Endotoxemia by Porphyromonas gingivalis Alters Endocrine Functions in Brown Adipose Tissue

Improvement of obesity is important for increasing longevity. The characteristics, size, and function of adipocytes are altered in patients with obesity. Adipose tissue is not only an energy storage but also an endocrine organ. Alteration of endocrine activities in adipose tissue, among them the functional decline of brown adipose tissue (BAT), is associated with obesity. Periodontal disease is a risk factor for systemic diseases since endotoxemia is caused by periodontal bacteria. However, the effect of periodontal disease on obesity remains unclear. Thus, this study aimed to investigate the effect of endotoxemia due to Porphyromonas gingivalis, a prominent cause of periodontal disease, on the BAT. Herein, endotoxemia was induced in 12-week-old C57BL/6J mice through intravenous injection of sonicated 108 CFU of P. gingivalis (Pg) or saline (control [Co]) once. Eighteen hours later, despite no inflammatory M1 macrophage infiltration, inflammation-related genes were upregulated exclusively in the BAT of Pg mice compared with Co mice. Although no marked histological changes were observed in adipose tissues, expressions of genes related to lipolysis, Lipe and Pnpla2 were downregulated after P. gingivalis injection in BAT. Furthermore, expression of Pparg and Adipoq was downregulated only in the BAT but not in the white adipose tissues, along with downregulation of Ucp1 and Cidea expression, which are BAT-specific markers, in Pg mice. Microarray analysis of the BAT showed 106 differentially expressed genes between Co and Pg mice. Gene set enrichment analysis revealed that the cholesterol homeostasis gene set and PI3/Akt/mTOR signaling gene set in BAT were downregulated, whereas the TGF-β signaling gene set was enriched in Pg mice. Overall, intravenous injection of sonicated P. gingivalis altered the endocrine functions of the BAT in mice. This study indicates that endotoxemia by P. gingivalis potentially affects obesity by disrupting BAT function.


INTRODUCTION
Obesity is a major public health concern worldwide and a risk factor for type 2 diabetes (Ozcan et al., 2004), and is also associated with multiple cancers (Key et al., 2003;MacInnis and English, 2006;Li et al., 2009;Lichtman, 2010). Furthermore, obesity itself increases mortality (Whitlock et al., 2009;Berrington de Gonzalez et al., 2010). Adipocytes in the adipose tissue produce various adipocytokines, including adiponectin and inflammatory cytokines (Hotamisligil, 2006;Kadowaki et al., 2006;Yoon et al., 2006), and play important roles in metabolism. Therefore, adipose tissue is considered an endocrine organ.
On the other hand, periodontal disease is also a global public health concern with its high prevalence (Pihlstrom et al., 2005). Periodontal disease results from chronic infections of periodontal bacteria, including Porphyromonas gingivalis, and leads to the destruction of bone and tissue around the teeth (Nassar et al., 2007). Periodontal disease results in not only tooth loss but also the aggravation of numerous types of systemic diseases including type 2 diabetes, cardiovascular disease, preterm low birth weight, and nonalcoholic fatty liver disease (Komazaki et al., 2017;Figuero et al., 2020;Genco and Borgnakke, 2020;Orlandi et al., 2020;Polak et al., 2020;Schenkein et al., 2020).
The association between obesity and periodontal disease has recently attracted increasing attention. Some studies reported that the prevalence of periodontitis is higher among individuals with obesity (Saito et al., 1998;Al-Zahrani et al., 2003;Katagiri et al., 2010). In addition, we previously reported that intravenous injections of sonicated P. gingivalis twice per week for 12 weeks to mice fed high-fat diet caused an increase in body weight and the accumulation of visceral and subcutaneous fat in mice (Sasaki et al., 2018). However, the direct effects of P. gingivalis on adipocytes remain unclear.
In this study, we investigated the effect of endotoxemia resulting from periodontal disease on adipose tissue upon intravenous injection of ultrasonicated P. gingivalis in mice, followed by a comprehensive analysis of gene expression in the brown adipose tissue (BAT).

Animals
C57BL/6J male mice (12-week-old; Sankyo Laboratory, Tokyo, Japan) were used in this study. The mice were provided ad libitum access to food and water throughout the study and housed under standard conditions on a 12-h light/dark (light: 8:00 to 20:00) cycle. Mice were randomly divided into two groups: those intravenously injected with 10 8 CFU of sonicated P. gingivalis suspended in 100 mL of saline (Pg group) and those receiving only saline (control [Co] group). Visceral white adipose tissue (eWAT), subcutaneous white adipose tissue (iWAT), and BAT from interscapular fat were harvested 18 h after P. gingivalis injection. All protocols regarding animal use and euthanasia were reviewed and approved by the Animal Care Committee of the Experimental Animal Center at Tokyo Medical and Dental University (A2020-054A). P. gingivalis Culture P. gingivalis (ATCC 33277) was cultured, as previously described (Sasaki et al., 2018;Udagawa et al., 2018), on trypticase soy agar (Difco Laboratories, Detroit, MI, USA) supplemented with 10% defibrinated horse blood, hemin, and menadione at 37°C under anaerobic conditions. After 48 h, P. gingivalis was inoculated in trypticase soy broth under anaerobic conditions and cultured at 37°C under anaerobic conditions to the mid-log phase, and then 10 9 CFU/mL of the bacterial suspension was sonicated at an amplitude of 20 kHz for 5 min on ice using a Qsonica Q700 sonicator (Waken Btech, Kyoto, Japan).

Quantitative Reverse-Transcription PCR Analysis
Total RNA was extracted from the eWAT, iWAT, and BAT (n = 7) using Trizol reagent (Invitrogen, Carlsbad, CA, USA) and NucleoSpin ® RNA kit (TaKaRa Bio, Shiga, Japan) in accordance with the manufacturer's instructions. Five-hundred nanograms total RNA was reverse-transcribed to cDNA, using the PrimeScript ™ RT Master Mix (TaKaRa Bio). Real-time PCR was performed using the Thermal Cycler Dice ® Real Time System II (TaKaRa Bio). PCR mixtures were prepared using TB Green Premix Ex Taq ™ II (TaKaRa Bio). PCR was carried out in accordance with the manufacturer's instructions. Gene expression levels were normalized to those of the reference gene, Rn18s. The PCR primers used herein are listed in Supplementary Table S1.

Isolation of Stromal Vascular Fractions From the BAT
Stromal vascular fractions (SVFs) were isolated from the eWAT, iWAT, and BAT (n = 6) 18 h after P. gingivalis injection. Adipose tissues were finely minced and digested with collagenase (Wako, Osaka, Japan) with Krebs-Henseleit-HEPES buffer (pH 7.4) supplemented with 1% BSA and 0.2% glucose at 37°C for 60 min. Thereafter, the samples were strained through a 100µm cell strainer and fractionated through centrifugation at 1,500 rpm for 5 min. The pellets were collected as cells in the SVF.

Microarray Analysis
The Agilent Low Input Quick Amp Labeling kit (Agilent Technologies, Santa Clara, CA, USA) was used, in accordance with the manufacturer's instructions, to generate complementary RNA (cRNA) from 200 ng total RNA for single-color (Cy3) microarray analysis (n = 4). Thereafter, cRNAs were analyzed through hybridization onto an Agilent SurePrint G3 Unrestricted Gene Expression 8 × 60 K Microarray (Agilent Technologies). Fluorescence signals were detected using the Agilent Microarray Scanner System (Agilent Technologies). Raw microarray data were extracted using Feature Extraction Software (ver. 11.0.1.1; Agilent Technologies).

Statistical Analysis
Data distributions were analyzed using the Shapiro-Wilk test, revealing all datasets were normally distributed. Unpaired t-test was performed to compare the two groups. Data were analyzed using R (ver. 3.6.0). Microarray data were quantile-normalized, log 2 -transformed, and identified differentially expressed genes (DEGs) by using R with the Limma Bioconductor package (ver. 3.40.6) (Ritchie et al., 2015). Benjamin and Hochberg's false discovery rate (FDR) was applied for multiple testing. DEGs were defined in accordance with an FDR q <0.1 and a |fold-change| >1.5. Overrepresentation enrichment analyses for DEGs were performed using the WEB-based Gene SeT AnaLysis Toolkit (http://www.webgestalt.org) (Wang et al., 2013) and the Database for Annotation, Visualization, and Integrated Discovery (DAVID) (http://david.abcc.ncifcrf.gov/) using the Gene Ontology (GO) and KEGG pathway databases. Gene set enrichment analysis (GSEA) (http://software.broadinstitute.org/ gsea/index.jsp) (Subramanian et al., 2005) was performed using hallmark gene sets (Liberzon et al., 2015).

P. gingivalis Injection Increased Inflammation-Related mRNA Expression in the BAT
The mRNA expression levels of tumor necrosis factor-a (Tnfa), monocyte chemotactic protein-1 (MCP-1) (Ccl2), and interleukin 1 beta (Il1b) were not significantly altered in the eWAT and iWAT following P. gingivalis injection. However, the expression of Tnfa and Il1b was significantly upregulated in the BAT in Pg mice compared to the Co mice ( Figures 1A, B). Interestingly, the mRNA expression level of Ccl2 was dramatically increased only in the BAT following P. gingivalis injection ( Figure 1C).

No Alteration in Macrophage Infiltration Was Observed in SVF From Adipose Tissues
Macrophage infiltration in the SVF from the eWAT, iWAT, and BAT was evaluated through flow cytometry, respectively. P. gingivalis injection caused no significant changes in the percentage of CD11b positive cells in the SVF from eWAT, iWAT, and BAT (Figures 2A, B). Few M1 macrophages were observed in eWAT, iWAT, and BAT of Co and Pg mice ( Figure  2C). The percentage of M2 macrophages in CD11b+ cells was 28.7% and 33.1% in the SVF from eWAT of Co and Pg mice, 49.3% and 50.7% in the SVF from iWAT of Co and Pg mice, and 39.2% and 42.9% in the SVF from BAT of Co and Pg mice, respectively. The percentage of M1 and M2 macrophages in eWAT, iWAT, and BAT did not significantly differ between Co and Pg mice (Figures 2D, E).

P. gingivalis Injection Downregulated mRNA Expression of Genes Related to Lipolysis and Metabolism in BAT
Although no marked histological changes were observed after P. gingivalis injection in adipose tissues ( Figures 3A, B), the expression of lipase, hormone sensitive (Lipe) and patatin-like altered in the eWAT and iWAT following P. gingivalis injection. However, the mRNA expression of Pparg in the BAT of Pg mice was significantly decreased compared to Co mice ( Figure 3F). Moreover, P. gingivalis injection significantly downregulated Adipoq expression in BAT ( Figure 3G).

P. gingivalis Injection Altered Gene Expression Patterns in the BAT
To investigate changes in gene expression in the BAT at 18 h after injection of sonicated P. gingivalis, microarray analysis was performed to obtain a comprehensive overview of gene expression profiles. All microarray data herein are available in the Gene Expression Omnibus database (www.ncbi.nlm.nih.gov/ geo) under GSE 153516. As shown in Figure 4A, among 106 DEGs (|fold change| > 1.5 and q < 0.1), 60 genes were upregulated, and 46 genes were downregulated. Gene expression patterns substantially differed between Co and Pg mice ( Figure 4B).

P. gingivalis Injection Altered BAT Metabolism
Gene ontology was assessed using GO slim for upregulated ( Figure 5A) and downregulated ( Figure 5B) DEGs. Notably, 42% of upregulated DEGs with GO terms were classified under "metabolic process" in the biological process category, whereas 79% of downregulated DEGs were classified under "metabolic process".
In GO analysis, genes were extracted in accordance with the GO term "lipid metabolic process" or "glucose metabolic process" from all detectable genes through microarray analysis. In total, 615 genes were identified under the GO term "lipid metabolic process" (Figure 6A), and 100 genes contained GO term "glucose metabolic process" ( Figure 6B). As shown in Figure 6, Co and Pg mice presented markedly different gene expression patterns for lipid metabolic process and glucose metabolic process.

P. gingivalis Injection Downregulated the Transcript Levels of Several Relevant BAT Marker Genes and Endocrine-Derived Factors from BAT
The expression levels of uncoupling protein 1 (Ucp1) and cell death-inducing DNA fragmentation factor, alpha subunit-like effector A (Cidea), which are BAT markers, were significantly downregulated in the BAT of Pg mice compared to those in Co mice, although the expression levels of fatty acid desaturase 3 (Fads3), a white adipocyte tissue marker, was comparable between Co and Pg mice ( Figure 9A). Interleukin 6 (Il6) expression in the BAT of Pg mice tended to be increased in Pg mice. The expression of fibroblast growth factor 21 (Fgf21) and ceramide synthase 3 (T3) (Cers3) was not significantly altered in the BAT after P. gingivalis injection. However, chemokine (C-X-C motif) ligand 14 (Cxcl14) and neuregulin 4 (Nrg4) expression were significantly downregulated in the BAT of Pg mice compared to those of Co mice ( Figure 9B).

DISCUSSION
A previous epidemiological study showed an association between periodontal disease and obesity in 1998 (Saito et al., 1998). In addition, another paper showed the causal effects of P. gingivalis infection that increased body weight in mice (Yoneda et al., 2012). We also reported that multiple intravenous injections of sonicated P. gingivalis increased body weight and the accumulation of visceral and subcutaneous fats in mice fed high-fat diet (Sasaki et al., 2018). However, the mechanisms, especially the direct effects of P. gingivalis infection on adipocytes still remained unclear, and this study was conducted to address this issue. In this study, endotoxemia was induced through an intravenous injection of ultrasonicated P. gingivalis to simulate endotoxemia by periodontal disease. Endotoxemia occurs in patients with severe periodontal diseases (Ide et al., 2004;Forner et al., 2006;Tonetti et al., 2007). P. gingivalis is a representative periodontal pathogen and contains various virulence factors including lipopolysaccharide, fimbriae, and enzymes (Kolenbrander et al., 2002). In addition, we have recently reported that only P. gingivalis infection, but not other periodontopathic bacteria including Aggregatibacter actinomycetemcomitans and Fusobacterium nucleatum, is significantly associated with intramuscular adipose tissue content in the lumbar muscles (Watanabe et al., 2020). Based on the above results, we focused on P. gingivalis infection. Although only a single oral administration of periodontal pathogen could not cause periodontitis in mice (Lalla et al., 1998;Lalla et al., 2000), the purpose of this study was to observe the differences in gene expression due to endotoxemia induced by P. gingivalis in adipocytes. Thus, we used mice fed normal chow and performed only a single injection of P. gingivalis.
Interscapular fat is different from other adipose tissues, such as visceral and subcutaneous adipose tissues. Interscapular fat is rich in brown adipocytes; whereas, visceral and subcutaneous adipose tissues are rich in white adipocytes (Wang and Seale, 2016). BAT differs from white adipose tissue which works as an energy storage, and BAT can generate heat and expend energy by consuming glucose and fatty acid. . Since several reports have shown that BAT is present in adult humans (Cypess et al., 2009;Saito et al., 2009), numerous studies have been performed to prevent and improve obesity and metabolic diseases by focusing on the BAT (van der Lans et al., 2013;Yoneshiro et al., 2013;Hiraike et al., 2017).
Macrophages themselves produce inflammatory cytokines including TNF-a and CCL2. M1 macrophages produce inflammatory cytokines whereas M2 macrophages suppress inflammation (Lumeng et al., 2007). Interestingly, Tnfa and Ccl2 expression were upregulated exclusively in the BAT after P. gingivalis injection without infiltration of inflammatory M1 macrophages, which suggests that P. gingivalis injection altered adipocytokine production in non-macrophage cells from the BAT, potentially brown adipocytes. The downregulation of Lipe and Pnpla2 expression in the BAT suggested that P. gingivalis injection altered lipid metabolism in BAT. However, only a single P. gingivalis injection did not alter adipocyte size, although there is a possibility that multiple and long-term P. gingivalis injections can cause hypertrophy of adipocytes in BAT. Adipoq and Pparg expression in BAT were downregulated by P. gingivalis injection, even though there was no significant difference in the expression of Adipoq and Pparg in eWAT and iWAT between Co and Pg mice. Adiponectin is an adipocytederived hormone that improves dyslipidemia and insulin resistance (Yamauchi et al., 2001), and activation of the adiponectin receptor can improve obesity-related diseases (Okada-Iwabu et al., 2013). Pparg regulates glucose and lipid metabolism (Berger and Moller, 2002) and directly regulates numerous genes involved in the functions of adipocytes, lipid transport, insulin signaling, and adipokine production (Lehrke and Lazar, 2005). Agonist PPARg ligands are already used for the treatment of type 2 diabetes (Lehmann et al., 1995).  Furthermore, adiponectin expression is controlled by PPARg. In obesity, inflammation is known to inhibit PPARg expression and function, thus inhibiting its direct targets such as adiponectin (Guilherme et al., 2008). Therefore, our result about downregulation of Adipoq could be an indirect effect of downregulation in Pparg expression. Microarray analysis also revealed the possibility of alterations in glucose and lipid metabolisms. GSEA supports these results since the "cholesterol homeostasis" and "PI3K/Akt/mTOR" gene sets were downregulated. Downregulation of PI3K/Akt/mTOR signaling gene set indicates downregulation of insulin signaling (Taniguchi et al., 2006). As previously reported, BAT is also associated with glucose homeostasis and insulin sensitivity. Therefore, this result suggests that P. gingivalis injection may cause a decrease in glucose homeostasis and insulin sensitivity in brown adipocytes (Stanford et al., 2013). Cholesterol imbalance has been recognized as a feature of obese adipocytes (Yu et al., 2010). Downregulation of cholesterol homeostasis gene set suggested that this can be extrapolated to obese mice. As previously reported, TGF-b levels correlate with obesity in mice (Samad et al., 1997;Samad et al., 1999) and humans (Alessi et al., 2000;Fain et al., 2005;Lin et al., 2009). Moreover, systemic blockade of TGF-b signaling protected mice from obesity, diabetes, and hepatic steatosis in a previous study (Yadav et al., 2011). Thus, upregulation of TGF-b signaling also indicates that the BAT in Pg mice may be the similar state to that in obese mice. In addition, we showed differential gene expression patterns for lipid metabolic process and glucose metabolic process between Co and Pg mice, and the percentage of downregulated "metabolic process" DEGs in the biological  process category were larger than upregulated DEGs. These results suggested that metabolic function in the BAT was altered by P. gingivalis injection. On the other hand, inflammation-related gene sets, TNF-a signaling via NF-kB gene set, IL6/JAK/STAT3 signaling gene set, and inflammatory response gene set, were enriched in the BAT, concurrent with the results of quantitative PCR, and supported evidence regarding inflammation in the BAT after P. gingivalis injection. Surprisingly, just a single intravenous injection of P. gingivalis downregulated the expression of Ucp1 and Cidea in the BAT. UCP1 is specific and responsible for heat production in the BAT and is involved in ATP synthesis (Nedergaard et al., 2005). Loss of UCP1 function enhanced obesity in mice on a high-fat diet (Feldmann et al., 2009). In addition, Cidea is a specific and important gene in BAT that regulates adipocyte differentiation and tight coupling of lipolysis and lipogenesis (Wang and Seale, 2016). CIDEA, a lipid droplet protein, and UCP1 are regulated by PPARg (Berger and Moller, 2002;Puri et al., 2008). Based on these results and the fact that BAT consumes high amount of glucose (Shibata et al., 1989;Liu et al., 1994), we suggest that the intravenous injection of sonicated P. gingivalis downregulated the expression of Pparg followed by Ucp1 and Cidea, potentially altering glucose and lipid metabolism. In this study, Il6 expression tended to be increased in BAT from Pg mice. IL-6 is widely recognized as a batokine, it is a BAT-derived secretable factor with pleiotropic actions, mediating glucose homeostasis, insulin sensitivity, and thermogenesis in BAT (Stanford et al., 2013;Villarroya et al., 2017), which is against classic inflammation-mediated effects and performs relevant metabolic functions by increasing insulin sensitivity in the muscle, promoting browning of WAT and M2 macrophage polarization (Mauer et al., 2014) or more recently, mediating stress responses and liver gluconeogenesis (Qing et al., 2020). In addition, just a single intravenous injection of P. gingivalis downregulated the expression of Cxcl14 and Nrg4 in the BAT. These endocrine-derived factors from BAT can affect distant organs. CXCL14 from BAT appears to influence the recruitment of M2 macrophages to subcutaneous WAT. In addition, a lack of CXCL14 in BAT correlated with an impairment of BAT function . Nrg4 from BAT acts on the liver to attenuate hepatic lipogenic signaling (Wang et al., 2014). Our results support the previous report that P. gingivalis infection promoted liver steatosis in mice (Sasaki et al., 2018). These results suggested that endotoxemia due to P. gingivalis directly affects the endocrine function in the BAT.
We previously reported that intravenous injections of sonicated P. gingivalis; twice per week for 12 weeks, increased visceral and subcutaneous fat, impaired glucose tolerance and insulin resistance, and resulted in liver steatosis and inflammation in mice administered a high-fat diet (Sasaki et al., 2018). In addition, a few other studies have reported the effect of P. gingivalis on adipocytes. Adiponectin was significantly downregulated in 3T3-L1 adipocytes treated with P. gingivalis LPS rather than with Escherichia coli LPS (Le Sage et al., 2017). Furthermore, successful periodontal treatment increased serum adiponectin levels in patients with type 2 diabetes (Bharti et al., 2013). Although the main contributor of systemic adiponectin level is the WATs, not the BAT (Giralt et al., 2016), in our results and from previous reports may indicate circulation levels of adiponectin may be different between Co and Pg mice. However, we feel it is difficult to detect differences in only one time point experiment at 18 h after P. gingivalis injection. Thus, further studies, especially long term and with multiple P. gingivalis injections evaluating systemic effects by altering BAT metabolism are required to determine the effect of periodontal infection on adipose tissues. In addition, BAT function is mainly controlled by brown adipose tissues, but also is controlled by several other cells including preadipocyes, endothelial cells, immune cells, and sympathetic neurons Villarroya et al., 2018). Therefore, it is unclear whether our results are direct effects of P. gingivalis endotoxemia on brown adipocytes. Further in vitro studies are required to determine the specific pathways of P. gingivalis endotoxemia on isolated brown adipocytes.
In conclusion, endotoxemia due to sonicated P. gingivalis may directly affect the endocrine function in the BAT. Moreover, periodontal bacterial infections potentially cause alterations in BAT functional markers and immunometabolic characteristics. To our knowledge, this is the first study to comprehensively evaluate gene expression profiles in BAT after endotoxemiainduction through P. gingivalis injection in mice. The present results suggest that endotoxemia by P. gingivalis affects obesity by disrupting adipocyte function.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi. nlm.nih.gov/geo/, GSE 153516.

ETHICS STATEMENT
The animal study was reviewed and approved by the Animal Care Committee of the Experimental Animal Center at Tokyo Medical and Dental University.

AUTHOR CONTRIBUTIONS
MH and YO performed most of the experiments and wrote the first draft of the manuscript. HNii, KM, YT, SY, TShim, NS, SM, TShib, TH, HTo, HTa, and TI assisted in some studies and reviewed the manuscript. SK and HNit supervised all the studies and the writing of the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by the Japan Society for the Promotion of Science (20H03863 to SK, 19K24062 to NS, 19K18989 to SM, and 19K19016 to TShib).