5
$\begingroup$

I'm reading this fantastic article on estimating body time: Molecular-timetable methods for detection of body time and rhythm disorders from single-time-point genome-wide expression profiles and one of the things that is not very clear to me is how the researchers estimated which genes are expressed and which ones are not:

Total RNA was prepared by using Trizol reagent (GIBCO􏱾BRL). cDNA synthesis and cRNA labeling reactions were performed as described (5). Affymetrix high- density oligonucleotide arrays (Murine Genome Array U74A, Version 1.0, measuring 9,977 independent transcripts) were hybridized, stained, and washed according to the Technical Manual (Affymetrix). Affymetrix software was used to deter- mine the average difference (AD) between perfectly matched probes and single-base-pair-mismatched probes. The AD of each probe was then scaled globally so that the total AD of each microarray was equal. The resulting AD values reflect the abundance of a given mRNA relative to the total RNA popu- lation and were used in all subsequent analyses

I'm not sure if I'm reading this correctly - did the researchers look at all RNA available in the cells and calculated the levels of messenger RNA produced by expressed genes? if not, how can the level of expression of a gene be estimated?

$\endgroup$

2 Answers 2

7
$\begingroup$

The technique described here is called microarray. Your question has given me an opportunity to put forth one of my opinions about certain problems of gene expression studies.

Gene expression is a measure of the activity of any gene. If the gene performs its activity in the form of a protein, then its expression should be a measure of the protein. If a gene makes a non-coding RNA then its expression is a measure of the RNA concentration.

[You can omit the cases of post-translational modification because of signal transduction. They are highly expressed and transiently but frequently used. ]

Like your example there are many studies which use mRNA concentration as a proxy for protein activity. This proxy works in many cases because transcriptional gene regulation is more frequently used mechanism for imparting stable changes. But the best strategy would always be to measure the proteins also.

Apart from microarray there are several techniques to measure RNA concentration:

  • RNA sequencing (high throughput)
  • Real-Time PCR (medium throughput)
  • Northern Blotting (Low throughput, semi-quantitative)

Many techniques exist for measuring proteins also:

  • Mass spectrometry (high throughput)
  • ELISA (Low throughput, quantitative)
  • Western blotting (Low throughput, semi-quantitative)

Usually the "proxy method" is used because protein quantification is comparatively more difficult. Antibody based techniques like ELISA and Western blotting have problems of cross comparison of proteins because of the variability of the antibody binding efficiency.

$\endgroup$
16
  • 3
    $\begingroup$ Unfortunately I don't remember the details but I recently saw a talk in a conference where it was demonstrated that mRNA levels don't really correlate with protein levels. I mean, we've always kind of known this but the numbers shown were shocking (less than 50%, perhaps as low as 30). I will try and get a hold of a reference and post back. $\endgroup$
    – terdon
    Apr 30, 2013 at 17:18
  • 2
    $\begingroup$ Wow, just found an article from 1998 stating that in yeast "We found that the correlation between mRNA and protein levels was insufficient to predict protein expression levels from quantitative mRNA data." And yet here we are, still using such data... $\endgroup$
    – terdon
    Apr 30, 2013 at 17:21
  • 1
    $\begingroup$ @terdon It varies dependent upon case, for many cases there is a relationship, but its hard to say apriori. This is difficult if you have a large number of probe sets of interest. As a rule of thumb, qPCR panels go up to a couple of score of genes or so and then they become difficult. $\endgroup$
    – shigeta
    Apr 30, 2013 at 17:22
  • 1
    $\begingroup$ It's my understanding that scientists often use these microarray experiments as high-throughput screening methods to identify either candidate genes or gene networks. If the data from one such experiment is truly to be trusted, it must always be followed up by traditional biochemical methods (mass spec, western blot, ELISA/RIA, microscopy). $\endgroup$
    – user560
    Apr 30, 2013 at 20:20
  • 1
    $\begingroup$ @leonardo: you are very right, nowadays that the microarray fad is gone not many journals would accept a study showing ONLY microarray data, without any further validation. Of course, the solution is to jump on the current fad, RNAseq :D $\endgroup$
    – nico
    May 1, 2013 at 12:47
2
$\begingroup$

Supplementary information from the Comments from the first answer:

I have perhaps found @terdon's reference? Not sure if this is it, but by measuring transcription events in individual cells, Taniguchi et al. certainly make the case that there is no correlation between protein and mRNA levels.

However microarray data (and most rnaSeq and qPCR experiments) are ensemble measurements; they measure the average mRNA concentration in a sample of millions of cells. In the aggregate, the paper also says that in the aggregate, the average mRNA correlates to the average protein product concentration. See this nice writeup in the OmeSpeak blog.

Overall, what is happening microscopically: translation is a somewhat stochastic process in the individual cell and events have to be modeled explicitly. Biology in the aggregate (averaged case in a cell population) is often what we have to work with; single cell techniques are quite arduous. (run a protein gel on a single cell!). I think there is room for both sorts of information when investigating biology; certainly publication of work measured on more than one cell at a time will continue for some time.

All the work cited here is in E coli, but in Eukaryotes one assumes that this is not going to be any simpler.

$\endgroup$
1
  • $\begingroup$ gene expression noise is also a variable feature.. in some gene networks, the noise is low whereas in others it is high.. $\endgroup$
    – WYSIWYG
    May 1, 2013 at 4:35

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .