RESEARCH LETTER
Selection of a DNA aptamer against norovirus capsid protein VP1 Rico Beier1, Claudia Pahlke2, Philipp Quenzel3, Anja Henseleit3, Elke Boschke3, Gianaurelio Cuniberti4 & Dirk Labudde1 Bioinformatics Group, Hochschule Mittweida – University of Applied Sciences, Mittweida, Germany; 2Institute for Materials Science and Max Bergmann Center of Biomaterials, TU Dresden, Dresden, Germany; 3Institute of Food Technology and Bioprocess Engineering, TU Dresden, Dresden, Germany; and 4Center for Advancing Electronics Dresden, TU Dresden, Dresden, Germany
1
Correspondence: Dirk Labudde, Technikumplatz 17, 09648 Mittweida, Germany. Tel.: +49 3727 58 1469; fax: +49 3727 58 1315; e-mail: [email protected] Received 5 November 2013; revised 16 December 2013; accepted 17 December 2013. Final version published online 13 January 2014. DOI: 10.1111/1574-6968.12366 Editor: Jeff Cole
MICROBIOLOGY LETTERS
Keywords SELEX; NGS; clustering; system biology.
Abstract The genetically and antigenically diverse group of noroviruses is the major cause of human viral epidemic gastroenteritis worldwide. Virus detection and control are thus crucial topics when aiming at containing and preventing the resulting large and often persisting outbreaks. Aptamers provide a promising alternative to antibodies concerning their ability to bind and thus detect and influence bioactive molecules. These small, single-stranded oligonucleotides are able to bind to a multitude of possible target molecules with high affinity. For a specific target the highest affinity aptamers are found by screening a randomized library. In this work a DNA aptamer capable of binding to the norovirus genotype II.4 capsid protein VP1 was found. The general approach is thereby not limited to norovirus capsid, but could be extended to almost any kind of biologically relevant molecule. The development of the library enrichment was further computationally analyzed in order to describe the enrichment during screening. This is the basis for a later extensive characterization of both target and aptamers that could lead to insights regarding the functional coherence of both partners. An abstract model describing this coherence could be utilized to generate a targetspecific library, from which future aptamer screening runs could benefit.
Introduction As its first detection in Norwalk, USA, in 1972, new sequencing and analysis techniques have led to the discovery of noroviruses of different genotypes all over the world. Noroviruses, belonging to the family Caliciviridae, are genetically and antigenically diverse. They can be classified into five major genogroups, each separated into numerous genotypes. Genogroups GI and GII, especially the genotype GII.4, are responsible for the majority of human infections, whereas the other genogroups contain bovine, murine, and further human pathogenic strains. Noroviruses are the major cause of viral epidemic gastroenteritis in children and adults worldwide, often resulting in large and persisting outbreaks. They are pathogenic and highly contagious, but no vaccine is currently available (Zheng et al., 2006). Norovirus genomes are single-stranded, positive-sensed RNA enclosed in a non-enveloped protein coat with distinct cup-shaped depressions. The icosahedral capsid ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
structure is formed by 90 dimers of the capsid viral protein 1 (VP1), which includes two domains. The inner S-domain forms a shell around the RNA, whereas the P-domain, subdivided into P1 and P2, protrudes on top of that shell (Prasad et al., 1999). Additionally, the capsid contains few copies of the minor capsid protein 2. This construct leads to a thermal stability, allowing the virus to survive temperatures up to 55 °C and a pH range of 3–7 (Ausar et al., 2006). There are conflicting opinions about the role of the P2 domain with its highly variable region as receptor for norovirus infection (Tan & Jiang, 2005; Murakami et al., 2013). At present, a norovirus infection is usually diagnosed by reverse transcription-PCR (RT-PCR) or enzyme-linked immunosorbent assay using anti-norovirus antibodies. The current commercial available immunoassays are not sufficiently sensitive to be used as a single, sufficient method for norovirus detection (Gray et al., 2007). They can only be utilized for screening where results are FEMS Microbiol Lett 351 (2014) 162–169
163
Selection of a DNA aptamer against norovirus capsid protein
confirmed by RT-PCR, which is the most sensitive method and thus considered as the ‘gold standard’. However, the genetic diversity of noroviruses does not allow testing for all genotypes in one assay. A recent development, an immunochromatographic detection assay based on antibodies, was rated to have a high sensitivity and specificity (Bruggink et al., 2011). Nevertheless, there is a strong need for point-of-care methods for norovirus detection. In place of antibodies even better recognition elements could be used. Small and stable nucleotide aptamers as receptor units would allow developing real-time, label-free and low-cost biosensor systems for norovirus detection. Aptamers are single-stranded nucleotide oligomers, which fold into complex three-dimensional structures. These structures are composed of helical parts and different kinds of loops like hairpins, inner loops and junctions, which allow branching of the structure. The great binding abilities of aptamers ranging from small inorganic and organic molecules to complex targets are a result of the formation of different non-covalent chemical bonds (Stoltenburg et al., 2007). Besides the composition of the specific binding regions, their physical orientation, which is determined by the overall structure of the aptamer, is important for the atomic fit. Furthermore, such norovirus binding molecules might also be used in vivo to control norovirus infection. One interesting approach to targeting the attachment and internalization of the virus would be to inhibit the binding of the P2 subdomain to its receptor by competitive interacting molecules. Noroviruses have been trapped in glycosylated hydrogels (Zhang et al., 2006), which could also be combined with norovirus-specific aptamers. They could also control the infection by blocking capsid rearrangements necessary for norovirus uncoating. However, more information concerning the amplification cycle of noroviruses in human body will be necessary to develop proof-of-principle antiviral strategies. This work is intended to find experimentally DNA aptamer capable of binding to the norovirus genotype II.4 capsid protein VP1 using the screening technology Systematic Evolution of Ligands by EXponential Enrichment (SELEX), which is commonly used to find unique aptamers capable of binding to a specific target. The SELEX screening process is universal and thus starts with a chemically synthesized library of nucleotide oligomers. Despite its large size, this library covers only a small random fraction of the possible sequence and structure space of the aptamers. During the experiment multiple subsequent selection rounds are performed, in which library and target molecules are incubated. As the number of target molecules offering binding occasions for the oligonucleotides is much smaller than the size of the library, the FEMS Microbiol Lett 351 (2014) 162–169
arising selection pressure results in a preferred binding of higher affinity oligonucleotides. After each round the non-binding candidates are washed out and the bound aptamers are prepared for the next round (Stoltenburg et al., 2007). This leads to the enrichment of a small number of aptamer clones featuring high affinity and specificity and thus a decrease in diversity in the resulting library can be observed. In the following analysis and binding experiments the target-binding aptamers have to be spotted and validated. For this analysis it is initially necessary to establish a data base suitable for validation and information extraction purposes. The conventional sequencing technologies, such as Sanger sequencing (Sanger et al., 1977), are often limited due to their relatively low sequencing coverage and the time-consuming cloning process. The introduction of next generation sequencing technologies facilitates the process of gathering larger amounts of sequence data in relatively short periods of time (Metzker, 2010). This enables sequence data to be obtained from all SELEX rounds rather than from only the final round. The bioinformatics analysis of information provided by observing the library development allows deeper understanding of the SELEX process. Sequence characteristics possibly relevant for target binding may be identified as one result. The process flow is schematically depicted in Fig. 1. Although in the current study the SELEX approach was used to find an aptamer specific to a norovirus capsid, the general approach can be applied to other pathogens, both prokaryotes and eukaryotes.
Materials and methods Selex
The sequences of the initial library contained a 49 nt long random section enclosed by necessary primer sequences required for sequencing. It was based upon a previously used template that was modified at position 5 of the forward primer to avoid the formation of primer homodimers (Avci-Adali et al., 2010): 5′-GCCTCTTGTGAGCC TCCTAAC-N49-CATGCTTATTCTTGTCTCCC-3′. To enforce a structural refold, the aptamer library (Ella Biotech) was denatured at 90 °C for 10 min, chilled on ice for 15 min and annealed to room temperature. The capsid protein VP1 of norovirus genotype II.4 (provided by Ribbox GmbH Radebeul, Germany) expressed as a recombinant protein with a polyhistidine tag was used as target. In the first round 2 nmol and in the following eleven rounds 0.2 nmol of the target was incubated with the aptamer solution in SELEX binding buffer (100 mM NaCl, 17 mM Na2HPO4, 3 mM KH2PO4, 5 mM KCl, 2 mM MgCl2, 1 mM CaCl2, 0.02%Vol Tween 20, pH 6). After 60 min at ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
164
R. Beier et al.
Fig. 1. Illustrates the schematic overall workflow. Beginning with the experimental parts, SELEX screening process (1), next generation sequencing (2) and binding validation for one aptamer candidate (9), the resulting data were prepared for further bioinformatics analysis (3). Besides the validation of aptamer enrichment (4), a clustering based on n-grams (5, 7, 8) was performed. Thereby the option of weighting the n-grams according to secondary structural information (6) was used. The gray arrows show a further usage of bioinformatics analysis to generate a target-specific SELEX starting library based on the clustering outcome or later only based on information about a desired new target molecule.
room temperature in a rotary blender, the mixture was loaded onto a HisSpinTrap column (GE Healthcare, UK). The column was washed with the washing buffer (binding buffer, 0.2%Vol BSA), and bound DNA-protein complexes were eluted with elution buffer (100 mM NaCl, 7 mM Na2HPO4, 5 mM KCl, 2 mM MgCl2, 1 mM CaCl2, adjusted to pH4 with citric acid, 0.02%Vol Tween 20). The chemicals used for the buffer were purchased from Merck KGaA, Germany. After every third selection round an additional negative selection was performed to remove aptamer candidates binding to background materials of the experiment or to a fecal extract, the later sample matrix. Therefore a stool suspension devoid of norovirus capsid proteins was applied to the column prior to multiple washing steps using the SELEX binding buffer. Aptamer candidates that bound to this prepared column were removed from the library due to their unspecific binding. The libraries of aptamer candidates from each round were amplified between each round of selection. The PCR protocol involved 20 cycles with an annealing temperature of 58 °C and an extension time of 30 s. Sequencing
In a first PCR preparation stage, sequencing and indexing primers were added to the aptamer sequence. In a second ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
stage, a necessary barcode for later identification and two flow cell attachment sites were attached to the construct. The samples taken after each selection round were then sequenced by the Deep Sequencing Group (Biotechnology Center TU Dresden, Germany) using an Illumina HiSeq2000 Next Generation Sequencer that generated 75 bp single reads. For each round the sequencer supplied a file containing the aptamer sequences annotated with a coded quality value for each base, which approximates the respective sequencing error probability. Experimental validation of target binding
Surface plasmon resonance spectroscopy exploits angle shifting to observe molecular interaction in real-time (Pattnaik, 2005). The monitoring of the interaction takes place within an evanescent field that propagates along a metaldielectric interface. The liSPR-system (capitalis technology GmbH, Germany) was used at 30 °C with a flow rate of 5 lL s 1 and the SELEX binding buffer as running buffer. Before each experiment the chip was cleaned by rinsing the gold layer with 10 drops of 65% fuming nitric acid, incubating the chip in neutralization solution (1 9 25% ammonia solution, 1 9 30% hydrogen peroxide and 5 9 ddH2O) for 2 min and finally flushing with ddH2O. The 0.5 mg mL 1 norovirus proteins dissolved in water FEMS Microbiol Lett 351 (2014) 162–169
165
Selection of a DNA aptamer against norovirus capsid protein
Fig. 2. Shows the clustering of different SELEX rounds as another opportunity of assessing the development of the library enrichment during the experiment. Due to memory limitation, only the 5000 most abundant sequences have been clustered. As all equal sequences were aggregated prior to clustering, the enrichment of a single sequence is not observable within the visual representation, but rather groups of non-equal sequences featuring high mutual similarity are shown. A heatmap representation of the distance matrix sorted according to the UPGMA clustering was chosen. Coloring indicates the similarity of sequences from white (not similar) over yellow and red to black (identical), so darker squares around the main diagonal represent clusters of sequences with high mutual similarity. Each clustering is annotated with the corresponding Shannon entropy H and modified Simpson index D′. Starting from round 4, clusters were growing and the number of sequences showing no similarity to other sequences decreased. As expected the step from round 5 to 6 was a large leap.
were incubated with the blank gold surface for 1 h at room temperature. The protein coated sensor chip was rinsed thoroughly with the running buffer. To evaluate the binding behavior, 10 lM of the selected aptamer clone were injected to the protein surface for 10 min. The clone corresponds to the DNA sequence GTCTGTAGTAGGGAGGATGGTCCGGGGCC CCGAGACGACGTTATCAGGC. Afterward the dissociation was monitored by the replacement of the sample solution with running buffer for about 3 min. Bioinformatics analysis
More details regarding the analysis can be found in the Supporting Information (see Data S1). Validation of sequence quality and enrichment
Due to sequencing errors, data filtering is required (Luo et al., 2012). All aptamer sequences that did not include the given primer sequences were rejected. In a second filtering step sequences that could be identified as faulty using their quality values were removed from the data set. The objective of a SELEX experiment is the enrichment of binding aptamers. The strength of enrichment can be FEMS Microbiol Lett 351 (2014) 162–169
numerically expressed by various diversity measures. Dividing the sequence dataset into different species two diversity measures, the Shannon entropy H and the modified Simpson index D′ have been employed to express and compare the library enrichment after different selection rounds (Keylock, 2005). Clustering
To find groups of sequences with high similarity, an alignment-free clustering was performed. A similarity measure based on the shared occurrence of n-grams, which are substrings of fixed length n, was used to derive a distance matrix for the obtained sequences. The common clustering method UPGMA was applied on this matrix. All sequences were then arranged in the order they appear in the UPGMA created tree structure. A visual representation of the distance matrix sorted this way allowed the interactive selection of clusters. Incorporation of secondary structure information into clustering
Loop regions of nucleotide aptamers are likely to interact with target molecules (Hoinka et al., 2012). As the ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
R. Beier et al.
166
Fig. 3. Shows two sensorgrams depicting the interaction of 10 lM of the investigated aptamer with the immobilized capsid protein VP1 of norovirus genotype II.4 and with murine norovirus polymerase. Immediately, after injection of the aptamer sample (2 min after starting the experiment), signals rose exponentially and became stable in the equilibrium phase. After another 11 min, sample was replaced by the running buffer (see above) and the dissociation phase was monitored.
Fig. 4. Includes two clusterings of the 1000 most abundant sequences of the last SELEX round, one using the secondary structure based weighting, the other using the unweighted variant. The coloring is according to that used in Fig. 2. The cluster that contained the one experimentally validated binding sequence is marked by two nested blue squares. The inner one denotes the so-called core cluster, which features very high mutual sequence similarity. The outer one shows a possible extension of the cluster including sequences that exhibited a high similarity to a large fraction of the sequences of the core cluster. As expected, the result of the secondary structure-dependent clustering is visually less clear.
unpaired nucleotides in loop regions do not take part in Watson-Crick or non-canonical nucleotide pairing, the related binding sites remain available for intermolecular chemical bonding. To incorporate this information into the clustering the ViennaRNA package was used to predict secondary structures of all obtained sequences (Lorenz et al., 2011). Based on a weighted set of predicted secondary structures, for every n-gram a propensity for being fully located on unpaired nucleotides was calculated and applied to the existing similarity measure. 3-D structure prediction and docking simulation
No structural data were available for the norovirus capsid protein VPI. A homology model was therefore generated using the I-TASSER server (Zhang, 2008; Roy et al., 2010). ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
The protein with PDB-ID 1ihmB, which has 46% sequence identity to VPI and a coverage of 94%, was used as template. Due to the lack of ssDNA tertiary structure prediction solutions, the aptamer sequences were converted to a corresponding RNA and tertiary structure was predicted using the iFoldRNA online server (Sharma et al., 2008). The software package HADDOCK was then used to dock the two predicted structures (Dominguez et al., 2003; de Vries et al., 2007). After the generation of 1000 initial docking variants, the best 500 complexes regarding their free binding energy were automatically subjected to different refinement steps by HADDOCK. The best refined structure was chosen.
Results Sequencing data for aptamers from the SELEX experiment were useful for about 98% of the raw sequences. FEMS Microbiol Lett 351 (2014) 162–169
167
Selection of a DNA aptamer against norovirus capsid protein
Fig. 5. Shows a simulated complex of the aptamer with sequence GTCTG TAGTA GGGAG GATGG TCCGG GGCCC CGAGA CGACG TTATC AGGC (green) and the target protein consisting of P-domain (blue) and S-domain (red). Both single structures were predicted using different software tools (see 3D structure prediction and docking simulation) and subsequently docked.
After application of the quality filter, 83% of the sequences were retained. Three main points of diversity decrease were observed. The first one from round 1 to 2 could not be confirmed in the sequence data set, possibly because of the small read size of the second round. The enrichment corresponding to the decrease from round 3 to 4 involved only a few sequences. As abundant sequences are lower weighted within the Shannon entropy, the enrichment is significant only in the Simpson index. The more significant enrichment from round 5 to 6 was related to a larger group of sequences and furthermore to a greater increase in sequence abundance of the top sequences of that group. In the other cases, the value changes were minor. Another way to obtain information about the enrichment is the consideration of the library development over several rounds as perceptible in sequence clusterings in Fig. 2. Beginning from round 4 first small groups of similar sequences appeared, which then grew until round nine. Afterward only internal variations occurred. The two largest clusters of the final SELEX round corresponded to sequences already found in 2 of 9 separate clusters in round 4. A separate binding study using surface plasmon resonance spectroscopy was completed for the most abundant aptamer clone from the 12th selection round. In the sensorgram the dimensions of the changes are shown as a function of time. In Fig. 3 it is shown that the investigated aptamer was able to differentiate between target and negative control. Murine norovirus polymerase was chosen because of its similarity to the target in pI (6.4/ 5.8) and molecular weight (59.4 kDa/59.8 kDa). The sequences from the final selection were clustered to determine which sequences are located in the same
FEMS Microbiol Lett 351 (2014) 162–169
cluster as the proved sequence. Figure 4 shows a heatmap representation of this clustering. Without taking into account the secondary structural information, clusters were visually emerging very clearly. The core cluster contained only highly similar sequences (< 4 single base mutations), which covered about 25% of the sequence pool, and is now referred to as main group. The extended cluster additionally contained some low percentages of other sequences. The greatest of these small groups was similar to the proved sequence in its first half and regarded as the side group. For the secondary structuredependent clustering, it was obvious that the clusters were not separated as clearly as before. Here, the core cluster contained both previously mentioned sequence groups, together occupying c. 21% of the sequences within the pool. As single base alterations can have major influence on secondary structure, the consideration of the extended cluster only added sequences to the main group that offered a very high sequence similarity. Figure 5 shows the best resulting structure of the simulated complex between the experimentally validated aptamer and the target protein based upon tertiary structure prediction and docking simulation. Further studies are needed to analyze the concrete aptamer–target binding.
Discussion The investigation has shown that after some observable enrichment steps later rounds of the SELEX experiment were characterized by a slow decrease in diversity. Sequences contained in the largest clusters of the last round could also be found in the enrichment in round 4. ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
R. Beier et al.
168
This underlines the possibility to use the analysis of an early SELEX round to computationally enrich the aptamer library. This strategy would reduce the number of necessary SELEX runs by introducing a sequence analysis step and afterward continuing the experiment with a library enriched on the basis of this analysis. Pursuing this concept, the base information for generating an advanced SELEX library does not necessarily need to originate directly from a previous SELEX experiment performed with the same target. Beginning with the observed characteristic of enrichment as first descriptor for the binding preference to the target, a more comprehensive description can lead to further insight. Here, other complex descriptors such as electrostatics or interaction preferences could be used. The correlation of these descriptors can be employed to design an abstract model describing the aptamer–target binding relationship. Many analyzed experimental data would be required to construct a model like this, which would also be able to predict potential characteristics of binding aptamers such as composition and architecture, only based on analysis of the desired target protein. This information could be used to design a specifically optimized initial SELEX library toward a given target protein. As the model only supplies an indication for preferred aptamer characteristics, a way has to be found to use this knowledge. Binding motifs and other predictable properties could be involved in a semirandom library generation, either by filtering random sequences according to the constraints or by modifying the underlying randomization. This fills the aptamers’ complex conformation space with diverse structures satisfying the predicted characteristics. The approach restricts the original universality of the SELEX process, which can be seen critically. Otherwise, a specific library involves the chance of finding higher affinity aptamers in less time and hence reduces demands and resources. An additional parameter controlling the degree of influence would allow balancing these advances and drawbacks individually. Reaching this point of development, the model is applicable to other, even structurally unknown target proteins and thus can contribute to either dry or wet laboratory investigations in fields of biosensor development and medical treatment.
Acknowledgement This work has been supported and funded by the Free State of Saxony, the European Social Fund and Saxon Ministry of Science and Fine Arts. We appreciate Dr Tajana Sch€ utze (Institute of Chemistry and Biochemistry, Freie Universit€at Berlin) and Dr Andreas Dahl (Deep Sequencing Group, SFB 655/BIOTEC) for helping us with their knowledge and ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
skills in DNA pool generation and sequencing, respectively. The authors declare that there is no conflict of interests regarding the publication of this article.
References Ausar SF, Foubert TR, Hudson MH, Vedvick TS & Middaugh CR (2006) Conformational stability and disassembly of Norwalk virus-like particles. Effect of pH and temperature. J Biol Chem 281: 19478–19488. Avci-Adali M, Paul A, Wilhelm N, Ziemer G & Wendel HP (2010) Upgrading SELEX technology by using lambda exonuclease digestion for single-stranded DNA generation. Molecules 15: 1–11. Bruggink LD, Witlox KJ, Sameer R, Catton MG & Marshall JA (2011) Evaluation of the RIDA QUICK immunochromatographic norovirus detection assay using specimens from Australian gastroenteritis incidents. J Virol Methods 173: 121–126. de Vries SJ, van Dijk AD, Krzeminski M, van Dijk M, Thureau A, Hsu V, Wassenaar T & Bonvin AM (2007) HADDOCK versus HADDOCK: new features and performance of HADDOCK2.0 on the CAPRI targets. Proteins 69: 726–733. Dominguez C, Boelens R & Bonvin AM (2003) HADDOCK: a protein-protein docking approach based on biochemical or biophysical information. J Am Chem Soc 125: 1731–1737. Gray JJ, Kohli E, Ruggeri FM et al. (2007) European multicenter evaluation of commercial enzyme immunoassays for detecting norovirus antigen in fecal samples. Clin Vaccine Immunol 14: 1349–1355. Hoinka J, Zotenko E, Friedman A, Sauna ZE & Przytycka TM (2012) Identification of sequence-structure RNA binding motifs for SELEX-derived aptamers. Bioinformatics 28: 215. Keylock CJ (2005) Simpson diversity and the Shannon–Wiener index as special cases of a generalized entropy. Oikos 109: 203–207. Lorenz R, Bernhart SH, Honer Zu Siederdissen C, Tafer H, Flamm C, Stadler PF & Hofacker IL (2011) ViennaRNA Package 2.0. Algorithms Mol Biol 6: 26. Luo C, Tsementzi D, Kyrpides N, Read T & Konstantinidis KT (2012) Direct comparisons of Illumina vs. Roche 454 sequencing technologies on the same microbial community DNA sample. PLoS One 7: 30087. Metzker ML (2010) Sequencing technologies – the next generation. Nat Rev Genet 11: 31–46. Murakami K, Kurihara C, Oka T et al. (2013) Norovirus binding to intestinal epithelial cells is independent of histo-blood group antigens. PLoS One 8: e66534. Pattnaik P (2005) Surface plasmon resonance: applications in understanding receptor-ligand interaction. Appl Biochem Biotechnol 126: 79–92. Prasad BV, Hardy ME, Dokland T, Bella J, Rossmann MG & Estes MK (1999) X-ray crystallographic structure of the Norwalk virus capsid. Science 286: 287–290.
FEMS Microbiol Lett 351 (2014) 162–169
169
Selection of a DNA aptamer against norovirus capsid protein
Roy A, Kucukural A & Zhang Y (2010) I-TASSER: a unified platform for automated protein structure and function prediction. Nat Protoc 5: 725–738. Sanger F, Nicklen S & Coulson AR (1977) DNA sequencing with chain-terminating inhibitors. P Natl Acad Sci USA 74: 5463–5467. Sharma S, Ding F & Dokholyan NV (2008) iFoldRNA: three-dimensional RNA structure prediction and folding. Bioinformatics 24: 1951–1952. Stoltenburg R, Reinemann C & Strehlitz B (2007) SELEX–a (r) evolutionary method to generate high-affinity nucleic acid ligands. Biomol Eng 24: 381–403. Tan M & Jiang X (2005) Norovirus and its histo-blood group antigen receptors: an answer to a historical puzzle. Trends Microbiol 13: 285–293. Zhang Y (2008) I-TASSER server for protein 3D structure prediction. BMC Bioinformatics 9: 40.
FEMS Microbiol Lett 351 (2014) 162–169
Zhang Y, Yao Q, Xia C, Jiang X & Wang PG (2006) Trapping norovirus by glycosylated hydrogels: a potential oral antiviral drug. ChemMedChem 1: 1361–1366. Zheng DP, Ando T, Fankhauser RL, Beard RS, Glass RI & Monroe SS (2006) Norovirus classification and proposed strain nomenclature. Virology 346: 312–323.
Supporting Information Additional Supporting Information may be found in the online version of this article: Data S1. Provides further detail regarding the methods used for analysis.
ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
Selection of a DNA aptamer against norovirus capsid protein VP1 Rico Beier1, Claudia Pahlke2, Philipp Quenzel3, Anja Henseleit3, Elke Boschke3, Gianaurelio Cuniberti4 & Dirk Labudde1 Bioinformatics Group, Hochschule Mittweida – University of Applied Sciences, Mittweida, Germany; 2Institute for Materials Science and Max Bergmann Center of Biomaterials, TU Dresden, Dresden, Germany; 3Institute of Food Technology and Bioprocess Engineering, TU Dresden, Dresden, Germany; and 4Center for Advancing Electronics Dresden, TU Dresden, Dresden, Germany
1
Correspondence: Dirk Labudde, Technikumplatz 17, 09648 Mittweida, Germany. Tel.: +49 3727 58 1469; fax: +49 3727 58 1315; e-mail: [email protected] Received 5 November 2013; revised 16 December 2013; accepted 17 December 2013. Final version published online 13 January 2014. DOI: 10.1111/1574-6968.12366 Editor: Jeff Cole
MICROBIOLOGY LETTERS
Keywords SELEX; NGS; clustering; system biology.
Abstract The genetically and antigenically diverse group of noroviruses is the major cause of human viral epidemic gastroenteritis worldwide. Virus detection and control are thus crucial topics when aiming at containing and preventing the resulting large and often persisting outbreaks. Aptamers provide a promising alternative to antibodies concerning their ability to bind and thus detect and influence bioactive molecules. These small, single-stranded oligonucleotides are able to bind to a multitude of possible target molecules with high affinity. For a specific target the highest affinity aptamers are found by screening a randomized library. In this work a DNA aptamer capable of binding to the norovirus genotype II.4 capsid protein VP1 was found. The general approach is thereby not limited to norovirus capsid, but could be extended to almost any kind of biologically relevant molecule. The development of the library enrichment was further computationally analyzed in order to describe the enrichment during screening. This is the basis for a later extensive characterization of both target and aptamers that could lead to insights regarding the functional coherence of both partners. An abstract model describing this coherence could be utilized to generate a targetspecific library, from which future aptamer screening runs could benefit.
Introduction As its first detection in Norwalk, USA, in 1972, new sequencing and analysis techniques have led to the discovery of noroviruses of different genotypes all over the world. Noroviruses, belonging to the family Caliciviridae, are genetically and antigenically diverse. They can be classified into five major genogroups, each separated into numerous genotypes. Genogroups GI and GII, especially the genotype GII.4, are responsible for the majority of human infections, whereas the other genogroups contain bovine, murine, and further human pathogenic strains. Noroviruses are the major cause of viral epidemic gastroenteritis in children and adults worldwide, often resulting in large and persisting outbreaks. They are pathogenic and highly contagious, but no vaccine is currently available (Zheng et al., 2006). Norovirus genomes are single-stranded, positive-sensed RNA enclosed in a non-enveloped protein coat with distinct cup-shaped depressions. The icosahedral capsid ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
structure is formed by 90 dimers of the capsid viral protein 1 (VP1), which includes two domains. The inner S-domain forms a shell around the RNA, whereas the P-domain, subdivided into P1 and P2, protrudes on top of that shell (Prasad et al., 1999). Additionally, the capsid contains few copies of the minor capsid protein 2. This construct leads to a thermal stability, allowing the virus to survive temperatures up to 55 °C and a pH range of 3–7 (Ausar et al., 2006). There are conflicting opinions about the role of the P2 domain with its highly variable region as receptor for norovirus infection (Tan & Jiang, 2005; Murakami et al., 2013). At present, a norovirus infection is usually diagnosed by reverse transcription-PCR (RT-PCR) or enzyme-linked immunosorbent assay using anti-norovirus antibodies. The current commercial available immunoassays are not sufficiently sensitive to be used as a single, sufficient method for norovirus detection (Gray et al., 2007). They can only be utilized for screening where results are FEMS Microbiol Lett 351 (2014) 162–169
163
Selection of a DNA aptamer against norovirus capsid protein
confirmed by RT-PCR, which is the most sensitive method and thus considered as the ‘gold standard’. However, the genetic diversity of noroviruses does not allow testing for all genotypes in one assay. A recent development, an immunochromatographic detection assay based on antibodies, was rated to have a high sensitivity and specificity (Bruggink et al., 2011). Nevertheless, there is a strong need for point-of-care methods for norovirus detection. In place of antibodies even better recognition elements could be used. Small and stable nucleotide aptamers as receptor units would allow developing real-time, label-free and low-cost biosensor systems for norovirus detection. Aptamers are single-stranded nucleotide oligomers, which fold into complex three-dimensional structures. These structures are composed of helical parts and different kinds of loops like hairpins, inner loops and junctions, which allow branching of the structure. The great binding abilities of aptamers ranging from small inorganic and organic molecules to complex targets are a result of the formation of different non-covalent chemical bonds (Stoltenburg et al., 2007). Besides the composition of the specific binding regions, their physical orientation, which is determined by the overall structure of the aptamer, is important for the atomic fit. Furthermore, such norovirus binding molecules might also be used in vivo to control norovirus infection. One interesting approach to targeting the attachment and internalization of the virus would be to inhibit the binding of the P2 subdomain to its receptor by competitive interacting molecules. Noroviruses have been trapped in glycosylated hydrogels (Zhang et al., 2006), which could also be combined with norovirus-specific aptamers. They could also control the infection by blocking capsid rearrangements necessary for norovirus uncoating. However, more information concerning the amplification cycle of noroviruses in human body will be necessary to develop proof-of-principle antiviral strategies. This work is intended to find experimentally DNA aptamer capable of binding to the norovirus genotype II.4 capsid protein VP1 using the screening technology Systematic Evolution of Ligands by EXponential Enrichment (SELEX), which is commonly used to find unique aptamers capable of binding to a specific target. The SELEX screening process is universal and thus starts with a chemically synthesized library of nucleotide oligomers. Despite its large size, this library covers only a small random fraction of the possible sequence and structure space of the aptamers. During the experiment multiple subsequent selection rounds are performed, in which library and target molecules are incubated. As the number of target molecules offering binding occasions for the oligonucleotides is much smaller than the size of the library, the FEMS Microbiol Lett 351 (2014) 162–169
arising selection pressure results in a preferred binding of higher affinity oligonucleotides. After each round the non-binding candidates are washed out and the bound aptamers are prepared for the next round (Stoltenburg et al., 2007). This leads to the enrichment of a small number of aptamer clones featuring high affinity and specificity and thus a decrease in diversity in the resulting library can be observed. In the following analysis and binding experiments the target-binding aptamers have to be spotted and validated. For this analysis it is initially necessary to establish a data base suitable for validation and information extraction purposes. The conventional sequencing technologies, such as Sanger sequencing (Sanger et al., 1977), are often limited due to their relatively low sequencing coverage and the time-consuming cloning process. The introduction of next generation sequencing technologies facilitates the process of gathering larger amounts of sequence data in relatively short periods of time (Metzker, 2010). This enables sequence data to be obtained from all SELEX rounds rather than from only the final round. The bioinformatics analysis of information provided by observing the library development allows deeper understanding of the SELEX process. Sequence characteristics possibly relevant for target binding may be identified as one result. The process flow is schematically depicted in Fig. 1. Although in the current study the SELEX approach was used to find an aptamer specific to a norovirus capsid, the general approach can be applied to other pathogens, both prokaryotes and eukaryotes.
Materials and methods Selex
The sequences of the initial library contained a 49 nt long random section enclosed by necessary primer sequences required for sequencing. It was based upon a previously used template that was modified at position 5 of the forward primer to avoid the formation of primer homodimers (Avci-Adali et al., 2010): 5′-GCCTCTTGTGAGCC TCCTAAC-N49-CATGCTTATTCTTGTCTCCC-3′. To enforce a structural refold, the aptamer library (Ella Biotech) was denatured at 90 °C for 10 min, chilled on ice for 15 min and annealed to room temperature. The capsid protein VP1 of norovirus genotype II.4 (provided by Ribbox GmbH Radebeul, Germany) expressed as a recombinant protein with a polyhistidine tag was used as target. In the first round 2 nmol and in the following eleven rounds 0.2 nmol of the target was incubated with the aptamer solution in SELEX binding buffer (100 mM NaCl, 17 mM Na2HPO4, 3 mM KH2PO4, 5 mM KCl, 2 mM MgCl2, 1 mM CaCl2, 0.02%Vol Tween 20, pH 6). After 60 min at ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
164
R. Beier et al.
Fig. 1. Illustrates the schematic overall workflow. Beginning with the experimental parts, SELEX screening process (1), next generation sequencing (2) and binding validation for one aptamer candidate (9), the resulting data were prepared for further bioinformatics analysis (3). Besides the validation of aptamer enrichment (4), a clustering based on n-grams (5, 7, 8) was performed. Thereby the option of weighting the n-grams according to secondary structural information (6) was used. The gray arrows show a further usage of bioinformatics analysis to generate a target-specific SELEX starting library based on the clustering outcome or later only based on information about a desired new target molecule.
room temperature in a rotary blender, the mixture was loaded onto a HisSpinTrap column (GE Healthcare, UK). The column was washed with the washing buffer (binding buffer, 0.2%Vol BSA), and bound DNA-protein complexes were eluted with elution buffer (100 mM NaCl, 7 mM Na2HPO4, 5 mM KCl, 2 mM MgCl2, 1 mM CaCl2, adjusted to pH4 with citric acid, 0.02%Vol Tween 20). The chemicals used for the buffer were purchased from Merck KGaA, Germany. After every third selection round an additional negative selection was performed to remove aptamer candidates binding to background materials of the experiment or to a fecal extract, the later sample matrix. Therefore a stool suspension devoid of norovirus capsid proteins was applied to the column prior to multiple washing steps using the SELEX binding buffer. Aptamer candidates that bound to this prepared column were removed from the library due to their unspecific binding. The libraries of aptamer candidates from each round were amplified between each round of selection. The PCR protocol involved 20 cycles with an annealing temperature of 58 °C and an extension time of 30 s. Sequencing
In a first PCR preparation stage, sequencing and indexing primers were added to the aptamer sequence. In a second ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
stage, a necessary barcode for later identification and two flow cell attachment sites were attached to the construct. The samples taken after each selection round were then sequenced by the Deep Sequencing Group (Biotechnology Center TU Dresden, Germany) using an Illumina HiSeq2000 Next Generation Sequencer that generated 75 bp single reads. For each round the sequencer supplied a file containing the aptamer sequences annotated with a coded quality value for each base, which approximates the respective sequencing error probability. Experimental validation of target binding
Surface plasmon resonance spectroscopy exploits angle shifting to observe molecular interaction in real-time (Pattnaik, 2005). The monitoring of the interaction takes place within an evanescent field that propagates along a metaldielectric interface. The liSPR-system (capitalis technology GmbH, Germany) was used at 30 °C with a flow rate of 5 lL s 1 and the SELEX binding buffer as running buffer. Before each experiment the chip was cleaned by rinsing the gold layer with 10 drops of 65% fuming nitric acid, incubating the chip in neutralization solution (1 9 25% ammonia solution, 1 9 30% hydrogen peroxide and 5 9 ddH2O) for 2 min and finally flushing with ddH2O. The 0.5 mg mL 1 norovirus proteins dissolved in water FEMS Microbiol Lett 351 (2014) 162–169
165
Selection of a DNA aptamer against norovirus capsid protein
Fig. 2. Shows the clustering of different SELEX rounds as another opportunity of assessing the development of the library enrichment during the experiment. Due to memory limitation, only the 5000 most abundant sequences have been clustered. As all equal sequences were aggregated prior to clustering, the enrichment of a single sequence is not observable within the visual representation, but rather groups of non-equal sequences featuring high mutual similarity are shown. A heatmap representation of the distance matrix sorted according to the UPGMA clustering was chosen. Coloring indicates the similarity of sequences from white (not similar) over yellow and red to black (identical), so darker squares around the main diagonal represent clusters of sequences with high mutual similarity. Each clustering is annotated with the corresponding Shannon entropy H and modified Simpson index D′. Starting from round 4, clusters were growing and the number of sequences showing no similarity to other sequences decreased. As expected the step from round 5 to 6 was a large leap.
were incubated with the blank gold surface for 1 h at room temperature. The protein coated sensor chip was rinsed thoroughly with the running buffer. To evaluate the binding behavior, 10 lM of the selected aptamer clone were injected to the protein surface for 10 min. The clone corresponds to the DNA sequence GTCTGTAGTAGGGAGGATGGTCCGGGGCC CCGAGACGACGTTATCAGGC. Afterward the dissociation was monitored by the replacement of the sample solution with running buffer for about 3 min. Bioinformatics analysis
More details regarding the analysis can be found in the Supporting Information (see Data S1). Validation of sequence quality and enrichment
Due to sequencing errors, data filtering is required (Luo et al., 2012). All aptamer sequences that did not include the given primer sequences were rejected. In a second filtering step sequences that could be identified as faulty using their quality values were removed from the data set. The objective of a SELEX experiment is the enrichment of binding aptamers. The strength of enrichment can be FEMS Microbiol Lett 351 (2014) 162–169
numerically expressed by various diversity measures. Dividing the sequence dataset into different species two diversity measures, the Shannon entropy H and the modified Simpson index D′ have been employed to express and compare the library enrichment after different selection rounds (Keylock, 2005). Clustering
To find groups of sequences with high similarity, an alignment-free clustering was performed. A similarity measure based on the shared occurrence of n-grams, which are substrings of fixed length n, was used to derive a distance matrix for the obtained sequences. The common clustering method UPGMA was applied on this matrix. All sequences were then arranged in the order they appear in the UPGMA created tree structure. A visual representation of the distance matrix sorted this way allowed the interactive selection of clusters. Incorporation of secondary structure information into clustering
Loop regions of nucleotide aptamers are likely to interact with target molecules (Hoinka et al., 2012). As the ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
R. Beier et al.
166
Fig. 3. Shows two sensorgrams depicting the interaction of 10 lM of the investigated aptamer with the immobilized capsid protein VP1 of norovirus genotype II.4 and with murine norovirus polymerase. Immediately, after injection of the aptamer sample (2 min after starting the experiment), signals rose exponentially and became stable in the equilibrium phase. After another 11 min, sample was replaced by the running buffer (see above) and the dissociation phase was monitored.
Fig. 4. Includes two clusterings of the 1000 most abundant sequences of the last SELEX round, one using the secondary structure based weighting, the other using the unweighted variant. The coloring is according to that used in Fig. 2. The cluster that contained the one experimentally validated binding sequence is marked by two nested blue squares. The inner one denotes the so-called core cluster, which features very high mutual sequence similarity. The outer one shows a possible extension of the cluster including sequences that exhibited a high similarity to a large fraction of the sequences of the core cluster. As expected, the result of the secondary structure-dependent clustering is visually less clear.
unpaired nucleotides in loop regions do not take part in Watson-Crick or non-canonical nucleotide pairing, the related binding sites remain available for intermolecular chemical bonding. To incorporate this information into the clustering the ViennaRNA package was used to predict secondary structures of all obtained sequences (Lorenz et al., 2011). Based on a weighted set of predicted secondary structures, for every n-gram a propensity for being fully located on unpaired nucleotides was calculated and applied to the existing similarity measure. 3-D structure prediction and docking simulation
No structural data were available for the norovirus capsid protein VPI. A homology model was therefore generated using the I-TASSER server (Zhang, 2008; Roy et al., 2010). ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
The protein with PDB-ID 1ihmB, which has 46% sequence identity to VPI and a coverage of 94%, was used as template. Due to the lack of ssDNA tertiary structure prediction solutions, the aptamer sequences were converted to a corresponding RNA and tertiary structure was predicted using the iFoldRNA online server (Sharma et al., 2008). The software package HADDOCK was then used to dock the two predicted structures (Dominguez et al., 2003; de Vries et al., 2007). After the generation of 1000 initial docking variants, the best 500 complexes regarding their free binding energy were automatically subjected to different refinement steps by HADDOCK. The best refined structure was chosen.
Results Sequencing data for aptamers from the SELEX experiment were useful for about 98% of the raw sequences. FEMS Microbiol Lett 351 (2014) 162–169
167
Selection of a DNA aptamer against norovirus capsid protein
Fig. 5. Shows a simulated complex of the aptamer with sequence GTCTG TAGTA GGGAG GATGG TCCGG GGCCC CGAGA CGACG TTATC AGGC (green) and the target protein consisting of P-domain (blue) and S-domain (red). Both single structures were predicted using different software tools (see 3D structure prediction and docking simulation) and subsequently docked.
After application of the quality filter, 83% of the sequences were retained. Three main points of diversity decrease were observed. The first one from round 1 to 2 could not be confirmed in the sequence data set, possibly because of the small read size of the second round. The enrichment corresponding to the decrease from round 3 to 4 involved only a few sequences. As abundant sequences are lower weighted within the Shannon entropy, the enrichment is significant only in the Simpson index. The more significant enrichment from round 5 to 6 was related to a larger group of sequences and furthermore to a greater increase in sequence abundance of the top sequences of that group. In the other cases, the value changes were minor. Another way to obtain information about the enrichment is the consideration of the library development over several rounds as perceptible in sequence clusterings in Fig. 2. Beginning from round 4 first small groups of similar sequences appeared, which then grew until round nine. Afterward only internal variations occurred. The two largest clusters of the final SELEX round corresponded to sequences already found in 2 of 9 separate clusters in round 4. A separate binding study using surface plasmon resonance spectroscopy was completed for the most abundant aptamer clone from the 12th selection round. In the sensorgram the dimensions of the changes are shown as a function of time. In Fig. 3 it is shown that the investigated aptamer was able to differentiate between target and negative control. Murine norovirus polymerase was chosen because of its similarity to the target in pI (6.4/ 5.8) and molecular weight (59.4 kDa/59.8 kDa). The sequences from the final selection were clustered to determine which sequences are located in the same
FEMS Microbiol Lett 351 (2014) 162–169
cluster as the proved sequence. Figure 4 shows a heatmap representation of this clustering. Without taking into account the secondary structural information, clusters were visually emerging very clearly. The core cluster contained only highly similar sequences (< 4 single base mutations), which covered about 25% of the sequence pool, and is now referred to as main group. The extended cluster additionally contained some low percentages of other sequences. The greatest of these small groups was similar to the proved sequence in its first half and regarded as the side group. For the secondary structuredependent clustering, it was obvious that the clusters were not separated as clearly as before. Here, the core cluster contained both previously mentioned sequence groups, together occupying c. 21% of the sequences within the pool. As single base alterations can have major influence on secondary structure, the consideration of the extended cluster only added sequences to the main group that offered a very high sequence similarity. Figure 5 shows the best resulting structure of the simulated complex between the experimentally validated aptamer and the target protein based upon tertiary structure prediction and docking simulation. Further studies are needed to analyze the concrete aptamer–target binding.
Discussion The investigation has shown that after some observable enrichment steps later rounds of the SELEX experiment were characterized by a slow decrease in diversity. Sequences contained in the largest clusters of the last round could also be found in the enrichment in round 4. ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
R. Beier et al.
168
This underlines the possibility to use the analysis of an early SELEX round to computationally enrich the aptamer library. This strategy would reduce the number of necessary SELEX runs by introducing a sequence analysis step and afterward continuing the experiment with a library enriched on the basis of this analysis. Pursuing this concept, the base information for generating an advanced SELEX library does not necessarily need to originate directly from a previous SELEX experiment performed with the same target. Beginning with the observed characteristic of enrichment as first descriptor for the binding preference to the target, a more comprehensive description can lead to further insight. Here, other complex descriptors such as electrostatics or interaction preferences could be used. The correlation of these descriptors can be employed to design an abstract model describing the aptamer–target binding relationship. Many analyzed experimental data would be required to construct a model like this, which would also be able to predict potential characteristics of binding aptamers such as composition and architecture, only based on analysis of the desired target protein. This information could be used to design a specifically optimized initial SELEX library toward a given target protein. As the model only supplies an indication for preferred aptamer characteristics, a way has to be found to use this knowledge. Binding motifs and other predictable properties could be involved in a semirandom library generation, either by filtering random sequences according to the constraints or by modifying the underlying randomization. This fills the aptamers’ complex conformation space with diverse structures satisfying the predicted characteristics. The approach restricts the original universality of the SELEX process, which can be seen critically. Otherwise, a specific library involves the chance of finding higher affinity aptamers in less time and hence reduces demands and resources. An additional parameter controlling the degree of influence would allow balancing these advances and drawbacks individually. Reaching this point of development, the model is applicable to other, even structurally unknown target proteins and thus can contribute to either dry or wet laboratory investigations in fields of biosensor development and medical treatment.
Acknowledgement This work has been supported and funded by the Free State of Saxony, the European Social Fund and Saxon Ministry of Science and Fine Arts. We appreciate Dr Tajana Sch€ utze (Institute of Chemistry and Biochemistry, Freie Universit€at Berlin) and Dr Andreas Dahl (Deep Sequencing Group, SFB 655/BIOTEC) for helping us with their knowledge and ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
skills in DNA pool generation and sequencing, respectively. The authors declare that there is no conflict of interests regarding the publication of this article.
References Ausar SF, Foubert TR, Hudson MH, Vedvick TS & Middaugh CR (2006) Conformational stability and disassembly of Norwalk virus-like particles. Effect of pH and temperature. J Biol Chem 281: 19478–19488. Avci-Adali M, Paul A, Wilhelm N, Ziemer G & Wendel HP (2010) Upgrading SELEX technology by using lambda exonuclease digestion for single-stranded DNA generation. Molecules 15: 1–11. Bruggink LD, Witlox KJ, Sameer R, Catton MG & Marshall JA (2011) Evaluation of the RIDA QUICK immunochromatographic norovirus detection assay using specimens from Australian gastroenteritis incidents. J Virol Methods 173: 121–126. de Vries SJ, van Dijk AD, Krzeminski M, van Dijk M, Thureau A, Hsu V, Wassenaar T & Bonvin AM (2007) HADDOCK versus HADDOCK: new features and performance of HADDOCK2.0 on the CAPRI targets. Proteins 69: 726–733. Dominguez C, Boelens R & Bonvin AM (2003) HADDOCK: a protein-protein docking approach based on biochemical or biophysical information. J Am Chem Soc 125: 1731–1737. Gray JJ, Kohli E, Ruggeri FM et al. (2007) European multicenter evaluation of commercial enzyme immunoassays for detecting norovirus antigen in fecal samples. Clin Vaccine Immunol 14: 1349–1355. Hoinka J, Zotenko E, Friedman A, Sauna ZE & Przytycka TM (2012) Identification of sequence-structure RNA binding motifs for SELEX-derived aptamers. Bioinformatics 28: 215. Keylock CJ (2005) Simpson diversity and the Shannon–Wiener index as special cases of a generalized entropy. Oikos 109: 203–207. Lorenz R, Bernhart SH, Honer Zu Siederdissen C, Tafer H, Flamm C, Stadler PF & Hofacker IL (2011) ViennaRNA Package 2.0. Algorithms Mol Biol 6: 26. Luo C, Tsementzi D, Kyrpides N, Read T & Konstantinidis KT (2012) Direct comparisons of Illumina vs. Roche 454 sequencing technologies on the same microbial community DNA sample. PLoS One 7: 30087. Metzker ML (2010) Sequencing technologies – the next generation. Nat Rev Genet 11: 31–46. Murakami K, Kurihara C, Oka T et al. (2013) Norovirus binding to intestinal epithelial cells is independent of histo-blood group antigens. PLoS One 8: e66534. Pattnaik P (2005) Surface plasmon resonance: applications in understanding receptor-ligand interaction. Appl Biochem Biotechnol 126: 79–92. Prasad BV, Hardy ME, Dokland T, Bella J, Rossmann MG & Estes MK (1999) X-ray crystallographic structure of the Norwalk virus capsid. Science 286: 287–290.
FEMS Microbiol Lett 351 (2014) 162–169
169
Selection of a DNA aptamer against norovirus capsid protein
Roy A, Kucukural A & Zhang Y (2010) I-TASSER: a unified platform for automated protein structure and function prediction. Nat Protoc 5: 725–738. Sanger F, Nicklen S & Coulson AR (1977) DNA sequencing with chain-terminating inhibitors. P Natl Acad Sci USA 74: 5463–5467. Sharma S, Ding F & Dokholyan NV (2008) iFoldRNA: three-dimensional RNA structure prediction and folding. Bioinformatics 24: 1951–1952. Stoltenburg R, Reinemann C & Strehlitz B (2007) SELEX–a (r) evolutionary method to generate high-affinity nucleic acid ligands. Biomol Eng 24: 381–403. Tan M & Jiang X (2005) Norovirus and its histo-blood group antigen receptors: an answer to a historical puzzle. Trends Microbiol 13: 285–293. Zhang Y (2008) I-TASSER server for protein 3D structure prediction. BMC Bioinformatics 9: 40.
FEMS Microbiol Lett 351 (2014) 162–169
Zhang Y, Yao Q, Xia C, Jiang X & Wang PG (2006) Trapping norovirus by glycosylated hydrogels: a potential oral antiviral drug. ChemMedChem 1: 1361–1366. Zheng DP, Ando T, Fankhauser RL, Beard RS, Glass RI & Monroe SS (2006) Norovirus classification and proposed strain nomenclature. Virology 346: 312–323.
Supporting Information Additional Supporting Information may be found in the online version of this article: Data S1. Provides further detail regarding the methods used for analysis.
ª 2013 Federation of European Microbiological Societies. Published by John Wiley & Sons Ltd. All rights reserved
