Type I Restriction and Modification Systems

DNA Cleavage

DNA binding

Type I restriction-modification enzymes recognise bipartite sequences, which consist of a 5' specific region of 3-4 bp and a 3' specific region of, usually, 4 bp, separated by a non-specific spacer of 6-8 bp (e.g. EcoR124I recognises GAAn₆RTCG - where R = purine and n = any nucleotide). DNA cleavage by Type I R-M enzymes occurs at non-specific sites far from their recognition sequence, which is mediated by ATP-dependent DNA translocation past the enzyme (Yuan et al., 1980; Studier & Banyopadhyay 1888; Dryden et al., 1997; Szczelkun et al., 1997.

The natural substrate for DNA cleavage is fully unmethylated DNA, which would normally be non-host DNA such as invading bacteriophage DNA or plasmid DNA. The enzyme must recognise the methylation status of the DNA by determining the presence or absence of a methyl group on the two adenines within the recognition sequence (one on each DNA strand). To accomplish this the enzyme flips the methylated base out of the double helix of the DNA (Mernagh et al 1998) and reads the methylation status. The enzyme must then compare both bases, so must flip out the other base and read the methylation status of that base. Restriction activity ONLY occurs if BOTH bases are unmethylated and must involve a switch in the function of the enzyme, such that methylation does not occur and DNA cleavage does occur (Firman et al 2000).

DNA cleavage

Type I R-M enzymes are unique amongst all known restriction endonucleases in that they cleave DNA in a random manner producing fragments of varying sizes from a few hundred basepairs to fragments as large as a few thousand basepairs.

This random cleavage of DNA by Type I restriction-modification systems is a direct result of translocation of DNA by the HsdR subunit. Translocation of DNA is an ATP-dependent process during which the HsdR subunit hydrolyses ATP and moves the DNA past the enzyme complex, which remains bound at the recognition site. The process is bi-directional, each HsdR subunit acting as a molecular motor, and proceeds until the process is blocked by some external event (usually another enzyme also translocating the DNA; although other blockages include DNA topology). Cleavage follows blockage of the translocation, which will produce a random cleavage site because the process of starting translocation and then blocking translocation is not in any way synchronized.

Blockage of DNA translocation can be because of a number of different reasons. As mentioned above the most common reason is collision between two Type I R-M enzymes. This led Studier and Bandyopadhyay (1988) to propose a model for DNA cleavage based upon collision between two translocating enzymes and he was able to clearly show DNA cleavage occurred half-way between two recognition sites on a two-site plasmid. However, our own studies have shown that cleavage of covalently closed circular DNA with a single site is an extremely efficient process, while cleavage of linear single-site plasmid DNA is an inefficient process (Szczelkun et al., 1997). This led us to suggest that DNA cleavage depends upon blockage of the translocation process, which is due to topological barriers to translocation with cccDNA - where translocation can wind all of the plasmid DNA into the expanding loop resulting in stalling due there being no more DNA to translocate. Janscak et al. (1999) have also shown that blockage can also be caused by a Holliday junction. However, many DNA binding enzymes appear to not effect the translocation process and can be displaced from the DNA. Recent studies in our own laboratory have also shown DNA binding drugs and compounds such as ethidium bromide also block translocation.

One can better understand the outcomes of DNA cleavage by considering the possible substrates:

(A) Linear DNA molecules:

A linear DNA molecule with a single binding site for the enzyme.
This DNA is not cleaved unless a massive excess of enzyme is present (enough to ensure non-specific binding occurs).
However, ATPase activity is normal and the DNA is translocated as normal.
A linear DNA molecule with two or more binding sites for the enzyme.
This DNA is efficiently cleaved into random sized fragments, but release of the two "end fragments" that are not cleaved. The random cleavage occurs between the binding sites for the enzyme and the result is observed as a smear on an agarose gel.

(B) Circular DNA molecules:

Covalently closed circular DNA (cccDNA e.g. plasmid DNA) with a single binding site for the enzyme.
This DNA is efficiently cleaved by the enzyme and results in linear fragments of full plasmid length, but with random end-points within the plasmid DNA sequence (produced by cleavage at random positions on the plasmid).
Covalently closed circular DNA (cccDNA e.g. plasmid DNA) with two or more binding sites for the enzyme.
This DNA is efficiently cleaved into random sized fragments, which are visible as a smear on an agarose gel.

DNA cleavage was proposed to be stoichiometric, with no turnover and a single cleavage event being performed by a single restriction endonuclease (Eskin & Linn, 1972a & b; Endlich Linn, 1985a & b), but more recently Bianco et al (2009) have shown turnover for these enzymes in a situation where the substrate DNA was digested, releasing the bound enzyme. This raises an interesting question as to whether dissociation of the enzyme from DNA, or dissociation of the subunits can produce turnover of the DNA cleavage.

Simons and Szczelkun (2011) have addressed this question in a recent paper and shown turnover does occur, but there are differences between the different families of Type I R-M enzymes and between the different types of substrates detailed above. For the Type IA and type IC enzymes EcoKI and EcoR124I respectively, the HsdR subunit can dissociate from the MTase, following DNA translocation, and from the DNA substrate provided there is a free end. However, the MTase of EcoKI and EcoAI can dissociate from the DNA independent of free DNA ends, but subsequent turnover requires freely available HsdR in solution. This type of subunit-induced provides an interesting mechanism for regulating restriction activity of Type I R-M systems in vivo.

Perhaps the most controversial concept of the Simons and Szczelkun paper is their proposal that release of HsdR from the DNA-bound MTase is accompanied by further translocation of HsdR toward the end of the DNA molecule (see figure opposite). The presence of DEAD box motifs in HsdR has suggested for some time that independent translocation should be possible, but there has been no direct evidence of this.

The model proposed by Simons and Szczelkun is that on DNA held in a Magnetic Tweezer setup (A) the translocating HsdR is unable to release the DNA, but free HsdR in solution can bind the released, DNA-bound MTase. However, on linear DNA, the free DNA end allows release of this translocating HsdR, into solution allowing it to rebind MTase. On circular DNA the lack of free ends to the DNA once again prevents release of the translocating HsdR.

What is very interesting from these observation is that cleavage of circular DNA, producing free ends from which HsdR should be able to "escape" does not allow turnover of the HsdR. The reason for this has yet to be elucidated, but two models are proposed:

The HsdR subunit remains bound to the DNA following cleavage so that it cannot be released.
The HsdR subunit is altered by the cleavage process in such a way that it cannot translocate again.