Recrystallization of Adenosine for Localized Drug Delivery

Abstract

Adenosine (ADO) is an endogenous metabolite with immense potential to be repurposed as an immunomodulatory therapeutic, as preclinical studies have demonstrated in models of epilepsy, acute respiratory distress syndrome, and traumatic brain injury, among others. The currently licensed products Adenocard and Adenoscan are formulated at 3 mg/mL of ADO for rapid bolus intravenous injection, but the systemic administration of the saline formulations for anti-inflammatory purposes is limited by the nucleoside’s profound hemodynamic effects. Moreover, concentrations that can be attained in the airway or the brain through direct instillation or injection are limited by the volumes that can be accommodated in the anatomical space (<5 mL in humans) and the rapid elimination by enzymatic and transport mechanisms in the interstitium (half-life <5 s). As such, highly concentrated formulations of ADO are needed to attain pharmacologically relevant concentrations at sites of tissue injury. Herein, we report a previously uncharacterized crystalline form of ADO (rcADO) in which 6.7 mg/mL of the nucleoside is suspended in water. Importantly, the crystallinity is not diminished in a protein-rich environment, as evidenced by resuspending the crystals in albumin (15% w/v). To the best of our knowledge, this is the first report of crystalline ADO generated using a facile and organic solvent-free method aimed at localized drug delivery. The crystalline suspension may be suitable for developing ADO into injectable formulations for attaining high concentrations of the endogenous nucleoside in inflammatory locales.

Graphical Abstract

INTRODUCTION

Adenosine (ADO) is an endogenous metabolite with diverse physiological effects in multiple organs, including tissues and cells in the cardiovascular,¹ nervous,² and immune systems.^3–5 Currently approved for use in humans as a vascular modulator and antiarrhythmic, ADO has been explored experimentally for suppressing protracted inflammation in the lungs and in the central nervous system (CNS). The physiological impact of exogenously administered ADO in the airway is multifaceted, with several studies demonstrating that the nucleoside confers tissue protection in models of acute respiratory distress syndrome (ARDS).^6–10 ADO has been shown to regulate alveolar fluid clearance in mice,¹¹ and pathologies of ARDS are exacerbated in animals lacking CD39 and CD73, ectonucleotidases that together generate extracellular ADO from ATP released by injured tissues.⁴ ADO also plays a role in osteogenesis by driving osteoblast differentiation.^12,13 In the CNS, ADO is neuroprotective, suppressing inflammation associated with traumatic brain injury (TBI).¹⁴ Moreover, the endogenous release of ADO limits brain tissue damage and seizure,¹⁵ and implantation of fibroblasts engineered to release ADO in the brain proved beneficial in TBI models.¹⁴ Indeed, ADO augmentation therapy (ATT) has been studied extensively for the suppression and prevention of seizures.^16–18 These physiological effects are mediated through G-protein-coupled adenosine receptors (A₁, A_2A, A_2B, and A₃) differentially expressed on multiple cell types, including leukocytes. Physiological actions of ADO are also attributed to receptor-independent, epigenetic mechanisms.^17,19 For instance, the intracellular accumulation of ADO terminates seizures by reducing DNA methylation; the increased expression of adenosine kinase shifts the equilibrium from S-adenosylmethionine to S-adenosylhomocysteine.^19,20 The transmembrane receptors and epigenetic targets add to the complexity of predicting ADO pharmacological actions based on extracellular ADO concentration alone.

The immunological effects of ADO have been documented extensively.⁴ In preclinical models, ADO steers macrophages into anti-inflammatory, regulatory phenotypes.^3,21–24 ADO expands regulatory T cells (Tregs), a lymphocyte subset by which cytokine storm can be attenuated.⁴ Evidence was sufficient insofar as a clinical trial designed to examine the effects of ADO on the release of cytokines was proposed (NCT00580905). While results from these and other studies point to opportunities for repurposing ADO as anti-inflammatory and immunomodulatory therapeutics,⁵ systemic drug infusion would result in significant hemodynamic adverse events due to the ubiquitous expression of adenosine receptors in the cardiovascular system.²⁵ The two currently licensed products Adenocard and Adenoscan, each containing 3 mg/mL of the active pharmaceutical ingredient (API) dissolved in saline, are formulated for bolus intravenous (IV) injection. Tissue-level concentrations that could be attained by administrating these IV formulations are limited because ADO is rapidly cleared in plasma, with a half-life estimated at <5 s.²⁶ Injecting the solution formulations directly into sites of inflammation would have limited impacts as ADO is rapidly eliminated in extracellular fluids due to enzymatic- and transporter-mediated mechanisms, as the nucleoside is eliminated by adenosine deaminase (ADA), an ubiquitous enzyme, and equilibrative and concentrative nucleoside transporters.²⁷ In intratracheal and intrathecal routes of administration, the volume that could be injected is limited by the anatomy, typically only a few milliliters in humans.^28,29 As such, the total drug exposure afforded by the saline formulation in acute inflammatory settings is limited. A concentrated ADO formulation would enhance the extent of drug exposure, potentially useful in ameliorating local inflammation.

Efforts have been made in constructing biomaterial systems of ATT.³⁰ Silk fibroin-based ADO-releasing implants have been optimized to control seizures in animal models of epilepsies.^1,15,31 The silk-based implants render the effective focal delivery of ADO into the brain, with demonstrated capacity to control drug release from weeks to years at rates of 0–1 μg per day.^16,32 In one design, zero-order ADO release was sustained for 2 weeks by encapsulating the drug molecules with eight layers of 8% (w/v) silk.^1,15 Feasibility is enhanced by the biocompatible and biodegradable nature of silk, which is as sutures in surgical procedures. ADO-functionalized polylactic acid (PLA)-based nanoparticles are found to be effective in impeding tissue damage in a rat model of osteoarthritis.³³ A composite patch made of polylactic-co-glycolic acid (PLGA) and gelatin renders the sustained release of ADO onto cardiac tissues and mitigates reperfusion injury.³⁴ An ADO-releasing scaffold protects the loss of transplanted islets in mice,³⁵ and a bone-targeting nanocarrier of ADO attenuated bone loss in an osteoarthritis model.³⁶ While these and other systems of ATT have been largely designed for attaining sustained concentrations of ADO in target tissues, their effectiveness in acute inflammatory settings in which bursts of high doses of the nucleoside are likely needed is uncertain. Moreover, the effort required for scale-up manufacturing of clinical grade of some of the delivery systems would involve extensive optimization and validation of not only the API but also the polymers. Herein, we report a carrier-free, crystalline form of ADO stabilized as a suspension in which concentrations more than two times that of the licensed products can be delivered in the same volume. Herein, we describe a facile, solvent-free recrystallization process in which particles of a different crystal habit than the starting material were obtained. Thermal data confirming a previously unreported monohydrate form of the ADO crystal are presented, evidence showing that the crystallinity is preserved in a protein-rich environment is shown, and a biological effect of the crystalline suspension is demonstrated in vivo.

MATERIALS AND METHODS

Chemicals.

Adenosine (purity > 99%) was purchased from Acros Organics (Waltham, MA). Bovine serum albumin, fraction V, heat shock isolated was purchased from Spectrum Chemical Mfg. Corp (New Brunswick, NJ). Lipopolysaccharide (LPS) from Escherichia coli was purchased from Novus Biologicals (Centennial, CO). Fetal bovine serum and rat alveolar macrophages, NR8383 [AgC11x3A, NR8383.1] (ATCC CRL-2192), were purchased from ATCC (Manassas, VA). Penicillin–streptomycin, l-glutamine, Dulbecco’s phosphate-buffered saline (DPBS), ACK lysis buffer, Nigericin sodium salt, and endotoxin-free sterile water and HPLC grade methanol were purchased from Thermo Fisher Scientific (Waltham, MA). Antimouse IL-6 (MP5-20F3) APC antibody was purchased from BioLegend (San Diego, CA). RPMI medium was purchased from Life Technologies (Gaithersburg, MD). Rat IL-1β uncoated enzyme-linked immune assay (ELISA) was purchased from R&D Systems (Minneapolis, MN). Mouse TNFα uncoated ELISA kit, antimouse CD16/CD32 and antimouse CD86 (B7-1) PE antibodies, and buffers used in flow cytometric analyses were purchased from Invitrogen (Waltham, MA).

Recrystallization of Adenosine.

A formulation of recrystallized ADO with water or BSA was developed using a multistep crystallization process, as depicted in Figure 1. The preliminary step was to form uniformly sized, nonaggregated crystals of ADO from its supersaturated aqueous solution. ADO was dissolved in water at 20 °C at a concentration of 50 mM. The saturated solution was then heated at 70 °C for 20 min to ensure the dissolution of the API. Supersaturation was introduced by cooling the hot solution on ice for 20 min during which nucleation was observed. The cooled ADO solution was then agitated at 200 rpm for 2 min, which almost instantly yielded crystals with a flocculated appearance. The agitation speed was increased to 500 rpm to break down the agglomeration and generate uniform and nonaggregated crystals.

Figure 1.

Schematic of the proposed method developed for the generation of crystalline ADO. A facile method for generating initial crystal suspension by cooling crystallization of ADO. The crystals were collected and resuspended in either distilled water or BSA solution (15% w/v) to yield the final suspensions (25 mM) rcADO and rcADO_BSA, respectively.

These crystals were stored at room temperature to equilibrate with the mother liquor for 72 h prior to the next step of collection of the crystals followed by resuspension with either water or BSA. The prepared ADO crystals (50 mM), which contained half of the added drug in the crystalline phase as confirmed by HPLC analysis, were collected by centrifugation at 2841g (g-force) for 5 min. The centrifuged crystals were separated from the supersaturated ADO solution (supernatant) and were resuspended with distilled water or aqueous BSA solution (15% w/v), as depicted in Figure 1. The resuspension volume for the crystals separated from the supernatant was kept the same as the volume of 50 mM crystal suspension centrifuged, which yielded a final crystal suspension containing 25 mM ADO. The crystalline formulations resuspended in water or BSA were allowed to equilibrate/stabilize at 4 °C for a period of 7 days and referred to as rcADO or rcADO_BSA, respectively.

Scanning Electron Microscopy.

The crystal morphology of the prepared formulations was examined using a Hitachi S-3400N-II scanning electron microscope (SEM) (Hitachi, Tokyo, Japan) equipped with an AXS XFlash detector 5010 (Bruker, Billerica, MA) and a Bruker Nano e-Flash 1000+ (Bruker, Billerica, MA) under vacuum. Images were collected under conditions of variable pressure with a 5.00 kV accelerating voltage, an 80.0 V probe current, and a working distance of 10.0 mm. Samples were prepared by filtration of the crystal suspension followed by air drying and mounting onto a carbon dot mount prior to imaging.

X-ray Powder Diffraction.

X-ray powder diffractometry (XRD) studies were performed on an X-ray diffractometer (Rigaku, MiniFlex II) with an X-ray generator for Cu Ka radiation (ka1 = 1.54056 A°) on filtered and air-dried crystal samples. Data were collected in the discontinuous scan mode using a step size of 0.01°(2θ). The scanned range was 3–40° (2θ).

Thermal Analytical Methods.

All differential scanning calorimetry (DSC) studies were performed on a DSC Q1000 equipped with a refrigerated cooling system (TA instruments) at a scan rate of 10 °C/min. All samples were separated from the solution using filtration and air-dried before all analyses. Two to five milligrams of each sample was weighed and placed on the sample holder, and samples were run under a nitrogen atmosphere. A dry purge of nitrogen gas (50 mL/min) was used for all runs. Thermogravimetric analysis (TGA) was performed using a TGA Q500 (TA instruments) under a nitrogen atmosphere (20 mL/min) in open platinum pans. Samples were heated from room temperature to 400 °C at a heating rate of 10 °C/min.

Light Scattering and Absorbance Measurements.

The colloidal nature of the crystal suspension was studied by evaluating the optical density at 600 nm using a NanoDrop One^C spectrophotometer (Thermo Fisher Scientific). The colloidal suspension was further explored for the presence of drug in the solution phase with increasing dilutions by reading the absorbance at 260 nm. The crystal suspensions were diluted serially with distilled water, vortexed, and analyzed using the same spectrophotometer. The dilution factor was expressed as the final volume after diluting with water for the analyzed sample relative to the initial volume (50 μL).

Fluorescence Quenching.

Fluorescence spectroscopic method was used to evaluate the binding interaction of BSA and ADO by measuring the intrinsic tryptophan fluorescence of BSA in the presence of ADO. The emission spectra for BSA from 300 to 400 nm were read using a TECAN infinite M1000 microplate reader (Männedorf, Switzerland) at an excitation wavelength of 280 nm. The BSA concentration was maintained constant at 0.1 mM, while the ADO amount added was varied from 0 to 50 mM. All samples were prepared in PBS, and scans were performed at different temperatures (293 K (19.85 °C), 300 K (26.85 °C), and 315 K (41.85 °C)). Reproducibility in measurements was ensured by performing each measurement in triplicate.

Infrared Spectroscopy.

A Nexus 470 Fourier transform infrared (FTIR) spectrometer equipped with a diffused reflectance infrared FT (DRIFT) attachment was used for FTIR analysis to evaluate the specific interactions involved in the binding of ADO-BSA and intermolecular interactions in rcADO and ADO in solution. The rcADO, rcADO_BSA, BSA solution, and ADO solution samples were dried on a stainless steel coupon (1 cm × 1 cm) and measured for comparison over the 4000–500 cm⁻¹ range under nitrogen gas purge conditions with 256 scans and a 4 cm⁻¹ resolution.

Dissolution.

The kinetics of ADO dissolution from the formulated crystals were studied using a Slide-A-Lyzer dialysis cassette (Thermo Scientific, 3.5 kDa cutoff) stirred at room temperature for 72 h in a beaker containing phosphate-buffered saline (PBS) under sink conditions. The dialysis cassettes were presoaked in the medium as recommended by the manufacturer followed by loading 1 mL of the respective samples in the dialysis cassette. The cassettes were immersed in a beaker containing PBS with continuous stirring. ADO solution (3 mg/mL) and ADO crystal suspension (25 mM) from the vendor served as controls. At various time points, 1 mL of the medium was collected for measurement using an Agilent 1200 HPLC equipped with a UV detector set at 280 nm. A TSK Gel 4000 SWXL precolumn was used. ADO was quantified using a reverse-phase HPLC method using a C 18 Hypersil GOLD column purchased from Thermo Scientific. The mobile phase used was phosphate buffer pH 4.5 and methanol (90:10) under isocratic flow conditions at a flow rate of 1.5 mL/min. ADO concentrations were quantified based on UV absorbance at 260 nm using a standard curve.

In Vivo Anti-Inflammatory Activity.

Adult male and female C57BL/6 mice in-bred at Duquesne University Animal Care Facility were used for this study. Each mouse was subcutaneously injected with 20 μL of the rcADO crystal suspension mixed with 20 μL of LPS solution (20 μg) in their right footpad. The control groups were injected with 20 μL of HyPure water and 20 μL of LPS solution. All animal use protocols were approved by Duquesne University Institutional Animal Care and Use Committee (IACUC). After 24 h, the popliteal draining lymph node from the injection-side foot and spleens were collected from each mouse. The collected spleens and draining lymph nodes (DLNs) were stored and processed using RPMI medium supplemented with 10% fetal bovine serum (FBS), penicillin–streptomycin, l-glutamine, and 50 μM of 2-mercaptoethanol. Spleens and DLNs were individually processed to obtain a single cell suspension, which was further used for staining for flow cytometric analysis of the treated cells. Briefly, each of the draining lymph nodes was crushed using a sterile syringe plunger against a 70 μm Corning Falcon cell strainer (Thermo Fisher). The cell suspension was transferred to a centrifuge tube, and the cells were collected by centrifugation. Cell pellets were resuspended in a 50 μL FACS flow cytometry staining buffer. The harvested spleens were cut at both ends and crushed by following a similar procedure as described above. The cell suspension was centrifuged, and the collected cell pellets were further subjected to treatment with red blood cell lysis buffer before resuspending in complete medium for cell culture experiments.

Flow Cytometric Analysis.

The flow cytometric staining procedure was performed in low-retention microcentrifuge tubes for both DLNs and splenocytes. DLN cells and splenocytes suspended in FACS buffer were incubated for 15 min with antimouse CD16/CD32 antibody (1 μg/10⁶ cells) to block nonspecific Fc binding. The cells were then stained with 0.125 μg/10⁶ cells of antimouse CD86 (B7-1) PE for 30 min on ice. Following washings, the cells were permeated and stained with 0.125 μg/10⁶ cells of antimouse IL-6 (MP5-20F3) APC antibody with room-temperature incubation for 30 min. After repeated washings with staining buffer, antibody-stained cell samples were analyzed with an Attune NxT flow cytometer (Thermo Fisher). Gating was applied to exclude debris and doublet cells. Color compensation was adjusted using samples processed similarly to other samples but contained just the individual antibodies against IL-6 and CD86.

Ex Vivo LPS Restimulation of Treated Splenocytes.

Splenocytes harvested from mice injected subcutaneously with LPS with either rcADO or water were harvested and processed, as described previously. Splenocytes were seeded in a 24-well cell culture plate in RPMI complete medium at a density of 1 × 10⁶ cells per well. Splenocytes were rechallenged by the addition of LPS at a concentration of 0.5 μg/mL. For each of the spleen samples, cells without the added LPS served as unstimulated controls. Splenocytes were cultured for a duration of 24 h after which the cells were harvested. The harvested cells were centrifuged and resuspended with FACS flow staining buffer. Flow cytometric analysis was performed by staining and analyzing the cells as previously described. The supernatants were collected and analyzed for TNFα concentrations by ELISA.

Anti-Inflammatory Effect of ADO on Activated Rat Alveolar Macrophages (RAMs).

RAMs were cultured in F12K medium containing 10% heat-inactivated FBS, 50 μM of 2-mercaptoethanol, and penicillin–streptomycin. The cells were seeded at a density of 10⁶ cells/mL in 48-well tissue culture plates in 500 μL of F12K complete medium. Cells were treated with the respective treatments, i.e., ADO solution (3 mg/mL), rcADO, rcADO_BSA, or 25 mM ADO suspension for 30 min prior to inflammasome activation. To study the effect of delivering a high drug amount in a limited treatment volume by the crystalline ADO formulation, the rcADO suspension was tested by either adding a fixed dose (748 μM) as the ADO solution or adding the same volume as the solution required for achieving the target dose. For all other ADO-containing controls, the added ADO concentration was 748 μM. Cells were stimulated for inflammasomes with LPS (1 μg/mL) for 2 h followed by 1 h of treatment with nigericin (20 μM final concentration). The supernatants were collected and analyzed for IL-1β concentrations by ELISA.

Statistical Analysis.

All data analysis was performed using GraphPad Prism 9 (San Diego, CA). Data were analyzed using a one-tailed unpaired t-test for comparison between two groups and one-way ANOVA with Tukey’s multiple comparison test. Asterisks are used to mark statistically significant values and are represented as nonsignificant (ns), p ≥ 0.05, *p < 0.05, and **p < 0.01.

RESULTS

We sought to develop highly concentrated formulations of ADO and made the serendipitous discovery that the nucleoside recrystallizes in supersaturated aqueous solutions upon cooling (Figure 1). This observation led to a series of experiments conducted to determine the physical properties of the crystals. The final suspension was obtained by collecting the precipitates by centrifugation and resuspending the resulting solids in either distilled water or in a solution of bovine serum albumin (BSA), with each containing 6.67 mg/mL of ADO, more than 2× that of the concentration in the licensed products. No organic solvent or polymeric carriers were used in the process or contained in the final suspension. Resuspending the precipitates in albumin (BSA, 15% w/v) was used to inform the behaviors of the ADO crystals in protein-rich environments. The crystals obtained in the recrystallization process are denoted hereafter as “rcADO”, for ADO suspended in water, or “rcADO_BSA” for crystals suspended in BSA (Figure 1).

Characterization of the New Crystal Form.

We first investigated the morphology of rcADO using light microscopy (Figure 2A), in which the photomicrographs revealed a needle-like material, with lengths averaging 88.84 μm and a standard deviation of 22.55 μm (Figure 2D); the materials were estimated to have an aspect ratio of less than 1:100. The size variation of the particles in rcADO, based on the percent coefficient of variance (%CV), was relatively narrow compared to that of the raw material (Figure 2C,F). The morphology and size distribution of rcADO_BSA resembled those of rcADO (Figure 2B,E).

Figure 2.

Microscopic analysis and size distribution of ADO crystals. Micrographs and size distribution of particles counted for rcADO (n = 322) (A, D), rcADO_BSA (n = 337) (B, E), and ADO (vendor; n = 267) (C, F) represented as median size ± standard deviation.

SEM images show a meshlike fibrous network formed by needle-shaped precipitates in rcADO and rcADO_BSA (Figure 3A,B), in contrast with the tabular crystals seen in the raw material (Figure 3C). Crystallinity of the needle-shape materials was confirmed by powder X-ray diffraction (XRD) analysis of air-dried samples of rcADO, rcADO_BSA, and the raw material (“ADO”; Figure 4A). The diffraction pattern of the raw material overlaps with the known crystalline form of ADO (ICDD PDF #00-035-1977), while the nonoverlapping peaks seen in rcADO (and rcADO_BSA) suggest that a new crystal form was obtained in the aqueous process (Figure 4A). Stability of the crystals is demonstrated in the microscopic imaging of suspensions stored at 4 °C after 72 days (Figure S6). DSC thermograms show that all samples registered an endothermic event at 508.85 K (235.7 °C; Figure 4B), corresponding to the known melting point of ADO. For rcADO and rcADO_BSA, but not the raw materials (ADO), an endotherm event occurred at 383.15 K (110 °C), which was followed by an exothermic event at 422.03 K (148.88 °C), which may be indicative of recrystallization.³⁷

Figure 3.

SEM micrographs of the crystalline ADO forms; images of filtered and air-dried samples of rcADO (A), rcADO_BSA (B), and ADO (crystals procured from the vendor) (C). The top and bottom panels represent images procured at 500× and 1000× magnifications, respectively. Samples were also air-dried directly on a carbon mount, and images show similar fiberlike morphologies as filtered samples (Figure S3). The scale bar can be seen inside micrographs.

Figure 4.

Evidence and characterization of the crystalline nature of the prepared formulations. XRD analysis indicating the crystalline nature of the formulations (rcADO and rcADO_BSA) as evidenced by sharp peaks at distinct positions than the raw material (ADO) (A), DSC thermograms showing thermal events in rcADO and rcADO_BSA samples with broad endothermic peaks centering 105 °C in comparison with ADO (B), and TGA analysis indicating the stoichiometric loss of water in the rcADO sample corresponding to the monohydrate form of crystals in comparison with ADO with no water loss indicating anhydrous form (C).

The distinct peaks in the infrared spectra of air-dried samples of ADO dissolved in water (“sADO”, 2.67 mg/mL) and rcADO suggest unique hydrogen-bonding patterns in the two (Figure 5A,B). Functional group assignments point to unstructured hydrogen bonding in sADO but a degree of orderliness of such in rcADO. Peaks consistent with the stretching of O–H in alcohol (3463 cm⁻¹) and ribose (3324 cm⁻¹) in rcADO were found merged in sADO,³⁸ while both samples show sharp peaks assigned for the N–H stretching. Broadening of the region 3200–3000 cm⁻¹ in sADO and the red shift of the peak in rcADO suggest loss of the hydrogen-bonding interaction of the amino group.³⁹ Collectively, the XDR, DSC, and IR results indicate that a unique crystalline form was generated in the process.

Figure 5.

FTIR analysis of ADO/BSA samples. Binding interactions show spectral differences in compared samples, with either individual or merged peaks observed in the respective spectra corresponding to free or hydrogen-bonded O–H and NH₂ stretching for ADO solution (A), rcADO (B), rcADO_BSA(C), and BSA (D).

Probing the Crystalline Form in rcADO.

The crystalline form of rcADO was investigated further using thermal methods. DSC thermogram of rcADO but not that of the raw material exhibits a broad endothermic peak at around 378.15 K (105 °C; Figure 4B), which would correspond to the loss of water. The exothermic peak at 421.15 K (148 °C) in rcADO, which is absent in the raw material, suggests a crystallization event. Together, these events indicate that rcADO underwent heat-induced dehydration followed by recrystallization. TGA shows loss in weight at 373.15 K (100 °C) for rcADO beyond which the weight remains constant up to the decomposition temperature (Figure 4C). The 6% reduction in weight corresponds stoichiometrically to a monohydrate form. No significant difference in thermal stability was found in either rcADO or the raw material each exhibits a single melting peak (T_m) at 509.15 K (236 °C) (ADO mp = 235.5 °C), United States Pharmacopeia, and a decomposition peak at 558.15 K (285 °C) (Figure 4C). The water loss-induced transformation was further evidenced in the hot-stage polarized light microscopic analysis (Figure S1). No visible change in the morphologies was seen during the heating phase up to the melting temperature, but a gradual loss of birefringence, in the analyzed orientation, was seen when heating beyond 423.15 K (150 °C), coinciding with the exothermic peak seen in rcADO. Melting was confirmed at 508.15 K (235 °C) in rcADO, indicating that the crystallinity remains despite the apparent loss of birefringence. Taken together, these data indicate that rcADO exists in a monohydrate crystalline form distinct from the starting raw material.

ADO Crystals in Albumin Solution.

We next examined the extent to which the ADO crystals would retain their physical form in concentrated protein environments since the extravasation of albumin is associated with inflammatory injuries, for example, in distressed airways.⁴⁰ The crystalline characteristics of the ADO crystals resuspended in BSA (15% w/v) resemble the crystals resuspended in water, as evidenced in the overlapping 2-theta angles (Figure 4A), and that the putative recrystallization peak was observed in the thermograms of both samples (Figure 4B). SEM images suggest albumin associating with but not disrupting existing crystals (Figure 3B), pointing to an adsorptive process reminiscence of the albumin coating of paclitaxel crystals.^41,42 Infrared spectroscopic data also indicate intermolecular interactions between ADO and BSA (Figure 5C), with the shifting of the peak centering at around 3500 cm⁻¹ (O–H stretching of alcohol group) in the BSA alone sample to 3417 cm⁻¹ in rcADO_BSA. Taken together, these results demonstrate the capacity of rcADO to retain its crystallinity in the protein-rich environment.

Molecular analysis of the ADO interaction with albumin has been described,^43–45 and we confirmed here in an aqueous solution of BSA (0.1 mM, 0.66% w/v) that the binding occurs in a concentration-dependent manner, quantified based on the quenching of intrinsic tryptophan fluorescence⁴⁶ (Figure 6A).

Figure 6.

Binding studies of ADO with BSA. Fluorescence quenching of the intrinsic fluorescence of BSA (0.66% w/v, 0.1 mM; λ_em = 280 nm and λ_ex = 330 nm) with an increasing concentration of ADO (0–50 mM, (a–i)) (A), positive deviation in the Stern–Volmer plot (F₀/F vs [ADO], M) at different temperatures (B), and modified Stern–Volmer plots for the determination of thermodynamic binding parameters studied at different temperatures (C). Temperatures studied are represented as 293 K (green circles), 300 K (red circles), and 315 K (blue circles).

The thermodynamic parameters for the ADO-BSA binding were calculated using the Stern–Volmer analysis

$$\text{log}\left(F_0-F\right)/F=\text{log}{K}_b+\text{log}\left[A\right]$$ (1)

where F₀ and F represent the fluorescence intensities of BSA in the absence and presence of ADO, respectively, [A] denotes the molar concentration of ADO, K_b represents the binding constant of ADO to BSA, and n is the number of binding sites. The double logarithmic function (log[(F₀ − F)/F] vs log [A]) plotted at the three temperatures (Figure 6C) tested was used to determine the values of K_b and n (Table 1). The negative ΔG values indicate that binding between ADO and BSA occurs spontaneously. The Stern–Volmer analysis (F₀/F vs conc.) yielded positive deviations at all three temperatures (Figure 6B), indicating the presence of transient intermolecular collision (dynamic) and stable binding (static) modes of interactions.^47–49 These results suggest that albumin may serve as a solubilizer of ADO in the crystalline suspension (Figure 1) and may be considered a method for producing cocrystals.⁵⁰

Table 1.

Thermodynamic Parameters of ADO Binding to BSA

temp (K)	Kb (M−1) × 106	n	R 2	ΔG (kJ M−1)
293	45.70	0.6791	0.9653	−42.963
300	84.527	0.8541	0.9603	−45.525
315	133.96	1.033	0.8978	−49.007

Dissolution of Crystalline ADO.

Dissolution of rcADO was studied using membrane dialysis and optical scattering (Figure 7). The same volume (1 mL) of rcADO suspension (6.67 mg/mL), rcADO_BSA, the raw material (ADO), or ADO solution (3 mg/mL) was introduced into separate dialysis cassettes (MWCO = 3.5 kDa), and concentrations of the nucleoside in the receiving reservoirs of buffered saline were measured using reverse-phase HPLC. As expected, twice the concentrations of ADO were recovered from rcADO relative to solution ADO (Figure 7A). The crystalline solubility was calculated to be 3.465 mg/mL for rcADO and 4.004 mg/mL for rcADO_BSA (Figure 7B,C), estimated based on light scattering and the corresponding increase in concentration using the method as described by Lindfors et al.⁵¹ Stepwise dilution of rcADO shows the dissolution (260 nm) coinciding with the disappearance of particulates (600 nm) (Figure 7B,C), suggesting a uniform suspension system. Intriguingly, no scattering was observed in the raw material (ADO; Figure 7D). This observation could be attributed to the non-homogeneous nature of the ADO suspension. Unlike rcADO, the ADO suspension was observed to form a cake of the sedimented ADO (Figure S2). The relative impact of rcADO and solution ADO (3 mg/mL) was tested in cultured alveolar macrophages stimulated with lipopolysaccharides (LPS), a proinflammatory substance derived from Gram-negative bacteria, and a Toll-like receptor 4 agonist. Given in the same volume, rcADO conferred superior inhibition of IL-1β production in the cells compared to solution ADO (Figure S4). These results indicate that rcADO exists as a suspension from which consistent bioactive doses of ADO can be expected.

Figure 7.

ADO dissolution kinetics from different formulations and optical light scattering analysis. Comparison of dissolution kinetics of ADO depicted as the cumulative ADO released using dialysis cassettes in release medium (125 mL, PBS 7.4) from rcADO (blue circles), rcADO_BSA (green circles), ADO from the vendor (red circles), and ADO solution of marketed strength (3 mg/mL) (purple circles) (A), changes in the scattering intensity (OD 600 nm) and presence of ADO in solution phase (OD 260 nm) as a function of dilution for rcADO (B), rcADO_BSA (C), and ADO (D). Simple linear regression was performed on OD 600 data for panels B and C, and the ADO concentration corresponding to the X-axis intercepts was determined as crystalline solubility. OD 600 values for ADO (D) could not be determined, probably due to sedimentation and cake formation (Figure S2).

Localized Anti-Inflammatory Effect In Vivo.

To determine the potential of rcADO as a modulator of loco-regional inflammatory responses in vivo, the crystalline suspension was injected subcutaneously into a mouse hind footpad with LPS (Figure 8A). The injection mixture contained rcADO or sterile water and LPS into one footpad in each mouse. Draining lymph nodes (popliteal) and spleens were retrieved 24 h post injection for flow cytometric analysis of cells expressing IL-6, a proinflammatory cytokine implicated in acute and chronic inflammatory pathogeneses.^52,53 The limited volume that could be injected into the footpad subcutaneous space served to assess the feasibility of using rcADO in localized drug delivery. Conventional ADO solution (3 mg/mL) would only render 60 μg of ADO, and the raw material (ADO) would likely yield lot-dependent dose variation, given the relatively wide size distribution (Figure 2). The expectation was that rcADO would dissolve as a function of convective fluid flux in the subcutaneous space, draining primarily to the popliteal lymph node adjacent to the stifle joint.

Figure 8.

Immunomodulatory effects of rcADO. Flow cytometric analysis of cell populations harvested from mice draining lymph nodes (DLNs) and spleens following footpad injections (40 μL) with LPS (20 μg) with or without rcADO (133.5 μg) (A), representative cell populations and gating applied for flow cytometric analysis (B), dot plots of %IL-6⁺CD86⁺ cells harvested from footpad draining lymph nodes in mice injected with water and rcADO, respectively, (C, D), effect of rcADO treatment on %IL-6⁺CD86⁺ from popliteal lymph node cell population (E), and production of TNFα in ex vivo cultured splenocytes challenged with LPS (F).

Cells in the popliteal lymph nodes injected with rcADO were found to exhibit lower frequencies of CD86⁺IL-6⁺ cells in the lymphocyte populations relative to cells recovered from mice injected with LPS alone (Figure 8C–E). The CD86⁺ cells in the gated lymph node population were most likely B cells (Figure 8B). The protection appears to extend to distant lymphoid organs, as ex vivo cultivated splenic cells in mice treated with rcADO produced lower levels of TNFα upon LPS challenge in vitro compared to control cells (Figure 8F), although no difference was observed in the CD86⁺IL-6⁺ population in these samples (Figure S5). These results demonstrate that rcADO injected in the subcutaneous space partially attenuated the effects of a proinflammatory insult. The lack of effect on the splenic CD86⁺IL-6⁺ population following rcADO treatment highlights the potential of the formulation to exert localized anti-inflammatory effects.

DISCUSSION

The ability to administer high-dose ADO into the confined tissue space is conducive to repurposing the nucleoside as an immunomodulator by avoiding hemodynamic side effects and rapid plasma clearance. The currently marketed ADO formulations are solutions at 3 mg/mL strength, requiring the administration of large volumes to achieve pharmacologically relevant concentrations, which is not feasible in certain inflammatory locales. Here, we report a crystalline formulation produced using a facile, organic solvent-free crystallization method, yielding a suspension in which more than 2× the ADO dose is contained in the same volume of the marketed formulations. The suspension contains acicular crystals with a relatively narrow particle size distribution, while the raw material exists as cubic, tabular shapes (monoclinic space group P21 with four molecules in the asymmetric unit),⁵⁴ with a wider size variation. The crystallinity of rcADO was not perturbed in the presence of high concentrations of albumin, which appeared adsorbing to the crystals,^41,42 with the implication that the formulation could be sustained in protein-rich bodily fluids, such as in the inflamed airway. Albumin is a biologically inert endogenous plasma protein with multiple binding sites for small organic molecules, including ADO.^45,48,49

It should be noted that after our initial discovery, a similar DSC/XPDR profile of ADO was reported in a patent application (C202110719209.3, China National Intellectual Property Administration), describing an entirely different process in which ADO crystals were generated using a tetrahydrofuran (THF)-evaporation process. Our work, however, provides previously unreported characteristics of a monohydrate crystal form of ADO, in addition to reporting an organic solvent-free crystallization method. Thermal analysis of rcADO indicates the loss of water that led to destabilization of the crystal lattice and subsequent molecular rearrangement in recrystallization (Figure 4). This was corroborated in hot-stage polarized imaging (Figure S1). Thus, the characteristics reported here defined the physical form of the needle-shaped crystalline ADO, rcADO. In addition to drug delivery applications, the process might be useful for removing impurities.⁵⁵

Evidence is accumulating in preclinical studies to justify repurposing ADO as immunomodulators.⁴ Notably, a recent study indicates that ADO can potentially attenuate the severity of ARDS in patients infected with the SARS-COV-2 virus by administering the drug solution (3 mg/mL) to patients through nebulization over a span of 10–15 min.^56,57 In intubated patients, however, high doses in limited volumes would need to be instilled intratracheally in a shorter time. While synthetic derivatives of ADO have been characterized pharmacologically, some of which have been evaluated in clinical studies, ADO remains an attractive candidate given its endogenous origin with an extensive safety profile in humans. While a twofold increase in concentration would not necessarily constitute a major change, it would mean the ability to deliver 100% more ADO in a given injection to tissue spaces in which limited volume could be accommodated. Reported AAT systems are largely aimed at achieving controlled concentrations for extended durations. We posit here that rcADO may be used to attain a burst of supraphysiological drug concentrations at the site of inflammation to overcome metabolic and transport clearance mechanisms. Relative to polymeric systems, the organic solvent-free suspension rcADO is produced using a facile method, which might be more conducive for scale-up manufacturing, and may serve as a starting point from which ADO delivery systems can be developed.

In conclusion, the crystalline ADO (rcADO) described here was generated using a solvent-free aqueous system exhibiting a previously unreported crystal form and habit. The data indicate that rcADO remained crystalline in a concentrated albumin solution, suggesting its potential as a carrier-free entity by which the site-specific delivery of relatively higher doses of ADO can be attained in small anatomical spaces in relatively short durations.