Development of a multiplex fluorescent reporter series, the Salmonella INtracellular Analyser (SINA) to distinguish different intracellular S. Typhimurium lifestyles

The distinct intracellular lifestyles of S. Typhimurium upon invasion of epithelial cells have been described either with regard to the specific pathogen localization or with regard to the bacterial growth dynamics. To date, fluorescent reporters are available to identify S. Typhimurium within vacuolar and cytosolic localizations; while the others measure the replication rate of the pathogen [18–20]. However, these localization and replication-rate reporters have not been coupled, as it has been generally assumed that the bacterial localization determines its replication rate. This notion has been challenged by different reports, for example on the different growth rates of S. Typhimurium within the cytosol depending on the targeting by autophagy [21–25]. A combined reporter system would enable a comprehensive elucidation of the intracellular lifestyle of a given intracellular pathogen. Therefore, we developed a novel fluorescent reporter series, the Salmonella INtracellular Analyzer (SINA). Our SINA1.1 reporter is composed of two separated modules to indicate the bacterial localization and replication rate. The localization module consists of two transcription reporters driven by localization-specific promoters, while the replication rate module carries a constitutively expressed fluorescent timer (Figs 1A and S1). At the molecular level, the localization module is composed of vacuolar (Vac) and cytosolic (Cyt) submodules, which utilize two characterized promoters, P_ssaG and P_uhpT to drive the expression of tagBFP and smURFP, respectively [18,19]. We confirmed the functionality of the fluorescent P_ssaG and P_uhpT reporters for our experimental setup during S. Typhimurium invasion of epithelial cells using a digitonin assay assayed by flow cytometry (S2 and S3 Figs). The replication rate module encodes Timer^bac, a DsRed mutant (S197T), which has been previously employed to differentiate S. Typhimurium subpopulations by their replication rates [20]. The emission spectrum of Timer^bac shifts from green to red as it matures, which reflects the bacterial metabolic activity (change in slope, Fig 1B top) as well as the replication rate (unvarying slope, varying green:red ratio, Fig 1B bottom) [26]. When Timer^bac is constitutively expressed, a metabolically active S. Typhimurium bacterium emits both green and red signals resulting from immature green and mature red fluorophores. In case such a bacterium experiences a metabolic halt, it eventually emits only red signals, due to the maturation of the existing green fluorophores in concert with ceased de novo synthesis of the green fluorophores. With SINA, we are able to simultaneously collect information on these replication rate changes of S. Typhimurium and its localization at single bacterium resolution, which enables a comprehensive and quantitative reflection of S. Typhimurium physiology inside an infected host.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. SINA enables precise determination of the different Salmonella intracellular lifestyles in human epithelial cells.

(A) Schematic diagram of the construction of subcellular localization and replication rate modules of SINA1.1. The subcellular localization module is composed of the vacuolar submodule (P_ssaG-tagBFP) and cytosolic submodule (P_uhpT-smURFP), while the replication rate module is composed of a constitutively expressed Timer^bac (P_ybaJ-Timer^bac) (B) (Top) Schematic diagram of the emission spectrum shifts of S. Typhimurium harboring Timer^bac as Timer^bac matures, where emission shifts from green to red (Bottom) Green:Red ratio increases with elevating S. Typhimurium replication rates. As S. Typhimurium divides, both Timer⁵¹⁰ and Timer⁵⁸⁰ fluorophores are diluted. With a higher production rate of Timer⁵¹⁰ than Timer⁵⁸⁰, fast dividing S. Typhimurium exhibits a higher Green:Red ratio. (C) Expected output by SINA as S. Typhimurium dwells in distinct subcellular localizations. Vacuolar S. Typhimurium are of lower replication rate (i.e. lower Green:Red ratio) and are expected to emit blue fluorescence; cytosolic S. Typhimurium are of higher replication rate (i.e. higher Green:Red ratio) and are expected to emit far red fluorescence (D) HeLa cells infected by S. Typhimurium harboring SINA1.1. Output of SINA from intracellular S. Typhimurium was detected by fluorescence microscopy at 1 h pi, vacuolar (arrowhead) and cytosolic (arrow) S. Typhimurium at 6 h pi. (3 independent experiments). Scale bars are 10 μm. (E) HeLa cells infected by S. Typhimurium harboring SINA1.1. Output of SINA from intracellular S. Typhimurium at 1 h and 6 h pi was detected by flow cytometry (3 independent experiments).

https://doi.org/10.1371/journal.ppat.1009550.g001

To validate the functionality of our SINA system during S. Typhimurium invasion of epithelial cells, we employed fluorescence microscopy and flow cytometry analysis (Fig 1C). With fluorescence microscopy, we observed intracellular S. Typhimurium simultaneously emitting both green and red signals (Timer^bac), but not Vac (P_ssaG-tagBFP) and Cyt (P_uhpT-smURFP) signals at 1 hour post-infection (pi) (Fig 1D). As these bacteria committed to vacuolar or cytosolic lifestyles at 6 hours pi, we observed that S. Typhimurium in cells with <10 bacteria emitted Vac signal during this time course. On the other hand, we observed a mixed population of S. Typhimurium in cells with >10 bacteria, where clusters of Cyt⁺ S. Typhimurium of low Timer^bac signals (arrow) and individual Vac⁺ S. Typhimurium (arrowhead) were detected. In the cells containing mixed S. Typhimurium populations, bacteria were either Vac⁺ or Cyt⁺ but not double positive, showing the presence of two populations with distinct discernible lifestyles (Fig 1D). We were also able to track the onset of bacterial division and signal output from SINA by time-lapse microscopy (S1–S3 Movies, S4B and S4C Fig).

We took advantage of the properties of our multiplex SINA reporter and devised a gating strategy to quantitatively analyze the bacterial lifestyles in single S. Typhimurium-infected cells using flow cytometry (S2 Fig). In brief, we first defined the infected cells by the size of the analyzed events (under SSC-A vs FSC-A plot), followed by the positive signals in the Timer⁵⁸⁰ vs Timer⁵¹⁰ plot (i.e. cells harboring S. Typhimurium). We then further classified the S. Typhimurium-infected cells into four sub-types according the signals of the localization module (tagBFP::SPI-2 vs smURFP::cytosolic plot), corresponding to cells with either vacuolar bacteria (Vac⁺Cyt^-) or cytosolic bacteria (Vac^-Cyt⁺) or cells with both vacuolar and cytosolic populations (Vac⁺Cyt⁺) or cells harboring S. Typhimurium that express only basal levels of the Vac and Cyt signals (S2 Fig). We observed that intracellular S. Typhimurium behaved as a population with a homogenous replication rate and basal expression levels of Vac and Cyt at 1 hour pi (Fig 1E). At 6 hours pi, this homogenous population segregated into Vac⁺ and Cyt⁺ subpopulations, with a Cyt⁺ distribution similar to that reported in the literature (10–20% cytosolic) (Fig 1E) [21]. The gradual separation of these subpopulations could be detected with SINA1.1 throughout the course of infection (S4A Fig). As we backgated the infected cells, we observed cells harboring only vacuolar S. Typhimurium (Vac⁺Cyt^-) and cells with both vacuolar and cytosolic bacteria (Vac⁺Cyt⁺), each forming a distinct population of different Green:Red ratio on the plot of Timer⁵¹⁰ against Timer⁵⁸⁰ (S5A Fig). Together, this demonstrated that our novel SINA1.1 reporter is capable of simultaneously and quantitatively distinguishing the S. Typhimurium lifestyles by their subcellular localization and replication rate at both single infected cell and single bacterium level using flow cytometry and fluorescence microscopy. The combination of the SINA reporter with flow cytometry fosters higher throughput analysis of S. Typhimurium lifestyles in infected cells as compared to microscopy, extending the possibility for rapid screening.

A novel dormant S. Typhimurium subpopulation in human epithelial cell

With the SINA1.1 reporter, we used the SPI-2 expression module to distinguish vacuolar S. Typhimurium from cytosolic bacteria. In the plot of the localization module, we identified an easily discernable population (~5–10%) of infected epithelial cells harboring Vac^-Cyt^- S. Typhimurium detectable as early as 2 hours pi, which became apparent at 6 hours pi (Figs 2A and S6). We backgated the Vac^-Cyt^- population, and we extracted physical parameters from the Timer^bac plot. This revealed that the Vac^-Cyt^- S. Typhimurium exhibit a similar replication rate (S5A Fig) but a reduced metabolic activity (S5B Fig) compared to Vac⁺Cyt^- S. Typhimurium as depicted by the green:red ratio and slope of Timer^bac plot, respectively. The capacities of Timer^bac in the measurement of the bacterial replication rate and metabolism have been well-elaborated in previous applications [20,27]. This Vac^-Cyt^- S. Typhimurium population was also visualized using live microscopy to confirm their presence using different detection approaches (S3 Movie). This was intriguing as metabolically inactive S. Typhimurium have not been reported in enterocytes so far. We thus infected polarized intestinal epithelial Caco-2 monolayers, and confirmed the presence of the Vac^-Cyt^- subpopulation with a shifted metabolic profile in a cellular model system for intestinal infections (S7 Fig). We also performed control infections in 3T3 (fibroblast model) and differentiated THP-1 cells (macrophage model) to test the sensitivity of SINA1.1 in these relevant cell types. As within epithelial cells, we also observed distinct S. Typhimurium populations in the macrophage and fibroblast models as described before (S8 and S9 Figs). To determine the intracellular localization of Vac^-Cyt^- S. Typhimurium, we further performed correlative light and electron microscopy (CLEM) by serial section transmission electron microscopy (TEM), and a digitonin assay demonstrating that this subpopulation is localized in a host vesicular compartment (Figs 2B, S4A–S4F and S10) [21]. Together, these results showed the presence of a novel intracellular S. Typhimurium population within epithelial cells that exhibits a lowered metabolic rate and resides in a host vesicular compartment, implicating a putative dormant phenotypic variant.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. S. Typhimurium displays a novel inactive intracellular lifestyle in epithelial cells.

(A) (Left and Middle) Timer^bac profile and distribution of single cells with no infection (black), infected cells with inactive bacteria (Vac^-Cyt^-) (red), infected cells with only vacuolar bacteria (Vac⁺Cyt^-) (blue) and infected cells with both vacuolar and cytosolic populations (Vac⁺Cyt⁺) (green) at 6 h pi. (Right) Abundance of S. Typhimurium-infected cells (Vac^-Cyt^-, Vac⁺Cyt^- and Vac⁺Cyt⁺) as illustrated in (A) (n = 3). (B) (Left) Brightfield and fluorescent microcopy (FLM) images of infected HeLa cells harboring Vac^-Cyt^- S. Typhimurium at 6 h pi. (Right) Serial sections of TEM images of Vac^-Cyt^- S. Typhimurium, arrowhead indicates host membrane structures of the SCV. (C) Schematic illustration for the constructions of SINA derivatives, SINA1.4 and SINA1.5. SINA1.4 was used for immunofluorescence staining against RAB5, RAB7, RAB11, LAMP1 and LC3; SINA1.5 was used for arabinose induction assay. (D) (Top) Responsiveness of intracellular S. Typhimurium towards an arabinose pulse between 5–6 h pi, uninduced control (black); arabinose-induced (red). (Bottom) Quantification on the responsiveness of Vac^- S. Typhimurium pulsed at different time intervals during the infection time course, dormant (black), inducible (maroon). Samples were all harvested at 6 h pi. (n = 3) (E) HeLa cells were infected with SINA1.4-harboring S. Typhimurium, harvested at 6 h pi, fixed and stained. Quantification of the presence of RAB5, RAB7, BAB11, LAMP1 and LC3 proximal to Vac^- and Vac⁺ S. Typhimurium at 6 h pi. (n = 3) (F) Representative images of Vac^- S. Typhimurium (arrow) quantified in (D); S. Typhimurium (green), Vac^- (blue), RAB5, RAB7, RAB11, LAMP1 and LC3 (grey), Phalloidin (red). (G) Designated populations of infected HeLa cells were enriched by cell sorting and plated for CFU. Quantification of CFU from dormant S. Typhimurium at 6 h, 24 h and 168 h pi and Vac⁺Cyt^- S. Typhimurium at 6 h pi. (n = 5 for Vac⁺Cyt^- 6 h, Dormant 6 h, 24 h; n = 3 for Dormant 168 h) (H) Quantification of SPI-2 activity using flow cytometry in enriched dormant S. Typhimurium at 6 h and enriched Vac^-Cyt^- infected cells re-plated until 24 h pi. (n = 3) (I) Survival percentage of dormant and Vac⁺Cyt^- intracellular S. Typhimurium against 3 h of CIP treatment, infected cells were harvested at 6 h pi, enriched by cell sorting and plated for CFU. (n = 4) At least a total of 1000 events of infected cells were analyzed by flow cytometry or 50 infected cells by microscopy in each experiment replicates. The bars represent the mean, statistics were performed using unpaired t test (**p<0.01).

https://doi.org/10.1371/journal.ppat.1009550.g002

Intracellular S. Typhimurium encounters a number of stresses upon uptake into host cells, including oxidative, pH and osmotic stress, which serve as key signals to trigger transcription reprogramming for the adaptation of an intra-host environment [28]. As the intracellular microenvironment and S. Typhimurium dormancy has been studied in some detail in macrophages, we decided to focus our comparison on this cell type in relation to the newly identified S. Typhimurium subpopulation in epithelial cells. During macrophage infections the SCV microenvironment drives a portion of S. Typhimurium into a dormant state that contributes to the elevation of antimicrobial persistence and polarization of infected macrophage [14,29]. We asked whether Vac^-Cyt^- S. Typhimurium shares similar physiologies with dormant S. Typhimurium inside macrophages, hence we determined the metabolic state of the Vac^-Cyt^- population. By replacing the cytosolic submodule of SINA1.1 with an arabinose inducible cassette to generate SINA1.5, we measured S. Typhimurium’s capacity to respond to arabinose treatment. This modification enabled us to directly monitor the metabolic activity of S. Typhimurium during the infection process (Fig 2C). The response of intracellular bacteria towards extracellular arabinose induction has been reported previously to characterize the metabolic state of macrophage-borne dormant S. Typhimurium [13]. With SINA1.5, we observed that approximately half of the Vac^- S. Typhimurium did not respond to arabinose induction at designated time intervals (Fig 2D). As the Vac^-Cyt^- bacterial niche may show limited arabinose accessibility, we corroborated our observations monitoring the reduced metabolism of this bacterial population via the signals of the Timer^bac reporter (S5B Fig). These data allowed us to propose that the Vac^-Cyt^- S. Typhimurium adopts a dormant state (coined as dormant S. Typhimurium hereafter) upon their internalization into epithelial cells.

Dormant S. Typhimurium resides in a unique vesicular compartment

We set out to determine whether the dormant S. Typhimurium localization is distinct from conventional SCVs. To determine this by immunofluorescence staining, we simplified SINA1.1 to SINA1.4 to free the red and far-red channels for indirect immunofluorescence staining of selected endocytic markers (Fig 2C). LAMP1 labels host lysosomes as well as the matured SCV, which is also present on the SCV of dormant S. Typhimurium in macrophages [7,13]. By fluorescence microscopy, we only observed minor recruitment of LAMP1 to the proximity of dormant S. Typhimurium within epithelial cells, in contrast to the high LAMP1 incidence proximal to Vac⁺ S. Typhimurium (Fig 2E and 2F). We also determined if dormant S. Typhimurium is localized in a SCV experiencing a halt in SCV biogenesis, therefore we tested a number of known early SCV markers (namely RAB5, RAB7 and RAB11). We did not observe the recruitment of any of these early SCV markers to the dormant S. Typhimurium (Fig 2E and 2F). This was intriguing as our previous ultrastructural and digitonin investigations clearly documented the dormant S. Typhimurium bacteria within a membrane-bound compartment. We also addressed if the dormant population is targeted by host autophagy, analyzing the localization of the autophagy marker LC3. We did not detect any localization of LC3 proximal to the majority of the dormant S. Typhimurium (Fig 2E and 2F). Therefore, we conclude that dormant S. Typhimurium are localized within a unique membrane-bound compartment distinct from the conventional SCV and that of dormant S. Typhimurium in macrophages, suggesting such dormancy formation is heavily governed by endocytic trafficking [13,30]. This compartment requires further characterization in future studies.

Dormant S. Typhimurium are viable, cultivable, resume metabolism and express virulence genes in host cells

Endocytic vesicles are either recycled or undergo fusion with the lysosomes for degradation. The same fate also applies to vacuolar S. Typhimurium, where SPI-2 deficient strains have a reduced survival capacity compared to SPI-2 competent strains [31]. We collected the infected cells harboring dormant S. Typhimurium by cell sorting at >90% purity and plated them for colony forming unit (CFU) measurement, a classical approach to determine the viability of the intracellular S. Typhimurium. We observed that dormant S. Typhimurium are viable and cultivable (Fig 2G), contrasting to the viable but not cultivable nature of dormant S. Typhimurium in murine macrophages [13,14]. To determine the fate of the dormant S. Typhimurium, we enriched and plated the viable infected cells harboring dormant S. Typhimurium, and monitored the bacterial behavior at 24 hours pi (S11 Fig). We observed that 50% of the dormant S. Typhimurium in infected cells collected at 6 hours pi became metabolically active and expressed SPI-2 at 24 hours pi, as demonstrated by the population shift in the Timer^bac plot, and becoming Vac⁺ (Fig 2H). To determine if the dormant S. Typhimurium persists in the host, we further enriched infected cells harboring dormant S. Typhimurium and monitored the presence and the viability of dormant S. Typhimurium at 7 days pi. The dormant S. Typhimurium were found to persist in cells and remained viable and cultivable over the whole period of 7 days (Fig 2G, “168h”). To further address whether the dormant intracellular S. Typhimurium phenotype confers a reduced sensitivity towards antibiotics, we supplemented ciprofloxacin (CIP) to the infected cell at 3 hours pi and determined the viability of dormant S. Typhimurium by CFU. We selected the in cellulo CIP supplementation to avoid artificial perturbations on the S. Typhimurium dormant status. Also, in cellulo treatment addresses the killing efficiency of dormant S. Typhimurium within the SCV, where subcellular distribution of this drug has been demonstrated to influence the bactericidal efficacy [32]. We observed a higher survival rate of dormant S. Typhimurium as compared to vacuolar S. Typhimurium (Fig 2I), similar to the observations made in the murine intestine [17]. These results demonstrated that dormant S. Typhimurium are viable, exhibit a delayed expression of SPI-2, persist in the epithelial host cells for up to 7 days and are less susceptible to antibiotics. Such unique metabolic and virulence reprogramming could serve as a strategic step for intestine-borne S. Typhimurium to prolong gut inflammation for community benefits and reservoir for relapse [33].

S. Typhimurium dormancy is not a result of the loss of T3SS2 effector secretion

We then studied whether the lack of T3SS2 effector secretion drives S. Typhimurium dormancy in epithelial cells. Using our SINA1.1 reporter in the S. Typhimurium SPI-2 secretion deficient mutant ΔssaV [34], we observed no significant difference in the proportion of Vac^-Cyt^- S. Typhimurium between wild type and the ΔssaV mutant (Fig 3A). Altogether, the formation of dormant S. Typhimurium is not a consequence caused by the lack of T3SS2 effector secretion during the infection of epithelial cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. S. Typhimurium dormancy is negatively regulated by SpoT.

(A) HeLa cells were infected with SINA1.1-harboring S. Typhimurium, the abundance of Vac^-Cyt^- population in wild type and SPI-2 mutant ΔssaV infected cells were quantified with flow cytometry at 6 h pi. (n = 3) (B) Schematic diagram for the construction of SINA derivative, SINA1.9, yielded from the introduction of an arabinose-inducible hilA expression cassette into SINA1.1. SINA1.9 was used to rescue the reduced invasiveness of ΔdksA, ΔrelA and ΔrelAspoT mutant strains. (C) HeLa cells were infected with SINA1.1 or SINA1.9-harboring S. Typhimurium, the abundance of Vac^-Cyt^- population in (p)ppGpp biogenesis and regulon mutants, Δlon, ΔdksA, ΔrelA and ΔrelAspoT were quantified by flow cytometry at 6 h pi (n = 3) (D) Distribution of Vac^-Cyt^-, Vac⁺Cyt^- and Vac⁺Cyt⁺ populations in hilA-expressing wild type (Left) and ΔrelAspoT mutant (Middle) infected HeLa cells at 6 h pi quantified by flow cytometry. Overlay Timer^bac profile (Right) of Vac^-Cyt^- (red) and Vac⁺Cyt^- (blue) populations of wild type and Vac^-Cyt^- population of ΔrelAspoT mutant (grey) in infected HeLa cells quantified by flow cytometry at 6 h pi. (3 independent experiments) At least a total of 1000 events of infected cells were analyzed by flow cytometry in triplicate experiments. Statistics were performed using unpaired t test. ns: not significant (P > 0.05), **P < 0.01, ****P < 0.0001.

https://doi.org/10.1371/journal.ppat.1009550.g003

S. Typhimurium dormancy is regulated by (p)ppGpp biogenesis

Class II toxin-antitoxin (TA) systems regulate the dormancy formation of non-pathogenic E. coli in laboratory conditions [35]. TA systems are comprised of a toxin and an antitoxin that counter-balances the toxin to regulate bacterial physiology, including growth arrest. A major TA system involves the stringent response mediated by the monofunctional (p)ppGpp synthases RelA and bifunctional (p)ppGpp synthases SpoT, after which (p)ppGpp binds to DksA to mediate transcription reprogramming for bacterial adaptation. The surge in (p)ppGpp levels also activates the ATP-dependent Lon protease to degrade Type II antitoxins to release the free toxins [36–38]. In recent reports, stringent response has been associated with slow growing S. Typhimurium populations, and TA systems are implicated in S. Typhimurium dormancy in macrophages [14,16]. Therefore, we assessed the links between the stringent response and S. Typhimurium dormancy in epithelial cells, studying the mutant strains (i) ΔrelA ((p)ppGpp synthase), (ii) ΔrelAΔspoT ((p)ppGpp synthases), (iii) ΔdksA ((p)ppGpp-binding transcription regulator) and (iv) Δlon (protease targeting antitoxin). With the Δlon mutant, we did not observe any difference in the level of dormant S. Typhimurium population in infected cells (~5–10%), suggesting that Lon protease is dispensable for S. Typhimurium dormancy in epithelial cells (Fig 3C). As ΔrelAspoT and ΔdksA were reported to suffer reduced invasiveness in epithelial cells due to the reduced SPI-1 expression, we thus constructed SINA1.9 (Fig 3B), a derivative of SINA1.1 with an additional cassette for an inducible expression of hilA to compensate the reduced invasiveness of the mutants (S12 Fig) following a previously published experimental strategy [39]. With the SINA1.9-complemented mutant strains, we obtained rescued invasiveness as compared to wild type S. Typhimurium. This allowed us to address the requirement of (p)ppGpp biogenesis and (p)ppGpp-regulated transcription for S. Typhimurium persistence. A significant increase in the Vac^-Cyt^- population was observed in ΔrelAspoT, whereas the increment was less pronounced in the ΔrelA single mutant and was indifferent in ΔdksA mutant, when comparing with the wild type strain (Fig 3C and 3D, Left and Middle panel). As SpoT has been reported to regulate SPI-2 expression [40], we further confirmed that the Vac^-Cyt^- population of ΔrelAspoT shared a comparable metabolic profile as the one observed in the wild type strain (Fig 3D, Right panel). Together, these results suggested that (p)ppGpp stringent response mediated by SpoT but not RelA is required to restrict dormancy entry of S. Typhimurium within epithelial cells independent of the DksA regulon, while SPI-2 effector expression and secretion and Lon protease are dispensable.

Mammalian cell culture

HeLa cervical adenocarcinoma cells, Caco-2 colorectal adenocarcinoma cells, 3T3 mouse fibroblasts and THP-1 acute monocytic leukemia cells were purchased from American Type Culture Collection (ATCC) and used within 20 passages of receipt. HeLa cells and 3T3 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM, high glucose, GlutaMAX Supplement, ThermoFisher) containing 10% (v/v) heat-inactivated fetal bovine serum (FBS, Sigma) and incubated at 37°C with 5% CO₂ and 100% humidity. Caco-2 cells were cultured in DMEM containing 10% FBS, 1% Non-essential amino acids (Gibco), 1% HEPES (Gibco), 1% Penicillin/Streptomycin (Gibco) and incubated at 37°C with 5% CO₂ and 100% humidity. THP-1 cells were cultured in RPMI-1640 medium (ThermoFisher) containing 10% FBS and incubated at 37°C with 5% CO₂ and 100% humidity. HeLa and 3T3 cells were seeded in 12-well tissue-culture treated plates (Corning Costar) at a density of 9x10⁴ cells/well 48 hours prior to infection. Caco-2 cells were polarized using Corning BioCoat Assay System (Corning) following manufacturer’s protocol. THP-1 cells were seeded in 12-well tissue-culture treated plates at a density of 9x10⁴ cells/well 96 hours prior to infection, and differentiated in 50 μg/mL phorbol 12-myristate 13-acetate (PMA, Sigma) for 24 hours, and incubated in RPMI-1640 + 10% for 72 hours. For immunofluorescence staining, HeLa cells were seeded on UV-treated glass coverslips (Marienfeld) in 12-well plates 48 hours prior to infection. For cell sorting experiments, HeLa cells were seeded in 10 cm tissue-culture treated dishes (Corning Costar) at a density of 1.8x10⁶ cells/well 48 hours prior to infection.

Bacterial strains

Bacterial strains and plasmids used in this study are listed in S1 and S2 Tables, respectively. All mutants were constructed using bacteriophage λ red recombinase system from parental stain S. Typhimurium Typhimurium strain SL1344 using primers listed in S3 Table [48]. HA-tagged T3SS2 effector strains were generated by transducing JL129 with P22 phage lysate (a generous gift from Stéphane Méresse, Centre d’Immunologie de Marseille-Luminy, France). Bacteria were cultured in Lysogeny broth (LB) supplemented with appropriate antibiotics, where necessary (Ampicillin 100 μg/mL; Kanamycin 50 μg/mL).

Plasmid construction

The replication rate module, Timer^bac is a generous gift from Dr. Dirk Bumann (University of Basel, Switzerland) [20]. To construct the localization module, tagBFP was amplified from pHRdSV40-NLS-dCas9-24xGCN4_v4-NLS-P2A-BFP-dWPRE using primers tagBFP_fw and tagBFP_rv, and replaced the GFP in pM973 to yield vacuolar module (pP_ssaG-tagBFP) [19,49]. For the cytosolic module (pP_uhpT-smURFP), smURFP-HO-1 and uhpT promoters were amplified from pBAD smURFP-HO-1 (smURFP_fw and smURFP_rv) and S. Typhimurium gDNA (uhpT_fw and uhpT_rv), respectively, and replaced the sfGFP and mxiE promoter in pTSAR1 [50,51]. The vacuolar (Vac_fw and tagBFP_rv) and cytosolic (uhpT_fw and Cyt_rv) modules were amplified and inserted into EcoRV and SmaI sites, respectively, of pBlueScript II KS (+) to generate pSINA-int. The localization module on pSINA-int was excised and inserted between SalI and SphI sites of pBR322 Timer^bac to yield pSINA1.1. pSINA1.4 was generated by replacing Timer^bac with GFP in pBR322 Timer^bac and inserted the vacuolar module at the SalI and SphI sites. pSINA1.5 was generated by inserting the amplified inducible smURFP cassette (Ara_fw and Ara_rv) and vacuolar cassette between the SalI and EagI sites of pBR322 Timer^bac. pSINA1.7 was constructed by reverting Timer^bac to DsRed by site-directed mutagenesis using DsRed_fw and DsRed_rv. pBAD hilA was generated by inserting the amplified hilA (hilA_fw1 and hilA_rv1) between the BamHI and PmeI sites of pBAD smURFP-HO-1. The inducible hilA cassette was amplified using primer hilA_fw2 and hilA_rv2 and inserted into the EcoRV site of pSINA1.1 to generate pSINA1.9.

Bacterial infections

Bacteria strains were streaked from glycerol stock on LB agar plates with appropriate antibiotics 2 days prior to infection. Three bacterial colonies were picked for overnight culture in LB medium supplemented with 0.3 M NaCl with shaking at 37°C. 150 μL overnight culture was subculture in 3 mL LB + 0.3 M NaCl (1:20 dilution) with shaking at 37°C for 3 h. For strains harboring pSINA1.9, 0.1% L-arabinose was supplemented to the subculture 1 h before harvest. Bacteria were harvested with centrifugation (1 mL, 6000 x g, 1 min, RT), washed once in 1 x PBS and resuspended in DMEM with no FBS. HeLa cells were infected at a MOI of ~100 for 25 min at 37°C. Extracellular bacteria were removed and washed with 1 x PBS (5X). Cells were then incubated in DMEM + 10% FBS for 1 h, washed with 1 x PBS (3X), incubated in DMEM + 10% FBS for 2 h, washed with 1 x PBS (3X) and then incubated in DMEM + 10% FBS supplemented with 10 μg/mL gentamicin for the remaining time course of the infection.

Flow cytometry

At designated time points, cells were washed with 1 x PBS (1X) and detached with 0.05% Trypsin for 5 min at 37°C. Detached cells were mixed with equal volume of DMEM + 10% FBS, passed through 40 μm strainer and collected by centrifugation (500 x g, 5 min, 4°C). Cell pellets were dislodged and fixed in 4% PFA (15 min, RT). Fixed cells were washed with 1 x PBS (2X) and resuspended in 200 μL 1 x PBS for further analysis. For digitonin permeabilization experiment, cells were permeabilized with 45 μg/mL digitonin (1 min, RT) or 0.25% saponin (30 min, RT), then washed and stained with anti-S. Typhimurium primary antibody and Alexa488-conjugated goat anti-rabbit secondary antibody [21]. The fluorescence intensities of the samples were assayed with LSR Fortessa (BD) (tagBFP Ex: 405 nm Em: 450/50 nm; Timer⁵¹⁰ Ex: 488 nm Em: 525/50 nm; Timer⁵⁸⁰ Ex: 562 nm Em: 582/15 nm; smURFP Ex: 633 nm Em: 670/30 nm) and analyzed with FlowJo (v10.0.4). The recorded events were gated according to the strategy described (S2 Fig).

Cell sorting

At designated time points, cells were washed with 1 x PBS (1X) and detached with 0.05% Trypsin for 5 min at 37°C. Detached cells were mixed with equal volume of DMEM + 10% FBS, passed through 40 μm strainer and collected by centrifugation (500 x g, 5 min, 4°C). Cells were washed with 1 x PBS (1X) and resuspended in DMEM + 10% FBS supplemented with 10 μg/mL gentamicin. SYTOX Green (ThermoFisher) was supplemented to differentiate dead cells when necessary. Cells were sorted with Aria III (BD) (tagBFP Ex: 405 nm Em: 450/50 nm; Timer⁵¹⁰ and SYTOX Green Ex: 488 nm Em: 530/30 nm; Timer⁵⁸⁰ Ex: 561 nm Em: 586/15 nm; smURFP Ex: 633 nm Em: 660/20 nm) to collect uninfected cells, infected cells with dormant or SPI-2 S. Typhimurium populations. The recorded events were gated according to the strategy described (S2 Fig).

Immunofluorescence microscopy

Cells seeded on coverslips were washed with 1 x PBS (1X) and fixed in 4% PFA (8 min, RT). After washing with 1 x PBS (3X), cells were permeabilized and blocked in 1 x PBS, 20% FBS, 0.25% saponin (30 min, RT). Coverslips were washed with 1 x PBS (3X) and incubated with anti-RAB5, anti-RAB7, anti-RAB11, anti-LC3 or anti-LAMP1 primary antibodies and phalloidin-rhodamine diluted in 1 x PBS, 2% FBS (60 min, RT), and then washed with 1 x PBS (3X) and incubated with Cy5-conjugated goat anti-rabbit secondary antibodies diluted in 1 x PBS, 2% FBS (60 min, RT). Stained coverslips were then washed with 1 x PBS (3X) and mounted on SuperFrost Plus microscope sides (Thermo Scientific) with ProLong Gold Antifade Mountant without DAPI (Invitrogen). Samples were imaged with Perkin Elmer Ultraview confocal spinning disk microscope equipped with Volocity software and a 20X/1.3 NA air objective. Images were analyzed with FIJI (NIH) [52] and figures were prepared using Inkscape (v1.0.1).

Colony forming unit plating

Infected HeLa cells were enriched by cell sorting, where 1000 infected cells were sorted for each sample. The cells were then collected by centrifugation at 500 x g for 5 min, and subsequently lysed in 0.1% Triton X-100 for 5 min at room temperature. The lysed cells were then serially diluted and plated on LB agar plates with appropriate antibiotics.

Dormant S. Typhimurium persistence assay

Infected HeLa cells harboring dormant S. Typhimurium were enriched by cell sorting using the gate Vac^-Cyt^-, and plated on 12-wells plates in DMEM + 10% FBS + Gen¹⁰. The medium was replaced with fresh DMEM + 10% FBS + Gen¹⁰ to avoid the growth of S. Typhimurium being released from dead cells. Cells were harvested at 24 h and 168 h pi for analysis and CFU plating.

Ciprofloxacin survival assay

A final concentration of 10 μg/mL of ciprofloxacin (CIP) were supplemented to the cell culture medium of the infected cells at 3 h pi. The cells were harvested at 6 h pi for cell sorting and CFU plating. CIP was administered at 3 h pi, which offered sufficient time for the infected population to differentiate into Vac^-Cyt^-, Vac⁺Cyt^- and Vac⁺Cyt⁺ for downstream enrichment of Vac^-Cyt^-, Vac⁺Cyt^- populations.

Serial sectioning transmission electron microscopy

At 6 h pi, infected HeLa cells harboring Vac^-Cyt^- S. Typhimurium were harvested by cell sorting. Enriched cells were allowed to adhere on specific dishes (MatTek) pre-coated with 50 mg/ml fibronectin (Sigma) for 3 hours. For EM sample preparation, the adhered cells were fixed with 4% PFA (EMS), 2.5% glutaraldehyde (Sigma-Aldrich) in 0.2 M HEPES for 1 hour at room temperature. Fixed samples were washed with 1 x PBS for three times, and position of interest were defined by fluorescent light microscopy. For further sample preparation, the fixed cells washed three times by the addition of fresh 0.1 M Caco buffer (pH 7.2) and post-fixed in 1% osmium (EMS) in 0.1 M Caco buffer (pH 7.2) enriched with 1.5% potassium ferrocyanide (Sigma-Aldrich) for 1 h. After three washes in 0.1M Caco buffer, samples were incubated in 0.2% of Tannic acid in water for 30 min at room temperature. Samples are post-fixed for a second time in 1% of osmium for 1 h, washed with water and incubated in 2% uranyl acetate dissolved in 25% ethanol for 1 h. Samples were then gradually dehydrated in an ethanol (Sigma-Aldrich) series ranging from 50% to 100%. Samples were embedded in PolyBed812 resin (EMS), followed by polymerization for 48 h at 60°C. The resin embedded samples were removed from the dishes after gentle heating.

For the correlative microscopy, the region of interest was determined thanks to landmarks printed below embedded samples. Embedded cells were sectioned with an ultramicrotome (Leica, UC7) with 70 nm thickness. Thin serial sections were collected on a single slot grid (Agar scientific). Serial sections were observed with a transmission electron microscope TEM Technai T12 (ThermoFisher) at 120kV.

Statistical analysis

Unless further specified in the figure legend, data were analyzed for statistical significance with a Mann-Whitney test using Prism 8.0 (GraphPad). P value of ≤ 0.05 is considered statistically significant. *P < 0.05, **P < 0.01, *** P < 0.001, **** P < 0.0001, ns: not significant/ P ≥ 0.05.