Mutate or die: Atomic structures explain bacterial SOS induction

About 50 y ago, Miro Radman proposed the SOS hypothesis: In response to significant amounts of DNA damage, a highly error-prone mode of DNA replication is induced in Escherichia coli under the control of the recA and lexA genes (1). The recA gene had previously been identified as playing a central role in homologous genetic recombination (2). The RecA protein was one of the most abundant proteins in E. coli after DNA damage and thus was the focus of intense interest. An early picture emerged that was relatively simple and at the same time puzzling: Single-stranded DNA (ssDNA) appears in the cell as a result of DNA damage, the RecA protein binds to this ssDNA and forms a filament (3, 4), and the resulting nucleoprotein filament catalyzes the cleavage of the LexA protein (5). Since the LexA protein is the repressor for recA and ~40 to 50 other genes, the cleavage of LexA inactivates its repressor function and induces the expression of these linked genes, which is termed the SOS response. One of the puzzles was that RecA protein can induce high levels of expression of the recA gene, a positive feedback loop that was initially considered inconceivable. The same RecA nucleoprotein filament that induces the cleavage of LexA also induces the cleavage of UmuD, and the resulting UmuD’ is a central part of the UmuD’₂C heterotrimer, an error-prone DNA polymerase that is known as pol V (6). Our understanding of the details and significance of the SOS response continues to grow, and a new cryoelectron microscopy (cryo-EM) study (7) provides a large amount of detail at the atomic level into how RecA interacts with LexA and UmuD as well as with the bacteriophage λ CI repressor and the regulatory DinI protein. All of these additional proteins bind in the deep helical groove of a RecA-DNA filament (Fig. 1), as shown in early low-resolution studies (8–10).

Fig. 1.

Atomic model for the complex of RecA (green ribbons), ssDNA (red spheres), and LexA dimers (yellow and magenta spheres) (7). The inner LexA is in yellow, and the outer LexA is in magenta. RecA will first polymerize on the ssDNA that may appear in the cell, typically when replication reaches a lesion. The LexA subunits bind in the deep helical groove of the RecA-DNA filament which stabilizes a conformation of the LexA protein that results in autocleavage. The cleaved LexA is inactive as a repressor, and 40 to 50 genes under the repression of LexA are now activated in the SOS response to DNA damage.

The new cryo-electron microscopy (cryo-EM) study provides a large amount of detail at the atomic level into how RecA interacts with LexA and UmuD, as well as with the bacteriophage λ repressor and the regulatory DinI protein.

While initial results suggested that RecA acted as a protease in the cleavage of LexA, it was subsequently shown that LexA cleavage was autocatalytic, with the site for the protease activity within LexA itself (11). Bacteriophages, in their continuing battle with their hosts, have developed repressors that mimic LexA. Thus, when a cell turns on the SOS response, it is taken as a signal to the phage that the host may be in great distress and that this is a good time to move from a lysogenic to a lytic phase. In other words, the phage might want to jump ship to insure its survival. As a result, the repressors for phages P22 (12), ɸ80 (13), and λ (12) have evolved to also undergo RecA-mediated self-cleavage during the SOS response. Similarly, UmuD autocleavage during the SOS response is assisted by RecA. While the autocleavage is intramolecular in LexA and the λ CI repressor, it is intermolecular in the UmuD dimer that binds to the RecA filament.

What the high-resolution cryo-EM study (7) shows is despite large structural differences between UmuD, DinI, the λ repressor, and LexA, all three bind to the same elements in the RecA-DNA filament. One of the main elements involves the L2 loop of RecA, which has previously been shown to invade double-stranded DNA (dsDNA) when a RecA filament is formed on ssDNA as part of the process of homologous recombination (14). They also show that the cleavage site regions in UmuD, the λ repressor, and LexA are intrinsically flexible, and binding to the RecA filament stabilizes these regions in a cleavage-competent conformation. This allows RecA to act as a coprotease for autocleavage reactions since these autocatalytic events would be greatly inhibited if the proteins were not bound in the deep groove of the RecA filament. The structures show how a RecA filament that was actively engaged in a strand exchange reaction with a homologous dsDNA molecule would be inhibited from catalyzing the cleavage of UmuD, the λ CI repressor, or LexA due to the fact that the L2 loop would no longer be available.

Mutations in DNA can be thought of as the engine of evolutionary change (15). We all know that DNA-damaging agents, whether chemicals or radiation, induce mutations in DNA and that most mutations are either neutral or deleterious. But the process is not simple or direct. A photon does not convert a T to a C, or a G to an A, for example. Rather, the damage to DNA by the photon or the chemical results in such a mutation during either the process of DNA repair or in the replication of damaged DNA. Some lesions in DNA would block normal replication, and Evelyn Witkin in 1967 suggested the presence of an inducible system in E. coli that could be triggered by only a few UV-induced pyrimidine dimers in the entire genome, with the induction allowing for both trans-lesion DNA synthesis as well as inactivation of phage repressors (16). This suggestion was extended by Radman into the SOS hypothesis (1). It is convenient to think of the activation of “sloppy” polymerases, such as pol V, that can on the one hand replicate past a lesion that would stall a normal polymerase (17) but on the other hand fail to replicate faithfully, as a necessary compromise in evolution. But an alternative picture has also emerged.

The strength of a population to survive in an environment that can change in unpredictable ways lies in the genetic diversity within that population, and mutations are one of the main mechanisms for the introduction of genetic diversity. Therefore, increasing the rate of mutations during periods of great stress can be an evolutionary advantage rather than a penalty (18, 19). Inaccurate DNA replication during adversity may have been selected for during evolution due to the increased fitness of a population with greater genetic diversity. Thus, the details in the study by Gao et al. (7) for how an error-prone polymerase is activated can be extended to suggest that there may be a direct evolutionary line between the proteins responsible for mutagenesis in bacteria to human cancers (20).