Knoevenagel Adducts as Trimethylenemethane Dipole Surrogates

Article abstract of DOI:10.1002/anie.201508100

Knoevenagel adducts derived from readily available acetoxyacetone and malonic acid derivatives served as trimethylenemethane surrogates for formal 1,3-difunctionalization through a sequence of selective γ-deprotonation/α-alkylation and palladium(0)-catalyzed allylic alkylation. Herein, we report the discovery and development of a three-component 1,3-difunctionalization of Knoevenagel adducts as well as a unique palladium(0)-catalyzed branch-selective allylic alkylation.

Full text of DOI:10.1002/anie.201508100

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201508100
German Edition: DOI: 10.1002/ange.201508100
Allylation
Knoevenagel Adducts as Trimethylenemethane Dipole Surrogates
Peter Vertesaljai+, Primali V. Navaratne+, and Alexander J. Grenning*
Abstract: Knoevenagel adducts derived from readily available
acetoxyacetone and malonic acid derivatives served as trime-
thylenemethane surrogates for formal 1,3-difunctionalization
through a sequence of selective g-deprotonation/a-alkylation
and palladium(0)-catalyzed allylic alkylation. Herein, we
report the discovery and development of a three-component
1,3-difunctionalization of Knoevenagel adducts as well as
a unique palladium(0)-catalyzed branch-selective allylic alky-
lation.
an allyl anion that can undergo a-alkylation to introduce
a new C C bond[3] as well as an allyl acetate functional group,
À
which is a productive electrophile for functionalization
(Tsuji–Trost reaction).[5] Thus, substrate 2 can directly partic-
ipate in a second functionalization reaction with nucleophiles
to yield 1,3-difunctionalized building blocks 3 with high
efficiency. Moreover, we envisioned intra- and intermolecular
variants of this formal 1,3-difunctionalization (Scheme 1).
Although there are numerous methods for either a- or g-
functionalization,[3,4] general strategies for the difunctionali-
zation of Knoevenagel adducts are scarce. Instead, 1,3-
difunctionalization is commonly achieved with reagents that
are used to generate trimethylenemethane (TMM), which has
been used extensively in [3+2] cycloaddition reactions
[Eq. (2); TMS = trimethylsilyl].[6] Herein we report our first
studies on Knoevenagel adducts as TMM surrogates.
K
noevenagel adducts are attractive chemical building blocks
owing to their environmentally friendly construction through
condensation[1,2] and their ability to undergo selective chem-
ical transformation.[3,4] For example, the g-C H deprotona-
tion/a-alkylation of Knoevenagel adducts results in a C C
À
À
bond as well as an alkene functional group for further
manipulation.[3] Grossman and Varner demonstrated that
ketone-derived Knoevenagel adducts can undergo alkylation
and then an ozonolysis/retro-Claisen condensation sequence
to yield monosubstituted malonic acid derivatives
[Eq. (1)].[3b,c]
The first challenge was the development of a g- versus g’-
selective deprotonation/a-alkylation sequence (Scheme 1,
1!2)). Though it is understood that allyl anions derived
from Knoevenagel adducts react exclusively at the a position
with alkyl halides and palladium–p-allyl electrophiles,[3]
regioselective deprotonation (g vs. g’) has yet to be examined.
The desired chemical events for our strategy are g-deproto-
nation and a-alkylation to give allyl acetates 2, whereas g’-
deprotonation/a-alkylation yields a vinyl acetate. The selec-
tivity challenge was made greater by the suspicion that the g’-
Rather than cleavage of the alkene functional group, we
reasoned that formal 1,3-difunctionalization of the acetoxy-
acetone-derived Knoevenagel adduct
1
was possible
(Scheme 1). Selective g- versus g’-deprotonation results in
À
C H atoms in 1 are several orders of magnitude more acidic
[7]
À
than the g-C H atoms.
The second challenge was that allylic acetates with
a quaternary carbon atom at the 2-position are underexplored
in Tsuji–Trost allylation chemistry.[8] Moreover, it was
reported that the palladium-catalyzed allylic substitution of
a 2-tert-butyl allyl acetate/carbonate derivative was sluggish
and low yielding, probably owing to the bulky 2-substituent.[8]
To begin our investigation, we synthesized a variety of
acyloxyacetone-derived Knoevenagel adducts 1a–d with
different ester (E) groups and acyloxy groups in two steps
from hydroxyacetone through acylation then Knoevenagel
condensation.[9] Notably, the Knoevenagel condensation
resulted in starting materials 1a–d as E/Z isomeric mixtures.[9]
Also, we chose bulky oxygen activating groups (pivaloyl and
tert-butoxycarbonyl, Boc) with the hope of sterically biasing
the deprotonation event.
Scheme 1. Proposed 1,3-difunctionalization of Knoevenagel adducts.
E=electron-withdrawing group, R=alkyl group, X=leaving group,
À
Y H=pronucleophile.
[*] P. Vertesaljai,[+] P. V. Navaratne,[+] Prof. A. J. Grenning
Department of Chemistry, University of Florida
P.O. Box 117200, Gainesville FL 32611-7200 (USA)
E-mail: grenning@ufl.edu
Homepage: http://grenning.chem.ufl.edu
[+] These authors contributed equally.
Supporting information and ORCID(s) from the author(s) for this
article are available on the WWW under http://dx.doi.org/10.1002/
anie.201508100.
With Knoevenagel adducts 1a–d in hand, we began
investigating conditions for kinetically selective deprotona-
Angew. Chem. Int. Ed. 2016, 55, 317 –320
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
317

Angewandte
Communications
Chemie
tion (g vs. g’). Unfortunately, standard deprotonation con-
ditions for kinetic selectivity (bulky bases at cryogenic
temperatures) failed to give the desired product cleanly.[10]
However, we were pleased to discover a simple protocol for g-
deprotonation/a-allylation with kinetic selectivity when allyl
tert-butyl carbonate was utilized as the electrophile in the
presence of a catalytic amount of [Pd(PPh3)4]. Knoevenagel
adducts 1a–d were allylated in good yield to give 2a–d on
a gram to multigram scale in all cases under operationally
simple conditions (08C!RT). Next, under standard Tsuji–
Trost conditions ([Pd(PPh3)4] (1 mol%), Cs2CO3, THF or
toluene) with para-methoxyphenol as the nucleophile, the
second allylic alkylation occurred without incident for
substrates 2a–d. We thus confirmed our hypothesis that
Knoevenagel adducts can undergo 1,3-difunctionalization
through the discovery of an iterative allylic alkylation
sequence.
We next examined other types of nucleophiles that can
react with allylic acetates. With little change from the
conditions in Scheme 2, an electron-rich phenol and 2-
Scheme 3. Scope of the reaction with respect to the nucleophile.
Reaction conditions: I) NaH (1 equiv), pronucleophile (1.5 equiv), [Pd-
(PPh3)4] (1 mol%), toluene, room temperature, 1.5 h, 100 mg scale;
II) NaH (1 equiv), pronucleophile (1.5 equiv), [Pd2dba3] (2 mol%),
DavePhos (8 mol%), toluene, 508C, 2–18 h. [a] Yield for the reaction
under conditions I. [b] The reaction was carried out in N,N-dimethyl-
formamide instead of toluene. [c] Yield for the reaction under con-
ditions II. [d] The catalyst was formed from [Pd(Cp)(allyl)] (10 mol%)
and DavePhos (10 mol%). [e] Cs2CO3 was used instead of NaH.
[f] The reaction was carried out in diethyl ether instead of toluene.
dba=dibenzylideneacetone, LG=leaving group.
Increased yields were also observed for the synthesis of 3g
and 3j under these conditions.
We next explored the viability of the commercially
available acyloin derivative 4 as a starting material. Knoeve-
nagel condensation and initial allylic alkylation occurred
without incident to yield 2e on a gram scale (Scheme 4).[9]
Scheme 2. Sequence development. Reaction conditions: I) [Pd(PPh3)4]
(1 mol%), THF, 08C!RT, 10 min (1–5 g scale); II) Cs2CO3 or NaH,
[Pd(PPh3)4] (1 mol%), CH2Cl2, room temperature, 1.5 h (0.5–1 g scale).
[a] The reaction was carried out with Cs2CO3. [b] The reaction was
carried out with NaH. Bn=benzyl, PMP=para-methoxyphenyl.
naphthol underwent successful alkylation to give 3e and 3 f,
respectively, in good yield (Scheme 3). Similarly, the use of
a malonate nucleophile yielded 3h in reasonable yield.
Diminished reactivity and decomposition were observed
with electron-deficient phenols (e.g. 4-chlorophenol) and
sterically demanding diethyl allylmalonate. Considering a pos-
sible mechanism involving outer-sphere palladium–p-allyl
substitution, we reasoned that the low yield could be due to
decreased nucleophilicity (electronically and/or sterically)
Scheme 4. Synthesis and transformation of an acyloin-derived Knoeve-
nagel adduct.
À
and a challenging C Nu bond-forming trajectory at the
Compound 2e underwent successful allylic substitution with
para-methoxyphenol in the presence of [Pd(PPh3)4] or Pd/
Davephos, whereby the Pd/Davephos combination was most
effective and promoted the formation of 3l in 61% yield
(Scheme 5).[11] Interestingly, the substitution occurred with
selectivity for the branched product with both palladium
catalysts (10– > 20:1 branched (3l)/linear (L)), which is
unusual for palladium catalysis.[12] For these few known
cases,[12] it is thought that the ligand on Pd dictates the
regiochemical outcome.[13] In our case, various Pd/ligand
combinations resulted in good selectivity for the branched
product, thus suggesting that the reaction is controlled
primarily by the substrate. In comparison, palladium–p-allyl
palladium–p-allyl intermediate because of
the adjacent quaternary center (Figure 1).[5]
For these more electronically and/or
sterically challenging nucleophiles, it was
discovered that a combination of palladium
and the ligand 2-dicyclohexylphosphanyl-
2’-(N,N-dimethylamino)biphenyl (Dave-
phos) was particularly effective for allylic
Figure 1. Mecha-
nistic challenge
of 2-quaternary
palladium–p-allyl
alkylation of the allylic electrophile with
a quaternary center. With the Pd/Davephos
combination, 3i was prepared in 68% yield,
as compared to < 5% with [Pd(PPh3)4].
substitution.
318
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 317 –320

Angewandte
Communications
Chemie
observed, and the desired product 3u was not detected. Thus,
we found this branched-selective allylic alkylation to be most
efficient with phenol derivatives without a steric bias as
nucleophiles.
Mechanistically, we suggest that the selectivity for the
branched product is due to an unusual geometry imparted by
the quaternary center in the palladium–p-allyl intermediate
(Figure 2). When the palladium–p-allyl intermediate bears
Scheme 5. Palladium-catalyzed branched-selective allylic alkylation.
Reaction conditions: I) NaH or Cs2CO3, [Pd(PPh3)4] (2 mol%), toluene,
room temperature, 2 h; II) Cs2CO3, [Pd(Cp)(allyl)] (2 mol%), DavePhos
(2 mol%), toluene, room temperature, 2 h.
intermediates generated from allylic gem-diacetates (or
related compounds) undergo substrate-controlled branched-
selective allylic alkylation owing to the electronic impact of
the terminal oxygen atom on the palladium–p-allyl inter-
mediate.[14] Generally speaking, common methods for
branched-selective allylic alkylation utilize iridium-based
catalysts.[15]
Figure 2. Mechanistic hypothesis for the unexpected selectivity for
branched products.
Regarding the scope of the reaction (Scheme 6),
palladium(0)-catalyzed branched-selective allylic alkylation
of 2e was observed with a variety of sterically and electroni-
a H atom at the 2-position and an aliphatic or aromatic group
at the 1-position (Figure 2), it is well-understood that the
palladium–p-allyl intermediate adopts a geometry in which
the 2-H and 1-R group are syn, and allylic alkylation occurs by
a sterically driven attack at the least-substituted position.[5] In
our case, the 2-quaternary center and the methyl group could
potentially adopt an anti arrangement, and we suggest that
the unusual selectivity results from this geometric orienta-
tion.[13]
To probe our hypothesis (Figure 2), we prepared cyclic
analogues 5a and 5b, as these substrates are locked in a syn
arrangement (Scheme 7) and should therefore undergo
linear-selective allylic alkylation (Figure 2). With [Pd(PPh3)4]
Scheme 6. Studies of the reaction scope.
cally differentiated phenols. With para- or meta-substituted
phenols, both good yields and good selectivity for the
branched product were observed. For example, all para-
substituted reagents tested reacted with 18–20:1 branched/
linear selectivity (products 3l, 3m, 3q). meta-Tolyl and 3,5-
dimethoxy aromatic rings were incorporated in good yield
and with 20:1 and 15:1 branched/linear selectivity, respec-
tively. Although a good yield (63%) was observed for the
incorporation of a 3-fluorophenyl moiety, the reaction
occurred with only 8:1 selectivity for the branched product.
A 2-naphthyl moiety was incorporated in good yield with
excellent selectivity for the branched product. Yields were
reasonable for the introduction of ortho-substituted aromatic
rings to give products 3o (2-methyl, 62%) and 3s (2-chloro,
41%), but the selectivity for the branched product decreased
below 10:1 in both cases. Surprisingly, 2,3-dimethoxyphenol
was not a competent nucleophile; only decomposition was
Scheme 7. Mechanistic probe and the synthesis of cycloheptenes.
Reaction conditions: I) Cs2CO3 (1.5 equiv), HOPMP (1.5 equiv),
[Pd(PPh3)4] (1 mol%); II) Cs2CO3 (1.5 equiv), HOPMP (1.5 equiv), [Pd-
(Cp)(allyl)] (4 mol%); III) Grubbs II catalyst (0.5 mol%), CH2Cl2;
IV) Pd/C, H2. [a] Yield for the reaction under conditions I.
or Pd/Davephos catalyst systems, 5a underwent clean b-
hydrogen elimination to the cycloheptyl triene, thus providing
no insight into the origins of our branched selectivity.
Fortunately, upon selective hydrogenation of the endocyclic
double bond, the hypothesized selectivity was observed when
[Pd(PPh3)4] was utilized as a catalyst. The Pd/DavePhos
combination was also effective, although the reaction was less
clean and lower yielding.
Angew. Chem. Int. Ed. 2016, 55, 317 –320
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
319

Angewandte
Communications
Chemie
ming, J. Chem. Soc. Perkin Trans. 2 2001, 497 – 507; f) Y. Sato, Y.
Oonishi, M. Mori, J. Org. Chem. 2003, 68, 9858 – 9860; g) D. Xue,
Y.-C. Chen, Q.-W. Wang, L.-F. Cun, J. Zhu, J.-G. Deng, Org. Lett.
2005, 7, 5293 – 5296; h) M. Bell, K. Frisch, K. A. Jørgensen, J.
Org. Chem. 2006, 71, 5407 – 5410; i) S. R. Waetzig, D. K.
Rayabharapu, J. D. Weaver, J. A. Tunge, Angew. Chem. Int.
Ed. 2006, 45, 4977 – 4980; Angew. Chem. 2006, 118, 5099 – 5102.
[4] Knoevenagel adducts can undergo g-selective substitution:
a) T. B. Poulsen, M. Bell, K. A. Jorgensen, Org. Biomol. Chem.
2006, 4, 63 – 70; b) T.-Y. Liu, H.-L. Cui, J. Long, B.-J. Li, Y. Wu,
L.-S. Ding, Y.-C. Chen, J. Am. Chem. Soc. 2007, 129, 1878 – 1879;
c) J.-W. Xie, W. Chen, R. Li, M. Zeng, W. Du, L. Yue, Y.-C. Chen,
Y. Wu, J. Zhu, J.-G. Deng, Angew. Chem. Int. Ed. 2007, 46, 389 –
392; Angew. Chem. 2007, 119, 393 – 396; d) S. Noritake, N.
Shibata, S. Nakamura, T. Toru, M. Shiro, Eur. J. Org. Chem.
2008, 3465; e) X. Wang, G. Liu, X.-H. Xu, N. Shibata, E.
Tokunaga, N. Shibata, Angew. Chem. Int. Ed. 2014, 53, 1827;
Angew. Chem. 2014, 126, 1858; f) J. Alemµn, C. B. Jacobsen, K.
Frisch, J. Overgaard, K. A. Jorgensen, Chem. Commun. 2008,
632.
[5] B. M. Trost, D. L. Van Vranken, Chem. Rev. 1996, 96, 395 – 422.
[6] a) B. M. Trost, D. M. T. Chan, J. Am. Chem. Soc. 1979, 101,
6429 – 6432; b) B. M. Trost, Angew. Chem. Int. Ed. Engl. 1986,
25, 1 – 20; Angew. Chem. 1986, 98, 1 – 20.
[7] a) The University of Wisconsin, Department of Chemistry has
compiled a helpful list of pKa values: http://www.chem.wisc.edu/
areas/reich/pkatable/.
[8] C. Commandeur, S. Thorimbert, M. Malacria, J. Org. Chem.
2003, 68, 5588 – 5592.
[9] See the Supporting Information for details.
As both steps of the two-step iterative allylic alkylation
described above are catalyzed by Pd0, we reasoned that a one-
pot 1,3-difunctionalization should be possible (Scheme 8).
Without any change in the reaction conditions, the Knoeve-
Scheme 8. One-pot 1,3-difunctionalization of Knoevenagel adducts.
Piv=pivaloyl.
nagel adducts 1a and 1 f underwent one-pot palladium-
catalyzed iterative allylic alkylation with allyl carbonate and
para-methoxyphenol in comparable yields to those observed
for the two-step sequence.
In conclusion, we have introduced a strategy for the 1,3-
difunctionalization of Knoevenagel adducts as TMM surro-
gates. We also uncovered a palladium(0)-catalyzed branched-
selective allylic alkylation, which may result from an anti
arrangement of the palladium–p-allyl intermediate. Cur-
rently, our major limitation is that the initial g-deprotona-
tion/a-alkylation can only be performed with an allyl
carbonate as the electrophile. Future studies are aimed at
intramolecular variants for efficient carbocycle formation, as
well as the application of this method in the synthesis of
complex molecules.
[10] J. dꢀAngelo, Tetrahedron 1976, 32, 2979 – 2990.
[11] See the Supporting Information for full details regarding the
development of the reaction in Scheme 5.
[12] a) B. M. Trost, M. J. Krische, R. Radinov, G. Zanoni, J. Am.
Chem. Soc. 1996, 118, 6297 – 6298; b) X.-F. Yang, W.-H. Yu, C.-
H. Ding, Q.-P. Ding, S.-L. Wan, X.-L. Hou, L.-X. Dai, P. -J. Wang,
J. Org. Chem. 2013, 78, 6503 – 6509; c) T. Hayashi, M. Kawat-
sura, Y. Uozumi, Chem. Commun. 1997, 561 – 562; d) A. J.
Blacker, M. L. Clarke, M. S. Loft, J. M. J. Williams, Org. Lett.
1999, 1, 1969 – 1971.
[13] a) B. Aakermark, S. Hansson, A. Vitagliano, J. Am. Chem. Soc.
1990, 112, 4587 – 4588; b) M. Sjoegren, S. Hansson, P. O. Norrby,
B. Aakermark, M. E. Cucciolito, A. Vitagliano, Organometallics
1992, 11, 3954 – 3964; c) P. O. Norrby, B. Aakermark, F. Haeff-
ner, S. Hansson, M. Blomberg, J. Am. Chem. Soc. 1993, 115,
4859 – 4867; d) M. P. T. Sjoegren, S. Hansson, B. Aakermark, A.
Vitagliano, Organometallics 1994, 13, 1963 – 1971; e) J. W. Faller,
N. Sarantopoulos, Organometallics 2004, 23, 2179 – 2185; f) P.
Fristrup, T. Jensen, J. Hoppe, P.-O. Norrby, Chem. Eur. J. 2006,
12, 5352 – 5360; g) U. Kazmaier, D. Stolz, K. Krämer, F. L.
Zumpe, Chem. Eur. J. 2008, 14, 1322 – 1329.
Acknowledgements
We thank the College of Liberal Arts and Sciences and the
Department of Chemistry at the University of Florida for
start-up funds.
Keywords: 1,3-difunctionalization · allylation ·
Knoevenagel condensation · palladium catalysis ·
trimethylenemethane
How to cite: Angew. Chem. Int. Ed. 2016, 55, 317–320
Angew. Chem. 2016, 128, 325–328
[14] a) B. M. Trost, J. Vercauteran, Tetrahedron Lett. 1985, 26, 131 –
134; b) B. M. Trost, C. B. Lee, J. M. Weiss, J. Am. Chem. Soc.
1995, 117, 7247 – 7248; c) N. Vicart, J. GorØ, B. Cazes, Tetrahe-
dron 1998, 54, 11063 – 11078.
[15] For reviews of iridium-catalyzed allylic alkylation, see: a) G.
Helmchen, A. Dahnz, P. Dübon, M. Schelwies, R. Weihofen,
Chem. Commun. 2007, 675 – 691; b) J. F. Hartwig, L. M. Stanley,
Acc. Chem. Res. 2010, 43, 1461 – 1475.
[1] a) E. Knoevenagel, Ber. Dtsch. Chem. Ges. 1898, 31, 2596 – 2619;
b) A. C. Cope, J. Am. Chem. Soc. 1937, 59, 2327 – 2330; c) A. C.
Cope, C. M. Hofmann, C. Wyckoff, E. Hardenbergh, J. Am.
Chem. Soc. 1941, 63, 3452 – 3456.
[2] G. Jones, Org. React. 1967, 15, 204 – 599.
[3] Knoevenagel adducts can undergo a-selective substitution:
a) A. C. Cope, K. E. Hoyle, D. Heyl, J. Am. Chem. Soc. 1941,
63, 1843 – 1852; b) E. M. Hancock, A. C. Cope, Org. Synth. 1945,
25, 44 – 46; c) R. B. Grossman, M. A. Varner, J. Org. Chem. 1997,
62, 5235 – 5237; d) H. Nakamura, H. Iwama, M. Ito, Y. Yama-
moto, J. Am. Chem. Soc. 1999, 121, 10850 – 10851; e) H. Karlsen,
P. H. Songe, L. K. Sunsby, L. C. Hagen, P. Kolsaker, C. Rom-
Received: August 29, 2015
Published online: October 22, 2015
320
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 317 –320