N-homocysteinylation of β-arrestins biases GPCR signaling and promotes platelet activation Free

Homocysteine (Hcy) is a sulfur-containing, intermediary amino acid involved in methionine metabolism.¹ Hyperhomocysteinemia (HHcy; circulating Hcy ≥15 μM) has been identified as an independent risk factor for a variety of cardiovascular diseases (CVDs),^2-5 and a growing body of evidence has established that HHcy independently contributes to thrombosis, particularly venous thromboembolisms.^6,7 Approximately 13% to 47% of patients with HHcy develop atherothrombotic vascular disease.⁶ Platelets are critical functional components of the cardiovascular system and are involved primarily in the coagulation process. Platelet hyperactivity promotes the occurrence and development of thrombotic diseases. Studies have shown that platelets isolated from the blood of patients with arterial thrombosis and HHcy are significantly more sensitive to agonists than platelets from healthy individuals.^8,9 The mechanisms by which HHcy causes these effects remain unclear. Although recent research has revealed that metabolic reprogramming,¹⁰ oxidative stress,¹¹ and inflammasome activation¹² are involved in the aggregation of pathophysiological processes mediated by HHcy, a full understanding of the pathogenic mechanism of HHcy has not been established, mostly because of the involvement of HHcy in the posttranslational modification (PTM) of numerous protein targets.¹³ 

As key modulators of the cardiovascular physiological system, G protein-coupled receptors (GPCRs) are inappropriately expressed or their functions are dysregulated in many CVDs; thus, GPCRs serve as important drug targets.¹⁴ In particular, the hyperactivation of GPCRs, such as protease-activated receptor 1/4 (PAR1/4), thromboxane A2 (TXA2) receptor (TP), or the purinergic receptors P2Y purinoceptor 1 (P2Y1) and P2Y purinoceptor 12 (P2Y12) can cause excessive platelet activation, thereby accelerating thrombosis progression.^15,16 Nevertheless, a systematic study of how HHcy affects a subset of GPCRs that modulate the same physiological or pathophysiological process in the cardiovascular system has not been carried out.

In Hcy-induced PTMs, N-homocysteinylation is considered a pathologic PTM.^13,17-21 Previous studies have shown that other diverse PTMs actively participate in the fine tuning of GPCR signaling and functions through the direct modulation of the activities of GPCRs or downstream effectors, such as G proteins and β-arrestins.^22-36 We speculated that HHcy may aggravate vascular diseases by affecting GPCR signaling, which is mediated by N-homocysteinylation. In this study, we noted that GPCRs amplified platelet activation under HHcy conditions by increasing the activity of Gi or Gq signaling. Further mechanistic studies revealed that N-homocysteinylation at K294/K296 within murine β-arrestin1/2 (m-β-arrestin1/2) inhibited β-arrestin recruitment to activated platelet GPCRs, thus blocking the desensitization of these receptors and ultimately leading to G protein overactivation and platelet hyperreactivity. Moreover, the switch to G protein–biased signaling by N-homocysteinylation at K294/K296 within m-β-arrestin1/2 is speculated to be a general mechanism after examination of a panel of GPCRs and was also found to participate in macrophage inflammation under HHcy conditions. By connecting HHcy to key components of GPCR signaling, our study revealed a previously uncharacterized mechanism of how HHcy affects the cardiovascular system.

The collection of human samples was approved by the Peking University Third Hospital Medical Science Research Ethics Committee (IRB00006761-M2020539) and abided by the principles of the Declaration of Helsinki. All animal studies were approved by the animal research ethics committees of the Peking University (DLASBD0157). See the supplemental Methods, available on the Blood website, for details.

For β-arrestin1/2 recruitment experiments, a bioluminescence resonance energy transfer (BRET) assay was performed using a Mithras LB940 microplate reader (Berthold Technologies) equipped with BRET filter sets, as previously described.³⁷ See the supplemental Methods for details.

GraphPad Prism version 8.0 was employed for statistical analyses, and all data were shown as the mean ± standard error of the mean. Statistical differences between groups were assessed using Student t tests and 1-way analysis of variance (ANOVA) with a Tukey test for multiple groups with only 1 variable. P values <.05 were considered significant, and significance was indicated as follows: ∗P < .05; ∗∗P < .01; ∗∗∗P < .001.

As an independent vascular risk factor, HHcy causes vascular injury and promotes thrombosis.⁶ Importantly, many soluble ligands cause platelet aggregation through the activation of a group of membrane receptors, that is, GPCRs (supplemental Figure 1A).^38,39 Platelets from patients with HHcy are more reactive to agonists of GPCRs,^8,40 but the underlying mechanism is unknown. Notably, Hcy can be metabolized to homocysteine thiolactone (HTL) by methionyl-tRNA synthetase (MARS; supplemental Figure 1B), which is readily detected in platelets (supplemental Figure 1C), and HTL can be linked to a protein via isopeptide bonds to lysine (Lys, K) residues to produce N-Hcy protein (supplemental Figure 1B).¹⁹ The aforementioned process is called N-homocysteinylation or K-Hcy modification. By using an anti–K-Hcy antibody, we found that the N-Hcy protein level was markedly increased in Hcy- or HTL-treated platelets (Figure 1A-B),¹⁸ suggesting that the effects of Hcy on platelets are caused by N-homocysteinylation.

Hcy and HTL alone did not induce platelet aggregation (supplemental Figure 1D), but preincubation with Hcy and HTL dose-dependently increased platelet aggregation in response to several GPCR agonists, and the half maximal effective concentration (EC₅₀) for Hcy and HTL in promoting platelet aggregation were close to 100 μM and 30 μM (Figure 1C-D; supplemental Figure 2A-B). Hcy and HTL also enhanced the production of active factors, including thromboxane B2 (TXB2), transforming growth factor-β (TGF-β1), and adenosine triphosphate, in platelets upon GPCR ligands stimulation (Figure 1E; supplemental Figure 1E-G). The effects induced by Hcy were specifically blocked by acetylhomocysteine thioether (AHT) or N-acetylcysteine (NAC), 2 MARS inhibitors that strongly inhibit N-homocysteinylation (Figure 1A,C,E).¹⁸ Notably, Hcy promoted platelet aggregation under both low-dose and high-dose agonist stimulation (supplemental Figure 2C-E), and intermediate to low doses of Hcy were also capable of augmenting platelet activation to some extent (supplemental Figure 2F-G).

We further assessed the role of HHcy in vivo using apolipoprotein E-deficient (Apoe^–/–) mice, a classic animal model of atherosclerosis.⁴¹ Optical coherence tomography angiography (OCTA) was explored to visualize and quantify mesenteric artery thromboses by tracking vascular diameter index changes, whereas Doppler ultrasonography was used to monitor the carotid artery occlusion times. We observed a significantly shortened tail bleeding time and accelerated arterial thrombus formation in both male and female HHcy mice, which were notably reversed by AHT (Figure 1F-I; supplemental Figure 1H-K). Moreover, platelets derived from patients with HHcy showed significantly higher reactivity than platelets from healthy individuals (Figure 1J-L). Collectively, these results suggest that a common mechanism that underlie the GPCR signaling pathway and N-homocysteinylation may contribute to the ability of HHcy to augment platelet activation both in vivo and ex vivo.

Both Gi- and Gq-mediated signaling contribute to platelet activation.³⁸ P2Y1 and TP are typical Gq-coupled receptors, whereas P2Y12 is Gi-coupled and PAR1 couples to both Gi and Gq (PAR4 can recruit Gq, and it is unclear whether PAR4 couples to Gi) (supplemental Figure 1A).¹⁵ The phosphorylation level at S473 of serine/threonine-protein kinase (AKT), which is a common downstream molecule of Gi, is significantly increased in platelets stimulated with Hcy or HTL (Figure 2A-B; supplemental Figure 3A-B). The adenine diphosphate (ADP)- or thrombin-induced decrease in cyclic adenosine monophosphate (cAMP) in platelets and HEK293 cells that overexpressed P2Y12 or PAR1 also significantly weakened in response to either HTL or Hcy incubation, indicating that the Gi signaling of P2Y12 or PAR1 also decreased (Figure 2C-D; supplemental Figure 3C-F).^34,42 In platelets, Gq activation mobilized intracellular Ca²⁺ and activated protein kinase C (PKC; evident by the phosphorylation of the substrate protein kinase D2 [PKD2] at position S916).⁴³ Importantly, the phosphorylation levels of PKD2 and the elevation of intracellular Ca²⁺ concentration induced by GPCR ligands significantly increased in platelets after Hcy and HTL treatment, indicating that pre-incubation with Hcy and HTL enhanced the activity of the platelet Gq pathway (Figure 2A-B, E-F; supplemental Figure 3A-B). Moreover, pretreatment with Hcy or HTL markedly increased Gq activation in response to ADP, U46619, or thrombin, as revealed by a transforming growth factor-α (TGFA) shedding assay in HEK293 cells overexpressing P2Y1, TP, or PAR1, respectively (supplemental Figure 3C, G-J).⁴⁴ Notably, the enhancement of Gq/Gi signaling by Hcy was reversed by both the MARS inhibitor NAC (2 mM) and AHT (2 mM) (Figure 2B-F; supplemental Figure 3A-B, E-F, H-J). Collectively, these results indicate that Hcy amplifies both the Gi and Gq signaling of several GPCRs in platelets through N-homocysteinylation.

Upon GPCR activation, not only G proteins but also β-arrestins are recruited to receptors.^30,37,45-48 β-arrestins are known to mediate GPCR internalization, desensitize G protein signaling, and carry out their own independent signaling.^49-53 We therefore assessed the effects of HHcy on β-arrestins.

By using Lck/yes-related novel tyrosine kinase-renilla luciferase (LYN-Rluc) as a plasma membrane marker⁵¹ (supplemental Figure 4A), we found that Hcy or HTL promoted the distribution of GPCRs across the plasma membrane, an effect that was inhibited by AHT or NAC (Figure 3A-B; supplemental Figure 4B-C). We also found that Hcy and HTL significantly reduced the recruitment of homo sapiens hs-β-arrestin1/2 to GPCRs in response to ADP, thrombin, or U46619 stimulation, an effect that was rescued by coincubation with AHT or NAC in a BRET assay (Figure 3C-F; supplemental Figure 4A, E-H).⁴⁹ These results indicate that N-homocysteinylation inhibits the functions of β-arrestin1/2.

We subsequently performed fluorescence colocalization analysis and observed that Hcy and HTL prevented hs-β-arrestins-mediated endocytosis of hs-P2Y12 (Figure 3G; supplemental Figure 4I). The efficiency of the activation of the Raf/MEK/ERK pathway downstream of hs-β-arrestins was also reduced in Hcy- or HTL-treated cells (supplemental Figure 4J-L), whereas AHT eliminated the effect of Hcy. Taken together, these findings suggest that N-homocysteinylation induced by Hcy or HTL inhibits the interaction between several platelet GPCRs (P2Y1, P2Y12, PAR1, and TP) and β-arrestin1/2, which may lead to the amplification of G protein signaling and the enhancement of platelet activation.

Arrestins are important signaling transducers downstream of many GPCRs.^48,51,54-56 We subsequently investigated whether the N-homocysteinylation of β-arrestin1/2 serves as a common mechanism by blocking interactions between arrestin and GPCR family members. In addition to platelet GPCRs, we further selected a panel of GPCRs that have important functions in the cardiovascular system, metabolism, and inflammation. Importantly, agonist-induced hs-β-arrestin1/2 recruitment to the corresponding receptors was significantly blocked by preincubation with HTL (Figure 3H; supplemental Figure 4M).

To investigate whether β-arrestin1 or β-arrestin2 plays a major role in the modulation of GPCR functions and the promotion of platelet aggregation induced by Hcy, we expressed ADP receptors in wild-type (WT) or hs-β-arrestin1/2 knockout HEK293 cells. The intracellular cAMP measurements and cAMP level in platelets revealed that the loss of β-arrestin2 significantly contributed to ADP-induced Gi signaling activation (Figure 4A-B), which indicated that β-arrestin2 was responsible for GPCR desensitization. In addition, platelet aggregation significantly increased in response to agonists, and the effects of HTL or Hcy treatment significantly decreased (Figure 4C-E) in platelets derived from Arrb2^–/– mice (supplemental Figure 5A), effects that may be a consequence of the decreased desensitization caused by m–β-arrestin2 deficiency. Therefore, these findings indicate that β-arrestin2 mainly mediates the effects of Hcy in vivo on agonist-induced platelet aggregation.

To provide further mechanistic insight into how Hcy or HTL treatment affects arrestin function, we used our recently developed alkynyl thioester probe AT-3⁵⁷ to enable the chemoselective labeling of N-Hcy proteins in platelets treated with Hcy or HTL (Figure 5A; supplemental Figure 6A).

We performed western blot analysis of N-Hcy proteins in platelets from healthy individuals and from patients with HHcy. Importantly, both hs-β-arrestin1 and hs-β-arrestin2 were detected in their N-homocysteinylated forms in patients with HHcy (Figure 5B). N-homocysteinylation was also detected in platelets from WT mice after Hcy/HTL administration (Figure 5C). To identify the specific HTL labeling sites on β-arrestin1/2, we incubated purified β-arrestin1/2 (β-arrestin1 from bovine and β-arrestin2 from rat) with HTL and identified the modified sites using liquid chromatography‒tandem mass spectrometry (LC-MS/MS) analysis. We revealed that K195 and K294 of β-arrestin1 (bovine) and K34, K296, and K314 of β-arrestin2 (rat) were N-homocysteinylated in vitro, whereas only K294 of β-arrestin1 and K296 of β-arrestin2 could be accurately detected in the MS/MS spectra because the m/z difference between y13⁺⁺ and y14⁺⁺ in β-arrestin1 and b8⁺ and b9⁺ in β-arrestin2 supported the expected modification of the LYS residue (Figure 5D-E; supplemental Figure 6B-D). The recruitment of arrestins by GPCRs is normally preceded by the phosphorylation of GPCRs by a group of GPCR kinases (supplemental Figure 6E).^{37,47,54,58-60} Interestingly, K294 of m-β-arrestin1 is located in the lariat loop and directly senses the phosphate of the receptor-phosphorylated C-tail, as demonstrated by the recently solved crystal and cryo-electron microscopy structures and by nuclear magnetic resonance studies (supplemental Figure 6F).^37,45,47,58 At the structural equivalent site of K294 of m-β-arrestin1 (Figure 5F), K296 of m-β-arrestin2 interacted with the atypical chemokine receptor 3 (CXCR7) phosphopeptide (C7pp; supplemental Figure 6G).^61,62 Because the interaction between the phosphorylated C-tail and arrestins plays key roles in arrestin recruitment by GPCRs, we speculated that the N-homocysteinylation of K294/296 within m-β-arrestin1/2 actively participates in the interactions between arrestins and GPCRs.

To investigate whether the N-homocysteinylation of the aforementioned sites by Hcy or HTL participated in the ligand-induced interaction between β-arrestin1/2 and GPCRs associated with platelet activation, we aligned the protein sequences of rat, human, and mouse β-arrestin2 with those of bovine, human, and mouse β-arrestin1 (Figure 6A; supplemental Figure 7A); in the figures, the N-homocysteinylation sites shown in the MS/MS spectra are marked in red. Notably, the K296 site of m-β-arrestin2 corresponds to K295 in hs-β-arrestin2. We subsequently evaluated the recruitment of β-arrestin1/2 to ligand-activated GPCRs using BRET assays. The results indicated that the designed LYS-to-arginine β-arrestin1/2 point mutations did not significantly change the structural integrity or recruitment of β-arrestin1/2 to these receptors under normal conditions (Figure 6B-G; supplemental Figure 7B-M).

Importantly, the inhibition of β-arrestin1/2 functions by preincubation with Hcy or HTL, including agonist-induced recruitment of β-arrestin1/2 to GPCRs (Figure 6B-G; supplemental Figure 7D-M) and GPCR endocytosis (Figure 6H-K; supplemental Figure 7N-Q), was markedly abolished by K294R or K295R mutations within hs-β-arrestin1/2, whereas K195R hs-β-arrestin1 and K34R/K313R hs-β-arrestin2 showed no difference from WT hs-β-arrestin1/2. Confocal immunofluorescence microscopy data confirmed that the N-homocysteinylation of K294/K295 within hs-β-arrestin1/2 suppressed the platelet GPCRs internalization mediated by β-arrestin1/2 (Figure 6L; supplemental Figure 7R).

Notably, ADP- or thrombin-induced cAMP reduction was amplified by Hcy or HTL in HEK293A cells transfected with WT hs-β-arrestin1/2 but not in cells transfected with K294R/K295R hs-β-arrestin1/2 mutants (Figure 6M-N; supplemental Figure 7S-T). We also performed a TGFA shedding assay and found that the K294R/K295R hs-β-arrestin1/2 mutation reversed the increase in Gq signaling induced by N-homocysteinylation (Figure 6O-Q; supplemental Figure 7U-W). Considering the desensitization roles of β-arrestin1/2, the augmentation of G protein signaling by Hcy or HTL downstream of these selected platelet GPCRs may be caused by the decreased desensitization functions of arrestins by the specific N-homocysteinylation of K294/K295 within hs-β-arrestin1/2.

To investigate whether the N-homocysteinylation of K296 in m-β-arrestin2 plays an important role in mediating the effects of Hcy on platelet functions in vivo, we generated m-β-arrestin2 K296R knock-in (KI) mice (Arrb2^K296R) using CRISPR-Cas9 gene editing (supplemental Figure 8A-D). Arrb2^K296R mice were further used for the assessment of platelet activity under Hcy or HTL stimulation, including the ex vivo experiments such as platelet aggregation assays and in vivo experiments involving the induction of FeCl₃-induced artery thrombosis (Figure 7A). Importantly, unlike platelets derived from WT mice, Hcy or HTL does not affect the aggregation rate of platelets derived from Arrb2^K296R mice under stimulation by platelet GPCR agonists, such as ADP, U46619, and thrombin (Figure 7B). We further investigated the impact of Arrb2^K296R on thrombosis in mice under conditions of HHcy. We induced arterial thrombosis in WT and Arrb2^K296R mice pretreated with Hcy (100 mg/kg) or vehicle. Notably, the Hcy-induced increase in carotid artery thrombosis and mesenteric arterial thrombosis was significantly blocked in the Arrb2^K296R mice when compared with their WT littermates (Figure 7C-E; supplemental Figure 8E). These results suggest that K296 is the key site for N-homocysteinylation within m-β-arrestin2 to mediate the effects of Hcy on the modulation of platelet function.

Previous studies have shown that mice with HHcy have increased plasma levels of inflammatory cytokines, such as tumor necrosis factor α (TNFA), interleukin 6 (IL-6), and IL-1β in the early stage.^63-66 HHcy also exacerbates macrophage inflammasome activation and M1 (proinflammatory phenotype) polarization, which play roles in the development of various CVDs.^65-68 Importantly, the activation of several GPCRs, including the ω-3 fatty acid receptor GPR120 and the bile acid receptor TGR5, is known to have therapeutic effects that antagonize inflammation.^69-72 We therefore investigated the effects of HHcy on the pharmacologic effects of different anti-inflammatory GPCRs.

We demonstrated that MARS was expressed in macrophages (supplemental Figure 1C) and that β-arrestin2 was N-homocysteinylated in peritoneal macrophages (PMs) (supplemental Figure 9A) after Hcy treatment; in addition, the N-homocysteinylation of K295 in hs-β-arrestin2 (corresponding to K296 in m-β-arrestin2) inhibited the interaction between GPR120 and β-arrestin2 under docosahexaenoic acid (C22:6n3; DHA) stimulation, as shown by the BRET assay results (supplemental Figure 9B). DHA is known to activate the GPR120-β-arrestin2 pathway to exert anti-inflammatory effects.⁶⁹ We found that in WT PMs, incubation with Hcy or HTL significantly inhibited the anti-inflammatory effects mediated by the DHA-induced activation of GPR120 (supplemental Figure 9C-D). However, β-arrestin2 deficiency not only prevented DHA from exerting its anti-inflammatory function but also abolished the proinflammatory effects of Hcy and HTL (supplemental Figure 9C-D). These findings suggest that β-arrestin2 may serve as a direct modification target for Hcy in the promotion of macrophage inflammation. Next, we reintroduced the expression of WT or K296R m-β-arrestin2 into PMs from Arrb2^–/– mice and then stimulated PMs with lipopolysaccharide and DHA with or without Hcy or HTL accordingly. The quantitative results for IL-6 and TNFA showed that the N-homocysteinylation of K296 within m-β-arrestin2 significantly weakened the anti-inflammatory effect mediated by GPR120 activation in response to DHA treatment, because the m-β-arrestin2 K296R mutation prevented Hcy and HTL from inhibiting the anti-inflammatory effects of DHA and amplified the inflammatory response in macrophages (supplemental Figure 9E).

INT-777 (also named the 6α-ethyl-23(S)-methyl-cholic acid), a Gs-biased ligand of TGR5,⁷³ is known to exert anti-inflammatory functions through the TGR5-Gs signaling pathway independently of β-arrestins.^70,72,74 Interestingly, unlike DHA, INT-777 had a pronounced anti-inflammatory effect on PMs under conditions of Hcy and HTL pretreatment. This occurred because, under the stimulation of INT-777, Hcy and HTL incubation failed to sharply increase the secretion levels of the inflammatory cytokine IL-6, suggesting that exogenous Hcy and its metabolite HTL do not significantly inhibit the anti-inflammatory function of INT-777 (supplemental Figure 9F). Thus, the Gs-biased ligand INT-777 has a stronger antagonistic effect against Hcy-induced inflammation exacerbation than DHA, which functions through the β-arrestin2 pathway, suggesting that G protein-biased ligands are more suitable for the treatment of patients with CVD with HHcy.

Approximately 5% to 7% of the general population suffer from HHcy, and HHcy has been identified as an independent risk factor for a variety of CVDs, especially thrombotic disorders.⁷⁵ The prevalence of HHcy in venous thrombus embolism (VTE) has been reported to be as high as 63%. Although supplementation with vitamins B6, B12, and folic acid has been shown to effectively lower plasma Hcy levels and prevent the occurrence of CVDs by promoting the conversion of Hcy to methionine or cysteine, several clinical trials have shown that administering folic acid or vitamins B6 and B12 does not produce significant therapeutic effects in patients with HHcy who already develop CVDs,^13,76,77 possibly because simply reducing the Hcy levels cannot eliminate the key pathogenic factors already induced by HHcy. Thus, identifying the common pathogenic mechanism of Hcy is crucial for developing new therapeutic strategies for HHcy. Interestingly, previous studies have shown that platelets from people with HHcy are more reactive in response to several GPCR agonists,^5,78 suggesting a strong connection between HHcy and GPCRs, which are the main regulators of the cardiovascular system in the clinic. Using platelets as the preliminary system, we revealed that HHcy leads to the N-homocysteinylation of β-arrestin1/2 in platelets, serving as an endogenous regulator of multiple GPCR signals to inhibit β-arrestin–mediated GPCR internalization and G protein signaling desensitization while biasing GPCR signaling via G proteins, consequently enhancing platelet reactivity and aggravating HHcy-related thrombosis. Furthermore, accumulating evidence indicates that HHcy may accelerate the progression of atherosclerosis or other vascular diseases through increased inflammation.^65,66 We further found that β-arrestin2 can be N-homocysteinylated in macrophages under Hcy stimulation, which suppresses the GPR120–β-arrestin2-mediated anti-inflammatory pathway and actively participates in the progression of atherosclerosis (AS), abdominal aortic aneurysm, and many other CVDs.

GPCRs located on macrophages mediate both proinflammatory and anti-inflammatory effects to adapt to changes in the environment and functions of macrophages.⁷⁹ For example, the IL-17A receptor and the proton-sensing receptor GPR132 induce the differentiation of proinflammatory macrophages,^79-83 whereas the IL-27 receptor, the omega-3 fatty acid receptor GPR120, and the bile acid receptor TGR5 inhibit the transformation of macrophages into an inflammatory phenotype.^69,70,79,84 The transition between proinflammatory and anti-inflammatory phenotypes in macrophages is highly complex, and the changes in signaling pathways and the underlying molecular mechanisms are not yet fully understood. Hcy has long been known to promote inflammatory responses in macrophages; however, previous studies on this pathologic phenomenon have yet to explain how Hcy disrupts the robust self-regulatory capacity of macrophages to lead to the polarization of inflammatory phenotypes. The results of this study revealed that β-arrestin N-homocysteinylation preferentially induces GPCR signaling pathways, causing imbalances in a series of GPCR signal transduction events that regulate macrophage phenotype switching. These findings led to the elucidation of the molecular mechanism through which Hcy promotes inflammation.

One important finding of this study is that HHcy could serve as a pathologically endogenous modulator of GPCR-biased signaling. G protein and β-arrestin signaling are 2 main downstream pathways that underlie the functions of GPCRs. Understanding the function of biased signaling and the development of biased ligands of GPCRs (either G or arrestin bias) has important therapeutic potential. For example, the Gi-biased opioid receptor agonist oliceridine has been approved for clinical use for postsurgical pain.^85-89 Herein, we determined that the N-homocysteinylation of β-arrestin1/2 by HHcy at the corresponding site K294/K296 could modulate the bias property of GPCR signaling by suppressing β-arrestin desensitization and, in turn, increasing G protein activation in a ligand-independent way (Figures 5 and 6; supplemental Figures 6 and 7). Therefore, this study provides important evidence to facilitate the evaluation of the therapeutic potential of prescription GPCR-targeted drugs for patients with HHcy, especially those drugs with G- or arrestin-biased properties.

Notably, the abilities of HHcy to increase platelet activity, thrombosis, and macrophage inflammation are not observed in Arrb2^K296R mice with impaired N-homocysteinylation of β-arrestin2 (Figures 7 and 8; supplemental Figure 8). Therefore, our data indicate that the imbalance between β-arrestin– and G protein–mediated GPCR signal transduction caused by N-homocysteinylation of m-β-arrestin2 at the K296 site may represent a previously uncharacterized pathogenic mechanism that underlie HHcy-related thrombosis and other cardiovascular diseases. Despite these useful mechanistic insights, our study has several limitations. Further verifying the inhibition of β-arrestins function by N-homocysteinylation and the functional modification sites of β-arrestins in platelets is challenging because the BRET assay, the most commonly used technique for detecting the interaction between GPCRs and β-arrestins, requires plasmid transfection into cells, followed by gene transcription and protein translation, which is difficult to achieve in platelets. Considering the low specificity of co-immunoprecipitation (Co-IP) technology (Co-IP cannot distinguish between direct and indirect protein interactions) and the lack of reliable anti-mouse PAR4, P2Y12, and TP antibodies, this method is also not suitable to verify the inhibitory effect of N-homocysteinylation on β-arrestins function and confirm the functional modification sites of β-arrestins at the platelet level in the current study. With the development of technology, we will continue to explore relevant research in the future.

N-homocysteinylation was first discovered in human serum albumin. N-homocysteinylation impairs the antioxidant capacity of albumin, promoting oxidative stress in the body.⁹⁰ In addition, apolipoprotein B of low-density lipoprotein (LDL) can be modified by HTL, thereby transforming LDL into N-Hcy-LDL, which is more readily taken up by vascular endothelial cells and macrophages and induces oxidative stress in endothelial cells and macrophages, facilitating the development and progression of AS.⁹⁰ Currently, the number of identified N-Hcy proteins is limited, and the pathogenic phenomena associated with HHcy cannot be fully explained. Moreover, most studies have not clarified the specific modification sites and their potential pathologic and physiological significance. Therefore, it is imperative to discover other N-Hcy proteins, understand the patterns of these modifications, and elucidate their involvement in the pathogenesis of HHcy.

Collectively, our findings reveal that HHcy-induced N-homocysteinylation of β-arrestins at specific sites downregulates β-arrestin function, enhances G protein signaling downstream of several GPCRs, and participates in both platelet-related and inflammatory processes. Pathologic conditions, such as HHcy, could have a broad impact on GPCR signaling systems and therefore could be important factors for the selection of therapeutics that target GPCRs.

This work was supported by the National Science Fund for Distinguished Young Scholars Grant (82425105 [J.-P.S.]), the National Natural Science Foundation of China (92168120 and 81974506 [L.T.]; 82204494 [Y.-Q.D.]), the New ComerstoneScience Foundation (J.-P.S.), and the Beijing Natural Science Foundation (Z200019 [J.-P.S. and L.T.]; 7232092 [Y.-Q.D.]).

Contribution: J.-P.S. and L.T. conceived and initiated the study; L.-Q.Z., J.-P.S., L.T., and X.W. designed most of the experiments; L.-Q.Z., C.-X.C., and Y.-Q.D. performed all the experiments and analyzed the data; L.-Q.Z., C.-X.C., Y.-Q.D., L.T., J.-P.S., and X.W. wrote the manuscript; C.W. designed and guided chemoselective labeling of the N-Hcy proteins; C.-Y.Z., Z.-Y.Z., and L.-L.H. guided platelet preparation and platelet activation detection; S.-M.H. and J.-L.W. contributed to β-arrestin2 purification; Q.-T.H. provided purified β-arrestin1 and contributed to plotting of data; N.C. guided the AT-3 labeling assay; L.Z. collected the samples from patients and controls; Z.Y. guided the bioluminescence resonance energy transfer assay of β-arrestins recruitment; F.Y. guided the cyclic adenosine monophosphate inhibition assay; Z.Y. and F.Y. provided critical materials, such as plasmids and ligands, used in this study; Y.-Q.D. and Y.-L.J. contributed to the animal experiments; S.-M.Z. provided the anti–K-Hcy antibody and the protocol of AHT synthesis, and guided the corresponding experiments; D.-M.Z. synthesized AHT; and all authors commented on the manuscript and approved the final manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Lu Tie, Peking University, 38 Xueyuan Rd, Haidian District, Beijing 100191, China; email: tielu@bjmu.edu.cn; and Jin-Peng Sun, Peking University, 38 Xueyuan Rd, Haidian District, Beijing 100191, China; email: sunjinpeng@bjmu.edu.cn.