- Research article
- Open Access
Single amino acid substitutions in the selectivity filter render NbXIP1;1α aquaporin water permeable
BMC Plant Biology volume 17, Article number: 61 (2017)
Aquaporins (AQPs) are integral membrane proteins that facilitate transport of water and/or other small neutral solutes across membranes in all forms of life. The X Intrinsic Proteins (XIPs) are the most recently recognized and the least characterized aquaporin subfamily in higher plants. XIP1s have been shown to be impermeable to water but permeable to boric acid, glycerol, hydrogen peroxide and urea. However, uncertainty regarding the determinants for selectivity and lack of an activity that is easy to quantify have hindered functional investigations. In an effort to resolve these issues, we set out to introduce water permeability in Nicotiana benthamiana XIP1;1α (NbXIP1;1α), by exchanging amino acid residues of predicted alternative aromatic/arginine (ar/R) selectivity filters of NbXIP1;1α for residues constituting the water permeable ar/R selectivity filter of AtTIP2;1.
Here, we present functional results regarding the amino acid substitutions in the putative filters as well as deletions in loops C and D of NbXIP1;1α. In addition, homology models were created based on the high resolution X-ray structure of AtTIP2;1 to rationalize the functional properties of wild-type and mutant NbXIP1;1α. Our results favour Thr 246 rather than Val 242 as the residue at the helix 5 position in the ar/R filter of NbXIP1;1α and indicate that the pore is not occluded by the loops when heterologously expressed in Pichia pastoris. Moreover, our results show that a single amino acid substitution in helix 1 (L79G) or in helix 2 (I102H) is sufficient to render NbXIP1;1α water permeable. Most of the functional results can be rationalized from the models based on a combination of aperture and hydrophobicity of the ar/R filter.
The water permeable NbXIP1;1α mutants imply that the heterologously expressed proteins are correctly folded and offer means to explore the structural and functional properties of NbXIP1;1α. Our results support that Thr 246 is part of the ar/R filter. Furthermore, we suggest that a salt bridge to an acidic residue in helix 1, conserved among the XIPs in clade B, directs the orientation of the arginine in the ar/R selectivity filter and provides a novel approach to tune the selectivity of AQPs.
Major Intrinsic Proteins (MIPs), commonly referred to as aquaporins (AQPs), constitute a large superfamily of channel proteins permeable to water and/or small uncharged solutes . AQPs are found in all forms of life and are particularly in high abundance in plants [2,3,4,5,6]. There are seven aquaporin subfamilies in plants, namely: Plasma membrane Intrinsic Proteins (PIPs), Tonoplast Intrinsic Proteins (TIPs), Nodulin 26-like Intrinsic Proteins (NIPs), Small basic Intrinsic Proteins (SIPs), Glycerol facilitator-like Intrinsic Proteins (GIPs), Hybrid Intrinsic Proteins (HIPs) and X Intrinsic Proteins (XIPs) [2, 7, 8].
The tetrameric aquaporin structure, wherein each monomer forms a functional pore, appears to be conserved among all AQPs. The monomer consists of six transmembrane helices (helix 1 – helix 6) connected by five loops (loop A – E), and both the N and C termini are located in the cytoplasm (Fig. 1). Loops B and E fold back into the membrane from opposite sides and form two short helices, denoted helix B and helix E, that jointly add a seventh α-helical transmembrane structure to the right-handed α-helical bundle of the monomer. These two half α-helices contain the asparagine-proline-alanine (NPA) aquaporin motif at their N-terminal ends that meet in the middle of the protein. Substrate permeability of AQPs is governed by two regions in the pore.
Firstly, the macro-dipoles of the two half-helices contribute to repulsion of positive charges at the conserved NPA region  and secondly, the variable aromatic arginine (ar/R) selectivity filter region modulates substrate specificity in AQPs by providing a substrate-specific aperture and specific interactions that fit the substrate . This second region was originally thought to consist of four amino acid residues [11,12,13,14], however, the recently solved X-ray crystal structure of the Arabidopsis thaliana TIP2;1 aquaporin (AtTIP2;1) at an atomic resolution of 1.18 Å revealed an extended ar/R filter with five amino acid residues at specific positions in the pore . In addition to the initially identified positions in helices 2, 5, E (denoted H2P, H5P, HEP) as well as in loop E (LEP), a position in loop C (LCP) was uncovered by this structure. Furthermore, in the ammonia and water permeable AtTIP2;1 the conserved arginine at position HEP of the ar/R selectivity filter was found in a new spatial location stabilized by a hydrogen bond to a histidine at position H2P. Mutational studies of human AQP1 showed that it was possible to mimic the TIP2-like substrate profile , suggesting that a similar orientation of the arginine may be induced by mutations also in other AQPs.
XIPs are easy to distinguish from other AQPs due to their conserved LGGC and NPARC motifs in loop C and in helix E, respectively . To date, XIPs have been found mostly in dicotyledonous plants, fungi and protozoa [5,6,7,16,8, 15–17]. Based on phylogenetic analyses plant XIPs are divided into four subgroups (I-IV) of which group I consists of XIPs from lower plants like moss and spikemoss . The XIPs of eudicots are divided into clades A and B, corresponding to subgroups II and III + IV, respectively. It is clear that the XIP subfamily has gone through a relatively recent lineage specific expansion, as some groups with multiple paralogs are only found in closely related taxa . In Nicotiana benthamiana there are, for example, three closely related genes, NbXIP1;1–1;3, of which NbXIP1;1 and 1;2 are alternatively spliced . Although purifying selection has been the main evolutionary force, it appears that plant XIPs have gone through periods of positive selection . Thus the evolutionary history of plant XIPs indicate that the different clades of XIPs have different functions. According to sequence similarity and proposed substrate selectivity, XIPs were originally thought to be related to NIPs [7, 8]. However, in primitive plants a functional redundancy of XIPs and TIPs has been suggested based on the occurrence of a histidine at position H2P of the selectivity filter . Furthermore, some support for a deep orthology with HIPs, TIPs and animal AQP8s (all with a histidine at H2P) has been put forward based on phylogenetic studies . Altogether this suggests that the structure of AtTIP2;1 should be a suitable template for homology modeling of XIPs.
AtTIP2;1 has been reported to be highly permeable to ammonia and water [14,23,24,, 22–25]. On the other hand, XIPs were predicted and later found to be impermeable to water but permeable to boric acid, glycerol, urea, and hydrogen peroxide [7, 8, 26]. In a recent study, the amino acid residue at the helix two position (H2P) in the ar/R filter showed a significant divergence among XIPs , suggesting that the residue at this position may have a strong effect on the substrate specificity of XIPs.
XIPs deviate from other AQPs in several ways and this has made it difficult to align them, especially in loops and in the region corresponding to helix 5. Based on the conservation of a glycine residue in helix 5, two alternative alignments have been suggested, placing a valine or a threonine at the H5P position of most dicot XIPs, whereof the former alignment is favoured by hydrophobicity plots and an apparent release of structural constraints . However, the identity of the residue in position H5P of the ar/R filter as well as the length of the flanking loops D and E are yet to be determined.
As mentioned above, the XIP1;1 pre-RNA in N. benthamiana can be alternatively spliced resulting in two different NbXIP1;1 proteins . The NbXIP1;1α protein studied here has a one amino acid shorter N-terminal region compared to the NbXIP1;1β protein. In a recent study we demonstrated that the NbXIP1;1α splice-variant is permeable to boric acid when expressed in the yeast Pichia pastoris, both by growth assays and when purified and reconstituted in proteoliposomes . However, attempts to quantify boric acid permeability in yeast spheroplasts were unsuccessful (unpublished results). Based on the modest boric acid permeability of the reconstituted NbXIP1;1α in proteoliposomes it was suggested that the protein was gated or only partially active. Interestingly, the heterologously expressed and purified protein was found to be phosphorylated at several positions in the N-terminal region, although the level of phosphorylation was not quantified . It is conceivable that the pore is occluded by loop C or D since both are relatively long in the XIPs. In contrast to other subfamilies, XIPs and PIPs have a D loop of similar length which is thought-provoking as this loop is central in the gating of PIPs . Depending on the exact alignment of helix 5, loop D in XIPs is either one amino acid residue shorter or three residues longer than in the PIPs.
Mutational studies of NbXIP1;1α have been hampered by the lack of an activity that is easy to quantify directly in spheroplasts, omitting the time consuming steps of purifying and reconstituting each mutant protein, which is needed when reconstituting AQPs in proteoliposomes. In an effort to introduce water permeability in the NbXIP1;1α and to gain in-depth understanding of the function of the different amino acids of the XIP selectivity filter, we set out to exchange the NbXIP1;1α ar/R filter for the AtTIP2;1 ar/R filter. In this study, we present functional results of amino acid substitutions in the filter as well as deletions in loops C and D of NbXIP1;1α. Furthermore, we have created and analysed homology models based on the AtTIP2;1 crystal structure to rationalize the functional properties of wild-type and mutant NbXIP1;1α.
Our characterization of the heterologously expressed NbXIP1;1α variants aims for a molecular understanding of the various XIP isoforms and of plant AQPs in general. Further studies in planta are required to elucidate the physiological role of the different clades of XIPs in the context of the whole plant.
Construction of NbXIP1;1α mutants
According to the Thermo Scientific Phusion Site-Directed Mutagenesis Kit manual, 5′- phosphorylated primers were designed and used in polymerase chain reactions employing a modified pPICZB vector containing NbXIP1;1α cDNA  as a template to generate pPICZB fragments harbouring the cDNA of NbXIP1;1α mutants. The modified pPICZB confers a His10-tag and a TEV protease cleavage site to the N-terminus of the NbXIP1;1α amino acid sequence. The primers used in the PCR and the resulting NbXIP1;1α mutants are shown in Additional file 1: Table S1 and Table 2, respectively. The resulting PCR products were circularized, and transformed into E. coli strain XL1-Blue MRF’ and the constructs were verified by sequencing of the purified plasmids.
Transformation of NbXIP1;1α mutants into Pichia pastoris
The plasmids were transformed into wild-type Pichia pastoris X-33 cells by electroporation according to the EasySelect™ Pichia Expression kit manual . The empty pPICZB was also transformed into Pichia pastoris X-33 cells as a negative control. Clones with potentially high copy numbers of NbXIP1;1α mutants were selected on YPDS (1% (w/v) yeast extract, 1% (w/v) peptone, 2% (w/v) dextrose, 1 M sorbitol) and YPD (1% (w/v) yeast extract, 1% (w/v) peptone, 2% (w/v) dextrose) agar plates containing different concentrations of zeocin as published [19, 29].
A small-scale expression screen was performed in order to analyze the expression levels in X-33 clones selected at the different antibiotic concentrations as described earlier . In brief, NbXIP1;1α mutant clones were cultured in 5 mL BMGY (1% (w/v) yeast extract, 2% (w/v) peptone, 100 mM potassium phosphate pH 6.0, 1.34% (w/v) yeast nitrogen base, 4 × 10−5% (w/v) biotin, 1% (v/v) glycerol) overnight to generate biomass. The Pichia cells were harvested and resuspended in 5 mL BMMY (1% (w/v) yeast extract, 2% (w/v) peptone, 100 mM potassium phosphate pH 6.0, 1.34% (w/v) yeast nitrogen base, 4 × 10−5% (w/v) biotin, 0.5% (v/v) methanol) to an optical density at 600 nm (OD600) of 1. Methanol was added to a final concentration of 0.5% (v/v) every 24 h to sustain protein induction. The cell cultures were incubated at 28 °C with continuous shaking at 245 rpm for 72 h. Cells corresponding to 40 OD600 units were harvested and lysed, by vortexing with glass beads, in 100 μL cold breaking buffer (50 mM NaPO4 pH 7.4, 1 mM EDTA, 5% (v/v) glycerol, 1 mM PMSF). The lysate was clarified by centrifugation and the supernatant, containing the crude cell extract was analyzed for NbXIP1;1 content by western blot.
Western blot analysis
The crude cell extracts were incubated in 3.33 × SDS loading buffer (250 mM Tris–HCl pH 6.8, 40% (v/v) glycerol, 8% (w/v) SDS, 2.4 M β-mercaptoethanol, 0.1% (w/v) bromophenol blue) for 30 min at room temperature. The proteins were separated on 12% SDS-PAGE gels and transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore). His-tagged NbXIP1;1α mutant proteins were visualized by probing with a monoclonal mouse 6xHis primary antibody (Clontech), using a horseradish peroxidase-conjugated polyclonal goat anti-mouse IgG as secondary antibody. Blots were developed by enhanced chemiluminiscence (ECL). For quantification of His-tagged NbXIP1;1α mutant proteins, Syngene PXi touch instrument was used to scan the protein signals on the membrane (see Additional file 2: Figure S1). The scanned images were analyzed with the ImageJ software .
Functional assay in Pichia spheroplasts
To test for water permeability in the NbXIP1;1α mutants, P. pastoris X-33 spheroplasts expressing NbXIP1;1α mutant proteins were prepared as previously described [19, 31]. Briefly, NbXIP1;1α protein production was induced in the transformed P. pastoris cells in BMMY with a starting OD600 = 1, as described earlier. After 26 h of induction, the cells were harvested, re-suspended and incubated in TE-buffer (100 mM Tris–HCl, pH 8.0; 1 mM EDTA) and 0.5% β-mercaptoethanol for 1.5 h to destabilize the cell wall. The cells were then washed and equilibrated in 20 mM Tris–HCl pH 8, 1.2 M sorbitol to a final OD600 of 5. Equilibrated spheroplasts were challenged with a hyperosmolar solution (20 mM Tris–HCl pH 8, 1.8 M sorbitol) and the shrinkage upon mixture was observed by increased light scattering in a stopped-flow apparatus (SF-61 DX2 Double Mixing Stopped-flow System, Hi-Tech Scientific) at 500 nm. Kinetic Studio version 2.28 (TgK Scientific Limited) was used to calculate the rate constants after fitting the average of at least 15 traces to a single exponential equation. This functional assay was repeated three times and the standard deviation of the rate constant for water permeability was calculated. To be able to compare water permeability among the NbXIP1;1α mutants, the rate constants were divided by the individual protein amounts estimated by western blot.
Unpaired t-test assuming Gaussian distribution with Welch’s correction was used for the analysis of the data in GraphPad Prism .
Homology modeling of the NbXIP1;1α was carried out with I-TASSER version 4.4  using the AtTIP2;1 structure (PDB ID 5I32)  as a template. Multiple sequence alignment of related AQPs used in the modeling was performed manually. Two alternative alignments of the helix 5 of NbXIP1;1α were submitted to the I-TASSER server, including constraints for the ICVAR variant of the ar/R selectivity filter (one letter code of residues in H2P, LCP, H5 P, LEP, HEP). Point mutations were introduced in the wild-type NbXIP1;1α (NbXIP1;1αwt) model using Coot  and the mutated residue and immediate surroundings were subsequently energy minimized using NAMD . The introduced H102 was singly protonated at Nδ. No attempt was made to model truncated loops or to further optimize the loops in the models. All models were supplemented with waters in the pore and subjected to geometry optimization using FG-MD . Validation of model geometry was done using MolProbity . The radius of the pore in the models was analysed by the program HOLE .
Design of NbXIP1;1α mutants
Structurally guided multiple sequence alignment of NbXIP1:1α to AQPs with structures deposited in the Protein Data Bank (PDB), resulted in two alternative alignments of residues in the ar/R selectivity filter of NbXIP1;1α (Fig. 2, showing last three of the five residues in the filter). Thus, it could be comprised of isoleucine, cysteine, valine, alanine and arginine (ICVAR) or isoleucine, cysteine, threonine, alanine and arginine (ICTAR; putative residue at H5P in bold) depending on how helix five was aligned. To introduce water permeability in NbXIP1;1α, two putative selectivity filters of NbXIP1;1α (I102, C175, V242/T246, A257, R263) were designed to mimic the AtTIP2;1 filter (H63, H131, I185, G194, R200; Table 1).
Due to the uncertainty in the alignment of loop C and the conservation of the residues at LEP and HEP only the two positions of the filter corresponding to H2P and H5P were considered. However, an L79G substitution in helix 1, that is not part of the selectivity filter, was added since a small amino acid residue is conserved at this position in AQPs with an aromatic amino acid residue in H2P and this appears to be a prerequisite to accommodate the introduced histidine at H2P in a TIP2;1-like orientation. Thus, two principal NbXIP1;1α mutants were constructed; NbXIP1;1αL79G/I102H/V242I and NbXIP1;1αL79G/I102H/T246I (Table 2; differing mutations putatively in H5P in bold). To exclude the possibility of poor water permeability due to occlusion of the pore by the long loop C and loop D, the corresponding NbXIP1;1α mutants with deletions in loop C and/or loop D were also generated. The truncations were guided by the two alternative alignments to achieve the same length of these loops as found in AtTIP2;1. To further investigate the contribution of the individual substitutions, six additional mutants (Table 2) were subsequently constructed based on the functional results presented below.
Heterologous expression of NbXIP1;1α mutants in Pichia pastoris
Mutant proteins of the NbXIP1;1α splice-variant were successfully expressed in P. pastoris, although at different levels. Hence, to obtain adequate protein amounts detectable by western blot, protein induction in P. pastoris was extended and more cells were used compared to the standard protocol . After this modification, crude cell extracts prepared from P. pastoris cells expressing the mutant proteins had appreciable amounts of His-tagged NbXIP1;1α mutant proteins, as shown in Fig. 3a and b. In general, protein expression was lower for constructs containing deletions and higher in cells expressing NbXIP1;1αL79G/I102H or NbXIP1;1αI102H/V242I (Additional file 2: Figure S1, Additional file 3: Table S2, Additional file 4: Table S3).
The triple mutant NbXIP1;1αL79G/I102H/V242I is permeable to water
To compare the water permeability of the yeast spheroplasts prepared from induced cells expressing the various constructs, a stopped-flow spectrometric assay was employed. As shown in Fig. 4, there was little or no difference in the rate of shrinkage between the control P. pastoris spheroplasts with the empty vector pPICZB and the spheroplasts overexpressing NbXIP1;1αwt, which suggests poor or no water permeability of the protein. The rate of shrinkage of spheroplasts expressing the mutant NbXIP1;1αL79G/I102H/T246I was only slightly increased whereas expression of NbXIP1;1αL79G/I102H/V242I resulted in a more than two-fold higher rate. The mean rate constants and standard deviations for the fitted curves were: 2.19 ± 0.25 s−1 (empty pPICZB plasmid), 2.20 ± 0.42 s−1 (NbXIP1;1αwt), 2.84 ± 0.48 s−1 (NbXIP1;1αL79G/I102H/T246I) and 5.52 ± 1.21 s−1 (NbXIP1;1αL79G/I102H/V242I). To compensate for differences in expression, the background corrected rates were related to the protein levels to obtain a specific activity in arbitrary units. The estimated protein amounts, rate constants and specific activities for the first set of NbXIP1;1α mutants are shown in Additional file 3: Table S2.
Considering the average specific activities, the NbXIP1;1αL79G/I102H/V242I triple mutant was almost 300-fold more permeable to water than the NbXIP1;1αwt. Although there is considerable variation between replicates, the corresponding relative specific rates, setting the specific rate of NbXIP1;1αL79G/I102H/V242I as unity, are significantly different as shown in Fig. 5 and Additional file 5: Table S4 (P < 0.05). In contrast, the relative specific rate of P. pastoris spheroplasts expressing NbXIP1;1αL79G/I102H/T246I was not significantly different from that of the NbXIP1;1αwt spheroplasts. However, the relative specific rate of P. pastoris spheroplasts expressing NbXIP1;1αL79G/I102H/T246I was significantly lower than that of the spheroplasts with NbXIP1;1αL79G/I102H/V242I.
To eliminate the possibility of the C and D loops blocking the pore, NbXIP1;1αL79G/I102H/V242I and NbXIP1;1αL79G/I102H/T246I mutants with loop C and/or loop D truncations were engineered. However, as shown in Fig. 5a, truncating the loops did not significantly improve the relative specific rates, as P. pastoris spheroplasts overexpressing NbXIP1;1αL79G/I102H/V242I/ΔC, NbXIP1;1αL79G/I102H/V242I/ΔD, NbXIP1;1αL79G/I102H/V242I/ΔC/ΔD, NbXIP1;1αL79G/I102H/T246I/ΔC or NbXIP1;1αL79G/I102H/T246I/ΔD showed large experimental variation with averages not significantly different from the wild type or the NbXIP1;1αL79G/I102H/T246I mutant without the loop deletions.
Single substitutions L79G and I102H are sufficient to render NbXIP1;1α water permeable
To ascertain whether all three amino acid substitutions in the NbXIP1;1αL79G/I102H/V242I mutant are required to induce water permeability in NbXIP1;1α, six additional NbXIP1;1α mutants were constructed as shown in Table 2. The estimated protein amounts, rate constants and specific activities for NbXIP1;1αwt, NbXIP1;1αL79G/I102H/V242I and the second set of NbXIP1;1α mutants were determined in the same way as for the first set and are reported in Additional file 4: Table S3. As displayed in Fig. 5b, when relating the specific rates to the triple mutant, all the other mutants showed lower specific rates that were significantly different from the triple mutant, except for NbXIP1;1αL79G/V242I (Additional file 6: Table S5). Still, all except spheroplasts overexpressing NbXIP1;1αV242I and NbXIP1;1αL79G/I102H, showed significantly higher relative specific rates than the NbXIP1;1αwt spheroplasts. Thus, even the single substitutions L79G and I102H are individually sufficient to increase the water permeability of the NbXIP1;1α.
Homology modeling and pore diameter
The wild-type NbXIP1;1α (NbXIP1;1αwt) was modeled using the AtTIP2;1 structure as a template. Even though two alternative alignments (denoted ICVAR and ICTAR; Fig. 2) were submitted to I-TASSER, only the ICTAR variant could be modeled using this approach. Regardless of which initial alignment that was used, similar NbXIP1;1α models were obtained with Thr 246 placed at the H5P position in the ar/R filter. All attempts to force the I-TASSER modeler with different constraints to place Val 242 at the H5P position proved futile, as it was energetically more favorable to place Thr 246 at the H5P position instead.
Overall the models are consistent with the general AQP structure and none of the loops appear to block the pore (Fig. 6a). Contrary to our expectation none of the models placed the highly conserved Cys 175 at the LCP position, instead the much less conserved Gly 186 was found at this position in all the models. It should however be noted that the fold and position of the long C and D loops are unreliable since no extra effort was made to optimize these regions neither in the models of the monomer nor in a tetramer.
An unexpected result of the modeling is the novel orientation of the arginine (Arg 263) of the ar/R selectivity filter. In all the models of the NbXIP1;1αwt and the mutants, Arg 263 forms a salt bridge to Asp 80 in helix 1 (Fig. 6b). Furthermore, the arginine makes hydrogen bonds to carbonyls of the backbone in loop E in all models except NbXIP1;1αL79G/V242I, in which it instead interacts with a carbonyl of the loop C backbone. As a result of the salt bridge to Asp 80 no hydrogen bonds are formed between the arginine and the introduced histidine at H2P. In the model of NbXIP1;1αL79G/I102H/T246I a TIP2-like orientation is adopted by the histidine but this is not sufficient to reorient the arginine to a similar position as found in the structure of AtTIP2;1.
The program HOLE  was used to estimate the radius of the pore in the models of NbXIP1;1αwt and the mutants (Fig. 7, Additional file 7: Figure S2). The result suggests that the pore of NbXIP1;1αL79G/I102H/T246I is most narrow followed by NbXIP1;1αI102H and NbXIP1;1αI102H/V242I, whereas the pore of the NbXIP1;1αwt and NbXIP1;1αL79G/V242I appears less restricted. The constriction in NbXIP1;1αL79G/I102H/T246I is indirectly formed by T246I since the isoleucine by steric hindrance and its hydrophobicity directs His 102 to the TIP2-like orientation at the same time as it is changing the location of loop E by interactions with Ala 257. In both NbXIP1;1αI102H and NbXIP1;1αI102H/V242I the pore radius is limited by the histidine and Leu 79.
The use of the Pichia pastoris expression system for the study of membrane proteins, especially AQPs, has gained a lot of interest in the past two decades. This is because, as a eukaryote, Pichia has the machinery to correctly express and fold eukaryotic proteins [29,40,, 39–41]. All the 13 human AQPs and a number of plant AQPs have been studied in P. pastoris [19, 29, 42, 43]. As expected from earlier studies of the NbXIP1;1αwt protein , expression of all the NbXIP1;1α mutant proteins in P. pastoris was successful. The observed differences in the expression levels of the mutants could, however, be due to differences in gene dosage of the construct encoding the mutant protein in the various Pichia clones as shown previously by Nordén et al. .
With the exception of the loop truncations which could potentially alter the overall protein fold and the stability of the protein, the mutations were not expected to introduce any adverse effect on the overall protein structure, as the substitutions constitute subtle changes in the pore of NbXIP1;1α. Since NbXIP1;1αwt was found in the P. pastoris plasma membrane in a previous study , it is likely that the NbXIP1:1α mutants are also localized to the plasma membrane of P. pastoris and therefore available for functional studies in spheroplasts.
In the present study we made an attempt to introduce water permeability in NbXIP1:1α in order to facilitate functional studies and to probe our understanding of the functional determinants in NbXIP1:1α, and this was indeed accomplished. The water permeable mutants as such support the idea that the overall AQP fold and trafficking to the plasma membrane is preserved in mutants harbouring subtle changes and allows different properties of NbXIP1:1α, like a putative gating, to be explored in a constructive way. In the same manner, predicted interactions in the ar/R filter region may be investigated by mutations to get a deeper understanding of how the selectivity can be tuned, which may be valid also for other isoforms.
Aquaporin channel gating involving loop D has been reported in the plant aquaporin SoPIP2;1 . As shown by the truncated mutants in spheroplasts assay, we found no support for our concerns that the long loop C and/or loop D could be occluding the pore. Conversely, the inability of the NbXIP1;1αL79G/I102H/V242I mutants with shortened loop C and/or D to facilitate the transport of water indicate non-functional conformations adopted by the mutants, possibly blocked by the truncated loops. It appears that the D loop is rather sensitive to mutations, as the deletion in this particular construct is only two amino acid residues long. However, it is possible that the folding in the membrane is compromised since the deletion removes a positively charged residue from the cytosolic side of the protein .
The stopped-flow spectrometric analysis of the water permeability of the NbXIP1;1α mutants in Pichia spheroplasts and the homology modeling of the NbXIP1;1α mutants with AtTIP2;1 as the template were performed concomitantly. Although it appeared from the initial mutational studies that we had succeeded in introducing water permeability by three TIP2-like substitutions in the ar/R selectivity filter, a more reasonable interpretation of our result is that V242I is a neutral mutation located outside the filter whereas T246I is positioned at H5P and not compatible with L79G/I102H and water permeability, favoring the second alternative alignment (Fig. 2b). Thus, results from both the experimental data and from the in-silico modeling support the idea that Thr 246 and not Val 242 is the amino acid residue that resides at the H5P position in the NbXIP1;1α ar/R filter. However, further studies including X-ray crystallographic structural determination of NbXIP1;1s are needed to definitively confirm this.
The NbXIP1;1αwt was impermeable to water in this study just as PtXIP1;1 and NtXIP1;1 could not facilitate the transport of water in Xenopus laevis oocytes [8, 26]. The inability of XIP1;1s to facilitate the transport of water has been largely attributed to the hydrophobic nature of their selectivity filters [7, 8]. This idea was further supported by the finding that NtXIP1;1s were permeable not only to glycerol, but also to urea and boric acid . It is not immediately clear from our model why NbXIP1;1αwt is not permeable to water. As the pore appears to be wide enough for water to pass, we assume that the presence of the hydrophobic residues (Ile 102 and Ala 257 at H2P and LEP, respectively) in the ar/R filter prevent the hydrogen bonding network needed for water permeability in the NbXIP1;1αwt.
The models of the NbXIP1;1αL79G/I102H/V242I and NbXIP1;1αL79G/I102H/T246I mutants revealed that the idea of instituting minimal changes in the NbXIP1;1α ar/R filter, to mimic the AtTIP2;1 ar/R filter in order to facilitate water permeability in NbXIP1;1α, was too simplistic. Additional substitutions seem to be needed to achieve this goal, especially since the residues at the LCP and H5P positions are not precisely known due to alternative alignment possibilities. In particular, the salt bridge between arginine in the ar/R filter and Asp 80 is of interest and relevant in other isoforms as well. Interestingly, a similar interaction between the arginine and an acidic residue at the corresponding position to Asp 80 is found in the structure of PfAQP, a glycerol facilitator of Plasmodium falciparum  (Fig. 6c). We note that an acidic residue at this position is not conserved among all groups of XIPs, e.g. in the XIPs of moss and spikemoss (XIP I; Table 3) that have a histidine or a glutamine in H2P, both of which may form a TIP2-like interaction with the arginine of the selectivity filter. Other XIPs lack both the salt bridge and a putative TIP2-like interaction between residues in H2P and HEP, suggesting that there are at least three different orientations of the arginine in HEP that are potentially tweaking the selectivity in the diverged phylogenetic groups of XIPs. There is also support for a similar modification of selectivity in TIP subgroups (Table 3). Both the TIP3s of angiosperms and the TIP1s of gymnosperms have a glutamate at the position that is aligning to Asp 80 in NbXIP1;1α. Interestingly, within the TIP1 sister clade to the TIP3s, the monocot/dicot TIP1s which all lack the arginine are also missing an acidic residue corresponding to Asp 80.
Contrary to our naïve expectations, only the ar/R filter in the NbXIP1;1αL79G/I102H/T246I mutant model resembled the AtTIP2;1 ar/R filter, although it was the NbXIP1;1αL79G/I102H/V242I mutant that was permeable to water. The latter could not only be attributed to the introduction of His 102 at the H2P position to increase the possibility of hydrogen bonding but also to the introduction of Gly 79 which makes room for water permeability in the NbXIP1;1αL79G/I102H/V242I. It was assumed that the V242I substitution in NbXIP1;1αL79G/I102H/V242I did not influence water permeability, since according to the model, it is located deep in the pore and does not form part of the ar/R filter. It is noteworthy that the Thr 246 was left untouched in the NbXIP1;1αL79G/I102H/V242I mutant and hence could provide hydrogen bonding possibilities to pore waters permeating the channel. The inability of the NbXIP1;1αL79G/I102H/T246I mutant to facilitate water permeability may therefore be attributed to the lack of Thr 246 as well as the small aperture indirectly caused by the isoleucine via His 102 and Ala 257 as mentioned in the result section.
With the exception of the NbXIP1;1αV242I mutant, all the single mutants engineered to investigate the individual contribution of the substitutions in the NbXIP1;1αL79G/I102H/V242I triple mutant were more permeable to water than NbXIP1;1αwt. As mentioned, it is not surprising that the NbXIP1;1αV242I mutant had poor or no water permeability as according to the models the NbXIP1;1αV242I ar/R filter is identical to that of the NbXIP1;1αwt. Consistently, the substitution V242I appears neutral also when combined with mutation L79G or I102H separately. However, it is intriguing that the V242I substitution still seems crucial for water permeability in the triple mutant NbXIP1;1αL79G/I102H/V242I, since the double mutant NbXIP1;1αL79G/I102H has a lower relative specific rate, which is not significantly different from that of NbXIP1;1αwt. We cannot explain this result based the models and it is possible that other factors like a proper folding of the protein and insertion in the plasma membrane come into play here, but from the recorded protein levels there is no clear pattern indicative of stability issues.
According to a recent study of the ar/R filter residues of XIPs, the residue at the H2P position varied significantly between XIP isoforms  and as such the residues at this position, could influence substrate specificity in XIPs. Our results corroborate the proposition of Ile 102 in the H2P position as the I102H substitution rendered NbXIP1;1α water permeable. The model of the NbXIP1;1αI102H mutant indicates that His 102 at the H2P position in the ar/R filter can contribute to the hydrogen bond network in the pore. However, the extended hydrogen bonding network might stabilize water molecules, making them less likely to shift position through the pore. This could be the reason that the water permeability in the NbXIP1;1αI102H mutant was not as high as in the NbXIP1;1αL79G/I102H/V242I mutant, where additional substitutions increase the radius of the pore.
Our results suggest that in addition to the five amino acid residues of the ar/R filter, other residues, especially those lining the pore, influence substrate specificity in NbXIP1;1α. Nonetheless, it was quite striking that the single amino acid substitution L79G in helix 1 made the NbXIP1;1αL79G mutant permeable to water. Superimposing the NbXIP1;1αL79G model on the NbXIP1;1αwt model revealed that Arg 263 was shifted in position in the NbXIP1;1αL79G mutant due to lack of steric hindrance as a result of the L79G substitution. It therefore appears that the new position adopted by Arg 263 and the extra space provided by the Gly 79 in the NbXIP1;1αL79G mutant are favorable for water permeability.
Subtle changes, like the single point mutations L79G or I102H, are sufficient to allow for water permeability in NbXIP1;1α. We conclude that NbXIP1;1α is not likely to be occluded when heterologously expressed in the spheroplasts and that Thr 246 most probably resides at the H5P position. Furthermore, most functional results can be explained from the models based on a combination of diameter and hydrophobicity of the ar/R filter. Our models suggest that a salt bridge between an acidic residue in a position corresponding to Asp 80 of NbXIP1;1α directs the orientation of the arginine in the ar/R filter and provides a new way to tune the selectivity of AQPs. Published structures and models, together with sequence analysis, predict that the arginine is positioned in three different orientations in various XIPs. The engineered water permeable variants will facilitate further investigations of the structural and functional properties of NbXIP1;1α in which e.g. the salt bridge, predicted residues in the LCP position, oxidation state, phosphorylation and a putative gating may be studied.
Aquaporin 8 s
Arabidopsis thaliana TIP2;1
Buffered glycerol-complex medium
Buffered methanol-complex medium
Fragment-guided molecular dynamics simulation
Glycerol facilitator-like intrinsic proteins
- H2P :
Helix 2 position
- H5P :
Helix 5 position
- HEP :
Helix E position
Hybrid intrinsic proteins
Isoleucine 102 substituted for histidine
Iterative threading assembly refinement
Leucine 79 substituted for glycine
- LCP :
Loop C position
- LEP :
Loop E position
Major intrinsic proteins
Nanoscale molecular dynamics
Nicotiana benthamiana XIP1;1α
Nicotiana benthamiana XIP1;1β
Nicotiana benthamiana XIP1;1
Nodulin 26-like intrinsic proteins
Asparagine-proline-alanine aquaporin motif
Nicotiana tabacum XIP1;1
- OD600 :
Optical density at 600 nm
Polymerase chain reaction
- PDB ID:
Protein data bank identifier
Protein data bank
Plasmodium falciparum aquaporin
Plasma membrane intrinsic proteins
Pichia pastoris vector for expression of proteins
Populus trichocarpa XIP1;1
Sodium dodecyl sulphate
Sodium dodecyl sulphate - polyacrylamide gel electrophoresis
Small basic intrinsic proteins
Spinacia oleracea PIP2;1
Threonine 246 substituted for isoleucine
Tris–HCl EDTA buffer
Tobacco etch virus
Tonoplast intrinsic protein 1 s
Tonoplast intrinsic protein 2-like
Tonoplast intrinsic protein 3 s
Tonoplast intrinsic proteins
Valine 242 substituted for isoleucine
X intrinsic protein 1
X intrinsic proteins
Yeast extract peptone dextrose medium
Yeast extract peptone dextrose medium with sorbitol
Gomes D, Agasse A, Thiebaud P, Delrot S, Geros H, Chaumont F. Aquaporins are multifunctional water and solute transporters highly divergent in living organisms. Biochim Biophys Acta. 2009;1788(6):1213–28.
Johanson U, Karlsson M, Johansson I, Gustavsson S, Sjövall S, Fraysse L, Weig AR, Kjellbom P. The complete set of genes encoding major intrinsic proteins in Arabidopsis provides a framework for a new nomenclature for major intrinsic proteins in plants. Plant Physiol. 2001;126(4):1358–69.
Chaumont F, Barrieu F, Wojcik E, Chrispeels MJ, Jung R. Aquaporins constitute a large and highly divergent protein family in maize. Plant Physiol. 2001;125(3):1206–15.
Sakurai J, Ishikawa F, Yamaguchi T, Uemura M, Maeshima M. Identification of 33 rice aquaporin genes and analysis of their expression and function. Plant Cell Physiol. 2005;46(9):1568–77.
Gupta AB, Sankararamakrishnan R. Genome-wide analysis of major intrinsic proteins in the tree plant Populus trichocarpa: characterization of XIP subfamily of aquaporins from evolutionary perspective. BMC Plant Biol. 2009;9:134.
Park W, Scheffler BE, Bauer PJ, Campbell BT. Identification of the family of aquaporin genes and their expression in upland cotton (Gossypium hirsutum L.). BMC Plant Biology. 2010;10:142.
Danielson JA, Johanson U. Unexpected complexity of the aquaporin gene family in the moss Physcomitrella patens. BMC Plant Biol. 2008;8:45.
Bienert GP, Bienert MD, Jahn TP, Boutry M, Chaumont F. Solanaceae XIPs are plasma membrane aquaporins that facilitate the transport of many uncharged substrates. Plant J. 2011;66(2):306–17.
Murata K, Mitsuoka K, Hirai T, Walz T, Agre P, Heymann JB, Engel A, Fujiyoshi Y. Structural determinants of water permeation through aquaporin-1. Nature. 2000;407(6804):599–605.
de Groot BL, Frigato T, Helms V, Grubmuller H. The mechanism of proton exclusion in the aquaporin-1 water channel. J Mol Biol. 2003;333(2):279–93.
Fu D, Libson A, Miercke LJ, Weitzman C, Nollert P, Krucinski J, Stroud RM. Structure of a glycerol-conducting channel and the basis for its selectivity. Science. 2000;290(5491):481–6.
Sui HX, Han BG, Lee JK, Walian P, Jap BK. Structural basis of water-specific transport through the AQP1 water channel. Nature. 2001;414(6866):872–8.
Zhu FQ, Tajkhorshid E, Schulten K. Collective diffusion model for water permeation through microscopic channels. Physical Review Letters. 2004;93(22):224501.
Kirscht A, Kaptan SS, Bienert GP, Chaumont F, Nissen P, de Groot BL, Kjellbom P, Gourdon P, Johanson U. Crystal structure of an ammonia-permeable aquaporin. PLoS Biol. 2016;14(3):e1002411.
Sade N, Vinocur BJ, Diber A, Shatil A, Ronen G, Nissan H, Wallach R, Karchi H, Moshelion M. Improving plant stress tolerance and yield production: is the tonoplast aquaporin SlTIP2;2 a key to isohydric to anisohydric conversion? New Phytol. 2009;181(3):651–61.
Reuscher S, Akiyama M, Mori C, Aoki K, Shibata D, Shiratake K. Genome-wide identification and expression analysis of aquaporins in tomato. Plos One. 2013;8(11):e79052.
Ariani A, Gepts P. Genome-wide identification and characterization of aquaporin gene family in common bean (Phaseolus vulgaris L.). Mol Genet Genomics. 2015;290(5):1771–85.
Venkatesh J, Yu JW, Gaston D, Park SW. Molecular evolution and functional divergence of X-intrinsic protein genes in plants. Mol Gen Genomics. 2015;290(2):443–60.
Ampah-Korsah H, Anderberg HI, Engfors A, Kirscht A, Norden K, Kjellström S, Kjellbom P, Johanson U. The aquaporin splice variant NbXIP1;1alpha is permeable to boric acid and is phosphorylated in the N-terminal domain. Front Plant Sci. 2016;7:862.
Anderberg HI, Kjellbom P, Johanson U. Annotation of Selaginella moellendorffii major intrinsic proteins and the evolution of the protein family in terrestrial plants. Front Plant Sci. 2012;3:33.
Abascal F, Irisarri I, Zardoya R. Diversity and evolution of membrane intrinsic proteins. Biochim Biophys Acta. 2014;1840(5):1468–81.
Maurel C, Reizer J, Schroeder JI, Chrispeels MJ. The vacuolar membrane protein gamma-TIP creates water specific channels in Xenopus oocytes. EMBO J. 1993;12(6):2241–7.
Jahn TP, Moller AL, Zeuthen T, Holm LM, Klaerke DA, Mohsin B, Kuhlbrandt W, Schjoerring JK. Aquaporin homologues in plants and mammals transport ammonia. FEBS Lett. 2004;574(1–3):31–6.
Loque D, Ludewig U, Yuan L, von Wiren N. Tonoplast intrinsic proteins AtTIP2;1 and AtTIP2;3 facilitate NH3 transport into the vacuole. Plant Physiol. 2005;137(2):671–80.
Dynowski M, Mayer M, Moran O, Ludewig U. Molecular determinants of ammonia and urea conductance in plant aquaporin homologs. FEBS Lett. 2008;582(16):2458–62.
Lopez D, Bronner G, Brunel N, Auguin D, Bourgerie S, Brignolas F, Carpin S, Tournaire-Roux C, Maurel C, Fumanal B, et al. Insights into Populus XIP aquaporins: evolutionary expansion, protein functionality, and environmental regulation. J Exp Bot. 2012;63(5):2217–30.
Törnroth-Horsefield S, Wang Y, Hedfalk K, Johanson U, Karlsson M, Tajkhorshid E, Neutze R, Kjellbom P. Structural mechanism of plant aquaporin gating. Nature. 2006;439(7077):688–94.
Invitrogen: EasySelect Pichia Expression Kit,version G. In.; 2005.
Norden K, Agemark M, Danielson JA, Alexandersson E, Kjellbom P, Johanson U. Increasing gene dosage greatly enhances recombinant expression of aquaporins in Pichia pastoris. BMC Biotechnol. 2011;11:47.
Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nat Methods. 2012;9(7):671–5.
Fischer G, Kosinska-Eriksson U, Aponte-Santamaria C, Palmgren M, Geijer C, Hedfalk K, Hohmann S, de Groot BL, Neutze R, Lindkvist-Petersson K. Crystal Structure of a Yeast Aquaporin at 1.15 ångstrom Reveals a Novel Gating Mechanism. Plos Biology. 2009;7(6):e1000130.
GraphPad Prism version 6.0a for Mac OS X. In. La Jolla California USA: GraphPad Software; 2012.
Yang J, Yan R, Roy A, Xu D, Poisson J, Zhang Y. The I-TASSER Suite: protein structure and function prediction. Nat Methods. 2015;12(1):7–8.
Emsley P, Lohkamp B, Scott WG, Cowtan K. Features and development of Coot. Acta Crystallogr D Biol Crystallogr. 2010;66(Pt 4):486–501.
Phillips JC, Braun R, Wang W, Gumbart J, Tajkhorshid E, Villa E, Chipot C, Skeel RD, Kale L, Schulten K. Scalable molecular dynamics with NAMD. J Comput Chem. 2005;26(16):1781–802.
Zhang J, Liang Y, Zhang Y. Atomic-level protein structure refinement using fragment-guided molecular dynamics conformation sampling. Structure. 2011;19(12):1784–95.
Chen VB, Arendall III WB, Headd JJ, Keedy DA, Immormino RM, Kapral GJ, Murray LW, Richardson JS, Richardson DC. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr Sect D. 2010;66(1):12–21.
Smart OS, Neduvelil JG, Wang X, Wallace BA, Sansom MS. HOLE: a program for the analysis of the pore dimensions of ion channel structural models. J Mol Graph. 1996;14(6):354–60. 376.
Karlsson M, Fotiadis D, Sjövall S, Johansson I, Hedfalk K, Engel A, Kjellbom P. Reconstitution of water channel function of an aquaporin overexpressed and purified from Pichia pastoris. Febs Lett. 2003;537(1–3):68–72.
Horsefield R, Norden K, Fellert M, Backmark A, Törnroth-Horsefield S, Terwisscha van Scheltinga AC, Kvassman J, Kjellbom P, Johanson U, Neutze R. High-resolution x-ray structure of human aquaporin 5. Proc Natl Acad Sci U S A. 2008;105(36):13327–32.
Moparthi L, Survery S, Kreir M, Simonsen C, Kjellbom P, Högestätt ED, Johanson U, Zygmunt PM. Human TRPA1 is intrinsically cold- and chemosensitive with and without its N-terminal ankyrin repeat domain. Proc Natl Acad Sci U S A. 2014;111(47):16901–6.
Agemark M, Kowal J, Kukulski W, Norden K, Gustavsson N, Johanson U, Engel A, Kjellbom P. Reconstitution of water channel function and 2D-crystallization of human aquaporin 8. BBA Biomembranes. 2012;1818(3):839–50.
Öberg F, Hedfalk K. Recombinant production of the human aquaporins in the yeast Pichia pastoris. Mol Membr Biol. 2013;30(1):15–31.
Sipos L, von Heijne G. Predicting the topology of eukaryotic membrane proteins. Eur J Biochem. 1993;213(3):1333–40.
Newby ZE, O’Connell 3rd JD, Gruswitz F, Hays FA, Harries WE, Harwood IM, Ho JD, Lee JK, Savage DF, Miercke LJ, et al. A general protocol for the crystallization of membrane proteins for X-ray structural investigation. Nat Protoc. 2009;4(5):619–37.
Omasits U, Ahrens CH, Muller S, Wollscheid B. Protter: interactive protein feature visualization and integration with experimental proteomic data. Bioinformatics. 2014;30(6):884–6.
We would like to thank Adine Karlsson for all help and support.
Carl Tryggers Stiftelse, Stiftelsen Olle Engkvist Byggmästare and the Swedish Research Council (VR) are gratefully acknowledged for financial support.
Availability of data and materials
The generated models presented in this article are available in the Model Archive repository via http://modelarchive.org/, using unique persistent identifiers. Wild type: 10.5452/ma-afgaz; Mutant 1: 10.5452/ma-arjyb; Mutant 2: 10.5452/ma-adeyx; Mutant 3: 10.5452/ma-a4od5; Mutant 4: 10.5452/ma-aehn9; Mutant 5: 10.5452/ma-acx2s; Mutant 6: 10.5452/ma-ajfz7; Mutant 7: 10.5452/ma-abd3g; Mutant 8: 10.5452/ma-a3cbx. See Fig. 7 legend for description of Mutant 1–8.
UJ, PK, HA-K, YS, AE and AK participated in the design of the study, HA-K constructed the NbXIP1;1 mutant P. pastoris clones, expressed the constructs, carried out the immunoblot assays, prepared spheroplasts and performed the functional studies. HA-K did all statistical analyses. HA-K and YS designed the homology modeling study, YS performed the homology modeling, UJ, YS and HA-K took part in the analysis of the homology modeling data. HA-K drafted the manuscript and UJ, PK, HA-K, YS, AE and AK was involved in the revision of the manuscript. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Consent for publication
Ethics approval and consent to participate
Primers used in the PCR reactions. (PDF 45 kb)
Expression of N-terminally His-tagged NbXIP1;1α mutants in P. pastoris. Western blots showing the expression levels of N-terminally His-tagged NbXIP1;1 mutants in P. pastoris X-33 clones. Blots were developed by enhanced chemiluminiscence in a Syngene PXi touch instrument. a. The first set of NbXIP1;1 α mutants. NbXIP1;1αst is an N-terminally truncated construct of NbXIP1;1α used as control for the western blot. b. The second set of NbXIP1;1α mutants. The control is 2 μg of purified NbXIP1;1αwt. (PDF 207 kb)
Estimated protein amounts, rate constants and specific activities for the first set of NbXIP1;1α mutants. (PDF 88 kb)
Estimated protein amounts, rate constants and specific activities for the second set of NbXIP1;1α mutants. (PDF 87 kb)
Results of unpaired t-test for comparisons indicated in Fig. 5a. (PDF 44 kb)
Results of unpaired t-test for comparisons indicated in Fig. 5b. (PDF 45 kb)
Cartoon representations of the homology models of NbXIP1;1αwt and NbXIP1;1α mutants. The pores in the models are shown in mesh representation from blue (widest) to red (narrowest). The HOLE program  was used to estimate the radius of the pore in the models. Starting from the top left to the bottom right; NbXIP1;1αwt, NbXIP1;1αL79G/I102H/V242I (mutant 1), NbXIP1;1αL79G/I102H/T246I (mutant 2), NbXIP1;1αL79G (mutant 3), NbXIP1;1αI102H (mutant 4), NbXIP1;1αV242I (mutant 5), NbXIP1;1αL79G/I102H (mutant 6), NbXIP1;1αI102H/V242I (mutant 7) and NbXIP1;1αL79G/V242I (mutant 8), respectively. See also Fig. 7. (PDF 187 kb)
About this article
Cite this article
Ampah-Korsah, H., Sonntag, Y., Engfors, A. et al. Single amino acid substitutions in the selectivity filter render NbXIP1;1α aquaporin water permeable. BMC Plant Biol 17, 61 (2017). https://doi.org/10.1186/s12870-017-1009-3
- Homology modeling
- Major Intrinsic Proteins
- Boric acid