Late last month, Stephen Russell and colleagues from the Mayo Clinic College of Medicine published a study in Nature Medicine demonstrating how viral tropisms can be modified by tissue-specific microRNAs.
Specifically, the investigators inserted target sequences complementary to muscle-specific miRNAs into the 3’ UTRs of an oncolytic virus that causes lethal myositis in tumor-bearing mice.
“The recombinant virus still propagated in subcutaneous tumors, causing total regression and sustained viremia, but could not replicate in cells expressing complementary miRNAs and therefore did not cause myositis,” they wrote in the study.
In addition to its potential in treating cancer, the approach may also “serve as a new modality for attenuating viruses for vaccine purposes,” the team added.
Last week, RNAi News spoke with Russell about the findings and their implications in treating cancer and potentially creating safer vaccines.
Let’s begin with the focus of your research.
The goal of my research lab is to use replication-competent viruses for cancer therapy — [that is,] viruses that selective replicate in and destroy tumor tissue — and translate those into clinical testing where possible.
So we do a lot of work engineering the tropisms of the various different viruses that we work on, and that is a reasonable spectrum of viruses: we work with measles virus, vesicular stomatitis virus, picornaviruses such as the coxsackievirus A21, adenovirus vectors, et cetera.
Where did the idea to start using microRNAs to tweak the tropism of these viruses come in?
I’m particularly interested in multiple myeloma as a target disease. Someone suggested I look at coxsackievirus A21, [which] was being studied in Australia by a group led by Darren Shafren [at the University of Newcastle]. He’d published a paper about three years ago showing that the virus had very potent oncolytic activity in melanoma models. You could grow a [subcutaneous] melanoma and give it tiny doses of virus, either intratumorally or intravenously, and the tumor went [away] very quickly. It appeared this virus had a stupendous capacity for amplification within the tumor.
We ended up trying that virus in multiple myeloma xenograft models, and we found that the tumors went [away] really fast, but then the animals died; just about the time when the tumor was regressing fast, the animals would develop severe paralysis. When we did an analysis of the animals, we found [the deaths were] due to myositis. We’d initially thought it was probably some kind of neurological infection because the CVA21 is closely related to poliovirus, but it wasn’t; it was myositis.
So here we had a virus that could destroy a tumor, but it had the unpleasant side effect of killing the animal in which the tumor was growing. What we wanted to do was eliminate the toxicity and retain the therapeutic efficacy.
We’ve done a lot of work in the past on how to engineer the host ranges of viruses by transcriptional targeting … but you can’t do that with CVA21 because it’s an RNA virus and its life cycle is entirely cytoplasmic — there is no DNA phase. Another approach we’ve used very successfully with measles is to engineer cell-targeting ligands into the virus’ coat, but [doing so] with picornaviruses is very difficult because they’re much smaller, much more compact, [and] the capsid structure is much more difficult to engineer. There is no technology available at this point for changing the tropism that way.
There is the interferon pathway stuff of somehow incapacitating the virus’ ability to combat the cellular interferon response, and we tried to do that, but it didn’t work. So we were left needing a new concept. About that time, I saw a presentation given by Luigi Naldini who is working in Italy [at the San Raffaele Telethon Institute for Gene Therapy]. He had introduced microRNA targets into lentiviral vectors and had found that was a good way to diminish the expression of the viral transgene in specific cell lineages.
He had been particularly concerned about lentiviral vectors transducing the mononuclear phagocytic cells in the liver, [which] results in antigen presentation and a powerful immune response against the encoded transgene. He wanted to engineer the vectors so they wouldn’t express in the macrophages but would express in the liver cells, so he’d introduced target sites for a hemapoietic microRNA. When he did that he got vectors that no longer expressed in the macrophages but expressed well in hepatocytes, and that was associated with an amelioration in the immune response.
I was listening to him and thought, “Wow, this is a great approach. It’s very economical — it only takes about 100 bases of sequence to give you the targeting — and it might work in a replication-competent virus.” And we had the ideal model system, if you like, in which to test it because we had oncolytic activity and then toxicity to muscle.
We were able to find some muscle-specific microRNA target sequences and incorporate them into the viral genome.
So this work in Nature Medicine is a sort of proof of concept for this approach?
Yeah. It had never before been shown that you could control the tropism of a replication-competent virus at the time we undertook this work. So it was a kind of blue-sky research project, but there was fairly good proof of principle in non-replicating systems that you could damp down gene expression.
Can you touch on the key findings of the paper?
What we did was to construct a recombinant CVA21 in which we had introduced into the 3’ untranslated region of the viral genome four copies of a microRNA target sequence. In fact, we did try different configurations, and the one that worked for us was one in which we had two copies of a sequence for one microRNA, [miR-133], and two copies of the target sequence for another microRNA, [miR-206], both of which are found exclusively in muscle.
That virus rescued and grew very nicely on HeLa cells, which is the cell line we grow these things up on, but in muscle cell lines, it had great difficulty in replicating, which was in contrast to the unmodified virus. When we transfected the HeLa cells with microRNA mimics corresponding to the two microRNA targets, the cells were no longer able to support the growth of the recombinant virus, whereas they could still support unfettered growth of the wild-type virus.
It appeared from those in vitro studies that the microRNAs that corresponded to the targets we put in the viral sequences very efficiently shut down viral replication. In vivo, what we did was grow subcutaneous tumors in SCID mice, either myeloma or melanoma tumors, and treat the mice with the recombinant virus or the wild-type virus. The wild-type virus, as expected, destroyed the tumors [but] killed the animals [as a result of] myositis, whereas the recombinant virus destroyed the tumors and the mice survived.
A few mice, very late after they had been treated [with the recombinant virus], did develop myositis. That was shown to be due to the emergence of a revertant [phenotype]. So the virus can, under certain circumstances, mutate or delete the inserted sequences and get back to its original tropism. But that is a pretty low-frequency event, and we were able to achieve this outcome of tumor disappearance [without] toxicity.
What’s the next step? Can this particular virus be clinically relevant?
Yes, this virus could be clinically relevant, and we’re exploring the possibility that we might take this engineered virus into the clinic for the treatment of patients with multiple myeloma. Multiple myeloma patients are quite severely immune compromised, and for that reason one wouldn’t wish to take any kind of risk using a virus that could cause disease in the patient. Although wild-type, unmodified CVA21 is being used in Australia as an experimental approach to treat patients with malignant melanoma, I don’t think one could justify using the unmodified virus in patients with myeloma. So we may actually proceed [with testing the modified virus in patients].
We do at the Mayo Clinic have a manufacturing capability to make clinical-grade virus. We also have the ability to do the necessary toxicology studies the [US Food and Drug Administration] requires.
We have already designed, built, preclinically tested, manufactured, and taken through to the clinic a couple of recombinant measles viruses. And we are at this time taking through a recombinant vesicular stomatitis virus. So this engineered CVA21 is something we are seriously considering putting through that same pipeline.
Beyond this virus, though, we think there are important applications elsewhere. One that we are actively working on at the moment [involves] vesicular stomatitis virus, [which] is a negative-strand rhabdovirus that infects cattle and has for many years attracted attention as an oncolytic virus in many animal models. But it has not yet been taken through to clinical testing because it has quite serious neurotoxicity.
There are various ways one can try to address the neurotoxicity, but this microRNA targeting may provide a really efficient way of doing that. So we’re currently making VSVs into which we’ve built neuron-specific microRNA targets in the hope that we can fully address the neurotoxicity of that virus and possibly make it suitable for in vivo use and clinical studies.
The other thing is that the creation of vaccines obviously requires that you somehow modify a virus in order to take away its natural pathogenicity but retain its immunogenicity. We think [our approach] may have potential for achieving just that goal with emerging or existing viruses.
One virus of interest in that regard is poliovirus. We know the polio vaccine can sometimes mutate and become neuropathogenic, and cases of polio that occur in the Western world are typically due to reversion of the vaccine-strain virus. It might be possible to introduce this type of modification to that vaccine strain in order to reduce the likelihood that it could revert to a neuropathogenic phenotype.
The implications for the vaccine world are definitely there.