G protein-coupled receptors (GPCRs) represent the largest family of membrane proteins within the human genome, comprising approximately 800 members. Based on sequence homology and evolutionary conservation, GPCRs are classified into different categories, including class A (rhodopsin-like), class B1 (secretin), class B2 (adhesion), class C (glutamate), class D1 (ste2-like fungal pheromone), class F (frizzled-like), and class T (taste 2). These GPCRs share a common structural framework characterized by a seven-transmembrane domain linked by three intracellular and three extracellular loops. Functioning as sensors for a wide variety of stimuli, ranging from photons and ions to neurotransmitters, small molecules, hormones and proteins, GPCRs are crucial to numerous physiological and pathophysiological processes (Wacker, Stevens, & Roth, 2017). Notably, approximately 36 % of drugs approved by the US Food and Drug Administration (FDA) target 121 specific GPCRs (Lorente et al., 2025). In fact, GPCRs and related genes account for around 12 % of all protein targets for approved drugs, underscoring their status as the largest family of drug targets.

GPCR signaling is primarily transmitted by transducer proteins, including G proteins, G protein-coupled receptor kinases (GRKs) and arrestins (Kolb et al., 2022). G proteins comprise 16 α, 5 β, and 12 γ subunits. The α subunits are classified into four families: Gi/o (Gi1, Gi2, Gi3, Go, Gt1, Gt2, Gt3 and Gz), Gs (Gs and Golf), Gq/11 (Gq, G11, G14 and G15) and G12/13 (G12 and G13). These subunits allow numerous combinations to form heterotrimeric G proteins. Among the four arrestins, arrestins 1 and 4 are found only in the retina, where they regulate rhodopsin and cone opsins, while arrestins 2 and 3 (also known as β-arrestin 1 and β-arrestin 2, respectively) serve as adaptor proteins that transduce signals for all other GPCRs (Rajagopal, Rajagopal, & Lefkowitz, 2010). Typically, for a GPCR, agonist binding induces a conformational change, promoting heterotrimeric G protein binding and the exchange of guanosine 5′-diphosphate (GDP) for guanosine 5′-triphosphate (GTP) on the Gα subunit. This leads to the dissociation of Gα from the Gβγ subunit, activating downstream effector proteins. To preserve biological homeostasis, negative feedback mechanisms deactivate prolonged second-messenger signaling by phosphorylating the active receptors with GRKs, which enables β-arrestin to bind and subsequently leads to receptor internalization (Goodman et al., 1996). The internalized receptors are either recycled or degraded, depending on factors including receptor type, post-translational modifications (e.g., ubiquitination), interaction with sorting proteins (e.g., β-arrestins), and ligand properties (e.g., persistent stimulation) (Hanyaloglu & von Zastrow, 2008). Besides acting as the negative regulators, β-arrestins also activate several mediators, such as mitogen-activated protein kinases (MAPKs) (Eichel, Jullie, & von Zastrow, 2016; Shenoy et al., 2006), the serine/threonine kinase AKT (Kendall et al., 2014), the tyrosine kinase SRC (Luttrell et al., 1999) and nuclear factor-κB (Gao et al., 2004), functioning independently of G protein signaling. Ultimately, the interaction of transducers and the receptor influences signaling outcomes in both time and space.

Given the pleiotropic nature of GPCR signaling, biased ligands can stabilize distinct active states of the receptor, facilitating interactions with specific transducers and thereby triggering selective activation of downstream signaling pathways, a phenomenon known as biased signaling (Kenakin, 2019). The initial discovery of biased signaling, as exemplified by pituitary adenylyl cyclase-activating polypeptide (PACAP-27) at the PACAP type-I receptor (Spengler et al., 1993), as well as ICI118551 and propranolol at the β2-adrenergic receptor (β2-AR) (Azzi et al., 2003), was ultimately recognized as a widespread occurrence across many GPCRs. Nowadays, ligand bias has been observed not only between G proteins and β-arrestins, but also among different families of G proteins (such as Gi, Gq and Gs) (Mao et al., 2023; Shang et al., 2023) and among subtypes within the same G protein family (such as Gi1, Gi2, Gi3, GoA, GoB and Gz of Gi family) (Faouzi et al., 2023). To quantify ligand bias, several methods have been proposed, with comprehensive description provided by Kenakin (Kenakin, 2019) and Onaran et al. (Onaran et al., 2017). Ligand bias is commonly calculated using Log (τ/KA) (Kenakin, Watson, Muniz-Medina, Christopoulos, & Novick, 2012) or Log (Emax/EC50) (Kenakin, 2017).

Over the past two decades, there has been a growing adoption of various strategies for the discovery of biased ligands, alongside increased interest in exploring the biological functions and potential applications in drug development. This interest stems from the therapeutic advantages that biased ligands can offer, which may not be achievable with balanced ligands. While several reviews have summarized GPCR biased ligands before 2018 (Kenakin & Miller, 2010; Tan, Yan, McCorvy, & Cheng, 2018), the continuous discovery of novel biased ligands and their unique biology and pharmacology necessitates an updated review. This analysis focused on class A GPCR biased ligands discovered between January 2003 and May 2024. Using “biased signaling” or “functional selectivity” or “biased ligand” and “GPCR” or “G protein coupled receptor” as search topics in Web of Science, we summarized 383 biased ligands for 60 class A GPCRs (available online at https://github.com/dicp2800/GPCR-Biased-Ligands) from approximately 3400 references, representing nearly double the number reported in the most recent 2018 review, which documented around 30 GPCRs with biased ligands (Tan et al., 2018). This review provides a brief overview of methods for discovering and assessing biased ligands, as well as their biological, preclinical, and clinical implications. Given that natural products are unique resources for drug development, their contributions to the discovery of biased ligands are examined.