No discussion of protein purifying would be complete if you excluded Size Exclusion Chromatography (SEC). Especially since, of all the types of protein chromatography you may be usin’, size exclusion tends to cause the most confusion! Because, unlike in gel electrophoresis (e.g. SDS-PAGE), it’s the bigger proteins that win the gel mesh drag race when it comes to the size exclusion case! And, no matter what type of SEC (aka gel filtration) you’re doing (i.e. preparative or analytical), choosing the right column is critical! 

video added 3/17/22

SEC is a column-based method used to separate proteins by size/shape – either for:

  1. purification (preparative SEC), often as a “polishing step” after you’ve done other separations (e.g. affinity and/or ion exchange) or 
  2. to see if proteins interact (analytical SEC). 

Today I want to tell (funnly) the basic premise behind how it works and then take you on a “behind the scenes” look at what’s going on at the molecular level (what’s actually in those columns anyway?!) and what goes on when we do it in practice (I’ll walk you through what my crazy protein-purifying day was actually like. As you might remember if you’ve been following me the last few days, today is day 4 of epic protein purification week (5 days, 6 protein purifications, >60L worth of insect cells, not enough time for coffee…). Today I finished 2 purifications and started a second one (the 6th and final one for the week) for a grand daily total of 1 gravity-flow affinity chromatography column, 1 ion exchange chromatography column, and 2 size exclusion columns – with 1 more on tap tomorrow. Grad student life is fun but it isn’t a walk in the park – and speaking of “parking,” let’s get into some of that jargon-y stuff with a car analogy, shall we? (we, or at least I, shall…) 

Proteins are molecules that serve as molecular workers and they’re made up of long chains of amino acid “letters.” These letters are like “building blocks” that have generic backbones that let them link together through peptide bonds and, because the different amino acids have different unique side chains with different properties, they cause the protein to fold up into shapes perfectly suited for carrying out various tasks. 

There’s a huge cellular advantage to making proteins out of premade parts – it’s quicker and easier than having to make everything “from scratch” each time. Imagine if a car factory waited until it got an order before even finding the nuts & bolts & cutting the metal. But from a protein biochemist’s standpoint, the similarity makes it harder to separate them and see what’s what because different “makes and models” of proteins have the same parts – kinda like how a recall of one car part can affect multiple makes and models but a recall of a boat part won’t affect your car.

For example, unlike how you can add DNase (DNA chewers) to remove DNA from protein preps or add protease (protein chewers) to remove protein from DNA preps, if you add protease to protein preps, you’ll chew up the protein you want along with the one you don’t.

So we need different ways to tell proteins apart. We commonly use a type of technique called PROTEIN CHROMATOGRAPHY that separates proteins based on how different proteins interact differently with different types of resin (little beads) that we fill cylindrical glass or plastic columns with. 

There are different types of chromatography and a lot of them involve the protein you want sticking to the column under initial conditions (due to charge-opposite charge interactions in ion exchange chromatography, or some specific feature like a genetically-engineered affinity tag in affinity chromatography). If your protein sticks but others don’t, you can wash all those others off and then change the conditions so your protein gets pushed off. 

To help make the process easier, we often use a liquid handling machine, like an AKTA. I’m in a protein biochemistry/structural biology lab so we do a LOT of protein purification, thus these machines are in high demand. Thankfully, our lab’s super fortunate in that we have 4 of these AKTA FPLC machines. FPLC stands for fast protein liquid chromatography & it uses pumps to push your proteins through the column. More here: 

Proteins need a “push” to get through the column. When we use gravity flow in hand-packed columns, we rely on gravity alone (which can take a really long time and often has to be done in the really cold cold room). The FPLC gives gravity a hand with pumps that provide some push (at a flow rate you can control way easier than fiddling with stopcock angles).

By controlling the pump settings you can control how fast the liquid flows BUT NOT (at least not directly) how quickly your proteins move through the column, since each protein will interact with the resin differently. And that’s the whole point – if they didn’t you wouldn’t be separating anything just moving the proteins from one place to another. 

Speaking of interacting with the resin, unlike with affinity chromatography and ion exchange chromatography, where the proteins actually bind (reversibly) to the resin (either because of some unique feature like an affinity tag (e.g. 6XHis or Strep) in the case of “affinity chromatography,” or because of charge in the case of ion exchange, with size exclusion chromatography (SEC, aka gel filtration (GF)) the proteins and beads don’t actually directly interact, they just take different routes. 

Smaller proteins have to go the long way, so it takes them longer to get through and they elute (come off the column) later – which we see as peaks on the chromatograph. This chromatograph shows light absorption measured as the protein passes through a UV detector after coming off (eluting from) the column en route to a fraction collector, which we can set to collect different fraction sizes (e.g 0.5-1 mL), typically in a 96-well deep-well “block.” Protein absorbance is typically measured at 280nm wavelength (and nucleic acids at 260nm) – the more protein that passes by the detector, the higher the absorption and what you want to see are nice sharp peaks showing individual proteins coming off. 

To picture what’s going on, imagine you have a bunch of cars you want to separate. Affinity chromatography is kinda like having snow-covered roads. Cars with snow tires and/or chains can go through, but other cars get stuck. Then you can add “snow plows” to clear the road so the proteins can continue on their way. The stronger the protein interacts with the column the more “snowed in” it is and the more plowing’s required.

The “snow plows” we add are typically competitor molecules that compete for binding sites (so it’s kinda like the snow plow actually takes the protein’s place – so not a perfect analogy but proteins aren’t really cars either!). So you can gradually change the conditions to separate the proteins even further. 

This is not what happens with SEC. In SEC, none of the roads are snow-covered, but differently-sized cars have to take different routes. You have lots of cars traveling to the same destination – the fraction collector (who knew it was such a destination location!). 

The routes have tunnels with different clearance heights. Bigger “cars” are too big to fit through the tunnels, so they’re “excluded” from a lot of the roads. Thus they miss out seeing a lot of the inside of boring beads and therefore they have shorter trips and get to the final destination more quickly. 

Instead of concrete tunnels under overpasses or through mountains, the tunnels in SEC are tunnels through porous beads. The beads are often made up of agarose or some other sugar or some polyacrylamide if you need something stronger and/or finer. More on different choices in a minute, but the key thing  these materials have in common is that, from the protein’s perspective, they’re pretty “boring” as is – like well-paved roads they won’t slow the proteins down. 

The “rules of the road” in SEC are that cars have to travel the longest route they can – if they can fit through a tunnel they have to go that way. On the bright side, at least those roads are less crowded, right? 

But imagine you have car-carrier truck. If it’s empty, it’s not that tall, so it can fit through the tunnels. And if the cars it will carry are on their own, they can fit through no problem as well. But put the cars on the truck and now you have a bigger, taller complex that can’t fit through anymore. So it’s route’s restricted. It can’t go through as many tunnels, so it takes a shorter route and arrives more quickly at the destination (fraction collector). This is the basic of ANALYTICAL SIZE EXCLUSION CHROMATOGRAPHY

Usually, like today, I’m using PREPARATIVE SEC – I separate *all* of my almost-pure protein by size to make it almost perfectly pure and keep all of that precious pure protein. But in analytical SEC, it’s more like PAGE (PolyAcrylamide Gel Electrophoresis) in that you’re using it on just a small bit of *sample* to “get a look.” SDS-PAGE, for those who aren’t familiar, is a common biochemistry technique for separating proteins in a little gel slab to look (after staining) at the size and number of proteins in the sample. It’s denaturing – the SDS & heat unfold (denature) the proteins – so if you see multiple bands you can’t actually tell if they represent proteins that were actually interacting, or they were just both present in the sample. more on SDS-PAGE: 

But with SEC, you don’t unfold the proteins, so protein-protein interactions can hold. The interactions have to be strong enough to withstand the journey. In analogy-terms, the cars on the truck bed have to be at least somewhat held down so they don’t fall off – thankfully in the tunnels there’s not far to go so the cars, if they temporally slip, can get back on fairly easily.

Of course, if 2 proteins elute together it could just mean that they’re similarly-sized. But if they are significantly differently-sized they shouldn’t come out together unless they interact. What you usually do is run a sample of each protein on its own. The smaller protein will elute at a larger volume (later timepoint) and the bigger at a smaller volume (shorter timepoint). When you mix them, the peaks will shift.

A leftward shift in the chromatograph indicates binding – but not “too big” of a shift – if the elution is really early, instead of a nice interaction it indicates aggregation – proteins clumping together so that they can’t go through any tunnels. The void volume, representing what was already in the beads before you started and what went through without going through any tunnels, depends on the column size.

Sometimes, depending on the size of the proteins and the resolution of the resin, it can be hard to see a dramatic shift in the bigger protein’s peak. But what’s more obvious should be a disappearance of the smaller peak. The peak-shifting is indicative of binding, and then you can take the corresponding collected fractions & run an SDS page to see if you can see multiple bands.

There are lots of different types of SEC columns you can choose from (at least in theory, if your lab is fortunate to have a bunch of types like we do!). The most obvious difference in columns is size – SEC columns range from super tiny ones used for the analytical stuff to super big columns used for preparative SEC (and taking cool selfies). 

The longer your protein travels, the more the lags from the extra “tunnel detours” add up, and thus the more separation. But also the more diluted your sample gets. Speaking of which – it’s important to concentrate your protein to a small volume before injecting it onto the column – concentrating was less critical with affinity chromatography since it doesn’t matter if a protein gets a head start – but here it does matter. The more you concentrate it, the sharper your peaks will be (but don’t concentrate it too far or your protein can “crash” (precipitate out). 

note: for the injection part, I inject into a sample loop which serves as a sort of parking spot for the protein until I tell the software to direct it onto the column. These loops come in different sizes (100uL, 500uL, 1mL, 2mL, 5mL, 10mL…) and super technical note but for preparative SEC, you want to use a loop size that’s about 4X larger than your injected volume (e.g. I usually concentrate to ~0.5mL use a 2mL loop). This will give you the best yield, making sure none of it gets lost (though you do dilute your sample a bit). For analytical, on the other hand, since you want to make sure you’re injecting the exact same amount of your samples, you want to use a loop that has a smaller volume than what you inject. This way the precise volume of the loop is filled with the protein at the concentration it was at before injecting. Sorry that got technical, but wanted to share in case it’s of use to someone. 

But back from that super-geek-out to our geek-out… In addition to size differences you can see, there are ones you cannot see – different column resins have different pore sizes (tunnel height/widths). Each bead has lots of pores & you buy based on “average” pore size, but each resin is HETEROPOROUS (has a range of pore sizes)). You want beads with the pore size best suited for separating proteins around the size of your desired protein. That is, you want your protein to be able to get into some but not all of the pores and you want things you don’t want to either go straight through or get caught up traveling the longest way possible so they don’t come out “elute” with your protein. In science jargon, you want sufficiently different “permeation rates”

By protein “size” we’re talking effective molecular radius (Mr). if your protein were spherical, this’d be its radius. Technically, what we actually care about is the hydrodynamic volume. Basically, proteins aren’t stationary, so you have to take into account a bit of flailing. No proteins are purely spherical (& some are *really* un-spherical), so Mr is just an estimate. But it’s way better than nothing! And you can use it when choosing resin to use & knowing around how many mL into the run it should elute.

Different resins have different “exclusion limits” – basically the biggest tunnel height – proteins that can’t fit in those tunnels get to go around instead of through all the beads, so they have the shortest possible trip and come off in the “void” along w/the stuff that was in the beads before you added your protein. 

Speaking of that “stuff” that was already there – really small things like salts have to take the super long way, so your protein elutes in whatever buffer (salt-water with pH-stabilizers) was already in the beads (which is why you have to equilibrate your column well by running buffer you want your protein to end up in through the column first) – this also makes SEC good for buffer exchange (e.g. as a faster alternative to dialysis). 

But what is it that our proteins are going through anyway? What are all these tunnels made up of? Often, agarose (strengthened a little). Wait a SEC – you use *agarose* to separate *proteins*? I thought agarose was for DNA? Turns out proteins also get to agarose-play, or actually *not play* because the great thing about agarose is its boringness. 

Agarose is a type of sugar, and it has always been associated in my mind with “DNA” (we often make a “slab” of agarose gel (a meshy matrix filled w/water) & use agarose gel electrophoresis to separate DNA fragments by size (longer DNA strands take longer to push through the gel’s pores) to look at them). 

BUT what I didn’t realize until a couple years ago was that I actually use agarose all the time to separate PROTEINS by size too! 

There are also different resin types w/various modifications to agarose to better suit various needs. When you make an agarose gel for electrophoresis, the gel forms because of “physical interactions” – agarose is a polysaccharide (long chain of sugar rings) & its strands stick together through individually weak, but collectively strong partial-charge based attractions called hydrogen bonds (H-bonds) 

Because those are just attractions, no electrons are being shared (sharing electrons is what makes covalent bonds, like those linking the sugar rings within a chain, so strong).  Agarose’s “gelation” is reversible – you can “reheat” an agarose gel & it will give you a liquid solution again because you break the H-bonds (but NOT the covalent bonds so your chains remain chains) – which is convenient if you forgot you had a flask heating in the microwave… 

Sometimes you want to *weaken* these H-bonds to more easily melt the gel around DNA to get the DNA out of the gel without hurting the DNA. This is done with various modifications to give you low melting point agarose:

But with SEC, you want to *strengthen* interchain interactions  so they can withstand pressure of liquid flowing through them at a faster rate without crushing. To do this you need to go beyond mere “attractions” & share some electrons! –  add in a few covalent bonds through chemical cross-linking.

The sugar subunits making up agarose have lots of hydroxyl (-OH) “legs” sticking out that help them stick to each other (& water) through H-bonds. Under the right circumstances these legs can also form covalent bonds (in fact, 2 of them team up through a covalent bond to link the individual sugar monomers together to get the agarose polymer). 

There are still plenty of legs left & with a little o-chem-ing you can modify them as you please including by getting them to form covalent bonds BETWEEN strands (CROSS-LINKS) by a leg from each strand grabbing on to different parts of a CROSS-LINKER like 2,3-dibromopropanol.

You can get agarose resins with different amounts of cross-linking. Cytiva (which used to be GE), I don’t work for or anything, but they make the AKTA and their columns. They sell “normal” cross-linked agarose under the brand name Sepharose™️ & a *super*-cross-linked version under name Superose™️. They also have “Superdex” columns which, in addition to cross-linked agarose, have covalently linked dextran chains. DEXTRAN (what’s cross-linked in Sephadex columns) is a different polysaccharide – it’s branched & has better separating properties, but it’s not as sturdy. To get the best of both worlds, you can stuff it into an agarose matrix that provides the strength needed to withstand the pressures generated when liquid flows through it at higher rates.

Today I used a Superdex column. It was the last step of a protein purification I started yesterday (and is on schedule for the new prep I started today). So now I want to tell you a bit more about how it fits into that protein purification story I’m walking you through.

So, I’m purifying a protein I want to study. I expressed the protein recombinantly, meaning that I stuck the genetic instructions for making that protein into “expression cells” to make it for me. Often bacterial cells are used for expression, but my protein (like many more complex proteins) prefers to be made in cells with more sophisticated processing machinery – so I expressed it in insect cells. 

A great thing about recombinant expression is that since you control the gene, you can control the protein (since it’s the order of DNA letters (and their RNA copy) that will control the order of protein letters (amino acids)). By adding a few extra DNA letters before my protein recipe, I can have the cells add on a few extra amino acids that can serve as an “affinity tag” to help me purify my protein out from all the other proteins using resin that binds that tag specifically, allowing me to get my protein to stick to a column while I was all the other stuff off, and then I can push my protein off using a tag mimic. And even cooler, I put in a sequence in between the tag end and my protein start that an endoprotease (protein-cutter) recognizes so I can cut off the tag once I’m done with it. 

So yesterday I did all that. Break cells containing protein open (lyse them) -> spin down really fast to pellet out all the membrane pieces and other insoluble gunk -> add the soluble stuff (supernatant) to affinity resin -> let bind -> wash -> compete off -> cleave tag

That left me with a solution of tag + untagged protein + endoprotease – as well as some lingering proteins that just liked the resin (nonspecific binders). And I want a solution of untagged protein without those others. So I used ion exchange chromatography to separate them based on charge. 

At this point my protein was almost pure, but not quite pure enough – I’m planning to use it for sensitive experiments so I want to make sure that I don’t have any “artifacts” skewing my results. So today I turned to size exclusion chromatography as a polishing step. 

After running SEC, I used the chromatograph to pick the fractions containing my protein – I combined them, concentrated them a bit more added a bit of glycerol as cryoprotectant to prevent ice crystals from forming, aliquoted them into single-use portions, and flash froze them.

Then it was back to the other prep!

For more practical protein-purification posts (and background/theory), check out the new page on my blog where I’ve collected some of my protein purification posts.

more on topics mentioned (& others) #365DaysOfScience All (with topics listed) 👉

helpful resources:

Cytiva, Fundamentals of size exclusion chromatography – link to handbook at bottom:

Ó’Fágáin C, Cummins PM, O’Connor BF. Gel-Filtration Chromatography. Methods Mol Biol. 2017;1485:15-25.

Leave a Reply

Your email address will not be published.