Genetic deletion of the alternative isoform I.1 of HIF-1α in T cells enhances anti-bacterial immune response and improves survival in the model of bacterial peritonitis in mice

Summary

Hypoxia-adenosinergic suppression and re-direction of the immune response has been implicated in the regulation of anti-pathogen and anti-tumor immunity, with Hypoxia-inducible factor 1α (HIF-1α) playing a major role. In this study, we investigated the role of isoform I.1, a quantitatively minor alternative isoform of HIF-1α, in anti-bacterial immunity and sepsis survival. By using the cecal ligation and puncture model of bacterial peritonitis we studied the function of I.1 isoform in T cells using mice with total I.1-isoform deficiency and mice with T cell-targeted I.1 knockdown. We found that genetic deletion of the I.1 isoform resulted in enhanced resistance to septic lethality, significantly reduced bacterial load in peripheral blood, increased M1 macrophage polarization, augmented levels of pro-inflammatory cytokines in serum, and significantly decreased levels of the anti-inflammatory cytokine IL-10. Our data suggest an immunosuppressive role of the I.1 isoform in T cells during bacterial sepsis that was previously unrecognized. We interpret these data as indicative that activation-inducible isoform I.1 hinders the contribution of T cells to the anti-bacterial response by affecting M1/M2 macrophage polarization and microbicidal function.

INTRODUCTION

Sepsis is a complex clinical syndrome representing a major healthcare obstacle with an unacceptably high mortality rate [1]. Despite massive efforts, the absence of effective therapies remains a critical barrier to improving survival, and failures in sepsis therapies point to the critical need for new treatment paradigms [2]. Sepsis includes two major stages: Systemic Inflammatory Response Syndrome (SIRS) characterized by a hyperinflammatory response to infection, and Compensatory Anti-inflammatory Response Syndrome (CARS) accompanied by production of anti-inflammatory cytokines that limits tissue damage associated with the inflammatory response [2–5]. The resulting period of immune suppression is characterized by inhibited macrophage function, diminished pro-inflammatory cytokine production, and sustained production of anti-inflammatory cytokines such as IL-10 [6, 7]. During the immunocompromised phase of sepsis, patients become increasingly susceptible to secondary hospital-acquired infections, which play a decisive role in the lethal outcome [8–12]. In humans inhibition of pro-inflammatory cytokines leads to increased later mortality due to opportunistic infections accompanied by macrophage deactivation, which was shown to be mediated by anti-inflammatory cytokines such as IL-10 [13–15].

Although macrophages and neutrophils play a central role in eliminating bacteria, T cells are required for efficient protection against bacterial infection via production of proinflammatory cytokines [16–20]. Within the first 24 hours of infection T cells can produce proinflammatory cytokines that can enhance the early innate immune response against bacterial infection [15, 21, 22], and inadequate activation of T cells results in an insufficient innate immune response leading to decreased bacterial clearance and survival [17]. Stimulation of macrophages through toll-like receptors in combination with Th1 cytokines leads to M1 polarization of pro-inflammatory and cytotoxic macrophages [23], while Th2 cytokines such as IL-10 create the alternatively activated macrophages of M2 phenotype, which are antiinflammatory and poorly microbicidal [23–25].

Hypoxia-inducible factor 1α (HIF-1α) is stabilized by hypoxia, which is associated with inflammatory tissue damage during sepsis [19, 26]. HIF-1α protein can be encoded by two alternative mRNA isoforms [27] (Fig. 1A). The difference between these isoform is due to a difference in the first exons, which results from alternative splicing (Fig. 1A). The conventional mRNA isoform contains I.2 exon and is ubiquitously expressed using constitutive I.2 promoter in various tissues [27, 29] and is indispensable for cell survival, glycolysis, and angiogenesis [28]. The alternative mRNA isoform starts with exon I.1 and is predominantly expressed in immune organs and testis [29, 30]. I.1 mRNA isoform encodes HIF-1α protein lacking the first 12 N-terminal amino acids, however this “short” I.1 isoform retains DNA-binding bHLH domain and has similar transcriptional activities as ubiquitous “full-size” I.2 isoform [30]. Importantly, the expression of I.1 promoter is strongly induced in antigen receptor-activated T cells as an immediate-early response gene [30]. Therefore I.1 isoform was dubbed “activation-inducible isoform” [31]. Previously, we demonstrated that genetic deletion of the I.1 isoform results in higher pro-inflammatory cytokine production by T cells [31], and macrophages in vitro [38].

Figure 1. Expression of HIF-1α I.1 isoform in TCR-activated T cells and its effect on pro-inflammatory cytokine production.

(A) Scheme of the differential expression of alternative isoform I.1 and ubiquitous isoform I.2 of murine HIF-1α. Exons are represented by squares. Dashed lines indicate alternative splicing. Horizontal arrows represent promoters. (B) I.1 mRNA expression in splenic T cells after TCR activation in vitro. T cells were isolated from spleen of C57BL/6 mice and activated by anti-CD3/anti-CD28 magnetic beads for 24 hrs under either normoxic (21 % O2) or hypoxic (1 % O2) conditions. Total RNA were subjected to RT-qPCR analysis of HIF-1α I.1 isoform. Data are shown as a mean ± SEM and representative of three independent experiments. Statistical significance between untreated ex vivo control and activated samples was determined using non-paired Student t-test. Two-tailed levels of significance are indicated by asterisk: **, p<0.01. (C) I.1 mRNA expression in splenic T cells after in vivo TCR activation. C57BL/6 mice were i.p. injected with 50 µg bacterial superantigen (SEB) for 3hrs. T cells were purified from spleens by anti-CD4/anti-CD8 magnetic beads and total RNA were subjected to RT-qPCR. Data are expressed as a mean ± SEM of 3 animals and representative of two independent experiments. Statistical significance between groups was determined using non-paired Student t-test. Two-tailed levels of significance are indicated by asterisk: *, p<0.05. (D) Effect of I.1 isoform deletion on cytokine production. Mice with total HIF-1α I.1 isoform knock-out were injected i.p with 50µg SEB. Plasma was collected 3 hrs after injection and serum levels of TNF-α, IL-2, and IL-6 were determined by ELISA. Data are expressed as a mean ± SEM of 3 mice per group and representative of three independent experiments. Statistical significance between groups was determined using non-paired Student t-test. Two-tailed levels of significance are indicated by asterisk: *, p<0.05; **, p<0.01.

In the present study we demonstrate a connection between the T cell-orchestrated antibacterial response and the activation-inducible I.1 isoform of HIF-1α. We show that activation-inducible I.1 isoform of HIF-1α negatively regulates T cell contributions into the anti-bacterial immune response during polymicrobial sepsis by affecting M1/M2 macrophage polarization.

RESULTS

Anti-inflammatory effects of TCR-activation inducible I.1 isoform in vivo

Our previous in vitro studies established that the I.1 isoform of HIF-1α is expressed as an immediate early response gene in T cells after activation through T cell receptor (TCR) stimulation [30, 31, 34] (Fig. 1A). We confirmed and extended these findings by showing that the I.1 isoform of HIF-1α is not only induced by TCR stimulation in vitro (Fig. 1B), but also in vivo when T cells are activated by bacterial superantigens (SAg) (Fig. 1C), which can activate large number of T cells by cross-linking their TCR with MHC Class II molecules of antigen presenting cells, thereby causing rapid polyclonal T cell proliferation and cytokine production [22]. We found that intraperitoneal (i.p.) injection of Staphylococcus aureus enterotoxin B (SEB) strongly induces I.1 isoform expression in T cells as early as early as 3 hours after SEB injection. Previously, we demonstrated that I.1-deficient T cells produce more proinflammatory cytokines in vitro after TCR-stimulation either by cross-linking mAb or by allogenic MHC in mixed lymphocyte culture [31]. However, the alternative isoform I.1 is a quantitatively minor isoform, which contributes to 10–15% of total HIF-1α mRNA [31]. It was not clear whether this minor alternative isoform plays a significant role in vivo by producing sufficient immunosuppression and affecting the anti-pathogen immune response. We reasoned that if the alternative HIF-1α isoform I.1 indeed regulates the overall intensity of the anti-pathogen immune response in vivo, then the elimination of this isoform in I.1-gene deficient mice would lead to reduced immunosuppression, higher pro-inflammatory cytokine production, stronger anti-bacterial response and improved survival in mouse models of sepsis. Our previous findings established that T cells can potentially contribute to the anti-bacterial immune response by secreting proinflammatory cytokines, which activate microbicidal properties of macrophages during sepsis [19]. Therefore, it was important to investigate whether elimination of the I.1 isoform in T cells leads to enhanced production of pro-inflammatory cytokines in vivo after TCR-activation with bacterial superantigens. Indeed, we show that pro- inflammatory cytokines are strongly induced in I.1-deficient mice after SEB injection (Fig. 1D). These results further extended our in vitro observations that the I.1 isoform acts as an immunosuppressive factor in activated T cells.

In addition, activation of peritoneal macrophages in vivo by LPS also induced expression of I.1 mRNA isoform (Suppl. Fig. 1A), which comes in agreement with previous reports of in vitro upregulation of I.1 isoform after TLR-mediated activation [38]. However, we did not detect significant changes in pro-inflammatory cytokine production in response to TLR-stimulation in vivo in I.1-deficient mice as compared to wild-type mice (Suppl. Fig. 1B).

Improved resistance of I.1-deficient mice to bacterial sepsis

To study the effect of I.1-deficiency in T cells on the anti-bacterial immune response during sepsis we adopted the cecal ligation and puncture (CLP) model. This murine model of bacterial sepsis creates a polymicrobial infection resulting from the leakage of enteric content as a result of intestinal perforation [35]. To study the contribution of T cells into the anti-bacterial immune response during sepsis, we had to avoid early lethal outcomes during the first 24 hours of systemic infection. This would allow sufficient time for T cells to not only get recruited and activated, but also to contribute to the anti-bacterial immune response to such an extent that prevention of T cell inhibition by genetic deletion of I.1 would be discernible. Therefore we adopted a CLP model of long-lasting sepsis with ~50% mortality because it would allow enough time for T cells to contribute to the anti-bacterial immune response [19] and because it closely mimics the mortality rate observed in human patients [1, 35]. In this model severity and sepsis mortality can be adjusted by changing the size of needle or number of punctures [19, 36].

While deficiency in conventional HIF-1α results in early embryonic mortality due to failure in vascularization [37], naïve gene-deficient I.1−/− mice did not show apparent differences when compared to the wild-type C57Bl/6 mice [31]. Furthermore, analysis of immune organs from untreated WT and I.1-deficient mice by flow cytometry revealed no significant differences in immune cells subsets in lymph nodes, thymus, and spleen. In addition, T cell repertoire in spleen showed no significant differences in CD4+, CD8+ T cells, NKT cells, and Treg cells proportions (data not shown). The likely explanation of this apparent lack of phenotypic differences is that since I.1 isoform is TCR-activation dependent, it is not expressed in naïve T cells. Thus, one can expect no significant differences in I.1-KO vs. wild-type mice in the absence of infection or inflammation [31, 32]. However, after developing CLP-induced polymicrobial sepsis, mice deficient of the I.1 isoform (I.1-KO) demonstrated significantly higher survival rates when compared with control HIF-1α-expressing mice (Fig. 2A). Accordingly, I.1-KO mice had a significantly decreased bacterial burden in the peripheral blood, peritoneal cavity, and spleen (Fig. 2B).

Figure 2. Effect of genetic deletion of I.1 isoform of HIF-1α on CLP-induced sepsis in mice.

(A) Effect of the I.1 isoform gene knockout on sepsis survival. Polymicrobial sepsis was induced in C57Bl/6 male mice by CLP using a single puncture 18 gauge needle and mice were observed for 15 days. Control sham surgeries were performed in three WT and I.1-KO mice. Survival data are pooled from three independent experiments using I.1-KO mice (36 total) to WT controls (36 total). Significance between groups was calculated using log-rank test, *p < 0.05. (B) Bacterial load measured as CFU from plated 25µL of peripheral blood drawn 24 hrs post CLP induced sepsis, CFU from 5mL of peritoneal lavage fluid collected 72 hrs after CLP, and CFU in homogenized spleen collected 72hrs post CLP. Results in are expressed as a mean ± SEM of 36 mice pooled from three independent experiments. Significance between groups was calculated using non-paired Student t-test. Two-tailed levels of significance are indicated by: *, p<0.05 (C) Cytokine levels in blood of septic mice after CLP. Plasma collected 6, 24 and 48 hours after CLP and cytokine levels were determined by ELISA. Results are expressed as a mean ± SEM of 12 animals per group and are representative of three independent experiments. For IL-6 and IL-10 significance between two groups (I.1KO vs. WT) was calculated using two-way ANOVA analysis. Levels of significance are indicated by asterisk: **, p<0.01. For MIP-2 ANOVA comparison return P-value >0.05, therefore non-paired Student t-test was used to compare 6h time points. Two-tailed levels of significance are indicated by: **, p<0.01

Deficiency in I.1 was also associated with higher serum concentrations of proinflammatory cytokines such as IL-6 and chemokines, such as MIP-2 (Fig. 2C). In addition, the concentrations of the anti-inflammatory cytokine interleukin 10 (IL-10) were found to be significantly lower in I.1-deficient mice (Fig. 2C).

Flow cytometry analysis of peritoneal lavage after 3 days of CLP-induced sepsis revealed that I.1 deficiency results in several changes in population of immune cells (Fig. 3A). I.1-deficient mice had higher percentage of neutrophils and lower number of macrophages, CD4+ and CD8+ T cells as compared to wild-type. Lymphocyte repertoire in spleens of septic mice showed higher percentage of natural killer cells in I.1-KO mice (Fig. 3B). Interestingly, I.1 deficiency was also associated with higher expression of Foxp3 in T regulatory cells, which is in agreement with a recently published study indicating HIF-1α as a factor, which downregulates Foxp3 expression and affects Th17/Treg polarization [51]. In contrast, T regulatory cells demonstrated lower levels of surface expression of ecto-5’-nucleotidase CD73, which creates extracellular adenosine (Fig. 3B). It is known that hypoxia-induced adenosine is increased during sepsis, and adenosine receptors negatively regulate the immune response to pathogen [36, 39, 52, 53]. The observed downregulation of CD73 can potentially increase pro-inflammatory response and increase neutrophil chemotaxis.

Figure 3. Flow cytometry analysis of immune cells after CLP.

(A) Analysis of peritoneal lavage cells. Peritoneal lavage was collected from mice, which were sacrificed after 72 hours post CLP procedure (single puncture 18 gauge needle). Neutrophils were stained as GR1-positive/F4/80-negative, macrophages as F4/80-positive/GR1-low, B cells as B220-positive. (B) FACS analysis of splenic T lymphocyte subsets. Splenocytes were collected from septic mice after 72h post CLP and stained for CD4+, CD8+, NK1.1+ (NK cells), CD4+CD25+Foxp3+ (Tregs), CD4+ Foxp3+CD39+, CD4+Foxp3+CD73+. Data are expressed as a mean ± SEM of 5 mice per group and representative of two independent experiments. Statistical significance between groups was determined using non-paired Student t-test. Two-tailed levels of significance are indicated by asterisk: *, p<0.05.

Development of mice with T-cell specific knockdown of I.1 isoform

To discriminate between the effects of the I.1 isoform in cells of the innate and adaptive immune systems, we decided to study the mice with T cell-specific I.1-knockdown. We created I.1-lox/lox_CD4Cre+/− conditional knockdown mice where I.1 exon is surrounded by loxP sites and Cre recombinase is expressed via CD4-promoter [41]. This results in Cre-mediated excision of the I.1 exon in thymocytes during their double-positive stage (Suppl. Fig. 2). Since the alternative isoform I.1 is shorter than the ubiquitous isoform I.2, but otherwise has the same amino acid sequence, the development of I.1-specific antibody for detection is impossible. Therefore, the efficiency of I.1 deletion in T cells was evaluated using RT-qPCR. RT-qPCR confirmed that while I.1 mRNA is strongly induced by TCR-activation in wild-type T cells, I.1 isoform is efficiently deleted in I.1-lox/lox_CD4Cre+/− mice (I.1CD4 mice) (Fig. 4A), while no difference in I.1 mRNA was detected in splenic B cells and macrophages (Fig. 4B). Untreated I.1CD4 mice did not show significant phenotypic differences in immune cell populations of thymus, spleen or lymph nodes (data not shown). Our initial hypothesis that the I.1 isoform of murine HIF-1α alpha acts as an immunosuppressive factor in activated T cells in vitro was further supported by data showing that T cells isolated form I.1CD4 mice produce more IFNγ when stimulated by anti-CD3/anti-CD28 mAb in vitro (Fig. 4C).

Figure 4. Effect of conditional knockdown of I.1 isoform in T cells.

(A) Evaluation of efficiency of I.1 exon deletion by RT-qPCR. Measurements of I.1 mRNA amounts in T cells. T cells were isolated by autoMACS from spleens of I.1-floxed (WT) and I.1-floxed_CD4Cre+/− (I.1CD4) mice and activated using beads carrying anti-CD3 and anti-CD28 mAb at either hypoxic (1% O₂) or normoxic (21% O₂) conditions for 48 hours. Total RNA was extracted and subjected to RT-PCR. (B) RT-qPCR analysis of RNA from splenic macrophages and B cells. Macrophages and B cells were isolated from spleen by autoMACS using CD11b–FITC and B220-FITC and anti-FITC magnetic beads. Data are expressed as a mean ± SEM of 3 mice per group and representative of two independent experiments. Statistical significance between groups was determined using non-paired Student t-test. Two-tailed levels of significance are indicated by asterisk: **, p<0.01. (C) Flow cytometry analysis of intracellular IFN-γ production by I.1 deficient CD4⁺ and CD8⁺ T lymphocytes activated in vitro with soluble anti-CD3 mAb for 24h, followed by PMA/ionophore for 2h. Results are depicted as a mean ± SEM of two independent experiments. Significance between two groups was calculated using two-way ANOVA. Asterisks indicates level of significance ** p<0.01%.

T-cell specific knockdown of I.1 isoform results in better resistance to bacterial sepsis

We predicted that the T-cell specific knockdown of the I.1 isoform would improve the anti-bacterial immune response and overall sepsis survival due to enhanced production of macrophage-stimulating pro-inflammatory cytokines by T cells. Indeed, the survival of I.1-lox/lox_CD4Cre+/− ( I.1CD4) mice was significantly higher as compared to wild-type (i.e. I.1-lox/lox_CD4Cre−/−) littermate controls (Fig. 5A). These results were similar to those obtained by using mice with total I.1 isoform knockout. Accordingly, during CLP-induced sepsis I.1CD4 mice showed lower levels of circulating bacteria (Fig. 5B) accompanied with higher IL-6 plasma levels and lower IL-10 levels as compared to wild type controls (Fig. 5C).

Figure 5. Effect of T cell-specific I.1 deficiency on polymicrobial sepsis survival and bacteremia after CLP.

(A) 15-day survival analysis of I.1-lox/lox_CD4Cre+ mice (I.1CD4) after CLP-induced sepsis. CLP was performed using a single puncture 18 gauge needle (n=16 per group). Control sham surgeries were performed in three WT and I.1CD4 mice. Presented survival data are pooled from three independent experiments. *p<0.05 when comparing WT to I.1CD4 mice by log rank test. (B) Evaluation of bacterial burden in blood. A 25ul aliquot of blood was collected 24 hrs post CLP to check for levels of circulating bacteria. Results are expressed as a mean ± SEM and representative of three independent experiments. Significance between two groups was calculated using non-paired Student t-test. Two-tailed levels of significance are indicated by: *, p<0.05 (C) IL-6 and IL-10 production in I.1CD4 mice. Plasma was collected at 6, 24 and 48 hours post CLP and serum cytokine levels were determined by ELISA. Results are expressed as a mean ± SEM of 5–10 animals per group and are representative of three independent experiments. Significance between two groups (I.1CD4 vs. WT) was calculated using two-way ANOVA analysis. Levels of significance are indicated by asterisk: **, p<0.01.

Expression of I.1 isoform in T cells affects polarization and anti-microbial functions of macrophages during bacterial sepsis

The observation of IL-10 downregulation in I.1-deficient mice during sepsis prompted further investigation of its mechanism. IL-10 can be secreted by Th2, Treg, NK cells, and also by macrophages and neutrophils [54]. Interestingly, LPS activation of peritoneal macrophages resulted in significantly lower IL-10 levels in peritoneal lavage (Fig. 6A), which implies that I.1-deficient macrophages may also contribute in IL-10-mediated regulation of the anti-bacterial response. To test whether activation-inducible isoform I.1 regulates IL-10 production in T cells, splenic T cells were activated with anti-CD3/anti-CD28-coated magnetic beads. We found, that I.1-deficient T cells produce significantly less IL-10 when compared to WT T cells (Fig. 6B).

Figure 6. Neutrophils and macrophage functions in mice with T cell-specific I.1 deficiency.

(A) Concentration of IL-10 in peritoneum after LPS activation. Five WT vs. I.1-KO mice were injected with 2ml of 3% thioglycollate i.p. for 2 days followed by 4 hours of 5mg/kg LPS i.p. Results are expressed as a mean ± SEM and representative of two independent experiments. Significance between two groups was calculated using non-paired Student t-test. Two-tailed levels of significance are indicated by: *, p<0.05 (B) Production of IL-10 by T cells in vitro. T cells were isolated from spleens of three WT and three I.1-KO mice and activated by beads coated with anti-CD3/anti-CD28 mAb for 48h. Results are expressed as a mean ± SEM and representative of three independent experiments. Significance between two groups was calculated using non-paired Student t-test. Two-tailed levels of significance are indicated by: **, p<0.01 (C) IL-10 mRNA expression in T cells activated in vivo. I.1KO and WT mice were injected i.p. with 25–100 µg of SEB for 3 hours. Data are expressed as a mean ± SEM of 3 mice per group and representative of two independent experiments. Statistical significance between groups was determined using non-paired Student t-test. Two-tailed levels of significance are indicated by asterisk: *, p<0.05; **, p<0.01. (D) Macrophage polarization in WT vs. I.1CD4 mice. Peritoneal macrophages obtained from mice with T cell specific HIF-1α I.1 isoform deficiency 72hrs after CLP-induced sepsis. Polarization states were determined by expression assessment of specific chemokine and cytokine mRNA with the use of RT-qPCR: M1-specific (iNOS) and M2-specific (CCL17). Data are expressed as a mean ± SEM of 3 mice per group and representative of two independent experiments. Statistical significance between groups was determined using non-paired Student t-test. Two-tailed levels of significance are indicated by asterisk: *, p<0.05. (E) The phagocytic capacity of macrophages from mice with T-cell conditional knockdown of I.1 isoform. Peritoneal macrophages were collected from WT and I.1CD4 mice after 5 days post CLP and their phagocytic functions were measured using in vitro system of pH-sensitive fluorogenic E. coli cells. F4/80-positive macrophages with ingested labeled E.coli were analyzed by FACS. Data are expressed as a mean ± SEM of 5 mice in two independent experiments. Significance between two groups was calculated using non-paired Student t-test. Two-tailed levels of significance are indicated by *, p<0.05; **, p<0.01

Since HIF-1α is a major transcriptional factor, it was rational to examine whether I.1 deficiency leads to changes in gene expression. However, I.1 is a minor isoform, and no significant changes were observed in the expression of HIF-1-dependent genes including VEGF, GLUT1, LDH etc. in T cells activated in vitro by anti-CD3 mAb or in vivo by SEB (data not shown). On the other hand, reverse transcription-quantitative PCR (RT-qPCR) measurement of cytokine gene expression after SEB injection revealed dramatic reduction of IL-10 mRNA in I.1-KO T cells (Fig. 6C) suggesting that the I.1 isoform may be involved in the regulation of IL-10 expression.

The observation of diminished IL-10 production by I.1-deficient T cells pointed to a possibility that M1/M2 polarization of macrophages may be affected by I.1 isoform in T cells during sepsis [14, 40, 42, 43]. Microbicidal M1 and anti-inflammatory M2 polarization states of macrophages can be assessed by expression of specific chemokines and cytokines [44]. To investigate whether I.1 deficiency in T cells would affect anti-bacterial properties of macrophages and their development during sepsis, peritoneal macrophages were isolated from T cell-specific I.1-knockdown mice (I.1CD4) and I.1-expressing WT mice, which underwent CLP-induced sepsis for five days. We found that macrophages derived from mice with T cell-specific I.1 knockdown were more M1-polarized as assessed by mRNA expression of iNOS (M1 phenotype [44]), and CCL17 (M2 phenotype [44]) (Fig. 6D) when compared to macrophages derived from wild-type septic mice. Furthermore, we analyzed phagocytic macrophages from T-cell-specific I.1-KO mice for their capacity to engulf bacteria. In vitro phagocytosis assay using peritoneal macrophages from five day-septic mice revealed significantly higher rates of phagocytosis in macrophages isolated from I.1CD4 mice as compared to the wild type mice (Fig. 6E).

Taken together, our data indicate that I.1 isoform deficiency in T cells leads to higher pro-inflammatory cytokine production and lower IL-10 production, which may enhance M1 polarization and augment the phagocytic capacity of macrophages, thereby improving the anti-bacterial immune response during sepsis.

DISCUSSION

This study extends our previous findings that the I.1 isoform of HIF-1α inhibits proinflammatory T cell functions [31–33].The presented here data point to an immunosuppressive role of the I.1 isoform in T cells during bacterial sepsis and implicate this isoform in hindering the anti-bacterial immune response.

Despite its disproportionately strong effects on T cell functions in vitro the quantitatively minor alternative isoform I.1 of HIF-1α has not been studied for its effect on the anti-pathogen response and survival in vivo. Since the HIF1a isoform I.1 contains a different first exon than the ubiquitous isoform I.2, and is expressed from its own promoter [29, 30] (Fig.1A), it is rational to suggest that the N-terminal difference between I.1 isoform and conventional HIF-1α isoform I.2 may result in different transcriptional partners and/or changes in sequence recognition. Furthermore, the phenomenon of rapid activation-induced upregulation of I.1 expression suggests a critical role of I.1 as an immunosuppressive “emergency brake”, which starts to apply immediately after T cells become activated [30, 31]. Data presented in this study suggest that the I.1 isoform may be involved in regulation of expression of cytokine genes like IL-10, which leads to attenuation of pro-inflammatory functions of immune cells. Likewise, other studies demonstrated that overexpression of HIF-1α induces IL-10 production [45, 46]. The role of IL-10 dependent inhibition of the anti-bacterial response is well established [14, 43]. Our study points to the involvement of the I.1 isoform of HIF-1α in regulation of IL-10 expression, which may represent the physiological regulation of the innate anti-bacterial response. Induction of activation-inducible isoform I.1 may lead to attenuation of the pro-inflammatory functions of T cells and macrophages, which may negatively affect M1 macrophage polarization, inhibit microbicidal functions of macrophages, and result in suppression of the anti-bacterial immune response.

It is now established that HIF-1α is essential for the efficient anti-bacterial functions of macrophages, which include migration to inflamed tissues and production of cytotoxic molecules such as NO [47–49]. Therefore, it is important to discriminate the differential roles of the alternative isoforms of HIF-1α. It is possible, that the indispensable role of HIF-1α in macrophages is due to the demand of the conventional isoform I.2, which is required for glycolytic metabolism in inflamed hypoxic tissues [48, 50]. On the other hand, it was shown that the alternative isoform I.1 of HIF-1α can attenuate TLR-activated macrophages [38] and TCR-activated T cells [31]. Therefore, it is plausible to suggest that while conventional HIF-1α isoform I.2 is required for metabolism and free radical production in macrophages, the I.1 isoform serves as a suppressor of pro-inflammatory responses. Further comparative studies using tissue-specific gene-knockdowns of HIF-1α isoforms can help to resolve the differential roles of HIF-1α in innate and adaptive immune responses.

The role of the activation-inducible HIF-1α isoform I.1 as an inhibitory factor in activated T cells can be potentially exploited for modulation of the inflammatory response and to prevent excessive collateral tissue injury. The present study also suggests that selective targeting of the I.1 isoform, which has restricted tissue expression, may prevent the unwanted side effects of elimination of HIF-1α. The targeted inhibition of the I.1 isoform may unleash powerful anti-bacterial functions of T cells that contribute into macrophage activation and bacterial clearance.

MATERIALS AND METHODS

Mice

All animal experiments were conducted in accordance with IACUC guidelines of Northeastern University and the National Institutes of Health guidelines on the use of laboratory animals (IACUC Permit # 11–0720R). Male mice 8–12 weeks old were used for experiments. HIF-1α I.1-deficient mice were generated as previously described [31]. Mice with T cell-targeted deletion of HIF-1α (I.1-lox/lox_CD4-Cre+/−) were generated by breeding homozygous I.1-floxed mice [31] with CD4-Cre transgenic mice [41]. Mice that were I.1-floxed and heterozygous for CD4-Cre had I.1-deficient T cells; control wild-type mice were their CD4-Cre-negative/I.1-floxed siblings. Mice were genotyped for the presence of CD4-Cre allele.

Surgical procedure of CLP

CLP was performed as previously described [19, 36]. Briefly, mice were anesthetized with 5% isoflourane for induction and 3% for maintenance. Fifty percent of the cecum was ligated distal to the ileocecal valve in order to prevent bowel obstruction. The cecum was perforated with a single puncture using an 18 gauge needle (BD Biosciences, San Jose, Cal). A small amount of feces (5mm) was manually extruded from the perforation site into the peritoneal cavity. Lactated Ringers solution plus 5% dextrose for fluid resuscitation were administered by sub-cutaneous injection in all studies to create a more clinically relevant sepsis model as this is standard care for human patients. Fluid resuscitation was initiated 1.5 h after CLP and received twice-daily for 5 days. Antibiotic therapy was withheld. Sham treated controls underwent the same surgical procedures (i.e. laparotomy and resuscitation), but the cecum was neither ligated nor punctured.

Blood collection, peritoneal content and aerobic bacteria quantification

Blood, peritoneal lavage (PL) and organ collection were performed as previously described [36]. Bacteremia was determined from whole blood taken from the facial vein. Briefly, the mandible was shaven and cleansed with 70% ethanol to prevent contamination of cultures from bacteria present on the skin surface. The submandibular vein was punctured using a sterile 25-gauge needle for blood collection. A total of 25 µl undiluted whole blood was immediately plated for culture. The local bacterial load in CLP mice was determined by diluting 100 µl of PL with sterile PBS in 10-fold increments to a maximum dilution of 1/10⁵. The peritoneum was lavaged with a total of 5 milliliters of PBS collected in 1mL intervals. To remove cellular debris, each mL collected was strained using a 70µm nylon filter (BD Biosciences, San Jose, CA). Bacterial counts were also performed on aseptically harvested livers and spleens. Tissues were weighted, homogenized, and debris pelleted. Supernatants were then serially diluted and plated for culture. All samples were plated on sheep blood agar 5% tryptic soy agar monoplates (Remel, Lenexa, KS) and incubated overnight at 37°C under aerobic conditions. CFUs were determined by manual counting and then multiplied by their dilution factor.

Reagents and antibodies

Cytokine levels were determined using IL-6, IL-2, IL-4, IL-10, MIP-2, TNFα, IFNγ DuoSet ELISA kits (R&D Systems, Minneapolis, MN). All antibodies were from BD Biosciences (Franklin Lakes, NJ). Cells were maintained in RPMI 1640 medium supplemented with 10% heat inactivated FCS, 100U/ml penicillin, 100 µg/ml streptomycin, 1mM HEPES, and 50 µM 2-ME. Staphylococcus aureus enterotoxin B (SEB) and LPS were purchased from Sigma Aldrich (St. Loius, MO).

Lymphocytes isolation and activation

T cells were isolated as described previously [31]. .After in vivo activation with SEB T cells were separated from splenocytes by autoMACS separator (Miltenyi Biotec, Auburn, CA) using FITC-conjugated anti-CD4, anti-CD8 antibodies (BD Biosciences) and anti-FITC magnetic beads (Miltenyi Biotec) according to manufacturer's protocol. For in vitro activation T lymphocytes were enriched by Petri dish adhesion and activated by Dynabeads T cell activator anti-CD3/anti-CD28 magnetic beads according to manufacturer’s manual (Invitrogen, Oslo, Norway). After 48hrs of activation T cells were gently separated using Dynamag-2 magnet separator (Invitrogen, Oslo, Norway). For in vivo activation mice (3–5 per group) were injected i.p. with 100µl of SEB solution in HBSS (25, 50, or 100µg per mouse). Bleeding and spleen harvesting were performed either at 3 or 24 hrs

Flow cytometry gating strategy

Signals were acquired on FACScalibur flow cytometer (BD Biosciences) using CellQuest Pro Software. Acquisition threshold was set such that platelets, dead cells and debris were not recorded. Lymphocytes were sequentially defined by gating on FSC-high/SSC-low, then by staining with anti-CD4 or anti-CD8 antibodies to determine T cells, anti-B220 mAb to determine B cells. NK cells and NKT cells were differentiated using anti-NK1.1 and anti-CD3 mAb. T regulatory cells were stained with Foxp3 staining kit (eBioscience). Treg cells were defined as CD4+CD25+Foxp3+. For CD39 and CD73 expression Treg cells were stained as CD4+Foxp3+CD39+CD73+. Macrophages and neutrophils were first gated by FSC-high/SSC-high and then stained with mAb for granulocyte marker GR1 and macrophage marker F4/80. Macrophages were determined as F4/80-positive/GR1-low, neutrophils as F4/80-negative/ GR1-high. FACS was used to analyze cell surface marker expression and determine the phagocytic capabilities of macrophages obtained from the peritoneal cavity of septic mice. PL cells were washed and blocked with CD16/32 for 10 min before staining.

Measurement of intracellular IFNγ production

Intracellular IFNγ production was measured by flow cytometry as previously described [31]. Briefly, splenocytes were incubated with 0.01–0.1 µg/ml of soluble anti-CD3 mAb for 24h, followed by addition of 20 ng/ml PMA (Sigma) and 200 ng/ml A23187 (Sigma) for 2 h. BD Cytofix/Cytoperm Plus kit with Golgi plug (BD Biosciences) was used for intracellular IFNγ staining.

Phagocytosis assay

To access macrophage-mediated phagocytosis we used heat-killed fluorescein-conjugated pHrodo E.coli (Invitrogen, Molecular Probes, Carsbad, CA) as previously described [36]. Briefly, pHrodo E.coli were opsonized with purified polyclonal IgG antibodies specific for the E. coli particles according to the manufacturer’s instructions. We collected approximately 7.5 × 10⁵ PL cells 5 days post CLP. Cells were then incubated with opsonized pHrodo E.coli for 45 minutes at a ratio of 1:20 (PL cell:bacteria). Phagocytosis was carried out according to the manufacturer’s instructions in serum free media to avoid artificial alteration due to the presence of exogenous factors.

LPS activation of peritoneal phagocytes

Peritoneal macrophages and neutrophils were recruited by i.p. injection of 2 ml 3% thioglycollate for 2 days. 5 mg/kg LPS in 1 ml PBS was injected i.p. for 4 hours.

RNA isolation, reverse transcription, and qPCR analysis

RNA was extracted using RNA-STAT-60 (Tel-Test, Friendswood, TX) according to manufacturer’s protocol, and cDNA was synthesized using SuperScript® III First-Strand Synthesis SuperMix for qRT-PCR (Invitrogen). Real-time PCR was performed using SYBR Green mix (ABI Prism) and L32 mRNA as a housekeeping gene reference. Relative amounts of I.1 mRNA were evaluated using comparative Ct analysis as previously described [31]. Analysis of IL-2, IL-4, IL-10, TNFα, CCL17, and iNOS mRNA was performed using RT² qPCR Primer Assay (SABiosciences, Frederick, MD).

Statistical analysis

Comparison of survival curves was estimated using log-rank (Mantel-Cox) test. Significance between two groups was calculated using non-paired Student t-test or two-way ANOVA (GraphPad Prism 5.0). All values are expressed in mean ± SEM. Two-tailed levels of significance are indicated by *, p<0.05; **, p<0.01.

Abbreviations

CFU

colony forming unit

CLP

cecum ligation and puncture

HIF-1α

hypoxia-inducible factor 1α

I.1CD4

I.1-lox/lox_CD4Cre+/− mice

I.1-KO

I.1-deficient mice

i.p.

intraperitoneal

LPS

lipopolysaccharide

PL

peritoneal lavage

RT-qPCR

reverse transcription-quantitative polymerase chain reaction

SAg

superantigen

SEB

Staphylococcus aureus enterotoxin B

WT

wild-type mice