The biosynthesis of branched dialkylpyrazines in myxobacteria

Article abstract of DOI:10.1002/cbdv.201000158

The biosynthesis of the volatiles 2,5- and 2,6-diisopropylpyrazine (2 and 3, resp.) released by the myxobacteria Nannocystis exedens subsp. cinnabarina (Na c29) and Chondromyces crocatus (strains Cm c2 and Cm c5) was studied. Isotopically labeled precursors and proposed pathway intermediates were fed to agar plate cultures of the myxobacteria. Subsequently, the volatiles were collected by use of a closed loop stripping apparatus (CLSA), and incorporation into the pyrazines was followed by GC/MS analysis. [²H ₈]Valine was smoothly incorporated into both pyrazines clearly establishing their origin from the amino acid pool. The cyclic dipeptide valine anhydride (16) - a potential intermediate on the biosynthetic pathway to branched dialkylpyrazines - was synthesized containing ²H₁ labels in specific positions. Feeding of [²H₁₆]-16 and [²H₁₂]-16 in both valine subunits mainly resulted in the formation of pyrazines derived from only one labeled amino acid, whereas only traces of the expected pyrazines with two labeled subunits were found. To investigate the origin of nitrogen in the pyrazines, a feeding experiment with [¹⁵N]valine was performed, resulting in the incorporation of the ¹⁵N label. The results contradict a biosynthetic pathway via cyclic dipeptides, but rather point to a pathway on which valine is reduced to valine aldehyde. Its dimerization to 2,5-diisopropyldihydropyrazine 36 and subsequent oxidation results in 2. The proposed biosynthetic pathway neatly fits the results of earlier labeling studies and also explains the formation of the regioisomer 2,6-diisopropylpyrazine 3 by isomerization during the first condensation step of two molecules valine aldehyde. A general biosynthetic pathway to different classes of pyrazines is presented.

Full text of DOI:10.1002/cbdv.201000158

CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
2129
The Biosynthesis of Branched Dialkylpyrazines in Myxobacteria
by Thorben Nawratha), Jeroen S. Dickschata), Brigitte Kunzeb), and Stefan Schulz*a)
a) Institut fꢀr Organische Chemie, Technische Universitꢁt Braunschweig, Hagenring 30,
D-38106 Braunschweig (phone: þ49-531-3917353; e-mail: Stefan.schulz@tu-bs.de)
b) Helmholtz-Zentrum fꢀr Infektionsforschung, Inhoffenstrasse 7, D-38124 Braunschweig
The biosynthesis of the volatiles 2,5- and 2,6-diisopropylpyrazine (2 and 3, resp.) released by the
myxobacteria Nannocystis exedens subsp. cinnabarina (Na c29) and Chondromyces crocatus (strains Cm
c2 and Cm c5) was studied. Isotopically labeled precursors and proposed pathway intermediates were fed
to agar plate cultures of the myxobacteria. Subsequently, the volatiles were collected by use of a closed
loop stripping apparatus (CLSA), and incorporation into the pyrazines was followed by GC/MS analysis.
[2H8]Valine was smoothly incorporated into both pyrazines clearly establishing their origin from the
amino acid pool. The cyclic dipeptide valine anhydride (16) – a potential intermediate on the biosynthetic
pathway to branched dialkylpyrazines – was synthesized containing 2H1 labels in specific positions.
Feeding of [2H16]-16 and [2H12]-16 in both valine subunits mainly resulted in the formation of pyrazines
derived from only one labeled amino acid, whereas only traces of the expected pyrazines with two labeled
subunits were found. To investigate the origin of nitrogen in the pyrazines, a feeding experiment with
[15N]valine was performed, resulting in the incorporation of the 15N label. The results contradict a
biosynthetic pathway via cyclic dipeptides, but rather point to a pathway on which valine is reduced to
valine aldehyde. Its dimerization to 2,5-diisopropyldihydropyrazine 36 and subsequent oxidation results
in 2. The proposed biosynthetic pathway neatly fits the results of earlier labeling studies and also explains
the formation of the regioisomer 2,6-diisopropylpyrazine 3 by isomerization during the first condensation
step of two molecules valine aldehyde. A general biosynthetic pathway to different classes of pyrazines is
presented.
Introduction. – Pyrazines are important odoriferous compounds that play a major
role in aroma and taste formation [1][2]. They are formed by the Maillard reaction
which takes place during heating of amino acids and carbohydrates [3][4]. Apart from
this chemical formation, pyrazines are also produced by many bacteria as volatiles or
during fermentation of food [1][2][5]. The naturally occurring pyrazines can be divided
into three different classes. The first class constituted by pyrazines with up to four alkyl
substituents, predominately Me or Et groups as in 2,3,5,6-tetramethylpyrazine (1),
whereas longer side chains are rare. They are frequently found in bacteria such as
Corynebacterium glutamicum that produces alkylated pyrazines in high amounts [5–7].
Proof for the involvement of bacteria in the production of pyrazines is sometimes
difficult, because especially the simple members of this class such as 2,5-dimethylpyr-
azine can be formed during autoclavation of culture media [8]. The second class of
pyrazines consists of compounds containing two branched side chains as represented by
2,5- and 2,6-diisopropylpyrazine (2 and 3, resp.) that are obviously derived from the
amino acids valine, leucine, or isoleucine [9]. Pyrazines with only one branched alkyl
group in the 2-position and, optionally, a Me or Et substituent in the 5- or 6-position,
ꢂ 2010 Verlag Helvetica Chimica Acta AG, Zꢀrich

2130
CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
e.g., 2-isopropylpyrazine (4), occur also [5]. Finally, methoxypyrazines constitute the
third pyrazine class that usually carry one or two alkyl substituents as found in 2-
methoxy-3-isopropylpyrazine (5) [5].
Only few efforts have been made so far to investigate the biosynthesis of pyrazines
in bacteria [2][9–15]. Recently, we have shown by a combination of gene knockout and
feeding experiments that the biosynthetic pathway to the first class of pyrazines in C.
glutamicum includes a sequence of transamination and oxidation of acetoin (¼ 3-
hydroxybutan-2-one; 6) and its higher homologues to a-amino ketones, followed by
condensation and oxidation [7] (Scheme 1).
Scheme 1. Biosynthesis of Tetramethylpyrazine (1) [7]
2-Alkyl-3-methoxypyrazines often have a low odor threshold and are characteristic
aroma components of different foods like peas or boiled potatoes [10][16]. These
compounds were initially proposed to be formed by the condensation of an amino acid
amide and glyoxal, followed by O-methylation [10]. Later, Cheng et al. suggested the
formation of 5 by a process involving condensation of valine (10) and glycine (11) to
form the piperazinedione 12 that, upon elimination of H2O, would yield a hydroxy-
pyrazine. Methylation during this sequence would lead to the methoxypyrazine 5
(Scheme 2). Feeding studies with labeled valine (10) and glycine (11) established the
incorporation of the former into 5, but glycine (11) was not incorporated [12], possibly
because it was metabolized before the pyrazine formation started [13]. To overcome
this problem, feeding experiments with differently 13C-labeled pyruvate isotopomers as
metabolic precursors of glycine and valine were carried out, supporting the proposed
bacterial pathway to the third class of pyrazines [13]. An identical mechanism was
proposed for the formation of 2-methoxy-3-(1-methylpropyl)pyrazine in Halomonas
venusta that showed increased formation of the respective piperazinedione intermedi-
ate when glycine and isoleucine were supplied [15].
Pyrazines of the second type were only recently found to be produced by bacteria
[1], and the biosynthetic pathway to these compounds is unknown. 2,5-Diisopropyl-
pyrazine (2) is the major pyrazine produced by Paenibacillus polymyxa. The addition of
valine (10) to the culture medium led to an increased production of 2 and an inhibition

CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
2131
Scheme 2. Proposed Biosynthetic Pathway to 3-Isopropyl-2-methoxypyrazine According to Cheng et al.
[13]
of the formation of all other pyrazines otherwise produced by this bacterium [9].
During the course of our program on the investigation of volatiles released by
myxobacteria, we identified several strains producing branched, dialkylated pyrazines.
Chondromyces crocatus (strains Cm c2 and Cm c5) and Nannocystis exedens (Na c29)
both release 2 among other pyrazines. Furthermore, small amounts of the regioisomeric
2,6-diisopropylpyrazine (3) are produced by C. crocatus [17–19].
We became interested in the biosynthesis of 2 and 3, because their amino acid
dependence pointed towards a different mechanism compared to the biosynthesis of
the highly alkylated pyrazines of the first class. Since these compounds, in contrast to
the amino acid-derived pyrazines from the third class, do not carry any O-containing
functional groups, a different pathway compared to the formation of methoxypyrazines
was expected. To shed more light on this biosynthetic pathway, differently labeled
amino acid derivatives were synthesized and fed to the myxobacteria N. exedens and C.
crocatus. These experiments resulted in the elucidation of a mechanism for the bacterial
formation of branched dialkylpyrazines via highly reactive amino aldehyde intermedi-
ates that is presented in this study.
Results and Discussion. – Our initial suggestion for the biosynthesis of 2 was based
on the proposal by Gallois et al. [12] and Cheng et al. [13] (Scheme 3), starting with self-
condensation of the amino acid valine (10) to the piperazinedione 16. Compared to the
proposed biosynthetic pathway to methoxypyrazines (Scheme 2), an additional
reduction step is needed to remove the excess O-atom. Reduction of the C¼O group
would furnish the hemiaminal 18 which, upon elimination of H2O, would give rise to the
dihydro-hydroxy-pyrazine 19. Rearrangement and H2O elimination would furnish the
target compound 2. The formation of its isomer 3 would require a rearrangement of the
C-backbone by an unknown mechanism.
To test this hypothesis, several feeding experiments with isotopically labeled
precursors and proposed intermediates of this pathway were carried out. These labeled
compounds were subjected to agar plate cultures of the bacteria, and their
incorporation into the pyrazines 2 and 3 was monitored using a closed-loop stripping
apparatus (CLSA) to collect the volatiles [17–19], followed by analysis of the

2132
CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
Scheme 3. Initially Proposed Biosynthetic Pathway to 2,5-Diisopropylpyrazine (2)
headspace extracts by GC/MS. The use of deuterium (2H1) labels was preferred because
of the shorter GC retention times of deuterated analogs compared to the parent
compounds [20]. This allowed the recording of almost clean mass spectra of deuterated
isotopomers that would not be possible with 13C or 15N labels. The results were used to
delineate the biosynthetic pathway.
Feeding Experiments with [2H8]Valine. Feeding of [2H8]valine to C. crocatus (Cm c5)
resulted in the one- and twofold incorporation into pyrazine 2. Predominantly, [2H7]-2
with one labeled i-Pr group was found (incorporation rate 22%), but traces of [2H14]-2
were also present (incorporation rate 2%). Similarly, [2H8]valine was incorporated into
3 with one or two labeled side chains, clearly establishing the formation of both
pyrazines from two valine units. Feeding of an isotopomer of the piperazinedione 16
with two labeled side chains could demonstrate its intermediacy in the proposed
pathway to 2 and could give additional insight into the pathway to 3. Therefore, the
extensively labeled piperazinedione 24 was synthesized as a mixture of both
diastereoisomers and fed to the bacteria.
The synthesis of 24 started with the alkylation of 2-[(diphenylmethylidene)ami-
no]acetonitrile (20) with 2-bromo[2H7]propane under basic conditions using NaOD in
D2O with simultaneous H/D exchange to yield 21 (Scheme 4). The diphenylmethyli-
dene protecting group was cleaved off under acidic conditions, and the nitrile 22 was
saponified to the carboxylic acid to yield rac-[2H8]valine (23) [21]. Its thermal
dehydration [22] that has recently been used for the formation of leucine anhydride
[23] resulted in a diastereoisomer mixture of 24, albeit in low yield only.
Feeding Experiments with [2H16]Valine Anhydride (24). Previous work with the
myxobacterium N. exedens (Na c29) showed that agar plate cultures of this bacterium
produced 2 as one of the major volatiles [18]. Accordingly, this strain was used to probe
whether 16 is an intermediate in the biosynthesis of 2. Feeding of 24 to an agar plate
culture as described in [18] clearly resulted in the incorporation of the isotopic labeling
into 2 (Figs. 1 and 2), but the outcome was unexpected. Interestingly, besides unlabeled
2, not only the expected [2H14]-2 was detected, but also [2H7]-2 was produced,
corresponding to only one deuterated i-Pr group. The unlabeled 2,5-diisopropylpyr-
azine (2) shows a molecular-ion peak at m/z 164 and a base peak at m/z 149 that arises
by loss of a Me group (Fig. 2). The corresponding ion peaks of [2H7]-2 are shifted to m/z

CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
2133
Scheme 4. Synthesis of [2H16]Valine Anhydride (24)
a) BnEt3NþClꢀ, toluene, NaOD, 2-bromo[2H7]propane; 54%. b) Et2O, 1m HCl. c) Conc. HCl, reflux;
83% (over two steps). d) SiO2, D, 10%.
Fig. 1. Total ion chromatogram of a headspace extract obtained after feeding of [2H16]valine anhydride
(24) to Nannocystis exedens (Na c29). The enlargement shows the ion traces of the molecular ion of
isotopomers of 2.
171, and 156 or 153, respectively, depending on the loss of a labeled or an unlabeled Me
group. The compound [2H14]-2 exhibits a molecular-ion peak at m/z 178 and a base peak
at m/z 160 representing the loss of a CD3 fragment from either of the labeled i-Pr
groups. These mass spectral data together with the slightly decreased retention times of
the deuterated compounds compared to those of unlabeled 2 unambiguously led to the
identification of the labeled pyrazines.
Compound 24 was also fed to C. crocatus (Cm c5) that is known for its ability to
produce a large variety of pyrazines, including the regioisomer of 2, 2,6-diisopropyl-
pyrazine (3) [17][19]. The results were similar to those obtained with N. exedens (Na
c29), showing unlabelled, 2,5-[2H7]-, and 2,5-[2H14]diisopropylpyrazine (2). Addition-
ally, weak incorporation of 24 into 3, a compound not produced by N. exedens, was
observed (Fig. 3), but, due to the low production rate of this volatile, only [2H7]-3 was

2134
CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
Fig. 2. Mass spectra of a headspace extract obtained after feeding of [2H16]valine anhydride (24) to
Nannocystis exedens (Na c29)
identified by the characteristically shifted ion peaks at m/z 153, 156, and 171, while
[2H14]-3 was missing.
The detection of [2H14]-2 is in accordance with the incorporation of the complete
precursor 24, whereas the formation of [2H7]-2 requires hydrolysis of the cyclic
[2H16]dipeptide 24 into two molecules of labeled valine. One labeled valine and another
unlabeled valine could then be used to rebuild [2H8]-16, meaning that the production of
[2H7]-2 and [2H7]-3 does not principally rule out the intermediacy of this piperazine-
dione in the biosynthesis of both pyrazines. However, the 2H1-incorporation rates for N.
exedens were 20% for [2H7]-2 and 1% for [2H14]-2, and, similarly, for C. crocatus 30%
for [2H7]-2 and 5% for [2H14]-2, raising doubts on the intermediacy of 16. If 16 would be
a true intermediate on the pathway to 2 and 3, and if a partial hydrolysis of [2H16]-16 to
[2H8]valine and back-formation of [2H8]-16 would give rise to [2H7]-2 and [2H7]-3, the
incorporation rates for these pyrazines, derived from one labeled valine, would

CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
2135
Fig. 3. Total ion chromatogram of a headspace extract of Chondromyces crocatus (Cm c5) obtained after
feeding of [2H16]valine anhydride (24). The enlargement shows the ion traces of the molecular ion of
isotopomers of 2 and 3.
expectedly be lower than the incorporation rates into the isotopomers derived from two
labeled valine units as present in the supplied 24. It seems more likely that the bacteria
completely cleave the cyclic dipeptide into two amino acid units, and then use the pool
of labeled and unlabeled valine to produce the observed isotopomeric mixture of 2 on
another pathway.
To differentiate between the two possibilities, labeled valine anhydride 34
containing differently labeled [2H4]valine and [2H8]valine units was synthesized and
used in feeding experiments. If 16 is a true intermediate, [2H10]-2 would be formed in
higher proportions relative to other isotopomers, while its hydrolysis into valine and
transformation via another pathway would lead to different isotopomers of 2 in low
concentration. Therefore, [2H12]valine anhydride 34 was synthesized, containing seven
2
and three H1-atoms in the i-Pr side chains.
Synthesis of [2H12]Valine Anhydride 34. Previously synthesized [2H8]valine (23) was
protected with a Boc group to give 25 as one building block of the target compound [24]
(Scheme 5). [2,4,4,4-2H4]Valine (31) was synthesized starting with a Grignard reaction
of [2H3]methylmagnesium iodide with acetaldehyde (26), followed by iodination of the
obtained [1,1,1-2H3]propan-2-ol (27) to 2-iodo[1,1,1-2H3]propane (28) [25] that was
used in the alkylation of 20. The resulting nirile 29 was transformed as described above
for the synthesis of 23 into rac-[2,4,4,4-2H4]valine (31) [21] that was subsequently

2136
CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
Scheme 5. Synthesis of [2H12]Valine Anhydride (34)
a) NaOD, D2O, Boc2O, dioxane; 12%. b) [2H3]MeMgI, Et2O. c) HI (57%), reflux. d) BnEt3NþClꢀ,
toluene, NaOD, 28; 37%. e) Et2O, 1m HCl. f) Conc. HCl, reflux; 6% (over two steps). g) SOCl2, MeOH;
31%. h) Boc-[2H9]Valine, ClCOO(i-Br), Et3N, CH2Cl2. h) 25, 1,2-Dichlorobenzene, reflux; 5% (over two
steps).
protected as methyl ester 32. Coupling of 25 and 32 formed dipeptide 33 [26][27] that
underwent deprotection and cyclization to 34 in refluxing 1,2-dichlorobenzene [27].
Feeding Experiments with [2H12]Valine Anhydride 34. Feeding of 34 to N. exedens
resulted only in the production of [2H3]-2 (molecular ion at m/z 167) originating from
the hydrolysis product [2H4]valine, and [2H7]-2 (m/z 171, from [2H8]valine) besides large
amounts of unlabelled 2 (Fig. 4). The isotopomer [2H14]-2 (m/z 178, from two units of
[2H8]valine) was detected in trace amounts only. The concentration was near the limit
of detection. The other expected isotopomers including [2H10]-2 that could directly be
formed from 34 as a hypothetical intermediate and [2H14]-2 could not be verified; their
concentration was most likely below the limit of detection. The relative ratios of 2,
[2H3]-2, [2H7]-2, [2H6]-2, [2H10]-2, and [2H14]-2 are roughly 90 :5 :5 :0.06 :0.13 :0.06,
when one assumes statistical use of valine from a pool of valine, [2H4]valine, and
[2H8]valine in a relative ratio of 95 :2.5 :2.5, showing the low probability of the
formation of 2 from two labeled valine molecules. The relative ratio of 2, [2H3]-2, and
[2H7]-2 actually found was ca. 95 :3 :2. The presence of only trace amounts of [2H14]-2
in the previous experiment together with the absence of [2H10]-2 in this experiment
indicate that the labeled isotopomers of 16 are indeed hydrolyzed to yield two labeled
valine molecules that serve as precursors for the biosynthesis of 2 in another pathway. A
direct transformation of 16 into 2 can be ruled out.

CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
2137
Fig. 4. Total ion chromatogram of a headspace extract of N. exedens (Na c29) obtained after feeding of
[2H12]valine anhydride (34). The enlargement shows the ion traces of the molecular ion of isotopomers
of 2.
The feeding experiment with 34 was repeated with C. crocatus (Cm c2) that also
produces both diisopropylpyrazines 2 and 3 [17], to see whether larger amounts of the
compounds were produced, compared to strain Cm c5 (Fig. 5). In this experiment, 2
Fig. 5. Total ion chromatogram of a headspace extract of C. crocatus (Cm c2) obtained after feeding of
[2H12]valine anhydride (34). The enlargements show the ion traces of the molecular ion of isotopomers
of 2 and 3.

2138
CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
and 3 with only one labeled i-Pr group were detected (molecular-ion peaks were
observed either at m/z 167 or 171, resp.), whereas all three possible isotopomers
derived from two labeled valine units were absent. These results similarly rule out the
intermediacy of 16 for this species.
Feeding Experiments with [15N]Valine. Upon addition of [15N]valine to cultures of C.
crocatus (Cm c2 and Cm c5), both strains showed an enhanced production of 2 and 3.
The one- and twofold incorporation of 15N-labeling was indicated by enhanced
intensities of the ions at m/z 165 and 166 compared to their natural abundance in the
unlabeled compounds (Fig. 6). The relative amounts of 2, [15N]-2, and [15N2]-2 were
55 :35 :10, calculated from the base peak at m/z 149, because the presence of an
abundant [Mꢀ1]þ ion in the mass spectrum of 2 made quantification by use of the
molecular ion ambiguous. A similar ratio, 52 :36 :12, was found for 3, [15N]-3, and
[15N2]-3 (Fig. 6). The ratio is roughly equal to the calculated ratio of isotopomers
(57:29 :14) when one fifth of the valine used by the bacteria was labeled with 15N.
The results clearly established the origin of the N-atoms in the pyrazines from valine
and excluded any participation of a transamination process during the biosynthesis of 2
and 3.
Fig. 6. Total ion chromatogram and mass spectra of 2 and 3 in a headspace extract of C. crocatus (Cm c5)
obtained after feeding of [15N]valine
The Biosynthetic Pathway to 2,5-Diisopropylpyrazine (2) and Its Regioisomer 3. The
results of the feeding experiments showed that the myxobacteria N. exedens and C.
crocatus metabolize cyclic valine dipeptide by hydrolysis and not by direct modifica-
tion. The piperazinedione serves merely as a source of the amino acid valine. No
evidence for an equilibrium between the cyclic dipeptide and the free amino acid was
found. Two units of valine are then combined to form the pyrazines 2 and 3, as shown by
the formation of [2H14]-2 and [15N2]-2 from [2H8]- and [15N]valine, respectively. Since an
excess of unlabeled valine originating from the medium or by de novo synthesis
compared to labeled valine was available to the bacteria in the feeding experiments, the

CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
2139
formation of unlabeled or mono-labeled pyrazines was favored over the doubly-labeled
compounds. These results rule out a piperazinedione intermediate in the biosynthesis
of dialkylated pyrazines of the second class as proposed in Scheme 2.
Instead, a pathway via an unstable a-amino aldehyde intermediate seems more
likely (Scheme 6), similar to the biosynthesis of alkylated pyrazines from acetoin and
related a-hydroxycarbonyl compounds [7]. Valine (10) is reduced to the highly reactive
a-amino-aldehyde intermediate 35 that can dimerize by two condensation reactions to
form the dihydropyrazine 36. Finally, 36 is oxidized enzymatically or nonenzymatically
to form the target pyrazine 2. The occurrence of dihydropyrazine intermediates in the
biosynthesis of pyrazines is supported by the recent identification of 3,6-dihydro-2,5-
dimethylpyrazine occurring besides 2,5-dimethylpyrazine in emissions of the fruit fly
Anastrepha serpentina [28]. Pathways requiring a valine amide as intermediate [10] or a
transamination step can be clearly ruled out because of the retention of the valine N-
atom.
Scheme 6. Proposed Biosynthetic Pathway to 2,5-Diisopropylpyrazines (2)
The presence of small amounts of 3 might be explained by a rearrangement during
the biosynthesis. a-Amino aldehydes are well-known to be autocatalytically trans-
formed into more stable amino ketones [29][30]. Since in the chemical reaction two a-
amino aldehydes react under exchange of their N-atoms, such a process seems unlikely
under physiological conditions. Instead, the isomerization would require an external
nitrogen source, possibly delivered by a transamination process. During this sequence,
the N-atom of the a-amino aldehyde would be lost and replaced by an N-atom from an
external source. Therefore, the 15N-content of the isomer [15N2]-3, obtained in the
feeding experiments with [15N2]valine, should be significantly lower than that of [15N2]-
2. This is not the case, indicating the absence of a 1-amino-3-methylbutan-2-one
intermediate in the biosynthesis of 3.
Instead, a rearrangement during the condensation of two a-amino-aldehyde units
35 as shown in Scheme 7 seems more plausible and is in accord with the observed 15N-
labeling pattern. The attack of the NH2 group of 35 on the C¼O group of another
molecule 35 leads to the tetraedric intermediate 37 that conventionally undergoes
elimination of H2O to form an imine. An identical reaction on the other carbonyl group
leads to compound 36, and eventually to 2. In the biosynthesis of 3, however, H2O is
eliminated, and the i-Pr group is rearranged to the former C¼O C-atom [31]. The
reaction commences with Hþ abstraction from the second N-atom, leading to the
primary imine 38. Tautomerization of the imine to the enamine furnishes the ene-
diamine 39 that condenses to the dihydropyrazine 40. The final oxidation furnishes the
rearranged pyrazine 3, without requirement for an external N-atom. Therefore, the 15N-
labeling patterns in compounds 2 and 3 remain identical.

2140
CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
Scheme 7. Proposed Biosynthetic Pathway to 2,6-Diisopropylpyrazines (3)
The proposed biosynthetic pathway may also be operative for other 2,5- and 2,6-
dialkylpyrazines, starting from the amino acids leucine (44), isoleucine (45), or others.
It explains also why no 2,3-dialkylpyrazines are formed, since the structure of the
precursor does not allow their formation. Interestingly, many of the educts of the
pyrazine biosynthesis such as acetoin (6) [7] or the highly reactive a-amino aldehyde 35
are toxic to bacteria. It seems reasonable that some bacteria use the formation of the
pyrazines as a detoxification mechanism to get rid of these compounds.
The biosynthetic pathway proposed here is similar to the pathway to alkylated
pyrazines of class one [7]. While the amino acids 10, 44, and 45 are used as starting
materials in the formation of branched dialkylpyrazines, the unbranched alkylated
pyrazines are formed via a-hydroxycarbonyl compounds such as acetoin (6),
glycolaldehyde (41), lactaldehyde (42), or probably glyoxal (43). The a-aminocarbonyl
compounds derived from these building blocks may be combined as shown in Scheme 8,
leading to a multitude of product pyrazines represented by the compounds 1, 2, 4, 5, and
46–49. a-Amino aldehydes are chemically unstable and may dimerize even without
enzymatic activity to form dihydropyrazines that either undergo disproportionation or
are oxidized, e.g., by air, to form pyrazines [11][32][33]. Nevertheless, it seems likely
that the dimerization process is enzyme-controlled because distinct pyrazine patterns
occur in different bacteria. Up to now, no enzyme responsible for the dimerization or
the following oxidation step is known. Finally, also the formation of methoxypyrazines
of type three can be explained by the described pathway. If one a-aminocarbonyl unit
in the condensation is replaced by an amino acid methyl ester, a dihydromethoxypyr-
azine would be formed that could be finally oxidized to the methoxypyrazine. The O-
Me group could also be added in later stages if the initial condensation is performed by
an amino acid instead of the methyl ester. The labeling studies by Gallois et al. [12] who
observed incorporation of 10, but not of glycine, and Cheng et al. [13] are in accord with
the biosynthetic pathway proposed in Scheme 8. The earlier proposal by Murray et al.
[10] suggesting the formation of methoxypyrazines by condensation of glyoxal with
valine amide is a variant of our proposal and can be differentiated by 15N-labeling
studies.

CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
2141
Scheme 8. General Biosynthetic Scheme for the Bacterial Formation of Pyrazines
In summary, our results strongly point to reactive a-aminocarbonyl species as key
intermediates in pyrazine biosynthesis. The results fit into a unifying pyrazine
biosynthetic pathway, explaining the formation of three important pyrazine classes
found in Nature.
We thank the Deutsche Forschungsgemeinschaft for financial support (DFG Schu 984-6). Funding by
the Deutsche Forschungsgemeinschaft within the Emmy Noether Program (to J. S. D. ) is gratefully
acknowledged.
Experimental Part
General. Chemicals were purchased from Sigma-Aldrich (Germany), Fluka (Switzerland), Acros
(Belgium), Merck (Germany), CDN (Canada) via Dr. Ehrenstorfer (Germany), or CS (Germany). All
solvents were purified by distillation and dried according to standard methods. TLC: 0.2-mm pre-coated
plastic sheets (Polygram Sil G/UV254; Macherey-Nagel). Column chromatography (CC): Merck silica gel
60 (SiO2; 70–200 mesh) and solvent mixtures as used for the determination of Rf values in TLC.
Compounds were detected using UV light or a ninhydrin soln., followed by heat-gun treatment. NMR
Spectra: Bruker DPX-200, AVII-300, DRX-400, or AVII-600 spectrometers; d in ppm rel. to Me4Si as
internal standard, J in Hz.
Media and Growth Conditions. The strains Nannocystis exedens Na c29, and Chondromyces crocatus
Cm c2 and Cm c5 were grown on VY/2 agar plates (5 g lꢀ1 baker yeast, 0.1 g lꢀ1 CaCl2 ·2 H2O, 0.5 mg lꢀ1
,
vitamin B12, 15 g lꢀ1 agar, pH 7.2). The feeding experiments were performed in plastic Petri dishes

2142
CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
containing 25 ml of VY/2 agar. Strikes of well-grown agar cultures were each placed centrally on fresh
agar plates, and the plates were incubated at 308 for 2 d (Cm c2) to 6 d (Cm c5 and Na c29) until
approximately one third to half of the plates were covered with cells. Then, the labeled precursors
(0.1 mm final concentration) dissolved in H2O and filter-sterilized were striked out from the end of the
grown cells on the surface area of the agar plates. The treated agar cultures were further incubated at 308
for 1–2 d (Cm c2 and Cm c5) and 7 d (Na c29), resp., and then analyzed.
Sampling of Volatiles. Volatile org. compounds emitted by liquid cultures of myxobacteria were
collected using the CLSA technique [17]. The volatiles were adsorbed on charcoal (Chromtech; Precision
Charcoal Filter, 5 mg) for 24 h and afterwards eluted with 30 ml of CH2Cl2. The solns. were immediately
analyzed by GC/MS, and stored at ꢀ308.
GC/EI-MS Analysis. GC/EI-MS Analyses were carried out on a HP-6890 GC system connected to a
HP-5973 mass-selective detector fitted with a BPX5 fused-SiO2 cap. column (25 m, 0.22 mm i.d., 0.25 mm
film; SGE), or on an Agilent 7890A GC system connected to an Agilent 5975C inert mass detector fitted
with a HP-5MS fused SiO2 cap. column (30 m, 0.25 mm i.d., 0.25 mm film; Agilent). Conditions for the
HP-6890/HP-5973 system were as follows: inlet pressure, 77.1 kPa; 23.3 ml He minꢀ1; injection volume,
1 ml; transfer line, 3008; electron energy, 70 eV. Conditions for the Agilent 7890A/Agilent 5975C system
were as follows: inlet pressure: 77.1 kPa, He 23.3 ml minꢀ1; injection volume: 2 ml; transfer line: 3008;
electron energy: 70 eV. The GC was programmed as follows: 5 min at 508, increasing at 58 minꢀ1 to 3208,
operated in either split or splitless mode (60-s valve time); He carrier gas at 1 ml minꢀ1 (HP-6890) or
1.2 ml minꢀ1 (Agilent 7890A). Retention indices (I) were determined from a homologous series of n-
alkanes (C8 –C35) [34]. Identification of compounds was performed by comparison of the mass spectra
with those in the Wiley-6 Library, and by comparison with those of synthetic standards and of retention
indices from the literature [17][18].
Derivatization with 2,2,2-Trifluoro-N-methyl-N-(trimethylsilyl)acetamide (MSTFA). Silylations
were performed by adding MSTFA to the corresponding compound without the addition of any solvent
as described in [35]. After 60 min at 608, the silylated compounds were concentrated under a stream of
N2, dissolved in CH2Cl2, and then analyzed by GC/MS.
Preparation of 2-[(Diphenylmethylidene)amino]-3-methylbutanenitriles. The following compounds
were prepared according to Baldwin et al. [21] either using commercially available 2-bromo[2H7]propane
or 2-iodo[1,1,1-2H3]propane (28). 2-Iodo[1,1,1-2H3]propane (28) was synthesized as described by Chay-
kovski et al. [25] using [2H3]MeI as starting compound. The reaction progress was monitored by GC/MS.
The crude product was used for the synthesis of 29, because isolation of 28 proved difficult due to its high
volatility.
2-Iodo[1,1,1-2H3]propane (28). EI-MS (70 eV): 173 (45, Mþ ), 158 (1), 155 (1), 127 (62), 46 (100).
2-[(Diphenylmethylidene)amino]-3-([2H3]methyl)[2,3,4,4,4-2H5]butyronitrile (21). Yield: 54%
(591 mg, 2.20 mmol). Rf (pentane/Et2O 10 :1) 0.23. GC: I 2047. 1H-NMR (400 MHz, CDCl3): 7.13–
7.82 (m, 10 H). 13C-NMR (100 MHz, CDCl3): 17.6 (m, 2 [2H3]Me); 32.5 (m, C[2H1]); 59.0 (m, C[2H1]);
118.9 (C); 127.4 (2 CH); 128.2 (2 CH); 128.9 (2 CH); 129.2 (2 CH); 131.0 (2 CH); 135.4 (C); 138.6
(C); 172.8 (C). EI-MS: 270 (7, Mþ ), 252 (2), 220 (60), 219 (5), 180 (4), 165 (21), 117 (100), 77 (14), 51
(6).
2-[(Diphenylmethylidene)amino]-3-methyl[2,4,4,4-2H5]butanenitril (29). Yield: 37% (530 mg,
1.99 mmol). Rf (pentane/Et2O 10 :1) 0.20. GC: I 2051. 1H-NMR (400 MHz, CDCl3): 1.01 (d, J¼6.8,
3 H); 1.12 (d, J¼6.8, 3 H); 2.11 (q, J¼6.6, 2 H); 7.18–7.82 (m, 20 H). 13C-NMR (100 MHz, CDCl3): 18.4
(m, Me, [2H3]Me); 18.8 (m, Me, [2H3]Me); 33.3 (2 CH); 118.9 (2 C); 127.5 (4 CH); 128.2 (4 CH); 129.0
(4 CH); 129.2 (4 CH); 131.0 (4 CH); 135.5 (2 C); 138.7 (2 C); 172.9 (2 C). EI-MS: 266 (3, Mþ ), 251 (1),
220 (44), 180 (10), 165 (30), 117 (100), 77 (26), 51 (16), 46 (15).
Preparation of Valine Isotopomers. The different valine isotopomers were prepared according to the
procedure of Baldwin et al. [21] by stirring either nitrile 21 or 29 (1 equiv.) for 12 h at r.t. in a mixture of
Et2O (8 ml) and 1m HCl (16 ml). Then, the org. layer was removed, and the aq. layer was extracted twice
with Et2O. Conc. HCl (16 ml) was added to the aq. layer, and the resulting soln. was heated to reflux for
12 h. Concentration of the product under vacuum gave a crude product that was dissolved in H2O and
subjected to ion-exchange CC (DOWEX 1ꢁ8, 200–400 mesh, Clꢀ ). Concentration in vacuo furnished
the product.

CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
2143
[2,3,4,4,4,4’,4’,4’-2H8]Valine (23). Yield: 83% (600 mg, 4.80 mmol). GC: I 1182 (Bis(trimethylsilyl)
(TMS) adduct). 13C-NMR (100 MHz, D2O): 18.2 (m, [2H3]Me); 19.7 (m, [2H3]Me); 30.9 (m, C[2H1]);
62.7 (m, C[2H1]); 177.1 (C). EI-MS (Bis-TMS adduct): 254 (3, [Mꢀ15]þ ), 226 (11), 219 (21), 164 (2),
133 (2), 119 (2), 101 (6), 73 (58), 60 (4), 45 (7).
1
[2,4,4,4-2H4]Valine (31). Yield: 6% (17 mg, 0.14 mmol). GC: I 1186 (Bis-TMS adduct). H-NMR
(300 MHz, D2O): 1.06 (d, J¼6.8, 3 H); 1.08 (d, J¼6.8, 3 H); 2.24 (q, J¼6.8, 2 H). 13C-NMR (100 MHz,
D2O): 18.9 (m, Me, [2H3]Me), 21.0 (m, Me, [2H3]Me); 33.0 (CH), 56.3 (m, C[2H1]), 179.2 (C). EI-MS
(Bis-TMS adduct): 250 (3, [Mꢀ15]þ ), 223 (3), 222 (11), 160 (2), 148 (100), 132 (5), 101 (6), 73 (61), 59
(3), 45 (8).
Preparation of [2H16]Valine Anhydride (24). As described by Basiuk [22], [2H8]valine (85 mg,
0.7 mmol) was mixed in a sublimation apparatus with 400 mg of SiO2 which was previously dried at 2508
in a high vacuum. The mixture was then heated to 2308 under high vacuum conditions for 45 min, upon
which the product (16 mg, 0.07 mmol) sublimed. The product was obtained in a diastereoisomeric mix-
ture (ratio 2 :1) with a yield of 10%.
Data of 24. GC: I 1332/1358. GC: I 1332/1358. 1H-NMR (400 MHz, CDCl3): 3.89 (d, J¼1.7, 1 H, H/D
exchange); 3.94 (d, J¼1.2, 1 H, H/D exchange); 5.90–5.94 (m, 4 H). 13C-NMR (100 MHz, D2O): 19.0
(m, 8 [2H3]Me); 31.1 (m, 4 C[2H1]); 61.2 (m, CH, C[2H1]); 62.2 (m, CH, C[2H1]); 169.3 (2 C); 172.7 (2 C).
EI-MS (70 eV) (I 1332): 213 (2, [MꢀH]þ ), 194 (4), 165 (100), 142 (12), 135 (25), 115 (60), 107 (13), 88
(20), 80 (45), 73 (11), 61 (19), 46 (25)./(I 1358): 214 (1, Mþ ), 194 (3), 164 (100), 141 (19), 135 (21), 123
(15), 115 (49), 106 (11), 86 (20), 79 (68), 61 (31), 46 (31).
Preparation of N-[(tert-Butoxy)carbonyl][N,2,3,4,4,4,4’,4’,4’-2H9]valine (25). According to the
procedure of Bayardon and Sinou [24], the Boc protecting group was introduced using NaOD in D2O
instead of NaOH to avoid any H/D exchange during the reaction. Aq. workup gave 50 mg (0.22 mmol) of
the pure product, containing a D-atom also bonded to N-atom.
1
Data of 25. Yield: 12%. GC: I 1489. H-NMR (400 MHz, CDCl3): 1.45 (s, 9 H); 9.80 (br. s, 1 H).
13C-NMR (100 MHz, CDCl3): 16.5 (m, [2H3]Me); 17.9 (m, [2H3]Me); 28.2 (3 Me); 30.0 (m, C[2H1]); 57.9
(m, C[2H1]); 80.0 (C); 155.8 (C); 177.0 (C). EI-MS: 226 (1, Mþ ), 180 (17), 152 (3), 124 (54), 105 (5), 80
(54), 57 (100), (41).
Preparation of [2,4,4,4-2H4]Valin Methyl Ester (32). This compound was prepared by using the
procedure of Herbert and Knaggs [26] for the synthesis of amino acid methyl ester hydrochlorides. The
obtained methyl [2H4]valinate hydrochloride was then isolated using ion-exchange chromatography
(DOWEX 1ꢁ8, 200–400 mesh, Clꢀ ) giving 95 mg (0.70 mmol) of the product containing some side
product.
1
Data of 32. Yield: 31%. GC: I 971 (MSTFA). H-NMR (400 MHz, CD3OD): 1.02–1.09 (m, 6 H);
2.21 (q, J¼6.7, 2 H); 3.84 (s, 6 H). 13C-NMR (100 MHz, CD3OD): 17.5 (m, Me, [2H3]Me); 18.5 (m, Me,
[2H3]Me); 30.4 (CH); 30.6 (CH); 53.3 (Me); 53.4 (Me); 59.1 (m, 2 C[2H1]); 170.6 (C); 171.5 (C). EI-MS
(MSTFA): 207 (1, Mþ ), 192 (4), 161 (62), 148 (100), 132 (7), 101 (6), 89 (14), 73 (70), 59 (8), 45 (7).
Preparation of N-[(tert-Butoxy)carbonyl)[2H12]divaline Methyl Ester 33. As described by Chen et al.
[27], the two molecules 25 and 32 were coupled, and the obtained crude product was used for the next
reaction without further purification.
Data of 33. GC: I 1904/I 1939. EI-MS (I 1904): 342 (1, Mþ ), 281 (4), 255 (2), 239 (13), 207 (22), 192
(10), 180 (8), 166 (4), 124 (54), 80 (100), 57 (61), 41 (20). EI-MS (I 1939): 342 (1, Mþ ), 286 (2), 269 (6),
255 (2), 209 (9), 180 (20), 152 (2), 134 (9), 124 (86), 102 (5), 80 (100), 76 (20), 57 (60), 41 (9).
Preparation of [2H12]Valine Anhydride 34. According to the procedure of Chen et al. [27], this
compound was prepared by heating 33 for 18 h at 1808 in 1,2-dichlorobenzene. The mixture was then
cooled to 508, MTBE was added cautiously, and then further cooled to r.t. The solid formed during the
reaction was centrifuged off, washed several times with CH2Cl2, dissolved in MeOH, and treated with
charcoal, which was filtered off afterwards. Removal of the solvent under vacuum finally gave the pure
diastereoisomeric product (ratio 2 :1).
1
Data of 34. Yield: 5% (4 mg, 0.02 mmol; over two steps). GC: I 1752/I 1778. H-NMR (600 MHz,
CD3OD): 0.82–0.85 (m, 12 H); 0.94 (d, J¼7.2, 3 H); 0.94 (d, J¼7.2, 3 H); 0.96 (d, J¼7.2, 3 H); 0.96 (d, J¼
7.2, 3 H); 2.16 (q, J¼6.9, 4 H); 2.21 (q, J¼7.1, 4 H ) . 13C-NMR (150 MHz, CD3OD): 16.6 (2 Me); 17.6 (m,
24 [2H3]Me); 17.8 (2 Me); 18.6 (2 Me); 19.2 (2 Me); 30.9 (m, 4 C[2H1]); 33.0 (m, 4 CH, 4 C[2H1]); 33.4 (m,

2144
CHEMISTRY & BIODIVERSITY – Vol. 7 (2010)
4 CH); 60.9 (m, 16 C[2H1]); 170.3 (8 C); 170.6 (8 C). EI-MS (I 1752): 210 (1, Mþ ), 192 (2), 165 (100), 136
(17), 116 (85), 104 (8), 88 (29), 80 (36), 76 (19), 58 (20), 50 (11), 46 (20). EI-MS (I 1778): 210 (1, Mþ ), 192
(3), 165 (100), 139 (22), 115 (77), 107 (7), 87 (32), 80 (58), 76 (40), 58 (25), 50 (11), 46 (23).
REFERENCES
[1] E. W. Seitz, in ꢃBioprocess Production of Flavor, Fragrance, and Color Ingredientsꢄ, Ed. A.
Gebelman, John Wiley & Sons, New York, 1994, p. 95.
[2] G. P. Rizzi, Food Rev. Int. 1988, 4, 375.
[3] H. Nursten, ꢃThe Maillard Reaction. Chemistry, biochemistry and implicationsꢄ, The Royal Society
of Chemistry, Cambridge, 2005.
[4] A. Adams, N. de Kimpe, Food Chem. 2009, 115, 1417.
[5] S. Schulz, J. S. Dickschat, Nat. Prod. Rep. 2007, 24, 814.
[6] A. L. Demain, M. Jackson, N. R. Trenner, J. Bacteriol. 1967, 94, 323.
[7] J. S. Dickschat, S. Wickel, C. J. Bolten, T. Nawrath, S. Schulz, C. Wittmann, Eur. J. Org. Chem. 2010,
2687.
[8] A. B. DeMilo, C.-J. Lee, D. S. Moreno, A. J. Martinez, J. Agric. Food Chem. 1996, 44, 607.
[9] H. C. Beck, A. M. Hansen, F. R. Lauritsen, FEMS Microbiol. Lett. 2003, 220, 67.
[10] K. E. Murray, J. Shipton, F. B. Whitfield, Chem. Ind. (London) 1970, 4, 897.
[11] G. P. Rizzi, J. Agric. Food Chem. 1988, 36, 349.
[12] A. Gallois, A. Kergomard, J. Adda, Food Chem. 1988, 28, 299.
[13] T.-B. Cheng, G. A. Reineccius, J. A. Bjorklund, E. Leete, J. Agric. Food. Chem. 1991, 39, 1009.
[14] G. P. Rizzi, in ꢃHeteroatomic Aroma Compoundsꢄ, Eds. G. A. Reineccius, T. A. Reineccius, ACS,
Washington, 2002, Vol. 826, p. 132.
[15] M. Bungert, T. Jahns, H. Becker, Flavour Fragrance J. 2001, 16, 329.
[16] M. A. Petersen, L. Poll, L. M. Larsen, Lebensm.-Wiss. Technol. 1999, 32, 32.
[17] S. Schulz, J. Fuhlendorff, H. Reichenbach, Tetrahedron 2004, 60, 3863.
[18] J. S. Dickschat, T. Nawrath, V. Thiel, B. Kunze, R. Mꢀller, S. Schulz, Angew. Chem., Int. Ed. 2007, 46,
8287.
[19] J. S. Dickschat, H. Reichenbach, I. Wagner-Dçbler, S. Schulz, Eur. J. Org. Chem. 2005, 4141.
[20] J. S. Dickschat, S. C. Wenzel, H. B. Bode, R. Mꢀller, S. Schulz, ChemBioChem 2004, 5, 778.
[21] J. E. Baldwin, R. M. Adlington, N. P. Crouch, C. J. Schofield, N. J. Turner, R. T. Aplin, Tetrahedron
1991, 47, 9881.
[22] V. A. Basiuk, Orig. Life Evol. Biospheres 1992, 22, 333.
[23] A. C. Durow, G. C. Long, S. J. OꢄConnell, C. L. Willis, Org. Lett. 2006, 8, 5401.
[24] J. Bayardon, D. Sinou, Tetrahedron: Asymmetry 2005, 16, 2965.
[25] M. M. Chaykovski, L. C. Bae, M.-C. Cheng, J. H. Murray, K. E. Tortolani, R. Zhang, K. Seshadri,
J. H. B. C. Findlay, S.-Y. Hsieh, A. P. Kalverda, S. W. Homans, J. M. Brown, J. Am. Chem. Soc. 2003,
125, 15767.
[26] R. B. Herbert, A. R. Knaggs, J. Chem. Soc., Perkin Trans. 1 1992, 109.
[27] J. Chen, S. P. Corbin, N. J. Holman, Org. Process Res. Dev. 2005, 9, 185.
[28] D. C. Robacker, M. Aluja, R. J. Bartelt, J. Patt, J. Chem. Ecol. 2009, 35, 601.
[29] L. Duhamel, P. Duhamel, A. Jarry, Tetrahedron Lett. 1970, 1053.
[30] L. E. Fisher, J. M. Muchowski, Org. Prep. Proc. Int. 1990, 22, 399.
[31] T. Ooi, A. Saito, K. Maruoka, J. Am. Chem. Soc. 2003, 125, 3220.
[32] H. Iida, K. Hayashida, M. Yamada, K. Takahashi, K. Yamada, Synth. Commun. 1973, 3, 225.
[33] M. Eckhardt, in ꢃScience of Synthesis Georgꢄ, Ed. J. Cossy, Thieme Verlag, Stuttgart, 2006, Vol. 26,
p. 492.
[34] H. van Den Dool, P. D. Kratz, J. Chromatogr. 1963, 11, 463.
[35] S. Schulz, C. Arsene, M. Tauber, J. N. McNeil, Phytochemistry 2000, 54, 325.
Received June 9, 2010