Planar-bilayer activities of linear oligoester bolaamphiphiles

Article abstract of DOI:10.3762/bjoc.7.184

Voltage-clamp experiments of eight oligoester bolaamphiphiles in two subclasses are described. Syntheses of three new terephthalate-based compounds were achieved in three linear steps. Together with five previously described, related compounds, the ion transport activity was assessed by means of the voltage-clamp technique. All of the compounds show multiple types of conductance behavior in planar bilayers, a subset of which was exponentially voltage-dependent. The varied and irregular activities were summarized with the aid of a recently developed "activity-grid" method.

Full text of DOI:10.3762/bjoc.7.184

Planar-bilayer activities
of linear oligoester bolaamphiphiles
Jonathan K. W. Chui, Thomas M. Fyles* and Horace Luong§
Full Research Paper
Open Access
Address:
Beilstein J. Org. Chem. 2011, 7, 1562–1569.
Department of Chemistry, University of Victoria, Victoria BC, Canada
doi:10.3762/bjoc.7.184
Email:
Received: 22 August 2011
Accepted: 26 September 2011
Published: 22 November 2011
Jonathan K. W. Chui - jkwchui@uvic.ca;
Thomas M. Fyles* - tmf@uvic.ca
* Corresponding author
This article is part of the Thematic Series "Supramolecular chemistry II".
Guest Editor: C. A. Schalley
§ Current address: Department of Chemistry, University of Manitoba,
Winnipeg MB, Canada
Keywords:
© 2011 Chui et al; licensee Beilstein-Institut.
License and terms: see end of document.
activity grid; single-molecule studies; synthetic ion channels; voltage
clamp experiment
Abstract
Voltage-clamp experiments of eight oligoester bolaamphiphiles in two subclasses are described. Syntheses of three new
terephthalate-based compounds were achieved in three linear steps. Together with five previously described, related compounds,
the ion transport activity was assessed by means of the voltage-clamp technique. All of the compounds show multiple types of
conductance behavior in planar bilayers, a subset of which was exponentially voltage-dependent. The varied and irregular activities
were summarized with the aid of a recently developed “activity-grid” method.
Introduction
Ion channels modulate conductance in cell membranes, and nents” of more complex systems. Since the pioneering report by
form the basic elements from which action potentials and other Tabushi [5] some three decades ago, hundreds of disparate syn-
complex electrophysiological behaviors are constructed [1]. thetic compounds have been shown to behave as ion channels,
With recent technological breakthroughs, such as stochastic by ensemble and/or single-molecule techniques [6,7]. Ensemble
sensing [2] and the construction of ionic networks [3,4], the techniques refer to the many variations of vesicle-based assays
control and understanding of ion-channel attributes at a single- [8]; the results obtained are comparable under identical condi-
molecule level takes on a new urgency. By being able to tailor tions, but often incommensurate with different variants of the
molecular features through synthesis, and potentially being method.
accessible in far larger quantity than membrane proteins,
de novo channels promise to be useful entities for The voltage-clamp technique, on the other hand, reports
structure–activity studies as well as practical “ionic compo- absolute flux under defined conditions [9]. In this method, a
1562

Beilstein J. Org. Chem. 2011, 7, 1562–1569.
planar bilayer is formed in an aperture between two electrically To provide a suitably focused yet sufficiently diverse suite of
connected aqueous compartments; this forms an insulator molecules in order to answer these questions, two classes of
between the electrodes, and no current flows despite an applied bolaamphiphiles were selected (Scheme 1). Both the phthalate
potential across the electrodes. Upon addition of the compound and solid-phase derived alkyl compounds derive from the
forming the active channel, a current on the order of pA can be de-macrocyclization of an earlier active channel [11]. Struc-
detected corresponding to the activity of single or small groups turally they similarly consist of a linear topology linked by
of molecules, and this current can be monitored with internal esters (Es); the classes differ by the presence of
millisecond time resolution or faster. Thus, this method can multiple aromatic rings in the phthalates, and the lack of
report not only the flux but also the structural dimensions and centrosymmetry in the alkyl compounds. These were expected
dynamic properties of the channel with high sensitivity; to be active in voltage-clamp experiments, as close analogues to
however, the voltage-clamp experiment suffers from a need for the phthalate compounds were found to be active in vesicles and
specialized equipment, as well as the daunting task of making planar bilayers [12], and the activities of the solid-phase com-
sense of all but the most regular of activities.
pounds in vesicle experiments were previously established [13].
Results and Discussion
We have recently developed an empirical classification scheme
for summarizing the range of voltage-clamp activities [10]. This Synthesis of new compounds
“activity grid” represents the conductance of an event as a The solid-phase synthesis of the alkyl oligoesters was previ-
vertical position on a logarithmic scale in half log-unit steps ously reported [14]. The phthalate compounds were prepared by
from 3 to 3000 pS, the duration of these events as horizontal a modular methodology (Scheme 2), beginning by reacting
position on a logarithmic scale from 10 ms to 100 sec, and the terephthaloyl chloride or compound 9 (prepared by a method
pattern of activity as a different color shading (green for adapted from [15]) with α,ω-bromoalcohols prepared by mono-
discrete “square-top” behavior, yellow for “flickering”, blue for brominating the corresponding diols as described [16]. This
discrete transitions with irregular opening, red for short dura- gave the bis-bromides, which were subsequently displaced by
tion spikes, purple for “erratic” activities from which no pattern the protected carboxylate anion of BOC-protected aniline 14 to
was discerned, and grey for regions of the grid where the give precursors 15, 16 and 17.
experimental setup precludes detection of events). A systematic
evaluation of published synthetic-ion-channel conductance Deprotection of the carbamate protecting groups under various
records revealed an underlying commonality over the wide acidic conditions also resulted in cleavage of internal esters in
range of structural types. However, published records are typi- 15. Selective cleavage was ultimately achieved by use of a
cally a brief excerpt of full activities and we wondered how the TMSCl–phenol reagent [17] in dichloromethane to give the
methodology would fare in the real world. How would the desired targets in three linear steps with modest overall yields.
method apply to full sets of unabridged records? And are there All three products were white solids that are freely soluble in
phenomena that are beyond the scope of this representation?
chloroform; compounds 2 and 3 are soluble in methanol
Scheme 1: Compounds studied with the voltage-clamp experiment.
1563

Beilstein J. Org. Chem. 2011, 7, 1562–1569.
Scheme 2: Synthesis of phthalate compounds.
whereas 1 is only sparingly soluble. All three compounds guaranteed; nonetheless, an analysis of different methods of
precipitate from water at micromolar concentration or above. compound introduction showed no substantial differences in
compound activity [18].
Voltage-clamp results
Voltage-clamp studies were performed as previously described Excerpts of experimental data are shown in Figure 1 and
[12]. The method of introducing the compounds is limited by Figure 2 for compounds 1 and 2, respectively. These two
their solubility: Water-soluble compounds (4–8) were generally figures establish that a great variety of activities coexist for
introduced as a solution to the bathing electrolyte, sparingly these two compounds, and that the regular “square top” behav-
soluble 2 and 3 by injection close to the proximity of the ior is a minority-type among many different types of irregular
bilayer, and the insoluble compound 1 was added by physical behavior. Complete experimental summaries are given in the
transfer from a brush. Physical transfer often results in Supporting Information File 1, from which it is clear that we
membrane rupture, and the integrity of the membrane cannot be could have made similar representations as Figure 1 and
Figure 1: Channel activities of compound 1. A: Regular “square-top” activity. Conditions: 1 M KCl buffered to pH 4.5, +200 mV; B: Large semiregular
activity. Conditions: 1 M CsCl unbuffered, +130 mV; C: Short-lived semiregular openings. Conditions: 1 M KCl unbuffered, +100 mV; D: Short-lived,
low-conductance bursts. Conditions: 1 M CsCl unbuffered, −50 mV; E: High-conductance bursts. Conditions: 1 M CsCl unbuffered, +120 mV; F:
Continuous erratic activity. Conditions: 1 M CsCl unbuffered, +160 mV.
1564

Beilstein J. Org. Chem. 2011, 7, 1562–1569.
Figure 2: Channel activities of compound 2. A: Rapid flickering. Conditions: 1 M CsCl, +100 mV; B: Semiregular openings. Conditions: 1 M CsCl,
+100 mV; C: Regular high-conductance bursts. Conditions: 1 M CsCl, +120 mV; D: Long-lived opening punctuated by regular bursts. Conditions: 1 M
CsCl, +100 mV; E: High-conductance, quasiregular activity. Conditions: 1 M CsCl, −120 mV; F: Abrupt closure following sustained bursts. Conditions:
1 M CsCl, pH 7, +120 mV; G: Short-lived, high-conductance bursts. Conditions: 1 M CsCl, pH 7, −120 mV; H: Brief spikes. Conditions: 1 M CsCl, pH
7, +180 mV; I: Long-lived regular opening, with conductance sublevels. Conditions: 1 M CsCl, pH 7, +170 mV.
Figure 2 with any other pair of compounds. Erratic behaviors current–time profiles, such as those given in Figure 1 and
may be aesthetically unappealing but should come as no Figure 2, into conductance–time profiles from the known
surprise: The hydrocarbon chains that are present in every com- applied potentials. A simple example is given in Figure 3
pound have significant conformational freedom; a variety of which, when viewed from the experimental current–time
aggregation states are accessible, and an anisotropic membrane perspective (top panel), shows (i) negative currents initially and
presents different local environments and other time-dependent positive currents over time, that (ii) events A and B are visually
degrees of freedom, which can be aptly collected under the of lower magnitude than events D–G, and (iii) events D–G are
umbrella term “dynamic supramolecular polymorphism” [8]. of decreasing magnitude. Correcting for changes in the applied
potential (middle panel) allows direct comparisons to be made,
A theoretical framework and tools for extracting mechanistic and it is clear from the bottom panel of Figure 3 that events A
information from the modeling of ideal “square-top” opening/ and B have a conductance at least twice as large as events D–G,
closing events is well-developed, and statistical approaches may and that the unit conductance of events D–G are in fact iden-
be appropriate when large numbers of events (> 104) are at hand tical. Simple cases such as this could be resolved by a skilled
[19]. The majority of the observations made in this study observer, but more complex cases require full conversion to a
conform to neither of these criteria, and we were unable to visu- conductance–time representation.
ally discern any obvious patterns; for this reason we turned to
the summary activity grids as a (potentially) model-inde- The second step involves a systematic inspection over several
pendent method for recording diverse behaviors.
orders of magnitude in time and conductance in order to recog-
nize and classify the conductance events that are encoded on the
Preparing the “activity grid” from empirical current–time traces activity grid of each record. A single experiment consists of
was facilitated by a suite of custom software, the development several records from a single day. Experiments were conducted
of which was described elsewhere [18]. There are two steps over a period of time, with varying conditions of electrolyte,
involved. The first involves the systematic conversion of sequence of applied potentials, etc. The entire dataset of activity
1565

Beilstein J. Org. Chem. 2011, 7, 1562–1569.
Figure 3: Conversion of a current–time profile of compound 5 (top) into conductance–time profile (bottom) by use of the applied potential (middle).
(Red bands indicate changes of the potential; the spikes appearing in these periods are the capacitive currents from the instrument. The conductance
is undefined when the applied potential is 0 mV, and thus this results in a break in the centre of the conductance–time profile.)
grids for a given compound was then summarized in a single set former. The second is that, despite being able to point out
of grids. This procedure answers the question: Was a particular empirical activity differences, these are almost impossible to
behavior of a particular conductance–duration ever observed? It rationalize in any simple way as the activities themselves are
does not provide information on the frequency of any particular multifacetted and do not admit simple explanations.
type of activity or conductance–duration. This results in an
activity profile for each compound, representative of the types In our systematic application of the “activity-grid” method to
of activity that were observed and their magnitude, from which published records we found that, from the great number of
a coarse-grained structure–activity can be mapped out unique architectures, there is an overall clustering characteristic
(Figure 4).
for the type of activity [10]. What we find here is preliminary
evidence that these relatively flexible molecules each individu-
The derived profiles are certainly consistent with our observa- ally also give rise to clusters of activity types. The extent of the
tions that almost every compound exhibits multiple types of clusters, as well as their magnitude in terms of duration and
activity. In this form it lacks a metric for the frequency of conductance, seem to be conserved between different linear
observation and is thus susceptible to both false positives (e.g., bolaamphiphiles and correspond to the influence of each unique
instrumental artifacts) as well as false negatives (i.e., absence of architecture on the overall system behavior. This is experi-
evidence is not evidence of absence). Nonetheless, the profiles mental support for the proposal [10] that there is some under-
reveal some interesting features when scrutinized through pair- lying energetic landscape that is as much a property of the
wise comparisons, and then from a “bird’s eye view”.
system (lipid, water, and compound) as it is a property of the
compounds we prepare and introduce.
Specific pairwise comparisons are revealing in two senses. The
first is that trends that were not obvious upon browsing through Voltage dependence
current–time traces stand out: For example, the comparison of Two types of voltage dependence have been reported previ-
the activity grids between the chain-reversed pair 5 and 8 shows ously. The first is a rectifying behavior, where ionic current
that they are capable of forming “square-tops” of similar dura- passes preferentially for one polarity of the applied potential in
tion and conductance, but the latter also has other modes of comparison to the other polarity [20,21]. The second is a
irregular or erratic activity that are simply not observed in the nonlinear current response to the applied potential, such as the
1566

Beilstein J. Org. Chem. 2011, 7, 1562–1569.
The current–time behavior is irregular, and therefore the current
was averaged over a 200 second interval. A plot of average
current as a function of applied potential appeared symmetrical
about zero, so the two branches for a given experiment were
combined as the average current as a function of the absolute
applied potential. Fits to a single exponential were adequate
given the erratic activity (r2 > 0.9), but both pre-exponential and
exponential terms varied considerably between different experi-
ments (Table 1). The exponent term of the fits can be used to
derive an apparent gating charge, corresponding in an idealized
model to the number of elementary charges required to move in
response to the applied potential [25]. The observed values lie
in the range from 1.2 to 3.5, with a cluster around 1.5. A physi-
cal interpretation of these values requires a detailed model, but
it is clear that several charged species, presumed to be the
protonated anilinium head groups, must move in response to the
applied potential through some significant portion of the bilayer
thickness. Neither the pre-exponential nor the exponent appears
to be a characteristic solely of the compound. Concentration of
the compound in the membrane presumably plays a large role,
but the general lack of suitable solubility precludes additional
studies.
Table 1: Voltage-dependent current (averaged over 200 s) for com-
pounds 1 and 2. A is the pre-exponential of the fitting, whereas b is the
exponent.
Compound
A
b
r2
Apparent gating
charge
1
1
1
1
2
2
2
0.0285 0.0645 0.9795 2.5
0.2062 0.0386 0.9671 1.5
0.4802 0.030
0.9702 1.2
0.0170 0.0422 0.9499 1.6
1.1919 0.0896 0.9199 3.5
0.0983 0.0382 0.9709 1.5
0.1655 0.032
0.9562 1.2
Qualitatively, both the open probability as well as unit conduc-
tance (of the somewhat regular “blue” multilevel openings)
appears to be enhanced at higher potentials. What could be the
underlying mechanism? One line of possibilities relates to the
underlying “system property” mentioned earlier. It is known
that pure lipids near their phase transitions can show single-
channel conductance activities; while the specific mode of
action depends on the identity of the lipid, discrete conductance
events indistinguishable from channels are possible [26]. It was
recently reported that an applied potential can effect lipid phase
Figure 4: Single-molecule structure–activity relationships for com-
pounds 1–8. Grids are given in the following sequence in each case:
Square-top (green); flicker (yellow); multiple-opening (blue); spike
(red); erratic (purple). Yellow arrows indicate hydrocarbon homologs;
green arrows indicate expansion of aromatic units; red arrows indicate
the removal of a polar head group; the beige arrow indicates the
reversal of chain polarity.
exponential dependence seen in alamethicin [22,23] or in a syn- transitions [27], one consequence of which is the creation of
thetic channel [24]. Two of the compounds in the terephthalate voltage-dependent lipid ion channels at the phase-transition
series, 1 and 2, exhibit voltage dependence in the second sense temperature. Pure diphytanoyl phosphatidylcholine is normally
(Figure 5).
stable to potentials exceeding 400 mV and does not show any
1567

Beilstein J. Org. Chem. 2011, 7, 1562–1569.
Figure 5: Voltage-dependent opening of compound 1. Inset: Exponential dependence of the mean current on the potential. The solid line fits the trace
shown in the figure.
detectible phase transitions [28,29]. It is conceivable that our loosely defined “system” comprising lipid, compound, water,
introduction of “impurities” creates local disorder akin to and electrolyte. A more focused investigation along these lines
domain separation at the melting transition and that fluctua- will be necessary to advance our understanding of how these
tions at the domain boundaries would give rise to voltage- simple channels “work”.
dependent channel activity.
The second conclusion concerns the “what compounds ought
Conclusion
we to think about” question. Many synthetic ion channels
Two subclasses of linear oligoester bolaamphiphiles, com- require heroic synthetic efforts, and one of the enduring motiva-
prising a total of eight compounds, were tested for conductance tions in various research groups is to find structures that are
activity in diphytanoyl phosphatidylcholine planar bilayers. All synthetically accessible such that structural variations are
compounds were active; most of them showed discrete “square- possible. This often translates into smaller “minimalist” channel
top” activity considered as a benchmark for the formation of molecules, and our suite of compounds was built along these
single-channels. One of these compounds 5 has “square-top” lines. This makes sense only if there is a defined and simple
activity as its predominant mode of activity, and two com- activity, such as the observed “square-top” events. In this study,
pounds in the terephthalate subclass show an exponential poten- we provided evidence that, although regular activities are not
tial dependence of the ionic currents. Aside from the potential- rare, their presence does not exclude other types of activity. The
dependent activity, the other types of conductance events can be frequent presence of other types of activity calls into question
documented in a standardized fashion by means of the “activity our ability to draw simple structure–function correlations.
grid” method. From these individual profiles a (sparse) struc- Rather, our attention should be focused on the entire range of
ture–activity map was constructed, and from this map two different activities, and the activity profiles give us a means to
conclusions were drawn.
do that. However, no obvious trend emerged from the pairwise
comparisons of this study. One possible reason is that, as com-
Firstly, the observations of clustered profiles match those that pounds become smaller, more individual molecules must come
emerged from our systematic literature survey [10]. Note together for active self-interactions or interactions with the
however, that the statistical support for this conclusion is very lipid, and site-specific changes are less able to target only a
weak due to the very small number of observations (fewer than single molecular parameter (rather than a confounding host of
200 on each side). Moreover, the irregular potential-depend- them). If this is true, then it suggests that “as small as possible”
ence of the ion channels observed is similar in characteristic to is not the optimal size for future investigations, and a larger
the voltage-dependent lipid-only channels. These two parallels study involving species with higher molecular weights may thus
suggest that these simple bolaamphiphile compounds are not lead to both more unique behaviors (with less dependence on
acting as channels on their own but rather in concert with a “system” properties) and deeper structure–activity insights.
1568

Beilstein J. Org. Chem. 2011, 7, 1562–1569.
21.Fyles, T. M.; Loock, D.; Zhou, X. J. Am. Chem. Soc. 1998, 120,
2997–3003. doi:10.1021/ja972648q
Supporting Information
22.Hall, J. E.; Vodyanoy, I.; Balasubramanian, T. M.; Marshall, G. R.
Biophys. J. 1984, 45, 233–247. doi:10.1016/S0006-3495(84)84151-X
23.Sansom, M. S. P. Eur. Biophys. J. 1993, 22, 105–124.
doi:10.1007/BF00196915
Supporting Information File 1
Synthesis procedures, spectroscopic characterization of
new compounds and voltage-clamp summary activity
records
24.Sakai, N.; Ni, C.; Bezrukov, S. M.; Matile, S. Bioorg. Med. Chem. Lett.
1998, 8, 2743–2746. doi:10.1016/S0960-894X(98)00481-8
25.Sakai, N.; Houdebert, D.; Matile, S. Chem.–Eur. J. 2003, 9, 223–232.
doi:10.1002/chem.200390016
[http://www.beilstein-journals.org/bjoc/content/
supplementary/1860-5397-7-184-S1.pdf]
26.Blicher, A.; Wodzinska, K.; Fidorra, M.; Winterhalter, M.; Heimburg, T.
Biophys. J. 2009, 96, 4581–4591. doi:10.1016/j.bpj.2009.01.062
27.Heimburg, T. Biophys. Chem. 2010, 150, 2–22.
doi:10.1016/j.bpc.2010.02.018
Acknowledgements
The ongoing support of the Natural Sciences and Engineering
28.Fyles, T. M.; Knoy, R.; Müllen, K.; Sieffert, M. Langmuir 2001, 17,
6669–6674. doi:10.1021/la0105937
Research Council of Canada is gratefully acknowledged.
29.Lindsey, H.; Petersen, N. O.; Chan, S. I. Biochim. Biophys. Acta 1979,
555, 147–167. doi:10.1016/0005-2736(79)90079-8
References
1. Hille, B. Ionic Channels of Excitable Membranes, 3rd ed.; Sinauer
Associates Incorporated: Sunderland, 2001; p 383.
2. Lu, S.; Li, W.-W.; Rotem, D.; Mikhailova, E.; Bayley, H. Nat. Chem.
2010, 2, 921–928. doi:10.1038/nchem.821
License and Terms
3. Hwang, H. L.; Chen, M.; Cronin, B.; Holden, M. A.; Bayley, H.
J. Am. Chem. Soc. 2008, 130, 5878–5879. doi:10.1021/ja802089s
4. Hwang, H. L.; Holden, M. A.; White, S.; Bayley, H. J. Am. Chem. Soc.
2007, 129, 11854–11864. doi:10.1021/ja074071a
This is an Open Access article under the terms of the
Creative Commons Attribution License
(http://creativecommons.org/licenses/by/2.0), which
permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
5. Tabushi, I.; Kuroda, Y.; Yokota, K. Tetrahedron Lett. 1982, 23,
4601–4604. doi:10.1016/S0040-4039(00)85664-6
6. Sisson, A. L.; Shah, M. R.; Bhosale, S.; Matile, S. Chem. Soc. Rev.
2006, 35, 1269–1286. doi:10.1039/b512423a
7. Matile, S.; Som, A.; Sordé, N. Tetrahedron 2004, 60, 6405–6435.
doi:10.1016/j.tet.2004.05.052
The license is subject to the Beilstein Journal of Organic
Chemistry terms and conditions:
8. Matile, S.; Sakai, N. The characterization of synthetic ion channels and
pores. In Analytical Methods in Supramolecular Chemistry;
Schalley, C. A., Ed.; Wiley-VCH: Weinheim, 2007; pp 381–418.
9. Fyles, T. M. Chem. Soc. Rev. 2007, 36, 335–347.
doi:10.1039/b603256g
(http://www.beilstein-journals.org/bjoc)
The definitive version of this article is the electronic one
which can be found at:
10.Chui, J. K. W.; Fyles, T. M. Chem. Soc. Rev. 2011.
doi:10.1039/c1cs15099e
doi:10.3762/bjoc.7.184
11.Fyles, T. M.; Loock, D.; van Straaten-Nijenhuis, W. F.; Zhou, X.
J. Org. Chem. 1996, 61, 8866–8874. doi:10.1021/jo961267c
12.Eggers, P. K.; Fyles, T. M.; Mitchell, K. D. D.; Sutherland, T.
J. Org. Chem. 2003, 68, 1050–1058. doi:10.1021/jo026415f
13.Luong, H.; Fyles, T. M. Org. Biomol. Chem. 2009, 7, 733–738.
doi:10.1039/b816649h
14.Luong, H.; Fyles, T. M. Org. Biomol. Chem. 2009, 7, 725–732.
doi:10.1039/b816648j
15.Mehdipour-Ataei, S.; Akbarian-Heizi, L. Eur. Polym. J. 2005, 41,
1280–1287. doi:10.1016/j.eurpolymj.2004.12.007
16.Chong, J. M.; Heuft, M. A.; Rabbat, P. J. Org. Chem. 2000, 65,
5837–5838. doi:10.1021/jo000291u
17.Kaiser, E.; Picart, F.; Kubiak, T.; Tam, J. P.; Merrifield, R. B.
J. Org. Chem. 1993, 58, 5167–5175. doi:10.1021/jo00071a028
18.Chui, J. K. W. A New Paradigm for Voltage-Clamp Studies of Synthetic
Ion Channels. Ph.D. Thesis, University of Victoria, Victoria, Aug 20,
2011.
19.Chui, J. K. W.; Fyles, T. M. Chem. Commun. 2010, 46, 4169–4171.
doi:10.1039/c0cc00366b
20.Goto, C.; Yamamura, M.; Satake, A.; Kobuke, Y. J. Am. Chem. Soc.
2001, 123, 12152–12159. doi:10.1021/ja010761h
1569