The roles of selectivity filters in determining aluminum transport by AtNIP1;2

ABSTRACT Aquaporins (AQPs) are channel proteins involved in transporting a variety of substrates. It has been proposed that the constriction regions in the central pores of the AQP channels play a crucial role in determining transport substrates and activities of AQPs. Our previous results suggest that AtNIP1;2, a member of the AQP superfamily in Arabidopsis, facilitates aluminum transport across the plasma membrane. However, the functions of the constriction regions in AtNIP1;2-mediated transport activities are unclear. This study reports that residue substitutions of the constriction regions affect AtNIP1;2-mediated aluminum uptake, demonstrating the critical roles of the constriction regions for transport activities. Furthermore, a constriction region that partially or wholly mimics AtNIP5;1, a demonstrated boric-acid transporter, could not render the boric-acid transport activity to AtNIP1;2. Therefore, besides the constriction regions, other structural features are also involved in determining the nature of AtNIP1;2’s transport activities. Abbreviations: AIAR: alanine-isoleucine-alanine-arginine; AIGR: alanine-isoleucine-glycine- arginine; AQP: aquaporin; Al-Mal: aluminum-malate; ar/R: aromatic/arginine; AVAR: alanine-valine-alanine-arginine; CK: control; H: helical domain; ICP-MS: inductively coupled plasma mass spectrometry; LA - LE: inter-helical loops A to E; NIP: nodulin 26-like intrinsic protein; NPA: asparagine-proline-alanine; NPG: asparagine-proline- glycine; NPS: asparagine-proline-Serine; NPV: asparagine-proline-valine; ORF: open reading frame; PIP: plasma membrane intrinsic proteins; SIP: small basic intrinsic proteins; TM: transmembrane helices; WIAR: tryptophan-isoleucine-alanine-arginine; WVAR: tryptophan-valine-alanine-arginine; WVGR: tryptophan-valine-glycine- arginine.


Introduction
Aquaporins (AQPs) are channel proteins generally believed to facilitate the permeation of water and small uncharged solutes across the plasma and intracellular membranes. 1 However, increasing evidence implicates that some AQP members potentially transport ions in plants. 2 Aquaporins exhibit highly conserved structural features. 3 Firstly, four AQP monomers form a biologically active tetramer embedded in cell membranes. 4,5 Secondly, each monomer contains an active pore region surrounded by six transmembrane helical domains (H1-H6) connected by five inter-helical loops, i.e., loop A to loop E (LA-LE). 5 Thirdly, two major constrictions in the pore are thought to play critical roles in the functional specialization of AQPs. 6 The pore's first constriction comprises two highly conserved asparagine-proline-alanine (NPA) motifs in the hydrophobic LB and LE. [6][7][8] Structurally, the asparagine (N) residues in the two NPA motifs fold back into the core of the protein to form one of the significant constrictions. 3,7,9 The second constriction is located near the extracellular end of the pore, designated as an ar/R (aromatic/arginine) region due to the high prevalence of aromatic and basic residues. 7 The ar/R region is comprised of four residues, one each from the helix 2 (H2) and helix 5 (H5) and two from the loop E (LE1 and LE2). 7,10,11 It has been postulated that the NPA motifs function as a primary filter against protons and other positive ions. 6,7,12,13 Structural and functional studies indicated that the NPA motif's polar asparagine (N) residues are involved in hydrogenbonding interactions with transport substrates. 7,14 For instance, the hydrogen-bonding interactions between the N residues of the NPA motifs and water molecules are essential for maintaining the connectivity of water flow in the pore of the AQP-1 water channel. 14 Hence, replacing the N residue with hydrophobic residues caused completely broken aqueous pathways of the AQP-1 channel. 14 In contrast, the ar/R constriction is proposed as the primary filter for substrate selectivity. 8,10,11,15 Furthermore, the highly conserved Arg (R) residue at the H2 position forms a hydrophilic surface and facilitates hydrogen bonding with the transport substrates. 8,16 Thus, the R residue of the ar/R tetrad is critical for substrate selectivity in some AQP members. 17,18 For instance, an exchange of the R residue altered substrate selectivity for some TIP members. 19 Based on their sequence similarity, plant AQPs can be classified into four subfamilies: plasma membrane intrinsic proteins (PIPs), tonoplast intrinsic proteins (TIPs), nodulin26like intrinsic proteins (NIPs), and small basic intrinsic proteins (SIPs). 20 The NIP subfamily is unique to plants, and its members facilitate transporting a diversity of substrates, including water, glycerol, 21,22 lactic acid, 23 urea, 24 formamide, 24 silicic acid, 25 selenite, 26 aluminum, 27 and metalloids such as arsenite (As(III)), antimonite (Sb(III)), and boron (B). [28][29][30][31][32][33] Arabidopsis has nine NIP members that can be further divided into two subgroups, i.e., NIP-I and NIP-II, based on their different ar/R tetrad patterns. 8,9,34 The NIP-I subgroup has six members, i.e., NIP1;1, NIP1;2, NIP2;1, NIP3;1, NIP4;1, and NIP4;2, with a conserved tetrad pattern of Trp (W) at H2, Val/Ile (V/I) at H5, Ala (A) at LE1, and Arg (A) at LE2. Moreover, the NPA-I members have an invariant NPA triad for the NPA1 motif, but variant NPA, NPG, or NPV triads in the NPA2 motif. 34 In contrast, three NIP-II members, i.e., NIP5;1, NIP6;1, and NIP7;1, have a similar tetrad pattern except that a smaller Ala (A) residue replaces the bulkier Trp (W) in the first residue of the ar/R tetrad, which results in a broader ar/R constriction that accommodates for transporting larger substrates. 8 Furthermore, it has been functionally demonstrated that the NPA-I members could transport formamide and glycerol besides water. However, NPA-IIs show very low water permeability but can transport urea and metalloids. 24,35 Our recent studies suggest that AtNIP1;2, an NPA-I member, mediates the permeation of aluminum (Al), possibly in the form of aluminum-malate (Al-Mal) complexes, across the plasma membrane (PM) in Arabidopsis. 27 AtNIP1;2 facilitates Al removal from the cell wall into the cytosol in the root-tip region and subsequent root-to-shoot translocation, which are critical steps for an internal Al-resistance mechanism in plants. 27,36,37 Furthermore, we have demonstrated that the Alactivated and AtALMT1-mediated malate release into the root cell wall 38-40 is a prerequisite for the AtNIP1;2-mediated function and resistance in Arabidopsis. 41 However, the roles of the NPA motifs and ar/R selectivity filter in AtNIP1;2's functions are still unclear.
AtNIP5;1 is the best-characterized NIP-II member that encodes a channel protein responsible for permeation of boric acid (B) into roots under B limitation. 33,42 In this study, we tested the roles of the NPA and ar/R constriction regions in determining substrate selectivity and transport activity of AtNIP1;2. Through a site-directed mutagenesis approach, critical residues in the constriction regions of AtNIP1;2 were subject to chemical nature changes, e.g., polar to nonpolar, or conversions to AtNIP5;1-like patterns. Evaluation of substrate selectivity and transport activity suggested that the constriction regions play critical roles in AtNIP1;2-mediated Al selectivity and transport. Furthermore, AtNIP5;1-like NPA and ar/R constriction regions are insufficient to render B transport activities to AtNIP1;2.

Site-directed mutagenesis
The open reading frame (ORF) of AtNIP1;2 was amplified by primers NIP1;2-F, 5ʹ-CTACggatccAAAATGGCGGAGATCTCGGGAAA-3ʹ and NIP1;2-R, 5ʹ-ATCCgcggccgcACGAGAGCTACCGTTTCGCA -3ʹ (the underlined sequences are restriction enzyme sites for BamH I and Not I, respectively). The PCR product was cut with restriction enzymes BamH I and Not I and cloned into the yeast expression vector, pYES2, to create a pYES2-AtNIP1;2 plasmid. AtNIP1;2 mutants were generated by site-directed mutagenesis using the following synthetic oligonucleotide primers. The lower case letters in the primer sequences represent mismatched nucleotides that introduced single amino-acid substitutions in the translated AtNIP1;2 proteins.
For NPA modification of AtNIP1;2: For ar/R modification of AtNIP1;2: In brief, high-fidelity PCR was performed to amplify the pYES2-AtNIP1;2 plasmid with Pfu DNA polymerase and primer pairs listed above. The resulting PCR products were checked by agarose gel electrophoresis. PCR products with the right size of 6.7 kb were digested with DpnI at 37°C for 3 h, which cut the methylated pYES2-AtNIP1;2 template into small fragments but left the non-methylated and circular PCR products intact. After DpnI digestion, the PCR products were transformed into competent E. coli strain TOP10. The purified plasmids from the transformed TOP10 cells were verified by sequencing. Additional runs of PCR-based site-directed mutagenesis were performed to generate multiple mutations.

Yeast Al sensitivity and uptake analysis
For Al sensitivity evaluation, the pYES2 empty vector or pYES2 carrying wild-type or mutant AtNIP1;2 open reading frames (ORFs) were transformed into the yeast (Sacchromyces cerevisea) strain BY4741. The resultant lines were first cultured in a liquid SD-Ura medium to the stationary phase. Cells were collected by centrifuge at 5,000 g for 5 min, followed by wash 3 times with ddH 2 O and 3 times with a low pH, low magnesium (LPM) medium, buffered with 5 mM Succinic acid to pH 4 For the drop assay, 5 μl of 10-fold serially diluted cell suspensions were spotted onto solid LPM plates (pH to 4.2), containing 0, 100, or 200 μM Al-Mal (1:2) and 2% galactose for induction of the GAL promoter. The plates were photographed after incubation at 30°C for 4 days.
For determining Al and B contents, yeast cells were cultured in an LPM liquid medium (-Ura, +2% galactose and 1% raffinose, pH 4.2) to a mid-exponential phase. Cells were harvested by centrifuge at 5,000 g for 5 min, followed by 3 time washes with an LPM medium (-Ura, +2% galactose, 1% raffinose). After washing, the cells were transferred to a new LPM medium (2% galactose, pH 4.2 or 7.0 adjusted by 5 mM succinic acid) to an OD600 value at 3.0. Next, Al-malate or H 3 BO 3 was added to the cell culture to a final concentration of 50 μM at pH 4.2 or 7.0, respectively. After 2 h incubation with gentle shaking, cells were harvested by centrifuge at 5000 × g for 5 min and washed 3 times with deionized water (ddH 2 O) (MilliQ; Millipore), dried, and then digested with 2 N HCl. The Al and B contents of each digested sample were determined by inductively coupled plasma mass spectrometry (ICP-MS) using an Agilent 7500 Series ICP mass spectrometer. Three biological replicates were conducted.

Influence of the NPA and ar/R constrictions on AtNIP1;2-mediated aluminum sensitivity in yeast
The asparagine (N) residue in the NPA1 and NPA2 motifs and the arginine (R) residue in the ar/R selective filter are conserved in the NIP subfamily. 34 To investigate the role of these conserved residues in AtNIP1;2-mediated Al-uptake activities, we constructed three AtNIP1;2 single mutants, designated as N111L, N230L, and R233G. In these mutants, the polar N residue of NPA1 and NPA2 was replaced by a nonpolar leucine (L), while the R residue of the ar/R tetrad was replaced by glycine (G).
Aluminum uptake mediated by the wild-type and mutant AtNIP1;2 proteins were investigated by assessing yeast [S. cerevisiae (BY4741)] sensitivity to Al toxicity. Yeast lines harboring an empty pYES2 vector (control, CK) or expressing AtNIP1;2 and its mutants were subjected to Al treatment on low pH, low magnesium (LPM) agar plates (pH 4.2) (Figure 2).
The growth of yeast cells transformed with the empty pYES2 vector or pYES2 containing the wild-type or mutant AtNIP1;2 could not be distinguished on the LPM plates that were not supplemented with Al ( Figure 2). This result indicated that heterologous expression of AtNIP1;2 or its mutants has no harmful effect on yeast growth under standard conditions. However, yeast cells that expressed wild-type AtNIP1;2 showed significantly reduced growth compared to those carrying the empty vector when exposed to Al stresses by adding 100 or 200 μM Al-malate (Al-Mal) to the growth medium ( Figure 2). The growth inhibition was presumed to be due to AtNIP1;2-mediated Al uptake and accumulation, as Wang et al. (2017) reported previously.
In contrast, N111L, N230L, and R233G grew similarly to the empty vector control (CK) line under -and + Al treatments ( Figure 2). These results indicated mutations in the conserved N and R residues on the NPA motifs and the ar/R region, respectively, abolished AtNIP1;2-mediated Al sensitivity in yeast (Figure 2), suggesting critical roles of these residues in AtNIP1;2-mediated Al-Mal uptake and accumulation.

Importance of the NPA and ar/R constriction regions for AtNIP1;2-mediated Al uptake
To confirm the differences in Al sensitivity were associated with Al uptake and accumulation in the yeast cells, shortterm (2 h) Al uptake was evaluated for individual yeast lines. As indicated in Figure 3, the yeast line harboring the native AtNIP1;2 construct had an ~3-fold higher Al uptake rate than the control (CK) line. This result is consistent with our previous observation that AtNIP1;2 facilitates across-PM Al uptake in yeast. 27 In contrast, the Al uptake in the yeast lines expressing AtNIP1;2 mutants (N111L, N230L, or R233G) was comparable to those of the CK line but not the native AtNIP1;2 line ( Figure 3), indicating that that the mutations caused a loss of AtNIP1;2-mediated Al-Mal uptake in yeast. Thus, the asparagine (N) residue in the NPA motifs and the arginine (R) residue of the ar/R selective filter are critical for AtNIP1;2-mediated across-membrane Al uptake.

Influence of the NPA and ar/R constrictions on substrate selectivity of AtNIP1;2
Except for the conserved 1 st residues of N in NPA1 and NPA2 motifs and the 4 th residue of R in the ar/R region, other residues show more diversity between the NIP-I and NIP-II subgroups. 34  . Influence of N/L substitution in the NPA motifs and R/G substitution in the ar/R selectivity filter on AtNIP1;2-mediated Al uptake. Yeast (BY4741) lines carrying the empty control (CK) vector pYES2 and expressing the native or mutated (i.e., N111L, N230L, and R233G) AtNIP1;2 were subject to short-term (2 h) Al uptake assays. Data are means ± SD (n = 3). Different letters above the columns indicate statistically significant differences at P < .05 by Tukey's test.

e1991686-4
For instance, AtNIP1;2, a NIP-I member, and AtNIP5;1, a NIP-II member, differ in the 3 rd residues of the NPA motifs and the first three residues of the ar/R tetrad (Figure 1).
To investigate the impact of the identity of pore constriction regions on uptake activities and substrate selectivity, the NPA motifs and ar/R selective filter of AtNIP1;2 were exchanged to partially or entirely mimic those in AtNIP5;1, a demonstrated B transporter of the NIP-II subgroup. 33 As indicated in Figure 4a, a single alanine-to-serine (A113S) exchange at the 3 rd residue of the NPA1 motif (NPA to NPS) abolished the AtNIP1;2-mediated Al-Mal transport activity (Figure 4a). This result indicates that the third residue of S is as critical as the first residue of N in the NPA1 motif for the AtNIP1;2-facilitated Al uptake (Figure 3). In contrast, a glycineto-valine (G232V) exchange at the 3 rd residue of the NPA2 motif had a much weaker impact on Al uptake: Al uptake activity decreased ~40% compared with the native AtNIP1;2 ( Figure 4a). Furthermore, the double AtNIP1;2 mutant (A113S/G232V), carrying NPA1/NPA2 motifs mimicked to those of AtNIP5;1 (NPS/NPV), also showed no increased Al transport activities (Figure 4a). This result indicates that the effect of A113S overrides that of G232V. Similarly, a single AVAR mutant, which contained a tryptophan-to-alanine (W91A) substitution that mimicks the first residue of the ar/R tetrad in AtNIP5;1, did not show any Al-uptake activity compared with the control line ( Figure 4b).
The strong effects of the A-to-S in NPA1 and W-to-A in the ar/R tetrad on AtNIP1;2-mediated Al transport could be because of the serine's hydroxyl group and tryptophan indole group affecting the folding and orientation of the channel protein. Or, the A-to-S substitution at NPA1 may affect posttranslational modification, e.g., phosphorylation. 43,44 Aluminum-uptake phenotypes differed in the other two single mutant lines, i.e., WIAR and WVGR (Figure 4b). The WIAR and WVGR lines contained valine-to-isoleucine (V218I) and alanine-to-glycine (A227G) exchanges that mimicked the 2 nd and 3 rd residues of the ar/R tetrad in AtNIP5;1. As shown in Figure 4b, Al uptake decreased ~38% in the WIAR and WVGR mutant lines than the wild-type (native) AtNIP1;2 ( Figure 4b). These results suggest that the 2 nd and 3 rd residues have fewer impacts on the AtNIP1;2-mediated Al uptake than the 1 st (Figure 4b) and 4 th (Figure 3) residues of the ar/R tetrad. a b Figure 4. Impacts of third residue substitutions in the NPA motifs and residue substitutions in the ar/R selectivity filter on AtNIP1;2-mediated Al uptake. Yeast (BY4741) lines carrying the control (CK) empty vector pYES2 and the native (NPA/ NPG/WVAR) or mutated AtNIP1;2 were subject to short-term (2 h) Al-uptake assays. (a) Red letters indicate substitutions of the third residue in the NPA1 and NPA2 motifs. (b) Red letters indicate residue substitutions of the ar/R tetrad. Data are means ± SD (n = 3). Different letters above the columns indicate statistically significant differences among groups at P < .05 by Tukey's test. a b Figure 5. AtNIP5;1-like NPA motifs and ar/R selectivity filter does not render B uptake activity to AtNIP1;2. Yeast (BY4741) lines carrying the control (CK) empty vector pYES2 or expressing AtNIP1;2 containing the native or mutated NPA and ar/R constriction regions were subject to short-term (2 h) B uptake assays. (a) Red letters indicate residue substitutions in the ar/R tetrad. (b) Residue substitutions (red letters) that transform the AtNIP1;2 NPA and ar/R constriction regions to an AtNIP5;1 type. Data are means ± SD (n = 3). Different letters above the columns indicate statistically significant differences at P < .05 by Tukey's test.
Furthermore, a double AIAR mutant and a triple AIGR mutant contained W91A/V218I and W91A/V218I/A227G residue substitutions partially and wholly mimicked the AtNIP5;1 ar/R tetrad. Both AIAR and AIGR lines resembled the AVAR single mutant in Al-uptake activities (Figure 4b). This result indicates that the tryptophan (W) residue at H2 has a dominant effect on influencing AtNIP1;2-mediated Al uptake over the valine (V) and alanine (A) residues at H5 and LE1, respectively.
AtNIP5;1-like constriction regions could not render boron transport activity to AtNIP1;2 To test whether an ar/R tetrad that mimics AtNIP5;1 can render B transport activity to AtNIP1;2, tryptophan (W) at H2, valine (V) at H5, or alanine (A) at LE1 in AtNIP1;2 were changed individually or in combination to alanine (A), isoleucine (I), or glycine (G), respectively (Figure 5a). No B uptake activities were observed for the native AtNIP1;2 and its single, double, or triple ar/R mutants (Figure 5a).
We then generated a quintuple mutant (NPS/NPV/AIGR, the underlines indicate residue substitutions) where the NPA motifs and ar/R selective filter mimics AtNIP5;1. However, as shown in Figure 5b, this AtNIP1;2 quintuple mutant could not transport B also.
These results indicate that 1) boric acid is not a transport substrate for AtNIP1;2 as the native AtNIP1;2 is incapable of transporting boric acid; 2) partial or complete mimics of the NPA motifs and the ar/R tetrad of AtNIP5;1 could not convert AtNIP1;2 from an Al to a B transporter.
In conclusion, the substrate selectivity of AtNIP1;2 is not simply controlled by the NPA motifs and ar/R selectivity filter. Other structural features and post-translational modifications, e.g., phosphorylation, 43,44 methylation and acetylation, 45 and glycosylation, 46 may also be necessary for determining the nature of the substrate specificity and transport activities.