Rationally designed chemical modulators convert a bacterial channel protein into a pH-sensory valve

Article abstract of DOI:10.1002/anie.200503403

Open channels: Chemical modulators are developed to convert a naturally occurring channel protein into a pH-actuated nanovalve. The pH interval, sensitivity, and activation of the valve by channel opening are tunable through the design of the modulators.

Full text of DOI:10.1002/anie.200503403

Communications
Biosensors
DOI: 10.1002/anie.200503403
Rationally Designed Chemical Modulators
Convert a Bacterial Channel Protein into a
pH-Sensory Valve**
Arma gˇ an Koçer, Martin Walko, Erna Bulten,
Erik Halza, Ben L. Feringa,* and Wim Meijberg*
Small-molecule modulators are very promising tools for the
exploration and manipulation of biological systems beyond
the limits of genetics. The modern molecular biology toolkit
provides a variety of methods that aid in exploring and
understanding the structure and function of biological
molecules. However, these methods have limitations, espe-
cially in the range of changes and responses that they can
accomplish. In this regard, chemical modification provides a
complementary approach. Variations can be introduced to the
protein to confer features not achievable with the 20 amino
acids that are genetically encoded. This is especially true
when a combination of properties, such as reversibility,
tunability, target specificity, sensitivity to external stimuli,
and control over the timescale of the effect, are desired all at
the same time.
A particularly interesting property to control at the
molecular level is transport across barriers such as biological
membranes, as such control could easily lead to applications
in, for example, sensing and detection and drug delivery. This
effect has been pursued in a number of studies on the
construction of functional nanopores with either synthetic
molecules or naturally occurring channels, the latter mainly as
[
1]
b-barrel structures. Herein, we describe the rational design
[
+]
[
*] M. Walko, E. Halza, Prof. Dr. B. L. Feringa
Department ofOrganic and Molecular Inorganic Chemistry
Stratingh Institute and Material Science Center
University ofGroningen
Nijenborgh 4, 9747 AG, Groningen (The Netherlands)
Fax: (+31)50-363-4296
E-mail: b.l.feringa@rug.nl
[
+]
Dr. A. Koçer, E. Bulten, Dr. W. Meijberg
BiOMaDe Technology Foundation
Nijenborgh 4, 9747 AG, Groningen (The Netherlands)
Fax: (+31)50-363-4429
E-mail: meijberg@biomade.nl
+
[
] These authors contributed equally to this work.
[
**] We acknowledge Dr. R. Friesen for early contributions to the work,
Prof. Dr. G. T. Robillard and Prof. Dr. Bert Poolman for fruitful
discussions, C. M. Jeronimus-Stratingh and A. van Dam for the ESI-
MS analyses, and the Neurobiophysics Group at the University of
Groningen for patch-clamp facilities. This work was supported by
the Netherlands Organization for Scientific Research (NWO-CW;
B.L.F.), the Materials Science Center (MSC plus), University of
Groningen (M.W. and B.L.F), and Nanoned, a national nano-
technology program coordinated by the Dutch Ministry ofEconomic
Affairs.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
3
126
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 3126 –3130

Angewandte
Chemie
and engineering of a semisynthetic a-helical channel protein
by altering its intrinsic properties, such that it senses changes
in ambient pH and converts this signal into the opening of a
pore (Figure 1). To do so, chemical modulators were devel-
oped that allow tuning of the pH interval and sensitivity of the
response (that is, steepness of the transition). Furthermore,
the introduction of a photocleavable protecting group results
in a light-activated pH-sensitive valve.
sensitivity toward membrane tension but did not open
[
10]
spontaneously in response to pH. To achieve pH-induced
opening we modified the channel with specifically designed
pH-responsive chemical modulators. The pH-sensitivity inter-
val of the channel could be tuned by varying the pK and
a
hydrophobicity of the modulators. Further control over the
timing, location, and amplitude of the channel opening was
obtained through the use of caged modulators, in which the
pH modulator was protected by a photoremovable group.
Channels thus modified are inert to the environmental pH
until exposure to long-wavelength UV irradiation, at which
point the pH responsiveness is immediately regained. This
additional feature resulted in more flexibility and precision
with respect to external control over the channel opening and
the associated release of molecules from carrier devices.
To selectively attach the chemical modulators to the
[
11]
protein, glycine-22 was mutated into cysteine (G22C). This
change resulted in five identical modification sites within the
hydrophobic constriction zone of the homopentameric pro-
tein. The chemical modulators were designed as sulfhydryl-
reactive molecules (Table 1) and were covalently attached to
Figure 1. Protonation-induced opening ofa mechanosensitive channel,
which shows four of the five subunits in the pentamer (conceptual
drawing). Protonation ofa hydrophobic pore region ofthe MscL
results in conformational changes and channel opening.
Table 1: The structure and properties ofsynthetic pH modulators.
The protein we used is one of the best-studied channel
proteins to date, that is, the mechanosensitive channel of large
conductance (MscL) from Escherichia coli. In nature, this
[
a]
Compound
pKa
Mass
[2]
calcd
exptl
homopentameric protein is embedded in the cytoplasmic
[
3,4]
membrane
and protects the bacterial cell against severe
[14]
1
2
5.23
7.10
15786
15789
15860
[
5]
osmotic shocks. Under hypo-osmotic conditions, cell turgor
leads to tension in the membrane and an altered lateral-
pressure profile, which in turn triggers large conformational
changes. The result is a 3-nm nonselective pore, through
which molecules can flow to balance the osmotic difference
between the interior and exterior of the cell.
An important parameter that influences the gating
transitions of the MscL is the polarity of the hydrophobic
constriction zone. An increase in the polarity or hydro-
philicity of the amino acid residue 22, located in this part of
the protein, shifts the tension threshold for channel opening
to lower values. The introduction of a cysteine residue at
this position followed by chemical modification with charged
sulfhydryl-reactive compounds caused the channel to gate
[15]
[16]
15860
[
6]
[7]
[
8]
3
4
5
7.75
7.85
7.35
15828
15842
15856
15830
15841
15855
[
b]
[
9]
[b]
[
10]
[
a] The masses refer to the monomer of MscL-G22C covalently modified
with the corresponding compound. [b] The pK values were estimated on
a
[
17–19]
the basis ofsimilar molecules of und in the literature.
.
[
11]
spontaneously.
Recently, we used covalently attached
photosensitive modulators to induce charge at this position,
thus leading to a channel in which gating could be switched on
and off in the absence of any applied tension in response to
detergent-solubilized MscL-G22C. It is known that the
constriction zone of the protein is not fully accessible to
chemical modification when the MscL is in its natural
[
12]
radiation of an appropriate wavelength.
[
20]
Herein, we coupled charge-induced channel opening of
the MscL to the ambient pH. In vivo this parameter varies
depending on the health status of the surrounding tissue and/
or cellular compartment. Around solid tumors, sites of
inflammation, endosomes, and lysosomes the pH is lower
membrane environment. ESI-MS analyses (see Supporting
Information) indicate that in detergent-solubilized systems
under our conditions all five subunits of the protein are
modified simultaneously, and that the modification proceeds
specifically and quantitatively within the detection limits of
the mass spectrometry measurements. Our results are indica-
tive of a more relaxed conformation of the channel in
detergent solution compared to the natural membrane
environment. In this regard, it is interesting to recall the
arguments about whether the Mycobacterium tuberculosis
[
13]
(
(
pH 6.8 to 5) than under normal physiological conditions
pH 7.4). The incorporation of pH-controlled membrane
channels in drug-carrying liposomes will allow selective
release at diseased sites only. This might increase the
effectiveness of the drug-delivery device, which is our
ultimate goal. In previous studies it has been shown that a
glycine-22-to-histidine (G22H) mutant exhibited increased
[
21]
crystal structure, which was obtained in the presence of
detergent, represents a fully or partially closed conformation
Angew. Chem. Int. Ed. 2006, 45, 3126 –3130
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3127

Communications
[
6,22,23]
of the protein.
tation.
Our results support the latter interpre-
designed to have a higher pK value, within the physiological
a
range as well as the pH interval of the efflux assay. Channels
modified with 2 did indeed show pH-dependent activity in
efflux assays (Figure 2a), a result confirmed with patch-clamp
measurements (Figure 2b and c). Thus, we showed for the
first time that it is possible to open the channel in response to
pH in synthetic lipids without applying any tension, in both
patch-clamp and liposomal efflux setups.
The effects of chemical modification of MscL have been
followed under isoosmotic conditions in a straightforward
liposomal efflux assay, by using fluorescence changes upon
the release of a reporter (see Supporting Information).
Briefly, chemically modified channels were incorporated in
liposomes consisting of synthetic lipids by a detergent-
mediated reconstitution method in the presence of calcein,
a self-quenching fluorescent dye. After chromatographic
removal of external dye, the resulting proteoliposomes
[
24]
It is known that the gating behavior of MscL channels can
be correlated with the hydrophilicity of the residues present in
[
10]
the pore constriction. From this point of view, the pyridine
derivatives are not ideal modulators as they are quite
hydrophobic. To address this issue and improve the liposomal
efflux efficiency, pH modulators based on a different struc-
tural motif (3–5) were designed. We aimed specifically at
(
average size 200 nm, as determined by dynamic light
scattering) were analyzed for channel activity at different
pH values. Channel opening led to the release of the dye from
the liposomal interior, and the resulting concentration
decrease and dequenching upon entering the surrounding
buffer could be monitored as an increase in fluorescence.
Controls with liposomes lacking MscL-G22C, or containing
unmodified MscL-G22C or wild-type MscL (with the C-
terminal His tag that is present in all proteins described here),
did not show any release activity (see Supporting Informa-
tion).
pH-induced release through MscL modified with 1 could
not be observed by the assay described above between pH 6
and 8, the pH limits of the assay. To access lower pH ranges,
pyridine-modified channels were analyzed at the single-
molecule level by measuring the ionic current flowing through
the channel in response to the pH in patch-clamp experi-
ments. Spontaneous channel opening was observed at pH 5.2,
but at higher pH values the channels still required tension to
increased hydrophilicity, a pK value tunable in the range
a
pH 6–8, synthetic availability, and specific and efficient
coupling to the protein.
Proteoliposomes containing channels modified with mod-
ulators 3–5 responded to pH and released more calcein at
pH values below the pK of the modulator used. In contrast,
a
control liposomes without MscL were stable at all pH values
analyzed (Figure 3). The pH interval in which the channel is
activated is dependent on the pK and hydrophobicity of the
a
modulator. When the hydrophobicity of the pH modulator
was higher as a result of the additional methyl groups, the
modified channel was harder to open at higher pH values,
where the modulator is mainly in its uncharged form. A
particularly interesting case is the channel modified with 5
(Figure 3c), which has the highest hydrophobicity of the three
open (see Supporting Information). Modulator
2
was
modulators as well as the lowest pK . At pH 7.4, the
a
Figure 2. Activity ofMscL-G22C modi if ed with 2. Modified MscL was reconstituted in 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) liposomes
at a protein/lipid ratio of 1:120 (wt/wt) with a detergent-mediated reconstitution method. The fluorescent dye calcein (final concentration 50 mm)
was added to the lipid–protein mixture and the detergent was removed with biobeads. External calcein was removed by column chromatography
on Sephadex G50 and the liposomal fraction was used in the efflux assay. Depending on the concentration, about 10 mL ofliposomes were diluted
in 2 mL isoosmotic buffer (330 mosm) with varying pH values, and the increase in fluorescence was monitored for 45 min. Maximum release was
reached within 30 min and is reported here relative to the release obtained after bursting the liposomes with excess Triton X100. Error bars
indicate the standard deviation of two separate experiments. a) pH-induced release in calcein efflux assays. At pH 8, the proteoliposomes released
about 8% ofthe dye, but at pH 6 the release increased to 16%. Patch-clamp measurements at b) pH 5.8 and c) at pH 8. Insets: typical channel
openings in an enlarged form. The channel opened spontaneously without applied tension at pH 5.8. The observed conductance was about
2
.5 nS, but iftension was applied gradually, the channel also gated to its uf ll conductance of3 nS with as little as 15 mmHg applied negative
pressure. Conversely, at pH 8 spontaneous gating was absent and only the tension-dependent openings were observed. At lower tensions (up to
À50 mmHg pressure in the patch) the channel opened to lower subconducting states (1 and 2 nS), whereas at higher levels ofapplied tension
the uf ll conductance was reached. Further application oftension resulted in the opening ofmore than one channel.
3
128
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 3126 –3130

Angewandte
Chemie
[
9]
barrier. Our results, in particular the dramatic increase in
channel activity upon a small amount of suction, appear to be
in agreement with this model.
Further control over the pH-induced opening process was
gained by chemically protecting the pH-sensitive group of the
modulator. This allows activation on command. It was
achieved by coupling the amino group of modulator 4 to a
photolysable group that can be removed by irradiation with
long-wavelength UV light (6; Figure 4a). In the dark, the
Figure 3. pH-dependent activity of modified MscL in calcein efflux
assays. Modified MscL was reconstituted in DOPC/cholesterol/DSPE-
PEG2000 (70:20:10m) liposomes in the presence ofcalcein. The
activity of the channel at different pH values was monitored to
determine the maximum release values. Release was invariably com-
pleted within 30 min. Error bars indicate the standard deviation from
four independent experiments. MscL-G22C modified with a) 3, b) 4,
and c) 5. DSPE=1,2-distearoyl-sn-glycero-3-phosphoethanolamine;
PEG=polyethylene glycol.
Figure 4. MscL-G22C modified with a caged pH modulator. a) Struc-
ture of 6 and removal ofthe protecting group by illumination.
b) Activity of the modified MscL-G22C channels in the efflux assay.
Channels modified with caged modulator 6 did not open at any
pH value in the dark (dark gray bars). After deprotection by irradiation
À2
at 366 nm for 10 min (spectroline long-life filter, 365 mWcm ),
channels became activated and released the liposomal content in
dependence on the ambient pH (light gray bars). Ifthe pH ofthe
protein sample, which was illuminated at low pH, was immediately
raised to 7.7Æ0.1, the release ofdye stopped (white bar with
horizontal lines) because ofclosing ofthe channel; ifthe pH was kept
constant at a low value, the channel stayed open (white bar) and
release continued. Channels modified with uncaged modulator 4
opened in response to pH (black bar). Error bars indicate the standard
deviation from three independent experiments.
physiological pH for humans, liposomes with these channels
retain their content at levels comparable to those of liposomes
without protein, but start to open and release content at lower
pH values, such as those found at cancerous and inflammation
sites. The results were confirmed by patch-clamp measure-
ments (see Supporting Information). As in the case of
modification with 2, the channel opened to a subconducting
state in response to pH in the absence of tension, and to its full
conducting state after the application of a small amount of
tension. It has been reported that transition from the closed to
the subconducting state represents the opening of the main
gate of the channel and is associated with the highest energy
modified channel remains closed and consequently calcein
efflux experiments show no release of dye, irrespective of the
environmental pH (Figure 4b, dark gray bars). After removal
of the protecting group at the amine moiety by UV
irradiation, the channel becomes responsive again to the pH
of the environment, which results in high release of the
liposomal content at low pH (Figure 4b, light gray bars).
Angew. Chem. Int. Ed. 2006, 45, 3126 –3130
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3129

Communications
Comparison of the performance of the illuminated sample
desired location, time, and dosage, for example, in micro/
nanosensory and delivery devices.
with an uncaged sample at the same pH (Figure 4b, black
bars) indicated that the expected activity of 4 after removal of
its protecting group was recovered, and therefore that
photocleavage was complete.
The caged modulator was also used to address the
reversibility of the pH-induced opening of the channel. To
this end, duplicate samples were illuminated at pH 5.7.
Subsequently, the pH of one of the samples was raised to
Received: September 26, 2005
Revised: December 7, 2005
Published online: April 4, 2006
Keywords: biosensors · drug delivery · membrane proteins ·
.
protein modifications · protonation
7
.7. Although release started during illumination in both
samples, the process was arrested immediately after the pH
was raised to 7.7 (Figure 4b, white bar with horizontal lines),
while continuing at low pH (Figure 4b, white bar). This
experiment shows that the modified pH-sensitive channels
can be closed through a protonation reaction. Unfortunately,
the time course and sensitivity of the liposomal release
experiment do not allow this cycle to be repeated a number of
times with the same sample. Therefore, further investigation
is required to conclusively prove that the pH-induced opening
of the channel is reversible, that is, that the closed state at high
pH after removal of the cage is identical to the closed state
after a cycle of low and high pH.
The pH modulators described herein were able to confer
pH responsiveness to the channel, as shown from patch-clamp
and efflux measurements. However, the measurements also
show that complete release of enclosed calcein was not
achieved. Although this effect is not fully understood, a
number of explanations can be put forward. For example, the
reconstitution procedure does not guarantee that an active
channel is associated with every liposome. Furthermore, in
the case of the pH modulators, the ionization of one group
[
1] H. Bayley, O. Braha, S. Cheley, L. Gu in Nanobiotechnology
Eds.: C. M. Niemeyer, C. A. Mirkin), Wiley-VCH, Weinheim,
2004, pp. 93 – 112.
(
[2] S. I. Sukharev, P. Blount, B. Martinac, F. R. Blattner, C. Kung,
Nature 1994, 368, 265 – 268.
[
[
[
3] C. Berrier, A. Coulombe, C. Houssin, A. Ghazi, FEBS Lett.
989, 259, 27 – 32.
4] P. Blount, S. I. Sukharev, P. C. Moe, M. J. Schroeder, H. R. Guy,
C. Kung, EMBO J. 1996, 15, 4798 – 4805.
5] N. Levina, S. Totemeyer, N. R. Stokes, P. Louis, M. A. Jones, I. R.
Booth, EMBO J. 1999, 18, 1730 – 1737.
1
[6] E. Perozo, A. Kloda, D. M. Cortes, B. Martinac, Nat. Struct. Biol.
002, 9, 696 – 703.
2
[
[
[
7] C. C. Cruickshank, R. F. Minchin, A. C. Le Dain, B. Martinac,
Biophys. J. 1997, 73, 1925 – 1931.
8] B. Ajouz, C. Berrier, A. Garrigues, M. Besnard, A. Ghazi, J. Biol.
Chem. 1998, 273, 26670 – 26674.
9] A. Anishkin, C. S. Chiang, S. Sukharev, J. Gen. Physiol. 2005,
125, 155 – 170.
[10] K. Yoshimura, A. Batiza, M. Schroeder, P. Blount, C. Kung,
Biophys. J. 1999, 77, 1960 – 1972.
[
11] K. Yoshimura, A. Batiza, C. Kung, Biophys. J. 2001, 80, 2198 –
206.
2
[
12] A. Koçer, M. Walko, W. Meijberg, B. L. Feringa, Science 2005,
309, 755 – 758.
will depress the pK of its neighbors in close proximity in the
a
homopentameric protein. The accumulation of five positive
charges will therefore only be achieved at much lower
pH values, and hence the overall activity of the protein will
appear to be lower. This proposal was confirmed by showing
that modification of the protein with [2-(trimethylammo-
[13] K. Ulbrich, V. Sˇ ubr, Adv. Drug Delivery Rev. 2004, 56, 1023 –
1050.
[
14] A. Fischer, M. J. Kung, F. P. Robinson, Can. J. Chem. 1978, 56,
072– 3077.
3
[
15] E. C. Kornfeld, J. Am. Chem. Soc. 1955, 77, 4425 – 4427.
[
16] S. Glasstone, E. F. Hammel, J. Am. Chem. Soc. 1941, 63, 243 –
+
nio)ethyl]methanethiosulfonate (MTSET ), which carries a
248.
charge under all experimental conditions used here and thus
results in five positive charges in the channel constriction,
leads to maximum release values close to 100% (see
Supporting Information). Finally, it has been reported that
when the number of MscLs was low in the patches, the
[
17] J. Martin, R. Mendez, T. Alemany, J. Chem. Soc. Perkin Trans.
1989, 223 – 226.
[18] J. Hine, C. D. Fischer, J. Am. Chem. Soc. 1975, 97, 6513 – 6521.
[
[
[
19] J. Hine, M. S. Cholod, R. A. King, J. Am. Chem. Soc. 1974, 96,
35 – 845.
20] J. L. Bartlett, G. Levin, P. Blount, Proc. Natl. Acad. Sci. USA
004, 101, 10161 – 10165.
8
[
25]
channel activity was followed by irreversible inactivation.
2
A combination of the latter two effects, that is, reduced
activity and inactivation with time, will lead to release values
below 100%.
In summary, we have converted a bacterial channel
protein into a pH-sensitive valve. When embedded in lip-
osomes, the modified channels sense the ambient pH and
conditionally release the liposomal content. The sensitivity
and pH interval for channel opening were tuned by varying
21] G. Chang, R. H. Spencer, A. T. Lee, M. T. Barclay, D. C. Rees,
Science 1998, 282, 2220 – 2226.
[22] I. Iscla, G. Levin, R. Wray, R. Reynolds, P. Blount, Biophys. J.
2004, 87, 3172– 3180.
[
[
23] G. Levin, P. Blount, Biophys. J. 2004, 86, 2862 – 2870.
24] J. L. Rigaud, B. Pitard, D. Levy, Biochim. Biophys. Acta 1995,
1
231, 223 – 246.
25] C. C. Hꢀse, A. C. Le Dain, B. Martinac, J. Biol. Chem. 1995, 270,
8329 – 18334.
[
1
the hydrophobicity and pK of the modulators. Additional
a
control over pH response was gained through the introduc-
tion of a photosensitive caging group. Gaining control over
MscL channel activity as presented herein is valuable for
further investigations into the mechanism of the channel
function. The pH-activated channel or caged channels can
also serve as control valves to release or mix the contents at a
3
130
www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 3126 –3130