Therapeutic Applications of Bias
Bias By Design: Biased agonism emerged as a pharmacologic force through pre-conceived notions of how agonist activity could be improved. These ideas emerged from consideration of the impact of various signals on physiology and the resulting outcomes. The first foray into this area was with the angiotensin receptor biased agonist/antagonist TRV027. This molecule was designed to improve the then current therapy of losartan (an angiotensin receptor blocker designed to block angiotensin-mediated vasoconstriction) (Violin and Lefkowitz, 2007); TRV027 provided angiotensin blockade of vasoconstriction with β-arrestin positive cardiac effects by being a biased agonist toward β-arrestin away from Gαq stimulation (Violin et al, 2006; 2010; 2014). From a theoretical point of view it is difficult to imagine a better exemplar biased molecule for the testing of this hypothesis in therapy. There have been many more prospective ideas presented for molecules with modified signaling profiles to improve drug therapy- see Table 2 of Kenakin, 2019. In general these ideas center on the emphasis of beneficial signaling pathways (eg. β-arrestin for PTH bone building in osteoporosis, [Ferrari et al, 2005]; β-arrestin-mediated glucagon-like peptide-1 insulin secretion in diabetes [Sonoda et al, 2008], or the elimination of negative signaling [opioid-mediated respiratory depression and or negative behavioral effects mediated by β-arrestin (Raehal and Bohn, 2011; Raehal et al, 2005; Bohn et al, 2003; Urs and Caron, 2014]). There are many more recent proposals for improved drug candidates based on the biased concept including agonists for muscarinic receptors (Randakova and Jakubic, 2021; McDonald et al, 2022), opioid receptors (Che et al, 2021; Conibear et al, 2020), Ghrelin (Mende et al, 2018), Neurotensin (Krumm et al, 2023), Glucocorticoids (Mao et al, 2023), and 5-HT2A (Pottie et al, 2023; Allen et al, 2011). Interestingly, efforts to achieve selectivity through allosteric modulation in studies for Alzheimer’s disease and schizophrenia fall short of yielding the selectivity needed without added biased signaling (in this case, induced bias through allosteric modulation) (van der Westhuizen et al, 2021).
Currently new approaches to the active quest for biased molecules have been developed (i.e. DNA-encoded libraries (Cai et al, 2023). At the receptor level techniques employing molecular dynamic simulations (Suomivuori et al, 2020), receptor structure (Cao et al, 2023; Sengmany et al, 2020; Vuckovic et al, 2023), receptor/β-arrestin/GRK structures (Chen and Tesmer; 2022), and combinations of techniques (eg atomic-level molecular dynamic simulations and functional assays) have been used to design biased molecules (El Daibani et al, 2023). Interestingly, a case where bias has been associated with structure is the identification of an allosteric binding site on PTH1R (and possible other members of class B GPCRs) which mediates only receptor interaction with G proteins and not β-arrestin (Zhao et al, 2023).
Interest in previously forbidden targets (where activation of the target initiates negative signaling responses) has been resurrected through considering biased signaling. An example of this is the Κ-opioid receptor which could be involved in cognition, reward, mood and perception. Agonists for these receptors have possible utility as antidepressants and anxiolytics in affective disorders, drug addiction, and psychotic disorders. However, this receptor also produces β-arrestin-mediated disturbing hallucinations thereby precluding therapeutic application. Analyses of biased signaling of Κ-opioid agonists suggests that the harmful effects of Κ-opioid agonism may be modulated through biased signaling and convert this receptor to be a viable therapeutic candidate protein (White et al, 2014; Che et al, 2021).
A variation on seeking bias in a new molecule is to intentionally control the intrinsic efficacy of the molecule to diminish an unwanted signaling effects. An added bonus of this approach is that the biased molecule will function as an antagonist of the negative signals that would have been produced by the natural agonist. This actually is an important part of the biased profile for the angiotensin molecule for heart failure TRV027; specifically, while it produces β-arrestin agonism with an EC50 of 10 nM, it also is a competitive antagonist at the same concentration blocking natural angiotensin mediated agonism of vasoconstriction at the same concentration (pKi=8.0: Violin et al, 2010 ). In fact the low intrinsic efficacy of the new biased opioid agonist TRV130 is postulated to be an important part of this molecule’s therapeutic profile (Che et al, 2021). Specifically, estimation of the relative efficacy of TRV130 (vs morphine) with the operational model of agonism (Black and Leff, 1983) shows that TRV130 has 33% of the efficacy of morphine for G protein and only 15% of the efficacy of morphine on β-arrestin (Singleton et al, 2021). Partial agonism of opioid receptors is being pursued as a means of bias for other molecules as well (Lutz et al, 2023). In addition, low efficacy for β-arrestin has been identified as an important feature of new GLP-1 biased agonists such as tirzepatide (Willard et al, 2020; Yuliantie et al, 2020; Jones; 2021).
Bias by design also may be approached through the mechanism of allosteric agonists. Specifically, there are data to show that allosteric agonists tend to have a different signaling bias from standard orthosteric agonists; for example, in a series of muscarinic m2 receptor agonists a bias plot for orthosteric and allosteric agonists producing [35S] GTP-S and ERK1/2 responses in CHO cells shows that the conventional orthosteric agonists are generally biased toward GTP-γ-S while the allosteric agonists are biased toward ERK1/2 ( Gregory et al, 2010).
Bias By Cross-Screening: Another approach is through the evaluation of biased signaling in agonist therapy in retrospect, i.e. cross screening molecular libraries to identify exemplar selective agonists that show preference to one pathway over another. This identifies molecules that specifically activate certain signaling pathways over others which then can be explored separately to identify possible therapeutic utility. For example, a comparative screen of adenosine A1 receptor agonists in calcium and cyclic AMP assays demonstrates a wide dispersion of biased compounds with little correlations (Aurelio et al, 2018). This approach plays on the natural expectation of allosteric probe dependence yielding bias thus reversing the question…’should an agonist program seek bias?’ into ‘…Bias will eventually seek an agonist program’.