The role of b cells in the pathogenesis of atherosclerosis

3.9.2 Evaluation of splenic extrafollicular responses in apoE -/- mice after ezetimibe treatment Having established that the total IgM and oxLDL-specific IgM autoantibodies in the plasm

3.9 Evaluation of humoral responses being associated with

hypercholesterolemic setting in the apoE -/- mice

In many studies, the deletion/defective of certain genes led to

autoimmune disorder For example, SLE mice models such as in MRL/lpr mice where CD95 is defective due to lpr mutation (Shlomchik et al., 1987)

lead to high autoantibodies production and in Roquinsan/san mice where the excessive accumulation of follicular T helper cells in the GCs lead to autoantibodies production (Linterman et al., 2009) Therefore, elevated titer of oxLDL-specific IgM antibodies could be a spontaneous consequence of lack

of APOE Therefore, it is imperative to investigate whether the elevated IgM antibodies response is directly associated with the progression of the disease and not due to the APOE deficiency

In order to investigate this hypothesis, we impeded the progression of the disease by using a blood cholesterol-lowering drug, ezetimibe The drug targets the molecule Niemann-Pick C1 Like 1 (NPC1L1) and, inhibits the uptake and absorption of dietary cholesterol in the intestine to be excreted out

(Garcia-Calvo et al., 2005) Thus, the administration of ezetimibe to apoE

-/-mice led to amelioration of the disease (Nakagami et al., 2009) Therefore, in

our disease regression model, 12 weeks old apoE -/- mice, which already exhibit certain extent of the disease, were administered with ezetimibe via oral gavage daily as therapeutic intervention for 12 weeks before analysis

3.9.1 Evaluation of total IgM and oxLDL-specific IgM autoantibodies in

apoE -/- mice after ezetimibe treatment

Plasma of the treated mice was examined after ezetimibe treatment

Non-treated apoE -/- and non-treated WT mice were administered with corn oil, the vehicle for ezetimibe Our ELISA data had shown that with ezetimibe

treatment, the amount of total IgM antibodies in the apoE -/- mice had

decreased to the level of WT control mice compared to non-treated apoE

-/-mice (Figure 32A) When oxLDL-specific IgM was examined in the plasma,

ezetimibe treated apoE -/- mice had lower titer compared to non-treated apoE

-/-mice and the titer was similar to that of control groups: non-treated WT and

WT mice (Figure 32B)

The amount of total IgM and oxLDL-specific IgM autoantibodies in the non-treated WT was similar to that of WT control mice, suggesting the

vehicle does not alter the level of total IgM (Figure 32A & 32B) Therefore,

the vehicle alone was no longer administered in non-treated mice and only 3 groups of mice were used for our subsequent disease regression experiments;

non-treated apoE -/- , treated apoE -/- and WT

Thus, our data on the plasma suggests that the elevated total IgM and

oxLDL-specific IgM autoantibodies in apoE -/- mice was associated with hypercholesterolemia

Figure 32 Total IgM and oxLDL-specific IgM autoantibodies were associated with hypercholesterolemia

(A) Quantification of total IgM in plasma determined by ELISA (mean ± SEM; n=8-10) Data were pooled from two independent experiments (B)

Plasma IgM titres against oxLDL determined by ELISA (n=5-7) * P < 0.05;

** P < 0.01; *** P < 0.001

3.9.2 Evaluation of splenic extrafollicular responses in apoE -/- mice after

ezetimibe treatment

Having established that the total IgM and oxLDL-specific IgM

autoantibodies in the plasma of apoE -/- mice had decreased after ezetimibe

treatment, we next examined if this was due to reduced extent of

extrafollicular responses in the spleen of the treated mice Indeed, our

immunofluorescence staining of the spleen to reveal extrafollicular responses

showed that there were a reduced number of extrafollicular response sites

being identified compared to the non-treated apoE -/- mice (Figure 33A)

When quantification of the extent of extrafollicular responses was performed,

we observed statistical significant decrease of the extrafollicular responses in

the treated apoE -/- mice, similar to that of the WT control mice (Figure 33B)

Therefore, robust extrafollicular responses in the spleen was a

consequence of hypercholesterolemia to produce elevated amount of total IgM

and oxLDL-specific IgM autoantibodies in the plasma of apoE -/- mice

Figure 33 Splenic extrafollicular responses were associated with hypercholesterolemia

(A) Representative images of area analyzed for extrafollicular sites identified

(white circles) in which CD138+ plasmablasts (red) colocalized with CD11c+DCs (blue) at bridging channel of B220+ follicles (green) between non-treated

apoE -/- (n=4), treated apoE -/- (n=6) and WT (n=6) Scale bar denotes 500 µm;

100X magnification (B) Quantification of extrafollicular responses sites in

spleen after ezetimibe treatment (n=4-6) Data were pooled from three independent experiments *** P < 0.001

3.9.3 Evaluation of IgM + antibody secreting cells in the spleen of apoE

-/-mice after ezetimibe treatment

Our earlier studies established that there was an increased frequency of

total IgM and oxLDL-specific IgM ASCs present in the spleen of apoE -/- mice

(Figure 19A & 19B, left) Together with the observations that in the spleen of

apoE -/- mice treated with ezetimibe had decreased extent of extrafollicular responses, prompted us to investigate if the frequency of IgM+ ASCs would be affected

Our ELISpot analysis on the frequency of total IgM ASCs in the

spleen of treated apoE -/- mice displayed similar frequency to that of

non-treated apoE -/- mice but a non-statistical significant increase in frequency

when compared to WT control mice (Figure 34A) In addition, when we

examined for oxLDL-specific IgM+ ASCs, the frequency of treated apoE was similar to that of non-treated apoE -/- and still maintained its statistical

-/-significant increase compared to WT control mice (Figure 34B)

While the frequency of total IgM and oxLDL-specific IgM ASCs

remained essentially unchanged between non-treated and treated apoE -/- mice,

we investigated if the amount of IgM antibodies secreted was affected to account for the decreased in total IgM and oxLDL-specific IgM antibodies in

circulation ELISA analysis of in vitro culture supernatant of splenocytes from treated apoE -/- revealed a non-statistical decrease but almost two-fold lower

production of total IgM antibodies compared to non-treated apoE -/- mice

(Figure 34C) ELISA analysis on the oxLDL-specific IgM autoantibodies

production from the in vitro culture supernatant of treated apoE -/- mice

revealed a statistical significant decrease in titer compared non-treated apoE

-/-mice

Therefore, our data on IgM+ ASCs indicates that while the frequency

of total IgM and oxLDL-specific IgM ASCs remained unchanged after ezetimibe treatment, the IgM antibodies production from ASCs from treated

apoE -/- mice was decreased This further suggests that IgM antibodies

production per cell basis was lower in the treated apoE -/- mice However, the spots identified to denote frequency of ASCs in ELISpot analysis do not necessarily equate one spot is equivalent to one cell In other words, cells that are capable of proliferation to form many cells could still form a single spot analogous to a bacteria colony forming assay experiment Together with earlier findings that splenic extrafollicular responses, where plasmablasts

proliferate and differentiate into plasma cells, were lower in treated apoE mice compared to non-treated apoE -/- mice, we reasoned that the lower IgM antibodies production from the culture supernatant of splenocytes from treated

-/-apoE -/- mice was due to a lack of proliferation of these IgM+ ASCs Indeed, in our EdU-pulsed chase experiment, we observed a lower percentage and number of B220-CD138+EdU+IgM+ plasma cells in the spleen of treated apoE - /- mice, supporting our hypothesis that proliferating IgM+ ASCs provided

massive amount of total IgM and oxLDL-specific IgM autoantibodies in apoE /- mice (Figure 34E & 34F)

-Figure 34 Elevated total IgM and anti-oxLDL IgM autoantibodies were mostly contributed by proliferating IgM + ASCs

(A-B) Frequency of splenic (A) total IgM and (B) anti-oxLDL specific IgM

per million cells in non-treated apoE -/- (n=15), treated apoE -/- (n=17) and WT mice (n=19) determined by ELISpot Data were pooled from five independent

experiments (C-D) Quantification of (C) total IgM and (D) anti-oxLDL IgM

in culture supernatant of splenocytes from non-treated apoE -/- (n=7), treated

apoE -/- (n=8) and WT mice (n=10) determined by ELISA (E-F) Comparative flow cytometry analysis on (E) percentage and (F) number of B220-

CD138+EdU+IgM+ plasma cells in spleen of non-treated apoE -/- (n=3), treated

apoE -/- (n=3) and WT mice (n=3) * P < 0.05

3.9.4 Evaluation of GC in the lymph node compartment of apoE -/- mice

From our observations, we noted LN hypertrophy in the LN

compartment in apoE -/- mice (Figure 14) This was associated with increased

GC reactions as the dominant antibody-producing pathway in these LNs

(Figure 16C, 16D, 17C & 17D) As we had established that the robust

extrafollicular responses to generate proliferating IgM+ plasmablasts in the

spleen of apoE -/- mice were associated with hypercholesterolemia (Figure 33

& 34), we next examined if hypertrophy and GC reactions in the LN

compartments were also associated with hypercholesterolemia

After ezetimibe treatment, the cellularity for treated apoE -/- mice decreased significantly in the axillary and brachial LN compared to non-

treated apoE -/- mice (Figure 35A) Similar decrease cellularity, but not

statistically significant, was also observed in the iliac LNs of ezetimibe treated

apoE -/- mice when compared to non-treated apoE -/- mice (Figure 35B)

Therefore, our disease regression model indicated LN hypertrophy was associated with hypercholesterolemia

Next, we investigated if ezetimibe treatment had an effect on the expansion of GC B cells in the LN compartment by flow cytometry Our results demonstrated that there were no differences in the relative percentage

of GC B cells between axillary and brachial, and iliac LNs in treated apoE and non-treated apoE -/- (Figure 36A & 36B, left) However, statistical

-/-significant decrease in relative number of GC B cells were noted in treated

apoE -/- mice, similar to WT mice when compared to non-treated apoE -/- mice

(Figure 36A, right) Similar observations of decreased relative number of GC

B cells were also observed in the iliac LNs, albeit non-statistically significant

(Figure 36B, right)

Lastly, we examined the effect of ezetimibe on the GC reactions in the

LNs of apoE -/- mice by immunofluorescence staining Our data showed that

GC reactions could still be observed in the LNs of treated apoE -/- mice but

they were noticeably smaller (Figure 37, inset) when compared to non-treated

apoE -/- mice (Figure 37)

Figure 35 Decreased lymph nodes cellularity after ezetimibe treatment in

apoE -/- mice

(A-B) Comparison cellularity count of (A) axillary and brachial LN, and (B)

iliac LNs between non-treated apoE -/- , treated apoE -/- and WT mice (mean ± SEM, n = 9-16) Data were pooled from four independent experiments * P < 0.05; *** P < 0.001

Figure 36 Decreased GC B cells in the lymph nodes of apoE -/- mice after ezetimibe treatment

(A-B) Comparison of B220+CD95+GL-7hi GC B cells in relative percentage

(left) and number (right) in the (A) axillary and brachial, and (B) iliac LNs by

flow cytometry between non-treated apoE -/- , treated apoE -/- and WT mice Data were pooled from four independent experiments * P < 0.05

Figure 37 Germinal centres were smaller in the lymph nodes of apoE -/- mice after ezetimibe treatment

Representative images of GC reaction sites in axillary and iliac LNs of non-treated apoE -/- (n=2), treated apoE -/- (n=2) and WT (n=2) mice Scale

bar represents 500µm

3.9.5 Evaluation of IgM + antibody secreting cells in the bone marrow of

apoE -/- mice after ezetimibe treatment

In our earlier studies, we also observed increased frequency of

oxLDL-specific IgM ASCs, but not total IgM ASCs, in the bone marrow of apoE

-/-mice (Figure 26E & 26F) Therefore, we investigated if the IgM+ ASCs

population was affected in the bone marrow compartment of apoE -/- mice after ezetimibe treatment

Our ELISpot analysis on the bone marrow compartment did not reveal any differences in the frequency of total IgM+ ASCs between non-treated

apoE -/- , treated apoE -/- and WT control mice (Figure 38A) In addition, when

we examined the IgM+ plasma cells in the bone marrow compartment by flow cytometry, we did not observed differences between treated and non-treated

apoE -/- mice (Figure 38C & 38D) However, the frequency of oxLDL-specific

IgM+ ASCs was decreased significantly in the treated apoE -/- mice (Figure 38B)

Therefore, our data showed that although no changes could be detected

in the total IgM+ ASCs/ plasma cells population in the bone marrow of treated

apoE -/- mice, we detected significant decrease of oxLDL-specific IgM ASCs suggesting that the formation of oxLDL-specific IgM ASCs in the bone marrow was associated with hypercholesterolemia

Figure 38 Decreased frequency of anti-oxLDL ASCs in the bone marrow

of ezetimibe treated apoE -/- mice

(A) Frequency of total IgM per million cells in non-treated apoE -/- (n=7),

treated apoE -/- (n=8) and WT mice (n=9) determined by ELISpot Data pooled

from 2 independent experiments (B) Frequency of total IgM per million cells

in non-treated apoE -/- (n=13), treated apoE -/- (n=14) and WT mice (n=14)

determined by ELISpot Data pooled from 3 independent experiments (C-D) Comparative flow cytometry analysis on (C) percentage and (D) number of

B220-CD138+EdU-IgM+ plasma cells in bone marrow of non-treated apoE (n=3), treated apoE -/- (n=3) and WT mice (n=3) * P < 0.05; ** P < 0.01

-/-3.9.6 Summary

Collectively, humoral responses in secondary lymphoid organs - both

GC reactions in the LN compartment and extrafollicular responses in the

spleen, were associated with hypercholesterolemia in apoE -/- mice

Furthermore, our data suggests robust splenic extrafollicular responses in

apoE -/- mice generated highly proliferative IgM+ plasmablasts to secrete

massive amount of antibodies in contributing to the elevated total IgM and

oxLDL-specific IgM autoantibodies Interestingly, the frequency of total IgM

and oxLDL-specific IgM ASCs did not decrease with ezetimibe treatment in

the spleen of apoE -/- mice However, ezetimibe treatment in apoE -/- mice led to

the decreased frequency of oxLDL-specific IgM ASCs in the bone marrow

Chapter 4 Discussions

summarized in Figure 39

In our study, we have identified the spleen as a major secondary lymphoid organ responsible for the generation of total IgM and oxLDL-specific IgM ASCs via the extrafollicular responses as the dominant antibody-producing pathway In contrast, the dominant antibody-producing pathway in

the LN compartment of apoE -/- mice was GC reactions and not extrafollicular responses Interestingly, although frequency of total IgM ASCs were increased, little or none oxLDL-specific IgM ASCs could be detected in the LNs

Humoral responses observed in apoE -/- mice was associated with hypercholesterolemia Extending our findings, the robust extrafollicular responses in the spleen was, at least in part supported by an increased amount

of 7α, 25-OHC oxysterol, the natural ligand for EBI2 chemokine receptor critically important for extrafollicular responses Our preliminary findings also suggest that splenic B1a, which are implicated in the production of oxLDL-specific antibodies, differentiated into IgM+ plasmablasts in the spleen Our findings also demonstrated the migration of IgM+ ASCs from the spleen to the bone marrow for long-term maintenance These IgM+ ASCs in the bone

marrow could also be another source of contribution to total and

oxLDL-specific IgM autoantibodies in circulation of apoE -/- mice

Figure 39 Schematic diagram of humoral responses in sustaining the

elevated IgM and oxLDL-specific IgM autoantibodies responses in apoE

4.2 Functional importance of IgM antibodies

Many studies have reported the observation of increased level of modified lipid-specific autoantibodies in CAD patients and in experimental models of atherosclerosis (Palinski et al., 1996; Palinski et al., 1989; Palinski

et al., 1995; Su et al., 2006; Tsimikas et al., 2007b) In our time-course analysis, we demonstrated that oxLDL-specific IgM autoantibodies in circulation increased during progression of atherosclerosis Furthermore, we demonstrated that the increase in total and oxLDL-specific antibodies were associated with hypercholesterolemia in our disease regression model These sets of observations were supported by published studies that reported that the titers of modified lipid-specific autoantibodies positively correlated with

increasing severity of the disease in ldlr -/- mice (Palinski et al., 1995)

Therefore, the important question to ask is what role does elevated total IgM and oxLDL-specific IgM autoantibodies serve? Studies in CAD patients suggest that modified lipid specific IgM autoantibodies were inversely correlated to cardiovascular risks (Su et al., 2006; Tsimikas et al., 2007b) Several animal studies have demonstrated the deposition of IgM antibodies in the plaque (Kyaw et al., 2012; Lewis et al., 2009) In the absence of serum

IgM antibodies, the lesion size area was higher in the aortic root of ldlr -/- mice (Lewis et al., 2009) Therefore, the prevailing view is that IgM antibodies are associated with protective functions in atherosclerosis

But what may be the underlying mechanisms confer by B cells? Studies in rabbits suggested that the formation of apoB-immune complexes could mediate rapid clearance of oxLDL from circulation, with the liver as a

major site for degradation (Wiklund et al., 1987) However, it was also reported that there was no apparent difference in clearance observed in

immune-deficient apoE -/- compared to immune-competent apoE -/- mice (Reardon et al., 2004) Therefore, whether the formation of immune complexes with oxLDL is involved in oxLDL clearance remained unresolved

Oxidized phospholipids are present on apoptotic cells and it was demonstrated that E06 oxLDL-specific antibodies could bind to these cells (Chang et al., 2004) Therefore, another functional possibility of IgM antibodies is the clearance of apoptotic cells present in the atherosclerotic plaque This is mediated through the activation of classical complement pathway initiated via recruitment of C1q molecules in IgM-dependent manner (Quartier et al., 2005) To support that the classical complement pathway

plays an important role, it was demonstrated that in C1q-deficient ldlr -/- mice had an increase accumulation of apoptotic cells in the plaque (Bhatia et al., 2007; Lewis et al., 2009)

In these aspects, the understanding of the generation and maintenance

of modified lipid-specific IgM autoantibodies in the progression of atherosclerosis is important and unexplored

4.3 Splenic humoral responses

B cells were found to have a protective role against atherosclerosis Atherosclerotic lesion size was increased in the proximal and distal aorta in

irradiated ldlr -/- mice after reconstitution with µMT (B cell-deficient) bone marrow (Major et al., 2002) The protective function of B cells was also

elegantly demonstrated by Caliguri et al, where splenectomized apoE -/- mice had increased lesion size while adoptive transfer of splenic B cells, but not T

cells, from aged apoE -/- mice to splenectomized recipients resulted in reduced lesion size (Caligiuri et al., 2002) Extending further, this protective immunity

was not conferred by splenic B cells of young apoE -/- mice (Caligiuri et al.,

2002) Subsequent adoptive transfer studies using splenectomized apoE -/- mice

as recipients, demonstrated that donor B1a cells, but not sIgM -/- B1a cells (does not secrete IgM antibodies), conferred protection against the disease (Kyaw et al., 2011) Taken together, these reports suggest an unknown protective mechanism, possibly in generating plasma cells, that took place

with aging and disease progression in apoE -/- mice Accordingly, this could

explain why a protective mechanism does exist in old apoE -/- mice but it takes place too late in the disease to have an overall protective effect

In our studies, we found that robust extrafollicular responses, rather than GC reactions, were the dominant antibody-producing pathway elicited in the spleen to account for the elevated IgM antibodies response in circulation

We did not detect any expansion of GC B cells nor defects in their isotypic

class-switching in the spleen of apoE -/- mice On the contrary, the extent of extrafollicular responses, identified by the colocalization of CD11chi DCs with

CD138+ plasmablasts, was increased significantly These plasmablasts were also IgM+ and were in proliferation phase as evident in our EdU-pulsed chase experiment Therefore, it is possible that extrafollicular responses in the spleen could be the mechanism by which B cells differentiate into plasma cells to confer the protective effect against atherosclerosis

4.4 Molecular cues in extrafollicular responses

Our findings beget the question why activated B cells differentiate into antibody producing cells via extrafollicular response pathway in atherosclerosis Using hen egg lysozyme (HEL) antigen of different affinities

to the HEL-specific SWHEL B cells, it was demonstrated that low antigen-BCR interaction led to the formation of GCs while high antigen-BCR interaction favored the formation of extrafollicular responses (Paus et al., 2006) Complementary to their findings, another group generated transgenic quasi-monoclonal mice with different BCR affinity to hapten (4-hydroxy-3-nitrophenyl)acetyl (NP) and obtained similar results too (O'Connor et al., 2006) Therefore, the B cell antibody responses are dependent on BCR affinity for the antigen and thus, it is possible that the nature of lipid antigens interaction with BCR led to the elicitation of extrafollicular responses in the

spleen of apoE -/- mice

The decision for B cells to undergo GC or extrafollicular responses is also dictated by the migration of activated B cells to different sites of the B cell follicles, which is mediated by chemokine receptor profiles Critically, GPR183/ EBI2 expression on activated B cells delineated their migration to either extrafollicular response or GC reaction pathways (Goodnow et al.,

2010) It was demonstrated that increased EBI2 expression directed activated

B cells to the bridging channel to differentiate into plasma cells in extrafollicular response pathway while activated B cells that lost EBI2 expression migrated into the B cell follicle to participate in the GC reactions (Gatto et al., 2009) The migration of activated B cells to the bridging channel

of the follicle suggested the presence of ligand for EBI2, which was recently identified to be oxysterol, 7α, 25-dihydroxycholesterol (7α, 25-OHC) (Hannedouche et al., 2011; Liu et al., 2011) Plasma cell responses in mice deficient in key enzymes for 7α, 25-OHC production and maintenance - CH25H, CYP7B1 or HSD3B7 were diminished in immunization studies when HEL-specific Hy10 B cells were adoptively transferred (Yi et al., 2012) In another independent study, the generation of IgG1+ plasma cells in response to

immunization with SRBCs was decreased on day 5 in ch25h -/- mice (Hannedouche et al., 2011) In our study, we did not detect an increased ebi2 mRNA expression in the spleen of apoE -/- mice which may suggest that B

cells from apoE -/- mice may not have an intrinsic propensity to skew humoral responses to extrafollicular response pathway However, this need to be

confirmed by specific analysis of ebi2 mRNA expression on sorted B cells Our data, however, revealed increased mRNA expression of ch25h, an

important enzyme in the synthesis of the oxysterol and no changes in mRNA

expression for hsd3b7, the enzyme important for the degradation or

conversion of 7α, 25-OHC into 4-cholesten-7α, 25-ol-3-one These observations of higher efficiency of 7α, 25-OHC synthesis while being converted to 4-cholesten-7α, 25-ol-3-one at normal rate potentially translates into increased amount of 7α, 25-OHC Indeed, we showed increased amount

of 7α, 25-OHC in the spleen of apoE -/- mice The amount of 7α, 25-OHC is important in migration because the titration of spleen extract resulted in a decreased migration of EBI2-transfected M12 cell line in a transwell chamber cell migration assay (Yi et al., 2012) Because of the colocalization of CD138+plasmablasts with CD11chi DCs in the bridging channel of follicle, it is anticipated that these DCs are an important source of 7α, 25-OHC to guide activated B cell migration Indeed, it was speculated that 7α, 25-OHC production could exceed the HSD3B7-mediated degradation present in CD11chiMHCIIhi DCs under activated conditions (Yi et al., 2012) Therefore, the bioavailability of 7α, 25-OHC could be important for governing the migration of activated B cells to participate in the extrafollicular responses in

apoE -/- mice

4.5 Presence of oxLDL in plasma

Besides addressing the issue on why is the extrafollicular responses in

the spleen of apoE -/- mice so robust, the next question is whether oxLDL are present in the plasma to initiate systemic humoral responses? Previous studies

described temporal changes of the amount of oxLDL in the plasma of apoE

-/-mice on chow diet - the level oxLDL detected in the plasma increased to the highest by 20 weeks old (Kato et al., 2009) However, the level of plasma oxLDL was observed to have decreased when lesion area at the aorta root increased after that time point (Kato et al., 2009) We speculate that the decreased amount of plasma oxLDL detected beyond 20 weeks old of age in their studies was due to the masking of oxidized epitopes, in their case oxidized phosphatidylcholine, on oxLDL due to elevated total IgM and

oxLDL-specific IgM autoantibodies detected in the plasma of apoE -/- mice To support our hypothesis, another group reported parallel increase of Cu2+oxLDL-specific IgM autoantibodies and apoB-immune complex with age in

apoE -/- mice (Dimayuga et al., 2006)

Administration of ezetimibe to lower circulating cholesterol levels in

apoE -/- mice from 12 weeks of age in our disease regression model demonstrated that splenic extrafollicular responses were strongly associated with hypercholesterolemia Interestingly, despite the reduction of splenic extrafollicular responses, we detected similar frequency of total IgM and

oxLDL-specific IgM ASCs in the spleen of treated apoE -/- mice compared to

non-treated apoE -/- mice However, these IgM+ ASCs from treated mice were not proliferating and did not secrete as much IgM antibodies as that from the

non-treated apoE -/- mice in our in vitro culture supernatant Taken together,

these sets of observations suggest that the formation of these IgM+ ASCs, and

as well as extrafollicular responses, could possibly taken place before our analysis

But how were these IgM+ ASCs generated in the treated apoE -/- mice? One possibility is that since our disease regression model is a therapeutic intervention against atherosclerosis, the formation of these IgM+ ASCs could have taken place before ezetimibe treatment Another possibility is that when the disease regresses, oxLDL are transported to the spleen to initiate humoral responses Indeed, previous studies have supported this hypothesis Even though total apoB-oxidized phospholipids (oxPL) level decreased, an increased ratio of plasma oxPL to apoB100 (density of oxPL per apoB

molecule) was detected after dietary reversion to normal chow diet in cynomolgus monkeys (Tsimikas et al., 2007a) The exact mechanism of how oxidized epitopes is being transferred to circulation when disease regression is taking place remained to be investigated This observation was not limited to animal studies but also reported in human studies Although total apoB-oxPL decreased in acute coronary syndrome patients treated with atorvastatin, ratio

of plasma oxPL to apoB100 was also found to increase significantly (Tsimikas

et al., 2004) Similar findings were reported with healthy volunteers when they were placed on low fat diet (Silaste et al., 2004) Therefore, although the total amount of oxLDL in the plasma decreased, the increased density of oxidized epitopes on LDL molecule associated with disease regression is still capable

of initiating splenic extrafollicular responses which is in line with observations that in addition to high BCR affinity to the antigen, epitope density could direct activated B cells to participate in extrafollicular responses (Paus et al., 2006)

4.6 Existence of tolerance mechanism?

Interestingly, the trend of increasing oxLDL level in the plasma till 20

weeks old in apoE -/- mice before it decreased (Kato et al., 2009) also suggests

a tolerance threshold for oxLDL before humoral responses take place which is

in agreement with our time course analysis where we observed elevated IgM and oxLDL-specific IgM autoantibodies beyond 16 weeks of age The tolerance mechanism remains to be explored and investigation on how B cells overcome such tolerance to differentiate into ASCs Interestingly, T cells were found to be reactive to native LDL rather than oxLDL and that oxidative

modification of LDL reduced T cell recognition (Hermansson et al., 2010) This strongly suggests that central tolerance to native LDL is incomplete and also implied the existence of peripheral tolerance mechanism to prevent autoreactive T cells specific for native LDL to be activated (Hansson and Hermansson, 2011)

Autoreactive B cells maintained peripheral tolerance by undergoing a state of anergy These autoreactive B cells, instead of being clonally deleted, tunes their responsiveness by down-regulating surface IgM, but not IgD, BCRs and elevated their basal calcium level when exposed to endogenous antigen (Goodnow et al., 1988; Zikherman et al., 2012) Further gene-

expression microarray studies elucidated apoE gene was significantly

upregulated in anergic B cells (Glynne et al., 2000) It is attractive to speculate

that the robust extrafollicular responses observed in the spleen of apoE -/- mice may be due to the inability of B cells to fall into the state of anergy However,

it remains to be determined if APOE protein is crucial in maintaining the anergy state of B cells

4.7 Role of B cell subpopulations

However, it remains unclear which B cell subpopulations differentiated into IgM+ plasma cells via extrafollicular response pathway Recently, Kyaw

et al elegantly demonstrated using adoptive transfer of B cell subpopulations into splenectomized apoE -/- mice recipients that the antibodies secreted by B1a cell population attenuated the disease progression (Kyaw et al., 2011) Whether PEC B1a cells spontaneously secrete natural IgM antibodies is controversial because little amount of secreted IgM could be detected and no

distinct spots could be seen in ELISpot assay, unlike their counterparts in the spleen (Choi et al., 2012; Holodick et al., 2010; Kawahara et al., 2003) The

work of Kawahara et al demonstrated that PEC B1a cells differentiation into

CD138+ ASCs does not take place in the PEC but into the spleen where they migrated (Kawahara et al., 2003) Our data supports this finding as we did not observed an increase in relative percentage and number of B1a IgM+plasmablasts in the PEC but noted an increased population of B1a cells differentiating into IgM+ plasmablasts in the spleen of apoE -/- mice However, when we examined for B1a differentiation into IgM+ plasma cells, there was a significant decreased population suggesting either that these newly differentiated IgM+ plasmablasts are either short-lived or they migrate out from the spleen into the bone marrow for long-term maintenance Colocalization of CD11chi DCs leads to increased survival of plasmablasts, by providing APRIL and BAFF, and their successful differentiation into plasma cells (Balazs et al., 2002; Garcia De Vinuesa et al., 1999) Therefore, although

we did not formally address the former possibility, increased apoptosis in B1a differentiated ASCs is an unlikely explanation We did, however, observed an increased frequency of oxLDL-specific oxLDL ASCs in the bone marrow in

apoE -/- mice, indicative of their subsequent migration into the bone marrow

Transferring of B2 cells into splenectomized apoE -/- mice, by contrast, did not attenuate atherosclerotic lesion size area and also failed to reconstitute total IgM level to that of sham-operated mice (Kyaw et al., 2011) The same group of investigators also demonstrated that B2 cells, but not B1a cells, were atherogenic population when atherosclerotic development was augmented

after transferring B2 cells into apoE -/- Rag2 -/- γc -/- (Kyaw et al., 2010)

Furthermore, their data suggest that B cells may be accountable for increased TNF-α mRNA expression in atherosclerotic lesion (Kyaw et al., 2010) Because anti-CD20 treatment after the establishment of atherosclerotic lesions

attenuates the disease development in apoE -/- mice without changing the levels

of plasma antibodies, the atherogenic effect of B2 cells seems to be mediated (Kyaw et al., 2010) This hypothesis was supported by other findings reporting that anti-CD20 treatment mediated protection against atherosclerosis leads to reduction in pathogenic T cell activation associated with decreased IFN-γ but increased IL-17 production (Ait-Oufella et al., 2010) Although these studies demonstrated that B2 cells are atherogenic, B2 cells comprised

cytokine-of marginal zone B cells and follicular B cells It remains to be demonstrated

if lipid-specific antibodies, particularly PC-specific IgM antibodies, produced

by marginal zone B cells could also have an overall protective effect in

atherosclerosis (van Leeuwen et al., 2009) Indeed, our findings in apoE

-/-mice showed that marginal zone B cell population expands, but it remained to

be demonstrated if this expansion leads to increase total IgM and

oxLDL-specific IgM autoantibodies in circulation (personal communication)

4.8 Absence of oxLDL-specific humoral response in lymph node

In our studies, we observed LN hypertrophy, which is indicative of immunological response In both skin draining and aorta draining LNs of

apoE -/- mice, we found GC center reactions and significantly increased GC B cells but little extrafollicular responses We confirmed that LN hypertrophy

and increased GC B cells in apoE -/- mice were associated with hypercholesterolemia in our disease regression model Unexpectedly, we

could not detect oxLDL-specific IgM+ ASCs in both skin and aorta draining LNs Therefore, the antigen specificity of these plasma cells generated in LNs remains unclear

So could DCs have not taken up oxLDL and migrate to the LN to initiate humoral responses? DCs are sentinel cells of the immune system that migrate to the secondary lymphoid organs after capturing foreign antigens to mount an immune response Indeed, oxLDL-specific T cells could be detected

in LNs suggesting DCs migration to the LNs to activate them (Zhou et al., 2001) However, as the disease progress, the emigration rate of DCs from the plaque to these organs decreased and resulted in their accumulation within the plaque region (Koltsova and Ley, 2011; Llodra et al., 2004) Extending these

findings was the recent human studies that reported in vitro findings that

oxLDL downregulated CCR7 mRNA and protein expression by DCs (Nickel

Given the absence of oxLDL-specific IgM+ ASCs in the LN

compartment of apoE -/- mice, it is attractive to speculate the antigenic specificity of humoral responses in the tertiary lymphoid structures at the adventitia layer of the aorta that arise in the late stage of the atherosclerotic

disease (Grabner et al., 2009; Moos et al., 2005) does not include that of oxLDL-specificity as well

4.9 Immune response or homeostatic response against oxidative modified lipids?

The high autoantibodies titer, in particular of IgM isotype, in apoE

-/-mice enabled the cloning of “E0” panel of oxLDL-specific IgM B cell hybridomas The monoclonal antibodies from the “E0” panel could also bind

to the atherosclerotic plaque, although giving rise to different patterns of staining with different clones, indicative of their relevance to the disease (Palinski et al., 1996) The prototypic E06 antibodies were found to be identical to that of classic PC-specific T15 idiotypic antibodies (Shaw et al., 2000) and could bind to the PC epitope on oxLDL but not native LDL (Horkko et al., 1999) Taken together, these sets of observations suggest PC-

specific clones and antibodies were increased in apoE -/- mice Unexpectedly,

our ELISA analysis on the plasma of apoE -/- mice is in disagreement with this assumption Although we could detect increased titer of oxLDL-specific IgM autoantibodies, we did not observe increased PC-specific IgM antibodies in

circulation of apoE -/- mice This suggests that PC-specific B cell clones in

apoE -/- mice did not expand in population and thus, it is logical that specific clones could also be isolated from WT mice Therefore, because of

PC-the lack of increased amount of PC-specific IgM antibodies in apoE -/- mice,

we reasoned that these antibodies were performing their respective functions, for example, binding to apoptotic cells for clearance, at a homeostatic level

On the other hand, we failed to detect increased oxLDL-specific IgM

autoantibodies in ldlr -/- mice at 24 weeks old Because atherosclerotic lesions were progressing at a slower rate, we examined these mice again at 45 weeks old when disease was advanced but we failed to observe the increased titer of oxLDL-specific IgM autoantibodies Surprisingly, we detected increased PC-

specific IgM antibodies in 24 weeks old ldlr -/- mice and its association with hypercholesterolemia when we treated these mice with ezetimibe (data not

shown) The reason for the differential antigenic responses in ldlr -/- from

apoE -/- mice is unclear

One of the drawbacks in antibodies studies in atherosclerosis field is the assumption that modified-lipid specific antibodies responses are the same

in apoE -/- and ldlr -/- mice For example, Cu2+ oxLDL instead of PC was used

in ELISA to measure oxLDL-specific antibodies in irradiated ldlr -/- mice reconstituted with µMT bone marrow to draw association with the disease outcome (Major et al., 2002) In light of our observations, we cannot expect

the antigenic B cell responses in apoE -/- and ldlr -/- mice are the same Therefore, our data suggest the possibility that decreased amount of oxidative modified lipid-specific antibodies detected could be interpreted as a decrease

in homeostatic level instead of decrease amount of antibodies that arise in response to the disease progression

4.10 IgM + plasma cells in bone marrow

Long-lived plasma cells are maintained in the bone marrow compartment and also serve as an important source of serum antibodies (Manz

et al., 1997; Radbruch et al., 2006; Shapiro-Shelef and Calame, 2005) In addition, these long-lived plasma cells are derived from GCs because the antibodies produced when examined, were of high affinity (Smith et al., 1997) Plasma cells generated via the extrafollicular pathway, however, do not migrate to the bone marrow and undergo apoptosis in the spleen after a few days (Smith et al., 1996) Therefore, it is generally accepted that long-lived plasma cells are the products of GCs but not of extrafollicular responses (Calame, 2001; Vinuesa et al., 2009) Our data revealing an increased frequency of oxLDL-specific IgM ASCs in the bone marrow associated with

splenic extrafollicular responses in apoE -/- mice, suggest, however, the possibility that plasmablasts generated from extrafollicular responses may be capable to migrate into the bone marrow for long-term maintenance similar to that of their IgG counterparts Indeed, we confirmed that this IgM+ plasma cells population was decreased when we disrupted humoral responses in our

CD11c-GFP-DTR chimeric ldlr -/- mice and that they were long-lived when we

maintained apoE -/- mice on BrdU in drinking water for a month Recent studies also reported the presence of IgM+ plasma cells and their importance

in the bone marrow compartment For example, it was found that IgM production in bone marrow provided long-term protection against fatal

infection challenge with intracellular bacterium, Ehrlichia muris (Racine et al.,

2011) Immunization of bacterial polysaccharide antigen, α1 à 3-dextran (DEX), resulted in the formation of cyclophosphamide and anti-CD20

treatment resistant IgM+ plasma cells in bone marrow (Foote et al., 2012) These IgM+ plasma cells were also found to be long-lived in the bone marrow after immunization with NP-CGG (Bortnick et al., 2012)

4.11 In situ generation of oxLDL-specific IgM plasma cells in bone

marrow?

The bone marrow compartment is often recognized for its role in haematopoesis as a primary lymphoid organ and its capability to initiate an immune response is often neglected The important question to ask next is the

possibility of in situ generation of IgM+ plasma cells, particularly specific IgM+ ASCs, in the bone marrow Circulating mature B cells, strategically localized to perisinusoidal niches in the bone marrow, were reported to be activated T-independently and differentiate into IgM+ plasma

oxLDL-cells in situ in response to S typhimurium as early as 1 day after microbial challenge (Cariappa et al., 2005) Furthermore, the number of S typhimurium-

specific IgM+ ASCs was dramatically increased in the bone marrow after

splenectomy, highlighting that bone marrow, could serve as a significant source of humoral response in absence of the spleen (Cariappa et al., 2005)

Indeed, our splenectomy data on apoE -/- mice also suggested the bone marrow could be a site of generation of IgM+ ASCs and oxLDL-specific IgM+ ASCs

in particular So could the oxLDL-specific IgM+ ASCs be generated in situ, rather than trafficking from the spleen into the bone marrow of apoE -/- mice?

Our data in our disease regression model showed that ezetimibe-treated apoE

-/-mice did not show increased frequency of oxLDL-specific IgM+ ASCs in the bone marrow while it did increased in the spleen argues against the hypothesis

that these cells were generated in the bone marrow Therefore, if humoral response against oxLDL does take place in the bone marrow of ezetimibe

treated apoE -/- mice, frequency of oxLDL-specific IgM+ ASCs should have also increased compared to WT controls because of survival factors to maintain plasma cells in the bone marrow

4.12 Trafficking to bone marrow

The increased frequency of oxLDL-specific IgM+ ASCs and increased population of EdU+IgM+ plasma cells observed in bone marrow in our EdU-pulsed chase experiment suggest the trafficking of these ASCs into the bone marrow compartment after proliferation in the splenic extrafollicular responses After immunization, the formation of antigen specific ASCs in the bone marrow peaks much later than the secondary lymphoid organs (Benner et al., 1981) Differences in the kinetics on the formation of antigen specific ASCs in secondary lymphoid organs and bone marrow after immunization implied the migration of these ASCs into the bone marrow compartment Such study of the kinetics of ASC in secondary lymphoid organs and bone marrow

is unfortunately challenging in a chronic inflammatory setting such as atherosclerosis

Therefore, to investigate the migration of IgM+ plasmablasts to the bone marrow, we depleted CD11chi DCs that are essential for the survival of differentiating plasmablasts in extrafollicular responses (Garcia De Vinuesa et

al., 1999) in chimeric ldlr -/- mice reconstituted with CD11c-GFP-DTR bone

marrow cells Construction of chimeric mice on ldlr -/- is essential on two

folds: 1) the generation of chimeric mice on apoE -/- background would revert

the hypercholesterolemia condition of these mice and 2) the generation of chimeric mice allowed for the repeated DT administration, which is not possible in CD11c-GFP-DTR mice as it results in animal death in 6-7 days after a single DT administration (Zammit et al., 2005) CD11c expression is not restricted to DCs Splenic marginal zone and metallophilic macrophages, and plasmablasts can express CD11c and thus can also be depleted with DT administration (Hebel et al., 2006; Probst et al., 2005) Therefore, splenic plasmablasts contribution to the bone marrow population may have been affected by 1) depletion of CD11chi DCs in providence of survival factors to proliferating IgM+ plasmablasts and 2) depletion of CD11c-expressing plasmablasts Our data had demonstrated that there was a decrease of splenic extrafollicular responses, concomitant with the observation that there was a decreased frequency of IgM+ ASCs and new migrant BrdU+IgM+ plasma cells

in the bone marrow when we administered DT to the chimeric mice This suggests that the formation or accumulation of IgM+ ASCs in the bone

marrow of apoE -/- mice may be a consequence of the humoral responses in the spleen in response to hypercholesterolemia

Another possibility to investigate splenic IgM+ plasmablasts contribution to the bone marrow compartment is to block their migration using FTY720, an immunosuppressive drug acting as the agonist for sphinogosine-1-phosphate receptor-1 (S1P1) receptor Egress of lymphocytes from secondary lymphoid organs is dependent on the expression of S1P1 into the blood, in response to S1P gradient, which is abundant in circulation but low in secondary lymphoid organs (Matloubian et al., 2004; Schwab et al., 2005) Thus, when immunized mice were treated with FTY720, the number of IgG

ASCs in the blood and bone marrow were reduced while localization within the secondary lymphoid organs was normal (Kabashima et al., 2006) However, it remained to be demonstrated if IgM plasmablasts also had similar chemokine receptor profile as IgG plasmablasts from secondary lymphoid organs into the bone marrow

4.13 Selection into bone marrow compartment

The selection of plasma cells into the long-lived compartment in the bone marrow is not entirely clear This selection of plasma cells requires changes in chemokine receptor profiles that enable their dislocation and ability

to egress out of the secondary lymphoid organs and their localization into the bone marrow One crucial requirement is the expression of CXCR4 on plasma cells for migration to colocalize with CXCL12-producing bone marrow stromal cells (Tokoyoda et al., 2004) However, the red pulp region of the spleen is also enriched for CXCL12 (Hargreaves et al., 2001) CXCR4-deficient plasma cells failed to locate themselves in the spleen and bone marrow, but remained in circulation (Hargreaves et al., 2001) Therefore, it is difficult to predict whether CXCR4 expression on plasma cells determine their retention in the spleen or homing to the bone marrow (Yoshida et al., 2010) Increased S1P1 expression and lower CXCR4 mRNA expression on IgG plasma cells have been shown to influence the migration of plasmablasts out

of the splenic compartment (Kabashima et al., 2006) Thus, the observation that the frequency of oxLDL-specific IgM ASC did not increase in the bone

marrow of ezetimibe treated apoE -/- mice raises the possibility that the splenic oxLDL-specific IgM ASCs did not acquire the changes in chemokine

receptors expression to allow their successful translocation to the bone marrow If so, under what conditions – either intrinsic factors of IgM plasmablasts or external environment factors influence, may confer their migratory capabilities?

4.15 Implication of study

Because of the advances in the understanding of the complexity of atherosclerosis, in addition to treatment of CAD patients with cholesterol lowering drugs such as statins and ezetimibe to impede the disease (2002; Phan et al., 2012), alternative strategies to module immune responses could lead to favorable outcomes

One such possibility is the IVIg treatment with total IgG antibodies isolated from a pool of healthy donors Although exact mechanisms of IVIg treatment are still being explored, studies have shown that the treatment was

able to reduce atherosclerosis and that atheroprotection in apoE -/- mice was due to the Fc region of the antibodies (Yuan et al., 2003) However, owing to the short half-lives of the infused immunoglobulin, sustained protective antibodies could be achieved through vaccination Indeed, multiple studies

involving immunization of either apoE -/- mice or ldlr -/- mice with lipid

antigens such as MDA-LDL and PC, or S pneumonia were associated with

decreased atherosclerosis (Binder et al., 2003; Caligiuri et al., 2007; George et al., 1998; Zhou et al., 2001) As such, our studies, although in a non-immunized settings, showed that IgM antibodies, in particular oxLDL-specific IgM autoantibodies, could be sustained in circulation due to the formation of long-lived plasma cell compartment in the bone marrow Therefore, it is

anticipated that vaccination in human with oxLDL will lead to sustained IgM antibodies responses against atherosclerosis

Although B cells in atherosclerosis were found to be protective in atherosclerosis, subsequent studies found that the pathogenic B cells, in particular B2 cells subset, exist (Ait-Oufella et al., 2010; Caligiuri et al., 2002; Kyaw et al., 2010; Major et al., 2002) B cell depletion using anti-CD20 antibodies has been effective in various autoimmune disorders and thus could also be useful in the treatment of atherosclerosis However, anti-CD20 treatment reduced the number of ASCs in the spleen but not those in the bone marrow due to higher CD20 expression in splenic ASCs (DiLillo et al., 2008)

In our work, we demonstrated robust extrafollicular responses in the spleen of

apoE -/- mice in generating IgM+ plasmblasts This is associated with the increase frequency of total IgM and oxLDL-specific IgM ASCs observed Therefore, we argue that with anti-CD20 treatment, the generation of these IgM+ ASCs to produce massive amount of total IgM and oxLDL-specific IgM autoantibodies to confer protection against the disease will be adversely affected

Therefore, a more selective strategy to target B2 cells is needed Given that mature B2 cell maturation and survival is dependent on BAFFR signaling,

the generation of baff-r -/- apoE -/- mice and irradiated ldlr -/- reconstituted with

bone marrow cells from baff-r -/- mice demonstrated that atherosclerotic lesions were reduced in these mice (Kyaw et al., 2012; Sage et al., 2012) A recent

study showed that treatment of apoE -/- mice with anti-BAFFR monoclonal antibodies also achieved reduced atherosclerotic lesions (Kyaw et al., 2013)