Salmonella effector driven invasion of the gut epithelium: breaking in and setting the house on ﬁre

Salmonella Typhimurium ( S .Tm) is a major cause of diarrheal disease. The invasion into intestinal epithelial cells (IECs) is a central step in the infection cycle. It is associated with gut inﬂammation and thought to beneﬁt S .Tm proliferation also in the intestinal lumen. Importantly, it is still not entirely clear how inﬂammation is elicited and to which extent it links to IEC invasion efﬁciency in vivo . In this review, we summarize recent ﬁndings explaining IEC invasion by type-three-secretion-system-1 (TTSS-1) effector proteins and discuss their effects on invasion and gut inﬂammation. In non-polarized tissue culture cells, the TTSS-1 effectors (mainly SopB/E/E2) elicit large membrane rufﬂes fueling cooperative invasion, and can directly trigger pro-inﬂammatory signaling. By contrast, in the murine gut, we observe discreet-invasion (mainly via the TTSS-1 effector SipA) and a prominent pro-inﬂammatory role of the host?"s epithelial inﬂammasome(s), which sense pathogen associated molecular patterns (PAMPs). We discuss why it has remained a major challenge to tease apart direct and indirect inﬂammatory effects of TTSS-1 effectors and explain why further research will be needed to fully determine their inﬂammation-modulating role(s).


Introduction
Salmonella enterica is a key global cause of foodborne diarrhea with Salmonella Typhimurium (S.Tm) as one of the main serotypes affecting humans.S.Tm also serves as a prototype for studying the general principles of Salmonella infection biology.
S.Tm invasion into the gut tissue is intimately linked to diarrhea and gut inflammation, two hallmarks of the disease [1][2][3][4].Gut inflammation provides S.Tm with a competitive advantage to outcompete the gut resident microbiota and to establish its niche in the gut lumen [5,6].Thus, to understand how the pathogen benefits from eliciting enteric disease, we need to assess the host cells which are targeted by the pathogen, the mechanisms of cell invasion and how this invasion process triggers inflammation in the host?"s gut.
Work in primarily calf and mouse infection models has shown that S.Tm enters the gut tissue via different routes, including the uptake by M cells or microbe-sampling dendritic cells and the active invasion of absorptive intestinal epithelial cells (IECs) [7].IEC invasion is best studied, thought to be the main driver of gut inflammation, and hence constitutes the focus of this review.This process strictly depends on the Salmonella pathogenicity island 1 (SPI-1), which encodes a needle-like injection structure, called the type-three-secretion-system-1 (TTSS-1), and several effectors for invasion of IECs [7,8].Upon entry, the infected host cell mounts a first line of pro-inflammatory responses.It is still not entirely clear to which extent pro-inflammatory responses are triggered by signal cascade manipulation via TTSS-1 effectors, or by the host cell?"s pattern recognition receptors (PRRs) that sense pathogen associated molecular patterns (PAMPs) such as peptidoglycan, lipopolysaccharide, flagellin and the TTSS-1 [9,10].Either way, IEC invasion represents an essential event in the pathogen?"sinfection cycle.
Here, we review the recent progress in our understanding of S.Tm IEC invasion, with a particular emphasis on the intact host gut.We focus on SPI-1 mediated invasion, since this is best understood at the molecular and cellular level and has been studied using various animal models including mice.We discuss how TTSS-1 effectors collaborate with other Salmonella virulence factors, how they contribute to IEC invasion, and their possible roles beyond.

Molecular basis for S.Tm invasion into nonphagocytic cells
Infections, biochemical experiments, and ectopic expression studies using transformed/immortalized cell line cultures have established the canonical view of S.Tm invasion into non-phagocytic cells, the molecular basis of which has been reviewed elsewhere [7].Briefly, TTSS-1 effectors interact with the host actin cytoskeleton to trigger lamellipodia and filopodia-containing surface protrusions (called membrane ruffles) which engulf S.Tm into a tight intracellular compartment [11][12][13].The size of these ruffles extends far beyond a single S.Tm, enabling the simultaneously internalization of several bacteria a phenomenon denoted cooperative invasion [14 ,15,16].The main TTSS-1 effectors driving this type of uptake are SopB, SopE and SopE2, which indirectly promote Arp2/3-dependent actin polymerization through nucleation promoting factors WAVE and WASH [17,18].SopB is believed to be a lipid phosphatase and/or a phosphotransferase/phosphoisomerase that manipulates phosphoinositide dynamics at the plasma membrane to recruit Arf1 [18][19][20].SopE and SopE2 are archetypical members of the WxxxE family of effector proteins [21], mimicking guanine nucleotide exchange factors to activate for example, Rac1 [15,22,23].Together, Arf1 and Rac1 govern the WAVE Regulatory Complex (WRC; a complex of WAVE and co-factors) to initiate Arp2/3 dependent actin polymerization [24] (Figure 1a).Importantly, the ectopic expression of SopB, SopE or SopE2 is sufficient to elicit pronounced membrane ruffling and/or facilitate the uptake of inert particles and non-invasive bacteria in transformed tissue culture cell models [15,23,25].In contrast to SopB/E/E2, which target upstream actin-regulatory processes, the TTSS-1 effectors SipA and SipC can directly bind to actin.They feature actin nucleating and bundling activities to target the actin cytoskeleton directly, which supports membrane ruffles, thus invasion [26][27][28][29].While SipC is a translocon component required for TTSS-1 function and invasion, deletion of SipA only has a minor impact on invasion in cultured cell lines.Notably, SipA can drive cell invasion into non-polarized tissue culture cells independent of SopB/E/E2 in a morphologically distinct process which lacks prototypical ruffles [30][31][32][33].
This classical view of SPI-1-triggered invasion has been extended with new TTSS-1-independent mechanisms in recent years, since SPI-1-deficient S.Tm mutants still feature a residual invasion capacity in cell lines.Rck is the best-studied TTSS-1-independent invasion factor [34].This is an outer membrane protein that binds to the epidermal growth factor receptor (EGFR) leading to Arp2/3 activation through Rac1 and Akt [35][36][37].Since EGFR localizes to the basolateral side of IECs, Rck could potentially promote IEC invasion from the lamina propria compartment [36].However, a Rck-associated phenotype in vivo remains to be shown.PagN is another outer membrane protein, which can drive cell invasion independent of TTSS-1 [38 ,39].Furthermore, a study postulated the existence of additional invasion factors, since a strain simultaneously deficient in TTSS-1 (invA), Rck and PagN was still able to invade certain cell lines [40].Together, this demonstrates that there is still room for discoveries regarding S.Tm invasion mechanisms.
Despite the detailed understanding of the molecular mechanisms underlying TTSS-1 effector-triggered invasion by S.Tm, the in vivo relevance and the associated invasion phenotypes into the gut tissue of an infected host has only received modest attention historically [41][42][43][44][45]. Recent publications partially fill this gap and provide new insights into how S.Tm invades IECs in vivo.Here, we summarize recent studies of invasion processes in vivo, particularly in murine infection models, which have provided the deepest insights.We discuss pre-invasion factors and how TTSS-1 effectors influence invasion, inflammation and intracellular location/survival within IECs.We focus on these early steps of the mucosal infection cycle, since SPI-1 expression has been shown to be promptly downregulated following traversal of the epithelium [46,47].

Pre-invasion factors prepare S.Tm for SPI-1dependent IEC invasion in vivo
Upon arrival in the gut lumen, S.Tm integrate a number of environmental inputs to trigger an elaborate signaling cascade that ramps up expression of flagella, TTSS-1 and the SiiE adhesin needed to attack the epithelium [48,49].Combined, these virulence factors enable S.Tm to swim to the epithelium, attach to its surface and invade into IECs.Most parts of the epithelium are covered with a thick layer of Muc2-containing mucus, studded with antimicrobial proteins, secreted IgA and numerous other factors [50] that reduce pathogen access to the epithelial surface.However, this mucus layer has channels and gaps which S.Tm exploits via flagella-driven near surface swimming to reach exposed IECs [51].The flagellum of S.Tm can be composed of two antigenically distinct flagellins, FljB and FliC.Compared to FljB, FliC-expressing S.Tm are slightly more invasive, which has been explained by a distinct near-surface swimming phenotype with more frequent stops along the surface of cultured cell lines [52].Furthermore, posttranslational methylation of FliC increases S.Tm invasion efficiency [53 ].The methylase FliB adds methyl groups on flagellasurface exposed lysine residues to increase hydrophobicity.In S.Tm strains locked for FliC expression (cannot switch to FljB expression), FliB deficiency leads to decreased IEC invasion.It was reasoned that hydrophobicity increases binding efficiency to phosphatidylcholine, the most abundant phospholipid present on host cell membranes (Figure 1b).Attachment is additionally supported by the giant adhesin SiiE encoded on SPI-4, shown to be important for IEC invasion in vivo [54 ,55].SiiE co-localizes with MUC1, a cell-surface protein expressed on the apical surface of epithelial cells.During infection of HT29-MTX cell layers, MUC1-deficient cells are more resistant to S.Tm as well as S. Enteriditis invasion [56 ].The combined data suggest that SiiE promotes invasion by binding to MUC1 glycans (Figure 1b).S.Tm genomes encode more than a dozen additional adhesins [57], several of which shown to affect intestinal colonization [58,59], but their specific impact on IEC attachment in vivo remains unclear.The TTSS-1 apparatus further promotes attachment by docking into the host cell membrane [60][61][62].
SPI-1 expression is costly to the pathogen, reducing its growth rate by as much as 50% [63].To limit these costs, virulence factor expression occurs only at times of need, that is, to initiate gut tissue invasion [64,65].It is regulated by a complex network of transcription factors including HilD [66], which also tunes S.Tm swimming behaviors [67].This is illustrated by LoiA (low oxygen induced factor A; encoded on SPI-14), which promotes SPI-1 expression through hilD at low oxygen condition as found in the gut lumen.S.Tm deficient for LoiA feature reduced invasion efficiency into CaCo-2 cells and are less virulent in vivo [68].In summary, pre-invasion factors including flagellin variants, adhesins, SPI-1-regulating Salmonella effector driven invasion of the gut epithelium Fattinger, Sellin and Hardt 11 transcription factors, and the TTSS-1 apparatus itself prime S.Tm to invade IECs of the gut mucosa (Figure 1b).
The SPI-1-driven IEC invasion step in the intact gut epithelium Upon docking to the epithelium, the pre-formed TTSS-1 effectors are delivered by TTSS-1 into the IEC cytosol to fuel invasion.In the infected gut, the individual contributions of the TTSS-1 effectors have been challenging to study, due to their redundancy and in vivo complexity.Recently, a neonate mouse model and a specific immune-deficient mouse line (Nlrc4  [70 ].Importantly, it was sufficient to delete only SipA in S.Tm SL1344 to detect a pronounced IEC invasion defect in adult mice [54 ].This effect was less obvious in neonate mice infected with the corresponding S.Tm 14028 mutant [70 ], suggesting strain specific variations or differences in epithelium maturation stage of neonate versus adult mice.Nevertheless, these two studies demonstrate that SopE/E2 and in particular SipA drive IEC invasion in vivo (Figure 1b).This appears surprising, as SipA contributes only slightly to S.Tm invasion in nonpolarized tissue culture models.Strains that lack SopB/ E/E2 rely on SipA in collaboration with SipC for a zipper-like invasion process [30,31].Based on these observations, S.Tm invasion into the mature mouse gut epithelium might proceed without expansive membrane ruffles.Indeed, in line with previous observations in ileal gut-lops from pigs and calves [41,72], IEC invasion by wt S.Tm in mice is characterized by smaller finger-like protrusions and does not fuel cooperative invasion, which is a distinct feature of ruffle-mediated entry (Figure 1c, [54 ]).These observations support that SipA can drive discreet IEC invasion in the mature epithelium of mice (Figure 1b).Moreover, in the complex in vivo environment, S.Tm exploits cell?cell junctional zones and cells neighboring goblet cells for efficient invasion (Figure 1b) [54 ].Differences between mouse epithelium and cell lines can be partially explained by the degree of host cell polarization [54 ].Accordingly, it was shown that SipA dependent manipulation of villin located at the brush border of polarized IEC can influence invasion in vivo [73].
Overall, the comparison of well-controlled in vivo and cell culture experiments demonstrates that the contribution of individual TTSS-1 effectors for invasion is contextdependent.The relative roles of SopB, SopE, SopE2, and SipA are strongly affected by the cellular environment, cell-polarity and cell-maturation.[14 ,54 ].Hence, we can expect continued progress in our understanding of SPI-1 driven IEC invasion and the specific impact a physiological host cell and tissue context has on this process.

Pro-inflammatory signaling elicited by the invading pathogen
SPI-1 driven IEC invasion leads to inflammatory responses.Since the induction of inflammation is crucial for outcompeting luminal microbiota, it has been suggested that TTSS-1 effectors control the degree of inflammation by interfering directly with pro-inflammatory signaling, cell death pathways and/or tight junctions [10,[83][84][85].Inflammatory NF-kB signaling can be affected by various TTSS-1 effectors such as SipA, SopB/E/E2 and AvrA [86][87][88][89].Accordingly, a recent study demonstrated that Rab8 GTPase can be targeted by the TTSS-1 effector SopD to interfere with inflammatory signaling [90 ].Cell death pathways were shown to be promoted or suppressed by SipA and SopB/E [91][92][93][94][95], and finally, tight junctions were suggested to be targeted by SipA, SopB/E/E2 and AvrA to either increase or decrease inflammation [96][97][98].These studies employed mainly tissue culture cell models and arrived at the conclusion that the TTSS-1 effectors are fine-tuning the infected host cell?"s response to the benefit of the pathogen.
Thus, TTSS-1 effectors might control inflammation by directly interfering with multiple host immune responses.However, verifying the role of the proposed mechanisms during gut inflammation in vivo remains a major challenge.This is explained by the different processes that can elicit pro-inflammatory responses in the infected cell (Figure 2).It is known that S.Tm-related PAMPs such as peptidoglycan, flagellin, TTSS components and LPS can trigger similar inflammatory signaling pathways as the ones which were suggested to be directly manipulated by TTSS-1 effectors (Figure 2) [80 , [99][100][101][102][103][104][105].Since TTSS-1 effectors influence IEC invasion efficiency, intracellular location and survival (discussed below), they will also determine the amount of PAMPs that can be recognized by PRRs within IECs and other cell types of the deeper tissue to elicit pro-inflammatory signaling (Figure 3).Notably, expression of PRRs and other innate immune signaling components can differ markedly between immortalized cell lines and the mature gut tissue [106].
Therefore, it appears plausible that epithelial cell line experiments have tended to particularly identify TTSS-1 effector-driven manipulation of host cell responses due to the partial or complete absence of PAMP-triggered proinflammatory cell death pathways.Furthermore, as discussed above, the contribution of TTSS-1 effectors for cell invasion is context-dependent, which implies that cell culture invasion data cannot be used as a proxy for in vivo IEC invasion [54 ,70 ].In summary, the points above make it difficult to draw definite conclusions and could explain some of the contradictory results pertaining to links between TTSS-1 effectors and inflammation.To formally prove effector-dependent induction of immune responses in the absence of PAMPs in vivo, transgenic mice with cell type-specific inducible expression of TTSS-1 effectors would be required, similarly to the investigations of the Helicobacter effector CagA [107].
However, even in such an experimental setup one would need to carefully control for possible contribution of gut luminal PAMPs.

TTSS-1 effector expression beyond invasion
Cell culture studies have established the concept that TTSS-1 effectors are expressed beyond invasion to Salmonella effector driven invasion of the gut epithelium Fattinger, Sellin and Hardt 13  The indirect contribution of TTSS-1 effectors to the degree of inflammation.TTSS-1 effectors influence invasion efficiency, intracellular location and intracellular survival/replication of S.Tm, which indirectly influence the amount of PAMPs that can be sensed by host immune receptors (red stars).Therefore, TTSS-1 effectors affect the pro-inflammatory signaling by both direct manipulation of the signaling cascades (Figure 2) and indirectly by increasing the density of bacteria/PAMPs within the infected cell.
control host cell machineries and thereby the intracellular fate of S.Tm.Host cell internalization leads to the formation of a Salmonella-containing-vacuole (SCV) [108], in which the membrane is tightly surrounding the bacteria [12].This SCV can either grow by fusion with macropinosomes or shrink by emanating membrane tubules promoting intracellular growth in the SCV or in the host cytosol, respectively [109 ].SopB influences the integrity of the SCV and together with SopE supports SCV escape into the cytosol [109 ,110,111 ].SopE/E2 are additionally associated with early intracellular replication [112].SipA also promotes initiation of intracellular replication and/or influences S.Tm survival and localization within epithelial cells [70 ,110,113,114 ].Finally, two recent studies reported new insights about a so far poorly characterized TTSS-1 effector, SopF [115 ,116 ].SopF was shown (i) to engage the V-ATPase-ATG16L1 axis thereby inhibiting xenophagy-dependent S.Tm restriction, and (ii) to interact with phospholipids and stabilize the SCV.Importantly, while some of these concepts such as intra-IEC replication and dual location (in SCV and/or cytosolic) were shown to have implications in vivo [46,47,69,70 ,103,117 ], others lack formal in vivo validation.Given that most infected IECs have a short lifetime and are promptly expelled into the gut lumen [75 ,101-103], some of these TTSS-1 effector-dependent intracellular fate modulations might have a minor impact in vivo.This argument is supported by the observation that intracellular SopB expression is mainly detected in expelled epithelial cells in the gall bladder of S.Tminfected mice [91].However, not all infected IECs are expelled, which would allow intracellular manipulation also over longer time periods at least within a fraction of the invasion foci.

Conclusions and perspectives
Extensive research has provided a detailed assessment of how S.Tm invades IECs with the help of TTSS-1 effectors.It is evident that S.Tm has evolved to invade host cells for which it can employ multiple invasion strategies and mechanisms.As SipA is encoded within SPI-1, while SopB, SopE, and SopE2, and the discussed adhesins are encoded elsewhere on the chromosome, it is tempting to speculate that SipA-mediated discreet-invasion represents the primordial IEC invasion mechanism and that initially, PAMP-triggered innate immune responses may have been the main mechanism(s) for eliciting gut inflammation.In this scenario, later acquisition of additional effectors would provide the ability to fine-tune these responses to the pathogen?"sbenefit.This may explain the broader redundancy of effectors and context-dependent differences.These circumstances pose technical challenges to the deciphering of how IEC invasion and the triggering of gut inflammation are interconnected in the host?"s gut.The authors show that non-transformed enteroids contract upon TTSS-1dependent S.Tm invasion, which densifies the IEC packing at the site of invasion.This response is induced by the epithelial NAIP/NLRC4 inflammasome and does not rely on other cell types. Figure1 Figure 2 Figure 3

14 .
Di Martino ML, Ek V, Hardt WD, Eriksson J, Sellin ME: Barcoded consortium infections resolve cell type-dependent Salmonella enterica Serovar Typhimurium entry mechanisms.mBio 2019, 10 This study compares the contribution of S.Tm TTSS-1 effectors for invasion into epithelial cells versus monocytes and macrophages.The authors use an internally controlled approach in which genetically tagged wild type and effector mutant S.Tm strains are assessed for their invasion efficiency.epithelium-autonomousNAIP/NLRC4 dependent expulsion of infected cells limits S.Tm transversal into the lamina propria, which otherwise induces a TNF hyper-response that destroys the epithelial barrier.76.ForbesterJL, Goulding D, Vallier L, Hannan N, Hale C, Pickard D, Mukhopadhyay S, Dougan G: Interaction of Salmonella enterica Serovar Typhimurium with intestinal organoids derived from human induced pluripotent stem cells.Infect Immun 2015, 83:2926-2934.77.Geiser P, Di Martino ML, Samperio Ventayol P, Eriksson J, Sima E, Al-Saffar AK, Ahl D, Phillipson M, Webb DL, Sundbom M et al.: Salmonella enterica Serovar Typhimurium exploits cycling through epithelial cells to colonize human and murine enteroids.mBio 2021, 12 This study uses three dimensional murine and human enteroids as a model to study the infection cycle of S.Tm.The study shows that S.Tm growth in the enteroid lumen is fueled by TTSS-1-dependent epithelial invasion followed by reemergence of intracellular S.Tm from expelled IECs.The results are complementary to Chong et al. [114].78.Holly MK, Han X, Zhao EJ, Crowley SM, Allaire JM, Knodler LA, Vallance BA, Smith JG: Salmonella enterica infection of murine and human enteroid-derived monolayers elicits differential activation of epithelium-intrinsic inflammasomes.Infect Immun 2020, 88. 79.Lawrence AE, Abuaita BH, Berger RP, Hill DR, Huang S, Yadagiri VK, Bons B, Fields C, Wobus CE, Spence JR et al.: Salmonella enterica Serovar Typhimurium SPI-1 and SPI-2 shape the global transcriptional landscape in a human intestinal organoid model system.mBio 2021, 12 In this study, human intestinal organoids were infected with S.Tm to explore the host transcriptional response.The results suggest that pro-inflammatory gene expression is largely independent of TTSS-1 mediated invasion, but that invasion affects cell cycle and DNA repair responses.80. Samperio Ventayol P, Geiser P, Di Martino ML, Florbrant A, Fattinger SA, Walder N, Sima E, Shao F, Gekara NO, Sundbom M et al.: Bacterial detection by NAIP/NLRC4 elicits prompt contractions of intestinal epithelial cell layers.Proc Natl Acad Sci U S A 2021, 118 [71]] mice) have provided new insights[54 ,69,70 ].Simultaneous deletion of SopB/E/E2/SipA in S.Tm SL1344[54 ], or of SopA/B/E2/SipA in S.Tm 14028 (which naturally lacks SopE)[70 ]abolished IEC invasion.In partial similarity to early results from bovine ligated ileal loops[71], in neonate mice, complementation of SopA/B/E2/SipA S. Tm 14028 with either SipA, SopE, or SopE2 partially rescued the IEC invasion defect, whereas SopA or SopB failed to do so.Moreover, triple mutants that retain exclusively SipA or SopE2 were still invasion-proficient.This indicates important, but redundant, roles of SipA and SopE2 for S.Tm 14028 invasion of neonate IECs