September 1, 2018 (Vol. 38, No. 15)

Method from New England Biolabs Allows More Fragments, Faster Assembly, and Higher Fidelity

Embraced by the synthetic biology community, Golden Gate assembly is commonly used to assemble 2–10 DNA fragments in a single “one pot” reaction to form complex, multi-insert modular assemblies that enable biosynthetic pathway engineering and optimization. However, current best practices for assemblies of more than 10 modules often rely on two-step hierarchical approaches using different type IIS restriction enzyme specificities at each step. Factors such as enzyme efficiency, stability, and buffer compatibility have placed practical limits on single- or two-step assemblies.

We have reduced these limitations through the use of engineered type IIS restriction enzymes and careful choice of function sequences guided by experimentally derived DNA ligase fidelity data. Our work demonstrates that it is now possible to achieve 20-plus fragment assemblies with both robust efficiency and accuracy (Figure 1).

Development of Test Systems

Our research focused on three different levels of assembly:

  • High efficiency and accurate single-insert assembly
  • Intermediate 5- or 12-fragment assembly
  • More complex 24-fragment assembly

Single-insert cloning based on the acquisition of a selectable antibiotic marker allowed fast throughput testing of efficiencies. This cloning was also compared to a similar-sized lambda amplicon to indicate any possible bias toward suppression of background by antibiotic selection. Screening of transformants by colony PCR confirmed the insertion of the lambda insert at the same high frequencies.

The 5-, 12-, and 24-fragment assembly systems were based on the ability to correctly assemble a lacI/lacZ cassette (designed by New England Biolabs (NEB) for use in optimization of assembly systems) to produce a blue colony phenotype upon growth on LB/Cam/X-gal/IPTG agar plates, indicating successful reconstruction of the coding sequence for β-galactosidase in the lacI/lacZ cassette. Additional confirmation of accurate assemblies was achieved by the sequencing of plasmids isolated from blue or white colonies. Sequencing of blue colonies showed the expected complete sequence for the lacI/lacZ genes, while sequencing of white colonies showed a mixture of misassemblies and, occasionally, uncut or cut/religated pGGA destination plasmids.

A final validation of the 5-, 12-, or 24-fragment test systems was performed by setting up assembly reactions in which a single component was purposefully omitted. Since any assembly is dependent on the presence of every module, destination construct, and functioning type IIS restriction enzyme and DNA ligase, any single omission should block the formation of a complete assembly that would result in a blue phenotype. Indeed, this was seen in all lacI/lacZ assembly test systems—no blue colonies were obtained if any single component was omitted.


All Golden Gate assemblies feature an inverse proportionality between the complexity of the assembly (number of inserts or modules) and the resulting efficiency of assembly (number of transformants)—thus, the greater the number of inserts, the lower the number of transformants. This relationship is often compensated for by plating greater volumes of the outgrowth on the selection plate to achieve enough transformants for downstream screening.

Representative transformation plates obtained from 1-, 12-, and 24-fragment assemblies of the lacI/lacZ cassette are shown in Figure 2. The volume of the 1 mL outgrowth spread on each transformation plate can be manipulated to result in appropriate levels of colony plating densities.

Figure 2. Representative transformation plates of Golden Gate assemblies featuring increasing complexities. Assembly reactions were transformed into competent Escherichia coli cell strains NEB 10-beta (1 fragment) and T7 Express (12 and 24 fragments) and incubated for 16 hours at 37°C. While many cell strains support assembly protocols and 10-beta is routinely recommended due to its ability to maintain large construct plasmid sizes stably, the non-alpha-complementing T7 Express cell strain was used for the lacI/lacZ cassette testing to avoid any possibility of alpha-fragment LacZ complementation.

Ligase Fidelity and Assembly Efficiency

Five-fragment lacI/lacZ cassette assembly was easily achievable with high levels of transformants and low backgrounds—so much so that there was little range for detectable improvements in the methodology. The test system was redesigned for 12 and 24 fragments. This redesign was guided by both advances in the re-engineering of the original BsaI-HF® type IIS restriction enzyme and the completion of T4 DNA ligase fidelity studies conducted at NEB.

While T4 DNA ligase, the mainstay of most biotechnological cloning efforts for over 50 years, prefers ligation of Watson-Crick base-pair substrates, it demonstrates significant activity on some mismatch-containing pairings. During Golden Gate assembly, ligation of mismatched pairs of overhangs can lead to incorrect assemblies, so care must be taken to minimize this possibility.

Recently, NEB researchers profiled the comprehensive fidelity of cohesive end ligation by this enzyme for all three- and four-base overhang sequences under standard reaction conditions. This dataset allows quantitation of sequence-dependent ligation efficiency and identification of mismatch-prone pairings. Using these four-base-pair overhang observations, NEB researchers designed and synthesized accurate “high-fidelity” junction sets for both the 12- and 24-fragment versions of the lacI/lacZ cassette.

In conjunction with BsaI-HFv2, re-engineered to provide improved Golden Gate performance, a series of optimization experiments for these more complex assemblies were performed. It was found that high efficiencies and accurate assembly levels were indeed possible, with correct, in-frame assembly proceeding in 99% of 12-fragment assemblies and in over 90% of 24-fragment assemblies (Figure 3, Table 1). The fidelity data can be applied to derive similar high-fidelity overhangs for any Golden Gate assembly design.

Figure 3. Golden Gate assembly of 24 fragments can be achieved with high efficiency and accuracy. While 30 cycles are sufficient to achieve 24-fragment assemblies, the stability of the BsaI-HFv2 and T4 DNA ligase allows continued assembly through 45 and 60 cycles with low background. (A) The efficiency of assembly versus cycle number. (B) The accuracy of assembly versus cycle number.

Table 1. Yields and fidelities for Golden Gate assemblies with BsaI-HFv2  and T4 DNA ligase.
The efficiency of assemblies per plate using outgrowth volumes described in Figure 3, with calculated yields from entire outgrowth built from 2 µL of the assembly reaction, and from the entire assembly reaction. All assembly protocols had a 5-min, 55°C terminal soak before transformation.

The Future of Assembly

DNA assembly methods are important tools for many areas of science, and researchers continue to test the limits of DNA assembly approaches with increasingly complex experimental conditions. The ability to construct more complex, multifragment assemblies, as shown in this work, will fuel additional efforts to push the technique forward.

Rebecca Kucera is an applications and product development scientist and Eric Cantor, Ph.D. ([email protected]), is a development group leader at New England Biolabs.
Previous articleDuchenne Muscular Dystrophy Mutation in Dogs Reversed by CRISPR-Cas9
Next articleSports Supplement Inhibits Growth of Drug-Resistant Breast Tumors