Oxidation-specific epitopes are dominant targets of innate natural antibodies in mice and humans

Article (PDF Available)inThe Journal of clinical investigation 119(5):1335-49 · May 2009with39 Reads
DOI: 10.1172/JCI36800 · Source: PubMed
Abstract
Atherosclerosis is a chronic inflammatory disease characterized by the accumulation of oxidized lipoproteins and apoptotic cells. Adaptive immune responses to various oxidation-specific epitopes play an important role in atherogenesis. However, accumulating evidence suggests that these epitopes are also recognized by innate receptors, such as scavenger receptors on macrophages, and plasma proteins, such as C-reactive protein (CRP). Here, we provide multiple lines of evidence that oxidation-specific epitopes constitute a dominant, previously unrecognized target of natural Abs (NAbs) in both mice and humans. Using reconstituted mice expressing solely IgM NAbs, we have shown that approximately 30% of all NAbs bound to model oxidation-specific epitopes, as well as to atherosclerotic lesions and apoptotic cells. Because oxidative processes are ubiquitous, we hypothesized that these epitopes exert selective pressure to expand NAbs, which in turn play an important role in mediating homeostatic functions consequent to inflammation and cell death, as demonstrated by their ability to facilitate apoptotic cell clearance. These findings provide novel insights into the functions of NAbs in mediating host homeostasis and into their roles in health and diseases, such as chronic inflammatory diseases and atherosclerosis.
Research article
The Journal of Clinical Investigation http://www.jci.org  Volume 119    Number 5  May 2009  1335
Oxidation-specific epitopes are dominant
targets of innate natural antibodies
in mice and humans
Meng-Yun Chou,
1
Linda Fogelstrand,
1
Karsten Hartvigsen,
1
Lotte F. Hansen,
1
Douglas Woelkers,
2
Peter X. Shaw,
1
Jeomil Choi,
1,3
Thomas Perkmann,
4,5
Fredrik Bäckhed,
6
Yury I. Miller,
1
Sohvi Hörkkö,
1
Maripat Corr,
1
Joseph L. Witztum,
1
and Christoph J. Binder
1,4,5
1
Department of Medicine and
2
Department of Reproductive Medicine, UCSD, La Jolla, California, USA.
3
Department of Periodontology,
Pusan National University, Pusan, Republic of Korea.
4
Center for Molecular Medicine (CeMM) of the Austrian Academy of Sciences and
5
Department of Medical and Chemical Laboratory Diagnostics, Medical University of Vienna, Vienna, Austria.
6
Sahlgrenska Centre for Cardiovascular and Metabolic Research, Wallenberg Laboratory, University of Göteborg, Göteborg, Sweden.
Atherosclerosis is a chronic inflammatory disease characterized by the accumulation of oxidized lipoproteins
and apoptotic cells. Adaptive immune responses to various oxidation-specific epitopes play an important role
in atherogenesis. However, accumulating evidence suggests that these epitopes are also recognized by innate
receptors, such as scavenger receptors on macrophages, and plasma proteins, such as C-reactive protein (CRP).
Here, we provide multiple lines of evidence that oxidation-specific epitopes constitute a dominant, previously
unrecognized target of natural Abs (NAbs) in both mice and humans. Using reconstituted mice expressing
solely IgM NAbs, we have shown that approximately 30% of all NAbs bound to model oxidation-specific epi-
topes, as well as to atherosclerotic lesions and apoptotic cells. Because oxidative processes are ubiquitous, we
hypothesized that these epitopes exert selective pressure to expand NAbs, which in turn play an important
role in mediating homeostatic functions consequent to inflammation and cell death, as demonstrated by their
ability to facilitate apoptotic cell clearance. These findings provide novel insights into the functions of NAbs in
mediating host homeostasis and into their roles in health and diseases, such as chronic inflammatory diseases
and atherosclerosis.
Introduction
Althoughhypercholesterolemiais necessaryfor the initiation 
and progressionof atherosclerosis, there is now abundantevi
-
dence that immune mechanisms are also central to all phases of 
lesion development (1–3). We and others have documented that, 
among several suggested immunogens present in the atheroscle
-
rotic plaque, oxidation-specific epitopes, as occur in oxidized LDL 
(OxLDL), are immunodominant. In turn, these lead to profound 
immune responses, including autoantibody generation, that mod
-
ulate lesion formation (4).
Many of these responses are adaptive in nature, responding to 
the myriad of new moieties generated in response to the complex 
neoepitopes formed when lipid peroxidation occurs. Surprisingly, 
innate immune recognition of these oxidation-specific epitopes 
is also prominent, and variousmacrophage scavengerreceptors 
bind to epitopes of OxLDL (5–7). In addition, we previously found 
that innate natural Abs (NAbs)bindto oxidizedphospholipids 
(OxPLs) of OxLDL. For example, cholesterol-fed apoE-deficient 
mice havevery highIgM titers toOxLDL, which enabled clon
-
ing of IgM-secreting hybridomas from the spleens of these mice 
with specificity forOxLDL(8). Alarge numberofthese bound 
to both the lipid and apoB moieties of OxLDL, andspecifically 
to the phosphocholine (PC) headgroup of OxPL, such as 1-pal
-
mitoyl-2-(5ʹ-oxovaleroyl)-sn-glycero-3-phosphocholine (POVPC), 
present as either a lipid or as an adduct bound to protein via the 
ε-amino groupof lysine. They did not bind to thePC of native 
phospholipids (9). Importantly, these antibodies, as represented 
by the prototypic antibody E06, inhibited the uptake of OxLDL by 
macrophage scavenger receptors CD36 and SR-BI (5, 6, 10), as did 
POVPC linked to BSA or a peptide. This demonstrates that the PC 
moiety of OxPL is a ligand for macrophage scavenger receptors, 
which are innate, cellular pattern recognition receptors (PRRs).
Because all of these cloned autoantibodies were IgM Abs, which 
arethought in largepart to represent NAbs in uninfectedmice(11), 
we sequenced the complementarity-determining regions (CDRs) 
determining their antigen-binding sites, which revealed them allto 
be genetically identical to a well-characterized B-1 cell clone, T15, 
described more than 30 years ago (12). T15 NAbs bind to PC cova
-
lently linked to the cell wall polysaccharide (C-PS) of pathogens 
and provide optimal protection to mice from lethal infection with 
Streptococcus pneumoniae (13). Furthermore, immunization of cho-
lesterol-fed 
Ldlr
–/–
mice withheat-killed S. pneumoniaeled to a nearly 
exclusive expansion of E06/T15 NAbs and atheroprotection (14).
Because NAbs are postulated to be conserved by natural selec
-
tion, itwas not apparentwhat theselecting agentmightbe, as 
Conflict of interest:Joseph L. Witztum is named as inventor in patents and patent 
applications from the UCSD for the potential commercial use of antibodies to oxi-
dized LDL.
Nonstandard abbreviations used:AP, alkaline phosphatase; CuOx-LDL, copper 
sulfate–oxidized LDL; ELISpot, enzyme-linked immunospot; FACS, fluorescence-acti
-
vated cell sorting; 4-HNE-LDL,4-hydroxynonenal–modified LDL; 4-HNE-MSA, 
4-hydroxynonenal–modified mouse serum albumin; ISC, IgM-secreting cell; 
MAA-BSA, malondialdehyde-acetaldehyde–modified BSA; MAA-MSA, 
malondialdehyde-acetaldehyde–modified mouse serum albumin; MDA-LDL, 
malondialdehyde-modified LDL; NAb, natural Ab; OxLDL, oxidized LDL; OxPL, 
oxidized phospholipid; PAMP, pathogen-associated molecular pattern; PC, phospho
-
choline; PC-BSA, PC-conjugated BSA; PC-KLH, PC-conjugated keyhole limpet hemo-
cyanin; PEC, peritoneal exudate cell; PRR, pattern recognition receptor; SPF, specific 
pathogen–free.
Citation for this article:
J. Clin. Invest.119:1335–1349 (2009). doi:10.1172/JCI36800.
research article
1336 The Journal of Clinical Investigation http://www.jci.org  Volume 119  Number 5  May 2009
oxidation of LDL and atherosclerosis per se should not exert any 
positive selective pressure. We postulated thatapoptotic cells, 
similar to OxLDL, would also display oxidation-specific epitopes 
on their surface, as cells undergoing programmed cell death are 
known to undergo enhanced oxidative processes (15, 16) and if 
not promptly cleared are likely to beproinflammatory (17, 18). 
Indeed, using mass spectroscopy, we demonstrated that apoptotic 
cells contained an enhanced content of OxPL in their membranes 
and that E06 bound prominently to their cell surface, consistent 
with this hypothesis (18, 19). We also demonstrated that C-reactive 
protein (CRP), an innate acute-phase protein, recognized the same 
PC moiety on OxLDL and apoptotic cells (20). These data strongly 
suggest that the PC moiety of OxPL, apoptotic cells, and the cell 
wall of bacteria constitute a pathogen-associated molecular pat
-
tern (PAMP) recognized by multiple arcs of innate immunity and 
that each could exert positive selective pressure.
A variety of such oxidation-specific epitopes, besides PC of OxPL, 
are likely to occur in abundance not only on apoptotic cells, but 
on shed microparticles, andin generalon membranes and even 
bacteria during inflammatory responses. We postulated that they 
might constitute a previously unrecognized but important class 
of PAMPs and in turn would be a major target of innate NAbs. In 
Figure 1
IgM Abs to oxidation-specific antigens are present in germ-free and conventional mice. (A) Conventional and SPF C57BL/6 mice have similar
IgM titers to oxidation-specific antigens. Plasma from 11-week-old female conventionally raised (n = 4) and SPF (n = 4) C57BL/6 mice were
tested by ELISA. Values are mean and SEM. (B) MDA-LDL–specific ISCs are dominant in the spleens of conventionally raised C57BL/6 mice.
Splenocytes from conventionally raised 12-week-old female C57BL/6 mice (n = 4) were tested by ELISpot assay for frequencies of ISCs as
described in Methods. Values represent the number of ISCs to indicated antigen as a percentage of total ISCs (mean and SD). Data are from 1
experiment representative of 3. **P < 0.01 compared with all other antigens (1-way ANOVA with Tukey-Kramer multiple comparison test). (C)
Binding curves of plasma IgM from germ-free Swiss-Webster mice to indicated antigens. Plasma samples were from 14- to 16-week old female
and male mice (n = 9). Values are mean and SEM. (D) Titers of IgM Abs to oxidation-specific epitopes are present in conventional and germ-free
Swiss Webster mice. Serum from 14- to 16-week-old female and male conventionally raised (n = 7), conventionalized (germ-free colonized with
bacterial flora) (n = 11), and germ-free (n = 9) mice were diluted 1:400 and tested for binding to the indicated antigens. Values are mean and
SEM. *P < 0.05, **P < 0.01, ***P < 0.002 compared with α1,3-dextran (1-way ANOVA with Tukey-Kramer multiple comparison test).
research article
The Journal of Clinical Investigation http://www.jci.org  Volume 119    Number 5  May 2009  1337
this article, we provide multiple lines of evidence suggesting that 
oxidation-specific epitopes are a dominant target of innate NAbs 
in both mice and humans.
Results
IgM Abs against oxidation-specific epitopes are present in normal and germ-
free mice.To characterize themurine humoralIgM responsesto 
defined oxidation-specific epitopes, we assessed specific IgM titers 
in plasma ofnaive, nonatherosclerotic C57BL/6 mice.As previ
-
ously observed (8), prominent IgM titers to oxidation-specific epi
-
topes, such as OxLDL (>1:1,350) and malondialdehyde-modified 
LDL (MDA-LDL) (>>1:1,350), and to 4-hydroxynonenal–modified 
mouse serum albumin (4-HNE-MSA)andPC-conjugated BSA 
(PC-BSA; 1:1,350), can be detected even in normal, conventionally 
housed mice, whereas IgM titers to “native LDL” are minimal or 
undetectable(Figure 1A) (seecomment on apparent binding to 
native LDLbelowunder the subhead 
IgM binding to native LDL). 
Among these, the IgM responses to MDA modifications were con
-
sistentlyfound to be the most robust, and the titerswere many fold 
higher than the titer (1:1,350) to the prototypic B-1 cell antigen 
α1,3-dextran. In addition, when comparing the plasma IgM titers 
to those in age-matched mice bred under specific pathogen–free 
(SPF) conditions, a similar response pattern was observed (Figure 
1A), suggesting that the basal titers of these IgM Abs are largely 
independent  of noncommensalexposureto microbial  patho
-
gens. Moreover, we found that oxidation-specific IgM levels were 
also present in T cell receptor–deficient (
Tcra
–/–
) mice — although 
slightly lower — indicatingthat inlarge part these responses do not 
require T cells (see Supplemental Figure 1; supplemental material 
available online with this article; doi:10.1172/JCI36800DS1).
In mice,IgM Abs arein large part derived from Ab-secreting cells 
in the spleen (21). Using enzyme-linked immunospot (ELISpot) 
analysis, we tested the frequenciesofIgM-secreting cells (ISCs) 
against candidate oxidation-specific epitopesin thespleens of 
conventionallyhoused C57BL/6 mice. ISCs with specificityfor 
oxidation-specific epitopes were equally prominent in the spleen, 
as was observed for the IgM in plasma, with up to 15% of all ISCs 
having specificity for MDA-LDL (Figure 1B).
Our data suggest thatIgM titers toan array of oxidation-spe
-
cificepitopes may in fact representIgMgeneratedeven inthe 
absence of response-eliciting antigen exposure. In confirmation 
of this, we demonstrated robust IgM titers to oxidation-specific 
epitopes in the serum of “germ-free mice,” which are completely 
free of gut bacteria (Figure 1C). In particular, IgM titers to MDA-
LDL (>1:1,600) and MAA-BSA (>1:1,600)were the most prominent 
among all IgM titers measured: MAA (malondialdehyde-acetalde
-
hyde adduct) is a specific and prominent chemical moiety gener
-
ated from 2 MDA and 1 acetaldehyde molecules reacting with the 
ε-amino group of lysine to form an adduct; in this case forming 
adducts with BSA. Titers to OxLDL (>1:1,250) and 4-hydroxynon
-
enal–modified LDL (4-HNE-LDL; 1:800) were also much higher 
than those to 
α1,3-dextran (1:400), while titers to PC-BSA, which 
shares molecular identity to the PC of OxPL (as found in OxLDL), 
were approximately 1:400.
Furthermore, reconstitution of germ-free mice with gut bacte
-
ria (“conventionalized mice”) for only 2 weeks led to increases in 
Figure 2
In vitro stimulation of B-1 cells induces increased natural
IgM Ab titers to oxidation-specific antigens. (A) Purified
B-1 cells were cultured in 24-well plates in triplicate at
a cell density of 1 × 10
6
cells per well in 500 μl culture
medium. Cells were stimulated with IL-5 (50 ng/ml),
KdO
2
-Lipid A (100 ng/ml), or TLR2 agonists (a combi-
nation of Pam3CSK4 [300 ng/ml] and FSL-1 [1 μg/ml])
and incubated at 37°C for 7 days. Control B-1 cells were
cultured in medium alone. Cell culture supernatants were
harvested after 7 days and IgM Ab titers analyzed by
ELISA at 1:45 dilution. Results were normalized to cell
number recovered after 7 days. Values are mean and
SEM. Data are from 1 experiment representative of 3.
*P < 0.05, **P < 0.01, ***P < 0.002 compared with α1,3-
dextran (repeated-measures ANOVA with Tukey-Kramer
multiple comparison test). (B) Natural IgM Abs produced
in vitro show specificity to MDA-LDL and CuOx-LDL. For
competition immunoassay, supernatants from purified
B-1 cell cultures stimulated with KdO
2
-Lipid A (100 ng/
ml) or IL-5 (50 ng/ml) were diluted to 1:20 and incubated
in the presence of the indicated concentrations of com-
petitors (Competitor conc.) overnight. After incubation,
IgM binding to MDA-LDL and CuOx-LDL was tested by
ELISA. Data are the mean of triplicate determinations,
expressed as ratio of IgM binding to MDA-LDL or CuOx-
LDL in the presence or absence of competitor (B/B
0
).
Data are from 1 experiment representative of 3.
research article
1338 The Journal of Clinical Investigation http://www.jci.org  Volume 119  Number 5  May 2009
research article
The Journal of Clinical Investigation http://www.jci.org  Volume 119    Number 5  May 2009  1339
many — but not all — of the oxidation-specific IgM levels (Figure 
1D), strongly suggesting molecularmimicry between many endog
-
enousoxidation-specific  epitopesandgut  bacterial epitopes. 
While IgM responses to PCand 4-HNE, as well as to 
α1,3-dex-
tran, were increased (more than 2-fold) in these mice, the MDA-
specific responses were found to be similar to those in germ-free 
mice. Moreover, similar specific IgM responses were measured in 
age-matched conventionally raised mice of the same genetic back
-
ground (Figure 1D). Note again that oxidation-specific IgM Abs 
constitute a major fraction of total IgMAbs, which were not differ
-
ent among wild-type, germ-free, or colonized mice (Figure 1D).
IgM binding to native LDL. In these studies, we tested IgM binding 
to antigens coated on microtiter wells using standard solid-phase 
ELISAtechniques (22). In some experiments, we saw low to modest 
levels of IgM binding to native LDL, which appeared to vary with 
different LDL preparations. However, in all cases, the binding to 
plated nativeLDL couldnot be competed by thesameprepara
-
tion of native LDL in solution (data not shown), suggesting that 
the LDL became modified in some way during the plating process. 
It should be emphasized that in competition immunoassays, we 
never observed native LDL competing for binding to any of the 
modified LDL preparations (e.g., as shown in Figure2B, Figure 
3A, and Figure 4A).
B-1 cells secrete IgM NAbs against oxidation-specific epitopes in vitro. To 
directly demonstrate that oxidation-specific IgM Abs are derived 
from innate B-1 cells, we isolated B-1 cells (both CD5
+
B-1a and 
CD5
 B-1b) from naive mice by fluorescence-activated cell sorting 
(FACS), stimulated them in vitro with various stimuli, and tested 
the culture supernatants for specific IgM Abs. IL-5, a TLR4 ligand 
(KdO
2
-Lipid A), as well as a combination of TLR2 ligands (FSL-1 
and Pam3CSK4) all induced B-1 cells to secrete IgM against MDA-
LDL, OxLDL (Figure 2A), and 4-HNE-LDL (data not shown); but 
alsoagainstthe prototypic B-1 cell antigen 
α1,3-dextran (Figure 2A). 
Utilizing B-1cellsfrom
Myd88
–/–
mice, weobserved thatthe respons-
esto both TLR4andTLR2agonistswerein large partMyD88 depen
-
dent (data not shown). Interestingly, the basal secretion of IgM to 
OxLDL and MDA-LDL was strikingly moreprominent thanthatof 
IgM against 
α1,3-dextran (Figure 2A), and in response to stimula-
tion with the TLR agonists, the oxidation-specific IgM increased to 
a greater extent than did the total IgM, or the IgM to
α1,3-dextran, 
for example (Supplemental Figure 2). Basal titers to PC-BSA were 
relatively low and increased only in response to IL-5. PC is an epit
-
opeforsome OxLDL-specific IgM, but not all.
By analogy to the robust anti-MDA responses in vivo, MDA-spe
-
cific IgM Abs were the dominant set of IgM Abs secreted by B-1 
cells in vitro. Although we tested only a narrowly selected set of 
antigens, the anti-MDA Abs constituted up to 30% of total IgM 
secreted. These Abs were highly specific for MDA modifications,as 
only MDA-LDL, butneitherOxLDL nor nativeLDL, competed for 
the binding (Figure 2B), while IgM Abs bound to plated OxLDL 
were competed by both OxLDL and MDA-LDL. We also calculated 
the binding avidities of the IgM in the supernatants for MDA-LDL 
and OxLDL usingthe Klotz method (23).The calculated 
K
d
s for 
MDA-LDL andOxLDL were 1.46 
×10
–7
mol/l and4.03 ×10
–9
mol/l, 
respectively, similar to values we previously determined for IgM in 
plasma of mice immunized against these epitopes (18).
B-1 cells secrete IgM NAbs against oxidation-specific epitopes in vivo. To 
test whether innate B-1 cells can alsosecreteoxidation-specific 
IgM in vivo, peritoneal B-1 cells from naive C57BL/6 mice were 
adoptively transferred intotheperitoneum of 
Rag1
–/–
recipient 
mice, which lack functional B and T cells.This led to the selective 
reconstitution of only B-1 cells in the peritoneum of 
Rag1
–/–
recip-
ients (Figure 3A, top row), and both B-1a (CD5
+
CD19
+
CD11b
+
) 
and B-1b cells (CD5
CD19
+
CD11b
+
) were detected (Figure 3A, 
middlerow), whereas conventional B-2 cells and T cells were typi
-
cally not found in recipient mice. The average percentagesof B-1a 
and B-1b cellsamong total cellsanalyzed in the peritoneum of 
Rag1
–/–
 B-1 recipients were 11.0% ± 2.6% and 8.5% ± 1.0% respec-
tively (
n= 12), compared with 18.2% ± 3.2% and 8.3% ± 2.3% (n= 4) 
in wild-type C57BL/6 mice.
The adoptive transfer also reconstituted the B-1 cell population 
in the spleens of 
Rag1
–/–
 B-1 recipients. B-1 cells (IgM- and CD43-
positive; Figure 3A, bottom row) constituted an average of about 
1% of total splenocytes analyzed (data ranged from 0.4% to 1.4%; 
n = 9), while in wild-type C57BL/6 mice, B-1 cells averaged about 
2.7% (
n = 3). Moreover, this reconstitution was also demonstrated 
by the number of splenic ISCs as measured by ELISpot: 60 ± 15 
ISCs vs. 152 ± 21 ISCs per 200,000 splenocytes in 
Rag1
–/–
B-1recipi-
ents (
n = 9) and C57BL/6 mice (n= 7), respectively.
The B-1cell–reconstituted 
Rag1
–/–
mice developed readily detect-
able plasma titers of IgM by the tenth week after transfer (Figure 
3B), while 
Rag1
–/–
mice that received only PBS did nothaveany 
plasma IgM. In a series of8 similar transfer experiments, the extent 
of reconstitution, as indicated by the levels of plasma IgM, varied 
and in part appeared to be related positively to the number of B-1 
cells transferred and the time after transfer studied.Remarkably, in 
all of the transfers, the recipient mice exhibited robust IgM titers 
against oxidation-specific epitopes (Figure3B), but notnativeLDL 
(data not shown), and the prevalence ofoxidation-specific IgM 
appeared similar to that observed in naive wild-type mice. In Fig
-
ure 3B, we show Ab binding dilution curves to oxidation-specific 
epitopes and other antigens fora typical transfer of approximately 
99% pure B-1 cells. Thetiters ranged from 1:800 to 1:6,400for 
oxidation-specific epitopes versus 1:400 for 
α1,3-dextran and PC-
BSA. Moreover, consistent with the fact that T15-idiotypic Abs are 
Figure 3
Characterization of Rag1
–/–
recipients adoptively transferred with B-1
cells. (A) Adoptive transfer of B-1 cells into Rag1
–/–
mice replenishes
B-1 cell population. Rag1
–/–
mice were injected with PBS (Rag1
–/–
+
PBS) or with B-1 cells (Rag1
–/–
+ B-1). Rag1
–/–
+ PBS: Lymphocyte
populations were absent in the peritoneal cavity (PEC, left). B-1 cells
(IgM
+
CD43
+
) were also absent from the spleen (Spleen, left). C57BL/6:
Peritoneal macrophages (CD11b
hi
CD5
) and T cells (CD11b
CD5
hi
)
were intact (PEC, upper middle). B cells could be divided into B-1a
(CD19
+
CD11b
int
CD5
int
), B-1b (CD19
+
CD11b
int
CD5
), and B-2 cells
(CD19
+
CD11b
CD5
) (PEC, lower middle). In the spleen, B-1 cells
were about 2.4% of total splenocytes (Spleen, middle). Rag1
–/–
+ B-1:
B-1 cell populations were reconstituted in the peritoneal cavity (PEC,
lower right) and spleen (Spleen, right), without B-2 cell or T cell con-
tamination (PEC, right). (B) IgM Abs to oxidation-specific epitopes are
present in the plasma of B-1 reconstituted Rag1
–/–
mice. Plasma col-
lected after 15 weeks from Rag1
–/–
+ B-1 (n = 8) or Rag1
–/–
+ PBS
(n = 6) and age-matched C57BL/6 mice (n = 7) were tested. Data
shown are from 1 transfer experiment representative of 6. Values
are mean and SEM. Numbers in the upper-right corner represent the
IgM titer to each antigen. (C) Natural IgM Abs produced in vivo show
specificity to MDA-LDL and CuOx-LDL. Data are the mean of triplicate
determinations, expressed as the ratio of IgM binding to MDA-LDL or
CuOx-LDL in the presence or absence of competitor (B/B
0
). Data are
from 1 experiment representative of 3.
research article
1340 The Journal of Clinical Investigation http://www.jci.org  Volume 119  Number 5  May 2009
predominantly secreted by B-1 cells (24), we found that adoptive 
B-1 cell transfer gave rise to E06/T15-idiotypic IgM in 
Rag1
–/–
mice 
as well (Figure 3B, bottom middle panel).
Whereasthe extent of plasmaIgM measured at 10 weekswas gen
-
erally lower in B-1 cell–reconstituted 
Rag1
–/–
than in wild-type mice 
in all the experiments noted above, inwhich greater than99% pure 
B-1 cells were transferred, in one experiment,in which a smallcon
-
tamination of T cellsinadvertently occurred (estimated to be <3%of 
allperitonealcells at sacrifice),the IgM levelsactually equaledthose 
of wild-type C57BL/6 mice (Supplemental Figure 3). Presumably, 
cotransferred T cellspromotedIgMsecretion in the recipient mice, 
possibly in part by secretionof cytokinessuch asIL-5, which, as we 
have previously shown, augments secretion of OxLDL-specific IgM 
by B-1 cells in a non-cognate manner (Figure 2A and ref. 25).
Again, oxidation-specificIgM Abs were a major fraction of the 
totalIgM. Competition immunoassays withpooled plasma of 
recipient mice demonstrated high specificity of the MDA-specific 
IgM Abs,as neither OxLDLnor native LDLcompeted for bindingto 
MDA-LDL (Figure 3C). We also calculated binding avidities for the 
IgM in the B-1 cell–reconstituted 
Rag1
–/–
 plasma for oxidation-spe-
cific epitopes as described above. The apparent 
K
d
s for MDA-LDL 
andOxLDLwere 9.9 
×10
–9
and 1.42×10
–7
mol/lrespectively. These 
values are similar to the 
K
d
s of 6.85×10
–8
 mol/l determined for the 
MDA-LDL–specificnaturalmAb NA-17,cloned from the spleen ofa 
B-1cell reconstituted 
Rag1
–/–
miceas describedbelow. Onlyvery low 
titers of IgG against oxidation-specific epitopes (or any other anti
-
gen) were found in theplasma of recipient mice,and theseAbswere 
predominantly of the IgG3isotype, which are known to be secreted 
by B-1 cells in a Tcell–independent fashion (26) (data not shown).
Oxidation-specific epitopes are dominant targets of natural IgM Abs. To 
directly addressthe extent to which oxidation-specific IgM Abs 
contribute to the total IgM NAb pool, we performed absorption 
studies using pooled plasma from 
Rag1
–/–
mice reconstituted with 
B-1cells.In theseexperiments, specific IgM Abs wereabsorbed 
from plasma with selected oxidation-specific model antigens and 
the amount of remaining IgM measured. Strikingly, MDA-LDL as 
well as OxLDL(containing a variety ofoxidation-specific epitopes) 
absorbed out approximately 10% of all IgM, while native LDL did 
Figure 4
Oxidation-specific epitopes are dominant targets of NAbs. (A) Preabsorption of plasma from Rag1
–/–
+ B-1 mice with oxidation-specific antigens
shows that oxidation-specific epitopes (OxEpitopes) are dominant targets for NAbs. Plasmas from Rag1
–/–
+ B-1 mice were preincubated in the
absence or presence of the indicated antigens (250 μg/ml total antigen) overnight and antigen-immune complexes pelleted by centrifugation.
Total IgM levels were then tested by ELISA. *P < 0.05, **P < 0.01, ***P < 0.002 compared with native LDL (ANOVA with Tukey-Kramer multiple
comparisons test). Data are means (and SEM) from 5 separate experiments, each using 3–7 plasma samples obtained from 5 different transfer
experiments, with each sample assayed in triplicate. (B) ELISpot assay of frequencies of MDA-LDL–specific ISCs in the spleens of wild-type
C57BL/6, Rag1
–/–
+ B-1, and Rag1
–/–
+ PBS mice. Results are from individual mice, and data are from 3 separate B-1 cell transfer experiments.
Horizontal bar represents the mean for the group.
P < 0.002 compared with Rag1
–/–
+ PBS (unpaired t test). (C) B-1 cell–derived natural mAb
NA-17. DNA sequences of VDJ splice sites of the V
H
and V
L
rearrangements expressed in NA-17 B-1 cell hybridoma and their relationship to the
most homologous germline V, D, J gene segments. Sequence analysis of NA-17 V
H
rearrangement did not reveal nucleotide variation to germline
genes. Sequence analysis of V
L
rearrangement revealed 1 nucleotide insertion between V
L
and J
L
germline gene segments.
research article
The Journal of Clinical Investigation http://www.jci.org  Volume 119    Number 5  May 2009  1341
Figure 5
Natural IgM Abs recognize oxidation-specific epitopes present on apoptotic cells and atherosclerotic lesions (A) Natural IgM Abs bind to apoptotic
thymocytes but not normal thymocytes. Apoptotic thymocytes from C57BL/6 mice were incubated with plasma from Rag1
–/–
+ B-1 or Rag1
–/–
+
PBS mice at 1:10 dilution, NA-17 at 2.5 μg/ml, or control IgM at 5 μg/ml. Top row: Deconvolution microscopy shows that NAbs in Rag1
–/–
+ B-1 as
well as NA-17 bound to apoptotic thymocytes. Bottom row: None of the IgM bound to normal thymocytes (quadrant 1 [Q1). NAbs in plasma from
Rag1
–/–
+ B-1 bound to both early (Q2) and late apoptotic thymocytes (Q3), while NA-17 bound prominently to late apoptotic cells (Q3). Scale
bar: 5 μm. 2°Ab, secondary Ab; Anti-ms-IgM-FITC, FITC-labeled anti-mouse IgM. (B) Natural IgM Abs are present in atherosclerotic lesions.
Endogenous IgM Abs were detected in aortic sections from cholesterol-fed B-1 cell–reconstituted Ldlr
–/–
Rag1
–/–
mice (bottom row), but not in
PBS-injected Ldlr
–/–
Rag1
–/–
mice (top row). Sections were also stained with MDA2 (5 μg/ml) for the presence of MDA epitopes. Red indicates
positive staining. Original magnification, ×160. (C) NA-17 recognizes oxidation-specific epitopes present in atherosclerotic lesions. Sections
of the brachiocephalic artery from cholesterol-fed Ldlr
–/–
Rag1
–/–
mice were stained with NA-17 (0.85 μg/ml) or a control natural IgM Ab, EN-2
(1.6 μg/ml). Original magnification, ×200. (D) NA-17 inhibits MDA-LDL binding to macrophages. Increasing concentrations of NA-17 or control
IgM were added with a fixed amount of biotinylated MDA-LDL (Bt-MDA-LDL; 2 μg/ml) to macrophages. Data are the average of 2 experiments,
expressed as the ratio of biotinylated MDA-LDL binding to macrophages in the presence or absence of IgM (B/B
0
).
research article
1342 The Journal of Clinical Investigation http://www.jci.org  Volume 119  Number 5  May 2009
not(Figure 4A). Furthermore,the MDA-specific MAA epitope, 
conjugated to mouse serum albumin (MAA-MSA), was capable of 
absorbing up to 25% of all IgM. Moreover, acombination ofall 
model antigens removed 35% of the NAbs in these plasmas (Figure 
4A). Thus, a surprisingly large percentageof B-1 cell–derived NAbs 
are directed against various oxidation-specific epitopes.
The  prominentrepresentationof  oxidation-specificIgM  in 
plasma was also reflected by the frequency of MDA-specific ISCs 
in the spleens of reconstituted 
Rag1
–/–
 mice. As expected, no ISCs 
were detected in the spleens of mice injected with PBS (Figure 4B). 
In contrast, in B-1 cell–reconstituted mice, approximately 12% of 
all ISCs were found to have specificity for MDA-LDL (Figure 4B). 
Interestingly, the frequency of MDA-LDL–specific ISCs in spleens 
of wild-type mice was found to be similarly high (Figure 4B), dem
-
onstrating that a large percentage of all splenic ISCs have specific
-
ity for MDA modifications, and a majority of these ISCs are likely 
derived from B-1 cells.
We furthercorroborated this notion bythecharacterization 
ofmAbs derived fromhybridomas preparedfromthe spleens 
of B-1 cell–reconstituted
Rag1
–/–
mice. From 2 separate fusions, 
we observed that 20%–30% of all IgM-secreting hybridomas had 
reactivity for MDA-LDL (data not shown). For example, the DNA 
sequence ofVDJ splice sitesof theV
H
andV
L
rearrangements 
expressedinone cloned MDA-specific hybridoma (NA-17) dis
-
playedcompletegermline gene usageof theV
H
rearrangement 
and only 1 nucleotide insertion (C) at the splice site of the V
L
 and 
J
L 
germline gene segments (Figure 4C; the completesequence is 
presented in Supplemental Figure 4).
B-1 cell–derived natural IgM Abs recognize oxidation-specific epitopes on
apoptotic cells and in atherosclerotic lesions. Oxidation-specific epitopes 
are ubiquitously present in inflammatory settings and are present 
on apoptoticcells(19, 27). As shownin Figure5A, plasma IgM from 
B-1 cell–reconstituted 
Rag1
–/–
mice recognized surface epitopes on 
apoptotic cells, as demonstrated by immunocytochemistry. Simi
-
larly, the MDA-LDL–specific NAb NA-17 strongly stained apop
-
totic cells. Neither plasma from PBS-injected 
Rag1
–/–
mice nor a 
keyhole limpet hemocyanin–specific (KLH-specific) control IgM 
bound apoptotic cells.
Figure 6
Human umbilical cord blood contains natural IgM Abs against oxidation-specific epitopes. (A) Left: Plasma titers of IgM in maternal and umbilical
cord plasma to native LDL, KLH, and oxidation-specific antigens measured by ELISA. Right: Data are plotted as ratio of antigen-specific IgM to
total IgM. ***P < 0.002 compared with maternal blood (Wilcoxon matched-pairs test and paired t test). Data shown are from 10 paired maternal-
infant samples, and each sample was assayed in triplicate. Values are mean and SEM. (B) Umbilical cord IgM binds to apoptotic cells in part via
binding to MDA. Apoptotic Jurkat cells, induced by UV exposure, were incubated with representative umbilical cord plasma (1:50 dilution) in the
absence and presence of MDA-LDL and native LDL (1 mg/ml). Abs bound were detected by FITC-conjugated anti-human IgM. Umbilical cord
IgM binding to apoptotic Jurkat cells (median fluorescence intensity [MFI], 1,917) was inhibited 45% by MDA-LDL (MFI, 1,047) while minimally
affected by native LDL (MFI, 1,514).
research article
The Journal of Clinical Investigation http://www.jci.org  Volume 119    Number 5  May 2009  1343
We also tested IgM binding to apoptotic cells by flow cytometry 
in relation to measures of cellular apoptosis (Figure 5A). Plasma 
IgM from reconstituted 
Rag1
–/–
mice bound to apoptotic cells with 
early (quadrant 2 [Q2]) and late (Q3) stages of apoptosis, but not 
viable cells(Q1). MonoclonalNA-17 stainedalmostthe entire 
population of late apoptotic cells. To further test whether these 
IgM NAbs recognize their cognate epitopes in vivo, we transferred 
B-1 cells into 
Ldlr
–/–
Rag
–/–
 mice that had been fed a high-cholester-
ol diet for 10 weeks to induce atherosclerotic lesions, which accu
-
mulate apoptotic cells and OxLDL (28, 29). The mice were main
-
tained on the atherogenic diet for an additional 6 weeks and then 
sacrificed, and atherosclerotic lesions were assessed for the depo
-
sition ofendogenous Abs(Figure 5B). As expected, neither IgM 
nor IgG Abs were present in lesions of 
Ldlr
–/–
Rag1
–/–
 mice injected 
with PBS alone (Figure 5B, top row). In contrast, IgMaccumulated 
in lesions of B-1 cell–reconstituted mice, at sites likely reflecting 
the edges of lesions at the time of engraftment of the transferred 
B-1 cells (Figure5B, bottom row). Inpart, these IgMAbs were 
binding to endogenous MDA epitopes in the lesions, which were 
found in comparable (but not identical) sites, asdemonstrated 
by immunostaining of adjacent sections using MDA2, an MDA-
specific murine monoclonal IgG we previously cloned (30) (Figure 
5B). Presumably MDA epitopes bound by endogenous IgM would 
also not be available to bind MDA2. Moreover,consistent with our 
earlier observation that only low IgG titers are secretedby B-1 cells, 
no IgG Abs were found in lesions of these mice.
Monoclonal NA-17 also recognized MDA epitopes in atheroscle
-
rotic lesionsof 
Ldlr
–/–
Rag
–/–
mice (Figure 5C), whereas a control 
natural IgMdid not.We also found that NA-17 couldsubstan
-
tially inhibit the binding of MDA-LDL to J774 macrophages in a 
dose-dependent fashion, whereas a KLH-specific IgMshowed only 
nonspecific inhibition (Figure 5D).
Natural IgM Abs in human cord blood recognize oxidation-specific epi-
topes. We previously showed that IgG and IgM titers to oxidation-
specific epitopes are present in nearly all adult human plasmas 
tested (4). However, we wanted to know whether the human IgM 
NAb repertoire displays binding to oxidation-specific epitopes 
similartothoseobserved in mice.IgM Absfoundinumbilical 
cord blood are exclusively from the infant and are considered to 
represent the humanequivalent ofnaive NAbs(31). Therefore, 
we characterized the binding properties of IgM in umbilical cord 
blood andinthe respective maternal plasma samples. Impor
-
tantly,umbilicalcordplasma contained prominent IgMtiters 
against MDA-LDL and OxLDL, but not native LDL or KLH (Fig
-
ure 6A). In contrast, maternal plasma also contained IgM against 
KLH, presumably due to exogenous antigen exposure (Figure 6A). 
Unlike IgM titers, IgG titers were found to be similar in maternal 
and umbilical cord plasma (data not shown), consistent with the 
ability of maternal IgG to be actively transferred across the pla
-
centa. As the totalIgM titers were found to be generally higher 
inmaternal samples, wecalculated the ratiosof oxidation-spe
-
cific IgM to totalIgM in individual samples. This revealed that 
MDA-LDL– and OxLDL-specific IgM — but not KLH-specific IgM 
— are relatively enriched in umbilical cord plasmas compared with 
maternal plasmas (Figure 6A), which typically also contain adap
-
tive IgM Abs (e.g., against KLH).
We further showed that human umbilical cord IgM also bound 
to apoptotic cells(Figure6B). This binding wasat least inpart 
mediatedthroughthe recognition ofMDAepitopes, as,in the 
example shown, MDA-LDL competed for up to 45% of the bind
-
ing of the umbilical cord IgM to apoptotic cells. Thus, oxidation-
specific epitopes also appear to be a dominant target for natural 
IgM Abs in humans.
NAbs facilitate apoptotic cell uptake by macrophages in vivo. Natural 
IgM Abs andthe MDA-LDL–specific NAb NA-17 recognize oxi
-
dation-specific epitopes on apoptotic cells (Figure 5A). When not 
promptly cleared, apoptotic cells are immunogenic and proinflam
-
matory (17, 18). To test the hypothesis that these NAbs maintain 
homeostasis against oxidatively modified structures, we examined 
their ability to mediate enhanced clearance of apoptotic cells in 
vivo. Using a previously described model (32), we compared apop
-
totic cell uptake by macrophages in vivo in 
Rag1
–/–
 mice 10 weeks 
after reconstitution with B-1 cells or with PBS. Mice were injected 
i.p. with fluorescently labeled apoptotic thymocytes 4 days after 
induction of sterile peritonitis with thioglycollate. Macrophage 
Figure 7
Natural IgM Abs against oxidation-specific epitopes facilitate apoptotic
cell uptake by macrophages in vivo. (A) Percentage of macrophages
that contained fluorescently labeled apoptotic thymocytes following i.p.
injection. In RAG + PBS mice, about 27% of macrophages phago-
cytosed apoptotic cells. The percentage was significantly increased,
to 33%, in RAG + B-1 mice (*P < 0.05, unpaired t test). Horizontal
bars denote means. (B) Apoptotic thymocytes were preincubated with
NA-17 or a control IgM that does not bind apoptotic cells before injec-
tion into Rag1
–/–
mice. The phagocytic uptake was significantly differ-
ent between the 3 groups (P = 0.01, 1-way ANOVA). The percentage
of macrophages taking up apoptotic cells was significantly increased
when the apoptotic cells were preincubated with NA-17 (RAG + NA-17),
compared with control IgM (RAG + control IgM) or without preincu-
bation (RAG) (32% vs. 20% vs. 23%; *P < 0.05, Bonferroni multiple
comparison test). Horizontal bars denote means.
research article
1344 The Journal of Clinical Investigation http://www.jci.org  Volume 119  Number 5  May 2009
uptake of apoptotic thymocytes was significantly enhanced in B-1 
cell–reconstituted 
Rag1
–/–
 mice compared with PBS controls (33% 
vs. 27%, 
P < 0.05; Figure 7A).
To directly test the impact of an oxidation-specific mAb, we pre
-
incubated the apoptotic cells with NA-17 or with a control IgM 
(specific for KLH) before injection into 
Rag1
–/–
 mice. The percent-
age of macrophages that had taken up apoptotic thymocytes was 
similarin 
Rag1
–/–
mice receiving apoptotic thymocytespreincu-
bated with control IgM or untreated apoptotic thymocytes (20% 
vs. 23%, 
P=NS; Figure 7B).Incontrast,the phagocytic uptake 
increased to 32% when the apoptotic cells were preincubated with 
NA-17 (
P< 0.05), demonstrating that binding of NA-17 facilitated 
apoptotic cell uptake by macrophages in vivo.
Discussion
Innate immune responses provide a vital and nonredundant role 
in the initial defense against invading pathogens and in maintain
-
ing homeostasis against a variety of self-antigens. The mediators 
of innate immunity are germline encoded,preformed, and utilize a 
limited set of PRRs to recognize a set ofcommon PAMPs. It is now 
widely recognized that one class of PRRs, the so-called scavenger 
receptors, such as SRA-1 and -2, CD36, SR-B1, LOX-1, PSOX, and 
others, allrecognize various oxidation-specific epitopes present on 
OxLDL and apoptotic cells (7). Indeed, recent evidence suggests 
that members of the innate TLR family also recognize such epit
-
opes as well (33, 34). The soluble innate protein CRP also binds to 
the PC of OxPL present on OxLDL and apoptotic cells, as well as 
to 
S. pneumoniae(20). Thus, oxidation-specific epitopes — as a class 
— constitute an important PAMP recognized by innate immunity, 
consistent with the oxygen-centric nature of life itself.
NAbs are part of the humoral arc of innate immunity. They are 
the product of naturalselection, exhibitinga geneticallydeter
-
mined and stable repertoire that is largely independent of exter
-
nal antigenic stimuli (35–37). NAbs have been defined in various 
ways, for example, as the IgM present in normal plasma (38), but 
more stringently, asgermline-encoded Abs that arise without any 
exogenous immune exposure or bacterial colonization of the gut 
(39). They are produced at tightly regulated and stable levels in 
healthy individuals and are found in all vertebrate species (11, 40). 
In normal, uninfected mice, most plasma IgM Abs are NAbs that 
are derived from innate B-1 cells, which differ from conventional 
B-2 cells in surface phenotype; anatomic location; restricted use 
of V
H
 genes that are minimally edited and reflective of germline 
usage; and, importantly, their capacity for self renewal (reviewed 
in refs. 11,41, 42).Despitethis, relatively little isknown about 
B-1cellontogeny, development,and,surprisingly,  function. 
Only recentlyhas evidence been presented to support a lineage 
independent from that of B-2 cells (43–46).Although the mecha
-
nisms leading to selection and expansion of B-1 cells are unclear, 
it is now appreciated that antigen selection during the fetal and 
neonatal period leads to
positive selection (11, 47, 48). Since this 
appears to occur equally well in mice raised in germ-free environ
-
ments (35, 49, 50), this selection is of necessity made by endog
-
enous “self-antigens.” This is in contrast to developing, self-reac
-
tive B-2 cells, in which such antigen encounter early in life leads 
tonegativeselectionviaapoptosis (aprocessknown asclonal 
deletion) or anergy. Thus, the repertoire of B-1 cells is selected to 
bind to evolutionarily important epitopes. Furthermore, because 
many, if not most, NAbs have dual specificities (11), typically with 
molecular epitopes on pathogens, an encounter later in life with 
such external pathogenscould further servetoexpand a given 
B-1 cell clone (12, 51). The example given in the Introduction that 
E06/T15, which binds OxPL of OxLDL and apoptotic cells, also 
binds to the PCs present on the cell wall of many pathogens nicely 
illustrates this paradigm. We report now that as a class, oxidation-
specific neoepitopes constitute a disproportionately large portion 
of such self-antigens.
In this article, we provide multiple lines of evidence suggesting 
that oxidation-specific epitopes are a dominant target of innate 
NAbsin both miceand humans. First, consistentwiththe view that 
theIgM poolof uninfected mice comprisesNAbs (11), weshowthat 
SPF-maintained mice as well as germ-free mice maintained under 
strict gnotobiotic conditions contained prominent IgM titers to 
oxidation-specific epitopes comparedwith titersagainst recognized 
TI-2 antigens, such as 
α1,3-dextran. Consistent with this, there was 
a high frequency ofISCs to oxidation-specific epitopes, chiefly 
MDA, in spleens of wild-typemice.Of interest, the introduction of 
normal gut flora into the germ-free mice (conventionalized mice 
in Figure 1D) increasedthe titers to OxLDL,PC-BSA, 4-HNE-LDL, 
and
α1,3-dextran.After birth,B-1cells are typically thought to 
have minimal response to exogenous antigenic stimuli, but some, 
such as E06/T15 (which binds to the PC of OxPL in OxLDL and 
PC-BSA), are known to be responsive (11, 14).
Second, B-1cells inculture, purified from naive uninfected mice, 
gave rise to anarray of oxidation-specific IgM Abs, whichwere 
much more prevalent than IgM to the classicantigen 
α1,3-dex-
tran (42). Indeed, such IgM Abs were already detectable in super
-
natants of unstimulated B-1 cells in culture, and the production 
oftheseIgM Abs in response tostimulation, especially that of 
TLRs, increased to a greater extent than that of total IgM or IgM 
to 
α1,3-dextran.Moreover, by analogy to the robust anti-MDA 
responses in vivo, MDA-specific IgM Abs were the most dominant 
set ofIgM Abs secreted in vitro, constituting up to 30% of total 
IgM. A similar specificity was seen for the IgM found in plasma of 
Rag1
–/–
 mice following adoptive transfer of B-1 cells. Absorption 
studies demonstrated that 20%–35% of these IgM Abs had specific
-
ity for combinationsofoxidation-specific epitopes, predominantly 
MDA-related epitopes. These data were further corroborated by 
ELISpotanalysisof splenocytes of 
Rag1
–/–
B-1recipients,which 
demonstrates that ISCs against MDA-LDL accounted for 10%–12% 
of all ISCs in the spleens of the reconstituted mice, similar to the 
frequency found in spleens of wild-type mice. In addition, study 
of hybridomas generated from the spleens of the B-1 cell–recon
-
stituted mice indicated that 20%–30% of all IgM-secreting clones 
bound to oxidation-specific epitopes, mostly MDA. Cloning and 
sequencing of a number of these confirmed their germline origin, 
as exemplified by the MDA-specific B-1 clone NA-17.
Third, these NAbs bound prominently to biologically relevant 
oxidation-specific epitopes. B-1 cell–derived IgM in the plasma of 
thereconstituted mice boundprominently to epitopes on apoptotic 
cells and to atherosclerotic lesions and colocalized with oxidation-
specific epitopes in vivo in atherosclerotic lesions. Furthermore, 
monoclonal NA-17 bound prominently to apoptotic cells and to 
atherosclerotic lesions. These data stronglysupport the hypothesis 
that oxidation-specific epitopes are a dominant target of NAbs in 
mice, which in turn bind to biologically relevant self-antigens.
Fourth, IgM Abs in umbilical cord blood are solely of fetal origin 
and represent such naive NAbs in humans (31). The presence of an 
enriched titer of oxidation-specific IgM Abs in human umbilical 
cord blood, which bind to apoptotic cells in part via MDA-LDL 
research article
The Journal of Clinical Investigation http://www.jci.org  Volume 119    Number 5  May 2009  1345
epitopes, strongly suggests that oxidation-specific epitopes are an 
important target of NAbs in humans as well. Interestingly, Merbl 
et al., using a sensitive antigen-array screening technique, recently 
tested the specificities of autoreactive NAbs in human cord blood 
(31)and reportedthat theantigen towhich suchNAbsmost 
commonly bound was native LDL. However, as noted in Results, 
though we often find low levels of binding to plated LDL, this can
-
not be competed by the same LDL in solution, suggesting that the 
LDL undergoes modification during the plating procedure (either 
structural, e.g., via oxidation, or even conformational). Indeed, the 
LDL used in the studies of Merbl et al. was bought from a com
-
mercial source, and we speculate that it may have been oxidized at 
the time of use. Thus, their data may be consistent with our obser
-
vations that oxidation-specific epitopes are a dominant target of 
NAbs found in humans.
The apparent high prevalence of oxidation-specific epitopes as 
targets of NAbs in mice and humans likely reflects the ubiquitous 
presenceof these epitopesconsequentto oxidative events. Cells 
undergoing apoptosis and the microparticles shed from them are 
rich in oxidation-specific epitopes(15, 18,52). Because this process 
is universal, we speculate that they could be fundamental select
-
ingantigens.Further, inflammatory events are associated with 
enhancedoxidative stress, and our different oxidation-specific 
Abs have been used to demonstrate the presence of these epitopes 
not only in atherosclerotic tissue, but in a variety of inflammatory 
settings, includingrenal,liver, andpulmonary disease(53–55), 
rheumatoid arthritis (our unpublished observation), CNS lesions 
found in multiple sclerosis (56), and Alzheimer disease (57), all of 
which would serve to provide stimulation later in life for particu
-
lar B-1 cell clones. Finally, many if not most NAbs also react with 
microbes that contain the same or similar antigenic self-determi
-
nants, leading to cross-reactivity between self-determinants and 
microbial antigens. The classic naturalIgM E06/T15, for example, 
binds PC of OxPL as well as PC as part of the capsular polysaccha
-
ride of pneumococci and many other microbes. Thus, postnatal 
stimulation of particular B-1 cell clones likely occurs upon expo
-
sure to molecular equivalents on pathogens.
Although the selection and expansion of B-1 cell clones clearly 
occurs in both fetal and postnatal life, the mechanisms leading 
to the activation of B-1 cells, conversion to plasma cells,and gen
-
eration of NAbs are poorly understood. The mode of presenta
-
tionofantigenthat resultsin productiveB-1 cellexpansionis 
generallyconsidered to be independent of cognate Tcellhelp. 
AlthoughsomeTI-2multivalent antigensmayleadto direct 
crosslinking of B cell receptor on B-1 cells and IgM production, 
in general most B-1 cell antigens do not (58). However, B-1 cells 
are known to be responsive to several non-cognate stimuli that 
also stimulate B-2 cells, including LPS and T cell cytokines such 
as IL-5. In our studies, we also found that NAbs specific for oxi
-
dation-specific epitopes are partiallydependent onT cell help. 
First, we demonstrated in culture that IL-5 stimulated B-1 cells 
to secrete IgM to oxidation-specific epitopes. Second, we found 
invivo
thatanti-OxLDLIgM titers were lower in age-matched 
T cell–deficientmice compared withwild-type mice(Supple
-
mental Figure 1) and that the B-1 cell transfer experiments with 
contaminating T cells resulted in more robust IgM titers in the 
recipients, including titers to OxLDL. Likely, the in vivo data can 
be explained in part by the secretion of IL-5, which we have pre
-
viously shown toprovide non-cognate help for the production 
ofE06/T15 IgMinvivoinatheroscleroticmice(25). However, 
both T cell–deficient miceand 
Rag1
–/–
 recipients of pure B-1 cells 
exhibited IgM Abs against oxidation-specific epitopes, indicating 
that T cell cytokines were not obligatory.
We also demonstrated that TLR4 and TLR2 agonists were par
-
ticularly effective in stimulating the generation of IgM by B-1 cells 
in culture, an effect that was in large part MyD88 dependent. A 
similar observation was reported by Genestier et al. while this arti
-
cle was in preparation (59). Remarkably, however, our data indi
-
cate that TLR activation seems tohave preferentiallyexpanded 
the numberof B-1 clones secreting IgM to oxidation-specific epi
-
topes as compared with thosesecreting total IgM or IgM to 
α1,3-
dextran (Figure 2A). This would imply that B-1 clones secreting 
such IgM are more responsive to activation by TLRs, suggesting 
evolutionary pressure linking generation of NAbs to oxidation-
specific epitopes and activation of innate PRRs. This is intriguing, 
as we and others have provided data suggesting that aside from 
products of exogenous pathogens, there are endogenous antigens 
capable of stimulatingsuch TLRs, including the oxidizedmoi
-
eties of OxLDL (60, 61). Obviously much study will be needed to 
explore such speculations.
NAbs havebeen said to be polyreactive and to bind tomany auto
-
antigens nonspecifically. However, we suggest that these observa
-
tions may be confounded by the fact that a given NAb against a 
specific oxidation-specific epitope may bind to the same structural 
modification on many different proteins or even lipids. For exam
-
ple, E06 binds to the PCs moiety on the surface of bacteria, as well 
as to the PC headgroup of OxPL, which is found in OxLDL in both 
the lipid phase and covalently bound to apoB. E06 binds to OxPL 
covalently bound to apo(a) as well as OxPL in the lipid phase of 
Lp(a) (62). It also binds to the OxPLs present on the cell surface 
of apoptoticcells (19) and to theOxPLs generated in the lungs 
of mice and humans infected with viruses (55). Thus, it binds not 
only to atherosclerotic lesions, but to a large number of tissues in 
which inflammation exists, presumably to the PC epitope gener
-
ated as result of lipid peroxidation and/or apoptosis. If onedidnot 
know the true identity of the epitope of the NAb E06, one would 
say this Ab was “polyreactive.” In this article, we identify an even 
more prominentset of naturalIgMdirectedagainst MDA epitopes 
(which are in fact a complex set of related epitopes) that are gener
-
atedas a consequence of lipid peroxidation. The highly reactive 
MDA may similarly modify a wide range of substrates, including 
proteins, lipids, and even DNA (63), generating MDA neoepitopes 
in a varietyofpathophysiological events,such as ischemia and 
reperfusion, diabetes, and atherosclerosis, as well as inflammatory 
events in the brain (64). Indeed, we speculate that many of the so-
called self-antigens that have been reported to be targets of NAbs 
may well be modified self-antigens.
Functionally,NAbs play an importantrole in providing the 
first line of defense against viral and bacterial pathogens (65–67). 
NAbs also have important activities in homeostasis (68), includ
-
ing immunoregulatory functions, tumor surveillance,and rec
-
ognition and removal ofsenescent cells, celldebris,and other 
self-antigens, which if persisting would be proinflammatory and 
immunogenic. Indeed,mice thatcannotsecrete IgM Abs and 
thus lack natural IgM have been found to more readily develop 
autoimmune disease when crossed on a susceptible background 
(69). Here we demonstrate that both murine natural IgM as well 
as human umbilical cord IgM bind apoptotic cells, but not nor
-
mal cells, in part through binding to oxidation-specific epitopes 
on theircell surface. IgM Abs in general have been shown to be 
research article
1346 The Journal of Clinical Investigation http://www.jci.org  Volume 119  Number 5  May 2009
required incomplement-mediated clearance ofapoptoticcells 
(70). In the present work we explicitly demonstrate that pure IgM 
NAbs facilitate macrophage uptake of apoptotic cells in vivo and, 
specifically, that the MDA-specific NAb NA-17 similarly promotes 
clearance. Previously, the OxPL-specificNAb E06/T15was also 
shown to facilitate efficient complement-mediated phagocytosis 
of apoptotic cells in vivo (71). These data demonstrate an impor
-
tant role for oxidation-specific natural IgM in mediating clearance 
of apoptotic cells. There are likely many other roles as well. Oxi
-
dized lipids have increasingly been shown to be proinflammatory. 
Apoptoticcells and their apoptotic blebs can activate endothelial 
cells,andE06can block theseeffects(19, 52). Recentlyit was 
shownthat multiplelung pathogens, such as chemical agents, 
H5N1 avian flu, or SARS, which are associated with high lethality 
due to acute respiratory distress syndrome, induce robust OxPL 
formation in the lung. In turn, the OxPLs induce lung injury and 
cytokine production by lung macrophages, and this latter effect 
was amelioratedbyE06/T15 invitro(55). As anotherexample, 
we have previouslyshown that enhancingthe titer ofE06/T15 
isatheroprotective(14), in part throughinhibiting the uptake 
ofOxLDLbymacrophages.We now show that NA-17alsohas 
the ability to inhibit binding of MDA-LDL in a similar manner. A 
great deal of work will be needed to assess the overall functional 
role of such oxidation-specific NAbs in atherosclerosis in particu
-
lar and in homeostasis in general.
Inconclusion,oxidation-specificepitopes constituteaprevi
-
ously unrecognized but important target of NAbs, and of innate 
immunity in general. NAbs may be beneficial not only for defense 
against pathogens but alsoinidentifying altered-self produced as a 
result of oxidative stress under inflammatory events. Understand
-
ing their role in health and disease may lead to novel diagnostic 
and possibly therapeutic approaches to deal with consequences of 
oxidative stress such as atherogenesis.
Methods
Animals. C57BL/6 and Rag1
–/–
 mice (The Jackson Laboratory) andLdlr
–/–
Rag1
–/–
mice (a gift from Godfrey Getzand Katherine Reardon,Univer-
sity of Chicago, Chicago, Illinois, USA), all on C57BL/6 background, were 
bred and maintained in our colony under SPF conditions unless otherwise 
noted. Plasma was obtained from 6-week-old female C57BL/6 T cell recep-
tor α
−/−
 (Tcra
–/–
) mice (courtesy of Stephen Hedrick, UCSD).
Fourteen- to 16-week-old germ-free Swiss Webster mice were maintained 
at the Göteborg University vivarium under strict gnotobiotic conditions. 
Sterility was monitoredbystool culturingand PCR for bacterial DNA. Con-
ventionally raised mice were transferred to gnotobiotic isolators at weaning 
and fed the same autoclaved chow diet. To obtain conventionalized mice, 
13- to 14-week-old germ-free mice were colonized with cecal content from 
12-week-old conventionally raised Swiss Webster donors for 14 days.
All experimental protocols were approved by the Animal Subjects Com-
mittee at UCSD; care and use of the germ-free mice was approved by the 
Göteborg University Animal Studies Committee.
B-1 cell isolation. Peritoneal exudate cells (PECs) from 6 -to 15-week-old 
naive C57BL/6 mice were harvested by peritoneal lavage using ice-cold 
PBS supplemented with 1% heat-inactivated FCS (Invitrogen). PECs were 
incubated with an anti–Fcγreceptor mAb (clone 2.4G2; BD Biosciences 
— Pharmingen)for 15minutesat4°Cbefore being stained with fluo-
rescently labeled mAbs to block nonspecific binding. PECs were stained 
withR-PE–labeled anti-CD19 (1D3),FITC-labeled anti-CD23(B3B4), 
and in some experiments PE-Cy5–labeled anti-CD3 (clone 145-2C11) (all 
from BD Biosciences— Pharmingen). B-1 cells were sorted to greater than 
99% purity using a FACSVantage SE cellsorter (BD) as the CD3
, CD19
+
, 
and CD23
 population.
B-1 cell cultures. Purified B-1 cells were seeded at 1 × 10
6
 cells per well in 
24-well flat-bottom plates in culture medium (RPMI 1640 medium con-
taining 10% heat-inactivated FCS, 10 mM HEPESbuffer, 2mM l-gluta-
mine, 0.05 mM 2-mercaptoethanol, 50 μg/ml gentamicin) in the presence 
and absence of the specific TLR4 agonist KdO
2
-Lipid A (100 ng/ml; Avanti 
Polar Lipids), a combination of the TLR2 agonists Pam3CSK4 (300 ng/ml) 
and FSL-1 (1μg/ml) (Invivogen), or vehicle control in triplicate in a final 
volume of 500 μl. Cells were incubated at 37°C/5% CO
2 
for up to 7 days.
Adoptive transfer of B-1 cells. B-1 cells were isolated from the peritoneumof 
6- to 15-week-old female donor C57BL/6 mice as described above. Purified 
B-1 cells were resuspended in PBS, and 0.5 or 1 × 10
6
 cells in 200 μl were 
injected into the peritoneal cavity of 6- to 15-week-old Rag1
–/–
 mice. Three 
to 4 donor mice were used for each recipient. Control Rag1
–/–
mice received 
an equal volume of PBS. Blood was collected via the retro-orbital plexus 
from recipient mice before and 4 and 10 weeks after transfer.
Flow cytometry. At the time of sacrifice, PECs and splenocytes were resus-
pended in staining buffer. After blocking with a specific anti–Fcγ receptor 
mAb (2.4G2) for 15 minutes at 4°C, 10
6
cells were stained with fluores-
centlylabeledmAbs specific forvarious surface markers (FITC-labeled 
anti–CD11b/Mac-1 [M1/70], PE-labeled anti-CD5, PerCP-Cy5.5–labeled 
anti-CD19 [1D3]; PE-labeled anti–mouse CD43 [S7], and APC- or FITC-
labeled anti-mouse IgM [II/41]; all from BD Biosciences — Pharmingen) in 
100-μl volumes of staining buffer for 30 minutes at 4°C in darkness, fol-
lowed byextensive washing. Cell populations were analyzed on a BD FACS-
Calibur or FACScan instrument. More than 0.5 ×10
5
 cells were analyzed 
per sample, with dead cells excluded by forward and side scatter. Surface 
marker analysis was performed using FlowJo software (Tree Star Inc.).
Measurement of Ab titers. Specific Ab titers to given antigens in plasma or 
cell culture supernatants were determined by chemiluminescent ELISA as 
previously described (14, 23). Ab dilution curves of Ab binding to plated 
antigens were determined by serial dilutions of plasma or culture superna-
tant, and a titer was defined as the highest dilution that yielded binding 
that was 2-fold greater than the background level. Purified rat anti-mouse 
IgM (II/41; BD Biosciences — Pharmingen) was used as capture Ab to mea-
sure total IgM levels. AP-labeled goat anti-mouse IgM (μchain specific) and 
anti-mouse IgG (γ chain specific) (Sigma-Aldrich) were used as detection 
Abs, as well as biotinylated rat anti-mouse IgM (R6-60.2; BD Biosciences 
— Pharmingen). To detect other Ig isotypes, rat anti-mouse IgG1 (A85-3), 
IgG2a/c (R11-89), IgG2b (R9-91), IgG3 (R2-38), and IgA (C10-3) were used 
as capture Abs; biotin-conjugated rat anti-mouse IgG1 (A85-1), IgG2a/c 
(R19-15), IgG2b (R12-3), IgG3 (R40-82), and IgA (C10-1) (all from BD Bio-
sciences — Pharmingen) were used as secondary Abs. To detect the levels of 
E06, a T15-specific anti-idiotype Ab (AB1-2) (72) was used as capture Ab, 
followed by incubation with AP-labeled goat anti-mouse IgM. Biotin-con-
jugated Abs were then detected with AP-conjugated neutravidin (Pierce, 
Thermo Scientific). Mouse anti-human IgG (G18-145) and IgM (G20-127; 
BD Biosciences — Pharmingen) were used as capture Abs to measure total 
IgM and IgG levels in humans. AP-labeled goat anti-human IgG and IgM 
(A3187 and A3437; Sigma-Aldrich) were used as detection Abs. The follow-
ing antigens were prepared as describedpreviously(29): copper sulfate–oxi-
dized LDL (CuOx-LDL), 4-HNE-LDL, MDA-LDL prepared from human 
LDL, and 4-HNE-MSA prepared from MSA. MAA-BSAwas prepared as 
described previously (73). α1,3-dextran was agift from John F. Kearney 
(University of Alabama at Birmingham, Birmingham, Alabama, USA). PC-
BSA and PC-KLH were from Biosearch Technologies Inc., and KLH was 
from Pierce Biotechnology.
Immunocompetition assays. The specificity of IgM Abs binding to MDA-
LDL was determined by competition immunoassays as described previ-
research article
The Journal of Clinical Investigation http://www.jci.org  Volume 119    Number 5  May 2009  1347
ously (14, 23). Plasma from B-1 cell–recipient Rag1
–/–
 mice was pooled and 
diluted to 1:100 for MDA-LDL and 1:200 for CuOx-LDL, or B-1 cell super-
natants were incubated overnight at 4°C in the presence and absence of 
increasing concentrations of competitors. Samples were then centrifuged 
at 15,800 g for 45 minutes at 4°C and supernatants analyzed for binding 
to the respective antigen by chemiluminescent ELISA.
To determine the percentage of total IgM in plasma binding to specific 
antigens, plasma samples were diluted in 1% BSA-TBS to yield a limit-
ingdilution foreachantigen, as determined in preliminary experiments. 
Diluted plasma sampleswere incubated overnight at 4°C intheabsence 
or presence of individual or combinations of antigens at a final concen-
tration of 250μg/ml. Thereafter, samples were centrifuged at 15,800 g
for 45 minutesat 4°Ctopelletimmunecomplexes andsupernatants 
analyzedfortotal IgMcontentbychemiluminescentELISA, usinga 
monoclonal ratanti-mouseIgM tocaptureanda polyclonalalkaline 
phosphatase–labeled (AP-labeled) goat anti-mouse IgM (μchain specific) 
for detection. Datawerecalculated astheamount of IgM remaining 
in the supernatant after absorption, expressed as a percentage of total 
IgM. In some experiments, we calculatedthe apparentAbavidity of IgM 
binding to a given antigen by analysis of the competition assays using 
the Klotz method (23).
ELISpot assay. The frequencies of total and antigen-specific IgM–secret-
ing splenocytes were quantified by ELISpot assay as described previously 
(14). Splenocytes were suspended in culture medium at 2 × 10
5
 and 1 × 10
5
cells/100 μl and cultured in triplicate in washed and blocked 96-wellMulti-
Screen-HA sterile nitrocellulose plates (Millipore) thathad been coated 
overnight with 4 μg/ml of rat anti-mouse IgM (II/41), MDA-LDL, OxLDL, 
4-HNE-LDL, orα1,3-dextran. After 22 hours incubation at 37°C/5% CO
2
, 
cells were removed by washing, and ISCsweredetected using biotinylated rat 
anti-mouse IgM, followed by HRP-streptavidin (Zymed Laboratories Inc., 
Invitrogen). Plates were developed using a tetramethylbenzidinemembrane 
substratesystem (KPL),andspotswerequantified using anautomated 
ImmunoSpot Image Analyzer (Cellular Technology Ltd.).
Apoptotic cells and immunofluorescence microscopy. Thymocytes harvested 
from C57BL/6 mice wereculturedin cell culture medium and induced 
to undergo apoptosis by 10 ng/ml PMA (Sigma-Aldrich) for 16 hours as 
describedpreviously (18).Plasma, NA-17, or control IgM (C48-6,anti-
KLH; BD Biosciences —Pharmingen) dilutedin 1% BSA-PBS was incu-
bated with apoptotic thymocytes for 30 minutes at 4°C. Cells were washed 
and incubated with FITC-labeled rat anti-mouse IgM (II/41) in 1% BSA-
PBS for 30 minutes and then washed again. For FACS analysis, cells were 
incubated with annexinV and 7-AAD (BD Biosciences— Pharmingen) 
for 15 minutes and immediately analyzed by BD FACSCanto. Umbilical 
plasma IgM binding to apoptotic Jurkat cells was detected by FITC-conju-
gated goat anti-human IgM (Jackson ImmunoResearch Laboratories Inc.). 
Apoptosis of Jurkat cells was induced with UV irradiation at 20 mJ/cm
2
, 
followed by further incubationof thecells in medium for 14–16hours 
before use. For immunofluorescence microscopy studies, cells were incu-
bated with2 μg/ml Hoechstdye (Sigma-Aldrich) for 10 minutes, fixed 
with 2% paraformaldehyde, and spun down on glass slides using a cyto-
spin (StatSpin). Images were captured by deconvolution microscopy using 
a DeltaVision deconvolution microscopic system operated by softWoRx 
software (Applied Precision) as described previously (18).
Immunocytochemistry. Frozen-embedded sections (embedded in Tissue-
Tek OCT compound [Sakura Finetek]) of aortic origin from cholesterol-fed 
Ldlr
–/–
Rag1
–/–
 mice were fixed with methanol, blocked with 2% goat serum, 
and stainedwith5 μg/mlMDA2, 1.6μg/mlEN2, or0.8 μg/ml NA-17 
(29), followed by a biotinylated goat anti-mouse IgG or anti-mouse IgM 
(Jackson ImmunoResearch Laboratories Inc.) to detect endogenous NAbs 
in the lesions. A Vectastain ABC-AP kit and a Vector Red AP chromogenic 
substrate(VectorLaboratories) were usedtovisualize the Abstaining. 
Slides were counterstained with Weigert’s iron hematoxylin (Richard-Allan 
Scientific, Thermo Scientific). Immunostaining of adjacentsections in the 
absence of primary Ab was used as a negative control.
Cloning and genetic analysis of hybridoma NA-17. Spleens of Rag1
–/–
 mice 
reconstituted with B-1cells were used to prepareB-1–derived hybrid-
omasusing techniques establishedinour laboratory(29). Productive 
hybridomas, by default, should only secrete NAbs. In brief, splenocytes 
from4  Rag1
–/
B-1recipientswerefused  with  theP3 ×63Ag8.653.1 
myeloma cellline. Primaryscreeningof supernatantswasperformed 
after 10 days of growth for the ability to secrete IgM and subsequently 
for IgMbinding to MDA-LDL by chemiluminescent ELISA. Selected 
hybridomas were thenclonedby limiting dilution(8).Here we report 
on clone NA-17, which was confirmed by DNA sequencing to be a NAb. 
Cloning and sequence analysis of IgM NA-17 hybridoma V
H 
and V
L
 was 
accomplished as previously described (12).
Human samples. Blood samples from healthy pregnant women and their 
newborn infants delivering at Magee-Womens Hospital (Pittsburgh, Penn-
sylvania, USA) were collected as part of a prospective study of preeclampsia, 
approved by the Institutional Review Board of the University of Pittsburgh, 
for which written informed consent was received from the mothers. Mater-
nal samples were collected in EDTA before delivery (mean, 10.5 hours) and 
cord venous samples by sterile aspiration of the umbilical vein after deliv-
ery of the infant. Samples were processed and aliquoted within 3 hours 
of collection and stored at –70°C until use. Ten uncomplicated nullipa-
rous maternal-infant pairs were selected for this preliminary analysis. The 
mothers (50% of mixed European descent and 50% African American) were 
23.4 ± 5.4 years of age, 39.9 ± 1.2 weeks gestational age (values are mean 
and SD). There were 6 male and 4 female infants.
In vivo uptake of apoptotic cells by peritoneal macrophages. The in vivo clear-
ance of apoptotic cells was assessed using a modification of the method 
described by Taylor et al. (32). Rag1
–/–
mice, adoptively transferred with 
B-1 cells or with PBS, were injected i.p. with 1 ml of sterile 3% thioglycol-
late to induce sterile peritonitis 15–19 weeks after B-1 cell transfer. Four 
days later, the mice were injected i.p. with 20 × 10
6
 fluorescently labeled 
apoptotic thymocytes (using 5-CMFDA; MolecularProbes, Invitrogen) 
in 200 μl PBS. The mice were sacrificed 40 minutes after injection, and 
peritoneal cells were recovered by lavage with 10 ml of ice-cold PBS with 
1% heat-inactivated FBS/10 mM EDTA. Macrophages were labeled with 
PE-conjugated F4/80 (BM8; eBioscience) and macrophage-specific uptake 
of apoptotic cells analyzed by FACS. Phagocytosis was expressed as the 
percentage of macrophages ingesting apoptotic cells. To test the ability 
of NA-17 tomediateenhanced uptake, fluorescently labeled apoptotic 
thymocytes were preincubated with cultured NA-17 hybridoma superna-
tant or anti-KLH IgM (C48-6; BD Biosciences — Pharmingen) at 5 μg/ml 
for 1 hour, washed, and then injected into 8- to 10-week-old Rag1
–/–
 mice.
Macrophage binding assay. Binding of biotinylated MDA-LDLtoJ774 
macrophages plated in microtiter wells was assessed by a chemilumines-
cent binding assay as described previously (60). Biotinylated MDA-LDL 
(2 μg/ml) was incubated in the absence or presence of NA-17 or anti-KLH 
IgM (C48-6; BD Biosciences — Pharmingen) at different concentrations 
overnight. Samples werethen centrifuged at 15,800 gfor 45 minutes at 
4°C. The harvested supernatants were then added to macrophages and the 
binding of biotinylated MDA-LDL determined by ELISA.
Statistics. Statistical tests used to analyze for significance are described in 
the figure legends. The tests used were: 1-wayor repeated measures ANOVA 
with Tukey-Kramer multiple comparison tests; Wilcoxon matched-pairs 
test; and paired ttest, using 2-tailed levels of significance. Results wereana-
lyzed with InStat 3 forMacintosh (GraphPad Software). A P value less than 
0.05 was considered significant.
research article
1348 The Journal of Clinical Investigation http://www.jci.org  Volume 119  Number 5  May 2009
Acknowledgments
This work was supported by American Heart Association (AHA) 
Scientist  Development  Awards  0530159N  (to  Y.I.  Miller), 
0630228N (to K. Hartvigsen), and 0430127N (to P.X. Shaw) and 
AHA Postdoctoral Award 0625133Y (to L.F. Hansen); NIH grants 
HL086559(to J.L.Witztum), P50 HL056989(to J.L. Witztum), 
and P01 HL088093 (to J.L. Witztum, Y.I. Miller, and C.J. Binder); 
the Austrian Academy of Sciences (to C.J. Binder); the Fondation 
Leducq (toJ.L. Witztum andC.J.Binder); and grants from the 
Swedish Heart and Lung Foundation, theSwedish Society of Med
-
icine, and the Gothenburg Medical Society (to L. Fogelstrand).
Received forpublicationJuly 16, 2008,and acceptedinrevised 
form February 25, 2009.
Address correspondence to: Joseph L. Witztum, University of Cali
-
fornia, San Diego, Basic Science Building, Room 1080, 9500 Gil
-
man Drive, La Jolla, California 92093, USA. Phone: (858)534-4347; 
Fax: (858) 534-2005; E-mail: jwitztum@ucsd.edu. Or to: Christoph 
J. Binder, CeMM and Department of Medical and Chemical Labo
-
ratory Diagnostics, Medical University of Vienna, Währinger Gür
-
tel 18-20, A-1090 Vienna, Austria. Phone: 43-1-40400-6441; Fax: 
43-1-40400-2097; E-mail: christoph.binder@meduniwien.ac.at.
1. Binder, C.J., et al. 2002. Innate and acquired immu-
nity in atherogenesis. Nat. Med.8:1218–1226.
2. Hansson, G.K.,andLibby, P. 2006. The immune 
response in atherosclerosis: a double-edged sword. 
Nat. Rev. Immunol.6:508–519.
3. Getz, G.S., Vanderlaan,P.A., and Reardon, C.A. 
2007. The immune system and murine atheroscle-
rosis. 
Curr. Drug Targets.8:1297–1306.
4. Binder, C.J., et al. 2005. The role of natural antibod-
ies in atherogenesis. J. Lipid Res.46:1353–1363.
5. Boullier, A., et al. 2005. Phosphocholine as a pat-
tern  recognition  ligand for  CD36. 
J. Lipid Res.
46:969–976.
6. Gillotte-Taylor, K., Boullier, A., Witztum, J.L., Stein-
berg, D.,andQuehenberger, O. 2001. Scavenger 
receptor classB type I as a receptor for oxidizedlow 
density lipoprotein. J. Lipid Res.42:1474–1482.
7. Plüddemann, A., Neyen, C., and Gordon, S. 2007. 
Macrophage scavenger receptors and host-derived 
ligands. 
Methods.43:207–217.
8. Palinski,W., etal.1996. Cloning of monoclonal 
autoantibodies to epitopes of oxidized lipoproteins 
from apolipoprotein E-deficient mice. Demonstra
-
tion of epitopes of oxidized lowdensity lipoprotein 
in human plasma. 
J. Clin. Invest.98:800–814.
9. Hörkkö, S., Miller, E., Branch, D.W., Palinski, W., 
and Witztum,  J.L.  1997. The  epitopesforsome 
antiphospholipidantibodies are adducts of oxi
-
dized phospholipid and beta2 glycoprotein  1 
(and other proteins). 
Proc. Natl. Acad. Sci. U. S. A.
94:10356–10361.
10. Boullier, A., et al. 2000. The binding of oxidized low 
density lipoprotein to mouse CD36 is mediated in 
part by oxidized phospholipids that are associated 
with both the lipidand protein  moieties of  the 
lipoprotein. 
J. Biol. Chem.275:9163–9169.
11. Baumgarth, N., Tung, J.W., and Herzenberg, L.A. 
2005. Inherent specificities in natural antibodies: a 
key to immune defense against pathogen invasion. 
Springer Semin. Immunopathol.26:347–362.
12. Shaw, P.X., et al. 2000. Natural antibodies with the 
T15 idiotype may act in atherosclerosis, apoptotic 
clearance, and protective immunity. 
J. Clin. Invest.
105:1731–1740.
13. Briles, D.E., Forman, C., Hudak, S., and Claflin, J.L. 
1982. Anti-phosphorylcholine antibodies of the 
T15 idiotype are optimally protective against Strep
-
tococcus pneumoniae. J. Exp. Med.156:1177–1185.
14. Binder, C.J., et al. 2003. Pneumococcal vaccination 
decreases atherosclerotic lesion formation: molec-
ular mimicry between Streptococcus pneumoniae 
and oxidized LDL. 
Nat. Med.9:736–743.
15. Kagan, V.E., et al.2004.Oxidative lipidomics of 
apoptosis:  redox  catalyticinteractionsof  cyto-
chrome c with cardiolipin and phosphatidylserine. 
Free Radic. Biol. Med.37:1963–1985.
16. Greenberg, M.E., et al. 2006. Oxidized phosphati-
dylserine-CD36 interactions play an essential role 
in macrophage-dependent phagocytosis of apop-
totic cells. J. Exp. Med.203:2613–2625.
17. Savill, J., Dransfield, I., Gregory, C., and Haslett, C. 
2002. A blast from the past: clearance of apoptotic 
cells regulates immune responses. 
Nat. Rev. Immu-
nol.2:965–975.
18. Chang, M.K., et al. 2004. Apoptotic cells with oxi-
dation-specific epitopes are immunogenic and pro
-
inflammatory. J. Exp. Med.200:1359–1370.
19. Chang,M.K., et al. 1999. Monoclonal antibodies 
againstoxidized low-density lipoprotein bind to 
apoptoticcellsand inhibittheirphagocytosis by 
elicited macrophages: evidence that oxidation-spe
-
cific epitopes mediate macrophagerecognition. 
Proc. Natl. Acad. Sci. U. S. A.96:6353–6358.
20. Chang, M.K., Binder, C.J., Torzewski, M., and Wit-
ztum, J.L.2002. C-reactive proteinbinds to both 
oxidized LDL and apoptotic cells through recog
-
nition of a common ligand: phosphorylcholine of 
oxidized phospholipids. Proc. Natl. Acad. Sci. U. S. A.
99:13043–13048.
21. Kawahara, T., Ohdan, H., Zhao, G., Yang, Y.G., and 
Sykes, M. 2003. Peritoneal cavity B cells are precur
-
sors of splenicIgM natural antibody-producing 
cells. J. Immunol.171:5406–5414.
22. Hörkkö, S., et al. 1999. Monoclonal autoantibod
-
ies specific for oxidized phospholipids or oxidized 
phospholipid-proteinadducts inhibit macrophage 
uptake of oxidized low-density lipoproteins. 
J. Clin.
Invest.103:117–128.
23. Friguet, B., et al. 1985. Measurements of the true 
affinity constant in solution of antigen-antibody 
complexes by enzyme-linked  immunosorbent 
assay. 
J. Immunol. Methods.77:305–319.
24. Masmoudi, H., Mota-Santos, T., Huetz, F., Coutin-
ho, A., and Cazenave, P.A. 1990. All T15 Id-positive 
antibodies (but not the majority of VHT15+ anti
-
bodies) are producedby peritoneal CD5+ B lym-
phocytes. Int. Immunol.2:515–520.
25. Binder, C.J.,etal. 2004.IL-5 linksadaptiveand 
natural immunity specific for epitopes of oxidized 
LDL  and  protects  from atherosclerosis. 
J. Clin.
Invest.114:427–437.
26. Manz, R.A., Hauser, A.E., Hiepe, F., and Radbruch, 
A. 2005. Maintenance ofserum antibody levels. 
Annu. Rev. Immunol.23:367–386.
27. Shaw, P.X., etal.2001. Human-derivedanti-oxi
-
dized LDL autoantibody blocks uptake of oxidized 
LDL by macrophages and localizes to atheroscle-
rotic lesions in vivo. Arterioscler. Thromb. Vasc. Biol.
21:1333–1339.
28. Tabas, I. 2005.  Consequences and therapeutic 
implications  of  macrophage apoptosis  in ath
-
erosclerosis: theimportanceoflesion stage and 
phagocytic efficiency. Arterioscler. Thromb. Vasc. Biol.
25:2255–2264.
29. Ylä-Herttuala, S., et al. 1989. Evidence for the pres
-
ence of oxidatively modified low density lipopro-
tein in atherosclerotic lesions of rabbit and man. 
J. Clin. Invest.84:1086–1095.
30. Palinski, W., et al. 1990. Antisera and monoclonal 
antibodies specific for epitopes generated during 
oxidative modification of low density lipoprotein. 
Arteriosclerosis.10:325–335.
31. Merbl,  Y.,  Zucker-Toledano,  M.,Quintana,F.J., 
and Cohen, I.R. 2007. Newborn humans manifest 
autoantibodies to defined self molecules detected 
by antigen microarray informatics. 
J. Clin. Invest.
117:712–718.
32. Taylor, P.R., et al. 2000. A hierarchical role for clas-
sicalpathway complement proteinsin the clearance 
of apoptotic cells in vivo. 
J. Exp. Med.192:359–366.
33. Miller, Y.I., et al. 2005. Toll-like receptor 4-depen
-
dent and -independent cytokine secretion induced 
by minimally oxidized  low-density  lipoprotein 
in macrophages. 
Arterioscler. Thromb. Vasc. Biol.
25:1213–1219.
34. Miller, Y.I., Chang, M.K., Binder, C.J., Shaw, P.X., 
and Witztum, J.L. 2003. Oxidized low density lipo-
protein and innate immune receptors. 
Curr. Opin.
Lipidol.14:437–445.
35. Haury, M., et al. 1997. The repertoire of serum IgM 
in normal mice is largely independent of external 
antigenic contact. Eur. J. Immunol.27:1557–1563.
36. Vasconcellos,R., Nobrega, A., Haury, M., Viale, A.C., 
and Coutinho, A. 1998. Genetic control of natural 
antibody repertoires: I. IgH, MHCand TCRbeta 
loci. 
Eur. J. Immunol.28:1104–1115.
37. Mouthon, L., et al. 1995. Invariance and restriction 
toward a limited set of self-antigens characterize 
neonatal IgM antibody repertoires and prevail in 
autoreactive  repertoiresof  healthy adults.
Proc.
Natl. Acad. Sci. U. S. A.92:3839–3843.
38. Jayasekera, J.P., Moseman, E.A., and Carroll, M.C. 
2007. Natural antibody and complement mediate 
neutralization of influenza virus in the absence of 
prior immunity. 
J. Virol.81:3487–3494.
39. Hooijkaas, H., Benner, R.,  Pleasants,  J.R.,  and 
Wostmann, B.S. 1984. Isotypes and specificities of 
immunoglobulins produced by germ-free mice fed 
chemically defined ultrafiltered“antigen-free” diet. 
Eur. J. Immunol.14:1127–1130.
40. Avrameas, S. 1991. Natural autoantibodies: from 
“horror autotoxicus” to “gnothi seauton.” 
Immu-
nol. Today.12:154–159.
41. Hardy, R.R., and Hayakawa, K. 2005. Development 
of B cells producing natural autoantibodies to thy
-
mocytes andsenescent erythrocytes. Springer Semin.
Immunopathol.26:363–375.
42. Kearney, J.F.  2005. Innate-like  B  cells. 
Springer
Semin. Immunopathol.26:377–383.
43. Montecino-Rodriguez, E., Leathers, H., and Dorsh-
kind, K. 2006. Identification of a B-1 B cell-speci-
fied progenitor. Nat. Immunol.7:293–301.
44. Herzenberg, L.A., and Tung, J.W.2006. B cell lineag
-
es: documented at last! Nat. Immunol.7:225–226.
45. Tung, J.W., Mrazek, M.D., Yang, Y.,Herzenberg, 
L.A.,and Herzenberg, L.A. 2006. Phenotypically 
distinct B cell development pathways map to the 
three B cell lineages in the mouse. 
Proc. Natl. Acad.
Sci. U. S. A.103:6293–6298.
46. Dorshkind, K., and Montecino-Rodriguez, E. 2007. 
Fetal B-cell lymphopoiesis andthe emergence of 
B-1-cell potential. 
Nat. Rev. Immunol.7:213–219.
47. Hardy,  R.R.,Wei,  C.J.,and  Hayakawa,K.  2004. 
Selection during development of VH11+ B cells: a 
model for natural autoantibody-producing CD5+ 
B cells. Immunol. Rev.197:60–74.
48. Qian, Y., et al. 2006. Autoreactive MZ andB-1 B-cell 
activation by Faslpr is coincident with an increased 
frequency of apoptotic lymphocytes and a defect in 
research article
The Journal of Clinical Investigation http://www.jci.org  Volume 119    Number 5  May 2009  1349
macrophage clearance. Blood.108:974–982.
49. Bos,N.A.,et  al.1987. Early development  of Ig-
secreting cells in young of germ-free BALB/c mice 
fed a  chemically  defined  ultrafiltered diet. 
Cell.
Immunol.105:235–245.
50. Thurnheer, M.C., Zuercher, A.W., Cebra, J.J., and 
Bos, N.A. 2003. B1 cells contribute to serum IgM, 
but not  to  intestinalIgA, production  in  gno
-
tobiotic  Ig allotypechimericmice.
J. Immunol.
170:4564–4571.
51. Kearney, J.F. 2000. Immune recognition of OxLDL 
in atherosclerosis. 
J. Clin. Invest.105:1683–1685.
52. Huber, J., et al. 2002. Oxidized membrane vesicles 
and blebs from apoptotic cells contain biologically 
active oxidized phospholipids that induce mono
-
cyte-endothelial interactions.Arterioscler. Thromb.
Vasc. Biol.22:101–107.
53. Houglum, K.,  et  al. 1997.  Excess iron induces 
hepatic oxidative stress and transforming growth 
factor beta1 in genetic hemochromatosis. 
Hepatol-
ogy.26:605–610.
54. Horie, K., et al. 1997. Immunohistochemical colo-
calization of glycoxidation products and lipid per
-
oxidation products in diabetic renal glomerular 
lesions. Implication for glycoxidative stress in the 
pathogenesis ofdiabetic nephropathy.J. Clin. Invest.
100:2995–3004.
55. Imai, Y.,etal.2008.Identification of oxidative 
stressandToll-like receptor 4 signaling asa key 
pathway of acute lung injury. Cell.133:235–249.
56. Newcombe, J., Li, H., and Cuzner, M.L. 1994. Low 
density lipoprotein uptake by macrophages in mul
-
tiple sclerosis plaques: implications for pathogen-
esis. Neuropathol. Appl. Neurobiol.20:152–162.
57. Dei,R., et al. 2002. Lipid peroxidation and advanced 
glycation endproductsin the brain in normalaging 
and in Alzheimer’s disease. 
Acta Neuropathol. (Berl.). 
104:113–122.
58. Berland, R., and Wortis,H.H. 2002. Originsand 
functions ofB-1cells with notes on  the  roleof 
CD5. 
Annu. Rev. Immunol.20:253–300.
59. Genestier, L., et al. 2007. TLR agonists selectively 
promote terminal plasma cell differentiationof 
B cell subsets specialized in thymus-independent 
responses. 
J. Immunol.178:7779–7786.
60. Miller, Y.I., et al. 2003. Minimally modified LDL 
binds to CD14, induces macrophage spreading via 
TLR4/MD-2, andinhibits phagocytosis of apop
-
totic cells. 
J. Biol. Chem.278:1561–1568.
61. Bae, Y.S., et al. 2009. Macrophages generate reactive 
oxygen species in response to minimally oxidized 
low-density lipoprotein. Toll-like receptor 4- and 
spleentyrosinekinase-dependent  activation  of 
NADPH oxidase 2. 
Circ. Res.104:210–218.
62. Bergmark, C., et al. 2008.  A novel function  of 
lipoprotein [a] as a preferential carrier of oxidized 
phospholipids in human plasma.
J. Lipid Res. 
49:2230–2239.
63. Blair, I.A. 2008. DNA-adducts with lipid peroxida-
tion products. J. Biol. Chem. 283:15545–15549.
64. Esterbauer, H., Schaur, R.J., and Zollner, H. 1991. 
Chemistry and biochemistry of 4-hydroxynonenal, 
malonaldehyde and related aldehydes.
Free Radic.
Biol. Med.11:81–128.
65. Ochsenbein, A.F., et al. 1999. Control of early viral 
and bacterial distribution and disease by natural 
antibodies. 
Science.286:2156–2159.
66. Boes, M., Prodeus,  A.P., Schmidt,  T., Carroll, 
M.C., and Chen,J.1998.A critical roleofnatu
-
ral immunoglobulinM in immediate defense 
againstsystemic bacterial infection. J. Exp. Med.
188:2381–2386.
67. Baumgarth, N., etal. 1999.Innateand acquired 
humoral immunities to influenza virus are medi
-
ated by distinct arms of the immune system. Proc.
Natl. Acad. Sci. U. S. A.96:2250–2255.
68. Lutz,H.U. 2007. Homeostatic rolesofnaturally 
occurring antibodies: an overview. 
J. Autoimmun.
29:287–294.
69. Boes, M., et al. 2000. Accelerated development of 
IgG autoantibodies and autoimmune disease inthe 
absence of secreted IgM.
Proc. Natl. Acad. Sci. U. S. A.
97:1184–1189.
70. Quartier, P.,Potter, P.K.,Ehrenstein,M.R., Walport, 
M.J., and Botto, M. 2005. Predominant role of IgM-
dependent activationof the classicalpathway in 
the clearance of dying cells by murine bone mar
-
row-derived macrophages in vitro. 
Eur. J. Immunol.
35:252–260.
71. Ogden, C.A., Kowalewski, R., Peng, Y., Montenegro, 
V., and Elkon, K.B. 2005. IGM is required for effi
-
cient complement mediated phagocytosis of apop-
totic cells in vivo. Autoimmunity.38:259–264.
72. Kearney, J.F., Barletta, R., Quan, Z.S., andQuintans, 
J.1981. Monoclonal vs.heterogeneous anti-H-8 
antibodies in the analysis of the anti-phosphoryl
-
choline response in BALB/c mice. Eur. J. Immunol.
11:877–883.
73. Xu,  D.,et  al.1997.  Epitope  characterizationof 
malondialdehyde-acetaldehyde adducts usingan 
enzyme-linked immunosorbentassay. 
Chem. Res.
Toxicol.10:978–986.
    • "We postulate that the greater predictive power of total Ig levels reflects a broad role of the polyreactive antibody repertoire in cardiovascular protection. On the one hand, the composite antibody repertoire can be expected to contribute to the safe disposal of a variety of other endogenous atherogenic autoantigens besides modified LDL, including apoptotic cells and microparticles (Chang et al., 1999; Chou et al., 2009; Tsiantoulas et al., 2015). On the other hand, the conventional role of antibodies in preventing bacterial infection may also be important for reducing the rate of atherosclerosis progression and reducing infection-triggered acute cardiac events (Smeeth et al., 2004). "
    [Show abstract] [Hide abstract] ABSTRACT: Aims: We aimed to determine whether the levels of total serum IgM and IgG, together with specific antibodies against malondialdehyde-conjugated low-density lipoprotein (MDA-LDL), can improve cardiovascular risk discrimination. Methods and results: The Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT) randomized 9098 patients in the UK and Ireland into the Blood Pressure-Lowering Arm. 485 patients that had cardiovascular (CV) events over 5.5years were age and sex matched with 1367 controls. Higher baseline total serum IgG, and to a lesser extent IgM, were associated with decreased risk of CV events (IgG odds ratio (OR) per one standard deviation (SD) 0.80 [95% confidence interval, CI 0.72,0.89], p<0.0001; IgM 0.83[0.75,0.93], p=0.001), and particularly events due to coronary heart disease (CHD) (IgG OR 0.66 (0.57,0.76); p<0.0001, IgM OR 0.81 (0.71,0.93); p=0.002). The association persisted after adjustment for a basic model with variables in the Framingham Risk Score (FRS) as well as following inclusion of C-reactive protein (CRP) and N-terminal pro-B-type natriuretic peptide (NtProBNP). IgG and IgM antibodies against MDA-LDL were also associated with CV events but their significance was lost following adjustment for total serum IgG and IgM respectively. The area under the receiver operator curve for CV events was improved from the basic risk model when adding in total serum IgG, and there was improvement in continuous and categorical net reclassification (17.6% and 7.5% respectively) as well as in the integrated discrimination index. Conclusion: High total serum IgG levels are an independent predictor of freedom from adverse cardiovascular events, particularly those attributed to CHD, in patients with hypertension.
    Full-text · Article · Jun 2016
    • "Even in the presence of highly immunogenic adjuvant CFA, the IgM bias could not be completely overcome; thus, it is likely that the bias represents an intrinsic propensity of the HMW4 epitope, not failure in class switch per se. In atherosclerosis, only oxidized lipids, such as the oxidized phosphatidylcholine in oxLDL, have been shown to display a similar IgM-bias [45]. In that case, the IgMs are produced by B-1 cells as " natural Abs, " considered to be atheroprotective in general. "
    [Show abstract] [Hide abstract] ABSTRACT: Background and aims: Anti-HMGB1 autoimmunity plays a role in systemic lupus erythematosus (SLE). Because SLE increases atherosclerosis, we asked whether the same autoimmunity might play a role in atherogenesis. Methods: We looked for the induction of HMGB1-specific B and T cell responses by a western-type diet (WTD) in the Apoe(-/-) mouse model of atherosclerosis. We also determined whether modifying the responses modulates atherosclerosis. Results: In the plasma of male Apoe(-/-) mice fed WTD, the level of anti-HMGB1 antibodies (Abs) was detected at ∼50 μg/ml, which was ∼6 times higher than that in either Apoe(-/-) mice fed a normal chow or Apoe(+/+) mice fed WTD (p ≤ 0.0005). The Abs were directed largely toward a novel, dominant epitope of HMGB1 named HMW4; accordingly, compared with chow-fed mice, WTD-fed Apoe(-/-) mice had more activated HMW4-reactive B and T cells (p = 0.005 and p = 0.01, respectively). Compared with mock-immunized mice, Apoe(-/-) mice immunized with HMW4 along with an immunogenic adjuvant showed proportional increases in anti-HMW4 IgG and IgM Abs, HMW4-reactive B-1 and B-2 cells, and HMW4-reactive Treg and Teff cells, which was associated with ∼30% increase in aortic arch lesions (p ≤ 0.01) by two methods. In contrast, Apoe(-/-) mice immunized with HMW4 using a tolerogenic adjuvant showed preferential increases in anti-HMW4 IgM (over IgG) Abs, HMW4-reactive B-1 (over B-2) cells, and HMW4-specific Treg (over Teff) cells, which was associated with ∼40% decrease in aortic arch lesions (p ≤ 0.03). Conclusions: Anti-HMGB1 autoimmunity may potentially play a role in atherogenesis.
    Full-text · Article · May 2016
    • "In healthy people, a significant portion of IgM is presented by natural IgM antibodies. In mice, 30% of all natural antibodies were reported to be reactive with OSEs (Chou et al., 2009; Miller et al., 2011). OSEs could be found in oxLDL, apoptotic cells, and modified proteins. "
    [Show abstract] [Hide abstract] ABSTRACT: Adaptive immune response plays an important role in atherogenesis. In atherosclerosis, the proinflammatory immune response driven by Th1 is predominant but the anti-inflammatory response mediated mainly by regulatory T cells is also present. The role of Th2 and Th17 cells in atherogenesis is still debated. In the plaque, other T helper cells can be observed such as Th9 and Th22 but is little is known about their impact in atherosclerosis. Heterogeneity of CD4+ T cell subsets presented in the plaque may suggest for plasticity of T cell that can switch the phenotype dependening on the local microenvironment and activating/blocking stimuli. Effector T cells are able to recognize self-antigens released by necrotic and apoptotic vascular cells and induce a humoral immune reaction. Tth cells resided in the germinal centers help B cells to switch the antibody class to the production of high-affinity antibodies. Humoral immunity is mediated by B cells that release antigen-specific antibodies. A variety of B cell subsets were found in human and murine atherosclerotic plaques. In mice, B1 cells could spontaneously produce atheroprotective natural IgM antibodies. Conventional B2 lymphocytes secrete either proatherogenic IgG, IgA, and IgE or atheroprotective IgG and IgM antibodies reactive with oxidation-specific epitopes on atherosclerosis-associated antigens. A small population of innate response activator (IRA) B cells, which is phenotypically intermediate between B1 and B2 cells, produces IgM but possesses proatherosclerotic properties. Finally, there is a minor subset of splenic regulatory B cells (Bregs) that protect against atherosclerotic inflammation through support of generation of Tregs and production of anti-inflammatory cytokines IL-10 and TGF-β and proapoptotic molecules.
    Full-text · Article · May 2016
Show more

    Recommended publications

    Discover more