"Extreme" Ugi reactions with some complex α-amino acids

Article abstract of DOI:10.1021/ol302379a

The Ti(IV)-catalyzed Ugi condensation of α-amino acids with electron-rich aromatic aldehydes performs adequately even with sterically demanding α-amino carboxylate salts. The reaction occurs diastereoselectively, in some cases with virtually complete dias

Full text of DOI:10.1021/ol302379a

ORGANIC
LETTERS
2012
Vol. 14, No. 18
4970–4973
“Extreme” Ugi Reactions With Some
Complex r‑Amino Acids
Charles Dylan Turner and Marco A. Ciufolini*
Department of Chemistry, University of British Columbia, 2036 Main Mall, Vancouver,
BC V6T 1Z1, Canada
ciufi@chem.ubc.ca
Received August 27, 2012
ABSTRACT
The Ti(IV)-catalyzed Ugi condensation of r-amino acids with electron-rich aromatic aldehydes performs adequately even with sterically
demanding r-amino carboxylate salts. The reaction occurs diastereoselectively, in some cases with virtually complete diastereoselectivity.
A stereochemical rationale for the reaction is proposed.
A need to alleviate at least some of the difficulties
encountered in the Ugi reaction of R-amino acids1 with
aromatic aldehydes, especially electron-rich ones, induced
us to devise a Ti(IV)-catalyzed variant of the process that
significantly improved kinetics and yields.2 Recently, an
opportunity arose to test the limits of this chemistry with
complex R-amino acids 1ꢀ2 as substrates (Figure 1). The
Ugi reaction of 1ꢀ2 piqued our interest because the NH2
Figure 1. Structure of R-amino acids 1 and 2.
group in either molecule is hindered, and its nucleophilic
character is inductively attenuated by the neighboring
NMe unit. Moreover, the free acid forms of 1ꢀ2 lactonize
easily. It will be seen shortly that the corresponding
lactones are not substrates for the Ugi reaction, while the
open forms are. The open structure of the substrates can
only be preservedby formation of a carboxylatesalt, which
therefore would have to be employed as such in the Ugi
step. However, this would deprive the medium of protonic
catalysis, hampering the overall process by retarding the
rate of imine formation (an acid-catalyzed process). The
problem would be especially acute with less electrophilic
aromatic aldehydes bearing electron-releasing substitu-
ents. A successful condensation of a carboxylate salt of
1ꢀ2 with an electron-rich aromatic aldehyde and an iso-
nitrile would thus constitute an “extreme” Ugi reaction,
the feasibility of which may seem doubtful.
Compounds 1 and 2 are required for an ongoing project
in total synthesis3 and may be reached from the known
lactone 3 (Scheme 1),4 starting with N-methylation to 4.
The direct azidation5 of the enolate of 4 was problematic,
but iodination proceeded efficiently to furnish 5 as a single
diastereomer6 and in nearly quantitative yield. The con-
figuration of 5, which segues from an approach of the
electrophilic agent from the convex face of the enolate,
was ascertained by the presence of dipolar coupling
(NOE) between Ha and the NMe hydrogens.7 Iodide 5
underwent facile nucleophilic substitution with azide ion.
(3) Turner, C. D. Dissertation; University of Bristish Columbia,
2012.
(4) Turner, C. D.; Ciufolini, M. A. Heterocycles 2012, 85, 85.
(5) Evans, D. A.; Britton, T. C.; Dorow, R. L.; Dellaria, J. F., Jr.
Tetrahedron 1988, 44, 5525. (b) Evans, D. A.; Britton, T. C.; Ellman,
J. A.; Dorow, R. L. J. Am. Chem. Soc. 1990, 112, 4011.
(6) Within the limits of 300 MHz 1H NMR.
€
(1) Reviews: (a) Domling, A. Chem. Rev. 2006, 106, 17. (b) De Graaf,
C.; Ruijter, E.; Orru, R. V. A. Chem. Soc. Rev. 2012, 41, 3969.
(2) Godet, T.; Bonvin, Y.; Vincent, G.; Ciufolini, M. A. Org. Lett.
2004, 6, 3281.
(7) Coupling constants were of little use in this case. For instance, in 5
the value of 3Jab, 1.4 Hz disallowed an unequivocal assignment.
r
10.1021/ol302379a
Published on Web 09/06/2012
2012 American Chemical Society

Scheme 1. Preparation of Diastereomeric Aminolactones 8 and 9
Scheme 2. Attempted Ugi Reaction with Lactone 9
However, product 6 was formed as the R-azido diastereo-
mer, as apparent again from an NOE correlation between
Ha and the NMe hydrogens. Lactone 7was thus a product of
nucleophilic substitution of Iꢀ with retention of configura-
tion, indicating that either the NMe unit had anchimerically
assisted the departure of the nucleofuge or epimerization at
the CdO R-position had occurred at some point during the
reaction. In the interest of evaluating the ease of epimeriza-
tion of the azidolactones, substance 6 was treated with DBU
in MeCN. This resulted in equilibration to a 1.2:1 mixture
of epimeric lactones 7 (major) and starting 6 (minor).
Evidently, the two epimers of the R-azido-lactone are nearly
isoenergetic, indicating that the conversion of 5into 6cannot
be attributed to equilibration after the substitution event,
and implying that in all likelihood the nucleophilic displace-
ment of iodide had occurred with anchimeric assistance.
Lactones 6 and 7 were readily separated by SiO2 flash
chromatography, and each was uneventfully hydrogenated
to furnish R-amino-lactones 8 and 9, the configuration of
which was ultimately confirmed by an X-ray structure of
a derivative (vide infra).
As indicated earlier, the lactones are incompetent Ugi
substrates. In this respect, attempted condensation of 9with
o-anisaldehyde, 10, and t-Bu-NC resulted in formation of
tetrahydropyridine 11 (22%) plus Passerini-type amide 12
(27%)8 as the sole identifiable products (Scheme 2). A brief
investigation of the unanticipated reaction leading to 11
revealed that the aldehyde is required,9 even though it is not
incorporated into the final product. An explanation might
be that 9 and 10 must first combine to afford imine 13,
wherein the increased acidity of the carbonyl R-proton, and
the stabilization of the incipient 14 by conjugation with
the aromatic ring, promote β-elimination of the pyrrolidine
Scheme 3. Saponification of Aminolactones 8 and 9
nitrogen. The aldehyde is no longer needed at this point: it is
released from 14, and the resulting 15 advances to 11.
The saponification of aminolactone 8 occurred smoothly
in the presence of aqueous LiOH to afford salt 18. By
contrast, the hydrolysis of 9 took place with a variable
degree of epimerization at the R-center. This undesirable
effect could be contained by carrying out the reaction with
aqueous KOH at 0 °C, whereupon a 2:1 mixture of salts 19
(major) and 20 resulted (Scheme 3; notice that compound
20 is the Kþ carboxylate version of 18). Separation of the
very polar isomeric salts was best avoided, and subsequent
experiments were carried out with mixture of the two.
Unsurprisingly, the Ugi reactions of 18ꢀ19 with electron-
rich aromatic aldehydes required more than the customary
5 mol % of Ti(IV). Still, reactions run at a 50 mol % catalyst
load returned the expected Ugi products in moderate
yield.10 Salt 18 combined with o-anisaldehyde and t-BuNC
(8) For a reaction leading to products similar to 12, see ref 2.
(9) In the absence of the aldehyde, only slow decomposition occurs.
(10) No reaction took place in the absence of TiCl4.
Org. Lett., Vol. 14, No. 18, 2012
4971

to afford a 7:1 mixture of products that were epimeric at the
level of the newly formed stereogenic center (39% from 8).
These were tentatively (vide infra) assigned as 21 (major)
and 22 (Scheme 4). No effort was made to further improve
these reactions, which afforded only lactone-type products:
evidently, the internal primary alcohol, instead external
MeOH, had intercepted a transient imino ester inter-
mediate.11 Also, the experiment described in Scheme 2
implies that 21ꢀ22 arise from an open form of the substrate,
and not from lactone 8.
Scheme 5. Ugi Reaction of Carboxylate Salt 19
Scheme 4. Ugi Reaction of Carboxylate Salt 18
The reaction of salt 19 (contaminated with ca. 33% of
20) proceeded in a similar fashion. However, product 23
(30% yield from 9, corresponding to 45% based on the
molar fraction of 19 in the starting mixture of salts) was
obtained substantially as a single diastereomer,6 while
contaminant 20 advanced to the anticipated 7:1 mixture
of 21 and 22 (Scheme 5). Silica gel flash chromatography
readily achieved separation of 23 from its isomers. The
HCl salt of 23 deposited diffraction quality crystals that
enabled an X-ray structural determination.12 This revealed
that the newly formed stereocenter had the (S)-configuration
(cf. ORTEP plot in Scheme 5).
Figure 2. Possible chelated model for Ti-catalyzed Ugi reaction.
Itseems appropriatetoventurearationaleforthe stereo-
chemical course of the above reaction, especially since
the literature provides no consistent picture for similar
transformations.1 Diastereoselectivity is probably deter-
mined by an irreversible attack of the isonitrile onto the
more favorable E-isomer of the imine arising upon
condensation of the amino acid with the aldehyde. The
isonitrile could possibly add to a chelate complex of the
imine (Figure 2). This may be tenable in the present case,
given the amount of TiCl4 present in the medium. How-
ever, such a model implies that the reaction of, e.g., an
R-(S)-amino acid (i.e., an L-amino acid) should always lead
to a product in which the newly formed stereocenter is of
R-configuration or, at the very least least, that the Lewis
acid should have an influence on the diastereomeric ratio
of the products. This is not the case.2 We believe instead
that the isonitrile adds to the more accessible face of
Figure 3. Nonchelated model for Ti-catalyzed Ugi reactions.
iminium ions 24ꢀ25 (Figure 3). An MMþ study13 of
closely related species14 revealed a preference for a con-
formation in which the carboxylate unit is approximately
normal ((9°) to the plane of the imino linkage and H-R
and H-β are nearly anti. In R-(S) amino acid derived 24,
the NMe moiety thus hinders access to the Re-face to a
greater degree that the COO(ꢀ) group hampers access
to the Si-face. Reaction from the Si-face is thus preferred.
€
(11) Compare: Ugi, I.; Demharter, A.; Horl, W.; Schmid, T. Tetra-
hedron 1996, 52, 11657.
(12) By contrast, various salts of 21 have yet to crystallize in a form
suitable for an X-ray diffractometric study.
(13) Calculations were carried out with the Hyperchem package,
available from Hypercube, Inc., Gainesville, FL.
(14) Details are provided as Supporting Information.
4972
Org. Lett., Vol. 14, No. 18, 2012

Further experiments indicated that aldehydes carrying
additional substitution on the aromatic nucleus react
normally, so long as they are not 2,6-disubstituted. For
instance, the reaction of aldehyde 26 with 4-MeO-C6H4-
NC and salt 18 produced 27 as the major component of an
8.5:1 mixture of diastereomers (26% overall yield from 8),
while an analogous reaction with 19 (contaminated with
about 33% of 20) and tert-Bu-NC or 4-MeO-C6H4-NC
afforded 28 (26% from 9, equivalent to 39% when cor-
rected for the molar fraction of 19 in the mixture of salts)
and 29 (32% from 9, equivalent to 48%), respectively
(Scheme 6; only the products derived from 19, essentially
single diastereomers,6 are shown). However, the reaction
of aldehyde 30 with 19ꢀ20 was inefficient (14%, equiva-
lent to 21%). This is attributable to further steric retarda-
tion of the rate of bothimine formation and ofnucleophilic
attack thereon, due to the 2,6-disubstituted nature of the
aldehyde. It is worth noting that product 31 was obtained
as a methyl ester, and not as a lactone.17
Scheme 6. Ugi Reactions with Aldehydes 26 and 30
In conclusion, the Ti(IV)-catalyzed Ugi reaction of
R-amino acids with aromatic aldehydes performs adequately
even with complex, hindered, carboxylate salts such as
18ꢀ19, so long as the aldehyde is not 2,6-disubstituted.
In the latter case, slower rates of imine formation and of
nucleophilic attack of the isonitrile depress overall yields
to synthetically unattractive levels. Ugi products thus
obtained may be valuable building blocks for the assembly
of certain alkaloidal frameworks.3,18 Research aiming
to optimize these transformations and to explore their
synthetic potential is ongoing. Pertinent results will be
disclosed in due course.
Conversely, in the iminium ion derived from the R-(R)
amino acid, it is the CHꢀCH2OH branch that blocks
access to the Si-face, promoting reaction from the
Re-face.15 Interestingly, a similar analysis leads to
accurate predictions regarding the sense of stereo-
selectivity of other Ugi reactions of R-amino acids re-
corded in the literature.16 On the basis of this model and
of the X-ray structure of 23, we tentatively assign the
configuration of Ugi products 21ꢀ22, as well as 27ꢀ29
and 31 (Scheme 6), as shown herein, pending confirmation
by X-ray diffractometry.
Acknowledgment. We thank the University of British
Columbia, the Canada Research Chair Program, NSERC,
CIHR, CFI, and BCKDF for financial support.
Supporting Information Available. Experimental pro-
cedures and characterization data for new compounds,
plus 1H and 13C NMR spectra. This material is available
free of charge via the Internet at http://pubs.acs.org.
(17) In addition to the expected 21ꢀ23, 27ꢀ29, and 31, the Ugi
reactions described herein returned mixtures of numerous uncharacter-
ized products. No evidence was ever garnered for the presence of starting
lactones 8ꢀ9 or tetrahydropyridine 11 in such mixtures.
(15) One may infer from the A-value of an NMe2 vs that of a CHMe2
(1.53 kcal/mol vs 2.52 kcal/mol: Eliel, E.; Wilen, S. H. Stereochemistry of
Organic Compounds; Wiley: New York, NY, 1994; pp 696ꢀ697) that the
steric demand of a CHꢀCH2OH branch is greater than that of an NMe
unit. Therefore, a CHꢀCH2OH should provide more effective steric
shielding of one of the imine faces than an NMe; i.e., it should promote
higher diastereoselectivity. This is consistent with observation.
(16) Details will be provided in a forthcoming full paper.
(18) For instance, those found in the quinocarcin/bioxalomycin/
cyanocyclin family of cytotoxic natural products: Scott, J. D.; Williams,
R. M. Chem. Rev. 2002, 102, 1669.
The authors declare no competing financial interest.
Org. Lett., Vol. 14, No. 18, 2012
4973