Restriction Enzyme

A restriction enzyme (also termed a restriction endonuclease) is a prokaryotic site-specific deoxyribonuclease (EC 3.1.21.3–5) that recognizes a specific short nucleotide sequence—typically 4 to 8 base pairs (bp) in length—within double-stranded DNA (dsDNA) and catalyzes the Mg²⁺-dependent hydrolysis of one phosphodiester bond on each strand to generate 5'-phosphate (5'-PO₄) and 3'-hydroxyl (3'-OH) termini. Evolved across bacteria and archaea as the effector cleavage arm of Restriction–Modification (R–M) innate immune systems that destroy invading bacteriophage genomes while sparing host chromosomal DNA protected by sequence-specific DNA methylation, restriction enzymes were discovered and characterized in the 1960s and early 1970s by Werner Arber, Hamilton O. Smith, and Daniel Nathans, who shared the 1978 Nobel Prize in Physiology or Medicine. Because Type II restriction enzymes cleave reproducibly within or adjacent to their recognition sites without requiring ATP, their pairing with DNA ligase and plasmid vectors by Stanley Cohen and Herbert Boyer in 1973 launched recombinant DNA technology and modern genetic engineering, where they remain foundational alongside programmable CRISPR-Cas systems.

Host-Controlled Restriction of Bacteriophages and the 1978 Nobel Prize

In the early 1950s, Salvador Luria, Giuseppe Bertani, and Jean Weigle observed a phenomenon termed host-controlled variation: when bacteriophage λ harvested from Escherichia coli strain C was used to infect E. coli strain K-12, the phage plating efficiency dropped by four orders of magnitude (~10⁻⁴)—the phage was "restricted" by the K-12 host. However, the few phage progeny that survived a single cycle in K-12 could subsequently reinfect K-12 with high efficiency (they had been "modified"), yet lost this protection as soon as they were passaged once back through strain C.

In 1962, Werner Arber and Daisy Roulland-Dussoix at the University of Geneva demonstrated that restriction and modification operate on the phage DNA molecule and proposed the Restriction–Modification (R–M) hypothesis: bacterial cells possess a restriction endonuclease (REase) that recognizes and cleaves unmethylated foreign DNA, paired with a cognate modification methyltransferase (MTase) that transfers a methyl group from S-adenosyl-L-methionine (SAM) onto specific adenine or cytosine bases within the same recognition sequence on the bacterium's own DNA. In 1968, Matthew Meselson and Robert Yuan isolated the first Type I restriction enzyme (EcoKI), which cleaved unmethylated DNA at variable distances far from its recognition site. Two years later (1970), Hamilton O. Smith and Kent Wilcox isolated the first site-specific Type II restriction endonuclease—HindII (from Haemophilus influenzae serotype d)—and proved that it cleaved within the symmetric 6 bp sequence 5'-GT(T/C)↓(A/G)AC-3'. In 1971, Daniel Nathans used HindII to cleave the circular 5,243 bp genome of Simian Virus 40 (SV40) into 11 discrete fragments resolved by gel electrophoresis, constructing the first physical restriction map of a genome.

Epigenetic Self vs. Non-Self Discrimination by Cognate Methyltransferases

To prevent auto-digestion of the bacterial chromosome, cognate R–M methyltransferases catalyze one of three base-specific methylations inside the recognition motif:

1. N⁶-Methyladenine (6mA): Addition of a methyl group to the exocyclic N6 amino nitrogen of adenine (such as M.EcoRI, which methylates the central adenine of 5'-GA*ATTC-3'), protruding into the major groove of B-form DNA and sterically blocking endonuclease binding.
2. N⁴-Methylcytosine (4mC): Addition of a methyl group to the exocyclic N4 amino group of cytosine (common in thermophilic bacteria, avoiding the mutagenic deamination risk of 5mC).
3. C⁵-Methylcytosine (5mC): Addition of a methyl group to carbon 5 of the pyrimidine ring (also protruding into the major groove). When chromosomal DNA replicates semi-conservatively, the nascent hemimethylated daughter duplex is resistant to Type II endonuclease cleavage and is converted to the fully methylated state by the maintenance methyltransferase.

Over 4,500 restriction enzymes have been characterized (cataloged in the REBASE database) and are divided into four major classes (Types I–IV):

Nomenclature Convention (Smith and Nathans, 1973)

Every restriction enzyme name follows a standardized code derived from its source organism: the first letter represents the bacterial Genus (E for Escherichia), the second and third letters represent the species (co for coli), an optional fourth letter designates the strain or plasmid (R for plasmid RY13 in EcoRI; d for serotype d in HindIII), and a Roman numeral (I, II, III...) indicates the chronological order of characterization from that strain.

Palindromic Recognition and the PD-(D/E)XK Catalytic Motif

Standard Type IIP restriction endonucleases (such as EcoRI, BamHI, HindIII, and EcoRV) function as C₂-symmetric homodimers that recognize rotational (inverted-repeat) palindromes—sequences of 4, 6, or 8 base pairs that read identically from 5' to 3' on both antiparallel strands (for example, 5'-GAATTC-3' / 3'-CTTAAG-5' for EcoRI). When the homodimer encounters its target site via one-dimensional sliding and hopping along DNA, it inserts recognition α-helices and loops into the major groove, forming 12 to 20 direct hydrogen bonds and van der Waals contacts with the base-pair edges (direct readout) while kinking or unwinding the DNA helix (indirect readout).

Most Type II restriction endonucleases share a conserved five-stranded mixed β-sheet core housing the PD-(D/E)XK catalytic motif, whose acidic Asp and Glu carboxylates coordinate one or two divalent Mg²⁺ ions (Ca²⁺ supports DNA binding without cleavage). Catalysis proceeds via an in-line S_N2 nucleophilic attack without a covalent enzyme–DNA intermediate: a Mg²⁺-activated hydroxide ion (OH⁻) attacks the target phosphorus atom from the backside (180° opposite the 3'-O leaving group), passing through a pentacovalent trigonal bipyramidal phosphorane transition state stabilized by Mg²⁺ and cleaving the 3'-C–O–P ester bond to yield 5'-phosphate (5'-PO₄²⁻) and 3'-hydroxyl (3'-OH) termini ready for resealing by DNA ligase.

Cohesive ("Sticky") Overhangs vs. Blunt Ends and Type IIS Golden Gate Assembly

Depending on the stagger of the two active sites across the dyad axis of the palindrome, Type IIP enzymes generate three terminal geometries:

• 5' Cohesive Overhangs ("Sticky Ends"): EcoRI (5'-G↓AATTC-3', 4 nt 5'-AATT overhang), BamHI (5'-G↓GATCC-3', 4 nt 5'-GATC overhang), and HindIII (5'-A↓AGCTT-3', 4 nt 5'-AGCT overhang).
• 3' Cohesive Overhangs: PstI (5'-CTGCA↓G-3', 4 nt 3'-ACGT overhang) and KpnI (5'-GGTAC↓C-3').
• Blunt Ends (Flush Termini): EcoRV (5'-GAT↓ATC-3') and SmaI (5'-CCC↓GGG-3').

In addition, Type IIS restriction enzymes—such as BsaI (5'-GGTCTC(N)₁↓(N)₅-3'), BsmBI, BbsI, and FokI (5'-GGATG(N)₉↓(N)₁₃-3')—consist of a monomeric sequence-recognition domain tethered to a nonspecific cleavage domain that dimerizes transiently to cut at a fixed distance outside their asymmetric recognition site. Because the 4 nt 5' overhang created by BsaI consists of user-defined nucleotides and the recognition site is spliced out during ligation, Type IIS enzymes enable scarless one-pot Golden Gate Assembly of 10 to 35+ DNA fragments, while fusing the FokI cleavage domain to zinc fingers or TALEs created ZFNs and TALENs.

Statistical Cleavage Frequency, Schizomers, and Star Activity

In a random DNA sequence with 25% base composition, the average distance between recognition sites of length n base pairs is 4ⁿ bp: 4-base cutters (AluI, MboI, TaqI) cleave every 4⁴ = 256 bp; 6-base cutters (EcoRI, BamHI, HindIII) cleave every 4⁶ = 4,096 bp; and 8-base rare cutters (NotI GCGGCCGC, PacI) cleave every 4⁸ = 65,536 bp.

• Isoschizomers: Enzymes from different species that recognize the same sequence and cut at the same position (e.g., MspI and HpaII both cut 5'-C↓CGG-3', though MspI cuts C^5mCGG whereas HpaII is blocked by CpG DNA methylation).
• Neoschizomers: Enzymes that recognize the same sequence but cut at different positions (SmaI cuts 5'-CCC↓GGG-3' blunt; XmaI cuts 5'-C↓CCGGG-3' 5' overhang).
• Isocaudomers (Compatible Cohesive Ends): Enzymes that recognize different sequences but generate identical overhangs (BamHI G↓GATCC and BglII A↓GATCT both yield a 5'-GATC overhang).
• Star Activity: Under non-standard conditions—low ionic strength (<25 mM salt), high pH (>8.0), high glycerol (>5% v/v), DMSO, or Mn²⁺ substitution—enzymes lose specificity and cleave sequences differing by 1 bp from their canonical site, a problem eliminated in modern engineered High-Fidelity (HF) enzymes.

How does DpnI selectively digest parental template plasmid DNA during site-directed mutagenesis?

In PCR-based site-directed mutagenesis, mutagenic primers are extended by a high-fidelity DNA polymerase around a circular parental plasmid template isolated from a standard dam⁺ Escherichia coli laboratory strain. In dam⁺ E. coli, Dam methyltransferase methylates the adenine (N⁶-methyladenine) within every 5'-G^m6ATC-3' sequence on the parental plasmid. In contrast, the newly synthesized mutant DNA strands produced in vitro by PCR contain unmethylated 5'-GATC-3' sites. Treating the reaction with DpnI—a Type IIM (Type IV) restriction enzyme that cleaves methylated 5'-G^m6A↓TC-3' DNA while leaving unmethylated DNA intact—digests the non-mutated parental plasmid template so that the synthetic mutant plasmid transforms competent bacteria.

Why do Type IIP restriction enzymes not require ATP to cleave DNA, whereas Type I and Type III enzymes do?

Hydrolysis of a DNA phosphodiester bond is thermodynamically exergonic (ΔG°' ≈ −20 to −25 kJ/mol) in aqueous solution; therefore, the chemical bond-breaking step itself never requires ATP energy input. Type IIP restriction enzymes (EcoRI, BamHI) find their palindromic target sites by passive one-dimensional thermal diffusion (sliding and hopping along the DNA contour driven by Brownian motion) and cleave directly at the recognition site using Mg²⁺ as a Lewis acid catalyst. Conversely, Type I (EcoKI) and Type III (EcoP15I) enzymes contain Superfamily 2 (SF2) helicase-like ATPase motor subunits (HsdR/Res) that use the free energy of ATP hydrolysis (~1 ATP/bp) to pump and translocate thousands of base pairs of DNA loops before triggering cleavage.

Why are sticky ends ligated more efficiently by T4 DNA ligase than blunt ends?

When two DNA fragments carry complementary 4 nt 5' or 3' cohesive overhangs ("sticky ends"), transient Watson–Crick base pairing (4 A–T or G–C base pairs) and base-stacking interactions between the overhangs hold the 5'-PO₄ and 3'-OH nicks together in duplex alignment at 16–25 °C, increasing the effective local concentration of compatible termini for T4 DNA ligase. In contrast, blunt-ended fragments lack any base-pairing overhang to stabilize their collision complex; ligation requires a simultaneous three-body encounter between the two flush DNA ends and the ligase molecule (K_M for blunt ends is ~50 μM vs. ~0.06 μM for sticky ends), necessitating 10–100× more T4 DNA ligase and crowding agents such as polyethylene glycol (PEG).

1. Smith, H. O., & Wilcox, K. W. (1970). "A restriction enzyme from Hemophilus influenzae: I. Purification and general properties." Journal of Molecular Biology, 51(2), 379–391. https://doi.org/10.1016/0022-2836(70)90149-X90149-X)
2. Pingoud, A., Fuxreiter, M., Pingoud, V., & Wende, W. (2005). "Type II restriction endonucleases: structure and mechanism." Cellular and Molecular Life Sciences, 62(6), 685–707. https://doi.org/10.1007/s00018-004-4513-1
3. Roberts, R. J., Belfort, M., Bestor, T., Bhagwat, A. S., Bickle, T. A., Bitinaite, J., ... & Xu, S.-Y. (2003). "A nomenclature for restriction enzymes, DNA methyltransferases, homing endonucleases and their genes." Nucleic Acids Research, 31(7), 1805–1812. https://doi.org/10.1093/nar/gkg274
4. Engler, C., Kandzia, R., & Marillonnet, S. (2008). "A one pot, one step, precision cloning method with high throughput capability." PLOS ONE, 3(11), e3647. https://doi.org/10.1371/journal.pone.0003647