Repurposing diphenylbutylpiperidine-class antipsychotic drugs for host-directed therapy of Mycobacterium tuberculosis and Salmonella enterica infections

The persistent increase of multidrug-resistant (MDR) Mycobacterium tuberculosis (Mtb) infections negatively impacts Tuberculosis treatment outcomes. Host-directed therapies (HDT) pose an complementing strategy, particularly since Mtb is highly successful in evading host-defense by manipulating host-signaling pathways. Here, we screened a library containing autophagy-modulating compounds for their ability to inhibit intracellular Mtb-bacteria. Several active compounds were identified, including two drugs of the diphenylbutylpiperidine-class, Fluspirilene and Pimozide, commonly used as antipsychotics. Both molecules inhibited intracellular Mtb in pro- as well as anti-inflammatory primary human macrophages in a host-directed manner and synergized with conventional anti-bacterials. Importantly, these inhibitory effects extended to MDR-Mtb strains and the unrelated intracellular pathogen, Salmonella enterica serovar Typhimurium (Stm). Mechanistically Fluspirilene and Pimozide were shown to regulate autophagy and alter the lysosomal response, partly correlating with increased bacterial localization to autophago(lyso)somes. Pimozide’s and Fluspirilene’s efficacy was inhibited by antioxidants, suggesting involvement of the oxidative-stress response in Mtb growth control. Furthermore, Fluspirilene and especially Pimozide counteracted Mtb-induced STAT5 phosphorylation, thereby reducing Mtb phagosome-localized CISH that promotes phagosomal acidification. In conclusion, two approved antipsychotic drugs, Pimozide and Fluspirilene, constitute highly promising and rapidly translatable candidates for HDT against Mtb and Stm and act by modulating the autophagic/lysosomal response by multiple mechanisms.

Middlebrook 7H9 broth (Becton Dickinson, Breda, the Netherlands) supplemented with 10% ADC (Becton Dickinson), 0.05% Tween 80 (Sigma-Aldrich), and with the addition of 50 µg/ml Hygromycin B in the case of DsRed-expressing H37Rv. Mtb suspensions were prepared from a running Mtb culture, which was one day prior to infection diluted to a density corresponding with early log-phase growth (OD600 of 0.25). Stm strain SL1344 was cultured in Difco lysogeny broth (LB) (Becton Dickinson). Stm was grown overnight in LB, subsequently diluted 1:33 in fresh LB and used after approximately 3 h of incubation, when log-phase growth was achieved (OD600 of 0.5). Bacteria (or liquid broth at equal v/v for mock-infection) were diluted in complete IMDM or complete RPMI without antibiotics for infecting cell lines or primary cells, respectively. We consistently used a multiplicity of infection (MOI) of 10 for both strains. Cell lines and primary cells, which had been seeded at a density of 10,000 or 30,000 cells per well, respectively, in 96-well flat-bottom plates 1 day prior to infection, were inoculated with 100 μl of the bacterial suspension. Cells were subsequently centrifuged for 3 min at 800 rpm and incubated at 37 °C/5% CO 2 for 20 min in case of Stm infection or 60  www.nature.com/scientificreports/ bacteria were then washed away with culture medium containing 30 μg/ml Gentamicin sulfate, incubated for 10 min at 37 °C/5% CO2, followed by replacement with medium containing 5 μg/ml Gentamicin sulfate and, if indicated, chemical compounds until readout. MOI of the bacterial inoculum was confirmed by a standard colony-forming unit (CFU) assay.
Chemical compound treatment. Cells were treated for 24 h, unless indicated otherwise, with chemical compound at a concentration of 10 µM, unless indicated otherwise, or DMSO at equal v/v in medium containing 5 μg/ml Gentamicin sulfate. Treatment regimens were designed not to exceed DMSO solvent end concentrations of 0.2%. Rifampicin and Isoniazid were used at a concentration of 0.05 µg/mL and 0.4 µg/mL, respectively. Higher concentrations of antibiotics showed almost complete eradication of Mtb growth which thus completely abrogates the window of detection for any additive/synergistic host-directed drug effects, while in contrast lower antibiotic concentrations showed a larger donor variability in the effect on Mtb outgrowth.
Colony-forming unit (CFU) assay. Cells were lysed in H 2 O containing 0.05% SDS (ThermoFisher Scientific). Lysates of Mtb-infected cells were serially diluted in steps of fivefold in 7H9 broth and 10 μl droplets were spotted onto square Middlebrook 7H10 agar plates. Plates were incubated at 37 °C for 12-14 days and bacterial colonies quantified using a microscope with a magnification of 2.5 times to enhance early detection of bacterial growth. Lysates of Stm-infected cells were serially diluted in LB broth, thereafter 10 μl droplets were spotted onto square LB agar plates and incubated overnight at 37 °C.
Liquid bacterial growth inhibition assay. Stm or Mtb cultures in logarithmic growth phase were diluted to an OD 600 of 0.1 in LB broth or 7H9 broth respectively, of which 200 μl per flat-bottom 96-well was incubated with chemical compound, antibiotic or DMSO at equal v/v at indicated concentrations. Stm growth was measured after overnight incubation at 37 °C, while Mtb growth was evaluated for 10 days of incubation at 37 °C. Absorbance was measured by optical density at 550 nm on a Mithras LB 940 plate reader (Berthold Technologies, Bad Wildbad, Germany) while shaking between measurements.
Flow cytometry. Infected cells were at experimental endpoint washed with 100 µl of PBS and detached by incubation in 50 µl of Trypsin-EDTA 0.05% for five minutes. Single cell suspensions were fixed by adding 100 µl of 1.5% paraformaldehyde and incubated for 60 min at 4 °C. Acquisition was performed using a BD FACSLyric™ Flow Cytometer equipped with BD FACSuite software (BD Biosciences). Data was analyzed using FlowJo software v10.
Cell viability assay. Cells seeded at a density of 30,000 cells/well in 96-well flat-bottom plates were stained in 50 μl RPMI without phenol red (Life Technologies-Invitrogen) containing propidium iodide (PI) (1:500, Sigma-Aldrich) and 100 μg/ml of Hoechst 33,342 (Sigma-Aldrich). Cells were incubated at room temperature (RT) for 5 min. Per well 3 images were taken using a Leica AF6000 LC fluorescence microscope combined with a 20 × dry objective. The number of dead cells (positive for PI) versus total number of adherent cells (positive for Hoechst) was quantified using CellProlifer version 3.0.0 43 . Immunostaining. Cells were seeded on poly-d-lysine coated glass-bottom (no. 1.5) 96-well plates (Mat-Tek, Ashland, Massachusetts, United States), pre-washed with PBS, at a density of 30,000 per well. After overnight incubation, cells were infected with DsRed-expressing Mtb at a MOI of 10 as previously described. At the indicated experimental endpoint, cells were washed three times with PBS and fixed for 60 min at RT using 1% methanol-free EM-grade formaldehyde (ThermoFisher Scientific) diluted in PBS. Cells were washed with PBS and remaining reactive formaldehyde was quenched using 100 μl of Glycine solution (1.5 mg/ml in PBS) for 10 min at RT. Cells were permeabilized in 0.1% Triton-X (Sigma-Aldrich) diluted in PBS for 10 min at RT and Fc-receptors were subsequently blocked using 5% human serum (HS; Sanquin Blood bank, Amsterdam, The Netherlands) for 45 min at RT. After removal of the 5% HS, cells were stained with primary antibody diluted in 5% HS for 30 min at RT, washed three times with 5% HS and incubated with secondary antibody in 5% HS for 30 min at RT in the dark. After washing three times with 5% HS, cells were counterstained with 50 ul of 2 μg/ ml Hoechst 33,342 and Phalloidin for 30 min at RT in the dark. Images, at least 3 per well, were acquired using a Leica TCS SP8 X WLL confocal system and 63X oil immersion objective. Hybrid detectors were used with a time gate to switch off data collection during the pulse. The fluorescent dyes LysoTracker Deep Red (ThermoFisher Scientific) (75 nM) and CYTO-ID 2.0 (Enzo LifeSciences) (1:500) were added to the cells 30 min prior to treatment endpoint. Cells were fixed at the experimental endpoint and counterstained with Hoechst and Phalloidin as described above except for permeabilization.
Colocalization analysis was performed as follows. Image background was subtracted using the rolling ball (20-pixel radius) algorithm with Fiji software 44 . CellProlifer 3.0.0 was used to first correct for non-homogenous illumination if necessary, then for the segmentation of both the fluorescent bacteria and marker of interest using global thresholding with intensity-based de-clumping 45 . For every experiment, segmentation was performed independently with both a range of thresholds and adaptive three-class Otsu thresholding to confirm segmentation results. Then per image the overlap of Mtb with marker of interest was calculated as percentage of object overlap (colocalization) or the integrated intensity of the marker of interest per single nucleus or bacterium was determined. Data in Figs. 3C and 4C are shown for 4 individuals, while two donors could not be analyzed due to high background levels.  46 . Relative expression levels were calculated by applying the formula ((2 −ΔCT(Target gene) )/(2 −ΔCT(GAPDH) )).
Statistics. Statistics were performed as described previously 23 . Normal distribution of the data was tested using the Shapiro-Wilk test. If the data was normally distributed, unpaired t-test with Welch's correction or one-way ANOVA with Dunnett's multiple comparison test were applied when assessing differences between 2 and 3 or more groups of unpaired data, respectively. Normally distributed paired samples were analyzed using repeated measure (RM) one-way ANOVA with Dunnett's multiple comparison test and Geisser Greenhouse correction. Two-way ANOVA with Tukey's multiple comparison test was used when the effect of two independent variables was tested simultaneously. Assessing differences between 2 or more groups of unpaired observations from not normally distributed data was performed using Kruskal-Wallis test, while paired observations were tested using the Wilcoxon matched-pairs signed rank test with post-hoc Benjamini-Hochberg correction. Analyses were performed using GraphPad Prism 8.

Results
Compounds were identified that contain autophagy-regulating activity and repurposing potential for HDT against Mtb and Stm. By employing our previously described flow-cytometry-based assay using Mtb-infected human MelJuSo cells 22 , we screened the Screen-Well Autophagy library that includes clinically approved molecules, to find new drugs with HDT activity against intracellular Mtb. Mtb-infected human cells were treated during 24 h with compounds. The PKA/PKB-Akt1 inhibitor H-89 was included as a positive control based on our earlier work 21,22 . We identified 25 compounds that strongly reduced bacterial burden (< 35% infected cells compared to control DMSO). From these 25 molecules we discontinued 15 because of undesirable significant host cell toxicity (cell-yield < 85% after 24 h treatment compared to control DMSO), and 2 because of a current lack of clinical approval (SB-216763 and Licochalcone A). Two additional compounds were not pursued further since their in vivo adjunct therapy potential had already been shown previously, independently validating our discovery approach (Chloroquine and its metabolite Hydroxychloroquine) 47 (Fig. 1A, and Table S1).
In our effort to focus on new candidates it was of interest that 2 of the 6 remaining compounds, notably Fluspirilene and Pimozide, are structural analogues of the diphenylbutylpiperidine-class of antipsychotic drugs. Both drugs have an extensive clinical safety profile and are already for several decades used to treat multiple disorders. To explore and compare mechanistics of the analogues, we decided to focus on these two drugs only in this study, while detailed work on the other hit compounds will be described elsewhere.
We observed no cellular toxicity following treatment with Fluspirilene and Pimozide, while H-89 treatment affected cell numbers slightly (Fig. 1B). To validate these initial screening results further in a physiologically more relevant in vitro model, we generated primary GM-CSF-derived pro-inflammatory macrophages (Mφ1) and M-CSF-derived anti-inflammatory macrophages (Mφ2) 22,42,48 , and tested the potential of the two diphenylbutylpiperidines to inhibit outgrowth of intracellular Mtb (using a classical colony forming unit (CFU) assay as read-out (Fig. 1C). Both Fluspirilene and Pimozide induced a significant decrease of Mtb outgrowth in Mφ1 and particularly Mφ2 (median reduction of bacterial outgrowth of 22% and 36% for Fluspirilene and 35% and 32% for Pimozide, respectively) ( Fig. 1C and Table S2).
To investigate potential broader-range applicability of Fluspirilene and Pimozide, we investigated their intracellular bacterial growth inhibitory activity against a different class of intracellular bacteria, using Stm-infected human Mφ (Fig. 1D 21,22 . Interestingly, both drugs vastly reduced Stm outgrowth (median reduction of bacterial outgrowth > 96% and > 98% in Mφ1 and Mφ2, respectively), suggesting these HDT drugs could be more broadly applicable.
To exclude that Fluspirilene and Pimozide acted in a direct anti-bacterial manner, extracellular Stm and Mtb were treated with Fluspirilene and Pimozide in liquid broth, in equal concentrations (10 µM) as used in the cellbased infection models. Fluspirilene and Pimozide did not affect Mtb or Stm growth, whereas control antibiotics Rifampicin (Mtb) and Gentamicin (Stm) inhibited bacterial growth (Fig. 1E, F and Table S2).
Taken together, we identified two clinically approved structural analogues displaying reported autophagyinducing capacities as novel safe candidate molecules for HDT, that inhibited intracellular Mtb as well as Stm in both pro-and anti-inflammatory human primary macrophages.
Fluspirilene and Pimozide HDT activity against intracellular Mtb was confirmed cross-species of host and mycobacterium. Since HDT drugs act on host rather than bacterial targets, they should inhibit outgrowth of drug-susceptible and multi-drug resistant (MDR)-Mtb strains similarly. To verify this, we measured the effect of the two diphenylbutylpiperidines on CFU outgrowth in human Mφ2 (we focused on Mφ2 since we found similar efficacy in both Mφ1 and Mφ2). The cells were infected with either Mtb Dutch outbreak strain 2003-1128 or Mtb Beijing strain 16,319, both MDR-Mtb-strains. Indeed, treatment with Fluspirilene and Pimozide significantly inhibited bacterial outgrowth of both MDR-Mtb strains highly efficiently ( Fig. 2A and Table S2), thereby emphasizing their clinical relevance.
Clinical application of HDT will most likely be considered as adjunct to standard antibiotic therapy for TB 14 . Since HDT and antibiotics have different targets, we investigated whether synergistic/additive effects between these treatments could be detected. Combining either Fluspirilene or Pimozide with a suboptimal dose of Rifampicin (0.05 µg/ml) inhibited Mtb outgrowth to a significantly larger extent compared to Fluspirilene, Pimozide, or Rifampicin treatment alone ( Fig. 2B and Table S2). Unexpectedly, this effect was not observed when Fluspirilene and Pimozide were combined with a suboptimal dose of Isoniazid (0.4 µg/ml) ( Fig. 2C and Table S2). While not further investigated here, a possible explanation for this could be that the diphenylbutylpiperidines alter host cell processes required for Isoniazid's conversion into its active form 49 .
To further strengthen the potential utility of Fluspirilene and Pimozide, we next tested their efficacy on Mtb-infected ex vivo non-human primate (NHP) bronchoalveolar lavage (BAL) cells which are a rich source of alveolar macrophages (AMφ), the primary target cells for Mtb. Cross-species efficacy of the drugs would suggest functional conservation of their host targets as well as translate and validate our above findings in human cells. Importantly, 24-h treatment with either one of the drugs decreased outgrowth of intracellular Mtb. This was reflected in the reduced percentage of infected cells (Fig. 2D, left panel), and especially in the reduced number of colony forming bacteria (Fig. 2D, middle panel, and Table S2). Notably, Fluspirilene and Pimozide treatment showed noy toxicity towards the ex vivo NHP AMφ (Fig. 2D, right panel).  www.nature.com/scientificreports/ Taken together, these results significantly strengthen the potential value of Fluspirilene and Pimozide in HDT approaches against intracellular infections, including MDR-Mtb bacteria, and demonstrate their activity also in (NHP) infected alveolar macrophages.
Fluspirilene and Pimozide regulate autophagy, induce a lysosomal response and enhance bacterial presence in autophago(lyso)somes. Because Fluspirilene and Pimozide are known to modulate and induce autophagy 50 , we investigated this as a potential mode of action against intracellular Mtb. Mφ2 were infected with Mtb, treated for 4 h with Fluspirilene and Pimozide, stained with CYTO-ID (a tracer that stains all autophagy-related vesicles) and visualized using confocal microscopy (Fig. 3A, left panel). Importantly, both Fluspirilene and Pimozide, although not statistically significant, tended to increase the CYTO-ID vesicle area and the bacterial localization in these vesicles (Fig. 3A, middle and right panel, and Table S2), lending support to the involvement of autophagy in Fluspirilene and Pimozide's mode of action against intracellular Mtb in infected human cells.
To assess whether the reduction in Mtb outgrowth induced by Fluspirilene and Pimozide could also be associated with an increase in lysosomal activity we first examined the nuclear accumulation of transcription factor EB (TFEB), a master regulator of the coordinated lysosomal expression and regulation (CLEAR) gene network, as well as autophagy 51,52 . Nuclear intensity of TFEB was quantified in Mtb-infected Mφ2 after 4 h of treatment with Fluspirilene, Pimozide or Torin-1 as positive control 53 , using confocal microscopy (Fig. 3B, left panel). A highly significant increase in nuclear TFEB was observed for both Fluspirilene and Pimozide, even to a similar extent as the positive control Torin-1 (Fig. 3B, right panel, and Table S2). To investigate the lysosomal response in more detail we employed the lysosomal tracer Lysotracker. DsRed-Mtb-infected Mφ2 were treated with Fluspirilene and Pimozide for 4 h and the lysosomal area and localization of bacteria in lysosomes was quantified (Fig. 3C, left panel). Pimozide tended to induce an increase in lysosomal area, in contrast to Fluspirilene (Fig. 3C, middle panel). Additionally, Fluspirilene and Pimozide induced a mild, though statistically not significant average increase in colocalization of bacteria and lysosomes compared to DMSO (Fig. 3C, right panel, and Table S2).
Collectively, these results suggest that the lysosomal response key regulator TFEB and likely also the autophagic response are involved in enhancing host defense induced by diphenylbutylpiperidines, which restricts outgrowth of intracellular Mtb. Notwithstanding, the observed limited effect sizes for some of these processes suggested to us that these mechanisms likely do not fully account for the potent HDT activity of Fluspirilene and Pimozide against intracellular Mtb. We therefore postulated that other mechanisms of action are likely involved in the control of intracellular bacteria upon treatment with Fluspirilene and Pimozide.

Fluspirilene and Pimozide inhibit STAT5 function and Pimozide additionally reduces the presence of cytokine-inducible SH2-containing protein (CISH) on Mtb phagosomes. Because
Pimozide has been reported to inhibit STAT5 function by dephosphorylation 54 , and work by Queval, Song 55 recently uncovered a STAT5-mediated control of phagosomal acidification induced by Mtb, we explored functional inhibition of STAT5 by diphenylbutylpiperidines as a potential mode of action to control intracellular Mtb. First, the effect of 4-h treatment with diphenylbutylpiperidines on nuclear presence of phosphorylated STAT5 (P-STAT5) was investigated in Mtb-infected Mφ2 using confocal microscopy (Fig. 4A, left panel). Both Fluspirilene and Pimozide significantly decreased the nuclear presence of P-STAT5 Fig. 4A, middle and right panel, and Table S2). To further corroborate our finding that Fluspirilene and Pimozide lowered nuclear P-STAT5 quantity and hence its transcriptional activity, we analyzed the expression levels of cytokine-inducible SH2-containing protein (CISH), a transcriptional target of STAT5 43 . We confirmed the observation by Queval, Song 25 that Mtb infection increases CISH transcript levels in our Mtb-Mφ2 model after 4 h (Fig. 4B). Unexpectedly, Fluspirilene increased CISH transcript levels, even to a higher extent than the positive control GM-CSF 56 , while Pimozide did not affect CISH transcript expression levels, despite the compound's' effects on P-STAT5 (Fig. 4B). We therefore quantified colocalization of CISH protein with Mtb, since the presence of CISH on the Mtb phagosome has been associated with bacterial survival 55 . Pimozide treatment caused a significant decrease of CISH fluorescence intensity per bacterium within the 4 h time-period, but Fluspirilene did not clearly show a difference compared to DMSO (Fig. 4C and Table S2).
Taken together, these data suggest that Fluspirilene and Pimozide both decrease P-STAT5 protein levels in the nucleus. In addition, although the effects on intracellular Mtb bacteria did not correlate with differential expression of the STAT5 target CISH at the transcript level, Pimozide lowered CISH protein presence on the Mtb phagosome. Thus, both Fluspirilene and Pimozide likely target intracellular Mtb by a mechanism involving P-STAT5, which for Pimozide, but not Fluspirilene, includes routing through the CISH effector pathway. These results suggest that structurally highly related HDT compounds can nevertheless subtly differ in their mode of actions against intracellular infection.

Pimozide induces ROS/RNS production and antioxidants impair bacterial killing by Pimozide as well as Fluspirilene. Next to autophagy and lysosomal degradation, reactive oxygen species (ROS) and
reactive nitrogen species (RNS) represent additional mechanisms that could play a key role in controlling intracellular Mtb. Of note, Pimozide has been reported to induce ROS production in various cell types [57][58][59] , prompting us to explore the role of radical species in the mode of action of both Fluspirilene and Pimozide. Using a general oxidative stress indicator (CM-H2DCFDA), we measured ROS/RNS production in uninfected Mφ2 (due to biosafety level 3-restrictions) during treatment with Fluspirilene and Pimozide. Pimozide but not Fluspirilene treatment clearly increased ROS/RNS production ( Fig. 5A and B, left panel). To confirm that probe conversion was caused by cellular produced radical species, the probe was incubated with Pimozide in a cell-free system, which showed no significant effect (Fig. 5C) www.nature.com/scientificreports/ production by Pimozide, we added the antioxidant N-acetylcysteine (NAC) to the Pimozide treatment regimen (Fig. 5B, right panel). In 7 out of 9 donors Pimozide-induced ROS/RNS production was impaired by the addition of NAC, indicating that the probe conversion indeed reflects elevated ROS/RNS production.
To study the importance of Pimozide-induced ROS/RNS production in its control of intracellular bacterial infection, Mtb-infected Mφ2 were treated with either Pimozide or Fluspirilene in combination with various ROS/ RNS inhibitors, namely: NAC, Ebselen (a NOX inhibitor and glutathione peroxidase mimic 60 ), MitoTEMPO (a mitochondria-targeted radical scavenger 61 ), and L-NMMA (a nitric oxide synthase (NOS) inhibitor 62 ) (Fig. 5D and Table S2). NAC, and L-NMMA, but not Ebselen and MitoTEMPO, significantly inhibited the HDT effects of Pimozide (Fig. 5D, right panel). Interestingly, we also observed a significant effect of MitoTEMPO but not NAC, L-NMMA or Ebselen, on Fluspirilene (Fig. 5D, left panel), indicating a -possibly differential-role for ROS/RNS in the mode of action of both Fluspirilene and Pimozide. NAC, MitoTEMPO, L-NMMA and Ebselen treated Mtb-infected Mφ2 all showed some reduced bacterial outgrowth, which was significant for NAC and Ebselen (Supplementary Fig. 1 and Table S2), further emphasizing the role of ROS/RNS in intracellular bacterial control, which is supported by their antagonistic effect on Pimozide and Fluspirilene induced infection control.
In conclusion, we have identified two clinically approved drugs, Fluspirilene and Pimozide, which we show to have novel and significant potential as host-directed therapeutics for the treatment of intracellular bacterial pathogens such as Mtb and Stm, mechanistically operating by multiple mechanisms of action. We show that both Fluspirilene and Pimozide significantly alter the lysosomal response and tend to increase autophagic targeting of Mtb, likely via nuclear translocation of TFEB. Furthermore, our findings support two additional modes of action, particularly associated with Pimozide-mediated activity. Firstly, Fluspirilene and Pimozide counteract Mtb-induced STAT5 phosphorylation, but only Pimozide thereby reduced the presence of CISH on the Mtb phagosome, thus probably regulating its acidification 55 . Secondly, Pimozide induced measurable ROS/RNS production and its efficacy was impacted by ROS/RNS inhibitors. Together these mechanisms likely work in concert (Fig. 6) to reduce intracellular bacterial outgrowth and thus could be targeted by adjunct HDT in difficult to treat intracellular infections such as Mtb and Stm and likely also related pathogens.

Discussion
In this study, we aimed to find new host-directed therapy (HDT) candidate drugs for the treatment of (drug resistant) TB and other difficult to treat intracellular bacterial infections, and identified two structural analogues of the diphenylbutylpiperidine-class, Fluspirilene and Pimozide, with strong efficacy against intracellular Mtb and Stm. We next validated their HDT potential in Mtb and Stm primary human macrophages with either pro-or anti-inflammatory phenotypes. Importantly, we corroborated their activity against intracellular Mtb in a cross-species non-human model using bronchoalveolar lavage fluid macrophages, which are a rich source of alveolar macrophages (AMφ), the primary target cells for Mtb. This cross-species efficacy underscored functional conservation of host target molecules and independently validated their effects in human Mtb infected cells. Additionally, the strong HDT activity of Fluspirilene and Pimozide was confirmed also against clinical isolates www.nature.com/scientificreports/ of multi-drug resistant (MDR)-Mtb strains, and moreover acted in a synergistic and/or additive mode with a suboptimal dose of Rifampicin. Extracellular bacterial proliferation itself was unaffected by these compounds, indicating that their targeting of host mechanisms is responsible for their activity. Furthermore, the high efficacy shown against Stm, indicates that these drugs could be repurposed as treatment against a wider range of intracellular pathogens as already shown for other HDTs and pathogens 63 . Interestingly, we propose to have found several modes of action employed by Fluspirilene and Pimozide, including modulation of the lysosomal response, autophagy, induction of ROS/RNS production and finally the inhibition of STAT5 with consequently decreased localization of CISH protein on the Mtb phagosome. We postulate that these mechanisms cooperate in concert via multiple pathways to reduce intracellular bacterial outgrowth. Drugs acting via multiple effector mechanisms might be desirable over single mechanism of action www.nature.com/scientificreports/ (MOA), because developing bacterial resistance against multiple host effector mechanisms simultaneously will be extremely hard to achieve (summarized in Fig. 6). Fluspirilene and Pimozide are potent inhibitors of several dopamine receptors and calcium channels and are used clinically to treat psychotic disorders, like schizophrenia, psychosis and Tourette syndrome 41,64 . Some antipsychotic drugs have already been shown to exert (host directed) antimicrobial activity. For example, phenothiazine-derived drugs, which include Trifluorperazine, as well as butyrophenone derivatives, with Haloperidol as a well-known example, have therapeutic potential against several pathogens as we and others have shown previously 22,26,65,66 . The diphenylbutylpiperidine-class of drugs has also been studied, revealing host-directed effects of Pimozide against both facultative and obligate intracellular pathogens, Listeria monocytogenes and Toxoplasma gondii respectively 67,68 . Also, Fluspirilene has been shown to potentiate antimicrobial activity against several bacteria, but only in a direct antibiotic manner 69,70 . Despite these observations, we found no direct antimicrobial effect of Fluspirilene (or Pimozide) on either Mtb or Stm. This is likely reconciled by the fact that Fluspirilene displayed minimum inhibitory concentrations (MICs) of > 80 µM for E. coli and K. pneumoniae which both belong to the same order as Stm, namely Enterobacterales, whereas in our work we employed significantly lower (10 µM) concentrations at which strong HDT but no direct anti-bacterial effects were observed 69 .
Our work suggests that four different host effector mechanisms are likely to play a role in the treatmentefficacy of Fluspirilene and Pimozide. A first mechanism we demonstrated was that Fluspirilene and Pimozide activated autophagy in Mtb-infected Mφ and increased the localization of Mtb in autophagic vesicles (Fig. 3A). Although this was not investigated in further detail, we hypothesize that Ca 2+ signaling could be important to this effector mechanism. Both Fluspirilene and Pimozide antagonize calcium channel activity and can lower intracellular Ca 2+ levels, thereby inactivating Calpain that leads to the induction of autophagy 41,64,71,72 . In support Although increased ROS/RNS production was significantly induced by Pimozide only, antioxidants impaired the efficacy of Pimozide and to a lesser extent also Fluspirilene (II). Fluspirilene and Pimozide both inhibited STAT5 activity and Pimozide consequently reduced CISH localization on the Mtb containing vesicle (III). Lastly, Fluspirilene and to a larger extent Pimozide, increased nuclear TFEB localization concomitant with an increased lysosomal response (IV). Black arrows indicate the general process of Mtb phagocytosis and subsequent events, white arrows show compound mediated activation or inhibition of targets and green and red lines depict signaling pathways and protein interactions with reduced or increased bacterial outgrowth as outcome, respectively. www.nature.com/scientificreports/ of this, a screen of Fluspirilene analogues showed that Ca 2+ channel inhibition was necessary for their ability to induce autophagy in a neuroglioma cell line, since analogues that lacked the ability to block Ca 2+ channels lacked autophagy-inducing activity 71 . Importantly, the host Ca 2+ signaling pathway is also manipulated directly by Mtb to inhibit autophagy 73 . Whether Ca 2+ -related effects also play a role in HDT activity of Fluspirilene and Pimozide against intracellular Mtb and Stm, however, remains currently unknown and will require future studies. Recently, new diphenylbutylpiperidine analogues with stronger autophagy-inducing capabilities were developed that represent interesting novel HDT for future studies 74 . The second mechanism that Fluspirilene and Pimozide are likely to employ is the lysosomal response (Fig. 3). An additional new finding with mechanistic implications was that treatment with Fluspirilene and Pimozide strongly increased the intranuclear presence of TFEB (Fig. 3B). This transcriptional regulator controls a coordinated lysosomal response 75 , as well as autophagy 76 . This ability of Fluspirilene and Pimozide was, to the best of our knowledge, hitherto unknown. TFEB translocation has been observed for drugs of the phenothiazineclass including Trifluorperazine 77 , suggesting this might represent common functionality of several different antipsychotic drugs.
The third mechanism we uncovered was phagosomal acidification, a process vital in the host response against intracellular bacteria, which is inhibited by the protein CISH that is under transcriptional control of STAT5 55 . This pathway is exploited by Mtb which induces GM-CSF to activate STAT5, leading to enhanced CISH expression and consequent V-ATPase degradation 55 . Although treatment with both Fluspirilene and Pimozide led to a reduction of nuclear STAT5, we could experimentally confirm the decrease in CISH protein on Mtb containing vesicles only for Pimozide ( Fig. 4A and C). Further research will be needed to further unravel the subtle differences in precise MOAs between Pimozide and Fluspirilene, but this is beyond the scope of the current work.
As a fourth effector mechanism, we found that Pimozide strongly induced ROS/RNS production. Mechanistically, the efficacy of both Pimozide and Fluspirilene towards intracellular Mtb was inhibited by ROS/RNS inhibitors. Of these two diphenylbutylpiperidines, only Pimozide is known to induce ROS/RNS production and to inhibit the expression of antioxidant genes, like catalase, in vitro as well as in vivo 57,78 . Interestingly, similar to the results from Cai, Zhou 57 , addition of n-acetyl-cysteine (NAC), a general ROS scavenger, partly inhibited the Pimozide induced increase in ROS/RNS levels by human macrophages, and reduced bacterial survival ( Fig. 5B and D). Since multiple signaling pathways could contribute to the ROS/RNS production caused by Pimozide, we investigated this in more detail. The respiratory burst induced upon phagocytosis did not seem to be involved since Ebselen, a NADPH oxidase (NOX) inhibitor 79 , did not restore bacterial survival. Because L-NMMA, selective for nitric oxide synthase (NOS) affected Pimozide's efficacy, a role for nitric oxide (NO) seems likely. It has become evident that the functions of ROS and NO are intertwined and that inhibition of one leads to crossinhibition of the other, supporting this possibility 80,81 .
Taken together, we have identified four potential mechanistic effector pathways each of which is regulated by Fluspirilene and/or Pimozide, and which likely act in concert to reduce intracellular bacterial outgrowth. Combinatorial treatment regimens using multiple approaches including HDT as well as anti-bacterial molecules are considered indispensable to eradicating TB 82 . An interesting finding was that although Fluspirilene and Pimozide displayed additive effects with a low dose regimen of Rifampicin, this was not observed with Isoniazid. Isoniazid is a prodrug and requires conversion by the Mtb catalase-peroxidase (KatG) enzyme to its active form, resulting in the production of bacteria-derived reactive species 49,83 . These radicals have both direct and indirect antimicrobial properties and in the latter case host ROS/RNS production can lead to the induction of autophagy, which is capable of targeting Mtb 49 .
The repurposing of drugs with known safety profiles presents attractive and significant advantages in terms of drug development, including safety and toxicity testing and clinical evaluation. One important consideration is the feasibility of using these drugs at an effective yet safe dose for new indications. To our knowledge, data on Fluspirilene is sparse, but information regarding dosing and toxicity for Pimozide is available. The FDA recommends a maximum dose of 10 mg/day 84 , and doses in the range of 1 to 12 mg/day reach a serum level of Pimozide in the nanomolar range 85 . Oral delivery might not be optimally effective as treatment for pulmonary TB, and inhalation therapy, e.g., by aerosolic nebulizers, would be one way to achieve an optimally effective local concentration, while circumventing the systemic toxicity often associated with orally administered drugs 86 . As a proof of concept, Gupta et al. have shown that inhalable particles consisting of Rapamycin (a HDT drug) and two antibiotics cleared Mtb in mice in which case the Rapamycin concentration achieved in the lungs was 100 times higher compared to intravenous injection 87 .
Since the continuing rise of global drug resistance will affect the treatment of many bacterial infections, approaches complementary to bacterial-directed strategies can offer important supplemental approaches. We have demonstrated the novel HDT potential of two approved antipsychotic drugs, Fluspirilene and Pimozide, against Mtb and Stm, two important but unrelated intracellular pathogens. Our results provide new insights into the molecular and functional effects of these two diphenylbutylpiperidines on key mechanisms of host defense, particularly: autophagy, lysosomal acidification, ROS/RNS generation and phagosome maturation via STAT5/ CISH inhibition. We show that Fluspirilene and Pimozide target these host responses, by similar but subtly different MOAs, which is accompanied by reduced intracellular outgrowth of Mtb and Stm in human cells. Based on these findings, we propose that the class of diphenylbutylpiperidines could be repurposed as novel HDT drugs for both TB and salmonellosis and further explored as host-directed compounds against other intracellular pathogens.

Data availability
The datasets generated and analyzed during the current study are not deposited in external repositories but are included as supplementary file or available from the corresponding author on reasonable request.