Xenopus laevis were housed and cared for in accordance and approval of the Institutional Care and Use Committees of the Mayo Clinic College of Medicine and Brigham and Women’s Hospital.

Constructs containing coding regions of indicated transporters were subcloned into the pGEMHE Xenopus expression vector and capped cRNA synthesized. Oocytes were injected with 50 nL cRNA (0.2 μg/μL, 10 ng/oocyte) or water as previously (Hirata et al., 2010; Romero et al., 2000) for other transporters and incubated at 16°C in OR3 media. Oocytes were studied 3–10 days after injection. Amount of cRNA injected was constant through all experiments and transporter expression was assumed to be constant as well.

All uptake experiments using Xenopus laevis oocytes were performed as described previously (Mandal et al., 2017; Mandal and Mount, 2019). Briefly, for [¹⁴C]-oxalate uptake experiments in oocytes, each oocyte was microinjected using a fine-tipped micropipette fitted in a microinjector (World Precision Instruments Inc. Sarasota, FL) with equal amount (∼25 ng/50 nL) of in vitro synthesized intact appropriate cRNA (checked by 1% formaldehyde-agarose gel electrophoresis in MOPS buffer, pH 8.0) and incubated in isotonic ND96 medium (96 mM NaCl, 2.0 mM KCl, 1.8 mM CaCl₂, 1.0 mM MgCl₂ and 5 mM HEPES, pH 7.4) containing 3 mM pyruvate and gentamycin (10 μg/mL at 16°C–18°C for approximately 48 h for the expression of respective protein from injected cRNA) in chloride-free medium (Cl⁻ was substituted by equimolar gluconate ion), pH 7.4 at 25°C. Oocytes were then washed four times with isotonic ND96 medium without pyruvate and gentamycin. After 1 h of incubation in the indicated chloride-free uptake medium (1 mL/well) containing [¹⁴C]-oxalate (20 μM; 51 μCi/μmol; Moravek Inc. Brea, CA) in 12-well plates at room temperature [25°C] in a horizontal shaker-incubator, oocytes (15–20 in each group) were washed three times with the ice-cold uptake medium to remove external radioisotope. Radioisotope content of each individual oocyte was measured by scintillation counter (LS 6500 multi-Purpose Scintillation Counter, Beckman Coulter, Brea, CA) after solubilization in 0.3 mL of 10% (v/v) SDS and addition of 2.5 mL of scintillation fluid. All uptake experiments included at least 15–20 oocytes in each experimental group; statistical significance was defined as p ≤ 0.05 by one-way ANOVA with Bonferonni’s multiple comparisons tests and results were reported as mean ± SD. All of the uptake experiments were performed more than three times for confirmation with appropriate controls; data shown for each figure are from a single representative experiment (units are pmol/oocyte/h).

Electrophysiology protocols were performed as we previously reported (Romero et al., 2000). All solutions were either ND96 (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl₂, 1 mM MgCl₂, and 5 mM HEPES, pH 7.5) with additional 1 mM Na₂SO₄ or 300 µM Na-oxalate as indicated. HCO₃ ⁻ solutions were continuously bubbled with 5% CO₂/33 mM HCO₃ ⁻ (pH 7.5). For unclamped Vm measurements, data were recorded with an OC-725C voltage clamp amplifier (Warner Instruments, Hamden, CT), filtered at 2–5 kHz, digitized at 10 kHz. For voltage clamp experiments, oocytes were clamped to −60 mV and then voltage rapidly shifted to −160 mV to +60 mV in 20 mV steps (70 ms sweeps, filtered at 1 kHz; HEKA Elektronik, Harvard Bioscience Inc.; https://www.heka.com/). The resulting linear IV-plots are then expressed as “slope conductance” in the appropriate non-CO₂/HCO₃ ⁻ solution (Chen et al., 2024). Full solution composition appears in Table 1. For ion-selective microelectrodes to monitor pH_i, we used a liquid-ion-exchange (lix) resin (H⁺ ionophore I, mixture B; Fluka Chemical, Ronkonkoma, NY, United States). Intracellular pH microelectrodes had slopes of at least −54 mV/pH unit (Romero et al., 2000). Whenever possible, the slope of pH changes (pH-electrode and pH-sensor) within an individual trace was calculated from regions with linear change and overlapping range of absolute pH to minimize the effects of intrinsic and HCO₃ ⁻-derived buffering on perceived rate of pH changes (Chesler, 1990; Rossano and Romero, 2017; Vaughan-Jones and Wu, 1990).

Flies were kept on standard medium in vials at 22°C, 12:12-h photoperiod, and 40% relative humidity. All lines were backcrossed into a w¹¹¹⁸ (Bonini) background and this line served as controls.

This was performed as described previously (Reynolds et al., 2024). To specifically knock down CG5002, we utilized the CapaR-GAL4 driver (Terhzaz et al., 2012) to restrict expression to the MT principal cells and crossed it VDRC lines KK100600 or GD10058 (Vienna, Vienna BioCenter Core Facility) containing a hairpin dsRNA sequence directed against CG5002.

Quantitative RT-PCR validation was performed as described elsewhere (Reynolds et al., 2024). mRNA was prepared from 7-day-old W¹¹¹⁸ or experimental tubules using Qiagen RNAeasy column. Superscript II and an oligo-dT were used for reverse transcription. For each sample, 500 ng of cDNA were added to 12.5 μL of 2 μM SYBR green reaction mix (Finnzymes, GRI, Essex, United Kingdom) and 1 μL of 6.6 μM forward and reverse primers. An Opticon 2 thermocycler was set as follows: 95°C for 15 min followed by 35 cycles of 95°C for 30 s, 58°C for 30 s, and 72°C for 30 s. The ribosomal protein L32 (RPL32) gene was used as a standard in all experiments. For each condition, we used three independent samples; each PCR was repeated three independent times to verify results. Full primer sequences are listed in Table 2.

Live epifluorescence imaging of pHerry (Rossano et al., 2017), a pseudo-ratiometric GEpHI consisting of a pH-sensitive superecliptic pHluorin (Miesenbock et al., 1998) (SepH; 488/510 nm ex./em.) and pH-insensitive mCherry (Shaner et al., 2004) (565/620 nm ex./em.) joined by a short peptide linker, was performed as previously described (Rossano and Romero, 2017). Briefly, UAS-pHerry-myr (encoding pHerry with an additional N-terminus myristylation sequence to aid in near-membrane targeting and nuclear exclusion) flies were crossed to CapaR-GAL4 flies to drive construct expression in the principal cells of the MT. Anterior MTs were extracted from 7d female flies and transferred to poly-l-lysine-coated slides. Preparations were moved to an inverted wide-field epifluorescent microscope with GFP (SEpH) and RFP (mCherry) filter sets (470/40 nm ex., 515 nm longpass em; 500 nm dichroic and 546/10 nm ex, 590 nm longpass em, 565 nm dichroic), a ×10/0.45 air objective, a monochromatic camera for live-image capture, and an in-line computer-controlled perfusion system. SEpH and mCherry images were collected in an interleaved fashion with 300 ms exposure for each channel and an additional 400 ms to allow for channel switching, yielding a total imaging rate of 1 Hz. MTs were continuously bathed in insect saline consisting of 121.5 mM NaCl, 20 mM KCl, 20 mM glucose, 8.6 mM HEPES, and 10.24 mM NaHCO₃ with pH 6.8 set by HCl and NaOH. In NH₄Cl pulse experiments NH₄Cl was provided in equimolar exchange for NaCl. pH_i calibration was performed as previously described (Rossano et al., 2013; Rossano et al., 2017; Thomas et al., 1979) using modified solution consisting of 130 mM KCl, 20 mM Glucose, and 8.6 mM MES, HEPES, or TAPS as appropriate for desired pH range. pH was adjusted with HCl and NMDG and solution was supplemented with 10 μM nigericin in DMSO. Additional point calibrations were used at the end of some imaging experiments. When possible, control and experimental tubules were image simultaneously with the MTs positioned side-by-side to minimize the effects of variable perfusion rates between experiments. Rates of change in pH_i were compared in segments of traces within the same pH_i to minimize the influence of the pH-dependence of intracellular buffering (Chesler, 1990; Rossano and Romero, 2017; Vaughan-Jones and Wu, 1990). Imaging stacks were acquired in Intelligent Imaging Innovations Slidebook software, background corrected and exported for fluorescence quantification in ImageJ. Full compositions of solutions are described in Table 3.

Adult flies (7d) were allowed to feed on standard food. Malpighian tubule (MT) from female flies were dissected in Schneider’s medium and transferred immediately to poly-l-lysine-coated slides with insect saline. MTs were allowed to settle for 10 min, at which point solution was swapped for saline supplemented with 10 mM Na-oxalate (equimolar exchange for NaCl).

CaOx crystals were quantified using ImageJ software as previously (Landry et al., 2016). Images were taken at ×10 magnification with an air objective and an inline polarizer. 600 μm was measured from the terminal end of the respective MT. Once measured, brightness and contrast were adjusted, resulting in only birefringent crystals in the designated 600 μm of tubule being visible, and the remaining tubule artifacts were blacked out. A threshold was then set, resulting in binary coloration so that crystals could be counted. Crystals were analyzed using the Analyze Particles feature in ImageJ with the minimum desired size of stone detection specified at 2 μm². Data generated included an accurate and specific crystal count. Data were then analyzed via one-way ANOVA with Bonferroni post hoc test (P < 0.05) or Student’s t-test (P < 0.05) using GraphPad Prism.

Data were compiled in GraphPad Prism and analyzed as indicated in individual data sets. Image analysis was performed in Intelligent Imaging Innovations Slidebook software as well as ImageJ. Curve fitting was performed in Microsoft Excel and GraphPad Prism. Representative micrographs were processed in Adobe Photoshop and figures were assembled with Adobe Illustrator.

Full Composition of all experimental solutions is described in Tables 1, 3.

All statistical calculations were performed in GraphPad Prism. Comparison across two groups was performed by unpaired Student’s t-test. Comparisons more than two groups were performed by ANOVA with Bonferroni multiple comparisons correction. Sample sizes for experiments were determined by number of oocytes or number of extracted Malpighian tubules (one tubule per fly). Additional details are provided in descriptions of individual experiments.

Previous work showed that apical protein dPrestin could transport oxalate into the Malpighian tubule lumen (Figure 1) (Hirata et al., 2012a). This dPrestin (Slc26a5/a6) is the only Drosophila Slc26 protein molecularly close to other known mammalian oxalate transporters (Hirata et al., 2012b) (Figures 2A,B). All others Drosophila Slc26 proteins, at the molecular sequence level, are most closely related to SLC26A11 (Figure 2C; Supplemental Figure S1), a Na⁺ independent SO₄ ²⁻ transporter (Vincourt et al., 2003) and a Cl⁻/HCO₃ ⁻ exchanger (Xu et al., 2011). SLC26A11/Slc26a11 proteins have not been reported to transport Ox²⁻.

To first test for oxalate transport, we used ¹⁴C-oxalate uptake experiments with candidate transporters expressed in Xenopus oocytes (Figure 3A). Known human oxalate transporters (SLC26A6, SLC26A2/DTDST) and dPrestin were tested for oxalate uptake and compared to water-injected controls, SLC26A11, and Drosophila Slc26 clones [CG5002 (Neat), CG5404, CG6125, CG6928, CG7005]. As previously shown, SLC26A6 (114.60 ± 18.30 pmol/oocyte/h), DTDST (48.001 ± 8.66) and dPrestin (109.16 ± 26.80) all show significant oxalate uptake vs. control (16.63 ± 2.99, p < 0.00001) with SLC26A6 and dPrestin having the highest oxalate uptake without extracellular Cl⁻ (Figure 3A). SLC26A11 does not show any statistically significant oxalate uptake compared to the control (17.15 ± 2.78, p = 1). Of the Drosophila Slc26 proteins, only Neat (53.43 ± 5.49, p < 0.00001) and CG5404 (32.74 ± 11.83, p < 0.05) showed significant oxalate uptake in Cl⁻ free solution. CG6125 (27.69 ± 3.69, p = 0.98), CG6928 (19.27 ± 3.17, p = 1) and CG7005 (21.62 ± 4.00, p = 1) did not differ from controls (16.63 ± 2.99).

Within these same uptake experiments, cis-inhibition by 10 mM SO₄ ²⁻ and 1 mM DIDS were evaluated (Figure 3B). As expected from the previously published activity of dPrestin (111.14 ± 27.28 pmol\oocyte/h), both DIDS (29.41 ± 3.64) and SO₄ ²⁻ (4.69 ± 0.73) cis-inhibit dPrestin oxalate uptake to water injected control levels (16.94 ± 3.04; SO₄ ²⁻, 7.74 ± 0.65; DIDS, 1.91 ± 0,38). A similar cis-inhibition pattern was found comparing CG5404 to control Ox²⁻ uptakes. For Neat (54.40 ± 5.59), cis-addition (bath) of 10 mM SO₄ ²⁻ (9.03 ± 1.22) fully blocked Ox²⁻ uptake (Figure 3B, left). However, 1 mM DIDS only reduced Ox²⁻ uptake by half (30.92 ± 4.29).

Many SLC26 protein family members transport chloride, and thus we evaluated if cis application of Cl⁻ in previously chloride-free solutions could inhibit Ox²⁻ uptake within the same experiments (Figure 3B, right sides of panels). Both Neat (54.40 ± 5.59) and CG5404 (32.74 ± 11.83) displayed no significant cis inhibition of Ox²⁻ uptake by Cl⁻ (66.29 ± 5.18 and 37.91 ± 4.64 respectively). Chloride inhibited Ox²⁻ uptake by dPrestin by approximately half (111.14 ± 27.28 to 68.52 ± 9.45). Similarly robust inhibition of oxalate uptake by cis application of Cl⁻ was observed in SLC26A6 (114.60 ± 18.30 to 52.82 ± 5.07). Taken together, these data support our ability to detect the effects of extracellular Cl⁻ on Ox²⁻ uptake and reveal Neat does not function as an Ox²⁻/Cl⁻ exchanger.

The next consideration was which Drosophila Slc26 oxalate transporter mRNAs or proteins have expression in MTs. Searching FlyAtlas2 (flyatlas2.org), we found that Neat was 175 ± 21 FPKM (adult male) and 191 ± 27 FPKM (adult female) while CG5404 is 16 ± 2.5 FPKM (male) and 15 ± 4.0 FPKM (female). These data indicate that Neat mRNA expression is ∼10-fold higher than cg5404, and that oxalate uptake is ∼2-fold higher. The combined data indicate that Neat is the Drosophila Slc26 transporter of most interest for a candidate basolateral Ox²⁻ transporter in the Malpighian tubule. CG5404 expression is increased relative to Neat in hindgut (140 vs 18 FPKM in adult females).

With Neat identified as the likely oxalate transporter of interest we further characterized its transport of Ox²⁻, HCO₃ ⁻, and SO₄ ²⁻ in a Xenopus oocyte heterologous expression system (Figure 4). Neat demonstrated electroneutral HCO₃ ⁻ influx in the presence of HCO₃ ⁻/CO₂ buffered saline (dpH/dt 0.0094 ± 0.0047 pH units per minute as compared to 0.0013 ± 0.0003 in water injected control oocytes; Figures 4A–C). This influx of HCO₃ ⁻ was stopped by the addition of 300 µM Ox²⁻ (ΔdpH/dt: −0.010 ± 0.004; Figure 4D) or 1 mM SO₄ ²⁻ (ΔdpH/dt: −0.010 ± 0.004; Figure 4E). Each of the substrates produced significant changes in dpH/dt compared to those observed in water injected oocytes (p < 0.005 by Student’s unpaired t-test in all cases). Expression of Neat did not significantly alter resting pH_i (Figure 4F). These electrophysiology data suggest Neat mediates electroneutral HCO₃ ⁻/Ox²⁻ exchange and HCO₃ ⁻/SO₄ ²⁻ exchange. Voltage clamp experiments revealed that dPrestin elicited conductance changes with 5 mM SO₄ ²⁻ as previously reported (Hirata et al., 2012b). However, neither Ox²⁻ nor SO₄ ²⁻ elicited a conductance change from baseline in Neat. Additionally, the overall Neat-oocyte conductance (5.94 ± 2.03 µS) was about 25% of baseline dPrestin conductance (19.44 ± 4.25 µS; Figure 4G). These data further support the conclusion that ion transport via Neat is electroneutral.

The fact that Neat mediates exchange of acid equivalents raised the possibility that we could investigate Neat activity in Drosophila MTs ex vivo through the use of genetically encoded pH indicators targeted to the cytosolic surface of the MT principal cell membrane. To this end, we used the GAL4/UAS binary expression system to express pHerry (Rossano et al., 2017), a pseudo-ratiometric genetically encoded pH indicator consisting of a pH-sensitive GFP variant (SEpH) and a pH insensitive RFP variant (mCherry) in the principal cells of adult Drosophila MTs and calibrated the biosensor to monitor pH_i in real time through live epifluorescent imaging as previously described (Rossano and Romero, 2017). Both the GFP and RFP components of pHerry were easily visualized in extracted MTs (Figure 5A), and direct manipulation of pH_i with NH₄Cl pulses confirmed the expected pH sensitivity of the SEpH component and the pH insensitivity of the mCherry component (Figures 5B,C). We used the high potassium/nigericin technique (Thomas et al., 1979) to calibrate the GFP/RFP ratio as a function of pH_i (Figures 5D,E). Boltzmann fit to these data yielded R² = 0.95 and pK_a = 7.24, which is in agreement with previously published ex vivo and in situ calibration of superecliptic pHluorin-based pH Indicators in Drosophila (Rossano et al., 2013; Rossano et al., 2017).

We next combined live imaging of pH_i with the robust RNAi toolkit available in Drosophila to evaluate changes in pH_i of MT principal cells attributable to HCO₃ ⁻, SO₄ ²⁻, and Ox²⁻ in the presence and absence of Neat knockdown with two separate RNAi constructs (KK100600 and GD10058 from the Vienna stock center). When CapaR-GAL4 was used to drive RNAi expression in MT principal cells quantitative PCR revealed knockdown of transcript level to 34% ± 0.5% and 40% ± 0.5% of control level in RNAi-1 (KK100600) and RNAi-2 (GD10058), respectively. Transcript levels were normalized to RPL32 in all cases and data were pooled across three runs from each genotype in RNA extracted from adult Drosophila MTs. Freshly extracted MTs expressing membrane-targeted pHerry were mounted on cover glass and perfused with insect saline buffered either with HEPES or HCO₃ ⁻/CO₂ with additional oxalate and sulfate to monitor changes in pH_i in response to basolateral addition of HCO₃ ⁻, Ox²⁻, and SO₄ ²⁻ (Figures 6A,B). Switching to HCO₃ ⁻ buffered saline produced an initial fall in pH_i followed by a gradual alkalinization mediated by HCO₃ ⁻ influx. This alkalinization was significantly slowed when Neat was knocked down with either RNAi construct as compared to that seen in control MTs (ΔdpH/dt: +0.14 ± 0.04 vs. +0.06 ± 0.04 in RNAi-1 experiments and +0.17 ± 0.06 vs. +0.07 ± 0.06 in RNAi-2 experiments, p < 0.01 in both cases by Students unpaired t-test; Figure 6C). This alkalinization was quickly reversed with the addition of 1 mM basolateral oxalate as well as 1 mM basolateral sulfate. In each case the change in dpH/dt with the addition of oxalate (−0.72 ± 0.18 vs. −0.41 ± 0.21 in RNAi-1 and -0.67 ± 0.10 vs. −0.28 ± 0.14 in RNAi-2) and sulfate (−0.19 ± 0.06 vs. −0.05 ± 0.05 in RNAi-1 and -0.15 ± 0.05 vs. −0.04 ± 0.03 in RNAi-2) was significantly reduced by knockdown of Neat (Figures 6D,E). Knockdown of Neat did not change resting pH_i in the principal cells of the transitional segment of MTs (Figure 6F). Taken together with our prior oocyte expression experiments these data strongly support the conclusion that Neat mediates significant HCO₃ ⁻/Ox²⁻ and HCO₃ ⁻/SO₄ ²⁻ exchange on the basolateral membrane of Drosophila MT principal cells of the transitional segment.

We hypothesized that if Neat was a significant mediator of basolateral Ox²⁻ transport in the Drosophila MT then knockdown of this transporter should significantly impair transepithelial transport of oxalate and formation of CaOx crystals in the MT lumen. Previous work has demonstrated that knockdown of the apical oxalate transporter in the MT significantly reduces luminal CaOx crystal formation when Ox²⁻ solution is perfused across extracted Drosophila MTs (Landry et al., 2016). To test if this was the case for the putative basolateral Ox²⁻ transporter, we assessed CaOx crystal burden in extracted MTs from control and Neat knockdown flies via polarizing birefringence. 60-min bath application of insect saline containing 10 mM Ox²⁻ was sufficient to induce prominent crystal formation (Figure 7A). CaOx crystal count was significantly reduced when Neat was knocked down by both RNAi-1 and RNAi-2 constructs (12.64 ± 4.14 and 21.5 ± 9.39 for each construct respectively vs. 33.20 ± 8.67 in control, p < 0.0001 by one-way ANOVA with Bonferroni multiple comparisons correction; Figure 7B). These data suggest Neat significantly contributes to transepithelial Ox²⁻ movement, and this contribution is of similar magnitude to that of the previously described apical Ox²⁻ transporter dPrestin (Landry et al., 2016).