Spatially resolved in vivo ratiometric imaging of mitochondrial quality control using co-registered confocal-OCT

Abstract

Mitochondrial quality control is essential for maintaining cellular bioenergetic homeostasis, but in vivo monitoring of its spatial dynamics remains difficult to quantify. Here we present a multimodal retinal imaging approach that enables spatially resolved ratiometric assessment of mitochondrial turnover using a hybrid confocal–optical coherence tomography (OCT) platform. Dual-wavelength confocal excitation of the pH-sensitive mt-Keima reporter is intrinsically co-registered with structural OCT and OCT angiography (OCTA), permitting compartment-specific quantification of excitation-dependent fluorescence within vascular and avascular retinal regions. In vivo imaging in mt-Keima transgenic mice demonstrated robust ratiometric separation of neutral and acidic mitochondrial environments and sensitivity to pharmacologic enhancement of mitophagy using 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR). OCTA-guided segmentation further revealed spatial heterogeneity in excitation-dependent mitochondrial signals between vascular and avascular retinal regions.

1. Introduction

Mitochondrial dysfunction is a central contributor to retinal metabolic stress and neurovascular pathology, reflecting the exceptionally high energy demand and continuous oxidative load of photoreceptors and inner retinal neurons [1–4]. These cells rely on tightly regulated mitochondrial turnover to sustain Adenosine Triphosphate (ATP) production, calcium homeostasis, and redox balance. Selective autophagic clearance of damaged mitochondria, termed mitophagy, is therefore a critical quality control mechanism that preserves cellular bioenergetic integrity and limits accumulation of dysfunctional organelles [5,6]. Disruption of mitophagy has been implicated in retinal aging, diabetic retinopathy, and other neurodegenerative retinal diseases, where impaired mitochondrial clearance has been reported to precede detectable structural degeneration and vascular dysfunction [7–10]. Despite its central biological role, longitudinal, noninvasive assessment of mitochondrial quality control within intact retinal tissue remains technically limited. In vivo retinal imaging is particularly constrained by depth-dependent scattering, vascular projection artifacts, motion-induced decorrelation, and fixation variability [11], and the absence of intrinsic molecular contrast capable of reporting organelle-specific processes.

Fluorescent mitophagy reporters such as mitochondrial-targeted Keima (mt-Keima) enable excitation-dependent discrimination between mitochondria residing in neutral mitochondrial environments and those delivered to acidic lysosomal compartments [6,12]. mt-Keima exhibits excitation maxima near ∼445 nm under neutral pH and ∼586 nm under acidic conditions, with emission centered near ∼620 nm (Fig. 1), enabling ratiometric discrimination of mitochondria residing in distinct intracellular environments. This excitation-shift property enables ratio-based normalization, reducing sensitivity to expression variability, illumination inhomogeneity, and scattering-related intensity bias [12]. Although mt-Keima has been extensively validated in cultured cells and ex vivo tissue preparations, translation to in vivo retinal imaging remains challenging due to optical scattering, limited depth discrimination, and the absence of intrinsic anatomical or vascular reference necessary for spatially resolved analysis [13]. Conventional confocal fluorescence imaging provides molecular specificity but lacks structural and vascular context, whereas optical coherence tomography (OCT) and optical coherence tomography angiography (OCTA) provide depth-resolved anatomical and angiographic information without molecular specificity. This disconnect underscores a methodological gap between molecular reporters and structurally contextualized retinal imaging modalities. These limitations highlight the need for an integrated imaging strategy that combines ratiometric mitophagy detection with co-registered anatomical and vascular guidance in vivo.

Fig. 1.

(A) Normalized excitation spectra of mt-Keima under neutral (green, pH ~∼7) and acidic (red, pH <5) conditions, illustrating a dominant excitation peak near 445 nm in the neutral mitochondrial environment and a shifted excitation band toward ~560–590 nm under acidic lysosomal conditions. Emission remains centered near ~620 nm (yellow dashed). (B) Transition of mitochondria from a neutral cytosolic environment to acidic lysosomal compartments during mitophagy, which underlies the excitation-dependent fluorescence shift.

To address this need, we developed a hybrid confocal OCT retinal imaging system that integrates dual-wavelength confocal fluorescence excitation (445 nm and 561 nm) with co-registered OCT and OCTA through a shared optical path and galvanometric scanning configuration. Excitation at 445 nm and 561 nm enables ratiometric assessment of mt-Keima fluorescence [6,7,10,14]. The shared optical architecture ensures intrinsic spatial co-registration between confocal fluorescence and OCTA-derived vascular information, enabling anatomically constrained ratio mapping within defined retinal compartments. The system was evaluated in vivo in transgenic mt-Keima mice under basal conditions and following pharmacological enhancement of mitochondrial turnover using 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR), an activator of AMP-activated protein kinase (AMPK) known to enhance mitophagy [15,16]. Co-registered OCTA enabled segmentation of distinct retinal vascular beds and surrounding avascular parenchyma within the same imaging field, permitting compartment-specific quantification of excitation-dependent fluorescence ratios. In vivo measurements demonstrated reproducible separation of mt-Keima signals associated with neutral and acidic mitochondrial environments and revealed condition-dependent increases in the 561 nm / 445 nm fluorescence ratio following AICAR administration. The platform further enables compartment-specific quantification of mitochondrial turnover within OCTA-defined vascular and avascular retinal regions via ratio analysis. In addition, we demonstrate in vivo sensitivity of the system to pharmacologic modulation of mitochondrial quality control using AMPK activation.

2. Materials and methods

2.1. Animal model and experimental preparation

All animal procedures conformed to the Association for Research in Vision and Ophthalmology (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Illinois Chicago (UIC). Seven-week-old C57BL/6J mice transgenic for the mt-Keima mitophagy reporter were used for in vivo imaging. A total of n = 7 mice, including both males (5) and females (2), were included in this study. Both eyes from each animal were imaged at baseline prior to any intervention, yielding a total of 14 eyes for basal measurements. Following baseline imaging, one eye from each animal received intravitreal AICAR injection, while the contralateral eye received vehicle (PBS), forming a paired-eye experimental design. Post-treatment imaging was performed on the same animals 24 hours after injection. For quantitative analysis, paired comparisons between AICAR-treated and PBS-treated eyes were performed within each animal (n = 7), with basal measurements serving as within-subject references. Animals were housed under pathogen-free conditions on a 12-hour light/dark cycle with free access to food and water.

Mitophagy was pharmacologically enhanced by intravitreal injection of AICAR. Injections were performed under anesthesia and visualization using a surgical microscope with a customized 33-gauge Hamilton needle (0.8 µL per eye, ∼2–3 mm posterior to the limbus). The right eye received AICAR (30 µg/µL; Cayman Chemical), while the contralateral eye received an equal volume of vehicle consisting of phosphate-buffered saline (PBS). Imaging was conducted 24 hours after injection.

Anesthesia was induced via intraperitoneal injection of ketamine (100 mg/kg) and xylazine (5 mg/kg). Body temperature was maintained using a thermostatically controlled heating pad. Pupils were dilated with 0.5% tropicamide, and sterile lubricant gel was applied to the cornea to maintain optical clarity. Mice were positioned to align the eye with the optical axis of the imaging system, and physiological stability was maintained throughout data acquisition.

2.2. Hybrid confocal OCT retinal imaging system

A custom hybrid confocal OCT retinal imaging system was developed for co-registered in vivo ratiometric fluorescence imaging and OCT acquisition. The system integrates dual-wavelength confocal fluorescence excitation with spectral-domain OCT through a shared optical path and common galvanometric scanning, enabling intrinsically aligned fluorescence, structural, and vascular datasets.

The spectral characteristics of mt-Keima excitation and emission are shown in Fig. 1. Based on these excitation properties, the confocal fluorescence subsystem utilized two diode lasers operating at 445 nm and 561 nm (OBIS LX 445 and OBIS LS-561; Coherent Inc., Santa Clara, CA, USA), corresponding to the neutral and acidic excitation states of mt-Keima.

The overall optical layout is illustrated in Fig. 2, showing co-axial alignment of the confocal excitation beams (445 nm and 561 nm) with the OCT sample arm via shared galvanometric scanners. The OCT subsystem employed a broadband superluminescent diode (SLD) (MD-810-HP; Superlum, Cork, Ireland) with a center wavelength of 810 nm and a bandwidth of 100 nm. Light was divided into sample and reference arms using a 75:25 fiber coupler (FC1) (TW805R3F2; Thorlabs, Inc., Newton, NJ, USA). Back-reflected light was detected using a custom spectrometer incorporating a 1200 lines/mm transmission diffraction grating (Wasatch Photonics, West Logan, UT, USA) and a 2048-pixel line-scan camera (Teledyne e2v, Chelmsford, UK) [17].

Fig. 2.

Schematic of the integrated optical system combining dual-wavelength confocal fluorescence excitation with spectral-domain OCT. CCD, charge-coupled device camera; CL, collimating lens; DM, dichroic mirror; FC, fiber coupler; F, filter; L, relay or imaging lens; LD, laser diode; PC, polarization controller; PMT, photomultiplier tube; SLD, superluminescent diode; SM, scanning mirrors. Confocal excitation at 445 nm and 561 nm is co-axially aligned with the OCT sample arm using shared galvanometric scanners, enabling intrinsically co-registered en face fluorescence and OCT acquisition. Retinal vascular contrast is derived from repeated OCT B-scans acquired at identical transverse locations.

The OCT subsystem operated at an A-line acquisition rate of 70 kHz. The axial resolution was ∼2.9 µm in air, corresponding to ∼2.1 µm in retinal tissue, assuming a refractive index of 1.38, and the lateral resolution at the retinal plane was ∼2 µm. Laser output power at the corneal plane was ∼46.45 µW (445 nm) and ∼68.26 µW (561 nm), measured with the scanning beam parked at the corneal plane. These powers are within the range reported for in vivo mouse retinal fluorescence imaging systems and were maintained below ANSI Z136.1 ocular safety limits for scanned retinal exposure.

Fluorescence emission centered near ∼620 nm was directed through the detection path via a dichroic mirror (DM1) (DMLP650R; Thorlabs, Inc., Newton, NJ, USA), followed by relay lenses (L7 and L8), and then reflected by a second dichroic mirror (DM2) (DMSP567R; Thorlabs, Inc., Newton, NJ, USA) toward the detection arm. The signal was subsequently collected by lenses (CL5 and CL6), spectrally filtered using an emission filter (F), and detected by a photomultiplier tube (PMT) (PMM01-1; Thorlabs, Inc., Newton, NJ, USA), as illustrated in Fig. 2. The confocal subsystem provided a lateral resolution of ∼2 µm and an axial section thickness of ∼34 µm, determined by the optical configuration and pinhole geometry. Confocal excitation and OCT illumination were co-axially aligned and raster-scanned using shared galvanometric scanners, ensuring intrinsic spatial correspondence between structural, vascular, and fluorescence datasets. The optical configuration maintained parfocal alignment between the confocal fluorescence detection path and the OCT imaging beam at the retinal plane, enabling spatially co-registered acquisition of fluorescence and OCT data.

2.3. Data acquisition

In vivo imaging was performed over a retinal field of view (FOV) of 1.2 mm × 1.2 mm. OCT volumes were acquired using 1000 A-lines per B-scan and 1000 B-scans per volume, corresponding to a transverse sampling interval of 1.2 µm in both scan directions. For OCTA acquisition, each transverse location was sampled with four repeated B-scans before advancing to the next position. For each experimental condition, five volumetric datasets were acquired sequentially, using identical scan parameters. Confocal fluorescence imaging was performed in an en face configuration using dual-wavelength excitation at 445 nm and 561 nm, acquired sequentially for each excitation wavelength to minimize spectral cross-talk and ensure accurate ratiometric analysis. Confocal and OCT datasets were acquired sequentially within the same imaging session without repositioning of the animal. All imaging parameters, including scan geometry and repetition schemes, were held constant across all conditions to ensure quantitative comparability of excitation-dependent fluorescence measurements.

2.4. Quantification of gene expression in the retinal vasculature

Retinas from wild-type, non-transgenic C57BL/6J mice were isolated from freshly enucleated eyes and incubated with 1 mM AICAR for 3 hours at 37 °C. Following incubation, retinal vasculature was isolated by hypotonic lysis in distilled water for 1 hour, followed by DNase treatment (200 U, Worthington #LS006330) for 10 minutes. The neuroretinal layer was then gently removed from the vascular plexus. Total RNA was extracted from the isolated retinal vessels (Qiagen), and cDNA was synthesized using standard reverse transcription protocols (HighCapacity cDNA Reverse Transcription Kit, ThermoFisher). The expression levels of mitophagy, mitochondrial biogenesis, and mitochondrial mass genes were then quantified by qPCR (QuantStudio 7 Flex system, Applied Biosystems) using Fast SYBR Green Master Mix (Applied Biosystems). The primer sequences used in these experiments are listed in Table 1.

Table 1. List of Primers used for qRT-PCR.

Gene	Forward Primer	Reverse Primer
Cox4	5’-GGA GTG TTG TGA AGA GTG AAG A-3’	5’-CTG GAT GCG GTA CAA CTG AA-3’
Tomm20	5’-CAT CTG GCT TGG TGT GTA TTT G-3’	5’-ATC AAG GAA GAA GGG CCT AAC-3’
Tfam	5’-AGG CCC AAG AAA CCA GTT AG-3’	5’-AGG CAG AAC TCA GCA TCA TC-3’
Opa1	5’-ATG ATG AGC ACC CAG CTT AC-3’	5’-CCT CTG CGA CAA AGG TTA CA -3’
Pink1	5’-GGC TAG TGA TCT CAG ACT TTG G-3’	5’-GGT ATC GGC TTT GCT GTA GT-3’
Actb	5’-GAG AGG TAT CCT GAC CCT GAA-3’	5’-TGA AGG TCT CAA ACA TGA TCT GG-3’

2.5. Image processing and quantitative analysis

En face OCTA contrast was derived from repeated OCT B-scans acquired at identical transverse locations using an intensity-variance approach. Temporal intensity variance across repeated B-scans was used to distinguish dynamic blood flow from static retinal tissue. Specifically, OCTA contrast at each voxel was computed as the sample variance of the OCT intensity across the repeated B-scan index (N = 4). En face vascular projection maps were generated by axial integration of the OCTA volume across the retinal depth and subsequently binarized using Otsu thresholding applied to the en face OCTA projection to distinguish flow-related signal from background tissue, followed by morphological filtering to remove isolated noise pixels and refine the vascular mask. The resulting binary masks defined vascular and complementary avascular parenchymal regions for compartmentalized analysis. This variance-based OCTA strategy follows previously established methodologies [18].

Confocal fluorescence images acquired under 445 nm and 561 nm excitation were processed using a custom MATLAB (MATLAB R2023b, MathWorks, USA)-based pipeline. Digitized detector signals were reconstructed into two-dimensional image frames and organized into temporal image stacks. Because confocal fluorescence and OCT/OCTA data were acquired using a shared optical path and common galvanometric scanning configuration, intrinsic spatial co-registration between modalities was maintained, ensuring pixel-level correspondence between structural, vascular, and fluorescence datasets. To correct for residual motion within the confocal image sequences, lateral alignment between consecutive frames was performed using a discrete Fourier transform (DFT)-based subpixel image registration method [19]. The aligned frames were then averaged temporally to improve signal stability and reduce frame-to-frame variability. Identical reconstruction and motion-correction procedures were applied independently to both excitation channels to maintain spatial correspondence prior to ratiometric analysis. Ratiometric mt-Keima maps were subsequently generated by computing the fluorescence intensity ratio of the 561 nm to the 445 nm excitation channels. These 561 nm/445 nm ratios provide a normalized measure of acidic-to-neutral mt-Keima signal while reducing sensitivity to absolute intensity fluctuations. Following motion correction and temporal averaging, OCTA-derived vascular masks were applied directly to the spatially corresponding confocal images for compartment-specific analysis. All processing steps were applied consistently across experimental conditions.

2.6. Statistical analysis

All imaging-derived quantitative data are presented as median and interquartile range unless otherwise noted. For qPCR-based gene expression analysis, data are presented as mean ± standard deviation. Due to the small sample size (n = 7) and to avoid assumptions of normality, non-parametric statistical methods were used. Paired comparisons between groups were performed using the Wilcoxon signed-rank test. Comparisons across multiple conditions (basal, PBS-treated, and AICAR-treated) were performed using the Friedman test, followed by post hoc pairwise comparisons using the Wilcoxon signed-rank test. A significance threshold of p < 0.05 was considered statistically significant. Statistical analyses were performed using MATLAB (MathWorks, USA) and OriginPro (OriginLab Corp., USA).

3. Results

Figure 3 shows representative in vivo imaging of mt-Keima mice under basal conditions. OCTA shows retinal vasculatures with microcapillary level resolution (Fig. 3(A1)). Binary OCTA processing was used to generate vascular (Fig. 3(B1)) and avascular (Fig. 3(C1)) maps of the imaged retinal field. These segmentation masks enabled compartment-specific analysis of mt-Keima fluorescence.

Fig. 3.

OCTA-guided regional segmentation and dual-excitation mt-Keima fluorescence imaging under basal conditions in an ONH-centered retinal field. (A1) En face OCTA image of the imaged retinal region centered on the optic nerve head (ONH). (A2–A3) Confocal fluorescence images acquired under 445 nm and 561 nm excitation, respectively. (B1) OCTA-derived vascular mask and (B2–B3) corresponding 445 nm and 561 nm confocal fluorescence images restricted to vascular regions. (C1) Inverse OCTA-derived mask defining avascular parenchyma and (C2–C3) corresponding confocal fluorescence images under 445 nm and 561 nm excitation.

Confocal imaging demonstrated that mt-Keima fluorescence was predominantly detected under 445 nm excitation, with comparatively minimal signal acquired under 561 nm excitation (Fig. 3(A2–A3)). Following OCTA-guided segmentation, vascular regions exhibited strong fluorescence under 445 nm excitation and sparse signal under 561 nm excitation (Fig. 3(B2–B3)). A similar excitation pattern was observed within avascular parenchymal regions, where 445 nm fluorescence remained prominent and 561 nm excitation contributed minimally to the overall signal (Fig. 3(C2–C3)). Across both vascular and avascular compartments, fluorescence intensity was consistently higher under 445 nm excitation compared with 561 nm excitation.

Figure 4 shows representative in vivo imaging of vehicle-treated mt-Keima mice. Intravitreal injection of the vehicle (PBS) caused a modest increase in excitation-dependent mt-Keima fluorescence (Fig. 4(A3)), compared with basal conditions (Fig. 3(A3)). Dual-excitation imaging showed that fluorescence across the retinal field remained predominantly driven by 445 nm excitation (Fig. 4(A2)), with comparatively lower signal detected under 561 nm excitation (Fig. 4(A3)). OCTA-guided segmentation revealed similar qualitative patterns in both vascular regions (Fig. 4(B2–B3)) and the avascular parenchyma (Fig. 4(C2–C3)), with the neutral 445 nm component remaining dominant despite a detectable acidic 561 nm signal.

Fig. 4.

OCTA-guided regional segmentation and dual-excitation mt-Keima fluorescence imaging of vehicle-treated mice (24 h). (A1) En face OCTA image of the imaged retinal region. (A2–A3) Confocal fluorescence images acquired under 445 nm and 561 nm excitation, respectively. (B1) OCTA-derived vascular mask and (B2–B3) corresponding confocal fluorescence images under 445 nm and 561 nm excitation, restricted to vascular regions. (C1) Inverse OCTA-derived mask defining avascular parenchyma and (C2–C3) corresponding confocal fluorescence images under 445 nm and 561 nm excitation.

Figure 5 shows representative in vivo imaging of mt-Keima mice that were treated with AICAR. Intravitreal injection of AICAR resulted in a detectable increase in excitation-dependent mt-Keima fluorescence in both vascular and avascular regions of the retina. Dual-excitation imaging demonstrated elevated signal under 561 nm excitation across the imaged retinal field (Fig. 5(A3)) relative to the corresponding 445 nm excitation (Fig. 5(A2)). OCTA-guided segmentation revealed increased 561 nm excitation signal within vascular regions (Fig. 5(B3)) and avascular parenchyma (Fig. 5(C3)), with 445 nm excitation remaining detectable in the same compartments (Fig. 5(B2), 5(C2)). Compared with control conditions, the relative contribution of the 561 nm excitation component was significantly increased in both vascular and avascular regions.

Fig. 5.

OCTA-guided regional segmentation and dual-excitation mt-Keima fluorescence in AICAR-treated mice (24 h). (A1) En face OCT angiography (OCTA) image of the imaged retinal region. (A2–A3) Confocal fluorescence images acquired under 445 nm and 561 nm excitation, respectively. (B1) OCTA-derived vascular mask and (B2–B3) corresponding confocal fluorescence images under 445 nm and 561 nm excitation, restricted to vascular regions. (C1) Inverse OCTA-derived mask defining avascular parenchyma and (C2–C3) corresponding confocal fluorescence images under 445 nm and 561 nm excitation.

Quantitative assessment of excitation-dependent mt-Keima fluorescence was performed by calculating the 561 nm / 445 nm fluorescence intensity ratio across experimental conditions (Fig. 6). Data are presented using boxplots showing the median, interquartile range, and individual paired measurements (n = 7). As shown in Fig. 6(A), fluorescence ratios increased progressively from basal to PBS-treated and AICAR-treated conditions. Median fluorescence ratios increased from 0.552 (IQR: 0.423–0.557) under basal conditions to 0.700 (IQR: 0.650–0.784) following PBS treatment and 2.536 (IQR: 2.489–2.704) under AICAR-treated conditions. A significant overall difference across conditions was observed (Friedman test), with pairwise comparisons confirming increases between groups (Wilcoxon signed-rank test).

Fig. 6.

Quantitative analysis of excitation-dependent mt-Keima fluorescence ratios (n = 7 samples per condition). (A) Distribution of 561 nm / 445 nm fluorescence intensity ratio measured over the entire imaged retinal region under basal, PBS-treated, and AICAR-treated conditions. Boxplots represent the interquartile range (25–75%) with the median indicated by the central line; whiskers extend to 1.5× the interquartile range, and individual data points correspond to paired measurements from the same animals. Statistical comparisons between conditions were performed using the Wilcoxon signed-rank test, and overall differences across conditions were evaluated using the Friedman test. (B) Comparison of the 561 nm / 445 nm fluorescence intensity ratio between OCTA-defined vascular and avascular regions for each experimental condition. Each pair represents measurements from the same eye. Statistical comparisons were performed using the Wilcoxon signed-rank test.

Compartmental analysis was performed to compare excitation-dependent fluorescence ratios between OCTA-defined vascular and avascular retinal regions (Fig. 6(B)). Under basal conditions, higher 561 nm / 445 nm ratios were observed in vascular regions (median: 0.724; IQR: 0.604–0.785) compared with avascular parenchymal regions (median: 0.433; IQR: 0.405–0.480). A similar distribution pattern was observed under PBS-treated conditions, with vascular regions exhibiting higher ratios (median: 1.167; IQR: 1.156–1.248) relative to avascular regions (median: 0.772; IQR: 0.651–0.979). Following AICAR administration, fluorescence ratios increased in both compartments, with vascular regions reaching (median: 3.046; IQR: 2.706–3.259) and avascular regions reaching (median: 2.208; IQR: 2.057–2.288). In all conditions, fluorescence ratios were consistently higher in vascular regions than in avascular regions.

To verify that AICAR stimulated mitophagy in our experimental setting, we assessed its effect on the expression of mitophagy-related genes. As expected, AICAR increased the expression of mitophagy-related genes (Pink1 and Opa1) (Fig. 7). In addition, the mitochondrial biogenesis gene Tfam increased along with the genes indicative of mitochondrial mass (Cox4, and Tomm20). The results of the traditional molecular approach (Fig. 7) aligns with our newly developed imaging-based approach to detect and quantify mitophagy (Fig. 6).

Fig. 7.

AICAR stimulated the expression of genes associated with mitophagy and mitochondrial biogenesis in the retinal vasculature. Retinal vessels were isolated after 3 h treatment with DMSO (control) or AICAR (1 mM, 37 °C). RNA isolated from retinal vessels was analyzed by qRT-PCR using gene-specific primers for the indicated targets. Actb served as the housekeeping gene for normalization. Relative expression is defined as the ratio of the gene of interest to the housekeeping gene, comparing AICAR-treated samples (n = 3) to controls. Statistical significance between control and AICAR groups was assessed using an unpaired Student’s t-test, *p < 0.05, **p < 0.01, and ***p < 0.001.

4. Discussion

This study establishes spatially resolved in vivo ratiometric imaging of mitochondrial quality control in the retina through integration of dual-wavelength confocal mt-Keima detection with co-registered OCT and OCTA. By combining excitation-dependent fluorescence quantification with vascularly guided segmentation, the system enables compartment-specific assessment of mitochondrial turnover within intact animals.

Mitophagy is a fundamental mitochondrial quality control process that maintains bioenergetic homeostasis by selectively removing damaged mitochondria through lysosomal degradation [3,12,20]. In metabolically demanding tissues such as the retina, where photoreceptors and inner retinal neurons require sustained ATP production and redox stability [1,4], precise regulation of mitochondrial turnover is essential. Although mt-Keima has enabled sensitive detection of mitophagy in cultured cells and ex vivo preparations [6,12], in vivo retinal application is complicated by optical scattering, tissue heterogeneity, and vascular confounds [21,22]. By incorporating OCTA-derived segmentation, the present system provides spatially constrained ratiometric analysis of mt-Keima fluorescence within defined vascular and avascular retinal regions, establishing excitation-dependent fluorescence as an in vivo functional readout of mitochondrial quality control.

Although mt-Keima exhibits distinct excitation peaks under neutral and acidic conditions, the spectral separation is not absolute. Neutral-state mt-Keima retains residual excitation at longer wavelengths, including near 561 nm, such that the 561 nm channel includes a baseline contribution from mitochondria outside acidic compartments. Accordingly, fluorescence acquired under 561 nm excitation should be interpreted as acidic-enriched rather than acid-exclusive. Under basal conditions, this overlap-derived background is expected to remain comparatively weak because the dominant neutral mitochondrial population is more efficiently excited near 445 nm. Therefore, the relatively low 561 nm signal observed under basal conditions (Fig. 3) is consistent with the known excitation properties of mt-Keima. The pronounced increase in 561 nm fluorescence and in the 561 nm / 445 nm ratio following AICAR treatment exceeds this baseline overlap and is consistent with increased mitochondrial localization to acidic compartments [6].

Compartmental analysis revealed reproducible differences in excitation-dependent fluorescence ratios between vascular and avascular regions across experimental conditions (Fig. 6). Under basal and vehicle-treated conditions, fluorescence ratios were low and relatively stable, consistent with expected basal mitochondrial turnover levels [3,5,20]. Following AICAR administration, a marked increase in the 561 nm / 445 nm fluorescence ratio was observed in both compartments, with a greater relative elevation within vascular regions. Although the cellular basis of this compartmental difference cannot be resolved by the current imaging configuration, these findings demonstrate that OCTA-guided segmentation enables detection of spatial heterogeneity in mitochondrial turnover in vivo, an aspect not accessible through conventional fluorescence imaging alone.

The imaging results were supported by independent molecular analysis of isolated retinal vasculature. AICAR treatment increased expression of the mitophagy regulator Pink1 and was associated with upregulation of genes related to mitochondrial dynamics and biogenesis, including Opa1, Tfam, Cox4, and Tomm20 (Fig. 7). These transcriptional changes are consistent with coordinated activation of mitochondrial quality control pathways [15,16], providing molecular validation that the excitation-dependent mt-Keima shifts detected in vivo correspond to biologically meaningful modulation of mitochondrial turnover. The concordance between ratiometric imaging and gene expression analysis strengthens the interpretation that the hybrid confocal–OCT platform detects genuine changes in mitochondrial quality control rather than nonspecific fluorescence variability.

A key advantage of this framework is its non-invasive, longitudinal compatibility. Unlike ex vivo or histological approaches, the integrated system enables repeated measurements within the same subject under identical physiological conditions, preserving vascular architecture and tissue integrity. Co-registered OCT and OCTA provide intrinsic structural and vascular context that is inherently compatible with future depth-resolved and disease-oriented extensions. This longitudinal capability is particularly relevant for retinal metabolic and neurovascular diseases, where mitochondrial dysfunction evolves progressively and spatial heterogeneity may influence disease trajectory.

Several limitations warrant consideration for future investigations. Ratiometric detection currently relies on mt-Keima reporter expression, restricting direct application to mt-Keima–expressing models. One potential solution is to correlate ratiometric mt-Keima changes with intrinsic OCT and OCTA signatures to enable translation to animal models and human subjects without the reporter. Furthermore, due to limited axial resolution, confocal acquisition does not provide sufficient resolution to characterize layer-specific mitochondrial activity. Future work may incorporate two-photon fluorescence to achieve depth-resolved detection.

In summary, this study establishes a spatially resolved, non-invasive approach for in vivo ratiometric assessment of retinal mitochondrial quality control. By integrating molecular fluorescence detection with OCT- and OCTA-derived structural context, the platform enables compartment-specific quantification of excitation-dependent mitochondrial dynamics within intact animals. This multimodal framework provides a quantitative imaging foundation for future investigations of mitochondrial dysfunction in retinal metabolic and neurovascular diseases and for evaluating, in vivo, therapeutic modulation of mitochondrial quality control.

5. Conclusion

This study establishes spatially resolved, non-invasive in vivo ratiometric imaging of retinal mitochondrial quality control through integration of dual-wavelength confocal mt-Keima detection with co-registered OCT and OCTA. The framework enables compartment-specific quantification of excitation-dependent mitochondrial turnover within vascularly and avascularly segmented retinal regions and demonstrates sensitivity to pharmacologic modulation of mitophagy. Concordant molecular validation further supports the biological relevance of excitation-dependent fluorescence shifts detected in vivo. By unifying molecular fluorescence contrast with structural and vascular context in intact animals, this multimodal approach provides a quantitative platform for investigating mitochondrial dynamics in retinal metabolic and neurovascular disease.

Funding

U.S. National Science Foundation https://ror.org/021nxhr62 ( CBET-2513952); National Eye Institute https://ror.org/03wkg3b53 ( R01 EY023522, R01 EY029673, R01 EY030101, R01 EY030842, P30 EY001792); Research to Prevent Blindness https://ror.org/04drjs621; Richard and Loan Hill Endowment; University of Illinois College of Medicine ( Thematic Project Grant Development Program).

Disclosures

The authors declare no conflicts of interest.

Data availability

All data supporting the findings and conclusions presented in this manuscript are included in the manuscript. Additional data may be made available upon reasonable request to the authors.