tmpA3A1 TMP
tmpA3A1 TMP
tmpA3A1 TMP
90
Ofcial journal of the Australasian Society for Immunology Inc.
www.nature.com/cti
THEORETICAL ARTICLE
In a pair of articles, we present a generalized quantitative model for the homeostatic function of clonal humoral immune system.
In this second paper, we describe how antibody production controls the saturation of antigens and the network of antibody
interactions that emerges in the epitome space with the establishment of the immune system. Efcient control of antigens, be it
self or foreign, requires the maintenance of antibody concentrations that saturate antigen to relevant levels. Simple calculations
suggest that the observed diverse recognition of antigens by natural antibodies is only possible by cross-reactivity whereby
particular clones of antibodies bind to diverse targets and shared recognition of particular antigens by multiple antibody
clones contribute to the maintenance of antigen control. We also argue that natural antibodies are none else than the result of
thymus-independent responses against immunological self. We interpret and explain antibody production and function in a
virtual molecular interaction space and as a network of interactions. Indeed, the general quantitative (GQM) model we propose is
in agreement with earlier models, conrms some assumptions and presumably provides the theoretical basis for the construction
of a real antibody network using the sequence and interaction database data.
Clinical & Translational Immunology (2017) 6, e131; doi:10.1038/cti.2016.90; published online 17 February 2017
THE GQM APPLIED TO ANTIBODY HOMEOSTASIS apparent afnity to indicate that factors like multiple binding sites
By denition, antigens are molecules recognized by antibodies. Most modulate the observed strength of the interaction.
denitions however fail to further elaborate what exactly is meant by Assuming that antibodies are produced with the intent of regulating
recognition. The strength of the interaction between the antigen- antigen availability, best control over antigen concentration is achieved
binding site (paratope) of an antibody and the antibody-binding site when the concentration of antibody is close to the KD (Figure 1). In
(epitope) of the antigen is characterized by afnity, kinetics of our map, this zone for a range of [Ab] and KD values is dened by a
association and dissociation, and binding energy. Antibodies often line, where [Ab] = KD, which is the line representing 50% saturation of
recognize more than one target. Immunological assays usually require the antigen (Figure 1). By lowering or increasing antibody production,
the titration of the antibody, which is the identication of lowest the host can release or capture antigens, and likewise by changing the
concentration that binds to the nominal target but does not bind to efciency of Ab binding, the host can modulate antigen saturation
others. This is quite logical for antibodies intentionally produced in (Figure 1). Various immunological mechanisms are responsible for
animals, but how we dene the target of an antibody in vivo? By removing antibodyantigen complexes, called immune complexes,
changing the concentration of antigen and antibody, saturation of any from the circulation.
can be achieved even when afnity of the interaction is low. The The range of [Ab] values we will be using in our model reect
absolute and relative concentration of antigens and antibodies does actual immunoglobulin concentrations in blood plasma, and start
matter and our GQM attempts to reveal antibody function by around the concentration that a single plasma cell could achieve by
addressing these factors. continuous secretion of antibody. The range of KD values includes
The general equation dening equilibrium dissociation constant KD: afnity constants usually observed1 for antibodyantigen interactions
(of KD = 10 610 10 M) but extends to both lower and higher values
AbAg to provide exibility for interpreting apparent afnities. Please note
KD
AbAg that these are exactly the same dimensions, which we use in our
tells us that KD = [Ab] when [Ag] = [AbAg]. That is when antigen is accompanying sister paper on B-cell development.2 Let us now analyze
half saturated, free antibody concentration is equal to KD. For the sake various characteristic immune responses in the order of increasing
of simplicity, we will regard [Ab] as the concentration of paratope and antibodyantigen interaction afnity. We will consider a single uidic
[Ag] as the concentration of epitope and we shall use the term compartment, the blood plasma for this theoretical framework,
1
R&D Laboratory, Diagnosticum zrt, Budapest, Hungary and 2MTA-ELTE Immunology Research Group, at Etvs Lornd University, Budapest, Hungary
Correspondence: Dr J Prechl, R&D Laboratory, Diagnosticum zrt, Attila ut 126, Budapest 1047, Hungary.
E-mail: jprechl@diagnosticum.hu
Received 26 August 2016; revised 15 November 2016; accepted 24 November 2016
A quantitative antibody homeostasis model II
J Prechl
2
Figure 1 Outlines of the GQM for regulation of antibody production. (a) Antibodies will saturate antigen by increasing their concentration or by increasing
apparent afnity. (b) Low concentrations of low-afnity antibodies do not bind antigen at relevant extent, antigen concentrations can be best controlled at
around 50% saturation, while elimination is achieved by increasing saturation further. (c) ASC produce antibodies to increase concentration, while prior
differentiation account for increased afnity or ability to remove antigen. Immune complexes are removed by different cells and silent or proinammatory
events.
however, with proper adjustments the model can be possibly extended As we have discussed in our accompanying article, our GQM
to include the extracellular space and mucosal surfacessites of key assumes that B1I cells develop from immature B cells in blood, as a
importance for immunological action. result of continuous BCR signaling triggered by blood-borne antigens.
Their precursors, immature B cells were selected based on their
NATURAL ANTIBODIES AND TI ANTIBODY RESPONSES polyreactive property, tested by self-antigen displayed on the devel-
Can low-afnity antibodies mediate any relevant biological effect at all? oping cells. Polyreactivity against common, shared molecular targets
For an antibody with KD = 10 6 M a concentration of 10 6 M should allows clusters of antibodies to cooperatively bind to highly abundant
be reached for substantial binding to its target, which is quite close to self-antigens, reaching relevant fractional saturation despite the
the total immunoglobulin concentration in plasma (Figure 2). relatively low afnity of the interactions. Any particular antibody
Obviously, no single antibody can dominate to such an extent can belong to several different such clusters, thereby increasing the
(except for pathological antibodies in disease, like monoclonal concentration of antibodies against that particular antigen. In other
gammopathies). Multiple antibody-binding sites on the antigen words, by shared, distributed recognition several different antigenic
increase the apparent afnity of the interaction by avidity effects, targets can be saturated by the same amount of antibody. Cross-
but not to the value required here. Most likely, a combination of these reactivity increases the apparent clonal diversity because clusters of the
effects, large cumulative concentration achieved by a large number of same network respond to different targets. Removal of cellular debris,
cross-reacting antibody producing B-cell clones and avidity might which contains a huge number of different molecules may become
confer effector functions to low-afnity antibodies. possible this way by a limited number of specicities and antibodies.
Natural antibodies are low-afnity antibodies, constitutively pro- Polyspecic, low-afnity, high off-rate interactions in the plasma result
duced by B1I cell populations that are relatively well characterized in in dynamic short-lived contacts, these natural antibodies acting like a
the mouse3 but only recently described in humans.4 Natural antibodies lubricant rather than like cement. On the other hand, molecular aging,
are polyreactive or polyspecic, binding to structurally different self often presenting as aggregation, results in polymerization of the target,
and microbial targets as well,5,6 these targeted epitopes being mostly increasing the avidity of interactions with natural antibodies and
non-protein molecules. The afnity of natural antibodies to mono- aiding removal. This removal process is continuous and silent. Natural
valent glycan has been determined to be in the range of 10 4106 M.7 IgM xes C1q, which in turn is captured and taken up by phagocytes
While mostly, of the IgM class,8 functionally similar antibodies throughout the body, utilizing different C1q receptors.1316 Natural
belonging to the IgA and IgG classes are also found. Besides providing antibodies may also cover and seal leaks in the endothelium,
immediate protection against invading microbial agents, natural promoting regeneration and healing.17
antibodies are known to play key roles in the clearance of self In addition to B1I cells, MZ B cells also contribute to the fast
molecules, thereby contributing to homeostatic control, suppressing production of antibodies upon challenge.18 These two populations are
inammation and autoimmunity.9,10 In humans, B1I cells represent the cells responsible for thymus-independent (TI) responses. Is there,
from less than 1 to 9% of the circulating B cells.11 Calculating with an then, a difference between natural antibodies and TI antibody
average 4 109 white blood cells per liter of blood, 5% B cells in white responses? We believe there is not. Natural antibodies are dened as
blood cells, and an average 5% of B1I of all the B cells, we arrive at an being produced in the absence of a known antigenic stimulus. This is
averaged 5 107 B1 cells in 5 l of blood, capable of a dominant possibly a wrong interpretation of the events. Accepting that B-cell
contribution to the plasma antibody pool. The number of circulating development involves selection of low-afnity, polyspecic self-
B1I cells indeed shows correlation with serum IgM levels.12 reactivity, the antigenic stimulus is self. Natural IgM antibodies are
Figure 5 Denitions of the antibody network. (a) Antibody Abi reacts with epitope A and cross-reacts with epitope B. Abj reacts with epitope A and cross-
reacts with epitope B. Distances, expressed as negative logarithm of KD are averaged to get the distance in the epitope space, providing an edge in the
network. (b) Free antibody concentration divided by KD reects antigen-binding capacity. This value is displayed as the radius of the node. Changes in
antigen concentration will trigger changes in node radius. New antigens entering the system may react with higher afnity with a given antibody and thereby
change its position in the epitope space.
Figure 6 Simplied schematic representation of the antibody-epitome interaction network. Only nodes of the network are shown, edges are not drawn for the
sake of simplicity but are represented as internodal distances. Blue nodes are natural antibody interactions, purple nodes are interactions with the accepted
environment, red nodes are TD interactions in the epitope space. Green area represents B cells in the quiescent state, not producing antibodies but
presenting themselves in the epitome space. Immunological self is a superconnected central component, while epitopes marked for elimination are extruded
and disconnected.
With the denitions we have built up of the immunological self. The reactivity networks generated by
Bransburg-Zabary et al.37 by antigen microarray analysis are probably
1. B cells respond to [Ag] by developing into ASC the experimental observations closest to our model. In these experi-
2. B cells respond to [Ag] by changing KD ments, correlation analysis of antibody reactivity between individuals
3. [Ab] is regulated to reach KD for control of [Ag] revealed the presence of connected specicities. The deconvolution of
4. Network node size is [Ab]/KD polyclonal antibody reactivity within individuals will be the key to
5. i-j internodal distance is |KDi-KDj| generate the experimental data directly supporting our model.
6. In a network with N = number of B cells with distinct antigen Quantitative, descriptive approaches to antibody networks currently
binding properties nodes rely on genetic information obtained by new-generation sequencing
technologies.3840 These can now generate antibody sequence
we have outlined a dynamic, weighted network model of the repertoires with immense depth and ne resolution. While it is
humoral immune system, shown in a simplied two-dimensional possible to sequence the B-cell repertoire of a whole zebra sh,41 we
representation in Figure 6. Because TI responses utilize polyclonal cannot do the same with a human B-cell repertoire. There will be a
responses, antibody production against a given epitope region is
sampling bias depending on where the B cells are obtained from: bone
always shared, [Ab]/KD is always o1. For oligoclonal or monoclonal
marrow, blood, tonsils, lymph nodes, spleen. Additionally, sequencing
TD responses, this value may exceed 1. The model shows the space
cannot provide antibody concentrations, even less, so free antibody
lling property with multi-sized spheres, the core representing
concentrations or antigen concentrations in the sampled organism.
immunological selfa giant superconnected component of the
Immunomics approaches will need to utilize the databases on
network of antibodies.
antibodyantigen interactions that have already been compiled
CONCLUDING REMARKS (the immune epitope database, allergome database). The development
We have introduced homeostatic antibody model based on the of novel immunological methods is required for further formal proof
assumption that the clonal humoral immune system seeks after an of this theory and for its application in various elds of immunology.
equilibrium between antibody and antigen. This requires that the By combining genetic, immunologic and immunoinformatics data,
membrane-bound form of antibody, the BCR, regulates the fate of we expect that a true bioinformatics modeling of the complete human
cells that produce antibodies, as shown in our previous article. In this antibody network, its dynamics, disease-associated changes will be
article, we argued that both the afnity and the concentration of the possible in the near future.
antibodies produced in the host are tuned for silently degrading,
carefully removing or aggressively eliminating antigen. CONFLICT OF INTEREST
An important message of this approach is that molecular abundance The author declares no conict of interest.
is the dening factor for immunological self. Everything abundantly
present in blood and in the bone marrow during the development of ACKNOWLEDGEMENTS
the system is self immunologically. Immunological homeostasis is This paper is dedicated to all immunology theorists, past and present,
about controlled catabolism and regeneration of self. Self is silently who advanced the eld of immune network theories. I wish to thank Kroly
Liliom for consulting us on enzyme kinetics and Anita Orosz and Krisztin
eliminated when ages and aggregates. Molecules binding with high
Papp for working with me on methodological approaches to quantitative
afnity to cells screened for low-afnity self-binding and accompanied
immunomics. I thank my family and my colleagues for their patience and
by danger signals are regarded to be eliminated. In this sense,
understanding in the moments when I was too immersed in thinking. We
low-afnity self-recognition is necessary not to avoid binding to self received Grant K109683 from the National Research, Development and
but rather to set a point of reference in the epitope space. Innovation Ofce of Hungary (http://nkh.gov.hu/english) supported this
There have been several different approaches and theories to work.
provide general, perhaps, mathematical explanations for the complex-
ity, the functioning of the immune system. The idiotype network
theory29 originally worked out by Niels Jerne,31 and the clonal
selection theory elaborated by Frank MacFarlane Burnet32 identied 1 Foote J, Eisen HN. Kinetic and afnity limits on antibodies produced during immune
different, seemingly contradictory concepts, which were resolved by responses. Proc Natl Acad Sci USA 1995; 92: 12541256.
Coutinho33 who suggested that clonal as well as network organization 2 Prechl J. A generalized quantitative antibody homeostasis model: regulation of B-cell
development by BCR saturation and novel insights into bone marrow function. Clin
co-exists in the immune system. Our GQM is in agreement with this Transl Immunology 2017; 6: e130.
observation, as clonal responses generate a network, even though this 3 Savage HP, Baumgarth N. Characteristics of natural antibody-secreting cells. Ann N Y
Acad Sci 2015; 1362: 132142.
network is a virtual antibodyantigen network of connectivity rather 4 Grifn DO, Holodick NE, Rothstein TL. Human B1 cells in umbilical cord and adult
than a physical network. In Zvi Grossmans horizontal networks,34 the peripheral blood express the novel phenotype CD20+ CD27+ CD43+ CD70 . J Exp Med
relative probability of maturation increases with antigen dose and with 2011; 208: 6780.
5 Seigneurin JM, Guilbert B, Bourgeat MJ, Avrameas S. Polyspecic natural
afnity, and proliferation can be uncoupled from differentiation antibodies and autoantibodies secreted by human lymphocytes immortalized with
under certain predictable conditions, which are key assumptions in Epstein-Barr virus. Blood 1988; 71: 581585.
our model as well. Grignolio et al.35 recently compiled factors that 6 Avrameas S, Guilbert B, Mahana W, Matsiota P, Ternynck T. Recognition of self and
non-self constituents by polyspecic autoreceptors. Int Rev Immunol 1988; 3: 115.
potentially inuence immunological identity, focusing on the changing 7 Bovin NV. Natural antibodies to glycans. Biochemistry Mosc 2013; 78: 786797.
nature of immunological self, which they conceptualize as liquid self. 8 Kaveri SV, Silverman GJ, Bayry J. Natural IgM in immune equilibrium and harnessing
their therapeutic potential. J Immunol 2012; 188: 939945.
The concept of immunological homunculus by Irun Cohen et al.36 9 Grnwall C, Silverman GJ. Natural IgM: Benecial Autoantibodies for the Control of
states that non-reactivity can only create chaos, self-reactivity is Inammatory and Autoimmune Disease. J Clin Immunol 2014; 34(Suppl 1): S12S21.
required for control and order. Our model suggests that the need 10 Vas J, Grnwall C, Silverman GJ. Fundamental roles of the innate-like repertoire of
natural antibodies in immune homeostasis. Front Immunol 2013; 4: 4.
for self-reactivity is quite profane: controlled catabolism. As long as 11 Grifn DO, Rothstein TL. Human b1 cell frequency: isolation and analysis of human
microbial epitopes are also efciently cleared, they may constitute part b1 cells. Front Immunol 2012; 3: 122.
12 Kraljevic K, Wong S, Fulcher DA. Circulating phenotypic B-1 cells are decreased in 29 Kieber-Emmons T, Monzavi-Karbassi B, Pashov A, Saha S, Murali R, Kohler H. The
common variable immunodeciency and correlate with immunoglobulin M levels. promise of the anti-idiotype concept. Front Oncol 2012; 2: 196.
Clin Exp Immunol 2013; 171: 278282. 30 Hiepe F, Drner T, Hauser AE, Hoyer BF, Mei H, Radbruch A. Long-lived autoreactive
13 Guan E, Robinson SL, Goodman EB, Tenner AJ. Cell-surface protein identied on plasma cells drive persistent autoimmune inammation. Nat Rev Rheumatol 2011; 7:
phagocytic cells modulates the C1q-mediated enhancement of phagocytosis. J Immunol 170178.
1994; 152: 40054016. 31 Jerne NK. Towards a network theory of the immune system. Ann Immunol 1974; 125C:
14 Benoit ME, Clarke EV, Morgado P, Fraser DA, Tenner AJ. Complement protein C1q 373389.
directs macrophage polarization and limits inammasome activity during the uptake of 32 Burnet FM. A modication of Jernes theory of antibody production using the concept of
apoptotic cells. J Immunol 2012; 188: 56825693. clonal selection. CA Cancer J Clin 1976; 26: 119121.
15 Ramirez-Ortiz ZG, Pendergraft WF, Prasad A, Byrne MH, Iram T, Blanchette CJ et al. 33 Coutinho A. Beyond clonal selection and network. Immunol Rev 1989; 110: 6387.
The scavenger receptor SCARF1 mediates the clearance of apoptotic cells and prevents 34 Grossman Z. Recognition of self, balance of growth and competition: horizontal
autoimmunity. Nat Immunol 2013; 14: 917926. networks regulate immune responsiveness. Eur J Immunol 1982; 12: 747756.
16 Prabagar MG, Do Y, Ryu S, Park JY, Choi HJ, Choi WS et al. SIGN-R1, a C-type lectin, 35 Grignolio A, Mishto M, Faria AM, Garagnani P, Franceschi C, Tieri P. Towards a liquid
enhances apoptotic cell clearance through the complement deposition pathway by self: how time, geography, and life experiences reshape the biological identity. Front
interacting with C1q in the spleen. Cell Death Differ 2013; 20: 535545. Immunol 2014; 5: 153.
17 Prechl J, Czirjk L. The endothelial deprotection hypothesis for lupus pathogenesis: the 36 Cohen IR, Young DB. Autoimmunity, microbial immunity and the immunological
dual role of C1q as a mediator of clearance and regulator of endothelial permeability. homunculus. Immunol Today 1991; 12: 105110.
[version 2; referees: 2 approved, 1 approved with reservations]. F1000 Res 2015; 37 Bransburg-Zabary S, Kenett DY, Dar G, Madi A, Merbl Y, Quintana FJ et al. Individual
4: 24. and meta-immune networks. Phys Biol 2013; 10: 025003.
18 Martin F, Oliver AM, Kearney JF. Marginal zone and B1 B cells unite in the early 38 Larsen PA, Smith TP. Application of circular consensus sequencing and network
response against T-independent blood-borne particulate antigens. Immunity 2001; 14: analysis to characterize the bovine IgG repertoire. BMC Immunol 2012; 13: 52.
617629. 39 Benichou J, Ben-Hamo R, Louzoun Y, Efroni S. Rep-Seq: uncovering the immunological
19 Pereira P, Forni L, Larsson EL, Cooper M, Heusser C, Coutinho A. Autonomous repertoire through next-generation sequencing. Immunology 2012; 135: 183191.
activation of B and T cells in antigen-free mice. Eur J Immunol 1986; 16: 685688. 40 Briney BS, Willis JR, McKinney BA, Crowe JE. High-throughput antibody sequencing
20 Gunti S, Messer RJ, Xu C, Yan M, Coleman WG Jr, Peterson KE et al. Stimulation of reveals genetic evidence of global regulation of the nave and memory repertoires that
toll-like receptors profoundly inuences the titer of polyreactive antibodies in the extends across individuals. Genes Immun 2012; 13: 469473.
circulation. Sci Rep 2015; 5: 15066. 41 Ben-Hamo R, Efroni S. The whole-organism heavy chain B cell repertoire from Zebrash
21 Vos Q, Lees A, Wu ZQ, Snapper CM, Mond JJ. B-cell activation by T-cell-independent self-organizes into distinct network features. BMC Syst Biol 2011; 5: 27.
type 2 antigens as an integral part of the humoral immune response to pathogenic
microorganisms. Immunol Rev 2000; 176: 154170.
22 Vinuesa CG, Chang PP. Innate B cell helpers reveal novel types of antibody responses. This work is licensed under a Creative Commons
Nat Immunol 2013; 14: 119126.
23 Kunkel EJ, Butcher EC. Plasma-cell homing. Nat Rev Immunol 2003; 3: 822829. Attribution 4.0 International License. The images or
24 Poulsen TR, Jensen A, Haurum JS, Andersen PS. Limits for antibody afnity maturation other third party material in this article are included in the articles
and repertoire diversication in hypervaccinated humans. J Immunol 2011; 187:
42294235.
Creative Commons license, unless indicated otherwise in the credit
25 Lanzavecchia A, Frhwirth A, Perez L, Corti D. Antibody-guided vaccine design: line; if the material is not included under the Creative Commons
identication of protective epitopes. Curr Opin Immunol 2016; 41: 6267. license, users will need to obtain permission from the license holder to
26 Vidarsson G, Dekkers G, Rispens T. IgG subclasses and allotypes: from structure to
effector functions. Front Immunol 2014; 5: 520. reproduce the material. To view a copy of this license, visit http://
27 Leong KW, Ding JL. The unexplored roles of human serum IgA. DNA Cell Biol 2014; creativecommons.org/licenses/by/4.0/
33: 823829.
28 Stone KD, Prussin C, Metcalfe DD. IgE, mast cells, basophils, and eosinophils. J Allergy
Clin Immunol 2010; 125: S73S80. r The Author(s) 2017