It's most likely to be contaminated. I suggest you repeat this assay in your neighbor lab first.

In addition, check your band in a long wavelength UV light and quickly, sometimes your dna could be broken by the UV.

Good luck.

Thanks Lin Bai! It is not rare to see the additional bands smaller than target band. Just the bigger 5kb band confuse me a lot.

Yes. It also confused me. If they are not contamination, a reasonable explanation may be that firstly your target DNA broke into the 2k and 1k bands, and then their ends pair with each other by coincidence :)

Are these bands also present in the sample from which you cut the PCR product?

Hi Alejandro. Yes, I also have those bands in my PCR product. Do you have similar experience?

You need to repeat the PCR with stricter condition, such higher temperature.

Xiang,

This is anecdotal evidence, so take it for what it's worth. Through the years, I've observed, repeatedly and under a variety of experimental conditions (different buffers, voltage gradients, intensities, etc) that if you overload an LMP agarose gel, there is a 'carry over' phenomenon whereby one band will somehow carry with it significant amounts of the other bands present in the sample. This has nothing to do with peak broadening, i.e. is not due to a loss of resolution,or to cross contamination while excizing the band, etc. The problem is solved simply by decreasing the amount of DNA loaded into the gel. We use a maximum of 5 ug per band per cm2 of lane cross-sectional area (only because it is a nice round number and the amount is safe; i.e. we have not systematically examined other values).

Hope it helps!