To the editor:

The paper by Kazuya Watanabe and colleagues published in the January issue (Nat. Biotechnol. 23, 88–93, 2005) is framed both in the paper itself and in the accompanying News & Views by Jo Handelsman (Nat. Biotechnol. 23, 38–39, 2005) as an astute, general genetic procedure for fishing out catabolic genes from environmental metagenomes. The proposed method exploits the concept of guilt-by-association by assuming that genes determining transcriptional regulators of pathways for metabolism of many chemical species lie adjacent to the very catabolic genes they regulate. On this basis, genes encoding potentially interesting enzymes can be captured through the fluorescence-activated cell sorting of substrate-induced Escherichia coli cells bearing random green fluorescent protein fusions to metagenomic DNA fragments. Although the approach has potential in selecting transcriptional factors that respond to given chemical compounds, for several reasons it is likely to work in only a very few instances in environmental screens for novel catabolic enzymes, as I outline below.

The first problem is that transcriptional regulators may not be associated with operons controlled in cis. Although the long-evolved regulatory network of E. coli1 does contain such gene arrangements, they are much less common in recently evolved metabolic pathways, where the most interesting and newest enzymes are to be found.

A second, more important, caveat is that catabolic promoters for xenobiotic-degrading operons may not be regulated and thus are expressed constitutively or semi-constitutively. It is very unlikely, therefore, that these types of genes would show up in the screen of Watanabe and colleagues.

A more troubling difficulty with the procedure is that transcriptional regulators are frequently activated by effectors that are not substrates of the pathways that they regulate. Conversely, substrates for interesting enzymes may not always induce expression of the corresponding genes; in fact, there is a considerable promiscuity in how substrates induce enzymes2,3,4, which is likely to endow the screening system described by Watanabe and colleagues with considerable noise from false positives and false negatives.

A final problem (which is shared by all E. coli-based strategies for surveying metagenomic libraries) is the uncertainty about expression of the genes encoded by environmental DNA sequences in the surrogate bacterial host. Although genes born by Gram-negative bacteria for catabolism of long-existing, natural hydrocarbons are likely to be picked up with this procedure, the most recently evolved genes encoding metabolic enzymes for more recent xenobiotics are likely to be missed because of the suboptimal evolution of their regulatory circuits in the short time frame2. Needless to say, the strategy of Watanabe and colleagues is also blind to archaeal DNA, fungal DNA and any other DNA encoding transcriptional factors that cannot work in concert with the RNA polymerase of E. coli. New procedures for screening the whole enzymatic contents of microbial metagenomes are still badly needed to tackle these challenges in environmental biotechnology.