Engineering Antibodies (1)
MSc Programme University of Nottingham
14th February 2005
by
Mike Clark, PhD
Department of Pathology
Division of Immunology
Cambridge University
UK
www.path.cam.ac.uk/~mrc7/
Antibody based immunotherapeutics
IgG is the preferred class
Schematic view of IgG domains
Antibody fragments can also be used
Antibodies can be derived from
immunised animals
The antibody immune response in-vivo can be
T-cell dependent or independent
Antibody fragments can also be selected from
in-vitro systems such as phage expression
Cycles of selection and mutation can give an
artificial in-vitro immune response based
simply on binding affinity
The Selection of IgG Fc Regions for
appropriate effector functions:
The role of isotypes and polymorphisms
Effector functions of human IgG
IgG1
IgG2
IgG3
IgG4
Complement activation
Classical pathway
Alternative pathway
+++
-
+
+
+++
-
-
Fc receptor recognition
Fc g RI
+++
-
+++
++
Fc g RIIa, 131R/R
Fc g RIIa, 131H/H
Fc g RIIb
++
+
++
+
-
++
++
++
+
+
+/ -
+
+/ -
Fc g RIII
Unlike mouse the human IgG subclasses
are very similar in sequence but they still
have different properties
The IgG receptor FcRn
Interaction with FcRn and with Protein A through similar region
FcRn is important for IgG
half-life and transport
Transferring
motifs between human subclasses
For FcgRI binding:
231
IgG1, IgG3
A P E L L G
IgG2
. . P V A IgG4
. . . F . .
sequence introduced into IgG1, IgG4
. . P V A or . . P V A G
For C1q binding:
318
IgG1, IgG3 E Y K C K V S N
IgG2
. . . . . . . .
IgG4
. . . . . . . .
sequence introduced into IgG1, IgG2
. . . . . . . .
238
G P
. .
. .
. .
. .
Db mutation
Dc mutation
327
331
K A L P A P
. G . . . .
. G . . S S
. G . . S S
Da mutation
Residues at key positions in mutated
constant regions
Antibody
G1
G1Da
G1Db
G1Dc
G1Dab
G1Dac
G2
G2Da
G4
G4Db
G4Dc
233
234
235
236
327
330
331
E
E
P
P
P
P
P
P
E
P
P
L
L
V
V
V
V
V
V
F
V
V
L
L
A
A
A
A
A
A
L
A
A
G
G
G
G
G
G
A
G
A
A
G
G
G
G
G
G
G
A
S
A
A
S
S
A
S
S
S
S
P
S
P
P
S
S
P
S
S
S
S
Summary of terminology
Mutant residues
Da
Residues 327, 330, 331 of IgG4
Db
Lower hinge of IgG2; omitting Gly236
Dc
Lower hinge of IgG2; including Gly236
Armour et al. Eur J Imm 1999; 29: 2613
Test systems: antibodies with CD52 and
a-RhD specificities
• Short, GPI-anchored glycoprotein
• Found on T cells and some B cells,
granulocytes and eosinophils
CD52
• About 45 x 104 molecules/cell
• Good target for CDC and ADCC
• Humanised variable domains of CAMPATH-1
antibody used
• Range of antibodies with same variable
domains already existed
Test systems: antibodies with CD52 and
a-RhD specificities
• Protein complex on erythrocyte membrane
• 1 - 3 x 104 molecules/cell
• Provides opportunities for use of agglutination
a-RhD
and rosetting assays
• Target for ADCC
• Used variable domains of Fog-1, a human IgG
isolated from hyperimmunised, RhD- blood
donor
Complement-mediated lysis of mononuclear cells
45
40
CAMPATH-1
antibodies
35
G1
% specific Cr release
G1D a
30
G1D b
G1D c
25
G1D ab
20
G1D ac
15
G2
10
G2D a
G4
5
G4D b
G4D c
0
-5
0.1
1
10
antibody, mg/ml
100
Complement-mediated lysis with
CAMPATH-1 G1(6.3 mg/ml),
inhibited by CAMPATH-1 G2Da
25
% specific Cr release
20
15
10
5
0
-5
1
10
100
G2Da concentration, mg/ml
1000
Binding to the FcgRI-bearing cell line, B2KA,
measured by fluorescence staining
G2
G1Da
CAMPATH-1H antibodies at
100 mg/ml
G4
G1Db
G1Dc
G1
Binding to the FcgRIa-bearing cell line,
B2KA, measured by fluorescent staining
CAMPATH-1
antibodies
160
140
G1
G1D a
G1D b
G1D c
G1D ab
G1D ac
G2
G2D a
G4
G4D b
G4D c
mean fluorescence
120
100
80
60
40
20
0
0.001
0.01
0.1
1
antibody, mg/ml
10
100
Chemiluminescent response of human
monocytes to sensitised RBC
Fog-1
antibodies
% chemiluminescence
140
120
G1
100
G1D a
G1D b
80
G1D c
60
G1D ab
40
G1D ac
20
G2
G2D a
0
G4
-20
G4D b
0
5000
10000
15000
20000
25000
30000
G4D c
antibody molecules/cell
Inhibition of chemiluminescent response to
clinical sera by Fog-1 G2Da
280
% chemiluminescence
240
200
G1
anti-D serum A
anti-D serum B
anti-D serum C
anti-D serum D
anti-D serum E
anti-C+D serum
anti-K serum
160
120
80
40
0
0
10
100
G2Da concentration, mg/ml
1000
Binding to the cell line 3T6 + FcgRIIa 131R,
measured by flow cytometry
100
Fog-1 antibodies
90
G1
50
G1Da
G1Db
G1Dc
G1Dab
G1Dac
G2
40
G2Da
G4
mean fluorescence
80
70
60
G4Db
G4Dc
G1Dg
IgA1,k
30
20
10
0.1
1
10
antibody concentration, mg/ml
100
Binding to the cell line 3T6 + FcgRIIa 131H,
measured by flow cytometry
90
Fog-1 antibodies
80
G1
G1Da
mean f luorescence
70
G1Db
G1Dc
60
G1Dab
50
G1Dac
G2
40
G2Da
G4
30
G4Db
20
G4Dc
G1Dg
10
0.1
1
10
antibody concentration, mg/ml
100
IgA1,k
Binding to the cell line 3T6 + FcgRIIb1*,
measured by flow cytometry
260
Fog-1 antibodies
G1
mean fluorescence
210
G1Da
G1Db
G1Dc
160
G1Dab
G1Dac
G2
110
G2Da
G4
60
G4Db
G4Dc
G1Dg
10
IgA1,k
0.1
1
10
antibody concentration, mg/ml
100
Binding to different forms of FcgRII
120
FcgRIIa 131R
FcgRIIa 131H
FcgRIIb1*
80
60
40
20
antibody constant region
G1Dg
G4Dc
G4Db
G4
G2Da
G2
G1Dac
G1Dab
G1Dc
G1Db
G1Da
0
G1
percentage of G1 binding
100
0% = binding of IgA1,k
Activity of Fog-1 antibodies in ADCC
120
100
% RBC lysis
80
G1
G1D ab
G2
G2D a
G4
G4D b
60
40
20
0
-20
0.1
1
10
100
1000
antibody concentration, ng/ml
10000
Inhibition by Fog-1 antibodies of ADCC
due to Fog-1 IgG1 (at 2ng/ml)
45
40
G2
35
G2Da
% RBC lysis
30
25
20
G1Da
15
10
5
0
0.0001
G1D c
G1Db, G1Dab, G1Dac,
{
G4, G4Db, G4Dc
0.001
0.01
0.1
1
10
inhibitor antibody concentration, ng/ml
100
1000
10000
Summary of antibody activities
Mutants of :
Binding to:
FcgRI
FcgRIIa R/R
FcgRIIa H/H
FcgRIIb1*
FcgRIIIb NA1
FcgRIIIb NA2
Monocyte actn
Complement lysis
ADCC
G1
wt
Da
Db
Dc
Dab Dac
G2
wt
Da
G4
wt
Db
Dc
Effect of mutations cannot always be
predicted from wildtype antibody activities
Complement lysis:
IgG2 activity is only ~3-fold lower than that of IgG1
but placing IgG2 residues in IgG1 (Db, Dc) eliminates lysis.
FcgRIIa 131H binding:
IgG1 and IgG2 show equal binding
but G1Db and G1Dc activities are 30-fold lower.
IgG1 binding may depend heavily on the mutated regions.
Other subclasses may have additional sites of interaction with
the effector molecules.
Db and Dc mutants
The 3 pairs of Db and Dc mutants show reduced activity in
all functions assayed but the residual levels of activity differ:
Db slightly more active in FcgRIIa 131H and 131R binding
Dc more active in FcgRI binding, monocyte activation
FcgRIIIb NA1and NA2 binding and ADCC
These mutants differ only by -/+ G236.
This must affect the ability of the FcR to accommodate the
IgG2 lower hinge.
What of the immunogenicity of
therapeutic antibodies?
Bad News
• Universal tolerance to all self-antigens does not
exist.
• Auto and allo-immunity are common observations
• Human proteins can be immunogenic in humans.
(e.g. recombinant insulin, EPO and Factor VIII)
• Human antibodies can be immunogenic in humans
(anti-idiotype and anti-allotype) and this applies to
chimeric, humanised and fully human antibodies.
Good News
• Auto and allo-immunity are common observations
but these immune reponses can be modified and
regulated.
• Human antibodies can be immunogenic in humans
but this immunogenicity varies from antibody to
antibody for complex reasons, and is probably
more dependent on the mode of action, and not
just the way they were made (i.e. chimeric,
humanised or fully human).
Antigenicity and Immunogenicity
• Antigenicity is simply an ability of a molecule to be
recognised by a pre-existing T-cell receptor (TCR)
or a B-cell receptor (antibody)
• But once an antigen is recognised by a receptor it
can either be immunogenic or tolerogenic.
• The same antigen can sometimes induce tolerance
and sometimes provoke an immune response
depending upon factors such as mode of
administration and uptake by and co-stimulation of
antigen presenting cells (APCs).
Immunogenicity
• Immunogenicity of T-cell dependent antigens relies
on presentation by professional APCs (e.g. Dendritic
cells).
• Dendritic cells (and other APCs) acquire antigen
through use of innate receptors including
complement receptors and Fc receptors, thus
allowing recognition and uptake of immune
complexes.
Induction of tolerance to therapeutic antibodies
Benjamin,R.J., Cobbold,S.P., Clark,M.R., & Waldmann,H. (1986) J. Exp. Med.
163, 1539-1552. Tolerance to rat monoclonal antibodies: implications for
serotherapy
Observation
• Relatively easy to tolerise mice, with de-aggregated human immunoglobulin or
with rat immunoglobulin, despite large differences in the constant region
sequences between mouse, human and rat.
• However in mice which are tolerant of soluble rat IgG2b, administration of
antibodies which bind to mouse cell surface antigens provokes a strong antiidiotype response.
Explanation
• Is this a function of the inherent immunogenicity of immune complexes?
•Aggregated antibody is more likely to activate complement and to bind to low
affinity Fc receptors.
Antibody selection and design
The choice of antibody constant region is
largely dictated by functional requirements of
the antibody.
But what about the V-regions ?
The V-region Mythology
Chimaeric
65% Human ?
Humanised
95% Human ?
• This commercial marketing mythology is based on an assumption
that mouse and human antibody sequences are unique.
• However a study of the Kabat database shows that there is high
sequence homology for antibodies from different species.
Kabat database variability of VH sequences
Human VH
Mouse VH
Are chimaeric, humanised and fully human antibodies so very
different in sequence?
Possible to select alternative V genes for humanisation
Gorman,S.D., Clark,M.R., Routledge,E.G., Cobbold,S.P. & Waldmann,H.
P.N.A.S. 88, 4181-4185 (1991) Reshaping a therapeutic CD4 antibody.
Routledge,E., Gorman,S., & Clark,M. in Protein engineering of antibody
molecules for prophylactic and therapeutic applications in man. (Ed. Clark,M. )
Pub. Academic Titles, UK (1993) pp. 13-44 Reshaping of antibodies for therapy.
• Gorman et al recognised that homology also extended through the CDR regions
not just the framework regions
• Homology to Kol was increased from 69% to 89% by the humanisation process.
The same strategy can be applied to almost any V-region
Sequence homologies of some rodent, humanised and human sequences
Antibody comparisons
FR
CDR
Whole V
Campath-1G
(68/87) 78%
(14/34) 41%
(82/121) 68%
Anti-Tac
(67/87) 77%
(14/29) 48%
(81/116) 70%
OKT3
(67/87) 77%
(12/32) 38%
(81/119) 68%
Campath-1H versus germline
(78/87) 90%
(8/34) 24%
(86/121) 71%
Anti-Tac versus germline
(77/87) 89%
(14/29) 48%
(91/116) 78%
OKT3 versus germline
(76/87) 87%
(6/32) 19%
(82/119) 69%
(77/87) 89%
(23/37) 62%
(100/124) 81%
murine versus human germline
humanised versus human germline
human versus human germline
Fog-1 RhD versus germline
Homologies for antibody heavy chain V regions
compared with human germline sequences
Sorted by homology
Antibody
Specificity
V-region
Homologous
VH
JH
Length
Matches
Homology
FOG-1
RhD
Human
V4-34
JH6A
124
100
0.807
anti-Tac
CD25
Humanised
HV1F10T
JH6a
116
91
0.784
anti-Tac
CD25
Mouse
HV1F10T
JH4D
116
84
0.724
TNF-alpha
Mouse
VI-4-IB
JH3B
117
84
0.718
Campath-1H
CD52
Humanised
DP-71_3D197D-
JH4D
121
86
0.710
Campath-1G
CD52
Rat
DP-34_DA-10
JH4D
121
85
0.702
OKT3
CD3
Humanised
b25
JH6a
119
82
0.689
OKT3
CD3
Mouse
DP-7_21-2-..
JH6a
119
81
0.681
HD37
CD19
Mouse
6M27
JH4D
124
83
0.669
anti-CD20
CD20
Mouse
DP-7_21-2-
JH2
121
81
0.669
anti-TNFa
What of the Emperor’s new clothes?
Appropriate selection of sequences of antibody Constant and
Variable regions is likely to be only one factor controlling the
immunogenicity of therapeutic antibodies.
However it is the final sequence of the antibodies which matters
and not the route by which they were made. For example it is
possible to come up with alternative humanised sequences for
the same antibody. Similar sequences can often be found for
mouse, rat and human variable regions within the databases.
Even fully human antibodies may contain unusual motifs or
structures as a result of the somatic recombination and junctional
diversity combined with somatic hypermutation.
What determines immunogenicity?
• Classical Self vs non-Self (Peter Medawar)
-
Aquired neonatal tolerance to antigens.
• Danger Hypothesis (Polly Matzinger)
-
Cell killing (inappropriate, non-apoptotic)
-
Inflammation (cytokine release)
• Pattern recognition (Charles Janeway)
-
Innate receptors for infectious pathogens
-
Complement activation and fixation of C3 (Fearon)
-
( Fc receptors for immune complexes)
The effect of aglycosylation on the
immunogenicity of a humanised therapeutic
CD3 monoclonal antibody Routledge et al
1995 Transplantation 60, 847-853
Normal Mouse
CD3 Transgenic
(no antigen)
(cell surface)
Human IgG1
-
+++
Aglycosyl IgG1
-
+
The effect of aglycosylation on the
immunogenicity of a humanised therapeutic
CD3 monoclonal antibody Routledge et al
1995 Transplantation 60, 847-853
The human IgG1in the CD3 transgenic mice was able to kill
target cells, to activate complement, to bind to FcR and to cause
cytokine release. Whereas the aglycosylated antibody was poor in
these functions and produced only a weak immune response.
Is this a special case or can it be generalised to other antibodies?
Is it consistent with the Matzinger “Danger Hypothesis” as
applied to therapeutic administration of recombinant antibodies?
Elimination of the immunogenicity of therapeutic
antibodies. Gilliland et al 1999 J.Immunol 162,
3663-3671
• Took CAMPATH antibody and mutated a
key residue in the CDR region so as to
prevent cell binding to CD52.
• This variant could be used to tolerise
CD52 transgenic mice so that they no
longer mounted an immune response to
the wild type CAMPATH
Factors likely to influence immunogenicity of therapeutic antibodies
Murine constant regions
V-region sequences
Human Ig allotypes
Unusual glycosylation
Method of administration
Frequency of administration
Dosage of antibody
Patients' disease status
Patients' immune status
Patients' MHC haplotype
Specificity of antibody
Cell surface or soluble antigen?
Formation of immune complexes with
antigen
Complement activation by antibody
Fc receptor binding by antibody
Inflammation and cytokine release
Will the idiotype always be immunogenic?
The idiotype will obviously always be unique and thus antigenic.
However it may be possible through mode of use to influence
whether this antigenic idiotype is immunogenic or tolerogenic!
Take home message to remember
In immunological terms antigenicity is certainly not the same as
immunogenicity!