randombio.com | Believe All Science | Science Is Truth | Science is All Sunday, January 24, 2021 Building and using a high-quality Western blot imaging systemYou can build an imaging system for the lab for 1/8 the cost, and get better results by understanding how they work |
n the past decade, chemiluminescent and fluorescent staining have almost entirely replaced colorimetric staining for Western blotting. However, many scientists don't take full advantage of their imaging systems. These devices are much more than simple cameras, and understanding how they work is essential in obtaining quantitative data.
I've been horrified to see people trying to analyze images in JPEG format or images taken on a cell phone, or even trying to use Photoshop to quantitate their bands. I've also had people complain that their imager was no good and demand that we buy a film processor, only to discover that they had set the lens aperture on their imager to f/22, effectively blocking 99.7% of the light.
In this article, I will describe what every researcher needs to know about getting valid data from fluorescent and chemiluminescent Western blot images. I will also describe how to upgrade a low-end DNA gel imager and how to build an inexpensive laboratory-grade imager.
Western blot of c-src in cultured cells imaged by chemiluminescence in a custom-built imaging system, 2-minute exposure. Three hot pixels are visible
Pixel depth The most important criterion is pixel depth, also known as bits/pixel, which determines the number of grayscale values a pixel can have. The pixel depth defines the maximum possible dynamic range. The actual dynamic range is set by the analog-digital converter, or ADC, in the camera. A low pixel depth will make faint bands impossible to see, while strong bands will be saturated, making them falsely appear to be identical. An imager with a resolution lower than 16 bits per pixel is not suitable for scientific imaging.
Even if the software produces a 16-bit image file, it does not necessarily mean you have a 16-bit ADC. Vendors might not know what their machine has, so it's a good idea to get a live demo. Create a histogram on an image and examine the actual numbers. A 16-bit ADC will produce all 65536 distinct grayscale pixel values. A 14-bit ADC will put three zeroes between each number. A 12-bit one will put 15. Of course, programmers could pad the values or low-pass filter the image, which would fill in the zeroes, so this technique isn't foolproof.
A 24-bit per pixel color image can appear to be grayscale, but it is not. A 24-bit color image has only 8 bits of dynamic range, for a total of 256 separate values. It's called a 24 bit image because there are three channels (red, green, and blue). If the R, G, and B are all the same, the image appears to be in shades of gray, but it is still technically a color image and is unsuitable for scientific purposes.
So, what you want is 65,536 shades of gray. Fifty is just not hard core enough!
Pixel value | 16-bit ADC | 14-bit ADC | 12-bit ADC |
---|---|---|---|
. . . | |||
32704 | 145 | 145 | 145 |
32705 | 144 | 0 | 0 |
32706 | 142 | 0 | 0 |
32707 | 139 | 0 | 0 |
32708 | 139 | 139 | 0 |
. . . |
Note: if the camera is trying to do video, it will produce fewer bits per pixel because ADC resolution is a trade-off with speed.
Lens diameter Lenses are described by their f/number and their focal length. For example, an f/1.2 50 mm lens has a large diameter lens and a focal length of 50 mm. A big lens, meaning a small f/number, is needed to avoid long exposures, which means more camera noise. This noise can be quite significant: with a small lens, it's often impossible to see any signal at all in a chemiluminescent blot regardless of how long you expose.
There are two parts on a lens that turn: one focuses, and the other is the aperture control, which changes the f/number by closing down the amount of light that gets through. For photographing charging wildebeest there may be times when you need that. For Western blots, keep it at its widest setting. In fact, I recommend gluing the aperture ring to its maximum setting so people can't change it by mistake while they're trying to focus.
Technically there are three parts that turn if you include the lens itself. Turning this one causes the lens to fall off and smash onto your blot. Don't laugh, it's happened.
Focal length determines distance from the blot and magnification. With a long telephoto lens you might have trouble fitting the entire blot into the image, even if you could focus on it.
Cooling For chemiluminescence, a cooled CCD camera is essential because you will be exposing for 5, 10, or even 30 minutes. Without cooling, you will still get an image, but it will have more hot pixels. Hot pixels make quantitation harder, and you need a dark frame correction or a software noise filter to remove them.
TIFF file format Use a computer to examine the smoothness of your image. It should look like the image above. If it looks grainy (or worse, blocky), or if there's visible banding, or if the file can only be saved as a JPEG, it means the camera is not good enough.
Time exposures Make sure the software allows you to take unlimited time exposures. Some cameras close the shutter automatically after 5, 20, or 60 minutes. Some don't have a shutter at all, but stop acquiring an image instead. Cheap cameras close the shutter after ten seconds, making them useless for chemiluminescence.
Using the right exposure is important: if the exposure is too long, the bands will be saturated and useless. If it is too short, all your pixel values will be bunched together and your quantitation will be inaccurate. The goal is to keep the pixels of interest halfway to saturation. If possible, expose all your images for the same length of time and, for goodness sake, write the exposure time down.
Binning The software should allow you to "bin" the image, which trades resolution and image size for sensitivity. Binning of 2×2 gives you four times the sensitivity but only 1/4 the number of pixels. Binning is common for chemi, as a 5000 × 5000 pixel image is hard to email and hard to analyze.
Monochrome vs color Most imagers use monochrome cameras, because color cameras are limited to three channels with broad, overlapping wavelength sensitivities. This is fine for taking cat photos, but it's useless for fluorescence multiplexing. A color camera would also need 16 bits per channel or 48 bits per pixel, which would make the file size three times bigger: an uncompressed 25 megapixel image at 48 bpp would need 150 megabytes of storage.
Western blot of actin in cultured neurons, imaged by CFL488 fluorescence in a modified commercial gel documentation system designed for ethidium bromide. Even after stacking of ten frames and flat-field correction it still appears grainy, but it is still usable
Fluorescent Westerns are easy to run: after washing off your excess primary antibody, you add a fluorescently-labeled secondary antibody, incubate for 1h, then dry the membrane and image it. With some modifications, it's possible to convert those $4000 fixed-focus dark boxes sold for use with EtBr-stained DNA gels to do fluorescent Westerns. However, as seen on the image at right, the results are often less than spectacular.
We converted ours to do 488 nm imaging as follows:
View of blue LED and filter changer (looking up inside imager toward camera). The filter changer is attached to an aluminum sheet (with a hole in front of the camera, not visible here). The filter changer comes in two sections, so it can be rotated down to change filters without removing it. The FITC emission filter (which appears blue) is in front of the camera lens. The blue LED and FITC excitation filter are mounted near the camera
At right shows how a blue LED might be mounted. It's essential to use a good, stable power supply, not batteries. Since blotting membranes are not shiny like gels, they won't reflect glare into the camera, so the light can be positioned next to the camera. Most bare LEDs require a current-limiting resistor. See here for the resistor sizing.
We tried making an emission filter out of orange transparent acrylic (Amber 2422, Delvies Plastics, $14.71 for 12×12×1/4 inch sheet), as its cutoff wavelength is almost as sharp as a dichroic filter (visible in photo at right). It worked almost as well as the $267.52 dichroic filter.
We also tried using a 100 mW 402 nm laser, which we happened to have lying around, as an alternative light source. Putting a diffractive diffusion filter front of it worked spectacularly well, and fluorescent bands were so intense they were easily visible by the naked eye. However, the diffraction filter introduced laser speckle, which made it impractical.
Although a lens produced a speckle-free pattern, the lens-expanded beam contained many imperfections due to the non-Lambertian nature of the laser. Alternatives such as homogenizing light pipes are said not to be suitable for lasers; a flat-top beam shaper, which costs around $5,000, is typically recommended. A 100-mW laser is also hazardous to work with and requires an interlock to prevent accidental exposure. Even reflection from a specular surface is hazardous. (A near-infrared laser is even worse, as you cannot tell whether you're being exposed to it until it's too late. Infrared-absorbing protective goggles are essential.)
These high-intensity blue LEDs are no picnic, either, and looking directly at them is strongly discouraged.
Background fluorescence is the biggest problem in fluorescent Westerns. Low-fluorescence PVDF must be used. Nitrocellulose and ordinary PVDF give unsatisfactory results.
We found that the camera in our gel documentation system was quite small and gave poor quality images. The manufacturer realizes this and limits the exposure time to ten seconds. (We found we could type any number of seconds in the text box, but the software automatically changed it to 7.955 when we tried to acquire an image.) An easy solution is to collect ten or more images and average them. This is called stacking and it's trivially easy for software like imal to do, as long as the blot doesn't move between exposures. The goal of stacking is to obtain a smooth image, which is essential for quantitation.
Flatfield dialog box in Imal. Image #1 is your blot and image #2 is your flatfield image. This corrects for any uneven illumination, which must be eliminated for quantitative Western blotting
For quantitation, it's also essential to correct for flatfield, otherwise some parts of your image will be brighter than others. To do this, put a piece of high-quality white plastic in the darkbox, collect ten images, and average them. This only needs to be done when your configuration changes or when you change filters. An ideal source of white plastic is the white sheet inside an old LCD computer monitor. (Wait, don't tell me you don't save those!)
Paper is no good, as it is impregnated with fluorescent dye and fluorescent fibers.
Here are the steps we use:
Custom-made imaging system for chemiluminescence. The only thing still missing is a door. That goes on next week
Chemiluminescence is much more sensitive than fluorescence, but it requires great care and clean equipment to make it work. Chemiluminescence works with PVDF or nitrocellulose but not Nytran.
It also requires an imager with a cooled CCD. Our building didn't have one, so we built one around an SBIG astronomy camera, a Nikkor 50 mm f/1.2 lens, and Nebulosity software. Our custom-built imager created images as good as the GE ImageQuant and Fujifilm imagers, and it was more flexible. Just as important, it only cost one-eighth as much.
The old GE ImageQuants used Apogee Alta U4000 cooled CCD astronomy cameras. Apogee is no longer around and GE spun off its imagers and Typhoon scanners to Cytiva. These babies cost more than an Audi A3, but unlike an Audi, if something breaks it's not so easy to get anybody to repair it.
Astronomers typically select cameras based on the sensor specs. My preference is for the KAF-8300, which is still available in some cameras. Don't forget, it must be monochrome, not color. If you need color images or color fluorescent multiplexing, get an RGB or custom λ filter set and a filter changer.
Imagers are easy to build, but there are a couple extra points to consider.
Chemiluminescent Westerns take more effort than fluorescent ones, as you have to block them overnight before staining and wash them 6 times after incubating with the HRP-linked secondary antibody. The chemi reagent is also expensive, but you need much less secondary (about 0.15μl / blot), or 66 times less secondary than a fluorescent blot.
In our custom imager, I used an SBIG STF-8300M with a Nikon adapter and 5-position filter changer with a Nikkor f/1.2 manual lens. It works amazingly well. One possible solution for the darkbox is to use a stainless steel kitchen garbage can. They're fairly cheap and reasonably rugged (see photo above) and it's easy to make them light-tight. As long as you remove the foot pedal no one will know what it is. Just don't forget to wash it out first.
jan 24 2021, 10:31 am. expanded jan 26 2021. last updated nov 21 2021.
Photographing Fluorescently-Stained 2D Gels
This article discusses some of the common fluorescent dyes and describes
how to build a light source for photographing fluorescent 2D gels using
a CCD camera.
Ultraviolet photography with a modified D90 DSLR
Modifying a DSLR camera for infrared also increases its sensitivity to
ultraviolet (UV) light. The modified camera can be used to take pictures
in the near-UV region down to about 320-330 nanometers.
Shortwave ultraviolet photography
Photography at short UV wavelengths (200-300 nm) poses
many challenges, but it also opens up a new universe of things
that can't be seen any other way.
Controlling an SBIG STF-8300M astronomy camera in Windows and Linux