The protective role of MLCP-mediated ERM dephosphorylation in endotoxin-induced lung injury in vitro and in vivo

The goal of this study was to investigate the role of MLC phosphatase (MLCP) in a LPS model of acute lung injury (ALI). We demonstrate that ectopic expression of a constitutively-active (C/A) MLCP regulatory subunit (MYPT1) attenuates the ability of LPS to increase endothelial (EC) permeability. Down-regulation of MYPT1 exacerbates LPS-induced expression of ICAM1 suggesting an anti-inflammatory role of MLCP. To determine whether MLCP contributes to LPS-induced ALI in vivo, we utilized a nanoparticle DNA delivery method to specifically target lung EC. Expression of a C/A MYPT1 reduced LPS-induced lung inflammation and vascular permeability. Further, increased expression of the CS1β (MLCP catalytic subunit) also reduced LPS-induced lung inflammation, whereas the inactive CS1β mutant increased vascular leak. We next examined the role of the cytoskeletal targets of MLCP, the ERM proteins (Ezrin/Radixin/Moesin), in mediating barrier dysfunction. LPS-induced increase in EC permeability was accompanied by PKC-mediated increase in ERM phosphorylation, which was more prominent in CS1β-depleted cells. Depletion of Moesin and Ezrin, but not Radixin attenuated LPS-induced increases in permeability. Further, delivery of a Moesin phospho-null mutant into murine lung endothelium attenuated LPS-induced lung inflammation and vascular leak suggesting that MLCP opposes LPS-induced ALI by mediating the dephosphorylation of Moesin and Ezrin.


MLCP is involved in endothelial barrier protection against LPS-induced lung vascular injury in vitro and in murine model of ALI.
Our group has previously shown that MLCP has a critical role in EC barrier strengthening induced by extracellular purines [16][17][18] and in protection against thrombin-induced EC barrier compromise in vitro 18 . To examine the role of MLCP in EC barrier compromise in a cell culture model we first analyzed LPS-induced permeability changes to identify the effective dose of LPS in human pulmonary artery endothelial cells (HPAEC). We show that LPS concentrations in the range of 0.2-1 μ g/ml induced comparable decreases in TER with maximal effect at 6-8 hours of treatment (Fig. 1A). Consequently, we used the lowest concentration (0.2 μ g/ml) in all subsequent experiments. To determine whether MLCP activity has a protective role in the LPS model of EC barrier compromise, we performed ECIS experiments with EC overexpressing the C/A truncated MYPT1 (C/A MYPT1 1-300). HPAEC were transduced with LacZ (control) or with C/A MYPT1-myc adenovirus. 24 hours after infection, HPAEC were treated with LPS or vehicle-control, and TER was monitored for 12 hours. We found that the overexpression of C/A MYPT1 attenuates LPS-mediated decrease in HPAEC barrier function (Fig. 1B). These results are consistent with our previously published observations indicating that depletion of MYPT1 significantly attenuated HPAEC barrier enhancement induced by extracellular purines (ref. 17,18). To further define the role of MLCP in LPS-induced ALI in vitro, we examined the effect of MYPT1 depletion on ICAM1 expression in LPS-treated HPAEC ( Fig. 1C-E). Our results demonstrate that LPS increases ICAM1 expression levels after 4 hours in a time-dependent manner in non-targeting siRNA-transfected control cells (Fig. 1C). MYPT1 depletion exacerbated the effect of LPS (Fig. 1D) strongly supporting the direct anti-inflammatory role of MLCP in this LPS model of endothelial injury.
To evaluate whether MLCP activity has a barrier-protective and anti-inflammatory role in LPS-induced vascular barrier injury in vivo, we utilized a LPS murine model of ALI in conjunction with a lung-targeted gene delivery method which we have recently developed 27 . In this approach, commercial (Polyplus) cationic polymer jetPEI (polyethyleneimine) transfection reagent forms a complex with the DNA of interest that specifically targets the lungs with preferential localization to the endothelium 27 . Utilizing this approach, plasmid encoding C/A MYPT1 1-300 or vector control (empty pcDNA 3.1) complexed with JetPEI was introduced intravenously (i.v.). Three days (72 hrs) later, LPS or vehicle was instilled intratracheally (i.t.). Capillary leak was estimated using Evans Blue Dye-conjugated albumin, and lung inflammation was evaluated by MPO staining as described in Materials and Methods. The efficiency of the in vivo gene delivery was confirmed by immunoblotting of lung homogenates with c-myc-specific antibody. Results in Fig. 2 demonstrate that overexpression of C/A MYPT1 significantly attenuates LPS-mediated lung vascular permeability increase ( Fig. 2A). Furthermore, ectopic expression of C/A MYPT1 dramatically decreased lung inflammation caused by LPS (Fig. 2B). Therefore, our results indicate that MLCP plays an anti-inflammatory and barrier-protective role against LPS-induced ALI in mice.
To further confirm our findings, we introduced expression plasmids containing either wild type CS1β (C/A) or the dominant negative (D/N) mutant CS1β (MLCP catalytic subunit) into mouse lungs using the JetPEI-assisted delivery approach. Our results demonstrate that wild type CS1β significantly reduces LPS-induced inflammatory processes in the mouse lung, as inferred from a reduction of protein content in BALF (Fig. 2C). Conversely, introduction of the D/N CS1β mutant significantly increases vascular leak (Fig. 2D). Collectively, our present results strongly implicate the involvement of MLCP activity in EC barrier strengthening in vitro and in a murine model of LPS-induced ALI.

LPS-induced increases in MLC and ERM phosphorylation involves activation of different signaling pathways.
We have previously shown that MLCP substrates, MLC and ERM proteins, are involved in EC barrier regulation in various endothelial permeability models 18,20,28 . To clarify the involvement of LPS in MLC and ERM phosphorylation and find a link between MLC and ERM phosphorylation level changes we analyzed the levels of di-phospho-MLC and phospho-ERM in HPAEC. ERM and MLC phosphorylation in LPS-challenged endothelial cells was analyzed by immunoblotting at discrete time points (0, 1, 2, 4, 6, 8 hours) utilizing an anti-di-phospho MLC antibody (Thr 18/Ser19) and a phospho-specific ERM antibody (phospho-Ezrin Thr567/ Radixin Thr564/Moesin Thr558). Our data demonstrate that LPS induced a gradual increase in Thr phosphorylation of ERMs which correlated with LPS-induced permeability changes (Fig. 1A). In contrast, the increase in MLC phosphorylation (maximum at 2-4 hours) preceded the maximal LPS-mediated EC barrier compromise (Fig. 1A). This data suggest that, similar to the thrombin model 29 , the increase in MLC phosphorylation may be involved in the initiation of LPS-caused EC barrier compromise, whereas increase in ERM phosphorylation is important for the development of the maximal response. Therefore, LPS-induced increases in MLC and ERM phosphorylation require activation of different signaling pathways. To confirm that ERM are substrates of MLCP in HPAEC, we depleted the MLCP catalytic subunit, CS1β , utilizing a siRNA approach. HPAEC were transfected with a specific-CS1β siRNA or a scrambled siRNA control, then treated with LPS for the indicated time periods, and the level of ERM phosphorylation was determined by immunoblotting with a phospho ERM antibody, as described in "Materials and Methods". Our results show that   To evaluate the kinases involved in LPS-induced MLC and ERM phosphorylation, we used pharmacological approach. We applied Rho-kinase inhibitor, Y-27632 (10 μ Μ ), PKC-inhibitor, Ro-31-7459 (10 μ Μ ) or DMSO-control (0.1%) one hour before adding LPS (0.2 μ g/ml) for six hours. Inhibition of PKC, but not Rho-kinase, abolishes LPS-induced ERM phosphorylation in human EC. In contrast, Rho-kinase, but not PKC, is involved in LPS-induced MLC phosphorylation (Fig. 5A). Therefore, we identified PKC, but not Rho kinase as a major kinase involved in ERM phosphorylation in LPS-challenged EC. In contrast, the major kinase responsible for MLC phosphorylation is Rho kinase, as expected.

The role of ERM proteins in LPS-induced EC barrier dysfunction. Our recently published results
demonstrate that ERM proteins play important, but different roles in the thrombin-induced EC barrier dysfunction: Moesin, and to a lesser extent Ezrin promote barrier dysfunction, and Radixin opposes it 28 . However, we have also shown that individual ERM protein depletion significantly attenuates the EC permeability increase induced by the microtubule disruptor, 2-methoxyestradiol (2ME) 20 . Further, HPAEC monolayers overexpressing phospho-null ERM mutants, exhibit less attenuation of 2ME-induced hyper-permeability, in response to PKC inhibition, implicating the phosphorylation of all ERMs in EC barrier dysfunction 20 . Collectively, this data suggest that involvement of ERM proteins in EC barrier regulation is likely to be agonist-specific. To examine the functional role of individual ERM proteins in LPS-induced EC barrier dysfunction, TER was used to detect permeability changes in Moesin-, Ezrin-or Radixin-depleted HPAEC. Non-targeting siRNA transfected cells were used as controls. The efficiency of the depletion of each protein was confirmed by immunoblotting. Our data demonstrates that Moesin (Fig. 6A) and Ezrin (Fig. 6B) depletion significantly attenuated the effect of LPS on TER, while depletion of Radixin slightly decreased basal TER, but did not significantly alter the LPS response (Fig. 6C). Therefore, our data demonstrates that ERM proteins have different roles in LPS-induced EC barrier compromise and confirms that the involvement of these proteins in EC barrier regulation may be agonist-specific. The role of Moesin phosphorylation in LPS-induced vascular barrier dysfunction in vivo. We have previously shown that Moesin is the predominant ERM protein in HPAEC 28 . It is involved in EC barrier compromise in several permeability models including LPS-caused EC barrier dysfunction in vitro 20,28 (Fig. 3A). Moesin is activated by phosphorylation at Thr558 25 . To assess the role of Moesin phosphorylation in vascular barrier regulation in vivo we generated a phospho-null Moesin mutant (Thr558 to Ala 18 ) and utilized an in vivo LPS murine model of ALI in conjunction with jetPEI delivery method, as we have recently described 27 . Mammalian expression plasmids coding either phospho-null Moesin mutant in pcDNA 3.1 or a control empty pcDNA 3.1 plasmid were injected into mice via a tail vein 27 . The overexpression of phospho-null Moesin mutant significantly decreased lung permeability inferred from EBD extravasation (Fig. 7A) and also decreased the protein level in BALF (Fig. 7B) from mouse lungs treated with LPS, indicative of lung inflammation. This data demonstrated for the first time that Moesin phosphorylation is involved in LPS-induced ALI in vivo, therefore supporting barrier-protective role of MLCP (which dephosphorylate Moesin) in vitro and in vivo.

Discussion
Endotoxin lipopolysaccharide (LPS), a constituent of the cell wall of Gram-negative bacteria, is a potent pro-inflammatory agent that directly disrupts EC barrier function in both macro-and microvascular EC 5,30 LPS-mediated murine lung injury, is a disease model that shares many characteristics of sepsis-induced ALI/ ARDS in humans. Specifically, the injury elicited is characterized by neutrophil infiltration into the lung in association with increased pro-inflammatory mediators including nuclear factor (NF)-κ B which leads to the up-regulation of pro-inflammatory cytokines such as tumor necrosis factor (TNF)-α , IL-6, IL-8, the up-regulation of ROS and RNS generators (NADPH oxidase, iNOS), increased formation of protein-rich alveolar fluid (edema), structural damage to the alveolo-capillary membrane and compromised lung function 31,32 .
It has been well established that edemagenic agents like thrombin or histamine induce cytoskeletal rearrangement via MLCK, PKC and activation of RhoA leading to the formation of paracellular gaps and the loss of EC barrier integrity 7,33,34 . The precise mechanisms by which LPS-induced signaling leads to EC barrier disruption are not yet fully clarified. However, it is clear that LPS acts through the activation of cell surface TLR4 which leads to reorganization of the actin cytoskeleton and increased EC permeability (reviewed in ref. 8). Similar to thrombin, LPS promotes cytoskeletal changes that are dependent, at least in part, on the activation of PKC and Rho signaling Pathways that can protect the endothelial barrier against LPS-induced lung injury are far less understood. Our recently published data have demonstrated barrier-protective effects of HSP-90 inhibitors, DDAH II (dimethylaminohydrolase II) overexpression, microtubule stabilization, extracellular β -NAD (β -nicotinamide adenine dinucleotide) and extracellular purines, ATP and adenosine 27,35,[41][42][43][44] . We have previously shown that purine-induced EC barrier strengthening is tightly linked to activation of MLCP and MLCP is involved EC barrier protection against thrombin-induced EC barrier compromise in vitro [16][17][18] . However, the role of MLCP in LPS-induced lung injury in vitro and in vivo has not been described and was the aim of the current study.
We demonstrated that the ectopic expression of a C/A MLCP regulatory subunit (truncated MYPT1, 1-300) significantly attenuates the LPS-induced decreases in TER reflecting EC barrier protection. In contrast, down-regulation of MYPT1 leads to exacerbation of LPS-induced expression of ICAM-1 suggestive of a direct anti-inflammatory function of MLCP. To further define the role of MLCP in LPS-induced ALI in vivo, we utilized a nanoparticle DNA delivery method which we recently developed 27 to specifically target lung EC. Using this method, we demonstrated that ectopic expression of C/A MYPT1 reduces lung inflammation (MPO staining) and attenuates LPS-induced vascular permeability (Evans Blue Dye albumin (EBA) extravasation into the lung). Consistent with these results, delivery of plasmids encoding the wild type CS1β (MLCP catalytic subunit) significantly reduced LPS-induced lung inflammation as determined by the reduction of protein content in bronchoalveolar lavage fluid (BALF). Conversely, the in vivo transfection of a catalytically inactive (D/N) CS1β mutant increases vascular leak. Collectively, these data convincingly demonstrate for the first time that MLCP directly provides EC barrier protection against LPS-induced EC barrier disruption in vitro and LPS-induced vascular leak in vivo. In addition, they support an anti-inflammatory role of MLCP in vitro and in vivo.
MLC is the classical substrate for MLCP. MLCP-mediated MLC dephosphorylation is primarily regulated by MYPT1 binding to myosin 14,15,45 . However, more recent studies indicate that additional cytoskeletal proteins may be targets for MLCP and suggest that MLCP may have a broader multidirectional involvement in the control of EC cytoskeleton and permeability 12,46,47 .
ERM (Ezrin/Radixin/Moesin) proteins are unconventional MLCP cytoskeletal substrates 19,48 ERM proteins act both as linkers between the actin cytoskeleton and plasma membrane, and as signal transducers in responses involving cytoskeletal remodeling [49][50][51] . In the unphosphorylated form, ERM proteins exist in auto-inhibited conformation. Phosphorylation induces and stabilizes the unfolded active conformation of ERMs, allowing them to simultaneously bind to membrane and actin filaments 51,52 . MYPT1 can bind to ERM proteins and facilitates MLCP-dependent ERM dephosphorylation 19 . Importantly, we and others have shown that ERM proteins are involved in EC barrier regulation 20,26,28,53 , however, their role in LPS-induced EC barrier compromise was unknown. Our data demonstrate that LPS-induced ERM phosphorylation occurs in parallel with the reduction in TER. In contrast, LPS-induced MLC phosphorylation is transient with a nadir at 2-4 hrs, which precedes maximal changes in permeability. These data suggest that LPS-induced MLC phosphorylation is involved in the initiation of the contractile response and mediates permeability increases, whereas ERM phosphorylation may be important for the magnitude and/or duration of permeability increases. Our data also demonstrate that LPS-induced increases in ERM phosphorylation are more prominent in CS1β -depleted cells and confirm that ERMs are cytoskeletal targets for MLCP in EC.
ERM proteins are phosphorylated at conserved Thr residues on the C-terminus (Thr 558 for Moesin, Thr567 for Ezrin and Thr564 for Radixin). The established literature indicate that multiple kinases are involved in the phosphorylation of ERMs in different cell types and in response to various stimuli. Initially Rho-kinase and PKC were the only kinases reported to phosphorylate ERM [54][55][56] . Lately, several other kinases have been identified that phosphorylate ERMs in different cell types 57 . Importantly, studies on lymphocytes demonstrate that the kinase(s) responsible for ERM phosphorylation may be cell-specific 57 .
According to a recent publications, ERM phosphorylation in EC is dependent on PKC, p38, and Rho-kinase either alone or in combination 26,53 . We have recently shown that thrombin-induced EC barrier dysfunction is accompanied by PKC-mediated ERM phosphorylation 28 . In addition, we have demonstrated that the microtubule disruptor, 2-methoxyestradiol (2ME), increases ERM phosphorylation in a p38/PKC-dependent fashion. The activation of p38 appears to occur upstream from the activation of PKC 20 . In contrast, TNFα -induced increases in ERM phosphorylation and EC barrier dysfunction likely involves parallel activation of p38 and PKC-mediated signaling 53 . Our data demonstrate that LPS-induced ERM phosphorylation is PKC, but not Rho-kinase dependent. At the same time we have demonstrated that LPS-induced MLC phosphorylation is Rho-kinase, but not PKC-dependent. These data indicate that LPS-induced ERM and MLC phosphorylation require the activation of different signaling pathways and support the hypothesis that MLC and ERM phosphorylation may be involved in different phases of LPS-induced EC permeability responses.
We and others have recently demonstrated that despite structural similarity among the ERM proteins, their individual roles in EC barrier function may vary and the degree of involvement may be agonist-specific 26,28 . These observations highlight the critical need to examine the role of individual ERM proteins in LPS-induced EC barrier disruption. Using short interfering RNA to specifically deplete Ezrin, Moesin and Radixin in EC, we demonstrated that depletion of Moesin and Ezrin significantly attenuates LPS-induced decrease in TER. In contrast, depletion of Radixin, but not Moesin and Ezrin modestly decreases baseline TER of unstimulated EC monolayers and has no significant effect on LPS-induced increases in permeability. The origin of the functional differences between highly homologous ERM proteins are not entirely clear, but most likely arises from differences in the interactions with partnering regulatory proteins. For example, Moesin and Ezrin differentially bind to L-selectin in lymphocytes 58 ; Radixin, but not Ezrin or Moesin bind to the integral membrane hyaluronan receptor, layilin 59 . Further, we have recently shown that ERM proteins have a different binding affinity toward the subunits of MLCP, CS1β and MYPT1. Moesin and to a lesser extent Ezrin, but not Radixin can be detected in CS1β immunoprecipitates. In contrast, Radixin, but not Ezrin or Moesin co-immunoprecipitates with MYPT1 in COS7 cells 18 . These differences in the binding capacities of ERM proteins for MLCP subunits may contribute to the different functional roles of ERM proteins in EC barrier regulation. However, the exact role of differential ERM-MLCP binding both at baseline and following LPS-stimulation remains to be determined.
Using phosphorylation-deficient ERM mutants that were ectopically expressed in EC, we have recently demonstrated a direct role of ERM phosphorylation in the loss of EC barrier integrity induced by the microtubule inhibitor, 2ME, in vitro 20 . To evaluate the role of ERM phosphorylation in LPS-induced vascular barrier dysfunction in vivo we delivered an expression plasmid encoding a phospho-null Moesin mutant into murine lung using a nanoparticle based DNA delivery method 27 . Our data demonstrate that ectopic expression of the phosphorylation-deficient Moesin significantly attenuates LPS-induced lung inflammation and vascular leak supporting a regulatory role of changes in Moesin phosphorylation in mediating changes in vascular barrier permeability in response to LPS. To the best of our knowledge, this is the first report directly showing a role of Moesin phosphorylation/dephosphorylation in the regulation of lung vascular barrier function in vivo.
Collectively, our data demonstrated that MLCP opposes LPS-induced lung injury in vitro and in vivo, at least in part, via dephosphorylation of its cytoskeletal targets, the ERM proteins.
Measurements of transendothelial electrical resistance of EC monolayer. Transendothelial electrical resistance (TER) was measured in response to the EC barrier disruptive agent LPS using an electrical cell substrate impedance sensing system (ECIS; Applied Biophysics, Troy, NY) as we have previously described 18,60 . HPAEC transfected with siRNA specific to Moesin, Ezrin or Radixin were plated on gold microelectrodes. TER was measured 72 h later.
Evans Blue Dye (EBD) albumin extravasation into mouse lung. Adult male mice were anaesthetized using 150 mg/kg ketamine and acepromazine. LPS solution (1 mg/kg body weight) or 0.9% sterile saline was introduced intratracheally via a 20-gauge catheter. Animals were allowed to recover for 18 hours. Bronchoalveolar lavage fluid (BALF) and lungs were collected and stored at − 80 °C. Evans blue-labeled albumin dye (EBD) was injected into the tail vein 120 minutes before the termination of the experiment in order to assess vascular leak.
In vivo DNA delivery into mouse lung endothelium with JetPEI. In vivo gene delivery was performed as we have previously described 27 . Briefly, wild type or D/N CS1β , C/A MYPT1, or phospho-null Moesin mutant in pcDNA 3.1 mammalian expression plasmids (40 μ g) or control empty plasmid pcDNA 3.1 (40 μ g) were incubated with glucose and jetPEI reagents, as per manufactures instruction for 15-30 minutes. Then, the cDNA/ jetPEI complexes were injected into 7-8 week old male mice via the tail vein 27 . At 72 hours post-injection, LPS (1 mg/kg) or sterile saline (0.9%) solution was instilled intratracheally (i.t.) via a 20-gauge catheter. Animals were allowed to recover for 18 hours. After the experiment, lungs were perfused with normal saline through the right ventricle, weighted and snap frozen in liquid nitrogen for further Western blot experiments. Bronchoalveolar lavage fluid (BALF) and lungs were collected and stored at −80 °C. EBD was measured in order to assess vascular leak.
Lung Histology. Lungs were perfused with PBS and immersed in 4% paraformaldehyde overnight. Sections (5 μ m) were cut from paraffin blocks and mounted on slides. Slides were deparafinized in xylene, and run through graded alcohol to distilled water. Slides were pretreated with Target Retrieval Solution (Dako, Carpinteria, CA) using a vegetable steamer and then were rinsed with distilled water. Lung sections were further used for myeloperoxidase staining.
Measurement of Myeloperoxidase (MPO) Activity. MPO assay kit was used for the staining as we have previously described 27 . Briefly, endogenous peroxidases were quenched with 0.3% H 2 O 2 for 5 minutes followed by distilled water for 2 minutes, and then washed with 1× PBS for 5 minutes. Slides were treated with Target Retrieval Solution, then incubated in Power Block Solution, rinsed in distilled water, placed in 1× PBS for 5 minutes then incubated with anti-myeloperoxidase (MPO) antibody for 30 minutes at room temperature. The slides were rinsed twice in PBS then incubated with a secondary peroxidase-labeled polymer conjugated to goat anti-rabbit IgG for 30 minutes, and then finally rinsed again in PBS. Bound antibody was detected with diaminobenzidine (DAB + substrate kit). Hematoxylin was used as counter-stain.

Statistics.
Values are expressed as mean ± SEM of at least 3 independent experiments. Analysis of variance (ANOVA) or Student's t test was used, as appropriate. A P value of < 0.05 was considered statistically significant.