I would strongly recommend to observe MIQE guidelines authored by several PCR Guru's (Bustin, Kubista, Pfaffl, Vandesompele, at least).
It is not only about the value of a publication, but also about normalization and reference genes, the modern and more appropriate term to be used instead of housekeeping genes.
From Chapter 8.1 (ref gene) utility must be experimentally validated for particular tissues or cell types and specific experimental designs (...) Reference gene mRNAs should be stably expressed, and their abundances should show strong correlation with the total amounts of mRNA present in the samples.
Normalization against a single reference gene is not acceptable unless the investigators present clear evidence for the reviewers that confirms its invariant expression under the experimental conditions described.
The optimal number and choice of reference genes must be experimentally determined and the method reported"
generally, so-called "house-keeping" genes involved in basic cellular functions are used for normalisation (16S, tus, rpoD, glyA, dnaB, gyrA, pykA/F, pfkA/B, mdoG, arcA). However, there is no guarantee that they are not affected by your experimental conditions (in fact, you would be surprised how often they are!), and thus need to be tested for stability of expression.
Typically, I would pick several candidate reference genes from the literature and, if available, from published gene expression/microarray datasets (for E. coli, use EcoArray http://www.ecogene.org/ and choose the genes that have low expression values in several gene expression studies, with fold-change below 2).
Then, I would run qPCR on cDNA from your control and treatment samples with primers for these candidate genes, and use geNorm, BestKeeper or NormFinder to check for stability of expression.
Choose those with the lowest variation in expression between conditions, ideally 3-5, but no less than 2 reference genes, as using a single reference gene is not reliable.
In our lab, we tend to stay away from 16S rRNA, simply because it is so much more abundant than our typical mRNA targets. This puts it out of the linear range for accurate quantification under the same conditions used for measuring the target RNA abundance. We have used rpoZ, secA, dnaK, but, as stated above, you will need to determine this empirically for each set of conditions you are testing.