Several structures of SgrAI bound to cognate (CACCGGTG) and noncognate (G
ACCGGTG) DNA, with Ca2+
, have been determined (Dunten et al. 2008 
and current work). In all, SgrAI forms a dimer very similar to those of Cfr10I and NgoMIV. Alignments of the structures show that SgrAI is more similar to Cfr10I, having some small deletions, and several insertions relative to Cfr10I 
. NgoMIV and Cfr10I form tetramers in the crystal structures, with the tetrameric interface on the side of the dimer opposite to that of the DNA binding site (i.e. tail-to-tail). The new structure of SgrAI described here shows a tetramer that is unlike the NgoMIV and Cfr10I structures, with the tetrameric interface of SgrAI at the DNA binding face of the dimer (i.e. head-to-head) stabilized by the swapping of the amino-terminal 24 residues of each subunit (space filling spheres, ). Residues 25–30 comprise the hinge loop that adopts a different conformation in the swapped form (, ). The SgrAI “swapping” domain is absent in NgoMIV and Cfr10I.
The biochemical data suggest that SgrAI can exist in at least two conformations, with one possessing an inherently greater DNA cleavage activity than the other. The observed stimulation of DNA cleavage activity could be accomplished by shifting the equilibrium from the low to the high activity form, possibly stabilized by higher order oligomers that favor the high activity conformation. The rate of DNA cleavage could be controlled by the positioning of groups in the active site, where the optimal alignment results in faster DNA cleavage kinetics. Analysis of the active sites of all SgrAI structures solved to date (Dunten et al. 2008 
and current work) shows very similar placement of all groups including the DNA in the various crystal structures, indicating that only a single conformation of the enzyme has been determined, which we have argued to be the low activity conformation 
To test the relevance of the domain swapped tetramer in the biochemical activity of SgrAI, two mutants were designed, P27G and P27W, predicted to destabilize the swapped conformation (), through introducing either increased flexibility with the glycine residue or steric conflicts with the large bulky tryptophan side chain. We found that both mutations disrupted the allosteric stimulation of DNA cleavage by SgrAI, without affecting the unstimulated DNA cleavage rate on the primary site sequence (), and without appreciably affecting binding affinity to uncleaved or precleaved primary site (). In addition, the activity of P27W SgrAI on plasmids containing either one or two primary site sequences shows that the presence of a second primary site does not appreciably accelerate DNA cleavage, as it does for the wild type enzyme (Text S1
, Figures S1
). Further, the cleavage pattern of P27W SgrAI does not involve concerted cleavage of the two primary site sequences (Figure S2
). Thus the plasmid assays also indicate that P27W SgrAI does not form the activated oligomer proposed to explain the fast, concerted cleavage by wild type SgrAI 
In addition to diminishing the ability of the SgrAI enzyme to be activated in DNA cleavage, the mutations were also found to eliminate the formation of HMWS under conditions where HMWS are formed by wild type enzyme (). These results support our previous hypothesis that the HMWS is the activated form of SgrAI 
. They also support a role for the interface between DBD seen in the structure of the domain swapped tetramer presented here in forming HMWS. Although species as small as tetramers are suggested by the accelerated cleavage of plasmids containing two primary site DNA sequences 
, our measurement of HMWS formed by wild type SgrAI and primary site containing DNA indicates species much larger than tetramers are formed 
. Therefore if the tetramer found in the crystal structure is a building block of the HMWS, a second interface between the DNA bound dimers in addition to the domain swapped interface must exist, in order to form run-on oligomers of the size and heterogeneity seen in HMWS; an attractive possibility is the interface used by NgoMIV and Cfr10I ().
The effect of the mutations on binding to secondary site DNA was unexpected. Wild type SgrAI binds to both primary and secondary site DNA very tightly, with slightly tighter affinity (~4-fold) for the primary sequence 
. Therefore wild type SgrAI seems to discriminate very little between the two sequences at the binding level. Yet these single site substitutions, P27W and P27G, affect affinity very strongly for the secondary, but not the primary, sequence where the KD
is shifted from nanomolar to micromolar. The origin of this effect is unknown and requires further investigation.
The SgrAI biochemical and structural data have some similarities to those of another type IIF restriction endonuclease, SfiI 
. SfiI is a tetramer in solution 
that cleaves two copies of its recognition sequence in a concerted manner. The crystal structure of SfiI with its recognition sequence DNA show a tetrameric arrangement similar to that of NgoMIV, although the subunit structure is more like the dimeric BglI 
. The conformation is identified to be in an inactive state since the DNA is mispositioned in the active site and only one of the predicted two divalent cations (Ca2+
in the crystal structure) is bound. The low pH of the crystallization conditions may be responsible for the lack of the second divalent cation binding 
. However, DNA cleavage data show that three recognition sites on the same DNA molecule are cleaved before enzyme dissociation, rather than the predicted two 
. These data were interpreted as dissociation of one of the two sites cleaved concertedly followed by reassociation and cleavage of the third site prior to enzyme dissociation. Given the model for SgrAI, it is tempting to speculate whether SfiI is fully active also only in oligomers higher order than tetramers, explaining the concerted cleavage of three sites and the inactive conformation of the tetrameric species solved in the crystal structure. However, no direct evidence of oligomerization beyond tetrameric species has been reported for SfiI.
The allosteric communication network has been investigated in Bse634I, another type IIF endonuclease that bears very close structural similarity to SgrAI 
. Bse634I is a tetramer in solution and cleaves DNA fastest when both DNA binding sites are occupied with its recognition sequence. However, when only a single site is occupied, the DNA cleavage rate is reduced. Hence it possesses both auto-inhibition and stimulation capacities. While we have shown that DNA cleavage by SgrAI is stimulated (>200-fold, 4-fold greater than the 50-fold stimulation of Bse634I), it is not known if auto-inhibition also occurs. For auto-inhibition like that in Bse634I to occur, SgrAI dimers not bound to DNA would need to associate with DNA bound SgrAI dimers and decrease the DNA cleavage rate. Although SgrAI is dimeric in the absence of DNA binding 
, we have shown by gel shift measurements that oligomerization of the SgrAI dimers occurs significantly only with significant concentration of DNA bound dimers (i.e. above 100 nM) and not with excess SgrAI that is not bound to DNA 
. However, the measurements of the stoichiometry of DNA binding by SgrAI performed with fluorescence anisotropy are suggestive of a second SgrAI dimer binding to the DNA bound SgrAI dimer. The single turnover DNA cleavage assays reported for SgrAI 
have been done with a substantial excess of SgrAI over the DNA, and if a second SgrAI dimer (without bound DNA) binds to the DNA bound SgrAI dimer, then all reported rate constants have been performed with this additional dimer associated with the enzyme-DNA complex, and with any concomitant auto-inhibition. Investigation of auto-inhibition awaits measurements done with 1
1 ratios of SgrAI and DNA.
To our knowledge, this is the first clear example of reversible domain swapping functioning to modulate the natural biological activity and specificity of an enzyme. Among previously reported examples 
, that of the RNase enzymes is strongest. RNase A, from bovine pancreas, forms oligomers during lyophilization in acetic acid 
, and the dimers and trimers have been shown to be domain swapped 
. Although the conditions for forming the oligomers are artificial, dimerization has been observed at pH 6.5 and 37°C with an equilibrium dissociation constant of 2 mM 
. The enzymatic activities of the oligomers indicate that hydrolysis of double stranded RNA is faster with the oligomeric forms than with the monomeric, however virtually no difference is seen in the activities of the dimeric and monomeric species 
. The related bovine seminal RNase, BS-RNase, exists as two interconverting dimers with only one stabilized by domain swapping. The enzyme exhibits cooperativity, but only at very high substrate concentrations (0.3 mM) and the effects are relatively small (1.2–1.3-fold) 
. The domain swapped form is required for immunosuppression activity, but this activity is not a natural biological function of the enzyme 
. Therefore, the potential use of domain swapping by SgrAI in a natural function of DNA cleavage rate stimulation and DNA sequence modulation may be the first clear case of a reversible domain swapping used to alter biological activity. This would also be the first case where DNA stimulates such domain swapping.
The unusual DNA cleavage activity of SgrAI may be a consequence of the large genome of Streptomyces griseus, from which it is derived. Restriction endonucleases are always coexpressed with a methyltransferase enzyme having the same sequence specificity, which functions to protect the host genome from the cleavage activity of the endonuclease. Hence the SgrAI methyltransferase must methylate all SgrAI recognition sequences within the genome before cleavage by the endonuclease can occur, and this requirement may be difficult due to the large size of the genome (over 8 million bp). The relatively long sequence recognized by SgrAI, 8 bp versus the usual 4–6, may have evolved due to this pressure, since the longer sequence greatly reduces the number of sites to be methylated in the host DNA. In addition, the inherently low cleavage activity of SgrAI in the absence of significant concentrations of unmethylated primary site DNA also reduces the pressure on host DNA, as well as the methyltransferase enzyme. However, such a long recognition sequence will also occur far less frequently in the phage DNA and hence place selective pressure on the enzyme for increased activity in order for adequate protection of the host from phage infection. It appears that one way in which the SgrAI enzyme activity is increased is through the stimulation of its cleavage activity with sufficient concentrations of unmethylated primary site DNA. Another way is through its secondary site cleavage activity, which will induce more cleavages in the phage DNA than at the primary sites alone, and hence could better protect the host. However, to protect against cleavage of the secondary sites in the host genome, the oligomerization may function to sequester activated SgrAI enzymes on the phage DNA and away from the host genome. It may also have an important role in sequestering the phage DNA itself or in rapidly communicating positive allosteric signals to multiple binding sites.