GADD45a is a novel candidate gene in inflammatory lung injury via influences on Akt signaling

Abstract

We explored the mechanistic involvement of the growth arrest and DNA damage-inducible gene GADD45a in lipopolysaccharide (LPS)- and ventilator-induced inflammatory lung injury (VILI). Multiple biochemical and genomic parameters of inflammatory lung injury indicated that GADD45a^−/− mice are modestly susceptible to intratracheal LPS-induced lung injury and profoundly susceptible to high tidal volume VILI, with increases in microvascular permeability and bronchoalveolar lavage levels of inflammatory cytokines. Expression profiling of lung tissues from VILI-challenged GADD45a^−/− mice revealed strong dysregulation in the B-cell receptor signaling pathway compared with wild-type mice and suggested the involvement of PI3 kinase/Akt signaling components. Western blot analyses of lung homogenates confirmed ∼50% reduction in Akt protein levels in GADD45a^−/− mice accompanied by marked increases in Akt ubiquitination. Electrical resistance measurements across human lung endothelial cell monolayers with either reduced GADD45a or Akt expression (siRNAs) revealed significant potentiation of LPS-induced human lung endothelial barrier dysfunction, which was attenuated by overexpression of a constitutively active Akt1 transgene. These studies validate GADD45a as a novel candidate gene in inflammatory lung injury and a significant participant in vascular barrier regulation via effects on Akt-mediated endothelial signaling.—Meyer, N. J., Huang, Y., Singleton, P. A., Sammani, S., Moitra, J., Evenoski, C. L., Husain, A. N., Mitra, S., Moreno-Vinasco, L., Jacobson, J. R., Lussier, Y. A., Garcia, J. G. N. GADD45a is a novel candidate gene in inflammatory lung injury via influences on Akt signaling.

Acute lung injury (ALI), a common and highly morbid inflammatory syndrome, is characterized by increased pulmonary endothelial and epithelial permeability leading to alveolar flooding (1) and a mortality >35% (2). While infectious causes, such as sepsis and pneumonia, are the most common ALI precipitants, supportive mechanical ventilation therapy may incite or worsen preexisting ALI, a syndrome known as ventilator-induced lung injury (VILI) (3, 4). Consistent with the concept that ventilatory support, while vital, is nevertheless potentially injurious (4, 5), multiple animal models and in vitro studies (6,7,8,9,10) support an injurious role for excessive mechanical stress and observational studies (4) suggest that ∼25% of critically ill patients without ALI at the initiation of mechanical ventilation will develop the syndrome during their first 5 days on the ventilator. Despite identification of several risk factors for the development of ALI (including sepsis, pneumonia, aspiration, and high tidal volumes; refs. 5, 11), only a minority of patients with these risk factors develop the syndrome. This marked heterogeneity, combined with the observed disparities in ALI between different ethnic and racial populations, lends support to the hypothesis that a genetic contribution may underlie ALI susceptibility (12, 13).

We previously searched for novel ALI/VILI candidate genes via orthologous gene expression profiling of in vivo (murine, rat, and canine) and in vitro [human pulmonary endothelial cells (ECs)] models of increased mechanical stress consistent with VILI (14). These studies generated a list of mechanosensitive candidate genes unidirectionally and differentially expressed across all species, including genes strongly implicated in ALI pathogenesis as well as novel candidates previously unassociated with lung injury, ventilation, or pulmonary pathophysiology (14, 15). One novel VILI-related candidate gene thus identified was the growth arrest and DNA damage-inducible gene GADD45a (14). GADD45a exhibits low constitutive expression with transcriptional activation by cellular genotoxic and nongenotoxic stressors, including ultraviolet and ionizing radiation, hyperoxia, and endotoxin [lipopolysaccharide (LPS); refs. 16,17,18,19]. GADD45a is recognized as a participant in the regulation of the cell cycle, apoptosis, maintenance of genomic stability, DNA methylation excision and repair, and regulation of Th1 differentiation (20,21,22,23,24,25,26). The involvement of GADD45a in inflammatory lung processes, however, is unknown. We utilized in vivo models of LPS-induced lung injury and VILI and genetically engineered mice with targeted GADD45a deletion to examine the participation of GADD45a in inflammatory lung injury. Our results are consistent with a significant role for GADD45a in inflammatory lung injury with genomic and cellular analyses, suggesting GADD45a participation in vascular barrier regulation via effects on Akt-mediated endothelial signaling.

MATERIALS AND METHODS

Cell culture and reagents

Standard reagents including LPS (batch O127B8) were obtained from Sigma-Aldrich (St. Louis, MO, USA) unless otherwise specified. Human pulmonary arterial endothelial cells (HPAECs) were obtained from Cambrex (Walkersville, MD, USA) and cultured as described previously (27). For SDS-PAGE, reagents were purchased from Bio-Rad (Richmond, CA, USA), Immobilon-P transfer membrane was from Millipore (Bedford, MA, USA), and gold microelectrodes were from Applied Biophysics (Troy, NY, USA). Primary antibodies for GADD45a as well as short interfering RNAs (siRNAs) specific for GADD45a, Akt1, and control sequence were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Primary antibodies for Akt, phospho-Akt (Ser-473), and ubiquitin, as well as secondary antibodies, were purchased from Cell Signaling Technology (Danvers, MA, USA). The Akt1/PKBα (active) cDNA expression vector was purchased from Millipore.

GADD45a-engineered mice

GADD45a^−/− mice (129/Ola background) generously provided by Dr. Michael O'Reilly (University of Rochester, Rochester, NY, USA) and Dr. Albert Fornace (Brigham and Women’s Hospital, Boston, MA, USA) were backcrossed onto the C57BL/6 background for eight generations (28). Wild-type (WT) C57BL/6 mice were obtained from Jackson Laboratories (Bar Harbor, ME, USA).

Preclinical model of VILI

All experiments were approved by the Animal Care and Use Committee at the University of Chicago. Mice were housed under standard conditions with free access to food and water. Mechanical ventilation experiments were performed in age-matched male WT C57BL/6 and GADD45a^−/− mice (8–12 wk) after anesthesia with inhaled isofluorane followed by intraperitoneal ketamine/acetylpromazine (150/15 mg/kg respectively). Mice were intubated (20-gauge catheter) and placed on mechanical ventilation (Harvard Apparatus, Holliston, MA, USA) at room air, tidal volume of 30 ml/kg (VILI groups, n=8/group), 65 breaths/min, and positive end expiratory pressure of 0 cm H₂O for 4 h (29). Control mice were allowed to breathe spontaneously for 4 h (n=6/group). Ventilated WT and GADD45a^−/− mice were monitored with intermittent blood pressure and arterial blood gas monitoring to ensure adequate perfusion and received 8 ml/kg 0.9% saline at the initiation of ventilation and at 2 h after the onset of ventilation. Deep anesthesia was maintained with ketamine/acetylpromazine throughout the experiment.

Preclinical model of LPS-induced lung injury

A second model was used to determine the involvement of GADD45a in inflammatory lung injury. Age-matched male C57BL/6 and GADD45a^−/− mice (8–12 wk) were anesthetized and intubated as described above. LPS (1.25 mg/kg; n=15–16/group) or an equal volume of water (1.25 ml/kg; n=8/group) was administered intratracheally as described previously (30, 31). After extubation, mice from the four experimental groups were allowed to breathe spontaneously for 18 h with free access to food and water and then underwent harvesting as described for the mouse VILI model.

Bronchoalveolar lavage (BAL) and lung tissue expression measurements

BAL was recovered as we previously described (30) for determination of cell counts and differentials, protein (Bio-Rad DC protein assay), albumin (ELISA; Bethyl Laboratories, Montgomery, TX, USA) (31) and levels of mouse cytokines [interleukin (IL)-1β, IL-6, keratinocyte-derived chemokine (KC), macrophage inflammatory protein (MIP)-2, and tumor necrosis factor (TNF-α)] using a Bio-Plex bead assay (Bio-Rad Laboratories; ref. 32). Lung tissue was homogenized in extraction buffer and assayed for albumin by ELISA (Bethyl Laboratories).

Histology and immunohistochemistry

To characterize VILI- or LPS-mediated alterations in lung morphology and to localize GADD45a expression, left lungs (2 animals/group) were excised from the mainstem bronchus at death and placed immediately in formalin overnight, followed by embedding in paraffin for histological evaluation by hematoxylin and eosin staining as we previously described (30). A masked lung pathologist (A.N.H.) graded the severity of each slide for edema, inflammation, and the presence of hyaline membranes utilizing the following criteria: 1) edema: absent, mild (<10% alveoli involved), moderate (involving 10–50% alveoli), or severe (involving>50% alveoli); 2) inflammation; absent, mild [<10 inflammatory cells/high power field (hpf)], moderate (10–50 inflammatory cells/hpf), or severe (>50 inflammatory cells/hpf); and 3) hyaline membranes: present or absent. Immunohistochemistry was performed on paraffin-embedded sections using rabbit anti-GADD45a (Santa Cruz Biotechnology) at a 1:40 dilution as described previously (33). Each sample was graded (n=2/group) from 0 to 3+ in the following locations: pulmonary epithelium (including airway epithelium and type II pneumocytes), pulmonary endothelium, and inflammatory cells.

RNA isolation and microarray analysis

Total RNA was isolated from whole lungs for expression profiling as described previously (34) using Affymetrix Mouse 430_2 arrays and protocols (Affymetrix, Santa Clara, CA, USA). Chip quality and “present” or “absent” expression calls were determined by the GeneChip Operating Software. Intensities and normalization of probe sets were calculated by Bioconductor software (GCRMA package; refs. 35, 36). The microarray data have been submitted to the National Center for Biotechnology Information’s Gene Expression Omnibus repository (GSE11662). To identify differentially expressed genes, two-group comparisons were conducted using significance analysis of microarrays (SAM) as described previously (37). Gene filtering parameters and results are summarized in Supplemental Table S1. “Dysregulated genes” are those that were differentially expressed with >2-fold change vs. control. For redundant probe sets representing the same Entrez Gene or UniGene ID, only the probe set with the lowest false discovery rate or highest fold change was included in the gene list.

Dysregulated genes were uploaded into the Ingenuity Pathway Analysis (IPA) software (http://www.ingenuity. com), a web-delivered application that utilizes the Ingenuity Pathways Knowledge Base (IPKB) containing a large amount of individually modeled relationships between gene objects, e.g., genes, mRNAs, and proteins, to dynamically generate significant regulatory and signaling networks or pathways. The genes submitted for mapping to corresponding gene objects in the IPKB are called “focus genes.” The significance of a canonical pathway is controlled by P value, which is calculated using the right-tailed (referring to the overrepresented pathway) Fisher’s exact test for 2 × 2 contingency tables. This is done by comparing the number of focus genes that participate in a given pathway, relative to the total number of occurrences of those genes in all pathways stored in the IPKB. The significance threshold of a canonical pathway is set to 1.3, which is derived by −log₁₀ [P value], with P ≤ 0.05.

Real-time RT-PCR and analysis

Transcript levels of CXCL1, CXCL2, PIK3CD, and Akt1 in homogenized mouse lungs from VILI-challenged and spontaneously breathing mice (n=3 per condition) were measured in 96-well microtiter plates with an ABI Prism 7700 Sequence Detector System (Applied Biosystems, Foster City, CA, USA) as we have described (38). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control for normalization. All primers and probes were purchased from Applied Biosystems in a 20× mixture. Experimental protocols were based on the manufacturer’s recommendation using the TaqMan Gold RT-PCR Core Reagents Kit (Applied Biosystems). Experimental parameters were 48°C for 30 min followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. A relative quantitative method was used to analyze changes in gene expression in a given sample relative to an untreated control sample, and specific mRNA transcript levels were expressed as fold difference, calculated by raising 2 to the power of the difference in geometric means. Comparisons of the fold change in each transcript were performed using unpaired t tests.

Immunoprecipitation and immunoblotting

Left lung homogenates were sonicated in immunoprecipitation buffer (50 mM HEPES, pH 7.5; 150 mM NaCl; 20 mM MgCl₂; 1% Nonidet P-40; 0.4 mM Na₃VO₄; 40 mM NaF; 50 μM okadaic acid; 0.2 mM phenylmethylsulfonyl fluoride; and 1:250 dilution of protease and phosphatase inhibitors (Calbiochem, San Diego, CA, USA). In companion experiments, HPAEC lysates were normalized for protein concentration followed by SDS-PAGE in 4–15% gradient polyacrylamide gels, transferred onto Immobilon membranes, and incubated with specific primary and secondary antibodies. Immunoreactive bands were visualized using enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ, USA). Standardized average gray values processed from ImageQuant software (Amersham Biosciences) were obtained from immunoreactive bands for quantification.

Transfection of siRNA and cDNA constructs

HPAECs were transfected with siRNA (Santa Cruz Biotechnology) specific for GADD45a (75 nM), Akt1 (75 nM), or a scrambled sequence (50 nM) using siPORT Amine (Ambion, Austin, TX, USA) as the transfection reagent according to the manufacturer’s protocol. The Akt1/PKBα (active) cDNA construct (Millipore, Bedford, MA, USA) was diluted 1:1 in FuGENE HD (Roche, Basel, Switzerland) and transfected concurrently with siRNA (scramble or GADD45a). Transfections were carried out 48 h before transendothelial electrical resistance (TER) measurements.

Measurement of TER

HPAECs were grown to confluence in polycarbonate wells containing evaporated gold microelectrodes, and TER measurements were performed using an electrical cell substrate impedance sensing system (Applied Biophysics) as we have previously described in detail (27, 39, 40). TER values from each microelectrode were pooled at discrete time points and plotted as the mean ± se.

Statistical methodology

Results are presented as means ± se, with the exception of the array and quantitative PCR (qPCR) gene expression data that are presented as mean fold change ± sd. Physiological means were compared using Stata (Stata, College Station, TX, USA) by one-way ANOVA and corrected for multiple comparisons using the Bonferroni method. Parameters demonstrating significantly unequal variances and non-normal distributions (Akt abundance) were analyzed by nonparametric testing (Mann-Whitney rank sum test; Stata). Gene expression data (array and qPCR) were transformed from absolute expression values to a log 2 scale by determining the expression ratio [=2^(experimental expression value−control expression value)]. When the expression ratio is positive, fold change equals the ratio, whereas a negative ratio is converted to fold change by the formula −1/expression ratio]. Fold changes were compared using Student’s t test. Predetermined significance level was P = 0.05.

RESULTS

Susceptibility of GADD45a^−/− mice to VILI

To assess the role of GADD45a in inflammatory lung injury, WT C57Bl/6 (WT) and GADD45a^−/− mice were exposed to extensive levels of mechanical stress via mechanical ventilation (4 h) at a high tidal volume (30 ml/kg). WT mice exhibited modest lung injury with increased BAL cellularity (Fig. 1) and enhanced microvascular permeability reflected by increases in BAL protein and albumin (Fig. 2). BAL levels of proinflammatory cytokines (KC, MIP-2, IL-1β, IL-6, and TNF-α) were also increased in VILI-challenged WT mice, with IL-6 achieving statistical significance (P=0.03; Fig. 3).

Figure 1.

GADD45a^−/− mice exposed to high tidal volume mechanical ventilation exhibit a neutrophilic alveolitis. A) Cellular content of BAL fluid from spontaneously breathing and VILI-challenged WT and GADD45a^−/− mice. VILI induces significantly higher cell counts in GADD45a^−/− mice compared with VILI-challenged WT mice (*P=0.002) or spontaneously breathing GADD45a^−/− mice (**P=0.02). B) Dramatic increases in BAL PMNs in VILI-challenged GADD45a^−/− but not in WT mice (*P<0.001, **P<0.001 vs. GADD45a^−/− controls). BAL PMNs comprise ∼70% of BAL cells in VILI-challenged GADD45a^−/− mice (n=8/VILI group; n=6/spontaneously breathing mice). C) Representative cytospin of BAL cells from a VILI-exposed GADD45a^−/− mouse. Inset: cytospin from a VILI-challenged WT mouse (×40).

Figure 2.

GADD45a^−/− mice demonstrate increased ventilator-induced vascular permeability. A) BAL protein, an index of alveolo-capillary permeability, is significantly elevated in VILI-challenged GADD45a^−/− mice compared with VILI-exposed WT mice (*P=0.02) or spontaneously breathing GADD45a^−/−controls (**P<0.001). BAL protein content in VILI-challenged WT mice was also significantly elevated relative to spontaneously breathing WT mice (***P=0.02). B) BAL albumin content was significantly elevated by VILI challenge in WT and GADD45a^−/− mice (**P=0.001, GADD45a^−/− mice; ***P=0.05, WT mice).

Figure 3.

Elevated BAL inflammatory cytokines in VILI-challenged GADD45a^−/− mice. A) BAL IL-6 levels are shown for all 4 experimental groups. IL-6 levels were negligible in spontaneously breathing animals of both genotypes. VILI-challenged WT mice elaborated significantly more BAL IL-6 than WT controls (***P=0.03), while VILI-exposed GADD45a^−/− mice demonstrated BAL IL-6 levels significantly higher than ventilated WT mice (*P=0.004) or GADD45a^−/− controls (**P=0.03). B) BAL cytokine levels in VILI-challenged GADD45a^−/− mice are shown relative to VILI-challenged WT mice (fold change above WT-VILI response). Ventilated GADD45a^−/− mice demonstrated a >2-fold increase over the WT VILI response of each cytokine tested; KC (*P<0.001), MIP-2 (*P<0.001), TNF-α (*P=0.001), IL-1β (*P<0.001), and IL-6 (*P=0.004; n=8/VILI group; n=6/spontaneously breathing group).

In contrast to WT mice, VILI-challenged GADD45a^−/− mice exhibited greater increases in BAL cellularity, especially in neutrophils [polymorphonuclear cells (PMNs); Fig. 1], as well as increased lung vascular permeability (BAL protein and albumin; Fig. 2). In addition, VILI-exposed GADD45a^−/− mice displayed striking and significant BAL elevations in innate immunity cytokine levels (KC, MIP-2, IL-6, IL-1β, and TNF-α) compared with similarly challenged WT mice (Fig. 3). Histological examination from spontaneously breathing WT and GADD45a^−/− mice did not detect evidence of lung edema formation whereas both WT and GADD45a^−/− ventilated groups demonstrated “mild” lung edema formation (fewer than 10% alveoli involved) with minimal PMN infiltration and without hyaline membrane formation (Supplemental Fig. S1).

Susceptibility of GADD45a^−/− mice to LPS-induced vascular leak and injury

We next examined the responses of GADD45a^−/− mice exposed to a model of LPS-induced lung injury as we have previously described (29,30,31). LPS-challenged WT and GADD45a^−/− mice both sustained dramatic but comparable increases in BAL cellularity and a prominent neutrophilic alveolitis (Supplemental Table S2). However, LPS-challenged GADD45a^−/− mice elaborated significantly greater BAL protein and lung tissue albumin levels compared with WT mice (Fig. 4), indicating that genetically engineered GADD45a^−/− mice demonstrate increased susceptibility to increased alveolar capillary permeability in two distinct models of inflammatory lung injury. Similarly, WT and GADD45a^−/− mice both demonstrated markedly elevated levels of innate immunity inflammatory cytokines post-LPS challenge, with significantly higher BAL cytokine levels in LPS-challenged GADD45a^−/− mice (Fig. 4). Histological examination of WT mice exposed to LPS demonstrated mild edema formation and “moderate inflammation” (10–50 inflammatory cells/hpf), where PMNs were the dominant inflammatory cell (Fig. 5). LPS-challenged GADD45a^−/− mice exhibited similar mild edema formation but were graded as “severe inflammation” (>50 inflammatory cells/hpf), again with prominent PMNs (Fig. 5).

Figure 4.

LPS-challenged GADD45a^−/− mice exhibit greater increases in alveolo-capillary permeability than WT mice. A) Depicted are several indices of lung microvascular permeability (BAL protein content, BAL albumin content, and lung tissue albumin content) in LPS-challenged GADD45a^−/−mice compared with LPS-challenged WT mice (fold increase over LPS-challenged WT mice). LPS evoked significant elevations in each index of permeability compared with water-treated controls. LPS-treated GADD45a^−/− mice demonstrated greater levels of BAL protein (*P=0.012) and lung albumin (*P=0.008) relative to LPS-treated WT mice. BAL albumin content did not differ significantly between LPS-treated GADD45a^−/− and WT mice. B) BAL levels of innate immunity cytokines in LPS-treated GADD45a^−/− mice were compared with LPS-treated WT mice (fold increase above WT-LPS levels). GADD45a^−/− mice exhibited significantly greater levels of BAL KC (*P=0.017), BAL MIP-2 (*P=0.031), and BAL TNF-α (not statistically significant, †P=0.123; n=14–15/LPS group; n=8/water group).

Figure 5.

GADD45a^−/− mice exhibit increased inflammatory lung histology after LPS exposure. A, B) LPS-challenged WT (A) and GADD45a^−/− (B) mice demonstrate diffuse interstitial neutrophilic infiltration, which is not evident in water-challenged lungs (insets: ×100). In LPS-challenged WT mice (×200), inflammation was graded as “moderate, 10–50 inflammatory cells/hpf,” whereas LPS-treated GADD45a^−/− mice (×200) exhibited “severe” inflammation with >50 inflammatory cells/hpf. PMNs were the predominant inflammatory cell visualized in both LPS groups, and inflammation was accompanied by mild edema without hyaline membrane formation (n=4/group). C) Immunohistochemical staining of GADD45a (brown; ×200) is shown for a water-challenged mouse where GADD45a expression was graded as 2+ in the epithelium (indicated by asterisk) and 0 to 1+ in pulmonary endothelium (indicated by arrows). Minimal inflammatory cells were observed. D) Section from an LPS-challenged WT mouse demonstrates 2+ GADD45a expression in the epithelium (asterisk) and 2+ in pulmonary endothelium (thick arrow), with prominent GADD45a immunoreactivity in inflammatory cells (3+; thin arrow; n=2/group).

Immunohistochemical localization of GADD45a expression in murine inflammatory lung injury

Immunohistochemical evaluation of spontaneously breathing or vehicle-treated WT mouse lungs revealed GADD45a expression in airway epithelium and type II pneumocytes (Fig. 5; inflammatory cells were not present in control lungs), which was unchanged by 4 h exposure to VILI (Supplemental Fig. S2). WT mice challenged by LPS demonstrated increased GADD45a expression in pulmonary endothelium and alveolar epithelium compared with control animals but also showed increased GADD45a expression in pulmonary endothelium and avid GADD45a staining in inflammatory cells present in LPS-challenged lung tissues (Fig. 5). Lungs of GADD45a^−/− mice were stained with GADD45a antibody to confirm specificity of the antibody and demonstrated negligible staining.

GADD45a influences VILI-regulated gene expression

Given the observed susceptibility of GADD45a^−/− mice to inflammatory lung injury (evoked by VILI or LPS), we next identified molecular signatures that characterize the role of GADD45a in the response to VILI-mediated mechanical stress. Expression profiling of WT lung tissues identified significant VILI-mediated gene dysregulation (571 genes) by two-group comparison (SAM software summarized in Supplemental Table S1). In contrast, few genes were dysregulated by GADD45a^−/− status alone (21 genes), whereas VILI-challenged GADD45a^−/− mice demonstrated the greatest number of dysregulated genes (658 genes; see full gene list: http://phenos.bsd. uchicago.edu/publication/GADD45). Consistent with the increased inflammatory injury in VILI-challenged GADD45a^−/− mice, gene expression analysis revealed significant up-regulation of VILI-associated marker genes, such as IL-6, IL-1β, CXCL1, and CXCL2, in GADD45a^−/− mice (Fig. 6; ref. 41). Fold changes of CXCL1 and CXCL2 by array were validated by real-time qPCR, as were fold changes of Akt1 and PIK3CD (Fig. 6). IPA revealed significant overlap between signaling pathways dysregulated by VILI-exposed WT mice and VILI-exposed GADD45a^−/− mice (Fig. 7). However, the targeted deletion of GADD45a resulted in prominent increases in signaling pathways related to the B-cell receptor and axonal guidance signaling with reduced dysregulation in integrin signaling. In addition, despite significant overlap in the gene lists regulated by VILI-challenged WT and GADD45a^−/− mice, direct comparison of the 2 groups demonstrated dramatic up-regulation of several immunoglobulin genes and members of the phosphoinositide-3-kinase (PI3K)/Akt signaling family in GADD45a^−/− mice (Supplemental Table S1, gene list 4).

Figure 6.

Fold changes of ALI biomarker genes (CXCL1, CXCL2, IL6, and IL1ß) and PI3K family members relative to the spontaneously breathing WT controls. The significance of gene dysregulation is determined by two-group comparison using a t test. A) Lung mRNA microarray. VILI-exposed GADD45a^−/− mice demonstrated significantly increased expression of each marker gene [CXCL1 (*P<0.001), CXCL2 (*P=0.001), IL6 (*P=0.005), and IL1ß (*P=0.004)] compared with WT controls. VILI challenged WT mice also demonstrated increased expression of CXCL1 (*P=0.003), CXCL2 (*P=0.006), and IL1ß) (*P=0.009), with IL6 failing to achieve statistical significance (P=0.10). Fold changes for PI3K family members Akt1 and PIK3CD are also shown. Both VILI groups showed increased expression of PIK3CD (*P=0.05 WT, *P=0.03 GADD45a^−/−), whereas neither VILI group demonstrated a change in Akt1 expression. B) Real-time qPCR validation of selected genes from lung homogenates. VILI-challenged WT and GADD45a^−/−mice demonstrated increased expression of CXCL1 (*P=0.02 WT; *P=0.002 GADD45a^−/−) and CXCL2 (*P=0.03 WT; *P=0.03 GADD45a^−/−), whereas neither VILI group demonstrated a significant change in Akt1 expression. The change in PIK3CD expression in VILI-exposed GADD45a^−/−mice did not achieve statistical significance (†P=0.21).

Figure 7.

Expression profiling reveals a signature of increased B-cell receptor signaling dysregulation in VILI-challenged GADD45a^−/− mice. IPA reveals canonical pathways enriched in dysregulated genes in VILI-challenged groups relative to spontaneously breathing WT control lungs. Dashed line represents threshold level for significance (1.3, negative log of P value 0.05 by Fisher’s exact test). The B-cell receptor signaling pathway exhibited the greatest dysregulation in the VILI-exposed GADD45a^−/− group, with components that include members of the PI3K/Akt pathway.

Role of PI3K and Akt in GADD45a-mediated susceptibility to lung injury

The B-cell receptor signaling pathway, containing both immunoglobulin genes and PI3K pathway genes (PI3K and Akt), emerged as a highly dysregulated canonical signaling pathway in VILI-challenged GADD45a^−/− mice, second only to specific B-cell components that encode various chains of an autoreactive immunoglobulin of the J558 family, (implicated in autoimmune disorders) (42,43,44,45). While the J558 immunoglobulin chains were dramatically elevated (100- to 1000-fold increased expression) in GADD45a^−/− mice (http://phenos.bsd.uchicago.edu/publication/GADD45), these autoreactive immunoglobulins were not viewed as likely key VILI targets, as their expression was modified almost exclusively by GADD45a status and not by VILI exposure. In addition, an immune complex-driven inflammatory lung injury phenotype would be unlikely to manifest within the 4 h of VILI exposure. Consequently, we focused on PI3K and Akt, members of the VILI-relevant highly dysregulated B-cell receptor pathway, as targets in VILI in GADD45a^−/− mice.

Initial studies investigated the contribution of the PI3K/Akt pathway to GADD45a^−/− VILI susceptibility by evaluating the abundance of total Akt in mouse lung homogenates. Marked differences in Akt levels were noted in lung homogenates from GADD45a^−/− mice compared with WT mice, with GADD45a^−/− lungs demonstrating a significant reduction in total Akt protein levels whether comparing lungs from spontaneously breathing mice (55±21% of total WT Akt expression; P = 0.02) or VILI-challenged mice (38±8% of WT total Akt expression; P = 0.04; Fig. 8A).

Figure 8.

GADD45a^−/− mice demonstrate decreased lung expression of Akt protein. A) Inset: representative immunoblot of spontaneously breathing and VILI-challenged lung homogenates probed for Akt with dramatic decreases in total Akt protein expression in GADD45a^−/− lungs. Bar graph: pooled Akt densitometric quantitation expressed as standardized average gray values, normalized for protein loading and standardized to WT control lungs. GADD45a^−/− lungs exhibit reduced Akt expression (*P=0.02 for spontaneously breathing GADD45a^−/− mice vs. WT mice, **P=0.05 for VILI- or LPS-challenged GADD45a^−/− lungs compared with WT mice). B) Relative Akt activation is depicted as the ratio of phosphorylated Akt (Ser-473)/total Akt. Inset: lung homogenates of spontaneously breathing, VILI-challenged, and LPS-challenged mice probed for Akt, p-Akt, and β-actin. In WT mice, VILI induced an ∼2-fold increase in Akt activation while LPS challenge provoked an ∼6-fold increase (**P=0.04). Despite the reduction of total and p-Akt in GADD45a^−/− lungs, LPS-challenged mice demonstrated increased Akt phosphorylation similar to LPS-challenged WT mice (*P=0.04). VILI challenge, in contrast, did not significantly increase the p-Akt/Akt ratio in GADD45a^−/− mice. C) Pooled results (n=3 mice/group) of relative ratio of immunoreactive ubiquitin and Akt (anti-Akt) obtained from Western blots of immunoprecipitates from WT and GADD45a^−/− mouse lung homogenates (*P=0.01). Inset: Western blots for ubiquitin and Akt in lung homogenates from 3 WT and 3 GADD45a^−/− mice.

To investigate the extent of Akt activation in these mouse models, p-Akt levels were detected by immunoreactivity to phospho-specific antibody recognizing Ser-473, an Akt site phosphorylated by PI3K. Akt activity varied by lung injury model in WT mice, with LPS challenge provoking an ∼6-fold increase in the ratio of activated (phosphorylated) Akt to total Akt (p-Akt/Akt), whereas VILI evoked only an ∼2-fold increase (Fig. 8B). In GADD45a^−/− lungs, despite the diminished total Akt abundance, LPS challenge still produced an approximate 6-fold increase in p-Akt/Akt levels, whereas it was difficult to detect any increase in Akt activation in VILI-challenged GADD45a^−/− lungs (Fig. 8B).

To address a mechanism for the observed reduction in Akt levels in GADD45a^−/− mice, we explored alterations in Akt transcriptional regulation but failed to observe differential regulation of Akt gene expression (Affymetrix array, qPCR) between GADD45a^−/− and WT lungs (Fig. 6). As these results suggested that reduced Akt levels may reflect post-translational modification and protein degradation, we next generated Akt immunoprecipitates, from WT and GADD45a^−/− mouse lung homogenates followed by Western blotting, to detect ubiquitin immunoreactivity consistent with Akt protein ubiquitination. Figure 8C depicts the markedly increased levels of ubiquitinated Akt in lung homogenates from GADD45a^−/− but not WT mice, strongly suggesting an increase in Akt protein degradation in GADD45a^−/− mice.

GADD45a involvement in LPS-induced EC barrier regulation via an Akt-dependent pathway

Given the observed significant increases in LPS- and VILI-induced vascular permeability and decreased Akt protein abundance in GADD45a^−/− mice, we next investigated the mechanistic contribution of Akt signaling to increased microvascular permeability during lung inflammation. Initial experiments examined the effects of GADD45a depletion (siRNA) on human lung EC total Akt levels. Similar to the reduced Akt levels in lung homogenates derived from GADD45a^−/− mice, in vitro GADD45a depletion resulted in decreased levels of total Akt (20±9%) compared with control ECs (scramble sequence siRNA; Fig. 9A, B). In addition, Akt activation increased after LPS stimulus in both control and GADD45a-silenced endothelium after LPS, although the total Akt abundance was decreased in GADD45a-silenced cells. We next utilized an in vitro model of vascular barrier regulation with human lung endothelium grown on gold microelectrodes with continuous measurement of TER across cell monolayers. Despite GADD45a depletion, unstimulated EC monolayers demonstrated normal basal barrier integrity (Fig. 9C, D). Control endothelial monolayers exposed to exogenous LPS (1 mcg/ml) exhibited modest declines in TER values peaking at 6 h; however, LPS challenge of endothelium with reduced GADD45a expression exhibited greater declines in TER values as well as slower and incomplete barrier recovery, indicating augmented vascular barrier dysfunction. A fall in TER values of a similar magnitude was observed when ECs were depleted of Akt1 (siRNA) and subsequently challenged with LPS (Fig. 9B, C). The simultaneous addition of siRNA oligonucleotides directed against both GADD45a and Akt did not result in further LPS-induced TER decrements compared with either siRNA alone, indicating the absence of additive or synergistic effects. However, overexpression of constitutively active Akt1/PKBα mitigated the LPS-induced declines in TER, even in the presence of GADD45a depletion (Fig. 9B–D). Transfection of the active Akt1/PKBα transgene did not alter TER in the control siRNA-treated cells. Together these results are consistent with a key role for Akt in GADD45a^−/− susceptibility to inflammation-induced lung vascular barrier dysfunction and support the hypothesis that the increased susceptibility to inflammatory lung injury and barrier dysfunction in GADD45a^−/− mice involves the critical loss of Akt and Akt-related signaling components.

Figure 9.

Expression of a constitutively active Akt transgene reverses the effects of GADD45a depletion in LPS-induced EC barrier dysfunction. A) HPAECs were treated with scramble siRNA, GADD45a siRNA, or AKT1 siRNA and immunoblotted for GADD45a or AKT, resulting in effective knockdown of GADD45a and AKT expression. Bottom bands demonstrate effective transfection of a Myc-His tagged constitutively active AKT1/PKBα transgene in HPAECs. B) Lung endothelium with GADD45a depletion (siRNA) or control siRNA was treated with LPS or vehicle and total Akt levels determined at 4 h. Depletion of GADD45a reduced Akt expression, though did not abolish Akt phosphorylation. C, D) Human ECs were transfected with siRNAs to deplete either GADD45a or AKT1 or both, or with an Akt/PKB cDNA to increase expression of constitutively active Akt. Treated cells were then assessed for TER after vehicle or LPS (arrow). Data are means ± se for 3–5 experiments/condition. C) TER tracings in which GADD45a depletion exaggerates LPS-induced declines in TER (gray circles). However, ECs depleted of GADD45a (siRNA) but with Akt overexpression (black solid line) failed to demonstrate the GADD45a siRNA-mediated exacerbation of LPS-induced declines in TER values, but instead mirrored the resistance of scramble siRNA-LPS treated cells, as shown in D. D) Normalized TER values at 6 h following LPS in endothelium transfected with either scrambled siRNA or GADD45a siRNA and 1 of 4 treatments: vehicle, LPS, AKT siRNA and LPS, or AKT overexpression and LPS. GADD45a depletion increased LPS-induced permeability at 6 h (*P=0.006 vs. siScr/LPS; **P<0.001 vs. siG45/vehicle). When Akt1 was depleted or cells were transfected with both Akt1 and GADD45a siRNAs concurrently, there were no differences noted between depletion of GADD45a alone, Akt1 alone, or the two in combination. Overexpression of c/a Akt1 abolished the effect of GADD45a depletion on LPS-induced barrier function (***P=0.043 vs. siG45/LPS), while it did not alter the resistance of scramble siRNA LPS-treated cells. Dashed line emphasizes the LPS effect on scramble siRNA-treated cells.

DISCUSSION

Mechanical ventilation, a life-saving intervention for critically ill patients with respiratory failure, exhibits the potential to contribute to inflammatory lung injury as highlighted by the decreased mortality in acute respiratory distress syndrome (ARDS) patients treated with a low tidal volume ventilation strategy (46). Furthermore, patients without ALI at the initiation of mechanical ventilation are at risk to progress to fully develop the syndrome, particularly if higher tidal volume ventilation strategies are utilized (4, 5, 47). Thus, a deeper understanding of the genetic factors involved in susceptibility to or severity of ALI/ARDS is key to our understanding of ALI pathogenesis and to the design of novel therapies (12, 13, 48,49,50,51). To our knowledge, our prior genomic studies (14, 52) were the first to suggest GADD45a as a novel ALI/VILI candidate gene, a suggestion strongly supported by our current studies indicating GADD45a as a novel therapeutic target in inflammatory lung injury. We used similar genomic approaches to successfully identify pre-B-cell colony enhancing factor (PBEF) as a novel ALI/VILI candidate gene and biomarker (33). Our recent studies validated PBEF as a potential VILI therapeutic target (38) and, interestingly, noted increased GADD45a expression in VILI- and PBEF-challenged mice with GADD45a expression strongly correlated with VILI severity (38). In the present study, GADD45a expression was also increased 1.98-fold in VILI-challenged WT mice (data not shown), a result consistent with prior VILI models (14, 52). While our immunohistochemical staining of VILI-challenged lungs did not demonstrate increased GADD45a protein expression (Fig. 5), we feel this may reflect the time course of our VILI challenge; lungs were harvested immediately after 4 h ventilation, which may have been insufficient to see changes in GADD45a translation. More recently, GADD45a expression was noted to be increased in the peripheral whole blood of patients during the acute stages of ARDS, declining in surviving patients during the recovery phase (53).

Although there is limited information as to the role or function(s) of GADD45a in lung homeostasis and disease, GADD45a is well recognized as a participant in cell cycle regulation, DNA repair, maintenance of genomic stability, as well as in regulation of p53-dependent and p53-independent apoptosis (18, 23, 24, 54). In addition, GADD45a interacts with innate and adaptive immune systems to regulate T-cell differentiation and proliferation (26). To link GADD45a availability and function to inflammatory lung injury, we exposed mice with targeted deletion of GADD45a to two models of lung inflammation and noted that GADD45a^−/− mice exhibited increased susceptibility to inflammatory lung injury after either LPS or excessive mechanical stress (VILI), the first characterization of a lung injury phenotype in GADD45a^−/− mice. Prior reports (55) describing exposure of GADD45a^−/− mice to hyperoxic oxidative stress revealed a similar phenotype as WT mice, highlighting the distinctive mechanisms underlying VILI-, LPS-, and hyperoxia-induced lung injury. In addition, whereas GADD45a expression in hyperoxia-challenged mice was confined to the lung epithelium (55), we observed GADD45a expression in epithelial cells and ECs after LPS and VILI challenge, with further increases in pulmonary endothelial expression 18 h after LPS. In both our LPS and VILI models of lung injury, GADD45a^−/− mice exhibited both increased neutrophilic inflammation and increased alveolo-capillary permeability. We feel these results strongly implicate a role for GADD45a in alveolar and microvascular barrier regulation during LPS-induced inflammatory lung injury and VILI.

To better dissect the role of GADD45a in inflammatory processes such as permeability, apoptosis, and leukocyte infiltration, we used genomic approaches that demonstrated significantly altered gene expression responses in GADD45a^−/− mice after ventilation. Remarkably, despite only modest physiological aberrations in VILI-challenged WT mice, 4 h of high tidal volume ventilation triggered dramatic alterations in gene expression with extensive up-regulation in a number of ALI candidate genes (CXCL1, CXCL2, IL-6, and IL-1β; refs. 41, 52). The molecular signature of ventilated GADD45a^−/− animals was qualitatively similar to that of ventilated WT mice but quantitatively amplified, with increased expression of each putative ALI marker gene (CXCL1, CXCL2, IL-6, and IL-1β) at levels that were at least two times greater than VILI-exposed WT mice (Fig. 6), results that were verified by qPCR and by BAL protein levels (Fig. 3B). These results appear to be consistent with the potential utility of these chemokines as early biomarkers of ALI or VILI (41).

We used IPA to identify biological pathways relevant to the mechanistic consequences of GADD45a deletion in the context of inflammatory lung injury. Several signaling pathways (ERK/MAPK, acute-phase response, IL-6, and axonal guidance; Fig. 7) were exclusively dysregulated in VILI-challenged GADD45a^−/− mice and may yield further insights into the role of GADD45a’s in resisting lung injury, which we hope to investigate further. It was also interesting to note that VILI-challenged GADD45a^−/− mice displayed only a modest increase in dysregulation of the NF-κB pathway, despite the marked increase in TNF-dependent inflammatory cytokines compared with ventilated WT animals. Examination of the pathway members annotated in the NF-κB pathway by the IPA software revealed that only IL-1β and IL1R2 are included in this pathway, whereas CXCL1 and CXCL2 are not.

We were struck by the dramatic dysregulation occurring in the B-cell receptor signaling pathway of ventilated GADD45a^−/− mice, with marked up-regulation of genes involved in the PI3K/Akt pathway. Included in this pathway are upregulated genes encoding the catalytic domain (p110δ) of PI3K (PIK3CD), the PI3K regulatory region, and a PI3K adaptor protein (PIK3AP or B-cell adaptor protein), which exhibited ∼20-fold increases in gene expression (http://phenos.bsd.uchicago.edu/publication/GADD45). PIK3AP is the tyrosine kinase substrate that bridges B-cell antigen receptor-associated kinases to PI3K activation (56). Activity of the PI3K p110δ catalytic domain (PIK3CD) is required for efficient neutrophil trafficking to cytokine-stimulated endothelium (57) and is the dominant PI3K isoform dictating cytokine secretion by natural killer cells (58), a possible link between innate and adaptive immunity (59,60,61).

A key target for the enzymatic activity of PI3K is the prosurvival signaling protein known as Akt, with both PI3 and Akt implicated in regulating responses to VILI. Recently, mice deficient in PI3Kγ were found to be physiologically and morphologically protected from ex vivo VILI (62). Pharmacological inhibition of PI3K was also protective in an ex vivo murine VILI model, whereas lung injury was potentiated in animals pretreated with an Akt inhibitor (63). We sought to determine whether alterations in the PI3K pathway were evident in VILI-challenged GADD45a^−/− mice and noted striking reductions in total Akt levels in GADD45a^−/− mice, an observation suggesting that alterations in Akt might underlie the GADD45a^−/−-induced susceptibility to VILI. Whereas several mechanisms exist that potentially couple the loss of GADD45a to dramatic reductions in Akt protein levels, both microarray analysis and qPCR failed to identify differences in Akt gene expression between groups, signifying that the GADD45a^−/− effect is unlikely to reflect altered Akt transcriptional regulation. Further investigation into the potential role of GADD45a in promoting Akt protein stability indicated that GADD45a depletion is strongly associated with increases in Akt ubiquitination, indicating that GADD45a modulates protein degradation pathways that affect Akt stability. The potential for GADD45a to modify the availability of Akt chaperone proteins (e.g., heat shock protein 90) or other posttranscriptional modifiers such as microRNAs is also currently being examined.

Our observations complement prior reports of potentiation of VILI in the presence of Akt inhibition as well as increasing recognition of the role of PI3K and Akt signaling in lung vascular barrier regulation (64,65,66). We examined whether GADD45a depletion affected human lung EC barrier function as suggested by in vivo results in LPS- and VILI-challenged GADD45a^−/− mice. We determined exaggerated declines in TER values after LPS treatment in human ECs treated with siRNA directed against GADD45a, suggesting a direct and significant role for GADD45a in EC barrier regulation after inflammatory stress. Interestingly, reductions in Akt1 expression (siRNA), followed by LPS challenges, produced TER values that essentially mirrored EC barrier responses after GADD45a depletion. Furthermore, reductions in the expression of both GADD45a and Akt failed to produce an additive effect, suggesting a common mechanism. Importantly, increased expression of Akt1 attenuated LPS-induced declines in endothelial monolayer integrity associated with GADD45a depletion without altering the TER of control cell monolayers. These data indicate that GADD45a regulates Akt availability via control of ubiquitination and protein degradation and that Akt is a key molecular target in mechanical stress- and inflammation-induced lung injury. As we have previously implicated Akt in the promotion of EC barrier function via transactivation of the sphingosine-1-phosphate receptor (S1P₁) resulting in Rac-dependent increases in cortical actin (64), we speculate that one mechanism by which GADD45a may contribute to vascular barrier function may be via stabilization of Akt interaction with S1P₁. Further investigation to determine the precise mechanisms by which GADD45a interacts with Akt may yield an enhanced understanding of the significance in Akt in the pathogenesis of VILI. In summary, this work strongly validates the candidate gene approach in identifying novel candidate genes such as GADD45a and elucidating gene involvement in inflammatory lung injury. Furthermore, this work supports a direct role for GADD45a in the regulation of lung microvascular barrier function. Future investigations should consider strategies designed to increase GADD45a and Akt lung expression as a platform for future targeted genotype-based therapy in critically ill patients at risk for ALI and VILI.