Enzymatic Assembly of Small Synthetic Genes with Repetitive Elements

Abstract

Gene synthesis efficiency has greatly improved in recent years but is limited when it comes to repetitive sequences, which results in synthesis failure or delays by DNA synthesis vendors. This represents a major obstacle for the development of synthetic biology since repetitive elements are increasingly being used in the design of genetic circuits and design of biomolecular nanostructures. Here, we describe a method for the assembly of small synthetic genes with repetitive elements: First, a gene of interest is split in silico into small synthons of up to 80 base pairs flanked by Golden-Gate-compatible overhangs. Then, synthons are made by oligo extension and finally assembled into a synthetic gene by Golden Gate Assembly. We demonstrate the method by constructing eight challenging genes with repetitive elements, e.g., multiple repeats of RNA aptamers and RNA origami scaffolds with multiple identical aptamers. The genes range in size from 133 to 456 base pairs and are assembled with fidelities of up to 87.5%. The method was developed to facilitate our own specific research but may be of general use for constructing challenging and repetitive genes and, thus, a valuable addition to the molecular cloning toolbox.

Introduction

Advances in DNA synthesis technologies have been pivotal for synthetic biology research. With the development of DNA assembly methods such as restriction-ligation,¹ polymerase chain assembly,² ligation-independent cloning methods^3,4 and in-vitro-⁵ or in-vivo-homology-based⁶ methods, it is possible to construct almost any synthetic gene, biosynthetic pathway, or genome. Of the many methods, the one-pot restriction-ligation method, Golden Gate Assembly⁷ (GGA), has become a standard method for the synthetic biology workflows due to its utility for rapid and high-throughput cloning. GGA utilizes type IIS restriction enzymes that cut distal to their recognition site, which allows overhangs to be designed with an arbitrary sequence for scarless cloning. Although DNA synthesis technologies have advanced drastically, the synthesis of DNA sequences with high or low GC content, stable secondary structures, and multiple repeats remain challenging,⁸ where especially repetitive DNA sequences result in synthesis failure or delays by vendors.⁹ To circumvent some of these challenges, it is possible to codon optimize or codon scramble the sequence of genes encoding proteins to enable successful synthesis with standard methods.¹⁰ However, this approach is not applicable for genes encoding repeats of functional noncoding RNA elements, since even a few sequence changes may change the folding and negatively affect the function of the RNA element. Utilization of functional RNA elements, either isolated from nature or de novo designed, has been increasingly popular in both the fields of synthetic biology¹¹ and RNA nanotechnology.¹² Thus, a DNA assembly method to successfully construct genes with repetitive sequences is needed.

Methods to assemble synthetic DNA with repeats have been developed previously. These techniques assemble shorter DNA sequences into larger repetitive DNA sequences using either restriction-ligation with type IIS restriction enzymes in multiple steps^13,14 or simply by annealing oligos¹⁵ to create sequence-defined overhangs for ligation. While these methods have proven useful for the scarless construction of repetitive DNA, the approach with type IIS restriction enzymes can be put into a GGA workflow for rapid assembly with minimal handling. GGA-based methods exist for the construction of at least 12 repeats for CRISPR guide-RNA arrays.¹⁶ However, these methods rely on preconstructed templates for the amplification of the scaffold part of the guide-RNA and are therefore not generalizable. Furthermore, these methods have not demonstrated the assembly of repetitive DNA, where specific linker or scaffold sequences with variable lengths are interspersed between the repeats such as for genes encoding RNA origami nanostructures¹² that display multiple copies of the same RNA element (Figure 1). Therefore, we wanted to build on this prior work to make a DNA assembly method to solve this challenge.

Figure 1.

Example of repetitive noncoding RNA gene. Schematic representation of the secondary structure of a two-helix RNA origami scaffold¹² (blue) with repeated elements (red) and its associated DNA template.

Results and Discussion

Here, we describe a two-step DNA assembly method for the construction of synthetic genes with repetitive elements that vendors could not provide. An overview of the method is shown in Figure 2. The method consists of splitting the gene of interest in silico into synthons of up to 80 base pairs (bp) using molecular biology design software such as Benchling.¹⁷ We chose to design our method around GGA since this method uses small overhangs (4 bp) that can be designed to minimize undesired annealing in contrast to homology-based approaches which require at least 15-bp of homology⁴ and since the synthons can be directly inserted into a desired destination vector. The genes are split into roughly equal-sized fragments depending on the sequence size starting from position 1 to position 70–80 and then from position 70–80 to position 140–150 and so on. To design the compatible 4 bp overhangs, we used the GGA tool from Benchling. Other tools for overhang design exist, such as the NEB Golden Gate tool¹⁸ or the Golden Gate overhang designer from the Edinburgh biofoundry.¹⁹ Final synthons were made by oligo extension with a DNA polymerase using oligos of up to 60 nucleotides (nt) that overlap with 18–25 bases on their 3′ ends for hybridization to reduce misannealing and allow for extension of short oligos to longer synthons. The method was developed to maximize the synthon sizes while minimizing oligo costs and keeping misannealing probabilities low. Shorter oligos can be used, but that would lead to an increased synthon number that would affect GGA efficiency. Longer oligos would not only increase production costs since they require an additional purification step but also increase the occurrence of repeated regions that could lead to misannealing. Using synthon sizes of 70–80 bp enabled the use of oligos of up to 60 nts, which is the lower price tier of IDT and other DNA synthesis vendors that offer the synthesis of custom DNA oligos. For a 60 nt oligo, only up to 25–30 nts can be unique, since the oligo sequence needs to include 10–11 nts for the type IIS restriction site and GGA overhang as well as 20–25 nts shared between oligo pairs for annealing purposes. With the oligomers synthesized, GGA can be used to assemble the gene and insert it into a desired entry vector.

Figure 2.

Overview of the two-step enzymatic synthesis of a synthetic gene. The designed gene with repeats (red) is split into 70–80 bp synthons in silico with 4-bp regions (orange, green, brown) that are used as overhangs for GGA. Oligos up to 60 nts are designed to cover the gene fragment with the type IIS restriction enzyme sites in the 3′ ends (blue). Oligos covering the termini of the genes include overhangs (purple and yellow) compatible with an entry vector. Double-stranded DNA synthons are synthesized by oligo extension and inserted into a vector by GGA.

To demonstrate the applicability of our method, we assembled synthetic genes encoding tandem repeats of RNA aptamers with unique linker sequences between each repeat and RNA nanostructures with multiple repeated aptamers. These linker sequences were designed for optimal folding of the given RNA constructs, either in the form of linear spacer sequences that do not interact with the remaining structure (2x(Cas6f-hp-MS2wt), 8xCas6f-hp, and 8xMS2) or in the form of stem loops and kissing loops for the RNA origami (remaining) constructs. Our RNA constructs displayed different combinations of known fluorogenic aptamers (Broccoli,²⁰ Corn,²¹ Pepper²²), protein-binding aptamers (MS2²³ hairpin and Cas6f²⁴ hairpin), and a recently selected aptamer (P6Ba) that targets the SARS-CoV2 spike protein.²⁵ Details regarding their secondary structures (Figure S1) and sequences can be found in the Supporting Information.

We chose these synthetic constructs because they require a well-defined sequence between the repetitive elements and therefore cannot be synthesized by methods used for concatemerization of repeats.¹⁴ Sequences ranging from 133 to 456 bp were designed. Several of the sequences failed initial screening for synthesis by the vendor Twist Bioscience,²⁶ and all sequences failed screening by Integrated DNA Technologies²⁷ (IDT) due to their content of repeated elements with repeat lengths ranging from 20 bp to 81 bp (Table S1). The sequences have a complexity score from IDT ranging from 30 to 139 (Table 1). The complexity scores which are proprietary calculations of sequence complexity are given as estimates for the likelihood of synthesis by IDT and are unique to their synthesis technology. Any sequence with a complexity score above 10 cannot be synthesized by IDT, and sequences with scores just below 10 can still be delayed. The sequence with the size of 133 bp was split into two synthons. The sequences ranging from 256 to 312 bp were split into four synthons. The sequence with 399 bp was split into five synthons, and the two sequences with the sizes of 452 and 456 bp were split into six synthons. To synthesize the constructs, we used oligos of up to 60 nts, and each synthon was synthesized in individual reactions to reduce oligo misannealing that can cause faulty assembly. The synthons were assembled by GGA into different destination vectors. The constructs 2H-3xCorn and 2H-3xBroccoli were assembled with a linearized backbone, while the remaining constructs were assembled with a circular backbone. This was done due to the lack of availability of a desired circular entry backbone for those specific constructs. Plasmid sequences with accompanying sequencing data can be found in Table S3, and the sequences for the genes, synthons, and oligos can be found in Table S4, Table S5, and Table S6, respectively.

Table 1. Summary of Assembly Fidelity for Our Eight Different Constructs with Complexity Scores from IDT.

name	length (bp)	synthons	complexity score	fidelity (correct of total)
2x(Cas6f-hp-MS2wt)	133	2	30.2	87.5% (7 of 8)
2H-3xCorn	256	4	51.4	20.0% (2 of 10)
8xCas6f-hp	280	4	139.2	62.5% (5 of 8)
2H-3xBroccoli	301	4	103.3	62.5% (5 of 8)
8xMS2	312	4	101	87.5% (7 of 8)
PX-Tri-P6Ba-UU	399	5	88	25.0% (2 of 8)
dAF14	452	6	123.9	12.5% (1 of 8)
dAF10	456	6	112.9	12.5% (1 of 8)

We successfully constructed all of the synthetic genes that could not be synthesized by the vendors (Table 1). For each construct, we sequenced at least eight clones, with fidelity reported as the percentage of completely correct clones (Table 1). We define any deviation from the designed sequences as a failure since any modifications to the sequences of our RNA constructs can potentially cause misfolding. However, some of these deviations might not be problematic for constructs expressing proteins due to the possibility of synonymous mutations. We found that a screening size of 8 was needed to get at least one successful clone for every construct. The synthesis fidelity reported here is an empirically determined success rate for our experiments based on the sequencing results and should generally correlate to the number of synthons, since Golden Gate reaction efficiency is the limiting factor. We observed that the synthesis fidelity decreased with an increasing number of synthons (Table 1). The highest observed fidelity of 87.5%, i.e., seven correct clones out of eight sequenced, was observed for the assembly of the two-synthon construct 2x(Cas6f-hp-MS2wt) and the four-synthon construct 8xMS2. For 2x(Cas6f-hp-MS2wt), the one failed assembly was due to mutations in one of the synthons, while for 8xMS2 a single mutation was found in a synthon.

In general, the 4-synthon constructs were assembled with a fidelity higher than 60% except for one construct, 2H-3xCorn, where only 20% of the selected clones were successful after two separate cloning attempts. For 2H-3xCorn, it appears that the linear backbone self-ligated, which resulted in colonies that propagated only the vector (Table S3). For 8xCas6f-hp, we observed a duplication of a synthon, together with several errors in another synthon, while for 2H-3xBroccoli we observed that one assembly lacked two synthons, and two of the failed assemblies had 1-bp and 2-bp deletions in different synthons. The 5-synthon construct PX-Tri-P6Ba-UU with an assembly fidelity of 25% had six failed assemblies that lacked one to three synthons. For the two 6-synthon constructs, dAF10 and dAF14, we were only able to recover one correct clone out of eight clones. For dAF10 several synthons were missing in the failed assemblies, while for dAF14 five assemblies lacked synthons and two assemblies had up to two single bp deletions. In summary, we observed that most failed assemblies lacked either partial or whole synthons (Figure S2). The lack of partial synthons indicates errors in the oligo extension that results in shorter synthons while the lack of whole synthons indicates that the overhangs are suboptimal for GGA. This could potentially be improved with computational tools for data-optimized design of Golden Gate overhangs.²⁸ The failed assemblies could also be due to misannealing of the wrong synthons during the GGA reaction or due to manual mishandling of synthons when preparing the assembly in the lab, e.g., using the wrong concentrations. We also observed failed assemblies with just single-bp mutations or deletions. These mutations or deletions could be either due to misannealing of oligos during the extension step or due to synthesis errors of the oligos. Prior work has shown that assembly of synthetic genes with chemically synthesized oligos can be improved after gel-purifying the oligos,²⁹ but this increases the price and turnaround. Furthermore, we did not sequence verify any synthons before assembling the final synthetic genes. Therefore, errors in the synthons could also have occurred during the oligo extension. However, by relying only on cheap and relatively short synthetic DNA oligomers that have a short turnaround time, it is possible to synthesize genes faster than synthetic gene synthesis and delivery from commercial vendors.

We note that there is no correlation between the assembly fidelity and complexity score of the genes. This could be explained by the difference between the gene assembly method that we present and the ones used by DNA synthesis vendors. It is likely that commercial DNA synthesis companies use polymerase chain assembly,³⁰ where multiple oligos are mixed in the same reaction for gene synthesis. This method is especially prone to errors when assembling repetitive elements due to misannealing. We perform our oligo extensions with two oligos as individual reactions, thus, minimizing this issue. Furthermore, we can split a gene to reduce the number of repetitive elements. By performing GGA, we instead shift the limiting factor to the GGA efficiency which is heavily affected by fragment number.³¹

Taken together, using simple design tools, we demonstrate the construction of synthetic genes of up to 456 bp with repetitive elements that are not possible for common vendors to manufacture through their standard platform due to high synthesis complexity scores. We report that the assembly fidelity can drop down to 12.5% in some cases; however, this should not be a concern in cases where gene synthesis is otherwise impossible due to high vendor complexity scores. In addition, the efficiency levels of our method are financially viable and nonlaborious even at higher synthon numbers. We did not go beyond 456 bp, as we already saw a large decrease in fidelity due to an increased synthon number. However, in RNA nanotechnology and RNA origami, in particular, when designing larger structures, there is more flexibility for the placement of repetitive motifs, which can make them easier to synthesize.

While this method was developed to facilitate the construction of genes encoding synthetic noncoding RNA devices and nanostructures with repetitive elements for applications in RNA synthetic biology and RNA nanotechnology, we also envision it as an important addition to the molecular cloning toolbox.

Materials and Methods

RNA Construct Design

2x(Cas6f-hp-MS2wt), 8xCas6f-hp, and 8xMS2 were all designed using the NUPACK design package. The secondary structures together with the sequences for the hairpins were used as design constraints. Linker sequences were designed as nonrepetitive sequences. The design with the lowest ensemble score was chosen for synthesis. 2H-3xCorn, 2H-3xBr, PX-Tri-P6Ba-UU, dAF14, and dAF10 were all designed using the ROAD¹² software and method for RNA origami design.

Design of Fragments and Oligos

Genes of interest are divided into 70–80 bp fragments in silico by using Benchling. Compatible 4-bp overhangs and initial primers with restriction site are designed with the Golden Gate assembly tool from Benchling. Initial designed oligos are then manually extended in the Benchling sequence editor to overlap by 18–25 bp for optimal annealing. Oligos are kept at a maximum length of 60 bases.

Oligo Extension for dsDNA Synthesis

DNA oligos (Integrated DNA technologies) for synthons were ordered to be resuspended in 1x IDT TE buffer. Synthons were assembled by oligo extension using Q5 DNA Polymerase (NEB) in 25 μL reactions consisting of 1x Q5 reaction buffer, 200 μM of each dNTP, 0.5 unit of Q5 DNA polymerase, and 200 nM oligos (100 nM of each). Thermocycling was performed as following: initial denaturation at 98 °C for 30 s, then six cycles of: denaturation at 98 °C for 10 s, annealing at synthon-dependent temperatures for 20 s, elongation at 72 °C for 10 s, ending with a final elongation at 72 °C for 2 min. Synthons were purified with the Nucleospin Gel and PCR cleanup kit (Macherey-Nagel) following the manufacturer’s protocol using a vacuum manifold. Synthon concentration was measured with a Denovix-11. Purified synthons were diluted to 100 nM.

Golden Gate Reaction

Golden Gate reactions were performed with equimolar amounts of DNA using 25 or 50 fmol of synthon DNA, 25 fmol of vector DNA, 0.5 μL of T4 DNA ligase (NEB), and 0.5 μL of either Esp3I (NEB) or BsaI (NEB) in 1 × T4 DNA ligase buffer with 10 μM ATP (NEB) in 10 μL reactions. Golden Gate reactions consisted of 10 min at 37 °C, followed by 15 cycles of 5 min at 37 °C and 5 min at 16 °C followed by heat-inactivation of the enzymes by a 5 min incubation at 50 and 80 °C.

The whole reaction was transformed into NEB Turbo cells following standard protocols, and cells were plated on LB-agar plates containing either 100 μg/mL carbenicillin or 34 μg/mL chloramphenicol. Eight to 10 nonfluorescent colonies for each construct were picked for plasmid propagation. Plasmids were propagated in cultures of terrific broth medium with appropriate antibiotics overnight at 37 °C and purified with a NucleoSpin miniprep kit (Macherey-Nagel) following the manufacturer’s prototol. Plasmids were verified by Sanger sequencing (Eurofins Genomics).

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acssynbio.3c00665.

• Figure S1 and Tables S1–S6: Secondary structure diagrams and sequences of the RNA structures, synthesis status for gene fragments at vendors tried, summary on assembly fidelity (Integrated DNA Technologies), list of DNA and oligo sequences used in this study with accompanying Benchling links (PDF)

• Figure S2: Sequence alignments of synthetic gene assemblies (PDF)

Author Contributions

M.N. conceived the method, designed experiments, and analyzed the data. M.N., M.G., N.S., and G.P. performed experiments. E.S.A. and G.P. supervised. All authors discussed and wrote the paper.

The authors declare no competing financial interest.