You could increase gene expression by putting the gene on a plasmid. Alternatively if you want to integrate another copy, then you could clone the gene adjacent to a sequence from elsewhere in the genome and integrate it at that location. 

 Thank you for your reply. I have expected that the gene will be integrated into another targeted site. But the results are always negative. Although several candidate strains which are expected to be integration of a gene are selected from the antibiotic selection plate,  all of the candidate strains represent negative results in colony PCR.

(In colony PCR, primers are designed to have band only when the gene is integrated into the right specific site, However, I can't get any band from the candidates)

So I think that the gene I want to integrate should have no homology sequence with any gene in the genome. Is it right?? (I can't not sure about it, but just expectation from my results)

That is why I wonder how to make integration even when the same gene exist in the genome. 

Assuming you have built your construct correctly, it should all be a matter of size. If you have two regions of homology, one being the gene itself and the other being the homology target elsewhere, the frequency of integration at each site should be proportional to the size of each region of homology. So if both are similar size, then you should get 50% at each site. 

First you should determine whether you can get any integration at all, perhaps there is a problem with the protocol or the strain. You might also just be having problems with the PCR itself, do you have any sort of control so that you know the PCR and the primers are working?

You could also design primers to see whether you are getting homology at the unwanted site (ie the gene itself in the genome). 

Homologous recombination should work if you overlapping regions and selection markers are well designed. We regularly insert native genes in alternate location in E.coli but the selection marker and ends are the key. You may want to provide larger flanking homologous regions to your constructs (the larger the better)

However, If the conventional homologous recombination does not work, Clone-integration is another approach to do this. 

https://www.addgene.org/45981/

This is one of the vectors, i have used in my studies and works well. You an insert any promoter in front of your gene by overlap extension PCR and integrate into the genome of E.coli or salmonella.

 Dear Himanshu Sharma 

Thank you for reminding me that the native gene also can be integrated in E.coli well.

I wonder that how long you make homology region??normally

How many colony you can get as selection??( Did you normally check it by colony pcr?)

And what was the percent of sucessfully integrated stain portion from the colony ???

Thank you  :)

I usually follow the method developed by datensko et. al.

 https://www.ncbi.nlm.nih.gov/pubmed/10829079

The method is explained for knock-outs but with some clever design you may use it for "knock-in"s as well.

I usually use a 50 bases region of homology on each side. But i had to use around 200-400 bases as well.

Based on my experience, the efficiency of knock-out depends on the vitality of the gene you are targeting.  Sometimes i had to screen only 2-3 colonies but sometimes more, so its hard to generalize.

For a knock-in you should not pose the problem of how important the gene is. Having said that,  the site of integration you want to target should be chosen wisely, try using phage integration sites if possible or replacing some unimportant gene. 

you can refer my recent paper for some detailed

info http://www.tandfonline.com/doi/full/10.1080/15476286.2016.1207037?scroll=top&needAccess=true

Thank you for your kind advice. 

If I make success in my experiment, I will share it

Thank you

 Sure, i would be glad if it works for you.