Endothelial glucocorticoid receptor is required for ... - Semantic Scholar

4 downloads 0 Views 729KB Size Report
Jan 2, 2013 - Julie E. Goodwina,1, Yan Fenga, Heino Velazquezb, and William C. Sessac,d,e,f ...... Orthner CL, Rodgers GM, Fitzgerald LA (1995) Pyrrolidine ...
Endothelial glucocorticoid receptor is required for protection against sepsis Julie E. Goodwina,1, Yan Fenga, Heino Velazquezb, and William C. Sessac,d,e,f a

Department of Pediatrics, dDepartment of Pharmacology, eDepartment of Cell Biology, cDepartment of Vascular Biology and Therapeutics Program, and Howard Hughes Medical Institute, Yale University School of Medicine, New Haven, CT 06520; and bDepartment of Internal Medicine, Veterans Affairs Hospital, West Haven, CT 06516 f

Edited by Michael A. Gimbrone, Brigham and Woman’s Hospital, Harvard Medical School, Boston, MA, and approved November 13, 2012 (received for review June 28, 2012)

The glucocorticoid receptor (GR) is ubiquitously expressed on nearly all cell types, but tissue-specific deletion of this receptor can produce dramatic whole organism phenotypes. In this study we investigated the role of the endothelial GR in sepsis in vivo and in vitro. Mice with an endothelial-specific GR deletion and controls were treated with 12.5 mg/kg LPS and phenotyped. Mice lacking GR showed significantly increased mortality, more hemodynamic instability, higher nitric oxide levels, and higher levels of the inflammatory cytokines, tumor necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6) compared with controls. There were no differences in rates of apoptosis or macrophage recruitment between the two groups. Both endothelial nitric oxide synthase (eNOS) and inducible nitric oxide synthase (iNOS) expression were increased after LPS challenge in mice with endothelial GR deficiency, and aminoguanidine, a specific iNOS inhibitor in mice was able to rescue hemodynamic collapse in these animals. In vitro, human umbilical vein cells (HUVECs) subjected to GR knockdown by siRNA showed increased expression of eNOS at baseline that persisted after treatment with LPS. Both eNOS and iNOS mRNA was increased by qPCR. In HUVECs lacking GR, NF-κB levels and NF-κB–dependent genes tissue factor and IL-6 were increased compared with controls. Thus, endothelial GR is a critical regulator of NF-κB activation and nitric oxide synthesis in sepsis.

T

he glucocorticoid receptor (GR) is a nuclear hormone receptor with wide-ranging roles in both health and disease. This receptor is a ligand-bound transcription factor that, in the absence of ligand, resides in the cytoplasm bound to Hsp90 and other stabilizing cofactors. Upon ligand binding the receptor–ligand complex translocates to the nucleus and affects gene transcription as well as a vast complement of downstream signaling pathways (1, 2). GR is the target of a number of synthetic steroids used as therapy for a wide array of autoimmune, inflammatory, and malignant conditions, as well as the receptor for the endogenous, adrenally produced steroid corticosterone. GR is present in nearly every tissue in the body and is widely conserved across species, highlighting its critical role in homeostasis and survival (3). This fact is underscored by the near uniform mortality observed in mice lacking global GR, likely due to severe lung hypoplasia (4). Thus, to address the cell-specific role of GR in mammalian systems, GR has been deleted in a tissuespecific manner. For example, deletion of GR in the central nervous system results in mice with profoundly altered hypothalamic–pituitary–adrenal (HPA) axes and 10-fold elevated circulating corticosterone levels as well as reduced anxietyrelated behavior (5). Tissue-specific excision of GR from hepatoctyes results in a severe growth deficit thought to be due to down-regulation of STAT5-mediated transcription (6, 7). Mice with tissue-specific deletion of GR in lung epithelial cells have been shown to have reduced viability (8). Therefore, the profound phenotypes observed in mice lacking GR underscores the importance of endogenous corticosterone in regulating normal homeostasis. The underpinnings of a common side effect of systemic glucocorticoid therapy, namely steroid-induced hypertension, have also been investigated in tissue-specific knockout (KO) mice. 306–311 | PNAS | January 2, 2013 | vol. 110 | no. 1

The loss of GR in the distal nephron did not protect against steroid induced hypertension (9), whereas mice lacking GR in vascular smooth muscle were initially protected but eventually became as hypertensive as controls (10). Recently, we have shown that mice deficient in endothelial GR were almost completely protected (11), demonstrating that cell-specific actions of GR are responsible for whole organism phenotypes. Given the relative resistance of endothelial GR knockout mice to steroid-mediated hypertension, as well as the key role of the endothelium in inflammatory states, here we studied the role of endothelial GR in the setting of lipopolysaccharide (LPS)induced sepsis, a state of severe hypotension and inflammation. We show that endothelial GR is a critical negative regulator of both nitric oxide (NO) release and NF-κB regulation and specifically that its elimination results in increased expression of endothelial nitric oxide synthase (eNOS) and inducible nitric oxide synthase (iNOS) and potentiation of an inflammatory milieu through prolonged activation of NF-κB. Thus, the permissive action of endogenous corticosterone acting via endothelial GR is critical for host protection from sepsis. Results Endothelial Cell GR Deficient Mice Are More Susceptible to LPSEC Induced Sepsis. Endothelial cell (EC) GR deficient mice (GR

KO ) were generated by breeding floxed (fl/fl) GR mice to Tie-1 Cre mice as previously described (11). GREC KO (n = 12), GR fl/fl (n = 13), and Tie-1 Cre mice (n = 7) were challenged with LPS (12.5 mg/kg, i.p.) and survival was monitored over a period of 96 h. GREC KO mice showed significantly increased mortality with only 42% survival 1 d post-LPS compared with both control groups with 85% and 100% survival, respectively (P < 0.05; Fig. 1A). In all cases, GREC KO mice demonstrated a more severe clinical phenotype compared with controls including diarrhea, conjunctivitis, decreased activity, and respiratory distress. As survival was similar between the control groups, we proceeded with subsequent experiments using only GREC KO and GR fl/fl mice. To investigate if the enhanced susceptibility to LPS was endothelial specific, GR was additionally excised in vascular smooth muscle (VSM) by breeding GR fl/fl mice to SM22-α Cre drivers, which target a putative calcium-binding protein present in smooth muscle, as described (10). As seen in Fig. 1B, the loss of GR in VSM (GRVSM KO) did not augment LPS-induced death with the same dose of LPS. Next, we performed dose–response experiments in control mice to determine the lethal dose of LPS. Controls (n = 5) were treated with 80 mg/kg LPS alone or pretreated (n = 6) with 2 mg/ kg dexamethasone 2 h before LPS injection. Consistent with several other studies, we found that a dose of 80 mg/kg was

Author contributions: J.E.G. and W.C.S. designed research; J.E.G., Y.F., and H.V. performed research; J.E.G. and Y.F. analyzed data; and J.E.G. and W.C.S. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1

To whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1210200110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1210200110

GR EC KO GR fl/fl Tie-1 Cre

80 60

*

40 20 0

0

1

2

3

4

100 80

*

60

20

0

1

Time (days)

2

3

4

Time (hours)

B

D 100 80

GR fl/fl

60 40

*

600

GR VSM KO

HR (bpm)

Percent survival (%)

GR EC KO GR /

LPS

40

500 LPS

GR EC KO GR /

400

20 0

0

1

2

Time (days)

3

4

300

0

1

2

3

Time (hours)

sufficient to result in >90% mortality over the 96-h time period (12, 13) (Fig. S1A). Pretreatment with dexamethasone was able to rescue nearly all mice from LPS-induced mortality at this dose. In contrast, pretreatment of GREC KO mice with the same dose of dexamethasone did not result in any improvement in survival (Fig. S1B). These data demonstrate that the heightened susceptibility of GREC KO mice to LPS is endothelial specific, the LPSinduced death in these mice was obtained with less than one-sixth of the lethal dose needed in control mice and that provision of dexamethasone provides no benefit in these animals. GREC

KO

Mice Have Lower Blood Pressure After Acute LPS Injections.

To investigate the early hemodynamic events after acute LPS administration, continual blood pressure monitoring was performed in anesthetized mice. The carotid arteries of GREC KO and control mice (n = 5 per group) were catheterized to allow indwelling pressure measurements and LPS was administered i.v. (5 mg/kg). Baseline mean arterial pressure (MAP) of GREC KO mice was slightly elevated as previously reported by telemetry (11); however, 4 h after LPS, MAP in the GREC KO mice had fallen to 40 ± 11 mmHg, whereas that in the control mice was 59 ± 4 mmHg (Fig. 1C). Analysis of the heart rate (HR) during this time showed no differences between the two groups at baseline or for the first 3 h after LPS treatment (Fig. 1D). However, at the 4-h time point, the HR of GREC KO mice had fallen to 418 ± 20 beats per minute (bpm), whereas control mice maintained their HR at 553 ± 34 bpm (P = 0.013). These data show that the loss of GR in endothelium augments LPS-induced hypotension. GREC KO Mice Have Increased Total Nitrate and Nitrite in Response to LPS. To assess why GREC KO mice had enhanced mortality after

LPS challenge, a variety of phenotypes were assessed. Initially, we examined cholesterol levels in fasted mice at baseline, as defects in the cholesterol transport pathway have been shown to contribute to the uncontrolled inflammation mediated by LPS (14). However, there were no differences between the groups. GREC KO animals (n = 7) had total cholesterol levels of 115 ± 2.5 mg/dL, whereas control mice (n = 9) had total cholesterol levels of 112 ± 2.2 mg/dL (P = 0.42, Fig. 2A). Similarly, GREC KO animals had HDL levels of 65 ± 3.8 mg/dL, whereas control mice had HDL levels of 70 ± 2.8 mg/dL (P = 0.27). Next the endogenous production of corticosterone was measured at baseline and after LPS challenge (n = 4–7 per group). At baseline, GREC KO mice (n = 7) had corticosterone levels of 404 ± 4.7 ng/mL, whereas control mice (n = 6) had levels of 256 ± 53 ng/mL (P = 0.067). Eight hours after LPS challenge, both groups showed a significant increase in corticosterone, demonstrating that the HPA axis and release of corticosterone was not different between the groups (Fig. 2B). Goodwin et al.

4

Fig. 1. Sensitivity and specificity of LPS-induced mortality in GREC KO mice with associated hemodynamic collapse. (A) GREC KO (n = 12), GR fl/fl (n = 13), and Tie-1 Cre+ mice (n = 7) were treated with LPS 12.5 mg/kg i.p. and monitored for the time shown. GREC KO mice had poorer survival compared with the control groups. (B) GRVSM KO mice (n = 5) treated with the same dose of LPS did not demonstrate increased mortality, suggesting the effect was specific to the endothelium. (C) GREC KO mice show significantly lower BP 4 h after treatment with 5 mg/kg i.v. LPS compared with controls. (D) Heart rate in GREC KO mice is also significantly lower than controls 4 h after i.v. LPS. *P < 0.05.

Because LPS promotes liver damage, the extent of liver injury was assessed via aspartate aminotransferase (AST) and alanine aminotransferase (ALT) levels at baseline and after LPS (n = 4–8/group, Fig. 2C). At this dose of LPS, there were no differences between the two groups, although there was a trend toward higher ALT levels in GREC KO mice. In addition to liver damage, we measured serum levels of tumor necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6), 8 h after challenge with LPS (Fig. 2D). At baseline there was no difference in the levels of either inflammatory cytokine between the two groups, with very low levels of TNF-α and undetectable levels of IL-6. After LPS, both groups showed a significant increase in TNF-α and IL-6 compared with baseline, but mice lacking endothelial GR exhibited a marked elevation of these cytokines. Finally, we investigated the sum of total nitrate and nitrite (NOx) in plasma as an index of nitric oxide synthase activity in both groups under each condition (Fig. 2E). At baseline, there were no differences between the groups. eNOS KO mice were included as a control and had markedly lower levels of NOx, as expected. Following LPS treatment, both groups showed a statistically significant increase in NOx levels; however, GREC KO mice had a more robust increase compared with controls. Loss of Endothelial GR Does Not Affect Apoptosis or Macrophage Recruitment After LPS. As the lung and liver are two organs

heavily affected by LPS-induced sepsis, apoptosis, via TUNEL staining, and macrophage infiltration, via CD68 staining, was examined in organs harvested from GREC KO and control mice 8 h after LPS injection. There were no differences between the two groups in baseline rates of apoptosis or degree of macrophage infiltration in either organ. In both the lung and the liver, LPS induced a statistically significant increase in apoptosis in both groups (Fig. S2 A and B). Macrophage infiltration in the lung and liver demonstrated a similar pattern with expected increases in macrophages after LPS (Fig. S2 C and D). Representative tissue sections are shown in Fig. S3. GREC KO Mice Have Increased Expression of iNOS After LPS Treatment.

In mice, LPS-induced hemodynamic responses are regulated, in part, via the NO system (15, 16). To determine if endogenous GR regulates the levels of vascular NOS, control and GREC KO mice (n = 3 per group) were treated with vehicle or LPS (12.5 mg/kg) and aortic gene expression via qPCR was examined at 8 h post-LPS treatment. GREC KO mice showed significantly elevated levels of iNOS mRNA following LPS treatment (Fig. 3A). As expected, iNOS mRNA was undetectable in vehicle-treated mice. The induction of iNOS was also confirmed via Western blotting at 4 and 8 h post-LPS in aortic tissue from GREC KO mice compared with control (Fig. 3B, with densitometry of iNOS to GAPDH below). PNAS | January 2, 2013 | vol. 110 | no. 1 | 307

MEDICAL SCIENCES

C

100

mean BP (mm Hg)

Percent survival (%)

A

100

Total cholesterol HDL

50 0

1000

*

600

Baseline LPS

400 200 0

D

75 Baseline LPS

25 0

Control

** 4000

*

3000 2000 1000

*

0

GR EC KO

Control GR EC KO

Baseline

150

LPS

100 50

IL-6 (ng/ml)

AST (U/L)

**

200

200

0

*

150

LPS

50

*

GR EC KO

Control

GR EC KO

**

40 30

Baseline

100

0

Control

Total NO ( M)

Baseline LPS

250

E

GR EC KO

Control

C 100 50

*

800

GR EC KO

Control

ALT (U/L

Corticosterone ng/ml

B

TNF - (pg/ml)

mg/dL

A 150

*

*

Baseline LPS

20 10 0

Control

GR EC KO

eNOS KO

Aminoguanidine, a Specific iNOS Inhibitor, Rescues GREC KO Mice from LPS. To test the possibility that iNOS-derived NO contributes to

cardiovascular collapse after LPS in mice, GREC KO mice were treated with the specific iNOS inhibitor, aminoguanidine (15 mg/ kg) at 2 and 6 h post-LPS treatment as previously described (17). Initially this treatment produced no change in the mortality of GREC KO mice challenged with LPS (Fig. 3C). However, when treatment with aminoguanidine was extended to include these early time points as well as at +24 and +30 h after LPS, we noted nearly 100% survival in these mice, demonstrating that inhibition of iNOS can rescue GREC KO mice from LPS-induced mortality. Endothelial GR Acts as a Negative Regulator of eNOS. Given the dramatic and rapid hemodynamic phenotype and marked increase in NOx in GREC KO mice after LPS challenge, and the concept that eNOS is necessary for iNOS induction in certain tissues (18), we also assessed eNOS expression in aortic tissue via qPCR 8 h post-LPS treatment. GREC KO mice showed significantly elevated levels of eNOS mRNAs following LPS treatment (Fig. 4A). To directly examine if the loss of GR influences eNOS mRNA levels in isolated endothelial cells, GR was reduced via siRNA in human umbilical vein cells (HUVECs) and cells were treated with LPS (10 μg/mL) for 24 h. As shown in Fig. 4B, LPS reduced eNOS mRNA levels (15) in cells treated with control siRNA (black bars); however, GR knockdown (white bars) increased eNOS mRNA levels both at baseline and after treatment with LPS. Similarly, Western blotting revealed higher levels of eNOS protein under the same conditions (Fig. 4C and quantified in Fig. 4D from multiple experiments). These data suggest that endogenous GR in EC acts as a negative regulator of eNOS in vitro. Loss of GR Prolongs Activation of NF-κB and Up-Regulates the Expression of NF-κB–Dependent Genes, tissue factor and IL-6. Previous work has

demonstrated that GR can negatively regulate NF-κB–dependent genes and suppress inflammation (19, 20). To determine whether 308 | www.pnas.org/cgi/doi/10.1073/pnas.1210200110

Fig. 2. Phenotyping of GREC KO and control animals. No differences were seen in (A) baseline cholesterol or HDL levels, (B) corticosterone levels, (C) ALT or AST levels. (D) TNF-α and IL-6 levels were similar at baseline and increased as expected after LPS. However, GREC KO mice had significantly greater levels of inflammatory cytokines after LPS. (E) NO levels after LPS increased in both groups but were significantly higher in GREC KO mice. All parameters were measured 8 h after LPS challenge.

endogenous GR regulates the extent of NF-κB activation, HUVECs were treated with TNF-α (10 μg/mL) over a range of time points and protein lysates Western blotted to assess total and phosphorylated NF-κB levels. As shown in Fig. 5A and quantified in Fig. 5B, control siRNA-treated HUVECs showed peak NF-κB activation after 15 min, which then declined over the remainder of the time points tested. In contrast, GR siRNA-treated HUVECs showed prolonged activation of NF-κB, as evidenced by higher phospho–NF-κB levels, extending until 120 min. Levels of total NF-κB were similar in both groups and did not change over the course of the experiment. Next we quantified two NFκB–dependent genes highly induced by LPS in HUVECs (21–23), tissue factor, and IL-6. To this end, mRNA levels were quantified at various time points (2–24 h) post-LPS (20 μg/mL) in HUVECs treated with control or GR siRNA. As shown in Fig. 5C, LPS induced tissue factor mRNA levels, an effect markedly increased in cells treated with GR siRNA. A similar pattern was noted for IL-6 mRNA levels in HUVECs treated with GR siRNA (Fig. 5D), although the magnitude of increase in transcript levels was not as great as with tissue factor. These data show that endogenous GR negatively regulates NF-κB activation and gene expression in HUVECs. Discussion The major finding of this study is that selective elimination of the endothelial GR results in prolonged activation of NF-κB in the setting of LPS administration. Mechanistically, continual activation of NF-κB results in increased expression of iNOS, TNF-α, and IL-6, which accounts for the increased mortality and hemodynamic changes observed in our mouse model. It is well known that glucocorticoids suppress NF-κB activation and hence they are widely used as therapeutic agents in a vast array of inflammatory and autoimmune conditions including asthma, arthritis, and lupus. However, this study uniquely demonstrates that the presence of the receptor itself is necessary for effective NF-κB suppression and highlights the constitutive role of the endothelial GR in mainGoodwin et al.

2

* Control GR EC KO

1

0

Vehicle

LPS

B

GR fl/fl None

1 hr

2 hr

4 hr

GR EC KO 8 hr

12hr

None 1 hr

2 hr

4 hr

8 hr 12 hr

iNOS GAPDH

C

Percent survival (%)

1

1.06

0.53 1.44 1.27 1.08

100

*

80 60

0.96 1.02 1.23 1.97 1.88 1.03 GR EC KO GR EC KO - 2 doses AG GR EC KO - 4 doses AG

40 20 0 0

1

2

3

4

Time (days)

Fig. 3. Increased iNOS expression in GREC KO mice following LPS challenge. GREC KO and GR fl/fl mice (n = 3 per group) were treated with either vehicle or LPS 12.5 mg/kg. After 4 and 8 h, the animals were killed and their aortas harvested and processed. (A) GREC KO mice show significantly higher iNOS mRNA 8 h after LPS treatment compared with controls. (B) Western blot of aortic homogenates shows increased iNOS expression in the GREC KO mice at both time points studied. Three separate experiments yielded similar results. (C) Aminoguanidine, a specific iNOS inhibitor, can rescue GREC KO mice after LPS. GREC KO mice were treated with LPS only (n = 11) or aminoguanidine at +2 and +6 h after LPS (n = 7) or aminoguanidine at +2, +6, +24, and +30 h after LPS (n = 7). Survival was monitored for the time shown. Mice treated with four doses of aminoguanidine had a statistically significant improvement in their survival compared with the other two groups (P = 0.04).

C

*

3 Control

2

GR EC KO

1 0

Vehicle

Con siRNA None LPS

eNOS GR GAPDH

LPS

GR siRNA None LPS

D

eNOS mRNA (fold change)

B 4

2.5

**

Control siRNA GR siRNA

1.5

*

1.0 0.5 0.0

No treatment

LPS

*

1.5 1.0

*

0.5 0.0 -

+

Control siRNA

Goodwin et al.

**

2.0

2.0

arbitrary units

A

eNOS mRNA (fold change)

taining homeostasis. In addition, we demonstrate that endotheial GR is critical for the protective actions of exogenous glucocorticoids in a model of LPS-induced sepsis. NF-κB is a sequence-specific, rapid response transcription factor whose activation is involved in a wide array of immune and inflammatory conditions (24). Glucocorticoids have long been recognized as effective therapy for many conditions mediated by NF-κB. Two hypotheses exist as to how glucocorticoids exert their effects: (i) through enhanced expression of IκB, an inhibitory protein, or (ii) through direct protein–protein interactions with NF-κB (24), and available data argue that these mechanisms may be cell specific. Whereas up-regulation of IκB has been shown to be important in lymphocytes and monocytes

(25–27), studies in endothelial cells suggest that protein–protein interactions are most important (28). However, the effect of glucocorticoids, in the absence of endogenous GR, has never been evaluated, nor have the actions of endogenous corticosterone been examined in response to LPS challenge in mice lacking cell-specific GR. Previous studies that have examined the role of constitutive NF-κB activation have typically found oncogenic phenotypes (29–31). With regard to permissive NF-κB activation in sepsis, data are scarce. However, the literature is replete with studies that have examined LPS-induced activation of NF-κB in endothelial cells as measured by tissue factor and/or plasminogen activator inhibitor-1 expression (22, 32–34). Our study is unique in that it demonstrates a time-dependent effect of endothelial GR knockdown on NF-κB activation, with the greatest effects, as measured by tissue factor and IL-6 mRNAs, at the early time points of 2 and 6 h. We presume the up-regulation of NF-κB observed at these early time points is directly responsible for the increased levels of iNOS that are observed both in vivo and in vitro. It is interesting to note that the mortality in the GREC KO mice occurs quite early in the observation period, usually less than 24 h after LPS administration. Whereas this phenotype may certainly be secondary to massive iNOS induction, we cannot rule out a role for eNOS here as well. Previous studies have suggested that increases in basal NO release, such as are observed in mice overexpressing eNOS, may reduce basal vascular tone in resistance vessels, as well as in large conduit arteries, thereby causing hypotension (35). Whereas GREC KO mice are not baseline hypotensive (11), perhaps the combined effects of LPS and eNOS upregulation produce such a swift and severe hemodynamic insult that mortality occurs very rapidly before hemodynamic compensatory mechanisms. Because the mice develop multiorgan dysfunction it is not possible to determine the ultimate cause of mortality, which undoubtedly includes hemodynamic dysfunction, but may involve overwhelming inflammation, coagulopathy, respiratory distress, and neurological insult. The up-regulation of NF-κB expression in GREC KO cells may incite a cascade of events that results in magnification of this multiorgan dysfunction resulting in the enhanced mortality. The second important finding of our study is that loss of endothelial GR results in up-regulation of iNOS mRNA levels in aortic tissue after LPS administration. iNOS is clearly induced via NF-κB and iNOS-deficient mice are protected from LPSinduced hemodynamic collapse (36). A recent investigation of the role of endothelial-specific NF-κB signaling in the pathogenesis of septic shock showed clearly that several hemodynamic and vascular parameters were markedly improved both in vivo and in vitro upon blockade of this cell-specific signaling pathway (37). Interestingly, mice with endothelial-specific expression of the NF-κB inhibitor IκB-α also showed reduced aortic iNOS

-

+

GR siRNA

LPS

Fig. 4. GR knockdown increases eNOS expression. (A) Controls and GREC KO mice (n = 3 per group) were treated with vehicle or LPS 12.5 mg/kg and aortas were processed 8 h later. eNOS mRNA was significantly higher in GREC KO mice. (B) At baseline GR knockdown in HUVECs resulted in increased eNOS mRNA. After 24 h of treatment with LPS 10 μg/mL control cells showed a decrease in eNOS mRNA, as expected, whereas GR knockout cells continued to show increased mRNA levels. (C) Representative Western blot showing increased eNOS expression in GR knockdown cells both at baseline and after LPS treatment. (D) eNOS densitometry for three separate experiments. *P < 0.05 compared with nontreated control, **P < 0.05 compared with similarly treated control.

PNAS | January 2, 2013 | vol. 110 | no. 1 | 309

MEDICAL SCIENCES

iNOS mRNA (fold change)

A

B

phospho/total NF- B

A

8 6

** **

4 2 0

0 5 15 30 60 120 0 5 15 30 60 120

tissue factor mRNA (fold change)

D **

400 **

100 50 40 30 20 10 0

Control siRNA GR siRNA * *

-

-

+

+ 2h

+

*

+ 6h

+

* + 12h

+

+

LPS

24h

mRNA fold change- IL-6

Control siRNA

C

**

8 6

GR siRNA

**

*

Control siRNA

4

*

GR siRNA

2 0 -

-

+

+ 2h

+

+

+

6h

expression when challenged with LPS (37) and improved hemodynamics. Our data showing that aminoguanidine (38–41), a selective iNOS inhibitor, rescues GREC KO mice from LPS-induced mortality suggests that augmented iNOS induction contributes to this phenotype. Multiple investigations have established a link between eNOS expression and glucocorticoids, most of them in pursuit of determining the underpinnings of glucocorticoid-induced hypertension. Studies in mice given dexamethasone, a synthetic glucocorticoid with extremely high affinity for the glucocorticoid receptor, have demonstrated significant down-regulation of eNOS expression (42, 43). More recently, the first glucocorticoid response element (GRE) was discovered in the eNOS promoter, thus strengthening the link between glucocorticoids and eNOS modulation (44). Therefore, it is not particularly surprising that elimination of this receptor may release a “brake” on eNOS expression in other settings, such as sepsis. In conclusion, our study identifies the endothelial GR as a critical regulator of NF-κB in LPS-induced sepsis. Elimination of the receptor appears to allow prolonged activation of endothelial NF-κB when challenged with an agonist such as LPS, thereby resulting in marked up-regulation of iNOS and inflammatory cytokines, which causes hemodynamic collapse and mortality. More importantly, because basal and LPS-stimulated levels of endogenous corticosterone are similar in both groups of mice, these data reveal the homeostatic importance of corticosterone signaling via endothelial GR as a critical regulator of vascular inflammation. Materials and Methods Mice. Male mice, age 8–12 wk, with tissue-specific excision of the endothelial glucocorticoid receptor, designated GREC KO mice, and littermate controls, designated GR fl/fl, were used for experimentation as described (11). These mice are fully congenic, having been backcrossed for more than 20 generations. For LPS-challenge experiments, a third group of mice, which expressed Tie-1 Cre but had an unfloxed glucocorticoid receptor, were generated. All experiments were performed according to a protocol approved by the Institutional Animal Care and Use Committee at Yale University School of Medicine and were consistent with the National Institutes of Health Guidelines for the Care of Laboratory Animals. Drug Injections. LPS from Escherichia coli strain O55:B5 (VWR International) was dissolved in PBS and prepared fresh before each experiment. A dose of 12.5 mg/kg was used according to several previously published studies (36, 45). Aminoguanidine (Cayman) was dissolved in saline and prepared fresh before experimentation and used at a dose of 15 mg/kg. Serum Measurements. Corticosterone measurement was performed by ELISA (Assay Designs) according to the manufacturer’s instructions. Nitric oxide was measured by the Total Nitric Oxide Assay kit (Assay Designs) according to the manufacturer’s instructions. Mouse IL-6 and TNF-α levels were measured by ELISAs from Pierce Biotechnology. Liver enzymes, cholesterol, and HDL

310 | www.pnas.org/cgi/doi/10.1073/pnas.1210200110

+ 12h

+

+ 24h

LPS

Fig. 5. GR knockdown in HUVECs prolongs NF-κB activation. (A) HUVECs were treated with TNF 10 μg/ mL. Both control and GR siRNA-treated cells showed similar peak NF-κB activation but GR siRNA-treated cells showed prolonged activation. (B) Densitometry of three separate experiments. (C and D) HUVECs were treated with LPS 20 μg/mL. After 2 and 6 h, GR siRNA-treated cells showed significantly increased tissue factor and IL-6 mRNA. This effect was abolished at 12 and 24 h. *P < 0.05 compared with nontreated control, **P < 0.05 compared with similarly treated control.

were measured by the Yale Mouse Metabolic Phenotyping Center by COBAS Mira Plus assay (Roche). Mice were fasted for 12–16 h before cholesterol determination. Apoptosis Determination. Tissue sections were mounted on glass slides and TUNEL staining was performed using the ApopTag Peroxidase In Situ Apoptosis Detection kit (Chemicon International). Five fields from each slide were analyzed and quantified by ImageJ software. On other slides CD68 (Serotec) staining was performed to identify macrophages, and images were also analyzed by ImageJ software. The same grayscale threshold was used for all images and the area of each tissue occupied by macrophages was calculated for three sections per sample and averaged. Blood Pressure Measurement. Male mice, 10–12 wk old, were maintained under 1.75% (vol/vol) isoflurane anesthesia. The carotid artery was catheterized and both a bladder catheter and peripheral IV line for intermittent saline injections were placed. After a 30- to 60-min equilibration period LPS, 5 mg/kg, was administered i.v. Mice were monitored continuously by oscillometric blood pressure measurement for 4–6 h after injection. Cell Culture. HUVECs were used at passage 4–8, and cultured in Endothelial Basal Medium-2 media with growth factors and 10% serum. Before experiments, they were treated with charcoal-stripped FBS (VWR International) for 24 h to eliminate the effects of small amounts of endogenous steroid in the culture media. Human GR-specific siRNA (Invitrogen) was used at a concentration of 1 nM for 48 h to effectively knock down GR. For Western blotting experiments, cells were treated with LPS 20 μg/mL for 24 h or 10 μg/mL TNF-α for the time points shown. For qPCR determination, cells were treated with 20 μg/mL LPS and treated for the time indicated. Western Blot. Tissues were snap frozen in liquid nitrogen, pulverized, and resuspended in lysis buffer (50 mM Tris·HCl pH 7.4, 0.1 mM EDTA, 0.1 mM EGTA, 1% Nonidet P-40, 0.1% sodium deoxycholate, 0.1% SDS, 100 mM NaCl, 10 mM NaF, 1 mM sodium pyrophosphate, 1 mM sodium orthovanadate, 1 mM Pefabloc SC, and 2 mg/mL protease inhibitor mixture; Roche Diagnostics). Cells were lysed on ice with lysis buffer. Protein concentrations were determined with the DC Protein assay kit (Bio-Rad Laboratories). Lysates were analyzed by SDS/PAGE and immunoblotted. Primary antibodies used include the following: GR (Thermo Scientific), eNOS (BD Biosciences), iNOS (Cayman) total NF-κB, phospho NF-κB (Cell Signaling), and Hsp90 and GAPDH (Affinity Bioreagents). Secondary antibodies were fluorescence-labeled antibodies (LI-COR Biosciences). Bands were visualized with the Odyssey Infrared LI-COR system. Quantitative PCR. Total RNA was isolated from cells using the RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. RNA was reverse transcribed using the Taqman Reverse Transcriptase kit (Applied Biosystems). qPCR was performed on a Bio-Rad iQ5 machine using the resultant cDNA, 2× SA Biosciences RT2 qPCR Master mix, and gene-specific primers. Primers used for the detection of mouse eNOS were: 5′ GTTGTACGGGCCTGACATTT 3′ and 5′ GGTCCTGTGCATGGATGAG 3′; for mouse iNOS, they were 5′ TTCTGTGCTGTCCCAGTGAG 3′ and 5′ TGAAGAAAACCCCTTGTGCT 3′; for human eNOS,

Goodwin et al.

Blood pressure and hemodynamic data were evaluated by repeated measures ANOVA with Bonferroni’s posttest. Statistical significance was set at a P value