TDP-43 facilitates milk lipid secretion by post-transcriptional regulation of *Btn1a1* and *Xdh*

The study investigates how milk lipid secretion is regulated at the molecular level, focusing on post-transcriptional regulation in mammalian lactation. While breastfeeding confers significant health benefits to infants and mothers, lactation insufficiency is common. Milk lipids are a major energy source for neonates, and proper secretion of lipid droplets is essential. Known components include BTN1A1 (encoded by Btn1a1) and XOR (encoded by Xdh), both implicated in milk lipid secretion. However, upstream post-transcriptional regulators are poorly understood. Given that RNA-binding proteins (RBPs) control RNA stability and processing, and lactation is a defining mammalian trait, the authors hypothesize that an RBP under positive selection in mammals might regulate key milk lipid secretion genes. They screen RBPs for signatures of positive selection and functionally test TDP-43 (encoded by TARDBP) as a candidate regulator of milk lipid secretion.

Prior work established BTN1A1 and XOR as key factors in milk fat globule (MFG) formation and milk lipid secretion. Btn1a1 knockout mice show failure in milk lipid secretion, large lipid droplets, and poor pup survival. XOR modulates lipid secretion and lactation initiation; mammary-specific deletion affects droplet docking but may not fully block secretion. Transcriptional regulation of Xdh has been described, but post-transcriptional regulation is less explored. Reports suggest RNA stability control is critical for lactation onset. RBPs, including TDP-43, mediate RNA processing and stability and have been implicated in various diseases and in breast cancer biology. This context supports investigating RBPs as post-transcriptional regulators of milk lipid secretion, specifically TDP-43 for its UG-rich RNA binding and potential roles in mammalian traits.

• Evolutionary analysis: Conducted phylogenomic positive selection analysis on 60 one-to-one orthologous RBPs across 15 vertebrates (8 mammals as foreground; 4 birds, 2 reptiles, 1 fish as background). Used PRANK for CDS alignment, Gblocks for conserved codon selection, and PAML branch-site models for LRTs with FDR correction to identify positively selected RBPs on the mammalian ancestral branch.
• Mouse genetics: Generated mammary epithelial-specific Tardbp knockout by crossing Tardbp floxed mice (deletion of exons 2–3) with WAP-Cre mice (activated mid-pregnancy to lactation in luminal MECs). Verified knockout by IHC, qRT-PCR, and Western blot at P17.5 and L10.
• Pup survival and growth: Monitored pup survival across litters in first and second lactation, with controlled litter sizes (6–8; adjusted to 7 at L2). Performed cross-fostering between WT and Tardbp−/− dams to assess maternal vs pup genotype effects. Tracked pup body weights over lactation.
• Milk collection and composition: Collected milk at L2 and L10 after oxytocin-induced letdown (10 U or 0.2 U depending on stage). Measured milk volume. Assessed milk protein levels (BCA, SDS-PAGE). Quantified milk triacylglycerols (TAGs); analyzed fatty acid profiles by GC.
• MEC lipid content: Isolated primary MECs; quantified intracellular TAG normalized to protein.
• Histology and imaging: Whole-mount carmine staining and H&E histology across pregnancy and lactation stages. TUNEL and Ki67 assays for apoptosis and proliferation. PLIN2 immunofluorescence with WGA and DAPI to visualize LDs and luminal borders. Transmission electron microscopy to examine LD accumulation and ultrastructure.
• Cell culture: HC11 mouse mammary epithelial cells grown and differentiated (DEX/insulin/prolactin). Assessed dome formation with and without Tardbp knockdown; rescue by BTN and XOR co-expression.
• Gene expression profiling: RNA-seq on MECs at L1 from WT and Tardbp−/−; mapped with STAR, counted with FeatureCounts, differential expression with edgeR (adjusted P<0.01; |log2FC|≥1). Identified lipid metabolism genes and UG-motif enrichment analyses to nominate direct TDP-43 targets.
• qRT-PCR and Western blot: Validated Btn1a1 and Xdh mRNA/protein levels at P17.5, L1, L10 in vivo and after Tardbp knockdown in HC11.
• RNA-protein interaction: Identified UG/TG-enriched sequences in 3′UTRs of Btn1a1 and Xdh. Performed RNA immunoprecipitation (RIP) in primary MECs and HC11 with TDP-43 antibody. Conducted biotin-RNA pull-down assays using in vitro-transcribed 3′UTR fragments to test TDP-43 binding. Domain-mapping with overexpressed Flag-tagged TDP-43 full-length and deletion mutants (ΔRRM1, ΔRRM2, ΔC) in RIP to define RNA-binding domains. Fragmented mRNA pull-down to map binding regions.
• mRNA stability assays: Treated primary MECs (P17.5, L1) and HC11 with actinomycin D to block transcription; measured decay kinetics of Btn1a1, Xdh (normalized to 18S rRNA), computed half-lives. Overexpressed TDP-43 FL or C-term to assess effects on mRNA half-life and protein expression.
• 3′UTR reporter assays: Constructed GFP reporters fused to mouse or human BTN1A1/XDH 3′UTRs and corresponding mutants lacking UG sites. Co-transfected with sh-TDP-43 in HC11; assessed GFP protein; conducted RIP-qPCR for GFP mRNA binding; measured GFP mRNA decay after actinomycin D.
• Involution assessment: Tracked mammary gland remodeling at L15, L18, L21 via whole-mount, H&E, and Ki67 to evaluate premature involution in Tardbp−/−.
• Human milk samples: Collected fresh milk (days 3–5 postpartum) from 60 healthy women. Isolated MFGs for RNA, verified MEC marker enrichment and low immune cell contamination. Compared TARDBP and HNRNPA1 mRNA levels across exclusive vs partial vs formula feeding groups via qRT-PCR. Controlled for collection time and delivery type.

• Evolutionary signal: Among 60 RBP orthologous groups, TARDBP and SRSF9 showed significant positive selection on the mammalian ancestral branch (FDR-corrected p=0.031 and 0.022, respectively); dN/dS for TARDBP=371.37; SRSF9=12.41.
• Lactation failure upon Tardbp loss: Maternal Tardbp−/− mice exhibited markedly reduced pup survival, with many pups dying before L2; survivors declined sharply after L15. Cross-fostering demonstrated the defect is maternal (lactation-related), not pup genotype-dependent. Pup weights were significantly lower when nursed by Tardbp−/− dams.
• Impaired milk secretion and lipid composition: Oxytocin-stimulated milk volume was significantly reduced at L2 and L10 in Tardbp−/− dams (n=7 per genotype). Milk TAG concentrations were markedly decreased at L2 (n=6/genotype), while MEC intracellular TAGs were increased at L2 (n=5/genotype) and L10 (n=6/genotype). Milk protein levels and expression of key milk protein genes were unchanged; fatty acid uptake indicators (PUFA/EFA) were similar between genotypes.
• LD phenotype: Milk fat globules from Tardbp−/− were significantly larger at L2 and L10 (P<0.001). MECs showed accumulation of large cytoplasmic lipid droplets and LDs filling alveolar lumens (PLIN2 IF; EM confirmed LD accumulation).
• BTN1A1 and XOR downregulation: RNA-seq at L1 identified lipid metabolism genes altered in Tardbp−/− MECs; Btn1a1 showed strongest UG-motif enrichment. qRT-PCR: Btn1a1 and Xdh mRNA significantly decreased at L1 and L10. Western blots: BTN and XOR proteins reduced at P17.5 and L1. In HC11 cells, Tardbp knockdown reduced BTN and XOR proteins and impaired dome formation; co-expression of BTN and XOR partially rescued dome formation.
• Direct binding of TDP-43 to 3′UTRs: UG/TG-enriched sequences present in 3′UTRs of Btn1a1 and Xdh. RIP in MECs and HC11 showed robust TDP-43 binding to Btn1a1 and Xdh mRNAs. Biotin-RNA pull-down confirmed binding of 3′UTR fragments (Btn1a1 nt 2781–3398; Xdh nt 4399–4623). Domain mapping showed RRM1 and RRM2 required for binding; C-terminus alone did not bind.
• mRNA stability regulation: In primary MECs, Btn1a1 and Xdh mRNA half-lives were reduced in Tardbp−/− vs WT. At P17.5: Btn1a1 t1/2 19.2 h (WT) vs 8.0 h (KO); Xdh t1/2 9.7 h (WT) vs 6.4 h (KO). At L1: Btn1a1 t1/2 17.5 h (WT) vs 7.3 h (KO); Xdh t1/2 11.5 h (WT) vs 6.7 h (KO). TDP-43 FL overexpression in HC11 increased mRNA half-lives (Btn1a1: 6.9→9.1 h; Xdh: 5.6→8.6 h), whereas C-terminal fragment had no effect. TDP-43 overexpression increased BTN and XOR proteins.
• 3′UTR reporter validation: TDP-43 knockdown decreased GFP fused to Btn1a1 or Xdh 3′UTRs; deleting UG-rich sites abolished this effect. RIP-qPCR showed reduced binding to mutant 3′UTRs. mRNA decay assays confirmed TDP-43-dependent stabilization for wild-type 3′UTRs, not mutants. Human BTN1A1/XDH 3′UTR GFP reporters behaved similarly upon TDP-43 knockdown.
• Premature involution: Tardbp−/− glands exhibited early remodeling with MFG accumulation, cell shedding at L15/L18, and collapsed alveoli by L21; Ki67 decreased at L21.
• Human relevance: In MFG RNA from human milk (days 3–5 postpartum; n=60), TARDBP mRNA was significantly higher in exclusive breastfeeding vs partial breastfeeding; HNRNPA1 showed no group differences. No differences due to collection time or delivery type. This suggests low TDP-43 expression associates with lactation insufficiency.

The study addresses the role of post-transcriptional regulation in milk lipid secretion by identifying TDP-43 as an RBP under mammalian positive selection that is necessary for efficient lactation. Functional loss of TDP-43 in mammary epithelial cells leads to decreased milk volume and lipid output, accumulation of large lipid droplets within MECs and alveolar lumens, and reduced offspring survival, indicating a secretion failure rather than lipid synthesis or uptake defects. Mechanistically, TDP-43 directly binds UG-rich motifs in the 3′UTRs of Btn1a1 and Xdh, enhancing their mRNA stability through its RNA recognition motifs (RRM1/2). This stabilizing effect maintains BTN1A1 and XOR protein levels required for docking and secretion of MFGs. The resultant decrease in Btn1a1/Xdh expression in Tardbp−/− MECs explains the secretion blockade and LD accumulation, aligning with phenotypes of Btn1a1 knockout models. The premature involution observed suggests prolonged secretion defects lead to tissue remodeling, though additional pathways may contribute. Human MFG transcript data support translational relevance, linking higher TARDBP expression to successful exclusive breastfeeding. Collectively, these findings position TDP-43 as a critical post-transcriptional regulator of milk lipid secretion and highlight RNA stability control as a key node in lactation biology.

TDP-43 has undergone positive selection in mammals and is essential for milk lipid secretion. Mammary-specific loss of TDP-43 causes lactation failure characterized by reduced milk volume, decreased milk TAGs, increased intracellular TAGs, enlarged lipid droplets, and poor pup survival. TDP-43 directly binds UG-rich elements in the 3′UTRs of Btn1a1 and Xdh to stabilize their mRNAs, maintaining BTN1A1 and XOR protein levels necessary for lipid droplet secretion. Human data indicate that higher TARDBP expression associates with exclusive breastfeeding, suggesting clinical relevance for lactation insufficiency. Future work should: (1) delineate additional TDP-43 targets and pathways influencing mammary lipid homeostasis; (2) dissect cell-type-specific roles within the mammary epithelium; (3) explore therapeutic modulation of TDP-43 or its mRNA targets to alleviate lactation insufficiency; and (4) validate findings in larger, longitudinal human cohorts with functional milk output measures.

• The WAP-Cre system targets luminal MECs from mid-pregnancy onward; earlier developmental roles or effects in other cell types were not assessed.
• While decreased Xdh contributed to the phenotype, premature involution in Tardbp−/− glands may involve additional mechanisms beyond Xdh downregulation.
• Some quantitative milk volume and composition data are limited to specific time points and sample sizes; detailed kinetics across lactation stages are not fully characterized.
• Human data are correlative, based on MFG RNA from a modest cohort (n=60; formula feeding n=2), without direct measurement of milk volume/output or causal inference.
• RNA stability assessments were performed in mouse MECs and HC11 cells; in vivo rescue experiments (e.g., Btn1a1/Xdh overexpression in Tardbp−/− glands) were not reported.