Restriction Enzyme Digest Setup: Reaction Table, Buffers, and Troubleshooting

A restriction digest that “doesn’t work” is almost never a mystery once you know where to look: the wrong buffer for a double digest, Dam/Dcm methylation blocking the site, too much enzyme causing star activity, or a vector that re-ligates because it was never dephosphorylated. This guide is written as a setup procedure and troubleshooting reference, not a definition — it covers how to build a digest reaction correctly the first time, what actually determines your cloning strategy (Type II vs Type IIS, blunt vs sticky ends), and how to read a failed digest by its symptom rather than guess.

Type II vs Type IIS Enzymes, and Why It Matters for Golden Gate

Most restriction enzymes used in routine cloning are Type II: they recognize a specific, usually palindromic sequence (commonly 4–8 base pairs) and cut within or immediately adjacent to that same sequence. EcoRI, BamHI, and HindIII are all Type II. Because the cut site and the recognition site are the same place, the enzyme’s recognition sequence always survives at the cut ends — which is exactly why two fragments cut with the same Type II enzyme can be ligated back together and re-cut by that enzyme again.

Type IIS enzymes are different in a way that matters a lot for assembly cloning: they recognize a non-palindromic sequence but cut at a fixed distance outside it. BsaI, for example, recognizes GGTCTC and cuts one nucleotide away on the top strand and five on the bottom, leaving a 4-nucleotide 5′ overhang whose sequence is not fixed by the enzyme — it’s whatever sequence the experimenter places at that position in the construct. Because the cut is offset from the recognition site, a correctly assembled junction no longer contains the recognition sequence and can’t be re-cut, while any unreacted or mis-ligated fragment still carries it and gets cut again. That’s the mechanism Golden Gate assembly (and MoClo-style modular cloning) relies on: enzyme and ligase run together in one pot, and the reaction is driven toward the correct multi-fragment assembly because only the correct product escapes further digestion. If your project involves multi-part assembly rather than a simple two-piece insert-into-vector clone, this is the property to design around, not the recognition sequence itself.

Recognition Sites and Overhangs You’ll Actually Use

Recognition sequences, cut positions, and methylation sensitivity for a given enzyme are catalogued authoritatively by REBASE (the restriction enzyme database maintained by New England Biolabs) and in vendor catalogs (NEB, Thermo Fisher, Promega) — always confirm the exact site and buffer for the specific enzyme and supplier you’re using rather than relying on a general table, since even enzymes with the same name (isoschizomers) can differ slightly in optimal conditions between suppliers. A few common Type II enzymes, for orientation:

Enzyme	Recognition sequence	Overhang produced
EcoRI	G^AATTC	5′ AATT (sticky)
BamHI	G^GATCC	5′ GATC (sticky)
HindIII	A^AGCTT	5′ AGCT (sticky)
XhoI	C^TCGAG	5′ TCGA (sticky)
SalI	G^TCGAC	5′ TCGA (sticky — compatible with XhoI)
NotI	GC^GGCCGC	5′ GGCC (sticky, 8-bp rare cutter)
SmaI	CCC^GGG	Blunt
EcoRV	GAT^ATC	Blunt

All recognition sequences are palindromic in Type II enzymes — the sequence reads the same 5′→3′ on both strands — which is what allows a single double-stranded enzyme dimer to cut both strands symmetrically.

Blunt vs Sticky Ends: How Overhang Choice Determines Your Ligation Strategy

The overhang your enzyme(s) leave behind isn’t incidental — it decides how the downstream ligation has to be designed:

• Sticky (cohesive) ends have a short single-stranded overhang that base-pairs with a compatible overhang before ligase seals the nick. This is far more efficient than blunt ligation because the overhang holds the two ends together while the enzyme works. Cutting a vector and insert with two different sticky-end enzymes gives you directional cloning for free — the insert can only go in one orientation because the two ends aren’t interchangeable.
• Compatible overhangs from different enzymes can be mixed: XhoI and SalI both leave a 5′-TCGA overhang, so a XhoI-cut fragment ligates cleanly to a SalI-cut fragment even though the two enzymes have different recognition sequences. The hybrid junction usually destroys both original recognition sites, which is useful to know if you were planning to re-cut there later. NEB publishes a compatible cohesive-ends chart that’s the standard reference for finding these pairings.
• Blunt ends have no overhang, so ligation depends entirely on random collision and blunt-specific ligase activity — it’s slower and less efficient, and it’s non-directional (an insert can go in either orientation unless you control for it another way). Blunt cloning generally needs a higher DNA and ligase concentration and a longer incubation than sticky-end ligation.
• Overhang mismatches from incomplete fill-in or chew-back can be created deliberately: T4 DNA polymerase or Klenow fragment can fill in a 5′ overhang or remove a 3′ overhang to convert a sticky end to blunt, when you need to force two otherwise-incompatible ends to ligate.

Whichever route you cut, dephosphorylation matters: a vector cut with a single enzyme, or with blunt ends, can re-circularize on itself with no insert at all, because the two ends are complementary to each other. Treating the linearized vector with a phosphatase — CIP (calf intestinal alkaline phosphatase) or SAP (shrimp alkaline phosphatase) — removes the 5′ phosphate ligase needs, so a self-closing vector can’t be sealed while a vector-plus-insert junction (which supplies its own phosphate from the insert) still can.

Setting Up a Digest: The Reaction Table and What a “Unit” Actually Means

A standard single-enzyme digest has five components: DNA, enzyme, reaction buffer, BSA (if the manufacturer recommends it for that enzyme), and water to bring the reaction to its final volume. A representative small-scale reaction, following the setup format used in Addgene’s restriction digest protocol:

Component	Amount
Plasmid DNA	~500 ng (diagnostic digest) to 1 µg (cloning digest)
Restriction enzyme(s)	1 µL each (see unit calculation below)
10× reaction buffer	3 µL (i.e. 1× final)
10× BSA (if recommended)	3 µL
Nuclease-free water	to 30 µL total

Total reaction volume commonly runs 10–50 µL depending on application, and is largely set by how much DNA volume you’re adding. Diagnostic digests (checking a plasmid’s identity) typically use less DNA and shorter incubations than digests feeding into cloning, which usually want at least 1 µg of clean DNA and a longer digest to ensure complete cutting.

What a unit means: by the standard definition used across suppliers, one unit of restriction enzyme is the amount required to completely digest 1 µg of DNA in a 50 µL reaction in one hour under the enzyme’s optimal conditions. That definition is the basis for scaling a reaction: if you’re cutting 2 µg of DNA, you need roughly twice the units (or twice the time) that 1 µg would require. In practice, most protocols use more enzyme than the strict calculation requires, for two reasons: unit activity is measured against very clean, ideal-condition DNA, and real preps are never quite that clean; and it’s mechanically hard to pipette much less than 0.2–0.5 µL of concentrated stock enzyme accurately, so that ends up being the practical floor regardless of what the unit math says. A modest excess of enzyme is normal and expected — a large excess (see star activity, below) is a different problem.

Mix gently by pipetting (don’t vortex — restriction enzymes are proteins in glycerol-heavy storage buffer and can be sheared or denatured by rough handling), and keep the enzyme on ice or in a cold block whenever it’s out of the freezer; heat exposure during setup is a common, easily avoided cause of reduced activity. Incubate at the enzyme’s specified temperature — commonly 37°C for mesophilic Type II enzymes, but not universally (SmaI, for example, is typically run at 25°C, and thermostable enzymes like TaqI run much hotter) — so check the specific enzyme’s data sheet rather than assuming 37°C by default. Standard incubation is about an hour, though diagnostic digests are often fine at 1–2 hours and cloning-scale digests (more DNA, needing complete cutting) often run 4 hours to overnight.

Double Digests: Solving the Buffer-Compatibility Problem

Cutting with two enzymes at once is where digests most often go wrong, because each enzyme has an optimal buffer and the two aren’t always the same. This is the most-searched practical sub-problem in restriction digestion, and there are three real ways to handle it:

1. Find a shared buffer with acceptable activity for both enzymes. Vendors publish double-digest compatibility tools for exactly this purpose — NEB’s Double Digest Finder is the standard one — which recommend the best available shared buffer and flag the expected percent activity for each enzyme in it. NEB’s newer High-Fidelity (HF) enzyme line and its rCutSmart buffer system are designed specifically to reduce this problem: a large share of NEB’s current catalog is formulated to run at or near full activity in a single common buffer, which removes the compatibility question for many common pairings. Always confirm the specific two enzymes in the tool rather than assuming compatibility from a prior project — buffer systems and enzyme formulations change between product generations.
2. Digest sequentially with a cleanup step in between when no shared buffer gives good activity for both enzymes. Cut with the first enzyme in its optimal buffer, heat-inactivate or column-purify to remove that buffer and stop the first enzyme, then resuspend in the second enzyme’s optimal buffer and cut again. This costs time and a cleanup step but avoids the accuracy tradeoff of running either enzyme outside its ideal conditions.
3. Accept a partial-activity tradeoff for a shared buffer that’s suboptimal for one enzyme, and compensate with more units of that enzyme and/or longer incubation time. This works for many pairings and is the fastest option, but it increases the risk of star activity from the enzyme now present in excess (see below) — it’s a reasonable choice for a quick diagnostic digest, less so for a digest feeding a cloning step you can’t easily repeat.

Star Activity: What Causes It and How to Avoid It

Star activity is relaxed sequence specificity — the enzyme cuts at sites similar to, but not identical to, its normal recognition sequence, producing extra cuts you didn’t design for. It’s usually caused by one or more of:

• Excess enzyme relative to the amount of DNA — a large unit excess (well beyond the modest surplus that’s normal practice) increases the chance the enzyme binds and cuts near-cognate sites.
• High glycerol concentration in the final reaction. Enzyme stocks are typically supplied in ~50% glycerol for storage; adding a large volume of enzyme (to compensate for low apparent activity, for instance) can push the reaction’s final glycerol concentration up. Keeping glycerol at or below roughly 5% of the final reaction volume is the standard vendor guidance for avoiding this.
• Non-standard buffer conditions — wrong ionic strength, incorrect pH, or substituting Mn2+ for the enzyme’s normal Mg2+ cofactor — can all relax specificity.
• Extended incubation well beyond what’s needed for complete digestion, especially combined with excess enzyme.
• Organic solvent contamination (residual DMSO or ethanol carried over from a DNA prep) can also induce star activity in susceptible enzymes.

The fix in most cases is straightforward: use the minimum enzyme amount that reliably completes the digest in the time you have, use the enzyme’s recommended buffer rather than a “close enough” substitute, and don’t let a digest run dramatically longer than necessary. NEB’s High-Fidelity (HF) enzyme variants are specifically engineered to reduce star activity risk relative to their standard counterparts and are worth using when available, particularly for digests that need to run overnight or use enzyme in excess for a double-digest tradeoff.

Methylation Sensitivity: Why a Digest Fails on Plasmid DNA from a Standard E. coli Strain

This is one of the more common invisible failure modes, because nothing about the reaction setup looks wrong. Standard cloning strains of E. coli (DH5α, JM109, and similar) carry two native methylases: Dam, which methylates the adenine in GATC sequences, and Dcm, which methylates the internal cytosine in CCWGG sequences (W = A or T). If a restriction enzyme’s recognition site overlaps a Dam or Dcm site, methylation at that position can block or reduce cutting entirely — and because the plasmid was still purified correctly and the reaction was set up correctly, this looks exactly like a generic “no cutting” failure unless you know to check for it.

A useful illustration of the same biology working in the opposite direction: DpnI cuts GATC only when the adenine is Dam-methylated, while its isoschizomer MboI cuts the same GATC sequence only when it is unmethylated. Plasmid DNA prepped from a standard Dam+ strain is a substrate for DpnI but not for MboI — which is exactly why DpnI is the standard reagent for eliminating methylated parental plasmid template after site-directed mutagenesis, since newly PCR-amplified product is unmethylated and survives, while the original Dam+ template is degraded.

If an enzyme isn’t cutting and you can’t otherwise explain why, check the enzyme’s methylation sensitivity against a vendor reference (NEB publishes a table of Dam/Dcm-sensitive restriction sites) before assuming the enzyme, buffer, or DNA prep is at fault. The fix, if methylation sensitivity is confirmed, is to propagate the plasmid in a Dam−/Dcm− E. coli strain (commercially available, e.g. NEB’s dam−/dcm− competent cells, or strains such as JM110/GM2163) instead of a standard cloning strain.

Stopping the Reaction: Heat Inactivation vs Column Cleanup

If the digested DNA is going straight into gel electrophoresis, you generally don’t need to stop the reaction at all — loading dye and the gel run do that. If it’s going into a downstream enzymatic step (ligation, a second digest in a different buffer, blunting), you typically need to either heat-inactivate or clean up first:

• Heat inactivation — commonly 65–80°C for around 15–20 minutes, though the exact temperature and time is enzyme-specific and should be confirmed against the manufacturer’s card, since not every enzyme is heat-labile enough to fully inactivate this way.
• Column or bead cleanup (a silica spin-column kit, or SPRI/magnetic beads) physically removes the enzyme and buffer salts rather than relying on denaturing the enzyme. This is the more reliable option when an enzyme isn’t heat-labile, when you need to fully exchange into a different buffer for a second digest, or when residual salt/enzyme could interfere with a sensitive downstream step like ligation.

For gel-based work, see this site’s agarose gel electrophoresis guide for running conditions and reading the resulting band pattern.

Partial Digests (When You Want Incomplete Cutting)

Occasionally the goal isn’t complete digestion — for example, when a plasmid has multiple copies of the same site and you want a population of linear fragments cut at just one of them, to build a partial digest library or to linearize at a subset of available sites. The standard approach is to deliberately under-power the reaction relative to a complete digest: reduce enzyme units, shorten incubation time, or lower the temperature, then stop the reaction at defined intervals (a time-course) and run each timepoint on a gel to identify the condition that gives predominantly single-cut product without complete digestion to the smallest fragments. Because this depends on enzyme lot activity and DNA quality, a partial digest generally needs to be empirically optimized for each specific plasmid and enzyme combination rather than following a fixed recipe.

Verifying the Digest on a Gel

Run the digest (alongside an undigested control and a sizing ladder) to confirm the cut matches what your plasmid map predicts. A few things to watch for beyond simple band-counting:

• Supercoiled, uncut plasmid runs anomalously fast on a standard gel relative to its actual size — an uncut control can look smaller than a linearized fragment of the same length. Don’t use an uncut lane to estimate size; use a ladder.
• The number of bands should match the number of cut sites for a circular starting molecule: N sites in a circular plasmid should produce N fragments (a single cut linearizes to one band at the plasmid’s full length; two sites produce two fragments whose sizes sum to the plasmid’s total length).
• Fragment sizes summing correctly is a stronger check than band count alone — two fragments that individually look plausible but don’t sum to the expected plasmid size usually indicate either a partial digest (uncut molecule co-migrating with a similarly-sized fragment) or an unplanned extra cut site.

For general gel-running parameters (percentage, buffer, stain choice, and reading the resulting bands), see the agarose gel electrophoresis protocol guide.

Troubleshooting by Symptom

No cutting at all

• Check methylation sensitivity (Dam/Dcm) against a vendor reference table before assuming anything else is wrong — see the section above.
• Confirm the buffer matches the enzyme’s requirement, especially in a double digest run in a compromise buffer.
• Confirm incubation temperature — not every enzyme runs at 37°C.
• Check DNA purity: residual phenol, ethanol, EDTA, or high salt carried over from extraction can inhibit enzyme activity even when the DNA concentration looks fine. Column-cleaning the input DNA is a fast diagnostic step.
• Confirm the enzyme itself is active — repeated freeze-thaw cycles, time left out of the freezer at room temperature, or an expired lot can all reduce or eliminate activity.

Partial cutting

• Insufficient units or time for the amount and topology of DNA present — supercoiled substrate digests more slowly than linear DNA of the same length.
• A mixed plasmid population from Dam+/Dam− heterogeneity in the prep, if the site is methylation-sensitive.
• DNA quality issues (nicking, contamination) reducing effective substrate availability.

Smearing instead of sharp bands

• DNA degradation before the digest even started — check an undigested control on the same gel.
• Star activity from enzyme excess or high glycerol, producing a range of non-specific extra cuts rather than the discrete expected fragments.
• Genomic DNA carryover in a plasmid miniprep, which appears as a high-molecular-weight smear independent of the digest itself.

Unexpected fragment sizes

• Star activity creating extra cut sites (more bands, and smaller than expected, than the map predicts).
• A plasmid map that doesn’t match the actual construct — a stale reference sequence after unlogged edits is a common, easily overlooked cause.
• Supercoiled uncut plasmid co-migrating with, and being misread as, a cut fragment — confirm against a ladder, not by eye against other lanes.

Vector self-ligation in downstream cloning

• Missing phosphatase (CIP/SAP) treatment on a vector cut with a single enzyme or with blunt ends — the single most common cause.
• A second, unintended cut site created by star activity, which can regenerate compatible ends the phosphatase step didn’t anticipate.
• Incomplete removal of a small stuffer or drop-out fragment before ligation, which can re-ligate back into the vector.

Frequently Asked Questions

How many units of restriction enzyme do I need for 1 µg of plasmid DNA?

By the standard unit definition, 1 unit fully digests 1 µg of DNA in a 50 µL reaction in one hour under optimal conditions, so in principle 1 unit covers 1 µg. In practice, most protocols use somewhat more than the strict calculation, both because real DNA preps are never as clean as the reference conditions the unit is defined against and because it’s difficult to pipette much below 0.2–0.5 µL of concentrated stock accurately.

Can I use two different restriction enzymes in the same buffer?

Only if their optimal buffers are compatible enough to give acceptable activity for both — check a vendor double-digest compatibility tool rather than guessing. If no shared buffer works well for both, digest sequentially with a cleanup step between reactions, or accept a partial-activity tradeoff and compensate with extra units and time for the less-favored enzyme.

Why didn’t my restriction enzyme cut my plasmid at all?

The two causes that look identical to “nothing happened” but have different fixes are methylation sensitivity (the site overlaps a Dam or Dcm site on DNA prepped from a standard cloning strain) and a buffer/temperature mismatch. Check both before troubleshooting DNA quality or enzyme activity.

What causes star activity in a restriction digest?

Most commonly: too much enzyme relative to DNA, high glycerol concentration from adding a large enzyme volume, a non-standard buffer, or an incubation much longer than needed. Using the minimum effective enzyme amount in its recommended buffer, for no longer than necessary, avoids most cases.

Do I need to heat-inactivate a restriction enzyme before ligation?

If the digested DNA is going into a downstream enzymatic step rather than straight onto a gel, yes — either heat-inactivate (if the specific enzyme is heat-labile at the temperature/time on its data sheet) or clean up via column or bead purification, which works regardless of an enzyme’s heat sensitivity and also removes buffer salts a second reaction might not tolerate.

What’s the difference between Type II and Type IIS restriction enzymes?

Type II enzymes cut within their own (usually palindromic) recognition sequence, so the recognition site survives at the cut end. Type IIS enzymes recognize a non-palindromic sequence but cut at a fixed distance outside it, leaving a programmable overhang and removing the recognition site from a correctly formed junction — the property that makes them the basis of one-pot Golden Gate and MoClo assembly.