GNA33 of Neisseria meningitidis is a lipoprotein required for cell separation, membrane architecture, and virulence.

Abstract

GNA33 is a membrane-bound lipoprotein with murein hydrolase activity that is present in all Neisseria species and well conserved in different meningococcal isolates. The protein shows 33% identity to a lytic transglycolase (MltA) from Escherichia coli and has been shown to be involved in the degradation of both insoluble murein sacculi and unsubstituted glycan strands. To study the function of the gene and its role in pathogenesis and virulence, a knockout mutant of a Neisseria meningitidis serogroup B strain was generated. The mutant exhibited retarded growth in vitro. Transmission electron microscopy revealed that the mutant grows in clusters which are connected by a continuous outer membrane, suggesting a failure in the separation of daughter cells. Moreover, sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis of culture supernatant revealed that the mutant releases several proteins in the medium. The five most abundant proteins, identified by matrix-assisted laser desorption ionization-time-of-flight mass spectrometry analysis, belong to the outer membrane protein family. Finally, the mutant showed an attenuated phenotype, since it was not able to cause bacteremia in the infant rat model. We conclude that GNA33 is a highly conserved lipoprotein which plays an important role in peptidoglycan metabolism, cell separation, membrane architecture, and virulence.

Full text

Page 1

INFECTION AND IMMUNITY, Apr. 2004, p. 1914–1919
0019-9567/04/$08.00?0 DOI: 10.1128/IAI.72.4.1914–1919.2004
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
Vol. 72, No. 4
GNA33 of Neisseria meningitidis Is a Lipoprotein Required for Cell
Separation, Membrane Architecture, and Virulence
Jeannette Adu-Bobie,1Pietro Lupetti,2Brunella Brunelli,1Dan Granoff,3Nathalie Norais,1
Germano Ferrari,1Guido Grandi,1Rino Rappuoli,1* and Mariagrazia Pizza1
IRIS, Chiron Vaccines,1and Unit of Electron Microscopy and Cryotechniques, Dipartmento Biologia Evolutiva,
University of Siena,253100 Siena, Italy, and Children’s Hospital Oakland Research Institute,
Oakland, California 946093
Received 8 August 2003/Returned for modification 17 October 2003/Accepted 18 December 2003
GNA33 is a membrane-bound lipoprotein with murein hydrolase activity that is present in all Neisseria
species and well conserved in different meningococcal isolates. The protein shows 33% identity to a lytic
transglycolase (MltA) from Escherichia coli and has been shown to be involved in the degradation of both
insoluble murein sacculi and unsubstituted glycan strands. To study the function of the gene and its role in
pathogenesis and virulence, a knockout mutant of a Neisseria meningitidis serogroup B strain was generated.
The mutant exhibited retarded growth in vitro. Transmission electron microscopy revealed that the mutant
grows in clusters which are connected by a continuous outer membrane, suggesting a failure in the separation
of daughter cells. Moreover, sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis of culture
supernatant revealed that the mutant releases several proteins in the medium. The five most abundant
proteins, identified by matrix-assisted laser desorption ionization–time-of-flight mass spectrometry analysis,
belong to the outer membrane protein family. Finally, the mutant showed an attenuated phenotype, since it was
not able to cause bacteremia in the infant rat model. We conclude that GNA33 is a highly conserved lipoprotein
which plays an important role in peptidoglycan metabolism, cell separation, membrane architecture, and
virulence.
The shape and osmotic stability of most bacteria are main-
tained by the polymeric glycopeptide murein (peptidoglycan),
a network of glycan strands interlinked by short peptides. Ex-
pansion of the cell wall during bacterial growth and splitting of
the septum requires the coordinate action of enzymes which
cleave covalent bonds within the murein sacculus and add new
peptidoglycan units. For Escherichia coli, it has been proposed
that a multienzyme complex (hydrolases and synthases) could
be responsible for such a task (9). Thus, murein hydrolases,
such as murein lytic transglycosylase A (MltA), are likely to be
involved in the enlargement of the murein net and therefore to
be essential for growth and division of the bacterial cell wall to
allow separation of the daughter cells. Lytic transglycolases are
a lysozyme subclass of murein hydrolases which not only cleave
the ?1-4 glycosidic bond between N-acetylmuramic acid and
N-acetylglucosamine but also transfer the glycosyl bond onto
the C-6 hydroxyl group of the terminal muramyl moiety, thus
forming a 1,6-anhydromuramic acid residue.
During screening of the Neisseria meningitidis serogroup B
genome to identify novel vaccine candidates and putative vir-
ulence factors, we identified a 47-kDa lipoprotein, genome-
derived Neisseria antigen 33 (GNA33 [NMB0033]), that is
highly conserved among N. meningitidis serogroup B strains as
well as other Neisseria species (18). Database sequence com-
parison shows that GNA33 shares 33% identity and 41% ho-
mology with the 38-kDa membrane-bound MltA of E. coli.
gna33 has been cloned and expressed in E. coli (18), and the
lipoprotein nature of the recombinant protein has been dem-
onstrated by incorporation of [3H]palmitate (11). The func-
tional homology of GNA33 with MltA has been confirmed,
since the recombinant protein is able to degrade both insoluble
murein and unsubstituted glycan strands (11). The formation
of 1,6-anhydrodisaccharide tri- and tetrapeptide reaction prod-
ucts demonstrated that GNA33 is a lytic transglycolase (11). In
E. coli an MltA deletion mutant as well as mutants with double
or triple mutations in the other lytic transglycosylases (MltB
and Slt70) showed no obvious phenotype (6, 16, 21). Com-
pared to the case for rod-shaped bacilli such as E. coli, little
work has been done to understand the cell division process in
gram-negative cocci.
To further define the function of GNA33 in N. meningitidis
and its role in growth and cell separation, a knockout mutant
was made. Here, we describe the construction and character-
ization of this mutant and show that GNA33 is essential for cell
separation, membrane architecture, and virulence.
MATERIALS AND METHODS
Bacterial strains and growth conditions. N. meningitidis strains MC58, NMB,
and BZ232 and their respective gna33 mutants were grown on GC agar base or
in GC broth with supplements at 37°C in 5% CO2,unless otherwise stated. For
selection of transformants, erythromycin was used at a concentration of 7 ?g/ml.
E. coli strain DH5 was used for cloning and grown on Luria-Bertani agar base or
in Luria-Bertani medium supplemented with ampicillin (100 ?g/ml) or erythro-
mycin (200 ?g/ml).
Construction of ?GNA33 mutants of N. meningitidis. Mutants of strains MC58,
NMB, and BZ232 (MC58 ?GNA33, NMB ?GNA33, and BZ232 ?GNA33,
respectively) in which the gna33 gene was replaced by allelic exchange with an
antibiotic cassette were prepared by transforming the parent strain with the
plasmid pBSUDGNA33ERM. This plasmid contains upstream and downstream
flanking regions for allelic exchange, a truncated gna33 gene, and the ermC gene
* Corresponding author. Mailing address: IRIS, Chiron Vaccines via
Fiorentina 1, 53100 Siena, Italy. Phone: 39 0577 243414. Fax: 39 0577
278508. E-mail: rino_rappuoli@chiron.com.
1914

Page 2

(encoding erythromycin resistance). The upstream flanking region (including the
start codon) from position ?867 to ?75 and the downstream flanking region
(including the stop codon) from position ?1268 to ?1744 were amplified from
MC58 by using the primers U33FOR (5?-GCTCTAGAGATGAGTCGAACAC
AATGAACAATGTCCTGA [the XbaI site is underlined]), U33REV (5?-TCC
CCCGGGCTCTTGCTTTGGCAGGCGGCGA [the SmaI site is underlined]),
D33FOR (5?-TCCCCCGGGCACGGGATATGTGTGGC [the SmaI site is un-
derlined]), and D33REV (5?-CCCGCTCGAGAGTAGGGACAACCGG [the
XhoI site is underlined]). Fragments were cloned into pBluescript (Stratagene,
Milan, Italy) and transformed into E. coli DH5 by using standard techniques
(19). Once all subcloning was complete, naturally competent Neisseria strains
MC58, NMB, and BZ232 were transformed by selecting a few colonies grown
overnight on GC agar plates and mixing them with 20 ?l of 10 mM Tris-HCl (pH
6.5) containing 1 ?g of plasmid DNA. The mixture was spotted onto a chocolate
agar plate, incubated for 6 h at 37°C with 5% CO2, and then diluted in phosphate
buffered-saline (PBS) and spread on GC agar plates containing 7 ?g of erythro-
mycin per ml. PCR with forward primer 5?-CGCGGATCCCATATGCAAAGA
AAGAGAATCCCAA and reverse primer 5?-GCTCTGAGGGCGACGACAG
GCGG was used to verify that correct allelic exchange with the chromosomal
gna33 gene had occurred, as indicated by the absence of an 800-bp PCR product,
indicating that the gene was replaced by a 1.2-kb fragment resulting in the
insertion of the erythromycin cassette. Lack of GNA33 expression was confirmed
by Western blot analysis as described below.
Preparation of outer membrane vesicles (OMVs). N. meningitidis strains
MC58, MC58 ?GNA33, BZ232, and BZ232 ?GNA33 were grown overnight at
37°C with 5% CO2on five GC plates, harvested with a loop, and resuspended in
10 ml of 20 mM Tris-HCl (pH 7.5)–2 mM EDTA. Heat inactivation was per-
formed at 56°C for 45 min, and the bacteria were disrupted by sonication for 5
min on ice (50% duty cycle, 50% output; Branson Sonifier with a 3-mm mi-
crotip). Unbroken cells were removed by centrifugation at 5,000 ? g for 10 min,
and the supernatant containing the total cell envelope fraction was recovered
and further centrifuged overnight at 50,000 ? g at 4°C. The pellet containing the
membranes was resuspended in 2% Sarkosyl–20 mM Tris-HCl (pH 7.5)–2 mM
EDTA and incubated at room temperature for 20 min to solubilize the inner
membranes. The suspension was centrifuged at 10,000 ? g for 10 min to remove
aggregates, and the supernatant was further centrifuged at 50,000 ? g for 3 h.
The pellet, containing the outer membranes, was washed in PBS and resus-
pended in the same buffer. The protein concentration was measured by the
Bio-Rad (Hercules, Calif.) detergent-compatible protein assay (modified Lowry
method), using bovine serum albumin (BSA) as a standard.
Preparation of whole-cell extracts. N. meningitidis strains MC58, MC58
?GNA33, BZ232, and BZ232 ?GNA33 were grown overnight on a GC plate at
37°C in 5% CO2. Colonies from each strain were collected and used to inoculate
7 ml of Mueller-Hinton broth, containing 0.25% glucose, to reach an optical
density at 620 nm (OD620) of 0.05 to 0.06. The culture was incubated for
approximately 2.5 h at 37°C with shaking until an OD620of 0.5 was reached and
then centrifuged for 10 min at 1,000 ? g. The supernatant was discarded, and the
pellet was resuspended in 500 ?l of PBS. Heat inactivation was performed at
56°C for 30 min.
TCA precipitation of proteins from culture supernatant. N. meningitidis
strains MC58 and MC58 ?GNA33 were grown in 200 ml of GC culture medium
in a humidified atmosphere containing 5% CO2to an OD620of 0.5. Bacterial
cells were removed by a 10-min centrifugation at 6,400 ? g, and the culture
medium was filtered through a 0.22-?m-pore-size filter (Millipore, Bedford,
Mass.). Proteins were precipitated with 10% trichloracetic acid (TCA)–0.4%
deoxycholate (30 min at 4°C), and the pellet was successively washed with 1%
TCA and acetone and dried with Speed-Vac (Savant, Holbrook, N.Y.). The
pellet was solubilized with 15 mM Tris-HCl (pH 8.8)–1 mM dithiothreitol.
SDS-PAGE and Western blot analysis. Total cell extract, TCA-precipitated
culture medium, and OMV preparations of strains MC58, MC58 ?GNA33,
BZ232, and BZ232 ?GNA33 were analyzed with a sodium dodecyl sulfate
(SDS)–10% polyacrylamide gel as described by Laemmli (14), using a Mini-
Protean II electrophoresis apparatus (Bio-Rad). Samples were suspended in
SDS sample buffer (0.06 M Tris-HCl [pH 6.8], 10% [vol/vol] glycerol, 2% [wt/vol]
SDS, 5% [vol/vol] 2-mercaptoethanol, 10 ?g of bromophenol blue per ml) and
heated to 100°C for 5 min before SDS-polyacrylamide gel electrophoresis (SDS-
PAGE) (3 ?g/lane). For Western blotting, the gel was equilibrated with buffer
(48 mM Tris-HCl, 39 mM glycine [pH 9.0], 20% [vol/vol] methanol) and trans-
ferred to a nitrocellulose membrane (Bio-Rad) by using a Trans-Blot (Bio-Rad)
semidry electrophoretic transfer cell. The nitrocellulose membranes were
blocked with 10% (wt/vol) skim milk in PBS containing 0.2% (wt/vol) sodium
azide. Anti-His tag-GNA33 monoclonal antibody (MAb) 25 (8) was diluted in
PBS containing 1% (wt/vol) BSA and 1% (wt/vol) Tween 20. Bound antibody was
detected by using rabbit anti-mouse immunoglobulin (immunoglobulins G, A,
and M)–alkaline phosphatase conjugate polyclonal antibody (Zymed, South San
Francisco, Calif.) and Sigma Fast 5-bromo-4-chloro-3-indolylphosphate–ni-
troblue tetrazolium substrate (Sigma Aldrich, St. Louis, Mo.).
MALDI-TOF mass spectrometry. Protein bands were excised from the gel,
washed with 100 mM ammonium bicarbonate-acetonitrile (50:50, vol/vol), and
dried with a Speed-Vac centrifuge (Savant). The dried spots were digested at
37°C for 2 h by adding 7 to 10 ?l of a solution containing 50 mM ammonium
bicarbonate, 5 mM dithiothreitol, and 0.012 ?g of sequencing-grade trypsin
(Promega, Madison, Wis.) per ?l. After digestion, 5 ?l of 0.1% trifluoroacetic
acid was added, and the peptides were desalted and concentrated with ZIP-TIPs
(C18; Millipore). Samples were eluted onto the mass spectrometer target (An-
chorchip 384 to 400 ?m; Bruker, Bremen, Germany) with 2 ?l of 2,5-dihydroxy-
benzoic acid (5 g/liter) in 50% acetonitrile–0.1% trifluoroacetic acid and allowed
to air dry at room temperature. Matrix-assisted laser desorption ionization–time-
of-flight (MALDI-TOF) spectra were acquired on a Biflex III MALDI-TOF
apparatus (Bruker) equipped with a 337-nm N2laser and a SCOUT 384 multi-
probe. Spectra were acquired in positive mode; acceleration and reflector volt-
ages were set at 19 and 20 kV, respectively. Typically, each spectrum was deter-
mined by averaging 100 laser shots. Spectra were externally calibrated by using
a combination of four standards (angiotensin II [1,046.54 Da], substance P
[1,347.74 Da], bombesin [1,619.82 Da], and ACTH18-39 clip human [2,465.20
Da]) spotted on the probe adjacent to the samples. Protein identification was
carried out by both automatic and manual comparisons of experimentally gen-
erated monoisotopic peaks of peptides in the mass range of 700 to 3,000 Da with
computer-generated fingerprints, using the Mascot program run on proprietary
databases.
Growth experiments. All growth experiments were carried out at 37°C with 5%
CO2. The suspension of organisms was adjusted to an initial OD620of 0.05 in 7
ml of GC medium and incubated with continuous end-over-end rotation. The
OD was measured every 20 min, and at each time point bacteria were plated out
in triplicate to obtain the corresponding CFU per milliliter. Growth studies were
preformed three times. To determine whether the ?GNA33 mutant is able to
grow in complement-inactivated serum, both the wild-type and mutant strains
were grown for 3 h in heat-treated (56°C, 30 min) rabbit serum. Serial dilutions
were carried out, and bacteria were plated out to determine CFU.
Transmission electron microscopy. Bacteria were grown in GC medium to an
OD620of 0.5 and centrifuged, and the pellets were fixed for 1 h in 2.0% glutar-
aldehyde diluted in PBS. After fixation, the pellets were washed with PBS and
then postfixed in 1% osmium tetroxide for 1 h at 4°C. Samples were then
dehydrated with a graded series of ethanol solutions and embedded in Epon-
Araldite. Ultrathin sections were routinely stained with uranyl acetate. The
samples were then viewed and photographed in a Philips CM10 transmission
electron microscope operated at 80 kV.
Virulence in infant rats. The ability of BZ232 ?GNA33 to cause bacteremia
was tested in infant rats challenged intraperitoneally (i.p.). The assay was pre-
formed as described previously (17). In brief, 5- to 8-day-old-pups from litters of
outbred Wistar rats (Charles River, Raleigh, N.C.) were randomly redistributed
to the nursing mothers. After overnight growth on chocolate agar, several colo-
nies were inoculated into Mueller-Hinton broth (starting OD620of ?0.1), and
the test organism was grown for approximately 2.5 h to an OD620of ?0.6 (early
log phase). The bacteria were washed once in PBS plus 1% BSA and diluted to
the required CFU per milliliter, and 100 ?l of the mixture was administered i.p.
Blood specimens were obtained 18 h after the bacterial challenge by puncturing
the heart with a needle and syringe containing approximately 10 ?l of heparin
(1,000 U/ml; Fujisawa USA, Deerfield, Ill.) without preservative. Aliquots of 1,
10, and 100 ?l of blood were plated onto chocolate agar. The CFU per milliliter
of blood was determined after overnight incubation of the plates at 37°C with 5%
CO2. To determine whether the mutant is sensitive to heparin, the parent and
mutant strains were resuspended in PBS and incubated with or without heparin
for 10 min. Bacteria were then plated out to determine the CFU.
RESULTS
GNA33 knockout characterization. A mutation of the gna33
gene was generated in strains MC58, NMB, and BZ232, and
the deletion of the gene was confirmed by PCR with chromo-
somal DNA. Protein expression was analyzed by immunoblot-
ting in bacterial total lysates or in OMV preparations from
wild-type and mutant strains, using anti-GNA33 MAb 25,
which recognizes GNA33 and a mimotope present in serotype
VOL. 72, 2004 N. MENINGITIDIS LIPOPROTEIN INVOLVED IN CELL SEPARATION1915

Page 3

P1.2 of PorA (8). The results reported in Fig. 1 show that on
the 2996 strain, used as positive control, the MAb recognizes
two bands of 41 and 47 kDa. The upper band of 47 kDa
corresponds to the GNA33 protein, whereas the lower band of
41 kDa corresponds to the PorA antigen, consistently with the
finding that the GNA33 antigen mimics a surface-exposed
epitope on loop 4 of PorA in strains of serotype P1.2 (8). The
47-kDa band, corresponding to the GNA33 protein, is present
in the OMV preparations and total lysates of MC58 (lanes 2
and 3, respectively) and is absent in the OMV preparation and
total lysate of MC58 ?GNA33 (lanes 4 and 5). In the case of
the BZ232 strain, there are two reactive bands in the total cell
lysates and OMV preparations (lanes 6 and 7), but only the
41-kDa band is present in the BZ232 ?GNA33 mutant (lanes
8 and 9), consistent with its P1.2 serotype. The antibodies were
not able to recognize GNA33 on the surface of the MC58
strain by fluorescence-activated cell sorter analysis, suggesting
that the protein is not surface exposed (data not shown).
The GNA33 mutant exhibits aberrant cell morphology and
decreased viability. To evaluate whether GNA33 has any in-
fluence on the growth rate or colony morphology of N. men-
ingitidis, the wild-type MC58, NMB, and BZ232 strains and the
mutant MC58
?GNA33, NMB
?GNA33 strains were grown on solid or liquid medium. Fol-
lowing overnight incubation of the strains on agar plates, no
differences in colony morphology between wild-type and mu-
tant strains could be visualized with the naked eye or with a
light microscope (data not shown). To evaluate the influence
on growth, the strains were grown for 5 h at 37°C, and samples
were collected every 20 min. Each sample was evaluated for
OD and for the number of bacterial colonies by plating on GC
agar plates. The results of the experiment related to the MC58
and MC58 ?GNA33 strains are reported in Fig. 2. As shown,
the ?GNA33 mutant exhibits a retardation in growth com-
pared to the parent MC58 strain (Fig. 2a). Furthermore, the
number of CFU of MC58 ?GNA33 per milliliter was much
lower than that of the wild-type MC58 strain (Fig. 2b). At an
OD620of 0.5, the numbers of viable colonies were 6.7 ? 108
CFU/ml for the MC58 strain and 4.2 ?107CFU/ml for the
?GNA33, and BZ232
mutant strain, indicating duplication times of approximately 20
min for MC58 and 40 min for MC58 ?GNA33, resulting in a
dramatic reduction in the viable counts. The same results were
obtained with the other strains (data not shown).
To verify whether GNA33 plays a role in cell size or cell
shape, transmission electron microscopy was used to examine
the morphology of the GNA33 mutant. As shown in Fig. 3, the
mutant morphology (Fig. 3c and d) is different from the typical
diplococcal morphology of the wild-type MC58 (Fig. 3a and b).
The mutant grows in clusters (Fig. 3c), and while the inner
membrane is separated, the outer membrane remains contin-
uous around the clusters (Fig. 3d), resulting in abnormal cell
morphology, with cells of various sizes with an undivided sep-
tum. The clusters are represented by cells which are unable to
separate. The fact that the mutant grows in clusters is likely to
result in misleading outcomes with the colony counts. The
bacterial colonies obtained with the mutant strain may be gen-
FIG. 1. Detection of GNA33 in Western blotting analysis of recom-
binant GNA33, total lysates, and OMV preparations with anti-GNA33
MAb 25. Lane 1, control strain 2996; lane 2, MC58 OMVs; lane 3,
MC58 total lysate; lane 4, MC58 ?GNA33 OMVs; lane 5, MC58
?GNA33 total lysate; lane 6, BZ232 OMVs; lane 7, BZ232 total lysate;
lane 8, BZ232 ?GNA33 OMVs; lane 9, BZ232 ?GNA33 total lysate;
lane 10, recombinant GNA33 from strain 2996. As expected, no reac-
tivity is observed with the mutant strains; however, the antiserum
cross-reacts with another band at 47 kDa. This band has been charac-
terized as PorA (8).
FIG. 2. Comparison of the viabilities of the wild-type MC58 and
the GNA33 knockout mutant on the basis of OD620(a) and CFU per
milliliter (b). The mutant MC58 ?GNA33 strain shows a significantly
reduced growth rate and lower CFU with respect to the parent strain.
1916 ADU-BOBIE ET AL.INFECT. IMMUN.

Page 4

erated by clusters of cells and not by individual bacteria. The
NMB and BZ232 mutants exhibited the same morphology
(data not shown), confirming that the phenotype observed is a
result of the deletion of the gna33 gene and not a strain-
dependent phenomenon.
The GNA33 mutant releases outer membrane proteins.
Since the mutant is impaired in cell separation, we investigated
whether the mutation in the gna33 gene may also affect mem-
brane protein assembly. To address this issue, we analyzed the
extracellular proteins present in strain MC58 and MC58
?GNA33 culture supernatants at an OD620of 0.5. The culture
medium was filtered through a 0.22-?m-size-pore membrane
and concentrated by TCA precipitation. Twenty micrograms of
protein was separated by SDS-PAGE (Fig. 4). In the culture
medium of the mutant strain MC58 ?GNA33 (Fig. 4, lane 4),
several proteins not present in the supernatant of the wild-type
strain (Fig. 4, lane 3) were visible on the gel. The five most
abundant proteins (Fig. 4, lane 4) were in-gel digested with
trypsin and analyzed by MALDI-TOF mass spectrometry for
protein identification. The proteins were identified as the outer
membrane protein PorA (NMB1429), the major outer mem-
brane PIB (NMB2039), the class 4 outer membrane protein
(NMB0382), the class 5 outer membrane protein (NMB1053),
and a putative adhesin protein (NMB2095). With the excep-
tion of NMB2095, the other proteins are also the major pro-
teins identified in the MC58 OMV preparation (Fig. 4, lane 5).
The only two faint bands visible in the culture supernatant of
the wild-type culture (Fig. 4, lane 3) were both identified as the
class 5 outer membrane protein (NMB1053). In conclusion,
the deletion in the gna33 gene seems to affect correct topolog-
ical organization of the N. meningitidis membrane, resulting in
the selective release of several outer membrane proteins.
Virulence of the GNA33 knockout mutant. We investigated
the ability of the mutant strain to cause bacteremia in infant
rats. Since previous experiments had shown that MC58 is un-
able to cause consistent and reproducible bacteremia in infant
rats (data not shown), the BZ232 strain and its GNA33 mutant
were used in the study. The virulence experiment was carried
out by challenging 5-day-old infant rats i.p with 102and 105
CFU of BZ232 or BZ232 ?GNA33. Blood samples were ob-
tained after 18 h and plated onto chocolate agar plates. The
results obtained are summarized in Table 1. At both challenge
doses of 2.5 ? 102and 2.5 ? 105CFU, the parent BZ232 strain
caused bacteremia, while the mutant was unable to cause bac-
teremia even at the highest challenge dose. As shown in Table
1, all three animals given a challenge dose of approximately 105
CFU of strain BZ232 ?GNA33 did not exhibit bacteremia
after 18 h. In a second experiment, infant rats were challenged
with 108CFU of BZ232 ?GNA33, and even at this dose the
mutant strain failed to cause bacteremia. It is worth noting that
the challenge dose of the mutant strain is based on the CFU of
the clusters of bacteria, and thus a more bacteria are inocu-
lated than with the wild-type strain.
To verify whether the ?GNA33 mutant is unable to cause
bacteremia because of a susceptibility to killing in presence of
heparin or to lyse in the serum, the wild-type and mutant
FIG. 3. Transmission electron microscopy of MC58 (a and b) and
MC58 ?GNA33 (c and d). While the wild-type parental strain MC58
exhibits a typical diplococci phenotype, the GNA33 mutant shows
abnormal cell morphology. Bars, 2 ?m (a), 0.4 ?m (b and d), and 4 ?m
(c).
FIG. 4. SDS-PAGE of TCA-precipitated culture media and OMV
preparation. Lane 1, molecular mass markers; lane 2, 20 ?g of GC
growth broth, lane 3, 20 ?g of MC58 culture medium (OD620, 0.5);
lane 4, 20 ?g of MC58 ?GNA33 culture medium (OD620, 0.5); lane 5,
6 ?g of MC58 OMVs.
TABLE 1. N. meningitidis strain BZ232 in which GNA33 has been
deleted is unable to cause bacteremia and death in infant rats
challenged i.p.
Exp Strain
Bacterial
dose
(CFU)
Results of blood culture
18 h after challenge
No.
positive/
total
Mean CFU/ml
1 BZ232 (wild type)2.5 ? 102
2.5 ? 105
2 ? 102
2 ? 105
4/4
3/3
0/4
0/3
2.52 ? 106
1 ? 107
BZ232 ?GNA33
2BZ232 ?GNA334.6 ? 108
0/4
VOL. 72, 2004 N. MENINGITIDIS LIPOPROTEIN INVOLVED IN CELL SEPARATION1917

Page 5

strains were incubated with or without heparin for 10 min or
grown in heat-treated serum for 3 h. For the wild-type strain,
14.3 ? 105bacteria were obtained in the absence of heparin,
compared to 13.6 ? 105in the presence of heparin. In the case
of the mutant strain, 2.6 ? 104bacteria were obtained in the
absence of heparin, compared to 1.3 ? 104bacteria when
incubated with heparin. These results show that the mutant
strain is not more susceptible to killing in the presence of
heparin than the parent strain. When the mutant was grown in
heat-inactivated serum, the numbers of CFU at time zero for
the wild-type strain and mutant strain were 2.3 ? 107and 8.5
? 105, respectively. After 3 h of incubation, the numbers of
CFU for the wild-type strain and mutant strain were 1.7 ? 109
and 4 ? 107, respectively. The results therefore suggest that the
mutant strain is able to grow in heat-treated serum. Moreover,
the mutant strain exhibits a retardation in growth compared to
the parent strain, consistent with data obtained when the
strains were grown in GC broth.
DISCUSSION
In this study we show that deletion of the N. meningitidis
gene coding for GNA33, which is homologous to the lytic
transglycosylase MltA of E. coli, affects cell separation (pre-
venting the formation of daughter cells) cell membrane assem-
bly (causing a release of outer membrane proteins), and viru-
lence. Instead of the characteristic diplococci, the GNA33
mutant shows abnormal cell morphology, which includes an
undivided septum, with cells of various sizes growing in clusters
of up to 17 cells. The bacterial cells in these clusters have a
double septum and a continuous outer membrane.
In E. coli, a model for the coordinated metabolism of the
murein layer has been suggested (9). The basis of the model is
a multienzyme complex consisting of bifunctional murein
transglycolsylases-transpeptidases (known as penicillin-binding
proteins), lytic transglycosylases (Slt70, MltA, and MltB), and
MipA (9). It has been suggested that there might be two such
complexes, one for the synthesis of murein and another for its
degradation. Formation of a complex could be a way for the
secure and effective control of potentially autolytic murein
hydrolases and to ensure that the growth of the sacculus occurs
to maintain the specific shape of the bacterium. Genome com-
puter analysis has shown that N. meningitidis has homologs of
a majority of the complex members. In addition, by using
affinity chromatography, interaction between GNA33 and pen-
icillin-binding protein 2 from N. meningitidis serogroup B has
been shown (11), further suggesting that GNA33 may be part
of a multienzyme complex, as is the case for MltA in E. coli.
Deletion of three lytic transglycosylases in E. coli did not in-
fluence growth and division of the mutant E. coli strains. In this
study we report that deletion of a protein of N. meningitidis
with homology to the E. coli lytic transglycosylase MltA and
with a similar in vitro enzymatic activity (11) causes a dramatic
effect on growth and cell division. One of the reasons for
different phenotypes induced by the deletion of MltA in E. coli
and GNA33 in N. meningitidis may be the different cell shape
of the bacterium: E. coli cells are rod shaped, while neisseriae
are diplococcal. Enzymes shaping the form of the cells may
have slightly different roles or be components of different com-
plexes. This observation may represent a first step in the un-
derstanding of the mechanisms which regulate cell shapes. The
cell division process in gram-negative cocci is essentially un-
characterized. This study suggests that we may learn more by
studying the cell shape in cocci. The MinD mutant of Neisseria
gonorrhoeae (N. gonorrhoeae MinD is homologous to the MinD
protein of E. coli [3, 4]) exhibits a phenotype similar to that of
our GNA33 mutant (7). Abnormal cell morphology has been
also observed in mutants with mutations in murein hydrolase
genes, such as tcp in N. gonorrhoeae (7), Cwl (lyt) in Bacillus
subtilis (10), and iap in Listeria (13), but we have not been able
to find any sequence homology between these enzymes and
GNA33.
The gna33 gene deletion not only perturbs the cell separa-
tion process but also seems to affect membrane architecture.
By analyzing the culture supernatant of the MC58 mutant
strain, we found that several proteins are specifically released
in the medium. The five most abundant released proteins,
unambiguously identified by MALDI-TOF mass spectrometry,
belong to the outer membrane family. Moreover, four of these
proteins are major outer membrane proteins of MC58. Further
analysis by two-dimensional electrophoresis has allowed the
identification of more than 50 membrane proteins (unpub-
lished data). At present we can only speculate on the possible
mechanism of selective membrane protein release, which can
occur either because of a physical dissociation of membrane
material leading to the formation of membrane vesicles or
because of selective leakage (secretion) of outer membrane
proteins. Shedding of membrane vesicles during the active
phase of growth is known to occur in many gram-negative
bacteria (12, 22, 23, 24), including N. meningitidis (5). Inter-
estingly, loss of membrane material is significantly increased in
the absence of a specific subset of inner membrane, periplas-
mic, and outer membrane proteins that share the property of
strongly interacting with the murein layer and form large com-
plexes which physically link the inner and outer membranes,
maintaining envelope integrity. Among these proteins are the
Braun’s (murein) lipoprotein (Lpp) and the proteins belonging
the Tol-Pal system. This system includes the peptidoglycan-
associated lipoprotein (PAL), a protein anchored to the outer
membrane; the periplasmic protein TolB; and the inner mem-
brane complex formed by the TolQ, TolR, and TolA proteins
(15). Any mutation in lpp and the tol-pal system perturbs the
outer membrane and causes the formation of OMVs (1, 2, 15).
A similar phenotype might be envisaged for the ?GNA33
mutant, reflecting the close interaction of the GNA33 protein
with the murein layer.
The GNA33 knockout mutant was unable to cause bactere-
mia in the infant rat model even at the highest dose used. This
attenuated phenotype is likely to be due to an inability of the
mutant strain to multiply in the bloodstream, probably because
of higher sensitivity of the bacterial cluster to complement-
mediated bacterial lysis. Using insertional mutagenesis, Sun et
al. (20) identified several genes essential for pathogenesis. One
of the genes identified, NMB1279, is the homolog of the E. coli
lytic transglycosylase MltB gene. However, the Neisseria mu-
tant with the deletion in the NMB1279 gene that we generated
exhibits a normal diplococcal phenotype and normal growth
and does not release outer membrane proteins (data not
shown).
In conclusion, we have shown that GNA33, a highly con-
1918ADU-BOBIE ET AL.INFECT. IMMUN.