DMEM/F12 (F12), Trypsin-EDTA, Igepal CA-930, Akt-IV, and SB216763 were purchased from Sigma–Aldrich, Inc. (St. Louis, MO, USA). Fetal calf serum (FCS) was acquired from Equitech-Bio, Inc. (Kerrville, TX, USA). A cocktail of sodium penicillin G, streptomycin sulfate, and amphotericin B was purchased from Gibco-BRL (Gaithersburg, MD, USA). Halt-TM Phosphatase inhibitor cocktail was purchased from Thermo Fisher Scientific (Waltham, MA, USA). Protease inhibitor cocktail was acquired from GE Healthcare Biosciences (Little Chalfont, UK). Trizol reagent and EXPRESS One-Step SYBR GreenER Universal Kit were purchased from Invitrogen (Carlsbad, CA, USA). Bovine IL-10 and IL-8 TSZ ELISA kit were purchased from Biotang (Waltham, MA, USA). Plasmid pLKO.1-GSK3β-#1 was a gift from Alex Toker (Addgene plasmid #32497) and PLKO.1 plasmid was a gift from Bob Weinberg (Addgene plasmid # 8453). pCF CREB M1 was a gift from Marc Montminy (Addgene plasmid # 22969). The generation of pCF empty vector was done by the enzyme digestion of pCF CREB M1 with SacI restriction enzyme, and then the purification and ligation of the 5-kb fragment were generated.

Antibodies against Laminin A/C, GAPDH, β-actin, CREB, CBP, and IgG-HRP-coupled antibodies were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). Antibodies against phospho-glycogen synthase (Ser641), phospho-GSK3α (Ser21), phospho-GSK3α (Ser9), phospho-NF-κB p65 (Ser536), phospho-CREB (Ser133), GSK3β, GSK3α, and NF-κB p65 were purchased from Cell Signaling Technology (Boston, MA, USA).

The strain of S. aureus used in this study is an isolate from a clinical case of bovine mastitis and was obtained from the American-Type Culture Collection (27543). Bacteria were cultured overnight in 3 ml of Luria-Bertani (LB) medium at 37°C with continuous agitation. The inoculum for infection assays was prepared by adding 1 ml of this pre-culture to 49 ml of fresh LB medium and grown at 37°C. This inoculum was added to cells until the culture reached the initial-middle log phase at an optical density of 0.3 at 600 nm and washed with F12 medium.

Endothelial cells used in this study were obtained from bovine umbilical veins and immortalized by transfection with an expression vector containing the E6–E7 oncogenes of human papillomavirus 16 (BVE-E6–E7) (31). These BECs were grown and maintained in F12 supplemented with 10% FCS and a cocktail of sodium penicillin G, streptomycin sulfate and amphotericin B, unless otherwise noted. For pretreatment with inhibitors, BECs were pre-incubated with a medium alone or with 1 µM of Akt-IV, 10 µM of SB216763, or 20 µM of Parthenolide before infection with bacteria.

Bovine endothelial cells were grown in six-well cell culture plates with F12 medium supplemented with FCS and antibiotics until 90–95% confluence. Then, cells were washed 2× with PBS and the medium was changed to F12 without serum and antibiotics. Cells were left in these conditions for at least 4 h. Bacteria were added at a multiplicity of infection (MOI) of 20 CFU/cell, and plates were centrifuged at 500 × g for 5 min and incubated in 5% CO₂ at 37°C for the times indicated.

For plasmids transfection containing siRNA against GSK3 isoforms, BECs were grown until 60–70% confluence in F12 medium without serum and antibiotics for 24 h. Then, transfection was performed with Lipofectamine 2000 from Thermo Fisher Scientific (Waltham, MA, USA); for siGSK3α, we added 3 µg of siRNA-GSK3α plasmid (Sigma–Aldrich) and 3 µg of PLKO.1; for siGSK3β, we added 3 µg of pLKO.1-GSK3β-#1 and 3 µg of PLKO.1; for siGSK3αβ, we added 3 µg of siRNA-GSK3α and 3 µg of pLKO.1-GSK3β-#1 plasmid; and for sicontrol, we added 6 µg of PLKO.1. After 24 h, the culture medium was changed to F12 with low serum (1%), and cells were incubated for 5 days. GSK3α and GSK3β gene-silenced cells were screened by Western blotting. For GSK3α, the targeting sequence was 5′-TACATCTGTTCTCGCTACTA-3′ (nucleotides 1535–1554, pLKO.1-GSK3α), and for GSK3β the targeting sequence was 5′-GAAGTCAGCTATACAGACACT-3′ (nucleotides 587–607, pLKO.1-GSK3β). To express the CREB-dominant-negative mutant, cells were cultured at 70–90% confluence in F12 medium without serum or antibiotics for 24 h; then, 350 ng of plasmid pCF-M1-CREB was transfected with Lipofectamine 2000 or 350 ng of pCF empty vector. pCF-M1-CREB plasmid encodes a CREB mutant in which Ser133 was replaced by Ala133. After 12 h, the medium was changed to fresh F12 and low serum, and cells were incubated for an extended period of 48 h. The medium was changed to F12 without serum or antibiotics, and cells were left for at least 4 h before stimulation with S. aureus. Transfection efficiency was monitored by detecting the FLAG tag present on the CREB-mutant construct in Western blot assays. The transfection efficiency of the empty pCF-M1 vector was monitored by the resistance of cells to Geneticin.

To evaluate the relative abundance of phosphorylated and non-phosphorylated proteins, the total protein (cytosolic and nuclear) from control and treated cells was obtained by washing cells 2× with ice-cold PBS and lysing them with 80 µl of cold lysis buffer (20 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1% Igepal CA-930, 10 mM Na-pyrophosphate, 50 mM NaF, and 1 mM Na-orthovanadate) supplemented with 1× protease inhibitor cocktail and 1× phosphatase inhibitor cocktail, which were added immediately before lysis. For subcellular protein fractionation, we followed the REAP method described by Suzuky et al. (32). Lysates were centrifuged at 16,000 × g for 20 min at 4°C, and the supernatant was recovered. Protein concentration was measured by Bradford using BSA as standard. Then, 30–40 µg of protein was separated by electrophoresis in 10% sodium dodecyl sulfate-polyacrylamide gels at 200 V for 1 h and electroblotted in a wet chamber onto 0.45 µm nitrocellulose membrane (Bio-Rad) at 250 mA for 1 h. Membranes were incubated in a blocking solution (5% non-fat milk solution) in TTBS (10 mM Tris–HCl, pH 7.5, 0.9% NaCl, and 0.1% Tween 20); primary antibodies to target proteins were added and membranes incubated at 4°C overnight. Membranes washed 1× with PBS and 2× with TTBS were incubated with secondary antibodies. Proteins were detected with the Immobilon Western Chemiluminescent HRP substrate kit from Millipore (Billerica, MA, USA). Protein loading was verified by stripping the membranes with NaOH of 0.2 N for 5 min and washed 2× with TTBS; after a blocking step, the membranes were re-probed with antibodies against the indicated housekeeping proteins and detected in the same way as described. Data presented in Figures 4 and 9C were obtained by scanning the membranes using a C-DiGit blot scanner (LI-COR, Lincoln, NE, USA).

After infection, BECs were washed 2× with ice-cold PBS, and the total RNA was extracted with Trizol reagent, following the isolation procedure described by the manufacturer. RNA was purified with RQ1 RNase-free DNase (Promega). One-step qRT-PCR was performed using the EXPRESS One-Step SYBR GreenER Universal Kit and the real-time StepOnePlus thermocycler from Applied Biosystems. Each reaction was performed with 100 ng of total RNA under the standard 20-µl reaction provided by Invitrogen. The one-step-cycling program conditions and oligonucleotide primers used were as described by Konnai et al. (2003) (33). Amplification of the expected single products was confirmed by single-peak analysis in temperature-melting curve and visualization on 1% agarose gels stained with ethidium bromide. Relative transcript levels of mRNA were calculated with the ΔΔCt method, using β-actin as the reference gene. Negative amplification of non-added RT enzyme reactions confirmed the absence of DNA in our assays.

Bovine IL-8 and IL-10 proteins in culture supernatants were quantitated for the indicated times by sandwich ELISA, according to the manufacturer’s instructions (Biotang, Waltham, MA, USA).

Co-IP assays were performed with the Pierce Protein A/G Magnetic Beads kit (Thermo Fisher Scientific). Briefly, the control and infected cells (~1 × 10⁷) were washed 2× with ice-cold PBS and lysed in washing/lysis buffer supplemented with 1× protease inhibitor cocktail and 1× phosphatase inhibitor cocktail. Cell debris was collected by centrifugation at 18,000 × g for 10 min at 4°C. The supernatant was recovered and quantitated for protein concentration. Then, 5 µg of CBP IP antibody was added to 500 µg of protein extract; the volume was adjusted to 500 µl with lysis buffer and incubated with continuous agitation overnight at 4°C. Then, 25 µg of protein A/G magnetic beads was added and incubated for 1 h with continuous agitation at room temperature (~23°C); beads were washed 2× with a washing buffer and 1× with water. The target antigen was eluted with 1× lane marker buffer and 50 mM DTT in a final volume of 100 µl. Then, 30 µl of sample in elution buffer was loaded, and proteins were detected by Western blot as described earlier.

The relative abundance of phosphorylated proteins was quantitated by densitometric analysis with the Image Processing and Analysis in Java Program ImageJ (http://rsbweb.nih.gov/ij). To calculate densitometric values, the intensity of the phosphorylated band was divided by the intensity of the non-phosphorylated one. These intensities were referred to a value of 1.0 that was arbitrarily assigned to the untreated control. The statistical significance of triplicate blots, IL-8 and IL-10 mRNA ΔΔCt value, and protein levels were evaluated with the Tukey HSD multiple-comparison test and one-way analysis of variance by using the JMP 6.0 program (SAS Institute). Differences between groups were considered significant at P < 0.05. All data are available on request.

The pro-inflammatory chemokine IL-8 is an early marker of bacterial infection (9). However, several authors have reported that S. aureus can induce or repress the IL-8 expression on different cellular physiological contexts, and this was correlated with the pathology produced by this bacterium (4–7). There is also evidence that S. aureus induces IL-10 expression, leading to immune tolerance (11). Therefore, we asked whether the expression of these cytokines depended upon the stage of infection. In order to evaluate the kinetics of IL-8 and IL-10 expressions, BECs were infected with S. aureus at a MOI of 20, which allows the survival of endothelial cells for 2, 4, and 6 h. Then, RNA was purified to analyze IL-8/IL-10 mRNA by qRT-PCR and their corresponding peptides in cell supernatants by ELISA. The expression of IL-8 mRNA increased by more than 2.5-fold in response to 2 h of S. aureus infection; however, at longer infection times of 4–6 h, IL-8 mRNA values returned to control levels (Figure 1A). Measurements of IL-8 peptide indicated an increase at 2 h post infection with similar values at 4 and 6 h (Figure 1B). In the case of IL-10 mRNA, an increase was observed at 4 h and remained without a statistical change for up to 6 h (Figure 1C) while its peptide reached a maximum at 4 h (Figure 1D). These data indicate that in BEC infected with S. aureus, the mechanism responsible for IL-8 expression precedes the mechanism leading to IL-10 expression.

The transcription factors NF-κB and CREB induce IL-8 and IL-10 expressions, respectively (26, 34). The transcriptional activation of NF-κB, in response to bacterial ligands, involves the phosphorylation of p65 subunit at Ser536 and its translocation to the nucleus. These events usually take place in the first 15–40 min after cell stimulation with different ligands (8, 14, 32, 35, 36). We observed that the phosphorylation of p65 at Ser536 reached a maximum level at 40 min after bacterial challenge followed by a gradual reduction as compared to uninfected control (Figure 2A). Several authors have demonstrated that CREB is transcriptionally activated by phosphorylation at Ser133 in response to various stimuli (28, 37, 38). Accordingly, a significant increase in phospho-CREB at Ser133 at 60 min post infection was detected with a maximum observed at longer infection times of 90–120 min (Figure 2B). Interestingly, these longer infection times that we tested to induce an increase in phospho-CREB-Ser133 were the same as those required for NF-κB to become dephosphorylated and inactivated. These results allowed us to select 40 and 120 min as the infection times to explore the effects of S. aureus infection on NF-κB p65 and CREB, respectively.

The transcription of NF-κB-specific target genes involves its translocation from the cytoplasm to the nucleus. To evaluate NF-κB nuclear translocation, we infected BEC with S. aureus at 40 and 120 min and detected the relative abundance of NF-κB p65 subunit present in the nucleus or cytosol by Western blotting. The NF-κB p65 subunit was detected in the nuclear fraction after 40 min of infection, whereas at 120 min, we could not observe it (Figure 2C). These indicate that NF-κB phosphorylation and nuclear translocation are processes that take place in the initial phase of infection and precede CREB activation.

GSK3α and GSK3β are constitutive active enzymes that regulate the activity of several transcription factors (15, 39). We have previously reported that PI3K/Akt-signaling pathway leads to the phosphorylation of GSK3α at Ser21 and GSK3β at Ser9 in S. aureus internalization by BEC (18). However, the role of each isoform in cytokine regulation was not evaluated. Therefore, we decided to investigate the contribution of GSK3α and GSK3β to the regulation of NF-κB and CREB activation in BEC infected with S. aureus. The time course of BEC infected with S. aureus showed that the phosphorylation of GSK3α (Ser21) and GSK3β (Ser9) increased significantly after 40 min of infection (Figures 3A,B), GSK3α phosphorylation (~8-fold compared with control) being higher than GSK3β (~2.5-fold compared with control). Interestingly, under basal conditions, BEC phosphorylation of GSK3α Ser21 is barely detected. Other authors have reported similar results about GSK3 relative phosphorylation under basal conditions in different cell types (40–43).

Next, we analyzed whether the increase in the relative abundance of phosphorylated GSK3α and GSK3β isoforms correlated with a decrease in the phosphorylation of glycogen synthase at Ser641 in BEC infected with S. aureus. A marked reduction of GSK3α/β kinase activity from 60 to 120 min post infection led to a gradual decrease in the relative abundance of phosphorylated glycogen synthase at Ser641 (Figure 3C). To further explore if GSK3α/β was associated with NF-κB p65 phosphorylation at Ser536, BECs were pretreated with SB to inhibit GSK3α/β activity before infection for 40 and 120 min. The inhibition of GSK3α/β with SB did not reduce the relative abundance of NF-κB p65 phosphorylated at Ser536 (Figure 4A, upper panel). In contrast to NF-κB, CREB phosphorylation was affected by the inhibition of GSK3α/β activity, because in S. aureus-infected cells, SB induced a substantial increase in the relative abundance of phospho-CREB-Ser133 compared to infected cells at 40 min (Figure 4A, middle panel). Interestingly, at 120 min, SB did not induce a greater increase in phospho-CREB-Ser133 compared to infected cells, indicating that the maximal increase of phospho-CREB-Ser133 in SB-pretreated cells peaked at 40 min post infection. Because these observations showed that GSK3 inhibition accelerated the phosphorylation of CREB-Ser133, we tested the relative abundance of p65 and CREB phosphorylation in BEC treated with Akt-IV, an Akt inhibitor. We observed that Akt inhibition did not affect NF-κB p65 phosphorylation level at 40 min of infection; however, at 120 min, a slight increase in phospho-p65 was observed, suggesting that GSK3α/β activity promotes NF-κB phosphorylation at later stages of infection (Figure 4B, upper panel). Akt inhibition completely abolished CREB phosphorylation, indicating that the activated state of GSK3α/β inhibits phospho-CREB-Ser133 increase (Figure 4B, middle panel). Finally, the nuclear accumulation of NF-κB p65 was not affected by GSK3α/β inhibition with SB (Figure 4C).

To confirm that GSK3α/β activity interferes with CREB phosphorylation at Ser133, the genes of each GSK3 isoform were siRNA-silenced for 120 h. After this period, we infected BEC with S. aureus for 40 min, a time in which GSK3α/β is still active and CREB phosphorylation was barely higher than the uninfected control (Figure 2B). GSK3α/β gene silencing was about 60–70% for GSK3α and 50–60% for GSK3β (Figure 5A), demonstrating the efficiency of the siRNA interference technique. Under these conditions, phospho-CREB-Ser133 showed a marked increase in GSK3α/β-silenced cells, compared to non-transfected or control vector cells (Figure 5B). These data demonstrate that (1) NF-κB p65 phosphorylation at Ser536 and its nuclear translocation are not directly affected by GSK3α/β activity, (2) GSK3α/β activity directly affects CREB phosphorylation at Ser133, and (3) Akt plays a major role in CREB phosphorylation.

A critical step in the transcriptional regulation mediated by NF-κB or CREB is the interaction of each of these transcription factors with the co-activator CBP. NF-κB and CREB compete for the limiting amounts of CBP to form a complex in the nucleus (14, 44). To test whether GSK3α/β activity was necessary for NF-κB-CBP or CREB-CBP complex formation in BEC infected with S. aureus, we carried out co-IP assays. When BECs were pre-incubated with SB and infected with S. aureus for 40 min, IP of CBP showed a marked reduction of the NF-κB-CBP interaction and a strong increase of the CREB-CBP complex (Figure 6A). Accordingly, the inhibition of Akt with Akt-IV (GSK3α/β remains active under this condition) favored NF-κB, but not CREB, interaction with CBP. The inhibition of GSK3α/β with SB at 40 min or control without SB at 120 min led to an increase in CREB-CBP interaction (Figure 6B). Moreover, at 120 min post infection, the Akt inhibition favored the interaction of NF-κB with CBP (Figure 6B). These data suggest that an initial stage of infection GSK3α/β activity indirectly favors NF-κB-CBP complex formation but not CREB-CBP interaction, whereas at later stages of infection, when GSK3α/β loses its kinase activity, the opposite occurs.

To identify which GSK3 isoform was predominantly affecting CREB-CBP or NF-κB-CBP complex formation, we performed an siRNA-silencing assay to independently reduce the expression of each isoform (Figure 6C). After infection of BEC with S. aureus for 40 min, we immunoprecipitated CBP, and CREB or NF-κB was immunodetected in the precipitated fraction. In S. aureus-infected cells, GSK3α was the primary isoform that affected CREB-CBP interaction, because in either GSK3α or GSK3α/β-silenced cells, a stronger interaction between CREB and CBP was observed compared to the amount of CREB-CBP complex seen in GSK3β-silenced cells (Figure 6D). This preferential association found between CREB and CBP in GSK3α-silenced cells repressed NF-κB-CBP interaction because of the reduced amount of NF-κB-CBP complex in these cells compared to the infected controls or the GSK3β-silenced cells.

To evaluate the role of NF-κB on IL-8 and IL-10 expressions, we incubated BEC with 20 µM Parthenolide, an NF-κB inhibitor, for 1 h before infection. Under these conditions, a reduced IL-8 expression was observed at 2 h post infection (Figure 7A), whereas an increased IL-10 expression was detected at 6 h post infection (Figure 7B), indicating that NF-κB activity causes a slight inhibition of IL-10 expression in BEC infected with S. aureus. Then, we evaluated the role of CREB on IL-8 and IL-10 expressions by transfecting BEC with a plasmid containing a CREB-Ser133Ala-dominant-negative mutant (DN-CREB-Ser133Ala) (Figure S1 in Supplementary Material). At 2 h post infection, the DN-CREB-Ser133Ala-expressing cells showed an increased expression of IL-8 (Figure 7C). By contrast, at 6 h post infection, the DN-CREB-Ser133Ala-expressing cells showed a significant reduction of IL-10 expression (Figure 7D). These results confirm that NF-κB and CREB compete for binding CBP and that they are essential for IL-8 and IL-10 expressions in BEC infected with S. aureus.

To investigate whether GSK3α/β activity influences NF-κB and CREB transcriptional activity and this, in turn, regulates IL-8 and IL-10 expressions, BECs were pre-incubated with SB or Akt-IV for 1 h. After infection with S. aureus for 6 h, the IL-8 and IL-10 peptides were quantitated by ELISA. We observed a reduction in IL-8 peptide compared to the infected control cells in the presence of SB, whereas in BEC treated with Akt-IV, a condition in which GSK3α/β isoforms are active, the IL-8 expression was higher after 6 h as compared to SB treated or infected control cells (Figure 8A). The expression of IL-10 increased when BECs were incubated with SB at 6 h post infection, whereas in the presence of Akt-IV, no change in IL-10 peptide compared to control was observed, indicating that active Akt (inactive GSK3α/β) was required to induce IL-10 synthesis (Figure 8B). These data suggest that GSK3α/β activity promotes IL-8 peptide synthesis during S. aureus infection and that the loss of GSK3α/β activity leads to a reduction of IL-8 and an increase in IL-10 expression.

We have presented evidence that in BEC infected with S. aureus, (1) GSK3α is highly phosphorylated, (2) GSK3α promotes CREB-CBP interaction, which is essential for IL-10 expression, and (3) GSK3α represses NF-κB-CBP interaction, which is critically indispensable for IL-8 expression. To confirm that GSK3α is the main isoform regulating these processes, we performed an siRNA-gene-silencing assay to reduce the expression of each GSK3 isoform in BEC infected with S. aureus for 6 h followed by the detection of IL-8 and IL-10 peptides by ELISA. We observed that GSK3α was the main isoform that regulates IL-8 and IL-10 expressions because GSK3α or GSK3α/β-silenced cells showed a stronger reduction of IL-8 expression and an increased IL-10 expression compared to GSK3β-silenced or control-infected cells (Figures 9A,B). Evidence of GSK3α and GSK3β gene-silencing efficiency is shown in Figure 9C. These data demonstrate that GSK3α is the primary isoform that regulates the expression of IL-8 and IL-10 in BEC infected with S. aureus.