Bone Marrow-Specific Knock-In of a Non-Activatable Ikkα Kinase Mutant Influences Haematopoiesis but Not Atherosclerosis in Apoe-Deficient Mice

Background The Ikkα kinase, a subunit of the NF-κB-activating IKK complex, has emerged as an important regulator of inflammatory gene expression. However, the role of Ikkα-mediated phosphorylation in haematopoiesis and atherogenesis remains unexplored. In this study, we investigated the effect of a bone marrow (BM)-specific activation-resistant Ikkα mutant knock-in on haematopoiesis and atherosclerosis in mice. Methods and Results Apolipoprotein E (Apoe)-deficient mice were transplanted with BM carrying an activation-resistant Ikkα gene (IkkαAA/AAApoe−/−) or with Ikkα+/+Apoe−/− BM as control and were fed a high-cholesterol diet for 8 or 13 weeks. Interestingly, haematopoietic profiling by flow cytometry revealed a significant decrease in B-cells, regulatory T-cells and effector memory T-cells in IkkαAA/AAApoe−/− BM-chimeras, whereas the naive T-cell population was increased. Surprisingly, no differences were observed in the size, stage or cellular composition of atherosclerotic lesions in the aorta and aortic root of IkkαAA/AAApoe−/− vs Ikkα+/+Apoe−/− BM-transplanted mice, as shown by histological and immunofluorescent stainings. Necrotic core sizes, apoptosis, and intracellular lipid deposits in aortic root lesions were unaltered. In vitro, BM-derived macrophages from IkkαAA/AAApoe−/− vs Ikkα+/+Apoe−/− mice did not show significant differences in the uptake of oxidized low-density lipoproteins (oxLDL), and, with the exception of Il-12, the secretion of inflammatory proteins in conditions of Tnf-α or oxLDL stimulation was not significantly altered. Furthermore, serum levels of inflammatory proteins as measured with a cytokine bead array were comparable. Conclusion Our data reveal an important and previously unrecognized role of haematopoietic Ikkα kinase activation in the homeostasis of B-cells and regulatory T-cells. However, transplantation of IkkαAA mutant BM did not affect atherosclerosis in Apoe−/− mice. This suggests that the diverse functions of Ikkα in haematopoietic cells may counterbalance each other or may not be strong enough to influence atherogenesis, and reveals that targeting haematopoietic Ikkα kinase activity alone does not represent a therapeutic approach.


Introduction
Cardiovascular diseases are the main cause of morbidity and mortality in western societies, with atherosclerosis being the underlying pathology triggering most of the cardio-and cerebrovascular incidents. Atherosclerosis is a chronic inflammatory disease of the vessel wall characterized by the activation of endothelial cells, the subendothelial accumulation of oxidized lowdensity lipoproteins (oxLDL) and the infiltration of inflammatory cells such as neutrophils, monocytes, dendritic cells (DCs) and lymphocytes [1,2].
A key regulator of inflammation and atherogenesis is the transcription factor nuclear factor kB (NF-kB) [3]. The NF-kB family has 5 members: p65 (RelA), c-Rel, RelB, NF-kB1 (p105, processed to p50) and NF-kB2 (p100, processed to p52) [4]. Under resting conditions, canonical NF-kB dimers (mostly p50/p65) are predominantly found in the cytoplasm bound to the inhibitory of kB (IkB)-a protein. Inflammatory signals such as TNF-a and oxLDL activate the IkB kinase (IKK) complex, which consists of two catalytically active kinases (IKKa/IKK1 and IKKb/IKK2) and one regulatory component (IKKc/NEMO). The phosphorylation of IkB-a by IKKb causes its ubiquitination and proteasomal degradation. This releases the NF-kB dimer, allowing a steady-state localization in the nucleus and the expression of many pro-inflammatory proteins [5]. These pro-inflammatory functions of canonical NF-kB activation have been linked to atherogenesis in vivo, showing reduced lesion formation in hyperlipidaemic Apolipoprotein E (Apoe)-deficient mice treated with the NF-kB inhibitor DHMEQ, which prevents the TNF-ainduced nuclear translocation of p65 [6]. Similarly, strongly reduced atherosclerotic plaque formation was observed in Apoe 2/2 mice with an endothelial cell-restricted inhibition of NF-kB activation through endothelial Ikkc-deficiency or through an endothelial cell-specific expression of a dominant-negative IkB-a transgene [7]. On the other hand, NF-kB activity in leukocytes also has an important role in the resolution of inflammation through the transcription of anti-inflammatory cytokines such as interleukin (IL)-10 [8] and the suppression of pro-inflammatory IL1-b secretion [9]. Furthermore, an anti-inflammatory role was described for IKKb by suppression of the classically activated (or M1) macrophage phenotype [10]. These anti-inflammatory properties of IKK/NF-kB signalling could explain the initially unexpected increase in atherosclerosis in hyperlipidaemic Ldl receptor (Ldlr)-deficient mice with a myeloid-specific deletion of Ikkb, which was linked to a significant reduction of the anti-inflammatory cytokine IL-10 in Ikkb 2/2 macrophages [11]. Thus, IKKb/ NF-kB signalling plays a complex role in inflammation and atherogenesis by driving both pro-and anti-inflammatory processes, and the outcome of NF-kB inhibition on atherosclerosis seems strongly dependent on the targeted cell type.
In contrast to IKKb and IKKc, the role of IKKa in atherogenesis has not been investigated. Although IKKa is not required for canonical IkB-a phosphorylation and subsequent NF-kB activation [12,13], the IKKa kinase exerts multiple NF-kBdependent and -independent functions that could potentially influence atherogenesis [4,14]. First, IKKa homodimers mediate alternative NF-kB activation through phosphorylation of NF-kB p100, triggering p100 processing and the release and nuclear localization of RelB-p52 dimers [4,15]. This pathway, induced by several members of the TNF-superfamily as BAFF and CD40 ligand, plays a central role in B-cell maturation and lymphoid organ formation and may thus implicate IKKa in atherogenesis through the recently appreciated role of B-cells in this pathology [16]. Secondly, nuclear IKKa modulates gene expression through phosphorylation of histone H3, mediating the expression of a subset of canonical NF-kB-dependent genes in TNF-a-stimulated mouse embryonic fibroblasts [17,18]. On the other hand, IKKa has also been associated with repression or termination of gene transcription. For example, an anti-inflammatory role has been described for IKKa kinase in macrophages by inducing the phosphorylation, promoter removal and degradation of the canonical NF-kB isoforms p65 and c-Rel [19]. Furthermore, diverse pro-inflammatory stimuli trigger the IKKa-mediated phosphorylation of the transcriptional repressor PIAS1, which then negatively regulates the expression of a predominantly proinflammatory subset of p65-and STAT1-dependent genes [20]. In addition, IKKa-mediated phosphorylation of TAX1BP1 triggers the assembly of the A20 ubiquitin-editing complex, which is an important negative regulator of canonical NF-kB activation [21]. In conclusion, the IKKa kinase can positively or negatively regulate cell signalling and pro-and anti-inflammatory gene expression by phosphorylating diverse substrates, and could thus play a complex role in atherogenesis as well.
As haematopoietic cells are crucial players in the inflammatory reactions driving atherosclerosis, this study investigated the role of haematopoietic IKKa kinase activation in haematopoiesis and atherogenesis after transplantation of atherosclerosis-prone Apoe 2/2 mice with bone marrow carrying an activation-resistant Ikka AA/AA mutant [22].

Nomenclature
The letter format of all gene and protein notations in this manuscript is in accordance with internationally agreed gene/ protein nomenclature guidelines: all letters of human genes/ proteins are in uppercase, whereas for mouse genes/proteins, only the first letter is in uppercase. Gene names are in italics.
Mouse model, bone marrow (BM) transplantation and ethics statement C57BL/6 Ikka AA/AA mice, which are homozygous for an activation-resistant mutant of Ikka through replacement of the serines 176/180 in the kinase activation loop with alanines [22], were crossed with atherosclerosis-prone C57BL/6 Apoe 2/2 to generate Ikka AA/AA Apoe 2/2 mice. BM cells (3610 6 /mouse) from Ikka AA/AA Apoe 2/2 mice or from Ikka +/+ Apoe 2/2 littermate controls were flushed from femur and tibia marrow cavities and were subsequently administered to female C57BL/6 Apoe 2/2 recipient mice by lateral tail vein injection one day after a lethal dose of whole-body irradiation (266.5 Gy). After four weeks of recovery, the mice were put on a high-fat diet containing 21% fat and 0.15% cholesterol (Altromin) for either 8 or 13 weeks, as indicated. All animal experiments were approved by local authorities (Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen, Germany; approval number 8.87-50.10.35.08.073) and complied with the German animal protection law. All surgery was performed under ketamine/xylazin anesthesia, and all efforts were made to minimize suffering.

Determination of chimerism
The degree of chimerism in BM-transplanted mice was determined by quantifying the mutated Ikka AA allele relative to the wild-type Ikka allele in genomic DNA isolated from blood cells. Real-time quantitative PCR analysis was performed using the Maxima SYBR Green qPCR Mastermix (Fermentas) in a thermal cycler 7900HT (Applied Biosystems) using specific primer pairs (Sigma-Aldrich): 59-CCTCTCAGTGGCTCACCTTT and 59-CAATGTTCCCACAAAAGATGTACAGAGACT (for Ikka); 59-CCTCTCAGTGGCTCACCTTT and 59-CAATGTTCCCA-CAAACGCTGTACAGAGCGC (for Ikka AA ); 59-CAAC-GAGCGGTTCCGATG and 59-GCCACAGGATTCCATACC-CAA (for b-Actin). b-Actin was used as a reference gene. The method was validated by analyzing a standard curve using genomic DNA from Ikka AA/AA Apoe 2/2 and Ikka +/+ Apoe 2/2 blood cells, mixed at different ratios.

Lipid measurement in blood serum
Cholesterol and triglyceride levels in the blood serum were quantified using enzymatic assays (Cobas, Roche) according to the manufacturer's protocol. High-density (HDL), low-density (LDL) and very low-density (VLDL) cholesterol profiles were determined after HPLC-based fractionation of pooled serum samples using a Superose 10/300GL column followed by in-line cholesterol detection as described by Parini et al. [23].
Cytokine levels in the blood serum were measured by flow cytometry using the BD Cytometric Bead Array Mouse Inflammation Kit (for Tnf-a, Mcp1, Il6, Ifn-c, Il12 and Il10; BD-Pharmingen). In addition, Il10 levels were measured using a mouse Il10 Quantikine ELISA kit (R&D Systems).

Atherosclerotic lesion analysis using histology and immunofluorescence staining
For atherosclerotic lesion analysis, the aortic root and thoracoabdominal aorta were stained for lipid depositions with Oil-Red-O. In brief, the heart with aortic root was embedded in Tissue-Tek for cryo sectioning. Atherosclerotic lesions were quantified in 5 mm transverse sections and averages were calculated from 3-5 sections. The aorta was opened longitudinally, mounted on glass slides and en face-stained. Macrophages, smooth muscle cells (SMCs) and T-cells in the atherosclerotic lesions were visualized by immunofluorescent staining for Mac2 (Cedarlane), Sma (Dako) and Cd3 (AbD Serotec), respectively, followed by a FITC-or Cy3-conjugated secondary antibody staining (Jackson ImmunoResearch). Appropriate IgG antibodies were used as isotype controls. Nuclei were counterstained by 49,6-diamidino-2-phenylindol (DAPI). Neutrophil presence was examined by Naphthol-AS-Dchloroacetate Esterase (ASDCL) staining. Apoptotic nuclei were detected by terminal deoxynucleotidyl nick-end labelling (TU-NEL-kit, Roche). Intracellular lipid deposits in aortic root lesions were stained using Nile Red (N-3013, Sigma). All images were recorded with a Leica DMLB fluorescence microscope and CCD camera. The quantification of lesion size and composition was performed using Diskus analysis software (Hilgers), whereas the Nile Red stainings were analyzed with help of Image J software. All analyses were performed without prior knowledge of the genotype.
To explore potential qualitative effects on atherosclerosis, the aortic root lesions were classified according to phenotype, as previously described [11]. Three categories were distinguished: (1) early lesions, containing only foam cells, (2) intermediate-type lesions, presenting foam cells, some necrosis and a fibrotic cap, (3) advanced lesions, showing extended fibrosis and necrosis and infiltration of the plaque into the media.

Preparation and labelling of oxLDL
OxLDL was prepared by oxidation of LDL (Calbiochem) with a 50 nM copper sulphate solution for 4 hours at 37uC for mildly oxidized LDL and overnight for heavily oxidized LDL, followed by purification over a PD 10 column (GE Healthcare). For lipid uptake experiments heavily oxidized LDL was labelled with 1,10dioctadecyl-3,3,3030-tetramethylindocyanide percholorate (Dil). After incubation of 0,5 mg/ml oxLDL with 20 mL of Dil (stock 3 mg/ml in DMSO) overnight at 37uC, the Dil-labelled oxLDL was purified over a PD 10 column and stored at 4uC for a maximum of 2 weeks.
BM-derived macrophages, nuclear extracts and NF-kB p65 DNA-binding ELISA BM-derived macrophages were generated as previously described [24]. Briefly, BM from the femurs and tibiae was flushed with ice-cold PBS using a 27-G needle, resuspended in PBS by repeated vigorous pipetting and filtered using a 70 mM cell strainer (BD Biosciences). The filtered solution was centrifuged, the resulting pellet resuspended in culture medium and the cells plated on 15 cm untreated culture dishes (Greiner). BM-derived macrophages from cryopreserved BM-cells were generated as previously described [25]. Culturing and stimulation of BMderived macrophages were performed in RPMI 1640 (+L-Glutamin) containing 10 mM Hepes, 10% FCS, 15% L929-cellconditioned medium and 100 U/ml gentamycin. After 7 days of culturing, differentiated macrophages were used for stimulation experiments and transferred onto untreated 6-well dishes (Greiner). The cells were left for 24 hours to adhere und were then stimulated with 100 ng/ml lipopolysaccharide (LPS) (Sigma-Aldrich), 10 ng/ml mouse Tnf-a (Peprotech) or 50 mg/ml mildly oxidized LDL, as indicated.
Nuclear extracts were isolated as described [11]. Briefly, BMderived macrophages were washed once with PBS and scraped of the culture dish using PBS with 5 mM EDTA. The cells were centrifuged and the cell pellet was resuspended in 100 ml buffer A (10 mM Hepes pH 7.8, 1.5 mM MgCl 2 , 0.5 mM DTT and 16 Complete EDTA-free protease inhibitor cocktail (Roche)). After 2 min incubation on ice, 100 ml of buffer A supplemented with 1.28% NP40 was added to the cell suspension and incubated on ice for 10 min. The cells were vortexed for 10 sec and centrifuged for 10 min at 2000 rpm. The supernatant was removed, the pellet dissolved in 50 ml buffer B (20 mM Hepes pH 7.8, 420 mM NaCl; 1,2 mM MgCl 2 , 0.2 mM EDTA, 25% glycerol, 0.5 mM DTT and 16 Complete EDTA-free protease inhibitor cocktail (Roche)) and incubated for 30 min on ice. Every 5 min, the sample was vortexed thoroughly. Then the solution was centrifuged for 15 min at 4000 rpm and the supernatant, being the nuclear lysate, was snap-frozen at 280uC. Protein concentration was determined using the Quick Start Bradford Protein Assay (Biorad).
NF-kB p65 DNA-binding activity in equal amounts of nuclear extracts was quantified using an oligonucleotide-based ELISA (TransAM NF-kB p65 ELISA, Active Motif) according to the supplier's instructions. Absorbance values were corrected for background by incubation with lysis buffer only.

In vitro macrophage foam cell formation and cytokine secretion
To quantify the uptake of DiI-labelled heavily oxidized LDL by BM-derived macrophages, macrophages were plated on 24-well plates and incubated overnight at 37uC. The next day, nonadherent cells were rinsed off with PBS and medium containing 1 mg/ml or 10 mg/ml Dil-oxLDL was added. To analyze whether oxLDL uptake occurred in an actin-dependent way, controls were pre-incubated for 1 h with 10 mM cytochalasin D, followed by a stimulation with 10 mg/ml Dil-oxLDL and 10 mM cytochalasin D. The cells were stimulated for 3 or 24 hours as indicated, washed with PBS and stained with F4/80 (clone BM8, eBioscience). Flow cytometric analysis was performed using a FACSCanto II and the data were analyzed using FlowJo software (Treestar). Data were calculated by subtracting the cell autofluorescence (cells without diI-oxLDL incubation) from the fluorescence of the diI-oxLDL-treated samples and were expressed as geometric mean fluorescence intensity (gMFI).
To measure cytokine and chemokine secretion from BMderived macrophages, cells were plated in 6-well plates, left for 24 h to adhere und were then stimulated with 10 ng/ml mouse Tnf-a (Peprotech) or 50 mg/ml heavily oxidized LDL. As control unstimulated cells were included. After 24 h the medium was harvested from the cells, and cytokine and chemokine levels were measured by flow cytometry using the BD Cytometric Bead Array Mouse Inflammation Kit (BD-Pharmingen). In addition, Mcp1 levels were measured using a mouse Mcp1 ELISA (DY479, R&D Systems).

Statistical analysis
All statistical analyses were performed using GraphPad Prism (GraphPad Software Inc.). Data are represented as means 6 SEM and were analyzed by 2-tailed Student's t-test or 2-way ANOVA with Bonferroni post-test, as appropriate. P,0.05 was considered statistically significant.

Bone marrow-specific loss of Ikka kinase activation affects B-and T-lymphocyte populations
To study the role of IKKa kinase activation in haematopoiesis in the context of atherosclerosis, BM was isolated from Ikka +/ + Apoe 2/2 or Ikka AA/AA Apoe 2/2 mice, the latter carrying an nonactivatable mutant of Ikka through replacement of serines 176/ 180 in its activation loop with alanine residues [22]. After transplantation of this BM into lethally irradiated Apoe 2/2 mice and a recovery period of four weeks, the mice received a highcholesterol diet for 13 weeks to accelerate atherosclerosis. Quantitative real-time PCR for the Ikka AA/AA vs Ikka +/+ allele in white blood cells of the transplanted recipient mice indicated that on average 96.1% (61.2) of leukocytes carried the mutant Ikka AA/AA allele, confirming a successful engraftment.
Interestingly, although total leukocyte counts were unaffected, Ikka AA/AA Apoe 2/2 BM chimeras showed a significantly reduced Cd19 + B-cell population in peripheral blood ( Figure 1A, Table 1). A similar decrease in B-cell number was seen in the BM and lymph nodes of the Ikka AA/AA Apoe 2/2 BM chimeras, whereas no difference was observed in the spleen B-cell population ( Figure S1).
In addition, also Cd3 + Cd4 + Cd25 + Foxp3 + regulatory T-cells (T reg ), which are associated with atheroprotection [26], were found to be significantly reduced in the Cd3 + T-cell population of the peripheral blood, thymus and secondary lymphoid organs of the Ikka AA/AA Apoe 2/2 -transplanted mice ( Figure 1B). Also among Cd45 + leukocytes, T reg frequency was markedly decreased in blood and spleens of the Ikka AA/AA Apoe 2/2 BM chimeras ( Figure 1B). However, this was not observed in the lymph nodes ( Figure 1B), which showed a significant overall increase of Cd3 + Tcells in the leukocyte population ( Figure S1B).
No changes were observed in the relative proportions of Cd4 + and Cd8a + T-cell subsets in blood or lymphoid organs ( Figure 1A, Figure S1). However, spleen and lymph nodes of Ikka AA/AA Apoe 2/2transplanted mice displayed a significant increase in the Cd3 + Cd44 low Cd62L high naive T-cell population, whereas Cd3 + Cd44 high Cd62L low effector memory T-cells, which mediate effector functions in inflamed tissue [27], were significantly reduced ( Figure 1C, Figure S2). Also, the Cd3 + Cd44 high Cd62L high central memory T-cell population, associated with the successive production of effector T-cells [27], was significantly decreased among splenic T-lymphocytes and total leukocytes ( Figure S3).
A similar effect on B-and T-cell populations was observed in a non-atherosclerotic context. C57BL/6 mice transplanted with Ikka AA/AA BM displayed a significantly reduced B-cell population in lymph nodes compared to controls, whereas both Cd4 + and Cd8a + T-cell subsets were increased ( Figure S4A,B). Similarly as observed before, T reg lymphocytes were markedly decreased ( Figure S5). Altogether, these findings indicate an important role for Ikka kinase activity in haematopoiesis, with reduced B-cells and T reg lymphocytes upon haematopoietic expression of an Ikka AA/AA mutant.
Atherosclerotic lesions are unaltered in Ikka AA/AA Apoe 2/2 bone marrow chimeras After 13 weeks of high-cholesterol diet, Ikka AA/AA Apoe 2/2 and Ikka +/+ Apoe 2/2 BM chimeras were sacrificed and lipid levels in serum and the extent of atherosclerosis in the aorta and aortic root were analyzed. Whereas body weight and serum triglyceride values were similar, cholesterol levels were significantly increased in Ikka AA/AA Apoe 2/2 vs Ikka +/+ Apoe 2/2 BM-transplanted Apoe 2/2 mice (Figure 2A,B). This could be attributed to an increase in VLDL and LDL lipoprotein fractions, as shown by a cholesterol analysis after HPLC-based lipoprotein size separation of pooled serum samples ( Figure 2C). Despite these differences in cholesterol levels, atherosclerotic lesion sizes in the aorta and aortic root were comparable in Ikka AA/AA Apoe 2/2 and Ikka +/+ Apoe 2/2 BM chimeras ( Figure 2D,E). To explore potential qualitative effects on atherosclerosis, the aortic root lesions were phenotypically classified into early, intermediate-type and advanced lesions, as previously described [11]. Again, no differences were found between both groups ( Figure 2F). Also, the cellular composition of aortic root lesions was comparable, presenting an equal content of macrophages (Mac2 + ) and SMCs (Sma + ) as shown by immunofluorescent stainings ( Figure 3A,B). Similarly, no significant differences were observed in lesional Cd3 + T-lymphocyte content ( Figure 3C). No neutrophils could be detected in any of the aortic root lesions. Finally, necrotic core sizes were comparable ( Figure 4A), and no significant differences were found in the content of TUNEL + apoptotic cells and apoptotic macrophages in aortic root lesions of Ikka AA/AA Apoe 2/2 and Ikka +/+ Apoe 2/2 BM chimeras ( Figure 4B,C). To investigate potential effects on earlier stages of atherosclerosis, a similar study was performed after a high-cholesterol diet fed for only 8 weeks. Successful engraftment of Ikka AA/AA Apoe 2/2 BM into lethally irradiated Apoe 2/2 mice was again confirmed by quantitative real-time PCR, showing the mutant Ikka AA/AA allele to be present in 94.3% (62.7) of blood leukocytes. After this shorter period of high-fat diet, no differences were obtained in body weight or lipid levels in the blood serum, and also HPLC-based lipoprotein fractionation of pooled serum samples showed comparable HDL-, LDL-and VLDL-associated cholesterol peaks ( Figure 5A-C). Similarly as for the 13 week-high-fat diet, Ikka AA/ AA Apoe 2/2 and Ikka +/+ Apoe 2/2 BM chimeras presented comparable atherosclerotic lesion sizes in the aorta and aortic root ( Figure 5D,E). Lesions were reduced in size with 39% and 56% in aorta and aortic root, respectively, compared to the longer diet course, as can be expected. Also, classification of the plaques into early, intermediate or advanced stages revealed comparable lesion phenotypes in Ikka AA/AA Apoe 2/2 and Ikka +/+ Apoe 2/2 BM-transplanted Apoe 2/2 mice, although the lesions were less advanced compared to the 13 week-diet study ( Figure 5F).
In summary, these data indicated that the transplantation of Ikka AA/AA Apoe 2/2 -mutant BM does not affect the size, phenotype or cellular content of atherosclerotic lesions in atherosclerosisprone Apoe 2/2 mice.
Haematopoietic knock-in of the Ikka AA/AA mutant does not affect systemic inflammatory gene expression in ApoE-deficient mice The canonical NF-kB pathway has been shown to play an important role in controlling atheroprogression, at least partly by balancing anti-and pro-inflammatory processes in macrophages [11]. Canonical NF-kB activation is dependent on IKKb and IKKc, but does not require IKKa kinase activity for IKKmediated IkB-a phosphorylation and degradation with subsequent NF-kB activation [12,13]. On the other hand, IKKa has been shown to terminate LPS-induced NF-kB activity in macrophages by promoting the phosphorylation and degradation of the NF-kB isoform p65, thereby abrogating p65-mediated gene transcription [19]. Therefore, we aimed to examine the direct effect of the Ikka AA/AA knock-in mutation on canonical NF-kB activity in the context of atherosclerosis by studying the DNA binding capacity of p65 in nuclear extracts of Ikka AA/AA Apoe 2/2 vs Ikka +/+ Apoe 2/2 BM-derived macrophages in vitro. Unexpectedly, the Ikka AA/AA knock-in mutation did not enhance or prolong p65 activity in Apoedeficient macrophages upon stimulation with Tnf-a or oxidized LDL (oxLDL), i.e. mimicking inflammatory or atherogenic challenges, and even slowed down Tnf-a-induced p65 activation. No differences were observed in LPS-induced p65 activity in Ikka AA/AA Apoe 2/2 vs Ikka +/+ Apoe 2/2 BM-derived macrophages ( Figure 6A).
Furthermore, the concentrations of the inflammatory proteins Tnf-a, Il-6 and Mcp1 in supernatants of Tnf-a-or oxLDLstimulated Ikka AA/AA Apoe 2/2 and Ikka +/+ Apoe 2/2 BM-derived macrophages were not significantly different, although Ikka AA/AA knock-in significantly enhanced the basal secretion of Mcp1, in contrast to Tnf-a or Il-6 ( Figure 6B). Neither Tnf-a nor oxLDL were able to induce secretion of Il-10 or Il-12p70 in vitro, although Il-12p70 secretion was found to be significantly higher in oxLDLstimulated macrophages upon Ikka AA/AA knock-in ( Figure S6). In addition, quantification of Tnf-a, Il-6 and Mcp1 in serum of Ikka AA/AA Apoe 2/2 vs Ikka +/+ Apoe 2/2 BM chimeras after 8 weeks of high-fat diet did not reveal significant differences ( Figure 6C), whereas the cytokines Ifn-c, Il-12 and Il-10 remained below the detection limit in both groups.
Given the importance of lipid uptake by macrophages in atherosclerotic lesions, we examined a possible effect of Ikka AA/AA knock-in on macrophage foam cell formation in vitro. With 2 different incubation times and oxLDL doses, we did not observe a significant difference in oxLDL uptake by Ikka AA/AA Apoe 2/2 vs Ikka +/+ Apoe 2/2 BM-derived macrophages ( Figure 7A). Cytochalasin D severely reduced the oxLDL-associated fluorescence signal, indicating that oxLDL was actively taken up by the cells in an actin-dependent way and not merely binding the cell surface ( Figure 7A). Next, we quantified lipid deposits in aortic root lesions of Ikka AA/AA Apoe 2/2 and Ikka +/+ Apoe 2/2 BM chimeras after 13 weeks of high-cholesterol diet using Nile Red staining, but no difference were observed ( Figure 7B,D). Also, quantification of Nile Red-positive macrophages after co-staining for Mac2 together with Nile Red dye revealed comparable intracellular lipid deposits in lesional macrophages and an equal amount of lipid-laden macrophages ( Figure 7C,D).
In conclusion, these data show that knock-in of Ikka AA/AA in an Apoe 2/2 background does not enhance NF-kB activity or majorly influence the secretion of important inflammatory proteins from Tnf-a-or oxLDL-stimulated macrophages, nor does it significantly affect macrophage foam cell formation.

Discussion
Atherosclerosis is characterized by a chronic inflammation of the vessel wall and all leukocyte subsets, including B-and Tlymphocytes, monocytes and monocyte-derived macrophages, neutrophils and DCs, contribute in their own specific ways to the pathogenesis of this widespread disease [1,2]. Therefore, we investigated the effect of a BM-specific non-activatable Ikka AA knock-in on haematopoiesis in conditions of atherosclerosis and identified a significant reduction in the B-cell population in the blood and lymph nodes of hyperlipidaemic Ikka AA/AA Apoe 2/2 BM chimeras in comparison with Ikka +/+ Apoe 2/2 BM-transplanted Apoe 2/2 controls ( Figure 1A, Figure S1). Comparable results were obtained in a non-atherosclerotic context, with a significantly reduced B-cell population in lymph nodes of C57BL/6 mice transplanted with Ikka AA/AA vs Ikka +/+ BM ( Figure S4), and are consistent with earlier observations of a reduced mature B-cell population in secondary lymphoid organs of Ikka AA/AA knock-in mice and in Ikka 2/2 and Ikka AA/AA BM chimeras [15,28]. Similar effects were seen in Baff 2/2 mice [29] and revealed a crucial role for the non-canonical Baff-Baffr-Nik-Ikka pathway in B-cell maturation and survival. Our observation that the Cd19 + B-cell population is also significantly decreased in the BM of Ikka AA/ AA Apoe 2/2 -transplanted Apoe 2/2 mice ( Figure S1) suggests that the Ikka kinase activity is also important in BM B-cell development. This corresponds to recent findings of Balkhi and colleagues, who identified a reduction in the Cd19 + , B220 + and Cd19 + B220 + B-cell population in the BM of kinase-dead Ikka (Ikka KA/KA ) knock-in mice and Ikka KA/KA BM chimeras, and also revealed less B220 + Cd19 + B-cells in the BM of irradiated Rag 2/2 mice reconstituted with Ikka 2/2 fetal liver cells [30]. Thus, our data support this important role for Ikka kinase activity in early B-cell development in the BM, which was revealed to involve both canonical and non-canonical NF-kB pathways [30]. Secondly, we observed a consistent increase in the Cd62L high C-d44 low naive T-cell population in secondary lymphoid organs of Ikka AA/AA Apoe 2/2 BM chimeras, whereas Cd62L low Cd44 high effector memory T-cells were decreased ( Figure 1C). This corresponds with a previous observation of Mancino and colleagues, who revealed that Ikka kinase activity in DCs is crucial for antigen-specific priming of naive T-cells in an in vivo delayed-type hypersensitivity model [31]. Similar effects on the ratio of naive vs effector memory T-cells were seen in mice deficient for Nik, the kinase activating Ikka in the non-canonical NF-kB pathway [32]. Likewise, Nik aly/aly mice, which carry a natural Nik mutant (Nik aly ) unable to bind Ikka and associated with reduced NF-kB activation [33][34][35], presented with reduced T-cell effector cytokine expression, which was ascribed to a Nikdeficiency in thymic DCs rather than to an intrinsic T-cell defect [36]. Despite equal DC numbers in the thymus of Nik aly/aly and Nik +/+ mice, thymic Nik aly/aly DCs showed a decreased expression of typical activation markers [36]. Similarly, also splenic Nik aly/aly DCs expressed a considerably lower level of MhcII compared to Nik aly/+ DCs [37], suggesting an inability of Nik aly/aly DCs to deliver T-cell costimulatory signals and contribute to the development of effector T-cells [36]. In line with this, our Ikka AA/AA Apoe 2/2 BM chimeras displayed a significantly reduced MhcII expression on splenic pDCs ( Figure 1D). Very recently, an increased ratio of naive vs effector memory T-cells was also seen in Nik 2/2 BM chimeras, but was linked with a cell-intrinsic role of Nik in the generation or maintenance of effector memory T-cells [38]. Given these new findings together with the previous observations of Mancino et al. [31], the relative importance of cell-intrinsic vs DCmediated effects on the increased ratio of naive vs effector memory T-cells in our Ikka AA/AA BM chimeras remains to be investigated.
In addition to reduced effector memory T-cells, Ikka AA/ AA (Apoe 2/2 ) BM chimeras displayed a significantly reduced T reg population ( Figure 1B, Figure S5). T reg cells develop in the thymus and in peripheral sites from naive CD4 + T-cells. Initially, defective alternative NF-kB activation in the thymic stroma and a disorganized thymic structure in Nik aly/aly mice was associated with a defective establishment of self-tolerance and reduced T reg numbers [39]. Later, a study of single Nik 2/2 vs mixed Nik 2/2 / Nik +/+ BM chimeras indicated an additional cell-intrinsic role for Nik in the maintenance of peripheral T reg T-cells [38]. Also, at least part of the T reg defect in Nik aly/aly mice was attributed to the reduced capacity of Nik aly/aly DCs to trigger T reg expansion and survival in vitro [37], and both NIK and IKKa were shown to be required in human DCs to trigger the development of T reg cells from naive CD4 + T-cells in vitro [40]. Subsequently, pDC-triggered T reg development from naive T-cells was associated with the induction of indoleamine 2,3-dioxygenase in pDCs by NIKmediated non-canonical NF-kB activation [40][41][42][43]. With Ikka a crucial player in this alternative NF-kB pathway, our data now directly confirm for the first time a role for haematopoietic Ikka kinase activation in the generation of T reg cells in vivo. However, the relative importance of T reg -intrinsic vs DC-mediated mechanisms remain to be investigated, both in the periphery as thymus, given that peripheral DCs can migrate into the thymus to actively contribute to thymic T reg generation [44]. Of note, the reduced T reg population in Ikka AA/AA Apoe 2/2 BM chimeras could be associated with the increased serum levels of VLDL observed in these mice, as depletion of T reg T-cells was recently discovered to enhance VLDL levels through reduced VLDL clearance [45].
Despite the clear effects of a BM-specific Ikka AA/AA Apoe 2/2 knock-in on haematopoiesis and the enhanced VLDL levels, no differences were observed in the size, phenotype and cellular composition of atherosclerotic lesions in hyperlipidaemic Apoe 2/2 mice transplanted with Ikka AA/AA Apoe 2/2 or Ikka +/+ Apoe 2/2 BM (Figure 2-5). The influence of the reduced B-cell population in the Ikka AA/AA Apoe 2/2 BM chimeras on atherosclerosis is unclear. On the one hand, an atheroprogressive effect could be suggested based on the observation that a transplantation of Ldlr 2/2 mice with Bcell deficient (mMT) BM aggravated atherosclerosis [46] and also several other studies indicated an atheroprotective role for B-cells [1,16]. However, different B-cell subsets have diverse functions in atherogenesis and B2 B-cells, in contrast to B1 B-cells, rather exacerbate atherosclerosis [16]. In this context, a lack of B2 Bcells, but a preserved B1 B-cell population was detected in mice with a deficiency of Baff [29] or Baffr [47], which can signal to NF-kB through Ikka. This was associated with reduced atherosclerosis in Baffr 2/2 Apoe 2/2 or Baffr 2/2 BM-transplanted Ldlr 2/2 mice compared to controls [48,49]. Although B-cell subsets were not examined in our study, Senftleben et al. predominantly found a reduction in the mature IgM low IgD high B-cell population in Ikka AA/ AA mice and Ikka AA/AA BM chimeras [15], with IgD being a surface marker of follicular B2 B-cells [16]. Thus, the reduced B-cell population in our Ikka AA/AA Apoe 2/2 BM chimeras may by itself provide atheroprotective effects through a potential decrease in B2 B-cells. Regarding T-lymphocytes, effector memory T-cells are present in atherosclerotic lesions [1] and were shown to positively correlate with the extent of atherosclerosis in atherogenic mice, whereas naive T-cells were inversely correlated with plaque size [50]. Furthermore, T reg T-cells have been shown to mediate atheroprotective functions [1,26]. Therefore, pro-atherogenic effects of a reduction in the T reg population in Ikka AA/AA Apoe 2/2 BM chimeras may be compensated by atheroprotective effects of the observed relative increase of naive versus effector memory Tcells in these mice. Comparable results were seen upon deficiency of Cd40-Traf2/3/5 signalling in Apoe 2/2 mice, which did not affect atherosclerosis despite an increase in both atheroprogressive effector memory T-cells and atheroprotective T reg cells in blood and secondary lymphoid organs [51]. Altogether, this suggests that simultaneous changes in lymphocyte subsets may completely balance individual atheroprogressive and -protective effects, without any net effect on atherosclerosis, similarly as observed in our study.
As macrophages play an important role in the uptake of modified lipids in atherosclerotic lesions, we examined the effect of an Ikka AA/AA knock-in on macrophage intracellular lipid accumulation. However, no significant differences were observed in vitro or in atherosclerotic lesions in vivo (Figure 7), which is similar as previously observed for Ldlr 2/2 mice with a myeloid-specific deletion of Ikkb [11]. Furthermore, the Ikka AA/AA knock-in mutation was previously shown to reduce macrophage apoptosis upon bacterial infection, which was associated with prolonged LPS-triggered NF-kB p65 activation in BM-derived macrophages [19]. As apoptosis is an important process in atherogenesis, being atheroprotective in early stages of disease but associated with plaque necrosis and atheroprogression in later phases [52], we investigated lesional apoptosis in Ikka AA/AA Apoe 2/2 and Ikka +/+ Apoe 2/2 BM chimeras. However, no significant differences were observed in cellular or macrophage apoptosis, and also necrotic core sizes were comparable ( Figure 4). Furthermore, we did not observe a differential activity of NF-kB p65 upon LPS stimulation of Ikka AA/AA Apoe 2/2 vs Ikka +/+ Apoe 2/2 BM-derived macrophages in vitro, and even detected a reduced response in p65 activation upon atherogenic (i.e. Tnf-a, oxLDL) exposure ( Figure 6A). Although these differential observations between our study and the one from Lawrence and colleagues [19] are unexpected at first sight, a major difference between the two studies is the use of Apoe 2/2 macrophages in our report. ApoE has been recognized as an important immunomodulator, which in macrophages promotes the anti-inflammatory M2 phenotype [53]. It is well-known that regulatory effects on NF-kB signalling behave cell type-specific, also in the context of Ikka-mediated NF-kB regulation. For example, the Ikka AA/AA knock-in mutation did not affect LPSinduced canonical NF-kB activation in BM-derived DCs [31], and even induced a small reduction in basal and LPS-induced NF-kB activation in B-cells [15]. Thus, the dissimilar effects of an Ikka AA/ AA knock-in mutation on LPS-induced NF-kB p65 activity in macrophages in our study compared to the one from Lawrence et al. [19] could be due to differential macrophage phenotypes induced by Apoe-deficiency or even by different culturing conditions. Furthermore, NF-kB regulatory mechanisms are often stimulus-dependent, as also exemplified by the observation that an Ikka AA/AA knock-in did not affect Tnf-a-mediated NF-kB activation in fibroblasts or mammary epithelial cells [22]. This could additionally explain why in our study no prolonged p65 activity could be observed in Tnf-a-or oxLDL-stimulated Ikka AA/AA macrophages in vitro.
Although the effect of Ikka on NF-kB stability in macrophages was only described for the isoforms p65 and c-Rel [19], also p50 activity has been reported in atherosclerotic lesions [54][55][56]. Even more, it has been suggested that p50-p50 homodimers represent the main NF-kB activity during inflammation resolution, at least in a rat carrageenin-induced pleurisy model [8], which could correspond to the higher inflammatory phenotype of atherosclerotic lesions in Ldlr 2/2 mice with a haematopoietic p50 deficiency [57]. Although beyond the scope of this study, it would be interesting to perform a detailed characterization of canonical NF- Figure 6. Ikka AA/AA knock-in does not enhance or prolong NF-kB p65 activity, or majorly influence cytokine expression in Apoe 2/2 macrophages in vitro. (A) BM-derived macrophages from Ikka AA/AA Apoe 2/2 and Ikka +/+ Apoe 2/2 mice were stimulated in vitro with 10 ng/ml Tnf-a, 50 mg/ml of mildly oxidized LDL or 100 ng/ml LPS for the indicated time. Activation of p65 was quantified in nuclear extracts using a TransAm p65 assay. Graphs represent the mean 6 SEM (n = 2); 2-way ANOVA with Bonferroni post-test, ***P,0.001. (B) BM-derived macrophages from Ikka AA/ AA Apoe 2/2 and Ikka +/+ Apoe 2/2 mice were stimulated in vitro for 24 h with 10 ng/ml Tnf-a or 50 mg/ml heavily oxidized LDL. Cytokine concentrations in the supernatants are displayed for Tnf-a, Il-6 and Mcp1. Graphs represent mean 6 SEM (n = 9 from 3 independent experiments); 2-way ANOVA with Bonferroni post-test, ***P,0.001. (C) Concentrations of Tnf-a, Mcp1 and Il-6 in serum of Apoe 2/2 mice transplanted with Ikka AA/AA Apoe 2/2 or Ikka +/ + Apoe 2/2 BM and receiving a high-fat diet for 8 weeks. Graphs represent the mean 6 SEM (n = 7-9). doi:10.1371/journal.pone.0087452.g006 kB isoform activity in different stages of atherosclerosis and investigate a potential regulation by the IKKa kinase under these specific atherosclerotic conditions in vivo. Simultaneously, the activity of non-canonical NF-kB isoforms in the course of atherogenesis could be readdressed. Although a single study reported the absence of p52 and RelB activity in isolated human atherosclerotic plaque cells [55], a recent report identified a significant upregulation of IKKa and p52 during human monocyte-macrophage differentiation in vitro. This enables p52mediated transcriptional repression under basal conditions, possibly preventing macrophage hyperactivation, but facilitates enhanced RelB-52 activity upon inflammation-induced RelB expression [58].
Our study did not reveal significant differences in inflammatory protein levels in the supernatants of Tnfa-or oxLDL-stimulated Ikka AA/AA Apoe 2/2 vs Ikka +/+ Apoe 2/2 BM-derived macrophages, with the exception of Il-12 which showed a significant increase in oxLDL-stimulated macrophages upon Ikka AA/AA knock-in ( Figure 6B, Figure S6). Furthermore, we could not detect significant differences in macrophage-related inflammatory cytokine and chemokine levels in the serum of atherosclerotic Ikka AA/ AA Apoe 2/2 vs Ikka +/+ Apoe 2/2 BM chimeras ( Figure 6C). Thus, despite the diverse roles of the Ikka kinase in modulating gene expression in an NF-kB-dependent or -independent manner [4,14], these functions do not seem strong enough in the context of atherogenesis to produce a major effect on systemic protein expression upon Ikka AA/AA knock-in.
Furthermore, it is important to remember that the identification of novel IKKa substrates continuously extends the molecular pathways and biological processes affected by this kinase [4,14]. Also, atherosclerosis is influenced by many leukocyte subsets [1,2]. Therefore, it is conceivable that our overall zero effect of the BMspecific Ikka AA mutation on atherosclerosis is at least partially the result of counterbalanced effects on different biological processes in macrophages or even different leukocyte subsets. For example, it would be interesting to study the effect of a DC-or T reg -specific Ikka AA/AA mutation on atherogenesis. Also, both IKKa and IKKb were shown to be important in neutrophil chemotaxis to HMGB1, a nuclear protein released by necrotic cells, but the functions of the IKKa kinase activity in neutrophil responses and molecular signalling in the context of inflammation and atherosclerosis have not yet been investigated. In addition, the role of IKKa in vascular cells remains to be investigated in more detail.
In conclusion, our data identify an important and previously unrecognized role for the haematopoietic Ikka kinase activity in Band T-cell homeostasis in conditions of atherosclerosis. However, the BM-specific Ikka AA knock-in in atherosclerotic mice did not affect the size or phenotype of atherosclerotic lesions. This indicates that the diverse functions of Ikka in haematopoietic cells may counterbalance each other or may not be strong enough to influence atherogenesis, and reveals that targeting haematopoietic Ikka kinase activity alone may not represent a suitable therapeutic approach. Although the overall zero effect on atherosclerosis is surprising at first sight, it has been observed before that deficiency of proteins with an important role in inflammatory signalling and biological processes does not induce any changes in the size or composition of atherosclerotic lesions, as for example described for BM-deficiency of Cd40 ligand [59,60] or Traf6 [61]. Also, atherosclerosis was not affected in Ldlr 2/2 mice with a BM p16 INK4a -deficiency [62], despite the fact that p16 INK4a is a regulator of macrophage activation and polarization and p16 INK4a -deficiency reduces LPS-induced NF-kB activation in BMderived macrophages [63]. Clearly, it would be interesting to address in the future the role of the IKKa kinase in atherosclerosis in different leukocyte subsets individually and in vascular cells.  Figure S3 Effect of a bone marrow-specific Ikka AA/AA knock-in on central memory T-cells. Shown is flow cytometric analysis of Cd44 high Cd62L high central memory T-cells in spleen and lymph nodes from Apoe 2/2 mice transplanted with Ikka AA/AA Apoe 2/2 or Ikka +/+ Apoe 2/2 BM and receiving a highcholesterol diet for 13 weeks. Data are represented as percentage of Cd3 + T-cells (left) and as percentage of Cd45 + leukocytes (right). Graphs represent the mean 6 SEM (n = 18-19), 2-tailed t-test, *P,0.05, **P,0.01, ***P,0.001. (DOCX) Figure S4 Effect of a bone marrow-specific Ikka AA/AA knock-in on B-and T-cell populations in a nonatherosclerotic context. Shown is flow cytometric analysis of spleen and lymph nodes from C57BL/6 mice transplanted with Ikka AA/AA or Ikka +/+ BM. Dead cells were excluded using Sytox Blue. (A) B220 + B-cell population as percentage of leukocytes, and the total number of B-cells in spleen and lymph nodes. (B) Cd4 + and Cd8a + T-cell subsets as percentage of leukocytes, and as percentage of Cd3 + T-cells. (C) Total number of Cd3 + Cd4 + and Graphs represent the mean 6 SEM (n = 3); 2-way ANOVA with Bonferroni post-test. (B-D) Intracellular lipid depositions in aortic root lesions were stained with Nile Red and co-stained with Mac2 in order to quantify lipid-laden macrophages. (B) Nile Red staining was quantified relative to the plaque area (left graph), and Nile Red + cells were quantified as percentage of total plaque cells (right graph). (C) Lipid uptake by macrophages was quantified as Nile Red + Mac2 + area as % of Mac2 + area (left graph), and as Nile Red + Mac2 + cells as % of Mac2 + -cells. (D) Shown are representative pictures from Nile Red staining (red fluorescence; left image). The middle and right image demonstrate Image J analyses. Displayed are the plaque area (thin red line), macrophage area (green), plaque cell nuclei (blue), Nile Red + area (yellow) and Mac2 + Nile Red + area (white, lipid deposits in macrophages). Graphs represent the mean 6 SEM (n = 5-6). doi:10.1371/journal.pone.0087452.g007