A phenotypic approach to probing cellular outcomes using heterobivalent constructs

Article abstract of DOI:10.1039/c9cc09595k

Various conjugation techniques are used to affect the intracellular delivery of bioactive small molecules. However, the ability to track changes in the phenotype when applying these tools remains poorly studied. We addressed this issue by having prepared a focused library of heterobivalent constructs based on Rho kinase inhibitor HA-100. By comparing the induction of the phenotype of interest, cell viability and cellular uptake, we demonstrate that various conjugates indeed lead to divergent cellular outcomes.

Full text of DOI:10.1039/c9cc09595k

ChemComm
View Article Online
View Journal
COMMUNICATION
A phenotypic approach to probing cellular
outcomes using heterobivalent constructs†
Cite this:DOI: 10.1039/c9cc09595k
Rohit Bhadoria, Kefeng Ping, Christer Lohk,‡ Ivar Jarving and Pavel Starkov
*
¨
Received 10th December 2019,
Accepted 10th March 2020
DOI: 10.1039/c9cc09595k
rsc.li/chemcomm
Various conjugation techniques are used to affect the intracellular changes in cell and organelle morphology, cytoskeletal rearrange-
delivery of bioactive small molecules. However, the ability to track ments or defects in cell division (Fig. 1).15
changes in the phenotype when applying these tools remains
We address the question of whether several known chemical
poorly studied. We addressed this issue by having prepared a modifiers of cellular uptake (the end-groups) are only respon-
focused library of heterobivalent constructs based on Rho kinase sible for getting small-molecule-cargo inside the cell or whether
inhibitor HA-100. By comparing the induction of the phenotype of they may also influence the overall cell-biological outcome.
interest, cell viability and cellular uptake, we demonstrate that
various conjugates indeed lead to divergent cellular outcomes.
Herein, we establish a chemically straightforward platform to
evaluate a focused library of heterobivalent compounds by
equipping small molecules, which induce well-characterised
Cell permeability is one of the biggest challenges faced when phenotypes, with different end-groups. In this work, we employ
developing multifunctional and high-molecular-weight chemical a polyethyleneglycol (PEG) linker 4,7,10-trioxa-1,13-tridecanedi-
probes.1–8 This can be improved by installing moieties that amine to ensure that the biomolecular interactions between the
help the constructs internalize, including cell-penetrating
peptides4,5 and chemical modifications by fatty acids, choles-
terol, asparagusic and lipoic acids.7,8 While some of the cellular
uptake-enhancing end-groups have been shown to target
treatment-resistant cancer cells and bacteria more effectively,9,10
it remains poorly understood whether and how such end-groups
may alter the cell-biological outcome at the phenotypic level.
Cellular uptake is often assessed by using constructs that
incorporate a fluorescent probe, which also serves as a cargo.11–14
This strategy has proven to be highly valuable for identification and
studying the mechanisms of cellular uptake.9–14 However, it does
not unambiguously demonstrate whether increased cell permeabil-
ity directly translates into sustained intracellular activity. Instead of
employing fluorescently labelled compounds as cargoes, it may,
perhaps, be more insightful to visualise cellular outcomes by cell
staining techniques. This approach relies on comparing holistic
changes in acquired vs. the default phenotype, for instance,
Department of Chemistry & Biotechnology, Tallinn University of Technology,
12618 Tallinn, Estonia. E-mail: pavel.starkov@taltech.ee
† Electronic supplementary information (ESI) available: Cell culture methods and
materials, cell imaging and quantification, cell viability and kinase assays,
Fig. 1 (A) Functional cellular uptake based on non-fluorescent hetero-
bivalent conjugates and changes in induced phenotypes visualized by cell
staining. (B) Design of bifunctional probes that are supplemented by
positive and negative controls. Scale bars, 20 mm.
synthetic procedures and characterisation of compounds. See DOI: 10.1039/
c9cc09595k
‡ Current address: Master’s Program in Biomedical Engineering, University
of Paris, 75006 Paris, France.
Chem. Commun.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Communication
ChemComm
Fig. 2 Heterobivalent Rho kinase inhibition probes. (A) Small molecule
Rho kinase inhibitors Y-27632, ripasudil and GSK269962A. (B) Hetero-
bivalent constructs derived from HA-100 used in this study. (C) ROCK2
kinase activity inhibition by Y-27632 (50 mM) and HA-100 (100 mM) and
HA-100 heterobivalent probes 2c–2l (100 mM). One-way ANOVA followed
by Bonferroni post-hoc test: ***p o 0.001 vs. control, (mean + S.D.; n = 3).
‘warhead’ and the target remain undisturbed even in the close
proximity of other biologically functional moieties.16–18
First, we turned our attention to Rho-associated coiled-coil-
containing protein kinase (ROCK) inhibitors. In mammalian
cells, there are two ROCK isoforms (ROCK1 and ROCK2), which
are instrumental in a myriad of cellular processes and dysre-
gulation of which is implicated in a wide spectrum of diseases.19
To date, several classes of ROCK inhibitors have been developed.20
Their scaffolds are chemically diverse (Fig. 2A), and we have
focused our efforts on modifying HA-100 (1) because of the ease
of chemical modification (i.e. linker extension and end-group
attachment) (Fig. 2B).21
Fig. 3 Different heterobivalent constructs lead to divergent phenotypes.
U2OS cells were treated with compounds for
5 h and stained for
F-actin (A) and a-tubulin (B). Green fluorescence (FITC, lex = 488 nm, lem
=
498–551 nm), red fluorescence (TRITC, lex = 561 nm, lem = 569–622 nm).
(C) Quantification of the number of cells with altered organization of F-actin
(left) and microtubules (right) for selected HA-100 constructs. (mean + S.D.;
n = 3). For cell viability data, see Fig. S1 and Tables S1 and S2, ESI.† Scale
bars, 20 mm.
Compounds 2c and 2d, with C11:0 and C15:0 end-groups,
The chemical extension of HA-100 was carried out by Michael respectively, did not induce the desired phenotype even at
addition of 1 to ethyl acrylate, hydrolysis of the resulting ester prolonged treatment times and were toxic at higher concentra-
and subsequent direct carboxamidation with a monoprotected tions (Tables S1 and S2, ESI†). Palmitic (C16:0; 2e) and stearic
4,7,10-trioxa-1,13-tridecanediamine to give linker-extended Boc- acid (C18:0; 2f) constructs did induce the phenotype of interest,
protected precursor 2b. It was then N-deprotected to give free albeit after rounds of optimisations (time vs. concentration).
amine 2a, which was coupled with corresponding unprotected The initial treatment over 14 h with 20 mM of 2e and 2f did not
carboxylic acids (i.e. fatty and bile acids, lipoic acid and biotin) to lead to dissolution of stress fibres, whereas at 30 mM both of the
give a library of compounds (2c–2l).22 All compounds were constructs resulted in cell death (no cells observed). Once the
subjected to the Rho kinase assay to confirm they inhibited duration of treatment was reduced to 5 h, we were able to
ROCK2 kinase activity (Fig. 2C).
observe the phenotype of interest for both compounds (2e and
The final constructs (2c–2l) were then assessed at different 2f at 30 mM). However, high cytotoxicity of these constructs has
concentrations and time durations in human osteosarcoma encouraged us to look for the alternatives (for cell viability data,
(U2OS) cell line (Fig. 3A). The inhibition of Rho kinase is known see Fig. S1 and Tables S1 and S2, ESI†).
to lead to a pronounced reduction in stress fibres – the bundles
Under standard U2OS cell culture conditions (DMEM, 10%
of filamentous actin (F-actin) – along with the characteristic FBS; 1% P/S), lipoic acid11-conjugate 2g was found ineffective
changes in cell shape.23 If linker-extended constructs 2c–2l were (5 h; 200 mM). The biotin construct (2h), a conjugation tool
both cell permeable and functional, we would expect to observe often used in biochemical follow-up studies,24 also proved
this default phenotype. Several well-established Rho kinase inactive (5 h; 200 mM).
inhibitors (Y-27632, GSK269962A, and HA-100) were used as
positive controls.
We have also profiled a selection of bile acid conjugates
(2i–2l). Bile acids belong to a class of cholesterol derivatives
Chem. Commun.
This journal is © The Royal Society of Chemistry 2020

View Article Online
ChemComm
Communication
known to facilitate digestion and absorption of lipids in
vertebrates.25,26 Although conjugation to bile acids is often
overlooked, they may hold the potential to improve cellular
uptake. First, bile acid conjugation is primarily used to target
and increase uptake of the constructs in the liver, biliary tract
and the intestines cell lines.27 Second, bile acids are chemically
diverse and can be modified to incorporate additional labelling
tags. Third, albeit they are structurally quite similar, they are
known to lead to divergent cell biological effects.28
Out of the four profiled bile acids, derivatives of deoxycholic
(DCA; 2i), chenodeoxycholic (CDCA; 2j) and ursodeoxycholic
acids (UDCA; 2l, albeit at higher concentration) brought about
the desired phenotypic changes, whereas the lithocholic acid
conjugate did not (LCA; 2k; 5 h; tested up to 200 mM). We also
confirmed that the blank bile acid–PEG constructs did not lead
to reduction in stress fibres on their own (Fig. S3, ESI†).
To quantify the effects observed, we counted the number of
cells that have acquired the desired phenotype at different
concentrations of HA-100 and its conjugates with C16:0 (2e),
DCA (2i), and CDCA (2j), and Y-27632, the most used Rho
inhibitor (Fig. 3C). Compounds with the DCA and CDCA end-
groups had higher activity than HA-100. The effect of palmitic
acid construct 2e could not be quantified beyond 35 mM
as it was cytotoxic and no cells were observed on coverslips
(cf. Tables S1 and S2, ESI†).
Importantly, we have observed a secondary effect on the
organisation of microtubules (Fig. 3C). DCA and CDCA con-
jugates 2i and 2j both induced bundling of microtubules,
whereas compound 2i induced abrupt change in the number
of affected cells when its concentration was increased from
25 mM to 30 mM. Although we cannot rule out synergistic effects
of our bifunctional constructs, it is known that actin cytoskeletal
rearrangements have a crosstalk effect on microtubules.29
Next, we wished to test whether these modifications would
also work with another kinase inhibitor (Fig. 4). We looked into
direct inhibition of myosin light chain kinase (MLCK/MYLK),
which phosphorylates myosin light chain (MLC).30 We modified
inhibitor ML-7 (4),31 which is structurally akin to HA-100 (1).
Treating the cells with 4 is known to have a pronounced effect
Fig. 4 Effects of myosin light chain kinase inhibitor ML-7-based con-
structs. (A) Chemical identity of ML-7 constructs. (B) Comparison of cells
treated by either ROCK or MYLK inhibitors. U2OS cells were treated with
compounds for 5 h and stained for F-actin (colourless) and p-MLC (green).
Green fluorescence (Alexa 488, lex = 488 nm, lem = 498–551 nm), red
fluorescence (TRITC, lex = 561 nm, lem = 569–622 nm). (C) Quantification
of the number of cells with altered organization of F-actin (left) and
microtubules (right) by ML-7 constructs (mean + S.D.; n = 3). Scale bars,
20 mm.
on F-actin/p-MLC colocalization and cell shape.32,33 Indeed, toxic than the parent compound or other conjugates (Tables S1
the parent compound (4) and its conjugates ML-7-C16:0 (5b) and S2, ESI†).
ML-7-DCA (5c) and ML-7-CDCA (5d) have led to similar changes
With several constructs demonstrating that conjugation to bile
in F-actin (i.e. dissolution of stress fibres) and in intracellular and fatty acids results in different outcomes, we wished to see to
distribution of phospho-MLC2 (Ser19) (p-MLC).34 We quanti- what extent the end-groups affect cell permeability. We prepared
fied the effects of ML-7 and its derivatives on the organisation fluorescent probes based on 7-dimethylaminocoumarin-4-acetic
of F-actin and microtubules (Fig. 4C). While the C16:0 construct acid (DMACA; Fig. 5). Both benzylamide of DMACA 6a and linker-
(5b) performed similarly to the parent compound, the DCA (5c) extended N-Boc protected construct 6b did not internalise, while
and CDCA (5d) conjugates led to reduction of stress fibres at C16:0, DCA and CDCA conjugates (6c 4 6d 4 6e) did. For
lower concentration. However, both 5c and 5d also induced compounds 6d and 6e, the efficiency of cellular uptake correlated
microtubule bundling (Fig. S4, ESI†). Notably, ML-7 and its with the onset of changes in phenotypes (Fig. 3C and 4C).
derivatives do not lead to the reduction of p-MLC in the cell
Having screened the two focused libraries of heterobifunc-
periphery, which indicates that the effect on actomyosin is tional probes using well-established ROCK and MYLK inhibitors,
achieved through inhibition of MLCK and not ROCK (cf. Rho we observe that there are complex relationships between the drug
kinase inhibitor GSK269962A vs. ML-7; arrows in Fig. 4B cores (warheads), the end-groups and the final heterobifunctional
and Fig. S2, ESI†).35 Interestingly, unlike the case of Rho constructs themselves. In addition to inducing the desired pheno-
kinase inhibitor HA-100, the C16:0 construct of ML-7 is less type, we show that such combinations may also lead to the
Chem. Commun.
This journal is © The Royal Society of Chemistry 2020

View Article Online
Communication
ChemComm
R. A. Copeland, B. Cravatt, J. L. Dahlin, D. Dhanak, A. M. Edwards,
M. Frederiksen, S. V. Frye, N. Gray, C. E. Grimshaw, D. Hepworth,
T. Howe, K. V. M. Huber, J. Jin, S. Knapp, J. D. Kotz, R. G. Kruger,
D. Lowe, M. M. Mader, M. Marsden, A. Mueller-Fahrnow, S. Mu¨ller,
R. C. O’Hagan, J. P. Overington, D. R. Owen, S. H. Rosenberg, R. Ross,
¨
B. Roth, M. Schapira, S. L. Schreiber, B. Shoichet, M. Sundstrom,
G. Superti-Furga, J. Taunton, L. Toledo-Sherman, C. Walpole,
M. A. Walters, T. M. Willson, P. Workman, R. N. Young and
W. J. Zuercher, Nat. Chem. Biol., 2015, 11, 536.
2 N. J. Yang and M. J. Hinner, Methods Mol. Biol., 2015, 1266, 29.
3 S. D. Conner and S. L. Schmid, Nature, 2003, 422, 37.
4 K. M. Stewart, K. L. Horton and S. O. Kelley, Org. Biomol. Chem.,
2008, 6, 2242.
5 G. Guidotti, L. Brambilla and D. Rossi, Trends Pharmacol. Sci., 2017,
38, 406.
´
´
6 J. Mosquera, I. Garcıa and L. M. Liz-Marzan, Acc. Chem. Res., 2018,
51, 2305.
7 G. Gasparini, E.-K. Bang, J. Montenegro and S. Matile, Chem.
Commun., 2015, 51, 10389.
8 M. Avadisian and P. T. Gunning, Mol. BioSyst., 2013, 9, 2179.
9 C. J. McKinlay, R. M. Waymouth and P. A. Wender, J. Am. Chem. Soc.,
2016, 138, 3510.
10 A. Antonoplis, X. Zang, M. A. Huttner, K. K. L. Chong, Y. B. Lee,
J. Y. Co, M. R. Amieva, K. A. Kline, P. A. Wender and L. Cegelski,
J. Am. Chem. Soc., 2018, 140, 16140.
11 G. Gasparini, G. Sargsyan, E.-K. Bang, N. Sakai and S. Matile, Angew.
Chem., Int. Ed., 2015, 54, 7328.
12 A. Nagel, P. S. Raschle and H. Wennemers, Angew. Chem., Int. Ed.,
2017, 56, 122.
13 G. H. Bird, E. Mazzola, K. Opoku-Nsiah, M. A. Lammert, M. Godes,
D. S. Neuberg and L. D. Walensky, Nat. Chem. Biol., 2016, 12, 845.
14 D. Zhang, L. Wei, M. Zhong, L. Xiao, H.-W. Li and J. Wang, Chem.
Sci., 2018, 9, 5260.
Fig. 5 Qualitative and quantitative comparison of cellular uptake of
heterobivalent constructs. (A) Chemical identity of DMACA constructs
tested. (B) Confocal images for DMACA constructs at 30 mM (DIC and
l
ex = 405 nm, lem = 435–478 nm). (C) Mean cellular fluorescence of cells
treated with DMACA constructs (mean + S.D.; n = 3). U2OS cells were
treated with compounds, after 1 h medium was exchanged and the cells
were imaged live on confocal microscope. Scale bars, 20 mm.
15 U. S. Eggert, Nat. Chem. Biol., 2013, 9, 206.
16 B. Z. Stanton, E. J. Chory and G. R. Crabtree, Science, 2018,
359, eaao5902.
17 T. W. Corson, N. Aberle and C. M. Crews, ACS Chem. Biol., 2008,
3, 677.
18 C. Borsari, D. J. Trader, A. Tait and M. P. Costi, J. Med. Chem., 2020,
63, DOI: 10.1021/acs.jmedchem.9b01456.
19 K. Riento and A. J. Ridley, Nat. Rev. Mol. Cell Biol., 2003, 4, 446.
20 Y. Feng, P. V. LoGrasso, O. Defert and R. Li, J. Med. Chem., 2016,
59, 2269.
21 S. P. Davies, H. Reddy, M. Caivano and P. Cohen, Biochem. J., 2000,
351, 95.
22 R. M. Lanigan, V. Karaluka, M. T. Sabatini, P. Starkov, M. Badland,
L. Boulton and T. D. Sheppard, Chem. Commun., 2016, 52, 8846.
23 S. Pellegrin and H. Mellor, J. Cell Sci., 2007, 120, 3491.
24 B. Lomenick, R. W. Olsen and J. Huang, ACS Chem. Biol., 2011, 6, 34.
25 F. G. Schaap, M. Trauner and P. L. Jansen, Nat. Rev. Gastroenterol.
Hepatol., 2014, 11, 55.
26 S. Zhuang, Q. Li, L. Cai, C. Wang and X. Lei, ACS Cent. Sci., 2017,
3, 501.
27 C. Faustino, C. Serafim, P. Rijo and C. Pinto Reis, Expert Opin. Drug
Delivery, 2016, 13, 1133.
28 J. D. Martinez, E. D. Stratagoules, J. M. LaRue, A. A. Powell,
P. R. Gause, M. T. Craven, C. M. Payne, M. B. Powell,
E. W. Gerner and D. L. Earnest, Nutr. Cancer, 1998, 31, 111.
29 A. Takesono, S. J. Heasman, B. Wojciak-Stothard, R. Garg and
A. J. Ridley, PLoS One, 2010, 5, e8774.
30 F. Hong, B. D. Haldeman, D. Jackson, M. Carter, J. E. Baker and
C. R. Cremo, Arch. Biochem. Biophys., 2011, 510, 135.
31 M. Saitoh, T. Ishikawa, S. Matsushima, M. Naka and H. Hidaka,
J. Biol. Chem., 1987, 262, 7796.
secondary effects (induction of microtubule bundling) and the
variations in cytotoxicity and cellular uptake. In order to quantita-
tively predict the structure–activity relationships and to better
understand the underlying molecular mechanisms, a larger sub-
set of modified small molecules, including allosteric inhibitors,
and modifiers (the end-groups) would be needed.
In conclusion, we have developed a confocal microscopy-
based assay for the ‘in-cell’ screening of heterobivalent con-
structs. This approach can be used to distinguish between
various constructs, even if their chemical alterations are very
minute. We have shown that both the end-groups and the
warheads contribute to the overall cellular outcome.
This research was funded by Estonian Research Council
Grant PUT1290 and TalTech Young Investigator Grant B62
(both to P. S.). Tallinn HRMS facilities are supported by
Institutional Research Funding IUT19-9 of the Estonian Minis-
try for Education & Research. We thank COST Action CA15135
MuTaLig members, Dr Pirjo Spuul (Tallinn) for helpful discus-
sions, and Mika Molin and Marko Crivaro for technical assis-
tance at the University of Helsinki Light Microscopy Unit. K. P.
acknowledges Estonian Smart Specialization PhD Fellowship.
Conflicts of interest
32 Y. Li, D. Lovett, Q. Zhang, S. Neelam, R. A. Kuchibhotla, R. Zhu,
G. G. Gundersen, T. P. Lele and R. B. Dickinson, Biophys. J., 2015,
109, 670.
There are no conflicts to declare.
33 S. B. Khatau, C. M. Hale, P. J. Stewart-Hutchinson, M. S. Patel,
C. L. Stewart, P. C. Searson, D. Hodzic and D. Wirtz, Proc. Natl. Acad.
Sci. U. S. A., 2009, 106, 19017.
Notes and references
34 K. Katoh, Y. Kano and Y. Noda, J. R. Soc., Interface, 2011, 8, 305.
35 G. Totsukawa, Y. Yamakita, S. Yamashiro, D. J. Hartshorne,
Y. Sasaki and F. Matsumura, J. Cell Biol., 2000, 154, 797.
1 C. H. Arrowsmith, J. E. Audia, C. Austin, J. Baell, J. Bennett,
J. Blagg, C. Bountra, P. E. Brennan, P. J. Brown, M. E. Bunnage,
C. Buser-Doepner, R. M. Campbell, A. J. Carter, P. Cohen,
Chem. Commun.
This journal is © The Royal Society of Chemistry 2020