Site-specific and stoichiometric modification of antibodies by bacterial transglutaminase

Article abstract of DOI:10.1002/anie.201004243

Spot on: Bacterial transglutaminase enables the site-specific modification of Gln side chains of tumor-targeting antibodies with various probes containing lysine or lysine surrogates. The method yields completely homogeneous immunoconjugates with a defined stoichiometry. In comparative in vivo studies with xenografted mice the pharmacological profiles for enzymatically conjugated antibodies were better than those of chemically modified analogues. Copyright

Full text of DOI:10.1002/anie.201004243

Angewandte
Chemie
DOI: 10.1002/anie.201004243
Protein Labeling
Site-Specific and Stoichiometric Modification of Antibodies by
Bacterial Transglutaminase**
Simone Jeger, Kurt Zimmermann, Alain Blanc, Jꢀrgen Grꢀnberg, Michael Honer,
Peter Hunziker, Harriet Struthers, and Roger Schibli*
The therapeutic efficacy of antibodies can be substantially
enhanced by conjugation of cytotoxic compounds such as
chemotherapeutics and particle-emitting radionuclides.[1]
Intuitively one would assume that the therapeutic index
would improve as the number of cytotoxic entities conjugated
to the antibody increases. However, recent studies on
auristatin–antibody conjugates in mice have demonstrated
that a drug/antibody molar ratio of 4:1 results in optimal
efficacy and in vivo tolerability.[2]
Unfortunately, conventional chemical strategies for pro-
tein modification are difficult to control and give rise to
heterogeneous populations of immunoconjugates with varia-
ble stoichiometries, each of which has its own in vivo
characteristics.[3] The introduction of artificial, bio-orthogonal
groups for site-specific and stoichiometric protein modifica-
tion offers a potential solution to this problem.[4–6] Such
strategies are en vogue but are often laborious and still risk
product heterogeneity.
Transglutaminases (TGs, E.C. 2.3.2.13) catalyze acyl-
transfer reactions between the g-carboxamide group of
glutamine (a side chain, which is otherwise chemically inert
under physiological conditions) and the primary e-amino
group of lysine, to form catabolically stable isopeptide bonds
(Figure 1a).[7] Most TGs are promiscuous with respect to the
lysine substrate and accept even simple 5-aminopentyl groups
as lysine surrogates. The criteria for a glutamine residue to be
recognized by the enzyme, however, are much more stringent:
it should be both located in a flexible region of the protein and
flanked by specific amino acids.[8] Given this inherent
selectivity, we hypothesized that TG would be an alternative
for the site-specific and stoichiometric functionalization of
antibodies.[9] For this study we used bacterial transglutami-
nase (BTG) because it is robust, inexpensive, and easy to
handle.
Figure 1. a) TG-mediated modification of Gln (Q) with a substrate
containing lysine or a lysine surrogate. b) Substrates used in this
study.
Our group is interested in radioimmunoconjugates for
diagnostic and therapeutic applications, where low off-target
accumulation of radioactivity is crucial. Earlier studies
performed with radiolabeled monoclonal antibodies (mAbs)
demonstrated that high numbers of metal chelators adversely
affect the biological behavior of radioimmunoconjugates.[10,11]
Therefore, we tested the features of BTG for the preparation
of immunoconjugates that are functionalized with different
metal chelators and radiolabeled with different diagnostic and
therapeutic radionuclides. Deferoxamine (DF, 1), an antidote
for metal poisoning, has recently been identified as a suitable
chelator for radionuclides such as 67Ga and 89Zr.[12] During the
course of our studies we recognized that without further
derivatization deferoxamine is already a potent BTG sub-
strate. Furthermore, the metal chelating system 4-(1,4,8,11-
tetraazacyclotetradec-1-yl)methyl benzoic acid (CPTA, 2)
was derivatized with a 1,5-diaminopentane (cadaverine)
spacer (Figure 1b; see the Supporting Information for details
on the synthesis). To probe the scope of the new strategy we
investigated other (model) substrates, which are of potential
[*] Dr. S. Jeger, Dr. M. Honer, Dr. H. Struthers, Prof. Dr. R. Schibli
Department of Chemistry and Applied Biosciences
ETH Zꢀrich, 8093 Zꢀrich (Switzerland)
E-mail: roger.schibli@pharma.ethz.ch
Homepage: http://www.pharma.ethz.ch
K. Zimmermann, A. Blanc, Dr. J. Grꢀnberg
Center for Radiopharmaceutical Science ETH-PSI-USZ
Paul Scherrer Institute, Villigen-PSI (Switzerland)
Dr. P. Hunziker
Functional Genomics Center, UNI ETH Zurich (Switzerland)
[**] This work was supported by the Swiss National Science Foundation
(No. 112437).
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201004243.
Angew. Chem. Int. Ed. 2010, 49, 9995 –9997
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
9995

Communications
interest for mAb modification such as the biotin–cadaverine
functionalization of the HC (see the Supporting Information).
For comparison, mass spectra of chCE7 and RTX chemically
functionalized at lysine with isothiocyanate or hydroxysucci-
nimide analogues of substrates 1–3 showed varying numbers
of molecules attached to both the HC and LC (see the
Supporting Information). Chemical conjugation therefore
yields a population of conjugates with two types of molecular
heterogeneity: first, species with different numbers of mole-
cules per mAb and second, isostoichiometic conjugates, in
which the substrates are linked to different sites of the mAb.
MALDI-TOF MS analyses unambiguously proved that Q295
and Q297 in chCE7agl are indeed the sites of BTG-mediated
conjugation. The tryptic fragment containing Q295 and Q297
([EEQ295YQ297STYR + H]+, calcd 1203.52, found 1203.59) of
the unconjugated chCE7agl was invariably modified with two
substrates. For example, the spectrum of the digested frag-
ment from the reaction with substrate 1 corresponds to
[EEQ295(1)YQ297(1)STYR + H]+ (calcd 2290.17, found
2290.41; Figure 2d).
Comparative B-factor analysis of the glycosylated Fc
fragment of RTX (DOI:10.2210/pdb1L6x/pdb) with that of a
deglycosylated Fc fragment of a human IgG1 (DOI:10.2210/
pdb3dnk/pdb) revealed significant shifts of the Ca positions in
the C/E loop towards a more open conformation (e.g.
approximately 5 ꢀ was observed for positions Q295 and N/
D297 in the deglycosylated variant; see the Supporting
Information). Based on these findings we concluded that
the differences between native and aglycosylated mAbs with
respect to enzymatic modification must be predominantly
related to the greater flexibility of the C/E loop. To prove this
hypothesis, we removed the glycans from chCE7 and RTX
conjugate 3, the galactopyranose–lysine derivative 4, and the
fluorescent dansyl–cadaverine derivative 5. The anti-L1-
CAM mAb chCE7 and the commercial anti-CD20 antibody
rituximab (RTX) were functionalized with these substrates in
the presence of BTG. The immunoconjugates were charac-
terized by LC–MS after reduction with dithiothreitol (for
experimental details see the Supporting Information).
To our surprise, no modification of native chCE7 or RTX
was observed with any of the substrates in the presence of
BTG, despite the fact that sequence analysis revealed
abundant Q residues (see the Supporting Information). In
parallel to these experiments we also tested the aglycosylated
mutant chCE7agl, in which asparagine at position 297 has
been replaced by glutamine to abort N-glycosylation.[13] This
mutation was originally engineered to reduce the serum half-
life of the radiolabeled immunoconjugate.[13] In the context of
the present study, the N297Q mutation has two interesting
consequences: 1) the antibody has an additional glutamine
residue, and 2) removal of the Fc glycans leads to an increased
mobility of the C/E loop (Q295–T299) as reported in the
literature.[14–16] The deconvoluted mass spectra of the
chCE7agl conjugates revealed no modification of the LC
but exactly two modifications of the HC with all of the
substrates tested (see the Supporting Information). Thus,
completely homogeneous immunoconjugates with a defined
substrate/mAb stoichiometry of 4:1 (the optimal drug/mAb
ratio found for auristatin immunoconjugates[2]) were formed
(Figure 2b,d). SDS PAGE and western blot analyses of
immunoconjugates enzymatically functionalized with the
biotin and the fluorescent substrates confirmed the exclusive
Figure 2. a) BTG recognizes exclusively Q295 located in the Fc region of deglycosylated IgGs (IgGdegl) as a site for modification with a suitable
substrate. This leads to an exact substrate/mAb ratio of 2:1. Deglycosylation: mAb (1 mgmLÀ1), PNGase F (100 U), PBS (pH 7.4), 378C,
overnight. Conjugation: mAB (1 mgmLÀ1), BTG (1 UmLÀ1), 1–5 (400 mm), PBS (pH 8), 378C. b) In the aglycosylated mutant chCE7agl (N297Q)
Q297 is recognized as a second site for modification resulting in a exact substrate/mAb ratio of 4:1 under the same reaction conditions.
c,d) MALDI-TOF mass spectra of the tryptic fragments containing Q295 or Q295 and Q297 (left) and the corresponding immunoconjugates
&
(right) proved the exact location and stoichiometry of the modification. e) Time-dependent BTG-mediated modification of chCE7degl ( ) and
*
chCE7agl ( ) with 1. The substrate/mAb ratio reaches a plateau after ca. 4 h.
9996 www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 9995 –9997

Angewandte
Chemie
with N-glycosidase F (PNGaseF). The deglycosylated anti-
bodies (chCE7degl and RTXdegl) were incubated with BTG
and substrates 1 and 2. MS analysis of the reaction products of
both deglycosylated antibodies showed products with exactly
one substrate per HC while the LC remained unmodified.
This gave rise to homogeneous immunoconjugates with a
substrate/antibody stoichiometry of exactly 2:1. MALDI-
TOF MS confirmed Q295 to be the site of modification
(Figure 2c).
To assess the effect of the moderate substrate/mAb ratio
and the homogeneity of the immunoconjugates on the in vivo
behavior, chCE7agl-(1)4 was radiolabeled with 67Ga and
injected into female nude mice bearing SKOV3ip human
ovarian carcinoma xenografts. Maximal mean uptake in the
tumor was 58.7% IDgÀ1 at 72 hours after injection (p.i.;
Figure 3a and Table S11 in the Supporting Information).
potential to be readily scaled up), and leads to homogenous
immunoconjugates with defined stoichiometries. The in vivo
characteristics of such immunoconjugates are superior to
those prepared using chemical coupling methods. Since
position 295 is located in the constant Fc region, the
enzymatic conjugation approach is applicable not only to
other human IgG1s, but also to mAbs belonging to subtypes
IgG2, IgG3, and IgG4, all of which conserve the Q295
residue.[17] Thus, the method is broadly applicable and permits
the systematic assessment and improvement of immunocon-
jugates.
Received: July 12, 2010
Revised: September 13, 2010
Published online: November 25, 2010
Keywords: bacterial transglutaminase · enzymes ·
.
protein labeling · radioimmunoconjugates
[1] P. Carter, Nat. Rev. Cancer 2001, 1, 118 – 129.
[2] K. J. Hamblett, P. D. Senter, D. F. Chace, M. M. C. Sun, J. Lenox,
C. G. Cerveny, K. M. Kissler, S. X. Bernhardt, A. K. Kopcha,
R. F. Zabinski, D. L. Meyer, J. A. Francisco, Clin. Cancer Res.
2004, 10, 7063 – 7070.
[3] T. U. Gerngross, Nat. Biotechnol. 2004, 22, 1409 – 1414.
[4] J. R. Junutula, H. Raab, S. Clark, S. Bhakta, D. D. Leipold, S.
Weir, Y. Chen, M. Simpson, S. P. Tsai, M. S. Dennis, Y. Lu, Y. G.
Meng, C. Ng, J. Yang, C. C. Lee, E. Duenas, J. Gorrell, V. Katta,
A. Kim, K. McDorman, K. Flagella, R. Venook, S. Ross, S. D.
Spencer, W. Lee Wong, H. B. Lowman, R. Vandlen, M. X.
Sliwkowski, R. H. Scheller, P. Polakis, W. Mallet, Nat. Biotech-
nol. 2008, 26, 925 – 932.
[5] P. Wu, W. Q. Shui, B. L. Carlson, N. Hu, D. Rabuka, J. Lee, C. R.
Bertozzi, Proc. Natl. Acad. Sci. USA 2009, 106, 3000 – 3005.
[6] W. Peters, S. Willnow, M. Duisken, H. Kleine, T. Macherey, K. E.
Duncan, D. W. Litchfield, B. Lꢁscher, E. Weinhold, Angew.
Chem. 2010, 122, 5296 – 5299; Angew. Chem. Int. Ed. 2010, 49,
5170 – 5173.
Figure 3. Time-dependent organ distribution of a) enzymatically and
b) chemically conjugated 67Ga-chCE7agl-(1)4 in tumor-bearing mice.
[7] L. Lorand, R. M. Graham, Nat. Rev. Mol. Cell Biol. 2003, 4, 140 –
156.
[8] A. Fontana, B. Spolaore, A. Mero, F. M. Veronese, Adv. Drug
Delivery Rev. 2008, 60, 13 – 28.
[9] M. Bartholomꢂ, J. Valliant, K. P. Maresca, J. Babich, J. Zubieta,
Chem. Commun. 2009, 493 – 512.
[10] K. Knogler, J. Grꢁnberg, I. Novak-Hofer, K. Zimmermann, P. A.
Schubiger, Nucl. Med. Biol. 2006, 33, 883 – 889.
[11] D. L. Kukis, G. L. DeNardo, S. J. DeNardo, G. R. Mirick, L. A.
Miers, D. P. Greiner, C. F. Meares, Cancer Res. 1995, 55, 878 –
884.
[12] E. C. Dijkers, J. G. Kosterink, A. P. Rademaker, L. R. Perk,
G. A. van Dongen, J. Bart, J. R. de Jong, E. G. de Vries, M. N.
Lub-de Hooge, J. Nucl. Med. 2009, 50, 974 – 981.
[13] K. Knogler, J. Grꢁnberg, K. Zimmermann, S. Cohrs, M. Honer,
S. Ametamey, P. Altevogt, M. Fogel, P. A. Schubiger, I. Novak-
Hofer, Clin. Cancer Res. 2007, 13, 603 – 611.
[14] S. Matsumiya, Y. Yamaguchi, J. Saito, M. Nagano, H. Sasakawa,
S. Otaki, M. Satoh, K. Shitara, K. Kato, J. Mol. Biol. 2007, 368,
767 – 779.
[15] M. J. Feige, S. Nath, S. R. Catharino, D. Weinfurtner, S.
Steinbacher, J. Buchner, J. Mol. Biol. 2009, 391, 599 – 608.
[16] S. Krapp, Y. Mimura, R. Jefferis, R. Huber, P. Sondermann, J.
Mol. Biol. 2003, 325, 979 – 989.
[17] P. Sondermann, R. Huber, V. Oosthuizen, U. Jacob, Nature 2000,
406, 267 – 273.
Nonspecific uptake in all of the off-target organs determined
was lower than 9% IDgÀ1 at all time points (with the
exception of blood). This gave rise to high tumor/liver and
tumor/kidney ratios of 14.0 and 10.7, respectively (72 h p.i.).
For comparison, the chemically prepared analogue (with a
calculated average of four ligands per mAb) had a maximum
tumor uptake of 21.0% IDgÀ1 at 48 hours p.i. and high levels
of radioactivity in the liver (between 6.0–10.2% IDgÀ1
)
(Figure 3b and Table S11 in the Supporting Information).
Positron emission tomography of mice injected with 89Zr-
labeled chCE7agl-(1)4 correlated well with the biodistribution
data of the 67Ga-labeled analogue (Figure S5 in the Support-
ing Information). A comparative biodistribution study of
RTXdegl, enzymatically and chemically modified with CTPA
and radiolabeled with 64/67Cu, in mice bearing CD-20 positive
Ramos xenografts also revealed improved target-to-nontar-
get ratios for the enzymatically prepared conjugate
(Table S12 in the Supporting Information).
In summary, we have shown that enzymatic modification
of mAbs using BTG is site-specific and versatile (with the
Angew. Chem. Int. Ed. 2010, 49, 9995 –9997
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 9997