1 Why we investigated this topic (written by Holden Karnofsky)
We were asked for our thoughts on a crowdfunding project1 raising money for biology experiments toward a potential class of therapies (called “DRACOs”) that “may be effective against all viruses.” I was instinctively skeptical because this seems like the sort of project that ought to have an easy time getting funding if it is indeed promising, and because I haven’t seen any signs that crowdfunding is a promising alternative to traditional funding sources for science. However, my model of “what science is and isn’t neglected by traditional funding sources” could easily be wrong, and having a good model on this question seems very important over the long run for giving well, since it will guide our choice of when to put time into investigating an idea vs. when to dismiss the idea along the lines of “Other funders are better positioned to evaluate this.” Since I saw excitement/interest from several people I believe are thoughtful about these issues, and since we are doing trial work with a senior scientific advisor (Chris Somerville), I thought it would be interesting to dig a little deeper on DRACO and see what we found.
My goal here is to work toward converging with others on the best model of broad market efficiency, rather than specifically to investigate DRACOs. As such, I’d like to make sure that time we put into investigating things like this is well-targeted at things that are likely to change someone’s mind, i.e. at good “tests” of our different models of broad market efficiency. A few hours of Chris’s time on DRACO didn’t give us a conclusive answer one way or the other on how promising this project is, but it did turn up some relevant observations, so we decided to write up what we have so far and ask others whether they think it would be a good idea to investigate further, or whether they’d rather see us spend that time investigating some other surface-level-promising idea for biomedical research.
2 Informal thoughts on this topic (written by Chris Somerville)
2.1 What DRACOs are
DRACOs (Double-stranded RNA Activated Caspase Oligomerizers) are broad-spectrum antiviral candidates developed by Dr. Todd Rider (a scientist who was at Lincoln Laboratory at MIT when the experiments were conducted). DRACOs work by detecting long double-stranded RNA (dsRNA) in cells and then triggering apoptosis (programmed cell death) when long double-stranded RNA is detected. According to the crowdfunding page for DRACOs, “virtually all virus-infected cells contain long double-stranded RNA, whereas healthy human and animal cells do not.”2
- Chris Somerville spent 2 hours reading about DRACOs and then discussed for about an hour with Holden.
- Chris, who is not a virologist, emailed a friend with knowledge of virology about DRACOs for a quick opinion.
2.3 Chris’s take on strengths
- The in vitro results with DRACOs are strong.
- The mechanism used by DRACOs is clever and broadly applicable.
2.4.1 Complicated delivery
- For DRACOs to work as a therapy, there needs to be an efficient method for therapeutically delivering proteins inside of cells in intact animals. Protein motifs (PTDs) that can induce uptake by cultured cells were used in the Rider paper3 and have been widely used experimentally. By contrast, delivery of therapeutic proteins into cells in live animals is relatively poorly described in the literature. In an early study when a TAT-fusion was injected intravenously into mice,4 the authors reported differential uptake into various cell types: “Histological staining showed very high tissue-associated activity in the liver, spleen, and heart. Low but detectable staining was seen in skeletal muscle and lung, little or no staining was seen in the kidney, and none was seen in the brain.” Kizaka-Kondoh et al., (2009)5 used a PTD construct to effectively target cancer cells in a mouse model of pancreatic cancer. Thus, the efficiency was apparently high enough that the amount of injected protein was tolerated (5mg/kg). In reviewing the use of PTDs for therapeutic uses, Van den Berg and Dowdy (2011)6 note: “Although it is still early in clinical trial development, by our count, there have been over 20 Phase I and Phase II clinical trials performed on over 2000 patients treated with PTD delivered peptide or protein cargo for a variety of indications by biotech companies and academic institutions, including: CapStone Therapeutics, CellGate, Kai Pharmaceuticals, Revance Therapeutics, Xigen Pharmaceuticals, the University of Maryland, and the University of Montreal. Impressively, Revance Therapeutics reported achieving statistical significance (P < 0.0001) in a Phase 2b trial of 180 patients, a first for PTD-mediated delivery.” Thus, it seems fair to say that, although more data is needed on which cell types are targeted by the various PTDs, the use of such constructs for medical applications is being explored by other groups and there is no evidence indicating it would not work for the kinds of applications envisioned for DRACOs.
- Whatever delivery method there is would likely have to be more complicated than a pill and might require skilled administration, limiting potential for widespread or quick use.
- For DRACOs to work, it would be necessary to deliver a lot of protein to cells (because DRACOs have to get into all cells that might be infected). Based on the mouse studies reported by Dr. Rider, it seems possible that a gram or more of DRACO would need to be administered to humans. The studies of Kizaka-Kondoh et al., (2009)7 used 5 mg protein per kg of mouse body weight. Delivering this large amount of protein could potentially trigger an immune response and cause anaphylactic shock, which can be fatal. Even human proteins used therapeutically exhibit this effect. In their review of immunological effects of therapeutic proteins, Kessler et al., (2006)8 write: “Most biopharmaceuticals induce immune responses (immunogenicity), which in many cases do not have clinically relevant consequences. However, in some cases the consequences can be severe and potentially lethal, causing a loss of efficacy of the drug or even worse, leading to autoimmunity to endogenous molecules.” Because DRACOs contain a non-human component, it is very likely that they will induce a strong immune response.
- Viruses sometimes infect large numbers of cells. If DRACOs succeeded in killing all human cells with viruses in them, that could mean killing very large numbers of human cells, potentially posing substantial health risks.
- Nature could have evolved a mechanism for triggering apoptosis in cells with dsRNA, but appears not do it naturally. One plausible explanation of this is that the costs of having this mechanism (i.e., cell death, possible off-target effects) would outweigh the benefits in most virus infections (eg., “colds”).
- There is a risk that some cells may naturally have enough dsRNA that they could potentially be attacked by DRACO despite being healthy. The proteins used in the DRACO system are most strongly activated by long dsRNA which is intended to make the mechanism specific for viruses. However, short dsRNA of 11-16 nucleotides can bind to the dsRNA binding sites and activate the proteins under some conditions.9 Thus, a lot of testing would be required to ensure that there are not any “off-target” effects in humans. Because the presence of dsRNA is species-specific, mouse studies do not provide compelling insights about effects on humans.
- Harmless viruses in cells could be targeted by DRACOs as well, posing additional risk (i.e., widespread cell death).
2.4.3 More limited applications than crowdfunding page implies
- In light of the safety concerns above, it may not make sense to use DRACOs to treat low-stakes conditions like rhinovirus (the common cold).
- While it might make sense to use a risky treatment against a very deadly virus, DRACOs would not work against HIV because this virus does not rely on dsRNA. However, the crowdfunding page lists HIV as a target. As an aside, VOCERO-AKBANI et al., (2010)10 describe a recombinant therapeutic protein that has some similarities with the DRACO system but targets HIV-infected cells by a different mechanism.
2.4.4 Research to date – and planned research – don’t seem to address these concerns
- All of the above concerns relate to delivery in animals, so they are consistent with DRACOs having good results with cultured cells, where nearly all research has taken place (including the only attempted replication). Cell types with dsRNA were probably not present, and there would be no challenges related to delivery or harm from large amounts of apoptosis.
- The Rider paper presented some preliminary mouse trials which indicated significant protection against life-threatening viral infections.11 However, the trials were quite short and there was no detailed analysis published of effects on the overall health or longevity of the mice. If further experiments were done, the emphasis should probably be on long-term health effects on a variety of animals with and without viral infections. A Facebook comment quoting an email from Dr. Rider addressed this, but we have not seen published results.
- The crowdfunding proposes additional in vitro testing (particularly in herpes), rather than animal testing that could address these concerns.
All of that said, DRACOs might be useful for something – perhaps a virus deadly enough to outweigh the potentially deleterious effects. Or perhaps the deleterious effects (if any) could be mitigated.
2.5 Chris’s take on broader negative signs
- Crowdfunding page cites the “valley of death” as an explanation for why the NIH is not funding work on DRACOs. Chris is skeptical of this explanation. He thinks that antivirals are a very hot area, and that it’s generally easy to get funding and interest around in vitro/cheap-animal trials of things that have worked before.
- Given above point, Chris thinks it’s a bad sign that there has been so little follow-up research since the Rider paper.12
- Given the points above, Chris’s guess is that there are further reasons and possibly further evidence for the unpromising-ness of DRACOs that we aren’t seeing. He thinks that enough investigation, such as talking to authors on the PLoS One paper, would be fairly likely to find such evidence.
2.6 Bottom line
- We don’t have enough information to say that DRACOs are a bad investment, but also don’t see DRACOs as an outstanding investment or something that we should examine more closely.
- We can see plenty of reasons why they don’t look as promising as crowdfunding page implies.
- A lot of NIH money goes to things that could turn out to massively reduce disease burden (e.g. oncology).
- We would guess that DRACOs would’ve gotten traditional funding, as well as interest from other academics, if they were highly promising (and that they will if future developments make them more promising). But we don’t know that.
3 Addendum: thoughts on a Facebook comment with some responses to concerns raised by others
After the above was completed, we came across a Facebook comment addressing some of the concerns that had been raised about DRACOs, which overlapped with our own. Chris’s thoughts on this comment are available here.
|Bevilacqua and Cech (1996)||Source (archive)|
|DRACO crowdfunding page||Source (archive)|
|Fawell et al. (1994)||Source (archive)|
|Kessler et al. (2006)||Source (archive)|
|Kizaka-Kondoh et al. (2009)||Source (archive)|
|Rider et al. (2011)||Source (archive)|
|van den Berg and Dowdy (2011)||Source (archive)|
|Vocero-Akbani et al. (2010)||Source (archive)|