Photostability study of natural high-potency sweetener monatin in a model beverage system and characterisation of the degradation products

Article abstract of DOI:10.1016/j.foodchem.2011.08.073

Photodegradation of the naturally occurring high-potency sweetener monatin was studied in a lemon-lime beverage model system under simulated conditions. Most of the monatin disappeared after treatment with the equivalent of four days of ultraviolet light exposure and resulted in the formation of a number of degradation products. These degradation products have been isolated and characterised in the present study. On the basis of these identified structures, a photodegradation pathway has been proposed suggesting that monatin gets oxidised on the indole C-2 position to result in 2-hydroxymonatin. It also undergoes decarboxylation resulting in a 4-oxopentanoic acid analog and a major rearrangement resulting in an isonicotinic acid analog. Monatin also degrades into 3-formylindole and indole-3-carboxylic acid. Besides these, a partial monatin dimer and a monatin lactone were identified as additional degradation products. Details of the isolation and characterisation are given.

Full text of DOI:10.1016/j.foodchem.2011.08.073

Food Chemistry 131 (2012) 413–421
Contents lists available at SciVerse ScienceDirect
Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem
Photostability study of natural high-potency sweetener monatin in a model
beverage system and characterisation of the degradation products
a
b
b
b
Mani Upreti a, , Kasi V. Somayajula , Dennis J. Milanowski , Peter Kowalenko , Ulla Mocek ,
⇑
Rafael San Miguel a, Indra Prakash a
a The Coca-Cola Company, P.O. Box 1734, Atlanta, GA 30301, USA
b AMRI, Bothell Research Center, 22215 26th Ave. SE, Bothell, WA 98021, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 17 May 2011
Received in revised form 18 August 2011
Accepted 24 August 2011
Available online 5 September 2011
Photodegradation of the naturally occurring high-potency sweetener monatin was studied in a lemon–
lime beverage model system under simulated conditions. Most of the monatin disappeared after treat-
ment with the equivalent of four days of ultraviolet light exposure and resulted in the formation of a
number of degradation products. These degradation products have been isolated and characterised in
the present study. On the basis of these identified structures, a photodegradation pathway has been pro-
posed suggesting that monatin gets oxidised on the indole C-2 position to result in 2-hydroxymonatin. It
also undergoes decarboxylation resulting in a 4-oxopentanoic acid analog and a major rearrangement
resulting in an isonicotinic acid analog. Monatin also degrades into 3-formylindole and indole-3-carbox-
ylic acid. Besides these, a partial monatin dimer and a monatin lactone were identified as additional deg-
radation products. Details of the isolation and characterisation are given.
Keywords:
High-potency sweetener
Monatin
Photodegradation
Ultraviolet
Ó 2011 Elsevier Ltd. All rights reserved.
Degradation products
1. Introduction
commercial supply for monatin. There are numerous reports in the
literature regarding the synthesis of monatin via various chemical
Natural high-potency sweeteners can be used in beverages to
produce all natural low calorie products. Monatin is one such natu-
rally occurring high-potency sweetener, an amino acid derivative
which has been isolated from the root bark of a spiny-leafed hard-
wood shrub called Sclerochiton ilicifolius. This plant naturally grows
in the northwestern Transvaal region of South Africa. Monatin has
two chiral centres leading to four potential stereoisomers; (2R, 4R)
monatin, (2S, 4S) monatin, (2R, 4S) monatin, and (2S, 4R) monatin.
The first structure elucidation of isolated natural monatin reported
it to be (2S, 4S)-2-amino-4-carboxy-4-hydroxy-5-(3-indolyl)-pen-
tanoic acid (Vleggaar, Ackerman, & Steyn, 1992). Lately, all four iso-
meric forms of monatin have been reported to occur naturally in this
plant from different regions (Bassoli, Borgonovo, Busnelli, Morini, &
Drew, 2005). Interestingly, these stereoisomers have been found to
have different sweetening characteristics. It has been reported that
the sweetness intensity is dependent on the optical purity of each
stereoisomer, and (2R, 4R) monatin is the sweetest among the four
optical isomers, being about 2700 times sweeter in comparison to
5% (wt./vol) sugar (Amino & Hirasawa, 2005).
(Nakamura, Baker, & Goodman, 2000) and chemoenzymatic routes
(Buddoo et al., 2009). Monatin is being developed by Cargill (Brady,
Steenkamp, Rousseau, Buddoo, & Steenkamp, 2009) and Ajinomoto
(Amino & Hirasawa, 2006). Development of a sweetener requires
hydrolytic, photo and thermal stability studies to assure viability
of the sweetener. However there are no such stability study reports
on monatin in the literature. Recently it has been reported that
degraded monatin solution have presence of ‘‘musty’’ off flavours
which is characteristic of 3-methyl indole (Skatole) and samples
get discoloured after exposure to UV light (Evans & Goulson,
2010). Stability of monatin is thus an issue as it tends to degrade
in solution at high temperatures and upon ultraviolet (UV) exposure
especially under low pH conditions. There are few reports on the
stabilisation of monatin in beverage formulations where photodeg-
radation of monatin gets reduced with the help of added
antioxidants (Roy, 2009), radical scavengers (Evans & Goulson,
2010) and with use of a UV-light shielding packaging (Mori, 2006).
We have an interest in studying the stability of natural sweeteners
under our product matrix and storage conditions. The present work
is focused on studying in detail the photodegradation of monatin un-
der UV irradiation and high temperature conditions in a lemon–lime
beverage model system. From the degraded monatin containing
beverage, seven of the resulting degradation products of monatin,
compounds 1–7, were isolated and characterised in order to
understand monatin’s degradation pathway.
Monatin is present in only trace amounts, up to 0.007% by mass in
the root bark of S. ilicifolius; therefore, it is not a source of potential
⇑
Corresponding author. Tel.: +1 404 676 2923; fax: +1 404 598 2923.
E-mail address: mupreti@na.ko.com (M. Upreti).
0308-8146/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.foodchem.2011.08.073

414
M. Upreti et al. / Food Chemistry 131 (2012) 413–421
1 min, 100–0% B over 1 min, 0% B for 9 min) at a flow rate of
5 mL/min. The injection volume was 500 L.
cvvHPLC method III employed Atlantis dC18 column
m, p/n 186003748, Waters Corporation, Milford,
MA, USA) and a binary solvent mobile phase (A, water with 0.05%
TFA; B, acetonitrile with 0.05% TFA) using a gradient (12% B to 50%
B over 30 min, 50–95% B over 5 min, 95% B for 5 min, 95–12% B over
1 min, 12% B for 9 min) at 5 mL/min flow rate. The injection volume
2. Materials and methods
2.1. Reagents and chemicals
l
(250 ꢀ 4.6 mm, 5
l
2.1.1. (2R, 4R)-monatin
For this study we obtained monatin sample from in-house
chemical synthesis of (2R, 4R) monatin in high optical purity
(+99%) using published procedures. Starting from indole first a
mixture of the R,R/S,S stereoisomers of monatin was obtained fol-
lowing published procedures (Holzapfel, Bischofberger, & Olivier,
1994). This mixture was subjected to optical resolution via re-
peated optical salt formation steps as described in a 2005 patent
application (Amino & Hirasawa, 2005). The enantiomeric purifica-
tion was followed by chiral HPLC to confirm the stereo-isomeric
enrichment (Buddoo, Rousseau, & Gordon, 2009).
was 500 lL.
2.2.3. High resolution mass spectra
These were recorded using a Waters Premier Q-T of Mass Spec-
trometer (Waters Corporation, Milford, MA) running on the Mass-
Lynx software platform, version 4.1. Samples were prepared in a
buffer consisting of H2O-acetonitrile (1:1, v/v) + 0.1% formic acid
and infused at a rate of 5 lL/min. MS parameters were as follows:
Capillary voltages, 3.5 kV (ESI+) or 2.5 kV (ESIꢁ); cone voltages,
35 V (ESI+ and ESIꢁ) default, adjusted to provide 60.1 ions per push
(IPP); source temperature, 80 °C; desolvation temperature, 200 °C;
desolvation gas flow 200 L/h; scanning from m/z 100–1200.
2.1.2. Monatin lactam derivative
It was prepared for direct comparison with one of the degrada-
tion product, compound 2. A sample of monatin was heated in
aqueous solution for several hours at 60 °C following a published
procedure (Ikeda, Kogiso, & Nakamura, 2007).
3-Formylindole and indole-3-caboxylic acid commercial sam-
ples for authentication and ammonium formate were purchased
from Sigma–Aldrich (St. Louis, MO, USA). Acetonitrile used was
HPLC grade. Water was purified using a Millipore system (Billerica,
MA, USA).
2.2.4. NMR
1H and 2D NMR experiments were performed on a 500 MHz,
Bruker Avance DRX NMR (Bruker BioSpin Corporation, Billerica,
MA, USA) gradient system, running on the Xwin NMR software
platform. Various probes were used including a 5 mm inverse de-
tected z-gradient probe and a capillary NMR probe with a 10
flow cell.
lL
Lemon–lime beverage samples at pH 2.6 without carbonation
were prepared as reported earlier (Clos, DuBois, & Prakash, 2008).
2.3. Photodegradation experiments in a sunlight chamber
2.2. Analytical instruments and conditions
An Atlas Sun-test chamber (Atlas Material testing technology
GmbH, Linsengericht, Germany) was used for simulating sun-light
exposure to beverage samples. The programme used was a 6 h per-
iod of UV exposure in one run. Such a run is equivalent to one day
of sun-light exposure in Arizona or 400 langley at 40 °C. Here lang-
ley is defined as, ‘‘a unit of energy per unit area, equal to 1 gram-
calorie/cm2 commonly employed in radiation theory’’ (Weast,
1983).
2.2.1. LC–MS
Experiments were carried out on a Sciex API 150 single-quadru-
pole LC–MS system (AB Sciex, Framingham, MA, USA) with APCI
and ESI sources running on the Analyst 4.1 platform. The LC–MS
system had Agilent 1100 pumps and diode array detector and uti-
lised a Gilson 215 autosampler. The LC–MS method employed Syn-
ergi Hydro-RP column (250 ꢀ 4.6 mm, 4
lm, p/n 00G-4375-E0;
Phenomenex, Torrance, CA, USA) with UV detection at 280 nm
and a binary solvent mobile phase (A, water with 50 mM ammo-
nium formate, pH 4.0; B, acetonitrile) using a gradient (12% B for
5 min, 12–34% B over 5 min, 34–95% B over 5 min, 95% B for
5 min, 95–12% B over 1.0 min, 12% B for 9.0 min) at a flow rate of
2.4. Monatin degradation and mass balance analysis
First a small scale degradation of monatin was carried out under
UV irradiation at 40 °C in a lemon–lime beverage model system at
30 lg/mL in the Atlas sun-test chamber using regular transparent
1 mL/min. The injection volume was 200 lL. For compound 5, sol-
vent phase A was modified to have only 5 mM ammonium formate
polyethylene terephthalate (PET) bottles. The total UV exposure
was carried out for 24 h (equivalent to four days of sun-light expo-
sure in Arizona). Black PET bottles were used as a control to stan-
dardise conditions. The degraded monatin solution was analysed in
triplicate using LC–MS. The average peak area was used to deter-
mine the calculated concentration using the response factor deter-
mined for monatin. The calculated concentrations assume that the
degradation products have the same response factor as monatin.
at pH 4.0.
2.2.2. HPLC
Waters 600 series semi-preparative HPLC system (Waters Cor-
poration, Milford, MA, USA) was used with UV Detection, diode ar-
ray detector (190–400 nm) monitored at 280 nm, running on the
Empower software platform. HPLC method I employed Synergi Hy-
dro-RP column (250 ꢀ 10 mm, 4
l
m, p/n 00G-4375-N0, Phenome-
A large scale UV-degradation was carried out at the 50 lg/mL
nex, Torrance, CA, USA) with a binary solvent mobile phase (A,
water with 5 mM ammonium formate, pH 4.0; B, acetonitrile)
using gradient (12% B for 5 min, 12–34% B over 5 min, 34–95% B
over 10 min, 95% B for 5 min, 95–12% B over 0.1 min, 12% B for
4.9 min) at a flow rate of 5 mL/min. The injection volume was
level in lemon-lime matrix using 300 mL size transparent PET bot-
tles. The calculated concentration of monatin remaining after deg-
radation was found to be approximately 20%. The combined
relative abundance for monatin and the peaks selected for isolation
and characterisation in this study was 89.2%. This suggested that
either the response factor varied significantly between monatin
and one or more of the degradation products or some portion of
the degraded monatin was not observed as integrated peaks within
the selected chromatographic window. The degraded monatin
samples showed a precipitate that settled out on the bottom of
the bottles after they were left undisturbed at 4 °C for several days.
100–400
HPLC method II employed Atlantis T3 C18 column
m, p/n 1860,03,694, Waters Corporation, Mil-
lL.
(250 ꢀ 10 mm, 5
l
ford, MA, USA) and a binary solvent mobile phase (A, water with
0.05% trifluoroacetic acid (TFA); B, acetonitrile with 0.05% TFA)
using a gradient (0% B to 50% B over 30 min, 50–100% B over

M. Upreti et al. / Food Chemistry 131 (2012) 413–421
415
This precipitate may account for some portion of the degraded
monatin that was not observed in the UV chromatogram.
after exposure had presence of a strong musty/fecal like odour
which was hard to miss. This result was the same as described pre-
viously (Evans & Goulson, 2010) where degraded monatin solution
was reported to have the presence of 3-methyl indole (Skatole)
which has a characteristic musty/fecal like odour.
2.5. SPE (solid phase extraction) concentration and degradant isolation
The degraded monatin solution was subjected to SPE concentra-
tion step to provide a smaller volume of concentrated sample that
was largely free of acid and unretained components. One 300 mL
bottle of degraded monatin solution was passed through a 5 g
C18 SPE cartridge that had been washed with 3 column volumes
of methanol and equilibrated with three column volumes of H2O.
After addition of the degraded monatin solution, the SPE cartridge
was washed with 3 column volumes of H2O and then eluted with 3
column volumes of methanol. The methanol layer was collected
and concentrated to dryness by rotary evaporation under reduced
pressure. The resulting concentrated degradant fraction was taken
up in 1.0 mL of H2O for subsequent HPLC fractionation. This proce-
dure was repeated for each bottle of monatin solution.
Preliminary HPLC fractionation was carried out following meth-
od I. The individual degradation product peaks were pooled from
multiple runs to yield the crude impurity fractions. Preliminary
analysis indicated that the crude impurity fraction containing com-
pound 1 required additional separation. Therefore, a second round
of chromatography using HPLC method II was undertaken to fur-
ther purify the sample. However, it resulted in two peaks eluting
at 14.3 and 15.3 min and both had the same [M + H]+ ion as ex-
pected for compound 1. Furthermore, re-injection of either peak
resulted in a mixture of both peaks. This suggested that 1 exists
as a mixture of isomers that readily interconvert. Therefore both
peaks (14.3 and 15.3 min) were collected, combined and concen-
trated for further analysis.
Compounds 2, 3, and 7 isolated from the preliminary HPLC frac-
tionation were found to be of sufficient purity for structural char-
acterisation and so were used without further purification. Crude
impurity fractions containing compounds 4 and 6 required an
additional purification and a second round of chromatography
was undertaken using HPLC method III to further purify sufficient
samples for structural characterisation.
Various attempts to purify the crude impurity fraction contain-
ing compound 5 via HPLC were not very successful and the purified
material was always found to contain multiple components result-
ing from the apparent decomposition of 5. An attempt was made to
isolate 5 from the crude fraction using the LC–MS method and the
modified solvent phase A (5 mM ammonium formate, pH 4.0).
NMR analysis of 5 thus obtained did not produce any useful data
and only MS/MS fragmentation was obtained.
3.2. Identification of the degradation products of monatin
The sample of degraded monatin beverage when concentrated
via SPE generated useful MS data from the LC–MS analysis (Table
1). A sample of the base matrix used was also analysed under sim-
ilar conditions to exclude peaks which were already present in the
base matrix from further analysis (Fig. 2A). Under this method, the
monatin peak was observed to elute at 6.42 min (Fig. 2B). A num-
ber of potential monatin degradation product peaks were observed
to elute between 2 and 17 min in the UV chromatogram (Fig. 2C).
The nine largest peaks, including monatin, which could not be
attributed definitively to the base matrix, are thus labelled in
Fig. 2C. One peak observed at 10.25 min was not observed right
after degradation but grew overtime as the samples were stored
at 4 °C and so was not included in the analysis. Out of these nine
peaks seven of the resulting degradation products of monatin,
compounds 1–7 (peaks 1–7), were isolated and characterised using
a combination of MS, MS–MS and 1D and 2D NMR techniques.
In order to have a reference for structural characterisation of the
isolated degradants, a detailed NMR dataset including 1H NMR, 13
C
NMR, 1H–1H COSY, HSQC and HMBC spectra, was acquired for
monatin. Accurate mass analysis using HRESI+ gave an [M + H]+
ion at m/z 293.1142 and the fragmentation pattern was acquired
for this ion by ESI + TOF MS/MS. All these assignments were found
to be in agreement with data reported in literature (Bassoli et al.,
2005; Holzapfel et al., 1994).
3.2.1. Identification of the degradation product, compound 1
The molecular formula of compound 1 was deduced to be
C
14H16N2O6 on the basis of the ESI + LC–MS analysis which showed
an [M + H]+ ion at m/z 309 along with a fragment ion at m/z 291
[M + H–H2O]+. The [M + H]+ ion was observed at m/z 309.1093 in
the HRESI + TOF data which indicated that the mass of the
[M + H]+ ion was in good agreement with the molecular formula
(calculated for C14H17N2O6: 309.1087, error: 1.9 ppm). This molec-
ular formula contained the net addition of one oxygen atom rela-
tive to monatin.
The fragmentation pattern acquired by ESI + TOF MS/MS select-
ing the [M + H]+ ion at m/z 309.0 for fragmentation was similar to
that observed for monatin. A notable difference from the monatin
fragmentation was observed for the fragment ions attributable to
the indole moiety. A pair of fragment ions corresponding to the in-
dole moiety was observed at m/z 134 and 146. The molecular for-
mula for the ion at m/z 146 contained one additional oxygen atom
relative to the ion observed at m/z 130 for monatin. These data
indicated that the additional oxygen atom present in compound
1 must be located within the indole moiety. This was further sup-
ported by NMR analysis (Table 2).
3. Results and discussion
3.1. Photodegradation of monatin
Monatin in regular transparent PET bottles degraded by 70% in
just one day’s equivalent of UV exposure while it was completely
gone by four days equivalent of UV exposure. However, in the dark
PET bottle which was our control in the sun-test chamber, monatin
was stable even up to 92% after four days of UV exposure showing
that blocking UV light was useful in stabilising monatin. Under vis-
ible light (regular laboratory light), monatin degraded only very lit-
tle and was present up to 90% in normal as well as dark PET bottles
(Fig. 1).
It was observed that samples in regular PET packaging after UV
exposure changed colour, from colourless in the beginning to yel-
low after exposure. A group of trained in-house expert flavourists
carried out odour analysis of the degraded beverage samples and
confirmed that the samples before exposure had no malodor but
The 1H NMR of 1 confirmed the presence of two isomers which
resulted in the appearance of two sets of signals in the NMR data.
One set of signals was of somewhat higher relative abundance and
was designated as the major isomer. For compound 1 (major), the
a-methine proton H-2 was assigned to a multiplet at 3.66 ppm.
Methylene protons at H-3 were assigned to a doublet of doublets
at 1.82 ppm (J = 7.3, 14 Hz) and a multiplet at 2.07 ppm. Methylene
protons at H-5 were observed as a multiplet at 2.08 and a doublet
at 2.30 ppm (J = 14.6 Hz). COSY correlations were observed
between the H-2/H-3 protons while the H-5 protons did not show
any additional COSY correlations. Assignment of the indole region
1
between 6.77–7.39 ppm by H NMR and COSY correlations (H40/

416
M. Upreti et al. / Food Chemistry 131 (2012) 413–421
Fig. 1. (A) (2R, 4R) Monatin and its UV–vis spectrum. (B and C) Concentration of monatin as a function of time in regular PET and black PET bottles under UV light and visible
light.
H50, H50/H60, H60/H70) was straightforward. The singlet correspond-
ing to H-20, observed at 7.12 ppm for monatin, was clearly absent
from the 1H spectrum (and HSQC data) acquired for 1. In addition,
the indole NH was observed as a singlet at 10.33 ppm and showed
no correlation in the COSY spectrum. These data indicated that the
addition of oxygen in 1 had occurred at C-20 of the indole moiety
resulting in the presence of a hydroxyl group at this position for
the major isomer. The carbon resonances were attributed using
two dimensional HSQC data. Carbons C-2, C-3, C-5, C-40, C-50, C60
and C-70 were assigned to 50.8, 40.2, 36.9, 125.0, 120.9, 127.0
and 108.5 ppm through their direct 1H–13C correlations with the
corresponding protons.
The assignment of the minor isomer of 1 was made in a similar
fashion. An additional COSY correlation was observed between the
H-5 methylene protons at 1.92 ppm and a methine proton at
3.67 ppm which was assigned as H-30 of the indole. This suggested
that the position of the C-2/C-3 double bond in the indole moiety
had shifted to yield the structure for the minor isomer illustrated
in Fig. 3. Thus compound 1 was established as 2-hydroxy monatin
which is present as a mixture of isomers depending on the position
of the unsaturation (Fig. 3).
3.2.2. Identification of the degradation product, compound 2
The molecular formula of compound 2 was deduced to be
C
14H14N2O4 on the basis of ESI + LC–MS analysis which showed
an [M + H]+ ion at m/z 275 and an [M + H]+ ion at m/z 275.1044
in the HRESI + TOF mass spectrum. The molecular formula con-
firmed that it has a structure resulting from the net loss of H2O rel-
ative to monatin. Monatin is known to exist in equilibrium with its
lactam and lactone derivatives. Therefore, 2 could likely be either
of these 2 compounds. The fragmentation pattern for this ion ob-
tained by ESI + TOF MS/MS showed a notable difference from the
fragmentation pattern for monatin in the absence of a fragment
ion corresponding to loss of one or more H2O units. This suggested
that 2 may not have a free hydroxyl group and so could correspond
to the lactone (ester) of monatin which lacks a free hydroxyl group
at C-4. In contrast, monatin lactam retains the hydroxyl group at C-
4 and so would be expected to show loss of H2O. Fragment ions at

M. Upreti et al. / Food Chemistry 131 (2012) 413–421
417
Table 1
HPLC, MS and MS/MS data of monatin and its degradation products.
Peak Rt
(min)
m/z(ES+)
m/z(ESꢁ)
Chemical
formulae
Compound
[M + H]+ [M + Na]+ Fragments
[M ꢁ H]ꢁ Fragments
1
4.12
6.42
309
293
291, 273, 263, 246, 228, 146,
134
C
14H16N2O6
2-Hydroxy monatin
Monatin
275, 257, 247, 230, 212, 158,
130, 132
229, 212, 183, 168, 158, 118
201, 230, 212, 184, 174, 158,
132, 118
C14H17N2O5
C14H14N2O4
C
C
C
2
3
7.55
10.51
275
247
Monatin lactone
13H14N2O3
13H10N2O3
27H26N4O8
(R)-2-Amino-5-(1H-indol-3-yl)-4-
oxopentanoic acid
2-(2-Formamidophenyl) isonicotinic
acid
4
5
11.73
12.61
243
535
215
241
197, 179, 169, 118
517, 418, 400, 374, 293, 243, 533
160
515, 489, 471, 416,
372, 354
116
Unknown
6
7
8
9
14.41
14.91
16.12
16.31
C9H7NO2
C9H7NO
Indole-3-carboxylic acid
3-Formyl-indole
146
n.d.
199
118
144
yielding a ketone at this position. Five protons between 7.01 and
m/z 229.0982 [M + H–COOH]+ and at m/z 212.0734 [M + H–COOH–
NH3]+ suggested the presence of a free amine and further sup-
ported 2 to be the ester rather than the lactam. A sample of mon-
atin lactam for direct comparison was prepared following a
reported procedure (Ikeda et al., 2007) and its structure was con-
firmed in comparison with published 1H NMR data. Analysis of
the lactam sample by LC–MS indicated the presence of a peak with
an [M + H]+ ion at m/z 275 that elutes after 2 and is not observed as
a significant peak in the degraded monatin sample confirming that
2 cannot be the lactam. Analysis of the 1H NMR, COSY and HSQC
data further supported compound 2 to be the monatin lactone
derivative (Fig. 4).
1
7.49 ppm in H NMR and COSY correlations (NH/H20, H40/H50,
H50/H60, H60/H70) were characteristic of the indole region indicat-
ing that it was unchanged in 3. The carbon resonances were attrib-
uted using the HSQC data. Carbons C-2, C-3, C-5, C20, C-40, C-50, C60
and C-70 were assigned to 51.5, 42.2, 40.5, 119.2, 120.1, 122.5 and
112.2 ppm through their direct 1H–13C correlations with the corre-
sponding protons. Thus, 3 results from the loss of the carboxylic
acid functionality at C-4 of monatin and has a ketone at this posi-
tion. It was assigned to be 2-amino-5-(1H-indol-3-yl)-4-oxopenta-
noic acid (Fig. 3). We later found that this compound was reported
before in a Japanese patent as a synthetic intermediate to monatin
(Amino, Kawahara, Funakoshi, & Sugiyama, 2003). This is the first
time reporting detail structural characterisation including COSY
and HSQC NMR data on this compound.
3.2.3. Identification of the degradation product, compound 3
The molecular formula of compound 3 was deduced to be
C13H14N2O3 on the basis of ESI + LC–MS analysis which showed
an [M + H]+ ion at m/z 247 and HRESI+ which displayed [M + H]+,
[M + Na]+, and [M + 2Na–H]+ ions at m/z 247.1087, 269.0909, and
291.0728, respectively. Accurate mass analysis indicated that the
mass of the [M + H]+ ion was in good agreement with the molecu-
lar formula (calcd for C13H15N2O3: 247.1083, error: 1.6 ppm). Rela-
tive to monatin, this indicated a net loss of CH2O2 (46 Da)
suggesting that one of the carboxylic acid groups present in mon-
atin has been lost during the formation of this degradation product.
The fragmentation pattern acquired by ESI + TOF MS/MS selecting
the [M + H]+ ion at m/z 247.0 for fragmentation showed fragment
ions at m/z 201.1026, [M + H–CH2O2]+, at m/z 184.0762 [M + H–
CH2O2–NH3]+ and 230.0820 [M + H–NH3]+. The fragment ions at
m/z 158.0598 and 174.0917 suggested that the degradant may lack
the carboxylic acid group present at C-4 in monatin, but were not
conclusive. The fragment ions at m/z 118.0958 and 132.0800 corre-
spond to the indole moiety. The 1H NMR data analysis (Table 3) of 3
3.2.4. Identification of the degradation product, compound 4
The molecular formula of compound 4 was deduced to be
C13H10N2O3 on the basis of ESI + LC–MS analysis which showed
an [M + H]+ ion at m/z 243 and HRESI+ which displayed an
[M + H]+ ion at m/z 243.0771 which was in good agreement with
the molecular formula (calculated for C13H11N2O3: 243.0770, error:
0.4 ppm). Relative to monatin this indicated a net loss of CH6O2.
The ESI + TOF MS/MS fragmentation acquired by selecting the
[M + H]+ ion at m/z 243.0 was limited to a single fragment ion at
m/z 215.0822 [M + H–CO]+. Loss of CO was not observed before
as a significant fragment ion for monatin or any of the degradants
discussed previously. Fragmentation using negative mode ESI and
selecting the [M ꢁ H]ꢁ ion at m/z 241.0 resulted in fragment ions
at m/z 197.0714 [M ꢁ H–CH2O2]ꢁ suggesting that 4 has at least
one carboxylic acid moiety and further losses led to ions at m/z
179.0608 [M ꢁ H–CH2O2–H2O]ꢁ and 169.0763 [M ꢁ H–CH2O2–
CO]ꢁ.
indicated the presence of an a-methine proton at 3.81 ppm which
was assigned as H-2. Two methylene protons at 3.00 and 3.21 ppm
showing COSY correlation with H-2 were assigned as H-3 and
seemed to be shifted downfield relative to monatin. This downfield
shift could be due to the loss of the C-1 carboxyl group resulting in
an unsaturation between C-2 and C-3, but this was not observed.
Another, less likely, scenario would be that the carboxylic acid at
C-1 has been lost with the additional unsaturation necessary to
satisfy the molecular formula located elsewhere in the structure,
but this would have resulted in a methylene group at C-2 rather
Analysis of the 1H NMR data (Table 3) indicated a lack of upfield
signals and suggested a significant rearrangement of C-1 through C-
5. Eight aromatic protons were observed in the 1H NMR between
7.26 and 8.87 ppm. Four protons, two triplets at 7.26 (J = 7.9 and
7.45 ppm (J = 7.7), and two doublets at 7.73 (J = 7.9) and 8.24 ppm
(J = 8.5 Hz) corresponded to an aromatic spin system and showed
COSY correlations which were assigned to protons H-30 to H-60,
respectively. The 13C NMR values for all the carbons were assigned
on the basis of HSQC and HMBC spectra. Analysis of the 13C NMR data
indicated that a 13C signal around 110 ppm typical of the C-7 posi-
tion of the indole moiety was absent suggesting that the indole
group was absent or significantly modified. HMBC correlations be-
tween the aromatic singlet at 8.32 ppm to the quaternary carbon
at 135.6 ppm and the presence of a signal at160.0 ppm in the 13C
than the observed
a-methine. These data indicated that the car-
boxylic acid group at C-4 must have been lost during the formation
of compound 3. Two additional isolated methylene protons ap-
peared as a broad singlet (3.90 ppm) and were assigned as H-5
which indicated that the unsaturation must be present at C-4

418
M. Upreti et al. / Food Chemistry 131 (2012) 413–421
Fig. 2. (A) LC–MS analysis of the base matrix alone showing the UV (280 nm) chromatogram. (B) LC–MS analysis of monatin sample before degradation showing the UV
(280 nm) chromatogram. (C) 20ꢀ SPE concentrated monatin degradant sample showing the UV (280 nm) chromatogram.
spectrum correlated to a singlet at 10.95 ppm suggested the pres-
ence of an N-formyl group. This was supported by the MS data above
which showed loss of CO. These data suggested that the indole group
had undergone ring opening between C-20 and C-30. This N-formyl
phenyl ring was found to be attached to a substituted pyridine on
the basis of HMBC correlations. HMBC correlations from the proton
at 8.11 ppm to C-2 of the pyridine and the quaternary carbon at
127.9 ppm allowed it to be assigned as H-3 within the pyridine moi-
ety. This proton also showed an additional HMBC correlation to a
carbon at 166.0 ppm which was assigned as a carboxyl substituent
at C-4. The remaining pair of aromatic doublets (J = 4.9 Hz) at 7.83
and 8.87 ppm showed a COSY correlation to each other and were as-
signed as H-5 and H-6, respectively. The various COSY and HMBC
correlations thus identified were (H-30/H-40, H-40/H-50, H-50/H-60,
H-5/H-6) and (CHO/C-20, H-30/C-10, H-50/C-10, H-60/C-20, C-2, H-3/C-
10 ,C-2, COOH, H-5/C-3, COOH, H-6/C-2, C-4), respectively. Therefore,
the structure of 4 was established as 2-(2-formamidophenyl)
isonicotinic acid (Fig. 3).
3.2.5. Identification of the degradation product, compound 5
The molecular formula for 5 was deduced to be C27H26N4O8
from the ESI + LC–MS analysis which an [M + H]+ ion at m/z 535.
The HRESI + TOF mass spectrum showed the accurate mass for
the [M + H]+ ion at m/z 535.1832 which was in good agreement
with the molecular formula (calculated for
C27H27N4O8:
535.1829, error: 0.6 ppm). Relative to monatin, this indicated a
net addition of C13H10N2O3. The fragmentation pattern acquired
by ESI + TOF MS/MS selecting the [M + H]+ ion at m/z 535.0 for frag-
mentation resulted in a fragment ion observed at m/z 293.1167
which corresponded to monatin and suggested that one half of
the structure is composed of monatin. The reciprocal fragment
ion was observed at m/z 243.0804 and had the same molecular

M. Upreti et al. / Food Chemistry 131 (2012) 413–421
419
Table 2
1H and 13C NMR data of the major and minor isomer of compound 1.
Degradant position no.
Compound 1 major (DMSO-d6)
dH 1H (ppm); J (Hz)
Compound 1 minor (DMSO-d6)
dH 1H (ppm); J (Hz)
d
13C
d
13C
2-Amino-4-hydroxy pentanedioic acid residue
1
–
–
–
–
2
3
4
3.66; m
1.82; dd (7.3, 14.0) 2.07; m
–
50.8
40.2
–
3.31; d (7.3)
1.65; dd (7.3, 14.0) 2.36; m
–
41.8
40.8
–
5
2.08; m 2.30; d (14.6)
36.9
1.92; dd (12.2, 15.3) 2.34; m
37.8
Indole moiety
10
10.33; s
–
–
7.39; d (7.3)
6.88; t (7.3)
7.12; t (7.3)
6.77; d (7.9)
–
–
–
–
10.26; s
–
–
–
40.5
126.1
120.9
127.0
108.4
–
20
30
3.67; m
7.58; d (7.3)
6.88; t (7.3)
7.10; t (6.7)
6.74; d (7.9)
–
40
125.0
120.9
127.0
108.5
–
50
60
70
3a0
7a0
–
–
–
–
Fig. 3. Compounds 1, 3 and 4 and their key COSY and HMBC correlations.
formula as compound 4. Fragment ions were also observed corre-
sponding to loss of an indole moiety ion at m/z 418.1268 and net
loss of an indole and CO2 at m/z 374.1406. Loss of H2O from the
[M + H]+ ion or the ion at m/z 418.1268 provided fragment ions
at m/z 517.1716 and 400.1158, respectively.
one half of the structure composed of a unit of monatin and the
second half having the same molecular formula as compound 4.
We hypothesise that the monatin moiety is likely attached through
the amine at C-2 as represented in Fig. 4.
The fragmentation pattern acquired by ESI-TOF MS/MS select-
ing the [M ꢁ H]ꢁ ion at m/z 533.0 resulted in fragment ions from
several losses of H2O and CH2O2 at m/z 515.1572, 489.1742, and
471.1662 besides fragment ions at m/z 416.1203 [M ꢁ H–indole
moiety]ꢁ, m/z 372.1216 [M ꢁ H–indole moiety–CO2]ꢁ and with
further loss of H2O yielding a fragment ion at m/z 354.1101. No
fragments corresponding to two halves were seen as was the case
in positive mode fragmentation.
3.2.6. Identification of the degradation product, compound 6
Compound 6 was given the molecular formula of C9H7NO2 on
the basis of ESI + LC–MS analysis which showed the [M ꢁ H]ꢁ ion
at m/z 160 in the positive ion mode. Accurate mass from HRESI-
TOF mass spectrum showed an [M ꢁ H]ꢁ ion at m/z 160.0392 (cal-
culated for C9H6NO2: 160.0399, error: ꢁ4.4 ppm). The fragmenta-
tion pattern acquired by ESI-TOF MS/MS selecting the [M ꢁ H]ꢁ
ion at m/z 160.0 was dominated by a single fragment ion at m/z
116.0504 resulting from the loss of CO2 and suggested the presence
of a carboxylic acid functionality in the structure of the degradant.
Analysis of the 1H NMR data indicated a lack of upfield signals and
Due to the limited stability of this degradant, NMR data were
not of use for assigning the structure. Thus, compound 5 is pro-
posed to have a structure which is a partial dimer of monatin with

420
M. Upreti et al. / Food Chemistry 131 (2012) 413–421
COOH
COOH
NH2
OH
O
O
N
H
Monatin
NH2
N
H
O
HO
6
O
COOH
COOH
NH2
COOH
N
H
N
H
OH
OH
7
2
N
H
COOH
NH2
1
O
O
COOH
NH
COOH
NH
N
H
OH
OH
3
C13H11N2O3
O
N
N
H
4
5
Fig. 4. Proposed monatin degradation pathway under UV conditions.
Table 3
1H and 13C NMR data of of compound 3 and 4.
Degradant position
Compound 3 (CD3OD)
dH 1H(ppm); J(Hz)
Compound 4 (DMSO-d6)
dH 1H(ppm); J(Hz)
d
13C
d
13C
1
–
–
2
3
4
5
3.81; d (7.1)
3.00; dd (7.1, 18.5) 3.21; d (18.5)
–
51.5
42.2
–
–
158.0
122.0
139.5
121.0
149.5
166.0
127.9
135.6
122.3
129.5
124.3
129.7
160.0
8.11; s
–
7.83; d (4.9)
8.87; d (4.9)
–
3.90; s
40.5
6
C4-COOH
10
–
–
–
20
7.20; s
–
7.49; d (7.8)
7.01; t (7.8)
7.09; t (7.6)
125.1
–
119.2
120.1
122.5
30
8.24; d (8.5)
7.45; t (7.7)
7.26; t (7.9)
7.73; d (7.9)
10.95; s, 8.32; s
40
50
60
NHCHO
70
7.34; d (7.8)
112.2
3a0
7a0
suggested it to be a 3-substituted indole structure. A comparison
with the commercially available compound (Sigma–Aldrich
284734, indole-3-carboxylic) provided a final confirmation of the
structure. The 1H NMR spectrum for the commercial standard
was found to be identical to that of the isolated sample. LC–MS
analysis indicated that compound 6 and the authentic sample of
indole-3-carboxylic acid have the same retention time and give a
single peak upon co-injection. Thus, the structure of compound 6
was assigned as indole-3-carboxylic acid.
146.0600 (calculated for C9H8NO: 146.0606, error: ꢁ4.1 ppm).
The fragmentation pattern acquired by ESI + TOF MS/MS selecting
the [M + H]+ ion at m/z 146.0 for fragmentation was dominated
by a single fragment ion at m/z 118.0584 [M + H–CO]+ suggesting
the presence of a formyl group in the structure of the degradant.
1H NMR data indicated a lack of upfield signals and suggested a
structure composed predominately of a 3-substituted indole simi-
lar to that observed for compound 6. The presence of a formyl
group was further confirmed by the presence of a singlet at
9.92 ppm in the 1H spectrum which was correlated to an aldehyde
carbon at 184.6 ppm in the HSQC spectrum. A comparison with the
commercially available compound (Sigma–Aldrich 12944-5, 3-for-
mylindole) was carried out. A 1H NMR spectrum for the commer-
cial standard was obtained and was found to be identical to that
of our isolated sample. LC–MS analysis indicated that compound
3.2.7. Identification of the degradation product, compound 7
Compound 7 was deduced to have a molecular formula C9H7NO
based on the ESI + LC–MS analysis which showed the [M + H]+ ion
at m/z 146 in positive ion mode. Accurate mass analysis from
HRESI + TOF mass spectrum showed an [M + H]+ ion at m/z

M. Upreti et al. / Food Chemistry 131 (2012) 413–421
421
7 and the authentic sample have the same retention time and give
a single peak upon co-injection. The structure of 7 was thus as-
signed as 3-formylindole.
have identified seven compounds as major degradation products
of monatin. Structures could be unambiguously determined for
six of the seven degradants. Two novel compounds were identified
namely 2-hydroxy monatin and 2-(2-formamidophenyl) isonicoti-
nic acid. A final structure for one of the degradation product has
been only partially determined due to its instability during the
purification. Two of the degradants 3-formyl indole and 3-carboxyl
indole were confirmed by direct comparison with authentic com-
mercial standards. Two remaining degradation products found
were monatin lactone and 2-amino-5-(1H-indol-3-yl)-4-oxopenta-
noic acid. Sensory analysis of the commercially available samples
of 3-formyl indole and 3-carboxyl indole indicated that former
had Skatole like odour same as observed in the sensory analysis
of the degraded monatin solution.
3.3. Sensory analysis on 6 and 7
Out of all these degradants, in-house sensory analysis was car-
ried out on commercially available 3-formyl indole and 3-carboxyl
indole in aqueous solution. In literature (Tolonen et al., 2000) 3-
formyl indole has been reported as a flavour enhancing compound.
It was found that in 0.1% (wt./vol) solution, 3-formyl indole had
Skatole like ‘musty’ smell same as was observed in the sensory
analysis of the degraded monatin solution while 3-carboxyl indole
had no such off flavour notes.
3.4. Possible mechanism and pathway for photodegradation of
monatin
Acknowledgements
The authors wish to thank Dr. Robert Peterson for providing
valuable sensory inputs.
Monatin is known to exist in equilibrium with its lactam and
lactone derivatives. However, in the present study no lactam deriv-
ative was found before or after the degradation. Only the lactone
derivative was found to be present to some extent (ꢂ1%) in the
monatin sample before degradation but increased in concentration
to about 4% after degradation.
References
Amino, Y., Kawahara, S., Funakoshi, T., Sugiyama, M. (2003). Method for preparation
of monatin derivative and its new intermediates. JP 2003171365.
Amino, Y., Hirasawa, K. (2005). Preparation of organic amine salts of glutamic acid
derivative (monatin), and their application as sweetening agents. EP 1533300.
Amino, Y., Hirasawa K. (2006). Method for producing monatin or salt thereof. WO
2006011613.
Bassoli, A., Borgonovo, G., Busnelli, G., Morini, G., & Drew, M. G. B. (2005). Monatin
and its stereoisomers: Chemoenzymatic synthesis and taste properties.
European Journal of Organic Chemistry, 8, 1652–1658.
Buddoo, S., Rousseau, A.L. Brady, D., Lalonde, J., Yao, Y., Wang, Y. (2009).
Chemoenzymic process for the production of monatin stereoisomers. WO
2009045383.
Buddoo, S., Rousseau, A.L., Gordon, G.E.R. (2009). Production of monatin
enantiomers. WO 2009044245.
Brady, D., Steenkamp, L.H., Rousseau, A.L., Buddoo, S., Steenkamp, P.A. (2009).
Production of monatin enantiomers. WO 2009044244.
Clos, J. F., DuBois, G. E., & Prakash, I. (2008). Photostability of rebaudioside A and
stevioside in beverages. Journal of Agricultural and Food Chemistry, 56,
8507–8513.
Evans, J.C., Goulson M.J. (2010). Shelf-stable monatin-sweetened beverage. WO
2010138513.
Holzapfel, C. W., Bischofberger, K., & Olivier, J. (1994). A simple cycloaddition
Thus under the present conditions of low pH in lemon–lime ma-
trix (pH = 2.6) and under sun-simulated exposure, we found mon-
atin degrading into seven main compounds. Compounds 1 and 4
were characterised fully as novel molecules but compound 5 was
found to be unstable and could not be characterised completely.
Major degradation products were compounds 1 and 7 with 13%
and 6% of the original starting concentration of monatin, respec-
tively. The possible pathway proposed for the degradation of mon-
atin is based on the structural identities of the various degradation
products that were obtained (Fig. 4). Oxidation of monatin resulted
in formation of 1 first which probably undergoes significant rear-
rangement by opening of the indole ring between C-20 and C-30
and ultimately yielding an N-formyl group derived from C-20. The
substituted pyridine appears to result from bond formation be-
tween the monatin amine and C-30 of the indole group. Compound
3 probably resulted from loss of CO and water from monatin. We
speculate that compound 5 is probably the result of an addition
of compound 4 to monatin through its amine at C-2 based upon
our partial characterisation results. Monatin probably also photo-
degrades to give 3-formyl derivative. There are reports in the liter-
ature about 2, 3-substituted indole derivatives giving correspond-
ing formyl derivatives upon photoxidation in acidic conditions
(Mudry & Frasca, 1973). Therefore, monatin and even 2 and 3
may oxidise to 6 which can further oxidise to 7.
approach to
a racemate of the natural sweetener monatin. Synthetic
Communications, 24, 3197–3211.
Ikeda, R., Kogiso, K., & Nakamura, K. (2007). Searching the functional group for
sweet expression and its reactivity. Nippon Aji to Nioi Gakkaishi, 14, 419–422.
Mori, K. (2006). Sweetener composition comprising monatin or a salt thereof and a
radical scavenger. US 20060083695.
Mudry, C. A., & Frasca, A. R. (1973). Photoxidation of indole derivatives. Tetrahedron,
29, 603–613.
Nakamura, K., Baker, T. J., & Goodman, M. (2000). Total synthesis of monatin.
Organic Letters, 2, 2967–2970.
Roy, G.M. (2009). Beverage composition and method of reducing degradation of
monatin, WO 2010036576.
Tolonen, M., Taipale, M., Viander, B., Pihlava, J.-M., Korhonen, H., & Ryhanen, E.-L.
(2000). Effect of fermentation on plant derived biomolecules in cabbage.
Mededelingen- Faculteit Landbouwkundige en Toegepaste Biologische
Wetenschappen, 65, 595–597.
4. Conclusion
Vleggaar, R., Ackerman, L. G. J., & Steyn, P. S. (1992). Structure elucidation of
monatin, a high-intensity sweetener isolated from the plant Schlerochiton
ilicifolius. Journal of the Chemical Society Perkin Transactions 1: Organic and Bio-
Organic Chemistry (1972–1999), 22, 3095–3098.
Weast, R. C. (1983). CRC handbook of chemistry and physics (64th ed.). Boca Raton, FL:
CRC Press Inc., p. F-92.
In present work the degradation of monatin in a model lemon–
lime beverage system has been studied to assess the viability of
this sweetener in our product conditions. We have found that upon
UV exposure monatin is unstable in a regular transparent PET
bottle while it survived well in the control dark PET bottle. We