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

Abstract

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.

Fig. 1. E2F1 plays a crucial role in miRNA regulation in Salmonella-infected and bystander cells.

a Heat map showing miRNA expression changes in Salmonella-infected cells and E2F1 knockdown cells. Results are shown as log2 fold change compared to mock-treated cells, or cells transfected with control siRNA, respectively. RNA-seq datasets from three independent experiments are shown. Only miRNAs with a number of reads ≥20 in both controls are shown. b Comparison of miRNA expression changes in Salmonella-infected cells and E2F1 knockdown cells. Color code for datapoints identical to panel a, based on average values. c E2F1 protein levels, determined by western blot, in HeLa cells infected with Salmonella or mock treated, analyzed at 1, 4, 8, and 20 hpi. d Schematic representation of the experimental design. HeLa cells infected with Salmonella-expressing GFP were collected at 20 hpi and subjected to FACS to separate the population of cells with internalized bacteria (Salmonella positive) and bystander cells. The total population of cells exposed to Salmonella collected at 20 hpi (unsorted) and mock-treated cells were used for comparison. e, f E2F1 expression, determined by western blot (e) or qRT-PCR (f), in the different cell populations obtained as described in panel d (total, Salmonella positive, and bystander). Results are normalized to mock‐treated cells. g Heat map showing miRNA expression changes upon Salmonella infection of HeLa cells, for the cells with internalized Salmonella (Salmonella positive) and bystander cells. Results are shown as log2 fold change compared to mock-treated cells. RNA-seq datasets from three independent experiments are shown. Only miRNAs with a number of reads ≥20 in the mock-treated cells are shown. h Comparison of miRNA expression changes in Salmonella-positive and bystander cells. Color code for datapoints identical to panel g, based on average values. i Expression levels of the mature forms of 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) determined by qRT-PCR in the total population, Salmonella-positive and bystander cells. Results are normalized to mock‐treated cells. j Expression levels of the mature forms of selected miRNAs determined by qRT-PCR in HeLa cells transfected with control or E2F1 siRNAs. Results are normalized to cells transfected with control siRNA. Salmonella infection was performed at MOI 100. Results are shown as mean ± s.e.m. of n = 5 (c, f, i, j) biologically independent experiments; western blots are representative of n = 3 (e) or n = 5 (c) biologically independent experiments; **P < 0.01 and ***P < 0.001 (statistical analysis is detailed in Supplementary Data 1); Source data are provided as a Source Data file.

Fig. 2. Reduction of E2F1 and miRNA expression is elicited by the secretome of Salmonella-infected cells.

a Schematic representation of the experimental design. Cells were infected with Salmonella and the secretome of cells was collected at 20 hpi (unless otherwise indicated). Secretome was centrifuged and filtered through 0.22 µm filters to exclude the presence of bacteria. The secretome was then used to treat naive cells that were analyzed at 1, 8, 14, and 24 h post-treatment. The secretome of mock-treated cells was used for comparison. b E2F1 expression, determined by western blot, in naive HeLa cells treated with the secretome of Salmonella-infected or mock-treated cells. Results are normalized to naive cells treated with the secretome collected from mock‐treated cells. c Heat map showing miRNA expression changes of Salmonella-infected cells (same as Fig. 1a) and naive HeLa cells treated with the secretome of Salmonella-infected cells (24 h post-treatment). Results are shown as log2 fold change compared to mock-treated cells, or to naive cells treated with secretome collected from mock‐treated cells, respectively. RNA-seq datasets from three independent experiments are shown. Only miRNAs with a number of reads ≥20 in both controls are shown. d Comparison of miRNA expression changes in Salmonella-infected cells and naive cells treated with the secretome of Salmonella-infected cells. Color code for datapoints identical to panel c, based on average values. e Expression levels of the mature forms of selected miRNAs, determined by qRT-PCR in naive HeLa cells treated with the secretome of Salmonella-infected cells, analyzed at 24 h post-treatment. Results are normalized to naive cells treated with secretome collected from mock‐treated cells. f E2F1 protein levels, determined by western blot, in naive HeLa cells treated with the secretome of Salmonella-infected or mock-treated cells and digested or not with proteinase K, analyzed at 24 h post-treatment. g Volcano plot of human proteins identified and quantified in the secretome of Salmonella-infected HeLa cells, expressed as log2 fold change compared to the secretome of mock-treated cells, determined by mass spectroscopy analysis. Proteins highlighted in red and blue present significantly increased or decreased in the secretome of infected cells, respectively. Highlighted proteins are associated with ER-stress response. Average from four independent mass spectroscopy experiments are shown. Secretome was collected at 14 hpi. h ASNS and HMGB1 protein levels, determined by western blot, in the secretome of Salmonella-infected or mock-treated cells, collected at 20 hpi. Ponceau staining of the membranes is shown. i–l Levels of BiP protein (i), IRE1 phosphorylation (j), ATF6 protein (k), and PERK phosphorylation (l), determined by western blot, in HeLa cells infected with Salmonella or mock-treated. m–p Levels BiP protein (m), IRE1 phosphorylation (n), ATF6 protein (o), and PERK phosphorylation (p), determined by western blot, in naive HeLa cells treated with the secretome of Salmonella-infected or mock-treated cells. Tunicamycin treatment was used as a positive control for ATF6 and PERK activation. Salmonella infection was performed at MOI 100. Results are shown as mean ± s.e.m. of n = 5 (b, e, f) biologically independent experiments; western blots are representative of n = 3 (h) or n = 5 (b, f, i–p) biologically independent experiments; **P < 0.01 and ***P < 0.001 (statistical analysis is detailed in Supplementary Data 1); Source data are provided as a Source Data file.

Fig. 3. Activation of the ER-stress response pathway by Salmonella infection or by the secretome of infected cells leads to E2F1 downregulation.

a Schematic representation of the experimental design for infection experiments. Cells were transfected with control, IRE1, or JNK siRNAs, and 48 h later were infected with Salmonella. Samples were analyzed at 1 and 20 hpi. b Schematic representation of the experimental design for secretome treatment. Cells were transfected with control, IRE1, or JNK siRNAs, and 48 h later were treated with the secretome of mock-treated or Salmonella-infected cells collected at 20 hpi. Samples were analyzed at 1 and 24 h post-treatment. c, d E2F1 protein levels, determined by western blot, in HeLa cells transfected with control, three independent siRNAs or a pool of siRNAs targeting IRE1, and either infected with Salmonella or mock-treated (20 hpi; c) or treated with the secretome of Salmonella-infected cells or mock-treated cells (24 h; d). e, f Expression levels of selected miRNAs, determined by qRT-PCR, in HeLa cells transfected with control or IRE1 siRNAs, and infected with Salmonella or mock-treated (20 hpi; e) or treated with the secretome of Salmonella-infected cells or mock-treated cells (24 h; f). Results are normalized to mock‐treated cells. g, i JNK phosphorylation, determined by western blot, in HeLa cells either infected with Salmonella or mock-treated (g) or treated with the secretome of Salmonella-infected cells or mock-treated cells (i). h, j E2F1 protein levels, determined by western blot, in HeLa cells transfected with control, JNK1, JNK2, or JNK1/JNK2 siRNAs, and either infected with Salmonella or mock-treated (20 hpi; h) or treated with the secretome of Salmonella-infected cells or mock-treated cells (24 h; j). k, m JNK phosphorylation, determined by western blot, in HeLa cells transfected with control or IRE1 siRNAs, and either infected with Salmonella or mock-treated (1 hpi; k) or treated with the secretome of Salmonella-infected cells or mock-treated cells (1 h; m). l, n E2F1 protein levels, determined by western blot, in HeLa cells treated with the proteasome inhibitor MG132 or control (DMSO), and either infected with Salmonella or mock-treated (20 hpi; l) or treated with the secretome of Salmonella-infected cells or mock-treated cells (24 h; n). Salmonella infection was performed at MOI 100. Results are shown as mean ± s.e.m. of n = 5 (c, e, f, l, n) or n = 6 (d) biologically independent experiments; western blots are representative of n = 3 (g, i, k, m), n = 5 (c, h, j, l, n) or n = 6 (d) biologically independent experiments; *P < 0.05, **P < 0.01, and ***P < 0.001 (statistical analysis is detailed in Supplementary Data 1); Source data are provided as a Source Data file.

Fig. 4. Regulation of E2F1 by the secretome of Salmonella-infected cells is triggered by engagement of HMGB1 with the receptor RAGE.

a Schematic representation of the experimental design for secretome treatment. Cells were transfected with control or RAGE siRNAs and 48 h later were treated with the secretome of mock-treated or Salmonella-infected cells collected at 20 hpi. b, c IRE1 phosphorylation (b) and E2F1 protein levels (c), determined by western blot, in HeLa cells transfected with control or RAGE siRNAs, and treated with the secretome of Salmonella-infected cells or mock-treated cells for 1 or 24 h, respectively. d Expression levels of selected miRNAs, determined by qRT-PCR, in HeLa cells transfected with RAGE siRNA, and treated with the secretome of Salmonella-infected cells or mock-treated cells for 24 h. Results are normalized to naive cells treated with secretome collected from mock‐treated cells. Compare with HeLa cells transfected with control siRNA (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 mock-treated 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. 5. E2F1 downregulation promotes Salmonella infection.

a Schematic representation of the experimental design. Cells were transfected with control or three independent siRNAs targeting E2F1 and 48 h later cells were infected with Salmonella. Infection was analyzed at 1 and 20 hpi. b, c Representative images (b) and quantification of intracellular Salmonella by cfu (c) of HeLa cells transfected with control or E2F1 siRNAs, and infected with Salmonella. Scale bar, 25 µm. d Schematic representation of the experimental design. The secretome of Salmonella-infected cells was used to treat naive cells for 24 h with control (DMSO) or KIRA6. Cells were then infected with Salmonella and analyzed at 1 and 20 hpi. KIRA6/control treatment was maintained during infection. The secretome of mock-treated cells was used for comparison. e, f Representative images (e) and quantification of intracellular Salmonella by cfu (f) of naive HeLa treated with the secretome of Salmonella-infected cells or mock-treated cells, treated with KIRA6/control, and then infected with Salmonella. Scale bar, 25 µm. g Schematic representation of the experimental design. Cells were pre-treated for 24 h with control (DMSO), the chemical inhibitor of IRE1 kinase activity KIRA6, or the ER-stress inducers tunicamycin or DTT and then infected with Salmonella. Treatments were maintained during infection. Infection was analyzed at 1 and 20 hpi. h, i Quantification of intracellular bacteria by cfu of HeLa cells infected with Salmonella, upon treatment with control/KIRA6 (h) or control/tunicamycin/DTT (i). Infection was performed at MOI 10. Results are shown as mean ± s.e.m. of n = 5 (c, h, i) or n = 6 (f) biologically independent experiments; microscopy images are representative of n = 3 (b, e) biologically independent experiments; *P < 0.05, **P < 0.01, and ***P < 0.001 (statistical analysis is detailed in Supplementary Data 1).

Fig. 6. E2F1 and miRNA downregulation upon Salmonella infection also occurs in vivo.

a Schematic representation of the in vivo Salmonella infection experiments. Four-week-old piglets were orally challenged with Salmonella; tissue samples of ileum and colon were collected at 2 and 6 days post-infection (dpi). Samples of untreated animals (mock) were collected for comparison. b–e Levels of E2F1 protein (b, c) and IRE1 phosphorylation (d, e), evaluated by western blot, in ileum and colon samples obtained from untreated animals (mock) or piglets challenged with Salmonella. Values below the western blots indicate E2F1:GAPDH, E2F1:β-actin, or IRE1-P:α-tubulin ratios; averages of the three control samples were set to 1. f, g Expression levels of the mature forms of selected miRNAs, determined by qRT-PCR, in the ileum (f) and colon (g) samples obtained from untreated animals (mock) or piglets challenged with Salmonella. Three animals were analyzed per condition/time point. Results are shown as mean ± s.e.m.; *P < 0.05, **P < 0.01, and ***P < 0.001 (statistical analysis is detailed in Supplementary Data 1); Source data are provided as a Source Data file.

Fig. 7. Model depicting the role of E2F1 in Salmonella infection and dissemination to bystander cells.

Salmonella internalization activates the ER-stress response (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).