Immunocomputing

1
Immunocomputing

• The natural immune system and its computational
  metaphors
• by
• Ian Nunn, inunn_at_digitaldoor.net, 2002

2
Organization of This Lecture

• Overview of the Immune System (IS)
• The innate IS
• The adaptive IS
• The immune response
• Antibodies and the Clonal Selection principle
• Immune network theory
• The key computational aspects
• The symbol ? means significant explanatory text
  or diagram follows later

3
Views of the Immune System

• A collection of lymphoid organs, cells
  principally leukocytes ( lymphocytes ? and
  phagocytes ?), and molecules that are
  interrelated in function
• A related collection of bodily defenses
• Physical barriers (skin, mucous membranes)
• Physiology (temperature, ph, enzymes in
  secretions)
• Innate IS ?(phagocytes) cellular level
• Adaptive IS ?(lymphocytes) molecular level
• The innate IS and the adaptive IS are interactive
• Its all molecular chemistry proteins and
  peptides

4
Anatomy of the Immune System1

• Primary lymphoid organs (black/red)
• Secondary lymphoid organs (blue/yellow)

5
The Antagonists

• Infectious foreign agents called pathogens
• Viruses (cold, influenza, smallpox)
• Bacteria (anthrax, E. coli)
• Multi-cellular parasites (malaria)
• Fungi
• Foreign proteins and toxins
• Pathogens express cell surface and soluble
  proteins called antigens (Ag)
• Agent identification problem how to detect and
  remove pathogens or harmful non-self elements
  without attacking beneficial self elements
  (autoimmune reaction)

6
Immune Defense1

• Response of various subsystems to pathogens

7
The Innate Immune System

• System available at birth, non-adaptive in
  makeup, providing an immediate response to
  invasion
• Principal components are
• Complement ? system, a class of 25 blood
  proteins
• Phagocytes that are scavenger cells including
  macrophages ? ingest foreign material and assist
  the adaptive immune response
• Natural Killer (NK) ? Cells, a type of lymphocyte

8
The Complement System

• Proteins that bind to the surface of certain
  types of bacteria
• Promotes two mechanisms ? of elimination after
  binding
• lysis the complement ruptures the cell membrane
• opsonization the bound complement marks the
  pathogen for destruction by macrophages
• Self cells have surface regulatory proteins that
  prevents complement binding

9
Macrophages

• Scavenger role
• Have receptors for
• Certain types of bacteria directly
• Complement on opsonized bacteria
• Activated partly by Th1 cell ? lymphokine ?IL-2
• Known as antigen presenting cells ?(APC)
• ingest and then digest pathogens and antigens
• present the Ag peptides at their surface to T
  cells ? via class II major histocompatibility
  complex (MHC) molecules that are contained only
  in IS cells

10
Macrophage Bacterial Ingestion1
Step 1 an opsonized antigen is ingested
Step 3 MHC/peptide complex presented on surface
Step 2 antigen peptides are bound by class II
MHC molecules
11
Macrophages (cont.)

• Secrete cytokines (IL-1) ? after activation
• Cytokines are a class of signalling molecules
  that
• Induce inflammatory response, physiological
  changes that facilitate the activity of IS cells
• Elevated temperature
• Increased blood flow and blood vessel
  permeability
• Trigger liver to produce acute phase protein
  (ATP) a complement molecule which binds to
  bacteria activating a macrophage response

12
Natural Killer Cells

• Bind to carbohydrates on surface of self cells
• Cant recognize specific antigens unlike Tk cells
  ? or killer cells of the adaptive immune system
• Healthy self cells express an inhibitory signal
• Virus-infected cells may lose inhibitory ability
  thus activating NK cells
• Activated NK cell injects chemicals that trigger
  apoptosis ( programmed cell death) or lysis

13
The Adaptive Immune System

• Characterized by a two-phase, primary and
  secondary response to pathogens ?
• Principal components are 1012 short-lived (4
  7 days) lymphocytes created in bone marrow at the
  rate of 107 per day
• T-cells ? which mature in the thymus gland
• B-cells ?(majority) which mature in bone marrow

14
Lymphocyte Growth

• Pleuripotent or common haemopoietic stem cells in
  bone marrow at birth
• Differentiate into progenitor cells including
• Myeloid type ? phagocytes
• Lymphoid type ? B and T lymphocytes
• These grow into immature precursor cells in bone
  marrow
• Precursor cells mature in primary lymphoid organs
• Mature cells
• Activate and differentiate in bodily tissue
• Some (B cells ?) multiply in secondary lymphoid
  organs

15
Lymphocyte Maturation
16
Cell Structure

• Each antigen exhibits many unique structural
  regions called epitopes (1016 possible
  varieties)
• Lymphocytes have (105) identical secretable
  surface protein receptors called antibodies (Ab)
  ?
• At any time the immune system has a set of 108
  different Ab types called its repertoire
• An antibody exhibits a unique structural region
  or binding site called a paratope expressing a
  range of affinities ? for binding a specific set
  of epitopes

17
Ag Epitopes and Ab Paratopes1
Antigen (Ag) showing epitopes and B cell
lymphocytes showing antibody (Ab) paratopes
(receptors)
18
Protein Structure (Folding)

• Courtesy www.stanford.edu/group/pandegroup/Cosm/

19
Shape-Space Representation

• Factors affecting Ab/Ag binding include
• Molecular shape of paratopes and epitopes
• Charge distribution
• Relationship of corresponding chemical groups
• Not covalent (chemical) bonding
• A binding site parameterizes an L-dimensional
  shape-space ?
• A paratope is at the center of a volume Ve of
  complementary epitopes with which it can bind
  called its recognition region
• e is called the affinity threshold ?

20
Shape-Space1

• Paratopes (), epitope complements (x) and
  affinity thresholds (e) in shape-space (V)

21
Spatial Distance

• Measures degree of interaction (affinity)
• For Ag(ag1, ag2 ,,agL) and Ab(ab1, ab2 ,,abL)
  expressed as vectors in shape-space

Euclidean
Manhattan
Hamming
22
Affinity Threshold (e)

• Distance or match score is inversely proportional
  to complementarity or affinity for binding

• A binding function measures affinity or strength
  of binding
• Its domain is the set of possible distances
• Its range is the set of binding values

• The affinity threshold e is that value of
  distance above which which binding actually
  occurs

23
Affinity Binding Functions1

1. Threshold (step) binding function

1. Sigmoid binding function

24
Activation Threshold

• A lymphocyte may bind multiple antigens
  (epitopes) of the same type
• A lymphocyte may also bind multiple antigens of
  related type
• A lymphocyte cant become activated before the
  number of receptors bound exceeds an activation
  threshold
• Different cell types have different activation
  thresholds
• Different cell types behave differently on
  activation

25
Lymphocyte Binding Different, Structurally
Related Antigens3
26
The Adaptive Immune Response

• Immune response (IR) in the adaptive immune
  system has two phases ?
• Primary response to antigen A (some Ab present)
• Initial lag phase
• Ab concentration then increases , levels off and
  falls
• Secondary response to antigen A
• Short lag phase
• faster buildup to greater maximum level with
  slower drop off
• Response to a new unknown but related antigen B
  after primary response to A
• Similar to secondary response to A but less
  pronounced. Called immunological cross-reaction

27
Primary and Secondary IR1
28
The Adaptive Immune Response (cont.)

• Demonstrates adaptation, reinforcement learning
  and associative memory needed for immunization -
  called a generalization capability
• Characteristics of an associative memory ? are
• Robust both to noise (binding occurs over a range
  of antigen types) and component failure
  (destruction of individual lymphocytes)
• Stored data recovered by reading same or similar
  data (IR)
• Restricted by Th cell dependency ?

29
The T Cell

• A lymphocyte that along with B cells ? are the
  major elements of the adaptive immune system
• Three major subclasses
• Helper (CD4, T4 or Th ) T cells ? assist a range
  of leukocytes in antigen identification
• Cytotoxic (killer or Tk) T cells ? destroy
  pathogens by lysis
• Suppressor (CD8) T cells express a negative
  effect on immune cell generation preventing
  autoimmune reaction
• Population diversity created in thymus by
  combinatorial rearrangement of genes but no
  somatic mutation ?

30
Th Cell Functioning

• Binds (recognizes) only linear conformations of
  epitopes on unfolded (digested) antigen
  peptide/MHC complex on a macrophage
• Secrete IL-2 lymphokines on activation and
  express IL-2 receptors
• IL-2 promotes cellular growth, activation and
  regulation - particularly of self, B cells and
  macrophages
• Th cells differentiate into
• Th1 cells that activate Tk cells and macrophages
  inducing an inflammatory response
• Th2 cells that activate B cells

31
Th Cells and Self-Tolerance

• Most self epitopes occur in the thymus and bone
  marrow
• An immature Th cell activated by binding a self
  epitope suffers apoptosis (clonal deletion or
  negative selection)
• Process called central tolerance
• Some may still be auto(self)reactive. A second
  mechanism, costimulation is required
• Signal I occurs when activation threshold
  exceeded
• Signal II IL-1cytokines provided by innate IS
• Signal I without II triggers apoptosis, a
  negative selection or down-regulatory signal

32
T Cell Tolerization3
33
Tk Cell Activation3
34
T Cell Activation2
35
The B Cell

• Two major subclasses
• Plasma cells ? that produce and secrete
  antibodies (no lymphokines), a defense reaction
• Memory cells ? that express the associative
  memory characteristic
• Antigen processing Plasma cells digest bound
  antigens and present Ag peptides at their surface
  via class II MHC molecules

36
B Cells and Self Tolerance

• Initial tolerization occurs in bone marrow
• Affinity maturation ? may produce autoreactive
  clones through somatic hypermutation ?
• Distributed tolerance (occurs in lymph nodes
  throughout body) by costimulation
• Signal I occurs when activation threshold
  exceeded
• Signal II provided by Th2 cell IL-2 lymphokines
  during antigen processing
• Signal I without II triggers apoptosis

37
B Cell Antigen Processing

• B cells digest bound antigens and present Ag
  peptides at their surface via class II MHC
  molecules
• A Th2 cell binds to the peptide/MHC complex and
  returns a signal II IL-2 lymphokine contributing
  to the B cells activation
• Activated B cells travel to the secondary
  lymphoid organs as part of the affinity
  maturation process

38
B Cell Affinity Maturation

• Affinity maturation is a cyclical process
  involving plasma B cells
• Selection activation by Th2 cell lymphokines and
  threshold regulated antigen binding
• Proliferation clonal division in lymph nodes
  expressing somatic hypermutation ?and receptor
  editing ?
• Differentiation after leaving lymph node, into
  plasma and memory cells

39
The Affinity Maturation Principle1
40
B Cell Adapted Population Diversity3
41
The Memory Problem

• B cells live only a few days (10 max)
• How is memory effected? Theories
• A long-lived variety of B cell
• Restimulation by long lived (years) traces of
  antigens in the body - a kind of low level
  chronic infection
• Both

42
Intracellular Pathogenesis

• Intracellular pathogens (viruses, some bacteria)
  are invisible to B cells
• Viral antigen is captured by a macrophage,
  presented to a Th cell which releases IL-2
  lymphokines
• Non-IS cells contain class I MHC molecules that
  transport internal viral peptides to the cell
  surface
• Class I MHC/peptide complexes are bound by killer
  T cells which are activated in part by
  costimulation by IL-2
• Tk cells kill infected cells by exercising an
  effector function (lysis, apoptosis induction,
  toxic chemical injection)

43
The Adaptive Immune Systems Response to
Infection1

1. Macrophage ingests Ag, presents MHC/peptide at
   surface, releases IL-1

1. T cell binds MHC/peptide and IL-1, activates

1. Activated T cell develops, releases IL-2

1. B cell binds antigen and IL-2, activates

1. Activated B cell clones, differentiates into
   plasma and memory cells

1. Plasma cell releases antibodies which bind
   antigens

44
The Antibody Molecule

• A soluble form of leukocyte receptor also called
  an immunoglobulin (Ig)
• Two identical light (L) and heavy (H) chains ?
• A constant region ? responsible for IS cell
  binding and available in a few varieties called
  isotypes that determine effector selection
  there are 5 Ig classes
• A variable region ? responsible for Ag binding
• Variable region is a concatenation of three
  genes, V (Variability), D (Diversity) and J
  (Joining) each from a separate gene library ?
• Binding is Ab paratope to Ag epitope

45
The Antibody Molecule1

1. Heavy (H) and light (L) chains variable (V) and
   constant (C) regions of the antibody molecule

1. The V, D and J gene libraries from which the
   antibody DNA is assembled

46
Combinatorial And Junctional Diversity

• Occurs in the bone marrow when lymphocytes are
  first created
• Expressed by random combinatorial joining of a D
  and J gene followed by a V gene in the VH chain
• Junction misalignment if amino acids dont line
  up causing some to be dropped (Frame shift). Many
  are non-translatable or unproductive and are
  dropped
• Productive recombinations result in cells which
  repeat with V and J genes of VL chain
• Estimated combinatorial diversity from both
  chains 5x107

47
Somatic Hypermutation

• Expressed in B cell (somatic) clonal reproduction
• Mutation rate 109 times normal (hyper)
• Types of mutation in Ab V region (receptor)
• Point mutations
• Short deletions
• Insertion of random gene sequence (receptor
  editing 25)
• Results
• Most non-functional or low affinity receptors
  eliminated by apoptosis (mechanism not
  understood) and negative selection
• Some are autoreactive and eliminated by negative
  selection
• A very few may have increased affinity due to
  conformational change and positive selection
• Total coverage of antigen repertoire thought to
  be complete

48
The Clonal Selection Principle

• Governs generation of new lymphocytes by plasma
  cells
• Clonal copies of parents under somatic
  hypermutation
• Elimination of autoreactive unproductive clones
• Proliferation and differentiation resulting from
  antigen activation of B cells
• Autoimmune disease the result of autoreactive
  clones resistant to early elimination by
  self-antigens
• The total number of lymphocytes kept relatively
  constant over time by regulation
• Responsible for maintaining Ab repertoire
  diversity recall repertoire is 108, 107
  replaced daily so complete replacement in 10 days

49
Combinatorial Diversity and Shape Space Coverage

• Generational diversity

50
Other Mutational Effects

• Clones may express different isotypes by
  recombination in the constant region of the
  antibody called isotype switching

51
Immune Network Theory

• A theory to explain the self-regulation and
  memory properties of the IS
• An Ab may display epitopes called idiotopes to
  distinguish them from Ag epitopes
• An Abs set of idiotopes called its idiotype
• An idiotope may be recognized by a set of
  antibody paratopes
• Direction of recognition results in activation or
  suppression in the network

52
The Immune Network1

1. Cascading recognition antibody view
2. Cascading recognition sets of idiotopes and
   paratopes

53
A Complex Adaptive System

• The IS is a complex adaptive system
• Large populations of several classes and
  sub-classes of agents with specific unique
  behaviors
• No central control
• Able to regenerate its (self) elements
• Overall population is self-regulating
• Able to adapt to any external influence through
  an ability to generate diversity
• A rich example of a multi-agent or swarm system

54
Features Of Computational Interest

• Pattern recognition through affinity binding
• Self-recognition by tolerization
• Intrusion detection by lymphocyte / antigen
  receptor binding
• Feature extraction through Macrophage and B cell
  antigen digestion and MHC/peptide (feature)
  presentation
• Reinforcement learning
• exhibited by the adaptive immune response (AIR)
  in response to repeated infection
• Clonal selection selects for cells with higher
  affinity
• Memory
• AIR exhibits associative memory in the
  cross-reactive response.
• A result of the affinity maturation principle by
  which high affinity long-lived memory cells are
  produced

55
Features Of Computational Interest (cont.)

• Self-Regulation through AIR and lymphocyte
  population regulation
• Generation of diversity
• Part of affinity maturation process (clonal
  production)
• One mechanism Combinatorial gene recombination
  and junction editing in primary lymphoid organs
• Another mechanism Somatic hypermutation in
  secondary lymphoid organs
• Point mutations (explore local optima)
• Short deletions
• Random insertions or receptor editing (escape
  from local optima)
• Adaptation IS can respond to novel conditions
  through AIR and ability to generate diversity

56
Features Of Computational Interest (cont.)

• Optimization the immune response and clonal
  selection result in increasingly better response
  to stimulus over time through higher affinity
  clones
• Distributed and Collaborative processing the
  various agents in the IS interact cooperatively
  with each performing a unique set of functions to
  achieve
• Distributed detection
• Distributed defense
• Robust and scalable
• Noise tolerant due to affinity binding
  (recognition region)
• Component failure tolerant affinity maturation
  ensures a large population of agents that carry
  desirable characteristics
• Garbage collection activity of phagocytes

57
Other Concepts

• Formalism allow mathematical treatment
• Shape space and identification parameters
• Binding functions and affinity threshold
• Stochastic nature of affinity binding
• Can model with differential equations
• Neat concepts
• Negative selection
• Costimulation and chemical signaling
• Helper and suppressor functions
• Threshold mechanisms (affinity and activation)
• Cross-reactivity

58
References

1. de Castro, L.N. and Von Zuben, F.J. Artificial
   immune Systems Part I Basic Theory and
   Applications. Technical Report TR DCA 01/99,
   December, 1999.
2. Eales, L.J. Immunology For Life Scientists A
   Basic Introduction. John Wiley Sons,
   Chichester, 1997.
3. Hofmeyr, S.A.Introduction to the Immune System.
   In Design Principles for the Immune System and
   Other Distributed Autonomous Systems. Oxford
   University Press, New York, 2001.