Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Angew Chem Int Ed Engl. Author manuscript; available in PMC 2010 May 20.
Published in final edited form as:
PMCID: PMC2873849

Symmetry-guided Design and Fluorous Synthesis of A Stable and Rapidly Excreted Imaging Tracer for 19F MRI**

1H and 19F are the most sensitive nuclei for nuclear magnetic resonance imaging (MRI), with the 1H signal suited for collecting information of the body[1,2] and the 19F signal suited for collecting information of drugs in the body.[3,4] Although 19F MRI[5] is only four years younger than 1H MRI,[6] it is not in clinical use. The progress of 19F MRI has been stalled by lack of suitable imaging agents. Current 19F imaging agents are perfluorocarbon (PFC) emulsions[713] and suffer severe shortcomings, including heterogeneity, instability, split 19F signal, complex formulation procedure and, most importantly, excessive organ retention for months or longer.[8,14] We developed a bi-spherical fluorocarbon molecule, denoted as 19FIT to stand for 19F imaging tracer, which overcame all the major deficiencies of PFC-based imaging agents. 19FIT, designed using the principle of modular spherical symmetry, is water-soluble and emits a single 19F signal from 27 fluorine atoms. The in vivo residence half-life of 19FIT measured in mice is about 0.5 day, and no evidence of organ retention or in vivo degradation was found. Our result shows that modular symmetry is a useful strategy for designing molecules with multi-functionalities. With suitable imaging agents like 19FIT, 19F MRI has the potential to play an important role in drug therapy, analogous to the role played by 1H MRI in disease diagnosis.

MRI has made significant contribution to medical diagnosis.[15] The value of MRI comes from its ability to collect in vivo information non-invasively without ionizing radiation.[16] 1H and 19F are respectively the most and second most sensitive stable nuclei for MRI. Due to the ubiquitous presence of the 1H signal and the complete absence of the 19F signal in the human body, 1H and 19F MRI complement each other in their information content. 1H MRI is suited for collecting information of the body (anatomy, physiology and biochemistry) and therefore is a valuable tool for disease diagnosis. 19F MRI, on the other hand, is a tracer-type technology and is suited for collecting information of drugs in the body (where, how much and in what form), and therefore has the potential to become a valuable tool for image-guided drug therapy.

Direct monitoring of drugs by 19F MRI requires the drug to be labeled by appropriate 19F imaging agents. For more than three decades, PFC emulsions have been used as 19F imaging agents.[5,713] However, PFC emulsions suffer severe drawbacks as imaging agents. First and foremost, PFCs are very lipophilic and accumulate excessively in internal organs (liver, spleen, lung) for months or longer.[8,14] Second, emulsion droplets are inherently heterogeneous, unstable, and are likely to disintegrate inside the body. In fact, the in vivo integrity of PFC droplets is difficult, if not impossible, to verify. Third, linear PFCs emit multiple 19F signals due to lack of symmetry. Signal splitting lowers signal intensity and can cause image artifacts. Macrocyclic PFCs (e.g., perfluoro-15-crown-5 ether, or PF15C5) will also emit multiple 19F signals if convalently modified because their cyclic symmetry will then be broken. Finally, the formulation of PFC emulsions is quite complex, requiring multiple surfactants and microfluidic devices.[1113] Complex formulation brings a range of difficulties for industrial production and regulatory approval. Indeed, there is currently only one injectable emulsion, Diprivan®, in the US market.[17]

All the aforementioned shortcomings of PFC-based imaging agents can be avoided if a hydrophilic stable molecule that emits a single 19F signal from multiple fluorine atoms can be developed. To achieve this goal, we employed a molecular design principle called modular spherical symmetry. A molecule with modular spherical symmetry comprises independent spherical cones joined by covalent bonds in the cone part. The fluorocarbon molecule, 19FIT, contains an F-spherical cone and an H-spherical cone (Figure 1). Spherical symmetry in the F-spherical cone ensures a single 19F signal from multiple fluorine atoms. Spherical symmetry in the H-spherical cone allows efficient incorporation of multiple hydrophilic groups. The advantage of spherical cones over macrocycles is that a spherical cone can be modified at the cone without breaking the symmetry of the sphere whereas a macrocycle will lose its cyclic symmetry upon modification. 19FIT comprises three building blocks, I, II and III, with I being the 19F signal emitter and II and III being aqueous-solubility enhancers (Figure 1). 19FIT was synthesized in a sequential manner that involves repetitive deprotection/condensation cycles (Scheme 1), analogous to solid-phase peptide synthesis. The synthesis proceeded from the F-sphere to the H-sphere so that all synthesis intermediates (2, I, 3, 4, 5, 6 and 7) contain nine -CF3 groups and thereby can be purified using the unique separation power of fluorous chemistry.[18,19] The final product 19FIT was purified using a combination of fluorous silica-gel chromatography and preparative HPLC (Figure S1).

Figure 1
Chemical structures of 19FIT and its three building blocks (I, II and III). 19FIT is designed to be a bi-spherical cone. The F-sphere contains 27 fluorine atoms for 19F signal generation. The H-sphere, currently made of 4 -OH groups, can be derivatized ...
Scheme 1
Synthesis of 19FIT. Reaction Conditions: (a) KH, BrCH2CO2tBu, THF, rt, 12 h; (b) TFA, anisol, CH2Cl2, rt, 2 h; (c) DIC, HOBt, DMF/THF (1/1), HN(CH2CO2tBu)2 (= II-(tBu)2), rt, 12 h; (c′) DIC, HOBt, DMF/THF (1/1), H2N(CH2CH2O)4Bn (= III-Bn), rt, ...

19FIT is soluble in phosphate buffered saline (PBS, Figure S2) and emits a single 19F signal from 27 fluorine atoms (Figure 2a). 19F NMR study shows that 19FIT forms micelle in PBS with a critical micelle concentration ca. 7 mM (Figure S3). Phantom experiments show that the 19F signal intensity is indeed proportional to 19F concentration (Figures 2b and 2c). The 19F longitudinal relaxation time, T1, of 19FIT is much shorter than that of PF15C5 measured under the same condition (163 ms vs. 1069 ms). Even with 30 mol% of relaxation enhancing Gd3+, the 19F T1 of PF15C5 reported in the literature[11] is still significantly longer than that of 19FIT. Shorter T1 brings another advantage of 19FIT over PFCs as it reduces data collection time and thereby increases signal intensity.[20,21]

Figure 2
a) 19F NMR spectrum of 19FIT in PBS. TFA is the internal 19F standard. b) 1H and 19F MRI images of phantoms (1–8) filled with 19FIT in PBS. 19F concentrations in the phantoms are: 4,050 mM, 2,025 mM, 1,012 mM, 506 mM, 253 mM, 126 mM, 63 mM, 32 ...

150 mM 19FIT PBS solution was administered to 16-week old BALB/c male mice at two dose levels: either 400 µL/mouse (ca. 60 mmol/kg 19F), or 200 µL/mouse (ca. 30 mmol/kg 19F). At both dose levels, the 19F signal decreased rapidly and became invisible in all organs except the bladder after 1–2 h (Figure 3 and Figure S4). The 19F signal intensity also decreased rapidly in whole-body 19F spectra and in urine samples collected from mice injected with 19FIT (Figure 4a and Figure S5). Based on whole-body and urine 19F signal intensity decays, the in vivo residence half-life, t½, of 19FIT is estimated to be ca. 0.5 day. In comparison, the in vivo residence half-lives of perfluorocarbons, also determined by 19F MRI, are months or longer.[8]

Figure 3
Superimposed 1H (white) and 19F (red) images (coronal view) of mice 1 (a) and 3 (b). Mice 1 and 3 were injected with 400 µL (60 mmol/kg 19F) and 200 uL (30 mmol/kg 19F) of 150 mM 19FIT PBS solution, respectively. For images beyond 2 h, see Figure ...
Figure 4
a) 19F signal intensity decay with time. Black symbols: whole-body 19F signals from mice 1 and 2 collected using a 3.0 T clinical MR scanner. Red symbols: mouse 6 urine 19F signal collected using an 11.7 T NMR spectrometer. For each series, the first ...

Whole-body 19F spectra in different mice showed only one 19F peak at all time points (Figure 4b and Figure S6). There is also only one 19F peak in all urine samples, with the same chemical shift as 19FIT (Figure S7). Comparison of HPLC profiles of urine samples collected before and after the injection of 19FIT showed only one peak contributable to fluorinated compounds (Figures 4c and 4d). The mass of this peak is 1,894.2 Da, the same as intact 19FIT. All these results are consistent with no in vivo degradation of 19FIT. Mice injected with 19FIT were observed for up to 45 days and showed no sign of acute toxicity or weight loss.

In summary, 19FIT overcomes major deficiencies of PFC-based 19F imaging agents, including heterogeneity, instability, split 19F signals, long 19F T1, complex formulation, and, most importantly, excessive organ retention. With suitable imaging agents, 19F MRI has the potential to play an important role in drug therapy, analogous to the role played by 1H MRI in disease diagnosis. From a chemistry standpoint, bi-spherical symmetry is a step forward from the unispherical symmetry employed in conventional dendrimer design.[22] In fact, modular symmetry is a general molecular design principle that can be extended to tri-spherical symmetry and beyond.

The application of 19FIT to drug or cell monitoring via 19F MRI still faces stiff challenges. One issue is sensitivity. The approach of using symmetry to generate a single 19F signal from multiple fluorine atoms can only go so far, on the order of 100 fluorine atoms. Other approaches, such as using paramagnetic ions to reduce the T1 and T2 of 19F,[11, 20, 21] are needed. Another issue is a dilemma common to all labeling tags. If the labeling is non-covalent, then the tag might dissociate from drugs or cells and signals from free and bound tags can hardly be distinguished. However, if the labeling is covalent, then the tag might alter the bioactivity of drugs or cells. One possible solution to this dilemma is to use the prodrug approach, i.e., making 19FIT-drug into a prodrug that replaces the free drug as the therapeutic agent. The pharmacologically inactive 19FIT-drug will be converted into the active free drug at the pathological site (e.g., tumor). The pharmacokinetics of 19FIT-drug can be monitored with 19F MRI right up to the point where 19FIT-drug is converted into the free drug at the disease site, with the 19FIT-drug →19FIT + drug conversion process monitored by 19F MRS. We are currently pursing these approaches.

Experimental Section


The synthesis of 19FIT involves the following steps:

An external file that holds a picture, illustration, etc.
Object name is nihms199651f6.jpg

where a is ether formation, b are b′ are deprotections, c and c′ are condensations.

tert-butyl mono-ester 2 (procedure a)

A suspension of potassium hydride (30%, 3.2 g, 24.0 mmol) was added slowly to a stirred solution of alcohol 1[23] (15.8 g, 20.0 mmol) in tetrahydrofuran (200 mL) at 0 °C. After 10 min, tert-butyl bromoacetate (5.9 mL, 7.8 g, 40.0 mmol) was added to the suspension in one portion at rt and the resulting mixture was stirred at rt overnight. After quenching the reaction with water (20 mL), the mixture was transferred into a separatory funnel and the lower phase was collected as a clear oil. Removal of low boiling point impurities from the oil under vacuum gave the mono-ester 2 as a clear oil (14.1 g, 78% yield). 1H NMR (400 MHz, CDCl3) δ 4.14 (s, 6H), 3.91 (s, 2H), 3.57 (s, 2H), 1.46 (s, 9H); 19F NMR (376 MHz, CDCl3) δ −73.51 (s); 13C NMR (100.7 MHz, CDCl3) δ 168.5, 120.2 (q, J = 293.4 Hz), 81.8, 79.1–80.0 (m), 69.2, 67.2, 66.2, 46.1, 27.9; MS (ESI) m/z 905 ((M+1)+); HRMS (MALDI-TOF) calcd for C23H20F27O6 905.0829, found 905.0823.

Mono-acid I (procedure b)

To a stirred solution of tert-butyl ester 2 (13.6 g, 15.0 mmol) and anisol (3.0 mL) in dichloromethane (100 mL) at rt was added trifluoroacetic acid (30 mL) and the resulting solution was stirred at rt for 2 h. After evaporated to dryness under vacuum, the residue was dissolved in methanol/toluene (50 mL/30 mL) and evaporated to dryness under vacuum to give the mono-acid I as a reddish oil (12.6 g, 99% yield) which was used in the next step without further purification. 1H NMR (400 MHz, Acetone-d6) δ 4.29 (s, 6H), 4.14 (s, 2H), 3.73 (s, 2H); 19F NMR (376 MHz, Acetone-d6) δ − 71.24 (s); 13C NMR (100.7 MHz, acetone-d6) δ 170.9, 121.2 (q, J = 292.5 Hz), 80.1–81.0 (m), 68.6, 67.5, 67.1, 47.1; MS (ESI) m/z 848 (M+); HRMS (MALDI-TOF) calcd for C19H12F27O6 849.0203, found 849.0197.

Mono-amide 3 (procedure c)

1,3-diisopropylcarbodiimide (6.6 mL, 5.4 g, 42.5 mmol) was added to a stirred solution of 1-hydroxytriazole (5.7 g, 42.5 mmol) and acid I (12.0 g, 14.2 mmol) in dry N, N-dimethylformamide (200 mL) at rt. After stirring for 15 min, di-tertbutyl iminodiacetate (II-(tBu)2) (10.4 g, 42.5 mmol) was added and the resulting mixture was stirred at rt for 12 h. Water (20 mL) was added to the reaction mixture and the resulting mixture was concentrated and purified by solid-phase extraction on Flouroflash® silica-gel, with H2O and CH3OH as eleunts, to give the mono-amide 3 as a clear oil (14.1 g, 92% yield) which was used in the next step without further purification. 1H NMR (400 MHz, CDCl3) δ 4.12 (s, 6H), 4.10 (s, 2H), 4.02 (s, 2H), 3.92 (s, 2H), 3.59 (s, 2H), 1.44 (s, 9H), 1.42 (s, 9H); 19F NMR (376 MHz, CDCl3) δ −73.29 (s); 13C NMR (100.7 MHz, CDCl3) δ 168.7, 167.9, 167.7, 120.1 (q, J = 293.4 Hz), 82.8, 82.0, 78.7–79.8 (m), 69.5, 67.7, 66.5, 49.9, 48.6, 46.0, 27.8, 27.76; MS (MALDI-TOF) m/z 1098 ((M+Na)+); HRMS (MALDI-TOF) calcd for C31H32F27NNaO9 1098.1544, found 1098.1538.

Di-acid 4 (procedure b)

This compound was prepared by employing the general procedure b. (reagents and solvent amounts were doubled compared to that of mono-acid I) with a 99% yield as a pale solid. 1H NMR (400 MHz, CD3OD) δ 4.22 (s, 8H), 4.16 (s, 2H), 4.14 (s, 2H), 3.60 (s, 2H); 19F NMR (376 MHz, CD3OD) δ −71.18 (s); 13C NMR (100.7 MHz, CD3OD) δ 172.5, 172.1, 171.7, 121.6 (q, J = 293.4 Hz), 80.4–81.5 (m), 70.3, 68.6, 67.9, 50.0, 47.4; MS (MALDI-TOF) m/z 964 ((M+1)+); HRMS (MALDI-TOF) calcd for C23H17F27NO9 964.0472, found 964.0476.

Tri-amide 5 (procedure c)

This compound was prepared by employing the general procedure c (reagents and solvent amount were doubled compared to that of mono-amide 3) with a 93% yield as an oil. 1H NMR (400 MHz, CDCl3) δ 3.80–4.13 (m, 20H), 3.47 (s, 2H), 1.28–1.31 (m, 36H); 19F NMR (376 MHz, CDCl3) δ −72.94 (s); 13C NMR (100.7 MHz, CDCl3) δ 169.2, 169.1, 168.2, 167.6, 167.5, 167.45, 167.3, 120.0 (q, J = 293.4 Hz), 82.9, 82.6, 81.8, 81.5, 78.7–79.8 (m), 68.3, 67.4, 66.6, 50.4, 50.1, 49.3, 48.9, 47.6, 46.6, 45.8, 27.5, 27.4; MS (MALDI-TOF) m/z 1456 ((M+K)+); HRMS (MALDI-TOF) calcd for C47H58F27KN3O15 1456.3074, found 1456.3327.

Tetra-acid 6 (procedure b)

This compound was prepared by using the general procedure b (reagents and solvent amount were quadrupled compared to that of mono-acid I) with a 99% yield as a pale wax. 1H NMR (400 MHz, CD3OD) δ 4.15 (s, 2H), 4.05–4.09 (m, 10H), 3.90–4.01 (m, 6H), 3.53–3.59 (m, 2H), 3.43 (s, 2H); 19F NMR (376 MHz, CD3OD) δ −71.13 (s); 13C NMR (100.7 MHz, CD3OD) δ 172.5, 172.4, 172.3, 172.15, 172.0, 171.6, 171.0, 121.6 (q, J = 292.6 Hz), 80.4–81.5 (m), 69.7, 69.1, 68.7, 68.2, 67.5, 50.5, 50.3, 49.8, 49.77, 47.3; MS (MALDI-TOF) m/z 1216 ((M+Na)+); HRMS (MALDI-TOF) calcd for C31H26F27N3NaO15 1216.0830, found 1216.0820.

Hepta-amide 7 (procedure c ′)

1,3-Diisopropylcarbodiimide (6.6 mL, 5.3 g, 42.0 mmol) was added to a stirred solution of 1-hydroxybenzotriazole (5.7 g, 42.0 mmol), 4 Å molecular sieve (5.0 g) and tetra-acid 6 (6.3 g, 5.3 mmol) in dry N, N-dimethylformamide (150 mL) at rt. After stirring for 15 min, 1-phenyl-2, 5, 8, 11-tetraoxatridecan-13-amine (III-Bn)[24] (10.0 g, 35.2 mmol) was added and the resulted mixture was stirred for 18h at rt. The mixture was filtrated, concentrated and purified by solid-phase extraction on Flouroflash® silica gel to give crude product. Then, a second coupling procedure as above mentioned was repeated with the crude product. After purification by solid-phase extraction on Flouroflash® silica-gel with H2O and CH3OH as eleunts, hepta-amide 7 was prepared as a clear oil (10.7 g, 90% yield). 1H NMR (400 MHz, CDCl3) δ 7.25–7.36 (m, 20H), 4.55 (s, 8H), 4.07–4.16 (m, 14H), 3.90–3.95 (m, 6H), 3.50–3.68 (m, 58H), 3.37–3.43 (m, 10H); 19F NMR (376 MHz, CDCl3) δ −73.47 (s); 13C NMR (100.7 MHz, CDCl3) δ 169.8, 169.7, 169.1, 168.9, 168.7, 168.3, 168.0, 138.0, 128.3, 127.7, 127.4, 120.0 (q, J = 293.3 Hz), 79.3–79.9 (m), 73.1, 70.8, 70.3, 70.0, 69.9, 69.8, 69.3, 69.1, 69.0, 68.3, 67.4, 66.4, 52.8, 52.7, 52.6, 48.3, 47.2, 45.8, 39.3, 39.2; MS (MALDI-TOF) m/z 2277 ((M+Na)+); HRMS (MALDI-TOF) calcd for C91H118F27N7NaO27 2277.7576, found 2277.7494.

19FIT (procedure b′)

A mixture of hepta-amide 7 (5.5 g, 2.5 mmol) and palladium on carbon (10%, 2.5 g) in methanol (200 mL) was stirred under an atmosphere of 50 bar hydrogen over 12 h at rt. After filtering the mixture through a pad of celite, the mixture was concentrated and purified by HPLC on preparative Fluoroflash® column with H2O and CH3OH as eleunts to give the pure 19FIT as a wax (4.5 g, 97% yield). The purity of 19FIT was verified using analytical HPLC (Figure S1). 1H NMR (400 MHz, CD3OD) δ 4.26 (s, 2H), 4.25 (s, 8H), 4.22 (s, 2H), 4.18 (s, 2H), 4.15 (s, 2H), 4.05 (s, 2H), 4.04 (s, 2H), 3.59–3.67 (m, 42H), 3.53–3.58 (m, 16H), 3.44 (t, J = 4.2 Hz, 4H), 3.38 (t, J = 5.6 Hz, 4H); 19F NMR (376 MHz, CD3OD) δ − 71.05 (s); 13C NMR (100.7 MHz, CD3OD) δ 172.3, 171.9, 171.3, 171.27, 171.1, 170.8, 170.5, 121.6 (q, J = 292.5 Hz), 80.3–81.5 (m), 73.7, 71.6, 71.4, 71.2, 71.18, 71.1, 70.4, 70.3, 69.5, 68.7, 68.3, 62.2, 53.2, 53.0, 52.5, 49.7, 47.2, 40.5, 40.4; MS (MALDI-TOF) m/z 1916 ((M+Na)+); HRMS (MALDI-TOF) calcd for C63H94F27N7NaO27 1916.5664, found 1916.5642.

MRI and Metabolic studies

See Supporting Information.

Supplementary Material



**This work was supported by NIH grants (EY015181 & EB004416), the Maryland Nano-Biotechnology Fund and the University of Utah seed incentive grant. We thank Dr. Yong-En Sun for assisting with animal handling.

Supporting information for this article is available on the WWW under or from the author.

Contributor Information

Dr. Zhong-Xing Jiang, Fischell Department of Bioengineering, University of Maryland, College Park (USA)

Mr. Xin Liu, Department of Physics, University of Utah, Salt Lake City, (USA)

Eun-Kee Jeong, Department of Radiology, University of Utah, Salt Lake City (USA)

Yihua Bruce Yu, Department of Pharmaceutical Sciences, University of Maryland, Baltimore (USA), Fischell Department of Bioengineering, University of Maryland, College Park (USA)


1. Lauterbur PC. Angew. Chem. 2005;117:1026–1034.Angew. Chem. Int. Ed. 2005;44:1004–1011. [PubMed]
2. Mansfeld P. Angew. Chem. 2004;116:5572–5580.Angew. Chem. Int. Ed. 2004;43:5456–5464. [PubMed]
3. Yu J-X, Kodibagkar VD, Cui W, Mason RP. Curr. Med. Chem. 2005;12:819–848. [PubMed]
4. Wolf W, Presant CA, Waluch V. Adv. Drug. Del. Rev. 2000;41:55–74. [PubMed]
5. Holland GN, Bottomley PA, Hinshaw WS. J. Magn. Reson. 1977;28:133–136.
6. Lauterbur PC. Nature. 1973;242:190–191.
7. Mason RP, Antich PP, Babcock EE, Gerberich JL, Nunnally RL. Magn. Reson. Imag. 1989;7:475–485. [PubMed]
8. Meyer KL, Carvlin MJ, Mukherji B, Sloviter HA, Joseph PM. Invest. Radiol. 1992;27:620–627. [PubMed]
9. Morawski AM, Winter PM, Yu X, Fuhrhop RW, Scott MJ, Hockett F, Robertson JD, Gaffney PJ, Lanza GM, Wickline SA. Magn. Reson. Med. 2004;52:1255–1262. [PubMed]
10. Ahrens ET, Flores R, Xu H, Morel PA. Nat. Biotech. 2005;23:983–987. [PubMed]
11. Neubauer AM, Myerson J, Caruthers SD, Hocket FD, Winter PM, Chen J, Gaffney PJ, Robertson JD, Lanza GM, Wickline SA. Magn. Reson. Med. 2008;60:1066–1072. [PMC free article] [PubMed]
12. Janjic JM, Srinivas M, Kadayakkara DKK, Ahrens ET. J. Am. Chem. Soc. 2008;130:2832–2841. [PubMed]
13. Kimura A, Narazaki M, Kanazawa Y, Fujiwara H. Magn. Reson. Imag. 2004;22:855–860. [PubMed]
14. Nosé Y. Artificial Organs. 2004;28:807–812. [PubMed]
15. Gore J. New. Eng. J. Med. 2003;349:2290–2292. [PubMed]
16. Willmann JK, van Bruggen N, Dinkelborg LM, Gambhir SS. Nat. Rev. Drug Disc. 2008;7:591–607. [PubMed]
17. Strickley RG. Pharm. Res. 2004;21:201–230. [PubMed]
18. Horváth IT, Rábai J. Science. 1994;266:72–75. [PubMed]
19. Curran DP. Synlett. 2001;9:1488–1496.
20. Ratner AV, Quay S, Muller HH, Simpson BB, Hurd R, Young SW. Invest. Radiol. 1989;24:224–227. [PubMed]
21. Lee H, Price RR, Holburn GE, Partain CL, Adams MD, Catheris WP. J. Magn. Reson. Imag. 1994;4:609–613. [PubMed]
22. Tomalia DA, Reyna LA, Svenson S. Biochem. Soc. Trans. 2007;35:61–67. [PubMed]
23. Jiang Z-X, Yu YB. Tetrahedron. 2007;63:3982–3988. [PMC free article] [PubMed]
24. Jiang Z-X, Yu YB. Synthesis. 2008:215–220. [PMC free article] [PubMed]