We have investigated the interplay between the adenylation A domain and the peptidyl carrier protein in the gramicidin S synthetase I EC 5.1.1.11 via partial tryptic digests, native PAGE
Trang 1the cross-talk between adenylation and peptidyl-carrier protein domains of nonribosomal peptide synthetases
Joachim Zettler and Henning D Mootz
Technische Universita¨t Dortmund, Germany
Introduction
A myriad of bioactive peptides is assembled by
non-ribosomal peptide synthetases (NRPSs) Examples of
such nonribosomal peptides (NRPs) include the
immu-nosuppressant cyclosporine A, the antibiotic
vancomy-cin, and the iron-chelating siderophore enterobactin
During the stepwise biosynthesis of the NRP, the
intermediates are covalently attached to the NRPS
template [1–4] Genetic and biochemical analysis of
NRPSs have revealed the modular organization of
these multifunctional mega-enzymes The
incorpora-tion of one building block into the growing peptide chain requires one module consisting of several spe-cialized catalytic domains [2–4] Figure 1A shows the interplay of individual domains during a catalytic cycle
at an elongation module In step 1, the adenylation (A) domain selects a cognate amino acid and activates
it by forming the corresponding amino acyl adenylate Then, as shown in step 2, the 4¢-phosphopantetheine moiety (Ppant) of the peptidyl-carrier protein (PCP) domain binds the activated acyl group as a thioester
Keywords
A-domain inhibitor; conformational change;
domain interaction; nonribosomal peptide
synthetase (NRPS); peptide antibiotics
Correspondence
H D Mootz, Technische Universita¨t
Dortmund, Fakulta¨t Chemie – Chemische
Biologie, Otto-Hahn-Str 6, 44227 Dortmund,
Germany
Fax: +49 0 231 755 5159
Tel: +49 0 231 755 3863
E-mail: Henning.Mootz@tu-dortmund.de
(Received 31 August 2009, revised 15
December 2009, accepted 16 December
2009)
doi:10.1111/j.1742-4658.2009.07551.x
Nonribosomal peptide synthetases serve as multidomain protein templates for producing a wealth of pharmaceutically important natural products For the correct assembly of the desired natural product the interactions between the different catalytic centres and the reaction intermediates bound
to the peptidyl carrier protein must be precisely controlled at spatial and temporal levels We have investigated the interplay between the adenylation (A) domain and the peptidyl carrier protein in the gramicidin S synthetase I (EC 5.1.1.11) via partial tryptic digests, native PAGE and gel-filtration analysis, as well as by chemical labeling experiments Our data imply that the 4¢-phosphopantetheine moiety of the peptidyl carrier protein changes its position as a result of a conformational change in the A domain, which
is induced by the binding of an amino acyl adenylate mimic The produc-tive interaction between the two domains at the stage of the amino acyl transfer onto the 4¢-phosphopantetheine moiety is accompanied by a highly compact protein conformation of the holo-protein These results provide the first biochemical evidence for the occurrence of conformational changes
in the cross-talk between A and peptidyl carrier protein domains of a multi-domain nonribosomal peptide synthetase
Abbreviations
A domain, adenylation domain; ANL, aryl and acyl CoA synthetases, NRPS A domains and firefly luciferases; A-PCP(D-4Cys), the gramicidin S synthetase I A domain and the PCP domain (C60F, C331A, C376S, C473A); A-PCP, the gramicidin S synthetase I A domain and the
PCP domain; ApCpp, adenosine-5¢-[(a,b)-methyleno] triphosphate; C domain, condensation domain ; E domain, epimerisation domain; eq., equivalent; GrsA, gramicidin S synthetase I; NRP, nonribosomal peptide; NRPS, nonribosomal peptide synthetase; PCP domain, peptidyl carrier protein domain; Ppant, 4¢-phosphopantetheine moiety; PP i, pyrophosphate; Sfp, 4¢-phosphopantetheine transferase involved in surfactin production; TAMRA, tetramethyl-rhodamine; TE domain, thioesterase domain; TycA, tyrocidine synthetase I; TycB, tyrocidine synthetase II.
Trang 2During steps 3 and 4, peptide-bond formation is
cata-lyzed at the acceptor position of an upstream
conden-sation (C) domain and at the donor position of a
downstream C domain In the case of an initiation
module, the upstream C domain is omitted, whereas in
the case of the last module of an NRPS assembly line
a thioesterase (TE) domain usually replaces the
down-stream C domain Additional domains can be included
within a module to achieve further diversification (e.g
epimerization, N-methylation and oxidation domains)
For most NRPSs, the arrangement of the modules
on the primary sequence is co-linear with the
assem-bled NRP However, modules can also be used
itera-tively, or the relationship between the module and
domain compositions and the NRP can be more
complex [5]
Crystallographic studies and NMR investigations
have revealed the 3D structures of representative
members of each essential NRPS domain in an
iso-lated form [6–11] For example, A domains belong,
together with acyl-CoA synthetases, aryl-CoA
synthe-tases and firefly luciferases, to the ANL superfamily
of adenylating enzymes Congeners of this class
con-tain two subdomains, the larger N-terminal
sub-domain (AN) (400–500 amino acids in size) and the
smaller C-terminal subdomain (AC) (100–150 amino
acids in size) Consistent with this enzyme class, crys-tallographic and mutational studies (for a review see [12]) have revealed that A domains probably adopt three different conformations during the catalytic cycle and use large-scale domain rotations [12] to catalyze the two half reactions, namely amino acyl adenylate formation and thioesterification onto the Ppant group of the C-terminal PCP (steps 1 and 2 in Fig 1A and see Fig S1A for representative structures
of the different conformations) [12–15] These three conformations include an open conformation with lit-tle contact between the subdomains when no sub-strates are present [12,13] Binding of ATP and the amino acid substrate results in a rotation of the subdomains towards the ‘adenylation conformation’ and closes the active site from bulk solvent [12,13] Breaking of the a,b-phosphodiester bond of ATP, and the subsequent release of pyrophosphate, induces
a 140 rotation of AC with respect to AN, giving a conformation where thioester formation takes place [12,13] However, a structure of an A domain in this
‘thioester-conformation’ with an interacting PCP is not available
PCP domains can also exist in different, intercon-verting conformations [8] Previous studies have shown that the structure of the PCP domain is dependent not
Fig 1 (A) Scheme of the reactions catalyzed by a minimal NRPS elongation module From a functional perspective, the PCP is the central domain in each module The 4¢-phosphopantetheine prosthetic group (Ppant) must interact in a minimal elongation cycle with the A-domain
to become acylated by the amino acyl adenylate intermediate, with the upstream C domain receiving the amino acyl or peptidyl group from the preceding module for peptide bond formation, and with the downstream C domain, or alternatively with the TE domain at the last mod-ule of the NRPS template, to deliver the peptidyl moiety and free the Ppant for the next elongation cycle Additionally, optional domains, which also need to specifically interact with the amino acylated PCP, might be incorporated in the module (B) Chemical structures of the A-domain inhibitors used in this study (5¢-O-[N-( L -phenyl)-sulfamoyl] adenosine) (1) and (5¢-O-[N-( L -prolyl)-sulfamoyl] adenosine) (2).
Trang 3only on its post-translational state (apo or holo) but
also on the presence or absence of PCP-interacting
domains or external enzymes [8,16] (for reviews see
[17,18]) Besides these intradomain dynamics, different
intramodular positions of the PCP domain, relative to
other NRPS domains, seem likely considering the
over-all architecture of an NRPS module Marahiel and
co-workers [19] recently reported a crystal structure of
the 144 kDa termination module, SrfA–C, involved in
the biosynthesis of surfactin, which is composed of
four domains (C-A-PCP-TE) In this structure, the
C domain and the ANform a structured platform onto
which the AC and the PCP domain are tethered
Fur-thermore, this SrfA–C construct lacks the attachment
site of the Ppant group (S1003A mutant) and shows
the PCP on the acceptor site of the C-domain The
dis-tance of the S1003A residue to the active-site histidine
of the C-domain is 16 A˚, making it within the reach of
the missing Ppant arm (capable of reaching 20 A˚
dis-tance at full linear extension) However, the disdis-tances
to the other catalytic centres of the SrfA–C module,
namely the A domain and the TE domain, are 57 and
43 A˚, respectively, too large to support the proposed
‘swinging arm model’ This model suggests that the
length and flexibility of the 18–20 A˚ Ppant arm is itself
sufficient to translocate the intermediates into active
sites from a central position [20,21] This finding
sug-gested that, in order to interact with another domain,
the PCP domain needs to translate relative to the
C–AN platform Similar domain translocations have
been suggested for the fatty acid synthase acyl carrier
proteins, which share a similar four a-helix bundle
topology with the PCPs [22] During catalysis, these
ACPs must travel distances of 50–80 A˚ between the
active centers [23–25] Although the conformational
changes of the PCP domains seem convincing and
were postulated in previous studies, to our knowledge
no direct biochemical evidence for these PCP
move-ments in a minimal elongation or initiation module
could be obtained until now
In this work, we provide the first biochemical
evi-dence for conformational changes of the PCP domain
relative to the A domain, which were dependent on the
reaction stage of the latter in an NRPS initiation
mod-ule We have investigated an A–PCP didomain model
construct by partial proteolytic digestion, gel filtration
and native gel electrophoresis, and studied the
accessi-bility of the sulfhydryl group of the Ppant-PCP from
the bulk solvent Taken together, our results support
significant conformational changes in the crosstalk
between A domains and PCP domains that reflect
dif-ferent states of the PCP domain in the catalytic cycle
of an NRPS module
Results
Amino acyl sulfamoyl adenosine inhibitors induce conformational changes in the A-PCP protein that are comparable to the effect of the native substrates
To investigate conformational changes in a catalyti-cally competent NRPS protein, we chose, as a model protein, a truncated construct of gramicidin S synthe-tase I (GrsA; EC 5.1.1.11), which consisted of the first two functional domains, namely the phenylalanine-specific A domain and the PCP domain The terminal
E domain was excised (the exact amino acid composi-tion of the investigated protein is shown in Fig S1B) With this protein, referred to herein as A-PCP, the first two reaction steps shown in Fig 1A can be studied [26] The latter reaction step must involve a productive domain–domain interaction between the
A domain and the PCP domain To trap the holo-enzyme in such a conformation, the natural substrates ATP and phenylalanine are not suitable because their use would lead to the formation of the amino acyl thioester on the Ppant of the PCP domain and prime the enzyme for the next step in the reaction sequence Therefore, we turned to the sulfamoyl-based inhibi-tor 1, which is a nonhydrolyzable analog of the phen-ylalanyl adenylate (see Fig 1B) and probably arrests the enzyme in a state destined for the productive interaction Previous studies determined the Ki values
of this inhibitor class to be in the nanomolar range [27,28]; hence, NRPS A domains bind these cognate inhibitors around two to three orders of magnitude more tightly than their amino acid or ATP substrates [29]
We first aimed to establish that binding of 1 induced similar effects on the conformation of the A domain in solution as the substrates ATP and l-Phe To this end,
we used a partial proteolytic digest, previously reported by Dieckmann et al [30] for the investigation
of the highly homologous protein tyrocidine synthetase
I (TycA) in its apo-form They found that the addition
of the substrates slowed down the proteolysis of the protein, mostly by decreasing the rate of cleavage between the two subdomains of the A domain These findings were later explained with the structural model that the smaller subdomain AC of the A domain rotates upon substrate binding and changes from an open conformation to a more compact one [11,12]
To identify the resulting protein fragments obtained from the partial tryptic digest, we performed in-gel tryptic digests and subsequent MALDI-TOF MS of the major protein bands (see Table S1) Additionally,
Trang 4we prepared a tetramethyl-rhodamine
(TAMRA)-loaded holo-protein through the Sfp-catalyzed reaction
of TAMRA-CoA with apo-A-PCP [31–33] (see the
Materials and methods) Because of the fluorescent
labeling of the PCP domain, the PCP-containing
frag-ments of the tryptic digest (AC-PCP and PCP) can be
visualized under UV illumination (see Fig S3) In
agreement with this previous work [30], we found that
trypsin cleaved the apo-A-PCP construct
predomi-nantly between the two subdomains of the A domain
(see Fig S2) The addition of the substrates ATP and
l-Phe dramatically changed the susceptibility of the
apo-protein to proteolysis by decreasing the rate of
cleavage, indicating that the complex with the amino
acyl adenylate shows less accessibility for the
protease-recognition sites at the solvent-exposed linkers (see
Fig S2) Control reactions revealed that addition of
pyrophosphate, AMP, or ATP alone changed neither
the rate nor the pattern of the proteolysis to a
detect-able extent (data not shown) The latter finding also
indicated that the protein preparations used in this
study were free of residual bound phenylalanine [34]
Addition of l-Phe slowed down the rate of the
proteol-ysis, an effect that was enhanced in the additional
presence of AMP and the nonhydrolyzable ATP
analog adenosine-5¢-[(a,b)-methyleno] triphosphate
(ApCpp) (data not shown)
A comparison of the effect of compound 1 with that
of the natural substrates ATP and l-Phe on the
apo-and holo-forms of the A-PCP construct in the partial
tryptic digest is shown in Fig 2 The apo-form and the
holo-form yielded similar results in this assay, but
sig-nificant differences were observed in the absence of
substrates (Fig 2A), and in the presence of ATP and
l-Phe or compound 1 (Fig 2B & C, respectively) The
addition of 1 resulted in tryptic digest patterns that
were qualitatively similar to those observed for ATP
and l-Phe addition (compare Fig 2B and 2C)
Inter-estingly, however, we had to increase the amount of
trypsin four-fold to observe a reasonable degree of
degradation in the former case These findings
indi-cated, within the resolution of this assay, that 1
induced the same conformational changes as the
sub-strates ATP and l-Phe, and is therefore suitable to
mimic the amino acyl adenylate The significantly
higher resistance to proteolysis of the protein–inhibitor
complex could be a result of the low Ki= 61 nm of
compound 1 [27] that results in a more effective
freez-ing of the conformation of the A domain in a
com-pact, closed state As the inhibitor lacks the b and c
phosphate groups, and release of the pyrophosphate
(PPi) is believed to precede domain alternation from
the adenylation into the thioester conformation, this
closed state is probably the thioester-forming confor-mation [13]
Although similar digest patterns were observed for the apo- and holo-forms of A–PCP we cannot rule out
a potentially different orientation or localization of the PCP domain relative to the A domain from these results The trypsin assay is probably not suitable to resolve such differences because the effect of the modi-fication on the susceptibility of trypsin-cleavage sites between the two domains might be too small Further-more, the fast degradation of the A domain into the two subdomains in the absence of substrates compli-cated quantitative interpretations with regard to the cleavage site(s) between the A domain and the PCP domain, because a PCP-domain fragment can originate from a complete didomain protein or from a previ-ously generated AC-PCP fragment
Evidence for different conformational states of the PCP domain relative to the A domain obtained from native PAGE and gel-filtration analysis
Next, we developed new assays, based on native PAGE and gel-filtration analysis, to monitor larger conformational changes, such as those predicted from the domain-alternation mechanism [12,13] in the A-PCP protein In contrast to the partial tryptic digest, these assays leave the protein intact Native PAGE can resolve different conformational states of a protein if these states exhibit different electrophoretic mobilities and are sufficiently stable under the electrophoretic conditions used Similarly, conformational changes can
be monitored via gel-filtration experiments if the over-all size and shape of the protein changes Before the analysis, we incubated the apo- and the holo-forms of the A-PCP protein with inhibitor 1, or without any ligand As shown in Fig 3, in the absence of ligands the electrophoretic mobility in native PAGE was slightly higher for the holo-form than for the apo-form (left lanes) This difference might reflect minor confor-mational changes but could also be explained by con-sidering the different chemical composition of the proteins Modification with Ppant introduces an extra negative charge into the protein, which should result in
a higher electrophoretic mobility (the calculated charge
of the apo-protein under the conditions of the native PAGE is approximately )20) Gel-filtration experi-ments supported the latter explanation because the apo-protein and the holo-protein eluted within the error margins at identical retention times (see Table 1; see Fig S5 for representative gel-filtration chromato-grams) Importantly, pre-incubation of the proteins
Trang 5with the cognate inhibitor 1 changed the migration
and retention behaviors of the complexes compared
with the free proteins in the native PAGE and in the
gel-filtration assays, respectively The binding of 1 to
the A domain was tight enough to survive both
sepa-ration processes (data not shown) In the native
PAGE, both complexes with the inhibitor, apo- and
holo-, migrated significantly faster than the ligand-free
protein (Fig 3, middle lanes) In the gel-filtration
anal-ysis, the complexes clearly eluted later (see Table 1)
Thus, the binding of 1 seemed to cause a
conforma-tional change leading to a more compact folding of
A-PCP This conformational change probably includes
the closing of the AC subdomain relative to the AN
subdomain, previously suggested in the literature
[12,13] and in agreement with our data from the par-tial proteolytic digests The native PAGE assay and the gel-filtration analysis are thus useful means to monitor these changes with intact proteins in solution Furthermore, a close inspection of the results showed differences between the apo-protein and the holo-pro-tein Interestingly, the presence of 1 led to a larger increase in the electrophoretic mobility of the holo-form compared to the effect seen for the apo-holo-form (Fig 3, compare left and middle lanes) Likewise, in the gel-filtration experiments the elution volume of the holo-protein in the presence of 1 was significantly lar-ger than the corresponding value of the apo-form in the presence of 1 (t-test with a significance level of 5%) The calculated shift differences to higher elution
A
B
C
Fig 2 Partial tryptic digests of apo-A-PCP
and holo-A-PCP under different conditions.
Digests are shown for the apo-A-PCP (left)
and holo-A-PCP (right) (A) Reactions were
performed in the absence of substrates at a
protein ⁄ protease ratio of 250:1 (w ⁄ w), (B)
under saturating conditions (2 m M each) of
ATP and L -Phe at a protein ⁄ protease ratio of
100:1 (w ⁄ w), and (C) in the presence of
inhibitor 1 (100 l M ) at a protein ⁄ protease
ratio of 25 : 1 (w⁄ w).
Trang 6volumes were 0.172 ± 0.034 mL for the apo-A-PCP
and 0.209 ± 0.029 mL for the holo-A-PCP Taken
together, these findings suggest that binding of the
inhibitor to the holo-form caused an additional
confor-mational change that further decreased the Stokes’
radius of the protein compared with the apo-form
This conformational change could be the result of the
PCP adopting a different position relative to the
A domain where the Ppant moiety is positioned to
reach the amino acyl adenylate and renders the overall
structure of the protein more compact This
interpreta-tion is in accordance with the logic of the
nonriboso-mal synthesis that only a Ppant-PCP is a substrate for
an A domain and also with the idea that binding of
the PCP to the A domain in a productive manner
increases the compactness of the protein
Further control reactions with a noncognate
inhibi-tor 2 of the A domain [(5¢-O-[N-(L-prolyl)-sulfamoyl]
adenosine) see Fig 1B] showed that this molecule had
no effect either on the electrophoretic mobilities in
native PAGE (Fig 3, compare left and right lanes) or
on the elution volume in the gel filtration (see Table 1) Furthermore, a similar construct from a truncated proline-activating module, tyrocidin synthe-tase II (TycB1), APro-PCP, was subjected to native PAGE after incubation with 1 or 2 In this case, only inhibitor 2 changed the electrophoretic mobility of this didomain protein (data not shown) Pre-incubation of apo-A-PCP and holo-A-PCP with l-Phe, together with ATP, AMP or ApCpp, did not change the electropho-retic mobilities of the proteins in the native PAGE (data not shown), presumably because the complexes formed are not stable under the electrophoretic condi-tions
Chemical modification reveals different spatial localizations of the Ppant, depending on the reac-tion state of the A domain
The model deduced from the above results suggested that the inhibitor 1 induced a significant conforma-tional change that leads to a compact holo-A-PCP protein Here, the holo-PCP domain interacts in a pro-ductive way with the A domain We decided to further test this model through the use of thiol-modifying agents In the productive conformation the Ppant is in the active site and therefore its thiol-group should be less accessible to the bulk solvent compared with a conformation in which the PCP is not destined to interact with the A domain A less-accessible Ppant moiety should be less prone to chemical modification
by thiol-modifying agents and therefore react more slowly To avoid undesired background labeling of sulfhydryl groups of cysteines, we first eliminated all cysteines in the protein by site-directed mutagenesis The PCP domain in our construct was free of cyste-ines; however, the A domain of GrsA contained four cysteine residues A sequence alignment with related
A domains revealed that two cysteine residues (Cys60 and Cys331) are not conserved in related A domains
We mutated Cys60 to phenylalanine because this is the most prominent residue at this position in the closely related A domains of the tyrocidine and bacitracin NRPS Cys331 is part of the binding pocket of the amino acid in the active site [11] The mutation Cys331Leu decreased the activity of the isolated GrsA
A domain to 26% compared with the wild-type protein [35] We performed the mutation Cys331Ala, which probably does not alter the activity or the specificity dramatically Cys376 is part of the sequence motif A6 and is conserved among NRPS A domains [2] How-ever, mutation of the corresponding cysteine residue to serine in the highly homologous TycA NRPS had no
Table 1 Elution volumes and apparent molecular weights of
different incubated A-PCP constructs in gel-filtration experiments.
Protein Elution volume (mL) m apparent (kDa)
Fig 3 Electrophoretic mobility of A-PCP monitored by native
PAGE A-PCP in the apo-form and in the holo-form was
pre-incu-bated without inhibitor or with compounds 1 and 2 (at 100 l M
each) and then subjected to native PAGE The gel was stained with
Coomassie Brilliant Blue Inh., inhibitor.
Trang 7effect on activity in previous studies [36] Therefore,
we also introduced a serine at this position Finally,
Cys473 is located in the subdomain AC of the
A domain and is moderately conserved Tyrosine and
alanine are the other amino acids frequently found at
this position; therefore, the mutation Cys473Ala was
used
Introduction of the four mutations (C60F, C331A,
C376S and C473A) had little effect on the activity of
the A domain, as determined by the ATP⁄ PPi
exchange assay As shown in Table 2, the Km of the
resulting construct A-PCP(D-4Cys) for l-Phe was
increased by two-fold, while the kcatwas reduced by
1.5-fold Importantly, the mutant protein, A-PCP
(D-4Cys), also showed a tryptic digest pattern
compa-rable to the reference construct A-PCP (data not
shown) and behaved similarly in native PAGE as the
reference construct (see Fig S4) Together, these
results indicated that the four mutations in A-PCP
(D-4Cys) had only a minor effect and thus this protein
was suitable for our studies
The only thiol group in holo-A-PCP(D-4Cys) belongs
to the Ppant moiety Addition of fluorescein-maleimide
and fluorescein-iodacetamide showed fast and
quantita-tive labeling, both when the protein was pre-incubated
with inhibitor 1 as well as in the absence of the small
molecule, indicating that the reactions were too fast to
observe any differences (data not shown) We therefore
tested the chemically less reactive and sterically more
demanding Texas-Red bromoacetamide, which indeed
resulted in differences in labeling velocity (see Fig 4)
The degree of labeling was determined from the
inten-sity of the fluorescent signal on an SDS⁄ PAGE gel MS
analysis confirmed that the chemical labeling took place
at the Ppant moiety (see Fig S6A) The chemical
modi-fication with the fluorophore proceeded most quickly
for holo-A-PCP(D-4Cys), without any ligands, and was
used as a relative reference In striking contrast, in the
presence of 1, the labeling reaction occurred at a
signifi-cantly slower rate (compare lanes 2 and 3 in Fig 4A at
the different reaction time-points and see Fig 4B for
the time-courses of the reactions) Substitution of 1
with substrates ATP and l-Phe led to only a low degree
of labelling, which was consistent with the formation of
the l-Phe-thioester blocking the Ppant group Each of
the two substrates alone decreased the labeling velocity only slightly, with l-Phe having a slightly stronger effect than ATP This finding is in agreement with the observed effect of these substrates in our partial proteolysis experiments (see above) A negative control with the noncognate inhibitor 2 showed that this molecule had no effect In another negative control, incubation of apo-A-PCP(D-4Cys) with Texas-Red bromoacetamide resulted only in the expected back-ground incorporation of the fluorophore, presumably because of minor unspecific reactions of the bromo-acetamide with other residues such as His or Met side chains
We conducted a further control experiment to rule out another possible mechanism of chemical labeling
of the Ppant thiol group Given a potential affinity of the aromatic fluorophore to the ATP-binding pocket,
Table 2 Kinetic parameters of the ATP-PPiexchange reaction for
L -Phe.
A
B
Fig 4 Chemical labeling of apo-A-PCP(D-4Cys) and holo-A-PCP (D-4Cys) with Texas-Red C 5 Bromoacetamide (A) A representa-tive SDS gel of the labeling reaction at different time-points under UV-light (top) and stained with Coomassie Brilliant Blue (below) Lane 1: apo-A-PCP(D-4Cys); lane 2: holo-A-PCP(D-4Cys); lane 3: holo-A-PCP(D-4Cys) + 100 l M 1; lane 4: holo-A-PCP(D-4Cys) + 7.5
m M ATP and L -Phe; lane 5: holo-A-PCP(D-4Cys) + 100 l M 2 (B) Densitometric analysis of band intensities after normalization with the total protein content.
Trang 8as observed for the smaller fluorescein [37], it was
con-ceivable that the alkylation reaction itself took place in
the active site of the A domain (instead of in the freely
accessible solvent) In this case, the effect of inhibitor 1
would only be competitive (displacing the labeling
reagent) and conclusions would be complicated As a
control we therefore performed the labeling assay in
the absence of inhibitor or substrates, but in the
pres-ence of increasing amounts of the free Texas-Red
fluorophore, which should compete with the
Texas-Red bromoacetamide for the binding site and slow
down the modification reaction However, the addition
of up to 16 eq of Texas-Red (compared with the
label-ing reagent) had no influence on the reaction velocity
of the labeling reaction (see Fig S6B), thus excluding
this alternative interpretation
From these experiments it cannot be completely
ruled out that the difference in accessibility of the
Ppant thiol group for the chemical labeling reagent
was a result of alternative pathways (e.g the opening
and closing of protein channels leading to the active
site without the necessity of PCP movement)
How-ever, considering the bulkiness of the labeling reagent,
such an interpretation seems very unlikely Taken
together, these results support the idea that the Ppant
sulfhydryl group is in two distinct locations, which are
dependent on the reaction stage of the A domain
Discussion
The interaction of the PCP with its neighbouring
domains in NRPS systems is crucial in the catalytic
cycle and in the directed product assembly in these
complex biosynthetic machineries How these
interac-tions are controlled has just recently begun to emerge
and the complete picture is still far from being
under-stood Mutational analysis, Ala-scanning mutagenesis
and directed protein evolution determined the residues
participating in the recognition interfaces on the PCP
domain for the interaction with the Ppant transferase,
adjacent TE domain and upstream C domain [38–41]
The same techniques were used to investigate the
rec-ognition interface of PCP domains with in trans-acting
A domains of siderophore-producing NRPS and
revealed a region of about eight amino acids
N-termi-nal to the Ppant attachment site as part of this
interac-tion [39] Interacinterac-tions of the PCP domain with the TE
domain, as well as the trans-acting Ppant transferase
and TEII enzymes, were also studied by NMR
spec-troscopy [8,16–18] These latter studies showed that
the PCP domain is intrinsically mobile and can adopt
multiple conformations, also dependent on the
pres-ence or abspres-ence of its post-translational modification
For the interaction with another domain or external enzyme, one of these pre-existing conformational states
is selected and stabilized It can be assumed that such conformational changes also play an important role in the productive interaction between the A domain and the PCP domain
The recently reported crystal structure of a termina-tion module from the surfactin NRPS, consisting of the domains C-A-PCP-TE, suggested that a movement
of the PCP is required to bridge the 57 A˚ distance between the Ppant attachment site on the PCP and the catalytic centre of the A domain; too far for the 18–
20 A˚ Ppant moiety [19] This structure also revealed a long linker of 15 amino acids, with little secondary structure, between the A domain and the PCP, which probably allows the PCP to travel between different catalytic centres In fact, this linker constituted the only connection between the two domains in the observed structure as there is no common protein– protein interaction surface A potential caveat for the interpretation of these structural findings is that the investigated termination module was crystallized in its inactive apo-form and that it provides only a single snapshot during the catalytic cycle
In this work, we have therefore collected biochemi-cal data from catalytibiochemi-cally competent proteins in solu-tion Our key findings are that (a) in our holo-A-PCP protein the thiol group of the Ppant cofactor can be present in at least two different environments that strongly differ in terms of the accessibility from bulk solvent, and that (b) the switch between these two positions is dependent on the reaction stage of the
A domain In the enzyme primed for amino acyl trans-fer, the Ppant thiol group is more shielded from the solvent Based on these data, we propose the following model Binding of inhibitor 1 induces the thioester conformation of the A domain, and the solvent-protected Ppant thiol points simultaneously into the catalytic centre of the A domain awaiting amino acyl transfer To adopt this conformation, the PCP domain has to dock on the A domain in an orientation whereby the Ppant attachment site faces the channel for this prosthetic group This specific conformation of the two domains is a highly compact form observed for the holo-form in our native PAGE and gel-filtra-tion analyses In contrast, in the absence of inhibitor 1, the Ppant moiety is labeled significantly faster, probably because it is oriented towards the bulk solvent This represents the open conformation of the A domain An open conformation of an A domain was observed in the above-mentioned structure of the C-A-PCP-TE module [19] Here, the residue corre-sponding to the Ppant attachment site points away
Trang 9from the A domain, consistent with the idea that the
Ppant arm should be accessible to bulk solvent
Furthermore, we collected evidence that the apo-PCP
domain does not interact in the same way with an
A domain poised for thioester formation, as if the
bind-ing interface between A and PCP domains is only
cre-ated for the holo-PCP This would be in agreement with
a critical contribution of the Ppant in the interaction of
A and PCP domains, as well as with the model of
differ-ent conformational states of the PCP domain predicting
that only one holo-state can support the binding to the
A domain [17] However, because this interpretation is
based on the difficult-to-resolve differences observed in
assays that only interrogate the globular structure of
the proteins, alternative techniques will be required in
the future to further investigate this point
Taken together, this work presents, to our
knowl-edge, the first biochemical evidence that changing the
reaction stage of the A domain (achieved by binding
of an inhibitor) affects the relative conformation of the
in cis interacting PCP domain It is conceivable that
most of this change is coordinated through a common
movement with the AC subdomain during the closing
of the subdomains However, the flexible linker
between the ACsubdomain and the PCP domain, and
the absence of a contact surface between these two
folded units [19], argue for a certain degree of
flexibil-ity and mobilflexibil-ity of the PCP domain relative to the
A domain Understanding these conformational changes
with higher atomic resolution will require further
structural or spectroscopic studies using catalytically
competent proteins We used a tightly binding
inhibi-tor of the A domain to achieve synchronization of the
protein ensemble This strategy appears to be
promis-ing for capturpromis-ing the proteins at the desired stage in
the reaction cycle Recently, the synthesis of
hydrolyti-cally stable phosphopantetheinyl analogs was reported
[42] These analogs might prove useful in fixing the
next reaction stage in the catalytic cycle of an
initia-tion or elongation module (i.e the interactions
between the PCP and condensation or modifying
domains) in multidomain NRPS enzymes
Materials and methods
General
Standard procedures were applied for PCR amplification,
purification of DNA fragments and cloning of recombinant
DNA [43] Oligonucleotides were from Operon (Cologne,
Germany) Unless otherwise stated, chemicals were
pur-chased from Applichem GmbH (Darmstadt, Germany) and
Roth (Karlsruhe, Germany) Inhibitors 1
(5¢-O-[N-(l-phenyl)-sulfamoyl] adenosine) and 2 (5¢-O-[N-(l-prolyl)-(5¢-O-[N-(l-phenyl)-sulfamoyl] adenosine) were kind gifts of M Hahn and M Marahiel [27]
Plasmid construction The gene fragment encoding GrsA A-PCP was PCR ampli-fied from the genomic DNA of Bacillus brevis ATCC 9999 using the primers P1 (5¢- tatccatggtaaacagttctaaaagtatattg) and P2 (5¢- tatagatctctcacttcttcttttactatc) The PCR product was subcloned into a pQE60 vector using NcoI and BglII sites to introduce a C-terminal His6-Tag The NcoI–HindIII fragment of this vector was then ligated into pET16b, result-ing in vector pJZ06 Site-directed mutagenesis was performed according to the Quick change protocol (Stratagene, La Jolla, CA, USA) Plasmid pJZ06 served as a template for two successive point mutations C60F and C473A were intro-duced using primers P4 (5¢-atgtagccattgtatttgaaaatgagcaact) and P5 (5¢- agttgctcattttcaaatacaatggctacat), and P6 (5¢- gaacagc cgtatttggccgcttattttgtatc) and P7 (5¢- gatacaaaataagcggccaaa tacggctgttc), respectively, to give pJZ12 In the same manner, the plasmid pJZ11, encoding the mutations C331A and C376S, was generated using primers P8 (5¢-ccctacggaaacaac gatcgctgcgactacatgggta) and P9 (5¢-tacccatgtagtcgcagcgatcg ttgtttccgtaggg), and P10 (5¢-tgaagctggtgaattatcgattggtggagaa ggg) and P11 (5¢-cccttctccaccaatcgataattcaccagcttca), respec-tively In order to combine all four mutations, an NdeI fragment containing the two mutations was excised from pJZ11 and ligated into pJZ12 to replace the corresponding NdeI fragment The resulting plasmid, pJZ13, encoded the A-PCP(D-4Cys) construct The correctness of the muta-tions in pJZ13 was confirmed by DNA sequencing (GATC, Konstanz, Germany) of the entire insert
Protein overproduction Escherichia coli BL21 (DE3) cells were transformed with the expression plasmids described above The expression and purification of the C-terminally His6-tagged apo-pro-teins were conducted as previously described [44] As judged by SDS⁄ PAGE, the proteins could be purified to apparent homogeneity by a single Ni2+-affinity chromatog-raphy step Fractions containing the recombinant proteins were pooled and dialysed against assay buffer [50 mm HE-PES (pH 8.0), 100 mm NaCl, 1 mm EDTA, 10 mm MgCl2] After the addition of 10% glycerine, the proteins were shock-frozen in liquid nitrogen and stored at )80 C The protein concentrations were determined using the calculated extinction coefficient at 280 nm
Synthesis of TAMRA-CoA The modification of CoA was carried out in accordance with a previously reported protocol [45] In short, 17.3 mg
Trang 10CoA trilithium salt dihydrate (21 lmol, 2.1 eq.) was
dis-solved in 2 mL of a 100 mm phosphate buffer (pH 7.0)
Five milligrams of tetramethylrhodamine-5-maleimide
(10 lmol, 1 eq.; purchased from Anaspec Inc., Fremont,
CA, USA) was dissolved in 800 lL of dimethylsulfoxide
The solutions were combined and the reaction mixture was
agitated at room temperature for 1 h in the dark followed
by purification with preparative HPLC on a reverse-phase
C18 column with a gradient of 5–50% acetonitrile in
0.05% trifluoracetic acid⁄ water over 30 min The purified
compound was lyophilized, and the identity was confirmed
by MALDI-TOF MS (negative mode): [M-H]) calculated
1247.3 gÆmol)1, observed 1247.5 gÆmol)1
Post-translational modification of the enzymes
Conversion of the apo-enzymes into the holo-enzyme or the
Ppant-TAMRA modified form was carried out in vitro by
adding 40 eq of CoA or 5 eq of TAMRA-CoA in the
pres-ence of 10 mm MgCl2 and 0.02 eq of the Bacillus subtilis
Ppant-transferase Sfp [46] overnight at 4C The excess of
CoA⁄ TAMRA-CoA was removed through dialysis
ATP-PPiexchange reaction
The ATP-PPiexchange assay was used to confirm the
ade-nylation domain activity [35] Reaction mixtures (final
vol-ume 100 lL) contained 50 mm HEPES (pH 8.0), 100 mm
NaCl, 1 mm EDTA, 5 mm MgCl2, 400 nm apo-enzyme and
1 lm–10 mm l-Phe After 10 min of incubation at 37C,
the reaction was started by the addition of 5 mm ATP,
25 lm Na4P2O7 and 0.015 lm Ci [32P-Na4P2O7] (Perkin
Elmer, Boston) and incubated at 37C for 45 s Reactions
were quenched by adding 0.5 mL of a stop mix [1.2%
(w⁄ v) activated charcoal, 0.1 m Na4P2O7 and 0.35 m
per-chloric acid] Subsequently, the charcoal was pelleted by
centrifugation, washed twice with 1 mL of water and
resus-pended in 0.5 mL of water After the addition of 3.5 mL of
liquid scintillation fluid (Roth, Karlsruhe, Germany), the
charcoal-bound radioactivity was determined by liquid
scin-tillation counting using a 1900CA TriCarb liquid
scintilla-tion analyzer (Packard, Meriden, CT, USA) The measured
values were corrected using the value of the negative
con-trol (without amino acid) and the maximal counts of the
radioactive pyrophosphate used Assuming that the amount
of radioactive PPicould be neglected compared with
nonra-dioactive PPi, the initial velocities of the reactions were
cal-culated The obtained values were analysed using a
Michaelis–Menten approach
Partial tryptic digest
The proteolysis reactions of the apo- and the holo-NRPS
proteins (6–12 lm) were performed in assay buffer at 37C
following a pre-incubation step of 10 min with substrates (ATP and l-Phe at a final concentration of 1 mm) or inhib-itors (final concentration 100 lm) The digest was started with the addition of a trypsin solution (0.08 lgÆlL)1 of modified trypsin; Promega, Madison, WI, USA) at a final protease⁄ protein ratio of 1 : 250 (w ⁄ w) Aliquots were with-drawn at different time-points and the digest was stopped
by the addition of SDS loading buffer To yield a reason-able digest in the presence of the cognate inhibitor 1, the protease⁄ protein ratio was raised to 1 : 25 (w ⁄ w) A control experiment using the unrelated protein, Sfp, for digestion with trypsin in the absence or presence of 1 ruled out that trypsin itself might be inhibited by 1 (data not shown)
Native PAGE Discontinuous native gel electrophoresis was performed similarly to the standard Laemmli SDS⁄ PAGE protocol, only without SDS [47] A 5% stacking gel and an 8% sepa-ration gel were used Before the loading buffer was added, proteins were pre-incubated for 15 min at 37C with or without substrates⁄ inhibitors The samples were not boiled before loading on the gel
Gel-filtration experiments
A Superdex 200 10⁄ 300 GL column (GE Healthcare, Chal-font St Giles, UK) was equilibrated with assay buffer [50 mm HEPES (pH 8.0), 100 mm NaCl, 1 mm EDTA,
2 mm dithiothreitol, 10 mm MgCl2] Following pre-incuba-tion with or without inhibitors for 10 min at 37C, 200 lL
of a 20 lm protein solution was applied onto the column and the absorption at 280 nm was recorded Column cali-bration was performed using the Gel Filtration Calicali-bration Kit – Low Molecular Weight (GE Healthcare)
Chemical labeling of holo-A-PCP(D-4Cys)
A 7.5-lm enzyme solution was mixed in assay buffer [50 mm HEPES (pH 7.0), 100 mm NaCl, 1 mm EDTA, 10 mm MgCl2] with 2 mm tris(2-carboxyethyl) phosphine (TCEP) and either 1 mm of substrates (ATP and⁄ or l-Phe) or
100 lm of the different inhibitors This mixture was pre-incu-bated for 10 min at 37C, and then incubated for 10 min at
25C The reaction was started through the addition of 8 eq (compared to the enzyme) of Texas-Red C5Bromoacetamide (purchased from Invitrogen, Carlsbad, CA, USA; 1 mm stock solution in dimethylsulfoxide) and conducted at 25C Aliquots were withdrawn at various time-points and the reaction was stopped with SDS⁄ PAGE loading buffer containing 20% (v⁄ v) mercaptoethanol The amount of incorporated fluorophore was visualized by UV-illumination
of the resulting SDS gel and analysed densitometrically using the program scion image (http://www.scioncorp.com) The