Theoretical Study, Synthesis, and Reactivity of Five-Membered-Ring Acyl Sulfonium Cations

Article abstract of DOI:10.1002/ejoc.201500749

The feasibility of the cyclization of γ-alkylthiobutyric acid derivatives to form previously unknown five-membered-ring acyl sulfonium cations was studied. Experimental results were in good agreement with our DFT calculations that predicted such cyclizations to be easy if starting with acyl iodides and mixed anhydrides of triflic acid. Particularly efficient were the reactions of γ-alkylthiobutyryl fluorides with trimethylsilyl triflate in CDCl₃ solution, which led to cyclic acyl sulfonium triflates. In some cases, the observed acyl sulfonium salts were stable enough to be characterized by NMR spectroscopy. They were found to be both alkyl-transfer reagents and acylating agents. They react with amines to form amides. These findings lend some weight to our hypothesis that the acyl sulfonium derived from methionine may have played a major role in the prebiotic synthesis of the first peptides on the primitive Earth. The reaction of trimethylsilyl triflate (Me₃SiOTf) with γ-alkylthiobutyryl fluorides leads to previously unknown acyl sulfonium cations. These reactive species react both as alkyl-transfer reagents (e.g., with sulfides) and as acylating reagents (e.g., with amines). Such reactivity suggests that the acyl sulfonium derived from methionine may have been able to do the same in the primitive ocean.

Full text of DOI:10.1002/ejoc.201500749

SHORT COMMUNICATION
DOI: 10.1002/ejoc.201500749
Theoretical Study, Synthesis, and Reactivity of Five-Membered-Ring Acyl
Sulfonium Cations
K. V. Raghavendra Rao,[a] Nicolas Caiveau,[a] Rolf David,[a] Ibrahim Shalayel,[a]
Anne Milet,[a] and Yannick Vallée*[a]
Keywords: Prebiotic chemistry / Sulfur / Cations / Alkylation / Acylation
The feasibility of the cyclization of γ-alkylthiobutyric acid de-
rivatives to form previously unknown five-membered-ring
acyl sulfonium cations was studied. Experimental results
acyl sulfonium triflates. In some cases, the observed acyl
sulfonium salts were stable enough to be characterized by
NMR spectroscopy. They were found to be both alkyl-trans-
were in good agreement with our DFT calculations that pre- fer reagents and acylating agents. They react with amines to
dicted such cyclizations to be easy if starting with acyl iod- form amides. These findings lend some weight to our hypoth-
ides and mixed anhydrides of triflic acid. Particularly ef- esis that the acyl sulfonium derived from methionine may
ficient were the reactions of γ-alkylthiobutyryl fluorides with
trimethylsilyl triflate in CDCl3 solution, which led to cyclic
have played a major role in the prebiotic synthesis of the first
peptides on the primitive Earth.
tulated by Truce and Abraham[2] to explain the formation
of a thiolactone upon heating 4-alkylthiobutyryl chlorides
Introduction
In 1993, Jakubowski[1] reported that cyclic acyl sulfon- (i.e., the transient acyl sulfoniums were found to be readily
ium compound 1 was formed from methionine during dealkylated). In 1977, Minato et al.[3] reported that the S-
tRNA aminoacylation catalyzed by E. coli methionyl tRNA alkylation of an aromatic thiol ester gave a sulfonium cation
1
synthetase (editing process). He stated that 1 was stable in that was detected by H NMR spectroscopy. Later, very re-
acidic water and proposed that it could have been a possible active three- and four-membered cyclic acyl sulfoniums, ob-
methyl donor in the primitive ocean at a time when S- servable only at low temperature, were described by Le-
adenosyl-methionine and folate cycles did not exist, and bedev et al.[4] In 2005, Kozhevnikov et al.[5] reported the
thus it may have played an important prebiotic role. How- isolation of two stable solid acyl sulfonium chlorides, the
ever, the pieces of evidence presented in support of the exis- stability of which was probably due to the bulky aromatic
tence of 1 were rather weak and consisted of TLC spots groups surrounding their sulfonium function.
and the argument that the compound producing these spots
We were thus intrigued by the possible existence of com-
reversed to methionine after basic hydrolysis. No spectro- pound 1. Had it existed in the primitive ocean, it could have
scopic data of 1 were presented, and from an organic chem- played a role not only as a methyl donor (as proposed by
ist point of view, the realness of such an acyl sulfonium Jakubowsky[1]) but also as an activated amino acid, able to
cation could be regarded as questionable.
react with other amino acids to give dipeptides, in the ab-
sence of any catalyst.[6–8]
In fact, acyl sulfoniums are poorly known compounds,
and most of those that have already been described or pro-
posed as intermediates are very reactive. In 1963, the inter-
mediacy of a five-membered cyclic acyl sulfonium was pos-
[a] Département de Chimie Moléculaire, Univ. Grenoble Alpes,
CNRS,
Campus, 38041 Grenoble, France
E-mail: yannick.vallee@ujf-grenoble.fr
http://dcm.ujf-grenoble.fr
Supporting information for this article is available on the
WWW under http://dx.doi.org/10.1002/ejoc.201500749.
Scheme 1. Synthesis and cyclization of acyl halides derived from
acid 2. DAST = diethylaminosulfur trifluoride, TfO = CF3SO3.
Eur. J. Org. Chem. 2015, 6125–6129
© 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
6125

Y. Vallée et al.
SHORT COMMUNICATION
To demonstrate that five-membered-ring acyl sulfonium Table 1. Standard free-energy change (ΔG) and activation energy
(ΔG‡) for cyclization and demethylation reactions.
cations can be stable species, we study herein the synthesis
of simple analogues of 1, for example 4, and demonstrate
that they can be both alkylating and acylating reagents
(Scheme 1).
Reaction
ΔG[a] [kcalmol–1] ΔG‡[a] [kcalmol–1]
3a Ǟ 4-Cl–
5.5
2.7
–0.1
–9.6
–14.4
–0.3
12.9
10.9
10.2
9.7
3b Ǟ 4-Br–
3c Ǟ 4-I–
3d Ǟ 4-TfO–
4-I– Ǟ 5 + MeI
4-TfO– Ǟ 5 + MeOTf
16.7
30.1
Results and Discussion
[a] Solvent (continuum): acetonitrile (ε = 35.688).[10]
Our DFT calculations (see Table 1, Figure 1, and the
Supporting information) predicted that acyl chloride 3a and
bromide 3b should not undergo cyclization (in solution at
room temperature). In contrast, we expected iodide 3c to the reaction of fluoride 3e, obtained by the reaction of the
cyclize readily to form the iodide salt of sulfonium 4. How- corresponding acid with DAST,[13] with Me3SiOTf (even
ever, in this case easy dealkylation of 4 was predicted, and though the formation of Me3SiF was clearly demonstrated
the probable final products would then be methyl iodide by 1H NMR, 13C NMR, and 19F NMR spectroscopy).[14]
and thiolactone 5. Cyclization was also anticipated to be In place of the singlet at about δ = 3.15 ppm[3] expected
easy and quick with mixed anhydride 3d as the starting ma- for sulfonium 4, we observed a signal at δ = 2.94 ppm, not
terial. Furthermore, the stabilities of 4-triflate and 5 + compatible with structure 4, but rather with a trialkylated
MeOTf were predicted to be similar. As the activation en- sulfonium species that we were unable to identify
ergy for this last transformation was calculated to be quite [Me2S+(CH2)3C(O)X?] but probably indicating migration of
high, we anticipated that the triflate salt of 4 could be the methyl group of 4 onto a sulfide. Similar migrations will
stable.
be presented later in this communication.
These predictions were confirmed by our synthesis of
As 4 was not observed, we envisaged to replace the
compounds 3a–d from acid 2.[9] As expected, 3a and 3b methyl substituent of 2 by bigger groups (Scheme 2). Treat-
were found to be stable compounds, but iodide 3c ment of fluoride 6 with Me3SiOTf[15] resulted in the forma-
(13C NMR: C=O at δ = 161.7 ppm), observed after treat- tion of a new species. At room temperature, its H NMR
1
ment of 3a with TMSI (a method known to yield acyl iod- spectrum consists mainly of a series of broad signals, but at
ides from chlorides[11]), was unstable at room temperature –20 °C, the spectrum is clearer and indicates the presence
in CDCl3 solution and was transformed into methyl iodide of a stereogenic sulfur atom.
(13C NMR: δ = –22.5 ppm) and thiolactone 5.[12]
Both the chemical shifts and the splitting of the observed
However, even if we did not observe any trace amount signals are in agreement with structure 7 (Figure 2, a), ac-
of MeOTf, we were unable to identify 4 as a product of companied with some thiolactone 5. In the 13C NMR spec-
Figure 1. Energy diagram for the reaction pathway from 3c to 5 (cyclization and demethylation).
6126
www.eurjoc.org
© 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2015, 6125–6129

Five-Membered-Ring Acyl Sulfonium Cations
1
Figure 3. H NMR spectroscopy study of the formation of 9 from
8 with time and at various temperatures: (a) 8; (b) just after the
addition of TMSOTf; (c) at 255 K after 1 h; (d) at 223 K; (e) at
215 K.
Scheme 2. Synthesis and evolution of acyl sulfonium cations.
trum, the C=O carbon atom is observed at δ = 187.1 ppm.
However, 7 has only a half-life of about 2 h at room tem-
perature, and it is cleaved into 5 and cyclohexene.[16] Under
similar conditions, from 8, chosen because in this case no β-
elimination would be possible from the expected sulfonium
cation, a new species for which we assumed structure 9 was
4. Thus, 13a was found to be stable in CDCl3 solution. Even
methyl derivative 13b was clearly obtained (1H NMR: SMe
at δ = 3.15 ppm). However, it was accompanied by two
other products that we identified as thiolactone 15[18] and
sulfonium 14b (13C NMR: C=O at
δ = 156.4 ppm,
1
obtained. Even though its H NMR spectrum is not com-
1
19F NMR: δ = 27.0 ppm; H NMR: SMe2 at δ = 3.07 ppm,
CH2S at δ = 5.05 ppm). The final ratio (reached after 5 h
at room temp.) of 13b and 15 was found to be about 1:1.
This is an unambiguous example of a cyclic acyl sulfonium
cation acting as a methyl donor. Another example of this
behavior was observed if the reaction was run in the pres-
ence of dibutyl sulfide. In this case, 15 and 16b[19] were ob-
tained in a nearly 1:1 ratio (a trace amount of 14b was also
present).
pletely resolved, even at low temperature, chemical shifts
and observable splitting patterns (Figure 3) added to the
signals observed in its 13C NMR spectrum (C=O at δ =
187.6 ppm) made us confident that the main species in solu-
tion was 9. This was further ascertained by the synthesis of
spiro derivative 11 from 10.[17] Here, the structure is rigidi-
fied, and the lack of a β-CH2 moiety simplifies the spec-
trum. Even though some impurities were present, the three
AB systems expected for 11 are clearly observed at –20 °C
(Figure 2, b). Compounds 9 and 11 are stable at least for
days in CDCl3 solution at room temperature.
Similar results were obtained from 12c. However, benzyl
migration was quicker than the previous methyl migration.
Thus, 13c was not observed, and 14c and 15 were obtained
in a 1:1 ratio. If dibutyl sulfide was present in the reaction
mixture, 15 and 16c[20] were the major products (≈1:1 ratio,
with some amount of 14c).
Having demonstrated that acyl sulfonium cations are
alkyl-transfer reagents, we tested their ability also to be
acylating reagents. Indeed, the LUMO that we calculated
for 4 (Figure 4) had its biggest coefficient on the carbon
atom of the carbonyl group ring, a strong argument in favor
of easy ring-opening reactions by nucleophilic species. Rep-
resentative results with amines and amides are presented in
Scheme 3.[21] Alicyclic sulfonium cations were found to re-
act with both types of nucleophiles, but aromatic cation 13a
was less reactive and reacted only with the amino group of
phenylalanine amide. These results confirm our prediction
and assumption that if methionine derivative 1 was present
on the early Earth, it could not only be a methyl donor but
1
Figure 2. H NMR spectra of 7 and 11 at 253 K.
Further evidence for the existence of stable cyclic acyl also a good starting material for the synthesis of dipeptides.
sulfonium cations was obtained with aromatic analogues of However, the stability of 1 in water is still far from certain.[6]
Eur. J. Org. Chem. 2015, 6125–6129
© 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
6127

Y. Vallée et al.
SHORT COMMUNICATION
Scheme 3. Reactions of stable acyl sulfonium triflates with amines and amides.
[6] a) A. Milet, A. Gouron, R. David, H. Jamet, Y. Vallée, Ab-
stracts of Papers, 246th ACS Meeting, Indianapolis, 2013,
COMP-63., b) Y. Vallée, A. Milet, K. V. Raghavendra Rao,
Phosphorus Sulfur Silicon Relat. Elem. 2015, DOI: 10.1080/
10426507.2015.1034314.
[7] The start codon is quite variable in Escherichia coli. Not only
AUG (Met codon) but also UUG, GUG, and AUU are used.
This may indicate that Met was “chosen” before the start co-
don: a) F. R. Blattner, G. Plunkett III, C. A. Bloch, N. T.
Perna, V. Burland, M. Riley, J. Collado-Vides, J. D. Glasner,
C. K. Rode, G. F. Mayhew, J. Gregor, N. W. Davis, H. A. Kirk-
patrick, M. A. Goeden, D. J. Rose, B. Mau, Y. Shao, Science
1997, 277, 1453–1462; b) C. Sacerdot, G. Fayat, P. Dessen, M.
Springer, J. A. Plumbridge, M. Grunberg-Manago, S. Blan-
quet, EMBO J. 1982, 1, 311–315.
Figure 4. Calculated LUMO for cation 4. (Left) Side view;
(right) top view.
[8]
Among the many plausible scenarios previously proposed to
explain the prebiotic formation of peptides in the primitive
ocean, and more broadly the appearance of life, no one attri-
butes a specific role to methionine. See, for example: K. Ruiz-
Mirazo, C. Briones, A. de la Escosura, Chem. Rev. 2014, 114,
285–366; S. Mann, Angew. Chem. Int. Ed. 2013, 52, 155–162;
Angew. Chem. 2013, 125, 166–173; G. Danger, A. Michaut, M.
Bucchi, L. Boiteau, J. Canal, R. Plasson, P. Robert, Angew.
Chem. Int. Ed. 2013, 52, 611–614; Angew. Chem. 2013, 125,
639–642; U. Schreiber, O. Locker-Grutjen, C. Mayer, Origins
Life Evol. Biospheres 2012, 42, 47–54; M. W. Powner, J. D.
Sutherland, Philos. Trans. R. Soc. B 2011, 366, 2870–2877;
C. G. Kurland, Bioassays 2010, 32, 866–871; J. L. Bada, Earth
Planet. Sci. Lett. 2004, 226, 1–15; M.-C. Maurel, L. Orgel, Ori-
gins Life Evol. Biospheres 2000, 30, 423–430.
Acknowledgments
The authors thank the Laboratoires d’Excellence ARCANE
(ANR-11-LABX-0003-01) for funding through a KVRR grant,
L’Institut du développement et des ressources en informatique sci-
entifique (IDRIS) (project number i214086670), and CECIC and
CIMENT for providing access to computing facilities.
[1] a) H. Jakubowski, J. Biol. Chem. 1993, 268, 6549–6553; b) H.
Jakubowski, FASEB J. 1993, 7, 168–172.
[2] W. E. Truce, D. Abraham, J. Org. Chem. 1963, 28, 964–967.
[3] H. Minato, T. Miura, M. Kobayachi, Chem. Lett. 1977, 609–
614.
[4] M. V. Lebedev, V. G. Nenaidenko, K. I. Smolko, E. S. Balen-
kova, Russ. J. Org. Chem. 2001, 37, 326–333.
[5] V. N. Kozhevnikov, D. N. Kozhevnikov, O. V. Shabunina, N. N.
Kataeva, S. A. Yushchuk, V. L. Rusinov, O. N. Chupakhin,
Russ. Chem. Bull. Int. Ed. 2005, 54, 2187–2196.
[9]
K. A. Williams, J. T. Doi, W. K. Musker, J. Org. Chem. 1985,
50, 4–10.
[10]
As experimental reactions were performed in deuterated
chloroform, calculations with CHCl3 integral equation formal-
ism–polarizable continuum model (ε = 4.711) were also per-
formed for all reactions. The found activation energies and
6128
www.eurjoc.org
© 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2015, 6125–6129

Five-Membered-Ring Acyl Sulfonium Cations
complexes energies were similar to those calculated with aceto-
nitrile (within 5 kcalmol–1). However, separated charged spe-
cies (4, Cl–, Br–, I–, and TfO– at an infinite distance) were po-
orly described in implicit CHCl3, because the scheme is mostly
a function of the dielectric constant of the solvent, ε. Calcula-
tions in acetonitrile were kept for their accuracy.
[11] A. H. Schmidt, M. Russ, D. Grosse, Synthesis 1981, 216–218.
[12] D. Bhar, S. Chandrasekaran, Tetrahedron 1997, 53, 11835–
11842.
[13] W. J. Middleton, J. Org. Chem. 1975, 40, 574–578.
[14] All reactions involving triflates were run in CDCl3 solution and
were followed by NMR spectroscopy. No attempts at isolating
products were made.
[15] Similar results were obtained by using Et3SiOTf and iPr3SiOTf,
as well as from the corresponding acyl chloride by using Me3Si-
OTf and Et3SiOTf.
[16] After some time, cyclohexyl triflate was also detected, see:
S. S. C. Koch, A. R. Chamberlin, J. Org. Chem. 1993, 58, 2725–
2737.
[17]
Compound 10 was obtained in two steps from commercially
available 1-(mercaptomethyl)cyclopropaneacetic acid: (1) alkyl-
ation of thiol with neopentyltriflate, (2) fluorination with
DAST.
I. Ryu, T. Okuda, K. Nagahara, N. Kambe, M. Komatsu, N.
Sonoda, J. Org. Chem. 1997, 62, 7550–7551.
[18]
[19]
T. J. Wenzel, J. Zaia, Anal. Chem. 1987, 59, 562–567.
[20] D. J. Procter, M. Thornton-Pett, C. M. Rayner, Tetrahedron
1996, 52, 1841–1854.
[21] Crystals of 17a were obtained after recrystallization from eth-
anol. CCDC-1055644 contains the supplementary crystallo-
graphic data for this paper. These data can be obtained free of
charge from The Cambridge Crystallographic Data Centre via
www.ccdc.cam.ac.uk/data_request/cif.
Received: June 7, 2015
Published Online: August 20, 2015
Eur. J. Org. Chem. 2015, 6125–6129
© 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
6129