Maybe your primers are contaminated, try changing ALL your reagents (included the 16S primers), if the band still there, maybe you have a contaminated working area (are you using PCR chamber? or something else?).

P.S: I can't see the attached file :(

Good luck! :D

I can't see the file but... I've been having similar problems after increasing Taq concentration and number of cycles (up to 35) with 27Fl/1494Rc primers and at the end seems that the contamination source is actually the water itself.

Get some commercial master mix (maybe you can get a sample), some molecular biology grade water and new primers. You should be able to trace your contaminant with those and several PCR combinations. BTW, load more PCR product in the gel and do up to 35-40 cycles if you want to be sure that no amplification is happening. Also, run sample only in alternate lanes on the gel and load first the negatives, sometimes if you have some strong + and some floats out of the well can be seen as a false +.

Plasticware and pipettes can be contamination sources, try using filter tips and DNase/RNase free tubes, many manufacturers do not recommend to autoclave tubes and tips, as DNA can be transferred during autoclaving in form of aerosols, especially if it's a shared autoclave where you kill bugs or autoclave waste.

Thank you Alejandro and Xabier, I will order some brand new pcr water and do different combination of trials, keep everything clean. let you know how it goes.

But I'm wondering if in the worst case, could I send the negative control for sequencing as well together with all my other samples? then in the sequencing analysis part, could I simply subtract the OTUs in the NTC from my real samples?

Hope this will not be the case anyway

If your PCR is for sequencing single products, e.g. 16S of a pure culture, and you have strong bands in your samples you may be able to sequence without problems as the contamination is a very small amount of the total DNA to sequence and it will appear as chromatogram noise. However, this is specially problematic for samples with limited DNA content and/or environmental samples that will be send for community analysis such pyrosequencing, as they can actually appear in your results as an important, but not real, part of the microbial community.

You are using universal bacteria primers so the chance to get contaminated is relatively high. Sometimes even the DNA polymerase may contain trace amount of DNA from the host bacteria that express this enzyme. This will cause the problem you mentioned. Besides using all new PCR reagents, one solution is to simply reduce PCR cycles to about 30 and you will get enough PCR yield while diminishing the negative controls. I find it work very well for me. 

I had similar problems before. It could be the Taq polymerase that contains E. coli DNA. Try to use a different Taq. Or you can use Montage PCR filter to filtrate each component (except DNA template) before PCR.

See my paper:

An anion-exchange method to concentrate dissolved DNA from aquifer water.

Hi, I did get it working with no band in the negative control by doing several things.

I changed the water to PCR grade water, also prepare the primer working solution with this water as well. also I decreased the volume of primers to decrease the probability 

then i did the test with my both old water and this new water for 10 samples, both are working.

But I got the problem when I started to use the 96 well plate to do more samples with the band appearing in ntc, so I decided to prepare master mix for every 8 samples only, so I have 12 tubes of individual master mix for the whole plate. And no band at all in ntc!

Thanks everyone for your contribution! :)