Homocysteine directly interacts and activates the angiotensin II type I receptor to aggravate vascular injury

Hyperhomocysteinemia (HHcy) is a risk factor for various cardiovascular diseases. However, the mechanism underlying HHcy-aggravated vascular injury remains unclear. Here we show that the aggravation of abdominal aortic aneurysm by HHcy is abolished in mice with genetic deletion of the angiotensin II type 1 (AT1) receptor and in mice treated with an AT1 blocker. We find that homocysteine directly activates AT1 receptor signalling. Homocysteine displaces angiotensin II and limits its binding to AT1 receptor. Bioluminescence resonance energy transfer analysis reveals distinct conformational changes of AT1 receptor upon binding to angiotensin II and homocysteine. Molecular dynamics and site-directed mutagenesis experiments suggest that homocysteine regulates the conformation of the AT1 receptor both orthosterically and allosterically by forming a salt bridge and a disulfide bond with its Arg167 and Cys289 residues, respectively. Together, these findings suggest that strategies aimed at blocking the AT1 receptor may mitigate HHcy-associated aneurysmal vascular injuries.

4 Telmisartan was used in most experiments to block AT1R, with a single exception that candersartan was used without providing rationale why either was used. More importantly, it is well-known that telmisartan has strong PPARgamma activation effect. Therefore, data interpretation should at least discuss the potential off-target effects. Figure 5: The authors performed ex vivo experiment and stated that AngII secretion from aortic ring explants was not altered by homocysteine within 12 hours, which does not rule out increase of AngII after 12 hours. Additionally, the authors measured mRNA of AGT, ACE, and AT1aR, but not renin, in the aortic explant. If renin is not present in the aortic ring, how was AngII produced?

Online
6. No immunofluorescent staining method is described. The specificity of all antibodies needs to be validated.
7. There is considerable skepticism about the accuracy of measuring angiotensin II. The authors have used this using a kit in which the reviewer can find no information. It is important to have a full description of this assay. For these measurements to be generally accepted, it would be helpful to provide validation studies.
8. It should be stated where the aortic rings are derived from. If these rings are derived from regions that are aneurysm-resistant, the relevance of findings in these tissues must be discussed. 9. On page 30 the last paragraph, the authors state that "Therefore, in addition to lowering total plasma Hcy, these patients may benefit from using AT1 receptor blockers rather than ACEIs, as our data showed that enalapril does not inhibit Hcy-induced AT1 receptor activation, while telmisartan does." There has no data to support this conclusion.

Reviewer #3 (Remarks to the Author):
This is a review of the computational portions of the paper. The authors used small molecule docking with AutoDock 4.2 to identify potential homocysteine binding sites on AT1 using both global docking and local refinement. AT1 is a membrane-embedded GPCR, so docking to lipidfacing regions is unusual, and any predicted conformations in the lipid bilayer region would be suspect. Fortunately, only cluster 1 looks to suffer from this problem. The most interesting cluster (cluster 3) was in the solvent-facing surface of AT1, so this is a reasonable prediction. The followup experiments support the binding location and the identification of Arg167 as a key binding residue. Additionally, MD simulations of the HCY at cluster 3 were stable.
The authors also suggest that HCY may make a disulfide bond with C289. In Fig 5J, these atoms are not close enough to bond, but it is conceivable that the HCY position would be dynamic enough to allow bonding.
Overall, this paper uses fairly standard computational methods in reasonable manners to support their stated conclusions.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): Using mouse aortic aneurysm models, the authors demonstrate that hyperhomocysteinemia (induced by supplementation of drinking water with homocysteine) aggravates AAA formation through a mechanism that is dependent on AT1. The in vivo parts of the manuscript are convincing and mechanistic, although somewhat incremental.

Response: Many thanks for the reviewer's positive comments.
For the radioactive ligand binding studies, were non-transfected HEK293 cells (e.g. cells that do not express AT1) used to control for non-specific binding? This might be particularly important to confirm that homocysteine actually has both orthosteric and allosteric effects on Ang II binding.
Response: Following reviewer's suggestion, we performed the radioactive ligand binding assay using non-transfected HEK293 cells as non-specific binding controls. We subtracted the non-specific bindings from the corresponding binding of AT1overexpressing HEK293 cells and recalculated K i and K D respectively. The final results were shown in Figure 3A-B, 5D and 5I in the revised manuscript. Our new data reinforced the notion that homocysteine has both orthosteric and allosteric effects on Ang II binding.
There are discrepancies between the Methods (page 7), the Results (page 14), and Online Figure 1 with regard to the age of the mice that were given homocysteine-water (8 weeks or 12 weeks?).

Response:
We are sorry about the editing error in the previous version of our manuscript. The mice were given homocysteine-water at 8 weeks of age. We modified the age of mice in the part of "Animal Treatments" on page 26 of the Methods section in the revised manuscript.
How was plasma homocysteine measured, and does the assay distinguish between free thiol and disulfide forms of homocysteine?  Figure 3C and paragraph 2 on page 12 of revise manuscript.
The conclusion that Arg167 is a homocysteine binding site is not adequately supported by the data. The R167A mutation had a minimal effect on Hcy binding ( Figure 5D).
Furthermore, since the mutation caused complete loss of AT1 signaling to both Hcy and Ang II, it is not correct to conclude that it is a specific Hcy binding site. inhibited in AT1 -/mice or application of AT1 blocker (Figure 1 and Online Figure 2-3), our results suggested that HHcy promotes AAA through direct AT1aR stimulation.

Response
Please refer to revised Figure 3C and paragraph 2 on page 12 of revise manuscript.
Comments 1. The major conclusion of this manuscript is that AT1a receptors are stimulated independent of angiotensin II to promote AAA. For this statement to be included, there must be some study in which AAA are formed in mice that have had a manipulation to reduce the production of angiotensin II. For example, administration with an ACE inhibitor.

Response:
Following with reviewer's helpful suggestion, we applied ACEI enalapril to exclude the role of Ang II production involved in HHcy-aggravated AAA formation in vivo. We compared HHcy-exacerbated AAA formation with and without administration of enalapril in both elastase-and CaPO4-induced aneurysmal models (Online Figure   12A). As expected, enalapril administration significantly decreased the blood pressure in both control and HHcy mice (Online Table 6). In contrast, administration of enalapril displayed no effect on HHcy-enhanced maximal aortic diameter enlargement in both aneurysmal models (Online Figure 12B-E), indicating that HHcy-aggravated AAA formation is independent on Ang II production. Please refer to line 11-18 on page 10 of revised manuscript.
2. In figure 1, data represented in figure 1B is that maximal aortic diameters were approximately 1 mm. Most measurements in this region for normal aorta would be in the 0.4 to 0.5 mm range. would be helpful to provide the starting size of each of these aortas. Based on these measurements, the infra renal aorta of all 4 groups is greatly expanded, with augmentation in the Hhcy group that are wild type fo At1a receptors.
However, Figure 1C and D seems to indicate there are minimal elastin breaks in 3 groups. How can this be reconciled? Therefore, data interpretation should at least discuss the potential off-target effects.

Response:
We agree with reviewer's concern for this important point. To evaluate whether other sartans had the similar functions of telmisartan on inhibition of Hcyinduced AT1 activation, candesartan and losartan were applied. As a result, both sartans markedly blocked Hcy-induced PKC and ERK1/2 phosphorylation and NFAT activation similar to telmisartan (Online Figure 10C-D). Thus, these sartans seems have the similar effects on Hcy-induced AT1 activation.
As reviewer suggested that telmisartan has strong PPARγ activation effect (Cardiovasc Res 2011;90(1):122-129), we applied PPARγ agonist rosiglitazone (RSG) to examine whether it can inhibit the activation of AT1 receptor by Hcy. Distinct with telmisartan, RSG did not inhibit Hcy-induced PKC and ERK1/2-MAPK phosphorylation and NFAT activation, implying that the inhibitive effect of telmisartan is independent on its side-effect on activating PPARγ (Online Figure 10A-B). Please refer to the text from line 18 on page 9 to line 3 on page 10 of revised manuscript. Figure 5: The authors performed ex vivo experiment and stated that AngII secretion from aortic ring explants was not altered by homocysteine within 12 hours, which does not rule out increase of AngII after 12 hours. Additionally, the authors measured mRNA of AGT, ACE, and AT1aR, but not renin, in the aortic explant. If renin is not present in the aortic ring, how was AngII produced?  Figure 5D). Accordingly, Ang II secretion of aortic ring explants was not altered by Hcy within 24 hours but significantly elevated under 48-hour treatment of Hcy (Online Figure 5E).