So we went ahead with the kit. The positive controls worked brilliantly, so did the competent cells that came with the kit. However every experiment we did failed; not even a single colony formed.
The next two months were spent in frustration and confusion. We tried many suggestions from various websites as well as colleagues, none of which worked. The situation seemed especially bleak at one point when the kit ran out and we had to wait for weeks for more material to arrive while doing nothing since we don't have the right constructs to work with.
Jumping ahead to the present day, we were able to achieve the goal by doing it the hard way. Looking back at our past attempts, a lot of the "helpful hints" we got from the internet were wrong; senior lab members at my university, while more experienced, did not necessarily understand the mechanism of mutagenesis either so their advice were of little help (one technician simply suggested blasting through with large doses of every ingredient, seems like he got lucky with his way on every occasion). In addition there were too many other issues with our methodology and record-keeping that lead us into many dead ends; our sterile techniques were also less than perfect that we wasted much time and resource fooled by several cases of false positive results due to contamination, etc etc.
So what went wrong? Let's begin with a short summary of various historical and contemporary approaches to site-directed mutagenesis.
The first site-directed mutagenesis methoddescribed by Hutchison et al was published in 1978. In essence the method involves the synthesis of a short oligonucleotide primer containing the mutated sequence, which is hybridised with the wild type ssDNA and extended with DNA polymerase before transformation into E coli.
The protocol did not eliminate the template and hence less than 20% of progeny will retain the mutation, requiring tedious selection, a major pain back in the 1970s with DNA sequencing in its infancy and Kary Mullis a few trips away from inventing PCR. It is not surprising that some better funded and less patient labs resorted to have the mutant gene completely synthesised at great expense or in a less extreme version known as cassette mutagenesis, partially synthesised with suitable cloning sites and inserted into the vector by ligation.
Annotated diagram of found on the internet
Fortunately for the less privileged, many improved protocols soon emerged to increase mutant yield and reduce the burden of selection. One of the most significant advancement was developed by Kunkel et al., in which...Oh well I'd be lazy for once and quote wikipedia directly:
"The plasmid to be mutated is transformed into an E. coli strain deficient in two enzymes, UTPase and uracil deglycosidase. The UTPase deficiency prevents the breakdown of UTP, a nucleotide that normally replaces dTTP in RNA, resulting in an abundance of UTP; the uracil deglycosidase deficiency prevents the removal of UTP from newly-synthesized DNA. As the double-mutant E. coli replicates the transformed plasmid, its enzymatic machinery incorporates UTP, resulting in a distinguishable copy. This copy is extracted, and then incubated with the Klenow fragment, dNTPs, DNA ligase, and an oligonucleotide containing the desired mutation, which attaches by base pairing to the complementary wild type gene sequence. The ensuing reaction replicates the UTP-containing plasmid using the oligonucleotide as primer, thus incorporating the desired mutation. This forms a chimeric plasmid, with one strand unmutated and containing UTP, and the other strand mutated and containing dTTP. When this plasmid is transformed into an E. coli strain with normal UTPase and uracil deglycosidase, the UTP-containing strand is broken down, whereas the mutation-containing strand is replicated, forming a plasmid lacking UTP but containing the desired mutation on both strands."
This method is seldom used nowadays because there are easier and better ways out there, nevertheless the idea of template elimination took hold and contributed the basis of all mutagenesis strategies used today.
Probably through pure coincidence, dam methylation of E coli was also characterised in the 70s. The neat system provides a simple way to distinguish daughter strands following replication through the lack of methylation on certain adenine bases. The daughter strand is then selectively removed by a methylation sensitive enzyme such as endonuclease MutH. A restriction enzyme called DpnI also acts on the same site and specifically digest the unmethylated strand. It did not take much imagination to realise that DpnI is the perfect tool to selectively remove the template without affecting unmethylated DNA generated in vitro.
Yet the easiest way to synthesise DNA, namely PCR, is riddled with random errors that it was limited to small fragments. Yields with the more commonly used T4 DNA polymerase were poor and selection was still carried out via phenotype or special vectors for the next decade. Finally the hurdle was removed with discovery of proof-reading thermostable DNA polymerases, which allowed the direct use of any dsDNA with thermal cycling, greatly reducing the complexity of DNA amplification.
Thanks to technology, most commercial "fast" mutagenesis kits use thermal cycling followed by DpnI digestion to remove the template before transforming them to cells on the same day, yet their internal mechanism may be more different than you think.
The simplest implementation popularised by Stratagene is known as Quickchange. Two overlapping primers containing the desired mutation are used with a proof-reading DNA polymerase to generate nicked plasmids; DpnI is added to remove methylated/hemimethylated source DNA, leaving behind only the mutated strands to transform E coli where the nicks are repaired.
Strategene's own (idealised) illustration
Theoretically the protocol is fast, easy and almost foolproof provided that your enzymes have not expired, yet in practice people (yours truly included) frequently run into all sorts to bizzare problems. The troubleshooting guide supplied with the kit point fingers in every possibly direction: Too much or too little template, bad plasmids, primer dimers, poorly controlled cycling, not enough respect for their super awesome competent cells (I am quoting verbatim) blah blah blah without a single mention of the biggest problem: excessive 5'->3' extension
In a normal PCR, extension requires little attention as long as there is enough time for the enzyme to finish building the new strand. Any initial overshoot will be corrected in the subsequent cycles by the other primer. However with a circular template and overlapping primers, the conditions of extension must be optimised to prevent the enzymes from displacing its own product and continuing to extend down the 3' end. The product will not form a nicked circle and hardly transforms into E coli if any. No polymerase is immune to the problem and the likelihood increases with each cycle as primers become used up, leaving the template free to anneal with the product. Unlike the nicked circles, over-extended ssDNA can act as DNA template further depleting primers otherwise available for correct binding. (Nicked circles cannot act as template for synthesis as the strand break is immediately downstream from the primer binding site)
Confusingly, many people report better results with longer extension and some versions of Stratagene's manual recommend 2min/kb extension instead of 1min/kb. Attempts to heed to these suggestions never worked, however the results did help identify the problem. I can't find the rather dramatic gel photo now but the 12-hour long amplification produced some extremely long ssDNA that would not even migrate in 0.7% agarose gel electrophoresis. Instead it formed a very bright halo around the wells that took a number of head-banging to figure out what is going on.
In the end we got it working like I said before. Instead of tinkering with the 9kb entry vector, the gene was shuttled into a smaller vector (3.5kb including the insert) which required a much shorter extension time and less room for error. Fortunately our gene of interest was maintained in a Gateway vector which made the process much less of an ordeal but others may not be so lucky.
If your insert is long or subcloning is difficult:
- Optimise your PCR conditions: reduce cycle number, decrease extension time and temperature, try a gradient of annealing temperatures, change template/primer ratio, increase pH of the reaction buffer, use additives such as DMSO (I add 4% for every PCR involving GC-rich genes, some difficult template may require as high as 10%) or betaine.
- Buy a specialised polymerase (variously referred to as "non-strand replacing" or "with DNA clamp") that is less likely to have issues with extension. PfuUltra seems to be a popular choice. A new product from Takara known as PrimeStar Max boasts an extension time of 5s/kb; a mutagenesis protocol tailored for the enzyme uses very small amount of template and skips DpnI digest entirely as it is easily outnumbered by products after an amazing 30 cycles.
- Redesign your primer to be partially overlapping. These primers are less likely to anneal to each other and may allow you to get away with a lower template concentration. I get a small yet noticable increase in colony numbers after I redesigned my primers with a short 3' overhang.
- Buy 5' phosphorylated primers and do a ligation after DpnI digestion. This might help stablising the double strand for some long mutations that does not like to form nicked circles.
A better system offered by Finnzyme known as Phusion uses an alternative strategy. Instead of using a pair of overlapping primers going in circles, it uses two "back to back" primers that amplifies away from the mutation site to form a linear strand which is subsequently phosphorylated and ligated before transformation.
Compared to Quickchange, it is much easier to use because
- The PCR product is linear and can be easily checked by gel electrophoresis; Quickchange usually ends with a smear leaving you wondering if it actually worked.
- Ligated DNA has better transformation efficiency
- No need for specialised enzymes and protocols, just follow the manual.
- The primers needs to be highly purified, e.g. HPLC or PAGE. The reason is that DNA oligos are synthesised from 3' to 5' and there are always some molecules with one or more bases missing from the business end and in our case, result in unwanted deletions. Extra purification and can cost a lot is the primers are long and negate the benefits of the method.
- Ligation of blunt-end strands usually requires overnight incubation, so one more day is needed before the DNA is ready for sequencing.
Dill's Refined Quickchange Mutagenesis Protocol
Before you start:
- Check your template by digesting it with DpnI and run a gel alongside the undigested plasmid. This step is not mentioned anywhere else but highly recommended because it establishes a few things that often go wrong: Your DpnI is active, your template is sufficiently pure, supercoiled and methylated.
- If your plasmid is stored in TE buffer, consider ethanol precipitation and re-dissolve in water as EDTA can affect polymerase activity, unless your template is very concentrated and only added in small quantities.
- Consult the documentations of your polymerase to work out the best reaction conditions. You often need more polymerase than normal to get the best yield. 100ng of template and 150ng of each primer appears to be a good starting point for point mutations on plasmids from 3kb to 6kb; outside these ranges you might need to experiment to find the best molar ratio.
- Design your primers. There are many way to do this and the best strategy is to use Takara's protocol as a starting point and apply common sense: Add a few more bases when in doubt, especially on the 5' end; try to end your primers on a C or G; avoid repetitive sequences that self-anneals. Strategene's recommendations and their web-based tool should be taken with a grain of salt. Melting temperature is not critical unless you are using some picky polymerase that you should not have bought in the first place; if you absolutely have to, use the nearest neighbor method on the non-mutagenic portions as a guide. Desalted or cartridge purified primers are fine in most cases, and the money is always better spent on a few extra bases on 5' end than more expensive purification steps.
- Get some competent cells with competency of 10^8 or higher. HsdR genotype is preferred since your DNA will be entirely unmethylated.
- Set up protocols according to the polymerase used.
- If there is a final extension step, don't include it.
- Do not exceed the extension time specified for your enzyme.
- 15-18 cycles should be enough, do not exceed 20 cycles.
- Add excess amount of DpnI, usually 10U per 50uL reaction but feel free to add more if have a lot of template. Not to mention the PCR buffer is usually not optimal for DpnI activity
- Vortex and centrifuge the tube to ensure good mixing. Incubate at 37C for 1-2 hours in a PCR machine with heated lid or cover with mineral oil to prevent evaporation. Vortex/spin at least once during incubation.
- (Optional) Add Proteinase K and incubate to stop the reaction. Do not heat inactivate since this might interfere with strand pairing.
- (Highly optional) Phosphorylate purified product and ligate with T4 DNA ligase. CSL recommends this step for all applications but I feel this is only for long mutations that you should not have used a Quickchange protocol to begin with.
- (Electroporation only) Desalt the product if your apparatus is prone to arching; chemically competent cells are better for the purpose.
- Add ligation product to competent cells, heat shock/electroporate , add SOC and shake at 37C for 2 hours instead of the usual 1 hour to allow cells to repair the plasmid. This is more important if your selection antibiotic is bacteriocidal such as kanamycin or streptomycin.
- Plate cells on appropriate selective plates. A good reaction should result in tens to hundreds of colonies.
- Pick 3-4 well-spaced colonies, grow them up, miniprep plasmids and sequence.
- For unknown reasons Quickchange sometimes results in random insertions. Nevertheless there should be at least one colony with the desired mutation and no other errors.
- If there is no colony at all, consider transforming with more products and plate a higher volume. I make my agar plates in advance and store them in the fridge without a bag. After one week they would have lost some of their water content and as a result up to 400ul can be spread easily.
- If no clones can be found after you screen 5-8 colonies, it is very likely that you have had contamination or template carry-over and you should start over and be more careful.
- The efficiency of the reaction can be accessed by blue-while screening compatible cell strains and plasmids. Normally more than 90% of colonies should be mutants - if not there is likely some issues with amplification/digestion.