Suppression of racemization in the carbonylation of amino acid-derived aryl triflates

Article abstract of DOI:10.1016/j.tetlet.2007.04.145

The carbonylation of enantiopure phenylglycine-derived aryl triflates was achieved to afford 4-carboxyphenylglycine analogs with high enantiomeric excesses (88 to >99% ee). Amide analogs of phenylglycine were well-tolerated in the hydroxy- and methoxycarbonylation processes, providing efficient access to benzoic acid and ester building blocks. The % ee of the product was dependent on the relative steric bulk of both the amino acid substrate and the requisite amine base, with ⁱPr₂NEt proving optimal in minimizing product racemization.

Full text of DOI:10.1016/j.tetlet.2007.04.145

Tetrahedron Letters 48 (2007) 4509–4513
Suppression of racemization in the carbonylation of
amino acid-derived aryl triflates
Jonathan B. Grimm,* Kevin J. Wilson and David J. Witter
Department of Drug Design and Optimization, Merck Research Laboratories, Boston, MA 02115, USA
Received 25 October 2006; revised 27 April 2007; accepted 30 April 2007
Available online 5 May 2007
Abstract—The carbonylation of enantiopure phenylglycine-derived aryl triflates was achieved to afford 4-carboxyphenylglycine ana-
logs with high enantiomeric excesses (88 to >99% ee). Amide analogs of phenylglycine were well-tolerated in the hydroxy- and meth-
oxycarbonylation processes, providing efficient access to benzoic acid and ester building blocks. The % ee of the product was
i
dependent on the relative steric bulk of both the amino acid substrate and the requisite amine base, with Pr2NEt proving optimal
in minimizing product racemization.
ꢀ 2007 Elsevier Ltd. All rights reserved.
The palladium-catalyzed carbonylation of aryl triflates
was first published 20 years ago1 and has since become
a common strategy to access benzoate derivatives from
phenols. The scope of this reaction includes the carbon-
ylation of triflates derived from tyrosine analogs2
(Scheme 1), and these conditions can successfully con-
vert an enantiomerically pure tyrosine triflate 4 into
the corresponding 4-carboxyphenylalanine derivative 6
with complete retention of enantiopurity. With the
accessibility of D- and L-tyrosine ((R)-2 and (S)-2), this
method represents a useful approach to enantiopure 4-
carboxyphenylalanines and has been used in the synthe-
sis of molecules with various biological properties3 as
well as for large scale preparation.4
During the course of an ongoing medicinal chemistry
program, we required access to various 4-carboxy-
phenylglycine intermediates 5 as single enantiomers.
Numerous methods for producing enantiopure phenyl-
glycine derivatives have been reported, including a
plethora of biocatalyzed processes5 and more traditional
synthetic organic approaches.6,7 However, since D- and
L-4-hydroxyphenylglycine ((R)-1 and (S)-1) are readily
available commercial products, it was hoped that the
target compounds might be accessed via a carbonylation
approach employing aryl triflate derivatives (3, Scheme
1). This route represents a more direct protocol than
existing methods, as it avoids the addition and removal
of chiral auxiliaries and reduces the number of protect-
ing group manipulations. It was anticipated that the
application of this methodology to phenylglycine ana-
logs might be complicated by racemization resulting
from the acidity of the benzylic proton. Nonetheless,
the required triflates were synthesized according to
Scheme 2 and were obtained in high enantiomeric excess
(>99% ee in most cases). Firstly, the amino acids were
protected as N-Boc intermediates8 7 and 8. Functional-
ization of the acids to the required amides 9a–e and 10b
was achieved using EDC and HOBt. The enantiomers of
methyl ester 9f were produced by methylation with
TMS–diazomethane. Facile conversion of the phenols
to the required triflates 11a–f and 12b using PhNTf2
completed the synthesis.
Scheme 1. Carbonylation route to 4-carboxyphenylalanine and 4-
carboxyphenylglycine derivatives.
Keywords: Carbonylation; Carboxyphenylglycine; Aryl triflates; Steric
effect; Amino acids.
*
Corresponding author. Tel.: +1 617 992 2369; fax: +1 617 992
2403; e-mail: jonathan_grimm@merck.com
0040-4039/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2007.04.145

4510
J. B. Grimm et al. / Tetrahedron Letters 48 (2007) 4509–4513
Table 2. Screening of hydroxycarbonylation conditions
Entry Amine base
Solvent system
Yielda eeb
(%)
(%)
1
2
3
Et3N
Et3N
Et3N
1:1 DMF:H2O
3:1 DMF:H2O
10:1 DMF:H2O
87
99
98
80
90
80
4
5
6
7
8
9
10
11
12
Et3N
3:1 DMSO:H2O 96
88
96
28
66
84
90
>99
98
N/A
Et3Nc
3:1 DMF:H2O
3:1 DMF:H2O
3:1 DMF:H2O
3:1 DMF:H2O
3:1 DMF:H2O
3:1 DMF:H2O
3:1 DMF:H2O
3:1 DMF:H2O
77
99
98
73
99
97
35
<5
DABCO
N-Methylpiperidine
N-Methylmorpholine
iPr2NH
iPr2NEt
2,6-Lutidine
DBU
Scheme 2. Reagents and conditions: (i) Boc2O, 1,4-dioxane, 1 N
NaOH, rt; (ii) R = NR1R2: amine, EDC, HOBt, DMF, rt or
i
amineÆHCl, EDC, HOBt, Pr2NEt, DMF, rt; (iii) R = OMe: TMS–
diazomethane, THF, EtOH, rt; (iv) PhNTf2, DCM, rt.
a Reactions stirred at 70 ꢁC until complete (or progression ceased) by
LC/MS (12–18 h).
N-Methyl amide (R)-11a was selected as a model sub-
strate to investigate the carbonylation reaction. Employ-
ment of a number of existing literature conditions gave
high yields but variable levels of racemization (Table 1).
Although the Pd(OAc)2/dppf/KOAc system of Cacchi
and Lupi9 afforded a high yield of acid (R)-13a, signifi-
cant racemization occurred during the reaction (entry
1).10,11 Exclusion of KOAc from the reaction (entry 2)
resulted in an increased ee, but a decreased yield. A
higher yield and an improved ee resulted from substitu-
tion of K2CO3 for KOAc and DMF for DMSO (entry
3).4 Similar results were obtained by using Et3N with
Pd(OAc)2/dppp in DMF/H2O (entry 4, 90% ee).2,3d
Given the unacceptable level of racemization still occur-
ring during the carbonylation, we undertook an investi-
gation to minimize racemization and maximize yields
for amino acid-derived substrates similar to 11a.
b Determined by chiral HPLC after conversion of the acids to the
methyl esters with trimethylsilyldiazomethane in THF/MeOH.
c Using only 1 equiv of base.
from the 2:1 ratio shown in Table 2 offered no apprecia-
ble advantage in yield, ee, or reaction time. Alteration of
the DMF:H2O solvent ratio had a noticeable effect on
the racemization level. A 3:1 ratio (entry 2) was found
to be ideal; decreasing or increasing this ratio adversely
affected the ee of product 13a (entries 1 and 3, 80% ee).
The use of DMSO (entry 4) in place of DMF had little
effect on the yield or ee of the carboxylic acid. Not sur-
prisingly, decreasing the amount of Et3N from 2 equiv
to 1 equiv resulted in an increase in ee from 90% to
96% (entry 5), but the reaction failed to reach comple-
tion (77% yield).12
Using the initial Pd(OAc)2/dppp/Et3N conditions as a
starting point and (R)-11a as the substrate, we per-
formed a brief screen of solvents, ligand/catalyst ratios,
and bases (Table 2). The reactions were performed at
70 ꢁC under a CO balloon until complete (or conversion
ceased) by LC/MS. In line with the literature prece-
dent,2,3,9 dppp was found to be the superior ligand with
regards to yield and conversion, though dppf was a com-
parable substitute. Changes in the ligand:catalyst ratio
The most significant effect on the racemization levels
was seen upon changing the base (Table 2, entries
6–12). In our screening of selected amines, DABCO,
N-methylpiperidine, and N-methylmorpholine (entries
6–8) all resulted in lower enantiomeric excesses than
i
Et3N. While Pr2NH was comparable to Et3N in yield
i
and ee (entry 9), Pr2NEt was vastly superior, affording
(R)-13a in >99% ee (entry 10). The employment of
Table 1. Hydroxycarbonylation of triflate (R)-11a
Entry
Conditionsa
Yield (%)
eeb (%)
1
2
3
4
Pd(OAc)2 (0.05 equiv), dppf (0.20 equiv), KOAc (3 equiv), DMSO, 60 ꢁC
Pd(OAc)2 (0.05 equiv), dppf (0.20 equiv), DMSO, 60 ꢁC
Pd(OAc)2 (0.10 equiv), dppf (0.20 equiv), K2CO3 (5 equiv), DMF, 60 ꢁC
Pd(OAc)2 (0.03 equiv), dppp (0.06 equiv), Et3N (2 equiv), DMF/H2O, 70 ꢁC
93
26
98
99
40
98
90
90
a Reactions stirred at indicated temperature until complete by LC/MS (16–24 h).
b Determined by chiral HPLC after conversion of the acids to the methyl esters with trimethylsilyldiazomethane in THF/MeOH.

J. B. Grimm et al. / Tetrahedron Letters 48 (2007) 4509–4513
4511
2,6-lutidine (entry 11) or DBU (entry 12) resulted in
poor conversion, with the latter yielding only trace
product. Examination of the relative amine basicities
indicated that there was essentially no correlation
between amine pKa and erosion of product enantio-
purity. For example, N-methylpiperidine and DABCO
are both less basic than Et3N, yet both racemized the
product to a greater extent than Et3N.13 This suggested
that the degree of racemization was largely determined
by the steric bulk surrounding the amine nitrogen. This
notion was supported by the excellent result obtained
with the bulky iPr2NEt, as well as the poor enantiopuri-
ties achieved with less hindered amines, for example,
N-methylpiperidine and DABCO.
(entry 2). Furthermore, the carboxylic acid products
were universally isolated in high purity without chroma-
tography through an aqueous base extraction and acid-
ification protocol. In contrast to the amide analogs, the
acid products afforded by carbonylation of phenyl-
glycine ester substrates were obtained as racemic mix-
tures (e.g., entries 14 and 15). It was not clear whether
this resulted from the lower pKa of the ester a-proton
or from the small size of the methyl ester moiety. A
future experiment involving larger ester groups (e.g.,
t
CO2 Bu) in place of the methyl ester could provide more
insight into the precise source of this effect.
In a preliminary effort to understand the mechanism of
racemization, hydroxycarbonylations of selected sub-
strates were performed in the presence of D2O. Carbon-
ylation of triflate (R)-11b with Et3N and D2O (Scheme
3) gave acid 15 (40% ee) with 60% deuterium incorpora-
tion at the benzylic position. This suggested that the
transient enolate intermediate was reprotonated by sol-
vent and not immediately quenched by an associated
ammonium species (i.e., the protonated base). Likewise,
D2O hydroxycarbonylation of (R)-11f afforded 16 as a
racemate with complete deuterium incorporation.
With these results in mind, the optimized hydroxycarb-
onylation conditions were applied to a variety of phenyl-
glycine- and phenylalanine-derived aryl triflates (Table
3).14 For the amide-containing phenylglycine triflates
(entries 1–11), the corresponding acids were uniformly
obtained in nearly quantitative yields and consistently
i
high enantiopurities when Pr2NEt was employed as
the base. Tertiary, secondary, and primary amides were
well-tolerated in the reaction. Notably, the primary
amide (R)-11b displayed an enhanced disparity in enan-
tiopurity between the Et3N (entry 3, only 28%) and iPr2-
NEt (entry 4, 92%) runs. Phenylalanine analogs (R)- and
(S)-12b (entries 12–13) were also smoothly converted to
carboxylic acids with excellent results (>90% yield,
>99% ee). The reaction was also found to be highly scal-
able, as demonstrated by the example performed on a
5 g scale with no appreciable decrease in yield or ee
One of the many benefits of the palladium-catalyzed car-
bonylation of aryl triflates is the ability to conduct the
reaction in the presence of an alcohol (instead of water)
to directly form a benzoate ester.2 Not surprisingly, the
same base-dependent racemization effect existed in the
methoxycarbonylation of the phenylglycine-derived aryl
triflates to form methyl benzoates. Methoxycarbonyl-
i
ation of these substrates with Pr2NEt gave, upon flash
chromatography, the desired methyl esters in excellent
yields and good to excellent enantiomeric excesses
(Table 4).15 For the sake of comparison, methoxycarb-
onylations of three of the substrates were also per-
formed with the less hindered Et3N. For the primary
amide (R)-11b (entry 1) and methyl amide (R)-11a (entry
3), high levels of racemization were seen with Et3N (18%
and 34% ee, respectively). However, methoxycarbonyl-
ation of the more hindered dimethylamide triflate (R)-
11c with Et3N gave a result (>99% ee, entry 5) nearly
Table 3. Hydroxycarbonylation of amino acid-derived aryl triflates
Entry R1
R2
R3
n
Yielda (%) eeb (%)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
NHMe NHBoc
H
0
0
0
0
0
0
0
0
0
0
0
1
1
0
0
97
95c
93
99
99
97
98
99
99
98
94
90
92
98
99
>99
>99
i
identical to that obtained with Pr2NEt (entry 6). This
NHMe
NH2
NH2
H
NHBoc
H
H
NHBoc
H
NHBoc
H
NHBoc
H
NHBoc
H
NHBoc
H
NHBoc
NHBoc
NHBoc
H
40d,e
outcome lended credence to the notion that the
steric bulk of both the amine base and the substrate
determined the level of racemization in the product.
Also of interest was the observation that unlike the
92d
95d
>99
>99
93
NH2
NMe2
NMe2
NHPh
NHPh
NHBoc
H
NHBoc
H
92
NHBn NHBoc
>99
>99
>99
>99
0
NHBn
NH2
NH2
H
NHBoc
H
OMe
OMe
NHBoc
H
0
a Reactions stirred at indicated temperature until complete by LC/MS
(12–18 h).
b Determined by chiral HPLC after conversion of the acids to the
methyl esters with trimethylsilyldiazomethane in THF/MeOH.
c Performed on a 5 g scale.
d Aryltriflate starting material ee = 98%.
i
e Reaction performed with Et3N instead of Pr2NEt.
Scheme 3. Hydroxycarbonylations in the presence of D2O.

4512
J. B. Grimm et al. / Tetrahedron Letters 48 (2007) 4509–4513
Table 4. Methoxycarbonylation of amino acid-derived aryl triflates
Weigele, M.; Narula, S. S. Biorg. Med. Chem. Lett. 2001,
11, 1665–1669; (d) Chen, L.; Tilley, J.; Trilles, R. V.; Yun,
W.; Fry, D.; Cook, C.; Rowan, K.; Schwinge, V.;
Campbell, R. Bioorg. Med. Chem. Lett. 2002, 12, 137–
140.
4. Cai, C.; Breslin, H. J.; He, W. Tetrahedron 2005, 61, 6836–
6838.
5. Recent examples include, but are by no means limited to:
(a) the transamination of phenylglyoxilic acid: Cameron,
M.; Cohen, D.; Cottrell, I. F.; Kennedy, D. J.; Roberge,
C.; Chartrain, M. J. Mol. Catal. B: Enzym. 2001, 14, 1–5;
(b) the hydration of phenylglycine nitriles: Hensel, M.;
Lutz-Wahl, S.; Fischer, L. Tetrahedron: Asymmetry 2002,
13, 2629–2633; (c) the hydrolysis of phenylglycine triflu-
oroacetamides: Arnusch, C. J.; Pieters, R. J. Eur. J. Org.
Chem. 2003, 3131–3138; (d) the hydrolysis of phenyl-
hydantoins: Wu, S.; Yang, L.; Liu, Y.; Zhao, G.; Wnag, J.;
Sun, W. Enzyme Microb. Technol. 2005, 36, 520–526.
6. For examples of an asymmetric Strecker approach, see: (a)
Yet, L. Angew. Chem., Int. Ed. 2001, 40, 875–877; (b)
Prasad, B. A. B.; Bisai, A.; Singh, V. K. Tetrahedron Lett.
2004, 45, 9565–9567.
Entry
R
Amine base
Yielda (%)
eeb (%)
1
2
3
4
5
6
7
8
NH2
NH2
Et3N
72
93
86
97
92
96
99
88
18c
88c
34
iPr2NEt
Et3N
NHMe
NHMe
NMe2
NMe2
NHPh
NHBn
iPr2NEt
Et3N
98
>99
>99
93
iPr2NEt
iPr2NEt
iPr2NEt
>99
a Reactions stirred at indicated temperature until complete by LC/MS
(2–8 h).
b Determined by chiral HPLC.
c Aryl triflate starting material ee = 98%.
7. For examples of an asymmetric reduction approach, see:
(a) Versleijen, J. P.; Sanders-Hovens, M. S.; Vanhomme-
rig, S. A.; Vekemans, J. A.; Meijer, E. M. Tetrahedron
1993, 49, 7793–7802; (b) Micskei, K.; Holczknecht, O.;
Hajdu, C.; Patonay, T.; Marchis, V.; Meo, M.; Zucchi, C.;
Palyi, G. J. Organomet. Chem. 2003, 682, 143–148; (c)
Kadyrov, R.; Riermeier, T. H.; Dingerdissen, U.; Tararov,
V.; Borner, A. J. Org. Chem. 2003, 68, 4067–4070; (d)
Nolin, K. A.; Ahn, R. W.; Toste, F. D. J. Am. Chem. Soc.
2005, 127, 12462–12463.
hydroxycarbonylation, the methoxycarbonylation was
highly sensitive to the ligand:catalyst ratio. A 1:1 ratio
proved to be optimal; higher ligand loadings resulted
in sluggish reactions and poor levels of conversion.
In summary, we have developed a convenient and
efficient protocol for the synthesis of enantiopure
4-carboxy derivatives of phenylglycine. These useful
building blocks can be prepared from the corresponding
aryl triflates through a palladium-catalyzed carbonyl-
ation. The enantiopurity of the carboxylate products
was found to be highly dependent on the steric environ-
ment of both the benzylic stereocenter and the amine
base. Excellent retention of ee was achieved when the
bulky amine iPr2NEt was employed in the reaction. This
method proved to be effective for a range of substrates
and could be performed on multi-gram scale without
diminishing yield or chiral purity.
8. Hirai, Y.; Yokota, K.; Momose, T. Heterocycles 1994, 39,
603–612.
9. Cacchi, S.; Lupi, A. Tetrahedron Lett. 1992, 33, 3939.
10. Because the enantiomers of carboxylic acid products 13a–f
and 14b were not separable by chiral HPLC, the % ee
values correspond to the methyl ester derivatives, obtained
by stirring each acid with excess TMS–diazomethane in
MeOH/THF for 5 min and subsequent removal of the
solvent.
11. Chiral HPLC conditions: The following conditions for the
methyl ester of 13a (i.e., 17a) are representative. Chiral
column: Chiralpak AD, 4.6 · 250 mm; mobile phase: 40%
IPA/60% hexanes, isocratic; flow rate: 0.75 mL/min; UV
detector: 254 nm; retention times: 6.02 min (for (S) enan-
tiomer), 8.26 min (for (R) enantiomer).
Acknowledgements
12. It was noted that briefly sparging the reaction mixture
with CO prior to (rather than after) amine addition gave
consistently higher yields. Bubbling the gas through the
reaction mixture after addition of the base, even for short
periods of time, inevitably resulted in a significant loss of
amine.
We thank Professor Samuel Danishefsky for insightful
mechanistic discussions relating to this work. We are
also grateful to Grace Bi for assistance with chiral
HPLC analysis of the intermediates and final products.
13. Relevant pKa values (as reported by the ACD Labs 8.0
pKa Database from Advanced Chemistry Development):
N-methylmorpholine 7.4, DABCO 8.1, N-methylpiperi-
References and notes
i
i
dine 9.9, Et3N 10.7, Pr2NEt 11.0, Pr2NH 11.1.
1. Cacchi, S.; Ciattini, P. G.; Morera, E.; Ortar, G. Tetra-
hedron Lett. 1986, 27, 3931–3934.
2. Dolle, R. E.; Schmidt, S. J.; Kruse, L. I. J. Chem. Soc.,
Chem. Commun. 1987, 904–905.
3. (a) Wrobel, J.; Dietrich, A. Tetrahedron Lett. 1993, 34,
3543–3546; (b) Klein, L. L.; Li, L.; Chen, H.-J.; Curty, C.
B.; DeGoey, D. A.; Grampovnik, D. J.; Leone, C. L.;
Thomas, S. A.; Yeung, C. M.; Funk, K. W.; Kishore, V.;
Lundell, E. O.; Wodka, D.; Meulbroek, J. A.; Alder, J. D.;
Nilius, A. M.; Lartey, P. A.; Plattner, J. J. Bioorg. Med.
Chem. 2000, 8, 1677–1696; (c) Sundaramoorthi, R.;
Siedem, C.; Vu, C. B.; Dalgarno, D. C.; Laird, E. C.;
Botfield, M. C.; Combs, A. B.; Adams, S. E.; Yuan, R. W.;
14. Hydroxycarbonylation procedure: The following procedure
for (R)-11a is representative. Triflate (R)-11a (500 mg,
1.21 mmol), Pd(OAc)2 (8.2 mg, 0.036 mmol), and dppp
(30 mg, 0.072 mmol) were taken up in DMF (3 mL), and
H2O (1 mL) was added. After sparging the solution with
i
CO for 10 min, Pr2NEt (313 mg, 2.43 mmol) was added.
A CO balloon was attached, and the reaction was stirred
at 70 ꢁC for 18 h. The mixture was then diluted with
EtOAc and extracted with saturated NaHCO3
(2 · 20 mL). The combined aqueous layers were acidified
to pH 2 with 1 N HCl and extracted with EtOAc
(2 · 40 mL). The combined organic extracts were washed

J. B. Grimm et al. / Tetrahedron Letters 48 (2007) 4509–4513
4513
with water and brine, dried (MgSO4), and evaporated to
added. A CO balloon was attached, and the reaction was
stirred at 70 ꢁC for 5 h. The mixture was subsequently
diluted with saturated NH4Cl and extracted with EtOAc
(2 · 30 mL). The combined organic extracts were washed
with brine, dried (MgSO4), and evaporated to a brown oil.
The residue was purified by silica gel chromatography (15–
75% EtOAc/hexanes) to afford methyl ester (R)-17a
afford clean acid (R)-13a (364 mg, 97%) as a colorless
1
solid. H NMR (DMSO-d6, 600 MHz) d 12.90 (1H, br s),
8.14 (1H, q, J = 4.7 Hz), 7.86 (2H, d, J = 8.2 Hz), 7.47
(2H, d, J = 8.3 Hz), 7.36 (1H, d, J = 8.5 Hz), 5.17 (1H, d,
J = 8.5 Hz), 2.55 (3H, d, J = 4.6 Hz), 1.34 (9H, s); 13C
NMR (DMSO-d6, 150 MHz) d 170.5, 167.7, 155.5, 144.6,
130.6, 130.0, 128.0, 79.2, 58.2, 28.8, 26.4. MS (ESI) Calcd
for C15H20N2O5 [M+Na]+ 331.1. Found: 331.1.
1
(381 mg, 97%) as a colorless solid. H NMR (DMSO-d6,
600 MHz) d 8.15 (1H, q, J = 4.5 Hz), 7.89 (2H, d,
J = 8.1 Hz), 7.50 (2H, d, J = 8.1 Hz), 7.38 (1H, d,
J = 8.6 Hz), 5.19 (1H, d, J = 8.4 Hz), 3.81 (3H, s), 2.55
(3H, d, J = 4.9 Hz), 1.34 (9H, s); 13C NMR (DMSO-d6,
150 MHz) d 170.4, 166.7, 155.5, 145.0, 129.8, 129.4, 128.2,
79.2, 58.2, 52.8, 28.8, 26.4. MS (ESI) Calcd for
C16H22N2O5 [M+Na]+ 345.1. Found: 345.1.
15. Methoxycarbonylation procedure: The following procedure
for (R)-11a is representative. Triflate (R)-11a (500 mg,
1.21 mmol), Pd(OAc)2 (13.6 mg, 0.061 mmol), and dppp
(25 mg, 0.061 mmol) were dissolved in a mixture of DMF
(4 mL) and MeOH (2 mL). After bubbling CO through
i
the solution for 5 min, Pr2NEt (313 mg, 2.43 mmol) was