Increasing the sensitivity of reverse phase protein arrays by antibody-mediated signal amplification
© Brase et al; licensee BioMed Central Ltd. 2010
Received: 18 December 2009
Accepted: 22 June 2010
Published: 22 June 2010
Reverse phase protein arrays (RPPA) emerged as a useful experimental platform to analyze biological samples in a high-throughput format. Different signal detection methods have been described to generate a quantitative readout on RPPA including the use of fluorescently labeled antibodies. Increasing the sensitivity of RPPA approaches is important since many signaling proteins or posttranslational modifications are present at a low level.
A new antibody-mediated signal amplification (AMSA) strategy relying on sequential incubation steps with fluorescently-labeled secondary antibodies reactive against each other is introduced here. The signal quantification is performed in the near-infrared range. The RPPA-based analysis of 14 endogenous proteins in seven different cell lines demonstrated a strong correlation (r = 0.89) between AMSA and standard NIR detection. Probing serial dilutions of human cancer cell lines with different primary antibodies demonstrated that the new amplification approach improved the limit of detection especially for low abundant target proteins.
Antibody-mediated signal amplification is a convenient and cost-effective approach for the robust and specific quantification of low abundant proteins on RPPAs. Contrasting other amplification approaches it allows target protein detection over a large linear range.
In recent years reverse phase protein arrays (RPPA) have proven themselves as useful experimental platform for the validation of biomarker candidate proteins in biological and clinical samples [1–6]. RPPA are considerably faster than conventional techniques such as mass spectrometry, western blotting, 2-D PAGE, and allow the analysis of hundreds of samples in parallel. In addition, measurements can be made with high accuracy and reproducibility. The basic idea of RPPA implies that all samples are spotted in parallel on solid-phase carriers. Samples can be printed either as serial dilutions or in a single concentration but as multiple replicate spots . The detection of a specific protein or a certain phosphorylation site is carried out with a single and highly specific antibody per slide, and the fraction of captured antibodies is mostly visualized with secondary antibodies. Recently, near infrared fluorescence-based detection was reported as useful for reverse phase protein microarrays [8, 9]. Routine applications involved analyzing the activation status of signaling pathways , protein profiling after RNAi-based silencing experiments , as well as of tumor biopsies [12, 13].
Critical for the outcome of RPPA-based measurements are first of all the sensitivity and specificity of primary antibodies but also of the signal detection method since samples are delivered as tiny droplets on solid phase carriers and no further separation steps are possible. In addition, proteins of interest may be expressed at a low level and their visualization requires means for signal amplification. Tyramide signal amplification (TSA) has been widely applied to increase the sensitivity of RPPAs [14–18]. However, TSA linked with a streptavidin-biotin strategy can produce unspecific signals that may interfere with the signal of interest .
Cell culture conditions and cell lysis
Human colon cancer cell lines HCT15 (CCL-225), HCT116 (CCL-247), H508 (CCL-253), SW620 (CCL-227), HT29 (HTB-38), RKO (CLR-2577), and SW480 (CCL-228) as well as the breast cancer cell lines HCC1954 (CRL-2338), MDA-MB-231 (HTB-26), BT-474 (HTB-20), SK-BR-3 (HTB-30), MCF-7 (HTB-22), and T47D (HTB-133) were obtained from American Type Culture Collection (Manassas, USA). Breast cancer cell lines were cultivated under conditions recommended by ATCC (ATCC serum and medium annotation, http://www.lgcpromochem-atcc.com). All colon cancer cell lines were cultivated in RPMI 1640 medium (ATCC) containing 10% FCS and 1% Pen/Strep (both Invitrogen, Karlsruhe, Germany). Cell lines were grown until 90% confluence and split 3 times per week. For lysis, cultured cells were incubated with trypsin (Sigma, Munich, Germany) and lysed in 1 ml M-PER (Pierce, Rockford, USA) containing 2 mM ortho-sodium-vanadate (Sigma), 10 mM NaF, and Complete mini protease inhibitor. The protein concentration was determined in duplicate by BCA Protein Assay (Pierce). Protein lysates were stored at -80°C until use.
Prostate tissue samples were collected at Martini-Clinic, Prostate Cancer Center (University Medical Center Hamburg-Eppendorf, Germany). Samples were acquired during radical prostatectomy and sectioned with a microtome. One slice of prostate tumor tissue was processed per experiment. Mouse liver tissue and collected prostate tissue were lysed with modified T-PER buffer (Pierce). 10 μl of lysis buffer were used per mg of tissue sample. Samples were homogenized 4 minutes with a tissuelyser (Qiagen, Hilden, Germany). Cell debris was subsequently pelleted (12 minutes, 13000 rpm) and the supernatant was passed through a Qiashredder tube (Qiagen).
Printing of protein microarrays
Protein lysates and spike-in experiments with recombinant His-tagged JNK (Invitrogen) were serially diluted with protein lysis buffer. To prepare samples for inkjet-spotting using the Sprint (ArrayJet, Roslin, Scotland) an equal volume of spotting-buffer (50% (v/v) Glycerol and 0.05% (w/v) Triton × 100 in ddH20) was added to each sample resulting in a final glycerol concentration of 25%.
A pin tool spotter, the 2470 Arrayer (Aushon, Billerica, MA, USA), was employed to print serial dilutions of cell line lysates (SW480, HCT15, HCT116, SW620, HT29, HCC1954, MDA-MB-231, BT-474, SK-BR-3, MCF-7 and T47D). Dilution series comprised 14 steps and were produced by diluting with lysis buffer by 33% per step. Samples were printed directly after adding Tween20 to a final concentration of 0.05% (v/v).
Samples were always printed onto nitrocellulose coated glass-slides (Grace Biolabs, Bent, OR, USA). Each spot corresponded to a final drop volume of 0.3 nl (pin tool spotter) or 0.6 nl (non-contact inkjet printer). The spot-to-spot distance was set to 320 μm and the spot diameter was 140 μm. Slides were stored at 4°C and used within a week.
Near infrared target protein detection on microarrays
Slides were blocked with a mixture of 33% Odyssey blocking buffer (LI-COR, Lincoln, USA), 1% BSA, and 0,02% NP40 in PBS overnight at 4°C. Primary antibodies were diluted in a buffer with background-reducing components (Dako, Glostrup, Denmark). Slides were incubated for 2 h with primary antibodies and subsequently immersed in wash buffer (1x PBS, 0.02% NP40 and 0.02% SDS) four times for 5 minutes. Next, slides were incubated with Alexa680-conjugated secondary antibody (Invitrogen, dilution 1:8,000) for 30 minutes. Washing was performed as described above. All washing and incubation steps were carried out at RT with gentle shaking. Finally, slides were rinsed in water and air-dried at room temperature.
The following detection antibodies were employed for western blotting and RPPA-based analysis: JNK (BD Biosciences, San Jose, USA), CDK4, Cyclin D1, GSK, PP2AA, PP2AB, PDK1, beta-Catenin, RB, Smad 2/3, SRC (Cell Signaling Technologies, Danvers, USA), ERK 1/2, NF-κB, PCNA, Stat3 (Santa Cruz, Santa Cruz, USA), KLK3 (Sigma), JNK, PKC, PLCγ (Abcam, Cambridge, USA), and pERK (R&D, Minneapolis, USA).
Antibody-mediated signal amplification on microarrays
Incubation with primary antibodies was carried out as described in the previous section. All working steps for antibody-mediated signal amplification were integrated into an automated procedure. A robotics protocol was established to increase and simplify the throughput. The 96-channel head robot Biomek FXP (Beckmann Coulter, Harbor Boulevard, USA) was used for all assay steps thus reducing hands-on time and minimizing experimental variation.
Anti-rabbit Alexa680-labeled (raised in goat) and anti-goat Alexa680-labeled (raised in rabbit) antibodies (both Invitrogen) were applied consecutively in a total of four cycles (dilution 1:8,000). Anti-mouse Alexa680 labeled (raised in goat) antibody was used in the first cycle for the detection of primary antibodies raised in mouse. Secondary antibodies employed for signal amplification were derived from commercial sources and of highest purity. Slides were washed four times for 5 minutes between automated incubation steps. Each secondary antibody was incubated for 30 minutes.
Western Blot detection
SDS-Page and Western immunoblotting analyses were performed using 5 to 20 μg protein lysate. Standard near infrared detection was applied as described elsewhere . The protocol for antibody-mediated signal amplification was adapted to a larger volume. Tyramide signal amplification was carried out in parallel to compare directly between both amplification methods. TSA Kit #11 (Invitrogen) was employed according to the manufacturer's instructions. Endogenous peroxidase was treated with 3% H2O2 for one hour. A 1:250 dilution of the HRP conjugated secondary antibody was applied. Peroxidase catalyzed signal amplification was carried out for 10 minutes and Streptavidin-Alexa680 (Invitrogen) was incubated for 20 minutes.
Slides were scanned with the Odyssey NIR scanner (LI-COR). Image analysis was carried out with GenePix-Pro 5.1 (Axon Instruments, Sunnyvale, USA). Spot intensity was corrected for background and noise due to unspecific antibody binding. Mean and SD were calculated for all replicates. Data sets were analyzed within the statistical computing environment R.
The detection limit was determined as follows: First, the signal intensity mean of a certain dilution step has to be greater than the sum of 2 x standard deviation (SD) of S0 and the mean of S0 with S0 corresponding to the lowest concentration of a dilution series. Thus, S0 was considered as background signal  ensuring a significant difference of signals from background. Second, in a dilution series signals must increase continuously with increasing proteins concentrations. The lowest concentration of a certain serial dilution complying with both requirements is defined as limit of detection (LOD). To sum up, the detection limit corresponded to the smallest concentration of a certain serial dilution that can be distinguished from the experimental background as well as from the next higher concentration. A detection limit is a specific measure for a specific detection antibody, detection method, and a specific sample. The curve fitting model and the normalization approach are described in detail in Additional file 1.
Optimizing AMSA to a robust and specific amplification method
Important experimental parameters such as blocking and wash buffer composition, primary and secondary antibody concentration, and the effect of background reducing agents were analyzed systematically to establish a new amplification approach for RPPA. For example, a spike-in approach with recombinant JNK1 protein was employed to assess the detection limit of the new amplification method. Beforehand, the specificity of all antibodies used in the amplifications steps was confirmed by probing suitable controls printed on microarrays such as human immunoglobulins and recombinant JNK protein. Signal detection omitting the target specific anti-JNK1 antibody yielded no signals (Additional file 1, figure 1a). Data also demonstrated that species-specific antibodies used as part of the amplification procedure did not cross react with human immunoglobulins potentially present in clinical samples of human origin. On the other hand, the presence of human immunoglobulins was confirmed by probing a replicate slide with anti-human IgG antibody (Additional file 1, figure 1b), and the specificity of the JNK1-detection was demonstrated as well (Additional file 1, figure 1c).
Aiming at a more comprehensive validation for a larger set of target proteins 14 highly expressed signaling proteins (CDK4, CyclinD1, ERK1/2, GSK, NF-κB, PCNA, PKC, PLCy, PP2AA, PP2AB, RB, SMAD2/3, SRC, STAT3) were chosen. Western blot analysis confirmed that all proteins are indeed expressed in the human cancer cell lines (Figure 3b). The samples were also printed on nitrocellulose-coated slides and probed with the same set of primary antibodies to compare standard NIR detection with antibody-mediated signal amplification for RPPA. The subsequent data analysis revealed a strong correlation between both data sets validating AMSA as robust and reproducible (Pearson r = 0.89, p < 2.2-16) (Figure 3c).
Comparison of AMSA to NIR-standard detection for the detection of low abundance proteins
Reverse phase protein microarray techniques allow for a highly parallel and quantitative analysis of signal transduction processes requiring only minute amounts of sample material. However, the protein of interest is frequently expressed at a low level and its detection requires highly sensitive methods. For this reason, a new NIR fluorescence-based amplification approach was developed [8, 9]. To our knowledge, this is the first description of an amplification system using successive incubation steps with two species-specific secondary antibodies on protein microarrays.
Analyzing the expression level of 14 different proteins with moderate to high abundance in seven different human cancer cell lines by RPPA revealed that AMSA results corresponded to those obtained with standard NIR detection. Thus, the high correlation coefficient indicated that the signal intensity gain mediated by AMSA is reproducible. Furthermore, side-effects such as background signal intensity gain were analyzed with respect to the number of amplification cycles. Signal-to-noise ratios improved considerably after three amplification cycles, and four rounds of amplification were found to deliver the best result. Thus, after four amplification cycles the target specific signal intensity gain was higher than the increase of the local background. Moreover, a direct comparison between AMSA and standard NIR confirmed a lower detection limit for a specific protein spiked into a cell lysate as well as for endogenous proteins expressed at a low level.
Different types of signal detection strategies were reported for protein microarray applications. Colorimetric, chemiluminescence, fluorescence detection techniques, linear waveguide technologies , and rolling circle amplification  were so far successfully adapted to protein microarrays. Tyramide signal amplification is one of the most frequently used techniques to increase the sensitivity of reverse phase arrays. TSA is based on the catalyzed deposition of tyramide in close proximity of the secondary HRP-linked antibody [24, 25]. The high velocity of enzymatic processes was suspected to cause signal saturation effects  and observed as sigmoidal curves in serial dilutions limiting the signal detection range on RPPA [7, 27]. However, time-resolved measurements involving systems biology applications benefit from a linear correlation between sample concentration and signal intensity. Besides that, insufficient specificity has been reported for tyramide signal amplification strategies [19, 28]. Ambroz et al. suggested that identical detection chemistries should be used for the RPPA-based analysis and for Western blot based antibody validation measures to reduce the risk of experimental artifacts . Western blot detection indicated that indeed the use of a biotin/avidin-based signal amplification strategy produces unspecific bands presumably due to insufficient blocking of endogenous biotin-modified proteins. Biotin-modifications were identified on histones and presumably play a role in epigenetic mechanisms . However, the use of AMSA circumvents the optimization of biotin-blocking procedures since no unspecific background signals were observed.
We have shown that antibody-mediated signal amplification is a powerful tool for the specific and sensitive detection of signals on reverse phase protein microarrays over a large linear range. AMSA is easy to implement and could in principle be also adapted to other immunoassay formats. In addition, assay costs are low compared to commercial amplification strategies since only low concentrations of secondary antibodies are required as reagents.
antibody-mediated signal amplification
infrared-based protein arrays with quantitative readout
reverse phase protein array
tyramide signal amplification
We thank Maike Wosch and Annika Bittmann for their excellent technical assistance and Johanna Fehling for graphical support. We thank Antonio R Garetta for the mouse liver tissue samples. This work was supported by the German Federal Ministry for Education and Research in the framework of the Program for Medical Genome Research (grants 01GS0890, 01GS08189, 01GS08107, and 01GS0864), the Program for Medical Systems Biology (grant 0315396B), as well as the Helmholtz Systems Biology Initiative (SBCancer). The authors are responsible for the contents of this publication. This paper is dedicated to the memory of Professor Annemarie Poustka, who was the founder and head of the Division Molecular Genome Analysis at the DKFZ. She was an inspiring scientist and a wonderful person.
- Paweletz CP, Charboneau L, Bichsel VE, Simone NL, Chen T, Gillespie JW, Emmert-Buck MR, Roth MJ, Petricoin IE, Liotta LA: Reverse phase protein microarrays which capture disease progression show activation of pro-survival pathways at the cancer invasion front. Oncogene 2001, 20: 1981–1989. 10.1038/sj.onc.1204265PubMedView Article
- Nishizuka S, Charboneau L, Young L, Major S, Reinhold WC, Waltham M, Kouros-Mehr H, Bussey KJ, Lee JK, Espina V, et al.: Proteomic profiling of the NCI-60 cancer cell lines using new high-density reverse-phase lysate microarrays. Proc Natl Acad Sci USA 2003, 100: 14229–14234. 10.1073/pnas.2331323100PubMed CentralPubMedView Article
- Sheehan KM, Gulmann C, Eichler GS, Weinstein JN, Barrett HL, Kay EW, Conroy RM, Liotta LA, Petricoin EF: Signal pathway profiling of epithelial and stromal compartments of colonic carcinoma reveals epithelial-mesenchymal transition. Oncogene 2008, 27: 323–331. 10.1038/sj.onc.1210647PubMedView Article
- Espina V, Mehta AI, Winters ME, Calvert V, Wulfkuhle J, Petricoin EF, Liotta LA: Protein microarrays: molecular profiling technologies for clinical specimens. Proteomics 2003, 3: 2091–2100. 10.1002/pmic.200300592PubMedView Article
- Boyd ZS, Wu QJ, O'Brien C, Spoerke J, Savage H, Fielder PJ, Amler L, Yan Y, Lackner MR: Proteomic analysis of breast cancer molecular subtypes and biomarkers of response to targeted kinase inhibitors using reverse-phase protein microarrays. Mol Cancer Ther 2008, 7: 3695–3706. 10.1158/1535-7163.MCT-08-0810PubMedView Article
- Wulfkuhle JD, Aquino JA, Calvert VS, Fishman DA, Coukos G, Liotta LA, Petricoin EF: Signal pathway profiling of ovarian cancer from human tissue specimens using reverse-phase protein microarrays. Proteomics 2003, 3: 2085–2090. 10.1002/pmic.200300591PubMedView Article
- Zhang L, Wei Q, Mao L, Liu W, Mills GB, Coombes K: Serial dilution curve: a new method for analysis of reverse phase protein array data. Bioinformatics 2009.View Article
- Loebke C, Sueltmann H, Schmidt C, Henjes F, Wiemann S, Poustka A, Korf U: Infrared-based protein detection arrays for quantitative proteomics. Proteomics 2007, 7: 558–564. 10.1002/pmic.200600757PubMedView Article
- Calvert VS, Tang Y, Boveia V, Wulfkuhle J, Schutz-Geschwender A, Olive DM, Liotta LA, III EFP: Development of Multiplexed Protein Profiling and Detection Using Near Infrared Detection of Reverse-Phase Protein Microarrays. Clinical Proteomics 2004, 1: 81–89. 10.1385/CP:1:1:081View Article
- Lobke C, Laible M, Rappl C, Ruschhaupt M, Sahin O, Arlt D, Wiemann S, Poustka A, Sultmann H, Korf U: Contact spotting of protein microarrays coupled with spike-in of normalizer protein permits time-resolved analysis of ERBB receptor signaling. Proteomics 2008, 8: 1586–1594. 10.1002/pmic.200700733PubMedView Article
- Sahin O, Lobke C, Korf U, Appelhans H, Sultmann H, Poustka A, Wiemann S, Arlt D: Combinatorial RNAi for quantitative protein network analysis. Proc Natl Acad Sci USA 2007, 104: 6579–6584. 10.1073/pnas.0606827104PubMed CentralPubMedView Article
- Haller F, Lobke C, Ruschhaupt M, Schulten HJ, Schwager S, Gunawan B, Armbrust T, Langer C, Ramadori G, Sultmann H, et al.: Increased KIT signalling with up-regulation of cyclin D correlates to accelerated proliferation and shorter disease-free survival in gastrointestinal stromal tumours (GISTs) with KIT exon 11 deletions. J Pathol 2008, 216: 225–235. 10.1002/path.2402PubMedView Article
- Haller F, Lobke C, Ruschhaupt M, Cameron S, Schulten HJ, Schwager S, von Heydebreck A, Gunawan B, Langer C, Ramadori G, et al.: Loss of 9p leads to p16INK4A down-regulation and enables RB/E2F1-dependent cell cycle promotion in gastrointestinal stromal tumours (GISTs). J Pathol 2008, 215: 253–262. 10.1002/path.2352PubMedView Article
- Grote T, Siwak DR, Fritsche HA, Joy C, Mills GB, Simeone D, Whitcomb DC, Logsdon CD: Validation of reverse phase protein array for practical screening of potential biomarkers in serum and plasma: Accurate detection of CA19–9 levels in pancreatic cancer. Proteomics 2008,8(15):3051–60. 10.1002/pmic.200700951PubMed CentralPubMedView Article
- Tibes R, Qiu Y, Lu Y, Hennessy B, Andreeff M, Mills GB, Kornblau SM: Reverse phase protein array: validation of a novel proteomic technology and utility for analysis of primary leukemia specimens and hematopoietic stem cells. Mol Cancer Ther 2006, 5: 2512–2521. 10.1158/1535-7163.MCT-06-0334PubMedView Article
- VanMeter AJ, Rodriguez AS, Bowman ED, Jen J, Harris CC, Deng J, Calvert VS, Silvestri A, Fredolini C, Chandhoke V, et al.: Laser capture microdissection and protein microarray analysis of human non-small cell lung cancer: differential epidermal growth factor receptor (EGPR) phosphorylation events associated with mutated EGFR compared with wild type. Mol Cell Proteomics 2008, 7: 1902–1924. 10.1074/mcp.M800204-MCP200PubMed CentralPubMedView Article
- Grubb RL, Calvert VS, Wulkuhle JD, Paweletz CP, Linehan WM, Phillips JL, Chuaqui R, Valasco A, Gillespie J, Emmert-Buck M, et al.: Signal pathway profiling of prostate cancer using reverse phase protein arrays. Proteomics 2003, 3: 2142–2146. 10.1002/pmic.200300598PubMedView Article
- Dupuy L, Gauthier C, Durand G, Musnier A, Heitzler D, Herledan A, Sakanyan V, Crepieux P, Reiter E: A highly sensitive near-infrared fluorescent detection method to analyse signalling pathways by reverse-phase protein array. Proteomics 2009.
- Ambroz KL, Zhang Y, Schutz-Geschwender A, Olive DM: Blocking and detection chemistries affect antibody performance on reverse phase protein arrays. Proteomics 2008, 8: 2379–2383. 10.1002/pmic.200700676PubMedView Article
- Korf U, Derdak S, Tresch A, Henjes F, Schumacher S, Schmidt C, Hahn B, Lehmann WD, Poustka A, Beissbarth T, Klingmuller U: Quantitative protein microarrays for time-resolved measurements of protein phosphorylation. Proteomics 2008, 8: 4603–4612. 10.1002/pmic.200800112PubMedView Article
- Sias PE, Kotts CE, Vetterlein D, Shepard M, Wong WL: ELISA for quantitation of the extracellular domain of p185HER2 in biological fluids. J Immunol Methods 1990, 132: 73–80. 10.1016/0022-1759(90)90400-PPubMedView Article
- Pawlak M, Schick E, Bopp MA, Schneider MJ, Oroszlan P, Ehrat M: Zeptosens' protein microarrays: a novel high performance microarray platform for low abundance protein analysis. Proteomics 2002, 2: 383–393. 10.1002/1615-9861(200204)2:4<383::AID-PROT383>3.0.CO;2-EPubMedView Article
- Schweitzer B, Wiltshire S, Lambert J, O'Malley S, Kukanskis K, Zhu Z, Kingsmore SF, Lizardi PM, Ward DC: Inaugural article: immunoassays with rolling circle DNA amplification: a versatile platform for ultrasensitive antigen detection. Proc Natl Acad Sci USA 2000, 97: 10113–10119. 10.1073/pnas.170237197PubMed CentralPubMedView Article
- Hunyady B, Krempels K, Harta G, Mezey E: Immunohistochemical signal amplification by catalyzed reporter deposition and its application in double immunostaining. J Histochem Cytochem 1996, 44: 1353–1362.PubMedView Article
- Bobrow MN, Harris TD, Shaughnessy KJ, Litt GJ: Catalyzed reporter deposition, a novel method of signal amplification. Application to immunoassays. J Immunol Methods 1989, 125: 279–285. 10.1016/0022-1759(89)90104-XPubMedView Article
- Ramaswamy A, Lin E, Chen I, Mitra R, Morrisett J, Coombes K, Ju Z, Kapoor M: Application of protein lysate microarrays to molecular marker verification and quantification. Proteome Sci 2005, 3: 9. 10.1186/1477-5956-3-9PubMed CentralPubMedView Article
- Hu J, He X, Baggerly KA, Coombes KR, Hennessy BT, Mills GB: Non-parametric quantification of protein lysate arrays. Bioinformatics 2007, 23: 1986–1994. 10.1093/bioinformatics/btm283PubMedView Article
- Aoki H, Iwado E, Eller MS, Kondo Y, Fujiwara K, Li GZ, Hess KR, Siwak DR, Sawaya R, Mills GB, et al.: Telomere 3' overhang-specific DNA oligonucleotides induce autophagy in malignant glioma cells. FASEB J 2007, 21: 2918–2930. 10.1096/fj.06-6941comPubMedView Article
- Kothapalli N, Camporeale G, Kueh A, Chew YC, Oommen AM, Griffin JB, Zempleni J: Biological functions of biotinylated histones. J Nutr Biochem 2005, 16: 446–448. 10.1016/j.jnutbio.2005.03.025PubMed CentralPubMedView Article
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.