- Email: [email protected]
In vitro selection of a 3′ terminal short protector that stabilizes transcripts to improve the translation efficiency in a wheat germ extract
Bioorganic & Medicinal Chemistry Letters xxx (xxxx) xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters journal homepage: www.elsevier.com/locate/bmcl
In vitro selection of a 3′ terminal short protector that stabilizes transcripts to improve the translation efficiency in a wheat germ extract Atsushi Ogawa , Akane Kutsuna, Masashi Takamatsu, Tatsuya Okuzono ⁎
Proteo-Science Center, Ehime University, 3 Bunkyo-cho, Matsuyama, Ehime 790-8577, Japan
ARTICLE INFO
ABSTRACT
Keywords: Cell-free translation Wheat germ extract In vitro selection RNA stabilization Translational enhancer
Wheat cell-free expression systems based on wheat germ extract (WGE) enable us to briefly synthesize various types of proteins in vitro merely by exogenously adding their mRNA templates. Moreover, it is possible to produce larger amounts of protein by thoroughly removing the endosperm, which contains many translation inhibitors, including ribonucleases (RNases). However, because small amounts of RNases are also present even in an endosperm-free, high-quality WGE (hqWGE), the in-vitro transcribed mRNA is rapidly degraded. In particular, 3′ exonucleases have been considered as the major RNases that degrade mRNA. We thus herein performed in vitro selection to find an effective, short 3′ protector sequence from a random RNA pool. The selected sequences stabilized in vitro transcripts in the hqWGE more effectively than the previously reported, longer 3′ protectors did. In addition, when one of these 3′ protectors was minimized and then fused to mRNA, the translation efficiency increased 5–6-fold in the hqWGE, mainly due to the mRNA stabilization.
Cell-free expression systems, which are generally composed of a cell extract, allow us to synthesize a protein in vitro merely by exogenous addition of the mRNA template (in vitro transcripts) thereinto.1 Among various types of cell-free systems, wheat germ extract (WGE) has a major advantage for the expression of a broad range of proteins derived from various types of organisms (not only eukaryotes but also viruses and prokaryotes), by virtue of its low codon preference.2 Although WGEs prepared by the traditional method show low productivity due to the presence of many contaminating translation inhibitors, such as ribosome-inactivating proteins, ribonucleases (RNases) and proteases, these contaminants can be considerably eliminated by extensively washing the endosperm from the wheat embryos to prepare a highquality WGE (hqWGE), permitting the production of much larger amounts of protein (several mg/mL).3 Nonetheless, even in an hqWGE, in-vitro transcribed RNA is still unstable and is generally digested within one hour.4 This means that there are non-negligible amounts of RNases derived from embryos. In particular, 3′ exonucleases have been considered as the major nucleases that degrade RNA in the hqWGE.5 In fact, we recently successfully stabilized in vitro transcripts (a 77-nt aptamer) for at least one hour in an hqWGE by ligating a tRNA-based rigid structure named GGCt86 (and a simple stem-loop structure named 5SL) to the 3′ terminus (and the 5′ terminus, respectively) to inhibit degradation by endogenous 3′ exonucleases (and 5′ exonucleases, respectively).4 However, we found that GGCt86 was unable to protect the
⁎
transcripts under longer incubation periods. In addition, GGCt86 is cumbersome due to its moderately long length (94 nt), and is inappropriate for protecting mRNA because of the complete tRNA frame, which could trap eEF-1α and/or the corresponding aminoacyl-tRNA synthetase. Therefore, we herein attempted to obtain a more effective, shorter 3′ protector through in vitro selection6 in order to stabilize transcripts including mRNA templates and ultimately to further improve the translation efficiency in the hqWGE. We first designed a 3′ protector library based on GGCt86 for in vitro selection as follows (Fig. 1A). Because a specific primer binding site is required at the 3′ terminus for amplification during selection, the 3′ terminal 20-nt sequence of GGCt86 was used as is. The 5′ terminal 6-nt sequence was also left under the expectation that it might form a stem with the 3′ terminal region to inhibit 3′ exonucleases. 20 random nucleotides were inserted between them. DNA sequences encoding this half-size 3′ protector library (46 nt) and 5SL (21 nt) were added (with the T7 promoter for the latter) to each side of a 105-bp sequence encoding a protected region (105PR),7 which was unintentionally chosen from a pMK-based plasmid, to prepare a random DNA pool including theoretically all possible sequences (approximately 1012 different DNA molecules; Fig. 1B, top). After the DNA pool was carefully transcribed so as not to decrease the diversity (Fig. 1B, (i)), 50 pmol (3 × 1013 molecules) of the transcripts (5SL/105PR/3′ protector library) were incubated at 26 °C in the hqWGE with several materials optimized for
Corresponding author. E-mail address: [email protected] (A. Ogawa).
https://doi.org/10.1016/j.bmcl.2019.06.058 Received 7 May 2019; Received in revised form 20 June 2019; Accepted 28 June 2019 0960-894X/ © 2019 Elsevier Ltd. All rights reserved.
Please cite this article as: Atsushi Ogawa, et al., Bioorganic & Medicinal Chemistry Letters, https://doi.org/10.1016/j.bmcl.2019.06.058
Bioorganic & Medicinal Chemistry Letters xxx (xxxx) xxx–xxx
A. Ogawa, et al.
Fig. 2. The stability of each of the selected RNAs. (A) The sequences of the random region in the top 8 most-abundant RNAs (4–11: 5SL/105PR/Xp (X = 4–11), wherein Xp is the 3′ protector). The read number in the sequencing and the residual rate after incubation at 26 °C for 24 h in the hqWGE are also shown in the rightmost two columns. (B) SDS-PAGE analysis of the three moststable RNAs (4, 8 and 9) before and after incubation at 26 °C for 24 h in the hqWGE. See Supplementary Fig. S4 for the other five RNAs.
the hqWGE, despite their shorter length. We next read the sequences of the selected RNAs using a nextgeneration sequencer without utilizing genetic engineering. 4,245 of the 32,270 read sequences were randomly chosen to cluster into several groups with the same sequence in the random region.9 Fig. 2A shows the sequences of the random region in the top 8 most-abundant RNAs (4–11: 5SL/105PR/Xp (X = 4–11), wherein Xp is the 3′ protector). To confirm the stability of these 8 RNAs individually, each RNA was incubated at 26 °C in the hqWGE. As shown in Fig. 2B and Supplementary Fig. S4, all transcripts investigated were as stable for 24 h as the selected library (their residual rates were 50–90%; Fig. 2A), meaning that we successfully obtained several effective 3′ protectors through in vitro selection. Because the 3′ protectors in the top 8 RNAs are G-rich without exception, we expected that they are folded into a G-quadruplex.10 We thus picked up two relatively stable RNAs (4 and 8) and checked whether they could form a G-quadruplex structure with a simple assay using Thioflavin T (ThT)11 in the absence of the hqWGE, which includes many biomolecules that could affect the assay. However, both RNAs increased the fluorescence of ThT only moderately, in the same way as a negative control RNA devoid of any 3′ protector, in contrast to a positive control RNA with the 3′ terminal TRF2 G-quadruplex (TRF2),11 which achieved a two-fold enhancement of the fluorescence (Supplementary Fig. S5A). These results indicate that 4 and 8 did not form a G-quadruplex, at least in the 3′ protector region. Although we cannot deny the possibility that the selected protectors were folded into a G-quadruplex in the hqWGE by interacting with endogenous molecules, the G-quadruplex structure at the 3′ terminus was not in itself sufficient for protecting transcripts: the positive control RNA with TRF2 was much more unstable in the hqWGE (Supplementary Fig. S5B). Therefore, it is assumed that the selected 3′ protectors form another rigid structure to prevent endogenous 3′ exonucleases in the hqWGE. To obtain a shorter 3′ protector, we next minimized a 3′ protector
Fig. 1. In vitro selection of 3′ terminal short protectors that stabilize transcripts in the hqWGE. (A) Design of a 3′ protector library (right) based on a tRNAbased 3′ protector, GGCt86 (left). The blue line with N20 represents the 20-nt random region. (B) Schematic illustration of the in vitro selection. In vitro transcripts in the random RNA pool are composed of a 5′ protector (5SL), the protected region (105PR), and the 3′ protector library including N20. (C) SDSPAGE analysis of the initial RNA pool (middle) and the resulting pool after the 11th round of selection (right) before and after incubation at 26 °C for 24 h in the hqWGE. As a control, an exogenous RNA-free solution was also resolved on the same gel (left). Some bands on the control lane that were also detected on other lanes are endogenous components in the hqWGE.4 The reason the band patterns differed between the initial pool and the selected pool is probably that the transcripts in the latter formed specific tertiary structures.
cell-free translation (Fig.1B, (ii)). The undigested RNA was then recovered and amplified by RT-PCR (Fig. 1B, (iii)). We repeated this transcription/incubation/amplification cycle 11 times with progressively increasing stringency (especially with respect to the incubation time; Supplementary Fig. S1). The residual amount of RNA in each round was evaluated with SDS-PAGE, which revealed that the stability of transcripts gradually increased as the selection proceeded (until the 10th round; Supplementary Fig. S1). The resulting RNA pool after the 11th round was stable at 26 °C for 24 h, whereas almost all of the initial pool was degraded within 1 h (Fig. 1C and Supplementary Fig. S2). To compare the protecting ability of the selected 3′ library with those of other 3′ protectors reported previously, we changed the 3′ library region (46 nt) to GGCt86 (94 nt)4 or a 59-nt triple helix (TH).8 Although both RNAs with these 3′ protectors were stable at 26 °C for at least 1 h as previously reported, they were completely digested in 24 h (Supplementary Fig. S3). These results clearly show that the selected protectors are much more effective for stabilizing in vitro transcripts in 2
Bioorganic & Medicinal Chemistry Letters xxx (xxxx) xxx–xxx
A. Ogawa, et al.
Fig. 3. Minimization of 4p in RNA 4. (A) The proposed secondary structures of 4p and the 10 nt-cropped ones (4p3-10 and 4p5-10). (B and C) SDS-PAGE analyses of 5SL/105PR/4p3-10 and 5SL/105PR/4p5-10, respectively, before and after incubation at 26 °C for 24 h in the hqWGE.
Fig. 4. Improvement of the translation efficiency in the hqWGE by protecting mRNA with 4p5-10 (and 5SL). (A) Schematic illustration of the two mRNAs used here: 5SL/CrPV-IRES_nLuc/4p5-10 and 5SL/CrPV-IRES_nLuc/ctrl. The sequence of ctrl is 5′ UAGCUAGCUAGAUAUCACUAGUUCUCGAGCUCGGUA 3′. (B) The relative translation efficiency of the two mRNAs at 26 °C for one to several hours in the hqWGE. (C) The residual amount of the two mRNAs that were incubated in the hqWGE under the same conditions as in (B).
4p in RNA 4, which was the most abundant RNA in the selected library. Because the 3′ terminal 20-nt sequence is the primer binding site, we cropped this region by 10 nt from 3′ or 5′ to prepare two shorter protectors (4p3-10 or 4p5-10, respectively; Fig. 3A). As a result, the 24-h protection ability at 26 °C in the hqWGE was lost in the former protector (Fig. 3B), while it was slightly decreased but mostly retained in the latter (Fig. 3C). This means the ten 3′ terminal nucleotides play an important role for protecting the transcripts. In fact, both when 4p5-10 was further shortened from 5′ by 5 nt (4p5-15) and when the five 3′ terminal nucleotides of 4p were deleted (4p3-5), the transcripts were completely degraded in 24 h (Supplementary Fig. S6). These results suggest that the several 3′ terminal nucleotides hybridize to the 5′ terminal 6-nt sequence to form not a simple stem-loop structure but an acceptor stem-like structure that could inhibit 3′ exonucleases, as expected in the design of the initial library (vide supra).12 To verify the versatility of the minimized 36-nt 3′ protector (4p510), we next altered the 105-nt central part (105PR), which had been used as the protected region, to either of two types of twice-longer (∼200-nt) completely different sequences: an internal ribosome entry site (IRES) of the cricket paralysis virus (CrPV)13 or a partial sequence derived from the Nanoluciferase (nLuc)14 gene (CrPV-IRES or nLuc1, respectively; Supplementary Fig. S7A). These transcripts were as stable at 26 °C for 24 h in the hqWGE as the original shorter one (Supplementary Fig. S7B), indicating that 4p5-10 functions independently from the protected region. However, when another ∼200nt sequence derived from the nLuc gene was used as the protected region (nLuc2; Supplementary Fig. S7A), the transcript was more severely degraded, though it could be detected in 24 h (the residual rate was approx. 13%; Supplementary Fig. S7B). These results suggest that an endogenous sequence- or structure-specific endonuclease (such as RNase Z, which was previously found to be highly active in the hqWGE4) first cleaved a certain site of nLuc2, and then exonucleases degraded the cleaved RNAs. Nonetheless, the fact that 13% of transcripts still remained undigested even after 24 h in the hqWGE reflected that they were moderately stabilized. Finally, we added 4p5-10 and 5SL to the 3′ and 5′ terminus, respectively, of an mRNA encoding the full-length nLuc gene to stabilize it and improve its translation efficiency in the hqWGE. Because a 5′ terminal stem-loop structure considerably inhibits the eukaryotic canonical translation,15 the CrPV IRES, which is well protectable with the terminal protectors as described above, was inserted between 5SL and the nLuc gene for IRES-mediated translation (Fig. 4A, above).16 We also
prepared a control mRNA, which has a randomly selected 36-nt sequence (ctrl) instead of 4p5-10 at the 3′ terminus (Fig. 4A, below). These two mRNAs were separately translated at 26 °C for 1, 3, or 5 h in the hqWGE, and then the activities of the expressed nLuc (i.e., the translation efficiencies) were evaluated with the substrate, furimazine (Fig. 4B).14 Comparing the activities between the two mRNAs in the same incubation time, the mRNA with 4p5-10 was 5–6-fold more translated than the control devoid of the 3′ protector in all cases.17 In contrast, when we added the previously reported, tRNA-based 3′ protector (GGCt86) to the 3′ terminus, no enhancement was observed (Supplementary Fig. S8).18 To confirm that 4p5-10 enhanced the translation efficiency by protecting mRNA from endogenous 3′ exonucleases, we then performed RT-qPCR to quantify the amount of residual mRNA transcripts in each period.19 The results showed that the amount of mRNA with 4p5-10 was approximately twice larger than that of the control mRNA in the same incubation time (Fig. 4C), indicating that the higher translation efficiency of mRNA with 4p5-10 can be attributed to its higher stability. The reason why the ratio of mRNA amounts was lower than the ratio of the translation efficiencies is that the dependence of the mRNA concentration on the translation efficiency differed between the mRNA with 4p5-10 and the control: as the concentration decreased, the translation efficiency of the control was suppressed more rapidly (Supplementary Fig. S9). Therefore, it seems probable that the higher stability of the mRNA with 4p5-10 made the main contribution to the higher translation efficiency, though it may also be possible that 4p5-10 directly enhanced the translation by recruiting a translational factor similar to a 3′ cap-independent translational element.5b,20 In summary, we selected several short (and thus easily manageable) 3′ protectors from a random RNA pool in order to inhibit the degradation of in vitro transcripts in the hqWGE. The obtained 3′ protectors stabilized transcripts much more effectively than the previously reported, longer 3′ protectors did. In addition, when one of these 3′ protectors was further minimized (4p to 4p5-10) and then fused to mRNA, the translation efficiency increased 5–6-fold, mainly due to the mRNA stabilization, though the stability of mRNA with 4p5-10 was moderate (compared to those of shorter transcripts with the same 3′ protector). The reason for the moderate stability is probably that 3
Bioorganic & Medicinal Chemistry Letters xxx (xxxx) xxx–xxx
A. Ogawa, et al.
endogenous sequence- or structure-specific endonucleases cleave the mRNA, judging from the fact that the stability is different depending on the protected sequence (Supplementary Fig. S7). This means that an introduction of silent mutations and/or an addition of decoy RNAs could further stabilize the mRNA and thus further improve the translation efficiency. Further studies are now underway along these lines.
(c) Robertson DL, Joyce GF. Nature. 1990;344:467;
7. 8.
Acknowledgments
9.
This work was supported by JSPS KAKENHI Grant Number 16K05846 and 19K05697. 10. 11. 12.
Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.bmcl.2019.06.058.
13. 14. 15. 16.
References 1. 2. 3. 4. 5.
Carlson ED, Gan R, Hodgman CE, Jewett MC. Biotechnol Adv. 2012;30:1185. Madono M, Sawasaki T, Morishita R, Endo Y. New Biotechnol. 2011;28:211. Madin K, Sawasaki T, Ogasawara T, Endo Y. Proc Natl Acad Sci USA. 2000;97:559. Ogawa A, Doi Y. Org Biomol Chem. 2015;13:1008. (a) Sawasaki T, Ogasawara T, Morishita R, Endo Y. PNAS. 2002;99:14652;
17. 18.
(b) Ogawa A, Tabuchi J, Doi Y. Bioorg Med Chem Lett. 2014;24:3724. 6. (a) Tuerk C, Gold L. Science. 1990;249:505;
19. 20.
(b) Ellington AD, Szostak JW. Nature. 1990;346:818;
4
(d) Sando S, Ogawa A, Nishi T, Hayami M, Aoyama Y. Bioorg Med Chem Lett. 2007;17:1216. The reason we chose a sequence of such short length to be protected is that it is easy to detect ∼170-nt RNA on a polyacrylamide gel in the presence of various endogenous biomolecules in the hqWGE. Wilusz JE, JnBaptiste CK, Lu LY, Kuhn CD, Joshua-Tor L, Sharp PA. Genes Dev. 2012;26:2392. In some sequences (more than 1500), the random region could not be identified because the sequences did not include the protected region (105PR). These sequences were probably derived from the endogenous wheat genome, given the fact that 200∼500-bp DNAs were amplified with RT-PCR of the hqWGE without any exogenous RNA. Fay MM, Lyons SM, Ivanov P. J Mol Biol. 2017;429:2127. Xu S, Li Q, Xiang J, et al. Sci Rep. 2016;6:24793. Although it is expected that 4p3-10 could also be folded into a stem-loop structure at the 3’ terminus, a simple stem-loop was previously found to be inefficient as a 3’ protector in the hqWGE.4. Pfingsten JS, Kieft JS. RNA. 2008;14:1255. Hall MP, et al. ACS Chem Biol. 1848;2012:7. Ogawa A. ChemBioChem. 2009;10:2465. The CrPV IRES-mediated translation is as efficient as the canonical translation in the hqWGE (unpublished results). mRNA with 4p instead of 4p-5 also showed high translation efficiency (93% of that by mRNA with 4p-5 in 5-h incubation). The tRNA-based 3’ protector adversely decreased the translation efficiency, probably because the tRNA frame trapped some protein necessary for translation (e.g., eEF1α). This result is different from our previous report,4 where an appropriate 5’ protector was not used. It was very difficult to analyze mRNAs with PAGE due to their long length. Ogawa A, Murashige Y, Tabuchi J, Omatsu T. Mol BioSyst. 2017;13:314.
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters journal homepage: www.elsevier.com/locate/bmcl
In vitro selection of a 3′ terminal short protector that stabilizes transcripts to improve the translation efficiency in a wheat germ extract Atsushi Ogawa , Akane Kutsuna, Masashi Takamatsu, Tatsuya Okuzono ⁎
Proteo-Science Center, Ehime University, 3 Bunkyo-cho, Matsuyama, Ehime 790-8577, Japan
ARTICLE INFO
ABSTRACT
Keywords: Cell-free translation Wheat germ extract In vitro selection RNA stabilization Translational enhancer
Wheat cell-free expression systems based on wheat germ extract (WGE) enable us to briefly synthesize various types of proteins in vitro merely by exogenously adding their mRNA templates. Moreover, it is possible to produce larger amounts of protein by thoroughly removing the endosperm, which contains many translation inhibitors, including ribonucleases (RNases). However, because small amounts of RNases are also present even in an endosperm-free, high-quality WGE (hqWGE), the in-vitro transcribed mRNA is rapidly degraded. In particular, 3′ exonucleases have been considered as the major RNases that degrade mRNA. We thus herein performed in vitro selection to find an effective, short 3′ protector sequence from a random RNA pool. The selected sequences stabilized in vitro transcripts in the hqWGE more effectively than the previously reported, longer 3′ protectors did. In addition, when one of these 3′ protectors was minimized and then fused to mRNA, the translation efficiency increased 5–6-fold in the hqWGE, mainly due to the mRNA stabilization.
Cell-free expression systems, which are generally composed of a cell extract, allow us to synthesize a protein in vitro merely by exogenous addition of the mRNA template (in vitro transcripts) thereinto.1 Among various types of cell-free systems, wheat germ extract (WGE) has a major advantage for the expression of a broad range of proteins derived from various types of organisms (not only eukaryotes but also viruses and prokaryotes), by virtue of its low codon preference.2 Although WGEs prepared by the traditional method show low productivity due to the presence of many contaminating translation inhibitors, such as ribosome-inactivating proteins, ribonucleases (RNases) and proteases, these contaminants can be considerably eliminated by extensively washing the endosperm from the wheat embryos to prepare a highquality WGE (hqWGE), permitting the production of much larger amounts of protein (several mg/mL).3 Nonetheless, even in an hqWGE, in-vitro transcribed RNA is still unstable and is generally digested within one hour.4 This means that there are non-negligible amounts of RNases derived from embryos. In particular, 3′ exonucleases have been considered as the major nucleases that degrade RNA in the hqWGE.5 In fact, we recently successfully stabilized in vitro transcripts (a 77-nt aptamer) for at least one hour in an hqWGE by ligating a tRNA-based rigid structure named GGCt86 (and a simple stem-loop structure named 5SL) to the 3′ terminus (and the 5′ terminus, respectively) to inhibit degradation by endogenous 3′ exonucleases (and 5′ exonucleases, respectively).4 However, we found that GGCt86 was unable to protect the
⁎
transcripts under longer incubation periods. In addition, GGCt86 is cumbersome due to its moderately long length (94 nt), and is inappropriate for protecting mRNA because of the complete tRNA frame, which could trap eEF-1α and/or the corresponding aminoacyl-tRNA synthetase. Therefore, we herein attempted to obtain a more effective, shorter 3′ protector through in vitro selection6 in order to stabilize transcripts including mRNA templates and ultimately to further improve the translation efficiency in the hqWGE. We first designed a 3′ protector library based on GGCt86 for in vitro selection as follows (Fig. 1A). Because a specific primer binding site is required at the 3′ terminus for amplification during selection, the 3′ terminal 20-nt sequence of GGCt86 was used as is. The 5′ terminal 6-nt sequence was also left under the expectation that it might form a stem with the 3′ terminal region to inhibit 3′ exonucleases. 20 random nucleotides were inserted between them. DNA sequences encoding this half-size 3′ protector library (46 nt) and 5SL (21 nt) were added (with the T7 promoter for the latter) to each side of a 105-bp sequence encoding a protected region (105PR),7 which was unintentionally chosen from a pMK-based plasmid, to prepare a random DNA pool including theoretically all possible sequences (approximately 1012 different DNA molecules; Fig. 1B, top). After the DNA pool was carefully transcribed so as not to decrease the diversity (Fig. 1B, (i)), 50 pmol (3 × 1013 molecules) of the transcripts (5SL/105PR/3′ protector library) were incubated at 26 °C in the hqWGE with several materials optimized for
Corresponding author. E-mail address: [email protected] (A. Ogawa).
https://doi.org/10.1016/j.bmcl.2019.06.058 Received 7 May 2019; Received in revised form 20 June 2019; Accepted 28 June 2019 0960-894X/ © 2019 Elsevier Ltd. All rights reserved.
Please cite this article as: Atsushi Ogawa, et al., Bioorganic & Medicinal Chemistry Letters, https://doi.org/10.1016/j.bmcl.2019.06.058
Bioorganic & Medicinal Chemistry Letters xxx (xxxx) xxx–xxx
A. Ogawa, et al.
Fig. 2. The stability of each of the selected RNAs. (A) The sequences of the random region in the top 8 most-abundant RNAs (4–11: 5SL/105PR/Xp (X = 4–11), wherein Xp is the 3′ protector). The read number in the sequencing and the residual rate after incubation at 26 °C for 24 h in the hqWGE are also shown in the rightmost two columns. (B) SDS-PAGE analysis of the three moststable RNAs (4, 8 and 9) before and after incubation at 26 °C for 24 h in the hqWGE. See Supplementary Fig. S4 for the other five RNAs.
the hqWGE, despite their shorter length. We next read the sequences of the selected RNAs using a nextgeneration sequencer without utilizing genetic engineering. 4,245 of the 32,270 read sequences were randomly chosen to cluster into several groups with the same sequence in the random region.9 Fig. 2A shows the sequences of the random region in the top 8 most-abundant RNAs (4–11: 5SL/105PR/Xp (X = 4–11), wherein Xp is the 3′ protector). To confirm the stability of these 8 RNAs individually, each RNA was incubated at 26 °C in the hqWGE. As shown in Fig. 2B and Supplementary Fig. S4, all transcripts investigated were as stable for 24 h as the selected library (their residual rates were 50–90%; Fig. 2A), meaning that we successfully obtained several effective 3′ protectors through in vitro selection. Because the 3′ protectors in the top 8 RNAs are G-rich without exception, we expected that they are folded into a G-quadruplex.10 We thus picked up two relatively stable RNAs (4 and 8) and checked whether they could form a G-quadruplex structure with a simple assay using Thioflavin T (ThT)11 in the absence of the hqWGE, which includes many biomolecules that could affect the assay. However, both RNAs increased the fluorescence of ThT only moderately, in the same way as a negative control RNA devoid of any 3′ protector, in contrast to a positive control RNA with the 3′ terminal TRF2 G-quadruplex (TRF2),11 which achieved a two-fold enhancement of the fluorescence (Supplementary Fig. S5A). These results indicate that 4 and 8 did not form a G-quadruplex, at least in the 3′ protector region. Although we cannot deny the possibility that the selected protectors were folded into a G-quadruplex in the hqWGE by interacting with endogenous molecules, the G-quadruplex structure at the 3′ terminus was not in itself sufficient for protecting transcripts: the positive control RNA with TRF2 was much more unstable in the hqWGE (Supplementary Fig. S5B). Therefore, it is assumed that the selected 3′ protectors form another rigid structure to prevent endogenous 3′ exonucleases in the hqWGE. To obtain a shorter 3′ protector, we next minimized a 3′ protector
Fig. 1. In vitro selection of 3′ terminal short protectors that stabilize transcripts in the hqWGE. (A) Design of a 3′ protector library (right) based on a tRNAbased 3′ protector, GGCt86 (left). The blue line with N20 represents the 20-nt random region. (B) Schematic illustration of the in vitro selection. In vitro transcripts in the random RNA pool are composed of a 5′ protector (5SL), the protected region (105PR), and the 3′ protector library including N20. (C) SDSPAGE analysis of the initial RNA pool (middle) and the resulting pool after the 11th round of selection (right) before and after incubation at 26 °C for 24 h in the hqWGE. As a control, an exogenous RNA-free solution was also resolved on the same gel (left). Some bands on the control lane that were also detected on other lanes are endogenous components in the hqWGE.4 The reason the band patterns differed between the initial pool and the selected pool is probably that the transcripts in the latter formed specific tertiary structures.
cell-free translation (Fig.1B, (ii)). The undigested RNA was then recovered and amplified by RT-PCR (Fig. 1B, (iii)). We repeated this transcription/incubation/amplification cycle 11 times with progressively increasing stringency (especially with respect to the incubation time; Supplementary Fig. S1). The residual amount of RNA in each round was evaluated with SDS-PAGE, which revealed that the stability of transcripts gradually increased as the selection proceeded (until the 10th round; Supplementary Fig. S1). The resulting RNA pool after the 11th round was stable at 26 °C for 24 h, whereas almost all of the initial pool was degraded within 1 h (Fig. 1C and Supplementary Fig. S2). To compare the protecting ability of the selected 3′ library with those of other 3′ protectors reported previously, we changed the 3′ library region (46 nt) to GGCt86 (94 nt)4 or a 59-nt triple helix (TH).8 Although both RNAs with these 3′ protectors were stable at 26 °C for at least 1 h as previously reported, they were completely digested in 24 h (Supplementary Fig. S3). These results clearly show that the selected protectors are much more effective for stabilizing in vitro transcripts in 2
Bioorganic & Medicinal Chemistry Letters xxx (xxxx) xxx–xxx
A. Ogawa, et al.
Fig. 3. Minimization of 4p in RNA 4. (A) The proposed secondary structures of 4p and the 10 nt-cropped ones (4p3-10 and 4p5-10). (B and C) SDS-PAGE analyses of 5SL/105PR/4p3-10 and 5SL/105PR/4p5-10, respectively, before and after incubation at 26 °C for 24 h in the hqWGE.
Fig. 4. Improvement of the translation efficiency in the hqWGE by protecting mRNA with 4p5-10 (and 5SL). (A) Schematic illustration of the two mRNAs used here: 5SL/CrPV-IRES_nLuc/4p5-10 and 5SL/CrPV-IRES_nLuc/ctrl. The sequence of ctrl is 5′ UAGCUAGCUAGAUAUCACUAGUUCUCGAGCUCGGUA 3′. (B) The relative translation efficiency of the two mRNAs at 26 °C for one to several hours in the hqWGE. (C) The residual amount of the two mRNAs that were incubated in the hqWGE under the same conditions as in (B).
4p in RNA 4, which was the most abundant RNA in the selected library. Because the 3′ terminal 20-nt sequence is the primer binding site, we cropped this region by 10 nt from 3′ or 5′ to prepare two shorter protectors (4p3-10 or 4p5-10, respectively; Fig. 3A). As a result, the 24-h protection ability at 26 °C in the hqWGE was lost in the former protector (Fig. 3B), while it was slightly decreased but mostly retained in the latter (Fig. 3C). This means the ten 3′ terminal nucleotides play an important role for protecting the transcripts. In fact, both when 4p5-10 was further shortened from 5′ by 5 nt (4p5-15) and when the five 3′ terminal nucleotides of 4p were deleted (4p3-5), the transcripts were completely degraded in 24 h (Supplementary Fig. S6). These results suggest that the several 3′ terminal nucleotides hybridize to the 5′ terminal 6-nt sequence to form not a simple stem-loop structure but an acceptor stem-like structure that could inhibit 3′ exonucleases, as expected in the design of the initial library (vide supra).12 To verify the versatility of the minimized 36-nt 3′ protector (4p510), we next altered the 105-nt central part (105PR), which had been used as the protected region, to either of two types of twice-longer (∼200-nt) completely different sequences: an internal ribosome entry site (IRES) of the cricket paralysis virus (CrPV)13 or a partial sequence derived from the Nanoluciferase (nLuc)14 gene (CrPV-IRES or nLuc1, respectively; Supplementary Fig. S7A). These transcripts were as stable at 26 °C for 24 h in the hqWGE as the original shorter one (Supplementary Fig. S7B), indicating that 4p5-10 functions independently from the protected region. However, when another ∼200nt sequence derived from the nLuc gene was used as the protected region (nLuc2; Supplementary Fig. S7A), the transcript was more severely degraded, though it could be detected in 24 h (the residual rate was approx. 13%; Supplementary Fig. S7B). These results suggest that an endogenous sequence- or structure-specific endonuclease (such as RNase Z, which was previously found to be highly active in the hqWGE4) first cleaved a certain site of nLuc2, and then exonucleases degraded the cleaved RNAs. Nonetheless, the fact that 13% of transcripts still remained undigested even after 24 h in the hqWGE reflected that they were moderately stabilized. Finally, we added 4p5-10 and 5SL to the 3′ and 5′ terminus, respectively, of an mRNA encoding the full-length nLuc gene to stabilize it and improve its translation efficiency in the hqWGE. Because a 5′ terminal stem-loop structure considerably inhibits the eukaryotic canonical translation,15 the CrPV IRES, which is well protectable with the terminal protectors as described above, was inserted between 5SL and the nLuc gene for IRES-mediated translation (Fig. 4A, above).16 We also
prepared a control mRNA, which has a randomly selected 36-nt sequence (ctrl) instead of 4p5-10 at the 3′ terminus (Fig. 4A, below). These two mRNAs were separately translated at 26 °C for 1, 3, or 5 h in the hqWGE, and then the activities of the expressed nLuc (i.e., the translation efficiencies) were evaluated with the substrate, furimazine (Fig. 4B).14 Comparing the activities between the two mRNAs in the same incubation time, the mRNA with 4p5-10 was 5–6-fold more translated than the control devoid of the 3′ protector in all cases.17 In contrast, when we added the previously reported, tRNA-based 3′ protector (GGCt86) to the 3′ terminus, no enhancement was observed (Supplementary Fig. S8).18 To confirm that 4p5-10 enhanced the translation efficiency by protecting mRNA from endogenous 3′ exonucleases, we then performed RT-qPCR to quantify the amount of residual mRNA transcripts in each period.19 The results showed that the amount of mRNA with 4p5-10 was approximately twice larger than that of the control mRNA in the same incubation time (Fig. 4C), indicating that the higher translation efficiency of mRNA with 4p5-10 can be attributed to its higher stability. The reason why the ratio of mRNA amounts was lower than the ratio of the translation efficiencies is that the dependence of the mRNA concentration on the translation efficiency differed between the mRNA with 4p5-10 and the control: as the concentration decreased, the translation efficiency of the control was suppressed more rapidly (Supplementary Fig. S9). Therefore, it seems probable that the higher stability of the mRNA with 4p5-10 made the main contribution to the higher translation efficiency, though it may also be possible that 4p5-10 directly enhanced the translation by recruiting a translational factor similar to a 3′ cap-independent translational element.5b,20 In summary, we selected several short (and thus easily manageable) 3′ protectors from a random RNA pool in order to inhibit the degradation of in vitro transcripts in the hqWGE. The obtained 3′ protectors stabilized transcripts much more effectively than the previously reported, longer 3′ protectors did. In addition, when one of these 3′ protectors was further minimized (4p to 4p5-10) and then fused to mRNA, the translation efficiency increased 5–6-fold, mainly due to the mRNA stabilization, though the stability of mRNA with 4p5-10 was moderate (compared to those of shorter transcripts with the same 3′ protector). The reason for the moderate stability is probably that 3
Bioorganic & Medicinal Chemistry Letters xxx (xxxx) xxx–xxx
A. Ogawa, et al.
endogenous sequence- or structure-specific endonucleases cleave the mRNA, judging from the fact that the stability is different depending on the protected sequence (Supplementary Fig. S7). This means that an introduction of silent mutations and/or an addition of decoy RNAs could further stabilize the mRNA and thus further improve the translation efficiency. Further studies are now underway along these lines.
(c) Robertson DL, Joyce GF. Nature. 1990;344:467;
7. 8.
Acknowledgments
9.
This work was supported by JSPS KAKENHI Grant Number 16K05846 and 19K05697. 10. 11. 12.
Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.bmcl.2019.06.058.
13. 14. 15. 16.
References 1. 2. 3. 4. 5.
Carlson ED, Gan R, Hodgman CE, Jewett MC. Biotechnol Adv. 2012;30:1185. Madono M, Sawasaki T, Morishita R, Endo Y. New Biotechnol. 2011;28:211. Madin K, Sawasaki T, Ogasawara T, Endo Y. Proc Natl Acad Sci USA. 2000;97:559. Ogawa A, Doi Y. Org Biomol Chem. 2015;13:1008. (a) Sawasaki T, Ogasawara T, Morishita R, Endo Y. PNAS. 2002;99:14652;
17. 18.
(b) Ogawa A, Tabuchi J, Doi Y. Bioorg Med Chem Lett. 2014;24:3724. 6. (a) Tuerk C, Gold L. Science. 1990;249:505;
19. 20.
(b) Ellington AD, Szostak JW. Nature. 1990;346:818;
4
(d) Sando S, Ogawa A, Nishi T, Hayami M, Aoyama Y. Bioorg Med Chem Lett. 2007;17:1216. The reason we chose a sequence of such short length to be protected is that it is easy to detect ∼170-nt RNA on a polyacrylamide gel in the presence of various endogenous biomolecules in the hqWGE. Wilusz JE, JnBaptiste CK, Lu LY, Kuhn CD, Joshua-Tor L, Sharp PA. Genes Dev. 2012;26:2392. In some sequences (more than 1500), the random region could not be identified because the sequences did not include the protected region (105PR). These sequences were probably derived from the endogenous wheat genome, given the fact that 200∼500-bp DNAs were amplified with RT-PCR of the hqWGE without any exogenous RNA. Fay MM, Lyons SM, Ivanov P. J Mol Biol. 2017;429:2127. Xu S, Li Q, Xiang J, et al. Sci Rep. 2016;6:24793. Although it is expected that 4p3-10 could also be folded into a stem-loop structure at the 3’ terminus, a simple stem-loop was previously found to be inefficient as a 3’ protector in the hqWGE.4. Pfingsten JS, Kieft JS. RNA. 2008;14:1255. Hall MP, et al. ACS Chem Biol. 1848;2012:7. Ogawa A. ChemBioChem. 2009;10:2465. The CrPV IRES-mediated translation is as efficient as the canonical translation in the hqWGE (unpublished results). mRNA with 4p instead of 4p-5 also showed high translation efficiency (93% of that by mRNA with 4p-5 in 5-h incubation). The tRNA-based 3’ protector adversely decreased the translation efficiency, probably because the tRNA frame trapped some protein necessary for translation (e.g., eEF1α). This result is different from our previous report,4 where an appropriate 5’ protector was not used. It was very difficult to analyze mRNAs with PAGE due to their long length. Ogawa A, Murashige Y, Tabuchi J, Omatsu T. Mol BioSyst. 2017;13:314.