You may suspend the colony in 1 ml nuclease free-water and boil for 10 min. Then, you can use 3ul of the suspension as PCR template. The rationale is very simple, to avoid too high concentration of PCR inhibitors.

Try extending the initial denaturation time in PCR reaction to 5 minutes.

Thanks, extending the initial denaturation time is a sensible step, and I've found the same suggestion floating around the web-boards, so it's going in my protocol. Suspending the colony in NFW to dilute PCR inhibitors also sounds like a good suggestion, but I'm concerned that doing so will also dilute the template copy density, which would be problematic for my assay, as I'm trying to extend the dynamic range at the low-end. Therefore, I'm considering washing the colony by suspending it in NFW, pelleting, then resuspending in as small a volume of NFW as possible (say 50 uL), and using that suspension as template in my reaction.

Since you are amplifying the single-copy gene from genomic DNA from limited cells (not multiply copies of plasmids), you should also consider using DNA polymerases specified for amplifying low-copy template. Low-copy templates require polymerases with high processivity and fidelity to prevent possible errors being amplified from a small pool of starting material.

Why not to grow a single colony for a while and then do a proper gDNA extraction. I have abandoned colony PCR due to frequent false negatives.

I agree. Why don't you enrich the bacteria first. Pick a pure colony from a agar plate, enrich overnight in 2 ml broth in a rocking incubator, centrifuge it, suspend pellets in 500 ul sterile distill water, boil and centrifuge, use 1ul supernatant for a PCR reaction. Overnight enrichment in broth will increase the gDNA ultimately the target while maintaining the purity of organism.

I could enrich overnight, wash, and process, but that takes enough time that it eliminates my primary impetus for performing colony PCR for transformant and recombinant screening: saving time. I could always simply grow a colony in selective media and perform kit-based gDNA purification, but it adds an overnight growth cycle to my procedure, which I'd like to eliminate for high-throughput screening. I too have had false positives with colony PCR, but I think that it's a tractable problem (at least that's my premise; we'll see). As for the polymerase, I am using a high-fidelity (55x Taq) pfu-based enzyme with high processivity, so it sounds like I don't need to worry on that count. Good to know :)

We have not had problems with colony PCR that we could trace back to PCR inhibitors, so unless you know (or can empirically prove to yourself) that your assay is struggling due to inhibitors, you could err on the side increased template.  We typically directly stab our pipettor tip-picked colonies into the PCR reaction mix (typically after first touching the tip lightly to a new agar plate to grow and save the clone).  You might try a side-by-side to see if you get better efficiency with diluted vs. undiluted PCR colones (with all that free time I'm sure you have in the lab ;).

Are you simply looking for "positive" colonies, or are you trying to sequence the fragment you amplify?  If it's the former, then you don't need a proofreading enzyme and could try good ol' Taq or other fast enzymes to see if they work better in your system.

Cheers

you need LB plates and could go for different colony PCR say 5 reactions and each CFU you use for colony PCR mark them  1,2,3,4,5  simultaneously streak it in LB Platesand mark them as 1,2,3,4,5  . now after amplification  check which colony is giving your desired product. say colony 5 given your product. now again go for colony PCR and take colony 5 as template.  

hope this will work all the best. 

Why not try using a direct PCR polymerase to bypass the DNA extraction portion in the nuclease-free water, thus allowing you to have the direct application of the template DNA in the PCR reaction? It may be worth a try, especially if you know that your template concentration may be low because it is a single-copy gene and you think that the nuclease-free water may be inhibiting that gene.

I use Takara/Clontech SapphireAmp master mix regularly to amplify >5 kb targets from K12 (MG1655) colonies. Use 10-uL tip to grab a bit of a fresh colony directly into the reaction (10 uL for screening; 50 uL to get enough DNA for several sequencing reactions). This polymerase works far better than any I've used in the past and it's cheap in master mix form. Also works for cerevisiae CPCR (same protocol). I follow the manufacturer's protocol except that I do longer preincubations at lower temperatures (4'/95 C) and sometime longer extensions or touchdown cycling if there's any problematic non-specific bands.

Kristin, thanks for the suggestions! I think that my extension time is sufficient for the DNAP that I'm using (it's highly processive; the manufacturer recommends 15s/kb, and not more than 1 min/kb because of the increased likelihood of secondary products), but I haven't yet played with reaction volume. To answer your question, I've been using small/young colonies because I've read that large/old colonies are enriched in PCR inhibitors.

Brendan, I have had similar problem in my experiments too. One thing I noted was the concentration of the bacterial you dilute and concentration you get after dilution makes a huge difference in your PCR results. I have found both excess and limited concentration of bacterial can cause the PCR product to decline. You might want to take separate colonies and dilute in different volumes ranging 20-100 ul of water and use different volume say 1-5 ul of the diluted template in the PCR reaction mix. Once you know the ranges for better results, you can optimize for you need. Give it a try if you have not, this has helped me maybe it will work for you too. And about the extension time, I am not sure about 15s extension for 1 kb amplicon size I normally do 45 seconds for 1 kb, so you may wanna try a minute if you are giving less extension time for you PCR reaction. I have known that you can also use DMSO to increase the specificity, but when I used it increased the specificity but decreased the amplicon amount (something you don't want but you can think about if this would help).

Thank you everyone for your responses and advice! My crowd-sourced protocol now looks like this:

Screening Colonies for Genomic Insert from Agar Plates

I may try this protocol; if I do, I will also do gDNA purification of stationary phase culture grown overnight from step 1. I can then compare the results to see how much coverage this protocol gives relative to a gDNA purification kit.

More steps can be added, such as the addition of a concentrated salt solution to cell suspension followed by centrifugation to pellet precipitated lipids, addition of proteinase K to denature histones, and ethanol precipitation of liberated DNA. If such careful extraction is required, I can use a genomic DNA purification kit (which should include all these additional steps in its standard protocol).

My motivation for this protocol is to screen colonies for a genomic insert with minimal sample preparation time, as the economics of my research currently dictate that time costs more than materials. I would like to eliminate the 8-hour growth cycle required for a gDNA purification protocol. I may be able to process the 100-μL cell suspension from step 1 without growth to stationary phase; for reasons of determining the dynamic range of my assay, I have previously prepared 10-fold serial dilutions of a sample of purified gDNA diluted to 10e+6 copies/mL, and detemined that the stochastic copy-number limit for my particular reaction was 10e+2 copies/mL. So I would only need (as a bottom limit) 1000 cells delivered to the suspension by pricking the colony. Sounds promising!

Caveat: I only have need to screen from a plate occasionally (historically, once-a-month), so by tweaking this protocol I stand to save around 8-hours per month. If my impetus for developing the protocol is saving time, it doesn't make sense at this point to put any more time into said development: the cost-benefit ratio doesn't support it. However, should I need to do more screening in the future, that balance will change, and I may perform the above experiments (or give them to an undergrad). If so, I will let you all know how things turn out.