Treatment of myocardial infarction (MI) focuses on rapidly re-establishing reperfusion following blockage in one or more coronary arteries. Despite successful thrombolytic and primary percutaneous coronary interventions (PCI) a significant number of patients still subsequently proceed to develop muscle damage and heart failure (1–3). This is partly due to reperfusion paradoxically leading to additional tissue damage, a condition termed ischaemia-reperfusion (IR) injury (4, 5). Since blood flow is normally restored in the occluded artery, tissue damage most likely occurs as a consequence of inadequate coronary microcirculatory perfusion (6). Indeed, sub-optimal myocardial perfusion can be observed in as high as 50% of patients, leading to worse outcomes than in patients with full perfusion recovery (7, 8). Increased clinical recognition of the importance of the coronary microcirculation has meant identifying strategies to improve detrimental perturbations within it have gained recent attention (9, 10). However, clinical imaging tools cannot spatially resolve blood vessels <200μm in diameter and so little is known about the full range of coronary microcirculatory responses to IR injury and whether potential strategies are vasculoprotective (11, 12). Moreover, in vivo experimental studies of the beating heart using intravital imaging have also been hampered due to obvious difficulties related to significant inherent contractile motion, movement of the heart brought about by nearby lungs and its location in an anatomically challenging position for microscopy (13).

Recently, we described an intravital microscopy technique that allowed imaging of the coronary microcirculation of the mouse beating heart with cellular resolution in vivo (14, 15). Using this technique, we showed that despite deceptive hyperaemic responses, increased microcirculatory flow heterogeneity and reduced functional capillary density (FCD) was observed post-reperfusion. This was primarily linked to marked neutrophil infiltration and occlusive platelet aggregate formation within coronary capillaries which consequently impacted infarct size. Therefore, therapeutic strategies that can ameliorate these perturbations and restore microcirculatory perfusion in the immediate aftermath of reperfusion are essential if salvaging viable myocardium is to be maximized.

Mesenchymal stromal cells (MSCs) have been at the forefront as candidates to treat cardiovascular diseases (CVDs). (16) This has been due to the ease with which they can be isolated and expanded in culture but mainly because of their potent anti-inflammatory and immunomodulatory capabilities. However, despite evidence that MSCs are beneficial in experimental models of myocardial IR injury, clinical trials have either failed to replicate these findings or have shown minor or temporary benefit. Therapeutic success is likely dependent on the ability of systemically injected cells to successfully migrate or home to the ischaemic heart and become retained by the local coronary microcirculation (17, 18). However, myocardial MSC homing is inefficient which may prevent their full therapeutic potential from being realised, necessitating the need to develop strategies to improve their recruitment (17, 18). Limited therapeutic success may also be attributable to increasing evidence demonstrating a significant heterogeneity, both morphologically and functionally in terms of differentiation and migratory capacity, amongst MSCs even if they are derived from the same tissue site (19). Recent studies have also suggested that 3D spheroid culture of MSCs utilising techniques such as non-adherent dishes, cell sheets, micro-fluidics and the hanging drop method, may be necessary to enhance their beneficial properties for CVDs (20, 21). Indeed, 3D cultured MSCs exhibit greater anti-inflammatory potential, stemness, angiogenic properties and survival rates in ischaemic tissue following their systemic administration (20, 22, 23).

Whether there is a difference in the myocardial homing ability and vasculoprotective efficacy of 2D or 3D cultured intra-tissue MSC sub-populations has not previously been investigated in any tissue let alone the heart, but may prove critical to improving outcomes for MI patients in the acute period post-PCI. To investigate this, the current study used two MSC sub-populations, both derived from the bone marrow (BM), with distinct phenotype and functionality and characterised by CD317 expression (24, 25). MSCs-Y201 are CD317^neg and exhibit potent tri-lineage differentiation capabilities. They also have a higher expression set of genes related to interactions with BM vasculature and in playing a role in vascular processes such as blood vessel remodelling, morphogenesis, development and patterning. In contrast, MSCs-Y202 are CD317^pos, poorly or non-differentiating and rarely associate with vascular cells or perivascular stroma (24, 25). Both sub-populations were cultured using conventional 2D and spheroid forming 3D methods and their ability to home to, become retained within, and subsequently vasculoprotect the coronary microcirculation was investigated intravitally in the IR injured mouse beating heart in vivo.

Reperfusion injury post-PCI remains one of the top 10 unmet clinical needs in cardiology (29). Since the smallest coronary blood vessels are increasingly recognised as a primary target of reperfusion injury, therapeutic interventions must be able to protect these delicate structures and maintain adequate perfusion in the heart. We have previously demonstrated, using novel intravital imaging of the anaesthetised mouse beating heart, a number of microcirculatory perturbations post-reperfusion injury including increased neutrophil recruitment, increased microthrombus formation and decreased FCD (14). The current study tested whether cellular therapy using MSCs, known for their potent anti-inflammatory properties, could successfully maintain the viability of the coronary microcirculation in the beating heart in vivo. However, despite this inherent property, clinical success using MSCs for treating various conditions has been limited. Therefore, we investigated intravitally whether we could enhance their potential vasculoprotective efficacy through 3D culture and the testing of different BM-derived MSC sub-populations for their coronary homing and anti-neutrophil/anti-platelet capacity.

MSCs are routinely grown in easy and relatively inexpensive 2D cultures in which they adhere to plastic surfaces, spread, become flat and greatly increase their size. However, infusion of such large cells reduces their ability to circulate beyond a single pass and reach target sites of injury including the heart. Indeed, <5% of infused MSCs reach the injured heart with most becoming entrapped within pulmonary capillaries (30). The current study firstly demonstrated that MSCs cultured using a simple scaffold-free, hanging drop method clumped together and formed compact spheroids that were 50-60% smaller in diameter than MSCs grown using conventional 2D culture. This is consistent with studies in which cellular diameter decreases of up to 70% were observed with 3D culture (31, 32). It was anticipated that after infusion, these smaller 3D cultured MSCs would escape non-specific entrapment and maintain a circulating pool of MSCs, thereby allowing more efficient delivery to the heart (33). However, our novel intravital imaging of the beating heart demonstrated that whilst 3D culture increased numbers of MSCs-Y201 and MSCs-Y202 circulating through the coronary microcirculation, this was were only observed immediately after their infusion and not thereafter. This is consistent with our previous intravital findings in other organs where circulating haematopoietic stem/progenitor cells (HSPCs) were also only observed on ‘first pass’ (34–36). The initial increase immediately after infusion may be due a transient reactive hyperaemic response in the ischaemic ventricle upon reperfusion which may have delivered more MSCs to the heart (14). Additionally, increased circulation may be linked to the recovery of critical stem cell homing receptors, such as CXCR4 which are lost during conventional culture, with 3D culture. This was indeed the case for MSCs-Y201 and has been described by others (37–39). The recruitment of 3D cultured MSCs-Y201 in the heart was very limited and therefore not sufficient to significantly reduce pulmonary presence when compared to the other groups. Indeed, the vast majority of the half a million MSCs that were injected became non-specifically entrapped in the lungs with no cells detected in other non-cardiac organs (data not presented).

MSC aggregation into tightly packed clusters creates an in vivo-like microenvironment which better preserves their phenotype and innate properties (40). Whether this impacted their ability to adhere to endothelial counterligands was tested in vitro and in the beating heart. 3D culture consistently enhanced adhesion of only MSCs-Y201 in static adhesion assays but improved retention of both sub-populations in vivo, albeit in small numbers. Interestingly, 1,731 genes have been identified to be overexpressed in hanging drop cultured human MSCs when compared to monolayer cultured cells, including those encoding critical MSC surface integrins and receptors involved in adhesion (41). Furthermore, similar to our own data, others have also shown the ability of 3D spheroid culture to enhance expression of adhesion integrins on various cell types, including a 3-fold increased expression of the ICAM-1 ligand lymphocyte function associated antigen-1 (LFA-1) on cancer cells and the VCAM-1 ligand very late antigen-4 (VLA-4) on HSPCs (31, 42, 43). In the current study more uniform distribution of adhesion molecules was also noted on 3D cultured MSCs which, when noted on leukocytes, has been known to increase the effectiveness of their recruitment within injured sites (44). Collectively, these changes may mechanistically explain the observed increased adhesion of 3D cultured MSCs.

Despite 3D culture benefitting the adhesion of both MSC sub-populations equally in vivo, only 3D cultured MSCs-Y201 modified coronary microcirculatory perturbations. To date, MSCs derived from a single site such as the BM have not been enriched for a particular intra-tissue sub-population prior to their clinical use for CVDs. However, our novel results suggest that a significant barrier to realize the therapeutic potential of MSCs is their intrinsic heterogeneity despite their same origin and not necessarily their local coronary retention, which even after 3D culture was limited. An anti-neutrophil effect was significant 45 minutes post-infusion, but only with 3D cultured MSCs-Y201, decreasing neutrophil presence by approximately 30%. Whether there is a difference in the secretory profile of the two tested MSC sub-populations warrants further detailed investigation. However, it is well established that MSCs can secrete multiple anti-inflammatory factors which can suppress an overactivated immune system within cytokine-rich environments and thereby attenuate neutrophil infiltration. These include interleukin-10, transforming growth factor-β, inducible nitric oxide synthase, prostaglandin-E2 (PGE2) and indoleamine pyrrole 2,3-dioxygenase. MSCs can also suppress ROS secreted by neutrophils which would protect the endothelium with which neutrophils make close adhesive contact (45).

MSCs, as well as being anti-inflammatory, can also exhibit pro-inflammatory properties in certain instances such as to stimulate pathogen clearance (45). Indeed the MSCs used in the current study have been shown to have varying immunomodulatory ability based on the expression of CD317 (24, 25). MSCs-Y202 have been identified as nullipotent CD317^pos cells that make up approximately 30% (CD317^dim ∼28%; CD317^bright ∼2%) of the BM MSC population. These specific MSCs are less effective at suppressing T cell proliferation and more capable of polarising them towards a pro-inflammatory type 1 T-helper (Th1) subset than MSCs-Y201. In contrast, CD317^neg MSCs-Y201, making up 70% of the BM MSC population, polarise T cells towards an anti-inflammatory type 2 T-helper (Th2) subset (25). CD317, also known as tetherin is a transmembrane protein best known for its ability to inhibit the spread of viral infection by ‘tethering’ them to the cell membrane. In addition, tetherin activates nuclear factor kappa B (NF-κB), a transcriptional activator that leads to the rapid expression of pro-inflammatory cytokines, thus contributing to antiviral defence (46). We therefore speculate that when a mixed population of BM-derived MSCs are infused for therapeutic purposes, the pro-inflammatory CD317^pos sub-population may ‘dilute’ the anti-inflammatory benefits of the CD317^neg sub-population and may even be harmful (25). Although this particular marker may molecularly identify a MSC subset with specific vasculoprotective properties in the injured heart, we show that their expansion in 3D cultures also remains equally important.

We have previously demonstrated that capillary occlusion by platelet aggregates is a key contributor to decreased perfusion in the IR injured heart (14). Since 3D cultured MSCs-Y201 did not significantly modify platelet behaviour, we investigated whether co-transplantation with an anti-coagulant could modify both platelet and neutrophil behaviour as previously demonstrated in the colon of colitis mice (47). Heparin was chosen as it is a widely used, safe drug which could be used clinically as an adjuvant if it was noted to improve MSC therapy. We firstly showed that heparin alone led to a gradual but worrying increase in neutrophil recruitment, an effect observed by others in different injuries such as in the peritonitis model (48, 49). It is plausible that heparin did indeed prevent microthrombus formation, unblocked occluded capillaries and thus improved perfusion. This may have consequently increased neutrophil trafficking and therefore their adhesion within a primed coronary microcirculation. However, dual therapy with heparin, whilst not improving the local presence of 3D cultured MSCs-Y201 (data not shown), did accelerate their anti-neutrophil ability, further decreased platelet presence and reduced infarct size more so than single therapy with MSCs alone. In this scenario, we postulate that heparin would have indirectly reduced microthrombus formation and the MSCs would have prevented any IR- and heparin-mediated neutrophil adhesion. Heparin can indirectly modify microthrombi presence through its actions as an anti-coagulant that reduces thrombin activity and consequently fibrin deposition. Lack of fibrin reduces the structural stability, strength and adhesive surfaces provided to platelet aggregates. Furthermore, the lack of thrombin, which is a highly potent platelet agonist, would also lead to reduced platelet activation and aggregation.

It is worth highlighting that the use of a single anti-coagulant therapy versus single cellular therapy versus dual therapy resulted in a 16%, 44% and 52% reduction in infarct size respectively. Although dual therapy did not significantly reduce infarct size statistically compared to cellular therapy alone, the modest salvage of an extra 8% more viable myocardium would have biological significance. Indeed, a 5% increase in infarct size observed clinically is independently associated with a 19% increase in mortality at 1-year (50). Furthermore, Pride and colleagues found clinically that every 5% increase in infarct size was associated with a 6.1% decrease in ventricular ejection fraction which correlated with poorer prognosis (51). Hence, we believe that even seemingly small improvements in infarct size can profoundly impact outcomes even if they fail to reach statistical significance (51, 52). Hence, overall our novel data suggested that 3D culture of the MSC-Y201 sub-population was superior in its ability to reduce inflammatory and thrombotic cell presence, improve flow and decrease infarct size after myocardial IR injury, with heparin co-administration providing further benefit at the level of the coronary microvasculature and infarct size.

Potential mechanisms by which MSCs-Y201 conferred vasculoprotection were also investigated. Oxidative stress following MI is associated with injury to the microvasculature which, through damage and denudation of ECs, can rapidly increase neutrophil and platelet activity. Indeed, we noted a remarkable increase in oxidative injury in 25% of coronary ECs when assessed flow cytometrically. Moreover, the coronary endothelium, rather than cardiomyocytes, appeared to be the primary and initial target of oxidative injury confirming our previous findings (14). Others have also shown this earlier and increased susceptibility of ECs to IR injury when compared to cardiomyocytes, with apoptotic death occurring at 5 and 60 minutes for ECs and cardiomyocytes respectively in isolated perfused hearts (53, 54). Collectively, these studies highlight the coronary microcirculation as an early and principal therapeutic target to prevent ensuing muscle damage. Growing evidence suggests MSCs can indirectly exert anti-oxidant effects through release of superoxide dismutase (SOD), catalase and glutathione peroxidase. For example, MSCs have recently been shown to enhance, rather than release themselves, SOD from cells exposed to oxidative stimuli in vitro and also in murine models of inflamed gut and ageing (55). Preventing EC damage may potentially explain their ability to reduce the recruitment of thromboinflammatory cells. It was therefore exciting to see that all MSC sub-populations could rapidly reduce the burden of oxidative stress on coronary ECs in vivo with MSCs-Y201 being most effective in vitro where they maintained an anti-oxidant effect even after 24 hours of H₂O₂ exposure.

The initial telomere length of the donor’s MSCs could therefore be a limiting factor of the MSCs therapeutic potential since it will affect the survival and integration ability of the trans-planted cells to the adult tissue The initial telomere length of the donor’s MSCs could therefore be a limiting factor of the MSCs therapeutic potential since it will affect the survival and integration ability of the trans-planted cells to the adult tissue.

It is well established that neutrophil adhesion to activated ECs is facilitated by endothelial expression of ICAM-1 and VCAM-1 with expression of the latter noted in res[ponse to oxidative damage (56). We confirmed previous findings that myocardial IR injury significantly increased both on CD31⁺ ECs in IR injured hearts. Interestingly, all MSC groups tested were able to reduce the overexpression of both ICAM-1 and VCAM-1, with greater and significant responses observed for the latter. Again, it seems that an ability to rapidly modify inflammatory markers in vivo appears to be an innate capability of BM-derived MSC subpopulations and is supported by other studies. Although there was no significant difference in the in vivo anti-oxidant and anti-adhesion molecule ability of all 4 MSC groups, 3D cultured MSCs-Y201 consistently demonstrated itself to be the most effective. These multiple beneficial effects at a molecular level may explain why this particular MSC group was able to significantly tame neutrophil recruitment within the injured beating heart in vivo. Similar reductions in endothelial ICAM-1/VCAM-1 have been observed by different groups, albeit using immunostaining and rt-PCR methods. Qui and colleagues showed a reduction of endothelial ICAM-1 expression after human umbilical cord MSC treatment in a renal model of IR injury. Rahbarghazi and colleagues more recently stated that there was limited data on the modulatory effects of MSCs on the expression levels of ICAM-1 and VCAM-1 in lung tissues and so investigated this in rat asthmatic models. They also found BM-MSCS decreased expression of both on pulmonary endothelium (57). In their study, they further tested whether this benefit could be conferred via a paracrine effect by performing parallel experiments using BM-MSC conditioned media. Although conditioned media was effective, the ability to decrease adhesion molecule expression was more evident in the cell-treated group.

2D cultured MSCs-Y201 did not significantly decrease any circulating serum cytokine levels in IR injured mice. It was therefore remarkable that simply culturing these specific cells in hanging drops dramatically converted them into the most effective cellular therapy at reducing pro-inflammatory cytokine levels. Further studies are required to determine how these decreases were elicited. However, preconditioning MSCs through 3D culture has previously been shown to modify their secretome such that there is an increased release of anti-inflammatory factors including PGE2 and TNFα stimulated gene-6 (TSG-6) which could subsequently suppress inflammatory cytokine secretion through paracrine actions on various cells (31, 58). An interesting observation in the current study was the universal ability of all MSC sub-populations to decrease to some degree levels of IL-1α, IL-12(p40) and CCL4/MIP-1β. It is possible therefore that targeting these specific inflammatory cytokines is a critical vasculoprotective mechanism. Although Shin and colleagues recently demonstrated that the secretome profile of MSCs from different sites varied, further investigation is required to determine if this is also the case for MSC sub-populations derived from the same site (59). Whilst endothelial oxidative damage and inflammatory adhesion molecule expression was equally impacted by 2D and 3D-cultured Y201 MSCs, the most striking observation was this differential effect on circulating inflammatory cytokines between 2D and 3D cultured MSCs-Y201. This effect, alongside the ability of 3D cultured MSCs-Y201 to preferentially reduce local neutrophil recruitment and improve FCD, appears to be the most likely explanation underlying their observed cardioprotection rather than direct beneficial effects on the coronary ECs per se. However, myocardial IR injury is a multi-faceted injury and therefore likely benefits from 3D cultured MSCs-Y201 exerting multiple protective mechanisms simultaneously to confer cardioprotection including effects on the microvessels themselves.

Only a limited number of studies have compared the benefits of 2D and 3D cultured MSCs in preclinical studies and currently no trials have yet evaluated their clinical efficacy. One of the major limitations of taking this approach to the clinic is the laborious preparation required to generate 3D spheroids of MSCs. This significantly limits their large-scale production for translational purposes. To circumvent this issue, future studies could consider testing the coronary vasculoprotective efficacy of MSCs-Y201 cultured on ultra low-attachment surfaces. These plates are coated in a way that inhibits immobilisation, forcing cells into a suspended state and thereby enabling spheroid formation. Importantly, they offer a less laborious technique for culturing MSCs as spheroids. Alternatively, increasing our knowledge of the paracrine factors released in the secretome from 3D cultured MSCs-Y201, and specifically identifying and testing proteins with vasculoprotective potential, would also be a worthwhile pursuit. Our study demonstrates that co-administration of heparin improved the vasculoprotective capacity of 3D cultured MSCs-Y201. Hence, future studies could also assess whether dual therapy of MSCs with other pharmacological drugs, such as contemporary anti-platelet drugs already used to treat patients post-PCI, may further improve their beneficial effects at the level of the coronary microcirculation.

To conclude, further improvements to MI patient outcomes will depend on addressing problems in the coronary microcirculation that arise during reperfusion. Therefore, developing strategies specifically directed at restoring microvascular function is of paramount importance. MSC cytotherapy is a promising strategy but clinical success has been limited. We show that 3D culture of BM-derived MSCs, a simple non-invasive and non-genetically engineered approach, improves their vasculoprotective ability without remarkably increasing their local presence in the IR injured heart. Moreover, 3D culture of a specific CD317^neg MSC sub-population, in combination with heparin, could offer a powerful new cellular treatment for improving multiple perturbations in the coronary microvessels post-reperfusion.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by REC 07/Q1105/9; University of York. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study. The animal study was approved by UK Home Office and University of Birmingham Animal Welfare and Ethical Review Body (AWERB). The study was conducted in accordance with the local legislation and institutional requirements.

KB: Formal Analysis, Writing – original draft, Data curation, Methodology. DK: Writing – review & editing. PG: Writing – review & editing, Resources. NK: Writing – review & editing, Conceptualization, Formal Analysis, Funding acquisition, Project administration, Writing – original draft.