Cross-specificity of protective human antibodies against Klebsiella pneumoniae LPS O-antigen

Abstract

Humoral immune responses to microbial polysaccharide surface antigens can prevent bacterial infection but are typically strain specific and fail to mediate broad protection against different serotypes. Here we describe a panel of affinity-matured monoclonal human antibodies from peripheral blood immunoglobulin M–positive (IgM+) and IgA+ memory B cells and clonally related intestinal plasmablasts, directed against the lipopolysaccharide (LPS) O-antigen of Klebsiella pneumoniae, an opportunistic pathogen and major cause of antibiotic-resistant nosocomial infections. The antibodies showed distinct patterns of in vivo cross-specificity and protection against different clinically relevant K. pneumoniae serotypes. However, cross-specificity was not limited to K. pneumoniae, as K. pneumoniae–specific antibodies recognized diverse intestinal microbes and neutralized not only K. pneumoniae LPS but also non–K. pneumoniae LPS. Our data suggest that the recognition of minimal glycan epitopes abundantly expressed on microbial surfaces might serve as an efficient humoral immunological mechanism to control invading pathogens and the large diversity of the human microbiota with a limited set of cross-specific antibodies.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.

from$8.99

All prices are NET prices.

Fig. 1: Identification of human K. pneumoniae O3–reactive PB memory B cells and intestinal LP plasmablasts through the use of biotinylated O-antigen.
Fig. 2: Relevance of somatic mutations for K. pneumoniae O3 reactivity.
Fig. 3: Cross-specificity of O3 antibodies with mannan-based K. pneumoniae O-serotypes.
Fig. 4: Functional activity of a selected cross-specific O3 antibody to various K. pneumoniae serotypes.
Fig. 5: Cross-specificity of O3-reactive antibodies with non–K. pneumoniae microbes.
Fig. 6: Cross-specificity of O3-reactive antibodies with intestinal microbes.

References

  1. 1.

    Julianelle, L. A. Immunological relationships of encapsulated and capsule-free strains of encapsulated pneumoniae (Friedlander’s Bacillus). J. Exp. Med. 44, 683–696 (1926).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  2. 2.

    Lancefield, R. C. A serological differentiation of human and other groups of hemolytic Streptococci. J. Exp. Med. 57, 571–595 (1933).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  3. 3.

    Shattock, P. M. F. The use of serology in the classification of micro-organisms. J. Gen. Microbiol. 12, 367–374 (1955).

    Article  PubMed  CAS  Google Scholar 

  4. 4.

    Tillett, W. S., Avery, O. T. & Goebel, W. F. Chemo-immunological studies in conjugated carbohydrate-proteins: III. Active and passive anaphylaxis with synthetic sugar-proteins. J. Exp. Med. 50, 551–567 (1929).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  5. 5.

    Vinogradov, E. et al. Structures of lipopolysaccharides from Klebsiella pneumoniae. Eluicidation of the structure of the linkage region between core and polysaccharide O chain and identification of the residues at the non-reducing termini of the O chains. J. Biol. Chem. 277, 25070–25081 (2002).

    Article  PubMed  CAS  Google Scholar 

  6. 6.

    Trautmann, M. et al. O-antigen seroepidemiology of Klebsiella clinical isolates and implications for immunoprophylaxis of Klebsiella infections. Clin. Diagn. Lab. Immunol. 4, 550–555 (1997).

    PubMed  PubMed Central  CAS  Google Scholar 

  7. 7.

    Hansen, D. S. et al. Klebsiella pneumoniae lipopolysaccharide O typing: revision of prototype strains and O-group distribution among clinical isolates from different sources and countries. J. Clin. Microbiol. 37, 56–62 (1999).

    PubMed  PubMed Central  CAS  Google Scholar 

  8. 8.

    Follador, R. et al. The diversity of Klebsiella pneumoniae surface polysaccharides. Microb. Genomics 2, e000073 (2016).

    Article  Google Scholar 

  9. 9.

    Podschun, R. & Ullmann, U. Klebsiella spp. as nosocomial pathogens: epidemiology, taxonomy, typing methods, and pathogenicity factors. Clin. Microbiol. Rev. 11, 589–603 (1998).

    PubMed  PubMed Central  CAS  Article  Google Scholar 

  10. 10.

    Kubinak, J. L. & Round, J. L. Do antibodies select a healthy microbiota? Nat. Rev. Immunol. 16, 767–774 (2016).

    Article  PubMed  CAS  Google Scholar 

  11. 11.

    Mond, J. J., Lees, A. & Snapper, C. M. T. T cell-independent antigens type 2. Annu. Rev. Immunol. 13, 655–692 (1995).

    Article  PubMed  CAS  Google Scholar 

  12. 12.

    Magri, G. et al. Human secretory IgM emerges from plasma cells clonally related to gut memory B cells and targets highly diverse commensals. Immunity 47, 118–134 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  13. 13.

    Bunker, J. J. et al. Natural polyreactive IgA antibodies coat the intestinal microbiota. Science 358, eaan6619 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  14. 14.

    Toellner, K.-M. et al. Low-level hypermutation in T cell-independent germinal centers compared with high mutation rates associated with T cell-dependent germinal centers. J. Exp. Med. 195, 383–389 (2002).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  15. 15.

    Whitfield, C., Richards, J. C., Perry, M. B., Clarke, B. R. & MacLean, L. L. Expression of two structurally distinct D-galactan O antigens in the lipopolysaccharide of Klebsiella pneumoniae serotype O1. J. Bacteriol. 173, 1420–1431 (1991).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  16. 16.

    Szijártó, V. et al. Both clades of the epidemic KPC-producing Klebsiella pneumoniae clone ST258 share a modified galactan O-antigen type. Int. J. Med. Microbiol. 306, 89–98 (2016).

    Article  PubMed  CAS  Google Scholar 

  17. 17.

    Stojkovic, K. et al. Identification of d-galactan-III as part of the lipopolysaccharide of Klebsiella pneumoniae Serotype O1. Front. Microbiol. 8, 684 (2017).

    Article  PubMed  PubMed Central  Google Scholar 

  18. 18.

    Guachalla, L. M. et al. Discovery of monoclonal antibodies cross-reactive to novel subserotypes of K. pneumoniae O3. Sci. Rep. 7, 6635 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  19. 19.

    Park, B. S. et al. The structural basis of lipopolysaccharide recognition by the TLR4–MD-2 complex. Nature 458, 1191–1195 (2009).

    Article  PubMed  CAS  Google Scholar 

  20. 20.

    Jansson, P.-E. et al. Structural studies of the O-antigen polysaccharides of Klebsiella O5 and Escherichia coli O8. Carbohydr. Res. 145, 59–66 (1985).

    Article  PubMed  CAS  Google Scholar 

  21. 21.

    Lipke, P. N. & Ovalle, R. Cell wall architecture in yeast: new structure and new challenges. J. Bacteriol. 180, 3735–3740 (1998).

    PubMed  PubMed Central  CAS  Google Scholar 

  22. 22.

    Doores, K. J. The HIV glycan shield as a target for broadly neutralizing antibodies. FEBS J. 282, 4679–4691 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  23. 23.

    Trama, A. M. et al. HIV-1 envelope gp41 antibodies can originate from terminal ileum B cells that share cross-reactivity with commensal bacteria. Cell Host Microbe 16, 215–226 (2014).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  24. 24.

    Bayhan, G. I., Senel, S., Tanir, G. & Ozkan, S. Bacteremia caused by Pseudomonas luteola in pediatric patients. Jpn. J. Infect. Dis. 68, 50–54 (2015).

    Article  PubMed  Google Scholar 

  25. 25.

    Zhou, Z.-H. et al. The broad antibacterial activity of the natural antibody repertoire is due to polyreactive antibodies. Cell Host Microbe 1, 51–61 (2007).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  26. 26.

    Polonskaya, Z. et al. T cells control the generation of nanomolar-affinity anti-glycan antibodies. J. Clin. Invest. 127, 1491–1504 (2017).

    Article  PubMed  PubMed Central  Google Scholar 

  27. 27.

    Fagarasan, S., Kinoshita, K., Muramatsu, M., Ikuta, K. & Honjo, T. In situ class switching and differentiation to IgA-producing cells in the gut lamina propria. Nature 413, 639–643 (2001).

    Article  PubMed  CAS  Google Scholar 

  28. 28.

    Lindner, C. et al. Diversification of memory B cells drives the continuous adaptation of secretory antibodies to gut microbiota. Nat. Immunol. 16, 880–888 (2015).

    Article  PubMed  CAS  Google Scholar 

  29. 29.

    Bergqvist, P. et al. Re-utilization of germinal centers in multiple Peyer’s patches results in highly synchronized, oligoclonal, and affinity-matured gut IgA responses. Mucosal Immunol. 6, 122–135 (2013).

    Article  PubMed  CAS  Google Scholar 

  30. 30.

    Di Niro, R. et al. Salmonella infection drives promiscuous B cell activation followed by extrafollicular affinity maturation. Immunity 43, 120–131 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  31. 31.

    Baumgarth, N., Tung, J. W. & Herzenberg, L. A. Inherent specificities in natural antibodies: a key to immune defense against pathogen invasion. Springer Semin. Immunopathol. 26, 347–362 (2005).

    Article  PubMed  CAS  Google Scholar 

  32. 32.

    Landsverk, O. J. B. et al. Antibody-secreting plasma cells persist for decades in human intestine. J. Exp. Med. 214, 309–317 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  33. 33.

    Macpherson, A. J., Köller, Y. & McCoy, K. D. The bilateral responsiveness between intestinal microbes and IgA. Trends Immunol. 36, 460–470 (2015).

    Article  PubMed  CAS  Google Scholar 

  34. 34.

    Coughlin, R. T. et al. Characterization of pneumococcal specific antibodies in healthy unvaccinated adults. Vaccine 16, 1761–1767 (1998).

    Article  PubMed  CAS  Google Scholar 

  35. 35.

    Peterson, D. A., McNulty, N. P., Guruge, J. L. & Gordon, J. I. IgA response to symbiotic bacteria as a mediator of gut homeostasis. Cell Host Microbe 2, 328–339 (2007).

    Article  PubMed  CAS  Google Scholar 

  36. 36.

    Tiller, T. et al. Autoreactivity in human IgG+ memory B cells. Immunity 26, 205–213 (2007).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  37. 37.

    Benckert, J. et al. The majority of intestinal IgA+ and IgG+ plasmablasts in the human gut are antigen-specific. J. Clin. Invest. 121, 1946–1955 (2011).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  38. 38.

    Tiller, T. et al. Efficient generation of monoclonal antibodies from single human B cells by single cell RT-PCR and expression vector cloning. J. Immunol. Methods 329, 112–124 (2008).

    Article  PubMed  CAS  Google Scholar 

  39. 39.

    Murugan, R., Imkeller, K., Busse, C. E. & Wardemann, H. Direct high-throughput amplification and sequencing of immunoglobulin genes from single human B cells. Eur. J. Immunol. 45, 2698–2700 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  40. 40.

    Lorin, V. & Mouquet, H. Efficient generation of human IgA monoclonal antibodies. J. Immunol. Methods 422, 102–110 (2015).

    Article  PubMed  CAS  Google Scholar 

  41. 41.

    Wardemann, H. et al. Predominant autoantibody production by early human B cell precursors. Science 301, 1374–1377 (2003).

    Article  PubMed  CAS  Google Scholar 

  42. 42.

    Szijártó, V. et al. Bactericidal monoclonal antibodies specific to the lipopolysaccharide O antigen from multidrug-resistant Escherichia coli clone ST131-O25b:H4 elicit protection in mice. Antimicrob. Agents Chemother. 59, 3109–3116 (2015).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  43. 43.

    Cerca, N. et al. Protection against Escherichia coli infection by antibody to the Staphylococcus aureus poly-N-acetylglucosamine surface polysaccharide. Proc. Natl Acad. Sci. USA 104, 7528–7533 (2007).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  44. 44.

    Szijártó, V. et al. Endotoxin neutralization by an O-antigen specific monoclonal antibody: A potential novel therapeutic approach against Klebsiella pneumoniae ST258. Virulence 8, 1203–1215 (2017).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  45. 45.

    Masella, A. P., Bartram, A. K., Truszkowski, J. M., Brown, D. G. & Neufeld, J. D. PANDAseq: paired-end assembler for illumina sequences. BMC Bioinformatics 13, 31 (2012).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  46. 46.

    Caporaso, J. G. et al. QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336 (2010).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  47. 47.

    R Development Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, Vienna, 2008).

  48. 48.

    Imkeller, K., Arndt, P. F., Wardemann, H. & Busse, C. E. sciReptor: analysis of single-cell level immunoglobulin repertoires. BMC Bioinformatics 17, 67 (2016).

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  49. 49.

    Warnes, G. R. et al. Various R programming tools for plotting data. R package version 2, 1 (2009).

    Google Scholar 

  50. 50.

    Wickham, H. ggplot2: Elegant Graphics for Data Analysis. (Springer-Verlag, New York, 2009).

    Google Scholar 

  51. 51.

    Gu, Z., Gu, L., Eils, R., Schlesner, M. & Brors, B. circlize Implements and enhances circular visualization in R. Bioinformatics 30, 2811–2812 (2014).

    Article  PubMed  CAS  Google Scholar 

Download references

Acknowledgements

We thank S. Kaluzewski (National Institute of Public Health–National Institute of Hygiene) for an unencapsulated mutant of a prototype O3:K55 strain 5505; S. Melegh (University of Pécs) for clinical isolates Kp2 (O3:K60) and Kp14 (O3:K35); A. Valverde (University Hospital Ramón y Cajal) for clinical isolate Kp81 (O3b:K non-typeable); A. Heidtmann, L. Kummer, B. Jocher and J. Braun for help with sample acquisition; C. Varga for performing animal experiments; C. Busse for help with experimental design; M. Nussenzweig (The Rockefeller University) for recombinant gp140 from strain YU2; and P. Sehr, R. Murugan, P. Thiele, C. Winter, D. Foster, A. Götze and the DKFZ High-Throughput Sequencing Unit of the Genomics and Proteomics Core Facility and the DKFZ and MPIIB Flow Cytometry Core Facilities for technical support. Supported by the European Funding Program Eurostars (E! 7563 – KLEBSICURE), powered by EUREKA and the European Community; the German Federal Ministry of Education and Research (H.W.); and the International Max Planck Research School for Infectious Diseases and Immunology (T.R.).

Author information

Affiliations

Authors

Contributions

T.R., V.S., J.L. and H.W. designed experiments and analyzed data; T.R., V.S., L.M.G., K.S., K. Hartl, L.S. and S. Kocher performed experiments; F.L., M.A.-S., J.S.-B., M.v.F., G.G. and P.H. identified patients and provided samples. S. Klein and K. Heeg provided experimental support; V.S., E.N., G.N. and H.W. conceived of the study; T.R. and H.W. wrote the manuscript; V.S., J.L., E.N. and G.N. edited the manuscript; and all authors approved the final version of the manuscript.

Corresponding author

Correspondence to Hedda Wardemann.

Ethics declarations

Competing interests

E.N., G.N., L.M.G., K. Hartl, L.S., V.S., T.R. and H.W. have filed patent applications on human antibodies to K. pneumoniae O-antigen; and E.N., G.N., L.M.G., K. Hartl, L.S. and V.S. are employees of Arsanis Biosciences and hold shares in Arsanis, the parent company of Arsanis Biosciences.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Integrated supplementary information

Supplementary Figure 1 Identification of O3-reactive B cells using biotinylated O3 O-antigen.

a, Schematic linear structure of biotinylated O3 K. pneumoniae O-antigen; repeating unit (outlined), MeP = Methylphosphate; Man = Mannose; KdO = 3-Deoxy-D-manno-oct-2-ulosonic acid; GlcNAc = N-acetylglucosamine; Hep = L-glycero-D-manno-heptose; PEG = Polyethylene glycol. b, Percentage of O3-reactive 7-AAD, CD19+, CD27+ peripheral blood B cells in eight donors compared to the background staining using non-biotinylated O3 (right). p-value was calculated using two-tailed Student’s t-test. Data are representative of at least two independent experiments.

Supplementary Figure 2 Immunoglobulin gene characteristics of O3-reactive PB and LP B cells.

a, IGH gene isotype distribution for O3-reactive PB memory B cells isolated by flow cytometric single cell sorting. b,c, IGHV gene family usage for O3-reactive PB memory B cells of the indicated isotypes from the same donors as shown in panel a (b) and LP plasmablasts isolated by flow cytometric single cell sorting from the indicated donors (c). d, Percentage of mutated and unmutated IGHV genes from O3-reactive PB memory B cells of the same donors as in panel a. e, Distribution of cell clusters of clonally related (gray boxes) compared to unique non-related (white) O3-reactive B cells in PB memory B cells or LP plasmablasts from the indicated donors. f, Clonal relationship of O3-reactive B cells of the indicated isotypes in PB memory B cells and LP plasmablasts of donors HD02, HD05, and HD10. Colors depict individual B cell clusters. n indicates numbers of analyzed sequences.

Supplementary Figure 3 O3-reactive PB memory B cell and LP plasmablast antibodies bind specifically to O3 O-antigen.

a, Immunoblot with whole O3 LPS for representative mAbs cloned from O3-reactive memory B cells from PB (036, 286) and from LP plasmablasts (601). b, O3 and streptavidin reactivity of individual recombinant mAbs from O3-reactive PB memory B cells (n = 41) and LP plasmablasts (n = 41) based on ELISA area under curve (AUC) measurements. c, Insulin ELISA reactivity of O3-reactive PB memory B cell antibodies (n = 41) under blocking conditions with 2% BSA. d, Polyreactivity ELISA with E. coli LPS, dsDNA and insulin under non-blocking conditions without BSA. Representative mAbs compared to the positive and negative control mAbs are shown. Dashed lines in indicate negative thresholds (b-d). Data are representative data of at least two independent experiments (a-d).

Supplementary Figure 4 Germline reversion of mutated O3-reactive antibodies.

Gene sequence alignment of IGH and IGK or IGL variable regions of the indicated mutated mAbs and their inferred germline counterpart compared to the respective germline V and J gene segments. CDRs are highlighted.

Supplementary Figure 5 O5-specific mAb 703.

a, mAb 703 shows specificity for O5 in immunoblot with whole LPS. b, Reactivity of mAb 703 with intestinal murine microbes as measured by flow cytometry. Sorting gate and percentage of antibody-bound bacteria is indicated. Boxes show the K. pneumoniae O-serotype, S. cerevisiae, and gp140 reactivity profile (strong-black; non-reactive-white). Data are representative of two independent experiments.

Supplementary Figure 6 Immunoglobulin gene features and antibody characteristics of O1-reactive PB B cells.

a, Schematic linear structure of biotinylated O1; repeating units (outlined); Gal = Galactose; KdO = 3-Deoxy-D-manno-oct-2-ulosonic acid; GlcNAc = N-acetylglucosamine; Hep = L-glycero-D-manno-heptose; PEG = Polyethylene glycol. b, Flow cytometric analysis of the frequency of 7-AAD, CD19+ PB B cells reactive to biotinylated O1 (right) compared to the negative control using non-biotinylated O1 (left) for a representative healthy donor detected by Streptavidin Alexa647 conjugate. Gates and percentages indicate the frequency of O1-reactive cells. c, Percentage of O1-reactive 7-AAD, CD19+ PB B cells (left) in eight donors compared to the background staining using non-biotinylated O1 (right). p-value was calculated using two-tailed Student’s t-test. d, IGH gene isotype usage for O1-reactive PB memory B cells from the same donors as in panel c. Means (red lines) and SD (black lines) are indicated. e, IGHV gene family usage for O1-reactive PB memory B cells of the indicated isotypes from the same donors as shown in panel d. n represents the number of analyzed IGH sequences per isotype. f, Percent of clonally-related O1-reactive B cells per sample in PB memory B cells of the same donors as in panel d. Means (red lines) and SD (black lines) are indicated. g, Number of somatic hypermutations in IGHV genes of the indicated isotypes from O1-reactive peripheral blood B cells. Data are pooled from the same eight donors as in panel d. Means (red lines) and SD (black lines) are indicated. n represents the number of analyzed IGH sequences per isotype. h, Percentage of mutated and unmutated IGHV genes from O1-reactive PB memory B cells. i, O1-reactivity of mAbs cloned from O1-reactive PB memory B cells as measured by ELISA. Green lines indicate the negative isotype control mAb and red lines the threshold for positivity. n indicates the number of tested mAbs. j, Immunoblot reactivity with O1, O2, and O3 LPS for a representative O1-reactive (anti-galactan II) mAb; + and - indicate serotypes expressing the galactan III or the galactan I variant, respectively. k, O1 and streptavidin reactivity of individual mAbs from O1-reactive PB memory B cells based on ELISA area under curve (AUC) measurements. l, IGHV gene somatic hypermutation counts for ELISA O1-specific mAbs (i, k) of the indicated isotypes (IGHM, n = 3; IGHA, n = 3). Red lines indicate means. m, Anti-O1 ELISA reactivity of the indicated mAb (top) and its predicted germline form (bottom). Green lines indicate reactivity of the negative isotype control mAb. Red lines indicate negative thresholds. n, Reactivity of the O1-reactive mAb 196 with murine intestinal microbes as measured by flow cytometry (Fig. 6a). Sorting gate and percentage of antibody-bound bacteria is indicated. Boxes show the K. pneumoniae O-serotype, S. cerevisiae, and gp140 reactivity profile (non-reactive-white). Red diagonal lines indicate non-tested. Data are representative of at least two independent experiments (b, c, i – k, m, n).

Supplementary Figure 7 Cross-specificity of individual K. pneumoniae O-antigen reactive antibodies with microbes from human intestine.

a,b, Reactivity of the indicated mAbs with intestinal microbes of 31/41 tested human stool samples (a) and from a single individual as measured by flow cytometry compared to the PBS negative control (b). Sorting gate and percentage of antibody-bound bacteria are indicated. The K. pneumoniae O-serotype, S. cerevisiae, and gp140 reactivity profiles of each antibody are shown. Data are representative of two independent experiments (b).

Supplementary Figure 8 Flow cytometry gating strategy.

a-e, Flow cytometry gating strategies for O3-reactive PB memory B cells (a; Fig. 1) and O1- and O3-reactive LP plasmablasts (b; Fig. 1) as well as for mAbs with reactivity for K. pneumoniae (c; Fig. 3), a representative human microbiota sample (d; Fig. 6), and P. luteola (e; Fig. 6).

Supplementary information

Supplementary Figures and Supplementary Text

Supplementary Figures 1–8 and Supplementary Tables 1-3

Reporting Summary

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Rollenske, T., Szijarto, V., Lukasiewicz, J. et al. Cross-specificity of protective human antibodies against Klebsiella pneumoniae LPS O-antigen. Nat Immunol 19, 617–624 (2018). https://doi.org/10.1038/s41590-018-0106-2

Download citation

Further reading