Protein stuck to your DNA? Denaturants like SDS will wash them away. Leave those out, & proteins are there to stay & you can use an Electrophoretic Mobility Shift Assay (EMSA) to track them on their way! We usually think about gel electrophoresis as a way to separate proteins OR nucleic acids (DNA or RNA). BUT what if you have a combination of both  EMSA, sometimes referred to as a gel shift assay can be used to detect RNA/DNA-protein interactions by putting a molecular “GPS monitor” on the RNA/DNA and seeing if it gives any other molecules “piggyback rides” that will slow it down on its journey through a gel. 

note: refreshed from 10/15/19 & video added on 11/24/21

In the past, we’ve looked at a method called the “slot blot” or filter-binding assay that I’ve used (way more than I’d have liked to) to measure RNA/protein (or DNA/protein) binding,

But that’s not the only way to measure it. Another common way is EMSA (gel shift assay) – so I’m gonna tell you more about it today!

Gel electrophoresis (which I’ve talked about A TON in past posts, so check those out if you’re unfamiliar) separates molecules by using ➖charge/➕charge attraction to send charged molecules through a gel mesh. The mesh acts as a sort of sieve – bigger things get slowed down more so they travel more slowly & won’t have traveled as far when you stop the run. The molecules move at a rate that depends on their charge (more charge, more motivation to push through the gel), their size (the bigger they are, the more friction they’ll encounter along the way), & the gel meshiness (tighter mesh will slow molecules down more).

Common types of gel electrophoresis are agarose gel electrophoresis (these horizontal gel slabs “knit” from sugar chains of agarobiose are good for separating big DNA pieces) and various types of PolyAcrylamide Gel Electrophoresis (PAGE) which are thinner, vertical gels made up of chains of polyacrylamide cross-linked to one another to form a mesh. You can make tighter meshes with polyacrylamide than you can with agarose so PAGE gels care good for separating small things and are often used to separate proteins and small nucleic acids) . More here:

We commonly use denaturing page when separating proteins – proteins start out as chains of amino acids but get folded up into complex 3D shapes. These shapes allow them to carry out various functions, but they will interfere with how the molecules travel through the gel. To get an accurate idea about how long a molecule is – a proxy for its molecular weight – which depends on the # of amino acids in the chain – you need to unfold the protein. You can do this with the detergent Sodium Dodecyl Sulfate (SDS), which coats the protein chain in a slippery coat of negative charge that prevents the protein from folding back up and ensures a consistent length/charge ratio. more here:

The idea behind adding SDS is that it makes all the proteins look the same except for their length. This “genericness” allows you to “subtract out” all the unique things about those proteins (like what amino acids its made up of) so you measure length only (like getting everyone to stand up straight, chin up, shoes off! before measuring their heights) 

But now you basically have a bunch of spaghetti noodles that aren’t recognizable by things that usually bind to it. So if you want to see if something usually binds to it, you have a problem… You need to run a non-denaturing (“native”) gel. DO NOT add SDS or urea (which erases RNA’s shape). Leave them be. This way the molecules can still interact (BUT it also means that you can’t use a molecular weight ladder to see how big they are). more on native PAGE here: ; YouTube:

To track it you label one of them, usually the DNA or RNA (more on *how* below) – we call this labeled thing the probe. So you mix the protein and probe, let them (potentially) bind, then send them on their travel through the gel. Usually you use non-denaturing TBE-polyacrylamide gels or (if your complex is really big, TAE-agarose gels)

Depending on how strong the binding is, they might separate along the way BUT thankfully we have a caging effect – even if the molecules come apart briefly on their way through the gel, the mesh’s pores trap them so close they’ll bind back again before they get separated – but you still will probably see a bit of a smear

If the probe binds another molecule, it will become bulkier so will travel slower through the gel &, when you stop the run, it will have traveled less far from the starting line than the free molecule, so it will show up as higher band.

How high? It’s not as easy as just comparing to a ladder because, unlike with SDS-PAGE, we have shape to take into account – & charge! SDS doesn’t only erase shape, it also provides an even ➖ charge. Proteins naturally have different charges because they’re made up of different amino acid building blocks, some of which can be charged (➖ or ➕) depending on things like the pH. Since charge difference is the motivating factor for moving through the gel, differences in charges will affect how the molecules travel. So a bigger binding partner might not necessarily slow a protein down more than a smaller binding partner – it depends on their size AND charge. Kinda like giving a piggyback ride to a heavier friend that’s shouting encouraging phrases as you travel. And the DNA/RNA adds a bunch of negative charge – so even more confounding of apparent sixes – so you’re gonna have to settle for comparisons (what does bound vs unbound run like)

You only have the tracker on one of the molecules, so you canNOT tell WHAT it’s interacting with BUT you can tell if it’s taking longer than usual to travel through your gel. If you start with pure protein & DNA, you know there’s only 1 type of friend to hang out with so it could “only” be with Bob.

But if you’re using unpure protein samples – things like “crude lysates” (basically cracked open cells) – you won’t know what that thing is – you can follow up and figure out what it is by using things like mass spec or, if you have a suspicion about what it probably is, do a Western blot which is another type of “probing” which uses antibodies against a protein to see if it’s there. Or add an antibody to the mix to look for a “supershift.”

But, once you determine where to expect it if it gives a piggyback ride to different friends or #s of friends, you can guess who they’re hanging out with based on where they are (kinda like if your GPS tells you they’re at a certain address & you search for that address in Google Maps and figure out it’s Bob’s house. Now, when you see that your protein’s at that address you can assume it’s hanging out with Bob. It’s not quite so simple with electrophoresis, because the addresses change depending on how long you run the gel for, but the relative positions should be similar.

Additionally, even if you know it has to be Bob. You don’t know how many Bobs there are piggybacking (what’s the stoichiometry, the number of one binding partner compared to the other. Sometimes a probe can bind multiple proteins so you’ll see multiple bands.

Will it hang out with Bob because it really likes Bob or because it really likes having friends? Any friends? Nucleic acids are – charged, so + charged regions of proteins might stick to them in a sequence-independent manner.

To make sure you’re only measuring *specific* interactions, you can add non-specific competitors – these are usually repetitive DNA fragments such as sonicated salmon sperm (seriously – if you ever wonder why biochemists have salmon sperm in their freezers this is probably why) or poly(dI-dC) – they “sop up” things that bind any ole DNA so you only see things that bind *specifically* to your probe. you want to add these BEFORE you add the labeled probe so they get sopped up first, before they have a chance to non-specifically stick to your probe

How much does it want to hang out with Bob? You run a series of increasing concentrations of the potential binding partner to get an idea of their relative affinity. If they’re really strong binding partners, you won’t need much for the molecules to link up but if the affinity’s weaker, you need a higher concentration – kinda like more peer pressure. 

What you’re usually looking for when you do binding studies is a value called the dissociation constant, Kd. This is the concentration of one of the binding partners at which half of the other is bound (e.g. the protein concentration at which half of the RNA/DNA is bound). So if you do that dilution series, the (apparent) Kd (Kd app)will be the protein concentration at which half of the RNA/funs normal & half runs up-shifted). more on Kd here:

This method typically isn’t as good as other methods like slot blotting (filter-binding assay) for determining Kd – but there are some advantages of the EMSA over the slot blot…

In slot blotting, you take a mix of nucleic acid & protein with different concentrations, like with the EMSA – but instead of using a gel to separate them, you use vacuum filtration – you use a vacuum suction to suck them through a stack of membranes – on top is a nitrocellulose membrane that traps protein (and any nucleic acid bound to it) but lets through free nucleic acids. Waiting below to capture those free nucleic acids is a nylon membrane whose + charge the nucleic acids find positively delightful. 

You can then compare the amount of nucleic acid on the top membrane (protein-bound) to the total amount (top + bottom) to get the proportion bound – plot them out & find the Kd. Since you have this clear binary readout – bound or unbound – it’s great for quantifying. Much easier than with EMSA smears but…

With the slot blot you get no size and/or stoichiometry indication – what matters is protein or not protein – it’s a yes/no readout. Good if you have a really pure protein with known binding stoichiometry (how many copies of 1 thing binds the other thing (e.g. 1:1 or 2:1)

But with EMSA, the more proteins bound to the RNA and/or the bigger the proteins, the bigger the shift you’ll see. So if 1 RNA piece can bind 2 proteins at once, you can get a “supershift” – it’ll get slowed down more than if just 1 protein were bound. But – if, for example, you find that 1 RNA is running as a complex with 2 proteins, you can’t tell if the RNA itself is bound to both proteins or if the RNA is only bound to one of them but that protein’s also bound to another protein. 

Now that we know what to look for, let’s talk more about how we can see it – our “GPS monitors”. We can’t see DNA or proteins with our naked eye, so we need a way to tell where in the gel they are. Common GPS monitors are radiolabels and fluorescent tags.

Radiolabels are like implanting a GPS tracking chip into the molecule – from the outside you can’t see its there – but if you look at the right thing you can detect its presence. Other molecules can’t tell its there so they interact with it as usual so this method gives you the most “actual condition”-ness

They work by replacing one (or more) of a molecule’s normal atoms with a radioactive isotope of that atom (it has the same biochemical properties as the “normal” one, but it has extra neutron(s) that make it unstable. Therefore, over time it gives off radiation that we can detect. We use ³²P (a phosphorus atom with 17 neutrons instead of the normal 16) to label the 5’ end. DNA & RNA have a sugar-phosphate backbone and at the 5’ end there’s a free phosphate. We can use a phosphatase to remove that normal phosphate and a kinase (T4 polynucleotide kinase) and “hot ATP” ([γ-³²P]ATP) to replace it with a phosphate containing ³²P. more on radiolabels here:

Another benefit is that these methods are super sensitive which means we don’t have to use very much radioactive stuff to detect very tiny amounts. Which is good because radiation can be dangerous, so we take a TON of precautions when working with it. We have a dedicated “hot room” where we do all our radioactive work, we work behind plexiglass shields, wear gloves, lab coats, safety goggles, etc., dispose of any radioactive products in dedicated containers, and use a Geiger counter to scan for radioactive contamination. Needless to say, this can dramatically lengthen how long it takes to run an experiment – and it can be stressful 

So many people have turned to safer methods like fluorescent tags. if you shine the right wavelength of light on the gel, the tags will absorb it and spit it back out at a different wavelength. more here:

They may be safer, but they do have disadvantages – these are more like ankle monitors. Sometimes, they don’t interfere with anything and the other molecules might not even notice those tags are there. BUT other times those other molecules notice the tag and stay away – it’s not that they’re biased or afraid of getting a bad rap for hanging out with “criminals” it’s just that that tag may get in the way of how they usually interact. If the other molecule usually attaches by grabbing onto that ankle and now there’s a bulky ankle monitor where it wants to grab on, it might not be able to. Or maybe it doesn’t recognize the tagged molecule. So, radioactive labeling is still the “gold standard”            

Leave a Reply

Your email address will not be published.