Brain angiogenesis inhibitor 1 (BAI1) is a pattern
and engulfment of Gram-negative bacteria
Soumita Dasa, Katherine A. Owenb, Kim T. Lyc, Daeho Parkc, Steven G. Blacka, Jeffrey M. Wilsona, Costi D. Sifria,
Kodi S. Ravichandranc, Peter B. Ernsta,c,1, and James E. Casanovab,c,1,2
Departments ofaMedicine,bCell Biology, andcMicrobiology, University of Virginia, Charlottesville, VA 22908
Edited* by Roy Curtiss, Arizona State University, Tempe, AZ, and approved December 27, 2010 (received for review October 1, 2010)
Bacterial recognition by host cells is essential for initiation of
infection and the host response. Bacteria interact with host cells
via multiple pattern recognition receptors that recognize microbial
products or pathogen-associated molecular patterns. In response
to this interaction, host cell signaling cascades are activated that
lead to inflammatory responses and/or phagocytic clearance of
attached bacteria. Brain angiogenesis inhibitor 1 (BAI1) is a re-
ceptor that recognizes apoptotic cells through its conserved type I
thrombospondin repeats and triggers their engulfment through
an ELMO1/Dock/Rac1 signaling module. Because thrombospondin
repeats in other proteins have been shown to bind bacterial
surface components, we hypothesized that BAI1 may also mediate
the recognition and clearance of pathogenic bacteria. We found
that preincubation of bacteria with recombinant soluble BAI1
ectodomain or knockdown of endogenous BAI1 in primary macro-
phages significantly reduced binding and internalization of the
Gram-negative pathogen Salmonella typhimurium. Conversely,
overexpression of BAI1 enhanced attachment and engulfment of
Salmonella in macrophages and in heterologous nonphagocytic
cells. Bacterial uptake is triggered by the BAI1-mediated activation
of Rac through an ELMO/Dock-dependent mechanism, and inhibi-
tion of the BAI1/ELMO1 interaction prevents both Rac activation
and bacterial uptake. Moreover, inhibition of ELMO1 or Rac func-
tion significantly impairs the proinflammatory response to infec-
tion. Finally, we show that BAI1 interacts with a variety of Gram-
negative, but not Gram-positive, bacteria through recognition of
their surface lipopolysaccharide. Together these findings identify
BAI1 as a pattern recognition receptor that mediates nonopsonic
phagocytosis of Gram-negative bacteria by macrophages and di-
rectly affects the host response to infection.
the innate immune response, in which pattern recognition
receptors (PRRs) play an important role (1). PRRs represent
a family of molecules that include Toll-like receptors (TLRs),
scavengerreceptors, C-type lectin receptors, andcytosolic sensors
such as NOD1 and NOD2 that allow cells to recognize and clear
unwanted particles or foreign molecules (2). For example, TLR2
recognizes the surface peptidoglycan (PG) of Gram-positive
bacteria, TLR4 recognizes the surface lipopolysaccharide (LPS)
of Gram-negative bacteria, while TLR5 recognizes conserved
determinants in bacterial flagellin (1). A second class of PRR
includes the scavenger receptor CD36, which recognizes a wide
range of proteoglycan and lipid-containing pathogen associated
molecular patterns (PAMPs) and also helps in the clearance of
apoptotic cells (3, 4). Although surface PRRs are important in
stimulating the host response to infection, the internalization of
bacteria may modify the host response further as the bacterial
PAMPs are displayed to intracellular sensing mechanisms.
BAI1 (brain-specific angiogenesis inhibitor 1) is a member of
the so-called adhesion-type family of 7-transmembrane receptors
ecognition of bacteria by host cells is essential for initiation of
(3). Its name derives from an initial observation that an extra-
cellular fragment of the receptor inhibited neovascularization
in an experimental brain tumor model (5). More recently it was
discovered that BAI1 is expressed on macrophages, where it acts
as a receptor for the clearance of apoptotic cells (6). A key feature
of BAI1 is the presence of five extracellular type 1 thrombo-
spondin repeats (TSRs) that bind to surface-exposed phosphati-
dylserine on apoptotic cells (6).
TSRs in other proteins have been shown to bind a variety of
bacterial products including LPS from Gram-negative bacteria
and PG from Gram-positive bacteria (7, 8). Importantly, the li-
gand binding specificities of different TSRs appear to vary;
whereas thrombospondin-1 (TSP1) interacts only with PG (8),
the single TSR in mindin/spondin-2 can also bind LPS and
lipoteichoic acid (LTA) (7). Although BAI1 contains five TSR
motifs, it is not known whether BAI1 recognizes bacterial prod-
ucts and, if so, what responses are triggered by this interaction.
between the BAI1 TSRs and bacterial surface LPS. Binding of
bacteria to BAI1 triggers activation of the Rho-family GTPase
Rac1 through an ELMO1/Dock/Rac signaling module, which
binds directly to the cytoplasmic domain of BAI1. Interestingly,
BAI1-mediated Rac activation is necessary not only for engulf-
ment of bound bacteria, but also for an efficient downstream
proinflammatory response. Together these observations suggest
that BAI1 functions not only as a receptor for apoptotic cells, but
also as a unique PRR for Gram-negative bacterial pathogens that
contributes directly to their internalization and the immunopa-
thogenesis of infection.
BAI1 Recognizes Bacteria Through Its TSRs. BAI1 has been shown
to recognize apoptotic cells by binding to exposed phosphati-
dylserine (6). To determine if BAI1 can also bind bacteria, we
examined its ability to recognize the Gram-negative enteric
pathogen Salmonella enterica serovar Typhimurium. For most of
our studies, we used a genetically engineered S. Typhimurium
strain (ΔinvG) that cannot actively invade host cells. Bacterial
attachment in the absence of internalization was measured in
murine J774 macrophages overexpressing BAI1, in which cells
were pretreated with cytochalasin D to block phagocytosis. As
Author contributions: S.D., K.A.O., K.T.L., P.B.E., and J.E.C. designed research; S.D., K.A.O.,
and K.T.L. performed research; D.P., S.G.B., J.M.W., C.D.S., and K.S.R. contributed new
reagents/analytic tools; S.D., K.A.O., K.S.R., P.B.E., and J.E.C. analyzed data; and S.D., K.S.R.,
P.B.E., and J.E.C. wrote the paper.
The authors declare no conflict of interest.
*This Direct Submission article had a prearranged editor.
1P.B.E. and J.E.C. contributed equally to this work.
2To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| February 1, 2011
| vol. 108
| no. 5www.pnas.org/cgi/doi/10.1073/pnas.1014775108
shown in Fig. 1A, cells overexpressing BAI1 bound 63% more
bacteria than controls (4.05 × 105cfu vs. 2.49 × 105; Fig. S1).
Because macrophages express other surface receptors that may
bind Gram-negative bacteria (e.g., scavenger receptor or TLRs)
(9), we assayed Salmonella binding to nonphagocytic fibroblastic
(i.e., CHO) cells (which do not express endogenous BAI1) in the
presence and absence of exogenous BAI1. As shown in Fig. 1B,
expression of BAI1 enhanced binding of S. Typhimurium by 2.3-
fold relative to mock-transfected cells. Taken together, these data
indicate that BAI1 can recognize determinants on the surface
of S. Typhimurium and mediate its binding to macrophages and
TSRs in other proteins have been shown to bind a variety of
bacterial products including LPS from Gram-negative bacteria
and PG and LTA from Gram-positive bacteria (7, 8). Impor-
tantly, the ligand binding specificities of different TSRs appear to
vary; whereas TSP1 interacts only with PG (8), the single TSR in
mindin/spondin-2 can bind PG, LTA, and LPS (7). As BAI1
contains five TSR motifs, we tested whether BAI1 recognizes
bacterial products through interactions with its TSRs. Previous
work has shown that binding of apoptotic cells to cultured
macrophages can be inhibited by preincubation with a soluble
BAI1 ectodomain fragment, which blocks binding to endogenous
receptors (6); this fragment contains the N-terminal arginyl-
glycyl-aspartic acid (RGD) motif and the five TSRs (RGD-TSR;
Fig. S2). To determine if this holds true for bacteria, S. Typhi-
murium (ΔinvG) was preincubated with GST alone, GST-RGD-
TSR, or a GST fusion containing the RGD motif but lacking the
five TSRs (RGD-ΔTSR). Bacterial binding to macrophages was
then measured as described earlier in the presence of cytocha-
lasin D. As shown in Fig. 1 C and D, preincubation with GST-
RGD-TSR inhibited Salmonella binding to J774 macrophages
(50.5 ± 16%) and primary bone marrow-derived macrophages
(BMDMs; 67 ± 5%) relative to GST alone. Importantly, GST-
RGD-ΔTSR had no inhibitory effect, indicating that inhibition
required the presence of the TSRs. A comparable level of in-
hibition was observed in macrophages freshly isolated from
murine small intestine (10), demonstrating the physiological
relevance of this interaction (Fig. 1E).
To confirm that the observed inhibition was caused by com-
petition for binding to endogenous BAI1, BMDMs were de-
pleted of BAI1 by using siRNA. As shown in Fig. 1G, BAI1
knockdown reduced surface binding of S. Typhimurium by 45 ±
5.8%. In contrast, knockdown of ELMO1, which is necessary for
BAI1-mediated engulfment (6), had no effect on surface binding.
Binding of Bacteria to BAI1 Triggers Engulfment. To assay bacterial
internalization, we used a standard gentamicin protection assay.
Briefly, cells were exposed to bacteria for 1 h at 37 °C, washed,
and incubated for an additional 90 min in the presence of the
membrane-impermeable antibiotic gentamicin. This treatment
kills extracellular bacteria, but intracellular bacteria remain vi-
able and are quantified as described earlier by measuring colony-
forming units in cell lysates. As shown in Fig. 2A, overexpression
of BAI1 in J774 cells increased internalization of the noninvasive
Salmonella invG mutant strain. Similarly, expression of BAI1 in
CHO cells resulted in a more than fourfold increase in bacterial
internalization, relative to vector controls (Fig. 2B). Together
these findings indicate that BAI1 binds S. Typhimurium at the
cell surface and mediates internalization of the bound bacteria,
even in heterologous, nonphagocytic cells.
Preincubation of noninvasive S. Typhimurium with GST-
RGD-TSR reduced internalization by 61 ± 16% in J774 mac-
rophages, whereas incubation with GST alone or GST-RGD-
ΔTSR had no effect (Fig. 2C). A corresponding inhibition was
observed in BMDMs depleted of endogenous BAI1 by siRNA
(Fig. 2D). Importantly, knockdown of ELMO1, which did not
inhibit bacterial binding to the cell surface (Fig. 1G), did inhibit
bacterial uptake, consistent with its known role in coupling BAI1
to the phagocytic machinery (6). Surprisingly, although an iso-
genic, invasive Salmonella strain (SL1344) bound BMDMs more
efficiently than the noninvasive strain, knockdown of BAI1 or
D D TSR
Δ Δ TSR
incubated with a noninvasive Salmonella Typhimurium strain (ΔinvG) in the presence of cytochalasin D to measure adhesion in the absence of internalization.
Numbers of bound bacteria (in cfu) were determined as described in Materials and Methods. (B) CHO cells were transfected with empty vector or a plasmid
expressing untagged full-length BAI1. Bacterial attachment was assayed as in A. (C–E) Inhibition of Salmonella attachment by the recombinant BAI1 ecto-
domain. Bacteria were preincubated for 15 min with 10 ng/μL GST alone, GST fused to an N-terminal BAI1 fragment containing the RGD motif and all five TSRs
(GST-RGD-TSR), or a similar fusion lacking the TSRs (GST-RGD-ΔTSR). Attachment to J774 cells (C), BMDMs (D), and intestinal macrophages (E) was assayed in
the presence of cytochalasin D as described in Materials and Methods. (F and G) Down-regulation of BAI1 inhibits bacterial attachment. BMDMs were de-
pleted of endogenous BAI1 or ELMO1 by using siRNA, and knockdown efficiency was determined by RT-PCR (F). Attachment of S. Typhimurium was assayed
as described earlier (G). In B–G, data represent the means (±SD) of three independent experiments. In A, cfu values (mean ± SD) indicated are from a rep-
resentative experiment with triplicate wells (*P ≤ 0.05).
Bacterial recognition mediated by the TSR domains of BAI1. (A) J774 macrophages expressing exogenous BA1 or transfected with empty vector were
Das et al.PNAS
| February 1, 2011
| vol. 108
| no. 5
ELMO1 reduced internalization efficiency to a similar extent
(Fig. S3). Together, these findings demonstrate that BAI1 not
only recognizes bacteria at the cell surface, but that engagement
of BAI1 triggers bacterial internalization.
BAI1-Mediated Bacterial Internalization Requires Rac1 Activation
Through an ELMO1-Dependent Mechanism. The ELMO1/Dock180
complexacts asa bipartite guanine nucleotideexchange factor for
theRho-familyGTPaseRac1,whichcoordinates theformation of
membranous pseudopods that drive particle internalization dur-
ing phagocytosis (11). Previous work showed that ELMO1 binds
to a conserved helical region in the cytoplasmic domain of BAI1,
and that ligation of BAI1 by apoptotic cells triggers the activation
of Rac1 in an ELMO1- and Dock180-dependent manner (6).
Mutation of three charged residues within this α-helix (RKR-
AAA) significantly reduces the binding of ELMO1 to BAI1 (6).
As shown in Fig. 2D, knockdown of ELMO1 attenuated bacterial
uptake by macrophages to the same extent as BAI1 knockdown,
suggesting that BAI1 and ELMO1 are also functionally linked
during bacterial internalization. To test this hypothesis, we
expressed WT BAI1 and mutant BAI1 (RKR-AAA) to compa-
rable levels (Fig. S4) in CHO cells and measured Rac1 activity
by using a pull-down assay, following addition of noninvasive
S. Typhimurium to the culture. As shown in Fig. 3A, addition of
bacteria to control CHO cells transfected with empty vector in-
duced a small but detectable activation of Rac1 during a 30-min
incubation. In cells expressing WT BAI1, the basal level of Rac1
activation was higher, but was dramatically increased upon ad-
dition of bacteria. In contrast, no activation of Rac1 was detected
in cells expressing mutant BAI1 (RKR-AAA) that cannot couple
to ELMO1. Rather, the mutant appeared to act as a dominant
negative, preventing even the low level of activation observed in
control cells. Importantly, the failure of this mutant to activate
Rac1 correlated with impaired internalization of bacteria (Fig.
3B). These data support the notion that the binding of bacteria
to BAI1 triggers their internalization through a mechanism in-
volving the activation of Rac1 by the ELMO/Dock180 complex.
Bacterial Interaction with BAI1 Triggers Proinflammatory Responses.
The clearance of apoptotic cells is typically noninflammatory,
and in fact often results in the production of anti-inflammatory
molecules such as TGF-β1 (12, 13) or IL-10 (14). In contrast,
recognition of bacterial PAMPs by TLRs or other PRRs typically
results in proinflammatory signaling, which is an important factor
in bacterial clearance from infected tissue. To determine if
bacterial recognition by BAI1 induces a proinflammatory re-
sponse, we measured the production of a major proinflammatory
cytokine, TNF-α, in BMDMs. As expected, incubation of non-
invasive Salmonella (ΔinvG) with control BMDMs led to a robust
induction of TNF-α (Fig. 3C). In contrast, BAI1-mediated in-
ternalization of the bacteria induced only a small increase in the
release of IL-10 (Fig. S5). The mean value for IL-10 induced by
infection was 132 pg/mL, compared with more than 3,000 pg/mL
reported in studies following the engulfment of apoptotic cells
(14). Remarkably, knockdown of BAI1 in BMDM cells reduced
the level of TNF-α release nearly 50% and the knockdown of
ELMO1 expression decreased TNF-α release even further. A
similar reduction of TNF-α was observed when BMDMs de-
pleted of BAI1 or ELMO1 were infected with invasive Salmo-
nella (SL1344; Fig. S6).
BAI1 Preferentially Recognizes Gram-Negative Bacteria.As described
earlier, the TSRs in other proteins exhibit distinct ligand specif-
icities: whereas TSP1 interacts only with the surface PG of Gram-
positive bacteria (8), the single TSR in mindin/spondin-2 can bind
PG and LTA of Gram-positive bacteria and the LPS of Gram-
negative species (7). To assess the binding preferences for BAI1,
CHO fibroblasts were engineered to stably express an epitope-
tagged form of the receptor. These cells were then incubated with
a number of different Gram-negative [S. Typhimurium, Escheri-
chia coli (DH5α), and Campylobacter jejuni] and Gram-positive
(Staphylococcus aureus, Streptococcus pneumoniae, and group A
Streptococcus) bacteria. For this assay, FITC-labeled bacteria
were incubated with parental CHO cells or their BAI1-expressing
derivatives, and bacterial binding was assessed by flow cytometry.
As shown in Fig. 4A, binding of all three Gram-negative species
was significantly enhanced in the presence of BAI1. In contrast,
none of the Gram-positive species exhibited binding greater than
The BAI1 TSRs Interact Directly with LPS. The most abundant
component of the outer membranes of Gram-negative bacteria is
LPS. To determine if the BAI1 TSRs can recognize LPS, we used
a solid-phase assay in which GST, GST-RGD-TSR, or GST-
RGD-ΔTSR were spotted onto nitrocellulose filters. These filters
were then incubated with LPS from different sources, washed,
and immunoblotted with antibody to LPS. As shown in Fig. 4B,
the RGD-TSR construct can bind LPS from S. enterica serovar
abortus equi and E. coli 055:B5 whereas the construct lacking the
TSRs cannot. To confirm the ability of BAI1 to recognize LPS in
intact cells, CHO cells were incubated with biotinylated ultra-
pure LPS (from E. coli 0111:B4) and surface binding was assayed
by flow cytometry. Fig. 4C shows that cells expressing BAI1
bound significantly more LPS than nontransfected controls.
LPS consists of three parts: the membrane-anchored lipid A
(which is recognized by TLR4), a positively charged “core” oli-
gosaccharide region, and a highly variable terminal oligosac-
charide chain referred to as O-antigen (Fig. S7). Our observation
that E. coli DH5α (which lacks O-antigen) and S. typhimurium
SL1344 (which expresses O-antigen) showed comparable binding
to BAI1 suggested that O-antigen is not necessary for this in-
D D TSR
Control BAI1 ELMO1
in CHO cells (B) expressing exogenous BAI1. Cells were incubated with S. Typhimurium (ΔinvG) for 1 h at 37 °C in the absence of cytochalasin D and in-
ternalization measured as described in Materials and Methods. (C) Competitive inhibition of bacterial internalization by the BAI1 ectodomain. S. Typhi-
murium (ΔinvG) was preincubated with GST alone, GST-RGD-ΔTSR, or GST-RGD-TSR, then incubated with J774 cells for 1 h at 37 °C. (D) Knockdown of BAI1 or
ELMO1 inhibits bacterial internalization. BMDMs were depleted of endogenous BAI1 or ELMO1 as described in Fig. 1. Internalization of S. Typhimurium
(ΔinvG) was assayed as described earlier. In A–D, data represent the mean ± SD of three separate experiments (*P ≤ 0.05).
BAI1 promotes bacterial internalization. (A and B) Bacterial internalization was measured by using the gentamicin protection assay in J774 cells (A) or
| www.pnas.org/cgi/doi/10.1073/pnas.1014775108 Das et al.
teraction. To determine if BAI1 recognizes the charged core
oligosaccharide, we tested the binding of recombinant BAI1
ectodomain to LPS variants (Fig. S7) containing the core [E. coli
055:B5 and Salmonella TV119 (Ra)] or lacking the core (Sal-
monella Minnesota Re595). For this assay, the LPS variants were
spotted onto membranes and probed with recombinant BAI1
ectodomain. As shown in Fig. 4D, BAI1 bound efficiently to LPS
species that contain the core oligosaccharide (Ra), but not to
a form that lacks it (Re). Moreover, BAI1 failed to bind other
bacterial products such as flagellin or the Gram-positive mem-
brane component PG. In contrast, TSP1, which has three TSRs,
bound PG but not LPS, as described previously by Rennemeier
et al. (8). Together, these findings suggest that BAI1 mediates
the binding of Gram-negative bacteria to macrophages through
a direct interaction of the core oligosaccharides of LPS with the
Finally, LPS is well known to promote proinflammatory sig-
naling through interactions with TLR4. To determine the extent
to which BAI1 may contribute to the overall inflammatory re-
sponse, we measured TNF-α release from control BMDMs or
cells depleted of endogenous BAI1 by RNAi. Surprisingly, we
found that down-regulation of BAI1 significantly attenuated
TNF-α production in response to LPS treatment (Fig. 4E). These
data suggest that BAI1 expression is required for efficient
proinflammatory signaling, and that it may cooperate with TLR4
in the response to LPS.
BAI1was previously shown tofunction asa receptorforapoptotic
cells, where it binds surface phosphatidylserine through its type I
TSRs and triggers their engulfment via an ELMO1/Dock180/
Rac1-mediated signaling pathway (6). Here we identify a unique
function for BAI1 in the binding and internalization of Gram-
negative bacteria through recognition of bacterial LPS by the
TSRs are conserved domains found in 41 human proteins,
many of which are involved in cell adhesion, migration, com-
munication, and tissue remodeling. The approximately 60 aa
TSR domains are comprised of an elongated, three-stranded
β-sheet (15). One face of the folded structure contains a helical
groove that is rich in positively charged side chains, and is
thought to represent the recognition face of the molecule. The
density and distribution of these positive residues differs widely
among TSRs from different proteins, and may have a significant
impact on ligand specificity. As noted earlier, the single TSR in
mindin/spondin-2 can bind a variety of ligands including LPS,
LTA, and PG, whereas the TSRs of TSP1 appears to be selective
for PG (7, 8) (Fig. 4D). The five BAI1 TSR domains vary sig-
nificantly in surface charge distribution; the domain pI values
vary from 5.2 to 8.7, suggesting that the different TSRs may
recognize distinct ligands. Alternatively, two or more of the
TSRs may act cooperatively to enhance binding affinity for
multivalent ligands such as bacterial membranes. Our results
indicate that the BAI1 TSRs bind Gram-negative, but not Gram-
positive, organisms, and they are therefore unlikely to recognize
LTA or PG, which are abundant in Gram-positive membranes.
In fact, a direct assay for PG binding showed that TSP1 bound
PG whereas BAI1 could not. Whether all five TSRs are impor-
tant for BAI1 interaction with LPS, or any differential affinity for
ligands exists among the TSRs, remain to be elucidated.
LPS is an important virulence factor for many Gram-negative
pathogens (16). It comprises a lipid A region, which inserts into
the membrane and contains five to seven acyl chains, depending
on the organism. This is linked to an inner core complex of eight
to 12 sugars, which is in turn linked to O-antigen, an oligosac-
charide chain of variable length and composition. It is the
composition of O-antigen that defines bacterial serotypes. TLR4
recognizes the lipid A moiety of LPS, in complex with the core-
ceptors CD14 and MD2 (17). In contrast, mindin/spondin-2 does
not bind lipid A, but instead appears to recognize carbohydrate,
as binding can be competed with simple sugars (e.g., mannose,
glucose). We found that an E. coli strain (DH5α) lacking O-
antigen in its LPS interacts with BAI1 in a cell-binding assay
(Fig. 4A), suggesting that O-antigen is not a determinant of BAI1
binding. Moreover, we found that purified LPS containing the
charged core oligosaccharide, but not O-antigen (Ra), bound BAI1
1 1.6 2.2 2.6 3.7 4.5 1.1 0.9 0.85
pull-down assay after incubation with S. Typhimurium (ΔinvG) for 0 min, 15 min, and 30 min. Pull-downs (Upper) and a fraction of the total lysates (Lower)
were blotted for Rac1. The ratio of GTP-bound active Rac1 to total Rac1 is expressed relative to uninfected control cells (control value is set to 1). Data are
representative of results from three experiments and the summary graph was plotted from the ratio of GTP-bound active Rac1 to total Rac1 of three in-
dependent experiments. (B) Cells expressing FLAG-BAI1 or FLAG-BAI1 (RKR-AAA) were incubated with S. Typhimurium (ΔinvG) for 1 h at 37 °C and in-
ternalization was determined by using the gentamicin protection assay. (C) Knockdown of either BAI1 or ELMO1 impairs proinflammatory signaling in
response to bacterial infection. TNF-α production was measured by ELISA in supernatants collected from control, BAI1-depleted, or ELMO1-depleted BMDMs
after 6 h of incubation with S. Typhimurium (ΔinvG). In B and C, data represent the mean ± SD of triplicate wells for each condition from three independent
experiments (*P ≤ 0.05).
Bacterial binding to BAI1 triggers Rac1 activation and bacterial internalization in an ELMO-dependent manner. (A) Rac1 activation was measured by
Das et al. PNAS
| February 1, 2011
| vol. 108
| no. 5
efficiently whereas LPS lacking the core (Re) did not (Fig. 4D).
This observation indicates that BAI1 interacts with a region of LPS
distinct from the lipid A moiety that is recognized by TLR4.
Whether BAI1 can recognize other carbohydrate-rich structures
such as fungal glycans remains to be determined.
carbohydrate-based determinants, including multiple scavenger
receptors (3) and the C-type lectin receptors (18). It is therefore
not surprising that knockdown of BAI1 in macrophages did not
completely abrogate bacterial uptake. However, our finding that
Salmonella internalization was reduced by 50% in BAI1-depleted
macrophages suggests that it is a quantitatively significant con-
tributor to bacterial clearance.
The decrease in TNF-α production we observed in BAI1-
depleted macrophages after bacterial infection or LPS treatment
signaling responses. TLR4 is also expressed in macrophages, and
LPS signaling is known to require TLR4. However, it is also known
that several phagocytic PRRs exhibit significant crosstalk with
TLRs and can amplify or otherwise modify their signaling output
(19, 20). It therefore seems likely that signals emanating from
ligated BAI1 intersect with those derived from TLR4 and that
BAI1 ligation positively regulates TLR4-mediated signaling. How-
ever, the mechanisms of this crosstalk remain to be defined.
In summary, we have identified a PRR, BAI1, that contributes
to the uptake of Gram-negative bacterial pathogens by macro-
phages. Bacterial recognition occurs via a direct interaction
between the BAI1 TSRs and the core oligosaccharide of bacte-
rial LPS, and results in both bacterial engulfment and a proin-
flammatory signaling response. As LPS is an important virulence
factor for Gram-negative pathogens, a more thorough knowl-
edge of these aspects of microbial recognition may lead to new
therapies that limit the survival of pathogenic bacteria and the
tissue damage induced by local inflammatory responses.
Materials and Methods
Bacterial Strains, Cell Culture, Transfection, and Plasmids. S. enterica serovar
Typhimurium (ΔinvG), a noninvasive isogenic mutant of WT strain SJW1103,
was used to measure bacterial binding and internalization (21). In some
experiments, the invasive S. enterica serovar Typhimurium strain SL1344 was
also used. SI Materials and Methods includes further details of bacterial
culture and cell maintenance.
Antibodies and Reagents. Antibodies used include the following: anti-LPS
(Abcam), anti-Rac1 monoclonal antibody (Upstate Biotechnology), anti-FLAG
(Sigma), anti-Tubulin (Abcam), HRP-conjugated anti-GST (GE Healthcare),
antithrombospondin (Abcam), and streptavidin-HRP high sensitivity (Pierce).
LPS from different bacterial species and S. aureus PG were purchased from
Sigma-Aldrich and biotinylated ultrapure E. coli O111:B4 LPS and flagellin
from Salmonella Typhimurium were obtained from InvivoGen. Recombinant
TSP1 was purchased from Cell Sciences. Cytochalasin D (Sigma-Aldrich) was
used at a concentration of 1 μg/mL to inhibit bacterial uptake by host cells.
% of Max
RGD Δ TSR
Index of Attachment
Salmonella E. coli C. jejuni Staph
Ra mutant LPS
Re mutant LPS
E. coli LPS
positive bacteria. (A) Equivalent numbers of Salmonella
Typhimurium (ΔinvG), E. coli (DH5α), C. jejuni, S. aureus, S.
pneumoniae, and group A Streptococcus were labeled with
FITC and incubated at a multiplicity of infection (MOI) of
100 with vector control CHO cells (open bars) or CHO cells
expressing FLAG-BAI1 (closed bars) for 1 h in the presence
of cytochalasin D to prevent internalization. Binding of
bacteria was analyzed by flow cytometry. For each bacte-
rial species, an index of attachment was determined
whereby background binding to non–BAI1-expressing CHO
cells was arbitrarily set to a value of 1. Data represent the
mean ± SD of three independent experiments. (B–D) The
BAI1 TSRs bind directly to LPS. (B) Purified, recombinant
GST (control), GST-RGD-ΔTSR, or GST-RGD-TSR (0.5 μg each)
were spotted onto nitrocellulose filters. After blocking,
filters were incubated with LPS from the indicated source
and blotted with anti-LPS antibody. (C) CHO cells contain-
ing empty vector or expressing FLAG-BAI1 were incubated
with biotin labeled LPS for 30 min at 37 °C. Cells were then
washed, incubated with streptavidin–phycoerythrin, and
analyzed by flow cytometry. (D) LPS from E. coli 055:B5,
Salmonella enterica serotype Typhimurium TV119 (Ra mu-
tant) and Salmonella enterica serotype Minnesota Re 595
(Re mutant), S. aureus PG, and Salmonella Typhimurium
flagellin were spotted on nylon membranes, blocked, and
incubated with GST-RGD-TSR (Left) or TSP1 (Right) fol-
lowed by HRP-conjugated anti-GST or anti-TSP1 antibody.
(E) Knockdown of BAI1 impairs proinflammatory signaling
after LPS treatment. TNF-α production was measured by
ELISA in supernatants collected from control or BAI1-de-
pleted BMDMs after 6 h of Salmonella Typhimurium LPS (5
μg/mL) treatment. Data represent the mean ± SD of du-
plicate wells for each condition from three independent
experiments (*P ≤ 0.05).
BAI1 recognizes Gram-negative but not Gram-
| www.pnas.org/cgi/doi/10.1073/pnas.1014775108Das et al.
RNA Preparation and RT-PCR. Total RNA was extracted using the RNeasy kit Download full-text
(Qiagen) and reverse transcribed by using the SuperScript kit (Invitrogen),
both according to the manufacturers’ instructions. Primer sequences are
included in SI Materials and Methods.
Preparation of Recombinant RGD-TSR. The RGD-TSR region of the BAI1
ectodomain (residues 202–585) was subcloned into the pGEX-4T2 vector and
purified as described previously (6).
Peptide and Inhibitor Incubation. Where indicated, GST-RGD-TSR, GST-RGD-
ΔTSR, or GST alone were preincubated with bacteria for 15 min at a con-
centration of 10 ng/μL before being added to cells in antibiotic-free DMEM
in a 37 °C CO2incubator.
BMDM Preparation. Primary BMDMs were derived from the femurs and tibia
of mice by using a modification of techniques described previously (10).
Detailed methods are provided in SI Materials and Methods.
Preparation of Intestinal Macrophages. Gut antigen-presenting cells were
isolated using techniques described previously (10, 22). Detailed methods are
provided in SI Materials and Methods.
siRNA Transfections. Nucleofection was used to introduce BAI1 or ELMO1
siRNA (ON-Target Plus SMART pool; Dharmacon) into BMDM cells using the
Amaxa mouse macrophage nucleofector kit (Lonza) using program Y-001.
After 48 h, RNA was prepared to monitor the level of BAI1 and ELMO1
expression by RT-PCR.
Gentamicin Protection Assay. Quantification of intracellular bacteria was
done by using the gentamicin protection assay as described previously (23).
Detailed methods are described in SI Materials and Methods.
Bacterial Attachment Assays. Infection of bacteria was performed as described
for the gentamicin protection assay but in the presence of 1 μg/mL cyto-
chalasin D to block internalization. After incubation for 1 h, cells were
washed to remove unbound bacteria and lysed in 1% Triton, and remaining
surface-bound bacteria were quantified by plating of cell lysates on Luria–
Bertani (LB) agar plates.
To measurebacterial attachment byflowcytometry, bacteria were labeled
with FITC dye (Sigma-Aldrich) as previously described (24). Cells were in-
cubated with labeled bacteria for 1 h at room temperature, washed, fixed
with 1% paraformaldehyde, and analyzed by flow cytometry by using a
Becton Dickinson FACSCalibur dual laser instrument as described (8).
Binding of LPS to BAI1. Biotinylated ultrapure E. coli O111:B4 LPS (5 μg;
InvivoGen) was incubated with 106cells in DMEM for 30 min at 37 °C. Cells
were then washed and stained with streptavidin–phycoerythrin before flow
Dot Blot. Purified control GST, GST-RGD-ΔTSR, and GST-RGD-TSR (0.5 μg)
were spotted onto nitrocellulose membrane and air-dried. Membranes
were blocked with 3% BSA and incubated with LPS at 1 μg/mL overnight at
4 °C. Membranes were then probed with anti-LPS antibody in 3% BSA fol-
lowed by HRP-conjugated secondary antibody, washed, and incubated in
In a similar assay, 0.5 μg of LPS from smooth (E. coli) and rough variants
(Ra and Re mutants), PG (from S. aureus), or flagellin (from S. typhimurium)
were spotted on nylon membrane, blocked, and incubated with 1 μg/mL of
GST-RGD-TSR or recombinant TSP1 overnight at 4 °C. After washing, mem-
branes were probed with HRP-conjugated anti-GST antibody (1:500) or anti-
TSP1 antibody followed by HRP-conjugated anti-mouse IgG and developed
by using Millipore detection reagent.
Western Blot. Cell lysates were separated by SDS/PAGE, transferred to a ni-
trocellulose membrane (Bio-Rad), and blocked in Tris-buffered saline solution
with Tween (20 mM Tris-HCl/136 mM NaCl, pH 7.5, with 0.05% Tween 20)
containing 5% nonfat dry milk. The binding of primary antibody was
detected with HRP-conjugated secondary antibody (Amersham Biosciences).
Cytokine Measurement. Supernatants were collected from BMDMs infected
with bacteria for the indicated times. Mouse TNF-α was measured by using an
ELISA kit (BD Pharmingen) according to the manufacturer’s instructions.
Assessment of Rac1 Activation. Rac1 activity was measured by a pull-down
assay by using GST-PBD (p21-binding domain of Pak1) beads as described
Statistical Analysis. Results are expressed as mean ± SD and were compared
by using two-tailed Student t test; differences were considered significant if
P values were lower than 0.05.
ACKNOWLEDGMENTS. We thank Joanne Lannigan and Michael Solga of the
University of Virginia Flow Cytometry Core Facility for expert technical
assistance in cell sorting and William Ross for flow cytometry. Research was
supported by National Institutes of Health Grants DK058536 (to J.E.C.);
AI08600, DK84063, and AI070491 (to P.E.); and GM64709 (to K.S.R.).
1. Iwasaki A, Medzhitov R (2004) Toll-like receptor control of the adaptive immune
responses. Nat Immunol 5:987–995.
2. Rakoff-Nahoum S, Medzhitov R (2008) Innate immune recognition of the indigenous
microbial flora. Mucosal Immunol 1(suppl 1):S10–S14.
3. Silverstein RL, Febbraio M (2009) CD36, a scavenger receptor involved in immunity,
metabolism, angiogenesis, and behavior. Sci Signal 2:re3.
4. Baranova IN, et al. (2008) Role of human CD36 in bacterial recognition, phagocytosis,
and pathogen-induced JNK-mediated signaling. J Immunol 181:7147–7156.
5. Nishimori H, et al. (1997) A novel brain-specific p53-target gene, BAI1, containing
thrombospondin type 1 repeats inhibits experimental angiogenesis. Oncogene 15:
6. Park D, et al. (2007) BAI1 is an engulfment receptor for apoptotic cells upstream of
the ELMO/Dock180/Rac module. Nature 450:430–434.
7. He YW, et al. (2004) The extracellular matrix protein mindin is a pattern-recognition
molecule for microbial pathogens. Nat Immunol 5:88–97.
8. Rennemeier C, et al. (2007) Thrombospondin-1 promotes cellular adherence of gram-
positive pathogens via recognition of peptidoglycan. FASEB J 21:3118–3132.
9. Taylor PR, et al. (2005) Macrophage receptors and immune recognition. Annu Rev
10. Wilson JM, et al. (2009) The A2Badenosine receptor impairs the maturation and
immunogenicity of dendritic cells. J Immunol 182:4616–4623.
11. Caron E, Hall A (1998) Identification of two distinct mechanisms of phagocytosis
controlled by different Rho GTPases. Science 282:1717–1721.
12. Xiao YQ, et al. (2008) Transcriptional and translational regulation of TGF-β production
in response to apoptotic cells. J Immunol 181:3575–3585.
13. Otsuka M, Negishi Y, Aramaki Y (2007) Involvement of phosphatidylinositol-3-kinase
and ERK pathways in the production of TGF-beta1 by macrophages treated with
liposomes composed of phosphatidylserine. FEBS Lett 581:325–330.
14. Voll RE, et al. (1997) Immunosuppressive effects of apoptotic cells. Nature 390:
15. Tan K, et al. (2002) Crystal structure of the TSP-1 type 1 repeats: A novel layered fold
and its biological implication. J Cell Biol 159:373–382.
16. Raetz CR, Whitfield C (2002) Lipopolysaccharide endotoxins. Annu Rev Biochem 71:
17. Park BS, et al. (2009) The structural basis of lipopolysaccharide recognition by the
TLR4-MD-2 complex. Nature 458:1191–1195.
18. Hollmig ST, Ariizumi K, Cruz PD, Jr. (2009) Recognition of non-self-polysaccharides by
C-type lectin receptors dectin-1 and dectin-2. Glycobiology 19:568–575.
19. O’Neill LA (2008) When signaling pathways collide: Positive and negative regulation
of toll-like receptor signal transduction. Immunity 29:12–20.
20. Lee MS, Kim YJ (2007) Signaling pathways downstream of pattern-recognition
receptors and their cross talk. Annu Rev Biochem 76:447–480.
21. Crago AM, Koronakis V (1998) Salmonella InvG forms a ring-like multimer that
requires the InvH lipoprotein for outer membrane localization. Mol Microbiol 30:
22. Denning TL, Wang YC, Patel SR, Williams IR, Pulendran B (2007) Lamina propria
macrophages and dendritic cells differentially induce regulatory and interleukin 17-
producing T cell responses. Nat Immunol 8:1086–1094.
23. Criss AK, Ahlgren DM, Jou TS, McCormick BA, Casanova JE (2001) The GTPase Rac1
selectively regulates Salmonella invasion at the apical plasma membrane of polarized
epithelial cells. J Cell Sci 114:1331–1341.
24. Smith LM, Laganas V, Pistole TG (1998) Attachment of group B streptococci to
macrophages is mediated by a 21-kDa protein. FEMS Immunol Med Microbiol 20:
25. Criss AK, Casanova JE (2003) Coordinate regulation of Salmonella enterica serovar
Typhimurium invasion of epithelial cells by the Arp2/3 complex and Rho GTPases.
Infect Immun 71:2885–2891.
Das et al.PNAS
| February 1, 2011
| vol. 108
| no. 5