Retention of peptide-based vesicles in murine knee joints after intra-articular injection

Abstract

While intra-articular injections continue to be utilized in the remediation of osteoarthritis, the efficacy of such treatment regimens continues to be plagued by the rapid clearance of drug therapeutics from the joint space, independent of whether the therapeutic is a macromolecule or a small-molecule pharmaceutical. Nanoparticle drug delivery vehicles have emerged as potential candidates to enhance the retention of therapeutics within the joint, with a wide variety of formulations that utilize retention strategies such as electrostatic interactions, collagen-type II binding domains, and passive targeting based on the physical size of the particles. However, despite these advances, significant issues regarding carrier and drug retention within the joint remain, so we sought to explore a novel nanoparticle retention scheme based on the integration of collagen peptides. In this work, we demonstrate the successful synthesis and fluorophore labeling of elastin-b-collagen peptide nanovesicles (ECnV). We further demonstrate the matrix/tissue retention and survivability of these ECnVs after novel tissue clearing processes that aid observation and detection of the nanovesicles in vitro. We also demonstrate the retention of these ECnVs in vivo and show successful localization of the ECnVs within murine knee joints. This study represents the first demonstration of the retention of elastin-b-collagen peptide-derived nanovesicles within joints, and suggests their potential for enhancing intra-articular treatment approaches for osteoarthritis.

1. Introduction

Osteoarthritis (OA), the most common form of arthroses [1], affects ~23 million people in the United States [2] and ~303 million cases have been reported worldwide [3]. While considered a disease of the whole joint [4–6], two hallmarks of OA are i) degradation of cartilage extracellular matrix (consisting of collagen II and proteoglycans), and ii) inflammation [7–13]. Left unchecked, these processes invariably lead to tissue degeneration/dysfunction, pain, joint stiffness, immobility/disability, and a markedly reduced quality of life [14–16].

Presently, therapeutic approaches for OA are limited to palliative symptomatic relief; predominantly via the use of non-steroidal anti-inflammatory drugs (NSAIDs) or anti-inflammatory glucocorticoids [12]. However, some pre-clinical work suggests that numerous therapeutics, including glucocorticoids, bisphosphonates, and biologics, may possess direct disease-modifying attributes [12,17,18]. Regardless of their mode of action, these treatment strategies typically rely upon parenteral delivery via intra-articular (IA) injections [19–23]. IA injections are favored for OA treatment because they localize therapeutics to the afflicted region and this in turn limits drug loads, enhances bioavailability, minimizes off-target effects, decreases costs, and increases patient compliance compared to orally or systemically administered therapeutics [22–24]. One exemplar approach is IA corticosteroid injections (IACS) [12,14,22]. Unfortunately, despite more than four decades of clinical IACS implementation [24], including recent ‘conditional’ to ‘strong’ recommendations for IACS injections (by the Osteoarthritis Research Society International and American College of Rheumatology, respectively [25,26]), the efficacy of IA corticosteroids remains controversial [27–30]. Indeed, several meta-analyses report only short-term pain improvement following injection (1–4 weeks), with long-term pain scores after IACS being largely indistinguishable from sham, saline, or no injection controls [31–34]. A potential factor in the shortcomings of IACS treatments, as well as most other IA therapies, is the lack of physical retention of therapeutic compounds, like corticosteroids, within the joint space; thereby limiting the in-joint bioactivity of the compounds, and necessitating higher injection doses or repeat injections [23].

Drug clearance across the synovium is very efficient due to a combination of rapid synovial fluid turnover and extensive venous and lymphatic drainage pathways [35,36]. The half-lives of IA administered NSAIDs and corticosteroids are reported to be on the order of just a few hours [23,37,38]. Similarly rapid clearance extends to macromolecular therapeutics; IA administered hyaluronates have retention half-lives ranging from 12 to 48 h depending on molecular weight and cross-linking [39,40]. Moreover, biologics such as the interleukin-1 receptor antagonist, Anakinra (branded Kineret), and the tumor necrosis factor antagonist, Adalimumab (branded Humira), are also rapidly cleared from the synovium after IA injection [23,41,42]. Thus, multiple IA injections and/or higher injection doses are typically needed to maintain IA drug concentrations above their effective therapeutic thresholds, but there is reluctance to perform/receive more than 2–4 IA injections per year owing to the potential for procedural complications and drug side-effects (e.g., chondrotoxicity) [23,43,44].

In aggregate, these findings suggest that new strategies are needed to not only increase the retention of drugs within the joint but also provide a sustained IA release that enables long-term therapeutic benefits. A common method for modifying the diffusive properties of therapeutics, and therefore their release and clearance from the joint, is to encapsulate them within or conjugate them to, a carrier system [38]. Several drug carriers for extending the release of therapeutics from within the joint have been explored, including: liposomes [45,46], hydrogels [47,48], polymeric nano- [49,50] and micro-particles [51,52], viral [53,54] and non-viral [55,56] gene delivery vectors, proteins [57] and protein-polyelectrolyte complexes [58]. Beyond providing a diffusive barrier to therapeutic release into the joint microenvironment, many drug carrier systems exploit the physical and chemical properties of the intra-articular environment to promote extended IA retention [59–61]. Cationic nanocarriers [62], dendrimers [63], and proteins [57] can be used to electrostatically deliver to and retain carriers and therapeutics in the anionic proteoglycan-rich cartilage of the joints. In addition to electrostatic targeting methodologies, bioactive targeting modalities such as antibodies [64] and peptides [65,66] that bind type II collagen have been assessed for bioactive-mediated joint retention. Similarly, hyaluronic acid (HA) that binds to CD44 (which is overexpressed in arthritic conditions) [67], has been explored as a targeting modality for increased retention [55]. While it has been suggested that targeting certain pharmaceuticals, such as biologics and glucocorticoids, to the full depth of cartilage tissue (e.g., via charge affinity) would be ideal for improving their therapeutic outcomes or even disease-modifying effects [68], sequestering of drug carriers within the synovium may also prove beneficial in alleviating pain and inflammation [66].

Despite the demonstrated pre-clinical advantages of the noted IA delivery approaches, limitations to their success remain. Microparticles are unable to effectively target and penetrate intra-cartilage tissue, and may be less effective at targeting biologics or corticosteroids to chondrocytes within the ECM where their disease-modifying effects might be best elicited [37,68]. Although cationic small drug carriers overcome such limitations, their use may be restricted to early primary (idiopathic) and secondary (post-traumatic) OA therapies [69]. This is because as OA progresses, the content of negatively charged proteoglycans decreases in cartilage and in turn, eliminates the inherent affinity that cationic carriers possess toward the oppositely charged ECM [12]. Indeed, degraded cartilage from diseased joints exhibits reduced cationic nanoparticle retention [70,71], and the cationic features of these carriers are not ideal for targeting synovial tissues.

Thus, an improved drug carrier system should be capable of: 1) long-term retention within the joint, 2) active targeting of specific joint tissues (e.g., cartilage, synovium, etc.) for enhanced local, cell-specific therapeutic efficacy, and 3) retention of this targeting capability despite (or even enhanced by) possible damage to native tissue ECM. Such active targeting has been accomplished via antibodies, however, in the context of drug carrier systems, their use is limited by their large size (~160 kDa) [72]. Additionally, antibodies are prone to immunogenic reactions, and can show reduced stability relative to other therapeutics [73]. Given these drawbacks, small bioactive/biomimetic peptides offer significant advantages as active targeting ligands [74,75].

Collagen-like peptides (CLPs) offer advantages for ECM-related applications, as they have widely been demonstrated to hybridize/bind to degraded collagen proteins [76–81]. A typical CLP possesses a (Glycine-X_AA-Y_AA)_n repeat sequence where n is typically between 6 and 10 repeats and the X_AA and Y_AA residues are usually proline and hydroxyproline, respectively [82,83]. The hybridization (or binding) of CLPs to, and their integration into, degraded/denatured collagen occurs through the triple helical folding of a monomeric strand(s) of CLP with the unfolded collagen triple helices in denatured collagen fibrils [84,85]. CLPs and other type–II–collagen-binding peptides can target and hybridize to denatured collagen in the joints of healthy, non-OA-bearing mice, which has been hypothesized to occur due to the accumulation of turnover products (i.e., degraded collagen) within the cartilage [71, 86,87]. It is reasonable to expect binding enhancement in the presence of the larger quantities of degraded and denatured collagen II and collagen I that is characteristic of OA cartilage and synovium, respectively, highlighting their potential use for active targeting, delivery, and retention of therapeutics to OA joints [88,89].

Our group has previously demonstrated that CLP sequences conjugated with an elastin-like peptide (ELP) yield ELP-CLP conjugates that are capable of self-assembling into nanovesicles with a bilayer comprised of a hydrophobic ELP core and hydrophilic CLP inner and outer shell [90,91]. We have also demonstrated that these elastin-collagen nanovesicles (ECnVs) are capable of hybridizing to type II collagen films [92]. Therefore our present objective was to investigate the localization and retention of ECnVs in the joints of mice through the use of our previously studied ELP-CLP sequence, (VPGFG)₆G-(GPO)₇GG [91]. To permit visualization of ECnVs within the joint tissue in vivo and ex vivo, the surface of the ECnVs was modified with a fluorophore. The conjugate was verified to have a T_m above that of physiological temperature by circular dichroism spectroscopy (CD) and the labeled ECnVs were characterized by fluorescence correlation spectroscopy (FCS) and transmission electron microscopy (TEM) to determine their size and morphology. We demonstrate not only that these ECnVs can bind to a collagen protein hydrogel in situ but that they survive the harsh tissue processing and clearing procedures required for the visualization and analysis of their distribution and binding in vitro/ex vivo. Furthermore, through IA injections into murine joints and subsequent longitudinal in vivo imaging and multi-scale ex vivo/in situ microscopy, we demonstrate greater than week-long retention of empty nanovesicles within the joint. Moreover, we find that these ECnVs target and distribute to numerous target tissues within the joint, and observed no accumulation in non-target organs such as the spleen, kidney, liver, etc. The work demonstrated herein highlights the exciting possibility of utilizing ECnVs to actively target drug therapeutics to degraded collagens that are typically found in OA joints.

2. Materials and methods

2.1. Synthesis and purification of peptides and peptide conjugates

The CLP peptides N₃-(GPO)₇GG-CONH₂, N₃-(GPO)₇GG-COOH (N₃ = N-terminal azide) and the ELP peptide NH₂-(VPGFG)₆G’ (G’ = alkyne-bearing propargyl glycine) were synthesized via standard Fmoc-based solid phase peptide synthesis (SPPS) techniques. The N₃-(GPO)₇GG-COOH CLP was further reacted with ethylene diamine ([EDA] Millipore Sigma, St. Louis, MO) to yield the modified CLP, N₃-(GPO)₇GG-EDA for use in fluorescently labeling the ECnVs. N₃-(GPO)₇GG-CONH₂ and N₃-(GPO)₇GG-EDA were each conjugated separately to the ELP NH₂-(VPGFG)₆G’ via copper(I)-catalyzed azide-alkyne cycloaddition, to yield two ELP-CLP conjugates, NH₂-(VPGFG)₆G-(GPO)₇GG-CONH₂ (henceforth F6-GPO7) and NH₂-(VPGFG)₆G-(GPO)₇GG-EDA (henceforth F6-GPO7-EDA), respectively. Details regarding the synthesis of peptides, their modifications, conjugations, and relevant purifications are provided in the Supporting Information (SI) Section 1.1. Figs. S1–S7 show representative total ion chromatograms and their corresponding mass spectra.

2.2. Circular dichroism spectroscopy of ELP-CLP conjugates

To inform ECnV formation and CLP-mediated ECnV hybridization to collagen tissues, the triple helical propensity and triple helical thermal stability of the ELP-CLP conjugate(s) were characterized. Since a 70:30 M ratio of F6-GPO7:F6-GPO7-EDA was used for ECnV formation, circular dichroism (CD) spectroscopy wavelength and temperature scans were performed on a 70:30 M ratio of F6-GPO7:F6-GPO7-EDA conjugates at a total concentration of 200 μM in pH 7.4 1xPBS (Thermo Fisher Scientific, Waltham, MA). It should briefly be noted that this ratio was chosen in order to minimize surface charge (from the amine) and preserve the self-assembly of F6-GPO7 alone (as in our previous studies), while also still including a minor fraction of a reactive moiety for fluorophore conjugation. All measurements were performed with a Jasco 1500 circular dichroism spectrometer (Jasco, Easton, MD). Detailed procedures and calculations are provided in the SI Section 1.2. Representative wavelength scans of the conjugates at 4 °C and 80 °C and temperature scans are shown in Fig. S8 and Fig. S9, respectively.

2.3. Formation of ELP-CLP nanovesicles (ECnVs)

A 70:30 M ratio (200 μM total in water) of F6-GPO7:F6-GPO7-EDA was utilized for ECnV formation, which was performed by a stepwise reduction in temperature in the cuvette well of a dynamic light scattering (DLS) Zetasizer Nano ZS instrument (Malvern Panalytical Inc., Westborough, UK). Detailed procedures for ECnV formation are provided in the SI Section 1.3. Representative correlation curves, measured intensity-average hydrodynamic diameters, and the polydispersity index for a dilute ECnV suspension are provided in Fig. S10.

2.4. Dye labeling and purification of ECnVs

After the ECnVs were formed, they were covalently labeled with a far-red fluorophore to permit their observation in matrices/tissues in vitro, in vivo, and ex vivo after optical clearing. Given that nearly 30% of the ECnVs in the current formulations carry amine functionality, the amine-reactive Alexa Fluor^® 633 N-hydroxy succinimidyl (NHS) ester dye (Thermo Fisher Scientific, Waltham, MA) was employed for ECnV labeling. Detailed labeling reaction and purification procedures are provided in SI Section 1.4. Representative images of a purified labeled ECnV pellet and a corresponding control are provided in Fig. S11.

2.5. Fluorescence correlation spectroscopy (FCS) characterization of labeled ECnVs

Successful fluorophore labeling of the ECnVs was determined qualitatively by physically viewing the centrifuged particle pellet for color (Fig. S11) and quantitatively by FCS. Photon count rates and fluorescence autocorrelation data were collected on a Zeiss 710 ConfoCor confocal microscope (Ex. 633 nm/Em 638–747 nm; Carl-Zeiss, Oberkochen, Germany). Particle concentration (particles/mL) was estimated by counting vesicle transits through the Gaussian observation volume, while hydrodynamic diameter was predicted, utilizing the Stokes-Einstein equation, based on particle diffusivity, which was determined by fitting the measured autocorrelation function of diffusing labeled particles with an appropriate three-dimensional diffusion model. Specific FCS equations, procedures and calibration methods are described in SI Section 1.5. The Alexa Fluor^® 633 calibration correlation function, fit, and residuals as well as the fit parameters (assuming a diffusion coefficient of 340 μm²/s) are provided in Fig. S12. Photon count rate data used for the calculation of ECnV concentration are provided in Fig. S13.

2.6. Transmission electron microscopy of labeled ELP-CLP nanovesicles

Labeled F6-GPO7-AF633 nanovesicles (AF633-ECnVs) were prepared and stained via a triple negative phosphotungstic acid staining protocol that was previously described [91]. Carbon film 400 mesh size copper grids were ionized with a PELCO easiGlow^® (Ted Pella Inc., Redding,CA) glow discharge unit prior to sample and stain spotting with neutralized 1% wt/wt phosphotungstic acid (Millipore Sigma, St. Louis, MO) in water solution. Sample and stained grids were imaged with a Carl Zeiss Libra 120 (Carl-Zeiss, Oberkochen, Germany) transmission electron microscope (TEM) with a high-tension voltage of 120 V. Image analysis and nanovesicle diameter determination was performed with ImageJ software (National Institutes of Health, Bethesda, MD) [93].

2.7. In vitro and ex vivo AF633-ECnV binding and chemical stability

The binding and chemical stability of AF633-ECnVs to collagen was assessed in harvested murine joint tissues as well as Geltrex^® matrix hydrogels (Thermo Fisher Scientific, Waltham, MA) comprising type IV collagen, laminin, entactin, and heparin sulfate proteoglycans that served as a tissue analog for initial testing. For the Geltrex^® matrix hydrogels, prior to the addition of AF633-ECnVs, the temperature of the AF633-ECnV solution was raised to 45 °C and held for 5 min to partially denature the CLP domains. 40 μL of AF633-ECnVs (~589 × 10^6 vesicles/mL (see SI Section 1.5)) were added to 200 μL of Geltrex^® and mixed to disperse the AF633-ECnVs in the hydrogel. This vesicle/gel solution was drawn into a 1 mL syringe and incubated at 37 °C for 20 min to elicit gelation. AF633-ECnV-laden gels were then extruded from the syringe, cut, fixed in 4% paraformaldehyde (Thermo Fisher Scientific, Waltham, MA), and washed for further assessment. Joint tissue harvesting and preparation is discussed below.

Microscopic observation of ECnVs and their target tissue interactions requires their survival through tissue processing procedures that aid the visualization of nanovesicle presence, distribution, and ECM binding in situ; this includes procedures for tissue fixation, optical tissue clearing (for 3D imaging) and paraffin embedded tissue processing (for histology). Following fixation, ECnV-laden gels were subjected to either aqueous or non-aqueous optical clearing protocols, including SeeDB [94] and thiodiethanol (TDE) [95], and benzyl alcohol-benzyl benzoate (BABB) [96] and tetrahydrofuran-dibenzyl ether (THF-DBE) [97] procedures, respectively. Additional samples were subjected to standard paraffin embedded histological processing steps, including decalcification (Formical-200-, StatLab, McKinney, TX), ethanol-based dehydration, xylene-clearing, and hot paraffin infiltration [98]. Specifics regarding clearing and processing protocols can be found in the SI Section 1.6. Following optical clearing/processing, hydrogels were mounted for in situ, far-red 3D fluorescent imaging on either a Zeiss AxioObserver.Z1 Apotome.2 (Ex. 580–604 nm/Em 625–725 nm) with super-resolution, structured illumination microscopy (SIM) capabilities (0.41μm/pixel); or a Zeiss LSM880 confocal microscope (Ex. 633 nm/Em 638–747 nm; 1.38 μm² × 10 μm/pixel, 50 to 100 slices collected).

2.8. Intra-articular injection and imaging of AF633-ECnVs

Male BALB/cJ mice (n = 15; Jackson Labs, Bar Harbor, ME) underwent intra-articular (IA) injection of AF633-ECnVs to permit the longitudinal study of nanovesicle retention within the joint via in vivo imaging. At ~12 weeks of age, mice were anesthetized using inhaled isoflurane, both lower limbs were shaved, and while under anesthesia, were transferred to an in vivo imager to capture baseline fluorescent intensity images of the animal and the antero-medial aspect of their knee joints (see below). Then a sterile approach was used to make a small (~2 mm) incision in the skin overlying the medial side of each limb (5–10 mm from patella). 6 μL of AF633-ECnVs (~589 × 10⁶ vesicles/mL in sterile saline; or 3.5 × 10⁶ particles/injection) were administered through this incision, via an anteromedial-to-posterolateral approach, with an ~10-degree vertical angle of insertion, into the intra-capsular space immediately medial to the patella using a micro-liter glass syringe remotely attached to a 36-gauge ‘micro’-needle (World Precision Instruments, Sarasota, FL). The left joint served as a sham operated control, receiving only IA saline. Incisions were closed with tissue glue (Gluture, Zoetis Inc, Kalamazoo, MI), and mice were immediately transferred to the in vivo imager to capture in vivo AF633-ECnV particle fluorescence emitted from the knee following IA injection. Bacitracin and buprenorphine were administered prior to surgery for prophylactic infection and pain control. All animal procedures were approved by the University of Delaware Institutional Animal Care and Use Committee (IACUC).

In vivo imaging of AF633-ECnV retention in the joint was performed on a UVP iBox Scientia imager (with integrated anesthesia), using far-red fluorescent excitation and emission filters (Ex. 600–645 nm/Em 700–740 nm) and a grey-scale CCD camera. All images were captured using identical excitation lamp intensities and exposure times (750 ms). Fluorescent images were captured immediately prior to IA injection, immediately after injection, and then at 2, 4, 8, 24, 48, and 72-h, and 4, 5, 6, 10, 14, and 21-day time points following IA injection. UVP software was used to segment and extract AF633 fluorescence signals from the injected and control knee joints, and to normalize the signal to both background fluorescence and the intensity of a fluorescent phantom included in each capture. At the 7, 21, and 28-day time points, intact lower limbs along with select organs (heart, lungs, liver/spleen, kidneys, and brain) were harvested for multi-scale ex vivo/in situ microscopy (n = 5 mice per group/time point) following standard euthanasia protocols.

2.9. Multi-scale imaging of AF633-ECnV distribution, targeting, and retention

Immediately after harvesting, excorticated joints and isolated organs were subjected to macro-scale, ex vivo far-red fluorescence imaging with an iBox Scientia as previously described. Intact joints and organs were then fixed and cleared—using the THF-DBE protocol—prior to being similarly imaged ex vivo for detailed macro-scale assessment. Next, cleared tissue were mounted for 3D confocal imaging (via a LSM880), before being ‘de-cleared’ for processing for paraffin-embedded histology. De-cleared and decalcified joints were embedded in paraffin and sectioned at 5 μm/section and placed atop charged microscope slides. Sections were left i) unstained or ii) stained with toluidine blue, and cover-slipped for fluorescent-ECV and histological evaluation on an upright wide-field brightfield and epi-fluorescent scope (Ex. 484–495 nm, 630–650 nm/Em. 501–623 nm, 655–755 nm; Axio. Imager, Zeiss) with a color CCD camera (MRc5; Zeiss).

3. Results and discussion

3.1. Preparation and characterization of AF633-ECnVs

Through the application of advanced optical tissue clearing techniques and longitudinal in vivo imaging/multi-scale bio-imaging approaches, we sought to investigate the ability of our ECnVs to be retained in both in vitro and in vivo collagenous environments via CLP mediated hybridization to native collagen protein and survive the requisite tissue processing approaches for such assessments. To this end, the peptides N₃-(GPO)₇GG-CONH₂, N₃-(GPO)₇GG-COOH, N₃-(GPO)₇GG-EDA and NH₂-(VPGFG)₆G’ were synthesized via SPPS and purified via reverse phase high performance liquid chromatography (Figs. S1–S5). From these peptides the ELP-CLP conjugates F6-GPO7 and F6-GPO7-EDA were synthesized using the common copper(I)-catalyzed azide-alkyne cycloaddition reaction and purified via reverse phase HPLC (Figs. S6 and S7) to generate ECnVs that could be labeled with fluorophores.

We previously demonstrated that the CLP domain of ELP-CLP conjugates is critical to the self-assembly of ECnVs [91]. The CLP domain folds into a triple helix upon cooling below its melting temperature (T_m), which then triggers hydrophobic collapse of the ELP domain above its inverse transition temperature (T_t), resulting in ECnV self-assembly [90, 91]. Using circular dichroism spectroscopy, we confirmed that a 70:30 M ratio (200 μM total concentration) of F6-GPO7:F6-GPO7-EDA formed triple helices at 4 °C but not 80 °C, as indicated by a positive ellipticity at 225 nm at 4 °C that was absent at 80 °C (Fig. S8) [83]. The T_m of the conjugate mixture was measured by monitoring the 225 nm ellipticity as a function of heating from 4 °C to 80 °C and was determined to be 46.5 °C (Fig. S9), a value marginally smaller than our previous reported value of 50 °C for F6-GPO7 [91] (without the EDA modification). The decrease is likely due to charge repulsion between amines of the EDA on the C-terminus of the conjugate [99,100]. It is briefly worth noting that this observation suggests that had there been more F6-GPO7-EDA conjugate in the total F6-GPO7 to F6-GPO7-EDA ratio that the T_m could have decreased further and possibly compromised the temperature-mediated self-assembly process of these conjugates into nanovesicles.

Despite this modest decrease in T_m, the circular dichroism data suggest that vesicle formation should be possible given the ability of the CLP conjugate to fold into triple helices, as per our previous work [90, 91]. Self-assembly and characterization of the 70:30 F6-GPO7: F6-GPO7-EDA conjugates was monitored via DLS (at a peptide concentration of 200 μM in water). Fig. S10a shows a decrease in the diffusion coefficient as the conjugate solution is cooled from 80 °C to 25 °C, indicating nanovesicle formation. These data indicate the formation of nanovesicles coincides with the formation of the triple helix secondary structure, consistent with our previous studies [91]. After formation, the ECnVs were diluted nearly ten-fold to a concentration of 77 μg/mL so as to be non-interacting and monodisperse, allowing for accurate measurement of their hydrodynamic diameter, which was determined to be ca. 135 nm at 25 °C (Fig. S10b). This size was comparable to the diameters of other ELP-CLP nanovesicles we have studied previously [90, 91].

After formation, the mixed 70:30 F6-GPO7:F6-GPO7-EDA nanovesicles were modified with an Alexa Fluor^® 633 NHS ester (AF633-NHS) fluorophore as depicted in Fig. 1a. It is worth noting that only a small portion of reactive amine (≤30 mol %) was included in the nanovesicle formulation to limit any possible interference of the dye label on CLP-mediated hybridization of the vesicles to collagen containing matrices. Incubation of the freshly made nanovesicles with solubilized AF633-NHS facilitated labeling of the ECnVs (Fig. 1a) which were purified and observed after ultracentrifugation (Fig. S11). Specific details of the reaction and purification are provided in Section 1.4 of the SI. The resuspended AF633-labeled ECnVs were then characterized by FCS. Representative AF633-ECnV autocorrelation data and the FCS analysis fit are shown in Fig. 1b. Via FCS analysis, the diffusion coefficient of the AF633-ECnVs was found to be 3.4 μm²/s, resulting in an estimated hydrodynamic diameter of 144 nm (per Stokes-Einstein equation). This diameter is comparable to the measured hydrodynamic diameter of the unlabeled ECnVs and is consistent with our previous studies [90,91]. The increase in hydrodynamic diameter between the unlabeled ECnVs and the AF633-ECnVs may be attributed to the influence of covalent surface modification with the relatively bulky AF633 dye [101]. More importantly, FCS only allows for the detection of fluorescent signals, allowing us to confirm that the ECnVs are successfully fluorescently labeled and that they retain their size and colloidal stability post labeling. Additionally, based upon linear trajectories analysis, one-dimensional mean square displacement calculations could be utilized to predict the number of vesicles/mL, which we estimate to be 588 million/mL. Detailed calculations and underlying theory for these calculations are provided in the SI, Section 1.5.

Fig. 1.

ECnV formation, fluorophore labeling and characterization via FCS and TEM. a) Schematic of nanovesicle formation and synthesis strategy for the conjugation of the Alexa Fluor^® 633 fluorophore to the ECnV surface. Two ELP-CLP conjugates, F6-GPO7 (amide bearing C-terminus) and F6-GPO7-EDA (amine bearing C-terminus) are combined in the vesicle formation process at a 70:30 M ratio, respectively. Effective labeling of the ECnVs proceeds by incubation with the Alexa Fluor^® 633 NHS ester. b) FCS autocorrelation data of AF633-ECnVs in 1xPBS. Raw correlation data is plotted as the black squares (left y-axis), the overall correlation function in red (left y-axis), and the fit residuals in blue (right y-axis). c) Transmission electron microscopy of AF633-ECnVs against a negative phosphotungstic acid stain. The scale bar is 200 nm and shown in black and white for clarity.

To complement FCS analysis, we characterized the morphology of the AF633-ECnVs via TEM. Fig. 1c shows a representative, negatively stained TEM image of AF633-ECnVs. The morphology of these nanovesicles is consistent with our previous work, indicating largely spherical nanoscale vesicles having a vesicular bilayer structure that is more opaque than the central portions of the ECnV [91]. Additionally, analysis of images (such as that in Fig. 1c) indicates that the AF633-ECnVs have a diameter of 122 ± 17 nm (n = 15), which is generally consistent with the FCS data, with a slightly smaller diameter that may be expected given that the TEM data are collected on dried samples.

3.2. In vitro imaging and survivability of AF633-ECnVs

After labeling the ECnVs, we wished to show that AF633-ECnVs hybridized to collagen matrices in vitro could be detected via fluorescent microscopy after having been subjected to optical tissue clearing/processing techniques. Solvent-based optical clearing and histological processing techniques employ chemicals that can alter both the retention and chemical integrity of bound tissue moieties such as nanoparticles and fluorophores [102,103]. Thus, in addition to demonstrating in vitro AF633-ECnV retention, we subjected AF633-ECnV-laden matrices to optical tissue clearing and formalin-fixed paraffin-embedded tissue processing techniques to establish their suitability for use in multi-scale in situ imaging and analysis.

Freshly prepared AF633-ECnVs (warmed to 45 °C, ≈T_m of the ECnVs (Fig. 9) to allow for CLP-mediated hybridization to collagens [76,85]) were added to collagen IV-containing Geltrex^® solution prior to gelation. The Geltrex^® hydrogels were then subjected to fixation and various optical tissue clearing/processing techniques (see Section 1.6 of the SI for details) Immediately following fixation, widefield epi-fluorescent imaging, structure illumination microscopy (SIM), and confocal microscopy were used to establish the baseline presence, distribution, fluorescent signal intensity, and integrity of AF633-ECnVs within the collagen hydrogels (n = 3 per clearing/processing approach; representative SIM images of gels, post-fixation, in water, are shown in the left column of Fig. 2).

Fig. 2.

Qualitative assessment of AF633-ECnV retention and survival in Geltrex^® gels following chemical processing for optical clearing via a) aqueous (TDE and SeeDB) and b) non-aqueous (BABB and THF-DBE) clearing protocols.

Longitudinal fluorescent imaging of the presence, distribution, and intensity of ECnV fluorescence throughout (not shown) and at the conclusion of optical clearing demonstrated, in a qualitative manner, the robustness of AF633-ECnVs to the aqueous TDE and SeeDB (Fig. 2a), and the non-aqueous THF-DBE-based (Fig. 2b) tissue clearing approaches; similar nano-vesicle retention, distribution, and sizes as their baseline (post-fixation) counterparts were observed among these. Each final disposition image (right column) shows paired pre- and post-clearing gels imaged under identical SIM parameters. At the termination of THF-DBE-based clearing, AF633-ECnVs exhibited similar distribution and appearance as their baseline condition; however, in DBE cleared and mounted samples, qualitative assessment suggests that the fluorescent intensity of the labeled vesicles appears to be modestly decreased. Despite this modest reduction in fluorescent signal intensity, the improved tissue clearing ability and depth of imaging permitted by DBE’s higher refractive index (R.I. 1.56, compared to 1.47 and 1.50 for TDE and SeeDB, respectively) partially offsets this loss in intensity [104]. The integrity of AF633-ECnV assembly and fluorescent intensity were also robust to extended time in TDE, SeeDB, and DBE, with no further observable changes detected over two-plus weeks in each final ‘mounting’ solution (data not shown).

The retention and stability of AF633-ECnVs in Geltrex^® gels was also observed through the decalcification and processing steps necessary for standard paraffin-embedding based histology. We observed that optical clearing via BABB drastically altered in situ nano-vesicle appearance and distribution (Fig. 2b); individual AF633-ECnV fluorescence was greatly diminished/lost at the conclusion of clearing, with the development of large, intense fluorescent aggregates suggesting a compromise of AF633-ECnV integrity. These changes appeared to be mediated by the terminal BABB solvent, as AF633-ECnVs survived all preceding dehydration steps through graded ethanol, as well as the xylene clearing and molten paraffin infiltration necessary for histological processing (data not shown). For the indicated reasons, and because of the difficulties in handling/imaging specimens within the high-viscosity SeeDB mountant, all subsequent optical tissue clearing and in situ imaging approaches herein utilized THF-DBE-based clearing.

3.3. In vivo imaging of intra-articular AF633-ECnVs

The utility of fluorescently labeled far-red nanovesicles for longitudinal in vivo ECnV imaging and assessment of their retention and clearance following intra-articular (IA) injection was investigated in a murine model. Following IA injection of 6 μL of AF633-ECnV containing solution into the right knee (~588,000 vesicles per μL in sterile saline), robust far-red fluorescent signals were detectable immediately (0 h) and then out to >10 days within depilated knees (images longitudinally tracking in vivo fluorescence in a representative IA AF633-ECnV and contralateral IA saline-injected animal are shown in Fig. 3a); no fluorescent signal was observed in saline-only injected knees. In the AF633-ECnV-injected limbs, AF633-ECnV signal often emanated from the small surgical incision site (see injected knee closeups in Fig. 3a). A surgical approach is required to administer therapeutics IA in the mouse due the relative size of the injection needle compared to the joint. Fluorescent signal at the incision location was due to a minor leakage of carrier from the joint space and needle end upon its retraction from the capsule, this carrier could become ‘entrapped’ in the closed incision (despite irrigation and sterile ‘swabbing’). For this reason, surgical incisions were made 5–10 mm medial to the patella, so that when the incision was moved laterally to visualize the injection site and then back to close the incision overlap of the IA and incision-derived signals could be prevented.

Fig. 3.

In vivo longitudinal imaging of IA AF633-ECnVs within murine knees. a) Longitudinal in vivo images from a representative mouse in which AF633-ECnVs were injected into the right knee; sterile saline was injected into the contralateral (control) limb. The upper images show the torso and lower limbs of the animal with the injected limb to the left and the contralateral limb, which exhibited no fluorescent signal, indicated via white arrows to the right. Fluorescence from the AF633-ECnVs (at 633 nm) is shown in red overlaid atop auto-fluorescence signals captured at 488 nm. The lower images focus on the AF633-ECnV injected limb demonstrating the intense fluorescent signals emanating from the injected joint, and to a lesser extent from the site of the surgical incision and i. a. injection approach (dashed boxes). b) Quantification of changes in background-corrected AF633 fluorescent intensity emanating from AF633-ECnV injected versus control joints. Data points indicate mean and standard deviation AF633 intensity from n = 15 mice; five mice each were imaged daily from 0 to 7 days (168 h, including imaging at 2, 4, and 6 h on d1), 0–21 days (504 h), and 0–28 days (672 h). c) Changes in AF633-ECnV injected limb fluorescence normalized to that of the contralateral limb. The half-lives and goodness-of-fit of the overall (i.e., global; t_1/2& r², respectively) and individual animal (i.e., replicates, ${\stackrel{-}{t}}_{1/2}$ and ${\stackrel{-}{r}}^2$ (see inset)) fits of the exponential decay curves (fitted from peak intensity at 24 h onward) are indicated for both the raw and normalized intensity data.

Quantification of in vivo AF633-ECnV fluorescent intensity over the course of up to 28 days following IA injection highlighted two intriguing behaviors. First, in analyzing the raw AF633-ECnV fluorescent intensity data, a linear increase in externally recorded fluorescent intensity (from ~14,000 to 22,000 a. u.) was observed in the knee over the first 8–24 h following IA injection [y = 795 a. u.*(t) + 16,318 (a.u), R² = 0.21, p = 0.007 (one sample t-test)]. This behavior suggests some degree of bulk ECnV redistribution, from deeper within the intra-articular space to closer to the capsular/joint/skin surface after IA injection (Fig. 3b), possibly due to transport within the joint during free ambulation following administration. Second, the fluorescent intensity emitted from the injected ECnVs, which we found to be quite stable under imaging conditions that mimic those encountered in our in vivo captures (10 × 750 ms-long imaging bouts, not shown), decayed exponentially with time from their 24-h intensity peaks, suggesting a degree of nanovesicle clearance/elimination from the joint. From the raw intensity data, the global fit of the mean decay response (Fig. 3b) predicted a signal intensity half-life of ~46 h [y = 24,225 * (1 – e^−0.015t), R² = 0.71], while a half-life of ~40 h (${\stackrel{-}{R}}^2=0.91$) was predicted based upon the mean of the individual specimen fits. It is briefly worth noting that this non-optimized result/retention time is at least as good as if not better than current accepted IA therapies, such as IA hyaluronic acid injections [38,39,66]. In the saline-injected joints, fluorescent intensity was effectively flat (at ~4000 a. u.) across the entirety of the study. When intra-articular AF633-ECnV intensity was normalized to the contralateral saline-injected joint (Fig. 3c), the initial increase in intra-articular AF633-ECnV intensity over the first 24 h was less apparent; however, the predicted signal intensity half-lives for AF633-ECnVs ranged from ~82 h (mean of individual fits; ${\stackrel{-}{R}}^2=0.85$) to ~104 h (fit of the mean response; ${\stackrel{-}{R}}^2=0.37$). Additionally, AF633-ECnV-injected joints exhibited statistically significant higher fluorescent intensities than their contralateral joints from initial injection out to 14 days post injection (p < 0.05; one-sample t-test). Such behaviors should not have been unanticipated given that the estimated size of the ECnVs is between 120 and 150 nm. The smallest diameters that are reported to allow synovial escape are on the order of 40–250 nm [105,106], suggesting that particles with diameters greater than 250 nm are required for reliable retention. Thus, the ECnVs are of a diameter which is likely to allow for at least some degree of escape from the joint, which is consistent with the predicted 50% IA clearance time between 64 and 128-h (predicted from our maximum intensity half-life times calculated from 24-h post-injection).

3.4. Multi-scale in situ imaging of AF633-ECnVs following intra-articular administration

Following longitudinal assessment of intra-articular AF633-ECnV retention/clearance in vivo, mice were necropsied at either 7-, 21-, or 28-days post IA injection (n = 5 mice per time point) for multi-scale assessment of IA administered AF633-ECnV distribution/localization within experimental and control limbs, and downstream tissues. Using ex vivo fluorescent imaging (via the iBox Scientia imager) strong AF633 fluorescence was macroscopically observed in both non-cleared (not shown) and optically cleared (in THF-DBE), intact, ex-corticated AF633-ECnV injected joints (a matched pair of saline-injected versus AF633-ECnV-injected limbs at 7 days post IA injection are shown in Fig. 4a). In all AF633-ECnV-injected joints, intense fluorescence signals were predominantly localized to the anterior aspect of the joint, distal to the patella; AF633 fluorescence was never observed in saline-injected contralateral limbs. Only one of the fifteen AF633-ECnV-injected limbs (at the 7-day time point) exhibited fluorescence outside of the articular joint proper; in this limb, two regions of fluorescence were detected in locations consistent with the draining lymphatics of the joint and the popliteal lymph node [107,108], as shown by arrows in the bottom row of images of Fig. 4a. No macroscopic evidence of AF633-ECnV presence/accumulation was observed among isolated organs that one might expect to accumulate systemically presented nanoparticles, e.g., liver/spleen, lungs, and kidneys (representative ex vivo images of intact and optically cleared organs are shown in Fig. 4b). Furthermore, no AF633 signals were identified in the ex vivo imaged carcass nor necropsy tissue remnants (not shown). Thus, ECnVs, possibly because of their relatively low concentration/number when administered IA, do not appear to appreciably accumulate in extra-articular off-target tissues; this behavior will need to be investigated further under higher IA ECnV dosing and be compared to systemic delivery. Nonetheless, the data from Figs. 3 and 4 suggest that ECnVs may be suitable for long-acting drug delivery post IA injection with limited risk of off target organ accumulation.

Fig. 4.

Macroscale ex vivo imaging of IA injected ECnVs within optically cleared mouse knees and organs. a) Ex vivo visible light and AF633 fluorescent imaging of intact, ex-corticated, and THF-DBE cleared lower limbs harvested 7-days following IA injection of either saline (left/contralateral limb) or AF633-ECnVs in sterile saline (right limb). No AF633 signal was recorded in any of the saline control limbs while robust AF633 signals were observed within the anterior and lateral aspects of the IA AF633-ECnV joints (see dashed ovals). In only this specimen (out of 15 total) was AF633 fluorescence observed away from the joint space proper, in what we presume to be the popliteal lymph nodes located posterior to the joint and femur (see white arrows). b) Ex vivo imaging of dissected and optically cleared brain, liver (and spleen), lung, and kidney tissues (tissue from the same 7-day time point specimen (as in (a)) is shown) demonstrated the absence of macroscale AF633 signal/accumulation within these tissues.

Following their ex vivo imaging, cleared, intact lower limbs from the 7-day time point post-injection groups were imaged via one-photon confocal microscopy to visualize the in situ 3D microscale distribution and localization of IA-administered AF633-ECnVs within the joint/limb (Fig. 5). Within the intact IA saline-injected joints, no AF633-ECnV fluorescence was observed (not shown). In contrast, AF633 fluorescence, and thus we assume nanovesicle presence, was easily observed across numerous regions of IA AF633-ECnV-injected joints. Because of differences in the optical depths at which tissues in the intact joint are imaged (and which could not be easily controlled/adjusted for), only a qualitative description of the in situ presence of fluorescent ECnVs and their distribution at the 7-day time point following IA injection is provided. Intense AF633 fluorescence was consistently found in the anterior region of the joint (Fig. 5), specifically in tissue consistent with the synovium and joint capsule. Within these capsular tissues, ECnV accumulation was pervasive, with regionally elevated accumulation appearing within the tissue that overlies the trochlear and femoral grooves distal to the patella; consistent with the postulate that joint articulation during ambulation could help to redistribute at least a portion of administered nanovesicles away from loaded soft-tissue contacts and toward the encapsulating synovium. Intense ECnV signal was also observed within/surrounding the capsular injection injury site (medial to the patella tendon), which is to be expected given that this is a site of tissue disruption/damage that undergoes rapid healing via soft tissue ‘scar’ formation and matrix remodeling while experiencing locally high ECnV numbers/concentration.

Fig. 5.

Confocal evaluation of IA AF633-ECnV presence within an intact and optically cleared AF633-ECnV injected mouse joint. a) A pseudocolored maximum intensity projection image of IA AF633-ECnV fluorescence from within an ~100 μm thick antero-medial region of an intact and THF-DBE cleared limb. The sagittal projection, captured through the medial aspect of the joint and at a depth of ~250 μm, qualitatively shows AF633 signal presence/intensity throughout a variety of intra-articular/intra-capsular tissues, including the synovium/capsule, fat pads/bursa, medial meniscus, and articular cartilage of the proximal medial femoral condyles and medial tibial plateau. Colored pixels within the image denote definitively positive AF633 signal; saline-injected contralateral limbs exhibited no AF633 presence and were completely ‘black’ (not shown). The image depicts an ~500 μm² region of interest. b) A three-dimensional rendering of AF633 fluorescent intensity within an optically cleared joint IA injected with AF633-ECnV. The image was reconstructed looking from the anterior-medial to posterior-lateral direction. The rendering illustrates the intense presence of AF633 signal associated with the capsular/synovial tissues, as well as the IA injection ‘injury’ site.

Within the intra-articular space, AF633-ECnV-derived fluorescence was observed in the intra-capsular fat pads, the supra- and infra-patellar bursa, and in association with the surfaces of the anterior and posterior horns of the medial and lateral menisci (Fig. 5a; only the anterior medial meniscal horn and medial femoral condyle and tibial plateau are shown in the pseudocolored intensity projection). AF633 fluorescence, at diminished intensity, was also observed within the articular cartilage proper of the anterior femoral condyles and anterior medial and lateral tibial plateaus, suggesting that ECnVs, or at the minimum AF633-labeled F6-GPO7, can be found at and within the articulating surfaces and cartilage proper of the joint. The in situ presence or absence of AF633-ECnV fluorescence deeper within the joint cannot be assessed given the depth-of-imaging limitations associated with the confocal microscope utilized. The presence of AF633-ECnV signal from within the draining lymphatics of the singular 7-day ECnV injected joint that was macroscopically observed (Fig. 4a), was confirmed by confocal microscopy (not shown). Additionally, a three-dimensional analysis of an optically cleared joint IA injected with AF633-ECnV is shown in Fig. 4b with AF633 fluorescence intensity shown in red, highlighting the presence of AF633 signal associated with the IA injection ‘injury’ site as well as the capsular/synovial tissues.

Lastly, the cleared 7-day time point AF633-ECnV and saline injected joints imaged via confocal microscopy were de-cleared, de-calcified, and processed for paraffin embedding for thin-section fluorescent imaging and histology. Far-red wide-field epi-florescent imaging of unstained sections in combination with visible light imaging of immediately adjacent toluidine blue stained sections permitted explicit demonstration that AF633-ECnV-derived fluorescent signals can survive the harsh processes of clearing, declaring, and paraffin processing/embedding. They also enabled the micro-scale confirmation of ECnV accumulation/presence throughout AF633-ECnV injected knee joints (a representative toluidine blue stained histology section from a IA AF633-ECnV injected joint and select examples of AF633-ECnV localization in an adjacent unstained section are shown in Fig. 6).

Fig. 6.

Fluorescent assessment of IA AF633-ECnV survival, presence, and localization within thin-sections from a decalcified and paraffin embedded murine knee. Epi-fluorescent imaging of far-red AF633-derived signals (Ex. 630–650/Em. 655–755) in combination with yellow-green tissue autofluorescence signals (Ex. 484–495/Em. 501–623; to visualize tissue/joint structure) highlighted the presence of robust AF633-ECnV-derived fluorescence throughout the representative 7-day post injection joint. The strongest fluorescent signals emanated from the anterior capsule/synovium and fat pad nearest the IA injection site, consistent with our multi-scale imaging findings. Clear AF633 signal was also observed in the interfaces between the patellar and femoral cartilages, the medial menisci and the femoral condyle and tibial plateau, and among/within other intra-capsular locations. AF633 signals were entirely absent from saline-injected contralateral limbs (not shown). The scale bar in each epi-fluorescent image denotes 50 μm. A thin section (5 μm) immediately adjacent to the imaged unstained section was stained with toluidine blue to provide a macroscopic overview of the joint geometry and the regions of the higher magnification (20×) epi-florescent images. Note: Fluorescent assessment of the presence of AF633-ECnVs in the histology sections were performed following optical clearing and in situ imaging; de-clearing, decalcification, standard ethanol dehydration processing, xylene-based clearing, and paraffin infiltration/embedding; thin sectioning; high temperature annealing/melting (60°C); and xylene-based de-paraffinization.

At 7-days post-IA injection, when looking across sagittal tissue sections from the most medial to most lateral aspects of the joint, ECnVs were clearly present. Indeed, signals from the AF633 fluorophore conjugated to the ECnVs was observed from just proximal to the patella (i.e., the quadriceps tendon insertion) to distal to the tibial plateau (i.e., the tibial synovial insertion), as well as across the entire anterior to posterior extent of the joint space (from the anterior to posterior joint capsule boundaries, and everywhere in between). Anteriorly within the joint, the most intense AF633-ECnV signals were localized to the capsule/synovium tissues between the patella and anterior menisci (forming a roughly trapezoidal region in the capsule); however, relatively strong AF633 fluorescence could be observed anywhere along the anterior capsule. This localized accumulation is consistent with the generalized ability of the synovium and capsule to efficiently ‘capture’ micro-/nanoscale vesicular carriers after their IA administration.

The next strongest signals encountered within the joint typically emanated from the anterior fat pad and bursa, again close to the initial injection site. Moving into the joint interface proper, positive, but less intense ECnV signals, were consistently observed between/within the articulating surfaces of the patella and trochlear groove, and occasionally at the quadriceps tendon articulation surface and surrounding the meniscal tibial ligaments. Centrally within the joint, unmistakable positive AF633-ECnV fluorescence, albeit less intense, was universally observed surrounding the anterior horns of the menisci, along the meniscal-femoral condyle and meniscal-tibial plateau surfaces, and at/within the femoral condyle-tibial plateau contact. Posteriorly, positive but weaker ECnV signals were observed on the meniscal margins and femoral condyles, and within the synovium/capsule. No AF633-fluorescence signal was ever observed in sections from saline-injected contralateral limbs (not shown). It is also noted that within the knees collected 7-days post-injection, there was no obvious/notable differences in the structure or composition of toluidine blue stained cartilage or other joint tissues among matched ECnV and saline injection joints, over this admittedly short observation window all experimental joints appeared qualitatively healthy. Further studies will need to be conducted to assess the potential for longer-term consequences of IA ECnVs on the health of cartilage and the joint.

Taken together, this pilot study assessed: 1) the capability of ECnVs to be successfully labeled with a tractable fluorophore while maintaining their morphology, 2) the sequestration of ECnVs in in vitro gels as well as in vivo tissues after exposure to state of the art tissue clearing protocols, 3) the retention of these AF633-ECnVs within murine knee joints post IA injection, and 4) the accumulation of AF633-ECnVs in tissues within the synovium/knee joint that are relevant for delivering future OA therapies to. The future use of additional imaging techniques will enable complete imaging of the whole joint in a 3D in situ manner for more comprehensive analysis of ECnV retention. The extended retention of our ECnVs in a healthy murine IA injection model, coupled with our prior demonstration of the controlled and triggered release of Dex from these nanocarriers, suggests that further developments of this CLP-based collagen-targeting nanovesicle platform offers significant opportunity to improve outcomes of the symptoms of OA.

4. Conclusion

In the presented work, we designed and characterized an alternative approach, using fluorophore-modified ECnVs, to enable retention of nanoparticles in collagenous matrices both in vitro and in vivo, for the purpose of establishing possible new vehicles for treating osteoarthritis (and other joint diseases). To the best of our knowledge, such an approach involving the binding of macromolecular moieties to cartilaginous collagen using CLP-mediated hybridization has not previously been studied for the purposes of treating/targeting OA.

Specifically, we demonstrate the successful surface labeling of our ECnVs and that the modification of the ECnVs with the fluorophore does not lead to a significant change in size or overall morphological characteristics of the nanovesicles. With the aid of advanced optical tissue clearing techniques, we also show that the labeled AF633-ECnVs are capable of being retained within collagen-containing hydrogel matrices. THF-DBE solvent-based tissue-clearing protocols yielded an optimal balance of ECnV stability/fluorescence and in situ visualization following tissue clearing and processing. We then assessed the retention of AF633-ECnVs within the murine knee joint. It was determined that the IA-injected, labeled ECnVs are retained in the murine knee joint environment for greater than 7 days with average half-lives ranging from ~40 h (raw data) to ~100 h (normalized data). These non-optimized results are at least as good as, if not superior to, existing macromolecular technologies that have previously been studied. Moreover, by studying the biodistribution of empty AF633-ECnVs following optical tissue clearing and in situ bioimaging, we could discern that IA injected ECnVs were not present in tissues that systemic nanoparticles typically accumulate, e.g., the liver and spleen. Rather, the AF633-ECnVs were strongly retained, in clearly observable quantities, within the injected knee joints of the mice. Closer examination revealed that nearly all the administered ECnVs could be found within the intra-articular/capsular space of the healthy knee joint, implying they were successfully retained within the joint. Lastly, we observed, through a combination of in situ and histological imaging approaches, that our non-optimized IA ECnVs appear to accumulate in/target several OA disease-relevant tissues within the joint, including the capsule/synovium, fat pad/bursa, articular cartilage surfaces, and other accessory tissue (e.g., menisci, tendons, ligaments, etc.). These findings have important implications for implementing a future ECnV system that is therapeutically active and capable of enhanced drug retention in OA joints. Future work will encompass studies not only of the retention and biocompatibility of IA ECnVs in healthy OA joints, but also the release of therapeutic cargo within the context of state-of-the-art OA models to improve outcome-driven OA treatment strategies.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.jddst.2022.103532.