We recently showed that Etk/Bmx, a member of the Tec family of nonreceptor protein tyrosine kinases, promotes tight junction formation during chronic hypoxic exposure and augments normoxic VEGF expression via a feedforward mechanism. Here we further characterized Etk's role in potentiating hypoxia-induced gene expression in salivary epithelial Pa-4 cells. Using transient transfection in conditionally activated Etk (ΔEtk:ER) cells, we demonstrated that Etk enhances hypoxia-response element-dependent reporter activation in normoxia and hypoxia. This Etk-driven reporter activation is ameliorated by treatment with wortmannin or LFM-A13. Using lentivirus-mediated gene delivery and small interfering RNA, we provided direct evidence that hypoxia leads to transient Etk and Akt activation and hypoxia-mediated Akt activation is Etk dependent. Northern blot analyses confirmed that Etk activation led to induction of steady-state mRNA levels of endogenous VEGF and plasminogen activator inhibitor (PAI)-1, a hallmark of hypoxia-mediated gene regulation. We also demonstrated that Etk utilizes a phosphatidylinositol 3-kinase/Akt pathway to promote reporter activation driven by NF-κB, another oxygen-sensitive transcription factor, and to augment cytokine-induced inducible nitric oxide synthase expression in endothelial cells. To establish the clinical relevance of Etk-induced, hypoxia-mediated gene regulation, we examined Etk expression in keloid, which has elevated VEGF and PAI-1. We found that Etk is overexpressed in keloid (but not normal skin) tissues. The differential steady-state Etk protein levels were further confirmed in primary fibroblast cultures derived from these tissues, suggesting an Etk role in tissue fibrosis. Our results provide further understanding of Etk function within multiple signaling cascades to govern adaptive cytoprotection against extracellular stress in different cell systems, salivary epithelial cells, brain endothelial cells, and dermal fibroblasts.
- inducible nitric oxide synthase
- nuclear factor-κB
cellular adaptation to diminished tissue O2 delivery (hypoxia) involves specific changes in gene expression profiles. Understanding the molecular mechanisms underlying the (patho)physiological responses to hypoxia may provide novel therapeutic approaches for treating diseases such as myocardial and cerebral ischemia, retinal disease, and cancer. Angiogenesis and neovascularization occur as a result of hypoxia-inducible factor 1 (HIF-1), a transcription factor that plays the role of master regulator of O2 homeostasis. The genes that are targeted by HIF-1 mediate adaptive responses to hypoxia and are involved in cellular processes that include vascular remodeling and angiogenesis and cell proliferation and survival. In the present study, we investigated the role of Etk/Bmx, a member of the Tec family of nonreceptor protein tyrosine kinases (PTKs), in regulating the expression of two genes, namely, VEGF and plasminogen activator inhibitor (PAI)-1. The expression of both PAI-1 and VEGF is stimulated by hypoxia, playing critical roles in wound healing and tumor progression.
Dysregulation of several proteolytic enzyme systems, such as tissue-type and urokinase-type plasminogen activators (tPA and uPA), has been associated with many diseases including keloids. Keloids, an aberration of cutaneous wound healing, behave clinically like a benign tumor that grows beyond the boundaries of original wound margins without evidence of spontaneous regression (reviewed in Ref. 31). The activities of tPA and uPA are regulated, in part, by the steady-state levels of PAIs. There is clinical and experimental evidence that the hypoxia-induced accumulation of PAI-1 messages and/or proteins is involved in governing fibrinolysis and cellular invasion (25, 37). However, excessive production of PAI-1 has been associated with both aberrant fibrin deposition in the development of thrombosis and a poor prognosis in the case of cancer patients (2, 18). Although the role of PAI-1 in many (patho)physiological processes has been studied extensively, the molecular mechanisms governing its expression are not fully understood.
Numerous studies have suggested that a hypoxic microenvironment develops in the early stages of wound healing (1) or in intratumor settings (9) where fibroblasts, endothelial cells, and cancer cells are stimulated by hypoxia to produce higher levels of growth factors, collagens, and VEGF (12, 19, 41, 45). These events eventually lead to neovascularization and extracellular matrix remodeling, which are largely contributed by the deregulated expression of VEGF and PAI-1. Our laboratory recently showed (15) that Etk/BMX elicits a protective effect by sustaining epithelial barrier function against chronic hypoxic exposure in epithelial cells. A plethora of studies have also documented the role of both cytoplasmic and receptor PTKs in mediating the hypoxia-induced signal transduction activation pathway (7, 11, 29, 48). Because phosphatidylinositol 3-kinase (PI3-kinase) has been shown to be indispensable for hypoxia-mediated induction of VEGF and PAI-1 (reviewed in Ref. 43) and PI3-kinase is also activated by Etk (10, 59), it is plausible that Etk is involved in hypoxia-mediated gene regulation. Additionally, because we have reported (58) that the activation of PTKs is involved in the signaling pathways leading to the hypoxic induction of PAI-1 in keloid fibroblasts, we hypothesized that Etk may also be involved in mediating the HIF-1-dependent induction of VEGF and PAI-1.
NF-κB has recently been considered as a target for antiangiogenic therapy because it is an oxygen-sensitive transcription factor and contributes to altered signaling events induced by hypoxia (3, 6). Other stressors such as proinflammatory cytokines (e.g., TNF-α and IL-1) have been demonstrated to mediate the activation of multiple signal transduction cascades that result in stimulating NF-κB activation (6). These pathways culminate with activation of IKK through the action of unidentified upstream components. The classic NF-κB activation pathway involves serine/threonine phosphorylation by IKK, followed by subsequent ubiquitination and proteolytic degradation by the 26S proteasome of the inhibitory IκB protein. The degradation of IκB enables NF-κB to translocate from the cytoplasm to the nucleus, where it binds to specific elements in the promoter regions of target genes to activate their expression (54). This includes genes that either govern immune and inflammatory responses, such as the inducible form of nitric oxide synthase (iNOS), or promote proliferation and apoptosis, such as cellular inhibitors of apoptosis (c-IAP)1 and -2 and TNF receptor (TNFR)-associated factors (TRAF)1 and -2, respectively.
Recently, Pan et al. (34) showed Etk to function as a TNFR2-specific kinase, playing a role in TNF-induced angiogenic events. Petro et al. (35) and Bajpai et al. (5) independently reported that Btk, an Etk-related member of the Tec family of kinases, is involved in signaling pathways leading to NF-κB activation. Importantly, Btk was proposed to be an upstream kinase of IKK. Moreover, studies have also demonstrated the involvement of the PI3-kinase/Akt pathway in mediating NF-κB activation, also lying upstream of IKK (8, 32). Akt can then phosphorylate and activate IKK (39), resulting in the subsequent activation of NF-κB that leads to cell survival and driving tumor progression. Because we showed previously (10) that Etk is upstream of PI3-kinase, it is conceivable that Etk may utilize a similar PI3-kinase-/Akt-dependent mechanism to activate the NF-κB signaling cascade.
Despite recent reports of the involvement of the Etk signaling network in the regulation of actin reorganization, cell migration, and tumor progression (4, 10, 15, 34), the role of Etk in activating downstream signaling cascades induced by stress conditions, such as hypoxia and inflammatory cytokines, remains to be established. We sought to utilize our inducible ΔEtk:ER system (10, 52) to establish the role of Etk activation in regulating VEGF and PAI-1 expression in hypoxia and to probe its cross talk with activation of the NF-κB pathway. Keloids represent a significant derangement in the wound healing process characterized by an excessive accumulation of extracellular matrix. Recent studies have shown that keloids constitutively express a higher level of PAI-1 and VEGF than their peripheral normal skin partners (14, 53, 57). Therefore, to establish the clinical relevance of Etk-regulated, hypoxia-mediated gene regulation, we examined the Etk expression in keloid and found that Etk is overexpressed in keloid tissues, and their derived primary cultures of fibroblasts, but not in normal skin tissues. These results indicate that VEGF and PAI-1 are downstream targets of Etk activation and that the overexpression of Etk may be involved in the pathogenesis of keloid. We propose that Etk plays a unique role in programming adaptive cytoprotection in different cell systems, salivary epithelial cells, brain endothelial cells, and dermal fibroblasts.
MATERIALS AND METHODS
Cell culture and constructs.
The rat parotid epithelial cell lines Pa-4 and Pa-4ΔEtk:ER were cultured as described previously (10, 26, 52, 55). Conditionally immortalized rat microvascular blood-brain barrier (TR-BBB) and brain-retinal barrier (TR-iBRB) endothelial cells (16) as well as TR-BBBΔEtk:ER (10) were maintained in a humidified atmosphere of 5% CO2-95% air at 33°C and cultured as previously described (10). For all experiments, phenol red-free medium and charcoal-adsorbed FBS were used to avoid constitutive activation of the estrogen receptor (ER) ligand binding domain by the weak ER agonist phenol red or estrogens present in untreated FBS. Primary cultures of fibroblasts were isolated from biopsied tissues obtained from patients undergoing excisional biopsies at the Dermatology Clinic at King Drew Medical Center (Los Angeles, CA). We collected five sets of tissues, keloid and normal skin matched by age, race, gender, and anatomic site, as previously described (47), following the established protocol approved by the Institutional Review Board of the Charles R. Drew University of Medicine and Science. Fibroblast cells derived from normal skin and keloid tissues were maintained in DMEM (GIBCO, Rockville, MD) supplemented with 10% FBS. All cultures were maintained at 37°C, 5% CO2, and 20% O2. Cells from passages 1–8 were used for experiments and routinely monitored for cell proliferation, morphology, and phenotype.
The NF-κB reporter construct was a kind gift from Dr. Ebrahim Zandi (University of Southern California). The HIF-1α expression plasmid and P11W and P11M reporter constructs, harboring the wild-type and mutated hypoxia-response element (HRE)-containing 47-bp DNA fragment located at 985–939 bp upstream of the human VEGF transcription initiation site (13), respectively, were kindly provided by Dr. Gregg Semenza (Johns Hopkins University), and the HIF-1β expression construct was a gift from Dr. Oliver Hankinson (University of California, Los Angeles).
Transient transfections and luciferase assays.
Pa-4ΔEtk:ER or TR-BBBΔEtk:ER cells were transiently transfected with 0.5 μg of P11W or P11M. One-tenth microgram of the Renilla luciferase pRL-TK plasmid was cotransfected as an indicator for normalization of transfection efficiency. Six hours after the start of transfection, cells were recovered overnight in 0.05% serum-stripped medium and subsequently treated with vehicle (−Etk) or 1 μM β-estradiol (E2) to activate ΔEtk:ER (+Etk). Cells were then treated for 16 h before cell lysates were harvested for luciferase assays. The relative luciferase activity from the firefly luciferase reporter gene was determined and normalized with Renilla luciferase activity with the Dual Luciferase Reporter Assay System (Promega). All fold induction data reported were calculated after normalizing the reporter activities with the activities from the indicator plasmid. In general, there was no obvious change in the basal activity in our cotransfection and inhibitor studies. Sometimes, a particular treatment may elicit a modulating effect on the general transcriptional machinery; however, these general effects were taken into account by normalizing with the activities of the indicator plasmid.
Whole cell lysates of TR-BBBΔEtk:ER cells, cultured fibroblasts, or keloid tissue samples were subjected to SDS-PAGE followed by immunoblotting with antibodies for iNOS (Transduction Labs) or Etk (Cell Signaling). Blots were visualized with an enhanced chemiluminescence detection kit (ECL Plus, Amersham Pharmacia Biotech), following the manufacturer's instructions, and the Versadoc 5000 Imaging System (Bio-Rad, Hercules, CA). Quantitative data were obtained and analyzed with Quantity One software.
Pa-4 and Pa-4ΔEtk:ER cells were cultured in 0.05% serum-stripped medium and exposed to either normoxic (20% O2) or hypoxic (1% O2) conditions for 4 h before total RNA isolation. Total RNA was extracted with the RNA-Stat reagent (Tel-Test, Friendswood, TX). Equal amounts of RNA from various samples were fractionated on a 1.5% agarose gel in the presence of 2.2 M formaldehyde and then stained with ethidium bromide to visualize the 28S and 18S ribosomal RNAs in order to analyze the quality and quantity of the RNA. Equal amounts of RNA were resolved through a denaturing agarose gel, transferred to nylon membranes (ICN Biomedicals), and UV cross-linked. The prehybridization and hybridization steps were carried out with QuickHyb solution (Strategene), following the manufacturer's recommendations. RNA blots were prepared and hybridization was carried out with 32P-labeled VEGF probe prepared from an isolated rat VEGF cDNA fragment (bases 1–560) or 32P-labeled PAI-1 probes (58). All blots, after stripping, were reprobed with 32P-labeled rat β-actin cDNA probe to ensure that the quality and quantity of RNA between lanes were comparable. Blots were exposed to films at −80°C overnight.
Small interfering RNA and lentivirus-mediated transduction.
The lentiviral small interfering (si)RNA against Etk (si-Etk) construct harbors the oligonucleotides of 5′-TGGAGCTGGGAAGTGGCCAGTTCAAGAGACTGGCCACTTCCCAGCTCCTTTTTTC-3′ and 5′-TCGAGAAAAAAGGAGCTGGGAAGTGGCCAGTCTCTTGAACTGGCCACTTCCCAGCTCCA-3′, downstream of U6 promoter, in the Lenti-Lox 3.7 (pLL3.7) vector as described in Ref. 21. As a negative control, the same vector was designed to encode the following sequences: 5′-TGATCTCTCGGTTCTATCACTTCAAGAGAGTGATAGAACCGAGAGATCTTTTTTGGATCC-3′ and5′-TCGAGGATCCAAAAAAGATCTCTCGGTTCTATCACTCTCTTGAAGTGATAGAACCGAGAGATCA-3′. The lentivirus, recombinant lentiviral vector, and packaging constructs were made as previously described (40). Briefly, human 293T cells (80–90% confluence) in T175 culture flasks were cotransfected by calcium phosphate precipitation with a total amount of 106 μg of DNA distributed in equal molar quantities of pLL3.7-si-Etk or empty vector pCMV-delta-8.7 [encoding human immunodeficiency virus (HIV) Gag-pol driven by cytomegalovirus (CMV) promoter] and pVSV-G [encoding vesicular stomatitis virus G (VSV-G) for pseudotyping] for viral packaging. Virus-containing supernatant medium from the transfected cells was harvested, concentrated through Centricon Plus-20 with Ultracel PL Membranes (Millipore) with a molecular weight cutoff of 30,000 and stored at −80°C. Titers of viral stocks were determined in 293T cells by serial dilutions and were in the range of 1–4 × 108 transduction units/ml. For lentiviral infection, Pa-4-Etk cells were plated 1 day before infection. When the culture reached 25% confluence, the culture medium was aspirated off, and fresh medium containing concentrated virus was added and incubated for 24 h in the presence of polybrene (9 μg/ml) at a multiplicity of infection of 20 to achieve >90% transduction efficiency as indicated by green fluorescent protein (GFP) expression monitored by fluorescence microscopy.
For the expression of Etk, RT-PCR analyses using RNA prepared from rat parotid epithelial Pa-4ΔEtk:ER cells and human dermal tissues, normal skin and keloid, were performed as described previously (51). Briefly, two degenerate primer pairs and one reverse transcription primer were selected from the most conserved regions that span the exon-intron junctions of the human Etk gene (51). P5 (5′-TGCCAGCAGCTGTACATGATCTGGTA-3′) was the primer for reverse transcription. DP1 (5′-TTCTTTCAGGARGCBCAGACTATGATGAA-3′) and DP4 (5′-TCCCACATYARGATHCCAAAWGCCCA-3′) were primer pairs for first-round PCR after reverse transcription. The second round of PCR was performed with nested primer pairs: DP1-DP3 (5′-TGGAGCWGACCACTTVACTGGAAACTT-3′) and DP4-DP2 (5′-ATGTGYTACGATGTSTGTGARGGCATG-3′), respectively. In the sequence of degenerate primers, R stands for nucleotides G/A, B for C/T/G, W for T/A, Y for C/T, H for C/T/A, S for C/G, and V for G/A/C.
Experiments were carried out in duplicate at least three times. One representative data set from these three independent experiments is presented where appropriate. The reporter activity shown is the mean ± SD based on at least three independent transfection experiments. Error bars represent SD. We used Student's t-tests, and P < 0.05 was considered significant in all the reporter gene assays presented.
Role of Etk in potentiating HRE-dependent VEGF expression.
To demonstrate the role of Etk in mediating HRE-dependent gene activation, reporter gene assays were performed in Etk conditionally inducible Pa4ΔEtk:ER cells. The reporter constructs used in our studies contained either wild-type (P11W) or mutant (P11M) sequences ligated to a heterologous SV40 promoter-luciferase expression construct. P11M contains a 3-bp substitution (AAA) of the HRE (5′-RCGTG-3′), thereby eliminating the ability for HIF-1 binding and subsequent transactivation in response to hypoxia (Fig. 1A, lane 5; Ref. 49). Pa4ΔEtk:ER cells were transiently cotransfected with either P11W or P11M in the presence or absence of HIF-1 expression constructs. The Renilla luciferase pRL-TK plasmid was cotransfected to normalize for transfection efficiency. Twenty-four hours after transfection, cells were cultured in 0.05% serum-stripped medium containing 1 μM E2 to activate Etk and then exposed to 1% O2 for an additional 24 h. Thereafter, cell lysates were collected for luciferase assays. As shown in Fig. 1A, Etk activation was able to modulate HRE-dependent VEGF expression in the normoxic condition (lane 2 vs. lane 1). After hypoxic exposure, Etk activation was found to further enhance the reporter activity to 15-fold (Fig. 1A, lane 2 vs. lane 1). In the presence of exogenous HIF-1α/-1β, the observed Etk-mediated P11W activation was less robust (Fig. 1A, lane 4 vs. lane 3). The data presented here illustrated that Etk can be a component of the hypoxia-mediated transcriptional activation and can synergistically enhance HIF-1-dependent gene expression. To our knowledge, this is the first demonstration that Etk can promote hypoxia-mediated HRE activation.
Given that PI3-kinase has been demonstrated to be a part of the hypoxia-mediated signaling pathway in augmenting VEGF gene expression (20), and because PI3-kinase also serves as a downstream effector of Etk (10, 59), we next determined whether the hypoxic induction of VEGF is indeed PI3-kinase- and Etk dependent. Reporter gene assays were performed as described above with the additional treatment of the kinase inhibitors LFM-A13 and wortmannin. LFM-A13 has been reported to be a selective inhibitor of Etk-related Btk and has been used to inhibit Etk activation (10). As shown in Fig. 1B, treatment of cells with either LFM-A13 or wortmannin reduced Etk- and HIF-1-driven reporter activity under normoxic as well as hypoxic conditions, suggesting that Etk and PI3-kinase indeed functionally converge within the same signaling cascade to mediate the HRE-dependent response.
Because we have already shown that Etk is involved in driving the hypoxia-induced transcriptional activation of VEGF, we next determined whether this event resulted in an enhancement in the steady-state level of VEGF mRNA. Northern analyses were performed to analyze the changes in endogenous VEGF expression level between Pa-4 and Pa-4ΔEtk:ER cells on Etk activation under both normoxic and hypoxic conditions. Cells were treated for 24 h in the presence or absence of 1 μM E2 to activate ΔEtk:ER before total RNA extraction. Blots were hybridized to 32P-labeled VEGF cDNA probes. All blots were reprobed for rat actin to ensure that the quality and quantity of mRNA between lanes were comparable. As shown in Fig. 1C, basal expression levels of VEGF mRNA were greater in Pa-4ΔEtk:ER than Pa-4 cells. The higher basal level of VEGF mRNA in the untreated Pa-4ΔEtk:ER cells was reproducibly observed. We suspect that this may have resulted from the residual E2 in our culture system. Notably, exposure of cells to 1% O2 further enhanced the VEGF steady-state level (Fig. 1D). These results support the conclusion that Etk may be involved in the signaling pathway mediating the hypoxia-induced response.
Etk activation is required and sufficient to induce PAI-1 expression in Pa-4ΔEtk:ER cells under both normoxia and hypoxia.
We demonstrated previously (58) that the activation of PTKs is required for hypoxia-dependent signaling events, such as ERK1/2 and PI3-kinase activation, leading to the induction of PAI-1. Thus we postulated that Etk might be the key PTK responsible for the hypoxia-mediated upregulation of PAI-1. To investigate this possibility, we determined whether Etk is involved in the hypoxia-induced stimulation of PAI-1. Northern blot analyses of RNA extracted from epithelial Pa-4 and Pa-4ΔEtk:ER cells were performed to determine the steady-state levels of PAI-1 mRNA. As shown in Fig. 2A, the basal level of PAI-1 mRNA in the parental Pa-4 cells (lane 1) was barely detectable under normoxia. Notably, activation of Etk increased PAI-1 expression in normoxia relative to the parental Pa-4 cells (Fig. 2A, lane 1 vs. lane 3). After 4 h of hypoxic exposure, Etk activation further induced the steady-state PAI-1 mRNA to a much higher level compared with the lack of induction in Pa-4 cells (Fig. 2A, lane 2 vs. lane 4). Importantly, the lack of hypoxia-mediated induction of PAI-1 in Pa-4 cells supports the proposed role for Etk activation in modulating hypoxia-mediated gene regulation.
To evaluate the role of PI3-kinase in mediating the Etk-dependent signaling pathway of PAI-1, Northern blot analyses were performed with RNAs prepared from cells treated as described above in the presence and absence of 100 nM wortmannin, a PI3-kinase inhibitor. As expected, wortmannin treatment resulted in a marked attenuation of hypoxia-induced PAI-1 induction in the Etk-activated cells (Fig. 2B, lane 4 vs. lane 3). However, the observed hypoxia-mediated induction of PAI-1 expression cannot be recapitulated by transfecting with either pGL3PAI-800 or pGL3PAI-4000 into Pa-4ΔEtk:ER cells under either normoxia or hypoxia (data not shown). Together these data suggested that Etk activation is required for the observed accumulation of PAI-1 mRNA in a transcription-independent manner or via an enhancer(s) that is not located in the 5′-flanking region of the PAI-1 gene.
Etk mediates activation of NF-κB.
NF-κB is an oxygen-sensitive transcription factor that can also elicit an adaptive cytoprotective response triggered by cytokine stimulation. Because Etk promotes cell survival and its family member Btk was found to stimulate NF-κB activity (5, 35), we next examined the effect of Etk on NF-κB activation in rat microvascular blood-brain barrier (TR-BBB) and blood-retinal barrier (TR-iBRB) endothelial cells through transient transfections and reporter assays using the NF-κB reporter plasmid (Igk)3TK-luc. The endothelial cell lines TR-BBB, TR-iBRB, and TR-BBBΔEtk:ER were cotransfected with the reporter construct and internal control plasmid (pRL-TK) to normalize for transfection efficiency. Cells were stimulated with a combination of the proinflammatory cytokines TNF-α and IFN-γ. As depicted in Fig. 3A, the results indicated that, compared with either parental cell line (TR-iBRB or TR-BBB), stably Etk-transfected TR-BBBΔEtk:ER cells were more responsive in a dose-dependent manner to TNF-α (but not IFN-γ) stimulation in mediating (Igk)3TK-luc reporter activation. To further confirm that Etk indeed is involved in the activation process, cells were cotransfected with either a kinase-dead Etk expression construct [Etk(kq)] or the dominant-negative Etk(DN) to inhibit the Etk-mediated activation of the NF-κB reporter gene. Etk(kq) harbors a point mutation (K445Q) in its ATP binding pocket, resulting in the loss of its kinase activity (22). A substitution of glutamate residue 42 to lysine (E42K) in the PH domain of Etk gives rise to the dominant-negative expression construct Etk(DN) (38). Both Etk(kq) and Etk(DN) efficiently attenuated the Etk-dependent activation of NF-κB in a dose-dependent manner (Fig. 3B).
We then determined whether the Etk-mediated activation of NF-κB involves the PI3-kinase/Akt pathway, because PI3-kinase is one of the downstream effectors of Etk (10) and Akt activation is known to activate IKK, leading to NF-κB activation and enhancement in cell survival (39). Rat TR-BBB cells were cotransfected with expression constructs harboring constitutively active forms of Akt (myr-Akt) or Etk (myr-Etk), followed by stimulation with TNF-α. Cell lysates were collected and analyzed for luciferase activities. The results revealed that Etk, coupled with cytokine stimulation in the presence of Akt, elicited the greatest fold induction of NF-κB reporter activity in a dose-dependent manner (Fig. 3C). Thus Etk and Akt elicit a synergistic effect on NF-κB. To ensure that Akt indeed mediates the Etk-dependent activation of NF-κB, TR-BBB cells were then cotransfected with constitutively active myr-Etk and expression constructs of dominant-negative Etk and Akt. As shown in Fig. 3D, Akt(DN) resulted in the attenuation of the Etk-stimulated reporter activity, suggesting that Akt is indeed a component of the Etk-dependent signaling pathway mediating the activation of NF-κB. Together, these results support a role for Etk in contributing to the activation of the NF-κB pathway.
Effect of Etk on NF-κB downstream target gene iNOS.
NO is endogenously produced by the NOS family, which includes three distinct isoforms. These are neuronal NOS (nNOS or NOS1), iNOS (or NOS2), and endothelial NOS (eNOS or NOS3). The iNOS gene is the only one of its family that is not constitutively expressed and is activated only on stimulation by inflammatory cytokines. Because we have clearly demonstrated that Etk activates NF-κB activity and the iNOS enzyme is one of the downstream target genes of this transcription factor, the following experiment was conducted to determine whether Etk activation of NF-κB results in driving the expression of iNOS. To analyze the inducibility of iNOS expression by Etk activation, TR-BBB and TR-BBBΔEtk:ER cells were stimulated with 25 ng/ml TNF-α for 24 h before protein harvesting for Western blot analysis. Whole cell lysates from unstimulated and stimulated cells were fractionated on an SDS-polyacrylamide gel and immunoblotted with an anti-iNOS antibody while being probed with an anti-actin antibody for equal loading. As shown in Fig. 3E, stably Etk-transfected cell lines exhibit higher basal expression levels of iNOS (lane 3 vs. lane 1) and greater inducibility of the protein level on cytokine stimulation compared with the parental TR-BBB cell line (lane 4 vs. lane 2). This is consistent with the report that Etk-related Btk is required for iNOS induction by LPS and IFN-γ (30). The fold induction of NO measured from the medium of TR-BBBΔEtk:ER cells treated with TNF-α/IFN-γ was reproducibly enhanced over that observed for the unstimulated TR-BBB cells (Urano K, Chau CH, and Ann DK, unpublished observation). The enhanced expression of iNOS in TNF-α-treated TR-BBBΔEtk:ER cells supports the notion that Etk indeed is involved in stimulating NF-κB activity and upregulating one of its subsequent downstream target genes.
Hypoxia activates Etk and Akt.
To test whether Etk activation is required for hypoxia-mediated signaling, Pa-4/Etk cells were treated with 1% O2 for 0, 1, 3, and 6 h. As a positive control, we transfected Pa-4/Etk cells with protein tyrosine phosphatase D1, which was shown previously to induce Etk activation (22). The whole cell lysates were immunoprecipitated with an anti-T7 antibody because Etk is tagged with T7 and subsequently probed with either anti-Etk or anti-phosphotyrosine antibody (4G10). It was apparent that tyrosine phosphorylation of Etk was transiently induced at 1 h after hypoxic treatment and decreased afterward (Fig. 4B, lane 2). As the tyrosine phosphorylation of Etk is very often correlated with its activation (21, 22, 52), we concluded that hypoxia treatment is able to transiently induce Etk activation.
Given that Akt is implicated in hypoxia-mediated signaling to exert its transcriptional activation of VEGF and PAI-1, it is possible that both Etk and Akt play a key role in the cellular response to hypoxia. We next set out to investigate the effect of Etk on hypoxia-mediated Akt activation. As shown in Fig. 4C, whole cell extracts were immunoblotted with an antibody specific for S473-phosphorylated Akt. Hypoxia induced an increase in phospho-Akt immunoreactivity at 6 h after exposure to hypoxia (Fig. 4C, top, lane 3 vs. lane 1). Intriguingly, the extent of basal and hypoxia-induced Akt activation was substantially attenuated in Pa-4/Etk/si-Etk cells (Fig. 4C, second panel, lanes 4 and 6 vs. lanes 1 and 3). The effect was not due to changes in the total amount of Akt protein in the Pa-4/Etk/si-Etk cells, as shown in Fig. 4C (third panel), suggesting that the lack of Etk activation, resulting from si-Etk, conveys the inability of cells to activate Akt on hypoxia treatment. The effect of si-Etk on the steady-state level of Etk was ascertained in Fig. 4A (lane 3 vs. lane 1). We concluded that hypoxia-mediated Etk activation, at least in part, accounts for the observed Akt activation.
Detection of Etk expression in keloid.
To provide a clinical correlation that Etk-dependent pathways are activated in response to stress cues such as hypoxia or proinflammatory cytokines, we extended our investigation of Etk expression in tissues of known chronic hypoxic stress. Keloid tissues have been suggested to be a useful model to investigate the mechanism underlying adaptive hypoxic response because of their state of local hypoxia as histologically described by Kischer (24) or as evidenced by an accumulation of HIF-1α in both keloid tissues and their derived primary fibroblast cultures (53).
To investigate whether Etk is expressed in keloids, we performed RT-PCR analyses using RNA prepared from fresh tissue lysates, two degenerate primer pairs, and one reverse transcription primer. In this study we collected five sets of tissues, keloid and normal skin, matched by race, age, gender, and anatomical site as control (47), and carried out tissue homogenate lysates for both RNA and protein extraction. A major RT-PCR product with a predicted size of ∼440 bp was obtained with the use of the keloid tissue RNA (data not shown). A second round of PCR using the nested primer pairs was performed to further confirm that the obtained RT-PCR product represented a bona fide expression of Etk. As shown in Fig. 5A, two PCR fragments with predicted sizes of ∼260 bp and ∼370 bp, respectively, were generated with RT-PCR product, using the keloid tissue RNA as template (lanes 3 and 6). To further substantiate the presence of Etk protein in the keloid tissues, Western blot analyses with an anti-Etk antibody were performed. As shown in Fig. 5B, keloid tissues expressed a higher level of the Etk protein (∼76 kDa) compared with normal skin tissues.
We recently demonstrated (58) that a genistein-sensitive tyrosine kinase-dependent signaling pathway indeed attenuated hypoxia-induced PAI-1 expression in keloid fibroblasts. Because hypoxia induction of VEGF and PAI-1 as well as the activation of NF-κB are all in existence in keloid fibroblasts (28), it is of great interest to determine whether Etk is indeed present in our keloid-derived fibroblasts. Primary cultures of fibroblasts were isolated from the same sets of tissues, keloid and matched normal skin, and were extracted for Western blot analyses. As shown in Fig. 5C, keloid fibroblasts demonstrated an elevated Etk protein level compared with matched normal skin fibroblasts.
Together these data suggest that the expression of Etk detected in keloids may contribute to part of the tyrosine kinase signaling pathways that are activated as a result of injury from stress.
The molecular basis of wound healing and tumor progression is governed by the interplay of cellular signaling pathways, the regulation of specific target gene activation, and the nature of the microenvironment for eliciting an adaptive response to environmental cues. In this report, we demonstrated that hypoxia (1% O2) treatment increases the steady-state level of VEGF and PAI-1 mRNA in salivary epithelial cells (Figs. 1 and 2). Etk activation alone is able to modulate HRE-dependent VEGF expression in both nonhypoxic and hypoxic conditions (Fig. 1). HRE-dependent reporter activity was attenuated with the use of wortmannin and LFM-A13, both of which are known inhibitors for the Etk signaling network (10, 59). We also demonstrated that Etk activation induces PAI-1 expression in epithelial cells (Fig. 2). Using keloid tissues as a model to investigate the mechanism underlying adaptive hypoxic response because of their state of local hypoxia, we also demonstrated an elevated basal level of Etk in both tissue lysates and their derived keloid fibroblasts compared with matched normal skin controls (Fig. 5). This observation is consistent with our previous report (58) that treatment with genistein led to a significant decline in hypoxia-induced PAI-1 mRNA and protein expression in keloid fibroblasts. The data reported here (Fig. 4) represent the first demonstration that hypoxia activates Etk and that this activation is necessary for subsequent Akt activation. However, in contrast to the PI3-kinase/Akt-stimulated VEGF transcriptional activation via HRE, Etk failed to robustly stimulate the activity of PAI-1 promoter/enhancer.
On the basis of our data, it is unequivocally demonstrated that hypoxia in conjunction with Etk synergistically enhanced the HRE-driven gene expression. Conceivably, Etk may be involved in the signaling pathways mediating the hypoxia-induced response. Given the large number of genes that have already been identified as targets under hypoxia-elicited regulation, HIF-1 is becoming a major transcription factor inducing the expression of gene products that allow for adaptation to hypoxia to occur. HIF-1 is a heterodimeric basic helix-loop-helix protein composed of HIF-1α and HIF-1β (also known as the aryl hydrocarbon nuclear translocator or ARNT) subunits (49, 50). Although HIF-1β is constitutively expressed, the steady-state level of HIF-1α is inducible and stabilized under hypoxic conditions (46) but subjected to ubiquitination and proteasome-mediated degradation under nonhypoxic conditions (17, 23, 27, 42). If Etk activation can also elicit a stimulatory effect to upregulate the expression of these HIF-1-dependent genes via a mechanism other than stabilizing HIF-1α, it is possible that Etk activation can complement and synergize the oxygen-dependent Siah/HIF prolyl hydroxylase domain-mediated HIF-1 regulation (reviewed in Ref. 44) in both normoxia and hypoxia.
Etk has also been proposed to play a role in evading apoptosis by mediating the activation of NF-κB, a transcription factor involved in upregulating antiapoptotic or prosurvival genes. In this report (Fig. 3), we have demonstrated that Etk activation resulted in the stimulation of NF-κB-dependent reporter activation and induction of iNOS, a known NF-κB downstream target gene. We predict that this in turn may stimulate the induction of downstream signaling cascades to elevate the prosurvival status of cells in response to noxious agents such as proinflammatory cytokines. The intermediary of this transcriptional upregulation involves Akt, functioning as a downstream effector of Etk (Fig. 4). Given that Etk is a cytoplasmic PTK, it is more likely that Etk is going to be a facilitator/mediator of signaling pathways downstream of extracellular stimuli. On the basis of the data in Fig. 4, we believe that hypoxia directly activates Etk, and Zhang et al. (59) and Pan et al. (34) have also independently demonstrated that TNF-α treatment induces Etk activation. This possibility is further supported by the finding that myr-Etk elicited a modest activation of NF-κB but cooperated with TNF-α treatment (Fig. 3D, lanes 2 and 3). The exact reason that myr-Etk elicited a modest activation is still unclear. It is possible that the membrane localization signal may affect its downstream signaling events. Although we do not have direct evidence of Etk stimulating the NF-κB consensus sequence within the promoter/enhancer of iNOS gene, we can assume the following series of events. Given that Etk can associate with the TNF-α receptor TNFR2 (34), the binding of TNF-α to its receptor triggers downstream signaling events that include Etk activation. This eventually results in the translocation of NF-κB into the nucleus to activate gene transcription of its target gene(s). Thus our Western blot analyses that showed greater enhancement of iNOS protein levels in stably Etk-transfected cells in both the absence and the presence of TNF-α stimulation compared with the parental cells would support our hypotheses of the signaling cascade as described above. Our data further show that interference with Etk activation with dominant-negative or kinase-dead Etk is able to attenuate NF-κB reporter activation (Fig. 3). Together with Figs. 1 and 2, we clearly demonstrate that Etk activation has a potent effect, leading to the stimulation of both hypoxia- and inflammatory cytokine-mediated reporter and gene activation.
The abnormalities in the NF-κB and HIF-1 pathways are also evident in various forms of human cancers, which often involve constitutively high nuclear expression levels of NF-κB and HIF-1 themselves. This study has elucidated the Etk-dependent signaling pathway involved in triggering the activation of both HIF-1 and NF-κB pathways and possibly the antiapoptotic response of cells to environmental cues. Conceivably, the resultant suppression of apoptosis depends on the regulation of genes involved in multiple aspects of growth control, allowing for cellular proliferation and/or propagation of tumor progression. Therefore, the observed concurrent activation of HIF-1 and NF-κB signaling events by Etk supports the unique role proposed for Etk in promoting not only wound healing but also possible tumor progression.
The exact nature of the cross talk between Etk and hypoxia on activating both VEGF and PAI-1 expression remains enigmatic. Notwithstanding the uncertainty concerning the detailed mechanism underlying this process, our results unequivocally demonstrate a crucial role of Etk in augmenting HIF-1 activation and in mediating NF-κB induction (Fig. 6). Although it is possible that these events are indirectly linked, our results suggest that they may functionally cooperate downstream from different extracellular stimuli, such as the hypoxia-mediated pathway and proinflammatory cytokine-dependent signaling. The most plausible model is that Etk activation is part of a central regulatory circuit, which directs both HIF-1- and NF-κB-triggered adaptive responses. It is still unclear whether hypoxic exposure leads to direct Etk activation. However, the reports by Zhang et al. (56) and Ping et al. (36) on the involvement of Etk/BMX expression and distribution in the setting of cardioprotection support our notion. Additionally, the finding by Pan et al. (34) that TNF via TNFR2 transactivates Etk supports the possibility that Etk can be indirectly activated. We are currently investigating both possibilities. We postulate that Etk-mediated pathways might have an important role in fine-tuning the response elicited by either hypoxia or proinflammatory cytokines, mediating the activation of both HIF-1 and NF-κB pathways.
Furthermore, recent transgenic studies by Paavonen et al. (33), using a K14 promoter to target Etk/Bmx expression in basal keratinocytes, support the notion that Etk function is associated with epithelial cell proliferation and migration, as well as acceleration of wound healing. The expression of Etk in the skin of these K14-Etk transgenic mice also induced chronic inflammation and angiogenesis. Together, our present data provide detailed molecular mechanisms underlying Etk function and linking Etk with NF-κB activation. Our findings clearly establish, for the first time, that Etk stimulates VEGF, PAI-1, and iNOS expression and suggest that Etk may serve a cytoprotective role against extracellular stresses and stimuli by activating Akt. These observations open a new avenue of investigation closely linking the Etk signaling network with epithelial and endothelial cell biology. Given the role of Etk in mediating various adaptive responses, Etk could be a novel therapeutic target for hypoxia- and inflammation-associated diseases such as keloids.
This work was supported in part by National Institutes of Health Grants R01-DE-10742 and DE-14183 (to D. K. Ann), HL-38658 (to K. J. Kim), and 1S11-AR-47359 (to A. D. Le), and Predoctoral Fellowship DE-07211 (to C. H. Chau).
We thank Dr. Ebrahim Zandi (University of Southern California, Los Angeles, CA), Dr. Oliver Hankinson (University of California, Los Angeles, CA), and Dr. Gregg Semenza (Johns Hopkins University, Baltimore, MD) for providing various reagents.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society