Evidence suggests flow-induced arterial remodeling involves factors released from cells that are intrinsic to the vessel wall and recruited from the bloodstream. Understanding the molecular details has been hampered by the need to study the process in vivo. The present findings suggest that HB-EGF, which has primarily been studied in epithelial and tumor cells, plays a pivotal role in low flow-induced negative hypertrophic remodeling (FINR) of the mouse carotid artery. Sustained low flow activated or increased the following elements within the HB-EGF signaling pathway: ROS, the ROS-sensitive HB-EGF sheddase TACE, expression of pro-HB-EGF, HB-EGF immunoreactivity, the HB-EGF receptor EGFR, Erk1/2, and the transcription factor NF-κB. These changes were associated with proliferation, increased leukocyte density, wall hypertrophy and lumen narrowing. Heterozygous and homozygous deletion of HB-EGF alleles caused “dose-dependent-like” inhibition of FINR (although inhibition of lumen narrowing was in some situations spared (see below), where proliferation and leukocyte accumulation were reduced and FINR almost abolished in HB-EGF-/-. Inhibition was also obtained by genetic and pharmacologic reduction in ROS generation and inhibition of EGFR activation. FINR was unaffected in mice deficient in the EGFR ligand family members, betacellulin and amphiregulin.
Interestingly, none of the above interventions affected high flow-induced eutrophic positive remodeling of the right carotid. This suggests that low and high flow-mediated remodeling are achieved by unique mechanisms. A second intriguing observation is that in FINR, significant inhibition of wall hypertrophy in HB-EGF+/- and p47phox-/- mice, and with AG1478 or apocynin treatment, did not inhibit lumen narrowing (an exception was in HB-EGF-/- mice, but their baseline diameters trended lower for unapparent reasons). This suggests that HB-EGF signaling is necessary for the hypertrophic response in FINR, but not for the mechanisms that cause lumen narrowing. The latter mechanisms may be driven in a negative feedback manner to normalize shear stress. This is the first report identifying a role for the HB-EGF pathway in flow-induced remodeling.
Although HB-EGF is well-known to activate EGFR and downstream mechanisms including Akt and Erk1/2 in SMCs, ECs and other cell types, few studies have investigated possible links between reduced shear stress, ROS and HB-EGF activation. Reduced shear stress causes rapid increase in HB-EGF expression in ECs19
. Low or disturbed shear stress promotes inflammatory-like conditions associated with increased ROS32
, in particular reduced nitric oxide35
, increased expression of attractant and adhesion molecules that promote leukocyte transmigration3
, and increased local angiotensin, endothelin and norepinephrine activity that induce ROS- and HB-EGF-dependent GPCR transactivation of EGFR14,15,23
. Although angiotensin and endothelin have not been examined in FINR, norepinephrine-induced trophic activity contributes significantly to FINR13
. In agreement with the prominent role of SMC proliferation in FINR7,10,15
, proliferation was inhibited in the media but not intima in HB-EGF-/-
mice ()— a finding consistent with HB-EGF's mitogenic action on SMCs but not ECs18
. Besides SMCs, leukocytes and ECs (and possibly adventitial fibroblasts) also release HB-EGF and express EGFR in vitro18,19
. Indeed, our histological results confirm ROS, TACE and EGFR activity in each vascular layer. Thus, autacrine/paracrine HB-EGF activity in FINR could involve all four cell types. Additional studies are required to determine their relative contributions. It should be noted that only associative evidence is provided for two of the elements (TACE, Erk1/2) in the signaling pathway proposed herein, because in the case of TACE specific antagonists and genetic models (TACE-/-
mice are embryonic lethal) are currently unavailable.
Primarily in vitro
studies have shown that pro-HB-EGF is cleaved by ADAM family proteases and MMPs18
. Evidence suggests that TACE (ADAM-17) is the primary mediator of HB-EGF shedding in many cells types, including SMCs where proliferation and hypertrophy result23
. In the present study, TACE immunoreactivity was elevated at the earliest time point examined after flow reduction (12 hr). ROS can increase MMP expression through Erk1/2, JNK, AP-1 and Ets-136
. While this pathway has not been examined for TACE, TACE may be directly activated by a ROS-sensitive cysteine switch mechanism37
. In the present study, systemic administration of the NAD(P)H oxidase inhibitor apocynin, as well as p47phox-/-
mice, resulted in potent inhibition of FINR, while increased mitochondrial ROS activity was not evident. Although the role of other sources of ROS and the specific ROS molecule(s) and pro-HB-EGF protease(s) that direct FINR remain to be determined, our findings demonstrate a primary role for NAD(P)H oxidase.
The observation that pro-HB-EGF expression occurs rapidly (hours) in response to reduced shear stress in ECs19
and to other stimuli in other cells, suggests that HB-EGF behaves as an immediate-early gene21
. The HB-EGF promoter contains a consensus site for NF-κB21
. Binding of NF-κB is redox sensitive, and activation of NF-κB by TNF-α and IL-β has been linked to induction of ROS molecules34
. Indeed, low shear stress in cultured ECs activates NF-κB by a NAD(P)H oxidase-dependent mechanism32
. Our results support the presence of this signaling sequence in arteries exposed to low flow in vivo
. Interestingly, in HB-EGF-/-
mice, activation of NF-κB was unaffected in ECs but abolished in SMCs (), suggesting that low shear-induced activation of NF-κB is upstream of HB-EGF in ECs but downstream in SMCs. Consistent with this, in SMCs NF-κB did not induce expression of EGFR mRNA or protein38
, whereas EGFR stimulation induced NF-κB activation39
. Others have found that NF-κB in SMCs resides downstream of EGFR in a pathway inducing Erk1/2 and Akt mediated proliferation and apoptosis40
. Also, peak activation of NF-κB occurred earlier in ECs than in SMCs (), possibly because ECs are more directly coupled mechanically to fluid shear stress than are SMCs.
Besides ECs and SMCs, FINR-induced leukocyte accumulation is also a potential source of HB-EGF18,26
. Activated macrophages exhibit increased HB-EGF expression and release18,26
. In the present study we confirmed earlier reports13
for leukocyte accumulation in intima, media and adventitia during FINR, and detected reduction in HB-EGF-/-
mice. Interestingly, HB-EGF promotes macrophage accumulation in ischemic heart31
. Thus, HB-EGF release from vascular wall cells may contribute to recruitment of inflammatory cells that further increase local HB-EGF activity. We did not find that reduced leukocyte density in HB-EGF-/-
mice was accompanied by reduced circulating leukocyte counts or migratory capacity. However, pharmacological depletion of macrophages reduced FINR of small branches of the mesenteric artery12
. Whether the association of reduced leukocyte density and reduced FINR in that study12
and the present study extends from less HB-EGF in the wall coming from fewer infiltrated cells— or from the attendant lower levels of cytokines such as IL1/6 and TNFα that induce HB-EGF in vascular wall cells— awaits studies using selective leukocyte depletion of HB-EGF.
Partial inhibition of FINR was obtained with chronic systemic administration of the EGFR antagonist AG1478-mesylate (). We did use a higher dose to test for incomplete blockade because of potential non-specific effects27
. However, ErbB receptor-ligand interactions could also underlie the partial inhibition. ErbB1 (EGFR) and ErbB4 have different biological functions18
. Proliferation is mediated by ErbB1, whereas chemotaxis is mediated by ErbB418
. Also, ErbB2 (for which no ligand has been identified) forms heterodimers with ErbB1 or ErbB4 which may induce cross-talk among downstream pathways promoting proliferation or migration18
. Neither ErbB2 nor ErbB4 are blocked by AG1478. In addition, Higashiyama and coworkers have recently identified an intracellular signaling pathway activated by pro-HB-EGF cleavage41
, wherein the intracellular carboxy-terminal remnant (HB-EGF-C) interacts with the transcriptional regulator, promyelocytic leukemia zinc finger (PLZF), resulting in increased proliferation. This mechanism, which is independent of EGFR— thus not inhibited by AG1478— could contribute to the residual FINR in the presence of AG1478 that we observed.
In conclusion, the present study has identified a central role for ROS→ HB-EGF→ EGFR signaling in hypertrophic low flow-induced arterial remodeling. Besides physiological remodeling, this pathway may also contribute to pathological processes. For example, oxidized LDL and remnant lipoproteins induce HB-EGF24
, and HB-EGF increases expression of oxidized LDL receptor25
. TACE, pro-HB-EGF and EGFR are increased in human atheromas and in those of animals with experimentally induced atherosclerosis17,26
. And plasma HB-EGF is increased in patients with atherosclerosis18
. It is therefore intriguing to hypothesize that increased HB-EGF signaling might impair the important adaptive outward remodeling response that occurs at sites of expanding atheromas along arteries3
. If this were true, inhibition of pathway activity might enhance this process and promote preservation of lumen area. Our finding that HB-EGF signaling is not involved in flow-induced positive remodeling suggests that blocking adverse effects of excessive HB-EGF activity may leave this important physiological mechanism intact.