When Archie Tse was a practicing oncologist at Memorial Sloan Kettering Cancer Center about a decade ago, he helped treat a woman with stage IV pancreatic cancer—an aggressive disease that can give a person only months to live. “Her condition was declining, and her prognosis was dismal,” Tse recalls.
Fluid began leaking into her abdomen, so a catheter was inserted to remove it. But when she developed an infection, the catheter was taken out, and she was sent home on hospice care.
A few months later, Tse saw the woman at the hospital again. “I did not expect her to be alive,” he says. The fluid in her abdomen had vanished, and yet she wasn’t on any therapy. “I thought, ‘What is going on?’ So I did a CAT scan to confirm, and sure enough, her tumors were gone.”
Doctors call this spontaneous remission. “If you see enough patients, you will have these anecdotal stories,” Tse says. In fact, doctors have been sharing anecdotes of tumors disappearing after an infection for over 100 years.
William B. Coley, also a doctor at Memorial Sloan Kettering but in the 1890s, discovered that a giant tumor slowly receded from a man’s face after an open wound became infected. That gave Coley the idea to inject his patients with dead bacteria in hopes that their tumors would regress.
The immune flare-up in response to Coley’s treatment may have primed immune cells to take notice of tumors as well as the injected pathogens. Sometimes it worked, but not surprisingly, injecting sick people with bacteria never came into mainstream medical fashion.
Now, instead of injecting tumors with bacteria, scientists are injecting people with treatments that promise to repeat Coley’s tumor-shrinking results. The target of these drugs: a protein called STING—for stimulator of interferon genes—that kicks into high gear during an infection.
STING is part of the innate immune system, the body’s first line of defense against pathogen invaders like bacteria and viruses. When activated, STING ramps up production of inflammatory proteins called interferons and cytokines. Through a complex cascade, these proteins jump-start the adaptive immune system, including an army of pathogen-destroying T cells.
In fact, modern medicine has already embraced the power of tapping T cells to fight cancer. Antibody-based checkpoint inhibitors such as Merck & Co.’s Keytruda and Bristol-Myers Squibb’s Opdivo work by releasing the brakes on T cells, which can learn to target and tackle cancer cells.
These checkpoint inhibitors work wonders in certain cancers, sometimes successfully treating 80% of patients, says Carl Decicco, head of discovery at BMS. “But most tumors are in the 20–30% range,” he adds. “There is certainly room for improvement.”
Companies are looking for therapies that could work alongside Keytruda and Opdivo, both expected to rake in more than $5 billion in sales this year. “We are constantly looking for other agents to combine with Keytruda,” says Tse, now an executive director of oncology clinical research at Merck. One of those agents is an activator, or agonist, of STING.
Checkpoint inhibitors don’t work for everyone because tumors are experts at keeping T cells away. And if there are no T cells near a person’s tumor to begin with, the drugs won’t do any good.
“You need a preexisting immune response for the checkpoint inhibitors to do their work,” says Gary Glick, CEO of IFM Therapeutics, which sold its STING agonist program to BMS last year. “The innate immune system gets the fire going.”
Put another way, activating STING creates an environment in which T cells can thrive. Then, with the aid of a checkpoint inhibitor, the T-cell army is primed to take down that tumor as well as other tumors lurking throughout the body.
That’s why, in just a few years, BMS, Merck, Novartis, and more than a dozen other companies have lunged at compounds that activate STING, betting that they can make checkpoint inhibitors work for more people and in more cancers. Preclinical STING agonist programs have already led to deals potentially worth more than $2 billion.
Some scientists think the first generation of STING agonists, which have to be injected directly into the tumor, won’t work for everyone. Thus many small firms are racing to develop and test the first small-molecule STING agonists that could be injected systemically or taken via pill.
The STING story
When Glen Barber discovered STING 10 years ago, he could not have predicted the excitement it would eventually create in cancer research. While studying immune system response to pathogens at the University of Miami, Barber found that cells missing the STING protein were highly vulnerable to viral infections (Nature 2008, DOI: 10.1038/nature07317).
Soon after, Russell Vance and colleagues at the University of California, Berkeley, discovered that a bacterial molecule called cyclic diguanylate monophosphate (c-di-GMP) binds to mammalian STING receptors and initiates an innate immune response in mammals (Nature 2011, DOI: 10.1038/nature10429).
c-di-GMP is a cyclic dinucleotide (CDN), a class of compounds made by linking two DNA building blocks called nucleotides. In 2012, Barber discovered that STING proteins can be activated by DNA leaking from dead cells of the same organism. But how STING recognized a bulky DNA molecule, very unlike the smaller CDN, was not clear.
Then, in 2013, Zhijian (James) Chen at the University of Texas Southwestern Medical Center discovered the missing piece of the puzzle. First, an enzyme called cyclic GMP-AMP synthase (cGAS) binds to DNA—whether from a virus, a bacterium, or the the same organism. cGAS then links two nucleotides to create a CDN molecule called cyclic-GMP-AMP (cGAMP), which is the natural agonist of STING (Science 2013, DOI: 10.1126/science.1232458). “Suddenly everything made sense,” Vance says. “Everything goes through STING.”
Our own DNA doesn’t normally trigger STING because DNA is safely packed away in a cell’s nucleus. If DNA leaks out of the nucleus into the cytosol where STING’s receptor faces, it can trigger an autoimmune reaction. Likewise, scientists recently discovered that radiation and chemotherapy may activate STING, thanks to the flotsam of DNA spilled from dying cancer cells.
Meanwhile, more STING surprises were brewing at Harvard Medical School. Lingyin Li was working in Tim Mitchison’s lab, which had basic research funding from Novartis. Mitchison wanted to use the money to study an anticancer compound called DMXAA that worked fantastically in mice but had just fallen flat in two Novartis-sponsored Phase III clinical trials.
“Tim said the Novartis people didn’t want to be in the same room when the drug was mentioned,” Li recalls. Neither did she. “I didn’t want to touch DMXAA,” Li says.
It took a while, but Mitchison convinced Li to take on the DMXAA project. “Once I caved, it got exciting very fast,” she says. Li unexpectedly showed that DMXAA activates STING. But unlike the natural CDN agonists of STING, which bind both human and mouse versions, DMXAA binds only to mouse STING (ACS Chem. Biol. 2013, DOI: 10.1021/cb400264n). Li had explained why the Novartis trials were unsuccessful. And that’s when STING fever hit.
A Berkeley-based company called Aduro Biotech showed that when synthetic CDNs are injected directly into a tumor implanted on a mouse’s flank, they activate STING, causing an innate immune response that warms up the tumor for a T-cell attack.
The Aduro team also gave the mice injections of tumor cells, which spread and covered their lungs. Astoundingly, injecting CDNs at one tumor activated T cells throughout the body, causing them to attack and clear cancer in other locations, including the tumor-riddled lungs (Cell Rep.2015, DOI: 10.1016/j.celrep.2015.04.031).
“That was a groundbreaking result because you don’t die of the tumor that you can cut out. You die of the tumors that micrometastasize,” says Aduro CEO Stephen Isaacs.
Novartis was watching this work closely. “They knew STING was a good target from years of priming due to our collaboration,” Li says. She published another paper about the design of a degradation-resistant cGAMP molecule that could be used as a STING agonist (Nat. Chem. Biol. 2014, DOI: 10.1038/nchembio.1661). Even though Li was the first to publish the structure, she was not able to patent it because Aduro had already done so.
Just months later, in March 2015, Novartis announced that it would help Aduro develop and commercialize its preclinical STING agonist program for cancer immunotherapy. The Swiss firm agreed to pay Aduro $200 million up front and potentially $500 million in milestone payments.
Li says the news was a validation. Her STING work helped her land a job at Stanford University School of Medicine, where she is now an assistant professor of biochemistry.
Although Li has shifted her focus to other proteins in the STING pathway, major drug companies frequently contact her to ask if she is still working on STING. “I think every single big pharma company is working on this,” she says.
Companies are swarming to develop STING agonists for cancer immunotherapy and STING antagonists for autoimmune diseases.
|Aduro Biotech and Novartis||Synthetic CDN STING agonist, intratumoral injection alone or with anti-PD1||Phase I clinical trial|
|Merck & Co.||Synthetic CDN STING agonist, IT injection alone or with anti-PD1 (Keytruda||Phase I clinical trial|
|Spring Bank Pharmaceuticals||Synthetic CDN STING agonist, IT or IV injection||Anticipated Phase Ib/II trial in 2018|
|Bristol-Myers Squibb||Synthetic CDN STING agonist, IT injection with anti-PD1 (Opdivo)||Preclinical|
|Curadev||Small-molecule STING agonists and antagonists||Preclinical|
|Mavupharma||Small-molecule STING pathwaya agonists and antagonists||Preclinical|
|StingInn||Small-molecule STING agonists and antagonists||Preclinical|
|StingInn and Vyriad||Oncolytic viruses encoding STING pathwaya activators||Preclinical|
|Synlogic||IT injection of E. coli bacteria engineered to produce high levels of the STING agonist c-di-GMP||Preclinical|
|Venn Therapeutics||IT injection of adenovirus that produces the bacterial STING agonist c-di-GMP||Preclinical|
|IFM Therapeutics||STING antagonist, cGAS modulators||Discovery|
|iTeos Therapeutics||Small-molecule STING pathway activatorsa, IV injection of molecules packaged in Cristal Therapeutics nanoparticles||Discovery|
|Nimbus Therapeutics||Small-molecule STING agonist||Discovery|
|Nimbus Therapeutics and Celgene||Small-molecule STING antagonist||Discovery|
|Sirenas||STING pathwaya antagonist||Discovery|
Note: Selected companies with drug development programs targeting the STING pathway. a Company hasn’t disclosed whether the drug target is STING or another protein in the STING pathway. CDN = cyclic dinucleotide. IT = intratumoral. IV = intravenous. Sources: Companies
In the past few years, STING fervor has continued unabated. In May 2016, Aduro and Novartis were the first to begin testing a STING agonist in a Phase I clinical trial. Merck launched its STING agonist and Keytruda combination trial in early 2017. Then in August, BMS paid $300 million up front, with $2 billion in potential milestones, to get preclinical programs, including one for a STING agonist, from IFM Therapeutics. And in September, Novartis began testing an experimental checkpoint inhibitor with Aduro’s STING agonist. Many future STING agonists will likely be tested in such combinations.
Some companies are also creating STING antagonists, molecules that would block overactive STING proteins in autoimmune diseases. In October, Celgene partnered with the computational chemistry company Nimbus Therapeutics to develop STING antagonists.
Nimbus also wants to design the first small-molecule agonists of STING, compounds that could be given intravenously, or even better, taken as a pill, unlike the CDNs from BMS, Merck, and Novartis, which will be injected directly into the tumor. “Small-molecule approaches for immuno-oncology are highly coveted by big pharma right now,” says Nimbus CEO Donald W. Nicholson.
CDNs are tough to turn into drugs since phosphodiesterase enzymes quickly break them down in the body. Chemical tweaks can make them more stable, but CDNs will probably never be packaged in a pill.
Using DMXAA—the mouse-specific STING agonist—as a starting point for developing small molecules has been “very successful” so far, Nicholson says. Mouse and human STING proteins are fairly similar, except for the spot where a CDN or small molecule binds. STING proteins pair up in cells, and the junction of this dimer is more splayed in humans than in mice. That makes it harder to find a small molecule that binds and activates the complex. The bulkier CDNs do it just fine.
Some firms have creative solutions for making CDNs work for more tumors. Spring Bank Pharmaceuticals says it can deliver modified, CDN-based STING agonists via a systemic IV injection. It expects to begin clinical trials of a liver cancer treatment late this year. The firm also has a partnership with an undisclosed antibody-drug conjugate developer to attach its STING agonist to an antibody that would home in on cancer cells.
Likewise, iTeos Therapeutics partnered with nanoparticle maker Cristal Therapeutics to deliver its drugs specifically to tumors. iTeos CEO Michel Detheux says the firm’s STING-activating compounds “are dependent on the presence of STING, but they don’t bind STING.” Detheux says the target is instead another protein upstream of STING and that his firm’s compound has the same end result as STING agonists.
Even if the systemic delivery of STING agonists—whether as CDNs or small molecules—is successful, companies will have to contend with the possibility of a dangerous and sometimes deadly event called a cytokine storm. If STING is activated too strongly, a massive, bodywide immune flare-up could run out of control. Such events have saddled other promising immunotherapies.
“This could open Pandora’s box,” says Spring Bank CEO Martin Driscoll. “We are very mindful of that in our development.”
Another toxicity concern was raised for the first time last year in a preclinical study from Tufts University that showed STING agonists could induce stress and death in T cells—the exact opposite of the drug’s intended effect. The authors wrote that a “re-evaluation of STING agonist-based therapies may be necessary” (J. Immunol. 2017, DOI: 10.4049/jimmunol.1601999).
Merck’s Tse says the high dose and duration of exposure to the STING agonist in the study might mean these results are not clinically relevant, but his team is closely watching T-cell levels in patients as they increase the dose in the Phase I study of Merck’s compound.
The first clinical trial data of a STING agonist could come from Merck or Novartis later this year. In the meantime, other big pharma companies, including GlaxoSmithKline, Janssen, and Pfizer, are quietly revealing their interest in the STING pathway through patents and scientific publications.
And in a full-circle return to Coley’s experiments from a century ago, there is renewed interest in injecting tumors with bacteria, this time genetically engineered to activate STING. The synthetic biology start-up Synlogic announced this month that it is testing a strain of Escherichia coli that produces high levels of the bacterial STING agonist c-di-GMP.
In a similar fashion, Venn Therapeutics is developing an adenovirus-based gene therapy that produces the same agonist when injected into tumors.
Researchers are still trying to work out the full details of the STING pathway and its role in infection and tumor immunity. “It is an exciting arena,” says Barber, the scientist who discovered STING. He has since started his own company, StingInn, to develop small-molecule agonists and antagonists of STING.
Even if the first generation of STING agonists don’t work as well as people hope, Barber thinks that future STING agonists, or maybe compounds targeting other proteins in the STING pathway, will take hold. “It is not going to go away soon,” he says. “I think interest in STING will be here to stay for many years.”