Cysteamine–bicalutamide combination therapy corrects proximal tubule phenotype in cystinosis

Abstract Nephropathic cystinosis is a severe monogenic kidney disorder caused by mutations in CTNS, encoding the lysosomal transporter cystinosin, resulting in lysosomal cystine accumulation. The sole treatment, cysteamine, slows down the disease progression, but does not correct the established renal proximal tubulopathy. Here, we developed a new therapeutic strategy by applying omics to expand our knowledge on the complexity of the disease and prioritize drug targets in cystinosis. We identified alpha‐ketoglutarate as a potential metabolite to bridge cystinosin loss to autophagy, apoptosis and kidney proximal tubule impairment in cystinosis. This insight combined with a drug screen revealed a bicalutamide–cysteamine combination treatment as a novel dual‐target pharmacological approach for the phenotypical correction of cystinotic kidney proximal tubule cells, patient‐derived kidney tubuloids and cystinotic zebrafish.


Introduction
Nephropathic cystinosis (MIM219800) is a lysosomal storage disease (LSD) caused by mutations in CTNS, a gene that codes for the lysosomal cystine/proton symporter cystinosin (Town et al, 1998). The loss of cystinosin leads to the lysosomal accumulation of cystine throughout the body and causes irreversible damage to various organs, particularly the kidneys (Cherqui & Courtoy, 2017). The first clinical signs develop during infancy (age of~6 months) in the form of renal Fanconi syndrome and over time patients develop chronic kidney disease and finally renal failure in the first or second decade of life (Gahl et al, 2002). For the past decades, great efforts have been directed towards reducing cellular cystine accumulation in cystinotic patients. The cystine-depleting drug cysteamine can efficiently lower the lysosomal cystine levels and postpone disease progression. However, it poses serious side effects and does not correct the established proximal tubulopathy associated with cystinosis (Gahl et al, 2002;Brodin-Sartorius et al, 2012).
Recent discoveries in lysosomal gene expressions and their regulatory networks have provided new insights in the studies of LSDs, including cystinosis (Gabande-Rodriguez et al, 2014;Lieberman et al, 2012;Canonico et al, 2016). Several studies based on in vitro and in vivo models have demonstrated that the loss of cystinosin is indeed associated with disrupted lysosomal autophagy dynamics, accumulation of distorted mitochondria, and increased reactive oxygen species (ROS) levels, leading to abnormal proliferation and dysfunction of proximal tubule cells Sansanwal et al, 2010;Sansanwal & Sarwal, 2012;Ivanova et al, 2015;Festa et al, 2018). Regardless of the observed cellular defects associated with cystinosis, the mechanisms linking cystinosin loss, lysosomal defects and epithelial dysfunction remain unknown, hampering the development of an enduring intervention to combat the disease.
In this work, we aimed to develop a novel pharmacological strategy to treat cystinosis by correcting its kidney tubule phenotype. By combining CRISPR/Cas9 technology and a dual proteomics and metabolomics approach, we identified alpha-ketoglutarate (aKG) as a novel metabolite affected in cystinosis which is also known to play a pivotal role in the regulation of autophagy and apoptosis. Using this knowledge, we discovered the bicalutamide-cysteamine combination treatment as a novel strategy for the phenotypical correction of cystinotic proximal tubule cells. This approach was validated further in cystinotic patient-derived kidney tubuloids and in cystinotic zebrafish, demonstrating the therapeutic potential for this combination therapy to treat patients with cystinosis.

CRISPR-generated CTNS À/À ciPTEC display increased cystine accumulation and impaired lysosomal autophagy dynamics
To understand the effect of CTNS loss on proximal tubule cell function, we used two well-characterized conditionally immortalized proximal tubule epithelial cell (ciPTEC) lines previously generated from urine samples of a cystinosis patient (ciPTEC CTNS Patient ) and an age-matched healthy volunteer (ciPTEC CTNS WT ) Wilmer et al, 2010;Wilmer et al, 2011). However, as these cell lines were obtained from different individuals, the genetic variation is likely to introduce phenotypical changes independent of CTNS loss. To overcome this limitation, we created an isogenic CTNSdeficient cell line (ciPTEC CTNS À/À ) by specifically knocking out CTNS in the control ciPTEC. A guide RNA (gRNA) targeting exon 4 of the CTNS gene was used to introduce mutations by CRISPR/Cas9 in the ciPTEC CTNS WT . After cell sorting and subsequent clonal cell expansion, three clones with biallelic mutations (lines 3, 7 and 35) were initially selected and all displayed a similar phenotype (Fig EV1), therefore, line 3 (hereafter referred to as CTNS À/À ) was used for subsequent experiments. This model enables direct evaluation of the effect of CTNS loss on proximal tubule epithelial cells independent of chronic exposure to other disease related changes in the body. As a reference, we also included the non-isogenic patientderived cystinotic ciPTEC CTNS Patient line bearing the homozygous 57-kb CTNS deletion (Peeters et al, 2011;Wilmer et al, 2011). In accordance with the pathophysiology of the disease, CTNS À/À cells displayed significantly increased levels of cystine compared to control CTNS WT cells (5.19 AE 0.30 versus 0.05 AE 0.02 nmol/mg protein), comparable to CTNS Patient cells ( Fig 1A). Next, we evaluated the effect of CTNS loss on mammalian target of rapamycin complex 1 (mTORC1) (Ivanova et al, 2015;Andrzejewska et al, 2016;Hollywood et al, 2020). Under normal conditions, mTORC1 is bound to the lysosomes and is responsible for regulating a wide range of cellular processes, including autophagy (Laplante & Sabatini, 2013;Martina & Puertollano, 2013). In the presence of nutrients (standard medium, fed condition), mTOR located to the lysosomal membranes of CTNS WT cells ( Fig EV2). Upon starvation (ÀAA), mTOR was released from the lysosomes and relocalized upon reintroduction of nutrients. In contrast, in CTNS À/À and CTNS Patient cells the fed condition revealed a less pronounced colocalization, and no difference was seen between the fed and starved condition ( Fig EV2). Accurate measurement of the lysosomal size and quantifying the colocalization with mTOR was not feasible due to high level of clusterization of endosomal vesicles. We further evaluated mTOR activity in the cells by tracking the subcellular localization of transcription factor EB (TFEB). If mTOR is deactivated, unphosphorylated TFEB can translocate to the nucleus, where it regulates gene transcription and activates autophagy. A~2.5-fold increase in TFEB nuclear translocation was observed after transfection with TFEB-GFP in CTNS À/À and CTNS Patient cells compared to CTNS WT cells ( Fig 1B). As TFEB will downregulate its own expression after activation (Rega et al, 2016), we found that the endogenous TFEB mRNA expression was also reduced in CTNS À/À and CTNS Patient cells (twofold) compared to control cells ( Fig 1C). During autophagy, LC3-II is recruited to autophagosomes and p62/SQSTM1 is degraded after the fusion of autophagosomes with the lysosomes (Tanida et al, 2008). Although levels of LC3-II and p62/SQSTM1 are similar at baseline, blocking the lysosomal fusion with bafilomycin showed a significant increase in both markers in CTNS À/À and CTNS Patient cells compared to CTNS WT cells (Fig 1D-H), indicating increased ▸ Figure 1. CRISPR-generated CTNS À/À ciPTEC demonstrate cystinosis phenotype.

A
Quantification of cystine levels (nmol/mg protein) in control (CTNS WT ) (n = 4), CRISPR-generated cystinotic cells (CTNS À/À ) (n = 6), and patient-derived cystinotic cells (CTNS Patient ) (n = 4). B Quantification of transcription factor EB (TFEB)-GFP nuclear translocation in CTNS WT , CTNS À/À , and CTNS Patient , respectively. Data are demonstrated as the ratio between number of the cells with nucleus-TFEB positive over the total number of TFEB-transfected cells. The ratios were then presented as a fold change compared to control cells (n = 3). C TFEB mRNA expression in CTNS À/À and CTNS Patient cells compared to control cells (n = 3). D, E Representative confocal micrographs and quantification of LC3-II accumulation in CTNS WT and CTNS À/À cells in presence and absence of 25 nM bafilomycin (BafA1) for 4 h, respectively (n = 3). Scale bars are 20 µm. F-H Western blotting and densitometric analyses for LC3-II/LC3-I ratio and SQSTM1 protein levels in CTNS WT , CTNS À/À and CTNS Patient cells cultured in the presence or in the absence of 25 nM BafA1 for 4 h, respectively (n = 3).
Data information: Data are expressed as mean AE SEM. *P-values < 0.05 were considered to be significant. One-way ANOVA with Dunnett's correction (A, B, C, E, G, H, I and J) or Unpaired t-test (E, G and H). Exact P-values and statistical tests are listed in Appendix Table S1. Source data are available online for this figure.
autophagic flux (Yoshii & Mizushima, 2017). Next, we evaluated the ability of these cells to process BODIPY dye-conjugated bovine serum albumin (DQ BSA), a dye that is endocytosed and becomes fluorescent after degradation inside the lysosomes. A delayed lysosomal cargo degradation (~2.5-fold) of CTNS À/À and CTNS Patient cells compared to control cells was observed ( Fig 1I). CTNS Patient cells, but not CTNS À/À cells, demonstrated a decreased BSA uptake compared to control cells (Fig 1J), indicating the defect is related to degradation rather than protein uptake in CTNS À/À cells. Overall, our data confirm that the CRISPR-generated cystinotic ciPTEC are a good isogenic model of healthy controls and they show a robust cystinotic phenotype of increased cystine accumulation, autophagy activation and reduced lysosomal cargo degradation.

Metabolomic and proteomic profiling reveal alpha-ketoglutarate accumulation in cystinosis
To gain in-depth knowledge of the affected cellular pathways and proximal tubule impairments in cystinosis, we performed targeted metabolomic profiling of 100 key metabolites (Dataset EV1) and untargeted proteomics in the three cell lines (Fig 2). Principal component analysis (PCA) of the metabolites and over 4,500 identified proteins (Fig 2A and F) showed that CTNS Patient cells account for most of the variability in the data, indicating that the different genetic background of the CTNS Patient cells affect the data more than the CTNS loss itself. This was further visualized by unsupervised hierarchical clustering in which the isogenic CTNS À/À cells clustered with CTNS WT rather than CTNS Patient cells ( Fig 2B, and Appendix Fig S1). To explore which pathways are directly linked to CTNS loss, we focused on the metabolites and proteins that were significantly altered in CTNS À/À cells compared to the CTNS WT . Pathway enrichment analysis of the metabolites distinctively affected in CTNS À/À cells revealed several affected pathways, including cystine/cysteine metabolism, the TCA cycle, glycerolipid metabolism and amino acid metabolism ( Fig 2C). Proteomic profiling, on the other hand, revealed a total of 337 proteins that were up-or downregulated in CTNS À/À cells compared to control cells ( Fig 2G). The differentially abundant proteins were then subjected to gene ontology (GO) classification via the Panther Classification System database (Thomas et al, 2003) to highlight their molecular role in the cell ( Fig 2H). The analysis showed an overall reduction in proteins involved in lipid metabolism and the breakdown of micromolecular cellular structures (peptidases, amino acid catabolism, reductases and hydrolases), and an increase in the glycolysis, TCA cycle, DNA replication and DNA repair, in line with the metabolomic data. Consistent with the reduced cargo degradation in cystinotic cells, we found an overall reduction in lysosomal catalytic proteins expression including, lysosomal acid lipase (LIPA), lysosomal acid phosphatase (ACP2), and the potentially representative cysteine protease, cathepsin S (CTSS) (Fig 2I). Moreover, cationindependent mannose 6-phosphate receptor (IGF2R) was found downregulated in CTNS À/À cells (Fig 2I). IGF2R is responsible for the delivery of many newly synthesized lysosomal enzymes from Golgi to the lysosome, and its downregulation could result in decreased lysosomal activity and production of ROS (Probst et al, 2006;Takeda et al, 2019). When cross-checking the affected metabolites of both CTNS À/À and CTNS Patient cells, we identified a set of key metabolites that were significantly changed as a result of CTNS loss: cystine, cysteine, aKG, serine, hypoxanthine and oleoyl-L-carnitine levels were increased, whereas the levels of betaine, L-carnitine and glucosamine-6-phosphate were decreased in both CTNS À/À and CTNS Patient compared to control cells ( Fig 2D and Appendix Fig S2). aKG was of particular interest as aKG is known to play a central role in the TCA cycle, maintaining the redox balance, regulating mTOR and autophagy. Moreover, aKG is a well-known antioxidant and its cellular levels can be increased in response to oxidative stress, contributing to the prevention and/or treatment of several disorders induced by such stress (Mailloux et al, 2009;Satpute et al, 2010;Starkov, 2013). This is done through downregulation of the mitochondrial enzyme alpha-ketoglutarate dehydrogenase (AKGDH), reported to be severely diminished in human pathologies where oxidative stress is thought to play a vital role (Starkov, 2013). Indeed, we found a decreased expression of AKGDH in CTNS À/À cells ( Fig 2I). We also observed a significantly increased level of aKG in plasma of cystinotic patients compared to healthy controls ( Fig 2E, and Table 1), signifying that loss of CTNS also leads to an increase in aKG levels in patients. This prompted us to evaluate further the effect of increased levels of aKG in cystinosis pathology.

Alpha-ketoglutarate regulates the phenotypic alterations in cystinosis
The link between cystinosin loss-of-function and increased oxidative stress has been long-established (Rizzo et al, 1999;Chol et al, 2004;Levtchenko et al, 2005;Wilmer et al, 2011). In line with this, both CTNS À/À and CTNS Patient cells presented increased ROS levels (~1.5fold) compared to control cells (Fed condition; Fig 3A). Starvation A Principal component analysis (PCA) of control (CTNS WT ), CRISPR-generated cystinotic (CTNS À/À ) and patient-derived cystinotic (CTNS Patient ) cells based on the metabolites measured. Each dot represents one sample, and the dots of the same colour are biological replicates (n = 3). B Heat map analysis of top 25 metabolites distinctively expressed in control and CTNS À/À cells. Metabolites significantly decreased (P < 0.05) were displayed in green, while metabolites significantly increased (P < 0.05) were displayed in red (n = 3). C Global test pathway enrichment analysis of the intracellular metabolic interactions distinctively affected in CTNS À/À cells compared to healthy control cells (n = 3).
Larger circles further from the y-axis and orange-red colour show higher impact of pathway. D List of metabolites that were shared and significantly altered in both the CTNS À/À and CTNS Patient cells compared to control cells (n = 6). E Plasma levels of aKG (µM) in healthy individuals (n = 4) and cystinotic patients (n = 6). All cystinotic patients were on cysteamine treatment at the time of blood sampling. For this experiment, non-parametric Mann-Whitney t-test was used to demonstrate the significance. F Principal component analysis (PCA) of the measured proteins in CTNS WT , CTNS À/À and CTNS Patient (n = 3). G Volcano plot illustrates differentially abundant proteins (n = 3). The -log 10 (Benjamini-Hochberg corrected P-value) is plotted against the log 2 (fold change: CTNS WT / CTNS À/À ). H Gene ontology (GO) analysis illustrates classes of proteins differing between CTNS WT and CTNS À/À cells (n = 3). I List of proteins that were significantly upregulated and downregulated in CTNS À/À cells compared to control cells (n = 3). AKGDH; Alpha-ketoglutarate dehydrogenase, LIPA; Lysosomal acid lipase, ACP2; Lysosomal acid phosphatase, CTSS; Cathepsin S, CTSC; Dipeptidyl peptidase, IGF2R; Cation-independent mannose-6-phosphate receptor, SORT1; Sortilin, CAT; Catalase, CTSA; Lysosomal protective protein and SOD; Superoxide dismutase.
Data information: Data are expressed as mean AE SEM. *P-values < 0.05 were considered to be significant. One-way ANOVA with Dunnett's correction (A, B, C, D, E, F, G, H and I). Exact P-values and statistical tests are listed in Appendix Table S1. Of note, exposure to DMKG for 4 h was not toxic in any of the three cell lines ( Fig EV3A). However, adding DMKG during the 24 h starvation had little to no effect in control cells, but led to massive cell death and apoptosis activation in cystinotic cells (Figs 3E-G, and EV3B and C). We, therefore, hypothesized that the increased level of aKG could be linked to autophagy and promoting cell death in cystinotic proximal tubule cells (Duran et al, 2012;Chin et al, 2014;Villar et al, 2017). DMKG exposure indeed resulted in an increased level of LC3-II/LC3-I ratio in CTNS À/À and CTNS Patient but not in control cells, confirming that DMKG induces autophagy in cystinotic cells (Fig 3H and I). In light of our data, we identified aKG as metabolite bridging cystinosin loss to increased oxidative stress, activation of autophagy and proximal tubule dysfunction in cystinosis.
Bicalutamide and cysteamine combination treatment phenotypically corrects CTNS À/À proximal tubule cells Cysteamine is used to efficiently reduce cystine accumulation in patients with cystinosis, but it cannot correct the established proximal tubulopathy associated with the disease. We hypothesized that this may be linked to the inability of this drug to lower aKG and target other metabolic pathways associated with CTNS loss. Cysteamine, as expected, boosted GSH and lowered cystine and cysteine levels. However, it had no significant impact on the other metabolites or on the proteome profiles (Figs 4A-E and EV4A). We, therefore, screened different candidate drugs based on their ability to reduce cystine and aKG levels and to restore the metabolic profile using metabolomics (Dataset EV1). The concentrations of the drugs tested were within a non-cytotoxic range ( Fig EV3D-J). Among the candidate drugs tested, bicalutamide, an anti-androgenic agent, did not restore the high cystine and cysteine levels, however improved the overall metabolic phenotype, including aKG, serine, betaine and oleoyl-L-carnitine ( Fig 4A). Treatment with bicalutamide alone also resulted in the upregulation of several metabolic enzymes involved in cysteine conversion, lysosomal degradation and the TCA cycle, including AKGDH (Figs 4C-E and EV4B). Furthermore, the cysteamine-bicalutamide combination treatment showed an additive effect and unsupervised clustering shows that the cells receiving the combination therapy are more similar to the CTNS WT cells than the other conditions ( Fig 4B). The combination treatment also led to the upregulation of more than 50 proteins, revealing an enrichment of proteins involved in the TCA cycle and in the metabolism of macromolecules such as lipids and vitamins that were shown to be ▸ Figure 3. aKG is a key metabolite responsible for impaired autophagy and proximal tubule dysfunction in cystinotic proximal tubule cells.
Data information: Data are expressed as mean AE SEM. P-values < 0.05 were considered to be significant. One-way ANOVA with Dunnett's correction (A, B, C, D, E, G and I) or Unpaired t-test (B, C, D, E and G). Exact P-values and statistical tests are listed in Appendix Table S1. Source data are available online for this figure.
6 of 20 EMBO Molecular Medicine 13: e13067 | 2021 ª 2021 The Authors downregulated in CTNS À/À cells (Figs 4C-E and EV4C). As bicalutamide reduced the elevated level of aKG in cystinotic cells, we hypothesized that it could also resolve aKG-mediated downstream effects. Indeed, bicalutamide (but not cysteamine) reduced the DMKG-mediated cell death and increase in LC3-II/LC3-I ratio in CTNS À/À cells (Fig 5A and B). Although cysteamine alone showed no effect on autophagy, combined with bicalutamide it additively reduced the DMKG-mediated increase in LC3-II/I in CTNS À/À cells compared to bicalutamide alone ( Fig 5B). Notably, cysteamine, bicalutamide and their combination had no effect on the basal autophagy activity in absence of bicalutamide ( Fig EV4D). We next tested whether bicalutamide alone or in combination with cysteamine could recover the antioxidant effect of aKG in cystinotic cells. Similar to cysteamine, bicalutamide was able to retrieve the effect of DMKG in CTNS À/À cells (2.1-fold; Fig 5C). Further reduction in ROS levels was observed when bicalutamide and cysteamine were given together. Of note, bicalutamide alone had no effect on GSH levels ( Fig 4A) but could significantly lower the ROS levels in presence of DMKG in cystinotic cells (Fig 5C), suggesting that bicalutamide, through a yet unknown mechanism, restores the antioxidant property of aKG in CTNS À/À cells.
Recently, bicalutamide has been patented for the treatment of several LSDs, promoting mTOR-associated autophagy and TFEBmediated cellular exocytosis (Farrera-Sinfreu et al, 2014), and therefore allowing the relief and/or treatment of the symptoms of many LSDs. Indeed, bicalutamide was able to induce TFEB nucleus translocation in both control and CTNS À/À cells and downregulate TFEB mRNA expression in control cells (and Fig EV4E and F). In agreement with the increased TFEB activity, bicalutamide also restored endocytic cargo processing in cystinotic cells (Fig 5D and E), allowing us to test whether the increased degradation of lysosomal cargo results in decreased cystine accumulation. We used a more sensitive and quantitative LC-MS/MS method (Jamalpoor et al, 2018) to measure cystine levels in cystinotic cells upon treatment with cysteamine, bicalutamide, or their combination. In line with metabolomics data, CTNS À/À cells treated with bicalutamide alone did not exhibit any reduction in cystine levels. Cysteamine alone resulted in a reduction of cystine in CTNS À/À cells (0.37 AE 0.08 nmol/mg protein; Fig 5F), but this was still higher (7.5-fold) than that found in control cells (0.05 AE 0.02 nmol/mg protein). Interestingly, combination treatment of bicalutamide and cysteamine resulted in a 2.8-fold decrease in cystine when compared to cysteamine alone (0.13 AE 0.003 versus 0.37 AE 0.08 nmol/mg protein; Fig 5F), bringing lysosomal cystine close to control. This effect was also found in CTNS Patient cells (Fig 5G).
Bicalutamide and cysteamine combination treatment efficiently lowers cystine and aKG levels in patient-derived cystinotic kidney tubuloids Next, we evaluated the safety and efficacy of cysteamine-bicalutamide combination treatment in patient-derived tubuloids. Tubuloids are an advanced in vitro model that was successfully used before to model genetic, infectious and malignant kidney disease and to screen for therapeutic efficacy (Schutgens et al, 2019). We established kidney tubuloids from urine samples of two paediatric cystinotic patients (CTNS Patient-1 and CTNS Patient-2 ) and compared their characteristics to tubuloids derived from healthy kidney tissue of two donors (CTNS WT-1 and CTNS WT-2 ). Urine-derived tubuloids were positive for paired-box gene 8 (PAX8) and negative for tumour protein p63 (Fig 6A), confirming that these structures indeed consisted of kidney epithelium (PAX8 + /TP63 À ) and not urothelium (PAX8 À /TP63 + ) (Saito et al, 2006;Albadine et al, 2010). Moreover, both patient and control tubuloids robustly expressed markers of the proximal tubule, loop of Henle, distal tubule and collecting duct ( Fig EV5A) (Schutgens et al, 2019). Both CTNS Patient tubuloids ◀ Figure 4. Bicalutamide-cysteamine combination treatment corrects the metabolome and proteome profile of cystinotic proximal tubule cells.
Data information: Data are expressed as mean AE SEM. P-values < 0.05 were considered to be significant. One-way ANOVA with Dunnett's correction (A,B, C, D and E). * significantly different from CTNS WT cells (P < 0.05). # significantly different from CTNS À/À cells (P < 0.05). Exact P-values and statistical tests are listed in Appendix Table S1.
Data information: Data are expressed as mean AE SEM. *P-values < 0.05 were considered to be significant. One-way ANOVA with Dunnett's correction (A, B, C, D, E, F and G) or unpaired t-test (B, C, E, F and G). Exact P-values and statistical tests are listed in Appendix Table S1. Source data are available online for this figure. displayed the cystinotic phenotype with increased cystine accumulation (~7-fold) compared to the controls (Fig 6B). In line with the previous findings in ciPTEC, treatment of CTNS Patient tubuloids with a non-nephrotoxic concentration of both cysteamine and bicalutamide (Fig EV5B-D) resulted in a more potent reduction in cystine levels (~2-fold) than with cysteamine treatment alone (Fig 6C and  D). Furthermore, we performed targeted metabolomic profiling (Dataset EV2) and identified that bicalutamide (but not cysteamine), either alone or in combination with cysteamine, reduced the aKG levels in CTNS Patient-2 tubuloids (Fig 6E).
Bicalutamide and cysteamine combination treatment efficiently lowers cystine levels and improves survival of cystinotic zebrafish Finally, we assessed the combination therapy in cystinotic zebrafish, a well-characterized in vivo model of cystinosis (Elmonem et al, 2017;Elmonem et al, 2018). Cystinotic zebrafish displayed significantly increased levels of cystine, dysmorphic features, delayed hatching and reduced survival compared to controls (Figs 6F-K and EV5E-G). As expected, cysteamine was able to reduce cystine levels (1.5-fold; Fig 6H) and a trend towards an additive effect of bicalutamide was observed (1.8-fold reduction for the combination), though this was not significant. However, bicalutamide either alone or in combination with cysteamine significantly improved the survival and reduced the percentage of dysmorphism in the cystinotic zebrafish without having an effect on their hatching rate, when compared to cysteamine monotherapy (Fig 6I-K). This indicates that the combination treatment does not induce toxicity and is beneficial in cystinotic zebrafish compared to cysteamine treatment alone.

Discussion
In this work, we identified a new therapeutic target in nephropathic cystinosis by evaluating the persistent cellular link between cystinosin loss-of-function and proximal tubule cell dysfunction. Using an omics-based strategy, we identified several proteins and metabolites to be consistently altered across the proximal tubule cell models used. Increased levels of aKG were particularly of interest as this metabolite is involved in autophagy regulation, oxidative stress and apoptosis, which are known to be dysregulated in cystinosis.
Our results also revealed that a bicalutamide-cysteamine combination treatment could provide a novel pharmacological approach for the phenotypical correction of cystinosis. Consistent with previous reports, we found that CTNS-deficient cells show high levels of cystine accumulation, activation of autophagy, delayed protein degradation and an increase in baseline ROS levels (Rizzo et al, 1999;Chol et al, 2004;Levtchenko et al, 2005;Settembre et al, 2008;Wilmer et al, 2011;Platt et al, 2012;Raggi et al, 2014;Ivanova et al, 2015;Festa et al, 2018). This was confirmed further by metabolic and proteomic analyses, where we found a reduction in lysosomal catabolic proteins and an upregulation of enzymatic antioxidizing agents such as catalase and superoxide dismutase in cystinotic cells, which was consistent with delayed lysosomal cargo degradation and increased oxidative stress, respectively. In addition, we found that aKG levels are increased in CTNSdeficient cells and in cystinotic patients, a metabolite known for its antioxidant properties and potential for the treatment of disorders induced by oxidative stress (Mailloux et al, 2009;Satpute et al, 2010;Starkov, 2013;Liu et al, 2018). During starvation, both control and CTNS-deficient cells showed a strong increase in ROS (after 4 h), and the CTNS-deficient cells seem to handle this well with only a slight reduction in cell viability compared to control cells after 24 h. Addition of DMKG (to increase intracellular aKG levels) during starvation led to a strong reduction in ROS in CTNS WT cells but had a minor effect in CTNS À/À cells and even an opposite effect in CTNS Patient cells. Even though DMKG had a minor effect on ROS levels in CTNS À/À cells during short-term exposure, it still led to massive cell death after 24 h in CTNS-deficient cell lines.
Many studies have investigated the role of aKG in response to ROS, mTOR and autophagy regulation, but the outcomes vary greatly depending on the context and the cell type studied. Overall, aKG is regarded as a health-promoting metabolite, reducing liver fibrosis (Zhao et al, 2016), delaying age-related disease and increasing life span in in vivo models such as Caenorhabditis elegans and mice (Chin et al, 2014;Su et al, 2019) and as a potential anti-cancer ▸ Figure 6. Cysteamine-bicalutamide combination treatment is effective in patient-derived cystinotic tubuloids and cystinotic zebrafish.
Data information: Data are expressed as mean AE SEM. *P-values < 0.05 were considered to be significant. One-way ANOVA with Dunnett's correction (B, C, D, E, G, H and J) or unpaired t-test (C, D and H). Exact P-values and statistical tests are listed in Appendix Table S1.  (2017) showed that addition of DMKG in cancer cells for 24 h inhibited autophagy and induced glutaminolysis-mediated apoptosis during starvation. In our cells, addition of bicalutamide but not cysteamine could prevent the DMKG induced cell death and reduced LC3-II/LC3-I levels. As cysteamine alone is able to efficiently induce glutathione synthesis in cystinotic cells, it is likely that the DMKG decreased cell viability was not due to a reduced availability of antioxidants. The marked increase in LC3-II/LC3-I levels in CTNS-deficient cells indicates activation of autophagy. However, the reduced protein degradation (shown here by DQ-BSA analysis and in previously published reports) (Ivanova et al, 2015;Festa et al, 2018) question the efficiency of this autophagic process. The high intracellular levels of aKG together with a defective autophagy may therefore induce glutaminolysis-mediated apoptosis during starvation specifically in cystinotic cells. As bicalutamide is known to upregulate the autophagic flux (Hao et al, 2017), the addition of bicalutamide could restore the autophagic flux in cystinotic cells and prevent this glutaminolysis-mediated apoptosis.
Several reports also indicate a role for mitochondria in cystinosis disease pathology, including affected mitochondrial function and dynamics (De Rasmo et al, 2019), as well as reduced autophagymediated clearance of damaged mitochondria resulting in increased ROS in their cells (Sansanwal et al, 2010;Festa et al, 2018). This is also in line with the decreased expression of AKGDH in our CTNS À/-À cells, which can then lead to the observed increase in aKG.
However, other TCA metabolites that we evaluated (cis-aconitate, succinic acid, fumarate, malate and citrate) remained unchanged. As there are multiple pathways in the cell that can either generate or catabolize aKG, it is also possible that another pathway is responsible for the increase in aKG. Studying the role of mitochondria in the progression of cystinosis remains important to further expand our knowledge on the complexity of the disease and prioritize drug targets for cystinosis.
Recent studies show that overexpression of TFEB activity stimulates lysosomal excretion and rescues the delayed endocytic cargo processing, allowing the relief and/or treatment of the symptoms of many LSDs (Platt et al, 2012). This in line with the study of Rega et al (2016), where stimulation of endogenous TFEB activity by genistein was shown to lower cystine levels and rescue the delayed endocytic cargo processing in cystinotic proximal tubule cells. Furthermore, inhibition of mTOR signalling by everolimus was shown to activate autophagy, rescue the number of large lysosomes, and in combination with cysteamine, reverse the cystine/cysteine loading defect in patient-specific and CRISPR-edited cystinotic induced pluripotent stem cells and kidney organoids (Hollywood et al, 2020).
Bicalutamide has recently been patented for the treatment of several LSDs, promoting mTOR-associated autophagy and cellular exocytosis (Farrera-Sinfreu et al, 2014). Here we find that in CTNS À/À cells, bicalutamide was able to increase endogenous TFEB activity, restore endocytic cargo processing and, in combination with cysteamine, normalize lysosomal cystine levels in both patientderived and CRISPR-edited cystinotic proximal tubule cells. Furthermore, cysteamine-bicalutamide combination treatment, but not cysteamine alone, reversed the proteomic and metabolic phenotype and resulted in reduction of aKG levels, and upregulation of AKGDH and IGF2R proteins, resolving the aKG-mediated downstream effects in cystinotic proximal tubular cells (Fig 7). So far, several murine models for cystinosis have been developed that recapitulate some of the clinical features seen in patients with cystinosis, including high tissue cystine levels (Johnson et al, 2013;Gaide Chevronnay et al, 2014;Gaide Chevronnay et al, 2015;Napolitano et al, 2015). However, the renal phenotype is often mild, has a late onset or lacks signs of a proximal tubulopathy or renal failure (Cherqui et al, 2002). Furthermore, unlike the human situation where treatment with cysteamine does not improve the renal phenotype and only improves some parts of the cystinotic phenotype (Cherqui, 2012), it can completely resolve any renal symptoms in mice. This means that the unmet medical need that is currently present in humans is not observed in Ctns À/À mice, making this model unsuitable for studying any drug combination therapy aimed at improving the renal phenotype. Hence, to extend our knowledge on the beneficial effect of the combination treatment and bring the therapy one step closer to clinical application, we evaluated the safety and efficacy in patient-derived tubuloids, an advanced in vitro model, and in vivo in cystinotic zebrafish. Tubuloids offer several advantages over (iiii) Tubuloids are grown in 3D, allowing neighbouring cells to interact in a more physiological way than in conventional 2D culture models. (iii) Tubuloids are not genetically modified or reprogrammed and can accurately mimic the donor genotype and phenotype, thereby allowing personalized medicine (Schutgens et al, 2019). Zebrafish models may generate symptoms that are not directly linked to clinical features in patients (Santoriello & Zon, 2012;Vliegenthart et al, 2014), but are valuable both for screening drug toxicity and as genetic disease models (MacRae & Peterson, 2015). The ctns À/À zebrafish presents a robust and versatile model of cystinosis with early phenotypic characteristics of the disease that can be used for the in vivo screening of novel therapeutic agents (Elmonem et al, 2017;Elmonem et al, 2018). The fact that the improved cystine lowering-efficacy of the combination treatment could be reproduced in tubuloids from two different patients and in cystinotic zebrafish, underlines the robustness of these findings and increases the likelihood that this treatment can be successfully extrapolated to cystinosis patients. Bicalutamide, a non-steroidal anti-androgenic agent, is also one of the most widely prescribed drugs for treating prostate cancer (Osguthorpe & Hagler, 2011). The use of any androgen deprivation therapy is generally associated with the risk of acute kidney injury (AKI) (Lapi et al, 2013;Peng et al, 2019), and up to 36% of patients who have taken bicalutamide for 1 to 6 months may experience AKI (Peng et al, 2019). However, this association is mainly driven by a combination of gonadotropin-or luteinizing-releasing hormone agonists with bicalutamide (Lapi et al, 2013). Bicalutamide-cysteamine combination treatment at an effective concentration was not only found to be safe in our models after short-term exposure, but also improved the survival and reduced the percentage of dysmorphism in the cystinotic zebrafish, confirming the safety profile of bicalutamide both in in vitro and in vivo. Anti-androgenic therapy is also known to reduce testosterone levels (Braga-Basaria et al, 2006;Lapi et al, 2013), complicating the beneficial effect of bicalutamide in cystinotic male patients with delayed maturation (Fivush et al, 1988;Chik et al, 1993;Winkler et al, 1993). Interestingly, bicalutamide is a racemate and its anti-androgenic activity is attributed solely to the (R)-enantiomer (Tucker & Chesterson, 1988;Mukherjee et al, 1996), while the (S)-enantiomer with little, if any, antiandrogenic activity mediates autophagy effects (Farrera-Sinfreu et al, 2014). Therefore, it would be of great scientific benefit to investigate further the effect of the (S)-bicalutamide and/or develop any structural analogues to treat cystinosis.
Taken together, we identified a new therapeutic target by evaluating the persistent cellular abnormalities using an omics-based strategy. We identified aKG as an important metabolite bridging cystinosin loss to lysosomal autophagy defect, apoptosis activation and proximal tubule cell impairment in cystinosis. Bicalutamide, but not cysteamine, was able to reduce aKG levels and resolve aKGmediated downstream effects in cystinotic cells. Bicalutamide in combination with cysteamine demonstrated additively reduced cystine levels both in vitro and in vivo, suggesting this combination therapy holds great potential to treat patients with cystinosis. Preclinical studies in a suitable animal model should determine whether cysteamine-bicalutamide combination therapy is able to improve or prevent the development of renal Fanconi syndrome and renal failure. Translation to the clinic should be further facilitated by the fact that bicalutamide is already an approved drug with a known safety profile, although a separate study should be performed to determine the appropriate dose for cystinotic patients.
Generation of CTNS À/À isogenic cell line of ciPTEC using CRISPR Guide RNAs (gRNAs) targeting exon 4 of the CTNS gene were designed using the online gRNA designing tool available at chopchop.cbu.uib.no. In order to maximize specificity, guide sequences with high scores for on-target efficiency and no predicted off-targets having at least 3 base pair mismatches in the genome were selected. Optimal gRNA (5 0 -GTCGTAAAGCTGGAGAACGG-3 0 ) was cloned into the pSPCas9(BB)-2A-GFP plasmid (Addgene #48138) as described previously by Ran et al (2013) and introduced into healthy ciPTEC using PolyPlus JetPrime. 72 h post-transfection, GFP-positive singlet cells were sorted using FACS Aria-II flow cytometer and expanded in 96-wells plate. The gRNA cut site was amplified with PCR using the primers flanking the cut region (F.CTNS_ex4 5 0 -GGCCTGTTTTCCTCCATCTCTG-3 0 ; R.CTNS_ex4 5 0 -AAGTGCCAACCAGCAGCTC-3 0 ). Knockouts were confirmed by Sanger sequencing followed by intracellular cystine accumulation.

CiPTEC culture
The ciPTEC used in this study (ciPTEC CTNS WT and ciPTEC CTNS patient ) were obtained from Cell4Pharma (Nijmegen, The Netherlands, MTA #A16-0147). These cell lines have been generated previously from proximal tubule cells isolated from urine samples of a cystinotic patient and an age-matched healthy control, followed by immortalization Wilmer et al, 2010;Wilmer et al, 2011). In brief, primary cells were immortalized by transfection with SV40T ts A58 (SV40T) and hTERT (human telomerase reverse transcriptase) followed by subcloning to obtain homozygous cell populations. Of each donor, one subclone was selected based on its proximal tubular characteristics including morphology, expression pattern and transport activity. In this study, we used the existing ciPTEC CTNS WT cell line to generate an isogenic CTNS knockout line, named CTNS À/À . Mycoplasma contamination was checked every 2 months and was found to be negative in all cell lines used.
Cells were cultured as previously described by Wilmer et al, 2010(Wilmer et al, 2010. The culture medium was Dulbecco's modified Eagle medium DMEM/F-12 (GIBCO) supplemented with foetal calf serum 10% (v/v), insulin 5 µg/ml, transferrin 5 µg/ml, selenium 5 µg/ml, hydrocortisone 35 ng/ml, epidermal growth factor 10 ng/ml and tri-iodothyronine 40 pg/ml. In short, cells were seeded at a density of 55,000 cells/cm 2 and grown at 33°C for 24 h to enable them to proliferate and subsequently cultured at 37°C for 7 days to mature into fully differentiated epithelial cells.

Cell treatment
Standard starvation medium was Hank's balanced salt solution (HBSS; GIBCO). When indicated, the cell permeable form of aKG, dimethyl a-ketoglutarate (DMKG) was added to a final concentration of 2 mM for 4 or 24 h. Cysteamine (100 lM), bicalutamide (35 lM), or the combination of cysteamine and bicalutamide (100 and 35 lM, respectively) were used as a treatment for 96 h (5 days).

Zebrafish maintenance and breeding
The animal care and experimental procedures were carried out in accordance with the ethical committee guidelines for laboratory animal experimentation at KU Leuven. Zebrafish (Danio rerio) were AB strain wild-type and ctns À/À mutant. Cystinotic and wild-type larvae were raised at 28.5°C in egg water (Instant Ocean Sea Salts, 60 lg/ml). At 48-h post-fertilization, larvae were treated with cysteamine 1,000 µM, bicalutamide 10 µM or their combination. Drugs were administered at 48-h post-fertilization in all experiments dissolved in the swimming water. The medium was refreshed every day, and dead embryos were sorted out.

Intracellular cystine quantification by HPLC-MS/MS
Cystine levels were quantified using HPLC-MS/MS; a rapid and sensitive assay that has been developed and validated in house (Jamalpoor et al, 2018). In brief, ciPTEC, tubuloids and zebrafish larvae pellets were suspended in N-Ethylmaleimide (NEM) solution containing 5 mM NEM in 0.1 mM sodium phosphate buffer (pH 7.4). The cell suspension was precipitated, and protein was extracted with sulfosalicylic acid 15% (w/v) and centrifuged at 20,000 g for 10 min at 4°C. Protein concentration was determined by the method of the Pierce TM BCA protein assay kit according to the manufacturer's protocol (Thermo Fisher, The Netherlands), and the cystine concentration was measured using HPLC-MS/MS. Data are expressed as the cystine values (nmol) corrected for total protein content (mg).
Tubuloids were grown in expansion medium for a few days and then differentiated for 7 days. Tubuloids were lysed and RNA was isolated using the RNEasy Mini Kit (Qiagen) according to the manufacturer's protocol. Quantitative real-time qPCR was performed using the iQ SYBR Green Supermix (Bio-Rad). A 384-well plate was used with a reaction volume of 12.5 ll and duplicates for each reaction. For the read-out, the CFX384 Touch Real-Time PCR Detection System (Bio-Rad) was used. Expression levels of the following genes were measured: RPS-13 (housekeeping gene), ANPEP, ABCC3, HNF1A, HNF4A, SLC12A1, SLC12A3, CALB1, AQP2 and AQP3. Primer sequences are provided in Table 2. Expression was normalized to the expression of RPS-13 within the same sample (ΔCt).

Immunofluorescence and confocal microscopy
To investigate mTORC1/LAMP1 colocalization, control and cystinotic ciPTEC lines were cultured on coverslips with the respective treatments. Thereafter, cells were fixed with 4% paraformaldehyde in PBS for 10 min, permeabilized with 0.1% Triton-X solution for 10 min and blocked with 1% bovine serum albumin (BSA) diluted in PBS for 30 min. Subsequently, cells were stained with the primary antibodies diluted in blocking buffer overnight at 4°C. After three washes with PBS, the cells were incubated for 2 h at room temperature with the corresponding secondary antibodies. Nuclei were stained with Hoechst 33342 (1 µM), and cells were imaged with a DeltaVision confocal microscope (Cell Microscopy Core, Department of Cell Biology, University Medical Centre Utrecht).
To assess TFEB intracellular distribution, cells were seeded in special optic 96-well plate until to reach 50% confluence. Cells were then transfected with the TFEB-GFP plasmid (a kind gift from Dr. Annelies Michiels (Viral Vector Core, Leuven, Belgium)) using Poly-Plus JetPrime reagent according to the manufacturer's instructions. After 48 h from transient transfection, cells were stained with Hoechst 33342 (1 µM) for 10 min and imaged using a Cell Voyager 7000 (CV7000) confocal microscope (Yokogawa Electric corporation, Tokyo, Japan). TFEB nuclear translocation data are expressed as number of cells with nucleus-TFEB positive over the total number of TFEB-transfected cells.
For tubuloid experiments, tubuloids were differentiated for 7 days. Tubuloids were fixed for 45 min with 4% formaldehyde and then permeabilized and blocked with 0.5% Triton X-100 plus 0.5% BSA in PBS for 30 min. Subsequently, tubuloids were rinsed with 0.5% BSA plus 0.1% Tween-20 in PBS and then stained with the primary antibodies diluted in 0.5% BSA plus 0.1% Tween-20 in PBS overnight at 4°C. Tubuloids were rinsed twice with 0.5% BSA plus 0.1% Tween-20 in PBS and then stained for 2 h at room temperature with the corresponding secondary antibodies and DAPI (1:1,000) in 0.5% BSA plus 0.1% Tween-20 in PBS. Tubuloids were washed twice with 0.5% BSA + 0.1% Tween-20 in PBS, mounted and imaged using a Leica SP8 confocal microscope.

Immunoblots
The ciPTEC were seeded in 6-well plates in triplicate with the respective treatments. Subsequently, cells were washed twice with PBS and lysed with RIPA buffer containing protease inhibitor cocktail (Roche). Protein quantification was performed using the method of the Pierce TM BCA protein assay kit according to the manufacturer's protocol. After the electrophoresis, the proteins were transferred to a nitrocellulose membrane (midi kit, Bio-Rad) with Trans-Blot Turbo Transfer System (Bio-Rad). Finally, membranes were imaged using the ChemiDoc TM XRS+ (Bio-Rad) and analysed using ImageJ software.

Endocytosis assay
The endocytic uptake was monitored in ciPTEC following incubation for 1.5 h at 37°C with 50 lg/ml of either BSA-AlexaFluor-647 (A34785, Thermo Fisher Scientific) or DQ Red BSA (D12051, Invitrogen). The cells were then fixed and stained with Hoechst 33342 (1 µM) for 10 min and imaged using a CV7000 confocal microscope (Yokogawa Electric corporation, Tokyo, Japan). Data were quantified with Columbus TM Image Data Storage and analysis software (PerkinElmer, Groningen, the Netherlands). Data are expressed as the number of BSA/DQ-BSA spots per cell.

Metabolomics
CiPTEC cells were washed with ice cold PBS, and metabolites were extracted in 1 ml lysis buffer containing methanol/acetonitrile/ dH 2 O (2:2:1). Samples were centrifuged at 16,000 g for 15 min at 4°C, and supernatants were collected for LC-MS analysis.
LC-MS analysis was performed on an Exactive mass spectrometer (Thermo Scientific) coupled to a Dionex Ultimate 3000 autosampler and pump (Thermo Scientific). The MS operated in polarityswitching mode with spray voltages of 4.5 kV and À3.5 kV. Metabolites were separated using a Sequant ZIC-pHILIC column (2.1 × 150 mm, 5 lm, guard column 2.1 × 20 mm, 5 lm; Merck) with elution buffers acetonitrile (A) and eluent B (20 mM (NH 4 ) 2 CO 3 , 0.1% NH 4 OH in ULC/MS grade water (Biosolve)). Gradient ran from 20% eluent B to 60% eluent B in 20 min, followed by a wash step at 80% and equilibration at 20%, with a flow rate of 150 ll/min. Analysis was performed using LCquan software (Thermo Scientific). Metabolites were identified and quantified on the basis of exact mass within 5 ppm and further validated by concordance with retention times of standards. Peak intensities were normalized based on total peak intensities, and data were analysed using MetaboAnalyst (Chong et al, 2019).
Peptide samples were analysed with an UHPLC 1290 system (Agilent technologies) coupled to an Orbitrap Q Exactive HF × mass spectrometer (Thermo Scientific). Peptides were trapped (Dr Maisch Reprosil C18, 3 lm, 2 cm × 100 lm) and then separated on an Flow was passively split to 300 nl/min. The mass spectrometer was operated in data-dependent mode. At a resolution of 35,000 m/z at 400 m/z, MS full scan spectra were acquired from m/z 375-1,600 after accumulation to a target value of 3e 6 . Up to 15 most intense precursor ions were selected for fragmentation. HCD fragmentation was performed at normalized collision energy of 27% after the accumulation to a target value of 1e 5 . MS/MS was acquired at a resolution of 30,000. Dynamic exclusion was enabled with an exclusion duration of 32 s. RAW data files were processed with MaxQuant (v1.6.0.16 (Cox, Mann 2008)), and MS2 spectra were searched with the Andromeda search engine against the SwissProt protein database of Homo Sapiens (20,259 entries, downloaded 31/01/2018) spiked with common contaminants. Cysteine carbamidomethylation was set as a fixed modification, and methionine oxidation and protein N-term acetylation were set as variable modifications. Trypsin was specified as enzyme and up to two miss cleavages were allowed. Filtering was done at 1% false discovery rate (FDR) at the protein and peptide level. Label-free quantification (LFQ) was performed, and "match between runs" was enabled. The data were further processed using Perseus 1.6.0.7 (Tyanova et al, 2016). 3 µl of the acidified ultrafiltrate was injected onto a peek-lined InertSustain AQ-c18 (2.1*100 mm dp 3µ) column using a I-Class Acquity (Waters). The column was run at 40°C in gradient mode using 0.5% acetic acid in H 2 O and Acetonitrile (initial conditions 100% 0.5% acetic acid in H 2 O at 250 µl/min). The column flow was directed to a Xevo TQSµ (Waters) fitted with an electrospray ionization probe operating in the negative mode at unit resolution. The capillary voltage was set at 0.6 kV. The temperature settings for the source and ionblock were, respectively, 550 and 150°C. As a drying gas, nitrogen was used at a flow rate of 800 l/h. The conegas flow was set at 50 l/h. The collision cell was operated with argon as the collision gas at a pressure of 0.35 Pa. An area response was generated by recording the MRM transitions of the neutral loss CO 2 for both 2 kg an d4-2 kg. The d4-2 kg was used to normalize the area response. Quantification was performed using a linear regression curve constructed from the normalized area response from the prepared calibration standards.

Cell viability assays
To estimate the percentage of cell death, cells were seeded in a 96well plate and after respective treatments, and the percentage cell death was assessed using Presto Blue Cell Viability Reagent according to manufacturer's instructions.
For tubuloid experiments, the drug safety screening was performed using a protocol derived from Driehuis et al (2019). In short, BME was dissolved by incubation in 1 mg/ml dispase II for 30 min at 37°C. Tubuloids were digested into tiny fragments using Accutase. These fragments were plated in BME, grown for 1 day in expansion medium and then switched to differentiation medium for 2 days. Next, BME was dissolved using dispase II as described above, after which tubuloids were sieved using a 70 lm cell strainer (BD Falcon) and counted. 500 tubuloids in 40 ll differentiation medium with 5% BME were plated in a 384-well plate using the Multidrop dispenser (Thermo Fisher) and treated with increasing concentrations of bicalutamide that were always combined with 100 lM cysteamine. DMSO volume was 1% in all conditions. Drugs were added using the D300E digital dispenser (Tecan). After 5 days, tubuloids were lysed with 40 ll Cell-Titer Glo (Promega). Luminescence was measured using the Sparkâ multimode microplate reader (Tecan) to determine ATP levels as an indicator of the amount of tubuloids that had survived the treatment. Tubuloid viability was normalized to the viability (ATP levels) upon treatment with cysteamine alone (= 100%).

Apoptosis assay
The ciPTEC were plated in an optic 96-well imaging plates and treated with DMKG in presence or absence of bicalutamide, cysteamine or their combination. Subsequently, the Cell Event Caspase-3/ 7 Green Detection Reagent (8 µM) was added and incubated for 30 min before imaging. Caspase-active cells were identified as described in manufacturer's instruction. Each well was imaged using the CV7000 confocal microscope and analysed with Columbus TM Image Data Storage and analysis software (PerkinElmer, Groningen, The Netherlands).

ROS detection assay
The ciPTEC were seeded in 96-well plates and ROS levels were assessed using general oxidative stress indicator (CM-H2DCFDA; Invitrogen) according to the manufacturer's protocol. Briefly, the cells were treated with the CM-H2DCFDA reagent (10 lM) and incubated in the dark. Following 20 min incubation at 37°C, cells were rinsed once with HBSS and incubated with the different treatment conditions as previously stated. Fluorescence was measured using a fluorescent microplate reader (Fluoroskan Ascent, Thermo Fisher Scientific, Vantaa, Finland) at excitation wavelength of 492 nm and emission wavelength of 518 nm. Data are expressed as the fluorescence values normalized to the protein concentration.

Statistical analysis
Statistical analysis was performed using GraphPad Prism 7.0 (GraphPad Software, Inc., USA). Data are presented as mean AE standard error of the mean (SEM) of minimally three independent experiments performed in triplicate. Significance was evaluated using one-way analysis of variance (ANOVA), or where appropriate, unpaired two-tailed Student's t-test was applied. P-values < 0.05 were considered to be significant. The zebrafish sample size was estimated based on the established method published before (Elmonem et al, 2017). In order to minimize the effects of subjective bias, the aquatic facility took care of the mating procedures independently of the subsequent analysis and for each strain, n.6 females and n.4 males were mated at the same time. For each group, zebrafish larvae were selected randomly from different pools of larvae.
For the human studies, cystinosis patients and control subjects were randomly selected and alfa-ketoglutarate measurements were performed by a blinded operator and analysed by an independent investigator. For the alfa-ketoglutarate measurements in the human subjects, an F-test was performed to compare variance between the cystinosis patients and controls. Since the variances were significantly different between the cystinosis and control subjects group, a non-parametric test was applied to test for significance between the cystinosis and control group. Informed consent was obtained from all patients and experiments conform the principles set out in the WMA declaration of Helsinki and the Department of Health and Human Services Belmont Report.

Data availability
The data sets produced in this study are available in the following databases: Proteomics data: PRoteomics IDEntifications database (PRIDE) partner repository with the dataset identifier PXD020046 (https:// www.ebi.ac.uk/pride/archive/projects/PXD020046).
Expanded View for this article is available online.
Papa S (2016) Defective autophagy, mitochondrial clearance and lipophagy in Niemann-Pick type B lymphocytes. PloS One 11: e0165780 The paper explained Problem Young children diagnosed with cystinosis receive cysteamine as a treatment to prevent the accumulation of cystine. This drug effectively lowers cystine values in their blood and tissues, but it is not able to prevent kidney function loss. In this study, we evaluated the effect of CTNS deficiency, the gene affected in cystinosis, in kidney proximal tubule cells, organoids and cystinotic zebrafish and determined the ability of different drug compounds to restore cell function.

Results
Loss of CTNS in proximal tubule cells reduces the ability to effectively degrade proteins, increases reactive oxygen species production and leads to large scale chances in intracellular metabolites and protein expression levels. Alpha-ketoglutarate (aKG) was one of the metabolites elevated in cystinotic cells as well as in patient plasma and may be linked to the increased autophagy and apoptosis in cystinotic cells. When evaluating the effect of different drug compounds on the cells, we found that a combination therapy of cysteamine with bicalutamide was far more effective in restoring cell functions than cysteamine alone and this finding was confirmed in different cystinosis models, including the kidney cell lines, kidney organoids and cystinotic zebrafish.

Impact
Cysteamine treatment alone cannot restore all cellular defects resulting from CTNS loss, and a combination therapy with a compound targeting the autophagic phenotype could therefore be highly beneficial in managing cystinosis. Our findings indicate that a cysteaminebicalutamide combination therapy is able to improve proximal tubule cell function in vitro and could potentially fulfil the unmet clinical need of preventing kidney failure in cystinotic patients.