- Email: [email protected]
Synthesis of non-canonical branched-chain amino acids in Escherichia coli and approaches to avoid their incorporation into recombinant proteins
Available online at www.sciencedirect.com
ScienceDirect Synthesis of non-canonical branched-chain amino acids in Escherichia coli and approaches to avoid their incorporation into recombinant proteins Christian Reitz, Qin Fan and Peter Neubauer Abstract In E. coli the non-canonical amino acids acids norvaline, norleucine, and b-methylnorleucine, which derive from an offpathway of the branched-chain amino acid synthesis route are synthesized and incorporated into cellular and recombinant proteins. The synthesis of these amino acids is supported by a high flux of glucose through the glycolytic pathway in combination with a derepression of the enzymes of the branched chain amino acid pathway, for example, when leucine-rich proteins are produced. Avoiding the synthesis and misincorporation of these amino acids has been challenging, especially in large-scale pharmaceutical processes where the problem is boosted by the typical fed-batch production and the technical limitation of mass transfer in the bioreactors.Despite its industrial importance, so far this issue has not been discussed comprehensively. Therefore this paper reviews, firstly, the specific pathway of the non-canonical branched chain amino acids starting at pyruvate, secondly, the molecular factors for their misincorporation, and thirdly, approaches to avoid this misincoporation. While the synthesis of these amino acids is difficult to prevent due to the broad promiscuity of the connected enzymes, recent studies on the control mechanisms of aminoacyl tRNA synthetases open new opportunities to avoid this misincorporation. Address Technische Universita¨t Berlin, Institute of Biotechnology, Department of Bioprocess Engineering, Ackerstr. 76, D-13355 Berlin, Germany Corresponding author: Neubauer, Peter ([email protected])
Current Opinion in Biotechnology 2018, 53:248–253 This review comes from a themed issue on Pharmaceutical biotechnology Edited by Amanda Lewis and Nripen Singh
https://doi.org/10.1016/j.copbio.2018.05.003 0958-1669/ã 2018 Elsevier Ltd. All rights reserved.
Introduction The genetic code is almost universal and other amino acids than the 20 canonical amino acids are generally not accepted in nature [1] with very few exceptions. Current Opinion in Biotechnology 2018, 53:248–253
Examples for such non-typical amino acids are selenocysteine and pyrrolysine in archae, that are incorporated into the peptide chain by reprogramming of one of the rarely used stop codons (UGA in connection to a downstream located specific mRNA hairpin structure for selenocystein, UAG without the need for a downstream element for pyrrolysine) [2]. Also, despite the universality of the genetic code, translational errors can occur, which is called mistranslation. The degree of mistranslation varies in different organisms and even can be beneficial in certain circumstances [3]. During the last decade also the efficient artificial translational incorporation of noncanonical amino acids in proteins became possible either by engineering of the genetic code or by replacement of the natural amino acid with a chemical homolog which is accepted by its aminoacyl tRNA synthetase [4]. in vitro investigations indicated a surprising plasticity of certain natural aminoacyl tRNA synthetases and ribosomal peptidyltransferases which even allows the incorporation of D-amino acids [5]. In non-engineered organisms translational misincorporation of non-canonical amino acids has been mainly found for the amino acids norvaline, norleucine and b-methylnorleucine which are side products of the branched chain amino acid pathway. Based on early studies with Serratia marcenscens the synthesis pathway for these amino acids was characterized [6]. As illustrated in Figure 1, norvaline, norleucine, and b-methylnorleucine are products of enzymatic elongation reactions beginning with a-ketobutyrate catalyzed by a-isopropylmalate synthase (PMS) (leuA), a-isopropylmalate isomerase (ISOM) (leuCD), and a-isopropylmalate dehydrogenase (IPMD) (leuB) and a final transamination step. Branching to norleucine includes a second cycle with these enzymes, while b-methylnorleucine synthesis proceeds from a-ketovalerate with the enzymes of the isoleucine/valine pathway. This pathway is triggered by conditions which lead to a lower intracellular level of leucine and conseqently cause a derepression of the leucine operon [7]. Additionally, the accumulation of a-ketobutyrate is a requirement since the affinity of a-isopropylmalate synthase (IPMS) for a-ketobutyrate is significantly lower compared to its normal substrate a-ketoisovalerate as characterised for S. marcescens [6]. Sycheva et al. [8] demonstrated that the synthesis of norvaline and norleucine also proceeds after inactivation of threonine deaminase (ilvA), indicating an alternative shortcut pathway, beside the aspartate threonine pathway, from pyruvate to a-ketobutyrate, a reaction which is www.sciencedirect.com
Incorporation of Non-canonical branched-chain amino acids Reitz, Fan and Neubauer 249
Figure 1
pyruvate
pyr CO2
ilvD
ilvC
AHAS
α-ketoisovalerate
α,β-dihydroxyisovalerate
α-acetolactate
ilvE, avtA
valine
leuA leuA leuCD leuB
ilvE, tyrB
leuCD
Isopropylmalate
leuB
AHAS
ilvA
ilvC
α-ketoisocaproate
ilvE, avtA
ilvD
pyr CO2
threonine
α-ketobutyrate
Leucine
Isoleucine
α-aceto-α-hydroxybutyrate α,β-dihydroxy-β-methylvalerate α-keto-β-methylvalerate
leuA leuCD leuB AHAS
ilvC
pyr CO2
ilvE
ilvD
α-ketovalerate α-aceto-α-hydroxyvalerate α,β-dihydroxy-β-methylcaproate α-keto-β-methylcaproate leuA
β-methylnorleucine
ilvE, avtA
leuCD leuB
Norvaline
ilvE, tyrA
α-ketocaproate
Norleucine Current Opinion in Biotechnology
Schematic view of predicted biosynthetic pathway of the modified branched-chain amino acids including norvaline, norleucine and b-methylnorleucine from pyruvate via the so-called “ketoacid chain elongation pathway” over a-ketobutyrate and a-ketovalerate to a-ketocaproate facilitated by the promiscuous enzymes of the (iso)-leucine biosynthetic pathway in E. coli.
also possible by the enzymes of the leuABCD operon (see Figure 1). While on one side the area of synthetic biology considers today the extension of the genetic code and the incorporation of non-canonical amino acids as an interesting issue to expand the functionality of proteins, the uncontrolled misincorporation of self-synthesized non-canonical proteins is a challenging problem in the production of biopharmaceuticals on the other side. Recently the problem of amino acid misincorporation was comprehensively reviewed in an excellent paper by Harris and Kilby [9]. The challenge in the pharmaceutical production of proteins is that such rare events of incorporation errors produce incorrect protein variants which eventually could induce immunological and other responses. Therefore it is a major concern to avoid or minimize these errors, which must be documented by the producer according to the ICH Q6B guideline. The previous review focused on the misincorporation of natural amino acids, however, www.sciencedirect.com
despite the industrial importance, misincorporation of non-canonical amino acids including norvaline, norleucine and b-methylnorleucine in recombinant E. coli cultures was not considered although it is a serious problem in the production of several pharmaceutically relevant proteins (for a detailed summary see Table 1).
Misincorporation of non-canonical amino acids into recombinant proteins Overall, the misincorporation of non-canonical amino acids in E. coli is influenced by cultivation conditions and also depends on the composition of the protein. Particularly the production of leucine-rich recombinant proteins seems to be critical [10]. This is due to the high need of leucine after induction. Consequently the intracellular concentration of leucine decreases and the leucine operon may be derepressed (i.e., activated). If this coincides with conditions which lead to an elevated synthesis of pyruvate, such as a high flux into the glycolytic pathway, the Current Opinion in Biotechnology 2018, 53:248–253
250 Pharmaceutical biotechnology
Table 1 Summary of references for the incorporation of modified amino acids derived from the branched chain amino acid pathway into recombinant proteins Amino acid
Product
NL NL NL
Hirudin in E. coli Bovine somatotropin in E. coli human brain-derived neurotrophic factor in E. coli interleukin-2 in E. coli Met-rich vaccine candidate Hirudin in E. coli Rec. Hemoglobin in E. coli Proinsulin in E. coli
NL NL b-MNL Norvaline Norvaline, ß-MNL, NL
Comment
AA composition
Ref.
in Met Positions
27 leu/191aa
[24] [7] [25]
in in in in
26 Leu/152 aa
Met Positions Met Positions Ile positions Leu positions
72 Leu/575 aa
[26,27] [28] [24] [16] Non-published own results
NL – norleucine, b-MNL – b-methylnorleucine.
non-canonical branched chain amino acids are synthesized by direct chain elongation of pyruvate by the enzymes of the leucine operon, and eventually incorporated into the recombinant product. This scenario is relevant, as some of the pharmaceutically interesting proteins have a high leucine content compared to an average E. coli protein; for example, recombinant hemoglobin and interleukin-2 have a leucine ratio of 13 and 17%, respectively, while an average E. coli protein only contains 8.4% (Table 2). Incorporation of norvaline, b-methylnorleucine, and norleucine into a peptide chain occur by the erroneous
Table 2 Relative content of the different branched-chain amino acids in pharmaceutically interesting proteins (calculated according to the NCBI CCDS database)
a
E. coli Interleukin-2 Human proinsulin Hemoglobin Human growth hormone Bovine Somatrotropin (MBS) Interferon a2 Interferon b1 a
Leucine
Valine
Isoleucine
8.4% 17.6% 17.3% 12.6% 14.7% 16.1% 14.3% 16.0%
7.9% 3.3% 6.9% 9.2% 3.2% 3.7% 5.3% 2.7%
5.4% 6.5% 2.3% 0.0% 3.2% 3.7% 3.7% 6.4%
Content of overall protein in E. coli according to [29].
Table 3 Alternative substrates for aminoacyl-tRNA synthetases tRNA synthetase Other analogs accepted MetRS
LeuRS IleRS
References
norleucine, cis-crotylglycine, [30] 2-aminoheptanoic acid, norvaline, 2-butynylglycine, allylglycine, ethonine norvaline [13,16,31] [24,32] ß-methyl-norleucine, (valine)
Current Opinion in Biotechnology 2018, 53:248–253
acceptance of these amino acids by different aminoacyltRNA synthetases despite very effective control mechanisms (summarized in Table 3). In summary, norvaline is mainly accepted instead of leucine by leucyl-tRNA synthetase (LeuRS), b-methylnorleucine by isoleucyl-tRNA synthetase, and norleucine by methionyl-tRNA synthetase (Figure 2). The specific mechanism is mainly known from recent work on LeuRS. Accumulated norvaline can be activated and transfered to tRNALeu because LeuRS does not discriminate well against norvaline in the synthetic reaction of aminoacylation. However, intrinsic hydrolytic editing activity of LeuRS prevents accumulation of norvalyl-tRNALeu for protein synthesis under normal conditions. Yet, perturbation of kinetic partitioning of aminoacylated tRNA between hydrolysis at the LeuRS editing site and dissociation from the enzyme may promote accumulation of mischarged norvalyl-tRNALeu [11]. This mischarged tRNA is not recognized during translational proofreading [12] and thus leads finally to norvaline-containing proteins. This misincorporation of norvaline causes strong growth inhibition. This was recently shown by supplementation of a cultivation medium with norvaline or cultivation under microaerobic conditions, that is, under oxygen limitation, of an E. coli strain defective in LeuRS editing [13]. The growth inhibition was clearly related to incorporation of norvaline at leucine places into the cellular proteome [14]. Interestingly, also a wild-type strain can misincorporate norvaline under microaerobic growth conditions at low but measurable extent (0.3%) [14]. This is interesting, as these conditions resemble the situation when norvaline is produced in industrial scale bioreactors. Experiments in a scale down twocompartment reactor where one of the compartments resembles the feed zone of a large industrial scale showed the intracellular accumulation of pyruvate in the feed compartment and subsequently also of norvaline [15]. In follow up studies also b-methylnorleucine and norleucine were detected in such scale down systems (non-published own data). www.sciencedirect.com
Incorporation of Non-canonical branched-chain amino acids Reitz, Fan and Neubauer 251
Figure 2
Leucine
Isoleucine
Methionine
Norvaline
β-Methylnorleucine
Norleucine
Current Opinion in Biotechnology
Leucine, isoleucine, and methionine in comparison to their isostructural analogs norvaline, b-methylnorleucine, and norleucine.
Approaches to limit misincorporation of branched-chain non-canonical amino acids into recombinant proteins Although conditions which cause the synthesis and incorporation of non-canonical branched-chain amino acids in E. coli cultivations are not entirely clear, different strategies can prevent the incorporation of norvaline or norleucine, respectively, into recombinant proteins produced in E. coli. For norleucine already Bogosian et al. [7] proposed the addition of methionine into the cultivation medium to increase the intracellular ratio between methionine and norleucine. Similiarly, also the addition of leucine could diminish norvaline incorporation into recombinant proteins [16]. These studies reveal the potential of supplementation as a controlling tool for guaranteeing the quality of recombinant proteins. Though, particularly in industrial scale bioprocesses extra feeding solutions would raise operative complexity and increase process costs.
[15]. The high influx of glucose under these conditions leads to an intracellular buildup of pyruvate, which is directly converted to a-ketobutyrate and further to these amino acids. In this case addition of the trace elements molybdenum, selenium, and nickel decreases the accumulation of norvaline and norleucine [18] as these are essential co-factors for the formic acid-hydrogen lyase metalloprotein complex, which is the key protein in the anaerobic mixed-acid fermentation pathway to reduce and prevent pyruvate accumulation in E. coli [19]. Also, when these trace elements are added, no formic acid accumulation in high cell density cultivations can be seen, as this enzyme splits formic acid into CO2 and H2 [20].
As an alternative to addition of the critical amino acid, overproduction of it can be achieved by strain engineering. As proof of principle a genetically engineered E. coli host was generated which overproduced methionine via several chromosomal mutations (metA, metK, metJ). Overproduction of methionine prevented norleucine incorporation in this strain without impacting cultivation performance or yield rates negatively [17].
The incorporation of the non-canonical amino acid is codon dependent. Recently Cveteic et al. [14] demonstrated that norvaline/leucine incorporations are higher at CTG codons (that also is the main isoacceptor) compared to the other Leu codons. Although the background is unclear, distinct different aminoacylation kinetics of the tRNALeu(CAG) isoacceptor or different interaction of this norvaline charged tRNALeu(CAG) isoacceptor with the translational apparatus have been hypothesized. Further investigations on this may provide a solution to avoid incorporation of norvaline into recombinant proteins. Unfortunately, this would not be a solution for methionine which only has one isoacceptor.
Cultivation conditions and composition of the medium can strongly influence the buildup of the non-canonical branched-chain amino acids. Norvaline and norleucine accumulation in recombinant E. coli is seen in cultivations with a concurrent excess of glucose and oxygen limitation
While it seems difficult to reduce the promiscuity of the enzymes of the leucine and valine/isoleucine synthetic routes to avoid the chain elongation to the noncanonical branched-chain amino acids, several control mechanisms have been established in the natural evolution at the level
www.sciencedirect.com
Current Opinion in Biotechnology 2018, 53:248–253
252 Pharmaceutical biotechnology
of aminoacyl tRNA synthetases (AARS). Different AARS families use different mechanisms to discriminate efficiently against non-cognate amino acids, either in the aminoacylation reaction or by post-transfer editing activities, that is, by hydrolysis of erroneously activated amino acids or misaminoacylated tRNAs [21]. While disabling the post-transfer editing of several tRNA synthetases (IleRS, LeuRS, ValRS) has been shown in several cases as a mechanism to facilitate the incorporation of noncanonical amino acids ([22] and citations therein), so far to our knowledge no investigations were performed to engineer or to evolute tRNA synthetases for an increased discrimination for their natural amino acids. New opportunities appear by the availability of the tRNA synthetase structures and the increasing understanding of the editing mechanisms. Also the editing mechanisms may be different for different organisms, as has been shown for IleRS [23]. IleRS from Streptomyces griseus does not have tRNAdependent pre-transfer editing while E. coli has, thus recombinant replacement strategies may be a feasible solution. Nevertheless, so far it remains an open but interesting question, whether it would be possible to produce an IleRS which is more accurate than that of E. coli.
Conflict of interest statement Nothing declared.
Acknowledgements The authors acknowledge a PhD grant to CR by Sanofi Chimie. We also thank to Ita Gruic-Sovulj for critical reading of the manuscript.
References and recommended reading Papers of particular interest, published within the period of review, have been highlighted as of special interest of outstanding interest 1.
Kubyshkin V, Acevedo-Rocha CG, Budisa N: On universal coding events in protein biogenesis. Biosystems 2017.
2.
Cobucci-Ponzano B, Rossi M, Moracci M: Translational recoding in archaea. Extremophiles 2012, 16:793-803.
3. Moghal A, Mohler K, Ibba M: Mistranslation of the genetic code. FEBS Lett 2014, 588:4305-4310. Comprehensive overview on natural events of mistranslation in prokaryotes and eukaryotes which includes also canonical replacements. Aside from a detailed summary of the quality control mechanisms of aminoacyl tRNA synthetases the analytical techniques are described. 4.
5.
6.
7.
Kubyshkin V, Budisa N: Synthetic alienation of microbial organisms by using genetic code engineering: why and how? Biotechnol J 2017, 12. Achenbach J, Jahnz M, Bethge L, Paal K, Jung M, Schuster M, Albrecht R, Jarosch F, Nierhaus KH, Klussmann S: Outwitting EFTu and the ribosome: translation with D-amino acids. Nucleic Acids Res 2015, 43:5687-5698. Kisumi M, Sugiura M, Chibata I: Biosynthesis of norvaline, norleucine, and homoisoleucine in Serratia marcescens. J Biochem 1976, 80:333-339. Bogosian G, Violand BN, Dorward-King EJ, Workman WE, Jung PE, Kane JF: Biosynthesis and incorporation into protein of norleucine by Escherichia coli. J Biol Chem 1989, 264:531539.
Current Opinion in Biotechnology 2018, 53:248–253
8.
Sycheva EV, Iampol’skaia TA, Preobrazhenskaia ES, Novikova AE, Matrosov NG, Stoinova NV: Overproduction of noncanonical amino acids by Escherichia coli cells. Mikrobiologiia 2007, 76:805-812.
Harris RP, Kilby PM: Amino acid misincorporation in recombinant biopharmaceutical products. Curr Opin Biotechnol 2014, 30:45-50. Summary of the mechanisms and documentation of appearances of misincorporation of natural amino acids in different organisms. Also the analytical methods are presented to verify these modifications.
9.
10. Fenton DM, Lai H, Lu HS, Mann MB, Tsai L: Control of norleucine incorporation into recombinant proteins. In: US Patent (1997). N, Perona JJ: Gruic-Sovulj I Kinetic partitioning 11. Cveteic between synthetic and editing pathways in class I aminoacyltRNA synthetases occurs at both pre-transfer and posttransfer hydrolytic steps. J Biol Chem 2012, 287:25381-25394. N, Akmacic I, Gruic-Sovulj I: Lack of discrimination 12. Cveteic against non-proteinogenic amino acid norvaline by elongation factor Tu from Escherichia coli. Croat Chem Acta 2013, 86:7382. N, Palencia A, Halasz I, Cusack S, Gruic-Sovulj I: The 13. Cveteic physiological target for LeuRS translational quality control is norvaline. EMBO J 2014, 33:1639-1653. The editing activity ofE. coli leucyl-tRNA synthetase is not required to prevent incorrect isoleucine incorporation, but to prevent misincorporation of norvaline. This translational quality control is a key feature for bacterial adaptive response to oxygen deprivation. N, Semanjski M, Soufi B, Krug K, Gruic-Sovulj I, Macek B: 14. Cveteic Proteome-wide measurement of non-canonical bacterial mistranslation by quantitative mass spectrometry of protein modifications. Sci Rep 2016, 6:28631. In vivo mistranslation was quantiatively analysed by shotgun proteomics combined with unbiased protein modification analysis in an E. coli strain with a defect in the editing mechanism of leucyl-tRNA synthetase and its isogenic wildtype. Misincorporation of norvaline on 10% of all measured leucine residues was detected under microaerobic conditions. Interestingly under these conditions also a significant misincorporation was found in the wildtype strain. 15. Soini J, Falschlehner C, Liedert C, Bernhardt J, Vuoristo J, Neubauer P: Norvaline is accumulated after a down-shift of oxygen in Escherichia coli W3110. Microb Cell Fact 2008, 7:30. 16. Apostol I, Levine J, Lippincott J, Leach J, Hess E, Glascock CB, Weickert MJ, Blackmore R: Incorporation of norvaline at leucine positions in recombinant human hemoglobin expressed in Escherichia coli. J Biol Chem 1997, 272:28980-28988. 17. Veeravalli K, Laird MW, Fedesco M, Zhang Y, Yu XC: Strain engineering to prevent norleucine incorporation during recombinant protein production in Escherichia coli. Biotechnol Prog 2015, 31:204-211. 18. Biermann M, Linnemann J, Knu¨pfer U, Vollsta¨dt S, Bardl B, Seidel G, Horn U: Trace element associated reduction of norleucine and norvaline accumulation during oxygen limitation in a recombinant Escherichia coli fermentation. Microb Cell Fact 2013, 12:116. 19. Yoshida A, Nishimura T, Kawaguchi H, Inui M, Yukawa H: Efficient induction of formate hydrogen lyase of aerobically grown Escherichia coli in a three-step biohydrogen production process. Appl Microbiol Biotechnol 2007, 74:754-760. 20. Soini J, Ukkonen K, Neubauer P: High cell density media for escherichia coli are generally designed for aerobic cultivations – consequences for large-scale bioprocesses and shake flask cultures. Microb Cell Fact 2008, 7:26. 21. Perona JJ, Hadd A: Structural diversity and protein engineering of the aminoacyl-tRNA synthetases. Biochemistry 2012, 51:8705-8729. 22. Vo¨ller JS, Dulic M, Gerling-Driessen UI, Biava H, Baumann T, Budisa N, Gruic-Sovulj I, Koksch B: Discovery and investigation of natural editing function against artificial amino acids in protein translation. ACS Cent Sci 2017, 3:73-80. A new way of systematic engineering of aminoacyl tRNA syntehtases for the incorporation of non-canonical amino acids is shown by the www.sciencedirect.com
Incorporation of Non-canonical branched-chain amino acids Reitz, Fan and Neubauer 253
translational incorporation of the fluorine containing amino acid trifluoroethylglycine with a post-transfer editing deficient mutant of IleRS. N, Dulic M, Bilus M, Sostaric N, Lenhard B, Gruic-Sovulj I: 23. Cveteic Naturally occurring isoleucyl-tRNA synthetase without tRNAdependent pre-transfer editing. J Biol Chem 2016, 291:86188631.
28. Ni J, Gao M, James A, Yao J, Yuan T, Carpick B, D’Amore T, Farrell P: Investigation into the misincorporation of norleucine into a recombinant protein vaccine candidate. J Ind Microbiol Biotechnol 2015, 42:971-975.
24. Muramatsu R, Misawa S, Hayashi H: Finding of an isoleucine derivative of a recombinant protein for pharmaceutical use. J Pharm Biomed Anal 2003, 31:979-987.
29. Neidhardt FC, Umbarger HE: Chemical composition of Escherichia coli. In Escherichia coli and salmonella Cellular and Molecular Biology. Edited by Neidhardt FC, Curtis III R, Ingraham JL, Lin CC, Low KB, Magasanik B, Reznikoff WS, Riley M, Schneider D, Umbardger HE. Washington D.C: American Society for Microbiology Press; 1996:13-16.
25. Sunasara KM, Cramer SM, Hauer CR, Rupp RG, Shoup VA: Characterization of recombinant human brain-derived neurotrophic factor variants. Arch Biochem Biophys 1999, 372:248-260.
30. Kiick KL, Weberskirch R, Tirrell DA: Identification of an expanded set of translationally active methionine analogues in Escherichia coli. FEBS Lett 2001, 502:25-30.
26. Lu HS, Tsai LB, Kenney WC, Lai PH: Identification of unusual replacement of methionine by norleucine in recombinant interleukin-2 produced by E. coli. Biochem Biophys Res Commun 1988, 156:807-813.
31. Tang Y, Tirell DA: Attenuation of the editing activity of the Escherichia coli leucyl-tRNA synthetase allows incorporation of novel amino acids into proteins in vivo. Biochem 2002, 41:10635-10645.
27. Tsai LB, Lu HS, Kenney WC, Curless CC, Klein ML, Lai PH, Fenton DM, Altrock BW, Mann MB: Control of misincorporation of de novo synthesized norleucine into recombinant interleukin-2 in E. coli. Biochem Biophys Res Commun 1988, 156:733-739.
www.sciencedirect.com
N, Bilus M, Gruic-Sovulj I: The tRNA A76 hydroxyl groups 32. Cveteic control partitioning of the tRNA-dependent pre- and posttransfer editing pathways in class I tRNA synthetase. J Biol Chem 2015, 290:13981-13991.
Current Opinion in Biotechnology 2018, 53:248–253
ScienceDirect Synthesis of non-canonical branched-chain amino acids in Escherichia coli and approaches to avoid their incorporation into recombinant proteins Christian Reitz, Qin Fan and Peter Neubauer Abstract In E. coli the non-canonical amino acids acids norvaline, norleucine, and b-methylnorleucine, which derive from an offpathway of the branched-chain amino acid synthesis route are synthesized and incorporated into cellular and recombinant proteins. The synthesis of these amino acids is supported by a high flux of glucose through the glycolytic pathway in combination with a derepression of the enzymes of the branched chain amino acid pathway, for example, when leucine-rich proteins are produced. Avoiding the synthesis and misincorporation of these amino acids has been challenging, especially in large-scale pharmaceutical processes where the problem is boosted by the typical fed-batch production and the technical limitation of mass transfer in the bioreactors.Despite its industrial importance, so far this issue has not been discussed comprehensively. Therefore this paper reviews, firstly, the specific pathway of the non-canonical branched chain amino acids starting at pyruvate, secondly, the molecular factors for their misincorporation, and thirdly, approaches to avoid this misincoporation. While the synthesis of these amino acids is difficult to prevent due to the broad promiscuity of the connected enzymes, recent studies on the control mechanisms of aminoacyl tRNA synthetases open new opportunities to avoid this misincorporation. Address Technische Universita¨t Berlin, Institute of Biotechnology, Department of Bioprocess Engineering, Ackerstr. 76, D-13355 Berlin, Germany Corresponding author: Neubauer, Peter ([email protected])
Current Opinion in Biotechnology 2018, 53:248–253 This review comes from a themed issue on Pharmaceutical biotechnology Edited by Amanda Lewis and Nripen Singh
https://doi.org/10.1016/j.copbio.2018.05.003 0958-1669/ã 2018 Elsevier Ltd. All rights reserved.
Introduction The genetic code is almost universal and other amino acids than the 20 canonical amino acids are generally not accepted in nature [1] with very few exceptions. Current Opinion in Biotechnology 2018, 53:248–253
Examples for such non-typical amino acids are selenocysteine and pyrrolysine in archae, that are incorporated into the peptide chain by reprogramming of one of the rarely used stop codons (UGA in connection to a downstream located specific mRNA hairpin structure for selenocystein, UAG without the need for a downstream element for pyrrolysine) [2]. Also, despite the universality of the genetic code, translational errors can occur, which is called mistranslation. The degree of mistranslation varies in different organisms and even can be beneficial in certain circumstances [3]. During the last decade also the efficient artificial translational incorporation of noncanonical amino acids in proteins became possible either by engineering of the genetic code or by replacement of the natural amino acid with a chemical homolog which is accepted by its aminoacyl tRNA synthetase [4]. in vitro investigations indicated a surprising plasticity of certain natural aminoacyl tRNA synthetases and ribosomal peptidyltransferases which even allows the incorporation of D-amino acids [5]. In non-engineered organisms translational misincorporation of non-canonical amino acids has been mainly found for the amino acids norvaline, norleucine and b-methylnorleucine which are side products of the branched chain amino acid pathway. Based on early studies with Serratia marcenscens the synthesis pathway for these amino acids was characterized [6]. As illustrated in Figure 1, norvaline, norleucine, and b-methylnorleucine are products of enzymatic elongation reactions beginning with a-ketobutyrate catalyzed by a-isopropylmalate synthase (PMS) (leuA), a-isopropylmalate isomerase (ISOM) (leuCD), and a-isopropylmalate dehydrogenase (IPMD) (leuB) and a final transamination step. Branching to norleucine includes a second cycle with these enzymes, while b-methylnorleucine synthesis proceeds from a-ketovalerate with the enzymes of the isoleucine/valine pathway. This pathway is triggered by conditions which lead to a lower intracellular level of leucine and conseqently cause a derepression of the leucine operon [7]. Additionally, the accumulation of a-ketobutyrate is a requirement since the affinity of a-isopropylmalate synthase (IPMS) for a-ketobutyrate is significantly lower compared to its normal substrate a-ketoisovalerate as characterised for S. marcescens [6]. Sycheva et al. [8] demonstrated that the synthesis of norvaline and norleucine also proceeds after inactivation of threonine deaminase (ilvA), indicating an alternative shortcut pathway, beside the aspartate threonine pathway, from pyruvate to a-ketobutyrate, a reaction which is www.sciencedirect.com
Incorporation of Non-canonical branched-chain amino acids Reitz, Fan and Neubauer 249
Figure 1
pyruvate
pyr CO2
ilvD
ilvC
AHAS
α-ketoisovalerate
α,β-dihydroxyisovalerate
α-acetolactate
ilvE, avtA
valine
leuA leuA leuCD leuB
ilvE, tyrB
leuCD
Isopropylmalate
leuB
AHAS
ilvA
ilvC
α-ketoisocaproate
ilvE, avtA
ilvD
pyr CO2
threonine
α-ketobutyrate
Leucine
Isoleucine
α-aceto-α-hydroxybutyrate α,β-dihydroxy-β-methylvalerate α-keto-β-methylvalerate
leuA leuCD leuB AHAS
ilvC
pyr CO2
ilvE
ilvD
α-ketovalerate α-aceto-α-hydroxyvalerate α,β-dihydroxy-β-methylcaproate α-keto-β-methylcaproate leuA
β-methylnorleucine
ilvE, avtA
leuCD leuB
Norvaline
ilvE, tyrA
α-ketocaproate
Norleucine Current Opinion in Biotechnology
Schematic view of predicted biosynthetic pathway of the modified branched-chain amino acids including norvaline, norleucine and b-methylnorleucine from pyruvate via the so-called “ketoacid chain elongation pathway” over a-ketobutyrate and a-ketovalerate to a-ketocaproate facilitated by the promiscuous enzymes of the (iso)-leucine biosynthetic pathway in E. coli.
also possible by the enzymes of the leuABCD operon (see Figure 1). While on one side the area of synthetic biology considers today the extension of the genetic code and the incorporation of non-canonical amino acids as an interesting issue to expand the functionality of proteins, the uncontrolled misincorporation of self-synthesized non-canonical proteins is a challenging problem in the production of biopharmaceuticals on the other side. Recently the problem of amino acid misincorporation was comprehensively reviewed in an excellent paper by Harris and Kilby [9]. The challenge in the pharmaceutical production of proteins is that such rare events of incorporation errors produce incorrect protein variants which eventually could induce immunological and other responses. Therefore it is a major concern to avoid or minimize these errors, which must be documented by the producer according to the ICH Q6B guideline. The previous review focused on the misincorporation of natural amino acids, however, www.sciencedirect.com
despite the industrial importance, misincorporation of non-canonical amino acids including norvaline, norleucine and b-methylnorleucine in recombinant E. coli cultures was not considered although it is a serious problem in the production of several pharmaceutically relevant proteins (for a detailed summary see Table 1).
Misincorporation of non-canonical amino acids into recombinant proteins Overall, the misincorporation of non-canonical amino acids in E. coli is influenced by cultivation conditions and also depends on the composition of the protein. Particularly the production of leucine-rich recombinant proteins seems to be critical [10]. This is due to the high need of leucine after induction. Consequently the intracellular concentration of leucine decreases and the leucine operon may be derepressed (i.e., activated). If this coincides with conditions which lead to an elevated synthesis of pyruvate, such as a high flux into the glycolytic pathway, the Current Opinion in Biotechnology 2018, 53:248–253
250 Pharmaceutical biotechnology
Table 1 Summary of references for the incorporation of modified amino acids derived from the branched chain amino acid pathway into recombinant proteins Amino acid
Product
NL NL NL
Hirudin in E. coli Bovine somatotropin in E. coli human brain-derived neurotrophic factor in E. coli interleukin-2 in E. coli Met-rich vaccine candidate Hirudin in E. coli Rec. Hemoglobin in E. coli Proinsulin in E. coli
NL NL b-MNL Norvaline Norvaline, ß-MNL, NL
Comment
AA composition
Ref.
in Met Positions
27 leu/191aa
[24] [7] [25]
in in in in
26 Leu/152 aa
Met Positions Met Positions Ile positions Leu positions
72 Leu/575 aa
[26,27] [28] [24] [16] Non-published own results
NL – norleucine, b-MNL – b-methylnorleucine.
non-canonical branched chain amino acids are synthesized by direct chain elongation of pyruvate by the enzymes of the leucine operon, and eventually incorporated into the recombinant product. This scenario is relevant, as some of the pharmaceutically interesting proteins have a high leucine content compared to an average E. coli protein; for example, recombinant hemoglobin and interleukin-2 have a leucine ratio of 13 and 17%, respectively, while an average E. coli protein only contains 8.4% (Table 2). Incorporation of norvaline, b-methylnorleucine, and norleucine into a peptide chain occur by the erroneous
Table 2 Relative content of the different branched-chain amino acids in pharmaceutically interesting proteins (calculated according to the NCBI CCDS database)
a
E. coli Interleukin-2 Human proinsulin Hemoglobin Human growth hormone Bovine Somatrotropin (MBS) Interferon a2 Interferon b1 a
Leucine
Valine
Isoleucine
8.4% 17.6% 17.3% 12.6% 14.7% 16.1% 14.3% 16.0%
7.9% 3.3% 6.9% 9.2% 3.2% 3.7% 5.3% 2.7%
5.4% 6.5% 2.3% 0.0% 3.2% 3.7% 3.7% 6.4%
Content of overall protein in E. coli according to [29].
Table 3 Alternative substrates for aminoacyl-tRNA synthetases tRNA synthetase Other analogs accepted MetRS
LeuRS IleRS
References
norleucine, cis-crotylglycine, [30] 2-aminoheptanoic acid, norvaline, 2-butynylglycine, allylglycine, ethonine norvaline [13,16,31] [24,32] ß-methyl-norleucine, (valine)
Current Opinion in Biotechnology 2018, 53:248–253
acceptance of these amino acids by different aminoacyltRNA synthetases despite very effective control mechanisms (summarized in Table 3). In summary, norvaline is mainly accepted instead of leucine by leucyl-tRNA synthetase (LeuRS), b-methylnorleucine by isoleucyl-tRNA synthetase, and norleucine by methionyl-tRNA synthetase (Figure 2). The specific mechanism is mainly known from recent work on LeuRS. Accumulated norvaline can be activated and transfered to tRNALeu because LeuRS does not discriminate well against norvaline in the synthetic reaction of aminoacylation. However, intrinsic hydrolytic editing activity of LeuRS prevents accumulation of norvalyl-tRNALeu for protein synthesis under normal conditions. Yet, perturbation of kinetic partitioning of aminoacylated tRNA between hydrolysis at the LeuRS editing site and dissociation from the enzyme may promote accumulation of mischarged norvalyl-tRNALeu [11]. This mischarged tRNA is not recognized during translational proofreading [12] and thus leads finally to norvaline-containing proteins. This misincorporation of norvaline causes strong growth inhibition. This was recently shown by supplementation of a cultivation medium with norvaline or cultivation under microaerobic conditions, that is, under oxygen limitation, of an E. coli strain defective in LeuRS editing [13]. The growth inhibition was clearly related to incorporation of norvaline at leucine places into the cellular proteome [14]. Interestingly, also a wild-type strain can misincorporate norvaline under microaerobic growth conditions at low but measurable extent (0.3%) [14]. This is interesting, as these conditions resemble the situation when norvaline is produced in industrial scale bioreactors. Experiments in a scale down twocompartment reactor where one of the compartments resembles the feed zone of a large industrial scale showed the intracellular accumulation of pyruvate in the feed compartment and subsequently also of norvaline [15]. In follow up studies also b-methylnorleucine and norleucine were detected in such scale down systems (non-published own data). www.sciencedirect.com
Incorporation of Non-canonical branched-chain amino acids Reitz, Fan and Neubauer 251
Figure 2
Leucine
Isoleucine
Methionine
Norvaline
β-Methylnorleucine
Norleucine
Current Opinion in Biotechnology
Leucine, isoleucine, and methionine in comparison to their isostructural analogs norvaline, b-methylnorleucine, and norleucine.
Approaches to limit misincorporation of branched-chain non-canonical amino acids into recombinant proteins Although conditions which cause the synthesis and incorporation of non-canonical branched-chain amino acids in E. coli cultivations are not entirely clear, different strategies can prevent the incorporation of norvaline or norleucine, respectively, into recombinant proteins produced in E. coli. For norleucine already Bogosian et al. [7] proposed the addition of methionine into the cultivation medium to increase the intracellular ratio between methionine and norleucine. Similiarly, also the addition of leucine could diminish norvaline incorporation into recombinant proteins [16]. These studies reveal the potential of supplementation as a controlling tool for guaranteeing the quality of recombinant proteins. Though, particularly in industrial scale bioprocesses extra feeding solutions would raise operative complexity and increase process costs.
[15]. The high influx of glucose under these conditions leads to an intracellular buildup of pyruvate, which is directly converted to a-ketobutyrate and further to these amino acids. In this case addition of the trace elements molybdenum, selenium, and nickel decreases the accumulation of norvaline and norleucine [18] as these are essential co-factors for the formic acid-hydrogen lyase metalloprotein complex, which is the key protein in the anaerobic mixed-acid fermentation pathway to reduce and prevent pyruvate accumulation in E. coli [19]. Also, when these trace elements are added, no formic acid accumulation in high cell density cultivations can be seen, as this enzyme splits formic acid into CO2 and H2 [20].
As an alternative to addition of the critical amino acid, overproduction of it can be achieved by strain engineering. As proof of principle a genetically engineered E. coli host was generated which overproduced methionine via several chromosomal mutations (metA, metK, metJ). Overproduction of methionine prevented norleucine incorporation in this strain without impacting cultivation performance or yield rates negatively [17].
The incorporation of the non-canonical amino acid is codon dependent. Recently Cveteic et al. [14] demonstrated that norvaline/leucine incorporations are higher at CTG codons (that also is the main isoacceptor) compared to the other Leu codons. Although the background is unclear, distinct different aminoacylation kinetics of the tRNALeu(CAG) isoacceptor or different interaction of this norvaline charged tRNALeu(CAG) isoacceptor with the translational apparatus have been hypothesized. Further investigations on this may provide a solution to avoid incorporation of norvaline into recombinant proteins. Unfortunately, this would not be a solution for methionine which only has one isoacceptor.
Cultivation conditions and composition of the medium can strongly influence the buildup of the non-canonical branched-chain amino acids. Norvaline and norleucine accumulation in recombinant E. coli is seen in cultivations with a concurrent excess of glucose and oxygen limitation
While it seems difficult to reduce the promiscuity of the enzymes of the leucine and valine/isoleucine synthetic routes to avoid the chain elongation to the noncanonical branched-chain amino acids, several control mechanisms have been established in the natural evolution at the level
www.sciencedirect.com
Current Opinion in Biotechnology 2018, 53:248–253
252 Pharmaceutical biotechnology
of aminoacyl tRNA synthetases (AARS). Different AARS families use different mechanisms to discriminate efficiently against non-cognate amino acids, either in the aminoacylation reaction or by post-transfer editing activities, that is, by hydrolysis of erroneously activated amino acids or misaminoacylated tRNAs [21]. While disabling the post-transfer editing of several tRNA synthetases (IleRS, LeuRS, ValRS) has been shown in several cases as a mechanism to facilitate the incorporation of noncanonical amino acids ([22] and citations therein), so far to our knowledge no investigations were performed to engineer or to evolute tRNA synthetases for an increased discrimination for their natural amino acids. New opportunities appear by the availability of the tRNA synthetase structures and the increasing understanding of the editing mechanisms. Also the editing mechanisms may be different for different organisms, as has been shown for IleRS [23]. IleRS from Streptomyces griseus does not have tRNAdependent pre-transfer editing while E. coli has, thus recombinant replacement strategies may be a feasible solution. Nevertheless, so far it remains an open but interesting question, whether it would be possible to produce an IleRS which is more accurate than that of E. coli.
Conflict of interest statement Nothing declared.
Acknowledgements The authors acknowledge a PhD grant to CR by Sanofi Chimie. We also thank to Ita Gruic-Sovulj for critical reading of the manuscript.
References and recommended reading Papers of particular interest, published within the period of review, have been highlighted as of special interest of outstanding interest 1.
Kubyshkin V, Acevedo-Rocha CG, Budisa N: On universal coding events in protein biogenesis. Biosystems 2017.
2.
Cobucci-Ponzano B, Rossi M, Moracci M: Translational recoding in archaea. Extremophiles 2012, 16:793-803.
3. Moghal A, Mohler K, Ibba M: Mistranslation of the genetic code. FEBS Lett 2014, 588:4305-4310. Comprehensive overview on natural events of mistranslation in prokaryotes and eukaryotes which includes also canonical replacements. Aside from a detailed summary of the quality control mechanisms of aminoacyl tRNA synthetases the analytical techniques are described. 4.
5.
6.
7.
Kubyshkin V, Budisa N: Synthetic alienation of microbial organisms by using genetic code engineering: why and how? Biotechnol J 2017, 12. Achenbach J, Jahnz M, Bethge L, Paal K, Jung M, Schuster M, Albrecht R, Jarosch F, Nierhaus KH, Klussmann S: Outwitting EFTu and the ribosome: translation with D-amino acids. Nucleic Acids Res 2015, 43:5687-5698. Kisumi M, Sugiura M, Chibata I: Biosynthesis of norvaline, norleucine, and homoisoleucine in Serratia marcescens. J Biochem 1976, 80:333-339. Bogosian G, Violand BN, Dorward-King EJ, Workman WE, Jung PE, Kane JF: Biosynthesis and incorporation into protein of norleucine by Escherichia coli. J Biol Chem 1989, 264:531539.
Current Opinion in Biotechnology 2018, 53:248–253
8.
Sycheva EV, Iampol’skaia TA, Preobrazhenskaia ES, Novikova AE, Matrosov NG, Stoinova NV: Overproduction of noncanonical amino acids by Escherichia coli cells. Mikrobiologiia 2007, 76:805-812.
Harris RP, Kilby PM: Amino acid misincorporation in recombinant biopharmaceutical products. Curr Opin Biotechnol 2014, 30:45-50. Summary of the mechanisms and documentation of appearances of misincorporation of natural amino acids in different organisms. Also the analytical methods are presented to verify these modifications.
9.
10. Fenton DM, Lai H, Lu HS, Mann MB, Tsai L: Control of norleucine incorporation into recombinant proteins. In: US Patent (1997). N, Perona JJ: Gruic-Sovulj I Kinetic partitioning 11. Cveteic between synthetic and editing pathways in class I aminoacyltRNA synthetases occurs at both pre-transfer and posttransfer hydrolytic steps. J Biol Chem 2012, 287:25381-25394. N, Akmacic I, Gruic-Sovulj I: Lack of discrimination 12. Cveteic against non-proteinogenic amino acid norvaline by elongation factor Tu from Escherichia coli. Croat Chem Acta 2013, 86:7382. N, Palencia A, Halasz I, Cusack S, Gruic-Sovulj I: The 13. Cveteic physiological target for LeuRS translational quality control is norvaline. EMBO J 2014, 33:1639-1653. The editing activity ofE. coli leucyl-tRNA synthetase is not required to prevent incorrect isoleucine incorporation, but to prevent misincorporation of norvaline. This translational quality control is a key feature for bacterial adaptive response to oxygen deprivation. N, Semanjski M, Soufi B, Krug K, Gruic-Sovulj I, Macek B: 14. Cveteic Proteome-wide measurement of non-canonical bacterial mistranslation by quantitative mass spectrometry of protein modifications. Sci Rep 2016, 6:28631. In vivo mistranslation was quantiatively analysed by shotgun proteomics combined with unbiased protein modification analysis in an E. coli strain with a defect in the editing mechanism of leucyl-tRNA synthetase and its isogenic wildtype. Misincorporation of norvaline on 10% of all measured leucine residues was detected under microaerobic conditions. Interestingly under these conditions also a significant misincorporation was found in the wildtype strain. 15. Soini J, Falschlehner C, Liedert C, Bernhardt J, Vuoristo J, Neubauer P: Norvaline is accumulated after a down-shift of oxygen in Escherichia coli W3110. Microb Cell Fact 2008, 7:30. 16. Apostol I, Levine J, Lippincott J, Leach J, Hess E, Glascock CB, Weickert MJ, Blackmore R: Incorporation of norvaline at leucine positions in recombinant human hemoglobin expressed in Escherichia coli. J Biol Chem 1997, 272:28980-28988. 17. Veeravalli K, Laird MW, Fedesco M, Zhang Y, Yu XC: Strain engineering to prevent norleucine incorporation during recombinant protein production in Escherichia coli. Biotechnol Prog 2015, 31:204-211. 18. Biermann M, Linnemann J, Knu¨pfer U, Vollsta¨dt S, Bardl B, Seidel G, Horn U: Trace element associated reduction of norleucine and norvaline accumulation during oxygen limitation in a recombinant Escherichia coli fermentation. Microb Cell Fact 2013, 12:116. 19. Yoshida A, Nishimura T, Kawaguchi H, Inui M, Yukawa H: Efficient induction of formate hydrogen lyase of aerobically grown Escherichia coli in a three-step biohydrogen production process. Appl Microbiol Biotechnol 2007, 74:754-760. 20. Soini J, Ukkonen K, Neubauer P: High cell density media for escherichia coli are generally designed for aerobic cultivations – consequences for large-scale bioprocesses and shake flask cultures. Microb Cell Fact 2008, 7:26. 21. Perona JJ, Hadd A: Structural diversity and protein engineering of the aminoacyl-tRNA synthetases. Biochemistry 2012, 51:8705-8729. 22. Vo¨ller JS, Dulic M, Gerling-Driessen UI, Biava H, Baumann T, Budisa N, Gruic-Sovulj I, Koksch B: Discovery and investigation of natural editing function against artificial amino acids in protein translation. ACS Cent Sci 2017, 3:73-80. A new way of systematic engineering of aminoacyl tRNA syntehtases for the incorporation of non-canonical amino acids is shown by the www.sciencedirect.com
Incorporation of Non-canonical branched-chain amino acids Reitz, Fan and Neubauer 253
translational incorporation of the fluorine containing amino acid trifluoroethylglycine with a post-transfer editing deficient mutant of IleRS. N, Dulic M, Bilus M, Sostaric N, Lenhard B, Gruic-Sovulj I: 23. Cveteic Naturally occurring isoleucyl-tRNA synthetase without tRNAdependent pre-transfer editing. J Biol Chem 2016, 291:86188631.
28. Ni J, Gao M, James A, Yao J, Yuan T, Carpick B, D’Amore T, Farrell P: Investigation into the misincorporation of norleucine into a recombinant protein vaccine candidate. J Ind Microbiol Biotechnol 2015, 42:971-975.
24. Muramatsu R, Misawa S, Hayashi H: Finding of an isoleucine derivative of a recombinant protein for pharmaceutical use. J Pharm Biomed Anal 2003, 31:979-987.
29. Neidhardt FC, Umbarger HE: Chemical composition of Escherichia coli. In Escherichia coli and salmonella Cellular and Molecular Biology. Edited by Neidhardt FC, Curtis III R, Ingraham JL, Lin CC, Low KB, Magasanik B, Reznikoff WS, Riley M, Schneider D, Umbardger HE. Washington D.C: American Society for Microbiology Press; 1996:13-16.
25. Sunasara KM, Cramer SM, Hauer CR, Rupp RG, Shoup VA: Characterization of recombinant human brain-derived neurotrophic factor variants. Arch Biochem Biophys 1999, 372:248-260.
30. Kiick KL, Weberskirch R, Tirrell DA: Identification of an expanded set of translationally active methionine analogues in Escherichia coli. FEBS Lett 2001, 502:25-30.
26. Lu HS, Tsai LB, Kenney WC, Lai PH: Identification of unusual replacement of methionine by norleucine in recombinant interleukin-2 produced by E. coli. Biochem Biophys Res Commun 1988, 156:807-813.
31. Tang Y, Tirell DA: Attenuation of the editing activity of the Escherichia coli leucyl-tRNA synthetase allows incorporation of novel amino acids into proteins in vivo. Biochem 2002, 41:10635-10645.
27. Tsai LB, Lu HS, Kenney WC, Curless CC, Klein ML, Lai PH, Fenton DM, Altrock BW, Mann MB: Control of misincorporation of de novo synthesized norleucine into recombinant interleukin-2 in E. coli. Biochem Biophys Res Commun 1988, 156:733-739.
www.sciencedirect.com
N, Bilus M, Gruic-Sovulj I: The tRNA A76 hydroxyl groups 32. Cveteic control partitioning of the tRNA-dependent pre- and posttransfer editing pathways in class I tRNA synthetase. J Biol Chem 2015, 290:13981-13991.
Current Opinion in Biotechnology 2018, 53:248–253