|Home | About | Journals | Submit | Contact Us | Français|
Caged antisense oligodeoxynucleotides (asODNs) are synthesized by linking two ends of linear oligodeoxynucleotides using a photocleavable linker. Two of them (H30 and H40) have hairpin-like structures which show a large difference in thermal stability (ΔTm = 17.5°C and 11.6°C) comparing to uncaged ones. The other three (C20, C30 and C40) without stable secondary structures have the middle 20 deoxynucleotides complementary to 40-mer RNA. All caged asODNs have restricted opening which provides control over RNA/asODN interaction. RNase H assay results showed that 40-mer RNA digestion could be photo-modulated 2- to 3-fold upon light-activation with H30, H40, C30 and C40, while with C20, RNA digestion was almost not detectable; however, photo-activation triggered >20-fold increase of RNA digestion. And gel shift assays showed that it needed >0.04 μM H40 and 0.5 μM H30 to completely bind 0.02 μM 40-mer RNA, and for C40 and C30, it needed >0.2 μM and 0.5 μM for 0.02 μM 40-mer RNA binding. However, even 4 μM C20 was not able to fully bind the same concentration of 40-mer RNA. By simple adjustment of ring size of caged asODNs, we could successfully photoregulate their hybridization with mRNA and target RNA hydrolysis by RNase H with light activation.
Manual regulation of gene expression is one of the most important scientific areas. Several technologies using oligonucleotides, such as antisense, antigene and RNA interference, have been developed. However, it is still very challenging to switch gene expression ‘on’ and/or ‘off’ and provide high spatial and temporal resolution. (1–15) Photolabile groups can be used to temporarily ‘cage’ oligonucleotides. Their activities can be restored by light irradiation which provides new possibilities for controlling gene expression in both space and time (2,9,16–28). Currently, there is still a strong need for more effective gene regulation tools.
For the regulation of gene expression, antisense strategies based on the hybridization of target mRNA with complementary DNA (antisense DNA) have been widely applied to gene silencing in many experimental systems and are being evaluated as treatments for cancers and other diseases in human clinical trials (29,30). Different strategies for photoregulation of gene expression have been developed. One strategy is based on application of multiple photoresponsive groups to modulate the interaction of antisense oligonucleotides (asODNs) and target RNAs. Monroe et al. (31) reported they could photomodulate DNA hybridization using a molecular beacon assay and determine the relative caging and uncaging percentages for 20-mer oligodeoxynucleotide with an average 14–16 nitrophenyl caging groups. While Komiyama et al. (32) incorporated azobenzene groups into the sense strand and controlled the hybridization of sense strand and antisense strand to regulate target RNA digestion through azobenzene photoisomerization. However, multiple azobenzene groups are a must for reasonable photomodulation. Recently, Deiters et al. (33) applied a new photocleavable nitrobenzyl derivative for caging thymidine at N′ position to disturb the A–T hydrogen binding. They could photoregulate luciferase expression in cells with 3 or 4 caging thymidines for 18-mer asODNs. For multiple photolabile groups, complete uncaging requires much high intensity UV light and long-time irradiation. Another strategy is focused on single photolabile linker for photomodulation. There were some reports about the mechanism of light-triggered strand break of oligonucleotides (34–36). For gene-related studies of interaction of asODN with DNA or RNA, pioneering work done by Taylor et al. (37) demonstrated the incorporation of a photoactive o-nitrobenzyl moiety bridging within the phosphate backbone of a DNA hairpin and the application to trigger DNA/DNA duplex formation. Near UV irradiation under ambient conditions triggered a strand break which released 18-mer oligodeoxynucleotide to bind a complementary DNA strand with a 9-fold greater affinity. More recent examples were presented by Dmochowski et al. (38–40) using a short complementary sense strand as the blocking moiety for an asODN with a heterobifunctional photocleavable linker. In these examples, multiple basepairs worked as multiple caging groups and the relative thermostability of asODN-PL-sODN and asODN/sODN duplex was used to control the asODN/RNA hybridization and RNA digestion in RNase H assay, and in cells and zebrafish. Based on this simple design, Friedman (41) and Chen (42,43) further respectively extended their applications in RNA interference in cells and morpholino oligonucleotides for targeting gene expression in zebrafish. However, this design was dependent on the relative thermostability of asODN-PL-sODN, asODN/sODN and asODN/RNA, which would be greatly effected in complicated cellar environments. Sometimes the best candidate in in vitro selection was not the best one in cells (40). More general and effective light-activated asODNs are still needed.
Based on many biological processes that involve the hybridization of DNA/DNA, DNA/RNA or RNA/RNA, we have sought to develop chemically synthetically facile and high-quantum efficient routes for photomodulating the hybridization of asODNs to target DNA or mRNA molecules (38). Most recently Dmochowski et al. (44) reported a photocleavable circular DNAzyme through enzymatic coupling of two ends of oligodeoxynucleotide for controlling RNA digestion. Here, we developed a general strategy of linking two ends of a linear asODNs using a photocleavable linker. These caged circular asODNs were restricted from extending to bind their targets and their activities were masked until light cleaved the linker and released two ends of asODNs. Previously, similar strategies used in photoregulation of asODN/RNA were based on relative thermostability of caged asODN, asODN/sODN and asODN/RNA. In this study, we aim to photomodulate asODN/RNA hybridization sterically and thermodynamically, as shown in Scheme 1.
Here, we reported a new design of caged asODNs. End-linked circular photolabile asODNs with 1-(2-nitrophenyl)-1,2-ethanediol (PL) photolabile moiety incorporated in the asODN sequences were designed to regulate hybridization of asDNAs and target RNA and/or photomodulate RNA digestion by RNase H. A linear asODN is usually flexible and has random structure, however when it hybridizes with a complementary RNA, the duplex will become much rigid and extended. Within a limited length of deoxynucleotides in an oligodeoxynucleotide ring, the formation of duplex will be efficiently inhibited. Five different length end-linked caged asODNs are synthesized and used in this work. Two of the circular caged asODNs (H30 and H40) have the hairpin-like structure. One is a 30-mer end-linked caged asODN (H30) with a 5-deoxynucleotide-paired stem and a 20-deoxynucleotide hairpin loop, the other one (H40) is a 40-mer end-linked asODN hairpin (H40) with another five non-complementary deoxynucleotides at both ends of H30. The other three caged asODNs (C20, C30 and C40) without stable secondary structures have different numbers of deoxynucleotides in the ring, but share the same 20-deoxynucleotide sequence in the middle. Caged asODNs, C20, H30 and H40, have 20, 30 and 40 paired bases with the 40-mer RNA target respectively, while C30 and C40 have the same middle 20 deoxynucleotides as C20 which are base-paired with the middle 20 nucleotides of the 40-mer RNA (Scheme 2). All these end-linked caged circular asODNs have restricted opening of oligodeoxynucleotide rings, which provides control over RNA/asODN duplex formation and targets RNA digestion by RNase H.
All single-stranded oligodeoxynucleotides were custom-synthesized with C7 amino modified CPG and purified with an Agilent 1200 HPLC system using a reverse-phase analytical HPLC column (reverse phase C18, 4.6 × 250 mm, 5 μm beads). The concentrations of all oligodeoxynucleotides were measured in water at 260 nm using a Varian Cary 300 UV/Vis spectrophotometer. Melting studies were carried out with a Varian Cary 300 UV/Vis spectrophotometer. Gels of RNA digestion were imaged using an Amersham Biosciences Storm 840 phosphorimager and quantified with ImageQuant software (ImageQuantTM TL v2005, GE Healthcare). All photoirradiation experiments with oligodeoxynucleotide samples were carried out with a xenon lamp (450 W) and monochromator (20 nm slit centered at 350 nm, 30 mW/cm2 at the sample). Standard 1× RNase H reaction buffer was defined as 20 mM Tris–HCl, 20 mM KCl, 10 mM MgCl2, 0.1 mM EDTA, and 0.1 mM DTT, pH 8.0. The melting temperature measurements used the same RNase H buffer.
CPG with 3′-end amino modification was used for oligodeoxynucleotide synthesis. All oligodeoxynucleotide sequences for H30, H40, C20, C30 and C40 were custom-synthesized according to the standard DNA synthesis. At the 5′-end of the sequences was attached with 1-(2-nitrophenyl)-1,2-ethanediol (PL) phosphoroamidite (see Supplementary Material for synthesis), following by the amino group (MMT protected) with C6 attached to sequences through phosphorodiester bonds. MMT was then removed by 4%TFA in CH2Cl2 to release 5′ amino group. Before cleavage from the resin, the amino group first reacted with about 100 fold excess succinic anhydride and N,N-diisopropylethylamine (DIPEA) in 0.5 ml DMF at room temperature overnight.
Sequences used in the study:
The oligodeoxynucleotides were then cleaved from resin and deprotected using concentrated ammonium hydroxide. After removal of ammonia, the oligodeoxynucleotides were purified with HPLC under reverse phase conditions: A, 0.05 M TEAA; B, acetonitrile; B, 0–15% in 30 min, 15–45% in 30 min. The product peak was collected and characterized by ESI-MS using negative mode (1% TEA in H2O/CH3CN). C20-COOH, calculated: 6801.0; measured: 6800.0; H30-COOH, calculated: 9890.3, measured: 9890.45; H40-COOH, calculated: 12981.3, measured: 12981.15; C30-COOH, calculated: 9875.0; measured: 9875.0; C40H-COOH, calculated: 12914.9; measured: 12915.6.
The oligodeoxynucleotides with amine and acid groups at two terminal ends were dried and ethanol-precipitated with 3 M NaCl to remove the TEAA residue. Twenty nanomoles precipitated oligodeoxynucleotides were dissolved in 2 ml 0.1 M MES buffer (pH = 6.5), 0.3 M NaCl and 10 mM MgCl2 with final concentration of 5 mM HOBt. Then ~1 mg 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide, hydrochloride (EDAC) was added into the oligodeoxynucleotide solutions. The mixture was vortexed and stood at room temperature for 10–14 h.
The solutions were then desalted using NAP-10 columns, following by RP HPLC purification (A, 0.05 M TEAA; B, acetonitrile; B, 0–15% in 30 min, 15–45% in 30 min, running temperature, 40°C). The retention time of product peaks were usually 1–2 min longer than those of starting oligodeoxynucleotides. The collected products were subject to dry in vacuum and characterized by ESI-MS. The yields were calculated by dissolved the caged circular oligodeoxynucleotides in water and measured the absorbance at 260 nm. The synthetic yields for caged circular oligodeoxynucleotides were 20–40% under our synthetic conditions. ESI-MS was carried out under negative mode (1% TEA in H2O/CH3CN). C20, calculated: 6782.0; measured: 6782.0; H30, calculated: 9872.3, measured: 9871.55 + n Na+; H40, calculated: 12 963.3, measured: 12 961.25 + n Na+, measured: C30, calculated: 9857; measured: 9854.8; C40, calculated: 12897.6; measured: 12 895.7.
Thermal denaturation studies were performed on H30, H40, their uncaged H30, H40, uncaged H30, H40 with target RNA in standard RNase H buffer. The concentration of each asODN was determined by dissolving it in pure water and measuring the absorbance at 260 nm. The solution was heated to 90°C for 5 min, and allowed to cool gradually to 20°C. Samples were monitored at 260 nm while heating or cooling at a rate of 0.5°C/min. Melting temperatures were determined from the peak of the first derivative plot of Abs260 versus temperature.
The 40-mer RNA target sequence (5′-CUUGUACAGAAAUACGGUCCGAAACGUUGGUCUGUUAUUG-3′) in HPLC-pure form, was purchased from Shanghai GenePharma Co, Ltd. Recombinant RNase H from E. coli and the reaction buffer were purchased from Epicentre Biotechnologies. The standard procedure for the RNase H assay was as follows: a caged circular asODN was incubated in 1× ribonuclease H reaction buffer at 37°C, [γ-32P]-labeled RNA oligonucleotide (200-fold excess) was added and incubated at 37°C for 20 min to allow RNA/DNA duplex formation. RNase H (2 U) was then added to the mixture and incubated at 37°C. Total reaction volume was 20 µl, and the final concentrations of asODNs and RNA were 0.02 µM and 4 µM, respectively.
To measure RNA degradation by RNase H after photoactivation, the DNA conjugates were UV-illuminated by Xe lamp through monochromator (350 nm UV light generated by 450 W Xenon lamp after passing through monochromator, ~30 mW/cm2), [γ-32P]-labeled RNA oligonucleotide was added and incubated at 37°C for 20 min to allow RNA/DNA duplex formation. RNase H assays were performed as described above. Time points were taken at 2, 5, 10, 15 and 30 min by sampling 4 µl of the reaction mixture, adding 6 µl gel loading buffer (50 mM EDTA, 90% formamide with bromphenol blue and xylene cyanol, total volume = 10 µl), and then heating the solutions to 95°C for 3 min to terminate the reaction.
All of the resulting solutions were subject to electrophoresis on a 20% polyacrylamide gel containing 7 M urea. Intensity values of gel bands were integrated in ImageQuant for each band with automated lane and band finding using a local method background correction in the gel lane. The relative amount of RNA digestion was determined by dividing the intensity of the band corresponding to cleaved RNA by the total intensity of the cleaved and uncleaved RNA bands.
To determine the binding of RNA to the caged circular asDNAs, gel mobility shift assays were performed as described in the literature (45). [γ-32P]-labeled 40-mer RNA was used in this study, and the binding of a fixed concentration of the 40-mer RNA (0.02 μM) with increased concentrations of caged circular asODNs was carried out in 9 μl RNase H buffer containing 20 mM Tris–HCl, 20 mM KCl, 10 mM MgCl2, 0.1 mM EDTA, 0.1 mM DTT, pH 8.0. Samples were annealed at 37°C for 30 min and then were put into ice water. Before the samples were loaded into the gel, 1 μl glycerol with bromphenol blue and xylene cyanol was added. All 10 μl solutions were loaded into 12% native polyacrylamide gels. The gels were then electrophoresed at 100 V for 2 h at 10°C, using 1 × TBE buffer (pH = 8.2). Gels were exposed and then imaged with a Storm phosphorimager.
The hybridization between an asODN and a sense RNA is one of the key factors in RNA digestion by RNase H. Previous works on related research with a single photolabile group in asODNs are based on the relative thermostability of three components in the system (caged asODN, asODN/sODN and asODN/RNA); however, it takes quite efforts to adjust the system to find the right pair. Steric inhibition of asODN/RNA is another way to achieve the same results. However, a single caging group attached to a 20-mer asODN is not enough to compensate the large thermodynamic driving force and have proven relatively ineffective (31,46,47).
Based on the natural properties of the oligodeoxynucleotide, a linear asODN is usually flexible and has random conformations. However, once it hybridizes with a complementary DNA or RNA strand, the duplex will become much rigid and extended. Within a limited length of deoxynucleotides in an oligodeoxynucleotide ring, formation of the duplex will be efficiently inhibited, and once the caged circular asODN is released by the light, it will quickly interact with its target RNA. Two different length end-linked caged asODNs (H30 and H40) are first synthesized. One is a 30-mer end-linked asODN (H30) with a 5 nucleobase-paired stem and a hairpin loop with 20 deoxynucleotides, the other one is a 40-mer end-linked caged asODN (H40) with another 5 non-complementary deoxynucleotides at both ends of H30. A 40-mer RNA that is complementary to H30 and H40 is used to hybridize with the asODNs and further evaluate the photomodulation of its digestion by RNase H. Another three end-linked caged circular asODNs (C20, C30 and C40) do not have stable secondary structures, and the middle 20 nt of C30 and C40 share the same sequence with C20 which is complementary to the middle 20 nucleotide sequence of the target 40-mer RNA. All these three caged asODNs are used to study their interaction with the target 40-mer RNA.
We measured the melting temperatures of uncaged asODNs (H30 and H40), caged asODNs (H30 and H40) and duplexes of uncaged asODNs (H30 and H40) with the 40-mer target RNA. The melting temperatures of uncaged H30 and H40 were both ~ 42°C under the same buffer conditions as RNase H assay, because these two oligodeoxynucleotides had exactly the same loop size and stem bases except that H40 had 5 extra non-complementary deoxynucleotides on each end. However for the end-linked caged H30 and H40, the melting temperatures were 59.5°C and 53.6°C, respectively, which were 17.5°C and 11.6°C higher than uncaged H30 and H40. And the caged H30 was more stable than the caged H40 due to the extra 10-base loop for H40 which destabilized the hairpin. We also measured the Tm of duplexes of uncaged H30 and H40 with the 40-mer target RNA, which were 77.3°C and 79.3°C, respectively.
RNA cannot be digested by RNase H without the existence of complementary antisense strand. Under current RNase H assay conditions with 4 µM [γ-32P]-labeled target 40-mer RNA, 0.02 µM asODN, and 2 U RNase H in standard RNase H buffer at 37°C, there was almost no RNA digestion detectable with caged asODN C20 up to 30 min incubation time before UV irradiation. However, UV irradiation (350 nm UV light generated by 450 W Xenon lamp after passing through monochromator, ~30 mW/cm2) triggered the chemical bond cleavage and promoted interaction between uncaged C20 and target RNA. Figure 1 showed a typical gel that the cleaved RNA at the lower band gradually increased, and RNA digestion reached up to 27.5% in 30 min, which was at least >20-fold increase in comparison to C20 in the dark. While with the fully complementary caged circular H30 and H40, about 79% target RNA was digested in 30 min after UV irradiation, which was about 3- and 2-fold increase of 40-mer target RNA digestion compared to H30 and H40 before light-activation, respectively. The results showed that RNA digestion did not increase after UV photoactivation even though H40 had 10 more paired bases than H30. However, ‘background’ RNA cleavage for H40 in the dark was much higher than H30 and C20. Clearly, H40 was fully complementary to the 40-mer target sequence, so the binding energy of the duplex was much larger than that of H30 or C20. H40 also had larger circle than H30 or C20, and it was much easier for 40-mer RNA to wind around H40 than H30 or C20, which should be less efficient to photomodulate target RNA digestion by RNase H.
To confirm which is the most important factor for photomodulation of RNA digestion between ring size and the number of paired bases, another two caged circular asODNs (C30 and C40) were introduced with 30 and 40 deoxynucleotides as H30 and H40, respectively, but only had the middle 20 deoxynucleotides complementary to the middle 20 nucleotide sequence of the 40-mer RNA target. Under the same RNase H assay conditions, around 50% 40-mer RNA was digested by RNase H in 30 min for both light-activated C30 and C40 which were 2- and 1.9-fold more than caged circular asODNs, C30 and C40 as shown in Figure 2. Comparing C30 with H30, C40 with H40, RNase H assay results showed that there was RNA digestion increase by RNase H (from 25%, 26% for C30 and C40 to 28%, 40% for H30 and H40) in the ‘caged’ state of circular asODNs, which was due to the more paired nucleotides and binding energy between asODNs and the 40-mer target RNA. In comparison with C20, C30 and C40 in the caged state, C20 with smaller 20-deoxynucleotide ring almost had no detectable 40-mer target RNA digestion, while there were 25% and 26% of target RNA cleaved for C30 and C40 by RNase H under the same conditions. And the photomodulation efficiency for C30 and C40 is much smaller than C20, which is at least >20-fold increase of RNA digestion with light activation when the ring was as small as a 20-mer oligodeoxynucleotide. To further prove that the ring size did matter for photoregulation of a certain length of RNA digestion by RNase H, we replaced the 40-mer target RNA with a shorter 20-mer RNA (5′-AAUACGGUCCGAAACGUUGG-3′) that was complementary to the sequence of C20. RNase H assay results with the 20-mer RNA showed that the 20-mer RNA was readily digested by RNase H even though C20 was in the ‘caged’ state (see Supplementary Figure S5). The result was not surprising, as we knew the short RNA would wind around the caged circular C20 more easily than the longer RNA due to the requirement of the helix structure for duplex formation. So the relative size of a cagd asODN ring and its target RNA was the most important factor to photomodulate the RNA digestion. For C20, it could not be used to photoregulate the 20-mer RNA digestion by RNase H. However, it was able to efficiently photocontrol a 40-mer target RNA digestion. For the large ring size of asODNs as H30, H40, C30 and C40, we expect they might readily photoregulate longer RNA digestion, such as mRNA in cells.
RNA cleavage by RNase H is a three-component system requiring the presence of a target RNA, an asODN and RNase H. RNase H binds to the RNA/asODN duplex, and then the RNA cleavage event can happen. To examine our design strategy, gel shift assays of interaction between the RNA and caged circular asODNs were used. We fixed the concentration of the 40-mer RNA at 0.02 μM in 10 μl RNase H buffer, then different ratios of asODN/RNA were used to determine the relative binding ability of the 40-mer RNA to caged circular asODNs. Unsurprisingly, the native gel assays indicated that only 0.04 μM of H40 was able to bind all of the RNA and form duplex, while it needed >0.5 μM H30 to form the duplex with the same amount of the RNA (see Supplementary Data). For caged circular asODNs such as C30 and C40, the native gel assays indicated that 0.2 μM C40 were needed for binding most of the 40-mer RNA and forming the stable duplex. Compared to H40, C40 had less complementary deoxynucleotides, which needed more caged asODN in order to fully occupy the 40-mer RNA. However, it needed more than 0.5 μM caged asODNs for both C30 and H30 to bind the same amount of the 40-mer RNA, which showed that the ring size of caged asODNs was the decision-maker for RNA binding. Gel shift assay result of C20 further confirmed that the ring size was the most important to binding affinity of the caged circular asODNs with the target RNA. For the 40-mer RNA, even 4 μM of C20 could not fully bind all of the RNA, and bound much less efficiently than C30 and C40, as shown in Figure 3. These results were consistent with RNase H assay experiments.
Interestingly, the position of RNA cleavage by RNase H was different for caged H40 from other caged circular asODNs, even though the antisense sequence of H30 covered the middle 30 deoxynucleotides of H40. The RNase H cleavage site was mapped using a 5′-end-P32 radiolabeled RNA probe. An alkaline RNA ladder was included to identify each nucleotide and enable sequence identification. The main cutting point was the phosphodiester bond of A and C for H30, and the phosphodiester bond of A and G for H40, as shown in Figure 4. To confirm that the short fragment of RNA with the H40 existence was not the second cleavage by RNase H, time dependent of RNA digestion was studied. From time point 2–30 min, the cleavage profile was the exact same without any intermediate cleavage fragment (See Supplementary Data). The results showed that RNase H could prefer binding the certain sequences and cutting the specific RNA phosphodiester bonds. The caged asODN and linear asODN had the same RNA cleavage patterns, and UV activation did not change the cleavage position of the RNA.
We designed a new type of caged asODNs with a single photolabile linker and end-linked circular structure. Two of the circular caged asODNs (H30 and H40) have the hairpin-like structures which are much more stable than their related uncaged antisesne ODNs with melting temperature increase as high as 17.5°C and 11.6°C, respectively. The other three caged circular asODNs (C20, C30 and C40) with different numbers of deoxynucleotides in the ring do not have stable secondary structures. All these caged circular asODNs have restricted opening of oligodeoxynucleotide rings, which provides control over RNA/asODN duplex formation. RNase H assay results showed that the 40-mer RNA digest was photomodulated about 2- to 3-fold upon light activation with caged H30, H40, C30 and C40, while with the presence of caged circular C20, RNA digestion was almost not detectable before light activation, however, photoactivation triggered >20-fold increase of RNA digestion by RNase H. And gel shift assays showed that it needed >0.04 μM and 0.5 μM of H40 and H30 in 10 μl solution to completely bind a fixed concentration (0.02 μM) of the 40-mer RNA. For caged C40 and C30, it needed >0.2 μM and 0.5 μM caged asODNs to fully bind the fixed concentration of the 40-mer RNA. However, even 4 μM of C20 was not able to completely bind the fixed concentration of the 40-mer RNA. But the replacement of the 40-mer RNA with a 20-mer RNA caused RNA to efficiently interact with caged C20, and assay results showed that the 20-mer RNA easily bound caged C20 and was readily digested by RNase H even without light activation. By simply adjusting the ring size of caged circular asODNs, we succeeded in photomodulating the 40-mer RNA hybridization and digestion by RNase H. Based on the results of C20, C30, C40, H30 and H40, the ability of their interaction to target RNA showed that using smaller loop of caged circular asODNs or longer RNA helped increase the photomodulation efficiency of RNA digestion by RNase H. We expect that these types of photolabile circular asODNs can be quickly released upon light activation. And the uncaged asODNs can subsequently interact with the target RNA, successfully regulate their hybridization with mRNA, and target RNA hydrolysis by RNase H. More studies of optimization of caged end-linked circular asODNs and their in vivo applications are on the way. And RNA degradation profile also showed that RNA digestion with the presence of caged circular asODNs did not change the RNA cleavage pattern by RNase H before and after light activation.
Supplementary Data are available at NAR Online.
Start-up fund came from State Key Laboratory of Natural and Biomimetic Drugs (Key Laboratory Grant) and Peking University “985” foundation (bmu-2009137-121). Funding for open access charge: Start-up fund.
Conflict of interest statement. None declared.
We thank State Key Laboratory facility for instrumental support and technical assistance.