Site-specific conjugation of cytotoxic drugs to antibodies ...1 Supplementary Materials for:...
Transcript of Site-specific conjugation of cytotoxic drugs to antibodies ...1 Supplementary Materials for:...
1
Supplementary Materials for: Site-specific conjugation of cytotoxic drugs to antibodies
substantially improves the therapeutic window
Jagath R Junutula, Helga Raab, Suzanna Clark, Sunil Bhakta, Douglas D Leipold, Sylvia Weir, Yvonne Chen, Michelle Simpson, Siao Ping Tsai, Mark S Dennis, Yanmei Lu, Y. Gloria Meng, Carl Ng, Jihong Yang, Chien C Lee, Eileen Duenas, Jeffrey Gorrell, Viswanatham Katta, Amy Kim, Kevin McDorman, Kelly Flagella, Rayna Venook, Sarajane Ross, Susan D Spencer, Wai Lee Wong, Henry B Lowman, Richard Vandlen, Mark X Sliwkowski, Richard H Scheller, Paul Polakis and William Mallet
Genentech Inc., 1 DNA Way, South San Francisco, CA 94080, USA.
Correspondence should be addressed to W.M. ([email protected]) or J.R.J.
2
Contents
Supplementary item and number
Title or caption Page Number
Supplementary Figure 1
Engineered Cys residues in the THIOMABs are blocked
with cysteinylation or glutathionylation during
fermentation process.
4
Supplementary Figure 2
Site-specific conjugation of biotin-maleimide to
engineered THIOMABs.
5
Supplementary Figure 3
Time course for re-oxidation of reduced antibody. 6
Supplementary Figure 4
Papain digest of LC-V110C THIOMAB shows evidence
for dimerization through inter-chain S-S bond between the
Fabs.
7
Supplementary Figure 5
Screening light chain variants to identify reactive thiol
groups in the Fab for site-specific conjugation.
8
Supplementary Figure 6
Anti-MUC16 3A5 humanization. 9
Supplementary Figure 7
Anti-MUC16 TDC binds only to MUC16-expressing cell
lines.
11
Supplementary Figure 8
Analytical characterization of TDC and ADC.
12
Supplementary Figure 9
Mapping the antibody region of cytotoxic drug attachment
to the TDC-Fab.
15
Supplementary Figure 10
Peptide-mapping – Identification of peptide(s) containing cytotoxic drug in the TDC-Fab.
16
Supplementary Figure 11
TDC and ADC have comparable in vitro activities. 17
Supplementary Figure 12
Anti-MUC16 TDC is active against multiple mouse
xenograft tumor models.
18
3
Supplementary Figure 13
The anti-MUC16 TDC produced using an optimized
conjugation process with 2 drugs/antibody is efficacious in
xenograft tumor models.
20
Supplementary Figure 14
Chimeric anti-MUC16 TDC is better tolerated in rats than
ADC.
21
Supplementary Figure 15
Anti-MUC16 ADC and TDC have similar pharmacokinetic
properties in tumor-bearing mice.
22
Supplementary Figure 16
Binding of human anti-MUC16 antibodies to FcRn. 23
Supplementary Figure 17
Dissociation of bound anti-MUC16 antibodies from FcRn.
24
Supplementary Table 1
Identification of additional sites that allow conjugation
specifically to LC or HC-Fab region of the antibody.
25
Supplementary Table 2
Comparison of humanized anti-MUC16 variants and their
binding to MUC16 extracellular domain (ECD).
26
Supplementary Table 3
MC-vc-PAB-MMAE is specifically conjugated to the HC-
Fab of the THIOMAB.
27
Supplementary Table 4
Peptide mapping – Identification of MC-vc-PAB-MMAE
conjugated peptide in the TDC-Fab.
28
Supplementary Table 5
Relative binding affinities of human anti-MUC16
antibodies to FcRn measured by ELISA.
29
Supplementary Methods
Cell lines and proteins; Anti-MUC16 antibody
humanization; Peptide mapping to identify MC-vc-PAB-
MMAE labeled peptides in anti-MUC16 TDC; Additional
efficacy studies, Competition for anti-MUC16 binding by
ECD proteins from different species; Surface plasmon
resonance analysis using BiacoreTM; FcRn binding
measurements by ELISA.
30
4
Supplementary Figure 1. Engineered Cys residues in the THIOMABs are blocked with
cysteinylation or glutathionylation during fermentation process.
SFig. 1a. Deconvoluted mass spectrum of 3A5 antibody. Theoretical masses for the
three major glycoforms of the antibody are: 147407, 147568 and 147731 daltons. The
observed masses were 147410, 147569 and 147734 daltons, in excellent agreement with
the expected masses. SFig. 1b. Deconvoluted mass spectrum of Thio-3A5 antibody.
Theoretical masses are 147469, 147631 and 147793 daltons. for the three major
glycoforms. Observed masses (heterogeneous and poorly resolved) are approximately
147774, 147904 and 148061 daltons. The heterogeneity is due to the presence of a
mixture of capping groups (usually cysteine or glutathionine) on the two newly
introduced cysteines. These sulfhydryl blocking groups are a result of the fermentation
process. These groups can be removed by reduction of the antibody under controlled
conditions. The antibody interchain disulfides are re-established by controlled oxidation
(SFig. 3); this process results in an antibody mass spectrum comparable to SFig. 1a.
5
Supplementary Figure 2. Site-specific conjugation of biotin-maleimide to engineered
THIOMABs.
Engineered cysteine residues were unblocked using the reduction/oxidation method
described in the Fig. 1. (SFig. 2a) SDS-PAGE analysis of the MAb upon reduction and
oxidation. Lane 1, intact THIOMAB; lane 2, THIOMAB after TCEP reduction; lane 3,
THIOMAB after TCEP reduction followed by size exclusion chromatography; lane 4,
THIOMAB after CuSO4 oxidation step. (SFig. 2b) Biotinylated THIOMABs were
analysed on reducing SDS-PAGE, and the conjugated biotin was detected on western blot
using streptavidin-horseradish peroxidase. Total antibody was detected with anti-IgG
horseradish peroxidase. (SFig. 2c) MAbs were deglycosylated and analysed on LC/MS
to quantitate the total number of biotin molecules per MAb. Lane numbers in SFig. 2b
correspond to the samples in SFig. 2c.
6
Supplementary Figure 3. Time course for re-oxidation of reduced antibody.
Trastuzumab was reduced and purified on a cation exchange column at pH 5.5. SFig. 3a
shows the reversed phase profile for the reduced antibody. The three peaks around 7
minutes (from left to right) are: light chain (LC), heavy chain (HC) and a small amount
of HC+LC that was not completely reduced. The purified, reduced antibody was allowed
to re-oxidize at pH 7, and aliquots were taken as a function of time. The SFig. 3b shows
that the majority of light and heavy chains have re-oxidized to form an intact antibody (~
10 min peak) containing 2 LC and 2 HC within 90 min. No additional formation of
intact antibody was seen at the 180 min reaction time point (SFig. 3c).
7
Supplementary Figure 4. Papain digest of LC-V110C THIOMAB shows evidence for
dimerization through inter-chain S-S bond between the Fabs.
Trastuzumab and its LC-V110C THIOMAB were digested with papain at a 250:1 ratio
(weight/weight) for 2 hrs at 37 °C to generate Fab and Fc fragments. The samples were
analyzed by LC/MS. SFig.4a and SFig.4b show the Fab portions of the THIOMAB and
Trastuzumab, respectively. Digestion of the THIOMAB sample resulted in a mixture of
monomeric Fab and Fab dimer (due to the disulfide bridge between two Fabs through the
engineered cysteines in the LC). Digestion of the Trastuzumab wild type (non-cysteine
engineered) antibody yielded the expected Fab monomer only.
8
Supplementary Figure 5. Screening light chain variants to identify reactive thiol groups
in the Fab for site-specific conjugation.
SFig. 5a. Phage ELISA assays were conducted to test the binding of biotinylated
hu4D5Fab-phage variants to HER2-ECD (filled orange bars) and streptavidin (filled
black bars). SFig. 5b. Derived thiol reactivity values (Streptavidin OD450/Her2 binding
OD450) were plotted against each Cys variant. HER2-ECD and streptavidin were
immobilized (200 ng) to Maxisorp-96 plates followed by incubation with hu4D5Fab-
phage and its variants (4 × 109 phage). The plates were washed, and bound phage was
measured and developed as described in Junutula et al., 2008, J. Immunol. Methods 332,
41-52.
9
Supplementary Figure 6. Anti-MUC16 3A5 humanization. S.Fig.6a. Light chain graft
S.Fig.6b. Heavy chain graft
10
S.Fig.6c. CDR-H3 sequences from selected 3A5 clones.
Murine anti-MUC16 (“mu3A5”) was grafted onto a human IgG1 framework (“3A5-
graft”), and then binding affinity was restored by CDR repair as described in the text,
yielding 3A5.v4. Sequences are aligned against the consensus human VLkappaI (SFig. 6a,
“HuKI”) and the consensus human VHsubgroupIII (SFig. 6b, “hum III”); differences from
the human consensus sequences are highlighted. The sequence, structural and contact
definitions of the CDRs used for the design of the 3A5 CDR graft are delineated by bars.
Changes made to the 3A5 CDR graft during humanization are boxed. (SFig. 6c) Changes
identified in CDR-H3 during CDR repair of the 3A5 graft that restore binding to MUC16.
All six CDRs in the 3A5 CDR graft were randomized with a bias towards their original
sequence. Following four rounds of selection against the MUC16 ECD construct, several
unique clones were identified in the CDR-H3 library. Their observed frequency out of 79
sequenced clones is indicated under “sibs”.
11
Supplementary Figure. 7. Anti-MUC16 TDC binds only to MUC16-expressing cell
lines.
MUC16-positive and MUC16-negative cell lines were identified using a combination of
quantitative real-time RT-PCR and protein detection methods. A subset of these lines
was evaluated for anti-MUC16 TDC binding by flow cytometry, using 2 µg/mL TDC or
binding buffer only, followed by a phycoerythrin-labeled secondary antibody. The extent
of binding is plotted as geometric mean fluorescence on a linear (SFig. 7a) or log10 scale
(SFig. 7b). MUC16-positive cells (OVCAR-3, OVCAR-3/luc, PE01, and transfected
PC3/MUC16) gave robust binding, whereas MUC16-negative cells (PC3/vector, IGROV-
1, SK-OV-3, and HEK 293S) were labeled at background levels.
12
Supplementary Figure 8. Analytical characterization of TDC and ADC. S.Fig.8a. Schematic diagram for the MC-vc-PAB-MMAE attached to the THIOMAB.
13
S.Fig.8b and 8c. Hydrophobic interaction chromatographic analyses for TDC and ADC.
6%
35%
59%
0 1 2
c
Thio-ch3A5-VC-MMAEb
0 1 23 4
68
12%
3%
42%
3%
32%
7%1%
ch3A5-VC-MMAE
14
S.Fig.8d. Hydrophobic interaction chromatographic analysis for TDC from large-scale
conjugation (multi gram scale).
Hydrophobic interaction chromatographic analyses were performed for the anti-MUC16
TDC (SFig. 8b), ADC (SFig. 8c), and TDC (SFig. 8d) prepared on a large scale using
the improved conjugation process. Samples were injected onto a Butyl HIC NPR column
and eluted with linear gradient from 0 to 70% B at 0.8 ml/min (Buffer A: 1.5 M
ammonium sulfate in 50 mM potassium phoshate, pH 7, Buffer B: 50 mM potassium
phosphate pH 7, 20% isopropanol). An Agilent 1100 series HPLC system equipped with
a multi wavelength detector and Chemstation software was used to resolve and quantitate
antibody species in the antibody. Peak identities were assigned based on LC/MS analysis
of ADC and TDC. For Fig. 8b and 8c, the DAR values are indicated below the
corresponding peaks.
15
Supplementary Figure 9: Mapping the antibody region of cytotoxic drug attachment to
the TDC-Fab.
SFig. 9a. A 280 absorbance spectrum of reduced, denatured light and heavy chain
portions of the MC-vc-PAB-MMAE-conjugated Fab. Deconvoluted masses of light
chain (SFig. 9b) and heavy chain (SFig. 9c) portions are consistent with drug labeling on
the heavy chain only. The mass of 24189 observed in SFig. 9c results from the loss of
MMAE-CO2 (minus 762 daltons), which is a characteristic fragmentation of the drug
(see Supplementary Table 2 for expected and observed masses).
16
Supplementary Figure 10. Peptide mapping – Identification of peptide(s) containing
cytotoxic drug in the TDC-Fab.
Peptide fragments from tryptic digests of maleimidylated purified Fabs of unconjugated
Thio-3A5 (SFig. 10a) and conjugated TDC (SFig. 10b). Peptides labeled with MC-vc-
PAB-MMAE elute later due to increased hydrophobicity. Four drug-conjugated peptides
(labeled with *) eluted at the end of the gradient. They can also be identified as cytotoxic
drug containing peptides by a characteristic in-source fragmentation ion (m/z 718.5) that
is observed in all MC-vc-PAB-MMAE containing mass spectra. SFig. 10c shows an
overlay of the extracted ion chromatogram from the unconjugated (black) and the
conjugated digests (pink). The strongest peaks coincide with the late eluting peaks of the
conjugate digest. All four peaks were identified as complete or partial tryptic cleavage
fragments located around the mutated cysteine in position 118 of the Fab-HC. The m/z
17
ions in the main peak at 29.05 min deconvoluted to give a mass of 3962 daltons, which is
the expected mass for peptide HC99-120 + 1 drug. The drug-containing peptide masses
did not map to any other region of the protein. The peak labeled with an arrow in SFig.
10a is the maleimide labeled peptide HC99-120. Supplementary Table 3 lists the
expected and observed fragment masses of the MC-vc-PAB-MMAE labeled peptides
isolated from TDC.
Supplementary Figure 11. TDC and ADC have comparable in vitro activities.
PC3/MUC16 (“MUC16”), PC3 empty vector (“neo”), and OVCAR-3 cells were
incubated with serial dilutions of humanized anti-MUC16 ADC and TDC as indicated in
the Methods. At the end of the incubation, viable cell numbers were determined using a
luminescence assay and were normalized to the “no treatment” values. IC50 values for
OVCAR-3 and PC3/MUC16 proliferation were determined using a four-parameter curve
fit and are indicated next to the corresponding symbol.
18
Supplementary Figure 12. Anti-MUC16 TDC is active against multiple mouse
xenograft tumor models.
S.Fig. 12a. Primary pancreatic tumor model.
19
S.Fig. 12b. PE01 cell line tumor model.
Xenograft tumors were implanted at a subcutaneous site (SFig. 12a: primary pancreatic
tumor transplant models) or in the right thoracic mammary fat pad (SFig. 12b: PE01 cell
line). Mice received single doses of the indicated conjugates or vehicle on Day 0, after
tumors were established; humanized antibodies were used throughout. Where indicated,
the “control” conjugate is a non-binding conjugate dosed to match the highest dose of
anti-MUC16 TDC in terms of mg/kg IgG (SFig. 12a) or µg/m2 MMAE (SFig. 12b).
Tumor burden is plotted as tumor volume. (SFig. 12a) Drug-antibody ratios are 1.6 for
the anti-MUC16 TDC and 2.0 for the non-binding TDC. (SFig. 12b) Drug-antibody
ratios are 2.0 for the anti-MUC16 TDC and 1.8 for the non-binding TDC.
20
Supplementary Figure 13. The anti-MUC16 TDC produced using an optimized
conjugation process with 2 drugs/antibody is efficacious in xenograft tumor models.
The anti-MUC16 TDC produced using an optimized conjugation process with 2
drugs/antibody (see Supplementary Fig. 8d) is efficacious in xenograft tumor models.
The refined process used to generate TDC with the desired stoichiometry of two drugs
per antibody retains the activity observed with the DAR = 1.6 species (main text Figure
3). The OVCAR-3 mammary fat pad model was employed as described in the main text.
Drug-antibody ratios are 2.0 for the anti-MUC16 TDC and 1.8 for the non-binding TDC.
21
Supplementary Figure 14. Chimeric anti-MUC16 TDC is better tolerated in rats than
ADC.
Sprague-Dawley rats were dose on Day 1 with vehicle or with chimeric anti-MUC16
ADC or TDC. Dose levels are given as µg/m2 MMAE and mg/kg IgG; e.g., “940/24.2”
indicates a dose of 940 µg/m2 MMAE and 24.2 mg/kg IgG. Blood was drawn on Days 5
and 12 for hematology (neutrophils; SFig.14a) and serum chemistry (serum AST;
SFig.14b); average values for each group are shown. Body weight was measured daily
and is plotted as average body weight change for each group (SFig.14c). SFig.14d: Day
12 bleeds were used to quantify total antibody (“Total IgG”) and antibody bearing at least
one MMAE (“Conjugate”), and the percent of antibody bearing at least one MMAE was
calculated as the ratio (“% Conjugated”). For convenience, both the absolute
concentrations and percent conjugated are plotted on the same graph; therefore, the Y-
axis indicates µg/mL or percent.
22
Supplementary Figure 15. Anti-MUC16 ADC and TDC have similar pharmacokinetic
properties in tumor-bearing mice.
Mice bearing OVCAR-3/mfp xenograft tumors were dosed once with 6 mg/kg
humanized ADC or TDC on Study Day 0. Blood was drawn at the indicated intervals,
and circulating levels of total and conjugated IgG were measured and are plotted over
time.
23
Supplementary Figure 16. Binding of human anti-MUC16 antibodies to FcRn.
Binding of human anti-MUC16 antibodies to human (SFig. 16a), cynomolgus monkey
(SFig. 16b), rat (SFig. 16c) and mouse (SFig. 16d) FcRn molecules at pH 6.0 measured
by ELISA (see Supplementary Methods). ADC and TDC showed similar binding to all
four FcRn molecules (see Supplementary Table 5).
24
Supplementary Figure 17. Dissociation of bound anti-MUC16 antibodies from FcRn.
Dissociation of bound anti-MUC16 antibodies from human (SFig. 17a), cynomolgus
monkey (SFig. 17b), rat (SFig. 17c) and mouse (SFig. 17d) FcRn molecules at pH 7.4
measured by ELISA. Anti-MUC16 antibodies were bound to FcRn at pH 6.0 in the FcRn
ELISA (see Supplementary Methods). After plates were washed, a pH 6.0 or 7.4 buffer
was added and the plates were incubated for 45 minute to allow dissociation. All
antibodies showed significant dissociation from all four FcRn molecules at pH 7.4: The
absorbance readings of 200 ng/ml antibody at pH 7.4 were less than those of 20 ng/ml
antibody at pH 6.0 (> 90% dissociation).
25
Supplementary Table 1. Identification of additional sites that allow conjugation
specifically to LC or HC-Fab region of the antibody.
THIOMABs % Biotinylation
Trastuzumab-wt 0
LC-V15C 94
LC-V110C 40
LC-S114C 95
LC-S121C 97
LC-S127C 91
LC-A153C 7
LC-S168C 91
LC-V205C 100
HC-S112C 100
HC-S113C 63
HC-A114C 100
HC-S115C 63
HC-T116C 100
Trastuzumab and its THIOMAB variants were expressed and purified from 293 cells as
described in the Methods. Purified proteins were conjugated with biotin-PEO-maleimide
as described in Fig.1. The amount of conjugated biotin to antibody was quantitated by
LC/MS analysis. Two biotin molecules/antibody is considered to be 100% biotinylation.
26
Supplementary Table 2. Comparison of humanized anti-MUC16 variants and their
binding to MUC16 extracellular domain (ECD).
Kabat # 95
96
97
98
99
100
102
CA125 MUC16 ECD OVCAR-3 MUC16 ECD
ch3A5 WD G G L T Y 0.3 2.3 0.3 0.7
3A5 graft WD G G L T Y nd nd 7.1 96
3A5v8 W T S G L D S 0.4 2.6 0.5 2.3
3A5v1 WA S G L D Y 0.6 3.0 0.5 2.1
3A5v7 WK S G L D S 1.2 3.3 0.8 nd
3A5v4 W T S G L D Y 0.3 2.4 0.8 0.9
CDR-H3 BIAcore KD (nM)ELISA IC50 (nM)
27
Supplementary Table 3. MC-vc-PAB-MMAE is specifically conjugated to the HC-Fab
of the THIOMAB.
Expected and observed masses of unconjugated and MC-vc-PAB-MMAE
conjugated anti-MUC16 Fab fragments. Species labeled with * are a result
of in-source fragmentation of the drug (loss of 762 daltons). LC-MS
analyses of the Fab and Fc fragments show drug-adducts only on the Fab
portion of the antibody and, more specifically, only on the HC of the Fab.
ID
predicted
mass
observed
unconjugated
sample
observed VC-
MMAE-conjugated
sample
non-reduced
FC 53302 53300 53300
Fab 47108 47108
Fab+VC-MMAE 48424 48424
Fab+VC-linker * 47664 47662
reduced
Fab-LC 23483 23483 23483
Fab-HC 23633 23633
LC+ VC-MMAE 24799
HC+VC-MMAE 24949 24950
HC+VC-linker * 24189 24188
28
Supplementary Table 4. Peptide mapping – Identification of MC-vc-PAB-MMAE
conjugated peptide in the TDC-Fab.
Observed peptide masses and residue locations from the tryptic digest of
maleimide (NEM)-treated TDC-Fab. The major observed mass (highlighted
in yellow) at elution time 29.05 min. corresponds to the HC peptide residues
99-120 (WTSGLDYWGQGTLVTVSSCSTK). All other identified masses
correspond to incomplete tryptic cleavages that incorporate peptide 99-120.
elution time start res # end res # peptide mass
31.25 68 120 7431.63
31.25 68 132 8599.97
31.25 73 120 6826.93
31.25 73 132 7995.27
31.66 77 120 6395.49
31.66 77 132 7563.83
30.21 88 120 5060.94
30.21 88 132 6229.29
29.05 99 120 3692.6
29.05 99 132 4860.94
29
Supplementary Table 5. Relative binding affinities of human anti-MUC16 antibodies to
FcRn measured by ELISA.
aRelative affinities were calculated as described in the Methods using data shown in
Supplementary Fig. 16.
bWe used antibody preparations with low amounts of aggregate (determined by size
exclusion chromatography) for this study since aggregation can increase the apparent
binding affinities in the FcRn ELISA. The relative binding affinities for human and
monkey FcRn molecules, but not for rat and mouse FcRn molecules, increased twofold
when the amount of aggregate in 3A5 increased from 0.4% to 5.2%. Relative binding
affinities for all four FcRn molecules did not increase significantly when the amounts of
aggregate in 3A5 ADC and 3A5 TDC increased from 0.6% to 3.2% and from 1.5% to
6.7%, respectively.
Relative binding affinity to FcRn at pH 6.0a Anti-MUC16
antibody
%
Aggregationb Human Monkey Rat Mou s e
3A5 0 . 4 1 . 0 1 . 0 1 . 0 1 . 0
AD C 0 . 6 2 . 1 1 . 5 0 . 9 0 . 8
Thio-3A5 1 . 1 1 . 3 1 . 2 1 . 0 1 . 0
T D C 1 . 5 2 . 1 1 . 7 1 . 0 1 . 0
30
Supplementary Methods
Cell lines and proteins
PC3 and OVCAR-3 cell lines were from the American Type Culture Collection
(Rockville, MD). PC3/neo and PC3/MUC16 cell lines have been described previously
(Chen, Y. et al. 2007, Cancer Res 67, 4924-4932). OVCAR-3 cells were cultured in
RPMI-1640 medium supplemented with 20% fetal bovine serum. PC3/neo and
PC3/MUC16 cells were cultured in a 50/50 mixture of Dulbecco’s modified Eagle’s
medium and Ham’s F-12 medium supplemented with glucose, 10% fetal bovine serum
(Sigma), and 250 µg/mL G-418 (Gibco). The PE01 ovarian adenocarcinoma cell line
was obtained from Cancer Research Technology Limited (London, UK) and cultured in
RPMI 1640 media plus 1% L-glutamine with 10% fetal bovine serum. Some binding
studies were performed using a MUC16 ECD protein derived from the “MUC16TMlong”
sequence reported previously by introducing a stop codon just prior to the transmembrane
domain (Chen, Y. et al. 2007, Cancer Res 67, 4924-4932). A commercial preparation of
CA125 (US Biologicals) was also used for some binding studies.
Anti-MUC16 antibody humanization
A combination of sequence (Kabat, E.A. & Wu, T.T. 1971, Ann N Y Acad Sci 190, 382-
393; Kabat, E. A., Wu, T. T., Perry, H. M., Gottesman, K. S. & Foeller, C. 1991, 5th edn
(National Institutes of Health, Bethesda)), structural (Chothia, C. & Lesk, A.M. 1987, J.
Mol. Biol. 196, 901-917) and contact (MacCallum, R.M., et al. 1996, J. Mol. Biol. 262,
732-745) CDR definitions were used to generate a CDR graft of 3A5 in the background
of the human consensus VLkappaI and VHsubgroupIII frameworks.
The 3A5 graft was displayed as a Fab on phage, and diversity was introduced into the
CDR regions using Kunkel mutagenesis (Lee, C.V. et al. 2004, Journal of Molecular
Biology 340, 1073-1093; Liang, W.-C. et al. 2007, Journal of Molecular Biology 366,
31
815-829). A mutagenesis strategy, introducing a mutation rate of approximately 50% at
each CDR position simultaneously, was employed in order to maintain a bias towards the
original 3A5 CDR sequences (Liang, W.-C. et al. 2007, Journal of Molecular Biology
366, 815-829; Gallop, M.A. et al. 1994, J. Med. Chem. 37, 1233-1251). Phage libraries
were panned separately against a MUC16 extracellular domain (ECD) protein. Several
selected clones were reformatted, expressed as IgG and characterized for binding to
antigen (Supplementary Table 4). The affinity of clone 3A5.v4 for MUC16 was greatly
improved compared to the 3A5 graft and comparable to that of the 3A5 chimera.
Additional changes (VL-S49Y and VH-N52S) were made to 3A5.v4 to improve
production, and these substitutions had no effect on binding affinity.
Peptide mapping to identify MC-vc-PAB-MMAE labeled peptides in anti-MUC16
TDC.
To simplify the peptide map generated from a tryptic digest of conjugated and
unconjugated 3A5 THIOMAB, the antibody was first subjected to a limited digestion
with endopeptidase Lys-C. This cleaves the antibody above the hinge region and
generates an Fc and a Fab portion. Deconvoluted masses from the LC-MS separation of
Fc and Fab show that only the Fab portion of the antibody is labeled with MC-vc-PAB-
MMAE. (Unlabeled Fab contains glutathione adducts that were not removed during the
initial reduction of the antibody, leaving the mutated cysteine unavailable for drug
labeling.) 3A5 THIOMAb-Fabs were purified by ion exchange chromatography using a 1
ml HiTrap SP FF column (GE Healthcare Bio-Science AB). The digest was loaded onto
the column in 50 mM Na-Acetate pH 5.5 and eluted with a 30 ml gradient to 0.3 M NaCl
in Na-Acetate. Further LC-MS analysis of the reduced Fab locates the drug exclusively
on the heavy chain portion (SFig. 9).
The isolated Fab fragment was diluted to 5.5 M guanidine in 100 mM Tris pH 8 and
reduced with 1mM DTT for 1 hr at 37 °C. The reduced cysteines were reacted with 2 mM
N-ethyl-maleimide for 30 min. LC-MS analysis shows the reduced Fab light chain mass
consistent with 5 NEM adducts on both the conjugated and unconjugated protein. The
heavy chain of the unconjugated Fab shows mass increase of 6 NEM molecules and the
32
MC-vc-PAB-MMAE conjugated HC has 5 NEM adducts as predicted. The NEM labeled
reduced Fabs were purified on 5 ml Zeba Desalt Spin Columns (Pierce) to remove the
guanidine.
Trypsin was added at a ratio of 1:50 (w:w) and the protein was allowed to cleave for 48
hrs. LC-MS analysis of the digests was performed on a TSQ Quantum Ultra (Thermo)
using a PLPR-S 300A, 3um column 50x2.1 mm (Polymer Laboratories). The mobile
phases were 0.05% TFA in water (A) and 0.04% TFA in acetonitrile (B). A gradient of 2-
60% B over 60 min. was used to elute the tryptic peptides. The peptide maps of the
conjugated and unconjugated proteins are shown in SFig. 10. Four drug-conjugated
peptides eluted at the end of the gradient. They can be identified by a characteristic
MMAE drug fragment ion (m/z 718.5) that is observed in all MC-vc-PAB-MMAE
containing mass spectra. This fragment was observed exclusively in the four late-eluting
peptides. (Fig 5 scan for 718 fragment across the spectrum). The deconvoluted masses
obtained from these peptides are all consistent with expected masses from complete or
partial tryptic cleavages of the region surrounding the A114C mutation as shown in
Supplementary Table 3.
Additional efficacy studies.
The studies with ovarian and pancreatic cancer transplant models were performed at
Oncotest GmbH (Freiburg, Germany) using female NMRI nu/nu mice (Taconic). The
PE01 xenograft model was established by inoculating female NCR.nude mice (Taconic)
with 5x106 cells per mouse into the right thoracic mammary fat pad after estrogen
supplementation. Single doses of antibody-drug conjugates were given on Day 0, after
tumors were established and mice were randomized according to tumor volumes.
Competition for anti-MUC16 binding by ECD proteins from different species.
Recombinant portions of the ECD of human, cynomolgus monkey, and rat MUC16
comprising exactly three SEA domains each were expressed using Chinese hamster ovary
cells. CA125 (US Biologicals, Swampscott, MA) was biotinylated (Bt-CA125) and thio-
3A5 was conjugated to ruthenium (Ru-3A5) using conventional methods. Bt-CA125,
33
Ru-3A5, and varying concentrations of ECD proteins were combined in solution and
incubated for one hour at room temperature. The analyte was then applied to a
streptavidin-coated plate (MSD, Gaithersburg, MD) to capture the Bt-CA125 and any
bound Ru-3A5. After washing, bound Ru-3A5 was detected using an MSD 6000 imager
(MSD, Gaithersburg, MD). Reduction in signal indicated inhibition of Ru-3A5 binding
to Bt-CA125 by the MUC16 ECD proteins. IC50 values for the ECD proteins were
calculated from the titration curves using standard methods.
Surface plasmon resonance analysis using BiacoreTM.
Binding affinities of anti-MUC16 Thio3A5 and the TDC (Thio3A5-MC-vc-PAB-
MMAE) to a recombinant MUC16 ECD were determined by surface plasmon resonance
analysis on a Biacore 3000 instrument (GE Healthcare Biacore, Inc.; Piscataway, NJ).
Antibodies were immobilized onto a CM5 sensor chip through an indirect capturing
reagent, goat anti-human IgG Fc. Various concentrations of MUC16 ECD in HEPES-EP
buffer (0.01 M HEPES, 0.15 M NaCl, 3 mM EDTA and 0.005% Polysorbate 20, pH 7.4)
were injected over immobilized anti-MUC16 antibodies at a flow rate of 100 µl/min for
2.5 minutes and the dissociation was allowed for 15 minutes. Regeneration was achieved
by injecting 25 µl of 10 mM Glycine, pH 1.5. Herceptin was used as a reference
antibody. Sensorgrams were generated after in-line reference cell correction followed by
buffer sample subtraction. Three independent experiments were carried out using two
separate sensor chips. The average dissociation equilibrium constant (KD) was calculated
based on the ratio of association and dissociation rate constants, obtained by fitting
sensorgrams with a first order 1:1 binding model using BiaEvalution Software (version
3.2).
FcRn binding measurements by ELISA.
Binding of human anti-MUC16 antibodies to FcRn at pH 6.0 was measured by ELISA.
Human FcRn ELISA was performed as described previously (Shields, R.L. et al. 2001,
J Biol Chem. 276, 6591-6604). Cynomolgus monkey, rat and mouse FcRn ELISA were
performed similarly. Briefly, Maxisorp 96-well plates were coated with NeutrAvidin
(Pierce, Rockford, IL) and blocked with bovine serum albumin. Biotinylated extracellular
34
domain of human, cynomolgus monkey, rat or mouse FcRn (Genentech, Inc.) was added
to the plates and incubated for one hour. After a wash step, anti-MUC16 antibodies (1.6-
200 ng/ml in twofold serial dilution at pH 6.0) were added in duplicate and incubated at
room temperature for 2 hours. Bound antibody was detected with horseradish peroxidase-
conjugated goat F(ab’)2 anti-human IgG F(ab’)2 antibody (Jackson ImmunoResearch,
West Grove, PA) followed by 3,3’,5,5’-tetramethyl benzidine (Kirkegaard & Perry
Laboratories, Gaithersburg, MD) as the substrate. Absorbance was read at 450 nm. For
data analysis, the average absorbance at 3.1 and 200 ng/ml anti-MUC16 3A5 antibody
(mid-OD) was calculated. Antibody concentrations corresponding to this mid-OD were
determined from the titration curves using a four-parameter regression curve-fitting
program. Relative affinities were calculated by dividing the mid-OD concentration of
anti-MUC16 3A5 by that of each antibody. To assess dissociation of bound antibodies
from FcRn at pH 7.4, the FcRn ELISA were performed as described above, except that a
dissociation step was added. After the sample incubation step, the plates were washed
and a pH 6.0 or 7.4 buffer was added. The plates were incubated for 45 minutes to allow
dissociation.