Salmonella Typhimurium manipulates macrophage cholesterol homeostasis through the SseJ-mediated suppression of the host cholesterol transport protein ABCA1

Abstract

Upon infection of host cells, Salmonella enterica serovar Typhimurium resides in a modified endosomal compartment referred to as the Salmonella-containing vacuole (SCV). SCV biogenesis is driven by multiple effector proteins translocated through two type III secretion systems (T3SS-1 and T3SS-2). While many host proteins targeted by these effector proteins have been characterized, the role of host lipids in SCV dynamics remains poorly understood. Previous studies have shown that S. Typhimurium infection in macrophages leads to accumulation of intracellular cholesterol, some of which concentrates in and around SCVs; however, the underlying mechanisms remain unknown. Here, we show that S. Typhimurium utilizes the T3SS-2 effector SseJ to downregulate expression of the host cholesterol transporter ABCA1 in macrophages, leading to a ~45% increase in cellular cholesterol. Mechanistically, SseJ activates a signaling cascade involving the host kinases FAK and Akt to suppress Abca1 expression. Mutational inactivation of SseJ acyltransferase activity, silencing FAK, or inhibiting Akt prevents Abca1 downregulation and the corresponding accumulation of cholesterol during infection. Importantly, RNAi-mediated silencing of ABCA1 rescued bacterial survival in FAK-deficient macrophages, suggesting that Abca1 downregulation and cholesterol accumulation are important for intracellular survival.

Introduction

Salmonella enterica serovar Typhimurium is a facultative intracellular pathogen that colonizes the intestinal epithelium and circulating phagocytes, causing disease ranging from mild gastroenteritis to severe septicemia. Upon internalization by host cells, Salmonella resides and proliferates within membrane-bound inclusions called Salmonella-containing vacuoles (SCVs). Pathogenic Salmonella strains express two type III secretion systems (T3SS-1 and T3SS-2), each of which translocate distinct arrays of effector proteins into the host cytosol. Although there is some functional overlap between T3SS-1 and T3SS-2, T3SS-1 effectors generally facilitate bacterial penetration of the intestinal epithelium, while T3SS-2 effectors promote the biogenesis of and bacterial survival within the SCV [1, 2].

Many host proteins targeted by these secreted effectors have been intensively studied, and in many cases their respective roles in invasion and/or intracellular survival are well characterized. In contrast, the role of host lipids and their manipulation by secreted effectors is less understood. Early studies reported that infection of both epithelial cells and macrophages by S. Typhimurium results in a significant increase in intracellular cholesterol and its precursors, much of which accumulates in and around SCVs [3]. The observed increase in cholesterol content requires a functional T3SS-2, indicating that it is actively induced by the pathogen, although specific effectors that may drive cholesterol accumulation have not been identified. Interestingly, although cholesterol accumulation occurs independently of de novo synthesis, later studies reported that one or more cholesterol precursors are necessary for optimal Salmonella survival [4]. Importantly, cholesterol has been reported on both the limiting membrane of SCVs and on Salmonella-induced filaments (SIFs) extending from them [5], suggesting an important role for cholesterol in SCV dynamics and intracellular survival.

While these early studies did not detail the mechanisms through which Salmonella perturbs cholesterol homeostasis, it is important to note that two T3SS-2 effectors, SseJ and SseL, have reported roles in cholesterol metabolism and transport. SseJ is a RhoA-dependent cholesterol acyltransferase containing both deacylase and glycerophospholipid:cholesterol acyltransferase (GCAT) activity. Key to SseJ activity are three catalytic residues, Ser141, Asp247, and His384N. Importantly, mutating any one of these three catalytic activities abolishes SseJ activity and attenuates Salmonella virulence in mice [6]. SseJ localizes to the cytosolic surface of the SCV where it catalyzes the transfer of acyl chains from phospholipids onto free cholesterol, generating cholesterol esters [6–9]. This function is homologous to eukaryotic ACAT, which esterifies cholesterol to generate cholesteryl esters within the endoplasmic reticulum (ER) and promotes the storage of excess cholesteryl esters in lipid droplets [10]. Additionally, both SseJ and SseL have been shown to interact with the host cholesterol transport protein oxysterol binding protein 1 (OSBP1) and recruit it to SCVs [11, 12]. OSBP1 is known to mediate the non-vesicular transfer of cholesterol from the ER to the Trans-Golgi Network (TGN) and endosomes, suggesting that SseJ and SseL play a role in redirecting cholesterol transport from the ER to the SCV. In agreement with this hypothesis, infection with an SseL-deficient strain of S. Typhimurium leads to an accumulation of large lipid droplets within the host cell [11, 13]. Importantly, bacterial strains deficient in either SseJ or SseL are attenuated in virulence in both macrophage and animal models [14, 15], demonstrating that modulation of host cell lipid homeostasis plays an important role during Salmonella pathogenesis.

Cholesterol is a critical component of eukaryotic membranes, whereas prokaryotic membranes are devoid of cholesterol. The distribution of lipids within eukaryotic membranes is heterogenous; plasma membrane (PM) and late endosomal lipid bilayers are enriched with cholesterol, while the ER contains low amounts of cholesterol at steady state [16]. Due to the importance of cholesterol in maintaining membrane stability and function, cellular cholesterol concentrations are normally under tight regulatory control.

Cholesterol homeostasis is maintained by balancing uptake (via low-density lipoprotein; LDL) and biosynthesis in the ER with the storage and export of excess cholesterol [17, 18]. Excess cholesterol is handled in two ways; it can be esterified and stored within ER-derived lipid droplets and/or exported out of the cell by a family of ATP-binding cassette (ABC) transporters, predominantly ABCA1 and ABCG1 [10, 19]. These proteins shuttle intracellular pools of cholesterol to the PM where it is transferred to apolipoprotein A-1 (via ABCA1), generating nascent high-density lipoprotein (HDL), or onto mature HDL (via ABCG1) [20–22]. It is becoming increasingly evident that many pathogenic microorganisms, including Salmonella Typhi and Salmonella Typhimurium, target cholesterol-enriched lipid domains as well as intracellular free cholesterol in order to facilitate microbial entry, intracellular localization, and other pathogenic functions [23–29].

Focal adhesion kinase (FAK) is a non-receptor tyrosine kinase typically associated with integrin-mediated signaling. We recently reported the surprising observation that S. Typhimurium recruits FAK to the cytoplasmic face of SCVs in infected macrophages in a T3SS-2-dependent manner. FAK activation at this site stimulates the Akt/mTOR signaling axis, thereby suppressing the autophagic capture and killing of intracellular Salmonella [30]. Remarkably, FAK and its downstream effector Akt have been shown to control cholesterol levels in fibroblasts by inversely regulating the expression of ABCA1 as a function of cell density [31]. In that study, the absence of FAK led to elevated Abca1 expression independent of cell density and correlated with a decrease in cellular cholesterol and sterol esters, demonstrating a novel role for FAK in the regulation of cholesterol homeostasis.

Here, we show that activation of the FAK/Akt signaling axis by intracellular S. Typhimurium induces cholesterol accumulation in macrophages by suppressing Abca1 expression. Bacterial survival is significantly attenuated in FAK-deficient macrophages as well as in wild-type (WT) macrophages treated with the pharmacological LXR agonist, T0901317, to induce Abca1 expression. Importantly, survival within FAK-depleted macrophages can be restored to wild-type levels by siRNA-mediated depletion of ABCA1. Mechanistically, we found that FAK/Akt activation is mediated by the T3SS-2 effector SseJ. In contrast to WT S. Typhimurium, an SseJ-deficient strain fails to activate FAK or Akt, does not downregulate Abca1 expression, and fails to accumulate cholesterol, resulting in reduced survival. Importantly, the catalytic activity of SseJ is necessary for cholesterol accumulation, as a loss-of-function mutation of the catalytic residue His384 failed to activate FAK or increase intracellular cholesterol. Together, these results define a novel mechanism through which Salmonella alters intracellular cholesterol homeostasis in a FAK- and ABCA1-dependent manner to promote survival in macrophages.

Results

Salmonella infection promotes cholesterol accumulation in primary mouse macrophages

Previously, S. Typhimurium had been shown to induce accumulation of cholesterol and its precursors in cultured immortalized macrophages through an unknown mechanism [3]. Here, we first confirmed that cholesterol accumulation occurs in primary bone marrow-derived macrophages (BMDMs) harvested from wild-type C57BL/6 mice. C57Bl/6 mice were chosen because 1) they are susceptible to Salmonella infection due to the lack of functional Nramp1 [32] and are widely used for this purpose and 2) to maintain consistency with our earlier work in which we generated mice conditionally lacking FAK in the myeloid lineage on the C57BL/6 background [30]. Since invasive Salmonella rapidly kills macrophages in a T3SS-1-dependent manner [33], we used a mutant of S. Typhimurium strain SL1344 that lacks a functional T3SS-1 (ΔinvG) [34]. As T3SS-1 and associated effectors are dispensable for Salmonella dissemination from the gastrointestinal tract and subsequent infection of phagocytic cells [35], this ΔinvG strain provides an established model of infection for macrophages which avoids potential T3SS-1-induced cytotoxicity.

To visualize cholesterol abundance and distribution in primary macrophages, BMDMs were incubated with BODIPY-cholesterol, a stable fluorescent cholesterol analog that is widely used to study the trafficking of cholesterol in live cells [36]. For this purpose, cells were infected with RFP-expressing ΔinvG Salmonella for a total of 18 hours prior to live-cell imaging. 90 minutes prior to imaging, cells were incubated for 45 minutes in serum-free culture media containing 0.5 μM BODIPY-cholesterol, followed by a 45 minute chase in the absence of labeled cholesterol. Images were then acquired of uninfected and infected cells in the same culture. As shown in Fig. 1A–B, infected BMDMs exhibited markedly increased BODIPY-cholesterol fluorescence compared to uninfected cells. Interestingly, in addition to BODIPY-cholesterol being found associated with SCVs, it was also highly enriched in other intracellular compartments, presumably elements of the endocytic pathway. Quantification of the mean fluorescence intensity (MFI) of BODIPY-cholesterol in individual cells revealed a 2.6-fold increase in MFI in infected BMDMs compared to uninfected cells (Fig. 1C), in agreement with previous findings in immortalized cells [3, 5].

Figure 1 -. Salmonella Typhimurium induces cholesterol accumulation in murine BMDMs.

Unless otherwise stated, n refers to the number of individual experimental replicates. (A-B) Bone marrow-derived macrophages (BMDMs) from WT mice were infected with RFP-expressing ΔinvG S. Typhimurium (red) at an MOI of 50. At 16.5 hours post-infection, cells were incubated with 0.5 μM BODIPY-cholesterol (pseudo-colored black) for 45 minutes and imaged after a 45 minute chase. Uninfected cells (A) were imaged from the same dish as infected cells (B). Cells are outlined by a dashed line. Scale bar represents 5 μm. (C) Mean fluorescence intensity (MFI) of BODIPY-cholesterol was measured from uninfected (filled circles) or infected BMDMs (open circles). Each data point represents a single cell from a representative experimental replicate, mean ± SD. Uninf. n=20 cells; Infected n=35 cells. (D) BMDMs were infected (open bars) at an MOI of 100 for 18 hours before endogenous cholesterol was extracted and quantified. Values are normalized to uninfected levels (filled bars), mean ± SD. n=3 experimental replicates. ***p<0.0005.

Next, we tested whether Salmonella alters the levels of endogenous cholesterol in BMDMs. Cells were infected or mock-infected for 18 hours before total cellular lipids were harvested using the Bligh and Dyer method [37]. Free cholesterol was then measured using a fluorometric cholesterol assay as previously described [38]. In agreement with the findings above, S. Typhimurium infection increased the content of free cholesterol by approximately 45% relative to uninfected BMDMs (Fig. 1D). As not all cells in the population become infected, cholesterol accumulation in infected cells likely exceeds 45% on a per-cell basis. Both the visual and biochemical cholesterol quantification results were recapitulated using immortalized BMDMs (iBMDMs) (Fig. S1A–E) [39]. Together, these results demonstrate that Salmonella Typhimurium induces the accumulation of cholesterol during the later stages of infection in both primary and immortalized bone marrow-derived macrophages.

Activation of FAK/Akt signaling downregulates expression of the cholesterol exporter Abca1

We recently demonstrated that the non-receptor tyrosine kinase FAK is recruited to the SCV by S. Typhimurium in a T3SS-2-dependent manner, and that this leads to activation of the serine/threonine kinase Akt on the surface of the vacuole [30]. Remarkably, FAK/Akt signaling has been shown to stimulate cholesterol accumulation in fibroblasts by downregulating transcription of the cholesterol export protein, ABCA1 [31]. As ABCA1 is the predominant cholesterol exporter in macrophages, we hypothesized that Salmonella-induced FAK/Akt signaling may similarly downregulate ABCA1 expression, leading to the cholesterol accumulation seen during macrophage infection. To test this hypothesis, iBMDMs were depleted of endogenous FAK using siFAK oligonucleotides or mock-depleted with non-coding siRNA, then infected with S. Typhimurium for either 5h or 24h. Cells were then lysed and immunoblotted to detect total and phosphorylated (active) FAK (pY397) or Akt (pS473) at each time point. In agreement with our previous findings, infection of untreated control macrophages resulted in robust phosphorylation of both FAK and Akt at 5h and 24h post-infection (Fig. 2A–C). As expected, depletion of FAK reduced overall FAK expression and substantially inhibited the downstream phosphorylation of Akt (Fig. S2A).

Figure 2 -. Salmonella suppresses Abca1 expression in a FAK- and Akt-dependent manner.

(A-C) Immortalized BMDMs (iBMDMs) were infected with ΔinvG S. Typhimurium (S.t.) at an MOI of 100 for 0h, 5h, or 24 hours. (A) Representative western blot of cell lysates immunoblotted to detect phosphor-FAK^Y397, total FAK, phospho-Akt^S473, and total Akt. Numbers indicate ratio of pFAK:total FAK and pAkt:total Akt at each time point. (B, C) Quantification of pFAK:total FAK (B) and pAkt:total Akt (C) normalized to uninfected control lysates averaged across 4 experimental replicates, mean ± SD. n=4. (D, E) iBMDMs were transfected with non-targeting siRNA (siControl, filled bars) or siFAK oligonucleotides (open bars) 48 hours prior to infection with ΔinvG (S.t.) at an MOI of 100 for 5 or 24 hours. Abca1 (D) or Abcg1 (E) mRNA was measured by qPCR. Values are normalized to uninfected levels (dotted line), mean ± SD. (D) n=4; (E) n=5. (F, G) iBMDMs were infected as described above for 5h in the presence (open bars) or absence (filled bars) of the Akt inhibitor triciribine (10 mM; +Akt-i). Cells were infected for 5h and Abca1 (F) or Abcg1 (G) mRNA was measured by qPCR. Values are normalized to uninfected levels (dotted line), mean ± SD. (F) n=5; (G) n=3. *p<0.05, **p<0.005, ns=not significant.

Next, we sought to assess whether S. Typhimurium regulates Abca1 expression during infection of control or FAK-depleted iBMDMs by measuring mRNA levels using qPCR. As shown in Fig. 2D, infection of control iBMDMs led to a 50% reduction in Abca1 mRNA at 5h and 72% at 18h post-infection, relative to uninfected cells (dashed line). In contrast, infection of FAK-depleted cells failed to reduce Abca1 mRNA below uninfected levels, and instead transiently increased Abca1 expression (Fig. 2D). These findings suggest that Salmonella may interfere with cholesterol efflux through the FAK-dependent downregulation of Abca1. To determine whether FAK-dependent downregulation was specific to ABCA1, we measured the expression of another ABC-family cholesterol transport protein, ABCG1. As shown in Fig. 2E, Abcg1 mRNA levels were suppressed during infection of both control and FAK-depleted cells, suggesting that S. Typhimurium also downregulates expression of Abcg1, but independently of FAK. To determine whether the suppression of Abca1 was specific to Salmonella invasion, iBMDMs were infected with non-pathogenic Escherichia coli (DH5ɑ) and mRNA expression was measured by qPCR as above. Remarkably, DH5ɑ failed to downregulate Abca1, instead inducing a significant increase in expression (Fig. S2B), indicating that the suppression of Abca1 is actively mediated by Salmonella and is not an innate host response to Gram-negative bacteria.

Since Akt signaling downstream of FAK is important for Abca1 suppression in fibroblasts, we sought to determine whether Akt signaling is also necessary for the downregulation of Abca1 during Salmonella infection. iBMDMs were treated with the pharmacological Akt inhibitor triciribine for 1 hour prior to and throughout a 5 hour infection with S. Typhimurium. As anticipated, triciribine did not affect FAK^Y397 phosphorylation or levels of total Akt during infection, but severely attenuated Akt^S473 phosphorylation (Fig. S2C). Similar to FAK depletion, Akt inhibition not only prevented Salmonella-mediated downregulation of Abca1, but actually increased mRNA levels by approximately 40% relative to uninfected controls, while Abcg1 expression was downregulated independently of Akt (Fig. 2F–G). Together, these results demonstrate that Salmonella specifically downregulates the expression of Abca1 in a FAK- and Akt-dependent manner.

Failure to downregulate Abca1 prevents cholesterol accumulation in infected macrophages

ABCA1 plays a major role in maintaining lipid homeostasis in macrophages and other cell types by facilitating cholesterol efflux [19, 40]. Since the findings above reveal that FAK/Akt signaling is required for Salmonella-induced downregulation of Abca1 expression, we reasoned that cholesterol retention would be reduced during infection of FAK-deficient macrophages. To test this hypothesis, we used BMDMs derived from mice conditionally lacking FAK in myeloid lineage cells, which we previously showed completely lack endogenous FAK [30]. BMDMs derived from these mice (FAK^−/− BMDMs) or their WT littermates were infected with RFP-expressing S. Typhimurium and incubated with BODIPY-cholesterol as described in Fig. 1. As observed above, infection of WT BMDMs resulted in a dramatic increase in BODIPY-cholesterol content by 18 hours post-infection relative to uninfected controls, with much of the fluorescence enriched in intracellular compartments (Fig. 3A–B). In contrast, BODIPY-cholesterol fluorescence remained similar between infected and uninfected FAK^−/− BMDMs (Fig. 3C–D). These morphological observations are reflected by the MFI analysis, where infection caused a 4.8-fold increase in BODIPY-cholesterol fluorescence in infected WT BMDMs relative to uninfected cells and no significant difference in FAK-deficient macrophages (Fig. 3E). In agreement with these fluorescence data, the fluorometric cholesterol assay indicated that infected WT BMDMs exhibited a ~45% increase in endogenous free cholesterol, while no change was observed in FAK^−/− BMDMs relative to uninfected cells (Fig. 3F).

Figure 3 -. Failure to downregulate Abca1 prevents cholesterol accumulation in infected macrophages.

(A-D) WT BMDMs (A-B) or FAK^−/− BMDMs (C-D) were uninfected (A, C) or infected with RFP-expressing ΔinvG S. Typhimurium (B, D; red) and incubated with BODIPY-cholesterol (pseudo-colored black) as described in Fig. 1. (E) MFI of BODIPY-cholesterol was measured from uninfected (filled) and infected (open) WT (circles) and FAK^−/− BMDMs (squares) as previously described. WT Uninf. n=20 cells; WT Infected n=27 cells; FAK^−/− Uninf. n=25 cells; FAK^−/− Infected n=16 cells. (F) WT and FAK^−/− BMDMs were infected (open bars) at an MOI of 100 for 18 hours before endogenous cholesterol was extracted and quantified. Values are normalized to uninfected levels (filled bars), mean ± SD. n=3. (G) iBMDMs were either untreated (filled bars) or pretreated with 10 μM T0901317 (open bars) for 24 hours prior to and maintained throughout infection. Cells were then infected at an MOI of 100 for 5 or 24 hours. Abca1 expression was measured via qPCR as previously described. n=7. (H-K) Control (H-I) or T0901317-pretreated BMDMs (J-K; +T090) were uninfected (H, J) or infected with RFP-expressing ΔinvG (I, K) and incubated with BODIPY-cholesterol as described above. (L) MFI of BODIPY-cholesterol was measured from uninfected (filled) or infected BMDMs (open) in the presence (+T090; squares) or absence (-Tx; circles) of T0901317 treatment as described above. -Tx Uninf. n=27 cells; -Tx Infected n=21 cells; +T090 Uninf. n=25 cells; +T090 Infected n=25 cells. (M) Control (-Tx) and T0901317-treated (+T090) BMDMs were infected with infected (open bars) at an MOI of 100 for 18 hours before endogenous cholesterol was extracted and quantified. Values are normalized to uninfected levels (filled bars), mean ± SD. n=3. **p<0.005, ***p<0.0005, ns=not significant.

Previous studies have shown that activation of the transcription factor liver X receptor (LXR) positively regulates ABCA1 expression and stimulates cholesterol efflux [41]. Among the pharmacological LXR agonists that have been developed, T0901317 has been shown to induce ABCA1 expression in murine macrophages [42]. We hypothesized that pharmacological induction of Abca1 expression might counteract the suppression of Abca1 triggered by Salmonella infection. To test this hypothesis, macrophages were either mock-treated or pretreated with 10 μM T0901317 18 hours prior to infection and maintained in the drug throughout the course of infection. Abca1 mRNA was measured by qPCR at 5h and 24h post-infection. As shown in Fig. 3G, treatment of cells with T0901317 increased Abca1 expression 6-fold at 5h and 8-fold at 24h post-infection, indicating that pharmacological activation of LXR can successfully bypass Salmonella-mediated Abca1 suppression. In contrast to control BMDMs, treatment of cells with T0901317 completely abrogated the accumulation of BODIPY-cholesterol in infected cells at 18h (Fig. 3H–L). As expected, these findings were mirrored by the fluorometric cholesterol assay, where T0901317 treatment caused the concentration of free cholesterol in infected BMDMs to remain near uninfected levels (Fig. 3M). As expected, these findings were recapitulated during T0901317 treatment of iBMDMs (Fig. S3A–F). Together, these observations strongly suggest that the Salmonella-mediated suppression of Abca1 promotes cholesterol accumulation in infected macrophages.

Suppression of Abca1 enhances Salmonella survival in macrophages

The findings described above suggest that the FAK-dependent accumulation of cholesterol is important for virulence of Salmonella in macrophages. To test whether Abca1 suppression and the subsequent increase in intracellular cholesterol enhanced S. Typhimurium survival in macrophages, we utilized a standard gentamicin protection assay to measure bacterial internalization and survival in immortalized and primary BMDMs. As shown in Fig. 4A, internalization of S. Typhimurium in iBMDMs pretreated with either 10 mM triciribine or 10 μM T0901317 was similar to control cells treated with vehicle alone. In contrast, both Akt inhibition and induced Abca1 expression reduced bacterial survival by ~50% at 18h post-infection (Fig. 4A). To ensure that the reduced bacterial survival upon T0901317 treatment was due to Abca1 induction and not-off target toxicity, we pretreated WT and FAK-deficient BMDMs with T0901317 and measured bacterial survival. As above, T0901317 treatment had no effect on S. Typhimurium internalization in either WT or FAK^−/− BMDMs, but survival was significantly reduced in WT cells (Fig. S4A). In agreement with our previous findings that Salmonella-mediated FAK activation suppresses killing of intracellular bacteria [30], bacterial survival in untreated FAK^−/− BMDMs was significantly reduced. Importantly, pretreatment of FAK-deficient BMDMs with T0901317 did not further reduce Salmonella survival in these cells (Fig. S4A), suggesting that reduced bacterial survival was not a result of off-target toxicity, but rather the inability of S. Typhimurium to downregulate Abca1.

Figure 4 -. Suppression of Abca1 enhances Salmonella survival within macrophages.

Internalization and survival rates were determined using a standard gentamicin protection assay as described in Materials and Methods. For all conditions, cells were infected with ΔinvG S. Typhimurium at an MOI of 75. (A) iBMDMs were untreated (filled bars) or pretreated with either 10 μM T0901317 for 24 hours (open bars; +T090) or with 10 mM triciribine (checked bars; +Akt-i) for 1 hour prior to and maintained throughout infection. Values are normalized to untreated control infection. Internalization rate = CFU at 1 hour/CFU at 0.5 hour. Survival rate = CFU at 18 hour/CFU at 1 hour. (B) WT or FAK^−/− BMDMs were transfected with siLuciferase (filled bars) or siABCA1 (open bars) oligonucleotides prior to infection. Values are normalized to WT BMDM siLuciferase control infection. All values are mean ± SD and n=3. *p<0.05, **p<0.005, ns=not significant.

To directly confirm that the decreased survival of S. Typhimurium in FAK-deficient macrophages is due to the failure to suppress Abca1, WT and FAK-deficient BMDMs were depleted of endogenous Abca1 using siRNA (Fig. S4B). As anticipated, bacterial internalization was unaffected by Abca1 depletion in either WT or FAK^−/− BMDMs compared to siLuciferase-treated control cells (Fig. 4B). Interestingly, Abca1 knockdown had no effect on intracellular survival in WT BMDMs, suggesting that any further reduction in Abca1 beyond that caused by the bacteria had no additional impact on survival. In contrast, depletion of Abca1 in FAK-deficient cells restored S. Typhimurium survival to levels found in WT cells, indicating that the observed decrease in bacterial survival is directly associated with the inability to downregulate Abca1 expression. These data demonstrate that the FAK-dependent suppression of Abca1 is important for Salmonella survival in macrophages.

The T3SS-2 effector SseJ is necessary for FAK activation and subsequent cholesterol accumulation

Within 2–6 hours of infection, conditions within the SCV trigger the expression of T3SS-2 and more than 30 effector proteins that are translocated through it into the host cytosol [1, 43]. We have previously shown that activation of FAK by S. Typhimurium requires T3SS-2 [30], but the specific effectors involved remained unidentified. Among the effectors translocated by T3SS-2, SseJ and SseL have been previously implicated in cholesterol regulation. To investigate whether these or other T3SS-2 effectors were involved in FAK activation, we infected iBMDMs with a panel of SL1344 T3SS-2 effector mutants. These mutants were all on a ΔorgA background which lacks a component of the sorting system for effector delivery through the T3SS-1. ΔorgA mutants therefore fail to produce a functional T3SS-1, similar to a ΔinvG mutant [44, 45]. At 24h post-infection, cells were then lysed and immunoblotted for pFAK^Y397 and pAkt^S473. As shown in Fig. 5A, infection with ΔsifA, ΔsseF, ΔsseI, and ΔsseL strains induced similar levels of FAK^Y397 phosphorylation relative to ΔinvG. In contrast, infection with a strain lacking both T3SS-1 and T3SS-2 or a specific ΔsseJ mutant failed to increase FAK phosphorylation above uninfected levels. In agreement with our previous observation that FAK activity is required for downstream phosphorylation of Akt, both the ΔT3SS-1/ΔT3SS-2 and ΔsseJ strains also failed to induce Akt^S473 phosphorylation during infection (Fig. 5B). These observations clearly demonstrate that the activation of both FAK and Akt during S. Typhimurium infection requires the specific T3SS-2 effector, SseJ.

Figure 5 -. The T3SS-2 effector SseJ is essential for activation of FAK and Akt, downregulation of Abca1, and cholesterol accumulation in macrophages.

(A-B) iBMDMs were infected with the indicated S. Typhimurium effector mutant strains at an MOI of 100 for 24 hours. Cells were lysed and immunoblotted for (A) total and phospho-FAK^Y397 or (B) total and phospho-Akt^S473. Quantification of pFAK:total FAK (A) and pAkt:total Akt (B) averaged across 4 experimental replicates is shown below representative western blots, mean ± SD. n=4. Values were normalized to uninfected controls. (C) iBMDMs were either mock-depleted (siControl; filled bars) or FAK-depleted (siFAK; open bars) as described in Fig. 2. Cells were then infected with ΔinvG or ΔsseJ S. Typhimurium at an MOI of 100 for 5 hours. Abca1 expression was measured via qPCR. Values are normalized to uninfected levels (dotted line), mean ± SD. n=4. (D) MFI of BODIPY-cholesterol was measured from uninfected (filled) or infected BMDMs (open) with RFP-expressing ΔinvG (circles) or RFP-expressing ΔsseJ (squares) mutants as previously described. ΔinvG Uninf. n=10 cells; ΔinvG Infected n=19 cells; ΔsseJ Uninf. n=13 cells; ΔsseJ Infected n=16 cells. (E) WT BMDMs were infected with ΔinvG (empty bar) or ΔsseJ (checked bar) at an MOI of 100 for 18 hours before endogenous cholesterol was extracted and quantified. Values are normalized to uninfected levels (filled bars), mean ± SD. n=3. (F) MFI of BODIPY-cholesterol was measured from uninfected (filled) or BMDMs infected (open) with RFP-expressing ΔsseJ::pSseJ (circles) or RFP-expressing ΔsseJ::pACYC184 (squares) as described above. ΔsseJ::pSseJ Uninf. n=10 cells; ΔsseJ::pSseJ Infected n=12 cells; ΔsseJ::pACYC184 Uninf. n=10 cells; ΔsseJ::pACYC184 Infected n=15 cells. (G) WT BMDMs were infected with ΔinvG (empty bar), ΔSseJ::pSseJ (checked bar), or ΔsseJ::pACYC184 (slashed bar) at an MOI of 100 for 18 hours before endogenous cholesterol was extracted and quantified. Values are normalized to uninfected levels (filled bars), mean ± SD. n=3.

Because SseJ is essential for the activation of both FAK and Akt, we next sought to determine whether SseJ is also necessary for suppression of Abca1 transcription. For this purpose, we infected control or FAK-depleted iBMDMs with either ΔinvG or ΔsseJ mutant strains and measured Abca1 mRNA by qPCR. As shown above (Fig. 2B), infection with the ΔinvG strain resulted in significantly reduced Abca1 expression in a FAK-dependent manner relative to uninfected controls (Fig. 5C). In contrast, infection with a ΔsseJ strain not only failed to downregulate Abca1, but rather increased its expression by two-fold, independent of FAK (Fig. 5C). Interestingly, this increase is similar to the response to non-pathogenic E. coli (Fig. S2), suggesting that in the absence of active suppression, the innate host response to Gram-negative infection increases Abca1 expression in macrophages.

Subsequently, we confirmed that SseJ is essential for cholesterol accumulation in S. Typhimurium-infected macrophages. As observed above (Fig. 1A–C), BMDMs infected with the T3SS-1-deficient ΔinvG strain contain significantly more BODIPY-cholesterol than uninfected cells (Fig. 5D). Conversely, infection with the ΔsseJ mutant did not detectably increase the level of BODIPY-cholesterol fluorescence relative to uninfected cells (Fig. 5D, S5A–B), phenocopying the loss of FAK or Akt activity as well as T0901317 treatment in which Abca1 is not downregulated. Similarly, biochemical analysis indicated that while infection with the ΔinvG strain increased endogenous free cholesterol by 50% at 18 hours post-infection, BMDMs infected with the ΔsseJ mutant exhibited no increase in cholesterol content relative to uninfected cells (Fig. 5E). Complementation of the ΔsseJ strain with a plasmid encoding SseJ under control of its endogenous promoter (pSseJ) restored accumulation of both BODIPY-cholesterol (Fig. 5F, S5C) and endogenous cholesterol (Fig. 5G) in infected cells, relative to the empty vector (pACYC184) (Fig. S5D). Together, these data suggest that SseJ-mediated activation of FAK is critical for the downregulation of Abca1 and the subsequent accumulation of cholesterol in infected macrophages.

Finally, and consistent with previous reports [14], survival of the ΔsseJ mutant was reduced by ~60% in BMDMs compared to ΔinvG (Fig. S5E), and largely restored by complementation with the pSseJ expression vector compared to the empty vector alone (Fig. S5F). As standard gentamicin protection assays are only able to assess bacterial survival in a population of cells, we used an immunofluorescence-based assay to determine S. Typhimurium survival within individual cells. Similar to the gentamicin protection assay, we found that both the ΔinvG and ΔsseJ strains infected a similar percentage of BMDMs at 1h and 5h. However, by 18h post-infection, a significantly smaller percentage of cells contained the ΔsseJ strain compared to ΔinvG (Fig. S5G). The same results were found by pretreating BMDMs with T0901317 prior to ΔinvG infection (Fig. S5G), suggesting that the failure to suppress Abca1 and the subsequent inability to accumulate cholesterol reduces S. Typhimurium survival by enhancing bacterial clearance from infected macrophages.

SseJ catalytic activity is important for S. Typhimurium-induced cholesterol accumulation

SseJ is a cholesterol acyltransferase, transferring acyl chains from glycerophospholipids onto free cholesterol, generating cholesteryl esters [6, 7]. This activity allows SseJ to modify the lipid content and potentially alter the biophysical properties of host membranes. As detailed in previous reports, mutating any of three specific catalytic residues (S151, D247, H384) eliminates both SseJ deacylase and acyltransferase activity and reduces Salmonella survival [6]. Therefore, to assess whether the catalytic activity of SseJ was necessary for FAK activation and cholesterol accumulation, we generated a mutation of His384N within the pSseJ rescue vector detailed above. Subsequently, iBMDMs were infected with ΔinvG, ΔsseJ::pSseJ, ΔsseJ::pACYC184, or ΔsseJ::H384N strains of S. Typhimurium for 18h prior to lysing cells for immunoblotting. As described above, activation of both FAK and Akt occurs in the presence, but not absence of, SseJ (compare ΔsseJ::pSseJ to ΔsseJ::pACYC184). Importantly, catalytically inactive SseJ (ΔsseJ::H384N) failed to induce either FAK or Akt phosphorylation (Fig. 6A–B), suggesting that the activation of FAK/Akt signaling depends on the catalytic activity of SseJ.

Figure 6 -. The catalytic activity of SseJ is critical for FAK/Akt signaling and cholesterol accumulation.

(A, B) iBMDMs were uninfected (−) or infected with ΔinvG, ΔsseJ::pSseJ, ΔsseJ::pACYC184, or ΔsseJ::H384N strains as indicated at an MOI of 100 for 24 hours. Cells were lysed and immunoblotted for (A) total and phospho-FAK^Y397 or (B) total and phospho-Akt^S473. Quantification of pFAK:total FAK (A) and pAkt:total Akt (B) normalized to uninfected control lysates averaged across 3 experimental replicates is shown below representative western blots, mean ± SD. n=3. Values represent the ratio of phospho/total protein for each condition and are normalized to uninfected controls. (C) WT BMDMs were infected with ΔinvG, ΔsseJ::pSseJ, ΔsseJ::pACYC184, or ΔsseJ::H384N as indicated at an MOI of 100 for 18 hours before endogenous cholesterol was extracted and quantified. Values are normalized to uninfected levels (filled bars), mean ± SD. n=3. *p<0.05, **p<0.005, ***p<0.0005, ns=not significant.

Finally, we determined the importance of SseJ catalytic activity in inducing cholesterol accumulation by infecting BMDMs with the strains outlined above and extracting lipids to measure endogenous cholesterol. As expected, infection with either ΔinvG or ΔsseJ::pSseJ resulted in a significant increase in cellular cholesterol levels, while the ΔsseJ::pACYC184 empty vector control failed to increase cholesterol above uninfected levels. Similarly, infection with the strain expressing catalytically inactive SseJ (ΔsseJ::H384N) completely failed to increase levels of free cholesterol in BMDMs relative to the empty vector ΔsseJ strain (Fig. 6C). Taken together, these results demonstrate that the catalytic activity of SseJ is critical to induce cholesterol accumulation in infected cells, through the FAK- and Akt-dependent suppression of Abca1 transcription.

Discussion

Cholesterol is an indispensable constituent of eukaryotic membranes and is widely acknowledged as one of the most important regulators of lipid organization. As such, cholesterol homeostasis is tightly regulated by balancing uptake and synthesis with storage and export. Unsurprisingly, numerous intracellular pathogens have been shown to manipulate cholesterol homeostasis as part of their virulence mechanisms [23–29]. In the case of S. Typhimurium, it is well established that infection of host cells results in a significant increase in cholesterol content, but how this is achieved has remained unknown. Here we show that the T3SS-2 effector SseJ induces cholesterol accumulation in infected macrophages by downregulating expression of the cholesterol export protein ABCA1. Mechanistically, we show that the catalytic activity of SseJ initiates signaling through the host tyrosine kinase FAK and the subsequent FAK-dependent activation of Akt. Remarkably, FAK/Akt signaling has been shown to suppress expression of Abca1 in fibroblasts through inhibitory phosphorylation of two transcription factors that drive transcription of Abca1, FOXO3 and TAL1 [31]. Here, we find that S. Typhimurium increases cellular cholesterol concentrations by downregulating Abca1 expression in a FAK- and Akt-dependent manner. Conversely, pharmacological induction of Abca1 during infection circumvents bacterial suppression and suppresses cholesterol accumulation in infected cells. An S. Typhimurium strain lacking SseJ fails to activate FAK or Akt, fails to downregulate Abca1, and fails to induce cholesterol accumulation. The inability of the ΔsseJ strain to suppress Abca1 and accumulate cholesterol significantly attenuates bacterial survival and is phenocopied by pharmacological induction of Abca1 during wild-type infection. As we previously reported [30], survival of S. Typhimurium in FAK-deficient macrophages is significantly attenuated. Here we show that depletion of endogenous Abca1 in FAK-deficient macrophages restores survival to control levels. Together, these results indicate that bacterial manipulation of cholesterol homeostasis through the SseJ/FAK/Akt-dependent suppression of Abca1 contributes significantly to intracellular survival.

Activation of FAK by SseJ

At present, it remains unclear how SseJ activates FAK during infection. SseJ localizes to the cytosolic surface of SCVs, both during infection and when ectopically expressed [7, 8]. We previously demonstrated that FAK is recruited to SCVs during infection, where it is activated in a T3SS-2-dependent manner [30]. FAK exists in the cytosol in an autoinhibited conformation in which an N-terminal FERM domain folds against the kinase domain [46]. Activation typically occurs by interaction of the FERM domain with integrins, other receptors, or by direct interaction with membrane phosphoinositides, thereby relieving autoinhibition [47–49]. Although it is possible that direct binding of SseJ to the FAK FERM domain could promote FAK activation, our observation that the catalytic activity of SseJ is essential for FAK activation suggests that modification of the local lipid environment, either through the generation of lysophospholipids via its deacylase activity or the extraction of cholesterol from host membranes through esterification, has an important role. Whether this leads to FAK activation directly or indirectly remains unknown and will be the subject of future investigation.

Cholesterol metabolism in Salmonella-infected cells

In uninfected cells, depletion of cholesterol induces de novo sterol synthesis, while increased cholesterol content leads to activation of nuclear LXR receptors. LXRs target genes that promote cholesterol clearance by coordinately downregulating expression of LDL receptors and upregulating cholesterol efflux genes, including ABCA1 [42, 50]. Current models propose a mechanism by which ABCA1 moves cholesterol from the cytoplasmic leaflet to the extracellular leaflet of the plasma membrane where it is transferred to circulating ApoA1, generating HDL in the plasma [19, 22, 51]. Loss of ABCA1, which occurs in Tangier’s disease, results in intracellular retention of cholesterol and prevents ER cholesterol sensing by impairing retrograde cholesterol transport from the PM to the ER [20, 21, 52]. This is notable, as previous studies using metabolic labeling have shown that de novo cholesterol synthesis is actually increased in response to S. Typhimurium infection despite elevated cholesterol levels [3, 4, 53]. Therefore, the downregulation of ABCA1 may also inhibit cholesterol sensing by the ER, allowing for the continuation of sterol synthesis during infection.

Although de novo synthesis accounted for a relatively small fraction of total cellular cholesterol in infected macrophages, these studies suggest that production of non-sterol cholesterol precursors, rather than mature cholesterol, contributes to intracellular survival. Statins, which inhibit an early stage of cholesterol biosynthesis, were found to reduce Salmonella survival in macrophages, while the addition of mevalonate restored survival to control levels [4]. However, inhibition of either squalene oxidase or Δ24 sterol oxidase, later enzymes in the sterol synthesis pathway, had no effect on survival [4, 53]. While these studies suggest that cholesterol biosynthesis is unnecessary for Salmonella survival, our findings suggest that a large fraction of the accumulated cholesterol is derived from the uptake of and inability to export excess cholesterol.

Role of cholesterol accumulation in Salmonella survival

How cholesterol accumulation in the endocytic pathway benefits S. Typhimurium remains uncertain; however, there are several possibilities that are not mutually exclusive. First, high cholesterol has been shown to inhibit autophagic flux by sequestering the SNARE proteins required for autophagosome/lysosome fusion [54, 55]. Flux can be restored by depletion of cholesterol with methyl-β-cyclodextrin or T090187 treatment [56, 57], suggesting that cholesterol accumulation may protect the bacteria from autophagic clearance. Furthermore, cholesterol delivery from the ER to the cytoplasmic leaflet of endolysosomes via OSBP promotes the recruitment and activation of mTORC1, leading to suppression of autophagy [58]. Thus, an important role of Salmonella-dependent cholesterol accumulation may be to inhibit autophagic clearance of the bacteria.

Second, recent evidence indicates that OSBP is specifically recruited to the SCV in epithelial cells through its interaction with both SseJ and SseL, where it is required for vacuolar stability [12]. Depletion or inhibition of OSBP leads to an increase in cytosolic bacteria that have escaped the vacuole. While this suggests that cholesterol delivery to the vacuole from the ER may be essential for SCV stability, the requirement for lipid transfer activity of OSBP was not tested directly.

Third, elevated cholesterol content inhibits lysosomal maturation and the acidification of late endosomes/lysosomes by inhibiting vacuolar ATPase pump activity [59]. High cholesterol has been reported to block recycling of mannose-6-phosphate receptors (MPRs) from endosomes to the TGN, resulting in extracellular secretion of lysosomal hydrolases and their depletion from lysosomes [60, 61]. It is well established that SCVs contain low levels of lysosomal hydrolases despite the presence of lysosomal membrane proteins in their limiting membrane [62]. This is due, at least in part, to the formation of a stable complex between the T3SS-2 effector SifA and host Rab9, which inhibits the Rab9-dependent recycling of MPR to the TGN [63]. However, accumulation of endosomal cholesterol in cells also inhibits MPR recycling by sequestering Rab9 in the absence of infection [60], and it is possible these two mechanisms act in tandem to deplete lysosomal hydrolases from SCVs. Therefore, cholesterol accumulation may prevent the acidification of endosomes and allow for S. Typhimurium to avoid exposure to antimicrobial hydrolases, ensuring bacterial survival.

Fourth, recent evidence suggests that SIFs are important for Salmonella growth and survival within host cells by facilitating access of intravacuolar bacteria to host nutrients and enhancing bacterial metabolism [64, 65]. SIFs are rich in cholesterol [5], but whether cholesterol is essential for their formation or function remains unknown. Although the role of SseJ in SIF formation has been previously investigated, the data are conflicting [14, 66, 67]. However, ectopic co-expression of SseJ and SifA is sufficient to induce the formation of SIF-like tubules from endolysosomal compartments in a manner that requires SseJ catalytic activity [68], suggesting that remodeling of the endolysosomal membrane by SseJ is essential for tubulation. These findings highlight the importance of cholesterol in regulating endolysosomal dynamics and suggest that Salmonella-induced cholesterol accumulation may promote SCV integrity and manipulate endolysosomal trafficking. Because cholesterol plays a major role in the regulation of membrane sorting and the partitioning of proteins between membrane domains [69, 70], it is reasonable to expect that the level of cholesterol accumulation seen during S. Typhimurium infection significantly impacts many host processes by altering endosomal trafficking. Future studies will be necessary to determine the exact role of cholesterol accumulation in endosomal transport.

As noted above, SseJ is a glycerophospholipid:cholesterol acyltransferase, catalyzing the transfer of sn-1 acyl chains from glycerophospholipids onto free cholesterol, in the process generating both lyso-phospholipids and cholesterol esters [6–9]. At present it is not clear which of these two activities is important for intracellular survival. Lyso-phospholipids regulate many cellular processes including epithelial permeability and phagocytosis [71, 72], however, the role of lyso-phospholipids during Salmonella infection remains unknown. It is possible that the generation of lyso-phospholipids in the SCV membrane is critical for regulating SIF formation and SCV membrane dynamics, and that the elevated local cholesterol content serves to sequester the free acyl chains removed from glycerophospholipids.

In summary, we elucidate a novel role for the Salmonella T3SS-2 effector SseJ in manipulating host cell cholesterol homeostasis by promoting cholesterol accumulation in macrophages. The catalytic activity of SseJ prevents bacterial clearance from BMDMs through the FAK- and Akt-dependent downregulation of Abca1 and the subsequent increase in intracellular free cholesterol. While we show that suppression of Abca1 enhances Salmonella Typhimurium survival, the specific role of cholesterol accumulation in bacterial pathogenesis remains uncertain. Our future studies will focus on determining how elevated cholesterol affects SCV membrane integrity and regulates endolysosomal trafficking.

It should be noted that these experiments were conducted in macrophages derived from C57BL/6 mice, which due to the lack of the phagosomal iron/manganese transporter Nramp are susceptible to systemic S. Typhimurium infection and are widely used to model human typhoid fever. However, S. Typhimurium causes a self-limiting enteritis in otherwise healthy humans and does not typically colonize circulating macrophages. Whether similar induction of cholesterol accumulation occurs during infection of intestinal epithelial cells and whether this is important for virulence in humans will also require further study.

Materials and Methods

Ethics statement

All experiments in this study were performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. Protocols were approved by the Institutional Animal Care and Use Committee at the University of Virginia (Protocol number 3488). All efforts were made to minimize animal suffering during the course of these studies.

Mice

The generation of myeloid-specific conditional FAK knockout mice and their control littermates have been described previously [73]. Mice were kept in pathogen-free conditions and allowed free access to food and water.

Cell culture

The hind leg bones of mice were isolated and bone marrow-derived macrophages (BMDMs) were extracted. BMDMs were plated onto untreated petri dishes and cultured in RPMI (Genesee Scientific) media containing 10% heat-inactivated fetal bovine serum (FBS), 1% penicillin/streptomycin/amphotericin B (PSF), and 10% L929-conditioned media as a source of colony stimulating factor-1 (CSF-1). iBMDM cells were a kind gift from Dr. Jonathan Kagan (Harvard University, Boston, MA). iBMDMs were grown on cell-culture treated petri dishes (Fisher) and cultured in Dulbecco’s modified Eagle medium (DMEM; Genesee Scientific) supplemented with 10% FBS and 1% PSF.

Bacterial strains and culture

The Salmonella Typhimurium SL1344 strain ΔinvG mutant (T3SS1-deficient), and the ΔorgA::tet, ΔspiA::kan (double T3SS1 and T3SS2 mutant; ΔT3SS1ΔT3SS2) strain have been described previously [74–76]. The T3SS2 effector mutants (ΔsifA, ΔsseF, ΔsseI, ΔsseL, and ΔsseJ) were all on a ΔorgA (T3SS1-deficient) background and were all a generous gift from Dr. Denise Monack (Stanford University, Stanford, CA). Bacteria were grown under non-invasion-inducing conditions. Briefly, a single colony was inoculated into LB broth with the proper antibiotics and grown statically overnight. Bacteria were harvested the following morning when their OD600 reached 0.6–0.8. Experiments using non-pathogenic Escherichia coli strain DH5ɑ followed the same protocol.

Bacterial infection

30 minutes prior to infection, cells were washed in PBS and cell culture media was replaced with fresh infection media containing 10% heat-inactivated FBS without antibiotics. Bacteria were diluted in this medium and cells were then infected with the indicated bacterial strain and multiplicity of infection (MOI) for 30 minutes. After 30 minutes, cells were washed in PBS and fresh infection medium containing gentamicin (100 μg/ml; Fisher Scientific) was added for 60 minutes to kill extracellular bacteria. After 60 minutes (90 minutes total), the concentration of gentamicin was reduced to 10 μg/ml for the remainder of the assay.

Microscopy

BMDMs were plated on fibronectin-coated (5 μg/ml) glass-bottom 3.5 mm imaging dishes (Mat-Tek Corp.), or fibronectin-coated glass coverslips in a 24-well dish. Cells were infected with the indicated bacterial strain at an MOI of 50. For BODIPY-cholesterol imaging, at 16h post-infection, cells were rinsed twice with PBS and serum-free DMEM containing 10 μg/mL gentamicin was added to the cells. 30 minutes later, cells were incubated with serum-free DMEM (10 μg/mL gentamicin) containing 0.5 μM BODIPY-cholesterol (TopFluor Cholesterol; Avanti Polar Lipids) for 45 minutes, rinsed twice in PBS, and chased for 45 minutes in phenol red-free imaging medium (140 mM NaCl, 2.5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 20 mM HEPES pH 7.4, with 10% heat-inactivated FBS and 0.45% glucose) containing 10 μg/mL gentamicin. Cells were held at 37 °C and imaged at approximately 18h post-infection using a 40x or 60x objective fitted to a Nikon TE 2000 microscope equipped with Yokogawa CSU 10 spinning disc and 512X512 Hamamatsu 9100c-13 EM-BT camera. Green fluorescence was excited with a 488 nm/100 mW diode laser (Coherent) and collected by a BP 527/55 filter. Red fluorescence was excited with a 561 nm/100 mW diode laser (Coherent) and collected by a BP 615/70 filter. Multicolor images were acquired sequentially and compiled using FIJI software [77]. Grayscale image LUTs were inverted to generate a white background with black fluorescence. For RFP-Salmonella images, LUT color was edited to generate red fluorescence on a white background while maintaining the same scaling as found in grayscale images. Images were merged using minimum fluorescence intensity to avoid conflicts between the different color schemes and did not notably change image appearance. Exposure time and capture settings were identical between infected and uninfected cells and cells with or without T0901317 treatment. All images (except Fig. S1C) were captured using a 60x objective and unless otherwise stated, no manipulation of brightness or contrast occurred for BODIPY-cholesterol images. RFP-ΔinvG was digitally adjusted for brightness and contrast to allow for better identification in merged images with black puncta. For low magnification imaging using a 40x objective, BODIPY-cholesterol image contrast was digitally increased to better show uninfected cells (High Contrast panels).

For percent infection experiments, cells were fixed at the indicated time point in 4% PFA and co-stained with an anti-Salmonella (Thermo) and AlexaFluor 647-phalloidin (Invitrogen). Images were captured using a 40x objective on a Nikon C1 Plus Confocal microscope. Mean fluorescence intensity of images was analyzed using NIS-Elements software (Nikon).

Endogenous cholesterol quantification assay

Cholesterol extraction was performed using the Bligh and Dyer method as previously described [37]. In brief, cells were lysed in a chloroform:methanol mixture and lipids were harvested and pelleted from the organic layer. Lipids were resuspended in an isopropanol:NP40 solution and cholesterol was analyzed following a protocol from Wang, et al., 2010 [38]. Lipid samples plated in a 96-well dish and were treated with bovine catalase (Sigma) before addition of cholesterol oxidase (Sigma), HRP (Sigma), and ADHP (Amplex Red; Invitrogen). Fluorescence was measured using a Cytation 1 plate reader (BioTek).

Drug treatments

Cells were pretreated either overnight with 10 μM T0901317 (Tocris) or for 1h with 10 mM triciribine (Sigma) prior to infection. Treatment was maintained throughout the course of infection.

siRNA and plasmid nucleofection

20 nM siRNA oligonucleotides targeting murine Abca1 (Dharmacon), FAK (Dharmacon), siLuciferase control (Dharmacon), and non-targeting controls (Invitrogen) were nucleofected into macrophages using program Y-001 on an Amaxa Nucleofector II. Nucleofection was performed 48 hours prior to infection.

Antibodies

Immunoblot analyses were performed using the following antibodies: FAK (Santa Cruz Biotechnology, Inc.), pFAK^Y397, Akt, pAkt^S473, HRP-linked anti-mouse IgG, HRP-linked anti-rabbit IgG (all Cell Signaling), IRDye® 680RD anti-mouse IgG, IRDye® 800CW anti-mouse IgG, IRDye® 680RD anti-rabbit IgG, and IRDye® anti-rabbit IgG (all LI-COR Biosciences). The mouse Abca1 antibody was a generous gift from Dr. David Castle (University of Virginia, Charlottesville, VA). Immunofluorescence staining was performed using rabbit anti-Salmonella (Thermo), donkey anti-rabbit AlexaFluor 488, and AlexaFluor 647 phalloidin (both Invitrogen).

Western blotting

Cells were seeded at 1.5×10⁶ cells/well onto 6-well culture dishes 18h prior to infection at an MOI of 100. Cells were then rinsed twice with PBS and lysed in Triton X-100 lysis buffer (50 mM Tris pH 7.4, 1% Triton X-100, 150 mM NaCl, and 2 mM EDTA) supplemented with 1 mM sodium vanadate, 50 mM sodium fluoride, and a cocktail of protease inhibitors (0.1 mM phenylmethylsulfonyl fluoride and 1 mg/ml each of pepstatin, leupeptin, and antipain). Samples were loaded onto 4–20% SDS-polyacrylamide gels and probed using the indicated antibodies. Densitometry was performed using Image Studio Lite software (LI-COR Biosciences).

RNA extraction and qPCR

Cells were seeded at 1.5×10⁶ cells/well onto 6-well culture dishes 18h prior to infection at an MOI of 100. Cells were infected at an MOI of 100 for 30 minutes, followed by gentamicin treatment for the indicated time points. RNA was extracted from iBMDMs using a Trizol (Ambion) extraction protocol as per the manufacturer’s instructions. cDNA production was performed using High Capacity cDNA Reverse Transcription kit (Applied Biosystems). Real-time qPCR analysis was performed using the ABI PRISM SDS7000 sequence detection system (Applied Biosystems) and the following Applied Biosystems TaqMan primer-probe sets: Mm00442646_m1 Abca1, Mm00437390_m1 Abcg1, and Hs99999901_s1 18S ribosomal subunit. PCR was performed using SYBR Green master mix (Applied Biosystems) as recommended by the manufacturer. The ΔΔCT method was used to quantify all relative mRNA levels as described (ABI user guide, 1997), using 18S RNA as the reference and internal standard.

Gentamicin protection assay

The gentamicin resistance assay has been described previously [57]. Briefly, cells were seeded at 1×10⁵ cells/well onto 24-well culture dishes 18h prior to infection at an MOI of 75. Cells were then lysed at 30 minutes, 1h, or 18h post-infection in 0.2% Triton X-100 in PBS and resuspended in additional PBS. CFUs were enumerated by plating 100 μL lysate dilutions onto LB agar.

SseJ complementation vector

Primers (IDT) were designed to specifically amplify SseJ (NCBI gene accession number NP_460590.1) and its putative promoter sequence [8] out of the Salmonella Typhimurium SL1344 genome. N-terminus primer: 5′-CCGCGCGGATCCGTCAGATAATATGTACCAGGC-3′; C-terminus primer: 5′-CGCCTCGACTTCAGTGGAATAATGATGAGC-3′. The SseJ PCR product and pACYC184 prokaryotic expression vector were digested using BamHI and SalI restriction enzymes (New England Biolabs) and ligated together using T4 DNA ligase (New England Biolabs) to generate the pSseJ complementation vector. NEB Q5 Site-Directed mutagenesis (New England Biolabs) was used to generate the H384N catalytically inactive SseJ vector derived from the SseJ complementation vector described above. Primers (IDT) were designed with a single point mutation, converting a His residue to an Asn residue. N-terminus primer: 5’-CGACCTTGTCaatCCAACCCA-3’; C-terminus primer: 5’-TTGAAGACGTATTGCGGAC-3’. The pACYC184 vector was a kind gift from Melissa Kendall (University of Virginia, Charlottesville, VA).

Statistical analysis

Student’s t test was used for the comparison of 2 independent groups. Two-way ANOVA with Tukey’s multicomponent post-test was used when comparing more than 2 independent groups. All tests were performed with Graphpad Prism8, and p-values of 0.05 were considered statistically significant.

Takeaways.

1. Salmonella Typhimurium infection of macrophages induces accumulation of intracellular cholesterol in a T3SS-2 dependent manner

2. The secreted T3SS-2 effector SseJ promotes downregulation of the cholesterol exporter ABCA1, by activating the host kinases FAK and Akt, in a manner that requires SseJ acyltransferase activity

3. Cholesterol accumulation enhances intracellular survival, and restoration of cholesterol homeostasis attenuates survival, indicating that active manipulation of host cholesterol content is an important virulence mechanism