Complete characterization of antibody specificities associated to natural infections is expected to provide a rich source of serologic biomarkers with potential applications in molecular diagnosis, follow-up of chemotherapeutic treatments, and prioritization of targets for vaccine development. Here, we developed a highly-multiplexed platform based on next-generation high-density peptide microarrays to map these specificities in Chagas Disease, an exemplar of a human infectious disease caused by the protozoan Trypanosoma cruzi. We designed a high-density peptide microarray containing more than 175,000 overlapping 15mer peptides derived from T. cruzi proteins. Peptides were synthesized in situ on microarray slides, spanning the complete length of 457 parasite proteins with fully overlapped 15mers (1 residue shift). Screening of these slides with antibodies purified from infected patients and healthy donors demonstrated both a high technical reproducibility as well as epitope mapping consistency when compared with earlier low-throughput technologies. Using a conservative signal threshold to classify positive (reactive) peptides we identified 2,031 disease-specific peptides and 97 novel parasite antigens, effectively doubling the number of known antigens and providing a 10-fold increase in the number of fine mapped antigenic determinants for this disease. Finally, further analysis of the chip data showed that optimizing the amount of sequence overlap of displayed peptides can increase the protein space covered in a single chip by at least ∼threefold without sacrificing sensitivity. In conclusion, we show the power of high-density peptide chips for the discovery of pathogen-specific linear B-cell epitopes from clinical samples, thus setting the stage for high-throughput biomarker discovery screenings and proteome-wide studies of immune responses against pathogens.
Detailed knowledge of antigens and epitopes recognized in the context of naturally acquired human infections has important implications for our understanding of immune system responses against pathogens, and of the immunopathogenesis of infectious diseases. This knowledge is also important for practical clinical applications such as the development of improved vaccines, intervention strategies, and diagnostics.
In the last decades, significant progress has been made in the discovery of antigens and epitopes thanks to a number of methodologies such as cDNA expression libraries (
1.
), combinatorial peptide libraries (2.
), and peptide and protein microarrays (3.
, 4.
). However, current knowledge of the B-cell antigens and the epitope repertoire recognized by the immune system in human infections is still scarce. Indeed, the Immune Epitope Database (5.
) currently contains an average of only 10 antigens with mapped B-cell epitopes recognized from naturally acquired human infections for bacterial or eukaryotic pathogens. The reasons for this are many, but can be largely attributed to different limitations in the mentioned screening technologies. Heterologous expression of cDNA libraries has been used to guide antigen discovery, but mapping of epitopes most often lags behind as it is a much more costly exercise. Similarly, combinatorial peptide libraries greatly facilitate the identification of peptides that are specifically recognized by antibodies, but these peptides have sequences that can greatly differ from those of the native epitopes (they are mimotopes), thus making it difficult to identify the original antigens. As a result, we currently have only limited detailed information on the fine specificities of the antibody response against complex pathogens.The number of tools for studying immune responses has recently expanded with the inclusion of peptide and protein microarrays, which have been used to identify pathogen-specific antigens and linear epitopes (
6.
, 7.
, 8.
, 9.
, - Crompton P.D.
- Kayala M.A.
- Traore B.
- Kayentao K.
- Ongoiba A.
- Weiss G.E.
- Molina D.M.
- Burk C.R.
- Waisberg M.
- Jasinskas A.
- Tan X.
- Doumbo S.
- Doumtabe D.
- Kone Y.
- Narum D.L.
- Liang X.
- Doumbo O.K.
- Miller L.H.
- Doolan D.L.
- Baldi P.
- Felgner P.L.
- Pierce S.K.
A prospective analysis of the Ab response to Plasmodium falciparum before and after a malaria season by protein microarray.
Proc. Natl. Acad. Sci. U.S.A. 2010; 107: 6958-6963
10.
, 11.
, 12.
, - Lessa-Aquino C.
- Rodrigues C.B.
- Pablo J.
- Sasaki R.
- Jasinskas A.
- Liang L.
- Wunder E.A.
- Ribeiro G.S.
- Vigil A.
- Galler R.
- Molina D.
- Liang X.
- Reis M.G.
- Ko A.I.
- Medeiros M.A.
- Felgner P.L.
Identification of seroreactive proteins of Leptospira interrogans serovar copenhageni using a high-density protein microarray approach.
PLoS Negl. Trop. Dis. 2013; 7: e2499
13.
). Although whole-protein arrays can successfully identify antigens recognized by antibodies, they present the typical difficulties associated with the production of recombinant proteins in heterologous or in vitro systems, do not provide information on the nature and precise location of the epitope(s) in a protein, and are more likely to suffer from nonspecific antibody binding because of the exposure of a large number of potentially antigenic regions. In contrast, peptide arrays can provide exquisite detail of epitope localization, but until now had other limitations mostly associated with their reduced capacity, preventing the complete scanning of large numbers of candidate proteins.Recent advances in computerized photolithography and photochemistry have led to the development of a novel high-density peptide microarray technology, where individual peptides can be synthesized in situ on a glass slide at high densities (
14.
, 15.
, 16.
, 17.
). This technology makes the production of high-density peptide arrays highly cost effective compared with previous approaches, while allowing the interrogation of complex immune responses with unprecedented throughput and mapping precision. Previous applications of this technology were limited to the fine mapping of epitopes in single proteins, using monoclonal antibodies, or using immunized animal sera as the source of polyclonal antibodies (16.
, 17.
, 18.
).Using these high-density peptide arrays, we here describe the first large-scale study of fine antibody specificities associated with Chagas Disease, which is an exemplar of a chronic human infectious disease. Chagas Disease, caused by the protozoan Trypanosoma cruzi, is an endemic disease of the Americas, affecting ∼8 million people (
19.
). The parasite invades and replicates within host cells, and briefly enters the bloodstream to reach other target tissues. Initially, the disease goes through an acute stage, characterized by patent parasitaemia and the appearance of antibodies against acute-phase antigens, such as SAPA (20.
), followed by a delayed specific humoral response. In general, the parasite-specific immune response mounted during T. cruzi infections is insufficient to completely eradicate the pathogen, leading to chronic infection (19.
). In this chronic stage circulating parasites are difficult to detect, even by extremely sensitive methods such as PCR. Therefore, detection of antibodies against whole-parasite extracts or defined antigens (21.
, 22.
) remains the standard for diagnosis of Chagas Disease.In this work, we screened high-density microarray slides containing peptides derived from T. cruzi proteins with mixtures of immunoglobulins purified directly from blood samples of Chagas Disease patients. This led to the identification of novel antigens and the simultaneous mapping of their linear B-cell epitopes, thus demonstrating the capacity and performance of this platform for studying antibody specificities associated with human infectious diseases.
DISCUSSION
This study provides so far the largest number of fine antibody specificities simultaneously measured for a human infection, with more than a 10-fold increment in throughput when compared with previous screenings using peptide microarrays (
7.
, 13.
, 42.
, 43.
). With only 20 μg of purified immunoglobulin directly obtained from human clinical samples, these high-density peptide microarrays allowed the recognition of specificities against ∼180,000 distinct pathogen-specific 15-mer peptides. Moreover, a sequential incubation protocol allowed subtraction of signal from healthy/non-infected human samples to obtain infection-specific signal in each microarray slide.We benchmarked the precision of our measurements using previously known antigens with mapped B-cell linear epitopes. These experiments showed that HD-Chips have an excellent epitope mapping performance, with AUCs values >0.96 in a single microarray and an average AUC of 0.972 after combining data from multiple samples. Further analysis showed that the deviation from a perfect AUC of 1 was only because of single-residue differences at the boundaries of the previously reported B-cell epitopes, particularly in the case of antigens with multiple repetitive epitopes. An independent validation in ELISA format of the TSSA antigen profile further supported the mapping precision of HD-Chips. In fact, we showed in this case that the major linear epitope within this antigen is located between residues 30 and 46 and not between residues 41 to 50 as detected using a less accurate technique (
39.
).The set of 15-mer peptides derived from randomly generated protein sequences was essential to define a nonspecific antibody-binding baseline distribution, and to normalize the data, allowing the subtraction of signal from incubations with negative samples, and bringing the data from different microarray replicates into a common scale. These data was also used to guide the selection of a reasonable threshold for classification of reactive peptides. Using this threshold, more than 2000 peptides, i.e. ∼1% of all screened 15-mer peptides, resulted seropositive in T. cruzi infected human sera pools and negative in the non-infected sera pools, with virtually no false positives (with the reactivity of all 24,000 randomly generated peptides below the cut-off). Although this threshold was defined based on the comparison against the reactivity of random peptides, it is already well established that both combinatorial libraries in phages (
1.
), or random peptides in high-density arrays (43.
, 44.
) can detect significant and reproducible binding by mimicking natural epitopes. Therefore, although we chose a very conservative threshold to identify the most reactive peptides, lowering this threshold can certainly provide additional Chagas-specific biomarkers.In our benchmark, epitope mapping performance was enhanced by subtracting reactivity signals measured in healthy donor samples, i.e. not related to the infection of interest, and perhaps caused by exposure of patients to other infectious agents or antigens with cross-reacting epitopes. Moreover, with this experimental setup the two samples were compared on the same physical spots, hence reducing the number of required chips by half and suppressing the variability associated with peptide synthesis and slide manipulation.
Based on the presented evidence it is clear that these HD-Chips are highly suitable for exhaustive mapping of the specificities of B-cell responses against human pathogens. Interestingly, antibody-binding to specific pathogen protein regions was also detected when using sera pools from healthy individuals (see for example the N-terminal region of the Ribosomal protein L19 in Fig. 1A, additional cases are also evident in supplemental Fig. S2). These antigenic determinants may be shared with other pathogens to which the donor has been exposed, or may represent cross-reactive determinants to other antigens (perhaps from vaccination antigens). In any case, the high-resolution scanning of antibody-binding across the full length of proteins is able to precisely map these regions, providing essential information for improving the specificity of recombinant antigens used in diagnostic applications.
In this context, it is interesting to compare the overall rate of antigen discovery provided by peptide and whole-protein arrays. Previous studies have profiled humoral immune responses in infected humans compared with healthy individuals for several pathogens using protein microarrays (
45.
). Some of the largest studies were performed for Mycobacterium tuberculosis (∼4000 proteins, representing almost the full proteome; with a 10% of seroreactive proteins) (- Davies D.H.
- Liang X.
- Hernandez J.E.
- Randall A.
- Hirst S.
- Mu Y.
- Romero K.M.
- Nguyen T.T.
- Kalantari-Dehaghi M.
- Crotty S.
- Baldi P.
- Villarreal L.P.
- Felgner P.L.
Profiling the humoral immune response to infection by using proteome microarrays: high-throughput vaccine and diagnostic antigen discovery.
Proc. Natl. Acad. Sci. U.S.A. 2005; 102: 547-552
8.
), Leptospira interrogans (∼3300 proteins, representing almost full proteome; with ∼5% of seroreactive proteins) (46.
), Brucella melitensis (∼3000 proteins, representing almost full proteome, with a ∼4% of seroreactive proteins) (47.
), and Plasmodium falciparum (∼1200 proteins or ∼23% of its proteome, with ∼13% of seroreactive proteins) (- Liang L.
- Tan X.
- Juarez S.
- Villaverde H.
- Pablo J.
- Nakajima-Sasaki R.
- Gotuzzo E.
- Saito M.
- Hermanson G.
- Molina D.
- Felgner S.
- Morrow W.J.W.
- Liang X.
- Gilman R.H.
- Davies D.H.
- Tsolis R.M.
- Vinetz J.M.
- Felgner P.L.
Systems biology approach predicts antibody signature associated with Brucella melitensis infection in humans.
J. Proteome Res. 2011; 10: 4813-4824
9.
). In our experiments we obtained a ∼1% global peptide positivity rate (∼2,031 positive out of 175,566 unique T. cruzi derived 15-mer peptides) when averaging signal from all HD-Chip experiments. This rate, goes down to ∼0.5% if we consider a non-redundant set of proteins with no prior serological characterization. In terms of full-length proteins, this corresponds to a ∼15% of the parasite proteins being targeted by the humoral immune response in infected subjects. Despite the fact that peptide microarrays are limited to identify linear epitopes, the estimated proportion of seroreactive proteins observed here for Chagas Disease is in line and even above those described in the aforementioned screenings using whole protein microarrays in other pathogens.- Crompton P.D.
- Kayala M.A.
- Traore B.
- Kayentao K.
- Ongoiba A.
- Weiss G.E.
- Molina D.M.
- Burk C.R.
- Waisberg M.
- Jasinskas A.
- Tan X.
- Doumbo S.
- Doumtabe D.
- Kone Y.
- Narum D.L.
- Liang X.
- Doumbo O.K.
- Miller L.H.
- Doolan D.L.
- Baldi P.
- Felgner P.L.
- Pierce S.K.
A prospective analysis of the Ab response to Plasmodium falciparum before and after a malaria season by protein microarray.
Proc. Natl. Acad. Sci. U.S.A. 2010; 107: 6958-6963
Therefore, we believe that high-density peptide microarrays offer an excellent platform for whole proteome screenings. In scenarios where the peptide space to screen is still too large, a combination of high-coverage whole-protein arrays followed by a detailed epitope scanning using high-density peptide chips could provide an excellent strategy to analyze antibody responses against pathogens. In this strategy, protein arrays would streamline the process of antigen identification, and HD peptide chips would precisely map the antigenic determinants in these antigens.
Besides the obvious practical impact of these large scale antigen screenings on the development of improved serodiagnostics and vaccines, unbiased and exhaustive studies of humoral immune response specificities in human infectious diseases are needed for a better understanding of pathogen immunogenicity and immunodominance, and to improve current antigenicity prediction tools (
48.
, 49.
). In our study, we successfully detected all antigenic determinants in high-confidence antigens with mapped B-cell epitopes. We were also able to discover and map the locations of epitopes for previously described antigens that lacked detailed B-cell epitope mappings. However, we failed to detect linear antigenic determinants for some of the proteins with previous serological evidence. Among other likely explanations, these antigens may contain only discontinuous, or low prevalence epitopes. Low prevalence of epitopes may be explained by the genetic heterogeneity of infecting parasites (50.
). Future studies using diverse serum samples (e.g. from diverse geographic origins, or from patients displaying different clinical manifestations of the disease) will help to answer this. We also note that the only epitopes that are readily mimicked by synthetic peptides are continuous epitopes. Discontinuous epitopes, which are made up of residues from separate stretches of the antigen polypeptide chain, can only be identified by x-ray crystallography of antigen-antibody complexes. However, we believe this limitation is largely balanced by the throughput and flexibility offered by high-density peptide microarrays.- Zingales B.
- Miles M.A.
- Campbell D.A.
- Tibayrenc M.
- Macedo A.M.
- Teixeira M.M.G.
- Schijman A.G.
- Llewellyn M.S.
- Lages-Silva E.
- Machado C.R.
- Andrade S.G.
- Sturm N.R.
{T}he revised {T}rypanosoma cruzi subspecific nomenclature: rationale, epidemiological relevance and research applications.
Infect. Genet. Evol. 2012; 12: 240-253
For many of these screenings seeking to identify antigens or defined antigenic determinants, a number of bioinformatic strategies based on defined protein features have been applied to narrow down the set of candidates displayed in the arrays (
25.
, 51.
, 52.
, 53.
). We have previously developed one such strategy to prioritize T. cruzi proteins and peptides for use in a spotted peptide microarray (25.
). In this work we found a significant ∼6 fold increment in the number of seroreactive peptides in this ranked set when compared with a random pick, thus supporting the utility of such approaches.In these experiments, we used a HD-Chip design were we simultaneously measured antibody-binding to >200,000 addressable spots in each slide. However, larger capacities (currently up to two million addressable spots) are possible by using a single mirror per peptide field. In this case, the larger capacity comes at the price of smaller fields and inherent larger measurement error. However, even using larger peptides fields, the screened protein space can be increased significantly by decreasing the sequence overlap between adjacent peptides in a protein sequence. The high signal correlation observed from peptides that are contiguous in the protein sequence suggested that it could be possible to further reduce the number of peptides required to completely scan each protein. In a posteriori computational experiments in which we re-analyzed the data simulating different peptide overlaps, we observed no significant epitope mapping performance loss down to an overlap of 12 residues between two 15-mers (i.e. a 3 residue shift when scanning a protein sequence) (not shown). This means that we should be able to achieve a ∼threefold increase in the protein search space using the same number of peptides per slide. Considering a large proteome such as that of T. cruzi, 33 HD-Chips would be required to screen the ∼6 million 15-mers in which this proteome can be broken down (200K peptide fields per chip, scanning proteins at maximal resolution). However, by scanning proteins with a 12-residue overlap and pushing the microarray density up to 500K peptides per slide, the complete proteome could be covered with only four HD-Chips, without sacrificing significant sensitivity.
In conclusion, by taking advantage of next-generation peptide arrays, we show that by screening ∼3% a large eukaryotic proteome we discovered and finely mapped more than 120 new antigenic determinants, providing essentially most of the linear B-cell epitopes currently known for this infectious disease. Our results show that it is now feasible to increase the pace of biomarker discovery for infectious diseases, and to further increase the scale and detail in the study of B-cell immune responses against human infectious diseases.
Acknowledgments
We thank Dr. Nick Thomson (Wellcome Trust Sanger Institute, Cambridge, UK) for critical reading of the manuscript, and Lic. María Gabriela Figini (IIB-INTECH, UNSAM) for technical support.
Supplementary Material
REFERENCES
- Lambda-display: a powerful tool for antigen discovery.Molecules. 2011; 16: 3089-3105
- Epitope discovery using peptide libraries displayed on phage.Trends Biotechnol. 1994; 12: 262-267
- Protein microarrays for highly parallel detection and quantitation of specific proteins and antibodies in complex solutions.Genome Biol. 2001; 2research0004
- Peptide microarrays for serum antibody diagnostics.Methods Mol. Biol. 2009; 509: 123-134
- Immune epitope database analysis resource.Nucleic Acids Res. 2012; 40: W525-W530
- From protein microarrays to diagnostic antigen discovery: a study of the pathogen Francisella tularensis.Bioinformatics. 2007; 23: i508-i518
- Pattern recognition in pulmonary tuberculosis defined by high content peptide microarray chip analysis representing 61 proteins from M. tuberculosis.PLoS One. 2008; 3: e3840
- Dynamic antibody responses tuberculosis to the Mycobacterium tuberculosis proteome.Proc. Natl. Acad. Sci. U.S.A. 2010; 107: 14703-14708
- A prospective analysis of the Ab response to Plasmodium falciparum before and after a malaria season by protein microarray.Proc. Natl. Acad. Sci. U.S.A. 2010; 107: 6958-6963
- Plasmodium immunomics.Int. J. Parasitol. 2011; 41: 3-20
- Profiling the humoral immune response of acute and chronic Q fever by protein microarray.Mol. Cell. Proteomics. 2011; 10M110.006304
- Identification of seroreactive proteins of Leptospira interrogans serovar copenhageni using a high-density protein microarray approach.PLoS Negl. Trop. Dis. 2013; 7: e2499
- Whole CMV proteome pattern recognition analysis after HSCT identifies unique epitope targets associated with the CMV status.PLoS One. 2014; 9: e89648
- Light-directed, spatially addressable parallel chemical synthesis.Science. 1991; 251: 767-773
- Individually addressable parallel peptide synthesis on microchips.Nat. Biotechnol. 2002; 20: 922-926
- High-resolution mapping of linear antibody epitopes using ultrahigh-density peptide microarrays.Mol. Cell. Proteomics. 2012; 11: 1790-1800
- Proteome-wide epitope mapping of antibodies using ultra-dense peptide arrays.Mol. Cell. Proteomics. 2014; 13: 1585-1597
- Identification and mapping of linear antibody epitopes in human serum albumin using high-density peptide arrays.PLoS One. 2013; 8: e68902
- Chagas disease.Lancet. 2010; 375: 1388-1402
- Identification of a Trypanosoma cruzi antigen that is shed during the acute phase of Chagas' disease.Mol. Biochem. Parasitol. 1989; 34: 221-228
- Assay for detection of Trypanosoma cruzi antibodies in human sera based on reaction with synthetic peptides.J. Clin. Microbiol. 1991; 29: 2034-2037
- Evaluation of a recombinant Trypanosoma cruzi mucin-like antigen for serodiagnosis of Chagas' disease.Clin. Vaccine Immunol. 2011; 18: 1850-1855
- TriTrypDB: a functional genomic resource for the Trypanosomatidae.Nucleic Acids Res. 2010; 38: D457-D462
- Assessing performance of orthology detection strategies applied to eukaryotic genomes.PLoS One. 2007; 2: e383
- Diagnostic peptide discovery: prioritization of pathogen diagnostic markers using multiple features.PLoS One. 2012; 7: e50748
- TcSNP: a database of genetic variation in Trypanosoma cruzi.Nucleic Acids Res. 2009; 37: D544-D549
- A genomic scale map of genetic diversity in Trypanosoma cruzi.BMC Genomics. 2012; 13: 736
- Genomic organization and expression profile of the mucin-associated surface protein (masp) family of the human pathogen Trypanosoma cruzi.Nucleic Acids Res. 2009; 37: 3407-3417
- The MASP family of Trypanosoma cruzi: changes in gene expression and antigenic profile during the acute phase of experimental infection.PLoS Negl. Trop. Dis. 2012; 6: e1779
- TcTASV: a novel protein family in Trypanosoma cruzi identified from a subtractive trypomastigote cDNA library.PLoS Negl Trop Dis. 2010; 4
- TcTASV-C, a protein family in Trypanosoma cruzi that is predominantly trypomastigote-stage specific and secreted to the medium.PLoS One. 2010; 8: e71192
- NNAlign: a web-based prediction method allowing non-expert end-user discovery of sequence motifs in quantitative peptide data.PLoS One. 2011; 6: e26781
- The outermost N-terminal region of tapasin facilitates folding of major histocompatibility complex class I.Eur. J. Immunol. 2009; 39: 2682-2694
- A comparison of normalization methods for high density oligonucleotide array data based on variance and bias.Bioinformatics. 2003; 19: 185-193
- ROCR: visualizing classifier performance in R.Bioinformatics. 2005; 21: 3940-3941
- Tracking repeats using significance and transitivity.Bioinformatics. 2004; 1: i311-i317
- Mapping Antigenic Motifs in the Trypomastigote Small Surface Antigen from Trypanosoma cruzi.Clin. Vaccine Immunol. 2015; 22: 304-312
- Structural analysis of B-cell epitopes in antibody:protein complexes.Mol. Immunol. 2013; 53: 24-34
- A Trypanosoma cruzi small surface molecule provides the first immunological evidence that Chagas' disease is due to a single parasite lineage.J. Exp. Med. 2002; 195: 401-413
- Immunological identification of Trypanosoma cruzi lineages in human infection along the endemic area.Am. J. Trop. Med. Hyg. 2011; 84: 78-84
- The genome sequence of Trypanosoma cruzi, etiologic agent of Chagas disease.Science (80-.). 2005; 309: 409-415
- On silico peptide microarrays for high-resolution mapping of antibody epitopes and diverse protein-protein interactions.Nat. Med. 2012; 18: 1434-1440
- Immunosignature system for diagnosis of cancer.Proc. Natl. Acad. Sci. U.S.A. 2014; 111: E3072-E3080
- Scalable high-density peptide arrays for comprehensive health monitoring.Nat. Commun. 2014; 5: 4785
- Profiling the humoral immune response to infection by using proteome microarrays: high-throughput vaccine and diagnostic antigen discovery.Proc. Natl. Acad. Sci. U.S.A. 2005; 102: 547-552
- Proteomic Features Predict Seroreactivity against Leptospiral Antigens in Leptospirosis Patients.J. Proteome Res. 2015; 14: 549-556
- Systems biology approach predicts antibody signature associated with Brucella melitensis infection in humans.J. Proteome Res. 2011; 10: 4813-4824
- Improved method for predicting linear B-cell epitopes.Immunome Res. 2006; 2: 2
- Designing immunogenic peptides.Nat. Chem. Biol. 2013; 9: 749-753
- {T}he revised {T}rypanosoma cruzi subspecific nomenclature: rationale, epidemiological relevance and research applications.Infect. Genet. Evol. 2012; 12: 240-253
- High throughput selection of effective serodiagnostics for Trypanosoma cruzi infection.PLoS Negl. Trop. Dis. 2008; 2: e316
- Serodiagnosis of Echinococcus spp. infection: explorative selection of diagnostic antigens by peptide microarray.PLoS Negl. Trop. Dis. 2010; 4: e771
- High-throughput prediction of protein antigenicity using protein microarray data.Bioinformatics. 2010; 26: 2936-2943
Article Info
Publication History
Published online: April 28, 2015
Accepted:
March 26,
2015
Received:
October 27,
2014
Footnotes
Author contributions: S.J.C. and F.A. designed research; S.J.C., C.S., V.B., C.A.B., and F.A. performed research; C.S. and J.A. contributed new reagents or analytic tools; S.J.C., M.N., J.M., V.T., A.C.F., O.C., C.A.B., and F.A. analyzed data; S.J.C., M.N., C.A.B., and F.A. wrote the paper.
DISCLOSURE DECLARATION. Claus Schafer-Nielsen is the owner and CEO of Schafer-N ApS. The other authors declare no conflict of interest.
Identification
Copyright
© 2015 ASBMB. Currently published by Elsevier Inc; originally published by American Society for Biochemistry and Molecular Biology.
User License
Creative Commons Attribution (CC BY 4.0) | How you can reuse 
Elsevier's open access license policy
Creative Commons Attribution (CC BY 4.0)
Permitted
- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article
- Reuse portions or extracts from the article in other works
- Sell or re-use for commercial purposes
Elsevier's open access license policy
