Gene therapy with AAV: too risky for a cure?

Blogging on Peer-Reviewed Research

Many medical conditions today are treated but never cured. Imagine, a child with a genetic disease like juvenile diabetes or hemophilia. This child will be taking expensive medications for their entire lives. In the case of some diseases the cost of the medications might be more than child or their parents can ever hope to earn in their lifetimes, much less spend on a life-saving drug.

This is one of the many reasons why people have placed such great hopes in gene therapy. If a disease results from a defective gene, and we could replace it or supplement it with a functional gene, perhaps we could have cures instead of treatments.

It seems simple.

But the many challenges are many. And we may have decide which risks are worth taking.

Obstacles to overcome
One challenge is the immune system. Our bodies are always ready to attack foreign proteins. Just like a child who was put up for adoption, isn't always recognized as family when they show up unexpected at the door, our bodies aren't likely to recognize a protein that we should have made but didn't. Even if it's missing because of genetic change, it will be seen as a foreign protein if it shows up unexpectedly, without proper warning.

Another challenge is cancer. One of the ways that cancer can begin is if a virus integrates into a chromosome and turns on genes that should be kept shut off. Human papilloma virus, the virus that causes cervical cancer, is one such case. HPV can insert itself into the genome and turn on genes for cell growth. If cells grow in the absence of their normal checks and balances, cancer can result.

How is this related to gene therapy?
A big part of gene therapy involves putting the correct gene into the genome to provide the protein that's missing from the gene that's defective. This requires getting the gene into the proper cells and somehow getting the gene into the chromosomal DNA.

How do you put a new gene into a chromosome? We look at how the experts do this in nature and see if we can't use the same techniques. In nature, the ones with the greatest expertise are viruses. AAV is one example.

AAV or adeno-associated viruses are viruses that can infect cells, and can integrate into chromosomal DNA through the work of enzymes that cut the chromosomal DNA and splice in the DNA from the virus.
Unfortunately, the genome is big and at this time, there's no way to control where the virus inserts itself. It's well known that viruses can integrate in innocuous positions or into parts of the chromosome are better left alone.

I'm not sure that we know much, right now, about why AAV picks one site over another.

In the July 27 issue of Science, researchers found some cases in mice, where the virus did indeed pick an unfortunate site to join a mouse chromosome (1). In their study, approximately one third of the mice that were treated with an AAV vector developed liver cancer. The researchers decided to investigate the sites where the virus joined the chromosomes.

They used PCR to create copies of the DNA at the junction sites, determined the DNA sequences from the PCR products, and used a sequence comparison program (probably BLAST although the article doesn't mention this) to determine where those DNA sequences were located in the mouse genome.

When the junction site DNA was examined more closely, they found that the integration sites from four tumor samples all mapped within the same 6, 000 base region in chromosome 12. Surprisingly (to me anyway), in the entire 3 billion base genome, two of the integration sites were only 12 bases apart, in the mir 341 micro RNA transcript.

I don't know if the researchers looked at all the sites where the virus integrated. To me, though, finding four integration sites so close together, makes me wonder if AAV integration isn't really that random.

It would be unfortunate if this location in chromosome 12 (or the human equivalent) is one that's preferred by AAV. Liver cancer isn't the sort of outcome that anyone would desire.

1. Donsante A, Miller DG, Li Y, Vogler C, Brunt EM, Russell DW, Sands MS. 2007. "AAV vector integration sites in mouse hepatocellular carcinoma." Science 27;317(5837):477.

More like this

You may want to add a note that all AAV gene therapy trials have been shut down since the patient's death on July 24. The death was unrelated to the concerns about AAV integration, though.


Are you sure all AAV trials are on hold?

As far as I've seen, only the specific AAV trial where the death occurred has been put on hold. The following is from the FDA statement issued on July 26:

FDA is not aware of similar adverse events occurring in other gene therapy trials either with this specific product or with those that use other genes in AAV vectors. However, as a precaution, the agency is further reviewing all ongoing trials involving any use of AAV.

Ah something I know about, so I can put my in my 2 cents. I was lucky enough to research using AAV to knock in GFP reporter gene research primary human foreskin fibroblasts. To use AAV, you package a maximum of 4.7kb with the AAV. In the front and rear of your construct you have sequence that is homologous to the gene of interest. The virus then drops in the intervening sequence with extreme specificity (which is the whole reason it was chosen to use in clinical trial). The virus itself is extremely efficient also, some labs like those at UofWashington have exceeded 90% in cell culture.

As far as I know the deaths in the AAV trials were due to the patient samples being exclusive to terminal subjects, and some problems with toxicity in the delivery system. Which has been the biggest hurdle to overcome to make AAV gene therapy viable.

I no longer am researching AAV, but in my experience researching using viral vectors, AAV holds huge promise.

Interesting post. But know insertion is NOT random but designed into the vector using homologous leading/lagging sequence to the destination. And AAV's are extremely efficient at delivering its vector and only infecting the targeted cells once.

By Alex Spencer (not verified) on 06 Aug 2007 #permalink

Thank Alex for an interesting twist.

The Science paper that I wrote about didn't say much about the flanking sequences that they used in constructing the vector. It sounds like, if you're getting homologous recombination, that the flanking sequences are really quite important and perhaps that's why this group saw the results that they did.

I agree that this holds a lot of promise. Hopefully, Targeted Genetics can figure out why the one patient died and how to address the problem.

If anyone is curious for more explicit information about AAV, below is the reference for a review written by David Russell, who oversaw the research I was part of at OHSU. It explains thoroughly how vectors are constructed, their integration, the use of helper viruses, and their applications.

Blood, Vol. 94 No. 3 (August 1), 1999: pp. 864-874

Adeno-Associated Virus Vectors and Hematology

By David W. Russell and Mark A. Kay

By Alex Spencer (not verified) on 06 Aug 2007 #permalink

Alex, I don't think that's right.

As far as I know, most AAV-based gene therapy vectors are not designed to integrate into a specific site. They don't contain the homologous targeting sequences you describe.

I believe most groups doing human gene therapy with AAV emphasize the potential for such vectors to persist in an extrachromosomal state. In fact, if you browse the Targeted Genetics web site, they claim AAV vectors don't integrate. (See the first paragraph under "Leadership in AAV.")

That seems incorrect, though. For example, the Russell & Kay review you cite states:

Cloning and sequencing of chromosomal junction fragments from cells transduced with neo shuttle vectors has shown that most vector genomes integrate by nonhomologous recombination at random locations as single-copy proviruses with terminal deletions.

This paper shows the same thing.

Note - whatever the truth on AAV integration, I agree with Ian that it's not likely related to the recent patient's death. In that case, the patient become ill only a week or so after being injected with the vector. It's hard to imagine a realistic scenario where aberrant integration could cause illness that fast.