Today I finished a couple of protein purifications that I started yesterday, and I wanted to figure out what my yield was – how much pure protein did I end up with? And, if I were to look inside the liquid in one of the tubes with my protein, how much of the total liquid volume would be my protein. That is, what were my protein concentrations? Like with many things in science, when it comes to protein measuring methods (aka “protein assays”), there are many you could choose to do – Bradford, UV-Vis spectrometry, BCA, and Lowry, just to name a few.
Apologies in advance that I’m going to use some jargon and not completely explain some of the background (which you can find in other posts of mine) but I hope this post will be helpful practical advice for others who might be in the same boat of trying to figure out how best to measure all of their protein treasure!
There are multiple ways to measure protein concentration, each with pros and cons for various circumstances. Some methods go straight to the source – UV absorption measures how a protein itself absorbs light. Other methods involve a dye binding to protein and the dye either changing color (colorimetric assays like the Bradford assay) or letting off light (fluorometric assays like with a Qubit). A whole different strategy is taken by copper-chelation based methods like the BCA & Lowry assays, which take advantage of proteins’ ability to “grab” (chelate) & reduce copper in what’s called a biuret reaction – that makes a light blue complex – and this part’s the same for BCA & Lowry methods – but then each method goes a (different) step further to amplify the signal, which you can then convert to a concentration by comparing to a “standard curve” of the signal from known quantities of proteins.
The main 2 I use (and which you’re likely to encounter in a biochemistry class lab) are the Bradford assay and UV-Vis spectroscopy, but I will introduce you to others as well.
There is no one “best” way to measure protein concentration – it all depends on the situation at hand. Some of many things to consider:⠀
- sensitivity (if you don’t have much protein you certainly don’t want to use it all up trying to figure out just how little you have!)⠀
- compatibility with the other things your protein’s hanging out with. In addition to the pH-stabilizing buffering agent (Tris, MOPS, HEPES, etc.) that gives buffers their name, and some salts, the buffer your protein’s in may contain things like detergents (artificial soaps Brij & Triton) & reducing agents (like DTT & β-mercaptoethanol), which can skew the readings of various assays and different methods have different Achilles’ heels⠀
- cost (financial, physical, time-wise)
- protein composition – many methods get most of their signal from particular amino acids (protein letters) – if your protein has more or fewer of the particular protein letters (amino acids) than an “average” protein, it can trick you into thinking there’s more or less of your protein than their actually is
This last point – protein composition – brings us to a discussion of how you go from your raw signal to an actual concentration. You need a way to convert the signal you record to the amount of protein it corresponds to. UV absorption goes straight to the source – it measures how a protein itself absorbs light. This is in contrast to the other methods, where you’re looking at an indirect readout – a dye binds to protein and changes color – and then you measure this color-changing. ⠀
For UV-Vis, if we just have a single protein, we can calculate something called an extinction coefficient which tells us how much a protein is predicted to absorb based on its molecular composition (in particular how big it is & how rich it is in the amino acid (protein letter) Tryptophan (Trp, W) which is the main absorber). (If we have a mix of proteins, we can use an “average” extinction coefficient. Speaking of averages, it’s harder to predict what you’d expect from the other methods, so instead of trying to calculate a specific conversion factor for each and every protein, we empirically determine (find out through experiment) how much signal different amounts of a “protein standard” such as Bovine Serum Albumin (BSA) or Bovine Gamma Globulin (BGG) produce. This allows us to create “standard curves” we can compare our signal to. ⠀
But, these curves aren’t perfect. Because there really is no “standard” protein. All proteins are unique! And this uniqueness comes because proteins are made up amino acid letters that have a generic peptide backbone part and one of 20 unique side chains (aka “R groups”) that sticks off kinda like a charm on a charm bracelet. Different proteins have different numbers and combos of letters and none of these methods only measures the generic part (BCA/Lowry comes closest, but even it can get skewed (by cysteine/cystine, tyrosine and tryptophans in particular)). Bradford gets skewed by arginine, lysine, and histidine. And UV absorbance by tryptophan (and tyrosine & phenylalanine to a lesser extent).
So depending on which unique parts the method measures and how many of those your protein has, you’ll get a different per-protein signal. If your protein has more of the skewer than the standard you’re comparing to, you’ll overestimate your protein’s true concentration, and if your protein has fewer a skewer, you’ll underestimate your protein’s true concentration. So, for example, lots of arginine? Bradford based on an “average protein” would overestimate. Lots of cysteine? BCA would overestimate the true concentration.
Speaking of “true concentration…”
When it comes to protein concentration, there are a couple of ways to report it. One is mg/mL. And if you have a mix of proteins, this is as far as you can go – but if you have a single protein, and you know its sequence, you can actually figure out how many molecules of the protein you have based on how much each molecule weighs (which can be calculate based on what atoms its made up of and how much each of its atoms weighs – don’t worry you don’t have to do this by hand – you can use free internet resources like UniProt or Expasy ProtParam).⠀
A single protein molecule, even a “big one” won’t weigh much, so instead we speak of how much a lot of them weigh – a lot a lot of them. Like a whole mol’s worth. A mol refers to Avagadro’s number (6.02 x 10²³) of something, and the molecular weight tells us how many grams one mol weighs. So, for example, if I look up BSA I see that it has a molecular weight of about 67,000 Daltons (Da), which means that a mol of it weights about 67,000 g (67 kg). ⠀
Definitely don’t have that much, but that doesn’t matter because that conversion ratio is the same whether you have 67 kilograms-worth of protein or a single molecule. So now we can convert from mg/mL to mol/mL – but what we really want is mol/L, which is called molarity. So, for instance, 1mg/mL of BSA would have a molarity of (puts on dimensional analyzer hat…)⠀⠀
(1 mg/mL) x (1 mol/67,000 g) x (1 g/1000 mg) x (1000 mL/1L) = ~0.000015 mol/L = ~0.000015 M, or 15 μM⠀
So how do we get mg/ml? Let’s start with the Bradford assay method. You might know Coomassie brilliant blue (CBB) better for its role in dyeing SDS-PAGE gels to visualize protein bands in a gel you’ve used to separate proteins by size. I did a whole Bri*fing on it if you want to refresh: http://bit.ly/2k28Lxy ⠀
That same protein-binding power that makes CBB good for staining proteins in a gel also makes it good for staining proteins in a well – or a cuvette (a rectangular “tube” with clear walls that you can shine light through) – and this is the basis of the Bradford protein assay. You take the “G-250” form of CBB, mix it with your protein, and then measure absorbance (how much shone light is “stolen” before it gets through your cuvette) at a shone wavelength of 595nm. Why there?⠀
CBB G-250 exists in 3 forms because it has 2 protonatable groups (parts that can give or take a proton (H⁺): both, 1, or neither might be protonated to give you a doubly-protonated and positively-charged cationic form (which looks red, maximal absorbance (Amax) at 470nm), a neutral form (which looks green, Amax at 650nm); and a negatively-charged anionic form that looks blue (Amax at 610nm). The red form predominates under acidic conditions, since there are lots of protons available. But when it binds to proteins, CBB gets stabilized in the blue form. So you can measure the blue-ness to figure out how much protein’s there. To quantify blueness to you typically measure absorbance at 595nM because the 2 forms’ absorbance peaks overlap and this is where there’s the biggest difference between them – you can also look at the ratio between those 2 wavelengths.⠀
But the blueness-to-protein conversion factor depends on the protein composition. Because Bradford doesn’t just like any ole amino acid – it has favorites – basically, CBB likes basic (i.e. normally positively-charged) amino acids – Lys, His, & Arg contribute most to binding, but van-der-Waals & hydrophobic interactions also contribute. So the standard you’re using should bind similarly to your protein or the average protein if you’re looking at a mix. ⠀
A couple of commonly used “standards” are Bovine Serum Albumin (BSA) & Bovine Gamma Globulin (BGG). They’re so commonly used that you can buy standardized “ampules” of them – little tubes filled with it at a precise concentration. BSA is a common standard for large reason because it’s cheap to get lots of it pure. But it gives a much higher Bradford response than most proteins, so it can lead you to underestimate how much of your protein is there. BGG is actually a mix of several immunoglobulins (a type of antibody), but it is closer to “average” in terms of Bradford binding. ⠀
note: there are also fluorescent-dye-based methods which work similarly but the dye gives off light when it binds rather than changing color (so you’re measuring light not light-stealing)⠀⠀
I made standard curves of BSA & BGG to compare them to one another and to my protein. ⠀
a standard curve is where you do a serial dilution of the standard of known concentration (e.g. 100 uL of 1 mg/mL + 100 uL buffer -> mix, mix, mix -> 100 uL of that mix to 100 uL of buffer -> mix, mix, mix…) and then you measure the signal from each of those values using whatever assay you’re using and then you plot that signal against concentration. Then you find the equation for the linear part of the curve. And then plug the value you measure for your unknown concentration sample into that equation to see what concentration that measured value corresponds to. It needs to be a value in the linear range! more on these here: http://bit.ly/2Nu9dAS ⠀
Bradford’s cheap and quick. But it has a narrow linear range and isn’t as protein-sensitive as the next methods – but thankfully it also isn’t as reducing agent sensitive – DTT & BME don’t interfere with it. Those reducing agents DO interfere with BCA & Lowry. Big time. And so do chelators like EDTA. Because BCA & Lowry are based on your protein chelating and reducing copper – so you want to make sure it’s only your protein doing that!
BCA and Lowry assays are based on the biuret reaction, which refers to a copper iron with a positive 2 charge (Cu²⁺) (cupric ion) getting grabbed (chelated) by molecules and gaining an electron (getting reduced), dropping copper’s charge down to give you Cu⁺ (a cuprous ion). The “biuret” name comes from this happening with an organic compound called biuret (NH₂- CO-NH-CO-NH₂) which is produced when excess urea and heat meet copper. But in proteins, it happens *mainly* from the backbone – it typically forms a tentradentate (4 binding points) complex with 4 peptide bonds, but it can react as long as you have at least 3 letters (any letter) in a row. As long as you do, you can use this method for peptides, not just proteins. And that’s a plus over the Bradford.
Another plus – because it’s *mainly* the backbone reacting, and that backbone’s the same no matter the letter, BCA & Lowry have less protein-to-protein variability than Bradford. BUT that “mainly” is in asterisks for a reason – in addition to the backbone, there are a few particular amino acids that can react with it above and beyond the backbone – if you have more cysteine/cystine, tyrosine and tryptophans, you’ll get a higher signal.
Speaking of getting a stronger signal…. the original biuret stops after the first reaction, when the copper’s reduced and protein-bound – and that has a light blue color. BCA & Lowry take things one reaction further…
BCA does this with bicinchoninic acid. The copper that gets chelated by proteins is only weakly bound, so you can add something that’s a stronger binder to grab it out – and you might want to do this to get a stronger signal. 2 BCA molecules latch onto the Cu⁺ and form a purply complex.
Lowry takes a different approach to signal strengthening – it uses something called Folin-phenol reagent (phosphomolybdic-phosphotungstic acid). This reagent gets reduced to make a bluer blue compound – and it does this without stealing the copper from the protein. So it’s an “endpoint” assay unlike the BCA, where copper can bind to the protein sites that get opened up when the BCA steals it, generating further signal. So, with BCA, timing really matters – you have to make sure you measure all your samples (including standards) after the same amount of time.
In addition to sensitivity, these methods have the “pro” that they are compatible with detergents, which can interfere with Bradford.
We’ve been using light-based measurements in all the above experiments, but I never really explained what light is – and that’s important to know for this next one, so here’s a quick summary. Light (electromagnetic radiation) (EMR) is little packets of energy traveling in waves. Different colors have different wavelengths of light with different energies (this is also true for “invisible” colors – wavelengths outside of the visible spectrum – like radio waves, which are lower frequency and ultraviolet (UV) waves which are higher frequency. Different molecules absorb different wavelengths of light to different extents, and this can be quantified by a number called the extinction coefficient (ε), which tells you how well a molecule absorbs light of a particular wavelength. More on *why* here: http://bit.ly/2CfaXbJ You can then use this equation called Beer’s law to figure out how much of a protein or DNA or anything that absorbs based on it⠀
But TLDR, it has to do with how much energy the outer electrons of the molecule have and how much more energy they need to get to the next level – electrons can be thought of as living in “houses” called molecular orbitals – it’s not that they “always” live in one place – they’re constantly zipping around, but these orbitals are where you have the greatest chance of finding them. Orbitals farther from the nuclei of the molecules (where the positive protons and neutral neutrons are held) require electrons to have higher energy to live there. If a molecule gets hit by a photon of the optimal energy, the photon can get absorbed and its energy used to promote a lower energy electron to a higher energy, further from the nucleus, orbital. ⠀
Electron delocalization through resonance involves the merging of some neighboring molecules’s orbits into that shared donut. And this lowers the cost to move to promote an electron – so the rings in aromatic amino acids can absorb UV light – Tryptophan (Trp, W), Phenylalanine (Phe, F), and Tyrosine (Tyr, Y) can all contribute, but Trp is the main contributor, so the UV280 absorbance per protein is gonna depend on how many Trps that protein has. “Abnormal” Trp numbers can trip you up! “Too many” and you can overestimate and “too few” and you can underestimate protein concentration if you go by the “average,” but if you know the protein’s sequence you can calculate the extinction coefficient for your exact protein⠀
Extinction coefficients can be given in terms of molarity if you know the sequence of the protein. but when you don’t (and even sometimes when you do) they’re given in mg/mL terms. They tell you the absorbance value (A) that corresponds to 10 mg/mL (1%) or 1 mg/mL (0.1%). Those percentages come from the weight/volume percentage convention that a 1% solution corresponds to 1 g/100 mL – more on why here: http://bit.ly/37LsTrq⠀
You can calculate the estimated extinction coefficient using free online software tools like Expasy ProtParam. I say estimated because context matters – the local environment around the absorbing part can influence how eager it is to absorb a photon⠀
When I do this for BSA I see this:⠀
Extinction coefficients are in units of M-1 cm-1, at 280 nm measured in water.⠀
Ext. coefficient 42925⠀
Abs 0.1% (=1 g/l) 0.638, assuming all pairs of Cys residues form cystines⠀
Ext. coefficient 40800⠀
Abs 0.1% (=1 g/l) 0.607, assuming all Cys residues are reduced⠀
First it tells me the values under oxidizing conditions and below that it tells me the values under reducing conditions (the intracellular environment is reducing and we usually add reducing agents like DTT or β-mercaptoenthanol) to protein solutions to keep them happy outside the cell). It tells me this because cysteine crosslinks can also absorb, where applicable.⠀⠀
Then I can plug this into Beer’s law (or have the computer do it for me) if I measure the absorbance. ⠀
You measure this absorbance using something called a spectrophotometer. Basically it shines light through a solution and measures to what extent different wavelengths make it through (are transmitted) versus don’t make it through (are absorbed). We looked at how this can be converted into concentration of solute (dissolved molecules) using Beer’s Law http://bit.ly/2N4nzXE⠀
The equation is: A = εcl⠀
A = absorbance⠀
ε = extinction coefficient (aka molar absorptivity coefficient) – specific for particular molecule & particular wavelength; units of L mol-1cm-1⠀
c = concentration (in mol/L) – this is molarity – a mole is just a chemist’s “baker’s dozen” – it’s Avogadro’s number (6.022 x 10^23) of something – solute molecules or donuts, it’s just a number http://bit.ly/2r4RnrX⠀⠀
l = path length (in cm)⠀⠀
For Beer’s Law, you only need the absorbance at a single wavelength – but there’s much more to learn if you look at the whole spectrum (or at least a couple key values). Because molecules have overlapping spectra (e.g. both DNA & RNA absorb light of 260nm wavelength) you look to ratios. ⠀
As I discussed, proteins absorb most strongly at 280, and this is where we typically calculate from. Proteins also absorb at 230nm and that absorbance is from the generic backbone part – corresponds to absorbance by the peptide bonds linking the letters. These peptide bonds also have resonance, but not as much as rings do, and they absorb ~190-230nm.⠀
Because DNA absorbs so strongly at UV260, where protein doesn’t, it’s relatively easy to see if you have DNA in your protein prep, but it’s harder to tell if you have protein in your DNA prep – 260 will dominate the 260/280 ratio⠀
Moral of the story: there’s no “one right way” to count your proteins & it’s important to carefully choose the one you use!⠀
For more practical protein-purification posts (and background/theory), check out this page on my blog where I’ve collected some of my protein purification posts. http://bit.ly/proteinpurificationtech ⠀
more on topics mentioned (& others) #365DaysOfScience All (with topics listed) 👉 http://bit.ly/2OllAB0