Combined dietary omega-3 and omega-6 fatty acids protect against hyperglycemia-associated retinopathy in neonatal mice

Abstract

Retinopathy of prematurity (ROP) with early vessel loss (Phase I) followed by uncontrolled vessel growth (Phase II) causes visual impairment in premature infants. Although supplementation with omega-3 (n-3) docosahexaenoic acid (DHA) alone shows mixed results in preventing ROP, supplementation with both n-3 DHA and n-6 arachidonic acid (ARA) in early postnatal life reduces severe ROP by 50 % (Mega Donna Mega study). In the Mega Donna Mega study, 146 (72.6 %) of 201 included infants had at least one hyperglycemic episode during the first 14 days of life, which is a strong ROP risk factor. We therefore evaluated the protective effects and mechanisms of combined dietary n-3 DHA and n-6 ARA in a neonatal mouse model of hyperglycemia-induced suppression of retinal vascular development (Phase I ROP). At postnatal day (P) 10, retinal vessel growth was improved in pups from mothers on diets enriched with 1 % DHA + 2 % ARA vs. 3 % DHA. Lipid changes in pup plasma and RPE complex (retinal pigment epithelium with choroid and sclera) were in accordance with maternal diets’ DHA and ARA levels, indicating that milk lipids reflected maternal diets. Proteomic retinal analysis revealed increased abundances of proteins related to mitochondrial respiration and glucose metabolism with the combined diet. Inhibition of mitochondrial ATP synthase negated the protective effects of the combined diet. In conclusion, combined DHA+ARA oral maternal supplementation protects against hyperglycemia-induced retinopathy in mouse neonates (Phase I ROP model) through enhanced retinal metabolism, suggesting the potential of balanced lipid supplementation for ROP prevention.

1. Introduction

Retinopathy of prematurity (ROP) is a potentially blinding eye disorder [1] characterized by two phases. In Phase I, physiological retinal vascular growth and neuroretinal development are suppressed [2]; in Phase II, pathological retinal neovascularization is triggered by both local hypoxia in vaso-obliterated retinal areas and oxygen-independent stimuli [3].

Metabolic factors modulate ROP development and progression [4,5]. In Phase I, perinatal hyperglycemia is common, and alterations in omega-3 (n-3) and omega-6 (n-6) long-chain polyunsaturated fatty acid (LCPUFA) levels, particularly n-3 docosahexaenoic acid (DHA) and n-6 arachidonic acid (ARA), occur, as the normal supply from the uterine environment is disrupted.

DHA and ARA, “conditionally essential” fatty acids, are crucial during fetal [6] and postnatal development [7], particularly in the retina [8]. DHA makes up ~ 50 % of total phospholipid fatty acids in retinal photoreceptor outer segments [9], while ARA is enriched in photoreceptors and retinal vessels [10]. After preterm birth, DHA and ARA levels fall dramatically due to disrupted in utero supply of these lipids. Reduced DHA and ARA blood levels persist for at least 4 weeks after birth [11,12] and are associated with increased ROP severity [13,14]. Clinical trials have explored the potential of nutritional/metabolic interventions in ROP [15]. Parenteral supplementation of n-3 LCPUFAs alone yields inconsistent results regarding ROP development and progression [16–22]. The DIAMOND (DHA Intake And Measurement Of Neural Development) study found that supplementation of DHA and ARA (1:2 ratio) in early life improved visual acuity at 12 months of age. Increased DHA supplementation, however, yielded no additional improvement of visual acuity [23]. The recent Mega Donna Mega study showed that enteral supplementation immediately after birth with n-3 DHA (50 mg/kg/d) and n-6 ARA (100 mg/kg/d) at a 1:2 ratio reduced severe ROP incidence by 50 % in extremely preterm infants (< 28 weeks of gestational age) [24], suggesting that the balance between DHA and ARA may be crucial in protecting against ROP. However, the underlying mechanisms are unknown.

Hyperglycemia in the early postnatal period after preterm birth, seen in 45 % of preterm infants with birth weight < 1000 g [25] and 80 % of preterm infants with birth weight < 750 g [26], is associated with increased ROP incidence and severity [27–32]. In this study, we examined postnatal hyperglycemic status in the Mega Donna Mega trial. We tested a balanced maternal rodent diet containing 1 % DHA + 2 % ARA vs. 3 % DHA in hyperglycemia-associated retinopathy in pups, modeling Phase I ROP [33–36]. We have previously shown that maternal diet composition is reflected in milk-fed pups’ retinas [37]. We assessed the impact of DHA+ARA vs. DHA maternal diet on pup retinal vascular development, fatty acid profile of retina, retinal pigment epithelium (RPE) complex, and plasma, as well as retinal proteome, to uncover protective mechanisms of combined nutritional n-3 and n-6 LCPUFAs as a potential strategy for prevention and treatment of Phase I ROP.

The primary aim of this study was to determine whether combined maternal DHA+ARA supplementation provides benefit over DHA alone in mouse models of ROP, as well as to investigate the underlying mechanisms. The 1:2 DHA:ARA ratio was chosen according to the Mega Donna Mega [24] and the DIAMOND [23] clinical trials. We hypothesized that maternal DHA+ARA vs. DHA may be protective in murine ROP models.

2. Materials and methods

2.1. Ethics statement

Mouse experiments followed the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publication 8523, revised 2011) and the ARVO Statement for Use of Animals in Ophthalmic Vision and Research. The work was approved by Boston Children’s Hospital’s Institutional Animal Care and Use Committee (protocol 0001619).

The previously published Mega Donna Mega study, an open-label, randomized controlled trial (ClinicalTrials.gov Identifier: NCT03201588) aimed to evaluate if a combined enteral fatty acid supplementation with DHA and ARA could reduce severe ROP [13,24,38]. The Regional Ethics Review Board in Gothenburg (303–11 and T570–15) and the Swedish Ethical Review Authority (2020–02381) approved the trial protocol. All procedures followed the Helsinki Declaration of 1975 (revised 2008). The WHO Guiding Principles on Human Cell, Tissue and Organ Transplantation are not applicable. Parents or guardians provided written informed consent. We analyzed longitudinal glucose blood samples of infants included in the study born before 28 weeks of gestational age at three neonatal intensive care units in Sweden between 2016 and 2019.

2.2. Mouse model of hyperglycemia-associated retinopathy and quantification

For all animal experiments, C57BL/6J mice (#000664, Jackson Laboratory, Bar Harbor, ME, USA) were bred in-house with 12-hour light/dark cycle and ad libitum chow. Because hyperglycemia in the early postnatal period is a strong ROP risk factor, and as most infants in our clinical study had at least one incidence of hyperglycemia, we used the neonatal hyperglycemia-associated retinopathy mouse model to study lipid intervention in early ROP. The dietary ratio of DHA:ARA (1:2) used in mice was aligned with the Mega Donna Mega clinical trial [24]. To induce hyperglycemia-associated retinopathy (a Phase I ROP model with suppression of physiological retinal vascular growth), mouse pups were intraperitoneally injected with 50 μg/g body weight streptozotocin (STZ, S0130, Sigma-Aldrich, St. Louis, MO, USA; 50 μg/μL in PBS) daily from postnatal day (P) 1–P9 [33]. C57BL/6J moms were fed fully defined rodent feed (Research Diets, Inc., New Brunswick, NJ, USA) including either 3 % DHA or 1 % DHA + 2 % ARA (DHA+ARA) of the diets’ lipid content (10 % of the chow) from P1–P10 (Table 1). DHA and ARA oils were sourced from DSM (Heerlen, Netherlands). At P10, after using a lethal intraperitoneal injection of ketamine/xylazine, plasma, RPE complex (retinal pigment epithelium with choroid and sclera), and retinas of pups with body weight ≥ 3.8 g were collected.

Table 1.

Composition of the rodent diets containing either 3 % DHA or 1 % DHA + 2 % ARA of the diets’ lipid content (10 % of the chow).

Diet name	1 % DHA + 2 % ARA		3 % DHA
Product #	D21032502-38		D21032503S
Macronutrients	gm%	kcal%	gm%	kcal%
Protein	21	20	21	20
Carbohydrate	59.9	58	59.9	58
Fat	10	22	10	22
Total		100		100
kcal/gm	4.1		4.1
Ingredients	gm	kcal	gm	kcal
Casein	200	800	200	800
L-Cystine	3	12	3	12
Corn Starch	337	1348	337	1348
Maltodextrin 10	132	528	132	528
Sucrose	100	400	100	400
Cellulose, BW200	50	0	50	0
Soybean Oil	0	0	0	0
DHASCO (43.3 % DHA)	2.3	21	6.8	61
ARASCO (42.2 % ARA)	4.6	41	0	0
Safflower Oil, High Oleic	90.1	811	90.3	813
t-ButyIhydroquinone	0.014	0	0.014	0
Mineral Mix S10022G	35	0	35	0
Vitamin Mix V10037	10	40	10	40
Choline Bitartrate	2.5	0	2.5	0
FD&C Blue Dye #1	0.04	0	0.04	0
FD&C Yellow Dye #5	0.01	0	0.01	0
Total	966.564	4001	966.664	4002
DHA (g/kg)	1		3
ARA (g/kg)	2		0

DHA, docosahexaenoic acid; ARA, arachidonic acid.

For inhibition of mitochondrial ATP synthase, oligomycin A (0.25 μg/g body weight) (#11342, Cayman Chemical, Ann Arbor, MI, USA; 0.25 μg/μL in PBS with 30 % dimethylsulfoxide) [39] was intraperitoneally injected daily from P7–P9 during deep retinal vessel formation. A higher concentration of oligomycin A (0.5 μg/g body weight) in neonatal mice caused postnatal growth retardation, which affects retinal vascular development (data not shown).

After fixation of the whole eye in 4 % paraformaldehyde, retinas were dissected and permeabilized using 1 % Triton X-100 in PBS. Isolectin B4 (Alexa Fluor 594, I21413, Invitrogen^™, Carlsbad, CA, USA) in 1 mM CaCl₂ in PBS was used overnight to stain vessels. Four images (one per quadrant) were taken between the optic nerve head and the leading vessel edge in the deep and superficial vascular layer on a Zeiss Axio Observer Z1 microscope (200x magnification) [33]. The retinal vascular status (number of meshes and total vessel length per field) was assessed with the “Angiogenesis analyzer” ImageJ plugin. At least 3 litters per group were analyzed.

2.3. Mouse model of oxygen-induced retinopathy and quantification

To validate findings from the hyperglycemia-associated retinopathy model, maternal DHA+ARA vs. DHA diet from P1–P17 was applied in the oxygen-induced retinopathy model. C57BL/6 J pups and their nursing dams were exposed to 75 % oxygen from P7–P12 with hyperoxia leading to retinal vessel regression (vaso-obliteration, VO; modeling oxygen-induced Phase I ROP) and return to room air at P12 resulting in relative hypoxia and subsequent formation of pathological retinal neovascularization (NV; modeling hypoxia-induced Phase II ROP) [40]. At P17, both eyes of pups with body weight 5.0–7.5 g [41] were collected after intraperitoneal injection of ketamine/xylazine.

Retinas were dissected and stained as described above. Images were acquired at 50x magnification with a Zeiss fluorescent microscope. Quantification of VO and NV was performed using ImageJ [42], and the proportions of NV and VO to the total retinal area were compared between diet groups.

2.4. Analysis of postnatal hyperglycemia

In the Mega Donna Mega study, premature infants receiving combined DHA+ARA (1:2 ratio) within 3 days after birth until 40 weeks postmenstrual age vs. standard of care without oral lipid supplementation showed a 50 % reduction in severe ROP [24]. As hyperglycemia in the first weeks of life is associated with increased ROP incidence and severity [27–32], we analyzed maximum daily glucose values over the first 14 postnatal days of infants born before 28 weeks of gestational age. Of the 206 randomized infants, 201 (97.6 %) had available glucose measurements during the first 14 days. Included infants’ (n = 201) mean gestational age was 25.5 weeks (SD ± 1.4), and mean birth weight was 792.1 g (SD ± 195.1). Routine blood samples for glucose measurement were drawn according to local guidelines and clinical indications. All glucose levels were retrieved from the infants’ medical records until 14 postnatal days. When multiple glucose levels were available for an infant on the same day, the highest value during that day was used to define presence of hyperglycemia (> 10 mmol/L) [43].

2.5. Fatty acid profiling

P10 retinal samples of known weight from pups after STZ and maternal DHA+ARA vs. DHA diet (n = 18 samples/group; both retinas pooled) were disrupted with glass homogenizers. For total lipid extraction [44], the following were sequentially added: 1.75 mL 0.88 % potassium chloride, 2 mL methanol, 4 mL chloroform. Heptadecanoic acid (17:0) was used as an internal standard. Total lipids were extracted overnight and then methylated with methanolic boron trifluoride (100 °C, 1 h). Fatty acid methyl esters (FAMEs) were quantified in a gas chromatograph with a flame ionization detector (Varian, Bruker, Billerica, MA, USA). FAMEs were eluted on a DB-FFAP column (30 m × 0.25 mm i.d. × 0.25 μm film thickness) by set program at initial oven temperature of 50 °C for 1 min, following with a ramp-up to 130 °C at a 30 °C/min rate, 175 °C at 10 °C/min, 230 °C at 5 °C/min, then held for 9.5 min, and a ramp-up to 240 °C at 5 °C/min for 11.13 min. Using peak comparison to the internal standard, 31 fatty acid species were quantified (8 species of saturated and 8 monounsaturated fatty acids, 8 species of n-6 and 7 species of n-3 PUFAs). At least 4 litters per group were analyzed.

2.6. Proteomics

Retinas from P10 pups after STZ and maternal DHA+ARA vs. DHA diet (n = 8 mice/group; both retinas pooled) were processed for label-free LC-MS/MS proteomics [39]. At least 2 litters per group were analyzed. An Orbitrap Fusion Lumos coupled to an Easy-nLC1000 HPLC pump (Thermo Fisher Scientific, Waltham, MA, USA) was used for mass spectrometry. Peptide spectra were analyzed with Proteome Discoverer (v2.5, Thermo Fisher Scientific). Included proteins had ≥ 2 unique peptides [39]. For each protein, logarithmic transformation, student’s t-test (p-value), and Benjamini-Hochberg method (FDR-adjusted p-value, called q-value) were performed (Qlucore Omics Explorer, v3.5, Qlucore, Lund, Sweden). Gene Ontology pathway enrichment (Biological Process, Molecular Function, Cellular Component) of significantly decreased or increased proteins (p < 0.05; |log2 fold change| > 0.2) was analyzed using EnrichR (retrieved 10/2024) [45–47]. Pathways with p < 0.05 (Fisher’s exact test) were considered significantly enriched.

Raw data are available on the ProteomeXchange Consortium via the PRIDE [48] repository (identifier PXD061307, https://doi.org/10.6019/PXD061307).

2.7. Statistics

Infant glucose values were displayed graphically as a longitudinal curve using locally estimated scatterplot smoothing (LOESS). For infant characteristics and neonatal morbidities, Fisher’s exact test was used for dichotomous variables, Mantel-Haenszel Chi-square trend test was performed for categorical ordered variables, and Mann-Whitney U test was applied for continuous variables.

Mouse data are presented as mean ± SEM. For non-omics data, normality (quantile-quantile plot) and variance (F-test) were tested, followed by parametric (unpaired t-test or Welch’s t-test) or non-parametric (Mann-Whitney U test) analysis (Prism v9.0, GraphPad Software, La Jolla, CA, USA). Statistical significance was considered at p < 0.05. Heatmaps were created using https://software.broadinstitute.org/morpheus/.

3. Results

3.1. In hyperglycemia-associated retinopathy in neonatal mice (modeling Phase I ROP), maternal DHA+ARA vs. DHA diet promoted physiological vessel growth

In the hyperglycemia-associated retinopathy model, the impact of maternal DHA+ARA vs. DHA diet was examined (Fig. 1A). The diets had no effect on pup body weight and blood glucose levels (Fig. 1B, C). Oral milk from mothers on DHA+ARA vs. DHA maternal diet resulted in increased density of the deep (Fig. 1D, E) and superficial (Fig. 1F, G) retinal vasculature in pups at P10, indicating that combined DHA+ARA vs. DHA lipid diet promoted physiological retinal vessel growth in mice with neonatal hyperglycemia-associated retinopathy.

Fig. 1.

DHA+ARA diet vs. DHA promoted physiological retinal vascularization in hyperglycemia-associated retinopathy in neonatal mice. (A) Schematic of mouse model. To induce hyperglycemia-associated retinopathy, mouse pups were injected daily with streptozotocin (STZ, 50 μg/g body weight intraperitoneally) in PBS from postnatal day (P) 1–P9. Nursing moms were fed a diet supplemented with docosahexaenoic acid alone (DHA) or DHA plus arachidonic acid (DHA+ARA) from P1–P10. (B) P10 mouse pup body weight and (C) blood glucose levels were measured. DHA pups, n = 9–10; DHA+ARA pups, n = 9; unpaired t-test; ns, not significant. (D–G) Retinal vessel analysis of STZ-induced pups at P10. (D, F) Number of vascular meshes (left) and total vessel length (right) per field in deep (D) and superficial (F) retinal vasculature were quantified. DHA pup retinas, n = 17–19; DHA+ARA pup retinas, n = 12–15; unpaired t-test; **** p < 0.0001. (E, G) Representative images of deep (E) and superficial (G) retinal vasculature. Retinal vessels were visualized with isolectin B4 (red). Magnification, 200x; scale, 50 μm.

3.2. In oxygen-induced retinopathy in neonatal mice, maternal DHA+ARA vs. DHA diet promoted physiological vessel growth

To validate findings in hyperglycemia-associated retinopathy, the maternal diets were tested in the oxygen-induced retinopathy mouse model (Supplemental Fig. 1A, B). Body weight was unchanged by the diets (data not shown). Maternal DHA+ARA vs. DHA diet resulted in a reduction of vaso-obliteration, suggesting improved physiological retinal vascularization (Supplemental Fig. 1C, D), while pathological neovascularization was unaffected (Supplemental Fig. 1E, F). These data confirm the protective effect of maternal DHA+ARA vs. DHA diet on physiological retinal vessel growth in two mouse models of Phase I ROP.

3.3. Hyperglycemia incidence in the first two weeks after preterm birth in the Mega Donna Mega study

In the Mega Donna Mega study, the median number of glucose blood samples per infant was 27 (IQR 17–47). The mean glucose level across all days was 7.9 mmol/L (SD ± 2.9), and, considering the maximum glucose levels, the mean was 9.5 mmol/L (SD ± 5.4). Of the 2298 samples, 770 (33.5 %) met the criteria for hyperglycemia (> 10 mmol/L), and 146 infants (72.6 %) had at least one hyperglycemic sample (Supplemental Fig. 2). Infants with hyperglycemia were more immature at birth and more often affected by neonatal morbidities, including ROP (Table 2).

Table 2.

Characteristics and neonatal morbidities of infants with and without hyperglycemia.

Variables	No hyperglycemia n = 55	Hyperglycemia n = 146	p-value
Birth characteristics
Gestational age (weeks)	26.6 (0.9)	25.1 (1.4)	< 0.0001
	26.7 (23.6; 27.9)	25.1 (22.3; 27.9)
	(26.1; 27.3)	(24.1; 26.1)
Birth weight (grams)	948.5 (179.4)	733.2 (166.5)	< 0.0001
	940 (580; 1330)	707 (420; 1345)
	(815; 1120)	(610; 829)
Birth weight SDS (Fenton)	0.38 (0.73)	−0.04 (0.82)	< 0.0001
	0.45 (−1.61; 1.81)	0.06 (−2.29; 2.34)
	(−0.04; 0.89)	(−0.51; 0.41)
Sex, boys	29 (52.7 %)	86 (58.9 %)	0.52
Morbidities
ROP stage			0.0012
No ROP	35 (63.6 %)	60 (41.1 %)
Mild ROP (stages 1 and 2)	14 (25.5 %)	41 (28.1 %)
Severe ROP (stage 3 and treated)	6 (10.9 %)	45 (30.8 %)
ROP treatment	2 (3.7 %)	34 (27.4 %)	0.0002
BPD (Jensen 1, 2, 3)	19 (35.2 %)	79 (64.8 %)	0.0005
NEC	2 (3.6 %)	19 (13.0 %)	0.07
Severe IVH (grades 3 and 4)	3 (5.5 %)	21 (14.4 %)	0.09
Neonatal course first month of life
Mechanical ventilation (days)	3.7 (7.4)	14.3 (10.2)	< 0.0001
	0 (0; 28)	15 (0; 28)
	(0; 4)	(4; 25)
CPAP (days)	16.9 (9.7)	12.0 (9.8)	< 0.0001
	19 (0; 28)	11 (0; 28)
	(7; 27)	(3; 20)
Sepsis	4 (7.3 %)	24 (16.4 %)	0.11
CRP, max (mg/L)	10.1 (19.9)	26.8 (42.6)	< 0.0001
	3.3 (0.3; 129.0)	13.0 (0.3; 300.0)
	(0.9; 11.0)	(4.0; 29.0)
Hb, min (g/L)	107.4 (15.6)	97.6 (14.8)	< 0.0001
	107.0 (56.0; 154.0)	100.0 (38.0; 151.0)
	(101.0; 113.0)	(95.0; 104.0)
Platelet count, min (109/L)	205.2 (79.1)	132.0 (76.8)	< 0.0001
	194 (28; 400)	130 (8; 391)
	(148; 259)	(72; 174)
Glucose, max (mmol/L)	8.5 (2.0)	20.2 (10.1)	< 0.0001
	9 (4; 15)	17 (10; 59)
	(7; 10)	(14; 23)

Infants with at least one blood glucose measurement > 10 mmol/L (n = 146) and infants without any episode of hyperglycemia (n = 55) during the first 14 postnatal days. Data are presented as mean (standard deviation), median (minimum, maximum), and (IQR), or number (percentage). For dichotomous variables Fisher’s exact test was used. For categorical ordered variables Mantel- Haenszel Chi-square trend test was performed. For continuous variables Mann- Whitney U test was applied. BPD, bronchopulmonary dysplasia; CPAP, continuous positive airway pressure; CRP, C-reactive protein; Hb, hemoglobin; IVH, intraventricular hemorrhage; NEC, necrotizing enterocolitis; ROP, retinopathy of prematurity.

3.4. Maternal diet composition was reflected in the pup lipid profile of plasma and RPE complex

Given the frequency of hyperglycemia and its association with ROP in premature infants in the Mega Donna Mega trial, we further focused on the hyperglycemia-associated retinopathy mouse model. At P10 in plasma and RPE complex, DHA levels were significantly increased in DHA diet group pups, while ARA was higher in DHA+ARA diet group pups (Fig. 2A, B), suggesting that fatty acids in the maternal diet can reach the pups via maternal milk and can incorporate into the pups’ ocular tissue. However, in the retina, DHA and ARA were comparable between the diet groups, and higher maternal dietary DHA (3 % vs. 1 %) did not further increase retinal DHA in pups (Fig. 2C), suggesting a tight regulation of local retinal lipid composition [49]. Other retinal (C14:0, C18:2n-6) and circulating (C14:0, C18:0, C20:5 n-3, C22:4n-6) fatty acids were changed by diets.

Fig. 2.

Maternal diet composition was reflected in the pup lipid profile of plasma and RPE complex. Targeted fatty acid analysis of neonatal mice with hyperglycemia-associated retinopathy. P10 (A) plasma, (B) RPE complex (retinal pigment epithelium with choroid and sclera), and (C) retina were isolated from mouse pups. Nursing moms were fed a diet supplemented with DHA or DHA+ARA from P1–P10. DHA pups, n = 18 (2 retinas pooled); DHA+ARA pups, n = 18 (2 retinas pooled); multiple unpaired t-test; * p < 0.05. Only comparisons with statistically significant changes are labeled. Fatty acids on the x-axis are denoted as Cx: yn-z, with x, number of carbons; y, number of double bonds; z, position of the first double bond from the omega (methyl) end.

3.5. The maternal DHA+ARA vs. DHA diet increased retinal proteins related to mitochondrial respiration

We assessed the global proteome of P10 STZ retinas isolated from DHA+ARA vs. DHA maternal diet pups to uncover mechanisms involved in the combined diets’ protective effect on retinal vessels. In total, 4203 proteins with ≥ 2 unique peptides were detected. Each diet group displayed a distinct proteome profile. DHA+ARA vs. DHA retinas showed 198 proteins with increased and 176 proteins with decreased abundance passing p < 0.05 (q < 0.37) and |log2 fold change| > 0.2 (Fig. 3A, B). Gene ontology pathways related to mitochondrial respiration, fatty acid and glucose metabolism were significantly (p < 0.05) enriched (Fig. 3C) with increased protein abundance of mitochondrial complex I subunits (NDUFS8, NDUFB6, NDUFS3) and of glucose transporter 1 (SLC2A1) (Fig. 3D). Pathways with decreased protein abundance in the DHA+ARA vs. DHA group were related to autophagy, pexophagy, mitochondrial apoptosis, lysosomes, endosomes, and negative regulation of growth factor signaling (data not shown).

Fig. 3.

DHA+ARA vs. DHA diet increased mitochondrial proteins in P10 retinas from mouse pups induced with hyperglycemia-associated retinopathy. Nursing moms were fed a diet supplemented with DHA or DHA+ARA from P1–P10. DHA pups, n = 8 (2 retinas pooled); DHA+ARA pups, n = 8 (2 retinas pooled). At P10, mouse retinas were isolated for proteomic analysis. (A) Principal component analysis of all detected proteins (≥ 2 unique peptides) in retinas isolated from mouse pups in DHA+ARA vs. DHA diet group. (B) Volcano plot of proteins with significantly (p < 0.05; q < 0.37; log2 fold change (log2FC) > 0.2 or < −0.2) increased (orange) or decreased (blue) abundance in retinas isolated from pups in DHA+ARA vs. DHA diet group. (C) Selected enriched (p < 0.05) gene ontology pathways associated with energy and lipid metabolism (biological process, BP; cellular component, CC; molecular function, MF) of proteins with significantly increased abundance (p < 0.05; log2FC > 0.2) in retinas isolated from pups in DHA+ARA vs. DHA diet group. Pathways related to oxidative phosphorylation are written in bold. Protein count refers to the number of distinct proteins with significantly increased abundance in the respective gene ontology cluster. Protein ratio describes the proportion of proteins with significantly increased abundance to all listed proteins in a cluster. (D) Heatmap depicting the altered proteins in the selected enriched pathways shown in (C).

3.6. Pharmacological inhibition of mitochondrial ATP production negated the maternal DHA+ARA vs. DHA diet promotion of retinal vessel growth

Next, we tested whether mitochondrial respiration contributed to the combined diet promotion of retinal vessel growth. In hyperglycemia-associated Phase I retinopathy, pups with maternal DHA+ARA or DHA diet received intraperitoneal injection of oligomycin A, inhibiting mitochondrial ATP synthase (Fig. 4A). No differences in body weight and blood glucose levels between maternal diet groups were observed (Fig. 4B, C). After pharmacological suppression of mitochondrial ATP energy production, no differences between DHA+ARA and DHA maternal diet groups in deep (Fig. 4D, E) and superficial (Fig. 4F, G) retinal vasculature at P10 were observed. These findings suggested that mitochondrial energy production was a key component in the combined DHA+ARA diet promotion of retinal vessel growth in neonatal hyperglycemia-associated retinopathy.

Fig. 4.

Pharmacological inhibition of mitochondrial ATP production negated the promotion of physiological retinal vascularization by dietary DHA+ARA. (A) Schematic of intervention. Mouse pups induced with hyperglycemia-associated retinopathy were treated with mitochondrial ATP synthase inhibitor oligomycin A (25 μg/g body weight intraperitoneally) in 30 % dimethylsulfoxide in PBS from P7–P9. Nursing moms were fed DHA or DHA+ARA diet from P1–P10. (B) Body weight and (C) blood glucose levels of P10 oligomycin A-treated pups with maternal DHA or DHA+ARA diet were measured. DHA pups, n = 5–6; DHA+ARA pups, n = 8; unpaired t-test; ns, not significant. (D–G) Retinal vessel analysis of STZ-induced pups at P10. (D, F) Number of vascular meshes (left) and total vessel length (right) per field in deep (D) and superficial (F) retinal vasculature were quantified. DHA pup retinas, n = 12; DHA+ARA pup retinas, n = 16; unpaired t-test or Mann-Whitney U test; ns, not significant. (E, G) Representative images of deep (E) and superficial (G) retinal vasculature. Retinal vessels were visualized with isolectin B4 (red). Magnification, 200x; scale, 50 μm.

4. Discussion

The LCPUFA supplementation approach preventing ROP development and progression should be refined. Our study demonstrates that combined dietary supplementation with DHA+ARA compared to DHA alone protects against hyperglycemia-associated retinopathy in neonatal mice (mimicking Phase I ROP associated with hyperglycemia) by enhancing retinal metabolism. Our findings highlight the importance of maintaining a balance between n-3 and n-6 LCPUFAs in promoting retinal vascular development during the critical early postnatal period.

While clinical trials of DHA supplementation have yielded inconsistent results regarding suppression of ROP [16–22], DHA vs. ARA supplementation is protective in a mouse model of neovascular ROP (oxygen-induced retinopathy) [37,50–52] by promoting physiological vascularization in Phase I and inhibiting pathological neovascularization in Phase II. Currently, the focus has shifted towards synergistic/complementary effects of oral DHA+ARA. Two clinical trials (Mega Donna Mega and DIAMOND) demonstrated that oral supplementation (starting close to birth) in Phase I ROP with DHA+ARA (1:2 ratio) improved retinal health in very preterm infants [23,24], but the underlying mechanisms are unclear. Our current work shows that combined DHA+ARA vs. DHA oral supplementation better promotes physiological vascular development in a Phase I ROP model of hyperglycemia-associated vascular development suppression. The mouse study results align with clinical data, providing the foundation for further mechanistic investigations.

This study focused on the hyperglycemia-associated retinopathy mouse model of Phase I ROP based on findings from the Mega Donna Mega trial showing that premature infants at risk for ROP exhibit a mean blood glucose level of 7.9 mmol/L. In contrast, published studies report that postnatal mean blood glucose levels in full-term infants range from 2.3 to 4.6 mmol/L at various timepoints [53–55]. In our clinical data, infants who experienced hyperglycemia during the early postnatal period were more likely to be of lower gestational age and birth weight and to present with comorbidities such as bronchopulmonary dysplasia. Infants with hyperglycemia were also more likely to receive mechanical ventilation or continuous positive airway pressure and to present with elevated CRP and reduced hemoglobin and platelet count. As some of the mentioned observations are known risk factors for ROP [56,57], it is likely that hyperglycemia cooccurs with other risk factors and that distinct risk factors modulate one another. This highlights the multifactorial nature of ROP and underscores the importance of employing different models of ROP focusing on distinct underlying factors, such as oxygen exposure or hyperglycemia, to further enhance our understanding of the disease’s multifaceted pathogenesis. To verify our findings, we therefore also carried out maternal diet intervention with DHA+ARA vs. DHA in the OIR mouse model, confirming the protective effect of maternal DHA+ARA diet on physiological vascularization in systems modeling Phase I ROP.

Photoreceptor metabolism and metabolic needs dictate retinal vascular development [33,58]. Our findings suggest that a combination of DHA+ARA vs. DHA alone promotes physiological retinal vascularization by improving retinal metabolism as indicated by proteomics data and confirmed by loss of the protective effect after inhibition of mitochondrial energy production. Of note, maternal DHA+ARA vs. DHA diet did not affect pup blood glucose levels, suggesting that maternal DHA+ARA supplementation does not prevent hyperglycemia, a factor leading to retinopathy, but instead protects the retina from the harmful effects of hyperglycemia. This effect could be mediated through enhanced retinal metabolism and potentially improved local glucose utilization, as indicated by the enrichment of the Gene Ontology pathway, glucose import, in the proteome. Additionally, maternal DHA+ARA diet might alter cell membrane compositions and thereby improve membrane fluidity and mitochondrial function [59–61]. The beneficial effects of DHA+ARA are not likely to be mediated through mitochondrial β-oxidation of these lipids for energy production. DHA and ARA can be oxidized by some cells such as murine astrocytes but to a much lower extent and at slower rates than other lipids, e.g., palmitic or oleic acid [62]. However, DHA+ARA might influence mitochondrial β-oxidation by modulating the supply of other saturated and unsaturated fatty acids. For example, we found significantly altered circulating (C14:0, C18:0, C20:5n-3, and C22:4n-6) and retinal (C14:0, C18:2n-6) fatty acids in DHA+ARA vs. DHA diet groups. Further studies are required to identify preferred lipids for retinal energy production and to discover whether mitochondrial β-oxidation of other lipids mediates DHA+ARA protection against neonatal retinopathy.

Balancing LCPUFAs has gained attention in studies of ROP [16–22] and other complications of preterm birth, including bronchopulmonary dysplasia [63], as well as other retinal disorders, such as diabetic retinopathy, age-related macular degeneration [64], type 3 Stargardt dystrophy [65], and dry eye syndrome [66]. Interestingly, our data revealed distinct diet-dependent changes in fatty acid profiles across different tissues, suggesting that plasma DHA and ARA levels are reflected in the RPE complex, while being tightly regulated in the retina [49]. Higher dietary DHA (3 % vs. 1 %) did not further increase retinal DHA levels. Notably, supplementation of DHA without n-6 ARA supplementation reduces serum ARA levels in infants [19] and animals [67,68], and low postnatal serum ARA levels are associated with increased ROP incidence [14]. Supplementation of DHA+ARA at a 1:2 ratio leads to increased DHA and ARA serum levels in infants [23,24], offering a promising approach for ROP prevention. These findings suggest a complex interplay between dietary LCPUFAs and endogenous fatty acid metabolism.

Our study has limitations that could be addressed in future research. First, while our mouse model of hyperglycemia-associated retinopathy provides valuable insights into mechanisms of neonatal hyperglycemic retinal disease, it does not fully recapitulate the complexity of human ROP. Second, the long-term effects of combined DHA+ARA supplementation on retinal function and structure beyond the neonatal period remain to be elucidated. Third, the optimal ratio of DHA to ARA for retinal protection requires further investigation, as it may depend on the specific developmental stage and pathological condition. Finally, the fatty acid profiling targeted lipids of interest and did not capture potential changes in other fatty acid species, particularly very long-chain fatty acids. A limitation to the generalizability of the animal study is that it did not consider gender/sex.

In conclusion, our findings provide evidence for the protective effects of combined DHA+ARA supplementation against hyperglycemia-associated retinopathy in neonatal mice through increased mitochondrial metabolism. This study offers mechanistic insights into the clinical observations of reduced ROP severity with oral DHA+ARA supplementation and highlights the importance of balanced LCPUFA intake for optimal retinal development, which may also have a beneficial effect on other retinal disorders.

Supplementary Material

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.phrs.2025.107877.

Data availability

Data will be made available on request.