Lambda Exonuclease, an ATP-independent molecular motor

Perkins TT, Dalal RV, Mitsis PG, Block SM. “Sequence-dependent pausing of single lambda exonuclease molecules,” Science. 2003 Sep 26; 301(5641):1914-8.

Lambda exonuclease is a simple and elegant processive molecular motor. This tiny enzyme is a donut-shaped homotrimer that degrades double-stranded DNA (dsDNA) in the 5'-3' direction to create long single stranded overhangs used in recombination by the bacteriophage lambda. In addition, Lambda-exo does not use an external fuel molecule as an energy source (like kinesin which utilizes energy from the binding and hydrolysis of ATP), but instead uses the energy released in the degradation of its DNA substrate to produce motion. It has been suggested that Lambda-exo may be an example of a Brownian ratchet because of its lack of fuel source and structural simplicity. By studying this enzyme, we hope to better understand how it converts energy from the hydrolysis of DNA into motion. Our approach is to examine Lambda-exo pausing using both single molecule and complimentary bulk biochemistry methods. Recently, we discovered a series of sequence-dependent pauses in lambda exonuclease motion.

Crystal structure of Lambda-Exonuclease.
(Kovall and Matthews, Science 1997)

In our single molecule measurements, the Lambda-exo was attached to a glass coverslip (sitting on a piezoelectric stage) and was bound to a piece of M13 DNA whose distal end was attached to an optically trapped bead (see below). We measured the stage position and bead position, from which we determined the length of the dsDNA as a function of time. Precision measurements (with 10 nm or ~ 30 basepair resolution) of Lambda-exo motion revealed sequence-dependent pausing. This pausing was found to be directionally dependent: i.e. running the enzyme along the same template in opposite directions produced pauses at different locations. This observation ruled out measures of DNA stability that affect both strands as possible causes of pausing.

Single molecule experiment with Lambda-Exonuclease. The optical tweezers allow us to: (i) track the length of the dsDNA as the enzyme degrades one strand and shortens the double stranded segment, and (ii) apply a constant force to the enzyme as it progresses, allowing for a detailed study of its kinetics.

To determine which sequence was most correlated with pausing, we performed a consensus analysis to compare the single molecule data dwell time histogram against functions of the DNA sequence. We created a scoring function for an n-base trial sequence of DNA (n=5-9 nucleotides); this scoring function was similar in concept to a sequence alignment scoring function. The scoring function was calculated for all 4n possible n-mers, and these were each compared against the single molecule dwell time histogram. This analysis revealed that the sequence most highly correlated with pausing was GGCGA.

Single molecule traces. (Left): Multiple traces from the forward direction show pausing at stereotyped locations. The pause at 900 nm is particularly strong. (Right): Traces from the backward direction (same piece of DNA, with enzyme moving along opposite strand of DNA as determined by location of 5’ phosphate) show that there is no strong pausing around 900 nm.

We also investigated pausing using bulk sequencing gel assays, which give single basepair resolution. In these assays, lambda exo digests short fragments of 3’ radiolabeled dsDNA; the reaction is stopped at various time points and the products are run on a large sequencing gel. Pauses on the sequencing gels are identifiable as distinct bands corresponding to locations where a significant fraction of enzymes halted. We ran two different sets of experiments. In the first set of experiments, DNA sequences flanking major pause sites identified through single molecule data were created by PCR and tested to see which pause sites would be visible in the gels; these experiments showed that only the 900 nm pause was visible on gels (other pauses were too weak).

Polyacrylamide gel analysis of pausing. (A) Exonuclease digestion of PCR products ( 300 bp) flanking several pause sites identified in single-molecule records, Pnnn is the position in nanometers. Numbers (red) above and below each lane specify the starting and ending bp of the M13 sequence. L, 10-bp ladder; C, control lane no enzyme; 90, 180, digestion by enzyme for the times in seconds indicated. Arrows (red) show locations of pauses in single-molecule data. The dominant pause site at 900 nm corresponds to a strong band (lanes P900); other sites do not. (B-D) Flip experiments. (B) WT displays the pause band at 900 nm in the forward (black box) but not in the backward direction. Afor and Aback show the identical pause pattern (red box). The band in Bback shows that the pause has been transferred to and offset in the backward strand (blue box). (C) 18for and 18back show that pause has been disrupted in the forward strand and successfully transferred to the backward strand (red box). (D) (Top) Cartoon showing enzyme orientation and strand digestion (dotted trace). Lower four drawings illustrate the flip sequences. Pause site, p (green); directions of digestion (black arrows); relative orientation of DNA sequences (red and blue arrows). WT, unmodified DNA sequence; FlipA and FlipB, sequences where 20 bp upstream or downstream of the pause site were swapped to the opposite strand, respectively; and Flip18, sequence where 18 bp on either side of the pause were swapped to the opposite strand.

In the second set of bulk experiments, we used 100 bp oligonucleotides that flanked the major pause at 900 nm. We flipped various segments of DNA before, after, or around the main pause site. These experiments revealed that the pause site was some subset of the sequence GGCGATTCT, the sequence immediately following the pause. This sequence agreed with the sequence determined from the single molecule correlation analysis (GGCGA). These experiments also revealed that pausing is caused by an interaction between nucleotides under the footprint of the enzyme and residues in the channel of the enzyme.
So, why might lambda exonuclease pause? One possibility is that pausing is a recombination related control mechanism. The left cohesive end of the lambda genome contains a nested GGCGA sequence (GGCGGCGA); we speculate that lambda exonuclease may partially explain an observation that recombination is more frequent from the right end of the lambda genome, although the packaging enzyme terminase is also known to bind to the left end of the genome. Another possibility is that pausing may allow binding of single stranded binding proteins such as the lambda beta protein or the E. coli RecA behind the enzyme. Alternatively, pausing may allow lambda exonuclease to dissociate from the substrate, although we did not see any evidence for this from our single molecule data. Finally, sequence specific pausing may be an evolutionary remnant that the enzyme inherited from its close cousins, the Type II restriction endonucleases (e.g. EvoRV, PvuII).

In future studies, we hope to learn more about the energetics of this interesting enzyme.