Hi Sam, this is a good question! Let's see if I manage to explain it clearly enough, although conceptually it is not difficult: if you want to compare two samples processed in the same way, you have to use same amounts (that's what you do when you compare an RNA from a control and a treated sample), but if you want to compare two samples processed differently, as in your case (RNA-IP vs. RNA extraction), you have use same proportions.

So, you do your RNA-IP from a certain sample. You have to quantify in some way this sample (mg if it is a tissue, cell number, total protein). Your input will be equal or a fraction of this starting material for your IP (in the same quantification units), let's say for example it is 1/10th.

You do your IP and you do your input extraction, now you have two possibilities:

1) You resuspend the IPed RNA in a certain volume and the input RNA in 1/10th of this volume. If you do so, you keep proportions respect to the starting material for the two processings and all the rest is easy, use equal volumes of RNA to produce the cDNA and equal amounts of cDNA in the qPCR (I suggest to quantify your input and base on this the volume to use for the RT, your IPed RNA will be much less, maybe not even detectable, this is normal, but will be highly enriched in some specific RNAs).

2) You resuspend in equal volumes and you use for the RT 1/10th volume of the input compared to the IPed RNA (in this way you restore proportions lost when resuspending). Then for qPCR it is as above.

You could also resuspend the two samples in different volumes, but you complicate your life to recalculate the proportions!

One more point. Since RT could have different efficiency in the very different samples you are retro-transcribing, it would be good practice to add equal amounts of an in vitro transcribed exogenous gene to the RT reaction and use this gene in qPCR to calibrate your specific gene(s).

Also don't forget to show that genes that are not known to bind to your protein are absent or very low in your RNA-IP compared to the input.

I hope I was clear enough! You can check this procedure in the paper Di Napoli et al on my page, although the protocol is not as detailed as here!

Good luck.

Great! Thank you this is very helpful.  Would it also be okay to base proportions off of RNA concentration, and say use 1/10 the concentration of input for the RT reaction? ie: if I use 1 ug of IP RNA for the RT reaction, could I use 100 ng of input RNA for the RT reaction?  Thanks!

Hi Sam, well, I have to apologise, I'm very sorry... I guess on Friday night I was accusing some fatigue... Your last question made me go through my first answer to your doubts, I re-checked our protocols... and I realize I have completely inverted things...

In fact, you do have to use same amounts of RNA, NOT same volumes... [If you use same volumes from the same starting material you will clearly bring no change in the concentrations, actually you might lose something in the IP due to partial efficiency of the IP itself...]. So YOU WERE RIGHT in the first place, you use same amounts of RNA. In the same amount of RNA the input will contain thousands of RNA species, while your IP only a few (if everything worked well), which means the few will be in much higher copy number to make up for the same amount of RNA... that's the enrichment.

In fact diluting one or the other of your samples would be arbitrary, because you could always find a dilution that works in the direction you want... I'm sure you can you see that...

So, go on as you were planning, 200 ng of each sample is fine, should be enough to see something in the input... Just try to calibrate your RT-qPCR with a known amount of an in-vitro transcribed exogenous RNA. And for the IgG control, since there should be almost nothing, in this case, yes, just elute in the same volume of your IP and use the same volume as for the IP RNA.

Sorry again... and good luck!

Yes, you can use same amount of RNA (200 ng) for RT reaction from RIP and Input. Since you are dealing with very little amount of RNA, it may be good idea to do Qubit estimation. Nanodrop can give you wrong estimate. Once you have done the qPCR you can always make a ratio as RIP/Input. Then you will get an idea whether your target is enriched in RIP or under-represented.

Thanks.

I met the same problem right now. I just curious how to calculate the qPCR? What kind of reference gene we should use? 

@Pietro Pilo Boyl: how to to calibrate your RT-qPCR with a known amount of an in-vitro transcribed exogenous RNA?

@Arunasalam Naguleswaran: how to make a ratio as RIP/Input? Just use Ct value? or have to calibrate to some genes?

Thanks!

You can create a standard curve from  serial dilutions of a template DNA (or cDNA) and  using this standard curve you can calculate  relative value for input and RIP samples and make a ratio. Since it is going to be a ratio you don't have to worry the units. Hope this helps you.

Thanks

Thank Arunasalam Naguleswaran and Nicolas Viphakone for the detailed answers. They really help me a lot. I really appreciated. 

Hi Xinxia, of course Nicolas's approach is one possibility. Other approach is what Sam was mentioning and I supported in my second post of this thread, to check the representation of a specific mRNA in equal amounts of total and IPed RNA... If an RNA is specifically IPed, it's representation in the IPed RNA pool will be much greater than in the starting extract to make up for the same amount of RNA, since the IPed RNA species will be so many less... but there is a sort of flaw with this approach if you use total RNA as starting material, since 90-95% of the RNA in that extract is ribosomal, so you actually work with 1/100th of mRNA, which might reduce the enrichment factor. The third possibility, which is what we preferred in our studies, is to add an in vitro synthesized RNA species, for example a GFP or a LacZ mRNA, to the RT mix. This RNA will be therefore distributed equally in all reactions, inputs and IPs. Independently of the amount of RNA you will use for the RT of the input and the IP, you can calculate the ratio between your specific gene and this exogenous gene in Input and IP, calculate the relative enrichment factor between input and IP (similarly to what Nicolas suggests, but in this case using an external normalization factor instead of relying on internal quantities and ratios) and you can then compare it to the same relative enrichment factors of a validated mRNA that interacts with your protein and a validated mRNA that does not interact. Depending if this relative enrichment factor is closer to or even higher than that of the validated interacting RNA or closer to that of the non-interacting RNA, you'll have a read out for your specific mRNA. You can also normalize the data to the validated binding mRNA, so that you will easily see for values of your mRNA around or above 1 that it binds well or for values below a threshold that you can set, for example 0.2, that it doesn't bind (in between it will mean it binds weakly). Of course you should try the same approach in a control IP, either with protein A beads only, and/or with an unrelated Ab, and/or on a knock-out or a knock-down extract using your Ab (in my opinion the best control, when possible), and you should find the relative enrichment factor significantly decreased...

The calibration curve Arunasalam talks about is independent from all this. You should run it of course for all the genes you will test in the qPCR (including the exogenous RNA, if you plan to use it) to make sure you will use the appropriate amount of RNA in both the input and IP to fall in the linear range for qPCR amplification allowing for some variability (higher/lower RNA amounts).

Hi. I would like to extrapolate Samantha's question to RIP-seq instead of qPCR. I have carried out RIP-seq to profile the mRNAs bound to my protein of interest (Ago2) in two samples-1&2 (undifferentiated and differentiated cells). Of late, we discovered that these two samples differ hugely in their total RNA content, which in turn has caused a bias in Ago2-pull down.

For eg., After doing Ago2-RIP-seq, mRNA-X is found to be enriched in sample-2 than sample-1 because of two possible reasons:

a)mRNA-X interacts more with Ago2 in sample-2 than sample-1

OR

b) mRNA-X is more abundantly expressed in sample-2 than in sample-1. So, it is more likely to be "pulled down" in sample-2 than sample-1

To normalize the bias in enrichment arising due to differential total-RNA levels in these two samples, we decided to do RNA-seq of 'Input'/'Total-RNA' to be used as normalization control for the Ago2-RIP-seq data . However, I have already done the Ago2-RIP-seq experiment and now, I need to do the 'Input'/Total RNA-seq.

Is there a requirement that the starting material/number of cells taken for Input/Total-RNA isolation should be the same as done for the Ago2-IPs for each sample? (Eg:For each sample, I'd used 20x106 cells per RIP-seq. Should I have to use 20x106 cells per Input/Total-RNA-seq, too?)

OR

I need to just take EQUAL AMOUNT of Input/Total-RNA from both samples-1&2 to make the libraries (irrespective of the starting cell number used for Ago2-IP experiment)? (Eg: Despite using 20X106 cells per RIP-seq, can I use just 5x106 cells for Input/Total-RNA seq as long as I take same number of cells for both sample-1 and sample-2)?

The rationale behind the 2nd option is that the final amount (in ng) of library DNA loaded for sequencing would be the same for Input or Ago2-IP libraries. So,what is more important is that EQUAL AMOUNT of starting material should be taken for THE TWO SAMPLES-1&2 UNDER COMPARISON, irrespective of the RNA species being sequenced (Input/Total-RNA or Ago2-RIP). Pls help me to know if this is the correct way to go about.