Survival phenotypes and lipid metabolic responses to DHAV-3 infection in Pekin duck hybrids with differential resistance Authors: Suyun Liang, Meijuan Zhang, Huihui Li, Bing Li, Mingtian Mao, Daewi Luo, Zhanbao Guo, Yi Tang, Shuisheng Hou Published: 2026 ABSTRACT Host metabolic reprogramming is increasingly recognized as a determinant of viral pathogenesis, yet its role in genetic disease resistance remains elusive. This study investigated the survival and lipid metabolic dynamics of Pekin duck combinations—pure resistant (RR), pure susceptible (SS), and reciprocal crosses (RS: R♀ × S♂; SR: S♀ × R♂)—following DHAV-3 infection. Phenotypic analysis revealed striking maternal inheritance in disease resistance: RS females exhibited the lowest mortality (22.7%), significantly outperforming SR females (51.4%) and even the resistant RR pure line (27.0%). To uncover the underlying mechanisms, we analyzed pre- and post-infection lipid profiles. Pre-infection liver biochemistry identified baseline cholesterol (CHOL) and LDL-C as robust predictors of viral susceptibility, with the highly susceptible SS group exhibiting the highest levels. Following viral challenge, plasma lipid dynamics and liver function indicators (ALT, AST) fluctuated significantly depending on cross-combination, sex, and time. Furthermore, the dynamic expression of key lipid metabolism genes, particularly cholesterol efflux regulators (ABCA1, ABCG1), was tightly orchestrated by genetic background and sex interactions. Overall, these findings demonstrate that baseline lipid profiles and post-infection lipid network reshuffling dictate the disease resistance differences among hybrid combinations. The exceptional resistance of the RS combination highlights the critical role of maternal effects and offers a metabolic perspective for strategic crossbreeding in poultry. Introduction Heterosis refers to the phenomenon in which hybrid offspring outperform the average of their parents in traits such as growth, reproduction, and disease resistance. Its genetic basis primarily involves non-additive effects, including dominance, superdominance, and epistatic effects. Early studies systematically evaluated heterosis, breed differences, and recombination effects in livestock such as pigs and cattle. In pigs, body weight at multiple ages and average daily gain are influenced by hybrid combinations. There is a strong negative correlation (approximately −0.90) between maternal heterosis and recombination effects (Cassady et al., 2002); however, this heterosis is significantly influenced by breed (Lo et al., 1992), and growth, carcass, and reproductive traits are genetically relatively independent (Petry et al., 2004). In dairy cattle, milk production and reproductive performance in Holsteins can be improved through crossbreeding combinations (Dezetter et al., 2015). In beef cattle, genomic heterozygosity shows a weak correlation with carcass traits, which can supplement pedigree information to enhance prediction accuracy; furthermore, non-additive SNPs exist on some chromosomes (Kenny et al., 2022). Adult body weight and body condition score in cows show a moderate positive genetic correlation, and hybrid vigor can significantly improve both traits, with marked differences between breeds (Ribeiro et al., 2022). With the advancement of molecular technologies, research on hybrid vigor has extended to the transcriptomic level. In chickens, using backcross and outcross populations (White Leghorns and Pekin Oil Chickens), it was found that sexual maturity traits (pubic bone distance, oviduct length) exhibited over 20% hybrid vigor, while age at first lay showed negative hybrid vigor; Non-additive gene expression (mRNA and lncRNA) predominates (44%–49%), with non-additive lncRNAs influencing the GnRH signaling pathway and gonadal development by targeting non-additive genes (e.g., CACNA1C, TGFB1), thereby driving hybrid vigor in sexual maturation (Wang et al., 2022a). Hybrid vigor in egg production (approximately 4%) was significantly negatively correlated with age at first lay and laying interval; although laying duration exhibited negative hybrid vigor, the laying pattern was improved (Wang et al., 2022b). Regarding feed intake and residual feed intake (RFI), YW hybrid chickens exhibited negative hybrid vigor. Non-additivity is the primary mode of gene expression in the hypothalamus, liver, and duodenum, and is closely associated with the oxidative phosphorylation (OXPHOS) pathway; Genes and lncRNAs exhibiting non-additive expression were enriched in mitochondrial components, and weighted co-expression network analysis further confirmed the key role of the OXPHOS pathway. ATP5I protein levels in the liver were consistent with gene expression, suggesting that hybrid chickens reduce ATP production by downregulating OXPHOS, thereby decreasing feed intake and feed efficiency (Yuan et al., 2023). In culled hens (100 weeks of age), hybrid chickens exhibited significantly higher body weight and slaughter weight than purebred lines, and different inbred-outbred combinations demonstrated varying degrees of hybrid vigor in breast muscle weight, leg muscle weight, and abdominal fat percentage; regarding meat quality, color indices of different hybrid combinations showed either positive or negative hybrid vigor (Yang et al., 2024a). Studies on circular RNAs (circRNAs) have similarly revealed tissue-specific non-additive expression. RFI-associated circRNA-mRNA co-expression modules were identified in the hypothalamus and liver, and these were also enriched in the OXPHOS pathway; Eight non-additive circRNAs were highly correlated with 16 non-additive genes; among them, circ-ITSN2 was significantly positively correlated with PGPEP1L in the liver, potentially exerting its effects through binding to microRNAs or RNA-binding proteins (Yuan et al., 2024). Furthermore, in meat sheep, direct and maternal hybrid vigor contribute significantly to growth traits (2.9%–6.1%), emphasizing the necessity of considering both hybrid vigor and breed effects in genetic evaluations (Duenk et al., 2020; Vargas Jurado et al., 2022). In the field of disease-resistant breeding for livestock and poultry, utilizing hybrid vigor to enhance animals’ resistance to pathogenic microorganisms has become a research hotspot. However, existing studies have primarily focused on growth performance and reproductive traits, while systematic analyses of hybrid vigor regarding antiviral immune traits remain relatively scarce. Particularly in waterfowl breeding, there is a lack of theoretical support based on molecular mechanisms regarding how to precisely control infectious disease risks and improve population health levels by optimizing hybrid combinations. Based on this, the present study hypothesizes that the combination of genetic backgrounds for disease resistance and susceptibility in parents will significantly influence the antiviral immune phenotype and transcriptional regulation of related genes in offspring. To this end, we established four hybrid populations: RR (resistant♂ × resistant♀), RS (resistant♀ × susceptible♂), SS (susceptible♂ × susceptible♀), and SR (susceptible♀ × resistant♂). Through artificial infection with DHAV-3, we systematically integrated multidimensional data on mortality phenotypes, serum biochemical and immunological indicators, and key gene transcription levels. The aim is to elucidate disease resistance differences among Pekin duck hybrid combinations, identify the core molecular regulatory pathways mediating antiviral hybrid vigor, and provide a theoretical basis for the molecular design of disease-resistant Pekin duck breeding lines. Materials and methods Virus The genotype III of duck hepatitis A virus (NCBI accession number: PQ777123.1) was isolated, purified and identified by our laboratory, and then stored at −80°C conditions (Liang et al., 2024). The titer of the DHAV-3 was determined to be 105.49 viral RNA copies. Animals The S2 line resistant to DHAV-3 Pekin ducks and susceptible ducks selected by our team were used as the experimental materials. The ducks were grouped and hatched according to the ratio of male to female = 1:4, and the eggs were collected. The F1 generation was obtained after hatching. The hybrid combinations of F1 generation were RR (resistant♂ × resistant♀), RS (resistant♀ × susceptible♂), SS (susceptible♂ × susceptible♀), and SR (susceptible♀ × resistant♂). All Pekin ducks were raised at the Sefiya Breeding Farm in Inner Mongolia (Liang et al., 2023). All experimental animals had free access to food and water during the experiment period. Experimental animals and breeding design All operations involving animals were guided and supported by the Animal Welfare and Ethics Committee of the Chinese Academy of Agricultural Sciences (IAS2025-134). Each duckling was intramuscularly inoculated in the leg muscle with 0.2 mL of DHAV-3 suspension containing 10 viral RNA copies, as determined by absolute quantitative PCR. After injection, the ducks were fed according to the conventional feeding and management methods, and the mortality rate of each group was counted. Six chicks from each group were collected for liver samples before infection, liver and blood samples were collected at 36, 48, and 72 hpi and three males for plasma biochemistry analysis, Oil Red O staining and RT-qPCR, additionally, three females were randomly selected from each hybrid combination at each time point. Biochemical index detection For liver samples, the weight of the animal tissue was weighed and added to 9 times the volume of the homogenization medium (0.9% normal saline) at a ratio of weight (mg): volume (ul) = 1:9. The samples were homogenized under an ice water bath, and a 10% homogenate was prepared. Centrifugation at 3500-4000 rpm for 10 minutes was performed, and the supernatant was taken for measurement. For blood samples, the samples were centrifuged at 3500 rpm for 10 minutes within 30 minutes after collection at 4°C, and the supernatant was taken. Six biochemical indicators were measured in liver and plasma samples: alanine aminotransferase (ALT, U/L), aspartate aminotransferase (AST, U/L), total cholesterol (CHOL, mmol/L), high-density lipoprotein cholesterol (HDL-C, mmol/L), triglycerides (TG, mmol/L), and low-density lipoprotein cholesterol (LDL-C, mmol/L). Oil Red O staining Take frozen liver tissue and prepare 6-10 μm thick frozen sections. Rinse them 3 times with distilled water, immerse them in 100% isopropanol for 5 minutes to remove moisture, then place them in the oil red O staining solution and stain at 60°C in a sealed environment for 8 minutes. Remove the sections and wash them successively with 85% isopropanol for 5 minutes, distilled water for 3 minutes, re-stain the cell nuclei with Mayer's hematoxylin staining solution for 30 seconds, rinse with running water for 3 minutes, wash with distilled water twice, absorb the surrounding moisture of the sections with filter paper, and seal with glycerol gelatin. The neutral fat in the sections appears orange-red or orange-yellow, and the cell nuclei are pale blue. After taking photos of the oil red O sections under a microscope (at 400x magnification), quantitative analysis was conducted using the Aipathwell® digital pathology image analysis software (Servicebio®, China). Quantitative Real-Time PCR (qPCR-RT) Analysis Total RNA was extracted from the livers of 4 hybrid combinations (RR, RS, SS, SR) at 36, 48 and 72 hpi after infection. The relative expression levels of ApoB (apolipoprotein B), LIPC (lipase C, hepatic type), ABCG1 (ATP binding cassette subfamily G member 1) and ABCA1 (ATP binding cassette subfamily A member 1) were detected by qRT-PCR. Three individuals of each sex and each time point were selected from each group. β-actin was used as the reference gene. The primer sequences are as follows: ApoB-F: AGTGGTGTTCTCACACAAGAAGA, ApoB-R: CTGTGGTACCTCCAGTTCAA; ABCA1-F: CGCAGAAACCTCAGATGGGA, ABCA1-R: GCCCTTCATCTGGTAGGAGC; ABCG1-F: AGCGTCTTGCCATTGCTTTG, ABCG1-R: CACTATCCAAGCCGCTGGT; LIPC-F: ACTGTTCCCGAGGAGTGAAT, LIPC-R: TTCACTGCTAAATTCTTGTGCC; β-actin-F: GGTATCGGCAGCAGTCTTA, β-actin-R: TTCACAGAGGCGAGTAACTT. The reactions were run on a qPCR instrument using the designed primers to measure the relative expression levels of the target genes. The PCR cycling conditions were as follows: initial denaturation at 95°C for 30 seconds, followed by 40 cycles of 95°C for 10 seconds, 60°C for 30 seconds (during which fluorescence was collected). After the cycling, a melting curve analysis was conducted: 95°C for 15 seconds, 60°C for 1 minute, and finally 95°C for 1 second, to confirm the specificity of the PCR products. The relative expression levels of the target genes were calculated using the ΔΔCt method, with the Ct value (Cycle Threshold) representing the point at which fluorescence exceeds the background level. The internal control gene (β-actin) was used to normalize the data and correct for experimental variation. Correlation analysis between baseline biochemical traits and mortality To evaluate the relationship between each biochemical indicator in the liver homogenate before infection and the mortality rate of the hybrid combinations, the mean values of each indicator in each group (n = 6, no gender distinction) at 0 hpi of the four combinations were calculated. The mortality data were the final mortality rate (%) at the end of the observation period after infection for each combination. The correlation between mortality and ALT, AST, CHOL, TG, HDL-C, and LDL-C was calculated using Spearman rank correlation analysis, and a correlation matrix heatmap was drawn. The strength of correlation was judged as follows: |r| ≥ 0.7 was considered strong correlation, 0.4–0.7 was considered moderate correlation, and < 0.4 was considered weak correlation. The significance level was set at P < 0.05. Principal Component Analysis To identify the main indicators that cause biochemical differences among different hybrid combinations, a principal component analysis (PCA) was conducted on the liver homogenate data before infection. The data included 4 combinations × 6 individuals, totaling 24 samples, with each sample containing 6 biochemical indicators. Before the analysis, the data was standardized using Z-score (mean = 0, standard deviation = 1). The PCA was completed using the FactoMineR package, and the variable contribution plot (Variables–PCA) was drawn using the factoextra package. The direction of the arrows in the Fig. indicates the correlation between the indicators and the principal components, and the length of the arrows represents the contribution of the indicators to the principal components. Based on the loadings of each indicator on the first principal component (Dim1) and the second principal component (Dim2), it was determined which indicators contributed to the separation between the combinations. Statistical analysis All statistical analyses were performed using R software (version 4.4.2) and the ggplot2 package was used to generate plots. Mortality was represented using cumulative mortality curves, and group comparisons were performed using either the chi-square test or Fisher's exact test, depending on the sample size. Three-way ANOVA was used to test the effects of the hybrid combination (RR, RS, SS and SR), time post-infection (36, 48 and 72 hpi), sex (male or female) and their interactions on plasma biochemical indicators. When a main or interaction effect was significant (P < 0.05), Tukey's HSD was applied for post-hoc multiple comparisons. Descriptive data are expressed as mean ± standard deviation (mean ± SD). For Oil Red O sections, the liver tissue area was automatically delineated using an artificial intelligence tracking function. Positive staining (lipid droplets) was identified based on the HSI colour model and the cumulative integrated optical density (IOD) was calculated. Two-way ANOVA was used to analyse the main effects and interactions of the combination (at each time point) and sex. If the main effect of the combination was significant, Tukey's HSD was used for pairwise comparisons between groups. Sex differences within the same combination were analysed using an independent sample t-test with Welch's correction. Gene expression data were analyzed using ART-aligned rank transformation for three-way ANOVA. The significance level was set at α = 0.05. Results Breeding design, F1 population, and mortality following viral challenge Four F1 genetic combinations were generated using resistant (R) and susceptible (S) ducks as parents: RR (R × R), RS (R × S), SS (S × S) and SR (S × R). The number of F1 individuals (male/female) was as follows: RR: 229 males and 256 females; RS: 257 males and 272 females; SS: 365 males and 317 females; SR: 252 males and 288 females (Fig. 1A). Following the challenge, the cumulative mortality rate over 240 hpi followed an S-shaped curve, peaking at 120–168 hpi. SS exhibited the highest and fastest mortality, RR the lowest and slowest, and the mortality of hybrids was intermediate, showing sex- and reciprocal-dependent differences (Fig. 1B). In males, RS mortality was consistently lower than SR and SS mortality, approaching RR mortality. However, SR mortality increased rapidly after 144 hpi, approaching SS mortality. In females, RS mortality was lower than that of both SR and RR females, with the latter approaching SS (Fig. 1C). The mortality curve of RS females fell below both RR and SS females, while that of SR females was close to SS females (Fig. 1D). At 240 hpi, mortality was 63.7% (SS ♂) and 66.2% (SS ♀); 30.6% (RR ♂) and 27.0% (RR ♀). RS females (22.7%) had lower mortality than the parental average (47.2%) and SR females (51.4%) (P < 0.01). SR males and females (∼51–52%) did not differ from the parental median (P > 0.05). There were significant reciprocal differences in females (22.7% vs. 51.4%, P < 0.001), but not in males (42.6% vs. 51.9%, P = 0.06) (Fig. 1E).Fig. 1Experimental design, population composition, and mortality kinetics of F₁ duck crosses following viral challenge. (A) The resistant ducks and susceptible ducks were used as parents for crossbreeding, resulting in four F₁ hybrid combinations: RR (R♀ × R♂), RS (R♀ × S♂), SS (S♀ × S♂), and SR (S♀ × R♂). The number of male and female offspring from each F₁ hybrid group has been indicated. The live ducks in the pictures represent the resistant ducks, while the dead ducks represent the susceptible ducks. (B) Cumulative mortality curves (hours post-infection, hpi) of the four crosses, stratified by sex. (C) Final mortality rates at the conclusion of the observation period, compared by cross and sex. (D) Comparative analysis of cumulative mortality between hybrids (RS, SR) and purebred lines (RR, SS). (E) Direct comparison of mortality kinetics between reciprocal hybrids (RS vs. SR), highlighting sex-dependent survival patterns. Experimental design, population composition, and mortality kinetics of F₁ duck crosses following viral challenge. (A) The resistant ducks and susceptible ducks were used as parents for crossbreeding, resulting in four F₁ hybrid combinations: RR (R♀ × R♂), RS (R♀ × S♂), SS (S♀ × S♂), and SR (S♀ × R♂). The number of male and female offspring from each F₁ hybrid group has been indicated. The live ducks in the pictures represent the resistant ducks, while the dead ducks represent the susceptible ducks. (B) Cumulative mortality curves (hours post-infection, hpi) of the four crosses, stratified by sex. (C) Final mortality rates at the conclusion of the observation period, compared by cross and sex. (D) Comparative analysis of cumulative mortality between hybrids (RS, SR) and purebred lines (RR, SS). (E) Direct comparison of mortality kinetics between reciprocal hybrids (RS vs. SR), highlighting sex-dependent survival patterns. Baseline hepatic lipid profiles strictly correlate with viral susceptibility prior to infection Pre‑infection liver tissue levels of ALT, AST, CHOL, TG, HDL‑C, LDL‑C were measured (n = 6/group, sexes combined). ALT ranged from 491.6 ± 133.5 (RS) to 584.0 ± 159.7 U/L (SS) (F₃,₂₀ = 0.95, P = 0.435) (Fig. 2A). AST ranged from 191.6 ± 22.52 (RR) to 234.8 ± 26.84 U/L (SS) (F₃,₂₀ = 3.72, P = 0.028), SS significantly higher than RR (P < 0.05) (Fig. 2B). CHOL highest in SS (0.552 ± 0.169 mmol/L), lowest in RR (0.226 ± 0.083 mmol/L) (F₃,₂₀ = 7.65, P = 0.0013), SS > RR, SR, RS (P < 0.01), RR < all others (P < 0.05) (Fig. 2C). TG highest in SS (0.538 ± 0.440 mmol/L), lowest in RS (0.205 ± 0.100 mmol/L), not significant (P = 0.174) (Fig. 2D). HDL‑C was low (0.003–0.058 mmol/L), no significant differences (P = 0.368) (Fig. 2E). LDL‑C highest in SS (0.778 ± 0.269 mmol/L), lowest in RR (0.344 ± 0.182 mmol/L) (F₃,₂₀ = 4.32, P = 0.016); SS > RR (P < 0.05) (Fig. 2F).Fig. 2Basal hepatic biochemical profiles in F₁ duck crosses prior to viral challenge. Concentrations of (A) alanine aminotransferase (ALT), (B) aspartate aminotransferase (AST), (C) total cholesterol (CHOL), (D) triglycerides (TG), (E) high-density lipoprotein cholesterol (HDL-C), and (F) low-density lipoprotein cholesterol (LDL-C) were measured in liver tissues of RR, SS, RS, and SR crosses (n = 6). Data are presented as mean ± SD. *P < 0.05, **P < 0.01. Basal hepatic biochemical profiles in F₁ duck crosses prior to viral challenge. Concentrations of (A) alanine aminotransferase (ALT), (B) aspartate aminotransferase (AST), (C) total cholesterol (CHOL), (D) triglycerides (TG), (E) high-density lipoprotein cholesterol (HDL-C), and (F) low-density lipoprotein cholesterol (LDL-C) were measured in liver tissues of RR, SS, RS, and SR crosses (n = 6). Data are presented as mean ± SD. *P < 0.05, **P < 0.01. Final mortality positively correlated with CHOL (r = 1), LDL‑C (r = 1), AST (r = 0.8), ALT (r = 0.4), HDL‑C (r = 0.4) (Fig. 3A). Dim1 explained 49.6% variance, Dim2 17.4% (cumulative 77.0%). CHOL and LDL‑C had largest positive loadings on Dim1; AST dominated Dim2 (Fig. 3B).Fig. 3Correlation analysis, PCA, and post-infection plasma enzyme kinetics. (A) Spearman rank correlation heatmap between pre-infection biochemical parameters and final mortality-related traits. Blue and red scales represent positive and negative correlations, respectively; color intensity indicates the strength of the correlation. (B) PCA variable contribution plot based on pre-infection hepatic biochemical data. Arrow direction and length represent the correlation with Dim1/Dim2 and the contribution magnitude of each indicator, respectively. (C) Comparison of the integrated optical density (IOD) of hepatic Oil Red O staining among different cross-combinations at 36, 48 and 72 hpi. (D) Gender-differentiated IOD of hepatic Oil Red O staining among combinations at 36, 48 and 72 hpi. (E) Representative images of Oil Red O-stained liver sections at 48 hpi. Correlation analysis, PCA, and post-infection plasma enzyme kinetics. (A) Spearman rank correlation heatmap between pre-infection biochemical parameters and final mortality-related traits. Blue and red scales represent positive and negative correlations, respectively; color intensity indicates the strength of the correlation. (B) PCA variable contribution plot based on pre-infection hepatic biochemical data. Arrow direction and length represent the correlation with Dim1/Dim2 and the contribution magnitude of each indicator, respectively. (C) Comparison of the integrated optical density (IOD) of hepatic Oil Red O staining among different cross-combinations at 36, 48 and 72 hpi. (D) Gender-differentiated IOD of hepatic Oil Red O staining among combinations at 36, 48 and 72 hpi. (E) Representative images of Oil Red O-stained liver sections at 48 hpi. Distinct patterns of hepatic lipid accumulation among cross-combinations following infection To further verify the relationship between lipid metabolism and the resistance combination, liver tissue sections at 36, 48, and 72 hpi were examined using Oil Red O staining. The results showed that only at 48 hpi was the difference between the combinations extremely significant (F₃,₂₀ = 8.290, P = 0.0009) (Fig. 3C). The IOD of the RS (95543 ± 12157) and SS (99033 ± 11875) combinations was significantly higher than that of the RR (66345 ± 16712) combination (P = 0.012 and P = 0.005), and it was also significantly higher than that of the SR (69066 ± 16909) combination (P = 0.025 and P = 0.010), while there was no significant difference between RR and SR (P = 0.988). When each combination was gender-differentiated, the results showed that only at 48 hpi was the main effect of the combination significant (F₃,16 = 6.831, P = 0.0036). In males, the IOD of the RS (93476 ± 3884) and SS (96799 ± 12314) combinations was significantly higher than that of the RR (63916 ± 13258) combination (P < 0.05), while there was no significant difference among the combinations in females (P > 0.05) (Fig. 3D). As showed in Fig. 3E, the cumulative optical density (IOD) of liver Oil Red O of the RS and SS hybrid combinations was significantly higher than that of the RR combination. Post-infection plasma biochemical dynamics are shaped by combination, sex, and time interactions Plasma ALT, AST, CHOL, TG, HDL‑C, LDL‑C at 36, 48, 72 hpi were analyzed by sex. Combination main effect on ALT was not significant (P = 0.646). At 72 hpi in females, RR (66.76 U/L) and SS (75.45 U/L) showed numerically higher than RS (32.08) and SR (30.28), but Tukey’s post‑hoc tests revealed no statistically significant pairwise differences (all P > 0.05) (Fig. 4A). Combination effect on AST was significant (P = 0.034); SS highest (e.g., 72 hpi females: 143.36 U/L), but Tukey post‑hoc tests showed no pairwise significance (Fig. 4B). For TG, the combination main effect was not significant (P = 0.107), but a significant combination × time × sex interaction was observed (P = 0.016). Descriptively, at 48 hpi females the RS group had the highest mean (2.82 mmol/L); at 72 hpi females the SR group had the highest mean (3.04); and at 72 hpi males the SS group had the highest mean (3.16) (Fig. 4C). For CHOL, the combination effect was highly significant (P = 0.0054); post‑hoc analysis showed that the SR group had significantly higher levels than the RS group (P < 0.05), e.g., in males at 36 hpi (5.39 vs. 2.60 mmol/L) and 48 hpi (5.10 vs. 2.71 mmol/L) (Fig. 4D). For HDL‑C, combination effect was highly significant (P = 0.00056) and combination × sex interaction significant (P = 0.048). Post‑hoc: in males, SR (3.48–3.74) > RS (1.59–2.97) (P < 0.01); in females, RR (2.84–3.99) > SS (2.11–2.32) (P < 0.05) (Fig. 4E). For LDL‑C, combination effect was not significant (P = 0.277), and mean levels were similar across groups (Fig. 4F).Fig. 4Spatiotemporal dynamics of plasma biochemical and lipid profiles across F₁ duck crosses, sexes, and infection phases. Temporal changes in plasma levels of (A) aspartate aminotransferase (AST), (B) alanine aminotransferase (ALT), (C) triglycerides (TG), (D) total cholesterol (CHOL), (E) high-density lipoprotein cholesterol (HDL-C), and (F) low-density lipoprotein cholesterol (LDL-C) at 36, 48, and 72 hpi. Within each panel, data are stratified by hybrid combinations (RR, SS, RS, SR) and sex (males: blue/solid lines; females: red/dashed lines) to illustrate three-dimensional variations. Significant differences between specific groups at a given time point are indicated by corresponding color-coded symbols. Data are presented as mean ± SD. *P < 0.05, **P < 0.01. Spatiotemporal dynamics of plasma biochemical and lipid profiles across F₁ duck crosses, sexes, and infection phases. Temporal changes in plasma levels of (A) aspartate aminotransferase (AST), (B) alanine aminotransferase (ALT), (C) triglycerides (TG), (D) total cholesterol (CHOL), (E) high-density lipoprotein cholesterol (HDL-C), and (F) low-density lipoprotein cholesterol (LDL-C) at 36, 48, and 72 hpi. Within each panel, data are stratified by hybrid combinations (RR, SS, RS, SR) and sex (males: blue/solid lines; females: red/dashed lines) to illustrate three-dimensional variations. Significant differences between specific groups at a given time point are indicated by corresponding color-coded symbols. Data are presented as mean ± SD. *P < 0.05, **P < 0.01. For time main effects: ALT was significant (P = 0.0059), 72 hpi > 36 hpi (P < 0.05) (Fig. 4A). AST significant (P = 0.015), 72 hpi > 36 hpi (P < 0.05) (Fig. 4B). CHOL not significant (P = 0.138) (Fig. 4D). TG not significant (P = 0.092) (Fig. 4C). HDL‑C marginally significant (P = 0.050), but post‑hoc showed no significant differences between any time points (all P > 0.05) (Fig. 4E). LDL‑C significant (P = 0.014), 48 hpi > 36 hpi (P < 0.05), 72 hpi not different from either (Fig. 4F). Dynamic Expression of Lipid Metabolism-Related Genes after Infection To evaluate the effects of combination, time, sex, and their interactions on lipid metabolism‑related gene expression, we performed aligned rank transform (ART) three‑way ANOVA. The results are shown in Table 1. For ABCA1, all main effects and interactions were highly significant (P < 0.001), with the sex main effect (F₁,₃₉ = 37.43, P < 0.001) and the combination × sex interaction (F₃,₃₉ = 27.82, P < 0.001) being particularly prominent. For ABCG1, significant effects were observed for combination (F₃,₄₂ = 6.71, P = 0.0008), time (F₂,₄₂ = 11.53, P = 0.0001), sex (F₁,₄₂ = 5.28, P = 0.027), and all two‑way and three‑way interactions (P < 0.01). For ApoB, significant main effects were detected for combination (F₃,₄₁ = 6.75, P = 0.0008) and time (F₂,₄₁ = 9.10, P = 0.0005), along with significant combination × time (F₆,₄₁ = 5.24, P = 0.0004), combination × sex (F₃,₄₁ = 5.42, P = 0.003), and time × sex (F₂,₄₁ = 8.05, P = 0.001) interactions; however, the three‑way interaction (P = 0.071) and the sex main effect (P = 0.130) were not significant. For LIPC, the combination (F₃,₄₀ = 4.99, P = 0.005) and time (F₂,₄₀ = 6.13, P = 0.005) main effects, as well as the combination × sex interaction (F₃,₄₀ = 3.05, P = 0.040), were significant; no other main effects or interactions reached statistical significance (P > 0.05).Table 1ART non-parametric three-way ANOVA statistics for lipid metabolism-related gene expression.GeneEffectF valueP valueABCA1combinationF3, 39 = 7.930.0003timeF2, 39 = 12.91< 0.0001sexF1, 39 = 37.43< 0.0001combination × timeF6, 39 = 7.10< 0.0001combination × sexF3, 39 = 27.82< 0.0001time × sexF2, 39 = 21.20< 0.0001combination × time × sexF6, 39 = 7.14< 0.0001ABCG1combinationF3, 42 = 6.710.0008timeF2, 42 = 11.530.0001sexF1, 42 = 5.280.0266combination × timeF6, 42 = 4.630.0011combination × sexF3, 42 = 9.84< 0.0001time × sexF2, 42 = 8.220.001combination × time × sexF6, 42 = 3.510.0066ApoBcombinationF3, 41 = 6.750.0008timeF2, 41 = 9.100.0005sexF1, 41 = 2.390.13combination × timeF6, 41 = 5.240.0004combination × sexF3, 41 = 5.420.0031time × sexF2, 41 = 8.050.0011combination × time × sexF6, 41 = 2.120.0714LIPCcombinationF3, 40 = 4.990.0049timeF2, 40 = 6.130.0048sexF1, 40 = 0.120.7339combination × timeF6, 40 = 1.560.1849combination × sexF3, 40 = 3.050.0396time × sexF2, 40 = 1.190.3156combination × time × sexF6, 40 = 1.820.1193 Table 1ART non-parametric three-way ANOVA statistics for lipid metabolism-related gene expression.GeneEffectF valueP valueABCA1combinationF3, 39 = 7.930.0003timeF2, 39 = 12.91< 0.0001sexF1, 39 = 37.43< 0.0001combination × timeF6, 39 = 7.10< 0.0001combination × sexF3, 39 = 27.82< 0.0001time × sexF2, 39 = 21.20< 0.0001combination × time × sexF6, 39 = 7.14< 0.0001ABCG1combinationF3, 42 = 6.710.0008timeF2, 42 = 11.530.0001sexF1, 42 = 5.280.0266combination × timeF6, 42 = 4.630.0011combination × sexF3, 42 = 9.84< 0.0001time × sexF2, 42 = 8.220.001combination × time × sexF6, 42 = 3.510.0066ApoBcombinationF3, 41 = 6.750.0008timeF2, 41 = 9.100.0005sexF1, 41 = 2.390.13combination × timeF6, 41 = 5.240.0004combination × sexF3, 41 = 5.420.0031time × sexF2, 41 = 8.050.0011combination × time × sexF6, 41 = 2.120.0714LIPCcombinationF3, 40 = 4.990.0049timeF2, 40 = 6.130.0048sexF1, 40 = 0.120.7339combination × timeF6, 40 = 1.560.1849combination × sexF3, 40 = 3.050.0396time × sexF2, 40 = 1.190.3156combination × time × sexF6, 40 = 1.820.1193 ART non-parametric three-way ANOVA statistics for lipid metabolism-related gene expression. Discussion The present study systematically investigated the resistance of four Pekin duck hybrid combinations to duck hepatitis A virus type 3 (DHAV-3) infection. By analysing mortality phenotypes, liver biochemical parameters, and the expression of lipid metabolism-related genes, we elucidated the impacts of hybrid combination, maternal origin, and sex on disease resistance, thereby providing critical insights for dissecting disease-resistant heterosis. With regard to mortality phenotypes, the RS hybrid combination exhibited the most pronounced disease resistance advantage. The mortality rate of female ducks in the RS group was found to be significantly lower than that observed in the SR group, as well as lower than both the RR group and the mid-parent value. This finding suggests the presence of a distinct overdominant effect. This finding suggests that non-additive genetic effects play a pivotal role in disease resistance. Conversely, the SR combination did not demonstrate a comparable advantage, suggesting that maternal origin exerts a decisive influence on the transmission of disease resistance. This is evidenced by the observation that resistant maternal lines are more efficient in conferring disease resistance to offspring. This observation is consistent with the established roles of maternal genetic background, mitochondrial inheritance, maternal effects, and sex chromosome-linked genes in the regulation of disease resistance. However, the present study merely corroborates the existence of non-additive effects, and it is unable to differentiate between dominance, overdominance, and epistasis. Further validation is necessary via gene segregation and quantitative trait locus (QTL) analysis. The present study has built upon previous research conducted on the subject of DHAV-3 resistance in Pekin ducks, with the present study focusing on lipid metabolism-related parameters. Viral infection has been demonstrated to induce substantial perturbations in lipid-associated pathways in Pekin ducks (Liang et al., 2025), indicating that host metabolic reprogramming may serve as a pivotal determinant of viral pathogenicity and host resistance heterogeneity. Therefore, a further systematic dissection of the differences in disease resistance among hybrid combinations was conducted from three perspectives: the baseline of lipid metabolism, biochemical dynamics, and related gene expression. A study was conducted to ascertain whether pre-infection liver biochemical baselines and PCA could identify CHOL and LDL-C as core indicators distinguishing pre-infection states across combinations. The study concluded that this was indeed the case, thus highlighting lipid metabolic homeostasis as a fundamental underpinning of disease resistance (Yang et al., 2024b; Zheng et al., 2024). Critically, the RS combination, despite exhibiting only moderate (rather than extreme) CHOL and LDL‑C levels, demonstrated optimal resistance. We interpret this as follows: a moderately low cholesterol state, particularly with respect to LDL‑C (Tang et al., 2021; Casiano Matos et al., 2024), is more conducive to viral defence, as many enveloped viruses, including DHAV‑3, exploit cholesterol‑rich lipid rafts for cellular entry and membrane fusion. Moderate cholesterol levels limit viral internalization efficiency without compromising the membrane fluidity required for functional immune receptor signaling (e.g., Toll‑like receptor assembly and type I interferon induction), thereby establishing a favorable immunological setpoint prior to infection (Jiao et al., 2024; Wang and Gao, 2024). In contrast, the elevated cholesterol levels observed in SR and some other combinations have been shown to be associated with unfavourable outcomes following infection (Mo et al., 2025; Wang et al., 2025), thus providing a biochemical rationale for the development of disease‑resistant breeding techniques. Post-infection plasma biochemical analyses revealed that DHAV-3 infection significantly disrupts lipid metabolism and liver function: ALT and AST levels increased significantly at 72 hpi, indicating exacerbated liver injury in the late infection phase; LDL-C peaked at 48 hpi and subsequently declined, showing a phase-specific fluctuation pattern. The temporal main effect of CHOL was non-significant, as this was primarily driven by genetic background. Furthermore, the investigation revealed that CHOL and HDL-C exhibited stable inter-combination differences, thus serving as reliable biochemical markers of genetic background. The present study revealed that the SR combination resulted in higher levels of CHOL and HDL-C, while the RS combination resulted in lower levels of both. However, these results must be interpreted with caution, as the study's findings do not fully align with previous research. Importantly, the RS combination did not simply show the lowest values; rather, it exhibited a dynamically balanced trajectory—moderate baseline levels, a controlled post‑infection peak, and a timely return toward homeostasis. This flexible response, which we term “metabolic flexibility”, appears to be the key protective attribute (Fig. 5). Further more, precise nutrition meets the demand for efficient production of high-quality and healthy meat, and has numerous recognized theoretical and practical application advantages (Hu et al., 2025). Our research results can provide certain reference significance for precise nutrition of meat ducks.Fig. 5Schematic model illustrating the proposed mechanism of resistance versus susceptibility in Pekin duck hybrids after DHAV‑3 infection. This conceptual diagram, created with BioRender.com, depicts a metabolic balance model comparing the RS (resistant) and SR (susceptible) hybrid combinations. The central balance scale represents lipid metabolic homeostasis as the key determinant of disease outcome. Schematic model illustrating the proposed mechanism of resistance versus susceptibility in Pekin duck hybrids after DHAV‑3 infection. This conceptual diagram, created with BioRender.com, depicts a metabolic balance model comparing the RS (resistant) and SR (susceptible) hybrid combinations. The central balance scale represents lipid metabolic homeostasis as the key determinant of disease outcome. At the molecular level, genes related to lipid metabolism, including ApoB, LIPC, ABCG1, and ABCA1, exhibited differential expression patterns post-infection, thereby further substantiating the aforementioned inferences. ApoB, a structural protein of atherogenic lipoprotein particles, has been demonstrated to reflect lipoprotein biogenesis and transport status (Sniderman et al., 2019; Borén et al., 2024). LIPC has been shown to be closely involved in reverse cholesterol transport and lipoprotein metabolism (Teng et al., 2019; Desrochers et al., 2022). ABCA1 (Schmitz et al., 2005; Zhang et al., 2022) and ABCG1 (Schmitz et al., 2001; Dahi et al., 2024) primarily mediate cholesterol efflux and lipid homeostasis maintenance. The differential expression of these genes across various combinations, time points, and sexes suggests that viral infection induces a reshaping of the host lipid metabolic network, which is strongly dependent on genetic background. In particular, the expression changes of ABCA1 and ABCG1 align with cholesterol homeostasis regulation in the RS combination, suggesting that the cholesterol efflux pathway may be a critical node mediating disease resistance differences. We further hypothesize that the RS background facilitates a “reset” of the lipid setpoint after the acute infection phase, as reflected by the return of LDL‑C and HDL‑C toward baseline levels at 72 hpi. It is noteworthy that the RS combination did not manifest an “extreme value” pattern (i.e., uniformly lowest or highest levels) in lipid profiles or gene expression; instead, it exhibited a more balanced metabolic phenotype. This finding is consistent with the observed lower mortality rate, suggesting that disease resistance does not depend on extreme alterations in individual lipid parameters but rather on the overall coordination of lipid homeostasis (Liu et al., 2021; Shin et al., 2025). Synthesizing these observations, we propose a coherent mechanistic narrative: the RS genetic background establishes a moderate baseline cholesterol state that attenuates viral entry; upon infection, it mounts a flexible transcriptional and biochemical response that efficiently clears excess lipids, curtails ER stress and mitochondrial damage, and limits hepatocellular necrosis, while rapidly restoring metabolic equilibrium. This integrated capacity—acting as a dynamic buffering system—accounts for the superior survival outcomes, and distinguishes the RS combination from both the metabolically “over‑reactive” SR and the inherently susceptible RR groups. Conclusions This study reveals that the variation in DHAV-3 resistance among Pekin duck cross-breeds is primarily caused by differences in host lipid metabolic homeostasis. Pre-infection baseline levels of CHOL and LDL-C are robust predictors of viral susceptibility. Furthermore, post-infection lipid dynamics and the coordinated expression of cholesterol efflux genes (e.g. ABCA1, ABCG1 and LIPC) are crucial biological mechanisms underlying host defense. Notably, the superior resistance phenotype of the RS combination underscores the pivotal role of specific reciprocal crossing and maternal genetic effects in disease inheritance. Strategic breeding designs can ultimately enhance offspring disease resistance by improving overall lipid metabolic networks, offering a novel approach to the molecular breeding of DHAV-3-resistant poultry. Author contributions SYL and YT conceived and designed the experiments; SYL, HHL, BL, MJZ, MTW, DWL and ZBG performed the experiments; SYL, YT and SSH supervised the work; SYL wrote the original draft; YT reviewed and edited the paper. All authors have read and agreed to the published version of the manuscript. Funding This work was supported by 10.13039/501100012166National key research and development program of China (2024YFF1000900) and intensive recruitment of high-level talents for scientific research projects in Hebei Province (2025HBQZYCXY017 and 2024HBQZYCXY031). Disclosures The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.