Limits...
Algorithmic co-optimization of genetic constructs and growth conditions: application to 6-ACA, a potential nylon-6 precursor.

Zhou H, Vonk B, Roubos JA, Bovenberg RA, Voigt CA - Nucleic Acids Res. (2015)

Bottom Line: This is compared to a 64-member full factorial library just varying expression (0.64 Mb of DNA assembly).Statistical analysis of the screening data from these libraries leads to different predictions as to whether the expression of enzymes needs to increase or decrease.This work introduces a generalizable platform to co-optimize genetic and non-genetic factors.

View Article: PubMed Central - PubMed

Affiliation: Synthetic Biology Center, Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA.

Show MeSH

Related in: MedlinePlus

Co-variance of genetic and media factors in a DOE library. (A) The monocistronic architecture is shown, where each gene has its own promoter (arrow), ribozyme (dashed line with circle), RBS (semicircle) and strong double terminator (TT). Two promoters are chosen for each gene that generate a low (−1) and high (+1) expression level. From left to right, these promoters are: (M4/M6), (M4/M6), (M4/M6), (M4/M6), (M1/M4), (M1/M4). The strengths are in arbitrary units of fluorescence and were evaluated by creating a new construct with the promoter at each position fused to mRFP (Materials and Methods). The part sequences are provided in Supplementary Tables S1 and S2. The errors were calculated as the standard deviation of three independent experiments performed on different days. (B) The 29-4 DOE library is shown, rank ordered by the titer. From left to right, the nine factors are shown according to their high/low expression state or presence/absence in the media (+1/−1). The associated construct is shown in the center, following the genetic part coloring and format as in part a. The titer and OD600 of each construct is shown to the right. The error bars were calculated as the standard deviation of two experiments performed on different days. (C) The normalized titers (Materials and Methods) are shown for the 29-4 DOE library (red lines) and the 26 full factorial library that only varies the expression levels (black lines). The full factorial library and screening data are shown in Supplementary Figure S1. The ‘optimal coding’ represents the predicted optimal state of each factor, used to build the optimal construct from the full factorial library (evaluated in Figure 3A).
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4666358&req=5

Figure 2: Co-variance of genetic and media factors in a DOE library. (A) The monocistronic architecture is shown, where each gene has its own promoter (arrow), ribozyme (dashed line with circle), RBS (semicircle) and strong double terminator (TT). Two promoters are chosen for each gene that generate a low (−1) and high (+1) expression level. From left to right, these promoters are: (M4/M6), (M4/M6), (M4/M6), (M4/M6), (M1/M4), (M1/M4). The strengths are in arbitrary units of fluorescence and were evaluated by creating a new construct with the promoter at each position fused to mRFP (Materials and Methods). The part sequences are provided in Supplementary Tables S1 and S2. The errors were calculated as the standard deviation of three independent experiments performed on different days. (B) The 29-4 DOE library is shown, rank ordered by the titer. From left to right, the nine factors are shown according to their high/low expression state or presence/absence in the media (+1/−1). The associated construct is shown in the center, following the genetic part coloring and format as in part a. The titer and OD600 of each construct is shown to the right. The error bars were calculated as the standard deviation of two experiments performed on different days. (C) The normalized titers (Materials and Methods) are shown for the 29-4 DOE library (red lines) and the 26 full factorial library that only varies the expression levels (black lines). The full factorial library and screening data are shown in Supplementary Figure S1. The ‘optimal coding’ represents the predicted optimal state of each factor, used to build the optimal construct from the full factorial library (evaluated in Figure 3A).

Mentions: Gibson assembly was used to build the T7 RNAP expression system shown in Figure 1C (23). MoClo was used to assemble the monocistronic pathway gene expression cassettes into a single backbone plasmid (22). Supplementary Table S1 and S2 list the parts and genes used in this study. The GenBank accession No. for NifV, AksD, AksE, AksF, KdcA and Vfl are P05342 (Azotobacter vinelandii), ABR55899 (Methanococcus aeolicus Nankai-3), ABR56236 (M. aeolicus Nankai-3), ABR57060 (M. aeolicus Nankai-3), AAS49166 (Lactococcus lactis), AEA39183 (Vibrio fluvialis) respectively (optimized codon sequences are provided in Supplementary Table S2). The scheme for DNA assembly from level 0 to level 1 to the final level 2 constructs is illustrated in Supplementary Figure S3. For level 0 parts-containing plasmids, the backbone plasmid pL0 was derived from pUC19. To be golden gate compatible, the BsaI and BbsI sites were removed by introducing silent mutations. Parts including T7 promoter-ribozyme, RBS-CDSs and double terminators were ligated into pL0 by restriction ligation using SmaI and T4 ligase. 10 ng of both insert and backbone plasmid were added to a 10 μl reaction containing 0.5 μl SmaI and 1 μl T4 ligase for incubation at room temperature for 2 h. For level 0 promoter plasmids, the inserts which contain a spacer, one T7 promoter and six different ribozymes were constructed by DNA oligo annealing. The spacer sequences were designed by the Random DNA Generator using a random GC content of 50% (http://www.faculty.ucr.edu/∼mmaduro/random.htm). The different T7 promoter variants were then introduced by primers via inverse PCR. The promoter-ribozyme parts are flanked by two BsaI sites with GGAG and TTAA as four nucleotides overlaps for the subsequent type IIS reaction to build level 1 plasmids. All of the engineered ribozyme sequences end with TTAA. The RBS-CDS constructs containing the enzymes were obtained from a concurrent study (unpublished results), which we mutated to eliminate BsaI and BbsI sites. Note that the RBS for kdcA in the monocistronic design is K005 (designed using the RBS Calculator) as compared to K007, which was used for the operon-based designs in Figure 1B and C. The level 1 plasmids are based on the pL1 backbone, originating from pMJS1CD (24) with Kanamycin resistance (also free of BsaI and BbsI sites). Then, 20 fmol of each level 0 plasmid are mixed with 5 U BsaI (New England Biolabs, #R0539S) and 5 U T4 DNA Ligase (Promega, #M1794) for a total of 10 μl 1× Promega T4 DNA Ligase Buffer and incubated. Two level 1 plasmids carrying expression cassettes with high (+1) and low (−1) expression levels for each of the six pathway genes were built. In total, there are 12 level 1 plasmids for six pathway genes and 12 containing mRFP in the pathway context for part characterization. Inverse PCR (iPCR) was used to generate the level 1 plasmids for the kdcA and aksF libraries using pL1-kdcA-TU2 and pL1-aksF-TU2 as templates. To build the level 2 plasmids, the six pathway cistrons are assembled in the order shown in Figure 2A. The cistrons are PCR amplified from the level 1 plasmids to give each construct specific cohesive ends upon BsaI digestion corresponding to the assigned position. The final expression plasmid backbone pL2 is derived from pAKP444 (unpublished results), where the BsaI sites in β-lactamase gene were eliminated. The seven assembly junction regions were sequence verified by Sanger sequencing. To build the kdcA and aksF libraries shown in Figure 3B and C, only the expression cassette for either kdcA or aksF was changed, while the other gene expression cistrons are the same as the #4 construct. To achieve higher expression, an extra copy of kdcA or aksF was introduced in a separate p15a Kanr plasmid (Supplementary Figure S4C). The T7 promoter M4 is used for both kdcA and askF expression. The same plasmid for overexpressing aksF or kdcA was introduced into the strain containing construct [−1, −1, −1, +1, +1, +1] individually for Figure 3A.


Algorithmic co-optimization of genetic constructs and growth conditions: application to 6-ACA, a potential nylon-6 precursor.

Zhou H, Vonk B, Roubos JA, Bovenberg RA, Voigt CA - Nucleic Acids Res. (2015)

Co-variance of genetic and media factors in a DOE library. (A) The monocistronic architecture is shown, where each gene has its own promoter (arrow), ribozyme (dashed line with circle), RBS (semicircle) and strong double terminator (TT). Two promoters are chosen for each gene that generate a low (−1) and high (+1) expression level. From left to right, these promoters are: (M4/M6), (M4/M6), (M4/M6), (M4/M6), (M1/M4), (M1/M4). The strengths are in arbitrary units of fluorescence and were evaluated by creating a new construct with the promoter at each position fused to mRFP (Materials and Methods). The part sequences are provided in Supplementary Tables S1 and S2. The errors were calculated as the standard deviation of three independent experiments performed on different days. (B) The 29-4 DOE library is shown, rank ordered by the titer. From left to right, the nine factors are shown according to their high/low expression state or presence/absence in the media (+1/−1). The associated construct is shown in the center, following the genetic part coloring and format as in part a. The titer and OD600 of each construct is shown to the right. The error bars were calculated as the standard deviation of two experiments performed on different days. (C) The normalized titers (Materials and Methods) are shown for the 29-4 DOE library (red lines) and the 26 full factorial library that only varies the expression levels (black lines). The full factorial library and screening data are shown in Supplementary Figure S1. The ‘optimal coding’ represents the predicted optimal state of each factor, used to build the optimal construct from the full factorial library (evaluated in Figure 3A).
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4666358&req=5

Figure 2: Co-variance of genetic and media factors in a DOE library. (A) The monocistronic architecture is shown, where each gene has its own promoter (arrow), ribozyme (dashed line with circle), RBS (semicircle) and strong double terminator (TT). Two promoters are chosen for each gene that generate a low (−1) and high (+1) expression level. From left to right, these promoters are: (M4/M6), (M4/M6), (M4/M6), (M4/M6), (M1/M4), (M1/M4). The strengths are in arbitrary units of fluorescence and were evaluated by creating a new construct with the promoter at each position fused to mRFP (Materials and Methods). The part sequences are provided in Supplementary Tables S1 and S2. The errors were calculated as the standard deviation of three independent experiments performed on different days. (B) The 29-4 DOE library is shown, rank ordered by the titer. From left to right, the nine factors are shown according to their high/low expression state or presence/absence in the media (+1/−1). The associated construct is shown in the center, following the genetic part coloring and format as in part a. The titer and OD600 of each construct is shown to the right. The error bars were calculated as the standard deviation of two experiments performed on different days. (C) The normalized titers (Materials and Methods) are shown for the 29-4 DOE library (red lines) and the 26 full factorial library that only varies the expression levels (black lines). The full factorial library and screening data are shown in Supplementary Figure S1. The ‘optimal coding’ represents the predicted optimal state of each factor, used to build the optimal construct from the full factorial library (evaluated in Figure 3A).
Mentions: Gibson assembly was used to build the T7 RNAP expression system shown in Figure 1C (23). MoClo was used to assemble the monocistronic pathway gene expression cassettes into a single backbone plasmid (22). Supplementary Table S1 and S2 list the parts and genes used in this study. The GenBank accession No. for NifV, AksD, AksE, AksF, KdcA and Vfl are P05342 (Azotobacter vinelandii), ABR55899 (Methanococcus aeolicus Nankai-3), ABR56236 (M. aeolicus Nankai-3), ABR57060 (M. aeolicus Nankai-3), AAS49166 (Lactococcus lactis), AEA39183 (Vibrio fluvialis) respectively (optimized codon sequences are provided in Supplementary Table S2). The scheme for DNA assembly from level 0 to level 1 to the final level 2 constructs is illustrated in Supplementary Figure S3. For level 0 parts-containing plasmids, the backbone plasmid pL0 was derived from pUC19. To be golden gate compatible, the BsaI and BbsI sites were removed by introducing silent mutations. Parts including T7 promoter-ribozyme, RBS-CDSs and double terminators were ligated into pL0 by restriction ligation using SmaI and T4 ligase. 10 ng of both insert and backbone plasmid were added to a 10 μl reaction containing 0.5 μl SmaI and 1 μl T4 ligase for incubation at room temperature for 2 h. For level 0 promoter plasmids, the inserts which contain a spacer, one T7 promoter and six different ribozymes were constructed by DNA oligo annealing. The spacer sequences were designed by the Random DNA Generator using a random GC content of 50% (http://www.faculty.ucr.edu/∼mmaduro/random.htm). The different T7 promoter variants were then introduced by primers via inverse PCR. The promoter-ribozyme parts are flanked by two BsaI sites with GGAG and TTAA as four nucleotides overlaps for the subsequent type IIS reaction to build level 1 plasmids. All of the engineered ribozyme sequences end with TTAA. The RBS-CDS constructs containing the enzymes were obtained from a concurrent study (unpublished results), which we mutated to eliminate BsaI and BbsI sites. Note that the RBS for kdcA in the monocistronic design is K005 (designed using the RBS Calculator) as compared to K007, which was used for the operon-based designs in Figure 1B and C. The level 1 plasmids are based on the pL1 backbone, originating from pMJS1CD (24) with Kanamycin resistance (also free of BsaI and BbsI sites). Then, 20 fmol of each level 0 plasmid are mixed with 5 U BsaI (New England Biolabs, #R0539S) and 5 U T4 DNA Ligase (Promega, #M1794) for a total of 10 μl 1× Promega T4 DNA Ligase Buffer and incubated. Two level 1 plasmids carrying expression cassettes with high (+1) and low (−1) expression levels for each of the six pathway genes were built. In total, there are 12 level 1 plasmids for six pathway genes and 12 containing mRFP in the pathway context for part characterization. Inverse PCR (iPCR) was used to generate the level 1 plasmids for the kdcA and aksF libraries using pL1-kdcA-TU2 and pL1-aksF-TU2 as templates. To build the level 2 plasmids, the six pathway cistrons are assembled in the order shown in Figure 2A. The cistrons are PCR amplified from the level 1 plasmids to give each construct specific cohesive ends upon BsaI digestion corresponding to the assigned position. The final expression plasmid backbone pL2 is derived from pAKP444 (unpublished results), where the BsaI sites in β-lactamase gene were eliminated. The seven assembly junction regions were sequence verified by Sanger sequencing. To build the kdcA and aksF libraries shown in Figure 3B and C, only the expression cassette for either kdcA or aksF was changed, while the other gene expression cistrons are the same as the #4 construct. To achieve higher expression, an extra copy of kdcA or aksF was introduced in a separate p15a Kanr plasmid (Supplementary Figure S4C). The T7 promoter M4 is used for both kdcA and askF expression. The same plasmid for overexpressing aksF or kdcA was introduced into the strain containing construct [−1, −1, −1, +1, +1, +1] individually for Figure 3A.

Bottom Line: This is compared to a 64-member full factorial library just varying expression (0.64 Mb of DNA assembly).Statistical analysis of the screening data from these libraries leads to different predictions as to whether the expression of enzymes needs to increase or decrease.This work introduces a generalizable platform to co-optimize genetic and non-genetic factors.

View Article: PubMed Central - PubMed

Affiliation: Synthetic Biology Center, Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA.

Show MeSH
Related in: MedlinePlus