The effect of proatherogenic pathogens on adipose tissue transcriptome and fatty acid distribution in apolipoprotein E-deficient mice

Background Chronic infections have been demonstrated to maintain low-grade systemic inflammation and associate with atherosclerosis. We studied the inflammation- and lipid homeostasis-related effects of Aggregatibacter actinomycetemcomitans (Aa) and Chlamydia pneumoniae (Cpn) infections on the epididymal and inguinal adipose tissue (AT) transcriptomes and fatty acid distribution in apolipoprotein (apo) E-deficient mice. Chow-fed apoE-deficient mice were exposed to 1) chronic intranasal infection with C. pneumoniae (Cpn group), 2) recurrent intravenous infection with A. actinomycetemcomitans (Aa group), 3) a combination of both types of infection (Cpn + Aa group), or 4) infection with the vehicle (control group). Epididymal and inguinal AT gene expression was analyzed using an Illumina Mouse WG-6 v2.0 platform and quantitative PCR (QPCR). Microarray data were analyzed using Gene Ontology enrichment analysis. AT fatty acid analysis was performed using gas–liquid chromatography. Results The transcriptomics data revealed significant enrichment in inflammation-associated biological pathways in both AT depots derived from the Aa and Cpn + Aa treated mice compared with the control group. The proportion of saturated fatty acids was higher in the inguinal AT in Aa (p = 0.027) and Cpn + Aa (p = 0.009) groups and in the epididymal AT in Aa group (p = 0.003). The proportion of polyunsaturated fatty acids was significantly lower among all Aa-infected groups in both depots. Chronic Cpn infection displayed only minor effects on transcriptomics and fatty acids of the AT depots. Conclusions Systemic infection with A. actinomycetemcomitans activates inflammation-related biological pathways and modulates cellular lipid homeostasis. The adverse changes in adipose tissues during chronic infection may promote atherosclerosis.


Background
Atherosclerosis is an inflammatory disease with complex multifactorial pathophysiology. In addition to the chronic inflammation in the vessel wall causing vascular endothelial dysfunction, the progression of the disease may be enhanced by persistent systemic infection, lowgrade inflammation, and metabolic disturbances of major organs such as liver and adipose tissue (AT).
AT is a multi-functional metabolic organ composed of a heterogeneous cell population. By storing and releasing fatty acids and secreting several important adipokines, AT plays a significant role in both energy metabolism and homeostatic regulation of the endocrine system. The AT dysfunction, followed by the spill-over of free fatty acids (FFAs) into the circulation via enhanced lipolysis and dysregulated expression of adipokines, modifies immune responses and contributes to metabolic disorders such as metabolic syndrome and type 2 diabetes [1,2].
AT depots are dispersed throughout the body and are closely associated with several organs [3]. Recent studies have reported many site-specific physiological characteristics in AT depots, i.e. differences in cell populations, microvasculature, innervation, and regional variation in metabolic functions including adipokine profiles, lipase activities, uptake of fatty acids, and the insulin response [4]. These differences have a major impact on the proinflammatory potential of the various AT depots.
In humans, visceral AT has been shown to generate angiotensin, interleukin (IL) -6, and plasminogen activator inhibitor (PAI)-1 at higher levels and leptin and adiponectin at lower levels than subcutaneous AT [5]. Differences in hormone-sensitive lipase activity and maximum lipolytic capacity between omental and subcutaneous AT depots have been reported; however, these phenomena seem to be strongly correlated to the adipocyte size and their capacity to expand [6]. In obese humans, perivisceral AT has been reported to contain more saturated fatty acids (SFAs) and monounsaturated fatty acids (MUFAs) than subcutaneous AT [7]. In mice, high-fat feeding results in a continuous increase in adipocyte size both in subcutaneous and visceral fat. Interestingly, large adipocytes contain relatively more stearic acid and oleic acid, and these cells produce more tumour necrosis factor (TNF)-α, leptin, angiotensinogen, adipsin, and PAI-1 compared with small adipocytes [8].
Although AT has considerable pro-inflammatory and pro-atherogenic potential, and contributes to systemic innate immune responses, the impact of chronic systemic infections on AT is rarely investigated. Many pathogens related to common chronic infections, such as gastritis, periodontitis, and pulmonary infections, are lipopolysaccharide (LPS) containing Gram-negative species. Although LPS promotes inflammation mainly through its effects on the monocyte/macrophage function, an in vitro study has demonstrated that subcutaneous human adipocytes secrete large quantities of TNF-α when activated by LPS [9]. Szalowska et al. 2011 have shown that LPS activation of human omental fat results in up-regulated expression of inflammation-and angiogenesis-related genes [10].
The periodontal pathogen Aggregatibacter actinomycetemcomitans and the persistent pulmonary infection agent Chlamydia pneumoniae are Gram-negative pathogens and associated with an increased risk for cardiovascular diseases [11][12][13]. Both pathogens have been found in human atherosclerotic lesions [11,14], accelerate atherosclerosis in mouse model [15,16], and are highly common among general population [17,18]. We have recently shown in apolipoprotein E (apoE)-deficient mouse model that systemic A. actinomycetemcomitans and C. pneumoniae infections may induce broad range pro-atherogenic effects including alterations in macrophage cholesterol homeostasis, endothelial function [19], and, moreover, in the liver morphology, fatty acid composition, and inflammation marker profile [20].
In the present study, we investigated the effects of systemic recurrent A. actinomycetemcomitans and chronic C. pneumoniae infections on epididymal and inguinal AT depots in apoE-deficient mice. In addition, combined infection model was used to study the impact of pathogen burden on these depots. ApoE-deficient mice display a proatherogenic plasma lipoprotein profile and they develop atherosclerotic lesions on a chow diet. In addition, due to decreased levels of LPS-neutralizing high-density lipoproteins, apoE-deficient mice show increased susceptibility to LPS-mediated inflammation compared to wildtype mice [21]. In order to include both visceral and subcutaneous AT and to identify their putative depot-specific pro-inflammatory and pro-atherogenic potential, we utilized both epididymal and inguinal AT depots. We focused on molecules associated with inflammation and lipid homeostasis, which were investigated on the level of gene expression and cellular fat composition.

Mice
Male apolipoprotein (apo) E-deficient mice (B6.129P2-Apoe tm1Unc /Crl) were purchased from Charles River Laboratories. The mice were fed regular mouse chow ab libitum and maintained in a germ-free environment at the National Public Health Institute Animal Facilities (Helsinki, Finland). Animal care and experimentation were conducted under Animal Care and Use Committee of National Public Health Institute ethical authorization and were in accordance with the guidelines of the Council of Europe.

Bacterial cultures, infection procedure, and serum analyses
The bacteria cultures and experimental design have been previously reported [20]. Briefly, the clinical A. actinomycetemcomitans strain AT445b (serotype b) was cultivated in a tryptic soy-serum-bacitracin-vancomycin medium at 37°C with 5% CO 2 for 48 h. The C. pneumoniae isolate, Kajaani 7, a Finnish epidemic strain, was isolated and cultured as previously described [22].
The infection procedure is presented in Table 1. The experimental design included chronic C. pneumoniae infection model, which was achieved by three consecutive inoculations, recurrent A. actinomycetemcomitans infection model mimicking repeated systemic exposure to the pathogen, and a combination of these two infection models to imitate the effects of infection burden. Chronic C. pneumoniae infection was induced in younger mice e in order to see the potential effect of persistence. Viable C. pneumoniae (Cpn) was administered intranasally as 2 × 10 6 inclusion forming units in 40 μl of sucrosephosphate-glutamic acid buffer (SPG), and viable A. actinomycetemcomitans (Aa) was administered intravenously (via the tail vein) as 1 × 10 7 colony forming units in 50 μl of 0.9% NaCl. The 10-week experiment included four groups: 1) recurrent Aa infection group (Aa group, n = 10) with ten Aa inoculations; 2) the chronic Cpn infection group (Cpn group, n = 10) with three Cpn inoculations; 3) the combined chronic Cpn and recurrent Aa infection group (Cpn + Aa group, n = 10) with three Cpn inoculations and ten Aa inoculations; and 4) the control group (n = 9) with three SPG inoculations and ten 0.9% NaCl inoculations.
The serum levels of nonesterified fatty acids (Waco Chemicals GmbH), the LPS activity (HyCult biotechnology b.v.), and the serum levels of TNFα (Pierce Biotechnology, Inc.) were measured as described in our previous publications [19,20].

Preparation of adipose tissues
Inguinal and epididymal AT depots were carefully dissected from sacrificed animals. The tissue specimens were stabilized with RNAlater™ (Sigma) or snap-frozen in liquid nitrogen. All the samples were stored at −70°C.
Isolation and quality evaluation of RNA Epididymal and inguinal AT homogenization was performed with a Mixer Mill MM 301 (Retsch GmbH) and total RNA was extracted using the RNeasy Lipid Tissue Mini Kit (Qiagen). DNase treatment was performed on a column using the RNase-Free DNase Set (Qiagen) and again after RNA extraction using the DNA-free™ Kit (Ambion). The RNA concentration was determined using a NanoDrop 1000 spectrophotometer (Thermo Scientific). The quality of total RNA was assessed by an Agilent 2100 Bioanalyzer using the Eukaryote Total RNA Nano Kit (Agilent Technologies).

Adipose tissue RNA microarray processing
Inguinal and epididymal AT gene expression was analyzed using the Illumina Mouse WG-6 v2.0 platform. Three biological replicates were used and the high quality total RNA (RNA integrity number ≥ 8.5) was amplified using the Illumina TotalPrep RNA Amplification Kit according to the manufacturer's instructions. Briefly, the total RNA was reverse-transcribed using T7 Oligo(dT) primers following the second strand cDNA synthesis.
After purification, the cDNA was in vitro transcripted into cRNA using biotin-labeled nucleoside triphosphates. The quality of biotinylated cRNA was assessed by an Agilent 2100 Bioanalyzer and 1.5 μg of cRNA from each sample was hybridized at 58°C overnight using the Illumina Whole-Genome Gene Expression Direct Hybridization Assay system with the six-sample BeadChip platform. The signal was developed using streptavidin-Cy3 and the BeadChips were scanned with an Illumina BeadArray Reader.

Microarray data analyses
The microarray data has been deposited in MIAMEcompliant format in Gene Expression Omnibus (http:// www.ncbi.nlm.nih.gov/geo/), accession number GSE50647. Gene ontology (GO) enrichment analysis for microarray data was performed using the Fisher's exact test, where the observed frequency of each present GO term was compared with the frequency in a reference gene set (control mice). P-values were corrected for multiple hypotheses using the Benjamini-Hochberg false discovery rate [23]. The analyses were performed using the R Base Package version 2.15.0. The maximum of 20 most enriched GO terms were reported for both epididymal and inguinal AT and ranked according to the p-value. To limit the data and focus on the biologically meaningful pathways and molecular functions, GO terms referring to cellular components were excluded. The fold-change (FC) limitation for the presented up-and down-regulated genes was 2.0 except when observing de novo lipogenesis-associated genes where all results are present.

Quantitative real-time PCR
Epididymal and inguinal AT cDNA was synthesized from total RNA using the ImProm-II™ Reverse Transcription system (Promega). The amount of total RNA in each reverse transcription (RT) reaction was 240 ng for the inguinal AT and 500 ng for the epididymal AT. The number of mice in each group was 10, except for the control group where 9 mice were used. Primers (Thermo Scientific) were designed using Beacon Designer (Premier Biosoft International) and National Center for Biotechnology Information Primer-BLAST (http://www.ncbi.nlm.nih.gov/tools/primer-blast/) The black squares represent Cpn inoculations and the black circles Aa inoculations. The controls received the respective vehicle (white squares and circles). †, sacrifice; Aa, A. actinomycetemcomitans; Cpn, C. pneumoniae.
for CD68, glyceraldehyde-3-phosphate dehydrogenase (Gapdh), monocyte chemoattractant protein 1 (Mcp-1), macrophage migration inhibitory factor (Mif), mannose receptor C type 1 (Mrc-1), myeloperoxidase (Mpo), and secretory leukocyte peptidase inhibitor (Slpi). Primer sequences are presented in Additional file 1: Table S1. Quantitative real-time PCR (QPCR) was performed using the Mx3005 Real-Time QPCR System (Stratagene). The analyses were conducted in 12.5 μl reaction mixtures, each containing 2 μl of 1:50 dilution of cDNA synthesized in RT reaction, 6.25 μl of Brilliant SYBR® Green QPCR Master Mix (Stratagene), 30 nM of ROX reference dye (Stratagene), and optimized concentrations of primers (100 nM, except 200 nM for Gapdh). The amplifications were performed using the following thermocycling program: initial denaturation at 95°C for 15 min, 40 cycles of 15 s at 95°C and 1 min at 60°C followed by melting curve analysis for 1 min at 95°C, gradual decrease to 55°C, 30 s at 55°C, gradual increase to 95°C, and 30 s at 95°C. All samples were analyzed as duplicates and the accepted intra-assay coefficient of variation for threshold cycle was ≤ 2.1%. The results were analyzed using Mx3005 Real-Time QPCR System (Stratagene) software and expressed as a log 2 fold change (log 2 FC) to the respective control mice (level set to zero in the analysis). Gapdh was used as a normalization gene and the software utilizes modified comparative quantitation method introduced by Pfaffl (2001) [24].

Adipose tissue fatty acid analysis
Total lipids were extracted from 30-80 mg of epididymal and inguinal adipose tissue with hexane-isopropanol (3:2) [25] and trans-methylated by heating with acidic methanol (5% H 2 SO 4 ) [26]. The resulting fatty acid methyl ester compositions were analyzed by gas-liquid partition chromatography using a Hewlett Packard 6890 Gas Chromatograph (ChemStation B.01.03 software) with a DB-225 GC column (30 m, diameter 0.32 mm, phase layer 0.25 μm; Agilent Technologies). The samples were split injected with hydrogen as a carrier gas and the temperature program was from 160°C to 230°C. The relative amounts of the methylated fatty acids were expressed as percentages of weight normalized to 100%.

Statistical analyses
The AT fatty acid data were expressed as mean percentages of weight with standard deviations (SDs). The gene expression data were expressed as log 2 fold change (log 2 FC) medians with interquartile ranges (IQRs). The significance of the differences between the experimental and control mouse groups was assessed using the nonparametric Mann-Whitney U test. Bivariate correlations were determined using the Spearman test. Analyses were performed using PASW Statistics 18 (Statistical Package for the Social Sciences).

Infection-enriched biological pathways in the inguinal AT
Infection-enriched biological pathways (i.e., GO terms) and the associated up-and down-regulated genes in the inguinal AT transcriptome are presented in Tables 2, 3 and 4. Compared with the control mice, the expression of "Protein binding" pathway (GO:0005515) was increased (p <0.001) most significantly in Aa group (Table 2). In addition, the inflammation-related "Antigen binding" (GO:0003823, p = 0.009), "Antibacterial humoral response" (GO:0019731, p = 0.009), and "Antimicrobial humoral response" (GO:0019730, p = 0.009) pathways were significantly enriched ( Table 2). Compared with the control mice, most of the enriched GO terms in the inguinal AT transcriptome of Cpn group were pathways associated with development-and morphogenesis ( Table 3). The most affected pathway was "Negative regulation of cellular process" pathway (GO:0048523, p = 0.036).
In Cpn + Aa group, a large number of up-regulated genes were immunoglobulin variables and the most upregulated was Ig gamma-2A chain C region secreted form-like (LOC100047788, FC = 37.73) (Additional file 7: Table S7). Lipid homeostasis -associated arachidonate 12lipoxygenase (Alox12, FC = 3.91) was also up-regulated.  model had little impact on de novo lipogenesis associated gene expression ( Figure 1C).

Alterations in QPCR-analyzed mRNA expression levels
The relative mRNA expression levels of neutrophil activation-(Slpi and Mpo) and macrophage phenotype-(macrophage type 2 marker Mrc-1) and activation-(Mcp-1 and Mif) related genes in inguinal and epididymal AT are presented in Table 8. In the inguinal AT, the relative expression levels of genes encoding Slpi and Mpo were significantly higher in the Aa group (Slpi: p = 0.003; Mpo: p = 0.009) than in control mice (level set to zero in the analysis). Cpn infection did not have a significant effect on the selected gene expression levels.
In the epididymal AT, the relative mRNA expression levels of Mcp-1, Mif, and Slpi in the Aa group, Mrc-1 and CD68 in Cpn group, and Mif in Aa + Cpn group were significantly altered but between −1 to 1 log 2 FC . The Cpn + Aa combination induced a reduction in the Mcp-1 mRNA levels (p <0.001).

Changes in AT fatty acid composition
The proportions of fatty acids in the epididymal and inguinal AT are presented in Figure 2. In the inguinal AT, Aa infection induced a significantly higher proportion of SFAs (Aa group: p = 0.027; Cpn + Aa group: p = 0.009; Figure 2A) and a lower proportion of PUFAs (Aa group: p = 0.006; Cpn + Aa group: p = 0.022; Figure 2C) compared to the control mice. The proportion of MUFAs was greater in the Cpn group (p = 0.018; Figure 2B).
In the epididymal AT, the proportions of SFAs were higher among all infected mice but the difference reached statistical significance only in the Aa group (p = 0.003; Figure 2A). The proportion of PUFAs was significantly lower in the Aa (p = 0.005) and Cpn + Aa groups (0.016) ( Figure 2C). When the two AT depots were compared, the proportion of SFAs was lower in the epididymal AT than in the inguinal AT in the control group (17.1 ± 1.8 vs. 20.7 ± 2.1% of total FA, respectively; p = 0.002) and the Aa + Cpn group (19.0 ± 2.3 vs. 24.2 ± 3.0% of total FA, respectively; p = 0.001).

Associations between serum and AT parameters
The associations between different serum and AT variables were determined by correlation analysis (data from all infection models and the controls combined) and presented as graphs in Figure 3. Serum LPS activity correlated positively with the serum FFA concentration (p < 0.001, Figure 3A) and with the proportion of SFA in the epididymal AT (p = 0.030, Figure 3B), and negatively with both inguinal (p = 0.045, Figure 3D) and epididymal (p = 0.013, Figure 3C) proportions of PUFA.

Discussion
In the present study, we examined the inflammationand lipid homeostasis-related effects of systemic A.    the serum levels of LPS activity, TNFα, and IgG antibodies against A. actinomycetemcomitans after 10 intravenous injections of live bacteria are significantly higher in the Aa group compared with the control group [19,20]. The persistence of C. pneumonia after three inoculations was also shown as the pathogen was recovered from the lungs of the mice [20]. Since atherosclerosis is a multiorgan disease, our aim was to broaden the study from the lipid depots in heart and liver to adipose tissue, which plays an important role in sustaining systemic inflammation and proatherogenic lipid profile.
Experimental acute endotoxemia has been shown to induce general insulin resistance and local and adaptive inflammation in human subcutaneous AT [27]. In our study, Aa infection enriched inflammation-related pathways in the inguinal AT. QPCR analysis verified upregulated expression of Mpo and Slpi, and the most upregulated genes in the microarray analysis were S100a8 and Ltf. These innate immunity-associated gene products are considered anti-inflammatory mediators and they exhibit various antimicrobial functions such as promotion of oxidative activity responses [28][29][30]. Human Slpi and Ltf have been shown to co-localize in secondary granules and they are co-released after neutrophil activation [31]. Interestingly, increased plasma Slpi levels were associated with progressive metabolic dysfunction in men [32]. Despite the up-regulation in neutrophil activation-related genes, we did not observe changes in the mRNA expression of genes associated with classically or alternatively activated macrophages. The response of Aa infection in the epididymal AT was quite distinct. As in inguinal AT, the humoral antibacterial response and immunoglobulin-related pathways were enriched. However, the myeloid cell-specific signs of enhanced oxidative stress were absent, expression of Slpi was somewhat down-regulated and, instead, several lipid homeostasis-related biological pathways were enriched.
On AT fatty acid level, we observed a potentially adverse increase in the proportion of SFAs among all Aa-infected mice. Mengting et al. 2010 reported that compared with lean mice, obese mice have a higher level of SFAs in their epididymal AT [33]. An interesting study by Suganami et al. 2007 demonstrated that SFAs released by hypertrophied adipocytes activate the NF-κB pathway in the macrophages with ensuing TNF-α release [34]. The subsequent lipolytic release of SFAs creates a positive feed-back loop between adipocytes and macrophages. Importantly, the macrophagederived TNF-α also resulted in NF-κB-dependent release of proinflammatory cytokines from the adipocytes. Both LPS and SFA have been shown to generate endoplasmic reticulum stress in primary human adipocytes [35]. Elevated levels of SFAs in both epididymal and inguinal AT depots in the Aa and Cpn + Aa groups suggested an increased proinflammatory potential of these AT depots. In addition, because PUFAs have been reported to inhibit the TLR dimerization induced by SFAs [36], the significantly decreased proportion of PUFAs among all A. actinomycetemcomitans-infected mice may intensify the effects of increased SFAs. The proportion of PUFAs also showed a significant inverse association with LPS serum activity. This result may be due to LPS-triggered enhanced oxidative stress in the AT, followed by the peroxidation of PUFAs. In addition to intracellular effects, LPS has been shown to elevate the levels of circulating FFAs in rodents in vivo [37]. This association was also found in our study as there was a strong positive correlation between serum LPS activity and the FFA level.
Lipid homeostasis-related biological pathways were enriched and four important de novo lipogenesis-related genes were moderately up-regulated in the epididymal AT in Aa group. The observed up-regulation of Pcx and Fasn could partially explain the increase in the SFA proportion. However, up-regulation of de novo lipogenesis genes was not observed in Cpn and Cpn + Aa groups, indicating the involvement of other regulatory pathways affecting SFA levels in the AT.
Chronic infection with C. pneumoniae did not enhance the pathways associated with humoral antibody response, and moreover, the effects on the AT depot fatty acid distribution were less evident than with Aa infection. However, the "Defense response" pathway which included four upregulated defensins was enriched in the epididymal AT depot. Previously, C. pneumoniae infection in vitro was reported to induce human β-defensin 2 production mediated by the TLR4 pathway in monocytes [38]. Although both C. pneumoniae and A. actinomycetemcomitans are Gram-negative pathogens, the effects of Cpn infection differed considerably from those of Aa infection. This result may be due to the distinct infection routes and C. pneumoniae being an obligatory intracellular pathogen. In addition, the difference in the infection time points may also influence the outcome and, altogether, the character of chronic infection differs from the recurrent one. The combined infection model displayed a humoral antibody response and changes in AT depot FA distribution similar to those with single Aa infection. Nevertheless, indicators of the potentially amplified de novo lipogenesis in epididymal AT remained absent.
The major limitation of the study was the small number of mice available for the microarray analysis, which resulted in non-significant gene FC-results after FDR correction. Although the apoE-deficient mouse model has been utilized widely in atherosclerosis research, it possesses certain drawbacks, such as a somewhat nonphysiological lipoprotein profile [39]. The adipocytes of apoE-deficient mice are smaller, contain fewer lipids, and accumulate less triglycerides from very low-density lipoproteins and chylomicrons compared with adipocytes derived from wild-type mice [40,41]. In addition, because apoE has been shown to protect against LPSderived sepsis, apoE deficiency actually exposes mice to the deleterious and possibly proatherogenic effects of LPS [21]. Nevertheless, the augmented susceptibility to LPS may produce a more human-like response, because rodents are generally more resistant to LPS-mediated effects than humans. Our study design included only males and therefore we could only draw gender-specific conclusions. It is noteworthy that, in mice, approximately 20% of AT genes exhibit sexual dimorphism (FC >1.2), and the enriched functional categories with FC >3 include "Lipid metabolism" [42].