Research ArticleImmunology

Nuclear PTEN enhances the maturation of a microRNA regulon to limit MyD88-dependent susceptibility to sepsis

See allHide authors and affiliations

Science Signaling  01 May 2018:
Vol. 11, Issue 528, eaai9085
DOI: 10.1126/scisignal.aai9085

PTEN and sepsis

The uncontrolled production of proinflammatory factors is a leading cause of organ dysfunction during sepsis. As well as being activated by microbial products, Toll-like receptors (TLRs) are activated by injury-associated danger signals. Almost all TLR-dependent cytokine production depends on the adaptor protein MyD88. Sisti et al. found that the abundance of the mRNA encoding the lipid and protein phosphatase PTEN was increased in mice after surgical induction of sepsis. Inhibition or knockdown of PTEN during sepsis resulted in increased inflammation, tissue injury, and mortality, which was associated with an increase in MyD88 abundance. PTEN activation induced the production of microRNAs that targeted Myd88 mRNA. Preventing the nuclear translocation of PTEN resulted in the cytosolic localization of a microRNA-processing complex and a failure to target MyD88. Together, these results suggest that the PTEN-dependent microRNA generation targets MyD88 to limit the damaging effects of sepsis.

Abstract

Sepsis-induced organ damage is caused by systemic inflammatory response syndrome (SIRS), which results in substantial comorbidities. Therefore, it is of medical importance to identify molecular brakes that can be exploited to dampen inflammation and prevent the development of SIRS. We investigated the role of phosphatase and tensin homolog (PTEN) in suppressing SIRS, increasing microbial clearance, and preventing lung damage. Septic patients and mice with sepsis exhibited increased PTEN expression in leukocytes. Myeloid-specific Pten deletion in an animal model of sepsis increased bacterial loads and cytokine production, which depended on enhanced myeloid differentiation primary response gene 88 (MyD88) abundance and resulted in mortality. PTEN-mediated induction of the microRNAs (miRNAs) miR125b and miR203b reduced the abundance of MyD88. Loss- and gain-of-function assays demonstrated that PTEN induced miRNA production by associating with and facilitating the nuclear localization of Drosha-Dgcr8, part of the miRNA-processing complex. Reconstitution of PTEN-deficient mouse embryonic fibroblasts with a mutant form of PTEN that does not localize to the nucleus resulted in retention of Drosha-Dgcr8 in the cytoplasm and impaired production of mature miRNAs. Thus, we identified a regulatory pathway involving nuclear PTEN–mediated miRNA generation that limits the production of MyD88 and thereby limits sepsis-associated mortality.

INTRODUCTION

Dysregulated inflammatory responses contribute to many pathological conditions, including sepsis (1). The inflammatory response during sepsis is initiated by the activation of Toll-like receptor (TLR) and cytokine signaling in resident cells (2). In addition to sensing microbial products, the TLR pattern recognition receptors are activated by endogenous danger signals produced during tissue injury (2). Systemic inflammatory response syndrome (SIRS), which is characterized by uncontrolled production of inflammatory mediators, is the main cause of sepsis-induced multiple organ dysfunction (1). The resulting tissue injury depends on the TLRs and cytokine receptors that initiate inflammatory pathways.

Activation and inactivation of an inflammatory response require regulation and coordination among numerous receptors, signaling pathways, and mediators. TLR family members and the interleukin-1 receptor (IL-1R) share a conserved cytoplasmic Toll–IL-1R (TIR) domain that recruits adaptor proteins, including myeloid differentiation factor 88 (MyD88), upon stimulation. MyD88 is a central adaptor protein that mediates signaling through all of the known TLRs except TLR3 (2). MyD88 is essential for cytokine production after stimulation with various ligands, such as cytokines (IL-1β and IL-18), danger-associated molecular patterns (DAMPs), and microbial pathogen-associated molecular patterns (PAMPs) (3). MyD88 deficiency increases susceptibility to a broad range of pathogens and inflammatory diseases (4). Expression of Myd88 is augmented by proinflammatory mediators (5) and is regulated at multiple levels, including transcription (6) and translation by microRNAs (miRNAs) (7, 8), which are small, endogenous RNA molecules, ~22 nucleotides in length, that regulate gene expression by targeting mRNAs for translational repression or degradation (9). Five miRNAs—miR125b, miR200b, miR200c, miR203, and miR155—inhibit MyD88 actions and TLR responsiveness (7, 8, 10, 11). Given the complexity of signaling and regulation involved in an inflammatory response, identifying molecular brakes with pleiotropic actions could potentially prevent the development of SIRS and organ damage by dampening harmful levels of inflammation.

Phosphatases prevent aberrant cellular activation by limiting the propagation of kinase signaling cascades and controlling transcriptional networks. The tumor suppressor phosphatase and tensin homolog (PTEN) is a lipid and protein phosphatase with widespread actions. The PTEN lipid phosphatase catalytic site is conventionally known to dephosphorylate phosphatidylinositol 3,4,5-trisphosphate (PIP3), a product of phosphatidylinositol 3-kinase (PI3K), which is required for the activation of protein kinase B (PKB; also known as Akt), a kinase that protects various cell types against apoptosis (12). The PTEN protein phosphatase catalytic site is less well characterized but has been proposed to regulate nuclear functions through facilitating the binding of transcription factors to promoter sequences and the maintenance of chromosomal integrity (13).

There is controversy concerning the role of PTEN in regulating the actions of TIR adaptors. Although macrophages from both PTEN heterozygous mice and mice specifically deficient in PTEN in myeloid cells are less responsive to TLR4 or TLR5 ligands (14, 15), pharmacological inhibition of PTEN enhances TLR-mediated cytokine production (16). In addition, myeloid-specific PTEN deficiency leads to increased susceptibility to lung infection by Gram-negative and Gram-positive bacteria (17, 18). Furthermore, macrophages from global PTEN−/+ mice exhibit lower lipopolysaccharide (LPS)−induced TLR4 membrane translocation compared to that in wild-type (WT) macrophages (19). PTEN is also a negative regulator of IL-1R signaling through the inhibition of nuclear factor κB (NF-κB), a transcription factor that mediates the inflammatory response (20). PTEN also impairs the killing and phagocytosis of non-opsonized bacteria and the yeast Candida albicans (18, 21).

Because PTEN is a pleiotropic molecule that regulates various pathways of the immune response, it may be a useful target for therapeutic intervention during sepsis. PTEN activation could lead to impairment of both IL-1R and TLR signaling and improve the outcome of SIRS. Although PTEN might represent an attractive target in the early stages of sepsis, where a molecular brake is needed to dampen the inflammatory response, constitutive PTEN activation might lead to an impairment in macrophage function, as observed in septic patients (22). Here, we investigated the role of PTEN in controlling the miRNA-mediated macrophage inflammatory response during sepsis, with the goal of identifying molecular pathways involved in this excessive inflammatory response.

RESULTS

PTEN inhibition drives mortality and increases lung injury in septic mice

Initially, we determined whether PTEN mRNA abundance changed during sepsis in both humans and mice. Peritoneal cells from mice undergoing polymicrobial sepsis induced by cecal ligation and puncture (CLP) had increased Pten mRNA abundance compared to that in cells from sham-treated mice (Fig. 1A). PTEN mRNA was increased in the blood leukocytes of adult septic patients who died (expired) compared to patients who survived (Fig. 1B). An increase in PTEN mRNA also occurred in pediatric patients who had septic shock compared to healthy controls (Fig. 1C), and PTEN abundance was less in pediatric septic patients who survived and developed severe comorbidities associated with sepsis (endotype A) compared to that in septic children who did not develop comorbidities (endotype B) (Fig. 1D).

Fig. 1 PTEN mRNA abundance is increased in murine sepsis and in the blood of septic pediatric and adult patients.

(A) Quantitative polymerase chain reaction (qPCR) analysis of Pten mRNA expression in murine peritoneal cells 6 hours after CLP-induced sepsis or in sham-treated controls. Data are from n = 8 mice per group and were analyzed by t test and Mann-Whitney U test. (B) qPCR analysis of PTEN mRNA expression in the blood of adult septic patients who either survived (n = 18) or died (expired; n = 22). (C) PTEN mRNA abundance in the blood of normal control subjects (n = 52) and septic pediatric patients (n = 180) as determined by qPCR analysis. (D) PTEN mRNA abundance in the blood of pediatric septic patients who either developed comorbidities (endotype A; n = 60) or not (endotype B; n = 160) as determined by qPCR analysis. Data in (B) to (D) were analyzed by analysis of variance (ANOVA), and corrections for multiple comparisons were performed using a Benjamini-Hochberg false discovery rate of 5%. Data are expressed as relative abundance to normal control subjects. For the appropriate panels, *P < 0.05 compared to sham, surviving patients, normal controls, or endotype B.

To study whether PTEN influenced the outcome of sepsis, we inhibited PTEN in vivo with the PTEN inhibitor Bpvic(OH) or knocked down PTEN expression with specific small interfering RNAs (siRNAs) before inducing CLP-mediated sepsis in mice. PTEN inhibition substantially reduced animal survival (Fig. 2A). This effect was associated with increased bacterial burden in peritoneal exudates 24 hours after CLP in comparison to that in control CLP mice administered either vehicle or scrambled siRNA (Fig. 2B). In addition, the numbers of neutrophils increased in the peritoneal cavity of CLP mice compared to that in sham-operated mice (Fig. 2C), whereas inhibition of PTEN with Bpvic(OH) or siRNA-mediated knockdown resulted in reduced numbers of neutrophils in the peritoneal cavity after CLP (Fig. 2, C and D). However, production of cytokines in peritoneal exudates 24 hours after CLP increased, including an increase in IL-1β, tumor necrosis factor–α (TNF-α), and IL-6, in PTEN-knockdown CLP mice in comparison to that in CLP mice that received scrambled siRNA (Fig. 2E).

Fig. 2 Myeloid PTEN inhibits SIRS and improves animal survival and bacterial clearance during sepsis.

(A) Survival rates of C57BL/6 mice treated with scrambled control siRNA or PTEN-specific siRNA or treated with vehicle or the PTEN inhibitor Bpvic(OH) before being subjected to moderate CLP. Survival was monitored for 7 days. Data are from n = 10 mice per group and were analyzed by log-rank (Mantel-Cox) test. (B to D) The peritoneal exudates of mice treated with the indicated inhibitors or siRNAs were isolated 6 hours after CLP-induced sepsis. Bacterial burden (B) and neutrophil recruitment (C and D) were determined. Data are from n = 7 mice per group and were analyzed by one-way ANOVA, followed by Bonferroni correction. PMN, polymorphonuclear neutrophil. (E) Concentrations of IL-1β, TNF-α, and IL-6 in the peritoneal exudates of mice treated as described in (A). Data are from n = 7 mice per group and were analyzed by one-way ANOVA, followed by Bonferroni correction. (F) Survival rates of PTENfl/fl (control) and PTENfl/fl_lysMcre mice subjected to moderate CLP-induced sepsis. Survival was monitored for 7 days. Data are from n = 14 mice per group and were analyzed by log-rank (Mantel-Cox) test. Inset: The abundance of PTEN and actin (internal control) in peritoneal cells from PTENfl/fl and PTENfl/fl_lysMcre mice was determined by Western blotting analysis. (G) Production of TNF-α, IL-1β, and IL-6 in the serum and peritoneal cavity 6 hours after CLP-induced sepsis in PTENfl/fl and PTENfl/fl_lysMcre mice. Data are from n = 7 mice per group and were analyzed by one-way ANOVA, followed by Bonferroni correction. For the appropriate panels, *P < 0.05 compared to sham mice, vehicle, scrambled siRNA control, or PTENfl/fl mice. PTEN inhib, Bpvic(OH); PC, peritoneal cavity.

Monocytes, macrophages, and neutrophils play a critical role in the pathogenesis of sepsis (23). We therefore used the Cre-Lox recombination system to cross PTENfl/fl and LysMcre mice, generating mice deficient in PTEN specifically in myeloid cells. Myeloid-specific PTEN deletion increased the mortality of septic mice (Fig. 2F), which correlated with increased TNF-α, IL-1β, and IL-6 abundance in the serum and the peritoneal cavity (Fig. 2G). To determine whether the increased bacterial load mediated organ injury, we administered antibiotic (ertapenem) to PTEN inhibitor–treated mice 1 hour after the induction of sepsis and then twice a day for 3 days. Although antibiotic treatment restored survival in septic mice treated with the vehicle control, the mice that were administered the PTEN inhibitor and antibiotic showed decreased survival, indicating that loss of PTEN function drives inflammation-mediated organ damage and mortality in septic mice (fig. S1).

Next, we determined whether PTEN was protective during peritonitis induced by methicillin-resistant Staphylococcus aureus (MRSA) (24). Mice were treated with PTEN inhibitor or vehicle control 24 hours before undergoing intraperitoneal infection with MRSA. PTEN inhibition resulted in reduced survival and substantially increased bacterial burden in the peritoneal exudates 24 hours after MRSA infection (fig. S2, A and B). This was accompanied by enhanced TNF-α production in the serum and peritoneal cavity (fig. S2C). Together, these data suggest that PTEN prevents an uncontrolled host defense response to infection and SIRS development, improving survival in two different models of sepsis. Systemic infection induced by sepsis can affect a number of organs, including the lung (25). Therefore, we analyzed PTEN abundance in lung sections and total lung homogenates 24 hours after CLP. Confocal microscopic analysis of lung sections showed that PTEN abundance increased in macrophages and bronchial epithelial cells after CLP (fig. S3A). In addition, Pten mRNA abundance was increased in total lung homogenates (fig. S3B).

To determine whether PTEN inhibition resulted in increased inflammation in the lung during sepsis, mice were treated with control or PTEN-specific siRNA before undergoing CLP, and 24 hours later, bronchoalveolar lavage (BAL) fluid and whole lungs were isolated. PTEN inhibition was accompanied by increased cell recruitment into the lungs, visualized in hematoxylin and eosin (H&E)–stained lung sections (fig. S3C). Although alveolar edema and capillary congestion, two measurements of pulmonary inflammation, were increased in CLP mice receiving scrambled siRNA, CLP mice with siRNA targeting PTEN had markedly increased amounts of these indicators of pulmonary inflammation (fig. S3, D and E). PTEN knockdown was also accompanied by the increased abundance of the proinflammatory cytokines IL-1β and TNF-α in the BAL fluid (BALF) of the CLP mice (fig. S3F). These data suggest that PTEN represses lung inflammation induced during polymicrobial sepsis.

Silencing PTEN increases MyD88 abundance and macrophage responsiveness

Phagocytes play an important role in the control of bacterial clearance and the inflammatory response. Although macrophages are required for the development of SIRS, neutrophils are key in controlling bacterial growth. Given that the lack of PTEN increased immune cell cytokine production, decreased neutrophil numbers at the site of infection, along with data (fig. S1) demonstrating that antibiotic treatment did not prevent enhanced mortality in PTEN inhibitor–treated septic mice, we anticipated that the role of PTEN in SIRS development would be restricted to macrophages. Therefore, we investigated the molecular pathways involved in increased cytokine production in myeloid-specific PTEN-deficient mice during sepsis. Specifically, we examined the role of PTEN in macrophage activation using thioglycollate-elicited peritoneal macrophages as a model of inflammatory cells. Macrophages were transfected with PTEN-specific siRNA (siPTEN) or scrambled siRNA control, which was followed by treatment with LPS, a TLR4 ligand that stimulates signaling pathways using both the MyD88 and TRIF adaptor proteins; Pam3CSK4, a TLR2 ligand that depends on MyD88 activity; or polyinosinic-polycytidylic acid [Poly(I:C)], a TLR3 ligand that signals exclusively through TRIF (26). In PTEN-silenced macrophages, LPS and Pam3CSK4, but not Poly(I:C), induced increased phosphorylation of NF-κB p65 compared to that in control siRNA–treated macrophages (Fig. 3A). In addition, LPS induced increased nitrite production in PTEN-silenced cells compared to that in control siRNA–treated cells (Fig. 3B). Further ruling out a role for PTEN in macrophage activation by TRIF-mediated TLR3 signaling, we found that PTEN deficiency did not influence LPS or Poly(I:C)-mediated phosphorylation of interferon regulatory factor 3 (IRF3) (fig. S4A), a transcription factor that is involved in TRIF-dependent gene transcription (27). In addition, PTEN deficiency did not affect the production of the TRIF-dependent cytokine interferon IFN-β and chemokine CXCL10 (fig. S4, B and C).

Fig. 3 PTEN inhibits MyD88 expression and prevents TLR2 and TLR4 activation in macrophages.

(A) Isolated macrophages were transfected with scrambled or PTEN-specific siRNA. Forty-eight hours later, the cells were stimulated with the indicated TLR agonists for 30 min. Left: Cell lysates were analyzed by Western blotting with antibodies against the phosphorylated form of NF-κB p65 (p-p65), PTEN, and actin. Right: Quantification of the normalized p-p65 band intensity from at least three independent experiments. Data are means ± SEM. (B) Macrophages transfected as described in (A) were left unstimulated or stimulated with LPS for 24 hours, and the amount of nitrite produced in the cell culture medium was determined by Griess assay. Data are means ± SEM of at least three independent experiments. (C) Left: Macrophages were transfected as described in (A), and cell lysates were analyzed by Western blotting to detect the indicated proteins. t-PTEN, total PTEN protein. Right: Quantification of relative PTEN and MyD88 band intensities from at least three independent experiments. Data are means ± SEM. (D) WT mice were injected intraperitoneally with scrambled siRNA or PTEN-specific siRNA. The abundance of PTEN and MyD88 proteins in resident peritoneal cells was determined by Western blotting analysis. Data are from two mice per group. (E) Left: Macrophages from PTENfl/fl and PTENfl/fl_lysMcre mice were challenged with LPS for the indicated times, and MyD88 abundance in the cell lysates was determined by Western blotting analysis. Right: Quantification of relative MyD88 band intensities from at least three independent experiments. Data are means ± SEM. For the appropriate panels, *P < 0.05 compared to scrambled siRNA control group or PTENfl/fl; #P < 0.05 compared to LPS-stimulated PTEN siRNA or PTENfl/fl_lysMcre compared to scrambled siRNA control group or PTENfl/fl macrophages. In all circumstances, at least three independent experiments were performed, and data were analyzed by one-way ANOVA, followed by Bonferroni correction.

IL-1 receptor–associated kinase 4 (IRAK4) and inhibitor of NF-κB kinase α (IKKα) are kinases that are downstream of MyD88 (28, 29). To determine whether PTEN directly dephosphorylated IRAK4 or IKKα, we performed an in-blot PTEN activity assay. Lysates from macrophages challenged with LPS, peptidoglycan, curdlan (a dectin-1 ligand), or Poly(I:C) for 30 min were transferred to membranes and then incubated with buffer alone, recombinant PTEN, or the positive control alkaline phosphatase. Membranes were probed for phosphorylated IRAK4 or IKKα. Membranes treated with buffer or recombinant PTEN yielded a prominent band for phosphorylated IRAK4 and IKKα in response to LPS (fig. S5). The phosphorylated proteins were undetectable in the alkaline phosphatase–treated membranes. When membranes were stripped and reprobed for actin, there was no observed difference among the treatment groups, excluding a role for PTEN protease activity. These data suggest that PTEN inhibits MyD88-dependent signaling through an effect on MyD88, rather than downstream effects on kinases involved in NF-κB activation.

To determine whether PTEN altered the abundance of TLR4 and the adaptor TIR, we compared the abundances of various adaptors and TLR4 in macrophages treated with scrambled siRNA or PTEN-targeted siRNA. PTEN-silenced macrophages had an increased abundance of MyD88, but not of TLR4 or other adaptors, compared to that in control siRNA–treated macrophages (Fig. 3C). Peritoneal cells from mice treated in vivo with siRNA targeting PTEN or from PTENflf/fl_lysMcre mice had an increased abundance of MyD88 in comparison to that of cells from control siRNA–treated or PTENfl/fl mice (Fig. 3, D and E). Furthermore, the LPS-induced increase in MyD88 in macrophages from mice in which PTEN was deleted (PTENflf/fl_lysMcre mice) was greater than that in cells from WT (PTENfl/fl) mice (Fig. 3E). We confirmed that PTEN reduced Myd88 mRNA abundance in macrophages transduced with active PTEN-expressing adenovirus (fig. S6, A and B). In transduction experiments with adenovirus expressing active PTEN, adenovirus expressing PTEN lacking lipid phosphatase activity (G129E mutant), or adenovirus expressing a dominant-negative form of PTEN that lacks both protein and lipid phosphatase activity (C214S mutant), we found that MyD88 protein abundance was undetectable in cells expressing active PTEN and was substantially reduced in cells expressing the G129E mutant (fig. S6C). Therefore, these results suggest that the lipid phosphatase activity of PTEN inhibits the production of MyD88. Together, these data (Fig. 3 and fig. S6) indicate that the lipid phosphatase activity of PTEN limits homeostatic and inducible MyD88 production and inhibits TLR and IL-1R signaling in macrophages.

Increased MyD88 in PTEN-deficient mice enhances mortality during sepsis

To further test our hypothesis that increased susceptibility to sepsis in PTEN-deficient mice depended on increased MyD88 abundance, we measured Myd88 mRNA abundance in peritoneal cells from septic mice treated with the PTEN inhibitor. Myd88 mRNA abundance was decreased 24 hours after sepsis, and not only PTEN inhibition reversed this effect, but also the abundance of the Myd88 mRNA exceeded that of the sham-operated mice (Fig. 4A). We performed rescue experiments to determine whether blocking MyD88 could overcome the enhanced mortality observed in PTEN-silenced mice. Administration of a MyD88-blocking peptide 1 hour after CLP substantially improved the survival of the PTEN-silenced mice (Fig. 4B), which was accompanied by decreased bacterial burden (Fig. 4C) and reduced proinflammatory cytokine production (Fig. 4D). We confirmed the specificity of the MyD88-blocking peptide in macrophages stimulated with LPS, Pam3CSK4, or Poly(I:C). Although all these agonists induced the production of nitrite (a nitric oxide metabolite), the MyD88-blocking peptide inhibited TLR4 and TLR2, but not TLR3, signaling, demonstrating the specificity of the peptide in blocking MyD88 action (fig. S7). These data point to a PTEN-MyD88 axis that controls inappropriate inflammation in polymicrobial sepsis. Furthermore, these results demonstrate an epistatic relationship between PTEN and MyD88, promoting a signaling balance to maintain homeostatic inflammatory responses during pathogen challenge.

Fig. 4 Blocking MyD88 prevents enhanced mortality, bacterial load, and SIRS in siPTEN-challenged mice.

(A) C57BL/6 mice were treated with PTEN inhibitor or vehicle control for 24 hours before being subjected to CLP-induced sepsis. Myd88 mRNA abundance was then determined in peritoneal cells 24 hours after sepsis by qPCR analysis. The abundance of Myd88 mRNA is expressed relative to that in sham-operated mice. Data are means ± SEM of n = 4 to 6 mice per group and were analyzed by one-way ANOVA, followed by Bonferroni correction. (B) WT mice were treated with scrambled siRNA control or PTEN-specific siRNA before undergoing CLP, which was followed by treatment with MyD88 peptide inhibitor for 1 hour. Survival was monitored for 7 days. Data are from n = 13 mice per group and were analyzed by log-rank (Mantel-Cox) test. (C and D) WT mice were treated as described in (B), and bacterial burden (C) and cytokine production (D) were determined in peritoneal exudates 24 hours after CLP. Data are means ± SEM of at least five mice per group and were analyzed by one-way ANOVA, followed by Bonferroni correction. For the appropriate panels, *P < 0.01 compared to sham, vehicle control, scrambled siRNA control, or PTENfl/fl; #P < 0.01 compared to PTEN siRNA group; &P < 0.05 compared to PTEN siRNA plus scrambled control. CFU, colony-forming unit.

PTEN regulates MyD88 expression in macrophages

Myd88 expression and MyD88 protein production are controlled at multiple levels, including signal transducer and activator of transcription 1 (STAT1) and NF-κB activation (6) and mRNA degradation by specific miRNAs (8, 10). To determine whether PTEN regulated MyD88 expression through an effect on STAT1, we determined STAT1 abundance and activation, measured as STAT1 phosphorylation on Ser727 or Tyr701, in PTEN-knockdown or siRNA control macrophages. We did not detect any change in total or phosphorylated STAT1 abundance in PTEN-silenced cells compared to that in cells treated with the control siRNA (fig. S8A). When alveolar macrophages were transduced with adenovirus expressing active PTEN, dominant-negative PTEN (C124S), or a lipid phosphatase–deficient PTEN (G129E), we saw no effect on STAT1 abundance or Tyr701 phosphorylation. However, overexpression of active PTEN inhibited the phosphorylation of STAT1 on Ser727 in a manner that was independent of its lipid phosphatase activity (fig. S8B).

We confirmed the lack of involvement of PTEN in STAT1 activation by showing that PTEN-deficient macrophages from the PTENflf/fl_lysMcre mice produced similar amounts of the STAT1-dependent chemokine CXCL1 in response to either LPS or Poly(I:C) (fig. S8C). To test the possibility that PTEN inhibited Myd88 expression by controlling the activity of other transcription factors involved in Myd88 expression, such as NF-κB and STAT3, we compared macrophages in which PTEN was knocked down or to macrophages treated with control siRNA and that had been treated with the Janus kinase 2 (JAK2) inhibitor AG490 (6), the NF-κB inhibitor (a p65 blocking peptide) (30), or the STAT3 peptide inhibitor STATTIC (31). Inhibition of these transcription factors did not influence the PTEN-mediated effects on Myd88 mRNA abundance in macrophages (fig. S8D). These results suggested that miRNAs might be involved in the PTEN-mediated regulation of MyD88 abundance in macrophages. Therefore, we evaluated the abundance of known inflammatory mature miRNAs, using an immunopathology-focused array (10), in peritoneal macrophages treated with siRNA targeting PTEN or scrambled siRNA. Of the 88 transcripts analyzed, 30 mature miRNAs were decreased in abundance in PTEN-silenced cells, including miR19a and miR19b, miR21, miR125a and miR12b, miR146a and miR146b, and miR203 (Fig. 5A), all of which have been reported to regulate TLR signaling (10, 32). In addition, 10 miRNAs known to regulate TLR responses (3336) were enhanced in PTEN-silenced cells, including let7g and miR155 (Fig. 5A). The remaining 48 miRNAs tested did not change in abundance. We verified the abundances of individual miRNAs in PTEN-silenced peritoneal cells by qPCR analysis (Fig. 5B). In addition, macrophages from naïve mice transduced with an adenovirus expressing constitutively active PTEN displayed increased miR125b and miR146b abundance and decreased miR155 abundance in comparison to cells transduced with control adenovirus (Fig. 5C).

Fig. 5 PTEN lipid phosphatase activity increases the abundance of miRNAs involved in regulating Myd88 expression and TLR activation in macrophages.

(A) Elicited macrophages were transfected with scrambled siRNA control or PTEN-specific siRNA. Twenty-four hours later, miRNAs were isolated and an miRNA-focused array was performed. (B) Elicited macrophages were transfected with PTEN-specific siRNA as described in (A), and the abundances of the indicated miRNAs were determined by qPCR analysis. Data are expressed relative to the miRNA abundance in the scrambled control siRNA samples. Data are means ± SEM. (C) Elicited macrophages were transduced with PTEN-expressing adenovirus (Ad) or empty vector. Forty-eight hours later, the abundances of indicated miRNAs were determined by qPCR analysis. Abundance relative to that in cells transfected with empty vector adenovirus (EV-Ad) is expressed as means ± SEM. (D) Macrophages from PTENfl/fl and PTENfl/fl_lysMcre mice were isolated, and the indicated miRNAs were detected by qPCR. Abundance relative to that in PTENfl/fl macrophages is expressed as means ± SEM. (E) CLP-induced sepsis was performed in PTENfl/fl and PTENfl/fl_lysMcre mice, and miRNA abundance was measured in peritoneal cells 24 hours after CLP by qPCR analysis. Abundance relative to that in the control siRNA–treated sham mice is expressed as means ± SEM. (F) Macrophages from PTENfl/fl and PTENfl/fl_lysMcre mice were treated with the PI3K inhibitor wortmannin (Wortm) or the mTOR inhibitor rapamycin (Rapam) for 24 hours, and the abundance of miR125b was determined by qPCR. Abundance relative to that in PTENfl/fl macrophages treated with vehicle control was expressed as means ± SEM. (G) Elicited peritoneal macrophages from WT mice were transfected with PTEN-specific siRNA or scrambled siRNA as described in (A) and with the indicated miRNA mimics. Twenty-four hours later, Myd88 mRNA abundance was determined by qPCR analysis. Abundance relative to that in macrophages treated with scrambled control siRNA and the mimic control is expressed as means ± SEM. (H) Macrophages from C57BL/6 mice were treated as described in (G) and challenged with LPS for 24 hours. Nitrite concentrations in the cell culture medium were determined by Griess assay. Data are means ± SEM of at least five mice per group repeated at three independent times and were analyzed by one-way ANOVA, followed by Bonferroni correction. *P < 0.05 compared to the scrambled siRNA control or PTENfl/fl mice; #P < 0.05 for LPS-stimulated PTEN siRNA compared to scrambled siRNA control or PTENfl/fl macrophages; &P < 0.05 compared to miRNA mimic alone.

To determine whether PTEN regulated the abundance of miRNAs in myeloid cells in vivo and to confirm the findings from our siRNA experiments, we analyzed the abundances of mature miRNAs in peritoneal macrophages from PTENfl/fl or PTENfl/fl_lysMcre mice. The abundances of miR125b, miR19a, and miR146b in peritoneal cells from PTEN-deficient mice were reduced, whereas there was no change in the abundance of miR134 (Fig. 5D). We observed a similar effect on the abundances of miR125b, miR19a, and miR134 in alveolar macrophages from PTENfl/fl_lysMcre mice (fig. S9). Examining PTEN-controlled miRNA abundance during sepsis showed that miR125a and miR125b, miR181a, and miR21 abundance increased after CLP in WT mice and that PTEN deficiency prevented miRNA production in both the sham and CLP groups (Fig. 5E). These data suggest that PTEN regulates miRNA production under both basal and inflammatory conditions.

The PTEN lipid phosphatase domain controls signaling through the PI3K-Akt pathway (12). Akt controls activation of mammalian target of rapamycin (mTOR), which is crucial for many cell processes, including mRNA translation, de novo nucleotide synthesis, and autophagy (12). To investigate whether the effects of PTEN on miRNA abundance depended on PI3K or mTOR activation, macrophages from PTENfl/fl and PTENfl/fl_lysMcre mice were treated with the PI3K inhibitor wortmannin or the mTOR inhibitor rapamycin. In WT cells, rapamycin, but not wortmannin, inhibited the production of miR125b (Fig. 5F), miR21, and miR203 (fig. S10, A and B), indicating a role for mTOR in regulating basal miRNA production. However, only PI3K inhibition prevented the reduction in miR203 abundance in PTENfl/fl_lysMcre macrophages, suggesting that the influence of PTEN on the production of this miRNA depended on its lipid phosphatase activity (fig. S10A). In contrast, PI3K inhibition by wortmannin did not rescue miR21 abundance in PTENfl/fl_lysMcre cells, indicating a role for PTEN in controlling this miRNA that is independent of the lipid phosphatase activity (fig. S10B).

We further analyzed changes in miRNA abundance to evaluate whether they accounted for enhanced Myd88 mRNA abundance and MyD88-mediated enhancement of macrophage activation in PTEN-silenced cells. Macrophages were first transfected with siRNA against PTEN or scrambled control siRNA. Forty-eight hours later, the cells were transfected with the miRNA mimics miR155, miR125b, and miR203, which were previously shown to be modulated by PTEN silencing for 24 hours. Myd88 mRNA abundance was then analyzed by qPCR. Although the miR155 and miR146b mimics had no effect on the PTEN-mediated reduction in Myd88 mRNA abundance, miR125b and miR203 decreased Myd88 mRNA only in PTEN-silenced cells (Fig. 5G). Confirming that miR125b and miR203 directly controlled Myd88 expression, the reduced expression of a luciferase reporter with the 3′ untranslated region (UTR) of Myd88 was observed in the macrophage cell line transfected with miR125b and miR203 alone, and introduction of both miRNAs further reduced expression (fig. S11A). In addition, Myd88 transcripts coprecipitated with Ago2, a member of the RNA-induced silencing complex (RISC), in macrophages transfected with miR125b or miR203 mimics (fig. S11B). Mimics of miR125b or miR203 also prevented the enhanced LPS-induced nitrite production in PTEN-silenced macrophages, an effect that was enhanced by combining both mimics (Fig. 5H). Together, these results indicated a previously unknown role for PTEN lipid phosphatase activity with both homeostatic and inducible production of miRNAs involved in regulating macrophage Myd88 expression and TLR signaling.

Nuclear PTEN regulates miRNA maturation

The abundance of miRNAs is regulated at numerous levels, including transcription, processing, and subcellular localization (37). To determine whether PTEN localization and the lipid phosphatase domain regulated miRNA production, we transfected human embryonic kidney (HEK) 293 cells with lentiviral vectors expressing the following PTEN constructs: (i) C124S, a dominant-negative PTEN construct (21); (ii) G129E, a construct lacking lipid phosphatase activity (21); (iii) K163R, a construct lacking an acetylation site important for PTEN membrane translocation (38); (iv) K254R, a construct that cannot localize to the nucleus (39); and (v) K289R, a construct that exhibits reduced nuclear PTEN (40). The expression of these constructs in HEK 293 cells was first confirmed by Western blotting analysis (fig. S12). Active PTEN increased the abundance of miR125b and miR21 and decreased the abundance of miR155 (Fig. 6A). PTEN lacking lipid phosphatase activity (G129E) did not affect the abundance of these miRNAs, and the K254R and K289R mutant forms of PTEN, which cannot localize properly to the nucleus, did not alter the abundance of miR125b, miR155, or miR21 (Fig. 6A). These data suggest that the nuclear localization of PTEN and its lipid phosphatase activity are necessary for miRNA processing.

Fig. 6 Nuclear PTEN drives miRNA processing.

(A) PTEN−/− MEFs were transduced with retrovirus-containing empty vector control (TRE) or the indicated WT or mutant PTEN constructs. Expression of the indicated miRNAs was determined by qPCR analysis. Abundance relative to that in untransfected cells (TRE) was expressed as means ± SEM of n = 3 independent experiments. Data were analyzed by one-way ANOVA, followed by Bonferroni correction. (B) Macrophages were isolated from PTENfl/fl and PTENfl/fl_lysMcre mice, and the abundances of Drosha, Dgcr8, Xpo5, and Dicer 1 mRNAs were determined by qPCR analysis. Abundance relative to that in macrophages from PTENfl/fl mice was expressed as means ± SEM of n = 4 mice per group. (C) The abundances of the indicated primary miRNAs in macrophages from PTENfl/fl and PTENfl/fl_lysMcre mice were determined by qPCR analysis. Abundance relative to that in macrophages from PTENfl/fl mice was expressed as means ± SEM of n = 4 mice per group. (D) Macrophages were treated with scrambled control or PTEN-specific siRNA, and the abundances of the indicated primary miRNAs were determined by qPCR analysis. Abundance relative to that in macrophages treated with control siRNA was expressed as means ± SEM of n = 5 mice per group. (E) Left: Macrophages from C57BL/6 mice were incubated with antibodies recognizing Drosha, Dgcr8, Dicer, or PTEN, as indicated, and visualized by confocal microscopy. Right: Overlap coefficients between PTEN/DICER, PTEN/Drosha, and PTEN/Dgcr8. Data are means ± SEM of three independent experiments, with ~100 cells analyzed in each experimental group. 4′,6-Diamidino-2-phenylindole (DAPI) staining is in blue. Each field is representative of 100 cells examined (original magnification, ×400) from each of three independent experiments with the values of the DICER/PTEN association set as 1. (F) Left: In situ PLA of PTEN/Drosha or PTEN/Dgcr8 complexes in elicited macrophages from PTENfl/fl and PTENfl/fl_lysMcre mice. PLA complexes are shown in red, and nuclei are shown in blue. Right: Numbers of specks/nucleus in at least 100 cells expressed as means ± SEM of at least three independent experiments. (G) Left: PTEN was immunoprecipitated (IP) from total macrophage cell lysates from PTENfl/fl and PTENfl/fl_lysMcre mice, and samples were then analyzed by Western blotting with antibodies against the indicated proteins. Right: The relative densities of bands corresponding to PTEN, Dgcr8, Drosha, and DICER were determined by densitometry analysis. Data are means ± SEM of three individual experiments, with the values of the PTENfl/fl control group set as 1. IB, immunoblotting. (H) Left: Macrophages from PTENfl/fl and PTENfl/fl_lysMcre mice were incubated with antibodies recognizing Drosha, Dgcr8, or Dicer, as indicated, and visualized by confocal microcopy. Confocal images were captured under identical settings to enable comparison of staining intensities. Images are from one experiment, which are representative of five independent experiments. Right: Overlap coefficients between DAPI (nuclear, blue) and Dgcr8, Drosha, or DICER (red). Data are means ± SEM of three independent experiments, with ~100 cells analyzed in each experimental group. Values from PTENfl/fl cells were set to 1. (I) Left: Nuclear and cytosolic fractions of macrophage lysates from PTENfl/fl and PTENfl/fl_lysMcre mice were isolated and analyzed by Western blotting with antibodies against the indicated proteins. Right: The relative intensities of bands corresponding to DGCR8, Drosha, and histone H3 were determined by densitometry analysis. Data are means ± SEM of three individual experiments, with the values of the cytosolic proteins from the PTENfl/fl control group set as 1. For the appropriate panels, *P < 0.05 compared to empty vector, control scrambled siRNA, or PTENfl/fl mice.

Because PTEN regulated the production of nuclear miRNAs, we investigated whether PTEN influenced the abundance of the enzymes involved in miRNA maturation, such as Drosha and Dgcr88 (proteins involved in the generation of precursor miRNA), Xpo5 (exportin 5; a protein that transports precursor miRNAs to the cytoplasm), and Dicer (the enzyme that generates mature miRNA) (41). We observed that the expression of these genes was similar in PTEN-deficient and WT macrophages (Fig. 6B). Next, we studied whether PTEN controlled the expression of primary miRNAs (several hundred nucleotide-long RNAs that will generate precursor miRNAs) in macrophages lacking PTEN or in WT macrophages treated with siRNA targeting PTEN or scrambled siRNA. PTEN deficiency increased the expression of primary miRNAs for miRNA125b, miRNA155, miRNA203, Let-7e, and Let-7d (Fig. 6, C and D), suggesting that PTEN participates in the maturation of miRNAs.

Next, we performed imaging experiments to determine whether PTEN interacted with Drosha and Dgcr8 in macrophages. Confocal microscopic analysis of macrophages from WT mice revealed nuclear and cytoplasmic colocalization of PTEN with Drosha and Dgcr8 (Fig. 6E). However, PTEN did not colocalize with Dicer in the cytoplasm of macrophages (Fig. 6E). Proximity ligation assay (PLA) confirmed that PTEN physically associated with Drosha and Dgcr8 in the nucleus of unstimulated macrophages from WT (PTENfl/fl) mice, but not in macrophages from PTENfl/fl_lysMcre mice (Fig. 6F). PTEN, Drosha, and Dgcr8 coimmunoprecipitated from macrophage lysates, confirming the results of the microscopy experiments (Fig. 6G). These results suggest that PTEN controls the nuclear translocation of Drosha and Dgcr8. Therefore, we investigated the localization of Drosha and Dgcr8 in macrophages from both PTENfl/fl and PTENfl/fl_lysMcre mice. We found that Dicer was located in the cytoplasm, but Drosha and Dgcr8 were localized mainly in the nucleus of WT macrophages. PTEN deficiency (PTENfl/fl_lysMcre mice) impaired the nuclear localization of Drosha and Dgcr8 without affecting the subcellular distribution of Dicer as evidenced by confocal microscopy (Fig. 6H) and by cellular fractionation of cytosolic and nuclear proteins (Fig. 6I). Together, these data indicate a role for PTEN in miRNA processing by directly regulating the distribution of miRNA-processing enzymes in the nucleus.

DISCUSSION

Through various cellular, molecular, genetic, and pharmacologic approaches, we identified a previously uncharacterized role for PTEN in regulating the maturation of nuclear miRNAs, which plays an important role in host innate immunity and decreased mortality and comorbidities associated with sepsis. In summary, our results demonstrated the following findings. First, Pten mRNA expression was enhanced in murine sepsis and in the blood of pediatric and expired adult septic patients. Furthermore, in septic patients who developed severe comorbidities, PTEN abundance was greatly decreased. Second, genetic and pharmacologic inhibition of PTEN resulted in animal mortality during sepsis, which was accompanied by increased bacterial load, inflammatory cytokine production, and lung injury. Third, PTEN deficiency led to exaggerated MyD88 abundance, and blocking the effects of MyD88 in PTEN-null mice prevented increased mortality and inflammatory responses during sepsis. Fourth, the actions of PTEN on MyD88 were due to the increased expression of miRNAs involved in macrophage activation in both naïve and septic mice. Fifth, PTEN lipid phosphatase activity and nuclear localization regulated miRNA expression. Finally, PTEN associated with and enabled the nuclear translocation of Drosha and Dgcr8. Together, our data identified previously uncharacterized mechanisms by which the tumor suppressor PTEN regulated miRNAs, affecting macrophage activation and improving sepsis outcomes.

Sepsis is characterized by the early development of SIRS, which leads to organ damage (1). Therefore, it is important to identify molecular pathways that prevent the development of SIRS and inhibit the organ damage induced during sepsis. We observed that PTEN expression was increased during sepsis, which suggests a regulatory role for PTEN in preventing overwhelming inflammation, which is mediated by regulating MyD88 abundance and preventing of aberrant TLR/IL-1R activation. When PTEN was deleted in phagocytes, a high mortality rate was observed during sepsis, similar to that seen in mice treated with a PTEN inhibitor or PTEN-specific siRNA. This indicates that PTEN in myeloid cells mediates the protective effects during sepsis.

In other studies, we demonstrated that genetic and pharmacological inhibition of PTEN improved the phagocytosis of C. albicans (21) and increased bacterial killing by alveolar macrophages infected with immunoglobulin G (IgG)–opsonized Klebsiella pneumoniae (42). PTEN also inhibits immune responsiveness to Gram-negative lung infections in vitro and in vivo (17, 18). Given that neutrophil migration during infection is essential for the control of sepsis (43), we speculate that the increased bacterial counts in PTEN-deficient mice were due to defects in neutrophil migration. In previous studies, PTEN deficiency led to impaired neutrophil migration (44), and severe sepsis decreased neutrophil migration to the site of infection in a manner dependent on CXCR2 down-regulation (43, 45, 46). However, how the effect of PTEN on CXCR2 regulates neutrophil migration to the focus of infection remains to be determined. Because PTEN inhibition decreases neutrophil recruitment to the site of infection, these data led us to hypothesize that macrophages are the key cells involved in the protective effect of PTEN during sepsis. To discriminate between PTEN effects on inflammation and bacterial growth, we cotreated mice with both the PTEN inhibitor Bpvic(OH) and an antibiotic (ertapenem). We found that treating mice with the antibiotic did not rescue the deleterious effect of the PTEN inhibitor with regard to animal lethality. These data suggest that PTEN deficiency leads to an overwhelming inflammatory response, organ damage, and death in response to systemic infection (fig. S1).

This study used various in vitro and in vivo techniques to identify MyD88 as a major target for PTEN. PTEN exerts both positive and negative actions during MyD88-dependent TLR/IL-1R activation. Genetic and pharmacological PTEN inhibition leads to enhanced LPS responsiveness in macrophages (47). Furthermore, PTEN deficiency increases TLR and IL-1R responsiveness in B cells, lung fibroblasts, and epithelial cells (48, 49). Aksoy et al. (50) demonstrated that PTEN+/− macrophages exhibit decreased membrane translocation of TLR4 during LPS stimulation. Here, we observed that MyD88-dependent phosphorylation of NF-κB p65 and nitrite production increased in macrophages after siRNA-mediated knockdown of PTEN. Although our in-blot activity assay showed that PTEN did not dephosphorylate IRAK4 or IKKα, the possibility that PTEN influences other downstream inflammatory effectors or directly influences downstream kinases involved in NF-κB activation cannot be ruled out.

We also further investigated whether PTEN controlled TRIF-dependent macrophage activation. PTEN deficiency did not influence either Poly(I:C)-induced IFN-β production or macrophage IRF3 phosphorylation. However, Li et al. (51) showed that PTEN increased the IFN-α/β–induced nuclear translocation of IRF3 and improved antiviral responses in mouse embryonic fibroblasts (MEFs). The difference between our data and those of the study by Li et al. could be due to the stimuli used, Poly(I:C) versus Sendai virus or IFN-α/β, or the cell lines tested, peritoneal macrophages versus HEK 293 cells.

MyD88 expression is controlled through transcriptional and posttranscriptional events (6, 10). PTEN did not affect the expression or phosphorylation of STAT1, a transcription factor responsible for MyD88 expression (6). Furthermore, pharmacologic approaches ruled out a possible participation of the transcription factors STAT1, STAT3, and NF-κB in PTEN-mediated inhibition of MyD88 expression. However, we found that PTEN inhibited MyD88 expression by directly controlling the expression of miRNAs that target MyD88 mRNA. PTEN deletion decreased the expression of many miRNAs, including miR125b and miR203, which inhibit MyD88 expression (7, 10). PTEN was also required for optimal miRNA expression in both homeostatic conditions and uncontrolled inflammation, such as sepsis. These data led us to study the role of PTEN in miRNA generation. We found that PTEN deficiency enhanced the expression of some miRNAs, including miRNAs involved in inflammatory responses, such as miR155 and let7g. The mechanisms involved remain to be determined, but PTEN may inhibit transcription factors that are involved in miRNA expression, including AP-1 and PU.1 (10, 52).

PTEN also mediated miRNA processing by directly associating with and facilitating the translocation of Drosha and Dgcr8 into the nucleus. PTEN shuttles between cellular compartments including the cytoplasm, mitochondria, plasma membrane, endoplasmic reticulum, and nucleus (53). PTEN is found in the nucleus in many normal and cancerous cells and tissues, and some of the molecular mechanisms regulating PTEN nuclear localization have been described previously (53). Although PTEN does not have a defined nuclear localization sequence, it has been proposed that either simple diffusion, a putative cytoplasmic localization signal, active shuttling by the RAN guanosine triphosphatase or major vault protein (MVP), phosphorylation-dependent shuttling, or monoubiquitylation- and SUMOylation-dependent nuclear import is responsible for the cellular location of PTEN (53). An alteration in the nuclear versus cytoplasmic distribution of PTEN has been linked with various diseases, including cancer and brain injury, and the mechanisms by which PTEN is localized in the nucleus in macrophages are under investigation.

The role of the phosphatase activity of PTEN is controversial, and protein-protein interactions have been suggested to play a major role (53). PTEN SUMOylation sites (K254R and K289R) are responsible for its membrane association and nuclear translocation, and this modification is also essential for the inhibition of the PI3K-Akt signaling pathway (39), a finding that correlates with our results showing that both lipid phosphatase activity and SUMOylation drive miRNA processing. Studies showed that SUMOylation is required for optimal miRNA processing and efficacy by influencing both Dicer (54) and the Argonaute-2/RISC complex activity (55, 56). Furthermore, SUMOylation increases the stability of Dgcr8 protein and promotes its association with pri-miRNAs (57, 58). However, SUMOylation does not influence the association between Drosha and Dgcr8 (57). Whether SUMOylation influences Drosha activity is unknown. Although PTEN SUMOylation positively controls PTEN phosphatase activity and nuclear translocation, it remains to be determined whether SUMOylation is required for miRNA expression and processing in macrophages in vivo. Furthermore, whether changes in PTEN SUMOylation also influence inflammatory responses needs to be studied. A well-studied PTEN partner, the tumor suppressor p53, also associates with Drosha and facilitates miRNA processing (59). Whether PTEN is part of a complex formed by p53 and Drosha and other proteins involved in miRNA processing remains to be determined.

Here, we dissected the PTEN-associated molecular pathways involved in sepsis. Our results identified a previously uncharacterized regulatory pathway that drives miRNA expression. This work has translational implications given the fact that both PTEN and miRNAs are key players in mediating or exacerbating a myriad of diseases, including cancer, asthma, atherosclerosis, and arthritis (6067). PTEN and miRNAs may work in concert to prevent aberrant inflammatory responses in different disease settings.

MATERIALS AND METHODS

Study design

In vitro and in vivo experiments were designed to test the hypothesis that PTEN prevents SIRS development during sepsis by enabling the expression of miRNAs that target MyD88 expression in phagocytes. For all experiments, the minimum sample size was determined to detect a difference between group means of twice the observed SE, with a power of 0.8 and a significance level of 0.05, using a power and sample size calculator (www.statisticalsolutions.net/pssZtest_calc.php). The calculated minimum sample sizes ranged from three to five, depending on the experiment. The average sample size for mouse studies was five per group. All samples were randomized but not blinded.

Patients and data collection

The institutional review boards of each participating institution, Cincinnati Children’s Hospital Medical Center and Yale School of Medicine, approved the study protocol. Children 10 years of age or younger admitted to the pediatric intensive care unit (ICU) who met pediatric-specific criteria for septic shock were eligible for enrollment (68). Age-matched controls were recruited from the ambulatory departments of participating institutions using published inclusion and exclusion criteria (68). Adults (n = 40) aged 19 years or older admitted to the medical ICU at Yale New Haven Hospital, Yale School of Medicine, who met the criteria for sepsis (69) were enrolled for RNA isolation obtained from whole blood, based on approval by the Human Investigational Committee. Baseline demographics of the adult septic patients, which included 20 survivors and 20 nonsurvivors, are outlined in table S1.

Mice

Female and male C57BL/6 mice (8 weeks old; weight, 18 to 23 g) were obtained from The Jackson Laboratory. PTENfl/fl and PTENfl/fl_lysMcre mice were bred at Indiana University School of Medicine and Vanderbilt University Medical Center. Mice were maintained according to National Institutes of Health (NIH) guidelines for the use of experimental animals with the approval of the Indiana University and Vanderbilt University Medical Center Committee for the Use and Care of Animals.

Transcriptomics data

PTEN mRNA expression data were extracted from an existing, microarray-based transcriptome database. Details of the study protocol were previously published (68), and the data are deposited in the Gene Expression Omnibus database (accession no. GSE66099). Briefly, the data reflect children with septic shock (n = 180) and normal controls (n = 52). The RNA used for microarray analyses was derived from whole blood obtained within the first 24 hours of meeting the criteria for septic shock. The original consent form for this protocol allows for the secondary analysis of clinical and biological data. The existing normalized transcriptome data were analyzed using GeneSpring GX 7.3 software (Agilent Technologies). All signal intensity–based data were used after robust multiarray average normalization, which specifically suppresses all but statistically significant variation among lower-intensity probe sets (70). All chips representing septic shock samples were then normalized to the respective median values of controls on a per-gene basis. Differences in mRNA abundance between study groups were determined using ANOVA, and corrections for multiple comparisons were determined using a Benjamini-Hochberg false discovery rate of 5%.

Polymicrobial sepsis

Sepsis was induced by CLP, as previously described (71), using a 26-gauge needle to induce moderate sepsis. Survival was monitored at 12-hour intervals for 7 days after CLP surgery. Mice showing signs of imminent death (inability to maintain an upright position, ataxia, or tremor or agonal breathing) were euthanized. For other protocols, mice were euthanized 6 or 24 hours after CLP surgery.

In vitro and in vivo pharmacological and genetic treatments

For in vivo experiments, mice were treated for 1 hour with the PTEN inhibitor Bpvic(OH) (1.5 mg/kg) or vehicle before being subjected to CLP. Sense and antisense strands of murine PTEN-specific siRNA [5′-AGAGAUCGUUAGCAGAAACTT-3′ (sense) and 5′-GUUUCUGCUAACGAUCUCUTT-3′ (antisense)] and scrambled (control) siRNA [5′-GCGCGCUUUGUAGGAUUCGTT-3′ (sense) and 5′-CGAAUCCUACAAAGCGCGCTT-3′ (antisense)] were used as previously described (72). All siRNAs were synthesized as 2′-deprotected, duplexed, desalted, and purified siRNA form (Integrated DNA Technologies). To avoid immune stimulation by synthetic siRNA, we incorporated 2′-O-methyl (2′OMe) uridine or guanosine nucleosides into one strand of the siRNA duplex. The siRNAs were conjugated with in vivo-jetPEI (Polyplus transfection) agent with 5% glucose. C57BL/6 mice were treated 48 hours before undergoing moderate CLP. In some experiments, mice were treated with the siRNAs and then treated with MyD88 blocking peptide (100 μg/kg; DRQIKIWFQNRRMKWKKRDVLPGT) or the scrambled control (DRQIKIWFQNRRMKWKK) with a palmitoyl modification in the N terminus 1 hour after undergoing CLP. Furthermore, WT mice were pretreated with the PTEN inhibitor as described earlier, which was followed by CLP. Septic mice were then treated or not with the antibiotic ertapenem sodium (30 mg/kg, subcutaneously) 1 hour after CLP and then every 12 hours thereafter. For in vitro experiments, macrophages were treated with the JAK2 inhibitor AG490 (1 μM), the STAT3 inhibitor peptide STATTIC V (20 μM), or the NF-κB p65 peptide inhibitor (NF-κB DRQIKIWFQNRRMKWKKQLRRPSDRELSE) or scrambled control (DRQIKIWFQNRRMKWKK) (100 μM each), and Bpvic(OH) (10 nM), PTEN siRNA, or scrambled control siRNA (30 nM each), as previously described (21). Macrophages (1 × 106) were incubated with LPS (100 ng/ml), Pam3CSK4 (2 μg/ml), or Poly(I:C) (50 μg/ml) for 1 hour before being analyzed by Western blotting or for 24 hours before being analyzed by qPCR assay.

Cell harvesting

Elicited macrophages were harvested from the peritoneal cavities of mice by lavage with phosphate-buffered saline (PBS) 4 days after the injection of 2 ml of 3% thioglycollate, as described previously (6).

Bacterial load and cell counts

Blood was collected from the orbital plexus of mice, and the peritoneal cavity was washed with PBS. Aliquots of serial log dilutions were plated in Mueller-Hinton agar dishes, as previously described (71). Leukocyte numbers were determined in the peritoneal cavity and BALF 6 hours after CLP or sham treatment using the Hemavet 950FS System (71).

Flow cytometry

Peritoneal cells were resuspended in PBS containing 2 mM EDTA and 0.5% fetal bovine serum (FBS). Fc receptor–mediated and nonspecific antibody binding was blocked by treatment with CD16/CD32 (clone 2.4G2, BD Biosciences Pharmingen) for 10 min at 4°C. The cells were stained with mouse anti–GR1–fluorescein isothiocyanate (FITC) (1:100; BD Pharmingen) for 30 min at 4°C, and cell surface expression was analyzed by flow cytometry (FACSCalibur). Data were analyzed with WinMDI and FlowJo version 7.6.4 software.

Cytokine measurement

The amounts of TNF-α, IL-1β, and IL-6 were measured using specific DuoSet ELISA kits (R&D Systems) according to the manufacturer’s protocol.

Histology

Mice were perfused with 10% formalin before lung harvesting. Tissues were fixed in 10% formalin, embedded in paraffin, cut into 5-μm sections, and stained with H&E, as previously described (71). Images were captured with an INFINITY1 camera attached to a Nikon Eclipse Ci microscope. Capillary congestion and alveolar edema were determined as previously described (71).

Western blotting

Western blots were performed as previously described (6). Protein samples were resolved by SDS–polyacrylamide gel electrophoresis (PAGE), transferred to a nitrocellulose membrane, and incubated with primary antibodies against MyD88, total or phosphorylated (S727 or Y701) STAT1, phosphorylated p65 NF-κB (Ser523), PTEN (all at a 1:1000 dilution; Cell Signaling), TIRAP, TRIF, TIRP, TLR4 (1:1000 dilution; Santa Cruz Biotechnology), or β-actin (1:10,000 dilution; Sigma-Aldrich). Densitometry analysis of bands of interest was performed as described previously (6).

RNA isolation and semiquantitative real-time RT-PCR

Total RNA from cultured cells was isolated with the GenElute Mammalian Total RNA Miniprep Kit (Sigma) according to the manufacturer’s instructions. Quantitative reverse transcription PCR (qRT-PCR) assays were performed as previously described (6). The primers for PTEN, MyD88, primary miRNAs, STAT1, and actin were all obtained from Integrated DNA Technologies. The relative abundance of the target of interest was calculated using the comparative threshold cycle (Ct) and expressed relative to the appropriate control or WT (ΔΔCt method).

Confocal microscopy and PLA

A total of 2 × 105 macrophages from PTENfl/fl or PTENfl/fl_lysMcre mice were plated in four-well chamber slides (Nunc) and then washed with PBS. Slide staining was performed as previously described (73) using various combinations of primary antibodies against Drosha, Dgcr8, Dicer, or PTEN (each used at a 1:200 dilution). Rhodamine- or FITC-conjugated goat anti-rabbit or anti-mouse secondary antibodies (1:200) (Sigma-Aldrich) were used. Cells were imaged on a Zeiss LSM 510 confocal microscope with an inverted Axiovert 100 M microscope stand using a C-apochromat 40×/1.2 W corr. Confocal images were taken with identical settings to enable comparison of staining. Z-stacked sections (10 to 22 slices) of the cells were captured in multitrack, and ImageJ software was used to reconstruct the images using the Z project plug-in. The extent of colocalization between the miRNA machinery and PTEN, as well as their intracellular localization, was quantitated using the JaCoP plug-in for ImageJ (74). The background of the collected images was corrected by the ImageJ rolling ball algorithm plug-in. The specific algorithm used was based on the Manders overlap coefficient (74), which ranges from 0 to 1, the former corresponding to non-overlapping images and the latter reflecting 100% colocalization between both images. For each experiment, at least 100 randomly selected cells were scored. PLA, which enables the visualization, localization, and quantification of individual protein-protein interactions at a range of 30 to 40 nm (75), was performed according to the manufacturer’s guidelines (Duolink orange detection system; Olink Bioscience). Briefly, cells were fixed with 4% paraformaldehyde at room temperature for 10 min, which was followed by cell membrane permeabilization with 0.2% Triton X-100 in tris-buffered saline (TBS) for 10 min. The cells were then blocked with 1% bovine serum albumin and 10% normal donkey serum in TBS for 1 hour at room temperature and incubated with the indicated primary antibody pairs (Drosha versus PTEN, Dgcr8 versus PTEN, and Dicer versus PTEN) overnight at 4°C. Oligonucleotide-conjugated secondary antibodies (PLA probe MINUS and PLA probe PLUS) against each of the primary antibodies were applied, and ligation and amplification were performed to produce rolling circle products. These products were detected with fluorescently labeled oligonucleotides, and the samples were counterstained with Duolink Mounting Medium with DAPI.

Nuclear isolation

Nuclear and cytoplasmic fractions of peritoneal cells from PTENfl/fl and PTENfl/fl_lysMcre mice were isolated as previously described (76) using a Subcellular Protein Fractionation kit (Pierce) according to the manufacturer’s instructions.

In-blot PTEN activity assay

PTEN-mediated dephosphorylation of IRAK4 and IKKα was examined by an in-blot phosphatase assay as described previously (21, 77). Briefly, histidine-tagged PTEN (His6-PTEN) was generated by inserting the complementary DNA (cDNA) encoding full-length PTEN into the pQE30 vector (Qiagen). The protein was purified with Ni-NTA beads (Qiagen) under denaturing conditions and then renatured by sequential dilution and concentration in renaturation buffer [PBS (pH 7.0) containing 2 mM MgCl2, 0.5 mM phenylmethylsulfonyl fluoride, 0.005% Tween 20, 10 mM dithiothreitol (DTT), and protease inhibitor mixture]. Purity (>90%) was confirmed by SDS-PAGE and Coomassie blue staining. Peritoneal macrophages (3 × 106) were plated overnight, and cells were stimulated with LPS for 1 hour and then lysed with radioimmunoprecipitation assay (RIPA) buffer. Equal amounts of proteins were subjected to 10% SDS-PAGE and electrotransferred to nitrocellulose. Blots were incubated with recombinant His6-PTEN (20 μg/ml) or alkaline phosphatase (500 U/ml) in 50 mM Hepes buffer (pH 7.0) containing 10 mM MgCl2 and 10 mM DTT at 30°C for 1 hour. Phosphorylated IRAK4 and IKKα were detected by Western blotting as mentioned earlier.

Immunoprecipitation

For immunoprecipitation of PTEN, macrophages were lysed with RIPA buffer, precleared with protein A–Sepharose for 30 min, and incubated overnight at 4°C with anti-PTEN antibody (1:80) as previously described (21). Protein A–Sepharose was added and incubated for 3 hours with rotation at 4°C. Immunoprecipitates were isolated and subjected to SDS-PAGE and Western blotting as described earlier. Membranes were then incubated with antibodies against PTEN, Drosha, Dgcr8, and Dicer, as described earlier.

Adenoviral constructs and infection

Adenovirus-containing cDNA constructs encoding constitutively active WT PTEN, dominant-negative PTEN (C124S), the lipid phosphatase activity mutant PTEN (G129E), and the empty viral vector (control) containing no cDNA insert were prepared as previously described (21). Adenoviruses were amplified in HEK 293 cells and purified by ultracentrifugation on a CsCl density gradient. Alveolar or peritoneal macrophages (2 × 106) were seeded in six-well plates in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FBS. Virus was added to the medium at a multiplicity of infection of 250. After 72 hours, the cells were harvested, and cell lysates were subjected to Western blotting analysis.

Retroviral transduction and generation of PTEN-mutant MEFs

PTEN−/− MEFs were cultured in DMEM with 10% FBS and penicillin-streptomycin. These cells were reconstituted with WT PTEN or mutants of PTEN (C124S, G129E, K254R, and K289R). Briefly, PTEN and mutant PTEN cDNAs were cloned into the pRev-TRE-hygromycin retroviral construct (Clontech). Phoenix 293 cells were transfected with the retroviral constructs with Lipofectamine. Medium was collected and replaced 24 hours after transfection. Medium was collected again after a further 3 days. The pooled virus-containing medium was centrifuged, polybrene (10 μg/ml) was added, and the medium was filtered. The filtered virus was added to PTEN−/− MEFs and incubated for 36 hours. The cells were then treated with hygromycin (100 mg/ml) for 1 week. The surviving cells were maintained in hygromycin (20 mg/ml) and expanded in number. The hygromycin-resistant cells were analyzed for the presence or absence of PTEN by Western blotting.

Targeted miRNA overexpression

For the inhibition of miR155, miR125b, miR146b, and miR203, macrophages were transfected with Lipofectamine siRNA Max transfection reagent with PTEN-specific siRNA or scrambled siRNA control as described earlier, which was followed by transfection with pre-miR155, pre-miR125b, pre-miR146b, or pre-miR–negative control 1 (pre-miR control), as was previously described (10). Transfected cells were treated with or without 30 nM miRNA mimic for 48 hours and, where indicated in the figure legends, were treated for 24 hours with LPS (100 ng/ml) before cell culture medium was collected or cell lysates were prepared for the isolation of RNA with an RNeasy kit (Qiagen).

miRNA analysis

We performed qRT-PCR analyses for miR125b, miR125a, miR146a, miR155, miR146b, miR451, miR19a, miR19b, miR203, miR30e, miR21, miR146b, miR181a, let7g, and RNU6 (used as normalization control) with TaqMan miRNA assays with reagents, primers, and probes obtained from Qiagen. We synthesized cDNA with a reverse transcription system (miScript II, Qiagen). We performed qPCR on the CFX96 Real-Time PCR Detection System (Bio-Rad Laboratories) as described previously (6).

Focused miRNA arrays

To determine the abundances of inflammatory miRNAs, macrophages from C57BL/6 WT mice were treated with PTEN-specific siRNA or scrambled siRNA as described earlier. RNA was then extracted from the cells with the RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. We synthesized cDNA with the RT2 miRNA First Strand Kit (Qiagen) and applied it to PCR-focused immunopathology miRNA array plates (Qiagen) as previously described (10). Normalization and statistical analysis of miRNA abundances were performed with the Online PCR Array Data Analysis Web Portal (SABiosciences) as previously described (10).

AGO2 immunoprecipitation

Macrophages were incubated with 30 nM miRNA mimic control, miR125b, and miR203 for 24 hours, and AGO2 was immunoprecipitated with an miRNA Target IP kit (Active Motif), according to the manufacturer’s instructions. Immunoprecipitation with anti-IgG antibody was used as a negative control. After the pull-down, coprecipitated mRNAs were converted to cDNA and subjected to qRT-PCR analysis with primers directed against the 3′ UTR of MyD88 and the 3′ UTR of β2-microglobulin (Integrated DNA Technologies).

Statistical analyses

Survival curves were expressed as percentage survival and were analyzed by a log-rank (Mantel-Cox) test. Bacterial load results are expressed as median values. Other results are expressed as means ± SEM and were analyzed by ANOVA, followed by Bonferroni analysis. For the pediatric patient group, we performed Mann-Whitney U statistical analysis. For the adult septic patients, PTEN mRNA expression values were compared by the unpaired Student’s t test. In all instances, differences were considered to be statistically significant when P < 0.05.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/528/eaai9085/DC1

Fig. S1. Antibiotic does not prevent PTEN inhibition–induced animal lethality during sepsis.

Fig. S2. PTEN controls MRSA-induced peritonitis.

Fig. S3. PTEN protects mice against lung injury during sepsis.

Fig. S4. PTEN does not control TRIF-dependent macrophage activation.

Fig. S5. PTEN does not dephosphorylate IRAK4 or IKKα in macrophage lysates.

Fig. S6. PTEN lipid phosphatase activity decreases MyD88 mRNA and protein abundance in alveolar macrophages.

Fig. S7. MyD88-blocking peptide prevents TLR4- and TLR2-mediated, but not TLR3-mediated, nitrite production.

Fig. S8. PTEN does not target transcription factors involved in basal Myd88 expression in macrophages.

Fig. S9. PTEN controls miRNA abundance in alveolar macrophages.

Fig. S10. Differential roles of mTOR and PI3K in miRNA expression.

Fig. S11. The miRNAs miR125b and miR203 directly reduce Myd88 mRNA abundance in macrophages.

Fig. S12. Efficiency of PTEN-expressing retrovirus in PTEN−/− MEFs.

Table S1. Adult sepsis patient demographics.

REFERENCES AND NOTES

Acknowledgments: We thank D. Mucida for suggestions. Funding: This work was supported by NIH grants (NIH HL-103777 and R01HL124159-01 to C.H.S., T32AI060519 to S.L.B., and GM099773 and GM108025 to H.W.), American Lung Association Senior Research Training Fellowship (RT-349159 to N.G.-B.), Coordination for the Improvement of Higher Education Personnel, and São Paulo Research Foundation (FAPESP) under grant agreement numbers 2011/19670-0 (Projeto Temático) and 2013/08216-2 (Center for Research on Inflammatory Disease). Author contributions: Conceptualized and designed experiments: F.S., L.D.M., J.C.A.-F., L.F., C.H.S., S.L.B., and N.G.-B. Performed experiments: F.S., Y.M.S., S.L.B., S.W., S.S., N.G.-B., C.S.D.C., L.F., and N.A. Data analyses: F.S., L.D.M., J.C.A.-F., L.F., C.H.S., S.L.B., N.G.-B., C.S.D.C., H.W., S.J., and F.Q.C. Manuscript preparation: N.G.-B. and C.H.S. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article