Squaramide-based synthetic chloride transporters activate TFEB but block autophagic flux

Cystic fibrosis is a disease caused by defective function of a chloride channel coupled to a blockade of autophagic flux. It has been proposed to use synthetic chloride transporters as pharmacological agents to compensate insufficient chloride fluxes. Here, we report that such chloride anionophores block autophagic flux in spite of the fact that they activate the pro-autophagic transcription factor EB (TFEB) coupled to the inhibition of the autophagy-suppressive mTORC1 kinase activity. Two synthetic chloride transporters (SQ1 and SQ2) caused a partially TFEB-dependent relocation of the autophagic marker LC3 to the Golgi apparatus. Inhibition of TFEB activation using a calcium chelator or calcineurin inhibitors reduced the formation of LC3 puncta in cells, yet did not affect the cytotoxic action of SQ1 and SQ2 that could be observed after prolonged incubation. In conclusion, the squaramide-based synthetic chloride transporters studied in this work (which can also dissipate pH gradients) are probably not appropriate for the treatment of cystic fibrosis yet might be used for other indications such as cancer.


Introduction
Cystic fibrosis is the most frequent monogenetic lethal disease affecting humans 1,2 . This pathology is caused by loss-of-function mutation in the cystic fibrosis transmembrane conductance regulator (CFTR), a chloride transporter 3,4 . Importantly, CFTR mutations do not only compromise chloride flux (resulting in increased chloride concentrations in sweat and alterations in the composition of mucus with subsequent bronchial infections) but also affect autophagic flux in respiratory epithelial cells and macrophages [5][6][7][8][9][10] . The perturbation of autophagy may be disease-relevant because autophagy induction by pharmacological inhibitors of E1A-associated protein p300 (where E1A = adenovirus early region 1A), best known as EP300, and transglutaminase-2 can favor the function of certain CFTR mutants (and in particular the del508 mutant protein) [11][12][13][14] and reduce lung inflammation in patients [15][16][17] .
Intrigued by these observations, we wondered whether direct perturbations of chloride homeostasis by means of synthetic chloride transporters (or anionophores) such as the squaramide-based compound SQ1 and the analogous SQ2 18,19 might affect the autophagic process. SQ1 and SQ2 incorporate into the plasma membrane (and perhaps other cellular membranes), thus causing chloride influx into cells and dissipating chloride gradients in intracellular compartments, and neutralizing lysosomal pH in the process 19 . Although this ultimately causes cell death (and SQ1 has indeed been developed as a first-in-class anticancer agent), SQ1 and SQ2 can modulate autophagy before cells die 19 , offering a window of opportunity for investigating their impact on the system.
Here, we report that squaramide-based synthetic chloride transporters dramatically inhibit autophagic flux although they activate a pro-autophagic transcription factor. Moreover, these chloride anionophores trigger the relocation of an autophagic marker (LC3) towards the Golgi apparatus.

Cytofluorometry measurement of ROS generation
The generation of ROS was monitored with hydroethidine (HE) (at a final concentration of 5 μM; stock 10 mM in DMSO; excitation wave length of 488 nM, emission 620 nM; Molecular Probe. Briefly, cells were collected after treatment with the testing compounds followed by co-staining, during 30 min at 37°C, with HE. Cytofluorometric acquisitions were carried out on a Milteny cytofluorometer (MACSQuant® Analyzer 10), and statistical analyses were performed by using the FlowJo software (LLC, Oregon, USA).

High content image acquisition
Cells were seeded in tissue culture-treated 384-well μClear imaging plates (Greiner BioOne, Frickenhausen, Germany) and incubated under standard tissue culture conditions during 24 h at 37°C. Then, cells were treated with the indicated compounds and after 6 or 24 h of incubation, cells were fixed with 4% formaldehyde solution containing 1 μM Hoechst 33342 overnight at 4°C. The fixative was changed to PBS, and the plates were subjected to automated image analysis. For automated fluorescence microscopy, a robot-assisted Molecular Devices IXM XL BioImager (Molecular Devices, Sunnyvale, CA, USA) equipped with Sola light sources (Lumencor, Beaverton, OR, USA), adequate excitation and emission filters (Semrock, Rochester, NY, USA), and a 16-bit monochromes sCMOS PCO.edge 5.5 camera (PCO, Kelheim, Germany) and a ×20 PlanAPO objective (Nikon, Tokyo, Japan) was used to acquire nine view fields/well, followed by image processing with the custom module editor of the MetaXpress software (Molecular Devices). For the latter, the images were segmented and analyzed for GFP, RFP granularity or global fluorescence intensity (depending on the dyes) by comparing the standard deviation of the mean fluorescence intensity of groups of adjacent pixels within the cytoplasm of each cell to the mean fluorescence intensity in the same ROI using the MetaXpress software (Molecular Devices).

Statistical analyses
Data are reported as means ± SD of n > 3 replicates and experiments were repeated at least twice yielding similar results. Data were analyzed using Prism (GraphPad Software, Inc., La Jolla, CA, USA), and statistical significance was assessed by means of two-tailed Student's t-test or ANOVA tests, as appropriate. Unless otherwise specified, data are reported as mean ± SEM. Statistical significance was analyzed using the Student's-test. Differences in treated and control cells were considered to be significant if *p < 0.05, **p < 0.01, ***p < 0.001.

SQ1 and SQ2 induce autophagic LC3 + puncta in U2OS cells
Our laboratory has made extensive use of U2OS biosensor cell lines to measure cellular stress responses including autophagy 12,[22][23][24][25][26][27][28] . We first determined the kinetics of potential cytotoxic effects of the synthetic chloride channels SQ1 and SQ2 on these cells to study autophagy in conditions in which U2OS cells fully conserve their viability. These measurements led to the conclusion that the SQ1 and SQ2 have cytostatic effects that become significant at 12 h of incubation and diminishes the number of cells below the control value, indicating cytotoxicity, from 18 h. However, at 6 h neither cytostatic nor cytotoxic effects were detectable (Fig. 1b). Hence, all subsequent experiments were done in this time frame (6 h). SQ1 and SQ2 both induced the lipidation of microtubule-associated proteins 1A/1B light chain 3B (hereafter referred to as LC3 II), as detectable by an increase in the electrophoretic mobility of LC3 (Fig. 1c). Moreover, both synthetic chloride anionophores (but not the control compound SQ3, which shares structural features with SQ2, yet lacks chloride transporter activity 19 ) inhibited the kinase activity of mechanistic target of rapamycin complex-1 (mTORC1), as indicated by a reduced phosphorylation of ribosomal protein S6 kinase beta-1 (S6K1, best known as p70S6K) (Fig. 1c). Commensurate with these effects, SQ1 and SQ2 (but not SQ3) stimulated the aggregation of green fluorescent protein (GFP)-LC3 fusion protein in cytoplasmic dots (Fig. 1d). This effect was found both in the absence and in the presence of bafilomycin A1 (BafA1), an inhibitor of the lysosomal vATPase (Fig. 1e). However, SQ1 and SQ2 did not induce autophagic flux, as indicated by two independent series of experiments. First, in the presence of SQ1 and SQ2, the abundance of the autophagic substrate sequestosome 1 (STQM1, best known as p62) in U2OS cells did not decrease (as this was observed with the positive control, torin, an inhibitor of mTORC1 and mTORC2) but increased (Fig. 1f, g). This increase probably reflects the impairment of lysosomal function due to the decrease of the activity of the pH-sensitive hydrolases. This loss of lysosomal function is due to abnormal pH increase in lysosomes when intracellular chloride concentration is dysregulated. Second, in the neuronal PC12 cell line engineered to express a doxycyclin-inducible autophagic cargo (namely a GFP fused to exon 1 of a pathogenic huntingtin protein variant that contains 74 glutamine repeats, Q74) 29 , SQ1 and SQ2 increased the level of the cargo, contrasting with the effects of torin or rapamycin that both caused its elimination (Fig. 2).
In sum, SQ1 and SQ2 stimulate LC3 lipidation, its redistribution towards puncta, yet do not stimulate autophagic flux, confirming previous observations made in HeLa cells 19 .

Golgi localization of LC3 after SQ1/SQ2 treatment
The GFP-LC3 puncta induced by SQ1 or SQ2 were larger than those induced by torin and tended to coalesce at one pole of the nucleus (Fig. 1). We therefore suspected that such puncta might aggregate at or close to the Golgi apparatus. To examine this hypothesis, we took advantage of a U2OS biosensor cell line expressing a red fluorescent protein (RFP)-LC3 fusion protein as a reporter of LC3 localization, as well as a GFP-galactose-1-phosphate uridylyltransferase (GALT) fusion protein as a reporter of Golgi location 20,22,24,30,31 . After treatment with SQ1 or SQ2, we found a remarkable overlap of both fluorescent signals indicating the translocation of LC3 to the Golgi, a phenotype that closely resembled the one induced by oleate 20,25 but differed from that induced by torin (Fig. 3).
To further investigate this phenomenon, we pretreated the cells for 1 h with brefeldin A (BFA), an inhibitor of GBF1 (brefeldin A-resistance guanine nucleotide exchange factor 1) that disrupts the Golgi apparatus causing its dispersion into ministacks all over the cytoplasm [32][33][34] . Preincubation with BFA reduced the colocalization of GFP-GALT and RFP-LC3, confirming the idea that LC3 coalesces at the Golgi, yet did not prevent the SQ1/SQ2-induced formation of RFP-LC3 dots, suggesting that the Golgi is not required for this phenomenon to occur (Fig. 4).

Contribution of TFEB to LC3 relocation
Since SQ1 and SQ2 inhibit mTORC1, and this inhibition has been involved in the activation of the proautophagic transcription factor EB (TFEB) [35][36][37][38][39] , we investigated whether these synthetic chloride transporters would cause the translocation of a GFP-TFEB reporter protein from the cytoplasm to the nucleus. Indeed, SQ1 and SQ2 (but not SQ3) were able to stimulate TFEB activation (Fig. 5a, b). TFEB is known to stimulate lysosomal biogenesis, and both SQ1 and SQ2 caused the overexpression of lysosomal-associated membrane protein 1 (LAMP-1) in U2OS cells (Fig. S1). Moreover, knockout of TFEB partially reduced the formation of LC3B puncta induced by SQ1 or SQ2, an inhibition that was far less prominent than that found for the knockout of ATG5 (Fig. 5c, d). Quantitation of GFP-LC3 puncta that co-stain with an antibody recognizing LAMP1 revealed that SQ1 and SQ2 did not inhibit the fusion between lysosomes and autophagosomes (Fig. S3e). These observations suggest that SQ compounds inhibit autophagic flux at a step downstream of the autophagosomelysosome fusion, presumably because SQ1 and SQ2 dissipate the (normally acidic) pH gradient on the lysosomal membrane 19 .
Intrigued by these observations, we decided to investigate the role of TFEB in the induction of LC3B puncta by inhibiting calcineurin, the phosphatase that dephosphorylates TFEB for its activation after mTORC1 inhibition 35,40,41 . For this, we either used the calcium chelator (BAPTA) that can be targeted into cells a cell-permeant acetoxymethyl ester derivative (BAPTA-AM) or, more specifically, enzymatic inhibitors of calcineurin such as cyclosporine A (CsA) and its non-immunosuppressive control compound cyclosporin H (CsH). Moreover, we used cycloheximide (CHX) as a general inhibitor of protein translation. Cells were preincubated for 1 h with these reagents and then were treated with SQ1 or SQ2. Of note, BAPTA-AM, CsA, and CsH strongly inhibited the formation of GFP-LC3 puncta (Fig. 6b, c). CsA and CsH failed to inhibit the formation of reactive oxygen species (ROS) induced by SQ1 and SQ2 (as measurable by quantifying the conversion of non-fluorescent hydroethidine into fluorescent ethidium), which however could be quenched by using the antioxidant N-acetyl cysteine (NAC) (Fig. S2). NAC was also capable of inhibiting LC3 puncta formation (Fig. 6b, c). Importantly, none of the inhibitors of GFP-LC3 puncta (i.e., BAPTA-AM, CHX, CsA, CsH, NAC) nor full inhibition of the autophagic process (in ATG5 KO cells) (Fig. S3C) were able to significantly reduce cell killing by SQ1 or SQ2, as measured at 24 h of incubation. Hence, the formation of GFP-LC3 puncta and later cell death can be uncoupled from each other. Furthermore, to evaluate the possibility that CsA and TFEB act on different steps of the cellular alterations induced by SQ1 and SQ2, we compared the SQ1 and SQ2 effects on the formation of GFP-LC3 dots between U2OS WT and U2OS TFEB KO cells in the presence or absence CsA. The CsA-mediated inhibition SQ1 and SQ2-elicited GFP-LC3 dots was maintained in TFEB KO cells. This argues in favor of the possibility that TFEB and CsA modulate the effects of SQ1 and SQ2 in a differential fashion (Fig. S3a, b).
Finally, we treated a cystic fibrosis bronchial epithelial (CFBE) cell line harboring the most frequent CFTR mutation (delF508) for 24 h with SQ1 or SQ2 in the absence or presence of bafilomycin A1. The results indicate that SQ1 or SQ2 induced the formation of GFP-LC3 dots and that this effect was not further enhanced by bafilomycin A1 (Fig. S3F), confirming that SQ1 or SQ2 inhibit autophagic flux in a disease-relevant cellular model.

Discussion
SQ1 and SQ2 have previously been shown to stimulate the lipidation of LC3 19 . Here, we confirm this finding and link it to the relocation of LC3 to discrete structure in the cytoplasm, which occurs in a strictly ATG5-dependent fashion. We found that SQ1 and SQ2 inhibit mTORC1, a phenomenon that is linked to the activation of TFEB (and other similar transcripition factors) 19 . Indeed, SQ1 and SQ2 promote the translocation of TFEB from the cytoplasm to the nucleus, and TFEB is partially required for the aggregation of LC3 in discrete cytoplasmic puncta. Although TFEB activation should stimulate autophagic flux, we found that SQ1 and SQ2 were unable to do so. Rather both SQ1 and SQ2 caused the accumulation of autophagic cargo in two different cell types, in human osteosarcoma U2OS cells (in which SQSTM1 became more abundant) and in rat neuronal PC12 cells (in which a mutant Huntingtin Q74 construct accumulated). SQ1 and SQ2 are synthetic chloride transporters that dissipate the lysosomal c-e Statistical analysis of RFP-LC3 puncta counts per cell, GFP-GALT marked Golgi surface per cell and the localization of LC3 and GALT (***p < 0.0001, compared to untreated control cells, Co) pH gradient, a phenomenon that previously has been suggested to account for the inhibition of autophagic flux 19 . Thus, although both agents activate a proautophagic transcription factor they ultimately fail to stimulate autophagic flux.
Several strategies designed to inhibit the calcineurinmediated activation of TFEB, namely intracellular calcium chelation or direct calcineurin inhibition by CsA largely prevented autophagy induction by SQ1 or SQ2, confirming the importance of this pathway for the observed phenotype. However, inhibition of protein neo-synthesis or the calcineurin pathway was unable to prevent SQ1 and SQ2-induced cell death in conditions in which such an inhibition fully suppressed the formation of LC3 puncta. We conclude from these results that the generation of such puncta apparently is not required for the cytotoxic action of synthetic chloride transporters, which must rely on other damage pathways to kill cells.
Surprisingly, SQ1 and SQ2 caused the relocation of LC3 into discrete perinuclear structures, usually presenting as one single cap in immediate vicinity of the nucleus. These structures turned out to co-localize with the Golgi apparatus. We previously found a similar Golgi-specific coalescence of LC3 in cells treated with unsaturated fatty acids (such as oleate) 20,42 , the photosensitizer redaporfin 22,43 , and the cytolytic peptide derivative LTX-401, which itself has a strong tropism for the Golgi 25 . Thus, structurally rather distinct compounds can induce a similar coalescence of LC3 in or at the Golgi apparatus. For all these compounds (SQ1, SQ2, oleate, redaporfin, LTX401), the Golgi-disrupting agent BFA prevented the relocalization of LC3 to perinuclear caps. The mechanisms explaining this organelle-specific pattern of LC3 redistribution remain to be elucidated.
Synthetic chloride transporters have been developed as hypothetical remedies against cystic fibrosis [44][45][46][47][48][49] . Indeed, at least theoretically, such ionophores might be used as substitutes to compensate for the defective function of CFTR 44,50 . As shown in this paper, consistent with precedent reports 19,51 , chloride ionophores that can dissipate pH gradients, such as the compounds studied here that have been shown to be capable of transporting both chloride and protons across lipid bilayers, block autophagic flux in a peculiar manner, arguing against their Fig. 5 Contribution of TFEB and ATG5 in LC3 relocation induced by the SQs. a, b Representative images of U2OS cells stably expressing GFP-TFEB fusion protein, treated with torin 300 nM, SQ1, SQ2, and SQ3 at 10 μM during 6 h. Data are expressed as means ± SEM of at least three independent experiments and demonstrate the average ratio between GFP-TFEB florescence intensity in the nucleus vs. cytoplasm. (*p < 0.05, **p < 0.01, ***p < 0.0001, compared to untreated cells, Co). c, d Representative images and statistics of U2OS GFP-LC3 cells, and the ATG5 KO and TFEB KO counterparts after 6 h of treatment with the mentioned compounds. Bars depicts the absolute number of GFP-LC3 + dots per cell. Data are expressed as means ± SEM of at least three independent experiments (*p < 0.05, ***p < 0.0001, compared to untreated cells, Co) therapeutic utility. Indeed, it appears to be advantageous to restore autophagic turnover in cystic fibrosis 12,15,17 , implying that autophagy inhibition would ultimately worsen the disease. These types of transporter should be exploited for therapeutic use against other diseases, for instance cancer, in which their cytotoxic potential might be advantageous. The development of chloride transporters which do not dissipate pH gradients is ongoing in Sydney 52 .
(see figure on previous page) Fig. 6 Effects of pharmacological inhibition of TFEB on LC3 relocation and cell death induced by SQ1 and SQ2. a Experimental design to assess LC3 puncta when the synthetic chloride transporters SQs were combined with different inhibitors. b Representative images of U2OS GFP-LC3 expressing cells when pretreated with the calcium chelator BAPTA-AM, the antioxydant NAC, the calcineurin inhibitor CsA, its analogue CsH, and the protein synthesis inhibitor CHX followed by the co-treatment with the synthetic chloride modulators. c Bars show the average number of GFP-LC3 + dots per cell. Data are expressed as means ± SD of one representative experiment *P < 0.05; **P < 0.01; ***P < 0.001. d Evaluation of cell death by double-staining with DAPI together with DiOC 6 (3) followed by cytofluorometric detection of dead cells (DAPI + ) and dying cells (DAPI + DiOC 6 (3) low ). Data are expressed as means ± SEM of at least three independent experiments (Co/Ctrl). *P < 0.05; **P < 0.01; ***P < 0.001