Cooperation of Angiopoietin-2 and Angiopoietin-4 in Schlemm's Canal Maintenance

Purpose Defects in the iridocorneal angle tissues, including the trabecular meshwork (TM) and Schlemm's canal (SC), impair aqueous humor flow and increase the intraocular pressure (IOP), eventually resulting in glaucoma. Activation of endothelial tyrosine kinase receptor Tie2 by angiopoietin-1 (Angpt1) has been demonstrated to be essential for SC formation, but roles of the other two Tie2 ligands, Angpt2 and Angpt4, have been controversial or not yet characterized, respectively. Methods Angpt4 expression was investigated using genetic cell fate mapping and reporter mice. Congenital deletion of Angpt2 and Angpt4 and tamoxifen-inducible deletion of Angpt1 in mice were used to study the effects of Angpt4 deletion alone and in combination with the other angiopoietins. SC morphology was examined with immunofluorescent staining. IOP measurements, electron microscopy, and histologic evaluation were used to study glaucomatous changes. Results Angpt4 was postnatally expressed in the TM. While Angpt4 deletion alone did not affect SC and Angpt4 deletion did not aggravate Angpt1 deletion phenotype, absence of Angpt4 combined with Angpt2 deletion had detrimental effects on SC morphology in adult mice. Consequently, Angpt2−/−;Angpt4−/− mice displayed glaucomatous changes in the eye. Mice with Angpt2 deletion alone showed only moderate SC defects, but Angpt2 was necessary for proper limbal vasculature development. Mechanistically, analysis of Tie2 phosphorylation suggested that Angpt2 and Angpt4 cooperate as agonistic Tie2 ligands in maintaining SC integrity. Conclusions Our results indicated an additive effect of Angpt4 in SC maintenance and Tie2 activation and a spatiotemporally regulated interplay between the angiopoietins in the mouse iridocorneal angle.


Immunofluorescence stainings
The eyes were fixed in 4% PFA for 30 min at room temperature (RT; or overnight at +4 °C) and stored in 1×PBS at +4 °C. For SC whole mount staining, corneal limbus region containing SC was dissected out either as whole corneal preparation or as two or three stripes, and the iris was carefully removed. 5% donkey serum-0.3% Triton X-100 in 1×PBS was used as the incubation solution in all staining steps. The samples were blocked for one hour at RT, incubated overnight with the primary antibodies at +4 °C, washed by changing the solution at least four times during two-three hours at RT, incubated overnight with secondary antibodies at +4 °C, washed, and flat-mounted with Immu-Mount (Thermo Fisher Scientific, Waltham, MA). The stripes were flat-mounted side by side in a vertical position the cornea always to the right to facilitate optimal imaging of the whole SC.
Alternatively, whole SC regions from Angpt4 +/Cre ; Rosa26 +/mTmG reporter mice were also directly flatmounted without any staining. Retina whole mounts were stained similarly as whole SC.
For sectioned tissue images of Angpt4 +/Cre ; Rosa26 +/mTmG mice, whole eyes were fixed with 95% ethanol overnight and processed for paraffin sectioning and dehydration as described elsewhere. 1 For cryosection staining of WT and Angpt2 -/mice, eyes were freshly frozen into Tissue- Tek OCT compound (Sakura, Alphen aan den Rijn, The Netherlands), cut to 10 µm sections and stored at -80 °C. Upon staining, sections were fixed with 4% PFA for 3 min, blocked with 5% donkey serum-0.3% Triton X-100 in 1×PBS for 30 min and incubated overnight with primary antibodies at +4 °C in a humified chamber followed by incubation with secondary antibodies for 2 h at RT and mounting with Immu-Mount.

Confocal microscopy and morphometrical analyses
The fluorescently labeled samples were imaged with Zeiss LSM780 confocal microscope (Zeiss, Oberkochen, Germany) using Plan Apochromat 10×/0.45 air, 20×/0.8 air, or 40×/1.4 oil objectives. 20×, and 0.21 µm × 0.21 µm with 40×, and the pinhole size was set to 1 AU. Z-stacks were taken at approximately 5 µm intervals for 10× and 20× images (morphometrical analyses) and with approximately 2 µm intervals for 40× images (Angpt expressions), and the images were processed to maximum intensity projections with Zen 2.3 lite software (Zeiss). 20× 1×2 tile images were taken carefully covering the whole SC (or as much as possible) for CD31 + SC area, Prox1 expression, and pTie2/Tie2 expression measurements, while 10× larger tile images were taken from CD31 and Lyve1 stainings to image the whole corneolimbal area. CD31 + SC area was measured from the 20× images using the Zen software. SC area was divided by the whole analyzed image area as a control, and the measurements from all images were averaged to obtain a single value for each eye. Prox1 + nuclei in the SC were analyzed from the 20× images using Fiji (ImageJ) software. CD31 staining was first used to determine SC area. Uneven local background fluorescence was subtracted from Prox1 staining by subtracting the duplicated median filtered image from the original image (corrected image = imagemedian filtered image, radius 50). CD31 staining masked region of interest (ROI) was then thresholded with constant pixel values (17-255, set to include all but the very faintest positive nuclei), and the binarized image was median filtered with a radius of 2 pixels to remove remaining noise.
Analyze particles-function was used to create ROI maps of the nuclei, and these ROI maps were then used to measure the number and area of Prox1 + nuclei from the background-subtracted, non-binarized image. Finally, the number of Prox1 + nuclei per mm 2 of CD31 + SC area and the percentage of total Prox1 + area from the CD31 + SC area per eye were calculated. Tie2 and pTie2 intensity mean values were measured from the 20× images with the Zen software using CD31 staining to determine SC area; the background intensity values measured from non-SC areas of the images were subtracted from the values, and after finding no differences in total Tie2 levels, pTie2/Tie2 ratio was reported. Lyve1 + corneal limbus LV area was quantified from the 10× images using the Zen software. LV area was normalized with SC length measured from 10× images using Fiji; conjunctival LVs and Lyve1 + limbal macrophages 4,5 were excluded from LV area analyses. Circulating limbal arteries and perilimbal veins which circulate the limbal area in a parallel manner, 6 collecting vein branches (including both collector channels with direct connections to the SC and perilimbal venous plexus branches at the sites of major vein drainage towards episcleral veins), 6 corneal arcades (loops of the limbal capillary plexus on the corneal side), 6 and SC narrowing points were calculated from the 10× images. SC narrowing points were determined as distinctive narrowings/convolutions/twists in the SC, and a new measurement point was assigned after every 100 µm if the SC continued as narrow.

scRNAseq data analyses
Previously published 7 scRNAseq data (data accessible at NCBI GEO database, 8 accession number GSE146188) were downloaded from the Broad Institute Single Cell Portal (http://singlecell.broadinstitute.org/) and reanalyzed in R using the Seurat package 9 following standard procedures as suggested by the package authors. The reanalyzed Seurat3 object is available upon request.

X-gal staining
Eyes from Angpt4 LacZ and control mice were fixed in 4% PFA for 10 min on ice, washed for 1 h with washing buffer (0.02% NP-40, 0.01% Sodium Deoxycholate, 2 mmol/L MgCl2, 5 mmol/L EGTA in 0.1 mol/L phosphate buffer, pH 7.4), and incubated overnight in 1 mg/mL X-gal/10 mmol/L potassium ferro-and ferricyanide at RT. After 1 h washing, eyes were postfixed in 4% PFA for 30 min at RT, embedded in OCT compound, frozen, cut to 10 µm sections and immediately mounted with Immu-Mount. Sections were imaged with Zeiss Axio Imager motorized brightfield microscope.

Quantitative PCR
RNA was isolated from tissues using Fibrous Tissue Mini Kit (Qiagen, Germantown, MD). 1 µg (anterior eye) or 3 µg (kidney, lung) of total RNA was transcribed to cDNA, and qPCR was performed as described previously 10 using β-actin or Gapdh as reference genes. Primers are reported in the Supplementary Table S1.

Transmission electron microscopy
TEM sample processing and imaging was performed as previously described. 10 The number of giant vacuoles per 100 µm of SC inner wall endothelium was measured with Fiji. Eyes for TEM analyses were always collected first after sacrificing a mouse to prevent inconsistent loss of vacuoles. A small hole was cut to the middle of the cornea after fixation to facilitate the access of the reagents during TEM sample processing.

Retinal ganglion cell and nerve fiber layer analysis
Eyes were fixed in Davidson's fixative (2% formalin, 30% ethanol, 10% acetic acid in distilled water) for 1-2 days at +4 °C and processed for paraffin sectioning (5 µm sections from the center of the eye containing the optic nerve head) and standard hematoxylin and eosin (H&E) staining. Sections were scanned with Hamamatsu NanoZoomer S60 slide scanner with a 40× objective. The retinal nerve fiber layer (RNFL) thickness was measured with NDP.view 2 software starting at approximately 300 µm away from the optic nerve head for 500 µm length of both left and right side of the retina from 10 points on each side with regular intervals. These 20 measurements were then averaged to obtain a single value for each eye. Retinal ganglion cell (RGC) nuclei were calculated from the same area from both left and right side of the retina and averaged to obtain a single value for each eye.   Animation of consecutive z-stacks taken with a confocal microscope across a corneal limbus sample, which has +/Cre mTmG/+ been mounted between two glass slides. In the Angpt4 ; Rosa26 mice, cells that express of have expressed Cre Angpt4 are permanently GFP-labeled (green). CD31 antibody staining (magenta) labels both the first appearing limbal vascular plexus and the later emerging Schlemm's canal (SC) endothelium (image plane moving from the + outer eye towards the inner eye). The GFP cells do not overlap with the SC endothelium since there is minimal amount of white signal which would indicate merging of green GFP and magenta CD31 signals; however, the + GFP cells locate closely to the SC, and also on the stroma nearer the outer eye.