Bioinspired Controlled Release of CCL22 Recruits Regulatory T Cells In Vivo

Deficiency of regulatory T cells (Treg), a type of lymphocyte that promotes immunological homeostasis in the healthy steady state, [1–3] is a causative factor for destructive inflammation and autoimmune disease. [2,4] Accordingly, increasing the presence of Treg at the site of autoimmunity has been suggested as a potential method to treat these types of disorders. [3,4] However, to our knowledge, techniques to increase the local presence of specific immune cell types in vivo do not yet exist. Herein, we develop and test a microscale controlled-release system for the recruitment of Treg that draws inspiration from a specific biological system: tumors. Specifically, a wide variety of tumors release the chemokine CCL22, [5,6] which is responsible for tumor-directed migration of Treg and corresponding tumor-specific immune evasion. Our hypothesis was that if a steady CCL22 release could be maintained, by tuning the material properties of a controlled-release system, Treg may be preferentially recruited to a local site in vivo. The first step towards testing this hypothesis was to design a release vehicle suitable for the steady release of CCL22. Although polymeric-microparticle-based controlled-release systems for proteins (including chemokines) have previously been developed, [7,8] the release of hydrophilic proteins from these particles typically follows a triphasic release profile. [9,10] Our goal was to achieve release of CCL22 without any periods of lag in order to produce a corresponding steady gradient of chemokine originating from a point source. To this end, new mechanistic descriptions of how controlled release from degradable materials [9,11] suggest that one of the major factors affecting the rate of release is the erosion of a dense polymer matrix into a porous macrostructure that allows for protein egress. The timing of this process and, in turn, the period of lag is determined by the type of polymer, its molecular weight (inherent viscosity) and the degree of porosity in the matrix. Accordingly, we hypothesized that addition of contiguous pores to the microparticles during fabrication would pre-establish pathways for diffusive protein egress and should bypass the need for erosion as a requisite for release. Thus, we used the polymer poly(lactic-co-glycolic acid) (PLGA) of a specific inherent viscosity (0.16–0.24 dl/g in 0.1% chloroform; in order to achieve release within a 3–8 week period) and modified the surface porosity using an osmotic gradient between the inner aqueous emulsion and the outer bulk aqueous phase (Supporting Information, Figure S1). We observed that porous microparticles prepared using a specific osmotic gradient (corresponding to 15 × 10−3 m NaCl in inner aqueous emulsion), in contrast to unmodified degradable particles ( Figure 1a), did not display the standard lag phase that would be expected with release of proteins and released CCL22 continuously over a 4-week period (Figure 1b). As an additional design parameter, the particles were made to be large enough to avoid uptake by phagocytic cells and to prohibit their movement across vascular endothelium (Supporting Information, Figure S2). As a consequence, particles would remain immobilized at the site of injection. Figure 1 CCL22MP Characteristics. a) Scanning electron micrographs of nonporous (left; top and bottom) and porous (right; top and bottom) CCL22 microparticles (CCL22MP). The top images were taken at 500× magnification, while the bottom images were taken ... In order to test the ability of rationally designed CCL22MP to attract Treg, an in vivo adoptive transfer model coupled with non-invasive live-animal imaging was used. Specifically, fluorescently labeled CCL22MP were injected into the triceps surae of normal FVB mice followed by intravenous (i.v.) infusion of ex vivo-alloactivated Treg [12] (AATreg) that constitutively expresses the luciferase gene. The hind limb muscles were chosen as the site for MP injections as these distal sites are not expected to produce large quantities of CCL22. The migration pattern of these bioluminescent AATreg was studied following the injection of non-labeled mature allogeneic dendritic cells (DC), which provide an activation stimulus (Supporting Information, Figure S3). Soon after DC stimulation, a significantly greater number of AATreg was recruited to the site of CCL22MP injection compared to an internal control of microparticles lacking CCL22 ( Figure 2a and Supporting Information Figure S4). The adoptively transferred Treg are expected to translocate to sites in the body (e.g., lungs, gut etc.) that secrete CCL22, as observed in Figure S4, but not sites such as the hind limb (as confirmed by our preliminary experiments (data not shown)). However, a significant number of Treg do migrate towards the hind limb muscles upon injection of CCL22MP but not upon addition of blank particles ( p < 0.05), suggesting that the migration must be due to the release of CCL22 from these particles. Importantly, several reports suggest that small changes in Treg numbers at local sites (leading to a change in the ratio of Treg to effector T cells) is sufficient to dramatically alter local immune responses. [13] In line with these reports, and as a test for the ability of CCL22MP to replace the function of CCL22 secreting cells in vivo (e.g., site-specific recruitment of regulatory cells), we also demonstrate that CCL22MP are effectively able to delay the rejection of transplanted allogeneic cells. Specifically, allogeneic luciferase-expressing Lewis lung carcinoma cells (which do not endogenously produce CCL22) were implanted subcutaneously into mice at the site of CCL22MP injections (or for comparison BlankMP or Bolus CCL22 as controls), and the time to rejection was recorded using non-invasive live imaging (Figure S3). We observed that the rejection rates were significantly slower in the CCL22MP group when compared to both bolus CCL22 and BlankMP controls (Figure 2b and Figure S5), supporting the hypothesis that establishing a CCL22 gradient in vivo using synthetic systems can help modulate local immune respons1es. Figure 2 CCL22MP in vivo. a) Representative fluorescence (red-gold) and luminescence images (blue-yellow) showing localization of particles and Treg, respectively; fluorescence images were used to outline (red-line) areas of particle localization (Igor Pro Living ... Potential therapeutic implications of such a bioinspired degradable controlled-release formulation capable of recruiting Treg in vivo are manifold. One obvious application is the use of CCL22MP in combination with an infusion of Treg expanded ex vivo. Current pre-clinical data suggest that freshly-isolated or ex vivo-expanded Treg infusion can prevent organ transplant rejection or suppress autoimmune diseases. [2,14,15] However, challenges such as obtaining adequate numbers and highly-purified populations of Treg have hindered progress into clinical trials. [14,15] Using formulations that release CCL22, it may be possible to lower the numbers of injected Treg, or potentially use populations with lower purity. Another use of CCL22MP would be to attract endogenous Treg populations, wherein these formulations have the potential to function as “off-the-shelf” therapeutics for the treatment of a wide variety of disorders associated with unrestrained immune reactivity. A potential drawback of using a CCL22 sustained release vehicle is that the receptor for this chemokine (CCR4) is expressed on both activated Treg and activated effector T cells, [16] suggesting that both these cell populations would be attracted towards CCL22MP. Yet in vivo studies show that CCL22 production associated with tumors [5,17] or long-surviving allografts [18] result in preferential recruitment of Treg leading to local immunosuppression. One possible explanation for these results is that Treg could express significantly more CCR4 than effector T cells [17] (Supporting Information, Figure S6). Further, it has been suggested that an optimal ratio of Treg to effector T cells, and not a complete absence of effector T cells, is necessary for effective local suppression of immunity. [14,19] Regardless, if local effector T cells abrogate a suppressive environment, CCL22MP can easily be modified to simultaneously release immunosuppressive agents (such as rapamycin) [20] to inhibit these effector T cells in situ, thereby assisting Treg to control adverse immune responses. In conclusion, we demonstrate that site-specific attraction of Treg leading to local immunomodulation can be achieved in vivo using CCL22MP. These bioinspired controlled-release formulations are particularly attractive as modular platforms for therapeutic development, as well as tools to study Treg-dependent modulation of immune responses in situ.