A Porphyromonas gingivalis haloacid dehalogenase
family phosphatase interacts with human
phosphoproteins and is important for invasion
Gena D. Tribble, Song Mao, Chloe E. James, and Richard J. Lamont*
Department of Oral Biology, University of Florida School of Dentistry, Gainesville, FL 32610-0424
Edited by Susan Gottesman, National Institutes of Health, Bethesda, MD, and approved June 6, 2006 (received for review November 10, 2005)
in prokaryotes and are generally involved in metabolic processes.
Porphyromonas gingivalis, an invasive periodontal pathogen, se-
cretes the HAD family phosphoserine phosphatase SerB653 when
in contact with gingival epithelial cells. Here we characterize the
structure and enzymatic activity of SerB653 and show that a
SerB653 allelic replacement mutant of P. gingivalis is deficient in
internalization and persistence in gingival epithelial cells. In con-
trast, mutation of a second HAD family serine phosphatase of P.
gingivalis (SerB1170), or of a serine transporter, did not affect
invasion. A pull-down assay identified GAPDH and heat-shock
protein 90 as potential substrates for SerB653. Furthermore, ex-
ogenous phosphatase regulated microtubule dynamics in host
cells. These data indicate that P. gingivalis has adapted a formerly
metabolic enzyme to facilitate entry into host cells by modulating
host cytoskeletal architecture. Our findings define a virulence-
related role of a HAD family phosphatase and reveal an invasin of
an important periodontal pathogen.
bacterial pathogenesis ? invasin ? microtubules ? intracellular survival
suffering from generalized periodontitis (1). The development of
between the host and microorganisms that colonize the gingival
sulcus. The Gram-negative anaerobe Porphyromonas gingivalis has
been strongly implicated in the initiation and progression of peri-
odontal disease and possesses a sophisticated array of virulence
factors, including those that allow the bacterium to adhere to and
invade host epithelial cells (2, 3). Invasion is accomplished through
manipulation of host signal transduction and remodeling of cy-
toskeletal architecture. Moreover, P. gingivalis localizes to the
perinuclear region, where it remains viable and blocks apoptosis of
the infected cell (4).
The molecular mechanisms used by P. gingivalis to facilitate
internalization and intracellular survival are only partially under-
stood. A screen of proteins secreted by P. gingivalis after contact
(SerB; refs. 5 and 6), corresponding to P. gingivalis ORF PG0653
(7). Based on motif searches, this protein is a member of the
haloacid dehalogenase (HAD) superfamily. Although this enzyme
family derives its name from the bacterial hydrolytic dehalogenases
(8), the group also includes phosphomutases, ATPases, phospho-
bers is low, but all are defined by the presence of three conserved
HAD family phosphatases use aspartate as the nucleophile, form
a phosphoenzyme intermediate during phosphoryl transfer, and
have an absolute requirement for a divalent ion cofactor (9).
HAD family enzymes are ubiquitous, with several thousand
members identified in genomic databases, but only a small number
of enzymes are well studied. The few members of the family that
have a defined function are associated with membrane transport,
metabolism, signal transduction, and nucleic acid repair. Recently,
eriodontal diseases are among the most common infections of
humans, with an estimated 5–20% of the world’s population
two eukaryotic HAD phosphatases, Eyes absent and Chronophin,
were characterized for their respective roles as a transcriptional
cofactor (10) and a regulator of actin dynamics (11). Such reports
begin to reveal the diverse functions managed by this widely spread
but poorly understood group of enzymes.
This report delineates our investigations into the role of the P.
gingivalis phosphoserine phosphatase during invasion of gingival
epithelial cells. We used biochemical analysis to confirm this
enzyme as a member of the HAD family of phosphatases. Study of
allelic exchange mutants in antibiotic protection assays indicated
that the SerB653, but not other HAD phosphatases or metabolic
enzymes, was required for maximum invasion efficiency. Further-
more, both internalization and intracellular survival were affected
epithelial cell extracts showed that the phosphatase may exert its
effect on invasion through components of membrane vesicular
a previously uncharacterized HAD family phosphatase that is
exploited by a prokaryote to facilitate an intracellular lifestyle.
as a member of the phosphoserine phosphatase subgroup of the
HAD family of hydrolases, based on the presence of three highly
SerB PG0653 (SerB653) contains an N-terminal ACT domain
and is commonly associated with enzymes subject to allosteric
regulation (12). PG0653 also has two putative transmembrane
domains and thus may be associated with the bacterial outer
membrane (Fig. 1B). The P. gingivalis genome encodes a second
predicted HAD family SerB enzyme, PG1170. The PG0653 and
PG1170 proteins are 36% identical and 61% similar in the phos-
phatase region and are both closely related to the putative phos-
phoserine phosphatase encoded by Bacteriodes thetaiotaomicron
VPI-5482, (71% and 41% identity, respectively) (7, 13). Sequence
of the PG0653 gene from strain 33277 is identical to that of the
sequenced strain W83, whereas the PG1170 ORF contained 21
nucleotide changes that resulted in five amino acid substitutions
motifs. We analyzed the two phosphoserine phosphatases in tan-
dem, to differentiate between potential pathogenic and metabolic
SerB653 Is Active on Phosphoserine and Phosphorylated Serine Pep-
tides. To determine the substrate reactivity profiles of SerB653 and
SerB PG1170 (SerB1170), purified his-tagged proteins were tested
Conflict of interest statement: No conflicts declared.
This paper was submitted directly (Track II) to the PNAS office.
Abbreviations: HAD, haloacid dehalogenase; GEC, gingival epithelial cells; HIGK, human
immortalized gingival keratinocytes.
*To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
© 2006 by The National Academy of Sciences of the USA
July 18, 2006 ?
vol. 103 ?
no. 29 ?
against phosphorylated amino acid and peptide substrates. Kinetic
favors phosphoserine 215-fold over phosphothreonine and 1,158-
fold over phosphotyrosine (Table 1). As suggested by the presence
is published as supporting information on the PNAS web site) of
SerB653 on phosphoserine is consistent with allosteric regulation,
with an H value of 3.7. SerB1170 is active on phosphothreonine
indicating SerB1170 would be more accurately classified as a
phosphothreonine phosphatase. Substrate preferences on phos-
phorylated peptides showed a similar pattern, with SerB653 most
active on the phosphoserine peptide, whereas SerB1170 worked
best on the phosphothreonine peptide (Fig. 2A). Although these
peptides are likely not ideal substrates for the phosphatases, their
on divalent cations as part of the active site. Addition of either
the cation-displacing inhibitor sodium fluoride or EDTA to
either phosphatase reaction resulted in a decrease in enzyme
activity, confirming the importance of cofactors for these en-
zymes (Table 2). The phosphate analog orthovanadate was also
that mimics the transition state of phosphoryl-transfer reactions
and competes with the substrate for interaction with the nucleo-
of these enzymes utilizes a covalent phosphoenzyme interme-
diate, a feature shared by other HAD phosphatases (15, 16).
Neither phosphatase was inhibited by up to 5 ?M okadaic acid,
an inhibitor specific for PPP family serine?threonine phospha-
tases. To further verify that SerB653 is a HAD-family phospha-
tase, site-directed mutagenesis was used to convert the predicted
nucleophile aspartate 198 to asparagine. Activity of the D198N
mutant enzyme was 9% of wild-type on the phosphoserine
substrate (Fig. 2B), verifying the importance of the aspartate 198
residue for catalysis.
SerB Gene Replacement Mutations and Effect on Invasion Efficiency.
Because the SerB653 protein, but not the SerB1170 protein, was
originally recovered extracellularly in the presence of gingival
epithelial cell (GEC) components (5), we hypothesized that
SerB653 might play a role in the invasion process of P. gingivalis.
serB653 and serB1170 single and double allelic replacement
mutants were thus constructed and tested for invasion of epi-
thelial cells. To control for the possibility that loss of serine
phosphatase activity affects serine production and metabolic
fitness, we also constructed a serine transporter mutant
sstT::pVA3000 to evaluate the importance of serine metabolism
for invasion (17). Growth curves in complex media were per-
formed for all mutants to test for changes in cell growth or
fitness. Although the double mutant serB653::ermF?
serB1170::tetQ showed a slight delay in entry to log-phase, all
other mutants displayed growth curves indistinguishable from
the wild type (not shown). Loss of serB653 gene product, either
in a single or double mutation, significantly (P ? 0.001) reduced
invasion by P. gingivalis of the human immortalized gingival
keratinocyte (HIGK) cell line (Fig. 3A). In contrast, loss of the
phosphatase SerB1170 alone, or of the serine transporter, did
not affect invasion efficiency. To ensure that the role of SerB653
proteins. (A and B) PG0653 and (C) PG1170. ACT domains in SerB653 are
black bars in 653 and 1170. HAD superfamily motifs are shown in bold italics,
A. Amino acid sequence of strain 33277 PG1170 is shown above substituted
Amino acid sequence, hydrophobicity, and structural motifs for SerB
Table 1. SerB activity on phosphorylated amino acids
1,5087.3 ? 10?1
3.4 ? 10?3
6.3 ? 10?4
1.1 ? 10?3
2.1 ? 10?1
5.3 ? 10?3
The phosphatase activity of His-tagged fusion proteins was assayed by measuring phosphate release from
phosphoserine (P-Ser), phosphothreonine (P-Thr), and phosphotyrosine (P-Tyr). Kmand kcatvalues were deter-
mined from substrate concentration curves analyzed with nonlinear regression.
www.pnas.org?cgi?doi?10.1073?pnas.0509813103Tribble et al.
was not strain- or cell line-specific, antibiotic protection assays
were also performed with P. gingivalis strain W83 and its isogenic
serB653 mutant and with both isogenic parent–mutant pairs in
a significant reduction in HIGK cell invasion efficiency com-
pared to its parental strain (Fig. 3B). Indeed, the reduction of
invasion with the W83 mutant was more pronounced than that
of the 33277 mutant, possibly the consequence of the W83 strain
the reduction in invasion by the 33277 serB653 mutant was less
than in HIGK cells (Fig. 3C), probably because transformation
altered signal transduction pathways involved in invasion of
HIGK cells. Parent-to-mutant ratios for W83 in GEC were
similar to HIGK cells. To confirm the involvement of SerB653
in the reduced invasion phenotype, we complemented the 33277
mutant with a plasmid containing the SerB653 gene (pTSerB).
Additionally, the parent plasmid pTCow was conjugated into the
wild-type 33277 strain and into the 33277 serB653::ermF mutant;
these strains were used as positive and negative controls for
invasion. The presence of plasmid impeded P. gingivalis invasion
efficiency; nonetheless, the complemented strain showed a
16-fold improvement in invasion over the mutant (Fig. 3D).
SerB653 Mutation Results in a Delay in Cell Penetration but Not
Adhesion.Invasion of eukaryotic host cells by bacterial pathogens
is a complex process, requiring bacterial attachment, penetra-
tion, and intracellular survival. The antibiotic protection assay
reports the collective outcome of these processes. To identify the
stage of invasion affected by the serB653 mutation, an adherence
assay and invasion time course were performed. No significant
difference in adherence to HIGK cells between the P. gingivalis
wild-type strains 33277 and W83 as compared to their isogenic
used fluorescent microscopy to quantitate the levels of intracel-
lular P. gingivalis. After 10 min of bacterial interaction with
HIGK cells, both parental and mutant strains showed similar low
levels of invasion (Fig. 4B). However, invasion by the wild type
was complete within 30 min, whereas the mutant required 2 hr
to complete the invasion process. The timing of the wild-type
invasion is consistent with previous studies in primary GECs, in
phosphopeptide RRApSVA (Upstate, Charlottesville, VA), threonine phosphopeptide RRApTVA (Promega, Madison, WI), and tyrosine phosphopeptide
RRLIEDAEpYAARG (Upstate). (B) Phosphatase activity of the SerB653 D198N mutant. Purified His-tagged enzymes were tested against the amino acid substrate
SerB enzymes are active on phosphorylated peptide substrates and use aspartate as the nucleophile. (A) Phosphatase activity on peptides: serine
Table 2. Inhibition of SerB activity
50% enzyme activity
Affect of common phosphatase inhibitors on enzyme activity. His-SerB 653
reactions were performed on phosphoserine, and His-SerB 1170 on phospho-
threonine. The concentration of inhibitor required to reduce activity to 50%
of wild type is shown.
graphs, invasion levels are represented as total colony-forming units per 2 ? 105
cells. Statistical significance between invasion levels was measured by using an
cells using P. gingivalis strain 33277 and the isogenic mutants described in
Materials and Methods. (B) Invasion of HIGK cells with P. gingivalis W83 and its
isogenic SerB653 mutant. (C) Invasion of primary GEC with 33277 and W83
mented SerB653 mutant. Statistical significance is between mutant and comple-
Allelic exchange mutants and gingival epithelial cell invasion. For all
Tribble et al.PNAS ?
July 18, 2006 ?
vol. 103 ?
no. 29 ?
which P. gingivalis was shown to complete invasion within 20 min
(20). These results indicate that the SerB653 mutant deficiency
is at the stage of host cell internalization. Additionally, although
fluorescent microscopy reveals the presence of the P. gingivalis
SerB653 mutant at wild-type invasion levels after 2 hr, the
antibiotic protection assay demonstrates there are less viable
mutant cells in the host at the same time point. Thus, in addition
to the deficiency in internalization, the SerB653 mutant is
compromised in its ability to survive inside host cells. Experi-
ments are ongoing to investigate interactions between the SerB
mutant and intracellular microbial sensing components such
as nucleotide-binding oligomerization domain (NOD) pro-
Interactions Between SerB653 and Host Proteins. To identify epi-
thelial cell proteins that might interact with SerB653 during the
invasion process, we used a bead-based pull-down kit for de-
tecting protein prey with his-tagged bait. Two proteins (85 and
37 kDa) from HIGK cell extracts were identified by the pull-
from the 85-kDa band (Fig. 5B), which matched mammalian
heat-shock protein 90 (HSP90). Five peptides sequenced from
the 37-kDa protein (Fig. 5C) identified it as human GAPDH.
Each of these proteins is known to be phosphorylated on serine
resides and thus able to serve as targets for serine phosphatases
clients, many of which are involved in signal transduction
pathways (24). Although the role of HSP90 phosphorylation is
not completely understood, it has been shown in several cases to
modulate the affinity of chaperone?client interactions (25).
GAPDH is primarily known for its role as a glycolytic enzyme,
but accumulating evidence suggests that this enzyme is involved
in a variety of activities unrelated to energy production, includ-
ing membrane fusion, microtubule bundling, DNA repair, and
apoptosis (26). GAPDH has been linked to the recruitment of
microtubules to membranes during vesicular trafficking, and it
has been hypothesized that phosphorylation regulates the inter-
action of GAPDH with tubulin during this process (27). Because
microtubule activity is known to be required for optimal P.
gingivalis invasion (2, 3, 28), and because microtubules play an
important role in cell internalization for other invasive patho-
gens (29), we decided to further investigate the possible link
between SerB653 and tubulin dynamics.
SerB Enzyme Alters Microtubule Dynamics During Invasion. Because
SerB653 can be released into the external milieu by P. gingivalis
(5), the response of HIGK microtubules to extracellular SerB653
was examined. Fluorescent microscopy (Fig. 6) revealed that
show a striking rearrangement of microtubules to the cell surface
compared to control cells. This effect was further exaggerated at
the 2-hr time point, where some cells with microtubule rear-
rangements also show significant changes in cell morphology.
Microtubule remodeling by SerB653 may provide the mecha-
nism by which the enzyme can modulate P. gingivalis penetration
and is consistent with GAPDH acting as a SerB653 substrate.
SerB653, however, likely has multiple effects on epithelial cell
SerB653, GAPDH, and tubulin is undergoing investigation in
Tightly regulated protein phosphorylation and dephosphoryla-
tion are an important means of signal transduction in both
prokaryotes and eukaryotes. In prokaryotes, protein phosphor-
ylation is mediated primarily by histidine kinases; however,
serine?threonine kinases and phosphatases have also been de-
scribed, and these enzymes are important for virulence in
cus agalactiae, and Listeria monocytogenes (30, 31). Moreover,
Salmonella, Shigella, and Yersinia use type III secretion machin-
ery to translocate kinases and phosphatases directly into host
cells, where they can interfere with host-cell signal transduction,
(A) Comparison of adherence efficiency between wild-type strains 33277,
W83, and selected mutants, by using formalin-fixed HIGK cells. An unpaired t
test comparing 33277 and the serB653::ermF mutant showed no significant
difference (P ? 0.36). A P. gingivalis 33277 fimbrial- (fimA) deficient mutant
(19) was used as a control for reduced adherence (P ? 0.001 compared to
33277, indicated by*). Results are from two independent experiments in
triplicate. (B) Quantitation of invasion over 3 hr. Fluorescent microscopy
conversion of fluorescent intensity to grain counts by using Morphometrics
image analysis software.
down assays were performed as described in Materials and Methods. Proteins
were separated by SDS?PAGE, and bands unique to the SerB653–HIGK inter-
and 37-kDa proteins. Peptides identified by MALDI-TOF analysis for each
protein band are shown.
SerB653 interaction with gingival epithelial cell extracts. (A) Pull-
www.pnas.org?cgi?doi?10.1073?pnas.0509813103Tribble et al.
often impacting cytoskeletal architecture (32). Although se-
creted bacterial tyrosine phosphatases and a serine?threonine
kinase have been identified (30, 31), the current report describes
a bacterial HAD family phosphoserine phosphatase important
for invasion of host cells.
SerB653, but not SerB1170, is secreted by P. gingivalis after
contact with epithelial cells or their components (5). Because
P. gingivalis lacks a type III secretion apparatus, and SerB653
does not possess a signal peptide, this enzyme may constitute
one of an emerging family of nonclassically secreted bacterial
proteins (33). SerB653 is also predicted to localize to the outer
membrane, because of the presence of two hydrophobic po-
tential membrane-spanning domains. These same putative
hydrophobic regions are similarly located in the SerB proteins
for Bacteroides fragilis and B. thetaiotaomicron, but not in the
P. gingivalis SerB1170 protein that was not involved in epithe-
lial cell interactions. Other members of the phosphoserine
phosphatase subfamily outside the Bacteroides taxonomic
group do not have the predicted C-terminal transmembrane
domains, indicating this may be an adaptation by this phylo-
genetic group of organisms.
The best-defined phosphoserine phosphatases belong to pro-
karyotes of the proteobacteria taxonomic group. These enzymes
are responsible for the biosynthesis of serine from 3-phospho-
serine and as such are part of the serine?threonine biosynthetic
pathway. As an asaccharolytic organism, P. gingivalis acquires
energy by uptake and catabolism of small peptides; biosynthesis
growth conditions. The lack of pressure on this enzyme for
fitness would make it eligible to develop new functions unrelated
to biosynthesis or metabolism.
The SerB653 mutant exhibits a normal growth rate in labo-
ratory medium but a reduced invasion efficiency for gingival
epithelial cells. Moreover, intracellular survival of the mutant is
also attenuated. Hence, SerB is a multifunctional invasin of P.
gingivalis that is involved both in entry and intracellular persis-
tence. In coprecipitation experiments, purified recombinant
SerB653 pulled down GAPDH and HSP90. Serine phosphory-
lation is thought to regulate microtubule?GAPDH interactions
in vesicular trafficking (23, 27, 34, 35), and interestingly this
model predicts the presence of an unknown eukaryotic phos-
phatase as part of the regulatory system. Thus our working
hypothesis is that the P. gingivalis SerB653 HAD-type phospha-
tase may be emulating a host enzyme to modulate the recruit-
ment of microtubules to host cell membranes. Consistent with
this, exogenous SerB653 recruits microtubules to the surface in
The role played by the SerB653 interaction with HSP90 remains
to be established. It is possible that modulation of HSP90 phos-
phorylation might change the subset of clients activated by the
chaperone. Alternatively, the interaction between these two pro-
teins might be driven by the heat-shock protein, which is known to
refold proteins that are insoluble or coaggregated (36).
In conclusion, analysis of P. gingivalis SerB653 has led to the
discovery of a previously undescribed role for a HAD family
phosphatase. SerB653 activity impacts microtubule dynamics
and is required for optimal invasion and intracellular survival of
the organism. The challenges now are to elucidate the mecha-
activity. Such investigations will provide new insights into the
multifactorial virulence mechanisms of P. gingivalis.
Materials and Methods
Bacterial Strains and Cell Culture. P. gingivalis strains were grown
anaerobically at 37°C in trypticase–soy medium supplemented
with hemin and menadione. Escherichia coli strains were grown
in LB media aerobically at 37°C. The E. coli strain DH5?
(Stratagene) was used as a host for cloning and plasmid purifi-
cation. E. coli strain TunerDE3 (Novagen) was used to purify the
His-tagged phosphatases. Primary cultures of GEC were ob-
tained from gingival explants and were maintained in culture as
obtained from D. Oda (University of Washington, Seattle, WA)
and cultured as described (38).
Plasmid Construction and Protein Purification. The phosphoserine
phosphatase ORFs (PG0653 and PG1170, www.tigr.org) were
PCR-amplified from genomic DNA of P. gingivalis 33277 by using
Pfu polymerase (Stratagene) and cloned into the plasmid pET30b
(Novagen), resulting in plasmids pET30b-653 and pET30b-1170.
These plasmids were sequenced to confirm the construct. A phos-
phatase-deficient mutant of the 653 enzyme was created by site-
directed mutagenesis, by using primers to alter the catalytic aspar-
tate 198 to asparagine. Proteins were purified by using a BioLogic
DuoFlow chromatography system loaded with an IMAC column
(Bio-Rad). Purity was ?90%, as determined by SDS?PAGE Coo-
by cloning the SerB653 wild-type gene in to the E. coli–Bacteroides
shuttle vector pTCow (39), and it was subsequently conjugated into
the PG0653::ermF mutant. (See Supporting Text and Fig. 8, which
are published as supporting information on the PNAS web site, for
Construction of Insertion and Allelic Exchange Mutants.AP.gingivalis
mutant in the serine uptake transporter sstT (17) was created by
cloning a 600-bp internal fragment of the gene into the P. gingivalis
suicide vector pVA3000 (40) and conjugation of the construct into
Allelic exchange mutants in PG0653 and PG1170 were created by
cloning ?2-kb gene fragments into pUC19, then inserting either
ermF (PG0653) or tetQ (PG1170) into the ORF. The resulting
constructs were digested with ScaI and the linear DNA electropo-
rated into P. gingivalis (41). Transformants were selected on eryth-
romycin or tetracycline and confirmed by PCR and Southern
hybridization. A double mutant in both 653 and 1170 was created
microtubule rearrangements. (A) Cells stained for ?-tubulin (red) and with
enriched tubulin at the cell surface. Controls contained enzyme buffer only.
tubulin rearrangements (not shown). (B) Magnification (?100) of cells at the
30-min time point.
Addition of exogenous His-SerB653 protein to HIGK cells induces
Tribble et al.PNAS ?
July 18, 2006 ?
vol. 103 ?
no. 29 ?