How SR-17018 splits the opioid signal at the μ-receptor.

SR-17018 is a small molecule that binds the μ-opioid receptor and triggers the analgesic G-protein arm of downstream signaling while recruiting the regulatory scaffold protein β-arrestin2 at a much lower rate than a balanced agonist would. In the original 2017 characterization, this pharmacology translated into a wider therapeutic window between analgesia and respiratory depression in mice.^[1] Follow-up work from the Bohn laboratory reported that chronic SR-17018 dosing does not produce the tolerance seen with morphine and can even reverse established morphine tolerance.^[2] A competing interpretation proposed by Gillis and colleagues in 2020 argues that the therapeutic-window benefit may be explained by low intrinsic efficacy rather than bias per se.^[3]Both readings are compatible with SR-17018's utility as a research tool; they differ in how the underlying biology should be described.

Subsequent literature complicated the early interpretation of SR-17018. Later work argued that some of its separation profile may reflect low intrinsic efficacy rather than bias alone, while newer studies also reported findings that tempered the original optimism from early rodent datasets.

The μ-opioid receptor (MOR; gene symbol OPRM1) is a seven-transmembrane, G-protein-coupled receptor expressed widely in the central and peripheral nervous system. On binding an agonist — endogenous β-endorphin, morphine, fentanyl, DAMGO in the laboratory — MOR engages heterotrimeric G-proteins of the Gαᵢ/o family. Activated Gαᵢ inhibits adenylyl cyclase, lowering intracellular cAMP; released Gβγ subunits open inwardly-rectifying potassium channels and close voltage-gated calcium channels. The net effect on a neuron is hyperpolarization and reduced neurotransmitter release. This is the biochemistry of opioid analgesia.

Within seconds of agonist binding, however, a second pathway begins. The receptor is phosphorylated on its C-terminal tail by G-protein-coupled receptor kinases (GRKs, principally GRK2 and GRK3 for MOR). Phosphorylated receptor is a substrate for β-arrestin2, which binds the receptor, sterically blocks further G-protein coupling, and recruits endocytic machinery — ending the signaling event and, often, internalizing the receptor.

β-Arrestin2 is not purely a brake, however. Once bound to the receptor, it acts as a scaffold for its own signaling cascades, including MAPK pathways. In the specific context of MOR, work from the Bohn laboratory in the late 1990s and early 2000s showed that genetic deletion of β-arrestin2 in mice enhanced morphine analgesia while reducing respiratory depression and gastrointestinal dysfunction.^[4,5] That genetic result is the foundation of the argument that biased agonism should, in principle, produce a safer opioid.

When a ligand binds a G-protein-coupled receptor, the receptor is not a simple on/off switch; it can adopt a range of conformations, and different conformations couple to different downstream effectors. A balanced agonist engages all available effectors in rough proportion to their natural coupling efficiencies. A biased agonist favors one pathway over another — in the MOR case, G-protein coupling over β-arrestin2 recruitment, or vice versa.

Quantifying bias is subtle. Two ligands might simply have different potencies or efficacies at a receptor, which is not the same as pathway-selective signaling. The currently accepted framework is the operational model of Black-Leff, reduced to a bias factor by Kenakin and colleagues.^[6] In practice, a reference agonist (commonly DAMGO for MOR) is chosen and assigned bias factor zero. Each test compound is characterized in at least two downstream assays — a G-protein assay such as [³⁵S]GTPγS binding or cAMP inhibition, and a β-arrestin recruitment assay such as BRET or PathHunter. The ratio of transduction coefficients across assays, relative to the reference, gives the bias factor: positive values indicate G-protein bias, negative values β-arrestin bias, zero indicates balanced signaling.

Two caveats apply to every bias calculation and become important in reading the SR-17018 literature. First, bias factor depends on the specific assays used; the same ligand can appear more or less biased depending on whether cAMP or GTPγS is the G-protein readout. Second, bias is relative to a chosen reference, so comparisons across studies that use different references are not straightforward. Both caveats are central to the 2020 re-analysis discussed below.

In the original Schmid 2017 characterization, SR-17018 was profiled alongside morphine, fentanyl, oxycodone, and TRV130 in matched G-protein and β-arrestin assays. Against DAMGO as reference, SR-17018 exhibited high-nanomolar potency for G-protein coupling and markedly reduced efficacy for β-arrestin2 recruitment — producing a bias factor substantially larger than the other compounds in the series.^[1]In the paper's framework, SR-17018 was the most biased MOR agonist they characterized.

The authors then computed a numeric therapeutic index in mice, defined as the ratio of the dose that produced respiratory depression to the dose that produced analgesia, and plotted this index against bias factor across their compound set. The correlation was positive and significant: higher bias, wider therapeutic window. SR-17018 sat at the top right of the plot. This is the result that launched the compound's reputation.

One useful way to understand the 2017 data is that SR-17018 is not a weak agonist in the traditional sense. It produces full or near-full efficacy for G-protein signaling at MOR — comparable to morphine. Its distinctive property is specifically the decoupling of that efficacy from β-arrestin2 recruitment. That decoupling is what the bias-factor framework is trying to quantify.

A full structure of MOR bound to SR-17018 has not, to our knowledge, been published at the time of writing. However, the broader question of what receptor conformations correspond to G-protein-biased versus β-arrestin-biased signaling has been addressed for several related ligands, and the working picture from that literature is informative for SR-17018 even without a compound-specific structure.

The prevailing view is that different MOR conformations present subtly different cytoplasmic surfaces to intracellular effectors. Some conformations favorably orient transmembrane helix 6 for G-protein engagement while being poorer substrates for GRK- mediated phosphorylation, which in turn reduces β-arrestin2 recruitment downstream. Compounds that stabilize such conformations are, operationally, G-protein biased.

A complementary line of evidence comes from engineered phosphorylation-deficient MOR mutants. Kliewer and colleagues showed that a receptor with its C-terminal phosphorylation sites removed displays reduced β-arrestin2 recruitment but — contrary to the simple bias hypothesis — shows increased tolerance and more pronounced respiratory depression in vivo.^[9] This result complicates the clean mapping from β-arrestin2 recruitment to opioid side effects and is part of the reason the field has softened its language around biased agonism since 2018.

For SR-17018 specifically, what can be said with confidence is that its binding produces the functional phenotype — high G-protein activation, low β-arrestin2 recruitment — regardless of whether that phenotype arises from a distinct receptor conformation, from low intrinsic efficacy, or from some combination of the two.

Newer pharmacology work suggests SR-17018 may achieve its signaling profile through a distinctive interaction mode rather than fitting a simple classic biased-agonist model alone.

In mice, SR-17018 was reported to produce dose-dependent analgesia in standard thermal nociception assays (hot-plate and tail-flick) at doses well below those that depressed respiration as measured by whole-body plethysmography. The ratio between the two dose-response curves — the therapeutic index — was larger for SR-17018 than for morphine, oxycodone, or fentanyl in the same study.^[1]

Grim and colleagues in 2020 extended this profile to chronic dosing. They reported that continuous SR-17018 administration by osmotic mini-pump did not produce the analgesic tolerance that morphine does over the same schedule, and that switching morphine-tolerant animals to SR-17018 progressively restored morphine responsiveness without triggering a precipitated withdrawal syndrome.^[2] The practical implication, if this pharmacology holds up broadly, is that SR-17018 is not merely a single-dose analgesic but a compound that interacts with the long-run regulatory biology of MOR differently from conventional opioids.

Gastrointestinal effects have been less extensively reported. The original 2017 paper noted attenuated colonic propulsion effects relative to morphine, consistent with the β-arrestin2 knockout phenotype.^[1,5] Reward liability and self-administration liability in rodents have received less attention in the published literature than the analgesia and respiration endpoints and remain an open area for characterization.

The conceptual question raised by the Gillis 2020 paper is worth stating precisely because it affects how SR-17018 is best described today.^[3] The Gillis argument is not that SR-17018 is not biased — the bias factor measurements are not in dispute — but that bias factor as measured in recombinant cell assays may not be the causal variable explaining the in-vivo therapeutic-window benefit. An equally good predictor, the Gillis group showed, is intrinsic efficacy for G-protein activation: a partial agonist with low intrinsic efficacy will look relatively biased in assays that amplify G-protein signaling more than β-arrestin recruitment, and will show a wider therapeutic window for reasons unrelated to bias per se.

The two interpretations make different predictions. If bias is causal, SR-17018 should behave differently from a simple partial MOR agonist at matched analgesic doses. If intrinsic efficacy is causal, SR-17018 should behave like a partial agonist with an unusual β-arrestin footprint — a distinction without a pharmacological difference in vivo. Experiments to distinguish the two are ongoing; for current purposes, the safer summary is that SR-17018 produces the expected in-vivo phenotype and that the mechanism by which it does so is under active revision.

A second open question concerns the durability of the tolerance findings reported by Grim 2020. As of the last review date at the top of this page, these results have not, to our knowledge, been independently replicated in a fully separate laboratory with an independent compound synthesis. Researchers using SR-17018 for tolerance studies should treat the published results as strong preliminary evidence rather than as an established phenotype.

SR-17018 is most usefully situated against two other much-discussed biased MOR agonists: TRV130 (oliceridine), developed at Trevena and reported by DeWire and colleagues in 2013,^[7] and PZM21, reported by the Manglik group using a structure-based discovery approach in 2016.^[8]

TRV130 is the only compound of the three to have reached the clinic — it was approved in 2020 as oliceridine (Olinvyk) for management of moderate-to-severe acute pain in hospital settings. In the Schmid 2017 dataset, TRV130 showed a positive bias factor but lower than SR-17018. Its clinical use is limited to short-term inpatient administration under monitoring.

PZM21 has had a more complicated scientific trajectory. The original 2016 paper reported analgesia without respiratory depression or reward liability in mice. A 2018 follow-up by Hill and colleagues reported that PZM21 does depress respiration and does produce tolerance under conditions that differ from the original study, specifically in a different mouse line and with higher doses.^[8] That finding, together with the 2020 intrinsic-efficacy reinterpretation, has made PZM21 a cautionary example in the biased-agonism literature and has indirectly raised the bar for accepting claims about any biased opioid candidate, including SR-17018.

Relative to all three, SR-17018 is distinguished by (a) the highest reported bias in the Schmid 2017 matched dataset, (b) the strongest claims for reduced tolerance on chronic dosing, and (c) the status of a research tool rather than a drug candidate — meaning that the scientific community, rather than a commercial developer, directs the characterization program.

More recent data narrowed the gap between the early “bias predicts safety” narrative and the alternative view that low intrinsic efficacy explains much of the observed profile.