Conversion of furostanol saponins into spirostanol saponins improves the yield of diosgenin from Dioscorea zingiberensis by acid hydrolysis

Xu Pang ab, Hong Zhi Huanga, Yang Zhaoa, Cheng-Qi Xionga, Li Yan Yub and Bai-Ping Ma*a
aDepartment of Biotechnology, Beijing Institute of Radiation Medicine, Beijing 100850, China. E-mail: mabaiping@sina.com; Tel: +86 10 68210077 ext. 930265
bInstitute of Medicinal Biotechnology, Academy of Medical Science & Peking Union Medical College, Beijing 100050, China

Received 19th October 2014 , Accepted 28th November 2014

First published on 28th November 2014


Abstract

Current production of diosgenin mainly depends on the acid hydrolysis of steroidal saponins from Dioscorea plants, and, in China especially, Dioscorea zingiberensis C. H. Wright (DZW) is used. The experimental results we obtained demonstrated that furostanol saponins, as the main constituents in DZW, were prone to generate 25-spirosta-3,5-diene as side product during the acid hydrolysis process, while spirostanol saponins hardly generated 25-spirosta-3,5-diene. This 25-spirosta-3,5-diene was the key reason leading to the low yield of diosgenin from DZW by acid hydrolysis. Effective conversion of furostanol saponins into spirostanol saponins can avoid the generation of 25-spirosta-3,5-diene so as to increase the yield of diosgenin, suggesting the importance of this preliminary conversion for improving the yield of diosgenin from DZW by acid hydrolysis. The conversion of furostanol saponins into spirostanol saponins can be performed by either enzymatic hydrolysis or spontaneous fermentation, whereas enzymatic hydrolysis is more controllable compared with spontaneous fermentation.


1. Introduction

Diosgenin is an important steroidal sapogenin, not only because it is used as a starting material for the synthesis of steroidal hormone drugs, but also due to its wide range of biochemical and pharmacological activities, including reducing the plasma level of cholesterol,1 anti-skin-aging,2 anti-inflammatory,3 antithrombosis,4 anticancer5 activity and suppressing acute lung injury.6 At present, diosgenin worldwide is mainly provided by China and Mexico. In China, Dioscorea zingiberensis C. H. Wright (DZW), which is a particular, resourceful cultivated plant especially in the areas of Shiyan and Enshi in Hubei province and Ankang in Shaanxi province, is always used as feedstock for the production of diosgenin. The conventional method for the production of diosgenin is acid hydrolysis of steroidal saponins from Dioscorea plants, generally using sulfuric acid.7 However, the acid treatment inevitably leads to a severe environmental pollution problem.8,9 Efforts in the application of biotechnology, instead of acid treatment, to the production of diosgenin have been reported,10–13 while no method without acid treatment has been industrialized so far and diosgenin is still produced by acid hydrolysis. Nevertheless, if the key factors to improve the yield of diosgenin based on the present production process could be determined, environmental pollution could be mitigated.

In this study, it was found that the yield of diosgenin from DZW by direct acid hydrolysis was very low, and much side product was observed in the acid hydrolyzate. The side product, identified as 25-spirosta-3,5-diene, was a steroid derivative supposed to derive from the starting steroidal saponins in DZW. This motivated us, and finally we clearly proved that furostanol saponins, with a high content level in DZW, were prone to generate 25-spirosta-3,5-diene in acid hydrolysis, while spirostanol saponins hardly generated it. Therefore, it was considered that 25-spirosta-3,5-diene mainly derived from furostanol saponins during the acid hydrolysis reaction might be the key reason leading to the low yield of diosgenin, and effective conversion of furostanol saponins into spirostanol saponins can improve the yield of diosgenin from DZW. Consequently, by pretreatment with β-glucosidase, the furostanol saponins of DZW were totally converted into spirostanol saponins, and the yield of diosgenin from DZW by acid hydrolysis was notably enhanced, suggesting the importance of this preliminary conversion for improving the yield of diosgenin from DZW by acid hydrolysis.

2. Results and discussion

2.1. 25-Spirosta-3,5-diene as side product leading to low yield of diosgenin by direct acid hydrolysis

The conventional method for diosgenin production is direct acid hydrolysis of steroidal saponins from Dioscorea plants. According to the conventional method, the yield of diosgenin from DZW (heat-dried DZW, the same as follows) by direct acid hydrolysis (3 mol L−1 sulfuric acid at 95 °C for 5 hours) was determined to be 1.26% in our laboratory, which is very low. In order to know the reason why the yield of diosgenin from DZW by conventional acid hydrolysis was so low, the hydrolyzate was subjected to HPLC-ELSD analysis. As Fig. 1A shows, diosgenin was not the only product; moreover, much side product identified as 25-spirosta-3,5-diene existed in the hydrolyzate and the ratio of 25-spirosta-3,5-diene to diosgenin obtained according to their HPLC-ELSD peak areas was 1.05[thin space (1/6-em)]:[thin space (1/6-em)]1, suggesting that 25-spirosta-3,5-diene generated in the acid hydrolysis process might be the key reason leading to the low yield of diosgenin.
image file: c4ra12709a-f1.tif
Fig. 1 HPLC-ELSD profiles of acid hydrolyzates from DZW (A) and β-glucosidase-treated DZW (B). (The DZW samples were obtained by heating fresh DZW at 80 °C.)

2.2. 25-Spirosta-3,5-diene derives from starting furostanol saponins during acid hydrolysis

Six typical compounds (the structures are drawn in Fig. 2) in DZW, including two furostanol saponins, namely, parvifloside and deltoside, three spirostanol saponins, namely, zingiberensis newsaponin, deltonin and prosapogenin A of dioscin, and the sapogenin diosgenin were subjected to acid hydrolysis reactions under typical conditions, namely, 3 mol L−1 sulfuric acid at 95 °C for 5 hours, before their hydrolyzates were analyzed by HPLC-ELSD. Based on the HPLC-ELSD profiles of acid hydrolyzates from the six compounds above, it was found that a large amount of 25-spirosta-3,5-diene was generated from parvifloside and deltoside (Fig. 3A and B) and zingiberensis newsaponin, deltonin and prosapogenin A of dioscin were mostly converted to diosgenin with little 25-spirosta-3,5-diene (Fig. 3C–E). In addition, it was found that diosgenin was not converted to 25-spirosta-3,5-diene under these reaction conditions (Fig. 3F), which is in contrast to a previous report stating that 25-spirosta-3,5-diene was derived from diosgenin by dehydration14. The ratios of 25-spirosta-3,5-diene to diosgenin produced from the six compounds are given in Table 1.
image file: c4ra12709a-f2.tif
Fig. 2 Structures of diosgenin (1), 25-spirosta-3,5-diene (2), parvifloside (3), deltoside (4), zingiberensis newsaponin (5), deltonin (6) and prosapogenin A of dioscin (7).

image file: c4ra12709a-f3.tif
Fig. 3 HPLC-ELSD chemical profiles of hydrolyzates from parvifloside (A), deltoside (B), zingiberensis newsaponin (C), deltonin (D), prosapogenin A of dioscin (E) and diosgenin (F) by acid hydrolysis in 3 mol L−1 sulfuric acid.
Table 1 Ratios of 25-spirosta-3,5-diene to diosgenin obtained by HPLC-ELSD from parvifloside, deltoside, zingiberensis newsaponin, deltonin, prosapogenin A of dioscin and diosgenin by acid hydrolysis with 3 mol L−1 sulfuric acid
Compound Peak area ratio (%) (25-spirosta-3,5-diene:diosgenin)
Parvifloside 127.77
Deltoside 115.05
Zingiberensis newsaponin 3.63
Deltonin 1.83
Prosapogenin A of dioscin 0.40
Diosgenin 0


Furthermore, the effect of acid concentration on the yield of 25-spirosta-3,5-diene was studied based on HPLC-ELSD analysis of acid hydrolyzates from parvifloside and deltoside under different acid concentrations (Fig. S1). A portion of parvifloside and deltoside were transformed into 25-spirosta-3,5-diene even with 0.5 mol L−1 sulfuric acid, suggesting that the conversion of furostanol saponins into 25-spirosta-3,5-diene during acid hydrolysis is inevitable. The yields of 25-spirosta-3,5-diene from parvifloside and deltoside in 0.5, 1 and 2 mol L−1 sulfuric acid solutions were lower than that in 3 mol L−1 sulfuric acid. However, our previous study has proved that sulfuric acid in concentrations less than 2.5 mol L−1 converted total saponins into diosgenin to a very limited extent (data not shown). That was the reason why 3 mol L−1 sulfuric acid was finally used in this work. The results clearly indicated that 25-spirosta-3,5-diene mainly derives from furostanol saponins during the acid hydrolysis process.

2.3. Proposed reaction pathways of furostanol and spirostanol saponins during acid hydrolysis

Fig. 4 presents the possible reaction pathways of furostanol and spirostanol saponins during acid hydrolysis. With furostanol saponins, after protonation of the oxygen atoms at C-3 and C-26, a portion of intermediate A generated intermediate B by cleavage between the sugar moiety and oxygen atom at C-3, and then intermediate B generated diosgenin by a dehydration reaction. Moreover, by cleavage between the oxygen atom and carbon atom at C-3, another portion of intermediate A generated intermediate C, and then intermediate D was derived from intermediate C by charge transfer. Furthermore, intermediate D lost the hydrogen atom at C-3 and afforded intermediate E, and finally intermediate E yielded 25-spirosta-3,5-diene by a dehydration reaction. However, with spirostanol saponins, after protonation of the oxygen atom at C-3, almost all intermediate F was converted into diosgenin by cleavage between the oxygen atom and carbon atom at C-3.
image file: c4ra12709a-f4.tif
Fig. 4 Main proposed reaction pathways of furostanol saponins and spirostanol saponins during acid hydrolysis.

2.4. Conversion of furostanol saponins into spirostanol saponins can avoid 25-spirosta-3,5-diene and increase the yield of diosgenin

It has thus been proved that 25-spirosta-3,5-diene mainly derives from furostanol saponins during acid hydrolysis. Therefore, effective conversion of furostanol saponins into spirostanol saponins before acid hydrolysis is important to obtain a high yield of diosgenin from DZW. Enzymatic treatment is an effective and controllable method for bioconversion, and β-glucosidase is most efficient at hydrolyzing furostanol saponins into spirostanol saponins. Accordingly, β-glucosidase isolated from Aspergillus flavus in our lab was used to treat DZW. HPLC-ELSD profiles of 70% ethanol extract from DZW and β-glucosidase-treated DZW are shown in Fig. 5. Parvifloside and deltoside were the main constituents of DZW (Fig. 5A), and in β-glucosidase-treated DZW, both of these were converted into prosapogenin A of dioscin (Fig. 5B). The yield of diosgenin after acid hydrolysis of β-glucosidase-pretreated DZW was enhanced to 2.82%, far more than that by direct acid hydrolysis. Moreover, an HPLC-ELSD profile of the acid hydrolyzate from β-glucosidase-pretreated DZW showed that, in addition to diosgenin, almost no 25-spirosta-3,5-diene was observed (Fig. 1B), clearly proving that 25-spirosta-3,5-diene was indeed the key factor leading to the low yield of diosgenin.
image file: c4ra12709a-f5.tif
Fig. 5 HPLC-ELSD profiles of DZW (A) and β-glucosidase-treated DZW (B) (The DZW samples were obtained by heating fresh DZW at 80 °C.).

2.5. Conversion of furostanol saponins into spirostanol saponins by spontaneous fermentation

The literature has reported that pretreatment of DZW by spontaneous fermentation before acid hydrolysis can improve the yield of diosgenin.15,16 In Ankang, Shaanxi province, China, the method of spontaneous fermentation was actually applied in a diosgenin production factory. A previous study reported that furostanol saponins can be hydrolyzed to the corresponding spirostanol saponins by endogenous glucosidase obtained from the same plant,17 which suggested that, by spontaneous fermentation, the furostanol saponins of DZW can be converted into spirostanol saponins by endogenous glucosidase, leading to an improved yield of diosgenin.

According to the common method in the literature, a heat-dried DZW sample was subjected to a spontaneous fermentation experiment and the yield of diosgenin by acid hydrolysis was finally determined to be 1.40% (for experimental details see ESI), so there was no notable improvement compared with the yield of 1.26% by direct acid hydrolysis. HPLC-ELSD analysis showed that parvifloside and deltoside were hardly converted into spirostanol saponins by spontaneous fermentation (Fig. S2) and in the acid hydrolyzate much 25-spirosta-3,5-diene still existed (Fig. S3). Generally, fresh DZW is always dried in air or by heating after collection. In order to better understand the conversion efficiency of spontaneous fermentation on DZWs under different conditions, the chemical constituents of fresh DZW, air-dried DZW and heat-dried DZW, as well as their spontaneously fermented residues, were compared by HPLC-ELSD (for experimental details see ESI). As shown in Fig. 6, among the three kinds of DZW, the main constituents were furostanol saponins, namely, parvifloside and deltoside. After spontaneous fermentation, the parvifloside and deltoside in fresh DZW were totally converted into the corresponding spirostanol saponins, namely zingiberensis newsaponin and deltonin (Fig. 7A), while little zingiberensis newsaponin and deltonin were detected in hydrolyzed air-dried and heat-dried DZWs (Fig. 7B and C), suggesting that the endogenous glucosidase in air-dried and heat-dried DZWs might be inactivated during the drying process, so that the furostanol saponins cannot be effectively converted. These results indicated that conversion of DZW furostanol saponins into spirostanol saponins could be conducted by spontaneous fermentation, but the conversion efficiency was influenced by the activity of endogenous glucosidase in the raw material.


image file: c4ra12709a-f6.tif
Fig. 6 HPLC-ELSD profiles of fresh DZW (A), air-dried DZW (B) and heat-dried DZW (C).

image file: c4ra12709a-f7.tif
Fig. 7 HPLC-ELSD profiles of hydrolyzed fresh DZW (A), hydrolyzed air-dried DZW (B) and hydrolyzed heat-dried DZW (C).

3. Experimental

3.1. Materials

Fresh tubers of Dioscorea zingiberensis C. H. Wright were collected from Ankang, Shaanxi province, China. Part of the fresh DZW tubers were stored at −20 °C to keep fresh (fresh DZW), and the others were dried in an oven at 80 °C (heat-dried DZW) and dried in air at room temperature (air-dried DZW) to a constant weight, respectively. β-Glucosidase was isolated from Aspergillus flavus in our lab previously.18 Pure diosgenin, 25-spirosta-3,5-diene, parvifloside, deltoside, zingiberensis newsaponin, deltonin, and prosapogenin A of dioscin were all obtained from DZW tubers and their acid hydrolyzates in our laboratory (purity > 95% detected by HPLC-ELSD), and identified by NMR and MS experiments (Fig. 2). All chromatographic grade solvents used were purchased from Fisher Scientific Co. Ltd, and solvents of analytical grades were all purchased from a Beijing chemical plant.

3.2. General procedures

Acid hydrolysis of DZW was carried out by depositing a sample in a flask with 3 mol L−1 sulfuric acid and keeping it at 95 °C for 5 hours under reflux conditions. Acid hydrolysis of compounds was performed by keeping each compound in an airtight test tube with sulfuric acid at 95 °C for 5 hours. Enzymatic hydrolysis of DZW was performed by depositing raw DZW mixed with water and β-glucosidase in an airtight triangular flask and incubating at 35 °C for 30 hours. Spontaneous fermentation was carried out by incubating raw DZW mixed with water at 35 °C for 30 hours. All chemical analysis was performed by HPLC-ELSD which was carried out on a Waters 2695 system equipped with a PL-ELS 2000 evaporative light-scattering detector (temp: 90 °C, gas: 1.6 L min−1) and Phecda C18 column (4.6 mm i.d. × 250 mm, 5 μm, ODS).

3.3. Determination of the yield of diosgenin from DZW and chemical characterization of the acid hydrolyzate

5 g heat-dried DZW was mashed and then mixed with 3 mol L−1 sulfuric acid (100 mL), followed by heating at 95 °C for 5 hours under reflux. After neutralizing with NaOH and filtration, the residue (together with filter paper) was dried at 80 °C and extracted with ethyl acetate at 85 °C under reflux three times (100 mL, 70 mL, 50 mL, 1 h each time). When the ethyl acetate was recovered, the residue was dissolved in 50 mL methanol and subjected to HPLC-ELSD analysis with a mobile phase of acetonitrile–water (94[thin space (1/6-em)]:[thin space (1/6-em)]6, v/v) to determine the yield of diosgenin. Moreover, the hydrolysate was analyzed by HPLC-ELSD with the gradient mobile phase consisting of methanol–water (0.00–14.00–14.01–35.00 min, 97–97–100–100% menthol).

3.4. Acid hydrolysis of typical compounds in DZW and chemical characterization

Parvifloside, deltoside, zingiberensis newsaponin, deltonin, prosapogenin A of dioscin and diosgenin (each 5 mg) were put into an airtight test tube with 3 mL of 3 mol L−1 sulfuric acid and heated at 95 °C for 5 hours, respectively. After neutralizing with NaOH, each hydrolyzate was extracted with ethyl acetate (3 mL). When the ethyl acetate layer was evaporated completely, the residues were dissolved with methanol and subjected to HPLC-ELSD analysis with the gradient mobile phase consisting of methanol–water (0.00–14.00–14.01–35.00 min, 97–97–100–100% menthol). Furthermore, in order to study the effect of acid concentration on the yields of diosgenin and 25-spirosta-3,5-diene from furostanol saponins of DZW, parvifloside and deltoside were hydrolyzed in sulfuric acid at concentrations of 0.5 mol L−1, 1.0 mol L−1, and 2.0 mol L−1 sulfuric acid. 15 mg parvifloside was dissolved in 5 mL of 0.5 mol L−1, 1 mol L−1, 2 mol L−1, and 3 mol L−1 sulfuric acid, respectively, and then heated at 95 °C for 5 hours. After neutralizing with NaOH, the hydrolyzates were extracted with ethyl acetate (3 mL). After the ethyl acetate was evaporated completely, each residue was dissolved in methanol for HPLC analysis. Likewise, deltoside was treated using the same conditions as parvifloside. The chemical characterization of acid hydrolyzates from the compounds above was performed using HPLC-ELSD with the gradient mobile phase consisting of methanol–water: (0.00–14.00–14.01–35.00 min, 97–97–100–100% menthol).

3.5. Determination of the yield of diosgenin from β-glucosidase-treated DZW and chemical characterization of the acid hydrolyzate

First, 5 g mashed heat-dried DZW mixed with 30 mL water and 89 mg β-glucosidase (based on the enzyme protein amount, the same as follows) was incubated at 35 °C for 30 hours to obtain a pretreated DZW sample. Then, the treated DZW was added to 70 mL of 3 mol L−1 sulfuric acid and an additional 6 mL of 18 mol L−1 sulfuric acid (finally 3 mol L−1 sulfuric acid) and then heated under reflux at 95 °C for 5 hours. After neutralization with NaOH and filtration, the residue (together with filter paper) was dried at 80 °C and extracted with ethyl acetate at 85 °C under reflux three times (100 mL, 70 mL, 50 mL, 1 h each time). When the ethyl acetate was recovered, the residue was dissolved in 50 mL methanol and subjected to HPLC-ELSD with a mobile phase of acetonitrile–water (94[thin space (1/6-em)]:[thin space (1/6-em)]6, v/v). Chemical characterization of the acid hydrolyzate from β-glucosidase-treated DZW was performed using HPLC-ELSD with the gradient mobile phase consisting of methanol–water (0.00–14.00–14.01–35.00 min, 97–97–100–100% menthol).

3.6. Chemical characterization of DZW and β-glucosidase-treated DZW

Heat-dried DZW and β-glucosidase-treated DZW (for sample preparation refer to the above section) were extracted with 70% EtOH under reflux at 95 °C for 1 hour, respectively, and then the extracts were subjected to HPLC-ELSD analysis. An acetonitrile–water solution (gradient, 0.00–17.00–19.00–20.00–25.00–25.01–35.00–40.00 min, 27–27–35–45–45–66–66–90% acetonitrile) was used as the mobile phase.

3.7. Chemical characterization of fresh, air-dried and heat-dried DZWs, together with their spontaneously fermented residues

Fresh, air-dried and heat-dried DZWs were extracted with 70% EtOH at 95 °C for 1 hour, respectively. After filtration, the extracts were subjected to HPLC-ELSD analysis directly. For chemical characterization of the spontaneously fermented residues, the pounded fresh DZW and the crushed air-dried and heat-dried DZWs were mixed with 30 mL water and incubated at 35 °C for 30 hours, respectively. Then each fermented residue was extracted with 70% EtOH at 95 °C for 1 hour. After filtration, the extracts were also subjected to HPLC analysis. HPLC analysis was performed using HPLC-ELSD with the gradient mobile phase consisting of acetonitrile–water (0.00–17.00–19.00–20.00–25.00–25.01–35.00–40.00 min, 27–27–35–45–45–66–66–90% acetonitrile).

4. Conclusion

For the first time, we demonstrated the importance of the preliminary conversion of furostanol saponins into spirostanol saponins for improving the yield of diosgenin from DZW by acid hydrolysis. Effective conversion of furostanol saponins into spirostanol saponins can avoid the generation of 25-spirosta-3,5-diene so as to increase the yield of diosgenin. Conversion of DZW furostanol saponins into spirostanol saponins can be conducted by enzymatic (such as β-glucosidase) hydrolysis or spontaneous fermentation. Spontaneous fermentation is a practical and economic conversion method to produce diosgenin, while its conversion efficiency is not highly controllable due to variable factors of endogenous glucosidase in raw DZWs. Enzymatic hydrolysis, when compared with spontaneous fermentation, is more effective and controllable, so more suitable for the industrialization of diosgenin production.

Notes and references

  1. H. Chen, C. Wang and C. T. Chang, Nutrition, 2003, 19, 646–651 CrossRef.
  2. Y. Tada, N. Kanda, A. Haratake, M. Tobiishi, H. Uchiwa and S. Watanabe, Steroids, 2009, 74, 504–511 CrossRef CAS PubMed.
  3. C. H. Huang, C. Y. Ku and T. R. Jan, Planta Med., 2009, 75, 1300–1305 CrossRef CAS PubMed.
  4. G. H. Gong, Y. Qin and W. Huang, Phytomedicine, 2010, 18, 458–463 CrossRef PubMed.
  5. S. Rajput and M. Mandal, Eur. J. Cancer Prev., 2012, 21, 205–215 CrossRef CAS PubMed.
  6. M. Gao, L. Chen, H. Yu, Q. Sun, J. Kou and B. Yu, Int. Immunopharmacol., 2013, 15, 240–245 CrossRef CAS PubMed.
  7. J. W. Rothrock, P. A. Hammes and W. J. Mcaleer, Ind. Eng. Chem., 1957, 49, 186–188 CrossRef CAS.
  8. H. Z. Zhao, P. Cheng, B. Zhao and J. R. Ni, Process Biochem., 2008, 43, 1427–1431 CrossRef CAS PubMed.
  9. P. Cheng, H. Z. Zhao, B. Zhao and J. R. Ni, Bioresour. Technol., 2009, 100, 2918–2925 CrossRef CAS PubMed.
  10. L. Liu, Y. S. Dong, S. S. Qi, H. Wang and Z. L. Xiu, Appl. Microbiol. Biotechnol., 2010, 85, 933–940 CrossRef CAS PubMed.
  11. W. Liu, W. Huang, W. Sun, Y. Zhu and J. Ni, World J. Microbiol. Biotechnol., 2010, 26, 1171–1180 CrossRef CAS PubMed.
  12. J. Lei, H. Niu, T. Li and W. Huang, World J. Microbiol. Biotechnol., 2012, 28, 1309–1314 CrossRef CAS PubMed.
  13. M. Wei, Y. Bai, M. Ao, W. Jin, P. Yu, M. Zhu and L. Yu, Bioresour. Technol., 2013, 146, 549–555 CrossRef CAS PubMed.
  14. Y. E. Peng, Z. H. Yang, Y. X. Wang, Z. Y. Liu, J. G. Bao and Y. Hong, Chem. Eng. Res. Des., 2011, 89, 2620–2625 CrossRef CAS PubMed.
  15. X. L. Tang, Z. L. Xu, B. Xia and Y. Y. Shi, J. Plant Resour. Environ., 2004, 13, 35–37 CAS.
  16. Z. J. Qiao and X. M. Li, Acta Bot. Boreali-Occident. Sin., 2006, 15, 212–215 Search PubMed.
  17. K. Inoue and Y. Ebizuka, FEBS Lett., 1996, 378, 157–160 CrossRef CAS.
  18. H. Z Huang, M. Zhao, L. Lu, D. W. Tan, W. B. Zhou, C. Q. Xiong, Y. Zhao, X. B. Song, L. Y. Yu and B. P. Ma, J. Mol. Catal. B: Enzym., 2013, 98, 1–7 CrossRef PubMed.

Footnotes

Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra12709a
These authors contributed equally to this work.

This journal is © The Royal Society of Chemistry 2015
Click here to see how this site uses Cookies. View our privacy policy here.