Controlled Release Formulations of IL-2, TGF-β1 and Rapamycin for the Induction of Regulatory T Cells

Abstract

The absence of regulatory T cells (Treg) is a hallmark for a wide variety of disorders such as autoimmunity, dermatitis, periodontitis and even transplant rejection. A potential treatment option for these disorders is to increase local Treg numbers. Enhancing local numbers of Treg through in situ Treg expansion or induction could be a potential treatment option for these disorders. Current methods for in vivo Treg expansion rely on biologic therapies, which are not Treg-specific and are associated with many adverse side-effects. Synthetic formulations capable of inducing Treg could be an alternative strategy to achieve in situ increase in Treg numbers. Here we report the development and in vitro testing of a Treg-inducing synthetic formulation that consists of controlled release vehicles for IL-2, TGF-β and rapamycin (a combination of cytokines and drugs that have previously been reported to induce Treg). We demonstrate that IL-2, TGF-β and rapamycin (rapa) are released over 3-4 weeks from these formulations. Additionally, Treg induced in the presence of these formulations expressed the canonical markers for Treg (phenotype) and suppressed naïve T cell proliferation (function) at levels similar to soluble factor induced Treg as well as naturally occurring Treg. Most importantly, we show that these release formulations are capable of inducing FoxP3⁺ Treg in human cells in vitro. In conclusion, our data suggest that controlled release formulations of IL-2, TGF-β and rapa can induce functional Treg in vitro with the potential to be developed into an in vivo Treg induction and expansion therapy.

Introduction

Over the past two decades, regulatory T cells (Treg) have been identified as one of the central components of the mammalian immune system [1-4]. The most commonly described, widely studied, and possibly most abundant regulatory T cells in the body are those that express CD4, CD25 [1, 5] and FoxP3 [6-8]. These CD4⁺ CD25⁺ FoxP3⁺ cells (Treg) play important roles in suppressing the activity of self-reactive immune cells and in re-establishing homeostasis following infection. Moreover, increased numbers of Treg suppress diverse inflammatory diseases such as autoimmunity [2, 9], transplant rejection [10, 11], dermatitis [12], psoriasis [13, 14] and even periodontitis [15, 16]. Given this evidence, it is not hard to perceive that strategies to boost local Treg numbers could be developed into potential therapeutics to treat these diseases.

Enhancing numbers of Treg at local tissue sites can be achieved by (i) ex vivo expansion of Treg followed by their local administration or systemic re-infusion, or (ii) in vivo manipulation of immune cells in order to tip the balance between Treg and effector T cells towards Treg. The latter approach is preferable given the stringency associated with ex vivo culture of human cells under Good Manufacturing Practice (GMP) conditions [17-19]. One possible means to achieve increased number of Treg in vivo is the use of biologic therapies that selectively enhance Treg numbers and function. Various antibodies (Abs), such as anti- IL-2 monoclonal (m) Ab [20], superagonistic anti-CD28 mAb [21], and agonistic anti-CD4 mAb [22], have been used in the past to increase in vivo Treg numbers. However, their exact mechanism of action has still not been characterized, and their safety in humans remains questionable. In fact, phase I clinical trials of the superagonistic anti-CD28 Ab (TGN1412) resulted in severe negative reactions (cytokine ‘storm’) in all 6 human subjects who received the Ab [23].

An alternative approach to increase Treg numbers in vivo is through the establishment of a local immunosuppressive environment that selectively favors Treg expansion. An environment rich in IL-2, transforming growth factor- β1 (TGF-β) and rapamycin (rapa), an inhibitor of the serine-threonine kinase mammalian target of rapamycin, has been shown to favor Treg development, even under inflammatory conditions [24-26]. However, providing a continuous presence of these factors in vivo, has proven difficult. Controlled release vehicles for such factors offer a potential solution to these problems.

In this study we describe the development and testing of controlled release formulations for IL-2, TGF-β and rapa. We show that a combination of these formulations (called FactorMP henceforth) is capable of Treg induction in vitro using either mouse or human cells. Further, we demonstrate that the FactorMP-induced Treg maintain their proliferative capacity and functional ability in vitro and express phenotypic surface markers that are consistent with soluble factor-induced Treg.

Materials and Methods

a. Mice

Six-eight week old C57Bl/6 (B6) and B6.SJL-Ptprca/BoyAiTac (CD45.1) were purchased from Taconic and used within two months. All animals were maintained under specific pathogen free conditions. Experiments were conducted in accordance with the National Institutes of Health Guide for Care and Use for Laboratory Animals and under Institutional Animal Care and Use Committee-approved protocols.

b. Microparticle Preparation

IL-2 and TGF-β microparticles (IL-2MP and TGFβMP, respectively) were prepared using the double emulsion-evaporation technique, as described [27, 28]. For the IL-2MP the following conditions were used. Five μg of recombinant (r) mouse IL-2 (from R&D Systems Minneapolis, MN, prepared in 50 mM ammonium acetate and 1 mM DTT) was mixed with 2 mg of BSA and 5 mM NaCl in 200 μl of de-ionized water. This solution was added to 4 ml of dichloromethane containing 200 mg of poly lactic-co-glycolic acid (PLGA; RG502H, 50% glycolate 50% lactate blend, viscosity 0.16-0.24 dl/g, Boehringer Ingelheim Chemicals Inc., Petersburg, VA), and the mixture was agitated using a sonicator (Vibra-Cell, Newton, CT) at 25% amplitude for 10 sec, creating the primary emulsion. This emulsion was then mixed with 60 ml of 2% polyvinyl-alcohol (PVA, MW ∼25,000, 98% hydrolyzed; Polysciences) under homogenization (L4RT-A, Silverson, procured through Fisher Scientific) at 3000 rpm for 1 min, creating the second emulsion. The resulting double-emulsion was then added to 80 ml of 1% PVA, and left for 3 hr spinning at 600 rpm. Subsequently, the microparticles were centrifuged (200g, 5 min, 4 °C), washed 4 times in de-ionized water, and lyophilized (Virtis Benchtop K freeze dryer, Gardiner, NY; operating at 80 mTorr).

For the TGFβMP the following conditions were used. One μg of r-human TGF-β (CHO cell-derived, PeproTech, Rocky Hill, NJ; prepared in 10 mM sodium citrate) was mixed with 10 mg D-mannitol, 1 mg of BSA, and 15 mM NaCl in 200 μl of de-ionized water. This solution was added to 4 ml of dichloromethane containing 200 mg of PLGA (RG502H), and the mixture agitated using a sonicator at 25% amplitude for 10 sec, creating the primary emulsion. This emulsion was then mixed with 60 ml of 2% PVA (containing 125 mM NaCl) under homogenization at 3000 rpm for 1 min, creating the second emulsion. The resulting double emulsion was then added to 80 ml of 1% PVA (containing 125 mM NaCl), and left for 3 hr spinning at 600 rpm. Subsequently, the microparticles were centrifuged (200g, 5 min, 4 °C), washed 4 times in de-ionized water, and lyophilized.

The rapaMP were prepared using the single emulsion-evaporation technique as described [29, 30]. Briefly, 1 mg of rapa (LC labs, Woburn, MA) dissolved in DMSO was mixed with 4 ml of dichloromethane containing 200 mg of PLGA (RG502H). This solution was mixed with 60 ml of 2% PVA under homogenization at 3000 rpm for 1 min creating the microparticle emulsion. The resulting emulsion was then added to 80 ml of 1% PVA and left for 3 hr spinning at 600 rpm. Subsequently, the microparticles were centrifuged (200g, 5 min, 4 °C), washed 4 times in de-ionized water, and lyophilized.

c. Microparticle Characterization and Release Assays

Scanning electron micrographs of the microparticles were obtained using a scanning electron microscope (JSM-6330F, JEOL, Peabody, MA). The size distribution of microparticles was determined using volume impedance measurements on a Multisizer 3 (Beckman Coulter, Brea, CA).

Release assays were conducted by incubating a suspension of particles on a roto-shaker at 37 °C; (i) 10 mg in 1 ml of cell culture media for IL-2MP and TGFβMP, and (ii) 10 mg in 1 ml of PBS containing 0.2% Tween-80 for rapaMP (due to the low solubility of rapa in aqueous solutions, release assays were conducted in PBS containing Tween-80 to avoid a obtaining a release profile that was dissolution dependant). At regular time intervals, particle suspensions were centrifuged (250g, 5min), the supernatant removed, and the particles re-suspended in 1 ml of appropriate solution. The amount of each cytokine in the supernatant was measured using a cytokine-specific ELISA (R&D systems, Minneapolis, MN), and the amount of rapa was measured using spectrophotometry (absorbance at 278 nm).

d. Mouse T cell isolation

Spleen and lymph nodes were dissected from B6 or CD45.1 mice. Following mechanical digestion, the tissue suspension was passed through a 70 μm nylon filter to obtain a single cell suspension of leukocytes. Predominantly naïve CD4⁺ T cells (> 90% pure) were isolated from this suspension with a CD4⁺ T cell negative isolation kit (Invitrogen, Carlsbad, CA) as per the manufacturer's instructions. These purified CD4⁺ T cells were used in cell culture and suppression assays. Natural regulatory T cells (nTreg) were isolated from this purified population of CD4⁺ T cells through positive selection of CD25 expressing cells using the CD25 microbead kit (Miltenyi Biotec, Auburn, CA) as per manufacturer's instructions.

e. Induction of mouse regulatory T cells (Treg)

For Treg induction experiments, naïve T cells were cultured either in direct contact with FactorMP (in 96-well round bottom cell culture plates), or separated from FactorMP by permeable transwell inserts (HTS Transwell®-96, 0.4μm pore size; Corning, Lowell, MA); (n=3 experiments with FactorMP in direct contact with T cells, and n=2 experiments in transwell® plates with particles separated from the cells). Dynabeads® mouse T-activator CD3/CD28 beads (Dynabeads®; Invitrogen, Carlsbad, CA) were used at a 2:1 (beads:T cells) ratio to activate T cells, and cultures maintained for 4 days. For cultures in the presence of soluble factors, the following concentrations of factors were used: 10 ng/ml IL-2, 5 ng/ml TGF-β and 10 ng/ml rapa (corresponds to a total amount of 2 ng IL-2, 1 ng TGF-β and 2 ng rapa). The following quantities of microparticles were used for the induction experiments in 200 μl of cell culture media: 2 mg TGFβMP, 0.5 mg IL-2MP, and ∼ 0.01 - 0.05 mg rapaMP. TGFβMP were pre-incubated in media for 18-22 days prior to use in the Treg induction experiments to account for the initial lag in release of TGF-β from the microparticles. To determine the phenotype of cells after culture, cells were stained with anti-CD4 (L3T4), anti-FoxP3 (FJK-16s), anti-CD25 (PC61.5), anti-glucocorticoid-induced TNFR-induced protein (GITR; DTA-1), anti-folate receptor-4 (FR4; eBio12A5) (antibodies from eBiosciences, San Diego, CA) and anti-cytotoxic T-lymphocyte antigen 4 (CTLA4; UC10-4B9, from Biolegend, San Diego, CA). To determine iTreg proliferation, naïve T cells were stained with carboxyfluorescein diacetate succinimidyl ester cell tracer (CFSE; Invitrogen, Carlsbad, CA) prior to activation with Dynabeads®. Stained cells were then analyzed on a BD-LSRII flow cytometer.

f. Suppression assay

Freshly-isolated naïve CD4⁺CD45.1⁺ T cells were stained with CFSE (Invitrogen, as per the manufacturer's instructions) and co-cultured with induced Treg (generated as described above) at different ratios in 96-well plates. For suppression assays, iTreg were generated in Transwell® cell culture plates, such that the iTreg could be easily separated from the FactorMP prior to co-culture with naïve CD4⁺CD45.1⁺ T cells. The number of naïve CD4⁺CD45.1⁺ cells was kept constant at 50,000 cells / well. For stimulation, 50,000 Dynabeads® were used per well. Co-cultures were carried out for 4 days, after which cells were stained for flow cytometry.

g. Human T cell culture

Anti-coagulated peripheral venous blood was obtained from healthy adult volunteers under a University of Pittsburgh Institutional Review Board approved protocol. CD4⁺ T cells were isolated from mono nuclear cells using the CD4 negative isolation kit (Miltenyi Biotec, Auburn, CA). Cells (500,000) were cultured in 0.5 ml of media (with serum) in the presence of human T cell activation beads (anti-CD2, anti-CD3 and anti-CD28 coated beads, Miltenyi Biotec, Auburn, CA). Cell culture media was supplemented with additional factors or FactorMP at the following concentrations: 500 U/ml recombinant-human IL-2, 10 ng/ml TGF-β, 2 ng/ml rapa, 8 mg/ml of TGFβMP, and/or 0.02 mg/ml of rapaMP. Following 4 days of culture, cells were collected, stained and Treg induction analyzed by flow cytometry. Soluble IL-2 was used in all of these cultures instead of IL-2MP, as IL-2MP encapsulated mouse rIL-2 and not the human protein.

Results

a. Microparticle Characterization

IL-2MP, TGFβMP, and rapaMP were all prepared under similar conditions, using the same polymer (RG502H, viscosity 0.16-0.24 dl/g). Scanning electron micrographs (Figure 1A) show that IL-2MP have a porous exterior surface (as these particles were formulated with a 20-30 mOsm higher soluble concentration in the primary emulsion when compared to the bulk aqueous phase), while the TGFβMP are slightly porous (a 0-5 mOsm difference in solute concentration) and the rapaMP were smooth without pores on the surface. Additionally, particle sections (Figure 1B) show emulsion pockets in the IL-2MP and TGFβMP, while there are no interior pockets in the rapaMP as these particles were formulated using the single emulsion-evaporation fabrication procedure. Further, the particles were prepared such that they were large enough (IL-2MP = 25.5 ± 7.5 μm; TGFβMP = 16.7 ± 6.3 μm; rapaMP = 16.7 ± 6.4 μm; errors indicate standard deviation from the mean for each particle set) to remain at a site of injection and not be taken up by phagocytic cells as shown in figure 1C. The size distribution observed in these microparticle formulations is characteristic of the double emulsion-evaporation procedure and has been previously described [27, 28, 31, 32]. Finally, as shown in figure 2 we observed a high initial burst followed by continuous release from IL-2MP (loading efficiency = 37.1 ± 6.1 %), a linear release of TGF-β (loading efficiency = 24.9 ± 9.9 %) following a ∼2 week lag phase, and a continuous release from rapaMP (loading efficiency = 63.9 ± 3.0 %).

Figure 1. Microparticle Characteristics.

Scanning electron micrographs of intact microparticles (A) and the cross-section of microparticles (B) showing the internal architecture. C – Microparticle size distribution determined using volume impedance measurements.

Figure 2. Release Profile.

In vitro release profiles of IL-2, TGFβ (in cell culture media) and rapa (in PBS containing 0.2% Tween-80). Error bars on release profiles are based on n = 6 measurements for IL-2MP and TGFβMP, and n= 3 measurements for rapaMP.

b. Treg Induction

Soluble IL-2, TGF-β and rapa have been shown previously to induce Treg (iTreg) [24, 25]. We wanted to determine if degradable polymer-based formulations designed to sustain the release of these factors could induce Treg reliably. Indeed, we observed that the microparticle formulations were similar to soluble factors in their in vitro Treg induction efficacy, as measured by FoxP3 expression (Figure 3A and 3B). Furthermore, we observed similar Treg induction efficiencies in experiments with microparticles that were in direct contact with T cells, and those with microparticles separated from T cells by transwell membranes. In addition, the FactorMP were capable of inducing Treg by releasing equivalent (2-3 ng IL-2 and 2-10 ng rapa), or reduced (0.2-0.4 ng of TGF-β) total amounts of the factors over 4 days of culture. Further, we observed that iTreg were capable of robust proliferation (as observed through CFSE dilution, Figure 3A).

Figure 3. FactorMP induce mouse Treg.

A – representative flow cytometry dot plots (gated on CD4-expressing cells) of naïve T cells stimulated in the presence of soluble factors or FactorMP. The X axis on these plots represents CFSE, which is a cell proliferation marker and the Y axis represents intracellular FoxP3, which is a definitive marker for mouse Treg. B – quantitative analysis of the percentage of CD4⁺ T cells that express FoxP3 after culture for 4 days under different conditions; * indicates p<0.05 based on n ≥ 3 independent experiments.

c. Phenotype and function of microparticle-induced Treg

In addition to FoxP3, Treg are known to express many other characteristic surface proteins. We tested for 3 canonical surface markers: CD25, FR-4 and GITR. CD25 is the high-affinity IL-2 receptor, which increases sensitivity to IL-2 and is important for Treg proliferation. FR-4, a folate receptor, is required for folic acid sensing and uptake, which in turn prolongs Treg survival. Finally, GITR is a surface receptor that has been suggested to play an important role in Treg survival and suppression. We observed that FactorMP-iTreg expressed these surface markers at levels equivalent to those on soluble factor-iTreg (Figure 4A). Importantly, although the expression of these surface proteins, along with FoxP3, suggests that these cells are Treg, it does not guarantee suppressive function. In order to test the ability of FactorMP-iTreg to suppress naïve T cell proliferation, we adopted an in vitro co-culture system described previously [33]. In these co-culture suppression assays, we observed that FactorMP-iTreg indeed possessed suppressive capabilities similar to Treg induced by soluble factors and natural Treg (Figure 4B).

Figure 4. FactorMP-iTreg express canonical Treg surface markers and suppress effector T cells.

A – representative flow cytometry dot plots (gated on CD4-expressing cells) showing the expression of surface markers and intracellular FoxP3 on naïve T cells stimulated in the presence of soluble factors or FactorMP. B – representative plots of CFSE dilution showing that the FactorMP-iTreg can suppress naïve T cell proliferation. Gates on individual plots indicate the percentage of proliferating cells. Ratios (1:1; 1:4 and 1:8) indicate the number of Treg in culture to the number of naïve T cells. Data are representative of at least 2 independent experiments.

d. Microparticle formulations induce Treg from human T cells

Human T cells isolated from peripheral blood mononuclear cells can also be induced to a Treg phenotype using soluble IL-2, TGF-β and rapa [34, 35]. For potential clinical application of our technology, we needed to determine if the microparticle formulations were capable of inducing Treg from human T cells. To this end, human T cells were cultured in the presence of soluble factors or FactorMP. We observed that the microparticles were equally capable of inducing Treg when compared to the soluble factors (Figure 5), while releasing equivalent (1-10 ng/ml rapa) or reduced (2-4 ng/ml TGF-β) amounts of factors.

Figure 5. Microparticle formulations generate human-iTreg.

A – representative plots displaying FoxP3 expression profile on human T cells cultured under different conditions. Numbers in plots represent the median fluorescence intensities (MFI). Grey plots indicate the FoxP3 expression in naïve unstimulated T cells. B – quantitative analysis of normalized FoxP3 MFI as determined from 2 independent experiments (n≥3). MFI was normalized by determining the ratio of experimental MFI and control (soluble IL-2 treated cells) MFI.

Discussion

Therapies that enhance Treg numbers and function have the potential to suppress transplant rejection and autoimmunity [2, 3]. Clinical trials are currently testing cellular therapies involving Treg as potential therapeutics for treating graft versus host disease [34, 36]. However, Treg-based cellular therapies face many challenges, which include, but are not limited to: (i) difficulties in isolating pure and homogenous populations and large quantities of Treg from the blood, (ii) inconsistent maintenance of the Treg phenotype and suppressive function post-proliferation, and (iii) the need for GMP facilities [17-19]. Hence, acellular therapies that can increase numbers and/or the suppressive potency of Treg without the need for ex vivo culture could be transformative.

One potential method to increase the ratio of Treg to effector T cells is to establish an environment rich in IL-2, TGF-β and rapa as described [24-26, 37]. Such an immunosuppressive, Treg-inducing environment can be attained through the sustained release of these factors at a local site. To this end, release formulations were fabricated using an FDA-approved, widely-used polymer, PLGA. We prepared porous IL-2MP (Figure 1) with a high initial burst, followed by a slow continuous release of the factor over a 5 week time frame (Figure 2). Such a release formulation for IL-2 was suitable for the induction of Treg, as it has been suggested that high initial doses of IL-2 might help Treg grow better and resist apoptosis [38, 39]. Additionally, we sought a way to continuously release TGF-β, as it has been shown that continuous presence of TGF-β is required for FoxP3 expression in naïve T cells [24, 40]. Other groups have reported release of TGF-β from similar PLGA microparticles [28, 32] with the expected burst and lag that typically accompanies protein release [41, 42]. In our formulations, we wished to avoid this initial burst of TGF-β, and indeed our formulations were successfully able to abstain from that initial burst (Figure 2). However, despite multiple attempts, we were unable to circumvent the 2 week lag phase. Unpublished data from our lab suggest that the initial lag phase could be due to ionic interactions between TGF-β (with a high isoelectric point of ∼8.6) and degrading PLGA polymer. Regardless, to overcome the problem of the initial lag phase, we were able to simply pre-incubate the TGFβMP (18-22 days) prior to their use in cell culture, which ultimately results in the initially targeted, linear release over a 3 week period of time (Figure 2). Finally, rapaMP were also formulated to be similar in size to the IL-2MP and TGFβMP and to release continuously over a 2-3 week time frame as previously demonstrated (Figure 2) [29, 30].

Importantly, the combination of these microparticle formulations (FactorMP) is as effective as soluble factors at inducing Treg from naïve T cells in in vitro cultures (Figure 3B). Additionally, we determined that the FactorMP iTreg were capable of robust proliferation (Figure 3A), expressed canonical surface markers representative of Treg (Figure 4A), and were able to suppress naïve T cell proliferation in an in vitro suppression assay (Figure 4B). Further, it was observed that Treg induction and proliferation occurred even when the cells were in contact with microparticles, suggesting that the microparticles do not have adverse on these cells. Finally, we observed that these microparticles are equally effective at inducing human Treg. The human-iTreg showed high expression of FoxP3 (Figure 5) and were also capable of proliferation (data not shown). Overall, our data suggest that these FactorMP have the potential to be used in vivo for local Treg induction at sites of transplant rejection or autoimmunity.

We envisage these particulate formulations could be explored as an ‘off-the-shelf’ therapeutic for creating a local immunosuppressive environment and increasing the presence of Treg at sites of inflammation. We are currently testing these particles in in vivo mouse models of destructive inflammation and autoimmunity where induction of immunological homeostasis may alleviate disease symptoms. Another possible application for such formulations would be to create an immunosuppressive lymph node like environment in vivo, when used in combination with formulations that can recruit [43] and activate [44, 45] naïve T cells.