Network Meta-Analysis: Effect of Cold Stress on the Gene Expression of Swine Adipocytes ATGL, CIDEA, UCP2, and UCP3

Cold stress significantly affects gene expression in adipocytes; studying this phenomenon can help reveal the pathogeneses of conditions such as obesity and insulin resistance. Adipocyte triglyceride lipase (ATGL); cell death-inducing deoxyribonucleic acid (DNA) fragmentation factor subunit alpha (DFFA)-like effector (CIDEA); and uncoupling protein genes UCP1, UCP2, and UCP3 are the most studied genes in pig adipose tissues under cold stress. However, contradictory results have been observed in gene expression changes to UCP3 and UCP2 when adipose tissues under cold stress were examined. Therefore, we conducted a meta-analysis of 32 publications in total on the effect of cold stress on the expression of ATGL, CIDEA, UCP2, and UCP3. Our results showed that cold stress affected the expression of swine adipocyte genes; specifically, it was positively correlated with the expression of UCP3 in swine adipocytes. Conversely, expression of ATGL was negatively affected under cold stress conditions. In addition, the loss of functional UCP1 in pigs likely triggered a compensatory increase in UCP3 activity. We also simulated the docking results of UCP2 and UCP3. Our results showed that UCP2 could strongly bind to adenosine triphosphate (ATP), meaning that UCP3 played a more significant role in pig adipocytes.


Introduction
Human metabolic syndrome affects individuals with a high body weight (BW) [1].The answer to the question of why individuals with a high BW are affected by human metabolic syndrome lies in the study of fat deposition mechanisms.Humans have two main types of fat tissues: white adipose tissue (WAT) and brown adipose tissue (BAT).WAT is primarily involved in energy storage, whereas BAT plays a role in heat production and the regulation of energy metabolism.BAT contains a large number of mitochondria [2].Research on fat tissues has recently increased, and the conversion of WAT to BAT in particular has attracted widespread attention [3].BAT plays an essential role in maintaining whole-body energy homeostasis by regulating insulin sensitivity and glucose metabolism [2].We are interested in investigating how pigs with high-fat ratios maintain whole-body energy homeostasis.Pigs are important animal models for human diseases [4].
Cold stress significantly affects gene expression in adipocytes [5].Expression of cold stress-related genes regulates the growth, differentiation, and function of adipocytes, thereby affecting the balance of energy metabolism in the body.In addition, WAT browning has demonstrated therapeutic potential for the treatment of human diseases [5].Further study of the influence of cold stress on adipocyte gene expression can help reveal the pathogeneses of obesity, insulin resistance, and other ailments [6], which will provide new insights into the treatment and prevention of these conditions [7].Short-term cold stimulation leads to cold stress, whereas long-term cold stimulation results in adaptation to the cold.However, the mechanism by which WAT is converted to BAT and the duration of this conversion remain unclear.In this study, we did not distinguish between cold stress and adaptation to cold; this article uses the term "cold stress" consistently for both.
Pigs exposed to cold stress experience significant changes in adipose tissue gene expression.The most studied genes in adipose tissue gene expression under cold stress include adipocyte triglyceride lipase (ATGL); cell death-inducing deoxyribonucleic acid (DNA) fragmentation factor subunit alpha (DFFA)-like effector (CIDEA); and uncoupling protein genes UCP1, UCP2, and UCP3.Pigs have a defect in UCP1 that prevents it from synthesizing protein [8]; however, that is not the focus of this study.UCP2 and UCP3 gene-encoded proteins are mainly located in the mitochondria and participate in energy metabolism [9].They regulate oxidative phosphorylation (OXPHOS) and uncoupling, thereby affecting energy balance within cells.ATGL encodes an enzyme responsible for breaking down triglycerides in adipocytes, producing free fatty acids (FFAs) and glycerol [10].Lipid metabolic products can also affect the expression of other genes.Adenosine triphosphate (ATP) regulates ATGL activity through the protein kinase A (PKA) pathway and inhibits or releases FFAs [11,12]; it also inhibits UCP2 and UCP3 activity in the mitochondria [13].CIDEA encodes an enzyme involved in fatty acid ester degradation, which further regulates fat metabolism [14].CIDEA helps regulate the fusion of lipid droplets, potentially regulating the process of adipocyte volume reduction [15].
It is reported that under cold stress, adipose-tissue UCP2 decreases and UCP3 increases [16].However, Zhou et al. found that under cold stress, adipose-tissue UCP3 decreased and UCP2 increased [17].These conflicting results led us to investigate, through a meta-analysis, the effect of cold stress on the expression of the swine adipocyte genes ATGL, CIDEA, UCP2, and UCP3.

Database Search Strategy and Study Inclusion
The articles included in this study were obligated to prioritize animal welfare by minimizing the uncomfortable reactions cold stimulation can cause in pigs.Two independent authors searched separately for the following keywords: "(pig OR, porcine OR swine OR hog OR boar OR sow OR piglet) AND (ATGL OR UCP2 OR UCP3 OR CIDEA) AND cold".We searched four databases-PubMed, ProQuest, ScienceDirect, and Web of Science-for literature published up to 1 October 2023.Table 1 presents the criteria for literature inclusion and exclusion, while Figure 1A illustrates the search and literature screening steps.).SAT: subcutaneous adipose tissue.VAT: visceral adipose tissue.The values of the invalid lines in the graph are 0, so when the 95% confidence interval (CI) was entirely greater than 0, cold stress was considered positively correlated with swine adipocyte gene expression.When the 95% CI was entirely less than 0, cold stress was considered negatively correlated with swine adipocyte gene expression.Green squares represent the weight of the study, and black diamonds represent the 95% CI. (C) Network diagrams comparing different genes.The thicker the line, the greater the number of studies.In this study, four datasets mentioned ATGL and UCP3 simultaneously, and three mentioned UCP2 and UCP3 simultaneously.

Data Extraction
We extracted expression levels of pig ATGL, CIDEA, UCP2, and UCP3 affected by cold stress from figures in the reviewed studies.GetData Graph Digitizer software (version 2.26) was used to extract the data.For example, we extracted Wuzhishan pig UCP3 expression data (control, 0.99; standard deviation [SD], 0.4) from the Lin 2017 report [19].The actual value of the control group was set at 1, and we carefully kept our error within a 95% confidence interval (CI).The extracted data included the number of experimental pigs, gene expression levels, and either SD or standard error (SE).If SE was obtained, it was converted to SD.Table 2 shows the characteristics of the included studies.We analyzed each treatment as an independent dataset [20]; for example, 10 h and 15 days in the Gao 2018 study were analyzed as two separate datasets [18].The articles present the results of relative messenger ribonucleic acid (mRNA) expression, transcript expression levels, and transcripts per million (tpm) data, which were extracted by artificial means by two authors of the current study.

Traditional and Network Meta-Analyses
The effects of cold stress on the expression of ATGL, CIDEA, UCP2, and UCP3 were determined through a traditional meta-analysis.We used Review Manager (Cochrane Collaboration v5.4; https://cochrane.org/accessed on 21 September 2020) to calculate the total effect size of gene expression.Continuous data were used to determine the level of gene expression, and the 95% CI was calculated using the mean difference (MD).
We compared the expression of ATGL, CIDEA, UCP2, and UCP3 under cold stress using a network meta-analysis, which was performed using the packages "coda", "rjag", and "gemtc" in R software (R Foundation for Statistical Computing, Vienna, Austria v4.3.2 31 October 2023).Gene expression data were normalized to fold change (FC) as follows (using UCP3 as an example): The corresponding 95% CIs were used to calculate effect sizes for continuous gene expression data.

Traditional Meta-Analysis
The four studies retained for our meta-analysis include seven treatments [16][17][18][19].Details are provided in Table 2. Due to one meta-analysis report including only one study [21], we decided to proceed with our analysis.

Network Meta-Analysis
The network diagram in Figure 1C compares gene expression among the swine adipocyte genes ATGL, CIDEA, UCP2, and UCP3.Trace plots with 25,000 iterations are shown in Figure 2A.Markov Chain Monte Carlo (MCMC) convergence was excellent.The normal distribution of density plots with 5000 iterations did not display a double-peak phenomenon (Figure 2B).Furthermore, the Brooks-Gelman-Rubin (BGR) diagnosis plot (Figure 2C) shows that the potential scale reduction factor was between 1.00 and 1.01 after 22,000 iterations, indicating that the running outcomes of the network meta-analysis were reliable.Figure 3 shows that cold stress had a consistent effect on the expression of ATGL, CIDEA, UCP2, and UCP3 without any order of priority.

Molecular Docking of UCP1, UCP2, and UCP3
Comparative results of human UCP1 and UCP3 protein docking are shown in Figure 4. UCP3 (Figure 4B) bound to ATP more tightly than UCP1 (Figure 4A), while porcine UCP2 (Figure 4D) bound to ATP more strongly than UCP3 (Figure 4C).UCP1 was more

Molecular Docking of UCP1, UCP2, and UCP3
Comparative results of human UCP1 and UCP3 protein docking are shown in Figure 4. UCP3 (Figure 4B) bound to ATP more tightly than UCP1 (Figure 4A), while porcine UCP2 (Figure 4D) bound to ATP more strongly than UCP3 (Figure 4C).UCP1 was more active than UCP3 in human cells.We believe this would result in increased UCP3 activity in pig adipocytes.active than UCP3 in human cells.We believe this would result in increased UCP3 activity in pig adipocytes.

Discussion
Cold stress-induced BAT activation is believed to improve human metabolic health [22].Cold stress increases levels of nicotinamide adenine dinucleotide (NAD + ); this can further promote mitochondrial heat production [22,23], which can enhance immunity.

Effect of Cold Stress on Adipocyte Gene Expression
Cold stress signals affect adipocyte gene expression by activating pathways in the nervous and endocrine systems.Research has found that cold stress can activate the hypothalamic-pituitary-adrenal (HPA) axis, promoting corticosterone secretion [24].Corticosterone can regulate adipocyte gene expression, thereby affecting adipocyte growth, differentiation, and function [25].Corticosterone can increase the expression of ATGL in broiler chicken WATs [26].However, our results showed a negative correlation between

Discussion
Cold stress-induced BAT activation is believed to improve human metabolic health [22].Cold stress increases levels of nicotinamide adenine dinucleotide (NAD + ); this can further promote mitochondrial heat production [22,23], which can enhance immunity.

Effect of Cold Stress on Adipocyte Gene Expression
Cold stress signals affect adipocyte gene expression by activating pathways in the nervous and endocrine systems.Research has found that cold stress can activate the hypothalamic-pituitary-adrenal (HPA) axis, promoting corticosterone secretion [24].Corticosterone can regulate adipocyte gene expression, thereby affecting adipocyte growth, differentiation, and function [25].Corticosterone can increase the expression of ATGL in broiler chicken WATs [26].However, our results showed a negative correlation between cold stress and ATGL, indicating that the cold stress mechanism in broiler chicks is different from that in pigs.
Expression of certain adipocyte genes changes under cold stress conditions [5,16].These genes are crucial in regulating the growth, differentiation, and function of adipocytes under cold stress.For example, heat shock factor 1 (HSF1) protects cell membrane stability, reduces intracellular oxidative stress (OS) levels, and alleviates cold stress damage to adipocytes [27].Changes in UCP2 expression are generally believed to be caused by cold stress.However, our results showed no relationship between cold stress and CIDEA and UCP2 expression.Additional research might be required to validate this finding.
Although cold stress promotes pig UCP1 gene expression [18], the pig UCP1 gene is defective and cannot synthesize protein [8], meaning that it can be detected only at the gene expression level.Similar studies have shown that protein kinase D1 (Prkd1) deletion in mouse BAT maintains thermogenesis after cold exposure [28], suggesting compensatory pathways in adipose tissue.Our results showed a positive correlation between cold stress and swine adipocyte UCP3 gene expression, indicating that the high activity of UCP3 might be due to the loss of function of UCP1 in pigs.

Transformation of Pig Adipocytes and Induction Factors
BAT contains more mitochondria than WAT [2].Genes expressed in pig WAT include those involved in BAT differentiation in the natural growth state [29].While there are no reports on pigs with BAT [30], pigs might possess beige (brite adipose tissue) [19], which could have a mutual-conversion relationship with WAT [5].Our published articles indicate that cold stress can promote WAT browning [18].In addition, rosiglitazone and rapamycin can induce WAT in beige adipocytes in vitro [31].Rosiglitazone, T3, and indomethacin can reportedly also induce beige adipocytes in mouse WAT [32].

Mutual Regulation of ATGL, CIDEA, UCP2, and UCP3
The proposed mechanisms of action of ATGL, CIDEA, UCP2, and UCP3 in swine adipocytes are shown in Figure 5.
cold stress and ATGL, indicating that the cold stress mechanism in broiler chicks is different from that in pigs.
Expression of certain adipocyte genes changes under cold stress conditions [5,16].These genes are crucial in regulating the growth, differentiation, and function of adipocytes under cold stress.For example, heat shock factor 1 (HSF1) protects cell membrane stability, reduces intracellular oxidative stress (OS) levels, and alleviates cold stress damage to adipocytes [27].Changes in UCP2 expression are generally believed to be caused by cold stress.However, our results showed no relationship between cold stress and CIDEA and UCP2 expression.Additional research might be required to validate this finding.
Although cold stress promotes pig UCP1 gene expression [18], the pig UCP1 gene is defective and cannot synthesize protein [8], meaning that it can be detected only at the gene expression level.Similar studies have shown that protein kinase D1 (Prkd1) deletion in mouse BAT maintains thermogenesis after cold exposure [28], suggesting compensatory pathways in adipose tissue.Our results showed a positive correlation between cold stress and swine adipocyte UCP3 gene expression, indicating that the high activity of UCP3 might be due to the loss of function of UCP1 in pigs.

Transformation of Pig Adipocytes and Induction Factors
BAT contains more mitochondria than WAT [2].Genes expressed in pig WAT include those involved in BAT differentiation in the natural growth state [29].While there are no reports on pigs with BAT [30], pigs might possess beige (brite adipose tissue) [19], which could have a mutual-conversion relationship with WAT [5].Our published articles indicate that cold stress can promote WAT browning [18].In addition, rosiglitazone and rapamycin can induce WAT in beige adipocytes in vitro [31].Rosiglitazone, T3, and indomethacin can reportedly also induce beige adipocytes in mouse WAT [32].

Mutual Regulation of ATGL, CIDEA, UCP2, and UCP3
The proposed mechanisms of action of ATGL, CIDEA, UCP2, and UCP3 in swine adipocytes are shown in Figure 5.  ATGL, CIDEA, UCP2, and UCP3 participate in regulating fat metabolism and energy balance.ATGL and CIDEA jointly regulate fat metabolism, produce metabolic products, and affect UCP2 and UCP3 expression.The main function of UCP2 and UCP3 in mitochondria is to produce heat [9].These genes work together to regulate fat metabolism and energy balance, which helps maintain homeostasis in organisms.Our network meta-analysis results showed no strong causal relationship among these four genes under cold stress conditions.
The results of our molecular-docking experiment indicated that in pig adipocytes, UCP2 is bound to ATP more strongly than UCP3.Similarly, in humans, UCP3 is bound to ATP more tightly than UCP1.This difference in ATP binding capacity between UCP1 and UCP3 in human adipocytes can reportedly result in UCP3 playing a role only when UCP1 cannot fulfill its physiological function [33].Therefore, we hypothesized that pig adipocytes have high UCP3 activity and that UCP2 plays a role only when UCP3 cannot fulfill its physiological function.If this is the case, then in pig adipocytes, UCP2 plays a more significant role than UCP3.
In summary, swine adipocyte UCP3 gene expression was found to be positively correlated with cold stress, while ATGL expression was negatively correlated with cold stress.We found no correlation between cold stress and the expression of UCP2 and CIDEA in swine adipocytes.During the early stages of cold stress, we hypothesized that changes in UCP2 and CIDEA expression that occur might cause this phenomenon.Then, adipocyte volume decreased with time.CIDEA was no longer needed to regulate the fusion of lipid droplets, and gene expression returned to normal levels.Similarly, expression of UCP2, which initially functioned together with UCP3, also returned to normal levels because UCP3 could independently fulfill its physiological function.

Limitation
The study used four published articles that employed relative gene mRNA expression, transcript expression levels, and transcripts per million data.The data underwent normalization during the network meta-analysis, which might have led to errors.The aim of the network meta-analysis was to identify the genes most affected by cold stress and then rank them.We saw no clear differences among ATGL, CIDEA, UCP2, and UCP3, which is a result we can accept.UCP3 is reported to have a higher expression level in skeletal muscle [34] and the liver.However, expression of UCP3 in skeletal muscle under cold stress was considered in the meta-analysis [16].

Conclusions
The gene expression of swine adipocyte UCP3 was positively affected, and that of ATGL was negatively affected, by cold stress.Network meta-analysis results showed that cold stress affected the expression of ATGL, CIDEA, UCP2, and UCP3.One included study shows that cold stress promotes expression of the pig UCP1 gene [18], but this gene cannot synthesize protein, and its failure to do so might lead to high UCP3 activity as a compensatory mechanism.

Figure 1 .
Figure 1.Summary of the study selection process and meta-analysis.(A) Literature selection process: Our search identified 32 publications based on keywords; ultimately, we used four articles.(B) Traditional meta-analysis of the effect of cold stress on ATGL, CIDEA, UCP2, and UCP3 gene expression (Gao 2018 [18], Zhou 2022 [17], Lin 2017 [19], Zhang 2022[16]).SAT: subcutaneous adipose tissue.VAT: visceral adipose tissue.The values of the invalid lines in the graph are 0, so when the 95% confidence interval (CI) was entirely greater than 0, cold stress was considered positively correlated with swine adipocyte gene expression.When the 95% CI was entirely less than 0, cold stress was considered negatively correlated with swine adipocyte gene expression.Green squares represent the weight of the study, and black diamonds represent the 95% CI. (C) Network diagrams comparing different genes.The thicker the line, the greater the number of studies.In this study, four datasets mentioned ATGL and UCP3 simultaneously, and three mentioned UCP2 and UCP3 simultaneously.

Figure 1 .
Figure 1.Summary of the study selection process and meta-analysis.(A) Literature selection process: Our search identified 32 publications based on keywords; ultimately, we used four articles.(B) Traditional meta-analysis of the effect of cold stress on ATGL, CIDEA, UCP2, and UCP3 gene expression (Gao 2018 [18], Zhou 2022 [17], Lin 2017 [19], Zhang 2022[16]).SAT: subcutaneous adipose tissue.VAT: visceral adipose tissue.The values of the invalid lines in the graph are 0, so when the 95% confidence interval (CI) was entirely greater than 0, cold stress was considered positively correlated with swine adipocyte gene expression.When the 95% CI was entirely less than 0, cold stress was considered negatively correlated with swine adipocyte gene expression.Green squares represent the weight of the study, and black diamonds represent the 95% CI. (C) Network diagrams comparing different genes.The thicker the line, the greater the number of studies.In this study, four datasets mentioned ATGL and UCP3 simultaneously, and three mentioned UCP2 and UCP3 simultaneously.

Figure 2 .
Figure 2. Network meta-analysis credibility.(A) Trace plots of network meta-analysis: Overlapping areas are numerous in MCMC, making it impossible to identify individual chains.(B) Density plots of network meta-analysis.With 5000 iterations, each result showed a single peak.We did not observe the double-peak phenomenon, indicating a high level of confidence in the computational results.(C) BGR diagnosis plot.After 5000 iterations, the shrink factor of each result quickly converged to 1.00 and remained stable, indicating that the model had achieved the expected level of convergence.

Figure 2 .
Figure 2. Network meta-analysis credibility.(A) Trace plots of network meta-analysis: Overlapping areas are numerous in MCMC, making it impossible to identify individual chains.(B) Density plots of network meta-analysis.With 5000 iterations, each result showed a single peak.We did not observe the double-peak phenomenon, indicating a high level of confidence in the computational results.(C) BGR diagnosis plot.After 5000 iterations, the shrink factor of each result quickly converged to 1.00 and remained stable, indicating that the model had achieved the expected level of convergence.

Figure 4 .
Figure 4. Molecular docking of UCP1, UCP2, and UCP3.The binding affinities of human UCP1 and UCP3 were −3.21 and −4.01 kcal/mol, respectively, while those of swine UCP3 and UCP2 were −4.29 and −5.33 kcal/mol, respectively.In swine, the number of amino acids involved in the binding pocket of molecular docking exceeded that in humans, and the degree of binding was also tighter than in humans.The dashed line in the figure represents a connection bond of <5 Å.

Figure 4 .
Figure 4. Molecular docking of UCP1, UCP2, and UCP3.The binding affinities of human UCP1 and UCP3 were −3.21 and −4.01 kcal/mol, respectively, while those of swine UCP3 and UCP2 were −4.29 and −5.33 kcal/mol, respectively.In swine, the number of amino acids involved in the binding pocket of molecular docking exceeded that in humans, and the degree of binding was also tighter than in humans.The dashed line in the figure represents a connection bond of <5 Å.

Figure 5 .
Figure 5. Proposed mechanisms of action of ATGL, CIDEA, UCP2, and UCP3 in swine adipocytes.Lipid droplet transfer between multiple adipocytes was accomplished using CIDEA; this transfer is mediated by the closure or opening of the CIDEA channel between adjacent adipose cells.PKA promotes phosphorylation of ATGL and, after the release of ATGL's phosphate, mediates the breakdown of TGs into DAGs and FFAs.UCP2 and UCP3, located on the surface of mitochondria, pump

Figure 5 .
Figure 5. Proposed mechanisms of action of ATGL, CIDEA, UCP2, and UCP3 in swine adipocytes.Lipid droplet transfer between multiple adipocytes was accomplished using CIDEA; this transfer is mediated by the closure or opening of the CIDEA channel between adjacent adipose cells.PKA promotes phosphorylation of ATGL and, after the release of ATGL's phosphate, mediates the breakdown of TGs into DAGs and FFAs.UCP2 and UCP3, located on the surface of mitochondria, pump hydrogen ions into the mitochondria and simultaneously release heat.FFA: free fatty acid.PKA: protein kinase A. TG: triglyceride.DAG: diacylglycerol.The red dashed line indicates inhibition.

Table 1 .
Inclusion and exclusion criteria.

Table 2 .
Characteristics of the studies included.