Author Topic: Quantify western blot with QUANTITY ONE  (Read 21818 times)

Bcrespo

  • Newbie
  • *
  • Posts: 33
Quantify western blot with QUANTITY ONE
« on: April 11, 2010, 05:00:00 PM »
Hi all,

I want ask you a question related with Western blot.

I have made WB to only see if there were proteins. But, know I need to Quantify them with Quantity One. I have my proteins bands, and the control (tubulin) in separate images files.

The question is for quantifying wich values are better the volume of the bands, the density of the bands. Do you rest   general backgound to all measeures, or do you get one backgorund in each lane of each band.

In summary,  HOw do you do to quantify your WB??


Jose

Bcrespo

ElkeKS

  • Full Member
  • ***
  • Posts: 158
Re: Quantify western blot with QUANTITY ONE (Bcrespo)
« Reply #1 on: April 13, 2010, 05:00:00 PM »
Hi Jose,

quantification of Western blots is inherently tricky, maybe more so than that of 2D spots.
I should say that I've not worked with QuantityOne a lot, so I don't know enough about that program. But I have some general questions.  What detection chemistry do you use? How do you generate the images, on a camera/scanner system, or with film? Do you run your samples in multiples, titrating to be sure to be in the linear range of your signal?

You can do one of two basic types of measurement to determine the raw intensities:
- a lane profile, where the background could be subtracted in a number of ways (straight line, rolling ball, etc.); this is especially useful if you have more than one band as a signal
- the volume of individual features, e.g. a box of equal size around each band, or a polygon that "cuts out" only the band, with background either from a box placed in a non-band area of a lane, or with a local background method from your software
The normalized intensity is then the ratio of background-corrected raw intensities of protein of interest / control.

But again, you need to be sure that your signals are in the linear range of your detection system, and in Western blots, there are a lot of steps that can saturate or be out of line. Think of loading capacity of your gel, transfer efficiency of the blot, protein binding capacity of the membrane, linearity of your detection chemistry, and of the detector (film, camera, scanner).

Hope that helps,
Elke


mysg

  • Guest
Re: Quantify western blot with QUANTITY ONE
« Reply #2 on: June 08, 2010, 09:27:06 AM »
Elke pointed out very good points: the linearity range of the detection, so let me chime in a little bit about what I usually do.

I usually use ECL+ reagent and detect signals using either STORM or TYPHOON. If you are DIGE and Typhoon user, this should be very easy for you. Since the linearity range of film scanning is narrower than STORM/TYPHOON data, I encourage to try this method. Just make sure signals are not saturated as you do the scanning for DIGE gels.

Once I get the images, I use Image J, which is pretty much NIH Image (free software made by NIH initiative), to open the gel images. I usually set up rectangular region of interest (ROI) with a consistent size. Move the ROI over to each band I want to quantitate, and use Ctrl+M (measurement) command, that gives me the average pixel intensity (averaged over the size of ROI). Then, I measure the background "adjacent" to the bands using the same ROI, but I don't do that for lane-by-lane. To be able to get single background ROI measurement, background needs to be relatively clean, homogeneous, which is very important. I use PVDF membrane from Whatman (Westran) that gives relatively low fluorescence and clean background for STORM scanning. Washing is also very important. I don't necessarily use low fluo PVDF membranes.

Once I get these measurement values from Image J, then I export/import text data to Excel. The number crunching in Excel is really simple. The average intensity of ROI that included bands reflect signals that is "above" the background. That means you can consider it as "volume" of the spot in 2DGE, but I have to subtract the background "volume" as well. So the formula will be:
((average intensity of the band containing ROI)-(average intensity of background)) x(size of the ROI)

I usually do these steps for normalization bands (usually tubulin in our lab), and normalize the bands of interest.
The final data will be just ratio between the bands normalized by tubulin data. Btw, we don't believe difference less than 50%, unless we run many gels and to the statistics.

Bcrespo

  • Newbie
  • *
  • Posts: 33
Re: Quantify western blot with QUANTITY ONE
« Reply #3 on: June 08, 2010, 02:58:17 PM »
Masaaki

Thansk very much, for your answer. It is very useful. I have been  had problems with the Image J, I cannot change from de next lane, i can only make new boxes up or down in the same lane.

I have a question about what feature measure. Do you measure volume or density? Why do you rest the tubulin and not divide the values?

thanks for your replies!! In may lab, nobody know how to do that!!!

Bcrespo
Bcrespo

mysg

  • Guest
Re: Quantify western blot with QUANTITY ONE
« Reply #4 on: June 09, 2010, 08:48:27 AM »
Masaaki

Thansk very much, for your answer. It is very useful. I have been  had problems with the Image J, I cannot change from de next lane, i can only make new boxes up or down in the same lane.

I have a question about what feature measure. Do you measure volume or density? Why do you rest the tubulin and not divide the values?

thanks for your replies!! In may lab, nobody know how to do that!!!

Bcrespo

Sorry for being tardy to respond, but let me answer as far as I can.

With regard to the Image J and the box, are you using some lane detection macro? What I'm doing is very simple method, which is just to use "rectangular selection" tool. In the Image J, there are buttons in the menu, and the left most button is the one I use. Simply click that, and make rectangular region of interest around your band. Now, to be able to do that, your bands need to be flat. If they are little bit tilted, then it's better to rotate your image, maybe a few degree. But, "smiled" bands are little hard to do this. Also, when the bands are touching each other, then it's tough. So, I usually use comb that has enough separation between lanes.

The trick here is to find out the best size of the ROI that accommodates all the bands of interest. If the ROI is tightly fit with one band but too small for another, then the size is not good. So, find out which band is the biggest, and find out the ROI size that fits with the biggest one. Once you do that, measure average pixel value using Ctrl+R command. By default, Image J measure average pixel value and the size of the ROI. Then, move that ROI to the next band, by simply grabbing the center of the ROI (if you grab the edge accidentally, then your ROI size will change, so grabbing the center is the key), and measure (Ctrl+R) again. That's how I do it.

I have to make some fresh Urea buffer....so let me answer other things later (but try ASAP). Masaaki

mysg

  • Guest
Re: Quantify western blot with QUANTITY ONE
« Reply #5 on: June 09, 2010, 03:33:22 PM »
I realized I hijacked this thread by talking about ImageJ, but....well still in the same perspective.

In terms of volume or density, if the size of the ROI is the same, they should mean the same thing. When I explain to myself, I always think like this. The band looks something like mountain range that stands up on the ground. Please look at my crappy line drawing, attached as jpg image. All I want to know is the "volume of the the mountain range, which should be proportional to the amount of the protein of interest. If the ROI size is the same across the bands and background, the (mean intensity value) x (size of the ROI) will give me the volume, as long as the "base" is subtracted.

Does this make all sense? I may be abstracting too much, but this figure sometimes helps to convey my image.

By the way, the number of pixcel intensity may come up reversed (white being zero, black being 255 - 8bit), and you want to "invert" the image in that case.

Bcrespo

  • Newbie
  • *
  • Posts: 33
Re: Quantify western blot with QUANTITY ONE
« Reply #6 on: June 10, 2010, 08:52:36 AM »
Many thanks Masaaki for your big explanation.
I try Image J as you mentioned and it worked perfectly, I was using the Gels analyze option. And It was became myself crazy.

I have other question: How do you pool data from different gels. I mean, I did to western to have replicates, the problem is that the bands of each western are not similar (for each replicate: due to, protein load, 1 o 2 antibody efficiency) and the standard desviation could be very high. I was thinking in treat easch western separately and see if the differences between samples in each western have similar tendency.

I have in some cases when I did : (sample value)-(tubulin value) a negative value!!!! do you have similar values?? How do you treat them? I made an image capture of my data in the excel with the negative values.

this discussion is very interest, i am learning a lot. Thanks guys.

Bcrespo
Bcrespo