On the Sunday of Hurricane Irene, I sat in the study in my home watching (or rather listening to) the trees whir, creak and groan around me. After the storm, walking through our neighborhood, we could see that many trees were down but some, like the hundred year old oak next to my house, were still standing.
What was I doing to keep my mind off Mother Nature during the storm? I was contemplating something far smaller in size. Cricetulus griseus to be exact – or the Chinese hamster. Chinese hamster ovary or CHO cells are the most commonly used cell line to produce therapeutic proteins. Pharmaceutical companies use CHO cells to produce tons of proteins on a global scale. Inevitably, some CHO cell proteins get in the mix. We are beginning to use our tools to determine whether having hamster-derived protein, or “host cell protein” (HCP) in the mix with protein drugs might be good, bad, or indifferent.
Since CHO-derived proteins are usually drugs for human use, they are purified to incredibly high standards and any amount of the HCP in the final product is tiny. But it is well known that CHO proteins are generated during the culture of the protein, and can therefore contaminate the final product. These HCP can potentially generate immune responses in the humans to which the drug is given.
Indeed, in two previous studies (one by Ingerslev et al., on FVIII-deficient patients, prior to and post factor VIII treatment), CHO proteins clearly induced immune responses. Anti-CHO antibodies have even been identified in 54% of individuals who have no history of exposure to protein therapeutics. The significance of these antibodies is unknown, particularly since the HCP against which they are directed is usually also unknown. If, by any chance, that HCP is similar to human (which it may well be, since both are mammals), then there is a slight chance that anti-HCP immune responses might cause problems for the human patient.
The publication of the CHO genome (see Xu X, Nagarajan H et al. in Nat Biotechnol) is a major step forward for many reasons. For one reason, it will be easier to target genes in the CHO genome and turn them on and off. For another, we can begin determining what CHO HCP are in the mix when protein drugs are produced.
Since we don’t yet have 2-D gels to work with, we selected a few highly secreted proteins to illustrate our approach. Once we selected the secreted proteins from the CHO genome, we applied EpiMatrix and ClustiMer, our standard immunogenicity screening tools. The proteins were then scored based on potential immunogenicity (setting aside Tregitopes, which is another chapter of the story).
Long story short? Considering that both humans and hamsters are mammals, the CHO proteins we looked at, that might be secreted (based on the bioinformatics signature) into cell culture supernatant are remarkably unlike human. Some have very few T cell epitopes (red bar in the negative zone) and cannot be expected to generate a very significant immune response, but some have lots of T cell epitopes (score higher than 20 in the graph). That’s not good, if you’re thinking about HCP from the immunogenicity perspective.
The good news? HCP potential for immunogenicity can be ranked, like any other protein, with immunoinoformatics tools, and the dominant (highest scoring) epitopes mapped and compared to human.
What does our ability to score CHO HCP and find epitopes mean? Well, right now, we don’t know, and we won’t know until we do some further in vitro studies. But should any of those HCP epitopes be much like human, and added to a drug – or even worse a vaccine (in which case the HCP would be administered with an adjuvant), the result could be autoimmunity . Or, if they are unlike human, and added to a drug, the result might be immunogenicity. Or perhaps the quantities of protein that are left over from CHO culture are so small as to render this hurricane afternoon contemplation completely irrelevant. But somehow I think not.
One thing I do know, is that because we are able and willing to contemplate these things, we may be better off. The trees that fell over during the hurricane were the big, tall, rigid ones. The ones that are still standing are the ones that are more flexible, and can bend in the wind.
Like the trees, we can bend with the wind. The CHO genome is now available. We have the option and perhaps, even the obligation to consider the CHO genome as a source of information for HCP. We now have tools in our hands to answer some critical questions. Is it Human? Is it Hamster? Is it potentially immunogenic?
For more thoughts about CHO and immunogenicity, catch me at one of my upcoming talks. They are posted here. Or write back to me (comment on this blog below) and give me your opinion. I’ve given you mine, and now I’m interested in hearing yours.