Discovery and optimization of potent and selective functional antagonists of the human adenosine A2B receptor

Article abstract of DOI:10.1016/j.bmcl.2009.08.040

We herein report the discovery of a novel class of antagonists of the human adenosine A2B receptor. This low molecular weight scaffold has been optimized to offer derivatives with potential utility for the alleviation of conditions associated with this receptor subtype, such as nociception, diabetes, asthma and COPD. Furthermore, preliminary pharmacokinetic analysis has revealed compounds with profiles suitable for either inhaled or systemic routes of administration.

Full text of DOI:10.1016/j.bmcl.2009.08.040

Bioorganic & Medicinal Chemistry Letters 19 (2009) 5945–5949
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery and optimization of potent and selective functional antagonists
of the human adenosine A2B receptor
Simon T. Bedford a, Karen R. Benwell a, Teresa Brooks a, Ijen Chen a, Mike Comer b, Sarah Dugdale b,
b
b
b
a
Tim Haymes b, Allan M. Jordan a, , Guy A. Kennett , Anthony R. Knight , Burkhard Klenke , Loic LeStrat ,
Angela Merrett b, Anil Misra b, Sean Lightowler b, Anthony Padfield b, Karine Poullennec b, Mark Reece b,
Heather Simmonite a, Melanie Wong a, Ian A. Yule b
*
a Vernalis (R&D) Ltd, Granta Park, Cambridge, CB21 6GB, UK
b Vernalis (R&D) Ltd, 613 Reading Road, Winnersh, Wokingham, RG41 5UA, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
We herein report the discovery of a novel class of antagonists of the human adenosine A2B receptor. This
low molecular weight scaffold has been optimized to offer derivatives with potential utility for the alle-
viation of conditions associated with this receptor subtype, such as nociception, diabetes, asthma and
COPD. Furthermore, preliminary pharmacokinetic analysis has revealed compounds with profiles suitable
for either inhaled or systemic routes of administration.
Received 17 July 2009
Revised 6 August 2009
Accepted 9 August 2009
Available online 14 August 2009
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
Adenosine receptor
Functional antagonist
Asthma
COPD
The human adenosine receptor is a G-protein coupled receptor
which is delineated into four subtypes, namely the A1, A2A, A2B and
A3 receptors. Whilst the therapeutic roles of modulators of the A1,
A2A and A3 receptor subtypes are comparatively well understood,1
the role of such compounds targeting the A2B receptor in human dis-
ease is less clear. The limited information obtained thus far indicates
that the receptor may play a role in diverse indications such as noci-
ception,2 diabetes,3 cancer4 and respiratory conditions such as asth-
ma and chronic obstructive pulmonary disease (COPD).5 However,
our present understanding of the true role of the A2B receptor is com-
plicated somewhat by the conflicting roles of the other adenosine
receptor subtypes in these conditions. This lack of understanding
is predominantly due to the paucity of low molecular weight aden-
osine A2B receptor antagonists which display appropriate functional
subtype selectivity and pharmacokinetic properties.
which recently completed a Phase I clinical trial as a potential
treatment for pulmonary disease.8
F
F
O
H
N
F
N
N
N
N
O
N
1
However, in our hands, this derivative suffered from a low vol-
ume of distribution and poor tissue penetration, limiting its appli-
cation as
a pharmacological tool. Alternate small molecule
derivatives with appropriate physicochemical properties are there-
fore much needed to fully elucidate the potential therapeutic util-
ity of this receptor subtype.
We recently described a series of thienopyrimidine derivatives
which demonstrated antagonistic activity at the human adenosine
A2A receptor. Some of these analogues, such as 2, demonstrated
moderate activity in models of Parkinson’s disease.9
Whilst this situation is gradually improving,6 there is a need for
compounds with appropriate properties which would allow a more
thorough understanding of the implications of selective adenosine
A2B receptor antagonism. The most advanced A2B antagonist in this
respect appears to be the xanthine-based derivative CVT-6883 1,7
O
S
S
N
* Corresponding author at present address: Paterson Institute for Cancer
Research, Wilmslow Road, Manchester, M20 4BX, UK. Tel.: +44 (0) 161 446 8079;
fax: +44 (0) 161 446 3109.
OH
N
H
N
2
E-mail address: ajordan@picr.man.ac.uk (A.M. Jordan).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.08.040

5946
S. T. Bedford et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5945–5949
In our routine binding protocols, many of these derivatives dis-
Table 2
Functional receptor antagonism data for C-4 ketoaryl derivatives11
played good subtype selectivity against the other adenosine recep-
tors, including the A2B receptor subtype. However, screening these
derivatives against the four human adenosine receptors in a fluoro-
metric calcium mobilization assay10 demonstrated that a small
subset displayed moderate functional antagonism of the adenosine
A2B receptor and, in contrast to their binding affinities, were
roughly equipotent when compared to their functional antagonism
of the A2A receptor. These derivatives generally contained a small
alkyl or alkylamino substituent at C-2 and to fully investigate the
utility of this effect at the A2B receptor, a series of derivatives were
prepared. Data from a selection of these compounds is shown in
Table 1.
O
Ar
N
S
N
N
N
H
Compound
Ar
A2B IC50 (nM)
A2A IC50 (nM)
12
20
21
22
23
24
25
26
27
28
29
30
5-Methyl-2-thienyl
5-Chloro-2-thienyl
5-Methyl-2-furyl
2-Thienyl
55
990
487
468
18
1541
1252
51
405
732
1715
1619
442
637
1168
1942
Small alkyl and alkylamino moieties gave only moderate activ-
ity against the A2B receptor and generally failed to yield notable
improvements in receptor subtype selectivity. The only notable
3-Thienyl
2-Thiazolyl
5-Thiazolyl
Phenyl
4-Methylphenyl
4-Methoxyphenyl
3-Pyridyl
148
176
148
511
215
1295
109
869
exception to this was the
a-methyl benzylamine derivative 10,
which gave a considerable increase in potency compared to the
parent 9, alongside a modest 2.5-fold selectivity over the A2A
receptor. However, whilst most benzylamine derivatives showed
a loss of potency (data not shown), the introduction of a more elec-
tron deficient pyridyl group gave a considerable increase in po-
tency and, in the case of 12, almost 20-fold functional selectivity
over the A2A receptor subtype. Both N-methylation (17) and con-
version of the pyridylmethylamine to the corresponding nicotin-
amide (14) were detrimental to activity, as was homologation of
4-Pyridyl
of potentially solubilizing moieties around the core thienopyrimi-
dine scaffold. Substitution of the 7-position of the thienopyrimi-
dine proved detrimental to functional antagonism of the A2B
receptor, whilst offering no improvement in receptor subtype
selectivity (data not shown). However, considerably better results
were observed with substitution at the adjacent 6-position (Table
3). Surprisingly in this case, the use of the 5-methyl-2-thieno moi-
ety at C-4, which had previously given a helpful increase in selec-
tivity against the A2A receptor, now had the opposite effect,
diminishing efficacy at the A2B receptor and restoring efficacy at
the A2A subtype. We therefore returned to the 2-thienyl moiety,
which restored both potency and some degree of selectivity. How-
ever, this data highlighted the delicate nature of the observed
selectivities between the two receptor subtypes.
Our findings demonstrated that only small substituents were
tolerated at the C-6 position. Interestingly, the introduction of ami-
no moieties not only gave a small increase in solubility but also
yielded a considerable increase in selectivity against the A2A recep-
tor, with the methylamino derivative 33 now displaying both sin-
gle-digit nanomolar antagonist efficacy and an almost 300-fold
difference in functional activity at the two receptor subtypes.
Whilst these findings were of interest, limited solubility was
still an issue with these derivatives. We believed this was attribut-
able to the high degree of crystallinity and planarity of these com-
the alkyl chain (15, 16). However, the combination of the a-methyl
substituent and the pyridyl moiety (to give 18 and 19) gave a use-
ful 2–3-fold increase in potency, albeit with a small reduction in
subtype selectivity.
In parallel with these studies, the nature of the aromatic moiety
at C-4 was also investigated. Data for representative derivatives are
given in Table 2. Whilst a variety of heterocyclic and phenyl ring
systems yielded moderate activity, only the 3-pyridyl derivative
29, the 5-thiazoloyl 25 and the 5-methyl-2-thienyl derivative 12
maintained some selectivity against the adenosine A2A receptor.
However, the derivatives described in Tables 1 and 2 suffered
somewhat with poor physicochemical properties, particularly
regarding their aqueous solubility, which hampered further stud-
ies. In an attempt to alleviate this, we investigated the placement
Table 1
Functional receptor antagonism data for initial thienopyrimidines11
O
S
S
N
N
R
Table 3
Functional receptor antagonism data for C-6 substituents11
Compound
R
A2B IC50 (nM)
A2A IC50 (nM)
3
4
5
6
–H
–Me
–Et
–NH2
5376
>10,000
3672
758
6685
7847
3852
220
O
S
7
8
9
–NHEt
–NHCH2-cyclohexyl
–NHBn
–NHCH(CH3)Ph
–NHCH2(2-pyridyl)
–NHCH2(3-pyridyl)
–NHCH2(6-pyrimidinyl)
–NHCO(3-pyridyl)
–NH(CH2)2(2-pyridyl)
–NH(CH2)2(3-pyridyl)
–NMeCH2(3-pyridyl)
–NHCH(Me)(2-pyridyl)
–NHCH(Me)(3-pyridyl)
1259
>10,000
5577
414
43
55
92
287
944
450
126
S
N
>10,000
7146
1077
186
990
116
161
351
177
2271
118
R
N
N
N
H
10
11
12
13
14
15
16
17
18
19
Compound
R
A2B IC50 (nM)
A2A IC50 (nM)
22
31
32
33
34
35
–H
–Cl
–NH2
–NHMe
–NHEt
–NMe2
18
16
32
3.5
19
39
51
62
>10,000
965
2485
1417
4142
19
17
99

S. T. Bedford et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5945–5949
5947
Table 4
and A3 receptor subtypes, with functional selectivities greater than
500-fold in both cases. The solubility of this compound offered a
reasonable pharmacokinetic profile after oral dosing in rats, with
a t1/2 of 2.3 h and an oral bio-availability of 84%. Whilst moderate
clearance was observed (2.6 L/h/kg), the volume of distribution
was reasonable, at 3.9 L/kg in rats. Furthermore, 44 displayed good
cellular permeability in a Caco-2 assay of 33 cm/s ꢀ 106 and mod-
erate (85%) binding to human plasma proteins. The compound also
Functional receptor antagonism data for 4-amide derivatives11
O
R
N
S
N
N
N
H
showed no adverse effects in a hERG patch clamp assay at 30
and a cell panel screen demonstrated no cellular toxicity at
40 M in a range of both normal and tumor cell lines. These toler-
ability studies were also supported by a micro-Ames study, which
lM
Compound
R
A2B IC50 (nM)
A2A IC50 (nM)
36
37
38
39
40
41
–NHMe
–NHEt
–NHiPr
–NHcPr
495
120
309
256
571
189
7107
2516
2610
710
3093
6136
l
observed no toxicity at 100 lM in two bacterial strains, rat liver
S9 fraction.13 Pleasingly, the compound also showed no notable ef-
fects in an Irwin behavioral study,14 with no evidence of A2A recep-
tor mediated effects (such as alteration of locomotor activity) after
oral dosing at 50 mg kgꢁ1 in mice.15
–NHcHexyl
Pyrrolidine
pounds, as evidenced by their high melting points.12 (For example,
33 has a melting point in excess of 240 °C.)
Although somewhat hindered by its limiting aqueous solubility,
33 demonstrated strong antagonism of the A2B receptor, though
selectivity against the other human adenosine receptors was
diminished compared to 44. Despite greater than 400-fold func-
tional selectivity against the A2A receptor, selectivities against the
A1 and A3 subtypes were somewhat diminished, at 85-fold and
20-fold, respectively.
A more detailed examination of the pharmacokinetics of 33 re-
vealed poor oral bio-availability of just 2% and a short half-life in
rats of 0.93 h after iv dosing. Though these properties preclude oral
dosing, when coupled with the high melting point of the solid, they
may make the compound appropriate for an inhaled route of
administration. Furthermore, given the compound displays a mod-
erate volume of distribution of 2.2 L/kg, high human plasma pro-
tein binding, moderate clearance of 1.7 L/h/kg and rapid
elimination, 33 would seem to be ideally suited for an inhaled
route of administration and would be expected to display minimal
levels of systemic exposure.
The synthesis of the compounds employed in this study is de-
scribed below.16,17 Compounds 3–5 were synthesized as previously
described.9 Derivatives 6–30 were prepared from the known
dichloropyrimidine 45.9 An Umpolung coupling of this core with
an aldehyde in the presence of N,N-dimethylimidazolium iodide18
gave the corresponding mono-chloro scaffolds which were con-
verted to the desired derivatives either by refluxing with the corre-
sponding amine or, in the case of the amide 14, via a Buchwald
coupling (Scheme 1).
In an effort to alleviate this issue, we investigated the effect of
disrupting the co-planar arrangement of the biaryl ketone system.
Alkyl, aminoalkyl and aminoacyl derivatives at C-4 all proved
highly detrimental to antagonistic efficacy (data not shown), indi-
cating a requirement for the keto-moiety at this position. However,
reversal of the amide moiety of the aminoacyl derivatives to main-
tain the position of the aryl carbonyl led to a series of moderately
potent and selective amide derivates. A selection of these amides
are shown in Table 4.
Whilst the N-ethyl derivative 37 displayed the best potency, the
pyrrolidine derivative 41 offered the best overall balance of po-
tency and selectivity, being 30-fold more efficacious at the A2B
receptor compared to the A2A receptor. More importantly, this sim-
ple replacement displayed a marked increase in measured aqueous
solubility, in excess of 500
to 14.5 M for compound 12 at pH 1.
Having previously observed a useful increase in selectivity upon
incorporating an -methyl moiety in earlier derivatives, we inves-
lM at both pH 1 and pH 6.8, compared
l
a
tigated whether the selectivity of 41 could be further improved in a
similar way. As detailed in Table 5, incorporation of this moiety to
give the racemic compound 42 gave a 15-fold increase in potency
and a corresponding sixfold increase in selectivity. Separation of
the enantiomers of 42 indicated that the (R)-isomer 43 was essen-
tially devoid of efficacy at the A2B receptor, with most activity
residing in the (S)-isomer 44 (Table 5).
Thienopyrimidines bearing a substituent at C-6 were prepared
in a similar way, from the trichlorothienopyrimidine scaffold 49
(Scheme 2). Nitration of dione 4 gave a mixture of 6- and 7-mono
nitrated products, which were readily separated. Treatment of 8
with PhPOCl2 not only introduced the 2,4-dichloro substitution
but simultaneously effected the ipso nitro-chloro exchange at the
C-6 position providing the trichlorinated scaffold 9 in good yield.
Exploiting the differential reactivity of the three chloro moieties al-
lowed stepwise installation of the keto-aryl moiety, followed by
introduction of the 3-picolylamine side chain at C-2 to give 31.
Given the interesting data obtained so far for both 44 and 33,
further studies were undertaken to assess in more detail the over-
all profile of these derivatives.
Though only displaying moderate 26-fold selectivity against the
A2A receptor, 44 demonstrated good selectivity against both the A1
Table 5
Functional receptor antagonism data for the optimization of the C-4 amide scaffold11
O
N
N
O
Ar
N
O
Ar
N
S
Cl
R
S
S
S
b or c
a
N
N
N
N
H
Cl
Cl
NRR'
N
N
N
45
6-30
Compound
R
A2B IC50 (nM)
A2A IC50 (nM)
41
42
43
44
H
189
13
3073
18
6136
2347
6760
468
Scheme 1. Reagents and conditions: (a) aryl-2-carboxaldehyde, N,N-dimethylim-
idazolium iodide, NaH, THF, reflux, 5 h, 33–83%; (b) (for compounds 3–13 and 15–
30) NHRR’, EtOH, reflux, 18 h, 8–80%; (c) (For amide compd 14), nicotinamide,
NaOtBu, Pd(OAc)2, xantphos, 1,4-dioxane, water, reflux, 24 h, 44%.
rac-Me
(R)-Me
(S)-Me

5948
S. T. Bedford et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5945–5949
O
O
S
CO2Me
a
S
S
b
NH
NH
O2N
NH2
N
H
47
O
N
H
48
O
46
c
S
N
S
N
O
Cl
O
f
d, e
N
S
S
N
S
Cl
R'RN
Cl
N
Cl
N
N
H
N
N
N
H
49
32-35
31
Scheme 2. Reagents and conditions: (a) urea, 180 °C, 5 h, 94%; (b) HNO3, H2SO4, rt, 20 min, 61% (as a 2:1 mixture of the 6- and 7-regioisomers); (c) PhPOCl2, 180 °C, 4 h, 45%;
(d) thiophene-2-carboxaldehyde, N,N-dimethylimidazolium iodide, NaH, THF, reflux, 3 h, 72%; (e) 3-picolylamine, n-BuOH, 100 °C, 2 h, 38%; (f) amine, DMA, sealed tube,
150 °C, 1 h, 10–68%.
EtO
EtO
O
N
Cl
against the other subtypes of adenosine receptors. Their differ-
ing physicochemical profiles offer the possibility of administra-
tion via inhalation or oral, systemic dosing and, as such, may
offer potential therapeutic options for the treatment of respira-
tory disorders such as asthma or COPD. Moreover, these agents
may offer useful pharmacological tools to further elucidate the
role of the adenosine A2B receptor in human disease
conditions.
b
a
S
S
S
N
N
N
Cl
N
Cl
N
Cl
45
50
51
c
O
NRR'
N
O
OMe
N
S
S
d
Supplementary data
N
N
H
N
N
N
H
N
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2009.08.040.
36-41
52
Scheme 3. Reagents and conditions: (a) (1-ethoxyvinyl)-tributylstannane, K2CO3,
PdCl2(PPh3)2, 1,4-dioxane, 100 °C, 30 min, 65%; (b) NaIO4, KMnO4, 1,4-dioxane,
water, 4 h, 54%; (c) (i) 3-picolylamine, NMP, 130 °C, 6 h; (ii) CHCl, 100 °C, 14 h; (iii)
SOCl2, MeOH, 60 °C, 2 h, 57% over three steps; (d) NHRR0, 1,4-dioxane, reflux, 1 h, 7–
88%.
References and notes
1. Jacobson, K. A.; Gao, Z.-G. Nat. Rev. Drug Disc. 2006, 5, 247.
2. Abo-Salem, O. M.; Hayallah, A. M.; Bilkei-Gorzo, A.; Filipek, B.; Zimmer, A.;
Mueller, C. E. J. Pharmacol. Exp. Ther. 2004, 308, 358.
3. Ruesing, D.; Mueller, C. E.; Verspohl, E. J. J. Pharm. Pharmacol. 2006, 58, 1639.
4. Panjehpour, M.; Karami-Tehrani, F. Oncol. Res. 2007, 16, 575.
5. Zhou, Y.; Schneider, D. J.; Blackburn, M. R. Pharmacol. Ther. 2009, 123, 105.
6. See, for example: Tabrizi, M. A.; Baraldi, P. G.; Preti, D.; Romagnoli, R.;
Saponaro, G.; Baraldi, S.; Moorman, A. R.; Zaid, A. N.; Varani, K.; Borea, P. A.
Bioorg. Med. Chem. 2008, 16, 2419; Baraldi, P. G.; Tabrizi, M. A.; Preti, D.; Bovero,
A.; Romagnoli, R.; Fruttarolo, F.; Zaid, N. A.; Moorman, A. R.; Varani, K.; Gessi,
S.; Merighi, S.; Borea, P. A. J. Med. Chem. 2004, 47, 1434.
More forcing conditions were then required to install the desired
functionality at the C-6 position to yield derivatives 32–35.
Preparation of the C-4 amide derivatives 36–44 was somewhat
less facile. A Stille coupling introduced an ethoxyvinyl moiety at
C-4 to yield 50, which could be oxidized to the corresponding es-
ter 51 under modified Lemieux conditions with catalytic potas-
sium permanganate and sodium periodate as re-oxidant.19
Introduction of the 3-picolylamine side chain at C-2 gave con-
comitant, partial formation of the amide at C-4, via displacement
of ethoxide. This necessitated hydrolysis of the amide and re-
esterification of the resultant acid with thionyl chloride and
methanol to give 52. Though slightly inefficient, efforts to circum-
vent this issue were unsuccessful. However, this process could be
performed without intermediate purification and proceeded in
reasonable overall yield. Finally, formation of the C-4 amides by
refluxing with the appropriate amines yielded the desired deriva-
tives 36–41.
The racemic derivative 42 and the single enantiomers 43 and 44
described in Table 5 were also prepared as described in Scheme 3,
employing racemic 2-(3-pyridyl)ethylamine in step (c), followed
by resolution by chiral HPLC. The stereochemical identity of each
isomer was later confirmed by repeating the synthesis on a small
scale using the known enantiopure amines.20
In summary, we report herein the discovery of a novel class
of adenosine A2B antagonists. These low molecular weight com-
pounds display promising functional activity and selectivity
7. Elzein, E.; Kalla, R. V.; Li, X.; Perry, T.; Gimbel, A.; Zeng, D.; Lustig, D.; Leung, K.;
Zablocki, J. J. Med. Chem. 2008, 51, 2267.
8. Harrison, C. Nat. Rev. Drug Disc. 2006, 5, 820.
9. Gillespie, R. J.; Cliffe, I. A.; Dawson, C. E.; Dourish, C. T.; Gaur, S.; Giles, P. R.;
Jordan, A. M.; Knight, A. R.; Lawrence, A.; Lerpiniere, J.; Misra, A.; Pratt, R. M.;
Todd, R. S.; Upton, R.; Weiss, S. M.; Williamson, D. S. Bioorg. Med. Chem. Lett.
2008, 18, 2916.
10. For more details, see: Gillespie, R. J.; Bamford, S. J.; Botting, R.; Comer, M.;
Denny, S.; Gaur, S.; Griffin, M.; Jordan, A. M.; Knight, A. R.; Lerpiniere, J.;
Leonardi, S.; Lightowler, S.; McAteer, S.; Merrett, A.; Misra, A.; Padfield, A.;
Reece, M.; Saadi, M.; Selwood, D. L.; Stratton, G. C.; Surry, D.; Todd, R.; Tong, X.;
Ruston, V.; Upton, R.; Weiss, S. M. J. Med. Chem. 2009, 52, 33. and the
supplementary data.
11. Values are the geometric mean of at least two independent determinations and
standard deviations for these values were generally within 30% of the mean
value.
12. Yang, G.; Ran, Y.; Yalkowsky, S. H. J. Pharm. Sci. 2002, 91, 517; Delaney, J. S. Drug
Discovery Today 2005, 10, 289.
13. CEREP mini-Ames, using 2 Salmonella strains, TA98 and TA100. For details, see
http://www.cerep.fr.
14. Irwin, S. Psychopharmacologia 1968, 13, 222.
15. Pharmacokinetic analysis after oral dosing at 2, 5, 30 and 50 mg/kg
demonstrated a generally linear increase in exposure with increasing dose,
as determined by comparison of the AUClast
.
16. All compounds gave satisfactory 1H NMR and LC–MS data in full agreement
with their proposed structures and were >90% purity, as determined by

S. T. Bedford et al. / Bioorg. Med. Chem. Lett. 19 (2009) 5945–5949
5949
examination of both the LC chromatograph (wavelength 235 nm) and the 1H
NMR spectrum.
18. Miyashita, A.; Obae, K.; Suzuki, Y.; Oishi, E.; Iwamoto, K.; Higashino, T.
Heterocycles 1997, 45, 2159.
17. Jordan, A.; Bedford, S.; Klenke, B.; Yule, I.; Poullennec, K.; PCT Int. Appl
WO2009037463, 2009; Jordan, A.; Bedford, S.; Klenke, B.; Yule, I.; Poullennec,
K.; PCT Int. Appl WO2009037468, 2009.
19. Lemieux, R. U.; von Rudloff, E. Can. J. Chem. 1955, 33, 1710.
20. Smith, H. E.; Schaad, L. J.; Banks, B. R.; Wiant, C. J.; Jordan, C. F. J. Am. Chem. Soc.
1973, 95, 811.