I just found this table in the NEB webpage showing activity of restriction enzymes directly in the PCR mix (at least some of them) it gives you an idea of the most sensitive enzymes even if you use a different polymerase.

https://www.neb.com/tools-and-resources/usage-guidelines/activity-of-restriction-enzymes-in-a-q5-taq-or-phusion-pcr-mix

It's best to purify it before digestion (and afterwards). You have left enough overhanging bp before the recognition site for the RE, right?

Yes.  Its better to purify the product and it takes only 2 min using the columns.  This gives you more confidence with cloning and avoid problems and suspisions in the future.

There are several considerations here.  Your PCR product is likely a mixture of many different molecules (e.g. the desired product, primer dimers, and possible a small amount of false product).  So direct cloning will give you some false inserts - possibly not enough to be too worried about; if you run your product out on an agarose gel, is the desired band the only visible band?  Second, your digestions may not perform optimally unless the product is suspended in the appropriate digest buffer.  My recommendation (which is old school, but still works) would be to band isolate the DNA first.  I used to make my own spin-column using nylon mesh; I'd just throw the band in a small amount of mesh, put the whole thing in an eppendorf tube (with the lid holding the mesh in place so that the entire band/mesh "sac" was suspended midway through the tube).  The resulting liquid is going to be essentially the correct product, with minimal junk - ready to digest in the appropriate buffer etc.

Yes, I always do gel-purification before using them for RE digestion. Run your PCR product on a gel to first check whether your expected band is amplified. If amplified, cut out the specific band and use a kit to gel-extract the DNA out of the gel block. I use Qiagen Gel Extraction kit (Cat. #28704) for this purpose (there are other methods without using a kit). And as Rumpf suggested, this practice might also reduce the chance to clone an non-specific amplified PCR product into your vector.

Maybe better not make confusions, here... Whatever is your PCR product, you could in principle digest it, but the PCR buffer will not allow you to do so, you'll need to substitute with a specific RE buffer. So, BEFORE RESTRICTION, as suggested by Alejandro and Prasad, two options: 1. Use a spin column to desalt your product and elute in 10 mM Tris/HCl pH 7.4, then add your RE buffer and digest, or 2. Just precipitate your PCR product in 0.3 M NaOAc and 2.5 vol. of ethanol, resuspend as above and digest.

AFTER digestion you will definitely need to GEL PURIFY your product, since REs cut palyndromic sequences and you will have the same overhang on your product and on the small endings of your product, but these little pieces will have a kinetic advantage during ligation and you will practically be unable to clone. In this step you will also get rid of any possible unspecific bands, as Robert and Yuan-Yeu were saying.

Pietro's got the most straight-forward approach - go with it.  :)

Yes, I agree. Pietro got a better approach. In my suggestion above, you need to get another run of purification later on after RE digestion. But, still, you need to make sure you have the PCR product first.

Yes, purify you PCR product from agarose gel, then digest it, after digestion you must see the results on agarose gel and purify it again. 

If you have single PCR product , digestion is recommended after purification(column), again purify it before ligation.Simple

I am repeating most of them mentioned here... Check about 1 ul of your PCR product make sure you have right size fragment....Then do column purification followed by digestion...anyway after digestion run on a gel and purify the product and setup for ligation...Initial gel purification is not necessary ...

If you have a single prominent band on a gel it is not absolutely necessary to purify the product before digestion - I recently did exactly the same experiment without doing so and it worked perfectly. But be sure to purify your DNA after digestion (preferably gel-extraction, but my guess is spin-columns that remove primers should also work just fine), as the quality of DNA is very important for ligation.

The PCR buffer could interfere with the digest (decrease efficiency or specificity), but if you'll dilute it sufficiently (like 5 ul PCR reaction into 25-50 ul digest) you should be OK.

As Ilya wrote, you should purify it afterwards, because the small pieces could interfere with ligation. However, be  careful with the choice of the method as we have often problems with gel purification.

Surprisingly, in my experience even with only 2 times dilution of PCR reaction digestion works just fine. But you never know how it will work with a different PCR buffer and a different restriction enzyme, so it is a concern you should have.

And as you might know with gel extraction you will lose ~50% of your DNA (or even more), so if you have a reason to be worried about amount of DNA for ligation, be sure to set up a few PCRs at the same time and pull the DNA together at some step.

It will depend on the salt sensitivity of the enzymes. Some restrictions enzymes are more stringent about their optimum activity than others. You can compare the activity in presence of amplification buffers once with purified product. 

Yes, you do.   Purify your specific PCR product on agarose gel, then cleavage this and following purification again.

Wow, wow! Thank you ALL for very timely contributions. They have been of great help.

Hello guys! Could you tell me how should pcr product look like after successful single digest? Thanks! 

Hi Altynay,

unless you cleave off large piece of DNA, it should be indistinguishable on agarose gel electrophoresis, if that is what you mean. What I mean is, that if you have linear PCR amplicon of e.g. 1 kbp and cleave off just the ends, so about 10 bp in total, you will not see difference between 1000 and 990 bp.

Of course, if you cleaved circular DNA, the situation would be different (but it is unlikely you will amplify whole plasmid with PCR and then digest, as that is done with site-directed mutagenesis).

If you cleaved off large piece of the DNA (say you'd amplify 1500 bp and cleave away 500 bp, there would be difference). But then why would you use such primers? But this is somewhat likely scenario.

I just found this table in the NEB webpage showing activity of restriction enzymes directly in the PCR mix (at least some of them) it gives you an idea of the most sensitive enzymes even if you use a different polymerase.

https://www.neb.com/tools-and-resources/usage-guidelines/activity-of-restriction-enzymes-in-a-q5-taq-or-phusion-pcr-mix

After the PCR run a 2 microlitre of the PCR product on the gel so as t check for the nonspecific bands or product amount. Then you can go ahead with the digestion if every thing looked fine on the gel. But do perform Gel purification after the digestion (PCR purification might lead to the contamination of the cut portion of the plasmid which can affect the ligation).

Totally agree with Pietro Pilo Boyl, he was given the most appropriate approach, and I would suggest you another way to purify the PCR product using phenol-chloroform purification as follow: [if the PCR product volume is 100 ul then it follows that you should add equal volume of Tris and chloroform]: 100 ul phenol (Tris), 100 ul chloroform, 12 ul of NaCl (5M), 70 ul isopropanol (100%) and then re-suspend in 1xTE, while you are finishing these steps you can go directly and digest your PCR product.

Another thing to consider is the NEB Cleavage Close to the End of DNA Fragments table here: https://www.neb.com/tools-and-resources/usage-guidelines/cleavage-close-to-the-end-of-dna-fragments

I wanted to digest my PCR product but the enzymes require more than one nucleotide at each end; I only introduced one extra nucleotide. :/