Biosensors and Bioelectronics 66 (2015) 50–54

Contents lists available at ScienceDirect

Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios

Sensitive and specific colorimetric DNA detection by invasive reaction coupled with nicking endonuclease-assisted nanoparticles amplification Bingjie Zou a,1, Xiaomei Cao a,1, Haiping Wu b, Qinxin Song c, Jianping Wang c, Tomoharu Kajiyama d, Hideki Kambara d, Guohua Zhou a,b,e,n a

Department of Pharmacology, Jinling Hospital, Medical School of Nanjing University, Nanjing 210002, China Huadong Research Institute for Medicine and Biotechnics, Nanjing 210002, China School of Life Science and Technology, China Pharmaceutical University, Nanjing 210009, China d Central Research Laboratory, Hitachi, Ltd., 1-280 Higashi-Koigakubo, Kokubunji-shi, Tokyo 185-8601, Japan e State Key Laboratory of Analytical Chemistry for Life Science, Nanjing University, Nanjing 210009, China b c

art ic l e i nf o

a b s t r a c t

Article history: Received 19 August 2014 Received in revised form 19 October 2014 Accepted 31 October 2014 Available online 6 November 2014

Colorimetric DNA detection is preferable to methods in clinical molecular diagnostics, because no expensive equipment is required. Although many gold nanoparticle-based colorimetric DNA detection strategies have been developed to analyze DNA sequences of interest, few of them can detect somatic mutations due to their insufficient specificity. In this study, we proposed a colorimetric DNA detection method by coupling invasive reaction with nicking endonuclease-assisted nanoparticles amplification (IR-NEANA). A target DNA firstly produces many flaps by invasive reaction. Then the flaps are converted to targets of nicking reaction-assisted nanoparticles amplification by ligation reaction to produce the color change of AuNPs, which can be observed by naked eyes. The detection limit of IR-NEANA was determined as 1 pM. Most importantly, the specificity of the method is high enough to pick up as low as 1% mutant from a large amount of wild-type DNA backgrounds. The EGFR gene mutated at c.2573 T4G in 9 tissue samples from non-small cell lung cancer patients were successfully detected by using IRNEANA, suggesting that our proposed method can be used to detect somatic mutations in biological samples. & Elsevier B.V. All rights reserved.

Keywords: Invasive reaction Nicking endonuclease AuNPs Colorimetric detection Ligation reaction Mutation detection

1. Introduction Sensitive and specific DNA detection is very important not only for biology research but also for clinical diagnostics. However, the concentration of DNA in biological samples is very low, thus, many DNA amplification methods have been developed, such as polymerase chain reaction (PCR), (Mi et al., 2009; Mullis and Faloona, 1987; Pollet et al., 2011) loop-mediated isothermal amplification (LAMP), (Hsieh et al., 2012; Notomi, 2000; Wong et al., 2014) rolling cycle amplification (RCA). (Cheng et al., 2009; Lizardi et al., 1998; Zhao et al., 2013) However, these methods are based on template amplification, which usually causes amplicon crosscontamination, and leads to false positive results. Signal amplification which amplifies target-specific signal instead of target itself n Corresponding author at: Department of Pharmacology, Jinling Hospital, Medical School of Nanjing University, No. 305, East Zhongshan Road, Nanjing 210002, China. E-mail address: [email protected] (G. Zhou). 1 These authors contributed equally to this work.

http://dx.doi.org/10.1016/j.bios.2014.10.077 0956-5663/& Elsevier B.V. All rights reserved.

is more suitable for DNA detection because no amplicon crosscontamination occurs. So far, many signal amplification-based DNA detection strategies have been proposed, including Invader Assay, (Hall et al., 2000; Lyamichev et al., 1999; Lyamichev et al., 2000) nicking endonuclease signal amplification (NESA) (Kiesling et al., 2007), cascade enzymatic signal amplification (CESA) (Zou et al., 2011, 2013) and so on. As these methods utilize dye-labeled probes, an expensive instrument is required for detecting amplified signals. Gold nanoparticles have useful optical property that derives from their surface plasmon resonance (Mirkin et al., 1996). The optical property is dependent on the size of the particles or the distance between each particle (Storhoff et al., 2000). So the aggregation of the gold nanoparticles could cause the change of the optical property. This characteristic can be used to develop a sensor for DNA detection. The main merit of the sensor is the visible detection by the naked eyes, and thus no equipment is required for the read-out. The key of the sensor is how to trigger the aggregation of the particles by the target DNA. The typical method is to design two gold nanoparticle probes (AuNPs) to

B. Zou et al. / Biosensors and Bioelectronics 66 (2015) 50–54

capture a target DNA at two different regions, causing the aggregation of the AuNPs (Mirkin et al., 1996; Storhoff et al., 1998). However, the sensitivity of the assay is just around 10 nM. Most importantly, the assay is limited to detect only short singlestranded oligonucleotides. To improve the sensitivity, nicking endonuclease-assisted amplification was coupled to nanoparticlebased sensor for DNA detection (termed as NEANA) (Xu et al., 2009). In this method, aggregation of AuNPs is achieved by adding a linker complementary to a target DNA as well as both AuNPs. The cleavage of linker probes is triggered by the target DNA when the target DNA and the linker form the duplex with the structure recognizable to a nicking endonuclease (NEase). Because one target DNA can yield around 1000 cleaved linker probes, the sensitivity is increased to 1000 folds in comparison to the conventional gold nanoparticle-based sensor. However, the target of interest should have a sequence recognizable to NEase. Consequently targets without the recognition sequence cannot be detected by NEANA. To improve the feasibility of the NEANA for detecting real biological samples with various sequences, RCA was combined with NEANA (Xu et al., 2012). Although this improvement allows NEANA to detect any target sequence, target-specific padlock probes are required; this might result in a high cost when detecting different targets. Herein, we proposed a colorimetric DNA detection by coupling invasive reaction with NEANA (termed as IR-NEANA for brief). A target DNA can produce so many flaps by invasive reaction, and then the flaps are ligated with a 5′ phosphated oligonucleotide to generate the template of NEANA. Because the invasive reaction just relies on over-lapping structure formed by the specific binding of an upstream probe and a downstream probe to a target DNA, unlike conventional NEANA, any DNA sequence can be detected by IR-NEANA. Most importantly, the flaps from invasive reaction are independent on target sequence, thus, the NEANA-based sensor in our method is universal to any target sequence, and only conventional probes used for forming an invasive structure are required for detecting different target sequence, bringing a very convenient experiment set-up.

51

and kept at 26 °C for 15 min. Then, 1 mL of Nb.BsmI (10 U/mL, New England Biolabs, USA) was added, and incubated at 60 °C for 2 h. 2.3. AuNPs hybridization After nicking reaction, 3 mL of each 13-nm nanoparticle probes (30 nM, prepared according to literature (Hill and Mirkin, 2006)), NaCl (666.7 mM) and water were added in a volume of 30 mL. Hybridization was performed at 55 °C for 30 min. The results were observed by naked eyes and the absorption spectra from 350 to 700 nm were measured by UV/vis Spectrophotometer (OneDrop Technologies, China) after centrifuging the products at 5000 rpm for 10 s.

3. Results and discussion 3.1. Principle In the proposed method, a target DNA is firstly hybridized with an upstream-probe (Up) and a downstream-probe (Dp), forming a one-base overlapping structure at the 3′-end of the Up (as shown in Fig. 1). Flap endonuclease (FEN) recognizes the overlapping structure and cleaves the flap, which is at the 5′-region of the Dp and not complimentary to the target DNA. As the reaction temperature is close to the melting temperature (Tm) of Dp, a cleavage cycle of dissociation of the cleaved Dp and annealing of an intact

2. Material and methods 2.1. Invasive reaction A set of reaction mixtures containing 10 mM MOPS (pH 7.5), 0.05% Tween-20, 0.05% Nonidet P40, 0.1 mM upstream probe (Table S1), 1 mM downstream probe (Table S1), 7.5 mM MgCl2, and various of concentrations (100, 20, 10, 2, 1, 0.2, 0.1 and 0 pM) of artificial target DNA (Table S1) were prepared. After 95 °C for 5 min, 100 ng of Afu flap endonuclease (prepared in our lab) was added to the above mixture with a total volume of 10 mL, followed by the incubation at 63 °C for 2 h. The usage of non-small cell lung cancer FFPE specimens was agreed by Ethics Committee of Jinling Hospital. The informed signed consents were obtained from the patient or from next of kin. 2.2. Ligation and nicking reaction One mL of invasive reaction mixture was added into ligation and nicking mixture containing 5 mL of 2  ligation buffer (132 mM Tris–HCl, 20 mM MgCl2, 2 mM DTT, 2 mM ATP, 15% PEG 6000, pH 7.6, New England Biolabs, USA), 0.5 mL of 2 mM p-oligo (Table S1), 0.5 mL of 2 mM molecular beacon (Table S1) and 1 mL of NEB buffer 3 (0.2 M NaCl, 20 mM Tris–HCl, 20 mM MgCl2, 2 mM DTT, pH 7.9, New England Biolabs, USA), followed by incubation at 45 °C for 5 min. A 1-mL of T4 ligase (17.5 U/mL, TaKaRa, China) was added,

Fig. 1. The principle of IR-NEANA. A target DNA produces so many flaps by invasive reaction; and then the flaps are ligated to P-oligo for producing targets of nicking reaction. Linkers are nicked into two parts, which cannot trigger the aggregation of AuNPs, leading to a red color of the reaction. On the contrary, when the target is absent, the linker is intact, leading to the aggregation of AuNPs, whose color is purple or achromasy.

52

B. Zou et al. / Biosensors and Bioelectronics 66 (2015) 50–54

Dp occurs, producing many flaps from a target DNA. Secondly, the produced flaps are ligated with a 5′ phosphated oligo (P-oligo) hybridized to the linker to generate the target of NEANA. Thirdly, NEANA is carried out by adding NEase. A nick is introduced to a linker while the ligated flap strand is alive to hybridize to another intact linker. The series of reactions are repeated to digest so many linkers. Then the reaction product is added with a solution containing AuNPs. If there are so many linker probes in the mixture, aggregation of nanoparticles occurs to change the color from red to purple. After centrifugation, the aggregation of AuNPs is precipitated from solution, the reaction mixture becomes colorless. When the number of linker probes decrease, the aggregation of DNA-modified AuNPs is reduced to give a small color change of the reaction mixture. When a target DNA is not present, no cleavage occurs, and therefore intact linkers produce the aggregation of AuNPs.

3.3. Optimization of conditions of nanoparticle probes aggregation In IR-NEANA, the presence of a target is detected by the color change of AuNPs through their aggregation. Therefore the precise control of the aggregation is needed. We firstly optimized the concentrations of the linker and NaCl to be 33.3 nM and 666.7 nM in the final hybridization mixture, respectively. The color changed clearly when the concentration of linker decreased to 16.7 nM from 33.3 nM (as shown in Supplementary data, Fig. S2). This means that a positive result should be observed when the amount of ligated flaps was large enough to cleave linkers more than 16.7 nM by NEase. Consequently, the production efficiency of ligated flaps is important. To investigate the ligation efficiency, a ligation-based NEANA was performed to detect flaps with different concentrations (10 nM, 1 nM, 0.1 nM, 0 nM), and the results were compared with the detection of synthesized ligated-flaps (10 nM, 1 nM, 0.1 nM, 0 nM) by ligation-based NEANA. As shown in Fig. S3, the detection limit of flaps was 1 nM as low as that of the synthesized ligated-flaps, indicating that all flaps could be ligated with p-oligo to generate the targets of NEANA.

3.2. Characterization of nanoparticle probes and aggregated nanoparticle probes In IR-NEANA, the read-out is the color change of nanoparticle probes due to the aggregation. We individually measured the UV spectra of nanoparticle probes and aggregated nanoparticle probes, and found that the aggregated AuNPs have a color change from red to purple with the absorption peak shift from 530 nm to 550 nm. The peak shift is around 20 nm and the color change from red to purple is not very obvious (in Supplementary data, Fig. S1A). For illustrating the aggregation of gold nanoparticle, transmission electron microscope (TEM) was performed for the free AuNPs and aggregated AuNPs. We observed that the size of aggregated AuNPs significantly increased (Fig. S1B). To clearly discriminate free nanoparticle probes from aggregated nanoparticle probes, we respectively centrifuged AuNPs and aggregated AuNPs at 5000 rpm for 10 s. It was found that the tube with aggregated AuNPs becomes colorless for the supernatant layer, and does not give any UV absorption spectrum (black curve in the Fig. S1C), while producing the obvious sediment of aggregated gold nanoparticle in the bottom of the tube (the tube near to the black curve in the Fig. S1C). However, the tube with free AuNPs is still red, and shows the UV absorption spectrum similar to that before centrifugation (red curves Fig. S1). This phenomenon indicates that the aggregated AuNPs have a large size, and are easy to be precipitated from solution at a centrifugal force. For an accurate detection, the discrimination of red color (positive) from the colorless (negative) is much sensitive. Therefore, we measure the absorption peak of centrifuged reaction product at 530 nm to monitor the free AuNPs.

A

Target (pM): Blank 0.1 0.2

1

2

10

20

3.4. Detection limit To investigate the sensitivity of IR-NEANA, we performed invasive reaction with various concentrations of target DNA (T1 in Table S1): 0.1 nM, 0.02 nM, 0.01 nM, 2 pM, 1 pM, 0.2 pM, 0.1 pM and 0 pM. After 2- h incubation, 1 μl of products was added into a ligation–nicking reaction mixture to perform ligation and nicking reactions; then the AuNPs were added for hybridization. As shown in Fig. 2A, the red color was observed only in the tubes in which the concentrations of target were higher than 1 pM. The absorption peak of AuNPs at 530 nm is related to the concentration of free AuNPs; and the concentration of free AuNPs is dependent on the amount of the flaps generated from target DNA-specific invasive reaction. So the intensity of the absorption peak of AuNPs at 530 nm (A530) is related to the concentration of a target DNA. As shown in Fig. 2 B, 1 pM target gave the absorbance A530 larger than the blank control did. However, the absorbance A530 from samples with 0.2 pM as well as 0.1 pM samples was very close to that of the blank control. Resultingly, the detection limit of IR-NEANA was determined to be 1 pM, which was 1000-fold lower than that of ligation-based NEANA for flap detection (1 nM, as shown in Fig. S3 A). Therefore, the combination of invasive reaction with NEANA not only enables the detection of targets with any sequences, but also significantly increases the detection sensitivity of NEANA.

B

100

0.20

0.2

0.1

0.1 nM 0.02 nM 0.01 nM 2 pM 1 pM 0.2 pM 0.1 pM Blank

0.15 A530

Absorbance

0.3

0.0 350 400 450 500 550 600 650 700 Wavelength (nm)

0.10 0.05 0.00

0.00.10.2 20 40 60 80 100 The concentration of target (pM)

Fig. 2. Sensitivity investigation of IR-NEANA by detecting targets at various concentrations (0.1 nM, 0.02 nM, 0.01 nM, 2 pM, 1 pM, 0.2 pM, and 0.1 pM). (A) Images and absorption spectra in a region of 350–700 nm. (B) Target concentration dependence of absorbance A530.

B. Zou et al. / Biosensors and Bioelectronics 66 (2015) 50–54

was much lower than those obtained templates having mutation at position e or f. It indicates that the downstream-probe (Dp) is more sensitive to the specificity than the upstream-probe (Up). Nevertheless the specificity of IR-NEANA is high enough to accurately discriminate one base difference in a sequence when the targeted base is at the overlapping position (a) or adjacent to overlapping positions (b and d). The capability of IR-NEANA for picking up a small amount of mutants from a large amount of wild-type DNA was investigated by detecting a series of samples artificially prepared by spiking various amounts of c.2573 T 4G mutant EGFR gene into wild-type EGFR gene (background), enabling the concentration of the mutant to be 50%, 25%, 10%, 5%, 2%, 1%, 0.5%, 0.1%, and 0% in a sample. The probe Dp was designed according to the sequences of c.2573 T4 G mutant EGFR gene, and the mutant base was just at the overlapping position (the sequences of probes and targets could be found in Table S1 in Supporting information). As shown in Fig. 4, the absorbance A530 of a sample containing 1% mutant gene was significantly higher than that of a sample containing the pure wild-type gene (0% mutant); but the absorbance A530 of a sample with 0.5% or 0.1% mutant gene was the same as that for wild-type gene. Therefore IR-NEANA is sensitive to discriminate 1% mutants from wild-type DNAs. As shown in Fig. 4B, the value of A530 increased almost linearly with the increase of the mutant concentration in a sample, but reached to a plateau for samples containing more than 10% mutant gene.

ATCTGGCCTGGTGC-3ಬ 5ಬ-ATGTCACTT CCCCTT GGTTCTCTCC 3ಬ-TACAGTGAAGGGGAACCAAGAGAGTAGACCGGACCACGTTATC -5ಬ f

c

e da b

0.20

Absorbance

Blank c

0.15

b

a

d

e

f

PC

PC

0.10

f e d

0.05

Blank

53

c a b

0.00 350 400 450 500 550 600 650 700 Wavelength (nm) Fig. 3. Images and absorption spectra of IR-NEANA for detecting various targets with one mutant base at different positions (a–f). All mutations were artificially formed by substituting the original base with its complementary one. The concentration of all targets is 0.1 nM. PC (positive control) and Blank (negative control) were obtained with the target without any mutation and without any target, respectively.

3.5. Specificity In IR-NEANA, the target DNA molecules produce so many flaps through invasive reaction and therefore its specificity of the reaction is important. As the ligation-based NEANA was target sequence-independent, the specificity was only depends on invasive reaction. To check the specificity of IR-NEANA, a series of artificial targets with one base mismatch at different locations (alphabetically indicated in Fig. 3, and the sequences as shown in Table S1) were detected by IR-NEANA. The results (Fig. 3) showed that the absorbance A530 from the templates with a mutant base at the position a, b, or d were close to that of blank control, but the templates with a mutant base at the position c, e, or f gave small background signals. This suggested that targets with one mutant base at different positions gave different efficiency in invasive reaction, yielding different amount of the flaps. Because the absorbance A530 is related to the flaps, the absorbance A530 from different targets varied. Moreover, it was found that the background signal from the template with a mutant base at position c

3.6. Analysis of biological samples To verify whether the IR-NEANA can be used to detect real biological samples, the mutation of the EGFR gene at c.2573 T4 G, which is a biomarker for guiding the use of Gefitinib, was employed as an example for the real sample. Nine formalin-fixed, paraffin-embedded (FFPE) specimens from non-small cell lung cancer patients were analyzed by IR-NEANA. It was found that 2 samples (S7, S8) carry the c.2573 T 4G mutation (Fig. S4). For comparison, these samples were also analyzed by pyrosequencing (Fig. S5). Although the same results were obtained by pyrosequencing, the peaks (marked as arrows in pyrograms) corresponding to the mutant are so low that it may lead to false negative results. Therefore IR-NEANA we have developed here is far superior to pyrosequencing in the detection of somatic mutations from tumor cells.

A

B 50% 25% 10% 5% 2% 1% 0.5% 0.1% 0%

0.2

0.1

0.20 0.15

A350

Absorbance

0.3

0.10 0.05

0.0 350 400 450 500 550 600 650 700

Wavelength (nm)

0

10

20

30

40

50

Mutant target (%)

Fig. 4. Affect of the presence of mutants on IR-NEANA. Absorbance was obtained by spiking various amounts of c.2573 T 4G mutant EGFR gene into 0.1 nM wild-type EGFR gene. The final concentration of the mutant in each sample was 50%, 25%, 10%, 5%, 2%, 1%, 0.5%, 0.1%, and 0%, respectively. (A) Absorption spectra and images of IR-NEANA. (B) Dependence of coexisting mutants on absorbance.

54

B. Zou et al. / Biosensors and Bioelectronics 66 (2015) 50–54

4. Conclusions

References

We proposed a sensitive and specific DNA detection method (IR-NEANA) by bridging invasive reaction with NEANA through ligation reaction. In IR-NEANA, the read-out can be observed by naked eyes, and no expensive instrument is required. As the invasive reaction is much specific, as low as 1% mutant DNA could be accurately picked up from a large amount of wild-type DNA backgrounds. Although only the EGFR gene mutated at c.2573 T 4G was demonstrated, the successful detection of 9 tissue samples from non-small cell lung cancer FFPE specimens suggests that our proposed method has a potential in clinical application, such as gene-guided personalized medicine. The method we developed here is sensitive to 1 pM target DNA, specific to one base difference, suitable to any target sequence, and inexpensive for instrumentation. Except for the probes required for invasive reactions, IR-NEANA-based sensor is highly universal when detecting different targets, bringing a very convenient experiment set-up.

Cheng, Y., Zhang, X., Li, Z., Jiao, X., Wang, Y., Zhang, Y., 2009. Angew. Chem. Int. Ed. Engl. 48 (18), 3268–3272. Hall, J.G., Eis, P.S., Law, S.M., Reynaldo, L.P., Prudent, J.R., Marshall, D.J., Allawi, H.T., Mast, A.L., Dahlberg, J.E., Kwiatkowski, R.W., de Arruda, M., Neri, B.P., Lyamichev, V.I., 2000. Proc. Natl. Acad. Sci. USA 97 (15), 8272–8277. Hill, H.D., Mirkin, C.A., 2006. Nat. Protoc. 1 (1), 324–336. Hsieh, K., Patterson, A.S., Ferguson, B.S., Plaxco, K.W., Soh, H.T., 2012. Angew. Chem. Int. Ed. Engl. 51 (20), 4896–4900. Kiesling, T., Cox, K., Davidson, E.A., Dretchen, K., Grater, G., Hibbard, S., Lasken, R.S., Leshin, J., Skowronski, E., Danielsen, M., 2007. Nucleic Acids Res. 35 (18), e117. Lizardi, P.M., Huang, X., Zhu, Z., Bray-Ward, P., Thomas, D.C., Ward, D.C., 1998. Nat. Genet. 19 (3), 225–232. Lyamichev, V., Mast, A.L., Hall, J.G., Prudent, J.R., Kaiser, M.W., Takova, T., Kwiatkowski, R.W., Sander, T.J., de Arruda, M., Arco, D.A., Neri, B.P., Brow, M.A., 1999. Nat. Biotechnol. 17 (3), 292–296. Lyamichev, V.I., Kaiser, M.W., Lyamicheva, N.E., Vologodskii, A.V., Hall, J.G., Ma, W.P., Allawi, H.T., Neri, B.P., 2000. Biochemistry 39 (31), 9523–9532. Mi, L., Wen, Y., Pan, D., Wang, Y., Fan, C., Hu, J., 2009. Small 5 (22), 2597–2600. Mirkin, C.A., Letsinger, R.L., Mucic, R.C., Storhoff, J.J., 1996. Nature 382 (6592), 607–609. Mullis, K.B., Faloona, F.A., 1987. Methods Enzymol. 155, 335–350. Notomi, T., Okayama, H., Masubuchi, H., Yonekawa, T., Watanabe, K., Amino, N., Hase, T., 2000. Nucleic Acids Res. 28 (12), E63. Pollet, J., Janssen, K.P., Knez, K., Lammertyn, J., 2011. Small 7 (8), 1003–1006. Storhoff, J.J., Elghanian, R., Mucic, R.C., Mirkin, Chad A., Letsinger, R.L., 1998. J. Am. Chem. Soc. 120, 1959–1964. Storhoff, J.J., Lazarides, A.A., Mucic, R.C., Mirkin, C.A., Letsinger, R.L., Schatz, G.C., 2000. J. Am. Chem. Soc. 122 (19), 4640–4650. Wong, J.K., Yip, S.P., Lee, T.M., 2014. Small 10 (8), 1495–1499. Xu, W., Xie, X., Li, D., Yang, Z., Li, T., Liu, X., 2012. Small 8 (12), 1846–1850. Xu, W., Xue, X., Li, T., Zeng, H., Liu, X., 2009. Angew. Chem. Int. Ed. Engl. 48 (37), 6849–6852. Zhao, B., Yan, J., Wang, D., Ge, Z., He, S., He, D., Song, S., Fan, C., 2013. Small 9 (15), 2595–2601. Zou, B., Ma, Y., Wu, H., Zhou, G., 2011. Angew. Chem. Int. Ed. Engl. 50 (32), 7395–7398. Zou, B., Ma, Y., Zhou, G., 2013. Methods Mol. Biol. 1039, 131–137.

Acknowledgment This work was supported by Jiangsu Province’s Clinical Science & Technology Special Project (SBL201230265), State key Basic Research Program of the PRC (2014CB744501), the National Natural Science Foundation of China (Grants 31200638, 21275161 and 81373485), National Key Science & Technology Special Project (2013ZX10004103-001), China Postdoctoral Science special Foundation (2013T60938) and Jiangsu Planned Projects for Postdoctoral Research Funds (1201023C).

Appendix A. Supplementary information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.bios.2014.10.077.