Non-endocrine cell types are important to islet function; predominantly endothelial cells which are a component of the network of intra-islet microvasculature which is five times denser than the exocrine pancreas alongside dendritic cells and nerve fibres (46). In human islets endocrine cell types are close to vasculature, in a non-patterned distribution along the vessel course. Each islet receives arterial supply from up to five arterioles which form a glomerulus-like network of capillaries within which endocrine cells are sited, closely located to vessels, with a double basement membrane (BM) between beta-cells and capillary endothelial cells, both of which heavily express lutheran glycoprotein (47, 48). Polarity of beta-cells related to vessels is exhibited on electron microscopy of rat islets, demonstrating how cell-matrix interactions with this BM can influence morphology (49). Islet capillaries demonstrate a thinner endothelium with significantly higher levels of fenestration, with these characteristics highly delineated at the islet-exocrine interface (50).

Native islet ECM producing cells are not well characterised, but evidence is emerging for the role of islet stellate cells (ISCs). Similar, but distinct from pancreatic stellate cell population, which are implicated in pancreatic fibrosis and exocrine disease, they possess a quiescent and activated state in which they are thought responsible for excessive ECM deposition and implicated in Type 2 Diabetes Mellitus (T2DM) (51, 52). However, when isolated ISCs are co-cultured with mouse islets, improved insulin production is seen (53). Others have described a pericyte cell type, with similar properties, capable of activation into a myofibroblast subtype. Such cells have actions on insulin production through control of microvasculature and directly and have also been demonstrated to produce ECM components (54, 55). The proposition that overactivity of these tissue resident fibroblasts impairs islet function in T2DM may draw comparison with the post-transplant microenvironment.

The extra-cellular matrix (ECM) is a complex environment, composed of multiple proteins, glycoproteins and proteoglycans. These networks, which are found across tissue types, can bind with cell surface receptors to influence cell differentiation, survival and function. This can be achieved via integrin receptors and through mechano-sensing of transmitted loads through frequency and intensity of cell-ECM interactions. The cytoplasmic tails of integrin receptors bind with the intra-cellular actin cytoskeleton and activate pathways involved with cell survival and function and allow bi-directional communication with the local micro-environment (56).

The importance of the islet ECM has led to the investigation of the impact of predominant BM components on beta-cell activity. Integrin signalling is important for the development of beta-cells, survival, and functional state. The composition of ECM and integrin expression changes through development and adulthood but in both mice and human developing islets, knock outs of integrin types can lead to reduced islet development and functionality (57, 58). Antibody blockade of integrin α₁ and integrin β₁ subunits reduce cell adhesion and migration of foetal and adult human islets in vitro (59). Culture of foetal islets with collagen IV also promotes insulin function compared to other ECM components and this action is reversed by integrin α₁β₁ antibody blockade (59).

Laminins and glycoproteins including fibronectin (FN) are also thought to impact survival and function. FN is a key component of developing islets and activates integrin α_vβ₁. Blockade with 270 RGD-peptide analogue in immunodeficient mice after renal subcapsular transplant of human foetal pancreas fragments, results in aberrant islet development with loss of conventional structure and a reduction in insulin positive cells (60). Whilst the ability of proteoglycans to bind and regulate availability of growth factors such as fibroblast growth factor, vascular endothelial growth factors (VEGF) and connective tissue growth factor has important implications for beta-cells (61). For example, reduction in expression of VEGF-A in mouse beta-cells causes impaired insulin production with reduction in fenestrations of intra-islet vasculature and reduces revascularisation at 1 month after these cells are transplanted under the kidney capsule of immunodeficient mice (62). FGFs activate the ERK1/2 pathway to increase insulin mRNA (63). Of note, the contextual influence of islet ECM is demonstrated as this effect is modulated by integrin binding to laminin which causes a reduction in FGF receptor expression (64). Appraisal of native islet architecture warrants consideration in comparison to the post-isolation and early-engraftment islet state to appreciate the impact of the cell-cell and cell-matrix post-transplant microenvironment. Developing this area of knowledge specific to islet transplantation will open avenues to improve isolation, culture and engraftment in transplantation to promote survival and function (65).

During islet isolation, native vascularisation and surrounding matrix is destroyed. Significant collagen destruction (60-80%) is required to facilitate successful islet isolation (66). Consequently, islets are distributed into the portal venous system devoid of this supporting framework, increasing their vulnerability to apoptosis and necrosis whilst reliant on diffusion (67). Due to reliance on diffusion and despite placement into the highly vascularised liver site, this is insufficient to sustain islet populations. Cells within the central core of isolated islets demonstrate progressive reduction in viability when cultured in vitro, leading to cell death (68). Interestingly, smaller islets demonstrated improved responses to glucose stimuli in vitro and increased islet index (IEQ/islet number) positively correlates with stimulated C-peptide 3 months following simultaneous islet-kidney transplant (69). Hypoxia is a significant cause of islet failure in emerging encapsulation techniques and devices with smaller pore diameter are not favoured due to the prevention of revascularisation (70).

To achieve function after IT, revascularisation is a necessary requirement for successful islet engraftment. New intra-islet vasculature is derived from both recipient and donor endothelium (71, 72). Angiogenesis begins early after IT, likely from hepatic arterial tree (73) between day 1 and three and completes at approximately day 14 (74) with a following medium to long term period of vascular and extra-cellular matrix remodeling (67). In response to hypoxia, cultured and transplanted islets express VEGF, before an increase in VEGF receptor expression (Flk-1/KDR & Flt-1) between day 3-5 with a similar later rise in VE-Cadherin; these signs of revascularisation are reduced in diabetic small animal models compared to normoglycaemic controls, suggesting that a hyperglycaemic environment is unfavourable for islet survival (75).

Morphological analysis of autopsy specimens in long term functioning islet transplants (>10 years) with positive serum C-peptide has revealed extensive capillary networks positively staining for CD31, with comparable vascular density, size of vessel and distribution to those seen on histological analysis of biopsy tissue from native pancreas (76). Elsewhere, regardless of implantation site, post-transplant intra-islet vascular density is reduced compared to the native pancreas with oxygen tensions, compared to the highly arterialized supply of the native pancreas (77). The mechanisms of revascularisation and neo-angiogenesis may alter between transplant sites.

To achieve successful islet isolation, destruction of the supporting framework of extra-cellular matrix vasculature is required ( Figure 1 ) (66). Collagenase digestion non-selectively targets ECM components which breakdown peripheral pancreatic and intra-islet matrix which is associated with cell death (78). It is logical that reinstating the cell-matrix interactions following IT would be beneficial for islet survival and function.

Native pancreatic ECM is composed of both interstitial matrix and basement membrane, which is predominantly laminin, fibronectin alongside collagen IV and V (79). Islet isolation with collagenase targets both the interstitial matrix and BM and both are shown to be lost immediately after isolation of human and canine islets (79–81). Loss of the islet BM in native islets of non-obese diabetic mice has been correlated with increased macrophage infiltration and development of insulinitis, suggesting that these initially BM devoid islets post-transplant have increased vulnerability (82). Integrin expression is also reduced progressively during periods of islet culture (78), but this reduction is diminished when islets are cultured with fibronectin or collagen. Furthermore, the addition of fibronectin and collagen significantly reduced apoptosis during islet culture demonstrating the benefits of maintaining islet-ECM interactions (79, 80, 83). Insulin expression reduces in intensity during culture of human islets parallel to simultaneous reduction in integrin expression (79). Recently, a human decellularized pancreatic ECM model has been used to culture isolated human islets. This method has resulted in improvement of in vitro insulin dynamics, with superior glucose stimulated insulin secretion and reduced apoptosis in extended culture. Transcriptomic analysis demonstrated increased expression of regulatory pathways associated with cell adhesion, ECM organisation and responses to external stimuli when compared to culture in suspension (84).

Avenues to improve engraftment include increasing selectivity of collagenase digestion, promoting neovascularisation through manipulation of islets, co-culture with ECM components, non-endocrine cells and promoting islet-ECM interaction post-transplant should be investigated to improve cellular therapies (65, 85).

The liver has many metabolic, digestive, and endocrine functions with complex micro-anatomical arrangement. The hepatic parenchyma receives dual blood supply from the portal venous system and hepatic arterial tree. Histologically, structures are arranged surrounding portal triads of portal venous sinusoids, arterial capillaries and bile duct branches with corresponding central veins which form the vascular outflow tract.

Following IT via the portal vein in man and primate models, islets will be spread amongst the liver with the majority sited within the venous lumen or closely related to it whilst only a minority become embedded within hepatic parenchyma (40, 76). Whereas, in small animal models, islets are most often seen within hepatic parenchyma, possibly due to endothelial breakdown as islets occlude the lumen in comparatively smaller vessels (76).

Intra-portal IT initiates the instant blood mediated immune response and triggers a response from the hepatic parenchyma (45). Embolic islets within the portal system result in areas of ischaemia inducing macroscopic changes in perfusion of the hepatic parenchyma in rodents and elevated hepatic enzymes in man (45). Hepatic stellate cells (HSC) located within the space of Disse in their quiescent state are responsible for ECM turnover and maintenance of the hepatic parenchymal environment (86). In response to injury, they undergo transformation into a myofibroblast that secretes tissue damaging ECM proteins ( Figure 2 ). In chronic liver disease, such as non-alcoholic fatty liver disease, activation of HSCs is responsible for the pathological deposition of excessive matrix proteins associated with fibrosis, and the loss of conventional tissue architecture and function. Moreover, HSC activity is correlated with progressive liver disease and severity (87–89).

HSC activation in progressive fibrosis is detrimental to tissue function, their presence and matrix deposition surrounding intra-portally transplanted islets may influence islet survival and function. This local microfibrosis may interfere with diffusion of insulin, increasing distances to capillary perfusion and paracrine activity and after islet isolation, which destroys native islet ECM, this post-transplant matrix deposition may result in abnormal matrix-cell signalling, which may contribute to impaired beta-cell function and survival. Of note, streptozotocin induced hyerglycaemic mice fed a high fat diet to induce steatohepatitis, have significantly lower rates of euglycaemia after intra-portal IT with significantly higher non-fasting capillary glucose readings compared to controls (90).

It may be that in some scenarios, HSC induced matrix deposition helps to restore the islet-ECM signalling necessary for function. The nature of this consequent post-transplant matrix deposition needs to be investigated, as when isolated HSCs are co-transplanted with islets under the mouse kidney capsule, improved glycaemic control is demonstrated through increased duration of graft function and superior islet dynamics assessed with intra-peritoneal glucose tolerance testing. Histological analysis reveals HSCs are sited surrounding islet clusters in this setting, which was also associated with increased islet insulin expression on immunofluorescent staining compared to solitary islet transplant controls (91). In mice, this pattern of fibrosis surrounding islet grafts has been associated with long term functioning portal venous grafts at 200 days post-transplant, whereas, fibrosis detecting through Masson’ trichrome staining within islet grafts has been seen in graft failure; a pattern which can be reduced with pre-treatment with nicotinamide (92). HSCs have also been demonstrated to have a positive immune tolerance role after subcapsular renal transplantation, which may improve islet survival from immune mediated cell death (93, 94).

Transplanted islets may also influence their microenvironment through paracrine activity. Local hepatic histological changes have been noted in an early cohort of IT recipients and focal peri-portal steatosis has been demonstrated by magnetic resonance imaging and on histology (40, 95). Presence of focal steatosis was associated with higher requirements of exogenous insulin. Interestingly, amongst the insulin independent recipients in this cohort, those with hepatic steatosis also had higher serum C-peptide levels, suggesting that better functioning grafts produced more steatosis through paracrine action. In one case, steatosis resolved after graft failure, supporting a paracrine insulin mechanism (95). Of note, markers of HSC activation, including expression of aSMA, fibronectin and a1 procollagen are increased when isolated rat HSCs are treated after 3-4 passages in vitro with a combination of insulin and 15 mM glucose for 24 hours (96). In addition, insulin and IGF-1 have been shown to increase HSC proliferation in vitro; culture with IGF-1 also promoted collagen-1 accumulation in media (97). This demonstrates the validity of investigating how the two-way relationship between transplanted islets and the hepatic parenchyma influences this post-transplant microenvironment.