The paper yields a comprehensive, almost exhaustive, database analysis of toxin-antitoxin genes from Prokaryotes (Bacteria and Archaea). The toxin genes of the known seven toxin-antitoxin (TA) gene families encode highly efficient inhibitors of cell growth. Most of the known toxins inhibit translation by a unique mechanism: they cleave mRNA positioned at the ribosomal A site and thereby block translation.

One enzyme, e.g., RelE, can cleave many mRNAs and thereby clog the translation apparatus very efficiently. Thus, the toxins encoded by TA genes are very efficient inhibitors of bacterial cell growth, including pathogenic bacteria.

The paper represents an enormous amount of database work—gene mining in the DNA databases. Via this comprehensive, almost exhaustive work, we elucidated a striking and unusual phylogenetic (evolutionary) pattern, and reached the incontrovertible conclusion that the seven known toxin-antitoxin gene families are abundant (almost ubiquitous) in free-living prokaryotes but virtually absent from obligate intracellular bacteria.

This is an important observation that points to the conclusion that free-living bacteria that encounter nutritional stress in changing environments benefit from having toxin-antitoxin genes. Moreover, we show that the toxin-antitoxin gene families have very complex phylogenetic patterns unlike any other known gene family.

This work puts a lot of genes (2 x 671) from 126 organisms into their right functional and phylogenetic context, and is therefore highly useful for the scientific community. We developed a novel method to exhaustively search for toxin-antitoxin genes in prokaryotic genomes. This method may be useful in other cases as well.

We have found a large number of new genes in Bacteria (and Archaea) that, when activated, can kill or stop the growth of the organisms in which they reside. So, either bacteria can kill themselves (some bacterial groups work to prove this "suicide" hypothesis) or bacteria benefit from having a large number of regulators that coordinate the speed of cell growth to nutrients in the environment—that is the theory that we have proposed and found experimental evidence to support.

I like to compare toxin-antitoxin genes with gears on a bike. Lance Armstrong needs many gears to adjust the speed when he encounters the steep uphill increment in certain mountains. Thus, one of the most slowly growing pathogens, Mycobacterium tuberculosis, has 60 TA gene pairs, an almost unreal number (Mycobacterium leprae, an obligatory intracellular pathogen, has none, even though the M. leprae genome was derived from that of M. tuberculosis by reductive evolution).

Many of my peers consider me as one of the pioneers of the toxin-antitoxin field. Way back in 1986, we published a unique antisense RNA-regulated mechanism that confers so-called "postsegregational killing" of bacteria, a term that I invented when I was writing my Ph.D. thesis. During the ’80s and ’90s, my group unraveled a beautiful and complex RNA folding mechanism controlled both by an antisense RNA and the folding pathway of the mRNA that is the target of the antisense RNA.

It's probably still the best-described regulatory mRNA folding pathway known. I stopped working with antisense RNAs since many of the general journals thought that our work was too esoteric. We encountered several depressing rejections of manuscripts describing highly sophisticated RNA "gymnastics." Therefore, I switched to the other type of toxin-antitoxin loci in which the toxins are counteracted, not by antisense RNAs, but by protein regulators.

We could rapidly see that this was also an exciting and fertile research field since toxin-antitoxin genes are ubiquitous in all Archaea and in all free-living Bacteria (but perhaps not as sexy as antisense RNA-regulated "killer" toxins).

Obviously the exhaustive mining and accurate annotation of 618 gene loci (2 x 618 genes) in 118 genomes was a frustratingly big task. Several times during the two and a half years that it took, we thought that we had them all. However, new branches kept popping up. But the new method that we developed made it easier and we now have exhaustively mined and accurately annotated 1,240 toxin-antitoxin loci in 218 prokaryotic genomes together with all plasmids in the National Center for Biotechnology Information (NCBI) genome database.

We have described that a significant fraction of highly pathogenic bacteria have many toxin-antitoxin loci that, when activated, can kill the harmful invaders—e.g., Mycobacterium tuberculosis has sixty (60!) toxin-antitoxin loci. Several, or perhaps many, groups are now working towards developing toxin-antitoxin genes in pathogenic bacteria as promising targets in the development of drugs aimed at eradicating persistent infections.