I screen, you screen, we (probably not all) screen precipitants and pH in hopes protein crystals will be seen! Protein crystallography aims to get a “look” at a protein’s structure by getting it to form crystals then beaming x-rays at it. Because the protein molecules are lined up nicely, they scatter the rays the same way and these scattered rays add up to give a pattern of spots called diffraction pattern we work backwards to figure out the proteins atomic architecture. But to do this we need to get the proteins to form crystals, which requires them to give up some protein-water interactions in exchange for some protein-protein ones, coming out of solution in an organized manner.
We (hopefully) can convince protein molecules to line up in these crystals using crystallization cocktails. These are combos of buffers (pH-stabilizers) and precipitants (solubility-lowerers including salts, organic solvents (like isopropanol, ethanol…), & polymers (like PEG). Proteins are picky & it can be hard to find the right cocktail. So we usually start by screening – testing out a wide range of conditions to find initial “hits” and then we optimize around those hits. I’m planning to release 2 videos to accompany this post – first the theory and then what it actually looks like in the lab and some practical details.
note: text mis-mashed and adapted from some of the bumbling biochemist’s first posts (pre-blog days) so they’re not great and the format’s kinda a bit all over the place, but hopefully better than nothing and I don’t have time for polishing because I really need to get some sleep… hopefully in combination with the video it will make sense. If not, sincere apologies!
You can think of protein crystallography as “game” taking place on crystallization diagram “field” with free protein concentration on the y-axis & precipitant on the x-axis & we have to scout out best field for each protein. When a protein’s under its solubility limit (in undersaturization zone of the crystallization diagram) there’s enough water for each protein molecule to get a full coat (hydration layer or solvation shell) & if you add more protein, the “newcomers” will get their own shell too. But when water runs out (you reach the solubility limit) newcomers will be turned away (or swapped for existing dissolved molecules) because a saturate solution is an equilibrium between dissolved & undissolved molecules – there’s a constant ratio between the two, so new molecules can dissolve only if others “undissolve.” To get crystals we need to push concentration of protein past solubility limit to a non-equilibrium state, where there’s a thermodynamic drive to leave solution & regain equilibrium – i.e. we want to reach a supersaturated state.
How? The solubility limit is specific to each set of conditions & each protein, so if we change conditions we can change the limit (why zone boundary’s curved) – on condition’s undersaturated is another’s supersaturated! If we lower solubility AFTER we’ve reached a high concentration of protein, we can create a supersaturated solution.
It’s like making rock candy, where you dissolve as much sugar as you can into boiling water to get saturated solution, then cool it. The solubility limit’s lower at the lower temp, so you get supersaturated solution & sugar forms crystals. Proteins are sensitive to temp & it’s not very practical, so instead we usually add chemicals called precipitants to lower solubility by lowering the amount of *available* water, making protein-protein interactions more attractive.
There are 3 main types we’ll look at:
- salts – compete for water; interact w/protein
- organic solvents – lower solvent’s “shielding” ability
- synthetic polymers – a bit of both of both of the above & increase crowding
Additionally, we can alter pH (at different pH’s proteins have different charges in different places, which changes the interactions they can have) and the temperature (which changes solubility and the speed at which the protein changes zones in the diagram).
We use screening to find initial “hits” of chemical combos (crystallization cocktails) with the capacity for growing crystals – as I alluded to above, this entails taking the protein/cocktail mixtures out of their happy places (their equilibria). Basically, if you have two different conditions (this could be a solid phase & a liquid phase of the same thing or 2 “populations” of different things) & you combine them, they’ll adjust until they find an equilibrium. At this point, the relative amounts in the two conditions aren’t necessarily the same (e.g. you could have more molecules in solid phase than liquid phase) but they’re “happy” w/ratio. SATURATION is an equilibrium state between dissolved & undissolved molecules. For a given number of protein molecules in a given set of conditions, a certain portion of them will be dissolved & a certain portion of them will be undissolved (hopefully as crystals & not just clumps).
The molecules may be happy but the crystallographer’s not. Instead, we see equilibrium more as their “lazy place” – chemical systems are driven by a desire to reach equilibrium and when they’re already at that happy place they lack motivation. So we motivate them to grow by taking them away from equilibrium into a supersaturate state where there’s more protein than the water can happily hold. In this state, since the water’s overfull, some protein molecules come out of solution (nucleate) to form a seed. These seeds are “hungry” – they want to take more molecules out of solution so the system returns to equilibrium. As they feed, crystals grow, but their “food source” (free protein “surplus” dwindles), so they get closer to equilibrium, their motivation dwindles & finally they reach equilibrium & stop growing.
We use crystallization cocktails with precipitants to make it easier to find the protein “food” the crystal is looking for. We can use precipitants like salts that take away some of water “shielding” protein molecules from one another and steal that water away from the protein molecules. And/or precipitants like PEGs which take up a bunch of space by flopping around, leading to a volume exclusion effect pushing the protein together. And/or we can make the food “tastier” by changing pH so that there are more potential protein-protein interactions http://bit.ly/2PF3TfH
Going back to our crystallization field diagram, our goal is to take our system from a field’s undersaturated zone, where crystals can’t form, to the supersaturated zone where they can. To do this, we can control pH, temperature & precipitants (chemicals that lower solubility) to move “zone lines” &/or get protein molecules (players) to move between them. At the end of the post I go into more details about how these various precipitants work, but the key thing to keep in mind is that each protein likes different combos and you basically just have to find it out empirically (through trial and error).
Unfortunately for us, there are an Infinite number of potential playing fields in the chemical universe. We can narrow the search by leaving out conditions that obviously won’t work for any protein (like instead of searching entire universe for aliens, leave out Venus…) We can also use what we know about *our* protein to guide us. Just like if someone can’t stand snow it’s waste of time to suggest they move to Arctic, if a protein’s unstable at a low pH, don’t test low pH combos. But there are still sooooo many options! So we need to screen somewhat strategically.
We can buy screens with premixed cocktails which sample playing fields (different combos of salts, buffers, organic solvents, polymers, etc). These screens sometimes come in tubes (more customizable) or pre-dispensed 96-well blocks (easier).
Some use a grid screen approach, where they change one condition (e.g. pH) across columns & another (e.g. salt concentration) down rows so you can test out each “pair.” In our analogy of neighborhood-searching for fields, this is the equivalent of going block by block checking for fields.
Other screens sample a wider range of conditions (less thoroughly) using a sparse matrix screen. These contain cocktails that’ve worked for *other* proteins, analogous to only going to addresses where you know there’s a field (but you don’t know if that field would work for *your* protein)(maybe you have footballs & it’s a baseball field). With these screens, since they’re based on previously discovered fields, you can’t discover any “new fields” with them, so some screens combine aspects of both grid & sparse screens.
Screening helps us find potential fields, but just bc crystals can form there doesn’t mean conditions are great. Kinda like how you could play football in field of weeds, but you will have a “better” game if you clean up field, spray on right lines, etc. The crystallization equivalent to this field renovation is optimization.
The idea with optimization is that we take a “hit” & do a finer screen around it (common to go +/- 1 pH unit & +/- 20% of precipitant concentration). For example, if we got a hit with HEPES (a buffer) pH 7 & 10% PEG4000 (a precipitant) we might try a grid screen from pH 6 – pH 8 & PEG4000 from 8-12%. Then we would take a hit from that second screen & do it again, sampling finer & finer (a method called successive grid screening).
With such screening, you can keep honing in on the “best” but there’s not necessarily one “best”… sometimes crystals of same protein can form in very different conditions & they often line up differently in the crystals (have different space groups, crystal packing, etc.). Some may be better diffracters, giving us better data, so we have to be careful not to get stuck in a rut of only trying what we know works.
How do we do this optimization in practice? Those commercial screens are great time-savers for finding initial hits, but then we have to customize your finer screens in the optimization step. But we don’t have to do it all manually, thank goodness!
We use a software called RockMaker to design our screens – we give it a recipe like “vary pH from 6 to 7.5 down the rows (i.e. 6, 6.5, 7, 7.5) & vary PEG4000 concentration from 10-16% across the columns (i.e. 10%, 11%, 12%, 13%, 14%, 15%, 16%). The RockMaker then calculates how much of each ingredient to add to each well & generates a barcode – we scan this barcode into a Formulatrix liquid handling machine & it follows the barcoded instructions to add right amounts of the different liquids into the wells of blocks or plates. Then it’s crystallization trial setup time!
Another practical note: I normally carry out initial screens in a 96-well format to maximize the “chemical space” I can sample, then later in the optimization (where there are fewer conditions I need to test) switch to a 24-well format where I can grow bigger crystals that are easier to harvest (fish out).
There’s never a true “end” to optimization, but we’re often limited by amount of protein (& time, stamina, etc.). Speaking of amount of protein, all that testing can take a ton of protein (at high concentrations so you start near saturation) – to try to get the most bang for the protein buck, we have liquid dispensing robots to help us do this with tiny tiny volumes so that we can test hundreds of combinations without needing a ton of protein (but you still need quite a bit because of the high concentration required).
How about that time/stamina factor? Don’t forget that the more trays you set up the more time you’ll have to spend checking the trays for crystals… Thankfully we have a microscope robot that takes pictures of our crystal trays for us over time so that we can see if crystals are forming in any of the wells. But the robot microscope sometimes mis-focuses so you end up having to look through all the pics one by one as well. And then when you’re unsure or something you can always look under the benchtop microscope – but you don’t want to fiddle around with the plates too much because the vibrations and stuff can cause problems.
Before I go, a very important point – we’ve talked about these cocktails as if they’re all that matters – but they’re definitely not! Instead, how we introduce these cocktails to the protein matters! If you have a ton of protein molecules in solution & you dump all the food on them all at once, lots of them will take some BUT they can’t take much because they quickly deplete the resource. That can give you tons of really tiny crystals that aren’t useful… BUT if you give out the food more slowly you get fewer seeds that are “greedier” (because it’s easier to join on to an existing seed than start one) so you can get nice big crystals.
How? The most obvious equilibrium in crystallography is that one we looked at before – between dissolved protein molecules & protein molecules in the solid form of a crystal. BUT there’s another important equilibrium taking place – that between the protein solution & the protein-less crystallization cocktail containing the precipitants. This second equilibrium “feeds” the dissolved/crystal equilibrium and it’s “controlled” by how you set up the experiment. Techniques fall into 2 main types:
- batch methods – these are conceptually the simplest – just mix your protein directly with the cocktail. In this case equilibrium between the protein solution & the cocktail happens really quickly, so the tricky thing with this technique is that you need to “know” the exact endpoint you want that’s ideal to get few, big crystals
- diffusion methods – in these methods you don’t need to “know” the exact endpoint – you start with the protein separated from the crystallization cocktail and gradually introduce them – as the solutions equilibrate they’ll “pass through” the “right” conditions and & when it does nucleation will occur – that seed will start draining growing, pulling more protein molecules out of solution and bringing the solution closer to equilibrium – once you hit it, the crystal stops growing
The diffusion method that I’ve used the most is “vapor diffusion,” mostly “hanging drop” crystallization. Basically you stick a drop of liquid containing your protein on a glass slide and then you flip the slide over and use it as the “roof” for a well of protein-less liquid (reservoir). Since this reservoir liquid is more concentrated than the drop liquid (because the reservoir hasn’t been diluted with your protein), water evaporates from the drop to help “dilute out” the reservoir (that’s not really its goal – it’s trying to escape the well altogether but there’s a lid, so it gets pulled in by the reservoir). This leaves less water available to surround the protein molecules. So the protein molecules start binding to each other instead – hopefully in the coordinated fashion that leads to nice crystals. ⠀
Once our crystals have grown and stop growing (could be days to weeks to months depending on the crystal) we have to fish them out with little loops, freeze them with cryoprotectants, and store them in liquid nitrogen dewars (basically really insulated giant thermoses) to keep them super cold until we’re ready to collect diffraction data from them. ⠀
more on the various precipitant types:
The most intuitive way to think about creating a supersaturated solution is physically removing water molecules, which is what evaporation does, which is why you get sea salt when water evaporates. But you can also reach a supersaturated solution without removing any water (and with diffusion crystallography both such dehydrations come into play). Precipitants “dehydrate” proteins without physically removing water.
One main class of precipitants is salts – when you add salts to a saturated protein solution, the salts compete for water (they need water coats to stay dissolved too) & the salts are stronger competitors, so they “use up” water. Therefore, it’s LIKE there’s less total water, but there’s REALLY the same amount you started with. But we can say that the salt has lowered water’s chemical activity (how much water it seems like there is). To help explain chemical activity in a quick intuitive sense, say a train station has 10 ticket machines – its “concentration” is 10 machines per station. BUT when you go to use one, expecting there to be 10, 8 are already in use, so it’s like there’s really only 2. There, we could say its ACTIVITY is 2 machines/station. The # of machines didn’t change, so *concentration* didn’t change BUT its “effective concentration” did & we call this the activity. Similarly, how much water there is doesn’t change when you add salt, but less water is available for the protein.
Another class of precipitants is organic solvents, which are beyond the scope of this post but you can read more on them here: http://bit.ly/2DDtyzC
bottom (jargon-y) line is when you add less-polar things (like these solvents) to water, they lower the solution’s overall dielectric, basically reducing the water’s ability to shield protein molecules from one another. One reason water can coat proteins and form tight water networks is it’s highly polar (has a directional charge imbalance) leaving positive and negative parts that attract. You can think of the organic solvent molecules interfering w/water’s “mega-dipole” (uneven charge distribution) so the water can’t arrange itself to optimally coat the protein. As a result, the protein’s less eager to take on a solvation coat and nearby protein molecules seem more attractive because they have charges they can be attracted to. So the proteins come out of solution & bind to each other (hopefully as crystals) A downside of these precipitants is that they’re volatile (evaporate easily) so you have to work quickly and add them manually (not in the pre-made screens).
The most popular precipitants are synthetic polymers like PEGs (Polyethylene glycols). Like salts, they compete w/proteins for water & like organic solvents, they lower the dielectric. But PEGs aren’t just copy-cats, they have an added bonus – volume exclusion! They act as macromolecular “crowding agents” – not only do they “steal” water, but they also “steal” space!. The excluded volume effect says 2 molecules can’t occupy same space at once. So if there’s a PEG molecule there, there can’t be a protein molecule there & pushy PEGs like to be “everywhere” so other molecules have to compensate by getting closer.
PEGs are long chains of ethylene oxide units & they act like inflatable tube men waving outside car dealerships. They wiggle around lots, taking up more space than they “deserve,” so heir excluded volume (the no-go zone for other molecules or “personal space”) is > than its “true” volume. PEG’s units are held together with flexible bonds, so it’s like there’s a 3D windmill at each joint. At any one time, a windmill’s blade can only be in one place, but it’s moving around so fast it’s like it’s taking up a big circle & other molecules can’t get close. To avoid “getting hit,” proteins huddle together (eery molecule has a “personal space” but when they bind together, they can “combine” parts of their personal space leading to a lower combined excluded volume to counteract PEG’s volume). If the proteins do this in an orderly fashion, we get crystals!
PEGs come in lots of lengths & we describe them by average weight. Each unit weighs ~44 Daltons (Da), so PEG 4000 would have ~91 units, etc. Different proteins like different weights & amounts, so you have to screen. Bacteria like PEGs too so we make it fresh from powder, then filter it which takes a long time because it’s super viscous (like syrup)
more on crystallization: https://bit.ly/gettingcrystals
much more on crystallography: http://bit.ly/xraycrystallography2