Schlemm’s canal-selective Tie2/TEK knockdown induces sustained ocular hypertension in adult mice

Abstract

Deficient Angiopoietin-Tie2 signaling is linked to ocular hypertension in glaucoma. Receptor Tie2/TEK expression and signaling at Schlemm’s canal (SC) is indispensable for canal integrity and homeostatic regulation of aqueous humor outflow (AHO) and intraocular pressure (IOP), as validated by conditional deletion of Tie2, its ligands (Angpt1, Angpt2 and Angpt3/4) or regulators (Tie1 and PTPRB/VE-PTP). However, these Tie2/TEK knockouts and conditional knockouts are global or endothelial, preventing separation of systemic and ocular vascular defects that impact retinal or renal integrity. To develop a more targeted model of ocular hypertension induced by selective knockdown of Tie2/TEK expressed in SC, we combined the use of viral vectors to target the canal, and two distinct gene-editing strategies to disrupt the Tie2 gene. Adeno-associated virus (AAV2) is known to transduce rodent SC when delivered into the anterior chamber by intracameral injection. First, delivery of Cre recombinase via AAV2.Cre into R26 ^tdTomato/+ reporter mice confirmed preferential and stable transduction in SC endothelium. Next, to disrupt Tie2 expression in SC, we injected AAV2.Cre into homozygous floxed Tie2 (Tie2^FL/FL) mice. This led to attenuated Tie2 protein expression along the SC inner wall, decreased SC area and reduced trabecular meshwork (TM) cellularity. Functionally, IOP was significantly and steadily elevated, whereas AHO facility was reduced. In contrast, hemizygous Tie2^FL/+ mice responded to AAV2.Cre with inconsistent and low IOP elevation, corroborating the dose-dependency of ocular hypertension on Tie2 expression/activation. In a second model using CRISPR/SaCas9 genome editing, wild-type C57BL/6J mice injected with AAV2.saCas9-sgTie2 showed similar selective SC transduction and comparable IOP elevation in course and magnitude to that induced by AAV2.Cre in Tie2^FL/FL mice. Together, our findings, demonstrate that selective Tie2 knockdown in SC is a targeted strategy that reliably induces chronic ocular hypertension and reproduces glaucomatous damage to the conventional outflow pathway, providing novel models of SC-Tie2 signaling loss valuable for preclinical studies.

Graphical Abstract

1. Introduction

Glaucoma impairs vision by neurodegeneration of retinal ganglion cells (Boland and Quigley, 2007; Libby et al., 2005; Quigley, 1999). Among multiple triggering events, chronic ocular hypertension conspires with genetics and advanced age to progressive vision deterioration (Alqawlaq et al., 2019). In adult-onset primary open-angle glaucoma (POAG), the most prevalent form of glaucoma, vision loss can be decelerated by effective intraocular pressure (IOP) lowering, which is mainly caused by dysregulated aqueous humor dynamics (Weinreb et al., 2014). Aqueous humor is constantly produced by the ciliary body, flows across the anterior segment chambers, and exits the eye (outflows) through two main passages: the trabecular meshwork (TM) into Schlemm’s canal (SC) and systemic ocular veins (conventional outflow pathway), and to a lesser extent across ciliary muscle, iris and sclera (uveoscleral outflow pathway) (Alm and Nilsson, 2009; Costagliola et al., 2020; Stamer et al., 2015; Tamm, 2009). Aqueous humor outflow (AHO) is significantly regulated at the TM-SC interface, where TM juxtacanalicular cells (non-endothelial) and endothelial SC cells can pathologically increase AHO resistance and thus elevate IOP (Abu-Hassan et al., 2014; Keller and Acott, 2013; Overby et al., 2009; Overby et al., 2014b; Stamer et al., 2015; Stamer and Clark, 2017; Tamm, 2009). However, the precise molecular mechanisms underlying the increased resistance to AHO are incompletely understood, and current treatments for AHO control are limited.

SC, a ring-shaped duct interposed between the TM and limbal vessels at the iridocorneal angle, regulates the transfer of aqueous humor and immune cells into systemic veins (Alvarado et al., 2010; Streilein, 2003). Its flat lumen is delimited by a lymphatic-like endothelium (inner wall) contiguous with the juxtacanalicular TM, and by a venous endothelium (outer wall) that extends collector channels (CCs) that connect with aqueous veins, which converge into systemic episcleral veins (EVs) at the corneal limbus (Aspelund et al., 2014; Kizhatil et al., 2014; Park et al., 2014; Patel et al., 2020; Ramos et al., 2007; van der Merwe and Kidson, 2014; van Zyl et al., 2020). Key molecular regulators of AHO and IOP homeostasis are expressed in SC of mice and humans, including the transcription factor Prox1 (Kim et al., 2017; Park et al., 2014), receptors VEGFR2/3 (Aspelund et al., 2014; Kizhatil et al., 2014), and receptors Tie1/2 (Du et al., 2022; Kim et al., 2017; Rajasundaram et al., 2023; Souma et al., 2016; Thomson et al., 2014; Thomson et al., 2017). Endothelial receptor tyrosine kinase Tie2 (tunica interna endothelial cell kinase, also known as TEK) is at the vanguard of research on SC signaling pathways as new drug target for AHO restoration, given its involvement in SC adult maintenance (Aspelund et al., 2014; Bernier-Latmani and Petrova, 2017; Thomson et al., 2021).

Tie2 is robustly expressed by SC of mice and humans (Kim et al., 2017; Patel et al., 2020; Thomson et al., 2014; van Zyl et al., 2020), with highest levels in SC-inner wall endothelial cells (Kim et al., 2017; Kizhatil et al., 2014). Tie2 is activated in SC by paracrine angiopoietin (Angpt) ligands, with Angpt1 as primary agonist and Angpt2/Angpt4 (or ortholog Angpt3 in mice) as cooperative agonists, as well as by the receptor Tie1 also expressed at SC (Du et al., 2022; van Zyl et al., 2020). SC is directly and constantly exposed to all three angiopoietins that are secreted by the TM, with Angpt2 and Angpt4 also secreted by corneal endothelium and cornea/sclera stroma, respectively (Kapiainen et al., 2022; Patel et al., 2020; Saharinen et al., 2017; van Zyl et al., 2020).

Defective Angpt-Tie2 signaling has been linked with SC deterioration and ocular hypertension in the context of human aging and glaucoma (Du et al., 2022; Kapiainen et al., 2022; Kim et al., 2017; Rajasundaram et al., 2023; Saharinen et al., 2017; Souma et al., 2016; Thackaberry et al., 2019; Thomson et al., 2020; Thomson et al., 2014; Thomson et al., 2017; Ujiie et al., 2023). Genetic studies in patients with primary congenital glaucoma (PCG) and potentially POAG have identified heterozygous loss-of-function mutations affecting Tie2 (Souma et al., 2016; Young et al., 2020) and Angpt1 (Choquet et al., 2018; Thomson et al., 2017), and linked risk loci in Tie2, Angpt1 and Angpt2 to ocular hypertension and POAG (Gao et al., 2018; Gharahkhani et al., 2021; Khawaja et al., 2018; MacGregor et al., 2018). Further, mouse genetics studies confirmed the dose-dependency of SC stability and function on Tie2 expression and activation, as both Tie2 heterozygosity and Angpt1 or Angpt2 deletion result in moderate SC defects (Kapiainen et al., 2022; Kim et al., 2017; Souma et al., 2016; Thomson et al., 2021; Thomson et al., 2017), while postnatal deletion of Tie2, Angpt1/Angpt2 or Angpt2/Angpt4 lead to SC regression and chronic IOP elevation resembling PCG (Kapiainen et al., 2022; Kim et al., 2017; Souma et al., 2016; Thomson et al., 2014; Thomson et al., 2017). Of clinical relevance, inhibition of Tie2 activation in SC prompts ocular hypertension in adult mice and nonhuman primates (Kim et al., 2017; Thackaberry et al., 2019), whereas activation of Tie2 supports recovery of SC integrity, AHO and IOP homeostasis (Kim et al., 2017; Li et al., 2020; Qiao et al., 2022; Thomson et al., 2019; Thomson et al., 2021). Collectively, this work established essential roles for the Angpt/Tie2 system in SC integrity and function as regulator of AHO and IOP.

The mouse models above relied on global or endothelial deletion of Tie2 or Angpts (Kapiainen et al., 2022; Kim et al., 2017; Souma et al., 2016; Thomson et al., 2020; Thomson et al., 2014; Thomson et al., 2021; Thomson et al., 2017), including inducible models designed to bypass the early embryonic lethality associated with the targeted genes which have critical roles in systemic and ocular endothelial development, stability and homeostasis (Augustin et al., 2009). Unfortunately, these highly valuable models manifested renal or retinal defects (Kenig-Kozlovsky et al., 2018; Kim et al., 2019).

Addressing these issues, here we developed targeted genetic approaches to disrupt Tie2 expression in the SC of adult mice. We provide detailed evidence that intracameral delivery of an AAV2 vector encoding Cre recombinase into homozygote floxed Tie2 mice (Tie2^FL/FL) induces selective Tie2 knockdown in SC-inner wall, which results in dysregulated aqueous outflow and sustained ocular hypertension. Furthermore, we demonstrate that AAV2-mediated delivery of CRISPR/SaCas9 targeting Tie2 (AAV2.SaCas9-sgTie) into wild-type C57BL/6J mice results in IOP elevation with comparable magnitude and timecourse. By driving selective Tie2 knockdown in the mouse SC we show that Tie2 levels at its inner wall regulate adult AHO and IOP homeostasis, potentially via pathogenic mechanisms related to human ocular hypertension. Thus, we provide novel genetic strategies to induce targeted Tie2 knockdown in SC of adult mice, and propose them as reproducible and reliable models for research on ocular hypertension pathogenesis and SC-targeted drug design.

2. Materials and methods

2.1. Mice

Mice were housed in an AAALAC-accredited animal facility at the University of Utah. Adult (1.5 to 2 months of age) balanced numbers of female and male littermates were assigned to experimental and control groups. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) and performed in strict accordance with the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research and the National Institutes of Health (NIH) Guidelines for the Care and Use of Laboratory Animals. All used mouse strains had a C57BL6/J (B6) background. Founder Tie2^FL/FL mice were kindly shared by Dr. Hellmut Augustin (Heidelberg University, Germany), bred in house by crossing to wild-type B6 mice (#000664, Jackson Laboratories, Bar Harbor, Maine, USA), and genotyped using published primers: TEKI3 5’-CAGGCTATCACTGTGACACTGGTAC and SDL2 5’-AAATACGCAGTTTCAG GGCTGGGA (Savant et al., 2015). Ai14 R26 ^tdTomato reporter mice (R26 ^tdTomato/+ for short of B6.Cg-Gt(ROSA)26 Sortm14(CAG-tdTomato)Hze/J) were a gift from Dr. Mario Capecchi (University of Utah) and purchased (007914, Jackson Laboratories).

2.2. AAV2 vectors

AAVs were recombinant, self-complementary, serotype 2 (scAAV2, referred to as AAV2), produced by Dr. William Hauswirth and collaborators (University of Florida, Ocular Gene Therapy Core). The AAV2 vector was a triple tyrosine mutant capsid variant (Y444+500+730F), with a truncated chimeric cytomegalovirus-chicken β-actin (CMV-smCBA) promoter (Bogner et al., 2015) driving expression of Cre recombinase or GFP, or empty, and with a CMV promoter driving expression of sgTie2/saCas9 or saCas9. AAV2s were titered by real-time PCR, assessed for endotoxin load and purified by silver-stained SDS-PAGE as originally described (Bogner et al., 2015). AAV stocks were stored at −80°C and discarded after two thaw-freeze cycles.

2.3. AAV2.CRISPR/SaCas9 design and in vitro target validation

To disrupt the Tie2/TEK gene, we utilized the University of Utah-Mutation Generation and Detection Core Facility and used CRISPOR (Concordet and Haeussler, 2018) to identify candidate saCas9 single-guide RNA (sgRNA) sequences (Suppl. Fig. 6A). The DNA encoding the sgRNA sequence targeting exon 1 was cloned into the single vector AAV-CRISPR/saCas9 system pX601 (61591, Addgene, Watertown, MA, USA) (Ran et al., 2015). To validate the effectiveness of this sgRNA at reducing Tie2 protein expression, CRISPR/saCas9 plasmids containing the candidate sgRNA sequence was transfected into HEK293T cells using the GenJet DNA reagent (SL100488, SignaGen Laboratories, Frederick, MD, USA) according to manufacturer’s instructions. Cells were cotransfected with 0.5 μg mouse Tie2 tagged with TurboGFP (Tie2-tGFP) plasmid (MG225327, OriGene Technologies, Rockville, MD, USA) and either with Tie2-knockdown CRISPR/saCas9 plasmid (0.5 μg), or with ctrTie2-CRISPR/saCas9 (0.5 μg) lacking the sgRNA sequence as control (Fig. 6B). Transfection efficiency was determined 48 hr posttransfection by immunostaining for HA to detect saCas9–3xHA expression. Cells were fixed with 4% PFA, incubated in blocking buffer (0.2% Triton-X, 10% BSA, 10% normal donkey serum in 0.01M PBS) for 1 hr at RT, and in primary antibodies, including rat anti HA (1:500; 11867423001, Millipore Sigma, St. Louis, MO, USA) and mouse anti-TurboGFP (1:100, TA150041S, OriGene Technologies) overnight at 4°C in 0.2% Triton-X, 5% BSA in 0.01M PBS. Coverslips were then washed three times with PBS and incubated with secondary antibodies (5% BSA in 0.01M PBS) for 2 hr at RT, proceeded by washes, Hoechst 33342 staining (1:1,000, H3570 Molecular Probes, Eugene OR, USA), rinses and mounting with Fluoroshield medium (F6057, Sigma-Aldrich). Cells were imaged using an inverted confocal microscope (see below). For HA and tGFP+ cell counts, three regions were analyzed and averaged (n=3). For in vivo delivery and validation, the saCas9-sgTie2 construct was packaged into AAV2(triple) and high-titer virus generated (5.97×10¹² vg/mL), as described above (2.2.) and previously (Bogner et al., 2015). To assess the efficiency of Tie2 knockdown in vivo, AAV2.saCas9-sgTie2 was delivered by intracameral injection into B6 mice, and at 3 wpi, we assessed the extent of saCas9–3XHA expression and the relative levels of Tie2 in SC by immunostaining.

Fig. 6. CRISPR-saCas9-mediated disruption via intracameral AAV2.sgTie2/saCas9 increases IOP in wild-type B6 mice.

(A) Schematic of the plasmid expressing both SaCas9-HA and sgRNA. (B) Experimental design of in vitro validation of Tie2-E1 plasmid, and in vivo testing of AAV2.sgTie2/saCas9 system and control AAV2.saCas9 in wild-type adult mice. ITR, inverted terminal repeats. (C) Confocal image of the SC from a corneolimbal wholemount coimmunostained for HA and CD31, representative of 3 wpi of AAV2.sgTie2/saCas9 (1×10¹⁰ vg/eye). Abundant cells expressing HA (green) overlay the SC endothelium, with many coexpressing CD31 (magenta), as evidenced by the YZ perspective views. (D) Coimmunostaining Tie2 (magenta) further corroborates colocalization of HA (green) to Tie2+ SC cells. (E) Mean IOPs show significant elevation at 1 wpi of unilateral AAV2.sgTie2/saCas9 (1×10¹⁰ vg/eye) (****P<0.0001, n=15 eyes) relative to both baseline and uninjected/contralateral eyes (uninj. L). Increased IOPs persisted with the same significance from 1 to 4 wpi relative to baseline and control littermates injected with unilateral AAV2.saCas9 (****P<0.0001; n=15). (F) Comparison of mean IOPs 1 to 4 wpi of unilateral AAV2.Cre or AAV2.GFP into Tie2^FL/FL mice versus unilateral AAV2.sgTie2/saCas9 or AAV2.saCas9 in wild-type mice (all AAVs 1×10¹⁰ vg/eye), demonstrate comparable patterns of IOP elevation, with identical significance and persistence.

2.4. Intracameral injection

Mouse anterior chambers were injected in mice 1.5 to 2 months of age with 2 μL of AAV2 or basal saline solution (BSS Plus; Alcon, Fort Worth, TX, USA) followed by 2 μL air to prevent leakage using a method previously described (Bogner et al., 2015). General anesthesia was induced by intraperitoneal (IP) injection of Avertin (1.3% 2,2,2-tribromoethanol and 0.8% tert-amyl alcohol, Sigma-Aldrich), ocular anesthesia by 0.5% tetracaine hydrochloride eyedrops (Alcon), and mydriasis by 1% tropicamide eyedrops (Somerset Therapeutics, Hollywood, FL, USA). Injection was performed via tri-beveled 36G needles and 5 μL Hamilton syringe (Nanofil system; World Precision Instruments, Sarasota, FL, USA) mounted on a micromanipulator. The procedure did not induce ocular bleeding, but leakage was observed in some eyes that were included in the study for not having discernible differences than those without leakage. A few eyes with lens damage or opacity during injection were excluded.

2.5. Intraocular pressure (IOP) measurement

IOP changes were tracked in the late morning using rebound tonometer (iCare Tonolab, Vantaa, Finland) by an operator masked to experimental group, as previously described (Bosco et al., 2018). Briefly, 6 sequential IOP readings per eye were recorded within 3 min of starting mouse sedation by isoflurane inhalation (2% in 2 L/min O₂) to avoid reported hypotensive effects (Ding et al., 2011). IOP was tracked in the same individual eyes 2- and 1-week preinjection as baseline (BL), then weekly from 1 to 4 or 8 wpi, and at 12 wpi. Data were reported as weekly mean IOP for left (L) and right (R) eyes and as mean IOP ± SEM per experimental group.

2.6. Outflow facility measurement

AHO facility was measured in vivo by a scientist masked to experimental group, using a previously established constant flow infusion technique (McDowell et al., 2022; Millar et al., 2011). Briefly, anesthesia was induced with IP ketamine/xylazine (100/10 mg/kg respectively; Fort Dodge Animal Health, Fort Dodge, IA, USA). IOP was recorded in both eyes before inducing topical anesthesia by 0.5% proparacaine HCl eyedrops (Alcaine, Alcon). The anterior chamber of both eyes was cannulated with 32G needles (Steriject TSK Laboratory, Japan) connected to a pressure transducer (BLPR2; World Precision Instruments, WPI, Sarasota, FL, USA) for continuous pressure determination. The opposing end of the transducer was connected to a 3-way valve, in turn connected to a PBS-filled 50-μl glass microsyringe (Hamilton Company, Reno, NV, USA) loaded into a microdialysis infusion pump (SP101i, WPI), and to an open-topped, variableheight PBS manometer. Following refilling of the anterior chamber from the manometer until IOP was equal to pre-cannulation tonometric IOP, the manometers were then closed to the perfusion lines, 30 min were allowed for pressure equilibration, then eyes were perfused at flow rates of 100 to 500 nl/min, in 100 nl/min stepwise intervals using the infusion pumps. A plot of mean equilibrated pressure versus flow rate was generated per eye. Linear regression of the slope yielded the total AHO resistance, and its reciprocal the total AHO facility. We excluded eyes where SC constriction at 500 nl/min flow rate increased total outflow resistance and thus caused deviation from linearity, and eyes with evident leakage due to extreme pressure variations or failure for pressure to equilibrate (19% and 6% of all eyes, respectively).

2.7. Corneolimbal wholemount and immunofluorescence

Mouse eyes were collected following transcardial perfusion with 4% paraformaldehyde (PFA; Sigma-Aldrich) as previously described (Bosco et al., 2018). The dorsal cornea was marked for later orientation, and the eye anterior segment was dissected out in ice-cold PBS, removing lens and posterior segment while maintaining intact iridocorneal tissue (TM, SC and iris) and ciliary body. Corneolimbal wholemount and immunostaining were performed with small modifications to previously published methods (Thomson and Quaggin, 2018). The cornea was flattened by 4 radial cuts delimiting each quadrant and the central iris was removed without disturbing the limbus. For immunostaining, eyes for all experimental groups were dissected and stained at the same time. Wholemounts were blocked and permeabilized by overnight incubation in 5% donkey normal serum, 2.5% bovine serum albumin, 0.5% Triton-X in Tris buffered saline (TBS) pH 7.5, then incubated for 3 days in primary antibodies, including rat anti-CD31 (1:50; 550274, BD Pharmingen, Franklin Lakes, NJ, USA), goat anti-Prox1 (1:250; AF2727, R&D Systems, Minneapolis, MN, USA), goat anti-Tie2 (1:200; AF762, R&D Systems), mouse anti-α-SMA (1:200; 14–9760, Thermofisher Scientific, Waltham, MA, USA), guinea pig anti-Iba1 (1:500; 234308, Synaptic Systems, Gottingen, Germany) or rabbit anti-HA (1:1,000, 3724T, Cell Signaling Technology, Danvers, MA, USA) and rabbit anti-cleaved caspase 3 (1:500, 570524 BD Biosciences, Franklin Lakes, NJ, USA). Afterwards, tissue was rinsed in TBST (TBS with 0.05% Tween 20) and incubated overnight in donkey secondary antibodies conjugated with Alexa Fluor 488, 555, or 647 nm (1:400, Invitrogen, La Jolla, CA, USA). Finally, wholemounts were mounted with Fluoromount-G (SouthernBiotech, Birmingham, AL, USA) with the external corneal surface facing the coverslip.

2.8. Confocal microscopy imaging and SC-Tie2 expression measurement

The entire SC circumference and distal limbal veins (dLVs) were imaged in corneolimbal wholemounts by resonance scanning confocal microscopy (A1R confocal system and Eclipse Ti inverted microscope; Nikon, Melville, NY, USA), a 20x objective and 3x optical zoom (0.41 μm/pixel resolution) and using image acquisition/analysis software (NIS-Elements C software, Nikon). Replicate samples with identical immunostainings were imaged with equal settings, and images only adjusted with ≤ 25% increased brightness for publication. For each SC quadrant, we collected multipoint images (0.8 μm z-step) that spanned from episcleral veins (EVs) throughout the SC. To allow sectorial analysis, the SC was divided in 12 radial segments (30° each, 3 segments per quadrant; Fig. 2D). Intensity of Tie2 immunofluorescence was quantified by “line densitometry” on single-slice images of the SC inner wall (SC-iw; recognized by CD31 coimmunostaining) in corneolimbal wholemounts. Densitometry was performed on three single-slice views of the SC-iw by manually tracing 3–5 nonoverlapping lines to collect 500–700 points per 30° canal segment (Suppl. Fig. 3A), and expressed as mean Tie2-intensity/segment. Tie2 densitometry was also quantified for all visible collector channels (CCs) and episcleral veins (EVs) for each SC, measured on maximum-intensity projection images by tracing 1 line across the CCs and 3 lines (spaced 10 μm apart) across EVs (Suppl. Fig. 3D). SC area was analyzed in maximum-intensity projection images of the CD31 channel, using the “polygonal line tool” to manually trace perimeter and length, and quantity area/length (μm²/mm) per SC quadrant (Suppl. Fig. 4A).

Fig. 2. Endothelial Tie2 expression is attenuated in SC, but not in collector channel (CCs) or episcleral veins (EVs), after injection of AAV2.Cre in Tie2^FL/FL mice.

(A) Experimental timeline. (B) Volume view of limbal vasculature in corneolimbal wholemount immunostained for Tie2, which is shown in pseudocolors by image depth (deeper SC and CC planes in white, and EVs in magenta). (C) Confocal images of SC quadrants from eyes representative of 4 and 12 wpi of AAV2.Cre and their uninjected (contralateral) eyes. Brackets mark SC segments with reduced Tie2 signal, noticeable by the predominance of red/CD31 signal over green/Tie2 (oriented is indicated). (D) Confocal image of the entire SC from a naïve eye (with some overlaying EVs) in a corneolimbal wholemount immunostained for Tie2. Diagram of the twelve 30 degree-SC sectors analyzed for Tie2 immunofluorescence intensity by line densitometry (arrows). (E) Bar graph of SC-Tie2 intensity, presented as mean ± SEM per sector (bars) and per experimental group (lines) for naïve eyes (n=8), 4 wpi Cre eyes (n=11) and uninjected eyes (n=3), and for 12 wpi Cre and uninjected eyes (n=3 each). Not significant, ns. (F) Histogram of frequency distribution (percent) for Cre/Uninj and Cre/Naïve ratios at 4 and 12 wpi (n=3 mice each).

2.9. Statistical analysis

Data analysis was performed in Prism 10 for macOS (GraphPad, San Diego, CA, USA). First, normal distribution of the data was evaluated by Kolmogorov-Smirnov and Shapiro-Wilk tests. Normal datasets with 2 means were compared by unpaired, two-tailed Student’s T-test or Mann-Whitney T-test (with Welch’s correction for unequal variances and Brown-Forsythe test), while multiple means were compared by 1-way ANOVA (or Kruskal-Wallis test), and by Tukey’s multiple comparison post hoc test. We applied mixed-effects analysis, 2-way ANOVA and post hoc Tukey’s multiple comparison test. Non-normal datasets were analyzed by Dunn’s post hoc test, and non-parametric statistical tests (Kolmogorov-Smirnov and Mann-Whitney). Datasets are reported as mean ± SEM, with P<0.05 set for rejecting the null hypothesis. Graphs indicate statistical significances as: *P<0.05, ** P<0.01, ***P<0.001, and ****P<0.0001.

3. Results

3.1. Schlemm’s canal endothelium is transduced by AAV2.Cre injected intracamerally in adult mice

To target the SC, we used self-complimentary, serotype 2 adeno-associated virus AAV2 (scAAV2) vectors, which have tropism for the SC inner-wall endothelium when delivered into the eye anterior chamber of live rodent eyes (Bogner et al., 2015; Buie et al., 2010; Lee et al., 2019), and ex vivo in human eyes (Rodriguez-Estevez et al., 2020). Unmodified AAV2 vectors are capable of transducing the rodent SC inner wall, although less efficiently than the enhanced AAV2 variant with triple-mutant capsid (Y444+500+730F), known as “AAV2(triple)” (Bogner et al., 2015; Buie et al., 2010), which we used here. First, to test whether Cre recombinase activation targeted SC, young adult mice of a Cre reporter strain (R26 ^tdTomato/+, 1.5-month-old) received a single intracameral injection of scAAV2(triple).Cre (referred to as “AAV2.Cre”; 3×10¹⁰ vg/eye in 2 μL; n=20 eyes per timepoint). Then, 3 and 4 weeks postinjection (wpi), we assessed the distribution of Cre recombinase activation in the SC and downstream limbal vessels by visualizing the expression of red fluorescent tdTomato (tdT) in corneolimbal wholemounts (Fig. 1A). Low-magnification inspection of whole eyes 3 wpi of AAV2.Cre showed tdT+ cells largely concentrated at the corneal limbus, which houses SC, and sparsely localized to the cornea and iris (Fig. 1B). Next, we immunostained wholemounts for CD31 (PECAM1), a canonical blood and lymphatic endothelial marker that labels SC endothelia, as well as blood and lymphatic limbal vessels (Aspelund et al., 2014; Kizhatil et al., 2014). Using confocal microscopy, we collected high-resolution images for each corneal quadrant encompassing the SC and distal limbal vessels (dLVs), and visualized SC and dLVs in separate maximum-intensity projections (Fig. 1C). SC planes consistently showed tdT+ cells overlapping most of the canal’s circumference and some cells flanking the canal’s anterior angle (Fig. 1D), whereas dLV planes exhibited few tdT+ cells along vessels (Fig. 1D’). To explore whether the tdT+ cells flanking SC were TM cells, R26 ^tdTomato/+ mouse corneolimbal wholemounts 4 wpi of Cre were coimmunostained for CD31 and alpha-smooth muscle actin (α-SMA), an established JCT-TM marker (de Kater et al., 1992; Overby et al., 2014a). Observation of single-slice images spanning from SC/TM interface to CCs confirmed the preferential concentration of tdT+ cells at the SC-inner wall (Suppl. Fig. 1B), abutting but not colocalizing with adjacent α-SMA+ TM cells (Suppl. Fig. 1A), and relatively scarce tdT+ at the SC outer wall and CCs (Suppl. Fig. 1C).

Fig 1. Cre recombinase activation is detectable in SC of adult R26 ^tTomato/+ mice after intracameral injection of AAV2.Cre.

(A) Experimental design. (B) Fluorescent microscopy of whole eyes (side view) representative of 3 wpi of AAV2.Cre in R26 ^tTomato/+ eye. Cells expressing tdTomato (tdT) predominantly localize to the corneal limbus, with few scattered across the cornea. (C) Confocal image of a limbus quadrant in a corneolimbal wholemount immunostained for CD31, split in two maximum-intensity projection of Schlemm’s canal (SC) and distal limbal vessels (dLVs) planes. (D) SC planes in eye representative of 4 wpi of AAV2.Cre (square in the cornea indicates the needle entry point). Abundant tdT-expressing cells (red) localize to SC endothelium (CD31, white) and few to the cornea (arrowheads). Zoomed in view (inset) reveals additional tdT+ cells adjacent to the proximal SC aspect (arrows). (D’) Distal LVs planes in the same wholemount exhibit very few tdT+ cells (inset, arrows) around aqueous and episcleral veins (AVs and EVs). (E) Single-slice view of SC immunostained for Prox1 (green) and CD31 (white), viewed shown as merged channels (1–3). Most tdT+ cells overlap CD31+/Prox1+ SC endothelial cells. (E’) Single-slice view of dLVs in the same wholemount. Scarce tdT+ cells (arrowheads) neighbor lymphatic vessels (Lym, Prox1+/green) and aqueous veins (AVs, arrowheads). (E“) Corresponding YZ-perspective views through SC (E) and dLV (E’) planes. Most tdT+ cells are confined to SC, few cells localize to dLV, and no cells are visible at depths immediately internal to SC or external to dLV. (F) High-resolution single-slice views of SC inner wall (iw) and corresponding YZ-perspective spanning both SC inner and outer walls (ow) from eye representative of 6 wpi of AAV2.Cre. Many cells expressing tdT coexpress Prox1+/CD31+ (asterisks), some lack one or both markers (Xs). Perspective views detects no tdT+ cell at the dLV plane.

To define whether tdT localized to SC endothelial cells, corneolimbal wholemounts were coimmunostained for Prox1 (prospero-related homeobox 1), a transcription factor expressed in SC-endothelial cells of the inner wall, as well as in limbal lymphatic endothelial cells (Aspelund et al., 2014; Jung et al., 2017; Kizhatil et al., 2014; Park et al., 2014; Truong et al., 2014). Visualization through SC and dLV planes as single Z-slices (0.8 μm) and YZ-perspectives revealed predominant tdT expression in SC cells coexpressing Prox1 and CD31 (Fig. 1E), and only occasional tdT+ cells devoid of CD31 apposed to limbal lymphatics and veins (Fig. 1E’). High-magnification inspection of single Z-slices (0.4 μm) and YZ-perspectives detected tdT+ cells localized to the SC-inner wall cells, which coexpressed both CD31 and Prox1 or neither of these SC markers, as shown for an eye representative of AAV2.Cre transduction in R26 ^tdTomato/+ mice (n=10) (Fig. 1F). These observations suggest that intracameral AAV2.Cre was able to selectively target the SC inner wall, resulting in stable expression of activated Cre recombinase in a subset of CD31/Prox1 double-positive SC endothelial cells, but not in the adjacent JCT trabecular meshwork or downstream limbal vasculature.

Since tissue-resident macrophages can non-specifically engulf AAVs by phagocytosis, altering the cellular distribution of viral transduction in diverse organs (Aalbers et al., 2017; Hamilton and Wright, 2021), we sought to evaluate whether tdT localized to myeloid cells, given that they reside and associate with the SC of mice and humans (Kizhatil et al., 2014; Margeta et al., 2018; Patel et al., 2020; van Zyl et al., 2020). Corneolimbal wholemounts from R26 ^tdTomato/+ mice were collected 4 wpi of AAV2.Cre, coimmunostained for myeloid cell marker Iba1 and CD31 and visualized throughout the limbal area by confocal microscopy. Myeloid Iba1+ cells concentrated at the corneal limbus and were abundant at the SC, but also present in dLVs and across the cornea (Suppl. Fig. 2A, 1). However, few Iba1/tdTomato double-positive cells were detected (Suppl. Fig. 2A, 2). This analysis suggests that intracameral delivery Cre recombinase via AAV2.Cre mainly transduces SC cells and is not significantly active in its associated macrophages. Altogether, these findings indicate that intracameral injection of scAAV2(triple).Cre in adult mice can selectively target and transduce Cre recombinase in SC endothelial cells, at its interface with the TM, without significantly reaching contiguous TM cells or downstream limbal endothelia.

3.2. Tie2 expression in SC inner wall is selectively attenuated in Tie2^FL/FL mice by intracameral AAV2.Cre

Tie2 signaling in SC has a well-established role in maintaining the canal’s integrity and function as a regulator of AHO and IOP (Bernier-Latmani and Petrova, 2017). Thus, we sought to knockdown Tie2 expression in SC of homozygous B6.Tie2 floxed (Tie2^FL/FL) mice (Savant et al., 2015) by unilateral injection of AAV2.Cre (3×10¹⁰ vg/eye) at 1.5 months of age. The intensity of Tie2 immunostaining was quantified across SC and dLVs 4 and 12 weeks after AAV2.Cre injection, in the uninjected contralateral eyes, and in eyes from naïve littermates (Fig. 2A). We immunostained corneolimbal wholemounts for Tie2 and CD31, to quantify levels of Tie2 expression at SC, collector channels (CCs) and episcleral veins (EVs), where Tie2 mRNA and protein have been detected in adult mice by both immunostaining and single-cell RNA sequencing (Kim et al., 2017; Kizhatil et al., 2014; Patel et al., 2020; Thomson et al., 2014; van Zyl et al., 2020). To resolve and distinguish endothelia corresponding to SC, CCs, EVs and other limbal vessels across the entire SC circumference, we collected multipoint, high-resolution (0.21 px/μm) confocal images spanning the entire SC and dLV planes for each corneolimbal quadrant (Fig. 2B). Visualization of max-intensity projection images of SC quadrants from Tie2^FL/FL mouse eyes 4 and 12 wpi of AAV2.Cre exhibited decreased Tie2 expression, in comparison to uninjected/contralateral eyes, evident as canal segments with dim and sparse labeling and reduced area and complexity, which localized to discrete or multiple zones in diverse quadrants (Fig. 2C).

To assess Tie2 expression levels in SC and detect possible Tie2 downregulation along the length of the canal, we quantified the intensity of Tie2 signal along 12 radial sectors (30° each, dorsal D1-D2-D3, nasal N1-D3, etc.; Fig. 2D), using line densitometry in single-slice images (0.4 μm) of the SC inner wall, as detailed in Methods and diagrammed (Suppl. Fig. 3A). Quantification of Tie2 mean intensity in SC per 30° sector (D1–3, N1–3, V1–3 and T1–3) across conditions (AAV2.Cre, uninjected/contralateral and naïve) detected significantly lower Tie2 expression 4 wpi of AAV2.Cre relative to uninjected eyes (796±24 vs. 1025±27 respectively, P<0.0001 by 2-way ANOVA with Tukey’s multiple comparisons test) and age-matched (~2.5 months) naïve eyes (954±19, P=0.0002) (Fig. 2E). Decreased Tie2 expression persisted by 12 wpi compared to uninjected eyes (748±37 vs. 1045±32, respectively, P=0.0002) and young naïve eyes (954±19, P=0.0012), while no significant differences existed between AAV2.Cre 4 versus 12 wpi, uninjected 4 versus 12 wpi, or naïve versus uninjected 4 or 12 wpi. To represent the nonuniform distribution of SC levels of Tie2 intensity along the canal circumference, we used radar charts to illustrate Tie2 levels per SC sector for individual eyes representative of naïve and 4 and 12 wpi uninjected and AAV2.Cre eyes (n=3 mice/group). This analysis further conveyed the asymmetric distribution of Tie2 levels along SC regions and the variable quadrant localization of downregulated Tie2 expression (Suppl. Fig. 3B).

To evaluate the efficiency of Tie2 knockdown in SC, we calculated the frequency distribution of AAV2.Cre eyes with Tie2 expression reduced ≥25% relative to uninjected/contralateral eyes (Cre/Uninj. Ratio) and to naïve littermate eyes (Cre/Naïve Ratio) 4 and 12 wpi (Fig. 2F). Ratio of 4 wpi Cre/Uninj. showed ≥25% lower Tie2 in SC of almost 60% of injected eyes (of which 15% had ≥50% Tie2 reductions), which expanded to 75% of eyes by 12 wpi (of which ~40% had ≥50% lower Tie2 expression). Thus, Tie2 knockdown was effectively induced in SC in over half of the eyes 4 wpi of AAV2.Cre, and in 3/4 at 12 wpi relative to their uninjected/contralateral eyes. Similarly, the ratio of Cre/Naïve showed 53% Cre eyes with decreased Tie2 expression 4 wpi and 72% 12 wpi, when 25% eyes had ≥50% lower Tie2 expression in SC than naïve controls. Depiction of ratios of Cre/Uninj. and Cre/Naïve per 30° SC-sector for individual eyes 4 and 12 wpi of AAV2.Cre (n=3 mice/wpi), confirmed the above observation that the induced knockdown of Tie2 expression in SC is segmental and asymmetric relative to uninjected/contralateral and naïve eyes (Suppl. Fig. 3C).

Because R26 ^tdTomato/+ mice injected with AAV2.Cre showed some transduced cells near limbal vessels distal to the SC, we asked whether Tie2 downregulation also affected the collector channels (CCs) or the episcleral veins (EVs). We performed line densitometry of Tie2 immunofluorescence, tracing 1 line across CCs (n=9–12/eye) and 3 lines across EVs (n=2–4/eye) in maximum-intensity projection images (Suppl. Fig. 3D). No significant CC-Tie2 expression differences were found between 4 wpi AAV2.Cre versus uninjected/contralateral eyes (1,392±61 and 1,380±52, respectively) or naïve eyes (1,241±41; by Mann-Whitney unpaired, 2-tailed T-test for 4 wpi and Brown-Forsythe and Welch ANOVA tests for 12 wpi) (Suppl. Fig. 3E). We also found no differences in EV-Tie2 expression between 4 wpi AAV2.Cre versus uninjected eyes (2,539±197 and 2,387±190, respectively) or naïve eyes (2,204±175 by Brown-Forsythe and Welch ANOVA tests) (Suppl. Fig. 3F). These data indicated that AAV2.Cre had little or no effect on Tie2 expressed by limbal vessel endothelia downstream from SC, further suggesting that Cre selectively affected Tie2 expression in SC inner wall within the conventional outflow pathway.

Lastly, to assess whether the induced Tie2 downregulation affected SC integrity, we estimated the area of SC at 4 wpi AAV2.Cre, relative to naïve and uninjected/contralateral eyes. Analysis of the linear area (μm²/mm) of SCs immunostained with CD31 revealed significant reduction in AAV2.Cre-injected versus naïve eyes (164±8.5 vs. 202±5.3 μm²/mm, P=0.0009 by Mann-Whitney unpaired, 2-tailed T-test) and versus uninjected eyes (189±6.1 μm²/mm, P<0.0165), but no statistical difference (P=0.1093) between naïve and uninjected eyes (Suppl. Fig. 4A). The detected SC shrinkage underscores the dependence of SC stability on Tie2 expression levels, resembling the dependence uncovered by Tie2/TEK dose reduction in heterozygote Tie2^WT/KO and aged eyes (Kim et al., 2017; Souma et al., 2016). Overall, these findings imply effective knockdown of Tie2 in SC by AAV2.Cre injection in Tie2^FL/FL mice, for which we refer to this mouse model as SC-Tie2 knockdown (or “SC-Tie2^KD”).

3.3. TM regression parallels SC-Tie2 knockdown in Tie2^FL/FL mice

Although the TM does not express Tie2, reduced Tie2 signaling not only leads to loss of SC integrity but also impacts the TM, as previously reported in congenital Tie2-hemizygotes (Souma et al., 2016) and Angpt1-KO mice (Thomson et al., 2017). We thus sought to determine whether the SC-Tie2 attenuation induced in adult mice could also impact the stability of TM cells that interface the SC-inner wall. We performed coimmunostaining for alpha-smooth muscle actin (α-SMA), an established TM marker (de Kater et al., 1992; Overby et al., 2014a) and for CD31, and collected corneolimbal wholemounts from naïve Tie2^FL/FL mice and littermates at 3 wpi of AAV2.Cre (1×10¹⁰ vg/eye) or PBS, and from R26 ^tdTomato/+ 4 wpi AAV2.Cre (n=4 mice each). Confocal images spanned the SC circumference and depths plus the juxtaposed TM region, including juxtacanalicular cells and lamellated beams. Control eyes, including naïve Tie2^FL/FL mice and 3 wpi of PBS showed similarly well-developed TMs, with uniformly intense and patchy α-SMA+ cells along most sectors, as well as regions with reduced or absent ⍺SMA staining (Fig. 3A), as previously described (Overby et al., 2014a). As control for IOP-independent effects of AAV2-Cre, 4 wpi in R26 ^tdTomato/+ mice resulted in intact TM with ⍺SMA patterns, comparable to naïve and PBS-injected Tie2^FL/FL mice (Suppl. Fig. 4B). In contrast, Tie2^FL/FL mice 3 wpi of AAV2.Cre showed TM with variable morphology, with sectors displaying typical ⍺SMA+ cells, sectors with drastic loss of ⍺SMA expression (Fig. 3A, insets), and sectors with reduced lamella and condensed cells (Fig. 3B). Notably, TM regions with low or nearly absent ⍺SMA labeling seemed to colocalize with SC sectors with reduced area and/or CD31 expression (Fig. 3A, insets). Preliminary observation of apoptotic cell death in corneolimbal wholemounts coimmunostained for cleaved caspase-3 (cCasp3), ⍺SMA and CD31, showed comparably scarce cCasp3+ cells along the juxtacanalicular TM or SC of both naïve Tie2^FL/FL mice and 3 wpi of AAV2.Cre (data not shown); however it is possible that apoptotic or necrotic cell death could contribute to earlier or gradual ⍺SMA+ TM and/or cell loss, a mechanism previously reported in human and rodent glaucoma (Stamer and Acott, 2012; Xia and Zhang, 2024). Together, these observations suggest that Tie2 knockdown in SC indirectly damages TM adult morphology, potentially through reduced cellularity, reproducing the hypoplastic TM phenotype of hemizygous Tie2^KO/+ mice (Souma et al., 2016).

Fig. 3. TM morphology is altered in Tie2^FL/FL mice after injection of AAV2.Cre.

(A) Confocal images of corneolimbal wholemounts coimmunostained for ⍺SMA and CD31, spanning from JCT-TM to dLVs. Control eyes, naïve and PBS, display strong ladder-like ⍺SMA expression throughout most of the TM, but also sectors with lower ⍺SMA expression (asterisks). AAV2.Cre eyes show multiple TM regions almost deprived of ⍺SMA+ cells (brackets). Insets) Single slice views of SC-TM interface at dorsal quadrants representative of each condition. The JCT-TM shows drastic loss of ⍺SMA+ cells 3 wpi AAV2-Cre. (B) Example of a TM sector with moderate to drastic loss of ⍺SMA+ cellularity 3 wpi of AAV2.Cre. Outline delineates SC.

3.4. Sustained ocular hypertension develops by AAV2.Cre transduction of SC in Tie2^FL/FL mice

Since genetic Tie2 signaling reduction has a demonstrated ocular hypertensive effect (Souma et al., 2016), we next investigated the functional impact of SC-Tie2 knockdown on IOP and AHO (Fig. 4A). First, longitudinal IOP changes were assessed in floxed Tie2 mice after bilateral injection of AAV2.Cre or AAV2.GFP (3×10¹⁰ vg/eye each). Homozygote Tie2^FL/FL mice showed significant IOP elevation after AAV2.Cre delivery compared to baseline (15.0±0.27 mm Hg; mean ± sem, n=20–40 eyes) as early as 1 wpi (20.8±0.41 mm Hg; P<0.0001 by Kolmogorov-Smirnov), which stabilized thereafter with identical statistical significance for the 8 weeks assessed (Fig. 4B). However, IOPs were unchanged 3 days after AAV2.Cre injection (15.18±6.6, n=12 eyes; data not shown), suggesting gradual IOP rise over the first week postinjection. As expected, AAV2.Cre injected eyes at all times assessed had significantly higher IOPs (P<0.0001) than AAV2.GFP eyes, controls that remained within baseline levels (16.1±0.46 mm Hg, n=28–50 eyes). In contrast, hemizygous Tie2^FL/+ mice injected with AAV2.Cre showed inconsistent and moderate IOP elevations from baseline that reached significance at some but not all timepoints evaluated (18.4±0.47 mm Hg at 1, 4, 5 or 8 wpi, P<0.05, n=20 eyes) (Fig. 4C).

Fig. 4. IOP and AHO facility are dysregulated in Tie2^FL/FL mice with AAV2.Cre titer-dependency.

(A) Experimental timeline. (B) Mean IOP in homozygote Tie2^FL/FL mice after bilateral injection of AAV2.Cre (3×10¹⁰ vg/eye). AAV2.Cre eyes show significantly raised IOPs from 1 to 8 wpi relative to baseline (****P<0.0001, n=20–40 eyes) and to AAV2.GFP controls (****P<0.0001, n=28–50 eyes). (C) IOPs in hemizygous Tie2^FL/+ mice injected with bilateral AAV2-Cre show less significant elevations at irregular timepoints (1, 4, 5 and 8 wpi) relative to baseline IOPs (*P<0.05; n=20 eyes). (D) Unilateral injection of AAV2.Cre (1×10¹⁰ vg/eye; R eyes) in homozygous Tie2^FL/FL mice results in significantly increased mean IOP from 1 to 8 wpi and 12 wpi (****P<0.0001, n=23–35 eyes) relative to baseline, uninjected (contralateral) eyes and AAV2.GFP controls (****P<0.0001; n=16–29 eyes each). (E) Mean IOPs in wild-type C57BL6/J mice (n=10) remain unchanged relative to baseline after unilateral AAV2.Cre. (F) Dose-response analysis of IOP over the course of a month after injection of three AAV2.Cre titers in Tie2^FL/FL mice. Weekly IOPs increase with 6×10⁰⁹ vg/eye similarly to 1×10¹⁰ vg/eye (P<0.0001; n=8 eyes), while 1.5×10⁰⁹ vg/eye increases IOP with lower significance (P<0.01 to p<0.05; n=8 eyes). (G) Box plots of outflow facility show significant reduction at 4 wpi of AAV2.Cre in Tie2^FL/FL mice compared with naïve and AAV2.GFP injected eyes (**P=0.0081, n=11 and ***P=0.0006, n=9 mice, respectively), which did not significantly differ with each other.

To assess the feasibility of using the uninjected contralateral eyes as intra-animal normotensive controls, IOPs were monitored after unilateral injection of AAV2.Cre or AAV2.GFP (1×10¹⁰ vg/eye) in Tie2^FL/FL mice. Similar to bilaterally injected mice, we observed significantly increased IOPs from baseline (14.8±0.20, n=23–35 eyes) at 1 wpi of AAV2.Cre (19.9±0.24 mm Hg; P<0.0001 by Kolmogorov-Smirnov), which persisted for 12 weeks, the longest timepoint studied (Fig. 4D). At all assessed timepoints, mean IOPs in AAV2.Cre right eyes (R) were significantly higher (P<0.0001) than in uninjected/contralateral left eyes (L) and littermates injected with AAV2.GFP R (n=29 eyes). Both controls maintained no significantly changed IOPs relative to baseline or among themselves at any timepoint.

To test for Cre activation-independent effects of AAV2.Cre in IOP, wild-type B6 mice (n=10) were injected with AAV2.Cre (2×10¹⁰ vg/eye), which showed no significant IOP changes from baseline over 4 weeks (Fig. 4E). In addition, to establish the lower AAV2-Cre titer required to induce significant IOP elevation in Tie2^FL/FL mice, we performed a dose response curve. Compared to the levels used in this study (3×10¹⁰ and 1×10¹⁰ vg/eye AAV2.Cre, P<0.0001 and P<0.001), unilateral injection of 6×10⁰⁹ vg/eye induced a significant mean IOP elevation from baseline from 1 to 4 wpi (P<0.0001, n=8 eyes) comparable to 1×10¹⁰ vg/eye, while 1.5×10⁰⁹ vg/eye AAV2.Cre elevated IOP but with lesser significance (P<0.001 to P<0.01, n=8 eyes) (Fig. 4F). This dose-range evaluation suggests 6×10⁰⁹ to 1×10¹⁰ vg/eye as adequate AAV2.Cre titers to effectively induce meaningful and steady ocular hypertension in Tie2^FL/FL mice.

Overall, these findings showed that a single AAV2.Cre intracameral injection (bilateral or unilateral) led most homozygous Tie2^FL/FL mice to consistently develop persistent and steady ocular hypertension for at least 2 months, but only inconsistent mild IOP elevations in hemizygous Tie2^FL/+ mice, suggesting an effective and scalable induction of SC-Tie2 knockdown that prompts IOP dyshomeostasis.

3.5. Aqueous outflow facility is impaired in SC-Tie2^KD Tie2^FL/FL mice

Since ocular hypertension is tightly coupled to dysregulated outflow resistance/facility at the conventional outflow pathway (McDowell et al., 2022; Stamer and Acott, 2012), we tested if IOP elevation was coupled to AHO facility decline. By in vivo quantification of total outflow facility in naïve Tie2^FL/FL and littermates at 4 wpi of bilateral AAV2.Cre or AAV2.GFP (3×10¹⁰ vg/eye), we detected significantly reduced mean outflow facility in AAV2.Cre eyes (11.13±1.06 nl/min/mm Hg, n=11) relative to naïve eyes (21.88±4.47 nl/min/mm Hg, n=6) and AAV2.GFP eyes (25.93±3.72 nl/min/mm, n=9) (Fig. 4G). Thus, AAV2.Cre eyes had 49% lower facility than naïve eyes (p = 0.0081 by 2-tailed/unpaired Student’s T-test) and 57% lower than GFP eyes (P=0.0006), while both controls had comparable outflow facility (P=0.50). The selective and uniform reduction in outflow facility induced by AAV2.Cre in Tie2^FL/FL mice is consistent with decreased aqueous drainage as a mechanism driving IOP increases.

3.6. Macrophages accumulate in the SC-TM interface of SC-Tie2^KD Tie2^FL/FL mice

Myeloid cells localize to the iridocorneal angle, and macrophages are highly enriched at the SC and TM of mice and humans (Kizhatil et al., 2014; Margeta et al., 2018; Patel et al., 2020; van Zyl et al., 2020), however, their potential involvement in conventional outflow and ocular hypertensive changes remains underexplored. To gauge whether intracameral AAV2.Cre injection in Tie2^FL/FL mice led to long-lasting effects on the density of macrophages associated with the SC/TM interface, we collected corneolimbal wholemounts 3 wpi of AAV2.Cre (1×10¹⁰ vg/eye), AAV2-empty (3×10¹⁰ vg/eye) or PBS, and from naïve littermates (n=4 mice, each), coimmunostained for Iba1 and CD31, and imaged by confocal microscopy throughout SC planes, which included the abutting TM outermost surface and some distal limbal vessels. Low-magnification images representative of naïve and PBS-injected eyes displayed comparably dense Iba1+ cells along the SC circumference, whereas visibly higher cell densities concentrated at SCs in AAV2.Cre eyes (Fig. 5A). High-magnification views at each SC quadrant showed PBS-injected controls with discrete Iba1+ cells with a distribution undistinguishable from naïve eyes, but vastly overlapping Iba1+ cells at 3 wpi of AAV2.Cre (Fig. 5B). Additionally, Tie2^FL/FL control mice injected with empty AAV2 vector, displayed patterns of SC-associated myeloid cells undistinguishable from naïve and PBS controls (Suppl. Fig. 5). These qualitative observations are consistent with a sustained innate immune response to SC/TM structural and/or functional alterations induced by SC-Tie2 knockdown. In support of this, the absence of long-lasting myeloid cell changes at the SC in control Tie2^FL/FL mice injected with PBS or empty AAV2, and in R26 ^tdTomato/+ mice injected with AAV2.Cre, respectively rule out an injury response to intracameral injection (PBS/Tie2^FL/FL), an anti-viral immune response elicited by the viral vector (empty-AAV2/Tie2^FL/FL), or Cre-toxicity driven SC and/or TM regression and myeloid recruitment (Cre/R26 ^tdTomato/+). The triggers and consequences of the macrophage expansion observed at the SC-TM interface of SC-Tie2^KD Tie2^FL/FL mice deserves further investigation, as the impact of these resident and/or mobilized myeloid cells on aqueous drainage and IOP is unclear.

Fig. 5. Myeloid cells accumulate at the SC/TM interface after Tie2 knockdown in SC of in Tie2^FL/FL mice.

(A) Confocal images through SC and outermost TM (JCT) in corneolimbal wholemounts coimmunostained for Iba1 (macrophages/monocytes) and CD31, from eyes representative of naïve Tie2^FL/FL mice and littermates 3 wpi of PBS or AAV2.Cre. Naïve and PBS-injected eyes exhibit prominent but comparable densities of Iba1+ cells (green) associated with SC and its boundaries (CD31, magenta). AAV2.Cre eyes show dense buildup of Iba1+ cells throughout SC and its adjacencies. Outline (white) marks SC boundaries. (B) High-magnification views demonstrate the comparable abundance of Iba1+ cells along the canal’s circumference in both naïve and PBS-injected eyes, and their drastic expansion 3 wpi AAV2.Cre.

3.7. SC-Tie2 disruption by CRISPR-saCas9 induces ocular hypertension in wild-type B6 mice

To develop an AAV-mediated approach for selective SC-Tie2 knockdown applicable to mice of any genetic background, we designed a CRISPR/Cas9 method targeting the Tie2/TEK gene expression at SC via intracameral delivery of AAV2(triple). We used the potent Staphylococcus aureus Cas9 (saCas9) (Ran et al., 2015), which is small enough to be packaged together with the single-guide RNA (sgRNA) into a single recombinant AAV vector (Chung et al., 2020), as depicted for sgTie2 (Fig. 6A). This approach was first tested in vitro to validate Tie2/TEK gene targeting and knockdown, then in vivo by intracameral injection of AAV2.sgTie2/saCas9 (or control AAV2.saCas9) in adult wild-type B6 mice (Fig. 6B). A SaCas9-compatible sgRNA sequence targeting exon 1 of Tie2/TEK was identified (Suppl. Fig. 3A), and the corresponding DNA was cloned into a single plasmid also encoding SaCas9 (pX601 CRISPR/SaCas9) (Ran et al., 2015). To test the effectiveness of this vector in reducing Tie2 protein expression, HEK293T cells were cotransfected with plasmid carrying mouse Tie2 tagged with TurboGFP (Tie2-tGFP) and either Tie2-CRISPR/saCas9 (Tie2-E1-px601-AAV-CMV, 7447 bp) or ctrTie2-CRISPR/saCas9 (pX601-AAV-CMV-saCas9-noPAM-GFP-B; pCMV6-AC-GFP, PS100010) plasmids. Since Sa-Cas9 endonuclease expression can be monitored via conjugated hemagglutinin (HA) epitopes (saCas9–3XHA), transfection efficiency was confirmed after 48 hr, when controls cotransfected with Tie2-tGFP and ctrTie2-CRISPR/Cas9 plasmid (pX601) showed a majority of cells with HA expression colocalizing with tGFP, in contrast to cells transfected with Tie2-tGFP and Tie2-CRISPR/Cas9 plasmid (Tie2-E1) that only exhibited HA expression and no tGFP expression (Suppl. Fig. 6B). Quantification of cells coexpressing HA and tGFP indicated significantly decreased Tie2-GFP expression using Tie2-E1 plasmid (**P<0.01 by Brown-Forsythe 1-way ANOVA) relative to control pX601 (Suppl. Fig. 6C).

To test CRISPR-mediated Tie2 knockdown in vivo, the Tie2-E1 construct was further packaged into AAV2.sgTie2/saCas9 and control AAV2.saCas9 prepared. First, to define the extent and localization of SaCas9 endonuclease expression at the SC, we examined SaCas9 expression in SC from wild-type B6 mice 3 weeks after unilateral AAV2.sgTie2/saCas9 intracameral injection (1×10¹⁰ vg/eye). For this, corneolimbal wholemounts were coimmunostained for HA and CD31 or Tie2 and imaged by confocal microscopy at high-resolution (0.4 μm/pixel), throughout the SC and dLVs. High-magnification views of maximum-intensity projection and YZ-perspectives showed abundant HA+ cells localized to SC and colocalized with CD31+ cells (Fig. 6C), and with Tie2+ cells (Fig. 6D), but very few HA+ cells in downstream limbal vessels (Suppl. Fig. 6D). These observations indicate that intracameral AAV2.sgTie2/saCas9 could selectively target and transduce SC cells, resulting in evident HA expression in SC endothelial cells.

To verify whether this CRISPR/SaCas9 system affected IOP levels in wild type C57BL6/2J mice, we monitored IOP before and after unilateral injection of AAV2.sgTie2/saCas9 or control AAV2-saCas9, both 1×10¹⁰ vg/eye (Fig. 6E). Eyes injected with AAV2.Crispr/SaCas9 showed significant IOP elevation above baseline (15.0±0.30 mm Hg; n=15 eyes) by 1 wpi (20.0±0.33 mm Hg; n=15 eyes; P<0.0001 by Kolmogorov-Smirnov) that was maintained with identical significance throughout 4 wpi, the latest time assessed, with a peak elevation 3 wpi that was significantly different from 1 wpi (21.2±0.41 mm Hg P<0.05 by 2-way ANOVA and Tukey’s multiple comparisons test). Moreover, weekly mean IOPs in AAV2.Crispr/SaCas9 eyes were also significantly higher (P<0.0001) than in AAV2.saCas9 controls (15.00 to 16.3±0.55 to 0.60 mm Hg n=15 eyes), which did not differ from baseline or contralateral eyes at any timepoint.

Comparison of the ocular hypertension induced in wild-type B6^WT/WT versus floxed Tie2^FL/FL mice by respectively injecting unilateral AAV2.sgTie2/saCas9 or AAV2.Cre showed no significant differences from 1 to 4 wpi, but highly significant differences versus all other conditions and timepoints (****P<0.0001 by 2-way ANOVA and Tukey’s multiple comparisons test) (Fig. 6F). Additionally, IOP data demonstrated that 1×10¹⁰ vg/eye AAV2.sgTie2/saCas9 in B6 mice resulted in rapid and sustained IOP elevation to ~20 mm Hg from 1 wpi to at least 4 wpi, equivalent to the IOP elevation induced by the same titer of AAV2.Cre in Tie2^FL/FL mice. Collectively, these findings indicate that CRISPR/saCas9 gene editing via intracameral AAV2.sgTie2/saCas9 efficiently induces ocular hypertension in wild-type B6 mice.

4. Discussion

This study establishes two targeted genetic approaches to selectively knockdown Tie2 expression in adult mouse SC, to bypass defects in other tissues associated with global or endothelial knockouts of Tie2/TEK, its ligands or regulators (Kenig-Kozlovsky et al., 2018; Kim et al., 2019). We show that SC-Tie2 targeted knockdown results in damage and dysfunction of the conventional outflow pathway, mimicking triggering events of human SC/TM damage and ocular hypertension. Given the roles of Tie2 pathway in human SC integrity during aging and POAG ocular hypertension, and the lack of strategies to modify SC-specific genes (Pang and Clark, 2020), these novel models may prove useful to test candidate drugs targeted to SC to control ocular hypertension. In addition, our CRISPR/SaCas9/AAV2.saCas9-sgTie2 approach allows these studies to be performed in mice of any background, and potentially in other species.

4.1. SC-inner wall Tie2 knockdown using Tie2^FL/FL mice phenocopies global/hemizygous and adult/endothelial Tie2-knockouts

Prior studies established the requirement of SC-Tie2 signaling for SC and TM development and maturation (Kim et al., 2017; Souma et al., 2016; Thomson et al., 2017), and for adult preservation of SC structure and interface with the juxtacanalicular TM (Kim et al., 2017; Li et al., 2020; Thomson et al., 2017). Furthermore, the degree of Angpt/Tie2 signaling deficiency calibrates the severity of SC (and TM) defects, outflow decline and ocular hypertension, whether triggered by impaired SC development or maturation (Aspelund et al., 2014; Kizhatil et al., 2014; Park et al., 2014), or adult dysfunction and regression.

We found that targeted knockdown of Tie2 in SC of Tie2^FL/FL mice reduced Tie2 expression at the inner wall, detected by immunostaining, to levels that were sufficient to disturb the structural integrity of both SC and TM. This regression impacted SM and/or TM functionality, resulting in a diminution in outflow facility and persistent IOP elevation. The levels of Tie2 protein knockdown in SC and severity of conventional outflow pathway defects are comparable to both Tie2^KO/+ hemizygotes (embryonic, global deletion) (Souma et al., 2016) and Tie2^{VE-cadherin-KO/KO} (adult, endothelial deletion) mice (Kim et al., 2017), suggesting effective knockdown of Tie2 in SC of adult Tie2^FL/FL mice. Considering the essential role of Tie2 activation for adult SC/TM integrity and AHO/IOP balance previously reported (Kim et al., 2017; Li et al., 2020), these SC-Tie2 KD approaches may serve as tunable models of SC-Tie2 expression/activation deficiency, SC/TM regression, and conventional AHO and IOP imbalance.

4.2. Selectivity of Tie2 attenuation in SC

The findings reported here provide direct and indirect evidence that intracameral injection of the scAAV2(Y444+500+730F) vector into adult mice selectively transduced SC in the conventional outflow pathway, allowing targeted SC-Tie2 knockdown via AAV2.Cre in Tie2^FL/FL mice and AAV2.saCas9-sgTie2 in wild-type mice. Direct evidence showed that AAV2.Cre injection into R26 ^tdTomato/+ Cre reporter mice selectively and stably transduced the SC since tdTomato, a reporter for Cre recombinase activation in R26 ^tdTomato mice, predominantly colocalized to inner wall-endothelial cells double-positive for CD31/Prox1 and to CD31+ cells with low or no Prox1 expression. Scarce tdTomato+ cells localized to endothelia downstream of SC (collector channels, blood and lymphatic limbal vessels) known to express Tie2, or in the JCT/TM or ciliary body, where Tie2 expression has been reported in discrete cell subsets (Kapiainen et al., 2022; Kim et al., 2017; Kizhatil et al., 2014; Li et al., 2020; Lou et al., 2022; Patel et al., 2020; Thomson et al., 2014; van Zyl et al., 2020). Moreover, previous finding of TM and ciliary body transduction by the same AAV2 vector used here (Bogner et al., 2015), provides indirect evidence that the SC-Tie2^KD approach primarily impacts Tie2 expression in SC and not in TM and ciliary body. Further direct evidence for AAV2-mediated effects being selective to SC arises from the efficient SC transduction by AAV2.sgTie2/saCas9 identified through detection of HA-tagged saCas9, which colocalized to CD31+ and Tie2+ SC endothelial cells. Together, the evidence suggests that the induced conventional outflow defects and ocular hypertension were triggered by Tie2 knockdown in the adult SC-inner wall endothelium.

4.3. Role of SC in adult TM maintenance

Although the Tie2^FL/FL SC-Tie2^KD approach selectively affected Tie2 expression in the SC, the JCT/TM showed regressive defects (decreased ⍺SMA+ cellularity) 3 weeks after AAV2.Cre intracameral injection. This defect resembles the hypoplastic TM of hemizygous Tie2^WT/KO (Souma et al., 2016) and the severely defective TM of Angpt1^KO/KO mice (Thomson et al., 2017). However, unlike these developmental models where the SC is absent or incompletely formed, the SC was slightly reduced in size after SC-Tie2^KD but preserved its overall morphology. Since the TM was not assessed after adult deletion of endothelial Tie2 (Kim et al., 2017), our adult SC-Tie2 knockdown exposes significant TM regression secondary to SC-specific Tie2 downregulation, but uncoupled from overt SC structural deterioration. Future quantitative analysis of SC/TM cell death over multiple timepoints during SC-Tie2^KD should define whether apoptotic or necrotic mechanisms contribute to TM regression. The current finding argues for the critical involvement of SC-Tie2 signaling in adult TM maintenance, mediated either directly by SC signaling or contact with the TM, or indirectly by dysfunctional SC outputs, mainly elevated IOP. The observed phenotype could additionally be linked to other ligand-receptor pairs that imply SC-to-TM communication (Balasubramanian et al., 2024).

4.4. Myeloid cells as potential players in SC-TM alterations

Myeloid cells dwell at the conventional outflow tissues, and include tissue-resident macrophages and macrophages/monocytes recruited to the SC or TM of mice and humans in health and glaucoma (Coupland et al., 1994; Kizhatil et al., 2014; Margeta et al., 2018; Patel et al., 2020; van Zyl et al., 2020; van Zyl et al., 2022). In addition, a dose-dependent immune response has been demonstrated for AAV-mediated ocular gene therapies (Bainbridge et al., 2015; Dimopoulos et al., 2018; Xue et al., 2018) and testing for inflammation is recommended (Chan et al., 2021; Yang et al., 2022), particularly after intracameral AAV injections (Bastola et al., 2020). Accordingly, we explored lasting changes in the number of myeloid cells at the SC-TM interface 3 wpi of AAV2.Cre injection in Tie2^FL/FL mice, and detected greater concentrations of Iba1+ cells than in eyes identically injected with PBS or with an empty AAV2 vector, or in R26 ^tdTomato/+ eyes at 4 wpi AAV2.Cre. This implies that the myeloid-cell build up at the SC-TM interface after SC-Tie2^KD was not a persistent antiviral response or an immune response to corneal damage by injection, since acute corneal incision leads to transiently increased myeloid cell densities that resolve by 3 weeks (Kiesewetter et al., 2019). The innate immune cells amassed at the SC/TM following SC-Tie2 knockdown are poised to affect the conventional outflow pathway stability and function. Previous work has proposed aqueous humor regulation by infiltrating macrophages or monocytes (Alvarado et al., 2010; Hamanaka et al., 2002; Kagan et al., 2014), but the roles they serve in AHO and IOP homeostasis remain undefined. A comprehensive analysis of SC/TM myeloid cell types and dynamics during IOP elevation by SC-Tie2^KD may inform whether and how they contribute to adult AHO and IOP dysregulation.

Supplementary Material

1

Suppl. Fig. 1. Cre recombinase activation is not detectable in JCT-TM cells of adult Cre reporter mice 4 wpi of intracameral AAV2.Cre. (A) Single-slice confocal images of the SC/TM interface, SC inner and outer walls, coimmunostained for ⍺SMA (green) and CD31(white), representative of R26 ^tTomato/+ mice 4 wpi of AAV2.Cre (n=8 eyes). Some tdT-expressing cells border the JCT-TM without overlapping its ⍺SMA+ cells (1), densely localize to the SC-iw (2), and only occasionally to the SC-ow and its CCs (3).

Suppl. Fig. 2. Cre recombinase activation is not detectable in myeloid cells associated with SC and TM. (A) Confocal image of SCs coimmunostained for Iba1 (green) and CD31 (white), representative R26 ^tTomato/+ eye 4 wpi of AAV2.Cre. Myeloid cells are abundant at the limbus and densely localized to the entire SC circumference. (1) Zoomed in view of the temporal SC quadrant (delineated) exposes the preferential association with SC, and the scarcity of cells with overlapping tdT and Iba1. (2) Similar view of the dorsal SC quadrant (delineated) with merged Iba1 and tdT channels. Few cells show colocalized Iba1+ tdT+ cells (boxed).

Suppl. Fig. 3. Tie2 expression analysis method, single-eye and CC/EV data.(A) Diagram of the densitometry analysis of Tie2 immunofluorescence intensity within 30° SC-segments (3 per quadrant). Examples of single-slice images used to quantify Tie2 intensity in the Tie2 channel, by tracing densitometry lines along the SC-inner wall (600–900 μm/30° sector), guided by the CD31 channel. Histograms were obtained until collecting 500–700 densitometry points. (B) Radar charts of mean SC-Tie2 intensity in eyes representative of naïve, 4 and 12 wpi of unilateral AAV2.Cre (n=3 pairs each; L, left eye; R, right eye). The SC circumference was analyzed as 12 30-degree sectors, 3 sectors per quadrant (D, dorsal; N, nasal, V, ventral, and T, temporal). (C) Individual eye ratios (AAV2.Cre/Uninj., AAV2.Cre/Naïve and Uninj./Naïve) per SC sector (3/quadrant, Dorsal, D, etc.) at 4 and 12 wpi (n=3 eyes/ratio-timepoint). (D) Method of line densitometry analysis of Tie2 expression in CCs and EVs. (E) Scatter/bar plots for Tie2 immunofluorescence intensity in CCs, showing mean ± SEM per group (bar) and for individual vessels (dots). (F) Same for EVs. Naïve (n=2 eyes), 4 wpi Cre and uninjected (n=4 eyes each).

Suppl. Fig. 4. SC and TM integrity.(A) Diagram of SC area analysis by measuring the linear area per SC quadrant, on max-intensity projection images. (B) Scattered-dot plot of SC linear area analysis, showing mean ± SEM per condition, and per quadrant (dots represent means per individual quadrant). Naïve (n=2 eyes), 4 wpi Cre and uninjected (n=4 eyes each). (C) Confocal images of corneolimbal wholemounts coimmunostained for ⍺SMA and CD31 representative of R26 ^tTomato/+ eyes 4 wpi of AAV2.Cre. Max-projection images of JCT-TM and SC reveal preserved TM morphology, comparable to those of naïve and PBS-injected controls (Fig. 3A). The same is suggested by single-slice views through the SC/TM interface.

Suppl. Fig. 5. Myeloid cell distribution at the SC is unaffected in Tie2^FL/FL mice following intracameral injection of an empty AAV2 vector. (A) Confocal images of a corneolimbal wholemount representative of Tie2^FL/FL eyes collected 3 wpi of an empty AAV2 vector and coimmunostained for Iba1 (macrophages/monocytes) and CD31. Max-intensity projection images of myeloid cells, viewed as single and merged channels, reveal cell density and aspect similar to those of naïve and PBS-injected control eyes (Fig. 5).

Suppl. Fig. 6. CRISPR-mediated Tie2 knockdown. (A) CRISPR/SaCas9 target sequences and protospacer adjacent motif (PAM) sequence (underlined) in Tie2/TEK exon 1. (B) Fluorescent microscopy image of HEK293T cell cultures 48 hr after cotransfection with Tie2-tGFP and either ctrTie2-CRISPR/Cas9 (pX601) or cTie2-CRISPR/Cas9 Tie2-E1 plasmids, and immunostaining for HA (magenta) to detect expression of saCas9–3xHAd. Tie2-E1 efficiently knockdown Tie2-tGFP expression is evidence by the loss of tGFP expression and persistence of HA immunostaining, relative to Px601-controls where both signals colocalize. (C) Quantification of tGFP/HA double-positive cells in three ROIs per sample (n=3 each), presented as the mean ± SEM (**P<0.01). (D) Confocal image of the same wild-type eye corneolimbal wholemount shown in C, here displaying the limbal vessels distal to the SC. Few HA+ cells (green) are detectable in the CD31+ endothelial cells (magenta).

Highlights.

• Intracameral AAV2 stably transduces Cre recombinase or SaCas9 in Schlemm’s canal (SC).

• AAV2-Cre in Tie2^FL/FL mice selectively attenuates Tie2 expressed in SC endothelium.

• Structurally, SC area reduces and the contiguous trabecular meshwork regresses.

• Functionally, steady ocular hypertension and impaired aqueous outflow develop.

• SC-Tie2 disruption by CRISPR-SaCas9 induces ocular hypertension in wild-type mice.