of PMIPs with distinct evolutionary relationships and structural properties. However, wet-lab experiments are needed to find out the possible solutes that are transported through the PMIP channels and the importance of these channels in the Phytophthora life cycles.
Author’s contributions
AKA and JA conceived the project. AKA designed the work. JA, AH, MAA, MMH and MH carried out the work and participated in analysis with the close supervision of AKA. JA participated in draft preparation of some parts of the manuscript. AKA supervised all procedures and wrote the manuscript. TI and YS critically read the manuscript. All authors approved the final version of the manuscript.
Competing interests
The authors declare that they have no competing interests.
Supplementary Figure S1: Grouping of MIPs from Phytophthora species based on phylogeny. The phylogenetic tree was generated as described in Figure 2. Each MIP group is shown with a specific background color to distinguish them from others.
Supplementary Figure S2: Multiple alignment of MIPs shown in Figure 2. The amino acid sequences were aligned using the Clustal Omega program. The TM helices and the dual NPA motifs are shown as gray and yellow, respectively. The residues in the ar/R selectivity filter and FP are green and cyan boxed, respectively.
Supplementary Figure S3: Multiple sequence alignment of MIPs in the six Phytophthora species. The amino acid sequences were aligned using the Clustal Omega program. Each MIP group is shown with a specific background color to distinguish them from others. The TM helices are shown within boxes with black lines and the dual NPA motifs are shown as gray. The residues in the ar/R selectivity filter and FP are red and blue boxed, respectively. The region of the TM helices and loops from which consensus sequences or motifs were depicted in Table 7 are green boxed. The pore-lining residues are indicated by arrows above the alignment and the conserved residues are indicated by stars (*) at the bottom of the alignment.
Supplementary Table S1: Predicted sub-cellular localization of PMIPs
Supplementary Table S2: Average pairwise sequence identity and similarity between different PMIP groups.
Acknowledgements
JA and AH were supported by stipends from The University Grant Commission of Bangladesh. There was no additional external funding received for this study.