Integration of chemical and rnai multiparametric profiles identifies triggers of intracellular mycobacterial killing

Cell Host & MicrobeArticle Integration of Chemical and RNAiMultiparametric Profiles IdentifiesTriggers of Intracellular Mycobacterial Killing Varadharajan Rico Nikolay Giovanni Jerome Gilleron,Inna Kalaidzidis,Felix Meyenhofer,Marc Yannis Kalaidzidis,and Marino 1Max-Planck Institute of Molecular Cell Biology and Genetics, 108 Pfotenhauerstrasse, 01307 Dresden, Germany2Belozersky Institute of Physico-Chemical Biology, Moscow State University, 119899 Moscow, Russia*Correspondence: Exploiting host processes to fight against infectious diseases is currently an underexplored approach. Yet, it presents Pharmacological modulators of host-microbial inter- several advantages. First, the host processes exploited by actions can in principle be identified using high- pathogens are well studied from both a mechanistic and content screens. However, a severe limitation of a physiological perspective ). Second, host this approach is the lack of insights into the mode kinase and phosphatase inhibitors that kill intracellular myco- of action of compounds selected during the primary bacteria have been successfully identified, thus validating the screen. To overcome this problem, we developed a approach (;Third, it could better prevent the emergence combined experimental and computational ap- of drug-resistant strains ( proach. We designed a quantitative multiparametric ). Additionally, critical host cellular responses are con- image-based assay to measure intracellular myco- served across diverse mycobacterial strains, including drug- bacteria in primary human macrophages, screened resistant ones (Given the complexity of a chemical library containing FDA-approved drugs, mycobacteria-host interactions, we reasoned that the target and validated three compounds for intracellular space for perturbations acting upon the host could be signifi- killing of M. tuberculosis. By integrating the multi- cantly larger than so far explored.
parametric profiles of the chemicals with those of Since mycobacteria block phagosome maturation siRNAs from a genome-wide survey on endocytosis, ; a substantial target we predicted and experimentally verified that two space could be related to endosomal/phagosomal trafficking.
compounds modulate autophagy, whereas the third This could be explored by cell-based small molecule screensmonitoring intracellular mycobacterial survival. However, a accelerates endosomal progression. Our findings severe constraint is that the primary assay typically does not demonstrate the value of integrating small molecules have the resolution to provide insights into the mode of function and genetic screens for identifying cellular mecha- of the selected hits ().
nisms modulated by chemicals. Furthermore, selec- In this study, we addressed this challenge by exploiting our tive pharmacological modulation of host trafficking knowledge of the molecular regulation of the endocytic path- pathways can be applied to intracellular pathogens way. We recently performed a cell-based genome-wide RNAi screen on endocytosis capturing various features of the endo-cytic system by quantitative multiparametric image analysis(QMPIA) (The QMPIA uncovered general principles governing endocytic trafficking that can also beapplied to the analysis of phagosome maturation. Here, we Tuberculosis is a disease characterized by an extensive array of first established a high-content assay to quantify intracellular interactions between mycobacteria and the host. To maintain mycobacterial survival and screened a library of pharmacologi- their intracellular niche, mycobacteria influence a variety of cally active compounds to identify host-specific modulators.
fundamental host cell processes including apoptosis ( Second, we developed an integrated experimental and com- putational pipeline to compare the multiparametric profiles (), trafficking ( from the chemical screen with those from the genome-wide ; ), and signaling ().
siRNA screen (This approach identified Therefore, a pharmacological modulation of the host cell mech- chemicals capable of overpowering intracellular mycobacterial anisms that are hijacked by the bacteria could, on the one hand, survival by modulating the cellular trafficking processes. Our be used to explore the processes altered during the infection results highlight the value of multiparametric phenotypic anal- and, on the other hand, lead to development of anti-TB ysis to extract functional information from combined chemical and genomic screens.
Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc. 129 Cell Host & Microbe Killing M. tuberculosis by Targeting the Host Validation of Three Compounds on M. tuberculosisTo validate the results from the screen, we selected three com- High-Content Chemical Screen to Identify Modulators of pounds within the cluster M5, namely Nortriptyline, Prochlorper- azine edisylate (PE), and Haloperidol on We developed an integrated experimental and computational the basis of (1) amplitude of the phenotypes ( strategy for screening, selecting and validating compounds en- showing reduced bacterial intensity, fewer bacteria, dowed with antimycobacterial activity by acting on the host and infected cells; and (2) lack of effect in the alamar blue screen.
machinery (A). We first established a high-content Additionally, these compounds did not show bacteriostatic assay in a human primary macrophage infection model activity on M. bovis BCG-GFP at neutral and acidic pH see the online).
B and S2C), or plate assay of M. tuberculosis (data not Briefly, cells were infected with M. bovis BCG expressing GFP and were incubated with To determine whether the phenotype observed on M. bovis chemicals for 48 hr. Since our primary interest was to identify BCG-GFP was reproducible on pathogenic M. tuberculosis, we chemical modulators disrupting the host-mycobacteria equilib- performed a colony-forming unit (CFU) assay on M. tuberculosis rium, chemicals were added after the infection. Cells were strain H37Rv and two clinical strains using the three selected then fixed and nuclei and cytoplasm stained and imaged using an automated spinning disc confocal microscope. We devel- Rapamycin ), and H89 oped further the QMPIA platform previously used to auto- as controls. All three selected compounds caused a significant matically identify cells, endosomes, and nuclei reduction in bacterial colonies in all the strains F), con- ) by implementing automated algo- firming that they impair intracellular survival of M. tuberculosis.
rithms to detect various features of intracellular bacteria andhost cell simultaneously ().
Integration of Multiparametric Chemical and Genetic We screened a bioactive library (Microsource Discovery Data Sets to Identify Cellular Processes Modulated by Inc., Spectrum collection, 2,000 compounds) comprising most FDA-approved drugs. The compounds produced repro- Since we aimed to identify compounds restoring mycobacteria- ducible alterations of various parameters such as bacterial arrested phagosome maturation, we explored the effects of the number, intensity, and bacterial and host cell morphology selected compounds on mycobacterial trafficking. We quantified the colocalization of M. bovis BCG-GFP with lysobisphosphati- We performed cluster analysis using an improvement of the dic acid (LBPA), a marker of late endocytic compartments ( method previously used ) based on density estimation (N.S., T. Galvez, C. Collinet, G.M., M.Z., and Y.K., and D and S2E). Infected cells treated with the unpublished data; The compounds selected compounds showed significantly higher colocalization grouped into six distinct clusters, each with a distinct pattern, than control cells G), indicating that the arrest in phag- i.e., phenotypic profile (and 1D, Table S1). The osome maturation was reverted.
strength of the phenotype is given by its phenoscore. Due to A significant challenge in phenotypic chemical screens is to the multiparametric nature of our screening output, we expected gain insights into the mechanism of action of the selected each cluster to capture distinct phenotypic features of mycobac- compounds. Multiparametric profiles are specific signatures of teria and the host. Cluster M5 (containing 115 compounds) was either chemical or genetic perturbations because they capture of particular interest, as it was characterized by a strong reduc- multiple facets of a biological system. Since random matches tion of bacterial parameters but relatively weak changes in host between such profiles are unlikely, clues into the molecular parameters reflecting toxicity, suggesting that the compounds in processes targeted by the chemicals could be obtained in this cluster are detrimental to bacteria but not significantly toxic principle by systematically searching for genetic alterations pro- to the host.
ducing similar phenotypes. Hence, we aimed to match the multi- Since we aimed to identify compounds that act upon the host, parametric profiles of the compounds from the mycobacteria we counterselected those directly killing mycobacteria by screen with those of the siRNAs from the genome-wide endocy- screening the MSD library using the alamar blue assay (see tosis screen (However, the two screens are the Interestingly, only six com- not directly comparable due to their distinct characteristics, i.e., pounds were shared with cluster M5, suggesting that most a chemical screen on intracellular mycobacterial survival in compounds in this cluster are not bacteriostatic. Additionally, human primary macrophages versus a functional genomics compounds functionally annotated as ‘‘antibacterial'' were not screen on endocytosis in HeLa cells. To overcome this problem, enriched in cluster M5 despite scoring in the alamar blue screen we bridged the two data sets via an intermediate screen, where (B). Standard antimycobacterial drugs were not en- the compounds from cluster M5 were screened on the endocy- riched in any particular cluster and yielded weak phenotypes tosis assay in HeLa cells (A). HeLa cells were pretreated (C–S1F, ). This is probably with the compounds prior to the internalization of fluorescently because our cell-based assay measures the reduction in GFP labeled epidermal growth factor (EGF) and Transferrin (Tfn) for fluorescence and therefore primarily reflects modulation of lyso- 10 min (), imaged, and multiple parameters somal degradation. Thus, the compounds of cluster M5 could extracted ().
modulate cellular processes at the host-pathogen interface Density estimation-based clustering separated the com- rather than acting directly on the bacteria.
pounds into two subclusters, E1 and E2 (Table S2), 130 Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc.

Cell Host & Microbe Killing M. tuberculosis by Targeting the Host Figure 1. High-Content Image-Based Screen for Intracellular Mycobacteria Survival(A) Flowchart of the experimental and computational pipeline used in this study.
(B) (Ba) Representative example of a human primary monocyte-derived macrophage infected with M. bovis BCG-GFP with cell contours (dashed line). (Bb)Segmentation of macrophage: intensity distribution of nuclear and cytoplasmic regions used to draw the contours of the cell. (Bc) Segmentation of bacteria:intensity distribution of bacteria in the region marked in the green square used for fitting the bacteria model.
(C) Heatmap representation of the mycobacteria high-content chemical screen after density estimation-based clustering. Rows denote cluster IDs and number ofcompounds. Columns are parameters of the high-content assay (). M5 is the cluster analyzed in this study.
(D) Representative images from clusters M1–M6.
See also and Table S1.
Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc. 131

Cell Host & Microbe Killing M. tuberculosis by Targeting the Host DAPI/Cell Mask Blue M. bovis BCG-GFP
M. bovis BCG-GFP with LBPA Colocalisation coef 0.1 Colocalisation threshold Figure 2. Validation of Three Selected Compounds for Intracellular Mycobacterial Survival(A–D) Representative images of infected cells treated with DMSO (A), Nortriptyline (B), prochlorperazine edisylate (PE) (C), and Haloperidol (D) at a concentrationof 10 mM for 48 hr after infection. Arrowheads indicate GFP-positive structures that are identified as ‘‘bacteria'' by the QMPIA. Scale bar, 10 mm.
(E) Quantitative multiparametric profiles of the phenotypes in (A)–(D) with the Z score (y axis) of the parameters (1–30) showing, e.g., decrease in number andintensity of intracellular mycobacteria. Error bars represent SEM.
(F) CFU analysis of human primary monocyte-derived macrophages infected with virulent M. tuberculosis H37Rv or two different clinical isolates CS1 and CS2and treated with the indicated compounds for 48 hr (pooled data from two independent biological replicates with at least three technical replicates each). Errorbars represent SD. D (DMSO 0.01%), Rf (Rifampin 2 mM), Rp (Rapamycin, 5.5 mM), PE (Prochlorperazine edisylate, 10 mM), N (Nortriptyline, 10 mM), H (Haloperidol,10 mM), H89 (H89, 10 mM). p values for all compounds tested (except H89) were <0.0001 and <0.01, for H37Rv and the two clinical strains, respectively.
(G) Colocalization of M. bovis BCG-GFP and LBPA in cells treated with the indicated drugs. Number of cells analyzed per condition is indicated. Error bars areSEM. Data are representative of at least two independent experiments. Abbreviations and concentrations are as in (F). See also 132 Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc.
Cell Host & Microbe Killing M. tuberculosis by Targeting the Host Figure 3. Integration of Chemical and RNAi Screen Data Sets(A) Scheme illustrating the integration of the primary chemical (green) and RNAi (blue) screen datasets via an intermediate screen on the chemicals fromcluster M5.
(B and C) Endocytosis (B) and autophagy(C) profiles of subclusters E1 and E2 from cluster M5. The profiles are average representations of Z scores of theindividual compounds within each subcluster. Parameters are grouped into different phenotype classes.
(D and E) Dose-response curves of mean integral intensities of M. bovis BCG-GFP (blue) and LC3 (magenta) for Nortryptyline (D) and prochlorperazine edisylate(E). Error bars are SEM, number of images is 25, data are representative of at least two independent experiments.
(F) Summary of IC50 values calculated from curves in (D) and (E). See also , Table S2, and Table S3.
suggesting that the compounds may act by at least two distinct sets of genes that are most likely to show similar phenotypes mechanisms. To identify them, we compared the multiparamet- (). We computed their representation ric profiles of the subclusters E1 and E2 to the genome-wide in KEGG pathways and GO terms (Table S3) and found that RNAi screen data set (and identified the ‘‘mTOR regulation'' and ‘‘regulation of autophagy and mTOR'' Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc. 133 Cell Host & Microbe Killing M. tuberculosis by Targeting the Host were significantly enriched in the E1 (p = 1.2 3 10
4 and tion (autophagic flux) (). To distinguish 9 3 10
4, respectively) but not the E2 (p = 0.15 and 0.4, respec- between these possibilities, we used two approaches. First, tively) subcluster.
we tested whether blockers of lysosomal degradation, e.g.,inhibitors of acidification (BafilomycinA1 or Chloroquine) that High-Content Autophagy Chemical Screen inhibit autophagic flux, had an additive effect with the com- Autophagy is a key process regulated by mTOR ( pounds on autophagosome accumulation ( and a known mediator of intracellular mycobac- ). HeLa cells were treated with Nortriptyline and PE in the terial survival ; The inte- presence or absence of BafilomycinA1, and LC3 lipidation gration approach predicted that compounds from E1 modulate (LC3-II band) was assessed by immunoblotting. Whereas both autophagy. To test it, we treated HeLa cells with the compounds, Nortriptyline and PE showed significant increase in the LC3-II immunostained for the autophagosome marker LC3, and band, additive effects of BafilomycinA1 were seen only upon extracted multiple parameters reporting diverse facets of auto- Nortriptyline, but not PE, treatment In addition, phagosomes ). The compounds of we generated HeLa cells stably expressing LC3 fused to GFP subcluster E1, but not E2, had a strong effect on autophagy under the control of the endogenous promoter through BAC re- (A), thus validating the prediction from the combineering (HeLa BAC LC3-GFP; The integration analysis. Of the three compounds validated in intra- results B and S5C) confirmed that Nortriptyline, but cellular mycobacterial killing, Nortriptyline and PE belong to not PE, had an additive effect with BafilomycinA1 and Chloro- subcluster E1, whereas Haloperidol belongs to E2. Consistent quine in increasing LC3-GFP vesicles.
with the data of Nortriptyline and PE, similar to Rapa- Second, we used an established assay for measuring auto- mycin or serum starvation (by HBSS), showed a strong effect on phagic flux based on a tandem mRFP-GFP-LC3 construct autophagy, whereas Haloperidol did not (and S3C).
(). Since GFP is quenched in the acidic envi- Human primary macrophages treated with Nortriptyline and PE ronment of autophagolysosomes whereas mRFP remains resembled HeLa cells in showing an increase in both number stable, a block in autophagy flux results in higher colocalization and integral intensity of LC3 vesicles (D and S3E), of mRFP and GFP compared to induction of autophagy. HeLa indicating that their effect of increasing autophagosomes is not cells stably expressing mRFP-GFP-LC3 ( ) were treated with the compounds, and the colocalization To determine whether the accumulation of autophagosomes between mRFP and GFP was quantified. Nortriptyline, similar is related to intracellular mycobacterial killing by Nortriptyline to starvation-induced autophagy (HBSS), showed a low mRFP and PE, we first computed their IC50 values in the two assays.
to GFP colocalization, whereas PE, similar to BafilomycinA1 For both compounds, the IC50 values for increased LC3 integral and Chloroquine, yielded higher colocalization A and intensity correlated well with decreased M. bovis BCG-GFP 5B). Similar results were obtained for the colocalization of LC3- (D–3F), arguing that the two phenotypes are linked.
GFP with LBPA in HeLa-LC3-GFP cells Based on Second, if autophagy modulation were involved in intracellular previous studies these results can be in- mycobacterial killing, we should visualize mycobacteria within terpreted as indicating that Nortriptyline induces autophagy autophagosomes. To test this, we analyzed mycobacteria- whereas PE inhibits autophagic flux.
infected cells by electron microscopy. Whereas in control cells In principle, blockers of autophagy flux should not have a detri- mycobacteria resided within single-membrane phagosomes mental effect on mycobacteria. Indeed, inhibitors of autophagy (cells treated with Nortriptyline and PE clearly flux that abrogate the function of lysosomes by blocking acidifi- showed the presence of mycobacteria within autophagosomes, cation did not affect the intensity of intracellular M. bovis defined as vesicles containing multiple membrane whorls BCG-GFP (E). In order to determine whether PE has and delimited by double membrane Haloper- a similar effect on lysosomal pH, we quantified the intensity of idol-treated cells, as expected, did not show this effect the acidotropic dye Lysotracker red over time. Unlike classical (A). Finally, we tested the effect of an inhibitor of auto- blockers of autophagy flux, PE did not have a negative effect phagy, 3-methyladenine (3MA) ). Upon on the acidity of lysosomes, but quite the contrary, it significantly treatment with Nortriptyline and PE in the presence of 3-MA, increased it (Moreover, this effect correlated with the mycobacteria were observed not in autophagosomes but in presence of mycobacteria within the lysotracker red-positive phagosomes (Additionally, 3-MA reverted vesicles C) and a concomitant reduction in bacterial the effect of Nortriptyline on mycobacterial survival in both intensity D).
CFU and image-based assays (data not shown), arguing that The increased acidification by PE is in apparent contradiction the effect of the compounds on mycobacteria is related to with the inhibition of autophagic flux. To solve this discrepancy, we treated HeLa cells stably expressing mRFP-GFP-LC3 withthe compounds for different periods of time. If autophagy flux Nortriptyline and PE Modulate Autophagy by Different were completely blocked, the colocalization between mRFP and GFP (corresponding to nonacidic autophagosomes) should Autophagy is a multistep process consisting of autophagosome increase with time, whereas if the flux were slowed down, then formation, maturation, and eventual degradation by fusion with the colocalization would be expected to decrease (i.e., autopha- lysosomes (autophagolysosomes). The increase in autophago- gosomes become acidic). The colocalization of mRFP and GFP somes mediated by Nortriptyline and PE could be due to either increased over time in cells treated with BafilomycinA1 ( a higher rate of formation or a block in maturation and degrada- E) or Chloroquine (data not shown). In contrast, in 134 Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc.

Cell Host & Microbe Killing M. tuberculosis by Targeting the Host <0.0001 <0.0001 Autophagosome area Figure 4. Ultrastructural Analysis of Autophagy and Intracellular Mycobacterial Survival(A–C) Electron micrographs of M. bovis BCG-GFP-infected human primary monocyte-derived macrophages treated with DMSO (A), 10 mM N (B), or PE (C) for 4 hr.
Black arrows indicate mycobacterial phagosomes, and black and white arrowheads indicate autophagosomes with and without mycobacteria, respectively.
Mitochondria (Mi), nucleus (Nuc), mycobacterial phagosome (MP), autophagosomes (AP), mycobacteria containing autophagosome (MAP).
(D–F) Quantification of number of autophagosomes (D) (n = 30), autophagosome area as percentage of total cell area (E) and total cell area (F) (n = 6) from theelectron micrographs. D (DMSO, 0.01%), N (Nortriptyline, 10 mM), PE (Prochlorperazine edisylate, 10 mM). Numbers denote p values in comparison to DMSO. Seealso Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc. 135

Cell Host & Microbe Killing M. tuberculosis by Targeting the Host Figure 5. Nortriptyline and PE Modulate Autophagy Differently(A) Representative images of HeLa cells stably expressing mRFP-GFP-LC3 treated with the indicated chemicals for 2 hr. DAPI/CMB denotes the cell nucleus andcytoplasm. Scale bar, 10 mm.
(B) Quantification of colocalization between the mRFP and GFP signals. The number of cells analyzed per condition is indicated. Results are representative of fourexperiments. Error bars are SEM. Abbreviations and concentrations are as in (A).
(C and D) Quantification of integral intensity of lysotracker red-labeled vesicles (C) and M. bovis BCG-GFP (D) in infected human primary macrophages treatedwith either DMSO or PE (10 mM) for the indicated periods of time. Total and bacterial LTR refer to integral intensity of all acidic compartments and those containingbacteria, respectively. Data are averaged from 45 images/condition/time point and normalized to DMSO to represent fold change (representative of twoindependent experiments).
(E) Colocalization between the mRFP and GFP signals in HeLa cells stably expressing mRFP-GFP-LC3 treated with the indicated drugs for different periodsof time. Colocalization at a single threshold of 0.35 was used to generate the time response curves. Data are from 50 images per condition and from fourexperiments. For (A), (B), and (E), D (DMSO, 0.01%), Hb (HBSS), N (Nortriptyline,10 mM), ChQ (Chloroquine, 25 mM), Baf (BafilomycinA1, 100 nM), PE(10 mM).
(F) Mean integral intensity of M. bovis BCG-GFP upon treatment of infected human primary macrophages with different concentrations of IFN-g for 4 hr prior totreatment with either 1 mM or 10 mM N or PE, for 48 hr. Significance was calculated by Student's t test. NS p > 0.05, *p < 0.05, **p < 0.001, ***p < 0.0001. Data areaveraged from 16 images/condition and are representative of four experiments. For (B)–(F), error bars indicate SEM. See also PE-treated cells the colocalization increased within the first hour induction. PE instead exerted a much weaker effect, providing but then progressively decreased (E), showing that the additional evidence that the two compounds interact differently autophagosomes acidify over time.
with the autophagy machinery.
To further characterize the effects of Nortriptyline and PE on autophagy, we tested whether they could synergize with an es- Haloperidol Accelerates Endolysosomal Trafficking tablished physiological inducer of autophagy, Interferon-gamma Haloperidol kills intracellular mycobacteria by a mechanism distinct from autophagy. Haloperidol belongs to subcluster E2 preactivated infected cells with IFN-g and assessed the effect (This subcluster is characterized by a strong alter- of the compound addition on mycobacterial survival. The results ation of the endocytic profile exemplified by an increase in were striking (F) and showed that Nortriptyline EGF and Tfn content, increased colocalization of EGF to Tfn synergized strongly with IFN-g, supporting its role in autophagy (reduced sorting), and accumulation of EGF in large endosomes 136 Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc.
Cell Host & Microbe Killing M. tuberculosis by Targeting the Host Figure 6. Analysis of Endocytosis Traf-ficking by Haloperidol(A and B) The kinetics of endocytic trafficking werequantified using a cell-based LDL traffickingassay. (A) Integral intensity of LDL in early endo-somes and (B) late endosomes. Values are Htreatment (10 mM) subtracted from DMSO andderived from an average of 1,200 cells/time point.
Error bars are SEM of number of images (averageof 40 per condition). Data are representative of twoexperiments.
(C and D) A431 cells stably expressing Rab5-GFPwere treated with either DMSO or Haloperidol(10 mM) for 2 hr and imaged live. Velocities ofminus-end (toward cell center) (C) and plus-end- directed movement (away from cell center) areshown for DMSO-treated cells (black line) andH-treated cells (green line). For curves in (B)–(D),the p value is for points denoted by asterisks (seethe ). For (C) and (D),error bars are standard deviations, n = 37 (DMSO),n = 29 (H).
(E) Mean square displacement curves for themovies (and 6D) were computed asdescribed in the . Thesecond part of the biphasic curve indicating longtimescale movements was fitted using linear least-square fit. Diffusion coefficient is derived from theslope of the fitted curves. See also Movie S1, and Movie S2.
results confirm the prediction that Halo-peridol leads to acceleration of traffickingalong the degradative pathway.
Phagosome maturation depends on the fusion with early, late endosomesand lysosomes (;) and is regu-lated by the motility of endocytic organ-elles along the cytoskeleton and therecruitment of the membrane tetheringfusion machinery. Both processes couldbe enhanced by Haloperidol. We first in the cell center. In accordance with the design principles of the tested the influence of Haloperidol on endosome motility.
endocytic system (; ), the Increasing concentration of Rab5 and displacement of early en- alterations observed argue for a faster degradation. Hence, we dosomes toward cell center are predictive of progression from hypothesized that Haloperidol could positively modulate endo- early to late endosomes ( cytic trafficking along the degradative route. To test this hypoth- ). We therefore quantified endosome motility in cells esis, we monitored the kinetics of transport of low-density stably expressing GFP-Rab5, as previously described ( lipoprotein (LDL) from early to late endosomes, using Rab5 ) (Movie S1 and Movie S2), and computed the velocity (and LBPA ) as markers, toward the cell center (minus end microtubule directed) and respectively (A). HeLa cells stably expressing GFP- periphery (plus end) as a function of GFP-Rab5 intensity. Halo- Rab5 from the endogenous promoter (HeLa BAC GFP-Rab5c) peridol-treated cells showed a much higher motility of early en- () were treated with Haloperidol and allowed dosomes harboring high levels of Rab5, and this enhancement to internalize LDL for different times prior to immunostaining was higher toward the cell center (and 6D). Next, with anti-Apolipoprotein B (ApoB) (to detect LDL) and LBPA anti- we computed the mean square displacement of endosomes bodies. Haloperidol significantly increased the intensity of LDL in over time, an established way to characterize motility patterns late, but not early, endosomes (and 6B). Similarly, (). In Haloperidol-treated cells, long, but not LDL-containing late, but not early, endosomes were larger, short, timescale movements of GFP-Rab5 endosomes were more numerous, and closer to the cell center in Haloperidol- significantly accelerated E). Thus, Haloperidol acceler- treated compared to control cells (B–S6G). These ates the centripetal movement of early endosomes.
Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc. 137

Cell Host & Microbe Killing M. tuberculosis by Targeting the Host M. bovis BCG with EEA1 (Haloperidol minus DMSO) Colocalisation coef M. bovis BCG with LBPA p value = 0.00011 (Haloperidol minus DMSO) Colocalisation coef Figure 7. Haloperidol Induces Alterations in Intracellular Mycobacterial Trafficking(A and C) Representative images of M. bovis BCG-GFP-infected human primary macrophages treated with 10 mM Haloperidol for the indicated times andimmunostained for EEA1 (A) and LBPA (C). The outline of infected cells is shown in dashed lines. Scale bar, 10 mm. The presence of EEA1 and LBPA onmycobacterial phagosome, as indicated by arrows and arrowheads (at 20 min), respectively, is mutually exclusive.
(B and D) Quantification of colocalization of M. bovis BCG-GFP with EEA1 (B) and LBPA (D). A stringency threshold of 0.35 was used. Data are averaged fromabout 1,000 infected cells and are representative of at least two independent experiments. Error bars are SEM. The p value denotes significance for theconsecutive points of the time series denoted by asterisks (see the Second, we tested whether Haloperidol could lead to increased delivery of the membrane-tethering and fusionmachinery to the arrested mycobacterial phagosome. Myco- In this study, we developed a pipeline of experimental and bacterial phagosomes contain Rab5, but not Rab7 ( computational methods to identify pharmacological modulators and lack PI(3)P ), of host cellular processes that trigger killing of intracellular myco- which mediates the recruitment of various Rab5 effectors, bacteria. First, we established a multiparametric image-based including the membrane tethering factor early endosomal assay to quantify mycobacterial survival in human primary antigen 1 (EEA1) ). Therefore, myco- macrophages. Second, we used a pharmacological approach bacterial phagosomes lack EEA1 ). If Haloper- to disrupt the host-pathogen equilibrium in favor of the host.
idol could restore such machinery, one would expect to see Third, we developed computational methods to compare the a small, transient increase in EEA1 on mycobacterial phago- multiparametric profiles of chemicals with those of siRNAs, to somes. We treated M. bovis BCG-infected cells with Haloper- identify the cellular mechanisms underlying the antibacterial idol for different times and quantified the colocalization of activities of the compounds. Fourth, we validated the predictions mycobacteria with EEA1 and LBPA. Whereas in control cells, from such comparison on the three selected compounds and es- as expected, mycobacterial phagosomes were devoid of tablished that two of them modulate autophagy via distinct EEA1 at all times (upper panels; B), in Halo- mechanisms and the third one accelerates endocytic trafficking.
peridol-treated cells a small but significant fraction of EEA1 Our results demonstrate the predictive value of comparing quan- localized to mycobacterial phagosomes 20–30 min posttreat- titative multiparametric profiles from chemical and genetic ment A, lower panels; Such an effect was screens to decipher the cellular mechanisms of chemical action.
specific for Haloperidol, as it was not observed in cells treated They further show that trafficking pathways can be modulated with Nortriptyline and PE (data not shown). Moreover, in Halo- pharmacologically to stimulate intracellular mycobacteria killing.
peridol-treated cells the colocalization of mycobacteria and The key feature of our intracellular mycobacterial survival LBPA increased with time C and 7D). Altogether, assay is the multiplicity and reliability of parameters capturing these data indicate that Haloperidol restores the delivery of crit- various features of bacteria and host at high resolution. Interest- ical components of the membrane-tethering machinery to the ingly, most compounds in the selected hit cluster (M5) are not mycobacterial phagosomes, thereby delivering mycobacteria directly antimycobacterial. Although we cannot rule out to lysosomes.
completely the possibility that the compounds may kill bacteria 138 Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc.
Cell Host & Microbe Killing M. tuberculosis by Targeting the Host directly within infected cells, we favor the interpretation that they ). PE may act at different stages, e.g., on the fusion act by modulating the host for the following reasons. First, the with lysosomes. Detailed characterization of the interactions of M5 cluster showed very little overlap with the compounds which Nortriptyline and PE with the autophagy machinery will lead to act directly on bacteria. Second, the three selected compounds a more precise understanding of their mode of action.
do not kill mycobacteria outside the host cell under the different The identification of Haloperidol as an antimycobacterial conditions tested. Third, they show similar effects on uninfected compound provides proof of principle for the pharmacological cells. Fourth, the effect of Nortriptyline and PE can be reverted by modulation of endocytic trafficking as a means of boosting the inhibiting a host cellular process (autophagy). Fifth, Nortriptyline host defense system to clear intracellular pathogens. Most showed strong phenotypic correlation with a structural analog, importantly, our data imply that such an approach is feasible Amitriptyline, in mycobacteria as well as endocytosis and auto- without inducing overt toxicity to cells. Haloperidol leads to phagy screens. Here, only structural analogs having a Tanimoto faster degradation of cargo by accelerating transport from early distance >0.9 were considered and to late endocytic compartments. Previous work has identified F–S2I). Sixth, Nortriptyline and PE enhance the acidity both general design principles and molecular details of the of lysosomes. This correlates with delivery of mycobacteria to machinery underlying endosome biogenesis and progression lysosomes and their survival. Additionally, standard antimyco- that are relevant to the pharmacological properties of Haloper- bacterial drugs showed weaker phenotypes compared to the idol. First, transport of cargo (LDL) along the degradative route selected compounds. Most likely, this is because our assay entails a conversion from Rab5- to Rab7-positive endosomes measures GFP fluorescence from mycobacteria that decays at (). Such conversion is based on a cut-out switch different rates, depending on whether the host degradation () that requires the endosomal levels machinery can be accelerated or not. In support of this, Rifam- of Rab5 to pass a threshold value prior to triggering a feedback picin did not show enhanced lysosomal acidity. Thus, our loop causing its removal and substitution with Rab7. Second, our screening platform has a bias for compounds that boost host recent systems survey on endocytosis has revealed that the mechanisms. Importantly, our data argue that such compounds centripetal movement of endosomes previously observed by are likely to be missed in a screen for antimycobacterial quantitative live-cell imaging is a master compounds outside the context of infected cells.
parameter that regulates several other properties of the early en- Whereas cell-based high-content screens have several dosome network, including cargo progression ( advantages over traditional target-based HT screens, under- ). Mycobacterium interferes with such a switch, keeping standing the mechanism of action of the identified compounds the phagosome in an arrested state. Remarkably, Haloperidol is a critical bottleneck (One exerts a positive influence on both centripetal movement of en- possible approach is to identify genes that when silenced dosomes and membrane recruitment of the Rab5 machinery.
produce a phenotype similar to that of small inhibitory molecules By shifting the molecular composition of the mycobacterial (However, there are significant caveats to phagosome, Haloperidol counteracts the fine balance between this approach: for example, small molecules usually act rapidly, phagosome activity and maturation arrest, restoring maturation.
whereas RNAi is gradual and requires longer times, possibly Thus, relatively subtle alterations in the general endocytic traf- allowing the system to adapt to a new steady state. Moreover, ficking machinery are sufficient to cause dramatic effects on the biological activity of small molecules may result from the intracellular mycobacterial survival.
modulation of multiple targets rather than Our results revealed antimycobacterial properties for three ex- specific ones as in RNAi, excluding off-target effects. We isting drugs. While Nortriptyline is an antidepressant, PE and reasoned that although it may prove difficult to identify targets Haloperidol are antipsychotics. The serum concentrations of small molecules by comparing the phenotypic profiles of achieved in patients for these drugs are lower than the in vitro chemicals and silenced genes, this approach may nevertheless antibacterial IC50 (). In principle, they would allow the identification of the cellular pathways affected, as long have to be made significantly more potent or achieve signifi- as the phenotypes are described with sufficient detail and spec- cantly higher serum levels to be effective in antimycobacterial ificity. Therefore, we captured the phenotypes through QMPIA, therapy. However, it is difficult to compare the effectiveness of and, by integrating chemical and functional genomics data the drugs under the particular conditions used in vitro with their sets, we differentiated hit compounds of the mycobacteria pharmacokinetics properties in patients, and further studies are screen into distinct groups on the basis of their phenotypes on necessary to corroborate their antimycobacterial activity in vivo.
endocytosis. Using this strategy, we could show both computa- Nevertheless, our work demonstrates that it is possible to find tionally and experimentally that one of the groups is significantly existing compounds with previously unrecognized antimyco- enriched for modulators of autophagy, a well-known antimyco- bacterial activity and mode of action. Interestingly, PE is also bacterial cell-defense mechanism ( used as antiemetic and administered in patients with MDR-TB ). Remarkably, autophagy phenotypes could thus be to alleviate ancillary symptoms of nausea and vomiting identified using a general endocytosis assay.
The finding that three neuromodulators are en- Nortriptyline and PE influence autophagy differently. Whereas dowed with membrane trafficking and antimycobacterial activity Nortriptyline induces the formation of autophagosomes, PE suggests that compounds can exert diverse biological effects slows down autophagic flux. The basal rate of autophagy may depending on the cellular context analyzed. Such influences be sufficient to accumulate autophagosomes in the presence are typically neglected in target-based screenings. Repurposing of PE. These autophagosomes progressively acidify over time existing drugs for additional applications could thus open new and become competent for mycobacteria degradation ( avenues and lead to rapid therapeutic advances Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc. 139 Cell Host & Microbe Killing M. tuberculosis by Targeting the Host However, it is not excluded that the effects on intracellular score >3 were considered hits. The screen was repeated twice (Z0 factors transport highlighted in this study may even be part of the mech- are 0.85 and 0.9 for each run). Hits from both runs were combined.
anism of action of the compounds in neuromodulation. The High-Content Autophagy Assay mechanistic clues obtained here open the possibility of exploring HeLa cells were treated with the compounds at a final concentration of 10 mM, the role of cellular processes, i.e., autophagy and modulation of incubated for 4 hr, fixed with 3.7% PFA, and immunostained with anti-LC3 endocytic trafficking, in the context of their original therapeutic antibody (clone PM036, MBL Labs). Cells were counterstained by DAPI and use. Indeed, many tricyclic antidepressants, regarded as medi- Cell Mask blue, imaged, and analyzed as described above.
ators of neurotransmitter signaling, show significant effects onautophagy (). Synergistic Dose-Response Curves activities of such compounds with other antibacterial and To calculate IC50 for mycobacterial survival, human primary monocyte-derivedmacrophages were infected with M. bovis BCG-GFP, treated with different immune modulators, such as IFN-g, could be exploited to yield concentrations of the compounds for 48 hr, and fixed. For autophagy assay, more potent pharmacological interventions. Finally, the methods the same preparation of cells was treated with the compounds for 2 hr, fixed, described here can be adapted to other diseases where host cell and immunostained for LC3. Cells were imaged and analyzed as described.
processes need to be targeted. The compounds identified maytherefore also be effective against pathogens using similar M. tuberculosis CFU Assay Human primary macrophages were infected with virulent M. tuberculosis strainH37Rv and two different clinical isolates (CS1 and CS2) for 1 hr at moi 1:10,washed, treated with Amikacin (200 mg/ml) for 45 min, washed again, and EXPERIMENTAL PROCEDURES treated with the compounds. Cells were lysed after 48 hr with 0.05% SDSand plated in multiple dilutions on 7H11 agar plates. Colonies were counted High-Content Screen for Intracellular Mycobacteria Survival after incubation at 37C for 3 weeks.
Human primary monocyte-derived macrophages were infected with M. bovisBCG-GFP for 1 hr at moi of 1:10, washed extensively (Biotec EL406). After Electron Microscopy adding the compounds (10 mM) (Beckman Fxp automatic workstation), cells Cells were fixed with 2.5% glutaraldehyde followed by postfixation in ferrocy- were incubated for 48 hr, fixed with 3.7% PFA (Sigma), stained with 1 mg/ml anide-reduced osmium following standard protocols () DAPI and 3 mg/ml Cell Mask Blue (Invitrogen), and imaged using the automated and embedded with epon. Then 70 nm thin sections were cut on a Leica ultra- spinning disk confocal Opera (Perkin Elmer). See the microtome, stained with uranyl acetate and lead citrate following classical for details.
procedures, and imaged on a 100 kV Tecnai 12 TEM microscope. For areameasurements, several overlapping high-magnification images covering the Data Analysis: Density-Based Clustering whole cell were acquired, and a montage was generated using ImageJ.
Multiparametric profiles from the chemical screen were clustered usingdensity-based clustering algorithm with von Mises-Fisher kernel used for Live-Cell Imaging density estimate (N.S., T. Galvez, C. Collinet, G.M., M.Z., and Y.K., unpub- A431 cells stably expressing GFP-Rab5 were treated with 10 mM Haloperidol lished data). Clustering was performed varying the kernel concentration for 2 hr and imaged live using Zeiss Duo scan microscope at 0.098 s/frame for parameter k. For every data set, the number of clusters obtained stabilized 2 min. Individual endosomes were identified and tracked over time as within a k range of 10–150. Further increase of k would not lead to an increase described (). To calculate mean square displacement, endoso- in the number of clusters. For each data set, a cluster was manually selected at mal tracks were first iteratively divided into nonoverlapping equal time intervals the first value of k that produced clusters that were stable in number and size.
from 0.1 to 70 s, and their displacements were computed for each interval.
k values for the mycobacteria and endocytosis screens were 30 and 90, Next, the square of displacement for all endosomal tracks for a given condition was averaged and plotted as a function of the time interval. In such a curve,early endosomes show a characteristic biphasic behavior, which we refer to Integrating Chemical and Genetic Screens as short timescale and long timescale movements.
Representative profiles of the two endocytosis subclusters were generated byaveraging the profiles of individual compounds in each subcluster. For these Estimation of p Value for Time Series two profiles, we assembled a list of genes from the endocytosis screen ( The null hypothesis tested is that deviation of a sequence of N measurements ) that had a Phenoscore >0.95 and a Pearson correlation of 0.7.
from zero is the result of normally distributed noise. Given that every measure- These lists were subjected to gene annotation enrichment analysis using ment i in the series has a value Di and variance s2i, we computed the probability a standard hypergeometric test. Gene annotations were obtained from Gene of the null hypothesis (p value) as Ontology and KEGG Pathway databases.
Intracellular Mycobacterial Trafficking Assay pvalue = 1 B 2 @1
erf@ 2@P Human primary monocyte-derived macrophages were infected with M. bovis BCG-GFP and stained for lysotracker red or LBPA as markers of lateendosomes and lysosomes (Colocalizations were scored over a range of stringency thresholds SUPPLEMENTAL INFORMATION varying from 0 (most relaxed) to 1 (most stringent).
Supplemental Information includes six figures, two movies, three tables, Alamar Blue Screen for Antibacterial Compounds Supplemental Experimental Procedures, and Supplemental References and To mid-log phase M. bovis BCG-GFP seeded into 384-well plates (OD can be found with this article at 0.03) in a volume of 90 ml 7H9, compounds were added to a final concentrationof 10 mM. DMSO and Amikacin (200 mg/ml) were included as negative and positive controls, respectively. Plates were incubated at 37C for 4 days, fol-lowed by addition of 10 ml of 1:1 mix of alamar blue (serotec) and 10% We thank Dr. J. Pieters (Biozentrum, University of Basel) for M. bovis BCG-GFP Tween-80. After 24 hr at 37C, fluorescence was read in the Tecan Pro (Ex strain, Dr. J. Gruenberg (University of Geneva) for anti-LBPA antibody, Drs. A.
535 nm; Em 590 nm). Analysis was done per plate. Compounds with a Z Ballabio and D. Medina (TIGEM, Naples) for HeLa mRFP-GFP-LC3 stable 140 Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc.
Cell Host & Microbe Killing M. tuberculosis by Targeting the Host clone, Drs. E. Jacobs and S. Monecke (TU, Dresden) for providing work to infer endosomal network dynamics from quantitative image analysis.
M. tuberculosis strains, and Dr. E. Fava (DKNZ, Bonn) for the initial work. We Curr. Biol. 22, 1381–1390.
are indebted to N. Tomschke, M. Stoeter, A. Giner, S. Seifert, C. Moebius, Fratti, R.A., Backer, J.M., Gruenberg, J., Corvera, S., and Deretic, V. (2001).
C. Andree, and the HT-TDS for expert technical assistance; I. Poser and Role of phosphatidylinositol 3-kinase and Rab5 effectors in phagosomal BAC recombineering facility for HeLa LC3-GFP BAC and HeLa Rab5-GFP biogenesis and mycobacterial phagosome maturation arrest. J. Cell Biol.
BAC cell lines; the High-Performance Computing Center at TUD, M. Chernykh, and the computer department for expert IT assistance; and E. Perini, M.
Gutierrez, M.G., Master, S.S., Singh, S.B., Taylor, G.A., Colombo, M.I., and McShane, and K. Simons for critical comments on the manuscript. This Deretic, V. (2004). Autophagy is a defense mechanism inhibiting BCG and work was supported by EU-FP7-funded projects NATT, PHAGOSYS, APO- Mycobacterium tuberculosis survival in infected macrophages. Cell 119, SYS, and the Max-Planck Society.
Received: August 2, 2012 Huang, R., Southall, N., Wang, Y., Yasgar, A., Shinn, P., Jadhav, A., Nguyen, Revised: November 2, 2012 D.T., and Austin, C.P. (2011). The NCGC pharmaceutical collection: a compre- Accepted: January 3, 2013 hensive resource of clinically approved drugs enabling repurposing and chem- Published: February 13, 2013 ical genomics. Sci. Transl. Med. 3. http://dx.doi.org/10.1126/scitranslmed.
3001862, 80ps16.
Imming, P., Sinning, C., and Meyer, A. (2006). Drugs, their targets and thenature and number of drug targets. Nat. Rev. Drug Discov. 5, 821–834.
Alix, E., Mukherjee, S., and Roy, C.R. (2011). Subversion of membrane trans- Jayachandran, R., Sundaramurthy, V., Combaluzier, B., Mueller, P., Korf, H., port pathways by vacuolar pathogens. J. Cell Biol. 195, 943–952.
Huygen, K., Miyazaki, T., Albrecht, I., Massner, J., and Pieters, J. (2007).
Armstrong, J.A., and Hart, P.D. (1971). Response of cultured macrophages to Survival of mycobacteria in macrophages is mediated by coronin 1-dependent Mycobacterium tuberculosis, with observations on fusion of lysosomes with activation of calcineurin. Cell 130, 37–50.
phagosomes. J. Exp. Med. 134, 713–740.
Jayaswal, S., Kamal, M.A., Dua, R., Gupta, S., Majumdar, T., Das, G., Kumar, Behar, S.M., Divangahi, M., and Remold, H.G. (2010). Evasion of innate immu- D., and Rao, K.V. (2010). Identification of host-dependent survival factors for nity by Mycobacterium tuberculosis: is death an exit strategy? Nat. Rev.
intracellular Mycobacterium tuberculosis through an siRNA screen. PLoS Microbiol. 8, 668–674.
Bucci, C., Parton, R.G., Mather, I.H., Stunnenberg, H., Simons, K., Hoflack, B., Karnovsky, M. (1971). Use of ferrocyanide-reduced osmium tetroxide in and Zerial, M. (1992). The small GTPase rab5 functions as a regulatory factor in electron microscopy. In Proceedings of the Eleventh Annual Meeting of the the early endocytic pathway. Cell 70, 715–728.
American Society for Cell Biology.
Caspi, A., Granek, R., and Elbaum, M. (2002). Diffusion and directed motion in Keane, J., Remold, H.G., and Kornfeld, H. (2000). Virulent Mycobacterium cellular transport. Phys. Rev. E Stat. Nonlin. Soft Matter Phys. 66, 011916.
tuberculosis strains evade apoptosis of infected alveolar macrophages.
J. Immunol. 164, 2016–2020.
Chen, M., Gan, H., and Remold, H.G. (2006). A mechanism of virulence: viru- Kimura, S., Noda, T., and Yoshimori, T. (2007). Dissection of the autophago- lent Mycobacterium tuberculosis strain H37Rv, but not attenuated H37Ra, some maturation process by a novel reporter protein, tandem fluorescent- causes significant mitochondrial inner membrane disruption in macrophages tagged LC3. Autophagy 3, 452–460.
leading to necrosis. J. Immunol. 176, 3707–3716.
Klionsky, D.J., Abeliovich, H., Agostinis, P., Agrawal, D.K., Aliev, G., Askew, Christoforidis, S., McBride, H.M., Burgoyne, R.D., and Zerial, M. (1999). The D.S., Baba, M., Baehrecke, E.H., Bahr, B.A., Ballabio, A., et al. (2008).
Rab5 effector EEA1 is a core component of endosome docking. Nature 397, Guidelines for the use and interpretation of assays for monitoring autophagy in higher eukaryotes. Autophagy 4, 151–175.
Collinet, C., Sto¨ter, M., Bradshaw, C.R., Samusik, N., Rink, J.C., Kenski, D., Kobayashi, T., Stang, E., Fang, K.S., de Moerloose, P., Parton, R.G., and Habermann, B., Buchholz, F., Henschel, R., Mueller, M.S., et al. (2010).
Gruenberg, J. (1998). A lipid associated with the antiphospholipid syndrome Systems survey of endocytosis by multiparametric image analysis. Nature regulates endosome structure and function. Nature 392, 193–197.
Koul, A., Herget, T., Klebl, B., and Ullrich, A. (2004). Interplay between myco- Del Conte-Zerial, P., Brusch, L., Rink, J.C., Collinet, C., Kalaidzidis, Y., Zerial, bacteria and host signalling pathways. Nat. Rev. Microbiol. 2, 189–202.
M., and Deutsch, A. (2008). Membrane identity and GTPase cascades regu-lated by toggle and cut-out switches. Mol. Syst. Biol. 4, 206. http://dx.doi.
Kuijl, C., Savage, N.D., Marsman, M., Tuin, A.W., Janssen, L., Egan, D.A., Ketema, M., van den Nieuwendijk, R., van den Eeden, S.J., Geluk, A., et al.
(2007). Intracellular bacterial growth is controlled by a kinase network around Desjardins, M., Huber, L.A., Parton, R.G., and Griffiths, G. (1994). Biogenesis PKB/AKT1. Nature 450, 725–730.
of phagolysosomes proceeds through a sequential series of interactions withthe endocytic apparatus. J. Cell Biol. 124, 677–688.
Kumar, D., Nath, L., Kamal, M.A., Varshney, A., Jain, A., Singh, S., and Rao,K.V. (2010). Genome-wide analysis of the host intracellular network that regu- Dhandayuthapani, S., Via, L.E., Thomas, C.A., Horowitz, P.M., Deretic, D., and lates survival of Mycobacterium tuberculosis. Cell 140, 731–743.
Deretic, V. (1995). Green fluorescent protein as a marker for gene expressionand cell biology of mycobacterial interactions with macrophages. Mol.
Laplante, M., and Sabatini, D.M. (2012). mTOR signaling in growth control and Microbiol. 17, 901–912.
disease. Cell 149, 274–293.
Dyer, M.D., Murali, T.M., and Sobral, B.W. (2008). The landscape of human Matsuzawa, T., Kim, B.H., Shenoy, A.R., Kamitani, S., Miyake, M., and proteins interacting with viruses and other pathogens. PLoS Pathog. 4, e32.
Macmicking, J.D. (2012). IFN-g elicits macrophage autophagy via the p38 MAPK signaling pathway. J. Immunol. 189, 813–818.
Eggert, U.S., Kiger, A.A., Richter, C., Perlman, Z.E., Perrimon, N., Mitchison, Pieters, J. (2008). Mycobacterium tuberculosis and the macrophage: main- T.J., and Field, C.M. (2004). Parallel chemical genetic and genome-wide taining a balance. Cell Host Microbe 3, 399–407.
RNAi screens identify cytokinesis inhibitors and targets. PLoS Biol. 2, e379.
Ponpuak, M., Davis, A.S., Roberts, E.A., Delgado, M.A., Dinkins, C., Zhao, Z., Virgin, H.W., 4th, Kyei, G.B., Johansen, T., Vergne, I., and Deretic, V. (2010).
Fairn, G.D., and Grinstein, S. (2012). How nascent phagosomes mature to Delivery of cytosolic components by autophagic adaptor protein p62 endows become phagolysosomes. Trends Immunol. 33, 397–405.
autophagosomes with unique antimicrobial properties. Immunity 32, 329–341.
Foret, L., Dawson, J.E., Villasen˜or, R., Collinet, C., Deutsch, A., Brusch, L., Poser, I., Sarov, M., Hutchins, J.R., He´riche´, J.K., Toyoda, Y., Pozniakovsky, Zerial, M., Kalaidzidis, Y., and Ju¨licher, F. (2012). A general theoretical frame- A., Weigl, D., Nitzsche, A., Hegemann, B., Bird, A.W., et al. (2008). BAC Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc. 141 Cell Host & Microbe Killing M. tuberculosis by Targeting the Host TransgeneOmics: a high-throughput method for exploration of protein function Seglen, P.O., and Gordon, P.B. (1982). 3-Methyladenine: specific inhibitor of in mammals. Nat. Methods 5, 409–415.
autophagic/lysosomal protein degradation in isolated rat hepatocytes. Proc.
Natl. Acad. Sci. USA 79, 1889–1892.
Regenthal, R., Krueger, M., Koeppel, C., and Preiss, R. (1999). Drug levels:therapeutic and toxic serum/plasma concentrations of common drugs.
Settembre, C., Di Malta, C., Polito, V.A., Garcia Arencibia, M., Vetrini, F., Erdin, J. Clin. Monit. Comput. 15, 529–544.
S., Erdin, S.U., Huynh, T., Medina, D., Colella, P., et al. (2011). TFEB linksautophagy to lysosomal biogenesis. Science 332, 1429–1433.
Rink, J., Ghigo, E., Kalaidzidis, Y., and Zerial, M. (2005). Rab conversion asa mechanism of progression from early to late endosomes. Cell 122, 735–749.
Shubin, H., Sherson, J., Pennes, E., Glaskin, A., and Sokmensuer, A. (1958).
Prochlorperazine (compazine) as an aid in the treatment of pulmonary tubercu- Rossi, M., Munarriz, E.R., Bartesaghi, S., Milanese, M., Dinsdale, D., Guerra- losis. Antibiotic Med. Clin. Ther. 5, 305–309.
Martin, M.A., Bampton, E.T., Glynn, P., Bonanno, G., Knight, R.A., et al.
Swinney, D.C., and Anthony, J. (2011). How were new medicines discovered? (2009). Desmethylclomipramine induces the accumulation of autophagy Nat. Rev. Drug Discov. 10, 507–519.
markers by blocking autophagic flux. J. Cell Sci. 122, 3330–3339.
Vergne, I., Chua, J., Lee, H.H., Lucas, M., Belisle, J., and Deretic, V. (2005).
Russell, D.G. (2001). Mycobacterium tuberculosis: here today, and here Mechanism of phagolysosome biogenesis block by viable Mycobacterium tomorrow. Nat. Rev. Mol. Cell Biol. 2, 569–577.
tuberculosis. Proc. Natl. Acad. Sci. USA 102, 4033–4038.
Schmid, S.L., and Cullis, P.R. (1998). Membrane sorting. Endosome marker is Zschocke, J., Zimmermann, N., Berning, B., Ganal, V., Holsboer, F., and fat not fiction. Nature 392, 135–136.
Rein, T. (2011). Antidepressant drugs diversely affect autophagy pathways Schwegmann, A., and Brombacher, F. (2008). Host-directed drug targeting of in astrocytes and neurons—dissociation from cholesterol homeostasis.
factors hijacked by pathogens. Sci. Signal. 1, re8.
142 Cell Host & Microbe 13, 129–142, February 13, 2013 ª2013 Elsevier Inc.

Source: https://publications.mpi-cbg.de/Sundaramurthy_2013_5321.pdf