Golden Gate assembly of BioBrick-compliant parts using Type II restriction endonucleases
Abstract
Aims: New methods of DNA recombination that capture the principal advantages of the BioBrick standard (ease of design) and Golden Gate assembly (decreased labor) are demonstrated here. Methods & materials: Both methods employ DNA methyltransferase expression vectors, available from Addgene, that protect selected sites on different plasmids from particular Type II restriction endonucleases. No other reagents are required. Results: The 4R/2M discontinuous DNA assembly is more efficient (produces more desired recombinant plasmids) and as specific (produces few undesired recombination products) as conventional subcloning. The 5RM continuous DNA assembly is approximately as efficient and specific as conventional Golden Gate assembly, even though in vivo methylation of one plasmid is incomplete. Conclusion: Both methylase-assisted methods streamline BioBrick assembly workflows without complicating the design of synthetic parts.
Method summary
Plasmids that comply with the BioBrick standard are used to transform Escherichia coli cells carrying DNA methyltransferase expression vectors. The methylated plasmids are purified and digested with Type II restriction endonucleases. All restriction fragments are ligated to each other without purification, but unwanted ligation products are linearized in a second round of restriction digestion before the mixture is used to transform E. coli.
Synthetic biology in practice is usually predicated on the assembly of short stretches of synthetic DNA into longer ones. Several DNA assembly methods have already been invented and described in the literature [1–5], each with particular strengths and shortcomings. Those that rely on PCR, for example, reduce subcloning steps, which require more time and effort. Yet, PCR products often contain unwanted random mutations, so PCR-based DNA assembly methods necessitate the re-sequencing of parts after they have been recombined and re-cloned. However, those that rely on restriction endonucleases to recombine fragments of plasmids do not.
One early innovation that galvanized synthetic biologists was the BioBrick standard [6]. Every DNA “part” is cloned into the same polylinker (EcoRI-XbaI-part-SpeI-PstI). All parts must be synthesized to exclude the restriction sites in the polylinker. Any two can be combined by double digesting one with Type II restriction endonucleases SpeI and either EcoRI or PstI, the other with XbaI (and EcoRI or PstI), and ligating the desired fragments with compatible overhangs together. The compatible SpeI and XbaI overhangs anneal to form a ligation scar (5′-ACTAGA-3′) that cannot be recognized by either enzyme. The desired ligation products thus end up with the same pattern of unique restriction sites around each part as each original, so that users can combine any number of parts, via iterative pairwise assembly, without ever running out of unique restriction sites. The most significant drawback to BioBrick assembly is that it originally relied on agarose gel purification of desired fragments, which is labor-intensive and recalcitrant to automation [7]. The other limitations (exclusion of restriction sites within parts, formation of ligation scars and pairwise ligations) cannot be circumvented by any BioBrick assembly method, including those described here.
Golden Gate assembly was subsequently invented as a labor-saving alternative [8]. This method utilizes Type IIS restriction endonucleases, such as BsaI, which recognizes a particular restriction site (5′-GGTCTC-3′) and introduces staggered nicks in both downstream strands regardless of their nucleotide sequences. DNA fragments are designed so that each will be cut on both ends by BsaI, creating complementary although not necessarily palindromic overhangs. A single enzyme can thus be employed to create any of 256 unique overhangs, thereby enabling the assembly of open reading frames and other genes without ligation scars. Restriction digestion separates the restriction sites recognized by BsaI from the desired parts, so ligation products are no longer recognized by that enzyme. The restriction endonuclease and T4 DNA ligase can thus work continuously (i.e., simultaneously) within the same reaction vessel, obviating any need for gel purification of desired fragments. Three or more restriction fragments (including recipient plasmids), each originally bound by BsaI sites that create overhangs that are uniquely compatible with that of one other restriction fragment, can be assembled in a single continuous reaction.
Golden Gate assembly, however, is not without drawbacks. BioBrick compatible parts are relatively easy to design because the prefix sequences upstream and suffix sequences downstream are identical for every part. The design of parts for Golden Gate assembly requires more time and effort. Type IIS restriction sites are non-palindromic so each part must be preceded by a site in the forward orientation and followed by a second in the reverse. Overhangs must, of course, complement only those of desired fusion partners. Certain overhang sequences anneal more efficiently than others, and ligation of mismatched overhangs can happen [9]. Many users rely on software to design DNA for Golden Gate assembly reactions [10]. Others, including the inventors of the MoClo system [11], have standardized overhangs but each has rules that take time to learn. If any assembly proves inefficient or inaccurate, new fragments must be designed, synthesized, and sequenced after assembly and cloning. Even if a construct assembles correctly, the original parts won't necessarily be compatible with other parts designed for other Golden Gate standards.
The BioBrick cloning standard is simpler, because parts are easier to design, and all parts are compatible with each other. The author previously introduced a way to reduce the labor required for assembly [12]. The 4R/2M (four restriction endonucleases, two DNA methyltransferases) assembly was inspired by prior work by Leguia et al. [13]. The recipient and donor plasmids are separately methylated in vivo, so that each is protected from the action of a single restriction endonuclease. Each plasmid is double digested separately with two restriction enzymes, which are subsequently heat denatured. The four restriction fragments are ligated to each other (and to other copies of themselves). The T4 DNA ligase is heat denatured, and the ligation products are digested with a pair of restriction enzymes so that only the desired recombinant plasmid remains uncut and capable of transforming competent E. coli efficiently.
The 4R/2M BioBrick assembly technique obviates gel purification, the step that is most labor intensive and recalcitrant to automation, so users can carry out larger numbers of pairwise BioBrick assemblies in parallel. The method nevertheless has two shortcomings that are addressed in this study. First, the M.PstI DNA methyltransferase expression vector failed to protect plasmids as completely as the corresponding M.EcoRI, M.XbaI or M.Ocy1 (M.SpeI orthologue) vectors. This problem complicated 4R/2M assembly schemes in which the donor plasmid was cut with EcoRI and SpeI, and the recipient with EcoRI and XbaI. The other shortcoming of 4R/2M is that restriction digestion and ligation occur in discrete sequential steps, necessitating a mixing step not required in conventional Golden Gate assembly. This article describes how in vivo plasmid methylation enables continuous 5RM assembly using SpeI, XbaI and other Type II restriction endonucleases.
Materials & methods
Materials
The synthetic methylase genes used in this study (M.XhoI and M.SalI) were purchased from Twist Biosciences (South San Francisco, CA, USA). M.MunI was purchased as a gBlock from IDT (IA, USA), as were others previously described (M.EcoRI, M.XbaI, M.Ocy1ORF8430P and M.PstI). All other materials and their sources were also described in the same earlier study [12].
Plasmid construction
Three new DNA methyltransferase expression vectors were constructed to enable two varieties of gel-free subcloning. Each vector was designed to exclude the restriction sites that could interfere with ligation reactions. Prham-M.EcoRI-p15A-aadA, Prham-M.Ocy1-p15A-aadA and Prham-M.XbaI-p15A-aadA were previously shown to methylate the EcoRI, SpeI and XbaI sites in high copy number plasmids to completion; the former two can be employed for 4R/2M BioBrick assembly. In principle, Prham-M.XbaI-p15A-aadA and Prham-M.PstI-p15A-aadA should also have worked for this purpose, but the latter vector was never able to methylate PstI sites to completion [12]. Shortly after the publication of that article, an undesired R214H missense mutation was discovered in the M.PstI gene. The gene was repaired via PCR-mediated site-directed mutagenesis, thereby creating a new Prham-M.PstI-p15A-aadA encoding a wild-type DNA methyltransferase gene. Two additional DNA methyltransferase expression vectors, Prham-M.XbaI-M.EcoRI-M.SalI-p15A-aadA and Prham-M.XhoI-M.MunI-M.Ocy1-p15A-aadA, were constructed to mediate 5RM Golden Gate assembly, as described in the following sections.
Four additional test plasmids were made to measure the efficiencies of the 4R/2M, Golden Gate and 5RM assembly protocols. Plasmids lacI-Ptac-lacO-IMBB4-pUC and tagRFP-IMBB4-pUC contain BioBrick-compliant inserts within the IMBB4 (Ichiro Matsumura BioBrick 4) polylinker: MfeI-EcoRI-XbaI-part-SpeI-PstI-XhoI-SalI. In other words, the tac promoter and Red Fluorescent Protein are both BioBrick compatible, but each is preceded by an extra MfeI site upstream and XhoI and SalI sites downstream (in comparison with the previously described IMBB2 [14]). Restriction endonucleases MfeI and EcoRI produce compatible cohesive overhangs. XhoI and SalI similarly create overhangs compatible with each other, although not with MfeI/EcoRI nor XbaI/SpeI. Plasmids lacI-Ptac-lacO-BsaI-BsaI-IMBB4-pUC and BsaI-tagRFP-BsaI-pUC-kan (not BioBrick compatible) were constructed to measure the cloning efficiency of Golden Gate. Each encodes BsaI sites with complementary overhangs (TATG and TGGT).
4R/2M (EcoRI) assembly
Donor plasmid lacI-Ptac-lacO-IMBB4-pUC was methylated in vivo in E. coli also carrying Prham-M.XbaI-p15A-aadA (in triplicate) by following an auto-induction protocol developed earlier [12]. The double transformants were propagated overnight in liquid LB supplemented with 100 μg/ml spectinomycin, 100 μg/ml ampicillin, 0.1% L-rhamnose and 0.001% glucose. The rhamnose promoter is subject to catabolite repression so M.XbaI is expressed only after limiting concentrations of glucose in the media have been depleted. The plasmids were purified by the QIAprep Spin Miniprep protocol (Qiagen), quantified by microspectrophotometry (using Take3 micro-volume plates in a BioTek Synergy2 microplate reader) and double digested overnight with EcoRI-HF and SpeI-HF (500 ng plasmid reacted with 10 units each in 25 μl 1× NEB CutSmart buffer, Ipswich, MA, USA).
Recipient plasmid tagRFP-IMBB4-pUC was similarly methylated with Prham-M.PstI-p15A-aadA, purified and digested with EcoRI-HF and XbaI. The restriction endonucleases in both digests were heat denatured (20 min at 80°C). The digested plasmids were mixed, and all four restriction fragments were ligated by T4 DNA ligase (80 ng of each plasmid, or control reactions with 80 ng of one plasmid and buffer, reacted with one Weiss unit of enzyme in 20 μl T4 DNA ligase buffer plus 5% polyethylene glycol, molecular weight 6,000, 500 cycles of 30 s at 10°C, 30 sat 30°C) to each other. The ligase was heat denatured (10 min at 65°C) and the ligation products (32 ng total DNA) were digested with XbaI and PstI-HF (12 units each in 32 μl 1× NEB CutSmart buffer, 1 h at 37°C); the desired lacI-Ptac-lacO-tagRFP-IMBB4-pUC should be the only viable construct resistant to both enzymes. All experiments employed the same batch of OmniMax 2 cells (ThermoFisher) made competent by the classical method of Inoue et al. [15]. Transformation efficiency was 3 × 107/μg, as determined by counting colonies after transformation with 10 pg of pUC19. All assembly experiments were conducted in triplicate, starting with six isolated colonies (three carrying recipient plasmids and three others carrying donors) and ending with nine transformations (recipient only, donor only, recipient and donor for each pair of parental plasmids). The colony counts were averaged and scaled to colony forming units (CFU) per nanogram. Values were rounded to the nearest integer, except for those less than one.
Golden Gate assembly
Plasmids lacI-Ptac-lacO-BsaI-BsaI-IMBB4-pUC and BsaI-tagRFP-BsaI-pUC-kan were purified and reacted with BsaI-HF v2 and T4 DNA ligase (200 ng of each plasmid, or one plasmid plus buffer, reacted with 20 units of restriction enzyme and 1.2 Weiss unit of ligase in 20 μl NEB CutSmart buffer containing 1 mM ATP) in a thermocycler programmed to oscillate between temperatures optimal for overhang annealing, ligation and restriction digestion (72 cycles of 5 min at 37°C, 10 nested cycle of 30 sat 10°C and 30 sat 30°C). The enzymes were heat denatured (10 min at 65°C), and the ligation products were diluted and digested again with BsaI-HF v2 (40 ng total DNA reacted with 16 units of restriction enzyme in 40 μl NEB CutSmart buffer for an h at 37°C).
5RM assembly
Each of the triply methylated plasmids were purified separately, then mixed together (200 ng each) with T4 DNA ligase (1.2 Weiss units) and Type II restriction endonucleases SpeI-HF, XhoI, XbaI and SalI-HF (5 units each) in NEB CutSmart buffer supplemented with 1 mM ATP (20 μl total volume). The reaction was thermocycled as in the Golden Gate assembly protocol, and the enzymes were subsequently heat denatured (10 min at 65°C). The ligation products were diluted and further digested with EcoRI-HF and XhoI (40 ng total DNA was reacted with16 units each in 40 μl 1× NEB CutSmart buffer, 1 h at 37°C).
Results & discussion
New M.PstI expression vector enables 4R/2M assembly using EcoRI
Test plasmids, lacI-Ptac-lacO-IMBB4-pUC (donor) and tagRFP-IMBB4-pUC (recipient), were used to demonstrate the utility of the Prham-M.XbaI-p15A-aadA and new Prham-M.PstI-p15A-aadA expression vectors for gel-free 4R/2M BioBrick assembly (Figure 1 & Table 1). The XbaI site of the donor was methylated in vivo (in triplicate). This plasmid was subsequently purified and double digested overnight with EcoRI and SpeI. The PstI site of the recipient was similarly methylated, and this plasmid was purified and digested with EcoRI and XbaI. The enzymes were heat denatured, and all the four restriction fragments were mixed and ligated to each other (or to other copies of themselves). The ligase was heat denatured, and the ligation products were digested with XbaI and PstI so that only the desired recombinant plasmid, lacI-Ptac-lacO-tagRFP-IMBB4-pUC, remained viable.
# | Name of DNA | Desired? | Viable? | Cut by | Protected from |
---|---|---|---|---|---|
1 | Parental donor | Initially | Yes | EcoRI, SpeI, PstI | XbaI |
2 | Parental recipient | Initially | Yes | EcoRI, XbaI, SpeI | PstI |
3 | Insert fragment | Transiently | No | XbaI | |
4 | Donor fragment | No | No | PstI | |
5 | Stuffer fragment | No | No | ||
6 | Recipient fragment | Transiently | No | SpeI | PstI |
7 | Insert homodimer | No | No | EcoRI, SpeI | XbaI |
8 | Donor homodimer | No | No | EcoRI, SpeI, PstI | |
9 | Stuffer homodimer | No | No | EcoRI, XbaI | |
10 | Recipient homodimer | No | No | EcoRI, XbaI, SpeI | PstI |
11 | Insert-stuffer heterodimer | No | No | EcoRI | XbaI |
12 | Stuffer-donor heterodimer | No | Yes | EcoRI, PstI | |
13 | Donor-recipient heterodimer | No | Yes | EcoRI, SpeI, PstI | |
14 | Insert-recipient heterodimer | Finally | Yes | EcoRI, SpeI | XbaI, PstI |
When one nanogram of digested ligation product was used to transform E. coli OmniMax2, 226 (average of three replicates) ± 48 (standard error) pink colonies were formed on LB agar plates supplemented with ampicillin indicating leaky expression of red fluorescent protein from the desired lacI-Ptac-lacO-tagRFP-IMBB4-pUC product (Table 2). Only 9 ± 1 white colonies, indicative of unwanted ligation products, were formed on the same plates. Colony counts on control plates that corresponded to assemblies that included only recipient plasmid (7 ± 3) or only donor plasmid (6 ± 1) were similarly low, suggesting that in vivo methylation and in vitro restriction digestion were complete or nearly so.
Assembly protocol | Vector only | Insert only | Vector + insert (red) | Vector + insert (white) |
---|---|---|---|---|
Gel purify (EcoRI) | 7 ± 2 | 61 ± 27 | 126 ± 44 | 11 ± 4 |
4R/2M (EcoRI) | 7 ± 3 | 6 ± 1 | 226 ± 48 | 9 ± 1 |
Golden Gate | 4 ± 0.4 | 0.2 ± 0.2 | 302 ± 58 | 93 ± 18 |
5RM (XhoI/SalI) | 147 ± 43 | 0.7 ± 0.3 | 417 ± 97 | 43 ± 10 |
5RM (EcoRI/MfeI) | 1 ± 0.4 | 5 ± 1 | 86 ± 9 | 10 ± 2 |
These colony counts are favorable to those produced by traditional subcloning, both in terms of cloning efficiency (number of pink colonies; Table 2) and specificity (absence of undesired background products). They are comparable to 4R/2M assembly reactions that were affected with other DNA methyltransferases, namely Prham-M.EcoRI-p15A-aadA and Prham-M.Ocy1-p15A-aadA, and other restriction endonucleases (SpeI, PstI, XbaI and EcoRI-HF) [12]. With the second pair DNA methyltransferase expression vectors, Prham-M.XbaI-p15A-aadA and the new Prham-M.PstI-p15A-aadA, parts in existing BioBrick-compatible plasmids can now be combined via 4R/2M assembly with either plasmid serving as donor or recipient. The workflow is significantly less labor intensive, and yet more efficient, than are traditional subcloning protocols.
Golden Gate assembly is also efficient when enzyme & substrate concentrations are optimal
Golden Gate assembly features continuous restriction digestion and ligation, so it is even less labor intensive than is the 4R/2M protocol. Plasmids lacI-Ptac-lacO-BsaI-BsaI-IMBB4-pUC and BsaI-tagRFP-BsaI-pUC-kan encode complementary overhangs produced by the Type IIS restriction endonuclease BsaI. The two plasmids were purified and reacted with BsaI-HF v2 and T4 DNA ligase at temperatures optimal for overhang annealing, ligation, and restriction digestion. The assembly process was efficient as expected. When one nanogram of product was used to transform E. coli OmniMax2, 302 ± 58 pink colonies were formed (Table 2). Background, mostly likely representing the parental recipient, lacI-Ptac-lacO-BsaI-BsaI-IMBB4-pUC, was surprisingly high, as 93 ± 18 white colonies formed on the same plates. Transformation of control reactions containing only recipient or donor plasmids produced few white colonies (4 ± 0.4 or 0.2 ± 0.2, respectively), so it is possible that restriction enzyme BsaI-HF v2 was substrate inhibited by the higher concentration of DNA in the experimental reactions.
5RM assembly is a Golden Gate-like reaction using Type II restriction endonucleases
Type II restriction endonucleases cleave within their recognition sequences. Ligations of two strands cut with the same Type II enzyme are thus normally susceptible to recognition and cleavage by the same enzyme, necessitating the elimination of that enzyme (by gel purification or heat denaturation) prior to ligation. Pairs of Type II restriction endonucleases that create compatible cohesive ends, however, can be used in Golden Gate assembly reactions. For example, XbaI and SpeI recognize and cleave different palindromic sites. The resulting restriction fragments ligate to each other to form a “scar” that is neither recognized nor cut by either enzyme. Since all BioBrick compatible parts are bound by a 5′ XbaI and 3′ SpeI, one part cut only with SpeI can be ligated to another cut only with XbaI, creating a new part that is bound by the same restriction sites. To enable continuous, directional assembly, the BioBrick standard was extended to include some new restriction sites (underlined): MfeI-EcoRI-NotI-XbaI-part-SpeI-NotI-PstI-XhoI-SalI. Type II restriction endonucleases MfeI and EcoRI recognize and cleave different sites to create compatible cohesive ends. XhoI and SalI similarly create cohesive ends that are compatible to each other, but not to those created by MfeI/EcoRI or XbaI/SpeI. In principle, three pairs of restriction sites, each cut to produce compatible cohesive ends, should suffice to enable Golden Gate-like assembly with the right subset of restriction enzymes (Figure 2). The desired ligation product will contain an MfeI/EcoRI or XhoI/SalI ligation scar, in addition to the XbaI/SpeI scar. Two DNA methyltransferase expression vectors, each designed to protect three sites, one of each pair (EcoRI/XbaI/SalI or MfeI/SpeI/XhoI) were constructed. E. coli carrying Prham-M.EcoRI-M.XbaI-M.SalI-p15A-aadA was co-transformed with test plasmid lacI-Ptac-lacO-IMBB4-pUC; isogenic cells carrying Prham-M.MunI (MfeI)-M.Ocy1 (SpeI)-M.SalI-p15A-aadA were co-transformed with tagRFP-IMBB4-pUC.
Each of the methylated plasmids were purified (in triplicate), mixed and reacted with T4 DNA ligase and SpeI, XhoI, XbaI and SalI in a thermocycled reaction otherwise identical to the Golden Gate assembly protocol (Figure 3 & Table 3). The ligation products were diluted and further digested with EcoRI and XhoI. Transformation of E. coli with one nanogram of total DNA led to the formation of 417 ± 97 pink colonies and 43 ± 10 white ones (Table 2). Assembly efficiency was thus greater than that of my conventional Golden Gate assembly and background was lower, thereby validating the design of the 5RM workflow. The vector only control, however, produced a surprisingly high 147 ± 43 colonies, which suggested that the Prham-M.EcoRI-M.XbaI-M.SalI-p15A-aadA expression vector failed to methylate the XbaI and SalI sites to completion. Very few colonies (0.7 ± 0.3) appeared on plates corresponding to the insert only control, as expected.
# | Name of DNA | Desired? | Viable? | Cut by | Protected from |
---|---|---|---|---|---|
1 | Parental recipient | Initially | Yes | SpeI, XhoI | EcoRI, XbaI, SalI |
2 | Parental donor | Initially | Yes | EcoRI, XbaI, SalI | SpeI, XhoI |
3 | Recipient fragment | Transiently | No | EcoRI, XbaI, SalI | |
4 | Stuffer fragment | No | No | ||
5 | Donor fragment | No | No | EcoRI | |
6 | Insert fragment | Transiently | No | SpeI, XhoI | |
7 | Donor homodimer | No | No | EcoRI, XbaI, SalI | |
8 | Stuffer homodimer | No | No | SpeI, XhoI | |
9 | Recipient homodimer | No | No | SpeI, XhoI | EcoRI, XbaI, SalI |
10 | Insert homodimer | No | No | SalI, XbaI | SpeI, XhoI |
11 | Donor-stuffer heterodimer | No | Yes | EcoRI | |
12 | Donor-recipient heterodimer | No | Yes | EcoRI | XbaI, SalI |
13 | Insert-stuffer heterodimer | No | No | SpeI, XhoI | |
14 | Insert-recipient heterodimer | Finally | Yes | EcoRI, XbaI, SpeI, XhoI, SalI |
When BioBrick parts are assembled, the user can select either parent plasmid as recipient vector. The selectively methylated plasmids can also be digested with one of two different sets of four Type II restriction endonucleases. The plasmid preparations described earlier were mixed (200 ng each) with T4 DNA ligase (1.2 Weiss units) and Type II restriction endonucleases EcoRI-HF, XbaI, MfeI-HF and SpeI-HF (5 units each) in the same buffer and thermocycled in the same way. The enzymes were heat denatured, and the ligation products were diluted and digested further with EcoRI-HF and XhoI. One nanogram of digested ligation product was used to transform chemically competent E. coli, leading to the formation of 86 ± 9 pink colonies and 10 ± 2 white ones on LB agar plates supplemented with ampicillin (Table 2). This assembly was adequate but produced fewer desired recombinant ligation products than did Golden Gate or 5RM using the other set of enzymes. The assembly was also specific, as control reactions containing only the recipient or donor plasmids produced only 1 ± 0.4 and 5 ± 1, respectively.
The efficiency and accuracy of 5RM reactions are predicated on the assumption that all of the DNA methyltransferases protect their target restriction sites specifically and completely. This assumption was tested by reacting the methylated plasmids used in 5RM assembly with individual restriction endonucleases at high concentrations for an extended amount of time. Each plasmid was also cut with BsaI, which recognizes a unique site in the beta-lactamase gene away from the BioBrick part, so that the digestion products could be easily distinguished via agarose gel electrophoresis. The tagRFP-IMBB4-pUC plasmid preparation, which also contains Prham-M.XhoI-M.MunI-M.Ocy1-p15A-aadA, was cut by EcoRI-HF, XbaI and SalI, but not by MfeI, SpeI or XhoI, as expected. In contrast, lacI-Ptac-lacO-IMBB4-pUC, co-transformed with Prham-M.XbaI-M.EcoRI-M.SalI-p15A-aadA was cut by MfeI-HF, SpeI and XhoI. It was protected from EcoRI-HF as expected but was partly digested by XbaI and SalI (Figure 4). The latter results were replicated when different plasmid preparations, including those purified on different days, were partially digested (data not shown).
Incomplete methylation confounds 5RM assembly
5RM assembly is predicated on complete methylation of complementary restriction sites on two plasmids. Unwanted digestion of unprotected sites within desired restriction fragments (insert and recipient vector) creates shorter double stranded DNAs that ligate to intact copies of insert or recipient plasmid, thereby preventing the formation of the desired recombinant product. The shorter DNAs can also ligate to form undesired products that survive the fifth restriction enzyme. The protection from XbaI and SalI conferred by the Prham-M.XbaI-M.EcoRI-M.SalI-p15A-aadA expression vector upon the cohabiting lacI-Ptac-lacO-IMBB4-pUC plasmid is incomplete (Figure 4). The other DNA methyltransferase expression vector, Prham-M.XhoI-M.MunI-M.Ocy1-p15A-aadA, protected most if not all cohabiting copies of tagRFP-IMBB4-pUC from MfeI, SpeI and XhoI.
This asymmetry was reflected in the colony count data (Table 2). In 5RM (XhoI/SalI), some of the recipient lacI-Ptac-lacO-IMBB4-pUC vector was cut with both XbaI and SpeI, or XhoI and SalI, which is probably why so many colonies formed on the vector only control plate. As noted above, the 5RM (XhoI/SalI) assembly reaction containing both donor and recipient plasmids was approximately as efficient and accurate as Golden Gate assembly. Some of the undesired fragments must have ligated to each other, forming unviable constructs. The 5RM (MfeI/EcoRI) assembly using the same preparations of plasmid were less efficient (fewer pink colonies) but more accurate (fewer white ones), suggesting that the undesired XbaI digestion of unprotected lacI-Ptac-lacO-IMBB4-pUC donor fragments reduced the quantity of properly cut insert fragments and increased that of unwanted but inviable ligation products.
It should be possible to achieve stoichiometric methylation of recombinant plasmids (and host chromosomes) by emulating natural design. All six DNA methyltransferase genes employed in this study were derived from operons that also contain the corresponding restriction endonuclease genes. The co-expression of a DNA methyltransferase and restriction endonuclease protects the cell from phages, whose chromosomes are initially unprotected. Restriction enzyme activity also serves as a quality control checkpoint for the DNA methyltransferase since any deficit of the latter would lead to catastrophic breaks in both strands of the chromosome. Ideally, these restriction-modification operons can either be cloned or re-synthesized, then integrated in combination into the E. coli chromosome so that expression of the encoded genes won't fluctuate as much as they would if multicopy plasmids were employed instead. The high copy number plasmids used in the study were co-purified with the low copy number DNA methyltransferase expression plasmids used to methylate them in vivo. Some bacteria, such as Neisseria meningitidis, express multiple restriction modification systems, so the integration of the three per strain for plasmid methylation prior to 5RM assembly is feasible.
4R/2M assembly is currently more reliable than 5RM
5RM assembly, like traditional Golden Gate reactions, is completed after a single mixing step, so it is amenable to high throughput or automated methods. The 4R/2M assembly protocol, however, is more reliable with existing DNA methyltransferase expression vectors for several reasons. First, it is easier to methylate a single restriction site within a population to completion than it is to methylate three. Second, 4R/2M assembly doesn't require complete methylation as 5RM does. Partial methylation simply leads to partial digestion of the desired insert-recipient construct after all the ligations have been completed. Third, the 4R/2M assemblies are also easier to troubleshoot and optimize. Each plasmid is cut with two enzymes prior to ligation, so agarose gel analysis between these reactions can show whether the DNA concentrations were estimated correctly and whether the digests were complete. Fourth, each reaction is carried out under optimal conditions (buffers, reactant concentrations, temperatures, and time) rather than some compromise tolerable to restriction endonucleases and T4 DNA ligase.
BioBrick users can obtain the four DNA methyltransferase vectors (Prham-M.EcoRI-p15A-aadA, Prham-M.Ocy1-p15A-aadA, Prham-M.XbaI-p15A-aadA and the new version of Prham-M.PstI-p15A-aadA described above) from Addgene, and to apply the 4R/2M assembly protocols to combine BioBrick compliant parts in their existing plasmids. The 5RM assembly method introduced here captures the strengths of the BioBrick cloning standard (ease of design, universal compatibility of parts) and the Golden Gate protocol (minimal labor). Those who wish to apply 5RM assembly can also obtain the two DNA methyltransferase expression vectors (Prham-M.XbaI-M.EcoRI-M.SalI-p15A-aadA and Prham-M.XhoI-M.MunI-M.Ocy1-p15A-aadA) and two test plasmids (lacI-Ptac-lacO-IMBB4-pUC and tagRFP-IMBB4-pUC) from Addgene. New parts will have to be subcloned into these test plasmids prior to 5RM assembly, and a variation of the 4R/2M protocol (in which the two plasmids are differentially methylated but initially cut with the same pair of enzymes) is well suited for this purpose. The now established 4R/2M protocol enables BioBrick assemblies without gel purifications. The novel 5RM method streamlines that process further into Golden Gate-like reactions.
Future perspective
In the future, DNA might be assembled in vivo through heterologous expression of DNA methyltransferases, followed by plasmid conjugation into strains expressing restriction endonucleases and T4 DNA ligase.
BioBrick-compliant parts are easy to design but normally labor-intensive to assemble.
Golden Gate assembly requires much less labor, but parts are more difficult to design.
DNA methyltransferases protect plasmids from subsequent restriction digestion.
Gel purifications can be circumvented by plasmid methylation and post-ligation digests.
4R/2M BioBrick assembly thus saves labor, and it is efficient, specific and reliable.
Two DNA methyltransferase expression vectors are required for 4R/2M assembly.
With two pairs of DNA methylases, any BioBrick part can become an insert or vector.
5RM enables continuous Golden Gate-like assembly of non-standard BioBrick parts.
5RM assembly is limited by incomplete plasmid methylation.
Accession numbers
The plasmids described in this study are available from Addgene: Prham-M.PstI-p15A-aadA (149340), Prham-M.XbaI-p15A-aadA (149338), Prham-M.XbaI-M.EcoRI-M.SalI-p15A-aadA (170767), Prham-M.XhoI-M.MunI-M.Ocy1ORF8430P-p15A-aadA (170768), tagRFP-IMBB4-pUC (170769), lacI-Ptac-lacO-IMBB4-pUC (170770), BsaI-tagRFP-BsaI-pUC-kan (170771) and lacI-Ptac-lacO-BsaI-BsaI-IMBB4-pUC (170772).
Author contributions
Ichiro Matsumura designed and conducted all experiments, and did all the writing.
Acknowledgments
The author is grateful to Cary Valley, Senior QC Scientist at Addgene, for discovering the unwanted R214H mutation in the original Prham-M.PstI-p15A-aadA vector.
Financial & competing interests disclosure
IM was supported by Emory University. The experiments were conducted with reagents left over from a project supported by the National Science Foundation (MCB 1413062). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.
Open access
This work is licensed under the Attribution-NonCommercial-NoDerivatives 4.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/
References
- 1. . Bricks and blueprints: methods and standards for DNA assembly. Nat. Rev. Mol. Cell Biol. 16(9), 568–576 (2015).
- 2. . Recent advances in DNA assembly technologies. FEMS Yeast Res. 15(1), 1–9 (2015).
- 3. . Overview of post Cohen-Boyer methods for single segment cloning and for multisegment DNA assembly. Curr. Protoc. Mol. Biol. 113(1), 3 26 21–23 26 20 (2016).
- 4. . DNA assembly standards: setting the low-level programming code for plant biotechnology. Plant Sci. 273, 33–41 (2018).
- 5. . In vivo DNA assembly using common laboratory bacteria: a re-emerging tool to simplify molecular cloning. J. Biol. Chem. 294(42), 15271–15281 (2019).
- 6. . Idempotent vector design for standard assembly of biobricks. MIT Synthetic Biology Working Group (2003).
- 7. . Why Johnny can't clone: common pitfalls and not so common solutions. BioTechniques 59(3), IV–XIII (2015).
- 8. . A one pot, one step, precision cloning method with high throughput capability. PLoS One 3(11), e3647 (2008).
- 9. Comprehensive profiling of four base Overhang ligation Fidelity by T4 DNA ligase and application to DNA assembly. ACS Synth. Biol. 7(11), 2665–2674 (2018).
- 10. . Enabling one-pot Golden Gate assemblies of unprecedented complexity using data-optimized assembly design. PLoS One 15(9), e0238592 (2020).
- 11. . A modular cloning system for standardized assembly of multigene constructs. PLoS One 6(2), e16765 (2011).
- 12. . Methylase-assisted subcloning for high throughput BioBrick assembly. Peer J . 8, e9841 (2020).
- 13. . 2ab assembly: a methodology for automatable, high-throughput assembly of standard biological parts. J Biol. Eng. 7(1), 2 (2013).
- 14. . Semi-automated Tip Snip cloning of restriction fragments into and out of plasmid polylinkers. BioTechniques 62(3), 99–106 (2017).
- 15. . High efficiency transformation of Escherichia coli with plasmids. Gene 96(1), 23–28 (1990).