Fidelity of NEB Restriction Enzymes
Though I haven’t done directional cloning in many years (I <3 Gibson), a coworker –who is helping me by building a nucleotide variant library, where fidelity is particularly important– was running into extra bands showing up in his digestions. I suspected star activity, so I took a pretty deep dive into learning about the fidelity of NEB restriction enzymes. I was having a hard time finding much information on the web, until I ran into this paper published by NEB in 2008 in Nucleic Acid Research. The authors calculate the fidelity index (FI) as the number of units of enzyme where you start seeing a star activity band, divided by the lowest number of units you need to see full cutting (at expected sites); thus, the larger the number, the better. I found Table 1 to be particularly informative (though a little hard to quickly interpret), so I created this more visual version for reference; colors reflect fidelity (yellow = bad, dark blue = good), and the area of the bubble reflects % of optimal activity. The last column is the classification the authors give each enzyme:
I thought this paper was great, and far more informative than what I could find on the NEB website. The data really shows just how variable the restriction enzymes are; some are absolutely hideous (ScaI, XmaI [more like Xm-ugh-1, amirite?], AgeI, SalI, HpaI, BamHI, KpnI), and some are naturally exquisite (HaeIII, MspI, XhoI, XbaI). What also surprised me was how underwhelming some of the (what I consider) classic restriction enzymes are; EcoRI can be quite terrible, depending on the buffer, and BamHI and KpnI are pretty awful. I find it quite a shame this data isn’t better incorporated into the NEB website; I can certainly imagine a case where, without this knowledge, someone in the past may have run into a ton of trouble from star activity coming from one of the above-mentioned enzymes, particularly in a suboptimal buffer (such as during a double digest).
One possible drawback is that the data is somewhat outdated (for example, it uses the older buffer system; though please note, the new buffers are actually quite similar to their previous versions, except the new buffers don’t have DTT, but they now have BSA). I compared the information in the paper with what I could find online, which you can see in this chart here (information from the website indicated by the suffix “-NEB”). For the most part, the two sets of information match pretty well, although there are some differences, which I can’t tell are from updated purification protocols, differences from the new buffer systems, or just poorly transcribed data (more on this in a second).
The best information on enzyme fidelity (or rather, absence of star activity) came from this handy chart NEB has published on their website. Unfortunately, I noticed some inconsistent information on the website, which I find to be rather sloppy. Here’s a short list of problems I noticed:
- BmtI “HF Factor” numbers don’t match the units listed, 2. The Math on BsrGI also doesn’t add up, 3. Assuming it’s not due to buffer differences, SalI-HF is listed as having a fidelity index of >32k on the above site, yet only >2k on the actual FAQ page for the enzyme.
In anycase, here’s the chart for the normal (website listed information) vs high fidelity data, if you’re interested:
In general, the high fidelity versions do seem to be moderate to vast improvements, though in a handful of cases, the HF versions are as specific (NsiI and EagI) or sometimes slightly less specific (MluI; at least on the provided values) than the original versions, with the only advantage afforded being that they are now active in Cutsmart buffer (which I’ll admit, is kind of convenient). With all the data publicly available now in hand, I’d like to point out that there are still many restriction enzymes for which I can’t find ANY information on fidelity (ie. AflII). Considering how variable RE fidelity appears to be, I find this rather disconcerting.
In the end, I found this exercise to be useful and informative (and rather surprising), and I think I have a better sense of what enzymes I’ll now try to favor, as well as others which I’ll now try to avoid. It’s quite clear now that not all restriction enzymes were created equal (and nowhere near absolute), and certain ones provide much larger windows of success than others. I hope you find this information useful in your own molecular biology experiments as well!