Klebsiella pneumoniae Reduces SUMOylation To Limit Host Defense Responses

K. pneumoniae decreases SUMO-conjugated proteins in epithelial cells.To investigate whether K. pneumoniae affects host cell SUMOylation, we compared the global pattern of proteins conjugated to SUMO1 or SUMO2/3 in uninfected cells with that of cells infected by the wild-type hypervirulent K. pneumoniae strain CIP52.145 (Kp52145 here). This strain harbors the genetic determinants associated with severe human infections (16). A549 epithelial cells showed a reduction in overall SUMO1-conjugated proteins of high molecular weight (>80 kDa) after 3 h of infection, with no changes in the overall SUMO2/3-conjugated proteins (Fig. 1A). We then validated the decrease in SUMOylation on the heavily SUMOylated RanGAP1 substrate. In Klebsiella-infected cells, we observed a clear reduction in SUMOylated RanGAP1 after 3 h postinfection (Fig. 1B). The decrease in SUMO1-conjugated proteins was also observed in NuLi-1 cells, a human primary-like airway epithelial cell line (see Fig. S1 in the supplemental material), demonstrating that the Klebsiella-induced decrease in overall SUMOylation was not cell type dependent. Kp52145-induced decrease in protein SUMOylation was dependent on live bacteria, because infection with either UV-killed or heat-killed bacteria caused an increase in SUMO1-conjugated proteins (Fig. 1C). To examine the relevance of these findings during infection, we assessed the pattern of SUMOylated proteins in lungs of infected mice. Although there was mouse-to-mouse variation, immunoblotting of lung homogenates revealed a decrease in the global SUMO1-conjugated proteome compared to that in homogenates from mock-infected mice (Fig. 1D). Together, our in vitro and in vivo findings reveal that K. pneumoniae infection leads to a decrease in the levels of SUMO-conjugated proteins.

• Open in new tab
• Download powerpoint

FIG 1

K. pneumoniae decreases SUMO-conjugated proteins in epithelial cells. (A) Immunoblot analysis of SUMO1, SUMO2/3, and tubulin levels in lysates of A549 cells infected with Kp52145 and the capsule mutant (strain 52145 Δwca_K2) for the indicated times. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to noninfected control cells. n.i., not infected; **, P ≤ 0.01; ***, P ≤ 0.001 by one-way ANOVA with Bonferroni’s multiple-comparison test. (B) Immunoblot analysis of RanGAP1 and tubulin levels in A549 infected with Kp52145 for the indicated times. (C) Immunoblot analysis of SUMO1 and tubulin levels in lysates of A549 cells infected with Kp52145 live, heat-killed for 30 min at 60°C, or UV-killed for 30 min at 10,000 J for the indicated times. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change at 5 h of infection compared to noninfected control cells. **, P ≤ 0.01; *, P ≤ 0.05 by one-way ANOVA with Bonferroni’s multiple-comparison test. (D) Immunoblot analysis of SUMO1 (antibody sc-9060) and tubulin levels in 50 μg of total protein obtained from whole-lung homogenate of C57BL/6 mice infected with Kp52145 or with PBS as a mock control for 24 h. Each lane represents a lung from a different mouse. SUMO1 smears were quantified from three independent blots using Image Studio Lite (LI-COR) and normalized to β-actin. The graph represents fold change of the Kp52145 lung homogenates compared to PBS controls. Two-tailed unpaired t test. (E) Immunoblot analysis of SUMO1 levels in lysates of cytochalasin D- (CytD; 5 μM, 1 h before infection) or dimethyl sulfoxide (DMSO; vehicle solution)-treated A549 cells infected with strain 52145 Δwca_K2 for 5 h. Membranes were reprobed for tubulin as a loading control. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to DMSO noninfected control cells. *, P ≤ 0.05 by one-way ANOVA with Bonferroni’s multiple-comparison test. In all panels, data are representative of at least three independent experiments.

In contrast to the wild-type strain, infection with an isogenic avirulent capsule mutant (strain 52145 Δwca_K2) induced an increase in SUMO1- and SUMO2/3-conjugated proteins (Fig. 1A). Since the capsule polysaccharide (CPS) abrogates the internalization of K. pneumoniae by A549 (17), we hypothesized that the capsule mutant-induced increase in SUMOylation is triggered by bacterial internalization. Indeed, when infections were performed in the presence of cytochalasin D, an inhibitor of Klebsiella internalization by epithelial cells (17), the CPS mutant did not increase the levels of SUMO1-conjugated proteins (Fig. 1E). Collectively, these results indicate that the CPS limits the increase in SUMOylation upon Klebsiella infection by preventing bacterial internalization.

The deSUMOylase SENP2 mediates K. pneumoniae-induced decrease in SUMO1-conjugated proteins in epithelial cells.We sought to gain insights into the mechanisms by which wild-type Klebsiella reduces the SUMOylation status in epithelial cells. Previous studies indicate that depletion of the E1 enzyme SAE2 and the E2 enzyme Ubc9 underlines bacterial pathogen-triggered decrease in SUMOylation (5–8). However, the levels of Ubc9 were not significantly affected in Kp52145-infected cells (Fig. 2A), and we did not detect any change in the levels of SAE2 or SAE1 (Fig. 2B). Although these results do not rigorously rule out that Klebsiella may affect the enzymatic activity of these enzymes, it prompted us to consider alternative possibilities to explain the Klebsiella-induced decrease in SUMOylation.

• Open in new tab
• Download powerpoint

FIG 2

The deSUMOylase SENP2 mediates K. pneumoniae-induced decrease in SUMO-conjugated proteins. (A) Immunoblot analysis of Ubc9 and tubulin levels in lysates of A549 cells infected with Kp52145 for the indicated times. Ubc9 bands were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to noninfected control cells. n.s., not significant versus n.i. determined using one-way ANOVA with Tukey’s multiple-comparison test. (B) Immunoblot analysis of SAE1, SAE2, and tubulin levels in lysates of A549 cells infected with Kp52145 for the indicated times. SAE1 and SAE2 bands were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to noninfected control cells. n.s., not significant versus n.i. determined using one-way ANOVA with Tukey’s multiple-comparison test. (C) Immunoblot analysis of SUMO1 and tubulin levels in lysates of control (AS), SENP1, or SENP2 siRNA-transfected A549 cells infected with Kp52145 for 5 h. AS, AllStars control, nonsilencing siRNA. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to AS-transfected noninfected control cells. n.s., not significant; ***, P ≤ 0.001; *, P ≤ 0.05 by one-way ANOVA with Bonferroni’s multiple-comparison test. (D) Fluorescence microscopy of pSENP2-GFP-transfected A549 cells grown on glass coverslips. Cells were infected with Kp52145 for 5 h or left uninfected (n.i.). Coverslips were stained with Hoechst (4′,6-diamidino-2-phenylindole [DAPI]) for nucleus identification. The percentage of SENP2 localized in either nucleus or cytoplasm is represented on the graph and is the result of independent counting of 100 cells from each of 3 independent experiments. (E) Immunoblot analysis of SENP2 and tubulin levels in cytosolic extracts of A549 cells infected with Kp52145 for 5 h or left uninfected. (F) Fluorescence microscopy of pSENP2-GFP transfected A549 cells grown on glass coverslips. Cells were treated with the proteasomal inhibitor MG262 (5 μM, 2 h before infection), the nuclear export inhibitor leptomycin B (LMB; 10 nM, 2 h before infection), or DMSO (vehicle solution) and infected with Kp52145 for 5 h or left uninfected (n.i.). Coverslips were stained with Hoechst (DAPI) for nucleus identification. The percentage of SENP2 localized in either nucleus or cytoplasm is represented on the graphs and is the result of independent counting of 100 cells from each of 3 independent experiments. (G) Immunoblot analysis of SUMO1 (antibody sc-9060) and tubulin levels in lysates of the nuclear export inhibitor leptomycin B- (5 μM, 2 h before infection) or DMSO (vehicle solution)-treated A549 infected with Kp52145 for 5 h or left uninfected. SUMO1 smears were quantified using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to DMSO noninfected control cells. ***, P ≤ 0.001; **, P ≤ 0.01; *, P ≤ 0.05 by one-way ANOVA with Bonferroni’s multiple-comparison test. (H) Immunoblot analysis of K48 linkage-specific polyubiquitin and GFP levels in immunoprecipitates of A549 infected with Kp52145 for 5 h. Cells were immunoprecipitated using anti-GFP antibody. Preimmune mouse IgG served as a negative control. (I) Immunoblot analysis of SENP2 and tubulin levels in cytosolic extracts of control (AS), RBX1, Skp1, Cul-1, βTrCP or Skp2 siRNA-transfected A549 cells left uninfected. (J) Immunoblot analysis of cullin-1 and tubulin levels in lysates of A549 cells infected with Kp52145 for 5 h or left uninfected. Lysates of A549 treated with 5 mM H₂O₂ for 10 min were used as a positive control for cullin-1 deNEDDylation. (K) Immunoblot analysis of Cul-1 and tubulin levels in lysates of control (AS) or CSN5 siRNA-transfected A549 cells infected with Kp52145 for 5 h or left uninfected. (L) Immunoblot analysis of K48 linkage-specific polyubiquitin and GFP levels in immunoprecipitates of control (AS) or CSN5 siRNA-transfected A549. Cells were transfected with a SENP2-GFP plasmid 24 h after the siRNA transfection and infected the following day with Kp52145 for 5 h or left uninfected. Cells were immunoprecipitated using anti-GFP antibody. Preimmune mouse IgG served as negative control. (M) Immunoblot analysis of SENP2 and tubulin levels in cytosolic extracts of control (AS) or CSN5 siRNA-transfected A549 cells infected with Kp52145 for 5 h or left uninfected. (N) Immunoblot analysis of SUMO1 and tubulin levels in lysates of control (AS) or CSN5 siRNA-transfected A549 cells infected with Kp52145 for 5 h or left uninfected. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to AS-transfected noninfected control cells. ***, P ≤ 0.001; **, P ≤ 0.01; *, P ≤ 0.05 by one-way ANOVA with Bonferroni’s multiple-comparison test. In all panels, data are representative of at least three independent experiments.

SENPs, a family of cysteine proteases, catalyze the deconjugation of SUMO from target substrates; therefore, we hypothesized that they might be involved in Klebsiella-triggered reduction of SUMO-conjugated proteins. SENP1 and SENP2 have nuclear localization with isopeptidase and hydrolase activity for SUMO1 and SUMO2/3 (18). SENP3 and SENP5 are localized in the nucleolus, while SENP6 and SENP7 are localized in the nucleoplasm, and all favor SUMO2/3 as the substrates (18). A small interfering RNA (siRNA)-based screen revealed an increase in the levels of SUMO1-conjugated proteins in infected SENP2 knockdown cells compared to that in cells transfected with nonsilencing control siRNA (AS) and the other SENPs (Fig. 2C; see also Fig. S2A), suggesting that SENP2 mediates the decrease in conjugated SUMO1 proteins in K. pneumoniae-infected cells. Control experiments showed that Klebsiella infection did not alter the transcription of any SENPs (Fig. S2B).

Owing to the global scale of the SUMOylation decrease in infected cells, it was intriguing that SENP2, a nucleus-associated deSUMOylase (19), accounts for the Klebsiella-triggered decrease in SUMOylation. We then sought to determine whether Klebsiella might affect the cellular distribution of SENP2 in infected cells. Fluorescence-based single-cell analysis confirmed that SENP2-green fluorescent protein (GFP) was mostly localized in the nuclei of noninfected A549 cells (Fig. 2D). In contrast, SENP2-GFP was mostly localized in the cytosol of infected cells (Fig. 2D). This finding was further supported by immunoblotting of cytosolic extracts demonstrating an accumulation of endogenous SENP2 in infected cells (Fig. 2E).

In human bone osteosarcoma epithelial cells (U-2 OS), it has been shown that SENP2 shuttles continuously between the nucleus and the cytoplasm, where it is degraded by the 26S ubiquitin proteasome (20). Corroborating these findings, pretreatment of A549 cells with the proteasome inhibitor MG262 led to the accumulation of SENP2-GFP in the cytosol (Fig. 2F). The localization of SENP2-GFP in the cytosol of infected cells was abrogated when infections were performed in the presence of leptomycin B, a nuclear export inhibitor (Fig. 2F), demonstrating that the presence of SENP2 in the cytosol of infected cells is dependent on nucleocytoplasmic shuttling. The Klebsiella-induced decrease in SUMOylation was not observed in leptomycin B-treated cells (Fig. 2G), indicating that Klebsiella-triggered SENP2 localization in the cytosol mediates the decrease in the levels of SUMO-conjugated proteins.

K. pneumoniae triggers loss of cullin-1 NEDDylation to increase SENP2 levels.To gain molecular insights into how Klebsiella increases the levels of SENP2 in the cytosol of infected cells, we hypothesized that K. pneumoniae may compromise the ubiquitin proteasome-mediated degradation of SENP2 occurring in the cytosol (20). The fact that SENP2 ubiquitylation is a necessary event to mark the protein for degradation by the 26S proteasome (20) prompted us to assess whether SENP2 is ubiquitylated in infected cells. In MG262-treated cells, GFP-tagged SENP2 was pulled down with an anti-GFP antibody, and K48-conjugated SENP2 was detected with an anti-K48 antibody. Ubiquitylated SENP2 was barely detected in Klebsiella infected cells in contrast to that in noninfected cells (Fig. 2H), sustaining the notion that SENP2 accumulation in the cytosol of infected cells is due to reduced proteasomal degradation, because SENP2 is not K48 ubiquitin tagged.

We next sought to identify the E3 ubiquitin ligase complex responsible for SENP2 ubiquitylation. The SCF (S-phase-kinase-associated protein 1 [Skp1], Cullin-1 (Cul-1), F-box protein) E3 ubiquitin ligases are the largest family of E3s in mammals (21). Reduction in the levels of the adaptor protein Skp1 and the scaffold protein Cul-1 by siRNA led to an accumulation of SENP2 in the cytosol of noninfected cells (Fig. 2I). The F-box protein determines the substrate specificity of the SCF, with three major classes of F-box proteins depending on the substrate interaction domain (WD40 repeats as βTrCP, leucine-rich repeats as Skp1 or Skp2, and other types such as cyclin F). siRNA-based experiments showed the accumulation of SENP2 only in the cytosol of βTrCP knockdown cells (Fig. 2I). Altogether, these results demonstrate that the SCF-E3 ligase Skp1-Cul-1-βTrCP mediates the degradation of SENP2 in the cytosol of cells.

The activity of SCF requires the conjugation of the ubiquitin-like polypeptide NEDD8 to Cul-1, suggesting the possibility that Klebsiella may trigger the loss of Cul-1 NEDDylation, resulting in the lack of transfer of ubiquitin to SENP2. Typically, Cul-1 appears as a doublet at ∼85 and ∼90 kDa, with the higher molecular band representing the NEDDylated form of Cul-1 (22). Immunoblotting experiments revealed the lack of Cul-1 NEDDylation in infected cells (Fig. 2J). As a positive control, cells were pretreated with H₂O₂, as this was shown to cause loss of Cul-1 NEDDylation (Fig. 2J) (23).

DeNEDDylation of Cul-1 is normally catalyzed by the COP9 signalosome (e.g., subunit 5 or CSN5/JAB1) (24). We then examined whether repression of CSN5 expression may inhibit Klebsiella-induced Cul-1 deNEDDylation. As shown in Fig. 2K, Klebsiella-triggered Cul-1 deNEDDylation was significantly reduced in CSN5 knockdown cells. In CSN5-silenced cells, Klebsiella did not reduce the K48-linked ubiquitylation of SENP2 (Fig. 2L), and, as anticipated, there was a reduction in SENP2 levels in the cytosol of infected cells (Fig. 2M). Finally, Klebsiella induced an increase in the global levels of SUMO1-conjugated proteins in CSN5-silenced cells (Fig. 2N).

In summary, these results demonstrate that K. pneumoniae impairs the cytosolic degradation of SENP2 by affecting its K48 ubiquitylation through CSN5-mediated deNEDDylation of the Cul-1 subunit of the SCF E3 ligase Skp1-Cul-1-βTrCP.

K. pneumoniae increases CSN5 levels through an EGFR-PI3K-AKT-ERK-GSK3β pathway.Transcription analysis showed that K. pneumoniae induced the expression of csn5 in A549 cells (Fig. 3A), and immunoblotting experiments demonstrated that Kp52145 increased the levels of CSN5 (Fig. 3B). The regulation of CSN5 is poorly understood, although there is evidence showing that CSN5 is a target of EGFR signaling (25). Previous work from our laboratory revealed that K. pneumoniae activates the EGFR-phosphatidylinositol 3-kinase (PI3K)-protein kinase B (AKT)-extracellular signal-regulated kinase (ERK)-glycogen synthase kinase 3 beta (GSK3β) signaling pathway to limit inflammation (14). We then investigated whether Klebsiella-induced expression of CSN5 is dependent on this signaling pathway. To test this hypothesis, we utilized pharmacologic inhibitors of EGFR (AG1478), PI3K (LY294002), AKT (AKT-X), or MEK (U0126) to block the pathway at four different levels. Immunoblotting revealed that Kp52145-triggered induction of CSN5 was reduced in cells pretreated with the inhibitors (Fig. 3C). Since we have shown that Kp52145 inactivates GSK3β by triggering the phosphorylation of serine 9 (17), we examined the role of inhibitory phosphorylation of GSK3β on CSN5 expression upon Kp52145 infection. Vectors encoding wild-type GSK3β (GSK3β-WT) or a constitutively active mutant (GSK3β-S9A) in which the serine 9 residue was changed to alanine were transiently transfected into A549 cells. As a control, cells were transfected with the empty vector pcDNA3. Klebsiella did not induce CSN5 in cells transfected with the GSK3β-S9A vector (Fig. 3C), demonstrating that the inhibitory phosphorylation of GSK3β is required for Klebsiella induction of CSN5. Next, we examined whether inhibition of this EGFR-controlled signaling pathway would abrogate Klebsiella-imposed reduction in SUMOylation. Indeed, pretreatment of the cells with the EGFR inhibitor AG1478 led to an overall increase in SUMO1 levels upon infection (Fig. 3D). Because Klebsiella CPS is the bacterial factor triggering the activation of EGFR (14), we assessed whether Klebsiella-induced CSN5 upregulation is dependent on the CPS. As anticipated, the cps mutant, strain 52145 Δwca_K2, did not increase CSN5 levels in A549 cells (Fig. 3E).

• Open in new tab
• Download powerpoint

FIG 3

K. pneumoniae increases CSN5 levels in an EGFR-PI3K-AKT-ERK-GSK3β-dependent manner. (A) csn5 mRNA levels, assessed by qPCR, in A549 cells left untreated (n.i.) or infected for up to 5 h with Kp52145. Values are presented as the means ± standard deviations (SDs) from three independent experiments measured in duplicates. ***, P ≤ 0.001 by one-way ANOVA with Bonferroni’s multiple-comparison test. (B) Immunoblot analysis of CSN5 and tubulin levels in lysates of A549 cells infected with Kp52145 for the indicated times. (C) Immunoblot analysis of CSN5 and tubulin levels in lysates of A549 cells infected with Kp52145 for the indicated times. Cells were pretreated with AG1478 (5 μM), LY294002 (20 μM), AKT-X (30 μM), U0126 (10 μM), or DMSO (vehicle solution) for 2 h before infection where indicated. Where indicated, cells were transfected with plasmids expressing GSK3β either wild-type or with a S9A mutation that renders the protein constitutively active or with pcDNA3, empty vector. The signaling pathway EGFR-PI3K-AKT-ERK-GSK3β activated in K. pneumoniae infection is depicted to represent the several steps inhibited in this figure. (D) Immunoblot analysis of SUMO1 levels in lysates of AG1478- (5 μM, 2 h before infection) or DMSO (vehicle solution)-treated A549 cells infected with Kp52145 for the indicated times. Membranes were reprobed for tubulin as a loading control. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change at 5 h of infection compared to DMSO noninfected control cells. **, P ≤ 0.01; ***, P ≤ 0.001 by one-way ANOVA with Bonferroni’s multiple-comparison test. (E) Immunoblot analysis of CSN5 and tubulin levels in lysates of A549 cells infected with Kp52145 or 52145 Δwca_K2 for the indicated times. In all panels, data are representative of at least three independent experiments.

Overall, these findings support the notion that Klebsiella CPS engages an EGFR-PI3K-AKT-ERK-GSK3β pathway to increase the levels of CSN5 to decrease the levels of SUMO-conjugated proteins.

K. pneumoniae decreases SUMOylation in macrophages.Having demonstrated that Klebsiella reduces SUMOylation in epithelial cells, we investigated whether Klebsiella affects SUMO-conjugated proteins in macrophages. A wealth of evidence has established that macrophages play a crucial role in host defense against K. pneumoniae, not only by clearing the bacteria but also by shaping the immune response against this pathogen (26, 27). MH-S mouse alveolar macrophages were infected with Kp52145, and the overall pattern of SUMO1-conjugated proteins was accessed by immunoblotting. Figure 4A shows that Kp52145 induced a decrease in the levels of SUMO1-conjugated proteins as well as a decrease in the levels of free SUMO1 (Fig. 4B). Similar results were observed in infected THP-1 human macrophages (Fig. 4C), indicating that the decrease in SUMOylation is not cell type dependent. Either UV-killed or heat-killed bacteria did not trigger a decrease in the level of SUMOylated proteins (Fig. S3), demonstrating that the reduction in the levels of SUMO1-conjugated proteins is mediated by live bacteria.

• Open in new tab
• Download powerpoint

FIG 4

K. pneumoniae decreases SUMOylation in macrophages. (A) Immunoblot analysis of SUMO1 and tubulin levels in lysates of mouse alveolar macrophages, MH-S cells, infected with Kp52145 for the indicated time points. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to noninfected control cells at the indicated time points. ***, P ≤ 0.001 by one-way ANOVA with Bonferroni’s multiple-comparison test. (B) Immunoblot analysis of SUMO1 and tubulin levels in lysates of MH-S cells infected with Kp52145 for the indicated time points. The band corresponding to free SUMO1 at 11 kDa was quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to noninfected control cells at the indicated time points. *, P ≤ 0.05; **, P ≤ 0.01 by one-way ANOVA with Bonferroni’s multiple-comparison test. (C) Immunoblot analysis of SUMO1 (antibody sc-9060) and tubulin levels in lysates of human monocytic THP-1 cells infected with Kp52145 for the indicated time points. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change at 180 min compared to noninfected control cells. Two-tailed unpaired t test. (D) Immunoblot analysis of SUMO1 and tubulin levels in lysates of MH-S cells infected with Kp52145 or the indicated mutants for 1 h. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to noninfected control cells. **, P ≤ 0.01; ***, P ≤ 0.001 by one-way ANOVA with Bonferroni’s multiple-comparison test. n.i., noninfected control. In all panels, data are representative of at least three independent experiments.

Previous work from this laboratory demonstrated the importance of the K. pneumoniae polysaccharides, CPS and lipopolysaccharide (LPS), in the interplay between the pathogen and macrophages (28). To examine whether CPS and LPS mediate the Klebsiella-triggered decrease in SUMOylation, we assessed the global pattern of SUMO1-conjugated proteins in macrophages infected with the CPS (52145 Δwca_K2) and LPS O-polysaccharide (52145 Δglf) mutants and the double mutant lacking both polysaccharides (52145 Δwca_K2 Δglf). Immunoblotting revealed an increase in the levels of SUMO-conjugated proteins in macrophages infected with the CPS and LPS O-polysaccharide mutants (Fig. 4D), indicating that both polysaccharides have a role in the Klebsiella-triggered decrease in SUMOylated proteins. Controls confirmed there were no differences in the levels of surface-attached capsule produced by the wild-type and the LPS O-polysaccharide mutant (174.4 and 174.8 μg glucuronic acid/10⁻⁹ CFU, respectively).

K. pneumoniae-induced decrease in SUMOylation in macrophages is mediated by TLR4-dependent type I interferon.Like in epithelial cells, Ubc9 levels were not decreased in Klebsiella-infected macrophages (see Fig. S4A). Cell fractionation experiments did not reveal increased levels of SENP2 in the cytosol (Fig. S4B), and NEDDylation of Cul-1 was not impaired in infected macrophages (Fig. S4C). Collectively, these observations indicated that Klebsiella may employ a different strategy in macrophages to decrease the levels of SUMO1-conjugated proteins.

The fact that the CPS and LPS O-polysaccharide mediated the Klebsiella-triggered decrease in SUMOylated proteins and that both polysaccharides are recognized by TLR4 (13, 29) led us to investigate whether Klebsiella targets TLR4 signaling to decrease SUMOylation. We observed an increase in SUMOylated proteins in infected tlr4^−/− immortalized bone marrow-derived macrophages (iBMDMs) (Fig. 5A). TLR4 signals via adaptors MyD88 and TRIF, with TRAM acting as a sorting adaptor controlling TRIF localization (30). To learn which TLR4 adaptor(s) mediates the Klebsiella-induced decrease in global SUMOylation, myd88^−/− and tram/trif^−/− iBMDMs were infected. We observed no differences in the global levels of SUMO1-conjugated proteins between wild-type and myd88^−/− iBMDMs (Fig. 5B). In contrast, there was an increase in the levels of SUMOylated proteins in tram/trif^−/− iBMDMs (Fig. 5C), indicating that Klebsiella-triggered decrease in SUMOylation is dependent on TLR4-TRAM-TRIF signaling.

• Open in new tab
• Download powerpoint

FIG 5

K. pneumoniae SUMOylation decrease in macrophages is mediated by TLR4-dependent type I interferon and let-7 miRNAs. (A) Immunoblot analysis of SUMO1 and tubulin levels in lysates of immortalized bone marrow-derived macrophages (iBMDMs) derived from C57BL/6 mice (WT) or Tlr4 knockout mice (tlr4^−/−) infected with Kp52145 for the indicated time points. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to wild-type (WT) iBMDM noninfected control cells. **, P ≤ 0.01; *, P ≤ 0.05 by one-way ANOVA with Bonferroni’s multiple-comparison test. (B) Immunoblot analysis of SUMO1 and tubulin levels in lysates of iBMDM cells derived from wild-type (WT) mice or MyD88 knockout mice (myd88^−/−) infected with Kp52145 for the indicated time points. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to wild-type (WT) iBMDM noninfected control cells. ***, P ≤ 0.001 by one-way ANOVA with Bonferroni’s multiple-comparison test. (C) Immunoblot analysis of SUMO1 and tubulin levels in lysates of iBMDM cells derived from WT mice or TRAM/TRIF double knockout mice (tram/trif^−/−) infected with Kp52145 for the indicated time points. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to wild-type (WT) iBMDM noninfected control cells. ***, P ≤ 0.001; **, P ≤ 0.01; *, P ≤ 0.05 by one-way ANOVA with Bonferroni’s multiple-comparison test. (D) Type I IFN levels determined in the supernatants of iBMDM wild-type (WT) or tlr4^−/− cells left untreated (n.i.) or infected for 3 h with Kp52145. The reporter cell line B16-Blue IFN-α/β was used for the quantification of levels of SEAP produced upon stimulation of the supernatants with the detection medium QUANTI-Blue and presented as IFN-α/β U ml⁻¹. IFN-β mRNA levels were assessed by qPCR and are presented as fold change against untreated wild-type (WT) cell levels. Values are presented as the means ± SDs from three independent experiments measured in duplicates. ***, P ≤ 0.001; **, P ≤ 0.01; *, P ≤ 0.05; n.s., not significant by one-way ANOVA with Tukey’s multiple-comparison test. (E) Immunoblot analysis of SUMO1 and tubulin levels in lysates of BMDM cells derived from wild-type (WT) or IFNAR1 knockout mice (ifnar1^−/−) infected with Kp52145 for the indicated time points or left untreated (n.i.). SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to wild-type (WT) BMDM noninfected control cells. ***, P ≤ 0.001; **, P ≤ 0.01; * ≤ 0.05 by one-way ANOVA with Bonferroni’s multiple-comparison test. (F) ifnb1, ifit1, and isg15 mRNA levels, assessed by qPCR, in MH-S cells left untreated (n.i.) or infected for 3 h with the K. pneumoniae strains indicated. Values are presented as the means ± SDs from three independent experiments measured in duplicates. ***, P ≤ 0.001 by one-way ANOVA with Tukey’s multiple-comparison test. (G) let-7 mRNA levels, assessed by qPCR, in BMDM cells derived from wild-type (WT) or ifnar1^−/− mice infected with Kp52145 for 1 h or left untreated (n.i.). Values are presented as the means ± SDs from three independent experiments measured in duplicates. ***, P ≤ 0.001 versus WT n.i. determined using two way-ANOVA with Holm-Sidak’s multiple-comparison test. (H) Immunoblot analysis of SUMO1 and tubulin levels in lysates of let-7 miRNA antagomir-transfected MH-S cells infected with Kp52145 for 1 h or left untreated (n.i.). Negative, Caenorhabditis elegans control sequence with minimal sequence identity in mouse cells. SUMO1 smears were quantified from three independent experiments using Image Studio Lite (LI-COR) and normalized to α-tubulin. The graph represents fold change compared to negative-transfected noninfected control cells. ***, P ≤ 0.001 by one-way ANOVA with Bonferroni’s multiple-comparison test. In panels A, B, C, E, and H, data are representative of at least three independent experiments.

The TLR4-TRAM-TRIF signaling pathway governs type I IFN production (31), making it relevant to investigate whether the Klebsiella-induced decrease in SUMOylation is mediated by type I IFN. Further confirming recent findings of our laboratory (32), Kp52145 induced IFN-β, ifnb1, and interferon-stimulated genes (ISGs) in wild-type iBMDMs, but those were significantly reduced in tlr4^−/− cells (Fig. 5D; see also Fig. S5A). To demonstrate the contribution of type I IFN to the Klebsiella-triggered decrease in levels of SUMO-conjugated proteins, we infected BMDMs obtained from ifnar1^−/− mice. Kp52145 increased the levels of SUMOylated proteins in ifnar1^−/− BMDMs (Fig. 5E), demonstrating that the Klebsiella-induced decrease in SUMOylation is indeed dependent on type I IFN signaling. We then investigated whether a reduced induction of type I IFNs by the CPS and LPS O-polysaccharide mutants may explain the high levels of SUMO1-conjugated proteins in infected macrophages. Indeed, the three mutants induced less ifnb1 and ISGs ifit1 and isg15 expression than the wild type (Fig. 5F). Furthermore, addition of type I IFN to macrophages infected with the double mutant lacking CPS and LPS O-polysaccharide rescued the decrease in the levels of SUMOylated proteins (Fig. S5B). Altogether, these findings provide compelling evidence to the notion that TLR4-TRAM-TRIF-induced type I IFN via IFN receptor 1 (IFNAR1)-controlled signaling mediates the Klebsiella-triggered decrease in the levels of SUMOylation.

let-7 miRNAs mediate type I IFN-mediated decrease in SUMOylation.IFNs regulate the expression of miRNAs from the let-7 family (33). miRNAs downregulate mRNA stability and protein synthesis through complementary elements in the 3′ untranslated regions (UTRs) of their target mRNAs. Sahin and coworkers reported that type I IFNs control SUMO expression and validated let-7 target sequences in 3′ UTRs of human and mouse sumo1, sumo2, and sumo3 transcripts (34), indicating an evolutionary conserved regulation of SUMO by let-7 family members. This evidence prompted us to investigate whether the Klebsiella-induced decrease in the overall levels of SUMO-conjugated proteins by type-I IFNs is through let-7 miRNAs. We reasoned that Kp52145 should upregulate the expression of a let-7 miRNA member(s) only in wild-type but not in ifnar1^−/− BMDMs. Figure 5G shows that Klebsiella upregulated the expression of let-7f, let-7g, and let-7i in wild-type BMDMs. These miRNAs were not upregulated in Klebsiella-infected ifnar1^−/− BMDMs (Fig. 5G), leading us to study whether these let-7 family members mediate the type I IFN-controlled decrease in SUMOylation in Klebsiella-infected macrophages. To validate this hypothesis, antagomirs for these let-7 family members were transfected into macrophages, and the global pattern of SUMOylation was determined by immunoblotting. As controls, we also tested a nontargeting control antagomir and a let-7b antagomir, since this miRNA was not significantly upregulated by Klebsiella in wild-type BMDMs (Fig. 5G). In contrast to cells transfected with a nontargeting control, inhibition of let-7f, let-7g, and let-7i led to a global upregulation of SUMOylation in infected cells (Fig. 5H). The effect was less prominent in let-7i-inhibited cells than in let-7f and let-7g ones (Fig. 5H). Collectively, these results reveal that Klebsiella-dependent type I IFN-mediated decrease in SUMOylation involves let-7 miRNAs.

SUMOylation increases host defenses against K. pneumoniae.The sophisticated strategies employed by Klebsiella to decrease the levels of SUMO-conjugated proteins in epithelial cells and macrophages strongly suggested that SUMOylation should play an important role in host defense against Klebsiella infections. Therefore, we sought to determine whether alterations in the levels of SUMO1-conjugated proteins might influence Klebsiella-cell cross talk. By using a SUMO1 expression plasmid, a common method to upregulate the SUMO status of the host proteins (5, 6), we transiently upregulated SUMOylation (see Fig. S6A and B). We and others have established that the dampening of inflammation is a hallmark of Klebsiella infection biology in vitro and in vivo (12, 14, 15, 35, 36). Therefore, we examined the levels of interleukin 8 (IL-8) and tumor necrosis factor alpha (TNF-α) released by infected epithelial cells and macrophages, respectively. IL-8 and TNF-α are well-established inflammatory readouts upon infection. In SUMO1-overexpressing A549 cells, we observed a significant increase in IL-8 in the supernatants of infected cells (Fig. 6A). Furthermore, knockdown of SENP2 also led to increased release of IL-8, strengthening the role of this protein in controlling SUMOylation levels and inflammation (Fig. 6B). SUMO1-overexpressing macrophages also produced more TNF-α upon infection (Fig. 6C). Interestingly, macrophages transfected with let-7f and let-7g antagomirs also released more TNF-α upon infection (Fig. 6D), further sustaining the connection between increased SUMOylation and inflammation in macrophages. Collectively, these findings suggest that the Klebsiella-triggered decreased in SUMOylation in epithelial cells and macrophages is a pathogen strategy to limit inflammation.

• Open in new tab
• Download powerpoint

FIG 6

SUMOylation increases host defenses against K. pneumoniae. (A) ELISA of IL-8 secreted by A549 cells transfected with either empty control plasmid (pcDNA3) or SUMO1-hemagglutinin (HA) tag plasmid (pSUMO1), which were left untreated (n.i.) or infected with Kp52145 for 3 h of contact, after which, the medium was replaced with medium containing gentamicin (100 μg ml⁻¹) to kill extracellular bacteria and incubated for a further 4 h. Values are presented as the means ± SDs from three independent experiments measured in duplicates. **, P ≤ 0.01 by two-way ANOVA with Holm-Sidak’s multiple-comparison test. (B) ELISA of IL-8 secreted by control (AS) or SENP2 siRNA-transfected A549, which were left untreated (n.i.) or infected with Kp52145 for 3 h of contact, after which, the medium was replaced with medium containing gentamicin (100 μg ml⁻¹) to kill extracellular bacteria and incubated for a further 4 h. Values are presented as the means ± SDs from three independent experiments measured in duplicates. **, P ≤ 0.01 by two-way ANOVA with Holm-Sidak’s multiple-comparison test. (C) ELISA of TNF-α secreted by empty control (pcDNA3) or SUMO1-HA plasmid-transfected MH-S, which were left untreated (n.i.) or infected with Kp52145 for 1 h of contact, after which, the medium was replaced with medium containing gentamicin (100 μg ml⁻¹) to kill extracellular bacteria and incubated for a further 4 h. Values are presented as the means ± SDs from three independent experiments measured in duplicates. ***, P ≤ 0.001 by two-way ANOVA with Holm-Sidak’s multiple-comparison test. (D) ELISA of TNF-α secreted by antagomir-transfected MH-S, which were left untreated (n.i.) or infected with Kp52145 for 1 h of contact, after which, the medium was replaced with medium containing gentamicin (100 μg ml⁻¹) to kill extracellular bacteria and incubated for a further 4 h. Values are presented as the means ± SDs from three independent experiments measured in duplicates. ***, P ≤ 0.001 by two-way-ANOVA with Holm-Sidak’s multiple-comparison tes; Neg, negative control. (E) Immunoblot analysis of IκBα and tubulin levels in lysates of empty (pcDNA3) or SUMO1-HA (pSUMO1) plasmid transfected A549 or MH-S cells, infected with Kp52145 for the indicated time points. (F) Immunoblot analysis of phosphorylated Jun N-terminal protein kinase (P-JNK), P-ERK, and P-p38 levels in lysates of pcDNA3- or pSUMO1-transfected A549 or MH-S cells infected with Kp52145 for the indicated time points. Tubulin immunoblotting was used as the loading control. n.i., noninfected control. (G) ELISA of IL-8 or TNF-α secreted by empty control (pcDNA3) or SUMO1-HA plasmid-transfected A549 or MH-S, respectively, which were treated with the NF-κB inhibitor BAY 11-7082 (5 μM, 2 h before infection) or DMSO (vehicle solution). Cells were left untreated (n.i.) or infected with Kp52145 for 1 h (MH-S) or 3 h (A549) of contact, after which, the medium was replaced with medium containing gentamicin (100 μg ml⁻¹) to kill extracellular bacteria and incubated for a further 4 h. Values are presented as the means ± SDs from three independent experiments measured in duplicates. ***, P ≤ 0.001; **, P ≤ 0.01; *, P ≤ 0.05 by two-way ANOVA with Holm-Sidak’s multiple-comparison test. (H) Percent intracellular survival in MH-S cells transfected with either empty control plasmid (pcDNA3) or SUMO1-HA tag plasmid (pSUMO1). Cells were infected with Kp52145 for 30 min (MOI, 100:1), and wells were washed and incubated with medium containing gentamicin (100 μg ml⁻¹) for 2.5 h. Intracellular bacteria were quantified by lysis, serial dilution, and viable counting on LB agar plates. Percent intracellular survival was determined by dividing the number of CFU obtained at 3 h of infection over the number obtained at 1 h and considering pcDNA3 as 100%. Values are presented as the means ± SDs from three independent experiments measured in duplicates. *, P ≤ 0.05 by unpaired t test with correction for Holm-Sidak’s multiple-comparison test. (I) Percent intracellular survival in antagomir-transfected MH-S cells. Cells were infected with Kp52145 for 30 min (MOI, 100:1), and wells were washed and incubated with medium containing gentamicin (100 μg ml⁻¹) for 2.5 h. Intracellular bacteria were quantified by lysis, serial dilution, and viable counting on LB agar plates. Percent intracellular survival was determined by dividing the number of CFU obtained at 3 h of infection over the number obtained at 1 h and considering the negative control as 100%. Values are presented as the means ± SDs from three independent experiments measured in duplicates. **, P ≤ 0.01; *, P ≤ 0.05 by one-way ANOVA with Holm-Sidak’s multiple-comparison test. In panels E and F, data are representative of at least three independent experiments.

To further investigate this notion, we examined the activation of NF-κB and MAP kinases by Klebsiella. NF-κB and MAP kinases govern the expression of inflammatory markers upon Klebsiella infection (13, 15). In the canonical NF-κB activation pathway, nuclear translocation of NF‐κB is preceded by phosphorylation and subsequent degradation of IκBα. IκBα levels were reduced in SUMO1-overexpressing infected epithelial cells and macrophages (Fig. 6E). No significant differences in the phosphorylation of MAPKs were observed (Fig. 6F), indicating that the activation status of MAPKs upon Klebsiella infection is not affected by SUMOylation. The inhibition of the canonical NF-κB pathway with BAY 11-7082 (an inhibitor of cytokine-induced IκBα phosphorylation) led to a marked decrease of cytokine release in SUMO1-overexpressing cells (Fig. 6G), connecting the increased release of inflammatory mediators in SUMO1-overexpressing cells with the activation of NF-κB.

Recently, we demonstrated that K. pneumoniae survives inside macrophages residing in the Klebsiella-containing vacuole (37). To learn whether SUMOylation plays any role in Klebsiella intracellular survival, SUMO1-overexpressing macrophages were infected, and the number of intracellular bacteria was determined 3 h postinfection. Figure 6H shows that the number of intracellular bacteria was significantly reduced in cells overexpressing SUMO. Klebsiella attachment was not affected in SUMO1-overexpressing cells (Fig. S6C), indicating that an increase in SUMOylation is detrimental for K. pneumoniae intracellular survival. Further sustaining this notion, we observed a significant decrease in bacterial numbers in macrophages transfected with let-7f and let-7g antagomirs (Fig. 6I). Klebsiella attachment was not affected in antagomir-transfected cells (Fig. S6D), demonstrating the importance of Klebsiella-induced let-7 family members for intracellular survival.

Overall, these data provide evidence that a Klebsiella-induced decrease in SUMOylation promotes bacterial infection.