Pyrazine biosynthesis in corynebacterium glutamicum

Article abstract of DOI:10.1002/ejoc.201000155

The volatile compounds released by Corynebacterium glutamicum were collected by use of the CLSA technique (closed-loop stripping apparatus) and analysed by GC-MS. The headspace extracts contained several acyloins and pyrazines that were identified by their synthesis or comparison to commercial standards. Feeding experiments with [²H₇]acetoin resulted in the incorporation of labelling into trimethylpyrazine and tetramethylpyrazine. Several deletion mutants targeting genes of the primary metabolism, were constructed to elucidate the biosynthetic pathway to pyrazines in detail. A deletion mutant of the ketol-acid reductoisomerase was not able to convert the acetoin precursor (S)2-acetolactate into the pathway intermediate (R)-2,3-dihydroxy-3-methylbutanoate to the branched amino acids. This mutant requires valine, leucine, and isoleucine for growth and produces significantly higher amounts and more different compounds of the acyloin and pyrazine classes. Gene deletion of the acetolactate synthase (AS) resulted in a mutant that is not able to convert pyruvate into (5)-2-acetolactate. This mutant also requires branched amino acids and produces only very small amounts of pyrazines likely from valine via the valine biosynthetic pathway operating in reverse order. A ΔASΔKR double mutant was constructed that does not produce any pyrazines at all. These results open up a detailed biosynthetic model for the formation of alkylated pyrazines via acyloins.

Full text of DOI:10.1002/ejoc.201000155

FULL PAPER
DOI: 10.1002/ejoc.201000155
Pyrazine Biosynthesis in Corynebacterium glutamicum
Jeroen S. Dickschat,*[a] Susanne Wickel,[a] Christoph J. Bolten,[b] Thorben Nawrath,[a]
Stefan Schulz,[a] and Christoph Wittmann[b]
Keywords: Natural products / Biosynthesis / Nitrogen heterocycles
The volatile compounds released by Corynebacterium glut-
amicum were collected by use of the CLSA technique
(closed-loop stripping apparatus) and analysed by GC-MS.
droxy-3-methylbutanoate to the branched amino acids. This
mutant requires valine, leucine, and isoleucine for growth
and produces significantly higher amounts and more dif-
The headspace extracts contained several acyloins and pyr- ferent compounds of the acyloin and pyrazine classes. Gene
azines that were identified by their synthesis or comparison
to commercial standards. Feeding experiments with
[2H7]acetoin resulted in the incorporation of labelling into tri-
methylpyrazine and tetramethylpyrazine. Several deletion
mutants targeting genes of the primary metabolism were
constructed to elucidate the biosynthetic pathway to pyr-
azines in detail. A deletion mutant of the ketol-acid reducto-
isomerase was not able to convert the acetoin precursor (S)-
2-acetolactate into the pathway intermediate (R)-2,3-dihy-
deletion of the acetolactate synthase (AS) resulted in a mu-
tant that is not able to convert pyruvate into (S)-2-aceto-
lactate. This mutant also requires branched amino acids and
produces only very small amounts of pyrazines likely from
valine via the valine biosynthetic pathway operating in re-
verse order. A ∆AS∆KR double mutant was constructed that
does not produce any pyrazines at all. These results open up
a detailed biosynthetic model for the formation of alkylated
pyrazines via acyloins.
phaga-Flexibacter-Bacteroides (CFB) phylum produces 2-
ethyl-3,6-dimethylpyrazine (2) and 2-ethyl-3,5-dimethylpyr-
azine (3).[5] Alkylpyrazines such as 7, 9, and 10 also occur
in insects, e.g. in the defensive spray of the phasmid insect
Phyllium westwodii.[6] They are furthermore known as ant
trail pheromones represented by 2 in Atta sexdens rubropi-
Introduction
Pyrazines occur frequently in Nature and are produced
by several bacteria.[1] The bacterium Paenibacillus polymyxa
releases several alkylated pyrazines with 2,5-diisopropyl-
pyrazine (1) as the major compound (Figure 1).[2] The
myxobacterium Chondromyces crocatus emits an exception-
ally complex pattern of volatile products composed of sev-
eral alkylated and methoxypyrazines. The major compo-
nents of headspace extracts are 1, 2-methoxy-3-(1-methyl-
ethyl)-6-(2-methylpropyl)pyrazine (15), and 2-methoxy-3-
(1-methylpropyl)-6-(2-methylpropyl)pyrazine (16) in strain-
specific proportions.[3] Several bacteria from the marine
Roseobacter clade also produce pyrazines. In the volatile
fraction from Sulfitobacter pontiacus a diverse array of pyr-
azines such as 3,6-dimethyl-2-(2-methylpropyl)pyrazine (7),
3,6-dimethyl-2-butylpyrazine (8), 3,6-dimethyl-2-(3-meth-
ylbutyl)pyrazine (9), 3,6-dimethyl-2-(2-methylbutyl)pyr-
azine (10), and the unusual compound 3,6-dimethyl-2-
(methylsulfanyl)pyrazine (17) have been detected.[3] The
compounds 8 and 9 are also released by cultures of Lok-
tanella sp.,[4] while a marine arctic bacterium from the Cyto-
[a] Institut für Organische Chemie, Technische Universität
Braunschweig,
Hagenring 30, 38106 Braunschweig, Germany
Fax: +49-531-3915272
E-mail: j.dickschat@tu-bs.de
[b] Institut für Bioverfahrenstechnik, Technische Universität
Braunschweig,
Gaußstraße 17, 38106 Braunschweig, Germany
Supporting information for this article is available on the
WWW under http://dx.doi.org/10.1002/ejoc.201000155.
Figure 1. Structures of known pyrazines and acetoin.
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 2687
Eur. J. Org. Chem. 2010, 2687–2695

J. S. Dickschat et al.
FULL PAPER
losa[7] or alarm pheromones such as 9 in Odontomachus has-
Early investigations by Kosuge et al. resulted in the iden-
tatus, or 3, 3,5-dimethyl-2-propylpyrazine (4), 2-butyl-3,5- tification of tetramethylpyrazine (11) and acetoin (18) in
dimethylpyrazine (5), and 3,5-dimethyl-2-pentylpyrazine (6) Bacillus natto.[13] Although no experimental proof was
in O. brunneus.[8] Methoxypyrazines such as 2-methoxy-3- given, the authors speculated about a biosynthesis of 11
(1-methylethyl)pyrazine
(12),
2-methoxy-3-(2-methyl- from 18. Demain et al. demonstrated a direct biosynthetic
propyl)pyrazine (13), and 2-methoxy-3-(1-methylpropyl)- link between pyrazines and the amino acids valine, leucine,
pyrazine (14) have been reported from several aposematic and isoleucine.[14] A genetically uncharacterised C. glut-
insects including Coccinella septempunctata.[9] The meth- amicum mutant was isolated that produced large amounts
oxypyrazines 12–14 are also found in some plants, e.g. in of 11 and was only able to grow with the supply of these
peas (Pisum sativum).[10]
amino acids. The biosynthesis of 11 was also shown to be
Knowledge about the biosynthesis of pyrazines is very thiamine-dependent, and based on these data a biosynthetic
limited, although this topic has long been questioned and pathway via 18 was suggested. Rizzi demonstrated the
controversally discussed in the literature. For the biosynthe- spontaneous formation of 11 from 18 in the presence of
sis of 2-methoxy-3-alkylpyrazines the condensation of an α- ammonium acetate or ammonium citrate at neutral to
amino acid amide such as valine amide with glyoxal fol- slightly acidic pH (pH 5.14), albeit at high non-physiologi-
lowed by O-methylation was suggested (Scheme 1).[10] In cal salt concentrations.[15] Inspite of these investigations a
contrast, Gallois et al. proposed valine and glycine as build- detailed understanding of the biosynthetic pathway to pyr-
ing blocks. However, feeding experiments with 13C-labelled azines was missing.
precursors demonstrated the incorporation of valine, but
Since the unravelling of the complete genome sequence
not of glycine.[11] Cheng et al. used [2-13C]pyruvate and [3- of C. glutamicum[16] a targeted gene deletion is possible.
13C]pyruvate resulting in the incorporation of the isotopic Here we describe the identification of volatiles released by
labelling in specific positions that is in accordance with a C. glutamicum. The headspace extracts contain several pyr-
biosynthetic pathway via valine, glycine, and S-adenosyl- azines and acyloins in different concentrations. Feeding ex-
methionine (SAM) for O-methylation.[12] The suggested periments with isotopically labelled precursors and gene
pathway proceeds via a cyclic dipeptide (dioxopiperazine), knockout experiments are described giving rise to a biosyn-
O-methylation, and elimination of water, and is more plaus- thetic pathway from glycolytic intermediates via acyloins to
ible than any other published route (Scheme 1).
a variety of alkylated pyrazines.
Results and Discussion
Volatile Products from Corynebacterium glutamicum
The volatiles released by agar plate cultures of the wild-
type strain of C. glutamicum have been collected in four
separate experiments by use of the CLSA (closed-loop
stripping apparatus) headspace technique and analysed by
GC-MS. The results of these analyses are presented in
Table 1 and the total ion chromatogram of one representa-
tive extract is shown in part A of Figure 2. Pyrazines
formed the predominant compound class represented by
methylpyrazine (27), 2,5-dimethylpyrazine (28), 2,3-dimeth-
ylpyrazine (29), large quantities of trimethylpyrazine (30),
2-ethyl-3,6-dimethylpyrazine (2), 2-ethyl-3,5-dimethylpyr-
azine (3), 2-propyl-3,5-dimethylpyrazine (4), tetramethyl-
pyrazine (11), ethyltrimethylpyrazine (31), 2-(hydroxy-
methyl)-5-methylpyrazine (34), and 2-(hydroxymethyl)-5,6-
dimethylpyrazine (35, Figure 3). The pyrazines were ac-
companied by acetoin (18) and its derivatives 3-hy-
droxypentan-2-one (19), 2-hydroxypentan-3-one (20), 2-hy-
droxyhexan-3-one (21), and 2,3-butanediol (23). The pyr-
azines 2, 11, 29, and 30 were consistently found in all four
replicates, whereas the other volatiles were only released by
one or a few cultures. The identity of all these compounds
was confirmed by comparison of the mass spectra (see Sup-
porting Information, Figures SI-1 and SI-2) and retention
indices to those of commercially available standards (11, 18,
23, and 27–30), or to synthetic references.
Scheme 1. Proposed biosynthetic pathways to 2-methoxy-3-alkyl-
pyrazines. A: Biosynthesis of 12 from valine and glyoxal as sug-
gested by Murray et al.[9] B: Biosynthetic pathway to 12 after
Cheng et al. from glycine, valine, and SAM.[12]
2688
www.eurjoc.org
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2010, 2687–2695

Pyrazine Biosynthesis in Corynebacterium glutamicum
Table 1. Compounds identified in the headspace extracts of C. glutamicum wild-type strain (WT) and mutant strains described in the text
(∆KR, ∆AS, ∆KR∆AS).
Compound[a]
I[b]
Ident.[c]
WT (1)
+
WT (2) WT (3) WT (4) ∆KR
∆AS ∆KR∆AS
Acetoin (18)
762
792
795
800
818
821
848
888
900
912
914
999
1002
1077
1082
1087
1137
1162
1169
ms, com
ms, com
ms, com
ms, syn
ms, syn
ms, com
ms, com
ms, com
ms, syn
ms, com
ms, com
ms, syn
ms, com
ms, syn
ms, syn
ms, com
ms, syn
ms, syn
ms, syn
ms, syn
ms, syn
ms, syn
+
++
+++
+
2,3-Butanediol (23)
++
2,3-Hexanedione (24)
+
+
+
3-Hydroxypentan-2-one (19)
2-Hydroxypentan-3-one (20)
Methylpyrazine (27)
+
+
+
+
+
+
Trimethyloxazol (26)
2,3-Heptanedione (25)
+
+
++
2-Hydroxyhexan-3-one (21)
2,5-Dimethylpyrazine (28)
2,3-Dimethylpyrazine (29)
2-Hydroxyheptan-3-one (22)
Trimethylpyrazine (30)
2-Ethyl-3,6-dimethylpyrazine (2)
2-Ethyl-3,5-dimethylpyrazine (3)
Tetramethylpyrazine (11)
2-(Hydroxymethyl)-5-methylpyrazine (34)
Ethyltrimethylpyrazine (31)
2-Propyl-3,5-dimethylpyrazine (4)
+++
++
++
+
+
++
++
++
++
+
+
+
+++
+
+++
++
+
+++
++
+++
+
+++
+
++
++
+++
+
++
+++
++
+
+
+
+
2-(Hydroxymethyl)-5,6-dimethylpyrazine (35) 1231
Trimethylpropylpyrazine (32)
Butyltrimethylpyrazine (33)
+
1243
1338
+
+
[a] Relative amounts of the volatile components in one extract are indicated by +: 0–2%, ++: 2–8%, +++: Ͼ8% of total area in the gas
chromatograms. Note that these indications for one specific sample are not comparable to the indications of another extract, i.e. the
indication of e.g. ++ for 21 in the ∆KR mutant represents a higher absolute amount than the indication +++ in the first wild-type extract
(compare Figure 1). [b] Retention indices I were determined from a homologous series of alkanes (C8–C36) on a HP-5 GC column. [c]
Compound identification was based on the mass spectrum (ms), comparison to a commercially available standard (com) or a synthetic
reference (syn, this work).
Compound 2 was synthesised from 28 (Scheme 2). Pyr- pyrazinecarboxylic acid (42) that was transformed into the
azines can be transformed into their N-oxides by using the respective methyl ester 44. The reduction of this ester to 34
peroxide reagent NaBO2·H2O2·H2O (simply named proved to be difficult. While LiAlH4 reduction failed, the
“NaBO3” in Scheme 2) in refluxing glacial acetic acid.[17] usage of NaBH4 gave 34 in poor yield. A known condensa-
This method was adopted to 28 to obtain 2,5-dimethylpyr- tion/oxidation sequence was carried out for the synthesis of
azine N-oxide (36) in good yield (85%) that was treated 35.[20] The reaction of diacetyl with 2,3-diaminopropionic
with POCl3 to result in 2-chloro-3,6-dimethylpyrazine acid proposedly resulted in a diimine intermediate that was
(37).[18] Fürstner and Leitner reported a method for the oxidised by the passage of an airstream through the reac-
iron-catalysed cross-coupling of aryl chlorides with Grig- tion mixture to give the heteroaromatic compound 43. Es-
nard reagents[19] that is also successful for the alkylation of terification to 45 and NaBH4 reduction furnished 35.
chloropyrazines.[3] The cross-coupling of 37 and EtMgBr
The acetoin derivatives 19–21 were synthesised as de-
yielded the desired pyrazine 2. The pyrazines 3 and 4 were picted in Scheme 3. Acetaldehyde was converted into 2-
equally synthesised from 2-chloro-3,5-dimethylpyrazine methyl-1,3-dithiane (46) that was used in a Seebach umpo-
(39) and EtMgBr or PrMgBr, respectively. The mass spectra lung with propanal to yield 2-(1-hydroxypropyl)-2-methyl-
and gas chromatographic retention indices of the re- 1,3-dithiane (50). Cleavage of the dithiane moiety with
gioisomeric alkylpyrazines 2 and 3 are very similar, indicat- AgNO3 in water, extraction of the reaction products with
ing the need for synthetic reference material of positional diethyl ether, and analysis of the dried organic layer by GC-
isomers. To rule out a confusion between 4 and its re- MS demonstrated the quantitative conversion of 50 to the
gioisomer 3,6-dimethyl-2-propylpyrazine (38), the latter hydroxy ketone 19. However, removal of the ether by distil-
compound was synthesised by the iron-catalysed alkylation lation resulted in a significant loss of the highly volatile
of 37 with PrMgBr. Comparison of both alkylpyrazines product and only small amounts of the pure product could
with the natural material clearly established the structure of be obtained. Similarly, the hydroxy ketones 20 and 21 were
4 for the volatile product released by C. glutamicum. Tri- synthesised from propanal and acetaldehyde, or butanal
methylpyrazine (30) served as starting material for the syn- and acetaldehyde, respectively.
thesis of 31. The N-oxidation yielded two regioisomeric N-
oxides 40a and 40b in a ratio of 2:3 not separable by column
chromatography. Therefore, the mixture was used in the
chlorination step to give chlorotrimethylpyrazine (41) that
was alkylated with EtMgBr resulting in 31. The (hy-
droxymethyl)pyrazine 34 was synthesised from 5-methyl- the results of our analyses showing the concomitant occur-
Pyrazine Biosynthesis
The observations made by Demain et al.[14] together with
Eur. J. Org. Chem. 2010, 2687–2695
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
2689

J. S. Dickschat et al.
FULL PAPER
Figure 2. Total ion chromatograms of Corynebacterium glutamicum headspace extracts obtained by CLSA. (A) C. glutamicum wild-type
strain, (B) mutant strain with a deletion of the ketol-acid reductoisomerase (∆KR), (C) mutant strain with a deletion of the acetolactate
synthase (∆AS), and (D) double mutant strain with a deletion of both the ketol-acid reductoisomerase and the acetolactate synthase
(∆KR∆AS). To allow for comparison of the wild-type and the mutant strains scaling of the ordinates is the same for all chromatograms
(A–D). Asterisks indicate artifacts also present in headspace extracts of agar plates containing the medium alone.
rence of acyloins and pyrazines in the headspace of C. glut- m/z = 142 (Figure SI-3 of Supporting Information). These
amicum suggest that the biosynthesis of several different data are in accordance with two deuterated methyl groups.
alkylated pyrazines proceeds via acyloin building blocks However, the incorporation proceeded with low rates (30:
(Scheme 4). To test this hypothesis, deuterated [2H7]acetoin 0.5%, 11: 1.0%; incorporation rates were determined from
was synthesised from [2H4]acetaldehyde in an umpolung re- the ratio of the peak areas of the molecular ions of the
action using 3-benzyl-5-(2-hydroxyethyl)-4-methyl-1,3-thi- deuterium labelled and the unlabelled isotopomers). The
azolium chloride and triethylamine. The labelled [2H7]-18 doubled incorporation rate observed for 11 compared to 30
was fed to the wild-type strain of C. glutamicum. Incorpora- reflects the statistically more likely uptake of one labelled
tion of the deuterium label was observed into the pyrazines acetoin molecule into 11 as it is formed from two acetoins,
30 and 11 indicated by a shift of the molecular ions of 30 whereas 30 requires only one acetoin unit (see discussion
from m/z = 122 to m/z = 128 and of 11 from m/z = 136 to below).
2690
www.eurjoc.org
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2010, 2687–2695

Pyrazine Biosynthesis in Corynebacterium glutamicum
According to textbook biochemistry[21] the biosynthesis
of amino acids valine and leucine needs two pyruvate units
(54a), one of which is transferred to the thiamine diphos-
phate (TPP, 55) cofactor under decarboxylation to give 2-
(1-hydroxyethyl)thiamine diphosphate (56a, Scheme 4).
This compound adds an acetyl group to the second pyr-
uvate unit to (S)-2-acetolactate (57a) in a reaction catalysed
by the acetolactate synthase (AS). The ketol-acid reducto-
isomerase (KR) catalyses the reaction of 57a to (R)-2,3-di-
hydroxy-3-methylbutanoate (58a) that is an intermediate to
valine and leucine. The AS and KR enzymes are also in-
volved in similar reactions in the biosynthesis of isoleucine.
The addition of an acetyl group from 56a to 2-oxobutano-
ate (54b) under catalysis of the AS results in (S)-2-aceto-
2-hydroxybutanoate (57b) that is converted by the KR to
(2R,3R)-2,3-dihydroxy-3-methylpentanoate (58b) en route
to isoleucine. Some organisms use an acetolactate decar-
boxylase to catalyse the formation of (R)-acetoin from (S)-
2-acetolactate, but this enzyme is not encoded in the ge-
nome of C. glutamicum, and therefore spontaneous decar-
boxylation of 57a to acetoin (18) or the presence of an un-
known type of acetolactate decarboxylase must be assumed.
Figure 3. Volatile products from Corynebacterium glutamicum.
Scheme 2. Synthesis of pyrazines from C. glutamicum.
Scheme 3. Synthesis of acyloins 19–22 from C. glutamicum.
Eur. J. Org. Chem. 2010, 2687–2695
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
2691

J. S. Dickschat et al.
FULL PAPER
Scheme 4. Proposed biosynthetic pathway to tetramethylpyrazine (11) via acetoin (18).
Equally 57b is converted to 19. The common biosynthesis 18 and a C3 unit such as hydroxyacetone/lactaldehyde or
of the intermediates 57a and 57b by the AS and of 58a and methylglyoxal that also coelute with the solvent. These C2
58b by the KR demonstrates some structural flexibility of and C3 building blocks can also serve for the formation of
the enzyme system with respect to their substrates. There- 27 (C2 + C3) and 28 (C3 + C3). The compounds 2 and 3
fore it seems reasonable that the action of the AS-KR sys- can be generated from the respective C3 unit and 19 or 20,
tem also accounts for the biosynthesis of 20 and 21. The 4 can arise from a C3 unit and 21, whereas 32 can be formed
reaction of 2-(1-hydroxypropyl)thiamine diphosphate (56b) from 18 and 19 or 20.
derived from 54b with 54a yields (S)-2-hydroxy-2-methyl-3-
The origin of the (hydroxymethyl)pyrazines 34 and 35
oxopentanoate (57c) that decarboxylates to 20. Equally, 2- found in one wild-type sample remains elusive. They may
oxopentanoate (54c) is transferred to the TPP cofactor re- either be generated by the incorporation of hydroxyacetone
sulting in 2-(1-hydroxybutyl)thiamine diphosphate (56c), or glycerinaldehyde as exemplarily shown for 35 in Scheme
which is together with 54a the precursor of (S)-2-hydroxy-2- SI-1 (see Supporting Information). Alternatively, these hy-
methyl-3-oxohexanoate (57d); decarboxylation of the latter droxylated pyrazines are formed from 28 and 30, respec-
results in 21.
tively, in a reaction catalysed, for instance by a mono-
The downstream steps from the acyloins to the pyrazines oxygenase. Another possibility is the spontaneous oxidation
are exemplarily shown for the formation of the D2h-sym- in the presence of oxygen, maybe via radical intermediates,
metrical compound 11 from two 18 molecules. The pathway although this mechanism seems unlikely, because the for-
to 11 might include the oxidation of 18 to butanedione (59) mation of several additional products could be expected.
by the acetoin dehydrogenase (AD) and transamination to The oxidation of any methyl group of 28 with molecular
3-aminobutanone (61) by an unknown aminotransferase. oxygen results in 34, whereas the oxidation of 30 should
Alternatively, a transamination reaction with 18 may first result in three different products represented by 35, 2-(hy-
proceed to 3-aminobutan-2-ol (60) that can be oxidised to droxymethyl)-3,6-dimethylpyrazine, and 2-(hydroxymethyl)-
61. Condensation of two units of 61 to tetramethyldihy- 3,5-dimethylpyrazine, but only 35 is found.
dropyrazine (62) and spontaneous oxidation in the presence
of oxygen can result in 11. As will be discussed below a
detailed analysis of our data suggest an enzymatic rather
than a non-enzymatic formation of the pyrazines from the
acyloin derivatives.
Volatiles from the ∆KR Mutant
In analogy to the biosynthesis of 11 the pyrazine 30 re-
To investigate the suggested pathway several deletion mu-
quires an acyloin building block in addition to a C2 build- tants were constructed and analysed for their production of
ing block such as glycolaldehyde or glyoxal (see Scheme SI- acyloin and pyrazines. A mutant with a gene deletion of the
1, Supporting Information). These metabolites are both not gene encoding the ketol-acid reductoisomerase (∆KR) was
found in the headspace extracts because they coelute with constructed. This mutant required valine, leucine, and iso-
the solvent in GC-MS. Compound 30 may be formed from leucine for growth and produced significantly higher
2692
www.eurjoc.org
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2010, 2687–2695

Pyrazine Biosynthesis in Corynebacterium glutamicum
amounts of acyloins and pyrazines compared to the wild-
Another volatile component found in traces in the head-
type strain (Table 1 and Figure 2, B), because acetolactate space extract of the ∆KR mutant was trimethyloxazol (26)
cannot be directed towards the amino acids valine and leu- that is the formal condensation product of 61 and acetic
cine. In consequence, a large acetolactate pool for the for- acid (Scheme 5), corroborating the proposed biosynthetic
mation of 18 and the pyrazines derived thereof is available. pathway to pyrazines via α-aminocarbonyl intermediates.
The derivatives 19–21 and 23 were also found, in addition Instead of acetic acid the more reactive acetyl-CoA may
to 2-hydroxyheptan-3-one (22) not present in the wild-type serve as the true reactant.
strain (mass spectra are depicted in Figure SI-2 of Support-
ing Information). For unambiguous identification com-
pound 22 was synthesised using the same approach as for
the shorter homologues from pentanal and acetaldehyde
(Scheme 3).
Similar to the proposed biosynthesis of the other acyloin
derivatives, 22 is presumably made from 2-oxohexanoate
(54d) by decarboxylation and addition to TPP (Scheme 4).
The reaction of 2-(1-hydroxypentyl)thiamine diphosphate
(56d) with pyruvate yields the (S)-2-hydroxy-2-methyl-3-
oxoheptanoate (57e) precursor of 22. Furthermore, the re-
lated compounds 2,3-hexanedione (24), and 2,3-heptane-
dione (25) were found. The presence of 24 and 25 supports
the biosynthetic pathway from the acyloin derivatives via
oxidation to the dicarbonyl compounds as depicted in
Scheme 4. The respective oxidation products of 18 (butane-
dione) as well as 19 and 20 both yielding 2,3-pentanedione
cannot be found in the headspace extracts, because these
compounds coelute with the solvent.
Scheme 5. Formation of trimethyloxazol (26) from 3-aminobut-
anone (61) and acetic acid.
Volatile Products from the ∆AS and ∆AS∆KR Mutants
A C. glutamicum mutant with a deletion of the acetolact-
ate synthase (∆AS) was constructed and analysed for the
production of acyloin and pyrazine derivatives (Table 1 and
Figure 2, C). This mutant required the addition of the
amino acids leucine, valine, and isoleucine to the growth
medium. The headspace extract contained only traces of
the pyrazines 11, 29, and 30 that are the most abundant
pyrazines found in the wild-type strain, whereas all other
pyrazines and the acyloin derivatives 18–22 produced by the
wild-type strain were absent in the ∆AS mutant. Since the
formation of acetolactate and 18 from pyruvate is impaired
in this mutant, the biosynthesis of these pyrazines may pro-
ceed via the valine and leucine biosynthetic pathway op-
erating in reverse order. It is remarkable that only the pyr-
azines derived from 18 (i.e., 11) and from 18 and a C2 or
C3 building block (29 and 30) were emitted, but no other
pyrazines originating from the acyloin derivatives 19–22
were found, pointing to the formation of all acyloins by the
acetolactate synthase in the wild-type strain.
The pyrazines produced by the ∆KR mutant included
the wild-type compounds 2, 11, 29, 30, and 31, ac-
companied by trimethyl-propylpyrazine (32) and butyl-tri-
methylpyrazine (33). The identity of these volatiles was es-
tablished by synthesis from the chloropyrazine 41 using
Fürstner’s and Leitner’s alkylation[19] with PrMgBr or
BuMgBr, respectively (Scheme 1). These pyrazines can be
generated from the acyloin derivatives produced by the
∆KR mutant. Compound 32 can arise from 18 and 21, and
33 can originate from 18 and 22 (Scheme 4). Further pyr-
azines such as 3,6-dimethyl-2,5-dipropylpyrazine that could
be formed by a dimerisation of 21 present in large quanti-
ties in the headspace extracts of the ∆KR mutant were not
found, pointing to an enzymatic formation of the pyrazines.
The enzyme seems to be able to catalyse the condensation
of the proposed aminocarbonyl building blocks derived
from 18 and 21, whereas the condensation of two such in-
termediates from 21 may exceed the limits of the enzyme.
This is in contrast to the suggested non-enzymatic forma-
tion of pyrazines such as 11 from acetoin and ammonium
salts at neutral to slightly acidic pH.[15] To exclude a non-
enzymatic mechanism for the formation of pyrazines in the
agar plate cultures of C. glutamicum the CG12½ medium
plates containing (NH4)2SO4 (1 gL–1) at different pH (pH
5.2, 6.5, and 7.2) were spiked with 18 (1 m). Analysis of
the volatiles released from these plates did not show the
production of the expected pyrazine 11 or any other pyr-
A C. glutamicum ∆AS∆KR double mutant was con-
structed and analysed for its profile of volatiles (Table 1 and
Figure 2, D). This mutant also required supply of the
branched amino acids and produced neither acyloin nor
pyrazine derivatives. Complementation of this mutant with
acetoin (1 m) resulted in the restored production of 11 and
30 (data not shown).
Occurrence of Identified Pyrazines in Nature
Alkylated pyrazines have frequently been described from
azine derivatives (data not shown). In case of a non-enzy- processed food. Their formation has been related to roast-
matic condensation of the acyloin derivatives and ammo- ing processes in which they are formed from amino acids
nium salts a complex pattern of several additional pyrazines and sugars in the Maillard reaction.[22,23] Several pyrazines
would be expected, formed from all possible combinations such as 28 can be formed in this reaction during the auto-
of the acyloins 18–22.
clavation process from medium ingredients,[1] but this was
Eur. J. Org. Chem. 2010, 2687–2695
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
2693

J. S. Dickschat et al.
FULL PAPER
ruled out for the pyrazines discussed here by analysis of the
Further investigations will be carried out in our laborato-
headspace compounds released from medium plates alone; ries to identify the genes and enzymes of the downstream
no pyrazines were found in the blank analysis. A hazelnut, steps of the biosynthetic pathway from the acyloins to the
cocoa, chocolate, rum, or roasted aroma has been attrib- pyrazine derivatives.
uted to several alkylated pyrazines such as 11 and 27–30
found in dark chocolate.[24] Several pyrazines described in
this article can also be found in insects, e.g. the mixture of
Experimental Section
2, 28, and 30 is the trail pheromone of ants of the Pogono-
Microorganisms: In the present work, the wild-type Corynebacte-
myrmex genus.[25,26] Among bacteria, Sulfitobacter ponti-
rium glutamicum ATCC 13032 (American Type and Culture Collec-
tion, Manassas, USA) and specifically designed mutants, con-
structed on basis of the wild-type, were investigated (Table SI-1 of
acus produces a blend of several pyrazines including 2, 11,
and 31.[3] Besides the compounds 2 and 3 already men-
tioned in the introduction a marine arctic bacterium from
the CFB phylum produces 11, 28, 30, and 31.[5] The hydrox-
ylated pyrazine 34 has been reported from the bat Tadarida
brasiliensis mexicana,[27] whereas its counterpart 35 and the
alkylpyrazines 32 and 33 are new natural products.
Supporting Information). Escherichia coli DH5α and NM522 were
obtained from Invitrogen (Karlsruhe, Germany) and used for plas-
mid amplification and DNA methylation, respectively.
Nucleic Acid Isolation: Cells were maintained at 30 °C on agar
plates, harvested after 2 d with a sterile inoculation loop and sus-
pended in 500 µL of sterile water. Cell disruption was performed
for 1 min at 30 Hz in a ribolyzer (MM301, Retsch, Haan, Ger-
many) after addition of glass beads (0.1–0.25 mm, Retsch, Haan,
Germany) and a mixture of phenol/chloroform/isoamyl alcohol
(700 µL, Carl-Roth GmbH, Karlsruhe, Germany). After separation
of aqueous and organic phase (10,000ϫg, 5 min), DNA from the
aqueous phase was precipitated by addition of sodium acetate
(65 µL, 3 , pH 5.5) and ethanol (100%, 1.3 mL) and a centrifuga-
tion step (10,000ϫg, 10 min). Subsequently, the supernatant was
removed. The precipitated genomic DNA was dissolved in sterile
water (100 µL). Isolation of plasmid DNA from E. coli DH5α and
NM522 was performed using the DNA isolation kits GFX Micro
Plasmid Prep (GE Healthcare, Piscataway, NJ, USA) and HiSpeed
Conclusions
The profile of volatiles released by Corynebacterium glu-
tamicum is composed of several pyrazine and acyloin deriv-
atives with trimethylpyrazine (30), tetramethylpyrazine (11),
and acetoin (18) as main compounds. Several additional
pyrazine and acyloin derivatives were identified from their
mass spectra and by comparison to commercially available
or synthetic references. Feeding experiments with [2H7]-18
clearly demonstrated that the pyrazines 11 and 30 are de-
rived from acetoin. Gene knockout mutants of relevant Plasmid Midi Prep (Qiagen, Hilden, Germany), respectively, ac-
cording to the instruction manual given by the manufacturer.
genes of the primary metabolism have been constructed to
further investigate the underlying pathways. Deletion of the
ketol-acid reductoisomerase (KR) of the biosynthetic path-
way to the branched amino acids resulted in a mutant that
required addition of valine, leucine, and isoleucine to the
medium and produced significantly higher amounts and
more different compounds of the pyrazine and acyloin
classes. Gene inactivation of the acetolactate synthase (AS)
gave a mutant that also required supply of the branched
amino acids and produced only minute amounts of the pyr-
azines 11, 29, and 30 explainable by their formation from
valine by the valine biosynthetic pathway operating in re-
verse order. The ∆AS∆KR double mutant is impaired in the
production of pyrazines and acyloins, but the production of
11 and 30 could be restored by the addition of acetoin to
the culture medium. These data strongly support the bio-
synthesis of alkylated pyrazines from acyloin derivatives.
The formation of 11 from 18 and ammonium salts has been
reported to proceed non-enzymatically under neutral to
slightly acidic pH and high salt concentrations. A similar
reaction was ruled out for the generation of pyrazines from
C. glutamicum by the addition of acetoin to agar plates with
medium alone at different pH values. Downstream steps
from the acyloin derivatives to the pyrazines likely proceed
via α-aminocarbonyl compounds. An indicator for this as-
sumption is the presence of trimethyloxazol (26) as a trace
component in the headspace extract of the ∆KR mutant
that likely arises from 3-aminobutanone (61) and acetyl-
CoA.
Strain Construction: Gene deletion was achieved by allelic replace-
ment of the wild-type gene by a shortened DNA fragment. The
knockout is usually carried out by a deletion of at least 300 bp to
be able to distinguish wild-type and knockout gene in gel electro-
phoresis. Here, 500 bp were deleted for single knockout of the ke-
tol-acid reductoisomerase. In the genome the large and small sub-
unit of the acetolactate synthase as well as the ketol-acid reducto-
isomerase are located next to each other. Due to this, the corre-
sponding knockouts were carried out by deletions containing more
base pairs. In the case of the large and small subunit of the aceto-
lactate synthase 1500 bp were deleted and in the case of the double
knockout of the acetolactate synthase and the ketol-acid reducto-
isomerase 2500 bp were deleted. The desired DNA fragment was
obtained in three steps by PCR. Sequences of the respective site-
specific primers P1-P4 for the deletions are given in Table SI-2 (see
Supporting Information). The obtained DNA fragments were sub-
sequently inserted into the vector pClik int sacB via two restriction
sites. Vector amplification was performed in E. coli DH5α and
NM522, carrying a plasmid-encoded copy of the C. glutamicum
specific DNA methyltransferase gene, to enable the appropriate
DNA methylation pattern. C. glutamicum was transformed by elec-
troporation as described previously.[28] Gene deletion was validated
by PCR analysis. As the genetic modification was restricted to the
internal coding sequence of each enzyme, no other genes were af-
fected.
Media and Growth Conditions: Cells were maintained at 30 °C on
agar plates containing complex medium with glucose (10 gL–1),
NaCl (2.5 gL–1), urea (2 gL–1), yeast extract (5 gL–1), beef extract
(5 gL–1), polypeptone (5 gL–1) and agar (20 gL–1). Afterwards,
cells were harvested by centrifugation (10,000ϫg, 15 min) and
2694
www.eurjoc.org
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Eur. J. Org. Chem. 2010, 2687–2695

Pyrazine Biosynthesis in Corynebacterium glutamicum
[4] J. S. Dickschat, I. Wagner-Döbler, S. Schulz, J. Chem. Ecol.
2005, 31, 925–947.
washed two times with 2.6% NaCl. Subsequently, cells were plated
on agar plates with CG12½ minimal medium that contained per
litre: glucose (10 g), K2HPO4 (16 g), KH2PO4 (4 g), (NH4)2SO4
(5 g), 3,4-dihydroxybenzoic acid (300 mg), CaCl2 (10 mg),
MgSO4·7H2O (250 mg), FeSO4·7H2O (10 mg), MnSO4·H2O
(10 mg), ZnSO4·7H2O (2 mg), CuSO4·5H2O (200 µg), NiCl2·6H2O
(20 µg), Na2MoO4·2H2O (20 µg), cyanocobalamin (100 µg), thi-
amine (300 µg), pyridoxal phosphate (4 µg), biotin (100 µg), and
agar noble (15 g). Additionally, the medium was supplemented with
valine, leucine, and isoleucine to a final concentration of 1 m due
to auxotrophy for these amino acids of the designed deletion mu-
tants. Agar plates were incubated for 3 d at 30 °C prior to CLSA
and GC-MS analyses.
[5] J. S. Dickschat, E. Helmke, S. Schulz, Chem. Biodiversity 2005,
2, 318–353.
[6] A. T. Dossey, M. Gottardo, J. M. Whitaker, W. R. Roush, A. S.
Edison, J. Chem. Ecol. 2009, 35, 861–870.
[7] J. H. Cross, R. C. Byler, U. Ravid, R. M. Silverstein, S. W. Rob-
inson, P. M. Baker, J. Sabino de Oliveira, A. R. Jutsum, J. M.
Cherrett, J. Chem. Ecol. 1979, 5, 187–203.
[8] J. W. Wheeler, M. S. Blum, Science 1973, 182, 501–503.
[9] B. P. Moore, W. V. Brown, M. Rothschild, Chemoecology 1990,
1, 43–51.
[10] K. E. Murray, J. Shipton, F. B. Whitfield, Chem. Ind. (London)
1970, 4, 897–898.
[11] A. Gallois, A. Kergomard, J. Adda, Food Chem. 1988, 28, 299–
Sampling of Volatile Compounds: The volatile products emitted by
the agar plate cultures were collected by using the CLSA headspace
technique.[29] Briefly, in a closed apparatus containing the agar
plates a circulating air stream was passed through a charcoal filter
(Chromtech GmbH, Idstein, Precision Charcoal Filter, 5 mg) for
24 h. The absorbed volatiles were eluted from the charcoal with
analytically pure dichloromethane (30 µL). The obtained solutions
were immediately analysed by GC-MS and stored at –80 °C. For
comparison agar plates containing the medium alone were investi-
gated. None of the volatiles shown in Table 1 have been detected
in these experiments.
309.
[12] T. Cheng, G. A. Reineccius, J. A. Bjorklund, E. Leete, J. Agric.
Food Chem. 1991, 39, 1009–1012.
[13]
[14]
T. Kosuge, T. Adachi, H. Kamiya, Nature 1962, 195, 1103.
A. L. Demain, M. Jackson, N. R. Trenner, J. Bacteriol. 1967,
94, 323–326.
[15]
[16]
G. P. Rizzi, J. Agric. Food Chem. 1988, 36, 349–352.
J. Kalinowski, B. Bathe, D. Bartels, N. Bischoff, M. Bott, A.
Burkovski, N. Dusch, L. Eggeling, B. J. Eikmanns, L. Gaigalat,
A. Goesmann, M. Hartmann, K. Huthmacher, R. Krämer, B.
Linke, A. C. McHardy, F. Meyer, B. Möckel, W. Pfefferle, A.
Pühler, D. A. Rey, C. Rückert, O. Rupp, H. Sahm, V. F. Wend-
isch, I. Wiegräbe, A. Tauch, J. Biotechnol. 2003, 104, 5–25.
GC-MS: GC-MS analyses were carried out on an Agilent 7890A
connected with an Agilent 5975C inert mass detector fitted with a
HP-5 (30 m, 0.25 mm i. d., 0.25 µm film) or BPX5 (25 m, 0.25 mm
i. d., 0.25 µm film) fused silica capillary column. Conditions were
as follows: inlet pressure: 77.1 kPa, He 23.3 mL min–1; injection
volume: 2 µL; transfer line: 300 °C; electron energy: 70 eV. The GC
was programmed as follows: 5 min at 50 °C increasing at 5 °C min–1
to 320 °C, and operated in splitless mode (60 s valve time). The
carrier gas was He at 1.2 mL min–1. Retention indices I were deter-
mined from a homologous series of n-alkanes (C8–C38). Identifica-
tion of compounds was performed by comparison of mass spectra
and retention indices with synthetic standards.
[17] A. Ohta, M. Ohta, Synthesis 1985, 216–217.
[18] B. Klein, N. E. Hetman, M. E. O’Donnell, J. Org. Chem. 1963,
28, 1682–1686.
[19] A. Fürstner, A. Leitner, Angew. Chem. Int. Ed. 2002, 41, 609–
612.
[20] E. Felder, D. Pitrè, S. Boveri, E. B. Grahitz, Chem. Ber. 1967,
100, 555–559.
[21] a) J. McMurry, T. Begley, The Organic Chemistry of Biological
Pathways, Roberts and Company Publishers, Greenwood Vil-
lage, 2005; b) J. McMurry, T. Begley, Organische Chemie der
biologischen Stoffwechselwege, Spektrum Akademischer Verlag,
München, 2006.
[22] H. Nursten, The Maillard reaction. Chemistry, biochemistry and
implications, The Royal Society of Chemistry, Cambridge,
2005.
[23] A. Adams, N. de Kimpe, Food Chem. 2009, 115, 1417–1423.
[24] C. Counet, D. Callemien, C. Ouwerx, S. Collin, J. Agric. Food
Chem. 2002, 50, 2385–2391.
Supporting Information (see also the footnote on the first page of
this article): Experimental procedures for the synthesis of com-
pounds and full characterisation data.
Acknowledgments
[25] B. Hölldobler, E. D. Morgan, N. J. Oldham, J. Liebig, J. Insect
Physiol. 2001, 47, 369–374.
[26] H. Torres-Contreras, R. Olivares-Donoso, H. M. Niemeyer, J.
Chem. Ecol. 2007, 33, 435–440.
[27] L. T. Nielsen, D. K. Eaton, D. W. Wright, B. Schmidt-French,
J. Cave Karst. Stud. 2006, 68, 27–31.
Funding by the Deutsche Forschungsgemeinschaft (DFG) within
the Emmy Noether programme (to J. S. D.) is gratefully acknowl-
edged.
[28] J. Becker, C. Klopprogge, O. Zelder, E. Heinzle, C. Wittmann,
Appl. Environ. Microbiol. 2005, 71, 8587–8596.
[29] J. S. Dickschat, S. C. Wenzel, H. B. Bode, R. Müller, S. Schulz,
ChemBioChem 2004, 5, 778–787.
[1] S. Schulz, J. S. Dickschat, Nat. Prod. Rep. 2007, 24, 814–842.
[2] H. C. Beck, A. M. Hansen, F. R. Lauritsen, FEMS Microbiol.
Lett. 2003, 220, 67–73.
[3] J. S. Dickschat, H. Reichenbach, I. Wagner-Döbler, S. Schulz,
Eur. J. Org. Chem. 2005, 4141–4153.
Received: February 3, 2010
Published Online: March 24, 2010
Eur. J. Org. Chem. 2010, 2687–2695
© 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
2695