Salmonella enterica serovar Typhimurium SL1344 and derivatives were used in all experiments (Table 1). Bacteria were grown on LB agar supplemented with streptomycin (100 μg/mL). Plasmids were introduced by electroporation and selection on carbenicillin (50 μg/mL). Strain stocks were frozen in 15% glycerol and stored at −80°C. Fresh plates were streaked from glycerol stocks every week and stored at 4°C except for hilA overexpressing strains, which, due to phenotype instability, were not frozen; experiments were done from fresh transformant colonies selected on carbenicillin plates. Overnight cultures were prepared by inoculating one colony into 2 mL LB-Miller broth with selective antibiotics, in a 14-mL polypropylene round-bottom tube (Becton Dickinson) with a loose cap. Cultures were incubated at 37°C in a shaking incubator (225 rpm) for 16–18 h. For sub-culturing (SPI-1 inducing conditions), 0.3 mL of the overnight culture was inoculated into 10 mL LB-Miller broth (no antibiotics unless indicated), in a 125 mL Erlenmeyer flask, and incubated at 37°C with shaking at 225 rpm for 3.5 h or until late log phase. For plate-reader growth assays, a Tecan Infinite 200 Pro plate reader was used. Overnight cultures were diluted 1:25 in fresh LB-Miller and 200 μL were aliquoted in triplicate into 96-well plates and grown with shaking at 37°C.

HeLa (human cervical adenocarcinoma, ATCCL-2) cells were grown at 37°C in 5% CO₂ in complete growth medium: Eagle's minimal essential medium (Mediatech) supplemented with 10% (v/v) heat-inactivated fetal bovine serum (Thermo Fischer Scientific), 2 mM L-glutamine and 1 mM sodium pyruvate. Cells were passaged every 3–4 days and used for experiments within 15 passages of receipt from ATCC.

The ProSeries promoter sequences were previously described (Davis et al., 2011). The transcriptional terminator (synTT) used was Bba_B0015 of the Registry of Standard Biological Parts (parts.igem.org). All plasmid ligation reactions were carried out using T4 DNA ligase (Promega). Restriction enzymes were obtained from New England Biolabs. The high-fidelity polymerase Phusion was used for all PCR reactions (New England Biolabs). PCR primers were sourced from Integrated DNA technologies. All plasmid constructs were verified by sequencing. Plasmids are listed in Table 1. Oligonucleotide sequences are listed in Table 2.

Salmonella were aliquoted into 96-well plates and incubated in the platereader with shaking. OD600, GFP fluorescence (excitation 478 ± 10 nm, emission 515 ± 20 nm) or mCherry fluorescence (excitation 555 ± 10 nm, emission 625 ± 20 nm) were measured every 15 min. For each sample, the promoter activity was calculated as described previously (Kelly et al., 2009). Briefly, the change in fluorescence between two readings during mid log phase of growth (1.5 and 2.5 h) was divided by the average OD600. This measure of promoter activity (per cell synthesis rate) was then normalized to the synthesis rate of the weakest promoter, ProA, resulting in RPU_A-relative promoter units.

Salmonella carrying pCHAR1-ProB.mCherry or pMPMA3ΔPlac-gfp were grown overnight and sub-cultured in flasks as described. Glucose-6-phosphate (G6P) (Sigma) was added to late log phase subcultures (3.5 h), aliquoted in triplicate into a 96-well plate, and incubated with agitation at 37°C in the plate reader. OD600, GFP, and mCherry fluorescence were read as described above every 10 min.

Primary antibodies used were mouse monoclonal anti-GFP (11E5, Molecular Probes), mouse monoclonal anti-mCherry (Clontech), mouse monoclonal anti-DnaK (8E2/2, Enzo Life Sciences), mouse monoclonal anti-HA (16B12, Covance), rabbit anti-panAkt (11E7, Cell Signaling), and rabbit anti-phospho Akt Ser473 (D9E, Cell Signaling). Secondary antibodies used were horseradish peroxidase (HRP)-conjugated goat anti-mouse or goat anti-rabbit (Cell Signaling). For chemiluminescent detection, the SuperSignal West Femto Substrate Kit was used according to the manufacturer's instructions (Thermo Fisher Scientific). Images were captured using a Carestream 4000M Pro Image Station and densitometry analysis was performed using ImageJ Software (W.S. Rasband, NIH, version 1.51).

Bacteria were washed once by centrifugation at 8,000 × g for 2 min, resuspended in HBSS and 10–20 μL bacteria were fixed in 500 μL 2.5% (w/v) paraformaldehyde at room temperature for 15 min, centrifuged and finally washed once in PBS. Bacteria were then stained with 10 μM Syto41 (Life Technologies) in PBS for 30 min at room temperature, washed once with PBS by centrifugation, and resuspended in 1 mL PBS for analysis on a BD LSR II flow cytometer (BD Bioscience). Data were analyzed using FlowJo software (Tree Star). Samples were gated on Syto41⁺ events and the % and mean intensity of GFP⁺ or mCherry⁺ events was measured.

These assays were done as described previously (Finn et al., 2017). Briefly, HeLa cells were seeded 20–24 h prior to infection in 24-well plates at 4.5 × 10⁴ cells per well. SPI1-induced Salmonella were collected by centrifugation at 8,000 × g for 2 min, washed and resuspended in Hank's buffered saline solution (HBSS) and used immediately to infect epithelial cells for 10 min at 37°C at an MOI of ~50. Inoculum CFU counts were checked by plating on LB agar plates. Extracellular bacteria were removed by washing with HBSS and cells were incubated in antibiotic-free complete growth media until 30 min post-infection (pi). Cells were then incubated for 1 h in complete growth media supplemented with L-Histidine (500 μg/mL) and gentamicin (50 μg/mL), followed by complete growth media supplemented with L-Histidine (500 μg/mL) and gentamicin (10 μg/mL) for the remainder of the infection. At indicated time-points, monolayers were lysed in 1 mL of 0.2% (w/v) sodium deoxycholate (DOC) in PBS and viable intracellular bacteria were enumerated by plating on LB agar.

HeLa cells were seeded in 6-well tissue culture plates at a density of 1.5 × 10⁵ cells/well 18–20 h pre-infection. On the day of infection, cells were serum starved for 3 h pre-infection and maintained in serum-free media throughout the assay. Cells were infected as described above. At 60 min pi, cells were washed once in ice-cold HBSS and lysed in ice-cold RIPA buffer (Sigma). Cell lysates were centrifuged at 4°C for 20 min at 16,000 × g to pellet cellular debris and the supernatant was transferred to a clean, pre-chilled tube. Total protein concentration was determined using the DC protein assay (BioRad) to ensure equal loading of samples for SDS-PAGE and immunoblotting.

For experiments in Figure 3C, HeLa cells were plated 20–24 h prior to infection on glass coverslips in 24-well plates (5.5 × 10⁴ cells per well) and infected as described above. At indicated time points pi, cells were fixed in 2.5% paraformaldehyde (w/v) in PBS for 10 min at 37°C. Cells were permeabilized and blocked in 0.1% (w/v) saponin plus 10% (v/v) normal goat serum in PBS (PBS-SS) for 30 min. Primary and secondary antibodies were rabbit anti-Salmonella LPS (1:300; Difco) and AlexaFluor 568-conjugated goat anti-rabbit IgG (1:500; Molecular Probes) diluted in PBS-SS. Coverslips were mounted on glass slides using AntiFade Gold + DAPI (Molecular Probes). Images were captured with the same gain and exposure for each sample on a Photometrics CoolSnap HQ camera using a 60 × /1.4N objective on a Nikon Ti epifluorescence widefield microscope. Post-acquisition analysis of fluorescence intensities was done using the ImageJ software Cell Counter Plugin. Bacteria were selected in the mCherry channel, blind to the GFP content, followed by pixel intensity measurement in the GFP channel. The average background intensity for each field was subtracted from the GFP pixel intensity for each bacterium.

HeLa cells were infected as described above. The differential permeabilization assay was performed as previously described (Finn et al., 2017). Briefly, infected HeLa cells were washed three times with KHM buffer (110 mM potassium acetate, 20 mM HEPES, 2 mM MgCl₂, pH 7.3), and the plasma membrane selectively permeabilized by incubation with 40 μg/mL digitonin (Sigma) in KHM buffer for 1 min at RT, followed by three washes with KHM buffer. Cells were then incubated for 12 min at RT with rabbit anti-Calnexin (Stressgen), to label the cytosolic face of the endoplasmic reticulum in permeabilized cells, and Pacific Blue-conjugated goat anti-Salmonella CSA1 antibodies (KPL), to detect cytosolic bacteria. After paraformaldehyde fixation, rabbit anti-Calnexin antibodies were detected with Alexa Fluor 647-conjugated anti-rabbit antibodies. Coverslips were then washed sequentially with PBS and distilled water, and mounted on glass slides in a MOWIOL solution supplemented with 2.5% (w/v) DABCO. Images were acquired with a Carl Zeiss LSM 710 confocal laser-scanning microscope equipped with a Plan APOCHROMAT 63X/1.4 N.A. objective, processed into maximum intensity projections using Zen 2012 SP1 software and assembled using Adobe Photoshop CC. Pixel area and fluorescence intensities were quantified using CellProfiler (Carpenter et al., 2006) from maximum intensity projections. The pixel area occupied by intracellular Salmonella was determined by mCherry signal. The 488 (GFP) and 455 (DAPI) nm intensities within the area occupied was quantified and the ratio of each fluorescence signal to pixel area was calculated.

HeLa cells grown on 24-well glass bottom plates (Greiner Bio-One) were infected as described above with either WT or ΔSPI2 Salmonella carrying pCHAR2-ProB.mCherry. Cells were imaged using a TiE inverted microscope with Perfect Focus System (Nikon) and custom laser launch (Prairie Technologies). Environmental control was maintained with a stage-top incubation system (Pathology Devices). Beginning at 3 hpi, images were collected on an iXon EMCCD camera (Andor) every 15 min until ~22 hpi using a Plan Fluor 40X/0.75NA objective. GFP intensity plotted vs. time for individual cells was used to determine the doubling time of bacteria in the cytosol. The data was analyzed using an exponential growth, nonlinear regression analysis (least squares fit). Vacuolar fold change was calculated as a ratio of the maximum mCherry intensity per cell vs. intensity at 3 hpi. All post-acquisition image analysis was done using ImageJ software, GraphPad Prism version 7.0a, and Adobe Photoshop (CC v2015.1.2 Adobe).

Statistical significance was determined using one-way ANOVA, followed by Tukey's multiple comparisons test, where indicated. A P-value of ≤ 0.05 was considered significant.

As a first step in developing a rationally designed expression system for use in Salmonella Typhimurium, we selected two unrelated fluorescent proteins, GFPmut3 and mCherry as reporters (Shaner et al., 2004). Both of these are widely used in the Salmonella field, often expressed constitutively under PrpsM in the same plasmid backbone (pFPV25.1 and pFPV.mCherry, respectively; Valdivia and Falkow, 1996; Drecktrah et al., 2008). We began by replacing PrpsM in pFPV25.1 with the synthetic promoters ProA, ProB, ProC, or ProD resulting in the plasmids pCON-ProA.gfp, pCON-ProB.gfp, pCON-ProC.gfp, and pCON-ProD.gfp, respectively. We also made the analogous series of mCherry plasmids pCON-ProA.mCherry, pCON-ProB.mCherry, pCON-ProC.mCherry and pCON-ProD.mCherry. Immunoblotting of lysates of late log phase bacteria harboring the pCON plasmids revealed that the levels of both GFP (Figure 1A) and mCherry (Figure 1B), correlated well with promoter strength, which increases from ProA to ProD (Davis et al., 2011).

Growth curves revealed that constitutive expression of either GFP (Figure 1C) or mCherry (Figure 1D) by any of these plasmids had no detectable effect on cell growth compared to Salmonella containing no plasmid. Under these conditions, log phase growth was observed from 1 to 3 h post-inoculation for all strains with a doubling time of 45–55 min. This is somewhat longer than we have previously reported (Ibarra et al., 2010), presumably due to the use here of a reduced-volume 96-well plate format to facilitate measurement of both fluorescence and OD600 for multiple strains over time. The fluorescence measurements allowed us to quantitatively evaluate promoter activity. The protein synthesis rate for each promoter was calculated by dividing the increase in fluorescence intensity between two time points in log phase (1.5 and 2.5 h) by the average OD600 (Kelly et al., 2009; Davis et al., 2011). To allow for better comparison between the different promoters the synthesis rate was normalized to that of ProA, yielding relative promoter units (RPU_A). In agreement with the published observations in E. coli (Davis et al., 2011), the range of synthetic promoter activity spanned two orders of magnitude when GFP was used as a reporter (Figure 1E). However, we were unable to distinguish a difference in promoter activity using mCherry as a reporter (Figure 1F). The reason for this is apparent when fluorescence intensity is plotted against OD600, which reveals that, unlike GFP (Figure 1E, Inset), mCherry fluorescence did not increase with bacterial growth during log phase (Figure 1F, Inset).

The second step in developing these optimized expression vectors was the addition of a synthetic transcriptional terminator (synTT) to each of the plasmids, resulting in the pCON1 series of plasmids: pCON1-ProA.gfp, pCON1-ProB.gfp, pCON1-ProC.gfp, pCON1-ProD.gfp, and pFPV25.1-TT along the analogous mCherry plasmids. The addition of synTT resulted in a three- to four-fold increase in fluorescence for bacteria bearing the GFP plasmids (Figure 2A) and a 2- to 2.5-fold increase for those bearing the mCherry plasmids (Figure 2B), at 2.5 h post inoculation. Although addition of the terminator resulted in higher protein production, promoter activity (RPU_A) was not increased (Figures 2C,D).

The above experiments show on a population level, how the synthetic promoters and terminator affect protein production, however, they provide no information on cell-to-cell variability. Therefore, to evaluate heterogeneity at the single cell level we used flow cytometry to compare the fluorescence intensity of individual bacteria over time. We selected the pCON1-ProC.gfp and pCON1-ProC.mCherry plasmids for single cell analysis, since the ProC promoter resulted in GFP fluorescence levels that were readily detectable by fluorescence microscopy without negatively impacting invasion (Figure 3A). Bacteria were harvested and fixed at 2, 3.5, 4, and 5 h post-inoculation, which under these growth conditions correspond to log phase, late-log phase, early stationary phase and stationary phase, respectively. The mean fluorescence intensity of Salmonella harboring pCON1-ProC.gfp was similar at all four time points (range from 36,601 ± 233 to 46,013 ± 2465, mean ± SD), whereas bacteria harboring pFPV25.1 were significantly brighter at 2 h (96,994 ± 8,395) than later time points (Figures 2E,F). In contrast, bacteria harboring the mCherry constructs, pFPV.mCherry or pCON1-ProC.mCherry fluctuated between each of the time points (Figures 2G,H). Thus, by both population based and single cell based analysis, the synthetic gfp expression construct produced the most consistent, predictable fluorescent protein production.

Given the wide range of GFP production levels obtained using synthetic elements, we next assessed which gfp constructs are compatible for infection studies using the well-established HeLa cell invasion model. As previously shown, strains harboring pFPV25.1 had a significant invasion defect (Knodler et al., 2005) and this was exaggerated by the addition of the synTT (Figure 3A), which increased GFP production (Figure 2A). For synthetic promoter driven GFP production, only the strain harboring pCON1-ProD.gfp, which together with pFPV25.1-TT produces the largest amounts of GFP (Figure 2A), had an invasion defect. All others, despite profound differences in GFP production spanning over two orders of magnitude, had no detectable effect on invasion (Figure 3A). Furthermore, we observed no impact on intracellular replication in HeLa cells (Figure 3B).

Altogether, the preceding experiments showed that Salmonella bearing the pCON expression vectors produced GFP over a wide spectrum of levels, without any effect on growth or invasion. Since our goal was to use these plasmids to study intracellular Salmonella, we also evaluated intracellular bacteria by fluorescence microscopy (Figure 3C and Table 3). The mean fluorescence intensity (MFI) of Salmonella producing GFP under the control of the weakest synthetic promoter, ProA, was slightly higher than that of bacteria bearing no plasmid (78 ± 25 vs. 40 ± 1, mean ± SD). Thereafter, the MFI increased with promoter strength (ProA < ProB < ProC < ProD) and was further increased, approximately three-fold, for each promoter in the presence of the synTT. The highest MFI (3,706 ± 1,750), was observed in bacteria bearing pCON1-ProD.gfp. Using a threshold of “3X the MFI of bacteria bearing no plasmid,” >90% of Salmonella bearing pCON1-ProB.gfp, pCON-ProC.gfp, pCON1-ProC.gfp, pCON-ProD.gfp, and pCON1-ProD.gfp plasmids were GFP positive (Table 3). Thus, GFP intensity correlated well with synthetic promoter strength and was further increased in the presence of the synthetic transcriptional terminator.

Thus far we have shown that the pCON plasmid series enables tunable, predictable and homogeneous production of GFP. Next, we wanted to take advantage of the pCON-based tunable expression system as a way to study Salmonella effector protein function. We selected SopB for these experiments, because, upon delivery into the host cell by T3SS1, the activity of this effector protein can be readily quantified by phospho-specific immunoblotting for phosphorylated AKT (Steele-Mortimer et al., 2000). To express SopB together with its cognate chaperone, SigE, we made constructs with the sopB-sigE operon under the control of each of the ProSeries promoters, and including a SopB C-terminal 2xHA tag, yielding pCON-ProA.sopB, pCON-ProB.sopB, pCON-ProC.sopB, and pCON-ProD.sopB. These plasmids were introduced into a SopB-SigE deletion mutant (ΔsopB; Knodler et al., 2006). To evaluate the levels of SopB-2xHA protein, lysates were prepared from SPI1-induced bacteria. Under these conditions SopB expressed under the control of its native promoter from a low copy number plasmid (pWSKDE-2xHA or PsopB.sopB) functionally complements ΔsopB (Knodler et al., 2009). Immunoblotting revealed that, production of HA-tagged SopB under the control of the ProSeries promoters was consistent with promoter strength, with ProA.sopB producing the lowest levels and ProD.sopB producing the highest levels (Figure 4A). Analysis of SopB levels produced by the pCON plasmids revealed that at late log phase (as used for invasion), the amounts of SopB produced by ProB.sopB were most similar to the amount produced by the native promoter (PsopB.sopB).

We next compared the ability of constitutively vs. natively expressed SopB to functionally complement the ΔsopB mutant. In HeLa cells, SopB-dependent Akt phosphorylation peaks at ~1 hpi (Steele-Mortimer et al., 2000; Cooper et al., 2011). When the ΔsopB mutant is complemented with plasmid borne SopB expressed under its native promoter (PsopB.sopB), Akt phosphorylation is restored (Figure 4B). Notably, ProB driven production of SopB, while producing similar levels of the effector to the native promoter, did not restore Akt phosphorylation (Compare Figures 4A,B). However, constitutively expressed SopB produced by either ProC or ProD restored the ability to induce Akt phosphorylation. Thus, the variable strength ProSeries of constitutive promoters can facilitate the identification of optimal constitutive expression levels needed to complement SopB function. Further fine-tuning, if necessary could be achieved by the incorporation of the synthetic TT as demonstrated for GFP and mCherry.

Constitutive expression of transcriptional regulators, such as those controlling Salmonella virulence genes, can be used to identify pathways affected by these regulators as well as to interrogate their roles in transcriptional cross-talk. HilA is a key regulator of the SPI1-regulon, and ultimately controls expression of genes encoding the T3SS1 secretion apparatus and secreted effectors (Bajaj et al., 1995, 1996). The level of SPI1 gene expression is dependent on the level of HilA, and high levels of expression of hilA have a fitness cost to the bacteria through increased production of T3SS1 (Sturm et al., 2011). Thus, this regulator is a good candidate for optimization of expression with a tunable promoter pCON plasmid series. Since we expected small changes in expression levels of hilA to have significant impacts on fitness, we introduced an additional level of tuning by using two plasmid backbones, one medium copy (ColE1 ori, ≈ 15–20 copies/cell) and one low copy (pSC101 ori, ≈ 5 copies/cell). The resultant plasmids were: pCON-(ProA thru ProD).hilA (medium copy number) and pCON2-(ProA thru ProD).hilA (low copy number). The plasmids were introduced into a ΔhilA background, to allow analysis of the impact of constitutive hilA expression. Immediately after transformation, we observed a growth defect in all transformants except those producing HilA under the control of ProA (Table 4). Expression of hilA under the control of the strongest promoter, ProD, resulted in punctiform colonies that grew poorly in liquid cultures. Due to these profound growth defects, the ProD constructs were excluded from further analyses. As shown in Figure 5A, a ΔhilA mutant replicates faster than WT Salmonella in broth cultures due to the lack of T3SS1 production (Sturm et al., 2011). Constitutive expression of hilA in the ΔhilA strain under the control of the weakest promoter (ProA) recapitulated the WT growth rate when the low copy number backbone was used. When constitutive expression of hilA was increased, either by using the stronger promoter ProB or by increasing the plasmid copy number, growth retardation was exacerbated. However, when hilA expression level was further increased using the ProC promoter the growth phenotype was inconsistent.

Overexpression of hilA under the control of its native promoter causes hyper-invasiveness into host cells (Bajaj et al., 1995; Sturm et al., 2011). To examine whether constitutive expression of hilA can recapitulate this effect, HeLa cells were infected with strains harboring the hilA plasmid series, and invasion was measured using the gentamicin protection assay. Immunoblotting of bacterial lysates, prepared from bacteria grown under SPI1-inducing conditions (as used for invasion), revealed that HilA protein levels reflected a combination of both promoter strength and plasmid copy number (Figure 5B). Although, it should be noted that, expression under the control of the strongest promoter, ProC, resulted in variable levels of HilA. In line with published work, our invasion assay confirmed that the ΔhilA strain is non-invasive (Figure 5C). Complementation of this strain with constitutive expression of hilA from the plasmids pCON2-ProB.hilA, pCON-ProB.hilA or pCON2-ProC.hilA lead to increased invasion by up to ≈ 5-fold compared to the WT strain. Further increase in HilA production above this level, by pCON-ProC.hilA, resulted in an invasion defect compared to WT, thus revealing a threshold of hilA expression above which invasion is compromised (Figure 5C). Thus, by combining the ProSeries of promoters with different copy number plasmids, we were able to fine-tune expression levels of hilA to find the optimum conditions for constitutive expression.

Finally, we wished to apply the synthetic elements approach to developing a bidirectional environmental sensor for use in Salmonella. Previously we have described a fluorescent reporter, PuhpT-gfp, that is specifically induced in the cytosolic subpopulation of Salmonella in epithelial cells (Finn et al., 2017). This GFP based cytosolic hexose phosphate activated reporter (CHAR) is induced by the presence of glucose-6-phosphate (G6P), which is found in the cytosol but not the SCV (Finn et al., 2017). While this single-color reporter enables detection of the cytosolic population, visualization of the total bacterial population requires a constitutively produced second fluorophore, which could have a negative impact on bacterial fitness.

To address the need for a two-color cytosolic reporter we designed two bidirectional vectors with constitutive mCherry production under the control of the synthetic ProB promoter and GFP production under the control of PuhpT (Figure 6A). SynTT was included at the 3′ end of both fluorescent genes. To add another level of tunability to this system we used both full-length and truncated (lacking 14 bp at the 5′ end) versions of the uhpT promoter resulting in pCHAR2-ProB-mCherry and pCHAR1-ProB.mCherry respectively. In broth grown cultures, GFP fluorescence in bacteria harboring either of these reporter plasmids responded in a dose dependent manner to exogenous G6P, while mCherry fluorescence was unchanged (Figures 6B,C). For a “no fluorescence control” we included bacteria bearing the promoterless pMPMA3ΔPlac-gfp (Pnull.gfp). Comparison of the maximal fluorescence intensity of GFP following induction with G6P revealed that bacteria harboring pCHAR2-ProB.mCherry were ~2.5-fold brighter than those harboring pCHAR1-ProB.mCherry (1,827 ± 44 vs. 689 ± 112), confirming that the full length uhpT promoter is stronger.

Having established that the pCHAR-ProB.mCherry reporters respond appropriately to G6P, we next assessed the functionality and fidelity of these reporters in intracellular bacteria. For this we used the HeLa cell infection model within which the cytosolic and vacuolar populations of Salmonella have been characterized. Using the gentamicin protection assay we found that neither invasion nor net intracellular replication were affected in Salmonella harboring pCHAR1-ProB.mCherry (Figure 7A) or pCHAR2-ProB.mCherry (Figure 7B). Furthermore, by immunofluorescence microscopy we showed that pCHAR1-ProB.mCherry had no effect on the percentage of infected cells containing >50 bacteria per cell at 6 hpi, indicating that the plasmid has no effect on the ability of Salmonella to initiate hyper-replication in the cytosol (Figure 7C).

Given that pCHAR1-ProB.mCherry contains the truncated PuhpT and therefore produces less GFP under inducing conditions, one concern was that it might not accurately report cytosolic bacteria. We therefore utilized fluorescence microscopy to evaluate GFP fluorescence of intracellular bacteria in cells infected with bacteria bearing pCHAR1-ProB.mCherry at 6 hpi (Figures 7D–F). For mCherry and GFP controls, we included pCHAR1 (which lacks the ProB.mCherry-TT cassette) and the promoterless GFP control pMPMA3ΔPlac-gfp (Pnull.gfp). Rather than measure the fluorescence of individual bacteria, which are difficult to accurately resolve in cells containing high numbers, we used a region of interest (ROI) approach to select populations of cytosolic bacteria (C = >50 bacteria/cell) or vacuolar bacteria (V = ≤ 20 bacteria/cell). For bacteria bearing pCHAR1-ProB.mCherry, mCherry fluorescence was used to define the ROI, whereas for bacteria bearing the single reporter pCHAR1 or Pnull.gfp control plasmid, immunolabeling with an anti-CSA (common structural antigen of Salmonella LPS) antibody was used. GFP fluorescence (intensity/ROI pixel area) was high in cytosolic, but not vacuolar populations of bacteria bearing either of the CHAR reporters (Figures 7D,F). No significant difference was observed in fluorescence intensity between GFP-positive bacteria bearing pCHAR1-ProB.mCherry or pCHAR1 (0.61 ± 0.1 and 0.53 ± 0.14, mean ± SD, respectively). GFP fluorescence in vacuolar bacteria bearing pCHAR plasmids was no higher than that of bacteria harboring Pnull.gfp. To further verify the fidelity of the cytosolic GFP reporters, i.e., that GFP is only produced in cytosolic bacteria, we used digitonin permeabilization assay to specifically immunostain cytosolic bacteria (Knodler et al., 2010; Finn et al., 2017; Figure 7E). Together, these results conclusively show the bidirectional sensor, pCHAR1-ProB.mCherry, accurately identifies cytosolic bacteria without affecting the ability of Salmonella to establish either intracellular niche.

One interest in our lab is the heterogeneity of intracellular Salmonella, for example the different growth rates in vacuolar vs. cytosolic populations (Malik-Kale et al., 2012). Previously, we performed live cell imaging using Salmonella constitutively producing either mCherry or GFP together with a vacuolar content marker (fluorescent dextran), which was internalized by fluid phase uptake into the endocytic pathway prior to internalization of bacteria (Malik-Kale et al., 2012). While this approach did demonstrate that cytosolic bacteria (dextran –ve) replicate faster than those in the SCV (dextran +ve), lack of dextran signal around a bacterium does not definitively prove that they are cytosolic. An advantage of the pCHAR-ProB.mCherry constructs, or similar dual color reporters, is that the bacteria themselves report their intracellular localization; in this case, cytosolic (red and green) vs. vacuolar (red only). Since, for live cell imaging, it is generally advantageous to have a stronger fluorescent signal we selected for these experiments pCHAR2-ProB.mCherry, which contains the full length PuhpT. Infected HeLa cells were imaged using a spinning disc confocal system, as previously described (Malik-Kale et al., 2012). WT Salmonella were compared to a SPI2 deletion mutant (ΔSPI2), which has a replication defect in the vacuolar compartment but not in the cytosol (Malik-Kale et al., 2012). As shown in Figure 8A, the total GFP intensity per infected HeLa cell started increasing between 180 and 240 min pi, which is when cytosolic replication is initiated. Since the fluorescence intensity of individual bacteria did not change during this period (see inset in Figure 8A), the increase in fluorescence was due to replication of bacteria. Doubling times of bacteria in the cytosol were calculated by measuring the total integrated GFP fluorescence intensity of each infected cell over time (Figure 8B). No significant difference was detected between the two strains (doubling rate 35 min ± 5.2 for WT vs. 35 min ± 6.9 for ΔSPI2). A slightly different approach was used for vacuolar bacteria (red only), which replicate more slowly. The maximal fold increase of this population was calculated per cell by dividing the maximal intensity (X hpi) by the intensity at the starting point of imaging (3 hpi), yielding the fold increase. This showed that, as expected, in the vacuole the SPI2 mutant replicated less than the WT, 3 ± 1.8 vs. 10 ± 8 (mean ± SD), respectively (Figure 8C). These results validate that pCHAR2-ProB.mCherry is well suited for live cell imaging of Salmonella.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.