Identification and characterization of a transporter complex responsible for the cytosolic entry of nitrogen-containing-bisphosphonates

Nitrogen-containing-bisphosphonates (N-BPs) are a class of drugs widely prescribed to treat osteoporosis and other bone-related diseases. Although previous studies have established that N-BPs function by inhibiting the mevalonate pathway in osteoclasts, the mechanism by which N-BPs enter the cytosol from the extracellular space to reach their molecular target is not understood. Here we implemented a CRISPRi-mediated genome-wide screen and identified SLC37A3 (solute carrier family 37 member A3) as a gene required for the action of N-BPs. We observed that SLC37A3 forms a complex with ATRAID (all-trans retinoic acid-induced differentiation factor), a previously identified genetic target of N-BPs. SLC37A3 and ATRAID localize to lysosomes and are required for releasing N-BP molecules that have trafficked to lysosomes through fluid-phase endocytosis into the cytosol. Our results elucidate the route by which N-BPs are delivered to their molecular target, addressing a key aspect of the mechanism of action of N-BPs that may have significant clinical relevance.

proteins. We detected an interaction between SLC37A3 and ATRAID in both pull-down 159 directions ( Figure 3F), confirming that SLC37A3 and ATRAID physically interact, likely 160 forming a lysosomal complex. 161 As it has been reported that the expression levels of certain solute carriers life, suggesting that ATRAID and SLC37A3 are mutually dependent for their stability. 177 Intriguingly, the deletion of ATRAID also altered the glycosylation pattern of SLC37A3. 178 In the absence of ATRAID, the mature, glycosylated population of SLC37A3 (around  Finally, we explored the mechanism by which the knockout of SLC37A3 and 186 ATRAID conferred resistance to N-BPs. Given the predicted function of SLC37A3 as a 187 transporter, we hypothesized that ATRAID and SLC37A3 transport N-BP molecules 188 across the lipid bilayer to inhibit FDPS. This hypothesis is consistent with our finding 189 that the role of SLC37A3 and ATRAID in the mechanism of action of N-BPs is specific 190 to the chemical properties of N-BPs. To test this hypothesis, we incubated wild-type, 191 ATRAID KO and SLC37A3 KO cells with radioactive ALN ( 3 H-ALN) and measured total 192 intracellular radioactivity. However, we observed no significant difference in 193 accumulation of radioactivity between the knockouts and wild-type cells ( Figure 4A). As 194 a previous study showed that N-BP molecules traffic to acidified endocytic 195 compartments through fluid-phase endocytosis (Thompson et al., 2006), we further hypothesized that N-BP molecules traffic to lysosomes and that SLC37A3 and ATRAID,197 together as a lysosomal complex, might function to release N-BP molecules from the 198 lumen of lysosomes into the cytosol. This model predicts that the total amount of 199 intracellular 3 H-ALN will remain the same in wild-type, SLC37A3 KO and ATRAID KO cells,200 but 3 H-ALN will not be able to exit lysosomes in knockout cells. To detect this potential supplement 1B-D) (Abu-Remaileh et al., 2017;Wyant et al., 2017) to assess the 212 lysosomal accumulation of 3 H-ALN. As our model predicted, we observed a significant 213 enrichment of 3 H-ALN in lysosomes purified from knockout cells compared to those from 214 wild-type cells ( Figure 4C). Additionally, consistent with the observation that ATRAID is 215 required for the stable expression of SLC37A3, ATRAID KO cells phenocopied 216 SLC37A3 KO cells in these uptake assays. Taken together, our results suggest that N-217 BPs traffic to lysosomes after internalization through endocytosis, and SLC37A3 and 218 ATRAID form a lysosomal transporter complex that releases N-BP molecules from the 219 lumen of lysosomes into the cytosol.

220
In summary, this study elucidates the route by which N-BPs enter the cytosol and 221 inhibit their molecular target. As a recent study has proposed that patients who harbor a 222 genetic variant of GGPS1 might be more prone to the side-effects of N-BP treatment 223 (Roca-Ayats et al., 2017), it is possible that patients with variants of SLC37A3 or 224 ATRAID, which are genes crucial for the action of N-BPs, might also exhibit non-225 canonical responsiveness to the drugs. Therefore, our results may bear significant 226 relevance to the clinical applications of N-BPs.

355
Cells were allowed to recover for two days before they were single-cell sorted into a 96-well plate. Clonal expansion and genotyping were then performed as described above 357 for HEK 293T cells.  was then centrifuged at 1000×g, 4°C for 2 min. Radioactivity in the supernatant was 582 then measured by scintillation counting.

583
All uptake assays were repeated independently for two times.   (A) Evaluation of the reproducibility of the CRISPRi screen. The enrichment score (ρ) of each sgRNA was calculated separately from two biological replicates of the CRISPRi screen and compared in a scatter plot. Data points corresponding to negative control sgRNAs are colored in gray. (B) Quantile-quantile plot comparing the distribution of observed average sgRNA enrichment scores (ρ scores) of each gene in the genome with a Gaussian distribution that has the same mean and standard deviation. The dashed gray line represents the predicted location of data points if the distribution of ρ scores is indeed Gaussian. The large deviations from the gray line observed at the two ends of the distribution indicate that the silencing of those genes has stronger effects than expected by pure Gaussian noise and is therefore likely to be biologically meaningful. The dotted lines are arbitrary thresholds set to select resistance hits (red dotted line) and sensitizing hits (blue dotted line) that deviate significantly from Gaussian predictions. 398 resistance hits and 28 sensitizing hits passed the thresholds. (C-D) Gene Ontology (GO) pathway enrichment analysis of the top 100 resistance hits (C) and the top 30 sensitizing hits (D) identified in the CRISPRi screen. Only the most specific subclasses that are statistically significant are shown. Both fold enrichment of pathway genes and P-values of fold enrichment are displayed.
Fold enrichment values were clipped at 100 fold. P-values were corrected for multiple testing using Bonferroni correction. Note that genes involved in the mevalonate pathway, which includes IPP biosynthesis, geranyl phosphate synthesis and farnesyl phosphate synthesis, are significantly enriched in top hits from the screen. IPP: isopentenyl pyrophosphate. (E) Volcano plot showing the results from a second CRISPRi screen using zoledronate, another representative N-BP, as the selection agent. Plot layout is the same as in Figure 1B. SLC37A3 and ATRAID are highlighted in red and cyan, respectively. Significant hits were defined as genes that had a fold enrichment with an absolute value larger than 0.1, and a P-value smaller than 0.05. Note that higher alendronate doses were used in F compared to E to induce detectable levels of unprenylated proteins. ALN: alendronate.     D and H) and Na + /K + -ATPase (G), and co-localization of SLC37A3-HA and lATRAID-V5 (E). Scale bars represent 10 µm. Each image displayed is the representative example chosen from at least five similar images. Note that in E, there exists a subpopulation of lATRAID-V5 that localizes to the plasma membrane but not with SLC37A3-HA. Such distribution is likely a result of a higher-thanendogenous expression level of lATRAID-V5 in KO 2 + SLC37A3-HA + lATRAID-V5 HEK 293T cells (A), as such distribution is not observed in ATRAID KO + lATRAID-V5 HEK 293T cells (G), which express lATRAID-V5 at a lower-than-endogenous level (A). KO 2 : ATRAID KO ; SLC37A3 KO . sATRAID: short isoform of ATRAID. lATRAID: long isoform of ATRAID. area (turquoise or blue, inside nuclei, where no stain should be present) and a signal area (orange or red) were selected and the distribution of pixel values within each area was plotted in a histogram. The outlines in the histogram are color-coded to match the boarders of selected areas.
(D-E) Polysome profiling experiment assessing the translation efficiency of SLC37A3 transcripts in SLC37A3 KO and KO 2 backgrounds. Lysates from indicated cell lines were analyzed on a gradient station and fractionated into five fractions: untranslated transcripts (UT), small and large ribosome subunits (SL), lowly translated transcripts (LT), medially translated transcripts (MT) and highly translated transcripts (HT). The level of SLC37A3 transcripts relative to the level of TBP (TATA-binding protein) transcripts in each fraction was measured and plotted. The SL fraction was excluded from the analysis. A total RNA fraction was included as a reference. No overall shift was observed in the distribution of SLC37A3 transcripts in the KO 2 background compared to that in the SLC37A3 KO  The band corresponding to an unglycosylated population of SLC37A3 that is present in the absence of ATRAID but not in the presence of ATRAD is marked with an asterisk. (G) Immunoblot comparing the glycosylation patterns of total cellular SLC37A3 and the sub-population of SLC37A3 that interacts with ATRAID.
In KO 2 HEK 293T cells over-expressing SLC37A3-HA and sATRAID-V5, proteins that interact with sATRAID-V5 were purified with immuno-precipitation against V5 epitope and compared with total proteins in the lysate. The pre-IP total lysate (lane 1-2) and anti-V5 IP eluate (lane 3-4) were either left untreated (lane 1 and 3) or treated with PNGase F (lane 2 and 4) and analyzed by blotting against SLC37A3-HA. IF: immunofluorescence. IB: immunoblot. KO 2 : ATRAID KO ; SLC37A3 KO . PNGase F: peptide: N-glycosidase F, an enzyme that removes all asparagine (N)linked sugar modifications from glycoproteins. Endo H: endoglycosidase H, an enzyme that only removes high mannose sugar moieties on ER glycoproteins that have not been processed by the Golgi apparatus.  Localization of HA tagged TMEM192 (TMEM192-HA), a lysosomal protein we expressed in wildtype (B), ATRAID KO (C) and SLC37A3 KO (D) HEK 293T cells and used as a handle to immunoprecipitate lysosomes, shown with a lysosomal marker, LAMP2, demonstrating correct localization of TMEM192-HA to lysosomes and, consequently, successful purification of lysosomes in the lysosome-purification-based uptake assay. Scale bars represent 10 µm. Each image displayed is the representative example chosen from at least five similar images.