This article describes the development of a method to induce acute or chronic dry eye disease in rabbits by injecting concanavalin A to all portions of the orbital lacrimal gland system. This method, superior to those already reported, generates a reproducible, stable model of dry eye suitable for the study of pharmacological agents.
Dry eye disease (DED), a multifactorial inflammatory disease of the ocular surface, affects 1 in 6 humans worldwide with staggering implications for quality of life and health care costs. The lack of informative animal models that recapitulate its key features impedes the search for new therapeutic agents for DED. Available DED animal models have limited reproducibility and efficacy. A model is presented here in which DED is induced by injecting the mitogen concanavalin A (Con A) into the orbital lacrimal glands of rabbits. Innovative aspects of this model are the use of ultrasound (US) guidance to ensure optimal and reproducible injection of Con A into the inferior lacrimal gland; injection of Con A into all orbital lacrimal glands that limits compensatory production of tears; and use of periodic repeat injections of Con A that prolong the state of DED at will. DED and its response to test agents are monitored with a panel of parameters that assess tear production, the stability of the tear film, and the status of the corneal and conjunctival mucosa. They include tear osmolarity, tear break-up time, Schirmer's tear test, rose bengal staining, and tear lactoferrin levels. The induction of DED and the monitoring of its parameters are described in detail. This model is simple, robust, reproducible, and informative. This animal model is suitable for the study of tear physiology and of the pathophysiology of DED as well as for the assessment of the efficacy and safety of candidate agents for the treatment of DED.
Dry eye disease (DED) is a chronic condition with high prevalence and morbidity1,2,3,4. Inflammation plays a key role in its pathogenesis5,6. The pathophysiology of DED is conceptualized as deriving from either under-production or over-evaporation of tears; the former is also known as aqueous-deficient DED7. Sjögren's syndrome, an extensively studied prototypical cause of DED, affects primarily the lacrimal glands (LGs) and is a striking example of their importance in the pathogenesis of DED. DED is often treated with artificial tears which provide temporary relief, or with cyclosporine or lifitegrast, both of which suppress ocular inflammation. None of the available treatments for DED are optimal, necessitating the development of new agents8,9.
The search for new therapeutic agents for DED is hampered by three major challenges: the lack of a recognized druggable molecular target, which may be elusive given the pathophysiological complexity of DED; the sparsity of promising agents; and the lack of animal models that recapitulate key features of DED.
As with most drug development efforts, informative animal models of DED are a crucial investigative tool, notwithstanding the axiomatic statement that no animal model completely recapitulates a human disease. Mouse, rat, and rabbit models of DED are the most commonly used while dogs and primates are used infrequently10,11. Most of the more than 12 rabbit DED models reported to date attempt to reduce tear production by either removing LGs or by impeding their function12,13,14,15,16. Such approaches include the surgical resection of the ILG; closure of its excretory duct; and impairing LG function by irradiation or injection of one of the following: activated lymphocytes, mitogens, botulinum toxin, atropine, or benzalklonium. Major limitations of these methods are their inconsistency and the frequent partial suppression of tear production.
Concanavalin A (Con A), a lectin of plant origin, is a potent stimulator T-cell subsets and has been used in experimental models of hepatitis17 and DED18. The original Con A-based model was reported to offer significant advantages, including its relative simplicity; inflammatory cell influx in the LGs, mimicking diseases such as Sjogren's; stimulation of the proinflammatory cytokines IL-1β, IL-8, and TGF-β1; reduced tear function monitored by measuring tear fluorescein clearance and tear break-up time (TBUT); and drug responsiveness shown for an anti-inflammatory corticosteroid.
When this promising method was applied, in addition to its advantages, limitations were identified that necessitated its overall revision and drastic improvements. Three critical shortcomings of the method are documented. First, the model was an acute one; the induced DED subsided after about 1 week. Second, the response of the animals was inconsistent. As demonstrated, in "blind" transcutaneous injections to the Inferior LG (ILG), Con A was delivered only randomly to the targeted gland. Detailed study of the anatomy of the ILG revealed that its size could vary as much as 4-fold19 making such injections "hit-or-miss" efforts. Finally, even when the ILG was injected, the superior LG (SLG) frequently compensated for the reduced tear flow, making the model problematic.
These key limitations were overcome by introducing three modifications to the method, generating a superior animal model of DED. First, the injection of Con A into the ILG was performed under ultrasound (US) guidance, ensuring that Con A entered the gland. The success of the injection was confirmed by obtaining a post-injection US image, as shown in Figure 1. Second, to remove the compensatory tear contribution of the SLG, both the palpebral and orbital portions of this gland were injected with Con A. Finally, this acute model of DED was converted to a chronic one by repeated injections of Con A every 7-10 days. DED of 2 months' duration is readily achieved in these rabbits. The success of this approach has been amply documented19.
As already mentioned, an important application of animal models of DED is to determine the efficacy and safety of candidate therapeutic agents. The utility of this model was demonstrated by the study of phosphosulindac (OXT-328), a novel anti-inflammatory small molecule20,21 administered as eye drops. Its efficacy was demonstrated based on a panel of parameters of DED19. The relative simplicity and informative nature of this model also allowed side-by-side comparison of phosphosulindac to the two FDA approved drugs for DED, cyclosporine and lifitegrast, demonstrating its strong preclinical superiority.
All animal studies were approved by the Institutional Review Board of Stony Brook University and performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.
1. Animals and housing
2. Methods of anesthesia and euthanasia
NOTE: All procedures require mild sedation except for Con A injection that requires moderate sedation.
3. Removal of nictitating membrane
4. Measurement of dry eye parameters and collection of tear samples
NOTE: Measure DED parameters based on study protocol needs (e.g., at baseline and specified time points thereafter). Measurements for DED should be done in the following order, with rigorous effort to faithfully replicate them each time. Test all animals at approximately the same time of day (± 1 h) to minimize circadian variations. These measurements usually require a team of two investigators.
5. Induction and treatment of dry eye
NOTE: Three portions of the orbital lacrimal gland system are injected.
6. Post-procedure care
Con A injections induced a strong inflammatory response in the lacrimal glands characterized by a dense lymphocytic infiltrate (Figure 13), accompanied by decreased tear production. All tear parameters were markedly altered (Table 1 and Table 2). In addition, tear lactoferrin levels were suppressed (control = 3.1±0.45 vs. Con A injected = 2.7±0.02 ng/mg protein (mean ± SEM); p<0.03). The end result was a compromised corneal and conjunctival epithelium, evidenced by increased rose bengal staining (Figure 6).
Injection of the three orbital LG tissues produced a consistent and uniform DED state unlike the states achieved by previous methods18,27. Key contributors to this result were the US-guided injection of the ILG and the injection of the OSLG. Table 1 summarizes the salient results of this method. All changes are consistent with severe DED.
A single set of Con A injections produces DED lasting about 1 week; all clinical parameters normalize by day 10 (Table 2). Sequential Con A injections about 1 week apart extend the duration of DED accordingly. For example, the second set of Con A injections on day 7 maintains DED for 2 weeks and so on. After approximately 5 sets of injections, the DED state often becomes permanent without the need for further injections.
When the rabbits with Con A-induced DED were treated with the novel agent phosphosulindac, it markedly suppressed the disease. For example, following one week's treatment with this agent TBUT increased markedly compared to vehicle-treated animals (43.6±4.0 vs. 12.2±2.8 s; p<0.001; mean ± SEM respectively, for these and following values) while tear osmolarity was normalized (294±4.6 vs. 311±2.0 mOsm/L, p<0.002). Mechanistically, phosphosulindac decreased the levels of two crucial interleukins, IL-1β (8.4±1.2 vs 21.2±6.6 pg/mg protein; p<0.03) and IL-8 (4.9±1.7 vs. 13.5±5.0 pg/mg protein; p<0.05)19.
Figure 1: Ultrasound image of the inferior lacrimal gland. Upper Panel: The ILG as it moves deeper in orbit to lie beneath the zygomatic arch. The dashed line represents the line on skin across which the US probe is swept. Middle Panels: As the hand-piece is swept across this line, the examiner looks for loss of the zygomatic bone echo that is present in the left image (arrow) and disappears in the right. Lower Panels: Images of the ILG taken before (left) and after (right) injection of Con A. Development of a large cystic space within the gland confirms proper delivery. Reprinted with permission19. Please click here to view a larger version of this figure.
Figure 2: Gas mask sedation. This photograph shows the gas mask providing brief moderate sedation with isoflurane. Please click here to view a larger version of this figure.
Figure 3: Removal of the nictitating membrane. (A) Injection of lidocaine/epinephrine. (B) Truncation of the nictitating membrane at its base with Westcott scissors. (C) The ocular surface visualized more readily after removal of the nictitating membrane. Please click here to view a larger version of this figure.
Figure 4: Tear break-up time measurement. (A) Uniform green tear film appearance of the corneal surface under blue light immediately following application of fluorescein drops. (B) Corneal surface that has already undergone marked break-up evidenced by multiple dark circles and linear streaks in the fluorescein. Break-up time is recorded as soon as the first dark spot or line develops. The two light blue circles are reflections of the light source off of the cornea. Please click here to view a larger version of this figure.
Figure 5: Schirmer's tear test. (A) The proper placement of the Weck-Cel sponge in the lower fornix to remove any residual topical lidocaine solution and baseline tears. By placing the posterior edge of the triangular sponge under the lower lid margin, one can maintain a very uniform technique to dry the ocular surface prior to placement of the tear strips. (B) A tear strip appropriately placed at the mid-position of the lower lid between the globe and lower lid (palpebral conjunctiva). Please click here to view a larger version of this figure.
Figure 6: Rose bengal staining. Upper: Photographs of the corneal surface. Left: No rose bengal staining is present before treatment with Con A. Right: Diffuse corneal and conjunctival staining is seen in the upper nasal quadrant post-injection (upper right). Lower: Conjunctival impression cytology from the superior bulbar conjunctiva. Left: Numerous goblet cells are present before treatment. Right: Epithelial cells are present but goblet cells are absent post-treatment. Please click here to view a larger version of this figure.
Figure 7: Preparation of rabbit for concanavalin A injections. (A) Small shears are used to remove fur, allowing easier visualization of landmarks to identify the orbital superior lacrimal gland. (B) Nair is used to remove hair that remains after shearing. Please click here to view a larger version of this figure.
Figure 8: Injection of the palpebral lacrimal gland. (A) The palpebral lacrimal gland, appearing as a bulbous elevation in the posterior temporal portion of the upper lid. Tears are seen streaming from the surface of this gland after applying a drop of 2% fluorescein. (B) The palpebral lacrimal gland is being injected while the rabbit is receiving moderate sedation. One investigator retracts the eyelid, optimizes exposure of the gland, and secures the mask while the second investigator injects the gland. Please click here to view a larger version of this figure.
Figure 9: Localization of the orbital superior lacrimal gland. Changes in skin contours indicate the location of the OSLG as it protrudes through the posterior incisure. Alternating medial pressure on the globe (large arrow) causes the superior orbital gland to prolapse, which is seen as a small elevation in the skin. This elevation will increase in size each time the pressure is applied (small arrows). The location of this gland is usually in line with the posterior orbital rim. Please click here to view a larger version of this figure.
Figure 10: Injection of the orbital superior lacrimal gland. (A) Application of gentle pressure to the skull with fine-toothed forceps in the area which prolapsed as in Figure 9. A thin slit-like opening in the skull can be palpated. Leaving a small indentation mark with the forceps greatly aids placement of the needle during injection. (B) The needle is being inserted perpendicularly through the incisure. If placed incorrectly, its passage is stopped by the bony skull. (C) The needle is in final position angled towards the lateral canthus. Please click here to view a larger version of this figure.
Figure 11: Localization of the inferior lacrimal gland. (A) The prominence of the superficial portion of the ILG seen through the lower lid. The curvilinear pen mark denotes the lower position of the gland. The vertical line, under the nasal limbus, denotes the approximate position where the ILG transitions to a deeper position within the orbit and serves as a visual reference for the US. (B) US hand-piece sweeping across the area of the vertical line; the US monitor will show where the zygomatic bone ends, where the ILG transitions and where the Con A injection should be given ("injection site"). Please click here to view a larger version of this figure.
Figure 12: Injection of the inferior lacrimal gland. Injection of the ILG is done at the location identified by US. The depth of injection is calculated as described in the text (step 5.8.6). Calipers (seen behind the needle) ensure that the needle is placed at the proper depth before injection. Please click here to view a larger version of this figure.
Figure 13: Histology of the lacrimal glands. Tissue sections of a normal inferior lacrimal gland with typical tubulo-alveolar structure (A) and following injection of Con A (B), showing marked lymphocytic infiltrate with effacement of structure. Similar inflammatory infiltrates are seen in the superior lacrimal glands. Please click here to view a larger version of this figure.
Injection Method | TBUT, sec | Tear Osmolality, Osm/L | STT, mm | |||
Baseline | Post-injection | Baseline | Post-injection | Baseline | Post-injection | |
mean ± SEM, % change | ||||||
ILG without US guidance | 58.5 ± 1.5 | 44.5 ± 7.7 | 297 ± 4.9 | 300 ± 3.8 | 15.2 ± 0.9 | 12.9 ± 2.2 |
%change | -23% | 1% | -15% | |||
p=0.05 | p=0.36 | p=0.21 | ||||
US-guided ILG + PSLG | 59.4 ± 0.6 | 11.4 ± 4.2 | 296 ± 4.7 | 326 ± 3.7 | 15.1 ± 1.3 | 10.7 ± 1.8 |
% change | -81% | 10% | -29% | |||
p<0.0001 | p<0.0001 | p<0.03 | ||||
US-guided ILG + PSLG + OSLG | 60 ± 0.0 | 6.2 ± 1.3 | 299 ± 2.9 | 309 ± 2.8 | 14.6 ± 0.9 | 9.9 ± 1.3 |
% change | -90% | 3% | -32% | |||
p<0.0001 | p<0.008 | p<0.002 | ||||
All values were measured on day 6 following a single injection of Con A into the glands listed. All comparisons were made to baseline. *ILG, inferior lacrimal gland; PSLG, palpebral portion of superior lacrimal gland; OSLG, orbital portion of superior lacrimal gland; US, ultrasound. |
Table 1: Effect of injection technique and number of injection sites on dry eye severity.
Baseline | Day 6 | Day 13 | Day 21 | |
1 injection | 2 injections | 3 injections | ||
TBUT, sec | 58.5 ± 1.5 | 44.5 ± 7.7 | 29.2 ± 7.8 | 12.8 ± 3.9 |
% change | -24% | -50% | -78% | |
p=0.17 | p=0.001 | p<0.0001 | ||
TOsm, Osm/L | 297 ± 4.9 | 300 ± 3.9 | 308 ± 4.9 | 313 ± 2.7 |
% change | 1% | 4% | 5% | |
p=0.36 | p=0.04 | p=0.003 | ||
STT, mm | 15.2 ± 0.9 | 9.3 ± 1.6 | 12.9 ± 1.6 | 7.4 ± 1.1 |
% change | -39% | -15% | -51% | |
p=0.17 | p=0.13 | p=0.008 | ||
Comparisons were made to baseline. |
Table 2: Effect of repeated Con A injections into ILG on duration of DED.
Rabbits are highly attractive for the study of DED. Their cornea and conjunctiva have a surface area closer to that of humans compared to mice and rats; their complement of drug metabolizing enzymes such as esterases, and histology of their lacrimal glands are similar to those of humans, and their eyes are large enough for informative pharmacokinetic studies. Compared to pigs and monkeys, with which they share similar features, they cost less and their experimental manipulation is easier. If mechanistic studies are contemplated, a relative drawback of the rabbit, compared to mice, is that fewer reagents (e.g., monoclonal antibodies) are available. On the other hand, the rabbit is far superior to mice for pharmacokinetic and biodistribution studies because individual tissues are easily dissected and of sufficient size for analytical work, avoiding "sample pooling."
A critical general parameter is the acclimation period of the rabbits. The animals are shipped from the vendor under conditions that often do not ensure a transportation environment of the appropriate temperature or humidity. Some animals may have already developed dry eye upon arrival. A two-week period of acclimation is recommended. Equally important is scrupulous attention to the humidity and temperature of the space where the study rabbits are housed in the vivarium. Deviations in either condition can induce huge variations in their eye status. Have back-up humidifiers and dehumidifiers on hand. If the central system fails, act quickly to restore ambient humidity using the back-up equipment. Bear in mind that such unfortunate developments are more common in the summer months. The three most critical steps, however, for successfully inducing DED in rabbits are: 1) the skillful use of US imaging to identify the ILG and to direct and confirm injection of Con A; 2) ensuring injection of both the ILG and the two parts of the SLG; and 3) reliably and reproducibly assaying the parameters of DED.
Developing the required experimental skill is not trivial but should not deter any serious investigator. Expect the learning curve to be completed within five iterations. An US imaging system of reasonable quality is essential. Recognition of anatomical hallmarks by US is important, therefore, the investigator should review the rabbit anatomy. The excellent description of rabbit anatomy by Davis25, a classic, can be immensely helpful. Also keep in mind the variation in the size of the ILG. The corollary to this is that the success of Con A must always be confirmed with follow up imaging. Variations in the response to Con A in a group of rabbits is most often due to the injection technique (unsuccessful or partially successful injection) or to ignoring the capacity of residual lacrimal gland tissues to compensate with overproduction of tears. For those who wish to master the injection technique, injecting methylene blue followed by prompt anatomic dissection can be helpful; visualization is achieved if it reaches the lacrimal gland or spills onto neighboring tissues. To date, this injection method has been performed over 270 times by the authors without a single complication.
Assaying the five parameters of DED presented above can be as tricky as is their determination in clinical practice. Although circadian variations have not yet been formally documented in any of them, there is enough background evidence of such phenomena in the eye28 that they should be assayed at the same time of day (± 1 h), especially when repeat assays are to be performed and compared to each other. Consistency in performing these assays is essential. A team of two is required. Four or more investigators in the same room participating in the assays can be disruptive, given that some steps require strict timing. Appropriate and high-quality photographic documentation, where indicated, is important.
This model is ideally suited for drug development studies. Mastery of the animal model and assay techniques ensured excellent reproducibility19 of efficacy and safety studies.
This is a powerful experimental approach because it eliminates the confounding variability of prior models, has streamlined the animal model and essentially standardized assaying the five parameters of DED. The successful application of this model to the study of a candidate therapeutic agent has affirmed its practical utility as an informative animal model for a disease in desperate need of novel agents and of a deeper understanding of its pathogenesis.
The authors have nothing to disclose.
All animal studies were completed in accordance and compliance with all relevant regulatory and institutional guidelines. All studies were approved by the Institutional Review Board of Stony Brook University and performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.
These studies were supported in part by a Targeted Research Opportunities grant from the Stony Brook University School of Medicine (Grant Number 1149271-1-82502) and a research grant from Medicon Pharmaceuticals, Inc., Setauket, NY. The authors thank Michele McTernan for editorial support.
100 mm macro lens | Canon EF 100mm f/2.8L IS USM | 3554B002 | |
26 gauge needles (5/8) | Becton Dickinson and Company, Franklin Lakes, NJ | 305115 | Needles for injecting ConA into the lacrimal glands |
27 gauge needles (5/8) | Becton Dickinson and Company, Franklin Lakes, NJ | 305921 | Needles for injecting ConA into the lacrimal glands |
Aceproinj (acepromazine) | Henry Schein Animal Health, Dublin, OH | NDC11695-0079-8 | 0.1ml/kg subcutaneously injection for rabbit sedation |
Anesthesia vaporizer | VetEquip, Pleasanton, CA | Item #911103 | |
Bishop Harmon Forceps | Bausch and Lomb (Storz), Bridgewater, NJ | E1500-C | Tissue forceps |
Caliper | Bausch and Lomb (Storz), Bridgewater, NJ | E-2404 | Caliper used to measure length of needle during ConA injection |
Concanavalin A | Sigma, St. Louis, MO | C2010 | Make 5mg/ml in PBS for injection into rabbit lacrimal glands |
DSLR camera | Canon EOS 7D DSLR | 3814B004 | Digital single lens reflex camera |
fluorescein | AKRON, Lake Forest, IL | NDC17478-253 | Dilute to 0.2% with PBS to measure TBUT |
Isoflurane | Henry Schein, Melville, NY | 29405 | |
Lactoferrin ELISA kit | MyBiosource, San Diego, CA | MBS032049 | Measure tear lactoferrin level |
lidocaine | Sigma, St. Louis, MO | L5647 | 1% in PBS for anesthesia agent |
macro/ring flash | Canon Macro Ring Lite MR-14EXII | 9389B002AA | |
Osmolarity tips | TearLab Corp., San Diego, CA | #100003 REV R | Measure tear osmolarity |
PBS (phosphate buffered saline) | Mediatech, Inc. Manassas, VA | 21-031-CV | |
Rabbit, New Zealand White or Dutch Belted (as described in text) | Charles River Labs, Waltham, MA | 2-3 kg | Research animals |
Rose Bengal | Amcon Laboratories Inc., St. Louis, MO | NDC51801-004-40 | 1% in PBS, stain the ocular surface |
Schirmer strips | Eaglevision, Katena products. Denville, NJ | AX13613 | Measure tear production |
Surgical Loupes +1.50 | Designs for Vision, Bohemia, NY | Specialty item | Provide magnificantion of ocular surface while observing tear break up and performing Concanavalin A injections. |
TearLab Osmometer | TearLab Corp., San Diego, CA | Model #200000W REV A | Measure tear osmolarity |
Ultrasound probe | VisualSonics Toronto, Ont | MX 550 S | Untrasonography-guide Con A injection for inferior lacrimal gland |