Non-pharmacological interventions for vascular health and the role of the endothelium

The vascular endothelium plays a central role in vessel function as the interface between the vessel and circulating blood, exemplified by the capacity for endothelial cells to respond to changes in blood flow. Early research during the 1980s identified the essential function of the endothelium in driving vasodilation, initially through endothelial ablation studies within canines (Pohl et al. 1986). Concurrently, in vitro research began to identify the functional components of the endothelium that drive relaxation of the smooth muscle cell layer and subsequent vasodilation. This included the identification and functional relevance of key vasoactive compounds, including nitric oxide (NO) (Palmer et al. 1987), prostacyclin (PGI₂) (Moncada et al. 1976), and endothelin-1 (Yanagisawa et al. 1988). Alongside these vasoactive compounds, the endothelium was also shown to respond to shear stress through the opening of ion channels and subsequent hyperpolarization (Olesen et al. 1988), a key component in endothelium-dependent hyperpolarization, leading to subsequent hyperpolarization and relaxation of smooth muscle (Garland and Dora 2017). Subsequent research has identified a highly complex system of mechanosensitive proteins within the endothelium, that can detect the haemodynamic forces of shear stress and cyclic strain, and modulate responses to flow through mechanotransduction pathways in a force-dependent manner.

Among the first proteins identified to be playing a role as mechanosensors were members of the glycocalyx glycoprotein layer that lines the internal surface of vessels at the interface between the blood and endothelial cell layer. These proteins were shown to respond to flow through the transduction of force to the cell membrane via glypican coreceptors or to the cytoskeleton via syndecan coreceptors (Baratchi et al. 2017). Numerous further mechanosensory proteins have since been shown to demonstrate an essential role in mechanotransduction, both in the detection of forces exerted on the cell and in the transduction of force from force sensors such as those within the glycocalyx. These include the VE-cadherin, platelet endothelial cell adhesion molecule 1 (PECAM-1) and vascular endothelial growth factor receptor 2 and 3 (VEGFR2/3) (VE-cadherin–PECAM-1–VEGFR2/3) mechanosensory complex (Tzima et al. 2005), caveolar proteins (Ariotti and Parton 2013), mechanosensitive ion channels (Baratchi et al. 2017), integrins (Janoštiak et al. 2014), and extracellular matrix component proteins (Fig. 2) (Baratchi et al. 2017).

While blood flow in vivo produces simultaneous shear stress and cyclic strain forces within blood vessels, these two forces are often addressed separately within in vitro research. Much of the research investigating the endothelial responses to mechanical force has been conducted using shear stress alone, due to the greater ease with which shear stress can be manipulated by the exposure of endothelial cells grown on a non-flexible surface to media under flow. As depicted in Fig. 2, at the onset of shear stress, the mechanical force exerted on the glycocalyx layer and other luminal mechanosensors activates the mitogen-activated protein kinase (MAPK) extracellular signal-related kinase 5 (ERK5), via MAP/ERK kinase 5 (MEK5) (Le et al. 2017). Simultaneously, elevated flow appears to drive shear-related responses within calveolae located within the endothelial membrane, through the activation of calveolin proteins, in particular calveolin-1 (Cav-1) (Sowa 2012). Both of these pathways have also been shown to play a key role in regulating responses in mechanosensitive ion channels and as such the potential for shear stress to drive calcium-dependent endothelial hyperpolarisation and subsequent activation of vasodilatory signalling mechanisms (Baratchi et al. 2017; Behringer 2017). Among these mechanosensitive ion channels, the transient receptor potential cation channel V4 (TRPV4) and Piezo1 have been identified as key mechanosensors and form the simplest mechanism for force-detection within the endothelium. Both Piezo1 and TRPV4 have been shown to increase endothelial permeability upon initiation of shear stress and appear to both play an essential role in subsequent calcium influx (Gerhold and Schwartz 2016). Recent research has also shown that TRPV4 is activated following the activation of Piezo1 (Swain and Liddle 2021) and is enhanced by activation and co-localisation with Cav-1 (Lu et al. 2017), suggesting that TRPV4 may play a central role in calcium-dependent flow responses.

Alongside the activation of luminal membrane proteins, flow-induced shear stress has also been shown to drive endothelial responses through the activation of adhesion molecules involved in cell–cell and cell–extracellular matrix (ECM) interactions. The most well studied of these pathways is the VE-cadherin–PECAM-1–VEGFR2/3 mechanosensory complex comprising of VE-cadherin, platelet endothelial cell adhesion molecule 1 (PECAM-1), and vascular endothelial growth factor receptor 2 and 3 (VEGFR2/3), which was first identified by Tzima et al. (2005). In response to increased shear, VE-cadherin transfers tension to PECAM-1 and facilitates the association between PECAM-1 and VEGFR2/3 (Conway et al. 2013). This leads to the activation of signalling cascades through phosphoinositide 3-kinase (PI3K) (Jin et al. 2003), integrin-associated pathways (Hahn and Schwartz 2009), and subsequently altering gene expression (Fig. 2; see “Flow-independent adaptation in vivo”). The essential role that this complex plays in mechanotransduction was demonstrated initially in bovine aortic endothelial cells (BAEC); in which loss of VE-cadherin or PECAM-1 resulted in loss of PI3K responses (Tzima et al. 2005). Subsequently, similar PI3K responses under flow have also been seen in human umbilical vein endothelial cells (HUVEC) under the inhibition of all three components within the VE-cadherin–PECAM-1–VEGFR2/3 complex (Jin et al. 2003; Coon et al. 2015; Russell-Puleri et al. 2017).

Evidence also suggests that shear stress can be detected through conformational changes in the extracellular matrix and ECM–cell interactions. However, this is less well understood, in large part due to the high level of complexity and regional specificity (Wheeler et al. 2014) that likely enables highly specific responses in different vascular beds (Baratchi et al. 2017). Although much research is still needed to understand the scope and role of the ECM in flow responses, integrins appear to play a pivotal role in this mechanosensory mechanism. Specifically, integrins mediate interactions between the ECM and mechanosensory properties of PECAM-1 (Collins et al. 2014), and enable ECM-dependent mechanotransduction and subsequent signalling pathways via the p130Cas scaffold complex (Janoštiak et al. 2014). First identified by Yamada et al. (2006), the p130Cas scaffold complex has been shown to associate with cell membrane integrins and other focal adhesions (Defilippi et al. 2006; Baratchi et al. 2017), through which changes in ECM conformation can drive phosphorylation of the CAS substrate domain and the induction of ERK and other signalling pathways (reviewed in Janoštiak et al. 2014), and subsequent changes in gene expression (see “Flow-independent adaptation in vivo”). This unique mechanosensory pathway has the potential to alter flow-dependent responses across different regions, due to the complexity and variability in ECM composition across the vasculature (Collins et al. 2014), and warrants further investigation, particularly when looking to understand flow responses within a specific vascular bed.

As mentioned, the majority of our understanding surrounding these mechanosensory pathways come from the application of shear stress. However, growing evidence indicates that cyclic strain has a unique impact on endothelial responses to mechanical force (Meza et al. 2019). Cyclic strain in isolation has been shown to activate a number of mechanosensory pathways, through stretch-activated ion channels, ECM–cell adhesions (via the activation of integrins), and cell–cell interactions [via the VE-cadherin–PECAM-1–VEGFR2/3 mechanosensory complex (Jufri et al. 2015)]. Collectively, these mechanisms enable cyclic strain to induce significant signalling responses and have been shown to induce changes in cell morphology and orientation in the absence of shear stress (Bernardi et al. 2018). When investigated in combination with shear stress, the two haemodynamic forces have an additive effect on mechanosensory signalling pathways, although this response is not fully synergistic in nature, most likely due to the overlap in the pathways responsible for detecting shear and strain (Meza et al. 2019). Furthermore, the interaction between these forces becomes more complex when considering the relative strength and dynamics of each force and the structural variability seen across the vasculature (Gray and Stroka 2017).

The endothelial response to changes in blood flow, and the associated mechanosensory response to changes in shear stress and cyclic strain, is highly dependent on the nature of the flow change itself (i.e., the pattern of flow) both in terms of acute signalling responses and subsequent adaptive changes. While the mechanosensory pathways are activated under different flow profiles and the nature of the relationship between different sensors is not well understood, clear differences are seen in subsequent signal pathway responses between flow patterns (Fig. 3) (Nakajima and Mochizuki 2017). In vitro modelling has been used to identify the nature of these differences, including mimicking in vivo disease states; such as ischemic reperfusion injury, or structural variations that can alter regional flow in vivo; such as stenoses, bifurcations or arched structures (e.g., aortic arch) (Gray and Stroka 2017).

Distinct endothelial phenotypes are seen across the vasculature, as a result of regional and tissue-specific variability in both local microenvironment and the function of the endothelium within different tissues (reviewed in Aird 2007a, b). Broadly speaking this heterogeneity has been observed in endothelial cells from different vascular beds (e.g., arteries vs. veins; Aird 2007b); across different specialised tissue beds (heart, lungs, kidneys, liver, and brain) (Cleuren et al. 2019; Jambusaria et al. 2020); and even in different regions within the same vessels (Simmons et al. 2012). Crucially the plasticity of the endothelium to adapt to changes in stimuli can result in the rapid loss of the in vivo phenotype when the cells are cultured in vitro (Aird 2007b; Kuosmanen et al. 2017). This plasticity can be beneficial in developing cell culture models, particularly where cells from relatively accessible sources, such as HUVEC isolated from umbilical cord, can be manipulated to express phenotypes akin to those found in less-accessible vascular beds. This is seen in HUVEC-based models of arterial flow, where arterial levels of shear stress (approx. 2.0 Pa or 20 dyn/cm³), of either a constant or pulsatile nature (Fig. 4), induce a well-characterised pattern of response (discussed below) comparable to those seen in BAEC (Conway et al. 2013) and other arterial cell types (Luu et al. 2010). Conversely, while plasticity enhances our ability to induce a particular phenotype through the manipulation of culture conditions, it also makes the maintenance of in vivo phenotypes in isolated cells or the modelling of specific vascular beds more complicated (Aird 2007b). For example, in modelling of the blood–brain barrier, researchers must consider the selection of appropriate endothelial cell sources; the essential role of astrocytes, pericytes, and other cells of the neurovascular unit; alongside appropriate simulation of the microenvironment on both sides of the barrier (Sivandzade and Cucullo 2018). To date, the majority of studies investigating flow and intervention responses have focused on a relatively small pool of endothelial cell types and generally under linear, arterial flow. While this is in keeping with the majority of human in vivo research, future considerations (see “Mechanisms driving vascular adaptation”) must look to expand the field in keeping with the advancement in technology facilitating the measurement of flow outside of conduit arteries (Baratchi et al. 2017).

Continuous elevation of flow induces rapid responses in the expression of key pro-quiescent transcription factors (Fig. 3a), most prominently krüppel-like factor 2 and 4 (KLF2 and KLF4) and nuclear-factor E2-related factor 2 (Nrf2) via the ERK5 signalling pathway (Fledderus et al. 2008). While elevation in the activity of nuclear-factor kappa-B (NF-κB) and the flow-responsive transcriptional co-regulator YAP, key factors in pro-inflammatory pathways, are also seen under continuous flow, this is highly transient and is followed by rapid reduction in activity over time (Nakajima and Mochizuki 2017). Combined with the rapid activation of p130Cas signalling pathways through the VE-cadherin–PECAM-1–VEGFR2/3 complex and ECM–cell-associated integrins, KLF2 mediates changes in vascular tone through the upregulation of endothelial nitric oxide (NO) synthase (eNOS) (Baratchi et al. 2017). Simultaneously, these signalling responses have been linked to upregulation of anti-inflammatory factors; including eNOS and thrombomodulin (Lin et al. 2005); suppression of inflammatory molecules; including intercellular cell adhesion molecule 1 (ICAM-1), VCAM-1 and E-selectin (Baratchi et al. 2017; Ajami et al. 2017); and improvements in redox balance via the Nrf2-antioxidant response (Fledderus et al. 2008).

Contrastingly, oscillatory or disturbed flow patterns (exemplified by repetitive changes from high-to-low flow or anterograde-retrograde flow, Fig. 4) drive relatively small responses in KLF2/4 and Nrf2. Disturbed flow patterns trigger significant elevations in pro-atherosclerotic and pro-inflammatory signalling pathways (Fig. 3b); most prominently via Nf-κB and YAP (Nakajima and Mochizuki 2017). Sustained elevation in both factors under disturbed flow leads to a significantly more-pronounced inflammatory response (Nakajima et al. 2017); including increases in ICAM-1, VCAM-1, E-selectin, and Monocyte Chemoattractant Protein-1 (MCP-1) (Chiu and Chien 2011). This difference in signalling response is driven by differences in mechanosensory activity between laminar and disturbed flow, demonstrated in the contrasting role of PECAM-1 between the two conditions. Under elevated laminar flow, acute increases in PECAM-1 signalling are essential for both acute shear-dependent vasodilatory responses (Bagi et al. 2005; Russell-Puleri et al. 2017) and subsequent vascular remodelling under exposure to chronic stimuli (Tzima et al. 2005). While under pathological and disturbed flow, PECAM-1 plays an essential role in NF-κB activation and downstream over-expression of adhesion molecules (Tzima et al. 2005). Furthermore, in PECAM-1 knock-out mice, disturbed flow has been linked to reductions in vessel diameter (Chen et al. 2010) and atherosclerosis severity (Harry et al. 2008). In combination, the changes induced by prolonged disturbed flow lead to the establishment of a dysfunctional, pro-atherosclerotic endothelium exhibiting elevated levels of inflammation, oxidative stress, and vasoconstriction (Chiu and Chien 2011). While this phenotype is likely typical under chronic, pathological conditions, such as atherosclerosis, a lack of current research makes understanding the impact of shorter periods of oscillatory flow difficult. Given the positive impact that pneumatic compression can have on vasodilatory function, it is likely that shorter, transient periods of flow of this nature have a different impact on endothelial response. This makes comparison between in vitro research and the potential mechanisms driving responses to intervention difficult (Table 1) and warrants further research into the impact of transient exposure to oscillatory flow.

Less well-studied in vitro is the endothelial response to changes in flow pattern in cultures grown under flow conditions and how short, transient changes may lead to subsequent adaptive changes in endothelial function. These transient changes in flow are a key focal point within in vivo human research and a central tenant of the mechanisms driving adaptations to a range of interventions including exercise (Green et al. 2017), passive heating (Cheng et al. 2019), and pneumatic compression (Credeur et al. 2019) (see “Ischemic preconditioning”). While limited research has been conducted to date, in vitro research has shown the potential for acute changes in flow to stimulate beneficial signalling pathways within the endothelium. In endothelial cells preconditioned to arterial levels of shear stress, further elevations in shear have been shown to stimulate increases in the expression of a range of genes including KLF2 and KLF4, alongside the upregulation of eNOS activity and the promotion of an anti-inflammatory and atheroprotective phenotype (Zhang and Friedman 2012). Furthermore, elevation in the frequency of pulsatile flow, as would be seen during interventions with the capacity to elevate heart rate, leads to increases in the expression of angiogenic and cell proliferation genes, as well as upregulation of VEGF (Zhang and Friedman 2013). These studies demonstrate that elevations or changes in flow conditions can drive significant mechanosensitive response, whether in cells previously cultured under static or continuous flow conditions. Additionally, they provide a potential mechanism for the vascular changes seen under interventions that drive continuous flow elevations (Table 1), and align with the improvements in NO-mediated vasodilatory function that are seen following interventions such as passive heating and exercise (Green et al. 2017; Cheng et al. 2019). The capacity for increases in both shear rate and pulsatile frequency to induce changes in endothelial responses also demonstrates the complex nature of flow responses within the endothelium and the need for specific models of intervention responses. Future research must also consider the impact of the duration of flow changes on subsequent responses, as most interventions induce changes in the magnitude and frequency of pulsatile flow, over a relatively short time-course and often in a non-continuous manner (see Fig. 1).

While most prominently studied from the perspective of disease, research has been conducted into the impact that reduction or loss of flow during ischemic reperfusion can have on previously perfused cells in vitro. Under ischemic reperfusion, the endothelium sees a rapid elevation in ROS and suppression of vasodilation via eNOS uncoupling, coupled with a transition into a more pro-thrombotic and pro-inflammatory state, via the upregulation of hypoxia-inducible factor 1 (HIF-1α) and NF-κΒ (Hernández-Reséndiz et al. 2018). While ischemic events in vivo consist of both a loss or significant reduction of flow and exposure to a hypoxic environment, cessation of flow alone has been shown to drive significant signalling responses including the activation of NADPH oxidase 2 (NOX2) and subsequent production of ROS, the induction of pro-inflammatory transcription factors (including HIF-1α and NF-κΒ) (Chatterjee 2018) and the downregulation of KLF2 (Matharu et al. 2008). Although prolonged, uncontrolled periods of ischemia can have severe pathological complications, including cellular dysfunction, cardiovascular complications and death, shorter controlled periods of ischemia in the form of ischemic preconditioning are thought to elicit beneficial adaptations. These adaptations include improving vascular function within both the periphery and cardiac circulation and improving outcomes in future ischemic reperfusion injury (Tapuria et al. 2008). It has been suggested that these benefits result from the lower level of cellular distress caused by short bouts of ischemia, which are insufficient to impact on cell viability and death, but capable of driving adaptive responses to condition the endothelium against the potential damage of further ischemic events (Kalogeris et al. 2012; Hernández-Reséndiz et al. 2018). This is seen in the time-course of the inflammatory responses seen following cessation of flow; as assessed by neutrophil adherence under TNFα exposure, in which relatively little change is seen over the first 30 min of flow cessation (Matharu et al. 2008). However, further research is needed to formally address endothelial responses to brief periods of ischemia, as used in IPC interventions, as well as in relation to hypoxia and loss of flow both separately and in combination.

Alongside the impact that environmental manipulation (e.g., hypoxic conditioning or passive heating) can have on blood flow (see “Ischemic preconditioning”), environment-based interventions also induce significant changes in the internal environment within blood vessels and the endothelium (e.g., temperature, oxygen tension/availability). In the case of ischemic preconditioning, much of our understanding regarding the impact that changes in temperature or oxygen availability have on the endothelium comes from research focused on pathological conditions. Interestingly, in vitro endothelial cells respond to hypoxia by elevating signalling via NF-κB and HIF-1α (D’Ignazio and Rocha 2016) in a similar manner to the responses seen under ischemic conditions (Hernández-Reséndiz et al. 2018), suggesting similar mechanisms of response whether oxygen is reduced via ischemia or environmental manipulation. In vivo research has demonstrated clear differences between moderate (9–16% inspired O₂) and extreme (2–8% inspired O₂) hypoxic “doses”; including beneficial vs. detrimental impact on blood pressure and systemic inflammation (Navarrete-Opazo and Mitchell 2014), akin to the differences seen between transient and sustained ischemic reperfusion (Kalogeris et al. 2012). In vitro, similar differences were seen when HUVEC were incubated at 0, 5, and 10% O₂, across 6-, 12-, and 24-h periods, with elevations in HIF-1α and NK-κB, subsequent increases in ICAM-1, VEGF and reductions in cell viability occurring in a time- and O₂ concentration-dependent manner (Baldea et al. 2018). While this study shows the potential impact of dose and exposure duration, the majority of in vitro studies have focused on extreme hypoxia reflective of pathological conditions, and more refined, dose-dependent response experiments are needed to understand the potential beneficial and detrimental effects of these differences at the cellular level (Pavlacky and Polak 2020).

Similarly, heat stress can alter endothelial signalling pathways and drive adaptive responses beyond those induced by changes in flow alone. The impact of hyperthermia on endothelial cells appears to be highly dependent on the temperature and conditions of the exposure. In addition to increases in eNOS expression and NO release (Harris et al. 2003), severe hyperthermia (≥ 41 °C) also triggers upregulation in adhesion molecules such as ICAM-1 (Lefor et al. 1994), and induction of endothelial apoptosis via a p53-dependent mechanism (Gu et al. 2014). This response appears to depend on the severity of the hyperthermic stimulus, as febrile temperatures (38–40 °C) drive little basal response in inflammatory cytokines (including IL-1Β, IL-6 and IL-8) or adhesion molecules (including ICAM-1, VCAM-1 and PECAM-1) (Shah et al. 2002). This difference in basal response may also relate to the anti-inflammatory effects febrile temperatures have on the endothelium, such as a reduction in TNFα-induced IL-6 secretion (Hasday et al. 2001). The duration of exposure to heat stress also appears to alter the severity and nature of endothelial responses, with little apoptotic cell death seen following 90 min at 42 °C (Liu et al. 2015) and a reduction in TNFα-induced cell adhesion molecule expression after 60-min at 42 °C compared to normothermic controls (Nakabe et al. 2007). NF-κB is thought to play a key role in regulating these responses to hyperthermia, both during exposure to heat stress and in the subsequent impact on endothelial function post-hyperthermia. Hyperthermia stimulates increases in NF-κB to protect against heat-induced apoptosis Liu et al. (2015), with NF-κB levels subsequently decreased following hypoxic-reoxygenation stimuli (Brunt et al. 2018) and were linked to reductions in adhesion molecule responses to TNFα treatment (Nakabe et al. 2007). It is possible that the divergent NF-kB responses to heat stress are analogous to those seen in response to changes in shear stress (see “Flow-independent adaptation in vivo”). Under this hypothesis, shorter or febrile temperature exposure may induce transient elevations in NF-κB that effectively “primes” the endothelium and promotes an anti-inflammatory response, while prolonged exposure to high temperatures leads to continuous elevation in NF-κB and promotion of a pro-inflammatory, apoptotic phenotype. Future research should examine this hypothesis and determine the role that magnitude and duration play in hyperthermic responses. This is particularly important if we wish to interpret the impact of heat-based interventions, as typically small changes in core temperature (~ 1–2 °C) are induced for short periods, while most in vitro research to date has focused on longer exposures above 40 °C.

To summarise, across the wide range of unique intervention strategies utilised for vascular health, clear similarities can be seen in the capacity for dose-dependent responses within the endothelium, whether the dose response is related to magnitude of flow change, duration of stimulus or the level of deviation from “normal”/ “healthy” conditions. These response patterns highlight the importance of specifically designing experiments to study the impact of interventions on the endothelium, rather than relying on the re-interpretation of findings from dysfunction or disease-based models.

Adding further complexity to our understanding of interventions and their effects on the endothelium is the manner in which different strategies may alter the circulating environment; specifically in the form of changes in circulating signalling molecules and vasoactive substances occurring in vivo. These changes can be: (1) induced locally by the endothelium itself in response to changes in flow; particularly in relation to the release of vasoactive substances and inflammatory cytokines; (2) produced by other tissues and released into the circulation, or (3) induced in an endocrine manner within the endothelium as a positive feedback loop response to circulating signalling factors released from other tissues.

In vivo the endothelium is rarely impacted in isolation and more commonly changes in flow are driven alongside changes in both proximal and distal tissues, making the role of circulating signalling factors central to understanding the holistic impact of non-pharmacological interventions. The importance of changes in circulating factors is seen across a range of interventions and is best evidenced in exercise and passive heating, which despite their differences in aetiology (e.g., active recruitment of muscle contraction vs passive exposure to a heat stimulus), both produce marked elevations in core temperature and blood flow (Naylor et al. 2011; Francois and Thomas 2016). Despite the methodological contrast between these two approaches, similar responses are seen not only in haemodynamic responses (see “Ischemic preconditioning”), but also in vascular biomarkers including interleukin-6 (IL-6), VEGF, and HSP72 (Henstridge et al. 2016; Hoekstra et al. 2018; Akerman et al. 2019). The similarities in the known acute response to these two strategies have led to suggestions that they may induce similar vascular benefits to one another, as a result of comparable elevations in heart rate and increases in both core and skeletal muscle temperature during acute bouts (reviewed in Hoekstra et al. 2020). Interestingly, acute circulating factor responses are also seen during hypoxic stimuli (Table 2), including elevations in both HSP72 and VEGF (Taylor et al. 2010; Burki and Tetenta 2014). These complementary acute responses across the three intervention strategies are matched in their responses to repeated exposure, most notably in the reduction in circulating markers of inflammation across exercise training (Palmefors et al. 2014), hot water bathing (Hoekstra et al. 2020), and dose-dependently in intermittent hypoxia (Navarrete-Opazo and Mitchell 2014).

One possible explanation for the similarities in both acute and chronic responses to a wide variety of intervention strategies is the central role that hypoxia-inducible factor 1α (HIF-1α) plays in angiogenesis and vascular function. In vivo human research has demonstrated that both acute and chronic exposure to heat or hypoxia results in similar elevations in extracellular HIF-1α (Lee et al. 2016), while elevations in skeletal muscle due to localised hypoxia during exercise have been suggested to play an essential role in adaptation to training (Lindholm and Rundqvist 2016). Although it is not possible to elucidate the impact that changes in HIF-1α have on the endothelium directly in vivo, HIF-1α plays a clear role in endothelial function in vitro. Indeed, the HIF-family of proteins have been highlighted as potential master regulators for angiogenesis during vascular development (Hashimoto and Shibasaki 2015), driving elevations in NO production via eNOS; the release of pro-angiogenic factors such as VEGF, and activation of inflammatory pathways via increased interleukin release (Krock et al. 2011). Research in rodent models has also shown that hypoxia-inducible factors are essential to the acute inflammatory response (including IL-4, IL-6, and IL-10) following ischemic preconditioning (Cai et al. 2013; Yang et al. 2018) and the subsequent beneficial signalling responses via Akt (Du et al. 2020), further demonstrates the central role that HIF signalling plays. The similarities and potential role of hypoxia-inducible factors across different intervention stimuli may be pivotal in the cross-over seen between many of the interventions discussed within this review, such as the cross-adaptation following heat acclimation interventions driving improvement in exercise tolerance in hot and hypoxic conditions (Lee et al. 2016; Gibson et al. 2017).

This is not to say that changes in the circulating environment are not highly dependent on the specific parameters of a given intervention, in much the same manner as is seen in the responses to flow between interventions. This is demonstrated in the variable responses to changing parameters within a single intervention type, such as the protocol-dependent responses in circulating IL-6, TNFα, VEGF, and other factors seen following exercise-based interventions (Palmefors et al. 2014; Ramos et al. 2015) and the temperature-dependent response in HSP70 seen following active heat acclimation (Lovell et al. 2007). The complexity of the circulating environment makes determining the full scope of a given intervention’s impact very difficult, let alone the potential variations in response within and between different intervention paradigms. This is particularly true in relation to the current approach taken to understanding the circulating environment. This is often done by targeting specific biomarkers of response (e.g., VEGF, HSP’s, HIF-1α, etc.), which by their very nature are likely to display similar responses to interventions that have been selected specifically for their ability to stimulate vascular responses. To address this dilemma, future research must look to expand the breadth of targets investigated within the circulation, as can be achieved through the utilisation of omics approaches that allow for the full scope of responses within the circulation to be determined (Hoffman 2017).

Our understanding of the impact that many circulating factors have on the endothelium at the cellular level suffers from a similar issue, in that the molecular response to many factors is well understood in isolation and in relation to pathological conditions. However, our knowledge of the cellular response to the collective changes seen during non-pharmacological interventions is much less developed. This is exemplified in our understanding of the impact that inflammatory cytokines have on the endothelium, which have been studied and reviewed at length in relation to vascular dysfunction and disease (Sprague and Khalil 2009). Elevated inflammatory cytokine levels have been shown to play a central role in endothelial dysfunction through disruption of vasodilatory mechanisms, sustained upregulation of angiogenic factors and activation of NF-κB signalling pathways (Sprague and Khalil 2009). Additionally, prolonged periods of cytokine elevation are thought to promote the development of hypertension (Pioli and de Faria 2019), atherosclerosis (Tedgui and Mallat 2006), and chronic vascular dysfunction (Steyers and Miller 2014). When we look to interpret the potential role that these mechanisms play within interventions, similarities can be drawn in the transient response to interventions such as exercise and passive heat exposure (Kasapis and Thompson 2005; Hoekstra et al. 2018); in which significant elevations in inflammatory cytokines likely elicit similar signalling responses to those seen in inflammatory disease or dysfunction. However, over repeated exposure to these strategies a reduction is seen in basal inflammatory cytokine profiles (Gleeson et al. 2011; Hoekstra et al. 2020), in stark contrast to the response seen in dysfunction. This is likely due to the transient nature of exposure during interventions compared to chronic elevations seen in pathological states. It has been suggested that this difference is key to the anti-inflammatory benefits of a number of interventions (Hoekstra et al. 2020), in a similar manner to the transient periods of reduced flow seen in ischemic preconditioning that drives beneficial adaptive responses in vascular function (Kalogeris et al. 2012; Hernández-Reséndiz et al. 2018). This also highlights the importance of considering differences in the nature, magnitude, and duration of changes in the circulating environment during interventions, especially when looking to compare the potential mechanisms driving endothelial responses from pathophysiological models and studies.

The importance of intervention-specific study design is made clear when changes in circulating environment are examined in concert with flow. For example, the endothelial cell response to even a small number of inflammatory cytokines (TNFα and IL-1β) is drastically altered when cells are preconditioned under arterial levels of shear stress (Luu et al. 2010). Furthermore, when the pattern of flow is modulated during exposure to TNFα, a significant difference is seen in the response. For example, significant increases in neutrophil adherence are seen when preconditioned cells are exposed to ischemic flow conditions (Sheikh et al. 2003), demonstrating the potential for markedly different responses dependent on blood flow conditions. Alongside this reciprocal relationship between circulating factors and flow, VEGF has the potential to interact with flow responses through its role in the assembly of the p130Cas complex (Evans et al. 2017) and interaction with other mechanosensory signal pathways via binding to VEGF receptors including VEGFR2/3 (Olsson et al. 2006). Adding further complexity still, alongside the interaction between flow and circulating factors, there is also significant potential for interplay between the factors themselves, this is seen in the case of IGF-1 and VEGF as, like many growth factors, they share common signalling pathways through PI3K–Akt and MEK1/2–ERK1/2, resulting in significant overlap in their signalling responses (Fig. 5) (Olsson et al. 2006; Rivera et al. 2011; Bach 2015). Similarly, heat shock proteins, particularly HSP90 (Fig. 5), have been shown to directly interact with HIF-1α and VEGF signalling; through stabilization of HIF-1α (Minet et al. 1999) and increased VEGFR2 receptor presentation (Bruns et al. 2012). Finally, as has been discussed in relation to differences in response to flow (see “Flow-independent adaptation in vivo”), endothelial heterogeneity can also impact endothelial responses to circulating signalling factors, including variability in the magnitude and scope of VEGF-induced MAPK signalling response across BAEC, HUVEC, and HMVEC cell types (Yashima et al. 2001). Taken together, these interactions further emphasise the importance of understanding the complete spectrum of changes seen within the circulation in vivo as well as the collective effects of these changes when modelling endothelial response in vitro.

The development of specific, well-designed cell culture experiments that incorporate core acute intervention factors/responses and better model the physiological environment of these scenarios is key to identifying how changes in circulating environment and flow responses drive beneficial vascular adaptations.