Unique Triphenylphosphonium Derivatives for Enhanced Mitochondrial Uptake and Photodynamic Therapy

Article abstract of DOI:10.1021/acs.bioconjchem.6b00682

In this study, unique methyl-functionalized derivatives (T?PP⁺) of the drug carrier triphenylphosphonium (TPP⁺) that exhibit significant enhancement of the accumulation of both the cation and its conjugated cargo in cell mitochondria are designed. We show that the presence of methyl group(s) at key positions within the phenyl ring results in an increase in the hydrophobicity and solvent accessible surface area of T?PP⁺. In particular, when the para position of the phenyl ring in T?PP⁺ is functionalized with a methyl group, the cation is most exposed to the surrounding environment, leading to a large decrease in water entropy and an increase in the level of van der Waals interaction with and partition into a nonpolar solvent. Therefore, stronger binding between the hydrophobic T?PP⁺ and mitochondrial membrane occurs. This is exemplified in a (hexachloro-fluorescein)-TPP⁺ conjugate system, where an ～12 times increase in the rate of mitochondrial uptake and a 2 times increase in photodynamic therapy (PDT) efficacy against HeLa and FU97 cancer cells are achieved when TPP⁺ is replaced with T?PP⁺. Importantly, nearly all the FU97 cells treated with the (hexachloro-fluorescein)-T?PP⁺ conjugate are killed as compared to only half the population of cells in the case of the (hexachloro-fluorescein)-TPP⁺ conjugate at a similar PDT light dosage. This study thus forms a platform for the healthcare community to explore alternative TPP⁺ derivatives that can act as optimal drug transporters for enhanced mitochondrially targeted therapies.

Full text of DOI:10.1021/acs.bioconjchem.6b00682

Subscriber access provided by Fudan University
Article
Unique Triphenylphosphonium Derivatives for Enhanced
Mitochondrial Uptake and Photodynamic Therapy
Zhang Hu, Ying Sim, Oi Lian Kon, Wai Har Ng, António Ribeiro, Maria João Ramos, Pedro
A. Fernandes, Rakesh Ganguly, Bengang Xing, Felipe Garcia, and Edwin K. L. Yeow
Bioconjugate Chem., Just Accepted Manuscript • DOI: 10.1021/acs.bioconjchem.6b00682 • Publication Date (Web): 03 Jan 2017
Downloaded from http://pubs.acs.org on January 5, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Bioconjugate Chemistry is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
Unique Triphenylphosphonium Derivatives for
Enhanced Mitochondrial Uptake and
Photodynamic Therapy
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Zhang Hu,1 Ying Sim,1 Oi Lian Kon,2 Wai Har Ng,2 António J. M. Ribeiro,3 Maria J. Ramos,3
Pedro A. Fernandes,*,3 Rakesh Ganguly,1 Bengang Xing,1 Felipe García,*,§,1 and Edwin K. L.
Yeow*,§,1
1School of Physical and Mathematical Sciences, Division of Chemistry and Biological
Chemistry, Nanyang Technological University, 21 Nanyang Link, 637371, Singapore
2Division of Medical Sciences, Laboratory of Applied Human Genetics, Humphrey Oei Institute
of Cancer Research, National Cancer Centre, 169610, Singapore
3UCIBIO, REQUIMTE, Departamento de Química e Bioquímica,Faculdade de Ciências,
Universidade do Porto, Rua do Campo Alegre s/n, 4169ꢀ007, Portugal
*Email: pafernan@fc.up.pt, fgarcia@ntu.edu.sg, edwinyeow@ntu.edu.sg
1
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 2 of 34
1
2
3
4
Abstract
In this study, unique methyl functionalized derivatives (T*PP+) of the drug carrier
triphenylphosphonium (TPP+) that exhibit significant enhancement in the accumulation of both
the cation and its conjugated cargo in cell mitochondria are designed. We show that the presence
of methyl group(s) at key positions within the phenyl ring results in an increase in the
hydrophobicity and solvent accessible surface area of T*PP+. In particular, when the paraꢀ
position of the phenyl ring in T*PP+ is functionalized with a methyl group, the cation is most
exposed to the surrounding environment; leading to a high drop in water entropy, and an increase
in Van der Waals interaction with and partition into a nonꢀpolar solvent. Therefore, a greater
binding between the hydrophobic T*PP+ and mitochondrial membrane is formed. This is
exemplified in a (hexachloroꢀfluorescein)ꢀTPP+ conjugate system, where an ca. twelve times
increase in mitochondrial uptake and two times increase in photodynamic therapy (PDT) efficacy
against HeLa and FU97 cancer cells is achieved when TPP+ is replaced by T*PP+. Importantly,
nearly all the FU97 cells treated with (hexachloroꢀfluorescein)ꢀT*PP+ conjugate are killed as
compared to only half the population of cells in the case of (hexachloroꢀfluorescein)ꢀTPP+
conjugate at similar PDT light dosage. This study thus forms a platform for the healthcare
community to explore alternative TPP+ derivatives that can act as optimal drug transporter for
enhanced mitochondriaꢀtargeted therapies.
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
ACS Paragon Plus Environment

Page 3 of 34
Bioconjugate Chemistry
1
2
3
4
INTRODUCTION
5
6
7
8
Mitochondria are essential cell organelles responsible for cellular respiration and apoptosis.
Several degenerative diseases including Parkinson’s disease, heart conditions, and cancer have
been attributed to dysfunctional mitochondria.1ꢀ3 Therefore, many efforts have been focused on
the design of optimal mitochondriaꢀtargeting drugs specifically delivered into the mitochondria.4ꢀ
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
11
There are several strategies available to targeted mitochondrial delivery including the use of
SzetoꢀSchiller peptides, mitochondrialꢀtargeting peptides and lipophilic cations (LC).12,13 LC are
capable of effectively permeating mitochondrial membranes due to their large hydrophobic
surface area and delocalized charge distribution.14,15 A large portion of LC, initially adsorbed on
the outer surface of mitochondria, are transported through the outer and inner membranes and
accumulated by severalꢀhundredfold due to the large membrane potential across the inner
mitochondrial membrane.16 One of the most commonly utilized LC for drug delivery is the
triphenylphosphonium cation (TPP+) (1 in Figure 1a) that is conjugated to a therapeutic agent via
a linker (e.g., alkyl chain).14,15 Previous studies have shown that modification of the
hydrophobicity of the TPP+ꢀdrug conjugate system by lengthening the alkyl chain between TPP+
and drug results in a significant improvement on the accumulation of drugs in the
mitochondria.17,18 However, the length and type of linkers have been known to affect the stability,
toxicity and usage (e.g., drug release by cleaving the linker between drug and carrier) of drug
conjugates.19,20 Moreover, there are no studies on the effect of phenyl substitution on the
mitochondrial uptake properties of TPP+. Hence, it is imperative to investigate alternative
strategies to control the hydrophobicity and mitochondrial uptake of TPP+ꢀdrug conjugates
without involving changes to the preꢀset linker.
3
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 4 of 34
1
2
3
4
5
6
Photodynamic therapy (PDT) of cancer involves the light irradiation of photosensitizerꢀ
stained tumour cells in the presence of molecular oxygen to produce reactive oxygen species
7
8
9
1
(e.g., singlet oxygen O2) that are capable of damaging and killing cells.21,22 Since mitochondria
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
are important regulators of cell death, PDTꢀinduced cytotoxicity within the mitochondria is a
promising cancer treatment with minimal damages to healthy tissues.21,23ꢀ26 Another advantage of
mitochondriaꢀtargeted PDT is the avoidance of defective signalling routes often encountered in
drugꢀinduced apoptotic responses that give rise to resistance in conventional chemotherapy.21 So
far, mitochondriaꢀtargeted PDT is still in its infancy, and the development of better
mitochondriaꢀtargeting photosensitizers (e.g., fluorescein) is instrumental for its future
progress.24ꢀ26
1
Fluorescein is capable of generating O2 when irradiated with light27 and is therefore
commonly used as a photosensitizer (PS) in both antimicrobial photodynamic therapy28 and
fluorophoreꢀassisted light inactivation (FALI) of endogenous proteins.27,29,30 In FALI, a strategy
employed to label proteins in vivo involves fusing FKBP12(F36V) to the targeted protein.29 A
fluoresceinꢀligand conjugate, where the engineered ligand is not only cell permeable but also
binds to FKBP12(F36V), results in efficient cellular protein labeling. The stained protein is
subsequently inactivated by 1O2 generated from the fluorescein conjugate during light
exposure.29 Therefore, fluorescein conjugated to suitable cellꢀpermeable carriers can effectively
function as a PDT agent.
Our approach in this study is based on the facile methyl functionalization of TPP+ to
enhance both delivery and mitochondrial uptake (see tri*phenylethylphoshonium cations T*PP+
2ꢀ7 in Figure 1a). We have demonstrated that the accumulation of the modified T*PP+ species in
the mitochondria of HeLa cells and FU97 gastric cancer cells increases significantly upon
4
ACS Paragon Plus Environment

Page 5 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
substitution of the phenyl rings. In addition, hexachloroꢀfluorescein conjugates 1a, 2a and 5a
(Figure 1b) were prepared and their mitochondrial uptake and PDT efficacy in HeLa and FU97
cells determined. Unconjugated fluorescein does not accumulate significantly in the
cell/mitochondria and therefore does not display any PDT effects. On the other hand, conjugates
1a, 2a and 5a permeate both plasma and mitochondrial membranes and are accumulated in the
mitochondria; hence allowing efficient PDT to occur. When compared to TPP+, T*PP+ cations
significantly increase the accumulation of photosensitizer in mitochondria which leads to a
nearly two times improvement in the PDT of cancer cells. This study underscores the importance
of the presence of the mitochondrial directing LC.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. (a) Structures of triphenylethylphosphonium 1 and tri*phenylethylphosphonium (T*PP+; where
*phenyl = monoꢀ (2, 3 and 4), diꢀ (5, and 6) and triꢀmethyl (7) functionalized phenyl); triꢀpꢀtolylꢀ
ethylphosphonium (2), triꢀmꢀtolylethylphosphonium (3), triꢀoꢀtolyethylphosꢀphonium (4), trisꢀ(3,5ꢀ
xylylethylphosphonium (5), trisꢀ(2,4ꢀxylylethylphosphoꢀnium (6) and trisꢀmesitylethylphosphonium (7).
(b) Structures of the (hexachloroꢀfluorescein)ꢀTPP+/T*PP+ conjugates: (hexachloroꢀfluorescein)ꢀ
triphenylethylphosphonium (1a), (hexachloroꢀfluorescein)ꢀtriꢀpꢀtolylethylphosphonium (2a) and
(hexachloroꢀfluorescein)ꢀtrisꢀ(3,5ꢀxylyl)ethylphosphonium (5a).
5
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 6 of 34
1
2
3
4
RESULTS AND DISCUSSION
5
6
7
8
Partition coefficient and mitochondrial uptake of cations 1ꢀ7. The octanolꢀwater partition
coefficients (P) for compounds 1ꢀ7 (i.e., ratio of the concentration of cation in 1ꢀoctanol to the
concentration of cation in water) were determined to establish the relative hydrophobicity of the
various cations (Figure 2a). T*PP+ cation 7, containing triꢀmethyl substituted phenyl rings, has
the highest P and is the most hydrophobic followed by cations with biꢀmethyl substituted (5 and
6) and monoꢀmethyl substituted (2, 3 and 4) phenyl rings. The least hydrophobic salt in the series
is the commonly used TPP+ 1. As expected, a positive correlation exists between hydrophobicity
and number of methyl groups present due to the additional nonꢀpolar substituents. Interestingly,
not only the number but also the phenyl ring substitution pattern plays a significant role in the
overall hydrophobicity of the cation. In the case of the monoꢀsubstituted phenyl, the
hydrophobicity increases in the order 4 < 3 < 2; where the phosphonium salt displays greatest
hydrophobicity when the paraꢀ position is substituted and least hydrophobic when the orthoꢀ
position is substituted.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In order to evaluate if the hydrophobic variation observed for compounds 1ꢀ7 would lead
to different mitochondrial uptake, their accumulation in isolated mitochondria was determined.
Basically, HeLa and FU97 mitochondria were isolated using a mitochondrial isolation kit
(Miltenyi Biotec), and the protein quantity in the mitochondria was determined using the
Bradford assay. Salt uptake was determined by first incubating freshly isolated mitochondria
(~0.135 mg of proteins per L) with the various cations 1ꢀ7 (500 µM) for 5 min. The
mitochondria were then lysed and the quantity of accumulated 1ꢀ7 (Cm) determined using HPLC.
The results for HeLa and FU97 cells are presented in Figure 2b and S11 (Supporting
Information), respectively. From Figure 2c, we note a positive correlation between Cm and Log P
6
ACS Paragon Plus Environment

Page 7 of 34
Bioconjugate Chemistry
1
2
(i.e., hydrophobicity) where T*PP+ uptake by the mitochondria is enhanced for the more
hydrophobic cations. In particular, there is a ca. 4 and 7 times enhancement in the uptake of 7 as
compared to 1 for HeLa and FU97 mitochondria, respectively.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. (a) Values of Log P and (b) amount of 1ꢀ7 accumulated (Cm) in isolated HeLa mitochondria. (c)
Positive correlation between Log P and Cm, and (d) negative correlation between Cm and ꢀGaq→oct/vdW.
The r2 values for the linear fits are provided.
Computational study. The free energy associated with the permeation of small charged
molecules across a membrane from an aqueous environment can be decomposed into the Born
free energy and hydrophobic (or neutral) free energy.14,31 The Born free energy WB = 339(Z2/r)
is inversely proportional to the radius (r) of the cation,14,31,32 whereas the hydrophobic free
energy was empirically found to be proportional to the accessible surface area of the cation (
∝
r2).14,31
7
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 8 of 34
1
2
3
4
5
6
7
8
9
To gain insight into the underlying cause of the trends observed in Section 2.1, molecular
dynamics (MD) calculations were performed for molecules 1ꢀ7 in a cubic box containing either
water or 1ꢀoctanol molecules. Thermodynamic integration was used to compute the free energy
ꢀGaq→oct associated with moving a solute molecule from water (aq) to 1ꢀoctanol (oct). ꢀGaq→oct
is defined to be the difference between the free energies for the annihilation of the solute in water
(ꢀGaq) and in 1ꢀoctanol (ꢀGoct) (i.e., ꢀGaq→oct = ꢀGaq ꢀ ꢀGoct). The values of ꢀGaq→oct with the
corresponding van der Waals (vdW) and electrostatic contributions to the free energy are
provided in Table 1. Furthermore, the solvent accessible surface area (SASA) of the cations are
almost equivalent in both water and 1ꢀoctanol (average unsigned difference of 0.5 Å2); showing
that the structural conformations adopted by 1ꢀ7 are maintained in the two solvents (Table 1).
Visualization of the trajectories confirms this observation.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 1. Free energy ꢀGaq→oct associated with moving a solute molecule from water (aq) to 1ꢀoctanol
(oct), and corresponding van der Waals (vdW) and electrostatic interaction contributions in kcal molꢀ1.
The solvent accessible surface area SASA for 1ꢀ7 are given
.
cation
SASA/Å2
535.8±0.1
626.1±0.1
623.0±0.1
566.3±0.1
708.6±0.1
653.3±0.1
673.6±0.1
ꢀGaq→oct
ꢀ17.7±0.2
ꢀ21.3±0.2
ꢀ20.7±0.2
ꢀ19.5±0.2
ꢀ23.6±0.3
ꢀ22.6±0.2
ꢀ23.6±0.3
vdW
electrostatic
ꢀ3.0±0.1
ꢀ3.7±0.1
ꢀ3.5±0.1
ꢀ3.4±0.1
ꢀ3.9±0.1
ꢀ3.3±0.1
ꢀ3.5±0.2
1
2
3
4
5
6
7
ꢀ14.7±0.2
ꢀ17.5±0.2
ꢀ17.1±0.2
ꢀ16.1±0.2
ꢀ19.7±0.2
ꢀ19.3±0.2
ꢀ20.1±0.2
8
ACS Paragon Plus Environment

Page 9 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
An almost negligible correlation exists between the electrostatic component of ꢀGaq→oct
with the experimental partition coefficient P and T*PP+ uptake Cm (Table 1). This is because the
calculated electrostatic solvation free energy is mostly dependent on the charge of the cation
which remains invariant for the cations studied here. Furthermore, assuming a spherical model
(calculated sphericity ~0.8 for 1 and 7), the calculated radius of 7 is only ~1.1 times larger than
that of 1 which suggests that changes in the Born free energy between the various cations is not
significant to fully explain the trends seen above.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. Positive correlation between ꢀGaq→oct and vdW energy (a) and negative correlation between
ꢀGaq→oct and SASA (b). The r2 values for the linear fits are provided and the corresponding cations for
each point are given.
On the other hand, we note that the magnitudes of both the calculated ꢀGaq→oct and the
vdW component of the free energy positively correlate with P and Cm (Figure 2d for HeLa cells).
The vdW component describes the nonꢀpolar (hydrophobic) interaction and is thus dependent on
the hydrophobicity of the cation. From Table 1, it is observed that a positive/negative correlation
exists between ꢀGaq→oct and the vdW/SASA term (Figure 3). As the number of substituted
9
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 10 of 34
1
2
3
4
5
6
7
8
9
methyl groups increases, a larger (hydrophobic) surface area of the cation is accessible to the
solvent. This leads to a higher drop in water entropy as more water molecules surrounding the
hydrophobic cation have less orientational freedom to form hydrogen bonds with other water
molecules; resulting in a poorer solvation in water. On the other hand, when the larger surface
area of the more hydrophobic cation is exposed to 1ꢀoctanol molecules, an increase in vdW
interaction leads to a favorable partition into the organic solvent (i.e., more negative ꢀGaq→oct).
A similar explanation is used to rationalize the trend observed within the monoꢀ
substituted cation group. In this case, the SASA increases in the order 4 < 3 < 2, indicating that
when the paraꢀ position is substituted, the methyl groups in 2 are better exposed to the solvent as
compared to the ones in 3 and 4. In particular, methyl groups at the orthoꢀ position in 4 are less
accessible to solvent molecules due plausibly to steric hindrance arising from the adjacent
phosphorous. Since more 1ꢀoctanol molecules are able to interact with 2, a stronger dispersion
interaction gives rise to the observed increased partition in 1ꢀoctanol (Figure 2a). In water, the
larger hydrophobic surface will give rise to a larger water entropy penalty. A similar driving
force is responsible for facilitating the uptake of the more hydrophobic cation in the
mitochondrial membrane where the greater binding of T*PP+ within the membrane may lead to
the observed higher mitochondrial uptake.18
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Mitochondrial uptake of dye conjugates increases with the hydrophobicity of the lipophilic
cation. We further examine the ability to deliver a hexachloroꢀfluorescein dye (i.e., cargo
sensitizer) into the mitochondria by T*PP+ of different hydrophobicity (i.e., carrier). In our study,
HeLa and FU97 cells were utilized and first incubated with similar amounts (20 µM) of
hexachloroꢀfluorescein, and conjugates 1a, 2a and 5a for 3 h. As control, the cells were also
10
ACS Paragon Plus Environment

Page 11 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Confocal fluorescence microscopy imaging of FU97 cells treated with 5a, 2a and 1a. The
fluorescence images upon 543 nm excitation of the dyeꢀT*PP+ conjugates and 633 nm excitation of
Mitotracker Deep Red are given in the first and second columns, respectively. Overlay images from the
first and second columns are given in the third column. The scale bar represents 20 ꢁm.
treated with Mitotracker Deep Red dye to monitor the progress of mitochondrial staining. After
washing the cells to remove free dyes, fluorescence imaging experiments were conducted using
laser light of 543 nm to excite hexachloroꢀfluorescein (first column in Figure 4) and 633 nm to
excite Mitotracker Deep Red (second column in Figure 4). No obvious fluorescence was detected
in the same cells after incubation with the unconjugated hexachloroꢀfluorescein; suggesting that
dye alone under the conditions employed here is not effectively delivered into the cell
mitochondria (Figure S12 and S13 in the Supporting Information for FU97 and HeLa cells,
respectively). In contrast, similar incubation with conjugates 1a, 2a and 5a leads to distinct
fluorescence images (Figure 4 and S13 in Supporting Information for FU97 and HeLa cells,
11
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 12 of 34
1
2
respectively). Images from PSꢀTPP+ 1a and PSꢀT*PP+ 2a and 5a overlap with the ones with
Mitotracker emission, suggesting that 1a, 2a and 5a are taken up by the cells and coꢀlocalized
with Mitotracker Deep Red in the mitochondria. Moreover, the fluorescence intensity from the
hexachloroꢀfluorescein moiety of the conjugates increases in the order: 1a < 2a < 5a (Figure S14
in the Supporting Information); suggesting that 5a is most readily taken up.
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We further quantify the uptake of PSꢀTPP+ and PSꢀT*PP+ by isolated HeLa and FU97
mitochondria using HPLC. After incubating freshly isolated mitochondria (~0.135 mg of
proteins per L) with 1a, 2a and 5a (500 µM) for 5 min, the amounts of accumulated 1a, 2a and
5a in HeLa/FU97 cells were determined to be 1.1±0.1/1.6±0.2, 7.3±0.3/8.1±0.5 and
13.2±0.6/20.2±0.9 mmol gꢀ1, respectively. These data positively correlate with the partition
coefficients (i.e., Log P of 1a, 2a and 5a are 0.19 ± 0.02, 0.91 ± 0.01 and 3.66 ± 0.02, respectively),
and increase in the order: 1a < 2a < 5a. In particular, we note that 5a is accumulated ~12 times
more than 1a in the mitochondria. Amongst the 3 conjugate systems, the most hydrophobic trisꢀ
(3,5ꢀdimethylphenyl)phosphonium (T*PP+ 5) in 5a acts as the best cargo transporter, followed
by triꢀpꢀtolyphosphonium (T*PP+ 2) in 2a and triphenylphosphonium (TPP+ 1) in 1a. The
fluorescence quantum yields and lifetimes of hexachloroꢀfluorescein, 1a and 5a in water did not
vary significantly and were measured to be 0.94 and 3.9 ns, respectively; indicating that the
photophysics of the sensitizer is not perturbed when linked to either TPP+ or T*PP+ (Figure S15
in Supporting Information). Therefore, it is most likely that the large accumulation of 5a in the
mitochondria leads to the higher emission intensity observed in the fluorescence cell image
(Figure 4) when compared to the other conjugates.
When the mitochondrial membrane potential (ꢀψm) was reduced by treating freshly
isolated HeLa mitochondria with FCCP (carbonyl cyanideꢀ4ꢀ(trifluoromethoxy)phenylhydrazone,
12
ACS Paragon Plus Environment

Page 13 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
500 nM) for 1 min before incubation with 5a, the latter was not detected in the mitochondria
using HPLC. This indicates that 5a is not accumulated in the mitochondria when ꢀψm is
abolished, and the accumulation of 5a is thus driven by the presence of the membrane potential.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
PDT efficacy is enhanced when hydrophobicity of cation increases. We next examine the
PDT efficacy of 3 compounds: hexachloroꢀfluorescein, 1a and 5a. Both HeLa and FU97 cells
were first incubated with different concentrations (i.e., 10, 20, 40 and 60 µM) of hexachloroꢀ
fluorescein, 1a and 5a for 3 h. The cells were then washed with PBS before exposure to white
light. Resazurin assay was employed to evaluate the viability of the lightꢀtreated cells after
incubation for 24 h. In the control experiment using free hexachloroꢀfluorescein, cell viability
remains high (> 90 %) for both HeLa and FU97 cells after 15 min of light irradiation (light
intensity of 14.3 J cmꢀ2). PDT is not effective because of the small amounts of dye accumulated
in the cells. In addition, the light dosages used here are insufficient to cause significant damages
to the cells.
In the absence of light irradiation, cells treated with 1a and 5a did not show obvious
cytotoxicity and high cell viability is observed (i.e., close to 100 %) (Figure 5). Conjugate
systems 1a and 5a are nonꢀlethal at concentrations as high as 60 ꢁM. The lack of dark
cytotoxicity is consistent with the IC50 values determined for 1a (192 ± 8 ꢁM for HeLa and
160 ± 12 ꢁM for FU97) and 5a (200 ± 11 ꢁM for HeLa and 207 ± 12 ꢁM for FU97). After 15 min
of light irradiation, cells treated with 5a clearly exhibited more cell deaths when compared to
those treated with an equal amount of 1a (Figure 5). For example, the viabilities of HeLa cells
treated with 60 ꢁM of 5a and 1a are 32 % and 68 %, respectively, after 15 min of light exposure
(Figure 5a). When FU97 cells were treated with 60 ꢁM of 5a, nearly all of the cells were killed
13
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 14 of 34
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Phototoxicity of 5a at various concentrations (0, 10, 20, 40 and 60 µM) and light irradiation
times (0, 5, 10 and 15 min) for (a) HeLa and (b) FU97 cells. The phototoxicity of 1a at different
concentrations after 15 min light irradiation is also displayed.
as compared to only 55.3 % of cells treated with the same amount of 1a (Figure 5b). Furthermore,
Figure 5 shows that elevated tumour cell damage occurs when the concentration of PSꢀT*PP+
conjugate used or light exposure time is increased; indicating that PDT is responsible for cell
death.
Upon light excitation, hexachloroꢀfluorescein, 1a and 5a in PBS react with the
surrounding O2 to produce singlet oxygen 1O2 as detected by the ABDA (9, 10ꢀanthracenediylꢀbis
(methylene) dimalonic acid) assay (Figure S16 in Supporting Information). Furthermore, the
1
absorbance of ABDA decreases with light irradiation time as the amount of O2 produced
1
increases. Previous studies have reported that late stages of apoptosis are triggered by O2
created during light irradiation of photosensitizers located in mitochondria.23 In this case,
mitochondrial permeabilization and photodamage of antiapoptotic protein bclꢀ2 can induce the
release of caspase activators such as cytochrome c or other proꢀapoptotic molecules (e.g.,
apoptosisꢀinducing factor) into the cytosol.33ꢀ35 PDTꢀmediated apoptosis occurs when conjugates
14
ACS Paragon Plus Environment

Page 15 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
1a and 5a, localized in the mitochondria, are exposed to light as indicated by positive Annexin V
staining. Annexin V emission from cells treated with 5a is, in general, brighter than the
corresponding fluorescence from cells treated with 1a (Figure 6 for HeLa cells); demonstrating
that at the same light dosage, the higher mitochondrial uptake of 5a leads to more cell deaths
when compared to 1a.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 6. Fluorescence imaging of HeLa cells treated with 1a and 5a (20 ꢁM), irradiated with white light
for 15 min, incubated for 1 day and then treated with Annexin V. Annexin V emission from cells treated
with 5a is, in general, brighter than the corresponding fluorescence from cells treated with 1a (first
column). Corresponding images from hexachloroꢀfluorescein conjugate fluorescence are given in the
second column. Overlay images from the first and second columns are given in the third column. Scale
bar represents 20 µm
.
15
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 16 of 34
1
2
3
4
CONCLUSIONS
5
6
7
8
In summary, we have demonstrated a readilyꢀaccessible, enhanced alternative to the
conventionally and widely used mitochondriaꢀtargeting triphenylphosphonium cation by simple
modification using commerciallyꢀavailable phosphine reagents. Our experimental and
computational studies have unambigiously demonstrated that increasing both the hydrophobicity
of the cation and solvent accessible surface area by methyl substitution at various positions of the
phenyl rings significantly improve the partition of T*PP+ cation in a hydrophobic environment,
and enhances the accumulation of conjugated cargo within the mitochondria. In particular, the
use of the more hydrophobic T*PP+ cationic carrier 5 facilitates a significantly better
mitochondrial uptake of conjugated hexachloroꢀfluorescein in HeLa and FU97 cells which in
turn leads to a two times improvement in PDT efficacy upon light irradiation. The LCs
developed in our study can be easily conjugated to other therapeutic cargos such as anticancer
prodrugs (e.g., doxorubicin) and antioxidants (e.g., ubiquinone).11
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
It is noteworthy to question, from our study, whether the inclusion of a broader range of
organic substituents onto lipophilic cations may additionally enhance their cell uptake efficiency
and hence further improve therapeutic applications. The rational design of optimal mitochondrial
delivery systems is both an exciting challenge in the area of synthetic chemistry and is relavent
to future mitochondrial therapies.
EXPERIMENTAL METHODS
Chemicals and materials. Phosphines 97% were purchased from Strem Chemicals, 6ꢀHEX
dipivaloate (98%) was purchased from Chemodex, the other chemicals were purchased from
Sigma Aldrich and all were used without further purification. Solvents were distilled over
sodium/benzophenone, dried and stored under 4 Å molecular sieves. Reactions were carried out
16
ACS Paragon Plus Environment

Page 17 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
using standard inertꢀatmosphere, Schlenk techniques or performed under argon. Human
epithelial carcinoma cell line (HeLa) was purchased from ATCC (ATCC no. CCLꢀ2); Stomach
carcinoma cell FU97 was obtained from Prof Oi Lian Kon’s research laboratory (National
Cancer Centre). MitoTracker Deep Red FM and the apoptosis imaging reagent (Annexin V
Alexa Fluor 488 conjugate) were purchased from Thermo Fisher Scientific. Mitochondria
Isolation Kit was purchased from Miltenyi Biotec.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
13
Instrumentation. H, C NMR (400/100 MHz) spectra were collected using a Bruker Avance
DPX400 spectrometer with the 1H, 13C NMR chemical shifts internally referenced to the relevant
residual solvent peaks. All NMR spectroscopic analysis were performed at room temperature
(300 K). Highꢀresolution mass spectra were obtained from Water QꢀTof Premier with ESI mode.
Melting points were determined using a SRSꢀOptimelt MPAꢀ100 apparatus using sealed glass
capillaries under argon and were uncorrected. Infrared spectra were recorded as Nujol mulls
using NaCl plates on a Shimadzu IR Prestigeꢀ21 FTIR spectrometer. Reverseꢀphase HPLC
analysis was performed on a Shimadzu HPLC system using an Alltima Cꢀ18 (250 × 10 mm)
column at a flow rate of 3.0 mL/min for preparation and a Cꢀ18 (250×4.6 mm) at a flow rate of
1.0 mL/min for analysis.
Diffractionꢀquality crystals were obtained by slow evaporation of solvent from solutions
in acetonitrile/water solvent mixtures at room temperature. The crystals were mounted onto
quartz fibers, and the Xꢀray diffraction intensity data were measured at 103 K with a Bruker
Kappa diffractometer equipped with a CCD detector, employing Mo K α radiation (λ = 0.71073
Å), with the SMART suite of programs.36 All data were processed and corrected for Lorentz and
polarization effects with SAINT and for absorption effects with SADABS.37 Structural solution
17
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 18 of 34
1
2
and refinement were carried out with the SHELXTL suite of programs.38 The structures were
solved by direct methods or Patterson maps to locate the heavy atoms, followed by difference
maps for the light, nonꢀhydrogen atoms. All nonꢀhydrogen atoms were refined with anisotropic
thermal parameters. The crystals of 5 had acetonitrile and 6 has a molecule of water as solvent of
crystallization. In 7 the bromide anion was modeled in two alternative sites (with 0.5 occupancy).
Crystallographic data is available from the Cambridge Crystallographic Database Centre (CCDC
1478282ꢀ1478291).
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Synthesis. TPP+ 1 and T*PP+ 2ꢀ7 (Figure 1a) were synthesized by dropwise addition of the
respective tertiary phosphine to a solution of ethylbromide in acetonitrile and reacted under
reflux or in a microwave. The (hexachloroꢀfluorescein)ꢀphosphonium salt conjugates 1a, 2a and
5a (Figure 1b) were prepared by reacting the respective tertiary phosphine precursors with 2ꢀ
bromoethylamine hydrobromide in acetonitrile to obtain their tri*phenylethylaminephosphonium
cation counterparts. The amino functionalized phosphonium cations were conjugated to a
hexachloroꢀfluorescein through the coupling reaction between the amine group on the
phosphonium salt and NHS group on the dye molecule. The products 1a, 2a and 5a were purified
by HPLC and further characterized by NMR and MS analysis. Detailed information on the
synthesis and characterization of the various compounds are provided in the Supporting
Information.
Determination of the octanolꢀwater partion coefficient. The extinction coefficients of 1ꢀ7 in
waterꢀsaturated 1ꢀoctanol and 1ꢀoctanolꢀsaturated water were determined using absorption
spectroscopy (Cary 100 UVꢀVis, Varian) and the BeerꢀLambert law. An approximate amount of
1.00 – 2.50 mg of sample was weighed and dissolved in 50 – 70 ml of saturated 1ꢀoctanol. The
18
ACS Paragon Plus Environment

Page 19 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
mixture is allowed to stir to ensure the formation of a homogenous solution. Next, with the stock
solution and 1ꢀoctanolꢀsaturated water, 3 sets of different volume ratios of 1ꢀoctanol to water
mixtures of each sample were prepared. In this study, the three volume ratios used were 1:5, 1:10
and 1:20 (1ꢀoctanol:water). Absorbance for each phase was recorded and used to calculate the
partition coefficient P values:
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
P = [LC]octanol/[LC]water
where [LC]octanol and [LC]water are the concentrations of the lipophilic cation in octanol and water,
respectively.
Determination of the half maximal inhibitory concentrations (IC50) of 1a, 2a and 5a. HeLa
and FU97 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) solution with
10 % fetal bovine serum (FBS) and 1 % Penicillin/Streptomycin under humidified atmosphere of
5 % CO2 at 37 °C. The IC50 was measured using Resazurin assay. The HeLa and FU97 cells
were seeded on a 96ꢀwells containing 10000 cells per well in 100 ꢁL DMEM media and
incubated overnight before the addition of 1a, 2a and 5a. Upon incubation at 37 °C for an
additional 72 h, the media were removed and the cells washed with PBS. Resazurin solution 100
µL (1:10 v/v ratio of resazurin solution:DMEM (no phenol red)) was added to each well before
incubation for 2 h at 37 °C. The samples were excited using a 560 nm light and the fluorescence
was recorded on a Tecan's Infinite M200 microplate reader using a 590 nm emission filter.
Different concentrations of 1a, 2a and 5a were used and for each concentration, the experiment
was repeated three times. The IC50 was determined from the plot of viability against
concentration of conjugates.
19
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 20 of 34
1
2
3
4
5
6
7
8
9
Isolated mitochondria preparation and mitochondrial quantification. Isolated mitochondria
were prepared using a mitochondria isolation kit (Mitochondria Isolation Kit, human 130ꢀ094ꢀ
532, Miltenyi Biotec). First, Hela or FU97 cells were lysed and the mitochondria magnetically
labeled with AntiꢀTOM22 microbeads. The monoclonal AntiꢀTOM22 antibody specifically binds
to the translocase of outer mitochondrial membrane 22 (TOM22) of human mitochondria. The
labeled cell lysate was loaded onto a LS column placed in a MidiMACS Separator. After
washing, only magnetically labeled mitochondria remained on the column which were then
removed from the separator. Functional human mitochondria were eluted from the column and
stored in storage buffer. Western blot was used to ascertain that proteins isolated were from cell
mitochondria.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Bradford Assay was used to determine the concentration of proteins in mitochondria. Mꢀ
PER mammalian protein extraction reagent was used to extract all protein constituents of the
mitochondria. Samples and diluted Bradford dye reagent (200 µL) were pipetted into separate
microtiter plate wells and incubated at room temperature for 10 minutes. UVꢀabsorbance was
measured at 595 nm and the concentration of proteins was determined from a calibration curve
obtained using a protein standard (i.e., BSA).
Mitochondrial uptake of salts. Freshly isolated mitochondria (0.135 mg protein) were
incubated with the salts (500 µM) in 2 mL plastic tubes in a shaking water bath at 25 °C. After 5
min, the mitochondria were lysed by centrifugation (7500×g) for 10 min, and the pellets
extracted by vortexing in 250 ꢁL Buffer B (99.9 % acetonitrile (ACN), 0.1 % trifluoroacetic acid
(TFA)) followed by centrifugation (7500×g) for 10 min. Supernatants (250 ꢁL) were removed
and stored at 4 °C in glass vials. All samples were diluted to 25 % ACN by addition of Buffer A
(99.89 % H2O, 0.1 % TFA), filtered using a Milipore Millex syringeꢀdriven filter unit (0.22 ꢁm)
20
ACS Paragon Plus Environment

Page 21 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
and subsequently injected into a 2 mL sample loop and separated by RPꢀHPLC; a C18 column
with a widepore C18 guard column were used. A gradient of buffer A and B was run at 1 mL
minꢀ1 (% B): 0–5 min, 5–15 %; 5–31 min, 15–100 %; 31–35 min, 100–5 %. Peaks were detected
at 220 nm using a UVꢀVis spectrophotometer. Known salt standards were used to determine
elution times and calibration curves. To reduce the mitochondrial membrane potential, freshly
isolated mitochondria (0.135 mg protein) was treated with 500 nM FCCP (carbonyl cyanideꢀ4ꢀ
(trifluoromethoxy)phenylhydrazone) for 1 min.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Fluorescence bioꢀimaging experiments. HeLa or FU97 cells were seeded in a µꢀdish (iBidi,
Germany) at a density of 100 000 cells (1 mL) per well one day prior to any measurements.
o
Hexachloroꢀfluorescein, 1a, 2a and 5a (20 µM) were incubated with the cells at 37 C for 3 h in
DMEM. Mitotracker 633 (100 nM) was added 30 min before the end of the incubation period.
Cells were then washed three times with PBS (1X) and imaged using a confocal microscope
(Nikon Eclipse TE2000) equipped with 543 nm argon and 633 nm lasers.
PDT and cytotoxicity assay after light irradiation. HeLa/FU97 cells were seeded in 96ꢀwell at
a density of 10 000 cells (100 µL) per well one day prior to experiments. The cells were treated
with different concentrations of hexachloroꢀfluorescein, 1a and 5a (i.e., 10, 20, 40 and 60 ꢁM)
for 3 h. Cells were irradiated with white light (400ꢀ800 nm light) for different time durations
(i.e., 5, 10 and 15 min), followed by 24 h incubation before performing the cell viability test (i.e.,
Resazurin assay). All experiments were repeated 3 times and the averaged taken.
Apoptosis imaging analysis. HeLa cells were seeded in a µꢀdish (iBidi, Germany) at a density
of 100 000 cells (1 mL) per well one day prior to experiments. The cells were incubated with 1a
21
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 22 of 34
1
2
3
4
5
6
7
8
9
and 5a conjugates (20 µM) for 3h in DMEM. Subsequently, drugꢀcontaining medium was
replaced with fresh culture medium and washed with PBS followed by irradiation with white
light (400ꢀ800 nm, light intensity of 14.3 J cmꢀ2) for 15 min. After an additional incubation time
of 24 h, the cells were washed with PBS and redistributed in fresh medium. The apoptosis
imaging reagent (Annexin V Alexa Fluor 488 Conjugate) was added and the cells further
incubated for 15 min. Live cell imaging was carried out using a Nikon Eclipse TE2000 Confocal
Microscope. Excitation filters were set as 488 nm and 543 nm for Annexin V and 1a/5a,
respectively.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Fluorescence quantum yield, lifetime and singlet oxygen detection. Fluorescence
quantum yields were performed using a fluorescence spectrophotometer (Varian Cary
Eclipse) using an excitation wavelength of 500 nm. The fluorescence lifetime was
recorded using a timeꢀcorrelated single photon counting spectrofluorimeter (FluoroCube,
Horiba Jobin Yvon) with a 466 nmꢀexcitation laser.
For the singlet oxygen detection, hexachloroꢀfluorescein, 1a and 5a (10 µM) were
mixed with 9, 10ꢀanthracenediylꢀbis (methylene) dimalonic acid (ABDA) (200 µM) in PBS
buffer (10 mM, pH 7.4) and placed in a 96ꢀwell. The ABDA absorbance at 378 nm was
measured. The samples were then illuminated with white light (400 ꢀ 800 nm) for different time
durations (i.e., 1, 5 and 10 min) and the absorbance of ABDA at 378 nm was reꢀmeasured.
The blank solution in the UVꢀabsorption experiment is the corresponding sample solution
without ABDA. ABDA solution (200 µM) in the absence of any sensitizer was used as in the
control experiment. The destruction of ABDA indicated the generation of singlet oxygen. Each
assay was performed twice and the averaged taken.
22
ACS Paragon Plus Environment

Page 23 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
Molecular dynamics calculations. Parametrization of the cations and 1ꢀoctanol: The
cations and 1ꢀoctanol structures were drawn in GaussView39 and firstly optimized at the
B3LYP/6ꢀ31+G(d) level of theory40ꢀ42 in Gaussian09.43 We then followed the default
antechamber44 protocol and settings to calculate all molecular mechanics parameters.
Antechamber attributes GAFF45 parameters to atoms, bonds and angles, and calculates
RESP point charges46 (at the HF/6ꢀ31G* level of theory)47 that are compatible with the
amber force field.48 Preparation of the models: A cubic box of 300 1ꢀoctanol molecules
with sides of 43 Å was built with the Packmol package.49 This software creates a packed
box of the specified molecule that minimizes both steric hindrance and empty space.50
This box was further relaxed in a 1 ns molecular dynamics simulation with the
isothermalꢀisobaric ensemble. The 300 1ꢀoctanol molecules optimized to a box of size of
43 Å x 43 Å x 43 Å and a density of 0.824 ± 0.003 (average over last half of the
simulation; error is the standard deviation), which compares very well with the
experimental density of 1ꢀoctanol at 20 ºC (0.827 g cmꢀ3). We prepared the simulation
boxes for every cationꢀsolvent combination by adding 20 Å of either TIP3P51 water
molecules or 1ꢀoctanol molecules (using the previous described box) in every direction
from the cation.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Minimization and molecular dynamics calculations: The protocol for the 14
molecular dynamics (MD) simulations (7 cations x 2 solvents) was the same: 1. Energy
minimization of the solvent molecules; 2. Energy minimization of both solute and
solvent; 3. Heating MD; 4. Equilibration MD; 5. Production MD. Both minimizations
consisted of 2500 steps with the steepest descent algorithm followed by 2500 steps with
23
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 24 of 34
1
2
3
4
5
6
the conjugate gradient algorithm. In the first minimization, the solute was restrained with
a force constant of 10 kcal molꢀ1 Åꢀ2.
7
8
9
The heating MDs run for 1 ns in the canonical ensemble (NVT). The temperature was
increased from 0 up to 300 K with the Langevin thermostat. We used the isothermalꢀisobaric
ensemble (NTP) for the equilibration MDs. Temperature was kept constant at 300 K with the
Berendsen thermostat, and the isotropic Berendsen barostat was used to keep the pressure
constant at 1 bar.52 We ran the equilibration MDs for 1 ns for the simulation with water and 10 ns
for the simulation with 1ꢀoctanol. These time lengths were enough to equilibrate the slowestꢀ
convergence property (solvent density) in the simulation boxes. Production MDs used the same
parameters as equilibration MDs and run for 10 ns for all soluteꢀsolvent combinations.
Explicit van der Waals interactions were truncated at 10 Å in all minimizations and
molecular dynamics, and the Coulombic interactions were calculated with the particle mesh
Ewald (PME) method,53 with the real part also truncated at 10 Å. We used the SHAKE algorithm
in all MDs, which constrains the length of the bonds containing hydrogen atoms, to allow the use
of a time step of 2 fs.54 All MD simulations were carried with the AMBER software.
Calculation of the accessible volume and surface area: The solvent accessible surface
area (SASA) and volume were calculated with the sasa command of Gromacs.55 A probe radius
of 1.4 Å was used for both area and volume measurements. Values were calculated for 1000
equally space snapshots of the 10 ns production MDs.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
24
ACS Paragon Plus Environment

Page 25 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
Thermodynamic integration calculations: We used thermodynamic integration
calculations to calculate the difference between hydration free energy and 1ꢀoctanol solvation
free energy for every compound. We calculated the free energy associated with two alchemical
transformations (Scheme 1): the annihilation of the solute in water and the annihilation of the
solute in 1ꢀoctanol. The difference between these two energies is equivalent to the
thermodynamic free energy of moving a solute molecule from water to 1ꢀoctanol.
The MD sampling for each transformation was done in two steps: the first is the
elimination of the atomic point charges, and the second the removal of the vdW components of
the solute. The first step was divided in 4 windows, which correspond to lambdas 0.2, 0.4, 0.6,
and 0.8. Since the effect of the neutralization of the atomic point charges in the potential energy
is typically linear, four points are enough to calculate the TI integral. The second step was
divided in 19 windows spaced by 0.05, from 0.05 up to 0.95. We used softcore potentials in the
second step.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For each window, 500 cycles of energy minimization and a 100 ps equilibration MD
preceded the production dynamics. The production MD run for 1 ns in the water solvent systems,
and 30 ns and 5 ns for the first step and second steps of the 1ꢀoctanol system. These larger
simulation times were required to obtain well converged energies. TI calculations were done in
AMBER 12. Besides the thermodynamic integration energy, we also calculated BAR and MBAR
energies with the alchemicalꢀanalysis package.
25
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 26 of 34
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 1. Free energy associated with moving a solute molecule from water to 1ꢀoctanol.
ASSOCIATED CONTENT
Supporting Information.
This material is available free of charge via the Internet at http://pubs.acs.org. Experimental
details on synthesis and characterization are provided.
AUTHOR INFORMATION
Corresponding Author
*Email: pafernan@fc.up.pt, fgarcia@ntu.edu.sg, edwinyeow@ntu.edu.sg
Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval
to the final version of the manuscript.
§F.G. and E.K.L.Y both jointly supervised the work.
ACKNOWLEDGMENT
F.G. would like to thank NTU for a startꢀup grant (M4080552) and MOE Tier 1 grant
(M4011441) for finantial support. P.A.F., A. J.M.R. and M.J.R. acknowledge financial support
from the European Union (FEDER funds POCI/01/0145/FEDER/007728) and National Funds
(FCT/MEC, Fundação para a Ciência e Tecnologia and Ministério da Educação e Ciência) under
the Partnership Agreement PT2020 UID/MULTI/04378/2013.
26
ACS Paragon Plus Environment

Page 27 of 34
Bioconjugate Chemistry
1
2
3
4
ABBREVIATIONS
5
6
7
8
LC, lipophilic cations; PDT, photodynamic therapy; PS, photosensitizer; TPP+,
triphenylphosphonium cation; T*PP+, tri*phenylethylphoshonium cation.
REFERENCES
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(1)
Frantz, M. ꢀC., and Wipf, P. (2010) Mitochondria as a target in treatment. Environ. Mol.
Mutagen. 51, 462ꢀ475.
(2)
Gaude, E., and Frezze, C. (2014) Defects in mitochondrial metabolism and cancer.
Cancer Metab. 2, 10.
(3)
Bae, S. K., Heo, C. H., Choi, D. J., Sen, D., Joe, E. ꢀH., Cho, B. R., and Kim, H. M.
(2013) A ratiometric twoꢀphoton fluorescent probe reveals reduction in mitochondrial H2S
production in parkinson’s disease gene knockout astrocytes. J. Am. Chem. Soc. 135, 9915ꢀ9923.
(4)
therapy. Nat. Rev. Drug Discov. 9, 447ꢀ464.
(5) Jean, S. R., Tulumello, D. V., Wisnovsky, S. P., Lei, E. K., Pereira, M. P., and Kelley S.
Fulda, S., Galluzzi, L., and Kromer, G. (2010) Targeting mitochondria for cancer
O. (2014) Molecular vehicles for mitochondrial chemical biology and drug delivery. ACS Chem.
Biol. 9, 323−333.
(6)
therapies. Discov. Med. 11, 106ꢀ114.
(7) Chamberlain, G. R., Tulumello D. V., and Kelley, S. O. (2013) Targeted delivery of
doxorubicin to mitochondria. ACS Chem. Biol. 8, 1389−1395.
(8) Han, M., Vakili, M. R., Abyaneh, H. S., Molavi, O., Lai, R., and Lavasanifar, A. (2004)
Smith, R. A. J., and Murphy, M. P. (2011) Mitochondriaꢀtargeted antioxidants as
Mitochondrial delivery of doxorubicin via triphenylphosphine modification for overcoming drug
resistance in MDAꢀMBꢀ435/DOX cells. Mol. Pharmaceutics 11, 2640−2649.
27
ACS Paragon Plus Environment

Bioconjugate Chemistry
Page 28 of 34
1
2
3
4
(9)
Yuan, L., Wang, L., Agrawalla, B. K., Park, S. –J., Zhu, H., Sivaraman, B., Peng, J., Xu
5
6
7
8
Q. –H., and Chang, Y. ꢀT. (2015) Development of targetable twoꢀphoton fluorescent probes to
image hypochlorous acid in mitochondria and lysosome in live cell and inflamed mouse model. J.
Am. Chem. Soc. 137, 5930−5938.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(10) Jung, H. S., Han, J., Lee, J. –H., Lee, J. H., Choi, J. –M., Kweon, H. –S., Han, J. H., Kim,
J. –H., Byun, K. M., Jung, J. H., Kang, C., and Kim, J. S. (2015) Enhanced NIR Radiationꢀ
Triggered Hyperthermia by Mitochondrial Targeting. J. Am. Chem. Soc. 137, 3017−3023
(11)
Smith, R. A., Hartley, R. C., Cochemé, H. M., and Murphy, M. P. (2012) Mitochondrial
pharmacology. Trends Pharmacol. Sci. 33, 341−352.
(12) Szeto, H. H., and Schiller P. W. (2011) Novel therapies targeting inner mitochondrial
membrane – from discovery to clinical development. Pharm. Res. 28, 2669ꢀ2679.
(13) Yousif, L. F., Stewart, K. M., and Kelley, S. O. (2009) Targeting mitochondria with
organelleꢀspecific compounds: strategies and applications. ChemBioChem 10, 1939ꢀ1950.
(14) Ross, M. F., Kelso, G. F., Blaikie, F. H., James, A. M., Cochemé, H. M., Filipovska, A.,
Da Ros, T., Hurd, T. R., Smith, R. A. J., and Murphy, M. P. (2005) Lipophilic
triphenylphosphonium cations as tools in mitochondrial bioenergetics and free radical biology
Biochem. (Moscow) 70, 273ꢀ283.
(15) Murphy, M. P. (2008) Targeting lipophilic cations to mitochondria. Biochim. Biophys.
Acta. 1777, 1028−1031.
(16) Murphy, M. P., and Smith, R. A. (2007) Targeting antioxidants to mitochondria by
conjugation to lipophilic cations. Annu. Rev. Pharmacol. Toxicol. 47, 629−656.
28
ACS Paragon Plus Environment

Page 29 of 34
Bioconjugate Chemistry
1
2
3
4
5
6
7
8
9
(17) AsinꢀCayuela, J., Manas, A. –R. B., James, A. M., Smith, R. A. J., and Murphy, M. P.
(2004) Fineꢀtuning the hydrophobicity of a mitochondriaꢀtargeted antioxidant. FEBS Lett. 571,
9ꢀ16.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(18) Ross, M. F., Prime, T. A., Abakumova, I., James, A. M., Porteous, C. M., Smith, R. A. J.,
and Murphy, M. P. (2008) Rapid and extensive uptake and activation of hydrophobic
triphenylphosphonium cations within cells. Biochem. J. 411, 633ꢀ645.
(19) Lu, J., Jiang, F., Lu, A., and Zhang, G. (2016) Linkers having a crucial role in antibodyꢀ
drug conjugates. Int. J. Mol. Sci. 17, 561.
(20) Ono, A., Miyauchi, S., Demura, M., Asakura, T., and Kamo, N. (1994) Activation
energy for permeation of phosphonium cations through phospholipid bilayer membrane Biochem.
33, 4312ꢀ4318.
(21) Agostinis, P., Berg, K., Cengel, K. A., Foster, T. H., Girotti, A. W., Gollnick, S. O.,
Hahn, S. M., Hamblin, M. R., Juzeniene, A., Kessel, D., Korꢀbelik, M., Moan, J., Mroz, P.,
Nowis, D., Piette, J., Wilson, B. C., and Golab, J. (2011) Photodynamic therapy of cancer: an
update CA Cancer J. Clin. 61, 250ꢀ281.
(22) Dolmans, D. E., Fukumura, D., and Jain, R. K. (2003) Photodynamic therapy for cancer
Nature Reviews Cancer 3, 380ꢀ387.
(23) Cuchelkar, V., Kopečková, P., and Kopeček, J. (2008) Novel HPMA copolymerꢀbound
constructs for combined tumor and mitochondrial targeting Mol. Pharmaceutics 5, 776ꢀ786.
(24) Zhang, C. –J., Hu, Q., Feng, G., Zhang, R., Yuan, Y., Lu, X., and Liu, B. (2015) Imageꢀ
guided combination chemotherapy and photodynamic therapy using a mitochondriaꢀtargeted
molecular probe with aggregationꢀinduced emission characteristics Chem. Sci. 6, 4580ꢀ4586.
29
ACS Paragon Plus Environment