Research Article

Improving the Expression of Human Granulocyte Colony Stimulating Factor in Escherichia coli by Reducing the GC-content and Increasing mRNA Folding Free Energy at 5’-Terminal End

Introduction

Granulocyte-colony stimulating factor (G-CSF), also known as colony-stimulating factor 3 (CSF3), is a glycoprotein that acts in haematopoiesis by controlling the production, survival, proliferation, differentiation and function of granulocytes that comprises 70% of white blood cells.^1-3 Recombinant human G-CSF has many pharmaceutical applications. In oncology, it has been used especially for the prevention of chemotherapy-induced neutropenia and also mobilisation of peripheral blood progenitor cells in a healthy donor before collection in hematopoietic stem cell transplantation.^4,5

Human G-CSF is a 19 kDa single-chain polypeptide consisting of 174 amino acids which contains five Cysteines forming two intramolecular disulfide bridges between Cys36-Cys42 and Cys64-Cys74, and one free Cysteine at residue 17.^6-8 Naturally occurred G-CSF has an O-linked glycosylation at Thr133 which protects the proteins from aggregation without affecting its biological activity.⁹

As a host organism for producing recombinant proteins, E. coli has many advantages including simple nutrient requirement, high growth rate on low cost substrate, well characterized on molecular genetics and cellular physiology.⁸ It also classified as safe organism for producing recombinant protein products.¹⁰ The non-glycosylated G-CSF produced in E. coli can be biologically active as produced in mammalian cell, but differs in the addition of N-terminal methionine.^11,12

As one of most important biopharmaceuticals worldwide, recombinant G-CSF production should be optimized in order to get a high-yield protein product with low production cost. In addition to the choice of expression system and other environmental conditions, several crucial factors at the genetic level should be considered such as codon usage, rare codons composition, mRNA secondary structure and GC-content near the translation initiation region (TIR).¹³

Many studies reported that nucleotide composition at 5’-terminal end of the open reading frame (ORF) or near the TIR is often a bottleneck in recombinant protein production. Theoretically, strong folding at 5’-terminal ends can cover the area of ribosome binding site and inhibits the translation initiation rate. Reducing the GC-content in this region is known can improve the protein expression level in E. coli.^14-16 Whereas, higher value of mRNA folding free energy indicates more unstable hairpin, which is preferred in expressing recombinant proteins.

In the present study, effects of reducing the GC content and increasing mRNA folding free energy at 5’-terminal end on G-CSF protein expression level were investigated. Two mutant variants of ORF were made by reducing the GC-content of eight codons immediately downstream the start codon from 75% to 62.5% and 54.2% which performed through synonymous codons substitutions. The expression of G-CSF was regulated under T7 promoter in E. coli expression system and monitored using SDS PAGE and western blot analyses.

Materials and Methods

Microbial strains and plasmid

The E. coli strain TOP10 was used for cloning and E. coli NiCo21(DE3) was used for protein expression. Plasmid pJexpress414 (ATUM, Newark, California, United States) was used as the expression vector.

Media

Lysogeny Broth (LB) medium [1% (w/v) Tryptone, 0.5% (w/v) Yeast Extract, and 1% (w/v) NaCl] was used for both cloning and expression step. Except for making E. coli competent cells and negative control, LB medium supplemented with 100 µg/mL ampicillin (LB-amp) was used in this whole experiment.

Primer design and template

PET21-hGCSF, which already constructed by Fuad et al,¹⁷ was used as the template for PCR amplification. Three forward primers were designed in this experiment (Table 1). One of them was designed based on the original hG-CSF template and not modified, while the others two were 5’-ends terminally modified and designed for synonymous codon substitution. However, the GC content reduction at 5’-ends were calculated using Oligoanalyzer 3.1 and minimum free energy (MFE) of local mRNA secondary structure formation near the rbs were monitored using KineFold and RNAfold web server.^18-20 To facilitate the cloning step, all forward primers were added with Nde I restriction site, whereas the reverse primer was added with Xho I restriction site. All primers were purchased from Integrated DNA Technologies (IDT, Singapore).

PCR and cloning

Three variants of the ORF encoding the same recombinant hG-CSF protein were amplified from pET21-hGCSF. PCR amplification was carried out separately for each pair of primers and conducted as follows: initial denaturation at 95°C for 5 minutes; 30 cycles of denaturation at 95°C for 1 minute, annealing at 66.8°C for 30 seconds, and extension at 72°C for 1 minute, then final extension at 72°C for 5 minutes. PCR product was purified using a PCR purification kit (Qiagen, Hilden, Germany). Purified PCR product and pJExpress414 expression vector (ATUM, Newark, California, United States) were exposed to Nde I/Xho I double digestion (3 IU/µg) (Thermo Fisher Scientific, Waltham, Massachusetts, United States). Digested products were examined with agarose gel electrophoresis and isolated from agarose gel using Gel/PCR DNA Fragments Extraction Kit (Geneaid, New Taipei City, Taiwan). Each ORF was cloned into an expression vector using T4-DNA ligase enzyme (2 IU/50 ng plasmid) (Kapa Biosystems, Inc., Wilmington, Massachusetts, United States) and carried out for 72 hours at 4°C. The plasmids obtained were named pJ414-hGCSF1, pJ414-hGCSF2 and pJ414-hGCSF3.

Transformation and characterization

The ligation product was transformed into E. coli TOP10 using one-step transformation and stock solution (TSS) by Chung et al,²¹ then continued with heat-shock method as described by Froger and Hall with minor modification.²² The transformants were selected in LB-amp medium at room temperature for 20 hours. Recombinant plasmids from E. coli TOP10 transformants were isolated using alkaline lysis method according to Green and Sambrook²³ and characterized by restriction analyses, subsequently sequenced to ensure that the synonymous substitution at 5’- terminal ends was successfully performed (1^st BASE, Selangor, Malaysia). The amino acid sequence of hG-CSF1, hG-CSF2, and hG-CSF3 were deduced using Translate Tools (ExPaSy) continued with alignment analysis using Pairwise Sequence Alignment Tools (EMBL-EBI).^24,25 Confirmed recombinant plasmids were then transformed into E. coli NiCo21(DE3) and then characterized by colony PCR. The colony PCR was done using 2 µL of culture as a template and PCR protocol as mentioned above.

Growth condition for expression of hG-CSF

Each variant of E. coli NiCo21(DE3) transformants was grown overnight in 1 mL LB-amp medium at 25°C, 150 rpm. This culture was diluted 1:50 with LB-amp medium and grown at 25°C, 150 rpm until 0D₆₀₀~1 was reached. After that, IPTG with a final concentration of 1 mM was added and growth was continued overnight at 25°C, 150 rpm. E. coli NiCo21(DE3) non-transformant was used as a negative control which had the same treatment as above.

Protein characterization

The sample for total proteins analyses was made by adding 50 µL of sample buffer and 50 µL of denaturing lysis buffer (8M urea, 10 mM Tris-Cl, and 100 mM NaH₂PO₄) into pellet cells from 1 mL culture, then boiled for 15 m. About 1 µL of sample was subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) according to Laemmli on 15% polyacrylamide gel.²⁶

Western blot analysis was carried out using a method as described in Dewi et al.²⁷ In this experiment, anti-G-CSF mouse monoclonal IgG antibody (1:2000 dilution, Santa Cruz Biotechnology Inc., Santa Cruz, California, United States) was used as primary antibody, while AP-conjugated goat anti-mouse IgG antibody (1:3500 dilution, Santa Cruz Biotechnology, Inc., Santa Cruz, California, United States) was used as secondary antibody. Both SDS-PAGE and Western blot analyses used 1.5 µg filgrastim (Neupogen®, Kirin-Amgen Inc., Thousand Oaks, California, United States) as a standard for recombinant h-G-CSF proteins. Finally, the amount of recombinant hG-CSF proteins which expressed from pJ414-hGCSF1, pJ414-hGCSF2, and pJ414-hGCSF3 recombinant plasmids were calculated using ImageJ software.²⁸

Results and Discussion

Synonymous codon substitution

To observe the effects of GC content on local mRNA secondary structure formation and hG-CSF expression level, a synonymous substitution was performed to alter several codon sequences immediately downstream the start codon. Codon usage table of Escherichia coli B from Kazusa Codon Database (http://www.kazusa.or.jp) was used to select the appropriate codons for synonymous substitution.²⁹ In this study, two mutant variants of hG-CSF1 ORF has been constructed. About eight codons encoding T P L G P A S S amino acids were undergone GC content reduction which calculated using Oligoanalyzer 3.1 web server.¹⁸ The codon adaptation index (CAI) of eight codon sequences was calculated using CAIcal web server.³⁰ After those analyses have been conducted, the forward primers were made for PCR amplification.

Table 2 shows that the GC content and CAI value of eight codons were decreased sequentially from hG-CSF1, hG-CSF2 and hG-CSF3. The MFE required for mRNA secondary structure formation around the area of ribosome binding site (rbs) was predicted using RNAfold and KineFold web server.^19,20 Table 3 shows that a higher GC content has a lower folding free energy. The higher value of MFE is considered the more unstable mRNA secondary structure which preferred for protein expression.³¹

Plasmid construction was performed to obtain recombinant plasmid carrying ORF encoding hG-CSF proteins. All variants of hG-CSF ORF were amplified from pET21-hGCSF and the result showed 0.53 kb of PCR product (Figure not shown). The expression vector and insert DNA were prepared by double digestion of pJExpress414 and PCR products using Nde I and Xho I restriction enzymes (). Lane 4 showed result of pJExpress414 digestion which consisting plasmid backbone (4.3 kb) and previous insert (0.75 kb).

Close

Figure 1. Preparation of expression vector and insert DNA by Nde I/Xho I double digestion. Lane 1-3, ORF of hG-CSF1, hG-CSF2 and hG-CSF3, respectively; Lane 4, recombinant pJExpression414 expression vector with previous insert DNA.

To ensure a good balance between stability of annealed DNA ends and activity of DNA ligase, thus ligation of cohesive DNA ends is normally carried out at 12-16°C for 24 hours.³² However, a lower temperature has advantage in decreasing the random motion in ligation mix, enable the cohesive ends to find their partner. The hydrogen bonds were subsequently formed and remained by the time DNA ligase helps phosphodiester bonds formation. Because a low temperature will reduce DNA ligase activity, thus a longer incubation time was used in this experiment.

The recombinant plasmids were isolated from growing E. coli TOP10 transformants and characterized by restriction and DNA sequencing analyses. Restriction analysis was performed to determine the actual size of the expression vector and insert DNA. shows double digestion of recombinant plasmids using Nde I and Xho I restriction enzymes, resulting 2 DNA bands with size of 4.3 kb and 0.53 kb which corresponds to the theoretical size of pJExpress414 expression vector and hG-CSF ORF, respectively. PJExpress414 with previous insert DNA was included in this analysis as a negative control.

Close

Figure 2. Restriction analysis of recombinant plasmids by Nde I/Xho I double digestion. Lane 1-3, pJ414-hGCSF1, pJ414-hGCSF2, pJ414-hGCSF3, respectively; Lane 4, recombinant pJExpress414 expression vector with previous insert DNA.

The DNA sequencing analysis and amino acids sequences alignment were carried out to ensure that synonymous substitution was successfully achieved. shows the nucleotide sequences of some initial codons which already been substituted. Those DNA sequences were translated into amino acid and then aligned to one and another to ensure that no amino acids have been changed. The result of these analyses concluded that the three variants of recombinant plasmids were successfully constructed without altering the amino acids composition.

Close

Figure 3. Pairwise sequences alignment between (A) hG-CSF 1 and hG-CSF2; (B) hG-CSF1 and hG-CSF3. The synonymous mutation was marked within a box.

All the confirmed recombinant plasmid were then transformed into E. coli NiCo21(DE3) and characterized by colony PCR using specific primers. This screening method was important because not all colonies appear following transformation carrying plasmid with the desired DNA fragment.³³ shows the colony PCR from colonies carrying pJ414-hGCSF1, pJ414-hGCSF2, and pJ414-hGCSF3 expression vector. The positive results were indicated by the presence of DNA bands with size of 0.53 kb ().

Close

Figure 4. Colony PCR analysis of E. coli NiCo21(DE3) transformants using specific primers. Lane 1-3, amplification fragment from colony carrying pJ414-hGCSF1, pJ414-hGCSF2, and pJ414-hGCSF3, respectively; lane 4, negative control.

Expression and characterization of hG-CSF proteins

To determine the expression level of recombinant hG-CSF proteins which already been genetically modified, the positive colonies were grown in LBamp and induced with IPTG. Neupogen as a standard was included in SDS-PAGE and Western Blot analyses. The hG-CSF band is deduced to have the approximate size of 18.8 kDa as it is shown on the electrophoregram (). However, there was no band with a size of 18.8 kDa detected in hG-CSF1 as it found in hG-CSF2 and hG-CSF3. It is indicated that expression of hG-CSF1 proteins did not occur or very weak expression. Therefore, Western blot analysis was carried out to prove that the prediction from SDS-PAGE analysis was correct.

Close

Figure 5. (A) SDS-PAGE and (B) Western blot analyses of hG-CSF proteins. Lane 1, total proteins of untransformed E. coli NiCo21(DE3); Lane 2-4, total proteins of hG-CSF1, hG-CSF2 and hG-CSF3, respectively; Lane 5, filgrastim as a standard protein.

Compared to SDS-PAGE analysis, Western Blot has better sensitivity and specificity to protein detection. This technique has the ability to detect picogram levels of protein in a sample. Besides, the specific interaction between antibody and antigen allows the detection of target protein in complex mixtures containing >100,000 different proteins.³⁴ However, the result of western blot analysis showed that no protein band with size of 18.8 kDa was detected in hG-CSF1 (), even though we used anti-hG-CSF monoclonal antibody which has a high specificity. This result proved that the hG-CSF1 proteins cannot be expressed in E. coli NiCo21(DE3) expression system. ImageJ software was used to calculate the total proteins of hG-CSF2 and hG-CSF3 in overproduction culture. The results of this analysis showed that the yield of hG-CSF2 and hG-CSF3 were 31.44 and 73.4 µg/mL culture, respectively.

In this study, we compared the heterologous expression of three variants of ORF encoding hG-CSF in E. coli NiCo21(DE3). This research was conducted based on many reports of previous studies which stated that there was a correlation between nucleotides composition at 5’-terminal ends and protein expression level.^14-16,35 In addition, it has been known that the distribution of codon usage frequency at the 5’-terminal ends was different to that observed in the rest of the ORF.^36-39 The obtained results provide important information in choosing strategies to increase protein expression level of hG-CSF through genetic improvement.

Kudla et al reported that there is a strong correlation between mRNA folding and protein synthesis.¹⁴ Their study suggested that tightly folded mRNA near the start codon will inhibit translation initiation rate. In this case, high GC content at the 5’-terminal ends were correlated with low mRNA folding free energy which increases the stability of mRNA secondary structure formation. It makes thermodynamic sense. GC pair has three hydrogen bonds while AU pair has only two, thus GC-rich mRNA tends to fold into more stable structure than AU-rich. Hence, synonymous substitutions around the start codon can regulate protein expression.

Their findings lead to the following prediction: Adding a stretch of codons with weak folded mRNA which indicated with high mRNA folding free energy at the 5’-terminal ends should increase protein expression, even if the additional codons have low CAI.¹⁴ Our results were consistent with their findings. We have found that lowering GC content of DNA encoding hG-CSF immediately downstream the start codon could increase mRNA folding free energy, and resulted in more unstable mRNA secondary structure which leads to increase protein expression level.

Our findings were also in line with research by Gu et al¹⁶ which suggested that decreased mRNA stability at the translation-initiation region facilitates efficient start-codon recognition.¹⁶ The ability of mRNA to sequester ribosomes will be inhibited when local mRNA secondary structure is present, thus lowering the effectiveness of translation initiation rate.¹³ Previously, Jia et al⁴⁰ were also stated that the stability of mRNA secondary structure is an important factor which influences protein expression levels.⁴⁰ Theoretically, if Shine-Dalgarno sequence and start codon itself tends to be base-paired to other nucleotides, then they will not interact efficiently with the pre-initiation complex/30S subunits.¹³

Moreover, we have found that the optimization of only 8 codons immediately downstream the start codon was enough in order to improve the expression level of recombinant proteins in E. coli expression system. This result was better than the previous study by Ghavim et al which stated that optimization of 20 first codons was enough to achieve high expression level of recombinant proteins in E. coli.⁴¹

According to our results, the highest protein expression level was achieved when GC content at 5’-terminal ends reduced from 75% to 54.2% along with increasing the MFE from -12.4 Kcal/mol to -7.3 kcal/mol using RNAfold web server and -15.9 kcal/mol to -10.8 kcal/mol using Kinefold web server. These results prove that GC content reduction and unstable mRNA secondary structure at 5’-terminal ends resulted in mRNA efficiency for translation.

Conclusion

In brief, our study proves that lowering the GC-content at 5’-terminal ends, exactly on eight codons downstream the ATG, has a great impact to improve the expression level of recombinant hG-CSF proteins in E. coli . Based on our findings, trials of GC-content reduction through synonymous codon substitutions near the start codon is recommended for improving the expression level of recombinant proteins in E. coli expression system, while preventing problems associated with the mRNA secondary structure formation.

Ethical Issues

Not applicable.

Conflict of Interest

There is no conflict of interest.

Acknowledgments

This research was supported by Indonesian Institute of Sciences Flagship Program during 2015-2017 fiscal years.

References

1. Han CH, Ryu HB, Hang B, inventors; University of Ulsan Foundation for Industry Cooperation, assignee. Water-soluble expression and purification method of human GCSF recombinant protein having biological activities. United States patent US 9,926,358. 2018.
2. Demetri GD, Griffin JD. Granulocyte colony-stimulating factor and its receptor. Blood 1991; 78(11):2791-808.
3. Lieschke GJ, Grail D, Hodgson G, Metcalf D, Stanley E, Cheers C. Mice lacking granulocyte colony-stimulating factor have chronic neutropenia, granulocyte and macrophage progenitor cell deficiency, and impaired neutrophil mobilization. Blood 1994; 84(6):1737-46.
4. Cetean S, Căinap C, Constantin AM, Căinap S, Gherman A, Oprean L. The importance of the granulocyte-colony stimulating factor in oncology. Clujul Med 2015; 88(4):468-72. doi: 10.15386/cjmed-531 [Crossref]
5. Pahnke S, Egeland T, Halter J, Hägglund H, Shaw BE, Woolfrey AE. Current use of biosimilar G-CSF for haematopoietic stem cell mobilisation. Bone Marrow Transplant 2019; 54(6):858-66. doi: 10.1038/s41409-018-0350-y [Crossref]
6. Lu HS, Boone TC, Souza LM, Lai PH. Disulfide and secondary structures of recombinant human granulocyte colony stimulating factor. Arch Biochem Biophys 1989; 268(1):81-92. doi: 10.1016/0003-9861(89)90567-5 [Crossref]
7. Abolghasemi Dehaghani S, Babaeipour V, Mofid MR, Divsalar A, Faraji F. An efficient purification method for high recovery of recombinant human granulocyte colony stimulating factor from recombinant E coli. Int J Environ Sci Dev 2010; 1(2):111-4. doi: 10.7763/ijesd.2010.v1.22 [Crossref]
8. Babaeipour V, Khanchezar S, Mofid MR, Pesaran Hagi Abbas M. Efficient process development of recombinant human granulocyte colony-stimulating factor (rh-GCSF) production in Escherichia coli. Iran Biomed J 2015; 19(2):102-10. doi: 10.6091/ibj.1338.2015 [Crossref]
9. Kraševec N, Milunović T, Lasnik MA, Lukančič I, Komel R, Porekar VG. Human granulocyte colony stimulating factor (G-CSF) produced in the filamentous fungus Aspergillus niger. Acta Chim Slov 2014; 61(4):709-17.
10. Broedel SE Jr, Papciak SM, Jones WR. The selection of optimum media formulations for improved expression of recombinant proteins in E coli. Athena Enzyme Systems Technical Bulletin 2001; 2:1.
11. Bönig H, Silbermann S, Weller S, Kirschke R, Körholz D, Janssen G. Glycosylated vs non-glycosylated granulocyte colony-stimulating factor (G-CSF)--results of a prospective randomised monocentre study. Bone Marrow Transplant 2001; 28(3):259-64. doi: 10.1038/sj.bmt.1703136 [Crossref]
12. Kim CK, Lee CH, Lee SB, Oh JW. Simplified large-scale refolding, purification, and characterization of recombinant human granulocyte-colony stimulating factor in Escherichia coli. PLoS One 2013; 8(11):e80109. doi: 10.1371/journal.pone.0080109 [Crossref]
13. Tuller T, Zur H. Multiple roles of the coding sequence 5’ end in gene expression regulation. Nucleic Acids Res 2015; 43(1):13-28. doi: 10.1093/nar/gku1313 [Crossref]
14. Kudla G, Murray AW, Tollervey D, Plotkin JB. Coding-sequence determinants of gene expression in Escherichia coli. Science 2009; 324(5924):255-8. doi: 10.1126/science.1170160 [Crossref]
15. Allert M, Cox JC, Hellinga HW. Multifactorial determinants of protein expression in prokaryotic open reading frames. J Mol Biol 2010; 402(5):905-18. doi: 10.1016/j.jmb.2010.08.010 [Crossref]
16. Gu W, Zhou T, Wilke CO. A universal trend of reduced mRNA stability near the translation-initiation site in prokaryotes and eukaryotes. PLoS Comput Biol 2010; 6(2):e1000664. doi: 10.1371/journal.pcbi.1000664 [Crossref]
17. Fuad AM, Yuliawati, Agustiyanti DF. Desain dan Konstruksi Gen CSF3 Sintetik (CSF3SYN) Mengandung Kodon Preferensi Escherichia coli Dengan Teknik PCR. Proceedings of National Seminar XXII “Kimia dalam Industri dan Lingkungan”; 2013. p. 87-98.
18. Owczarzy R, Tataurov AV, Wu Y, Manthey JA, McQuisten KA, Almabrazi HG. IDT SciTools: a suite for analysis and design of nucleic acid oligomers. Nucleic Acids Res 2008; 36:W163-9. doi: 10.1093/nar/gkn198 [Crossref]
19. Gruber AR, Lorenz R, Bernhart SH, Neuböck R, Hofacker IL. The vienna RNA websuite. Nucleic Acids Res 2008; 36:W70-4. doi: 10.1093/nar/gkn188 [Crossref]
20. Xayaphoummine A, Bucher T, Isambert H. Kinefold web server for RNA/DNA folding path and structure prediction including pseudoknots and knots. Nucleic Acids Res 2005; 33:W605-10. doi: 10.1093/nar/gki447 [Crossref]
21. Chung CT, Niemela SL, Miller RH. One-step preparation of competent Escherichia coli: transformation and storage of bacterial cells in the same solution. Proc Natl Acad Sci U S A 1989; 86(7):2172-5. doi: 10.1073/pnas.86.7.2172 [Crossref]
22. Froger A, Hall JE. Transformation of plasmid DNA into E coli using the heat shock method. J Vis Exp 2007(6):253. doi: 10.3791/253 [Crossref]
23. Green MR, Sambrook J. Preparation of plasmid DNA by alkaline lysis with sodium dodecyl sulfate: minipreps. Cold Spring Harb Protoc 2016; 2016(10). doi: 10.1101/pdb.prot093344 [Crossref]
24. Gasteiger E, Gattiker A, Hoogland C, Ivanyi I, Appel RD, Bairoch A. ExPASy: the proteomics server for in-depth protein knowledge and analysis. Nucleic Acids Res 2003; 31(13):3784-8. doi: 10.1093/nar/gkg563 [Crossref]
25. Kanz C, Aldebert P, Althorpe N, Baker W, Baldwin A, Bates K. The EMBL nucleotide sequence database. Nucleic Acids Res 2005; 33(Suppl_1):D29-D33. doi: 10.1093/nar/gki098 [Crossref]
26. Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970; 227(5259):680-5. doi: 10.1038/227680a0 [Crossref]
27. Dewi KS, Retnoningrum DS, Riani C, Fuad AM. Construction and periplasmic expression of the anti-EGFRvIII ScFv antibody gene in Escherichia coli. Sci Pharm 2016; 84(1):141-52. doi: 10.3797/scipharm.ISP.2015.06 [Crossref]
28. Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nat Methods 2012; 9(7):671-5. doi: 10.1038/nmeth.2089 [Crossref]
29. Nakamura Y, Gojobori T, Ikemura T. Codon usage tabulated from international DNA sequence databases: status for the year 2000. Nucleic Acids Res 2000; 28(1):292. doi: 10.1093/nar/28.1.292 [Crossref]
30. Puigbò P, Bravo IG, Garcia-Vallvé S. E-CAI: a novel server to estimate an expected value of Codon Adaptation Index (eCAI). BMC Bioinformatics 2008; 9:65. doi: 10.1186/1471-2105-9-65 [Crossref]
31. Ringnér M, Krogh M. Folding free energies of 5’-UTRs impact post-transcriptional regulation on a genomic scale in yeast. PLoS Comput Biol 2005; 1(7):e72. doi: 10.1371/journal.pcbi.0010072 [Crossref]
32. Lund AH, Duch M, Pedersen FS. Increased cloning efficiency by temperature-cycle ligation. Nucleic Acids Res 1996; 24(4):800-1. doi: 10.1093/nar/24.4.800 [Crossref]
33. Bergkessel M, Guthrie C. Colony PCR. Methods Enzymol 2013; 529:299-309. doi: 10.1016/b978-0-12-418687-3.00025-2 [Crossref]
34. Ghosh R, Gilda JE, Gomes AV. The necessity of and strategies for improving confidence in the accuracy of western blots. Expert Rev Proteomics 2014; 11(5):549-60. doi: 10.1586/14789450.2014.939635 [Crossref]
35. Kucharova V, Skancke J, Brautaset T, Valla S. Design and optimization of short DNA sequences that can be used as 5’ fusion partners for high-level expression of heterologous genes in Escherichia coli. Appl Environ Microbiol 2013; 79(21):6655-64. doi: 10.1128/aem.01676-13 [Crossref]
36. Bulmer M. Codon usage and intragenic position. J Theor Biol 1988; 133(1):67-71. doi: 10.1016/s0022-5193(88)80024-9 [Crossref]
37. Qin H, Wu WB, Comeron JM, Kreitman M, Li WH. Intragenic spatial patterns of codon usage bias in prokaryotic and eukaryotic genomes. Genetics 2004; 168(4):2245-60. doi: 10.1534/genetics.104.030866 [Crossref]
38. Makhoul CH, Trifonov EN. Distribution of rare triplets along mRNA and their relation to protein folding. J Biomol Struct Dyn 2002; 20(3):413-20. doi: 10.1080/07391102.2002.10506859 [Crossref]
39. Clarke TF 4th, Clark PL. Increased incidence of rare codon clusters at 5’ and 3’ gene termini: implications for function. BMC Genomics 2010; 11:118. doi: 10.1186/1471-2164-11-118 [Crossref]
40. Jia M, Li Y. The relationship among gene expression, folding free energy and codon usage bias in Escherichia coli. FEBS Lett 2005; 579(24):5333-7. doi: 10.1016/j.febslet.2005.08.059 [Crossref]
41. Ghavim M, Abnous K, Arasteh F, Taghavi S, Nabavinia MS, Alibolandi M. High level expression of recombinant human growth hormone in Escherichia coli: crucial role of translation initiation region. Res Pharm Sci 2017; 12(2):168-75. doi: 10.4103/1735-5362.202462 [Crossref]