Suprachoroidal Delivery in Rats and Guinea Pigs Using a High-Precision Microneedle Injector

Abstract

Purpose: Methods of injection into the suprachoroidal space (SCS) have been developed for larger animals and humans, but reliable administration to the SCS of rodents remains challenging given their substantially smaller eyes. Here, we developed microneedle (MN)-based injectors for SCS delivery in rats and guinea pigs.

Methods: We optimized key design features, including MN size and tip characteristics, MN hub design, and eye stabilization, to maximize injection reliability. Performance of the injection technique was characterized in rats (n = 13) and guinea pigs (n = 3) in vivo using fundoscopy and histological examinations to validate targeted SCS delivery.

Results: To enable SCS injection across the thin rodent sclera, the injector featured an ultrasmall, hollow MN measuring 160 µm in length for rats and 260 µm for guinea pigs. To control MN interaction with the scleral surface, we incorporated a three-dimensional (3D) printed needle hub to restrict scleral deformation at the injection site. A MN tip outer diameter of 110 µm and bevel angle of 55° optimized insertion without leakage. Additionally, a 3D printed probe was used to secure the eye by applying gentle vacuum. Injection by this technique took 1 minute to perform, was conducted without an operating microscope, and yielded a 100% success rate (19 of 19) of SCS delivery determined by fundoscopy and histology. A 7-day safety study revealed no notable adverse ocular effects.

Conclusions: We conclude that this simple, targeted, and minimally invasive injection technique can enable SCS injection in rats and guinea pigs.

Translational Relevance: This MN injector for rats and guinea pigs will expand and expedite preclinical investigations involving SCS delivery.

Introduction

The suprachoroidal space (SCS) has recently been recognized as a useful site of drug administration, driven by the introduction of simple, safe and reliable SCS injection using a microneedle (MN) delivery system¹ and demonstration of safety and efficacy of SCS injection in clinical trials.² The U.S. Food and Drug Administration recently approved a product for SCS injection of triamcinolone acetonide to treat macular edema³ and preclinical studies indicate that SCS delivery may be a preferred alternative for many ocular therapies that might otherwise be administered intravitreally, subretinally or by other routes of administration.⁴ 

The SCS is a potential space between the choroidal vasculature and the scleral wall of the eye in the posterior segment (Fig. 1A). SCS injections provide widespread, peripheral treatment coverage to the chorioretinal layer, enabling further transport to neighboring tissues like the retina.^5–9 Unlike its intravitreal counterparts, SCS injection is not hindered by the eye's inner limiting membrane,¹⁰ often resulting in a higher treatment concentration and gene transfection in target tissues.^6,7,10,11 Compared with subretinal injection, SCS injection does not involve vitrectomy, retinotomy or separation of photoreceptors from the retinal pigment epithelium, and it can be performed in an outpatient setting.⁵ Owing to the technique's less invasive nature, SCS injection may also have a lower risk of ocular complications like glaucoma, cataracts, and retinal detachment, which can occur with intravitreal or subretinal injections.^2,7,12–14 

Figure 1.

 

Ocular anatomy relevant to SCS injection. (A) An illustration of SCS injection by MN. Material (blue) injected between the sclera (white) and choroid (pink) travels circumferentially toward the posterior pole by expanding the SCS. (B) A relative comparison of eye size (mean axial length) and tissue thickness (mean scleral thickness) in human^15,16 versus rodent eyes.^17,18

Delivery to the SCS can be achieved via transscleral injections using either standard hypodermic needles or hollow MNs, which characteristically measure approximately 1 mm in length for use in humans and shorter in smaller animals.^19,20 Because their length can be controlled precisely to cross the sclera but not penetrate the choroid or retina, MNs are typically simpler to use and less invasive than hypodermic needles.^1,13,21 Administration to the SCS using MNs was pioneered in rabbits^1,8,13 and later executed in pigs,²² rhesus macaques,²¹ and horses.²³ Clinical trials also have used MN injectors to deliver triamcinolone acetonide to the SCS safely to treat patients with macular edema owing to noninfectious uveitis^3,24,25 and retinal vein occlusion,²⁶ without causing apparent changes to the underlying choroid.²⁷ 

Despite these benefits, SCS injections have largely been studied only in larger animals and have yet to be performed reliably in smaller animals like guinea pigs, rats, and mice. Rodents, however, are the most widely used animal models for preclinical research owing to ethical considerations, their availability, and lower cost. Additionally, there are more ocular disease models developed in rodents than in larger mammals.^28–30 Guinea pigs can model multiple vascular insults that occur with diabetic retinopathy and macular edema,³⁰ and their thin sclera facilitates the observation of drug diffusion in the eye.⁹ Wistar, SD, and RCS rats have been used to model glaucoma,³⁰ wet and dry age-related macular degeneration,^28,30 diabetic retinopathy,^29,30 macular edema,³⁰ and retinitis pigmentosa.^28,29 

The major barrier to SCS injections in rodents is the small size of their eyes (Fig. 1B). Although the sclera of rabbits, large pigs, and humans are roughly the same thickness at the ora serrata region (i.e., 400–450 µm),^15,16,31 the scleral thickness of guinea pigs and rats is two to four times thinner (i.e., 189 µm and 104 µm, respectively).^17,18 Researchers have used hypodermic needles (30G–34G) to perform rodent SCS injections under a dissecting or operating microscope, but these methods have lacked precise control over penetration depth and require a high level of training and skill for success, thus precluding the injection technique from widespread use.^5,11,12,32 

In this study, we sought to scale down the MN injector for SCS injection pioneered in larger animals for simple and reliable use in rodent models to make the injection technique more accessible and support the preclinical development of treatments that would benefit from SCS delivery. In doing so, we developed a high-precision MN injector that incorporates three-dimensional (3D) printed components for stabilizing the eye. Using this device, we demonstrated reliable SCS administration in rats and guinea pigs using a simple, fast, and minimally invasive procedure that does not require a microscope. 

Methods

SCS Injector Design and Fabrication

To fabricate hollow MNs, fire-polished aluminosilicate glass pipettes (part #A100-64-10, O.D. 1 mm, I.D. 0.64 mm, Sutter Instrument, Novato, CA) were pulled using a micropipette puller (P-97, Sutter Instrument). The resulting MNs were beveled at desired angle using a beveler device (BV-10, Sutter Instrument). Ethanol was then flushed through the MNs followed by two flushes of deionized water to clear the lumen from glass debris. Finally, MNs were individually housed in a 12-mm-long piece of stainless steel tubing (outer diameter [OD], 1.47 mm; wall thickness, 0.2 mm; McMaster-Carr, Douglasville, GA) and connected to a 10-µL Hamilton syringe (#7653-01, Hamilton, Reno, NV) via a fine screw fitting (M3-0.1, Base Lab Tools, Stroudsburg, PA). The extremely small thread on the screw fitting stabilized the MN in the hub and enabled fine adjustment of the MN length protruding from the tubing by moving the steel tubing forward and backward along the needle length. More specifically, the MN length was decreased by turning the fine screw clockwise, which pushed the steel tube hub forward and that, in turn, decreased the MN length protruding out of the hub. Turning the screw in the opposite direction pulled the steel hub back, thereby increasing the protruding MN length. 

The needle hub and vacuum eye stabilizer were designed via computer-aided design (Solidworks, Waltham, MA) and fabricated using a 3D printer (SLA Form 2, Formlabs, Somerville, MA). Because of their contact with ocular surfaces, these parts were printed with the highest resolution to provide a smooth surface finish, which was confirmed by inspection through a stereomicroscope (Olympus SZX16, Olympus, Tokyo, Japan). 

Animal Studies

All animal procedures complied with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by Georgia Institute of Technology Institutional Animal Case and Use Committee. Wistar rats (either sex, 6–9 months old), Brown Norway rats (either sex, 6–9 months old), and Hartley guinea pigs (either sex, 6–9 months old) were obtained from Charles River Laboratories (Wilmington, MA). 

SCS Delivery Optimization Studies

In this terminal procedure, MN injectors of various designs and specifications were tested for SCS administration in Wistar rats and Hartley guinea pigs. Each experimental variable was tested in three rats (6 injection attempts per group, 1 per eye) and two guinea pigs (4 injection attempts, 1 per eye). The injection solution contained 200-nm diameter green fluorescent nanoparticles (excitation, 505 nm; emission, 515 nm; FluoSpheres, Thermo Fisher Scientific, Waltham, MA) at 0.05% (w/v) concentration in Hank's balanced salt solution (Mediatech, Manassas, VA). In some cases, a blue dye, methylene blue (Sigma-Aldrich, St. Louis, MO), was added at 0.2% (w/v) concentration for visualization of fluid spread in the SCS without the need for fluorescence optics. Eyes were imaged by fundoscopy immediately after injection to assess SCS delivery and were then harvested for histological examinations after euthanasia. 

SCS Delivery Characterization Study

Using the developed injection technique, the same formulation as above containing nanoparticles in Hank's balanced salt solution was unilaterally (left eye) injected into the SCS of 10 Wistar rats, 3 Brown Norway rats and bilaterally in 3 Hartley guinea pigs. Eyes were imaged as described above immediately following injections. Animals were then euthanized, and eyes harvested for post mortem analysis. 

SCS Delivery Safety Study

A 0.025% (w/v) solution of fluorescein (AK-fluor 10%, Akorn, Lake Forest, IL) in Hank's balanced salt solution was injected, unilaterally (left eye) into the SCS of four Wistar rats and three Hartley guinea pigs, with the contralateral eye remaining naïve as a negative control. Eyes were imaged via fundoscopy at several timepoints before and after injection up to 7 days to monitor possible complications associated with the procedure. Upon termination of the study, animals were euthanized, and eyes were collected for histopathological analysis. 

Suprachoroidal Injection and Fundoscopy

After induction of general anesthesia in a chamber using 5% isoflurane in 800 mL/min oxygen flow, animals were placed on a heating pad and maintained under inhalation anesthesia (2%–3% isoflurane) for the duration of the procedure. In guinea pigs, a combination of inhaled isoflurane and injected ketamine:xylazine (60 mg/kg:7.5 mg/kg) was used, as needed, to induce full anesthesia, where the ketamine:xylazine was injected intramuscularly after a stable plane of anesthesia was achieved on isoflurane. 

One drop of topical anesthesia (0.5% Tetracaine Hydrochloride Ophthalmic Solution, Amici Pharmaceuticals, Melville, NY) was applied to each eye, followed by 1% tropicamide drops (Tropicamide Ophthalmic Solution, Henry Schein, Melville, NY) to dilate the eye. Excess fluid was gently wiped away using a cotton tip. Eye lubricant (GenTeal Tears, Alcon, Fort Worth, TX) was applied on both eyes to prevent dehydration. 

To expose the scleral surface, one eye was proptosed using the latex glove method.^33,34 Briefly, a small square piece (1.5 × 1.5 cm) was cut from a latex glove with a slit in the center into which the eye was placed such that it supported the eye and kept it proptosed. To secure the eye while injecting, the vacuum pump (Part # D2028B, AIRPO) was turned on and the custom-made eye stabilizer was placed on the inferior cornea to apply a gentle vacuum. 

The liquid formulation was injected by inserting the MN across the scleral supratemporally 1 mm posterior to the limbus. The MN was kept in place after the injection for 30 seconds to prevent reflux, after which the MN and latex square were removed, and the vacuum was turned off. Each injection took approximately 1 minute, starting from the glove proptosis until the vacuum was turned off and the glove removed. After injection, fundus images (fluorescence and brightfield) were collected using a RetCam II system (Clarity Medical Systems, Pleasanton, CA). 

For terminal procedures, the animals were euthanized via CO₂ asphyxiation. For survival studies, animals were allowed to recover at 37°C from anesthesia and returned to their cages when ambulatory. Animals were monitored throughout the course of study for any procedure-associated adverse effects. 

Histology

We used two separate techniques to process eyes after animal euthanasia. In SCS delivery optimization and characterization studies, eyes were enucleated and fixed in 10% formalin (Triangle Biomedical Sciences, Durham, NC) for 48 hours. Eyes were then cryopreserved by transfer into a solution of 10%, 20%, and 30% sucrose (MP Biomedicals, Solon, OH) sequentially and were kept in each solution until the tissue sank (6–12 hours). Before sucrose substitution, samples were rinsed twice with deionized water, and a small hole was made in the cornea to allow for faster sucrose infiltration and to prevent tissue deflation. Eyes were embedded in optimal cutting temperature media (Tissue-plus compound, Thermo Fischer Scientific), frozen using liquid nitrogen, and sectioned 10- to 20-µm-thick sections every 100 µm using a cryotome (CryoStar NX70, Thermo Fischer Scientific). Sections were counterstained with DAPI (Vectashield, Antifade Mounting Media, Burlingame, CA) and imaged via confocal microscopy (LSM700, Carl Zeiss Microscopy, White Plains, NY). 

For the safety study, eyes were incubated in a fixation solution (97% methanol, Sigma-Aldrich; 3% acetic acid, Thermo Fischer Scientific) at −80°C for 10 days.³⁵ Eyes were allowed to warm up to room temperature, and the cornea and lens were then gently removed using a razor blade to prevent tissue deformation during dehydration. Samples were then processed by sequential immersion in 100% ethanol (reagent grade alcohol, VWR, Radnor, PA; for 4 hours, repeated twice), xylene (VWR; for 4 hours, repeated twice), a first paraffin bath (4 hours), and a second paraffin bath (overnight). Finally, eyes were paraffin embedded, cut into 7-µm sections every 100 µm using a rotary microtome (HM 355 S, Thermo Fisher Scientific), and stained with hematoxylin and eosin via an autostainer (Autostainer XL, Leica, Buffalo Grove, IL). Sections were imaged using a brightfield microscope (Axiocam 506 color, Carl Zeiss Microscopy). Despite better tissue preservation using the paraffin embedding method, eyes from SCS delivery optimization and characterization studies were cryosectioned because organic solvents used in paraffin processing would dissolve the injected fluorescent particles, thus impairing our ability to confirm SCS delivery in tissue sections. 

Results

MN Injector Development

The key to achieving a successful SCS delivery is precise insertion of the MN across the entire scleral tissue, but not deeper.¹³ Conventional hypodermic needles fail to deliver this precision because the needle dimensions are substantially larger than the thickness of sclera, that is, just the needle orifice on a small gauge hypodermic needle can be many times larger than rodent scleral thickness (Fig. 2A-1). Even conventional hollow MNs developed for SCS injection, despite their microscopic size, are still disproportionately large when compared with rodent eye features (Fig. 2A-2). Thus, a new injector tailored for tiny rodent eyes is needed. 

Figure 2.

 

MN characteristics for injection into rodent eyes. (A) Development stages of high-precision MN injector. Illustration and representative images of (A-1) hypodermic needles currently used for SCS injection in rodents (design #1), (A-2) a conventional hollow metal MN developed for SCS injection in large animals (design #2), (A-3) an ultrasmall glass MN housed in a metal tube (1 mm OD) as the hub (design #3) and (A-4) an ultrasmall MN with improved hub design for better control of MN interaction with sclera (design #4). The lower aspect ratio of hub width to needle length (\(\frac{{\boldsymbol{\beta }}}{{\boldsymbol{\alpha }}}\)) enabled precise scleral penetration and increased SCS delivery success rate. Representative thickness of rat sclera is shown in some images for comparison. (B) Illustrations of MN interactions at the scleral tissue interface. The relatively flat and wide hub of Design #3 increased the likelihood of wide tissue deformation (B-1) and oblique insertion (B-2), resulting in incomplete MN penetration. (B-3) Design #4 reduced those issues by restricting tissue deformation to a small area localized at the injection site and enabling perpendicular insertion through better visualization of the MN-tissue interface during injection.

Ultrasmall Glass MN to Match Scleral Thickness

We scaled down the MN dimensions by fabricating MNs out of glass using techniques adapted from micropipettes used to inject material into individual cells (e.g., as used for in vitro fertilization).³⁶ Using a micropipette puller and beveler to grind the needle tip at a predetermined angle, we made MNs with lengths as short as 100 µm and bevel angles of 30° and 65°. In our initial studies, using design #3, the MN was housed in a hollow steel tube that served as a hub and was connected to a 10-µL Hamilton syringe for injection (Fig. 2A-3). 

First, we identified the appropriate MN length for SCS delivery in rats in vivo by varying MN length between 100 µm (similar to rat scleral thickness) and 200 µm. SCS delivery was determined via imaging injected fluorescent nanoparticles in the SCS by in situ fluorescence fundoscopy, followed by post mortem histology analysis for confirmation, if fundoscopy indicated success. 

With MN design #3, which involved a MN mounted on a wide hub, none of the injected eyes showed fluorescent signal in the fundus imaging, indicating a failed delivery (Supplementary Fig. S1A). In 31% of cases (11 of 36 trials), the fluid could not even be ejected out of the MN owing to strong resistance against flow (Supplementary Fig. S1B), which is a hallmark of intrascleral injection, as reported previously.^1,13 In those cases, the eyes showed a microscopic scleral hole at the injection site denoting partial scleral penetration consistent with our intrascleral delivery argument (Supplementary Fig. S1C). In other cases, fluid flowed out of the MN, but it leaked onto the scleral surface (Supplementary Fig. S1B), and inspection of the scleral tissue showed no evidence of MN penetration (Supplementary Fig. S1D). There was no apparent correlation between the MN length and MN penetration into sclera as assessed by fluid flow (Supplementary Fig. S1B). Even for the longest MNs (i.e., 200 µm), where the needle was twice the thickness of rat sclera, a scleral hole was found at the injection site in just one of the six attempts. These observations suggest that another factor, beside the MN length, plays a role in determining injection outcome. 

Needle Hub Design to Control MN Interaction With Sclera

We next hypothesized that the poor penetration of MNs into rat sclera may arise from unwanted interaction between the injection device and sclera at the tissue interface. We considered the following elements that may interact at the tissue interface: (1) the MN, (2) the needle hub, and (3) the sclera–conjunctiva tissue that has a curved surface and is elastic. Because we already worked to optimize the MN and the scleral tissue is difficult to change, we focused on needle hub design. We suspected that the high aspect ratio of the needle hub width (β) to the needle length (α) in design #3 (Fig. 2A-3) may give rise to a poorly controlled MN–tissue interaction that could impede MN penetration. 

In one failure mode associated with an excessively wide hub (Fig. 2B-1), we expect that scleral deformation can extend across a large area of tissue, confounded by the innate curvature of the ocular surface, and leave a gap between the MN and tissue surface at the site of injection. The resulting tissue buckling can lead to poor MN penetration. Another failure mode (Fig. 2B-2) concerns insertion of the MN at an oblique angle that could result in poor scleral perforation because the needle has to traverse a longer distance to reach the SCS compared with perpendicular insertion. In this case, a wide hub (\(\frac{\beta }{\alpha } \gg 1\)) makes it difficult to achieve perpendicular insertion locally where the MN contacts sclera, leading to unpredictable insertions at uncontrolled oblique angles. 

To overcome these issues, we designed and 3D-printed a new hub with low aspect ratio (\(\frac{\beta }{\alpha } \cong 1\)) according to design #4 (Fig. 2A-4, Fig. 3A). This design minimizes hub contact area on the ocular surface and restricts tissue buckling to a highly localized area immediately surrounding the injection site. In addition, it promotes perpendicular insertion naturally by providing better visualization of the MN contacting the ocular surface (Fig. 2B-3). We tested the performance of this new design in rat eyes with MNs ranging from 100 to 200 µm in length. These results showed MN puncture into sclera in 50% of eyes (18 of 36 trials) (Supplementary Fig. S2B) and successful SCS delivery in 28% of the injected eyes (10 of 36 trials), as determined by detecting fluorescence signal in the SCS by fundus and histology images (Supplementary Fig. S2A). Although much better than design #3 with the wide hub, the results with design #4 were still variable and showed no apparent dependence on MN length, which indicated that additional factors needed to be addressed (Supplementary Fig. S2). 

Figure 3.

 

Delivery technique developed for SCS injection in rats and guinea pigs. (A) An injector featuring an ultrasmall MN and a needle hub designed for precise penetration into sclera (based on design #4); magnified views provide the MN with greater resolution. (B) A custom-made 3D-printed vacuum probe used for eye stabilization during injection. (C) Schematic of MN with optimized geometry including needle length, bevel angle and tip OD. (D) Representative image showing injection of a blue dye, methylene blue (MB), and green-fluorescent nanoparticles in an albino guinea pig eye in vivo by MN using a vacuum probe as eye stabilizer. (E) Representative image of a rat eye in vivo immediately after injection showing the injection site and visible dye spread withing the SCS.

Eye Stabilization to Control the MN Interaction With Sclera

Having observed frequent eye movement during injections, we surmised that this motion may contribute to variable injection success. Given the submillimeter length scales of the injection process, even the slightest eye movement could cause misalignment at the MN–tissue interface (e.g., as shown in Fig. 2B), thereby decreasing the effectiveness of optimizing the MN and hub design. We, therefore, hypothesized that stabilizing the eye to prevent movement during injection would improve the success rate. 

To that end, we designed a 3D-printed stabilizer probe comprised of a vacuum port and an eye cup placed on the inferior corneosclera (Fig. 3B). The vacuum port was connected to a pump to generate a gentle vacuum to secure the eye. To lower the likelihood of corneal irritation, we (1) fabricated the device using a high-resolution 3D-printer to minimize surface roughness on the eye cup that could scratch the cornea, (2) designed the shape of the eye cup to minimize tissue strain and deformation by conforming to the corneal surface by mimicking the corneal curvature, and (3) limited the vacuum duration to less than 1 minute per eye, respectively. It is worth noting that we have used this technique in rodent eyes more than 50 times and have never observed acute or long-term adverse effects associated with vacuum stabilization. 

Next, we repeated the test of SCS injection in rats, this time combining eye stabilization with MN design #4 (Fig. 2A-4). Unlike previous trials, this combination generated consistent data. Injection using MNs with lengths of 100 µm and 120 µm exhibited high resistance without achieving SCS injection, indicating that the MN length was too short to cross the sclera fully (Supplementary Fig. S3). Increasing the length to 140 µm produced successful SCS delivery in four of the six trials, with high injection resistance encountered in the two failed cases. Longer needles (160 µm) enabled successful SCS delivery in all cases with no injection resistance. Although we did not try still longer needles, owing to concern of possible choroidal puncture, we concluded that the optimal length for SCS injection in the rat is 160 µm, which was enabled by tight control over the hub–tissue interface interaction and the stabilization of eye movement. 

MN Tip Geometry to Facilitate MN Insertion and Injection

The MNs used in the studies so far had a tip bevel of 45° and a tip OD of 120 µm. We hypothesized that these MN tip parameters could be optimized further, such that sharper tips could facilitate better MN insertion into the scleral tissue, but a longer MN tip could lead to injection leakage (i.e., as tip length approaches the scleral thickness) (Fig. 3C). 

To assess tip sharpness, we tested four bevel angles of 30°, 45°, 55°, and 65° and found that the sharpest bevel angle (30°) facilitated tissue penetration, but extraocular leakage was seen five out of six times (Supplementary Fig. S4). This leakage was likely due to the large tip opening associated with the steep bevel angle that spanned the entire scleral thickness, which enabled a fluid pathway to the scleral surface (Supplementary Fig. S4D-1). In contrast, we observed high resistance to injection in all cases when using MNs with the bluntest tips (65°), which was indicative of incomplete scleral penetration owing to insufficient sharpness (Supplementary Fig. S4D-3). The MNs with 45° bevel tips performed moderately well, with extraocular leakage seen in only two of the six cases (Supplementary Fig. S4). MNs with 55° bevel tips worked best of all, with all attempts achieving low-resistance injection into the SCS without leakage (Supplementary Fig. S4). 

Last, we assessed the effect of tip OD ranging from 90 to 150 µm (Fig. 3C). Independent of the tip OD, all MNs in this study penetrated scleral tissue well, owing to their optimal 55° bevel angle. However, an MN with a larger tip OD (130 and 150 µm) sometimes exhibited extraocular leakage, whereas the smaller tip OD (90 and 110 µm) reliably injected into the SCS without leaking (Supplementary Fig. S5). Although no MN tips ever broke in our hands, we selected the 110-µm tip OD as the optimal size because the 90-µm needles were more fragile (Table). 

Table.

 

Optimal MN Injector Parameters

 

Suprachoroidal Injection in Guinea Pigs

Guided by optimization in rats, we tested injection into the SCS of guinea pig eyes using MN with a 55° bevel angle and a 110-µm tip OD administered using hub design #4 with the aid of suction-based ocular stabilization. Because the guinea pig sclera is thicker than that in the rat (approximately 170 µm), we tested three MN lengths and found consistently high injection resistance when using 170- and 230-µm MNs, whereas the 260-µm MNs produced successful SCS injections without leakage in all four trials (Supplementary Fig. S6, Table). 

Characterization of Suprachoroidal Delivery by MN Injection

To further characterize SCS injection in rats and guinea pigs using optimized MNs, we imaged the fundus by brightfield and fluorescence imaging after injection of green fluorescent nanoparticles in vivo. Consistent with literature on SCS injection by MNs in larger animals,¹³ we observed radial and posterior flow of injected nanoparticles from the site of injection in the SCS (Fig. 4). SCS localization was confirmed in part because the choroidal vasculature could be seen shadowing the nanoparticle fluorescence, indicating that the nanoparticles were located behind the choroid. SCS localization was further confirmed by histological examination of the ocular tissues after injection. In both rat and guinea pig, the injected green fluorescent nanoparticles were found in a location between the sclera and the choroid (i.e., in the SCS) (Fig. 5). 

Figure 4.

 

Representative images of the fundus in rat and guinea pig after SCS injection by MN in vivo. Brightfield and fluorescence fundus images taken before and after SCS injection of a solution containing green fluorescent nanoparticles in rat (A) and guinea pig (B). Complete rendering of fluorescence spread in fundus of rat (C) and guinea pig (D) created by stitching together fundus images captured at different locations on the fundus. Eyes each received a 3 µL (rat) or 8 µL (guinea pig) injection of Hank's balanced salt solution containing 200 nm green fluorescent particles on the superior side of the eye using MNs with optimal parameters (Table). i, inferior; N, nasal; S, superior; T, temporal. Circle indicates the optic nerve. Arrow indicates the injection site.

Figure 5.

 

Representative histological tissue sections of rat and guinea pig eyes after SCS injection in vivo imaged by confocal microscopy. Images show retina stained blue with DAPI stain and injected nanoparticles with green fluorescence in rat (A) and guinea pig (B) eyes. Insets feature fluorescence and brightfield magnified sections of retina and neighboring tissues without (A-1) and with (A-2, B-1) injected nanoparticles. Insets are labeled with relevant anatomical structures. Eyes each received a 3 µL (rat) or 8 µL (guinea pig) injection of Hank's balanced salt solution containing 200 nm green fluorescent particles on the superior side of the eye using MNs with optimal parameters (Table).

We performed additional characterization using a larger cohort of animals, including 13 rats (10 albino Wistar rats and 3 pigmented Long Evans rats) and 3 albino guinea pigs. Fluorescence fundoscopy and fluorescence microscopy of histological tissue sections both confirmed successful SCS delivery in all injected eyes among the albino animals (Fig. 6, Supplementary Video S1 and S2). Brightfield images taken immediately after delivery seemed to be normal, without signs of hemorrhage or other adverse effects. This result was anticipated; the MNs were designed to be short enough to avoid puncturing choroid or retinal vessels that could cause bleeding. All of the injections had low injection resistance and no leakage onto the scleral surface, indicating complete MN penetration across the sclera. There was no evidence of fluorescent nanoparticles in the subretinal space or the vitreous. These results show reliable SCS administration in rats and guinea pigs with a 100% success rate and no acute safety signals associated with the technique. 

Figure 6.

 

Representative fundus and histological images after SCS injection into rat and guinea pig eyes. Injections were performed on 10 albino Wistar rats unilaterally (A) and 3 albino guinea pigs bilaterally (B) in vivo with brightfield (BF) fundus imaging before and immediately after injection, fluorescence (Fluor) fundus imaging immediately after injection and fluorescence imaging of histological tissue sections taken from enucleated eyes frozen within 30 minutes after injection. Eyes each received a 3 µL (rat) or 8 µL (guinea pig) injection of Hank's balanced salt solution containing 200 nm green fluorescent particles on the superior side of the eye using MNs with optimal parameters (Table). Green signal emitted by the injected nanoparticles confirms SCS delivery in all injected eyes. Histological sections were counterstained with DAPI stain (blue) to identify location of retina.

We similarly carried out SCS injection in the three Long Evans rats and achieved SCS injection in each case (Supplementary Fig. S7). Although the albino animals allowed visualization of injected nanoparticles in the SCS owing to the lack of pigmentation in the retinal pigment epithelium, the pigmented Long Evan rats enabled demonstration of SCS targeting by the lack of visualization of injected nanoparticles, hidden by the pigmented retinal pigment epithelium layer. Although histological tissue sections confirmed successful SCS administration (Supplementary Fig. S7B), fundus images had no fluorescent signal, consistent with targeted SCS delivery without chorioretinal puncture or intravitreal injection (Supplementary Fig. S7A). 

Circumferential Spread of Fluid in the SCS

SCS injection in albino animals revealed that a single injection of 3 µL of fluid containing nanoparticles immediately posterior to the superior limbus resulted in fluid spread covering most of the superior hemisphere and extending posteriorly toward the optic nerve. Interestingly, we noticed an apparent blockage of flow in the SCS across the long posterior ciliary artery (LPCA) into the inferior hemisphere, which limited the greater circumferential spread of nanoparticles within the SCS (Fig. 7). This finding is plausible, because the LPCA is a major artery located at the choroid–sclera interface that originates near the optic nerve head and runs anteriorly to the ciliary body. Similar observations in rabbit eyes were previously reported by Chiang et al.,³⁷ who concluded that the LPCA tightly adheres the choroid to the sclera, thereby physically blocking particles cross-over. 

Figure 7.

 

Representative fundus images fluid distribution in the SCS after injection of green-fluorescent nanoparticles 1 mm posterior to the superior limbus in rat eyes in vivo. (A) Brightfield and fluorescent fundus images display the effect of the LPCA on SCS spread, blocking nanoparticle cross-over into the inferior hemisphere. (B) Additional images showing fluid flow spreading in the superior SCS but blocked by the LPCA (dashed lines). Eyes each received a 3-µL injection of Hank's balanced salt solution containing 200 nm green fluorescent particles on the superior side of the eye using MNs with optimal parameters (Table). i, inferior; ON, optic nerve; S, superior. Data are representative of eyes from 5 rats.

Safety of Suprachoroidal Delivery by MN Injection

Although SCS injection has been well-tolerated in larger animals and humans,^13,34 we assessed the safety of SCS injection by MN in rats and guinea pigs using brightfield and fluorescence fundoscopy, clinical examinations, and post mortem histological analyses. 

Fundus Examination

Brightfield fundus imaging revealed no notable findings in all animals studied. No intraocular hemorrhage or other complications were seen in any of the animals throughout the week-long study. This outcome is expected for an injection targeted to the SCS without puncturing choroid or deeper tissues.¹³ 

Although nanoparticles are known to remain in the SCS,^13,38 fluorescent fundus imaging showed that injection of fluorescein was mostly cleared from the SCS within 1 hour after injection (Fig. 8), probably through choriocapillaris blood circulation. By 3 days, the fluorescein was completely cleared. These results are consistent with clearance measurements in rabbits, where total SCS collapse and significant fluorescein clearance was reported within 1 hour after injection.³⁸ 

Figure 8.

 

Representative fundus images from a week-long tolerability study after SCS injection in rat and guinea pig eyes in vivo. Brightfield and fluorescent fundus images collected before injection and at designated times after injection in the rat (A) and guinea pig (B). Eyes each received a 3 µL (rat) or 8 µL (guinea pig) injection of Hank's balanced salt solution containing fluorescein on the superior side of the eye using MNs with optimal parameters (Table). Data are representative of eyes from four rats and three guinea pigs.

Clinical Examinations

Visual clinical examinations of proptosed eyes of rats and guinea pigs, aided by magnified imaging with a cell phone camera, produced no notable findings over the course of 1 week after SCS injection (Supplementary Fig. S8). Results showed no conjunctival hemorrhage after the injection, and the microscopic sclerotomy created by MN insertion was self-resolved within 3 days after the procedure. Furthermore, no notable changes in external features such as hyperemia, irritation, swollen conjunctive or eyelids, corneal scratch, or scar were seen in any of the injected eyes for the duration of the study, indicating the good tolerability of the overall procedure, including the application of gentle vacuum to the cornea. 

Histological Analyses

At the end of the week-long study, eyes were enucleated and examined histologically. Inspection of tissue sections from the superior hemisphere near the injection site showed that the SCS had returned to its native, collapsed state,³⁸ as evidenced by apposition of the sclera to the choroid (Fig. 9). This finding is consistent with our observations from fluorescence fundoscopy data that showed rapid clearance of fluorescein, corresponding with SCS collapse. 

Figure 9.

 

Representative histological sections at the end of a week-long tolerability study after SCS injection in rat and guinea pig eyes in vivo. Hematoxylin and eosin–stained histological sections of rat (A) and guinea pig (B) eyes 1 week after injection. No structural abnormalities, perforations or intraocular hemorrhage were visible. Eyes each received a 3 µL (rat) or 8 µL (guinea pig) injection of Hank's balanced salt solution containing fluorescein on the superior side of the eye using MNs with optimal parameters (Table). Data are representative of eyes from four rats and three guinea pigs.

Histology showed no evidence of adverse effects in ocular tissues including choroidal puncture, hemorrhage, retinal perforation or disorganization, or optic nerve abnormalities near the injection site or at other locations throughout the eye (Fig. 9). 

Discussion

Unmet Need for Suprachoroidal Delivery in Rodents

Since Patel et al.¹ first introduced SCS delivery via MNs as a new route of administration to the eye, this technology has been studied widely in large animals (e.g., rabbits, pigs, and monkeys) and was recently approved for clinical use to deliver a corticosteroid, triamcinolone acetonide, suprachoroidally to treat uveitic macular edema.³⁹ Distinguishing features of SCS delivery include (1) posterior segment delivery via circumferential flow within the SCS toward the posterior pole, (2) widespread fundus coverage, (3) targeted chorioretinal delivery that minimizes off-target effects (notably compared with intravitreal injection), and (4) minimally invasive nature of the procedure. 

Research on SCS delivery in rodents has been underused mainly owing to the challenges of working with their extremely small eyes. This limitation excludes access to the breadth of rodent disease models available for preclinical development. For the lack of a better delivery method, researchers have resorted to using hypodermic needles, despite their disproportionately large size compared to rodents’ ocular tissues. This delivery method has several drawbacks, including the need for (1) specially trained personnel to perform injections, (2) an operating surgical or dissecting microscope to visualize the insertion process, and (3) a two-step process to make a scleral incision and then insert a blunt needle for injection. With this approach, there is also a higher risk of failure owing to possible hand tremor, eye movement, slightest error in insertion angle, and other stochastic factors. The lack of control over needle depth penetration increases the chances of chorioretinal perforation, choroidal hemorrhage, off-target delivery (e.g., intravitreal injection or subretinal injection), and other unintended outcomes. All these factors underscore a substantial need for a better SCS delivery method in rodents. 

High-Precision MN Injector Design and Fabrication

We hypothesized that the keys to reliable SCS administration in rodents are two-fold: (1) the use of a needle scaled to dimensions similar to that of rodent eyes, and (2) precise control over the depth of penetration into the ocular tissue. We achieved the former by fabricating MNs out of glass, which can be pulled and beveled into various shapes and sizes using techniques developed for intracellular injections and other micropipette applications. Although we found the glass MNs easy to make and handle, and have never had them break during use, alternatively ultrasmall MN could be fabricated out of other materials using other methods.^40,41 

We achieved a precise tissue penetration depth by controlling interactions at the interface between the injector device and the elastic scleral tissue. Optimization of this interaction was critical because the margin of error in rodent eyes is extremely small. In rats, for example, sclera is approximately 104 µm thick¹⁸ and the choroid is approximately 50 µm thick,^42,43 and in guinea pigs sclera is approximately 189 µm thick¹⁷ and the choroid is approximately 70 µm thick.⁴⁴ This means that a penetration error of just tens of microns can either cause chorioretinal perforation (by going too deep) or intrascleral delivery (by not penetrating deep enough). We, therefore, conducted a series of studies to optimize factors that determine penetration accuracy, including (1) MN features (length, tip bevel angle, tip length), (2) hub design, and (3) eye stabilization. 

MN Length

We found the optimal MN length for SCS delivery to be 160 and 260 µm for rats and guinea pigs, respectively. These lengths not only account for the total tissue thickness (sclera and conjunctiva), but also the additional effect of scleral deformation at the injection site. 

MN Tip Sharpness

We varied MN bevel angle between 30° and 65°, and found 55° to outperform the others, resulting in successful SCS delivery in all injected eyes. Blunter tips (65°) were too dull and did not penetrate the sclera completely. They also required greater force to insert, which could potentially increase risk of tissue damage. Sharper tip angles (30° and 45°) enabled easier insertion, but caused fluid leakage while injecting because their tip opening spanned the scleral tissue thickness. 

MN Tip OD

Another factor influencing insertion was the needle tip OD. The smaller the tip, the easier it was to insert the MN. Smaller tips minimize the needle track in the tissue, which could mean less tissue disruption, and hence better safety (smaller sclerotomy). Our goal was to find the smallest MN that yielded the highest success rate without clogging or leaking. Testing tip ODs ranging from 90 µm to 150 µm, we found the MNs with a 110-µm tip OD to be optimal. Tips of 90 µm OD performed similarly well, but were more fragile, so we chose the slightly larger needle tip to decrease possible needle breakage while injecting. 

Hub Design

The effect of needle hub design on penetration accuracy was manifested by its effect on the (1) extent of tissue deformation caused upon contacting the sclera, and (2) the MN insertion angle into the sclera. We defined a design parameter, hub width-to-needle length ratio (\(\frac{\beta }{\alpha })\), as a basis for hub design and compared two scenarios: (1) when the hub width was much bigger than needle length (\(\frac{\beta }{\alpha } \gg 1\)), and (2) when that ratio was close to unity (\(\frac{\beta }{\alpha }\sim 1\)). The former design had variable success owing to the resulting tissue deformation across large areas which may not be centered around the site of insertion, thus exposing sclera to a partial needle length. Additionally, the large hub's footprint made it difficult to know the insertion angle locally around the injection site, possibly resulting in uncontrolled oblique angle insertions that could prompt failed injections. In contrast, we had complete success with the \(\frac{\beta }{\alpha } \cong 1\) hub shape. Here, tissue deformation was limited to a small area immediately adjacent to the insertion site, and it provided better visualization to facilitate perpendicular MN penetration. 

Eye Stabilization

Eye movement was another confounding factor that could disrupt MN penetration dynamics. As such, we designed a probe that applied a gentle vacuum to the cornea, which secured the eye in position while injecting. This probe was fabricated using high-resolution 3D printing to minimize surface roughness that could scratch the cornea. 

Reliability and Targeted Delivery Analysis

Having incorporated several device design elements to achieve high-precision scleral insertion, we hypothesized that the optimized device would achieve a high success rate of SCS injection. Indeed, our study including 16 animals (19 injections) indicated 100% success rate in two rodent species, namely, rats and guinea pigs, in both albino (Wistar rat, Hartley guinea pig) and pigmented (Brown Norway rat) animals. We found no evidence of intravitreal injection, subconjunctival delivery, or fluid leakage out of the eye, denoting the targeted nature of the injection method. The procedure was straightforward to carry out, took approximately 1 minute to perform per eye, and did not require an operating microscope. We expect a minimal learning curve for others to master the technique; the design elements implemented into the device mitigate the risk of failure caused by human error or stochastic factors (hand tremor, eye movement, etc.), thus minimizing dependence on human skill or experience. 

Safety Analysis

Numerous studies in the literature have reported the safety and tolerability of SCS MN injections in both large animals and humans.^24,45 We, too, reached the same conclusion in our analyses in rodents. Clinically, all injected eyes seemed to be normal throughout the study, without signs of hyperemia, redness, irritation, swelling, corneal scar or opacity, cataract, or scleral hole. Fundus imaging revealed no intraocular hemorrhage or abnormalities. Similarly, post mortem histological analyses showed no notable changes at the chorioretinal level, consistent with the minimally invasive nature of this procedure. Future studies are needed to assess safety, including inflammatory responses, in greater detail. 

Injection Spread Within the SCS

The SCS is a potential space, meaning that fluid must expand it by separating the sclera and choroid to accommodate the injection flowrate. Injecting 3 µL of fluid with nanoparticles resulted in SCS expansion and spread across the majority of the superior hemisphere. Notably, however, nanoparticles did not cross into the inferior hemisphere in an area intersected by the LPCA, a large arterial vessel in the SCS. This finding suggests that the LPCA impedes flow, possibly by creating a stronger bond between the sclera and choroid, which makes it more difficult for fluid to expand the SCS. 

Impact and Future Work

Ophthalmology researchers often need to perform procedures that require a high degree of precision in execution given the delicate and complex structure of the eye. We addressed this challenge, in the case of rodent SCS delivery, by taking an engineering approach to control scleral penetration with precise spatial control. Our approach relied on (1) identifying unfavorable interactions at the scleral tissue interface as the root cause of the problem, (2) breaking down the interactions into individual components (MN features, hub design, deformable sclera, and eye movement), and (3) optimizing each component to find the best-performing configuration. 

We propose a technique that makes rodent SCS administration much more accessible, with no need for sophisticated equipment, highly specialized personnel, or detailed training. We hope that this work will increase the throughput of preclinical investigations, decrease the associated costs and complexity by decreasing the need to use large animals in early stages, and accelerate the discovery and development of new ocular therapies. We established proof of concept of this SCS delivery method by injecting fluorescent solutions, but various formulations including therapeutic agents, stem cells, viral vectors, controlled release systems, and more can be tested in future studies that can address pharmacokinetics, pharmacodynamics, and safety. 

A natural next step of great value would be SCS injection in mice. Given the commonalities, we believe that our design and optimization approach could be used in mice, but the still smaller size of the mouse eye compared with rats and guinea pigs will require even smaller MNs and length scales of control. We did not attempt SCS injection in mice in this study. 

Another area of interest with similar challenges is intravitreal injection in rodents. In this case, the issue of small eye size is further exacerbated by the large lens, which necessitates precise insertion angle without nicking the lens. We believe our technique could be modified to enable controlled oblique, instead of perpendicular, insertion to be suitable for intravitreal injection. 

Limitations

Although this study included dozens of animals in method development and feasibility studies, and 16 animals in the final device reliability testing, additional studies are needed not only in more animals, but also in more types of rodents (i.e., not just rats and guinea pigs), and in the hands of other investigators in other laboratories. Notably, mice were not addressed in this study, because they were beyond the scope of our particular research needs and represent a particularly challenging case because of their small size. 

We were not able to collect cross-sectional images of ocular tissue in live animals. Rodent eyes were too small for ultrasound imaging or optical coherence tomography using the instruments to which we had access. Such real-time data might contain additional information, such as the dynamics of SCS expansion and collapse, and other safety related data. Additionally, this work does not address immune response to the injection procedure (e.g., microglia activation and infiltration), which may reveal additional insight regarding the safety and tolerability of the procedure. 

We did not track intraocular pressure in this study. Previous work in rabbits revealed an immediate elevation followed by a transient decrease in the intraocular pressure after injection that eventually returned to baseline.^19,46 We expect, but do not know, if such a trend exists in rodents, which will require further investigation. 

Conclusions

This study developed a robust MN delivery technique for SCS injection in rats and guinea pigs by controlling and optimizing MN dimensions, tissue–hub interactions, and eye stabilization during injection. Targeted delivery was accomplished with high success rate in rats and guinea pigs using a simple, one-step, minimally invasive procedure that takes approximately 1 minute per injection (i.e., from proptosis to end of injection) and requires no surgical microscope. The presented technique can equip ophthalmology researchers with a tool to reliably deliver and test a variety of formulations such as stem cells, genetic cargos, antibodies, sustained release systems etc., thus amplifying the throughput of preclinical investigations and development of new ocular therapies that benefit from SCS administration. 

Acknowledgments

The authors thank Donna Bondy for administrative support, Richard K. Noel for veterinary expertise, Richard Shafer for microfabrication and 3D printing expertise, and Brandon G. Gerberich and Jae Hwan Jung for helpful discussions. 

Supported by National Institutes of Health grants (R01EY02209, R01EY028450, R01EY021592, P30EY006360, and T32EY07092). 

Disclosure: A. Hejri, None; I.I. Bowland, None; J.M. Nickerson, None; M.R. Prausnitz, Clearside Biomedical (O, P, R) 

Supplementary Material

Supplementary Video 1. 

Supplementary Video 2.