Developing safer opioids remains a long-held goal of pain researchers, who have especially set their sights on the mu opioid receptor (MOR). Agonists that bind MORs can activate multiple signaling pathways inside the cell. The most prominent pathway is mediated by heterotrimeric G proteins and is responsible for pain relief. Another pathway, controlled by β-arrestin, a scaffolding protein that regulates G protein-coupled receptors such as MORs, is responsible for opioid-related side effects including respiratory depression. To make opioids safer, researchers are redesigning these drugs to preferentially activate the G protein pathway over the β-arrestin pathway.
Although this approach, called biased agonism, has shown promise, investigators didn’t have a clear picture of how much bias is needed to make a less dangerous opioid—until now.
A years-long study led by Laura Bohn, Scripps Research Institute, Jupiter, US, presents a series of MOR agonists to explore how the degree of signaling bias impacts the therapeutic window, which here represents the range of doses at which an agonist produces anti-nociception while avoiding respiratory depression. Using cell-based assays that independently measure activation of the G protein versus the β-arrestin pathway, Bohn and colleagues designed agonists with varying degrees of bias. They found that the extent of bias correlates with the therapeutic window. Indeed, in two mouse models of thermal pain, the agonists provided analgesia similar to morphine and fentanyl, but the higher the bias factor, the less likely they were to affect respiration.
“I love reading stuff like this. They show the highest degree of biased agonism in the entire group of G protein biased ligands I’ve seen. They clearly cause much less respiratory depression than other opioid-related compounds while still producing strong anti-nociception,” says William Schmidt, NorthStar Consulting, Davis, US, an expert on drug development who was not involved in the study.
The new research was published online November 16, 2017, in Cell.
Biased agonism becomes more complicated
In 2005, Bohn found that genetically deleting β-arrestin 2 in mice could dramatically lessen the effect of morphine on respiratory depression while enhancing its analgesic capacity (Raehal et al., 2005). This research helped lay the groundwork for a biased agonist approach to opioid design (see PRF related news here and here).
“We had all this evidence that eliminating β-arrestin 2 recruitment would lead to a better analgesic with fewer side effects,” said Bohn. “So, we were very interested in making ligands that would selectively activate the MORs but not recruit β-arrestin 2.”
Last year, the biopharmaceutical company Trevena completed late-phase clinical trials with their own MOR biased agonist, TRV130 (Singla et al., 2017).
“This was a landmark compound because it was the first biased agonist to go into human testing,” said Schmidt.
But although the compound relieved pain just as well as morphine, it only avoided respiratory depression at the lowest tested dose; at higher doses, it was no better at avoiding this side effect than morphine was. But this work left open a question: When separating opioid analgesia from opioid side effects, is it enough to consider only the presence of biased agonism, or does the degree of bias matter, too?
Exploring a range of bias factors
To address this question, Bohn’s team synthesized a series of new MOR agonists that each had a slight modification of its core chemical structure.
“I moved to Scripps because it’s well known for its chemistry. It took us two to three years to really find compounds that were highly biased,” explained Bohn. “Then we kept working on them—looking for better selectivity and running them through off-target screens. It was a very iterative process.”
They calculated the bias factor for each compound by using two cell-based assays. The first measured G protein pathway activation by examining recruitment of a radioactively labeled guanosine triphosphate (GTP) analog to the cell membrane following MOR activation; GTP recruitment to an activated GPCR is a process required for G protein signaling. For this assay, the investigators used Chinese hamster ovary (CHO) cells that expressed human MORs. The second assay, based on enzyme fragment complementation, measured the activity of the β-arrestin pathway.
Each compound could then be compared to one another relative to DAMGO, a highly selective MOR agonist that activates each pathway maximally. If a particular compound activated the G protein pathway more than the β-arrestin pathway, then it would have a bias factor greater than 1 and be considered biased toward the former.
The researchers found that fentanyl had a bias factor of less than 1, meaning that it was biased toward the β-arrestin pathway. Morphine, however, had a bias factor slightly greater than 1, meaning it was somewhat biased toward the G protein pathway. But, remarkably, the team found five compounds that had bias factors ranging from 11 to 85. For context, Trevena’s TRV130 had a bias factor of 3.
Given that the context in which a GPCR is activated can influence its downstream signaling, the researchers wanted to confirm the bias factors in a different system. This time, they used a cyclic AMP (cAMP) accumulation assay in CHO cells expressing human MORs as a measure of bias toward the G protein pathway; cAMP is a so-called second messenger whose levels are increased after agonists bind to GPCRs like the MOR.
Once again, morphine and fentanyl bias factors hovered at about 1. Although the exploratory compounds still had high bias factors, the range dropped to 2.5 to 40 using the cAMP assay. Demonstrating consistency, however, the ranked order of the compounds, based on the bias factor, was preserved. Similar results were found using mouse MOR.
Does increasing signaling bias enhance the therapeutic window?
Next, the authors systemically administered the agonists to mice over six hours and confirmed that the compounds made their way to the brain, just as morphine and fentanyl do. Similar to morphine, plasma levels of the agonists peaked within 30 minutes and were still present one hour after administration.
The team then tested the compounds’ acute analgesic capacity with the hot plate and warm water tail flick assays.
“We chose these two assays because they are a gold standard with mice and have been used for nearly a hundred years,” said Bohn. “If you have a new compound, you should test it in an old standard assay that reliably models the effects of a known drug.”
Each agonist decreased pain sensitivity just as much as morphine or fentanyl. But measurements of arterial oxygen saturation and breathing frequency revealed that the agonists with the highest bias factor had the smallest effect on respiration.
Importantly, the anti-nociception produced by each compound was dose dependent, and the dose needed to produce 50 percent of the maximal effect was similar to that of morphine. Unlike morphine or fentanyl, however, this dose had minimal to negligible effects on respiration.
Using these data, the researchers then calculated the therapeutic window for each compound. A correlation was revealed: the greater the bias factor, the larger the therapeutic window.
“It comes down to fundamental pharmacology, where the poison is in the dose,” explained Bohn. “What’s the dose that will give maximal efficacy to treat pain, and is that going to give enough of a window of separation from the side effects?”
What about chronic pain?
The new compounds represent a paradigm shift in the way researchers might think about biased agonism, along with their promise as potential treatments in people.
“This is a tremendous win, because it demonstrates an important concept in the field while also pointing toward some new compounds for therapeutic development,” remarked Schmidt.
It may even explain why Trevena’s drug succeeded in relieving pain but only avoided respiratory depression at lower doses. That compound’s bias factor of 3, although significant, may not have produced a therapeutic window large enough to prevent the side effects.
One limitation of the current study, Schmidt says, is that it could have looked to a more familiar drug for a better understanding of the novel agonists. “We already know that buprenorphine, an agonist-antagonist analgesic, produces less respiratory depression than morphine, fentanyl, and other traditional opioids,” he said. “And recent studies have shown that buprenorphine is a biased agonist, so including this within the current study would have provided a benchmark against which we could have calibrated the significance of the newer compounds.”
As for future research, while the current study examined acute responses to biased MOR agonists, whether a biased agonism strategy will work in the setting of chronic pain remains unclear. In fact, a recent study from Ru-Rong Ji, Duke University, Durham, US, and colleagues using β-arrestin 2 knockout mice suggested that biased agonists might have chronic effects by promoting hyperalgesia and tolerance, through an impact on the NMDA receptor (Chen et al., 2016; also see comments on the current paper, including a comment from Ji).
“We really need to know what happens with chronic dosing of these compounds because there are adaptive changes that can occur,” says Schmidt. “It would be wonderful to see these same studies repeated with cells that have been incubated continuously with the highly biased compounds.”
Nathan Fried is a postdoctoral fellow at the University of Pennsylvania, Philadelphia, US.
Image credit: Schmid et al., 2017, with permission from Elsevier.