Supra-lacrimal protein-based carriers for Cyclosporine A reduce Th17-mediated autoimmunity in murine model of Sjögren’s Syndrome

Abstract

Sjögren’s syndrome (SS) is a multifactorial autoimmune disease with principal symptoms including inflammation and loss of function of lacrimal glands (LG) and salivary glands. While glandular infiltrates includes both B- and T- cells, CD4+ T cells are strongly implicated. Utilizing the male non-obese diabetic (NOD) mouse model of SS, this work: 1) identifies clinically-relevant elevations in cytokines (IL-17A, IL-2) in LG-derived CD4+ T cells; and 2) explores tissue-specific immunosuppression of SS using a novel protein-based drug carrier to concentrate cyclosporine A (CsA) directly in the LG. A potent immunosuppressant, topical ophthalmic CsA is approved for dry eye disorders; however, it cannot effectively resolve inflammation due to limited accumulation in the LG. Systemic CsA has dose-limiting side effects that also limit its ability to block LG inflammation. Using elastin-like polypeptides (ELPs) fused genetically to cyclophilin, the intracellular cognate receptor of CsA, this manuscript reports a sustained-release formulation of CsA that maintains therapeutic drug concentrations in the LG and extends intervals between doses. This formulation blocked both in vitro Th17 cell differentiation and IL-17A secretion. In vivo treatment significantly decreased the abundance of Th17.1 cells, a helper cell population sharing phenotypes of both Th17 and Th1, in the LG of diseased NOD mice. Treatment with even a single dose of the sustained-release formulation was effective enough to improve basal levels of tear production. Thus, this sustained-release formulation suppressed local LG inflammation driven through IL-17 dependent pathways, while improving ocular surface function.

Introduction

Sjögren’s syndrome (SS) is a chronic systemic autoimmune disease affecting approximately 0.3% to 0.6% of the total population. SS is prevalent in 4 million individuals in the United States alone, with the highest female-to-male ratio of 9:1 of any known autoimmune diseases [1]. SS is characterized by lymphocytic infiltration of the exocrine glands, specifically lacrimal gland (LG) and salivary gland (SG) [2, 3]. This infiltration is associated with the principal clinical hallmarks of SS, persistent dry eyes and dry mouth, which leads to severe corneal damage and compromised oral health. SS patients may also experience systemic manifestations including interstitial nephritis [4], liver disease [5] and obstructive bronchitis [6] in its early stage. Continued development of the disease may lead to B cell lymphoma, resulting in serious morbidity or death [7]. SS-mediated dry eye disease (DED) manifestations and the underlying LG inflammation are the focus of this study.

Although B cell dysregulation in SS has been implicated in more severe manifestations in later courses of the disease, initial glandular tissue infiltration is dominated by T lymphocytes, of which 70% are CD4+ helper T (Th) cells [8]. Historically, Th cells have been categorized into two main subsets, Th1 and Th2, based on their distinct pattern of cytokine secretion [9]. Th1 cells are characterized by the secretion of IFN-γ and IL-2, but not IL-4 or IL-5. In contrast, Th2 cells are characterized by the secretion of IL-4, IL-5 and IL-13, but not IFN-γ or IL-2. This Th1 or Th2 fate is regulated by transcription factors T-bet or GATA3, respectively [10]. The pathogenesis of SS has been considered Th1-mediated, a premise supported by high levels of Th1 cytokines in SG [11] and serum [12] of SS patients. Previous studies have suggested that the balance between Th1/Th2 favors Th1 in the SG of SS patients [13]. The Th1/Th2 ratio, as represented by IFN-γ/IL-4 and TNF-α/IL-4 ratios in saliva, is significantly higher in primary SS patients than in non-SS DED patients [14].

More recently, the discovery of a distinct subset of helper T cells, which is characterized by increased secretion of IL-17A, has led to a re-examination of infiltrating helper T cells in SS [15]. This new subset has been named Th17 based on this signature cytokine. As a proinflammatory cytokine, IL-17A overexpression is implicated in the pathogenesis of many autoimmune diseases including multiple sclerosis [16, 17], rheumatoid arthritis [18, 19], inflammatory bowel disease [20], psoriasis [21, 22], systemic lupus erythematosus [23] and others. In pSS patients, abnormal expression of IL-17A was observed in SG lesions as well as in plasma [24, 25]. When immunized with SG protein as an autoantigen to induce experimental SS, IL-17A knockout (KO) mice were resistant to induction of SS-associated changes in the SG. Moreover, infusion with IL-17A-expressing Th17 cells expedited the onset of autoimmune sialoadenitis in these IL-17A KO mice, causing markedly reduced salivary secretion and profound SG inflammation [26]. Notably, IL-17A production has also been implicated in SS development in CD25 knockout mice [27] and in a non-obese diabetic (NOD)-derived SS disease model [28].

Initially, T helper cell type 17 (Th17) cells were reported as the only IL-17A secreting lineage in CD4+ T cells [29-31]. Later, IL-17A-producing CD4+ T cells were recovered from synovial fluid of patients with juvenile idiopathic arthritis and analyzed for their cytokine and surface marker expression. Relative to Th17 cells, these synovial IL-17A-producing cells exhibited uniform Chemokine Receptor 6 (CCR6) expression but distinct CCR4 expression. Some of these cells with an intermediate phenotype between Th1 and Th17 expressed both IL-17A and IFN-γ [32]. These IFN-γ-expressing Th17 cells, or Th17.1 (also called Th17/1, Th1Th17 or Th1/Th17) were also identified in the gut of patients with Crohn's disease [33] and in the central nervous system during experimental autoimmune encephalomyelitis [34, 35]. The majority of IL-17A-secreting cells in the joints of arthritis patients were also found to express cytokines characteristic of both Th17 and Th1 cells [36]. TGF-β may be a key regulator responsible for this developmental plasticity of Th17 cells. Th17 cells require TGF-β for sustained expression of IL-17F and IL-17A, two prominent members of the IL-17 family. The expression of IFN-γ is enhanced in the absence of TGF-β, leading to the Th17.1 phenotype [37]. The phenotypic conversion from Th17 to Th17.1 cells has been demonstrated both in vitro [36] and in vivo [38].

Cyclosporine A (CsA) exerts pharmacological activity upon binding to its cytoplasmic cognate receptor, cyclophilin. This complex interferes with the phosphatase activity of calcineurin, impeding the dephosphorylation of nuclear factor of activated T cells (NFAT). This event interrupts proinflammatory cytokine release, in particular of IL-2, and reduces lymphocyte proliferation [39]. Although Th1 cells are often thought to be its primary target, CsA is reported to inhibit IL-17A production [40-42] and to attenuate Th17 cells [43], likely via inhibition of T cell activation [44], supporting its potential use to modulate Th17 and related cells and reduce local IL-17 treatment for SS-associated DED. This possibility is explored in this study. However, a necessary additional step to test the potential of CsA to target LG-specific Th17 and related cells is the development of a formulation that allows CsA accumulation in the LG.

Derived from human tropoelastin, elastin-like polypeptides (ELPs) are biocompatible, biodegradable and of low immunogenicity [45], which makes them prospective candidates for developing new biopharmaceutics. A characteristic property of ELPs is their thermal responsiveness, with an adjustable transition temperature (T_t) controlled by the choice of guest residue, X, and by the number of repetitive units, n, in their pentameric amino acid repeat sequence of (VPGXG)_n. When the temperature rises above the T_t, ELPs phase separate from their highly water-soluble form into coacervates [46]. They revert to the fully water-soluble state again when the temperature falls below the T_t. Previously, we reported the development of a construct for systemic delivery of CsA with ELPs [47]. Cyclophilin A (CypA), the cytosolic receptor of CsA, was fused to the N-terminus of an ELP called A192 with the amino acid sequence of G(VPGAG)₁₉₂Y. The resultant fusion protein, CA192, maintained nanomolar binding affinity to CsA. CA192-CsA exhibited a strong pharmacological efficacy both in vitro by inhibiting IL-2 secretion from activated Jurkat cells and in vivo by increasing stimulated tear production in NOD mice when given subcutaneously [47]. To extend the concept of ELP-mediated drug delivery based on CypA-CsA binding, the current report makes two significant enhancements: 1) improved drug loading capacity; and 2) sustained drug release through phase separation following supra-LG injection in response to physiological temperature.

The male NOD mouse is a well-established murine model of autoimmune dry eye disease (DED). This model recapitulates the lacrimal component of human SS by spontaneously developing a progressive autoimmune dacryoadenitis (inflammation of LG) at 8-12 weeks in age, which parallels reduction in tear flow and production of tear proteins characteristic of SS-associated DED [48-52]. In the current study, using this established model of SS-associated DED, we have identified IL-17A as a proinflammatory cytokine that is markedly and selectively increased in the LG in parallel with disease development. We further show that sustained release of CsA via the supra-LG injection of a novel ELP depot reduces Th17.1 cell infiltration of the LG and is accompanied by improvement of other manifestations of SS-associated DED.

1. Materials and Methods

1.1. Murine model

Both NOD/ShiLtJ mice (Stock No: 001976) and BALB/cJ mice (Stock No: 000651) were purchased from The Jackson Laboratory (Bar Harbor, ME). Animal use was in compliance with protocols approved by the University of Southern California Institutional Animal Care and Use Committee an in accordance with the Guide for the Care and Use of Laboratory Animals 8^th edition [53].

1.2. Lymphocyte isolation from LG, spleen and lymph nodes.

This isolation method was revised from methods in a previous report [54]. Sterile filtered solutions (0.2 μm) were prepared as follows: 1) s-Ham’s, which is Ham’s F12 medium (11765054, Gibco, Grand Island, NY), supplemented with penicillin (100 U/mL), streptomycin (100 μg/mL) (15070063, Gibco, Grand Island, NY), L-glutamine (2 mM) (25030081, Gibco, Grand Island, NY), n-butyric acid (2 mM) (B2503, Sigma-Aldrich, St. Louis, MO), linoleic acid (0.3 μM) (L1376, Sigma-Aldrich, St. Louis, MO), soybean trypsin inhibitor (50 μg/mL) (T9003, Sigma-Aldrich, St. Louis, MO), bovine serum albumin (BSA) (5 mg/mL) (A3912, Sigma-Aldrich, St. Louis, MO), and 2-[4-(2-hydroxyethyl)piperazin-1-yl]ethanesulfonic acid (HEPES) (10 mM) (11344041, Gibco, Grand Island, NY) adjusted to pH 7.6 with 1 M NaOH (221465, Sigma-Aldrich, St. Louis, MO); 2) CHD-Ham’s, which is s-Ham’s further supplemented with 350 U/mL collagenase (17100017, Gibco, Grand Island, NY), 300 U/mL hyaluronidase (LS02592, Worthington Biochemical Corp, Lakewood, NJ), and 40 kU/mL DNase (07469, STEMCELL Technology, Cambridge, MA); 3) s-Hank’s, which is Hank’s balanced salt powder (9.5g/L) (H4891, Sigma-Aldrich, St. Louis, MO), Ethylenediaminetetraacetic acid (EDTA) (2 mM) (E5124, Sigma-Aldrich, St. Louis, MO) and HEPES (10 mM) adjusted to pH 7.6 with 1 M NaOH in a volume of 1 L; 4) and negative selection buffer, which is 1X Dulbecco’s Phosphate Buffered Saline (PBS) (14190144, Gibco, Grand Island, NY), supplemented with 0.5% (w/v) BSA and 2 mM EDTA. To isolate lymphocytes from tissue samples, freshly isolated LGs were minced into 1 mm³ pieces and sequentially incubated in s-Hank’s and CHD-Ham’s in a shaking water bath set at 37°C for 10 min each. 3mL of each solution was used to digest a pair of LGs from one mouse. CHD-Ham’s incubation was repeated once to allow adequate enzymatic digestion. The supernatant was collected after every incubation. The pooled supernatant was filtered through a 10 μm cell strainer (43-50010-03, pluriSelect, El Cajon, CA) to enrich lymphocytes in the filtrate. After washing twice in PBS, enriched lymphocytes were resuspended either in negative selection buffer (480017, BioLegend, San Diego, CA) at a density of 10⁸ cells/mL for negative selection detailed in section 2.3 or in PBS at a density of 10⁷ cells/mL for flow cytometry analysis detailed in section 2.13.

For preparation of lymphocytes from lymph node or spleen for negative selection, freshly collected spleen or lymph nodes were placed in a 70 μm cell strainer pre-wetted with PBS on top of 50 mL centrifuge tube. The thumb side of a 10 ml syringe plunger was then used to smash the spleen and lymph nodes, while constantly adding up to 5 mL PBS. The isolated cells were washed twice with PBS and resuspended in negative selection buffer at a density of 10⁸ cell/mL.

1.3. Negative selection of CD4+ T cells.

CD4+ T cells were negatively selected from LG-derived lymphocyte and lymphoid tissues, such as spleen and lymph nodes, using the Mouse CD4+ T Cell Isolation Kit (480033, BioLegend Inc, San Diego, CA). Lymphocytes isolated as described above were first adjusted to ≤10⁸ cells/mL in negative selection buffer. A biotin-antibody cocktail composed of biotin-labeled anti-CD8a, CD11b, CD11c, CD19, CD24, CD45R/B220, CD49b, CD105, I-A/I-E (MHC II), TER-119/Erythroid, and TCR-γδ was diluted 1:10 into cell suspension. After 15 min incubation on ice, an equal volume of freshly vortexed streptavidin magnetic nanobeads was added into the cell suspension, followed by another 15 min incubation on ice. The cell suspension was then transferred to 12 x 75 mm round bottom polystyrene tubes, in which the total volume of cell suspension was increased to 2.5 mL with negative selection buffer. The tube was placed in the magnet (MAG-4902-10, ThermoFisher Scientific Inc., Waltham, MA) for 5 min for magnetic separation, prior to collection of fluid containing unbound cells. This process was repeated with another 2.5 mL of negative selection buffer. The unbound fractions, which are enriched CD4+ T cells, were pooled.

1.4. mRNA purification and quantitative real-time PCR

mRNA was extracted from negatively isolated CD4+ T cells using the RNeasy Mini kit (74104, Qiagen, Germantown, MD), followed by reverse transcription using TaqMan Reverse transcription Reagents (N8080234, ThermoFisher Scientific Inc., Waltham, MA). Gene expression levels were measured using TaqMan gene expression assays on a QuantStudio 6K Flex Real-time PCR system (ThermoFisher Scientific Inc., Waltham, MA) with the following primers: IL-17A (Mm00439618_m1), IL-2 (Mm00434256_m1), aquaporin 5 (Mm00437578_m1) and GAPDH (Mm99999915_g1). GAPDH was used as an endogenous control. The following equations were used to calculate relative quantification (RQ):

$$\Delta C_t=C_t(targetgene)-C_t(GAPDH)$$ (Eq.1)

$$\Delta \Delta C_t=\Delta C_t(exprimentalgroup)-\Delta C_t(controlgroup)$$ (Eq.2)

$$RQ(experimentalgroup∕controlgroup)=2^{-\Delta \Delta C_t}$$ (Eq.3)

A serial of nonparametric Kruskal-Wallis tests was conducted to capture significant difference in gene expression levels in CD4+ T cells isolated from the same tissue but different strains of mice and from different tissues in the same group of mice. Dunn’s multiple comparison test was then used for multiple comparison.

1.5. In vitro differentiation of Th17-like cells

Tissue culture-treated plates were first coated with 5 μg/mL anti-mouse CD3ε antibody (100314, BioLegend, San Diego, CA), at 37°C for 4 hr. Lymphocytes were harvested from mouse spleen through a 70 μm cell strainer as described above and resuspended in PBS. Then, these CD4+ T-cells were negatively isolated from splenocytes as described above. The isolated CD4+ T cells were resuspended into culture media (RPMI 1640 (11875093, ThermoFisher Scientific Inc., Waltham, MA) + 10% FBS (26140079, ThermoFisher Scientific Inc., Waltham, MA) + 10 mM HEPES + 1X Antibiotic-Antimycotic (15240062, ThermoFisher Scientific Inc., Waltham, MA) supplemented with 5 μg/ml anti-mouse CD28 antibody (102112, BioLegend, San Diego, CA), 50 ng/ml recombinant mouse IL-6 (575704, BioLegend, San Diego, CA), 5 ng/ml recombinant human TGF-β1 (580702, BioLegend, San Diego, CA), 10 ng/mL recombinant mouse IL-23 (589002, BioLegend, San Diego, CA), 10 μg/mL anti-mouse IL-4 antibody (504108, BioLegend, San Diego, CA), and 10 μg/mL anti-mouse IFN-γ antibody (505812, BioLegend, San Diego, CA) to a cell density of 0.8 - 1 x 10⁶/mL and incubated at 37°C with 5% CO₂ for 2 days. Fresh medium with the same supplements was added into each well to increase the initial volume by 50% and incubation was continued for another 2 days.

1.6. Biosynthesis and biophysical characterization of CAC and CVC

A96 and V96 plasmids were previously synthesized by recursive directional ligation in a modified pET-25b(+) vector, which contains a unique BseRI restriction sequence upstream of start codon and overlapping with the ribosome binding site [55]. Plasmids containing E. coli biased codons encoding N-term or C-term CypA flanked by restriction sites of NdeI (CA∣TATG) and BamHI (G∣GATCC) were purchased from Integrated DNA Technologies Inc. (IDT) (Coralville, IA). The N-term CypA sequence consists of:

5’- CA∣TATGATGGTTAACCCGACCGTTTTCTTCGACATCGCTGTTGACGGTGAACCGCTGGGTCGTGTTTCTTTCGAACTGTTCGCTGACAAAGTTCCGAAAACCGCTGAAAACTTCCGTGCTCTGTCTACCGGTGAAAAAGGTTTCGGTTACAAAGGTTCTTGCTTCCACCGTATCATCCCGGGTTTCATGTGCCAGGGTGGTGACTTCACCCGTCACAACGGTACCGGTGGTAAATCTATCTACGGTGAAAAATTCGAAGACGAAAACTTCATCCTGAAACACACCGGTCCGGGTATCCTGTCTATGGCTAACGCTGGTCCGAACACCAACGGTTCTCAGTTCTTCATCTGCACCGCTAAAACCGAATGGCTGGACGGTAAACACGTTGTTTTCGGTAAAGTTAAAGAAGGTATGAACATCGTTGAAGCTATGGAACGTTTCGGTTCTCGTAACGGTAAAACCTCTAAAAAAATCACCATCGCTGACTGCGGTCAGCTGGAAGG∣TTACTGATCTCCTCG∣GATCC-3’

The C-term CypA consists of:

5’-CA∣TATGGGTATGGTTAACCCGACCGTTTTCTTCGACATCGCTGTTGACGGTGAACCGCTGGGTCGTGTTTCTTTCGAACTGTTCGCTGACAAAGTTCCGAAAACCGCTGAAAACTTCCGTGCTCTGTCTACCGGTGAAAAAGGTTTCGGTTACAAAGGTTCTTGCTTCCACCGTATCATCCCGGGTTTCATGTGCCAGGGTGGTGACTTCACCCGTCACAACGGTACCGGTGGTAAATCTATCTACGGTGAAAAATTCGAAGACGAAAACTTCATCCTGAAACACACCGGTCCGGGTATCCTGTCTATGGCTAACGCTGGTCCGAACACCAACGGTTCTCAGTTCTTCATCTGCACCGCTAAAACCGAATGGCTGGACGGTAAACACGTTGTTTTCGGTAAAGTTAAAGAAGGTATGAACATCGTTGAAGCTATGGAACGTTTCGGTTCTCGTAACGGTAAAACCTCTAAAAAAATCACCATCGCTGACTGCGGTCAGCTGGAAGGTTGATAATGATCTTCAG∣GATCC-3’

This allowed their ligation into the multiple cloning sites of a normal, modified pET-25b(+) vectors for the N-term, C-term sequences respectively. Within the N-term CypA sequence, a BseRI site (CTCCTC) was placed before the BamHI site to allow further ligation with the ELP-encoding sequence. The CypA sequence was ligated into the modified vector containing the BseRI site overlapping with the ribosome binding site. Both CypA-pET-25b(+) vector and A96 or V96-modified pET-25b(+) vectors were digested with BseRI and BssHII to allow CypA insertion to the 5’ end of the ELP encoding sequence. For C-term CypA a glycine-encoding codon, GGT, was inserted into the 5’ terminus of the CypA encoding sequence and the BseRI restriction site at the 3’ terminus was removed. This modified C-term CypA sequence was first inserted to the modified pET-25b(+) vector using NdeI and BamHI, as described above. To complete the ligation of CAC and CVC, the CypA-A96/V96-modified pET-25b(+) vectors or the C-term CypA-modified pET-25b(+) vector were digested with BssHII and AcuI, or BseRI and BssHII, respectively. The stepwise cloning strategy is illustrated in Supplementary Figure S1 and the open reading frame for each confirmed gene is in the supplementary materials.

For protein expression, ClearColi BL21(DE3) electrocompetent cells (60810, Lucigen, Middleton, WI) were transformed by CAC or CVC encoding plasmids for recombinant expression following the manufacturer’s protocol [56]. Expressed CAC and CVC were purified from cell lysates using inverse transition cycling [57]. More than 98% purity can be obtained by 3 rounds of ELP collection by cycling above and below the phase transition temperature of the fusion protein. For concentration measurements, purified ELPs (CAC and CVC) were diluted 1X with 8M guanidine-hydrochloride and assayed for absorbance at 280 nm using a UV-Vis spectrophotometer with a one-centimeter light path (DU800, Beckman Coulter Inc.).

$$C_{ELP}=O{D}_{280}∕\epsilon$$ (Eq.4)

The molar extinction coefficient, ε, of CAC and CVC, was estimated to be 17,960 (M⁻¹ cm⁻¹) [58] based on the following equation:

$$\epsilon =125n_{Cysteine}+5500n_{Tryptophan}+1490n_{Tyrosine}$$ (Eq.5)

SDS-PAGE was used to resolve proteins, which were then stained with Bio-Safe™ Coomassie Stain (1610786, Bio-Rad Laboratories) to verify the molecular weight and determine the purity of the fusion proteins. Their transition behavior was characterized by measuring optical density at 350 nm through a controlled temperature gradient from 25 to 75°C at 1°C/min using DU800 UV/Visible Spectrophotometer (Beckman Coulter, Brea, CA) The transition temperature (T_t) was defined as the temperature at which the maximum first derivative of the optical density with respect to the temperature was reached.

1.7. Characterization of ELP monomeric/oligomeric state and drug binding affinity

To obtain and characterize mono-dispersed materials for functional investigations, CAC and CVC isolated by ELP-mediated phase separation were further purified by size exclusion chromatography (SEC) to separate different oligomeric states included in these ELP solutions.

For characterization purpose, 100 μL of CAC or CVC at 25 μM was first resolved with a size exclusion HPLC column (Shodex Protein KW-803, Showa Denko America, Inc, New York, NY) at a 0.5 mL/min isocratic flow of PBS. Elution was subject to analysis by three in-line detectors: 1) a variable wavelength detector (SYS-LC-1200, Agilent, Santa Clara, CA) at 210 nm; 2) a multiangle light scattering detector (MALS) (DAWN HELEOS, Wyatt Technology Corporation, Santa Barbara, CA); and 3) differential Refractive Index (dRI) detector (OPTILAB rEX, Wyatt Technology Corporation). ASTRA 6 software was used for data analysis and molar mass estimation. For preparative scale chromatography, a BioLogic Duo-Flow system (Bio-Rad Laboratories, Inc., Hercules, CA) and HiLoad^® 26/600 Superdex^® 200 pg SEC column (28989336, Cytiva, Marlborough, MA) was used to purify mono-disperse material. In detail, 5mL of CAC or CVC with no more than 300μM concentration were resolved at 2.0mL/min isocratic flow of PBS.

The drug binding kinetics of CAC and CVC were studied using isothermal titration calorimetry (ITC) (MicroCal PEAQ-ITC, Malvern Instruments Ltd, Northampton, MA). Both constructs were diluted to 25 μM to avoid phase separation at room temperature, which was the temperature used for the assay. Both CsA and ELP constructs were equilibrated in the same buffer (3% v/v DMSO in PBS) to minimize the background heat released by buffer mismatch. CAC or CVC was titrated 12 times, in 3 μL aliquots, into 280 μL of a 5 μM CsA solution. Each injection generates a heat pulse that is integrated with respect to time and normalized for concentration to generate a titration curve of kcal/mol vs molar ratio (ligand/sample). The resulting isotherm is fitted to an “one set of sites” binding model to generate the affinity (KD), stoichiometry (n) and enthalpy of interaction (ΔH) using MicroCal ITC analysis software (Malvern Instruments Ltd, Northampton, MA).

1.8. Drug encapsulation and loading efficiency measurements

Nine volume parts of CAC or CVC solution in PBS was mixed with one volume part of CsA in ethanol (T038181000, ThermoFisher Scientific Inc., Waltham, MA), at three times molar excess, dissolved in ethanol. The mixture was stirred for one hour at 4°C to maximize drug loading, followed by high-speed centrifugation at 16,100 x g and ultrafiltration through syringe filters (25 mm, 0.2 μm) (4612, Pall Corporation, Port Washington, NY) to remove unbound, insoluble drug. Ethanol was removed by dialysis against PBS (~5 mL against 1 L incubated at 4 °C with 2-4 changes of buffer over 1-2 days). The loading efficiency was determined by RP-HPLC (1260 Infinity II, Agilent, Santa Clara, CA). The binary mobile phase was composed of water and methanol, each containing 0.1% trifluoroacetic acid (TFA) (1081780050, Sigma-Aldrich, St. Louis, MO). ELPs and ELP-bound CsA was resolved by a C4 column (BU12S05-1546WT, YMC CO., Devens, MA), eluted by a gradient flow of methanol (A452, ThermoFisher Scientific Inc., Waltham, MA) /water (40:60) to methanol/water (95:5) for the first 5 min and then an isocratic flow of methanol/water (95:5) for another 5 min at a flow rate of 1 mL/min, and detected at 210 nm.

1.9. Endotoxin removal

Drug-loaded CAC and CVC were subject to endotoxin removal by filtration through Acrodisc^® Mustang E syringe filters (MSTG25E3, Pall Corporation, Port Washington, NY) at 4 °C, prior to any efficacy and pharmacokinetic evaluation. 0.2 μm pore size of these filters also ensured sterility suitable for in vivo injection. Per recommendation of United States Pharmacopeia, the remaining endotoxin burden was measured with chromogenic Limulus Amebocyte Lysate (LAL) assay (C1500, Associates of Cape Cod, Inc., East Falmouth, MA) following manufacturer’s protocol.

1.10. CsA mediated inhibition of IL-17A secretion from Th17 cells

In vitro differentiated Th17-like cells were stimulated with 20 ng/mL PMA and 1 μg/mL ionomycin, immediately following treatment with free CsA dissolved in DMSO, CAC-CsA or CVC-CsA at CsA concentrations from 10 pM to 1 μM for 5 hr at 37°C. IL-17A concentration in culture medium was measured using ELISA (432504, BioLegend, San Diego, CA) on a SpectraMax iD3 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA) at 450 nm.

1.11. Pharmacokinetic study

CsA-loaded CAC and CVC were fluorescently labeled with rhodamine. 3 times molar excess of NHS-Rhodamine (46406, ThermoFisher Scientific Inc., Waltham, MA) dissolved in DMSO was mixed with 200-300 μM CAC-CsA or CVC-CsA in PBS. The mixture was protected from light and incubated with constant rotation for 1.5 hr at 4°C. Unbound free dye was then removed by Zeba desalting column (89893, ThermoFisher Scientific Inc., Waltham, MA) and sink condition dialysis against PBS at 4°C. As reported previously [47], the labeling efficiency was determined to be 177% for CAC-CsA 137% for CVC-CsA.

To compare the pharmacokinetic profiles of CAC and CVC, rhodamine-labeled CsA-loaded CAC was injected into 12-week-old male BALB/c mice intravenously (IV), or subcutaneously (SC) to an area of skin overlaying the LG (supra-LG injection) CsA-loaded CVC was also injected supra-LG SC. The injected dose was 1095 nanomoles of drug-loaded protein/kg BW. 20 μL of blood was collected from the tail vein by tail nicking at various time points up to 120 hrs. The collected blood was immediately added to 80 μL of heparinized PBS at a heparin concentration of 1000 U/mL. Diluted plasma was collected after spinning down blood cells at 1,000 x g for 10 min. Fluorescence intensity (Excitation/Emission: 540/580 nm) was measured by a SpectraMax iD3 Multi-Mode Microplate Reader to estimate plasma concentration of CAC and CVC.

The area under the plasma concentration-time curve (AUC) was first calculated with the trapezoidal method to reflect total body exposure to drugs administered, i.e., CAC IV, CAC SC or CVC SC, which subsequently enabled non-compartmental computation of the area under the first moment curve (AUMC), mean residence time (MRT), mean absorption time (MAT) after SC injection, the SC bioavailability, F, the plasma clearance (CL) with following equations.

$$MRT=AUMC∕AUC$$ (Eq.6)

$$MAT=MR{T}_{SC}-MR{T}_{IV}$$ (Eq.7)

$$CL∕F=Dose∕AUC$$ (Eq.8)

$$F=AU{C}_{SC}∕AU{C}_{IV}$$ (Eq.9)

Due to the phase transition behavior of CVC at physiological temperature, CVC IV was not explored, as that was expected to form micron-sized droplets in the bloodstream. The MRT_IV and AUV_IV of CVC were assumed to be equal to those of CAC. The last two time points of each profile were fit to log-linear decay to estimate the terminal half-life, t_1/2, terminal.

Besides circulating concentration measurements, the local retention after supra-LG injection was monitored using an in vivo imaging system (IVIS) (Lumina Series III, Perkin Elmer, Waltham, MA). Epifluorescence images of mice treated with SC CAC or CVC were captured at different points from 4 hr to 14 days. The total fluorescence intensity within the region of interest (ROI) against the time profile was fit to a two-phase decay model. The terminal half-lives of both constructs after supra-LG injection were estimated and compared.

The depot-forming construct, CVC, was expected to extend the duration of drug retention at SC injection site. To confirm this hypothesis, CAC-CsA or CVC-CsA were injected into BALB/c mice SC via supra-LG injection with n=6 at a drug dose of 2.7mg/kg. Both LG and whole blood were collected from three animals in each group on day 7 and from the remaining animals on day 14 for LC-MS analysis to determine CsA concentration.

1.12. Therapeutic study

Male NOD mice typically develop a SS-like autoimmune dacryoadenitis from 8-12 weeks, which is well-established by 14 weeks of age [47]. In the first therapeutic study, four groups of 14-week-old male NOD mice with n=10/group were treated with either CAC-CsA, CVC-CsA, or A192 + Sandimmune^® at a CsA dose of 2.0 mg/kg or with PBS once a week for two weeks, via supra-LG SC injection. The Sandimmune^® Injection formulation of CsA was used as a free drug control; however, it was supplemented with additional free ELP to control for the presence of ELPs in the CAC treatment. A192, which has the sequence G(VPGAG)₁₉₂Y, was used, which shares a similar MW as CAC and CVC, but lacks the drug-binding specificity. A different dosing regimen was used in the second therapeutic study. Male NOD mice with n=15/group were treated with CAC-CsA, or CVC-CsA at a CsA dose of 3.0mg/kg or with PBS once via supra-LG injection. Two weeks after the first injection, lymphocytes were prepared from LGs under the conditions in each therapeutic study using the methods described above and subjected to analysis with flow cytometry. In the second therapeutic study, total number lymphocytes isolated from LGs were counted with Bio-Rad TC10 Automated Cell Counter. Individual gland weight was also recorded. Statistical analyses were performed using one-way analysis of variance (ANOVA) followed by Tukey’s posthoc test using Prism 9 (GraphPad Software, La Jolla, CA). A p-value < 0.05 was considered statistically significant.

1.13. Flow cytometry

To monitor in vitro Th17-like cell differentiation, Th17-like cells were stimulated with 50 μg/mL phorbol 12-myristate 13-acetate (PMA) (P8139, Sigma-Aldrich, St. Louis, MO) and 1 mg/mL ionomycin (I0634, Sigma-Aldrich, St. Louis, MO) in the presence of 5 μg/mL Brefeldin A (BFA) (420601, BioLegend, San Diego, CA), the protein transport inhibitor, for 5 hr at 37°C to allow intracellular accumulation of IL-17A. Cells collected from each well of 12-well plate were collected, washed once with 1mL PBS and resuspended in 1mL fresh PBS, into which 1 μL of reconstituted Aqua dead cell stain (L34957, ThermoFisher Scientific, Waltham, MA) was added. This cell suspension was briefly vortexed and incubated on ice for 30 minutes in the dark. Cells were then washed 1X with 1 mL PBS, followed by surface antigen staining. Cells were resuspended in 100 μL PBS, and mixed well with 0.25 μg PerCP anti-mouse CD45 antibody (103129, BioLegend, San Diego, CA) before incubating on ice for 30 min in the dark. Cells were then washed 1X with 1mL PBS and fixed with 500 μL of 4% paraformaldehyde at room temperature for 20 min in the dark. The cell membrane was permeabilized by washing cells 1X with an intracellular stainingpermeabilization wash buffer (421002, BioLegend, San Diego, CA). Cell were then resuspended in 100 μL permeabilization buffer, mixed well with 0.25 μg FITC-labeled anti-mouse IL-17A antibody (506908, BioLegend, San Diego, CA) and incubated on ice for 30 min in the dark. Finally, cells were washed 1X with permeabilization buffer and resuspended in 500 uL PBS before being acquired on a Fortessa X-20 Cell Analyzer (BD Bioscience, San Jose, CA). CD45+ cells with positive intracellular IL-17A staining were considered as Th17-like cells. FlowJo™ V10 software (BD Biosciences, San Jose, CA) was used for data analysis.

To evaluate cells obtained from the therapeutic studies, helper T cell subtypes were distinguished based on their surface markers. LG-derived lymphocytes were first subject to Aqua staining as described above, following by surface staining with serial antibodies purchased from BioLegend (San Diego, CA): Brilliant Violet 421™ anti-mouse CD196 (CCR6) (129817), PerCP/Cy7 anti-mouse CD194 (CCR4) (132213), PE anti-mouse CD183 (CXCR3) (126505), APC anti-mouse CD25 (102011), BV605 anti-mouse CD127 (IL-7Rα) (135025), Alexa Fluor® 700 anti-mouse CD3 (152316), FITC anti-mouse CD4 (100406), APC/Cyanine7 anti-mouse CD8a (100713) and PerCP anti-mouse CD45 antibodies (103129) at the manufacturer’s recommended concentration. Cells were acquired on a Fortessa X-20 Cell Analyzer and data analyzed with FlowJo™ V10 software.

1.14. Basal thread test of tear production

Tear production was assessed as described [59], A ZoneQuick phenol red-embedded thread was applied in both eyes at the canthus of the ocular surface for 10 sec, while the animal was under light anesthesia with isoflurane. Basal tear volume was recorded as the length of thread wetting by basal tears in mm.

2. Results

2.1. Cytokine expression by CD4+ T cells present in healthy LG versus LG with autoimmune dacryoadenitis.

To better understand the spectrum of immune cells infiltrating the LG upon establishment of autoimmune dacryoadenitis, the infiltrating lymphocytes and glandular epithelial cells were prepared from LG from age-matched male NOD and BALB/c mice via enzymatic digestion. Lymphocytes were subsequently enriched by filtration through a 10 μm cell strainer, from which the CD4+ T cell population was negatively selected with 81.6 ± 9.7% (mean ± SD) purity as shown in Supplementary Fig. S2. The CD4+ T cell population was then negatively selected from total lymphocytes. Similarly, CD4+ T cells were also isolated from the LG-draining cervical lymph nodes (LN) and spleen (SP). qRT-PCR was conducted to measure gene expression levels of two pro-inflammatory cytokines of interest, IL-2 and IL-17A, the signature cytokines of Th1 and Th17 cells, from these purified CD4+ T cells. As shown in Figure 1A, a considerably higher IL-17A gene expression was observed in CD4+ T cells isolated from NOD mouse LG relative to LG from age-matched BALB/c mice. Due to the lack of lymphocytic infiltration, the relative abundance of CD4+ T cells to glandular epithelial cells was markedly lower in BALB/c mice. CD4+ T cells derived from BALB/c mouse LG were thus subject to higher epithelial cell contamination, as exhibited by higher expression of the membrane water channel, Aquaporin 5 (AQP5) (Figure 1C). Unlike CD4+ T cells isolated from the LG, IL-17A gene expression levels were similar in age-matched NOD and BALB/c mice in LN and SP. Notably, when comparing IL-17A gene expression among different organs, CD4+ T cells from NOD mice also exhibited dramatically higher IL-17A gene expression in LG than LN and SP, which was not observed for BALB/c mice (Figure 1A).

Figure 1. IL-17A and IL-2 gene expression is elevated in CD4+ T cells isolated from NOD mouse LG.

mRNA was extracted from CD4+ T cells, negatively isolated from LG, LN or SP, and used to quantify A) IL-17A, B) IL-2 gene expression and C) epithelial cell contamination in the LG samples quantified by the expression of the water channel protein, AQP5. BALB/c mice were analyzed at 11 and 22 weeks (B11, B22). NOD mice were analyzed at these timepoints as well as at 7 and 15 weeks (N7, N11, N15, N22) to better characterize the development of disease. A) IL-17A was significantly higher in LGs of NOD mice older than 7 weeks in comparison with age-matched BALB/c mice. In contrast to LG, IL-17A was expressed to the same extent in CD4+ T cells from LN and SP for both strains. B) CD4+ T cells from LG also showed notably higher IL-2 expression than cells from LN and SP in older NOD mice. C) The expression of the epithelial-cell-specific membrane water channel protein, AQP5, was used to evaluate the relative contamination of isolated T cells with epithelial cells. Due to the limited abundance of T cells in LGs with low lymphocyte infiltration (B11,B22,N7), the epithelial cell contamination is greater for LG preparations from BALB/c mice at 11 and 22 weeks compared to NOD mice at 22 weeks (N22). N=7. Error bar represents SD. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001.

The expression profiles of IL-2 and IL-17A in tissue-infiltrating CD4+ T-cells were distinct. As shown in Figure 1B, LG IL-2 expression in these cells are comparable in NOD and BALB/c at all time points measured, while IL-2 expression showed a trend to an increase in CD4+ T-cells from LN and SP from BALB/c versus NOD mice. Although IL-2 expression from infiltrating CD4+ T-cells was still much higher than in LN and SP in older NOD mice, the fold increase of 2-3 in this tissue was much less than the ~100-fold increase for IL-17A in the same cell population. Since the elevation of IL-17A is much more prominent than IL-2, we decided to focus on IL-17 and IL-17 secreting cells in the rest of this study.

2.2. Characterization of recombinant CAC and CVC, two protein-based carriers of CsA.

As depicted in Figure 2A, by fusing CypA to both the N and C-termini of an ELP backbone with the same number of pentameric repeats but differing hydrophobicity, CAC and CVC were cloned and expressed from E. coli. As reported earlier [57], CAC and CVC were purified from cell lysates by repeatedly inducing ELP-dependent phase separation. The molecular weights of the purified constructs were resolved by SDS-PAGE and staining with Coomassie Blue (Figure 2B). The absolute MW of CAC and CVC were confirmed by MALDI-TOF mass spectroscopy to be 72.8 kDa and 75.4 kDa, respectively (Figure 2C, Table 1).

Figure 2. A cyclophilin-based drug-carrier assembles an ELP depot for cyclosporine at physiological temperatures.

A) Cartoon depicting the ELP fusion proteins, CAC or CVC, which specifically bind CsA and behave differently in response to physiological temperature when injected subcutaneously; B) Compared to a MW ladder (lane 1), SDS-PAGE of purified CAC (lane 2) and CVC (lane 3) confirmed their expected MW and high purity; C) The absolute MW of both CAC and CVC was confirmed by MALDI-TOF mass spectroscopy to be 72.8 kDa and 75.4 kDa, respectively, very close their expected MW. D) ELP phase separation was monitored using optical density as a function of temperature and concentration and plotted as a phase diagram. Fusions assemble coacervates in solution above a log-linear (Eq. 3-1) fit line. Dashed lines represent the 95% confidence interval. At relevant injection concentrations (~100 μM), CAC, but not CVC, remains soluble at physiological temperature. E, F) SEC-MALS was used to analyze the oligomeric state of both constructs in solution. Two populations of CAC were separated with the 2^nd fraction being the major population, with an estimated MW of 77.6 kDa. Unlike CAC, CVC appeared homogeneous, with an estimated MW of 75.6 kDa. By comparing the observed and expected (Table 1) MW, both constructs were concluded to be primarily monomeric in solution. G) The binding thermodynamics between CAC or CVC and CsA were analyzed by ITC. The K_d of CsA was 207 ± 133 nM for CAC and 46 ± 35 nM for CVC. The binding stoichiometry of CsA was 0.48 ± 0.04 to CAC and 0.33 ± 0.02 to CVC, which confirmed the availability of two binding sites per fusion protein. The binding enthalpy between CsA was −236 ± 39.2 kJ/mol for CAC and −109 ± 11.9 kJ/mol for CVC.

Table 1.

Amino acid sequence and thermo-responsive behavior of CAC and CVC.

Label	Amino Acid Sequence	1 M.W. [kDa]	2 Est. M.W. by MALS [kDa]	3 Est. M.W. by MS [kDa]	4 Slope, m [°C/Log10(μM)]	4 Intercept, b [°C]	5 Est. Tt at 300μM [°C]
CAC	M-CypA-(VPGAG)96-CypA	72.9	77.6	72.8	4.3 ± 0.8	55.2 ± 1.2	44.5
CVC	M-CypA-(VPGVG)96-CypA	75.6	75.6	75.4	7.3 ± 1.6	39.9 ± 2.4	21.8

CypA amino acid sequence:

MVNPTVFFDIAVDGEPLGRVSFELFADKVPKTAENFRALSTGEKGFGYKGSCFHRIIPGFMCQGGDFTRHNGTGGKSIYGEKFEDENFILKHTGPGILSMANAGPNTNGSQFFICTAKTEWLKHVVFGKVKEGMNIVEAMERFGSRNGKTSKKITIADCGQLEG

¹Expected molecular weight based on the open reading frame for the expressed protein.

²Estimated molecular weight by MALS, data shown in Figure 2E, 2F.

³Estimated molecular wight by MALDI-TOF mass spectrometry, data shown in Figure 2C

⁴The transition temperatures of both constructs follow Eq.10, yielding an intercept, b, at 1 μM, and a slope, m, representing the change in temperature upon a 10-fold change in concentration. Mean ± 95% CI.

⁵The estimated transition temperature at the injection concentration used in in vivo therapeutic evaluation described in section 2.5, based on Eq.10.

Consistent with our previous finding [47], the T_t of both constructs were a log-linear function of concentration as follows (Figure 2D):

$$T_t=b-m{\log }_{10}[C_{ELP}]$$ (Eq.10)

where the intercept, b, represents the transition temperature at 1 μM, the slope, m, is the decrease in Celsius for a 10-fold increase in concentration and [C_ELP] represents the fusion protein concentration. Values for b and m as summarized in Table 1 were used to estimate the transition temperature (T_t) of the in vivo injection formulation, which are 44.5, 21.8 °C for CAC, CVC respectively. These results show that CAC is expected to remain soluble at the physiological temperatures, while CVC is expected to phase separate.

Previously, we reported that two populations of CA192 with different oligomeric status could co-exist in solution: a nano-aggregate and a dimeric species, as determined by SEC-MALS, while only the dimeric form bound to CsA [47]. Similar to CA192, as shown in Figure 2E, CAC also exhibited two oligomeric states, despite a much smaller calculated nano-aggregate fraction. Unlike CA192, the 2^nd low MW fraction of CAC retained by SEC was monomeric in solution, with an absolute MW estimated to be 77.6 kDa. This is very close to its expected MW, which was confirmed by mass spectrometry (Table 1). In contrast to CAC, a nano-aggregate fraction eluting in the void volume was not observed during SEC of CVC; however, CVC also existed in solution as a homogeneous monomer with an absolute MW of 75.6 kDa (Figure 2F).

Additionally, the binding kinetics between CAC/CVC and CsA were analyzed by isothermal titration calorimetry (ITC) at 25°C, the temperature at which both constructs remain soluble (Figure 2G). The equilibrium dissociation constant (K_d) was determined to be 207 ± 133 nM and 46 ± 35 nM for CAC and CVC, respectively, comparable with that of CA192 of 189 ± 87 nM [47]. The stoichiometry of CAC and CVC was determined to be 0.48 ± 0.04 and 0.33 ± 0.02, respectively, consistent with the designed architecture, by which each CAC or CVC molecule can maximally bind two CsA molecules. When binding with CsA, the binding enthalpy of CAC was determined to be −236 ± 39.2 kJ/mol, stronger than that of CVC of −109 ± 11.9 kJ/mol.

Finally, prior to efficacy and pharmacokinetic evaluation, CAC-CsA and CVC-CsA were first subject to endotoxin removal. The remaining endotoxin level at 300μM protein concentration was determined to be 40 EU/mL for CAC-CsA and 43 EU/mL for CVC-CsA.

2.3. In vitro efficacy of CAC-CsA and CVC-CsA against Th17-like T cells.

IL-17A producing Th17-like cells can be differentiated and polarized from CD4+ T cells isolated from NOD mouse splenocytes when incubated with a cocktail of cytokines and antibodies. After a 4-day incubation, cells with positive intracellular staining for IL-17A were identified by flow cytometry. As shown in Figure 3, about 5% of these helper T-cells were polarized into Th17-like cells. This frequency was significantly reduced to below 1% when co-incubated with 10 nM CsA, either as free drug, or delivered by CAC or CVC.

Figure 3. CAC-CsA and CVC-CsA effectively inhibit Th17-like cell differentiation in vitro.

CD4+ T-cells obtained by negative selection from NOD mouse splenocytes can be differentiated into Th17-like cells in vitro when treated with a cocktail of cytokines and antibodies (IL-6, TGF-β1, IL-23, anti-mouse CD28, anti-mouse IL-4, and anti-mouse IFN-γ) on anti-mouse CD3ε antibody-coated plate. As CD4+ T cells were pre-selected prior to the differentiation, pan-leukocyte marker CD45 was used here in place of CD4. A subset of these helper T cell population, Th17-like cells were captured by flow cytometry based on their intracellular accumulation of IL-17A upon activation. Normally, 4.9 ± 1.6% (mean ± SD from n=3) of CD4+ T-cells can be differentiated into Th17-like cells. However, Th17-like cell abundance was significantly reduced to 0.6 ± 0.1%, 0.8 ± 0.2 and 0.9 ± 0.1 when co-incubated with 10 nM CsA, delivered by CAC, CVC or in its free form, respectively. **p<0.01.

These in vitro differentiated Th17-like cells, when activated by ionomycin and phorbol myristate acetate (PMA), secrete large amounts of IL-17A in the absence of CsA. The cytokine release from activated Th17-like cells differentiated as described above was also observed to be effectively inhibited by CAC-CsA and CVC-CsA in a dose-dependent manner as shown in Figure 4. The half maximal inhibitory concentrations (IC₅₀) of CAC-CsA and CVC-CsA were determined as 28.5 ± 9.1 nM and 9.1 ± 0.6 nM, respectively, slightly higher than that of free CsA which was 1.0 ± 0.4 nM.

Figure 4. CAC-CsA and CVC-CsA effectively inhibit IL-17A secretion from activated Th17-like cells in vitro.

Upon activation with PMA and ionomycin, Th17-like cells produce a large amount of IL-17A, which is secreted into the culture medium. CsA, regardless of its delivery vehicle, inhibits this IL-17A secretion in a dose-dependent manner. The IC₅₀ of free CsA, CAC-CsA and CVC-CsA were estimated to be 1.0 ± 0.4 nM, 28.5 ± 9.1 nM and 9.1 ± 0.6 nM (mean ± SD from n=3), respectively.

2.4. Pharmacokinetic profiles of CAC and CVC in mice.

To enhance accumulation in the target tissue, the LG, and to more effectively control the local elevation of IL-17A expression in LG associated with IL-17A expressing immune cells, we developed a new route of administration, supra-LG injection, which consists of a subcutaneous (SC) injection to the region overlaying the LG (Figure 5A).

Figure 5. The depot-forming CVC-CsA has extended PK profile relative to soluble CAC-CsA.

A) The cartoon illustrates supra-LG injection (subcutaneous injection to the region overlaying the LG). This new route of administration was developed to maximize the local effect of CsA; B) Using a rhodamine-labeled approach, the plasma concentration of CAC was determined after IV administration and compared to CAC or CVC after supra-LG injection. Supra-LG administration of CVC gave the lowest relative bioavailability and the longest terminal half-life (15.2 hr for CAC SC and 56.2 hr for CVC SC) (Table 2). Error bars represent mean ± SD from n = 5 BALB/c mice; C) Whole animal IVIS imaging was used to compare the local retention of CAC (left) and CVC (right) on day 5 post supra-LG injection. CAC was completely eliminated by day 5, while CVC remained easily detectable. D) Image analysis was used to quantify absorption of CAC and CVC from the supra-LG injection site, which follows a two-phase decay model with a terminal half-life of 28 hr for CAC and 92 hr for CVC. E, F) HPLC-MS was used to quantify CsA drug concentrations in plasma and LG at 7 and 14 days after supra-LG administration. CVC-CsA maintained a stable CsA concentration in whole blood (panel E) and a much higher level in the LG lysate (panel F) for at least 14 days, the longest time measured. In contrast, by these timepoints CsA in the CAC-CsA formulation was undetectable.

The pharmacokinetic (PK) profiles of CAC and CVC after supra-LG injection and CAC after IV injection were studied in healthy BALB/c mice. Plasma concentrations of both constructs versus time profiles are depicted in Figure 5B. Non-compartmental analysis was conducted under the assumption that the PK profile of CVC given IV is roughly the same as that of CAC given IV, considering their similar architecture and MW. CVC was not given IV since the coacervation may disrupt venous circulation. As summarized in Table 2, the depot-forming construct, CVC, exhibited a 4-times longer mean residence time (MRT) of 52 hr relative to its soluble counterpart, CAC, of 12.6 hr when given supra-LG. The slower absorption phase of CVC is expected to be the rate-limiting step, supported by the mean absorption time (MAT) extension from 1.6 hr to 40.9 hr. The bioavailability, F, after supra-LG SC injection was calculated to be 40.9% and 24.4% for CAC and CVC, respectively. Additionally, the depot-forming CVC demonstrated a terminal half-life of 56.2 hr, significantly longer than the 15.2-hr terminal half-life of CAC after supra-LG SC administration.

Table 2.

Comparison of pharmacokinetic parameters observed for CAC after IV or SC administration and CVC after SC administration.

Parameters (Unit)	CAC IV	CAC SC	CVC SC
CL/F (mL/hr)	0.3 (0.04)	0.76 (0.12)	1.34 (0.23)
AUC (μM hr)	108.6 (13.8)	44.4 (3.4)	26.5 (3.9)
AUMC (μM hr2)	1196.7 (180.9)	559.8 (67.7)	1366.4 (213.7)
MRT (hr)	11.0 (1.2)	12.6 (0.9)	52.0 (7.5)
MAT (hr)	-	1.6 (0.9)	40.9 (7.5)
F (%)	100	40.9 (3.1)	24.4 (3.6)
V/F (mL/30g BW)	0.99 (0.27)	*0.99 (0.27)	*0.99 (0.27)
T1/2, Terminal (hr)	41.0 (4.3)	15.2 (2.7)	56.2 (5.4)

Values are indicated as mean (SD)

^*Volume distribution estimated from IV analysis were assumed in SC analysis.

Use of an in vivo imaging system (IVIS) confirmed the longer retention of CVC at the injection site. Fluorescently labeled CAC-CsA and CVC-CsA were administered via supra-LG injection. As shown in Figure 5C, by day 5 after injection, CAC is no longer detectable at the injection site, while CVC is still easily detectable. Based on the remaining fluorescence intensity retained at injection, absorption of CAC and CVC from the injection site follows a two-phase decay model with a terminal half-life of 28 hr (CAC) and 92 hr (CVC) (Figure 5D). More importantly, the depot-forming CVC achieved sustained drug release and retained much higher CsA both in whole blood (Figure 5E) and LG lysate (Figure 5F).

2.5. Therapeutic effect of CAC versus CVC on disease pathogenesis in the NOD mouse model of autoimmune dacryoadenitis

Initially, 14-week-old male NOD mice were divided into four treatment groups: PBS, A192 + Sandimmune^® (US-FDA approved IV formulation for CsA), CAC-CsA and CVC-CsA, with n = 10. The PBS group served as a negative control, while the A192 + Sandimmune group served both as an ELP carrier control and a free drug control. Mice received supra-LG injection once a week for two weeks at a CsA dose of 2.0 mg/kg. Two weeks after the first injection, infiltrating lymphocytes were prepared from LG and subsequently analyzed by flow cytometry to understand how different treatments affected helper T cell composition.

Surprisingly, based on our gating strategy for different helper T cell subsets illustrated in Figure 6A, we observed that Th17 cells, traditionally defined as CD3+ CD4+ CCR6+ CCR4+ CXCR3−, were fairly rare. Instead, the CCR6+ helper T-cell population was predominantly CCR4−. Therefore, our focus became the CCR6+ CCR4− CXCR3+ Th17.1 cell population (also called Th1Th17 cells) as the likely origin of the IL17A expressed by CD4+ T cells in the LG in disease (Figure 1).

Figure 6. Two treatments with CVC-CsA significantly reduced Th17.1 and Th2, but not Th1 or Treg, cells in the LG-infiltrating CD4+ population, as well as glandular IL-17A accumulation.

14-week-old male NOD mice were treated twice (weekly) by supra-LG administration and compared to controls. In addition to PBS, a combination of ELP (A192) that lacks the ability to bind CsA combined with an approved formulation of CsA known as Sandimmune^® was also used as a control. A) Different CD4+ T cell subsets were gated based on surface antigen expression. With prior gates set on CD3+ CD4+, Th17 cells were defined as CCR6+ CCR4+ CXCR3−, Th17.1 as CCR6+ CCR4− CXCR3+, Th1 as CCR6− CCR4− CXCR3+, Th2 as CCR6− CCR4+ CXCR3− and Treg as CD25+ CD127−. B) Relative to control groups treated with PBS or A192 + Sandimmune^®, only CVC-CsA treatment reduced the frequency of Th17.1 cells obtained from the diseased NOD mouse LG. CAC-CsA treatment was ineffective in reducing the Th17.1 cell population. C) As an alternative to isolation and flow cytometry, the total IL-17A content of LG lysates was measured by ELISA. Compared with the PBS group, only the CVC-CsA group showed significantly lower IL-17A concentration in LG lysate. CVC-CsA treatment also increased the frequency of D) Th2 cells in LG-infiltrating CD4+ T cell population but didn’t effectively alter the frequency of E) Th1 or F) Treg. Error bars represent mean ± SD from n=10. Statistical analysis was performed using a one-way analysis of variance (ANOVA) followed by Tukey’s posthoc test. *p<0.05; **p<0.01.

As shown in Figure 6B, relative to PBS controls and an equivalent CsA dose given as free Sandimmune^® plus ELP, only CVC-CsA was able to significantly reduce Th17.1 cells. CAC-CsA given supra-LG did not exert a therapeutic effect. It was not surprising that CAC-CsA or Sandimmune^®, even at an equivalent CsA dose, did not significantly reduce Th17.1 cell frequency. Supra-LG injection was intended to maximize the CsA accumulation in LG. Unlike the depot-forming CVC-CsA, CAC-CsA remained fully soluble after being injected. As illustrated in Figure 5C and D, absorption of CAC-CsA from the injection site was much faster than CVC-CsA, making it incapable of exerting a sustained effect on the LG. In addition to being quickly absorbed from the injection site, free CsA in Sandimmune^® is also subject to rapid renal filtration, making it even less bioavailable to the LG than CAC-CsA.

With respect to their yield from treated LG in the experimental groups, we also investigated other T cell subtypes including Th1, Th2 and regulatory T cells (Treg) which is known to suppress immune response and maintain self-tolerance. No significant differences in the frequencies of Th1 cells (Figure 6E) and Treg cells (Figure 6F) were seen. However, as summarized in Figure 6D, the anti-inflammatory Th2 cells were increased in the CVC-CsA group relative to the PBS and A192 + Sandimmune^® treated groups. The CAC-CsA treated group also increased this Th2 cell population relative to the PBS group. No significant difference between the CAC-CsA and A192 + Sandimmune^® groups were observed.

As shown in Figure 5, CVC-CsA administration resulted in high CsA concentration both locally in LGs and systemically in the blood for more than 14 days, potentially allowing an even longer dosing interval. Moving forward, we further tested the in vivo efficacy of CVC-CsA when given as a single supra-LG injection over a two-week period. Three treatment groups with n=15 per group were included in this study: PBS, CAC-CsA given supra-LG SC and CVC-CsA given supra-LG SC. At 14 days after injection, CD4+ T cell composition was examined by flow cytometry. As shown in Figure 7A, CVC-CsA significantly outperformed CAC-CsA, as well as PBS, by reducing the frequency of Th17.1 cells recovered in the LG-infiltrating CD4+ T cell population. However, IL-17A accumulation was not significantly reduced by either treatment, possibly due to less drug exposure than when a weekly dose was given (Figure 7B). Consistent with the weekly dose treatment findings in Figure 6, both CAC-CsA and CVC-CsA increased the prevalence of the pro-repair Th2 cell population (Figure 6D) while not eliciting obvious effects on Th1 cells (Figure 6E) or Treg cells (Figure 6F). In addition to modulating Th17.1 and Th2 cell levels in LG, we measured functional recovery of tear flow in response to therapeutic treatments using a basal thread test. As shown in Figure 8, CVC-CsA given supra-LG SC significantly increased tear production in these mice, but CAC-CsA did not improve tear production. We speculate that the tear production was restored by reducing the number of pathogenic Th17-like cells in NOD mice LGs (Figure 7A). Healthy BALB/c control mice were not included in this study as they lack LG lymphocytic infiltration; however, the dose-dependent effects of CVC-Rapa will need to be evaluated in healthy LGs in future toxicological studies. Despite the change in T cell composition and improvement in tear secretion, the total numbers of lymphocytes collected from LGs were not found to be significantly different among three groups (Supplementary Figure S3).

Figure 7. One treatment with CVC-CsA over a two-week period effectively altered the LG-infiltrating helper T cell composition.

A) CVC-CsA appeared to be more effective than CAC-CsA in reducing the abundance of Th17.1 cells; B) IL-17A was not reduced by either CAC-CsA or CVC-CsA, likely due to reduced drug exposure from the weekly dose treatment. Consistent with findings in Figure 6, CVC-CsA also increased C) Th2 cell abundance but didn’t alter the frequency of D) Th1 or E) Treg in NOD mice LGs. Error bars represent mean ± SD from n=15. **p<0.01.

Figure 8. One treatment with CVC-CsA increased basal tear production. 14-week-old male NOD mice were treated with CAC-CsA or CVC-CsA at a CsA dose of 3.0mg/kg or with PBS once for two weeks via supra-LG injection.

14-week-old male NOD mice were treated with PBS, CAC-CsA or CVC-CsA as a single injection over a two-week period. The injection dose of CsA in CAC-CsA and CVC-CsA groups was 3mg/kg. The basal tear production in both eyes was measured by thread test before and after the treatment. No significant difference was observed in the (A) PBS group or the (B) CAC-CsA group. A significant increase in tear production was only observed in (C) CVC-CsA group. Error bars represent mean ± SD from n=30 (both eyes of 15 animals). **p<0.01.

3. Discussion

The therapeutic approach reported here was inspired by initial findings of increased expression of the proinflammatory cytokines, IL-17A and IL-2, which are characteristic of Th17 and Th1 cells, in diseased NOD mice LG-derived CD4+ T cells. Overexpression of IL-17A in particular has been strongly correlated with pathogenesis of several autoimmune diseases [16-23]. Male NOD mice have been previously reported to develop lymphocytic infiltration of the LG as early as 8 weeks of age and fully established autoimmune dacryoadenitis by 12 to 14 weeks of age [50]. Our findings from the analysis of gene expression levels of IL-17A in LG-infiltrating CD4+ T cells suggests that increased IL-17A is established as early as 7 weeks of age, suggesting it as one of the earliest events in the onset of LG pathogenesis. Additionally, we observed that CCR6+ CXCR3+ Th17.1 cells, the “alternative” Th17 cells, are more abundant within the LG-infiltrating CD4+ T cell population, and are likely the principal source of the IL-17A accumulated in the LG. These “alternative” Th17 cells have also been shown to be enriched in the target organs of several other autoimmune diseases [60, 61], likely being particularly pathogenic in tissue inflammation and autoimmunity [62].

In addition to Th1, Th2 and Th17, another independent CD4+ T cell lineage, follicular helper T (Tfh) cells, has also been identified, defined by surface expression of C-X-C chemokine receptor type 5 (CXCR5) and the induced costimulatory molecule (ICOS). Through the production of IL-21, Tfh cells exhibit strong effects in stimulating the differentiation of B cells into antibody-producing cells [63]. It is interesting to note that Th17 cells share several features, including the expression of IL-21 [64] and ICOS, triggering B-cell proliferation and promote the formation of germinal centers [65]. We speculate that Th17.1 cells may share these properties and function as strong B-cell helpers in the LG of SS. Additionally, T cells may also express B-cell activating factor (BAFF) in inflamed glandular tissue of patients with SS [66]. BAFF is essential for B cell maturation and survival [67] and found to be elevated in serum and SGs of SS patients. Taken together, we speculate that Th17 cells, both classical and alternative, may be responsible for the dysregulation of B cells and B cell expansion in autoimmune lesions in SS.

Patients with SS-associated dry eye disease (DED) represent 11 % [68] of the total DED patients, as defined by symptoms alone. This includes patients with many different forms of the disease (non-autoimmune-mediated aqueous deficient dry eye, evaporative dry eye, etc). DED is highly prevalent in the population, affecting as many as 17% of women and 11.1% of men in the United States [69]; however, DED is multifactorial and patients with different etiologies are often not distinguished in clinical trials. The few US-FDA approved medications for DED such as topical cyclosporine A (Restasis^®) and lifitegrast (Ziidra^®) provide relief for some patients with SS-associated DED by suppressing ocular surface inflammation. However, they have not been shown to suppress the lymphocytic inflammation of the LG, the principal driver of SS-associated DED. Thus, topical medications treat the symptoms and not the major cause of SS-associated DED. There are currently no US-FDA approved treatment options for SS-associated DED that suppress the underlying lymphocytic infiltration of the LG. Our injectable sustained release construct represents a first step in addressing this limitation.

As an immunosuppressant, CsA has been widely explored for use in various autoimmune indications, such as psoriasis [70], rheumatoid arthritis [71], atopic dermatitis [72], uveitis [73], systemic lupus erythematosus [74] and others. Its systemic use is has been correlated with dose-dependent systemic side effects including nephrotoxicity [75], hepatotoxicity [76], neurotoxicity [77] and hypertension [78]. Organic additives such as polyoxyethylated castor oil (Cremophor EL) are often required for CsA formulation as solubilizers and emulsifiers to enable parenteral administration which may also cause anaphylactoid reactions. Previously, using ELPs as a drug delivery platform, we successfully constructed a first generation ELP-based CsA carrier, named CA192. CA192 carries CsA via non-covalent binding and was internalized into Jurkat and Hela cells where it co-localizes with low-pH compartments, consistent with lysosomes [47]. This finding is consistent with another publication from our lab [79]. CA192-CsA, the drug with this carrier, exhibited strong activity in inhibition of IL-2 release in vitro in an activated Jurkat cell system comparable to free CsA through the cytosolic calcineurin-NFAT signaling pathway. When administered every other day at a drug dose of 2.5 mg/kg, CA192-CsA significantly increased stimulated tear production and mitigated nephrotoxicity compared [47]. The second generation ELP-based CsA carriers, CAC and CVC, reported here retained an additional CypA fused to the C-terminus of the ELP backbones, A96 and V96, exhibiting a higher drug loading capacity. Another advantage over CA192 was that CAC and CVC produced a lower extent of protein aggregate. As shown in Figure 2E and 2F, 22.5% of CAC by AUC in UV chromatogram was in the nanoparticle fraction. Interestingly, CVC had no detectable protein aggregates. As a comparison, 57.8% of CA192 was in the nanoparticle fraction. Just like CA192, CAC and CVC were expressed as soluble proteins from E. coli with a yield of 60-70mg/L for CAC and 20-25mg/L for CVC without the need for refolding. More importantly, the characteristic thermo-responsiveness of ELPs was utilized to achieve sustained drug release. Specifically, the temperature sensitive construct, CVC, undergoes temperature-dependent phase transition at the injection site at physiological temperature upon SC injection, exhibiting sustained drug release for more than 2 weeks, and functioning much like a depot implant. This 2-week duration of effect was not exhibited by its temperature-insensitive counterpart CAC.

As in this work (Fig. 8), CsA delivered by our first generation ELP-based CsA carrier, CA192, improved tear flow [47]; however, histological analysis did not reveal a reduction in total lymphocytic infiltration. We speculate this may be due to the immunomodulatory potential of extracellular CypA via its interaction with CD147 [80]. Increases in extracellular CypA has been correlated with pro-inflammatory diseases, such as rheumatoid arthritis, sepsis and asthma. Some evidence suggests that recombinant CypA obtained from E. Coli exhibits reduced immunologic activity due to the lack of post-translation acetylation [81]. In future studies we will take additional measures to study and block CypA/CD147 interactions; however, in this report we instead introduced a new flow cytometry-based alternative to histology to more accurately quantify changes in the relative composition of the lymphocytes, in particular T cells (Fig. 7).

In this proof-of-concept study, we have administered our CsA sustained release formulation via “supra-LG injection”, SC injection to the region overlaying the LG, which may be feasible in human patients at least for a significant part of the gland. This may be further resolved by incorporating a targeting moiety to our drug carriers. Previously in our lab, we have successfully exercised this concept and enhanced inflamed LG accumulation of another ELP-based drug carrier with ICAM-binding peptide [82].

The abundance of IL-23 and TGF-β have been reported to be skewing factors deciding whether naïve T cells differentiate to “classical” Th17 cells or “alternative” Th17.1 cells. Both cytokines were present during in vitro differentiation of Th17 cells, likely leading to the co-existence of two phenotypes. The incomplete inhibition of both effector cell differentiation and cytokine secretion that was observed in Figures 3 and 4 suggest that CsA may inhibit the development of a specific Th17-expressing cell phenotype. Taken together with in vivo findings that percentage of Th17.1 cell in CD4+ T cells was reliably reduced by CVC-CsA, we conclude that this susceptible population may be the Th17.1 cells, a finding which will be confirmed in future studies.

4. Conclusion

Here, using the male NOD mouse as a model of SS-associated dry eye disease, we identified increased IL-17A expression by LG-infiltrating CD4+ T cells as an early potential pathogenic factor in development of disease. ELPs in fusion with cyclophilin can deliver CsA to exert an inhibitory effect on the principal cell type responsible for IL-17 expression, the Th17.1 cell, impeding their cellular development and cytokine secretion. More importantly, in response to physiological temperature, ELP-mediated depot formation allows the sustained release of CsA, enhancing its in vivo efficacy against SS-mediated ocular surface manifestations.

Data availability

The authors declare that raw data required to reproduce these findings are available upon request.