Biomaterials‐based nanoparticles conjugated to regulatory T cells provide a modular system for localized delivery of pharmacotherapeutic agents

Abstract

Type 1 diabetes (T1D) presents with two therapeutic challenges: the need to correct underlying autoimmunity and restore β‐cell mass. We harnessed the unique capacity of regulatory T cells (Tregs) and the T cell receptor (TCR) to direct tolerance induction along with tissue‐localized delivery of therapeutic agents to restore endogenous β‐cell function. Specifically, we designed a combinatorial therapy involving biomaterials‐based poly(lactic‐co‐glycolic acid) nanoparticles co‐loaded with the Treg growth factor, IL‐2, and the β‐cell regenerative agent, harmine (a tyrosine‐regulated kinase 1A [DYRK1A] inhibitor), conjugated to the surface of Tregs. We observed continuous elution of IL‐2 and harmine from nanoparticles for at least 7 days in vitro. When conjugated to primary human Tregs, IL‐2 nanoparticles provided sufficient IL‐2 receptor signaling to support STAT5 phosphorylation for sustained phenotypic stability and viability in culture. Inclusion of poly‐L‐lysine (PLL) during nanoparticle‐cell coupling dramatically increased conjugation efficiency, providing sufficient IL‐2 to support in vitro proliferation of IL‐2‐dependent CTLL‐2 cells and primary murine Tregs. In 12‐week‐old female non‐obese diabetic mice, adoptive transfer of IL‐2/harmine nanoparticle‐conjugated NOD.BDC2.5 Tregs, which express an islet antigen‐specific TCR, significantly prevented diabetes demonstrating preserved in vivo viability. These data provide the preclinical basis to develop a biomaterials‐optimized cellular therapy to restore immune tolerance and promote β‐cell proliferation in T1D through receptor‐targeted drug delivery within pancreatic islets.

1. INTRODUCTION

Current medicines have largely failed to prevent or reverse the approximately 80 autoimmune diseases that are characterized by loss of immune self‐tolerance, including type 1 diabetes (T1D). ¹ , ² Collectively, these disorders affect an estimated 23.5 M individuals in the United States, ³ with T1D prevalence continually on the rise over the past several decades. ⁴ Hence, curative therapies that target the underlying disease process are desperately needed.

Regulatory T cell (Treg) defects contribute to the development of T1D in humans and in the non‐obese diabetic (NOD) mouse model. ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ The efficacy of transferred Tregs as a therapeutic modality has been demonstrated in animal models of systemic lupus erythematosus, ¹² multiple sclerosis, ¹³ inflammatory bowel disease, ¹⁴ oophoritis, ¹⁵ and T1D. ¹⁶ , ¹⁷ This successful groundwork inspired the translation of this approach to human T1D patients, with a phase I clinical trial demonstrating the safety and feasibility of ex‐vivo expanded, autologous polyclonal Tregs. ¹⁸ However, a rapid early loss of Tregs was observed upon adoptive cell transfer (ACT), and intervention did not result in robust preservation of endogenous C‐peptide production, ¹⁸ highlighting an ongoing need to both preserve Treg engraftment and restore endogenous β‐cell mass.

The destruction of β‐cells observed in T1D is unique in its tissue‐restriction, with β‐cell specificity conferred by the human leukocyte antigen (HLA) locus ¹⁹ and its influence over the T cell repertoire. ²⁰ , ²¹ The HLA class II region constitutes the strongest risk determinant in T1D with selected haplotypes conferring disease risk across a staggering range (0.1 to ~50 times the reference genotype). ²² , ²³ These concepts form the basis for applying MHC‐restricted T cell receptors (TCRs) capable of recognizing islet peptides presented by the highest risk HLA‐DR3/4 and DQ2/8‐trans alleles covering ~90% of all T1D cases. ²⁴ In support of this notion, Tregs derived from either TCR transgenic mice or endogenous antigen‐specific T cells are dramatically more potent (10–100‐fold) at suppressing and reversing murine autoimmune disease, as compared to polyclonal Tregs. ¹⁶ , ¹⁷ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ Until recently, the isolation of sufficient quantities of islet antigen‐specific Tregs has presented a challenge in translating these findings. ³¹ The transformative demonstration that human T cells can be manipulated ex vivo has created a new class of “living drugs”—opening novel avenues for genetic modifications of T cells to redirect specificity. ⁵ , ²⁴ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ TCR or chimeric antigen receptor (CAR) gene modified T cells are in clinical use, representing a paradigm shift in the field of ACT for combatting cancer, ³⁸ but this approach has yet to be fully realized in the treatment of autoimmune diseases such as T1D.

Treg survival and function are IL‐2 dependent. ³⁹ Indeed, data from the literature suggest Tregs require concurrent delivery of IL‐2 for in vivo persistence and optimal engraftment. ⁴⁰ , ⁴¹ Low‐dose IL‐2 treatment has been shown to prevent diabetes in NOD mice. ⁴² In humans, however, the systemic administration of IL‐2 can also cause harmful off‐target effects by augmenting the proliferative and cytotoxic activities of effector T cells and natural killer (NK) cells, ⁴³ , ⁴⁴ and impairing β‐cell function when administered in combination with rapamycin. ⁴⁵ This suggests novel methods for localized growth factor delivery to sustain Tregs post‐infusion may augment engraftment. In a series of elegant studies, the Irvine laboratory pioneered the nanoparticle (NP)‐coupled T cell approach for autocrine and targeted paracrine delivery of cytokines as well as TLR agonists to enhance anti‐tumor ACT. ⁴⁶ , ⁴⁷ , ⁴⁸ We hypothesized that optimization of islet antigen‐specific Tregs, through biomaterial‐based drug delivery, would enhance their stability and fitness, directing Tregs to the islets and pancreas draining lymph node (pLN) where they would not only exert tolerance induction but also, deliver targeted payloads of drugs to restore endogenous β‐cell mass and prevent T1D. Expanding upon our established methods for generating poly(lactic‐co‐glycolic acid) (PLGA) microparticles and NPs as biocompatible vehicles for controlled in vivo delivery of therapeutic agents, ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ we herein report the cellular conjugation of NPs loaded with IL‐2 and harmine, a compound putatively reported to promote β‐cell proliferation, ⁵⁸ as well as their in vivo evaluation using the NOD mouse model of T1D.

2. MATERIALS AND METHODS

2.1. IL‐2 NP fabrication

NPs containing IL‐2 were fabricated using a modified solvent evaporation technique newly developed in our laboratory. Briefly, a 3% solution of poly(lactic‐co‐glycolic acid) (PLGA)—comprised of a 75:25 ratio of DL lactide/glycolide (Corbion), except for experiments involving human Tregs which used a 50:50 ratio of DL lactide/glycolide (Resomer® RG 504H, Sigma‐Aldrich)—was prepared by dissolving 50 mg of PLGA in 1.5 ml of methylene chloride. To this solution, 0.5 ml of a 1.5% solution of PLGA—b‐poly(ethylene glycol)‐Malemide (PLGA/PEG‐Mal) in methylene chloride was added. To form the primary emulsion, a 200 μl solution containing the proteinaceous payload comprised of either 500 μg recombinant human IL‐2 (rhIL‐2, ProSpec) and 1.5 mg of BSA (Sigma) or 1.5 mg BSA alone were added to the dissolved polymers in a drop‐wise fashion and sonicated at 50% output using a 55 W sonicator (Qsonica) for 60 seconds. To form the secondary emulsion, the primary emulsion was added to a 1.5% polyvinyl alcohol (Mowiol 8‐88, Sigma) solution in a dropwise fashion and sonicated at 40% output for 90 s. The resulting emulsion was then transferred to a rapidly stirring 0.2% polyvinyl alcohol solution and continuously stirred for 3 h to allow for the evaporation of methylene chloride and formation of the particles. Resultant NPs were pelleted at 12,000xg then washed and pelleted four times using an excess of molecular grade water to remove all traces of polyvinyl alcohol. Assessment of separate lyophilized fabrication batches revealed that this protocol consistently yielded an average of 23 mg of NPs (±4.1 mg, n = 4). NPs were used immediately following fabrication.

2.2. IL‐2/harmine NP fabrication

As harmine is hydrophobic in nature ⁵⁹ and IL‐2 is hydrophilic, ⁶⁰ we predicted that it would be possible to encapsulate both agents simultaneously within a single NP. As before, NPs were fabricated using a modified solvent evaporation technique newly developed in our laboratory. Briefly, a 3% solution of PLGA comprised of a 75:25 ratio of DL Lactide/Glycolide (Corbion Purac) was prepared by dissolving 50 mg of PLGA in 1.5 ml of methylene chloride. To this solution, 0.5 ml of a 1.5% solution of PLGA/PEG‐Mal in methylene chloride was added. 40 mg of harmine suspended in 200 μl of ethanol was added the PLGA/PEG‐Mal solution and gently agitated for 30 min. To form the primary emulsion, 200 μl rhIL‐2 (10 mg/ml) was added to the dissolved polymers containing harmine in a drop‐wise fashion and sonicated at 50% output using a 55 W sonicator (Qsonica) for 60 s. To form the secondary emulsion, the primary emulsion was added to a 1.5% polyvinyl alcohol (Mowiol 8‐88, Sigma) solution in a dropwise fashion and sonicated at 40% output for 90 s. The resulting emulsion was then transferred to a rapidly stirring 0.2% polyvinyl alcohol solution and continuously stirred for 3 h to allow for the evaporation of methylene chloride and formation of the particles. Resultant NPs were pelleted at 12,000xg then washed and pelleted four times using an excess of molecular grade water to remove all traces of polyvinyl alcohol, then immediately used in the assays described in the following sections.

2.3. NP physicochemical characterization

IL‐2 and IL‐2/harmine NPs were re‐suspended between 5 and 10 mg/ml in deionized H₂O and their diameters measured by dynamic light scattering using the Malvern Panalytical Zetasizer Ultra system. To determine the stability of the colloidal dispersions, the Malvern Panalytical Zetasizer Ultra was also used to measure the zeta potential of NPs in suspension.

2.4. NP payload elution

To assess release kinetics, 2 mg of IL‐2 NPs or IL‐2/harmine NPs were suspended in 600 μl sterile PBS in sealed tubes and repeatedly inverted at 37°C. At 24 h and days 3, 7, 14, and 28, NPs were pelleted, their supernatants decanted and re‐suspended with corresponding volumes of sterile PBS before being returned to agitation at 37°C. IL‐2 levels in collected supernatants were measured by ELISA (Invitrogen) and presented as a cumulative release curve. As harmine possesses a well‐characterized fluorescent spectra, ⁶¹ we determined the levels of harmine present at each time point by first measuring the fluorescence intensity of the sample (excitation 300 nm, emission 435 nm) against a known standard of harmine in aqueous solution (Figure S1).

2.5. NP‐cell conjugation

To optimize NP conjugation to the cell surface, we tested two protocols. In Protocol 1, NPs were conjugated to the surface of CTLL2 cells at a ratio of 4 mg NP to 20 × 10⁶ cells for 30 min in conjugation buffer (85 mM EDTA 11.9 mM Phosphate 129 mM NaCl) before undergoing Ficoll density gradient separation and repeated washes with HBSS to remove unbound NPs. In Protocol 2, newly fabricated NPs were incubated in a 0.01% solution of poly‐L‐lysine (PLL) in HBSS on a rotator at room temperature for 30 min. NPs were washed twice with HBSS to remove excess PLL, then incubated with CTLL2 cells or primary murine Tregs at a ratio of 4 mg NP to 20 × 10⁶ cells in conjugation buffer (85 mM EDTA 11.9 mM Phosphate 129 mM NaCl) at 37°C for 30 min with gentle inversion every 10 min. After incubation, the cells were washed twice with HBSS to remove the unbound NPs, at which point the cells were freshly utilized in the various assays described below.

2.6. IL‐2 NP‐conjugated CTTL2 cell viability and proliferation in vitro

IL‐2 dependent CTLL2 cells were labeled with CellTrace Violet (CTV) proliferation dye (ThermoFisher Scientific) according to manufacturer instructions, then conjugated to IL‐2 loaded NPs using either Protocol 1 or Protocol 2, or left unconjugated for positive and negative control conditions. Cells were plated in non‐adherent round bottom plates at 5 × 10⁵ cells/well in 2 ml complete RPMI media (cRPMI, RPMI1640 with 10% FBS and 1% Pen/Strep), then incubated for up to 1 week at 37°C, 5% CO₂. Positive control CTLL2 cells received 8 or 500 IU/ml soluble rhIL‐2 (ProSpec) on days 1 and 3 (as indicated in the figure legends) while negative control cells received neither soluble IL‐2 nor IL‐2 NPs. On the days indicated in the figures, all cells were collected from the plate and stained with eFluor 780 fixable viability dye (eBioscience). The percentage of live cells and proliferation were assessed by flow cytometry on a BD LSRFortessa as previously described. ²⁴ Additionally, live CTLL2 cells were enumerated using acridine orange and propidium iodide (AO/PI) dyes on a Cellometer Auto 2000 (Nexcelom) as indicated in the figure legends.

2.7. Human Treg isolation, expansion, and NP conjugation

T cells were isolated from leukopheresis products, obtained from healthy, anonymous donors at the LifeSouth Blood Bank. Peripheral blood leukopaks were enriched for CD4⁺ T cells with RosetteSep® Human CD4⁺ T Cell Enrichment Cocktail (StemCell Technologies) according to the manufacturer's instructions, then diluted with PBS at 1:1 ratio and layered over Ficoll‐Paque Plus (GE Healthcare). CD4⁺‐enriched cells were collected from the resulting interface, following density‐gradient centrifugation. Remaining red blood cells were lysed with ACK Lysing Buffer (Life Technologies) for 4 min at room temperature. T cells were washed with PBS, then stained with CD4‐PerCP/Cy5.5 (Clone: RPA‐T4, Biolegend), CD25‐BB515 (Clone: 2A3, BD Horizon™) and CD127‐PE (Clone: A019D5, Biolegend). Following washing with PBS, cells were resuspended in PBS/2%FBS at 20 × 10⁶/ml, and CD4⁺CD25^hiCD127^lo Tregs were FACS sorted, using an S3e cell sorter (BioRad). Freshly sorted Tregs were washed with PBS, counted and seeded at 0.5 × 10⁶ cells/ml in cRPMI supplemented with 600 IU/ml rhIL‐2 (National Institute of Cancer). Tregs were activated with anti‐CD3/CD28 beads (MACS GMP ExpAct Treg Kit, Miltenyi Biotec) at 4:1 bead to cell ratio. The culture was split and rhIl‐2 added every other day for 10–11 days. Expanded primary human Tregs were conjugated to IL‐2 NPs or BSA NPs using Protocol 1, then washed at least three times.

2.8. In vitro IL‐2 signaling in IL‐2 NP‐conjugated human Treg

NP conjugated Tregs were seeded in cRPMI at 1 × 10⁶/ml. To some BSA‐NP‐coupled cultures, exogenous soluble IL‐2 was added at 600 IU/ml as a positive control. On the indicated days, cells were stained for phosphorylated STAT5 (pSTAT5), using BD Biosciences Phosflow™ reagents and instructions. Briefly, cells were first stained with Live/Dead Near IR dye (Invitrogen), then CD25‐BB515 (Clone: 2A3, BD Horizon™). Following washing, cells were fixed by BD Cytofix™ (BD Biosciences) and permeabilized with BD Perm buffer III (BD Biosciences). Cells were stained pSTAT5‐Alexa Fluor® 647 (Clone: 47/Stat5 pY694, BD Biosciences).

2.9. In vitro stability and viability of IL‐2 NP‐conjugated human Treg

NP conjugated Tregs were re‐stimulated with anti‐CD3/CD28 Dynabeads® (1:1 ratio, Thermo Scientific) and seeded at 0.5 × 10⁶/ml in cRPMI. To some BSA‐NP‐coupled cultures, rhIL‐2 (600 IU/ml, National Institute of Cancer) was added every other day as positive control. On days 3, 4, and 5, cells were stained with Live/Dead Near IR dye and with CD25‐BB515, as described above. Intracellular staining was performed using the Foxp3 Staining Buffer Set (eBiosciences) according to the manufacturer's instructions. Cells were then stained with FOXP3‐AF647 (Clone: 206D, Biolegend) and Helios‐PE (Clone: 22F6, Biolegend).

2.10. Confocal microscopy

NPs were fabricated to encapsulate IL‐2 and BSA labeled with Alexa Fluor 488 (AF488), then conjugated to CTLL2 cells by either Protocol 1 or Protocol 2 described above. Conjugated cells were fixed and the cell surface and nuclei labeled with wheat germ agglutinin AF647 and DAPI, respectively. Confocal microscopy images were acquired on a Leica SP8 confocal laser‐scanning microscope. Conjugation efficiency was visually assessed from low magnification confocal microscopy images.

2.11. NOD Treg isolation and expansion

Spleens and lymph nodes (apical, axillary, and inguinal) were excised from 8‐week‐old female NOD or NOD BDC2.5 TCR transgenic mice and dissociated into a pooled single cell suspension. CD4⁺ cells were isolated via negative selection using the EasySep Mouse CD4 T Cell Isolation Kit (Stem Cell Technologies, Vancouver Canada). CD4⁺CD62L^HiCD25^Hi naïve Tregs were further isolated using an S3e Cell Sorter (Bio‐Rad, Hercules, CA) utilizing the following cell surface antibodies: CD4‐FITC (Clone RM4‐5, BioLegend), CD25‐PE (Clone PC61, BioLegend), and CD62L‐PE‐Cy7 (Clone MEL‐14, BioLegend). An aliquot of the sorted cells was stained with FoxP3‐APC (Clone FJK‐16s, BD Biosciences) using FoxP3/Transcription Factor Buffer kit (eBioscience) according to the manufacturer's instructions, and purity was assessed by flow cytometry on a BD LSRFortessa. Sorted cells were routinely 97–99% FoxP3⁺ (Figure S2). The remaining cells were expanded in Treg growth media (DMEM with 10% FBS plus 2000 U/ml rhIL‐2 [Teceleukin; Biogen]) using CD3/CD28 Dynabeads (Invitrogen) at a 3:1 bead to cell ratio with media replenished with additional rhIL‐2 on days 2, 4, 6, and 8. Cells were collected on Day 9 and purity verified by intracellular FoxP3 and Helios staining using the FoxP3/Transcription Factor Buffer set (eBioscience) and FoxP3‐AF488 and Helios‐BV421 (BD Bioscience). Tregs were then cryopreserved for future use or freshly utilized for in vivo studies described below.

2.12. IL‐2 NP‐conjugated Treg viability, proliferation, and phenotyping

Naïve polyclonal NOD Tregs were labeled with CTV then conjugated to IL‐2 NPs or BSA NPs via Protocol 2 as described above, plated in non‐adherent round bottom plates at 2.5 × 10⁵ cells/well in 1 ml media, and incubated at 37°C, 5% CO₂ for 4 days in triplicate. Unconjugated positive control Tregs received 2000 IU/ml soluble IL‐2 on day 0 while negative controls received no IL‐2 supplementation. In all groups, cell culture media was replaced with fresh media containing no soluble IL‐2 after 24 h. Each day, all Tregs were collected and stained with eFluor 780 fixable viability dye as described above, as well as for the intracellular transcription factor FoxP3 using the FoxP3‐APC (Clone FJK‐16s, BD Biosciences) as described above for flow cytometric analysis of Treg viability (dye exclusion), proliferation (CTV dye dilution), and phenotypic stability (FoxP3 MFI), as previously described. ⁶² , ⁶³

2.13. NP‐conjugated Treg migration assay

IL‐2 NP‐conjugated or unconjugated NOD Tregs were seeded at 5 × 10⁵ cells in 100 μl media/well in the upper chamber of transwell bare‐filter tissue culture well inserts containing polycarbonate membranes (6.5 mm diameter, 3 μm pore size). The lower chamber held 500 μl migration media containing 200 ng/ml CCL21 (Peprotech). Cells were incubated at 37°C, 5% CO₂ for 18 hours, at which time, the number of live cells that had migrated to the bottom chamber were counted using a hemocytometer and trypan blue exclusion.

2.14. IL‐2 NP‐Polyclonal Treg T1D prevention study

IL‐2 NPs were conjugated to polyclonal NOD Tregs using Protocol 2. Female 8‐week‐old NOD mice (Jackson Laboratory) were randomized into three treatment groups: IL‐2 NP‐Tregs (7.5 × 10⁵ cells, n = 13), unconjugated Tregs (7.5 × 10⁵ cells, n = 13), or saline (n = 15). Treatments were administered in 200 μl PBS by tail vein injection. Animals were monitored weekly via tail prick for blood glucose using a one‐touch glucose monitor until they reached 25 weeks of age or a diagnosis of diabetes, defined as blood glucose ≥240 mg/dl at screening followed by confirmatory blood glucose ≥240 mg/dl taken 24 h later.

2.15. IL‐2/harmine NP‐NOD BDC2.5 Treg T1D Prevention Study

IL‐2/harmine NPs were conjugated to NOD BDC2.5 TCR transgenic Tregs using Protocol 2. Normoglycemic, female 12‐week‐old NOD mice (Jackson Laboratory) were randomized into three treatment groups: IL‐2/harmine NP‐BDC2.5 Tregs (1 × 10⁶ cells, n = 11), unconjugated IL‐2/harmine NPs (n = 9), and saline (n = 12). For the unconjugated IL‐2/harmine NP group, the NP mass was identical to the mass of NPs used to conjugate 1 x 10⁶ BDC2.5 Tregs and represents as accurate a particulate control as possible. Two mice from both the IL‐2/harmine NP‐BDC2.5 Treg group and the unconjugated IL‐2/harmine NP group were excluded from the study as they were found to be hyperglycemic (blood glucose >240 mg/dl) at enrollment. Treatments were administered in 200 μl PBS by tail vein injection. Animals were monitored weekly for blood glucose until they reached 27 weeks of age or a diagnosis of diabetes, as described above.

2.16. Ethics and approval

Experiments using animals were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Florida. Experiments using human cells from purchased blood products (LifeSouth Blood Bank) were approved by the University of Florida Institutional Review Board (IRB) as non‐human exempt research.

2.17. Statistics

Figures were generated using GraphPad Prism version 9.3.1 (GraphPad Software, San Diego, CA, USA). Data were analyzed by One way ANOVA with Bonferroni or Tukey's post hoc test for multiple comparisons, multiple t tests, or Mantel–Cox log‐rank test with post hoc Bonferroni correction as indicated in the figure legends. p values <.05 were considered significant.

3. RESULTS

3.1. PLGA/PEG‐Mal NPs provide sustained release of IL‐2 alone or with harmine

Broadly biocompatible polymers, such as PLGA, provide a modular mechanism for the delivery of both hydrophobic and hydrophilic therapeutic agents for continuous release. ⁵⁰ , ⁶⁴ We employed a water–oil–water (w‐o‐w) double emulsification technique to fabricate poly(lactide‐co‐glycolide)‐b‐poly(ethylene glycol) maleimide (PLGA/PEG‐Mal) NPs loaded with hydrophilic IL‐2 or co‐loaded with both IL‐2 and hydrophobic harmine. We theorized that the PLGA/PEG‐Mal formulation would support subsequent cellular conjugation of NPs via the formation of thioether linkages to thiols on the cell surface. The fabrication process produced IL‐2 PLGA/PEG‐Mal NPs between 350 and 500 nm in diameter, with the volumetric majority at 419 nm (Figure 1A, black line). Co‐loading of IL‐2 and harmine produced slightly larger particles (350 and 650 nm) with the volumetric majority at 472 nm (Figure 1A, red line). We next measured the zeta potential of NPs in solution, which indicates the degree of electrostatic repulsion between adjacent, similarly charged particles in a dispersion. It is generally recognized that a zeta potential value >30 or <−30 mV is desired for an electrostatically stabilized suspension. ⁶⁵ The zeta potentials for IL‐2 and IL‐2/harmine NPs measured −69.2 and −61, respectively, indicating their stability and resistance to aggregation in solution. Scanning electron microscopy (SEM) visually confirmed the size, shape, and morphology of the NPs (Figure 1B,C).

FIGURE 1.

Physical characteristics of nanoparticles (NPs) encapsulating IL‐2 and IL‐2/harmine. (A) NPs were re‐suspended at 750 μg/ml in deionized water (diH₂O) and their average size determined using the Malvern Zetasizer Ultra (n = 7). Scanning electron microscopy (SEM) images of (B) IL‐2 NPs and (C) IL‐2/harmine NPs were captured under vacuum using a SU5000 Schottky Field Emission SEM (Hitachi) at 6.0 kV, 12.9 mm × 10.0 k magnification revealing intact NPs with uniform shape and size on a porous filter membrane (scale bars = 5 μm with hash marks at 0.5 μm intervals)

We next assessed the in vitro release kinetics of IL‐2 and harmine from NPs to determine whether the respective payloads would be released in a sustained fashion at levels considered sufficient to promote Treg survival and β‐cell regeneration in vivo. IL‐2 loading was nearly 10 times greater in the IL‐2/harmine NPs resulting in an increased burst release over the initial 24 h of the in vitro release assay (Figure 2A,B). However, IL‐2 was continuously released from both IL‐2 (Figure 2A) and IL‐2/harmine NPs (Figure 2B) over 7 days. Harmine was also continuously released over the seven‐day period (Figures 2C and S1). We hypothesized that the localized release of IL‐2, from both formulations, beyond 24 h would be sufficient to promote Treg survival. ³⁹

FIGURE 2.

Release kinetics of IL‐2 and harmine from PLGA/PEG‐Mal NPs over the course of 1 week. (A,B) Supernatant IL‐2 levels were measured via ELISA. (C) Harmine levels were measured via fluorescent intensity against a known harmine standard curve (excitation: 300 nm, detection: 435 nm). Data are plotted as mean cumulative release with standard deviation (IL‐2 NP, n = 6; IL‐2/Harmine NP, n = 3)

3.2. PLGA/PEG‐Mal IL‐2 NPs promote CTLL2 cell proliferation as well as IL‐2 receptor signaling on human Treg

IL‐2 NP were next conjugated to the surface of IL‐2‐dependent CTLL2 cells, and in vitro survival and proliferation assessed as compared to CTLL2 cells cultured in the presence or absence of soluble IL‐2 (Figure 3). CTLL2 cells were co‐labeled with CellTrace Violet (CTV) proliferation dye and eFluor 780 fixable viability dye prior to NP‐cell coupling. To account for the burst release of IL‐2 from the NPs, all groups were washed to remove any soluble IL‐2 on day 1. Importantly, soluble IL‐2 (8 IU/ml) was provided at a level that approximated the amount of IL‐2 released from IL‐2 NP between 24 and 72 h (Figure 2A), six times lower than the amount of soluble IL‐2 typically required to maintain the CTLL2 cell line, and soluble IL‐2 was replenished on days 1 and 3. At this low IL‐2 level, neither the IL‐2 NP nor soluble IL‐2 condition was sufficient to promote robust cell survival; however, approximately 40% of IL‐2 NP‐coupled cells remained viable in culture for 7 days. Dye dilution peaks indicative of cellular proliferation were comparable among IL‐2 NP‐coupled cells (Figure 3A) and the soluble IL‐2 positive control condition (Figure 3B) while CTLL2 cells deprived of IL‐2 did not divide (Figure 3C). To support their translational potential, IL‐2 NPs were next conjugated to expanded primary human Tregs. Importantly, IL‐2 NPs facilitated IL‐2 receptor (IL‐2R) signaling in vitro, as evidenced by STAT5 phosphorylation initially comparable to positive control Tregs cultured in the presence of high levels of soluble IL‐2 (600 IU/ml), but by day 3 in culture, pSTAT5 levels were comparable between IL‐2 NP and BSA NP bound cells (Figure 4A,B). With this, IL‐2R signal strength was sufficient in both intensity and duration to support phenotypic stability (Figure 4C,D) and viability (Figure 4E) of human Tregs for at least 4 days following re‐activation with anti‐CD3/CD28 coated beads. By day 5, however, both the viability and proportion of FOXP3⁺Helios⁺ thymically derived Tregs:FOXP3⁺Helios⁻ cells, which may include peripheral Tregs and conventional T cells, were significantly lower for IL‐2 NP Tregs versus soluble IL‐2 positive controls (Figure 4C–E). Hence, IL‐2 NPs coupled to the cell surface promoted survival, proliferation, and IL‐2R signaling in IL‐2 dependent CTLL2 cells as well as human Tregs, yet the efficiency was lower than expected, potentially suggesting insufficient NP conjugation and/or unequal distribution of NPs on daughter cells.

FIGURE 3.

IL‐2 remains biologically active following encapsulation in nanoparticles (NPs). To assess sustained delivery of functional IL‐2, IL‐2‐loaded NPs were conjugated to the IL‐2‐dependent CTLL2 cell line. 5 × 105 cells were cultured for 7 days in 2 ml media. The media was replaced after 24 h to remove the contribution of burst released IL‐2 from cell‐conjugated NPs. CTLL2 cell proliferation was evaluated by CTV dye dilution via flow cytometry on days 0 (red), 2 (orange), 4 (green), and 7 (purple). (A) The fluorescent signal in IL‐2 NP conjugated cells diminished over the course of the study in a similar manner to (B) cells supplemented with soluble IL‐2, indicative of cellular proliferation. (C) CTTL2 cells did not divide in the absence of IL‐2

FIGURE 4.

Cell‐conjugated IL‐2 NP facilitate IL‐2 signaling to maintain phenotypic stability and viability of human Tregs. (A,B) Previously expanded human Tregs were conjugated with IL‐2 NP or BSA NP as a control. NP conjugated cells were cultured at 1 × 106/ml without or with soluble IL‐2 (sIL‐2, 600 IU/ml) as indicated on the figure. STAT5 phosphorylation (pSTAT5) was measured by flow cytometry on days 1, 2, and 3 then normalized to fluorescence minus one (FMO) controls (n = 6, One way ANOVA with Tukey's multiple comparison test). (C–E) Previously expanded human Treg were conjugated with IL‐2 NP or BSA NP as a control, then re‐stimulated with anti‐CD3/CD28‐coated beads and cultured at 0.5 × 106 cells/ml without or with sIL‐2 (600 IU/ml). Treg stability (percentage of live cells expressing FOXP3 and Helios) and viability (dye exclusion) were assessed by flow cytometry on days 3, 4, and 5 (n = 4, One way ANOVA with Bonferroni multiple comparison test). Error bars represent SEM of the mean. **p < .01; ***p < .001; ****p < .0001

3.3. Poly‐L‐lysine enhances PLGA/PEG‐Mal NP‐cell coupling efficiency

To visually confirm NP‐cell coupling, fluorescent dye‐labeled PLGA/PEG‐Mal IL‐2 NPs were coupled to CTLL2 cells, then imaged by confocal microscopy revealing variable NP density on coupled cells (Figure 5A–C). There is a large body of literature demonstrating that the positively charged synthetic homopolymer, PLL, can be used to augment adhesion of various negatively charged substrates to the surface of cells, as recently reviewed. ⁶⁶ Hence, we predicted that PLL would enhance the electrostatic interaction between the cell membrane and the NPs and thereby, promote Maleimide‐mediated covalent coupling of the NP to cell surface thiol groups. ⁶⁷ Indeed, inclusion of PLL in the NP‐cell coupling process dramatically increased NP conjugation resulting in high NP density on coupled cells (Figure 5D–F). Moreover, PLL‐mediated conjugation of PLGA/PEG‐Mal IL‐2 NP promoted the expansion of CTLL‐2 cells in vitro resulting in an increased number of viable cells on day 6 as compared to CTLL‐2 cells coupled to IL‐2 NP without PLL and soluble IL‐2 control (Figure 5G). Therefore, PLL was included in the PLGA/PEG‐Mal NP‐cell coupling protocol for all subsequent experiments.

FIGURE 5.

Incorporation of poly‐L‐lysine (PLL) into nanoparticle (NP) conjugation protocol dramatically increases conjugation efficiency and CTLL2 cell viability. IL‐2 loaded NPs, co‐loaded with AF488 fluorescent dye labeled BSA (green), were conjugated to CTLL2 cells (cell membrane labeled with wheat germ agglutinin AF647 [red], nucleus labeled with DAPI [blue]) in the absence (A–C) or presence (D–F) of PLL, and assessed by confocal microscopy 24 h after conjugation. (G) 5 × 105 cells were cultured for 6 days in 2 ml media. CTLL2 cells conjugated to IL‐2 NPs in the presence of PLL (triangle down) were more abundant in culture as compared to CTLL2 cells conjugated to IL‐2 NPs in the absence of PLL (triangle up) or unconjugated CTLL2 cells maintained in the presence (square) or absence (circle) of soluble recombinant IL‐2 (500 IU/ml). Live cells were counted by trypan blue dye exclusion using an automated cellometer on the indicated days (n = 1)

3.4. IL‐2 NPs conjugation maintains Treg viability without negatively impacting cell migration

IL‐2 NPs were coupled to the cell surface of CTV‐labeled Tregs from NOD mice and assessed for viability, proliferation, and phenotypic stability in vitro as compared to Tregs cultured in the presence or absence of soluble IL‐2. Bovine serum albumin (BSA) loaded NPs were conjugated to NOD Tregs as an additional negative control. IL‐2 NP conjugation promoted Treg survival and proliferation over the course of 4 days following removal of burst release IL‐2 on day 1 (Figure 6A,B). Additionally, IL‐2 NP coupled Tregs displayed stable expression of the lineage‐defining transcription factor, FoxP3 (Figure 6C).

FIGURE 6.

Survival, proliferation and phenotypic stability of Tregs following conjugation to IL‐2 NPs. CTV‐labeled NOD Tregs were conjugated to IL‐2 NPs (triangle up), BSA NPs (triangle down), or left unconjugated in the presence (square) or absence (circle) of soluble IL‐2 (2000 IU/ml). 2.5 × 105 cells were cultured for 4 days in 1 ml media. Cells were assessed daily by flow cytometry for (A) viability via dye exclusion, (B) proliferation via CTV dye dilution, and (C) FoxP3 mean fluorescence intensity (MFI; n = 1)

Stephan, et al. previously reported that PEG‐Mal NP conjugation, performed in the absence of PLL, did not impair the in vitro migration efficiency of CD8⁺ T cells. ⁴⁶ Considering the high density of IL‐2 NPs on the cell surface following coupling in the presence of PLL (Figure 5), we assessed the impact of NP conjugation on Treg migratory capacity in vitro with potential implications for extravasation in vivo. IL‐2 NP‐coupled Tregs or Tregs alone were seeded onto transwell tissue culture inserts with the lower chamber containing the Treg chemoattractant, CCL21. ⁶⁸ Importantly, after 18 h, no significant difference in Treg migration efficiency was observed (Figure 7) supporting the hypothesis that IL‐2 NP conjugation would enhance Treg survival and function, but not interfere with their ability to migrate to sites of inflammation when administered as a therapeutic.

FIGURE 7.

Retention of Treg migratory capacity following nanoparticle (NP) conjugation. (A) Scheme of the experimental setup. 5 × 105 unconjugated or IL‐2 NP conjugated polyclonal non‐obese diabetic Tregs were plated onto transwell, bare‐filter tissue culture inserts (3 μm pore size). Tregs were allowed to migrate for 18 h into the bottom reservoir containing 200 ng/ml CCL21. (B) After 18 h, cells were collected from lower (checkered) and upper (gray) chambers, and live cells were enumerated by trypan blue exclusion using an automated hemocytometer. Migration efficiency was comparable between unconjugated (Treg) and IL‐2 NP conjugated Treg (NP‐Treg, n = 3, p = .095 upper, p = 0.372 lower, multiple t tests)

3.5. Optimization of NP‐Treg ACT to prevent T1D onset in NOD mice

To investigate their therapeutic potential to prevent T1D onset, polyclonal NOD Tregs were coupled to IL‐2 NPs and administered intravenously to 8‐week‐old, pre‐diabetic female recipient NOD mice. IL‐2 NP conjugated polyclonal Tregs did not prevent or delay diabetes onset as compared to the unconjugated polyclonal Treg and saline control groups (Figure 8A). We speculated that this failure to disrupt autoimmune T1D pathogenesis might be due to the lack of specificity for homing to the insulin‐producing pancreatic islets. To address this, we utilized BDC2.5 transgenic NOD T cells, which express a pancreatic islet‐specific TCR that promotes cell trafficking to the pancreatic islets. ⁶⁹ The NP‐Treg formulation was further optimized by dual loading of IL‐2 with harmine, which stimulates insulin‐producing β‐cell replication. ⁷⁰ , ⁷¹ IL‐2/harmine NPs were coupled to BDC2.5 Tregs and administered to 12‐week‐old, pre‐diabetic female NOD mice, representing late stage intervention when significant insulin‐producing cell loss has already occurred. ⁷² While unconjugated IL‐2/harmine NPs did not prevent diabetes, animals that received IL‐2/harmine NP‐coupled BDC2.5 Tregs had significantly reduced T1D incidence versus the saline control group (Figure 8B).

FIGURE 8.

Intravenous infusion of NPs conjugated Tregs for diabetes prevention in NOD mice. (A) Female 8‐week‐old NOD mice received 7.5 × 105 IL‐2 NP‐conjugated polyclonal NOD Tregs (blue), unconjugated polyclonal NOD Treg (purple), or saline (gray) via tail vein injection. Blood glucose levels were monitored weekly until 25 weeks of age. No significant difference in diabetes incidence was observed (p = 0.773, Mantel–Cox log‐rank test). (B) 1 × 106 NOD BDC2.5 Tregs conjugated to IL‐2/harmine NPs (green), unconjugated IL‐2/harmine NPs (blue), or saline (gray) were injected into the tail vein of normoglycemic female NOD mice at 12 weeks of age. Blood glucose levels were monitored weekly for until 27 weeks of age. Diabetes progression was significantly different across the three treatment groups (p = .011, Mantel–Cox log‐rank test), with the largest difference between IL‐2/harmine NP‐Treg versus saline (**p = .0029, pair‐wise Mantel–Cox log‐rank test with post hoc Bonferroni correction for multiple comparisons)

4. DISCUSSION

Current therapeutic interventions have failed to reverse disease after onset, perhaps, in part due to the notion that T1D presents with two distinct therapeutic challenges. Specifically, there is a need to correct the underlying autoimmune pathogenesis, coupled with a need to restore endogenous β‐cell mass. ⁷³ These two challenges provided the rationale to generate engineered Tregs capable of exerting tissue‐targeted immunoregulation, along with the delivery of a tunable payload of therapeutic agents that could improve engraftment and restore β‐cell function.

Tregs have emerged as a form of “living drug” for the correction of autoimmunity and have been demonstrated to exert protection from a broad array of immune‐mediated diseases in multiple model systems by limiting innate and adaptive effector mechanisms. ⁷⁴ , ⁷⁵ , ⁷⁶ , ⁷⁷ , ⁷⁸ However, there remain a number of challenges to optimizing Tregs to direct both their specificity and function. Studies from the NOD mouse model highlight the need for antigen‐specificity to exert potent immunoregulatory function in late stages of disease. ¹⁶ , ¹⁷ , ⁷⁹ Furthermore, there is an emerging notion that the process of in vitro expansion may make Tregs highly dependent on exogenous IL‐2, which upon adoptive transfer into an immune replete host, would place Tregs in the position of insufficient survival signal for maintenance. ⁷⁸ Indeed, Bluestone and colleagues demonstrated that adoptively transferred Tregs peaked in peripheral circulation on day 7–14 followed by a rapid loss with sampling over a period of 1 year. ¹⁸ A follow‐up trial demonstrated improved Treg survival when ACT was performed in combination with low‐dose systemic IL‐2 administration, but the observed off‐target expansion of activated NK, mucosal associated invariant T (MAIT), and CD8⁺ T cells, remains of concern. ⁴⁴ Thus, we first sought to determine if we could provide Tregs with a “backpack” of IL‐2, as a means to provide exogenous growth factor in an autocrine manner to improve initial engraftment and mitigate expansion of cytotoxic populations. For these studies, we iterated a fabrication process that resulted in biocompatible NPs of acceptable sizes and electrostatic properties allowing for robust conjugation to the surface of viable murine Tregs.

We next set out to add therapeutic agents that could capitalize on the ability of Tregs to traffic to sites of inflammation (in general), and more specifically be retained in sites when activated through their cognate TCR. ⁸⁰ , ⁸¹ While a number of agents, including harmine, have been demonstrated to induce β‐cell proliferation, ⁵⁸ , ⁷⁰ this is not without the potential for central nervous system side‐effects associated DYRK1A inhibition. Thus, we sought to harness a biomaterials approach to deliver both the autocrine growth factor IL‐2 along with the tissue restricted delivery and localized release of harmine in vivo. We generated PLGA/PEG‐Mal NPs that could be simultaneous loaded with bioactive IL‐2 and the drug harmine for continuous elution of both molecules for at least 1 week.

Initial efforts at cell surface labeling through PEG‐Mal linkages did not provide sufficient quantities of IL‐2 to support robust CTLL2 cell survival in vitro. To increase the efficiency of NP conjugation, we optimized our protocol by increasing adsorption through PLL coating prior to incubation with Tregs. This resulted in viable Treg cultures that retained their capacity to traffic and proliferate, thus, providing an opportunity to enhance Treg function beyond endogenous pathways for tissue regeneration and localized release of agents that would otherwise have potentially deleterious off‐target side effect profiles. Interestingly, we note that the IL‐2 and IL‐2/harmine NPs possessed slightly different size distributions, zeta potential and morphologies. We speculate that the increased size of the co‐loaded IL‐2/harmine NPs was likely due to increased IL‐2 loading (supported by greater burst release over the first 24 h) and the incorporation of ethanol in the methylene chloride (oil phase) as the solvent of the hydrophobic harmine molecule in the w‐o‐w double emulsification process.

We next evaluated whether the NP‐coupled Tregs could be used as a means to impact autoimmune diabetes development in the NOD mouse model. For these experiments, we tested the impact of polyclonal Tregs linked with IL‐2 NPs, as well as antigen‐specific BDC2.5 Tregs linked with IL‐2/harmine NPs as an initial test of concept. While infusion of polyclonal Tregs conjugated to IL‐2 NPs did not prevent diabetes onset in NOD mice, the IL‐2/harmine NP formulation, when conjugated to islet antigen‐specific Tregs, demonstrated the potential to slow or halt progression of T1D. There are technical limitations to generating sufficient numbers of Tregs in mice to test all conditions in parallel including attrition at each processing step, along with limitations on murine Treg expansion and increased plasticity ¹⁷ that are not observed to the same degree with human Tregs. ⁶³ , ⁸² For these reasons, the current study lacked a BDC2.5 Treg only control. Nonetheless, these data demonstrate that BDC2.5 Tregs conjugated to NPs do not interfere with the ability of Tregs to prevent and reverse autoimmune diabetes in NOD mice. ¹⁶ , ²⁵ , ²⁷ Hence, while we cannot currently conclude that our conjugated IL‐2/harmine NP‐BDC2.5 Tregs are optimal over unconjugated antigen‐specific Tregs alone, these studies offer a modular platform for other investigators to test the capacity of NPs conjugated to Tregs to augment various aspects of tissue repair and regeneration. Further research is needed to determine the optimal number of expanded Tregs, the effects of harmine on Treg survival and function, and potentially, additional drugs loaded in cell conjugated NPs to significantly prevent and/or reverse diabetes. Moreover, the capacity of NP‐Tregs to traffic to the pLN and pancreas, and induce in vivo β‐cell proliferation will require additional confirmation through detailed mechanistic studies.

5. CONCLUSION

The findings presented here demonstrate an early proof‐of‐concept that controlled release of biologics and drugs from biomaterials can be used in combination with antigen‐specific Tregs to augment/enhance cellular function in settings of immune tolerance induction and regenerative medicine. The broad capacity to encapsulate both hydrophobic and hydrophilic agents in PLGA thus enables a combinatorial therapeutic approach. We note that we have previously demonstrated the ability to redirect the specificity of human and mouse Tregs through either TCR gene transfer ²⁴ or through chimeric antigen‐receptor (CAR) delivery. ⁸³ Thus, our approach has the potential to extend the therapeutic ability of Tregs to deliver a payload of drugs in a tissue‐targeted and localized manner to avoid toxicities that may result from systemic administration.

AUTHOR CONTRIBUTIONS

Gregory Marshall: Methodology, Validation, Investigation, Data Curation, Formal Analysis, Writing‐Original Draft, Writing‐ Reviewing & Editing, Visualization, Project Administration, Funding Acquisition. Judit Cserny: Investigation, Data Curation, Writing‐ Reviewing & Editing, Visualization. Chun‐Wei Wang: Methodology, Validation, Investigation, Data Curation, Writing‐ Reviewing & Editing. Benjamin Looney: Investigation, Data Curation, Writing‐Original Draft, Writing‐ Reviewing & Editing, Visualization. Amanda L. Posgai: Formal Analysis, Writing‐Original Draft, Writing‐Reviewing and Editing, Visualization. Rhonda Bacher: Formal Analysis, Writing‐Reviewing & Editing. Benjamin Keselowsky: Conceptualization, Methodology, Resources, Writing‐Reviewing & Editing. Todd M. Brusko: Conceptualization, Methodology, Resources, Writing‐Reviewing & Editing, Supervision, Project Administration, Funding Acquisition.

ACKNOWLEDGEMENTS

These studies were supported by the National Institutes of Health (NIH) small business innovation research (SBIR) grant R43 AI131850 and R21 AI133067 to Gregory Marshall, as well as P01 AI042288 to Todd M. Brusko with additional funds from a JDRF Pilot Innovation Award (5‐2011‐469) and pilot and feasibility grant from the Diabetes Research Connection to Todd M. Brusko. The funders had no role in the design or conduct of the study, nor the decision to prepare and submit the manuscript for publication.

CONFLICT OF INTEREST

At the time the research was conducted, Gregory Marshall, Chun‐Wei Wang, and Benjamin Looney were employees, Benjamin Keselowsky an advisor, and Todd M. Brusko the chief operating officer of OneVax, LLC (now InspiraTherapeutics, Inc), a preclinical biotechnology company focusing on the use of biomaterials‐based micro‐ and nanoparticles for the delivery of tolerogenic immunotherapies. We also acknowledge the presence of a patent (#9913830) pertaining to the use of NP coupled Treg cell therapy for treatment of immune and/or autoimmune disorders by Todd Brusko and Benjamin Keselowsky. The remaining authors declare no conflicts of interest.

Supporting information

Appendix S1 Supplementary Information

Marshall GP, Cserny J, Wang C‐W, et al. Biomaterials‐based nanoparticles conjugated to regulatory T cells provide a modular system for localized delivery of pharmacotherapeutic agents. J Biomed Mater Res. 2023;111(2):185‐197. doi: 10.1002/jbm.a.37442

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.