PCR is based on the principle of DNA replication. If inside the cells, DNA is replicated from a single copy (or two, in 2n), I guess that by PCR you can also do it, specially because PCR doesn't have all the regulatory elements that cells posses to control DNA replication.
every method has a sensitivity limit. In the case of PCR, it depends on the way of the detection of amplicons and the lengh of your target and number of PCR cycles (more cycles can also produce nonspecificity).
Maybe you could find interesting to study a single-cell PCR approach.
yes with good primers 30 cycles is one billion times more amplimer and the use of nested pcr primers gives even more product. Many people and labs worldwide have done it
Theoretically it is possible, but in practice, it is difficult to achieve. I addition to all the above suggestions, do take care of your primer sensitivity and template topology. Linear templates are amplified more efficiently and with higher sensitivity. if you are using a circular super-coiled template, such as that of bacteria, try to linearize it with a RE that cuts outside the amplicon region.
Just a word of caution, the more you increase sensitivity, the more is the assay susceptible to contamination.
I think a more interesting question is, how can you know a priori if you have a single copy of your target gene in your PCR reaction? Because that's prone to stochastic effects and is Poisson distributed, right?
Let's do a thought experiment.
Firstly let's assume that you're going to add one ul of your DNA template to your reaction, and on average there is one copy of your target gene per ul of your template DNA.
When you pipette one ul of your template there is a probability that you'll capture zero copies of your target, one copy of your target, or more than one copy of your target, and the Poisson distribution tells us that in that case you have about a one in three chance of each of those outcomes.
In other words, assuming end point PCR, if you get no amplification, it does not necessarily mean your target is not present in your template DNA, it only means that by random chance there was none in your reaction. And if you get some amplification you cannot know if that is due to a single copy, or more than a single copy. So you'll need to run at least three technical replicates.
But also from the Poisson distribution we know that if you have on average three copies of your target gene per ul of your template DNA, then you have a 95% probability of capturing at least one copy in your PCR reaction.
These things are easier to assess with qPCR where one clearly sees an increase in the variation in Cq among technical replicates when the number of copies of your target drops to ten and below.
Relatedly, in my experience of qPCR the Cq of a single gene copy is around about 32 - 34. So if you are attempting to detect 1-3 copies of your target in an end point PCR to run on a gel, your're probably going to want to run 35 cycles. If you do have a single copy of your target in your reaction, and you run 30 cycles, you probably aren't going to see that on a gel, even if your amplification worked. On the other hand if you use some stain such as Biotum GelRed, which claims to be far more sensitive than EtBr or SYBR Safe, then you will increase the sensitivity of your PCR.
Anyway, you have to think about PCR on very low copy number targets probabilistically.
That analysis is absolutely correct for solutions of dna Alun Parsons so possibly the solution is to add 1 cell (easier to view) directly into the pcr tube before cycling
The amplification of sample DNA depends on the number of primers added in PCR solution. If add less or only one primer, then it may be possible to only a single of amplified DNA.