School of Dentistry, University of Queensland
Name: Agnieszka Lichanska
Title: Lecturer, School of Dentistry, University of Queensland
Professional Background: 2005-present, lecturer, oral biology and pathology, University of Queensland; 2002-2005, post-doc, Institute for Molecular Bioscience, University of Queensland; 1999-2001, post-doc, Queen's University of Belfast.
Education: 1998 — PhD, biochemistry, University of Queensland.
VANCOUVER, BC — Agnieszka Lichanska is a member of the University of Queensland's functional genomics group and a lecturer in its department of oral biology and pathology in UQ's School of Dentistry in Brisbane, Australia.
While the primary focus of Lichanska's work these days is in studying periodontitis in humans and metabolomics in humans and mice, Lichanska also worked for several years on projects studying growth hormone receptor in mice and came to know firsthand the reasons why spotted arrays are often discarded in favor of more expensive commercial platforms.
During her presentation at the World Microarray Congress held here last week, Lichanska discussed some of the results she and other researchers at UQ, including UQ Institute for Molecular Bioscience professor Michael Waters, have had in studying growth hormone receptor, and why struggles with the IMB's in-house arrays led the group towards commercial options. BioArray News spoke with Lichanska following her presentation.
Perhaps you can give me some information about your background and how you wound up studying GHR in mice?
I finished my PhD in Brisbane and then I went over to Northern Ireland, to Queen's University of Belfast to work as a post-doc. We started doing some microarrays there in addition to our work on inherited, conventional diseases. And when I came back [to Australia] I got a job working with Mike Waters in IMB and started looking at the growth hormone receptor, because that's what Mike has been doing for 30 years or so.
This was a time when microarray technology was just starting to be used, and IMB started an internal resource center where they were printing their own microarrays. For the first study the platform we used was printed in-house. They were doing this with cDNA and later Compugen [array oligonucleotides], but we had issues with those arrays.
When I started working with Mike, the mice were specifically being bred for this project. You have to breed them five generations before they actually become homogenous — before you can use them in experiments. We knew that the results were that they were smaller and that they had big problems with growth and stuff. But we had to wait until we could run the arrays.
What kind of samples are you taking from them?
We use liver tissue. So you kill the mice to get the liver out. And we were putting it directly into RNA [samples], rather than freezing because freezing is too slow, unless you chop it into really small bits.
You mentioned that in your experiments you found some correlation between genes involved in GHR and cancer genes. Can you tell me a little more about that?
What we looked at was in some situations GHR can be translated to the nucleus, but in general it doesn't happen in a normal biological situation. However, if you do a partial hepatoectomy, the liver can regenerate, and during this time you see a relocation of the GHR to the nucleus and it actually activates proliferation. So the same thing has been absorbed in cancers — where the GHR is actually in the nucleus in a lot of cancer cells. But nobody knew why it's there and what the significance of that is. The work we've done shows that there is correlation between the nuclear localization of receptor and the expression of cancer markers in blood cells and also proliferation genes such as survivin. We also saw down-regulation of some tumor suppressive genes, and up-regulation of cancer markers.
Have you published that?
We are waiting for a few more experiments. We sent it off to Proceedings of the National Academy of Sciences, but they wanted to see a few more experiments done. Hopefully, it will be accepted.
You said that you started off with spotted but then moved to a commercial platform. Why did you make that decision?
We were hoping to use spotted arrays, but we had issues with the quality of the arrays that we were getting. IMB wanted everyone to use their in-house microarrays and that's why we were using their arrays. The second problem was that the way the Compugen arrays were prepared was that each gene was spotted only once. So something that usually happened during hybridization was that you lost a spot — and we couldn't rely on those. Another thing we saw was that fold changes were really compressed on those arrays, so when you actually looked at your data it would change about 1.5 fold and every time you did a Northern blot it was like five or six fold difference. So the changes were quite compressed. Most arrays do compress the changes. But I knew that once we put the mice on, which had really been stimulated, we were going to lose a lot of the changes. I mean we only identified 20 genes that were differentially expressed on those BaF cell lines. And that's not a lot. I would expect a lot more than that.
Why are you using Northern blot instead of real-time PCR to validate your results?
It was easy and quick. We had a lot of material to start off with — so you can run Northerns — that's not a problem. We had more experience running Northerns than having to prepare optimized real-time assays. And, unless you are using TaqMan, which at the time was quite expensive, you had to optimize every primer set. It just did not work and it cost a lot of money as well. It's easier because you can also target the region that you are interested in by using the Affymetrix array or by the Compugen [oligos].
So what were the advantages of using Affy's platform? What results did it have on your work?
[The results were] huge. We can actually see small fold changes that we know we can trust because we know from other sources that they are real. Secondly, they are much better annotated. We had annotation problems with the spotted arrays because I don't know what they did, but [IMB] didn't follow the annotation that was developed wasn't the actual annotation that should have been there.
What happened a few times was that they didn't print the slides in the order that they said they printed them. So we worked with the annotation for the chips, but two weeks after we already did some analysis they would come and say, "Oh no, that was wrong."
Also, when you actually have to prepare the probes, it does take a long time. Also the technicalities, putting the hybridization solution and inverting it — it was technically quite challenging to do, so when we switched to Affy it was quicker and easier.
During your presentation, though, you said commercial arrays had some short-comings. That there needed to be better annotation, an optimized analytical approach, and better access to raw data.
[There needs to be greater access to] raw data especially because we have been working on growth hormone and there's been a lot of work on growth hormone. And they have been done across a lot of platforms, which introduces other problems. But most of the data at the US National Center for Biotechnology Information is in a soft format, which means it has already been normalized. And there's a difference between that and the raw data that you can actually take and apply your own statistical analysis, your own normalizations and do it the way you would do it for your own array. So that limits what you can do with the data.
Has there been any progress in changing that situation?
Yes. In [NCBI's Gene Expression Omnibus database] they now encourage you to submit raw data, rather than analyzed data. So for our recent stuff we have used their raw data.
So what's the next stage for the study with mice?
We will be completing the experiments for the PNAS paper and we will also be looking at mice to analyze the signal in the receptor. We will see if those mice get tumors early, compared to the wild types or whether there will be other problems. We don't know what will happen.
Growth hormone is known for the post-natal regulation of growth but it is present in the embryo, but no one knows what it does. It is there, the receptor is there.
And what's the objective of this work?
We were hoping to find some genes so that [someone] could design drugs to prevent proliferation, but whether that will happen or not — at the moment it doesn't look likely that that will happen — has yet to be determined. But growth hormone receptor has been linked to a number of cancers. It is likely that there is a link. So, the ideal is to prevent this proliferative stage but keep all the goodies from the signaling — so we are really looking at inhibitors of particular signaling pathways.
And what is next for you?
I am actually working on two other array projects. One is very similar — we are looking at metabolic profiling of mice and we are hoping to expand that to humans, but the other project, quite a lot of it is periodontal investigation. That's totally different as I am part of the School of Dentistry right now.