Use of a Multicomponent Reaction for Chemoselective Derivatization of Multiple Classes of Metabolites

Full text of DOI:10.1002/cbic.201200035

DOI: 10.1002/cbic.201200035
Use of a Multicomponent Reaction for Chemoselective Derivatization of
Multiple Classes of Metabolites
Kyle A. Totaro, Babajide O. Okandeji, and Jason K. Sello*[a]
The wide range in quantity, stability and structure of metabo-
lites in biological samples presents challenges in systems biol-
ogy and in diagnostic medicine where it is often the goal to
analyze dozens or hundreds of these molecules at a time.
Often, the analysis of low-molecular-weight metabolites is sim-
plified by selective derivatization based on their constituent
functional groups. Typically, a specific reagent is used to deri-
vatize one structural class of metabolites. This “one reagent–
one functional group” method can be technically cumbersome
when the objective is to analyze constituents of varying struc-
tures in a biological sample. Here, we demonstrate that the
Ugi 4-component reaction can be used to selectively derivatize
metabolites with either amine, carboxylic acid, aldehyde, or
ketone moieties under a single set of reaction conditions. The
reaction is compatible with aqueous conditions in which me-
tabolites originate and it yields stable products that can be
separated and ionized in LC-MS analysis. The inclusion of a spe-
cially designed UV-active isocyanide substrate in the Ugi reac-
tion ensures that all the reaction products have a chromophore
and thus can be easily detected. This derivatization of structur-
ally and functionally diverse metabolites (e.g., neurotransmit-
ters, hormones, and cofactors) bodes well for the analysis of
complex biological samples.
imine-forming reactions have been utilized in the derivatiza-
tion of aldehydes and ketones.[4] Derivatization of carboxylic
acids can be performed directly by reagent-based silylation or
methylation, or indirectly using carbodiimide-promoted
esterification.[4] Although these “one reagent–one functional
group” methods were designed for the characterization of one
or a small number of metabolites of interest, many groups
have attempted to adapt the reagents for holistic studies of
metabolism. In 2007, Carlson and Cravatt reported a new strat-
egy for metabolomic analyses termed “metabolite enrichment
by tagging and proteolytic release” (METPR) that exploits these
classical derivatization reagents.[5] In their method, solid sup-
ports functionalized with specific derivatization reagents were
used for the selective capture of metabolites from complex
biological samples. Subsequently, the captured metabolites
were released for analysis by the action of a protease that
cleaves the linker coupling the metabolites to the support.
While this method has advantages with respect to simplifying
the analytes, it still requires a unique reagent and set of reac-
tion conditions for each derivatization.
The inherent limitations of the classical reagents and the
expanding demands of modern day metabolomics warrant the
development of new methods for metabolite derivatization.
Ideally, these methods would be: 1) operationally simple,
2) chemoselective, 3) tolerant of aqueous conditions, and 4) ca-
pable of providing stable and separable products that can be
easily characterized with existing analytical instrumentation.
We propose that most of these criteria can be met by using
the Ugi 4-component reaction (U-4CR) for derivatization.[6] In
this multicomponent reaction (Scheme 1), an amine, an alde-
hyde/ketone, a carboxylic acid, and an isocyanide react sponta-
Historically, scientists have isolated and characterized one or
a handful of low-molecular-weight metabolites of interest at
a time. Present day analyses of organismal physiology (i.e., sys-
tems biology) and disease diagnosis have necessitated the
identification, characterization and/or measurement of dozens
or hundreds of metabolites in a biological sample. The general
term for these types of analyses is called metabolomics.[1] Al-
though non-destructive methods such as NMR have been ap-
plied, often metabolomic analyses involve the direct isolation
of metabolites and determination of their masses.[1a] With
some exceptions,[2] these analyses are typically facilitated by
derivatization.
Chemical derivatization has been useful in both GC-MS and
LC-MS-based metabolomics as a means to impart metabolites
with physicochemical properties that facilitate analysis.[3] Che-
moselectivity in the derivatizations is achieved by the use of
reagents with defined reactivity. For example, metabolites with
amines have been derivatized with electrophilic reagents in-
cluding 2,4-fluorodinitrobenzene, isothiocyanates, isocyanates,
and N-hydroxysuccinimide esters.[4] Oxime, hydrazone and
Scheme 1. The Ugi 4-component reaction.
neously to yield a stable peptide-like product (a-acylamino car-
boxamide). This reaction is generally high yielding and has
a broad substrate scope; most metabolites have one or more
functional groups that can participate in the reaction. Indeed,
there are examples in the literature wherein a-amino acids and
proteins have been used as substrates in U-4CRs.[7,8] Addition-
ally, the U-4CR is compatible with water[9] and solvents typical-
ly used for metabolite extractions (e.g., acetonitrile, chloroform
and methanol)[1g] and their products are easily characterized by
mass spectrometry. Most importantly, chemoselectivity can be
[a] K. A. Totaro,+ B. O. Okandeji,+ Prof. J. K. Sello
Chemistry Department, Brown University
324 Brook Street, Providence, RI 02914 (USA)
E-mail: jason_sello@brown.edu
[+] These authors contributed equally to this work.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.201200035.
ChemBioChem 2012, 13, 987 – 991
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
987

achieved by adding the other three reactants in slight molar
excess over the analyte of interest.
Table 1. Isolated yields of amino acid derivatization and retention times.
The detection of metabolites in LC-MS is greatly facilitated if
the derivatization reaction utilizes a chromophoric substrate
with an absorbance that is distinct from most metabolites.
Aside from facilitating detection, the chromophore enables the
relative and absolute quantifications of derivatized metabolites
to be determined spectrophotometrically. Since most commer-
cial LC-MS instruments have UV/Vis detectors, we sought to
use U-4CR reactants that have chromophores that can be
detected in the UV range. We elected to use a UV-active iso-
cyanide for the derivatization because the isocyano functional
group is rarely observed in nature;[6,10] accordingly, this isocya-
nide can be used in Ugi reactions to derivatize metabolites
containing either an amine, carboxylic acid, aldehyde or
ketone moiety. We envisioned that chromophoric isocyanides
could be prepared using the reaction of ring-opened oxazo-
lines with UV-active acyl chlorides or isocyanates.[11] Of the four
different UV-active isocyanides that we prepared and tested
(Scheme 2), the p-nitrophenyl-carbamate isocyanide was the
most appropriate for metabolite derivatization. It can be pre-
Amino acid Yield [%] tR [min]
Amino acid
Yield [%] tR [min]
alanine
arginine
59
n.d.
31
68
n.d.
4.9
n.d.
5.6
leucine
lysine[c]
methionine
68
69
54
9.4
4.5
8.9
11.8
6.1
5.3
7.3
9.9
8.3
10.7
asparagine
aspartate[a]
cysteine
glutamate[a] 70
glutamine[b] 87
13.3, 13.5 phenylalanine 60
n.d.
13.0, 13.2 serine
4.8
proline
52
37
44
44
72
76
threonine
glycine
histidine
isoleucine
39
96
71
4.5
4.6
13.7
tryptophan
tyrosine
valine
Reactions were performed at room temperature for two days. Chromato-
graphic conditions: 2.5ꢁ100 mm C18 reversed-phase column with
a linear gradient of 30–90% acetonitrile/0.1% formic acid in water over
16 min with UV detection at 320 nm. In most cases, these chromato-
graphic conditions did not separate the product diastereomers. [a] Yields
refer to isolated tandem Ugi–Passerini product and the retention times
reflect diastereomers. [b] Yield refers to isolated piperidinedione product.
[c] Yield refers to isolated e-lactam product. n.d.= not detected.
a-amino acids with isocyanides and aldehydes in
methanol at À308C.[7] This variant of the canonical
Ugi reaction is called a “4-component 5-center” reac-
tion because of the bifunctional a-amino acid sub-
strate and the participation of the methanol solvent
as a reactant (Scheme 3).[7] Despite our use of differ-
ent reaction conditions and isocyanide substrates,
the products we observed were equivalent to those
reported by Ugi,[7,12] except in the reactions with his-
tidine and glutamine (Scheme 4). In our reaction with
histidine, the product is linear rather than cyclic as re-
ported by Ugi.[7] In the reaction reported by Ugi, the
carbonyl of the isocyanoacetate substrate acts as
a hydrogen-bond acceptor that induces a conforma-
tion of the initial product that favors cyclization. In
contrast, the isocyanide substrate in our reaction with histidine
lacks an appropriately oriented hydrogen-bond acceptor for
cyclization. In our reaction with glutamine, observation of a pi-
peridinedione structure is indicative of a ring-closure involving
the amino acid’s primary amide side chain. It is not obvious
why we observed this cyclization and Ugi did not. In any case,
the yields and retention times of the products are reported in
Table 1. Most amino acids reacted with moderate yields,[13]
with the exception of arginine and cysteine.[14] We were further
gratified to find that all of the derivatized a-amino acids had
unique retention times under a
Scheme 2. The synthesis of UV-active isocyanides.
pared in high yield and has a lmax value of 320 nm that is dis-
tinct from that of most metabolites, enabling an enhancement
of the signal-to-noise ratio. Moreover, it does not have the typ-
ical noxious odor of isocyanides and is a solid that is stable on
the bench for weeks at room temperature.
In our first set of proof-of-principle experiments, a series of
reactions in which the p-nitrophenyl-carbamate isocyanide and
isobutyraldehyde were reacted with proteinogenic a-amino
acids in methanol at room temperature (Scheme 3 and
Table 1). Ugi and co-workers reported analogous reactions of
single chromatographic method;
this is not the case for those
that are not derivatized.
While the a-amino acid deriva-
tizations required methanol as
a substrate,[7] we sought to de-
termine if the reactions could
tolerate water. Ugi reactions are
Scheme 3. Derivatization of amino acids.
reported to be enhanced in
988
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 987 – 991

Scheme 4. Unexpected products of glutamine and histidine derivatization. The products reported by Ugi are shown on the left and the products observed in
this work are shown on the right (see also the Supporting Information).
aqueous conditions,[9] but we found that the yield of a model
reaction with serine was significantly lower when the metha-
nol/water ratio exceeded 3:1. In any case, the water tolerance
of this reaction is clearly advantageous in the context of proc-
essing biological samples.
thylamine, phenylpyruvic acid, phenyllactic acid, and phenyl-
acetic acid) in their urine.[18] Currently, there is no reliable
urine-based method for diagnosis of PKU due to difficulties in
detection of these metabolites. We successfully derivatized
these metabolites using the U-4CR (Table 2). In the LC-MS anal-
ysis, all of the derivatized metabolites could be detected at
Given our success with a-amino acids, we explored derivati-
zation of a wide range of structurally and functionally diverse
metabolites containing carboxylic acid, amine, aldehyde, or
ketone functional groups. In these reactions, analytes with the
functional group of interest were treated with the p-nitrophen-
yl-carbamate isocyanide and the two other requisite substrates
of the U-4CR.[15] As illustrated in Scheme 5, we efficiently deriv-
atized fatty acids (1a–c), intermediates in central metabolism
(2a–b), polyamines (3a–c), neurotransmitters (4a–d), a coen-
zyme (5), and hormones (6a–h). In most cases, we were
pleased to observe good yields in these reactions, regardless
of the structural complexity of the metabolite. However, con-
sistent with literature precedents,[16] we found that reactions
with ketone-containing metabolites had lower yields. Given
the complexity of the metabolite reactants, our observations
provide new insights into the substrate selectivity and toler-
ance of the U-4CR.[6] For example, we observe exclusive reac-
tion at the unconjugated ketone carbonyl of steroid hormones
and not their conjugated ketone. Furthermore, U-4CR derivati-
zations of molecules with phenols do not undergo competing
Ugi–Smiles reactions in which an isocyanide, aldehyde, amine
and phenol react with one another.[17]
Table 2. Diagnostic metabolites of PKU—isolated yields of derivatization
and retention times of products.
PKU metabolites
Yield
[%]
tR
[min]
PKU metabolites
Yield
[%]
tR
[min]
phenethylamine
phenylpyruvic acid
76
78
7.4
7.7
phenyllactic acid
phenylacetic acid
94
90
6.6
7.2
Chromatographic conditions: 2.5ꢁ100 mm C18 reversed-phase column
with a linear gradient of 45–95% acetonitrile/0.1% formic acid in water
over 10 min with UV-detector set to 320 nm.
320 nm and exhibited close, but distinct retention times. Fur-
thermore, all of the derivatized metabolites were readily de-
tected in the positive-ion mode in electrospray ionization-mass
spectrometry (ESI-MS). The singularity of ion mode in the ESI-
MS analysis highlights an advantage of our derivatization
chemistry because the underivatized, cationic phenethylamine
is best detected under positive-ion mode, but the anionic car-
boxylate metabolites are better detected under negative-ion
mode. The coalescence of the products’ retention times and
ionization conditions that result from the derivatization is
clearly advantageous with respect to analysis. While we have
not derivatized these metabolites in an actual urine sample
from a PKU patient, the ease of analysis suggests the utility of
this derivatization method in biomarker detection.
In the aforementioned derivatizations, the metabolites were
associated with normal physiology. We considered the possibil-
ity that this strategy could be used in clinical chemistry where
metabolites are often biomarkers of disease, particularly in
metabolic disorders.[1c] To demonstrate that this chemistry is
compatible with metabolites associated with disease, we per-
formed reactions with biomarkers of phenylketonuria (PKU).
This metabolic disorder results from a deficiency in phenylala-
nine hydroxylase, the enzyme that catalyzes conversion of phe-
nylalanine into tyrosine. Individuals with PKU have excess phe-
nylalanine in the blood and phenylalanine catabolites (phene-
In conclusion, the chemical derivatization of metabolites has
been and will continue to be an important tool in studies of
metabolism. Here, we demonstrate that the U-4CR is a compel-
ling alternative to the “one reagent–one functional group” ap-
ChemBioChem 2012, 13, 987 – 991
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
989

lites containing amines, carboxylic acids, aldehydes, or ketones
at room temperature under a single set of reaction conditions,
even in the presence of water. Through the use of a specially
designed UV-active isocyanide, we demonstrate that the U-4CR
yields products that can easily be detected, separated, and ion-
ized under defined conditions, irrespective of metabolite struc-
ture. These findings provide yet another application of the U-
4CR, which has been extensively been used in chemical syn-
thesis[6] and in bioconjugation of proteins.[8] Having validated
the derivatization chemistry on metabolite standards, work is
now in progress in our laboratory to demonstrate its utility on
real-life biological samples for metabolomic analyses and dis-
ease diagnosis.
Experimental Section
Procedures for reagent preparation and metabolite derivatization
and spectral data for the corresponding products are provided in
the Supporting Information.
Acknowledgements
This work was generously supported by funds from Brown Uni-
versity, including a R.B. Salomon New Faculty Award to J.K.S. and
a Frontier in Chemistry Award and a NSF Career Award
(MCB1053319) to J.K.S. The authors acknowledge Dr. Tun-Li Shen
for expert technical advice and support for LC-MS analyses.
Keywords: biomarkers
·
isocyanide
·
metabolites
·
metabolomics · Ugi reaction
[1] a) W. B. Dunn, D. I. Ellis, TrAC, TrAC Trends Anal. Chem. 2005, 25, 285–
294; b) W. Lu, B. D. Bennett, J. D. Rabinowitz, J. Chromatogr. B 2008,
871, 236–242; c) N. Vinayavekhin, E. A. Homan, A. Saghatelian, ACS
Chem. Biol. 2010, 5, 91–103; d) R. Madsen, T. Lundstedt, J. Trygg, Anal.
Chim. Acta 2010, 659, 23–33; e) R. Wei, G. Li, A. B. Seymour, Anal. Chem.
2010, 82, 5527–5533; f) A. Nordstrçm, R. Lewensohn, J. Neuroimmunol.
Pharmacol. 2010, 5, 4–17; g) S. S. Rubakhin, E. V. Romanova, P. Nemes,
J. V. Sweedler, Nat. Methods 2011, 8, S20-S29.
[2] B. D. Bennett, E. H. Kimball, M. Gao, R. Osterhout, S. J. Van Dien, J. D. Ra-
binowitz, Nat. Chem. Biol. 2009, 5, 593–599.
[3] J. M. Halket, D. Waterman, A. M. Przyborowska, R. K. P. Patel, P. D. Fraser,
P. M. Bramley, J. Exp. Bot. 2005, 56, 219–243.
[4] G. Lunn, L. C. Hellwig, Handbook of Derivatization Reactions for HPLC;
Wiley, New York, 1998.
[5] E. E. Carlson, B. F. Cravatt, Nat. Methods 2007, 4, 429–435.
[6] a) A. Dçmling, I. Ugi, Angew. Chem. 2000, 112, 3300–3344; Angew.
Chem. Int. Ed. 2000, 39, 3168–3210; b) A. Dçmling, Chem. Rev. 2006,
106, 17–89.
[7] I. Ugi, A. Demharter, W. Hçrl, T. Schmid, Tetrahedron 1996, 52, 11657–
11664.
[8] T. Ziegler, S. Gerling, M. Lang, Angew. Chem. 2000, 112, 2202–2205;
Angew. Chem. Int. Ed. 2000, 39, 2109–2112.
[9] M. C. Pirrung, K. D. Sarma, J. Am. Chem. Soc. 2004, 126, 444–445.
[10] P. J. Scheuer, Acc. Chem. Res. 1992, 25, 433–439.
Scheme 5. Metabolites and isolated yields of their derivatization reactions.
[11] T. Lindhorst, H. Bock, I. Ugi, Tetrahedron 1999, 55, 7411–7420.
[12] In reactions of aspartic acid and glutamic acid, the a-amino acid moiet-
ies participate in a Ugi 4-component 5-center reaction, while the car-
boxylate side chain participated in a Passerini reaction. This is a so-
called tandem Ugi–Passerini reaction.
proach that has been used for more than a half century in me-
tabolite derivatization. Using this multicomponent reaction
with commercially available or easily prepared substrates, one
can chemoselectively derivatize structurally complex metabo-
[13] Except for reactions with tryptophan, tyrosine and serine, our yields are
higher than those reported by Ugi. In the aforementioned cases, Ugi
990
www.chembiochem.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2012, 13, 987 – 991

included triethylamine presumably to enhance the nucleophilicity of
methanol. These findings imply that these reactions need not be per-
formed at À308C.
[16] S. T. M. Simila, S. F. Martin, Tetrahedron Lett. 2008, 49, 4501–4504.
[17] L. El Kaim, M. Gizolme, L. Grimaud, J. Oble, Org. Lett. 2006, 8, 4019–
4021.
[14] Arginine’s basic guanidinium group facilitates solvolytic cleavage of the
isocyanide carbamate, yielding a methyl carbamate. The poor solubility
of cysteine in methanol and acetonitrile precludes efficient reaction.
[15] Benzylamine, 4-pentenoic acid, and/or isobutyraldehyde were selected
for use as the other U-4CR substrates in these derivatization experi-
ments.
[18] K. Michals, R. Matalon, Am. J. Clin. Nutr. 1985, 42, 361–365.
Received: January 13, 2012
Published online on April 13, 2012
ChemBioChem 2012, 13, 987 – 991
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
991