An enzyme-assisted self-delivery system of
lonidamine–peptide conjugates for selectively
killing cancer cells
A self-delivery system consisting of lonidamine and a self-assembling
peptide was designed for the selective killing of phosphataseoverexpressing cancer cells, which was mediated by both enhanced
cellular uptake of LND–peptide and enzyme-triggered intracellular
fiber formation, thereby providing a generalized strategy to develop
cancer-targeting systems of drug conjugates.
Lonidamine (LND) belongs to a broad-spectrum of antineoplastic agents that acts on mitochondria and inhibits glycolysis
in cancer cells.1–3 It has been widely used for the treatment of a
variety of cancers, such as lung cancer, breast cancer and liver
cancer.4–8 However, clinical practice utilizing the drug has been
hindered because of its poor solubility and low bioavailability
that leads to inadequate accumulation in tumor tissues and
severe side effects in normal tissues (e.g. hepatic toxicity).9–11
Thus a drug delivery system (DDS) is urgently needed to deliver
the drug selectively to cancer cells.
Recently, self-assembled nanofibers that consist of peptide
derivatives have attracted considerable attention as ideal drug
delivery vehicles thanks to their high aspect ratio, flexible
structural tunability and excellent biocompatibility.12–16 To
improve the solubility of LND, we propose the conjugation of
LND to a self-assembling peptide motif to confer a LND–
peptide conjugate as a self-delivery system for LND. Compared
with traditional DDS, the self-delivery system of LND–peptide
guarantees 100% drug loading, and increases cellular uptake and
retention of drugs through a nanofiber-mediated endocytosis
process.17–19 Furthermore, enzyme-instructed self-assembly (EISA)
processes have been widely utilized to trigger the intracellular
formation of nanofibers for selectively killing cancer cells.20–24
Therefore, the integration of LND with EISA in the LND–peptide
construct would not only increase the solubility and cellular
uptake of the drug, but also enhance the selectivity and potency
of LND–peptide against cancer cells.
Herein, we design the self-delivery system of LND–peptide to
contain a sequence of lonidamine–Gly–Phe–Phe–pTyr (denoted
as LND–GFFpY hereafter): (i) LND serves as the capping group
of the peptide to increase its overall hydrophobicity with a
chemical linkage liable to protease-mediated cleavage; (ii) a
moiety of Phe–Phe–Tyr is selected to enhance self-assembling
capability of LND–peptide; and (iii) a phosphorylated Tyr (pY) is
chosen as the terminus and acts as the substrate of alkaline
phosphatase (ALP). We envisage that LND–GFFpY is dephosphorylated by cell-borne ALP to yield LND–GFFY (i.e. hydrogelator) that self-assembles to form drug-bound nanofibers for
further cellular uptake (Fig. 1A). Cancer cells with a higher
expression of ALP would show a faster conversion rate and yield
more hydrogelators compared to normal cells, leading to selective
accumulation of LND–peptide in cancer cells (Scheme 1). In the
cytoplasm, it is hypothesized that two anti-cancer pathways
converge to increase the selectivity and potency of the selfdelivery system: (1) LND–peptide is cleaved by proteases to yield
free LND for inhibition of glycolysis in mitochondria and
(2) further self-assembly of LND–peptide disturbs the cytoskeleton
Fig. 1 (A) Chemical structures and enzymatic transformation of LND–GFFpY
into LND–GFFY with ALP; (B) TEM images of 1.0 wt% LND–GFFpY (pH 7.4)
after adding 3 units per mL ALP for 2 h; and (C) TEM images of HepG2 cell
lysates with 12 h pre-treatment of 150 mM LND–GFFpY.
a Department of Chemistry, China Pharmaceutical University, Nanjing 210009,
China. E-mail: [email protected], [email protected]
b Key Laboratory of Biomedical Functional Materials, China Pharmaceutical
University, Nanjing 210009, China
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9cc06204a
COMMUNICATION
Published on 19 November 2019. Downloaded by Auckland University of Technology on 1/3/2020 4:52:49 AM. View Article Online View Journal | View Issue
This journal is © The Royal Society of Chemistry 2019 Chem. Commun., 2019, 55, 14852–14855 | 14853
(i.e. microtubules) and disrupts cellular hemostasis leading to cell
death (Scheme 1).
We synthesized LND–GFFpY, LND–GFFY and GFFpY by
standard solid phase peptide synthesis. The structures of all
compounds were verified using the mass spectrum (MS) and
1
H NMR spectrum respectively (Fig. S1–S3, ESI†). The solubility
of LND–GFFpY in aqueous solution was confirmed to be
B109 mM, which was 560 fold that of free LND (0.1947 mM).
With 3 units per mL ALP, 1.0 wt% of LND–GFFpY formed a
stable hydrogel at room temperature in PBS (pH 7.4) after 2 h
incubation (Fig. 1B, inset). Further studies revealed that the
minimum gelation concentrations (MGCs) of LND–GFFpY were
1.0 wt% and 9.0 wt% in PBS with and without ALP respectively.
Consistently, the critical micelle concentration (CMC) of LND–
GFFpY was found to decrease from 105 to 54 mM after addition
of 3 units per mL ALP (Fig. 2A). In comparison, the control
compound LND–GFFY showed a higher MGC of 2.0 wt% with a
heating-and-cooling process. And a physical mixture of LND
and GFFpY formed a hydrogel at 5.0 wt% with the same ALP.
These results suggested that enzyme-instructed self-assembly
(EISA) of LND–GFFpY occurred more efficiently than that of
LND–GFFY or a mixture of LND and GFFpY in aqueous solution.
Transmission electron microscopy (TEM) confirmed an
entangled nanofibrillar network for the LND–GFFpY/ALP hydrogel
(Fig. 1B), whereas LND–GFFY hydrogels that were formed after
heating and cooling demonstrated obvious aggregation of insoluble
peptides in TEM examinations (Fig. S4, ESI†). The rate of LND–
GFFpY dephosphorylation in EISA was further investigated using
HPLC elution curves. As shown in Fig. S5 (ESI†), the peak height of
LND–GFFYp diminished after incubation with ALP, while a new
peak appeared corresponding to the elution of LND–GFFY. After
12 h of ALP treatment, it was found that 490% of LND–GFFpY was
converted to LND–GFFY (Fig. S6, ESI†).
With the hydrogelation of LND–GFFpY/ALP confirmed
in vitro, we proceeded to examine the intracellular gelation of
LND–GFFpY with cancer cells that overexpress ALP. First, the
ALP expression levels of cancer cells HepG2 and HeLa were
visualized in confocal laser microscopy (CLMS) by utilizing a
4-nitro-2,1,3-benzoxadiazole (NBD)-capped ALP substrate NBDGFFpY that was synthesized according to previous report
(structure characterization: Fig. S7, ESI†).25 Quantitative analyses
revealed that the relative mean fluorescence intensities of NBDGFFpY-treated cells were 9.8, 7.3, 1.6, and 1.0 for HepG2, HeLa,
LO2 and NIH3T3 cells respectively (Fig. S8 and S9, ESI†), thus
confirming the overexpression of ALP in HepG2 and HeLa. Next,
intracellular gelation of LND–GFFpY with HepG2 was studied.
After 12 h incubation with LND–GFFpY, the cell pellets of HepG2
were rinsed with PBS to remove free gel precursors outside cells and
then lysed for TEM examinations, which revealed an entangled
fibrous network in the cell lysate (Fig. 1C). These results proved the
gelation capabilities of LND–GFFpY with intracellular ALP in cancer
cells.26
We further examined in situ gelation of LND–GFFYp in the
subcutaneous tissues and tumor tissues of mice. Aqueous
solution of LND and rhodamine disappeared in o3 h after injection
(Fig. S10B and C, ESI†). In contrast, rhodamine-incorporated LND–
GFFpY/ALP hydrogels were clearly visible 3 h and 12 h after injection
in both subcutaneous areas (Fig. S10E and F, ESI†) and peritumoral
regions (Fig. S10H and I, ESI†). In the absence of ALP, we also
observed gelation of LND–GFFpY in peritumoral regions at 3 h
(Fig. S10K, ESI†), which was diminished at 12 h (Fig. S10L,
ESI†), suggesting the plausible contribution of ALP from the
tumor microenvironment.
Characterization of LND–GFFpY/ALP hydrogels with circular
dichroism (CD) revealed a negative peak centered at 210 nm
and a positive peak at 195 nm (Fig. 2B), illustrating a typical b-sheet
structure for LND–peptide. Further rheological tests demonstrated
that values of G0 dominated those of G00 in dynamic time sweep
with a weak dependence on frequency (Fig. S11A and B, ESI†),
suggesting an elastic hydrogel network for the LND–peptide hydrogel. Dynamic strain sweep results confirmed the shear-thinning
properties of the hydrogel, for decreased values of G0 with
increasing strain, with a critical strain of 6.4% reached where
gel-to-sol transition began (Fig. 2C). Cyclic strain sweeps were
Scheme 1 Schematic illustration of the self-delivery system of LND–
GFFpY for the selective killing of cancer cells.
Fig. 2 Characterization of LND–GFFpY/ALP hydrogels (1.0 wt%, 3 units
per mL ALP). (A) MGC and CMC of LND–GFFpY with 3 units per mL ALP or
without ALP; (B) CD spectra; (C) dynamic strain sweep; and (D) cumulative
release of intact LND moieties from 1.0 wt% LND–GFFpY hydrogels with or
without proteinase K (0, 10, and 30 units per mL).
Communication ChemComm
Published on 19 November 2019. Downloaded by Auckland University of Technology on 1/3/2020 4:52:49 AM. View Article Online
14854 | Chem. Commun., 2019, 55, 14852–14855 This journal is © The Royal Society of Chemistry 2019
further employed to characterize the thixotropic behavior of
hydrogels. When the strain decreased from 50% to 2%, G0 and
G00 restored to initial values at 2% strain that could be repeated
for three cycles (Fig. S11C, ESI†). These properties of hydrogels
promise the further development of hydrogels as injectable
materials for drug delivery.
Then we proceeded to examine the release of intact LND
moieties from the self-delivery system LND–GFFpY/ALP. As
shown in Fig. 2D, the solution of free LND displayed burst
release of drugs, with a cumulative release of B72% in 12 h. In
contrast, less than 5% of LND moieties were released from
LND–peptide hydrogels in PBS buffer within 72 h, suggesting
the excellent stability of covalent conjugations between LND
and the peptide. To test the feasibility of drug release under
lysosomal compartments where proteases are abundant, 10 to
30 units per mL of proteinase K was employed to trigger the
release of intact drugs from hydrogels. Quantifications of drug
release by high performance liquid chromatography (HPLC)
revealed that the cumulative release of intact LND moieties was
11% and 28% in 72 h for release buffers containing 10 and 30
units per mL of enzymes respectively, confirming the release of
intact drugs from the self-delivery system.
Next, we compared the cell inhibitory efficacies of LND–
peptide in cancer cells (HepG2 and HeLa) with those of free
LND in vitro. It was found that the efficacy of LND–GFFpY
surpassed that of free LND in both cancer cells (Fig. 3A and B).
The IC50 values of LND–GFFpY were calculated to be 146.8 and
194.8 mM for HepG2 and HeLa, respectively, whereas those
values of free LND were 384.4 and 398.7 mM for the two cells
(Table S1, ESI†). In addition, we also examined the cell inhibitory
effects of two control compounds (LND–GFFY, or a mixture of
LND and GFFpY) both of which showed less efficacy in selfassembly (Fig. S12, S13, ESI†). The IC50 values of LND–GFFY were
378.6 and 376.5 mM for HepG2 and HeLa, respectively, while those
values of LND in physical mixtures of LND and GFFpY were
350.7 mM and 358.7 mM for the two cells (Table S1, ESI†). These
results suggested the indispensable role of enzyme (i.e. ALP)
instructed self-assembly (EISA) in the increase of the potency of
LND–GFFpY against cancer cells.
Since cancer cells (HepG2 and HeLa) demonstrated higher
levels of ALP expression than normal liver cells (LO2) and
fibroblast NIH3T3 (Fig. S8 and S9, ESI†),25 we further examined
the selective inhibitory efficacy of LND–GFFpY against cancer
cells over normal cells. We observed a much lower inhibition
rate of LND–GFFpY when it was incubated with normal cells
LO2 or NIH3T3 (Fig. 3C and D). Compared to HepG2, the IC50
values of LND–GFFpY were found to increase 3.1-fold in LO2
cells. Consistently, similar selective cytotoxicity was also observed
in HeLa cells (IC50 = 194.8 mM) versus NIH3T3 (IC50 = 508.8 mM)
(Table S1, ESI†). Interestingly, the control compounds (LND–
GFFY, or a mixture of LND and GFFpY) didn’t show much
selectivity towards cancer cells (Fig. S12 and S13, ESI†).
To investigate the mechanism involved in LND–GFFpYinduced selective killing of cancer cells, we first quantified the
cellular uptake of LND–GFFpY (also intracellular LND–GFFY
after dephosphorylation) and free LND with pre-established
calibration curves from HPLC (Fig. S14, ESI†). As shown in
Fig. 4A, the uptake of LND–GFFpY in HepG2 cells at 37 1C was
91.8 nmol/106 cells, which was 3.2-fold that of free LND. A
similar enhancement of the uptake of LND–GFFpY was also
found in HeLa. Interestingly, the uptake of LND–GFFpY was
much lower in normal cells (LO2: 36.4 nmol/106 cells and
NIH3T3: 33.9 nmol/106 cells) than in cancer cells, whereas the
uptake of free LND showed no preference between cancer cells
and normal cells. The enhanced uptake of LND–GFFpY in
cancer cells could be plausibly attributed to the ALP-mediated
EISA process, which continuously converts LND–GFFpY to
nanofibers thereby driving the uptake equilibrium towards the
cytosol. Further studies of cellular uptake performed at 4 1C
revealed a significant decrease of LND–GFFpY uptake at lower
temperatures (Fig. 4B), suggesting that the cellular entry of
LND–GFFpY was mediated by an energy-dependent pathway
(i.e. endocytosis). In contrast, temperature changes did not lead
to obvious changes in the uptake of free LND, suggesting a
diffusion-based mechanism for the cellular entry of free drugs.
Intracellular nanofiber formation is another factor that could
contribute to the selective cytotoxicity of LND–GFFpY to cancer cells.
Thus, we examined the morphological changes of the cytoskeleton
in LND–GFFpY-treated HepG2 cancer cells with specific staining of
actin filaments and microtubules.21 Confocal laser scanning
Fig. 3 Viabilities of HepG2 (A), HeLa (B), LO2 (C) and NIH3T3 (D) cells after
48 h treatment of LND and LND–GFFpY (37.5–600 mM).
Fig. 4 Cellular uptake of 350 mM LND–GFFpY and LND in cancer cells
(HepG2, HeLa) and normal cells (LO2 and NIH3T3) after 4 h incubation at
37 1C and 4 1C. One-way ANOVA, mean SD, ***P o 0.001.
ChemComm Communication
Published on 19 November 2019. Downloaded by Auckland University of Technology on 1/3/2020 4:52:49 AM. View Article Online
This journal is © The Royal Society of Chemistry 2019 Chem. Commun., 2019, 55, 14852–14855 | 14855
microscopy (CLSM) images demonstrated a significant decrease
in actin/microtubule staining (i.e. red fluorescence, Fig. S15G,
ESI†) in LND–GFFpY-treated HepG2 cells.
In contrast, LND treatment didn’t induce obvious changes
in actin networks (Fig. S15D, ESI†) as compared to control (i.e.
without any treatment, Fig. S15A, ESI†). These results further
illustrated that the inhibition of the cytoskeleton could be one
plausible mechanism for EISA-induced cell death.
We further studied the anti-cancer efficacies of LND–GFFpY
hydrogels in vivo by utilizing a HeLa-cell-xenografted tumor
model in mice. Peritumoral administration of LND–GFFpY
hydrogels led to significant tumor regression, whereas injection
of LND solution at the same dose failed to do so (Fig. 5A).
Histological analyses (H&E stain) further confirmed that LND–
GFFpY treatment conferred a lowest density of tumor cells in
tumor tissues compared to those of the PBS or LND group (Fig. S16,
ESI†). In addition, we didn’t observe significant changes in body
weight in mice after hydrogel treatment (Fig. 5B), suggesting an
excellent biocompatibility for the self-delivery system.
In summary, we reported the design and synthesis of a selfdelivery system of LND–GFFpY for the selective killing of cancer
cells. The LND–GFFpY conjugate displayed a 560-fold increase
in water solubility compared to free drug LND. It demonstrated
superior gelation capabilities (MGC = 1.0 wt%) through an
enzyme-instructed self-assembly (EISA) process under the catalysis
of alkaline phosphatase (ALP). Furthermore, endogenous ALPs from
cancer cells were found to trigger the intracellular gelation of
LND–GFFpY in HepG2 cells. Compared to free LND, LND–
GFFpY displayed increased potency against cancer cells both
in vitro and in vivo, with the selective killing of cancer cells over
normal cells. Cellular uptake experiments and intracellular
staining of the cytoskeleton revealed two plausible mechanisms
involved: (i) enhanced cellular uptake of LND–GFFpY in cancer
cells over normal cells and (ii) disruption of cytoskeleton networks
of HepG2 due to intracellular nanofiber formation. Through this
work, we provide a generalized strategy to construct cancer cellselective self-delivery systems that are based on EISA of peptide
derivatives. In future, a variety of hydrophobic drugs could be
utilized for conjugations with self-assembling peptides to enhance
their water solubility and cancer-targeting capabilities.
This work was supported by the ‘‘Double First-Class ‘‘University
project (No. CPU 2018GY25) and the Young Teachers’ Research
Funding from College of Science, China Pharmaceutical University
(No. 2018CSYT003).
Conflicts of interest
There are no conflicts to declare.
Notes and references
1 S. Fulda, L. Galluzzi and G. Kroemer, Nat. Rev. Drug Discovery, 2010,
9, 447–464.
2 K. Nath, L. L. Guo, B. Nancolas, D. S. Nelson, A. A. Shestov, S. C. Lee,
J. Roman, R. Zhou, D. B. Leeper, A. P. Halestrap, I. A. Blair and
J. D. Glickson, Biochim. Biophys. Acta, Rev. Cancer, 2016, 1866, 151–162.
3 G. Cheng, Q. Zhang, J. Pan, Y. Lee, O. Ouari, M. Hardy, M. Zielonka,
C. Myers, J. Zielonka, K. Weh, A. Chang, G. Chen, L. Kresty,
B. Kalyanaraman and M. You, Nat. Commun., 2019, 10, 2205.
4 M. De Lena, V. Lorusso, A. Latorre, G. Fanizza, G. Gargano,
L. Caporusso, M. Guida, A. Catino, E. Crucitta, D. Sambiasi and
A. Mazzei, Eur. J. Cancer, 2001, 37, 364–368.
5 L. Ricotti, A. Tesei, F. De Paola, C. Milandri, D. Amadori, G. L.
Frassineti, P. Ulivi and W. Zoli, J. Chemother., 2003, 15, 480–487.
6 M. Spada, M. Giampaglia, C. Milandri, D. Amadori, W. Zoli,
V. Lorusso and M. D. Lena, Ann. Oncol., 2004, 15, 75.
7 A. G. Assanhou, W. Y. Li, L. Zhang, L. J. Xue, L. Y. Kong, H. B. Sun,
R. Mo and C. Zhang, Biomaterials, 2015, 73, 284–295.
8 H. Choi, T. Liu, K. Nath, R. Zhou and I. W. Chen, Nano Today, 2017,
13, 15–22.
9 G. S. Price, R. L. Page, J. E. Riviere, J. M. Cline and D. E. Thrall,
Cancer Chemother. Pharmacol., 1996, 38, 129–135.
10 N. Li, C.-X. Zhang, X.-X. Wang, L. Zhang, X. Ma, J. Zhou, R.-J. Ju,
X.-Y. Li, W.-Y. Zhao and W.-L. Lu, Biomaterials, 2013, 34, 3366–3380.
11 S. Ganapathy-Kanniappan and J. F. H. Geschwind, Mol. Cancer,
2013, 12, 152.
12 S. Fleming and R. V. Ulijn, Chem. Soc. Rev., 2014, 43, 8150–8177.
13 L. Adler-Abramovich and E. Gazit, Chem. Soc. Rev., 2014, 43, 6881–6893.
14 X. Du, J. Zhou, J. Shi and B. Xu, Chem. Rev., 2015, 115, 13165–13307.
15 K. Sato, M. P. Hendricks, L. C. Palmer and S. I. Stupp, Chem. Soc.
Rev., 2018, 47, 7539–7551.
16 J. Gao, J. Zhan and Z. Yang, Adv. Mater., 2019, e1805798.
17 C. Liang, D. Zheng, F. Shi, T. Xu, C. Yang, J. Liu, L. Wang and
Z. Yang, Nanoscale, 2017, 9, 11987–11993.
18 Y. Cai, H. Shen, J. Zhan, M. Lin, L. Dai, C. Ren, Y. Shi, J. Liu, J. Gao
and Z. Yang, J. Am. Chem. Soc., 2017, 139, 2876–2879.
19 Y. Ji, Y. Xiao, L. Xu, J. He, C. Qian, W. Li, L. Wu, R. Chen, J. Wang,
R. Hu, X. Zhang, Z. Gu and Z. Chen, Adv. Sci., 2018, 5, 1700867.
20 J. Zhou, X. Du and B. Xu, Angew. Chem., Int. Ed., 2016, 55, 5770–5775.
21 Z. Feng, H. Wang, X. Chen and B. Xu, J. Am. Chem. Soc., 2017, 139,
15377–15384.
22 J. Zhou, X. Du, X. Chen, J. Wang, N. Zhou, D. Wu and B. Xu, J. Am.
Chem. Soc., 2018, 140, 2301–2308.
23 C. F. Wu, R. Zhang, W. Du, L. Cheng and G. L. Liang, Nano Lett.,
2018, 18, 7749–7754.
24 J. Zhan, Y. B. Cai, S. S. He, L. Wang and Z. M. Yang, Angew. Chem.,
Int. Ed., 2018, 57, 1813–1816.
25 J. Zhou, X. Du, C. Berciu, H. He, J. Shi, D. Nicastro and B. Xu, Chem,
2016, 1, 246–263.
26 Z. Feng, X. Han, H. Wang, T. Tang and B. Xu, Chem, 2019, 5,
2442–2449.
Fig. 5 (A) Changes in tumor volumes in HeLa-xenografted mice after
peritumoral administration of varying Lonidamine formulations (PBS, free LND or LND–
GFFpY). (B) Plot of average body weight of mice after drug treatment.
(One-way ANOVA, mean SD,P o0.001.)
Communication ChemComm
Published on 19 November 2019. Downloaded by Auckland University of Technology on 1/3/2020 4:52:49 AM. View Article Online