Metabolic shifts in murine inner ears mimicking erythrocytes and plasma reveal diagnostics and early predictors for age-related hearing loss

Summary

Age-related hearing loss (ARHL) is a major public health concern, driven by the interplay of multiple factors. Here, we reveal spatiotemporal metabolic shifts in murine inner ears mirroring erythrocytes and plasma. Isotope-labeled glucose tracing demonstrates a metabolic rerouting favoring glycolysis over the pentose phosphate pathway, alongside downregulation of the tricarboxylic acid cycle, indicating impaired energy production and redox homeostasis. Accumulation of medium- and long-chain acylcarnitines further exacerbates lipotoxicity. Notably, age-dependent depletion of arginine, lysine, proline, and glycine disrupts the arginine-polyamine-urea cycle. Translationally, UK Biobank plasma metabolomics links omega-6 fatty acids, linoleic acid, glycine, and albumin to ARHL resilience, while branched-chain amino acids, tyrosine, creatinine, glycoprotein acetyls and urea confer risk. Sex differences in ARHL were linked to fatty acid metabolism divergence. These bioenergetic disruptions in the inner ear are mirrored in erythrocytes and plasma, highlighting potential biomarkers for early ARHL diagnosis and treatment.

Graphical abstract

Highlights

• •Metabolic shifts in murine inner ears mimic erythrocytes and plasma during aging
• •Omega-6 fatty acids, linoleic acid, glycine, and albumin protect against ARHL
• •BCAAs, tyrosine, creatinine, GlycA, and urea are risk factors for ARHL
• •Sex differences in plasma fatty acids and inner ear β-oxidation during aging

Diagnostics; Human metabolism

Introduction

Aging population is rapidly increasing worldwide. For this reason, the frequency of age-related hearing loss (ARHL) is increasing and is a leading public health issue. Epidemiological studies indicate that the risk of ARHL appears as early as 45 years, with one-third of individuals over 65 experiencing some degree of hearing loss, and more than half of adults over 75 exhibiting severe hearing impairment that affects daily communication. Moreover, the incidence of ARHL is higher in males than in females.^1,2,3 Thus, ARHL is considered as the earliest onset aging-related disease and a leading public health concern affecting approximately 25% of American adults over 60 and millions worldwide. Untreated ARHL can progress to moderate-to-severe levels with serious psychosocial, neurological, and economic ramifications. Although the pathogenesis of ARHL is highly complex, emerging evidence indicates that ARHL is closely linked to increased oxidative stress,^4,5,6,7 chronic inflammation,^8,9,10 mitochondrial dysfunction,¹¹ and imbalanced energy regulation.¹² However, the metabolic changes that accompany aging and how these changes affect ARHL are poorly understood. Moreover, a sensitive, convenient and robust means to accurately define early age-related hearing decline prior to hearing loss is missing.

Auditory transduction initiates in the inner ear, a complex structure filled with endolymph and perilymph, and relies on the coordinated activity of sodium, potassium, and calcium ion channels, which mediate mechano-electrical transduction, signal modulation, and neurotransmitter release. The proper function of these ion channels is essential for auditory signal transduction and relies heavily on sustained ATP availability. Therefore, the inner ear has high metabolic demands since accurate auditory perception relies on continuous energy supply to maintain ion gradients, signal transduction and auditory function. Due to the limited end-arterial supply and glucose serving as the primary energy source, inner ears are extremely sensitive to hypoxia. Under oxygen-rich conditions, glucose is fully oxidized via oxidative phosphorylation to generate sufficient ATP to support the complex processes of hearing and balance perception. Additionally, glucose metabolism via the pentose phosphate pathway (PPP) protects the inner ear from oxidative stress and provides nucleotide precursors for cell growth and repair. However, any conditions or diseases affecting oxygen availability or glucose metabolism can impact inner ear function. Previous studies showed that reprogramming of glucose metabolism in the inner ear is closely related to morphogen signaling in developing hair cells.¹³ Moreover, early studies revealed that diabetes-related glucose metabolism impairments and inner ear insulin resistance can lead to inner ear dysfunction.^14,15,16 However, the metabolic mechanisms of the inner ear, the most unique and crucial auditory component related to early age-relatedhearing decline, are least explored.

Erythrocytes are the most abundant and structurally simplest cells in the human body. Mature erythrocytes lack a nucleus and are incapable of HIF-mediated transcriptional regulation or de novo protein synthesis, relying entirely on metabolic regulation for their function.¹⁷ The inner ear, similar to the peripheral circulation system, is filled with fluidic lymphatic compartments analogous to the liquid-flowing erythrocytes and plasma. Moreover, glucose serves as the primary energy source for erythrocytes. Therefore, employing advanced mass spectrometry and in vivo isotope-labeled ¹³C₆-glucose flux analysis, combined with precise amino acid (AA) quantification, here we sought to initially conduct murine studies across the lifespan to explore whether the metabolic shifts in erythrocytes as well as plasma in the peripheral circulation reflect the metabolic reprogramming driven by ARHL in inner ear; Subsequently, we extended to large human retrospective and prospective studies using NMR-based UK Biobank metabolomics data to validate our murine studies with a goal to identify potential aging-sensitive circulating metabolites for diagnosing and predicting ARHL.

Results

Characterization of a mouse model of ARHL

To elucidate the functional consequences of metabolic alterations in the inner ear from adolescence to aging, we set out to measure auditory function of C57BL/6 mice at four distinct age stages: 1-month-old (1M), 2-month-old (2M), 10-month-old (10M), and 16-month-old (16M). This murine model has been employed in prior studies to investigate ARHL.¹⁸ To establish the timeline of auditory decline in mice, auditory brainstem response (ABR) testing was conducted to measure hearing thresholds, amplitudes, and latency periods (Figures 1A–1D). ABR thresholds were similar in 1M and 2M mice, whereas a significant threshold decrease was observed in 10M mice, particularly at high frequencies (24 kHz, 32 kHz). At 16M, mice exhibited near-complete deafness with thresholds reaching approximately 90 dB (Figure 1B). As mice are most sensitive to frequencies of 8 kHz and 16 kHz, further analysis of other auditory parameters at these frequencies revealed similar amplitude and latency of peaks I and II in 1M and 2M mice, whereas significant decreases were observed in 10M mice (Figures 1C and 1D), which is consistent with the human report that AR-hearing decline occurs as early as 45 years in humans. Thus, we demonstrated progressive age-related functional decline in cochlear hair cell mechanoelectrical transduction, spiral ganglion neuron (SGN) discharge frequency, and SGN to cochlear nucleus (CN) auditory transmission. Using this mouse model of ARHL we investigated inner ear metabolism of C57BL/6 mice at four distinct age stages: 1-month-old (1M), 2-month-old (2M), 10-month-old (10M), and 16-month-old (16M), aiming to identify metabolic changes associated with ARHL.

Figure 1.

Progressive hearing loss in aging mice and age-dependent changes in glucose metabolism in erythrocytes and the inner ear determined by [U-¹³C₆]-glucose tracer analysis

(A) Schematic of experimental setup involving ABR Hearing Test for mice of diverse age groups, [U-¹³C₆]-glucose infusion into the circulatory system followed by tracer analysis in euthanized mice at three different time points.

(B) ABR thresholds of 1-, 2-, 10-, and 16-month-old mice in response to different frequencies. The ABR thresholds of 10-month-old mice were compared with those of 2-month-old mice, and similarly, the ABR thresholds of 16-month-old mice were compared with those of 10-month-old mice. Data are expressed as mean ± SEM. p values were assessed by repeated measures two-way ANOVA, followed by a multiple-comparison test. N = 8 mice per group. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

(C) Peak I and II amplitudes were measured in 1-, 2-, 10-, and 16-month-old mice in response to different sound intensities at frequencies of 8 and 16 kHz. Peak I and II amplitudes of 10-month-old mice were compared with those of 2-month-old mice. Data are expressed as mean ± SEM. p values were assessed by repeated measures two-way ANOVA, followed by a multiple-comparison test. N = 8 mice in each group. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

(D) Peak I-II latency were measured in 1-, 2-, 10-, and 16-month-old mice in response to different sound intensities at 8 and 16 kHz. Peak I-II latency of 10-month-old mice were compared with those of 2-month-old mice. Data are expressed as mean ± SEM. p values were assessed by repeated measures two-way ANOVA, followed by a multiple-comparison test. N = 8 mice per group. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

(E–G) Increased incorporation of glucose-derived ¹³C into glycolytic metabolites; glucose (E), pyruvate (F), and lactate (G), in erythrocyte and the inner ear with aging. Data are expressed as mean ± SEM. N = 5–8 mice in each group.

(H) Increased incorporation of glucose-derived ¹³C into alanine in erythrocyte and the inner ear with aging. Data are expressed as mean ± SEM. N = 5–8 mice in each group.

(I) Decreased incorporation of glucose-derived ¹³C into PPP metabolite (Ru5P) in erythrocytes and the inner ear with aging. Data are expressed as mean ± SEM. N = 5–8 mice in each group. Ru5P, D-Ribulose 5-phosphate.

(J) The ratio of lactate (M+3) to lactate (M+2) increases in erythrocyte and the inner ear with aging. Data are expressed as mean ± SEM. p values were assessed using the ordinary one-way ANOVA, followed by a multiple-comparison test. N = 5–8 mice in each group. ∗p < 0.05.

(K) Incorporation of glucose-derived ¹³C into TCA cycle and amino acid metabolites, citrate, succinate, malate, aspartate, glutamate, glutamine, in erythrocyte and the inner ear from adolescence to aging. Data are expressed as mean ± SEM. N = 5–8 mice in each group.

(L) Schematic of age-dependent changes in glucose metabolism in erythrocyte and the inner ear. Erythrocytes and the inner ear exhibit similar alterations in glucose dynamics, with glucose metabolism shifting from the pentose phosphate pathway (PPP) toward glycolysis, leading to a reduction in antioxidant capacity. Additionally, the downregulation of the tricarboxylic acid (TCA) cycle in the inner ear results in decreased energy supply, which impacts auditory signal transmission. Red font indicates increased metabolic pathways and metabolites, while blue font depicts decreased metabolic pathways and metabolites.

Rapid glucose metabolism in erythrocytes and the inner ear

Because glucose is the main energy source for both inner ears and erythrocytes, we hypothesized that changes in glucose metabolism in erythrocytes mirrors glucose metabolism in inner ears across lifespan and is correlated with auditory functional decline from adolescence to aging. To test this hypothesis, we comprehensively traced glucose metabolic fate and reprogramming in vivo across lifetimes by administering ¹³C₆-glucose via intravenous injection, followed by isolation of erythrocytes, plasma and inner ears at three different time points (0.5, 1.5, 2.5 h) after injection. We tracked ¹³C₆-glucose at two levels: (1) at cellular and tissue level, to define glucose metabolic fate in erythrocytes and the inner ear from adolescence to aging; and (2) at the systemic level in plasma, to elucidate overall glucose dynamics and reprogramming from adolescence to aging.

Specifically, we monitored ¹³C₆-glucose metabolism through glycolysis, the pentose phosphate pathway (PPP), the tricarboxylic acid (TCA) cycle, and amino acid synthesis from 0.5 to 2.5 h post-injection (Figure S1A). Labeled lactate (M+3) rapidly appeared in erythrocytes and the inner ear as the end product of glycolysis at 0.5 h post-injection, indicating a high rate of glucose metabolism in erythrocytes and the inner ear (Figures S1C and S1E). Additionally, at 0.5 h in the inner ear pyruvate generated through glycolysis was further converted to acetyl-CoA, channeling into the TCA cycle and rapidly producing labeled intermediates including citrate (M+2) and amino acids (glutamate (M+2) and glutamine (M+2)), which are closely associated with the TCA cycle (Figure S1F). Similarly, we detected a series of labeled TCA cycle intermediates such as citrate (M+2), malate (M+2), succinate (M+2) in both erythrocytes and the inner ears (Figure S1D). Furthermore, ¹³C₆-glucose was rapidly metabolized into glycolysis and TCA cycle intermediates in erythrocytes, plasma, and the inner ear from 0.5 to 1.5 h, with these intermediates reaching undetectable levels by 2.5 h post-injection (Figures S1B–S1H). Overall, our findings revealed that ¹³C₆-glucose is rapidly metabolized within 0.5 h in both inner ears and erythrocytes and is nearly undetectable by 2.5 h.

Erythrocytes largely mirror the changes in glucose metabolism within the inner ear during ARHL

Glucose metabolism exhibited the most significant age-related changes by 0.5 h post-injection (Figures S1B–S1H). Therefore, we selected this time point to precisely analyze and compare the reprogramming of glucose metabolism in erythrocytes and the inner ear across the lifespan. Intriguingly, the reprogramming of glucose metabolism in the inner ear mirrored that of erythrocytes during aging. We observed an increasing trend in labeled glucose and glycolytic intermediates including pyruvate (M+3) and lactate (M+3) in both erythrocytes and the inner ear from 1 month-old up to 16-month-old mice (Figures 1E–1G). Additionally, labeled alanine (M+3), derived from pyruvate, also showed an upward trend in both erythrocytes and the inner ear (Figure 1H), and glycolytic intermediates labeled G3P (M+3) and 3PG (M+3) were elevated in erythrocytes (Figure S1I). In contrast, the labeled intermediate Ru5P (M+5) of PPP from glucose was reduced in both erythrocytes and the inner ears during the progression of ARHL (Figure 1I). Moreover, the increased ratio of lactate (M+3) from glycolysis to lactate (M+2) from the PPP further confirmed that glucose is preferentially metabolized toward glycolysis over PPP through aging (Figure 1J). Thus, our in vivo glucose flux tracing analyses demonstrated that erythrocytes largely mirror the glucose metabolic reprogramming within the inner ear during aging, with glucose flux increasingly shifting from the PPP to glycolysis.

Moreover, we observed labeled glucose-derived ¹³C-carbon labeled TCA cycle intermediates, such as citrate (M+2), succinate (M+2), and malate (M+2), along with a rise in associated amino acids like glutamate (M+2), glutamine (M+2), and aspartate (M+2) in inner ears during adolescence (Figure 1K). Conversely, the levels of these TCA cycle intermediates were reduced in 10-month-old and further reduced in the inner ears of 16-month-old mice (Figure 1K). Moreover, TCA cycle-derived AAs including glutamate (M+2), glutamine (M+2), and aspartate (M+2) were maintained at high levels in 10-month-old as in 2-month-old mice, while their levels were substantially lower in 16-month-old mice (Figure 1K). This pattern indicates an enhanced flux of glucose into the TCA cycle for fully oxidative phosphorylation to promote the production of ATP and amino acids to meet the increased energetic and nutritional demands required for inner ear maturation during adolescence, but decompensated and dropped in inner ears during aging and thus ARHL.

Notably, we observed a consistent upward trend in labeled glycolytic metabolites including glucose (M+6), pyruvate (M+3) and lactate (M+3) in plasma like erythrocytes and inner ears (Figures S1J–S1L), implicating an intimate crosstalk among erythrocytes, plasma and inner ears during ARHL development.

Different from inner ears, we observed that labeled TCA cycle intermediates, such as citrate (M+2), malate (M+2), and 2-oxoglutarate (M+2), along with their resultant amino acids glutamate (M+2) and glutamine (M+2), were elevated, while aspartate (M+2) levels were downregulated in both erythrocytes and plasma during aging (Figure S1K), implicating the alterations in TCA cycle intermediates and related amino acids in erythrocytes likely originated from metabolic changes in plasma. Taken together, we elucidated how erythrocytes and plasma largely mirror changes in glucose metabolism within inner ear during aging and thus leading to ARHL in multiple ways including: (1) glucose rapidly channels into glycolysis, the PPP, and the TCA cycle; (2) glucose metabolism in the inner ear and erythrocytes shifts more toward glycolysis over the PPP during aging, leading to a lower capacity to combat oxidative stress; and (3) glucose channeled toward TCA cycle is enhanced in the inner ear during adolescence, whereas its TCA is downregulated during aging, resulting in insufficient energy supply (Figure 1L).

The comprehensive metabolic landscape of murine erythrocytes, plasma, and inner ears across lifespans

Next, to comprehensively map the metabolomic profiles of the inner ear and erythrocytes and further analyze the interconnections between their metabolic characteristics during adolescence to aging, we conducted an in-depth analysis by untargeted metabolomics. Ultimately, we identified 351, 335, and 311 metabolites in erythrocytes, plasma, and the inner ear, respectively (Figures S2A and S2B). Following the Human Metabolome Database (HMDB) categorization scheme, these metabolites were classified into eight chemical categories. The distribution of metabolites across these categories, with lipids and amino acids being the most predominant in erythrocytes, plasma, and the inner ears (Figure S2C). The quantity and percentage of metabolites within each chemical category are illustrated accordingly, highlighting the prominence of lipids and amino acids in these biological matrices.

Partial Least Squares Discrimination Analysis (PLS-DA) of the inner ear revealed metabolomic profiles progressively diverging with age, exhibiting distinct clustering between 1-month-old (1M) and 2-month-old (2M) mice, and between 10-month-old (10M) and 16-month-old (16M) mice. This pattern mirrors our auditory phenotype observations, where 1M and 2M mice demonstrated superior hearing compared to the 10M and 16M mice exhibiting hearing decline (Figure S2A). Furthermore, metabolomic analyses of erythrocytes and plasma displayed similar clustering characteristics, with more pronounced age-related differences than those observed in the inner ear (Figure S2A).

The transformation in metabolic profiles is also reflected by the variance in specific metabolites, as illustrated. We observed substantial alterations in metabolites within erythrocytes, plasma, and the inner ear throughout the adolescent and aging processes (Figure S2B). The influence on metabolism was markedly more profound during adolescence than in aging, with an escalation in the number of differential metabolites observed at 16 months old comparing to 10 months old (Figure S2B). To ascertain the specific pathways targeted by these differential metabolites, enrichment analysis revealed that the alterations in metabolic pathways within erythrocytes, plasma, and the inner ear predominantly converged on glycolysis, fatty acid metabolism, the urea cycle, polyamines, and amino acid metabolism (Figures S2C and S2D). Altogether, the murine erythrocytes, plasma, and the inner ear demonstrate similar age-dependent alterations of metabolic pathways.

Accumulation of acylcarnitines is common metabolic feature in inner ears, erythrocytes and plasma during aging

Besides glucose metabolic impairment during aging, our comprehensive metabolomic profiling revealed a significant downregulation in the levels of medium and long-chain acylcarnitines during adolescence, but a progressive increase in erythrocytes, plasma, and the inner ear during aging (Figures 2D–2F). Interestingly, carnitine, the transporter for medium and long-chain fatty acids and substrates for acylcarnitine synthesis, exhibited heterogeneous metabolic changes in erythrocytes, plasma, and the inner ear. During adolescence, plasma levels of carnitine remained unchanged, erythrocyte levels increased, and inner ear levels decreased. In contrast, during aging, plasma and erythrocyte carnitine levels gradually decreased, while inner ear levels increased (Figures 2A–2C). In cells containing mitochondria, the build-up of long-chain acylcarnitines is viewed as harmful, as these intermediates can harm oxidative phosphorylation, thereby causing decreased ATP production and increased oxidative stress.¹⁹ However, the erythrocyte contains no mitochondria and endoplasmic reticulum. Thus, the erythrocyte lacks de novo synthesis of phospholipids and do not perform β-oxidation. Notably, the Lands cycle is a critical defensive system for erythrocytes to counteract oxidative damage membrane phospholipids by transferring the long-chain acylcarnitine to lysophospholipids to maintain the membrane integrity. Thus, our findings revealed that during adolescence, erythrocytes exhibit a low demand for acylcarnitine due to low oxidative stress at young age. Thus, erythrocytes have a sufficient higher level of carnitine but lower level of acylcarnitine due to minimal demand and less oxidative stress during adolescence. Conversely, during aging, the need for long- and medium-chain acylcarnitines rises to facilitate the Lands cycle for repairing oxidized membrane phospholipids, resulting in the significant carnitine consumption. The inner ear effectively oxidizes medium and long-chain acylcarnitines for energy production and growth demand during adolescence, but β-oxidation of medium and long-chain acylcarnitines is impaired with aging, causing lipotoxic damage. Finally, a marked accumulation of medium and long-chain acylcarnitines was observed in the plasma, possibly originating from both erythrocytes and the inner ear in aging (Figure 2G).

Figure 2.

Carnitine and long and medium chain acylcarnitines: Age-dependent changes in erythrocytes, plasma and the inner ear

(A–C) Age-dependent changes in carnitine in erythrocytes, plasma, and the inner ear. Data are expressed as mean ± SEM. p values were assessed using the ordinary one-way ANOVA, followed by a multiple-comparison test. N = 14–20 mice per group. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

(D–F) Heatmaps representing age-dependent changes in long and medium chain acylcarnitines in erythrocytes, plasma, and the inner ear. N = 14–20 mice per group.

(G) Schematic graph showing cross-tissue metabolite communication involving carnitine and long and medium chain acylcarnitines within erythrocytes, plasma and the inner ear in adolescence and aging. Red font indicates increased metabolites, and blue font depicts decreased metabolites.

Erythrocytes reflect the age-dependent amino acid changes in inner ear

Our glucose tracing and untargeted metabolomics revealed significant age-related changes in amino acids within erythrocytes and the inner ear, prompting us to precisely quantify 15 amino acids (Figure 3A). We found that most amino acids in the inner ear exhibit an age-dependent decline (Figure 3B). Glutamate and glycine, crucial neurotransmitters, notably decrease during inner ear aging, potentially impairing neural transmission and auditory function. Interestingly, four amino acids (arginine, lysine, proline and glycine) showed an age-dependent decline across erythrocytes, plasma, and the inner ear (Figures S3A and S3B). Amino acids are well-known for their crucial roles in protein synthesis, energy supply, neurotransmitter synthesis, metabolic and immune regulation, and various pathophysiological processes. Specific amino acids and their metabolites can serve as biomarkers for disease diagnosis and treatment. The decline in amino acid levels within the inner ear is likely to contribute to ARHL by impacting energy metabolism, neurotransmitter synthesis and antioxidant defense.

Figure 3.

Urea cycle and polyamine metabolism: Age-dependent changes in erythrocytes, plasma and the inner ear

(A) Experimental design for mouse sample collection and amino acids targeted metabolomics study.

(B) Heatmap representing changes in amino acid levels in erythrocyte, plasma and the inner ear determined by targeted metabolomics. Data are represented relative to the average of 1-month-old mice samples. N = 14–20 mice per group.

(C) Schematic diagram of the urea cycle and polyamine metabolism.

(D) Bar graphs for of urea cycle age-dependent changed metabolites (arginine, citrulline, ornithine, urea) in erythrocytes, plasma and the inner ear determined by untargeted metabolomics. Data are expressed as mean ± SEM. p values were assessed using the ordinary one-way ANOVA, followed by a multiple-comparison test. N = 14–20 mice per group. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

(E) Bar graphs depicting age-dependent changes in polyamine metabolites (putrescine, spermidine, spermine) in erythrocytes, plasma and the inner ear determined by untargeted metabolomics. Data are expressed as mean ± SEM. p values were assessed using the ordinary one-way ANOVA, followed by a multiple-comparison test. N = 14–20 mice per group. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001.

(F) Schematic graph showing cross-tissue metabolite communication in urea cycle metabolites and polyamines within erythrocytes, plasma and the inner ear during adolescence and aging. Red font indicates increased metabolites, and blue font depicts decreased metabolites.

PSR, polyamine stress response.

Erythrocytes mark the changes in urea cycle and polyamine metabolism in the inner ear across lifespan

The ultimate metabolism of amino acids leads to the production of ammonia, which is then processed through the urea cycle to convert toxic ammonia into non-toxic urea for excretion. Ornithine, an important intermediate in the urea cycle, also serves as a precursor for polyamine synthesis. Ornithine can be converted to putrescine by ornithine decarboxylase, which is further metabolized to produce spermidine and spermine (Figure 3C). Intriguingly, our untargeted metabolic pathway analysis revealed similar changes in the urea cycle and polyamine metabolism in erythrocytes and the inner ear across lifetimes. Through the integration of untargeted metabolomics and precise amino acids quantification, we observed a significant decrease in the levels of three amino acids related to the urea cycle —arginine, citrulline, and ornithine— and polyamine metabolites (putrescine, spermidine) in erythrocytes, plasma, and the inner ear during adolescence. However, the levels of the three urea cycle-related amino acids in erythrocytes, plasma, and the inner ear continued to decline, while polyamine metabolites in erythrocytes and the inner ear generally increased during aging (Figures 3D and 3E). Notably, there were differences in polyamine metabolites in plasma during aging, with putrescine and spermidine concentrations remaining relatively stable, while spermine levels significantly decreased (Figure 3E). The urea cycle by-product urea remained constant in erythrocytes and the inner ear across all four age groups of mice, while its concentration in plasma progressively increased with age (Figure 3D).

Overall, erythrocytes mirror the changes in urea cycle and polyamine metabolism in the inner ear from adolescence to aging. During adolescence, the urea cycle and polyamine metabolism are significantly downregulated in both erythrocytes and the inner ear. During aging, the urea cycle is further downregulated, leading to decreased detoxification capacity, while polyamine levels increase, triggering a polyamine stress response aimed at maintaining homeostasis. The polyamine stress response (PSR) plays a crucial role in cellular adaptation to environmental stress and the maintenance of cellular function. The PSR not only impacts the physiological state of individual cells but also affects energy homeostasis at the tissue and organ levels (Figure 3F).

Human translational retrospective studies by analyzing the UK Biobank metabolomic profiling in ARHL

To validate our murine findings, we analyzed human retrospective and prospective studies using UK Biobank metabolomic profiling to address whether plasma metabolites, especially those identified metabolites in our murine studies during the progression of ARHL, could effectively diagnose and predict ARHL. The cohort, recruited between 2006 and 2010, comprised 502,665 participants, with 110,786 undergoing plasma metabolomics analysis via NMR (Figure 4A). ARHL was defined based on self-reported questionnaires from UK Biobank participants, and further classified into hearing difficulty and hearing aid user groups (For detailed information, please refer to the STAR Methods section of the UK Biobank Database). The questionnaire has been validated in large-scale hearing loss studies.²⁰ Among the 502,665 recruits, 100,743 had hearing difficulty, and 14,963 used hearing aids. Supporting the aging-induced hearing decline, the prevalence of hearing difficulties was gradually escalated after the age of 45 (Figure 4B), the necessity for hearing aids rose sharply beyond 55 years (Figure 4C).

Figure 4.

Analysis of age-related plasma metabolites associated with ARHL: Data from the UK Biobank

(A) Schematic of experimental setup and analytic workflow involving analysis of age-related blood biomarkers from the UK Biobank associated with age-related hearing loss.

(B) The incidence of hearing difficulty increases with age in the UK Biobank cohort.

(C) The incidence of hearing aid use increases with age in the UK Biobank cohort.

(D) Forest plot of plasma metabolites with aging in individuals with hearing difficulty from the UK Biobank database. Red color indicates a significant increase. Blue color indicates a significant decrease. Black indicates no significant change.

(E) Forest plot of plasma metabolites with aging in individuals with hearing aid use from the UK Biobank. Red color indicates a significant increase. Blue color indicates a significant decrease. Black indicates no significant change.

(F) Multi-plasma metabolites logistic prediction model for hearing difficulty from the UK Biobank database.

(G) Multi-plasma metabolites logistic prediction for hearing aid use from the UK Biobank.

TFA, total fatty acid; USFA, unsaturated fatty acid; PUFA, polyunsaturated fatty acid; MUFA, monounsaturated fatty acid; SFA, saturated fatty acid; LA, linoleic acid; Ala, alanine; Gln, glutamine; Gly, glycine; His, histidine; Leu, leucine; Val, valine; Phe, phenylalanine; Tyr, tyrosine; GLU, glucose; Lac, lactate; Pyr, pyruvate; CA, citrate; 3HB, 3-hydroxybutyrate; ACA, acetoacetate.

Next, we compared the metabolome of total around 11,000 individuals including 61,764 controls and 39,498 individuals with hearing difficulties (Figure 4A). Among all of identified metabolites, monounsaturated fatty acids (MUFAs), branched chain amino acids (BCAAs, isoleucine, leucine, valine), tyrosine, creatinine, urea, and glycoprotein acetyls (GlycA) were significantly elevated in individuals with hearing difficulties, which are potential risk factors for hearing impairment, while unsaturated fatty acids (USFAs) and glycine were significantly dropped in individuals with hearing difficulties, which are protective factors against hearing loss (Figure 4D).

Given that individuals using hearing aids exhibit more severe ARHL, we performed a detailed analysis of the plasma metabolomics for this group. A total of 3,218 participants were assigned to the hearing aid users, while 19,904 participants served as controls (Figure 4A). In the case of hearing aid users, USFAs, omega-6 fatty acids, polyunsaturated fatty acids (PUFAs),linoleic acid, glycine, lactate, pyruvate, and albumin were protective against hearing loss, while MUFAs, alanine, glutamine, phenylalanine, tyrosine, BCAAs, glucose, acetoacetate, creatinine, and GlycA were risk factors (Figure 4E).

Furthermore, we constructed diagnostic models integrating diverse metabolite combinations identified in the forest plot, with the goal of evaluating its potential to distinguish clinical subgroups. Our analysis indicated that various metabolite combinations exhibited suboptimal diagnostic performance for the hearing difficulty group, as evidenced by a logistic regression model with an area under the curve (AUC) of merely 0.5664 (95% CI: 0.5608–0.5681; Figure 4F). In contrast, the diagnostic model for individuals with more severe hearing impairment, particularly those reliant on hearing aids, demonstrated significantly improved discriminatory capacity, with the optimal metabolite combination achieving an area under the curve (AUC) of 0.6778 (95% CI: 0.6679–0.6877; Figure 4G).

Sex difference in the levels of fatty acids and β-oxidization gene expression links with the sex dimorphism in ARHL

Given the limited diagnostic capacity of plasma metabolites for hearing difficulties, our subsequent analyses focus on hearing aid users, incorporating sex differences and a prospective cohort study design. Epidemiological evidence robustly supports a higher prevalence of ARHL in males compared to females.^21,22 Similarly, our cohort analysis revealed a markedly increased prevalence of hearing aid usage among males aged 55 and older relative to females (Figure 5A). The hearing aid users comprised 1,826 men and 1,392 women. Building on these findings, we conducted a focused investigation into sex-specific plasma metabolite differences in individuals with ARHL. We identified notable sex-specific differences in plasma metabolites among individuals with ARHL, particularly in fatty acids. TFAs, MUFAs, and SFAs were identified as risk factors for hearing loss in females. In contrast, several fatty acids, including omega-6 fatty acids, PUFAs, and linoleic acid, demonstrated protective effects against hearing loss in males (Figure 5B). Moreover, age-related analyses revealed a positive correlation between fatty acids, including TFAs, MUFAs, PUFAs, SFAs, omega-6 fatty acids, and linoleic acid, and age in females, while these associations were inversely correlated with age in males (Figure 5C). Other plasma metabolites exhibited consistent trends in the age-related analysis across genders. Alanine, phenylalanine, tyrosine, glucose, citrate, creatinine, and GlycA were positively correlated with age in both men and women, whereas, albumin, and glycine showed a negative correlation with age (Figure S4).

Figure 5.

Sex dimorphism in humans with ARHL links with sex differences in plasma fatty acids levels in humans and β-oxidization gene expression in murine inner ears

(A) The incidence of hearing aid use increases with age for men and women. After age 50, the prevalence in men rose significantly faster than in women.

(B) Forest plot of plasma metabolites with aging in men and women with hearing aid use from the UK Biobank. Red color indicates a significant change in female. Blue color indicates a significant change in male. Black indicates no significant change between genders.

(C) Correlations of plasma fatty acids levels with age in men and women from the UK Biobank.

(D) Age-dependent changes in Pparα, Cpt1a, Cpt2 and Octn2 mRNA levels for both female and male in the inner ear. Data are expressed as mean ± SEM. p values were assessed using the t test. N = 4 mice per group. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

The disparities in plasma fatty acid levels in ARHL likely reflect sex-specific differences in the activity of enzymes regulating fatty acid β-oxidation in the inner ear. Because it is impossible to have an access to human inner ears, we took advantage of our murine ARHL model. Given the complete hearing loss observed in 16-month-old (16M) mice, our investigation focused on β-oxidation related gene expression changes during the aging process from 2 to 10 months (2M–10M). To examine sex-specific differences, we conducted the gene expression profiling including a key transcription factor regulating fatty acid metabolic gene expression (Pparα), carnitine uptake (Octn2), fatty acid oxidization (Cpt1a, Cpt2) in the inner ears of male and female mice. First, no difference of the expression levels of these genes were detected in murine inner ears at 2M between females and males; Second, their expression was only markedly dropped in inner ears of males at 10 months (10M), while their expression levels were still maintained at similar levels as 2 months (2M) up to10 months (10M) (Figure 5D). These findings provide the precedent murine evidence of an early decline in β-oxidization related gene expression in male inner ears, implying a previously unrecognized potential molecular bases underlying sexual dimorphism in a higher incidence of ARHL in males.

Human prospective translational study: Identify biomarkers for predicting ARHL

Lastly, we performed the analysis of the prospective cohort study to identify the biomarkers for predicting ARHL. Briefly, we included a total of 6,289 participants enrolled between 2006 and 2010 and followed up using a standardized hearing questionnaire for the diagnosis of developed hearing decline in 2014. Among these individuals, 588 were identified as hearing aid users (Figure 4A). Next, multivariable analyses were performed and revealed that omega-6 fatty acids, linoleic acid, glycine, and albumin were significantly associated with a protective effect against the development of hearing loss (all p < 0.05). Conversely, BCAAs, tyrosine, creatinine, glycoprotein acetyls, and urea were independently associated with an increased risk of hearing loss (all p < 0.05; Figure 6A). Furthermore, a logistic regression model incorporating these metabolite combinations demonstrated moderate discriminatory capacity, with an area under the receiver operating characteristic curve (AUC) of 0.715 (95% CI: 0.6937–0.7362; Figure 6B). These findings provide human strong evidence conferring these metabolites serving as a biomarker predicting the risk of the development of hearing loss during aging.

Figure 6.

Forest plot analysis of plasma metabolites with aging in individuals with hearing aid use: prospective cohort study from the UK Biobank

(A) Forest plot analysis of plasma metabolites with aging in individuals with hearing aid use model from the UK Biobank. Red color indicates a significant increase. Blue color indicates a significant decrease. Black indicates no significant change.

(B) Multi-plasma metabolites logistic prediction model for hearing aid use cohort prospective study from the UK Biobank.

(C) Schematic graph showing cross-tissue energy metabolism reprogramming in erythrocytes, plasma and the inner ear with aging.

Taken together, retrospective case-control and prospective cohort studies of the UK Biobank hearing impaired population revealed the following: (1) Omega-6 fatty acids, linoleic acid and glycine were protective against ARHL. (2) BCAAs, tyrosine, creatinine, glycoprotein acetyls, and urea were risk factors for ARHL. (3) In females, most plasma fatty acids increase with age and serve as risk factors for ARHL. In males, however, most plasma fatty acids decrease with age, acting as protective factors against ARHL.

Discussion

Although the inner ear dysfunction has been well accepted as an early contributor for ARHL, how inner ear metabolic changes driven by ARHL remains unexplored and whether there are common metabolic shifts among inner ears, systemic circulating erythrocytes and plasma is unclear. Here, we report that murine erythrocytes and plasma as well as the murine inner ear share similar metabolic characteristics featured with age-dependent glucose metabolic reprogramming preferentially toward glycolysis over PPP coupled with the accumulation of long or medium-chain acylcarnitine and decreased arginine, lysine, proline and glycine in murine ARHL. Leveraging metabolome of a large-scale human cohort from UK Biobank including retrospective and prospective studies, we validated our murine findings and identified multiple beneficial (Omega-6 fatty acids, linoleic acid, glycine and albumin) and detrimental metabolites (BCAAs, tyrosine, creatinine, GlycA, and urea) during development of ARHL. Moreover, by integrating factors such as age and sex, we uncovered significant sex-specific differences in fatty acid levels and their associations with the higher risk of ARHL in males. Lastly, we provide strong human evidence that these identified metabolites are sensitive biomarkers to predict the risk of the development of hearing loss during aging (Figure 6C). Overall, those specific metabolic changes in erythrocytes and plasma mark the metabolic dynamics of inner ears, track the aging and predict the early age-related hearing decline in humans, immediately suggesting a therapeutic possibility.

Rapid glucose metabolism and reprogramming of glucose metabolism in both erythrocytes and inner ears with aging

It is well known that glucose, as the primary energy source for erythrocytes and the inner ear, plays a crucial role in maintaining their functions. Erythrocytes generate energy via glycolysis or produce NADPH through the PPP. The energy and reducing equivalents are crucial for maintaining erythrocyte morphology and function. ATP produced by glycolysis drives ion pumps that maintain electrolyte balance and ion concentrations inside and outside the cell, ensuring membrane stability and functionality. NADPH generated through the PPP is an essential cofactor for producing the antioxidant glutathione, protecting erythrocytes from oxidative damage. The metabolism, functionality, and deformability of erythrocytes are closely associated with age-related functional decline and ischemic hypoxic conditions.^23,24 Similarly, glucose, as the primary energy source for the inner ear, plays a critical role in energy supply, antioxidant protection, maintaining electrochemical gradients, and neurotransmitter homeostasis. Dysfunction in energy metabolism is a major pathogenic factor in ARHL. Studies have shown that a decline in glucose metabolism in the auditory cortex is also a significant cause of auditory aging. Enhancing protein levels of enzymes related to glycolysis, TCA cycle, and oxidative phosphorylation can improve glucose metabolism in the auditory cortex, thereby delaying auditory aging in guinea pigs.²⁵ Thus, it is critical to define the in vivo glucose metabolic utilization and reprogramming in erythrocytes and inner ears across the lifetime. We employed large-scale in vivo isotope-labeled glucose tracing to elucidate the differences in glucose utilization in both erythrocytes and inner ears of mice at different ages. The samples of erythrocyte and inner ear were collected at three time points after ¹³C₆-glucose injection, revealing rapid glucose metabolic kinetics in both erythrocytes and inner ears, with complete metabolism within 2.5 h. We identified the 0.5h time point as the most reliable for tracking the metabolic fate of glucose in both erythrocytes and inner ears, providing a detailed and comprehensive understanding of the impact of reprogramming changes on glucose metabolism. We observed consistent changes in glucose metabolism flux in both erythrocytes and inner ears during aging, with a shift in glucose metabolism flux from the PPP toward glycolysis. Erythrocytes partially mirror the changes in glucose kinetics within the inner ear during aging.

The reprogramming of glucose metabolism results in decreased antioxidant capacity in both erythrocytes and inner ears during aging, correlating with reported declines in glucose-6-phosphate dehydrogenase (G6PD) activity associated with aging.²⁶ Overexpression of G6PD enhances the PPP, protecting the inner ear from ROS-derived oxidative damage, reducing DNA damage and cell apoptosis, and thereby effectively delaying the progression of ARHL and aging.^27,28 This aligns with studies reporting a decline in G6PD activity in erythrocytes with aging, which disrupts metabolic reprogramming mechanisms. Moreover, supplementation with NAD⁺ and related antioxidants has been shown to prevent ARHL.^29,30,31 We elucidated the molecular mechanisms of oxidative stress in ARHL from the perspective of glucose metabolism reprogramming during aging. Recent research indicates that glycolysis in aging cells is enhanced in a PDK4-dependent manner. Pyruvate dehydrogenase kinase 4 (PDK4) inhibits PDH activity, facilitating a shift from mitochondrial oxidation to cytoplasmic anaerobic glycolysis, subsequently increasing lactate production. Moreover, lactate induces ROS production through NADPH oxidase 1 (NOX1) activation in aging cells,³² aligning with our observations of glucose metabolism remodeling in the inner ear. Studies have reported that the expression of glycolysis-related enzyme LDHB (lactate dehydrogenase B) is suppressed, disrupting the conversion of lactate to pyruvate. This reduction in pyruvate, a substrate for the TCA cycle in mitochondria, leads to mitochondrial dysfunction and ATP insufficiency, ultimately promoting ARHL.¹¹ Based on this, we further analyzed the incorporation of ¹³C-glucose into the TCA cycle and related amino acids. During aging, the TCA cycle in the inner ear shows a downregulated trend, resulting in inadequate inner ear function and impaired auditory signal transmission. In contrast, during adolescent stages, the upregulation of the TCA cycle in the inner ear is likely due to the high energy requirement for rapid growth and development. Thus, the regulation of the TCA cycle plays a critical role throughout the inner ear life cycle. Furthermore, research has shown that dietary intake of deuterium oxide (D₂O) can slow the breakdown of glucose and amino acids and reduce endogenous oxidative stress in the cochlea, thereby delaying the progression of ARHL in mice.⁶ This suggests that adjusting metabolic flux could be a potential target for mitigating ARHL in the future.

The metabolism of acylcarnitines in both erythrocytes and inner ears undergoes substantial changes from adolescence to aging

Acylcarnitines play a critical role in fatty acid metabolism and significantly impact erythrocyte and inner ear function. They are essential for energy supply, maintaining cellular membrane integrity and function, and lipid signaling. While lacking mitochondria, erythrocytes depend crucially on their membrane structure and function to maintain normal physiological activities. Since erythrocytes lack the ability to synthesize lipids de novo, they rely on the Lands cycle, which uses phospholipase-mediated removal of damaged acyl chains and replacement with undamaged fatty acids. This process depends on the balance between coenzyme A-conjugated acyl chains and acylcarnitines to repair the cell membrane.³³ Enhanced glycolysis and reduced PPP in erythrocytes indicate diminished antioxidant capacity, resulting in significant ROS accumulation during aging. ROS activation leads to lipid peroxidation and protein damage in the cell membrane.³⁴ The observed decrease in L-carnitine and increase in acylcarnitines in aging mice suggest membrane damage in erythrocytes during aging. This necessitates the production of acylcarnitines for proper membrane phospholipid remodeling and repair through the Lands cycle, maintaining the dynamic balance and function of erythrocyte membranes. Therefore, our findings suggest that the accumulation of medium and long-chain acylcarnitines during the aging process in mice may be closely associated with the remodeling of red blood cell membrane phospholipids.

Our findings suggest that the accumulation of long-chain acylcarnitines in the inner ear during aging result from dysfunctions in the transport of acylcarnitines into mitochondria. For example, inherited metabolic disorders such as carnitine palmitoyltransferase II (CPT II) deficiency affect fatty acid β-oxidation, leading to energy production defects impacting inner ear function. Additionally, the decline in antioxidant capacity and the increase in reactive oxygen species (ROS) during aging can impair mitochondrial function, further contributing to acylcarnitine accumulation. Long-chain acylcarnitines (LCACs) have been shown to significantly affect the Ca²⁺ homeostasis within cardiomyocytes,³⁵ which is essential for maintaining axonal structure and function. Notably, abnormal increases in intracellular Ca²⁺ levels within axons often lead to severe axonal degeneration, a phenomenon particularly common in peripheral neuropathies. Previous studies have demonstrated that when mitochondrial function is impaired, acylcarnitines accumulated in Schwann cells (SCs) are released into the surrounding axons, thereby disrupting axonal calcium homeostasis and stability. The accumulation of toxic lipid intermediates, such as LCACs, in mitochondrially dysfunctional SCs represents a potential pathological mechanism of peripheral neuropathy.³⁶ LCACs could accumulate and damage neurons by inducing astrocytic mitochondrial dysfunction in ischemic stroke.³⁷ Furthermore, related studies have confirmed that the abnormal accumulation of acylcarnitines is closely associated with poor myocardial ischemic prognosis, decreased insulin sensitivity, and chronic inflammatory responses, among other disease states.^19,38 The accumulation of acylcarnitines not only indicates impaired fatty acid utilization in the inner ear but also suggests that the buildup of long-chain acylcarnitines can lead to axonal degeneration and demyelinating neuropathies. We hypothesize that excessive acylcarnitine accumulation in the inner ear may cause auditory nerve degeneration, impeding auditory signal transmission.

Remodeling of amino acid metabolism during aging

Amino acids, fundamental to life activities, not only participate in protein synthesis but also play crucial roles in energy metabolism, signal transduction, and immune regulation. However, the precise quantification of inner ear amino acid metabolism throughout the lifespan remains unexplored. We observed significant declines in arginine, lysine, proline and glycine in erythrocytes, inner ear, and plasma with advancing age. These four amino acids also show marked downregulation in other aging tissues. For example, proline levels decrease with aging in mouse muscle and skin.^39,40 Proline supplementation can maintain mitochondrial homeostasis through mitophagy, enhance myogenic differentiation of aged mesenchymal stem cells,⁴¹ improve redox state, and increase nitric oxide (NO) bioavailability, thereby preventing or delaying angiotensin II (AngII)-induced hypertension and potentially extending lifespan by slowing cellular aging.^42,43 Similarly, lysine levels decline with aging in skeletal muscle and lens.⁴⁴ Research indicates that the reduction of lysine in aging tissues is associated with age-related metabolic pathologies such as lysine oxidation and lysine succinylation.^45,46 Arginine is crucial in metabolism as a precursor for NO, ornithine, creatine, and polyamines. A significant decrease in arginine may lead to reduced NO production, resulting in ischemic damage to cochlear neural structures.^47,48 Additionally, reduced NO impairs vasodilation,^49,50 potentially causing constriction of terminal vessels in the inner ear, affecting blood supply and leading to hair cell apoptosis. In a noise-induced hearing loss mouse model, L-arginine supplementation increased NO levels, protecting cochlear hair cells from oxidative stress by shifting glucose metabolism flux from glycolysis to the pentose phosphate pathway through S-nitrosylation and inhibition of pyruvate kinase M2 (PKM2).⁵¹

Glycine stands out as a key factor, with its levels not only declining in mice but also exhibiting an age-dependent reduction in plasma from the UK Biobank cohort. Both retrospective case-control studies and prospective cohort analyses indicate that glycine acts as a protective factor against ARHL. As a precursor to glutathione, an essential intracellular antioxidant, glycine plays a vital role in slowing the aging process.⁵² Clinical trials have demonstrated that the combination of glycine and N-acetylcysteine (NAC) can reverse glutathione synthesis deficits in older adults, resulting in improved muscle strength and cognitive function.^53,54 Our findings suggest that glycine serves as a crucial protective factor against ARHL, and its supplementation may alleviate oxidative stress in the inner ear via the glutathione synthesis pathway, thereby slowing the progression of ARHL. Therefore, our findings reveal that the levels of arginine, lysine, proline and glycine in the inner ear exhibit a consistent decline with those in erythrocytes, plasma, and the body overall during aging. These findings lead to future investigation of the effects of depleting arginine, lysine, proline, and glycine in ARHL through large-scale clinical trials, including dietary supplementation studies.

It is well known that the urea cycle is the primary pathway for elimination of nitrogen, an end product of amino acid metabolism, by converting toxic ammonia into urea for excretion. Studies have found that with aging, the urea cycle function in the inner ear weakens, leading to elevated ammonia levels. This process involves multiple physiological mechanisms, including direct neurotoxicity of ammonia, increased oxidative stress, and enhanced inflammatory responses. The decline in urea cycle efficiency raises ammonia concentration in the inner ear environment, causing cellular damage and auditory dysfunction. Extensive research indicates that polyamines have significant potential as biomarkers for age-related diseases. Elevated polyamine levels in blood and brain tissues are closely associated with age-related neurodegenerative diseases, such as Alzheimer’s and Parkinson’s.⁵⁵ Moreover, elevated serum spermidine levels in patients with mild cognitive impairment (MCI) who are at risk of developing Alzheimer’s disease further demonstrate the potential of polyamines as biomarkers for the progression from MCI to Alzheimer’s disease.^56,57 Additionally, some studies suggest that polyamine accumulation in age-related neurodegenerative diseases represents a chronic maladaptive polyamine stress response (PSR).⁵⁸ PSR is a biological response involving changes in polyamine metabolism and regulation under various stressors, including oxidative stress, inflammation, heat shock, or pathogen infection. Cellular senescence is an adaptive response to stress, and PSR plays a crucial role in the pathogenesis of age-related neurodegenerative diseases. As small organic cations, polyamines are critical in cellular proliferation, differentiation, apoptosis, gene expression regulation, ion channel activity, and antioxidant protection.⁵⁹ Our research indicates that polyamine metabolism in the inner ear is enhanced during aging, possibly reflecting a polyamine stress response. As the inner ear cells face increased stress due to aging, their demand for polyamines rises. Thus, inner ear cells may boost polyamine synthesis and utilization to counteract oxidative stress and maintain cellular function. Enhanced polyamine metabolism helps protect inner ear cells from oxidative damage, stabilize cell membranes, and ensure normal ion channel function, thereby mitigating the progression of hearing loss.

Intriguingly, recent epidemiological evidence suggests that dysregulation of acylcarnitine, the pentose phosphate pathway, and arginine and ornithine metabolism in the blood of elderly Chinese individuals is closely associated with a decline in intrinsic capabilities, including hearing and cognitive functions.^60,61 This finding strongly corroborates our research results. Moreover, extensive parabiosis experiments in mice have shown that the exchange of the blood system can improve the cognitive function and aging phenotypes of skeletal muscle, kidney, and liver in aging mice.^62,63,64 This indicates that factors and metabolic small molecules in the blood, along with erythrocytes, play an indispensable role, yet studies have not yet reported their effects on hearing improvement.

Identification of plasma small-molecule metabolic biomarkers with ARHL based on the UK Biobank

Based on the integrated findings from our retrospective and prospective cohort studies, we demonstrate that Omega-6 fatty acids, including linoleic acid, exert protective effects against ARHL. While the protective role of polyunsaturated fatty acids (PUFAs) in ARHL has been preliminarily confirmed, the majority of current research has focused on Omega-3 PUFAs. A 3-year prospective cohort study conducted in the Netherlands was the first to identify a significant negative correlation between plasma levels of ultra-long-chain Omega-3 PUFAs—particularly docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA), both of which possess neuroprotective properties—and the incidence of ARHL, with improvements in low-frequency hearing being particularly pronounced. Researchers have hypothesized that the protective mechanism may involve enhancements in cochlear microcirculation,⁶⁵ a theory supported by the Australian Blue Mountains Hearing Study, which found that regular consumption of Omega-3 PUFA-rich fish significantly reduces the risk of hearing loss. Mechanistic studies in animal models have uncovered several protective pathways for PUFAs.⁶⁶ In C57BL/6J mouse models, long-term supplementation with Omega-3 fatty acids delayed hearing loss through mechanisms such as the maintenance of IGF-1 signaling, modulation of the inflammatory factor balance, and support of homocysteine metabolism via BHMT-mediated methylation.⁶⁷ The Fat-1 transgenic mouse model further confirmed that endogenous enrichment of omega-3 fatty acids significantly delays ARHL progression.⁶⁸ However, clinical studies examining the relationship between Omega-6 fatty acids and ARHL are lacking, and several critical questions remain, including the efficacy differences between PUFA subtypes, the optimal ω-6/ω-3 ratio, and their specific roles in ARHL.

Remarkably, analyses of the UK Biobank database and mouse models uncovered a pronounced sex dimorphism in ARHL, characterized by differences in plasma fatty acid metabolites and the expression of fatty acid metabolism-related enzymes (Pparα, Cpt1a, Cpt2) in the inner ear. Additionally, prior studies have shown that female mice display enhanced resistance to the detrimental effects of high-fat diets on body weight, metabolism, and hearing, highlighting the critical role of sex-specific metabolic pathways in modulating hearing loss progression.⁶⁹

Additionally, our study revealed that metabolites such as branched-chain amino acids, tyrosine, creatinine, GlycA, and urea are significantly associated with increased ARHL risk, while glycine and albumin exhibit protective effects. Of particular interest is GlycA, a biomarker of acute and low-grade chronic inflammation, which has demonstrated significant value in various disease studies in recent years. Numerous clinical cohort studies have demonstrated that GlycA serves as a biomarker for subclinical cardiovascular diseases, chronic heart failure, psoriasis, inflammatory bowel disease, and acute pancreatitis.^70,71,72,73 Given that chronic inflammation is a key mechanism underlying ARHL, our research provides the first evidence of the potential value of GlycA as a plasma biomarker for ARHL. Overall, our research revealed multiple small metabolites are capable to predict ARHL, leading to future investigation to explore their potential prevention and treatment in the disease.

In summary, our study uncovered significant similarities in the metabolic remodeling of erythrocytes, plasma, and the inner ear during adolescence to aging, encompassing shifts in multiple metabolic pathways including glycolysis, PPP, oxidative stress, urea cycle and the synthesis of acylcarnitines, polyamines, and amino acids. Through UK Biobank-based metabolomics large data analyses, we identified a series of plasma biomarkers with protective and risk-associated effects on ARHL. Additionally, we observed gender-specific differences in the influence of various fatty acids on ARHL. From mouse models to large cohort human studies in ARHL, we integrated metabolomic data from erythrocytes, plasma, and the inner ear, emphasizing the role of circulating metabolites in spatiotemporal communication across these compartments. Overall, our study demonstrates that metabolic shifts in erythrocytes and plasma mimics the inner ear, offering diagnoses and predictions of ARHL. In the future, we aim to explore therapeutic targets and develop intervention strategies based on those identified metabolic markers. We will prioritize clinical trials involving dietary supplementation with various amino acids and the development of targeted metabolic modulation drugs, while also investigating synergistic treatment approaches in conjunction with traditional hearing rehabilitation methods.

Limitations of the study

This study acknowledges certain limitations. While the C57BL/6J mouse is a widely used model for ARHL, the early-onset hearing loss associated with the Ahl mutation in this strain does not fully recapitulate the progressive pathology observed in human ARHL. Future investigations could explore non-human primate models (such as rhesus or cynomolgus monkeys) in conjunction with metabolomic analysis of inner ear tissue derived from human autopsies, facilitating a more accurate elucidation of the metabolic regulatory mechanisms driving ARHL.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Yang Xia (yang.xia1106@csu.edu.cn).

Materials availability

The study did not generate new unique reagents.

Data and code availability

• •Data: All data have been deposited at the web (https://data.mendeley.com/datasets/v58pb4dhkc/1) and are publicly available as of the date of publication. Mendeley Data:https://doi.org/10.17632/v58pb4dhkc.1.
• •Code: The code is not currently accessible to the public, as it is associated with ongoing research that remains unpublished by our group.
• •Other items: Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We would like to acknowledge Hong Liu for substantial contributions to the analysis of UK Biobank data. This work is supported by the NSFC (82230023 to Y.X.) and Feifan Scholar Fund of Xiangya Hospital, Central South University (Y.X.); the Fuqing Project Program of Xiangya Hospital of Central South University (2209090555464 to C.L.); National Natural Science Foundation of China (82400873 to C.C.), Postdoctoral Fellowship Program (Grade B) of China Postdoctoral Science Foundation (GZB20240866 to C.C.), Natural Science Foundation of Hunan Province (2025JJ60551 to C.C.); National Natural Science Foundation of China (82301514 to F.Y.), the Natural Science Foundation of Hunan Province (2023JJ41018 to F.Y.); the National Natural Science Foundation of China (youth program, 82401827 to W.L.), the China Postdoctoral Science Foundation (2023M743968 to W.L.), and Hunan Provincial Natural Science Foundation (youth program, 2023JJ40919 to W.L.).

Author contributions

Q.G. designed and conducted the experiments, drew the figures, and wrote the manuscript; Z.L. collected and analyzed the experimental data; C.L. performed UK Biobank data curation and analyses; C.C. took charge of infusing [U-¹³C₆] glucose in vivo; X.L. and Y.W. collected inner ear tissue samples; Y.Z. and X.L. managed the collection and preliminary processing of erythrocytes and plasma; W.L. and F.Y. handled the initial data processing; Q.Q. and Y.Z. guided the process of metabolomics data analysis; R.E.K. provided critical feedback on the manuscript and expertise in metabolism; Y.X. oversaw the design of experiments and interpretation of results, the writing and organization of the manuscript, and did final editing.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Biological samples
Mouse erythrocyte samples	This paper	N/A
Mouse plasma samples	This paper	N/A
Mouse inner ear samples	This paper	N/A
Deposited data
Raw metabolomics data	This paper	Mendeley Data: https://doi.org/10.17632/v58pb4dhkc.1
Experimental models: Organisms/strains
C57BL/6J mice	Aniphe Biolaboratory Inc
Software and algorithms
MetaboAnalyst	iCarbonX Diagnostics (Zhuhai) Company Limited	https://www.metaboanalyst.ca/
Graphpad Prism 10	Graphpad Software
R	R Foundation for Statistical Computing, Vienna, Austria	https://www.r-project.org

Experimental model and study participant details

Experimental model

All animal protocols followed the guidelines of the World Organization for Animal Health and national standards for animal research in China. All procedures were approved by the Ethics Review Committee of Xiangya Hospital, Central South University (CSU-2023-0336). Animal husbandry and euthanasia were conducted in accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals. Upon acquisition, a total of 80 wild-type C57BL/6J mice at four different age groups (1M, 2M, 10M, 16M) were selected, with 10 male and 10 female mice per group. Gender had no impact on the outcomes.The mice were housed under controlled conditions for one week at 24 ± 1°C with a 12-hour light/dark cycle and provided with water and standard rodent food ad libitum.

Study participant

The UK Biobank represents the largest repository of genetic and environmental data in the UK, aimed at elucidating disease etiology and preventive strategies. This comprehensive database includes genetic profiles, blood samples, and longitudinal health records from 502,665 volunteers aged 37–74, recruited between 2006 and 2010, with decades of follow-up. Among them, 110,786 individuals who underwent plasma metabolomics profiling using nuclear magnetic resonance (NMR) were included in the present study. The study, approved by the NHS National Research Ethics Service (Ref: 11/NW/0382), obtained written informed consent from all participants. Participants completed self-reported questionnaires on hearing health, enabling the classification of ARHL cases and controls as outlined in the table below.

Study reference	Definition	Detail criteria
		Case	Control	Exclusion
UK Biobank	Self-reported hearing difficulty	If answered “yes” to Question 1 or Question 2	If answered “no” to both Question 1 and 2	Conductive hearing loss
	Self-reported hearing aid use	If answered “yes” to Question 3	If answered “no” to all these questions

Questionnaires

• Question 1: Do you have any difficulty with your hearing? [UK Biobank field ID: 2247]

• Question 2: Do you find it difficult to follow a conversation if there is background noise (such as TV, radio, or children playing)? [UK Biobank field ID: 2257]

• Question 3: Do you use a hearing aid most of the time? [UK Biobank field ID: 3393]

If a participant answered, ‘Yes’ to ‘Do you have any difficulty with your hearing?’ or ‘Do you find it difficult to follow a conversation if there is background noise (such as TV, radio, children playing)?’, they were classified as a hearing difficulty case. Participants who answered ‘No’ to these questions were classified as controls. Individuals who selected the answer, “I don’t know’or did not respond or declined to answer were excluded. If a participant answered, ‘Yes’ to ‘Do you use a hearing aid most of the time?’ they were classified as a hearing aid user case. Participants who answered ‘No’ to the question were classified as controls.

In the retrospective analysis, two independent case-control studies were conducted based on the UK Biobank standardized self-reported hearing questionnaire: 1) a hearing difficulty group (n=39,498) and a control group (n=61,764); 2) a hearing aid users group (n=3,218) and a control group (n=19,904). To further investigate gender differences, hearing aid users were stratified by sex (1,826 males, 1,392 females). The prospective cohort analysis utilized 2014 follow-up data, including 588 newly initiated hearing aid users and a control group of 5,701 individuals.

Method details

ABR

Mice were deeply anesthetized via intraperitoneal injection with a mixture of ketamine (100 mg/kg) and xylazine (10 mg/kg). Auditory tests were conducted in a soundproof chamber, with mice placed on a heating pad to maintain body temperature. Post-anesthesia, response signals were captured using three electrodes: the recording electrode needle was inserted subcutaneously at the vertex of the mouse skull, the reference electrode needle was placed subcutaneously behind the mouse’s ear at the mastoid region, and the ground electrode was positioned along the midline of the back. Auditory stimuli were generated by System 3 digital signal processing hardware and software (Tucker-Davis Technologies), consisting of pure tone bursts ranging from 4 to 32 kHz (0.1 ms rise/fall time, 2 ms duration, and 21 presentations/second). ABR waveforms were recorded at initial sound pressure levels of 90 dB SPL, decreasing in 5 dB intervals, until waveforms were barely distinguishable. Thresholds were defined as the minimum stimulus level eliciting clear peaks for waves I to V. Amplitudes (I and II) were measured by subtracting trough values from peak values, and latencies were calculated by subtracting the onset time of peak amplitude of wave I from that of wave II.

¹³C₆-glucose tracing and tissue collection

Mice were intravenously injected with 5% D-Glucose (U-¹³C₆) at a dose of 5 ml/kg through the tail vein. Anesthesia and cardiac blood collection were performed at 0.5h, 1.5h, and 2.5h post-injection. The collected venous blood was centrifuged at 2000xg for 5 minutes at 4°C. The supernatant plasma and lower layer erythrocytes were separated and stored at -80°C. Following cardiac blood collection, the mice were perfused, and the temporal bones were promptly harvested. The otic bullae and surrounding skull bones were isolated and placed in a dish containing PBS. Adherent muscles and soft tissues were carefully removed to obtain comprehensive inner ear tissue, including the cochlea, vestibular organs, organ of Corti, sensory epithelium, sensory neurons, and associated nerves. The entire inner ear was rapidly frozen in liquid nitrogen and stored at -80°C.

Sample preparation

Preparation of erythrocyte and plasma samples: Thaw erythrocyte and plasma samples on ice. Pipette 20 μl of erythrocytes or plasma, then add pre-cooled extraction solution (Methanol:Acetonitrile:Water,5:3:2,v/v/v), with 180 μl added to erythrocytes and 230 μl to plasma. Shake for a few seconds, then place on a shaker at 4°C for 30 minutes. Centrifuge at 18300xg for 10 minutes at 4°C. For erythrocytes, take 100 μl of the supernatant, and for plasma, take 200 μl of the supernatant for analysis.Preparation of Tissue Samples: Weigh 3-10 mg of inner ear and add pre-cooled extraction solution (Methanol:Acetonitrile:Water,5:3:2,v/v/v) to prepare a final concentration of 15 mg/mL in the buffer. Add small steel beads and homogenize using a grinder at 70 Hz for 5 minutes at 4°C. Then, place on a shaker at 4°C for 30 minutes. Centrifuge at 18300xg for 10 minutes at 4°C. Take 100 μl of the supernatant for analysis.

UHPLC-MS metabolomics

Metabolomic analysis was conducted using the ThermoFisher Dionex UltiMate 3000 UHPLC coupled with a Q Exactive HF MS. A 5-minute gradient analysis method from the literature was employed, utilizing a Kinetex C18 column (2.1 × 150mm, 1.7μm) under the following chromatographic conditions: flow rate of 0.45 ml/min, column oven temperature of 45°C, sample temperature of 4°C, and injection volume of 10 μl. In the positive ion mode, the mobile phase consisted of solvent A (0.1% formic acid in water) and solvent D (0.1% formic acid in acetonitrile), with the elution gradient set as follows: 0-0.5 min, 5% D; 0.5-1.1 min, 5%-95% D; 1.1-2.75 min, 95% A; 2.75-3 min, 95%-5% A; 3-5 min, 5% A. In the negative ion mode, the mobile phase included solvent B (95% acetonitrile/1 mM ammonium acetate in water) and solvent C (5% acetonitrile/1 mM ammonium acetate in water), with the elution gradient set as: 0-0.5 min, 100% C; 0.5-1.1 min, 100% B; 1.1-2.75 min, 100% B; 2.75-3 min, 100%-0% B; 3-5 min, 100% C. Mass spectrometry parameters were configured as follows: resolution 60,000, scan range 65-900 m/z, maximum injection time 200 ms, automatic gain control (AGC) 3 ×10ˆ6 ions, sheath gas 45, auxiliary gas 15, and sweep gas 0. All samples were randomly injected and independently analyzed in both positive and negative ion modes.

Amino acid quantification

Stable isotope-labeled internal standards for 15 amino acids (arginine, lysine, proline, histidine, glutamate, aspartate, glycine, phenylalanine, tyrosine, leucine, valine, alanine, methionine, serine, and threonine) were added to the extraction solution (Methanol:Acetonitrile:Water,5:3:2,v/v/v=5:3:2). During the preparation of erythrocyte, plasma, and inner ear tissue samples, the extraction solution containing the stable isotope-labeled internal standards was added. The remaining steps followed the protocol for untargeted metabolomics.

Quantitative RT-PCR

Total RNA was isolated from the inner ear using TRIzol reagent (Invitrogen). Reverse transcription was performed using a cDNA synthesis kit, followed by quantitative real-time PCR (qRT-PCR) analysis with SYBR Green PCR Master Mix (Qiagen) on a LightCycler 480 Instrument II (Roche). Sequences were as follows: Mouse Pparα, (forward,5′-TGCAAACTTGGACTTGAACG-3′; reverse, 5′-GAT CAGCATCCCGTCTTTGT-3′), Mouse Cpt1a, (forward,5′-GGTCTTCTCGGGTC GAAAGC-3′;reverse 5′-TCCTCCCACCAGTCACTCAC-3′),Mouse Cpt2, (forward, 5′-CAAAAGACTCATCCGCTTTGTTC-3′;reverse,5′-CATCACGACTGGGTTTGG GTA-3′), Mouse Octn2, (forward,5′-GGACCAGAAACTTAACAACGACG-3′; reverse ,5′-CAGGCTGTGTGAATGGACCT-3′).

UK Biobank database

Baseline characteristics were summarized using descriptive statistics, including means and standard deviations for continuous variables and frequencies and proportions for categorical variables. The prevalence of hearing impairment and hearing aid utilization was calculated across predefined age strata.

Metabolite concentrations were categorized into tertiles (low, medium, high) based on their distribution. The distribution of cases and controls within each tertile was examined using contingency tables. Disease risk associated with per standard deviation increase in metabolite levels was estimated through univariate logistic regression models, with corresponding 95% confidence intervals (95% CI) and linear P-values calculated for each metabolic marker. An additive interaction model was employed to assess effect modification by sex on the association between metabolite levels and hearing aid utilization, with interaction P-values calculated.

Longitudinal analysis of hearing aid utilization was conducted using follow-up data from 2014. A two-tailed P-value <0.05 was considered statistically significant. The discriminatory performance of each metabolite was evaluated using Harrell’s concordance (C) statistic, a rank-order measure of predictive accuracy ranging from 0.5 (no discrimination) to 1.0 (perfect discrimination). All statistical analyses were performed using R software, version 4.4.2 (R Foundation for Statistical Computing, Vienna, Austria).

Data processing and analysis

Raw data files were converted to mzXML format using RawConverter (The Scripps Research Institute) and analyzed with Maven (Princeton University, Princeton, NJ). Processed data were further analyzed on MetaboAnalyst. Sample data underwent sum normalization, log transformation, and autoscaling before Partial Least Squares Discriminant Analysis (PLS-DA). Differential metabolites were identified based on a p-value less than 0.05 and a fold change greater or less than 1. Pathway enrichment analysis was performed on the identified differential metabolites. Additionally, heatmaps of acyl-carnitine were generated using the average values of the raw data via MetaboAnalyst. Heatmaps for amino acid quantification were created using ratios to the relative mean concentration of 1M, also utilizing MetaboAnalyst.

Quantification and statistical analysis

Data are presented as mean ± standard deviation. For comparisons across multiple groups, one-way or two-way ANOVA was employed as indicated in the figure legends, followed by multiple comparisons to determine the p-value for comparisons between groups. All statistical analyses were performed using GraphPad Prism 10.0. A p-value of less than 0.05 was considered statistically significant and is denoted as ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001, and ∗∗∗∗ p < 0.0001.

Additional resources

No additional resources were generated in the study.

Published: August 5, 2025