Overexpression of hepatic serum amyloid A1 in mice increases IL-17-producing innate immune cells and decreases bone density

Serum amyloid A (SAA) is an acute-phase protein produced primarily in the liver that plays a key role in both the initiation and maintenance of inflammation. Rapidly secreted SAA induces neutrophilia at inflammatory sites, initiating inflammation and inducing the secretion of various cytokines, including TNF-α, IL-6, and IL-17. IL-17 is expressed in several inflammatory cells, including innate immune cells such as γδT cells, ILC3 cells, and neutrophils. Increased IL-17 levels exacerbate various inflammatory diseases. Among other roles, IL-17 induces bone loss by increasing receptor activator of nuclear factor-κB ligand (RANKL) secretion, which stimulates osteoclast differentiation. Several studies have demonstrated that chronic inflammation induces bone loss, suggesting a role for SAA in bone health. To test this possibility, we observed an increase in IL-17-producing innate immune cells, neutrophils, and γδT cells in these mice. In 6-month-old animals, we detected increased osteoclast-related gene expression and IL-17 expression in bone lysates. We also observed an increase in neutrophils that secreted RANKL in the bone marrow of TG mice. Finally, we demonstrated decreased bone mineral density in these transgenic (TG) mice. Our results revealed that the TG mice have increased populations of IL-17-producing innate immune cells, γδT cells, and neutrophils in TG mice. We additionally detected increased RANKL and IL-17 expression in the bone marrow of 6-month-old TG mice. Furthermore, we confirmed significant increases in RANKL-expressing neutrophils in TG mice and decreased bone mineral density. Our results provide evidence that chronic inflammation induced by SAA1 causes bone loss via IL-17-secreting innate immune cells.

Serum amyloid A (SAA) is an acute-phase protein produced primarily in the liver that plays a key role in both the initiation and maintenance of inflammation. Rapidly secreted SAA induces neutrophilia at inflammatory sites, initiating inflammation and inducing the secretion of various cytokines, including TNF-α, IL-6, and IL-17. IL-17 is expressed in several inflammatory cells, including innate immune cells such as γδT cells, ILC3 cells, and neutrophils. Increased IL-17 levels exacerbate various inflammatory diseases. Among other roles, IL-17 induces bone loss by increasing receptor activator of nuclear factor-κB ligand (RANKL) secretion, which stimulates osteoclast differentiation. Several studies have demonstrated that chronic inflammation induces bone loss, suggesting a role for SAA in bone health. To test this possibility, we observed an increase in IL-17-producing innate immune cells, neutrophils, and γδT cells in these mice. In 6-month-old animals, we detected increased osteoclast-related gene expression and IL-17 expression in bone lysates. We also observed an increase in neutrophils that secreted RANKL in the bone marrow of TG mice. Finally, we demonstrated decreased bone mineral density in these transgenic (TG) mice. Our results revealed that the TG mice have increased populations of IL-17-producing innate immune cells, γδT cells, and neutrophils in TG mice. We additionally detected increased RANKL and IL-17 expression in the bone marrow of 6-month-old TG mice. Furthermore, we confirmed significant increases in RANKL-expressing neutrophils in TG mice and decreased bone mineral density. Our results provide evidence that chronic inflammation induced by SAA1 causes bone loss via IL-17-secreting innate immune cells.
Of the cytokines secreted by SAA signaling, IL-17 plays an important role in increasing inflammation by activating neutrophils and attracting inflammatory cells to sites of inflammation (10). Thus, increased IL-17 expression is associated with the exacerbation of inflammatory disease (11). Various innate immune cells, including γδT cells, group 3 innate lymphoid cells (ILC3), and neutrophils are also known to secrete IL-17 (10).
Bone is maintained via a balance of osteoclast and osteoblast cell populations, the functions of which are to degrade old bone and generate new bone, respectively (12). Receptor activator of nuclear factor-κB ligand (RANKL) is a signaling molecule that promotes osteoclast differentiation by binding to its receptor RANK, which is expressed in osteoclast precursors (13). IL-17 secretion increases the proportion of osteoclasts by increasing RANKL secretion from osteoblasts. Also, IL-17producing cells, including Th17 cells and γδT cells, express RANKL themselves (14,15). Moreover, several studies have reported that neutrophils can also secrete RANKL (16,17); therefore, IL-17 alters bone homeostasis in a direction that induces bone loss. Also increased IL-17-mediated bone loss has been reported in several chronic inflammatory diseases, including psoriasis (18,19).
In this study, we sought to determine whether increased SAA induced bone loss via increased IL-17 signaling, using SAA1-overexpressing transgenic (TG) mice. We observed an increase in IL-17-producing innate immune cells, neutrophils, and γδT cells in these mice. In 6month-old animals, we detected increased osteoclastrelated gene expression and IL-17 expression in bone lysates. We also observed an increase in neutrophils that secreted RANKL in the bone marrow of TG mice. Finally, we demonstrated decreased bone mineral density in these TG mice. Therefore, we concluded that SAA induced bone loss upon chronic inflammation via IL-17-production in innate immune cells . The SAA1 TG mouse model may  have additional value as a disease model for IL-17mediated bone loss. Results SAA1-overexpressing TG mice display increased IL-17expressing innate immune cells In a previous study, we observed that SAA1 increased IL-17 expression via the Toll-like receptor 2 (TLR2) in γδT cells (20). In this study, we sought to determine whether IL-17 was secreted by other immune cells, in addition to γδT cells in TG mice. Using flow cytometry, we analyzed the population of IL-17 + γδTCR + cells (IL-17-producing γδT cells, Fig. 1A), IL-17 + CD4 + cells (IL-17-producing Th17 cells, Fig. 1B), and lineage − CD45 + CD90.2 + IL-17A + cells (IL-17-producing ILC3, Fig. 1C) in lymph nodes. As observed, IL17 + γδTCR + cells were significantly increased in TG mice when compared with WT mice. However, the proportion of Th17 and ILC3 cells did not differ significantly between these mice. Therefore, we confirmed that SAA1 overexpression in mice led to an increased γδT cell population, but not Th17 or ILC3 cells.

SAA1-overexpressing TG mice display increased neutrophil populations
It was previously reported that SAA increased neutrophilia by increasing G-CSF (1). Neutrophils themselves can express IL-17 (10). To confirm this in our TG mice, we measured G-CSF expression by western blot of serum proteins. We observed increased G-CSF expression in TG mice compared with WT mice ( Fig. 2A). We next measured neutrophil population via flow cytometry. We confirmed the presence of a circulating CD11b + Ly6G + population (neutrophils) in blood, and this was significantly increased in TG mice when compared with WT mice (Fig. 2B). In addition, neutrophils in TG mice were found to secrete IL-17 (Fig. 2C). Consequently, SAA1 TG mice were confirmed as secreting IL-17 from neutrophils and γδT cells.
Increased osteoclast-related mRNA expression with increased IL-23 and IL-17 in SAA1-overexpressing TG mice As other studies have reported bone loss mediated by chronic inflammation (18,19), we also investigated whether bone loss occurred as a result of chronic inflammation induced by SAA1 and IL-17. First, we measured increased SAA1 expression in the serum of 6-month-old mice by western blot. We confirmed increased SAA1 secretion in 6-month-old TG mice compared with WT mice (Fig. 3A). We investigated the expression of the following genes: those related to osteoclast activity; Ctsk, Tnfsf11, Nfatc1, and Acp5 (21), osteoblast/osteocyte activity; Tnfrsf11b, Alpl, Bglap, and Runx2 (22), and Wnt signaling related genes required for proper osteoblast function (23,24); Axin1 Ctnnb1, and Ccn4. No significant differences were observed in the mRNA expression of osteoclast and osteoblast-related markers in whole bone lysates from 7-weekold mice (data not shown). However, when we examined 6month-old TG mice to confirm the effects of chronic inflammation on bone from increased SAA1 and IL-17, we observed that the osteoclast-related genes, Tnfsf11 and Ctsk were significantly increased in TG mice when compared with WT mice (Fig. 3B). We observed no significant differences in osteoblast-associated gene levels between TG and WT mice; however, for Axin1 and Ctnnb1 genes associated with Wnt signaling, and Tnfrsf11b associated with osteoblast activity, these levels tended to decrease in TG mice (Fig. 3, C and D). We measured the expression levels of Il-17a and Il-23a mRNA in bone lysates and confirmed a 367-fold and 11-fold increase, respectively, in TG mice when compared with WT mice (Fig. 3E). Therefore, we concluded that chronic inflammation in 6-month-old TG mice induced an increase in osteoclastrelated gene expression in relation to IL-17. The absolute numbers of IL-17A + γδ TCR + , Th17, and ILC3 cells per 5000 lymph node cells (P1) are shown (n = 4). Detailed gating strategies and information for P1 are shown (Fig. S1). Data are expressed as the mean ± standard deviation, and unpaired two-tailed Student's t-test was applied to compare two groups. *p < 0.05 and **p < 0.01 compared with WT mice.

Serum amlyoid A causing bone loss
Increased Tnfsf11 mRNA expression in the bone marrow from SAA1-overexpressing TG mice A recent report suggested that osteocytes could also express RANKL (i.e., Tnfsf11) when stimulated by IL-17 (15). To investigate RANKL expression in osteocytes, we separated bone and bone marrow. As a result, bone that located osteocyte expressed Tnfsf11 mRNA in both WT and TG mice, and there was no significant difference (Fig. 4A). For bone marrow, . Detailed gating strategies and information for P1 are shown (Fig. S3). Data are expressed as the mean ± standard deviation, and unpaired two-tailed Student's t-test was applied to compare two groups. *p < 0.05 compared with WT mice. The red arrow indicates that A is gated to B. MW, molecular weight. Figure 3. Increased osteoclast-related gene expression and IL-17 and IL-23 expression in whole bone lysates from 6-month-old SAA1overexpressing TG mice. A, upper, circulating SAA1 levels measured by western blot using equal volumes (1 μl) of serum from 6-month-old mice. The molecular weight of SAA1 is 12 kDa (A, bottom). Densitometry analysis of western blots showing SAA1 quantities, n = 3. The ponceau S staining to show that the same volume of serum was loaded is shown (Fig. S2). Relative mRNA expression levels of (B) Ctsk, Tnfsf11, Nfatc1, and Acp5 (C) Tnfrsf11b, Alpl, Bglap, and Runx2 (D) Axin1, Ctnnb1, and Ccn4, and (E) Il17a and Il23a, normalized to glyceraldehyde-3-phosphate dehydrogenase (Gapdh) mRNA expression levels, and relative to WT mice in whole bone lysates. For B-E; WT; n = 4 and TG; n = 8. Data are expressed as the mean ± standard deviation, and unpaired two-tailed Student's t-test was applied to compare two groups. *p < 0.05 and **p < 0.01 compared with WT mice. MW; molecular weight.

Serum amlyoid A causing bone loss
Tnfsf11 mRNA expression was not detected by qPCR in WT mice. However, in TG mice we detected Tnfsf11 expression (Fig. 4B). Therefore, we confirmed that increased Tnfsf11 expression in TG mice was expressed by bone marrow cells.

Increased RANKL expression in bone marrow from SAA1overexpressing TG mice
We sought to determine whether RANKL protein levels were also increased in TG mice and which bone marrow cells expressed these levels. Firstly, we measured RANKL expression in bone marrow using flow cytometry. Our results indicated that RANKL expression was significantly increased in TG mice when compared with WT mice (Fig. 5A). RANKL is known to be expressed by osteoblasts derived from the mesenchymal stem cell lineage (CD45 − cells) and inflammatory cells derived from the hematopoietic stem cell (HSC) lineage (CD45 + cells). Therefore, we investigated CD45 + RANKL + cell and CD45 − RANKL + cell populations to identify the major RANKL-secreting cells between these lineage cells. In TG mice, we observed that RANKL expression increased in CD45 + cells, but not in CD45 − cells. Therefore, we concluded that increased RANKL expression levels in TG mice were secreted from CD45 + inflammatory cells of the HSC lineage, rather than from osteoblasts. Furthermore, we observed that CD45 − cells were significantly reduced in TG mice (Fig. 5B). These results indicated that RANKL expression was increased in inflammatory cells from SAA1 TG mice.
Increased RANKL-producing neutrophils in bone marrow from SAA1-overexpressing TG mice IL-17-expressing cells themselves are known to be capable of expressing RANKL. Therefore, we ascertained whether IL-17-expressing innate immune cells were increased in TG mice. Bone marrow neutrophil and γδT cell populations were determined by flow cytometry. From these data, we observed that γδT cell and neutrophil populations were significantly increased in TG mice when compared with WT mice (Fig. 6). Therefore, we examined RANKL expression in γδT cells and neutrophils in the bone marrow of TG mice using flow cytometry. We confirmed that γδT cells did not express RANKL (Fig. 7A). However, in neutrophils, RANKL expression was significantly increased in TG mice when compared with WT mice (Fig. 7B), indicating that IL-17-expressing neutrophils secreted RANKL and contributed to osteoclast differentiation in SAA1 TG mice.

Reduced bone mineral density and increased TRAP-positive cells in SAA1-overexpressing TG mice
Finally, we investigated whether increased RANKL in SAA1 TG mice resulted in bone loss. Bone mineral density was measured using micro-computed tomography (μCT) in 6month-old TG mice. We confirmed that bone mineral density, bone volume/total volume, trabecular thickness, trabecular number, and cortical thickness were all significantly decreased in TG mice (Fig. 8A). Histological analysis was also performed using TRAP staining in mouse tibia sections; we detected increased TRAP-positive staining in TG mice when compared with WT mice (Fig. 8B). Thus, we confirmed that bone mineral density decreases were correlated with increased osteoclasts in TG mice.

Discussion
It was reported that SAA signaling promotes increased IL-17 expression (4, 5). In our previous study, we demonstrated that SAA1 overexpression in TG mice led to spontaneous psoriasis (20), in which IL-17 plays a key role (25). In this study, we demonstrated that SAA stimulated γδT cells and neutrophils to secrete IL-17. Long-lasting inflammation in several chronic inflammatory diseases, including psoriasis, is related to bone loss (18,19). Therefore, we investigated whether bone loss occurred in SAA1 TG mice, via chronic inflammation with IL-17. We observed that 6-month-old TG mice had decreased bone mineral density and increased RANKL expression, when compared with WT littermate controls. We also observed RANKL-producing neutrophils in these SAA1 TG mice, suggesting that chronic IL-17-secreting neutrophils contributed to decreasing bone mineral density via RANKL signaling in SAA1 TG mice. SAA induces neutrophilia by promoting G-CSF secretion (1, 5). IL-17 accelerates neutrophil infiltration at inflammation sites to both initiate and maintain inflammatory responses (10). Therefore, it is conceivable that interactions between SAA and IL-17 result in neutrophilia in SAA1 TG mice. Despite their short life span, activated neutrophils can produce considerable protein and lipid levels, which participate in inflammatory responses (26,27). Similarly, neutrophils also produce RANKL (16,17). It is known that RANKL + neutrophils interact directly with osteoclasts (17). In addition, it was confirmed in vivo that RANKL + neutrophils contributed to osteoclastogenesis (28). In this study, we were unable to demonstrate the direct effects of neutrophils on bone formation. However, data from previous studies

Serum amlyoid A causing bone loss
have allowed us to speculate on the effects of RANKL + neutrophils on bone homeostasis (17,28). Therefore, our observations that increased numbers of neutrophils, as well as increased IL-17 and RANKL expression levels on surface membranes, suggested that neutrophils played a major role in bone loss in our TG mice.
However, the precise role of IL-17 signaling in osteoclasts and osteoblasts is controversial. IL-17 was recently shown to  4). B, right, detailed gating strategies and information for P1 are shown (Fig. S4). Data are expressed as the mean ± standard deviation and unpaired two-tailed Student's t-test was applied to compare two groups. *p < 0.05 and **p < 0.01 compared with WT mice.  (Fig. S5). Data are expressed as the mean ± standard deviation and unpaired twotailed Student's t-test was applied to compare two groups. *p < 0.05 and **p < 0.01 compared with WT mice.

Serum amlyoid A causing bone loss
reduce Wnt signaling when associated with osteoblast function; this decreased activity resulted from a relative increase in osteoclast function (19). Our results demonstrated that Tnfrsf11b mRNA expression, which is related to osteoblast function (22), and Axin1 and Ctnnb1, which are related to Wnt signaling (23,24), were reduced in SAA1 TG mice. In addition, CD45 − populations were significantly decreased in TG mice, suggesting that IL-17 may have decreased osteoblast  and (B, right, bottom), the counts for RANKL + cells per 5000 CD11b + Ly6G + cells in bone marrow (E1) are shown (n = 4). Detailed gating strategies and information for P1 and E1 are shown (Fig. S6). Data are expressed as the mean ± standard deviation and unpaired two-tailed Student's t-test was applied to compare two groups. *p < 0.05 compared with WT mice. Serum amlyoid A causing bone loss populations via Wnt signaling, leading to a relative increase in osteoclast levels. Further studies are required to fully elucidate the effects of IL-17 on osteoblasts and osteoclasts.
In contrast with our data, several studies showed that SAA suppresses osteoclast differentiation (29,30). The inhibitory effects of SAA in osteoclastogenesis were discovered in vitro; however, our experimental data were collected in vivo. Also, several studies have shown that IL-17 induces osteoclastogenesis. Therefore, we speculate that the direct effects of SAA on osteoclastogenesis are weaker than any indirect effects of SAA interactions with IL-17 on osteoclast differentiation.
In this study, we revealed that SAA induced bone loss via an increase in IL-17-producing innate immune cells. These data suggested that IL-17-mediated inflammatory responses and bone loss may decrease when SAA levels are reduced. Therefore, SAA appears to be a valuable therapeutic target for some diseases characterized by inflammatory responses and bone loss. In addition, we also demonstrated the value of SAA1-overexpressing TG mice as a potential disease model for bone loss caused by chronic IL-17 inflammation.

Experimental procedures
Mice SAA1 TG mice were previously generated in a C57BL/6 background (31). Age-matched (7-9-week-old and 6-monthold) TG mice and littermates (WT mice) were used. Mice were raised and maintained under conventional conditions. They were given ad libitum access to food and water and housed under 12 h light/dark conditions. All animal experiments were conducted in accordance with the guidelines for animal experimentation and with permission from the Animal Use and Care Committee of Kyungpook National University.

Quantitative polymerase chain reaction (q-PCR) analysis
To separate bone marrow from bone, the femur at both ends was severed with sterile scissors. Bone marrow was rinsed using a 23-gage needle and a 10 cc syringe filled with ice-cold phosphate buffered saline. Bone tissue was harvested and ground using a MagNA Lyser (Roche Diagnostics). A TaKaRa Mini-BEST Universal RNA Extraction Kit (TaKaRa) was used to isolate RNA according to manufacturer's instructions. Total RNA was reverse transcribed at 60 C for 1 h using a Prime-Script first strand cDNA Synthesis Kit (Takara). Gapdh was used as an internal control for quantification. Messenger RNA (mRNA) levels were expressed as the fold change in expression when compared with untreated control mice. The primers are shown (Table 1). Expression levels of Il-17a, Il-23a, Acp5, Nfatc1, Ctsk, Runx2, Tnfrsf11b, Tnfsf11, Ctnnb1, Axin1, Alpl, Bglap, Ccn4, and Gapdh in samples were determined using SYBR green fluorescence (Takara), using the CFX96 TM Real Time System (Bio-Rad).

Western blot analysis
Protein levels of SAA1 and G-CSF in sera were measured by western blot. Goat anti-mouse SAA1 antibody (Abcam) diluted at a ratio of 1:1000, Goat anti-G-CSF primary antibodies (Santa Cruz Biotechnology) diluted at a ratio of 1:2000 were used, followed by HRP-conjugated anti-goat IgG secondary antibodies (Santa Cruz Biotechnology). Immunoblots were visualized using the ECL Detection System (GE Healthcare).

Micro-computed tomography (CT)
Intact tibia were dissected from mice, and muscle and soft tissues were removed. The bone was fixed in 10% formalin for 24-48 h and stored in 70% ethanol. A micro-CT scan was conducted using a Quantum GX2 micro-CT Imaging System (PerkinElmer, Waltham, MA, USA) using the following parameters: 90 kV, 88 μA, and with a voxel size of 50 μm.

Histology
Tibia tissues were decalcified in 0.25 M EDTA (pH 7.5) and 4% paraformaldehyde solution for 4 weeks. The solution was refreshed every week. Paraffin-embedded tibia were sectioned and stained using a TRAP staining kit (Takara) according to manufacturer's instructions. This was followed by methyl green counterstaining.

Fluorescence flow cytometry analysis
A single-cell suspension was obtained from lymph node, blood, and bone marrow cells. Blood and bone marrow cells were used after removing red blood cells (RBCs), using RBC lysis buffer (i.e., 155 mM ammonium chloride, 12 mM sodium bicarbonate, and 0.1 mM EDTA). For intracellular staining, single-cell suspensions were fixed with 4% paraformaldehyde solution and treated with permeabilization buffer (90% methanol). The following antibodies were used; fluorescein isothiocyanate (FITC)-conjugated anti-CD4 (RM4-5; Thermo Table 1 Primers used for quantitative polymerase chain reaction (qPCR)

Gene
Primer sequence

Statistical analysis
Data are expressed as the mean ± standard deviation from at least three independent experiments. Significant differences between groups were calculated using two-tailed Student's ttest. For all analyses, p < 0.05 was considered statistically significant.

Data availability
All data are contained within the article.
Supporting information-This article contains supporting information.