GenBank file upload.

To start the partitioning design process, the user first specifies a single GenBank file for server upload at the web interface front page. The input sequence must be smaller than 10 Mb in file size. Splitting of the DNA design into smaller sequences is recommended when processing larger input files. The Genome Partitioner algorithm is primarily intended for the partitioning of prokaryotic genome-scale DNA constructs. However, DNA designs derived from eukaryotic sequences can also be processed as long as no discontinuous features or long interspersed repeat sequences are present.

Partitioning parameter settings.

After upload, a new graphic interface appears (Fig 2) where the user defines partitioning specific parameters. Direct and inverted repeat sequences and non-unique sequence stretches within terminal homology regions can lead to misalignment and imprecise end joining of assembly units resulting in short sequence deletions, duplications or formation of non-desired assembly side products. The Genome Partitioner algorithm uses build in rules to generate terminal homology regions between adjacent assembly units that are devoid of sequences that impede recombination-based homologous end joining. Build-in parameter settings are used for the detection of sequence features such as direct repeats, hairpin structures and non-unique sequence stretches that occur in multiple terminal homology regions (Table 1). For each assembly tier, the user can specify the maximal size in base pairs for segments (input field 1), blocks (input field 5) and subblocks (input field 9). This is in particular useful for optimization of synthesis costs, when restrictions in sequence length apply. The maximal unit size is defined as the partition sequence plus flanking 5’ prefix and 3’ suffix adapter sequences derived from current and subsequent higher order assembly tiers. Further, the size of terminal homology regions (termed 'overlaps') can be specified for segments (input field 2), blocks (input field 6), and subblocks (input field 10). Overlaps of 25 bp are sufficient at the lowest tier where in vitro assembly methods are preferentially used for homologous end joining of subblocks. At higher tiers of block and segment assembly where yeast homologous recombination is the method of choice, preset values for overlaps are 80 bp and 120 bp respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Overview of the Genome Partitioner web interface.

In the left panel, the parameter input interface allows the user to customize partitioning optimization criteria, to specify size and homology overlaps of assembly units, and define preset or customized adaptor sequences. The user can optionally generate specific PCR primers to test for correct assemblies at the block, segment and genome level and select between an equidistant partitioning and an advanced mode where the partitioning at the segment level follows the boundaries of biological parts. The right panel provides a flowchart illustrating the four-tier assembly workflow. Assembly units of increasing size are formed using overlap-based DNA assembly methods.

https://doi.org/10.1371/journal.pone.0177234.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. List of the built-in optimization parameters used by the Genome Partitioner algorithm.

https://doi.org/10.1371/journal.pone.0177234.t001

Design of adapter sequences.

Further, the user can select preset or specify customized 5’ prefix and 3’ suffix adapter sequences that permit the integration of assembly units into corresponding destination vectors (input fields 3,4,7 and 8). Adapter sequences also contain recognition sequences for type IIS restriction enzymes that facilitate the release of assembly units from cloning vectors. The predefined adaptor sequences contain all necessary type IIS recognition sites and can be used for the assembly of blocks into pXMCS-2 and segments into pMR10Y cloning vectors. At the level of subblocks, a single type IIS recognition site such as the predefined adaptor for BbsI is sufficient (input fields 11 and 12) for the release of subblocks from plasmid cloned DNA. If subblocks are provided in form of linear DNA, no 5' prefix and 3' suffix adaptor sequences are required and input fields 11 and 12 remain empty. When designing adapter sequences, the user must select restriction sites that are absent in the DNA construct. The predefined adaptor sequences are BbsI for subblocks, BspQI for blocks, PacI and PmeI for segments. To assist in the removal of internal restriction sites, genome-scale DNA design tools such as the Genome Calligrapher web tool [13] can be used (https://christenlab.ethz.ch/GenomeCalligrapher).

Primer design for PCR verification of assemblies.

In input field 13, the users can optionally generate sets of diagnostic PCR primers that allow for the testing of correctly assembled constructs and sequence verification across assembly junctions. The Genome Partitioner algorithm calculates suitable PCR primer-pairs surrounding the assembly junctions of segments, blocks and subblocks. For each assembly junction, forward and reverse primers, each 20 bp in size, are generated that anneal within a pre-defined search window of 150 bp flanking the terminal homology region. PCR primers are analyzed to define suitable primer pairs that pass the following design-rules: 1) Each primer pair must fall within a predefined melting temperature interval (∆Tm < 1°C). 2) The sequence of the last 8 bases from the 3’ end of each primer must be unique within the search window. 3) Primer pairs show no homologies and self-homologies. 4) Each primer must be devoid of hairpins and repeats larger than 4 bases. For each assembly junction, the first pair of primers passing the design rules is written into the corresponding output files segment_primers.txt, block_primers.txt and subblock_primers.txt. If no suitable primer-pair is found, the program will increase the search window by 50 bases followed by increasing the annealing temperature range by 1°C using a maximum of six iterations. If still no primer-pair is found, then the unresolved region is reported in the output file for separate processing. A statistics output of the primer design steps is written in the log file (logfile.txt), showing the number of discarded primer pairs and a description of the optimization problems encountered.

Annotation of biological parts.

In addition to the default mode, which produces equidistant partitioning designs at the segment level, partitioning can be performed on advanced mode where segment boundaries follow boundaries of biological parts (input field 13). The advanced mode prevents splitting of biological parts into different segments, which is particularly useful for grouping together biological functions on the level of segments. Partitioning in the advanced mode requires biological parts to be annotated using the feature type 'source' of the GenBank format.

FEATURES             Location/Qualifiers

            source           1..182

                                /mol_type = "genomic DNA"

                                /note = "synthetic part:1"

                                /organism = "Caulobacter syntheticus v3_19”

            source         183..2484

                                /mol_type = "genomic DNA"

                                /note = "synthetic part:2"

                                /organism = "Caulobacter syntheticus v3_19"

            source         2485..4847

                                /mol_type = "genomic DNA"

                                /note = "synthetic part:3"

                                /organism = "Caulobacter syntheticus v3_19"

Partitioning algorithm.

The partitioning algorithm uses a four steps process. In the first step, segments not exceeding a user-defined size limit are generated. The algorithm restrains the variation of the segment sizes to less than 10%. Each segment shares terminal homology regions (called THR hereafter) to adjacent segments. The length of THRs at the segment level is user-defined (in the range of 35 to 200 bp). Segments are flanked by user-defined, segment specific 5’ and 3’ adapter sequences.

In the second step, all segments are further subdivided into blocks with a user-defined size limit (in the range of 2000 to 5000 bp). The size limit of blocks is set not to exceed a size variation larger than 10%. To maximize the size of each block, each segment is divided into the least number of blocks. If necessary, the size of blocks is adjusted to accommodate for subsequent addition of 5’ and 3’ adapter sequences. Blocks that overlap with termini of segments also carry the appropriate adapter sequences of segments. Each block shares THRs with adjacent blocks. The length of THRs at the block level is user-defined (size range between 15 to 200 bp).

For all block THRs of a given segment, the Genome Partitioner algorithm analyses the occurrence of unfavorable secondary structures and sequence patterns prone to interfere with the recombination based DNA assembly process. In a first optimization routine, hairpins and direct repeats are identified within THRs. The maximal length of direct and indirect repeats within THRs must be smaller than 8 bp. If larger repeats are detected, the algorithm generates alternative block partitioning variants with terminal homology sequences shifted up or downstream of the original partitioning site. In a second optimization routine, the algorithm searches for identical substrings occurring multiple times (i.e. non-unique sequences) within THRs of DNA blocks from each segment. Non-unique sequences within THRs larger than 8 bp are removed by generating a set of partitioning variants that no longer include non-unique sequence substrings. These partitioning design variants are iteratively evaluated for occurrence of repeat, hairpins and non-unique sequence substrings within multiple THRs. A metric is used to identify the optimal partitioning design variant that i) shows absence of repeats and ii) is unique in sequence and iii) requires the least repositioning of THR regions. The THR optimization procedure follows a non-recurring search path until repeat sequences and non-unique sequence substring larger than 8 bp have been removed. If repeat sequences, hairpins and non-unique sequence substrings remain after 20 iteration cycles, the last position of THRs is written in the output file.

In the third step, each block of the partitioning design is further subdivided into subblocks. Subblocks deviate by less than 10% from a user-defined size limit (in the range 500 to 2000 bp) including the length of any adapter sequences. To maximize the size of each subblock, each block is divided into the least number of subblocks. If necessary, the size of subblocks is adjusted to accommodate for subsequent addition of 5’ and 3’ adapter sequences. The length of terminal homology regions at the subblock level is user-defined (in the range of 15 to 200 bp). THRs of DNA subblocks are optimized according to the same procedure as used for the THR optimization of blocks. Similar to blocks, subblocks harbor adapter sequences for higher-order assembly. Subblocks that correspond to termini of segments carry composite adapter sequences composed of the innermost segment adapters, followed by the block adapters and the outermost subblock adapter sequences. Similarly, subblocks that upon assembly form the termini of blocks carry composite adapters composed of inner block adapter and outer subblock adapter sequences. All other subblock types only carry subblock adapter sequences. The addition of composite adapter sequences permits release and subsequent assembly of subblocks into higher order assembly units.

In the fourth step, sequences of segments, blocks and subblocks including 5’ and 3’ adapter sequences are written into FASTA files and annotated into the GenBank output file. While smaller DNA constructs, less than 100 kb in size, are partitioned within seconds, larger genome designs may take a few minutes to process. During the partitioning process and overlap optimization, a progress bar shows the percentage of completed assembly units for each partitioning level.

Download of partitioning results in FASTA and GenBank file format.

After completing the partitioning optimization, the user is relocated to a graphical output interface where FASTA files for assembly units and a GenBank file with the annotated partitioning design can be downloaded (S2 Fig). Downloaded FASTA sequence files can be directly submitted to a commercial provider of de novo DNA synthesis. In addition, partitioning parameter settings, statistics and primer files for the verification of DNA assemblies are provided for download. A graphic map of the generated four-tier assembly design is shown. A table summarizes the size range of assembly units and the effective DNA synthesis effort needed for each assembly tier. Furthermore, a comparison of assembly constraints prior and after the optimization is shown (S2 Fig).

PCR amplification of subblocks.

Subblocks were ordered from a commercial provider of de novo DNA synthesis (Gen9, Inc. Cambridge, MA, USA) and provided as sequence confirmed plasmid cloned constructs. Subblock sequences were PCR amplified from maintenance vectors (pG9m-2) in a 25 μl reaction volume using 0.25 μl (2.5 u) Phusion® high-fidelity DNA polymerase (New England Biolabs, USA), 5 μl 5x Phusion® HF reaction buffer (NEB), 0.25 μl (~ 25 ng) plasmid template DNA, 0.125 μl 100 μM forward primer (#1) (S1 Table), 0.125 μl 100 μM reverse primer (#2), 2.5 μl dNTPs (2 mM each) (Thermo Fisher Scientific Inc., USA), 0.75 μl DMSO (Fisher Scientific, UK), and 16 μl ddH₂O. The PCR reaction was run on a BIORAD S1000 thermal cycler (Bio-Rad Laboratories Inc., USA) using the following protocol. (1) Initial denaturation 3:00 min at 95°C, (2) denaturation 30 s at 95°C, (3) primer annealing 30 s at 58°C, (4) elongation 1:30 min at 72°C, (5) repeat steps 2–4, 25 times, (6) final elongation 5 min at 72°C.

Digestion of subblocks and pXMCS-2 target vector.

To release subblocks from adapter sequences, PCR products were digested with BbsI type IIS restriction enzyme. Digestion reactions contained four times 2.5 μl of each of the four subblocks directly taken from the PCR reaction mixture, 0.5 μl (5 u) BsbI type IIS restriction enzyme (NEB, USA), 2 μl 10x NEBuffer 2.1 (NEB, USA), and 7.5 μl nuclease-free H₂O (Promega, USA). The digestion reactions were incubated at 37°C overnight and subsequently column purified using the NucleoSpin® Gel and PCR clean up Kit (Macherey-Nagel, Switzerland). The pXMCS-2 maintenance vector [20] was digested with NdeI and NheI-HF restriction enzymes in a 40 μl digestion reaction volume composed of 20 μl (294.4 ng/μl) pXMCS-2, 0.5 μl (10 u) NdeI (NEB, USA), 0.5 μl (10 u) NheI-HF (NEB, USA), 4 μl 10x CutSmart® buffer (NEB, USA), and 15 μl nuclease-free H₂O (Promega, USA). The digestion reaction was incubated at 37°C for 4 h. The digested vector was separated on a 1% agarose gel (UltraPure™ Agarose, Invitrogen, USA) and run for 40 min at 120 V, extracted from the gel and eluted in 20 μl using the NucleoSpin® Gel and PCR clean up Kit (Macherey-Nagel, Switzerland).

DNA assembly of subblocks into blocks.

The BbsI-digested subblocks were assembled into corresponding blocks and integrated in maintenance vector pXMCS-2 (S2 Table) using isothermal assembly reactions consisting of: 4 μl 5x isothermal reaction buffer, 0.008 μl (0.08 u) T5 exonuclease (NEB, USA), 0.25 (2.5 u) Phusion® high-fidelity DNA polymerase (NEB, USA), 2 μl (80 u) Taq DNA Ligase (NEB, USA), 8.742 μl nuclease-free H₂O (Promega, USA), 4 μl (~ 20 ng/μl) of purified BbsI-digested subblock pool, and 1 μl (108 ng) of NdeI and NheI-HF digested pXMCS-2 vector. Ligation reaction mixtures were incubated at 50°C for 1 h. 5 μl of each of the pXMCS-2::block[0–4] assemblies were dialysed on 0.025 μm VSWP MF™ membrane filters (Merck Millipore Ltd., IRL) for 20 min and subsequently electroporated into competent E. coli DH5α (90 μl aliquots, OD ~ 15) at 1.75 kV, 400 Ω, and 25 μF using 0.1 cm electrode gap Gene Pulser® cuvettes (Bio-Rad Laboratories, USA). The pulse was applied at time constants between 8.6 and 8.8 ms. Immediately after the electroporation, transformed DH5α cells were rescued in 1 ml SOC medium and incubated at 37°C for 1 h. 100 μl of each rescued cell sample cells was plated onto selective LB medium supplemented with kanamycin (20 μg/ml) and incubated at 37°C overnight.

Verification of block assemblies by PCR.

Correct block assemblies were verified by PCR of subblock junctions using primer sets generated by Genome Partitioner algorithm. PCR reactions were performed in 20 μl reaction volumes composed of 10 μl 2x GoTaq® G2 Green Master Mix (Promega, USA), 0.5 μl 100 μM forward primer (fw primers of #3–32) (S1 Table), 0.5 μl 100 μM reverse primer (rv primers of #3–32), 1 μl DH5α pXMCS-2::block[0–4] liquid culture, and 8 μl ddH₂O. The PCR protocol consisted of (1) initial denaturation 3:00 min at 95°C, (2) denaturation 30 s at 95°C, (3) primer annealing 30 s at 60°C, (4) elongation 30 s min at 72°C, (5) repeat steps 2–4, 25 times, (6) final elongation 5 min at 72°C.

BspQI-mediated block release.

The pXMCS-2::block[0–4] plasmids were isolated from DH5α using the GeneJET Plasmid Miniprep Kit (Thermo Scientific, USA). Subsequently, blocks were released from the pXMCS-2 backbone via BspQI restriction digestion. Each digestion reaction contained 10 μl (> 5 μg) pXMCS-2::block[0–4] plasmid, 1μl (10 u) BspQI type IIS restriction enzyme (NEB, USA), 4 μl 10x NEBuffer 3.1 (NEB, USA), and 25 μl nuclease-free H₂O (Promega, USA). The digestions were incubated at 50°C for 1.5 h and the reactions was stopped by heat inactivation at 80°C for 20 min. Subsequently, digested constructs were purified over column using the NucleoSpin® Gel and PCR clean up Kit (Macherey-Nagel, Switzerland).

DNA assembly of blocks into segment 25.

Purified pXMCS-2::block[0–4] digestion mixtures were used for assembly of segments into yeast vector pMR10Y (pMR10::ARS::URA3). S. cerevisiae strain VL6-48N (BC3347) was grown in YPD medium supplemented with 80mg/L adenine to OD₆₀₀ of 0.7. 2 ml of the yeast culture was pelleted, washed with 1 ml 0.9% NaCl-solution and pelleted again. 10 μl of fish sperm DNA (10μg/μl, single stranded from salmon testes, D7656, Sigma-Aldrich, USA) was added to the cells. Subsequently, 540 ng of linearized pMR10Y (digested with PacI and PmeI) and 300 ng of each of the five BspQI digested DNA block (block[0–4]) were added to the pellet. After thorough vortexing, the pellet was resuspended in 500 μl transformation mixture composed of 400 μl 50% PEG solution, 50 μl 1M Lithium acetate, 50 μl ddH₂O. Next, 57 μl DMSO were added to the transformation reaction followed by 15 min incubation at RT and 15 min at 42°. Transformed cells were pelleted, resuspended in 100 μl ddH₂O and plated onto yeast drop-out medium lacking uracil, supplemented with glucose (10 g/L), adenine (80 mg/L) and incubated at 30°C for three days till colonies appeared.

Verification of segment assemblies by PCR.

Correct segment assembly was verified by PCR amplifying each block junction using the Genome Partitioner's automatically designed primers. Six colonies were picked and resuspended in liquid yeast synthetic drop out medium lacking uracil and supplemented with glucose (10 g/L) and adenine (80 mg/L). Colony PCR was performed in a 20 μl reaction volume containing 10 μl 2x Phire Green Hot Start II PCR Master Mix (Thermo Scientific, USA), 0.5 μl 25 μM forward primer (fw primers #33–40), 0.5 μl 25 μM reverse primer (rv primers #33–40), 1μl yeast liquid culture, and 8 μl ddH₂O. The PCR protocol consisted of (1) initial denaturation 3:00 min at 98°C, (2) denaturation 5 s at 98°C, (3) primer annealing 5 s at 62°C, (4) elongation 20 s min at 72°C, (5) repeat steps 2–4, 40 times, (6) final elongation 1 min at 72°C.