Opioid abuse, addiction, and overdose present a major public health crisis internationally. Now more than ever, there is a critical need for safer effective painkillers with lower abuse potential. Researchers from academia and industry have turned their attention to the concept of biased agonism, which describes the phenomenon where a ligand stabilises a subset of receptor active conformations, giving rise to differential or preferential activation of different cellular responses when compared to a reference agonist and reference pathway. For the μ‐opioid receptor, the aim is to develop novel agonists that stimulate analgesia with the exclusion of adverse effects, ultimately to offer an improved therapeutic index when compared to commonly used μ‐opioid receptor agonists such as morphine and fentanyl. A prevailing hypothesis in the field has been that μ‐opioid receptor signalling via β‐arrestin2 mediates the adverse effects including respiratory depression and constipation. The study from Kliewer et al. (2020) presents a highly collaborative effort to convincingly demonstrate that β‐arrestin2 knockout mice retain the phenotype for morphine‐ and fentanyl‐induced respiratory depression and constipation. These findings contradict an earlier report (Raehal, Walker, & Bohn, 2005), challenging the paradigm that μ‐opioid receptor signalling linked to adverse, as distinct from therapeutic, effects occurs through β‐arrestin and G protein respectively, which has major implications for discovery efforts focused on biased μ‐opioid receptor agonists as novel analgesics.

A major strength of the study from Kliewer et al. (2020) is the independent replication of respiratory depression experiments from laboratories spanning three different countries. The authors in each location use different methods to measure respiration in mice. All consistently showed that complete knockout of β‐arrestin2 in C57BL/6J mice had no effect on morphine‐induced respiratory depression. The study design reported here is particularly strong with three independent cohorts of animals used and experimenters blinded to both genotype and treatment. The authors are unable to pinpoint why the outcomes here did not replicate the original report (Raehal, Walker, & Bohn, 2005). However, their hypothesis that differences in breeding and the use of mice with mixed genetic backgrounds in the original study likely contributes to these differences is highly probable. In this respect, the study highlights some of the inherent limitations of using knockout animals to dissect the mechanistic basis of drug action. Namely, the potential effects of background strain and inbreeding on the functional response/s under investigation or possibly compensatory mechanisms masking the effects of knocking out a single protein. This study demonstrates the importance of transparency in animal research, including recognition of the need to publish research outcomes that fail to confirm a hypothesis, as well as independent replication studies.

Additional evidence by Kliewer et al. (2020) supports their conclusions regarding the role of β‐arrestin2 in opioid‐induced adverse effects. Fentanyl caused a similar degree of respiratory depression in β‐arrestin2 knockout mice compared to WT littermates (Kliewer et al., 2020). In addition to respiratory depression, both morphine and fentanyl induced constipation in WT and β‐arrestin2 knockout mice to a similar degree. These data are consistent with μ‐opioid receptor knock‐in mice that contain phosphorylation‐deficient mutations (that display decreased β‐arrestin2 recruitment) also retaining constipation and respiratory depression phenotypes in response to fentanyl and morphine (Kliewer et al., 2019). This is particularly noteworthy in light of the recent report that, relative to DAMGO, a synthetic μ‐opioid receptor agonist, fentanyl is generally biased towards β‐arrestin2 recruitment over GTPγS binding. However, the reported relative efficacy of fentanyl is very much dependent on species (rat vs. human), response or assay used to quantify the effect (Rivero et al., 2012; Schmid et al., 2017). Kliewer et al. (2020) did not explore other possible mechanisms for morphine‐ or fentanyl‐induced respiratory depression and constipation, although a compensatory increase in β‐arrestin1 expression in the β‐arrestin2 knockout mice was ruled out. Multiple independent studies support the conclusion that knockout or knockdown of β‐arrestin2 enhances morphine analgesia and prevents tolerance development (reviewed in Raehal, Schmid, Groer, & Bohn, 2011). Collectively, these studies support the notion that other, different, mechanisms contribute to μ‐opioid receptor‐mediated cellular responses that give rise to the various side effect liabilities associated with prototypical μ‐opioid receptor agonists. Indeed, for biased ligand discovery across the entire GPCR superfamily, parsing out the cellular responses linked to therapeutic, rather than adverse, effects presents a major challenge. For many GPCRs the “optimal” in vitro bias profile to translate to greater safety and maintain efficacy is unknown.

Often GPCR signalling bias is over‐simplified as a choice between signals from just two transducers: G protein‐dependent or arrestin‐dependent. While convenient, this mis‐represents the spectrum of transducers that contribute to GPCR‐stimulated cellular responses. Biased ligands should always be defined as relative to a reference agonist and pathway, because the degree and magnitude of bias are influenced by system and observational factors. The preferred reference ligand is often the endogenous agonist. However, this may not always be a feasible or simple selection, for example, GPCRs with multiple endogenous ligands such as the opioid receptors, in vitro systems where multiple subtypes of a receptor are present, or when studying orphan GPCRs. Arguably, in vitro bias profiles are best used as a means to differentiate ligands, where ligands with distinct bias profiles can then be clustered and used to correlate with in vivo effects. Indeed, clustering ligands based on bias profiles determined from in vitro assays may provide a means to overcome the confounding influence of species differences, where in vitro profiles are performed on human receptors whereas preclinical studies are performed in rodents. Recently, μ‐opioid receptor in vitro bias profiles were correlated with differential in vivo therapeutic windows for anti‐nociception efficacy versus those for inducing respiratory depression (Schmid et al., 2017). Of particular note, the bias factor (relative to the synthetic agonist DAMGO and β‐arrestin2 recruitment) for GTPγS binding was more strongly correlated than for cAMP accumulation. Coupled with the findings reported by Kliewer et al. (2020), this suggests that classifying μ‐opioid receptor‐biased agonists based on binary bias analyses within a discovery pipeline is insufficient. Moving forward, we are presented with an opportunity to change the GPCR‐biased signalling paradigm by broadening our view beyond known canonical transducers and exploiting new pharmacological tools with well‐defined bias fingerprints.

阿片类药物的滥用，成瘾和药物过量在国际上是一个重大的公共卫生危机。现在比以往任何时候都更迫切需要更安全，有效的止痛药，而这种止痛药的使用可能性较低。来自学术界和工业界的研究人员已将注意力转移到偏向激动的概念上，该现象描述了一种现象，其中配体稳定了一部分受体活性构象，与参考激动剂和参考相比，会引起不同细胞应答的差异性或优先激活。途径。对于μ阿片受体激动剂，目的是开发一种新型激动剂，以刺激镇痛并排除不良反应，与吗啡和芬太尼等常用μ阿片受体激动剂相比，最终提供更高的治疗指数。该领域的一个普遍假设是，通过β-arrestin2产生的μ阿片受体信号介导了包括呼吸抑制和便秘在内的不良反应。来自Kliewer等人的研究。（2020年）进行了高度合作，以令人信服地证明β-arrestin2基因敲除小鼠保留了吗啡和芬太尼诱导的呼吸抑制和便秘的表型。这些发现与较早的报道相矛盾（Raehal，Walker和Bohn，2005年），挑战了与阿片类药物受体信号转导的不良作用有关的范式，这与治疗效果不同，分别通过β-arrestin和G蛋白发生，这具有重要意义对于发现工作的关注集中在有偏见的μ阿片受体激动剂作为新型镇痛药。

Kliewer等人的研究的主要优势。（2020）是来自三个不同国家的实验室对呼吸抑制实验的独立复制。每个位置的作者使用不同的方法来测量小鼠的呼吸。所有研究均一致表明，在C57BL / 6J小鼠中完全敲除β-arrestin2对吗啡诱导的呼吸抑制没有作用。此处报道的研究设计特别强大，使用了三个独立的动物队列，并且实验者对基因型和治疗均不知情。作者无法确定为什么这里的结果不能复制原始报告（Raehal，Walker和Bohn，2005年））。但是，他们的假说很可能是原始研究中的繁殖差异和具有混合遗传背景的小鼠的使用差异的原因。在这方面，研究突出了使用基因敲除动物剖析药物作用机理的一些固有局限性。即，背景菌株和近交对正在研究的功能性反应或可能的补偿机制的潜在作用掩盖了敲除单个蛋白质的作用。这项研究证明了透明度在动物研究中的重要性，包括认识到需要发表未能证实假设的研究成果以及独立的复制研究。

Kliewer等人的其他证据。（2020）支持他们关于β-arrestin2在阿片类药物引起的不良反应中的作用的结论。与WT同窝仔相比，芬太尼在β-arrestin2基因敲除小鼠中引起相似程度的呼吸抑制（Kliewer等，2020）。除呼吸抑制外，吗啡和芬太尼在WT和β-arrestin2基因敲除小鼠中诱发便秘的程度相似。这些数据与μ阿片受体敲除小鼠一致，这些小鼠含有磷酸化缺陷突变（显示出减少的β-arrestin2募集），并且还保留了对芬太尼和吗啡起反应的便秘和呼吸抑制表型（Kliewer等人，2019年））。鉴于最近的报道，这尤其值得注意，相对于DAMGO，一种合成的阿片类受体激动剂，芬太尼通常偏向于GTPγS结合的β-arrestin2募集。但是，已报道的芬太尼的相对功效在很大程度上取决于物种（大鼠与人类），用于量化效应的反应或分析方法（Rivero等人，2012 ; Schmid等人，2017）。Kliewer等。（2020年）没有探讨吗啡或芬太尼诱导的呼吸抑制和便秘的其他可能机制，尽管排除了β-arrestin2基因敲除小鼠中β-arrestin1表达的补偿性增加。多项独立研究支持以下结论：敲除或抑制β-arrestin2会增强吗啡镇痛作用并阻止耐受性的发展（综述见Raehal，Schmid，Groer和Bohn，2011年））。总的来说，这些研究支持这样的观点，即其他不同的机制也促进了阿片受体介导的细胞反应，从而引起了与典型的阿片受体激动剂相关的各种副作用。实际上，对于在整个GPCR超家族中发现有偏差的配体而言，解析与治疗作用而非不良作用有关的细胞应答是一项重大挑战。对于许多GPCR，未知的“最佳”体外偏倚曲线可转化为更高的安全性并保持功效。

通常，GPCR信号偏倚被过分简化为仅来自两个传感器的信号之间的选择：依赖于G蛋白或依赖于抑制蛋白。虽然很方便，但是它错误地代表了有助于GPCR刺激的细胞反应的换能器谱。偏差配体应始终定义为相对于参考激动剂和途径，因为偏差的程度和大小受系统和观察因素的影响。优选的参考配体通常是内源性激动剂。但是，这可能并不总是可行或简单的选择，例如，具有多个内源性配体（例如阿片受体）的GPCR，存在多个受体亚型的体外系统，或在研究孤儿GPCR时。可以说，体外偏倚曲线最好用作区分配体的方法，然后可以将具有明显偏倚特征的配体聚在一起，并用于与体内效应相关。实际上，基于从体外测定中确定的偏倚概况的聚集配体可提供克服物种差异的混杂影响的方法，其中对人类受体进行体外概况而在啮齿动物中进行临床前研究。最近，阿片类药物受体的体外偏倚曲线与抗伤害感受的体内治疗窗口与诱导呼吸抑制的体内治疗窗口不同有关（Schmid等，其中对人类受体进行体外分析，而对啮齿动物进行临床前研究。最近，阿片类药物受体的体外偏倚曲线与抗伤害感受的体内治疗窗口与诱导呼吸抑制的体内治疗窗口不同有关（Schmid等，其中对人类受体进行体外分析，而对啮齿动物进行临床前研究。最近，阿片类药物受体的体外偏倚曲线与抗伤害感受的体内治疗窗口与诱导呼吸抑制的体内治疗窗口不同有关（Schmid等，2017）。特别值得注意的是，GTPγS结合的偏倚因子（相对于合成激动剂DAMGO和β-arrestin2募集）比cAMP积累更紧密相关。加上Kliewer等人报告的发现。（2020年），这表明在发现管道中基于二元偏倚分析对μ阿片类受体偏向激动剂进行分类是不够的。展望未来，我们将有机会改变GPCR偏向的信号范式，方法是将我们的视野扩大到已知的典型换能器之外，并利用具有明确定义的偏倚指纹的新药理学工具。