Genetic and Functional Characterization of ANGPTL7 as a Therapeutic Target for Glaucoma


 Glaucoma is a leading cause of blindness. Current glaucoma medications work by lowering intraocular pressure (IOP), a risk factor for glaucoma, but most treatments do not directly target the pathological changes leading to increased IOP, which can manifest as medication resistance as disease progresses. To identify physiological modulators of IOP, we performed genome- and exome-wide association analysis in >129,000 individuals with IOP measurements and extended these findings to an analysis of glaucoma risk. We report the identification and functional characterization of rare coding variants (including loss-of-function variants) in ANGPTL7 associated with reduction in IOP and glaucoma protection. We validated the human genetics findings in mice by establishing that Angptl7 knockout mice have lower (~2 mmHg) basal IOP compared to wild-type, with a trend towards lower IOP also in heterozygotes. Conversely, increasing mAngptl7 levels via injection into mouse eyes increases the IOP. We also show that acute gene silencing via siRNA knockdown of Angptl7 in adult mice lowers the IOP (~2-4 mmHg), reproducing the observations in knockout mice. Collectively, our data suggest that ANGPTL7 is important for IOP homeostasis and is amenable to therapeutic modulation to help maintain a healthy IOP that can prevent onset or slow the progression of glaucoma.


INTRODUCTION 62
Glaucoma is a leading cause of irreversible blindness, with a global prevalence of 3.54% 63 in individuals 40-80 years of age, and is projected to affect more than 111.8 million people by 64 2040 (1). Classified as a neurodegenerative disease, glaucoma is characterized by the progressive 65 loss of retinal ganglion cells in the eye and thinning of the neuroretinal rim of the optic nerve head. 66

148
In summary, we observed a significant association of Gln175His in ANGPTL7 with 149 reduced IOP and a sub-threshold association, in the same direction and of similar magnitude, with 150 pLOF variants in ANGPTL7. Assuming that the pLOF variants indeed cause a loss of protein 151 function, our data suggests that loss of ANGPTL7 can lead to lower IOP. 152 153 IOP-associated variants in ANGPTL7 are protective against glaucoma 154 To understand if carriers of variants in ANGPTL7 would also be protected against 155 glaucoma, we performed an association analysis of Gln175His with glaucoma in UKB, GHS Estonia (EstBB), the HUNT study from Nord-Trøndelag, Norway (HUNT), and the Copenhagen 160 General Population Study/Copenhagen City Heart Study from Copenhagen, Denmark (CGPS-161 CCHS). A meta-analysis across these eight cohorts showed a significant reduction in glaucoma 162 risk for Gln175His carriers (odds ratio (OR allelic ) = 0.77, P = 2.7 × 10 -6 , Figure 3A). We also 163 analyzed glaucoma risk in carriers of the rarer Arg177*/Trp188* variants in a cross-ancestry meta-164 analysis and observed a consistent trend towards reduction in risk (OR allelic = 0.87, P = 4.1 × 10 -1 , 165 Figure 3B). Taken together, the associations of missense and pLOF variants in ANGPTL7 with 166 reduced IOP and the association of the missense variant with reduced glaucoma risk suggest the 167 hypothesis that loss of ANGPTL7 confers protection against glaucoma, and that this effect is 168 mediated through the regulation of IOP. 169

ANGPTL7 variants are associated with corneal measures 171
We performed a phenome-wide association analysis (PheWAS) to understand whether 172 other traits were associated with a burden of pLOF and deleterious missense variants in ANGPTL7. 173 We tested the ANGPTL7 variant aggregate for association with 14,050 and 10,032 binary and 174 quantitative traits in UKB and GHS, respectively. No associations reached phenome-wide 175 significance (P < 2 × 10 -6 after multiple-testing correction for 24,082 total traits) in GHS. The only 176 significant associations in UKB were with ocular traits (Table 1), specifically with decreased 177 IOPcc, decreased corneal resistance factor (CRF) and increased corneal refractive power along 178 both weak and strong meridians measured at 3 and 6mm diameters. The effect of these variants on 9 IOPcc was slightly attenuated (-0.17 SD) compared to that on IOPg (-0.22 SD in UKB), which 180 suggests that ANGPTL7 has some impact on corneal properties that are known to affect the IOPg 181 measurements (20). The association observed with decreased CRF is also consistent with a corneal 182 effect of ANGPTL7. 183

184
We used the autorefraction measurements at 3mm diameter to derive measures of clinical 185 interest, namely, mean corneal refractive power (mCRP), corneal astigmatism, and refractive 186 astigmatism (Supplementary Methods) and checked for association with ANGPTL7. We observed 187 a significant association with increased mCRP (beta allelic = 0.16, P = 1.1 × 10 -13 , Table 1) but no 188 association with corneal or refractive astigmatism. We also did not observe associations with mean 189 spherical equivalent (MSE; measure of refractive error) or myopia (either derived from MSE or 190 via ICD-10 diagnosis), which could result from increased mCRP (Table S2). Overall, our PheWAS 191 results show that while ANGPTL7 is associated with changes in corneal anatomy/biomechanics-192 related quantitative measures, we did not detect an increased risk for any related disease outcomes 193 that we could test. In addition, pLOF and deleterious missense variants in ANGPTL7 are not 194 associated with any systemic quantitative traits or binary outcomes. To understand the impact of Gln175His, Arg177*, and Trp188* variants on the expression 198 and secretion of ANGPTL7, we transiently transfected constructs expressing the wild type (WT), 199 Gln175His, Arg177*, and Trp188* variant proteins in HEK293 cells. We measured mRNA levels 200 by Taqman, which showed similar Gln175His and Arg177* transcript levels and a trend towards 201 decreased levels of the Trp188* transcript compared to WT (P = 0.08; Figure S5). However, 202 analysis of intracellular, steady-state protein in whole-cell lysate by western blotting and ELISA 203 revealed increased levels of Gln175His compared to WT (P < 1 ´ 10 -4; Figure 4C). As expected, 204 Arg177* and Trp188* encoded lower molecular weight proteins (~30-32 kDa). No significant 205 difference in the protein levels of these two mutants was revealed by ELISA in comparison to WT 206 ( Figure 4A, C). Because ANGPTL7 is a secreted protein, we next determined the levels of WT, 207 Gln175His, Arg177*, and Trp188* in the cellular supernatant. Protein analysis by Western blot 208 showed that the Arg177* and Trp188* variants were not detectable and the Gln175His was 209 drastically reduced in the supernatant compared to WT (P < 0.05; Figure 4B). ELISA assay further 210 corroborated the severely reduced levels of Gln175His and the inability of Arg177* and Trp188* 211 to reach the extracellular space ( Figure 4C

Angptl7 KO mice have lower basal IOP than WT 236
We generated and characterized Angptl7 -/-(KO) mice. We did not observe any ocular 237 changes on anterior segment optical coherence tomography (OCT), or a difference in corneal 238 thickness between the two genotypes ( Figure S6). We also monitored the IOP in KO, Angptl7 +/-239 (Het) and WT mice and observed a dose-dependent decrease in IOP across the three genotypes 240 ( Figure 7A). The mean IOP was lowered in KO mice (mean ± standard error of the mean (SEM): 241 15.39 ± 0.25 mmHg) by 11% (1.96 mmHg, P < 1× 10 -4 ) compared to WT (17.36 ± 0.23 mmHg). 242 Het mice (16.26 ± 0.43 mmHg) showed a smaller (6%, 1.1 mmHg, P = 0.02) but significant 243 reduction in IOP compared to WT. We confirmed via RNAscope that Angptl7 mRNA was not 244 expressed in any ocular tissue in KO mice whereas it was expressed in TM, cornea, and sclera of 245 WT mice ( Figure S7; Supplementary Methods). 246 247

siRNA-induced knockdown of Angptl7 mRNA and lowering of IOP in WT Mice 248
To investigate whether knockdown of Angptl7 with small interfering RNA (siRNA) can 249 also lower IOP, we tested six different siRNAs targeting Angptl7 in C57BL/6J mice and monitored 250 IOP over time. We injected C57BL/6J mice intravitreally with 15µg of siRNAs and performed 251 qPCR six weeks later on limbal rings dissected from mouse eyes enriched for the TM. IOP was 252 significantly lowered 2 weeks post-injection in mice treated with two of the six siRNAs (siRNA 253 #3 and #5) compared to the PBS and Naïve (no injection) groups ( Figure 7B). Naïve and PBS-254 treated animals maintained their IOPs at baseline for the duration of the study (weeks 0-6). In mice 255 treated with siRNA#3 and #5, IOP was lowered by 2-4 mmHg starting at week 2 compared to 256 PBS-treated mice ( Figure 7B). At the end of the study, we collected the eyes, carefully micro-257 dissected the limbal ring and performed qPCR. We observed the highest level of knockdown 258 (>50%) of Angptl7 mRNA with siRNAs #3 and #5 compared to PBS-treated mice, which is 259 consistent with the IOP lowering observed in mice injected with these two siRNAs ( Figure 7C). 260 These results suggest that acute inhibition of Angptl7 expression also lowers IOP. 261

DISCUSSION 262
In this study, we present genetic and functional evidence for a role for ANGPTL7 in the 263 physiological control of IOP and as a potential target for glaucoma therapy. Through genetic 264 association analyses across 8 cohorts in Europeans, we confirmed the association of a rare 265 missense variant, Gln175His (rs28991009), in ANGPTL7 with a decrease in IOP and with 266 decreased risk for glaucoma (15,19). We further identified pLOF variants in ANGPTL7, Arg177* 267 (rs143435072) and African ancestry-enriched Trp188* (rs145750805), that also associated with a 268 decrease in IOP, suggesting that Gln175His carriers are protected from glaucoma through a loss 269 or reduction in ANGPTL7 activity. Through cell-based expression assays, we found that 270 Gln175His, Arg177*, and Trp188* were severely defective in secretion when compared to wild-271 type and, while not proof, this observation is consistent with the hypothesis that they result in a 272 loss of protein function. We also identified predicted-deleterious ANGPTL7 variants in burden 273 analyses associated with reduced IOP, indicating that there may be other ultra-rare ANGPTL7 274 variants that confer protection from glaucoma. Supporting this hypothesis, a recent report 275 The above studies suggest a strong correlation of ANGPTL7 expression with glaucoma 297 disease-state, however, they are not evidence for a causal relationship between ANGPTL7 levels 298 and elevated IOP/glaucoma. Human genetics finding of an association between ANGPTL7 loss 299 and lower IOP suggests that ANGPTL7 is physiologically important for IOP regulation. We 300 validated this hypothesis in mice where we performed reciprocal experiments: measuring IOP after 301 increasing ANGPTL7 levels via injection of mAngptl7 into mouse eyes, and after removing all 302 mAngptl7 protein by generating Angtpl7 KO mice. Our findings show that increasing mAngptl7 303 results in increased (~2-4 mmHg) IOP and decreasing mAngptl7 through KO mice reduces basal 304 IOP levels (~2 mmHg), establishing that ANGPTL7 functions in vivo to maintain IOP 305 homeostasis. We further recapitulated the reduction in IOP observed in KO mice by injecting WT 306 mouse eyes with siRNA against mAngptl7, which not only replicates the observation in genetic 307 15 mutant mice but also illustrates that the effect of mAngptl7 on IOP continues post-development 308 and is amenable to modulation by therapeutics in adulthood. Our results of lower IOP in Angptl7 309 KO mice are highly consistent with observations from human genetics. Based on this, we could 310 extend the siRNA knockdown findings to humans and surmise that inhibition of ANGPTL7 in 311 adulthood could be an efficacious way to lower IOP and, eventually, the risk for glaucoma. As all studies, this study has limitations. First, we did not account for IOP-lowering 345 medication use in the IOP analysis. We did exclude from the IOP analysis individuals with a 346 glaucoma diagnosis, greatly reducing the effect of medication use, however this would not exclude 347 those individuals on IOP-lowering medications without a glaucoma diagnosis. In addition, the IOP 348 measurement in UKB for most participants was only taken once, and therefore can be prone to 349 errors that a median of multiple measurements might buffer against. Both of the above factors 350 would influence the estimation of the effect of rare genetic variants on IOP. Second, since a 351 glaucoma diagnosis can often be made well after the onset of disease, we can expect that a certain 352 percentage of controls have undiagnosed glaucoma, which may affect the effect size estimate. 353 Third, the ANGPTL7 gene in humans lies within intron 28 of MTOR (39), therefore variants in 354 ANGPTL7 could also affect MTOR function. While a possibility, it is unlikely that the effect on 355 IOP and glaucoma is due to MTOR, as: (I) the variants are predicted to be protein-altering in 356 ANGPTL7 whereas there is no obvious predicted functional effect on MTOR; (ii) we show data 357 suggesting that the variants have a functional impact on ANGPTL7; and (iii) we show mouse data 358 that establish a role for ANGPTL7 in IOP regulation. 359

360
In summary, our genetic and pharmacological results indicate that ANGPTL7 participates 361 in the normal physiological regulation of IOP in humans and mice. Since excessive amounts of 362 ANGPTL7 protein in the eyes of experimental animals cause IOP to elevate to pathological levels, 363 upregulation of ANGPTL7 in humans may be responsible for the elevated IOP that leads to POAG. 364 Therefore, we propose ANGPTL7 as an excellent candidate to explore as a therapeutic target for 365 POAG. which has been the gold standard for measuring IOP, while IOPcc provides a measure of IOP that 397 is adjusted to remove the influence of corneal biomechanics (47). For this study, we focused on 398 IOPg as this measurement is the most comparable to IOP measurements in other cohorts, and 399 herein IOPg will be referred to as IOP. IOP in POAAGG was measured using a GAT. In GHS, 400 IOP measurements were obtained from several instruments including GAT, analyses of IOP, we excluded individuals with: 1) a glaucoma diagnosis; 2) IOP measures that 407 were more than 5 standard deviations away from the mean; 3) more than a 10-mmHg difference 408 between both eyes. We derived a mean IOP measure between both eyes for each individual. IOP 409 of only one eye was used in instances where IOP measures for both eyes were not available.         The Primary Open-Angle African American Glaucoma Genetics Study: Baseline Demographics, 606 Ophthalmology 122, 711-720 (2015).