No mapped reads to IFNg in macrophages?

[DonDolowy]

Dear all,

I am doing an RNA-seq analysis of 50bp PE of macrophages stimulated with LPS at 0hrs, 2hrs and 24hrs. The dataset consists of WT and KO samples and duplicates of each sample.

I have now mapped it using Tophat 2.0.8 and Bowtie2, but when inspecting the results in IGV, I do not get any mapped reads at all to Interferon Gamma, which I know is expressed in macrophages. This is also verified by qPCR.

I am using iGenome mm9 and the command I run is:

tophat -p 8 -r 200 --library-type fr-unstranded -G genes.gtf -o WTA.R1 genome R1.fastq R2.fastq

Do you know what went wrong?

Thanks a lot.

Best,

Kevin Dalgaard

[JamieHeather]

Hmm that is odd.

Have you tried just manually looking for IFNg containing reads in your data? You could just try BLASTing or grepping for a short sequence, just to check that there are reads containing IFNg sequences.

This probably isn't the proper way to check it, but it's what I'd do. Then if there were IFNg reads in the data, you could see what they end up mapping to (if anything).

[A.N.Other]

Do things other than IFNg behave as predicted - is this just isolated to this gene or is it a global issue? Check out some stuff in the same pathway. Could be that you have a mismatched genome and gtf file, for example, which would mean some (but not necessarily all) annotations would be wrong.

[DonDolowy]

I went back to the raw fastq files and grep´ed for sequences that that are unique to IFNg. No hits at all, whereas I got hits from different kind of house keeping genes, and macrophage specific genes. Just not IFNg (and also the other interferons).

For another RNA-seq dataset we have very little (or no expression) as well, so this suggest that there might be a technical problem when constructing the library (TruSeq). We will try to do qPCR on the library fraction and see if we get anything there.

[JamieHeather]

Out of interest, what kind of macrophages are these? Do have any functional readouts from them post-stimulation (or ideally any IFNg ELISAs or something) or verification of phenotype? I'm just wondering if maybe these are MDMs where the differentiation might have gone wrong, and you're just looking at monocytes or something on those lines.

Are there any other macrophage related genes missing in your analysis, other than the interferons?

[DonDolowy]

They are bone marrow derived macrophages stimulated with LPS. Samples are from unstimulated (0hrs), after 2hr and 24hrs of LPS stimulation.

We also have ELISA data showing that IFNg for sure is present. TFN expression in the RNA-seq dataset is as expected.

[JamieHeather]

Hmm, very odd!

[DonDolowy]

Just got the qPCR results from the library leftover. It shows no presence expression of IFNg as well. I will try to get hold of the RNA samples from our collaborator and check them.

Since we are not a immunology lab, we contacted some colleagues who have performed RNA-seq and microarray on primary BMDM from C57BL6 and from GM-CSF cultured BMDM. In their dataset they see very low expression of IFNg for primary BMDM, whereas for the cultured ones they also have no expression at all.
Same samples just analyzed with microarray show some expression, but it is so low that it is below the threshold for what one would consider as being expressed.

Some thoughts are that we did polyA enrichment. IFNg should be polyadenylated, but maybe the transcript has some unique properties? My first concern was that IFNg mRNA is relatively short, so with a library size selection of 300bp it would maybe not be included in the library, but with its size of 1231nt that should not be the case. Unless, it for some reason is resistant to fragmentation during the library preparation.

[kevinrue]

interferon type-1 absent in infected bovine alveolar macrophages

Dear all,

I have a similar observation for type 1 interferons in our bovine alveolar macrophage (I saw you mentioned "the other interferons").

Did anyone get to the bottom of this "interferon absence" issue ?

I looked at all aligned reads (even those mapping to multiple/too many loci using STAR): I can only see a handful of our 20 million paired-end 2x90bp reads per library mapping in the region of interferon genes, but most landing seemingly randomly in the intergenic regions.

As a "positive control", there is one gene flanking this "interferon region" which shows significant expression (many reads piled up and well aligned to exon junctions).


Among our 127 samples, only a few of them show non-zero counts for type-1 interferon genes (less than 5 reads out of 20 million anyway).

Also, I looked at similarly short transcripts (plotted the log2CPM against the gene length), and I could find several examples of short(er) genes with detectable (even high) expression.

Meanwhile, using RT-qPCR, we did observe detectable expression of IFNB1 in a group of biological replicates (infected macrophages 48 hours post-infection). This was even enough to detect statistically significant up-regulation of IFNB1 relative to uninfected cells.

Given the tight regulation of interferon genes, is it possible that their turnover is so rapid that they are degraded during library preparation? (assuming they all have poly-A tails, does it get chopped too quickly before our poly-A purification step?)

Note:
we did find detectable expression of IFN-gamma (detectable = over 1 cpm in at least 10 biological replicates among our 127 individual samples). Even so, the expression remains close to the lower bound of detection.


I can see that the latest post is from 2013. I was hoping that by now someone may have further insights into the issue?

Many thanks in advance