CXCR4 and its ligand CXCL12 are expressed in cardiac myocytes and fibroblasts, and myocardial ischemia significantly upregulates CXCL12 (Pillarisetti and Gupta, 2001; Yamani et al., 2005; Hu et al., 2007). Different studies have revealed a protective role for CXCL12/CXCR4 signaling after MI and MI/IRI through survival effects on resident cardiomyocytes and recruitment of protective circulating cells. Intracardiac or intramyocardial injection of CXCL12 reduced infarction size and increased cardiac function after MI/IRI (Hu et al., 2007) and MI (Segers et al., 2007; Saxena et al., 2008), and cardioprotective effects were blocked with AMD3100 (Hu et al., 2007). CXCL12-induced cardioprotection was associated with improved survival of hypoxic myocardium and increased neo-angiogenesis, and was linked with anti-apoptotic AKT (also known as PKB = protein kinase B) and MAPK3/1 (also known as ERK1/2) signaling in cardiac myocytes and endothelial cells (Hu et al., 2007; Saxena et al., 2008). Also, delivery of CXCL12 triggered upregulation of vascular endothelial growth factor (VEGF) in the infarcted area and in cardiac endothelial cells, with VEGF an important regulator of angiogenesis and progenitor cell recruitment (Saxena et al., 2008). Furthermore, multiple studies have explored the effect of CXCL12 delivery into the myocardium through local treatment with CXCL12-overexpressing adenovirus, CXCL12-transgenic skeletal myoblasts or CXCL12-releasing hydrogels. Such exogenous CXCL12 delivery was associated with enhanced recruitment and incorporation of CXCR4⁺ stem and progenitor cells in the infarcted area (Abbott et al., 2004; Elmadbouh et al., 2007; Segers et al., 2007; Purcell et al., 2012).

Stem cell transplantation during MI seems promising to improve cardiac outcome (Sanganalmath and Bolli, 2013), but it is debated whether this is primarily mediated through direct regeneration of cardiac myocytes or through protective paracrine effects on remodeling or preservation of injured tissue (Liehn et al., 2013b). For example, transplantation of endothelial progenitor cells (EPCs) was associated with increased neovascularization and improved cardiac function after MI and MI/IRI, despite variable effects on inflammation and apoptosis (Schuh et al., 2008, 2012). Overexpression of CXCL12 in transplanted EPCs further increased angiogenesis, however without significant improvement of cardiac function (Schuh et al., 2012). In a complementary approach, transplantation of mesenchymal stem cells (MSCs), which have been described as cardiac precursors, is widely investigated as a potential therapy after MI (Hatzistergos et al., 2010; Dong et al., 2012; Liehn et al., 2013b). Overexpression of CXCL12 in transplanted MSCs improved survival of cardiomyocytes after MI, however without evidence for cardiac regeneration (Zhang et al., 2007). MSCs with transgenic CXCR4 expression displayed increased incorporation into the ischemic area, which was associated with increased angiogenesis, myogenesis and cardiac function (Zhang et al., 2008; Huang et al., 2012). In vitro experiments demonstrated hypoxia to increase CXCR4 expression on MSCs, and CXCR4-mediated migration of MSCs to CXCL12 was shown to require PI3K/AKT signaling (Yu et al., 2010). The CXCL12/CXCR4 axis was also at least partly responsible for the beneficial effects of VEGF-overexpressing MSCs by mediating the recruitment of cardiac stem cells to the infarcted region (Tang et al., 2011). Furthermore, cardioprotective effects of transplanted MSCs were shown to require myocardial CXCR4 expression (Dong et al., 2012).

The efficiency of AMD3100 in mobilizing progenitor cells, including EPCs, from the bone marrow was associated with enhanced accumulation of progenitor cells in the infarcted tissue, enhanced neovascularization and improved cardiac function after single AMD3100 treatment in an MI and MI/IRI model (Jujo et al., 2010, 2013). Similarly, cardioprotective effects were reported of daily AMD3100 injections after MI in rats (Proulx et al., 2007). However, two independent groups revealed a reduced cardiac outcome after MI upon chronic AMD3100 administration (Dai et al., 2010; Jujo et al., 2010). This was either associated with a reduced incorporation of progenitor cells in the infarcted region despite enhanced mobilization (Jujo et al., 2010), or with an increased proliferation of resident cardiac progenitor cells. This latter observation raised the suggestion whether increased proliferation may be linked to reduced differentiation, so whether cardioprotective CXCR4 signaling may be required to direct cardiac progenitors to cardiac commitment to ensure their participation in repair of injured myocardium (Dai et al., 2010).

Interestingly, platelet-surface binding of CXCL12, which correlated with platelet activation, was significantly increased in patients with acute coronary syndrome (ACS) compared to patients with stable angina pectoris, and correlated with the number of circulating hematopoietic progenitor cells (Stellos et al., 2009). Similarly, surface expression of CXCR7 but not CXCR4 was found to be significantly enhanced on platelets from patients with ACS compared to subjects with stable CAD (Rath et al., 2014). These data may be supported by recent findings that CXCL12 upregulates CXCR7 surface availability on platelets (Chatterjee et al., 2014), as will be discussed in more detail later. Of note, platelet CXCR7 surface expression levels above average in patients with ACS positively correlated with an increase in the left ventricular ejection fraction as a measure of recovery after MI, suggesting a beneficial effect of CXCL12/CXCR7 signaling on functional recovery in ACS patients (Rath et al., 2014).

Despite cardioprotective functions of the CXCL12/CXCR4 axis in the ischemic heart, Cxcr4-heterozygosity in mice reduced infarct size after MI, however without affecting cardiac function. This was explained by a counterbalance of on the one hand reduced neovascularization, and on the other hand reduced inflammation with less neutrophils and a preferential recruitment of Gr1^low over inflammatory Gr1^high monocytes (Liehn et al., 2011). Likewise, adenovirus-mediated overexpression of CXCR4 in the heart increased infarct size and reduced cardiac function. This was associated with an enhanced recruitment of inflammatory cells, enhanced tumor necrosis factor (TNF) α expression and increased apoptosis of cardiomyocytes (Chen et al., 2010a). On the other hand, deficiency of Cxcr4 specifically in cardiac myocytes did not affect heart function or remodeling after MI (Agarwal et al., 2010).

Together, these studies demonstrate a double-edged role of CXCR4 in the ischemic heart, and require further investigation of the role of CXCR4 and its chemokine ligands in the inflammatory processes associated with MI. In this context, also the alternative CXCR4 ligand MIF is upregulated in myocardium and plasma after MI (Yu et al., 2001, 2003) and can exert cardioprotection. Underlying mechanisms include activation of AMP-activated protein kinase (AMPK) (Miller et al., 2008), reduction of oxidative stress (Koga et al., 2011) or inhibition of c-Jun N-terminal kinase (JNK)-mediated apoptosis (Qi et al., 2009). An important role for the chemokine receptor CXCR2 on resident cardiac cells in MIF-mediated myocardial protection after ischemic injury was recently shown (Liehn et al., 2013a), but it remains unclear whether also MIF/CXCR4 signaling in cardiomyocytes may contribute to cardioprotection. Furthermore, MIF-induced recruitment and differentiation of protective EPCs through CXCR4 and CXCR2 may be involved in the cardioprotective effects of MIF (Simons et al., 2011; Asare et al., 2013; Kanzler et al., 2013). On the other hand, adverse effects of MIF through myocardial infiltration of inflammatory cells were revealed after prolonged ischemic injury in both MI and MI/IRI (Gao et al., 2011; White et al., 2013). Although these studies did not examine which MIF receptor was involved, a recent report demonstrated an important role for CXCR2 in mediating MIF-triggered monocyte recruitment in the ischemic heart (Liehn et al., 2013a). But also here, an additional involvement of MIF/CXCR4 interaction in ischemic inflammatory cell recruitment remains unclear. In conclusion, recent data indicate that similarly as CXCR4, MIF plays a double-edged role in myocardial ischemia. However, the relative importance of CXCR4 vs. other MIF receptors as CXCR2, in MIF-mediated effects remain unclear.

In contrast to conditions of MI and injury-induced restenosis, not much is known about a potential involvement of bone marrow-derived vascular progenitor cells in native atherosclerosis, as has been recently summarized for EPCs (Du et al., 2012) and vascular smooth muscle progenitor cells (SPCs) (Merkulova-Rainon et al., 2012). It was shown that infusion of EPCs as well as treatment with AMD3100, which triggered EPC mobilization, enhanced plaque regression after normalization of plasma lipid levels in Reversa mice (Yao et al., 2012). In contrast, a study in Apoe^−/− mice did not reveal an atheroprotective effect of systemic AMD3100 treatment, but rather found AMD3100 to abolish beneficial effects of apoptotic body treatment on atherosclerosis. Endothelial apoptotic bodies were shown to contain miRNA126, which is transferred to neighboring ECs to induce the expression and release of CXCL12 by unleashing autoregulatory CXCR4 signaling. Injection of EC-derived apoptotic bodies into Apoe^−/− mice increased CXCL12 expression in atherosclerotic lesions and promoted progenitor cell mobilization and their recruitment to the endothelial lining of the lesions. Lesions of mice treated with apoptotic bodies were generally smaller in size and exhibited a less inflammatory phenotype with reduced macrophage and apoptotic cell content (Zernecke et al., 2009). Thus, despite contradictory findings on the effect of AMD3100 treatment, both studies suggest an atheroprotective function for CXCL12/CXCR4 signaling through mobilization of protective EPCs. Interestingly, it was revealed that patients with CAD show lower levels and a decreased migratory response of circulating EPCs (Vasa et al., 2001). In addition, systemic treatment of mice with CXCL12 in a partial ligation model—which induces advanced atherosclerotic lesions with an unstable phenotype—enhanced the recruitment of Lin⁻Sca1⁺ SPCs and promoted a more stable plaque phenotype (Akhtar et al., 2013). Such plaque-stabilizing role for SPCs was also suggested earlier by Zoll et al, who showed that injection of SPCs reduced atherosclerotic lesion size and improved lesion stability (Zoll et al., 2008). Together, these studies reveal an atheroprotective function for CXCL12/CXCR4 signaling through recruitment of protective EPCs and plaque-stabilizing SPCs.

However, others reported on an atheroprogressive role for vascular progenitor cells. Inducing apoptosis of rare lesional bone marrow-derived SMCs substantially decreased plaque size (Yu et al., 2011a), and George et al. found EPC transfer to increase atherosclerosis in mice (George et al., 2005). Contradictory findings on the presence or function of vascular progenitor cells in chronic atherosclerosis may be related to the atherosclerosis model, lesion stage, but also on the vague definition of such progenitor cells. Further research is definitively needed to improve phenotypical and functional characterization of vascular progenitor cell subsets in context of atherosclerosis before a role of the CXCR4/CXCL12 axis in their mobilization and potential functions in this pathology can be investigated in detail.

Interestingly, MI was recently revealed to accelerate atherosclerosis in mice. MI reduced CXCL12 expression in bone marrow through sympathetic nervous system activity and signaling through β 3 adrenergic receptor (β 3AR or ADRB3). In this way, MI enhanced mobilization of HSPCs from bone marrow niches and their hosting in the spleen, triggering myelopoiesis and increased atherosclerosis up to 3 months after coronary ligation (Dutta et al., 2012). Although the latter study did not address the underlying mechanisms of CXCL12 upregulation upon β 3AR blocking nor examine potential effects on CXCR4 expression or function, LaRocca et al. revealed a direct (physical) interaction of β 2 adrenergic receptors with CXCR4 resulting in the modification of the contractile nature of cardiomyocytes (Larocca et al., 2010). It is further known that β 2- and β 3 adrenergic receptors cooperate during progenitor cell mobilization with partial functional redundancy under stress (Mendez-Ferrer et al., 2010). Hence, one may speculate that Cxcr4 may also functionally interact with other adrenergic receptors, like β 3AR, in a direct or indirect way.

Monocytes and macrophages have been proven to be of outstanding importance in the progression and development of mature atherosclerotic lesions and their depletion has been recognized atheroprotective already 20 years ago (Ylitalo et al., 1994). As the picture grows it becomes more and more evident that depletion of individual cell subsets does not serve as a realistic therapeutic approach, hence understanding the details of cell–cell interactions and communication gains importance (Weber and Noels, 2011).

Macrophages and foam cells. CXCR4 is expressed on all monocyte subsets, with highest expression on classical human monocytes. This contrasts with observations in mice, which show highest CXCR4 levels on non-classical monocytes (Ingersoll et al., 2010), as will be discussed later in more detail. Gupta et al. revealed high expression of CXCR4 on human blood monocytes, which declined while they differentiated into macrophages, but restored again after 24 h, peaking at 7 days. Interestingly, CXCR4 expression on macrophages could be further upregulated by stimulation with oxidized low-density lipoprotein (oxLDL). From here the authors conclude that, although there is no direct evidence, restoration of CXCR4 expression on lesional macrophages and its further up-regulation by oxLDL during foam cell formation may contribute to migration of intimal foam cells and the subsequent progression of plaque growth (Gupta et al., 1999). Furthermore, CXCL12/CXCR4 signaling was linked with enhanced macropinocytosis in leukocytes (Tanaka et al., 2012), suggesting that a lack of CXCR4 may also influence (modified) lipid accumulation in macrophages and other lesional cells. In contrast, a recent study found CXCL12 to induce phagocytosis and the uptake of acetylated LDL in THP1-derived macrophages specifically through binding CXCR7 but not CXCR4 (Ma et al., 2013). In this context, the CXCR7 agonist CCX771 was recently shown to increase the uptake of very low-density LDL (VLDL) in adipocytes. Correspondingly, treatment of Apoe^−/− mice with CCX771 reduced the levels of circulating VLDL and decreased atherosclerosis (Li et al., 2014). Whether similar mechanisms can be identified in other cell types as macrophages remains to be examined, as are the exact mechanisms underlying CXCR7-mediated uptake of VLDL or modified lipids.

Patients with CAD. In patients with stable and unstable angina pectoris CXCR4 surface expression on peripheral blood mononuclear cells (PBMCs) was decreased and CXCL12 levels in patients with unstable angina pectoris were explicitly low. However, in vitro treatment of PBMCs from these patients with high concentrations of CXCL12 reduced mRNA and protein levels of chemokine ligands CCL2 and CXCL8, MMP9 and tissue factor, while increasing tissue inhibitor of metalloproteinases (TIMP) 1. Therefore, high (local) concentrations of CXCL12 may mediate anti-inflammatory and matrix-stabilizing effects promoting plaque stabilization and may be beneficial in angina pectoris and ACSs (Damas et al., 2002). Another study showed autocrine CXCL12 signaling to downregulate expression of runt-related transcription factor (RUNX) 3 in human monocytes/macrophages, thereby promoting a pro-angiogenic, but immunosuppressive phenotype of these cells (Sanchez-Martin et al., 2011).

Compounds regulating CXCR4. Several compounds were identified to modify CXCR4-mediated immune responses in vitro. Glucocorticoids have been demonstrated to upregulate CXCR4 expression on human blood monocytes and Caulfied et al. assume that increased CXCR4 expression sensitizes monocytes to tissue resident CXCL12, guiding monocytes away from sites of inflammation with supposingly lesser local CXCL12 release (Caulfield et al., 2002). The latter most likely contradicts another study, which revealed high expression of CXCL12 in SMCs, ECs and macrophages in human atherosclerotic plaques but not in normal vessels (Abi-Younes et al., 2000). This would rather argue for a chemotactic gradient of CXCL12 toward the site of inflammation, although this might be disease- and cell type-dependent.

As a second example, monocyte CXCR4 expression has been shown to be modulated by hydrogen sulfide (H₂S) donors. H₂S donors have lately been recognized as vasoprotective agents and changes in H₂S may affect atherosclerosis. Interestingly, a synthetic slow H₂S releaser (GYY4137) inhibited oxLDL-induced foam cell formation and cholesterol esterification in RAW264 cells and primary human monocytes, which was accompanied by decreased CXCR4 expression (Liu et al., 2013). In contrast, angiotensin-converting enzyme (ACE) inhibitors, widely used to treat high blood pressure by interfering with the renin-angiotensin system, did not affect CXCR4 expression on primary human monocytes and THP1 cells (Apostolakis et al., 2007). The same group also assessed if angiotensin I and II treatment would have a direct impact on chemokine receptor expression on THP1 cells, again CXCR4 expression was not altered (Apostolakis et al., 2010).

Conjugated linoleic acids (CLA) were shown to influence human peripheral blood monocyte function by suppressing CD18 expression, thereby reducing the number of β 2-integrins expressed on the external surface and decreasing adhesion to activated ECs. In addition, CLA reduced CXCR4 expression, resulting in an only minor initiation of “inside out” signaling. As a result, partial, but incomplete, activation of β 2-integrins further reduces adherence of leukocytes and their migration to CXCL12 (De Gaetano et al., 2013).

Statins, which lower intracellular cholesterol synthesis, are the gold standard to treat hyperlipidemia-associated atherosclerosis, but have also been reported to cause numerous other pleiotropic effects. To examine if statin withdrawal would affect human monocyte subsets in patients with stable CAD, statin treatment was cut off for 2 weeks. Subsequent evaluation of blood monocyte subsets did not reveal any differences in numbers, but downregulation of Toll-like receptor (TLR) 4 on all subsets and decreased expression of CXCR4 on classical monocytes (CD14⁺⁺ CD16⁻) (Jaipersad et al., 2013). Interestingly, high doses of statin treatment were also shown to reduce general CXCL12 plasma levels in hyperlipidemic patients (Camnitz et al., 2012). However, it remains elusive whether statin treatment and a subsequent increase in CXCR4 expression, but decreased CXCL12 titers, point at a direct pro- or anti-atherosclerotic role of the CXCL12/CXCR4 axis.

Another mechanism, recently recognized to drive atherosclerotic lesion growth, is hypoxia (Marsch et al., 2013). Notably, hypoxia-induced upregulation of the transcriptional activator HIF1 triggers CXCR4 mRNA and protein expression in human monocytes. In addition, these cells showed increased migration toward CXCL12 under hypoxic conditions. Based on these findings the authors conclude that the hypoxia–HIF1–CXCR4 pathway may regulate cell trafficking and localization into hypoxic tissues, such as atherosclerotic lesions (Schioppa et al., 2003).

In contrast, CXCR4 expression on macrophages was suppressed in the presence of M-CSF. Increased M-CSF titers have been implicated in the pathogenesis of atherosclerosis and it was shown that M-CSF delivers a pro-atherogenic signal to human macrophages by stimulation of cholesterol accumulation and pro-inflammatory chemokine secretion. Thus, M-CSF-induced suppression of macrophage CXCR4 expression may point at an atheroprotective function of downstream CXCR4 signaling events (Irvine et al., 2009).

CXCR4 expression and potential functions: mouse studies. The above-mentioned studies on human monocytes and macrophages suggest diverse potential roles for CXCR4 in atherosclerosis, still mouse studies are equally elusive, without a clear indication for a pro- or anti-atherogenic role for CXCL12/CXCR4. In contrast to human monocytes, CXCR4 in mouse is higher expressed on non-classical monocytes and it is not clear if this does necessarily imply functional differences in individual mouse and human monocyte subsets or not (Ingersoll et al., 2010).

One study reported an interesting finding using a mutant non-heparin sulfate-binding CXCL12 (HSmCXCL12). In vitro, HSmCXCL12 failed to promote transendothelial migration of PBMCs if used as chemoattractant in the bottom well of transwell plates, and inhibited the haptotactic response to wild-type CCL7, CXCL12, and CXCL8. Further, intravenous administration of HSmCXCL12 into mice also repressed the recruitment of lymphocytes and mononuclear phagocytes to air pouches injected with CXCL12. Moreover, repetitive administration of HSmCXCL12 in vivo reduced leukocyte-surface expression of CXCR4, and CXCL12-induced chemotaxis and adhesion. From here the authors conclude that non-heparin sulfate-binding variants of CXCL12 can mediate a powerful anti-inflammatory effect through induction of chronic CXCR4 internalization on leukocytes in vivo. Subsequently, this leads to receptor desensitization putatively explaining the functional deficits of these leukocytes (O'boyle et al., 2009). Hence, it could be interesting to carefully dissect the differential functional consequences of receptor desensitization through receptor internalization compared to CXCR4 blocking with AMD as reported by Zernecke et al. (2008).

A potential pro-inflammatory role for wild-type CXCL12 may also be deduced from findings from Liu et al., who revealed decreased CXCR4 expression on RAW264 cells and primary human monocytes treated with the H₂S releaser GYY4137, which was introduced above. Administration of GYY4137 into Apoe^−/− mice receiving a high-fat diet for 4 weeks decreased atherosclerotic plaque formation and partially restored aortic endothelium-dependent relaxation. Further, intercellular adhesion molecule (ICAM) 1, TNF-α and interleukin (IL) 6 mRNA expression as well as superoxide generation in the aorta declined in mice treated with GYY4137 (Liu et al., 2013). Similarly, and again paralleling in vitro studies with human monocytes, CLA treatment of Apoe^−/− mice with already pre-established atherosclerosis induced lesion regression by reducing leukocyte adhesion and decreasing CD18 expression on classical monocytes (De Gaetano et al., 2013).

The above studies may point at a pro-atherogenic role of CXCR4 signaling in atherosclerosis; however this may strongly depend on the binding partner interacting with CXCR4. As already described, CXCL12 is not the only ligand for CXCR4 and MIF, an alternative ligand for CXCR4, has a strong pro-atherogenic impact. In this context, Bernhagen et al. revealed that antibody mediated-neutralization of MIF, but not CXCL12, induced atherosclerotic lesion regression in Apoe^−/− mice (Bernhagen et al., 2007). In line, knock-out of Mif in Ldlr^−/− mice did also result in diminished atherosclerosis (Pan et al., 2004). This suggests potential different roles of CXCR4 and its ligand CXCL12 in atherosclerosis through interplay of CXCR4 with MIF. Similarly, interplay of CXCL12 with other signaling molecules may modify its inflammatory effects. For example, hetero-complexes of high mobility group box (HMGB) 1 and CXCL12 were reported to induce inflammatory cell recruitment to injured tissue, which was not the case for each compound alone. Further, these complexes exclusively bound to CXCR4 inducing a conformational rearrangement of CXCR4, which differed from the single binding of CXCL12 to its receptor (Schiraldi et al., 2012).

Treatment of mice with the CXCR4 antagonist AMD3100 induced cell egress from the bone marrow (Schiraldi et al., 2012), which is in line with findings by Zernecke et al. who described increased leukocytosis, mostly neutrophil mobilization, and enhanced lesion formation in Apoe^−/− receiving a cholesterol-rich diet for 12 weeks while supplemented with AMD. Interestingly, monocyte numbers were only moderately enhanced in these mice and, according to the authors, lesion growth was mainly attributable to increased plaque neutrophils and enhanced apoptosis (Zernecke et al., 2008).

Notably, a growing body of evidence underlines the role of neutrophils in atherogenesis (Drechsler et al., 2010, 2011; Doring et al., 2012) and it was recognized that the CXCL12/CXCR4 axis maintains neutrophil homeostasis primarily by regulation of neutrophil release from the bone marrow in a cell-autonomous fashion (Eash et al., 2009). It was further implicated that senescent neutrophils in the periphery expressing high levels of CXCR4 home back to the bone marrow to be cleared (Martin et al., 2003). In contrast, activated neutrophils downregulate CXCR4 expression putatively postponing their clearance (Bruhl et al., 2003; Martin et al., 2003).

CXCR4 on lymphocytes plays an essential role during B-cell development (Nagasawa et al., 1996) and T-cell homeostasis (Bleul et al., 1996b; Zou et al., 1998). Furthermore, the CXCL12/CXCR4 axis drives chemotaxis or fugetaxis of T-cells in various pathophysiological settings (Poznansky et al., 2000; Dunussi-Joannopoulos et al., 2002; Fernandis et al., 2003; Okabe et al., 2005; Zhang et al., 2005). Similarly, CXCL12 is able to trigger B-cell chemotaxis in vitro through CXCR4 (Klasen et al., 2014).

It became evident that the impact of B- and T-cells in atherosclerosis is strongly subset-dependent. While e.g., Th1 responses are known to be pro-atherosclerotic, regulatory T-cells haven been proven to be protective. Similarly, B1- and B2-cells exhibit diverse functions in lesion development (Weber and Noels, 2011). Nevertheless, studies dissecting the role of CXCR4 on T- and B-cells in the context of atherosclerosis are scarce. Two independent groups showed that lysophosphatidylcholine (LPC), a main phospholipid component of oxLDL, upregulates CXCR4 expression on Jurkat cells and human blood CD4⁺ T-cells. Further, the chemotactic ability of CD4⁺ T-cells toward CXCL12 and their production of pro-inflammatory cytokines were increased in the presence of LPC. Hence, the ill alliance of LPC and CXCL12 in atherosclerotic lesions may amplify pro-inflammatory responses by stimulation of CD4⁺ T-cells and subsequent plaque growth (Han et al., 2004; Hara et al., 2008). Interestingly, it was also shown that excess of mineralocorticoids, mainly aldosterone, drive CAD by cardiac and renal fibrosis, as well as hypertension. Here, Chu et al. imply a specific role of the CXCL12/CXCR4 axis in the detrimental consequences of mineralocorticoid excess and render CXCL12 explicitly responsible for the accumulation of T-cells in fibrotic tissue (Chu et al., 2011). From patients with abdominal aortic aneurysm (AAA) we further learn that T- and B-cells recruited to sites of AAA express high levels of CXCR4 and exhibit a pro-inflammatory signature. Hence, CXCR4/CXCL12 interactions may strongly impact on the recruitment and retention of inflammatory lymphocytes infiltrating AAAs (Ocana et al., 2008). In contrast, acute stress induced by public speaking did not enhance the number of CXCR4 expressing T-cells, but increased the frequency of T-cells expressing CXCR2, CXCR3, and CCR5. Therefore, cardiac sympathetic activation may lead to EC and T-cell activation subsequently driving acute flooding of atherosclerotic lesions with pro-inflammatory mediators (Bosch et al., 2003).

In addition to CXCL12-triggered effects, CXCR4 is able to mediate MIF-induced B-cell chemotaxis, and T-cell chemotaxis and arrest in vitro (Bernhagen et al., 2007; Klasen et al., 2014). Blockade of MIF in Apoe^−/− mice on high-fat diet resulted in the formation of smaller atherosclerotic lesions displaying a reduced macrophage and T-cell content, supporting a role for MIF in T-cell chemotaxis also in the context of atherosclerosis (Bernhagen et al., 2007).

CXCR4 expression (mRNA, protein) was reported on platelets (Wang et al., 1998; Kowalska et al., 1999) and although platelets lack nuclei and many organelles and are mainly known for their important role in blood coagulation, their impact on immunological and inflammatory responses, in particular atherosclerosis, should not be underestimated (Lievens and Von Hundelshausen, 2011). Addition of CXCL12 to platelets from healthy donors induced platelet aggregation, which could be inhibited by blocking CXCR4. The latter implies an atherogenic, pro-thrombotic, and plaque-destabilizing role for the CXCL12/CXCR4 axis in vivo (Falk et al., 1995; Abi-Younes et al., 2000). In contrast, others report CXCL12 to be a weak platelet agonist, however still amplifying platelet activation, adhesion and chemokine release triggered by low doses of primary platelet agonists, such as adenosine diphosphate (ADP) and thrombin, or arterial flow conditions (Kowalska et al., 1999; Gear et al., 2001). Furthermore, CXCL12 gradients could induce platelet migration and transmigration in vitro involving PI3K signaling (Kraemer et al., 2010). In addition, recent work showed CXCL12 to trigger CXCR4 internalization and cyclophilin A-dependent CXCR7 externalization on (mouse and human) platelets, resulting in prolonged platelet survival. Mice lacking the cytosolic chaperone cyclophilin A showed less CXCL12-induced rescue of platelets from activation-induced apoptosis through CXCR7 engagement. Hence, differential regulation of CXCR4/CXCR7 surface expression on platelets upon CXCL12 exposure at sites of platelet activation/accumulation may orchestrate platelet survival, subsequently impacting on platelet-mediated physiological mechanisms (Chatterjee et al., 2014).

CXCR4 expression in arterial ECs. Expression of CXCR4 on various types of vascular ECs has been widely reported (Hillyer et al., 2003), however, it should be emphasized that ECs are a very heterogeneous population, with ECs from different anatomic sites differing in basal gene expression, localization and function (Aird, 2007). Further, one has to carefully distinguish between expression on venous and arterial ECs (Dela Paz and D'amore, 2009). Unfortunately, many studies extrapolate e.g., in vitro findings generated with human umbilical vein endothelial cells (HUVECs) to arteriosclerosis, which is at least daring.

In a study examining CXCR4 expression following vessel wall injury in porcine coronary arteries, CXCR4 expression could be shown 24 h to 7 days after injury, but only in lymphocytes, granulocytes and myelo-fibroblasts entering the injured tissue (Jabs et al., 2007). However, others investigated CXCR4 expression in human carotid artery specimens, where CXCR4 was abundantly expressed by lesional ECs and only marginally in minimally diseased endothelium (Molino et al., 2000; Melchionna et al., 2005). Similarly, Gupta et al. showed CXCR4 mRNA expression in human coronary artery ECs, although it is not clear if these cells originated from inflamed or steady-state endothelium (Gupta et al., 1998). CXCR4 was also shown to be expressed (mRNA and protein) by cultured bovine aortic ECs (BAECs), in cryo-sections of rabbit thoracic aortas (Volin et al., 1998) and in mouse aortic endothelium (Melchionna et al., 2010). For BAECs it was further demonstrated that resting BAECs accumulate CXCR4 protein in cytoplasmic granules, while migrating BAECs display a diffuse surface expression of CXCR4 (Feil and Augustin, 1998).

In addition, in vitro studies with human aortic ECs (HAECs) revealed enhanced CXCR4 surface expression and CXCL12-induced chemotaxis in the presence of VEGF or basic fibroblast growth factor (bFGF) 48 h after stimulation. Notably, interferon (IFN)-γ, lipopolysaccharide (LPS) or CXCL12 did not elevate the surface expression of CXCR4 (Salcedo et al., 1999).

CXCR4 expression in venous and microvascular ECs. Upregulation of CXCR4 surface expression after addition of VEGF and bFGF was also seen in human microvascular ECs and HUVECs (Salcedo et al., 1999, 2003). In contrast, Schutyser et al. do not report changes in CXCR4 mRNA expression in human microvascular ECs after treatment with VEGF, but describe augmented CXCR4 expression (mRNA and protein) after serum starvation and/or hypoxic treatment of microvascular ECs (Schutyser et al., 2007). Further, hypoxia is also an important regulator of the CXCL12/CXCR4 axis in HUVECs by enhancing CXCR4 expression (Ceradini et al., 2004; Jin et al., 2012). In general hypoxia does cause lowering of local pH, and pH changes are also known to occur upon physical exercise and hemodynamic shear stress, as well as in pathological states including cardiac ischemia. In this context Melchionna et al. reported that acidosis decreased CXCR4 surface expression on mouse aortic ECs in vivo and on HUVECs in vitro in a HIF1α-dependent manner (Melchionna et al., 2010).

Notably, mouse microvascular ECs were also shown to augment CXCR4 expression in vitro in response to erythromycin, an anti-inflammatory antibiotic drug used for treatment of chronic inflammatory diseases. The authors assume that beneficial effects of erythromycin are partly due to CXCR4-expressing ECs recruited to sites of tissue injury (Takagi et al., 2009). However, since microvascular ECs also comprise a broad variety of ECs, for example of dermal, brain, heart or pancreatic islets origin, it remains elusive how any differential regulation of CXCR4 expression described above would impact on atherosclerotic plaque development.

Role of CXCR4 in ECs? Concerning possibly relevant functions of endothelial CXCR4 signaling in context of atherosclerosis, several putatively athero-relevant findings were described in HUVECs, but confirmations in arterial ECs are pending. For example, laminar shear stress suppresses CXCR4 expression in HUVECs while low shear stress favors CXCR4 expression, subsequently resulting in increased EC apoptosis and CCL2 and CXCL8 release (Melchionna et al., 2005).

Another study revealed enhanced release of CXCL12 by HUVECs after oxLDL treatment and a subsequently increased migratory and adhesive response of MSCs (Li et al., 2010). It was further demonstrated that MIF facilitates leukocyte rolling on stimulated HUVECs while siRNA-mediated knock-down of endothelial MIF resulted in decreased expression of E-selectin, ICAM-1, vascular cell adhesion molecule (VCAM) 1, CXCL8, and CCL2 (Cheng et al., 2010).

In addition, it seems that not only VEGF may regulate CXCR4 expression, but it was also shown that CXCL12 treatment increased VEGF protein expression in human microvascular ECs after serum starvation (Saxena et al., 2008), underlining the role of CXCL12 in angiogenesis. Given the CXCL12/CXCR4 axis to be angiogenic in general (Ara et al., 2005; Unoki et al., 2010), but also in tumor development (Domanska et al., 2013) and potentially in atherosclerotic lesions (Di Stefano et al., 2009), and angiogenesis being considered to increase plaque vulnerability, the question still remains if this reflects an important unfavorable role of CXCR4 in atherosclerosis. In a different approach, blocking of TLR2 resulted in increased angiogenic capacity of HUVECs, putatively mediated via association of TLR2 with CXCR4 and subsequently enhanced CXCR4 signaling. Consequently, knock-down of CXCR4 in the presence of TLR2 blocking antibodies revealed less angiogenesis. From these data the authors conclude that TLR2 blocking might serve as a promising therapeutic approach in e.g., MI or peripheral artery disease by promoting revascularization. However, considering angiogenesis pro-atherogenic, TLR2 blocking might have detrimental consequences in the context of plaque stability and development, as already mentioned above (Wagner et al., 2013). Nevertheless, it was also shown that the inflammatory mediators IFN-γ and TNF-α decrease CXCR4 and CXCL12 expression in HUVECs, thereby decreasing their angiogenic capacity (Gupta et al., 1998; Salvucci et al., 2004). Interestingly, the latter contradict findings from Salcedo et al, who showed no difference in CXCR4 expression in HAECs after IFN-γ treatment (Salcedo et al., 1999). Yet, it should be emphasized again that HUVECs and HAECs might exert totally different responses in the presence of the same stimulus, underlining again the importance of caution in generalizing findings from ECs of different origin.

VSMCs are highly specialized cells controlling contraction and regulation of blood vessel diameter, blood pressure, and blood flow. Moreover, VSMCs play a critical role in secretion of extracellular matrix components, which determine the mechanical properties of mature blood vessels. Differentiated VSMCs in adult blood vessels proliferate at very low rates and retain high plasticity, which enables changes in phenotype, referred to as phenotypic switching (Owens et al., 2004). Phenotypic switching of VSMCs is considered an important pathophysiological mechanism in atherosclerosis (Gomez and Owens, 2012).

CXCR4 expression in vascular SMCs. Not much is known about the expression and function of CXCR4 in mature VSMCs. Several studies reported no CXCR4 expression on human or bovine (aortic) SMCs (Gupta et al., 1998; Volin et al., 1998). In contrast, Schecter et al. were the first to claim a functional CXCR4 expression on human aortic SMCs, as the addition of CXCL12 or envelope proteins of HIV specifically binding CXCR4 did induce tissue factor activity in human aortic SMCs (Schecter et al., 2001). Similarly, it was demonstrated that HIV can infect arterial (lesional) human SMCs and blocking CXCR4 in human aortic SMCs in vitro did significantly reduce their viral load. Again, direct expression of CXCR4 on aortic SMCs was not investigated. Yet, the authors still claim that HIV infection of VSMCs through CXCR4 may be one reason why HIV patients are more susceptible to develop atherosclerosis (Eugenin et al., 2008). Nevertheless, Li et al. revealed CXCR4 protein expression on human saphenous vein SMCs (Li et al., 2009) and others later reported CXCR4 expression (RNA, protein) on mouse medial SMCs (Nemenoff et al., 2008), rat aortic SMCs (Jie et al., 2010; Pan et al., 2012) and human aortic SMCs (Weber et al., unpublished data).

Role of CXCR4 in vascular SMCs? As mentioned before, phenotypic switching of VSMCs from a contractile to a synthetic secretory phenotype and their assumed migration from the medial to the intimal arterial wall, where they secrete pro-inflammatory mediators, is considered a pathophysiological mechanism in atherogenesis (Gomez and Owens, 2012). On the other hand, intimal SMCs do also stabilize atherosclerotic lesions by fibrous cap formation. This contrasts the pathology of injury-induced restenosis and neointimal hyperplasia, which are mainly driven by SMC proliferation. Hence, the contribution of SMCs to lesion formation is strongly context-dependent.

In a rat model of diabetes, a metabolic disorder associated with a higher prevalence of atherosclerosis, high glucose levels were shown to trigger activation, proliferation, and enhanced chemotaxis of VSMCs via stimulation of the CXCL12/CXCR4 axis (Jie et al., 2010). Correspondingly, salvianolic acid B (Salvia miltiorrhiza), used to treat cardiovascular diseases in traditional Chinese medicine, was shown to inhibit CXCL12/CXCR4-mediated proliferation, migration and subsequently neointimal hyperplasia by VSMCs in a balloon angioplasty-induced neointima formation model in rats. Here salvianolic acid B directly decreased surface expression of CXCR4 on aortic rat SMCs (Pan et al., 2012).

Further, lesion reduction in Mif^−/−Apoe^−/− mice was attributed to a reduction in lesional SMC proliferation, cysteine protease expression, and elastinolytic and collagenolytic activities (Pan et al., 2004). Notably, oxLDL, supposed to be an important trigger of atherogenesis and plaque growth, was shown to induce rat aortic SMC proliferation. This effect could even be further enhanced by addition of CXCL12 and came along with diminished SMC apoptosis. It remains open if this effect would be beneficial through plaque stabilization, or detrimental because of intimal hyperplasia, as discussed above (Li et al., 2013). As mentioned before, vein graft failure after bypass grafting is a major problem and may be associated with neointimal hyperplasia and accelerated atherosclerosis. In this context, Zhang et al. demonstrate that CXCL12/CXCR4 signaling might be a crucial step in vein graft atherosclerosis and contribute to SMC-mediated vein graft neointimal hyperplasia in mice. Furthermore, CXCR4-mediated recruitment of inflammatory (progenitor) cells to the vein graft may add to this picture (Zhang et al., 2012).

Also, it was described that CXCL12 stimulates pro-MMP2 expression in human aortic SMCs via CXCR4 in association with the epidermal growth factor receptor in vitro. The authors conclude that CXCR4 expands its signaling repertoire by cross-talking with other receptors, pointing at an important role of ligands engaged in receptor cross-talk as critical players in CAD (Kodali et al., 2006).