Optimized high gradient magnetic separation for isolation of Plasmodium-infected red blood cells
© Bhakdi et al; licensee BioMed Central Ltd. 2010
Received: 9 July 2009
Accepted: 2 February 2010
Published: 2 February 2010
Highly purified infected red blood cells (irbc), or highly synchronized parasite cultures, are regularly required in malaria research. Conventional isolation and synchronization rely on density and osmotic fragility of irbc, respectively. High gradient magnetic separation (HGMS) offers an alternative based on intrinsic magnetic properties of irbc, avoiding exposure to chemicals and osmotic stress. Successful HGMS concentration in malaria research was previously reported using polymer coated columns, while HGMS depletion has not been described yet. This study presents a new approach to both HGMS concentration and depletion in malaria research, rendering polymer coating unnecessary.
A dipole magnet generating a strong homogenous field was custom assembled. Polypropylene syringes were fitted with one-way stopcocks and filled with stainless steel wool. Rbc from Plasmodium falciparum cultures were resuspended in density and viscosity optimized HGMS buffers and HGMS processed. Purification and depletion results were analysed by flow cytometer and light microscopy. Viability was evaluated by calculating the infection rate after re-culturing of isolates.
In HGMS concentration, purity of irbc isolates from asynchronous cultures consistently ranged from 94.8% to 98.4% (mean 95.7%). With further optimization, over 90% of isolated irbc contained segmented schizonts. Processing time was less than 45 min. Reinfection rates ranged from 21.0% to 56.4%. In HGMS depletion, results were comparable to treatment with sorbitol, as demonstrated by essentially identical development of cultures.
The novel HGMS concentration procedure achieves high purities of segmented stage irbc from standard asynchronous cultures, and is the first HGMS depletion alternative to sorbitol lysis. It represents a simple and highly efficient alternative to conventional irbc concentration and synchronization methods.
Malaria remains one of the world's major health burdens. With 2.5 billion people at risk it affects an estimated 500 million people annually, causing one to three million deaths, the majority of which occurs in children under five years of age. Improved methods facilitating research in the field are urgently needed [1, 2].
Isolation of infected red blood cells (irbc) is a crucial step in basic and applied malaria research. For the past three decades, isolation has been performed mostly by Percoll® density gradient separation, exploiting the fact that density of irbc decreases with parasite maturation . A further refinement of this method are hypertonic, discontinuous Percoll®-sorbitol gradients, where particular fractions of irbc can be obtained. Hypertonicity causes cell shrinkage of rbc, while irbc swell back due to an influx of sorbitol through new permeability pathways. This increases the density gaps between the different developmental stages and allows better separation than in pure Percoll® gradients . Purification results, however, depend on a variety of factors, including individual research experience. Gelatin sedimentation is used as an alternative concentration method, however, it is useful only for parasite strains exhibiting knobs .
Frequently, not only highly purified but also stage-synchronized parasite cultures and isolates are required. Synchronization of cultures is performed by isotonic sorbitol lysis of late-stage irbc, as described 30 years ago . Time-consuming synchronization cycles by repeated sorbitol lysis and/or Percoll® isolation are required to obtain synchronized and pure irbc suitable for downstream applications [6–8]. While sorbitol selectively lyses late-stage irbc, it imposes sub-lytic osmotic stress in younger stage irbc and likely enters these cells . Whether or not exposure of irbc to synthetic chemicals and osmotic stress, respectively, has unwanted consequences on parasites remains an open question.
In principle, high gradient magnetic separation (HGMS) offers a way to concentrate or deplete malaria irbc from suspensions, relying solely on their intrinsic magnetic properties. Particularly late-stage irbc are known to behave as paramagnetic particles . In a paramagnetic particle, magnetic poles are induced only when exposing the particle to a magnetic field, the removal of which leads to immediate de-magnetization. Due to the very small distance separating the particle's respective north- and south-poles, very high magnetic field gradients are required to create a net magnetic force, which is able to attract or repel the particle. Such gradients are generated by placing thin filamentous or spherous ferromagnetic material as a matrix into a strong homogenous magnetic field, which is usually provided by rare-earth dipole magnets or electromagnets. With this technology, magnetic gradients up to 100 Tesla/cm can be created at the surface of the matrix .
Paramagnetism in malaria irbc results from the hemoglobin catabolism of intra-erythrocytic malaria parasites. Free haem as a toxic by-product is de-toxified by polymerization and by oxidation of the molecule's central iron atom . Oxidized iron [Fe+3] carries five unpaired electrons in its d-orbitals, rendering the molecule and the resulting polymer paramagnetic [10, 13, 14]. Deposition and accumulation of polymerized haem (haemozoin) in the parasite's food vacuole result in a continuously increasing magnetic susceptibility of the irbc .
Successful, but not highly efficient HGMS of late-stage irbc from malaria cultures was first described in 1981 . Later, commercially available, polymer coated HGMS columns were shown to offer improved results [15–18]. Recently, successful synchronization was demonstrated with polymer coated HGMS columns. However, sorbitol-pretreatment of Plasmodium cultures was essential . Generally, the use of HGMS in malaria research is still hampered by limited column capacity, inconsistent separation purities and high costs.
Firstly, this study presents a buffer-optimized HGMS concentration system for irbc, as an alternative to Percoll® density gradient separation. Secondly, a HGMS-BSA gradient system is introduced for HGMS depletion of late-stage irbc, as an alternative to Sorbitol lysis.
Buffer-optimized HGMS was developed by identifying the three key variable parameters relevant to HGMS for cell separation. Briefly, these are derived as follows:
where is the magnetic force, the drag force and the gravitational force acting on the particle.
where m is the mass of the particle and the gravitational acceleration.
where d p is the density of the particle, d f the density of the carrier fluid, the gravitational acceleration, r the radius of the particle and η the viscosity of the carrier fluid .
Analysing equation (1)-(5), it becomes obvious that apart from the constant intrinsic parameters of the particle, there are only four extrinsic parameters governing the HGMS system. These are a) the gravitational acceleration, b) the velocity of the carrier fluid, c) the viscosity of the carrier fluid, and d) the density of the carrier fluid.
It was hypothesized that in an ideal buffer-optimized HGMS system, the density of the carrier fluid can be adjusted to the density of rbc, which would render the influence of the gravitational acceleration on the particle negligible, setting the particle sedimentation velocity close to zero. In this case, particle velocity would be perfectly controllable solely by adjusting the velocity of the carrier fluid. Additionally, the density-concurrent increase in viscosity η would favourably support laminar flow conditions in the column.
To test this hypothesis, commonly available raw materials were employed to perform selective, buffer-optimized HGMS of P. falciparum late-stage irbc from asynchronous, standard routine malaria cultures, avoiding exposure to anything but incomplete malaria culture medium (RPMI 1640) or isotonic sucrose solution, substituted with either BSA or gelatine, respectively. A concentration protocol was developed as an alternative to Percoll® separation and HGMS with polymer-coated columns. We show that with further refinement, optimized HGMS results in unprecedented purities of segmented-stage irbc from standard asynchronous cultures.
The second protocol demonstrates efficient depletion of late-stage irbc from Plasmodium cultures. Here, a BSA density gradient was introduced within the HGMS separation column, which prevented uncontrolled sedimentation of rbc during incubation within the matrix of the column. It is demonstrated that HGMS depletion alone, without HGMS concentration, can be used as an alternative to Sorbitol lysis for culture synchronization, potentially further widening the application of HGMS in malaria reserach.
The P. falciparum TM 267 laboratory strain was kindly provided by the Hospital of Tropical Diseases, Faculty of Tropical Medicine, Mahidol University, Bangkok. Parasites were passaged in human rbc blood group O in RPMI 1640 medium containing HEPES 5.94 g/l, glucose 1 g/l, hypoxanthine 50 mg/l, 5% sodium bicarbonate and 10% human AB serum at 5% haematocrit in an incubator under 5% CO2 atmosphere at 37°C.
Hydroethidine (HE) and acridine orange (AO) were purchased from Invitrogen, Molecular Probes (Eugene, OR). Sheath fluid for flow cytometry was from Becton Dickinson Bioscience (BDB) (San Jose, CA), all other reagents were from Sigma-Aldrich (St. Louis, MO). Human blood for parasite cultures was from volunteers in our laboratory and human AB serum from the Blood Bank of Siriraj Hospital, Bangkok.
Depletion of late-stage irbc by sorbitol lysis
Sorbitol lysis was performed as described previously . Briefly, 50 μl of packed cells from P. falciparum culture where re-suspended in 1 ml of aqueous 5% D-sorbitol solution and incubated at 37°C for 15 min, followed by centrifugation at 900 g for 5 min. The supernatant and the dark layer of free late-stage parasites from lysed irbc were discarded. The remaining cells were washed once in 10 ml RPMI 1640 and re-cultured.
HGMS apparatus, columns and buffer
A dipole magnet was constructed as described previously [11, 14]. Briefly, two 50 × 30 × 12 mm neodymium magnets (Neotexx, Berlin, Germany) were mounted on a custom made, U-shaped iron-yoke, generating a uniform field strength of a minimum of 0.7 Tesla between the poles. For HGMS concentration, a 3 ml polypropylene syringe (Terumo, Ayutthaya, Thailand) was filled with 1 g of stainless steel wool of 30-50 μm diameter (Oscar Weil, Lahr, Germany) and was fitted with a one-way stopcock (Value Plastics, Fort Collins, CO). For HGMS depletion, a 5 ml syringe was filled with 4 g of stainless steel wool. HGMS buffers were isotonic sucrose with 0.75% gelatin (concentration buffer) or RPMI 1640 with 0.2% and 1% BSA (depletion buffers) (all chemicals from Sigma-Aldrich, St. Louis, MO). Columns were autoclaved for re-culturing experiments.
HGMS protocol for concentration of irbc
For isolation of late-stage irbc, the column was filled with concentration buffer in upright position to evacuate air by upward displacement. After removing remaining air bubbles by gentle finger tapping, a 20G/1-inch injection needle was connected to the stopcock. The column was placed between the poles of the dipole magnet and equilibrated for 5 min. Different volumes of cultures containing malaria irbc were centrifuged at 800 g for 5 min, washed in RPMI 1640, resuspended in concentration buffer at a final haematocrit of 1-5%, and oxygenated against ambient air for 5 min. After oxygenation of rbc was completed, the stopcock of the HGMS column was opened and the rbc suspension applied. Without interrupting the flow, the column was rinsed with 45 ml of concentration buffer. After this, the stopcock was closed and the column was removed from the magnet. The needle was removed from the stopcock and the column retrogradely flushed with 12 ml of PBS or isotonic sucrose solution. The eluate containing isolated late-stage irbc was collected, centrifuged at 800 g for 5 min, washed once with RPMI 1640 analysed and re-cultured.
HGMS protocol for depletion of irbc
For depletion of late-stage irbc, the column was filled with RPMI 1640 containing 1% BSA in upright position to evacuate air by upward displacement. After removing remaining air bubbles by gentle finger tapping, a 23G/1 inch injection needle was connected to the stopcock. The column was placed between the poles of the dipole magnet and equilibrated for 5 min. Different volumes of cultures containing malaria irbc were centrifuged at 800 g for 5 min, washed in RPMI 1640, resuspended in RPMI 1640 containing 0.2% BSA at a final haematocrit of 3.3%, and oxygenated against ambient air for 5 min. The difference in BSA concentration created a density gradient sufficient to stop uncontrolled sedimentation of rbc within the matrix. After oxygenation of rbc was completed, the stopcock was opened and the rbc suspension applied to the top of the column. When the rbc suspension had completely entered the matrix, the flow was stopped by closing the stopcock. The column was left between the poles of the dipole magnet for 30 min, which in pilot experiments was determined to be the optimal incubation time. After incubation, the column was rinsed with 8 ml of RPMI 1640/0.2% BSA and the flow-through collected, centrifuged at 800 g for 5 min, washed once and re-cultured.
Analysis of purity and synchronization
Purity of isolates and synchronization of re-culture was analysed by Giemsa stained blood smears and by flow cytometer (BD FacsCalibur). For microscopic counting, stages were defined as follows:
ring-stage trophozoites: clearly visible ring-shaped cytoplasm, and one to two chromatin dots
young trophozoites: small solid cytoplasm (occupying the lesser fraction of the erythrocyte's cytoplasm), one nucleus and little, light-brown pigment
late trophozoites: large solid cytoplasm (occupying the larger fraction of the erythrocyte's cytoplasm), one nucleus and solid, dark-brown pigment
schizonts: parasites with two or more nuclei
For flow cytometry, staining was performed with HE or AO, as described previously. For the first, 50 μl of P. falciparum culture with a haematocrit of 5% were mixed with HE at a final concentration of 5 μg/ml in a final volume of 55 μl, followed by incubation in the dark at 37°C for 30 min. 500 μl of PBS were added and data acquired in the flow cytometer. Logarithmic orange fluorescence was observed through a 585/42 nm band pass filter in the FL-2 channel and compensation set to 0 . For the second, 5 μl of P. falciparum culture with a hematocrit of 5% were mixed with AO at a final concentration of 0.5 μg/ml in a final volume of 200 μl. The mixture was incubated in the dark at room temperature for 5 min. 500 μl of PBS were added and data acquired in the flow cytometer. Logarithmic green was detected using a 530/30 nm bandpass filter and logarithmic red fluorescence using a 630 nm longpass filter in the FL-1 channel and FL-3 channel, respectively. Compensation was set to 0 .
Concentration of late-stage irbc from cultures by buffer-optimized HGMS
Yield and purity of isolates in six independent experiments
[total no. of irbc purified]
Purity irbc/total cells
Depletion of trophozoites and schizonts from cultures by buffer-optimized HGMS
HGMS depleted fraction:
Rings+early trophozoites [%]
Late trophozoites + schizonts [%]
Rings+early trophozoites [%]
Late trophozoites + schizonts
Comparison of cultures after HGMS depletion and sorbitol lysis
After HGMS depletion or sorbitol lysis, for each method identical numbers of ring-stage irbc were re-cultured. Initial parasitaemia was adjusted to 0.5-1% with fresh rbc ("day 0"). For example, 500 million rbc with a differential parasitaemia of 4.0% ring and young trophozoite irbc and 4.6% late trophozoite and schizont irbc were processed by HGMS-depletion over one column or by sorbitol lysis. The flow-through or the remaning cells of sorbitol lysis, respectively, were washed and resuspended in complete culture medium. For one 25 ml cell culture flask, 250 million rbc with a total of 10 million ring-stage irbc were re-cultured by adding 75 μl of fresh rbc which resulted in a final ring-stage parasitaemia of 1.0%.
Although irbc concentration by Percoll® gradient centrifugation, gelatin sedimentation and depletion by sorbitol lysis have remained the gold standards for separation of late-stage irbc for the last decades, all three methods still harbour certain disadvantages. Separation results are not always consistent, depending on a variety of factors including the individual experience of the investigator, and gelatin sedimentation can be used only for parasite strains exhibiting knobs.
HGMS represents an alternative method for isolation of malaria irbc. In theory, irbc concentration as well as irbc depletion should be achievable. As to the latter, no reports are available. As to the first, after the initial report of Paul et al in 1981 , few other studies focussed on the subject. Employing commercially available polymer-coated HGMS columns (MACS®), Staalsohe et al  and Uhlemann et al , reported separation purities of irbc of 70-90%, with purities over 90% only achievable through the use of a second column. Data for maximum column capacity are not given. Another study describes successful concentration of trophozoits, schizonts and gametocytes from different malaria species, with purities up to 98% for P. falciparum . Again, total column capacities are not documented. Purities of irbc of up to 98% were achieved using polymer coated HGMS columns for concentration of Plasmodium berghei irbc from mice . A recently published report employed HGMS for synchronization of P. falciparum cultures. However, pre-treatment of cultures by sorbitol lysis was essential for synchronization, and overall purities of HGMS concentration were rather variable, ranging from 13.60% to 85.28% . Other publications report the use of HGMS for purification of Plasmodium irbc, but rely on the protocols described in the studies mentioned above, and do not focus on the subject of HGMS as such.
This study presents a detailed report of new approaches to HGMS concentration and depletion, which were developed for malaria research.
For HGMS concentration, the introduction of density and viscosity optimized buffers renders polymer coating of the HGMS matrix unnecessary and regularly achieves irbc purities of over 95% from initial parasitaemias as low as 1.7%.
For HGMS depletion of late-stage irbc, the protocol offers a first viable alternative to sorbitol lysis for late-stage depletion of malaria cultures. For this, however, some limitations and handling challenges should be considered to help researchers decide under which circumstances it might be the method of choice:
First of all, capacities of columns are limited to a certain maximum number of cells. In our setting, a total of 22 million late-stage irbc was determined as capacity limit of the columns. In a conventional culture with about 1 to 2.5% late-stage parasitaemia and a total parasitaemia of 5%, this would correspond to loading roughly 109 total rbc (100 μl packed cells) onto the column, which in our laboratory is equivalent to a culture from a 25 ml flask. For this case, the processing time of 50 min is similar to sorbitol lysis. Upscaling of columns is under way and will be able to offer larger capacities in the future.
Comparison between existing irbc concentration and depletion methods and buffer optimized HGMS
Cost per sampleA
Special equipment required
Percoll ® :
Magnet, HGMS columns
Buffer optimized HGMS:
Magnet, HGMS columns
Buffer-optimized HGMS has the potential to become a useful tool in malaria research. When compared to conventional methods and to commercially available HGMS systems, the high quality of irbc purifications, ease of handling and low cost should render buffer-optimized HGMS attractive for malaria researchers in a wide range of fields.
List of abbreviations
bovine serum albumin
high gradient magnetic separation
infected red blood cells
red blood cells.
The authors acknowledge the support given by the staff of the Center of Excellence for Flow Cytometry, Office for Research and Development, Siriraj Hospital, and the staff of the Department of Pathobiology, Faculty of Science; Mahidol University. Many thanks to Sucharit Bhakdi for fruitful discussions and revision of the manuscript. Part of this study contributes to the MD thesis of AH. Sources of funding: German Research Foundation (Sebastian Bhakdi); Cusanuswerk Germany (Annette Ottinger); Siriraj Grant for Research and Education; Grant number: 001 (II)/50; Thailand Research Fund - Senior Research Scholarship (Kovit Pattanapanyasat)
- Sachs RS, Malaney P: The economic and social burden of malaria. Nature. 2002, 415: 680-685. 10.1038/415680a.View ArticlePubMedGoogle Scholar
- Guerra CA, Snow RW, Hay SI: Mapping the global extent of malaria in 2005. Trends Parasitol. 2006, 22: 353-358. 10.1016/j.pt.2006.06.006.PubMed CentralView ArticlePubMedGoogle Scholar
- Wahlgren M, Berzins K, Perlmann P, Björkman A: Characterization of the humoral response in Plasmodium falciparum malaria. I. Estimation of antibodies to P. falciparum or human erythrocytes by means of microELISA. Clin Exp Immunol. 1983, 54: 127-134.PubMed CentralPubMedGoogle Scholar
- Kutner S, Breuer WV, Ginsburg H, Aley SB, Cabantchik ZI: Characterization of permeation pathways in the plasma membrane of human erythrocytes infected with early stages of Plasmodium falciparum: Association with parasite development. J Cell Physiol. 1985, 125: 521-527. 10.1002/jcp.1041250323.View ArticlePubMedGoogle Scholar
- Goodyer ID, Johnson J, Eisenthal R, Hayes DJ: Purification of mature-stage Plasmodium falciparum by gelatine flotation. Ann Trop Med Parasitol. 1994, 88: 209-211.PubMedGoogle Scholar
- Lambros C, Vanderberg JP: Synchronization of Plasmodium falciparum erythrocytic stages in culture. J Parasitol. 1979, 65: 418-420. 10.2307/3280287.View ArticlePubMedGoogle Scholar
- Inselburg J, Banyal HS: Synthesis of DNA during the asexual cycle of Plasmodium falciparum in culture. Mol Biochem Parasitol. 1984, 10: 79-87. 10.1016/0166-6851(84)90020-3.View ArticlePubMedGoogle Scholar
- Schuster FL: Cultivation of Plasmodium spp. Clin Microbiol Rev. 2002, 15: 355-364. 10.1128/CMR.15.3.355-364.2002.PubMed CentralView ArticlePubMedGoogle Scholar
- Krugliak M, Ginsburg H: The evolution of the new permeability pathways in Plasmodium falciparum-infected erythrocytes-a kinetic analysis. Exp Parasitol. 2006, 114: 253-258. 10.1016/j.exppara.2006.03.015.View ArticlePubMedGoogle Scholar
- Moore LR, Fujioka H, Williams PS, Chalmers JJ, Grimberg B, Zimmerman PA, Zborowski M: Hemoglobin degradation in malaria-infected erythrocytes determined from live cell magnetophoresis. FASEB J. 2006, 20: 747-749.PubMed CentralPubMedGoogle Scholar
- Oberteuffer J: High gradient magnetic separation. IEEE T Mag. 1973, 9: 303-306. 10.1109/TMAG.1973.1067673.View ArticleGoogle Scholar
- Francis SE, David J, Sullivan DJ, Goldberg DE: Hemoglobin metabolism in the malaria parasite Plasmodium falciparum. Annu Rev Microbiol. 1997, 51: 97-123. 10.1146/annurev.micro.51.1.97.View ArticlePubMedGoogle Scholar
- Pauling L, Coryell C: The magnetic properties and structure of hemoglobin, oxyhemoglobin and carbonmonoxyhemoglobin. Proc Natl Acad Sci USA. 1936, 22: 159-163. 10.1073/pnas.22.3.159.PubMed CentralView ArticlePubMedGoogle Scholar
- Paul F, Roath S, Melville D, Warhurst DC, Osisanya JO: Separation of malaria-infected erythrocytes from whole blood: use of a selective high-gradient magnetic separation technique. Lancet. 1981, 2: 70-71. 10.1016/S0140-6736(81)90414-1.View ArticlePubMedGoogle Scholar
- Trang DT, Huy NT, Kariu T, Tajima K, Kamei K: One-step concentration of malarial parasite-infected red blood cells and removal of contaminating white blood cells. Malar J. 2004, 3: 7-10.1186/1475-2875-3-7.PubMed CentralView ArticlePubMedGoogle Scholar
- Staalsoe T, Giha HA, Dodoo D, Theander TG, Hviid L: Detection of antibodies to variant antigens on Plasmodium falciparum-infected erythrocytes by flow cytometry. Cytometry. 1999, 35: 329-336. 10.1002/(SICI)1097-0320(19990401)35:4<329::AID-CYTO5>3.0.CO;2-Y.View ArticlePubMedGoogle Scholar
- Uhlemann A, Staalsoe T, Klinkert M, Hviid L: Analysis of Plasmodium falciparum infected red blood cells. MACS & more. 2000, 4: 7-8.Google Scholar
- Ribaut C, Berry A, Chevalley S, Reybier K, Morlais I, Parzy D, Nepveu F, Benoit-Vical F, Valentin A: Concentration and purification by magnetic separation of the erythrocytic stages of all human Plasmodium species. Malar J. 2008, 7: 45-10.1186/1475-2875-7-45.PubMed CentralView ArticlePubMedGoogle Scholar
- Ahn SY, Shin MY, Kim YA, Yoo JA, Kwak DH, Jung YJ, Jun G, Ryu SH, Yeom JS, Ahn JY, Chai JY, Park JW: Magnetic separation: a highly effective method for synchronization of cultured erythrocytic Plasmodium falciparum. Parasitol Res. 2008, 102: 1195-1200. 10.1007/s00436-008-0893-8.View ArticlePubMedGoogle Scholar
- Lamb H: Rectilinear motion of a sphere, resistance, terminal velocity, streamlines. Case of a liquid sphere, and of a solid sphere, with slipping. Hydrodynamics. Edited by: Sir H. 1932, Lamb. Cambridge University Press, 597-601. 6Google Scholar
- Pattanapanyasat K, Walsh DS, Yongvanitchit K, Piyawatthanasakul N, Wanachiwanawin W, Webster HK: Robust in vitro replication of Plasmodium falciparum in glycosyl-phosphatidylinositol-anchored membrane glycoprotein-deficient red blood cells. Am J Trop Med Hyg. 2003, 69: 360-365.PubMedGoogle Scholar
- Bhakdi SC, Sratongno P, Chimma P, Rungruang T, Chuncharunee A, Neumann HP, Malasit P, Pattanapanyasat K: Re-evaluating acridine orange for rapid flow cytometric enumeration of parasitemia in malaria-infected rodents. Cytometry. 2007, 71: 662-667. 10.1002/cyto.a.20406.View ArticlePubMedGoogle Scholar
- Paul F, Melville D, Roath S: The morphology and viability of blood components processed by high-gradient magnetic separation. Clin Lab Haematol. 1985, 7: 43-53.View ArticlePubMedGoogle Scholar
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.