Fig 1.
(A) A linear structural model of the human ROMK1 monomer. Residues important for potassium transport in the pore helix (yellow) and residues tested in this study (Groups 1–6) are highlighted. ROMK1 shares a common tetrameric architecture with other inward rectifying (Kir) potassium channels: two transmembrane (TM) domains TM1 and TM2, a conserved potassium selectivity filter and cytoplasmic N- and C- terminal domains. The linear model was generated using Protter [52] (UniProt accession no. P48048). (B) A ROMK1 homology model (aa 38–365) shows the oligomerization of four ROMK1 subunits to form a functional channel with a central pore through which the potassium ions pass. The homology model was built based on the crystal structure of Kir2.2 (PDB ID 3SPG), which is 47.4% identical to ROMK1 (which is well within the range suitable for accurate comparative modeling). Images were rendered using PyMOL (v2.1.0). (C) The ROMK1 potassium selectivity filter contains the indicated “T141IGYG145” motif. Main-chain oxygens in this motif line the pore and facilitate potassium entry. V140, which is located between the selectivity filter and the pore-lining helix TM2, contributes to single channel conductance and barium block [28]. In TM1, K80 is important for the interaction between TM1 and TM2 and controls channel gating [5, 34]. (D) The cytoplasmic extended pore includes residues D254, E258, N259, and D298. Panels C and D display portions of two subunits. The other two were removed for visual clarity.
Fig 2.
In silico saturation mutagenesis analysis of ROMK variants.
The predicted pathogenicity probabilities obtained with Rhapsody are shown in a heat map (divided in two parts) with a color code ranging from red (deleterious, >0.5) to blue (neutral, ≤0.5). Light yellow entries correspond to wild-type (WT) amino acids. The bottom panels corresponding to each heat map represent the residue-averaged pathogenicities predicted by Rhapsody (red), EVmutation (green) and PolyPhen-2 (blue). The pink shade indicates the range of Rhapsody results for the 19 specific substitutions for each residue. The locations of 31 variants selected for experimental testing and two controls are marked by black squares on the map, and callouts colored by the measured phenotype, with the same color scheme as in Figs 3 and 5. On the right, the residue-averaged predictions from Rhapsody are displayed for a single monomer in the human ROMK1 homology model using the same color code as in the map.
Fig 3.
Scatter plots comparing computational predictions and identifying consensus and discordant data.
The two scatter plots compare the predictions from Rhapsody and EVmutation (left) and those from Rhapsody and PolyPhen-2 (right), allowing the outputs to be grouped in three different categories: consensus neutral, consensus deleterious, and discordant, as indicated by the labels. Cutoff values between neutral/deleterious predictions for each method are represented by dashed lines. Note that EVmutation’s ΔE score anticorrelates with the expected pathogenicity of variants. The 31 variants selected for experimental validation and two controls (see text, Fig 5 and S1 Table) are labelled in the two plots and marked by different symbols and colors based on the experimentally observed phenotypes. Results for those variants which cannot be evaluated using EVmutation, due to the absence of a suitable Pfam domain and/or MSA, are shown in the right plot only, with abscissa values based on PolyPhen-2 scores. Labels written in square brackets refer to rat ROMK1 variants that were experimentally tested after substituting for the counterparts in the human sequence.
Fig 4.
A yeast-based assay to assess the activity of a heterologously expressed potassium channel.
(A) Schematic of the yeast-based assay. A yeast strain lacking its endogenous potassium transporters, Trk1 and Trk2, is unable to grow on medium containing low potassium but can be rescued by the expression of a human potassium channel. (B) Controls for yeast viability assays on solid (top panel) and in liquid (bottom panel) medium. Yeast cultures were transformed with an empty expression vector as a negative control, or with a plasmid expressing Kir2.1, WT ROMK1, or ROMK1 with the V140M or K80M mutation. Kir2.1, ROMK1 V140M and K80M were used as controls, and V140M and K80M exhibit increased growth compared to cells expressing the WT channel. For the viability assay on solid medium (top panel), saturated overnight cultures of yeast were serially diluted and spotted on SC-Leu medium supplemented with dextrose and containing the indicated concentration of potassium. Plates shown were imaged after a two-day incubation at 30°C and are representative of three independent experiments. For the viability assay in liquid medium (bottom panel), yeast grown overnight to saturation were diluted to an OD600 of 0.20 with medium supplemented with 25mM KCl. OD600 readings were recorded, normalized to wells containing only medium, and these values were subtracted from the reading at t = 0 (see Materials and Methods). Graphs were made using GraphPad Prism (v8. 1. 2), and data represent results from four independent experiments (n = 2–3 each), ± the range of the data. The growth of K80M on 100mM KCl is reduced because K80M enhances potassium uptake and intracellular potassium concentration, which is toxic.
Fig 5.
Growth phenotypes of trk1Δtrk2Δ yeast expressing the ROMK1 variants.
Representative yeast viability assays with control strains and the ROMK1 variants in group 6 were performed on (A) solid and (B) liquid medium. (A) Ten-fold dilutions of overnight, saturated cultures of yeast were inoculated on medium as described in Fig 4. Images were taken after two days of incubation at 30°C. (B) Saturated yeast cultures were diluted to a starting OD600 of 0.2 with assay medium supplemented with 25mM KCl and grown at 30°C. OD600 readings were recorded and data were standardized as described in the Materials and Methods and in Fig 4. Data represent results from two independent experiments (n = 2–3 each), ± the range of the data. (C) Table summarizing the growth phenotypes of the six ROMK1 groups described in the text. The predicted consequence of each group is denoted. Growth phenotypes were obtained in a blinded fashion, compared to the growth of trk1Δ trk2Δ yeast expressing WT ROMK1, and the results are color-coded: Red denotes a severe growth defect, orange denotes a moderate growth defect, green denotes no growth defect (WT-like), and blue denotes a slight increase in growth compared to the WT control. These classifications were performed by visual inspection. For example, as shown in Fig 5A and 5B, the P265R variant exhibited levels of growth that matched the vector control (i.e., the errors overlapped in Fig 5B), and hence this mutation was designated “severe growth defect” in part (C). In contrast, the P265Y variant exhibited levels of growth that were intermediate to that of the vector control and “Wild-type” (Fig 5B). Hence, this mutation was designated “moderate growth defect” in part (C).
Fig 6.
Tested variants on the 3D structure of ROMK.
Left, residues associated with variants tested experimentally are color-coded on the ribbon representation according to their measured phenotypes, as in Fig 3. Right, the dominant twisting deformation computed from an elastic network model of the tetramer is indicated by red arrows. K80 is positioned near the interface between the two oppositely rotating (transmembrane and cytoplasmic) regions.
Table 1.
List of known mutations associated with Bartter syndrome and their computationally predicted classification.