已建立且具有高度预测性

应用

凭借适用于各种生物测定的简单协议，MatTek 的应用程序被广泛应用于世界各地的生命科学行业。

+ 应用程序 50 多个相关应用 皮肤 – 医疗器械提取物的皮肤刺激 (ISO 10993-23:2021)、皮肤刺激刺激 (MTT ET-50)、皮肤刺激 (OECD TG 439)、皮肤腐蚀 (OECD TG 431)、抗衰老皮肤免疫 Cell – 基础 DC 研究、微生物感染、过敏性测试 免疫细胞 肠道 – 上皮恢复、肠道纤维化、药物代谢、药物输送（肠道）、炎症 肠道 眼部 – 眼睛刺激 (MTT ET-50)、眼睛刺激测试 (OECD TG 492) 眼部口腔粘膜 – 口腔刺激、吸收、口腔念珠菌病、药物输送（口腔）粘膜） 口腔粘膜 呼吸道 - 呼吸道感染、药物输送、吸入毒理学、炎症和纤维化 呼吸道 阴道 - 炎症、女性卫生、性病感染、杀菌剂测试 阴道Pri mary Human Cells & Media

MatTek 从事细胞分离工作已超过 25 年。适用于 2D 和 3D 人类生物学的 3D 认证细胞。

我们的产品 皮肤细胞

表皮角质形成细胞、黑色素细胞、真皮成纤维细胞

气道细胞

支气管上皮细胞、支气管成纤维细胞

免疫细胞

树突状细胞、浆细胞样树突状细胞、T 细胞

您的购物车目前是空的。

返回商店

我的账户登录注册信息请求

感谢您请求有关 MatTek 产品的信息！代表将很快与您联系。

**如果您想订购 MatTek 产品，请联系客户服务**

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. AbstractModification of membrane receptor makeup is one of the most efficient ways to control input-output signals but is usually achieved by expressing DNA or RNA-encoded proteins or by using other genome-editing methods, which can be technically challenging and produce unwanted side effects. Here we develop and validate a nanodelivery approach to transfer in vitro synthesized, functional membrane receptors into the plasma membrane of living cells. Using 尾2-adrenergic receptor (尾2AR), a prototypical G-protein coupled receptor, as an example, we demonstrated efficient incorporation of a full-length 尾2AR into a variety of mammalian cells, which imparts pharmacologic control over cellular signaling and affects cellular phenotype in an ex-vivo wound-healing model. Our approach for nanodelivery of functional membrane receptors expands the current toolkit for DNA and RNA-free manipulation of cellular function. We expect this approach to be readily applicable to the synthesis and nanodelivery of other types of GPCRs and membrane receptors, opening new doors for therapeutic development at the intersection between synthetic biology and nanomedicine. IntroductionMembrane proteins play important roles in all aspects of cellular signaling and are the interface through which a cell responds to extracellular cues. One of the most important subsets of the cellular membrane protein makeup are G-protein鈥揷oupled receptors (GPCRs), which sense a variety of external cues to orchestrate a broad range of cognitive and physiological responses and are targets for approximately 30% of all clinically-approved drugs1. Therefore, modulation of GPCRs is a powerful means to manipulate cellular signaling and phenotype, enabling a broad spectrum of research and clinical applications. Unfortunately, expression of functional membrane proteins can only been achieved through the introduction of DNA or RNA-based genetic material into cells, which can be technically challenging in primary cells and offers poor control over the amounts of protein being produced2.To obtain pure and soluble GPCRs while maintaining proper folding and transitions between conformational states, individual receptor molecules need to be in a native-like environment, which can be provided with the use nanolipoprotein particles (NLPs), also known as nanodiscs3,4. Nanodiscs hold great promise as a platform for efficient cellular or in vivo delivery of membrane-associated molecules, due to their highly bio-compatible properties (e.g. stability, lack of toxicity, biodistribution)5. However, their use has been limited to the delivery of drugs6, phospholipids7 and personalized cancer vaccines8, while nanodelivery of functional membrane receptors has not been reported.Traditionally, nanodisc-solubilized GPCRs have been produced using cell-based methods that require expression in cell hosts followed by detergent extraction and nanodisc reconsitution, which is time-consuming and labor intensive3,9. Cell-free synthesis provides a viable alternative for large-scale production of membrane proteins and is emerging as a competitive choice due to its increasing production yields (up to mg/ml reaction amounts in the most optimized systems) and to its lower expression costs9,10,11,12,13,14,15, especially with the adoption of alternative energy regeneration systems for protein synthesis16.Here, we developed a platform integrating cell-free production and nanodelivery of functional GPCRs to the plasma membrane of cells. Using the well-studied 尾2 adrenergic receptor (尾2AR) as a prototypical GPCR17, we validated the utility of this platform by demonstrating that membrane-delivered 尾2AR responds to ligand binding and triggers cAMP production to rescue the wound healing defects of 尾1/尾2AR double knockout primary cells. We expect this platform to be readily adapted to the cell-free production and nanodelivery of a broad range of GPCRs and other membrane proteins for DNA and RNA-free manipulation of cellular functions.Full-length production of 尾2AR-NLPs by cell-free, co-translational approachesTo achieve large-scale, cell-free expression of functional 尾2AR, we first codon-optimized the cDNA encoding human 尾2AR protein for expression using Expressway鈩?Maxi Cell-Free E. coli Expression System (Supplementary Fig.聽1). To induce spontaneous insertion of 尾2AR into 螖49A1-induced NLPs, we co-expressed codon-optimized 尾2AR with human apolipoprotein A-1 lacking the amino-terminal 49 amino acids (螖49A1) in the presence of 1,2-ditetradecanoyl-sn-glycero-3-phosphocholine (DMPC) (Fig.聽1a), as previously described11,18.Figure 1Cell-free production of 尾2AR-NLPs. (a) Co-translational production of 尾2AR-NLPs was achieved by incubating 尾2AR and ApoAI DNAs in the presence of synthetic lipid nanoparticles (1,2-dimyristoyl-sn-glycero-3-phosphocholine, DMPC) with the cell-free reaction mixture at 30鈥壜癈 for 18鈥塰ours. (b) Optimization of the GPCR-NLPs in vitro translation system. 尾2AR was cell-free produced in the absence of 螖49A1 cDNA and titrated in the presence of 螖49A1 cDNA (10:1, 20:1, 40:1). FluoroTect鈩?GreenLys (Promega) was added for visualizing newly synthesized proteins. Cell-free reactions were centrifuged at 10,000鈥塺pm for 10鈥塵inutes. 5鈥壩糽 aliquots of sample before centrifugation (total fraction, T) and of the supernatant (soluble fraction, S) were used for SDS-PAGE. Gel images were taken using Molecular Imager庐 Gel Doc鈩?XR System from Biorad. (c) Band intensity ratio of the soluble vs. total cell-free produced 尾2AR was utilized to determine solubility. Results are shown as mean鈥壜扁€塖.E. (d) The identity and molecular weight of cell-free expressed 尾2AR was verified by side-by-side comparison with cell-expressed 尾2AR, via western blot analysis using anti-flag labeling.Full size imageWe then maximized the production of NLP-solubilized full-length 尾2AR protein by varying the ratio of p尾2AR to p螖49A1 plasmids in the cell-free co-expression reaction mix (Fig.聽1b). We found that a DNA ratio of 10:1 for p尾2AR to p鈭?9A1 promoted about 70% solubilization of 尾2AR in the total translated protein fraction (Fig.聽1c and Supplementary Fig.聽2a). Using western blot analysis, we further confirmed the expression of full-length 尾2AR (~47鈥塳Da) using an anti-flag antibody (Fig.聽1d and Supplementary Fig.聽2b).To conclusively demonstrate 尾2AR association with NLPs, we first purified particles by affinity chromatography via the 螖49A1 6XHis tag, followed by characterization using size-exclusion chromatography and transmission electron microscopy (TEM). The chromatographic analysis showed two clearly distinct elution peaks representing 尾2AR integrated into NLPs and empty NLPs (Supplementary Fig.聽3), suggesting stable association of 尾2AR with NLPs. TEM further confirmed that approximately 27.5% of the NLPs in the purified sample contained integrated 尾2AR (Fig.聽2a,b). 尾2AR-integrated NLPs displayed a significantly larger diameter than empty NLPs when imaged by TEM (尾2AR-NLPs: 33.0鈥壜扁€?.0, empty NLP: 21.5鈥壜扁€?.4, P鈥?鈥?.8641E-15). These results indicate that cell-free expressed 尾2AR successfully integrates into the NLP-supported bilayer.Figure 2Transmission Electron Microscopy (TEM) characterization of 尾2AR-NLPs. TEM images of stain-embedded empty NLPs with average diameter of 21.5鈥塶m and 尾2AR-NLPs with average diameter of 23鈥塶m. Scale bar: 50鈥塶m. (c) Particle size distribution of empty and 尾2AR-NLPs from TEM data. Standard deviation for empty and beta-adrenergic NLPs are 1.4 and 3.0, respectively (p鈥?lt;鈥?.0001, paired t test).Full size imageThe 尾 2AR-NLP complex is functional in vitro We next sought to investigate the functional integrity of NLP-bound 尾2AR. To do so we assayed the ligand-induced conformational changes of NLP-bound 尾2AR using a single-molecule pull-down assay (SiMPull)19,20, tapping into a 尾2AR conformational biosensor (Nb80-GFP) that specifically stabilizes the active conformation of 尾2AR21. Immobilized 尾2AR-NLPs or empty NLPs were first pre-incubated with either antagonist ICI118551 (ICI, 10鈥壩糓) or agonist isoproterenol (ISO, 10鈥壩糓) before pulling down purified Nb80-GFP. Only in the 尾2AR-NLP sample in the presence of the agonist we could detect green fluorescent spots, representing 尾2AR-bound Nb80-GFP (~313 molecules). We observed a small number of pull-down Nb80-GFP either in the presence of ICI (~31 molecules) or under the conditions when empty NLPs were immobilized (~17 molecules) (Fig.聽3a,b). As an additional confirmation of the specificity of binding, we also pulled down mCherry-labeled Gs protein in the presence of ISO, but not ICI (Fig.聽3c,d). Importantly, we observed different association rates of Nb80-尾2AR binding in the presence of ligands with different pharmacological properties (Supplementary Fig.聽4a). The association rates for two full agonists (ISO and epinephrine, EPI, 10鈥壩糓) and one partial agonist (dobutamine, DOB, 10鈥壩糓) are 5.3*108, 1.4*109 and 1.8*108鈥塎鈭? min鈭?, respectively, which correlate with the known efficacy of the drugs examined (Supplementary Fig.聽4b)22. These results suggest that cell-free expressed 尾2AR maintains functional integrity when integrated into NLPs and can distinguish the pharmacological properties of different ligands.Figure 3Functional characterization of 尾2AR-NLPs for binding to Nb80-GFP or cognate Gs-protein. (a,c) Purified 尾2AR-NLPs or empty NLPs were immobilized on a chip via anti-flag antibody (M2, Sigma). Nb80-GFP or Gs-mcherry-containing HEK cell lysates were flown over the immobilized 尾2AR in the presence of either the 尾2AR agonist isoproterenol (ISO, 10鈥壩糓) or the 尾2AR antagonist ICI118551 (ICI, 10鈥壩糓). (b,d) Average number of fluorescent molecules per imaging area, Nf. Results are shown as mean鈥壜扁€塻.d. (n鈥?gt;鈥?; ***p鈥?lt;鈥?.0001).Full size imageBecause our TEM data allows us to determine only the percentage of NLPs-bound 尾2AR versus empty NLPs, we further evaluated the fraction of functional receptors out of the NLP-bound 尾2AR pool using SiMPull (Supplementary Fig.聽5a). To do so, we compared the number of pulled down Nb80-GFP molecules in the presence of 10鈥壩糓 ISO to the total number of immobilized 尾2AR-NLPs, visualized with an anti-6xHis-555 antibody (Supplementary Fig.聽5b,c). Assuming that 75% GFP is fluorescent, as previously reported20,23, we estimated that the percentage of functional 尾2AR particles was 10% of the total purified sample, which corresponds to 30鈥塶g (330鈥塮mol; 2*1011 molecules) per microliter (ul) cell-free reaction.Functional insertion of cell-free expressed 尾 2AR-NLPs in living cellsRecent studies have described the application of NLPs to both in vitro and in vivo delivery of hydrophobic drugs and membrane antigens5,6,8. Thus, we tested whether cell-free expressed 尾2ARs could be delivered via NLPs onto cell membranes and maintain their function. When DiO-labeled empty NLPs were incubated with HEK 293 cells in serum-free medium, green fluorescence could be readily detected on the cell membrane after 14鈥塰r incubation, suggesting efficient transfer of the lipid component of NLPs onto cell membranes (Supplementary Fig.聽6a). To test nanodelivery of 尾2AR, we applied purified NLP-bound 尾2AR onto 293 cells (100鈥塶M) for different amounts of time, followed by live-cell staining and imaging with an anti-flag antibody conjugated to Alexa-546 (Fig.聽4a). After just 6鈥塰r of incubation, red fluorescence intensity could already be detected at the cell membrane, indicating 尾2AR incorporation onto the cell membrane and continued to increase up to 18鈥塰r of incubation, when it reached 10-fold higher intensity compared to the 6鈥塰r timepoint (Fig.聽4b).Figure 4Nanodelivery of in vitro translated 尾2AR on cells. Time-course of 尾2AR-NLPs insertion onto HEK293 cell membranes was performed by live-cell anti-flag staining at various time points of incubation with 尾2AR-NLPs. (a) Representative fluorescence images showing flag staining intensity after 18鈥塰ours incubation. (b) Analysis of flag cell staining indicates efficient membrane insertion after overnight incubation (18鈥塰ours) (n鈥?鈥?0 cells, p鈥?lt;鈥?.0001). Results are shown as mean鈥壜扁€塖.E. (c) Isoproterenol induced membrane-relocalization of Nb80-GFP in HEK293 cells after 尾2AR nanodelivery (18鈥塰ours, n鈥?鈥? cells). (d) Representative traces of fluorescence from line indicated in c.Full size imageThe conformational biosensor Nb80-GFP has been previously used for direct visualization of 尾2AR activation in living cells21. We thus used this approach to verify that nanodelivered 尾2AR maintained functional integrity as demonstrated by efficient cytosol-to-membrane recruitment of Nb80 and production of cyclic AMP (cAMP) upon agonist exposure (Fig.聽4c). After 18 h of 尾2AR nanodelivery on 293 cells expressing Nb80-GFP, we performed dual-color, time-lapse imaging of flag staining and Nb80 relocalization in response to 10鈥壩糓 ISO. We observed the rapid relocalization of Nb80-GFP from the cytoplasm to the membrane, confirmed by a marked increase in green fluorescence intensity on the cell membrane (~75%), while the red fluorescence intensity on the membrane remained constant (Fig.聽4d and Supplementary Fig.聽6b).These results suggest that cell-free鈥損roduced 尾2AR can be functionally delivered to the cell membrane, resulting in a conformational response to agonist stimulation. More importantly, when stimulated by agonist, nanodelivered 尾2AR triggered a downstream signaling cascade comparable to that of endogenous 尾2AR receptors (Fig.聽5a). We measured cAMP production with the fluorescent resonance energy transfer (FRET)-based cAMP biosensor ICUE3 in living cells24. As isoproterenol is a nonselective 尾1AR/尾2AR agonist and most cell types endogenously express 尾ARs, we isolated primary neonatal cardiac myocytes from both 尾1/尾2AR double-knockout mice (dKO) and wild-type mice. Upon isoproterenol application, we observed cAMP production demonstrated by 15% FRET changes only in wild-type, but not in dKO myocytes (Fig.聽5b). When 尾2AR was nanodelivered onto dKO primary myocytes for 14鈥塰r, we observed 14% FRET changes in response to ISO, which was similar to that observed in wild-type cells. In contrast, 14鈥塰r incubation of dKO myocytes with empty NLPs failed to generate any detectable FRET changes (Fig.聽5c).Figure 5Nanodelivered 尾2AR can drive cellular signaling. (a) Schematic of GPCR nanodelivery to cellular membranes to alter cellular properties. (b,c) ICUE measurement of cAMP production in primary 尾1/尾2AR double-knockout neonatal cardiac myocytes (dKO myocytes) shows that nanodisc-delivered 尾2AR is capable to elicit cellular signaling in response to the agonist isoproterenol at a level comparable to endogenous 尾2AR (**p鈥?lt;鈥?.01, unpaired t test). Results are shown as mean鈥壜扁€塖.E.Full size imageReceptor nanodelivery alters cellular phenotypeThe 尾2AR signaling has been directly linked to fibroblast migration during wound-healing processes25,26. Thus, we next asked whether 尾2AR nanodelivery could be used to induce a pro-migratory phenotype in cultured fibroblasts. To do so, we nanodelivered 尾2AR onto mouse embryonic fibroblasts (MEFs) from our dKO mice and then assayed their wound healing properties using an in vitro scratch assay. At 24 and 48鈥塰r after scratch application, dKO MEFs displayed 33% and 57% gap closure, respectively. Upon 尾2AR nanodelivery, both with and without agonist stimulation (10鈥壩糓 ISO), the wound-healing properties of these cells were significantly increased (Fig.聽6a). 尾2AR-insertion without agonist application resulted in 50% and 79% gap closure at the two timepoints, indicating that basal activity of the receptor is sufficient to elicit a significant wound-healing effect (p鈥?鈥?.014 after 24鈥塰r). Agonist stimulation in addition to 尾2AR nanodelivery led to a further increase in gap closure, reaching 66% and 95% at the 24 and 48鈥塰r time points, respectively (Fig.聽6b, p鈥?鈥?.0002 after 24鈥塰r and p鈥?鈥?.008 after 48鈥塰r, n鈥?鈥?). These results validate the feasibility of GPCR nanodelivery and prove that this is a suitable strategy to selectively modulate cellular signaling and phenotype.Figure 6Nanodelivered 尾2AR increases rescues wound healing defects of 尾1/尾2AR dKO mouse embryonic fibroblasts (MEFs). (a) Representative images of dKO MEFs treated with 尾2AR-NLPs or empty NLPs with or without isoproterenol and subjected to scratch assay for cell migration. Wound healing was analyzed at 24 and 48鈥塰ours. Red lines represent the cell front. (b) Wound healing was quantified as percentage of gap closure. Results are shown as mean鈥壜扁€塖.E. (*p鈥?lt;鈥?.05, **p鈥?lt;鈥?.01, n鈥?鈥?, unpaired t test).Full size imageDiscussionIn summary, we demonstrated for the first time a method for nanodelivery of membrane proteins, in which an in vitro one-step, cell-free system can be used to synthesize functional nanodisc-solubilized GPCRs that can be further delivered onto plasma membranes of cells to trigger signaling cascades and alter cellular function. To demonstrate the feasibility of GPCR nanodelivery and provide proof of concept for the therapeutic potential of this new delivery system, we cell-free produced nanodisc-solubilized 尾2AR and fully characterized its function both prior and after delivery on cell membranes. We validate nanodelivery of 尾2AR in three different cell types, including hard to transfect primary cells, where we show that nanodelievered 尾2AR triggers intracellular cAMP production in a similar manner to naturally occurring endogenous 尾2AR and imparts novel functional properties to the receiving cells, as evidenced by increased cell migration in an ex vivo wound-healing model using 尾1/尾2AR double knockout cells.Our method greatly expands the existing toolbox for direct protein transduction to living cells independent of DNA or RNA-based techniques2 and affords the benefits of cell-free, large-scale, in vitro protein expression (e.g., speed, purity, lower costs and no need for detergents).Since a decade ago when membrane protein synthesis in cell-free systems started27, the yield-on-cost ratio has been continuously increasing. Membrane protein synthesis in cell-free systems (e.g. E. coli and wheat germ based cell-free systems) has started to reach levels comparable to soluble proteins15,28, especially with the development of a continuous exchange cell-free dialysis system (CECF) that provides a prolonged reaction time and freshly supplied reaction components14. Though the CECF reaction is approximately 10 times more expensive compared to a typical batch reaction, with the increased productivity, it is possible to achieve a significant reduction of the total costs by a factor of 10. The main costs of cell-free synthesis systems arise through the addition of T7 RNA polymerase, energy in the form of adenosine and guanosine triphosphate and an energy regeneration system. The costs of exogenous added T7 RNA polymerase might be decreased by using in house production. The development of alternative energy-rich components and the energy regeneration systems is already in process16. The yield-on-cost ratio of cell-free system for membrane protein synthesis will continue to increase given the increasing productivity of cell free systems in combination with the reduced costs for lysates and the energy regeneration system.In the current study, we adopted an innovative approach to estimate the amount of functional GPCR product in the final expression product, which takes advantage of a protein binding assay (SiMPull). Because of the intrinsically different experimental procedures used to determine the yield of functional GPCR between this and previous studies, an exact comparison among yields produced by the different systems would not be accurate. However, the relatively overall lower efficiency of functional GPCR expression using in vitro translation (ng per ul cell-free reaction) compared to the broadly used insect cell-baculorivus expression system ( 1鈥塵g/L total purified proteins)28,29 is possibly due to the intrinsic limitations of the system. In fact, the tRNAs, amino acids, ribosomes and other components of the mixture are limited in an in vitro translation reaction, which further limits the time length of reaction efficacy. In addition, selection of different lipid types may further increase the solubility of the protein, though extensive characterization has been performed in the past15,28. Successful strategies to further optimize cell-free synthesis of GPCRs and membrane proteins in general have been done on a variety of lipid structures like liposomes, micelles, bicelles and nanodiscs27,30,31. Future work will focus on further increase the yields by applying modifications to the cell-free reaction conditions, such as integrating continuous exchange cell-free dialysis system and systematic optimization in selection of lipid types.Given the efficiency of membrane protein incorporation strictly depends on the concentration of nanodisc-solubilized proteins applied onto the cells, our system also provides a much higher degree of precision in fine-tuning the membrane levels of delivered receptors than those achieved by DNA expression. The high simplicity and flexibility of this system may allow nanodelivery of a broad range of membrane proteins, such as other GPCRs, optogenetic actuators and reporters in various cell lines. Furthermore, It is promising that this system may allow nanodelivery of GPCRs to living organisms for both research and clinical uses, such as revision restoration32,33. However, extensive research focusing on substantial optimization, such as in vivo stability of the receptor complex, the bioavailability in certain tissue types and the specificity of the nanodelivery, is necessary to realize the full potential of this system for in vivo applications.MethodsCell-free, in vitro translation reaction for GPCR-NLP self-assemblyCell-free reactions using the ExpresswayTM Maxi Cell-Free E. coli Expression System (Life Technologies) were carried out as previously described18. The gene for 尾2AR was codon-optimized for E. coli expression. A Flag tag was positioned after the start codon and NdeI and SmaI restriction sites were positioned at the beginning and the end of the construct, respectively. The sequence was ordered as a Geneblock (Integrated DNA Technologies) and cloned using the NdeI-SmaI sites into pIVEX2.3d and pIVEX2.4d (Roche). The pIVEX2.3d-尾2AR construct was used to prepare 尾2AR-NLPs for cell-membrane insertion, followed by live staining, which required the flag tag to be exposed at the very beginning of the receptor N-terminus; the pIVEX2.4d-尾2AR construct resulted in a protein product with an N-terminal 6xHis tag preceding the flag tag and was used for all other applications. The base changes that were made during codon optimization are shown in Supplementary Fig.聽3. Small, unilamellar vesicles of DMPC (liposomes) were prepared by sonication of a 25鈥塵g/ml water鈥揇MPC solution on ice until optical clarity was achieved (usually 10鈥塵in), followed by 2鈥塵in of centrifugation at 14,000 rcf to remove metal contamination from the sonication probe tip. DMPC small, unilamellar vesicles were added to the cell-free reaction mixture prior to starting the reaction at a final concentration of 2鈥塵g/ml. Addition of FluoroTect鈩?GreenLys (Promega) was done at the beginning of the reactions to facilitate visualization and quantification of synthesized proteins. For membrane protein and NLP coexpression in a 200鈥壩糽 reaction, 2.5鈥壩糶 of plasmid DNAs were added to the lysate mixture. The ratio of 尾2AR to 螖49A1 plasmid DNA (containg a N-terminal HIS tag) was kept constant at 20:1. The reactions were incubated at 30鈥壜癈 for 18鈥塰. Empty NLPs were generated by omitting the 尾2AR plasmid DNA.Affinity purification of cell-free鈥損roduced NLPsEmpty NLPs or 尾2AR-NLPs, both containing 6xHIS tags on their ApoAI component, were purified from their respective reaction mixtures by Ni/NTA affinity purification. Briefly, we used immobilized metal ion affinity chromatography (IMAC) tips on an automatic pipette (80鈥壩糽 resin volume, Mettler-Toledo). The resin was first equilibrated with native buffer (Imidazole 20鈥塵M, NaCl 300鈥塵M, NaHPO4 50鈥塵M, pH 8.0), then incubated with the NLP-containing reaction mixture and washed three times before elution of the NLPs in native buffer with 400鈥塵M imidazole. Eluted NLPs were then dialyzed in 2鈥塋 of phosphate-buffered saline (PBS) buffer using 3.5鈥塊Da molecular weight cutoff D-tube dialyzers (Millipore). Total protein concentration was measured by bicinchoninic acid (BCA) assay (Thermo) on a Synergy2 plate reader (BioTek), using a standard curve with known concentrations of BSA. SDS-PAGE and fluorescence imaging of the gels were performed as previously described18.Nb80-GFP quantificationRecombinant enhanced green fluorescent protein (EGFP; 1鈥塵g/ml) was purchased from Vector Laboratories. Nb80-GFP containing 5xHis-tag was expressed in HEK293T cells and purified using IMAC resin tips (Rainin). A series of dilutions of EGFP and Nb80-GFP was made using Tris-EDTA buffer (10鈥塵M Tris-HCl, 10鈥塵M EDTA, pH 8.0) as the diluent. Fluorescence was determined by a Synergy 2 fluorescence microplate reader (BioTek Instruments) using a 485鈥塶m, 20鈥塶m bandpass excitation filter and a 528鈥塶m, 20鈥塶m bandpass emission filter with an instrument sensitivity setting of 50. Quantitation of EGFP was determined using a calibration curve determined by linear regression (r2鈥?鈥?.998).Transmission electron microscopyEmpty NLPs and 尾2AR-NLPs were diluted to a final concentration of 0.2鈥塵g/mL. A deep-staining approach was utilized for sample embedding immediately prior to examination with a JEOL 1230 TEM. All samples were mixed with 16% ammonium molybdate and 0.1% trehalose and immediately transferred to a carbon-coated copper grid. The grids were further prepared for imaging with a single wash using PBS buffer and were air-dried. All TEM micrographs were captured at 60,000x magnification and processed for size distribution measurements in FIJI package. From 8 micrographs, 140 particles were measured by the long axis. The measurements for empty and 尾2AR-NLPs were plotted and standard deviations were calculated using Microsoft Excel.Single-molecule pull-down assay (SiMPull)For the 尾2AR functionality test, cells were plated onto 100鈥塵m petri dishes (VWR) at a density of 200,000 per dish and grown for 24鈥塰 before being transfected with 3鈥壩糶 of Gs-mCherry plasmid DNA. Cell lysates were prepared by cell titration in apyrase reaction buffer (NEB) supplemented with protease inhibitors cocktail (Sigma) (20鈥塵M MES, 50鈥塵M NaCl, 5鈥塵M CaCl2, 1鈥塵M DTT, 0.05% Tween-20, pH 6.5). Complete cell lysis was ensured by probe-sonicating on ice for 30鈥塻. Cell debris was removed by 10鈥塵in of centrifugation at 14,000 rcf (4鈥壜癈). Nearly all guanine nucleotide triphosphates and diphosphates (NTPs and NDPs) were hydrolyzed to monophosphates (GMPs) by incubating the lysate with apyrase at 30鈥壜癈 for 1鈥塰. SiMPull chips were first washed twice with T50 buffer (50鈥塵M NaCl, 10鈥塵M Tris-HCl, pH 7.8). Next, NeutrAvidin (Thermo) was added and slides were incubated for 5鈥塵in at room temperature (RT). After washing twice with T50, biotinylated anti-flag antibody (10鈥塶M) was added and incubated for 10鈥塵in at RT, then washed twice with T50 again. Subsequently, 1鈥壩糶 of purified cell-free鈥損roduced NLPs (either containing flag-tagged 尾2AR or empty nanodiscs as a negative control) was added, followed by 10鈥塵in incubation at RT. Unbound samples were removed from the chamber by washing twice with T50 buffer. Cell lysates expressing Nb80-GFP or Gs-mCherry were 1:10 diluted into T50 buffer containing 10鈥壩糓 isoproterenol (Sigma) or ICI (Sigma), then added to the SiMPull chamber and incubated for 10鈥塵in at RT, followed by washing twice with T50 with isoproterenol or ICI. Proteins immobilized on the slides were visualized using a prism-based, total internal reflection fluorescence microscope equipped with excitation laser 488鈥塶m (GFP) and 561鈥塶m (mCherry) and DV2 dichroic 565dcxr dual-view emission filters (520/30鈥塶m and 630/50鈥塶m). Mean spot counts per image and standard deviations were calculated from images taken from 5 to 10 different regions using a script written in Matlab. For drug-specific Nb80 to 尾2AR ON rate determination: 1鈥壩糶 of cell-free鈥損roduced 尾2AR-NLP was immobilized on SiMPull chips and preincubated with 10鈥塶M drug (isoproterenol, dobutamine, or epinephrine) for 15鈥塵in. Immediately after injection of 1鈥塶M purified Nb80 (diluted in T50 buffer containing 10鈥塶M drug), several images were taken at each time point (0, 30鈥塻 and 1, 2, 3, 4, 5, 10, 15, 20 and 25鈥塵in). 蟿1/2 was determined by fitting the data with a one-phase association curve (GraphPad Prism 6).Cell culture, DNA constructs and transfectionHEK293 cells (ATCC #CRL-1573) were grown in Dulbecco鈥檚 modified Eagle鈥檚 medium (DMEM, Invitrogen) supplemented with 10% fetal bovine serum and 1鈥塵g/ml penicillin-streptomycin. The pEGFP_N1 plasmid containing Nb80-GFP was a gift from Dr. Huang Bo (UCSF). The original rat Gs protein pcDNA was from Addgene (55793). The pcDNA plasmid containing Gs-mCherry was made by restriction cloning using a single BamHI site inserted after amino acid 71 in the Gs protein. The mCherry insert was amplified by PCR with the addition of two BamHI sites and two flexible linker regions (GGGS) on each side prior to restriction cloning. The correct insertion of mCherry was confirmed by sequencing. Cell transfections were performed with Effectene庐 Transfection Reagent (Qiagen) according to manufacturer鈥檚 protocol. All cell-culture reagents were from Life Technologies, unless otherwise noted.Live-cell confocal imagingFor insertion of NLPs onto cell membranes, the cells were plated onto 35鈥塵m glass bottom microwell dishes (MatTek) at a density of 10,000 per well and transfected with 100鈥塶g of Nb80-GFP plasmid DNA. After 14鈥塰, the medium was switched to serum-free medium with the addition of 10鈥壩糶 of purified total NLPs (corresponding to a final concentration of approximately 55鈥塶M). The removal of serum is meant to prevent absorption of NLPs onto serum albumin and increase the efficiency of NLP insertion onto cell membranes. To maximize the insertion onto cell membranes, cells were incubated with the NLPs overnight. For surface-labeling of cells with membrane-inserted receptors, cells were washed three times in HBSS (Life Technologies) and supplemented with 2鈥塵M Ca2+ and 15鈥塵M HEPES (pH 7.4), prior to being incubated at 37鈥壜癈 for 15鈥塵in with M1 anti-flag antibody (1:1000, Sigma) conjugated to Alexa 546 (A-20002, Thermo Fisher). The cells were imaged in HBSS buffer with a 40鈥壝椻€塷il-based objective on an inverted confocal microscope (Observer, Zeiss). Images were acquired with a CCD camera driven by the Zen software (Zeiss). Isoproterenol (Sigma) was carefully added directly onto the dish during the imaging session (final concentration 10鈥壩糓). Time-lapse images were analyzed on Image J and Nb80-GFP membrane relocalization was confirmed by drawing a line across the cell membrane and using the plot profile function. The analysis of green and red fluorescent signal change (membrane 螖F/F0) was done using a custom-made script written in Matlab. Briefly, the green and red fluorescence intensities were measured in a mask generated using the signal from the red channel, indicating the membrane location of flag-尾2AR. Fluorescent signal change for each wavelength was calculated as (F-F0)/F0, with F0 being the fluorescence intensity prior to agonist stimulation.Fluorescent resonance energy transfer (FRET) measurementWild-type or 尾1/尾2AR dKO myocytes were infected with adenovirus overnight for expression of the cAMP biosensor ICUE324. dKO myocytes were further incubated with 尾2AR-NLPs or empty NLPs from 6 to 24鈥塰 before cells were used in FRET recording. Two adenoviruses expressing ICUE3 and flag-尾2AR were used to co-infect dKO cells overnight, for the cell-expressed 尾2AR data. Cells were imaged on a Zeiss Axiovert 200鈥塎 microscope with a 40脳/1.3NA oil-immersion objective lens and cooled CCD camera. Dual emission ratio imaging was acquired with a 420DF20 excitation filter, a 450DRLP dichroic mirror and two emission filters (475DF40 for cyan and 535DF25 for yellow). The acquisition was set with 200鈥塵s exposure in both channels. Images in both channels were subjected to background subtraction and ratios of yellow-to-cyan color were calculated at different time points.Scratch assay尾1/尾2AR dKO MEFs were plated onto 6-well plates in DMEM medium containing 10% FBS and incubated at 37鈥壜癈 to create a confluent monolayer. 尾2AR-NLPs or empty NLPs were added to the cells at 20鈥塶M concentration and incubated for 8鈥塰. The cell monolayer was scraped in a straight line with a pipet tip to create a gap. The debris was removed and the cells were washed once to smooth the edge of the scratch with 1鈥塵l of the growth medium, then incubated with 4鈥塵l of medium. Immediately following the scratch, isoproterenol or PBS vehicle was added to cells, with a final isoproterenol concentration of 10鈥壩糓. Images of the scratches were taken under a phase-contrast microscope. Additional pictures were taken at the 24- and 48-h time points. The images acquired at each time point were analyzed for percentage gap closure on ImageJ.Statistical analysisTo quantify GPCR insertion efficiency into NLPs from 尾2AR TEM images, a semi-automated custom-made particle-selecting tool was used. To quantify the particles based on diameter distribution, particle analysis tools from Image J were employed. The statistical distribution of NLPs revealed that both the mean and the mode for the size of GPCR-filled NLPs were 23.5鈥塶m, with about 27% of insertion efficiency. For quantification of 尾2AR insertion on the cell membrane, statistical analisis was performed using one-way ANOVA with Dunnett鈥檚 post-test, for comparison of every condition versus control condition (Empty-NLPs). During scratch assay, on the outer bottom of the dish an ultrafine ink marker was used to lay reference points close to the scratch to facilitate retrieving the same field for all image acquisitions. For TEM nanodisc size distribution statistical analysis was performed using paired t-test. All other statistical analyses were performed using unpaired t test. References1.Ghosh, E., Kumari, P., Jaiman, D. Shukla, A. K. Methodological advances: the unsung heroes of the GPCR structural revolution. Nat Rev Mol Cell Biol 16, 69鈥?1, https://doi.org/10.1038/nrm3933 (2015).CAS聽 Article聽 PubMed聽 Google Scholar聽 2.D鈥橝stolfo, D. S. et al. Efficient intracellular delivery of native proteins. Cell 161, 674鈥?90, https://doi.org/10.1016/j.cell.2015.03.028 (2015).Article聽 PubMed聽 Google Scholar聽 3.Denisov, I. G. Sligar, S. G. Nanodiscs for structural and functional studies of membrane proteins. Nat Struct Mol Biol 23, 481鈥?86, https://doi.org/10.1038/nsmb.3195 (2016).CAS聽 Article聽 PubMed聽 Google Scholar聽 4.Leitz, A. J., Bayburt, T. H., Barnakov, A. N., Springer, B. A. Sligar, S. G. Functional reconstitution of Beta2-adrenergic receptors utilizing self-assembling Nanodisc technology. Biotechniques 40, 601鈥?02, 604, 606, passim (2006).5.Fischer, N. O. et al. Evaluation of nanolipoprotein particles (NLPs) as an in vivo delivery platform. PLoS One 9, e93342, https://doi.org/10.1371/journal.pone.0093342 (2014).ADS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 6.Ghosh, M. Ryan, R. O. ApoE enhances nanodisk-mediated curcumin delivery to glioblastoma multiforme cells. Nanomedicine (Lond) 9, 763鈥?71, https://doi.org/10.2217/nnm.13.35 (2014).CAS聽 Article聽 Google Scholar聽 7.Numata, M. et al. Nanodiscs as a therapeutic delivery agent: inhibition of respiratory syncytial virus infection in the lung. Int J Nanomedicine 8, 1417鈥?427, https://doi.org/10.2147/IJN.S39888 (2013).PubMed聽 PubMed Central聽 Google Scholar聽 8.Kuai, R., Ochyl, L. J., Bahjat, K. S., Schwendeman, A. Moon, J. J. Designer vaccine nanodiscs for personalized cancer immunotherapy. Nat Mater 16, 489鈥?96, https://doi.org/10.1038/nmat4822 (2017).ADS聽 CAS聽 Article聽 PubMed聽 Google Scholar聽 9.Corin, K. et al. A robust and rapid method of producing soluble, stable and functional G-protein coupled receptors. PLoS One 6, e23036, https://doi.org/10.1371/journal.pone.0023036 (2011).ADS聽 CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 10.Gao, T. et al. Characterization of de novo synthesized GPCRs supported in nanolipoprotein discs. PLoS One 7, e44911, https://doi.org/10.1371/journal.pone.0044911 (2012).ADS聽 CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 11.Yang, J. P., Cirico, T., Katzen, F., Peterson, T. C. Kudlicki, W. Cell-free synthesis of a functional G protein-coupled receptor complexed with nanometer scale bilayer discs. BMC Biotechnol 11, 57, https://doi.org/10.1186/1472-6750-11-57 (2011).CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 12.Shilling, P. J., Bumbak, F., Scott, D. J., Bathgate, R. A. D. Gooley, P. R. Characterisation of a cell-free synthesised G-protein coupled receptor. Sci Rep 7, 1094, https://doi.org/10.1038/s41598-017-01227-z (2017).ADS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 13.Orban, E., Proverbio, D., Haberstock, S., Dotsch, V. Bernhard, F. Cell-free expression of G-protein-coupled receptors. Methods Mol Biol 1261, 171鈥?95, https://doi.org/10.1007/978-1-4939-2230-7_10 (2015).CAS聽 Article聽 PubMed聽 Google Scholar聽 14.Quast, R. B., Sonnabend, A., Stech, M., Wustenhagen, D. A. Kubick, S. High-yield cell-free synthesis of human EGFR by IRES-mediated protein translation in a continuous exchange cell-free reaction format. Sci Rep 6, 30399, https://doi.org/10.1038/srep30399 (2016).ADS聽 CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 15.Shinoda, T. et al. Cell-free methods to produce structurally intact mammalian membrane proteins. Sci Rep 6, 30442, https://doi.org/10.1038/srep30442 (2016).ADS聽 CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 16.Anderson, M. J., Stark, J. C., Hodgman, C. E. Jewett, M. C. Energizing eukaryotic cell-free protein synthesis with glucose metabolism. FEBS Lett 589, 1723鈥?727, https://doi.org/10.1016/j.febslet.2015.05.045 (2015).CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 17.Rasmussen, S. G. et al. Crystal structure of the beta2 adrenergic receptor-Gs protein complex. Nature 477, 549鈥?55, https://doi.org/10.1038/nature10361 (2011).ADS聽 CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 18.He, W. et al. Cell-free expression of functional receptor tyrosine kinases. Sci Rep 5, 12896, https://doi.org/10.1038/srep12896 (2015).ADS聽 CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 19.Jain, A., Liu, R., Xiang, Y. K. Ha, T. Single-molecule pull-down for studying protein interactions. Nat Protoc 7, 445鈥?52, https://doi.org/10.1038/nprot.2011.452 (2012).CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 20.Jain, A. et al. Probing cellular protein complexes using single-molecule pull-down. Nature 473, 484鈥?88, https://doi.org/10.1038/nature10016 (2011).ADS聽 CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 21.Irannejad, R. et al. Conformational biosensors reveal GPCR signalling from endosomes. Nature 495, 534鈥?38, https://doi.org/10.1038/nature12000 (2013).ADS聽 CAS聽 Article聽 PubMed聽 Google Scholar聽 22.Baker, J. G. The selectivity of beta-adrenoceptor agonists at human beta1-, beta2- and beta3-adrenoceptors. Br J Pharmacol 160, 1048鈥?061, https://doi.org/10.1111/j.1476-5381.2010.00754.x (2010).CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 23.Ulbrich, M. H. Isacoff, E. Y. Subunit counting in membrane-bound proteins. Nat Methods 4, 319鈥?21, https://doi.org/10.1038/nmeth1024 (2007).CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 24.DiPilato, L. M. Zhang, J. The role of membrane microdomains in shaping beta2-adrenergic receptor-mediated cAMP dynamics. Mol Biosyst 5, 832鈥?37, https://doi.org/10.1039/b823243a (2009).CAS聽 Article聽 PubMed聽 Google Scholar聽 25.Le Provost, G. S. Pullar, C. E. beta2-adrenoceptor activation modulates skin wound healing processes to reduce scarring. J Invest Dermatol 135, 279鈥?88, https://doi.org/10.1038/jid.2014.312 (2015).Article聽 PubMed聽 Google Scholar聽 26.Pullar, C. E. Isseroff, R. R. The beta 2-adrenergic receptor activates pro-migratory and pro-proliferative pathways in dermal fibroblasts via divergent mechanisms. J Cell Sci 119, 592鈥?02, https://doi.org/10.1242/jcs.02772 (2006).CAS聽 Article聽 PubMed聽 Google Scholar聽 27.Kalmbach, R. et al. Functional cell-free synthesis of a seven helix membrane protein: in situ insertion of bacteriorhodopsin into liposomes. J Mol Biol 371, 639鈥?48, https://doi.org/10.1016/j.jmb.2007.05.087 (2007).CAS聽 Article聽 PubMed聽 Google Scholar聽 28.Saarenpaa, T., Jaakola, V. P. Goldman, A. Baculovirus-mediated expression of GPCRs in insect cells. Methods Enzymol 556, 185鈥?18, https://doi.org/10.1016/bs.mie.2014.12.033 (2015).CAS聽 Article聽 PubMed聽 Google Scholar聽 29.Schutz, M. et al. Directed evolution of G protein-coupled receptors in yeast for higher functional production in eukaryotic expression hosts. Sci Rep 6, 21508, https://doi.org/10.1038/srep21508 (2016).ADS聽 CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 30.Bayburt, T. H. Sligar, S. G. Membrane protein assembly into Nanodiscs. FEBS Lett 584, 1721鈥?727, https://doi.org/10.1016/j.febslet.2009.10.024 (2010).CAS聽 Article聽 PubMed聽 Google Scholar聽 31.Lyukmanova, E. N. et al. Lipid-protein nanodiscs for cell-free production of integral membrane proteins in a soluble and folded state: comparison with detergent micelles, bicelles and liposomes. Biochim Biophys Acta 1818, 349鈥?58, https://doi.org/10.1016/j.bbamem.2011.10.020 (2012).CAS聽 Article聽 PubMed聽 Google Scholar聽 32.Gaub, B. M., Berry, M. H., Holt, A. E., Isacoff, E. Y. Flannery, J. G. Optogenetic Vision Restoration Using Rhodopsin for Enhanced Sensitivity. Mol Ther 23, 1562鈥?571, https://doi.org/10.1038/mt.2015.121 (2015).CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 33.Gaub, B. M. et al. Restoration of visual function by expression of a light-gated mammalian ion channel in retinal ganglion cells or ON-bipolar cells. Proc Natl Acad Sci USA 111, E5574鈥?583, https://doi.org/10.1073/pnas.1414162111 (2014).ADS聽 CAS聽 Article聽 PubMed聽 PubMed Central聽 Google Scholar聽 Download referencesAcknowledgementsThis work was supported by funding from NIH to L. T. (DP2MH107056), M.C. (R01CA155642), Y.K.X (NIH HL127764, NIH HL112413 and VA Merit 01BX002900), RHC (AI095382, EB021230, CA198880, National Institute of Food and Agriculture and Finland Distinguished Professor program) and A.S. (AHA postdoctoral fellowship). This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. We thank Gerard Joey Broussard from the Tian lab and Ruensern Tan from the Xiang lab at the University of California, Davis, for help with developing the MatLab scripts and the Alexa-546 anti-flag M1 antibody labeling, respectively. We would also like to thank Feliza Bourguet for technical assistance with lipid and expression screening.Author informationAuthor notesTommaso Patriarchi and Ao Shen contributed equally to this work.AffiliationsUniversity of California Davis, School of Medicine, Department of Biochemistry and Molecular Medicine, Davis, California, USATommaso Patriarchi聽 聽Lin TianUniversity of California Davis, School of Medicine, Department of Pharmacology, Davis, California, USAAo Shen聽 聽Yang K. XiangLawrence Livermore National Laboratory, Livermore, California, USAWei He聽 聽Matthew A. ColemanUniversity of California Davis, Department of Molecular and Cellular Biology, California, USAMo Baikoghli聽 聽R. Holland ChengVA Northern California Health care system, Mather, California, USAYang K. XiangUniversity of California Davis School of Medicine, Radiation Oncology, Sacramento, California, USAMatthew A. ColemanAuthorsTommaso PatriarchiView author publicationsYou can also search for this author in PubMed聽Google ScholarAo ShenView author publicationsYou can also search for this author in PubMed聽Google ScholarWei HeView author publicationsYou can also search for this author in PubMed聽Google ScholarMo BaikoghliView author publicationsYou can also search for this author in PubMed聽Google ScholarR. Holland ChengView author publicationsYou can also search for this author in PubMed聽Google ScholarYang K. XiangView author publicationsYou can also search for this author in PubMed聽Google ScholarMatthew A. ColemanView author publicationsYou can also search for this author in PubMed聽Google ScholarLin TianView author publicationsYou can also search for this author in PubMed聽Google ScholarContributionsT.P., A.S., L.T. and M.A.C. were involved in designing the research and discussion of the results; T.P., A.S., W.H., M.B. conducted the experiments; T.P., A.S., W.H., M.B. performed the analyses; Y.K.X., R.H.C., M.A.C. and L.T. provided critical scientific and technical expertise regarding protocols and reagents; T.P. and L.T. wrote the manuscript.Corresponding authorsCorrespondence to Matthew A. Coleman or Lin Tian.Ethics declarations Competing Interests The authors declare no competing interests. Additional informationPublisher\'s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Electronic supplementary material Supplementary InformationRights and permissions Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article鈥檚 Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article鈥檚 Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Reprints and PermissionsAbout this articleCite this articlePatriarchi, T., Shen, A., He, W. et al. Nanodelivery of a functional membrane receptor to manipulate cellular phenotype. Sci Rep 8, 3556 (2018). https://doi.org/10.1038/s41598-018-21863-3Download citationReceived: 18 September 2017Accepted: 12 February 2018Published: 23 February 2018DOI: https://doi.org/10.1038/s41598-018-21863-3 CommentsBy submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate. Sign up for the Nature Briefing newsletter 鈥?what matters in science, free to your inbox daily.