How has the bumbling biochemist been spending her day? Measuring interactions between protein & RNA! I’ve talked a lot about proteins, and a lot about RNA – but what’s even cooler is when the 2 come to play! But how to study interaction between protein & RNA? The filter-binding assay provides us a way! You may have noticed some revitalized reposts, but idle I’ve been not – I’ve been spending most of my days wrangling with this SLOT BLOT!
I love proteins & RNA – and sometimes specific ones love each other too and they can “get married” (bind aka associate) – so [protein] + [RNA] -> [protein-RNA] but they also can get “divorced” (unbind aka dissociate) – so [protein-RNA] ->[protein] + [RNA] and this is all really dynamic – molecules marrying and divorcing all the time.
Say you want to know how much a protein and an RNA molecule like each other. If you had a single protein molecule and a single RNA molecule you could mix them together then track their “marriages” & “divorces” & marriages & redivorces. If you were only to check in on them occassionally, the more they like each other the more likely they are to be married when you spy on them.
But it’s really hard to track single molecules – and in your cells there are tons of molecules anyway – so we can scale things up – use lots of copies of the protein & the RNA and instead of tracking them over time we take a single “census” after we give the molecules enough time to come to a dynamic equilibrium (rates of marriage & divorce are constant so there’s no net change even if the couples themselves are changing). Since all the copies are “the same” the more you see that are “married” compared to single, the greater the affinity between the two.
How many are married depends on the concentration of the partners. When the protein/RNA couples “break up” they go their separate ways and they can only get “re-married” if they find new partners – and like them enough to bind. And in order to stay stably bound they need to like them enough to stay married.
In order for an RNA to be protein-bound when you go looking, it has to
- find a protein to run into
- decide to stick to the protein
- stay stuck the protein
The more protein there is around, the more likely 1) is to happen – and the higher the affinity of the interaction, the more likely 2) & 3 are. 2 & 3 depend on the inherent chemical properties of the interaction partners (one molecule’s “Prince Charming” is another’s warty toad) and the environment
You can think about each time an RNA meets a protein it has a certain probability of sticking and each couple has a certain probability of divorcing – and those probabilities aren’t gonna change if you change the concentrations but the more copies you have the more likely you are to be able to detect them so your signal will be higher. But 1 does change with concentration – because the less protein there is, the less likely an RNA is to run into it.
So you can do a serial dilution of the protein (try out a number of protein concentrations (e.g. 200nM, 100nM, 50nM…) while keeping the RNA concentration constant to decrease the chances of the RNA running into a protein so we can pull some numbers out of the data. more on serial dilution – cereal-style here: http://bit.ly/2P4GcOR
One way we can take a molecular census of married vs divorced protein-RNA (or DNA but I’m an RNA person) couples is with an experiment called a “Slot Blot” or a “Dot Blot” or a “Filter-Binding Assay” – the basic idea is you mix protein and RNA, let them reach a steady marriage/divorce rate (binding equilibrium), then separate the couples from the singles & compare. You do the separation by using vacuum suction to pull them through a membrane sandwich – on top is a nitrocellulose membrane that the protein *can’t* get through – but free RNA *can*. And, waiting bellow it to capture that free RNA is another membrane that RNA *can* bind
But RNA can only make it to the lower membrane if it’s free – remember the protein sticks to the top – so if the RNA’s stuck to the protein and the protein’s stuck to the top membrane, the RNA will get stuck up there too. And then you compare how many singles vs couples there are by comparing how much RNA is trapped where
RNA is invisible to us so we need a way to “see it” – Usually we label the 5’ end cuz it’s easiest. We add a radioactive phosphate group to the end. more here: http://bit.ly/2lb0O9U
Note: you can’t use fluorescent labels with these membranes because the linker is hydrophobic so it sticks to the nitrocellulose membrane & doesn’t flow through (so all your stuff would get trapped on top)
You compare amount of RNA on top membrane to total amount of RNA (top + bottom) to get the proportion of RNA that’s bound. And you plot the concentration of the protein vs. the proportion bound to get a curve. Ideally you set up the experiment to cover a range so that at the highest protein concentration you have 100% of the RNA bound (all on top membrane) and at the lowest protein concentration you have “0”% of it bound. The point you’re most interested in is the “halfway point” – the concentration of protein at which half the RNA is bound & half is unbound (so equal amounts of radioactivity on both membranes). We call this the Kd (aka dissociation constant) and you can learn more about it here: http://bit.ly/2JyBbuj
On a more practical note, here’s the gist of how we do it (adapted from when I was writing up a guide for some colleagues so it’s a bit technical sorry). These are those kinds of experiments where the multichannel pipet and color-coding play a key role! Because there’s a lot of wells – if you do a serial dilution with 8 concentrations, that’s 1/12 of a block – and you can only blot 1/2 a block per 48-slot-blot. and if you want to do replicates…let’s just say it’s been an exhausting few weeks and there’s a lot more to come! http://bit.ly/2m5nzfV
First you do a serial dilution of the protein – deep well blocks are good for this – You start with WAY more of the protein than the radiolabeled RNA (even at your lowest concentration point) This way, when RNA binds protein there’s still a ton of protein left to bind. So in the whole [protein] + [RNA] <-> [protein-RNA] scheme, when you take some protein out of commission by moving it to the [protein-RNA] side, it’s like removing a drop from a bucket – so you can think of the concentration of free protein as constant in each of the wells – if protein’s Prince Charming, you don’t need to worry about copies of the RNA “competing” for Prince Charmings. Instead, what you want is each RNA deciding for themselves whether to bind based on how much they like the Prince, not how many Princes there are.
This is only true if the concentration of the radiolabeled RNA is way below the Kd of the interaction. If the RNA concentration is too high, so much of the protein will get bound that it lowers the amount of free Princes in a meaningful way, so you get what’s called “ligand depletion” – to avoid this you want to stay at least 10x under the Kd. And you want the protein concentration series to range from ~100-fold less – ~100-fold more than the RNA concentration (all these concentrations are molarity-wise because we care about the # of molecules and if we went by weights we’d be deceived by bigger molecules)
And speaking of ways to be deceived – you need your proteins to be really pure – which, thankfully in this lab, we’re pretty good at :). If your proteins aren’t pure you can’t be sure that it’s actually the protein you think doing the binding. And even if it is, there could be other stuff in there with it that affects how well it binds. So you usually use proteins that are recombinantly expressed (genetic instructions for the protein are cloned into circular pieces of DNA called plasmid vectors that we can stick into cells to make protein from it that we then purify out using various forms of protein chromatography) more here: http://bit.ly/2jURpCH
Once you’ve done your dilution, you add radiolabeled RNA and mix. And wait – you need it to reach that dynamic equilibrium so that molecules are marrying & divorcing at the same rate so at the same time one couple’s divorcing, another’s getting married.
So, once you reach this equilibrium you can take a census at any time and you’ll get the same numbers of singles vs. couples (even if the couples themselves aren’t the same as they were a millisecond ago). So once you reach equilibrium that count won’t change, but what that count is depends on how much of each you started with. note: this isn’t a “true” equilibrium technique because you disrupt things when you separate them & stuff – but it’s “close enough”
While all that dating drama’s going on you can start prepping for the “census-taking.” We use a vacuum filtration apparatus (slot blot). The sample gets pulled through with the help of vacuum suction and the hardest part to this assay (experiment) is getting an even suction… We have a Biorad Bio-Dot® and Bio-Dot SF Microfiltration Apparatus. The slot format (SF) lets you do 48 samples. The bio-dot format lets you do 98 but I could never get much success with it – it’s really hard to get an even vacuum. Even with the SF format, it’s a lot of fidgeting with knobs and trial and error.
There’s a “generic” base part that gets used with both the SF & biodot “upper parts.” The base gets attached to the vacuum suction line and there’s a valve you use to control air flow. Depending on whether you’re using the SF or the biodot format, you put in the appropriate middle plastic part and rubber gasket. Pre-wet a few filters (empty tip-box lids make good wetting baths). You might have to play around with the number of filters to get the best vacuum, but 3’s been my go-to lately.
Put the filters in, then put a pre-wet nylon membrane, followed by a pre-wet nitrocellulose, being careful not to introduce bubbles. Then put on the upper frame. Apply the vacuum and tighten the screws stepwise in a diagonal fashion for an even suction. When you get a good suction you can hear a change in the vacuum noise to a smoother, less screechy noise. At that point I loosen the screws and tighten them so they’re just “not loose” – at this point the suction’s doing the real work and if they’re too tight you get uneven suction
Speaking of which – you check this by first pipetting wash buffer into all the slots and checking that the liquid goes through evenly – this pre-washing is also important for getting the membranes ready. If it’s not even, try setting up the vacuum again.
If even, you can proceed to load your sample. Let the sample flow through and then pipet in wash buffer again. Once it’s all flowed through let it keep going for a short bit (minuteish) to make sure it’s really all flowed through and help it dry so you don’t get bleeding between the wells. Then unscrew the top, remove the membranes (using tweezers), and turn off the vacuum.
Transfer the membranes to a piece of saran wrap, allow them to air dry, and wrap them. Then expose them to a phosphor capture screen in a cassette. Let it expose, then scan it & quantify the bands.
One of the reasons I had so much fun reading about the studies on cracking the genetic code – Nireberg & Leder used similar experiments to test radiolabeled tRNA binding to ribosomes – only tRNAs that matched the mRNA sequence would bind the ribosome and (since ribosomes (although they’re largely made of RNA) are pretty bulky – the tRNAs would then get stuck on the membrane) 10.1126/science.145.3639.1399
more on those experiments: http://bit.ly/2lT8jma