*For correspondence:
ngsharaf@stanford.edu (NGS);
dcrees@caltech.edu (DCR)
Present address:
†
Department
of Biology, Stanford University,
Stanford, United States
Competing interests:
The
authors declare that no
competing interests exist.
Funding:
See page 17
Received:
24 April 2021
Preprinted:
04 May 2021
Accepted:
18 August 2021
Published:
19 August 2021
Reviewing editor:
Janice L
Robertson, Washington
University in St Louis, United
States
Copyright Sharaf et al. This
article is distributed under the
terms of the
Creative Commons
Attribution License,
which
permits unrestricted use and
redistribution provided that the
original author and source are
credited.
Characterization of the ABC methionine
transporter from
Neisseria meningitidis
reveals that lipidated MetQ is required
for interaction
Naima G Sharaf
1,2†
*, Mona Shahgholi
1
, Esther Kim
1
, Jeffrey Y Lai
1,2
,
David G VanderVelde
1
, Allen T Lee
1,2
, Douglas C Rees
1,2
*
1
Division of Chemistry and Chemical Engineering, California Institute of Technology,
Pasadena, United States;
2
Howard Hughes Medical Institute, California Institute of
Technology, Pasadena, United States
Abstract
NmMetQ is a substrate-binding protein (SBP) from
Neisseria meningitidis
that has
been identified as a surface-exposed candidate antigen for meningococcal vaccines. However, this
location for NmMetQ challenges the prevailing view that SBPs in Gram-negative bacteria are
localized to the periplasmic space to promote interaction with their cognate ABC transporter
embedded in the bacterial inner membrane. To elucidate the roles of NmMetQ, we characterized
NmMetQ with and without its cognate ABC transporter (NmMetNI). Here, we show that NmMetQ
is a lipoprotein (lipo-NmMetQ) that binds multiple methionine analogs and stimulates the ATPase
activity of NmMetNI. Using single-particle electron cryo-microscopy, we determined the structures
of NmMetNI in the presence and absence of lipo-NmMetQ. Based on our data, we propose that
NmMetQ tethers to membranes via a lipid anchor and has dual function and localization, playing a
role in NmMetNI-mediated transport at the inner membrane and moonlighting on the bacterial
surface.
Introduction
The substrate-binding protein (SBP) NmMetQ from the human pathogen
Neisseria meningitidis
has
been identified as a surface-exposed candidate antigen for the meningococcal vaccine (
Pizza et al.,
2000
). Subsequently, NmMetQ has been shown to interact with human brain microvascular endothe-
lial cells (
Ka
́
nova
́
et al., 2018
), potentially acting as an adhesin. However, the surface localization of
NmMetQ challenges the prevailing view that SBPs reside in the periplasm of Gram-negative bacteria
(
Thomas and Tampe
́
, 2020
), binding and delivering molecules to cognate ATP-Binding Cassette
(ABC) transporters in the inner membrane (IM). Several questions arise from these studies: Has
NmMetQ lost its ABC transporter-dependent function in the IM? How does NmMetQ become
embedded in the outer membrane (OM) surface of the bacterium?
The ABC transporter-dependent role of SBPs has been well characterized for multiple ABC trans-
porter systems (
Hollenstein et al., 2007
;
Oldham et al., 2013
;
Sabrialabed et al., 2020
;
Liu et al.,
2020
;
Nguyen et al., 2018
;
de Boer et al., 2019
). These studies reveal conserved SBP-dependent
characteristics, including that the SBP is largely responsible for substrate delivery to the ABC trans-
porter, with concomitant stimulation of the transport coupled ATPase activity. Structural studies
have shown that SBPs dock to the periplasmic surface of the transporter’s transmembrane domains,
with the substrate-binding pocket juxtaposed with the translocation pathway of the transporter.
While many SBPs have only been assigned ABC transporter-dependent functions, a few SBPs have
also been shown to have both ABC transporter-dependent and ABC transporter-independent
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
1 of 21
RESEARCH ARTICLE
functions (often referred to as moonlighting functions) (
Adler, 1975
). For example, the
E. coli
malt-
ose SBP (MBP) binds and stimulates its cognate ABC transporter during the maltose import cycle
(
Davidson et al., 1992
). In addition, the MBP-maltose complex is also a ligand for the chemotaxis
receptor, triggering the signaling cascade involved in nutrient acquisition (
Hazelbauer, 1975
;
Manson et al., 1985
). Other SBPs have also been assigned ABC transporter-independent functions
(
Mu
̈
ller et al., 2007
;
Castan
̃
eda-Rolda
́
n et al., 2006
;
Matthysse et al., 1996
), including NspS from
Vibrio cholerae
, which has been shown to play a role in biofilm formation (
Young et al., 2021
) and
not transport (
Cockerell et al., 2014
). Additionally, two MetQ proteins,
N. gonorrhoeae
(Ng)
NgMetQ and
Vibrio vulnificus
(Vv) VvMetQ have also been identified as putative adhesins, mediating
bacterial adhesion to human cervical epithelial cells (
Semchenko et al., 2017
) and to human intesti-
nal epithelial cells (
Lee et al., 2010
;
Yu et al., 2011
), respectively. Evidence that these MetQ SBPs
bind and stimulate their cognate ABC transporters, however, is lacking. Whether NmMetQ has lost
its ATP transporter-dependent function or whether it plays roles at both the IM and OM cannot be
determined through amino acid sequence alone and must be experimentally verified.
Since SBPs are not membrane proteins, the detection of NmMetQ at the cell surface of the bacte-
rium suggests it must be tethered to the OM. In Gram-negative bacteria, the paradigm that SBPs
translocate into the periplasm where they diffuse freely between the IM and OM can be traced back
to early experiments by Heppel showing that the osmotic shock of Gram-negative bacteria leads to
the release of SBPs (
Heppel, 1969
). While many SBPs in Gram-negative bacteria have been identi-
fied as secreted proteins (
Willis and Furlong, 1974
;
Ahlem et al., 1982
), several studies have also
identified a few lipid-modified SBPs (lipo-SBP) (
Tokuda et al., 2007
). However, the presence of lipo-
SBPs in Gram-negative bacteria has not been generally appreciated and the role that lipid modifica-
tions play in SBP surface localization remains unexplored.
Although ABC transporter-dependent functions of NmMetQ, VvMetQ, and NgMetQ are not well
studied, the homologous SBP from
E. coli
, EcMetQ, is well characterized. Studies show that the
E.
coli
methionine uptake system consists of EcMetQ and its cognate ABC transporter EcMetNI (
Kad-
ner, 1974
;
Kadner, 1977
). Structures of both EcMetQ and EcMetNI alone and in complex are avail-
able (
Kadaba et al., 2008
;
Johnson et al., 2012
;
Nguyen et al., 2015
). EcMetNI comprises two
transmembrane domains (TMD), which form a substrate translocation pathway, and two nucleotide-
binding domains (NBD), which couple transport to the binding and hydrolysis of ATP. In the absence
of EcMetQ, EcMetNI adopts the inward-facing conformation, with the TMDs open to the cytoplasm
and NBDs separated. The available crystal structures of EcMetQ reveal two domains connected by a
linker that form the methionine-binding pocket (
Nguyen et al., 2015
). Of note, EcMetQ has been
experimentally verified to be a lipoprotein by radioactive palmitate labeling (
Tokuda et al., 2007
).
Additionally, Carlson et al. found that wild-type EcMetQ remains associated with recombinantly
expressed his-tagged EcMetNI when solubilized in detergent and peptidiscs but not when its N-ter-
minal cysteine is mutated to prevent lipidation (
Carlson et al., 2019
). This study shows that EcMetQ
association with EcMetNI depends on its N-terminal lipid. Structures of EcMetQ are also available,
however, the lipid modification is not present in EcMetQ structures. A structure of the EcMetQ:
EcMetNI complex is also available and shows EcMetNI in the outward-facing conformation, with the
TMDs and NBDs close together. In this structure, EcMetQ is docked to the periplasmic surface of
the TMDs with the binding pocket open to the central cavity (
Nguyen et al., 2018
). These struc-
tures, together with in vivo functional assays (
Nguyen et al., 2018
;
Kadner, 1974
), show that
EcMetQ is intimately involved in EcMetNI-mediated methionine transport.
Although the interaction between EcMetQ and EcMetNI is well characterized, less is known about
the corresponding system in
Neisseria meningitidis
. To date, there have been no biochemical or
structural studies reported for NmMetNI. Recently determined structures of NmMetQ are in the
ligand-free, L-methionine-, or D-methionine-bound states, and binding assays show L-methionine
binds NmMetQ with greater affinity than D-methionine (
Nguyen et al., 2019
). These studies were
carried out with an NmMetQ protein that lacks the native N-terminal signal sequence, establishing
that the N-terminal signal sequence is not necessary for ligand binding. However, NmMetQ is pre-
dicted to be lipoprotein based on the N-terminal protein sequence (Uniprot entry Q7DD63)
(
UniProt Consortium, 2019
). Experimental evidence confirming this modification, however, has not
been reported. Thus, a full understanding of the post-translational modification of NmMetQ and its
interactions with NmMetNI are lacking. To better understand NmMetQ and the role it plays in
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
2 of 21
Research article
Biochemistry and Chemical Biology
Structural Biology and Molecular Biophysics
methionine transport, a detailed characterization of both NmMetNI and NmMetQ with its native
N-terminal signal sequence is required.
In this work, we characterized NmMetQ and NmMetNI using multiple biophysical methods. Using
mass spectrometry and site-directed mutagenesis, we demonstrate that full-length NmMetQ,
recombinantly-expressed in
E. coli
, is a lipoprotein (lipo-NmMetQ). Functional assays showed that
both lipo-NmMetQ and L-methionine are required for maximal stimulation of NmMetNI ATPase
activity. NmMetNI was also stimulated to a lesser extent by pre-protein NmMetQ (a variant with an
unprocessed N-terminal signal peptide) with L-methionine and by lipo-NmMetQ with select methio-
nine analogs. We also determined the structures of NmMetNI in the presence and absence of lipo-
NmMetQ to 6.4 A
̊
and 3.3 A
̊
resolution, respectively, using single-particle electron cryo-microscopy
(cryo-EM). Using a bioinformatics approach, we also identified MetQ proteins from other Gram-neg-
ative bacteria that are predicted to be modified with lipids. This analysis suggests that the lipid mod-
ification of MetQ proteins is not restricted to
N. meningitidis
and
E. coli
.
Based on our data, we propose that lipo-NmMetQ, and more generally lipo-MetQ proteins in
other Gram-negative bacteria, possesses dual function and localization: ABC transporter-dependent
roles at the IM and a moonlighting ABC transporter-independent role (or roles) at the OM. Our find-
ings highlight the complexity of the cell envelope and that much remains to be understood about
the rules governing protein localization in Gram-negative bacteria and the moonlighting functions of
SBPs on the surface of the cell.
Results
N. meningitidis
MetQ is a lipoprotein
While lipoproteins and secreted proteins both must traverse the inner cell membrane during biogen-
esis, their maturation occurs through different mechanisms depending on the N-terminal signal
sequence (
Figure 1A
). Lipoproteins are synthesized in the cytoplasm as pre-prolipoproteins, inserted
into the IM, and then anchored via their N-terminal signal sequence to the cytoplasmic membrane
(
Okuda and Tokuda, 2011
). While tethered to the IM through the signal sequence, pre-prolipopro-
teins are subsequently modified by three enzymes: (1) phosphatidylglycerol transferase (Lgt), which
transfers the diacylglycerol group preferentially from phosphatidylglycerol (PG) to the cysteine resi-
due via a thioester bond of the pre-prolipoprotein, producing a prolipoprotein (
Mao et al., 2016
)
(2) signal peptidase II (LspA), which cleaves the prolipoprotein N-terminal signal sequence to yield a
diacylated lipoprotein with the N-terminal cysteine
Hussain et al., 1982
;
Vogeley et al., 2016
; and
(3) apolipoprotein N-acyl transferase (Lnt), which N-acylates the cysteine residue preferentially using
an acyl group of phosphatidylethanolamine (PE) to produce a triacylated lipoprotein. (
Noland et al.,
2017
;
Wiktor et al., 2017
). Similar to lipoproteins, secreted proteins are synthesized in the cyto-
plasm as pre-proteins with an N-terminal signal sequence. These pre-proteins serve as substrates for
signal peptidase I (Spase I), which cleaves the N-terminal signal sequence to yield the mature
secreted protein (
Karla et al., 2005
;
Paetzel et al., 1998
).
NmMetQ is predicted to be a lipoprotein by SignalP 5.0, a deep neural network algorithm that
analyzes amino acid sequences to predict the presence and location of cleavage sites
(
Almagro Armenteros et al., 2019
). To validate this prediction, we expressed NmMetQ using an
E.
coli
expression system with the native N-terminal signal sequence and a C-terminal decahistidine
tag.
E. coli
has been previously used to produce lipid-modified
N. meningitidis
proteins
(
Fantappie
`
et al., 2017
). We purified NmMetQ in the detergent n-dodecyl-
b
-D-maltopyranoside
(DDM) using an immobilized nickel affinity column followed by size-exclusion chromatography (SEC).
The SEC elution profile shows one main peak with an elution volume of 66 mL (
Figure 1A
). An analy-
sis of the peak fraction by liquid chromatography mass spectrometry (LC/MS) revealed two major
deconvoluted masses of 31,662 and 31,682 Da (
Figure 1B
). These masses correspond well with the
theoretical masses of two lipoprotein NmMetQ proteins: one with a triacyl chain composition of
16:0, 16:0 and 16:0 (31,661 Da) and another with a triacyl chain composition of 16:0, 16:0 and 18:1
(31,685 Da), respectively (
Figure 1A
, top). We calculated the intact masses of the lipo-NmMetQ pro-
teins using a combination of 16:0 and 18:1 acyl chains because these were the major species found
in previous studies of recombinantly expressed lipoproteins (
Hantke and Braun, 1973
;
Luo et al.,
2016
).
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
3 of 21
Research article
Biochemistry and Chemical Biology
Structural Biology and Molecular Biophysics
To confirm that lipid attachment site occurs at the N-terminal Cys 20 on NmMetQ, we generated
a Cys-to-Ala NmMetQ mutant (NmMetQC20A). We hypothesized that this mutation would prevent
lipid attachment and lead to the accumulation of pre-protein NmMetQ, containing an unprocessed
N-terminal signal sequence and the C20A mutation. The NmMetQC20A protein was expressed and
purified in DDM as previously described. The SEC elution profile reveals two major peaks with dis-
tinct elution volumes, 78 ml and 100 mL for peak 1 and 2, respectively (
Figure 1C
). For peak 1, anal-
ysis of the fraction containing the highest peak revealed a deconvoluted mass of 32,804, which
correlates well with the theoretical intact mass of the pre-protein NmMetQ (32,802 Da). For peak 2,
the deconvoluted mass was 30,840, which agrees with the theoretical intact mass of a secreted
Figure 1.
Mass spectrometry (MS) analysis of lipo-NmMetQ and NmMetQC20A proteins. (
A
) (Top) Schematic of lipoprotein maturation pathway. Inset
contains a schematic of a lipoprotein with acyl chain composition [16:0,16:0,16:0]. Acyl chains are grouped in a dotted line box and their average
masses are calculated. Below the schematic are the theoretical masses for the lipo-NmMetQ proteins (in italics) assuming triacylation occurs via the
canonical lipoprotein maturation pathway due to the sequential action of three enzymes (Lgt, LspA, and Lnt). The numbers in the brackets correspond
to the total number of carbons and double bonds, respectively, present in the fatty acyl chains of the lipid. (Bottom) Schematic of various
NmMetQC20A proteins with example theoretical average masses, shown in italics, assuming cleavage occurs between A19 and A20, possibly by signal
peptidase I (SPase I). N-terminal signal peptides are represented by a green rectangle. (
B
) Characterization of lipo-NmMetQ. Size-exclusion
chromatogram and mass spectra of peak 1. The molecular masses of the major species correspond within 1 Da to the predicted mass for two
triacylated NmMetQ species, one with acyl chain composition [16:0, 16:0, 16:0] (31,661 Da) and the other with [16:0, 16:0, 18:1] (31,685 Da). (
C
)
Characterization of NmMetQC20A. Size-exclusion chromatogram and mass spectra of the major species from peak 2 and peak 3. The molecular masses
of the major species of peak 2 and 3 correspond to the pre-protein NmMetQ (32,802 Da) and secreted NmMetQ (30,839 Da), respectively. These
measured masses are within 3 Da of the predicted masses for each species. Assigned NmMetQ species are depicted in cartoon form on the
chromatograms.
The online version of this article includes the following figure supplement(s) for figure 1:
Figure supplement 1.
DLS and SEC-MALS measurements of NmMetQ proteins.
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
4 of 21
Research article
Biochemistry and Chemical Biology
Structural Biology and Molecular Biophysics
NmMetQ protein cleaved between Ala 19 and Ala 20 (30,839 Da) (
Figure 1A
, bottom). The produc-
tion of the secreted NmMetQ was surprising since we only expected the accumulation of the pre-
protein NmMetQ. However, these data suggest that the Cys-to-Ala mutation created a noncanonical
cleavage site, possibly allowing Spase I to inefficiently cleave the pre-protein to yield secreted
NmMetQ. Together, these data clearly demonstrate that the major species of recombinantly-
expressed NmMetQ is heterogeneously triacylated at Cys 20. Mutating Cys 20 to Ala prevents the
production of lipoprotein NmMetQ, leading to the formation of pre-protein NmMetQ and secreted
NmMetQ. The location of cleavage site, position of lipid attachment, and heterogeneous triacyl
chain composition of NmMetQ in this study are consistent with previous studies characterizing other
lipoproteins produced in
E. coli
(
Luo et al., 2016
;
Kwok et al., 2011
).
These data also reveal an interesting property of each DDM-solubilized NmMetQ variant: lipo-
NmMetQ, pre-protein lipo-NmMetQ, and secreted NmMetQ proteins elute at different volumes
despite their similar molecular masses (between 31 and 33 kDa). Specifically, lipo-NmMetQ and pre-
protein NmMetQ proteins elute at much higher apparent mass than secreted NmMetQ on a HiLoad
16/600 Superdex 200 (GE healthcare) column (
Figure 1B,C
). To further investigate the properties of
the NmMetQ proteins, we used dynamic light scattering (DLS) to measure their hydrodynamic radii
(R
h
) and calculate their theoretical molecular weights assuming a folded globular protein. We found
that the R
h
values and molecular weight estimates were larger for lipo-NmMetQ (R
h
= 7.9
±
0.2 nm,
Mw-R = 430
±
20 kDa) and pre-protein NmMetQ (R
h
7.7
±
0.06 nm, Mw-R = 400
±
7 kDa) than for
secreted NmMetQ (R
h
3.0
±
0.013 nm, Mw-R = 43.0
±
0.3 kDa) (
Figure 1—figure supplement 1
).
These proteins were also analyzed using Size Exclusion Chromatography with Multi-Angle Light Scat-
tering (SEC-MALS). For both lipo-NmMetQ and pre-protein NmMetQ, estimated molar masses were
lower with SEC-MALS when compared to DLS, with lipo-NmMetQ measurements of 111
±
0.3 versus
430
±
20 kDa and pre-protein NmMetQ measurements of 105
±
0.3 versus 400
±
7 kDa, respectively.
These data suggest that both lipo-NmMetQ and pre-protein NmMetQ aggregate and that the mass
of the aggregate depends on the precise condition of the experiment. However, molar masses for
the secreted NmMetQ protein were more similar (26
±
1 versus 43.0
±
0.3 kDa), suggesting that
unlike lipo-NmMetQ and pre-protein NmMetQ, secreted NmMetQ does not associate with DDM
micelles. Based on the size-exclusion chromatograms, DLS, and SEC-MALS data, we propose that
both lipo-NmMetQ and pre-protein NmMetQ aggregate with DDM to form protein-DDM micelle-
like complexes.
The ATPase activity of NmMetNI is maximally stimulated in the
presence of both lipo-NmMetQ and L-methionine
Figure 2A
shows that in the presence of 1
m
M NmMetNI alone (black trace) and in the presence of
50
m
M L-methionine (blue trace), the ATPase activity was low, demonstrating that L-methionine
alone is not sufficient to stimulate NmMetNI ATPase activity. However, in the presence of both 1
m
M
lipo-NmMetQ and 50
m
M L-methionine, a marked stimulation of ATPase activity was observed
(
Figure 2A
, green trace). To exclude the possibility that the stimulation of ATPase activity is medi-
ated by either the lipid-moiety or the unliganded NmMetQ protein subunit, the experiment was
repeated in the absence of L-methionine (NmMetNI and unliganded lipo-NmMetQ only) (
Figure 2A
,
magenta trace). Under these conditions the ATPase activity is low, showing that unliganded lipo-
NmMetQ is not sufficient to stimulate NmMetNI activity. Given these findings, we conclude that
NmMetNI ATPase activity is tightly coupled, requiring both L-methionine and lipo-NmMetQ for max-
imum stimulation. This result strongly suggests that lipo-NmMetQ plays a role in methionine-medi-
ated NmMetNI ATP hydrolysis.
Next, we characterized the effect of different NmMetQ proteins (lipo-NmMetQ, pre-protein
NmMetQ, and secreted NmMetQ) on the ATPase activity of NmMetNI.
Figure 2B
demonstrates
that in the presence of 50
m
M L-methionine, the NmMetNI ATPase activity increases with increasing
concentration of lipo-NmMetQ up to 2
m
M, after which the activity starts to plateau (green trace).
The same protocol was performed with pre-protein NmMetQ, which contains an N-terminal signal
sequence but without the lipid modification. Addition of pre-protein NmMetQ also led to stimula-
tion of ATPase activity, although to a lesser extent than observed for lipo-NmMetQ (orange trace).
Addition of secreted NmMetQ, however, had little effect on the ATPase activity (cyan). Together,
these data establish that the lipid moiety of lipo-NmMetQ is required for maximal NmMetNI
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
5 of 21
Research article
Biochemistry and Chemical Biology
Structural Biology and Molecular Biophysics
stimulation, although the N-terminal signal sequence of pre-protein NmMetQ could partially mimic
its stimulatory effect.
A comparison of NmMetNI’s ATPase activity with that of the previously characterized EcMetNI
reveals that these transporters have different ligand-dependent ATPase activities. When L-methio-
nine and SBP are absent, NmMetNI has no detectable basal ATP activity. However EcMetNI has a
basal ATPase rate of 300 nmol Pi/min/mg (
Kadaba et al., 2008
). These transporters also differ in
their response to L-methionine. In the presence of L-methionine, the ATPase activity of EcMetNI
decreases due to the binding of L-methionine to the C2 domain, which is responsible for the regula-
tory phenomenon of transinhibition. For NmMetNI, however, no such effect was detected, as antici-
pated from the absence of the C2 autoinhibitory domain in NmMetNI.
A comparison of NmMetNI SBP-dependent ATPase stimulation to other ABC importers also
reveals some similarities and differences. For NmMetNI, only liganded-SBP maximally stimulated
NmMetNI ATP hydrolysis. Maximal stimulation by liganded-SBPs is also a mechanistic feature shared
by the ABC importers EcMalFGK
2
(
Davidson et al., 1992
) and EcHisQMP
2
(
Ames et al., 1996
). In
contrast, for the ABC importer EcYecSC-FliY (
Sabrialabed et al., 2020
), full stimulation of ATPase
can be achieved in both the liganded-SBP and the unliganded-SBP. Although the origin of these dif-
ferences are unclear, our data show that NmMetNI is tightly coupled and highlight the mechanistic
differences between ABC importers.
N-formyl-L-methionine, L-norleucine, L-ethionine, and L-methionine
sulfoximine are potential substrates for the lipo-NmMetQ:NmMetNI
system
To identify potential substrates of the NmMetQ-lipoprotein MetQ system, we determined the rela-
tive binding affinities of several methionine analogs to NmMetQ. For these measurements, we used
Fluorine chemical shift Anisotropy and eXchange for Screening (FAXS) in competition mode, a pow-
erful solution NMR experiment that monitors the displacement of a fluorine-containing reporter mol-
ecule by a competing ligand. An important feature of FAXS is that fluorine modification of the
competing ligand is not required (
Dalvit et al., 2003
;
Dalvit and Vulpetti, 2019
). As previously dis-
cussed (
Gerig, 1994
;
Dalvit and Vulpetti, 2019
), the fluorine nucleus has several properties that are
advantageous for NMR:
19
F is 100% abundant, possesses a spin 1/2 nucleus, and has high
Figure 2.
ATP hydrolysis of NmMetNI in the presence and absence of L-methionine and NmMetQ proteins. (
A
) ATP hydrolysis was measured in the
presence of 1
m
M of DDM-solubilized NmMetNI alone (black trace), 50
m
M L-methionine (blue trace), 1
m
M lipo-NmMetQ (magenta trace) and both 50
m
M L-methionine and 1
m
M lipo-NmMetQ (green trace). Insert shows representative measurements of absorbance versus time (black dots) and the
linear fits (green lines) for NmMetNI ATPase activity in the presence of lipo-NmMetQ and L-methionine at increasing ATP concentrations (0.2, 0.4, 0.8,
1.2, 1.6, 2.0, and 4.0 mM) (
B
) Specific activity of NmMetNI with increasing concentrations of various NmMetQ proteins: lipo-NmMetQ (green trace), pre-
protein NmMetQ (orange trace), and secreted NmMetQ (cyan trace) with 50
m
M L-methionine. Vmax values were determined by fitting the Michaelis-
Menten equation to a plot of ATPase activity versus ATP concentration (0.2, 0.4, 0.8, 1.2, 1.6, 2.0, and 4.0 mM) at different MetQ protein concentrations
( 0.5, 1, 2, 4, 5, 8
m
M). N=3 error bars represent standard error of the mean (SEM). These data show the NmMetNI ATPase activity is tightly coupled,
requiring both L-methionine and lipo-NmMetQ for maximal NmMetNI ATPase stimulation.
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
6 of 21
Research article
Biochemistry and Chemical Biology
Structural Biology and Molecular Biophysics
gyromagnetic ratio, which results in high sensitivity (83% of
1
H). It also has a large chemical shift
anisotropy (CSA), allowing higher responsiveness to change in molecular weight, such as those that
occur during a protein-ligand binding event. Additionally, since fluorine atoms are not present in
most commonly used buffer systems and virtually absent from all naturally occurring biomolecules,
background interference in fluorine NMR experiments is minimal.
To optimize the FAXS experiment, we considered several factors. As shown in
Figure 1B
, lipopro-
tein NmMetQ may multimerize, possibly through an association with the hydrophobic acyl chains,
increasing its apparent molecular weight. Because FAXS is sensitive to the apparent molecular
weight of the protein, we chose to use a NmMetQ construct lacking its native N-terminal signal
sequence and is therefore not modified with lipids (referred to here as NLM-NmMetQ). Trifluoro-
methyl-methionine was selected as a reporter molecule and the fluorine signal intensity was moni-
tored in the presence of NLM-NmMetQ and several methionine analogs (
Figure 3A
). For these
studies, we optimized the concentration of the reporter molecule, NLM-NmMetQ, and the relaxation
time (T2) for the NMR measurement. A reporter molecule concentration of 2 mM was chosen here
to decrease acquisition time. Additionally, 43
m
M NLM-NMMetQ was chosen as a compromise
between using less protein and increasing the fraction of reporter bound to the protein. The relaxa-
tion time T2 = 120 ms was chosen for its ability to strongly attenuate but not eliminate the reporter
signal in the presence of 43
m
M NLM-NmMetQ. As previously described (
Dalvit et al., 2003
;
Dalvit and Vulpetti, 2019
), for all experiments, two fluorine spectra (1D and Car-Purcell-Meiboom-
Gill (CPMG) filtered) were acquired. The intensity signals of the reporter molecule measured in both
spectra and the ratio -ln(CPMG/1D) were calculated. We anticipated that analogs that bind to NLM-
NmMetQ would lead to the displacement of the reporter molecule, resulting in a decrease in the -ln
(CPMG/1D) ratio.
Our results for the competition binding experiments are shown in
Figure 3C
. The plot shows the
signal intensity ratio of the reporter molecule in the presence of each methionine analog. Since dis-
placement of the reporter molecule by the analog correlates to the analog’s binding affinity, methio-
nine analogs with higher affinity will be positioned toward the left side of the plot, while lower
affinity methionine analogs will appear on the right side. As controls, we measured the -ln(CPMG/
1D) ratios with the reporter molecule alone and reporter molecule plus NLM-NmMetQ. As expected,
the reporter molecule alone has a low -ln(CPMG/1D) ratio, while the reporter molecule plus NLM-
NmMetQ has a high -ln(CPMG/1D) ratio (less free reporter molecule due to NLM-binding).
Next, we carried out the FAXS experiments in the presence of various methionine analogs. We
first added L-methionine, a known high-affinity ligand of NmMetQ (Kd 0.2 nM
Nguyen et al., 2019
).
As expected for a higher affinity ligand, L-methionine completely displaced the reporter molecule.
We then examined two methionine analogs with amino group substitutions: (1) N-formyl-L-methio-
nine, which is used by bacteria to initiate translation and (2) N-acetyl-L-methionine, which is present
in bacteria (
Schmidt et al., 2016
) and human brain cells (
Smith et al., 2011
;
Figure 3C
, circles).
Addition of these analogs led to the respective complete or near complete displacement of the
reporter molecule, indicating that changes to the amino group do not dramatically affect the sub-
strate’s ability to bind tightly to NLM-NmMetQ. D-methionine displaced less reporter than L-methio-
nine, consistent with its lower binding affinity (3.5
m
M
Nguyen et al., 2019
), while N-acetyl-D
methionine failed to displace the reporter molecule. These results suggest that modifications to
both the amino group and stereochemistry lead to significantly weaker binding than the singly modi-
fied derivative.
Similar to our observations with D-methionine and N-acetyl-D methionine, changes to the car-
boxyl group resulted in less displacement of the reporter molecule than L-methionine. Specifically,
L-methioninol, with the carboxyl group reduced to an alcohol, failed to displace the reporter mole-
cule while L-methionine ethyl ester only partially displaced the reporter molecule (
Figure 3C
,
circles).
Lastly, changes to the L-methionine side-chain exhibited varying effects. Methionine analogs with
changes to the sulfur atom, including seleno-L-methionine, L-methionine sulfoximine, and L-norleu-
cine were well tolerated, with a greater displacement of the reporter molecule than D-methionine,
which has an estimated Kd of 3.5
m
M (
Nguyen et al., 2019
). However, L-ornithine failed to displace
the reporter molecule, suggesting that binding of ligands with a charged amino group is energeti-
cally unfavorable. Side-chain length also plays a role in methionine analog affinity to NLM-NmMetQ.
Increasing the side-chain length by an addition of a methylene group (L-ethionine) was better
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
7 of 21
Research article
Biochemistry and Chemical Biology
Structural Biology and Molecular Biophysics
tolerated than decreasing the length by one carbon (S-methyl-L-cysteine). Shorter thiol derivatives
(L-cysteine and L-homocysteine) were ineffective at displacing the reporter molecule. Together, our
data establish that NLM-NmMetQ can accommodate variability in the binding of methionine ana-
logs, including modifications to the amino group, D-stereochemistry, and changes in the side-chain
(to a limited extent), while exhibiting little tolerance for variations in the carboxyl group.
Figure 3.
Characterization of the interaction of methionine analogs with NmMetQ using FAXS and ATPase experiments. (
A
) Schematic diagram of the
FAXS experiment. The intensity of the fluorine signal decreases in the presence of NLM-NmMetQ. Addition of the methionine analog causes the
fluorine signal intensity of the reporter molecule to increase due to its displacement from NLM-NmMetQ. (
B
) Chemical structures of the methionine
analogs used in this study. (
C
) (Top) Ordering of methionine analogs by their binding affinity to NLM-NmMetQ. (Bottom) Schematic representation of
FAXS experiment depicted in bulk solution. Methionine analogs with higher affinity are positioned toward the left side of the plot, while lower affinity
methionine analogs are positioned toward the right. (
D
) ATPase activity of NmMetNI at 2 mM ATP in the presence of lipo-NmMetQ and methionine
analogs at 1:8:50 molar ratio, respectively. N=3 error bars represent SEM.
The online version of this article includes the following source data for figure 3:
Source data 1.
The measured -ln(Icpmg/I1D) values: NMR.xlsx.
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
8 of 21
Research article
Biochemistry and Chemical Biology
Structural Biology and Molecular Biophysics
To determine whether methionine analogs could serve as potential substrates for the lipo-
NmMetQ:NmMetNI system, we measured NmMetNI ATPase activity in presence of lipo-NmMetQ
and several methionine analogs. For these assays, we chose methionine analogs identified by FAXS
to bind NLM-NmMetQ with an affinity similar or higher than D-methionine, a known substrate for
E.
coli
NmMetNI. Since substrate-stimulated ATPase activity is a hallmark of ABC transporters
(
Bishop et al., 1989
;
Mimmack et al., 1989
), we expected methionine analogs that are substrates
for this system would stimulate NmMetNI ATPase activity.
Figure 3D
shows the results for the
methionine analog stimulation of NmMetNI ATPase activity. As a negative control, we tested L-cys-
teine, where, as expected, no substrate-stimulated ATPase stimulation was detected. Our data show
that the following methionine analogs led to substrate-stimulated ATPase activity: N-acetyl-L-methi-
onine, N-formyl-L-methionine, L-norleucine, L-ethionine, and L-methionine sulfoximine. However, no
correlation was seen between affinity to NLM-NmMetQ and NmMetNI stimulation. This data sug-
gest that binding to NmMetQ is necessary to initiate transport; however, this step alone does not
determine the magnitude of NmMetNI ATPase stimulation. Taken together, our FAXS and ATPase
experiments suggest that N-formyl-L-methionine, L-norleucine, L-ethionine, and L-methionine sulfox-
imine are potential substrates for the
N. meningitidis
lipo-NmMetQ:NmMetNI system.
Structures of
N. meningitidis
MetNI in the inward-facing conformation
and
N. meningitidis
lipo-NmMetQ:NmMetNI complex in the outward-
facing conformation
To gain insight into the role of lipo-NmMetQ in the NmMetNI transport cycle, we determined struc-
tures of NmMetNI in different conformational states by single-particle cryo-EM. By varying the nucle-
otide analog and concentration, multiple conditions were screened to identify ones that promoted
different conformational states. With 1 mM AMPPNP (below the Km), the structural analysis of an
equimolar mixture of lipo-NmMetQ and NmMetNI revealed that NmMetNI was captured in the
inward-facing conformation at 3.3 A
̊
resolution (
Figure 4A
); no densities for either AMPPNP and
lipo-NmMetQ were observed. For this data set, the two dimensional class averages showed clear
structural features, suggesting a high level of conformational homogeneity (
Figure 4—figure sup-
plement 1
). The overall architecture of NmMetNI is similar to previously determined structures of
EcMetNI, comprising two copies of each TMD and NBD, encoded by
MetI
and
MetN
, respectively
(
Kadaba et al., 2008
;
Johnson et al., 2012
). Each MetI subunit contains five transmembrane helices
per monomer for a total of ten transmembrane helices per transporter (
Figure 4B
).
A comparison between NmMetNI and EcMetNI reveals similar subunit folds, with the root mean
square deviation (RMSD) of 2.4 A
̊
over 843 C
a
atoms. As predicted from the primary sequence, the
NmMetN subunits lack the C2 autoinhibitory domain. As a result, the interfaces of NmMetNI and
EcMetNI are distinct. In the inward-facing conformation of NmMetNI, the NBDs interact directly. In
contrast, in EcMetNI, the inward-facing conformation forms an interface through the C2 autoinhibi-
tory domains, with a slight separation between the NBDs (
Figure 4—figure supplement 2A
). A simi-
lar increase in NBD:NBD distance, defined as the average distance between C
a
of glycines in the P
loop and signature motifs, is observed in the previously determined molybdate ABC transporter
structures,
Methanosarcina acetivorans
ModBC (MaModBC) and
Archaeoglobus fulgidus
ModBC
(AfModBC) (
Hollenstein et al., 2007
;
Gerber et al., 2008
;
Figure 4—figure supplement 2B
). To
date, these are the only other reported pair of homologous structures, one with an autoinhibitory
domain and one without. For AfModBC, which lacks the autoinhibitory domain, the NBD:NBD dis-
tances are ̃17 A
̊
and 21 A
̊
for each AfModBC in the asymmetric unit. For MaModBC, which does
have an autoinhibitory domain, this distance increases to ̃27 A
̊
. A comparison of these structures
suggests that type I ABC importers share a common quaternary arrangement in the inward-facing
conformation such that the presence of a regulatory domain increases the separation of the NBD:
NBD distance in comparison to the homologous structure without a regulatory domain
Figure 4—
figure supplement 2D
.
Reasoning that increasing the nucleotide concentration to above the Km for MgATP would pro-
mote complex formation, we mixed equimolar lipo-NmMetQ and NmMetQ in the presence of 5 mM
ATP. Under these conditions, we were able to determine the single-particle cryo-EM structure of the
complex to 6.4 A
̊
resolution (
Figure 4
). Despite extensive efforts, we were unable to improve the
resolution of this complex. The structure was modeled by rigid-body refinement of both NmMetNI
in the inward-facing conformation (traced from the 3.3 A
̊
resolution reconstruction) and the
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
9 of 21
Research article
Biochemistry and Chemical Biology
Structural Biology and Molecular Biophysics
previously determined soluble NmMetQ structure in the substrate-free conformation (PDB:6CVA).
Our model shows lipo-NmMetQ docked onto the NmMetI subunits and the NmMetN subunits in a
closed dimer state. No clear density was seen for the lipid moiety of lipo-NmMetQ or ATP (
Figure 4
).
Of note, Liu et. al. were also not able to observe the lipid moiety of lipo-SBP in complex with an
ABC transporter, despite these cryo-EM structures resolving at higher resolutions (3.30, 3.44, 3.78
A
̊
)
Liu et al., 2020
. A comparison between NmMetNI and EcMetNI in the outward-facing
Figure 4.
Architecture of NmMetNI and lipo-NmMetQ:NmMetNI complex. (
A
) The 3.3 A
̊
resolution cryo-EM map and NmMetNI in the inward-facing
conformation in two views. (
B
) Transmembrane localization of NmMetI, showing NmMetI contains five transmembrane helices per monomer (
C
) The 6.4
A
̊
resolution cryo-EM map and model of NmMetNI in complex with lipo-NmMetQ in the presence of ATP. NmMetNI is in the outward-facing
conformation. NmMetI is shown in light/dark blue, NmMetN in light/dark grey and lipo-NmMetQ in light pink. The membrane is represented by a gray
box.
The online version of this article includes the following figure supplement(s) for figure 4:
Figure supplement 1.
Cryo-EM map generation and data processing refinement of NmMetNI in the inward-facing conformation.
Figure supplement 2.
Comparison of type I ABC transporters.
Figure supplement 3.
Cryo-EM map generation and data processing refinement of lipo-NmMetQ:NmMetNI complex in the outward-facing
conformation.
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
10 of 21
Research article
Biochemistry and Chemical Biology
Structural Biology and Molecular Biophysics
conformation in complex with their respective MetQ proteins reveals they have similar conforma-
tions, with RMSD of 2.2 A
̊
over 1048 C
a
atoms (
Figure 4—figure supplement 2C
). In contrast to the
inward-facing conformation, the NBD:NBD arrangement is similar for both EcMetNI and NmMetNI.
Lipo-MetQ proteins may be present in a variety of other Gram-negative
bacteria
We used a bioinformatics approach to determine if other Gram-negative bacteria could have lipid-
modified MetQ proteins. For the analysis, we chose predicted MetQ protein sequences from the
InterPro family IPR004872 (of which NmMetQ is a member), restricting the search to Proteobacteria,
Taxonomy ID 1224, and 90% identity. The amino acid sequence of the MetQ proteins were then ana-
lyzed using SignalP 5.0.
Figure 5
summarizes the results. Our data revealed that lipid-modified
MetQ proteins may be present in all classes of Proteobacteria (Alpha, Beta, Gamma, Delta, and Epsi-
lon), with varying number of lipid-modified MetQ proteins in each Order (magenta vs white). These
results suggest that lipid modification of MetQ proteins are not restricted to
N. meningitidis
(this
work) and
E. coli
(
Tokuda et al., 2007
;
Carlson et al., 2019
) and are likely present in a wide variety
of Gram-negative bacteria.
Discussion
NmMetQ has been previously identified as an OM surface-exposed candidate meningococcal vac-
cine antigen (
Pizza et al., 2000
), possibly playing a role in bacterial adhesion to human brain endo-
thelial cells (
Ka
́
nova
́
et al., 2018
). However, the presence of NmMetQ at the OM challenges the
prevailing view that SBPs reside in the periplasm, freely diffusing between the IM and OM
(
Thomas and Tampe
́
, 2020
). To better understand whether NmMetQ has lost its ABC transporter-
dependent function at the IM and how NmMetQ remains at the surface of the bacterium, we used
multiple biophysical techniques to characterize the structure and function of NmMetQ and
Figure 5.
Distribution of lipid-modified MetQ proteins in different classes of Proteobacteria, a major phylum of Gram-negative bacteria. Plot of the
number of MetQ proteins analyzed in each Order, grouped by Proteobacteria. Predicted lipid-modified and secreted MetQ proteins are shown in
magenta and white, respectively.
The online version of this article includes the following source data for figure 5:
Source data 1.
Distribution of lipid-modified MetQ proteins: lipoproteins.xlsx.
Sharaf
et al
. eLife 2021;10:e69742.
DOI: https://doi.org/10.7554/eLife.69742
11 of 21
Research article
Biochemistry and Chemical Biology
Structural Biology and Molecular Biophysics