Seamless Assembly of Biological Parts into Functional Devices and Higher Order Multi-Device Systems

A new method is described for the seamless assembly of independent, prefabricated and functionally tested blunt-end, double strand nucleic acid parts (DNA fragments) into more complex biological devices (vectors) and higher order multi-device systems. Individual parts include bacterial selection markers, bacterial origins of replication, promoters from a variety of different species, transcription terminators, shuttle sequences and a variety of “N” and “C” terminal solubility/affinity expression tags. Pre-assembly modification of parts with DNA modifying enzymes is not required. Seamless assembly of multiple parts is accomplished in a single step using a specialized thermostable enzyme blend in about 30 minutes. Combinatorial assembly of parts is an inherent feature of the new process, substantially simplifying device and system optimization. To underscore the utility of the new process, parts were assembled into several protein expression devices in order to identify the optimal expression construct for a model target gene, as an example of the utility of the assembly process, and a higher order multi-device system is also described, for the over-expression of a four-enzyme bio-synthetic pathway, and optimized for end-product accumulation in E. coli as a paradigm for how this assembly process could be used to address the assembly of more complex biological pathways.

experimental design and significantly slow the process of identifying optimal devices and systems.
Providing prefabricated and functionally validated parts to researchers without the need for retrieval from destination vectors, combined with a seamless protocol conducive to combinatorial assembly of parts into devices and higher order systems, would represent a significant improvement in synthetic biology molecular cloning. This paper describes such a system. Prefabricated parts are provided such that a wide variety of devices can be rapidly assembled. Appropriately chosen parts are combined and assembled in a single-tube reaction creating molecules that transform/transfect and properly function as devices in E. coli, mammalian and yeast (S. cerevisiae) cells. Multiple assembles can be performed in parallel to generate a collection of unique devices that can be used in combination to optimize novel biological systems. The utility of this technology, referred to as "SureVector" (SV), was validated by assembling parts into a collection of devices (plasmids) in order to perform a protein expression screen to identify the best expression tag combination for a model gene of interest (GOI). To further demonstrate the applicability of the SV process, a multi-device system was designed and constructed to reconstitute a four-enzyme biosynthetic pathway in E. coli. This multi-device system enabled the identification of several over-expressing bacterial clones within one week. and allow specific assembly of parts into devices (example shown in Table 1) and ultimately systems.

XP1 and XP2 Expansion Parts
XP1 parts contain either the yeast autonomous replication sequence (yARS or 2-micron circle) allowing plasmid replication in S. cerevisiae or a linker derived from a unique nucleotide sequence not possessing a function other than to tether a chosen bacterial origin of replication to an XP2 fragments during parts assembly.
The XP2 parts include either a non-functional tether sequence, as above, or the lacI repressor found in E. coli, or mammalian selection markers or yeast auxotrophic markers, allowing yeast to grow in the absence of an essential amino acid. XP2 parts designed specifically for use in mammalian cells include blasticidin, puromycin or hygromycin selection markers. For yeast, XP2 parts include the auxotrophic markers URA3 and HIS3 and the selection marker for hygromycin resistance. XP2 parts also contain a transcription terminator; either the bovine growth hormone polyA signal (bGH pA) specific for mammalian cells or a rho independent terminator for bacterial transcripts provided by the strong hairpin forming sequence (5'-GCCGCCAGCGGAACTGGCGGC-3'). These terminator sequences are positioned and oriented within the XP2 part to terminate transcripts from an upstream GOI.

Large Scale Production and Purification of SV Parts
Large scale parts synthesis was performed by PCR in 96-well plates using sequence verified master plasmid templates. Each reaction contained 1 ng of plasmid template, 1x Herculase II reaction buffer (Agilent Technologies, Inc.), 0.25 mM of each dNTP, 0.4 µM of each PCR primer and 2 µl of pre-formulated Herculase II enzyme (Agilent Technologies, Inc.) in a final volume of 100 µL. Thermocycling conditions were: 1 cycle at 95 ºC for 2 min.; 30 cycles at 95 ºC for 20 sec., 55 ºC for 20 sec. and 72 ºC for 30 sec.; 1 cycle at 72 ºC for 3 min. Contents of multiple 96-well plates for each SV part were pooled and purified using AMPure XP magnetic beads according to the manufacturer's instructions (Beckman-Coulter). Correct lengths and purities of SV parts were assessed with an Agilent Technologies, Inc. BioAnalyzer.

General Assembly of SV Parts into Devices
SV parts designed to assemble into a desired device were combined with a SV adapted GOI part and other reaction components were added as follows: 1x SureVector reaction buffer, 0.25 mM of each dNTP, 5.0 nM of each part (e.g. SM + OR + XP1 + XP2 + T + GOI + P) and 1 µL of pre-formulated enzyme in a final volume of 20 µL. Thermocycling of these components consisted of 1 cycle at 95 ºC for 2 min.; 8 cycles at 95 ºC for 20 sec., 55 ºC for 20 sec. and 68 ºC for 30 sec.; 1 cycle at 68 ºC for 3 min. Following thermocycling, one unit of Dpn I restriction enzyme was added to the reaction and incubated for 5 minutes at 37 °C. One µL of the reaction was transformed into XL1-Blue Supercompetent E. coli cells (Agilent Technologies, Inc.) according to instructions and varying amounts (10, 20 and 50 µL) of the transformation mixtures were spread onto LB agar plates containing the appropriate antibiotic and incubated at 37 ºC until colonies were easily visualized (12 -16 hrs.). Device DNA was purified from select colonies and either analyzed by restriction digestion, or sequenced to verify correct parts assembly, or used directly in downstream processes.

SV Parts Assembly into Nedd5 Protein Expression Devices
To demonstrate an obvious use of the new SV cloning method, the human Nedd5 gene was chosen as a model GOI for performing an expression screening experiment. Nedd5 is a mammalian septin known to associate with actin-based structures such as the contractile ring and stress fibers and is involved in the process of cytokinesis in human brain tumors (11), although the specific nature of Nedd5 is not pertinent to this paper. The Nedd5 gene containing a start and a stop codon was adapted by PCR to be SV compatible by using the primers listed below: N-terminal forward primer 5'-GGTGGCGGAGGTTCTGGAGGCGGTGGAAGTATGGGATCCATGTCTAAGCAACAACCAACTC-3' and N- The Nedd5 gene containing a start codon but lacking a stop codon was adapted by PCR to be SV compatible by using the following primers: C-terminal forward primer:

SV Nedd5 Expression Devices Screening
Nedd5 clones with different expression tags were cultured overnight at 37ºC with shaking at 250 rpm in 1 ml of LB broth containing ampicillin (50 µg/ml). The following day, 10 ml cultures of LB broth containing ampicillin (50 µg/ml) were inoculated with these clones and incubated at 37ºC with shaking at 250 rpm until the OD 600, a measure of bacterial growth, reached 0.6 (approximately 1 hour). Protein expression from the P Tac promoter was induced by the addition of IPTG to a final concentration of 0. ctcgaggagatattgtacactaaaccaaatgTCAGGCCTTGATGGCTTTCAATAC Two sets of bi-cistronic devices were designed and assembled, one containing the ribA and ribD genes and the other containing the ribB and ribE genes. A ribosome binding site (RBS) was included in the 3' region of the ribA and ribB genes downstream of their native stop codon and the same sequence was also included in the 5' region upstream of the ATG start codon of the ribD and ribE genes. This RBS sequence was used as the overlap by which the ribD and ribE genes were positioned downstream of ribA and ribB genes, respectively. The intended outcome was to place two rib genes under control of one promoter and couple expression of the upstream and downstream rib genes via a second RBS between the two rib genes. This was done to attempt to balance expression levels of the individual rib genes. This second RBS overlap region between the rib genes was designed in such a way that the downstream rib gene was not in the same reading frame as the upstream rib gene thus preventing two gene products in the same device from becoming physically linked. An additional stop codon was also added to each upstream rib gene to further guard against translation read-through. Bi-cistronic vectors of this type have been used previously for preparation of nuclear receptor partners RAR and RXR (12) and for the analysis of NFΦB p50/p65 heterodimer (13). Devices lacking either the upstream or downstream rib gene parts were correctly assembled into circular molecules using 90 bp Nterminal and C-terminal "Non-Coding" linker parts NC-N and NC-C, respectively, and were made by overlap extension (14): N-terminal "NC" replaces ribA or ribB parts.
rib -"NC" N-term_Forward ggtggcggaggttctggaggcggtggaagtgaaactgcactcatcgtccctcgaggagct rib -"NC" N-term_Reverse gaaattgttaaattatttctagattcgaagagctcctcgagggacgatgagtgcagtttc C-terminal "NC" replaces ribD or ribE parts. rib -"NC" C-term_Forward ttcgaatctagaaataatttaacaatttcacataaaggaggtatagacagcatacgagtc rib -"NC" C-term_Reverse ctcgaggagatattgtacactaaaccaaatgactcgtatgctgtctatacctcctttatg Bi-cistronic devices that only contain a single rib gene required either a NC-N or NC-C part in lieu of the corresponding rib gene part. Three standard parts were used in both sets of rib devices; the T7 promoter-HIS 6 , XP1 linker and XP2 lacI. These 3 standard parts were used in various combinations with either the ampicillin or kanamycin selectable markers and the pBR322 or p15a bacterial origins of replication. Therefore, all SV rib devices were assembled from just seven SV parts. A total of 18 device level plasmids were constructed. These system level devices were designated by letter-number codes. "K" devices consisted of the kanamycin resistance marker (kan), the p15a origin of replication and either zero, one or two rib genes. "A" devices consisted of the ampicillin resistance marker (amp), the pBR322 origin of replication and either zero, one or two rib genes ( Table 2). Higher order systems were created using various combinations of these device level plasmids by the co-transformation of two devices. For example, devices K6 and A7 resulted in co-expression of ribA-ribD genes from the K6 device and ribB-ribE genes from the A7 device. Combinations of devices were transformed into Agilent BL21(Gold) DE3 E. coli and spread onto LB-agar plates containing 100 µg/ml each of kanamycin and ampicillin (LB-kan-amp) plus 0.5 mM IPTG. Plates were incubated at 37°C for 12 to 18 hours and examined under UV light to identify DMRL expressing clones (systems) as evidenced by fluorescent colonies surrounded by fluorescent halos.

Validation of DMRL System Synthesis -Assay, Purification and MS Analysis
Monitoring clones for DMRL synthesis on agar plates was straightforward as the DMRL fluoresces within colonies and is secreted into the surrounding media creating fluorescent halos around DMRL positive colonies. DMRL production from clones cultured in liquid media was also straightforward as the compound possesses a characteristic visible light absorption spectrum with λ max OD 490 (15) that can be measured in cell free supernatants. Validation of DMRL synthesis required its production, purification and analytical characterization. Clone K6A7 produced pronounced fluorescent colony-halos and was chosen for this purpose. A single colony was inoculated into 3 ml of LB-kan-amp liquid media and incubated overnight at 37°C with shaking at 250 rpm. A 2.0 ml sample of this culture was inoculated into a 500 ml Erlenmeyer flask containing 100 ml of LB-kan-amp and incubation continued as before until the OD 600 value reached 0.35. IPTG was added to a final concentration of 0.5 mM and incubation continued for an additional 18 hours. Cells were removed by centrifugation and acetic acid added to the supernatant to a final concentration of 5%. This sample was applied to a 2.5 x 3.5 cm column of Florisil (Sigma-Aldrich) equilibrated with 5% acetic acid. The column was washed with one liter of 5% acetic acid and DMRL eluted with 100 ml of 3% pyridine. Solvent was removed by evaporation and the residue suspended in 5 ml of water. This sample was applied to a 2.5 x 35 cm column of chromatography grade cellulose (Sigma-Aldrich) equilibrated with water and DMRL was eluted with water. Forty milligrams of DMRL were recovered from 0.8g of cells

Quantities and Rates of DMRL Synthesis from Devices and Systems
Single colonies from devices and systems listed in Table 2 were purified by re-streaking onto fresh LB-kan-amp plates without IPTG. Three colonies from each plate were cultured overnight at 37ºC with shaking at 250 rpm in separate tubes containing 3 ml of LB-kan-amp. The next day, one hundred microliters of each culture were added to 4.9 ml of LB-kan-amp and incubated until

Design Features of SV Parts and Assembly Process
The key design features of SV parts, ensuring precise and ordered joining, are the unique 30 bp sequences incorporated into each PCR primer used to generate a part (see Materials and Methods section). Figure 1 schematically represents how seven SV parts align and overlap due to this design feature.
Assembling parts listed in Table 1  The assembly mechanism of linking parts into higher order devices is represented in Figure 2.
Parts are denatured and adjacent parts anneal due to the 30 bp overlaps. Exposed 3'-OH ends are partially extended by a polymerase resulting flaps that are digested by an endonuclease and covalently joined by a ligase.

Pathway Construction
DMRL pathway construction was accomplished by first assembling SV parts into devices (Table   2) containing zero, one and two ("bi-cistronic") rib genes. PCR primers used to amplify rib gene  (Table 2). Ribosome binding sites (RBS) were designed and incorporated into the 3'and 5' flanking regions of the rib genes and subsequently used as unique overlaps between the genes such that ribD and ribE genes were positioned downstream of the ribA and ribB genes, respectively. The intended outcome of placing two rib genes under control of one T7 promoter and coupling expression of the upstream and downstream rib genes was balanced expression levels. The RBS placed in the 5' regions of both ribD and ribE genes promoted downstream rib gene translation efficiency. The RBS overlap between the upstream and downstream rib genes was designed so that the downstream rib genes were out of frame with the upstream rib genes thus preventing two gene products in the same device from becoming physically linked. An additional stop codon was also added to each upstream rib gene as an added prevention to translation read-through. Bi-cistronic devices of this type have been used previously for preparation of nuclear receptor partners RAR and RXR (13) and for the analysis of NFΦB p50/p65 heterodimer (14). SV devices lacking one or both rib genes were also assembled using N-terminal and C-terminal "Non-Coding" parts; NC-N and NC-C, respectively (Table 2). Zero or single rib gene control devices required either an NC-N or NC-C part, or both in the case of the double negative control, in lieu of a rib gene part. Common SV parts used in both sets of rib devices were the T7 promoter-HIS 6 , XP1 linker and XP2 lacI (allowing IPTG induction of rib gene devices expression) in addition to various combinations of ampicillin and kanamycin selectable markers with pBR322 or p15a bacterial origins of replication. Therefore, all 18 SV rib device plasmids were assembled from just 7 seven SV parts, not including the rib gene parts. Table 2 devices were transformed either individually or in various co-transformation scenarios (Table 3)

Characterization of DMRL Synthesized by SV rib System K6A7.
DMRL synthesized and purified from system K6A7 ( Table 3)

Quantities and Rates of DMRL Synthesis from DMRL Devices and Systems
Figure 7a confirms that only SV systems K6A7, K7A6, K8A9, K9A8, each containing two rib gene devices expressing all four rib genes, produced DMRL. Clones K6A7 and K7A6 produced the highest levels of DMRL and most rapidly (Figures 7a and b). These systems are composed of devices K6 and A6 expressing enzymes from ribA and ribD genes, respectively, both in the NCR path. ribA is positioned immediately downstream of the T7 promoter. Devices K7 and A7 express enzymes from ribB and ribE genes, respectively, with ribB in the SCR path and ribE in the CCR path. ribB is positioned immediately downstream of the T7 promoter. Total DMRL accumulation and slower synthesis rates are similar in systems K8A9 and K9A8. These systems are composed of devices K8 and A8 expressing enzymes from ribA and ribE genes, respectively, with ribA in the NCR path and ribE in the CCR path. ribA is positioned immediately downstream of the T7 promoter. Systems composed of devices K9 and A9 express enzymes from ribB and ribD genes, respectively, with ribB in the SCR path and ribD in the NCR path. ribB is positioned immediately downstream of the T7 promoter. How these system configurations resulted in different totals and rates of DMRL synthesis were not examined.

Discussion
A new method (SureVector) for seamless assembly of biological parts into functional devices and higher order systems is presented that takes advantage of principles learned from prefabrication engineering design and assembly. These are: (1) Functional DNA parts are manufactured and quality controlled at a location away from the assembly site (Agilent Technologies, Inc.; (2) Parts, other requisite assembly materials and detailed assembly instructions are delivered to the site of construction (the research laboratory) for assembly; (3) from a synthetic biology standpoint, processes (1) and (2) enable rapid and reliable combinatorial assembly of desired devices destined for introduction into E. coli, mammalian and yeast cells.
The salient features of this new method were demonstrated by constructing a set of protein expression devices to identify the best device for the expression of a target GOI (Nedd5) and a higher order system was constructed to recreate a four-gene biosynthetic pathway system (DMRL). The combinatorial assembly power of this process was utilized in developing the DMRL expression system. A total of 18 protein expression devices were assembled in one day possessing zero, one or two rib gene (bi-cistronic) parts. Expression screening was initiated the following day and positive systems were selected for structural and functional testing in less than one week. While assembled devices were being sequenced for validation of their structural integrity, DMRL was purified and characterized by mass spectrometry. Experiments were also performed to determine optimal DMRL synthesis rates and maximum production. The same type of project out sourced to a third-party vendor would have taken several months to complete (previous experience). While optimizing DMRL production was not the goal of these experiments, it is worth noting that 40 mg of DMRL was purified from 100 ml of media collected after growth of 0.8 g (wet weight) of E. coli clone K6A7. In stark contrast, Maley and Plaut (15) required 5 kg of the mold A. gossypii to obtain 160 mg of DMRL.
The combinatorial power, simplicity and assembly accuracy of the SureVector process will facilitate building many "multi-device" systems including unique biochemical synthetic pathways and novel regulatory circuits. Production of fine chemical intermediates and endproducts represents an obvious high-value application.         Table 2) were cotransformed into Agilent BL21(Gold)DE3 E.coli and spread onto LB-kan-amp-IPTG plates.
Resulting colonies were examined under unfiltered UV light. All resulting K6A7 systems contained four rib genes and produced DMRL as evidenced by fluorescent colonies with fluorescent halos. Control (device K5A5) through three rib genes (see Table 2) did not produce DMRL as no fluorescent halos were detected.

Figure 7 (a) Total Quantites of DMRL Synthesis from Devices and Systems.
Single colonies from devices and systems listed in Table 3 were purified by re-streaking onto fresh LB-kan-amp plates without IPTG. Three colonies from each plate were cultured in separate tubes containing 3 ml of LB-kan-amp. One hundred microliters of each culture were added to 4.9 ml of LB-kan-amp and incubated until OD600 reached between 0.3 and 0.5. IPTG was added to a final concentration 0.5 mM and incubation continued for three hours after which OD600 values were re-measured. Cultures were centrifuged to remove cells and OD409 values of supernatants obtained. OD409 values were normalized relative to OD600 values measured post IPTG addition and the resulting numbers compared. DMRL was synthesized exclusively by systems containing two devices each expressing two rib genes -systems K6A7, K7A6, K8A9 and K9A8.   Table 2. SV Devices Required to Re-Create the DMRL Biosynthetic Pathway. Each device contains zero, one or two rib genes, and either the kanamycin resistance marker and p15a origin of replication prefabs (labeled "K") or the ampicillin resistance marker and pBR322 origin prefabs (labeled "A"). Single rib gene control plasmids (K1 -K4; A1-A4) require either N-or Cterminal "non-coding" parts (NC-N and NC-C, respectively) to assemble complete devices. Zero rib gene control devices (K5 and A5) contain both NC-N and NC-C parts in place of both rib genes to assemble into complete devices. Co-transformation of one "K" device (e.g. K6 through K9) with one "A" device (e.g. A6 through A9) results in a higher order system expressing either 0, 1, 2, 3 or 4 rib genes.   Table 2. SV Devices Required to Re-Create the DMRL Biosynthetic Pathway. Each device contains zero, one or two rib genes, and either the kanamycin resistance marker and p15a origin of replication prefabs (labeled "K") or the ampicillin resistance marker and pBR322 origin prefabs (labeled "A"). Single rib gene control plasmids (K1 -K4; A1-A4) require either N-or Cterminal "non-coding" parts (NC-N and NC-C, respectively) to assemble complete devices. Zero rib gene control devices (K5 and A5) contain both NC-N and NC-C parts in place of both rib genes to assemble into complete devices. Co-transformation of one "K" device (e.g. K6 through K9) with one "A" device (e.g. A6 through A9) results in a higher order system expressing either 0, 1, 2, 3 or 4 rib genes.