Hand-held Photometer Based on Liquid-Core Waveguide Absorption

Mar 29, 2010 - An in-line spectrophotometer on a centrifugal microfluidic platform for real-time protein determination and calibration. Zhaoxiong Ding...
0 downloads 0 Views 4MB Size
Anal. Chem. 2010, 82, 3394–3398

Hand-held Photometer Based on Liquid-Core Waveguide Absorption Detection for Nanoliter-scale Samples Jian-Zhang Pan, Bo Yao, and Qun Fang* Institute of Microanalytical Systems, Department of Chemistry, Zhejiang University, Hangzhou, 310058, China This paper reports a fully integrated hand-held photometer based on the liquid-core waveguide (LCW) detection principle for nanoliter-scale samples. All components of the photometer including light-emitting diode (LED) light source, LCW flow cell, photodiode detector, dropper pump, electronic circuit, liquid-crystal display screen, and battery were fully integrated into a small-sized (12 × 4.5 × 2.1 cm) instrument. A bent optical coupler was developed to conduct the detection light into or out of the LCW flow cell through its sidewall. This design allowed the sampling probe, input and output optical couplers, and LCW flow cell to be integrated in a single Teflon AF capillary, which significantly simplified system structure, improved working reliability, and reduced sample consumption. Two UV-LEDs were used as light source in the photometer to achieve dual wavelength detection at 260 and 280 nm, which was applied to assess on-site the quality and quantity of DNA samples. The effective optical path length of the photometer was ∼15 mm with a sample consumption of only 350 nL. The potential of the photometer applied in point of care testing was also demonstrated in the measurement of total cholesterol in serum samples. Absorption detection is a well-documented and widely used analytical technique. However, most of the conventional cuvettebased absorption spectrometers are usually used in the measurements of milliliter-scale samples, rather than microliter or nanoliter scale samples. Miniaturization has been one of the most important development trends of analytical instruments. In recent years, the microfluidic analysis technique has shown great advantages1-3 in low sample consumption and instrument miniaturization. However, many miniaturized absorption detection systems suffered loss of detection sensitivity4-6 compared with conventional spectrophotometers, due to the limited optical path length. * To whom correspondence should be addressed. E-mail: fangqun@ zju.edu.cn. Phone: +86-571-8820 6771. Fax: +86-571-8827 3572. (1) Cheng, X. H.; Irimia, D.; Dixon, M.; Sekine, K.; Demirci, U.; Zamir, L.; Tompkins, R. G.; Rodriguez, W.; Toner, M. Lab Chip 2007, 7, 170–178. (2) Salim, M.; O’Sullivan, B.; McArthur, S. L.; Wright, P. C. Lab Chip 2007, 7, 64–70. (3) Sia, S. K.; Linder, V.; Parviz, B. A.; Siegel, A.; Whitesides, G. M. Angew. Chem., Int. Ed. 2004, 43, 498–502. (4) Splawn, B. G.; Lytle, F. E. Anal. Bioanal. Chem. 2002, 373, 519–525. (5) Balslev, S.; Jorgensen, A. M.; Bilenberg, B.; Mogensen, K. B.; Snakenborg, D.; Geschke, O.; Kutter, J. P.; Kristensen, A. Lab Chip 2006, 6, 213–217. (6) Malic, L.; Kirk, A. G. Sens. Actuators, A 2007, 135, 515–524.

3394

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

To improve the detection sensitivity of the miniaturized absorption detection systems, various approaches for extending the optical path length employing axial-direction,7-11 multireflection,12-15 cavity ring-down spectroscopy16,17 and liquid-core waveguide (LCW)18-20 detection modes were developed. Steigert et al.9 developed a chip-based system for alcohol measurement, using axial-direction detection mode along the microchannel to increase the path length to 10 mm. Mishra and Dasgupta15 reported a multireflection absorption detector based on a silvercoated capillary and got a 50-fold gain in signal-to-noise compared to the same capillary using the perpendicular-direction detection mode. Waechter et al.17 developed an absorption detection system based on fiber-loop ring-down spectroscopy and obtained a detection limit of 5 µM for tartrazine with a minor detection volume of only 6 nL. With the LCW absorption detection mode, the low refractive index materials were used to construct the flow cell to achieve the total internal reflection of the detection light within the liquid filled in it. The LCW mode has been widely applied in the construction of high-sensitive absorption detector in the conventional systems21,22 due to its excellent waveguide ability, ease of extending the path length, and commercially available LCW capillary (such as Teflon AF 1600/2400 capillary). This technique was also used in microfluidic systems. Duggan et al.20 embedded (7) Collins, G. E.; Lu, Q.; Pereira, N.; Wu, P. Talanta 2007, 72, 301–304. (8) Noda, T.; Takao, H.; Yoshioka, K.; Oku, N.; Ashiki, M.; Sawada, K.; Matsumoto, K.; Ishida, M. Sens. Actuators, A 2006, 119, 245–250. (9) Steigert, J.; Grumann, M.; Brenner, T.; Riegger, L.; Harter, J.; Zengerle, R.; Ducree, J. Lab Chip 2006, 6, 1040–1044. (10) Petersen, N. J.; Mogensen, K. B.; Kutter, J. P. Electrophoresis 2002, 23, 3528–3536. (11) Doku, G. N.; Haswell, S. J. Anal. Chim. Acta 1999, 382, 1–13. (12) Sun, Y.; Yu, X.; Nguyen, N. T.; Shum, P.; Kwok, Y. C. Anal. Chem. 2008, 80, 4220–4224. (13) Llobera, A.; Demming, S.; Wilke, R.; Buttgenbach, S. Lab Chip 2007, 7, 1560–1566. (14) Salimi-Moosavi, H.; Jiang, Y. T.; Lester, L.; McKinnon, G.; Harrison, D. J. Electrophoresis 2000, 21, 1291–1299. (15) Mishra, S. K.; Dasgupta, P. K. Anal. Chim. Acta 2007, 605, 166–174. (16) van der Sneppen, L.; Ariese, F.; Gooijer, C.; Ubachs, W. Annu. Rev. Anal. Chem. 2009, 2, 13–35. (17) Waechter, H.; Bescherer, K.; Du ¨ rr, C.; Oleschuk, R. D.; Loock, H.-P. Anal. Chem. 2009, 81, 9048–9054. (18) Du, W. B.; Fang, Q.; He, Q. H.; Fang, Z. L. Anal. Chem. 2005, 77, 1330– 1337. (19) Du, W. B.; Qun, F.; Fang, Z. L. Chem. J. Chin. Univ.-Chin. 2004, 25, 610– 613. (20) Duggan, M. P.; McCreedy, T.; Aylott, J. W. Analyst 2003, 128, 1336–1340. (21) Waterbury, R. D.; Yao, W. S.; Byrne, R. H. Anal. Chim. Acta 1997, 357, 99–102. (22) Yao, W. S.; Byrne, R. H. Talanta 1999, 48, 277–282. 10.1021/ac100257z  2010 American Chemical Society Published on Web 03/29/2010

a Teflon AF tubing in a sandwich chip to form a LCW flow cell with a 5 mm path length. In 2005, our group18 developed a microchip-based absorption detection system using a Teflon AF capillary coupled to the chip channel as the detection flow cell. An effective path length of 17.4 mm was obtained with a detection volume of only 90 nL. However, in most of the absorption detection systems based on LCW capillaries, two tee connectors were usually used at the inlet and outlet ends of the capillary to deliver the liquid as well as to guide the detection light into and out of the capillary. The tee connector has the problem of relatively large dead volume existing at the joint region between the LCW capillary, optical fiber, and connecting tube or microchannel. During measurement, some gas bubbles are easily trapped at the dead corner and are difficult to be eliminated. These bubbles generate serious scattering of the detection light and significantly deteriorate the system detection accuracy and working reliability. Furthermore, the existence of the dead volume in the tee connector also increases the difficulty of washing the flow cell after each measurement. Additionally, these previous designs18-22 used the lab-bench components, and no attempt has been made at integration into a single device yet. In the present work, a hand-held photometer with a size of 12 × 4.5 × 2.1 cm was developed, in which all of the components were fully integrated, including the LCW flow cell, light-emitting diode (LED) light source, photodiode detector, dropper pump for liquid driving, electronic circuit for instrument control and data processing, liquid-crystal display (LCD) screen, and battery. A Teflon capillary was employed to construct the LCW flow cell to obtain high detection sensitivity and nanoliter-scale sample consumption. Instead of the tee coupler used in most of the LCW systems, a U-shaped bent coupler was developed based on a Teflon LCW capillary to simplify the system structure and improve working reliability. The design of a U-shaped, bent coupler for solid optical fibers has been widely used in the construction of evanescent field absorption sensors.23-25 In this work, this concept was applied in a liquid-core optical-waveguide capillary system for the first time. Since the optical transmission behavior in the LCW-capillary bent coupler is quite different from that of solid optical fibers, it was investigated by both experiment and ray tracing simulation methods. By combining the bent coupler with two UV-LEDs, dual wavelength detection at 260 and 280 nm was achieved for measurements of nanoliter-scale DNA samples. EXPERIMENTAL SECTION Chemicals and Reagents. All chemicals used were of analytical grade unless mentioned otherwise, and deionized water was used throughout. The standard Lambda DNA stock solution of 0.3 mg/mL was obtained from Fermentas Co. (Shenzhen, China). This solution was also used as concentrated solution to calibrate the photometer. A series of standard solutions was prepared by sequentiallydilutingtheDNAstocksolutionwithTris-hydrochloride buffer (pH 8.0). Three DNA samples were isolated from African green monkey kidney fibroblast cells, mouse cortical astrocytes, and the leaf of moso bamboo, respectively (see Supporting (23) Gupta, B. D.; Dodeja, H.; Tomar, A. K. Opt. Quantum Electron. 1996, 28, 1629–1639. (24) Yan, Q. G.; Tao, S. Q.; Toghiani, H. Talanta 2009, 77, 953–961. (25) Guo, H. Q.; Tao, S. Q. Sens. Actuators, B 2007, 123, 578–582.

Figure 1. Appearance (A), layout (B), and schematic diagram of the absorption detection module of the hand-held photometer (C).

Information). The cholesterol assay kit (see Supporting Information) was obtained from Biosino Bio-Technology & Science Inc. (Beijing, China). The working solutions were prepared by mixing 1 mL of reagent solution with 10 µL of cholesterol sample solution and incubated at 37 °C for 6 min. A blank and a concentrated solution for photometer calibration were prepared by mixing 1 mL of reagent solution with 10 µL of water and 1% H2O2, respectively. Apparatus. The hand-held photometer (Figure 1A,B) consisted of modules including absorption detection, dropper pump, and electronic control systems. Absorption Detection Module. The schematic diagram of the absorption detector is shown in Figure 1C. A 10 cm-long Teflon AF 2400 capillary (90 µm i.d., 345 µm o.d., Random Tech., San Francisco, CA) was bent at point A and B to form two U-shaped bends which served as input and output optical couplers, respectively. Before bending, the capillary was heated to 275 °C by a hot air gun. The U-shaped bend was produced by first bending the heated capillary into a U-shape, then fitting it into a 1 mm gap between two jaws of a vernier caliper to obtain a bend radius (Rb) of 0.3 mm, and finally allowing it to cool to room temperature for shape setting. The inlet end of the Teflon capillary was used as a sampling probe, and its outlet was connected to the dropper pump. In DNA measurements, two UV-LEDs (UVTOP260TO39BL, λ ) 260 nm, full width at half-maximum, fwhm ) 12 nm, UVTOP280TO39BL, λ ) 280 nm, fwhm ) 12 nm, SET Inc., Columbia, SC) were positioned 10 mm from point A with the Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

3395

emitted light beams focused on the coupler tip with an incident angle of 30°. In cholesterol measurement, a bluish-green LED (NEPE510AS-E, λ ) 497 nm, fwhm ) 25 nm, Nichia Co., Tokushima, Japan) was used as a light source with the same incident angle. A photodetector (OPT301, Texas Instruments, Tucson, AZ) was fixed with its detection window facing point B. The AB distances in DNA and cholesterol measurements were 1.5 and 2.5 cm, respectively. The whole absorption detection module was shielded from ambient light. Dropper Pump. The dropper pump consisted of an elastic Tygon tube (11 cm long, 0.25 mm i.d., 1.8 mm o.d.) and a plastic piston (6 mm width). One end of the Tygon tube was connected to the flow cell, and the other was blocked by epoxy glue. The nanoliter-scale sample was introduced into the photometer by first pressing the piston to squeeze the Tygon tube, then inserting the sampling probe into the sample solution, and releasing the piston to aspirate the sample into the flow cell by negative pressure. Electronic Control System. A microcontroller (MSP430F169, Texas Instruments, Dallas, TX) with a self-developed program written in C language was used to control data acquisition, processing, and display (see Supporting Information). A fourbutton keypad and a liquid crystal display were employed for user control and result display, respectively. A constant current source was designed to supply the LEDs. The two LEDs were lit alternatingly at 2 Hz in the DNA measurement. The whole photometer was powered by a 3.6 V Li-ion battery. A data processing program was developed to achieve automated photometer calibration and sample measurements. The absorbance of the sample was calculated according to eq 1. Asample ) -log

[

Isample - Ibackground Iblank - Ibackground

]

(1)

where Iblank, Ibackground, and Isample are the transmitted light intensities when the flow cell filled with the blank, concentrated, and sample solutions, respectively. For DNA measurement, Asample was further normalized by eq 2. A' ) Asample · LSP /LEP

(2)

where LEP is the effective optical path length (mm) which was measured by determining the same DNA standard with the present photometer and a conventional spectrometer, and LSP is the standard path length of 10 mm. The purity of DNA sample was evaluated by the ratio of the normalized absorbances measured at 260 and 280 nm, A′280 nm/A′260 nm. The DNA concentration was calculated by eq 3.26 CDNA ) 50 × A′260nm(µg/mL)

(3)

In the cholesterol measurement, the cholesterol concentration was calculated by eq 4. Ccholesterol )

Cstandard ·A Astandard sample

(4)

(26) Sambrook, J.; Russell, D. W. Molecular Cloning: A Laboratory manual; 3rd ed.; Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, 2001.

3396

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

Figure 2. Schematic simulation result of ray tracing at the bent coupler with three modes of ray transmission. The arrow indicates the transmission direction of the rays.

The Astandard of a cholesterol standard with concentration of Cstandard was premeasured with the hand-held photometer before the sample measurement. In comparison experiments with conventional spectrophotometry, a UV-vis spectrophotometer (WFZ800-D2C, Rayleigh Analytical Instrument Co., Beijing, China) was used. TracePro software (Lambda Research Co., Littleton, CO) was used to simulate ray transmission in the LCW capillary (see Supporting Information for details). Procedures. Before sample measurement, the hand-held photometer was calibrated by introducing the concentrated and blank solutions, respectively. The sample solution was introduced into the flow cell using the dropper pump, and the measurement results were displayed on the screen. The operations for parameter setting, photometer calibration, and sample measurement were user-controlled via the keypad and menu-based interface. Safety Considerations: The fabrication of the bent coupler should be carried out in a ventilated hood with the heating temperature less than 360 °C. RESULTS AND DISCUSSION LCW Absorption Detection Module. Design of Bent Coupler. In the design of the absorption detection module, the main objective was to construct optical couplers based on a single Teflon capillary. In the straight section of the LCW capillary, detection light rays transmitted along the capillary channel by total internal reflection (TIR) when their incident angles were larger than the critical angle of the water/Teflon interface. Due to this characteristic, it is difficult to couple light rays directly through the wall of a straight LCW capillary. However, the situation for a bent LCW capillary is quite different. On the basis of the results of ray tracing simulation, there are three ray transmission modes in the bent region (Figure 2). Rays I and II were reflected by interfaces of water/Teflon and Teflon/air, respectively, because their incident angles on the interface were larger than the corresponding critical angles (θwater/Teflon ) 76°, θTeflon/air ) 51°). Rays III with incident angles less than θwater/Teflon and θTeflon/air traversed both water/ Teflon and Teflon/air interfaces into the air. This means that the bend could be used to couple rays out of the LCW capillary through its wall. According to the principle of ray reversibility, this bend could also couple rays into the LCW capillary. This deduction was further verified in the following experiments. Such

Figure 3. (A) Schematic diagram of the device used in the optimization of the bend radius of the bent coupler. (B) Schematic simulation result for output rays from a 0.3 mm-Rb output coupler. The arrow indicates the transmission direction of the rays. The blue rays represent that their intensities were reduced to 33%-66% of the initial red rays, and the green rays were reduced to 0%-33%. (C) Data of coupling efficiency of output couplers with different Rb obtained in both experiment and ray tracing simulation.

bent couplers monolithically produced on a capillary not only eliminated the unreliable factors of the tee couplers but also significantly simplified the detection system structure by integrating the flow cell, optical couplers, and sampling probe into a single LCW capillary. Optimization of Bend Radius of Coupler. On the basis of the above simulation results, the variation of the coupler bend radius Rb had significant influence on the coupler performance. Therefore, the effect of Rb of an output bent coupler on the output coupling efficiency (Eoc) was investigated with the device (see Supporting Information) shown in Figure 3A. The Eoc value was defined as the ratio of light intensities measured at the bend (Figure 3A) and at the outlet end of the straight LCW capillary. Both the experiment and simulation showed that the Eoc increased with the decrease of Rb in the range of 0.3-10 mm (Figure 3B). Figure 3C shows a schematic simulation result of output rays from a coupler with 0.3 mm Rb. A higher Eoc is correlated to a larger signal-to-noise ratio at the detector. According to the principle of ray reversibility, the result is also applicable to the input coupler; thus, 0.3 mm Rb was chosen for both the output and input couplers. Optimization of the Incident Angle of Light. The effect of the incident angle of the LED light in the range of 0°-70° on the coupling efficiency of the input coupler was investigated using the device (see Supporting Information) shown in Figure 4A.

Figure 4. (A) Schematic diagram of the device used in optimization of the incident angle of the LED light. Image series of the output light at the outlet end of the LCW capillary obtained in both (B) experiment and (C) ray tracing simulation at different incident angles. (D) Quantitative data of the output light flux through the capillary channel on the imaging plane in image series (B) and (C). (E) Schematic raytracing simulation result for input rays at 30° incident angle on the central axis plane of the LCW capillary with a bend radius of 0.3 mm.

Figure 4B,C show two image series of the output light at the outlet end of the LCW capillary obtained in both experiment and ray tracing simulation at different incident angles, respectively. The quantitative data of the output light flux through the capillary channel on the imaging plane in these images are summarized in Figure 4D. Both the experimental and simulation results showed similar trends. The output light flux, being proportional to the input coupling efficiency of the coupler, increased with the incident angle in 0°-30°, reached maximum at 30°, and decreased in 30°-70°. Therefore, the incident angle of 30° for the LED light was employed. Figure 4E shows a schematic ray tracing result on the central axis plane of the capillary at 30° incident angle. There are three light transmission modes in the LCW capillary after the external light was coupled into the capillary. In mode a, the incident rays were guided only within the capillary wall without passing through its channel, which had no contribution to the absorbance signal but produced relatively high background. Such background signals could be eliminated by subtraction of the background signal recorded without a sample from the measurement with the Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

3397

Table 1. Results of the Handheld Photometer and Conventional Photometer hand-held photometer sample 1 sample 2 sample 3 a

conventional photometer

A260 nm

A280 nm

A260 nm/A280 nma

CDNA (µg/mL)

A260 nm

A280 nm

A260 nm/A280 nma

CDNA (µg/mL)

0.112 ± 0.001 0.211 ± 0.002 0.092 ± 0.001

0.060 ± 0.001 0.112 ± 0.001 0.052 ± 0.001

1.84 ± 0.03 1.88 ± 0.04 1.79 ± 0.04

5.60 ± 0.05 10.55 ± 0.10 4.60 ± 0.05

0.109 ± 0.001 0.203 ± 0.001 0.097 ± 0.001

0.058 ± 0.001 0.104 ± 0.001 0.054 ± 0.001

1.79 ± 0.02 1.94 ± 0.02 1.82 ± 0.02

5.45 ± 0.05 10.15 ± 0.05 4.85 ± 0.05

The ratio of A260

nm

and A280

nm

for a pure DNA sample ranges from 1.8 to 2.0.

sample before sample measurement. In mode b, the rays were guided within the capillary channel. In mode c, the rays were transmitted through the capillary wall and channel alternately. The effective absorbance pathlengths of the rays under mode c were shorter than those under mode b, since the transmission of the rays in the capillary wall did not generate effective absorption. DNA Quality Assessment. Quick quality assessment of nucleic acid or protein samples are frequently carried out in molecular biology or biochemistry laboratories using the data of A260 nm and A280 nm to estimate the purity and concentration. The photometer with two UV-LEDs (260 and 280 nm) was applied in the measurement of DNA samples. The size of the photometer was 12 × 4.5 × 2.1 cm with a weight less than 70 g. The effective path length of the photometer was ∼15 mm with a detection volume of ∼100 nL. After the sample was introduced, the data of A260 nm, A280 nm, A280 nm/A260 nm, and DNA concentration were automatically displayed on the LCD screen. Linear responses of A260 nm ) 2.11 × 10-2C (µg/mL) - 9.1 × 10-5 (r2 ) 0.998) and A280 nm ) 1.10 × 10-2C(µg/mL) - 5.5 × 10-5 (r2)0.998) in the range of 1.0-20.0 µg/mL DNA standard were obtained (see Supporting Information for details). The limit of detection for double stranded DNA based on 3 times the standard deviation of the blank values was 0.1 µg/mL. The precision of the system was 2.3% RSD (n ) 11) for 10.0 µg/mL DNA standard. The sample consumption for each measurement was 350 nL. The photometer was also applied to assess three real DNA samples. For comparison, these samples were also measured using a conventional spectrophotometer with a 100 µL cuvette (10 mm path length). The results are listed in Table 1. There is no obvious difference between the two sets of results. However, the present photometer had 1.5-fold path length and only onethree-hundredth sample consumption of the spectrophotometer. Compared with the commercial spectrophotometer for tiny samples, such as NanoDrop ND-1000 from Thermo Fisher Scientific Co. (Waltham, MA), the present photometer had onethird of sample consumption, 15-fold longer path length, compact size, and lower cost. Cholesterol Measurement. The photometer with a single bluish-green LED source was used to perform measurement of total cholesterol in serum samples, to demonstrate its potential application in point of care testing. Due to the use of a relatively cheap visible-LED source, the total cost of the photometer was reduced to ca. US $35. A linear response A ) 1.02 × 10-1C - 8.7 × 10-3 (r2 ) 0.996) in the clinically relevant concentration range of 0.26-5.22 mM cholesterol was obtained. The precision was 1.2%

3398

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

RSD (n ) 11) for 0.26 mM cholesterol standard. The photometer was also applied in the determination of total cholesterol in real human serum samples from three healthy donors. The measurement results were 3.79 ± 0.03, 3.64 ± 0.04, and 2.62 ± 0.04 mM cholesterol (n ) 3). The same samples were also measured using a conventional biochemical analyzer from Zhejiang University Hospital, with corresponding results of 3.85, 3.56, and 2.73 mM cholesterol. There is no significant difference between the two sets of results. CONCLUSIONS The present photometer has a low sample consumption and is sensitive, self-contained, and portable. It is suitable for nanoliter sample analysis and on-site analysis. Benefiting from the use of integrated LCW flow cell, the sample consumption was reduced to several hundred nanoliters without evident loss in detection sensitivity, and if required, it could be further reduced by shortening the lengths of the sampling probe and flow cell. The photometer achieved dual wavelength detection at 260 and 280 nm using two UV-LEDs, which would be quite useful in routine quality assessment not only for DNA but also for RNA and protein samples in laboratories of molecular biology and biochemistry. On the basis of this design, multiwavelength detection could be readily achieved in the photometer using multiple LEDs with different emitting wavelengths as the light source. In addition to the above applications, the photometer could also be easily coupled to various miniaturized flow analysis systems, such as flow-injection analysis, sequential-injection analysis, or microfluidic analysis, to serve as an absorption detector. ACKNOWLEDGMENT Financial support from the National Natural Science Foundation of China (Grants 20775071, 20825517, and 20890020) and the Ministry of Science and Technology of China (Grants 2007CB714503 and 2007CB914100) is gratefully acknowledged. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text, figures and a movie of the working process of the photometer. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review January 28, 2010. Accepted March 17, 2010. AC100257Z