I'm studying a gene, which shows significant poor clinical outcome according to TCGA database in HCC. However, when I investigated its functions, it displayed tumor suppressor phenotype, like inhibiting proliferation, migration, self-renewal... but it can also upregulate some oncogene targets. I have no clue how to explain this inconsistant results with TCGA clinical samples.
I appreciate if anyone could give me some advices.
Hi Vyvi Li,
the answer below does not respond you straightforwardly, because I don know nothing about this gene, but I think it gonna open an avenue to pursue the answer.
Your gene shows significant poor clinical outcome according an specific database, then you noticed this gene is a tumor suppressor, i.e., this gene should work decelerating carcinogenesis, but it expression is correlated with malignancy.
Both experimental and theoretical results show that high cell death rates promote increasing in tumor size in long term. Expression of a strong anti-apoptotic protein Bcl-2, another mechanism to avoid apoptosis, was found to be associated with a favorable prognosis. It is a paradox, one may expect a tumor with high expression of Bcl-2 be associated with large size and and malignancy, but indeed those tumor with low expression of Bcl-2 present growth in long term and malignancy.
If you need an enlightenment about this aspect of cancer biology, I can send you a couple of manuscripts.
Cheers!
1. What is the direction of the association between your gene and clinical outcome in the TCGA database of HCC?
Specifically, is it upregulation or downregulation?
2. What is the proportion of affected cases compared to all patients in the poor outcome group?
(Boxplot or hazard ratio of a regression model should make above obvious).
Now ask;
3. Is the direction (effect) of your gene in the TCGA data different from your functional analysis results?
4. If yes, does the proportions in 2 justify a single mechanism or suggest the potential existence of divergent machnisms?
5. Have you considered other factors?
a. TCGA samples are from bulk tumours with a variety of cell types and signals.
b. Different isoforms of the same gene can have divergent functions, which may or not lead to a poor outcome.
Consider a hypothetical scenario with a gene "A".
1. A's main function is to promote proliferation, with overexpression to 20 units causing cancer.
2. Overexpression of A to 40 units then causes further growth, increasing somatic mutations and poor outcome.
3. But, A has a second context-dependent function in cell differentiation, with down-regulation supporting epithelial to mesenchymal transition and poor outcome.
4. Additionally, A has two binding partners "B" and "C", binding with B promotes migration and metastasis while binding with C induces apoptosis. Here, even with increased expression of A, the phenotype depend on the interaction with either B or C.
Once you have the direction of your effect, then ask the question of how well your in-vitro model reflects the complex TCGA mixture.
You can apply a deconvolution algorithm on the RNA data to adjust for tumour purity.
Good Luck
Hi Alexandre Sarmento Queiroga,
Does that mean I should concern about the long term study? Actually I do need some manuscripts about this aspect of cancer biology, especially how to design this kind of long term study both in vitro and in vivo. Another question confused me that do you think whether the overexpression fold would change the phenotype? Like overexpressed 10000-fold of target gene in cells vs overexpressed 100-fold of target gene in cells would have different functions?
My email address is [email protected].
Thank you so much for your apply!
Alexandre Sarmento Queiroga
Hi Chinedu Anthony Anene,
1. The gene is upregulation in TCGA database of HCC;
2. The Boxplot make above obvious;
3-4. I think it may exist divergent mechanisms;
5. I used stable A -overexpressed HCC cells and stable A -knockdown HCC cells for in vitro study, not sure whether this common cell model could reflect the complex TCGA mixture or not. Do you think whether the overexpression fold would change the phenotype? Like overexpressed 10000-fold of A in cells vs overexpressed 100-fold of A in cells would have different functions?
6. I don’t know how to apply a deconvolution algorithm on the RNA data to adjust for tumour purity…
Thank you so much for your apply!
Chinedu Anthony Anene
Hi Vivy Li
Yes, overexpression levels can influence phenotypes. Using your example, 10000-fold may cause resource overload, expression imbalance and unselective protein-protein interactions, while the same gene at 100-fold will have a balanced physiological effect. See this paper for more information Article Quantitative nature of overexpression experiments
.Regarding RNA based tumour purity, you can download pre-calculated ESTIMATE scores for TCGA data from https://bioinformatics.mdanderson.org/estimate/ the original paper is https://www.nature.com/articles/ncomms9971.
I can lend a hand if you need help with a fresh calculation.
Cheers
Hi Vivy Li,
about the questions bellow:
1- Another question confused me that do you think whether the overexpression fold would change the phenotype? Like overexpressed 10000-fold of target gene in cells vs overexpressed 100-fold of target gene in cells would have different functions?
Chinedu Anthony Anene answered this question straight to the point.
2 - Does that mean I should concern about the long term study?
I don't know exactly. You have suppressor tumor gene that is related to poor prognosis, and it somehow surprise you (as far as I can tell), than I just make a comment that could help you to go deeper on your discovery. I hope my comment indeed made sense for you question.
- Badges
- Science topic
- Similar topics
- Biological Science
- Histology