Inclusion of restriction sites in PCR primers

[Harry Witchel]

Kevin Mulcahy <K.Mulcahy at sheffield.ac.uk> wrote:
>I am trying to clone a cDNA using RT-PCR and have designed a primer pair 
>through the help of a computer programme (MacVector). This programme 
>apparently (so long as the settings are approriate) designs primer pairs 
>which do not form primer-dimers or other primer-primer/primer-product 
>artifacts which may inhibit successful PCR amplification. It also 
>suggests an optimal annealing temperature to use for the PCR as a 
>starting point when using the primers for the first time.
>
>However, I added a further 10 bp onto the 5' ends of each primer so that 
>they contained a restriction enzyme site (a different site for each 
>primer) to be used for digestion of the amplified product and cloning 
>into an appropriate vector for sequencing. I attempted to perform the 
>RT-PCR with these modified primers using a range of annealing 
>temperatures, annealing times and extension times but I could never 
>amplify the desired product.
>
>Therefore, does anyone know what rules should be followed in the design 
>of primers when adding restriction sites? Also, are there any general 
>rules or suggestions for determining the cycling parameters for PCR when 
>using these modified primers?
>
>

Dear Kevin --
   The first rule of thumb with PCR is that the computer programs provide 
guidelines and suggestions, but computer programs DO NOT PROMISE SUCCESS. 
 You do not say whether you also tried primers which did not have the 10 
base pair sequence for restriction cutting; it is likely that your 
primers would not have worked whether or not you add the restriction 
sequences.  Here are:

RULES FOR ADDING RESTRICTION SITES TO PCR PRIMERS:

1)  Choose enzymes which cut near the end of DNA (esp. EcoRI, BamHI).  
There is a table in the back of both the New England Biolabs and 
Stratagene catalogues comparing the ability of different enzymes to cut 
right at the end of a piece of DNA; the other enzymes that are listed 
(but I personally have not tested out) are PstI, KpnI, & StuI.

2)  Put four bases 5' to the cutting site.  You have already done this.  
I tend to use GAGA, but any combination which does not use 4 identical 
bases should be OK.

3)  Don't worry about what your computer says about the sites.  Most 
programs should pick out that your site is palindromic -- just enter your 
primer onto the computer without the cutting site.  Since the palindrome 
is at the 5' end, you will probably not have any interference during the 
pCR reaction.  Real PCR problems are created at the 3' end of the primer. 
 I have heard of people using primers which are 70 bases long, with only 
the last 15 bases matching the sequence to be amplified, and the PCR 
still worked (albeit inefficiently).  
   I personally ignore computers entirely when designing PCR -- you 
should just follow the same rules as the computer: no long runs (5 or 
more) or the same base, avoid palindromes near the 3' end, and make 
certain not to have any possible matching between the two primers at the 
3' end (the very last base should not be complementary, and any alignment 
between the last 4 bases should be checked by eye).

4)  Spend money on new primers.  If you go into the PCR business, be 
prepared to spend money, otherwise you are wasting your time.  I have 
repeatedly found that when a primer set did not work the first time 
(trying 3 different temperatures), doing all the PCR optimisation was a 
glorious waste of time whereas buying new primers often resulted in 
instant success.  New primers may mean choosing a completely different 
region, or it may mean offsetting your sequence by 3 or 4 bases.  PCR is 
very finicky, especially with regard to primer choice and template 
quantity -- a set of primers might work well with a purified plasmid as 
template, but when you go to the cDNA or genomic DNA there is nothing.  
Commercial primer sets should cost you #50-60; if not, Perkin-Elmer (no 
affiliation) is quick, cheap, and reliable.

5)  Use feedback from the reaction.  On the gel of your reaction, did you 
get a large smudge at 50 bases -- this is primer dimers, and it means 
that you need to raise the annealing temp, lower the cycle number, or 
design new primers.  If your gel was completely empty, then you genuinely 
got nothing, in which case you should lower the annealing temp, raise the 
cycle number (up to 40 or 50 cycles, or even doing a secondary PCR 
reaction with internal primers), or change primers.  Don't be bashful 
about changing primers.

6)  Check your template.  If your RNA is chewed up (quite possible), or 
your cDNA reaction is not working (less likely), then you will not get 
any good PCR.  One way to test template quality is to use your template 
to amplify glyceraldehyde phosphate dehydrogenase (GAPDH) using the 
primers: 

GAPDH 248: ACC GGG AAG CCC ATC ACC & 

GAPDH 672: CAG CCT TGG CAG CAC CAG

These produce a 400 bp fragment from most mammals (mouse rat rabbit human 
etc).

7)  PCR a smaller region.  Technically speaking, normal PCR should work 
on anything from 1-2 kb, but if your template is rare, primer quality 
uncertain, and there are a large number of other templates in the 
reaction, you may find that the PCR reaction tends to favour reaction 
products with lengths of about 400 bp or smaller.  If you design a 
reaction to make a product of 200 bp, you will be much more likely to get 
what you are looking for when starting with cDNA or genomic DNA.  Small 
PCR products are much more easily analysed on acrylamide gels than on 
agarose, and acrylamide does not turn all the small stuff into a smudge.

ALTERNATIVES TO RESTRICTION SITES IN PCR:

I personally love using Bam and Eco sites in a directional cloning 
strategy for PCR.  At the end of the reaction I phenol chloroform, 
ethanol precip, digest, acrylamide gel, crush elute, and ligate.  It 
works for me.
   If it doesn't work for you, I know people who swear by the TA cloning 
systems.  You ligate immediately after the PCR, and there is no faffing 
about.  I personally have less luck with this method, but it may work 
every time for you.

Write if you have problems.
   Harry