Time-of-day difference in iron supplementation in iron-deficient pregnant model mice

Abstract

During pregnancy, iron supplementation is commonly recommended once daily, partly because hepatic hepcidin secretion following iron intake suppresses subsequent iron absorption. However, the optimal timing of iron administration during maternal iron deficiency remains unknown. We established a pregnant mouse model of iron deficiency and compared the effects of iron (FeSO₄, 1 mg/kg) supplementation at the beginning (ZT12) and end (ZT0) of the daily active phase of pregnant mice on maternal, placental, and fetal outcomes. Supplementation at ZT12 significantly enhanced placental iron transport-related mRNA expression, increased placental iron storage, and improved fetal weight and survival compared to the iron-deficient or iron supplementation at ZT0. Both iron deficiency and iron supplementation markedly altered the maternal gut microbial composition; in particular, Proteobacteria, which are widely reported to be associated with intestinal inflammatory status, were significantly elevated in mice supplemented at ZT0. The rhythmicity of key placental iron transport genes expression was altered by iron deficiency or Clock mutations. This study proposes a chrono-nutritional strategy to maximize iron transport and fetal benefits.

Introduction

Pregnancy anemia remains a critical global public health challenge. According to the World Health Organization, 36% (95% uncertainty interval, 34–39) of pregnant women aged 15–49 years worldwide were affected by anemia in 2019¹. This condition is strongly associated with adverse maternal outcomes, including mortality, preterm birth, placental abruption, and neonatal complications, such as stillbirth, growth restriction, and neurodevelopmental deficits, which pose severe risks to both maternal and fetal health^2–4. Iron deficiency is the primary cause of anemia during pregnancy. Therefore, most international guidelines recommend universal iron supplementation (100–200 mg daily) starting in the first trimester to prevent and manage this condition⁵. Oral iron therapy, particularly low-cost iron salts, such as ferrous sulfate, remains the cornerstone of treatment owing to its proven efficacy, safety, and affordability, despite frequent reports of gastrointestinal side effects^6,7.

Iron absorption is initiated in the duodenum, where the divalent metal transporter 1 (DMT1) mediates luminal iron uptake into enterocytes. Basolateral export via ferroportin (FPN1) is coupled with transferrin (Tf) binding for systemic distribution⁸. During pregnancy, increased maternal blood volume and high placental–fetal demand enhance both intestinal absorption and hepatic mobilization of iron⁹. The syncytiotrophoblast layer of the placenta expresses transferrin receptor (Tfr) on its maternal-facing membrane to internalize Tf-bound iron, which is subsequently transported to the fetus via placental FPN1¹⁰. At the cellular level, iron-responsive element binding protein 1 and 2 (IRP1/2) bind iron‑responsive elements (IREs) in target mRNAs to control the translation and stability of DMT1, Tfr, and FPN1, thereby maintaining cellular iron homeostasis¹¹. Although intestinal iron absorption is tightly regulated, a portion of dietary iron remains unabsorbed and persists in the intestinal lumen, where it becomes accessible to resident microorganisms¹². Excess luminal iron can catalyze the formation of reactive oxygen species, damage the intestinal barrier, and alter microbial ecology by favoring the expansion of siderophilic bacteria¹³. Conversely, iron deficiency can also disrupt the gut ecosystem by limiting bacterial growth and reducing the abundance of gut microbial composition¹⁴.

Importantly, multiple physiological processes involved in iron homeostasis during pregnancy may be influenced by circadian regulation. The circadian clock, an approximately 24-h endogenous oscillation system, is maintained even in the absence of external cues, such as light or food¹⁵. The circadian clock regulates not only sleep–wake cycles but also diverse physiological processes, including feeding behavior, autonomic nervous system activity, and endocrine function¹⁶. In mammals, circadian rhythms are organized by a hierarchical clock system, in which central and peripheral clocks coordinate daily physiological processes across multiple tissues. The core molecular mechanism of the circadian clock is a transcriptional–translational feedback loop involving clock genes such as Clock (circadian locomotor output cycles kaput), Bmal1 (brain and muscle ARNT-like 1), Per (period), and Cry (cryptochrome)^15,16. These genes exhibit ~24-h rhythmic expression and are essential for circadian regulation.

In the intestine, the absorption of nutrients and pharmaceuticals exhibits diurnal fluctuations governed by the circadian clock^17,18, suggesting that the absorption of elements, such as iron, might be controlled by the circadian rhythm. A study on piglets revealed that the transcriptional levels of iron-related intestinal transporters exhibit time-dependent rhythms¹⁹. Hepcidin, a hepatic hormone that degrades FPN to limit iron efflux, decreases systemically during pregnancy and enhances maternal iron absorption and placental transfer. Oral iron intake triggers an immediate hepcidin surge, creating a feedback loop that suppresses subsequent absorption. Consequently, an excessive single dose or frequent daily supplementation may paradoxically diminish iron uptake²⁰. Notably, circulating hepcidin exhibits a distinct diurnal rhythm, remaining low in the morning and increasing throughout the day^21,22, independent of dietary iron intake²¹. This rhythmic pattern suggests that morning iron supplementation, when hepcidin levels are lowest, may facilitate more efficient absorption. These findings underscore the need to optimize the iron formulation, dosage, frequency, and timing of antenatal care. Although prior studies have explored dose and frequency adjustments^20,22,23, significant gaps persist in understanding the optimal timing of iron supplementation, such as the circadian rhythm window and its regulatory effects on placental–fetal iron homeostasis.

Whether core circadian genes directly regulate this process, and how pregnancy, a state with elevated iron demand, modifies circadian-driven iron absorption remain unclear. Three sequential experiments were conducted in this study. In experiment 1, we established an iron-deficient (FeD) pregnant mouse model by comparing the mice’s subjective morning (beginning of the daily active phase of pregnant mice, ZT12) and evening (end of the active phase, ZT0) iron supplementation. In experiment 2, we mapped the circadian patterns of placental iron metabolism and clock genes. In experiment 3, we used clock-mutant mice to understand the role of clock genes in the rhythm of iron metabolism during pregnancy.

Results

Iron supplementation at ZT12 was more effective than at ZT0 in a pregnant model of iron deficiency

FeD model mice were induced from 4 weeks of age by feeding an iron-deficient diet, and iron supplementation at zeitgeber time (ZT; ZT0 and 12 were defined as the light on and off times, respectively) 0 or 12 was initiated starting from gestational day 7 (Fig. 1A). Specifically, ZT12 (lights off, RT 8 PM) corresponds to the beginning of the active phase (the “morning” for mice), and ZT0 (lights on, RT 8 AM) corresponds to the end of the active phase (the “evening” for mice). Samples were collected on embryonic day 17 (E17) at ZT6. After conception, the weight of pregnant mice gradually increased during gestation (Fig. 1B). The control group demonstrated the highest weight compared to the FeD (p < 0.0001), ZT0Fe (p < 0.0001), and ZT12Fe (p < 0.05) groups. Although iron supplementation failed to restore weight to control levels, ZT12Fe (p < 0.01) but not ZT0Fe (p > 0.05) showed greater weight recovery than FeD. At E17, the maternal serum iron levels were significantly lower in the FeD model than in the control (p < 0.0001), and no recovery was observed in the supplemented groups (Fig. 1C). Serum ferritin, a common indicator of iron storage, showed a significant decrease in the FeD (p < 0.0001 vs. control), with incomplete restoration in the supplemented groups (Fig. 1D). Maternal serum UIBC and TIBC showed no significant changes among the FeD and iron-supplemented groups, suggesting that maternal transferrin saturation and systemic iron transport capacity were not markedly affected by the timing of iron supplementation (Fig. 1E, F). Although serum ferritin levels increased after iron supplementation, they did not fully reach control values, indicating that dams remained mildly iron-deficient at the time of sampling. This partial repletion was expected, given the reduced supplementation dose. Visual inspection revealed pale coloration and developmental abnormalities in FeD fetuses, with frequent stillbirths retaining intact placentas (Fig. 1G). The stillbirth rate decreased from 7 (7.7%) in the FeD group to 2 (1.9%) in the ZT0Fe group, with complete elimination in the ZT12Fe group (Fig. 1H). Stillbirth rates differed significantly among groups (Fisher’s exact test, p = 0.016). The FeD group exhibited a markedly higher stillbirth rate compared with both the Control (p = 0.0045) and ZT12Fe groups (p = 0.0047), whereas the difference between ZT0Fe and ZT12Fe was not statistically significant (p = 0.246). These findings indicate that iron supplementation, particularly at ZT12, markedly reduced stillbirth incidence to a level comparable with the Control group. Total uterine mass (including fetuses and placentas) relative to maternal weight showed no intergroup differences (Fig. 1I). After isolating the fetuses and removing the placenta, individual fetal weights were recorded. The average fetal weight in the FeD group was significantly lower, with partial recovery in the supplemented groups (Fig. 1J). ZT12Fe fetuses exhibited higher mean weights than ZT0Fe fetuses (p < 0.0001). Four placentas and their corresponding fetuses were randomly selected from each dam for further analysis. Placental acid-extractable iron levels were significantly reduced in the FeD groups, with no restoration in the supplemented groups (Fig. 1K, L). Furthermore, tissue iron storage (ferritin) recovered in the ZT12Fe groups (p < 0.01 vs. FeD; p < 0.05 vs. ZT0Fe), whereas the ZT0Fe groups showed no improvement (Fig. 1M). Fetal ferritin levels were significantly lower in the FeD group and remained decreased in all supplemented groups (Fig. 1N). In summary, morning iron supplementation (ZT12Fe) demonstrated superior efficacy in improving FeD-related maternal and fetal changes compared to evening iron supplementation (ZT0Fe).

Fig. 1. Morning iron supplementation (ZT12) is more effective than evening supplementation (ZT0) in a pregnant model of iron deficiency.

A Schematic of the experimental design for evaluating iron supplementation at zeitgeber time (ZT) 12 and ZT0 in a pregnant iron-deficient (FeD) mouse model. Samples were collected at ZT6 on embryonic day 17 (E17). B Maternal body weight during gestation. Two-way ANOVA with Tukey’s multiple-comparison test (within each time point) is shown in the table. C Maternal serum iron levels at E17. D Maternal serum ferritin levels. E Maternal serum UIBC levels. F Maternal serum TIBC levels. G Representative photos of fetus in control, FeD, and Embryolethality. H Table showing fetal number and fetal death rate per group. I Total uterine mass (fetuses + placentas) to maternal weight ratio. J Fetal average weight. K Placental acid-extractable iron levels. L Fetal acid-extractable iron levels. M Placental ferritin levels. N Fetal ferritin levels. Data are presented as mean ± SEM. Statistical analysis was performed using one-way or two-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Group allocation is shown in (A). B–F include 32 pregnant mice divided into four groups (n = 8 per group). J shows fetal numbers as detailed in Table H. For (K, L), four embryos and their corresponding placentas were randomly selected from each pregnant mouse (32 mice total), yielding 128 fetal/placental samples (32 per group). For (M, N), two embryos and their corresponding placentas were randomly selected from each pregnant mouse, yielding 64 fetal/placental samples (16 per group).

Timing of iron supplementation differentially affected changes in maternal gut microbiota

Because both iron deficiency and iron supplementation affect the bacterial composition of the mouse intestine²⁴, maternal colonic luminal contents collected at E17 were analyzed to characterize the maternal gut microbiota. Sequencing data aligned to the SILVA database revealed distinct microbial community structures across groups at the phylum (level 2) and genus (level 6) levels (Fig. 2A, B). Species diversity analysis revealed that FeD significantly reduced α-diversity (p < 0.0001 vs. control), which was not restored by iron supplementation (Fig. 2C). Bray–Curtis analysis showed that microbial community structures were distinctly different among the four groups, including ZT0Fe vs. ZT12Fe groups (Fig. 2D). Thus, dietary iron deficiency and subsequent iron supplementation altered both the host iron status and the composition of the gut microbiota. We further examined the intergroup abundance of the gut microbiota at the phylum level (Fig. 2E). Proteobacteria were almost absent in the controls, although their abundance increased significantly in the FeD groups (p < 0.0001 vs. control). Evening iron supplementation at ZT0 further amplified Proteobacteria abundance compared with FeD (p < 0.05), whereas ZT12Fe maintained FeD-like levels (p < 0.05 vs. ZT0Fe). Linear discriminant analysis effect size (LEfSe) (linear discriminant analysis [LDA] threshold = 4) analysis of phyla (level 2) further indicated that Proteobacteria contributed the most to differences between the supplementation groups (Fig. 2F). Given that Proteobacteria are associated with proinflammatory responses²⁵, evening iron supplementation may favor the expansion of inflammation-associated taxa.

Fig. 2. Timing of iron supplementation differentially affects maternal gut microbiota composition.

A Taxonomic classification of microbial communities at the phylum level. B Taxonomic classification at the genus level. C Alpha-diversity analysis, including Shannon index, Observed OTUs, Faith’s PD, and Pielou’s Evenness. D Beta-diversity visualized by Bray–Curtis Principal Coordinate Analysis (PCoA), showing distinct clustering among the four groups. E Relative abundance of major phyla. F Linear discriminant analysis effect size (LEfSe), indicating that Proteobacteria contributed most to the differences between supplementation groups. Statistical analysis was performed using the Kruskal–Wallis test. *p < 0.05, **p < 0.01, ****p < 0.0001. Group allocation is as shown in Fig. 1A. Data were obtained from 32 pregnant mice divided into four groups (n = 8 per group).

To refine the taxonomic resolution, the LDA threshold was increased to four to identify the most contributory taxa from the phylum to the genus (Fig. 3A). Subsequent analysis of intergroup differences revealed that FeD increased the abundance of Parabacteroides, Akkermansia, Hungatella, Clostridium innocuum_group, Lachnoclostridium, Bifidobacterium, Lactococcus, Blautia, and Muribaculaceae (vs. control, Fig. 3B). Although evening (ZT0) supplementation restored Akkermansia and Lactococcus to control levels, it failed to normalize proinflammatory Escherichia–Shigella, which exhibited ZT0Fe-specific enrichment (Fig. 3B). Among the functional bacterial groups, a study on maternal gut microbiota (including humans and mice) revealed that the abundance of Bifidobacterium typically increases during late pregnancy and contributes to neonatal gut preparation and immune priming²⁶. In our study, Bifidobacterium abundance was significantly reduced in both the iron-deficient and iron-supplemented groups compared with controls. Muribaculaceae, a core commensal family in the mouse gut known for its polysaccharide-degrading and propionate-producing capacity²⁷, was likewise markedly decreased in iron-deficient and iron-supplemented groups. Lactobacillus and Lactococcus, commonly recognized as beneficial genera²⁸, declined in the iron-deficient group, showed slight recovery in the ZT0 supplementation group, but remained low in the ZT12 group. Analysis of the functional bacterial groups indicated that anemia during late pregnancy significantly altered the maternal gut composition, accompanied by a reduction in beneficial taxa. Despite the partial improvement in fetal outcomes, low-dose iron supplementation failed to completely normalize the maternal gut microbiota. Collectively, iron supplementation at ZT0 may influence the microbial composition in a manner that favors iron-responsive and potentially proinflammatory taxa.

Fig. 3. Genus-level analysis reveals specific taxa influenced by iron supplementation timing.

A LEfSe cladogram identifying differentially abundant taxa from phylum to genus. B Relative abundance of key genera. Evening (ZT0) supplementation resulted in a ZT0Fe-specific enrichment of proinflammatory Escherichia–Shigella. Statistical analysis was performed using the Kruskal–Wallis test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Group allocation is as shown in Fig. 1A. Data were obtained from 32 pregnant mice divided into four groups (n = 8 per group).

Gene expression related to iron absorption and transport aligned with fetal changes induced by iron supplementation, and related to circadian clock

The expression of iron absorption-related genes was analyzed using tissue RNAs collected at ZT6 (Fig. 1A). In the maternal duodenum, iron deficiency (FeD) upregulated Dmt-1 expression compared to the control (Fig. 4A), with a high expression persisting in ZT12Fe (p < 0.0001 vs. control; p < 0.05 vs. ZT0Fe). The expression levels of Fpn1 and its isoforms remained unchanged. Hepatic Hamp, which encodes hepcidin, showed no significant differences (Fig. 4B). The expression of Tfr, which mediates iron uptake into the liver for storage, was significantly elevated under iron deficiency (p < 0.01, vs. control) and remained high following supplementation in the liver (Fig. 4B). Iron-responsive element-binding proteins (Irp1 and Irp2), which serve as sensors of iron status, showed no significant differences between the groups. In the placenta, Tfr expression was similar between the FeD and control groups but was significantly reduced in the ZT0Fe group (p < 0.05, vs. FeD or ZT12Fe; Fig. 4C). When comparing the supplementation groups, ZT12Fe exhibited significantly higher Tfrc expression than ZT0Fe (P < 0.05). The sole iron exporter, fpn1, which is responsible for transferring iron from the placenta to the fetus, was significantly increased in the ZT12Fe group (p < 0.01, vs. ZT0Fe). In addition, the expression of Irp1 and Irp2 levels was significantly elevated in the ZT12Fe group (p < 0.0001 vs. FeZT0). To explore the potential link between iron absorption and the circadian clock regulation, we assessed the expression of core clock genes. At ZT6, no significant differences were observed in the expression of clock genes in the maternal duodenum or liver (Fig. S1A, B). However, the ZT12Fe group demonstrated a specific increase in placental Bmal1 and Clock expression compared to the other groups (Fig. S1C). Although these results suggest a potential relationship between iron absorption and circadian rhythms, the limited sampling time points preclude definitive conclusions.

Fig. 4. Expression of iron metabolism genes in maternal and placental tissues is aligned with fetal changes.

A Relative mRNA expression of iron absorption-related genes (Dmt-1, Fpn1) in the maternal duodenum. B Relative mRNA expression of iron metabolism-related genes (Hamp, Tfr, Irp1, Irp2) in the maternal liver. C Relative mRNA expression of iron transport-related genes (Tfr, Fpn1, Fpn1a, Fpn1b, Irp1, Irp2) in the placenta. Statistical analysis was performed using one-way or two-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Group allocation is as shown in Fig. 1A. A, B include 32 pregnant mice divided into four groups (n = 8 per group). For (C), 3–4 embryos and their corresponding placentas were randomly selected from each pregnant mouse, yielding a total of 123 fetal/placental samples (Control, n = 28, FeD, n = 31, ZT0Fe, n = 32, ZT12Fe, n = 32).

Iron deficiency altered daily oscillation of iron-related gene expression in the placenta

To elucidate the mechanism of time-dependent effects of iron supplementation, we investigated the circadian rhythmicity of clock- and iron-related gene expressions in the placentas of control and FeD pregnant mice (Fig. 5A). The mean fetal weights were significantly reduced in the FeD groups across all time points (p < 0.0001 vs. control) without rhythmic variations (Fig. 5B). Maternal serum iron analysis showed no intergroup differences owing to the limited sample size; however, rhythmicity was observed in the FeD groups (goodness of fit: 0.001, p < 0.001; Fig. 5C). From each dam, four placentas and their corresponding fetuses were randomly selected. Placental and fetal iron levels in FeD were lower at ZT0 and ZT6 than in the control, although they increased at ZT12 with no difference among the groups (Fig. 5D, E). Placental ferritin levels remained consistently higher in the control group (p < 0.001 vs. FeD at all time points), whereas FeD placentas displayed rhythmic fluctuations (goodness of fit: 0.025, p < 0.02) peaking at ZT12 (Fig. 5F). Fetal ferritin exhibited time × diet interactions; control levels peaked at ZT0 (Goodness of fit: 0.031, p < 0.02), whereas FeD peaked at ZT12 (goodness of fit: 0.003, p < 0.001; Fig. 5G). Collectively, these measurements indicate that placental iron content exhibits rhythmic variation. Iron deficiency altered these rhythms, enhancing oscillations in fetal acid-extractable iron and placental ferritin levels, and shifting the peak of fetal ferritin to the same phase as placental and fetal iron levels. Clock gene expression rhythms in the placenta, especially Clock, Dbp, and Cry1, changed owing to iron deficiency with increased amplitude (Fig. 5H). Irp1/2 showed rhythmic expression, peaking at ZT0 in the control, whereas the peak shifted to ZT12 in the FeD group. To confirm these transcriptional changes at the protein level, we examined placental IRP2 protein expression (Fig. S2). After normalization to GAPDH, IRP2 protein levels were significantly higher in FeD placentas than in controls at ZT12, whereas no difference was observed at ZT0. This result is consistent with the mRNA profile, providing protein-level support for the rhythmic upregulation of Irp2 under iron deficiency. Fpn1 includes an IRE domain in its Fpna isoform (but not in Fpnb)²⁹. Fpn1 expression in the control peaked at ZT0, whereas in the FeD, the peak occurred at ZT12. Isoform analysis revealed that Fpna (IRE-containing) but not Fpnb (IRE-lacking) showed the same peak phase at ZT12. Hamp showed no rhythmicity but increased in FeD at ZT18 (p < 0.01). Tfr in the control displayed high expression at ZT0, whereas FeD exhibited rhythmic expression with a peak at ZT12. In the FeD group, the placental Irp1, Irp2, Fpn1, Tfr, and clock genes expressions peaked at ZT12, which peaks were highly consistent with fetal iron and ferritin levels.

Fig. 5. Iron deficiency alters the daily oscillation of iron-related gene expression in the placenta.

A Schematic of the experimental design for Experiment 2, investigating circadian rhythmicity in control and FeD pregnant mice. B Fetal average weights at four time points across the day. C Maternal serum iron levels. Rhythmicity was observed in the FeD group. D Placental acid-extractable iron levels. E Fetal acid-extractable iron levels. F Placental ferritin levels. G Fetal ferritin levels. H Rhythmic expression of clock and iron metabolism genes in the placenta. In the FeD group, the expression peaks of placental Irp1, Irp2, Fpn1, Tfr, and clock genes shifted to ZT12. The gray shaded areas represent the dark phase of the light/dark cycle. Rhythmicity was analyzed using the cosinor procedure. The values for goodness of fit are presented in the table below. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Data were analyzed by two-way ANOVA followed by Sidak’s post-hoc test. Group allocation is shown in (A). Involves 16 pregnant mice. Under iron deficiency (FeD) and control conditions, samples were collected at four Zeitgeber time points (ZT0, ZT6, ZT12, ZT18), resulting in eight groups (Control-ZT0, Control-ZT6, Control-ZT12, Control-ZT18, FeD-ZT0, FeD-ZT6, FeD-ZT12, FeD-ZT18; n = 2 pregnant mice per group). 4 embryos and their corresponding placentas were randomly selected from each pregnant mouse, yielding 64 fetal/placental samples (8 per group).

Clock mutation altered fetal iron condition and placental gene expression rhythms

To investigate the role of circadian clock regulation in placental iron transport, we analyzed pregnant Clock mutant mice. Because of the difficulty of pregnancy in Clock mutant mice under FeD conditions, the study was based on a normal diet with pregnancy in wild-type (WT) and mutant mice, and samples were collected at two critical phases (ZT0 and 12; Fig. 6A). Clock mutations altered fetal viability or weight at each time point (Fig. 6B) and attenuated diurnal variations in maternal serum iron compared to the control (but not significantly because of the number of mice, Fig. 6C). Placental acid-extractable iron levels in the mutants remained comparable to those in the wild type (Fig. 6D), whereas fetal acid-extractable iron decreased at ZT12 (p < 0.05, vs. WT; Fig. 6E). Clock mutants exhibited elevated placental ferritin levels (p < 0.01, two-way analysis of variance [ANOVA]) and significantly increased fetal ferritin at ZT12 (p < 0.01, vs. WT), suggesting aberrant fetal iron accumulation especially at ZT12 (Fig. 6F, G). WT mice maintained the placental rhythmic gene expression of Per1, Clock, Cry1, Irp1, Irp2, Fpn1, and Tfr, whereas rhythmicity was lost in the Clock mutant mice (Fig. 6H). These findings indicate that loss of Clock function disrupts both the placental circadian rhythm and the rhythmic expression of iron metabolism-related genes.

Fig. 6. Clock mutation alters fetal iron condition and placental gene expression rhythms.

A Schematic of the experimental design for Experiment 3, using wild-type (WT) and Clock mutant (Hom) mice. B Fetal average weight. C Maternal serum iron. D Placental acid-extractable iron levels. E Fetal acid-extractable iron levels. F Placental ferritin levels, which were elevated in Clock mutants. G Fetal ferritin levels. H Relative mRNA expression of clock and iron metabolism genes in the placenta. The rhythmic expression of Per1, Clock, Cry1, Irp1, Irp2, Fpn1, and Tfr was abolished in the Clock mutant mice. The gray shaded areas represent the dark phase of the light/dark cycle. Statistical analysis was performed using two-way ANOVA followed by Fisher’s LSD test for post-hoc comparisons. *p < 0.05, **p < 0.01, ****p < 0.0001. Group allocation is shown in (A). 12 pregnant mice were included. Under wild-type (WT) and Clock mutant (Hom) genotypes, samples were collected at two Zeitgeber time points (ZT0 and ZT12), generating four groups (WT-ZT0, WT-ZT12, Hom-ZT0, Hom-ZT12; n = 3 pregnant mice per group). Four embryos and their corresponding placentas were randomly selected from each pregnant mouse, yielding 48 fetal/placental samples (12 per group).

Discussion

This study demonstrated that the timing of iron supplementation during pregnancy critically influences the maternal, placental, and fetal iron status. Morning supplementation (ZT12Fe, at the beginning of the dark/active phase, subjective morning) was more effective than evening supplementation (ZT0Fe, at the beginning of the light/rest phase, subjective evening) in enhancing placental iron storage, increasing fetal weight, and reducing stillbirths. These physiological benefits were accompanied by the circadian-dependent regulation of iron-related genes and attenuation of proinflammatory shifts in the maternal gut microbiota by iron supplementation.

The dose selection was based on both human clinical guidelines and previous mouse studies^30,31. Ferrous sulfate was selected as a representative and widely used form of oral iron supplementation, considering its availability and cost-effectiveness in both experimental and clinical settings. Here, we used a reduced supplementation dose (1 mg/kg body weight FeSO₄)^32–34. Although some parameters (e.g., maternal ferritin levels) were not fully restored, indicating that dams remained mildly iron-deficient at the time of sampling. This approach avoided the confounding effects of potential iron overload. Collectively, morning supplementation provided greater maternal benefits, as evidenced by the gestational weight gain (Fig. 1B). Fetal recovery analysis revealed that fetuses prioritized iron acquisition even with maternal anemia. Consistently, in a rat model of maternal iron deficiency, iron supplementation during late pregnancy did not restore maternal hepatic iron stores but preferentially delivered iron to the fetuses, in line with our present findings³⁵. Morning supplementation nearly restored fetal weight to normal levels and eliminated stillbirths (Fig. 1H, J). As a critical interface for maternal–fetal iron transfer, placental iron storage directly affects fetal health. Placental ferritin levels (reflecting iron reserves) were higher in the morning group, indicating enhanced placental iron buffering (Fig. 1M). Morning supplementation demonstrated superior outcomes across the maternal–placental–fetal axis.

The most common side effect of iron supplementation is that unabsorbed supplemental iron exceeding absorptive capacity remains in the intestinal lumen, generating reactive oxygen species that damage the intestinal mucosa¹². Once the mucosal barrier is compromised, bacteria or toxins can easily translocate into the host and exacerbating inflammation³⁶. Increased unabsorbed supplemental iron also alters the gut microbiota¹⁴. Iron serves as a nutrient source for many pathogens, and iron supplementation can promote their growth, while inhibiting beneficial bacteria, leading to dysbiosis. Both iron deficiency and unabsorbed supplemental iron can alter the intestinal microbial composition and may promote inflammatory responses, as reported in studies and reviews describing iron-induced shifts toward proinflammatory gut microbiota in adults, including women of reproductive age^7,24,37. In this study, we analyzed the maternal colonic contents at E17 and observed reduced microbial diversity during FeD. Although iron supplementation failed to restore α-diversity, β-diversity analysis revealed distinct compositional differences between the supplementation groups, highlighting the profound impact of timing on microbiota structure (Fig. 2C, D).

At the phylum level, Proteobacteria contributed the most to the intergroup differences (Fig. 2F). Under iron-deficient conditions, many pathogenic enteric bacteria require iron acquisition via iron carriers before they can acquire bacterial virulence and gastrointestinal colonization¹⁴. Under conditions of iron deficiency, Proteobacteria dominated through siderophore-mediated iron competition, which was consistent with their increased abundance in the FeD group (Fig. 2E). Proteobacteria are well-established markers of gut dysbiosis and inflammation. The relative abundance of this phylum is significantly increased in conditions such as inflammatory bowel disease and metabolic syndrome²⁵. The lipopolysaccharide derived from the outer membrane of Gram-negative Proteobacteria activates host proinflammatory cytokine release, exacerbating mucosal inflammation³⁸. Evening iron supplementation appeared to further increase the abundance of Proteobacteria and was associated with enrichment of inflammation-related taxa. Genus-level analysis identified Escherichia–Shigella as the most differentially abundant genus, with its abundance increasing under iron deficiency and further increasing with evening iron supplementation, paralleling Proteobacteria trends (Fig. 3). Escherichia–Shigella belongs to the Enterobacteriaceae family within the Proteobacteria phylum and is classified under the Gammaproteobacteria class. Multiple studies have demonstrated a strong association between elevated Escherichia–Shigella abundance and intestinal inflammation. For example, in human studies, an expansion of Escherichia–Shigella has been reported in patients with IgA nephropathy, where its relative abundance correlated with disease onset and treatment response³⁹. Pathogenic Escherichia–Shigella strains can suppress autophagy in epithelial and immune cells, evade immune surveillance, and induce inflammation. Its abundance positively correlates with inflammatory markers, such as C-reactive protein, white blood cell count, and tumor necrosis factor-α, and negatively correlates with serum albumin levels, further supporting its proinflammatory role in gut pathology⁴⁰. In a study on anemic piglets, parenteral administration of iron significantly increased gut Proteobacteria and Escherichia–Shigella⁴¹. These findings aligned with our experiment’s nighttime iron dosing, although the authors did not specify the timing of the injections. Thus, evening iron supplementation (ZT0Fe, at the beginning of the light/rest phase, subjective evening) may cause unabsorbed iron to enter the gut. In maternal anemia, pathogens prioritize iron acquisition via siderophores to exploit the limited iron resources. Unabsorbed supplemental iron perpetuates this hierarchy, allowing pathogens to sequester iron first, with residual iron only later, thus benefiting commensal bacteria⁴⁰. A plausible explanation for the potential presence of unabsorbed luminal iron following evening supplementation (ZT0Fe) is the well-established circadian variation in intestinal motility and bile acid secretion. Studies have indicated that bile acids, which are frequently secreted during active period feeding, enhance digestion and intestinal motility, whereas rest period conversion to secondary bile acids suppresses excessive gut movement^42,43. Thus, at the beginning of the active period, iron supplementation may accelerate iron absorption and facilitate rapid clearance of unabsorbed iron via active period motility. Conversely, at the beginning of the rest period, supplementation coincided with slow motility, leading to prolonged luminal retention of unabsorbed iron. Although residual iron may partially aid in the recovery of certain gut microbiota from anemia, this occurs at the cost of pathogenic overgrowth. However, the present study did not directly quantify residual unabsorbed iron in the intestinal lumen; therefore, only tentative interpretations can be proposed based on the gut microbiota profiles. Specifically, despite identical iron doses, evening iron supplementation (ZT0Fe, at the beginning of the light/rest phase, subjective evening) was associated with an increased abundance of pathogenic bacteria, which may be related to the presence of unabsorbed luminal iron.

When comparing the gut microbiota of late-pregnant mice in our study with findings from previous literature, we observed several striking consistencies despite differences in model and physiological context. For instance, consistent with results from a male colitis model⁴⁴, we also found a significant increase in the phylum Pseudomonadota (Proteobacteria), including potentially pathogenic genera such as Escherichia–Shigella, in iron-deficient pregnant mice. This cross-model convergence suggests that the expansion of Pseudomonadota represents a conserved microbial response to iron deficiency, likely driven by their efficient iron acquisition systems. However, we also identified model-specific discrepancies. For example, the increase in Actinobacteriota reported in male iron-deficient mice was not observed in our pregnant model⁴⁵, and the direction of change in Verrucomicrobiota varied across studies⁴⁴. We observed that the abundance of Verrucomicrobia varied with the timing of iron supplementation. Although its representative genus, Akkermansia, is not considered a classical iron-dependent bacterium¹³. One possible explanation is that increased luminal iron favors iron-responsive taxa, thereby indirectly constraining the niche of Verrucomicrobia at the community level. However, these mechanisms were not directly tested in the present study. These differences highlight the profound influence of pregnancy on the intestinal microenvironment. For the non-pathogenic taxa, we observed that beneficial bacterial groups such as Bifidobacteria, Muribaculaceae, Lactobacillus, and Lactococcus were significantly reduced in both the iron-deficient and iron-supplemented groups. However, the present study cannot disentangle the independent contributions of iron deficiency and supplementation timing to these microbial changes. This finding suggests not only a disruption of maternal physiological adaptation under anemic conditions but also the profound and lasting impact of long-term iron deficiency (induced by the low-iron diet) on both the mother and offspring. Although iron supplementation at ZT12 partially improved fetal and placental parameters, the abundance of beneficial bacteria was not restored, indicating that low-dose iron supplementation could not fully reverse the maternal microbial imbalance caused by anemia (Fig. 3B).

The relative expression of Dmt1 in the maternal duodenum provides a mechanistic explanation for reduced efficiency of iron absorption at night (Fig. 4A). The placenta absorbs maternal iron via the Tfr and exports it to the fetus through the Fpn1. In our RT-PCR analysis of placental tissue, morning iron supplementation significantly upregulated both Tfr and Fpn1 expression, which correlated with the higher ferritin content observed in these placentas (Figs. 1L and 4C). The Irp1/2 exhibited no changes in the maternal liver but displayed rhythmicity in the placenta (Fig. 4B, C). Direct quantification of placental iron transfer using isotope tracing would provide the most definitive evidence and should be incorporated in future studies to validate and expand these findings.

Irp2 expression lacks circadian rhythmicity in normal murine cells but exhibits Clock-dependent oscillations in cancer cells⁴⁶. Our results further demonstrated that placental Clock gene expression exhibited rhythmic patterns analogous to those of Irp1/2. A study has shown that placental cells share striking similarities with cancer cells in their proliferative, migratory, and invasive properties⁴⁷. Trophoblast and cancer cells share similar molecular programs involving cell adhesion, extracellular matrix remodeling, and epithelial–mesenchymal transition (EMT), but this similarity represents an evolutionary parallel or analogy⁴⁸. Comparative analyses across species suggest an association between lower placental invasiveness (e.g., cattle and horses) and a reduced incidence of metastatic cancers relative to species with highly invasive placentation, including humans, cats, and dogs⁴⁹. Another study using whole-genome sequencing of human placental somatic cells demonstrated that human placentas exhibit a high level of somatic mosaicism and mutational burden, showing clonal expansion patterns analogous to those observed in tumors⁵⁰. Inspired by the mechanistic similarities between placental invasion and cancer metastasis described in prior studies, we propose a hypothesis centered on iron, a nutrient with high cellular demand. We hypothesized that placental Irp1/2 may participate in the circadian regulation of iron metabolism, so we conducted multipoint sampling experiments. Strikingly, under iron deficiency, the circadian expression rhythms of placental iron metabolism genes were completely inverted compared with those under normal conditions. Under normal conditions, Clock, Irp1, Irp2, Fpn1, and Tfr expression peaked at night (ZT0), whereas iron deficiency shifted all peaks to the morning (ZT12) (Fig. 5H). This phase reversal may explain the superior iron absorption efficiency of the morning supplementation. Although the underlying mechanism remains unclear, this iron deficiency-induced phase shift may reflect a form of circadian rhythm remodeling, a phenomenon that has been increasingly reported in recent studies. Nutritional status and physiological conditions, such as pregnancy, modulate peripheral clocks^51,52. Iron deficiency-induced alterations in placental phase may also involve oxygen-sensing pathways, such as hypoxia-inducible factor (HIF) signaling. Previous studies have demonstrated that anemia and tissue hypoxia can activate HIF signaling⁵³, and that HIF pathways interact with several core clock genes⁵⁴. Therefore, HIF-mediated oxygen signaling may contribute to the circadian reprogramming of the placenta under iron deficiency, although the precise molecular mechanisms remain to be elucidated in future studies. Therefore, maternal iron deficiency serving as a metabolic cue capable of reprogramming the placental circadian phase even in the absence of changes in the central clock is plausible. When interpreting the time-of-day–dependent effects of iron supplementation, feeding-related influences on iron bioavailability should also be considered. Although iron was administered at fixed Zeitgeber times under controlled light–dark conditions with ad libitum feeding, we cannot fully exclude a contribution of feeding-associated modulation of iron absorption. Thus, the observed effects are more likely to reflect an interaction between circadian timing and physiological context rather than a single dominant mechanism. Although the precise signaling pathways remain unclear, plasticity of the circadian system has been observed in response to dietary and hormonal changes.

Extensive studies have established that Irp1/2 binds to IREs on target mRNAs (e.g., Tfr and Fpn1) to regulate their translation, thereby establishing rhythmic expression patterns^55–57. For Tfr, IRPs directly bind to IREs located in the 3’untranslated region (3’UTR) of Tfr mRNA, regulating its mRNA stability. This mechanism is consistent with the similar rhythmic mRNA expression patterns observed for Irp1/2 and Tfr in the present study and may represent a plausible mechanism underlying the circadian regulation of placental iron uptake. In contrast, the IREs present in the 5’UTR of Fpn1a primarily confer IRP-mediated post-transcriptional regulation at the level of translation efficiency, rather than directly determining mRNA abundance. Although we observed similar rhythmic mRNA expression patterns between Irp1/2 and Fpn1a, these findings do not establish a direct transcriptional regulatory relationship between IRPs and Fpn1a. Whether Fpn1a translation is directly regulated by IRPs in a rhythmic manner requires further validation at the protein level, which was not addressed in the present study. Therefore, we propose that Tfr, as a well-characterized direct IRP target gene regulated through IRE-mediated control of mRNA stability, may serve as a more appropriate indicator of IRP-dependent circadian regulation at the transcriptional level. In the current study, we speculate that this regulatory mechanism may also operate in the placental tissue.

Finally, to examine whether the rhythmicity of placental iron metabolism depends on core clock genes, we used Clock mutant mice. The transcription factors CLOCK and BMAL1 form a core circadian transcription complex¹⁵. Bmal1 knockout mice are unsuitable as they exhibit placental vascular defects that lead to intrauterine fetal death⁵⁸. Clock knockout mice trigger compensatory upregulation of the CLOCK homolog NPAS2, masking the expected phenotype⁵⁹. By contrast, the Clock^Δ19 mutant disrupted the CLOCK function more completely via a dominant‑negative effect, thereby avoiding NPAS2‑mediated compensation⁶⁰. Moreover, a previous study has reported that IRP2 circadian rhythms in cancer cells are controlled by CLOCK, leading us to select Clock^Δ19 mutants for our study⁴⁶. Upon loss of rhythmic Clock expression, the rhythmic expressions of Irp1/2, Tfr, and Fpn1 were abolished in the mutants, confirming that placental iron metabolism rhythms depend on core clock genes.

This study had some limitations. First, most of iron metabolism genes were analyzed only at the transcriptional level. Protein-level validation is required to exclude post-transcriptional modifications and further mechanistic discussions. Second, single-time-point microbiota sampling precludes the dynamic assessment of gestational microbial changes, and functional data (e.g., metatranscriptomics) are required to dissect host–microbe iron competition. Fecal iron concentration was not quantified because it was technically unfeasible to accurately separate unabsorbed iron derived from different supplementation times within a 24-h period. Nevertheless, systemic and tissue iron measurements, together with gut microbiota data, provided indirect but informative indicators of iron utilization and intestinal iron exposure. Additionally, direct quantification of iron transfer using isotope tracing would provide the most definitive evidence. Third, Clock mutant studies were confined to iron-sufficient conditions because combining FeD with circadian disruption causes pregnancy failure, precluding the analysis of anemia–circadian interactions. Finally, we did not employ an intestine-specific clock gene knockout model, which would help disentangle the maternal gut clock contributions from placental rhythmicity. It is important to note that mice are nocturnal, and their feeding and absorption rhythms are inverted relative to humans. Thus, ZT0 and ZT12 in mice do not map directly onto human morning and evening supplementation. While the underlying circadian regulation of iron absorption is conserved, caution is needed when extrapolating these findings to human dietary practices.

This study provides mechanistic insights into gestational iron timing, placental circadian iron metabolism, and microbial interactions. morning supplementation (ZT12Fe) outperformed evening dosing (ZT0Fe) in improving the maternal–placental–fetal iron status while mitigating proinflammatory dysbiosis. We further identified phase-shifted placental iron rhythms in FeD governed by core clock genes. These findings highlight the importance of circadian timing in the regulation of placental iron metabolism and suggest the potential benefits of time-targeted nutritional interventions in pregnancy models. Future clinical studies should investigate the influence of time-of-day–dependent iron supplementation on maternal and fetal outcomes to validate the translational relevance of these findings.

Methods

Experiment 1

Female Cr1:CD1 ICR mice (3-week-old) were divided into four groups (N = 32, N = 8 per group): control, iron-deficient anemia model (FeD), nighttime iron-supplemented group during pregnancy after successful FeD model establishment (ZT0Fe), and morning iron-supplemented group during pregnancy after successful FeD model establishment (ZT12Fe). To establish an iron-deficient anemia (FeD) model, female mice were fed a low-iron diet (3.2 ppm iron, F2FeDD, AIN-93M base, Oriental Yeast Co., Ltd., Japan) for 3 weeks after 1 week of acclimation. Control mice were fed a standard diet (45 ppm iron; AIN-93M; Oriental Yeast Co., Ltd., Japan). All mice were housed alone in clean cages with free access to food and water. At 7 weeks of age, female mice were co-housed overnight with age-matched male Institute of Cancer Research (ICR) mice. Vaginal plugs were checked at 8:00 AM the next morning, and the presence of a plug-confirmed pregnancy (defined as embryonic day 0 [E0]). Given that numerous studies have identified the period from E7 to E15 as a critical window for iron transport and the peak period of iron demand in mouse embryos, we chose to administer iron supplementation for 10 consecutive days^61,62. From E7 to E16, the ZT0Fe and ZT12Fe groups received 1 mg/kg body weight FeSO₄ (Wako Chemicals) once daily at either ZT0 or ZT12, respectively. Although 2 mg/kg body weight FeSO₄ has been commonly used in other studies^32,33, we reduced the dose to 1 mg/kg to avoid iron overload and to better detect the time-dependent effects on iron absorption. To minimize the stress from daily gavage during pregnancy, FeSO₄ was dissolved in condensed milk (10 mg/kg body weight), following the protocol described by Wang et al.⁶³. To control for potential feeding effects, all groups, including controls, received condensed milk at both ZT0 and ZT12, with iron administered only to the treatment groups. The mice were acclimated to condensed milk from feeding dishes for 3 days prior to the intervention. Iron supplementation was terminated on E16 to eliminate acute effects during sample collection. On E17, pregnant mice were euthanized at ZT6, and the maternal serum, liver, duodenum, colon contents, placenta, and fetuses were collected for analysis.

Experiment 2

Female and male Cr1:CD1 ICR mice (3-week-old) were divided into two groups (N = 16, N = 8 per group): control and FeD. The FeD model establishment and pregnancy confirmation methods were identical to those used in Experiment 1. All the mice had ad libitum access to food and water. On E17, pregnant mice were euthanized at ZT0, ZT6, ZT12, and ZT18 (N = 2 per time point), and maternal serum, placenta, and fetuses were collected for analysis. Samples were collected at four Zeitgeber time points (ZT0, ZT6, ZT12, and ZT18), representing 6-h intervals over a 24-h cycle. This four-point design is commonly used in circadian rhythm studies to capture key phases of diurnal oscillation^64,65, while minimizing the number of experimental animals.

Experiment 3

Heterozygous (Het, Clock^Δ19/+) mice were crossed to generate F1 offspring with three genotypes: wild-type (WT, Clock^+/+), heterozygous (Het, Clock^Δ19/+), and homozygous (Hom, Clock^Δ19/Δ19). To establish stable lines for experimental use, wild-type (WT) and homozygous mutant (HOM) mice were bred. Subsequently, female and male Clock^Δ19/Δ19 mutant mice (3-week-old, Cr1:CD1 ICR background) were divided into two groups (N = 12, N = 6 per group): wild-type (WT) and Hom. Owing to pregnancy difficulties in Hom mice under FeD conditions, no FeD model has been established. All the mice were fed a standard diet (AIN-93M; Oriental Yeast Co., Ltd., Japan) using pregnancy confirmation methods identical to those used in Experiment 1. All the mice had free access to food and water. On E17, pregnant mice were euthanized at ZT0 and ZT12 (N = 3 per time point), and maternal serum, placenta, and fetuses were collected for analysis.

Animals

All procedures strictly adhered to the guidelines of The Hiroshima University Committee for Animal Experimentation (approval nos. A23-20-2 and A23-72-3). All the experiments were performed in accordance with the Japanese Government Act for Humane Treatment and Management of Animals (Law No. 105, 1973), and Standards Relating to the Care and Management of Experimental Animals (Notification No. 6, 1980). Pregnant mice were anesthetized with 5% isoflurane delivered via inhalation in an induction chamber. Adequate anesthesia was confirmed by loss of pedal withdrawal reflex before any procedure. Terminal blood collection was performed via unilateral orbital enucleation under deep anesthesia. Mice were subsequently euthanized by cervical dislocation, followed by rapid tissue collection. All procedures were conducted to minimize animal suffering. The mice for experiments 1 and 2 were obtained from Charles River Laboratories, Japan, Inc. Clock^Δ19/Δ19 homozygous mutant used in Experiment 3 were purchased from The Jackson Laboratory (strain #002923, C57BL/6J‑Clock^m1Jt/J, https://www.jax.org/strain/002923). The male mice were maintained on a normal iron diet and used only for mating. The iron-deficient diet was administered exclusively to female mice to establish the maternal iron-deficiency model. Since 2004, C57BL/6J background Clock mutants have been backcrossed with Cr1:CD1 ICR mice at Waseda University for over 15 generations⁶⁶. All animals were housed at 23 ± 1 °C and 60% ± 5% relative humidity under a 12-h light/12-h dark cycle, with ad libitum access to food and water. Zeitgeber time 0 (ZT0) was defined as the time of lights-on, corresponding to the beginning of the rest phase for the nocturnal mice. Conversely, ZT12 (lights-off) marks the beginning of their active phase. In the text, the terms “morning” and “evening” are used from the subjective perspective of the mice (i.e., morning at ZT12 and evening at ZT0).

Sample collection and processing

Maternal Samples: For serum preparation, at least 1 mL of blood was collected in a 1.5-mL microcentrifuge tube and allowed to clot at room temperature for 30 min, followed by centrifugation at 3000 × g for 20 min (25 °C). The supernatant (serum) was aliquoted and stored at −80 °C until analysis. The duodenum and liver were excised and rinsed thoroughly in ice-cold saline (PBS) to remove any residual blood. The duodenum was placed on ice using the TRIzol reagent. The liver was divided into two portions: one was snap‑frozen in liquid nitrogen and stored at −80 °C; the other was placed in TRIzol on ice. For microbiota analysis, colonic contents were collected from the distal colon.

Placental and Fetal Samples: After euthanasia, the uteri were excised and weighed. The fetuses and their corresponding placentae were collected in a left-to-right order (maternal abdominal orientation), rinsed in ice-cold PBS, and weighed individually. The samples were photographed sequentially. Each placenta was then bisected longitudinally: one half was snap-frozen in liquid nitrogen and stored at –80 °C; the other half was placed in TRIzol on ice. Fetuses were euthanized immediately after collection, snap-frozen in liquid nitrogen, and stored at –80 °C. TRIzol-preserved tissues were homogenized using a Tissuelyser II (Qiagen, Hilden, Germany). Homogenates were centrifuged at 1500 × g (4 °C, 10 min) to remove debris. The supernatants were aliquoted and stored at –80 °C.

Biochemical assays

Serum, liver, placental, and fetal iron concentrations were determined using a Metallo Assay Iron LS (Ferrozine Method) kit (FE31M; AKJ Global Technology, Tokyo, Japan). This assay quantifies the acid-extractable iron pool, including both free (labile) iron and non-heme iron released under acidic conditions, according to the manufacturer’s instructions. Serum unsaturated iron-binding capacity (UIBC) was measured using the UIBC Kit (UIB02A; Metallogenics Co., Ltd., Chiba, Japan) according to the manufacturer’s instructions, based on the bathophenanthroline colorimetric method. Total iron-binding capacity (TIBC) was calculated as the sum of serum iron and UIBC values (TIBC = serum iron + UIBC). In fetal iron concentration studies, to reduce tissue heterogeneity and focus on major iron storage organs, such as the liver and spleen, the head and limbs were removed during the analysis. Ferritin levels in the maternal serum and placental and fetal homogenates were assessed using a ferritin (mouse) enzyme-linked immunosorbent assay kit (KA1941; Abnova Corporation, Taiwan). There are currently no standardized reference ranges for serum ferritin in mice; therefore, maternal iron status was evaluated based on relative differences from the control group rather than fixed thresholds⁶⁷.

Real-time RT-PCR

Total RNA was isolated from the samples using a phenol–chloroform isoamyl alcohol-based phase separation method. RNA purity was assessed by measuring the A260/A280 ratio using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA), followed by concentration normalization to 50 ng/μL using DEPC-treated water. The adjusted concentrations were verified before downstream applications. Amplification reactions were performed on a Thermal Cycler Dice Real-Time System III (Takara Bio, Shiga, Japan) using the One-Step TB Green PrimeScript RT-PCR Kit (Takara Bio, Shiga, Japan). Gapdh and Tbp served as endogenous controls for data normalization. Gene expression differences were quantified using the comparative Ct method with the primer sequences listed in Supplementary Table 1.

16S rRNA gene sequencing analysis

Maternal colonic contents collected in Experiment 1 were submitted to Genome Read Co. (Kagawa, Japan) for analysis. Samples were thoroughly mixed with lysis buffer and transferred to tubes containing EZ-Beads (Promega, Tokyo, Japan). Mechanical disruption was performed using a MM400 bead mill at maximum speed for 30 s to ensure efficient cell lysis. The resulting lysates were purified using the GeneFind v2 system (Beckman Coulter, Brea, CA, USA), which utilizes magnetic beads for DNA extraction. PCR amplification targeting the V3–V4 hypervariable regions of the 16S rRNA gene using specific primers. Purified amplicons were used to construct sequencing libraries, and paired-end sequencing (301 bp per read) was performed on an Illumina MiSeq platform (Illumina, San Diego, CA, USA). Raw sequencing data quality was assessed using FastQC (v0.12.1; https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The read counts per sample ranged from 48,913 to 89,208. Based on per‑base quality profiles, a slight drop in quality was observed at the 3′ end of read 2; therefore, low-quality regions were truncated as follows: read 1 was truncated to 270 bp from the 5′ end, and read 2 was truncated to 220 bp from the 5′ end. Primer sequences were removed by trimming 18 and 22 bp from the forward and reverse primers, respectively. Paired-end reads were processed using the DADA2 algorithm in QIIME2 (v2023.2) for error correction, denoising, and amplicon sequence variant (ASVs)^33,68. A minimum overlap of 20 bp was required for reliable paired‑end merging. Chimeric sequences were removed, yielding merged sequences approximately 490 bp in length. The final feature table comprised 619 high-confidence ASVs after chimera removal. Denoising statistics, including the number of non-chimeric reads used for ASV construction, are summarized in sup5. Interactive visualization of these files is accessible through QIIME2 View (https://view.qiime2.org). To account for uneven sequencing depths across samples, rarefaction curves were generated using subsampling reads to evaluate the impact of sequencing depth on the observed ASVs and alpha diversity metrics. Based on the feature table and representative sequences, alpha diversity was assessed by calculating Faith’s phylogenetic diversity, Shannon index, Pielou’s evenness, and observed feature counts. Statistical comparisons were performed using the Kruskal–Wallis test. Beta diversity was analyzed using Bray–Curtis and Jaccard distances, as well as unweighted and weighted UniFrac distances, and statistical comparisons were assessed using permutational multivariate analysis of variance. Principal coordinate analysis was performed to visualize clustering patterns. Taxonomic assignment was performed using the SILVA 138 99% OTUs full-length reference database⁶⁹. Stacked bar plots depicting microbial community composition at the genus level were generated. All-level relative abundance tables were made accessible using the QIIME2 View. LEfSe was applied to identify differentially abundant taxa between the groups at the phylum (L2) and genus (L6) levels. LEfSe and cladogram plots were generated using online platform (https://www.bioincloud.tech). All analytical outputs, including diversity indices, statistical summaries, and interactive visualizations, were stored in QIIME2 artifacts (.qzv), to ensure transparency and reproducibility. As the fecal samples contained high bacterial biomass, no mock community or negative extraction control was included, and sequencing quality assessment confirmed the absence of contamination.

Western blotting

Placental tissues were homogenized using the Easy Lysis Kit (Atto, Tokyo, Japan) following the manufacturer’s instructions. The concentration of extracted proteins was determined with the BCA Protein Quantification Kit (Abcam, Cambridge, UK). Equal quantities of protein were then combined with 2× Laemmli sample buffer, heated at 100 °C for 5 min, and loaded for electrophoresis. Protein separation was performed on c-PAGEL Neo SDS–PAGE gels (Atto, Japan) and transferred onto PVDF membranes (Clear Blot P+ Membrane, Atto). The membranes were blocked with 5% EzBlock Chemi (TBS-T containing 0.05% Tween-20) for 1 h at room temperature, divided into two sections, and subsequently incubated overnight at 4 °C with primary antibodies: anti-IREB2 (Proteintech, Cat# 29976-1-AP; dilution 1:1000) or anti-GAPDH (Cell Signaling Technology, Cat# 2118S; 1:1000). After washing, HRP-conjugated secondary antibody (Multi-rAb™ HRP-Goat Anti-Rabbit Recombinant Antibody, Proteintech, Cat# RGAR001; 1:10,000) was applied for 1 h at room temperature. Chemiluminescent signals were visualized with EzWestLumi One reagent (Atto) using the Lumazone in vivo Imaging System (Shoshin EM Corp., Japan). Signal intensities were processed in VisiView (version 2.1.4, Visitron Systems GmbH, Germany) and quantified in ImageJ. Protein levels were normalized to GAPDH expression.

Statistical analysis

Sample size determined from preliminary experimental data and published studies^37,44,46,70. Data are presented as means ± standard errors of the mean (SEMs). Statistical analyses were performed using GraphPad Prism version 10.3.1 (GraphPad Software, San Diego, CA, USA). Depending on sample size, the Shapiro–Wilk and D’Agostino & Pearson tests were used to assess normality, and the Brown–Forsythe test was used to evaluate homogeneity of variances. Parametric analyses were conducted using one-way or two-way ANOVA with Tukey’s multiple comparison test for post-hoc analysis. Nonparametric analyses were performed using the Kruskal–Wallis test with Dunn’s multiple comparison test for post-hoc analysis. Circadian rhythm was analyzed using the single cosinor procedure program (Acro.ex version 3.5, Dr. Refinetti) to determine the goodness of fit of the normalized data⁷¹. Statistical significance was set at p < 0.05. The significance levels are indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Author contributions

Conceptualization: N.L., S.S., and Y.T.; data curation: N.L. and Y.H.; formal analysis: N.L. and Y.L.; funding acquisition: S.S., H.F., and Y.T.; investigation: N.L., K.H., Y.K., and Y.T.; methodology: N.L., S.S., and Y.T.; project administration: N.L.; resources: T.K., S.S., H.F., and Y.T.; supervision: N.L. and Y.T.; validation: N.L. and Y.H.; visualization: N.L.; writing—original draft: N.L.; writing—review and editing: S.S. and Y.T. All authors read and approved the final manuscript.

Data availability

All data are available in the main text or Supplementary Materials. Raw 16S rRNA sequencing data of the gut microbiota were deposited in the NCBI Sequence Read Archive (SRA) under the accession number PRJNA1271692. The Clock mutant mouse strain can be provided by the corresponding author (Y.T.) pending scientific review and a completed material transfer agreement. Requests for this strain should be submitted to Y.T.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41538-026-00860-1.