Systemic hypertension is not protective against chronic intraocular pressure elevation in a rodent model

High intraocular pressure is the most well documented glaucoma risk factor; however many patients develop and/or show progression of glaucoma in its absence. It is now thought that in some instances, ocular perfusion pressure (blood pressure – intraocular pressure) may be as important as intraocular pressure alone. Thus, systemic hypertension would be protective against glaucoma. Epidemiological studies, however, are inconclusive. One theory of why hypertension may not protect against elevated intraocular pressure in spite of increasing ocular perfusion pressure is that with time, morphological changes to the vasculature and autoregulatory failure outweigh the benefits of improved perfusion pressure, ultimately leading to poor retinal and optic nerve head blood supply. In this study we showed the presence of increased wall:lumen ratio and wall area of the ophthalmic artery in rats with chronic hypertension in addition to failure of retinal autoregulation in response to acute modification of ocular perfusion pressure. Subsequently we found that in spite of dramatically increasing ocular perfusion pressure, chronic systemic hypertension failed to protect retinal structure and function from a rodent model of glaucoma.


S1 Blood pressure profile in subcutaneous ANG II infusion model of hypertension
: Blood pressure profile. Effect of chronic ANG II infusion on SBP (mean ± SEM) over 12 weeks in normotensive (blue symbols) and hypertensive (red symbols) animals. (SBP: systolic blood pressure, IOP: intraocular pressure).
A two-way repeated measures ANOVA found no significant interaction between time and group for either hypertensive or normotensive animals (both p = 0.5), which we interpret to mean that the different procedures performed in Group-1 and Group-2 did not significantly affect BP.   Figure S2C there was no significant interaction (p = 0.4) or BP (p = 0.2) effect on bipolar cell amplitude (Vmax). Figure S2D shows that was no significant interaction (p = 0.3) or BP effect (p = 0.4) on ganglion cell function (pSTR). For all three parameters, there was a significant time effect (p<0.001 for all three parameters) which may be due to the effects of ageing and repeated general anaesthesia. Figure S3: Effect of chronic hypertension on ERG waveform timing in ANG II treated (red bars) and saline control (blue bars) animals. A: Photoreceptoral sensitivity (s). B: Isolated P2 peak time. C: pSTR peak time. The kinetics of the ERG waveform were also not significantly altered by chronic systemic hypertension ( Figure S3). As shown in Figure S3A, there was no significant interaction (p = 0. 4) or blood pressure (p = 0.4) effect on the sensitivity of the of photoreceptoral response, which is reflective of the slope of the fast component of the P3 1-3 . There was no significant interaction (p = 0.8) or blood pressure (p = 0.6) effect on the peak time of the isolated P2 after filtering of the oscillatory potentials ( Figure S3B). Similarly, for the pSTR response, shown in Figure S3C, there was no significant interaction (p >0.9) or blood pressure effect (p = 0.1). Figure S4: Effect of chronic hypertension on oscillatory potential responses. A: Oscillatory potential peak amplitude. B: Representative OP waveforms from one hypertensive (red trace) and normotensive (blue trace) animal at week 12. C: Oscillatory potential peak time. D: Correlation between BP integral over 12 weeks and OP peak amplitude (% change from baseline) found at week 12. The solid black line represents the Deming regression (Y = -0.094X + 148.6, rs = -0.50, p = 0.04). Figure S4A shows that there was a significant decline in the OP peak amplitude in hypertensive rats (p = 0.02), whereas OP amplitude in normotensive animals did not change appreciably over the 12 weeks. Post hoc analysis revealed that OP amplitude was significantly reduced at weeks 8  Figure S4C). Representative waveforms collected at week 12 are shown in Figure   S4B. Relative change in OP amplitude was significantly correlated with the BP integral across the 12 weeks (p = 0.04, Figure S4D).

S3 Effect of 12 weeks of chronic systemic hypertension on oscillatory potential (OP) amplitudes
Oscillatory potentials are considered to be reflective of the integrity of inner retinal circulation [4][5][6][7] and have been shown to be altered in human studies in hypertension also [8][9][10] . Our findings in the rat are consistent with these human studies. Figure S5: Experimental timeline. In both groups, BP was measured daily for 5 days to establish baseline. After BP elevation, measurements were taken weekly. Dashed red arrows indicate osmotic minipump implantation for BP elevation at weeks 0, 4 and 8. Animals in Group-1 underwent acute blood pressure manipulation at week 12 to assess autoregulatory capacity (blue dashed arrow). All animals underwent cardiac perfusion at the end of experiments prior to tissue collection. Animals in Group-2 underwent circumlimbal suture implantation to elevate IOP at week 4. The circumlimbal suture was left in place for the next 8 weeks. Electroretinography and optical coherence tomography imaging was performed every four weeks in Group-2 (blue dashed arrows). (BP: blood pressure, IOP: intraocular pressure, ERG: electroretinogram, OCT: optical coherence tomography).

S5 Procedure for Alzet Osmotic Pump implantation
Under isoflurane anaesthesia (3% induction, 1.5% maintenance, 1.5 L min -1 ) and aseptic conditions, a 1 cm incision was made in the skin between the shoulder blades, and a subcutaneous pocket created by blunt dissection. The pump was inserted into the pocket, and the skin closed using interrupted sutures. The 2ML4 pump reservoir holds 2 mL of preloaded solution and takes four weeks to release its contents. Therefore, a new pump was implanted and the old pump explanted at weeks four and eight. CA, USA). Each stack of 250 images was then registered and averaged using ImageJ software.

S6
As the Long Evans rat heart rate is approximately 350 beats per minute 11 , the image stack encompasses one full cardiac cycle, so any confounding effects of variation in arteriole diameter within the cardiac cycle are removed. Between 30 to 50 image stacks were captured for each animal, and after registration and averaging of each stack, the average images were then registered once more, to allow arteriole diameter to be measured in the same location throughout the BP manipulation procedure. Three to four first order arterioles approximately one disc diameter superior to the optic nerve head were imaged and measured in each animal. The same area was imaged in each animal to minimise any confounding effect of regional variation in blood flow and vessel diameter throughout the retina.
Arteriole diameter was measured using the Diameter plugin in ImageJ software 12 , which measures the width of the vessel by plotting the image intensity profile along a line perpendicular to the vessel. The distance between the half decay and half rise of the intensity profile is taken as the vessel width. The plugin measures the diameter along the line selected by the masked experimenter by re-sampling five times at parallel one pixel intervals. This process was repeated five times for each image to produce a total of 25 diameter measurements for each vessel at each time point, which were then averaged.

S9 Electroretinogram (ERG) stimulus characteristics and procedure for signal acquisition and analysis
Animals were dark adapted overnight and light exposure during set-up was minimised to achieve optimal conditions for eliciting the ganglion cell specific scotopic threshold response 16 . Animals were anaesthetised using intramuscular ketamine: xylazine (60:5 mg kg -1 ). Topical anaesthesia of the cornea (0.5% proxymetacaine) and pupil mydriasis (0.5% tropicamide, Alcon Laboratories, Frenchs Forest, NSW, Australia) was also performed.  Hz, -3 dB) was used to extract the OPs, revealing the isolated P2 amplitude, reflective of ONbipolar cell activity [24][25][26] . The relationship between stimulus energy and P2 amplitude was modelled using a saturating hyperbolic function, which returns a maximal response amplitude (Vmax), reflective of bipolar cell function. A range of 18 stimulus energies between -6.35 and 2.07 log cd.s.m -2 were used to model the P2 response to extract Vmax [27][28][29] . A twin flash paradigm at 2.07 log cd.s.m -2 was employed to extract the isolated rod contribution to the P2 waveform [30][31][32] . Retinal ganglion cell function was assayed using the positive scotopic threshold response (pSTR), measured using very dim stimulus energies [33][34][35] . In this study, the amplitude at 110 ms after stimulus onset in response to a -5.31 log cd s m -2 flash was used to indicate ganglion cell function.

S10 Procedure for optical coherence tomography (OCT) signal acquisition and analysis
Animals were placed on an alignment stage and Genteal Gel (Novartis, North Ryde, NSW, Australia) was applied to the eye as a coupling interface between the eye and the rat-specific