Global microRNA profiles in extracranial and intracranial malignant germ cell tumours

Bookmark and Share
Published: 29 May 2013
Views: 5630
Dr Matthew Murray - Cambridge Cancer Centre, UK

Dr Matthew Murray presents data on profiling extracranial and intracranial malignant GCTs at the 3rd International CNS Germ Cell Tumour Symposium in Cambridge, UK.

With regards to germ cell tumours, as we all know, they’re clinically and histologically very heterogeneous and that causes problems for us clinically when trying to classify these tumours for tumour diagnosis and for subsequent treatment purposes. Now obviously despite that one of the theories is that all of these tumours derive from a common precursor cell which gives rise to the different malignant subtypes we see. If that is the case then what we might hope to derive from biological examination of these tumours are fundamental biological abnormalities that are shared across the malignant subtypes. If so, such biological changes may be useful and attractive biomarkers for full tumour classification and diagnosis.

Recently our lab has been investigating microRNA expression in these tumours. MicroRNAs are short, non-protein coding RNAs that post-transcriptionally regulate gene expression. This regulation occurs through binding of the microRNA shown here with the messenger RNA target and what’s been shown is that predominantly results in messenger RNA degradation.

Important work has also shown that microRNA profiles reflect the developmental lineage of a large range of different tumours and therefore we undertook global microRNA profiling in malignant germ cell tumours hoping to identify common biological abnormalities.

So this was part of the study that we’ve done. It’s a CCLG approved study shown here and we have access to the UK bank of fresh frozen and formalin-fixed tissue of both extracranial and intracranial tumours. For this study we looked at 42 of those tumours; they reflected the range of histological subtypes we see in clinical practice. They were predominantly extracranial tumours but also included a small number of intracranial tumours too. We also profiled six teratomas and eight developmental and gonadal control samples.

After extracting RNA from these samples we hybridised it to the ExCom microarray platform, had it interrogated for 615 known human microRNAs and all the data which we derived was analysed using Bioconductor in the statistical software system R.

So this is the data for the 42 samples, so this is a heat map with overexpression of microRNAs shown in red and under-expression shown in blue. Each one of these columns represents an individual sample and each one of the rows represents a differentially expressed microRNA. Shown here are 65 microRNAs which were most differentially expressed in this group. The most important thing to notice is at the top of this heat map with the samples segregating completely with the non-malignant samples shown here on the right, so that’s gonadal and developmental control samples in green and the teratomas which are shown here in brown, and on the left hand side are the malignant samples constituting the yolk sac tumours, germinomas and the embryonal carcinoma samples.

The most striking feature in terms of dark red and dark blue is at the top of the heat map for these eight microRNAs here which in terms of the non-malignant samples have very low levels or no level, no expression at all, and are the microRNAs from the miR-371-373 and miR-302 clusters. Because this finding was so striking what we then wanted to do was look at how these samples clustered based on the expression of just these eight microRNAs. The paediatric data shown here on the left, this is the array data, we’ve got each of the eight microRNAs from these two clusters and what we see is that the expression of those microRNAs drives the samples into two clusters once again with non-malignant samples on the left and malignant samples on the right hand side.

We then went on and re-analysed data published, QRTPCR data now, published for microRNAs in adult gonadal tumours which included ovarian and testicular tumours. For the same eight microRNAs we see exactly the same findings with robust segregation between non-malignant samples here and the malignant samples on the right.

So this finding of over-expression of these two microRNA clusters occurs regardless in these tumours of patient of age, whether that be paediatric or adult, it occurs regardless of the malignant histological subtype and it also occurs regardless of the anatomical site of the tumour, whether that be gonadal or extra-gonadal. And, as I say, within that subset there are a small number of intracranial tumours which also behaved the same way as the extracranial ones with regards to these microRNAs.

We also looked at the individual malignant subtypes and in addition to identifying the same miR-371-373 and miR-302 overexpression in each of the individual subtypes, we also identified subtype specific microRNAs which might be useful in the differential diagnosis and classification of these tumours. For example, miR-375 is very elevated in yolk sac tumours, the miR-182-183 cluster in germinomas and, for example miR-520g in embryonal carcinoma.

So, to summarise the expression, and this is going back to the original histological classification slide, what we know from murine studies is that these microRNAs are at very high levels not only in embryonic stem cells but also in primordial germ cells. Under normal differentiation then the levels become very low or non-expressed and during somatic differentiation, for example in teratoma formation, the levels we see are also very low or at very low levels of expression. In the malignant subtypes that we’ve analysed, and obviously choriocarcinoma is rare and we haven’t looked at that, but in all the other malignant subtypes we see very high levels of expression and it remains to be elucidated whether this expression persists right the way through aberrant primordial germ cells or whether that becomes switched off and back on again to cause malignant progression. So all of these findings are consistent with a common origin theory of these tumours.

One of the things we wanted to do next was just have a look to see whether there were any evidence of microRNA differences by anatomical site. In the germinoma samples we had three intracranial tumours and ten extracranial tumours. So we had a little look at those cases in more detail and that data is shown here. We did find some differentially expressed microRNAs, three listed here, which were under-expressed in the intracranial samples and two microRNAs, miR-451 and miR-144, which originate from the same cluster, that were overexpressed in intracranial germinomas. Overall, though, the total number was only 5 out of 615 microRNAs analysed on the array so just under 1% so there certainly weren’t a significant number of microRNAs that we identified.  That notwithstanding, the changes that we saw were quite significant so this is validation separately by QRTPCR for miR-451 and we see that the levels range from approximately eighty-fold to over a thousand-fold higher than in the extracranial samples. The basis and the significance of these observations remain to be elucidated.

So, to summarise this section of the talk, with regards to commonalities what we notice is that the miR-371-373 and miR-302 clusters are overexpressed in all malignant germ cell tumours regardless of patient age, histological subtype or anatomical site of disease. This is consistent with a common origin theory.

Importantly, the miR-371-373 and 302 clusters, the expression is either non-existent or very low in normal tissues including differentiated tissues such as teratomas and that lends itself to the potential use for these microRNAs as clinical biomarkers. With regards to looking at some of the differences, the number of differences we saw for germinomas between intracranial and extracranial was small. Importantly, the microRNAs we identified were not neural specific microRNAs so they didn’t just represent the analysis of normal brain tissue alongside tumour tissue. So, for example, miR-124 is a brain specific microRNA and levels of that weren’t different between the sets of samples. So this suggests that there may be some small level, potentially, of biological divergence in these tumours compared to extracranial germinomas.

I mentioned a slide ago about the potential for these microRNAs as biomarkers for this disease and the properties of microRNAs are such that they lend themselves to clinical use in the hospital setting. So this is a study published a number of years ago now which is one of the first studies of the demonstration of microRNA detection in the serum back in 2008. What this group showed was that actually the sample can be left out on the bench without processing for up to 24 hours and these levels shown here are the CT values from PCR for three different microRNAs. It shows that the levels are remarkably stable even if the sample is left out and that is not something that would be the case with, for example, attempting to measure messenger RNA levels. Similarly the sample can be freeze-thawed numerous times and the results are very reliable. So this work suggests that looking at microRNAs clinically may be useful.