Upregulated hepatic lipogenesis from dietary sugars in response to low palmitate feeding supplies brain palmitate

The developing brain accumulates saturated fatty acids in parallel with growth both in utero and postnatally, with palmitic acid (PAM; 16:0) comprising a major proportion of brain fatty acids and playing roles in myelination, membrane structure, protein palmitoylation, and signaling. PAM can be sourced from the diet or synthesized de novo from substrates such as glucose. Infant diets (human milk) provide PAM, often esterified at the sn-2 position, enhancing absorption. However, preclinical data on the origin of brain PAM during development are limited and conflict with adult studies: tracer studies in rat pups suggested negligible incorporation of dietary PAM into brain lipids and reliance on endogenous synthesis, whereas in adults dietary PAM is incorporated into brain lipids via several routes. Compound-specific isotope analysis (CSIA) of δ13C leverages natural isotopic differences between dietary sugars (C4 plants; 13C-enriched) and dietary fat (C3 plants; 13C-depleted) to infer fatty acid origins without tracers. Prior adult mouse work indicated that approximately 70% of brain PAM is maintained by DNL from dietary sugars, with a substantial local brain contribution, and that low dietary PAM augments DNL contributions. The present study investigates during development whether brain PAM originates from dietary PAM versus DNL from dietary sugars, and identifies tissue-specific pathways regulating brain PAM maintenance.

Previous studies in developing rodents using labeled PAM (e.g., 3H-palmitate) reported little to no entry of intact dietary PAM into brain lipids, implying predominant local synthesis during development. In contrast, adult rodent studies using 14C-PAM via feeding, intravenous, intracarotid, or perfusion approaches demonstrated incorporation into brain lipids. CSIA at natural abundance has been applied in adults to show that most brain PAM derives from DNL from dietary sugars, with 44–58% from local brain DNL and augmentation under low dietary PAM conditions. Broader literature also indicates metabolic flexibility in liver versus brain in response to essential fatty acid deprivation (e.g., upregulated hepatic conversion of ALA to DHA in deficiency without parallel changes in brain), supporting a model where liver adapts to dietary supply while brain maintains a basal DNL program.

• Study design and animals: BALB/c dams (4 weeks old) were randomized to isocaloric low-PAM (LP), medium-PAM (MP), or high-PAM (HP) diets for 4 weeks pre-breeding, bred, and maintained on the same diets through lactation until weaning (P21). Offspring continued on the respective diets post-weaning to P35. Male pups (primary outcome) and when possible female pups were sampled at P0, P10, P21, and P35; dams were sampled at offspring P21. Animal care adhered to approved protocols.
• Diets: Three isocaloric diets (16.8% fat, 19.4% protein, 63.9% carbohydrate) differed only in fatty acid composition. LP and HP provided 15.6% energy from oleic acid (OLA) or PAM, respectively; MP split 7.8% energy from PAM and 7.8% from OLA. All diets contained 1.2% energy from soybean oil (essential PUFA). Diet fatty acid profiles and δ13C values were validated by GC-FID and GC-C-IRMS; dietary carbohydrates and protein δ13C were measured by EA-IRMS. Weighted dietary sugars were 13C-enriched (−11.15 ± 0.65 mUr) relative to PAM (≈ −29.4 to −29.7 mUr) and OLA (≈ −28.3 mUr), enabling isotopic discrimination.
• Sample collection: Brain and liver tissues were snap-frozen at collection (P0, P10, P21, P35 pups; dams at 15 weeks/offspring P21). Milk exposure pre-weaning was verified via pup stomach content fatty acid profiles (P0, P10) reflecting low, medium, and high PAM.
• Lipid extraction and fractionation: Total lipids were extracted (Folch method). For subsets, brain phospholipid classes (ChoGpl, EtnGpl, PtdIns, PtdSer, CerPCho) and liver neutral lipids (TAG, PL, CE, MAG, DAG, FFA) were separated by TLC. Internal standards were used for quantification.
• Fatty acid quantification: Fatty acid methyl esters (FAMEs) were quantified by GC-FID using appropriate columns and temperature programs. Peaks were identified via external standards and quantified relative to internal standards.
• Compound-specific isotope analysis: δ13C of PAM and other fatty acids (OLA, POA, STA, LNA, ARA, DHA) in tissues and diets were measured by GC-C-IRMS, normalized to VPDB using USGS reference materials and multipoint linear normalization (R2 > 0.999). Methylation correction for FAMEs used EA-IRMS to determine the δ13C of the added methyl group (−41.56‰) and applied mass-balance correction.
• RNA sequencing: Day 35 male brain and liver RNA (n=5/diet/tissue) were extracted, quality-checked (average RIN 8.27 ± 0.68), poly-A selected, and libraries prepared (NEBNext Ultra II Directional). Sequencing was performed on Illumina NovaSeq 6000 (paired-end 150 bp). Reads were trimmed and QC’d (FastQC, Trim Galore), screened (FastQ Screen), aligned to mouse genome (STAR, GENCODE m27), and quantified (HTSeq; DESeq2 for FPKM). PCA assessed sample clustering.
• Differential expression and pathway analyses: DEGs were identified with DESeq2 (adjusted p < 0.05 or 0.1; log2FC > 1.2). GSEA (iDEP.96; FGSEA) used DESeq2 fold-changes with GO Biological Process sets (FDR 0.2). WGCNA identified co-expression modules among top 1000 variable genes; GO enrichment performed for modules.
• Behaviour: Maternal nest-building during gestation days 15–18 scored using a naturalistic nest scoring system. Pup sensorimotor development assessed via geotaxis and righting reflex at P0–P10.
• Statistics: Two-way ANOVA for pup multi-timepoint outcomes with Tukey’s post-hoc when interactions present; one-way ANOVA or Kruskal–Wallis for single timepoint or dam comparisons; significance at p < 0.05. Outliers assessed by ROUT.

• Brain versus liver PAM levels during development:
  • Male pup brain total PAM: no significant main effect of diet; significant effect of time on relative percentage and concentration (both p < 0.0001). Relative % peaked at P10; concentration increased from P10 to P21 and P35.
  • Male pup liver total PAM: significant main effects of diet (relative % p < 0.0001; concentration p = 0.0351) and time (relative % p < 0.0001; concentration p = 0.0024). Dose-response reduction with lower dietary PAM. Liver total PAM peaked at P10, decreased by P21 and P35.
  • Separated lipid fractions: In liver at P35, dose-response reduction in relative % PAM persisted in TAG (p < 0.0001) and CE (p = 0.0135) but not PL, MAG, DAG, FFA. Brain PL class-specific PAM levels differed by class (p < 0.0001) without diet effects.
  • Female pups showed similar patterns to males for brain and liver PAM.
• Isotopic origin (δ13C-PAM):
  • Brain δ13C-PAM (male pups): significant diet × time interaction (p = 0.0008). Across time, lower dietary PAM led to more 13C-enriched δ13C-PAM, indicating greater DNL from dietary sugars. LP vs HP brains were 4–11% more enriched at all timepoints (adjusted p 0.0476 to < 0.0001). LP vs MP were 4–7% more enriched at P0, P21, P35 (p = 0.0085 to < 0.0001). MP vs HP differed at P35 (4% more enriched; p = 0.0066). Absolute brain δ13C-PAM values were closer to dietary sugars (−11.15 ± 0.65 mUr) than dietary PAM (≈ −29.5 mUr), supporting a predominant DNL contribution.
  • Liver δ13C-PAM (male pups): significant main effects of diet and time (both p < 0.0001) with a pronounced dose-response enrichment under lower dietary PAM. LP vs HP: 18–31% more enriched at all timepoints (p < 0.0001). LP vs MP: 13–24% more enriched (p < 0.0001). TAG fraction at P35 showed similar pattern (p = 0.0002). Female pup δ13C-PAM patterns mirrored males.
  • Other fatty acids: Brain and liver δ13C for lipogenic fatty acids showed diet/time effects consistent with enhanced DNL under low PAM (e.g., brain and liver δ13C-POA main effect of diet p < 0.0001; brain δ13C-STA diet × time p = 0.0036; liver δ13C-STA diet × time p = 0.0076). Non-lipogenic PUFA (LNA, ARA, DHA) δ13C values were unaffected by diet in brain or liver.
• Dams: No diet effect on total brain or liver PAM levels. δ13C-PAM in dam brain (p < 0.0001) and liver (p = 0.0013) showed dose-response enrichment with lower dietary PAM, indicating increased DNL from dietary sugars.
• Behaviour: No differences among diet groups in maternal nest quality or pup geotaxis/righting reflex development P0–P10.
• RNA sequencing (day 35 males):
  • PCA showed distinct clustering by tissue (liver vs brain), with liver exhibiting greater expression variance; no distinct clustering by diet within tissues, but intra-group correlations > 0.9.
  • DEGs: Numerous liver DEGs for LP vs MP and LP vs HP (adjusted p < 0.05 or 0.1; |log2FC| > 1.2). No DEGs detected in brain across diet contrasts; none for MP vs HP in liver. • Top liver DEG HP vs LP: cyp7a1 upregulated in HP (log2FC 3.97; p = 5.9e−10). • Top liver DEG MP vs LP: pla2g4f upregulated in MP (log2FC 3.66; p = 2.88e−05).
  • GSEA (FDR 0.2): In liver HP vs LP, lipid metabolism-related pathways were differentially expressed: upregulated import across plasma membrane and cholesterol biosynthesis; downregulated acyl-CoA and thioester biosynthesis. cyp7a1 appeared in 10 significant pathways. No significant pathway changes in brain for HP vs LP; MP vs LP liver pathways were largely cell-cycle related and not lipid metabolism.
  • WGCNA (liver): 9 modules identified. Yellow module (138 genes) enriched for lipid metabolic processes (lipid/sterol/small molecule biosynthesis; acyl-CoA, thioester, fatty acid metabolism). Top genes included acot2, sqle, cyp51, elovl6, scd2, fasn, acly. Blue module (185 genes) enriched for lipid metabolism (includes pla2g4f, cytochrome P450s, carboxylesterases). Green module included cyp7a1 and was enriched for homeostatic processes. Brain modules showed no enrichment for lipid metabolism.
• Overall interpretation: Despite large dietary differences, brain total PAM levels were maintained, while isotope data indicate the majority of brain PAM during development derives from DNL using dietary sugars. Under low dietary PAM, hepatic DNL is upregulated and likely supplies the brain with PAM, while brain transcriptional programs remain relatively unchanged, indicating a liver-driven compensatory mechanism.

The study addressed whether developing brain PAM is derived from dietary PAM or synthesized de novo from dietary sugars and which tissues respond transcriptionally to maintain brain PAM. Isotopic analysis showed brain δ13C-PAM values closer to dietary sugars than dietary PAM across developmental stages, indicating DNL from dietary sugars maintains most brain PAM. This contribution was augmented under low dietary PAM. In parallel, liver δ13C-PAM was strongly enriched under low PAM, and transcriptomic analyses at day 35 revealed extensive hepatic, but not brain, changes: lipid metabolism pathways (including cholesterol biosynthesis and substrate import) were upregulated, while acyl-CoA and thioester biosynthesis processes were downregulated in HP versus LP comparisons, and co-expression modules enriched for DNL and fatty acid/cholesterol metabolism genes were identified (fasn, acly, elovl6, scd2, cyp genes). Together, these results support a compensatory hepatic response that enhances DNL from dietary sugars to supply PAM to the brain when dietary PAM is low, maintaining brain PAM homeostasis. The absence of brain DEGs and lipid metabolism pathway changes suggests the developing brain relies on a relatively stable basal DNL program and on peripheral (hepatic) supply. The findings align with literature showing hepatic metabolic flexibility versus relative brain stability under dietary manipulations and suggest that isotopic approaches can inform substrate usage during development. Temporal patterns in δ13C-PAM also imply potential contributions from other substrates (e.g., ketone bodies) earlier in development, though glucose remains the predominant source for DNL maintaining brain PAM.

Using natural abundance δ13C analysis and transcriptomics, the study demonstrates that during development the majority of brain palmitic acid is maintained by de novo lipogenesis from dietary sugars. Under low dietary PAM, hepatic DNL is upregulated, supplying PAM to the brain and preserving total brain PAM pools, whereas brain gene expression related to lipid metabolism remains stable. This identifies a liver-driven compensatory mechanism critical for brain PAM homeostasis in development. The work establishes δ13C-based CSIA as a feasible measure of tissue fatty acid origin in developmental contexts. Future research should quantify contributions of non-glucose substrates to acetyl-CoA pools for DNL, examine post-transcriptional and post-translational regulation of lipogenesis, extend transcriptomic profiling to earlier developmental stages, and consider multi-isotope approaches (e.g., hydrogen isotopes) to estimate the fraction of newly synthesized fatty acids.

• Temporal limitation of transcriptomics: RNA-seq was performed only at day 35; gene regulation at earlier developmental stages (P0–P21) was inferred from δ13C patterns but not directly measured.
• Regulatory scope: Analyses focused on mRNA expression. Post-transcriptional, post-translational, and allosteric regulation of lipogenesis (e.g., SREBP-1c, ChREBP pathways) were not assessed.
• Substrate attribution: While δ13C distinguishes sugars from fats/protein, it cannot quantify contributions of other acetyl-CoA sources (e.g., ketone bodies, ketogenic amino acids). The study could not resolve exact fractional contributions of non-glucose substrates to DNL.
• Fractional synthesis quantification: CSIA at natural abundance does not directly provide the fraction of newly synthesized PAM; hydrogen isotope methods were suggested to address this.
• Maternal adaptation: Dams on low PAM likely upregulated hepatic and possibly mammary DNL, altering milk fatty acid composition; while documented, this adaptation complicates strict isolation of pre-weaning dietary PAM effects.
• Blinding: Diet administration could not be blinded, potentially introducing procedural bias, though analytical procedures and QC steps mitigate this risk.
• Sex balance: Primary outcomes were powered in males due to higher male pup numbers; female data were included when possible but were not the primary focus.