A multi-hierarchical approach reveals D-serine as a hidden substrate of sodium-coupled monocarboxylate transporters

Transporter research primarily relies on the canonical substrates of well-established transporters. This approach has limitations when studying transporters for the low-abundant micromolecules, such as micronutrients, and may not reveal physiological functions of the transporters. While D-serine, a trace enantiomer of serine in the circulation, was discovered as an emerging biomarker of kidney function, its transport mechanisms in the periphery remain unknown. Here, using a multi-hierarchical approach from body fluids to molecules, combining multi-omics, cell-free synthetic biochemistry, and ex vivo transport analyses, we have identified two types of renal D-serine transport systems. We revealed that the small amino acid transporter ASCT2 serves as a D-serine transporter previously uncharacterized in the kidney and discovered D-serine as a noncanonical substrate of the sodium-coupled monocarboxylate transporters (SMCTs). These two systems are physiologically complementary, but ASCT2 dominates the role in the pathological condition. Our findings not only shed light on renal D-serine transport, but also clarify the importance of non-canonical substrate transport. This study provides a framework for investigating multiple transport systems of various trace micromolecules under physiological conditions and in multifactorial diseases.


Introduction
(IPA) revealed alterations of the biological function in Molecular Transport, which are 146 categorized by the types of their canonical substrates (Table supplement 2). Transport of "all 147 micromolecules" was shown to be drastically decreased from 4h IRI (Figure 2A: Top). This 148 decrease was due to the dramatic reduction of transport systems for heavy metals and organic 149 compounds ( Figure 2A). Figure 2B shows membrane transport proteins corresponding to the 150 annotation in Figure 2A. This proteomic result unveiled the whole picture of the molecular 151 targets of the injury. Furthermore, the results revealed the key membrane transport proteins 152 behind AKI metabolite biomarkers. It is noted that some transporters were upregulated, 153 suggesting the compensatory mechanism repair processes for cell survival from the injury 154 ( Figure 2B).

245
The transport functions of SMCTs are inhibited by non-steroidal anti-inflammatory 246 drugs (NSAIDs; e.g., ibuprofen and acetylsalicylate) (Itagaki et al., 2006;Gopal et al., 2007). 247 To verify the contribution of SMCTs to D-serine-reduced cell growth, we added ibuprofen to 248 the growth assay. In contrast to mock, in which ibuprofen has no effect, the growth suppression 249 by D-serine treatment was gradually attenuated by ibuprofen in both FlpInTR-SMCT1 and 250 FlpInTR-SMCT2 cells, indicating that the cell growth suppression by D-serine treatment is the 251 result of SMCT1 and SMCT2 functions ( Figure 5A). 252 We characterized SMCTs mediating D-[ 3 H]serine transport by using SMCT-stable cell  Figure 5D). The excess of non-radioisotope-labeled D-serine also inhibited D-[ 3 H]serine 258 uptake in both SMCTs. Still, its inhibitory effect in SMCT2 was much less than in SMCT1, 259 suggesting the lower affinity of D-serine in SMCT2 than in SMCT1 ( Figure 5D). The substrate selectivity of SMCT1 was investigated using synthetic biochemistry to 266 avoid the interference of amino acid and carboxylate transports by endogenous transporters in 267 cells. We established a cell-free assay system using proteoliposomes, in which purified proteins 268 are reconstituted into liposomes and all substances in the system are controllable. We purified 269 human SMCT1 by affinity column and reconstituted SMCT1-proteoliposome (SMCT1-PL) 270 ( Figure 6A). The function of SMCT1 in SMCT1-PL was verified by the transport of lactate, The transport of D-[ 3 H]serine in SMCT1-PL was Na + -dependent and reached the 278 stationary phase at approximately 5 min ( Figure 6B). The amount of D-serine transport was 279 slightly lower than lactate and propionate but higher than L-serine and the transport was 280 inhibited by ibuprofen, confirming SMCT1-mediated D-serine transport ( Figure 6C). Amino 281 acid selectivity revealed that SMCT1 recognized both L-and D-serine over other small amino 282 acids (L-and D-alanine), acidic amino acids (L-and D-glutamate), and large neutral amino acids 283 (L-and D-tyrosine) ( Figure 6D). Therefore, we concluded that serine is the substrate of SMCT1 284 and the recognition is more stereoselective for the D-than the L-isomer.

285
ASCT2 and SMCTs contribute to D-serine transport in renal proximal tubular epithelia 286 We next examined the contributions of ASCT2 and SMCTs to D-serine reabsorption in 287 the kidney using an ex vivo transport assay. ASCT2 is an antiporter, influx one amino acid with 288 efflux of another, while SMCTs are symporters (Ganapathy et al., 2008;Scalise et al., 2018). Therefore, we measured D-serine transport in the presence or absence of L-Gln preloading and 290 with or without ibuprofen to distinguish the functions of mouse Asct2 and Smcts. The

293
Accordingly, we tested 10 µM D-[ 3 H]serine transport in BBMVs derived from normal mice. In 294 the non-preloading condition, D-[ 3 H]serine uptake in Na + dependence gradually but 295 continuously increased and the uptake was remarkably inhibited by ibuprofen, suggesting that 296 the transport was mainly attained from Smcts functions ( Figure 7A). In L-Gln preloaded 297 BBMVs, D-[ 3 H]serine uptake in Na + -dependence arose quickly and reached the saturated point 298 at 1 -2 min ( Figure 7B). The uptake at early time points (£ 30 sec) was ibuprofen-insensitive 299 but became partly ibuprofen-sensitive from 1 min to the stationary phase ( Figure 7B). These 300 results indicate that the D-serine transport was mediated by both Asct2 and Smcts. At early 301 time points, D-[ 3 H]serine was transported by Asct2, which occurred highly and rapidly.

302
Meanwhile, Smcts functions (ibuprofen-sensitive) initiated later but had prolonged functions 303 as seen in the non-preloading condition. Taken altogether, we suggested that, in the normal 304 kidney where renal proximal tubular epithelial cells contain intracellular L-Gln, D-serine 305 reabsorption is derived from combinational functions of both ibuprofen-sensitive (e.g. Smcts) 306 and ibuprofen-insensitive (e.g. Asct2) transporters.

307
The localization of Smct1 in the kidney was unclear. Gopal et al. reported the 308 localization of Smct1 at S3 and Smct2 in all segments, while single-nucleus RNA sequencing 309 (snRNA-Seq) showed Smct1 at S2 -S3 and Smct2 at S1 segment (Figure supplement 7: 310 Controls) (Gopal et al., 2007;Kirita et al., 2020). We generated an anti-Smct1 antibody and 311 examined the localization of Smct1. The result showed that Smct1 was mainly localized at the 312 S3 and slightly at the S1 + S2 segments (Figure supplement 8).
Under IRI conditions, we observed low urinary but high plasma levels of D-serine, 315 suggesting high D-serine reabsorption (Figure supplements 1A, 2A). Our membrane 316 proteomics revealed that Asct2 was increased while and Smct1 and Smct2 were decreased in 317 both 4h and 8h IRI (Table 1). We then examined the functional contributions of Asct2 and

349
In this study, we presented an approach to investigate the transport systems for 350 micronutrients and metabolites under physiological conditions and in multifactorial diseases. 351 We selected AKI as a model study because AKI is known to target renal proximal epithelia

363
Several studies have reported the non-canonical substrates of membrane transport 364 proteins, such as the recognition between amino acids, carboxylates and amines among SLCs 365 (Metzner et al., 2005;Matsuo et al., 2008;Schweikhard and Ziegler, 2012;Wei et al., 2016).  Figure 5E). Besides 379 transport affinity, the expression level is another important factor in evaluating the transport 380 capacity. Regarding chromatogram intensities in the proteomics data, we found that the 381 intensities of Smct1 (2.9 x 10 9 AU) and Smct2 (1.6 x 10 8 AU) were remarkably higher than 382 Asct2 (1.5 x 10 7 AU) in the control mice (Table 1:

392
Several pieces of evidence support SMCT1 at S3 segment and SMCT2 at S1 + S2 393 segments on renal D-serine handling. First, our enantiomeric amino acid profile from the IRI 394 model, which agrees with the previous reports in CKD patients and AKI animal models, Second, previous studies indicated renal D-serine transport takes place at the proximal tubules 399 by the distinct transport systems between S1 + S2 and S3 segments. Both systems exhibited 400 the characteristics of Na + dependency, electrogenicity, low affinity (mM range), and partial 401 stereoselectivity (Kragh-Hansen and Sheikh, 1984;Shimomura et al., 1988;Silbernagl et al., 402 1999). These properties of the D-serine transport suit well to SMCTs and convince us of the 403 contribution of SMCTs to renal D-serine transport in addition to ASCT2. Moreover, continuous  Combining all the results, we propose the model of D-serine transport systems in renal 413 proximal tubules ( Figure 8). In the normal kidneys, the expression levels of SMCT2 at S1 + S2 414 segments and SMCT1 at S3 segment are high, whereas ASCT2 is low. Both SMCTs and 415 ASCT2 are involved in D-serine reabsorption. Nonetheless, the net reabsorption levels appear 416 to be relatively low due to the low affinities of SMCTs. It is likely that reabsorption at the S3   Table S1). It is possible that low L-serine reabsorption in IRI is a result of B 0 AT3 435 reduction.

436
In proteomics, some transporters including ASCT2 were found to be increased in the 437 AKI conditions ( Figure 2B), yet the mechanism behind this change is still unclear. snRNA-seq 438 study revealed that the cell clusters in all proximal tubular segments during early AKI were 439 drastically different from the healthy kidney suggesting that the new distinct cell clusters in 440 AKI were de novo synthesized for the repair process, as early AKI is a reversible condition 441 (Kirita et al., 2020). ASCT2 was found to be only one type of small amino acid transporter 442 which was increased during AKI conditions ( Figure 2B). ASCT2 was reported to be highly  Thus, we postulate that ASCT2 is a nascent transporter to provide small amino acids and glutamine for cell growth, as a molecular mechanism in de novo cellular synthesis for the 448 recovery processes. In this study, we used the cDNA of human SLC5A8/SMCT1 clone "NM_145913". We 508 generated the clone NM_145913 from the clone AK313788 (NBRC, NITE, Kisarazu, Japan).

509
At first, SMCT1 from AK313788 was subcloned into p3XFLAG-CMV14 (Sigma) via HindIII 510 and BamHI sites. The clone "pCMV14-SMCT1" NM_145913 was subsequently generated by  Anti-ASCT2 Ab was purified from the anti-sera using HiTrap Protein G HP (GE 543 Healthcare) following the manufacturer's protocol. After purification, the Ab was dialyzed 544 against PBS pH 7.4 and adjusted concentration to 1 mg/mL.

548
To purify anti-Asct2(NT) Ab, the anti-sera was subjected to the first purification by using the       (Table S1).

640
Mass spectrometry of HEK293 cells was analyzed from crude membrane fractions.

652
Effect of D-serine on cell growth HEK293 cells were seeded into 96-well-plate at 10,000 cells/well. Transient 654 transfection was performed 12 hours after seeding followed by L-or D-serine treatment at 12 655 hours after that. In the case of FlpInTR-stable cell lines, if needed, ASCT2 siRNA was 656 transfected 12 hours after seeding. Dox was added one day after seeding followed by treatment 657 with L-or D-serine (in the presence or absence of ibuprofen as indicated) 10 hours after adding 658 Dox. The cells were further maintained for two days. Cell growth was examined by XTT assay.

663
Cell growth in serine treatment samples was compared to the control (without serine treatment).

664
For transporter screening, the growth of the transfected cells at a specific D-serine concentration 665 was compared to that of Mock after normalization with no treatment. After termination of the assay, the cells were lysed. An aliquot was subjected to measure 673 protein concentration, and the remaining lysate was mixed with Optiphase HiSafe 3 674 (PerkinElmer). The radioactivity was monitored using a β-scintillation counter 675 Hitachi). In the transport assay with or without Na + , Na + -HBSS, or Na + -free HBSS (choline-Transport assay in FlpInTR-stable cell lines was performed in a similar way to HEK293 678 cells. After cell seeding for one day, SMCT1 and SMCT2 expression were induced by adding

1188
Area and colors represent -log10(p-value) and annotated functions (z-score), respectively.   against -log10 of p-value. Three proteins with log2 fold more than 4.0 (see values in Table S1) 1214 are omitted for the better view. The value of Asct2 (red dot) was set as a cut-off value to select   Prior to uptake measurement, the BBMVs were preloaded with 4 mM L-Gln or buffer (no 1274 preload). The uptake was measured for 1 min in Na + -free buffer (K + ), Na + -containing buffer