Cold-induced serum amyloid A3 amplifies cAMP–PKA signaling to drive adaptive thermogenesis in brown adipocytes

Abstract

Serum amyloid A3 (SAA3) is typically associated with inflammation, but its role in brown adipose tissue (BAT) thermogenesis remains elusive. Here, we identify SAA3 as a cold-inducible, locally acting amplifier of BAT thermogenesis. Using lentiviral modulation and BAT-specific knockout mice, we demonstrate that SAA3 is essential for cold tolerance and energy expenditure. Mechanistically, secreted SAA3 augments thermogenic signaling via a GPR3-dependent cAMP/PKA axis, driving robust UCP1 induction. Notably, this activation triggers transcriptional upregulation of endogenous Saa3 and Gpr3, establishing an autocrine/paracrine amplification loop that accelerates thermogenic output. Consequently, SAA3 loss blunts thermogenic responses in vivo and in vitro. These findings establish SAA3 as a BAT-derived regulator linking cold exposure to energy dissipation via a GPR3-mediated mechanism, highlighting its therapeutic potential for metabolic diseases.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s13105-026-01235-6.

Introduction

The global rise in obesity and its associated comorbidities, including type 2 diabetes and cardiovascular disease, presents a major public health challenge. Adipose tissue plays a central role in maintaining energy homeostasis and is a key contributor to the development of obesity and related metabolic disorders [1]. Among these, brown adipose tissue (BAT) has garnered considerable attention due to its capacity to dissipate energy by promoting triglyceride clearance, glucose uptake, and heat generation through thermogenesis [2, 3]. The thermogenic function of BAT is primarily driven by mitochondrial uncoupling protein 1 (UCP1), which uncouples oxidative phosphorylation to produce heat instead of ATP. UCP1-mediated adaptive thermogenesis not only contributes to body temperature regulation but also enhances overall energy expenditure [3–5]. The discovery of active BAT in adult humans has sparked intense interest in its potential as a therapeutic target for obesity and metabolic disease by promoting energy dissipation and limiting weight gain.

Among the G protein–coupled receptors implicated in BAT function, GPR3 (G protein-coupled receptor 3) is distinguished by its exceptionally high constitutive activity and ability to activate the cAMP/PKA pathway in a ligand-independent manner [6, 7]. This property positions GPR3 as a noncanonical thermogenic regulator that acts in parallel to β-adrenergic signaling. However, whether endogenous extracellular cues, such as adipokines, modulate GPR3 activity during adaptive thermogenesis remains unclear.

Serum amyloid A (SAA), a member of the acute-phase protein family, has emerged as a critical link between metabolic dysfunction and inflammation. Of the four known SAA subtypes, circulating SAA1/SAA2 levels rise markedly during acute inflammation (up to 1,000-fold). These levels are also elevated in chronic metabolic disorders such as obesity, where SAA functions as an inflammatory adipokine and serves as a marker for insulin resistance [8, 9]. SAA1 and SAA2 are 96% homologous and are predominantly expressed in the liver in response to inflammatory stimuli in both humans and mice. These subtypes are analogous to murine SAA1.1 and SAA2.1 [10, 11]. Intriguingly, human adipose tissue is a major source of SAA1/SAA2 under non-acute-phase conditions, suggesting a tissue-specific role in metabolic regulation [12–14]. Mouse SAA3 shares 69% amino acid identity with human SAA1 and is primarily expressed in extrahepatic tissues, including adipocytes and macrophages [10]. SAA3 expression is markedly elevated in the adipose tissue of obese mice and is implicated in macrophage recruitment, white adipose tissue (WAT) inflammation, and adipogenesis [15–19]. Circulating SAA3 in mice has only been observed under severe inflammatory conditions [20] and is absent in humans due to a pseudogene [21, 22]. SAA4, on the other hand, is constitutively expressed by most cell types and exhibits limited responsiveness to inflammatory cues [23].

Despite the absence of functional SAA3 in humans, murine SAA3 is considered the functional analog of human extrahepatic SAA1/2 due to high sequence homology and conserved adipose tissue expression [10, 12–14]. In humans, the overlapping expressions of SAA1 and SAA2 in both hepatic and extrahepatic tissues complicate efforts to dissect their tissue-specific roles. By contrast, the predominant extrahepatic expression of SAA3 in mice enables more precise differentiation between hepatic and extrahepatic SAA functions. Although SAA3 levels in adipose tissue positively correlate with obesity and diabetes, its physiological role in regulating adaptive thermogenesis remains poorly understood. Crucially, while previous studies focus on circulating SAA as an endocrine factor, we identify SAA3 as a locally acting extracellular amplifier of thermogenic signaling, contrasting it with the systemic roles of SAA1/SAA2 [17, 20, 24]. The distinct expression pattern and tissue specificity of murine SAA3 offer a unique opportunity to explore the functional role of extrahepatic, BAT-derived SAA in thermogenesis.

Here, we employ in vivo BAT-specific Saa3 knockdown/overexpression models, UCP1-Cre–driven brown adipocyte-specific Saa3 knockout (UCP1-SAA3^−/−) mice, and isolated brown adipocytes to define SAA3’s role in adaptive thermogenesis. We demonstrate that SAA3 is indispensable for maintaining BAT thermogenic activity and systemic energy homeostasis. Critically, we establish SAA3 as a locally acting extracellular amplifier that autonomously drives the expression of Ucp1 and Fgf21. Mechanistically, SAA3 enhances thermogenic signaling via GPR3-dependent activation of the cAMP/PKA axis, creating a self-amplifying loop that reinforces thermogenic gene expression. Our findings position SAA3 as a BAT-derived mediator of thermogenesis and reveal a GPR3-mediated amplification mechanism with therapeutic implications for metabolic disease.

Results

Cold-induced SAA3 acts as a local autocrine/paracrine signal in BAT

C57BL/6 mice were housed at thermoneutrality (TN) or exposed to cold (4 °C, CE) for 2, 6, or 11 days. Immunoblotting showed that UCP1 protein was significantly induced after 2 days of CE and remained elevated through day 11, whereas SAA3 protein exhibited a transient peak at 2–6 days and declined by day 11. Consistently, qPCR analysis revealed sustained induction of Ucp1 mRNA across 11 days, while Saa3 and Fgf21 mRNA peaked between days 2–6 (Fig. 1A and 1B). In cultured WT-1 brown adipocytes, norepinephrine (NE) and the β3-adrenergic agonist CL316,243 (CL) upregulated both SAA3 and UCP1 protein expression (Fig. 1C).

Fig. 1.

SAA3 expression in BAT is induced by cold and correlates with thermogenesis. (A and B) Protein A and mRNA B levels of SAA3, FGF21, and UCP1 in BAT of male C57BL/6NCrl mice housed at thermoneutrality (TN, 30 °C) or cold exposure (CE, 4 °C) for 2, 6, or 11 days. C Immunoblot analysis of SAA3 and UCP1 protein expressions in differentiated WT-1 brown adipocytes treated with norepinephrine (NE, 1 μM) or CL316,243 (CL, 10 μM) for 6 h. D Time-course protein expression of thermogenic markers in WT-1 cells in response to recombinant SAA3 (SAA3rec, 1 μg/mL). Data are presented as mean ± SEM; individual biological values are plotted in all bar graphs. For all panels, n = 3–4 biological replicates per group. Immunoblots are representative of 3 independent biological experiments (molecular weight markers shown). Statistical significance for A and B was determined by two-way ANOVA with Bonferroni post-hoc test; statistical significance for C and D was determined by one-way ANOVA; *p < 0.05, **p < 0.01, ***p < 0.001 vs. TN A, B or control group C, D

Moreover, recombinant SAA3 (SAA3rec) not only increased UCP1 and FGF21 expression in a time-dependent manner (Fig. 1D), but also significantly increased the expression of endogenous SAA3 protein within the cell lysate (Fig. 1D). Crucially, this protein accumulation was paralleled by a robust upregulation of Saa3 mRNA (Figure S1A), demonstrating that SAA3rec stimulates de novo SAA3 synthesis and establishing an autocrine/paracrine amplification mechanism. Notably, SAA3rec stimulation selectively induced Saa3 but did not alter combined Saa1/2 expression (Figure S1B), indicating isoform specificity. To investigate whether SAA3 functions as an endocrine factor or a local signal, we measured SAA3 concentrations in both circulating plasma and conditioned media. Circulating SAA3 levels in C57BL/6 mice plasma were below the detection limit of the assay in both TN and CE conditions (Figure S2A), indicating that the cold-induced upregulation of SAA3 in BAT does not lead to detectable systemic spillover. Similarly, in vitro, SAA3 protein concentrations in the conditioned media from differentiated WT-1 brown adipocytes treated with the ꞵ-adrenergic agonist CL316,243 remained extremely low (Figure S2B), despite the clear induction of intracellular SAA3 protein (Fig. 1C). Together, these findings indicate that SAA3 is induced in BAT by sympathetic activation and that SAA3 itself is sufficient to drive thermogenic gene expression, suggesting an autocrine/paracrine amplification loop mediated by secreted SAA3, thereby positioning it as both a target and mediator of adaptive thermogenesis.

Local modulation of SAA3 in BAT alters thermogenic capacity in vivo

To investigate the role of SAA3 in BAT thermogenic regulation, C57BL/6 mice were locally injected with a lentiviral control vector, full-length SAA3 cDNA (SAA3OE), or shRNA targeting SAA3 (shSAA3) directly into interscapular BAT. After 5 days of recovery, mice were exposed to either cold (4 °C, CE) or thermoneutral (30 °C, TN) conditions for 11 days (Fig. 2A). Mice with SAA3-overexpressing BAT (SAA3OE) exhibited significantly improved cold tolerance compared with controls (Fig. 2B), without differences in body weight gain, food intake, or fat depot weights (Figure S3A–S3C). Notably, at thermoneutrality—where endogenous thermogenic drive is minimal—SAA3OE BAT displayed robust SAA3 expression (Fig. 2C and 2D), which was sufficient to drive the upregulation of key thermogenic genes and proteins, including UCP1, FGF21, and PGC-1α (Fig. 2C–2G). This mimicry of cold signaling was accompanied by enhanced phosphorylation of PKA, ATF2, and p38 MAPK (Fig. 2H) and the upregulation of lipolysis-related proteins, such as pHSL, ATGL, and ABHD5 (Fig. 2I). Notably, because phosphorylated HSL (pHSL) reflects the enzyme's activated state rather than its total abundance, this finding indicates that SAA3 promotes not only lipolytic protein expression but also the functional activation of the lipolytic machinery. However, under cold exposure, where the endogenous machinery is already maximally stimulated, the induction of thermogenic genes in control mice was not further augmented in SAA3OE mice, suggesting a physiological ceiling effect. Supporting a cell-autonomous role in lipid mobilization, we further observed that treatment of differentiated brown adipocytes with recombinant SAA3 (SAA3rec) significantly upregulated the fatty acid transporter CD36 and lipolytic proteins in vitro (Figure S4).

Fig. 2.

SAA3 overexpression in BAT improves thermogenic capacity and cold tolerance. A Schematic and experimental timeline of local lentivirus injection (scrambled control or full-length Saa3 cDNA) into the BAT of male C57BL/6NCrl mice and subsequent housing at TN (30 °C) or CE (4 °C) for 11 days. B Core body temperature changes measured during the 11-day cold exposure period. C Protein levels of SAA3, UCP1, FGF21, and PGC-1α in BAT. D–G mRNA expression levels of Saa3 D, Ucp1 E, Fgf21 F, and Pgc1a G in BAT. H and I Phosphorylation levels of key signaling proteins H and expression of lipolysis-related proteins I in BAT. Data are presented as mean ± SEM; individual biological values (n = 4–8 mice per group) are plotted in all bar graphs. For mRNA analysis, three technical replicates were performed per sample. Immunoblots are representative of independent biological experiments; molecular weight markers (kDa) are indicated on the right of each blot. Statistical significance for the time-course data in (B) was determined by two-way repeated-measures ANOVA with Bonferroni post-hoc test. For (C–I), statistical significance was determined by two-way ANOVA with Bonferroni post-hoc test. ∗ p < 0.05, ∗ ∗ p < 0.01, ∗ ∗ ∗ p < 0.001 vs. indicated groups

Conversely, mice with BAT-specific SAA3 knockdown (shSAA3) displayed impaired adaptive thermogenesis. Upon acute cold challenge, these mice failed to maintain core body temperature, showing a significant drop after 2 h of CE (Fig. 3A). This BAT-specific knockdown did not affect systemic metabolic parameters, including body weight gain, food intake, or fat depot weights, under either CE or TN conditions (Figure S3D–S3F). Mechanistically, the suppression of cold-induced SAA3 in BAT (Fig. 3B and 3C) resulted in a marked reduction in thermogenic gene and protein expression (Fig. 3D–3I), along with blunted downstream phosphorylation of PKA, ATF2, and p38 MAPK (Fig. 3J). Furthermore, protein levels of the lipolytic enzymes ATGL and ABHD5 were significantly reduced in BAT from shSAA3 mice specifically under cold stress (Fig. 3K). Together, these findings demonstrate that SAA3 is sufficient to initiate thermogenic signaling at thermoneutrality and is required for the full activation of BAT-mediated adaptive thermogenesis during cold exposure.

Fig. 3.

SAA3 knockdown in BAT impairs cold tolerance and reduces thermogenic activation. A Core body temperature changes of mice locally injected with scrambled control or shSaa3 lentivirus in BAT during the 11-day experimental period at TN (30 °C) or CE (4 °C). B Protein levels of SAA3, UCP1, FGF21, and PGC-1α in BAT. C–F mRNA expression levels of Saa3 (C), Ucp1 (D), Fgf21 (E), and Pgc1a (F) in BAT. G and H Phosphorylation levels of signaling proteins G and expression of lipolysis-related proteins H in BAT. Data are presented as mean ± SEM; individual biological values (n = 3–6 mice per group) are plotted in all bar graphs. For mRNA analysis, three technical replicates were performed per sample. Immunoblots are representative of independent biological experiments; molecular weight markers (kDa) are indicated on the right of each blot. Statistical significance for the time-course data in A was determined by two-way repeated-measures ANOVA with Bonferroni post-hoc test. For B–H, statistical significance was determined by two-way ANOVA with Bonferroni post-hoc test. ∗ p < 0.05, ∗ ∗ p < 0.01 vs. indicated groups

Conditional deletion of SAA3 in brown adipocytes impairs cold-induced adaptive thermogenesis and reduces total energy expenditure

To further examine the physiological role of SAA3 in BAT, we generated brown adipocyte-specific knockout mice (UCP1-SAA3^−/−) by crossing Saa3^fl/fl mice with Ucp1-Cre mice (Fig. 4A). SAA3 protein was efficiently depleted in BAT, while expression in WAT, liver, lung, heart, and kidney remained unchanged (Figure S5A and S5B), confirming tissue specificity. At room temperature (~ 25 °C), UCP1-SAA3^−/− mice exhibited comparable body weight gain and fat distribution to their wild-type (WT) littermates (Figure S5C and S5D). In WT mice, SAA3, FGF21, and UCP1 protein expression in BAT increased within 2 days of an 11-day cold exposure, whereas no such induction occurred in inguinal WAT (iWAT) (Fig. 1A; Figure S6).

Fig. 4.

SAA3 deficiency in brown adipocytes impairs thermoregulation and energy expenditure. A Schematic overview of the brown adipocyte-specific SAA3 knockout (UCP1-SAA3^−/−) mouse model generated by crossing Saa3^fl/fl mice with Ucp1-Cre mice. B and C Core body temperature B and body weight C of WT littermates and UCP1-SAA3^−/− mice during a 5-h acute cold challenge at 4 °C (CE) or thermoneutrality (TN, 30 °C). D mRNA levels of thermogenesis-related and mitochondrial genes in BAT assessed by qPCR after 5-h cold exposure. E and F Protein levels of key thermogenic markers E and mitochondrial oxidative phosphorylation (OXPHOS) complexes F in BAT under TN or CE conditions. (Note: Panels E and F share the same HSP70 loading control as proteins were detected on the same membrane.). G–J Metabolic parameters including oxygen consumption (VO₂, G), carbon dioxide production (VCO₂, H), respiratory exchange ratio (RER, I), and total energy expenditure (EE, J) measured by indirect calorimetry. Data are presented as mean ± SEM; individual biological values (individual mice) are plotted in all bar graphs. For D–F, n = 3–4 independent biological experiments with 3 technical replicates per sample. For B, C, and G–J, n = 6 individual mice per group. Representative immunoblots in E, F are from 3 independent biological experiments; molecular weight markers are indicated on the right of each blot. Statistical significance for the time-course data in B was determined by two-way repeated-measures ANOVA with Bonferroni post-hoc test. For C–F, statistical significance was determined by two-way ANOVA with Bonferroni post-hoc test. Statistical significance for metabolic cage data G–J was determined by ANCOVA to account for variations in body weight. ∗ p < 0.05, ∗ ∗ p < 0.01 vs. indicated groups

Functionally, loss of SAA3 compromised acute cold tolerance. During a 5-h cold challenge, UCP1-SAA3^−/− mice showed a more rapid and severe decline in core body temperature compared with WT controls (Fig. 4B), without differences in body weight (Fig. 4C). At the molecular level, the cold-induced upregulation of Saa3, Ucp1, Fgf21, and Ppargc1a mRNA in BAT was significantly blunted in UCP1-SAA3^−/−mice (Fig. 4D). Moreover, expression of mitochondria-related genes, including Ppara, Tfam, and Atp5a1, was markedly reduced in BAT from knockout mice after cold exposure (Fig. 4D). Consistent with these transcriptional changes, protein expressions of UCP1 and FGF21, as well as the phosphorylation of p38 MAPK and ATF2, were significantly diminished in UCP1-SAA3^−/− BAT compared with WT (Fig. 4E). Since cold exposure typically enhances mitochondrial oxidative phosphorylation (OXPHOS) capacity, we examined mitochondrial protein complexes. While complexes IV and V were robustly upregulated in WT mice upon cold exposure, this adaptive increase was significantly attenuated in UCP1-SAA3^−/− mice. Complex III showed a similar trend that did not reach statistical significance (Fig. 4F), indicating impaired mitochondrial respiratory capacity. To assess systemic metabolic consequences, mice were placed in metabolic cages (Promethion) following acclimation at thermoneutrality. Cold exposure increased whole-body oxygen consumption (VO₂) and carbon dioxide production (VCO₂) in both genotypes, but these responses were significantly attenuated in UCP1-SAA3^−/− mice (Fig. 4G and 4H). The respiratory exchange ratio (RER) did not differ between groups (Fig. 4I), indicating no shift in fuel substrate preference. Notably, the cold-induced rise in total energy expenditure was markedly blunted in UCP1-SAA3^−/− mice compared with WT controls (Fig. 4J). Together, these results demonstrate that loss of SAA3 in brown adipocytes compromises BAT mitochondrial function, impairs cold-induced adaptive thermogenesis, and reduces whole-body energy expenditure.

SAA3 as a modulator of UCP1-dependent thermogenesis in brown adipocytes

To assess whether SAA3 is required for β3-adrenergic–mediated thermogenic activation, we mimicked cold-induced stimulation using the β3-adrenergic agonist CL316,243. In differentiated primary brown adipocytes, CL316,243 markedly increased SAA3, UCP1, PGC-1α, and FGF21 protein expression, together with the phosphorylation of ATF2 and p38 in WT cells, but failed to induce these responses in UCP1-SAA3^−/− brown adipocytes (Fig. 5A). Consistently, UCP1-SAA3^−/− brown adipocytes exhibited blunted cAMP accumulation and reduced phosphorylation of PKA substrates after CL316,243 stimulation compared with WT controls (Fig. 5B and 5C), indicating impaired β3-adrenergic receptor signaling efficiency. To examine functional changes in thermogenic capacity, we used the small-molecule thermosensitive fluorescent dye ERthermAC in live cells [25]. Functionally, UCP1-SAA3^−/− brown adipocytes displayed reduced thermogenic capacity, as evidenced by higher ERthermAC fluorescence (lower temperature) after CL316,243 stimulation, a defect that was rescued by recombinant SAA3 (SAA3rec) treatment (Fig. 5D). Likewise, UCP1-SAA3^−/− brown adipocytes showed reduced basal, maximal, and spare respiratory capacity, all of which were restored by SAA3rec (Fig. 5E).

Fig. 5.

SAA3 drives thermogenesis via a GPR3-dependent cAMP–PKA signaling axis. A Western blot analysis and quantification of thermogenesis-related proteins in primary brown adipocytes from WT and UCP1-SAA3^−/− mice treated with or without CL316,243 (CL, 10 μM) for 6 h. B and C Intracellular cAMP levels B and phosphorylation of PKA substrates C in primary brown adipocytes following CL treatment. D Representative ERthermAC fluorescence images and quantification of thermogenic capacity in primary brown adipocytes treated with CL or recombinant SAA3 (SAA3rec). Scale bar, 50 μm. E Oxygen consumption rate (OCR) in primary brown adipocytes with sequential additions of SAA3rec, oligomycin, FCCP, and rotenone/antimycin A (RA). F Effect of Saa3 siRNA on CL- or SAA3rec-induced protein expression in WT-1 cells. G Protein expression in WT-1 cells treated with the ꞵ-blocker Bupranolol (10 μM) with or without CL or SAA3rec. H and I Intracellular cAMP levels H and PKA substrate phosphorylation I in WT-1 cells treated with CL or SAA3rec. J Effect of PKA inhibitor H89 (5 μM) on SAA3rec-mediated signaling in WT-1 cells. K–N mRNA levels of Gpr3 (K), cAMP levels L, signaling proteins M, and thermogenic gene expression N in WT-1 cells transfected with control or Gpr3 siRNA followed by SAA3rec treatment. Data are presented as mean ± SEM; individual biological values (independent cell preparations) are plotted in all bar graphs. For all panels, n = 3–4 independent biological experiments (independent cell preparations) with 3 technical replicates per sample. Representative immunoblots and fluorescence images are from 3 independent biological experiments; immunoblots of p-PKA substrates C are shown without densitometric quantification or statistical comparison, given the qualitative nature of this readout; molecular weight markers are indicated on the right of each blot. Statistical significance for A G and K–N was determined by two-way ANOVA with Bonferroni post-hoc test. For B, D–F, H, I, and J, statistical significance was determined by one-way ANOVA. ∗ p < 0.05, ∗ ∗ p < 0.01 vs. control group; #p < 0.05 vs. SAA3rec group

In the WT-1 brown adipocyte cell line, both CL316,243 and SAA3rec increased UCP1, PGC-1α, and FGF21 expression, whereas Saa3 siRNA abrogated these effects (Fig. 5F). Bupranolol partially attenuated CL316,243-induced thermogenic gene expression, particularly UCP1 and PGC-1α, but SAA3rec-induced UCP1 expression was unaffected by Bupranolol co-treatment (Fig. 5G). In addition, SAA3rec alone significantly elevated intracellular cAMP and the phosphorylation of PKA substrates to levels comparable to those of CL316,243, indicating that SAA3 can directly activate the cAMP–PKA pathway independently of β3-adrenergic receptor stimulation (Fig. 5H and 5I). To further determine whether SAA3 activates thermogenic signaling through PKA, we pretreated WT-1 brown adipocytes with the PKA inhibitor H89 prior to stimulation. Both CL316,243 and SAA3rec markedly increased the expression of thermogenic proteins, including FGF21, and enhanced the phosphorylation of p38, ATF2, and PKA substrates. However, these effects were substantially attenuated by H89 pretreatment, confirming that SAA3 signals through a PKA-dependent mechanism (Fig. 5J).

Given the observed increase in cAMP and sensitivity to PKA inhibition, we hypothesized that SAA3 might act via a Gs-coupled GPCR. Indeed, SAA3rec significantly increased Gpr3 mRNA expression in WT-1 brown adipocytes (Fig. 5K).

To further investigate the functional requirement of GPR3 in SAA3-mediated signaling, we performed genetic loss-of-function experiments using siRNA targeting Gpr3 in brown adipocytes. First, we examined the second messenger response. While SAA3rec treatment robustly elevated intracellular cAMP levels in WT-1 brown adipocytes, this response was significantly attenuated by Gpr3 knockdown (Fig. 5L). Consistently, the SAA3-driven downstream signaling was also markedly suppressed; Gpr3 silencing prevented the induction of thermogenic proteins (UCP1) and the phosphorylation of PKA, p38 MAPK, and ATF2 (Fig. 5M). Furthermore, at the transcriptional level, while SAA3rec robustly induced the expression of Ucp1, Fgf21, and Pgc1a in control cells, this induction was substantially reduced in Gpr3-deficient cells (Fig. 5N). These genetic data demonstrate that GPR3 is a critical mediator of SAA3-driven thermogenic signaling. However, these loss-of-function experiments establish GPR3 as a functionally required downstream component of the pathway rather than direct evidence of a physical SAA3–GPR3 interaction, which remains to be demonstrated.

To rule out the possibility that the blunted thermogenic response in UCP1-SAA3^−/− cells resulted from defective adipogenesis, we examined differentiation markers. Contrary to previous reports linking Saa3 silencing to impaired PPARγ expression [22, 26], protein levels of adipogenic markers (aP2, PPARγ, and C/EBPα) did not differ between UCP1-SAA3^−/− and WT brown adipocytes, irrespective of CL316,243 treatment (Figure S7A). Cell morphology was also comparable between genotypes (Figure S7B), indicating that SAA3 does not influence brown adipocyte differentiation.

In summary, our in vitro studies establish SAA3 as a potent activator of thermogenesis in brown adipocytes. We demonstrate that SAA3 is required for the full thermogenic response to ꞵ-adrenergic stimulation and, more importantly, can independently initiate the cAMP-PKA signaling cascade via a GPR3-dependent mechanism—even in the absence of β-adrenergic agonists.

Discussion

Understanding the regulatory mechanisms of BAT-mediated adaptive thermogenesis is essential for identifying potential targets to combat obesity [27, 28]. In this study, we identify SAA3 as a previously unrecognized, locally derived regulator of cold-induced adaptive thermogenesis in BAT. We demonstrate that SAA3 is rapidly and transiently induced by cold exposure and β3-adrenergic stimulation in vivo and in vitro, and that recombinant SAA3 alone is sufficient to enhance UCP1 and FGF21 expression. Functional studies using local lentiviral manipulation and brown adipocyte–specific knockout mice provide direct evidence for a causal role of SAA3 in maintaining cold tolerance, mitochondrial respiratory capacity, and systemic energy expenditure. Mechanistically, SAA3 promotes cAMP–PKA signaling and enhances thermogenic gene expression through a GPR3-dependent pathway, distinct from canonical sympathetic activation. This positions SAA3 as a critical autocrine/paracrine regulator that reinforces BAT thermogenic output by boosting a noncanonical, GPR3-dependent signaling route. Together, these results establish SAA3 as an integral component of the local regulatory network that amplifies thermogenic activation in brown adipocytes.

Previous studies have reported divergent results regarding the role of SAA3 in metabolic regulation, likely reflecting differences in disease models and the absence of tissue-specific knockout approaches. For example, in db/db mice, Saa3 is predominantly expressed in white adipocytes but not in the liver [13]. Loss of extrahepatic Saa3 in obese mice reduced WAT inflammation, improved systemic lipoprotein profiles, and conferred resistance to HFD-induced obesity, particularly in females [20]. In contrast, global deletion of Saa3 rendered mice more susceptible to HFD-induced obesity, characterized by increased adiposity [29]. Furthermore, triple knockout of Saa1, Saa2, and Saa3 had no effect on body weight or adiposity under HFD feeding but impaired glucose and insulin tolerance [30]. These inconsistent findings highlight the complexity of SAA3 function and suggest that its metabolic role is highly context- and tissue-dependent. By employing a brown adipocyte–specific Saa3 knockout model, our study clarifies this discrepancy, demonstrating that BAT-derived SAA3 functions distinct from its systemic counterparts. Unlike the chronic, maladaptive role often attributed to SAA3 in obesity-associated inflammation, we reveal a physiological role for SAA3 in regulating acute, cold-induced adaptive thermogenesis. Consistent with a protective, tissue-intrinsic role, we have separately shown, using the same brown adipocyte-specific knockout model, that BAT-derived SAA3 also defends against short-term HFD-induced weight gain by promoting diet-induced thermogenesis through an ATGL/CPT1A-dependent fatty acid oxidation pathway, [31] indicating that SAA3 supports adaptive energy dissipation across both cold- and diet-induced thermogenic stimuli.

To address the physiological distinction between shivering and non-shivering thermogenesis (NST), we acknowledge that the acute cold tolerance test primarily assesses the body's total thermogenic capacity, incorporating contributions from shivering muscle [32]. However, our focus on the acute 2-h phase was a deliberate strategy driven by our time-course analyses, which revealed that the maximal phenotypic divergence in cold tolerance between groups occurred sharply at this time point. This window allows us to capture the immediate impact of SAA3 loss on thermoregulation before confounding factors from prolonged chronic adaptation set in. We propose that the rapid hypothermia observed in SAA3-deficient models—despite the presence of functional muscle machinery—reflects a specific failure in the BAT-mediated component required to supplement shivering. This interpretation is robustly substantiated by converging molecular evidence: the significantly blunted whole-body VO₂ consumption and the failure to induce BAT-specific thermogenic markers (e.g., UCP1) in the SAA3-deficient mice. Thus, SAA3-mediated BAT function is confirmed as an indispensable early contributor to acute cold defense. Our findings indicate that the cold-induced rise in SAA3 in brown adipocytes is largely driven by sympathetic nervous system (SNS) activation. Importantly, recombinant SAA3 was able to induce UCP1 and PGC-1α expression even in the presence of the β-blocker bupranolol, suggesting that while the SNS serves as an upstream initiator, SAA3 itself functions as a critical endogenous amplifier of thermogenesis. Several non-sympathetic activators of BAT have also been identified, including bone morphogenetic proteins (BMPs), [33, 34] retinoids, [35] atrial natriuretic peptide (ANP) from the heart, [36] and irisin from skeletal muscle [37, 38]. These endocrine or paracrine signals can act directly on BAT or modulate sympathetic activity. However, unlike these factors, which often originate from distal tissues, SAA3 is induced locally within BAT, creating a rapid autocrine/paracrine amplification mechanism. At the biochemical level, SAA3, like other members of the SAA family, contains an N-terminal signal peptide [10]. Consistent with this structural feature, we detected SAA3 protein in the conditioned media of β3-adrenergic agonist–stimulated brown adipocytes (Figure S2B), albeit at concentrations below the assay's lower limit of quantification. This low extracellular abundance, together with the absence of detectable systemic SAA3 elevation during cold exposure (Figure S2A), supports a model in which SAA3 functions as a locally confined, short-range autocrine/paracrine signal rather than a bulk-secreted endocrine factor. Nevertheless, while cold exposure remains the most robust driver of BAT thermogenesis, it is often accompanied by discomfort and increased risk of respiratory infections [39]. Similarly, β-adrenergic agonists have shown limited efficacy in clinical settings and are associated with adverse cardiovascular effects [40]. In this context, our findings suggest that targeting brown adipocyte-derived SAA3 signaling may represent a promising strategy to potentiate BAT activity, bypassing the drawbacks associated with systemic sympathetic activation or environmental cold stress.

Mechanistically, our data demonstrate that SAA3 enhances the thermogenic program by activating the cAMP–PKA signaling axis, leading to phosphorylation of p38 MAPK and ATF2, which drives the transcriptional induction of Ucp1 and Fgf21. These effects are PKA-dependent (abrogated by H89) and, importantly, occur independently of ꞵ-adrenergic receptor activity. Crucially, we identify GPR3 as an essential mediator for this signal. Since SAA3rec stimulation also transcriptionally upregulates both Gpr3 and endogenous Saa3, this establishes an amplification loop. In this model, SAA3 acts as a rapid signal accelerator to maximize thermogenic output via GPR3, a potent cold-inducible constitutive GPCR in BAT [6, 41]. While pharmacological tools for GPR3 remain limited, our genetic loss-of-function studies using siRNA conclusively demonstrated that GPR3 is indispensable for SAA3-mediated cAMP accumulation, PKA/ATF2 phosphorylation, and subsequent UCP1 induction.

The activated (phosphorylated) lipolytic machinery — pHSL, ATGL, and ABHD5 — was consistently regulated by SAA3 gain- and loss-of-function across four independent systems, together with SAA3-dependent, SAA3rec-rescuable mitochondrial respiratory capacity (Fig. 5E), providing functional evidence that SAA3 promotes lipid mobilization and oxidative utilization. SAA3 also upregulated CD36 (Figure S4), consistent with enhanced fatty acid uptake, though this was not directly assayed. Notably, ABHD5 was reduced in SAA3OE BAT specifically under cold exposure relative to thermoneutrality (Fig. 2I), consistent with the physiological ceiling effect described above: since cold exposure already maximally activates the endogenous lipolytic cascade, the additional lipolytic drive from SAA3 overexpression may trigger a compensatory downregulation of ABHD5 once substrate becomes limiting. This remains a plausible interpretation warranting future validation.

While the precise molecular coupling between SAA3 and GPR3 remains to be fully elucidated, we propose a "metabolic relay" hypothesis to explain the upstream activation. We observed that SAA3rec treatment significantly upregulates the fatty acid transporter CD36 and lipolytic markers in brown adipocytes (Figure S4). In light of recent studies showing that lipolysis-derived fatty acids can serve as endogenous ligands to modulate GPR3 constitutive activity [6, 42]. We speculate that SAA3 functions as a metabolic trigger. By promoting local lipolysis and lipid uptake, SAA3 may generate a lipid-rich microenvironment that potentiates GPR3 signaling, which subsequently feeds into the cAMP/PKA axis to drive thermogenesis. We present this as a plausible upstream mechanism that warrants future structural investigation. Nevertheless, further studies are needed to clarify whether SAA3 also exerts GPR3-independent effects through other previously identified SAA receptors, such as TLR2, TLR4, or SR-BI, in brown adipocytes [31].

Limitations of the study

Several limitations should be considered when interpreting these findings. First, we acknowledge the limitations associated with the Ucp1-Cre driver line (JAX stock #024670), as ectopic expression or transgene integration effects have been reported [43]. While we cannot fully exclude background effects, our key findings were consistently recapitulated using acute lentiviral-mediated knockdown and overexpression approaches in C57BL/6 mice, as well as by recombinant protein rescue experiments. These independent validation strategies strongly suggest that the thermogenic defects are specifically attributable to SAA3 deficiency rather than transgenic artifacts. Second, our study was conducted exclusively in murine models. Although SAA3 is a pseudogene in humans, murine SAA3 is considered the functional analog of human extrahepatic SAA1/2 due to high sequence homology and conserved adipose tissue expression. Therefore, our findings are likely to provide relevant insights into the function of human adipose-derived SAA. Third, although our data support a locally acting, paracrine mode of SAA3 action based on circulating plasma and conditioned-media measurements (Figures S2A and S2B), we did not directly quantify SAA3 concentrations in BAT interstitial fluid under thermoneutral versus cold-exposed conditions. Direct interstitial sampling would provide the most rigorous confirmation of the local concentration gradient we currently infer and represents an important next step for future studies. Finally, while we identified GPR3 as a downstream effector, further studies are needed to elucidate the precise structural interactions among SAA3, lipid mediators, and the GPR3 receptor complex.

Despite these limitations, our findings establish SAA3 as a brown adipocyte–derived regulator of thermogenesis, linking cold-induced sympathetic activation to enhanced cAMP–PKA signaling and mitochondrial function. By revealing an autocrine/paracrine pathway that amplifies thermogenic responses, this work not only advances the fundamental understanding of BAT biology but also opens alternative avenues for targeting non-adrenergic mechanisms to harness BAT in the treatment of obesity and metabolic disease.

Methods

Animals

Male C57BL/6NCrl mice, aged between 5–6 weeks, were procured from BioLASCO Taiwan Co., Ltd, Taiwan. We obtained transgenic mice, which express Cre recombinase under the control of the Ucp1 promoter (B6.FVB-Tg (UCP1-Cre)1Evdr/J; stock #024670), from the Jackson Laboratories. These UCP1-Cre mice are designed to specifically express Cre recombinase in UCP1-positive cells, resulting in the selective inactivation of gene expression in brown and specific beige adipocytes [44]. We bred double-floxed Saa3 mice, generated from the Transgenic Mouse Model Core Facility of the National Laboratory Animal Center, with UCP1-Cre mice. This breeding resulted in the production of brown adipocyte-specific SAA3 knockout mice (UCP1-SAA3^−/−). Unless otherwise noted, mice were housed in a pathogen-free animal facility at 22 °C with a 12 h light/dark cycle from 6 AM to 6 PM and fed ad libitum a standard chow diet (D12450B, Research Diets Inc.). The mice were weighed weekly. The food intake was calculated by subtracting the remaining food from the added food in both the food hopper and the cage every 7 days. The weights of the BAT, inguinal WAT (iWAT), and epididymal WAT (eWAT) depots were recorded upon tissue harvest. The Institutional Animal Care and Use Committee (IACUC) of the National Defense Medical University, Taipei, Taiwan, approved all animal experimental procedures, and the studies were conducted in compliance with the approved protocol IACUC-21–286 at the same institute. During all experimental procedures, the number of animals and their suffering due to treatments were minimized. Only male mice were used in this study to minimize the potential confounding effects of hormonal fluctuations on metabolic and thermogenic parameters. The potential influence of sex on SAA3-mediated thermogenesis warrants further investigation.

Cell lines

The immortalized mouse brown preadipocyte WT-1 cell line was used. The WT-1 cell line was obtained from Sigma-Aldrich and has been authenticated by the manufacturer. All cells were regularly tested and confirmed to be negative for mycoplasma contamination. For differentiation, confluent cultures (Day 0) were treated with induction media containing high glucose DMEM (4.5 g/L), 5 μM dexamethasone, 0.5 μg/ml insulin, 0.5 mM IBMX, 1 nM T3​, and 10% FBS. The media was refreshed on Day 2. From Day 3 until harvest (Day 6 or 7), cells were maintained in media with insulin, T3​, and 10% FBS.

Primary cell cultures

Primary brown preadipocytes were isolated from the stromal vascular fraction of BAT from 2-week-old male WT and UCP1 − SAA3^−/− mice via collagenase digestion. Cells were filtered through 70-μm and 40-μm strainers and centrifuged at 500 × g for 5 min. They were cultured in high-glucose DMEM (4.5 g/L) with 10% FBS. Upon reaching confluency, cells were differentiated for 48 h using an adipogenic cocktail (0.5 mM IBMX, 5 μM dexamethasone, 0.125 mM indomethacin, 0.5 μg/ml insulin, and 1 nM T3). Post-differentiation, cells were maintained in media with 0.5 μg/ml insulin and 1 nM T3 and harvested on Days 6–8. Early passage primary brown preadipocytes (p ≤ 3) were used to avoid senescence.

Cell treatments and RNA interference

Differentiated primary adipocytes were treated for 6 h with 10 μM CL316,243 or 1 μg/mL recombinant mouse SAA3 (Cusabio) before harvest. For signaling pathway analysis, WT-1 brown adipocytes were pretreated with the PKA inhibitor H89 (5 μM) for 1 h prior to stimulation. In separate experiments, differentiated WT-1 cells were treated for 6 h with 10 μM CL316,243 or 10 μM Bupranolol. For RNA interference, On-Target plus SMARTpool siRNA targeting mouse Saa3 (catalog # L-055794–01–0050) and siRNA targeting mouse Gpr3 (catalog # M-045861–01–0005) were purchased from Dharmacon. WT-1 and primary brown adipocytes were treated with siRNA using Lipofectamine RNAiMAX (Invitrogen) according to the manufacturer’s instructions.

Cold exposure and temperature measurements

For cold exposure (CE) experiments, C57BL/6NCrl mice, aged 12–16 weeks and weighing 26–28 g, were used. In the time-course study, mice were acclimated to thermoneutrality (30 °C) for 1 week before being exposed to 4 °C for 2, 6, or 11 days. For thermogenesis gain and loss-of-function experiments, mice were acclimated to thermoneutrality (30 °C) for 11 days, while control groups were maintained at 30 °C. In a separate cold challenge experiment, UCP1-SAA3^−/− and their WT littermate controls were acclimated to 30 °C for 7 days before acute cold exposure. Mice were then placed in a 4 °C environment for 5 h. Core body temperature was measured by taking rectal temperature readings from conscious mice in their basal state and upon cold exposure using a Thermalert model TH-5 temperature monitor (Physitemp, Clifton, NJ).

Body composition and energy expenditure

Body composition, including fat and lean mass, was determined using magnetic resonance technology (EchoMRI-100–700; Echo Medical Systems, Houston, TX, USA). Energy expenditure (EE) and other metabolic parameters were measured using a rodent Promethion metabolic cage system (Sable Systems International). Mice were individually housed in metabolic cages and acclimated for 2–3 days prior to measurements. All metabolic studies were conducted at controlled temperatures of 30 °C or 4 °C under a 12-h light/dark cycle. Real-time metabolic measurements, including EE (kcal/hr), oxygen consumption (VO₂), carbon dioxide expiration (VCO₂), and respiratory exchange ratio (RER), were continuously recorded by the Sable System data acquisition software (IM-3 v.20.0.3).

Computed tomography (CT) scan

Computed tomography was utilized to measure adipose tissue quantitatively. A three-dimensional dataset of tissue attenuation values was created using X-rays and reconstructed into multiplanar images. Adipose tissue was quantified using the Hounsfield scale.

Intra-BAT lentivirus injection

Lentiviral vectors were constructed by cloning the full-length murine SAA3 cDNA or shRNA targeting SAA3 into the pLnt1 vector. A control vector with no transgene was also prepared. Lentivirus production was performed as previously described [45]. For in situ injections, [46] mice were anesthetized and placed on a workbench. A 1 cm incision was made on their backs after shaving and disinfecting the area. The fascia and surrounding white adipose tissue were carefully separated to expose the BAT. SAA3 lentivirus (1 × 10¹⁰ infectious units/mouse) or vehicle was injected at 3–4 points on both sides of the BAT using a 31G pen needle (BD) in a total volume of 50 μl. The incision was sutured and disinfected upon completion.

Thermogenic capacity measurement

The thermogenic capacity of differentiated brown adipocytes was measured using the ERthermAC dye (Emdmillipore, SCT057) [25]. Cells were incubated in DMEM with 250 nM ERthermAC for 30 min at 37 °C. After replacing the medium with fresh phenol red-free DMEM, fluorescence was measured every 5 min for 120 min at 37 °C using a GloMax Discover Multimode Detection System (Promega) after adding 10 μM CL316,243. Increased fluorescence intensity indicates a lower cellular temperature.

RNA and protein analyses

Total RNA was extracted from mouse adipose tissue or isolated primary adipocytes using TRIzol reagent (Invitrogen). Two micrograms of total RNA were reverse transcribed with the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Waltham, MA, USA). Quantitative PCR (qPCR) was performed using SYBR Green (Life Technologies, Carlsbad, CA, USA) with primer sequences listed in Suppl. Table S1. 18S rRNA was used as the reference gene. For protein analysis, tissues or cells were homogenized in RIPA buffer with a protease and phosphatase inhibitor mixture (Roche Diagnostics). Proteins were separated by SDS-PAGE and transferred to polyvinylidene fluoride membranes. Membranes were incubated overnight at 4 °C with the indicated antibodies. After washing, membranes were incubated with an HRP-conjugated secondary antibody. Where multiple targets were probed on the same membrane, strips were sequentially stripped and reprobed; HSP70 or α-Tubulin served as the loading control for each individual panel. Band intensities were quantified using ImageJ software. Primary and secondary antibody information, including source, catalog number, and RRID, is provided in Suppl. Table S2.

Adipocyte oxygen consumption rate (OCR) Measurement

Cellular oxygen consumption was measured using an XF24 analyzer (Seahorse Bioscience). Isolated brown preadipocytes were plated and differentiated into mature brown adipocytes in an XF24-well microplate. OCR was measured at 37 °C and normalized on protein content, as measured by the BCA assay in each well. The measurement included a baseline phase followed by sequential injections of Oligomycin (1.5 μM), FCCP (4 μM), and Rotenone/antimycin A (0.5 μM).

Sources and catalog numbers for all chemicals, recombinant proteins, commercial assay kits, oligonucleotides, cell lines, mouse strains, and software used in this study are provided in Suppl. Table S3.

Statistical analysis

Data are presented as the mean ± SEM, as described in the figure legends. An unpaired two-tailed Student's t-test was used for two-group comparisons. For comparisons among more than two groups, one-way ANOVA or two-way ANOVA (including two-way repeated-measures ANOVA for time-course data involving repeated measurements in the same animals) with Bonferroni multiple comparisons test was applied, as indicated in the figure legends. Analysis of covariance (ANCOVA) was used for the analysis of metabolic cage data (Fig. 4G–4J) to account for variations in body weight. A minimum of three independent experiments was performed for all analyses. Statistical significance was defined as *p < 0.05, **p < 0.01, or ***p < 0.001, with analyses performed using GraphPad Prism 6.0 (GraphPad Software, San Diego, CA, USA).

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ATF2

Activating transcription factor 2

BAT

Brown adipose tissue

EE

Energy expenditure

Epi

Epididymal

NE

Norepinephrine

FGF21

Fibroblast growth factor 21

GPR3

G protein-coupled receptor 3

PKA

Protein kinase A

RER

Respiratory exchange ratio

SAA

Serum Amyloid A

Sub

Subcutaneous

TN

Thermoneutral

UCP1

Uncoupling protein-1

WAT

White adipose tissue

Author contributions

P.-C.C.: Data curation, Investigation, Methodology, Validation, Conceptualization, Writing- original draft. H.-C.C.: Formal analysis, Investigation, Visualization. P.-R.L.: Formal analysis, Investigation, Visualization. Y.-F.T.: Funding acquisition, Supervision. P.-S.H.: Writing-review & editing, Supervision, Funding acquisition, Project administration, Conceptualization.

Data availability

Data and code availability The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work, the authors used Gemini in order to improve English phrasing and ensure compliance with journal formatting requirements. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the publication's content.

Competing interests

The authors declare no competing interests.