Abstract

   RNA splicing dysregulation and the involvement of specific splicing
   factors are emerging as common factors in both obesity and metabolic
   disorders. The study provides compelling evidence that the absence of
   the splicing factor SRSF1 in mature adipocytes results in whitening of
   brown adipocyte tissue (BAT) and impaired thermogenesis, along with the
   inhibition of white adipose tissue browning in mice. Combining
   single‐nucleus RNA sequencing with transmission electron microscopy, it
   is observed that the transformation of BAT cell types is associated
   with dysfunctional mitochondria, and SRSF1 deficiency leads to
   degenerated and fragmented mitochondria within BAT. The results
   demonstrate that SRSF1 effectively binds to constitutive exon 6 of
   Ndufs3 pre‐mRNA and promotes its inclusion. Conversely, the deficiency
   of SRSF1 results in impaired splicing of Ndufs3, leading to reduced
   levels of functional proteins that are essential for mitochondrial
   complex I assembly and activity. Consequently, this deficiency disrupts
   mitochondrial integrity, ultimately compromising the thermogenic
   capacity of BAT. These findings illuminate a novel role for SRSF1 in
   influencing mitochondrial function and BAT thermogenesis through its
   regulation of Ndufs3 splicing within BAT.

   Keywords: dysregulatred splicing, impaired thermogenesis, mitochondria
   dysfunction, splicing regulator, whitening of brown adipocyte tissue
     __________________________________________________________________

   Eliminating SRSF1 genetically triggers the whitening of brown adipocyte
   tissue (BAT) and compromises its thermogenic function, consequently
   causing metabolic disorders in mice. Combining single‐nucleus RNA
   sequencing with transmission electron microscopy, the study highlights
   that the absence of SRSF1 disrupts mitochondrial homeostasis, leading
   to an overproduction of mitochondrial reactive oxygen species in BAT
   due to alterations in the splicing of Ndufs3.

   graphic file with name ADVS-11-2306871-g001.jpg

1. Introduction

   Obesity, characterized by the accumulation of excess body fat resulting
   from an imbalance between energy intake and energy expenditure (EE),^[
   [38]^1 ^] underscores the significance of enhancing EE in key metabolic
   organs like adipose tissue as a potential strategy for obesity
   prevention.^[ [39]^2 ^] Adipose tissue comprises two primary types:
   white adipocyte tissue (WAT), functioning primarily as an energy
   reservoir, and BAT, responsible for non‐shivering thermogenesis and
   capable of utilizing surplus lipids and glucose for energy
   expenditure.^[ [40]^3 ^] Studies have demonstrated that external
   stimuli can induce the transformation of a subset of WAT cells into
   brown‐like adipocytes, referred to as beige cells, in a process known
   as white fat “browning”.^[ [41]^2 ^] Notably, substantial quantities of
   metabolically active BAT have been observed in adult humans,
   underscoring the potential of stimulating the development and function
   of both brown and beige fat as a promising approach for combatting
   obesity.^[ [42]^4 ^]

   Proper mitochondrial function is essential for the metabolic stability
   and performance of BAT.^[ [43]^5 ^] The mitochondrial respiratory
   chain, encompassing the mitochondrial electron transport chain (mETC)
   and Fo‐F1ATP synthase, constitutes a series of protein complexes
   (complexes I, II, III, IV, and V) situated within the within the inner
   mitochondrial membrane (IMM), playing a central role in oxidative
   phosphorylation (OXPHOS). Electrons required for OXPHOS are sourced
   from metabolic pathways involving carbohydrates, lipids, and amino
   acids, and are channeled into the mETC.^[ [44]^6 ^] Complex I (CI),
   composed of 45 subunits encoded by both mitochondrial DNA (mtDNA) and
   nuclear DNA (nDNA), serves as the primary and intricate enzyme within
   the respiratory chain. Its function involves catalyzing the transfer of
   electrons from NADH oxidation to ubiquinone.^[ [45]^1 , [46]^6 , [47]^7
   ^] Additionally, CI plays a crucial role in generating reactive oxygen
   species (ROS). ROS are normal byproducts of respiration and are
   necessary for intracellular signaling pathways.^[ [48]^8 ^] However, an
   excessive mtROS resulting from mETC dysfunction can promote protein
   unfolding and aggregation, ultimately leading to a compromised and
   dysfunctional state of the mitochondria.^[ [49]^9 ^] These consequences
   can contribute to a spectrum of diseases and disorders, including
   neurodegeneration, metabolic syndrome, and aging.^[ [50]^10 ^]

   CI is initially formed by a subcomplex consisting of ≈89 kDa, which
   includes the core subunit NDUFS3, in conjunction with NDUFS2 and
   NDUFA5.^[ [51]^11 ^] While NDUFS3 itself does not possess direct
   catalytic activity, it plays a pivotal role in CI assembly. Missense
   mutations in the NDUFS3 gene have been linked to various diseases, such
   as Leigh syndrome, and deficiency of NDUFS3 can result in mitochondrial
   myopathy and optic atrophy.^[ [52]^11 , [53]^12 ^] These prior findings
   underscore the critical importance of an appropriate structure and
   composition for the biogenesis and assembly of CI. Such precision is
   essential to enable the complex to efficiently perform catalytic
   functions while minimizing potential harm to the mitochondria or the
   cell.

   Serine/arginine‐rich splicing factors (SRSFs) have undergone extensive
   investigation due to their significant contributions to constitutive
   splicing and alternative splicing (AS).^[ [54]^13 ^] Typically, SRSFs
   consist of one or two RNA recognition motifs (RRMs) and one
   serine/arginine‐rich (RS) domain. Among these splicing regulators,
   SRSF1, serving as the archetype member of the SR protein family,^[
   [55]^14 ^] has the capability to modulate AS isoform generation by
   binding to specific RNA sequences and influencing spliceosome
   activity.^[ [56]^15 ^] While SRSF1 is recognized for its involvement in
   various biological processes and diseases, including the cell cycle,
   apoptosis, adipocyte differentiation, stress response, cancer, and
   neurodegeneration,^[ [57]^16 ^] its precise function in mature
   adipocytes still lacks clarity.

   In the study, we investigate the role of SRSF1 in adipose tissue by
   generating adipose‐specific SRSF1 knockout mice. Our analysis revealed
   that the absence of SRSF1 had a disruptive impact on mitochondrial
   integrity, leading to a significant whitening of BAT characterized by a
   reduction in the number of mitochondria. By employing single‐nucleus
   RNA sequencing, we identify shifts in cell type composition and
   alterations in differentiation pathways within BAT in the absence of
   SRSF1. Further investigation reveals the pivotal role of SRSF1 in
   promoting the inclusion of constitutive exon 6 of Ndufs3, a crucial
   subunit of mitochondrial complex I. In the absence of SRSF1, exon 6 of
   Ndufs3 is skipped, resulting in diminished production of functional
   protein and disrupted mitochondrial function. Notably, the
   overexpression of native Ndufs3 alleviates mitochondrial defects and
   restores their function in SRSF1‐knockdown adipocytes. These findings
   underscore the important role of SRSF1‐mediated splicing of Ndufs3 in
   mitochondrial homeostasis and brown adipocyte thermogenesis, offering
   novel insights into the significance of splicing regulation in
   metabolic processes.

2. Results

2.1. Inactivation of SRSF1 in Adipose Tissue Triggers the “Whitening” of BAT
and Leads to Metabolic Disorders in Mice

   To investigate the role of SRSF1 in adipose tissue and whole‐body
   energy metabolism, we generated an adipose‐specific SRSF1‐knockout (KO)
   mouse model by breeding SRSF1^flox/flox mice with transgenic C57BL/6J
   mice expressing Cre recombinase controlled by the mouse adiponectin
   (Adipoq) promoter active in mature adipocytes.^[ [58]^17 ^] The KO mice
   carried the Adipoq‐Cre;SRSF1^flox/flox genotype, while the
   SRSF1^flox/flox mice served as the wild‐type (WT) control group (Figure
   [59]S1A, Supporting Information). Confirmation of SRSF1 deletion was
   achieved through western blot analysis in BAT, inguinal WAT (iWAT), and
   epididymal WAT (eWAT) (Figure [60] 1A, left panel). The
   stromal‐vascular fraction (SVF) consists of preadipocytes, endothelial
   cells, fibroblasts and other cell types within adipose tissue,
   excluding mature adipocytes. Notably, the retention of SRSF1 in SVF
   cells emphasizes the specificity of the adipocyte‐targeted SRSF1
   deletion (Figure [61]1A, right panel).

Figure 1.

   Figure 1
   [62]Open in a new tab

   Deletion of SRSF1 in adipose tissue caused whitening of BAT and
   metabolic disorders in mice. A) Protein levels of SRSF1 were assessed
   via western blot in the BAT, iWAT, and eWAT of 8‐week‐old WT and KO
   mice (Left). SRSF1 protein levels were analyzed in the SVF of BAT,
   iWAT, and eWAT from 8‐week‐old WT and KO mice (Right). B)
   Representative pictures of BAT, iWAT, eWAT from 8‐week‐old WT and KO
   mice were shown on the left. The ratio of BAT, iWAT, and eWAT relative
   to body weight in 8‐week‐old WT and KO mice fed a chow diet was
   measured and shown on the right graph (n = 13). C) Representative HE
   images of BAT, iWAT, and eWAT from 8‐week‐old WT and KO mice were shown
   on the left. The presence of large unilocular lipid droplets (LDs) in
   BAT was indicated by arrowheads. Scale bar, 50 µm. The LD diameters in
   BAT, iWAT, and eWAT for both WT and KO mice were measured and presented
   as the mean ± SEM of 400 cells per sample on the right panel (n = 4
   mice per group). D) The ratio of liver relative to body weight of
   8‐week‐old WT and KO mice fed a chow diet was measured. E)
   Representative pictures of liver samples from 8‐week‐old mice were
   shown on the left. The middle images show the liver sections stained
   with HE, while the right images show the liver sections stained with
   BODIPY 493/503. The yellow color represents LDs, and the nuclei are
   stained with DAPI. The scale bar is 25 µm for the middle panel and
   50 µm for the right panel. F) The lipid profile, including plasma
   levels of TG, cholesterol, HDL‐C, and LDL‐C, was measured from
   8‐week‐old WT and KO mice on a chow diet (n = 8). G) Blood glucose
   levels were measured in 16‐hour‐fasted WT and KO mice at 8 weeks of age
   on a chow diet (n = 12). H) Blood glucose concentrations were monitored
   during a GTT (n = 6). The area under the curve (AUC) is depicted in the
   right graph. I) Blood glucose levels were represented as a percentage
   of basal glycemia during intraperitoneal ITT in both WT and KO mice.
   The AUC is shown in the right graph. J) Western blot analysis was
   conducted to evaluate total Akt and Akt phosphorylation in BAT and
   liver extracts obtained from 8‐week‐old WT and KO mice that were
   maintained on a chow diet. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001 by
   Student's t‐test. Data represent the mean ± SEM.

   Both KO mice and WT mice exhibited comparable body weights and food
   intake level when fed a standard chow diet (Figure [63]S1B,C,
   Supporting Information). However, KO mice displayed a notable increase
   in BAT mass, a slight decrease in eWAT mass, and no significant change
   in iWAT mass compared to WT mice (Figure [64]1C). Interesting, KO mice
   exhibited a striking whitening effect in BAT, characterized by the
   accumulation of enlarged lipid droplets (LDs) (Figure [65]1C). In eWAT,
   there was variation in adipocyte size, with smaller LDs observed in KO
   mice compared to WT mice, while no significant difference was observed
   in iWAT between the two groups (Figure [66]1C). Additionally, KO mice
   manifested a pronounced fatty liver phenotype, evidenced by a
   significant increase in the liver‐to‐body weight ratio (Figure [67]1D).
   This lipid droplet accumulation was confirmed through hematoxylin and
   eosin (HE) staining and BODIPY staining (Figure [68]1E). The
   substantial lipid presence in BAT and liver tissues pointed toward a
   disruption of lipid metabolism in KO mice.

   We conducted assessments of the lipid profile by measuring serum levels
   of triglycerides (TGs), cholesterol, high‐density lipoprotein (HDL),
   and low‐density lipoprotein (LDL). Our findings revealed elevated
   levels of TGs and cholesterol in KO mice when compared to their WT
   counterparts (Figure [69]1F). However, there was no significant
   differences in HDL and LDL levels between the two groups. These results
   substantiate the presence of a lipid metabolism disorder in
   adipose‐specific SRSF1 deficient mice. Moreover, when we compared
   fasting and refeeding serum glucose levels, KO mice exhibited
   significantly higher glucose levels than WT mice fed on a chow diet
   (Figure [70]1G). Glucose and insulin tolerance tests (GTTs and ITTs,
   respectively) unveiled that KO mice had severe glucose intolerance and
   insulin resistance (Figure [71]1H,I). In line with the ITT data, the
   response of insulin on Akt phosphorylation in BAT and liver were
   diminished in KO mice (Figure [72]1J). In conclusion, our results
   demonstrate that SRSF1 deficiency in adipose tissue leads to BAT
   whitening, hepatic steatosis, hypertriglyceridemia, and insulin
   resistance

2.2. SRSF1 Knockout Mice Exhibited Pronounced Cold Intolerance and Impaired
Thermogenesis

   Considering the notable increase in lipid accumulation and the
   pronounced whitening phenotype observed in the interscapular BAT of KO
   mice, we decided to investigate how SRSF1 deficiency affects BAT
   function. To accomplish this, we conducted high‐throughput RNA‐seq on
   the BAT samples collected from 8‐week‐old WT and KO mice. Our analysis
   clearly revealed that key thermogenic marker genes, including Dio2,
   Cidea, and Cox8b, were downregulated in KO mice, while markers
   associated with inflammation, such as Itgax and Adgre1, were
   upregulated (Figure [73] 2A). Furthermore, gene set enrichment analysis
   (GSEA) indicated that the downregulated genes in the BAT of KO mice
   were predominantly enriched in pathways related to thermogenesis and
   OXPHOS (Figure [74]2B). Correspondingly, KO mice exhibited a
   significant decrease in their core body temperature during both feeding
   and fasting periods when kept at room temperature (RT) (Figure [75]2C;
   Figure [76]S2A, Supporting Information).

Figure 2.

   Figure 2
   [77]Open in a new tab

   SRSF1 deficiency impairs the thermogenic function of BAT. A)Volcano
   plot showing DEGs in BAT of both WT and KO mice using RNA‐seq data.
   Biomarkers relative to thermogenesis and inflammation were highlighted
   using different colors. B) Gene Set Enrichment Analysis (GSEA)
   identifying the most significantly downregulated pathways, including
   thermogenesis and OXPHOS. C) The rectal temperature of 8‐week‐old WT
   and KO mice was measured at RT (n = 6). D) The rectal temperature of
   8‐week‐old mice, individually housed, was measured after subjecting
   them to cold exposure (4 °C) (n = 8). No food was provided to the mice
   during this period. E) Survival curves was plotted for WT and KO mice
   exposed to cold (4 °C), with the mice housed individually in separate
   cages without food (n = 16). F)The core body temperature of 8‐week‐old
   WT and KO mice was measured under cold stimulation conditions (4 °C).
   During this period (24 h), mice were co‐housed together and provided
   with food (n = 8). G) Survival curves were also plotted for mice
   co‐housed and exposed to cold for 48 h (n = 20). H)Thermal images,
   along with the surface temperature and iBAT temperature of 8‐week‐old
   WT and KO mice, were displayed under different conditions. In the top
   panel, the mice were kept at RT. In the middle panel, mice were
   individually housed at 4 °C for 4 h without food. In the bottom panel,
   WT and KO mice were co‐housed at 4 °C for 24 h with food. The color
   scale represents the temperature range, with blue indicating cold and
   red indicating hot. I,J) Representative images of HE staining (I) and
   Ucp1 immunohistochemistry (J) of BAT from mice housed at RT or 4 °C for
   24 h are shown. Scale bar: 50 µm (I), 100 µm (J). K,L) The protein
   levels (K) or mRNA levels (L) of thermogenic genes in BAT of WT mice
   and KO mice were measured after exposing the mice to cold temperature
   (4 °C). WT and KO mice were co‐housed at 4 °C for 24 h (n = 4), and the
   results were compared to mice housed at RT. The q‐PCR data analysis
   involved measuring amplicon abundance via the 2^−ΔΔCT method,
   normalized to β‐Actin. Each gene's value in WT‐RT was set as 1 to
   assess the relative mRNA levels of target genes across the KO‐RT,
   WT‐4 °C, and KO‐4 °C groups. ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001.
   Data represent the mean ± SEM.

   To further explore the impact of SRSF1 deficiency on adaptive
   thermogenesis, we conducted cold stimulation experiments for varying
   durations on mice. For acute cold exposure without food, individual
   mice were placed in separate cages at 4 °C. Notably, KO mice
   experienced a rapid decline in core body temperatures over time,
   ultimately leading to mortality after 3 h (Figure [78]2D,E). In
   contrast, WT mice maintained core body temperatures above 30 °C and
   exhibited no mortality under the same conditions (Figure [79]2D,E).
   When provided food during 6 h cold exposure (housed separately) and
   extended periods of cold exposure (co‐housed), KO mice consistently
   displayed lower rectal temperatures than WT mice (Figure [80]S2B,
   Supporting Information; Figure [81]2F). Furthermore, while KO mice
   survived the 24 h stimulation period, they exhibited a higher mortality
   rate after the 24 h period (Figure [82]2G). Thermal imaging analysis
   unveiled that the surface temperatures and brown fat temperatures in
   the interscapular region of KO mice were reduced when they were housed
   at RT compared to WT mice. Furthermore, this reduction in temperature
   became even more pronounced in KO mice at 4°, particularly when KO mice
   were housed individually (Figure [83]2H).

   HE staining revealed that cold exposure led to a reduction in fat
   droplet size in the BAT of WT mice, indicating enhanced lipolysis and
   thermogenesis. In contrast, KO mice did not exhibit a significant
   reduction in lipid accumulation within BAT (Figure [84]2I). Ucp1, a
   mitochondrial transmembrane protein, functions by decoupling OXPHOS
   from ATP production.^[ [85]^18 ^] Mice lacking Ucp1 are susceptible to
   developing hypothermia when exposed to sudden cold conditions.^[
   [86]^19 ^] The tightly controlled expression of Ucp1 in brown and beige
   adipocytes established Ucp1 as a pivotal indicator of thermogenesis.^[
   [87]^20 ^] Immunohistochemical analysis further revealed reduced
   immunoreactivity of Ucp1 in the BAT of KO mice following cold exposure
   (Figure [88]2J). Additionally, western blot analysis demonstrated a
   significant increase in the levels of thermogenic proteins, including
   Adrb3, Pgc‐1α, and Ucp1 in the BAT of WT mice after exposure to cold
   stimulation. In contrast, KO mice did not exhibit a significant
   induction of these proteins following cold stimulation, and their
   expression remained lower in KO mice even when kept at RT, compared to
   WT mice (Figure [89]2K). Furthermore, the mRNA levels of thermogenic
   genes (Cidea, Ucp1, Ppargc1a), genes involved in OXPHOS (Aco2 and
   Ndufb8), as well as the lipolysis genes (Lipe and Mgll), were found to
   be decreased in the BAT of KO mice. Moreover, the expression of these
   genes was not stimulated in response to cold exposure in KO mice
   compared to WT mice (Figure [90]2L). Collectively, these findings
   strongly support the conclusion that the reduced expression of
   thermogenic genes led to impaired thermogenesis in KO mice.

   To further substantiate this, we isolated primary brown adipocytes from
   the BAT of WT and KO mice at 1 day old and cultured them in
   differentiation medium until maturity. As shown, the brown adipocytes
   lacking SRSF1 exhibited significantly lower levels of
   thermogenic‐related proteins (Figure [91]S2C, Supporting Information).
   Even treatment with Forskolin (FSK,) a cAMP‐PKA activator, failed to
   restore their expression (Figure [92]S2D, Supporting Information). This
   points to dysfunction in brown adipocytes as the primary cause of
   impaired BAT thermogenesis in the KO mice.

2.3. SRSF1 Deficiency Impairs β‐Adrenergic Agonist‐Induced Thermogenesis in
BAT

   CL‐316243, a selective β3‐adrenergic receptor agonist, is commonly used
   to mimic cold exposure and activate BAT activity.^[ [93]^21 ^] To
   investigate the significant role of SRSF1 in systemic metabolism, we
   conducted a study employing indirect calorimetry to examine the acute
   metabolic response of WT and KO mice to adrenergic agonists.
   Interestingly, there were no differences in oxygen consumption and
   carbon dioxide production between KO and WT mice without CL‐316243
   stimulation, suggesting that the reduced OXPHOS and thermogenic
   capacity of KO mice did not significantly impact whole‐body metabolism
   in a resting state (Figure [94] 3A,B). However, the administration of
   CL‐316243 resulted in a substantial increase in oxygen consumption and
   carbon dioxide production in WT mice, while having a minimal effect on
   KO mice (Figure [95]3A,B). This finding indicated that SRSF1 is
   essential for the metabolic response to CL‐316243.

Figure 3.

   Figure 3
   [96]Open in a new tab

   SRSF1 deficiency impairs β3‐adrenergic agonist‐induced thermogenesis of
   BAT. A,B) O[2] consumption (A) or CO[2] production (B) were recorded in
   WT and KO mice during a 48 h light‐dark cycle, both before and after an
   injection of CL‐312643 at a dosage of 1 mg kg^−1. O[2] consumption or
   CO[2] production in WT and KO mice was also measured before and after
   1 h of CL‐312643 treatment, and the results are shown on the right. C)
   A schematic design of the CL‐312643 treatment regimen is presented.
   Mice were kept on a standard diet for 8 weeks, followed by
   intraperitoneal injections of either saline or CL‐316243 (1 mg kg^−1)
   in both WT and KO mice for 7 consecutive days. The experiments were
   conducted on the 8th day, after a 7‐day injection period. D)
   Representative images of BAT in both KO and WT mice were obtained after
   continuous injection of CL‐316243 for 7 days. E,F). Representative
   images of HE staining (E) and Ucp1 immunohistochemistry (F) of BAT from
   mice treated as described in (C). Scale bar: 50 µm (E); 100 µm (F). G)
   The relative expression of thermogenic genes in BAT was assessed in
   mice, as detailed in (C) (n = 3), employing the method described in
   Figure [97]2L. H) The expression of thermogenic proteins in BAT was
   evaluated in mice treated as described in (C). ^* p < 0.05, ^** p <
   0.01, ^*** p < 0.001. Data represent the mean ± SEM.

   Research has shown that adrenal hormone stimulation of BAT promotes
   tissue remodeling by enhancing enhanced mitochondrial biogenesis and
   Ucp1 expression.^[ [98]^22 ^] To investigate whether SRSF1 is essential
   for BAT remodeling in response to adrenergic stimulation, mice were
   injected with CL‐316243 or saline (control) for 7 days (Figure [99]3C).
   After 7‐days treatment with CL‐316243, BAT from KO mice still exhibited
   a whitening effect compared to BAT from WT mice (Figure [100]3D). HE
   staining indicated a significant reduction in the size of BAT lipid
   droplets in WT mice, while the BAT of KO mice showed excessive
   accumulation of lipid droplets (Figure [101]3E). Immunohistochemical
   staining further demonstrated a significant increase in Ucp1 expression
   in WT mice treated with CL‐316243, while this induction was compromised
   in KO mice (Figure [102]3F). Moreover, the administration of CL‐316243
   in WT mice resulted in a considerable upregulation of genes related to
   thermogenesis, lipolysis, and OXPHOS in BAT, whereas this response was
   significantly impaired in the BAT of KO mice (Figure [103]3G). Western
   blot analysis provided further support for these findings, revealing a
   significant increase in the expression of thermogenic proteins in the
   BAT of WT mice following agonist treatment. However, this induction was
   markedly compromised in the BAT of KO mice (Figure [104]3H).

2.4. SRSF1 Regulates the Browning of WAT Following Treatment with a
β3‐Adrenergic Receptor Agonist

   In addition to its role in regulating BAT thermogenesis, the
   stimulation of β3‐adrenergic receptor agonist can also induce the
   browning of WAT.^[ [105]^2 ^] Therefore, we investigated the potential
   of CL‐316243 to induce browning of WAT in both WT and KO mice. Our
   experimental approach involved treating both KO and WT mice with
   CL‐316243 for 7 days. Interestingly, CL‐316243 treatment resulted in a
   significant increase in multilocular adipocytes and the characteristic
   browning in the iWAT and eWAT of WT mice. However, the effect of
   CL‐316243 on adipocytes in KO mice was significantly reduced (Figure
   [106]S3A,B, Supporting Information). Furthermore, treatment of with
   CL‐316243 led to enhanced expression of Ucp1 proteins in both iWAT and
   eWAT from WT mice, while Ucp1 induction in KO mice was noticeably
   reduced (Figure [107]S3A,B, Supporting Information). Additionally, the
   induction of both thermogenic proteins and thermogenic gene expression
   was significantly impaired in KO mice treated with CL‐316243 (Figure
   [108]S3C–F, Supporting Information). In conclusion, our results
   demonstrated that SRSF1 deficiency hindered the induction of beige
   adipocytes mediated by adrenergic receptor agonists in WAT,
   highlighting the important role of SRSF1 in regulating the browning
   process and the thermogenic response of adipocytes in WAT.

2.5. Single Nucleus RNA Seq (snRNA‐seq) Unveils Distinctive Cell Types in the
BAT of KO Mice

   To investigate cell population heterogeneity and explore the molecular
   and pathway characteristics linked to thermogenic function, we
   conducted snRNA‐seq analysis on BAT samples obtained from 8‐week‐old WT
   and KO mice. In total, we collected 24341 nuclei, and after quality
   control we further analyzed data from 19020 single nuclei (Figure
   [109]4A). The initial clustering and marker gene annotation identified
   13 major cell populations (Figure [110]S4A, Supporting Information).
   Based on the representative markers, we classified these clusters into
   6 distinct cell types, namely stromal cells (Fgf14, mt‐Nd2; 19.3%),
   adipocytes (Plin1, Pparg, Cidec; 25.6%), adipose stem and progenitor
   cells (ASPCs) (Pdgfra, Dcn; 10.0%), mural cells (Trpc3, Myh11; 2.7%),
   endothelial cells (Esam, Pecam1; 21.4%), and immune cells (Adgre1,
   Lyz2; 21.0%) (Figure [111]4B–D).

Figure 4.

   Figure 4
   [112]Open in a new tab

   snRNA‐seq uncovers cellular population heterogeneity in BAT between WT
   and KO mice. A) Overview of the experimental workflow: BAT was isolated
   from both WT and KO mice, dissociated into single‐nucleus suspensions,
   followed by snRNA‐seq and subsequent bioinformatic analyses. B) t‐SNE
   plot illustrates the distribution of various cell types within BAT. C)
   The heatmap displays the top 10 DEGs within each cell type of BAT,
   chosen according to the gene‐diff criteria. D) Expression patterns of
   representative marker genes are depicted for each cell type. The scale
   bar represents gene expression levels, with deeper red shades
   indicating higher expression (denoted as H) and lighter colors
   indicating lower expression (denoted as L). E) Separate t‐SNE plots are
   generated for the WT and KO datasets to explore potential differences
   in cellular distribution. F) Comparative analysis reveals the
   proportion of different cell types between BAT samples from WT and KO
   mice.

   Interestingly, the adipocyte population in KO mice exhibited a notable
   shift in its clustering distribution on the t‐SNE plot, indicating
   distinct molecular and functional features of these adipocytes
   (Figure [113]4E). Analysis of WT and KO cell proportion in each cluster
   showed a decrease in the frequencies of stromal cells, adipocytes, and
   endothelial cells in the BAT of KO mice. The KO mice exhibited a
   dramatic increase in the presence of ASPCs and immune cells. No
   significant differences were observed in the proportion of mural cells
   in the KO mice (Figure [114]4F).

   Examination of the top 20 differentially expressed genes (DEGs)
   revealed extensive downregulation of genes involved in thermogenesis
   (Ucp1, Cidea, Dio2, Ppargc1a, and Elovl6) in KO mice compared to WT
   mice (Figure [115]S4B, Supporting Information). In contrast,
   inflammation‐related genes such as Alcam, Dock2 and Adgre1 were
   upregulated in KO mice (Figure [116]S4B, Supporting Information). We
   annotated the DEGs identified in the comparison between WT and KO mice
   using the KEGG pathway analysis. This analysis revealed that the
   upregulated pathways were primarily associated with the inflammatory
   response, whereas the predominantly downregulated pathways were related
   to thermogenesis and OXPHOS (Figure [117]S4C, Supporting Information).
   Overall, these findings highlight the significant impact of SRSF1 on
   the cellular landscape and molecular pathways associated with BAT
   function.

2.6. A Distinctive Subcluster of Brown Adipocyte, Designated as BA‐3,
Demonstrates Metabolic Dysfunction in the BAT of KO Mice

   To analyze the population of adipocytes and adipocyte stem/progenitor
   cells, we conducted unsupervised clustering in both WT and KO mice. As
   a result, we identified a total of nine unique subclusters, labeled as
   C1 to C9 (Figure [118] 5A). When the t‐SNE plot was segregated, a
   significant difference in the distribution of brown adipocytes (C1‐C3)
   was observed (Figure [119]5B). This disparity indicates a potential
   alteration in the function of brown adipocytes in KO mice compared to
   WT mice. As illustrated in Figure [120]5C, subclusters (C1, C2, C3)
   characterized by the marker Plin1 were identified as mature brown
   adipocytes. C1 and C2, which displayed relatively higher expression of
   Ppargc1a, Ucp1, and Dio2, were predominantly observed in WT mice and
   designated as brown adipocytes‐1 and brown adipocytes‐2 (BA‐1 and BA‐2)
   (Figure [121]5B,C). Conversely, C3 was exclusively present in KO mice
   and exhibited a unique gene expression pattern, with notable
   upregulation of stress‐induced genes such as Trib3, Marc1, and Mreg
   (Figure [122]5B,C). Therefore, this cluster was labeled as BA‐3. C5,
   displaying elevated expression of Lep and Slc7a10, well‐known markers
   for white adipocytes, were identified as white adipocytes (WA). Adipose
   stem cell markers, Dpp4 and Cd55, were exclusively expressed in C7. C8,
   characterized by high levels of Fmo2 and Gdf10, was annotated as the
   adipogenesis regulator (Areg) cluster. Finally, C4, C6, and C9,
   characterized by dominant expression of Col4a1, Icam1, or Cyp1b1, were
   identified as preadipocytes (PreA) clusters. (Figure [123]5A,B).
   Additionally, the KO mice showed a reduction in the number of brown
   adipocytes, accompanied by an increase in the number of WA, ASCs,
   Pre‐A, and Areg cells (Figure [124]5D).

Figure 5.

   Figure 5
   [125]Open in a new tab

   Identification of a distinct subpopulation of brown adipocytes with
   impaired metabolism in SRSF1‐deficient BAT. A) t‐SNE analysis reveals
   the existence of nine distinct subclusters of adipocytes. B) The
   density plot demonstrates differences in the distribution subclusters
   between WT and KO samples. C) The dot plot displays the expression
   levels of selected marker genes for each subcluster, with dot size
   indicating the percentage of cells expressing a specific gene and color
   intensity representing the average expression level. D) A stacked bar
   plot provides a visual representation of the proportion of cells from
   WT or KO samples within each adipocyte subcluster. E) GSVA‐GO analysis
   of biological processes highlights the functional characteristics and
   distinctions of BA‐3 in comparison to other adipocyte subtypes. F)
   Violin plots depict the expression levels of genes related to
   thermogenesis, OXPHOS, and FAO among BA‐1, BA‐2, and BA‐3.

   Gene Ontology enrichment analysis (GO) using gene set variation
   analysis (GSVA) revealed that BA‐3 primarily participated in the
   transport of TGs and fatty acids, as well as in the cellular responses
   to oxidative stress and inflammation (Figure [126]5E). Notably, a
   substantial reduction in the expression of genes associated with
   thermogenesis (Ucp1, Cidea, Dio2, Adrb3), OXPHOS (Aco2, Cox5b, Atp5a1,
   Ndufb8), and fatty acid oxidation (FAO) (Hadha, Acadm) was observed in
   BA‐3 compared with the BA‐1 and BA‐2 (Figure [127]5F). In contrast,
   BA‐1 primarily engaged in amino acid and acetyl‐CoA metabolism, along
   with hormone response (Figure [128]S5A, Supporting Information). BA‐1
   exhibited predominant expression of genes related to lipogenesis (Acly,
   Fasn, Acaca), indicating its primary role in lipid synthesis and energy
   storage within BAT (Figure [129]S5B, Supporting Information). BA‐2 was
   mainly associated with FAO (Figure [130]S5A, Supporting Information).
   BA‐2 displayed significantly higher expression of genes linked to
   thermogenesis (Dio2) and FAO (Hadha, Hadhb, Cpt1b, Ppara) compared to
   BA‐1 (Figure [131]S5C,D, Supporting Information). These findings
   strongly suggest that BA‐2 plays a pivotal role in adipocyte
   thermogenesis.

   The results indicate that the deficiency of SRSF1 contributes to
   alternations in cell type composition within BAT, transitioning from
   functional brown adipocytes in WT mice to impaired brown adipocytes and
   white adipocytes in KO mice.

2.7. Pseudotime Analysis Reveals Distinct Differentiation Trajectories and
Mitochondrial Dysfunction in brown Adipocytes Following SRSF1 Deficiency

   Next, we wanted to utilize pseudotime analysis to reconstruct the
   trajectories of brown adipocytes and evaluate the cellular dysfunction
   arising from SRSF1 deletion. To ensure accurate trajectory analysis of
   brown cell differentiation, our approach involves excluding the white
   adipose cell population.

   Monocle2 organized ASCs (C7), PreA (C4, C6, and C9), Areg (C8), BA‐1
   (C1), BA‐2 (C2), and BA‐3 (C3) along a shared trajectory that
   bifurcated into two distinct branches (Figure [132] 6A). ASCs and PreA
   cells were primarily located at the early part of the pre‐branch
   (Pre‐b), while BA‐1 and BA‐2 cells resided at or near the terminus of
   the primary branch b1. Likewise, BA‐3 cells occupied at or near the
   terminus of the main branch b2 (Figure [133]6A,B). Furthermore, all
   cells were categorized into three states. State 1 consisted of cells
   from both WT and KO mice, whereas state 2 and state 3 were
   predominately occupied by cells from WT and KO mice, respectively
   (Figure [134]6C). This observation suggests that brown adipocytes in WT
   and KO mice display distinct molecular characteristics and functional
   properties within the analyzed trajectory.

Figure 6.

   Figure 6
   [135]Open in a new tab

   Pseudotime analysis reveals abnormal differentiation pathways and
   mitochondria dysfunction within BAT of KO mice. A) The pseudotime
   trajectory map displays the identities of the subclusters. B)
   Pseudotime progression is represented from black to light yellow. C)
   The map illustrates the relationship of all subclusters to the three
   developmental stages (states) based on gene expression. The composition
   of each state in WT and KO samples is shown in a pie chart, with each
   section divided equally according to the pseudotime of the cells. D) A
   heatmap arranges all genes with dynamic expression patterns along
   pseudotime (left). Significantly enriched GO terms for each module are
   presented on the right. E–G). Scatterplot graphs depict the expression
   dynamics of the selected genes along pseudotime, providing a detailed
   view at the resolution of individual cells. Expression levels of genes
   related to mitochondrial DNA replication (E), OXPHOS (F), and
   thermogenesis (G) are plotted against pseudotime. Solid lines represent
   the b1 branch (WT), while dashed lines represented the b2 branch (KO).

   Upon close examination of the trajectory, we identified genes that
   exhibited significant differential expression over pseudotime. These
   genes were then categorized into four modules based on their temporal
   expression patterns, as depicted in the heatmap in Figure [136]6D.
   Specifically, genes in Module 1 (e.g., Clk1, Pi16, Tram1) displayed
   high expression levels at the initial Pre‐b stage, with their
   expression gradually decreasing along pseudotime. These genes were
   found to be enriched in terms related to phosphorylation, cellular
   proliferation, and protein transport, indicating their role in
   precursor adipocytes. Interestingly, both WT and KO mice exhibited
   similar gene expression patterns, suggesting no significant differences
   in the characteristics of these precursor adipocytes (Figure [137]6D).

   Module 2, in contrast, exhibited a significant enrichment of genes
   associated with mitochondrial functions, including mt‐Atp6, mt‐Nd1,
   Uqcrfs1, and Cox8b. The expression of these genes displayed a gradual
   increase along the b1 branch, without notable accumulation at the end
   of the b2 branch (Figure [138]6D–F). This observation strongly
   indicated impaired mitochondrial function in cells from KO. Module 3
   focused on thermogenic genes such as Ucp1, Ppargc1a and Dio2, which are
   typically highly expressed in mature brown adipocytes. The expression
   of these genes progressively increased as pseudotime advanced, reaching
   peak levels during the late maturation stage in cells situated at the
   end of the b1 branch. Conversely, cells at the end of the b2 branch
   exhibited decreased expression of thermogenesis‐related genes in Module
   3, providing clear evidence of impaired BAT function in KO mice
   (Figure [139]6D,G). Module 4 was predominantly localized in cells at
   the terminus of the b2 branch. Functional characterization of Module 4
   through GO analysis indicated its participation in pathways related to
   oxidative stress response and apoptosis (Figure [140]6D). These
   findings strongly suggested a significant stress response in cells from
   KO mice.

2.8. SRSF1 Deletion Compromises Mitochondrial Integrity and Triggers an
Excessive mtROS Production

   Pseudotime analysis revealed mitochondrial dysfunction in brown
   adipocytes of KO mice. The results of GO analysis for BA‐1, BA‐ 2, and
   BA‐3 further emphasized the pronounced mitochondrial dysfunction in
   cells from BA‐3 compared to BA‐1 and BA‐2 (Figure [141]S6, Supporting
   Information). Based on these findings, we conclude that the compromised
   function of adipocytes in KO mice can be primarily attributed to
   mitochondrial dysfunction.

   We then conducted transmission electron microscopy on BAT samples from
   both WT and KO mice. Our findings unveiled severe mitochondrial
   degeneration in brown adipocytes of KO mice, characterized by a
   transformation into small dot‐like structure, accompanied by the
   accumulation of large‐sized lipid droplets, in stark contrast to the
   control group (Figure [142] 7A). Additionally, we observed a
   significant reduction in both mitochondrial abundance and mtDNA copy
   number (Figure [143]7B). At high magnifications, mitochondria in WT
   mice displayed a well‐preserved inner membrane structure with neatly
   aligned cristae. Conversely, cells from KO mice exhibited severely
   disrupted inner mitochondrial membranes, along with a higher percentage
   of mitochondria containing fractured inner membranes
   (Figure [144]7C,D). Additionally, we assessed the expression of key
   genes involved in mitochondrial membrane fusion and division, including
   Mfn1, Mfn2, Opa1, Dnm1l, and Fis1. Notably, we observed lower
   expression levels of these genes in KO mice compared to WT mice, with
   the exception of Mfn2 (Figure [145]7E). In conclusion, our results
   strongly indicate that deficiency of SRSF1 in BAT leads to compromised
   mitochondrial integrity.

Figure 7.

   Figure 7
   [146]Open in a new tab

   Deletion of SRSF1 disrupted mitochondrial homeostasis, leading to
   accumulated mtROS. A) Transmission electron microscopy (TEM) images of
   BAT from 8‐week‐old WT and KO mice are shown at low magnification.
   Lipid droplets are labeled as “LD”, and mitochondria are indicated as
   “M” with arrows. Scale bar, 10 µm. B) Comparison of mitochondrial
   numbers in the BAT of KO and WT mice (left). Measurement of the
   relative content of mtDNA in the BAT of these mice (right). C) TEM
   images of BAT at high magnification, highlighting mitochondria. Scale
   bar, 1 µm. D) Presentation of the percentage of damaged mitochondria in
   both groups. E) Analysis of mitochondrial fusion and fission genes in
   BAT samples from 8‐week‐old WT and KO mice using q‐PCR. F) A dot plot
   illustrates the expression of stress‐related and antioxidant‐related
   genes in BA‐1, BA‐2, and BA‐3 subpopulations. G,H) q‐PCR analysis of
   antioxidant genes (G) and stress‐related genes (H) in BAT samples from
   the same mice mentioned above. H) Measurement of mtROS levels in BAT
   samples from 8‐week‐old WT and KO mice using MitoSOX (n = 3). Data are
   presented as arbitrary fluorescence units (AFU). ^* p < 0.05, ^** p <
   0.01, ^*** p < 0.001. Data represent the mean ± SEM.

   Furthermore, the elevated expression of stress‐related genes and the
   decreased expression of antioxidant genes was observed in BA‐3 when
   compared to BA‐1 and BA‐2 (Figure [147]7F), which was consistent with
   the phenotype of disrupted mitochondria in brown adipocytes of KO mice.
   Comparable alterations in gene expression were also evident in the BAT
   of KO mice (Figure [148]7G,H). To explore whether SRSF1 deficiency led
   to elevated mtROS production, we assessed mitochondrial oxidative
   stress levels using MitoSOX, a probe specifically designed for
   detecting mitochondrial superoxide. Our analysis revealed a pronounced
   accumulation of mitochondrial superoxide in the BAT of KO mice
   (Figure [149]7I). These findings emphasized the crucial role of SRSF1
   in preserving mitochondrial integrity and function in brown adipocytes.

2.9. SRSF1 Regulates the Inclusion of Constitutive exon 6 of Ndufs3,
Promoting the Production of its Functional Protein in BAT

   To elucidate the mechanism underlying the impact of SRSF1 deficiency on
   mitochondria dysfunction in brown adipocytes, we analyzed the RNA‐seq
   data available to identify potential splicing events regulated by SRSF1
   in BAT of KO mice compared with controls. As shown in Figure [150]
   8A,B, the absence of SRSF1 resulted in significant exon skipping in
   genes involved in the assembly of mitochondrial complexes, such as
   Ndufs2, Ndufs3, Atp5d and Atp6ap1, as well as genes related to
   mitochondrial function, including hmox2, Nfs1, Emc2, and Cyp27a1.
   Interestingly, we found that the regulated exons of Ndufs2, Ndufs3,
   Hmox2, and Emc2 are constitutively spliced (Figure [151]S7, Supporting
   Information). This suggests that SRSF1 plays a role in regulating the
   splicing of these constitutive exons in the BAT.

Figure 8.

   Figure 8
   [152]Open in a new tab

   SRSF1 plays a key role in regulating the inclusion of constitutive exon
   6 within Ndufs3 pre‐mRNA. A,B) Validation of representative exon
   inclusion or exclusion events influenced by SRSF1 was conducted through
   RT‐PCR. The regulated exon, identified with its exon number, is
   depicted in the green box. The exclusion/inclusion (Ex/In) ratios of
   RNA products are quantified (n = 3). C) A schematic diagram
   illustrating the Ndufs3 and Ndufs3‐T isoforms, with or without exon 6.
   Note that SRSF1 deletion led to the production of Ndufs3‐T. D) Analysis
   of the expression pattern of Ndufs3 and Ndufs3‐T isoforms in BAT
   samples collected from WT and KO mice using RT‐PCR analysis. E) Western
   blot analysis conducted on BAT samples obtained from WT and KO mice
   using anti‐Ndufs3 antibodies. F) Schematic diagrams illustrating the
   potential binding sites for SRSF1 on exon 6 of Ndufs3, indicated by the
   red‐marked sequence. The binding motifs of SRSF1, predicted by the RBP
   Suite website, are displayed at the bottom. G) q‐PCR analysis performed
   on Ndufs3 pre‐mRNA in samples of immunoprecipitation (IP) using
   anti‐SRSF1 antibodies or IgG control (left) in brown adipocytes. The
   western blot analysis shows SRSF1 protein levels in whole cell extracts
   (input) and post‐IP using the anti‐SRSF1 antibodies (right). H)
   Diagrams of the Nduf3 minigene construct and two of its derivatives are
   presented. The sequence within the exon6, highlighted in red, have been
   substituted with the sequences marked in yellow in the mutation
   minigene. I) Brown adipocytes were co‐transfected with the
   SRSF1‐overexpressing plasmid and either the Ndufs3 minigene or
   derivatives. In vitro splicing analysis of Ndufs3 E6 was performed
   using RT‐PCR, and the resulting Percent Spliced In (PSI) values were
   presented on the right graph. PSI = Inclusion/(Inclusion + Exclusion).
   ^* p < 0.05, ^** p < 0.01, ^*** p,0.001. Data represent the mean ± SEM.

   In this study, we focused on examining the Ndufs3 gene for two primary
   reasons: its established role in mitochondrial function and the notable
   impact observed due to SRSF1 deficiency (Figure [153]8A). Our findings
   revealed that the deletion of SRSF1 caused exon 6 skipping from Ndufs3
   pre‐mRNA, resulting in the formation of a truncated isoform (referred
   to as Ndufs3‐T) (Figure [154]8C). Although the presence of Ndufs3‐T
   mRNA in the BAT of KO mice was confirmed through RT‐PCR analysis
   (Figure [155]8D), western blot analysis revealed a significant decrease
   in the levels of Ndufs3 protein in the BAT of KO mice. The protein
   isoform of Ndufs3‐T was not detected in the KO group, which suggests
   potential instability of the Ndufs3‐T mRNA caused by the omission of
   constitutive exon 6 (Figure [156]8E).

   To investigate the mechanism by which SRSF1 regulates the inclusion of
   exon 6 in Ndufs3, we first utilized the RBP Suite prediction website to
   determine the potential binding motif for SRSF1 in the regions of
   Ndufs3 spanning exon 5 to exon 7. The motif analysis conducted on
   SRSF1‐activated AS events showed a significant enrichment of
   purine‐rich sequences, particularly AGAAGGAC. In fragment exon 6 of the
   Ndufs3 gene, we identified a similar sequence, AGAAGGAT, which could
   potentially serve as a binding site for SRSF1 (Figure [157]8F).
   Subsequently, we performed RNA immunoprecipitation (RIP) experiments in
   brown adipocytes and confirmed the presence of SRSF1 bound to exon 6 of
   Ndufs3 pre‐mRNA (Figures [158]8G). To further investigate the
   association between the binding of SRSF1 to exon 6 and exon inclusion,
   we created a minigene reporter plasmid, which contained a genomic DNA
   fragment of Ndufs3 exons 5–7 (Figure [159]8H). When brown adipocytes
   were co‐transfected with the minigene and SRSF1‐expessing plasmids,
   Ndufs3 exon 6 was completely included (Figure [160]8I, lane 1). We then
   generated a mutation minigene by replacing AGAAGGAT with AT‐rich
   sequences in the exon 6. Brown adipocytes transfected with the Ndufs3
   mutant showed a significant reduction in exon 6 inclusion
   (Figure [161]8I, lane 2). Interestingly, when the SRSF1 binding
   sequence was restored in the Add‐in minigene, exon 6 skipping was
   effectively suppressed in these cells (Figure [162]8I, lane 3 compared
   with lanes 1, 2). These findings suggested that the binding of SRSF1 is
   necessary for inclusion of Ndufs3 exon 6.

2.10. SRSF1 Preserves Mitochondrial Function by Controlling the Inclusion of
exon 6 in Ndufs3 in Primary Brown Adipocytes

   Given that Ndufs3 is a constituent of mitochondrial complex I,
   responsible for electron transport and the generation of mtROS, our
   study aimed to investigate the impact of Ndufs3 deficiency on
   mitochondrial function. To achieve this, we designed and employed two
   siRNAs, namely si‐Ndufs3‐1 and si‐Ndufs3‐2, to knock down Ndufs3. The
   efficiency of this knockdown was confirmed through qPCR conducted in
   brown adipocytes (Figure [163]S8A, Supporting Information).

   In comparison to cells treated with siNC, brown adipocytes treated with
   siNdufs3 displayed markedly fragmented mitochondria. These mitochondria
   did not exhibit the typical elongated rod shape but appeared swollen
   and spherical (Figure [164] 9A). Furthermore, the expression levels of
   genes associated with mitochondrial dynamics exhibited a significant
   decrease in Ndufs3‐knockdown cells (Figure [165]9B). Consistently,
   Ndufs3 knockdown resulted in an elevated production of mtROS, as
   evidenced by MitoSOX staining, a critical method used for assessing
   mitochondrial oxidative stress levels (Figure [166]9C). Using the
   cyanine dye JC‐1 to measure membrane potential, we further observed a
   significant reduction in membrane potential in adipocytes upon Ndufs3
   knockdown (Figure [167]9D).

Figure 9.

   Figure 9
   [168]Open in a new tab

   SRSF1 regulates mitochondrial homeostasis through controlling the
   inclusion of Ndufs3 exon 6 in brown adipocytes. A) Primary brown
   adipocytes were transfected with either Ndufs3‐specific siRNAs
   (siNdufs3‐1, siNdufs3‐2) or control siRNA (siNC) for 48 h.
   Representative images of cells stained with Mitotracker green are
   displayed. Enlarged images reveal the presence of fragmented and
   degenerated mitochondria. Nuclei were stained with Hoechst. Scale bar,
   5 µm. B) q‐PCR analysis was performed to assess the expression of genes
   involved in the regulation of mitochondrial dynamics in cells described
   in (A). C) mtROS detection was carried out using MitoSOX staining (red)
   in cells described in (A). Scale bar, 20 µm. The quantification of
   relative fluorescence units (RFU) was shown in the right graph. D)
   Mitochondrial membrane potential was assessed using JC‐1 staining in
   cells described in (A). Green fluorescence indicated low membrane
   potential, while red fluorescence indicated high membrane potential and
   accumulation of JC‐1 in mitochondria. Scale bar, 50 µm. The ratio of
   red/green fluorescence was quantified and shown in the right graph. E)
   Brown adipocytes were transfected with SRSF1‐specific siRNAs
   (siSRSF1‐1, siSRSF1‐2) or siNC for 48 h. Representative images of cells
   stained with Mitotracker green are displayed. Enlarged images reveal
   the presence of fragmented and degenerated mitochondria. F) mtROS
   detection was carried out using MitoSOX staining (red) in cells
   described in (E). Scale bar, 20 µm. The quantification of relative
   fluorescence units (RFU) was shown in the right graph. G) Brown
   adipocytes were co‐transfected with siNC and vector plasmid, siSRSF1
   and vector plasmid, or siSRSF1 and Ndufs3 plasmid for 48 h.
   Representative images of Mitotracker staining (green, top panel) and
   MitoSOX staining (red, bottom panel) were shown. Scale bars: 10 µm (top
   panel), 20 µm (bottom panel). H) Mitochondrial membrane potential was
   detected using JC‐1 staining in the cells described in panel (G). I)
   Quantification of RFU for the experiment described in (G) was
   illustrated on the top graph. Quantification of the red/green
   fluorescence ratio for the experiment described in (H) was presented on
   the bottom graph. ^* p < 0.05, ^** p < 0.01, ^*** p,0.001. Data
   represent the mean ± SEM.

   Similarly, the knockdown of SRSF1 in primary brown adipocytes also had
   detrimental effects on mitochondrial homeostasis. These effects
   included a significant increase in the presence of fragmented and
   degenerated mitochondria (Figure [169]9E), an elevation in mtROS
   production (Figure [170]9F), and perturbation of mitochondrial membrane
   potential (Figure [171]S8B,C). Notably, these observations appeared to
   be more pronounced compared the effects observed in primary adipocytes
   upon Ndufs3 knockdown, emphasizing the critical roles of both SRSF1 and
   Ndufs3 in maintaining mitochondrial homeostasis and function.

   To investigate deeper into the role of Ndufs3 as a mediator of SRSF1's
   influence on mitochondria within brown adipocyte, we conducted an
   experiment involving the introduction of Ndufs3 plasmids or
   empty‐vector controls into SRSF1‐knockdown cells (Figure [172]S8D,
   Supporting Information). Intriguingly, our observations demonstrated
   that the expression of native Ndufs3 alleviates the adverse
   mitochondrial effects resulting from SRSF1 knockdown. These
   improvements included the reduction of fragmented mitochondria, the
   mitigation of increased mtROS production and the restoration of
   disrupted membrane potential (Figure [173]9G–I). Collectively, these
   findings underscored the crucial role of SRSF1‐regulated splicing
   control on Ndufs3 in maintaining mitochondria integrity and function in
   primary brown adipocytes.

2.11. SRSF1‐Regulated Splicing of Ndufs3 Influences Thermogenesis in Mature
Brown Adipocytes

   We next aimed to investigate the role of SRSF1‐regulated splicing of
   Ndufs3 in the regulation of thermogenesis in mature brown adipocytes.
   To investigate this, we conducted several experiments. First, we
   transiently transfected brown adipocytes with either Ndufs3 siRNA or
   siNC and allowed them to differentiate into mature adipocytes. As shown
   in Figure [174] 10A,B, both si‐Ndufs3‐1 and si‐Ndufs3‐2 significantly
   reduced the expression of thermogenic genes such as Ucp1, Dio2 and
   Prdm16 after brown adipocyte differentiation compared to cells treated
   with si‐NC.

Figure 10.

   Figure 10
   [175]Open in a new tab

   SRSF1‐regulated splicing of Ndufs3 plays a significant role in
   influencing thermogenesis in mature brown adipocytes. A,B) Primary
   brown adipocytes were transiently transfected with Ndufs3 siRNA or
   siNC, followed by induction of differentiation into mature adipocytes
   for the following experiments. Ndufs3 knockdown efficiency was
   confirmed by western blot (A). The expression of thermogenic genes was
   then assessed using q‐PCR in these cells (B). C) Brown adipocytes were
   co‐transfected with siNC and vector plasmid, siSRSF1 and vector
   plasmid, or siSRSF1 and Ndufs3 plasmid respectively, followed by
   induction of differentiation into mature adipocytes. The expression
   levels of genes involved in thermogenesis, mitochondrial OXPHOS, and
   FAO were measured by q‐PCR. The expression level of each gene was
   normalized to β‐Actin as an internal control (n = 3). D) Western blot
   analysis was conducted to assess the protein levels of Srsf1, Ndufs3,
   and Uc1 in the mature brown adipocytes described in panel (C). E) A
   functional model for SRSF1's role in the regulation of Ndufs3 splicing
   and its control over thermogenesis in BAT. Our model demonstrates that
   SRSF1 effectively binds to exon 6 of Ndufs3 pre‐mRNA, thereby promoting
   its inclusion in the final mRNA transcript. Conversely, the deficiency
   of SRSF1 results in impaired splicing of Ndufs3 and subsequently
   reduced levels of Ndufs3 proteins. Consequently, this deficiency
   triggers an increase in mtROS generation and disrupts mitochondrial
   function within brown adipocytes, ultimately compromising the
   thermogenic capacity of BAT. ^* p < 0.05, ^** p < 0.01, ^***p,0.001.
   Data represent the mean ± SEM.

   As expected, SRSF1 knockdown affected the expression of genes such as
   Ndufs3, Ucp1, Dio2, Shdb and Cpt1b in mature brown adipocytes.
   Interestingly, overexpression of Ndufs3 in SRSF1‐knockdown cells
   significantly restored the expression of these genes (Figure [176]10C).
   In addition, western blot analysis further demonstrated that knockdown
   of SRSF1 resulted in significantly reduced levels of Ndufs3 and Ucp1,
   whereas the overexpression of Ndufs3 in SRSF1‐knockdowon cells led to a
   notable increase in the protein levels of Ucp1 without affecting SRSF1
   levels (Figure [177]10D). These findings strong demonstrated that
   SRSF1‐regulated splicing of Ndufs3 influences thermogenesis in mature
   brown adipocytes.

3. Discussion

   Emerging evidence underscores the significance of mutations or altered
   expression patterns in splicing regulators, along with aberrant
   splicing events, as common factors in diet‐induced obesity and
   metabolic dysregulation.^[ [178]^23 ^] Among these regulators, SRSF1
   assumes a pivotal role as a splicing factor that governs AS in numerous
   genes crucial for essential cellular functions.^[ [179]^24 ^] However,
   its contribution to brown adipocyte thermogenesis and white adipocyte
   browning, two critical physiological processes tightly linked to energy
   expenditure, remains largely unexplored.

   In this study, we introduce a transgenic mouse model, which allows us
   to investigate the specific effects of SRSF1 deficiency in mature
   adipocytes. Genetic ablation of SRSF1 results in the whitening of BAT
   and a reduction in thermogenic function, leading to metabolic disorders
   in mice. Furthermore, SRSF1 deficiency impedes the browning of WAT in
   response to cold challenge and β3‐agonist treatment. Importantly, our
   research reveals that SRSF1 deficiency disrupts mitochondrial
   homeostasis, leading to an excessive production of mtROS in BAT through
   the modification of the splicing of Ndufs3 (Figure [180]10E).

   Mitochondria function is vital in maintaining the energy metabolism and
   thermogenic activity within BAT. Its malfunction leads to BAT
   whitening, characterized by the loss of its characteristic
   mitochondrial structure and function.^[ [181]^25 ^] Previous research
   has elucidated that deficiencies in APOO or Trib1 disrupt mitochondria
   structure and hinder oxidative phosphorylation in brown adipocytes,
   ultimately inducing BAT whitening.^[ [182]^26 ^] Similarly, our study
   unveiled noticeable BAT whitening and mitochondrial dysfunction in
   SRSF1 knockout mice. Through snRNA‐seq analysis, we observed that
   mature brown adipocyte subpopulations BA‐1 and BA‐2 predominantly
   occupied the BAT of WT mice, while they were replaced by white
   adipocyte subpopulation (WA) and mature brown adipocyte subpopulation
   BA‐3 within the BAT of KO mice. Although we noted increased adipocyte
   stem and progenitor populations in the BAT of KO mice, adipocyte
   differentiation appears similar between the two groups. Thus, BAT
   whitening in KO mice doesnot appear to result from abnormal adipocyte
   differentiation but rather emerges from mitochondria dysfunction, which
   possibly triggers the transformation of BA‐1 and BA‐2 subpopulations.

   Previous research has investigated NOVA's role in governing the AS of
   JNK1 and JNK2 genes, integral component of the signaling pathway that
   regulates thermogenic gene expression.^[ [183]^27 ^] Additionally,
   another splicing factor, RBM4a, has been shown to oversees the AS of
   PRDM16, a pivotal transcription factor in brown adipogenesis, enhancing
   its impact on the development of brown adipocytes.^[ [184]^28 ^] In
   line with these previous findings, our research highlights the critical
   role of SRSF1 in regulating mitochondrial function and BAT
   thermogenesis in mice. Notably, our investigation reveals that the
   deficiency of SRSF1 results in alterations in the inclusion of
   alternative or constitutive exons in genes related to mitochondria
   function within BAT. We have provided substantial evidence that SRSF1
   plays a critical role in promoting the inclusion of constitutive exon 6
   of Ndufs3 pre‐mRNA. The skipping of exon 6 results in a decrease in the
   production of functional proteins, a critical subunit of mitochondrial
   complex I, leading to mitochondrial structural damage and dysfunction.
   Previous research emphasized SRSF1's role in adipocyte differentiation
   by influencing the AS of key genes such as Rps6kb1 and Pparγ.^[
   [185]^29 ^] Yet, in Adipoq‐Cre‐mediated SRSF1 knockout WAT and BAT
   (Figure [186]S9, Supporting Information), we did not observe abnormal
   isoforms for these two genes. This suggests that SRSF1 likely regulates
   different splicing events to maintain the physiological functions of
   mature adipocytes, differing from its role during adipocyte
   differentiation process.

   D'Angelo et al. conducted an extensive investigation into the role of
   Ndufs3 and presented compelling evidence of how its deletion leads to a
   partial reduction in CI activity.^[ [187]^11b ^] Even a modest decrease
   in CI activity can rapidly results in partial depolarization of the
   mitochondrial membrane, leading to uncontrolled accumulation of mtROS
   and alterations in mitochondrial morphology and dynamics. These changes
   invariably impact overall mitochondrial function.^[ [188]^30 ^]
   Similarly, we observed that direct knockdown of Ndufs3 or the skipping
   of exon 6 in Ndufs3 in the absence of SRSF1 led to a decrease in Ndufs3
   proteins. This reduction disrupted mitochondrial homeostasis, resulting
   in increased mitochondrial fragmentation, excessive accumulation of
   mtROS, and mitochondrial membrane depolarization. Interestingly, SRSF1
   knockdown resulted in more severe impairments in mitochondrial
   functions. Nevertheless, overexpressing Ndufs3 in SRSF1‐knockdown brown
   adipocytes effectively restored mitochondrial functions and increased
   the expression of thermogenic genes. Therefore, our findings underscore
   the vital significance of SRSF1‐regulated inclusion of Ndufs3 exon 6 in
   maintaining mitochondria homeostasis and BAT activity.

   It is well‐established that WAT can undergo a phenomenon referred to as
   “beiging”, wherein it adopts brown adipocyte‐like characteristics in
   response to specific environmental or hormonal cues. This process plays
   a crucial role in thermogenesis, energy expenditure, and glucose
   homeostasis, making it as a promising target for addressing obesity and
   diabetes.^[ [189]^31 ^] Previous studies have suggested that
   mitochondrial dysfunction is linked to a phenotype marked by the
   suppression of WAT browning.^[ [190]^26 , [191]^32 ^] In eWAT of
   SRSF1‐deficient mice, we observed exon 6 skipping in Ndufs3 pre‐mRNA
   along with increased apoptosis‐related proteins, indicating potential
   impairment in WAT functions (Figure [192]S10, Supporting Information).
   It is possible that mitochondria dysfunction plays a crucial role in
   the impaired browning of WAT observed in SRSF1‐deficient mice in
   response to cold exposure or agonist stimulation. However, further
   investigation is needed to uncover the precise mechanism by which SRSF1
   deficiency affects the beiging process within WAT. Nevertheless, our
   findings provided valuable insights into the complex regulatory
   mechanisms governing beige adipocyte formation, highlighting the
   significant role for SRSF1 in this intricate process.

   In conclusion, our study has unveiled a novel role for SRSF1 in the
   control of BAT thermogenesis. Our findings illustrate how SRSF1
   influences the splicing of Ndufs3, a key player in mitochondria
   homeostasis and BAT function. These results contribute significantly to
   our understanding of the molecular intricacies governing adipose tissue
   metabolism and function, providing valuable insights into its
   physiological functions.

4. Experimental Section

Animals

   All experiments were conducted in accordance with the guidelines of the
   Institutional Animal Care and Use Committee of Shanghai Institute of
   Nutrition and Health, Chinese Academy of Sciences. SRSF1 ^flox/flox
   mice were described previously^[ [193]^33 ^] and provided by Dr. Fu XD
   (Westlake University). SRSF1^flox/flox mice were crossed with
   transgenic C57BL/6J mice expressing Cre recombinase driven by the
   adiponectin (Adipoq) promoter, which is active exclusively in adipose
   tissue. Littermates without the Cre gene (SRSF1^flox/flox) were used as
   controls. Only males were used for further experiments, and littermates
   were compared in all studies. Body weight was measured weekly, food
   intake was calculated by measuring daily food consumption. For
   metabolic studies, mice were individually housed in metabolic cages
   (CLAMS‐16, Columbus Instruments), with free access to food and water.
   The oxygen consumption rate was monitored for 48 h. In the CL316, 243
   treatment experiment, the mice were injected intraperitoneally with
   CL316, 243 at a dose of 1 mg·kg^−1 at the 29.5 h mark during the 48 h
   period. The [194]CL316243 continuous injection experiment involved
   daily intraperitoneal injection of [195]CL316243 (1 mg kg^−1) into mice
   daily for 7 days starting at 8 weeks of age, with a follow‐up
   experiment on the 8th day. For tissue collection, mice were sacrificed
   and perfused with ice‐cold PBS. Different adipose tissue depots
   (inguinal WAT, epididymal WAT, and interscapular BAT), and the liver
   were collected, weighed, and immediately frozen in liquid nitrogen.

Blood Biochemical Analysis

   Plasma triglyceride (TGs), total cholesterol (TC), high‐density
   lipoprotein (HDL‐C), and low‐density lipoprotein (LDL‐C) levels were
   enzymatically measured using an automatic analyzer (EnSpire Multimode
   Plate Reader, PerkinElmer, American). The reagents used for the
   analysis were purchased from Nanjing Jiancheng Bioengineering
   Institute, China. Blood glucose levels were measured using an Optium
   Xceed meter (Abbott Diabetes Care).

Glucose and Insulin Tolerance Test

   In the glucose tolerance test experiment, mice were fasted for 16 hours
   and then injected intraperitoneally with D‐glucose (1 g kg^−1). In the
   insulin tolerance test experiment, mice were fasted for 8 h and then
   injected intraperitoneally with human insulin (Novolin) (0.75 U kg^−1).
   Blood glucose levels were measured at indicated times using an Optium
   Xceed meter, from the tail vein.

Histology and Immunohistochemistry

   For H&E staining, iWAT, eWAT, BAT, and liver tissues were fixed in 10%
   neutral‐buffered formalin, then embedded in paraffin, and cut into 7 µm
   sections. HE staining was performed on these sections. Representative
   images were obtained from at least three independent experiments. To
   measure fat cell size, the H&E‐stained sections of at least three
   individual samples in each group were analyzed. The average diameter of
   adipocyte cells was determined using NIH Image J software from the
   National Institutes of Health. For immunohistochemistry, BAT, iWAT, and
   eWAT were fixed in 4% PFA overnight, dehydrated, and embedded in
   paraffin for sectioning. Antibodies for UCP1 (U6382, Sigma) were used
   for immunohistochemical staining. For immunofluorescence staining, the
   mice were anesthetized with isoflurane and perfused with phosphate
   buffered saline (PBS) through the aorta. Afterward, they were perfused
   with chilled 4% paraformaldehyde in phosphate buffer (pH 7.0). Brown
   adipose tissue was dissected, stored overnight at 4 °C in 25% sucrose,
   embedded in OCT compound, and cut at 15–35 µm using a cryostat. The
   tissue sections were then collected on slides. The slides were first
   washed in 0.1 m PBS and then blocked in a solution of 5% normal goat
   serum with 0.1% Triton X‐100, which was prepared in 0.1 m PBS, for a
   duration of 1 h. After that, the slides were incubated overnight at
   4 °C with primary antibodies. The nuclei were stained with DAPI
   (1 µg mL^−1). For BODIPY (D3922, Invitrogen) staining, 10 µm thick
   liver cryostat sections were used. The sections were rinsed with PBS
   and then stained with 1 µg mL^−1 BODIPY at room temperature for 30 min.
   Immunofluorescence images were captured using a Zeiss LSM880NLO
   Confocal Microscope.

Transmission Electron Microscopy Imaging

   For transmission electron microscopy, BAT samples were fixed overnight
   at 4 °C with 2.5% glutaraldehyde (pH 7.4), and then postfixed at room
   temperature for 1 h with 2.0% osmium tetroxide. Thin sections were
   stained with uranyl acetate and lead citrate before being examined
   using a transmission electron microscope (Hitachi H‐7650, Japan). The
   representative images were obtained from three independent experiments.

RNA‐Seq and Data Analysis

   Total RNA isolated from KO and WT mice was used for paired‐end RNA‐Seq
   using an Illumina HiSeq 2000 system, following the manufacturer's
   instructions. Reads mapping and data analysis between the two samples
   were conducted as previously described. The raw sequence data have been
   submitted to the Gene Expression Omnibus under accession number
   PRJNA999070.

Cold‐Stress Experiments

   To induce short periods of cold stimulation, mice were individually
   placed in cages inside a freezer set at 4 °C without food. For longer
   durations of cold stimulation, mice were exposed to 4 °C for 24 h,
   during which they were adequately provided with food. The core body
   temperature was monitored using a rectal probe (TES‐1310, TES
   Electrical Electronic Corp).

Infrared Thermography

   An infrared camera (Magnity Electronics Co., Ltd, China) was utilized
   to measure interscapular surface temperature or entire body area or of
   mice. The thermal imaging was performed on all animals immediately
   after anesthesia. Multiple infrared photographs were taken of each
   animal during the light phase. The infrared analysis software (ThermoX)
   was employed to identify the maximum temperatures in the interscapular
   region from all photos of the same animal. A minimum of four photos
   were captured and analyzed for each animal/group.

Protein Extraction and Western Blot Analysis

   Tissue and cell lysates were prepared using RIPA buffer supplemented
   with a proteinase inhibitor cocktail (Roche), PMSF, and a phosphorylase
   inhibitor (Roche). Immunoblotting was performed using the following
   primary antibodies: SRSF1(MABE163, Merck, Kenilworth); AKT (4691; Cell
   Signaling Technology, Massachusetts); pAKT(S473) (9271, Cell Signaling
   Technology, Massachusetts); PGC‐1α (AF7736, Beyotime, Shanghai); UCP1
   (U6382, Sigma, St Louis); ADRB3 (ab52623; Abcam); NDUFS3 (ab177471;
   Abcam); Caspase‐3 (AC030, Beyotime, Shanghai); BAX (A19684, Abclonal
   Technology, Wuhan); Bcl2 (68103‐1‐Ig, Proteintech, Wuhan); GAPDH
   (A19056, ABclonal, Wuhan); β‐Actin (sc‐47778; Santa Cruz Biotechnology,
   Dallas).

RNA Isolation, RT‐PCR, qPCR, and Mitochondrial DNA Copy Number Quantification

   Total RNAs were prepared using Trizol (Thermo Fisher Scientific,
   Massachusetts) following the manufacturer's instructions. Complementary
   DNA (cDNA) was synthesized by reverse transcription using oligo (dT)
   priming and M‐MLV Reverse Transcriptase (Promega Corporation) according
   to the manufacturer's protocol. The PCR products were analyzed by
   agarose gel electrophoresis and stained with ethidium bromide. Gene
   expression was measured using standard qPCR methods with the SYBR
   Premix Ex Taq kit (YEASEN Technology Company, Shanghai). Analysis was
   conducted on a QuantStudio Real‐Time PCR System (Applied Biosystems).
   The PCR primers used are listed in Table [196]S1 (Supporting
   Information). To quantify the copy number of mitochondrial DNA, total
   DNA was extracted from brown adipose tissues using the Animal tissue
   DNA kit (3 101 050; Simgen, Hangzhou). To determine the copy number of
   mitochondrial DNA (mtDNA), a quantitative real‐time PCR‐based method
   was utilized, as described previously.^[ [197]^34 ^] The mtDNA content
   was quantified by calculating the ratio of the copy number of a
   mitochondrial gene (ND4) to that of a single‐copy nuclear gene (18S
   rRNA). The target mitochondrial DNA copy number was normalized to the
   nuclear gene 18s rRNA. The primers used for Real‐Time PCR are listed in
   Table [198]S1 (Supporting Information).

Isolation of Primary Brown Adipocytes and Adipocyte Differentiation

   The stromal vascular fraction from interscapular BAT of male mice at
   postnatal day 1–2 was isolated.^[ [199]^35 ^] The tissues were minced
   in isolation buffer (Table [200]S3, Supporting Information) and then
   filtered through a 70 µm cell strainer (352 350, Corning FALCON). The
   preadipocytes obtained were cultured in primary culture medium (Table
   [201]S3, Supporting Information). Adipocyte differentiation was
   performed following a standard protocol. Two days after the
   preadipocytes reached confluence (day 0), differentiation was initiated
   by adding the induction medium (Table [202]S3, Supporting Information).
   After 3 days, the medium was replaced with maintenance medium (Table
   [203]S3, Supporting Information). The medium was changed daily until
   day 8.

siRNAs and Plasmids

   Different siRNA oligos were synthesized by Gene Pharma (Shanghai,
   China). Ndufs3 overexpression plasmids were purchased from Genescript
   (Nanjing, China). Minigenes were constructed by amplifying genomic
   sequences spanning exons 5–7 of Ndufs3 gene, which was then cloned into
   the PCDNA3.1 vector. Deletion or add‐in mutant derivatives were created
   based on the minigene plasmids. For transient transfection, siRNA or
   plasmids were transfected into primary adipocytes using either
   Lipofectamine RNAiMAX (Invitrogen, California) or Lipofectamine 2000
   (Invitrogen, California), following the manufacturer's protocol. The
   sequences of the siRNA and plasmids can be found in Tables [204]S1 and
   [205]S2 (Supporting Information).

Detection of Mitochondrial ROS Production in BAT

   Detection of brown adipose tissue mtROS production was conducted
   following the previously described method.^[ [206]^36 ^] Briefly,
   freshly collected brown adipose tissue was immediately sectioned into
   25 µm cryostat sections without fixation. The sections were then
   stained with 5 µm MitoSOX Red (YEASEN, Shanghai, China), a
   mitochondrial superoxide indicator, at 37 °C for 10 min. Subsequently,
   the nucleus was stained with DAPI for 10 min and the samples were
   visualized immediately using a Zeiss LSM880NLO Confocal Microscope.
   Quantification of fluorescence intensity was performed by examining
   five random fields per section and analyzed using ImageJ software.

Mito Tracker Green, Mito SOX Red, and JC‐1 label in Primary Brown Adipocytes

   To assess mitochondrial morphology and content in primary brown
   adipocytes, the cells with 100 nm green‐fluorescing MitoTracker Green
   (YEASEN Technology Company, Shanghai) were loaded for 30 min at 37 °C.
   Afterward, the cells were washed with warm complete culture medium. For
   measuring mitochondrial reactive oxygen species (ROS) levels, the cells
   with 5 µm MitoSOX Red were loaded for 10 min at 37 °C. Additionally,
   the nucleus was stained with Hoechest3342 for 10 min at 37 °C (ShareBio
   Technology Company, Shanghai). To examine mitochondrial membrane
   potential, the cells with 10 µg mL^−1 JC‐1 (YEASEN Technology Company,
   Shanghai) were stained at 37 °C for 15 min in the dark. Subsequently,
   the cells were imaged using the confocal microscope immediately after
   washing them with dilution buffer. The excitation and emission
   wavelengths for each fluorescent dye were selected following the
   manufacturer's instructions. All data were obtained from experiments
   with a minimum of three replicates.

RNA Immunoprecipitation

   In exploring the interaction between SRSF1 protein and Ndufs3 pre‐mRNA,
   RNA immunoprecipitation (RIP) experiments in primary brown adipocytes
   were conducted using a specific antibody against SRSF1 (MABE163, Merck,
   Kenilworth). The experiments were performed following the guidelines
   provided by the manufacturer using the Magna RIP RNA‐Binding Protein
   Immunoprecipitation kit (Merck, Kenilworth). Cell lysates were prepared
   and incubated with beads‐antibody complexes. RNAs were extracted from
   the beads and reverse transcribed as mentioned earlier. The resulting
   cDNA products were then analyzed using gene‐specific primers for Ndufs3
   through RT‐PCR or qPCR.

Nuclei Isolation From Mouse Tissue and snRNA‐seq

   The preparation of nucleus suspensions for experiments was conducted
   under constant ice‐cold conditions. Interscapular brown adipose tissue
   was harvested from 8‐week‐old control and adipose tissue‐specific
   SRSF1‐knockout mice. Fresh tissue samples were cut and lysed in 1 mL of
   nucleus lysis solution (NST) (Table [207]S3, Supporting Information) on
   ice for 7 min. Once complete nucleus lysis was confirmed, 1 mL of ST
   Wash buffer (Table [208]S3, Supporting Information) was added to
   resuspend the nuclei. The resuspension was then filtered through a
   40 µm cell sieve, and the resulting filtrate was transferred to a 15 mL
   centrifuge tube. The cell sieve was washed with an appropriate amount
   of ST Wash buffer, and the wash buffer was combined with the nuclei
   filtrate. To ensure proper resuspension, the nuclei were resuspended.
   Finally, the nuclei were stained with Tissue Blue and counted through
   microscopic examination.

   Nuclei were resuspended in PBS+1% BSA at a concentration of
   700–1200/µL. The cDNA library amplification and DNA library
   construction were carried out using the 10x Genomics Chromium Next GEM
   Single Cell 3ʹ Reagent Kits v3.1 (1 000 268) and the Chromium Single
   Cell 3′/5′ Library Construction Kit (1 000 020), respectively,
   following the manufacturer's instructions. The libraries were sequenced
   on an Illumina Nova 6000 platform using PE150 mode.

snRNA‐seq Data Processing

   To process the data, the Cell Ranger software pipeline (version 3.1.0)
   provided by 10x Genomics was utilized. This pipeline facilitated
   several steps including demultiplexing of cellular barcodes, read
   mapping to the genome and transcriptome using the STAR aligner, and
   down‐sampling of reads to generate normalized aggregate data across
   samples. This resulted in a matrix of gene counts versus cells. BAT
   samples were collected separately from both WT and KO mice, resulting
   in two distinct samples. The reads obtained were aligned to the GRCm38
   reference genome to generate unique molecular identifier (UMI) count
   matrices. The data from these sequenced samples were then aggregated
   using the aggr pipeline in Cell Ranger. Once the feature‐barcode matrix
   was obtained, consisting of 24341 nuclei, downstream analyses were
   performed using the R package Seurat. These analyses included quality
   control assessments, clustering of cells, cell annotation, and
   identification of differentially expressed genes (DEGs). The quality of
   the nuclei was evaluated based on metrics such as total UMI counts per
   nucleus, the number of detected genes per nucleus, and the proportion
   of mitochondrial gene counts. These quality control measures were
   crucial in ensuring the integrity and reliability of the subsequent
   analyses.

   The identification of top variable genes across single cells followed
   the methodology outlined in Macosko et al.^[ [209]^37 ^] This involved
   calculating the average expression and dispersion for each gene, which
   were then categorized into multiple expression‐based bins. To reduce
   the dimensionality of the log‐transformed gene‐barcode matrices of the
   top variable genes, principal component analysis (PCA) was performed.
   Subsequently, cells were clustered using a graph‐based clustering
   approach, and the results were visualized in a 2D space using
   t‐distributed stochastic neighbor embedding (tSNE). To identify
   significantly differentially expressed genes between clusters, a
   likelihood ratio test was employed. This test simultaneously assessed
   changes in mean expression and the percentage of expressed cells. For
   unbiased cell type recognition of scRNA‐seq data and inference of the
   cell of origin for each single cell, the R package SingleR^[ [210]^38
   ^] was utilized, a novel computational method. By applying SingleR, the
   cell types of each single cell were independently determined and
   identify their respective cell of origin. This approach allowed for
   unbiased cell type characterization and classification.

   Differential expression analysis was conducted using the Seurat^[
   [211]^39 ^] package to identify genes that exhibited significant
   changes. For a gene to be considered differentially expressed, it had
   to meet the following criteria: p value <0.05 and |log2foldchange| > 1
   (or |log2foldchange| > 0.58). GO enrichment and KEGG pathway enrichment
   analysis of DEGs were performed using R and the hypergeometric
   distribution.

GSVA Enrichment Analysis

   To perform pathway enrichment analysis, the background gene set files
   from the KEGG database ([212]https://www.kegg.jp/) was initially
   obtained. These files were downloaded and organized using the GSEABase
   package (v1.44.0). Next, the signal pathway activity values for
   individual cells using the GSVA package (v1.30.0) were calculated. This
   enabled us to assess the activity levels of various signaling pathways
   within each cell. To compare the differences in signal pathway activity
   among different groups, the LIMMA software package (v3.38.3) was
   employed. This package facilitated the statistical analysis and
   calculation of differential pathway activity. By comparing the activity
   levels of specific pathways between different groups, significant
   differences in pathway activation were successfully identified.

Cell Development Trajectory

   To infer the trajectory of cell differentiation from the Seurat object,
   the Monocle2 (v2.9.0) package was utilized. This package enables to
   conduct pseudotime analysis on the cells. Initially, the Seurat object
   to a CellDataSet object using the importCDS function of the Monocle2
   package was converted. Subsequently, the ordering genes (qval < 0.01)
   were identified that exhibited differential expression among cell
   clusters. This was achieved by employing the differential GeneTest
   function of the Monocle2 package, which utilizes a negative binomial
   generalized linear model to assess gene expression changes. The
   reduceDimension function of the Monocle2 package was then employed to
   reduce the dimensionality of the CellDataSet object. This function
   employs unsupervised learning algorithms, such as independent component
   analysis or non‐negative matrix factorization, to identify latent
   factors that capture the primary sources of variation in the data.
   Finally, the differentiation trajectory of the cells was inferred using
   the orderCells function of the Monocle2 package. This function utilizes
   a minimum spanning tree algorithm to connect cells based on their
   similarity in the reduced dimensional space. Additionally, it assigns a
   pseudotime value to each cell based on its position along the
   trajectory.

Data Availability

   The snRNA‐seq data sets have been deposited in the GEO database under
   the accession number [213]GSE238009.

Statistical Analyses

   Statistical analysis was performed using GraphPad Prism 8.0 software.
   The histograms display data from at least three independent experiments
   and were presented as the mean ± SEM. The sample size (n) for each
   statistical analysis is specified in the figure legend. Statistical
   significance between groups was determined using the two‐tailed
   unpaired Student's t‐test. Differences between values were considered
   statistically significant when ^* p < 0.05, ^**p < 0.01, and ^*** p <
   0.001.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   N.Y. and L.S. contributed equally to this work and are co‐first
   authors. Conception and design were performed by N.Y. and Y.F. Data
   acquisition and analysis were performed by N.Y., L.S., Q.P., R.S.,
   Z.W., Z.X., X.Y., and Y.F.. Drafting of the manuscript was performed by
   N.Y. and Y.F..

Supporting information

   Supporting Information
   [214]ADVS-11-2306871-s001.pdf^ (1.2MB, pdf)

Acknowledgements