Cell delivery and survival is a formidable challenge. To achieve functional retinal regeneration, an efficient and reliable means of cell delivery is required. Bolus injection of cell suspension is the typical method. However, this technique results in disorganized or incorrectly localized grafts. Bolus injection also contributes to poor cell survival due to shearing forces during injection and reflux of cells from the injection site [1]. Our experience, as well as that of many others, suggests the survival of grafted cells is quite low by bolus injection, with less than 1% of the grafted cells surviving 2–4 weeks after transplantation [2,3]. In contrast, our work delivering cells to the retina via a degradable poly(L-lactic acid)/poly(lactic-co-glycolic acid) (PLLA/PLGA) scaffold demonstrated a decrease in the number of cells lost to reflux and/or shearing forces by nearly 50%. Four weeks after transplantation, the scaffold improved overall survival by 10-fold over cell suspension grafts, corresponding to a survival rate of 78% [1]. This suggests the use of polymer scaffolds can solve both cell delivery and survival problems, resulting in an improved graft outcome.

The limited integration of grafted cells is another problem. This is in large partly due to the barrier formed by inhibitory extracellular matrix (ECM) and cell adhesion molecules associated with reactive gliosis after degeneration or injury [4-6]. A previous study used adult mice deficient in glial fibrillary acidic protein and vimentin as the host, and demonstrated that the grafted cells integrated robustly into the host retina with distinct neuronal identity and appropriate neuronal projections [7]. This suggests removal of the barrier produced by reactive glial cells in the degenerative host retina is important in encouraging cell integration. In an attempt to increase cell integration, different drugs with the ability to disrupt the outer limiting membrane or degrade ECM molecules or their hyaluronan binding targets have been tried, such as DL-alpha-aminoadipic acid, chondroitinase ABC and MMP2 [8-10]. The transplantation of different cell types has also been investigated. Gage et al. demonstrated that adult rat hippocampus-derived neural progenitor cells (NPCs) could integrate to a high degree in most layers of the retina when injected into the eyes of neonatal rats or mature rats with active retinal degeneration [11,12]. The high migratory potential of these cells could be attributed to the expression of matrix metalloproteinase-2 (MMP2) [10]. We isolated retinal progenitor cells (RPCs) from postnatal mammalian neuroretina and showed they also possessed integrative abilities similar to NPCs [2]. Transplantation of RPCs stimulated the host retina to increase MMP2 secretion, partly from activated glial cells. In turn, increased MMP2 expression induced degradation of the inhibitory glial scar related molecules, such as CD44 and neurocan, deposited at the outer limits of the degenerative retina. This resulted in enhanced host donor integration following retinal transplantation [13]. This study provided the first identification of an MMP2-dependent mechanism in cell integration. As the endogenous MMP2 was not enough to achieve sufficient cell integration in vivo, we then transplanted a biodegradable MMP2-PLGA scaffold, in conjunction with RPCs, to the subretinal space of Rho–/– mice and demonstrated enhancement of cell integration with evident degradation of CD44 and neurocan [14] (Figure 1). Similarly, when RPCs were co-transplanted with biodegradable MMP2-PLGA microspheres, an increased number of cells were integrated with the degenerative host retina [15]. Maclaren et al. used photoreceptor precursors from the developing retina at the peak of rod genesis and showed enhanced cell integration, suggesting the post-mitotic developmental donor stage may facilitate cell integration [3]. Our work with induced pluripotent stem cells (iPS) derived-photoreceptors demonstrated significant integration of grafted cells with host retina [16], suggesting photoreceptor-committed cells may also promote cell integration.

A third problem is controlling the differentiation of grafted cells to a specified cell type, which is an essential requirement for functional cell replacement. Despite high integration and morphological evidence of differentiation, NPCs cannot express retinal specific markers [12], which may in part be attributed to cell intrinsic factors. Photoreceptor precursors derived only from immature post-mitotic rod precursors and not from proliferating progenitor cells could differentiate into mature rod photoreceptors, form synaptic connections and improve visual function [3]. We transplanted iPS-derived photoreceptors into Rho–/– mice and demonstrated that integrated cells in the degenerative retina could express photoreceptor specific markers recoverin, rhodopsin and opsin, form synaptic connections with the inner retina and improve visual function [16]. RPCs were also able to differentiate into mature photoreceptors, express photoreceptor markers recoverin, rhodopsin and opsin, as well as exhibit photoreceptor morphological similarities [2]. Moreover, recipient mice showed improved visual function as determined by light-mediated behavior over a 25-week period when only a few photoreceptors left. However, no rhodopsin positive cells could be found in the Rho–/– mice model whether these cells were transplanted alone or co-transplanted with MMP2 microspheres [15]. This may be related to the complete absence of rhodopsin expression in Rho–/– hosts. Polymer scaffolds have been demonstrated to promote differentiation of RPCs in vitro prior to transplantation. When RPCs cultured on polymer scaffolds for one week were transplanted, they were able to express rhodopsin even in Rho–/– mice [1,13] (Figure 2). These results suggest post-mitotic photoreceptor precursors and photoreceptor-committed cells (differentiated from iPS, RPCs or other cells), may facilitate differentiation into mature photoreceptors regardless of host settings. Limited source is a problem for photoreceptor precursors. In vitro expansion of proliferating donor cells can overcome this problem. Using polymer scaffolds to direct stem cells towards photoreceptor-committed cells may facilitate the desired outcome while simultaneously inhibiting undesirable proliferation.

The most common synthetic biodegradable polymers are poly(-lactic acid) (PLA), poly(glycolic acid) (PGA), and their copolymer, PLGA. These polymers are degraded by hydrolysis of the ester linkage into lactic and glycolic acids, which are subsequently metabolized by the body. By controlling the crystallinity, molecular weight, and the ratio of lactic to glycolic acid subunits, the degradation time of PLGA can be tailored over a wide range [42]. PLGA has been approved by the FDA for uses in drug delivery, diagnostics and other applications of clinical science research, including cardiovascular disease, cancer, vaccine and tissue engineering [43-45]. Furthermore, PLGA has been shown to exhibit a high degree of biocompatibility in the eye [34]. We have shown that PLGA/PLLA scaffolds improved the survival of RPCs in different retinal degenerative models, promoted differentiation of RPCs and provided physical guidance, resulting in a more normal anatomical organization [1,46]. RPC/PLGA scaffold constructs showed at least a 16-fold and 10-fold increase in cell delivery and survival respectively, thereby promoting cellular integration. RPCs delivered via the PLGA scaffolds could express photoreceptor markers recoverin and rhodopsin [1]. These results suggest PLGA scaffolds facilitate the survival, integration and differentiation of the grafted cells. Despite the aforementioned benefits, PLGA scaffolds have extremely limited flexibility, even when blended with PLLA. Their thickness (150–250 μm) is substantially greater than the photoreceptor layer of the rodent retina (∼30 μm), thus resulting in damage to the host retina and presenting a substantial barrier to functional photoreceptor replacement in rodent models [1,46]. In addition to retinal detachment, other complications have also been reported, such as inflammation, fibrosis and foreign body response [47,48] (Table 1).

In order to reduce the incidence of trauma, ultra-thin PMMA scaffolds (6 μm) were introduced. Due to their thinness, these scaffolds increased the ease of delivery and reduced the risk of trauma, thereby enhancing potential integration with the host. The grafted RPCs were able to express at least three markers of mature retinal cells. Moreover, no foreign body responses were seen four weeks after transplantation [49]. However, PMMA cannot be degraded and remains in the subretinal space permanently or until removed. We suggest that this may interfere with retinal reattachment and increase the risk of scarring and inflammatory response (Table 1).

A biodegradable and elastomeric PGS scaffold of intermediate thickness (45 μm) was then introduced [50]. Compared to PLGA, PGS has several advantages: (1) PGS has improved biocompatibility, including less inflammation and fibrosis and no foreign body giant cell response [48,50]. (2) PGS has an ideal degradation profile. PGS has been shown to be degraded by surface erosion, which results in preservation of geometry, no detectable swelling, and slow loss of mechanical strength relative to mass, whereas PLGA underwent bulk degradation accompanied by extensive deformation, swelling, and faster loss of mechanical strength than mass. PLGA lost > 98% of its mechanical strength after 7 days which was approximately the cell seeding time prior to transplantation, whereas PGS lost only about 8% over the same time despite a greater loss in mass. Significant loss of mechanical strength increased fragility of the scaffolds and difficulty of transplantation. Moreover, PGS is a rapidly degrading polymer with an in vivo degradation time of about 4–8 weeks [51]. (3) PGS is an elastic and soft material with an elastic modulus of 1.66 ± 0.23 MPa and a maximum strain at failure of 113 ± 22%, whereas PLLA/PLGA is a brittle material with an elastic modulus of 9.0 ± 1.7 MPa and a maximum strain at failure of only about 9% [46,52]. As retinal tissue has an elastic modulus of 0.1 MPa and a maximum strain at failure of about 83% [53], PGS possesses improved mechanical properties that are more similar to those of retinal tissue. The elastomeric property of PGS allows the polymer scaffolds to be scrolled and transplanted via syringe injection, thereby protecting cells from shearing forces and reducing trauma to retinal tissue (Table 1). We have shown that PGS scaffolds promoted RPC survival, differentiation and glutamate induced calcium influxes. By using PGS scaffolds as a cell delivery vehicle, 13% and 7% of grafted cells migrated into the C57BL/6 and Rho–/– mouse retinal explants at one week. 1.5% and 0.8% of grafted cells migrated into the C57BL/6 and Rho–/ – mouse retina four weeks post-transplantation [54].

Another potential scaffold material is PCL, which has also been utilized extensively for the delivery of drugs or proteins [55-57]. To date, the PCL scaffolds are the thinnest biodegradable substrates used in retinal tissue engineering (5 μm) [58]. They are fifty times thinner than the PLGA scaffolds and nine times thinner than the PGS scaffolds. They can be placed into the subretinal space with minimal physical distortion. Moreover, PCL is highly permeable, allowing for the passage of physiologically significant molecules. However, it takes about 2–3 years to degrade completely. PCL degradation occurs gradually from its surfaces, the same as PGS, with no pathologic increase in local acidity [58,59] (Table 1). We showed that structured PCL scaffolds increased the expression of mature photoreceptor and bipolar markers and that the PCL-delivered RPCs migrated into both normal and degenerative retina [60].

In order to achieve functional retinal regeneration, controlled cell differentiation is required. Studies of photoreceptor precursors and iPS suggest photoreceptor-committed cells may facilitate differentiation into mature photoreceptors after transplantation [3,16]. An ideal polymer scaffold should not only deliver an appropriate number of healthy cells to the retina, but also promote cell differentiation towards photoreceptor lineage prior to transplantation in order to improve the yield of desired cellular phenotypes and decrease the probability of undesired mitotic activity post-transplantation. All the above mentioned polymer scaffolds have demonstrated the ability to promote differentiation of RPCs and other cell types [61-65]. To achieve this, cells must attach to the polymer scaffolds first. This can be more difficult for some cell types, such as RPCs, which often form non-adherent neurospheres and do not readily adhere to a growth substrate. Generally, an extracellular matrix protein, such as laminin [2], is used to modify the surface for cell attachment. Laminin is known to influence differentiation of progenitor cells toward mature retinal phenotypes [66,67]. Other proteins, including fibronectin, collagen and poly-L-lysine, are also used. Significantly longer outgrowth of neurites occurred on laminin coated fibers compared to fibronectin coated fibers [68]. Collagen-functionalized nanofibers improved cell adhesion and migration compared to non-functionalized nanofibers [69]. Polymer scaffolds made from PLGA [1,46], PMMA [49] and PGS [52,54] required protein modification of the surface with either laminin or a combination of laminin and poly-L-lysine to achieve RPC attachment. Additionally, PLLA/PLGA scaffolds needed a concentrated cell solution in order to achieve appropriate levels of cell attachment [46].

In addition to changing surface chemistry, polymer surface topography and architecture can also be altered. Microscale and nanoscale topographies have been found to induce reorientation and differentiation of multiple cell types, including neural progenitor cells [62], astroglial cells [70], smooth muscle cells [71] and mesenchymal stem cells [72]. Furthermore, surface topography can direct cell morphologic adaptations and subsequent changes in protein expression levels [62,73-76]. Surface topographic cues similar to endogenous ECM can direct cell morphogenesis via cell-surface contact [60,75]. It is suggested that the influence of surface topography on differentiation is via cell–cell surface receptors which act as mechanochemical transducers, activating signal transduction pathways and ultimately modulating gene expression and protein expression [76]. In the intact retina, the orientation of photoreceptors and retinal bipolar cells is radial with respect to the curvature of the globe. To approximate this highly ordered physical characteristics of the retina, a variety of techniques have been attempted. RPCs attached to the PLGA porous scaffolds created by phase-inversion casting and solid-liquid phase separation [77] demonstrated down-regulation of immature markers and up-regulation of markers of differentiation. These cells also exhibited morphologies consistent with photoreceptors, including a high degree of polarization of the cells and high survival [46]. PMMA scaffolds microfabricated [78] to provide precisely defined pores demonstrated enhanced RPC adherence during transplantation and allowed for greater process outgrowth and cell migration into the host retinal layers. Integrated cells expressed the mature neuronal marker neurofilament-200, the glial marker glial fibrillary acidic protein and the retinal-specific marker recoverin [49]. Utilizing microfabrication processes and soft lithography (or creation of a polydimethylsiloxane (PDMS) inverse replica) in combination with polymer molding, porous scaffolds can be structured precisely, simply, rapidly, and at low-cost without the further use of specialized microfabrication equipment. Scaffolds of varying porosity and pore structure can be quickly and easily produced by this technique. PGS scaffolds fabricated using this method demonstrated increased differentiation of RPCs towards mature phenotypes as evidenced by changes in mRNA, protein levels, and enhanced sensitivity to glutamate. Transplanting composites into the subretinal space of C57BL/6 and Rho–/– mice resulted in long-term RPC survival, and integrated cells exhibiting mature retinal markers [52,53]. RPCs grown on PCL scaffolds molded from a PDMS inverse replica exhibited an increase in gene expression for the photoreceptor markers recoverin and rhodopsin [61]. PCL scaffolds fabricated by template synthesis [58] (or inverse replication) from a nanoporous anodized aluminum oxide substrate comprised a surface topography of vertically oriented nanowires which facilitated RPC adhesion, growth, long-time survival and differentiation [60]. Recently, we have investigated a PCL scaffold templated from nanostructured silicon wafers. RPCs cultured on these scaffolds had a tendency to differentiate towards photoreceptors, expressing photoreceptor markers crx, recoverin and rhodopsin (Figure 3).