Cell surface clicking of antibody-recruiting polymers to metabolically azide-labeled cancer cells

Article abstract of DOI:10.1039/c9cc03379c

Triggering antibody-mediated innate immune mechanisms to kill cancer cells is an attractive therapeutic avenue. In this context, recruitment of endogenous antibodies to the cancer cell surface could be a viable alternative to the use of monoclonal antibodies. We report on antibody-recruiting polymers containing multiple antibody-binding hapten motifs and cyclooctynes that can covalently conjugate to azides introduced onto the glycocalyx of cancer cells by metabolic labeling with azido sugars.

Full text of DOI:10.1039/c9cc03379c

ChemComm
View Article Online
View Journal
COMMUNICATION
Cell surface clicking of antibody-recruiting polymers
to metabolically azide-labeled cancer cells†
Cite this:DOI: 10.1039/c9cc03379c
a
a
b
cd
Annemiek Uvyn,‡ Ruben De Coen,‡ Olivier De Wever, Kim Deswarte,
cd
a
Received 2nd May 2019,
Accepted 19th August 2019
Bart N. Lambrecht and Bruno G. De Geest
*
DOI: 10.1039/c9cc03379c
rsc.li/chemcomm
Triggering antibody-mediated innate immune mechanisms to kill cancer destruction by the innate immune system. Very recently, we demon-
cells is an attractive therapeutic avenue. In this context, recruitment strated that constructs composed of multiple hapten motifs installed
of endogenous antibodies to the cancer cell surface could be a viable onto a hydrophilic polymer backbone (termed antibody-recruiting
alternative to the use of monoclonal antibodies. We report on polymers or ARPs) largely outperform monovalent antibody-
5
antibody-recruiting polymers containing multiple antibody-binding recruiting constructs comprised of only a single hapten motif.
hapten motifs and cyclooctynes that can covalently conjugate to
The way antibody-recruiting hapten motifs are typically installed
azides introduced onto the glycocalyx of cancer cells by metabolic onto a cancer cell surface is either by non-specific hydrophobic
6
labeling with azido sugars.
anchoring of a lipid tail to the phospholipid cell membrane, by
7
ligand-receptor mediated recognition or by a combination of the
8
The immune system has evolved to provide several lines of defense latter with covalent conjugation. Both the latter processes are
against invading pathogens. When mucosal or epithelial barriers dynamic. Indeed, lipid motifs are known to be mobile on cell
are crossed, innate immune mechanisms comprise the second line surfaces and can shuttle between cells and bind to hydrophobic
of defense. Amongst others, these rely on adsorption of antibodies pockets of serum proteins, such as albumin, and be carried away
onto a target pathogen surface, thereby triggering complement by the bloodstream. Ligand-receptor binding is characterized by
1
9
activation, direct killing by NK cells and phagocytosis by macro- avidity, meaning that the residence time on a cell surface is limited
phages. The latter two mechanisms directly depend on recognition under highly diluted conditions, especially when using monovalent
of the antibody Fc-domains by Fc receptors on the cell surface of ligands with an avidity that is typically orders of magnitude lower
2
NK cells and macrophages. The source of antibodies that bind to a than monoclonal antibodies.
pathogen surface originates from immunization campaigns,
Hence, covalent ligation to the cell surface could be a viable
acquired immunity due to prior infection by the same pathogens, approach to cope with the above-listed challenges. However,
or by external administration of antibodies as passive immune- none of the canonical amino acids, nor carbohydrate motifs,
therapy. Alternatively, one could exploit the virtually unlimited pool exhibits orthogonal reactivity to achieve conjugation to a cell
of endogenous antibodies against hapten motifs such as dinitro- surface only, without random conjugation to abundantly pre-
phenol, rhamnose or a-Gal that are found population-wide in sent proteins in the extracellular environment.
human blood due to pesticide exposure, commensal bacteria in
the gut or meat consumption.
In this regard, metabolic labeling of cells with azido sugars is a
powerful strategy to introduce azide motifs as ‘unnatural orthogonal
3,4
In the context of cancer immune-therapy, strategies are developed handles’ into the glycocalyx that can selectively be conjugated by
to efficiently recruit such endogenous antibodies from the blood cyclooctynes through strain-promoted azide–alkyne cycloaddition
10
stream to the surface of a target cancer cell to mark this cell for (SPAAC), forming a stable triazole bond (Scheme 1).
In this paper we aim to explore in the context of ARPs, the
a
introduction of an unnatural cell surface receptor through metabolic
labeling of the glycocalyx with azido sugars. As a proof-of-concept we
focus on azido sugars that show no selectivity towards cancer cells.
But interestingly, novel glycobiology developments by the Cheng
group, have led to the development of a next generation of azido
sugars that allow for selective metabolic labeling of cancer cells that
Department of Pharmaceutics, Ghent University, Ghent, Belgium.
E-mail: br.degeest@ugent.be
b
c
Department of Human Structure and Repair, Ghent University, Ghent, Belgium
Laboratory of Mucosal Immunology and Immunoregulation, VIB Center for
Inflammation Research, Ghent, Belgium
d
Department of Internal Medicine and Pediatrics, Ghent University, Ghent, Belgium
Electronic supplementary information (ESI) available: Experimental details and
†
11
over-express specific enzymes. This would allow one to introduce
an unnatural cell surface receptor by systemic administration of an
additional data. See DOI: 10.1039/c9cc03379c
Both authors contributed equally.
‡
This journal is ©The Royal Society of Chemistry 2019
Chem. Commun.

View Article Online
Communication
ChemComm
biotin-PABTC. The synthesis route is depicted in Scheme 2A and
the polymer properties are summarized in Table 1. Experimental
characterization data are listed in the ESI,† section.
Subsequently, biotin-polyPFPA was substituted with an
amine-functionalized cycloalkyne, targeting a degree of substitution
(
5
DS; number substituted PFP-ester repeating units per 100) of 2 and
. Among different types of cycloalkynes, dibenzocyclooctyne (DBCO)
was chosen based on its optimal balance between reactivity towards
15
azides and inertness towards other groups, in particular nucleo-
philes. Note that we did not attempt to increase the DS further due
to solubility issues given the hydrophobicity of the DBCO moiety. To
generate a hydrophilic polymer backbone, all unreacted PFP esters
were converted into hydrophilic repeating units by treatment of the
polymer with an excess of 2-aminoethanol.
Scheme 1 Antibody recruiting polymer (ARP) concept. First, the glyco-
calyx of a target cancer cell is engineered with azides by metabolic labeling
with an azido sugar. In a second step, ARPs are conjugated to the cell
surface by SPAAC between azides in the glycocalyx and cycloalkyne motifs
on the ARP backbone. Cell surface conjugation of ARPs triggers recruit-
Prior to testing the conjugation of the DBCO-polymers to
ment of endogenous antibodies to the cell, thereby flagging the cell for cells, we first determined an optimal experimental window for
destruction by innate immune effector mechanisms.
metabolic azide labeling and SPAAC conjugation. Hereto, Jurkat T
cells (a model human lymphoma cell line) were cultured for 2 days
azido sugar followed by systemic administration of a cycloalkyne- with different concentrations of the azido sugar N-azidoacetyl-
containing ARP.
mannosamine (Ac ManN ), followed by staining with DBCO-EG -
4 3 4
We designed ARPs based on an activated ester polymer scaffold Fluor 545, a red fluorescent DBCO conjugate. Confocal microscopy
that could be subsequently functionalized with SPAAC-reactive and flow cytometry (Fig. S20, ESI†) revealed a dose-dependent effect
and antibody-recruiting motifs. We reasoned that, contrary to the of both the azido sugar and the fluorescent DBCO conjugate.
13
use of polymer with a SPAAC-reactive end-group, multiple of such
From these data we concluded that a 25 mM concentration of
motifs along a polymer backbone could enhance the binding azido sugar is sufficient to achieve good cell surface labeling
efficiency to metabolically azide labeled cells. Hereto, pentafluoro- with excellent availability/reactivity towards SPAAC conjuga-
12
phenyl acrylate (PFPA) was polymerized by reversible addition– tion. Importantly, confocal microscopy (Fig. S21, ESI†) proved
14
fragmentation chain transfer (RAFT), using 2-(butylthiocarbo- a homogenous density of the fluorescence over the cell surface
nothioylthio)propanoic acid (PABTC) as chain transfer agent and absence of non-conjugated cells (cf. overlay between
(CTA). A degree of polymerization (DP) of 100 was targeted, fluorescence and transmitted light (TL) channel), which was
followed by removal of the trithiocarbonate RAFT end-group by also confirmed by flow cytometry (Fig. S21, ESI†). These data
treatment with an excess of 4,4-azobis(4-cyanovaleric acid) (ACVA). show that non-specific binding of dye to cells that were not
For detection purpose, a biotin-functionalized (CTA) was used, i.e. metabolically labeled with azido sugars is negatable. By contrast,
Scheme 2 (A) Synthesis of DBCO-modified polymers by RAFT polymerization followed by trithiocarbonate end-group removal and post-
polymerization modification with functional amines. For immobilization or detection purpose, a biotin-functionalized RAFT chain transfer agent was
used. Here m equals 0,2 or 5 while n equals 98, 96 or 93. (B) Synthesis of ARPs with PABTC chain transfer agent composed of DNP and DBCO. Control
polymers were synthesized in identical way, but by omitting one or both of the functional amines. Fluorescent labeling was done using rhodamine-
cadaverine (represented by a red sphere). Here a and c equal 0 or 5, d equals 1. Therefore b equals 74, 79 or 84.
Chem. Commun.
This journal is ©The Royal Society of Chemistry 2019

View Article Online
ChemComm
Communication
Table 1 Characterization of poly(PFPA) synthesized by RAFT polymeriza- DBCO-polymers with a DS of 2, at equal dose of polymer. Confocal
tion with (biotin-PABTC) and without biotin-functionalized (PABTC) chain
microscopy confirmed (Fig. 1B) the flow cytometry data, showing
selective binding of DBCO-polymers onto the surface of azido sugar
labeled cells.
Taking into account our observations that DS 5 DBCO-polymers
exert selective cell surface binding to azide-labeled cells, we
transfer agent
a
NMR b
SEC c
SEC c
Conv.
[%]
M
n
M
n
M
w
b
d
CTA
DP
DP
[kDa]
[kDa]
[kDa]
Ð
Biotin-PABTC 110 89
98
85
23.9
20.5
11.8
12.2
15.7
14.9
1.33
PABTC
100 85
1.22 designed ARPs based on a DP 100 polyPFPA backbone substituted
a
19
b
with a DS 5 of DBCO and a DS 5 of DNP. Recently, we have
Conversion determined by F-NMR (282 MHz; CDCl
3
).
M
n
: number
average molecular weight, based on conversion measured by NMR. demonstrated that a polymer backbone containing 5 DNP units per
c
d
Determined by SEC (THF), M
dispersity.
w
: weight average molecular weight. Ð =
1
00 monomeric repeating units confers an almost 4 log increase in
5
binding avidity to anti-DNP antibodies. This combination also
ensures good water-solubility of the resulting polymers as an
azide-labeled cells exhibit a full decade monomodal shift in elevated DNP/DBCO content also increases the overall hydro-
fluorescence intensity. phobicity. Note that in the context of this work, we elaborated on
Next, we investigated the effect of the DBCO DS on the efficiency anti-DNP as a source of endogenous antibodies due to the com-
of cell surface conjugation. Hereto, cells were cultured in presence mercial availability of (fluorescently labeled) anti-DNP antibodies,
of 25 mM of azido sugar, or in absence of azido sugar as control, which strongly facilitates experimental readout compared to the
followed by pulsing with DBCO-polymers and counterstaining with use of full human serum as a source of endogenous antibodies.
AF488-labeled streptavidin. Flow cytometry (Fig. 1A) showed a
A DP 100 polyPFPA polymer was again synthesized by RAFT
highly specific binding of DBCO-polymers to azido sugar labeled polymerization using PABTC as CTA. Following trithiocarbo-
cells with an extent of non-specific binding equal to the back- nate end-group removal, the polymer was fluorescently labeled
ground fluorescence of the cells. Moreover, DBCO-polymers with with tetramethylrhodamine-cadaverine (DS 1). The polymer was
a DS of 5 showed greatly enhanced cellular conjugation than then divided in 4 equal parts and functionalized with DNP and/
or DBCO, prior to addition of an excess of 2-ethanolamine. A
control polymer without DNP and DBCO was also prepared.
The synthesis of these polymers is depicted in Scheme 2B. The
reason that we did not use a biotinylated CTA for these
experiments was to avoid steric interference between streptavi-
din binding and anti-DNP recruitment to polymer-treated cells.
The polymer properties are summarized in Table 1, while
experimental characterization is shown in the ESI,† section.
To test the in vitro ability of these polymers to recruit
antibodies to the surface of metabolically azide-labeled cells,
cells were cultured in the presence of azido sugar and pulsed
with different concentrations of the polymers. We first con-
firmed (Fig. S22, ESI†) that the presence of DNP on the polymer
backbone does not impair the ability of the polymers to bind
through SPAAC to azido sugar labeled cells. To investigate the
antibody-recruiting properties of these polymers, polymer trea-
ted cells were washed to remove unbound polymer, followed by
pulsing with AF488-labeled anti-DNP antibodies and analysis
by flow cytometry. Fig. 2A unambiguously demonstrates that
only polymers that contain both DBCO and DNP are capable to
recruit antibodies to the cell surface. Importantly, control cells
that were not cultured with azido sugar, did neither bind
polymer nor antibody. These findings indicate high selectivity
of DBCO/DNP-polymers towards azides on the cell surface and
low non-specific binding. Confocal microscopy (Fig. 2B) pro-
vides further proof of the presence of AF488-labeled anti-DNP
on the cell surface of azide-labeled cells treated with DBCO/
DNP-polymers but not in any of the controls.
To test whether this approach also holds potential for anti-
body recruitment in a context of solid tumors, we made use of
tumor spheroids that are 3D spherical structures that represent
an in vitro mimicry of a solid tumor. As cancer cell line we made
Fig. 1 (A) Flow cytometry analysis (n = 3) and (B) confocal microscopy of
Jurkat T cells, with or without metabolic labeling with the azido sugar
Ac
biotin end-group). Cells were counterstained with streptavidin-AF488
green fluorescence). Scale bar represents 10 micron.
4 3
ManN , followed by SPAAC conjugation to DBCO-polymers (with
(
use of mouse 4T1 cells which is a triple negative breast cancer
This journal is ©The Royal Society of Chemistry 2019
Chem. Commun.

View Article Online
Communication
ChemComm
spheroids surface, again with high selectivity for azide-labeled
spheroids.
One could argue on the limited penetration depth of the
antibodies. However, the ARP approach relies on the use of
endogenous antibodies that are thus continuously replenished
from the blood stream which can be considered as an unlimited
source of antibody, compared to administration of a single dose
of monoclonal antibody. Hence, strong binding of antibody
to the surface of ARP-conjugated tumors is expected to drive
potent innate immune effector responses that can then further
propagate deeper into the tumor.
In summary, we have demonstrated the ability of polymers
containing multiple DBCO moieties as target cell binding motif
and DNP moieties as antibody recruiting motif. The feasibility
of this approach was demonstrated to be highly efficient on 2D
and 3D cell cultures. In our current endeavors we are testing
this approach for in vivo targeting using cancer cell selective
azido sugars and elucidating the innate effector mechanisms
involved in cell killing.
This project has received funding from the European
Research Council (ERC) under the European Union’s Horizon
2
020 research and innovation programme (grant agreement
Fig. 2 (A) Flow cytometry analysis of anti-DNP antibody recruitment to
azido sugar labeled cells pulsed with polymers. For both cases, non-
labeled cells were used as control. (n = 3) (B) Corresponding confocal
microscopy images. Red fluorescence (polymer) and green fluorescence
No. 817938).
Conflicts of interest
(AF488-anti-DNP) are shown separately together with the overlay of both
fluorescence channels and the transmitted light (TL) channel.
There are no conflicts to declare.
References
cell line and also categorized as a so-called ‘cold tumor’ which
shows in vivo low infiltration of immune cells under steady state
conditions. Hence, strategies that can alter the inflammatory state
in the tumor microenvironment, such as antibody-mediated innate
responses triggered by ARP binding to tumor cells might be of great
relevance. Spheroids were grown and cultured for 48 h with azido
sugar prior to addition of ARPs and anti-DNP antibody. Spheroids
grown in absence of azido sugar were used as control. Confocal
microscopy (maximum intensity projections are shown in Fig. 3,
z-stack in Fig. S23, ESI†) show that ARPs show good penetration
into the azide-labeled spheroids and bind to the surface of
individual cells, whereas only a minor fraction of non-specific
binding is observed for spheroids that were not treated with azido
sugar. Furthermore, massive anti-DNP binding was observed to the
1
A. Iwasaki and R. Medzhitov, Science, 2010, 327, 291–295.
2 A. Pincetic, S. Bournazos, D. J. DiLillo, J. Maamary, T. T. Wang,
R. Dahan, B.-M. Fiebiger and J. V. Ravetch, Nat. Immunol., 2014, 15,
7
07–716.
3
4
P. J. McEnaney, C. G. Parker, A. X. Zhang and D. A. Spiegel, ACS
Chem. Biol., 2012, 7, 1139–1151.
C. E. Jakobsche, C. G. Parker, R. N. Tao, M. D. Kolesnikova,
E. F. Douglass and D. A. Spiegel, ACS Chem. Biol., 2013, 8,
2404–2411.
5 A. Uvyn, R. De Coen, M. Gruijs, C. W. Tuk, M. van Egmond and
B. G. D. Geest, Angew. Chem., Int. Ed., DOI: 10.1002/anie.201905093.
R. T. C. Sheridan, J. Hudon, J. A. Hank, P. M. Sondel and
6
L. L. Kiessling, ChemBioChem, 2014, 15, 1393–1398.
7 A. F. Rullo, K. J. Fitzgerald, V. Muthusamy, M. Liu, C. Yuan,
M. Huang, M. Kim, A. E. Cho and D. A. Spiegel, Angew. Chem., Int.
Ed., 2016, 55, 3642–3646.
8
S. Li, B. Yu, J. Wang, Y. Zheng, H. Zhang, M. J. Walker, Z. Yuan,
H. Zhu, J. Zhang, P. G. Wang and B. Wang, ACS Chem. Biol., 2018, 13,
1
686–1694.
9
T. Fritz, M. Voigt, M. Worm, I. Negwer, S. S. M u¨ ller, K. Kettenbach,
T. L. Ross, F. Roesch, K. Koynov, H. Frey and M. Helm, Chem. – Eur.
J., 2016, 22, 11578–11582.
1
0 N. J. Agard, J. A. Prescher and C. R. Bertozzi, J. Am. Chem. Soc., 2004,
26, 15046–15047.
1
1
1 H. Wang, R. Wang, K. Cai, H. He, Y. Liu, J. Yen, Z. Wang, M. Xu,
Y. Sun, X. Zhou, Q. Yin, L. Tang, I. T. Dobrucki, L. W. Dobrucki,
E. J. Chaney, S. A. Boppart, T. M. Fan, S. Lezmi, X. Chen, L. Yin and
J. Cheng, Nat. Chem. Biol., 2017, 13, 415–424.
1
1
2 A. Das and P. Theato, Chem. Rev., 2016, 116, 1434–1495.
3 R. F. Thomas, B. Martyn, T. L. Bailey and M. I. Gibson, ACS Macro
Lett., 2018, 7, 1289–1294.
Fig. 3 Confocal microscopy images (maximum intensity projections) of
1
4 C. Boyer, V. Bulmus, T. P. Davis, V. Ladmiral, J. Liu and S. Perrier,
Chem. Rev., 2009, 109, 5402–5436.
4
T1 mouse breast cancer spheroids, cultured with (A) or without (B) azido
sugar and treated with DCBO/DNP-polymer (red fluorescence) and 15 J. Dommerholt, F. P. F. T. Rutjes and F. L. van Delft, Top. Curr.
AF488-anti-DNP (green fluorescence). Scale bar represents 50 micron.
Chem., 2016, 374, 16.
Chem. Commun.
This journal is ©The Royal Society of Chemistry 2019