A Primate-Only Long Noncoding RNA LINC01021 May Promote Cell and Organism Aging by Destabilizing RBMX Through DAZAP1

LncRNA sequences were obtained from the NONCODE database (version 6) for multiple representative species spanning primates, mammals, vertebrates, and invertebrates (Zhao et al. 2021). NONCODE annotates lncRNAs based on transcriptomic evidence, transcript structure, length, and computational assessment of coding potential (Zhao et al. 2021). LINC01021 is annotated in NONCODE as transcript NONHSAT100800.2, comprising 1,112 nt across five exons. Additionally, the transcript exhibits a negative CNCI score (−0.1515520), supporting its low protein‐coding potential and corroborating its annotation as a lncRNA in the NONCODE database. Human tissue‐level RNA‐seq expression data were downloaded from the Genotype‐Tissue Expression (GTEx) project (Consortium 2013).

To evaluate evolutionary conservation of lncRNAs across species, all species were subjected to exhaustive pairwise sequence comparisons. For each species pair, nucleotide‐level alignments were performed using BLASTN. For each query lncRNA, significant alignments were retained based on sequence similarity, alignment length, and coverage thresholds to ensure biologically meaningful matches. For each query transcript, only the highest‐scoring alignment in the target species was retained. Reciprocal best hit (RBH) analysis was then performed to define putative orthologous lncRNA pairs. A lncRNA pair was considered conserved between two species if each transcript represented the best hit of the other in reciprocal searches. The number of RBH‐supported lncRNA pairs was summarized for all species combinations to characterize the global conservation landscape. LncRNAs were defined as primate‐conserved if they showed reciprocal best‐hit support among primate species but lacked detectable RBH relationships in non‐primate species included in this study. These primate‐specific conserved lncRNAs were subsequently used for downstream aging‐associated expression analyses.

Human tissue RNA‐seq expression data were obtained from the GTEx project. Genomic coordinates of primate‐conserved lncRNAs were intersected with GTEx gene annotations to identify corresponding transcripts, requiring at least 90% reciprocal overlap to ensure high‐confidence mapping. Only lncRNAs satisfying this criterion were retained for downstream analyses. For each tissue, samples with available age information were included. Age was modeled as a continuous variable, and sex was incorporated as a covariate when applicable. Lowly expressed lncRNAs were excluded before modeling to enhance statistical stability. To characterize age‐dependent expression dynamics, generalized additive models (GAMs) were fitted independently for each lncRNA within each tissue according to the following model:where s(Age) represents a penalized regression spline capturing non‐linear age effects, and Sex was included when applicable. The statistical significance of the smooth age term was evaluated for each lncRNA. Multiple testing correction was performed using the Benjamini–Hochberg procedure. LncRNAs with significant age‐dependent effects after correction were defined as aging‐associated lncRNAs. Expression i β 0 β 1 Sex i ε i = + + + s Age i

Human embryonic lung fibroblasts (HELFs) were obtained from the Kunming Cell Bank of Type Culture Collection (Kunming, China). Primary human dermal fibroblasts (HDFs) were isolated from circumcised foreskins of healthy male donors (aged 5–20 years), as previously described (Xiao et al. 2022). Cells were cultured in Dulbecco's modified Eagle medium (DMEM; C11995500BT, Gibco, MD, USA) supplemented with 10% fetal bovine serum (FBS; RYS‐F22‐05, Royacel Biotechnology, Lanzhou, China) and 1% penicillin–streptomycin (15140–122, Gibco, MD, USA) at 37°C in a humidified incubator with 5% CO₂. Routine monitoring for morphology and contamination was performed.

For siRNA or plasmid transfection, cells were seeded in 6‐well plates (3–5 × 10⁵ cells per well) and transfected at ~80%–90% confluence using Lipofectamine 2000 (11668019, Thermo Fisher Scientific, MA, USA) according to the manufacturer's protocol. For siRNA transfection, 100 pmol of siRNA was used per well, while 2 μg of plasmid DNA was used for plasmid transfection. Medium was replaced with fresh DMEM 6 h after transfection.

Cells transfected with the LINC01021‐overexpression (LINC01021‐OE) plasmid were harvested 6 d post‐transfection, while other conditions were collected after 3 d. siRNA sequences are listed in Table S1.

RNA sequencing (RNA‐seq) datasets (accession number: GSE63577↗) were obtained from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/↗). These datasets include transcriptomic profiles of young and senescent fibroblast cell lines, including human foreskin fibroblasts (HFF), BJ foreskin fibroblasts (BJ), IMR‐90 fetal lung fibroblasts (IMR90), and MRC‐5 lung fibroblasts (MRC5). Differential expression analysis was conducted in RStudio (R 4.4.3) using the DESeq2 package. Genes significantly upregulated in senescent cells were defined as those with an adjusted p value < 0.05 and log₂ Fold Change (log₂FC) > 0.

Senescent cells were induced by three methods: ionizing radiation (IR), doxorubicin (DOX) treatment, and replicative exhaustion, as previously described (Y. He et al. 2020). Low‐passage HELF and HDF fibroblasts (< 25 passages) were used as non‐senescent controls or for senescence induction. IR‐induced senescence was established by exposing cells at ~70% confluence to 15 Gy of ionizing radiation using a small animal radiation research platform (Xstrahl Inc., Camberley, UK). Cells were subcultured at a 1:3 ratio 3 d after irradiation, with full senescence typically observed ~7 d later. DOX‐induced senescence was established by treating cells with 250 nM DOX (HY‐15142, MCE, Shanghai, China) for 24 h. After treatment, the medium was replaced with standard DMEM, and cells were cultured for an additional 7 d. Replicative senescence was induced by serially passaging HELF cells until cell division ceased, typically after approximately 35 passages.

SA‐β‐Gal staining was performed using a Senescence β‐Galactosidase Staining Kit (C0602, Beyotime Biotechnology, Shanghai, China) following the manufacturer's instructions. Briefly, cells were seeded in six‐well plates and washed with PBS, fixed at room temperature for 7 min, and then incubated with the staining solution overnight at 37°C. The stained cells were imaged using a microscope (SMZ‐171‐BLED, Motic, Xiamen, China). Blue‐stained cells were considered senescent.

Total RNA was isolated from cultured cells using TRIzol reagent (15596018, Invitrogen, CA, USA) according to the manufacturer's instructions. RNA concentration and purity were assessed spectrophotometrically. Equal amounts of RNA were reverse‐transcribed into cDNA using the HP All‐in‐One qRT Master Mix II Kit (RT203, YoungGen, Kunming, China). Quantitative PCR (qPCR) was performed using 2 × TSINGKE Master qPCR Mix (SYBR Green I) (TSE201, TSINGKE, Beijing, China) on a CFX Connect Real‐Time PCR System (Bio‐Rad, CA, USA), with GAPDH serving as the internal control. Relative gene expression levels were calculated using the 2^^−ΔΔCt method. Primer sequences used in this study are provided in Table S2.

Cell counts were determined manually using a hemocytometer under a light microscope. Cell suspensions were mixed and diluted to an appropriate concentration for counting. An aliquot of the diluted sample was loaded onto a hemocytometer, and cells were counted in duplicate at 100× magnification. The average number of cells per grid was used to calculate the final concentration using the standard hemocytometer formula.

Cell proliferation was assessed using the Cell‐Light EdU Apollo567 In Vitro Kit (C10310↗ ‐1, RiboBio, Guangzhou, China) according to the manufacturer's instructions. Cells were seeded into 96‐well plates at a density of 3000 cells per well and allowed to adhere overnight. The cells were then incubated with the EdU working solution for 2 h, fixed with 4% paraformaldehyde for 15 min at room temperature, and permeabilized with 0.5% Triton X‐100. EdU incorporation was detected by staining with Apollo567, and nuclei were counterstained with DAPI. Fluorescence images were acquired using a fluorescence microscope.

Cells were washed twice with cold 1 × phosphate‐buffered saline (PBS; B540627‐0500, Sangon Biotech, Shanghai, China) and lysed on ice in RIPA buffer (BP‐115DG, Boston BioProducts, MA, USA) supplemented with PMSF (ST507, Beyotime Biotechnology, Shanghai, China) for 30 min. The lysates were sonicated at 35 W for 10 s, followed by centrifugation at 13,000 × g for 15 min at 4°C to remove insoluble debris. Protein concentrations were determined using a BCA Protein Assay Kit (P0010, Beyotime Biotechnology, Shanghai, China). Equal amounts of protein were mixed with 5 × Omni‐Easy Instant SDS‐PAGE Sample Buffer (LT101, EpiZyme, Shanghai, China), denatured at 95°C for 5–10 min, and separated by SDS‐PAGE on 12% polyacrylamide gels. Proteins were transferred onto 0.22‐μm PVDF membranes (Millipore, Bedford, MA, USA) using a wet transfer system (Bio‐Rad, CA, USA). Membranes were blocked with 5% (w/v) skim milk in TBS containing 0.1% Tween‐20 (TBS‐T) for 2 h at room temperature and incubated with primary antibodies overnight at 4°C. After washing with TBS‐T, membranes were incubated with HRP‐conjugated secondary antibodies for 1 h at room temperature. Protein bands were visualized using the SCG‐W3000 Plus chemiluminescence imaging system (Servicebio, Wuhan, China) and quantified with ImageJ software (NIH, MD, USA). All antibodies used in this study are listed in Table. S3

To generate overexpression plasmids, the full‐length sequences of LINC01021, RPL23A, SKP1, EPB41L2, and DAZAP1 were individually cloned into the pCDH‐CMV‐MCS‐EF1‐puro‐3×FLAG‐3×HA vector.

RNA was transcribed in vitro using T7 RNA polymerase (K102A01, Perfect mRNA, Hangzhou, China) and labeled at the 3′‐end with biotin using the Pierce RNA 3′ End Desthiobiotinylation Kit (20163, Thermo Fisher Scientific, MA, USA), according to the manufacturer's protocol. RNA pull‐down assays were conducted as previously described (P. Wang et al. 2014). Briefly, 50 pmol of biotin‐labeled RNA was refolded in RNA structure buffer (10 mM Tris, pH 7.0; 100 mM KCl; 10 mM MgCl₂) by heating at 85°C for 5 min and snap‐cooling on ice. RNA‐protein complexes were captured using the Pierce Magnetic RNA‐Protein Pull‐Down Kit (20164, Thermo Fisher Scientific, MA, USA), washed, eluted, and denatured in 5 × SDS loading buffer. The recovered proteins were separated by gradient SDS‐PAGE and analyzed by MS for protein identification.

LINC01021‐OE HELFs were crosslinked by incubating with 1% formaldehyde for 5 min at 37°C, followed by quenching with 0.125 M glycine for 10 min at room temperature. Cells were then washed twice with ice‐cold PBS containing PMSF and lysed using SDS lysis buffer from the ChIP Assay Kit (P2078, Beyotime Biotechnology, Shanghai, China). Chromatin was fragmented using a Bioruptor Plus sonicator (Diagenode, Seraing, Belgium) at 40 W for 30 cycles (10 s on/10 s off) at 4°C. The sheared chromatin extracts were precleared with Protein A/G agarose beads pre‐blocked with salmon sperm DNA for 30 min at 4°C. After removing 2% of the input as a control, the lysates were incubated overnight at 4°C with either an anti‐RBMX antibody (F1965, Selleckchem, TX, USA) or a control IgG (A7016, Beyotime Biotechnology, Shanghai, China). Immune complexes were captured with Protein A/G agarose beads for 2 h at 4°C and washed sequentially with low‐salt, high‐salt, LiCl, and TE buffers. DNA–protein complexes were eluted with 1% SDS / 0.1 M NaHCO₃ and decrosslinked by incubation with 0.2 M NaCl at 65°C for 4 h. The samples were treated with proteinase K at 45°C for 1 h, and the purified DNA was isolated using a commercial DNA purification kit (D6492, OMEGA BIO‐TEK, GA, USA). Quantified DNA was analyzed by qPCR, with primer sequences provided in Table S2.

Protein stability was assessed as previously described (Q. Zhang et al. 2020). Cells transfected with the indicated plasmids were treated with 50 μg/mL cycloheximide (CHX; GC17198, GLPBIO, CA, USA) at the specified time points. At each time point, cells were harvested, lysed in RIPA buffer with protease inhibitors, and denatured in SDS loading buffer at 95°C for 5 min. Protein levels were analyzed by immunoblotting with the specified antibodies.

Cells transfected with the indicated plasmids were treated with 10 μg/mL Actinomycin D (Act D; GC16866, GLPBIO, CA, USA) to inhibit transcription. Total RNA was extracted at the indicated time points, and RNA stability was assessed by RT‐qPCR using specific primers. Relative RNA levels at each time point were normalized to the 0 h time point and plotted to evaluate RNA decay.

HELFs were transfected with the LINC01021‐OE plasmid for 6 d to achieve robust overexpression. Cells were then treated with 1 μM MG132 (HY‐13259, MCE, Shanghai, China), 10 μM chloroquine (HY‐17589AR, MCE, Shanghai, China), or 5 μM Q‐VD‐OPh (HY‐12305, MCE, Shanghai, China) for 12 h before being harvested for Western blot analysis.

Transgenic mice with a knock‐in (KI) allele for LINC01021‐OE were obtained from Shanghai Model Organisms Center Inc. (Shanghai, China). These mice were generated by CRISPR/Cas9‐mediated homologous recombination, inserting a CAG‐LINC01021‐polyA (PA) cassette into the Rosa26 locus on a C57BL/6J background. The CAG promoter drives constitutive and ubiquitous expression of LINC01021 across tissues, independent of endogenous regulatory elements (Hasegawa et al. 2013). Briefly, Cas9 mRNA and guide RNA (gRNA) were synthesized by in vitro transcription, and a donor vector containing a 3.3 kb 5′ homology arm, the CAG‐LINC01021‐PA cassette, and a 3.3 kb 3′ homology arm was constructed via In‐Fusion cloning. Cas9 mRNA, gRNA, and donor vector were microinjected into C57BL/6J zygotes to generate F0 founders, which were confirmed by PCR and Sanger sequencing. These founders were backcrossed to wild‐type C57BL/6J mice to obtain heterozygous F1 progeny. All mice were maintained under specific pathogen‐free (SPF) conditions at the Kunming Institute of Zoology, Chinese Academy of Sciences. Animal experiments were approved by the Animal Ethics Committee of the Kunming Institute of Zoology, Chinese Academy of Sciences (IACUC‐RE‐2025‐08‐008) and conducted following the Guide for the Care and Use of Laboratory Animals.

Mouse FI was assessed using a validated clinical deficit accumulation approach, as previously described (Feridooni et al. 2015). The FI comprised 23 health deficits spanning multiple physiological systems, including integument, musculoskeletal, auditory, ocular, digestive, urogenital, and body weight (see Table S4 for detailed scoring criteria). Each deficit was scored as 0 (absent), 0.5 (mild), or 1 (severe) according to predefined criteria. Body weight was recorded as a continuous numerical variable for each mouse. Forelimb grip strength was measured three times using a grip strength meter, and the mean value was used for analysis. These measurements were reported as raw physiological readouts and were not converted into standard deviation‐based categorical scores. The FI for each mouse was calculated by summing the individual deficit scores and dividing by the total number of assessed items (23), yielding a final FI value ranging from 0 (no deficits) to 1 (all deficits present).

In vivo senescence imaging was performed using the Senescence‐Tracker Near‐Infrared (NIR) probe (C0603, Beyotime Biotechnology, Shanghai, China), as previously described (Hu et al. 2023). The probe was dissolved in DMSO to generate a 10 mM stock solution and diluted in 1% DMSO, 2% Tween‐80, and 97% sterile saline to a final concentration of 5 μM. Mice received a single intravenous tail‐vein injection of 100 μL probe solution. Whole‐body fluorescence images were acquired 24 h post‐injection using an IVIS Lumina 3 imaging system (PerkinElmer, CT, USA) under identical acquisition settings. Regions of interest (ROIs) were manually defined over each mouse, and fluorescence intensity was quantified as total radiant efficiency using Living Image Software v4.4. All imaging and analysis parameters were kept constant across experimental groups.

Data are presented as mean ± standard error of the mean (SEM). Comparisons between two groups were performed using an unpaired two‐tailed Student's t‐test, with Welch's correction applied when variances were unequal. Comparisons among more than two groups were conducted using one‐way analysis of variance (ANOVA). A p‐value < 0.05 was considered statistically significant (* p < 0.05, ** p < 0.01, *** p < 0.001; ns, not significant).

LncRNAs constitute a major class of regulatory transcripts in eukaryotic genomes but are characterized by rapid evolutionary turnover and pronounced lineage specificity (Mattick et al. 2023; Statello et al. 2021). To systematically assess the evolutionary distribution of lncRNAs, we compiled annotated lncRNA catalogs from the NONCODE database across multiple representative species. As shown in Figure 1A, the number of annotated lncRNAs varies dramatically among species, ranging from only 52 in yeast to more than 96,000 in humans. Notably, substantial discrepancies were also observed among closely related primates: chimpanzee, gorilla, and orangutan each harbor approximately 6,000–15,000 annotated lncRNAs, markedly fewer than reported in humans. These differences are unlikely to reflect true biological absence, but instead largely arise from variation in sequencing depth, annotation strategies, and research intensity, underscoring the limitations of absolute lncRNA counts for inferring functional relevance across species.

To move beyond catalog size and directly evaluate evolutionary relationships, we performed pairwise comparisons of lncRNAs across species to assess sequence conservation (Figure 1B). Consistent with prior observations, lncRNAs exhibited strong species specificity, with only a small fraction shared between any two species. Importantly, when human lncRNAs were used as the reference, a substantially larger fraction showed detectable conservation within primate species compared with non‐primate mammals. Based on this primate‐skewed conservation pattern, we identified a set of 7098 human lncRNAs that are conserved in at least one other primate species but absent from non‐primate species, hereafter referred to as primate‐conserved lncRNAs (Figure 1C). These transcripts combine evolutionary constraints with lineage restriction, suggesting that they may encode regulatory functions preferentially relevant to primate biology.

To investigate whether primate‐conserved lncRNAs are implicated in aging, we analyzed their age‐associated expression patterns across multiple human tissues using GAMs. Across all tissues examined, a subset of primate‐conserved lncRNAs exhibited significant associations with chronological age (Figure 1D), with the colon displaying the largest number of aging‐associated candidates. Integrative analysis further revealed that many of these lncRNAs showed age‐dependent expression changes across multiple tissues, with some exhibiting significant associations in more than 16 tissues (Figure S1A), indicating potential roles in systemic or organism‐wide aging regulation.

To define a high‐confidence set of aging‐associated candidates, we applied more stringent criteria requiring both statistical significance (FDR < 0.05) and substantial age‐dependent expression changes (log₂FC > 1). This analysis identified a core cohort of primate‐conserved lncRNAs with robust and recurrent age‐associated expression across tissues (Figure 1E), such as LINC01021, CTC‐575 l10.1, CTA‐150C2.13, RP11‐305F18.1, RP11‐700 N1.1, RP4‐541C22.5, and RP11‐291B21.2. Among these, LINC01021 emerged as the most broadly associated transcript, exhibiting significant age‐related expression changes across seven tissues, including the adrenal gland, cervix uteri, and colon (Figure 1E). In addition to LINC01021, several other primate‐conserved lncRNAs, such as CTC‐575 l10.1, CTA‐150C2.13, and RP11‐305F18.1, also displayed consistent age‐associated expression patterns across multiple tissues (Figure 1E). Together, these findings position LINC01021 as a representative exemplar of a broader class of aging‐associated primate‐conserved lncRNAs rather than an isolated outlier. Based on its breadth of age association, recurrent induction across independent aging contexts, and suitability for functional interrogation, LINC01021 was selected for subsequent mechanistic investigation.

Following the identification of primate‐conserved lncRNAs with recurrent age‐associated expression across human tissues, we next examined whether LINC01021, as a representative candidate, exhibits consistent induction in established human cellular models of senescence. We first analyzed a publicly available RNA‐seq dataset comprising four human fibroblast strains (HFF, BJ, IMR90, and MRC5) subjected to either replicative or stress‐induced senescence (Marthandan et al. 2016). Across these well‐established models, which recapitulate core features of cellular aging, LINC01021 emerged as a robust aging‐associated primate‐conserved lncRNA that was reproducibly upregulated across all four fibroblast strains under both senescence paradigms (Figure 2A–D).

To experimentally validate these observations, we established three independent senescence models in HELFs, including IR‐induced senescence, DOX‐induced senescence, and replicative senescence induced by serial passaging. Successful induction of senescence was confirmed by a marked increase in SA‐β‐Gal positive cells (Figure 2E–H) and significant upregulation of canonical senescence regulators CDKN2A and CDKN1A at the mRNA level (Figure 2I–N). Consistent with the bioinformatic analyses, qRT‐PCR revealed a robust and highly reproducible increase in LINC01021 expression across all three senescence models (Figure 2O–Q). Notably, the magnitude of LINC01021 induction was comparable to, and in some conditions exceeded that of CDKN2A and CDKN1A, further supporting its tight association with the senescent state.

Given the consistent and reproducible induction of LINC01021 across multiple cellular senescence models, we next investigated whether LINC01021 functionally contributes to the establishment of senescent phenotypes in human fibroblasts. Gain‐of‐function experiments were first performed in HELFs, in which overexpression of LINC01021 was confirmed by qRT‐PCR (Figure 3A). Compared with control cells, LINC01021‐OE HELFs exhibited a significantly increased proportion of SA‐β‐Gal positive cells, a reduced fraction of EdU‐positive proliferating cells, and an overall impairment of proliferative capacity, as reflected by slower cell growth rates (Figure 3B–F). Consistent phenotypic effects were also observed in HDFs, in which LINC01021‐OE similarly increased SA‐β‐Gal positivity and reduced EdU incorporation (Figure S2A–D), indicating that the pro‐senescent effects of LINC01021 are not restricted to a single fibroblast lineage.

At the molecular level, LINC01021 overexpression induced the expression of core senescence effectors, including CDKN2A and CDKN1A, and promoted the transcription of multiple senescence‐associated secretory phenotype (SASP) factors, such as IL1β, IL6, and CXCL2 (Figure 3G–K). Concordantly, protein levels of P16 and P21 were markedly increased following LINC01021 overexpression (Figure 3L–N). Notably, P53 protein abundance was also elevated under these conditions (Figure 3L,O), a pattern that contrasts with previous reports describing a suppressive role of LINC01021 on P53 activity in cancer cells, highlighting potential context‐dependent regulation (Han et al. 2021).

To complement these gain‐of‐function analyses, loss‐of‐function experiments were performed across three independent senescence models. Knockdown of LINC01021 consistently reduced the expression of CDKN2A and CDKN1A and attenuated the induction of SASP‐associated genes (Figures 3P–S and S2E–R). At the protein level, P16 and P21 were correspondingly decreased across all senescence models upon LINC01021 depletion, whereas P53 protein levels decreased upon knockdown and increased upon overexpression, remaining positively correlated with LINC01021 expression (Figures 3T–W and S2S–Z). Together, these gain‐ and loss‐of‐function analyses demonstrate that LINC01021 actively modulates key cellular senescence features in human fibroblasts and reveal a context‐dependent relationship between LINC01021 and P53 regulation that differs from its previously reported role in cancer cells (Kaller et al. 2024; Li et al. 2017).

To explore the molecular mechanism underlying LINC01021‐mediated senescence regulation, particularly its impact on P53‐associated pathways, we first examined the subcellular localization of LINC01021. Subcellular fractionation analyses revealed that LINC01021 is predominantly enriched in the nucleus (Figure 4A), prompting us to focus on nuclear‐associated proteins in subsequent mechanistic investigations. We therefore performed RNA pull‐down coupled with MS to identify LINC01021‐interacting proteins, which yielded 8 candidate binding partners with annotated nuclear localization (Figure 4B,C).

To assess the functional relevance of these candidates in LINC01021‐associated senescence, we examined their expression changes in LINC01021‐OE HELFs. Among the 8 candidates, the mRNA levels of genes encoding 3 candidate proteins (NUDT21, DAZAP1, and RBMX) were negatively correlated with LINC01021 expression, 3 (RPL23A, EPB41L2, and SKP1) were positively correlated, and 2 (FBL and DKC1) showed no significant changes (Figure 4D). To functionally interrogate their roles, negatively correlated candidates were individually depleted, whereas positively correlated candidates were ectopically expressed in HELFs. Notably, knockdown of RBMX, but not NUDT21 or DAZAP1, resulted in robust induction of the senescence regulators CDKN2A and CDKN1A, along with multiple SASP genes, including CXCL1, CXCL3, IGFBP5, MMP1, and CCL2 (Figures 4E and S3A,B). In contrast, overexpression of RPL23A, EPB41L2, or SKP1 failed to elicit comparable senescence‐associated transcriptional responses (Figure S3C–E) indicating RBMX as the primary functional mediator among the identified LINC01021‐interacting candidates.

Consistent with these transcriptional changes, RBMX knockdown markedly increased the proportion of SA‐β‐Gal‐positive HELFs and elevated protein levels of the senescence effectors P16 and P21 (Figure 4F–K). Notably, RBMX knockdown also led to an increase in P53 protein (Figure 4H,L), closely mirroring the molecular phenotype observed upon LINC01021 overexpression. To assess P53 dependency, we knocked down p53 via siRNA in LINC01021‐overexpressing cells and examined P53/P21 pathway activity. p53 depletion markedly attenuated LINC01021‐induced P21 activation (Figure S3F,G) indicating that LINC01021 promotes senescence at least in part through P53‐dependent signaling downstream of RBMX. Together, these findings position RBMX as a key effector downstream of LINC01021 that mediates senescence regulation via P53, and underscore the essential role of P53 in this pathway.

Given that RBMX functions as a key downstream mediator of LINC01021‐induced cellular senescence, we next investigated how LINC01021 regulates RBMX abundance. AlphaFold‐based structural modeling predicted a putative three‐dimensional association between LINC01021 and RBMX (Figure 5A); however, ChIP‐qPCR using an anti‐RBMX antibody revealed only low but reproducible enrichment of LINC01021 (Figures 4B and 5B), suggesting an indirect chromatin‐associated relationship rather than a stable RNA–protein interaction. Consistent with a regulatory effect, LINC01021 overexpression in HELFs dramatically reduced RBMX protein levels (Figure 5C,D) and shortened the RBMX protein half‐life (Figure 5E), indicating that LINC01021 negatively regulates RBMX abundance.

To assess whether RBMX destabilization involves proteolytic pathways, we examined the major mechanisms of protein degradation. Pharmacological inhibition of the proteasome using MG132 marginally restored RBMX protein levels in LINC01021‐OE cells (Figure 5F,G), whereas inhibition of lysosomal degradation or caspase‐dependent proteolysis had no detectable effect (Figures 5H,I and S4A,B). SKP1, a core adaptor component of SCF (SKP1‐CUL1‐F‐box) E3 ubiquitin ligase complexes that mediate proteasomal protein turnover (Zeng et al. 2025), was identified among candidate LINC01021‐interacting proteins in the RNA pull‐down assay; however, SKP1 knockdown failed to rescue RBMX protein levels (Figures 5J and S4C), indicating that proteasome‐mediated degradation contributes weakly and in an SKP1‐independent manner. In parallel, qPCR analysis revealed that LINC01021 overexpression significantly reduced RBMX mRNA levels (Figure 4D), prompting us to examine post‐transcriptional regulation.

Act D chase experiments demonstrated that LINC01021 markedly accelerated RBMX mRNA decay (Figure 5K), indicating a substantial contribution of mRNA destabilization to RBMX downregulation. Among proteins identified in the RNA pull‐down screen, the RNA‐binding protein DAZAP1, previously implicated in mRNA stability control (Qiu, Li, et al. 2025; Wang et al. 2021; Zhang et al. 2024), was significantly reduced at both mRNA and protein levels upon LINC01021 overexpression (Figure 4D; Figures 5L and S4D). Functionally, DAZAP1 depletion decreased RBMX protein and mRNA levels and accelerated RBMX transcript decay (Figures 5M–O and S4E,F), whereas DAZAP1 overexpression robustly increased RBMX protein abundance (Figures 5P and S4G,H). Together, these findings indicate that LINC01021 suppresses RBMX expression predominantly through a DAZAP1‐dependent post‐transcriptional mechanism governing RBMX mRNA stability, with a limited contribution from proteasome‐mediated protein turnover.

To assess whether ectopic expression of the primate‐specific lncRNA LINC01021 is sufficient to elicit aging‐associated phenotypes in vivo, we generated a humanized LINC01021 KI mouse model using CRISPR‐Cas9. Correct transgene integration was confirmed by PCR genotyping, and LINC01021 expression in the spleen was verified by qRT‐PCR (Figures 6A,B and S5A), as rodents lack an endogenous ortholog of this primate‐specific lncRNA. Body weight measurements revealed no significant differences between KI and WT mice at the examined ages, indicating that overall growth was not significantly affected (Figures 6C and S5B). To further quantify senescence burden in vivo, we applied near‐infrared senescence imaging using the activatable near‐infrared β‐galactosidase probe, a β‐galactosidase‐responsive fluorescent sensor that emits in the NIR spectrum upon enzymatic cleavage by SA‐β‐Gal, enabling real‐time detection of senescent cells in deep tissues (Hu et al. 2023). This analysis revealed a significantly elevated senescence signal in young adult KI/KI mice compared with WT controls, whereas aged KI mice exhibited a trend toward increased signal intensity that did not reach statistical significance (Figures 6D,E and S5C,D), suggesting that LINC01021 expression may contribute to the onset of cellular senescence in vivo.

Consistent with increased senescence burden, comprehensive phenotypic analyses uncovered signs of accelerated functional decline in LINC01021‐expressing mice. Frailty indices increased more steeply in young KI/KI animals relative to WT littermates, while aged transgenic mice displayed a similar upward trend without statistical significance, likely reflecting increased inter‐individual variability (Figures 6F and S5E). In parallel, both young and aged KI mice showed a tendency toward reduced whisker number compared with WT controls (Figure S5F,G). At the molecular level, both heterozygous (KI/WT) and homozygous (KI/KI) young mice displayed marked upregulation of the senescence markers P16 and P21 in the spleen (Figure 6G–I), indicating that ectopic LINC01021 expression enhances senescence‐associated gene expression in vivo, thereby linking organismal phenotypes to underlying cellular senescence processes.

In addition, male transgenic mice exhibited early‐onset coat depigmentation beginning around 8 months of age, with homozygous KI/KI animals developing pronounced whitening by 10 months (Figure 6J,K), further supporting premature aging‐associated phenotypes.

Functional assessments further revealed aging‐related impairments in motor performance and systemic immune homeostasis. From 10 months of age onward, transgenic mice exhibited reduced grip strength compared with WT controls, with the most pronounced decline observed in KI/KI mice compared to their WT counterparts, while younger KI mice showed a similar downward trend (Figures 6L and S5H). In beam‐walking assays, aged transgenic mice required more time to traverse the beam and paused more frequently, accompanied by a trend toward increased foot slips (Figures 6M–O and S5I–K). In contrast, rotarod and treadmill performance did not differ significantly between WT and KI mice (Figure S5L–S), suggesting that motor deficits associated with LINC01021 expression are task‐specific rather than globally impaired. Hematological analyses revealed systemic aging‐associated immune alterations, including increased white blood cell and lymphocyte counts in both KI genotypes (Figures 6P,Q and S5T,U), with additional elevations in granulocytes and monocytes in KI/KI mice (Figures 6R,S and S5V,W), while platelet and red blood cell counts remained unchanged (Figures 6T and S5X–Z).

Collectively, these findings demonstrate that ectopic expression of human LINC01021 contributes to multiple aging‐like phenotypes in vivo, linking primate‐specific LINC01021 activity to organismal senescence and functional decline.