In search of a mouse model of Spaceflight Associated Neuro-ocular Syndrome using one-carbon genetics

Abstract

Spaceflight Associated Neuro-ocular Syndrome (SANS) represents a group of ocular signs and symptoms affecting some astronauts, with genetic and metabolic factors associated with its development. This study investigated whether genetic disruption of one-carbon metabolism in a murine model, specifically via Shmt1 knockout (KO) and folate restriction (FR), could induce retinal changes mimicking SANS. Wild-type (WT) and Shmt1 KO mice were maintained on either a folate-sufficient (FS) or FR diet, and retinal thickness was assessed via optical coherence tomography at multiple time points. Red blood cell folate concentrations were significantly lower in FR mice but were not further affected by Shmt1 KO. Retinal thickness declined with age in FS WT mice, while FR WT mice exhibited early reductions in retinal thickness, suggesting a congenital effect. Shmt1 KO mice on FR diets showed delayed retinal thinning, potentially due to compensatory metabolic mechanisms. No overt retinal pathology was observed, as expected, without other environmental stressors. These findings suggest that folate availability influences retinal structure and may contribute to SANS susceptibility. Additional research is required to examine the combined impact of diminished folate availability alongside a spaceflight stressor, such as radiation, CO₂, or microgravity, on retinal thickness. Such work could potentially lead to a murine model of SANS.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-50609-9.

Introduction

As spaceflight missions have increased in duration, ophthalmic changes, including optic disc edema, chorioretinal folds, retinal nerve fiber layer (RNFL) thickening, hyperopic refractive shifts, and globe flattening, have been identified in some astronauts during and after flight¹. These ocular changes, collectively referred to as Spaceflight Associated Neuro-ocular Syndrome (SANS)², likely involve a complex interplay of factors related to the unique environment of space including radiation, gravity, and CO₂ levels³. Both the etiology and pathophysiological cascade that lead to this range of visual and neurological signs and symptoms are unknown.

In 2012, Zwart et al. identified higher circulating concentrations of homocysteine and other biomarkers of one-carbon (1-C) metabolism in astronauts who developed SANS compared to those who did not⁴; it was later revealed genetic variants in the 1-C pathway may sensitize some astronauts to SANS⁵. Specifically, an elevated number of minor alleles of methionine synthase reductase (MTRR) 66 and major alleles of serine hydroxymethyltransferase 1 (SHMT1) 1420 were associated with an increased risk of development of SANS signs in some astronauts. These findings contributed to the hypothesis that genetics and B-vitamin status predispose some astronauts to develop SANS^6,7.

Folate status can impact vision by influencing retinal health and function⁸. Specifically, adequate folate levels support the health of the retinal pigment epithelium (RPE) and other critical components of the visual system while folate deficiency can lead to impaired retinal cell function and increased oxidative stress, contributing to visual impairments and congenital disorders^9,10. The expression of folate binding protein 1 during eye development, observed as early as embryonic day 10.5 in the neural folds and tubes¹¹, underscores the critical role of folates during early ocular development. Evidence from animal studies suggest that maternal folate deficiency particularly affects embryonic ocular structures derived from the optic cup and cranial neural crest, with these effects becoming more pronounced over prolonged periods of deficiency⁹.

Elevated circulating homocysteine concentrations have also been implicated in retinal diseases in humans and animals, including retinal vessel occlusive disease, optic atrophy, and macular degeneration^12,13. Mechanistically, homocysteine may contribute to elevated oxidative stress within the retina, which is already prone to oxidative damage due to its high concentration of mitochondria that support the degree of metabolic activity and reactive oxygen species development. Furthermore, homocysteine impairs endothelial function by reducing the production of nitric oxide¹⁴. This impairment can lead to reduced blood flow and compromised vascular integrity within and around the eye. Homocysteine also induces structural alterations in ocular tissues, affecting components such as the RPE, optic nerve, and cornea¹⁵. While folate deficiency often leads to homocysteinemia, folate deficiency can occur without an accompanying increase in homocysteine levels¹⁶. Furthermore, tissues can be folate-deficient even with adequate systemic circulating folate due to issues such as impaired folate transport, increased tissue demand, or metabolic dysfunctions that limit folate utilization¹⁷. Factors such as genetics, enzyme deficiencies, competition with other nutrients, medications, or localized disease processes can contribute to this mismatch.

The SHMT1 gene is highly conserved between humans and mice. In humans, there are single nucleotide polymorphisms in SHMT1 that can affect enzyme function, nuclear localization¹⁸, and folate metabolism, which may be linked to various diseases, including cancer and cardiovascular diseases^19,20. A genetic knockout (KO) of Shmt1 in mice disrupts cytoplasmic 1-C metabolism, leading to impaired nucleotide synthesis and neural tube defects^17,21. While some compensation by Shmt2 occurs in the mitochondria, it is insufficient to fully restore normal folate metabolism, resulting in a range of physiological consequences²¹.

To date, no animal models of SANS exist. There are, however, multiple murine models that exhibit altered 1-C biochemistry^22–24. In this study, we sought to determine whether genetic variation of 1-C metabolism in a murine model could be employed to mimic SANS. We hypothesized that mice with genetically-induced alterations in 1-C metabolism (via Shmt1 KO) and/or altered B-vitamin status (via folate-restricted diet) would result in increased retinal thickness and/or other ophthalmic pathologies (Fig. 1). Development of an animal model for SANS would greatly expand research capabilities informing space medicine, with spinoff applications for terrestrial medicine.

Fig. 1.

Study design by genotype and diet. Schematic overview of experimental cohorts used to examine the effects of folate sufficient (FS; 2 mg/kg) and folate restricted (FR; 0 mg/kg) diets in wild-type (WT) and cytosolic serine hydroxymethyltransferase 1 (Shmt1) knockout (KO) mice. (A) Primary cohort (n = 91) used for optical coherence tomography (OCT) imaging (faded black lines) and terminal histopathological analysis (solid black lines). Mice were maintained on FS or FR diets from gestation through 9 months of age, with total retinal thickness (TRT) measured by OCT at 3, 6, and 9 months in independent age groups. Mice of the same color represent animals that were utilized for repeat OCT measurements. (B) Separate, age-matched cohort (n = 33) used for red blood cell (RBC) folate measurements at 6 months of age. A subset of animals (n = 15) overlapped between the histopathology (solid black lines) and RBC folate cohorts. Body weight measurements at 9 months (n = 107) were obtained from an independent cohort with no overlap with the OCT cohort (not pictured). Sex distribution per group and timepoint is indicated. Images created using BioRender.com.

Results

Impact of folate restriction and Shmt1 KO on RBC folate levels

Consumption of a folate restricted diet (FR) significantly reduced RBC folate concentration in mice compared to a folate sufficient diet (FS). This effect is evident across both WT and Shmt1 KO genotypes (Fig. 2). The Shmt1 KO genotype did not exhibit an additional significant reduction in RBC folate levels under FR conditions. Similarly, RBC folate levels in Shmt1 KO mice fed a FS diet were comparable to WT-FS mice.

Fig. 2.

RBC folate levels by diet and genotype. RBC folate status (mean + SD) for age-matched mice fed either a folate sufficient (blue) or folate restricted (red) diet stratified by genotype (WT vs. Shmt1 KO). n per group is indicated and represents number of animals. Statistical significance was assessed using two-way ANOVA (diet × genotype). A significant main effect of diet was observed (p < 0.0001).

Total retinal thickness trends with age, diet, and genotype

In FS-fed mice, TRT by OCT measurement was significantly lower at age 9 months when compared to both 3-month-old (p < 0.05) and 6-month-old (p < 0.05) WT mice (Fig. 3). TRT trended similarly in Shmt1 KO mice but did not reach significance. Across all age groups, TRT values in Shmt1 KO-FS mice were consistently lower than those in in WT-FS mice (Fig. 3).

Fig. 3.

Retinal thickness trends with age, diet, and genotype. Total retinal thickness (defined as distance between inner limiting membrane and retinal pigment epithelium, as measured by OCT-based global thickness maps) in µm as mean + SD for mice aged 3, 6, and 9 months across experimental groups (WT-FS, WT-FR, Shmt1 KO-FS, and Shmt1 KO-FR). Symbols indicate genotype (WT: square; Shmt1 KO: open triangle), and color indicates diet (FS: blue; FR: red). n per group is indicated and represents the number of animals (single eye per animal). Statistical significance was determined using two-way ANOVA (age × experimental group), followed by Tukey’s HSD post-hoc test for multiple comparisons.

Under FR conditions, the decrease in TRT with age was not apparent in WT but was observed in Shmt1 KO mice (Fig. 3, p < 0.05). Three-month-old WT mice fed FR diets exhibited significantly lower TRT compared to age-matched FS-fed WT mice (p < 0.05). Shmt1 KO mice fed FR diets at 3 months exhibited significantly higher TRT compared to their WT counterparts (p < 0.05).

It is also important to mention that diet significantly affected the body weight of mice at 9 months, with FR-fed mice weighing less than FS-fed mice irrespective of genotype (p < 0.05; Fig. S3). However, the weights of the FR-fed mice were still within the expected weight range for 9-month-old mice²⁵. There was no discernible retinal or ocular pathology observed across diet groups or genotypes. H&E-stained sections from each group are shown in Figs. S4–S7, and independent review by an ophthalmic pathologist (R.R.) identified no abnormalities in retinal or other ocular structures.

Interobserver variability and measurement methodology

Bland-Altman analysis comparing TRT measurements between two observers revealed minimal interobserver variability (average percent difference was 1.1%), affirming the reliability of manual segmentation across OCT images (Fig. S2). Prior reports^26,27 have demonstrated a strong agreement between retinal thickness measurements obtained via OCT and histopathological imaging. TRT obtained via histopathological imaging was compared to OCT-derived values by aligning inferior and superior histological cross-sections with the global thickness map, focusing on sections corresponding to 0.4 mm from the optic nerve head (Table S1). The average percent difference in TRT was 5%, though variability was higher in histopathological measurements, consistent with known effects of tissue processing on retinal morphology. TRT variability increased closer to the optic nerve head for both techniques, with histopathological imaging exhibiting significantly greater variability overall. It must be acknowledged that the two different methodologies were applied to two different groups of mice.

Discussion

SHMT1 is essential for cytosolic folate metabolism, and its impairment has been linked to genomic instability due to uracil misincorporation into DNA, especially under folate-deficient conditions. Given its central role in nucleotide biosynthesis and cellular methylation, SHMT1 deficiency could plausibly affect development and maintenance of retinal structures, where nucleotide demand is high. Our findings align with MacFarlane et al., demonstrating dietary folate strongly influences systemic folate levels²¹. However, the absence of a significant effect of Shmt1 deficiency on RBC folate suggests that dietary folate availability is the primary determinant of RBC folate status. This underscores the importance of dietary folate in maintaining RBC folate levels irrespective of Shmt1 genotype and highlights the potential for compensatory mechanisms, such as increased thymidylate synthase (TYMS) and thymidine kinase 1 (TK1) expression previously observed in Shmt1 KO mice²⁸. These compensatory increases may contribute to mitigation of the anticipated detrimental effects of Shmt1 KO on systemic folate levels and nucleotide synthesis, preserving cellular homeostasis and genome stability. While the present study identifies associations between folate status, genotype, and retinal structure, it was not designed to directly interrogate the underlying molecular or biochemical mechanisms.

Age-related thinning of the retina, particularly within the RNFL, has been well-documented²⁹ and is attributed to loss of retinal ganglion cells and their axons³⁰. Our results corroborate this, with significant decreases in TRT observed in FS-fed WT mice from 3 to 9 months of age. The lack of a similar age-dependent trend under FR conditions, however, suggests that dietary folate may influence the retinal aging process. Notably, 3-month-old WT-FR mice had reduced TRT compared to FS counterparts, consistent with the hypothesis that elevated homocysteine under folate restriction may exacerbate thinning¹⁰. In contrast, Shmt1 KO mice fed FR diets displayed TRT values comparable to FS-fed WT mice at this time point. These results suggest an earlier onset of retinal thinning in WT-FR mice, whereas Shmt1 KO-FR mice exhibit a pattern more consistent with later-onset thinning. While this distinction may reflect differences in underlying biological processes, the cross-sectional design of this study does not allow definitive differentiation between developmental and degenerative mechanisms and should therefore be interpreted as hypothesis-generating. The absence of significant retinal thinning in Shmt1 KO mice under FR conditions may reflect robust compensatory mechanisms of increased expression of TYMS and TK1 that would be expected. TYMS inhibition has been shown to inhibit mitosis of various cell groups of the involved ocular structure, leading to ocular pathologies, including decreased retinal thickness³¹. Compensatory increases in TYMS activity in Shmt1 KO mice may underscore the adaptive capacity of the retina to buffer disruptions in 1-C metabolism. Furthermore, SHMT1 deletion has been proposed to favor remethylation of homocysteine to methionine, as Methylenetetrahydrofolate reductase and SHMT1 compete for methylene tetrahydrofolate³², thereby potentially limiting the rise in homocysteine levels that may contribute to retinal pathologies. Because different animals were assessed at each time point, longitudinal progression within individual mice cannot be determined, limiting the ability to distinguish between developmental and degenerative processes.

Our findings partially align with reports by Ceylan et al., which observed thinner RNFL in elderly patients with serum folate deficiency³³, but suggest that genetic context and age may critically modify retinal vulnerability to folate insufficiency. While our study did not detect significant differences in TRT between 9-month-old WT and Shmt1 KO mice, a trend toward reduced TRT under FR was seen primarily in Shmt1 KO mice, supporting the role of folate in maintaining retinal structure. This pattern is consistent with prior observations in this study that age-related retinal changes may not be uniform across genotypes and may reflect differential adaptive or inflammatory responses rather than structural degeneration alone. The lack of retinal pathology, though, suggests that the observed TRT changes in this study are not associated with overt retinal damage. Additionally, it is noteworthy that serum folate as reported by Ceylan et al. and RBC folate measurements reported herein may reflect different aspects of systemic folate status, with RBC folate providing a longer-term indicator of folate availability. Given the retina’s dependence on sustained folate levels for nucleotide synthesis and cellular maintenance³⁴, RBC folate may serve as a more reliable indicator of folate status in the retina compared to serum folate, which reflects short-term dietary intake. This distinction is critical for interpreting folate deficiency in both human and murine models. The slight but significant reduction in body weight observed in 9-month-old FR-fed mice contrasts with prior studies that reported no weight differences under normal fat diets with or without folate supplementation. Although we cannot establish a direct correlation between weight and TRT in our study due to separate cohort measurements, other mouse models of disease such as hyperglycemia, have observed a correlation between decreased body weight and reduced retinal thickness³⁵. Further studies are warranted to elucidate these complex relationships and better understand the role of folate metabolism in retinal health.

A recent study by Zarrinbakhsh et al.³⁶ used a hindlimb unloading mouse model to simulate microgravity-induced headward fluid shifts and reported transient elevations in intraocular pressure (IOP) accompanied by significant thinning of the RNFL, particularly in the peripapillary region during the unloading phase. Notably, RNFL thinning occurred despite expectations that fluid shifts might promote tissue thickening, highlighting that microgravity analogs can induce neuroretinal structural loss rather than uniform retinal edema. In contrast, the present study did not measure IOP and did not resolve layer-specific thickness changes but instead quantified TRT at regions distal to the optic nerve head. Thus, TRT reductions observed under folate-restricted and Shmt1-deficient conditions may reflect contributions from multiple retinal layers, including but not limited to the RNFL. It is also possible that metabolic or inflammatory stress initially produces transient retinal thickening or edema that is not captured at the time points or spatial resolution of this study, but ultimately results in loss of neurons and nerve fibers, leading to net retinal thinning. Such temporal progression cannot be resolved with the current dataset but represents a plausible mechanism linking disrupted 1-C metabolism to later structural loss.

Collectively, concordance with published human RNFL thinning under folate deficiency, agreement between OCT and histopathological measurements, and independent pathology review support the biological validity of the observed retinal structural changes and provide a mechanistic foundation for integrating metabolic susceptibility into future SANS-relevant models. While these results support the use of 1-C metabolic perturbation as a biologically meaningful contributor to retinal health, several methodological and translational limitations warrant consideration before its use as a SANS model. First, the present study was designed to examine genetic and nutritional susceptibility factors within an adult physiological context rather than to replicate the temporal profile or environmental complexity of a spaceflight mission. As such, this model represents a component of a broader framework rather than a complete phenocopy of spaceflight-induced ocular disease.

From a measurement standpoint, TRT was derived from global thickness maps approximately 400 μm from the optic nerve head due to limitations in resolving the Bruch’s membrane opening in murine OCT images. Since SANS ocular changes in astronauts are most prominent in the immediate peripapillary and optic nerve head regions, these features cannot be directly assessed in mice with the current approach. Accordingly, global TRT measurements obtained distal to the optic nerve head may not capture localized peripapillary changes or layer-specific alterations characteristic of SANS. In addition, opposing structural changes across retinal layers, such as thinning of neuronal layers with concurrent thickening of vascular or inflammatory compartments, may obscure localized effects when TRT is used as the primary outcome measure. Small group sizes reduce statistical power, and the use of independent animals at each time point precludes longitudinal analysis, limiting the ability to distinguish developmental from degenerative processes. Furthermore, the use of separate cohorts for OCT imaging and histopathological analysis constrained direct cross-modality comparisons. Although histopathological measurements showed general concordance with OCT-derived values, formalin fixation and paraffin embedding are known to introduce non-uniform tissue shrinkage, limiting the precision of absolute thickness comparisons. From a biological and mechanistic perspective, our designation of “folate restriction” rather than “depletion” acknowledges the potential contribution of endogenous folate sources, such as bacterial production via coprophagy; however, the physiological implications of this distinction remain unclear. Additionally, we did not quantify key biomarkers of 1-C metabolism, including homocysteine, S-adenosylmethionine, or S-adenosylhomocysteine, which would further clarify mechanistic links between folate status and retinal structural changes.

Finally, SANS is hypothesized to follow a multi-hit model^7,37 in which genetic susceptibility interacts with environmental stressors unique to spaceflight. The present model captures the genetic and nutritional components of this framework but does not incorporate key environmental factors such as microgravity, elevated CO₂, or radiation exposure. Together, these limitations highlight important areas for refinement in future studies. Incorporation of layer-specific and peripapillary OCT analyses, direct measurement of one-carbon metabolic intermediates, and integration of spaceflight-relevant stressors, such as hindlimb suspension, hypercapnia, or low-dose radiation, will be essential for developing a more comprehensive and translationally relevant murine model of SANS and for evaluating potential countermeasures.

Conclusion

This study demonstrates that dietary folate availability is the primary determinant of systemic folate status when comparing the effects of dietary folate and the Shmt1KO genotype. Dietary folate influences retinal structure in a murine model only when combined with a Shmt1 deficiency in younger animals. Folate sufficient WT mice exhibited age-dependent declines in TRT and folate restricted mice had earlier reductions, suggesting that folate availability modulates retinal structural integrity in younger animals. While these findings do not recapitulate canonical SANS-associated ocular changes during spaceflight, they identify 1-C metabolism as a biologically relevant susceptibility factor that can influence retinal structure. SANS remains a significant concern for astronauts on long-duration space missions, and although ground-based and spaceflight human studies are ongoing, the development of an animal model would provide a complementary platform to accelerate mechanistic discovery. This work establishes a framework for integrating metabolic and genetic risk factors into future models of SANS and represents an initial step toward developing a translationally relevant murine model.

Methods

Experimental animals and diets

All animal experiments were approved by the Institutional Animal Care and Use Committee (Texas A&M University, College Station, TX; AUP# 2019 − 0278, approved 2020) and followed National Institutes of Health (NIH) guidelines for the Care and Use of Laboratory Animals and all applicable federal and state laws. The study is reported per ARRIVE guidelines (https://arriveguidelines.org).

Mice were originally purchased from Jackson Laboratories (Bar Harbor, ME). The establishment and breeding of the Shmt1 KO mouse strain have been previously described²¹. Briefly, male and female Shmt1^+/− breeder mice were crossed to produce Shmt1^−/−, Shmt1^+/−, and Shmt1^+/+ offspring; only Shmt1^−/− (Shmt1 KO) and Shmt1^+/+ (WT) offspring were matured for the purpose of this study. The experimentally-defined diets used in this study were control AIN93G containing 2 mg/kg folic acid (FS) and AIN93G lacking folic acid (Dyets; Bethlehem, PA). Due to the likelihood of coprophagy and bacterial folate availability through gut fermentation, we designated this group folate restricted (FR) instead of folate deficient. The breeder mice were housed according to their diet in cellulose-bedded cages and fed their respective diet over 8 weeks before breeding pairs were established to obtain the experimental Shmt1 KO and WT mice. Experimental diets were randomly assigned and continued through gestation and from weaning until sacrifice. Study design with number of animals per group are included in Fig. 1.

The primary study cohort consisted of mice (n = 91) allocated to OCT imaging and terminal histopathological analysis. Mice assessed at 3, 6, and 9 months of age represent independent age groups rather than repeated longitudinal measurements within the same animals. RBC folate measurements (n = 33) and body weight data at 9 months (n = 107) were obtained from separate, age-matched cohorts that were part of parallel studies conducted under identical breeding, dietary, and housing conditions. The majority of these animals were not generated specifically for the present study; rather, available samples and data were incorporated to minimize additional animal use in accordance with the principles of Reduction. A subset of animals (n = 15) was additionally bred to support histopathological analyses, and RBC samples collected from these animals at 6 months were included in the RBC folate dataset. The body weight cohort was independent and did not overlap with the OCT imaging cohort.

RBC folate analysis was incorporated after initiation of the study to confirm systemic folate status in response to dietary manipulation, and therefore was performed using available age-matched cohorts rather than the primary longitudinal imaging cohort. Body weight data were similarly derived from an existing cohort maintained under identical experimental conditions. All cohorts were standardized with respect to genotype, diet, housing conditions, and study timelines and were studied over overlapping timeframes.

All animals were euthanized using carbon dioxide inhalation, consistent with American Veterinary Medical Association Guidelines for the Euthanasia of Animals. CO₂ was introduced gradually into the chamber to minimize distress, at a fill rate of 20–30% of the chamber volume per minute, followed by cervical dislocation to ensure death, following institutional and ethical guidelines. Euthanasia was performed following 9-month OCT measurement(s), with histopathological samples collected shortly thereafter.

Genotyping

WT and Shmt1 KO mice were identified by PCR genotyping of tail snips with expected amplicon sizes of 780 bp for the wild-type allele and 460 bp for the deletion allele (Texas A&M Institute of Genomic Medicine; College Station, TX). The gene-specific primers have been detailed elsewhere²¹. Sex was determined by PCR genotyping of tail snips using established primers targeting the Sex-determining Region Y (SRY) gene (280 bp amplicon size for male) and later confirmed by gross examination at weaning. This approach ensured accurate sex assignment during early postnatal stages when mice were being allocated to experimental cohorts and maintained on specific diets.

Optical coherence tomography

At designated timepoints (Fig. 1), mice were placed in an induction chamber and anesthetized with isoflurane (3%; Dechra; KS, USA). The mice were then placed on a heating pad and maintained on isoflurane at a 2% flow rate for the duration of imaging. One drop of 1% tropicamide ophthalmic solution (Akorn; IL, USA) was placed in the left eye for 2 min to allow for full dilation. A micro contact lens (PMMA, 3.2 mm diameter, Cornea Surfaces R1.70 mm; Cantor+Nissel; UK) was rinsed clean using sterile buffered solution (Boston Simplus Multi-Action Solution, Bausch+Lomb; NY, USA) and placed on the fully dilated left eye. A sterile eye wash solution (Advanced Eye Relief, Bausch+Lomb; NJ, USA) was used to keep the cornea and contact lens hydrated throughout the imaging. A drop of 0.3% hypromellose gel (GenTeal Tears; Alcon Laboratories; TX, USA) was placed on the eye after contact lens removal for corneal hydration/lubrication. The contact lens was cleaned between each mouse using sterile buffered solution.

Heidelberg Engineering Spectralis HRA + OCT imaging platform (Heidelberg Engineering; Heidelberg, Germany) and Heidelberg Eye Explorer software were used to acquire in vivo retinal image scans³⁸. The IR + OCT infrared setting was selected with the following parameters: OS left eye, 30° angle, 42.00 D focus, IR 100% intensity power, and OCT volume mode. For OCT control, eye setting “small” was selected and enhanced depth imaging mode was turned off. Adjustments to the focus and brightness of the camera and/or positioning of the mouse head ensured clear image resolution and quality. The retinal image scan was acquired with the optic nerve head centered (in the fundus image box) and the peripapillary retina in view. Specific scan parameters were as follows: horizontal scans, ART 100 frames, 20 × 30, 25 B-scan sections. All OCT imaging was performed by a trained analyst using a standardized acquisition protocol on the same imaging platform. Imaging conditions were standardized across animals and time points, including anesthesia depth, pupil dilation, use of a micro contact lens, optic nerve head centering, and fixed scan parameters. Retinal layer segmentation of the retinal pigment epithelium (RPE) and the inner limiting membrane (ILM) were performed using the Eye Explorer software with manual correction. A standard operating procedure for manual segmentation of murine retinal layers was written and adopted prior to initiation of full data analysis to ensure consistent application of segmentation criteria across all images.

Traditionally, OCT captures images with the human corneal curvature of 7.7 mm. The murine corneal curvature is approximately 1.6 mm³⁹. Adjusting the cornea curvature in the software resulted in no change in vertical distance, thereby suggesting that total retinal thickness (TRT) measurements obtained by vertical distance differences between the ILM and RPE are reliable between human and mouse models. However, adjusting the corneal curvature resulted in a 2.5-fold lateral distance difference, and thus when producing a global thickness map over the optic nerve head with the ETDRS grid in the Eye Explorer™ software, the standard human eye settings of 1, 2, and 3 mm from the Bruch’s membrane opening (BMO) equate to approximately 0.4, 0.8, and 1.2 mm for a mouse. The BMO cannot be clearly resolved in murine OCT images. Subsequently, to calculate average TRT from the global thickness map with the ETDRS grid, the center region is excluded as it is expected to contain the BMO while all other regions are averaged. This calculation results in TRT at approximately 0.4 mm from the BMO (Fig. S1). Thus, TRT represents an averaged global retinal thickness measurement distal to the optic nerve head rather than a peripapillary or ONH-centered metric. The primary analyst for OCT spectral image processing was blinded to diet group and genetics. To ensure reliability in spectral imaging measurements, a secondary analyst performed manual segmentation and measurements on five randomly selected spectra (from different timepoints and genetics). Each analyst produced a global thickness map, and the averaged TRT was calculated as described above. Inter-observer reproducibility of OCT-derived TRT measurements was assessed as described in the Statistical Analyses section.

Red blood cell (RBC) folate analysis

Ethylenediaminetetraacetic acid (EDTA) whole blood was collected via tail stick from 18 WT and 16 Shmt1 KO mice (Fig. 1B). Blood was mixed with 1% L-ascorbic acid (Sigma Aldrich; St. Louis, MO) (prepared fresh on the day of use) in a 1:11 ratio and then frozen at −80 °C. The hemolysate was shipped from Texas A&M laboratories to the NASA Johnson Space Center on dry ice and frozen at −80 °C until analysis (within 30 days of collection). RBC folate concentrations were quantified by using a chloramphenicol-resistant Lactobacillus rhamnosus microbiological assay kit from the Centers for Disease Control (Total Folate Kit, Centers for Disease Control and Prevention; Atlanta, GA). Outliers were evaluated for using the IQR strategy described previously. One outlier was removed.

Histopathological analysis of mouse retina

Enucleation and tissue collection were performed at age 9 months on a single eye per animal, corresponding to the eye used for OCT imaging (left eye in all cases except one animal, for which the right eye was collected due to technical considerations) (Fig. 1). Harvested samples were placed in acetic zinc fixative (AZF) solution (Excalibur Pathology; Norman, OK) and stored and shipped at room temperature until they were further processed (within 1 week) for formalin-fixed paraffin embedding (FFPE) and sectioning at Excalibur Pathology. FFPE sections were stained with H&E and then TRT was measured using Aperio ImageScope imaging system (Leica Biosystems; Deer Park, IL) for 20 cross sections. Pathological analysis was also conducted in the retina and in other ocular tissues, including the cornea, lens, ciliary body, and optic nerve. Pathological evaluation of the retina and other ocular tissues, including the cornea, lens, ciliary body, and optic nerve, was performed by Excalibur Pathology and independently reviewed by a board-certified ophthalmic pathologist (R.R.).

Statistical analyses

Male and female mice were included in all experimental groups, with sex distribution by genotype, diet, and timepoint provided in Fig. 1. Sexes were combined for statistical analyses as the study was not powered to detect sex-specific differences, and no a priori sex-based hypotheses were defined. Effects of diet and genotype on RBC folate status and weight were assessed via two-way analyses of variance (ANOVA). TRT was analyzed using two-way ANOVA with age (3, 6, and 9 months) and experiment group (combined genotype and diet: WT-FS, WT-FR, Shmt1 KO-FS, and Shmt1 KO-FR) as factors, followed by Tukey’s HSD post-hoc testing where appropriate. Since genotype and diet were combined into a single factor for this analysis, interaction effects between genotype and diet were not independently assessed. Normality was assessed with the Shapiro-Wilk’s test and was satisfied for all analyses. Groups were considered statistically different when the p-value ≤ 0.05.

Outliers were identified using the interquartile range (IQR) method, defined as values falling outside Quartile 1 or Quartile 3 ± (1.5 × IQR), and were removed prior to analysis. This resulted in the exclusion of two outliers from TRT analyses and one outlier from RBC folate analyses. A Bland-Altman plot was used to assess the difference in averaged TRT calculation between the two OCT analysts. All statistical analyses were performed in GraphPad Prism V.10.2.3 (GraphPad Software; Boston, MA).

Conference presentation

Preliminary data was presented at the 2024 NASA Human Research Program Investigators’ Workshop and the 2024 American Society for Nutrition conference.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

H.B. conceived the study, designed the methodology, conducted the investigation, curated the data, performed formal analysis, and wrote the original draft. C.A. contributed to the investigation and participated in drafting and revising the manuscript. T.M. contributed to the investigation and data curation and participated in drafting and revising the manuscript. R.R. contributed to methodological development and manuscript revision. D.P. contributed to the investigation and manuscript revision. T.H. contributed to methodology, provided resources, and revised the manuscript. S.Z. co-conceived the study and contributed to drafting and revising the manuscript. S.S. co-conceived the study and contributed to drafting and revising the manuscript. P.S. co-conceived the study, acquired funding, and contributed to drafting and revising the manuscript. All authors reviewed and approved the final manuscript.

Data availability

The authors declare that the data supporting the findings of this study are available within the paper and its Supplementary Information files. Should any raw data files be needed in another format, they are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.