Reprogramming of microRNA expression via E2F1 downregulation promotes Salmonella infection both in infected and bystander cells

Cells infected with pathogens can contribute to clearing infections by releasing signals that instruct neighbouring cells to mount a pro-inflammatory cytokine response, or by other mechanisms that reduce bystander cells’ susceptibility to infection. Here, we show the opposite effect: epithelial cells infected with Salmonella Typhimurium secrete host factors that facilitate the infection of bystander cells. We find that the endoplasmic reticulum stress response is activated in both infected and bystander cells, and this leads to activation of JNK pathway, downregulation of transcription factor E2F1, and consequent reprogramming of microRNA expression in a time-dependent manner. These changes are not elicited by infection with other bacterial pathogens, such as Shigella flexneri or Listeria monocytogenes. Remarkably, the protein HMGB1 present in the secretome of Salmonella-infected cells is responsible for the activation of the IRE1 branch of the endoplasmic reticulum stress response in non-infected, neighbouring cells. Furthermore, E2F1 downregulation and the associated microRNA alterations promote Salmonella replication within infected cells and prime bystander cells for more efficient infection.

M icroRNAs (miRNAs), due to their instrumental role as post-transcriptional regulators of gene expression 1 , are increasingly recognized as major players in the interaction between host cells and bacterial pathogens 2 . Although miRNAs were initially shown to be part of the host immune response to fight infection, emerging evidence demonstrates that the host miRNA pathway can also be subverted by bacterial pathogens for their own benefit.
Advances in RNA sequencing have contributed to revealing that infections by various bacterial pathogens induce extensive changes of the host miRNome. Along this line, it has also been shown that even closely related bacterial species can lead to distinct host miRNome changes 3 (e.g. virulent vs. avirulent Mycobacterium tuberculosis 4 or Francisella tularensis subspecies tularensis vs. subspecies novicida 5 ). Conversely, a common set of miRNAs was shown to be consistently regulated upon infection with six bacteria 6 . Notwithstanding, the molecular mechanisms underlying most of the described miRNA regulations remain poorly understood. Furthermore, the comparative analysis of miRNA profiles in cells with internalized bacterial and neighboring non-infected bystander cells, as well as the evaluation of their potential relevance for infection, has yet to be investigated.
It is becoming increasingly clear that cells infected with pathogens can signal to bystander cells. In this context, cumulating evidence shows that infected cells have the ability to alert and instruct bystander cells to mount a pro-inflammatory cytokine response, thus contributing to clearing infections more effectively. The process of bystander cell activation is indeed emerging as a crucial evolutionary adaptation in metazoan innate immunity, relevant to viral, parasite, and bacterial infections, in a broad range of hosts [7][8][9] . Although the relevance of bystander innate immune responses has been mostly described for viral infections, bystander cell activation by bacterial pathogens has also been described. A seminal report by Kasper et al. 10 revealed that infection by Shigella flexneri activates the production of the neutrophil chemotactic factor interleukin-8 (IL-8) in bystander cells, whereas its expression is impaired in infected host epithelial cells. IL-8 production is activated through a mitogen-activated protein kinase (MAPK)-dependent mechanism in response to a yet unidentified molecular signal transmitted to adjacent cells via gap junctions. Increased IL-8 production in bystander cells has also been observed in response to infection by the bacterial pathogens Salmonella Typhimurium and Listeria monocytogenes 10 . Paracrine bystander cell activation mediated by soluble signals has also been reported in response to bacterial pathogen infection. An interesting study revealed that during Listeria infection the chemokines CXCL2 and CXCL5 are primarily produced by epithelial bystander cells, rather than by infected cells where their expression is dampened 11 . In this case, reactive oxygen intermediates mediate the paracrine activation of bystander cells. Similarly, during Chlamydia trachomatis infection the response of infected cells to interferon-γ is blocked, whereas that of bystander cells is unhindered contributing to limiting the spread of infection 12 . Along the same line, we have shown that Shigella flexneri infection induces plasma membrane remodeling in infected and bystander cells, through the activation of the acid sphingomyelinase and strong accumulation of ceramide at the cell surface 13 . These changes of membrane composition determine a depletion of permissive bacterial binding sites, constituting a cell-autonomous defense mechanism that protects cells from infection by non-motile bacteria.
As described above, the study of bystander cells in the context of infection by bacterial pathogens has mainly focused on their role in host defense mechanisms, particularly innate immunity. However, given the constant evolutionary pressure of host and pathogen to obtain a competitive advantage, it is conceivable that both have evolved alternative strategies to exploit bystander cell functions.
Here, we analyzed the role of the transcription factor E2F1 in the regulation of miRNAs upon infection with the bacterial pathogen Salmonella enterica serovar Typhimurium (hereafter, Salmonella). E2F1 is a central player in numerous processes, including cell cycle progression 14 , DNA damage response 15 , senescence 16 , apoptosis 17 , and metabolism 18 . Indeed, E2F1 binds to hundreds of promoter regions of genes involved in numerous cellular pathways [19][20][21] , including miRNA genes [22][23][24][25][26] . We have previously shown that E2F1 expression is decreased upon Salmonella infection 27 . We have now determined that E2F1 is a pivotal player in the regulation of the miRNome during Salmonella infection, both in infected and bystander cells. Mechanistically, we show that the secretome of Salmonella-infected cells is sufficient to induce E2F1 and miRNA regulation in naive cells, through the activation of the endoplasmic reticulum stress response pathway, involving intercellular signaling mediated by secreted HMGB1 and its interaction with the transmembrane receptor RAGE (receptor for advanced glycation end products). Collectively, this work reveals that the downregulation of E2F1 and consequently of miRNAs promotes infection, by promoting on one hand the bacterial replication within infected cells as well as by priming bystander cells for efficient Salmonella infection.

Results
E2F1 transcription factor is a major player in miRNA regulation during Salmonella infection. To investigate the importance of E2F1 downregulation to the miRNA expression changes occurring during Salmonella infection, we have performed a comparative analysis of miRNA expression datasets obtained from small RNA sequencing of Salmonella-infected HeLa cells and E2F1 knockdown cells. HeLa cells are an epithelial cell line widely used to study infection by bacterial pathogens, including Salmonella. Upon Salmonella infection, 258 miRNAs (47% of the total detected miRNAs) were downregulated and 112 miRNAs (20% of total) were upregulated (at least 1.4-fold compared to mock-treated cells; Fig. 1a). In E2F1 knockdown cells, 245 miR-NAs (44% of total) were downregulated and 41 miRNAs (7.5% of total) were upregulated (at least 1.4-fold compared to cells transfected with control siRNA; Fig. 1a). Most importantly, the comparison of the miRNA expression profiles revealed a clear trend for miRNAs decreased in E2F1 knockdown cells to also be decreased in Salmonella-infected cells (Fig. 1b), with 144 (26% of total; 56% of Salmonella downregulated) miRNAs downregulated in both conditions. Of note, only miRNAs with ≥20 reads in both control conditions were considered for the analysis (546 miR-NAs). Efficient knockdown of E2F1 was confirmed by western blot (Supplementary Fig. 1a). Overall, these results indicate that the transcription factor E2F1 plays a critical role in the regulation of miRNA expression during Salmonella infection.
These results prompted us to further characterize the regulation of E2F1 expression during infection. Firstly, we determined the kinetics of E2F1 modulation upon Salmonella infection, in samples collected at early (1 and 4 hpi), intermediate (8 hpi), and late stages (20 hpi) of infection. This analysis revealed that E2F1 protein levels are decreased at 4 hpi, gradually lowering with the progression of infection (8 and 20 hpi; Fig. 1c). E2F1 mRNA levels were also decreased at 4, 8, and 20 hpi ( Supplementary  Fig. 1b). A similar decrease of E2F1 levels was observed in Salmonella-infected HCT-8 cells ( Supplementary Fig. 1c, d), a colon epithelial cell line also used as a model for bacterial infection studies.
Additionally, we observed that E2F1 expression remains unchanged in HeLa cells treated with purified Salmonella LPS ( Supplementary Fig. 1e) or heat-killed Salmonella (Supplementary Fig. 1f). Indeed, the decrease of E2F1 was dependent on Salmonella internalization, since a Salmonella mutant strain defective in invasion (ΔSPI-1 strain) was not able to elicit this phenotype, whereas infection with a strain able to invade but defective in intracellular replication (ΔSPI-2 strain) led to E2F1 downregulation ( Supplementary Fig. 1f and ref. 27 ). Finally, we determined that Shigella flexneri or Listeria monocytogenes infection do not affect E2F1 expression, either at early or late times post-infection ( Supplementary Fig. 1g, h). These results demonstrate that the strong regulation of E2F1 expression is not a broad and/or unspecific host cell response to bacterial infection, but it is rather restricted/specific to Salmonella infection.
Regulation of E2F1 and miRNA expression by Salmonella infection occurs in infected as well as in bystander cells. Given the strong downregulation of E2F1 expression observed upon Salmonella infection (30% of control at 20 hpi), and considering that only about 25-30% of the cells are infected in the experimental conditions analyzed, we hypothesized that E2F1 downregulation could occur not only in infected cells but also in the bystander cell population. Thus, we evaluated E2F1 expression in    the fraction of HeLa cells with internalized bacteria (Salmonella positive), as well as in bystander cells, separated by fluorescenceactivated cell sorting (FACS; Fig. 1d). Interestingly, we observed a strong decrease in E2F1 protein and mRNA levels in both cell populations (Fig. 1e, f). Furthermore, we analyzed miRNome changes in these cell populations-of the 563 detected miRNAs (≥20 reads in mock-treated cells), 186 (i.e. 33%) were commonly downregulated in Salmonella-positive and bystander populations (at least 1.4-fold compared to mock-treated cells; Fig. 1g, h). A panel of eight miRNAs (miR-15a-5p, miR-15b-5p, miR-16-5p, miR-22-3p, miR-421, miR-744-5p, let-7i-3p, and let-7i-5p) was selected for further validation by quantitative reverse transcription PCR (qRT-PCR). The selection of these miRNAs was motivated by the miRNA expression datasets described above, our analysis of miRNA function during Salmonella infection 27,28 , as well as on having been previously described as E2F1-dependent miRNAs 22,23,25 . In perfect agreement with E2F1 downregulation (Fig. 1e, f), we observed that the expression of these miRNAs was decreased both in Salmonella-positive and bystander cell populations (Fig. 1i). Of note, we confirmed that the expression of miR-15a-5p, miR-15b-5p, miR-16-5p, miR-22-3p, miR-421, miR-744-5p, let-7i-3p, and let-7i-5p is also decreased upon E2F1 knockdown of naive cells (Fig. 1j). In addition, we showed that the levels of these miRNAs are restored to normal levels in cells overexpressing EGFP-E2F1 and then infected with Salmonella ( Supplementary Fig. 2a, b), providing further evidence of a causal relationship between the levels of these miRNAs and E2F1.
Taken together, these data show that the E2F1 downregulation and the consequent decrease of miRNA expression occur both in cells with internalized Salmonella, as well as in the bystander cell population.
The secretome of Salmonella-infected cells is sufficient to elicit E2F1 and miRNA regulation in naive cells. Considering that extracellular Salmonella could not trigger per se E2F1 downregulation in epithelial cells (Supplementary Fig. 1f and ref. 27 ), we reasoned that the reduction of E2F1 expression in bystander cells should likely be prompted by a signaling mechanism between the cells with internalized bacteria and the neighboring bystander cells, presumably mediated by the secretion of extracellular factor(s). In agreement with this hypothesis, treatment of naive HeLa cells with the secretome of Salmonella-infected cells (collected at 20 hpi; Fig. 2a) resulted in a strong decrease of E2F1 protein levels, which was particularly noticeable at the later times post-treatment (14 and 24 h; Fig. 2b), when compared to treatment with secretome of mock-treated cells. Similar results were obtained at the mRNA level ( Supplementary Fig. 2c). The addition of the supernatant of Salmonella cultures (WT, ΔSPI-1, or ΔSPI-2 strains) to naive cells did not affect E2F1 levels (Supplementary Fig. 2d), demonstrating that the secretome of infected cells, and not of the bacteria, is responsible for triggering the observed effects on bystander cells.
Global analysis of miRNA expression revealed that treatment of naive cells with the secretome of Salmonella-infected cells results in a reduction of expression of 278 miRNAs (at least 1.4fold compared with cells treated with secretome of mock-treated cells; 51% of the 542 detected miRNAs) (Fig. 2c). Importantly, 154 (28% of total; 59% of Salmonella downregulated) miRNAs were commonly downregulated in naive cells treated with the secretome of Salmonella-infected cells, and in cells infected with Salmonella (Fig. 2c, d). Similarly, 158 miRNAs were commonly downregulated in bystander cells and in naive cells treated with the secretome of Salmonella-infected cells, which corresponds to 64% and 57% of the miRNAs downregulated in either of them, respectively ( Supplementary Fig. 2e, f). In agreement with this, treatment of naive cells with the secretome of Salmonella-infected cells resulted in decreased expression of the selected panel of E2F1-dependent miRNAs (Fig. 2e). This very strong overlap strongly supports the existence of a common mechanism (E2F1 downregulation) that orchestrates the downregulation of miRNAs in both experimental conditions. Nevertheless, differences between the miRNome profiles of bystander cells and cells treated with secretome can be observed, likely explained by: (i) bystander cells, contrary to cells treated with the secretome, are initially exposed to the bacteria inoculum, even if the bacteria are not internalized or are eliminated from inside the cells in the course of the infection; (ii) the bystander cells are exposed to increasing concentrations of the factors secreted to the extracellular space by infected cells during the 20 h of infection, whereas the naive cells treated with the secretome are exposed to the full secretome for 24 h. As such, the existence of other mechanisms acting on miRNA expression, in addition to E2F1 regulation, cannot be excluded. Regarding the higher number of miRNAs downregulated in E2F1 knockdown cells (245 miRNAs) compared with the miRNAs commonly downregulated in naive cells treated with the secretome and infected cells (154 miRNAs), this is arguably explained by the fact that E2F1 knockdown cells accumulated for a longer time the consequences of decreased levels of the transcription factor. In fact, E2F1 knockdown cells were collected 48 h after siRNA transfection, whereas the Salmonella-infected cells were collected at 20 hpi and the cells treated with the secretome were collected 24 h post-treatment. This is particularly pertinent taking into consideration that miRNAs have long half-lives 29,30 .
Overall, these results demonstrate that the secretome of Salmonella-infected cells is sufficient to elicit E2F1 and miRNA downregulation of naive cells, explaining a major fraction of the miRNA regulation observed in bystander cells during Salmonella infection. E2F1 downregulation is triggered by activation of the endoplasmic reticulum stress response, induced by infection or by the secretome of infected cells. To understand whether the decrease of E2F1 expression prompted by the secretome of Salmonella-infected cells was due to the secretion of peptide(s)/protein (s), we performed a thorough digestion of the secretome with proteinase K. Interestingly, proteinase K digestion completely impaired the ability of the secretome to trigger E2F1 downregulation (Fig. 2f). In face of this observation, we next analyzed the protein composition of the secretome by mass spectrometry. The analysis was performed for the secretome of infected cells collected at 14 hpi to minimize changes in the secretome triggered by cell death/cytotoxicity associated with the late stages of Salmonella infection. The majority of the identified proteins were significantly enriched in the secretome of Salmonella-infected cells compared to mock-treated cells (722 proteins; Fig. 2g and Supplementary Data 2). Functional analysis of these proteins revealed enrichment, among others, for proteins associated with the endoplasmic reticulum stress response (ER-stress response, a.k.a. UPR, unfolded protein response; Supplementary Data 3), including asparagine synthetase (ASNS) 31 , high mobility group box1 (HMGB1) 32 , calreticulin 33 , glucose-regulated protein 94 (GRP94, HSP90B1) 34,35 , glucose-regulated protein 75 (GRP75, HSPA9) 35 , binding immunoglobulin protein (BiP, GRP78, HSPA5) 35,36 , and heat-shock protein 110 (HSP110) 37 . The increase of ASNS and HMGB1 levels in the secretome of Salmonella-infected cells was validated by western blot (Fig. 2h). Only one Salmonella protein was identified in the secretome of infected cells-glycerol-3-phosphate transporter periplasmic binding protein (gene ugpB). To elucidate the contribution of the ER-stress response to the decrease of E2F1 and miRNA expression during Salmonella infection, we first examined whether this pathway is activated upon Salmonella infection and/or upon treatment of naive cells with the secretome of Salmonella-infected cells. We observed that ER-stress response is indeed activated in both conditions, as assessed by the increased expression of the ER chaperone BiP 36 (Fig. 2i, m) and phosphorylation of the ER-stress sensor IRE1 (ref. 38 ) (Fig. 2j, n and Supplementary Fig. 2g, h). IRE1 phosphorylation occurs very early on during infection (1 hpi) or secretome treatment (1 h), and although phosphorylation decreases with time, it is maintained at high levels compared to control cells (Fig. 2j, n and Supplementary Fig. 2g, h). The increase of BiP occurred later than IRE1 phosphorylation (starting at 8 hpi or 8 h post-treatment with the secretome). This is consistent with the increase of BiP protein levels occurring only after prolonged ER stress 39 . Interestingly, ER-stress response was only triggered by infection or treatment with the secretome of invasive Salmonella strains (WT and ΔSPI-2), but not with the non-invasive strain ΔSPI-1 ( Supplementary Fig. 2i, j), suggesting a causal relationship between ER-stress response and E2F1/ miRNA regulation. Of note, Salmonella infection or treatment with the secretome of Salmonella-infected cells did not activate the two additional ER-stress sensors-ATF6 (Figs. 2k, o) and PERK (Fig. 2l, p).
The effect of ER-stress response as a trigger for E2F1 regulation was confirmed by the strong decrease of E2F1 levels observed in HeLa cells treated with well-described inducers of ER stress, specifically DTT and tunicamycin ( Supplementary Fig. 3a), in agreement with a previous report 40 . Importantly, knockdown of IRE1 (siRNA pool and three independent siRNAs) or inhibition of its kinase activity by the chemical inhibitor KIRA6 (ref. 41 ) strongly impaired the E2F1 downregulation prompted by Salmonella infection (Fig. 3a, c and Supplementary Fig. 3b, d) or by treatment of naive cells with the secretome of Salmonellainfected cells (Fig. 3b, d and Supplementary Fig. 3c, e). Moreover, preventing the activation of the IRE1 branch of the UPR pathway also compromised the downregulation of the selected miRNAs (miR-15a-5p, miR-15b-5p, miR-16-5p, miR-22-3p, miR-421, miR-744-5p, let-7i-3p, and let-7i-5p) (Fig. 3e, f and Supplementary Fig. 3f, g). Interestingly, ASNS levels were dramatically reduced in the secretome of IRE1 knockdown cells infected with Salmonella, when compared to the secretome of control-infected cells ( Supplementary Fig. 3h). This result demonstrates that in   Fig. 3o), which is dependent on IRE1 endonuclease activity 42 . Finally, we confirmed that activation of the ER-stress response pathway induced by treatment with the secretome of Salmonellainfected cells was impaired in IRE1 knockdown cells or KIRA6treated cells ( Supplementary Fig. 3p, q), as evaluated by BiP expression. The Salmonella effector protein SlrP has been shown to partially co-localize and interact with the ER chaperone ERdj3, reducing the interaction of ERdj3 with denatured proteins, suggesting that it could increase the accumulation of unfolded proteins in the ER, and thus activate ER-stress response 43 . However, no differences in IRE1 activation and E2F1 downregulation were observed between infection with WT or ΔslrP Salmonella strains ( Supplementary Fig. 4a, b), excluding this effector protein as the trigger for ER-stress activation.
Collectively, these results demonstrate that the IRE1 branch of the ER-stress response pathway is triggered during Salmonella infection of epithelial cells and that the secretome of Salmonellainfected cells is sufficient to induce this phenotype in naive cells. Importantly, the activation of the IRE1 pathway triggers E2F1 and miRNA downregulation.
IRE1 activation induced by infection or by the secretome of infected cells leads to JNK pathway activation and subsequent E2F1 proteasomal degradation. Following activation, the IRE1 endoribonuclease activity processes the mRNA encoding the transcription factor XBP1, leading to the expression of an active transcription factor (XBP1s) that transactivates a subset of UPR target genes 44 . IRE1 also degrades specific mRNAs through a process denominated regulated IRE1-dependent decay (RIDD) [45][46][47] . In addition, it is well described that the TNF receptor-associated factor 2 (TRAF2) protein binds to phosphorylated IRE1, facilitating the recruitment and activation of the JNK MAPK pathway 48 .
Monitoring of downstream targets of IRE1 activation, specifically splicing of XBP1 and expression levels of typical IRE1/XBP1 targets (e.g. ERdj4/DNAJB9, EDEM1, SEC24D) by qRT-PCR in Salmonella-infected cells and naive cells treated with secretome of Salmonella-infected cells revealed differential splicing of XBP1. Interestingly, XBP1 splicing and the consequent activation of downstream targets did not occur in Salmonellainfected cells ( Supplementary Fig. 4c, d); conversely, both XBP1 splicing and activation of its target genes occurred in naive cells treated with secretome of infected cells (Supplementary 4f, g). The result in Salmonella-infected cells is in apparent contradiction with our observation that IRE1 activation occurs in both experimental conditions (demonstrated by IRE1 phosphorylation; Fig. 2j, n). However, we determined that the expression of the RNA ligase RTCB, responsible for the XBP1 exon ligation [49][50][51] , is markedly decreased in Salmonella-infected cells, while it is maintained in cells treated with the secretome ( Supplementary Fig. 4e, h). These data suggest that, although IRE1 activation occurs in both conditions, XBP1 splicing is impaired in Salmonella-infected cells presumably due to altered expression of the cytoplasmic splicing machinery.
We have clearly demonstrated that IRE1 activation is crucial for E2F1 decreased expression during Salmonella infection and secretome treatment (both in IRE1 knockdown cells and cells treated with the IRE1 kinase inhibitor KIRA6; Fig. 3c, d and Supplementary Fig. 3d, e). However, the lack of XBP1 splicing in Salmonella-infected cells ( Supplementary Fig. 4c, d), as well as the observation that the inhibition of IRE1 endonuclease activity, using the 4µ8C inhibitor 52 , did not affect the extent of E2F1 downregulation observed upon Salmonella infection or by treatment with the secretome of Salmonella-infected cells ( Supplementary Fig. 4i, j) led us to hypothesize that IRE1 could act through a mechanism independent of its endonuclease activity. Of note, the efficiency of the 4µ8C inhibitor was confirmed by quantifying XBP1 splicing in cells treated with tunicamycin ( Supplementary Fig. 4k).
According to the literature, the activation of the JNK pathway by the IRE1 pathway is independent of its endonuclease activity 48 . It is already described that the JNK pathway is activated during Salmonella infection of epithelial cells 53,54 , independently of innate immune receptors such as TLRs. Our data confirm that this occurs in a time frame compatible with IRE1 activation (Fig. 3g). Moreover, JNK activation also occurs in cells treated with the secretome of Salmonella-infected cells       (Fig. 3i). Adding to the importance of JNK, we also demonstrate that JNK knockdown prevents the E2F1 downregulation observed during Salmonella infection or secretome treatment (Fig. 3h, j and Supplementary Fig. 4l-n). Moreover, in both conditions JNK activation is not detected in IRE1 knockdown cells (Fig. 3k, m). Together, these results demonstrate that JNK activation mediates the E2F1 regulation during Salmonella infection/secretome treatment.
Regarding the mechanism of degradation of E2F1 upon Salmonella infection or secretome treatment, we tested the relevance of proteasome degradation to this process. Inhibition of the proteasome by MG132 treatment prevented the E2F1 downregulation elicited by Salmonella infection or secretome treatment (Fig. 3l, n). Of note, it is described that E2F1 regulates its own promoter activity 55 and therefore the decrease in the E2F1 protein level likely explains the decrease observed at the mRNA level ( Supplementary Figs. 1b, d and 2c).
Overall, our data indicate that IRE1 activation upon Salmonella infection or secretome treatment leads to JNK activation, which phosphorylates 56,57 and thus inactivates E2F1. The inactive transcription factor is then degraded by the proteasomal machinery.
Secretome of Salmonella-infected cells leads to RAGEdependent ER-stress response pathway activation. Considering that the protein HMGB1 is highly enriched in the secretome of Salmonella-infected cells (Fig. 2g, h) and that this protein has been shown to activate the ER-stress response by a mechanism dependent on its interaction with RAGE 58-60 , we evaluated the involvement of RAGE in the downregulation of E2F1 (Fig. 4a). Firstly, we confirmed that activation of the ER-stress response pathway elicited by treatment with the secretome of Salmonellainfected cells was impaired in RAGE knockdown cells, as evaluated by IRE1 phosphorylation status and/or BiP expression  (Fig. 3f). e Schematic representation of the experimental design for the treatment of naive cells with the secretome of HMGB1-depleted cells. Secretome of cells transfected with control or HMGB1 siRNAs and Salmonella-infected or mocktreated were used to treat naive cells. f, g IRE1 phosphorylation (f) and E2F1 protein levels (g), determined by western blot, in naive HeLa cells treated with the secretome of Salmonella-infected or mock-treated HMGB1-depleted cells for 1 or 24 h, respectively. h, i IRE1 phosphorylation (h) and E2F1 protein levels (i), determined by western blot, in HeLa cells treated with recombinant HMGB1 protein for 1 or 24 h, respectively. Naive cells treated with the secretome of Salmonella-infected cells or mock-treated cells are shown for comparison. Salmonella infection was performed at MOI 100. Results are shown as mean ± s.e.m. of n = 5 (c, d, g) biologically independent experiments; western blots are representative of n = 3 (b, f, h, i) or n = 5 (c, g) biologically independent experiments; *P < 0.05 and ***P < 0.001 (statistical analysis is detailed in Supplementary Data 1); Source data are provided as a Source Data file.
( Fig. 4b and Supplementary Fig. 5a). Importantly, the knockdown of RAGE also prevented the E2F1 downregulation upon treatment with the secretome of infected cells (Fig. 4c). Concurring with these results, the downregulation of the panel of selected miRNAs upon treatment with the secretome of Salmonellainfected cells was impaired in RAGE knockdown cells (Fig. 4d). Moreover, IRE1 activation and E2F1 downregulation did not occur in naive cells treated with the secretome of Salmonellainfected HMGB1-depleted cells (Fig. 4e-g). Accordingly, treatment of naive cells with recombinant HMGB1 induced both IRE1 activation and consequent decrease of E2F1 levels (Fig. 4h, i). Efficient knockdown of RAGE or HMGB1 was confirmed by western blot (Supplementary Fig. 5b, c).
Overall, these results demonstrate that the ER-stress response activation and consequent E2F1 regulation induced by the secretome of Salmonella-infected cells is dependent on the engagement of RAGE, elicited by the HMGB1.

Downregulation of E2F1 promotes Salmonella infection.
Considering the magnitude of E2F1 downregulation and its specificity during Salmonella infection, we pondered whether it could impact the infection process. Indeed, E2F1 knockdown increased Salmonella infection, specifically the bacterial invasion (1 hpi) and replication (20 hpi), compared to cells transfected with control siRNA (Fig. 5a-c). Moreover, cells pre-treated with secretome from Salmonella-infected cells also presented increased bacterial invasion and replication when compared to cells pretreated with secretome from mock-treated cells (Fig. 5d-f).
Conversely, preventing E2F1 downregulation by blocking the activation of the IRE1 pathway (KIRA6 treatment) inhibited Salmonella infection (Fig. 5g, h) and thwarted the increase in Salmonella infection elicited by pre-treatment with the secretome of Salmonella-infected cells (Fig. 5d-f). Moreover, when the ERstress response was prompted independently of infection, by tunicamycin or DTT treatment, an increase of Salmonella invasion and replication was observed when compared to control cells (Fig. 5g, i).
Together, these results show that the decrease of E2F1 expression upon Salmonella infection that ensues from activation of the ER-stress response pathway is required to sustain a productive bacterial replication inside host cells, as well as to promote infection of bystander cells.
Salmonella infection triggers IRE1 activation and E2F1 downregulation in vivo. Having shown that Salmonella infection elicits a strong downregulation of E2F1 in vitro with important consequences for the outcome and dissemination of infection, we examined whether E2F1 regulation could be observed in vivo. We used a piglet model of Salmonella infection and compared E2F1 expression in mucosal samples of ileum and colon tissues of control or Salmonella-infected animals, collected at 2 and 6 days post-infection (Fig. 6a). The levels of E2F1 were strongly decreased in the ileum and colon samples from Salmonellainfected animals compared to control animals (Fig. 6b, c). Conversely, and in agreement with the data obtained in in vitro infection models, IRE1 phosphorylation was significantly increased in tissues collected from infected animals (Fig. 6d, e).
Overall, these results demonstrate that also in an animal model, Salmonella infection results in a strong downregulation of E2F1 expression, likely triggered by activation of the IRE1 branch of the ER-stress response pathway, as well as the consequent downregulation of E2F1-dependent miRNAs further supporting their relevance for the infection process.

Discussion
Although miRNAs are increasingly recognized for their crucial role in the interaction between host and bacterial pathogens, the exact mechanisms underlying their regulation during infection remain poorly understood. In this study, we identified and characterized the transcription factor E2F1 as a major player in the regulation of miRNAs upon Salmonella infection of epithelial cells (Fig. 7). On a global scale, E2F1-dependent miRNAs account for more than 50% of the miRNAs downregulated during Salmonella infection. Salmonella infection elicits a strong decrease of E2F1 expression, which appears to be restricted to this pathogen, not occurring in response to infection by other intracellular pathogens (Shigella flexneri and Listeria monocytogenes). Of note, E2F1 downregulation is dependent on bacterial invasion rather than on extracellular Salmonella, given that purified LPS, heatkilled Salmonella, and the invasion-deficient ΔSPI-1 Salmonella mutant failed to induce regulation. These results clearly establish E2F1 regulation upon Salmonella infection as a specific effect, rather than a broad and/or unspecific response to bacterial components/infection.
Interestingly, we demonstrate that E2F1 downregulation and subsequent miRNA regulation is not limited to infected cells, but it also occurs, and to a remarkably comparable extent, in bystander cells. This is particularly interesting in light of previous studies focused on various intracellular bacterial pathogens that revealed that, far from being passive observers, bystander cells play a critical role in controlling infection [7][8][9] . Noteworthy, our study proposes a shift of the current paradigm by demonstrating that changes elicited in bystander cells can positively contribute to the progression of infection. Indeed, we demonstrate that E2F1 downregulation in bystander cells renders these cells more permissive to infection, by promoting Salmonella invasion, and intracellular replication. The net effect of E2F1 and miRNA downregulation resulting in increased Salmonella infection is likely explained by the fact that several of these miRNAs have been shown to counteract Salmonella infection. Based on a genome-wide microscopy-based functional screening approach we have identified miRNAs controlling infection of epithelial cells by Salmonella 27,28 . Among others, the miRNAs selected for validation in the present study, namely miR-15a-5p, miR-15b-5p, miR-16-5p, miR-22-3p, miR-421, miR-744-5p, let7i-3p, and let-7i-5p, are among the miRNAs strongly inhibiting Salmonella infection. Of note, miRNA independent effects of E2F1 downregulation may also contribute to the observed increase of Salmonella infection.
An aspect of particular relevance is the trigger/signal that mediates the signaling between infected and bystander cells, ultimately leading to E2F1 and miRNA downregulation. The mechanism of intercellular communication is determinant to the range of action of the signal: signals transmitted only to cells in the immediate vicinity of the infected cell likely rely on direct cellto-cell communication via gap junctions, whereas long-range communication likely reflects the release of soluble signals or secretion of exosomes or microvesicles. Taking into consideration that the extent of E2F1 downregulation was comparable in infected and bystander cells, and the relatively low percentage of Salmonella-infected cells (25-30%) in the experimental conditions tested, we reasoned that a long-range communication mechanism was likely involved. This was unequivocally demonstrated by showing that the secretome of Salmonella-infected cells, in particular proteins present therein, is sufficient to trigger E2F1 downregulation and consequent miRNA regulation in naive cells.
Analysis of the protein composition of the secretome by mass spectrometry revealed a dramatic enrichment of protein abundance in the secretome of Salmonella-infected cells, when compared to that of mock-treated cells. Importantly, there were several proteins previously shown to be involved in or secreted upon activation of the ER-stress response, including asparagine synthetase (ASNS) 31   secretome of infected cells. Based on this, we hypothesize that ERstress response could be induced in Salmonella-infected cells. It has been well described that ER-stress response promotes inflammatory pathways and is involved in a wide range of related pathologies, such as diabetes, obesity, inflammatory bowel disease, and atherosclerosis 61,62 . Recent evidence has also implicated this pathway in the infection by bacterial pathogens [63][64][65][66] . Although the activation or inhibition of the ER-stress response has been reported for several bacterial pathogens (e.g. Brucella abortus 67 , Chlamydia pneumoniae 68 , group A Streptococcus 69 , Campylobacter jejuni 70 , Legionella pneumophila 71 ) in most cases its role for pathogenesis remains unclear. Here, we demonstrate that the activation of the ER-stress response, and consequent E2F1 and miRNA downregulation, is beneficial for Salmonella replication and dissemination of infection. Particularly, we showed that ER-stress response is activated both in Salmonellainfected cells and in cells treated with the secretome of infected cells. Most importantly, we demonstrated that E2F1 downregulation is triggered by ER-stress response, establishing a causal relationship between induction of ER-stress response in infected/ bystander cells and increased Salmonella infection.
Mechanistically, we show that Salmonella infection and treatment with secretome of infected cells specifically activates the ERstress sensor IRE1, whereas the two other ER-stress response branches mediated by ATF6 and PERK are not activated (Fig. 7). Our results are in apparent contradiction with a previous report indicating that ER-stress response, specifically the PERK branch, is triggered during Salmonella infection of mice 72 . This can potentially be explained by the different organisms evaluated, namely mouse tissue 72 versus human cells and piglet tissue (this study), and/or infection procedure/protocol. As such, it is possible that in distinct cellular/organism models other branches of ERstress response pathway are activated upon Salmonella infection. Following IRE1 activation upon Salmonella infection/secretome treatment, our data support a model in which the JNK pathway is activated by IRE1 leading to phosphorylation of E2F1, among other targets. The inactive E2F1 is then degraded by proteasomal degradation (Fig. 7). JNK was previously shown to phosphorylate E2F1, inactivating the transcription factor 56,57 .
Regarding the stimulation of ER-stress response in naive cells upon treatment with the secretome of Salmonella-infected cells, we identified the activation of RAGE transmembrane receptor as the trigger for the induction of ER-stress response. RAGE activation has been previously shown to mediate ER-stress response in various cell types [58][59][60][73][74][75] , upon interaction with various ligands including the protein HMGB1 (refs. [58][59][60] ). In the case of Salmonella infection, our results demonstrate that RAGE activation is engaged by the binding of the protein HMGB1. Indeed, we show that HMGB1 is secreted to the extracellular milieu by Salmonellainfected epithelial cells, and that this secreted protein is both required and sufficient to bystander cell activation. Of note, it has been previously shown that HMGB1 is released to the intestine of piglets infected with Salmonella 76 , corroborating our in vitro observations and our in vivo data showing activation of ER-stress response in intestinal samples of Salmonella-infected piglets.
IRE1 has been shown to cleave precursor miRNAs, specifically those of miR-17, miR-34a, miR-96, and miR-125a, mediating their degradation 77 . As such, it is conceivable that IRE1-mediated degradation directly regulates the levels of these specific miRNAs during infection. However, it is not likely that IRE1-mediated degradation is contributing to the regulation of the high number of miRNAs downregulated during Salmonella infection, given that the effect of IRE1 on miRNAs appears to be specific to motifs present in those particular precursor miRNAs 77 . It is conceivable that for miR-34a and miR-125a-5p, miRNAs that are not significantly changed in E2F1 knockdown cells but that are downregulated during Salmonella infection and secretome treatment, there is a contribution of IRE1-mediated degradation.
In summary, we demonstrate that downregulation of the transcription factor E2F1 and consequent miRNome changes are crucial for Salmonella infection, by promoting Salmonella replication in infected cells and priming bystander cells for more efficient bacterial infection. Moreover, our study challenges the current paradigm by demonstrating that reprogramming of bystander cells during bacterial infection can positively contribute to the dissemination of infection.
For LPS stimulation, cells were treated with LPS purified from Salmonella Typhimurium (Sigma, L6143) at 1 and 10 μg/ml for 24 h.
For treatment with ER-stress-inducing compounds, cells were incubated with DTT (1 mM; Roth, 6908.4) or tunicamycin (0.5 μg/ml; Sigma, T-7765) for 24 h. For infection, cells were treated with DTT and tunicamycin for 24 h prior to infection and during infection (with 10 μg/ml gentamicin). For XBP1 splicing analysis, cells were treated with tunicamycin (2 μg/ml) for 6 h.
For IRE1 kinase activity inhibition, cells were treated with the KIRA6 (0.5 µM; Merck, 5322810001) as follows: for 24 h prior to infection and during the infection (with 10 μg/ml gentamicin); for 24 h during secretome treatment and during the infection (with 10 μg/ml gentamicin); for 24 h during the secretome treatment. For IRE1 endonuclease activity inhibition, cells were treated with 4μ8C (Merck, 412512) for 1 h prior to infection/secretome treatment (50 μM) and during infection (with 10 μg/ml gentamicin) or secretome treatment for 24 h (25 μM). For XBP1 splicing analysis, cells were treated with 4μ8C for 1 h (50 μM) and then simultaneously treated with 4μ8C (25 μM) and tunicamycin (2 μg/ml) for 6 h.  (1), specifically the IRE1 branch, which activates the JNK MAPK pathway (2) leading to E2F1 downregulation (3) and a consequent decrease in expression of a significant number of miRNAs (4). Given that many of these miRNAs were shown to counteract infection, the net outcome of miRNA regulation is an increase of Salmonella intracellular replication (5). Infection of cells by Salmonella leads to increased secretion of proteins to the extracellular milieu (several of which are related to the ER-stress response), including HMGB1 (6). HMGB1 binds to the RAGE receptor at the surface of bystander cells (7), triggering an IRE1-mediated ER-stress response in bystander cells (8). This, in turn, activates the JNK pathway (9) resulting in a decrease of E2F1 expression in bystander cells (10). Echoing the effect in infected cells, E2F1 downregulation decreases the expression of numerous miRNAs (11), with a net effect of priming these cells for Salmonella infection, by favoring bacterial binding to bystander cells, as well as by promoting bacterial internalization and intracellular replication (12).
For inhibition of the proteasome, cells were treated with MG132 (20 μM; Sigma, 474787) during infection (with 10 μg/ml gentamicin) and for 24 h during the secretome treatment.
Bacterial infections. For Salmonella, Shigella, and Listeria infections, overnight cultures were diluted 1:100 in LB (Salmonella or Shigella) or 1:50 in BHI medium (Listeria) and grown at 37°C with shaking until OD 600 2 (Salmonella), OD 600 0.4 (Shigella), or OD 600 0.7 (Listeria). Bacteria were then harvested by centrifugation for 2 min at 12,000g and resuspended in complete medium. Cells were infected with the bacteria at the MOI indicated in the figure/figure legend. After the addition of bacteria, cells were centrifuged at room temperature (RT) for 10 min at 250g (Salmonella or Listeria) or for 15 min at 2000g (Shigella) and incubated at 37°C in a 5% CO 2 humidified atmosphere for 20 min (Salmonella or Listeria) or 15 min (Shigella). Extracellular bacteria were killed by replacing the medium with fresh medium containing 50 μg/ml gentamicin for 30 min. The medium was then exchanged to medium containing 10 μg/ml gentamicin, until analysis.
The fraction of Salmonella-infected cells was determined by flow cytometry and by fluorescence microscopy followed by automated image analysis. In the experimental conditions used for the majority of the experiments in the manuscript, specifically infection of HeLa cells with Salmonella at MOI 100, both analyses show that the percentage of infected cells is typically 25-30%.
For stimulation with heat-killed Salmonella, bacteria were prepared as described above, except that before infection, bacteria were washed twice in PBS, incubated for 2 h at 80°C, and then resuspended in complete medium. Treatment was performed under the same conditions as for live bacteria and the cells were collected 20 h post-treatment.
For treatment with bacterial secretome, Salmonella was grown in LB medium as described above until OD 600 2. Bacterial culture was centrifuged for 2 min at 12,000g, the supernatant was harvested, centrifuged for 5 min at 12,000g and then the supernatant was filtered through 0.22 µm syringe filters (Millipore, SLGV033RB). Cfu assays confirmed the absence of Salmonella in the secretome samples after processing. Treatment was performed in HeLa cells seeded in 24-well plates at a dilution of 1:5 of bacterial supernatant in cell culture medium for 24 h.
To quantify bacterial invasion or intracellular replication by cfu assays, cells were washed three times with PBS and lysed with PBS containing 0.1% Triton-X-100. Cell lysates were serially diluted in PBS and plated on LB (Salmonella or Shigella) or BHI plates (Listeria).
Secretome collection and cell treatment. For secretome collection, HeLa cells were seeded in six-well plates and were mock-treated or infected 48 h later with Salmonella (MOI 100) as described above, except that the medium exchange after antibiotic treatment was done with 2 ml of serum-free medium containing 10 μg/ ml gentamicin. At 20 hpi, the secretome was collected and centrifuged for 10 min at 300g at 4°C. The supernatants were filtered through 0.22 µm syringe filters (Millipore, SLGV033RB) and supplemented with 10% fetal bovine serum (Biochrom, S0115). Cfu assays confirmed the absence of Salmonella in the secretome samples after processing.
To treat naive cells with the secretome, HeLa cells were seeded at a density of 6.0 × 10 4 or 2.4 × 10 5 cells per well in 24-or 6-well plates, respectively, and treated with the secretome 48 h later (500 µl or 2 ml per well in 24-or 6-well plates, respectively). Twenty-four hours after treatment cells were collected and further processed for western blot or RNA isolation as described below.
For Salmonella infection of cells treated with secretome, HeLa cells were seeded in 24-well plates, treated with the secretome the following day for 24 h, and then infected with Salmonella (MOI 10) as described above.
To digest proteins from the secretome, the secretome collected as described above was treated with Proteinase K-agarose (50 µl/ml of secretome; Sigma, P9290) for 2 h at 37°C.
Cell sorting. HeLa cells were seeded in six-well plates (2.4 × 10 5 cells per well) and infected 48 h later with Salmonella (MOI 100) as described above. At 20 h after infection, cells were trypsinized and collected in PBS. Sorting of the GFP-negative cells (bystander) and GFP-positive (Salmonella positive) cells was performed using a FACSAria III flow cytometer (BD Biosciences) based on the signal intensity from the FITC-A channel. Mock-treated cells were also subjected to the same procedure. Sorted cells (3.0 × 10 5 cells for each fraction) were collected and further processed for western blot and RNA isolation as described below.
E2F1 overexpression. For the overexpression of E2F1, the coding sequence of E2F1 was excised from HA-E2F-1 wt-pRcCMV (Addgene plasmid # 21667, a gift from William Kaelin) and cloned as a C-terminal fusion of the GFP protein in pEGFP-C1 plasmid (Clontech), in the BamHI restriction site.
1 × 10 6 HeLa cells were electroporated with 4 µg of pEGFP-E2F1 or pEGFP-C1 (empty vector), using a Gene Pulser Xcell Electroporation System (Biorad), according to the manufacturer's instructions. After 48 h, cells were infected with Salmonella as described above. Cells were collected for western blot or RNA isolation at 20 hpi.
Protein extracts and western blot. Cells were washed with PBS, lysed in Laemmli's sample buffer, sonicated, and separated in SDS-PAGE followed by western blotting. For phosphorylated JNK samples, cells were lysed in RIPA buffer containing PhosSTOP (Sigma, 4906845001).
For piglet tissue, intestinal sections of ca. 1 cm 2 were processed as previously described 81 . Briefly, mucosa scrapings were homogenized in Laemmli's sample buffer containing Proteinase Inhibitor Cocktail (Sigma-Aldrich, P8340) using prechilled mortar and pestles, sonicated, and centrifuged at 10,000g for 15 min. Protein samples were separated in SDS-PAGE, followed by western blotting.
Of note, long-term treatment with tunicamycin results in a blockade of Nlinked glycosylation, which results in a fully unglycosylated ATF6 detectable as a faster migrating form of ATF6 (approx. 100 kDa). This unglycosylated ATF6 form cannot undergo proteolytic processing into the active ATF6 transcription factor 82,83 .
Quantification of western blots was performed with ImageJ. Uncropped images of immunoblots are included as Source Data.
Sample preparation, mass spectrometry data acquisition, and analysis. Twenty-two milliliters of cell culture medium from HeLa cells mock-treated or infected with Salmonella (MOI 100) were collected at 14 hpi and processed as described above. Secretomes were concentrated by centrifugation using Amicon filters with a 3 kDa cut-off value (Merk-Millipore, UFC900308). Samples from four independent experiments were collected.
Samples were processed for mass spectrometry at the EMBL Proteomics Core Facility. Disulfide bonds were reduced by adding 10 mM final concentration of DTT in 50 mM HEPES, pH 8.5, and incubating the samples for 30 min at 56°C. Subsequently, an alkylation step was performed by adding 20 mM 2-chloroacetamide in 50 mM HEPES, pH 8.5, and incubating the samples in dark for 30 min at RT. Four washes with 50 mM HEPES were performed in Amicon filter tubes (3 kDa cut-off) to concentrate the sample, remove the phenol red, and exchange the sample buffer to optimal digestion conditions. For digestion, volume was adjusted to 100 µL with 50 mM HEPES and trypsin (Promega, V5111) was added at a 1:25 enzyme to protein ratio for overnight digestion at 37°C. Peptides were eluted by centrifugation (10 min, 12,000g), followed by a second elution with 100 µL of 50 mM HEPES. Peptides were labeled with TMT10plex Isobaric Label Reagent (ThermoFisher, 90110) 84 , according to the manufacturer's instructions. For further sample clean-up an OASIS HLB µElution Plate (Waters, 186001828BA) was used. Replicates were subjected to offline high pH reverse phase fractionation, carried out on an Agilent 1200 Infinity HPLC system, equipped with a Gemini C18 column, resulting in six fractions 85 .
Peptides were separated using the UltiMate 3000 RSLC nano LC system (Dionex) fitted with a trapping cartridge (µ-Precolumn C18 PepMap 100, 5 µm, 300 µm i.d. × 5 mm, 100 Å) and an analytical column (nanoEase™ M/Z HSS T3 column 75 µm × 250 mm C18, 1.8 µm, 100 Å, Waters). Trapping was carried out with a constant flow of solvent A (0.1% formic acid in water) at 30 µL/min onto the trapping column for 6 min. Subsequently, peptides were eluted via the analytical column with a constant flow of 0.3 µl/min with an increasing percentage of solvent B (0.1% formic acid in acetonitrile) from 2 to 4% in 4 min, from 4 to 8% in 2 min, then 8 to 28% for a further 96 min, and finally from 28 to 40% in another 10 min. The outlet of the analytical column was coupled directly to a QExactive plus (Thermo) mass spectrometer using the proxeon nanoflow source in positive ion mode. The peptides were introduced into the mass spectrometer QExactive plus via a Pico-Tip Emitter 360 µm OD × 20 µm ID; 10 µm tip (New Objective) and a spray voltage of 2.3 kV was applied. The capillary temperature was set at 320°C. Full mass scan was acquired with mass range 350-1400m/z in profile mode in the FT with a resolution of 70,000. The filling time was set at a maximum of 100 ms with a limitation of 3 × 10 6 ions. Data-dependent acquisition was performed with the resolution of the Orbitrap set to 35,000, with a fill time of 120 ms and a limitation of 2 × 10 5 ions. A normalized collision energy of 32 was applied. A loop count of 10 with count 1 was used and a minimum AGC trigger of 2e 2 was set. Dynamic exclusion time of 30 s was used. The peptide match algorithm was set to "preferred" and charge exclusion "unassigned", charge states 1, 5-8 were excluded. MS2 data were acquired in profile mode 86 .
IsobarQuant 87 and Mascot (v2.2.07; Matrix Science, London, UK) were used to process the acquired data, which was searched against the Uniprot Homo sapiens proteome database (UP000005640) containing common contaminants and reversed sequences. The following modifications were included into the search parameters: Carbamidomethyl (C) and TMT10 (K) (fixed modification), Acetyl (N-term), Oxidation (M), and TMT10 (N-term) (variable modifications). For the full scan (MS1) a mass error tolerance of 10 ppm and for MS/MS (MS2) spectra of 0.02 Da was set. Further parameters were set as follows: trypsin as a protease with an allowance of maximum two missed cleavages; a minimum peptide length of seven amino acids; at least two unique peptides were required for protein identification. The false discovery rate on peptide and protein level was set to 0.01.
The raw output files of IsobarQuant (protein.txt-files) were processed using the R programming language. Only proteins that were quantified with at least two unique peptides were considered for the analysis. Moreover, only proteins that were identified in two out of two mass spec runs were kept. Nine hundred and forty-eight proteins passed the quality control filters. Raw signal-sums (signal_sum columns) were first cleaned for batch effects using the "removeBatchEffect" function of the limma package (v. v3.34.5) 88 and further normalized using vsn (variance stabilization normalization 89 ). The four replicates of the HeLa wild-type condition were normalized separately to maintain the higher protein abundance in this condition. Proteins were tested for differential expression using the limma package. A protein was annotated as a hit with a false discovery rate (fdr) smaller 5% and a fold change of at least 100% and as a candidate with an fdr below 20% and a fold change of at least 50%.
Functional analysis of the proteins significantly increased in the secretome of Salmonella-infected cells compared to mock-treated cells (722 proteins) was conducted using the Ingenuity Pathway Analysis Software (IPA, Ingenuity Systems). Datasets were uploaded into IPA and analyzed for functional enrichment in terms of "Canonical pathways", based on the information in the Ingenuity Knowledge Base. Enrichments were calculated by IPA using multiple hypothesis correction based on the Benjamini-Hochberg method; corrected P values are shown.
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 90 partner repository with the dataset identifier PXD018026 (http://www.ebi.ac.uk/pride/archive/projects/PXD018026).
Confocal microscopy images, shown as maximum projected Z-stack images, were acquired with a Leica SP5 laser scanning confocal microscope (Leica Microsystems).
RNA isolation, PCR, and quantitative real-time PCR. For total RNA isolation, including small RNA fraction, cells were lysed in TRIzol (Invitrogen, 15596026) and RNA was extracted by phenol-chloroform followed by isopropanol precipitation. For the piglet ileum and colon tissue, total RNA was extracted using mirVana miRNA Isolation kit (Invitrogen, AM1561) according to the manufacturer's instructions.
For quantification of gene expression, total RNA was reverse transcribed using hexameric random primers and M-MLV reverse transcriptase (Invitrogen, 28025021), according to the manufacturer's instructions. qRT-PCR was performed using SsoAdvanced Universal SYBR Green Supermix (BioRad, 172-5274) according to the manufacturer's instructions. The primer pairs used are indicated in Supplementary Table 1. Expression was normalized to β-actin or RPL37a.
For the XBP1 splicing assay, PCR was performed using Taq DNA polymerase (NEB, M0273), according to the manufacturer's instructions. The primer pairs used are indicated in Supplementary Table 1. Amplified fragments covering the XBP1 intron and flanking exon fragments were separated on 6% polyacrylamide gels.
qRT-PCR was performed using a CFX96 TouchTM Real-Time PCR detection system (BioRad). Relative gene expression was calculated using the 2 −ΔΔCt method.
RNA sequencing and computational RNA-seq analysis. Library preparation and deep-sequencing were performed by the EMBL Genomics Core Facility. Briefly, RNA integrity was checked using the RNA Nano 6000 Assay Kit (Agilent Technologies, 5067-1511) in a Bioanalyzer 2100 system (Agilent Technologies), and concentration was measured with Qubit RNA Assay Kit (Invitrogen, Q10211) in a Qubit 2.0 Fluorometer (Invitrogen). Small RNA-seq libraries were prepared manually from 250 ng of total RNA using the NEBNext Multiplex Small RNA Library Prep kit (New England Biolabs, E7300L). Obtained libraries that passed the QC step, which was assessed on the Agilent Bioanalyzer system, were pooled in equimolar amounts. Ten picomolar solution of each pool of libraries was loaded on the Illumina sequencer HiSeq2500 and sequenced uni-directionally, generating 190 million reads per run, each 62 bases long. For data analysis, the quality of the reads was analyzed using FastQC (version 0.11.8). The reads were trimmed and adapter sequences were removed using Trimmomatic (version 0.38) 91 . The reads in fastq format were converted to fasta format using fastq_to_fasta commands in the FASTX Toolkit (version 0.0.13). Further read processing, including polyA-tail removal, size filtering (minimal length 12 nt. after clipping), alignment of the reads, computation of alignment statistics, and annotation, was performed using the READemption pipeline (version 0.4.3) 92 with default parameters, with Segemehl (version 0.2.0) 93 . Reads were mapped against the human mature miRNA sequences provided by miRBase (release 22). For normalization, read counts of each miRNA species were normalized by those of miR-29a-3p (a miRNA that does not change with Salmonella infection or E2F1 knockdown). This normalization strategy has been previously used for the normalization of small RNA-seq datasets in which global changes of miRNA expression are observed 94 .
The demultiplexed FASTQ files and coverage files have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession numbers GSE147362 (Salmonella-infected cells), GSE147361 (E2F1 knockdown samples), and GSE147363 (cells treated with secretome).
Piglet infections and sample preparation. Salmonella infection of piglets and sample preparation were performed as described previously 28,95 . Briefly, six piglets were challenged orally with 10 8 cfu of Salmonella Typhimurium, whereas three piglets (mock group) received sterile medium. The mock group was necropsied prior to the infection group and subsequently, three animals were necropsied at 2 or 6 days post-infection (dpi). Sections of mucosal tissue from the ileum and colon were independently sectioned in pieces of 10 cm 2 and immediately frozen in liquid nitrogen for protein and RNA isolation.
All procedures involving animals were performed at the animal facility of the University of Leon, Spain, and were approved by the institutional bioethical committee of the University of Leon, Spain (license number ULE_003_2005, approval date 25 January 2005) and performed according to European regulations regarding animal welfare and protection of animals used for experimental and other scientific purposes. Statistical consideration was not used to determine the animal sample size. Animals were randomly assigned to experimental groups; data blinding was not performed.
Statistical analysis. Unless otherwise indicated, data are presented as mean ± standard error of the mean (s.e.m.), with the exact number of experiments performed indicated in figure legends. Statistical analysis was performed using Prism Software (GraphPad). Normal distribution of the data was assessed by the Shapiro-Wilk test. For statistical comparison of datasets from two conditions, twotailed Student's t-test or multiple t-test corrected for multiple comparison using the Holm-Sidak method were used; for data from three or more conditions/groups, one-way ANOVA with Tukey's or Dunnett's post hoc test or two-way ANOVA with Tukey's multiple comparison test was used. For non-parametric data, Wilcoxon signed rank or Kruskal-Wallis with Dunn's multiple comparison tests were used. Values of P < 0.05 were considered significant. Heatmaps were plotted using TIBCO Spotfire software. Statistical analyses are detailed in Supplementary Data 1.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The data supporting the findings of this study are available from the corresponding author upon reasonable request. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD018026. The demultiplexed FASTQ files and coverage files have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession numbers GSE147362 (Salmonella-infected cells), GSE147361 (E2F1 knockdown samples), and GSE147363 (cells treated with secretome). The Uniprot Homo sapiens proteome database (UP000005640) can be accessed at https://www.uniprot.org/ proteomes/UP000005640. The human mature miRNA sequences can be accessed at http://www.mirbase.org/. Source data are provided with this paper.