CHEMISTRY & BIODIVERSITY Vol.

3 (2006)

A Broad Diversity of Volatile Carboxylic Acids, Released by a Bacterial

Aminoacylase from Axilla Secretions, as Candidate Molecules for the
Determination of Human-Body Odor Type
by Andreas Natsch*, Samuel Derrer, Felix Flachsmann, and Joachim Schmid
Givaudan Schweiz AG, Ueberlandstrasse 138, CH-8600 Duebendorf, Switzerland
(phone: 41-44-824-2105, fax: 41-44-824 2926, e-mail: andreas.natsch@givaudan.com)

Human body odor is to a large part determined by secretions of glands in the axillary regions. Two
key odoriferous principles, 3-methylhex-2-enoic acid (3MH2; 4/5) and 3-hydroxy-3-methylhexanoic acid
(HMHA; 6) have been shown to be released from glutamine conjugates secreted in the axilla by a
specific Na-acyl-glutamine aminoacylase (N-AGA) obtained from axilla isolates of Corynebacteria sp.
However, the low number of different odorants reported in humans stands in contrast to the observed
high inter-individual variability in body odors. Axilla secretions of individual donors were, therefore,
analyzed in detail. The secretions were treated with N-AGA, analyzed by GC/MS, and compared to
undigested controls. Over 28 different carboxylic acids were released by this enzyme from odorless axilla
secretions (Table 1). Many of these body odorants have not been reported before from a natural source,
and they include several aliphatic 3-hydroxy acids with 4-Me branches, 3,4-unsaturated, 4-Et-branched
aliphatic acids, and a variety of degradation products of amino acids. The odor threshold of some of the
acids was found to be in the range of 1 ng. Most of these compounds were present in all donors tested, but
in highly variable relative amounts, and they are, thus, candidate molecules as key components of a
compound odor determining the individual types of human body odor.

1. Introduction. Body odors have been shown to be a key criterion in mate

selection in a variety of species. Most intriguingly, the odor-dependent mate selection
leads to couples with dissimilar alleles in the major-histocompatibility-complex (MHC)
loci, thus maximizing the frequency of MHC-heterozygotes in the offspring, which
prevents inbreeding. This phenomenon has been investigated in detail in rodents [1],
but recently also in sticklebacks and sand lizards [2] [3]. In addition, several studies
with humans judging the body odors on worn T-shirts point to a preference for body
odors of MHC-dissimilar persons [4] [5]. All these studies led to the conclusion that
individuals have a distinct body-odor type, which is determined, at least partly, by their
inherited MHC alleles, although the underlying mechanism is unknown. In most of
these studies, behavioral responses to complex body secretions (urine in mice, and
sweat on worn T-shirts in the human studies) were recorded, but the chemistry of these
secretions and the mechanisms involved in their formation have received relatively
little attention.
Mice can recognize the body odor of a potential mate with dissimilar MHC based
on urine samples [6], and Singer et al. [7] have made attempts to characterize the
chemistry of the odor determinants. They found different, mostly unidentified
2006 Verlag Helvetica Chimica Acta AG, Zrich

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

carboxylic acids (such as phenylacetic acid) to be important discriminators, whose

relative abundance is associated with different MHC types, and they concluded that the
type of body odor results from a compound odor of several acids, with the relative
abundance of different acids, but not their presence/absence, generating the body-odor
differences in mice. In addition, Yamazaki et al. [8] have shown that mice can
discriminate the odor of blood-plasma samples after treatment with the proteolytic
enzyme pronase, but no discrimination is achieved with untreated plasma samples. In
mouse urine, phenylacetic acid is secreted as a conjugate with the amino acids glycine
and taurine [9] and, thus, amino acid conjugates are likely candidates to be odorant
precursors.
In humans, the strong odor emanating from the axillary region appears to make the
most-important contribution to the individual body odor, and, therefore, several
studies have addressed its chemical composition. The focus of early studies on the
odoriferous principles in human axilla secretions was mainly on odoriferous steroids
after the boar pheromone (5a-androst-16-en-3-one), and a related odoriferous steroid
(5a-androst-16-en-3a-ol) had been detected in human axilla secretions [10] [11]. The
first detailed analysis of the chemical composition of axillary odor was presented by
Zeng et al. [12]. They concluded that (E)-3-methylhex-2-enoic acid (3 M2 H) is the key
odor component, but they also reported several other carboxylic acids. In our previous
work [13], we had shown that 3 M2 H is found in hydrolyzed axilla secretions along
with its hydrated analogue (RS)-3-hydroxy-3-methylhexanoic acid (HMHA). It was
reported by others [14] that HMHA has a very pungent axillary odor, and it was
confirmed to be the most-abundant odorant in axilla secretions. However, these two
dominant odorants alone, found in different studies, cannot yet rationalize the interindividual differences and the individual types of body odor.
Sweat that is secreted by axillary glands is odorless, and the action of skincommensal Corynebacteria is needed to generate the odoriferous principles from nonsmelling molecules present in these secretions [15 17]. We had previously isolated
non-odoriferous precursors and specific enzymes of Corynebacteria that transform
them into volatile substances: a specific Zn-dependent Na-acyl-glutamine aminoacylase
(N-AGA) triggers the release of the acids 3M2 H and HMHA from Na-acyl-glutamine
conjugates secreted in the axilla [13], and a b-lyase releases sulfanyl alcohols from
synthetic cysteine conjugates and from axilla secretions [18]. In addition, sulfanyl
alcohols were recently found to be bound to the Cys-Gly dipeptide in axilla secretions
[19].
Here, we report the identification of a series of novel volatile acids contributing to
human body odor. We show that all these compounds can be released from odorless
axilla secretions by the same N-AGA isolated from human skin bacteria and present
evidence that they are secreted as glutamine conjugates in the axilla. In addition, we
show that their relative abundance varies widely between individuals and, therefore, a
compound odor of these different acids is a likely candidate for determination of
individual types of human body odor.
2. Results. 2.1. Identification of Carboxylic Acids in Human Axilla Secretions
Treated with Na-Acyl-Glutamine Aminoacylase (N-AGA). The cotton pads sampled
from five male donors during physical exercise were almost odorless immediately after

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

collection, and no major odoriferous compounds could be extracted from the aqueous
extracts without hydrolysis. In contrast, the lyophilized, resuspended extracts treated
with N-AGA released a very pungent odor. The organic phases obtained from
extraction of both N-AGA-treated and -untreated, acidified samples of axilla
secretions were, therefore, first evaluated olfactorily by gas chromatography (GC),
using an apparatus equipped with a sniff port. Zones or distinct peaks of effluents that
reminded of axillary odor were marked on the gas chromatograms. These fractions
were then further investigated by GC/MS. A typical chromatogram obtained from an
individual panelist is shown in Fig. 1 for an enzymatically treated sample (a) and the
corresponding untreated sample (b). Most of the signals marked in Fig. 1, a could be
unequivocally assigned to individual compounds (1 27) 1). For a few, however, only
general structure types were identified (see below).
The dominant compounds (Table 1) in terms of abundance in all donors were the
known acids (E)- (4) and (Z)-3 M2 H (5), and HMHA (6), the latter being more
abundant in all donors. Many of the other GC/MS peaks could be attributed to known
compounds on the basis of their MS properties, as corroborated by comparison to an inhouse and a commercial (NIST02) library. However, several GC/MS signals did not
match any known fragmentation pattern. Based on GC observations (unsymmetrical
peak shape typical for rather polar compounds on nonpolar columns) and MS
characteristics (key mass fragment at m/z 89 accompanied with only weak signals in the
higher-mass range for a whole series of unsymmetrical peaks), several of the unknown
compounds were postulated to be 3-hydroxy acids, while others were presumed to be
unsaturated congeners in position 3, and possibly branched in position 4 or higher. In
total, 19 different reference compounds with these structural features and matching
molecular weights were, therefore, synthesized.
By comparison of the GC/MS data of the axillary probes with those of the freshly
synthesized samples, some of the unknown components could be unequivocally
identified as 3-hydroxy-4-methylhexanoic acid (7), 3-hydroxy-4-methylheptanoic acid
(14), 3-hydroxy-4-methyloctanoic acid (19), 3-hydroxyoctanoic acid (17), and (Z)- and
(E)- 4-methyloct-3-enoic acid (11 and 13, resp.). Based on reference spectra, two
further hydroxy acids, namely 3-hydroxy-3-methylheptanoic acid (12) and 3-hydroxy-3methyloctanoic acid (18a) were also identified. All these novel compounds were only
found to be present in samples treated with N-AGA, and for all of them, a glutamine
conjugate appears, therefore, to be the precursor secreted in the axilla.
A few more GC/MS signals could not be identified by comparison with existing
spectra or reference samples, but through the application of the principles of homology
to the observations of both the GC properties (more or less equidistant retention times
tR between homologues) and the mass-spectroscopic features (key peaks, indication of
molecular weight). The following compounds belong to this group of hypothetically
identified homologues: (Z)- and (E)-4-methylnon-3-enoic acid (16 and 18b, resp.), 3hydroxy-4-methylnonanoic acid (21), and 3-hydroxydecanoic acid (23).
Furthermore, two diacids, octanedioic acid (24) and nonanedioic acid (27), as well
as the corresponding partially reduced compound 9-hydroxynonanoic acid (25), were
identified based on spectroscopic-library matches in enzyme-treated, but not in
1)

For convenience, the peak numbers were also taken as compound numbers (see Table 1).

Fig. 1. GC/MS Analysis of Hydrolyzed Axilla Secretions. a) Total-ion-current chromatogram of a typical sample of fresh axilla secretion from one individual
donor after treatment with the enzyme N-AGA, and b) without enzymatic treatment. For associated compounds, see Table 1.

4
CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

3-Hydroxy-4-methylhexanoic acid

3-Hydroxy-3-methylheptanoic acid

3-Hydroxy-4-methylheptanoic acid

3-Hydroxyoctanoic acid

3-Hydroxy-3-methyloctanoic acid

3-Hydroxy-4-methyloctanoic acid

16.33

17.94

19.18

20.29

21.17

22.27

12

14

17

18a

19

Name

3-Hydroxy-3-methylhexanoic acid

tR [min]

a) 3-Hydroxy Acids
6
15.65

Compound a )
Structure

5.09

0.09

1.48

4.42

0.99

0.18

0.12

0.42

100.00

6.43

0.35

0.73

100.00

0.49

0.39

6.66

0.70

0.16

0.34

100.00

Relative abundance [%]

0.20

0.48

1.30

0.74

0.20

0.24

100.00

1.86

1.28

1.37

3.16

0.58

0.78

100.00

Table 1. GC Retention Times (tR ), Structures, and Relative Amounts of Constituents Enzymatically Released by N-AGA from Odorless Axilla Secretions
Taken from Five Different Donors ( A E ). The axilla secretions were treated with N-AGA, extracted under acidic conditions, and analyzed by GC/MS (for
details, see Exper. Part). The percentage of the individual constituents, listed according to chemical class, is given relative to the most-abundant compound,
i.e., HMHA (6), which was arbitrarily set to 100%.

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

5

( E )-3-Methylhex-2-enoic acid

4-Methyloct-4-enoic or 4-methylideneoctanoic acid

( Z )-4-Methyloct-3-enoic acid

( E )-4-Methyloct-3-enoic acid

( Z )-4-Methylnon-3-enoic acid

12.71

16.79

17.69

18.17

20.01

11

13

16

3-Hydroxydecanoic acid

26.73

23

( Z )-3-Methylhex-2-enoic acid

3-Hydroxy-4-methylnonanoic acid

25.43

21

b) Unsaturated Acids
4
11.16

Name

tR [min]

Compound a )

Table 1 (cont.)
Structure

0.89

4.27

1.22

0.14

24.24

1.36

0.72

3.71

0.04

0.26

0.15

10.42

0.45

0.20

0.07

0.00

0.57

0.18

0.17

41.06

1.57

1.15

1.70

Relative abundance [%]

0.09

0.37

0.10

14.64

0.54

1.05

2.70

0.74

0.35

16.91

0.87

0.16

1.24

6
CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

3-Methyl-2-oxopentanoic acid

4-Methyl-2-oxopentanoic acid

4-Ethylheptanoic acid

Phenylacetic acid

4-Ethyloctanoic acid (goat acid)

7.68

7.81

16.93

17.55

19.84

10

15

c) Amino Acid Degradation Products and Miscellaneous Acids

1
7.10
2-Hydroxypropanoic acid (lactic acid)

3.60

n.i. b )

26.12

22

0.99

( E )-4-Methylnon-3-enoic acid

21.17

0.13

0.04

1.98

0.01

0.71

5.52

0.61

0.09

1.21

0.42

0.68

2.17

42.29

B
0.39

0.25

0.61

2.45

28.53

23.87

Relative abundance [%]

18b

Structure

Name

tR [min]

Compound a )

Table 1 (cont.)

5.17

0.48

0.06

0.11

2.39

7.99

29.72

0.85

0.29

0.22

0.56

2.53

9.74

7.16

1.28

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

7

Octanedioic acid (suberic acid)

9-Hydroxynonanoic acid

(4-Hydroxyphenyl)acetic acid

Nonanedioic acid (azelaic acid)

27.08

27.25

27.61

29.92

24

25

26

27

Structure

) Compound and peak numbers (see Fig. 1) are identical. b ) Not identified ( Mr 154 g/mol).

8-Hydroxyoctanoic acid

24.06

20

Name

tR [min]

Compound a )

Table 1 (cont.)

1.77

0.06

0.08

6.56

1.16

1.10

0.36

0.42

0.27

2.24

0.82

3.17

0.89

0.39

Relative abundance [%]

A

0.98

2.67

0.65

0.19

0.15

0.32

0.58

0.23

0.11

0.21

8
CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

untreated samples. 8-Hydroxyoctanoic acid (20) is a related compound postulated on

the basis of its MS properties (homology to 25) and GC retention time. For the
dominant peak 22 (tR 26.12 min, Mr probably 154), the structure could not yet be
identified, but the most likely hypothesis is a doubly unsaturated 4-methyloctanoic
acid. The minor peak corresponding to 8 (tR 16.79 min) could also be described only by
a structural hypothesis, i.e., to an unsaturated 4-methyloctanoic acid, spectroscopically
similar to 11 and 13, with the position of the CC bond not clearly assigned.
Based on a library search, the putative amino acid degradation products 3-methyl-2oxopentanoic acid (2), 4-methyl-2-oxopentanoic acid (3), phenylacetic acid (10), and
(4-hydroxyphenyl)acetic acid (26) could be clearly identified to be present in enzymetreated axilla secretions, but not, or only in traces, in control samples. Finally, the
previously reported [12] 4-ethylheptanoic (9) and 4-ethyloctanoic acid (15) were found
in the enzyme-treated samples only. Therefore, it is very likely that these compounds
are also secreted as glutamine conjugates. A special case is lactic acid (1). For three
panelists, this compound was only present in enzyme-treated samples, which indicates
the presence of a glutamine conjugate for this simple metabolite; however, in two
panelists, the lactic acid level in the untreated sample was already high, but still further
enhanced by enzymatic treatment. It, thus, appears to be secreted in the axilla both in
the free and the glutamine-bound form.
In summary, the compounds specifically released by N-AGA can be grouped into a)
3-hydroxy acids (6, 7, 12, 14, 17, 18a, 19, 21, 23), b) unsaturated, branched acids (4, 5, 8,
11, 13, 16, 18b, 22), and c) amino acid degradation products from Leu, Ile, Phe, and Tyr,
as well as some miscellaneous structures such as Et-branched acids, dioic acids, and
terminally hydroxylated acids (1 3, 9, 10, 15, 20, 24 27). For each of these compounds,
the peak area for each donor was calculated relative to the dominant peak (HMHA;
6), and listed in Table 1. It can clearly be seen from this analysis, that almost all
compounds were present in detectable amounts in samples of all donors tested, but that
the relative abundances differed, indicating that inter-individual odor differences
appear to be determined by the relative proportions of the same compounds, rather
than to the presence or absence of certain constituents.
To verify that the release of all the novel acids is, indeed, due to N-AGA, and not to
a potentially contaminating minor enzyme from the preparation of N-AGA, some
samples were treated with N-AGA in combination with the N-AGA-specific inhibitor
4-carbamoyl-2-[(3,7-dimethyloctyl)(hydroxy)phosphinoylmethyl]butyric acid, and
compared to the samples treated with N-AGA only. The amount of the novel acids
formed in the presence of the inhibitor was reduced by at least 80%, indicating that
their formation, indeed, is due to N-AGA activity. For example, the amount of 14 was
reduced by 98.4%, and the amounts of 17, 19, and 10 were reduced by 88.8, 96.7, and
73%, respectively, in the inhibitor-treated samples compared to the samples treated
with the enzyme only.
2.2. Enzymatic Treatment and Analysis of Other Body Secretions. The secretion of
such a wide variety of different glutamine conjugates in the axillary region leads to the
question whether these compounds are systemically present in the blood and secreted
by sweat in all body regions or in the urine. Sweat samples obtained from the neck, the
chest, the back, and the face were, thus, sampled on cotton pads, extracted, and treated
with the enzyme N-AGA, as described for the axillary samples. In addition, blood-

10

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

plasma and urine samples were acid-extracted with BuOH, and the extracts were
treated with N-AGA.
When N-AGA-treated sweat samples from other body regions were compared to
untreated controls, no additional peaks were detected, and not even trace amounts of
HMHA (6), the dominant compound in axilla sweat, were found. Lactic acid was
detected in high amounts in sweat from other body regions, but the amount was not
enhanced by N-AGA treatment, indicating that, contrary to the axilla, it is released in
the free form only at other body sites. N-AGA did also not release detectable amounts
of HMHA or other acids from the BuOH extracts of plasma samples, although the
method was sensitive to low amounts of the HMHA glutamine conjugate in spiked
plasma samples. In urine extracts treated with N-AGA, a very large peak of
phenylacetic acid (10; ca. 0.9 mm in urine) was identified, which was not present in
untreated urine extracts, but no other newly formed GC/MS peaks were observed. The
glutamine conjugates for the diverse acids reported above appear, therefore, to be
secreted specifically in the axilla, and only the glutamine conjugate of 10 is excreted in
urine.
2.3. Synthesis and Sensory Data for Some Novel Acids Identified in Axilla Secretions.
The novel acids were synthesized as described in the Exper. Part. All these compounds
are strongly odorant, although, as individual compounds, they do not only remind of
axillary odor, as does HMHA (6). The isomeric mixture of 11 and 13 is perceived as
acidic, green, and fresh, and 14 is described by the panelists as green, powdery, fruity,
and woody. For these latter two compounds and for 3 M2 H (4/5), GC odor-threshold
values (sniff analysis) were determined. For 3 M2 H, the average odor threshold was
1.9 ng. For the isomeric mixture 11/13, a very similar detection threshold of 2.2 ng was
found. For the hydroxy acid 14, a threshold of 10.9 ng was determined. However,
dynamic dilution olfactometry gave rise to a significantly lower value 2 ). An average
panelist was able to perceive a concentration of 0.17 ng/l of 14, with some panelists still
smelling a concentration of 0.04 ng/l.
2.4. Synthesis of Glutamine Conjugates of Two Novel Key Odorants, and Cleavage
by Axilla Bacteria and N-AGA. For the isomeric mixture 11/13 and for 14, the
corresponding glutamine conjugates 28 and 29 were prepared from the pertinent
carboxylic acids by means of DCC (dicyclohexylcarbodiimide) activation, followed by
reaction with l-glutamine in the presence of Et3N. It is noteworthy to mention that both
the activation and coupling with the hydroxy acid 14 could be carried out without
protection of the secondary OH function.

2)

Note that the odor threshold determined by these two methods is defined differently. For details,
see Exper. Part.

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

11

The conjugates 28 and 29 were fed to stationary cultures of various strains of

Corynebacterium and Staphylococcus species, which had originally been isolated from
the human axilla. As for the conjugates reported before [13], Corynebacteria, but not
Staphylococci, did release the corresponding acids (Table 2). Interestingly, the velocity
of the cleavage was clearly higher for the two novel substrates as compared to the
glutamine conjugate of HMHA (6).
Table 2. Cleavage of Novel Glutamine Conjugates by Axilla Bacteria. Stationary-phase cultures (OD600
1) were amended with 2 mm of the substrate, and analyzed after 7 or 24 h of incubation (for details, see
Exper. Part).
Strain

Species assignment

Velocity of cleavage [mmol l

a

Ax1
Ax 6
Ax 9
Ax 3
Ax 7
Ax 15
Ax 19
Ax 20
Ax 21
a

Staphylococcus capitis
S. epidermidis
Micrococcus luteus
Corynebacterium bovis
Corynebacterium (group G )
C. jeikeium
Corynebacterium sp.
C. striatum
C. bovis

h 1]

6 )

10 )

14 (from 29)

11/13 (from 28)

0
0
0
0
0
48.6
93.9
153.6
1

4.8
6.1
0
18.0
5.2
27.9
50.7
166.9
473

0
0
0
13.6
0
146.3
205.2
233.7
43.0

0
0
0
14.7
0
74.9
138.1
208.0
32.3

) From the corresponding glutamine conjugate.

A similar observation was made when the substrates were incubated with the
recombinant enzyme: pure N-AGA efficiently cleaved these novel substrates, and the
Km value for 28 was found to be 0.115 mm, with a velocity Vmax of 0.83 mmol min 1 mg 1
protein, compared to the main unsaturated glutamine conjugate of 3M2 H (4/5), which
has a Km of 0.125 mm and a Vmax of 0.14 mmol min 1 mg 1 protein. The Km of 29 was
found to be 0.25 mm, with a Vmax of 0.51 mmol min 1 mg 1 protein, which is, thus,
similar to the main hydroxylated conjugate of HMHA (6; Km 0.3 mm, Vmax 0.57 mmol
min 1 mg 1). N-AGA had the highest affinity for both the conjugate of phenylacetic
acid (10) and the nonphysiological substrate Na-lauroyl-Gln, both giving rise to a Km
value of 0.05 mm (Fig. 2).
2.5. Identification of Two Novel Glutamine Conjugates in Axilla Secretions. The two
novel synthetic glutamine conjugates 28 and 29 were used as references to search for
these minor conjugates in axilla samples, where they had not been detected before.
Concentrated axilla extracts were fractionated size-selectively by gel filtration, and
then subjected to LC/MS analysis. Peaks with associated matching molecular weights
and identical retention times were detected in both axilla samples investigated, clearly
indicating the presence of 28 and 29 in these samples. In addition, treatment of the LC/
MS samples with N-AGA, and subsequent re-analysis of the same samples by the same
methodology showed that the specific peaks with the correct masses had disappeared.
The analytical data for 29 as an example are shown in Fig. 3. It cannot be excluded,
however, that the signal at m/z 288 1 may also be due to the potentially co-eluting

12

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

Fig. 2. Kinetics Analysis of the Enzymatic Cleavage of Glutamine Conjugates by N-AGA. The following
substrates were tested at varying concentrations: 28 (&); 29 (~); Gln conjugates of 6 (&), of 4/5 (~), of 10
(*), and of lauroic acid (*).

hypothetical glutamine conjugates for compounds 12, 17, and 20, which would have
identical molecular weights.
The above analytical approach was finally also used to search in the same samples
for even further glutamine conjugates, for which no synthetic reference compounds
were available. A peak associated with a molecular mass of 300, corresponding to the
glutamine conjugate of 15, was present in both samples, and disappeared upon N-AGA
treatment. A peak with mass 302, which would correspond to the conjugates of
compounds 18a, 19, 24 or 25, was equally present in both samples, und vanished after
enzymatic treatment. Finally, a small peak giving rise to a molecular mass of 254,
corresponding to the conjugates of the amino acid degradation products 2 and 3, was
identified, and also found to be sensitive to N-AGA treatment. This result strongly
suggests that all these compounds exist as their glutamine conjugates. Besides these
novel compounds, the investigated samples contained large quantities of the glutamine
conjugates of HMHA (6) and 3 M2 H (4/5), as reported before [13].
3. Discussion. It has long been known that human axilla secretions are odorless,
and that the odoriferous principles are released from non-smelling molecules upon the
action of Corynebacteria [15 17]. Indeed, the initially odorless samples obtained in
this study from donors during physical exercise were found to release a plethora of
different carboxylic acids upon incubation with a single bacterial enzyme. The
structures of a total of 26 acids could be assigned, but there remain further minor,
unidentified peaks in the GC/MS, so further analysis could reveal even more
odoriferous principles in axillary secretions. Nine acids, compounds 7, 12, 14, 18a, 19,

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

13

Fig. 3. Identification of the Novel Glutamine Conjugate 29 in Fresh Axilla Secretions by LC/MS. The LC/
MS traces for the ion at m/z 288 1 are shown for a) the synthetic reference 29, b) the fraction eluting at
17.5 min from gel filtration of concentrated axilla secretions, and c) the same sample after treatment with
N-AGA.

21, 11, 13, and 16, had never been reported from nature before, and they could be
unique to human physiology. The 3-hydroxy-3-methylheptanoic and -octanoic acids are
homologues of the previously reported key component HMHA (6). Many of the other
newly identified acids have, as a unifying principle, a Me branch in position 4, and are
either unsaturated or hydroxylated at position 3. Whereas the two straight-chain, evennumbered 3-hydroxy-acids 17 and 20 are known intermediates of common b-oxidation
of longer-chain fatty acids [20], the biosynthetic origin of the Me-branched acids
remains unknown. Interestingly, the 4-Et-branched acids had been known to be
contained in axilla secretions before [12]. 4-Ethyloctanoic acid (15), nicknamed goat
acid due to its pungent goat-like odor, had been originally isolated from secretions of
the sebaceous glands of male goats [21]. The fact that the 4-Et-branched acids are fully
saturated on the one hand, and the abundance of unsaturated and hydroxylated acids
with a Me branch at the same position on the other hand, led to the initial hypothesis
that the two unknown acids with molecular masses of 174 and 170 could be the
analogues 4-ethyl-3-hydroxyheptanoic and 4-ethyloct-3-enoic acid, respectively. These
reference compounds were, thus, synthesized, but the retention times (tR ) and MS
reference fractionation patterns of the synthetic compounds did not match with any
peak in the sweat samples.

14

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

Similarly to the Et-branched acids, the novel Me-branched ones have a very
pungent odor, and are perceived by the human nose at very low concentrations, as
observed by GC odor-threshold measurement and by olfactometry. In addition, the
strong odors of the individual peaks were already detected by direct evaluation at a
sniff port during GC analysis of the enzyme-treated axilla-secretion samples. In the
current study, no attempts were made to determine the chirality of the novel hydroxy
acids. For HMHA (6), a (S)/(R) ratio of 75 : 25 had been reported, with the (S)-isomer
having a more-pungent odor [14], and we could confirm this observation. For the novel
compounds 7, 14, 19 and 21, four stereoisomers are possible, and it will be interesting to
determine in future studies whether a single isomer is dominant in the axilla secretions,
and whether the dominant isomer is also perceived most strongly, as is the case for
HMHA (6) [14].
Interestingly, the samples obtained from the different donors exhibited distinct
odors, despite the fact that the two main compounds are the same in all individuals.
After our detailed analysis, this initial observation can be rationalized by the
observation that all the minor components are present in very different relative
proportions in the different samples. This reminds of the hypothesis for body-odor
determination in mice put forward by Singer et al. [7], who concluded that types of
body odor in mice are determined by the specific relative proportions of different
carboxylic acids generating a compound odor, rather than by the simple presence/
absence of individual chemical compounds.
The acids reported in this work are all released from axilla-secretion samples by the
same recombinant N-AGA enzyme, indicating their secretion as glutamine conjugates.
This conclusion, indirect in its nature, is further corroborated by the following
observations: 1) two predicted glutamine conjugates were, indeed, present in the
odorless secretions, as shown by LC/MS analysis in comparison to synthetic reference
compounds; 2) the LC/MS signals of these conjugates disappeared upon treatment of
the LC/MS samples with N-AGA; and 3) the N-AGA-specific inhibitor 4-carbamoyl-2[(3,7-dimethyloctyl)(hydroxy)phosphinoylmethyl]butyric acid strongly reduced the
formation of the novel acids, indicating that N-AGA, and not a potential contaminating
enzyme, releases the different structural principles.
Finally, the two novel glutamine conjugates 28 and 29, and the corresponding
conjugate of 10, were efficiently cleaved by isolates of Corynebacteria and the
recombinant N-AGA, further demonstrating the broad substrate specificity of this
enzyme for the acyl part of the substrate (but with a very high specificity for the Gln
residue), as reported before [13]. Interestingly, by measuring Michaelis Menten
kinetics for the different substrates, it was found that Vmax for 28 is much higher
compared to Vmax of the structurally related glutamine conjugate of 3 M2 H (4/5). The
same observation made on the pure enzyme is also true for the turnover of the novel
substrates by isolated axilla bacteria: Corynebacteria isolated from axilla did cleave
these novel substrates faster than the main component, the glutamine conjugate of
HMHA (6). This indicates that, despite their lower availability, these substrates can
still make a significant contribution to the overall body odor.
The glutamine conjugates of phenylacetic acid (10) and different drugs are known
to be formed in the liver by the mitochondrial enzyme Acyl-CoA-l-glutamine Nacyltransferase [22], and a first hypothesis would be that the conjugates secreted in the

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

15

axilla are also products of liver metabolism. In this case, systemic presence in plasma
and secretion in sweat at different body regions or excretion in the urine would appear
likely. However, attempts to release HMHA (6) with N-AGA from BuOH extracts of
plasma and urine, or from sweat obtained from different body regions, failed, although
the method used did corroborate the presence of high amounts of the glutamine
conjugate of 10 as a liver metabolite in urine, as reported before [9]. These
observations point to a specific synthesis of these conjugates locally in the axillary
glands, or to a potential binding in the blood of the conjugates to longer peptides/
proteins from which they cannot be released by N-AGA, which needs the free COOH
group in Gln for activity [13]. This latter hypothesis would be especially interesting in
the context of the peptide-microflora hypothesis on the body odors determined based
on MHC, as put forward by Penn and Potts [23], stating that the most-likely hypothesis
is: MHC molecules influence odor by binding unique subsets of peptides, which are
carried to the preputial, coagulating, axillary region, or other microbe harbouring glands
where their metabolites are made volatile by the commensal microflora. The result that
the different acids that make up a compound odor are all linked to glutamine with a
peptide bond, which is specifically cleaved by commensal microorganisms, gives a first
hint in this direction. However, MHC binding would require that Gln is bound to a
longer peptide in the blood, which is cleaved off before/upon secretion in the axilla.
The use of a single enzyme for the release of odoriferous principles from axilla
secretions is admittedly a focus on a narrow window. Indeed, we had previously
reported different sulfur-containing odorants in axilla secretions, and had presented
first evidence that they are released by a bacterial b-lyase [18], and conjugates of
odorants with the dipeptide Cys-Gly had been reported [19]. Whether further enzymes
will be discovered in these Corynebacteria, which release yet other series of odoriferous
principles, remains an open question. The recent publication of the first genome
sequence of an axilla isolate of Corynebacteria will certainly accelerate such studies
[24].
In conclusion, the detailed chemical information and the enzymatic methods for
analysis of axilla secretions presented in this and in our previous work [18] will facilitate
further studies. These are needed to bridge the gap between studies on the biochemistry
of the human body odors with the fascinating studies linking behavioral data
(individual liking of body odors) to genotyping of the MHC locus. The potential
quantitative correlation of the novel chemical principles with different MHC genotypes
is the next major challenge in this direction.
We thank H. Gfeller for conducting LC/MS analyses, K. Grman and H. Koch for measuring sensorial
data, donors of axilla secretions for their efforts, and Dr. B. Schilling, Dr. G. Frater, and Dr. M. Gautschi
for critical discussions.

Experimental Part
1. General. All reagents and solvents were purchased from Fluka (puriss. or purum), used without
further purification. IR Spectra: Perkin-Elmer 681 and Nicolet-510 FT-IR spectrometers, in cm 1. Optical
rotation: Perkin-Elmer 241 polarimeter. 1H- and 13C-NMR Spectra: Bruker AM-400 spectrometer, d
in ppm rel. to Me4Si or TPS, J in Hz. MS Spectra: Finnigan MAT95 (for extract samples) and Hewlett
Packard 5973 MSD (for synthetic samples), with LC/MS system Finnigan LCQ (APCI).

16

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

2. Collection of Axilla Secretions and Treatment with the Recombinant N-AGA. Axilla secretions of
individual donors were sampled on cotton pads fixed in the axillary region during physical exercise. Pads
were immediately frozen after collection. Pads from both axillae were then pooled, and extracted with
EtOH/H2O 1 : 1 (50 ml). The extracts were lyophilized, and resuspended in dist. H2O (2 ml). They were
extracted once with t-BuOMe (MTBE) and once with hexane to remove interfering lipids, and then split
into two 1-ml samples. The enzyme N-AGA was produced and purified as described before [13]. Briefly,
the open reading frame (Gene Bank accession number AF 534871) was expressed from the vector pBAD/
gIIIA (Invitrogen) in E. coli TOP10; arabinose induced cultures were lyzed in Buffer A ( 50 mm NaCl,
50 mm NaH2PO4/K2HPO4 ; pH 7), and the lyzate was purified on a phenyl-sepharose hydrophobicinteraction-resin column (Pharmacia; linear gradient from 1 to 0m of (NH4 )2SO4 in Buffer A) and on a
Mono-Q strong anion-exchange column on an FPLC system (Pharmacia; gradient from 0 to 0.8m KCl in
Buffer A). The resulting enzyme was over 95% pure, as judged by SDS-Page. The enzyme (1.8 mg) was
added to one axilla sample, and after 30 min at 368, both the enzyme-treated and the control samples
were acidified with 1m aq. HCl (50 ml), extracted with MTBE (0.5 ml), and the org. phase was
concentrated to ca. 80 ml, before being analyzed both by GC (equipped with a sniff-port and an FID) and
GC/MS. In control experiments, samples were split in two parts: one aliquot was amended with the
enzyme N-AGA, the other was exposed both to the enzyme and the inhibitor 4-carbamoyl-2-[(3,7dimethyloctyl)(hydroxy)phosphinoylmethyl]butyric acid (final concentration 500 mm), which specifically blocks N-AGA [25], to verify that the release of acids is not due to an enzymatic impurity in the
preparation of N-AGA. The same experiments as described above were also performed on sweat samples
collected with cotton pads during physical exercise from other body regions: back, neck and chest, as well
as from the face.
3. GC/MS Analysis of Axilla Secretions. For GC/MS analysis of the org. axilla extracts, a combination
of a Hewlett-Packard 5890 II gas chromatograph and a Finnigan MAT95 mass spectrometer (low
resolution, EI (70 eV), ion-source temp. 2308) was applied. A VF-5ms column (Varian), with a length of
30 m (i.d. 0.25 mm; film thickness 0.25 mm) was used. An aliquot (1.0 ml) of the MTBE soln. was injected
in splitless injection mode (2108). The temp. of the column oven was initially set to 308 for 3 min, and
subsequently increased to 658 (at 358/min), and then to 2708 (at 48/min).
4. Detection of Potential Glutamine Conjugates in Blood Plasma and Urine. Plasma samples (25 ml
each of five different donors) or urine samples (20 ml from one donor) were first extracted once with
MTBE, and then once with hexane to remove interfering lipids. The extracts were acidified with HCl to
pH 2, saturated with NaCl, and then extracted with an equal volume of BuOH. The org. phase was
separated, evaporated, the residues dissolved in H2O (4 ml), and adjusted to pH 7. The samples were split
into equal parts. One part was treated with N-AGA, and extracted and analyzed for released acids as
described above for the sweat samples. In parallel, plasma samples were spiked with synthetic Na-3hydroxy-3-methylhexanoyl-glutamine (HMHA-Gln; 20 mm), extracted, digested, and analyzed for
released HMHA (6) and other acids in the same way to verify that BuOH extraction was suited to
efficiently detect potential glutamine conjugates.
5. Fractionation and LC/MS Analysis of Aqueous Axilla Secretions. A lyophilized, nonhydrolyzed
sample of human axilla secretions (24 ml) sampled during several sessions of physical exercise and
pooled from two donors, and a second sample (10 ml) from a third donor, were separated by gel filtration
(Superdex Peptide 10/300 GL; Pharmacia), with (NH4 )2CO3 (100 mm) as elution buffer. The fractions
between 17- and 20-ml elution volume were collected, concentrated, and subjected to LC/MS analysis
(Finnigan LCQ mass spectrometer, APCI mode; equipped with a Flux Rheos 2000 HPLC pump). HPLC
Separation was performed on a C18 reverse-phase column modified for proteins and peptides (VYDAC,
Hesperia, CA, USA). The mobile phase consisted of H2O (A) and MeOH (B) each containing 1%
AcOH (v/v). The solvent flow was 0.3 ml/min, and the following gradient was used: 0 1 min, 90% A; 1
4.5 min, 90% A 98% B; 4.5 9 min, 98% B; 9 10 min, 98% B 90% A; 10 12min, 90% A.
6. Determination of Odor-Threshold Values. The GC-sniff threshold values and the dynamicdilution-olfactometry threshold values were determined as described by Neuner-Jehle and Etzweiler [26].
The olfactometric method determines the minimal concentration (in ng/l) that can be perceived by an
average panelist, whereas the GC-sniff threshold is the minimal amount (in ng) leaving the GC column at

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

17

a GC sniff port that can be perceived by an average panelist. Five panelists were used for the GC study,
and 14 panelists were used for the olfactometry study.
7. Bacterial Strains, Culture Conditions, and Enzyme-Kinetics Assays. The isolation of a collection of
axilla bacteria has been described before [13]. The strains were grown for 24 h in Mueller Hinton broth
(Difco) amended with 0.05% Tween 80, harvested by centrifugation, and resuspended to an optical
density at 600 nm of 1.0 in Buffer A. Aliquots (500 ml) of this stationary culture were then amended with
a final concentration of 2 mm of various substrates. After 7 or 24 h of incubation (with shaking at
300 r.p.m. at a temp. of 368), the samples were acidified, and extracted with MTBE (250 ml), and the
amount of released acids was determined by GC. Enzymatic assays to determine enzyme kinetics with
the various substrates were done in Buffer A with the recombinant N-AGA at a final concentration of
4.5 ng/ml, and by means of a fluorescent test: the enzyme reaction was stopped by adding 0.5 volumes of
MeCN containing 2 mm fluorescamine (Fluka), thus derivatizing the free NH2 group of the released lglutamine. Fluorescence was then measured on a FlexStation apparatus (Molecular Devices), with an
excitation wavelength of 381 and an emission wavelength of 470 nm.
8. Synthesis of Reference Compounds. 8.1. 3-Hydroxy Acids. General Procedure (GP 1). Under an
atmosphere of N2 , a stirred soln. of (i-Pr)2NH (10.1 g, 0.1 mol) in THF (50 ml) was cooled to 08, and a
BuLi soln. (2.7m in heptane, 27 ml, 0.1 mol) was added dropwise within 10 min. The resulting soln. was
allowed to warm to r.t., and stirring was continued at this temp. for ca. 30 min. The obtained lithium
diisopropylamide (LDA) soln. was cooled to
108, and a soln. of AcOH (3.0 g, 0.05 mol) in THF
(15 ml) was added dropwise and under vigorous stirring within 15 min. Subsequently, the mixture was
slowly warmed to 358, and kept at this temp. for 20 min. The resulting light-beige and possibly gelatinous
suspension was cooled to 208, and a soln. of the required aldehyde (0.05 mol) was added within 10 min
at the same temp. The mixture was subsequently allowed to warm to r.t., and stirred for ca. 1 h. The
reaction was quenched with ice/water (250 ml). The neutral parts were extracted with MTBE (2 
100 ml). The aq. phase was acidified with 2m H2SO4 (70 ml), and extracted with MTBE (3  100 ml). The
combined org. phases were washed with H2O (2  50 ml) and brine (50 ml), dried (Na2SO4 ), and
concentrated in vacuo to afford the crude 3-hydroxy acids as reference compounds.
3-Hydroxy-4-methylhexanoic Acid (7). Obtained according to GP 1 from 2-methylbutanal (4.3 g).
Yield: 2.8 g (38%; 53 : 47 diastereoisomeric mixture). Colorless oil. IR (film): 3400m (br., OH), 3500
2500m (br., COOH), 2963s, 2934m, 2878m, 1706vs (CO), 1462w, 1405m, 1278m, 1180s, 1051m, 1012m.
1
H-NMR (400 MHz, CDCl3 ): 8.1 7.3 (br. s, OH, COOH); 4.02 3.88 (m, H C(3)); 2.55 2.41 (m,
CH2(2)); 1.61 1.42 (m, H C(4), 1 H of CH2(5)); 1.24 1.12 (m, 1 H of CH2(5)); 0.96 0.86 (m,
Me C(4), Me(6)). 13C-NMR (100 MHz, CDCl3 ; major diastereoisomer): 178.0 (s, CO); 71.2 (d, C(3));
39.7 (d, C(4)); 38.7 (t, C(2)); 25.3 (t, C(5)); 13.7 (q, Me C(4)); 11.6 (q, C(7)). 13C-NMR (100 MHz,
CDCl3 ; minor diastereoisomer): 178.1 (s, CO); 71.7 (d, C(3)); 39.7 (q, C(4)); 37.7 (t, C(2)); 24.9 (t, C(5));
14.3 (q, Me C(4)); 11.4 (q, C(7)). EI-MS: 128 (2, [M H2O] ), 110 (4, [M 2 H2O] ), 89 (100), 71
(77), 57 (29), 43 (46), 29 (28).
3-Hydroxy-4-methylheptanoic Acid (14). Obtained according to GP 1 from 2-methylpentanal
(5.0 g). Yield: 4.0 g (50%; 54 : 46 diastereoisomeric mixture). Yellow oil. IR (film): 3500 2500m (br.,
OH, COOH), 2959s, 2931m, 2874m, 1708vs (CO), 1459w, 1407m, 1288m, 1182s, 1035m. 1H-NMR
(400 MHz, CDCl3 ): 8.0 7.4 (br. s, OH, COOH); 4.01 3.89 (m, H C(3)); 2.56 2.41 (m, CH2(2)); 1.72
1.04 (several m, H C(4), CH2(5), CH2(6)); 0.95 0.82 (m, Me C(4), Me(7)). 13C-NMR (100 MHz,
CDCl3 ; major diastereoisomer): 177.6 (s, CO); 71.5 (d, C(3)); 38.6 (t, C(2)); 37.6 (d, C(4)); 34.8 (t, C(5));
20.2 (t, C(6)); 14.2 (q, Me C(4)); 14.1 (q, C(7)). 13C-NMR (100 MHz, CDCl3 ; minor diastereoisomer):
177.7 (s, CO); 72.0 (d, C(3)); 37.7 (q, C(4)); 37.6 (t, C(2)); 34.4 (t, C(5)); 20.1 (t, C(6)); 14.6 (q, Me C(4));
14.0 (q, C(7)). EI-MS: 161 (2, [M H] ), 124 (5, [M 2 H2O] ), 89 (100), 71 (57), 43 (56).
3-Hydroxyoctanoic Acid (17). Obtained according to GP 1 from hexanal (5.0 g). Yield: 3.8 g
(47.5%). Yellow oil. IR (film): 3470m (br., OH), 3300 2500m (br., COOH), 2959m, 2930s, 2860m,
1708vs (CO), 1410m, 1176m, 1126m, 1044m. 1H-NMR (400 MHz, CDCl3 ): 7.8 7.1 (br. s, OH, COOH);
4.08 4.01 (m, H C(3)); 2.54 (dd, J 16.4, 3.6, 1 H of CH2(2)); 2.45 (dd, J 16.4, 8.8, 1 H of CH2(2));
1.61 1.21 (several m, CH2(4), CH2(5), CH2(6)); 0.89 (t, J 6.8, Me(8)). 13C-NMR (100 MHz, CDCl3 ):
177.0 (s, CO); 68.2 (d, C(3)); 41.1 (t, C(2)); 36.3 (t, C(4)); 31.6 (t, C(5)); 25.1 (t, C(6)); 22.5 (t, C(7)); 13.9
(q, C(8)). EI-MS: 124 (3, [M 2 H2O] ), 89 (97), 83 (33), 71 (73), 56 (65), 43 (100).

18

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

3-Hydroxy-4-methyloctanoic Acid (19). Obtained according to GP 1 from 2-methylhexanal (5.7 g).

Yield: 4.1 g (47%; 56 : 44 diastereoisomeric mixture). Yellow oil. IR (film): 3500 2400m (br., COOH),
3380m (br., OH), 2957s, 2928s, 2873m, 1708vs (CO), 1461w, 1406m, 1286m, 1182s, 1040m. 1H-NMR
(400 MHz, CDCl3 ): 7.3 6.5 (br. s, OH, COOH); 4.02 3.86 (m, H C(3)); 2.61 2.42 (CH2(2)); 1.77
1.08 (several m, H C(4), CH2(5), CH2(6), CH2(7)); 0.96 0.83 (m, Me C(4), Me(8)). 13C-NMR
(100 MHz, CDCl3 ; major diastereoisomer): 178.1 (s, CO); 71.4 (d, C(3)); 38.7 (t, C(2)); 38.0 (d, C(4));
32.3 (t, C(5)); 29.4 (t, C(6)); 22.9 (t, C(7)); 14.8 (q, Me C(4)); 14.0 (q, C(8)). 13C-NMR (100 MHz,
CDCl3 ; minor diastereoisomer): 178.2 (s, CO); 71.9 (d, C(3)); 38.1 (d, C(4)); 37.7 (t, C(2)); 31.9 (t, C(5));
29.3 (t, C(6)); 22.9 (t, C(7)); 14.4 (q, Me C(4)); 14.2 (q, C(8)). EI-MS: 156 (1, [M H2O] ), 114 (7,
[M AcOH] ), 89 (100), 71 (44), 54 (23), 43 (68), 29 (14).
(E/Z)-4-Methyloct-3-enoic Acid (13/11). Under an atmosphere of N2 , a flask was charged with
hexan-2-one (25 g, 0.25 mol), (2-carboxyethyl)triphenylphosphonium chloride (100 g, 0.27 mol), THF
(250 ml), and 1-methylpyrrolidin-2-one (250 ml). To this mixture, cooled to 08, was added NaH (20 g,
0.5 mol) in small portions under stirring over 30 min, during which the temp. was allowed to rise to 258
(H2 evolution). Then, gentle heating to 30 358 was applied, until the slightly exothermic reaction
started, and no further heating was needed to keep the temp. above 308 during ca. 1 h. Stirring was then
continued for 5 h at r.t., after which the H2 evolution ceased and the brownish mixture became more fluid
again. This mixture was then poured into ice/water (1 kg), and extracted with MTBE/hexane 1 : 1 (3 
400 ml). The aq. phase was acidified to pH 2 with diluted H2SO4 (ca. 300 ml), and extracted with hexane
(2  300 ml). The combined org. phases were washed with hot H2O (2  200 ml) and dried (Na2SO4 )
overnight, during which the majority of the remaining triphenylphosphine oxide precipitated and was
filtered off. The solvent was removed under reduced pressure, and the product was dried at 0.1 mbar for
5 h to give a crude mixture of 11 and 13. Yield: 19.2 g (50%; (E)/(Z) 2 : 1 (GLC analysis)), with traces of
residual triphenylphosphine oxide. Pale yellow oil. IR (film): 3400 2400m (br., COOH), 2958m, 2929s,
2860m, 1705vs (CO), 1412m, 1379w, 1294m, 1218m, 1129w. 1H-NMR (400 MHz, CDCl3 ): 11.2 (br. s,
OH); 5.30 (tt, J 7.2, 1.4, H C(3)); 3.09 (dd, J 7.2, 0.8, CH2(2)); 2.02 (t, J 7.2, CH2(5)); 1.74 (s,
Me C(4) of (Z)-isomer); 1.63 (s, Me C(4) of (E)-isomer); 1.42 1.23 (m, CH2(6), CH2(7)); 0.91 (t, J
7.2, Me(8) of (Z)-isomer); 0.89 (t, J 7.2, Me(8) of (E)-isomer). 13C-NMR (100 MHz, CDCl3 ; (Z)isomer): 179.2 (s, CO); 140.3 (s, C(4)); 115.2 (s, C(3)); 33.3 (t, C(2)); 31.7 (t, C(5)); 29.9 (t, C(6)); 23.3 (q,
Me C(4)); 22.6 (t, C(7)); 13.9 (q, C(8)). 13C-NMR (100 MHz, CDCl3 ; (E)-isomer): 179.2 (s, CO); 140.1
(s, C(4)); 114.6 (s, C(3)); 39.2 (t, C(5)); 33.5 (t, C(2)); 29.9 (t, C(6)); 22.3 (t, C(7)); 16.2 (q, Me C(4));
13.9 (q, C(8)). GC/EI-MS ((E)-isomer): 156 (18, M ), 114 (34), 96 (61), 81 (53), 69 (94), 55 (100), 43
(44), 41 (81). GC/EI-MS ((Z)-isomer): 156 (20, M ), 114 (47), 96 (66), 81 (58), 69 (100), 55 (98), 43
(44), 41 (80).
8.2. Synthesis of Glutamine Conjugates. N2-[(3E)-4-Methyloct-3-enoyl]-l-glutamine (28). To the
isomeric mixture of 11 and 13 (1.68 g, 8.6 mmol) 3 ) in 1,4-dioxane (23 ml), N-hydroxysuccinimide (1.13 g,
9.8 mmol, 1.1 equiv.) was added. The soln. was cooled with an ice bath, and a soln. of dicyclohexylcarbodiimide (DCC; 1.87 g, 10.2 mmol, 1.2 equiv.) in 1,4-dioxane (10 ml) was added over 10 min. A
white suspension formed, which was warmed to r.t. under stirring, then left standing overnight without
agitation. The mixture was filtered, and the solvent was evaporated to leave 2.49 g (100%) of a crude,
which was dissolved in THF/H2O 1 : 1 (40 ml). To this soln. was added l-glutamine (1.52 g, 10.4 mmol, 1.1
equiv.), followed by Et3N (1.46 g, 14.4 mmol, 1.7 equiv.), and the mixture was stirred at r.t. during 18 h.
The THF was evaporated, the aq. layer was acidified with 30% aq. KHSO4 soln. (40 ml), and extracted
with AcOEt. The org. layer was washed with brine, and dried (MgSO4 ). The crude product was purified
by flash chromatography (FC) (SiO2 ), and the isolated product was washed repeatedly with warm
toluene to remove residual traces of triphenylphosphine oxide, and dried in high vacuum. Yield: 1.63 g
(71%; 1 : 1 (E)/(Z) mixture). Viscous oil. [a]22
D 2.1 (c 1.7, EtOH). IR (film): 3349 (br), 2931m,
2497w, 1715m, 1651vs, 1440m. 1H-NMR: (400 MHz, CD3OD; diastereoisomeric mixture): 5.35 5.28
(sym. m, H C(3')); 4.42 4.37 (sym. m, H C(2)); 3.00 (d, J 7.2, CH2(2')); 2.35 1.90 (series of m,
CH2(3), CH2(4), CH2(5')); 1.74, 1.66 (2s, Me C(4')); 1.45 1.25 (m, CH2(6'), CH2(7')); 0.91, 0.90 (2t, J
5.2, Me(8')). 13C-NMR (100 MHz, CD3OD; diastereoisomeric mixture, partly resolved): 177.7 (s, C(1));
3)

Containing traces of residual triphenylphosphine oxide.

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

19

174.9 (s, C(5)); 174.9 (s, C(1')); 141.0, 140.0 (s, C(4')); 118.1, 117.7 (d, C(3')); 53.3, 53.3 (d, C(2)); 40.3 (t,
C(5')); 36.1, 35.3 (t, C(2')); 32.7, 32.6 (t, C(4)); 31.2 (t, C(6')); 28.6, 28.5 (t, C(3)); 23.7, 23.6 (t, C(7')); 23.4,
16.3 (q, Me C(4')); 14.4, 14.3 (q, C(8')). ESI-MS (pos.): 285 ([M H] ).
N2-[(3S)-3-Hydroxy-4-methylheptanoyl]glutamine (29). The procedure described for 28 was
repeated with 14 (1.60 g, 10.0 mmol) in 1,4-dioxane (30 ml), with N-hydroxysuccinimide (1.0 g,
10.0 mmol) and DCC (2.10 g, 10.2 mmol) in 1,4-dioxane (12 ml). The soln. obtained after filtration
was directly diluted with H2O (40 ml), followed by addition of l-glutamine (1.70 g, 11.6 mmol) and Et3N
(1.62 g, 16 mmol). Reaction and workup were carried out as described above. The crude product (1.40 g,
49%) was purified by FC [C18; Merck LiChroPrep, 40 63 mm (3  15 cm column); H2O ! H2O/EtOH
1 : 1]. The EtOH was evaporated, and the aq. soln. was lyophilized to yield 390 mg (14%) of 29. Sticky,
very hygroscopic white solid (mixture of 4 diastereoisomers). [a]22
0.9 (c 1.1, EtOH). IR (film):
D
3352 (br.), 2959m, 2481w, 1717m, 1650vs, 1455m. 1H-NMR (400 MHz, CD3OD; diastereoisomeric
mixture): 4.43 (m, H C(2)); 3.95 3.80 (m, H C(3')); 2.42 1.85 (3m, CH2(3), CH2(4), CH2(2')); 1.65
1.05 (series of m, H C(4'), CH2(5'), CH2(6')); 0.93 0.90 (m, Me C(4'), Me(7')). 13C-NMR (100 MHz,
CD3OD; diasteroisomeric mixture, partly resolved): 177.8 (br. s, C(1)); 174.9 (br. s, C(5)); 174.8 (br. s,
C(1')); 73.7, 73.6, 72.9 (3d, C(3')), 53.2 (br. d, C(2)); 41.8, 40.7 (2t CH2(2')); 39.8, 39.7, 39.4, 39.3 (4d,
C(4')); 36.4, 35.6, 35.5 (3t, CH2(4)); 32.7 (br. t, CH2(5')), 28.7, 28.6 (2t, CH2(3)); 21.4 (br. t, CH2(6')); 15.4
(q, Me(7)); 14.7, 14.6, 14.5, 14.4 (4q, Me C(4')). ESI-MS (pos.): 577 ([2M H] ), 289 ([M H] ).

REFERENCES
[1] K. Yamazaki, E. A. Boyse, V. Mike, H. T. Thaler, B. J. Mathieson, J. Abbott, J. Boyse, Z. A. Zayas, L.
Thomas, J. Exp. Med. 1976, 144, 1324.
[2] T. B. H. Reusch, M. A. Haberil, P. B. Aeschlimann, M. Milinski, Nature 2001, 414, 300.
[3] M. Olsson, T. Madsen, J. Nordby, E. Wapstra, B. Ujvari, H. Wittsell, Proc. R. Soc. London, Ser. B
2003, 270 (Suppl. 2), S254.
[4] C. Wedekind, T. Seebeck, F. Bettens, A. J. Paepke, Proc. R. Soc. London, Ser. B 1995, 260, 245.
[5] C. Wedekind, S. Furi, Proc. R. Soc. London, Ser. B 1997, 264, 1471.
[6] M. Yamaguchi, K. Yamazaki, G. K. Beauchamp, J. Bard, L. Thomas, E. A. Boyse, Proc. Natl. Acad.
Sci. U.S.A. 1981, 78, 5817.
[7] A. G. Singer, G. K. Beauchamp, K. Yamazaki, Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 2210.
[8] K. Yamazaki, G. K. Beauchamp, A. Singer, J. Bard, E. A. Boyse, Proc. Natl. Acad. Sci. U.S.A. 1999,
96, 1522.
[9] M. O. James, R. L. Smith, R. T. Williams, M. Reidenberg, Proc. R. Soc. London, Ser. B 1972, 182, 25.
[10] R. Claus, W. Alsing, J. Endocrinol. 1976, 68, 483.
[11] B. W. L. Brooksbank, R. Brown, J. A. Gustafsson, Experientia 1974, 30, 864.
[12] X. N. Zeng, J. J. Leyden, H. J. Lawley, K. Sawano, I. Nohara, G. Preti, J. Chem. Ecol. 1991, 17, 1469.
[13] A. Natsch, H. Gfeller, P. Gygax, J. Schmid, G. Acuna, J. Biol. Chem. 2003, 278, 5718 .
[14] Y. Hasegawa, M. Yabuki, M. Matsukane, Chem. Biodiv. 2004, 1, 2042; M.Yabuki, Y. Hasegawa, M.
Matsukane, International Symposium on the Chemistry of Essential Oils, Terpenes and Aromatics,
Tokushima Bunri University, October 18 20, 2002, Tokushima, Japan, p. 124; Y. Hasegawa, M.
Furukawa, M. Matsukane, to Kao Corp., Japanese Appl. JP 10025265 (Chem. Abstr. 1998, 128,
127752).
[15] W. B. Shelley, H. J. Hurley, A. C. Nichols, Arch. Derm. Syphilol. 1953, 68, 430.
[16] J. J. Leyden, K. J. McGinley, E. Hoelzle, J. N. Labows, A. M. Kligman, J. Invest. Dermatol. 1981, 77,
413.
[17] N. Shehadeh, A. Kligman, J. Invest. Dermatol. 1963, 41, 1.
[18] A. Natsch, J. Schmid, F. Flachsmann, Chem. Biodiv. 2004, 1, 1058.
[19] C. Starkenmann, Y. Niclass, M. Troccaz, A. J. Clark, Chem. Biodiv. 2005, 2, 705.
[20] C. G. Costa, L. Dorland, U. Holwerda, I. Tavares de Almeida, B. Poll-The, C. Jakobs, M. Duran, Clin.
Chem. 1998, 44, 463.
[21] T. Sugiyama, H. Sasada, J. Masaki, K. Yamashita, J. Agric. Biol. Chem. 1981, 45, 2655.

20

CHEMISTRY & BIODIVERSITY Vol. 3 (2006)

[22] L. T. Webster, U. A. Siddiqui, S. V. Lucas, J. M. Strong, J. J. Mieyal, J. Biol. Chem. 1976, 251, 3352.
[23] D. Penn, W. Potts, Adv. Immunol. 1998, 69, 411.
[24] A. Tauch, O. Kaiser, T. Hain, A. Goesmann, B. Weisshaar, A. Albersmeier, T. Bekel, N. Bischoff, I.
Brune, T. Chakraborty, J. Kalinowski, F. Meyer, O. Rupp, S. Schneiker, P. Viehoever, A. Phler, J.
Bacteriol. 2005, 187, 4671.
[25] F. Schroeder, M. Fournie-Zaluski, A. Natsch, to Givaudan Schweiz AG, PCT Int. Appl. WO
2004048394 A1 (Chem. Abstr. 2004, 141, 38737).
[26] N. Neuner-Jehle, F. Etzweiler, in Perfumes: Art, Science and Technology, Ed. P. M. Mller, D.
Lamparsky, Blackie Academic and Professional, London, 1991, p. 153.
Received October 24, 2005