Abstract

   Bronchopulmonary dysplasia (BPD) is a prevalent chronic respiratory
   condition in preterm infants with an increasing incidence, severely
   affecting their survival rate and quality of life. Exploring the
   underlying mechanisms of BPD helps to develop novel effective
   therapeutic strategies. In this study, integrated metabolomic analyses
   of tracheal aspirates (TAs) from BPD infants and non-BPD infants, along
   with lung tissues from hyperoxia-induced experimental BPD neonatal rats
   and control rats, demonstrated that BPD was associated with a
   significant reduction in 3-hydroxyanthranilic acid (3-HAA), which was
   confirmed to be partly caused by tryptophan-metabolizing enzyme
   disorders. In vivo and in vitro models were subsequently established to
   assess the efficacy and underlying mechanisms of 3-HAA in relation to
   BPD. Compared with the BPD group, 3-HAA nebulization improved lung
   development and suppressed inflammation in rats. Limited
   proteolysis-small molecule mapping (LiP-SMap) proteomic analysis
   revealed the involvement of the ferroptosis pathway in the underlying
   mechanism by which 3-HAA alleviated hyperoxia-induced BPD injury.
   Ferroptosis was identified by detecting Fe^2+ levels, malondialdehyde
   (MDA), 4-HNE, total aldehydes, mitochondrial morphology,
   ferroptosis-associated protein and mRNA expression, and this
   dysregulation was indeed ameliorated by 3-HAA nebulization in vivo.
   Furthermore, a combination of LiP-SMap, molecular docking, SPR and
   Co-IP analyses confirmed that 3-HAA can bind directly to FTH1 and
   disrupt the nuclear receptor coactivator 4 (NCOA4)-FTH1 interaction. In
   conclusion, our study is the first to reveal that BPD is linked to the
   reduction of 3-HAA, and 3-HAA could inhibit the ferroptosis pathway by
   targeting FTH1, thereby alleviating hyperoxia-induced injury in rats
   and alveolar type II epithelial cells, highlighting the potential of
   targeting 3-HAA and ferroptosis for clinical applications in BPD.

   Keywords: Bronchopulmonary dysplasia, Neonatal rat,
   3-Hydroxyanthranilic acid, Ferroptosis, Ferritin heavy chain 1,
   Alveolar type II epithelial cells

Graphical abstract

   [37]Image 1
   [38]Open in a new tab

Highlights

     * •
       BPD is associated with a significant reduction in the tryptophan
       metabolite 3-hydroxyanthranilic acid (3-HAA).
     * •
       3-HAA protects against hyperoxia-induced injury by inhibiting
       ferroptosis in lung tissue and AECII.
     * •
       3-HAA inhibits the ferroptosis pathway by targeting FTH1 and
       disrupting the NCOA4-FTH1 interaction.
     * •
       This study first reveals the dysregulation of tryptophan metabolism
       in BPD and the mechanism of 3-HAA in alleviating BPD.

1. Introduction

   Bronchopulmonary dysplasia (BPD) is a prevalent chronic respiratory
   disease in preterm infants [[39]1], and its incidence has increased
   with the increasing survival rates of extremely premature newborns. BPD
   is characterized mainly by abnormal alveolar development, alveolar
   simplification, impaired pulmonary microvascular development, oxidative
   stress, and pulmonary inflammation [[40]2], which severely affect the
   survival rate and quality of life of preterm infants [[41]3,[42]4]. The
   pathogenesis of BPD remains incompletely understood, although it is
   thought to be the result of an imbalance between lung injury and repair
   processes in preterm infants with immature lung development [[43]5].

   Despite advancements in perinatal care and the implementation of
   various therapeutic interventions in neonatal intensive care units
   (NICUs) to address BPD, no individual therapy has been shown to
   significantly decrease the incidence or severity of the condition.
   Considering the widespread occurrence of BPD and the limitations of
   existing treatments [[44]6,[45]7], it is essential to explore the
   mechanisms of BPD and to create novel, effective therapies that can
   counteract airway inflammation and alveolar abnormalities in affected
   infants.

   Cumulative research has indicated that BPD is associated with complex
   metabolic dysregulation [[46][8], [47][9], [48][10], [49][11]],
   including disturbances in essential amino acid metabolism. Changes in
   metabolites are believed to accurately reflect molecular
   characteristics that are closely linked to BPD, integrating the
   internal biological state and its external environment [[50]12].
   Tryptophan, which has gained much attention in recent years, is
   metabolized through the 5-hydroxytryptamine (5-HT), kynurenine, and
   indole pathways, and its metabolites possess a variety of biological
   activities, such as anti-inflammatory, anti-tumor, and antioxidant
   effects [[51]13,[52]14]. Most papers address the contributions of
   tryptophan and its metabolites to pathologies of the central nervous
   system, the digestive system, and the immune system. However, their
   contribution to the respiratory system has almost been ignored and
   understudied [[53]15]. Some evidence indicates the potential role of
   indole-3-acetic acid, a tryptophan metabolite, in modulating
   inflammation, emphysema and decreased lung function in chronic
   obstructive pulmonary disease (COPD) [[54]16], which shares some
   pathological features with BPD. However, the complex functions and
   mechanisms of tryptophan metabolites in neonatal BPD remain to be
   explored.

   Given that tryptophan metabolism is affected under pathological
   conditions, tryptophan and its metabolites are very attractive
   biomarkers for diagnosis, prognosis, and treatment. In addition, the
   biological effects of tryptophan metabolites and their changes in
   disease suggest that they may serve as potential therapeutic targets.
   In this study, we identified a reduction in the tryptophan metabolite
   3-hydroxyanthranilic acid (3-HAA) in both the tracheal aspirates (TAs)
   of infants with BPD and the lung tissues of hyperoxia-induced
   experimental BPD rats. Mechanistically, our data revealed that 3-HAA
   inhibited the ferroptosis pathway by targeting ferritin heavy chain 1
   (FTH1) and disrupting the NCOA4-FTH1 interaction, thereby alleviating
   hyperoxia-induced damage in rats and alveolar type II epithelial cells
   (AECII). Collectively, our findings offer new insights into the
   management and treatment of BPD, potentially leading to targeted
   therapies that address the metabolic dysfunctions associated with BPD.

2. Materials and methods

2.1. TA samples and sample size

   Infants with a gestational age (GA) < 32 weeks who underwent invasive
   mechanical ventilation (IMV) on the first day of life at the NICU of
   the Children's Hospital of Chongqing Medical University (CHCMU) between
   June 2023 and October 2023 were included in this study, and parents
   provided informed consent. TA samples were collected during routine
   tracheal aspiration in infants receiving mechanical ventilation as
   previously described [[55]17] and frozen at −80 °C until tryptophan
   metabolites were analyzed. Sample size estimation was performed using
   G∗Power 3.1 software [[56]18] with the following parameters: two-tailed
   α = 0.05, Cohen's effect size d = 1.5 (based on previous research
   [[57]19,[58]20]), and power (1 - β) of 0.8 [[59]21]. The calculation
   indicated that a sample size of at least 9 infants per group was
   required to detect significant differences between the groups. Relevant
   clinical data were collected from electronic medical records. The
   infants were followed up until 36 weeks postmenstrual age and
   categorized into the BPD group and the non-BPD control group. BPD was
   diagnosed according to the National Institutes of Health workshop
   definition in 2018 [[60]22]. After confirming that there were no
   differences between the two groups in terms of postmenstrual age or
   weight at the time of TA sample collection (postmenstrual age:
   30.9 ± 1.2 vs 31.2 ± 0.8 weeks, p = 0.49; weight: 1404.0 ± 290.3 vs
   1473.0 ± 258.4 g, p = 0.58), 10 infants with BPD and 10 non-BPD
   controls were finally enrolled and TA samples were subsequently
   analyzed. The clinical characteristics of the infants are summarized in
   [61]Table S1. The Institutional Review Board of CHCMU approved this
   study (2023–258), which was registered on [62]chictr.org
   (ChiCTR2400087152).

2.2. Animals and interventions

   The Animal Ethics Committee of CHCMU approved the animal experiment
   protocol (No. CHCMU-IACUC20240412007), which complied with the
   [63]ARRIVE guidelines. The hyperoxia-induced experimental BPD neonatal
   rat model [[64]23] is recognized as an appropriate animal model for
   investigating the pathogenesis of BPD and exploring new therapeutic
   approaches. Neonatal rats are born during the saccular stage of lung
   development, which is closely similar to that of preterm infants, and
   exhibit pathological features of BPD under hyperoxic conditions, and
   have genetic heterogeneity observed in human populations. Pregnant
   Sprague‒Dawley (SD) rats were obtained from Chongqing Medical
   University (Chongqing, China). They were housed individually and
   provided unrestricted access to food and water under conditions of
   20–25 °C with a 12:12-h light‒dark cycle, awaiting the birth of the
   neonates.

   Newborn SD rats in the BPD group, together with their nursing dams,
   were placed in a hyperoxia chamber within 12 h of birth that maintained
   an oxygen concentration of 85 % for 14 days as previously described
   [[65]8]. The rats in the control group were exposed to air containing
   21 % oxygen. The rats in the BPD + 3-HAA group were nebulized every
   other day (P2, P4, P6, P8, P10, P12, and P14) for 30–50 min each time
   during the 14-day hyperoxia modeling period. During the nebulization
   process, newborn rats were placed in a specialized chamber connected to
   a nebulizer (403S, yuwell), which delivered an aerosolized form of
   3-HAA. To minimize maternal and nutritional variations affecting lung
   development, the number of newborn rats per chamber was equalized, and
   nursing dams were rotated across all experimental groups.

2.3. Preparation of 3-hydroxyanthranilic acid-nebulized solutions

   After identifying the differential tryptophan metabolite 3-HAA, we
   prepared a solution of 3-HAA for nebulized intervention in neonatal
   rats. 3-HAA (CAS No. 548-93-6, AbMole) was dissolved in dimethyl
   sulfoxide (DMSO) and brought to a final concentration of 0.5 mg/mL in
   phosphate-buffered saline (PBS) immediately before use, with a DMSO
   concentration of 1.7 %. To explore the appropriate dose and
   effectiveness of 3-HAA nebulization, newborn rats were randomly divided
   into eight groups: control, control + DMSO, BPD, BPD + DMSO, BPD+3-HAA
   (25 mg/kg), BPD+3-HAA (50 mg/kg), BPD+3-HAA (75 mg/kg), and BPD+3-HAA
   (100 mg/kg). Finally, 75 mg/kg was chosen for further experiments.

2.4. Lung sample collection

   The rats were anesthetized via inhalation of isoflurane on the 15th day
   after birth, after which trachea and lung tissues were isolated. The
   lung tissues were lavaged through the right ventricle with precooled
   PBS (4 °C), and then the right lung tissues were excised and stored at
   −80 °C for subsequent biochemical analyses. The left lobe was inflated
   and fixed in situ via the trachea with 4 % paraformaldehyde (PFA) at a
   pressure of 20 cmH[2]O [[66]24,[67]25] and then incubated in 4 % PFA at
   4 °C overnight, enabling subsequent hematoxylin‒eosin (HE) staining,
   immunohistochemical (IHC) staining and immunofluorescence (IF)
   staining.

2.5. Metabolomic analysis

   A targeted metabolomics approach was employed to analyze the levels of
   tryptophan metabolites in infant TA samples and rat lung tissues.
   Metabolites were extracted and analyzed via ultra-high-performance
   liquid chromatography coupled with tandem mass spectrometry
   (UHPLC-MS/MS) metabolomics as previously described [[68]26,[69]27].

2.6. Identification of proteins interacting with 3-HAA

   Identification of proteins interacting with 3-HAA was conducted using
   the limited proteolysis-small molecule mapping (LiP-SMap) chemical
   proteomics approach, following established protocols [[70][28],
   [71][29], [72][30]]. First, lung tissue lysates in the BPD group were
   aliquoted into 6 equivalent volumes, each containing 100 μg of protein.
   To determine the proteins that interact with 3-HAA, 0.33 nmol/(μg total
   protein) 3-HAA was introduced to 3 of the 6 independent replicates and
   DMSO was added to the other 3 replicates, which were then incubated at
   25 °C for 10 min. Proteinase K was added to all samples at a 1:100
   enzyme/substrate ratio and incubated at 25 °C for 5 min. The protein
   fragments generated from the proteinase K-limited proteolysis step were
   further digested with trypsin at a 1:50 trypsin/substrate ratio to
   produce peptides for mass spectrometry analysis using a nano-UPLC
   (nanoElute2, Bruker) coupled to a timsTOF Pro2 instrument (Bruker).

2.7. Cell models

   Primary human AECII are difficult to passage and expand in vitro, and
   we used immortalized human AECII (ZQY095, ZQXZ-bio, China) induced by
   the SV40 gene fragment to study human AECII biology [[73]31]. The cells
   were grown in H-DMEM (Gibco, USA) supplemented with 10 % FBS, 100 IU/mL
   penicillin and 100 μg/mL streptomycin (C0222, Beyotime) and maintained
   in an incubator containing 5 % CO[2] at a suitable temperature (37 °C).
   AECII were divided into the control group (21 % O2) and hyperoxia group
   (85 % O2) to confirm the level of 3-HAA. Then in vitro experiments were
   performed to further demonstrate the mechanisms underlying the
   modulation of ferroptosis in AECII responded by the 3-HAA. AECII were
   randomly divided into the control (C) group, hyperoxia (H) group,
   hyperoxia+3-HAA (HA) group, hyperoxia+3-HAA + erastin (ferroptosis
   agonist, MCE, HY-15763) (HAE) group, and hyperoxia+3-HAA + deferoxamine
   (DFO) (ferroptosis inhibitor, MCE, HY-B1625) (HAD) group. The
   concentrations of 3-HAA, erastin and DFO were determined on the basis
   of cell viability with a cell counting kit-8 (CCK-8) assay.

2.8. Cell viability assay

   To explore the appropriate concentrations of 3-HAA, erastin and DFO,
   they were dissolved in DMSO respectively and brought to different
   concentrations in serum-free medium immediately before use. AECII were
   plated in 96-well plates and divided into control and hyperoxia groups.
   Subsequently, 3-HAA (20, 40, 60, 80, 100, 120, 140, and 160 μM), DFO
   (10, 25, 50, 75, and 100 μM) and erastin (1, 2, 5, 10, and 20 μM) were
   administered to both groups, respectively, for 24 h. Cell viability was
   assessed using the CCK-8 (K1018, ApexBio) assay as described previously
   [[74]32]. The working concentration of the CCK-8 solution was prepared,
   and the solution was added to each well. A microplate reader (BioTek
   Synergy H1) was used to analyze the absorbance (450 nm) after culturing
   for 2 h at 37 °C.

2.9. Histological analysis, IHC staining, and IF staining of lung tissue

   The 4 % PFA-fixed lung tissues were dehydrated with gradient ethanol
   and xylene before being embedded in paraffin. Paraffin-embedded tissues
   were sliced into 4 μm sections for morphological analysis using HE
   staining. The radial alveolar count (RAC) was used to assess the
   development of alveoli [[75]2], and the mean linear intercept (MLI) was
   used to assess the volume‒to‒surface ratio of alveolar airspaces
   [[76]33]. Histological images were examined via an optical microscope
   (Nikon, Japan), with three random fields of view selected per sample.

   Paraffin-embedded lung tissue sections were deparaffinized, hydrated,
   microwaved in Tris-EDTA buffer for antigen retrieval, incubated with
   3 % hydrogen peroxide for 30 min, and then blocked with 3 % BSA for 1 h
   at room temperature. For IHC staining, the sections were incubated with
   primary antibodies of interest at 4 °C overnight. Then, goat
   anti-rabbit IgG or goat anti-mouse IgG was added, and the samples were
   incubated for 30 min at room temperature. Subsequently, the DAB
   reaction was performed. Finally, the sections were counterstained,
   mounted, and examined under an optical microscope (Nikon, Japan). For
   IF staining, the sections were incubated with specific primary
   antibodies at 4 °C overnight and then incubated with an HRP-conjugated
   secondary antibody for 30 min at room temperature and with tyramide
   signal amplification (TSA) fluorescent dyes (AFIHC024, Aifang) [[77]34]
   for 10 min, followed by DAPI staining for 10 min. Images were captured
   using a fluorescence scanner (KF-FL-020, KFBIO), with three random
   fields of view selected per sample. The antibodies used are presented
   in [78]Table S2.

2.10. IF staining and live-cell imaging of AECII

   AECII were fixed in 4 % PFA for 20 min and subsequently washed three
   times with PBS. The cell slides were blocked with 10 % goat serum for
   10 min following incubation with 0.1 % Triton X-100 for 20 min if
   necessary and then incubated overnight at 4 °C with the primary
   antibodies of interest. The appropriate HRP-conjugated secondary
   antibody was used for 50 min at room temperature, and TSA fluorescent
   dyes (AFIHC024, Aifang) were added for 5 min, followed by DAPI staining
   for 6 min. Images were captured using a fluorescence scanner
   (KF-FL-020, KFBIO), with three randomly selected fields of view per
   sample. The antibodies used are presented in [79]Table S2.

   The reactive oxygen species (ROS) level in AECII was measured by the
   fluorescent probe DCFH-DA (S0033S, Beyotime) [[80]35,[81]36], and lipid
   peroxidation was assayed by Liperfluo [[82]37,[83]38] (L248, Dojindo).
   Briefly, cells in different groups were stained with 10 μM DCFH-DA at
   37 °C for 20 min or 1 μM Liperfluo at 37 °C for 30 min. After being
   washed with serum-free medium three times, images were acquired and
   analyzed using NIS-Elements software at an excitation wavelength of
   488 nm (Nikon, Japan).

2.11. Amplex Red hydrogen peroxide (H[2]O[2]) assay

   The extracellular H[2]O[2] levels were measured using the Amplex Red
   Hydrogen Peroxide Assay Kit (A22188, Invitrogen), following the
   manufacturer's instructions [[84]39]. Cell culture supernatant of
   different experimental groups was collected and incubated with 100 μM
   Amplex Red reagent containing 0.2 U/mL horseradish peroxidase for
   30 min at room temperature. Fluorescence was detected using a
   microplate reader with excitation at 540 nm and emission at 590 nm, and
   concentrations were determined based on a standard curve.

2.12. Measurement of redox- and ferroptosis-associated markers

   The content of total aldehydes in the lung tissues and AECII cells was
   detected using an Amplite™ Fluorometric Aldehyde Quantitation Kit
   [[85]36]. The levels of malondialdehyde (MDA) and Fe^2+ in lung tissues
   and AECII samples were detected by using commercially available assay
   kits. Reduced glutathione (GSH) levels and superoxide dismutase (SOD)
   activity in AECII were detected by the corresponding commercial
   reagents. These markers were detected following the manufacturers’
   instructions, and the commercial kits used are presented in [86]Table
   S3.

2.13. Enzyme-linked immunosorbent assay (ELISA)

   Lung tissues were homogenized in PBS, and supernatants were obtained by
   centrifugation at 12,000 rpm and 4 °C for 10 min for analysis of
   transforming growth factor-beta 1 (TGF-β1), tumor necrosis factor-alpha
   (TNF-α), vascular endothelial growth factor (VEGF), and interleukin 6
   (IL-6) levels. The cell culture supernatant was centrifuged at 1000×g
   and 4 °C for 20 min for TNF-α and IL-6 analysis. These factors were
   detected by ELISA kits according to the manufacturers' instructions.
   The total protein levels in the lung tissues were quantified by a BCA
   assay (P0010, Beyotime). The absorbance at 562 nm was measured with a
   microplate reader (BioTek Synergy H1). The ELISA kits used are
   presented in Table S3.

2.14. Western blot (WB)

   The lung tissues and AECII samples were lysed with RIPA lysis buffer
   and then centrifuged at 4 °C and 12000 rpm for 10 min, after which the
   supernatant was collected. Protein was quantified using a BCA assay kit
   (P0010, Beyotime) and then diluted with loading buffer, and boiled for
   10 min. After electrophoresis on SDS-polyacrylamide gels at 80 V for
   30 min and 120 V for 60 min, the proteins were transferred to
   polyvinylidene difluoride (PVDF) membranes with a pore size of
   0.45 μm at 400 mA for 30 min by NcmBlot Rapid Transfer Buffer. The PVDF
   membrane was quickly blocked for 15 min at room temperature and then
   incubated with primary antibodies overnight at 4 °C and the appropriate
   secondary antibody goat anti-rabbit IgG (511203, Zenbio) at room
   temperature for 1 h. The images of protein bands were captured by
   ChemiDoc™ Touch Imaging System (BIO-RAD) using an enhanced
   chemiluminescence (ECL) solution kit (17047, Zenbio), and analyzed
   using Image lab and ImageJ software. The antibodies used for WB are
   presented in [87]Table S2, with β-actin serving as a reference.

2.15. Real-time quantitative PCR (RT‒qPCR)

   Total RNA was isolated and purified from rat lung tissue and AECII
   samples following the protocol provided by Accurate Biotechnology
   (AG21023). The RNA yield and quality were evaluated by a NanoDrop
   spectrophotometer (Thermo Fisher Scientific) with OD260/OD280 ratios
   between 1.8 and 2.2. Evo M-MLV RT Premix for qPCR (AG11734, Accurate
   Biotechnology) and the SYBR Green Premix Pro Taq HS qPCR Kit (AG11733,
   Accurate Biotechnology) were used for reverse transcription and
   amplification, respectively. Gene expression analysis was performed on
   a BioRAD Real-Time PCR System, with β-actin serving as a reference, and
   relative mRNA expression levels were calculated via the 2 ^(−ΔΔCt)
   method. The primers used for RT‒qPCR are shown in [88]Table S4.

2.16. Transmission electron microscopy (TEM)

   Fresh lung tissues and collected cells were fixed in 2.5 %
   glutaraldehyde, followed by 2 h of fixation with 1 % osmium tetroxide,
   and subsequently dehydrated using a graded ethanol series. The samples
   were embedded in acetone overnight and sectioned into ultrathin slices
   (60–80 nm). Finally, the slices were stained with lead citrate and 2 %
   uranyl acetate. The sections were observed using an HT-7800 TEM
   (HITACHI, Tokyo, Japan) or a Thermo Scientific™ Talos™ L120C.

2.17. Molecular docking

   The interaction between 3-HAA and FTH1 was calculated using a
   computational docking method [[89]40,[90]41] (Autodock vina 1.1.2). The
   molecular structure of the 3-HAA ligand was sourced from PubChem (CID:
   86). The FTH1 structure used for docking was obtained from a Protein
   Data Bank (PDB) (PDB ID: [91]4OYN). After subjecting the ligand 3-HAA
   and the receptor FTH1 to hydrogenation and structure optimization,
   3-HAA was docked into the potential binding pocket of FTH1.

2.18. Surface plasmon resonance (SPR)

   The SPR technique, which is a robust method for analyzing real-time
   biomolecular interactions without the need for labeling, was utilized
   to identify the interaction affinity between the recombinant human FTH1
   protein and compound 3-HAA. The FTH1 protein was directly coupled to
   the CM5 chip (BR-1005-30, China) for determination. Following
   incubation, a gradient of 3-HAA was introduced into the FTH1
   protein-CM5 system. Similarly, to explore whether human FTH1 interacts
   with human NCOA4, the recombinant human NCOA4 protein was directly
   coupled on the CM5 chip, and a gradient of FTH1 protein was introduced
   into the NCOA4 protein-CM5 system. To determine whether 3-HAA affects
   the binding of FTH1 to NCOA4, FTH1 protein with and without 3-HAA was
   introduced into the NCOA4 protein-CM5 system. Binding interactions were
   analyzed using BIAcore T200 evaluation software (v.2.0, GE Healthcare,
   USA). The procedure followed the methodology described in previous
   studies [[92]42,[93]43].

2.19. Co-immunoprecipitation (Co-IP) assay

   A Co-IP assay was used to investigate the interaction between NCOA4 and
   FTH1 in AECII following treatment with 3-HAA. The cells were lysed with
   IP buffer containing protease inhibitor cocktails, and the total
   protein concentration was assessed with a BCA assay kit. The lysate
   supernatant was incubated with 10 μL anti-NCOA4 (ab314553, Abcam)
   antibody or regular IgG overnight at 4 °C, and repeatedly washed
   protein A/G agarose beads were added to capture the immunocomplexes.
   Then, the immunocomplexes were washed 3 times with IP lysis buffer, and
   the samples were boiled in SDS‒PAGE loading buffer and subjected to WB
   analysis [[94]44].

2.20. Statistical analysis

   Statistical analyses were conducted with SPSS software 26.0 and
   GraphPad Prism 9.0. All experiments were repeated at least three times
   biologically, data were presented as mean ± standard deviation (SD).
   Statistical significance between two groups was determined by Student's
   t-test, whereas comparisons across three or more groups were performed
   by one-way analysis of variance (ANOVA) with Tukey's post hoc test for
   multiple comparisons. Cohen's d was employed as a measure of effect
   size to quantify the degree of intergroup differences in metabolites,
   calculated by dividing the mean difference between groups by the pooled
   standard deviation. Spearman's rank correlation coefficient was
   performed to quantify the associations between metabolites and clinical
   characteristics. The Kyoto Encyclopedia of Genes and Genomes (KEGG)
   database was utilized for pathway evaluation. A p value < 0.05 was
   considered statistically significant throughout the study (∗p < 0.05,
   ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 in the figures).

3. Results

3.1. BPD was associated with decreased 3-HAA

   In our study, the scheme for identifying differential tryptophan
   metabolites is shown in [95]Fig. 1A. UHPLC-MS/MS metabolomics analysis
   was performed to explore the tryptophan metabolites in the TAs of BPD
   infants and non-BPD infants. The orthogonal projections to latent
   structures discriminant analysis (OPLS-DA) model revealed the
   distinction between the two groups (n = 10) ([96]Fig. 1B). Twenty-one
   metabolites were identified in the TAs, among which 3-HAA (P = 0.033,
   log[2] fold change (FC) = -2.355, Cohen's d = 1.084) and ILA
   (P = 0.003, log[2] FC = −0.996, Cohen's d = 1.567) were decreased in
   the BPD group compared with the control group ([97]Fig. 1C).
   Correlation analysis among tryptophan metabolites and clinical
   characteristics revealed that there was no correlation between 3-HAA
   levels and the duration of antibiotic use ([98]Fig. S1).

Fig. 1.

   [99]Fig. 1
   [100]Open in a new tab

   BPD was associated with decreased 3-HAA. (A): Scheme for the
   identification of differential tryptophan metabolites in BPD. (B):
   OPLS-DA statistical model for tryptophan metabolites in TA samples.
   Each point represents a single sample, colored by group. (C): Volcano
   plot of detected metabolites in TAs. Red indicates an upregulation in
   BPD, blue shows a downregulation in BPD, and the circle sizes refer to
   variable importance in projection (VIP). (D): Representative HE-stained
   lung tissue sections from control and BPD group rats, scale
   bar = 25 μm. (E): OPLS-DA statistical model for tryptophan metabolites
   in lung tissues. Each point represents a single sample, colored by
   group. (F): Volcano plot of detected metabolites of rat lung tissues.
   Red indicates an upregulation in BPD, blue shows a downregulation in
   BPD, and the circle sizes refer to the VIP.

   Subsequently, to investigate whether tryptophan metabolic dysregulation
   also exists in an experimental setting, we utilized a well-established
   SD rat model of hyperoxia-induced BPD. Compared with the control group,
   the BPD group presented significant alveolar simplification, indicating
   that there was impaired lung development ([101]Fig. 1D). Therefore,
   this preclinical rat model is suitable for further exploration of
   tryptophan metabolites and their role in BPD. As expected, the OPLS-DA
   model revealed the distinction between the BPD group and the control
   group (n = 8) ([102]Fig. 1E). Twenty-seven metabolites were identified,
   and more metabolites were altered in lung tissue, among which 3-HAA
   (p = 0.0001, log[2] FC = −1.328, Cohen's d = 2.611), NAS (p = 0.015,
   log[2] FC = −3.851, Cohen's d = 1.603), and 5-HTOL (p = 0.017, log[2]
   FC = −1.302, Cohen's d = 1.440) decreased in the BPD group compared
   with the control group ([103]Fig. 1F). Taken together, only 3-HAA was
   reduced in both the TA samples and the lung tissues from the BPD group
   and was subsequently selected for further investigation.

3.2. Dysregulation of related metabolic enzyme led to decreased 3-HAA

   IHC staining for 3-HAA ([104]Fig. 2A) was then performed on rat lung
   tissues, while IF staining for 3-HAA ([105]Fig. 2G) was conducted on
   AECII, confirming the reduction in 3-HAA in both in vivo and in vitro
   hyperoxia models. 3-HAA is an important intermediate metabolite in the
   Kynurenine (KYN) metabolic pathway. KYN is converted into
   3-Hydroxykynurenine by kynurenine 3-monooxygenase (KMO), and then
   catalyzed by kynureninase (KYNU) to generate 3-HAA. In the subsequent
   metabolic process, 3-HAA is further catalyzed by downstream enzymes
   such as 3-hydroxyanthranilate 3,4-dioxygenase (HAAO) to generate
   Quinolinic acid, etc. ([106]Fig. S2). To investigate the underlying
   reasons for the differences in 3-HAA levels between groups, WB analysis
   ([107]Fig. 2B and C) and IHC staining ([108]Fig. 2D–F) of the 3-HAA
   metabolism-related enzymes KMO, KYNU, and HAAO were performed on rat
   lung tissues. The results demonstrated that the protein expressions of
   KMO and KYNU were decreased in the BPD group, which led to a decrease
   in the production of 3-HAA, whereas the elevated protein expression of
   HAAO led to an increase in the downstream metabolism of 3-HAA.
   Additionally, WB analysis ([109]Fig. 2H and I) and IF staining
   ([110]Fig. 2J-L) of AECII revealed a similar pattern: KYNU and KMO
   decreased and HAAO increased in the hyperoxia group. The RT‒qPCR
   results were also consistent in both in vivo and in vitro hyperoxia
   models ([111]Fig. S3A), providing a reasonable explanation for the
   observed differences in both the in vivo and in vitro models. Overall,
   BPD is associated with decreased 3-HAA, which was confirmed to be
   partly caused by dysregulation of related metabolic enzyme.

Fig. 2.

   [112]Fig. 2
   [113]Open in a new tab

   Dysregulation of related metabolic enzyme led to decreased 3-HAA. (A):
   IHC staining and the quantitative results in lung tissues of 3-HAA in
   the control and BPD groups, scale bar = 25 μm. (B): THE expression of
   KMO, KYNU, and HAAO validated by WB in lung tissues. (C): The
   quantitative results of KMO, KYNU and HAAO protein expression in lung
   tissues. The levels of β-actin were used as internal standard control.
   (D). IHC staining and the quantitative results in lung tissues of KMO
   in the control and BPD groups, scale bar = 25 μm. (E). IHC staining and
   the quantitative results in lung tissues of KYNU in the control and BPD
   groups, scale bar = 25 μm. (F). IIHC staining and the quantitative
   results in lung tissues of HAAO in the control and BPD groups, scale
   bar = 25 μm(G): IF staining and the quantitative results in AECII of
   3-HAA (red) and DAPI (blue) in the control and hyperoxia groups, scale
   bar = 60 μm. (H): THE expression of KMO, KYNU, and HAAO validated by WB
   in AECII lysate. (I): The quantitative results of KMO, KYNU and HAAO
   protein expression in AECII. The levels of β-actin were used as
   internal standard control. (J): IF staining and the quantitative
   results in AECII of KMO in the control and hyperoxia groups, scale
   bar = 60 μm. (K): IF staining and the quantitative results in AECII of
   KYNU in the control and hyperoxia groups, scale bar = 60 μm. (J): IF
   staining and the quantitative results in AECII of HAAO in the control
   and hyperoxia groups, scale bar = 60 μm. For animal-related
   experiments, n = 6 per group. n = 3–4 per group. The data are presented
   as the means ± SD (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001).

3.3. 3-HAA alleviated hyperoxia-induced lung injury in BPD rats

   We next investigated whether supplementation with 3-HAA could alleviate
   hyperoxia-induced BPD. Neonatal rats in the BPD + 3-HAA group were
   nebulized with 3-HAA every other day (P2, P4, P6, P8, P10, P12, and
   P14) for 30–50 min each time during the 14-day hyperoxia modeling
   period ([114]Fig. 3A). To explore the appropriate dose and
   effectiveness of 3-HAA nebulization, newborn rats were randomly divided
   into eight groups: control, control + DMSO, BPD, BPD + DMSO, BPD+3-HAA
   (25 mg/kg), BPD+3-HAA (50 mg/kg), BPD+3-HAA (75 mg/kg), and BPD+3-HAA
   (100 mg/kg). Our results demonstrated that 3-HAA nebulization improved
   histopathological alterations, as the RAC increased and MLI decreased
   compared with those of the BPD group ([115]Fig. S4A). In addition,
   compared with BPD, 3-HAA increased the expression of SPC ([116]Figs.
   S4B and S4E) and AQP5 ([117]Fig. S4C and S4E), which are specific
   markers of AECII and AECI, respectively. 3-HAA also decreased the
   expression of MPO ([118]Figs. S4D and S4E), which can increase
   oxidative stress and tissue injury. A dose of 75 mg/kg exhibited the
   best effect and was selected for further experiments.

Fig. 3.

   [119]Fig. 3
   [120]Open in a new tab

   3-HAA alleviated hyperoxia-induced lung injury in BPD. (A): Schematic
   representation and timeline of the intervention for the three groups of
   rats. (B): HE-stained lung tissues and the quantitative results of MLI,
   scale bar = 25 μm(C): IHC staining of 3-HAA performed on lung tissues
   and the quantitative results, scale bar = 25 μm. (D): IF staining
   showed the percentage of SPC + cells relative to all DAPI + cells in
   the control, BPD and BPD+3-HAA groups per high-power field (HPF), scale
   bar = 100 μm. (E): THE expression of SPC, MPO, and COX2 validated by WB
   in lung tissues and the quantitative results. The levels of β-actin
   were used as internal standard control. For animal-related experiments,
   n = 6 per group. The data are presented as the means ± SD (∗p < 0.05,
   ∗∗p < 0.01, ∗∗∗p < 0.001).

   In further experiments, we confirmed that 3-HAA nebulization decreased
   the MLI ([121]Fig. 3B) and effectively increased the 3-HAA level
   ([122]Fig. 3C). In the present study, 3-HAA partially protected against
   the reduction in SPC^+ cells after hyperoxia ([123]Fig. 3D). In
   addition, the percentages of CD68^+ inflammatory macrophages,
   CD11c^+ dendritic cells and MPO^+ neutrophils in BPD rat lungs were
   significantly increased [[124]45] when compared with the control group.
   However, 3-HAA protected myeloid cells from influx to the lung
   ([125]Figs. S5A–C). The ELISA results revealed that the levels of VEGF
   increased, whereas the levels of inflammatory cytokines (including
   TNF-α, IL-6, and TGF-β) in the lung tissues were significantly lower in
   the BPD + 3-HAA group compared to BPD group ([126]Fig. S3B), which was
   consistent with the observed mRNA expression patterns ([127]Fig. S3C).
   At the protein level ([128]Fig. 3E) and mRNA level ([129]Fig. S3D),
   3-HAA reduced the expression of MPO and COX2, suggesting an attenuated
   inflammatory response [[130]46], and increased the expression of SPC,
   implying an amelioration of AECII damage. Collectively, these findings
   confirmed that 3-HAA protected against lung damage in a
   hyperoxia-induced BPD rat model by improving alveolar complexity and
   suppressing inflammation.

3.4. 3-HAA alleviated ferroptosis in lung tissue caused by hyperoxia-induced
BPD

   To explore the underlying mechanism by which 3-HAA alleviates
   hyperoxia-induced BPD in neonatal rats, chemical proteomics by the
   LiP-SMap approach was performed ([131]Fig. 4A). After data
   preprocessing, 20,519 detected peptides were retained. Differentially
   expressed peptides were screened according to a p < 0.05 and FC ≤ 0.1
   or FC ≥ 10. Further analysis revealed enrichment of a specific form of
   cell death, known as ferroptosis, which is an iron dependent, lipid
   peroxidation-driven form of programmed cell death, as indicated by KEGG
   pathway analysis and Gene Ontology biological process (GO-BP) terms
   ([132]Fig. 4B and C). Fe^2+ participates in the Fenton reaction,
   exacerbating lipid peroxidation and cellular damage. Mitochondrial
   membranes, which are rich in phospholipids, are particularly
   susceptible to oxidative damage, leading to mitochondrial dysfunction
   and impaired energy metabolism. Aldehydes such as MDA and 4-HNE are
   byproducts of lipid peroxidation and are produced when free radicals
   cause oxidative damage to lipids in cell membranes. We next assessed
   ferroptosis by evaluating ferroptosis-associated protein and mRNA
   expression, aldehydes and Fe^2+ concentrations, and mitochondrial
   morphology. Our results revealed that the contents of total aldehydes,
   ferrous iron, MDA and 4-HNE were increased in lung tissue because of
   hyperoxia-induced BPD and were significantly reduced in the 3-HAA + BPD
   group ([133]Fig. 4D). Furthermore, after 3-HAA treatment, long-chain
   acyl-CoA synthetase-4 (ACSL4), transferrin (TF), FTH1, NCOA4, and
   LC3B-II were decreased, whereas GPX4 was increased compared with those
   in the BPD group ([134]Fig. 4E). The RT‒qPCR results were consistent
   ([135]Fig. S3E). Mitochondrial alterations in AECII from rat lung
   tissues were observed using TEM, revealing typical
   ferroptosis-associated changes, such as mitochondrial pyknosis and loss
   of cristae, and 3-HAA treatment attenuated these mitochondrial
   abnormalities ([136]Fig. 4F). Collectively, these findings suggested
   that 3-HAA inhibited ferroptosis in the hyperoxia-induced BPD rat
   model, thereby attenuating lung injury.

Fig. 4.

   [137]Fig. 4
   [138]Open in a new tab

   3-HAA alleviated BPD by inhibiting ferroptosis (A): Flow chart
   depicting the LiP-SMap assay. (B): Bubble plot shows the KEGG pathway
   enrichment analysis of the LiP-SMap assay. (C): Bubble plot shows Gene
   Ontology biological process (GO-BP) terms enriched in the LiP-SMap
   assay. (D): The content of total aldehydes, MDA, Fe^2+ and 4-HNE in
   lung tissues. (E): Western blotting of 4-HNE, GPX4, ACSL4, TF, FTH1,
   NCoA4, LC3B-II and their quantification in lung tissues. The levels of
   β-actin were used as internal standard control (n = 6). (F):
   Representative TEM images of mitochondria in AECII of rat lung tissue
   (x8000). The data are presented as the means ± SD (∗p < 0.05,
   ∗∗p < 0.01, ∗∗∗p < 0.001).

3.5. 3-HAA inhibited the hyperoxia-induced ferroptosis pathway in AECII

   A large amount of evidence suggests that persistent damage to AECII is
   closely related to the occurrence and development of BPD [[139]47]. The
   aforementioned in vivo experiments revealed mitochondrial
   ferroptosis-like changes in AECII from rat lung tissues, so the
   regulatory mechanism by which 3-HAA affects ferroptosis in AECII in an
   in vitro model was further confirmed by administering the ferroptosis
   agonist erastin and the corresponding inhibitor DFO. The optimal
   concentrations of 3-HAA (140 μmol/L), DFO (75 μmol/L), and erastin
   (2 μmol/L) were selected by CCK-8 for further experiments ([140]Figs.
   S6A–F). The results of IF staining of AEC-II ([141]Figs. S6G–H)
   indicated reduced SPC expression under hyperoxic conditions.
   Conversely, the increased expression of AQP5 indicates the gradual
   differentiation of AECII into AECI [[142]48]. While 3-HAA and DFO
   treatment attenuated the cell transformation induced by hyperoxia, the
   ferroptosis agonist erastin reversed the 3-HAA response.

   In the hyperoxia-induced cellular model of AECII, the ROS level
   measured by the fluorescent probe DCFH-DA ([143]Fig. 5A) and the extent
   of lipid peroxidation measured by the fluorescence emitted by Liperfluo
   ([144]Fig. 5B) were significantly elevated compared with those in the
   control group, and supplementation with 3-HAA and DFO reduced excessive
   ROS and lipid peroxidation generation, which could be reversed by
   erastin. A similar pattern was observed for the contents of redox and
   ferroptosis-associated indicators, such as Fe^2+, total aldehydes, MDA,
   and H[2]O[2], which increased in the H and HAE groups and decreased
   following 3-HAA and DFO treatment. In contrast, GSH and SOD levels were
   decreased in the H and HAE groups but increased in the HA and HAD
   groups ([145]Fig. 5C–E). The levels of TNF-α and IL-6 in the cell
   culture supernatant were increased in the hyperoxia group, whereas the
   levels of inflammatory markers were decreased after 3-HAA and DFO
   treatment, indicating that 3-HAA and DFO mitigated hyperoxia-induced
   inflammation. Conversely, erastin counteracted the anti-inflammatory
   effects of 3-HAA and increased TNF-α and IL-6 levels ([146]Fig. S3F).
   Compared with those in the H group, the protein levels of FTH1, NCOA4,
   COX2 and LC3B-II in AECII also decreased, whereas GPX4 expression
   increased after 3-HAA and DFO treatment ([147]Fig. 5F), which was
   consistent with the mRNA expression patterns ([148]Fig. S3G). However,
   the antiferroptotic effect of 3-HAA was reversed by erastin.
   Mitochondrial shrinkage and increased mitochondrial membrane density
   [[149]49] were observed in AECII cells due to hyperoxia exposure.
   Treatment with 3-HAA and DFO attenuated mitochondrial alterations given
   their crucial role in preventing ferroptosis. In contrast, erastin
   activated ferroptosis and blocked the protective effects of 3-HAA
   ([150]Fig. 5G). Collectively, these findings suggested that 3-HAA
   alleviated hyperoxia-induced ferroptosis in AECII.

Fig. 5.

   [151]Fig. 5
   [152]Open in a new tab

   3-HAA inhibited the hyperoxia-induced ferroptosis pathway in AECII.
   (A): Live cell fluorescence imaging of DCFH-DA (green) in AECII and the
   quantitative results, scale bar = 10 μm. (B): Live cell fluorescence
   imaging of LiperFluo (green) in AECII and the quantitative results,
   scale bar = 10 μm. (C): The content of total aldehydes, Fe^2+, MDA, GSH
   in AECII. (D): The content of SOD in AECII. (E): Extracellular H[2]O[2]
   levels in cell culture supernatant. (F): THE expression of GPX4, FTH1,
   LC3BII, NCOA4, and COX2 in AECII validated by WB and the quantitative
   results. The levels of β-actin were used as internal standard control.
   (G):TEM images of mitochondria in AECII, scale bar = 1 μm. For the
   cellular experiments, n = 3 per group. The data are presented as the
   means ± SD (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001).

3.6. 3-HAA inhibited the ferroptosis pathway by targeting FTH1

   Different from some other studies showing a decrease in FTH1 during
   ferroptosis [[153]50], our results indicated that upregulation of FTH1
   was observed during hyperoxia-induced ferroptosis in vivo and in vitro,
   which guided us to focus on this protein. We reasoned that elevated
   levels of FTH1 may serve as a protective mechanism against
   hyperoxia-induced oxidative stress by binding excess iron. However,
   simultaneously elevated NCOA4 acts as a cargo receptor for ferritin,
   specifically recognizing and binding to FTH1 and facilitating the
   autophagic degradation of ferritin [[154]51]. 3-HAA improved this
   phenomenon. IF staining of AECII revealed that 3-HAA and DFO inhibited
   ferritinophagy, as indicated by the reduced expression of LC3B, NCOA4,
   and FTH1, which indicates impaired autophagic degradation of ferritin
   ([155]Figs. S6I–J).

   Indeed, by using the LiP-SMap method mentioned above, we identified
   FTH1 as a potential binder of 3-HAA, as shown in [156]Fig. 6A (with
   p = 0.04 and log[2] FC = −5.6). SPR was subsequently performed to
   provide insights into the binding kinetics and molecular affinities
   between 3-HAA and FTH1. The data indicated that 3-HAA exhibited a
   strong, dose-dependent affinity for FTH1 (KD = 5.87E-06) ([157]Fig.
   6B). Subsequently, we conducted molecular docking to clarify the
   interaction mode between FTH1 and 3-HAA ([158]Fig. 6C), demonstrating
   that 3-HAA bonds to SER31 and SER59 through hydrogen bonding (2.0 Å and
   2.6 Å). The docking score was −5.1 kcal/mol. Given that the NCOA4-FTH1
   interaction is essential for ferritinophagy, we performed SPR to
   investigate whether 3-HAA alleviated ferritinophagy by disrupting the
   NCOA4-FTH1 interaction. As shown by SPR, recombinant FTH1 bound the
   recombinant NCoA4 protein with high affinity ([159]Fig. 6D). However,
   SPR revealed that 3-HAA was able to prevent FTH1-NCOA4 binding
   ([160]Fig. 6E). We validated this finding using a Co-IP assay
   [[161]52]. As shown in [162]Fig. 6F, 3-HAA treatment substantially
   weakened the interaction between NCOA4 and FTH1 compared to the AECII
   in the hyperoxia group, as evidenced by decreased FTH1 in the
   immunoprecipitates pulled down by anti-NCOA4. Collectively, these
   phenomena indicated that the exogenous administration of 3-HAA
   successfully alleviated ferroptosis by directly targeting FTH1 and
   disrupting the NCOA4-FTH1 interaction, thereby inhibiting
   hyperoxia-induced AECII injury in vitro and in vivo.

Fig. 6.

   [163]Fig. 6
   [164]Open in a new tab

   3-HAA inhibited the ferroptosis pathway by targeting FTH1 (A): Volcano
   plot of differentially expressed peptides by LiP-SMap approach. Red
   indicates an upregulation in the BPD+3-HAA group, and blue shows a
   downregulation in the BPD+3-HAA group. (B): The binding between human
   recombinant FTH1 and varying concentrations of 3-HAA was assessed by
   SPR. (C): The illustrations of molecular docking represent the binding
   modes of 3-HAA with the human FTH1, depicted as 3D binding and 2D
   binding views. (D): The binding between human recombinant NCoA4 and
   varying concentrations of FTH1 was assessed by SPR. (E): Recombinant
   human NCoA4 was coupled on the chips. The binding between FTH1 and
   NCoA4 in the presence or not of 3-HAA was assessed by SPR. (F): Co-IP
   assay of NCOA4 and FTH1 interaction in AECII and corresponding
   quantitative results.

4. Discussion

   BPD is the most common chronic lung disease in preterm infants who
   require respiratory support. The absence of effective prevention and
   treatment strategies results in higher mortality rates and poorer
   prognoses in BPD infants than in non-BPD premature infants, imposing
   heavy economic burden on families and society [[165]53,[166]54].
   Understanding the underlying pathogenesis of BPD is crucial for
   developing targeted interventions [[167]55]. Recent studies have
   highlighted the role of metabolic dysfunctions in contributing to lung
   injury caused by hyperoxia [[168]56]. In this study, we aimed to
   investigate the potential role of tryptophan metabolites in preventing
   hyperoxia-induced BPD and to explore the potential mechanism involved.
   Therefore, we analyzed tryptophan metabolic profiles and identified a
   reduction in 3-HAA in experimental and clinical BPD. Treatment with
   3-HAA suppressed inflammation and improved lung development in BPD
   rats. Our findings indicated that both in vivo and in vitro, 3-HAA
   modulated the dysregulated expression of ferroptosis-related molecules
   and mitigated the signs of mitochondrial ferroptosis in
   hyperoxia-induced models. In addition, molecular docking, SPR and Co-IP
   assays were performed to confirm that 3-HAA inhibited the ferroptosis
   pathway by targeting FTH1, thereby alleviating hyperoxia-induced damage
   in rats and AECII.

   From a developmental perspective, premature infants who are prone to
   BPD have distinct metabolic signatures compared with other patient
   groups. For example, Piersigili [[169]57] reported that the
   bronchoalveolar lavage fluid (BALF) of BPD infants had relatively high
   levels of serine, taurine, and citrulline on the first day after birth.
   Another study showed sphingolipid metabolites in TAs were significantly
   elevated in BPD infants during the first few days after birth
   [[170]58], which led to mitochondrial dysfunction [[171]59,[172]60] and
   abnormal ROS production. These biomarkers of BALF and TAs accurately
   reflect what is happening in the lungs. Owing to the complexity of BPD
   pathogenesis, it is necessary to develop sufficient biomarkers to
   identify premature infants who are susceptible to BPD. Emerging
   evidence [[173]61] has revealed that tryptophan metabolites are
   essential for modulating pathophysiological functions such as
   inflammation, metabolism, and oxidative stress. Interestingly, studies
   [[174]62,[175]63] suggest that metabolites from tryptophan may
   significantly influence lung health and modulate immune responses in
   the respiratory tract, which are critical in conditions like BPD. Our
   results showed that BPD was associated with dysregulation of tryptophan
   metabolism and a significant reduction in its metabolite 3-HAA. As
   discussed earlier, 3-HAA has demonstrated multiple biological
   activities in different experimental disease models, including
   anti-inflammatory effects via the inhibition of PI3K/Akt/mTOR
   activation induced by LPS [[176]64], the ability to modulate immune
   responses by increasing the binding of NCoA7 and aryl hydrocarbon
   receptors [[177]65], and the ability to regulate apoptosis in cancer
   cells by binding to YY1 [[178]66]. Furthermore, 3-HAA significantly
   facilitates tumor cell escape from ferroptosis [[179]67]. Our study
   revealed its ability to inhibit hyperoxia-induced ferroptosis by
   binding to FTH1 and disrupting the NCOA4-FTH1 interaction. In our
   opinion, 3-HAA can be used as a potential biomarker and potential
   therapeutic strategy for preventing BPD progression, providing new
   perspectives on BPD management and treatment.

   In recent years, ferroptosis, a form of regulated cell death associated
   with iron accumulation and lipid peroxidation, has gained increasing
   attention [[180]68,[181]69] because it is closely associated with
   several diseases. There is increasing evidence that ferroptosis plays
   an important role in many respiratory disorders [[182]70]. Iron
   metabolism is crucial for the regulation of ferroptosis. Ferritinophagy
   [[183]71,[184]72], a specialized autophagic process, breaks down
   ferritin to release free iron within the cell. This iron can engage in
   the Fenton reaction, exacerbating lipid peroxidation and cellular
   damage, key features of ferroptosis. For example, a recent report has
   shown that cigarette smoke induces labile iron accumulation through
   ferritinophagy, leading to ferroptosis in lung epithelial cells
   [[185]73]. The evidence indicates that gene dysregulation associated
   with ferroptosis and iron accumulation in lung tissues may contribute
   to BPD [[186][74], [187][75], [188][76]]. Disruption of iron
   homeostasis and ferritinophagy could lead to increased vulnerability of
   lung tissues to oxidative stress, exacerbating this condition. Not
   surprisingly, our study revealed that 3-HAA inhibited hyperoxia-induced
   ferroptosis by reducing the protein and mRNA expression of ACSL4, TF,
   FTH1, NCOA4, and LC3B-II while increasing GPX4 expression. The
   ferroptosis inhibitor DFO attenuated cell damage and ferroptosis, but
   the ferroptosis agonist erastin reversed the 3-HAA-induced anti-damage
   response. FTH1 is a key component of the ferritin complex and plays a
   crucial role in regulating iron storage and the process of ferroptosis,
   converting toxic ferrous iron (Fe^2+) into its less reactive ferric
   form (Fe^3+) [[189]77,[190]78]. Elevated levels of FTH1 may serve as a
   protective mechanism against iron-induced oxidative stress by binding
   excess iron [[191]79]. However, simultaneous increases in NCOA4 and
   LC3BII promote ferritinophagy in hyperoxia-induced models. NCOA4 acts
   as a cargo receptor for ferritin, specifically recognizing and binding
   to FTH1, facilitating the autophagic degradation of ferritin [[192]51].
   When ferritin is tagged for degradation, LC3BII helps encapsulate the
   autophagic cargo, promoting the fusion of autophagosomes with lysosomes
   for degradation and subsequent free iron release, leading to the Fenton
   reaction and cellular damage. Indeed, a combination of LiP-SMap,
   molecular docking, SPR and Co-IP analyses confirmed that 3-HAA can bind
   directly to FTH1 and disrupt the NCOA4-FTH1 interaction. This study
   discovered for the first time that 3-HAA effectively inhibited the
   ferroptosis pathway by targeting FTH1, thereby alleviating
   hyperoxia-induced injury in vitro and in vivo.

   AECII cells have the ability to self-renew and differentiate into AECI
   cells, which is crucial for alveolar structure recovery and maintenance
   following lung injury [[193]80]. In the context of BPD, the disruption
   of alveolar maturation may stem from impaired proliferation and
   transition of AECIIs. Reports have shown varying results regarding the
   changes in AECII in experimental BPD lungs. Our immunofluorescence
   analysis of SPC in lung tissue revealed that the proportion of
   SPC^+ cells was reduced in the BPD group. Hyperoxia-induced in vivo and
   in vitro models demonstrated that ferroptosis and a reduction in 3-HAA
   in AECII were closely related to the progression of BPD. These findings
   could be crucial for the development of targeted therapeutic strategies
   against BPD in which ferroptosis of AECII plays a critical role. Though
   3-HAA protected myeloid cells such as CD68^+ inflammatory macrophages,
   CD11c^+ dendritic cells and MPO^+ neutrophils from influx to the lung,
   whether 3-HAA can similarly inhibit ferroptosis in these cell types in
   the lung has not been explored, and further studies are needed to
   clarify these mechanisms.

   There are some limitations in our current study. Firstly, the clinical
   part is a single-center, small sample size exploratory study aimed at
   screening potential biomarkers. It has been challenging to find
   suitable infants for controls, and the collection of an effective
   amount of TA samples has been difficult, resulting in limited ability
   to balance the groups [[194]81]. There are differences in gestational
   age and duration of antibiotic use between the BPD and non-BPD infants.
   However, no significant differences were observed between the groups in
   terms of birth weight, postmenstrual age and weight at the time of TA
   sample collection, which helps control for some potential confounding
   factors. Moreover, we found no correlation between 3-HAA levels and the
   duration of antibiotic use in the correlation analysis among tryptophan
   metabolites and clinical characteristics. Then we have validated the
   differences and effects of 3-HAA using experimental models, partially
   explored the role of 3-HAA in the pathogenesis of BPD, reflecting true
   biological effects. These multidimensional strategies significantly
   compensated for the limitations in the sample size of the clinical
   preliminary exploratory study. Future studies will involve prospective
   cohorts to further investigate 3-HAA as a potential biomarker for the
   clinical diagnosis of BPD. Increasing the sample size and conducting
   subgroup analysis based on gestational age helps mitigate the impact of
   baseline differences. Secondly, we established an in vitro
   hyperoxia-induced model using immortalized human AECII. However, it is
   important to note that these cells may not fully replicate the
   characteristics of primary human AECII. Despite these limitations, this
   study provides new insights into the mechanisms of BPD development and
   interventions.

5. Conclusions

   Our study is the first to reveal that BPD is associated with
   significantly reduced 3-HAA levels, indicating dysregulation of the
   tryptophan metabolic pathway. Supplementation with 3-HAA inhibits the
   ferroptosis pathway by targeting FTH1, thereby alleviating
   hyperoxia-induced injury in rats and AECII. This work highlights the
   potential of targeting 3-HAA and ferroptosis for clinical applications
   and provides new insights into the management and treatment of BPD.

CRediT authorship contribution statement

   Qiqi Ruan: Writing – original draft, Visualization, Resources,
   Methodology, Funding acquisition, Formal analysis, Conceptualization.
   Yingqiu Peng: Writing – review & editing, Methodology, Formal analysis.
   Xuanyu Yi: Writing – review & editing, Methodology, Formal analysis.
   Jingli Yang: Writing – review & editing, Resources. Qing Ai: Resources,
   Data curation. Xiaochen Liu: Writing – review & editing, Methodology.
   Yu He: Writing – review & editing, Resources, Project administration,
   Funding acquisition, Conceptualization. Yuan Shi: Writing – review &
   editing, Resources, Project administration, Funding acquisition,
   Conceptualization.

Authors’ consent for publication

   All authors of the manuscript titled “The tryptophan metabolite
   3-hydroxyanthranilic acid alleviates hyperoxia-induced bronchopulmonary
   dysplasia via inhibiting ferroptosis” agree to submit it for
   publication in Redox Biology. We hereby grant editors the right to edit
   the manuscript and publish it. We confirm that this manuscript is an
   original work, has not been published before, and is not under
   consideration for publication elsewhere. We further confirm that all
   authors have contributed significantly to the manuscript. We have
   reviewed the final version of the manuscript and approved it for
   publication.

Ethics approval

   The Institutional Review Board of CHCMU approved the clinical research
   section (2023-258), which was registered on [195]chictr.org
   (ChiCTR2400087152). The Animal Ethics Committee of CHCMU approved the
   animal experiment protocol (No. CHCMU-IACUC20240412007) which complied
   with the [196]ARRIVE guidelines.

Funding

   This work was supported by the National Key Research and Development
   Program of China (No. 2022YFC2704803), the National Natural Science
   Foundation of China (No. 82001602), Clinical Research Project for the
   Summit Program of Children's Hospital of Chongqing Medical University
   (CHCMU-2024-XKDF-1002), Key Projects for Technological Innovation and
   Application Development of Chongqing (CSTC2021jscx-gksb-N0015), Special
   Funding for Postdoctoral Research Projects of Chongqing (No.
   2022CQBSHTB3085), China Postdoctoral Science Foundation (No.
   2023MD744152), Natural Science Foundation of Chongqing (No.
   CSTB2024NSCQ-MSX0491), and Chongqing Graduate Research Innovation
   Project (No. CYB23211).

Declaration of competing interest

   The authors declare that they have no known competing financial
   interests or personal relationships that could have appeared to
   influence the work reported in this paper.

Acknowledgments