ere's my rant for today: I don't know about anyone else, but I've
had it up to here with the unreliability of Western blotting. I'm
mad as hell and I'm not going to do them anymore.
A while back, we had a medical resident in our lab who said he wanted to help out with our research. So we showed him how to run Western blots. He went through an entire box of gels, putting them in backwards at first, before he got one that was acceptable. Finally he ran almost screaming from the lab, never to be seen again. I think he's changed careers: he's probably now a monk sitting on a mountaintop in the Himalayas, rocking back and forth, muttering “Gels . . . gels. . . Atto . . . fluorescent blot . . .”
A colleague of mine once bought an antibody against some ion channel he was studying. Unfortunately, it had the same MW as vimentin, a highly abundant protein that tends to stick to other proteins. He started getting suspiciously strong bands and finally did an immunoprecipitation and had it analyzed. Sure enough, it was 100% vimentin. The commercial antibody was no good and there was no trace of the protein he'd spent two months working on.
Nowadays you have to use either fluorescence (preferably infrared) or chemiluminescence to detect the bands and you have to do a loading control such as actin. You also have to show the marker bands, otherwise reviewers, desperate to find something wrong with your paper, will nail you.
This is something they don't teach in grad school. Sometimes the band of interest shows up but the actin doesn't. Sometimes only actin shows up. And sometimes you just get a smear. With rodents and cultured cells, it's not too bad. If you have a rare protein and you're dealing with human samples, which are so variable they might as well be seven billion different species, it's a nightmare.
Alternatives
There are alternatives. I once did some tests on a TSQ triple-quadrupole mass spectrometer and found that it's just as sensitive as ELISA and—unlike Western blots—it's virtually 100% specific. You proteolyze your sample with trypsin, run it through a nano-HPLC column, and do MRM on the major peptide. Other proteases give longer peptides but they're less robust.
All you need is the peptide sequence. You don't need an antibody or even a protein reference sample. You just calibrate and identify the fragmentation pattern using a synthetic peptide. This eliminates the need to wait until some vendor gets around to selling an antibody.
Unfortunately, it's also a royal pain and the equipment is eye-wateringly expensive. When something breaks, it can cost $35,000 just to fix it. The sample has to be absolutely clean: only peptides and solvent.
Here's what we need: we want to put our sample in a box and have it tell us the amount of every protein accurate to at least 16 bits. Counting PTMs and assuming ten peptides per protein and ten fragments per peptide, that means over 10,000,000 data points for each sample.
They can do this easily with DNA and even RNA. Why not proteins? Fragmenting a sample with trypsin makes it easy to handle, but would create ten or more peptides for each protein, giving us a million peptides in a cell extract. Even UHPLC only separates, at best, 10,000 peptides over 48 hours. That's orders of magnitude too small and 48 times too slow.
Maybe something with quantum dots, narrowband radio waves, or quantum computers. If you're a physicist or equipment designer and have any ideas, it's a great chance to become rich.
Mice and rats up the wazoo
Now on to my rant about mice and rats.
A few months ago while on a grant review committee, one of those boring tasks I discovered are routine in the nightmare world of academia, I came to realize just how pernicious the current dogma is.
There is now a standard way to study a gene: create a knockout mouse and measure how it affects the biological parameter you're interested in. That's it. No other paradigm comes close.
On animal protocol forms, we have to state that we were using twenty animals per group—ten males and ten females. This is a direct result of the push by feminists to include female animals in research. It had some benefits for science, but it also doubled our use of animals.
It takes a toll. One female postdoc literally cried when she had to euthanize her rats at the end of an experiment. When I was doing Morris water maze experiments, watching the little rats happily swimming around in the water was great. They were like little pets. The day I had to euthanize them was one of the worst days I ever had.
When I worked in an industry-related place, our regulatory guy came up with a plan to do a tox study. A big part of it was the use of beagles. It was to be done by a CRO. Dogs are standard in pharma: it is not voluntary. The US government and the FDA set up rules that can only be satisfied by using dogs.
On our animal protocol forms, there is a space where we have to indicate where we considered alternatives. No one ever writes anything of substance there because there are none.
NIH is the worst, but it's the same everywhere: you either do animal experiments or a human study or you do nothing. Inventing new technology is a low priority in biology, but why? Physicists get funding for pieces of equipment that cost billions of dollars. Until somebody creates a virtual mouse so we can dispense with animals altogether, we need something like that in biology. Otherwise, we're not pushing the frontier; we're just filling the blanks: using fifty-year-old technology to find something new about things others discovered long ago.
The term “scraping the dregs” comes to mind. Now, I like dregs as much as anybody, but I hate scraping them. And I dread working with rats and mice.
Want to save the animals? Stop nattering and give us an alternative.
apr 02 2022, 6:22 am
