Eur. J. Biochem.

248, 481-487 (1997)

0 FEBS 19Y7

Identification, isolation and biochemical characterization

of a phosphopantetheine:protein transferase
that activates the two type-I fatty acid synthases of Brevibacterium ammoniagenes
Hans-Peter STUIBLE, Sandra MEIER and Eckhart SCHWEIZER
Lehrstuhl fiir Biochemie, Universitat Erlangen-Niimberg, Germany

(Received 11 June 1997) - EJB 97 0825/4

Upon heterologous expression of the Brevibacreriuni ammoniagenes type-I fatty acid synthase FAS-A
in Escherichia coli, only the pantetheine-free apoenzyme is synthesized. Activation of FAS-A to its
holoform was achieved by transformation with a second B. ammoniagenes gene, PPTI, encoding a type-I
FAS-specific phosphopantetheine transferase. PPTI was identified as a coding sequence located immedi-
ately downstream of the second FAS gene present on the B. ammoniagenes genome, fasB. Due to this
linkage, PPTl was part of the cloned fasB DNA region and, consequently, FAS-B but not FAS-A was
synthesized as holoFAS in E. coli. PPTI encodes a protein of 153 amino acids and has a calculated
molecular mass of 16 884 Da. The PPTl gene product contains 25 76 identical and 42 % conserved amino
acids compared with the type-I1 acyl-carrier-protein- activating enzyme of E. coli. Although there is
essentially no intergenic region between jusB and PPTI, the PPTase gene is autonomously expressed in E.
coli if flanked by 200 bp of its endogenous 5‘ DNA. The structural independence of Pptlp was confirmed
immunologically, as specific antibodies react with the purified PPTase but not with FAS-B. Overexpres-
sion and purification of the His-tagged Pptlp allowed the in vitro activation of apoFAS-A. This holoen-
zyme synthesis requires, in addition to Pptlp, CoA and Mg” and leads to a specific FAS activity coinpa-
rable to that of natural B. ammoniugenes FAS-A. The reactivity of the in vitro-activated FAS-A was
verified by the optical FAS assay and by analysis of its in vitro products. In agreement with the known
overall colinearity of B. ammoniagenes FAS-B and the Saccharomyces cerevisiae FASI and FAS2 gene
products, a PPTI-like sequence is also observed at the C terminus of FAS2. However, in contrast to B.
arnnioniagenes PPTI, this sequence is an integral part of the yeast FAS2 gene. Thus, activation of type-I
fatty acid synthases may be accomplished by distinct truns-acting PPTase enzymes and by intrinsic ci.r-
acting PPTase domains.
Keywords: fatty acid synthase ; phosphopantetheine transferase ; apoprotein ; in vitro activation ; Brevihact-
erium ammoniagenes.

Several multienzyme systems, such as fatty acid synthases phosphopantetheine attached to an ACP-like domain within their
(FAS), polyketide synthase (PKS) and various peptide synthases, multifunctional proteins [ l ] . The mechanism of activation of
characteristically contain enzyme-bound 4-phosphopantetheine apoACP to holoACP has been elucidated by Vagelos and co-
as a prosthetic group [I, 21. Binding of acyl intermediates to workers for the type-I1 ACP component of Escherichia coli FAS
the terminal thiol of the cofactor provides the chemical energy [5]. According to these studies, 4-phosphopantetheine is trans-
necessary for subsequent transacylation reactions. In addition, ferred from CoA to the hydroxyl group of a specific serine resi-
the extended molecular structure of this prosthetic group may due of ACP by the enzyme, phosphopantetheine : protein trans-
allow it to act as a long flexible ‘arm’, thereby controlling the ferase (PPTase).
successive processing of intermediates by the various active sites
in the multienzyme system [3]. In dissociated type-I1 FAS or CoA + apoACP holoACP + 3‘,5’-ADP . (1)
PKS systems consisting of several monofunctional and structur- wra\e

ally independent component enzymes, 4-phosphopantetheine is
bound to a specific, low-molecular-mass acyl carrier protein
(ACP) [4]. In contrast, type-I FAS or PKS systems contain the
Correspondence to E. Schweizer, Lehrstuhl fur Biochemie, Uni-
versitat Erlangen-Nurnberg, Staudtstr. 5, D-91058 Erlangen, Germany
Fax: +49 9131 858254.
E-mail: eschweiz@biologie.uni-erlangen.de
Ahhreviations. FAS, fatty acid synthase; PPTase, 4’-phosphopante-
theine:protein transferase; ACP, acyl carrier protein ; PKS, polyketide
synthase.
Enzymes. Fatty-acid synthase (EC 2.3.1.85); 4’-phosphopantetheine
:protein transferase.
This enzyme, also designated ACP synthase, has been puri-
fied recently from an overproducing E. coli strain after the corre-
sponding gene had been cloned and sequenced [6]. Correspond-
ing transferases activating multifunctional type-I fatty acid syn-
thases have not been identified. Although the overall amino acid
sequences of type-I1 FAS or PKS acyl carrier proteins on the one
side, and the pantetheine-carrying domains of the corresponding
type-I multienzymes on the other, exhibit a rather low similarity,
there are a few positions close to the phosphopantetheine-bind-
ing serine residue that are highly conserved throughout most
systems [I]. Thus, it may depend on the substrate specificity
of the respective PPTases whether a type-I FAS apoenzyme is
482 Stuible et al. (Eur J. Biochern. 248)

Table 1. Plasmids used in this study.

Plasmid Essential properties Reference

pHPS 1 pBlue Script derivative containing thejusA gene and 2.5 kb of its 5' and 1 kb of its 3' flanking regions.
The,JiisA region is flanked by two Nor1 restriction sites 191
pGM44 pBlue Script derivative containing t h e f a d gene and 0.8 kb of its 5' and 3' flanking regions. The,fusB
region is flanked by ClnI and Not1 restriction sites 191
pHPS5 pGM44 derivative containing the entire fusA region. The &sA fragment was cloned into the single Not1
restriction site this work
pFASA/B A 1 pHPSS derivative containing a central 3-kb NsiI-XbaI deletion in the ,fa& reading frame this work
pFASAlB A2 pHPS5 derivative containing a central 7-kb BornHI deletion in the ,fu.sB reading frame this work
pFASA/B A3 pHPS5 derivative containing 2 kb of the C-terminal region of the ,fad3 insert this work
pHPS6 pHPSl derivative containing the PPTl gene and 200 bp of its 5' and 50 bp of its 3' flanking regions.
The PPTI reading frame was cloned into a unique XbuI site 2.5 kb upstream of thefusB initiation codon this work

activated in a heterologous host containing an endogenous type-

I1 system. For instance, the production of enzymatically active
fungal methylsalicylic acid synthase and human FAS have been
reported upon expression of these type-I enzymes in Streptomy-
ces coelicolor and E. coli, respectively [7, 81. As the phospho-
pantetheine-binding site sequences of the Brevibacterium ammo-
niugenes type-I and E. coli type-I1 FAS proteins differ consider-
ably, however, heterologous activation of the B. ammoniagenes
FAS was not expected a priori. Expression of the multifunc-
tional B. ammoniagenes FAS-A protein in E. coli resulted in the
production of enzymatically inactive apoenzyme [9]. In contrast,
heterologous expression of the second B. ammoniagenes type-I
FAS complex, FAS-B, led to the production of a fully active
holoenzyme. These findings, suggesting the existence of a
PPTase acting specifically on this type-I fatty acid synthase,
prompted us to search for the respective gene. As reported in
this study, a PPTase coding sequence was identified immediately
downstream of the B. ammoniagenes fusB gene. The enzymatic
activity of this novel phosphopantetheintransferase is demon-
strated by in vivo and in vitro apoFAS-A activation.
MATERIALS AND METHODS
Bacterial strains, culture condition and plasmids. For
plasmid amplification and B. ammoniugenes FAS expression the
E. coli strain DH5a (Gibco) was used. For heterologous expres-
sion of B. ammoniagenes FAS-A and FAS-B the respective E.
coli transformants were grown at 30°C in a medium containing
20 g/l casamino acids, 10 g/l yeast extract, 10 g/l NaC1, and
50 mg/l ampicillin, pH 7.5. The plasmids used in this study are
listed in Table 1.
DNA manipulations and biochemical techniques. Stan-
dard DNA-manipulation techniques were performed as de-
scribed elswhere [lo]. PCR were performed using Vent DNA
polymerase (Biolabs) according to the recommendations of the
manufacturer. Purification of B. ammoniagens FAS, FAS activ-
ity assays, Litro synthesis of fatty acids, and gas chromato-
graphic analysis of fatty acid methylesters were performed as
described 191.
Overexpression and purification of Pptlp. The PPTI gene
was amplified from pGM44 (Table 1) by PCR with Vent DNA
polymerase (Biolabs). The PCR primers created novel restriction
sites for Sphl at the N-terminal methionine and BamHI substitut-
ing the stop codon. The reading frame was cloned into the ex-
pression vector pQE70 (Qiagen) and transformed into the E. coli
strain M I 5 [pREP4] (Qiagen). The resulting construct expressed
Pptlp fused with a C-terminal His tag.
Fig. 1. Coexpression of the two B. umrnoniugenes FAS variants in E.
coli. Western blot analysis of extracts from pHPS5-transformed E. coli
cells was performed using specific antisera against B. nmnioniagenes
FAS-A (anti-A) and FAS-B (anti-B).
For Pptlp purification, a 50-ml overnight culture of E. coli
M15 transformants grown in Luria Broth with 50 mg/l kanamy-
cin and 100 mg/l ampicillin was harvested, suspended in 200 ml
fresh medium (without kanamycin) and induced with isopropyl-
1-thio-p-D-galactopyranosideat 3 mM. The induced cells were
grown for 7 h at 3O"C, collected by centrifugation and stored
at -70°C. The cell pellet was suspended in 2 0 m l 5 0 m M so-
dium phosphate pH 7.8, 300 mM NaCl and l mg/ml phenyl-
methylsulfonyl fluoride (buffer A). Cell lysis was performed by
sonication with a Branson B-I 2 sonifier. Insoluble components
were removed by centrifugation (Beckmann JA20, 18000 rpm,
20 min, 4"C), and 2 ml of Niz+-agarose (Qiagen) equilibrated
with buffer A was added. This preparation was stirred on ice for
60 min. Subsequently, the Ni" resin was washed twice with
buffer A and twice with 50 mM sodium phosphate, 300 mM
NaC1, 10% glycerol, pH 6.0 (buffer B), and loaded onto a 5-ml
column. The protein was eluted with a gradient from 0.05 M
to 0.5 M imidazole dissolved in buffer B. 1-ml fractions were
collected, and aliquots were analyzed on a an SDS/IS% poly-
acrylamide gel. The Ppt lp-containing fractions were stored in
buffer B at -70°C.
In vitro activation of apoFAS-A. For in vitro activation,
apoFAS-A was prepared freshly from E. coli DH5a as described
191. In vitro activation was carried out at 20°C in of 1 ml 0.3 M
potassium phosphate, pH 7.3, 3 mM dithiothreitol, 2.6 mM CoA
and 60 mM MgSO,. As MgSO, leads to precipitation of insolu-
ble material it was added immediately before starting the reac-
tion. 1.5 pg Pptlp (dissolved in 10 p1 buffer B) were sufficient
to fully activate 1 mg FAS within 5 min.
 aJ
In
FAS
FAS-A/B
18:l
Stuible et al. (Eur J . Biochenz. 248)

A
pHPS5

pFASNBA 1
-- fas 0

. . ...
...
PPTl fas A
483

C
0
Q
v) pFASNBA 2
!?!
ij -
0
m _ _ _ __ _..)..,I
pFASNBA3 ...... . _ _ _ _ __ ATER MPTACP KRKS

0
U

PPTl fas A
Control
* -
(1
~

pHPS6 ATER MPTAGP KRKS

Fig. 3. Functional identification of PPTl by successive deletion of the

b tase; MPT, rnalonyl-palmitoyl transferase; KR, ketoreductase; KS, /I-
retention lime
Fig. 2. Analysis of in vitro products synthesized by the purified FAS-
A/FAS-B complex isolated from E. coli pHPS5 transformants. After
transesterification, fatty acid methylesters were identified by gas chro-
matography. As a control, a reaction mixture with no malonyl-CoA
added was analyzed to identify unspecific signals.
Immunological techniques. The production of domain-spe-
cific antisera against synthetic B. ammoniagenes FAS peptides
was described previously [ 111. The Pptlp-specific antiserum
was raised against the internal peptide HIPGFAEQLSRPGST-
FEQV (synthesized by D. Palm, University of Wurzburg, Ger-
many). Experimental conditions were as described recently for
the antisera against FAS-specific peptides [I I].
RESULTS
Activation of B. ammoniagenes FAS-A and FAS-B in E. coli.
Only one of the two B. ammoniugenes type-1 FAS multien-
fuse coding sequence. Boxed areas indicate the fusA (open) and ,fusS
(dark) DNA sequences contained in pHPS5. Arrows delimit the .fnsA,
fusB and PPTl reading frames. The order of functional domains encoded
by the respective genes or gene fragments is listed. (A) The three dele-
tion constructs derived from pHPS5 are depicted. (B) The construct
shown contains the PPTI coding sequence incorporated, as a PCR frag-
ment, into pHPSl at a position 2.5 kb upstream of the fulsil translational-
start codon. The DNA sequence and deduced amino acid sequence of
this insert is depicted in Fig. 4. AT. acetyl transferase; ER, enoyl reduc-
ketoacyl synthase.
zymes, FAS-B, is expressed in E. coli as enzymatically active
holo FAS, while FAS-A is synthesized as phosphopantetheine-
less apo-enzyme [9]. As the presence of an endogenous 2nd
FAS-B - specific E. coli PPTase appears unlikely, these findings
suggested that the observed PPTase activity was encoded by the
transforming FAS-B plasmid itself. To study the capacity of this
PPTase to activate, in addition to FAS-B, the second B. anzmo-
niagenes FAS complex, FAS-A, both FAS enzymes were coex-
pressed in E. coli. For this purpose, the two B. ammoniagenes
FAS genes were combined, together with their flanking DNA
regions, within the 28-kb plasmid, pHPS5. Western blot analysis
of E. coli pHPS.5 transformant extracts with specific FAS-A and
FAS-B antisera revealed the heterologous expression of both ,fas
genes at high and comparable rates (Fig. 1).
The FAS-AIFAS-B enzyme mixture isolated from the fa.sA/
jusB cotransforniants exhibited a specific overall FAS activity
of 370 mU/mg, which corresponds to that of the homogeneous
FAS-A and FAS-B enzymes. This finding suggested that both

Table 2. In vitro products of purified holoFAS-A obtained by different activation procedures. FAS-A was either activated in E. coli by
coexpression of the PPTase-encoding DNA contained in the indicated plasmids, or was activated in vitro by incubation with purified Pptlp. As a
control, the in vitro products of FAS-A and FAS-B purified from B. arnnioniagenes and E. coli, respectively, were determined. n.d., not detectable.

Enzymes Source Transformed plasmid Fatty acid products

16:O 18:O 18: 1

FAS-A B. ummotiiagenes none 5 15 80

FAS-A E. coli pHPS 1 n.d. n. d. n.d.
FAS-5 E. coli pGM44 92 8 n.d.
FAS-A and FAS-B E. coli pHPS.5 24 21 55
FAS-A E. coli pFASAJB A 1 5 10 85
FAS-A E. coli pFASA/B A 2 8 14 78
In vim-activated FAS-A E. coli pHPS 1 13 16 71
484 Stuible et al. ( E M J. Biochem. 248)
Table 3. In vivo and in vitro activation of apoFAS-A. In vivo activation
was achieved in E. coli upon transformation with the indicated fiisB
deletion constructs. In vitro activation was performed by incubation of
apoFAS-A with purified Pptlp (Materials and Methods). FAS activity
was determined using the partially purified preparations resulting after
the ammonium sulfate fractionation and ultracentrifugation steps [I 11.
Enzyme sources were as described in Table 1. For quantification of over-
all FAS activity, the inherent and phosphopantetheine-independent re-
duction of acetoacetic acid ethyl ester by the /l-ketoacyl reductase com-
ponent activity was used as an internal standard.
Enzyme Plasmid FAS activity
% of ketoreduc-
tase activity
FAS-A pHPS 1 0 tgcggtaccagcggtaaaaaccgatcgggaaattgccgcaattagagcgcagctatttgatga
FAS-B pGM44 34
FAS-A pFASAlB A 1 30
FAS-A pFASAIB A2 37
FAS-A pFASAIB A3 24
FAS-A pHPS6 18
In vitro-activated FAS-A pHPSl 32
FAS variants rather than only FAS-B were enzymatically active
in the mixture. Otherwise, only half of this specific activity
should be expected. Biosynthesis of enzymatically active FAS-A
upon expression with FAS-B was further confirmed by dem-
onstrating the capacity of the FAS-A/FAS-B mixture to synthe-
size oleic acid (Fig. 2). This unsaturated fatty acid is the main
product of FAS-A, while FAS-B synthesizes exclusively satu-
rated fatty acids 191. The activity of FAS-A in the FAS-A/FAS-B
enzyme preparation was also manifested by the relative amount
of palmitic acid synthesized (Table 2). Palmitic acid is the main
product of FAS-B but constitutes only a minor component
among the FAS-A products. Therefore, its relative amount
should decrease as the activity of FAS-A in the FAS-A/FAS-B
preparation increases. The type and relative amount of fatty
acids synthesized by the FAS-A/B enzyme mixture isolated from
E. coli essentially corresponds to the arithmetic sum of the prod-
ucts synthesized by the homogeneous and enzymatically active
FAS-A and FAS-B complexes (Table 2). In agreement with the
immunological data (Fig. 1) a slight predominance of FAS-A
over FAS-B is indicated by the in vitro products obtained with
the FAS-A/FAS-B mixture. The reported results indicate that
FAS-A is activated in E. coli when expressed with FAS-B, while
being synthesized as apoFAS in the absence of FAS-B. Thus,
activation of the FAS-A and FAS-B apoenzymes in E. w l i was
obviously an intrinsic capacity of the cloned fasB DNA.
Deletion analysis of fasB in the fasA/fasB construct pHPS5.
The PPTase coding sequence apparently contained within the
cloned fusB DNA fragment was identified by deleting selected
parts of this DNA. The three deletion constructs thus obtained
are depicted schematically in Fig. 3A. In pFASA/Bdl and
pFASAIBA2, internal segments of 3 kb and 7 kb, respectively,
were deleted from fasB. In both cases, catalytic domains with
essential functions in fatty acid synthesis were removed. In
pFASA/BA3, a 10-kb deletion encompasses essentially the entire
fasB coding region and all of its 5' flanking DNA. In all con-
structs, the,fasA DNA remained intact and continued to be asso-
ciated with the undeleted parts of ,fusB. The three fasA/dfasB
plasmids were transformed into E. coli and the heterologous
FAS complex synthesized in the transformants was isolated. Due
to their grossly altered protein structures, it was not expected
that the truncated fnsB products, if synthesized at all, were still
contained in the FAS preparations thus obtained. Instead, only
9
v)
Q R L T S A I S G G P A L Y E R
cag~gactaacctcgqctatc~ccggtgg~~cggcgttgtatgagcgccccgtggaccgcaa~
P V D R N
3
L G G T G D V V K E R E A A V L L D D A A
~tgggtggaaccggtgacgtagtcaaggaacgcgaagcagcagtattgcttgacgat~cg
R L R G A V L E P S T S A E S G K Q Q * M L
cggctgcgcggagccgtc~tcgagccgagcacttccgcagaatc~aagcagcagtagtgc -
i
t
D N R E A M T V G V D L V H I P G F A E Q
tcgacaaccgtgaagcgatgaccgtgggtgtggacttggtccacatcccc~tttgccgagc
L S R P G S T F E Q V F S P L E R R H A Q
aattgtcgcgccctggttcgacttttgagcaagtgttttcgccgttggaacgtcgtcatgctc
T R R D A A A D A T N S S L A
aaacgcgccgtgacgctgcagcggatgctacgaattcgagccttgcgggttcacggactgagc
L A G R W A A K E A F I K A W S Q A I Y G
G S R T E H
-k
acctggctgggcggtgggcggcaaaagaagcgttcatcaaggcgtggtcgcaagcgatctacg
K P P V I E P D L V N F A E I E V L P D R
gcaagccaccagtgattgaaccagacctggtgaacttcgcagagatcgaagtcttgcccgacc
W G R V A L Q L K G E V A A K L Q E S I G
4ctggggcagggtagcgctgcagcttaaaggtgaagttgctgcaaaacttcaggaatcaata~
D V E L A L S I S H D G D Y A T A Q C L L
gcgacgtggagctggcgctgagcatcagccaegatgatggcgattacgccaccgcgcagtgcctgc
R Y Q R '
gtgcattgctcccatggt
Fig. 4. Corrected DNA and deduced amino acid sequences of the C-
terminal part of the B. arnntoniagenes fuse gene. The three bases,
which have been added to the previously published sequence (EMBL
data base accession no. X64795) are indicated in black. *, stop codons.
The translational start site of the PPTl reading frame and its putative
ribosomal-binding site are underlined.
the mutationally unaffected FAS-A complex should be isolated
from these transformants. The various FAS preparations were
assayed for their enzymatic activities and their in vim-product
spectra. If the putative PPTase was part of the FAS-B multien-
zyme, it should be inactivated with FAS-B by the extensivefasB
deletions contained in the mutant plasmids (Fig. 3A). As a con-
sequence, only apoFAS-A with no overall FAS activity should
be produced. However, all FAS-A preparations isolated from the
fasB deletion constructs elicited essentially the same overall
FAS activity (Table 3). Furthermore, the in vitro-product spectra
of the FAS preparations obtained from the pFASA/Bdl and
pFASA/Bd2 transformants were identical to those of intact
FAS-A isolated from B. ammoniagenes fusB disruptants (Table
2). As the capacity of activation of apoFAS to holoFAS was
retained even after deletion of essentially the entire FAS-B read-
ing frame, the PPTase coding sequence was located to the short
segment of fu.sB DNA present in the pFASA/BA3 construct
(Fig. 3A).
Identification of the B. ammoniagenes PPTase coding region.
The PPTase-encoding DNA segment in pFASA/BA3 was scruti-
nized by examination of its previously determined nucleotide
sequence. This analysis uncovered two sequencing errors that
may have been caused in the earlier work by the high G + C
content of this area. The corrected fasB sequence terminates
273 bp upstream of the original translational stop site. Immedi-
ately downstream of the fasB stop codon another reading frame
is now initiated, encoding a protein of 153 amino acids (Fig. 4).
This protein shares 25% identical and 4 2 % conserved amino
acids with the E. coli ACP synthase (Fig. 5 ) and includes two
peptide motifs typical for several other phosphopantetheine
transferases. In agreement with the failure of FAS-A activation
in E. coli there is no comparable coding sequence associated
with the B. nmmmoniagenes fasA gene. Experimental data to
support the suggested PPTase-encoding function of the reading
frame was obtained by its PCR amplification and subsequent
expression with fasA. The respective construct, pHPS6, contains
the putative PPTase coding sequence and 200 bp of its 5' DNA
integrated into the fasA plasmid, pHPS1, at a position 2.5 kb
upstream of the fusA translational-initiation site (Fig. 3 B). The
partially purified FAS-A preparation isolated from E. coli cells
transformed with this plasmid exhibited overall FAS activity,
 Stuible et al. (Eur: J. Biochem. 248) 485

E.coli ACPS
B.aWm. P P T l
-1 1
. . ....
s.cer. FA.52 . . . . . . . . . . . . .vs
P * P a t - PAS2 ............ .TE

* *** * ***
E.coli ACPS . . . .....RN
B-aWm. P P T l P -.DA
S.cer. FA82 S&PS
P.pat. FAS2 P,€LE.&S .......
* * * * * * ** * * *
E.coli ACPS S’
B-aWm. P P T l YQR’
S.cer. FAS2 KK’
P - P a t . FASZ F‘

Fig. 5. Sequence comparison of the E. coli phosphopantetheine:acyl-carrier-proteinsynthase with the B. ammoniagenes PPTl product and
the putative PPTase sequences at the C termini of S. eerevisiae and f? patuZrcnt FASZ *, positions that are conserved in all four proteins.
Positions being identical or conserved between the bacterial and at least one of the two fungal sequences are marked in black.

A B C
45 -
36 -
29 -
24 -
20 -
14 -

Fig. 7. Purification and immunological identification of Pptlp. The

Fig.6. Western blot analysis of partially purified FAS-B with anti-
bodies raised against FAS-B (B) and Pptlp (C) peptide sequences.
The nitrocellulose blots used for immuno-staining were from different
lanes of the same gel and contained identical amounts of FAS-B protein.
(A) A third lane of this gel was subjected to Coomassie staining. The
arrow indicates the position of FAS-B (320 kDa).
PPTI reading frame in the pQE70 His-tag expression vector was ex-
pressed in E. coli and purified by chromatography on Ni*+-agaroseas
described in Materials and Methods. The eluted fractions were analyzed
by SDSPAGE on a 15% gel. The fractions containing the homogenous
PPTase were subjected to Coomassie staining (A) and Western blot
analysis (B) using antibodies raised against an internal Pptlp peptide
sequence.

which resulted from the presence of the PPTase-like reading

frame (Table 3). Therefore, this gene was designated PPTl. Cor- Table 4. In vitro activation of B. ammoniagenes apoFAS-A by puri-
respondingly, FAS activity was abolished upon elimination of fied Pptlp. Protein extracts from non-transformed and,fusA-transformed
E. coli cells were prepared as described in Table 3. ApoFAS-A was
this fragment from pHPS6 (data not shown).
purified from E. coli DH5n as described [9]. Experimental conditions
The structural independence of FAS-B and Pptlp, as indi- were as described in Materials and Methods.
cated by the above observations, was further demonstrated using
Pptlp-specific antibodies. The antiserum raised against the in- Experimental Specific FAS activity of
ternal peptide HIPGFAEQLSRPGSTFEQV of Pptlp exhibited conditions
no cross-reaction with partially purified FAS-B (Fig. 6), but apo FAS-A E. cofi protein extmct
showed a strong signal with the purified PPTase (Fig. 7). Under
the same experimental conditions, an intensive immunological ,$USAtrans- non-trans-
reaction was observed between FAS-B and an antiserum directed formed formed
against a FAS-B-specific peptide (Fig. 6). Expression of PPTZ
mU/mg
in E. coli was low, since in crude extracts of E. colifasB trans-
formants no low-molecular-mass protein was detected, which Complete system 320 130 0
cross-reacts with the anti-Pptlp Ig. This result corresponds to Without Pptlp 0 0 0
data published on the endogenous E, coli ACP synthase which, Without Mg”, CoA 0 0 0
i n contrast to the high abundance of ACP, is expressed at an
extremely low level 161.
Heterologous expression of Pptlp and in vitro activation
of apoFAS-A. For purification and in vitro analysis, Pptlp was essentially homogeneous Pptlp preparation was obtained, which
expressed with a C-terminal His tag using the E. coli expression cross-reacted with the specific anti-Pptlp serum and exhibited
vector pQE70 (see Materials and Methods). The His-tagged the expected molecular mass of about 18 kDa (Fig. 7). The
Pptlp was isolated from the E. coli cell extract by affinity chro- PPTase activity of the eluted protein was demonstrated by using
matography on Ni2+-agaroseunder non-denaturating conditions. it for in vitro activation of apoFAS-A. The pantetheine-free apo-
In addition to the enzymatic activity, this characteristic indicated FAS-A enzyme isolated from the E. colifusA transformants was
the native structure of the heterologously expressed protein. An efficiently converted to enzymatically active holoFAS upon in-
486 Stuible et al. (Eui: J. Biochem. 248)
cubation with CoA, Mg" and the purified PPTase. No FAS acti-
vation was observed if either one of the components, Pptlp,
CoA or Mg' ' , were incubated separately with apoFAS-A (Ta-
ble 4). In addition to purified apoFAS-A, a partially purified
preparation from E. c d i &A transformants could be activated
under these conditions. In this case, the PPTase specificity of
the reaction became evident from the finding that an extract of
non-transformed E. coli cells was inactive (Table 4). The spe-
cific activity of in vitro-activated holoFAS-A (320 mU/mg) was
comparable to that observed with FAS-A isolated from B. um-
moniageizes [9], suggesting the same degree of FAS-A panteth-
einylation in vivo and in v i m . Evidence for the in vitro conver-
sion of apoFAS-A to the native holoenzyme was derived from
the analysis of its in vitvo-synthesized products. These products
were identical to those synthesized by holo FAS-A isolated from
B. ammoniagenes (Table 2).
DISCUSSION
The present paper describes an experimentally well docu-
mented phosph0pantetheine:protein transferase that specifically
activates a type-I FAS with a structurally integrated ACP do-
main. Several procaryotic PPTases acting on individual type-I1
acyl or peptidyl carrier proteins have been described recently
1121 but all attempts to identify a corresponding, type-I FAS-
specific enzyme had been unsuccessful. The B. ammoniagenes
PPTase characterized in this study that activates the type-I FAS,
FAS-A and FAS-B, in the heterologous E. coli system is sug-
gested to have the same function in the natural host, B. ammo-
iziugenes. For FAS-A, formation of the holoenzyme was also
demonstrated in vitro using purified apoFAS-A and Pptlp.
The PPTl gene encoding the B. ammoniagenes PPTase was
uncovered after correction of two sequencing errors contained
in the previously published fasB gene [13]. The corrected fu.sB
sequence ends now at a position similar to the end of,fasA, i.e.
91 amino acids earlier than before. The reading frame for PPTl
begins immediately downstream of fa.sB. The lack of an in-
tergenic region between jasB and PPTl raises the question on
the mechanism of in vivo transcriptional initation of PPTl. The
transcription of PPTI with fasB into a bicistronic mRNA or an
overlapping of the PPTl transcriptional initiation site with the
end of fasB appears possible. In E. coli, transcription of the B.
ummoniugenes PPTl gene from various fazB deletion constructs
was independent of ,fasB demonstrating that transcription of the
two genes is not necessarily coupled. The same holds true for the
insertional inactivation of fusB in the B. ainmoniugenes genome,
which allows the production of enzymatically active FAS-A 191
and, thereby, demonstrates the expression of PPTl being inde-
pendent of jusB. Thus, PPTl may normally be transcribed with
fasB but may, alternatively, be transcribed independently offusB
from surrogate initiation sites in its upstream region. Even
though expression from these sites is possibly less efficient, this
effect may be compensated by the gene dosage of the multicopy
plasmids used in this study. Other than transcription, the transla-
tion of PPTl being initiated from its well conserved Shine-
Dalgarno sequence at the end ofjuxB, should not be complicated
by the absence of an intergenic region. In Corynebucteriu suc-
cessful translation is possible even from overlapping reading
frames, as demonstrated recently for the genes of two asparto-
kinase subunits in the related organism, Corynebacterium glu-
tumicum [14]. Although insertion of a single nucleotide at the
end of fasB would eliminate the translational discontinuity be-
tween the ,fuss and PPTl reading frames, a corresponding se-
quencing error can be excluded by the immunological data ob-
tained with the Pptlp-specific serum. Therefore, the in vivo
function of Ppt l p as a trans-active PPTase for FAS-A and FAS-B
is demonstrated. Correspondingly, the in vitro activation of apo-
FAS-A using His-tagged Pptlp occurred very readily.
As was reported previously, the B. uinmoniugenes FAS pro-
teins are colinear to a hypothetical head-to-tail fusion of the two
yeast FAS subunits, p and u [13]. A significant degree of se-
quence identity is exhibited over nearly the entire length of the
yeast and B. ammoniagenes FAS proteins. In FAS-B this se-
quence similarity extends beyond the end of fusB and includes
the PPTl coding sequence (Fig. 5). The C terminus of yeast FAS
subunit a and the PPTl gene product share, within a 140-amino-
acid overlap, 28% identical and 42% conserved positions. No
other yeast protein exhibits a comparably high degree of simi-
larity to the type-I FAS-activating phosphopantetheine :protein
transferase of B. atnmoniagenes. Among the various sequences
conserved in the four known or suggested PPTases (Fig. 5 ) , there
are two regions which appear to be of a particular functional
importance. The sequence motifs, (Vor1)GXD and (WorF)XX-
KE(AorS)XXK have recently been reported to be present in all
known and mostly procaryotic PPTases [12]. These findings
suggest that the PPTase function may represent an integral part
of the yeast FAS complex. Previous results from our laboratory
are in agreement with this conclusion. For instance, no yeast
mutants that produce a pantetheine-free FAS enzyme and are
non-allelic to FAS2 have been isolated (Schweizer, E., unpub-
lished data). On the other hand, a specific class of yeast FAS2
mutants producing a pantetheine-free FAS protein maps within
the C-terminal region of FAS2 [15]. Some of the respective mu-
tant alleles have been identified by DNA sequencing and were
found to be restricted, with no exception, to one of the two
highly conserved motifs mentioned above (data not shown). In
particular, the conserved glycine residue within the (Vor1)GXD
motif was affected in several cases, indicating the importance of
this position, as suggested earlier 1151. In agreement with the
sequence similarity of the known fungal type-I FAS, a PPTI-
like domain is present in the u chains of these FAS proteins [12].
As an example, the corresponding Penicillium putulum FAS2
sequence is listed in Fig. 5. Thus, a PPTase function seems to
be associated not only with B. ammoniagenes FAS-B but also
with the structurally related fungal FAS enzymes. The extent
of this association, however, differs in the two systems. In B.
ammoniagenes, FAS-B and PPTase are encoded by distinct read-
ing frames which may or may not be transcribed coordinately
into a bicistronic mRNA. In yeast, the two reading frames are
fused and therefore the PPTase may represent an additional do-
main of the FAS multienzyme. Hence, the potential capacity of
self-activation observed among these fungal type-I fatty acid
synthases seems to represent a novel quality of this many-sided
class of multifunctional enzymes.
This work was supported by the Deutsche Fnr.schungsgemeinschuft
and by the Fonds der Chenzischen Industrie. In addition, financial sup-
port from the Biotech (B102CT-942067)and Human Capital and Mobil-
ity (ERBC HRXCT 940570) programs of the European Commission is
gratefully acknowledged. We thank Dr Jorn Kalinowski, University of
Bielefed, for sharing with us his expertise in the molecular genetics of
Coqnehacteria,
REFERENCES
1. Hopwood, D. A. & Sherman, D. H. (1990) Molecular genetics of
polyketides and its comparison to fatty acid biosynthesis, Annu.
Rev. Genet. 24, 37-66.
2. Kleinkauf, H. & von Dohren, H. (1996) A non-ribosomal system of
peptide biosynthesis, ELMJ. Biochem. 236, 335-351.
3. Schweizer, E., Piccinini, F., Duba, C., Gunther, S., Ritter, E. & Ly-
nen, F. (1970) Die M~ilon~l-Bindung.sstellen des Fettsauresynthe-
tase-Komplexes der Hefe, EUKJ. Biochem. 15, 483-499.
 Stuible et al. ( E M J . Biocheni. 248)
4. Magnuson, K., Jackowski, S., Rock, C. 0. & Cronan, J. E. (1993)
Regulation of fatty acid biosynthesis in Escherichia coli, Micro-
biol. Rev. 57, 522-542.
5. Prescott, D. J., Elovson, J. & Vagelos P. R. (1975) Acyl carrier
protein synthetase, Methods Enzyrnol. 25, 95 - 101.
6. Lambalot, R. H. & Walsh, C. T. (1995) Cloning, overexpression and
characterization of the Escherichia coli holo-acyl carrier protein
synthase, J. Biol. Chern. 270, 24658-24661.
7. Bedford, D. J., Schweizer, E., Hopwood, D. A. & Khosla, C. (1995)
Expression of a functional fungal polyketide synthase in the bac-
terium Streptomyces coelicolor A3(2), J. Bacteriol. 177, 4544-
4548.
8. Jayakumar, J., Huang, W. Y., Raetz, B., Chirala, S. S. & Wakil, S.
J. (1996) Cloning and expression of the multifunctional human
fatty acid synthase and its subdomains in Escherichia coli, Proc.
Nut1 Acad. Sci. USA 93, 14509-14514.
9. Stuible, H . 2 , Meurer, G. & Schweizer, E. (1997) Heterologous ex-
pression and biochemical characterization of two different type I
fatty acid synthases from Brevibacteriurn arnmoniagenes, Eul: J.
Biochem. 247, 268-273.
10. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) Molecular clon-
ing: a laboratory manual, 2nd edn, Cold Spring Harbor Labora-
tory Press, Cold Spring Harbor NY.
487
11. Stuible, H.-P., Wagner C., Andreou, I., Huter, G., Haselmann, J. &
Schweizer, E. (1996) Indentification and functional differentiation
of two type I fatty acid synthases in Brevihacteriuni umnio-
niagenes, J. Bactrriol. 178, 4787 -4793.
12. Lambalot, R. H., Gehring, A. M., Flugel, R. S., Zuber, P.. LaCelle,
M., Marahiel, M. A., Reid, R., Khosla, C. & Walsh, C. (1996) A
new enzyme superfamily - The phosphopantetheinyl transfer-
ases, Chemistry and Biology 3, 923-936.
13. Meurer, G., Biermdnn, G., Schutz, A,, Harth, S. & Schweizer, E.
(1992) Molecular structure of the multifunctional fatty acid syn-
thase gene of Brevibacteriurn ammoniagenes: its sequence of ca-
talytic domains is formally consistent with a head to tail fusion
of the two yeast genes FASl and FAS2, Mol. Gen. Genet. 232,
106- 116.
14. Kalinowski, J., Cremer. J., Bachmann, B., Eggeling, L., Sahm, H. &
Puhler, A. (1991) Genetic and biochemical analysis of the aspar-
tokinase from Corynebacterium glutamicuni, Mol. Microbiol. 5,
1197- 1204.
15. Schorr, R., Mittag, M., Muller, M. & Schweizer, E. (1994) Dif-
ferential activities and intramolecular location of fatty acid syn-
thase and 6-methylsalicylic acid synthase component enzymes, J .
Plant Physiol. 143, 407-415.