An efficient cloning method to expand vector and restriction site compatibility of Golden Gate Assembly

Summary

Golden Gate Assembly is an efficient and rapid cloning method but requires dedicated vectors. Here, we modified Golden Gate to expand its compatibility to a broader range of destination vectors while maintaining its strengths. Our Expanded Golden Gate (ExGG) assembly adds to the insert(s) type IIS restriction sites that generate protruding ends compatible with traditional type IIP sites on the recipient vector. The ligated product cannot be cleaved again, owing to a single-base change near the junction. This allows the reaction to proceed in a single tube without an intermediate purification step. ExGG can be used to introduce multiple fragments into a vector simultaneously, including shorter fragments (<100 bp) and fragments with shared sequences, which can be difficult to assemble with other fast cloning strategies. Thus, ExGG extends the convenience of Golden Gate to a much larger space of pre-existing vectors designed for conventional cloning.

Graphical abstract

Highlights

• •We expanded the compatibility of Golden Gate Assembly to a broader range of vectors
• •Expanded Golden Gate retains the efficiency and speed of Golden Gate Assembly
• •Expanded Golden Gate allows for digestion and ligation in a single reaction
• •Expanded Golden Gate can assemble multiple insert fragments

Motivation

Molecular cloning is a technique fundamental to many fields of biological research. Conventional cloning proceeds by separate digestions of vector and insert, followed by gel purification, ligation, and transformation. Golden Gate Assembly is a fast and efficient cloning method, wherein digestion and ligation of all fragments occur in a single reaction. However, Golden Gate Assembly requires dedicated destination vectors and cannot be used on the majority of existing plasmids. We have developed an Expanded Golden Gate (ExGG) strategy that retains the strengths of Golden Gate Assembly and extends them to a much larger range of destination vectors.

Sorida and Bonasio develop an Expanded Golden Gate (ExGG) strategy that allows rapid insertion of one or more inserts in a broader range of destination plasmids with high efficiency and accuracy. ExGG is performed in a single tube, making routine molecular cloning fast and easy.

Introduction

Molecular cloning is the process by which multiple DNA fragments are recombined together within plasmids that can be propagated in bacterial hosts, typically Escherichia coli.¹ Cloning is a fundamental experimental technique that facilitates a myriad of downstream molecular applications in a wide range of biological research fields, including expression of proteins and RNAs as well as the generation of recombinant DNA sequences for genomic editing. For decades, cloning has been performed mostly by cutting donor and recipient DNA molecules with type IIP (for “palindromic”) restriction endonucleases (REs) that cut within specific recognition sites, isolating the desired fragments based on their sizes, and then “stitching” them together with T4 DNA ligase. This strategy, which we refer to as “conventional” or “classic” cloning, usually leverages the complementarity of protruding sticky DNA ends generated by type IIP enzymes,¹ although in some cases single base overhangs (e.g., in TOPO-TA cloning²) or less desirable blunt ends can be used.

Although it has been in use for decades, conventional cloning is still a somewhat laborious process, requiring multiple experimental steps, including the isolation of intermediate DNA fragments, and it is prone to failure. Because of this, alternative methods have been developed over the years, including Gateway cloning³ and in vivo assembly in bacteria⁴ or yeast.^5,6 Two methods have gained popularity in recent years, as they provide streamlined protocols to assemble more than two fragments in a single reaction and are available commercially: Gibson Assembly⁷ and Golden Gate Assembly.^8,9

Golden Gate Assembly relies on the use of type IIS (for “shifted” cleavage) REs, such as BsaI, that recognize a non-palindromic sequence and cleave the DNA away from the recognition site, leaving ssDNA overhangs.^10,11,12,13 In Golden Gate, simultaneous RE digestion and ligation using T4 DNA ligase occur on insert and plasmid DNA. However, Golden Gate cloning can only be performed with destination vectors that have been specifically designed to contain one (or more) appropriately oriented pair of recognition sites for a type IIS RE. This is required to facilitate vector-insert ligations and to suppress self-ligation of the empty vector while the reaction proceeds in a single tube (“one-pot”), but it limits the applicability of Golden Gate Assembly, because most of the existing plasmids do not have such type IIS sites. In addition, only some type IIS REs, specifically BbsI, BsaI, BsmBI, Esp3I (isoschizomer of BsmBI), and SapI, are used in Golden Gate Assembly (see key resources table), further restricting vector availability. Although Golden Gate Assembly is a powerful cloning method, as long as certain conditions are fulfilled,^{14,15,16,17,18} it would be even more versatile and powerful if it could be expanded to a broader range of plasmids.

Here, we describe an Expanded Golden Gate (ExGG) method, which allows for Golden Gate-like efficient assembly of multiple DNA fragments into plasmids that have conventional multiple cloning sites (MCSs) for type IIP REs, but do not contain the type IIS sites that are required for conventional Golden Gate Assembly. In ExGG, a destination vector is digested by type IIP REs, while type IIS sites are introduced into one or more DNA inserts such that, upon digestion, they generate sticky ends compatible with those of the cut vector, but upon ligation, they destroy the original type IIP RE site, allowing for rapid, one-pot assembly of multiple DNA fragments. Our method retains the versatility and convenience of Golden Gate Assembly and expands its functionality to work with almost all common plasmid vectors.

Results

One-pot, one-step Expanded Golden Gate assembly

ExGG assembly requires PCR products and a plasmid; these are digested by different REs and ligated by T4 DNA ligase in a single tube (Figure 1A). No gel extraction of the digested plasmid or PCR product is required. As an example, we constructed a plasmid for bacterial expression of AsCas12f1, a miniature Cas protein that targets dsDNA.¹⁹ We utilized BsaI to digest the PCR product (AsCas12f1) and EcoRI and XhoI to digest the destination plasmid (pET-RB) (Figure 1B), two REs that are active in the reaction buffer provided with commercial T4 DNA ligase.²⁰ The PCR primers to amplify the insert fragment were designed to add a BsaI recognition site on each side and generate 5′ protruding ends (5PEs) compatible with those generated by EcoRI and XhoI. A key aspect of this primer design is that we introduced a nucleotide adjacent to the 5PE on each side that would prevent restoration of the EcoRI or XhoI recognition site after ligation into the vector. In this case, the EcoRI site GAATTC was changed to GAATTA, and the XhoI site CTCGAG was changed to GTCGAG (Figure 1A). We named this altered base “recut blocker.”

Figure 1.

One-pot, one-step Expanded Golden Gate assembly

(A) Scheme of one-step Expanded Golden Gate (ExGG). The primers were designed to introduce a BsaI recognition site that can generate an overhang compatible with EcoRI or XhoI but that cannot be re-digested after ligation because of the adjacent modified base (“recut blocker”). RE digestion and ligation were simultaneously conducted in a single tube. MCS, multiple cloning site; ROI, region of interest.

(B) Design to construct the pET-RB::AsCas12f1 plasmid. The insert fragment acquires 5PE of EcoRI and XhoI and recut blockers by BsaI digestion. BbsI, BsmBI, and SapI recognition sites reside in the vector plasmid.

(C) Points indicate the number of colonies obtained in the indicated reaction conditions. Minus insert is a negative control in which the insert fragment was replaced by water.

(D) Assembly validation. Colonies obtained by six cycles of 37°C–25°C reaction with Hi-T4 DNA ligase were tested by colony screening PCR, RE cut test, and Sanger sequencing. RE cut test was applied to clones that yielded the correct-size product by colony screening PCR; Sanger sequencing was performed on plasmid DNA from clones that showed the correct pattern by RE digestion, although all clones tested were positive.

After combining DNA molecules and enzyme in a single tube and incubating at 37°C for 1 h, we transformed suitable E. coli cells and obtained more than 5-fold greater numbers of colonies when the AsCas12f1 insert fragment was included as compared with the vector alone (Figure 1C, left). We also tested six cycles of alternating 37°C and 16°C incubations for 5 min each,^8,21 which resulted in efficiencies comparable to those obtained at 37°C (Figure 1C, left).

Next, we tested whether a thermostable ligase (Hi-T4 DNA ligase, key resources table), would perform better in this assay. Although the number of positive colonies did not increase, these reactions displayed slightly lower background in the vector-only controls, suggesting improved cloning efficiency (Figure 1C, right). To validate the integrity of the constructs obtained with ExGG, we performed PCR-based colony screening (45 plasmids), restriction mapping (9 plasmids), and Sanger sequencing (9 plasmids). In all cases, the plasmids obtained by ExGG were correctly constructed (Figure 1D).

Activity of type II restriction endonucleases in T4 DNA ligase reaction buffer

Similar to Golden Gate Assembly, the key advantage of ExGG over conventional cut and ligate cloning methods is that restriction digestion and ligation occur at the same time (one-step) and in the same tube (one-pot). This requires that the RE and ligase enzymes are active in the same buffer. As shown above, this was the case for EcoRI and XhoI. We sought to determine what other enzyme combinations might be suitable for ExGG. We tested the activity of 11 type IIP REs that generate four-base, 5′ protruding ends, as well as two type IIS REs in T4 DNA ligase reaction buffer (Figure S1A). Our results show that 9/11 of type IIP were fully active and 11/11 were at least partially active. As for type IIS REs, both tested enzymes gave satisfactory cleavage activity, although BbsI-HF resulted in incomplete digestion.

Two-step ExGG further expands compatibility of insert and vector sequences

Next, we sought to assemble a plasmid to express Un1Cas12f1, another miniature Cas protein.²² However, the Un1Cas12f1 protein-coding sequence contains BsaI and SapI recognition sites, while the destination vector, pET-RB, contains sites for BbsI, BsmBI, and SapI, so that no reliable type IIS RE could be used with these fragments in a one-step, one-pot reaction (Figure 2A). To overcome this problem, we adopted a one-pot, two-step reaction strategy (Figure 2B). In the first step (step 1), the insert fragment was digested at 37°C for 1 h with a type IIS RE (BbsI-HF), which was subsequently heat inactivated at 65°C for 20 min. Then, the temperature of the reaction was ramped down to 4°C at the rate of −0.1°C per second to rehybridize the strands of the insert fragment. In the second step (step 2), the vector was digested by type IIP REs (EcoRI-HF and NotI-HF) and ligated to the insert fragment by Hi-T4 DNA ligase with six cycles of 37°C–25°C incubation. After transformation, we obtained 134 and 139 colonies from two replicates when the Un1Cas12f1 insert fragment was included and zero colonies in reactions with the vector alone (Figure 2C). All the plasmids tested consisted of the desired construct as determined by PCR-based colony screening (30 out of 30), RE mapping (6 out of 6), and Sanger sequencing (6 out of 6) (Figure 2D). Although the two-step protocol is slightly longer than the one-step reaction, it is still simpler and faster than conventional cut and ligate cloning and, as we show here, results in efficient generation of the desired constructs.

Figure 2.

One-pot, two-step Expanded Golden Gate assembly

(A) Design to construct pET-RB::Un1Cas12f1 plasmid. The insert fragment acquires 5PE of EcoRI and NotI and recut blockers by BbsI digestion. BbsI, BsmBI, and SapI recognition sites reside in the vector plasmid, while BsaI and SapI recognition sites reside in the Un1Cas12f1 insert fragment.

(B) Overview of the two-step ExGG. In step 1, the insert fragment was digested by a type IIS RE, BbsI-HF. Then, the BbsI-HF was heat inactivated. In step 2, the digested insert fragment was mixed with the second reaction premix containing the vector, type IIP REs (EcoRI-HF and NotI-HF), and Hi-T4 DNA ligase. The vector was digested by these type IIP REs and ligated with the insert fragment.

(C) Points indicate the number of colonies obtained in the indicated reaction conditions. Minus insert is a negative control in which the insert fragment was replaced by water.

(D) Accuracy of assembly. Colonies obtained from insert-plus plates were tested by colony screening PCR, RE cut test, and Sanger sequencing.

ExGG with multiple insert fragments

Another desirable feature of conventional Golden Gate Assembly is that it allows for the simultaneous assembly of multiple fragments into a vector. To determine whether ExGG can also be employed with multiple fragments, we performed a one-pot, one-step ExGG assembly with three insert fragments simultaneously.

We sought to introduce one fragment encoding a twin FLAG-HA epitope tag, one encoding a monomeric form of streptavidin (mSA2), and one encoding AsCas12f1, into pET-RB as a destination vector (Figure 3A). We designed primers for the amplification of these insert fragments to introduce compatible 5PE after BsaI digestion (Figures 3A and S1B). Using the same protocol as above, consisting of six cycles of 37°C for 5 min and 25°C for 5 min in presence of Hi-T4 DNA ligase, we obtained 95, 85, and 83 colonies in three replicate reactions containing inserts and virtually no background when the inserts were omitted (Figure 3B, left); however, only 15% (7 out of 46) of the clones tested contained all three desired inserts, as determined by colony screening PCR (Figure 3C). Note that we obtained three replicates to test whether the low accuracy was reproducible, because ExGG had performed well so far.

Figure 3.

Expanded Golden Gate assembly with multiple inserts

(A) Design to construct the pET-RB::FH-mSA2-AsCas12f1 plasmid by one-step ExGG. The insert fragments were digested by BsaI, while the vector was digested by EcoRI and XhoI.

(B and E) Points indicate the number of colonies obtained in the indicated reaction conditions. In (E) the number of insert fragments is indicated.

(C and F) Assembly validation. Colonies obtained from insert-plus plates were tested by colony screening PCR, RE cut test, and Sanger sequencing.

(D) Design to construct the pET-RB::FH-mSA2-Un1Cas12f1 plasmid by two-step ExGG. The insert fragments were digested by BbsI, while the vector was digested by EcoRI and NotI.

Given that increased numbers of cycles can improve efficiency and accuracy of traditional Golden Gate Assembly (key resources table), we repeated the reaction, increasing the number of cycles to 30. We obtained 96 and 97 colonies in two replicates and again virtually no background when inserts were omitted (Figure 3B, right). However, this time 100% (30 out of 30) contained the three inserts as determined by PCR, and all tested plasmids (12 out of 12) were also validated by RE mapping and Sanger sequencing (Figure 3C). These results indicate that ligation of multiple insert fragments can be achieved with ExGG one-pot, one-step reactions, but it requires extended cycling to maximize accuracy.

Then, we sought to assemble the FLAG-HA, mSA2, and Un1Cas12f1 fragments and the pET-RB plasmid by one-pot, two-step reaction (Figure 3D). We obtained few colonies, and extending the number of cycles to 30 in this case was not successful (Figure 3E, top). Because BbsI-HF was not fully active in T4 DNA ligase reaction buffer (see Figure S1A), we assumed that incomplete digestion of even just one of the three insert fragments may be responsible for the failed reactions. We first reduced the number of the insert fragments from three to two and found that these could be successfully assembled (Figure 3E, top) and validated (Figure 3F).

Given that BbsI-HF is assumed to perform better in the rCutSmart buffer than in the T4 DNA ligase buffer, we conducted the entirety of reactions in rCutSmart. We obtained colonies but not in excess of those found in the vector-only control (Figure 3E, bottom); however, the correct plasmid was still obtained 82% of the times (23 out of 28) with six cycles and 100% of the times (12 out of 12) with 30 cycles (Figure 3F). These results indicate that enhanced RE digestion improves efficiency of assembly and that the one-pot, two-step approach is applicable to ligation of three insert fragments by using the rCutSmart buffer.

Comparison of ExGG with Golden Gate and Gibson Assembly

We compared the efficiency of ExGG with two popular fast cloning methods: Golden Gate and Gibson. We designed an insert that could be cloned in the same vector plasmid with all three methods and utilized the same molar amounts in all reactions. The plasmid vector, pUC19-GG, contains a Golden Gate-compatible cloning site as well as an MCS for type IIP REs (Figure 4A). We amplified an insert fragment, PB-ITRs, by PCR introducing 20-bp homology arms for Gibson Assembly, BsaI recognition sites for Golden Gate, and BbsI recognition sites for one-step ExGG (Figure S2). For ExGG, the plasmid was digested by SalI-HF and HindIII-HF, while the insert fragment was digested by BbsI-HF; these were ligated by Hi-T4 DNA ligase simultaneously for 1 h (6 cycles of 37°C for 5 min and 25°C for 5 min) in the rCutSmart buffer (Figure 4B). For Golden Gate, the plasmid and insert fragment were simultaneously digested by BsaI-HFv2 and ligated by T4 DNA ligase for 1 h (6 cycles of 37°C for 5 min and 16°C for 5 min) in T4 DNA ligase reaction buffer. For Gibson, the plasmid and insert fragment were subjected to SalI-HF and HindIII-HF digestion for 1 h at 37°C in the rCutSmart buffer and assembled by Gibson reaction over another 1 h at 50°C.

Figure 4.

Comparison of ExGG, Golden Gate, and Gibson Assembly

(A) Design to construct the pUC19:PB-ITRs plasmid by one-step ExGG, Golden Gate, and Gibson Assembly using the same DNA molecules. The insert fragment has ExGG-, Golden Gate-, and Gibson-compatible ends. The destination plasmid has a Golden Gate-compatible BsaI cloning site and SalI and HindIII recognition sites.

(B) Scheme of the comparison experiment. The same amounts of insert fragment and vector plasmid were applied to each reaction.

(C) Points indicate the number of colonies obtained by the indicated assembly methods.

(D) Assembly validation. Colonies obtained from insert-plus plates were tested by colony screening PCR, RE cut test, and Sanger sequencing. GG, Golden Gate; GA, Gibson Assembly.

All three methods yielded at least 20 times greater numbers of colonies when the insert was included compared to the negative control (Figure 4C). PCR-based colony screening, RE mapping, and Sanger sequencing showed high accuracy of all three methods (Figure 4D). These results indicate that the efficiency of ExGG is comparable to Golden Gate and approaching that of Gibson Assembly in a one-insert-fragment assembly. However, the accuracy was comparable across the three methods.

Discussion

In this study, we extended the utility of Golden Gate Assembly to a much larger set of existing destination plasmids lacking recognition sites for type IIS REs. Given that a myriad of plasmids have been generated over decades to be utilized with conventional cloning strategies, based on type IIP REs, our technique enables accelerated and simplified cloning throughout the broad range of plasmids generated in the history of molecular biology. The most convenient application of ExGG is a simple one-pot, one-step reaction, whereby RE digestion and DNA ligation occur simultaneously (Figure 1). We also addressed the cases in which no type IIS RE can be employed in the one-step reaction and developed a substitute two-step reaction system (Figure 2). One-step and two-step ExGG require 1 h and 2.5 h of incubation, respectively, without purification of RE-digested fragments (Figure 5A). We found that one-step and two-step ExGG allow for assembly of at least three insert fragments; accuracy of one-pot ExGG and efficiency of two-pot ExGG were increased by longer incubation and by use of the rCutSmart buffer, respectively. Because molecular cloning is a necessary tool for most fields of biological research, we believe that ExGG will provide an option to facilitate and accelerate the work of many researchers.

Figure 5.

Timeline and comparison

(A) Reaction time of cloning from after purification of PCR products to before transformation. Gel extraction is indicated by dark gray. ExGG requires incubation once the reaction mix is assembled.

(B) Advantages and disadvantages of ExGG and major cloning methods.

Strengths of ExGG

Given that ExGG is based on Golden Gate Assembly, they share the advantage of ligating one or more inserts into a destination vector in a single, fast reaction, without purification steps. In addition, because of its design, ExGG can be used on a much broader range of destination vectors: while Golden Gate Assembly can only use destination vectors having a pair of the appropriately oriented recognition sites of a type IIS RE, ExGG uses destination vectors with two type IIP REs recognition sites, which are found in virtually all available cloning vectors, typically conveniently located within an MCS. For these vectors, conventional cut and ligate cloning remains a valid option, albeit one that is more time-consuming and less streamlined than ExGG, because the digested vectors and insert fragments must be separated from type IIP REs by agarose gel extraction. Therefore, ExGG combines advantages from conventional cloning and Golden Gate Assembly.

Comparison of ExGG and Gibson Assembly

Gibson Assembly is a widely used alternative to conventional cloning and Golden Gate, but it too has limitations (Figure 5B). It is generally not recommended to use with inserts smaller than 100–200 bp, possibly because of the risk that the exonuclease might digest the whole insert (key resources table). Moreover, the homology arms that guide assembly must be unique among the ends of the DNA fragments. For example, when multiple nuclear localization signals (NLSs), flexible linkers, or 2A peptides are part of the homology regions, the assembly efficiency and accuracy might decrease. Finally, inserts that form stable secondary structures or contain repetitive sequences can reduce the assembly efficiency.²³ Conversely, Golden Gate can assemble insert fragments as small as 24–25 nt, as for example with hybridized DNA oligos for CRISPR targeting,²⁴ as well as repetitive insert fragments including poly-glutamine stretches.²⁵

Given that ExGG is based on Golden Gate, it should similarly be capable to assemble these type of DNA fragments that are not a good substrate for Gibson Assembly. Indeed, the FLAG-HA fragment used in our multi-insert assembly was 91 bp in length after BsaI digestion, and the SV40 NLSs, which reside on one of the ends of three insert fragments, share the same or highly similar DNA sequences (Figure S1B), and yet they were successfully assembled. Furthermore, because Gibson Assembly uses a 5′ exonuclease to digest DNA fragments followed by a DNA polymerase to fill the gap after annealing, it could theoretically introduce mutations in the destination vector in the vicinity of the inserted fragment, which might go undetected and cause problems downstream. Conversely, and similar to conventional cloning and Golden Gate, ExGG cannot introduce alteration in the destination plasmid, so that sequencing-based validation can be limited to the inserted fragment(s). Finally, Gibson Assembly, depending on the particular cloning strategy, might need separation of REs used for linearization of the destination plasmid mostly by gel extraction under UV, whereas ExGG does not require gel separation and extraction.

Comparison with additional cloning strategies: Pyrite and ASAP

Although less widespread, additional one-pot cloning strategies have been developed. Pyrite cloning offers an improvement over conventional cloning, whereby after 2 h of digestion by type IIP REs, 10 cycles of ligation and heat inactivation of REs were conducted in a single tube.²⁰ Mai et al. reported a similar cloning method executing RE digestion, heat inactivation of REs, and ligation in series.²⁶ However, these methods theoretically cannot prevent self-ligation and require an overnight reaction program.

Essentially, ExGG leverages the idea of isocaudamers, restriction enzymes with different recognition sites but compatible protruding ends, and extends it by pairing a type IIS with a type IIP enzyme. Previously, Zuckermann et al. established a Golden Gate-based method called ASAP-cloning¹⁶ by utilizing an isocaudomer pair of XbaI (T|CTAGA) and NheI (G|CTAGC), whereby the backbone vector was linearized by XbaI, while insert fragments were digested by BbsI to make up the NheI-digested end in the head and tail of fragments. By adding NheI to the reaction, self-ligation among insert fragments was prevented, while the fragments ligated into the XbaI-digested destination vector escaped re-digestion by XbaI. However, ASAP-cloning requires an isocaudomer pair; there are only a few options such as BamHI-BglII, SalI-XhoI, and XbaI-SpeI. Another cloning strategy that leverages isocaudamers is BioBrick, where the XbaI-SpeI pair is exploited in a modular, idempotent cloning strategy.²⁷

Assembly of multiple inserts by ExGG

Although one-step ExGG yielded sufficient numbers of colonies in three-insert fragment ligation in 1 h, its accuracy was low (Figure 3C). According to colony screening PCR, the band size of the negative clones was similar to that observed in the negative control, i.e., the empty pET-RB destination vector (Figure S3). This suggests that the destination vector was self-ligated or not fully digested. However, we rarely saw colonies in a no-insert control, suggesting that activity of type IIP REs (EcoRI and XhoI) toward the vector decreased in the presence of insert fragments. This limitation was overcome by extending the incubation time.

We were also able to assemble three insert fragments simultaneously in a two-step reaction, albeit with some modifications. Specifically, we only obtained sufficient numbers of colonies containing the desired construct when we replaced the ligation buffer with rCutSmart buffer, suggesting that the bottleneck was a limited activity of BbsI-HF in the former (Figures 3E and 3F). We suspect one-side-digested insert fragments would be irreversibly ligated to fully digested fragments or vectors, generating “dead-end” molecules. Other potential solutions besides use of rCutSmart are a reduction in the concentration of DNA fragments to a half or quarter, doubling BbsI-HF, or a use of 2.5 times more concentrated BbsI-HF. Because enzymatic cleavage and ligation are opposing reactions, it is difficult to know for certain the mechanisms by which self-ligation products and desired products are generated.

Additional directions for future improvement

Some additional rounds of optimization might further extend the utility or efficiency of ExGG. For example, we did not attempt to shorten incubation time or decrease the number of 37°C–25°C (or 37°C–16°C) cycles in either one-step or two-step reactions. We also did not extend the final digestion at 60°C for 5 min, but there is a report showing that an additional incubation for 10 min lowers background colonies.²¹

We only attempted to assemble one to three insert fragments. Ligation of four or more insert fragments is theoretically feasible but may require a methodological improvement. We suspect that one of the potential obstacles is self-ligation of the insert fragment with a palindromic protruding end (e.g., AATT, 5PE of EcoRI-digested dsDNA), which is inevitable for ExGG because the self-ligated product will not be re-digested by an RE (EcoRI in this case) due to the recut blocker (Figure S4). Considering such a “dead” molecule, an increase in concentration of the insert fragments that results in palindromic protruding end after type IIS RE digestion might improve efficiency of assembly.

We did not use BsmBI, Esp3I, or SapI. For BsmBI, since its preferential reaction temperature is 55°C, this temperature could be applied to step 1 of the two-step reaction. 42°C–25°C cycles could be used for one-step ExGG (key resources table).

We did not attempt to assemble unpurified PCR products. Theoretically, purification of PCR products can be skipped, although ideally primers and potential primer dimers should be removed.

We applied ramp-down of temperature at the rate of −0.1°C/s after 65°C heat inactivation of BbsI-HF in step 1 of the two-step reaction to let possibly melted double-stranded DNAs rehybridized (see method details). We did not apply a fast ramp-down, such as −6°C/s. The necessity of this slow ramp-down is unknown and may depend on the length of insert fragments. Melting temperature (Tm) of fragments may be a criterion to determine the speed of ramp-down.

In the case of the lack of suitable type IIP RE restriction sites on the vector, inverse PCR, which allows for introduction of a suitable type IIP RE site on each end of the desired backbone, followed by DpnI digestion, would allow for cloning by ExGG, although in this case a similar approach would also allow cloning by other fast methods such as Golden Gate and Gibson Assembly. However, inverse PCR is undesirable as it carries the potential to introduce mutations in the backbone, requiring whole-plasmid sequencing.

A popular use of Golden Gate is in the context of multi-level assemblies, as, for example, in the Modular Cloning (MoClo) strategy.^14,15 While we did not attempt to develop a multi-level version of ExGG, it would be applicable to MoClo-like and other new strategies. In addition, the existing multi-level assemblies could incorporate an ExGG step. For these purposes, the four-base overhangs of the various fragments could be selected based on their ligation efficiency and fidelity^17,18 to maximize chances of success in multi-fragment and multi-level assembly.

Limitations of the study

We note two potential limitations of ExGG. First, at least two bases have to be altered to prevent re-digestion, one on each side of the insert, to block re-digestion after ligation. However, given that these recut blocker bases reside outside the inserted DNA, the inserted functional region remains intact. Second, ExGG requires type IIP REs generating four-base 5′ protruding ends but cannot be used with enzymes that generate 3′ protruding ends (e.g., KpnI and PstI) or different lengths of 5′ protruding ends (e.g., NdeI), because all reliable type IIS REs generate exclusively four-base 5′ protruding ends. This is however a minor limitation since there are several type IIP REs that generate four-base 5′ protruding ends, and these are very commonly found in most available vectors. We also note that the restriction sites of the type IIP REs utilized to digest the vector must be absent in the insert fragments.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and virus strains
XL10-Gold	Agilent	200314
Chemicals, peptides, and recombinant proteins
KAPA HiFi HotStart ReadyMix	Kapa Biosystems	KK2602
T4 DNA Ligase Reaction Buffer	NEB	M0202L
rCutSmart Buffer	NEB	B6004S
GeneRuler 1 kb Plus DNA Ladder	Thermo Scientific	SM1331
BsaI-HFv2	NEB	R3733S
BbsI-HF	NEB	R3539S
AgeI-HF	NEB	R3552S
BamHI-HF	NEB	R3136S
EcoRI-HF	NEB	R3101L
HindIII-HF	NEB	R3104S
NcoI-HF	NEB	R3193S
NotI-HF	NEB	R3189S
SalI-HF	NEB	R3138S
SpeI-HF	NEB	R3133S
XbaI	NEB	R0145S
XhoI	NEB	R0146S
XmaI	NEB	R0180S
KpnI-HF	NEB	R3142S
NdeI	NEB	R0111S
Smal	NEB	R0141S
BbsI-HF	NEB	R3539M
T4 DNA Ligase	NEB	M0202L
Hi-T4 DNA Ligase	NEB	M2622S
100 mM ATP	NEB	P0756S
NEBuilder HiFi DNA Assembly Master Mix	NEB	E2621
LB Broth	Fisher Scientific	BP9723
Agar	Fisher Scientific	BP1423
OneTaq DNA Polymerase	NEB	M0480L
5X OneTaq Standard Reaction Buffer	NEB	M0480L
SYBR Safe DNA Gel Stain	Thermo Scientific	S33102
Glycerol	Fisher	BP229
Bromophenol blue	Sigma-Aldrich	114391
Tris	Sigma-Aldrich	T1503
EDTA 2Na ⋅ 2H2O	Sigma-Aldrich	E5134
Type IIS Restriction Enzymes	NEB	https://www.neb.com/tools-and-resources/selection-charts/type-iis-restriction-enzymes
Critical commercial assays
MinElute PCR Purification Kit	Qiagen	28006
NucleoBond Xtra Midi Plus	Macherey-Nagel	740412.50
NEBridge® Golden Gate Assembly Kit (BsaI-HF® v2)	NEB	E1601L
NEBridge® Golden Gate Assembly Kit (BsmBI-v2)	NEB	E1602L
Oligonucleotides
Primers used in this study	This paper	See Table S1
Recombinant DNA
pCMV-AsCas12f1	Addgene	RRID:Addgene_171614
Cas12f-GE ver4.1	Addgene	RRID:Addgene_176544
pAc5.1B-EGFP	Addgene	RRID:Addgene_21181
pET-RB	This study	N/A
pUC19-GG	This study	N/A
Software and algorithms
Benchling	Benchling	https://www.benchling.com
Other
NanoDrop One C	ThermoFisher	ND-ONEC-W
Gibson Assembly Cloning Guide 2nd Edition	SGI	https://www.biocat.com/bc/files/Gibson_Guide_V2_101417_web_version_8.5_x_11_FINAL.pdf
Gibson Assembly Cloning Kit Instruction Manual	NEB	https://www.med.unc.edu/pharm/sondeklab/wp-content/uploads/sites/868/2019/10/gibson-cloning.pdf
In-Fusion HD Cloning Kit User Manual	Takara	https://www.takarabio.com/documents/User%20Manual/In/In-Fusion%20HD%20Cloning%20Kit%20User%20Manual_102518.pdf
Gibson Assembly Cloning	Addgene	https://www.addgene.org/protocols/gibson-assembly/
Nucleic Acid Data	NEB	https://www.neb.com/tools-and-resources/usage-guidelines/nucleic-acid-data
Cleavage Close to the End of DNA Fragments	NEB	https://international.neb.com/tools-and-resources/usage-guidelines/cleavage-close-to-the-end-of-dna-fragments

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Roberto Bonasio (roberto@bonasiolab.org).

Materials availability

Plasmids used and generated in this study can be provided upon request to the lead contact.

Experimental model and study participant details

Bacterial culture

The bacterium XL10-Gold was cultured on LB plates or LB liquid medium at 37°C.

Method details

PCR amplification and purification of insert fragments

The insert DNA of AsCas12f1 and Un1Cas12f1 for one insert fragment ligation was amplified from pCMV-AsCas12f1¹⁹ and Cas12f-GE ver4.1,²² respectively, with KAPA HiFi HotStart ReadyMix using the 2-step program composed of 95°C 3 min, 30x (98°C 20 sec/72°C 50 sec), 72°C 1 min, 4°C constant. The PB-ITRs fragment was amplified from our plasmid containing the PB-ITRs sequence (574 bp) by the same procedure. The primers used are listed in Table S1. Note that AsCas12f1 and Un1Cas12f1 products were also able to be obtained by a 3-step program with annealing at 60°C for 15 second with similar quality and quantity to the 2-step program. For three inserts ligation, the fragments of FLAG-HA and mSA2 were amplified from a gBlocks (IDT) DNA fragment containing the FLAG-HA or mSA2 sequence; AsCas12f1 and Un1Cas12f1 DNA fragments were amplified from the same template plasmids, using the same material and program as described above. PCR was performed in a 25 μl scale and 2 μl of the reaction was run on 1% agarose gel to check the band. The rest of DNA was purified by MinElute PCR Purification Kit, eluted with 30 μl of 1x TE buffer (10 mM Tris-HCl pH 8.0 at 25°C, 1 mM EDTA-NaOH pH 8.0 at 25°C), and measured for concentration by NanoDrop One C. We assume contaminant products such as primer dimers would reduce the accuracy of assembly; thus, we recommend obtaining a single band by increasing annealing temperature of PCR or by gel extraction. We recommend SYBR Safe DNA Gel Stain and a blue light for gel excision rather than a UV light to reduce a risk of DNA damage. The gBlocks template sequences are shown below.

gBlocks PCR template
Name	Top strand sequence (5′ to 3′)
NLS-FLAG-HA	cccaagaagaagaggaaagtcggcagcggagactacaaggatgacgacgataagggcagcggatacccatacgacgtgcctgactacgcc
mSA2	gcggaagcgggtatcaccggcacgtggtacaaccagcatggttctaccttcaccgttaccgcgggtgcggacggtaacctgaccggtcagtacgaaaaccgtgcgcagggcactggttgccagaactctccgtacaccctgaccggtcgttacaacggtaccaaactggaatggcgtgttgaatggaacaactctaccgaaaactgccactctcgtaccgaatggcgtggtcagtaccagggtggtgcggaagcgcgtatcaacacccagtggaacctgacctacgaaggtggttctggtccggcgaccgaacagggtcaggacaccttcaccaaagttaaa

One-step ExGG

To construct the AsCas12f1 expression plasmid, we assembled a reaction by mixing 2 μl T4 DNA Ligase Reaction Buffer, 0.2 μl BsaI-HFv2, 0.2 μl EcoRI-HF, 0.2 μl XhoI, 0.4 μl T4 DNA Ligase, 0.05 pmols of circular vector (pET-RB), 0.1 pmols of insert (AsCas12f1), and MilliQ water bringing up to 20 μl. For multiple inserts, we added 0.1 pmols for each insert. We generally assembled a master mix of reactions without insert(s) on ice and then add inserts or MilliQ water for + insert and – insert, respectively. Note that the master mix can be assembled at room temperature; we assembled it on ice to compare multiple conditions as precisely as possible. We aliquot T4 DNA Ligase Reaction Buffer and store at –20°C to avoid freeze-thaw cycles that might degrade ATP. The mixture was incubated at 37°C for 1 h or 6 cycles of (37°C 5 min/16°C 5 min) followed by incubation at 60°C for 5 min and kept at 4°C. For Hi-T4 DNA Ligase, 6–30 cycles of (37°C 5 min/25°C 5 min) followed by incubation at 60°C for 5 min and then kept constant at 4°C. The reaction can be stored at –20°C. For transformation, 2 μl of the reaction was applied. Required volume of vector and insert was calculated by this formula: (required volume in μl) = [(required pmols) x (base pairs x 650 g/mol)] / [1,000 x (concentration in ng/μl)] (Key Resources Table). For vectors, the required amount is 0.05 pmols; for insert fragments, 0.1 pmols. The recipe and incubation program of one-step ExGG are summarized below.

One-step ExGG
Reaction
10x T4 DNA Ligase Reaction Buffer∗	2 μL
Type IIS RE	0.2 μL
Type IIP RE 1 (5PE)	0.2 μL
Type IIP RE 2 (5PE)	0.2 μL
Hi-T4 DNA Ligase	0.4 μL
Vector	0.05 pmol
Insert fragment(s)	Each 0.1 pmol
Nuclease-free water	Bring up to 20 μL
Incubation
37°C 5 min 25°C 5 min	6–30 cycles
60°C 5 min	1 cycle
4°C ∞	1 cycle

∗Buffer can be replaced by 10x rCutSmart Buffer; add 2 μL 10 mM ATP.

Two-step ExGG

Step 1: to construct the Un1Cas12f1 expression plasmid, we assembled a first reaction by mixing 1 μL T4 DNA Ligase Reaction Buffer, 0.5 μL BbsI-HF, 0.1 pmol of insert fragment (Un1Cas12f1), and MilliQ water bringing up to 10 μL. For multiple inserts, we added 0.1 pmol for each insert fragment. The mixture was incubated at 37°C for 1 h, at 65°C for 20 min, and then kept at 4°C. The reaction can be stored at −20°C. Note that the heat inactivation temperature varies according to the Type IIS enzyme.

Step 2: we assembled a second reaction premix by mixing 1 μL T4 DNA Ligase Reaction Buffer, 0.2 μL EcoRI-HF, 0.2 μL NotI-HF, 0.4 μL Hi-T4 DNA Ligase, 0.05 pmol of circular vector (pET-RB), 1 μL of 10 mM ATP made from 100 mM ATP and MilliQ water bringing up to 10 μL, and then added the second reaction premix to the first reaction. The mixture was incubated at 37°C for 1 h or 6–30 cycles of (37°C 5 min/25°C 5 min), 60°C for 5 min, and kept at 4°C. The reaction can be stored at −20°C. For transformation, 2 μL of the reaction was applied.

In some reactions, T4 DNA Ligase Reaction Buffer was replaced by rCutSmart Buffer in steps 1 and 2 and 2 μL of 10 mM ATP was added in step 2. The recipe and incubation program of two-step ExGG are summarized below.

Two-step ExGG
Step 1	Reaction
	10x T4 DNA Ligase Reaction Buffer∗	1 μL
	Type IIS RE	0.2 μL
	Insert fragment(s)	Each 0.1 pmol
	Nuclease-free water	Bring up to 10 μL
	Incubation
	37°C 1 h	1 cycle
	65°C or 80°C 20 min	1 cycle
	−0.1°C/sec to 4°C	1 cycle
	4°C ∞	1 cycle
Step 2	Premix
	10x T4 DNA Ligase Reaction Buffer∗	1 μL
	Type IIP RE 1 (5PE)	0.2 μL
	Type IIP RE 2 (5PE)	0.2 μL
	Hi-T4 DNA Ligase	0.4 μL
	10 mM ATP	1 μL∗
	Vector	0.05 pmol
	Nuclease-free water	Bring up to 10 μL
	Incubation
	37°C 5 min 25°C 5 min	6 cycles
	60°C 5 min	1 cycle
	4°C ∞	1 cycle

^∗Buffer can be replaced by 10x rCutSmart Buffer; add 2 μL 10 mM ATP in Step 2.

ExGG vs. Golden Gate vs. Gibson Assembly

We made a DNA mix containing 0.025 μM pUC19-GG vector plasmid and 0.05 μM PB-ITRs insert fragment and added 2 μl to each reaction, providing 0.05 pmol and 0.1 pmol, respectively. For negative control, the insert fragment was replaced by MilliQ water. One-step ExGG reaction was assembled in 20 μl, including 2 μl 10x rCutSmart Buffer, 2 μl 10 mM ATP, 0.2 μl BbsI-HF, 0.2 μl SalI-HF, 0.2 μl HindIII-HF, and 0.4 μl Hi-T4 DNA Ligase, incubated at 6 cycles of (37°C 5 min/25°C 5 min), 60°C for 5 minutes, and kept at 4°C. Golden Gate reaction was assembled in 20 μl, including 2 μl 10x T4 DNA Ligase Reaction Buffer, 0.6 μl BsaI-HFv2, and 0.4 μl T4 DNA Ligase and incubated at 6 cycles of (37°C 5 min/16°C 5 min), 60°C for 5 minutes, and kept at 4°C. For Gibson Assembly, the reaction was first assembled in 10 μl, with 1 μl 10x rCutSmart Buffer, 0.2 μl SalI-HF, and 0.2 μl HindIII-HF, incubated at 37°C for 1 hour, and kept at 4°C; then the reaction was mixed with 10 μl 2x NEBuilder HiFi DNA Assembly Master Mix, incubated at 50°C for 1 hour, and kept at 4°C. For transformation, 2 μl of the reaction was applied.

Primer and plasmid design

We used Benchling to design primers and edit DNA. Primers were purchased from IDT (USA) and GENEWIZ (Azenta Life Sciences) (USA) (Table S1). Since BsaI-HFv2 and BbsI-HF requires at least one nucleotide from the end followed by their recognition sequence to perform 50–100% activity in a suitable buffer (Key Resources Table), we added two T’s at the 5’ end of the primers. In general, inclusion of additional bases might be beneficial.

Restriction endonuclease activity test

We used the pAc5.1B-EGFP plasmid as substrate DNA. We transformed the bacterial strain XL10-Gold with pAc5.1B-EGFP and recovered by midi-prep using NucleoBond Xtra Midi Plus following the manufacturer’s instruction. The concentration of pAc5.1B-EGFP (6089 bp) was 2176.4 ng/μl, measured by NanoDrop One C. The reaction was assembled by mixing 2 μl T4 DNA Ligase Reaction Buffer, 0.1 pmols (0.18 μl) pAc5.1B-EGFP, 0.5 μl each RE, and MilliQ water to bring up to 20 μl. For undigested control, RE was replaced by MilliQ water. The reaction was incubated at 37°C for 1 hour, 5 μl of which was mixed with 6x loading dye (10 mM EDTA-NaOH pH 8.0, 40% Glycerol, 0.003% bromophenol blue), loaded onto 1% agarose gel with 0.5 μg/ml ethidium bromide, and separated in the TBE buffer (0.089 M Tris, 0.089 M boric acid, 2 mM EDTA) at 150V. We loaded 0.25 μg GeneRuler 1 kb Plus DNA Ladder. DNA was visualized and imaged under a ultra violet light. Regarding REs, the reaction contained 10 units (U) BsaI-HFv2, 10 U BbsI-HF, 10 U AgeI-HF, 10 U BamHI-HF, 10 U EcoRI-HF, 10 U HindIII-HF, 10 U NcoI-HF, 10 U NotI-HF, 10 U SalI-HF, 10 U SpeI-HF, 10 U XbaI, 10 U XhoI, or 5 U XmaI.

Colony screening PCR

We picked single bacterial colonies, inoculated a replica plate, suspended in 50 μl MilliQ water, and incubated at 94°C for 10 min. The replica plate, which was made by LB with 1% agar and 50 μg/ml kanamycin or 100 μg/ml ampicillin, was incubated at 37°C for 16–18 hours and stored at 4°C. For pET-RB-based plasmids, the PCR mix was assembled with 5 μl suspended bacteria, 4 μl 5X OneTaq Standard Reaction Buffer, each 0.2 mM dNTP, 0.2 μM two forward primers (F1 and F2), 0.2 μM reverse primer (R), 0.1 μl OneTaq DNA Polymerase, and MilliQ water to bring up to 20 μl. The reaction was incubated at 94°C for 2 min, 35 cycles of 94°C 30 sec, 55°C 30 sec, and 68°C 30 sec, 68°C for 5 min, and 4°C infinite. For pUC19-GG-based plasmids, one forward and one reverse primers were added and the polymerase extension time was prolonged to 50 sec. 5 μl of the reaction was mixed with 6x loading dye. The methods to run gel and detect DNA are described in Restriction endonuclease activity test. The primers for pET-RB consist of pET-F, pET-R, and As-F specific to AsCas12f1 or Un-F specific to Un1Cas12f1, in which the pET-F and pET-R primers attach pET-RB flanking its MCS, while the As-F or Un-F primer attaches AsCas12f1 or Un1Cas12f1, respectively (Figure S3). The expected size from correctly assembled pET-RB::AsCas12f1 or FH-mSA2-AsCas12f1 and pET-RB::Un1Cas12f1 or FH-mSA2-Un1Cas12f1 plasmids was 572 bp and 596 bp, respectively, while that from the intact pET-RB was 251 bp. The primers for pUC19-GG consist of pUC19-F and pUC19-R, which attach pUC19-GG backbone. The expected size from correctly assembled pUC19::PB-ITRs-ExGG, pUC19::PB-ITRs-GG, and pUC19::PB-ITRs-GA was 794, 810, and 832 bp, respectively, while that from the intact pUC19-GG was 232 bp.

Diagnostic restriction digestion

The pET-RB plasmid into which AsCas12f1 or FH-mSA2-AsCas12f1 and Un1Cas12f1 or FH-mSA2-Un1Cas12f1 was inserted was recovered from bacteria by mini-prep and digested with KpnI-HF and BamHI-HF, respectively. The pUC19::PB-ITRs plasmids were digested with NdeI and Smal. The reaction included 1 μl 10x rCutSmart Buffer, 0.5–1 μg of the recovered plasmid, 0.5 μl (10 units) of KpnI-HF, 0.5 μl (10 units) of BamHI-HF, or each 0.25 μl (5 units) of NdeI and Smal and MilliQ water to bring up to 10 μl, and was incubated at 37°C for 1 hour. 5 μl of the reaction was mixed with 6x loading dye. The methods to run gel and detect DNA are described in Restriction endonuclease activity test. The expected size of pET-RB::AsCas12f1 and pET-RB::FH-mSA2-AsCas12f1 yielded by KpnI-HF digestion was 5971 + 649 and 5971 + 820 + 282 bp, although the 282 bp band was hardly visible. The expected size of pET-RB::Un1Cas12f1 and pET-RB::FH-mSA2-Un1Cas12f1 yielded by BamHI-HF digestion was 5964 + 1016 and 5964 + 1469 bp. We digested pET-RB by KpnI-HF or BamHI-HF for comparison, which is expected to yield a 5306 bp band. The expected size of pUC19::PB-ITRs-ExGG, pUC19::PB-ITRs-GG, and pUC19::PB-ITRs-GA yielded by NdeI and Smal digestion was 2671 + 590, 2717 + 560, and 2690 + 609 bp, respectively.

Sanger sequencing

The plasmids recovered by mini-prep were sent to GENEWIZ (Azenta Life Sciences) (USA), following their instruction. We sequenced the ligation junctions but did not entirely sequence the inserted DNA. Alignment analysis was performed by Benchling. We used the primers pET-F and pET-R or pUC19-F and pUC19-R for sequencing. For some plasmids, F1-ori-R was additionally used.

Replicates

For one-step and two-step reactions, we obtained 2 or 3 replicates among which we used the same purified insert fragments and plasmid but assembled separate reaction master mixes. We used the same master mix between + and – insert. The batch of competent cells was not necessarily same among replicates but was same between + and – insert. The number of colonies were counted by eyes three times and the median was taken. For colony screening PCR, we picked approximately 30 clones (15 clones from each replicate) to test the accuracy. For RE cut test, we basically chose 6 clones (3 clones from each replicate) with PCR screening positive. For Sanger sequencing, we sequenced the plasmids with RE cut test positive, although we did not find any plasmid that tested negative in RE cut test.

Quantification and statistical analysis

This study does not contain statistical analyses.

Published: August 22, 2023

Data and code availability

• •The data reported in this paper will be shared by the lead contact upon request.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.