All animal experimental procedures and protocols were carried out according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The protocol was approved by the Ethics Committee on the Use of Animals of the State University of Campinas (Protocols 6131-1/2022; 6131-1(A)/2023; 6392-1/2024) and followed the Ethical Principles in Animal Experimentation validated by the Brazilian College of Animal Experimentation. Adult female C57BL/6J and transgenic Townes HbSS mice (JAX Strain #:013071) 8 months old, were kept under controlled temperature conditions (22–25 °C), relative humidity (55%), 12-h photo period and free access to water and standard diet. The number of animals used is described in the figure legend of each correspondent result.

The electroretinography (ERG) was performed in both experimental groups, C57BL/6J and Townes HbSS, using the Ganzfeld stimulator (Q450SC, Roland Consult) coupled to the RetiPort System (Roland Consult, Brandenburg, Germany). The technique detects variations in electrical potential between the cornea and the retina in response to light stimuli of different intensities. Prior to binocular ERG recordings, the animals were dark-adapted for at least 12 h and then anesthetized with an intraperitoneal (i.p.) injection of ketamine hydrochloride (80 mg/kg) and xylazine hydrochloride (10 mg/kg). The pupils were fully dilated with 1% tropicamide (Mydriacyl®; Alcon, São Paulo, Brazil) and 2% methylcellulose (Ophthalmos®; São Paulo, Brazil) was used to prevent corneal injury. Goldring electrodes (Ø 1 mm; Roland Consult, Brandenburg, Germany) were positioned on the corneas (active) and needle electrodes (Roland Consult, Brandenburg, Germany) placed subcutaneously medial to the two ears (reference) and at the base of the tail (ground electrode). All signals were amplified 100,000 times, filtered with a bandpass filter between 1 and 300 Hz, and digitized at a rate of 512 Hz. During recordings, mouse body temperature was maintained at 38 °C using a water-heated platform. All preparation and electrode placement was conducted under dim red light.

The ERG protocol was an extended version of the ISCEV protocol (International Society for Clinical Electrophysiology of Vision). Protocols and procedures have been described in detail elsewhere [10–12]. Briefly, under anesthesia, dark-adapted responses (rod and mixed rod-cone response) were recorded to flashes of −3.8, −2.8, −1.8, −0.8, and 0.2 log cd·s/m² (white light) strengths on a dark background. The number of repeats (sweeps) decreased with increasing flash strength (12, 10, 8, 8, and 4, respectively). The inter-stimulus interval (1, 2, 5, 10, and 20 s) and the interval between conditions (10–120 s) were increased accordingly to maintain the dark-adapted state. The light-adapted response (cone response) was recorded using flashes of 0.3 log cd·s/m² on a background of 25 cd/m² (white light) after 2 min of light adaptation. Twenty flashes were averaged with 1 s inter-stimulus interval.

ERG components from both eyes were analyzed by detecting the troughs and peaks of the a- and b-waves using the MATLAB program routines (The MathWorks Inc., Natick, Massachusetts, USA), written and kindly provided by Professor Jan Kremers (University Hospital Erlangen, Erlangen, Germany) and with Excel spreadsheets (Microsoft Office 2010, © Microsoft Corporation, Redmond, WA, USA). The oscillatory potentials (OPs) were isolated and removed from the scotopic recordings by a variable filter method [13]. ERGs without OPs were used to measure a- and b-wave parameters. Isolated OPs were also analyzed. The amplitude of the scotopic a-wave was defined as the difference in μV between the baseline (average of recordings 16 ms before the occurrence of the flash) and the minimum within 50 ms after stimulus onset. The amplitude of the scotopic and photopic b-wave was the difference in μV between a-wave trough and the peak of the b-wave. The implicit times corresponded to the intervals between the stimulus onset and the peaks of the a-wave and of the b-wave. Abnormal responses and aberrant signals were excluded after inspection of the raw signal from each animal. The photopic a-waves presented very small amplitudes and were therefore not considered in the analysis.

Next, euthanasia was performed via intraperitoneal injection of ketamine hydrochloride (240 mg/kg) and xylazine hydrochloride (30 mg/kg), followed by enucleation of both ocular globes. The retinas of the left ocular globes were isolated for molecular analyses, and the right ocular globes were subjected to morphological analysis.

Blood collection was performed after euthanasia using EDTA tubes (MiniCollect® - EDTA K3, 1mL, Vacuette). The hematological parameters included red blood cells (RBC), hemoglobin (HB), hematocrit (HCT), white blood cell (WBC), platelets (PLT) and reticulocytes using Sysmex XN-Series XN™ (Roche, XN). The metabolic evaluation consisted in lactate dehydrogenase (LDH) and bilirubin quantification in the AU5800 Series™ (Beckman-Coulter) equipment.

The ocular globe was fixed in Davidson’s solution (2% formaldehyde, 35% ethanol, 10% acetic acid diluted in water) for 24 h. The tissue was then dehydrated, and 5 µm sections were produced in a rotating microtome (Leica, RM2125, Germany) and collected on silanized slides. The slides were stained with hematoxylin and eosin (HE) for general morphology.

A pool of 2 retinas of different animals per sample were dissected with lysis buffer (Urea 8M, Tris-HCl 1M pH 7.2, PMSF 100nM, dithiothreitol (DTT) 65 mM and cOmplete™ Mini (Merck # 11836153001) protease inhibitor cocktail) in a PowerLyzer 24 homogenizer (Qiagen). Samples were centrifuged (4 °C, 10,000 rpm for 20 min) and the supernatant was collected. Total protein concentration was determined by Bradford protein assay [14].

For protein digestion, samples were treated with 8M urea (U0631, Sigma-Aldrich), followed by protein reduction with DTT (D0632, Sigma-Aldrich), 5 mM for 25 min at 56 °C, and alkylation with iodoacetamide (IAA) (I6125, Sigma-Aldrich), 14 mM for 30 min at room temperature in the dark. For protein digestion, urea was diluted to a final concentration of 1.6 M with 50 mM ammonium bicarbonate (A6141, Sigma-Aldrich), and 1 mM of calcium chloride (C4901, Sigma-Aldrich) was added to the samples. For trypsin digestion, 2 ng/uL of trypsin (Sequencing Grade Modified Trypsin—V5111—Promega) was added. The samples were incubated for 16 h at 37 °C. The reaction was quenched with 1% formic acid (100264, Merck), and peptides were desalted with C18 stage tips (Octadecyl C18-bonded sílica—3 M Empore™ extraction disks), dried in a vacuum concentrator, reconstituted in 0.1% formic acid (100264, Merck), and stored at −20 °C for subsequent analysis by LC-MS/MS.

An aliquot of each pool of samples was loaded onto an LTQ Velos Orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) connected to a nanoflow LC (nLC-MS/MS) by an EASY-nLC system (Proxeon Biosystems, West Palm Beach, FL, USA) through a Proxeon nanoelectrospray ion source. The peptides were separated by a 2%–90% acetonitrile (ACN) gradient in 0.1% formic acid using an analytical column PicoFrit Column (20 cm × ID75 mm, 5 mm particle size, New Objective, Woburn, MA) at a flow of 300 nL/min over 212 min. The nanoelectrospray voltage was set to 1.7 kV, and the source temperature was 275 °C. All instrument methods for the LTQ-Orbitrap Velos were set up in the data-dependent acquisition mode. Full-scan MS spectra (m/z 300–1,600) were acquired by the Orbitrap analyzer after accumulation to a target value of 1 × 10^-6. The resolution of the Orbitrap was set to r = 60,000, and the 20 most intense peptide ions with charge states ≥2 were sequentially isolated to a target value of 1,000 and fragmented in the linear ion trap by low-energy collision-induced dissociation (CID) as the dissociation or fragmentation method, with a normalized collision energy of 35%. The signal threshold for triggering an MS/MS event was set to 500 counts. Dynamic exclusion was enabled with an exclusion size list of 500, an exclusion duration of 30 s and a repeat count of 1. An activation q = 0.25 and an activation time of 10 ms were used.

LC-MS/MS runs were conducted for each region raw data using MaxQuant v1.3.0.3 software. MS/MS spectra were searched against The Mouse UniProt database using the Andromeda search engine. As search parameters, a tolerance of 10 ppm was considered for precursor ions (MS search) and 1.0 Da for-fragment ions (MS/MS search), with a maximum of two missed cleavages. Carbamidomethylation of cysteine was considered a fixed modification, and oxidation of methionine and protein N-terminal acetylation were considered variable modifications. A maximum of a 1% false discovery rate (FDR) was set for both the protein and peptide identification. Protein quantification was performed using the LFQ algorithm implemented in MaxQuant software with a minimal ratio count of 1 and a window of 2 min for matching between runs. The global normalization was used in the MaxQuant LFQ algorithm for label-free quantification (LFQ) [15]. Statistical analysis was performed with Perseus v1.6.7.0 software, which is available in the MaxQuant package. Identified protein entries were processed, excluding reverse sequences, “only by site” entries, and contaminants. Raw LFQ values were log transformed, and the mean of normalized intensities were compared between groups. Supplementary Table S6 lists all proteins along with their corresponding FDR and p-values. Since no protein reached significance under the FDR cutoff, proteins with p < 0.05 were selected for subsequent analysis, allowing an exploratory overview of biologically meaningful candidates. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository [16] with the dataset identifier PXD082336.

To better understand the biological relevance of the differentially expressed proteins (DEPs), the complete list was separated into upregulated (Log2Fold Change >0) and downregulated Log2Fold Change <0) for functional enrichment and protein-protein interaction (PPI) network construction. KEGG molecular pathway enrichment was performed using ShinyGO (p < 0.05) [17]. Next, String DB database [18] (confidence interval = 0.7) was used to construct the PPI network for DEPs present in the most relevant enriched pathways.

To further evaluate the molecular expression profile of relevant DEPs, RT-qPCR was performed. Total RNA was extracted from the isolated retinas with the RNeasy® Mini kit (Qiagen, 74104, Germany), followed by purification with DNAse I Amplification Grade (Invitrogen, USA) to remove remaining genomic DNA. The quantity and purity of the extracted RNA were measured using NanoDrop2000 spectrophotometer (Thermo Scientific). Two micrograms of RNA were reverse transcribed with the High-Capacity Kit RNA-to-cDNA (Life Technologies, 4387406, Carlsbad, CA, USA). All the protocols followed the manufacturer’s instructions.

Next, the RT-qPCR reaction was performed using the SYBR Green PCR Master Mix® (Applied Biosystems, Foster City, CA, USA) detection method in an ABI StepOne Plus equipment (Applied Biosystems, Foster City, CA, USA). Specific primers were used for Slc24a1 and Cnga1 genes (Table 1). Briefly, the PCR amplification was carried out in a 10 μL reaction volume containing 2 μL of cDNA (1:10 dilution) and 8 μL of reaction mix, containing 200 or 400 nM of specific primers diluted in injection water and 3.15 μL of SYBR Green PCR Master Mix. All samples were run in duplicate with the following cycling conditions: 1 cycle of 95 °C for 10 min, 40 cycles of 95 °C for 15s, and 40 cycles of 60 °C for 1 min. Relative quantification of each gene was performed using the 2^-ΔΔCT method, according to Livak et al. [19]. The Ct values obtained for each sample were normalized by the ratio of the target gene and the beta-actin (Actb) and glyceraldehyde-3-phosphate dehydrogenase (Gapdh) reference genes mean value. Results were obtained by the expression ratio of the Townes HbSS/C57BL/6J groups.

Statistical analysis was performed in GraphPad Prism® (version 10.00, GraphPad, Inc., San Diego, CA). The Shapiro-Wilk normality test was applied in all data. Two-way ANOVA was used for the ERG scotopic data, and Mann–Whitney test for the ERG photopic data. RT-qPCR, hematological and metabolic parameters were compared by Student’s t-test. All values were considered statistically significant when p < 0.05.

To confirm the SCA phenotype, hematological parameters were evaluated in Townes HbSS and C57BL/6J. The Townes HbSS mice showed hematological features consistent with disease, including a reduction of 58.23% in erythrocyte count (5.1 ± 0.9 and 8.8 ± 0.9 10⁶/uL); 52.10% in hematocrit (24.4 ± 4.1 and 46.9 ± 5.3%) and 46.49% in hemoglobin concentration (5.9 ± 1 and 12.6 ± 1.3 g/dL), in addition to expressive raise of 998% in reticulocyte count (41 ± 6.7 and 4.1 ± 1.1%) and of 702.33% in white blood cell count (30.5 ± 7.8 and 4.3 ± 2.1 10³/uL), all comparisons with p < 0.0001 (Figure 1).

Regarding the evaluation of hemolysis markers, Townes HbSS group showed a significant increase of 852.83% in total bilirubin (1.4 ± 0.3 and 0.2 ± 0.02 mg/dL); 1,095.75% in direct (0.4 ± 0.1 and 0.03 ± 0.00 mg/dL) and 784.43% in indirect bilirubin (0.1 ± 0.2 and 1 ± 0.02 mg/dL), all comparisons with p < 0.0001, with no statistically significant difference in LDH levels between groups (Figure 2).

Retinal responses to light stimuli were evaluated by electroretinography (ERG) in all animals. Under scotopic conditions (Figure 3A), for the a-wave, referring to photoreceptor activity, the Townes HbSS mice showed smaller amplitude (153 ± 59.4 µV) when compared to the controls (182.8 ± 62.8 µV), presenting a reduced response to the stimulus at the highest light intensity (0.3 log cd·s/m²) (p = 0.016) (Figure 3B). A significant reduction was also observed in Townes HbSS implicit time in −0.7 log cd·s/m² light intensity (20.1 ± 1.20 ms) compared to controls (20.82 ± 1.33 ms) (p = 0.0095) (Figure 3D). No statistically significant differences were observed between the experimental groups in either b-wave amplitude (Figure 3C) or implicit time (Figure 3E), which are mainly attributable to bipolar cell activity. Finally, compared to C57BL/6J (23.32 ± 7.18 µV), Townes HbSS mice (20.34 ± 7.49 µV) also exhibited reduced amplitude of OPs at maximum light intensity (0.3 log cd·s/m²) (p = 0.0274), (Figures 3F,G), indicating an early impairment of the inner retina.

For the photopic response, qualitative evaluation of the averaged responses of both groups shown in Figure 4A indicates that Townes HbSS showed a reduced amplitude of the photopic negative response (PhNR), possibly related to innermost retina alterations, specifically the ganglion cells layer. There was no significant difference in the median b-wave amplitude between Townes HbSS (69.69, 49.87-95.70 μV) and C57BL/6J (73.04, 50.74-97.07 μV) (Figure 4B). However, Townes HbSS (38.24, 36.63-39.49 ms) showed a higher implicit time of the b-wave compared to controls (37.21, 35.67 – 38.10 ms) (p = 0.0096), as shown in Figure 4C, indicating a delayed response in bipolar cells.

In the HE staining, qualitative assessment of retinal morphology indicated disruptions in the outer nuclear layer (ONL) in C57BL/6J and Townes HbSS mice (Figure 5). Discrete alterations are observed in the inner nuclear layer (INL) in Townes HbSS sections (Figure 5B), findings that may be related to retinal degeneration. In addition, blood vessels were observed in the ganglion cell layer (GCL) in Townes HbSS compared to controls, as illustrated by the black arrow in Figure 5B.

Comparative proteomic profiling was performed to compare the retinal protein expression patterns between Townes HbSS mice and control animals. The mice retina proteome encompassed 2,248 proteins, of which 163 were exclusively identified in controls (C57BL/6J) and 43 in Townes HbSS. Although no protein remained statistically significant after FDR correction, we selected 268 differentially expressed proteins (DEPs) as candidates (nominal p-value <0.05): 63 upregulated and 205 downregulated in sickle cell mice compared with controls (Supplementary Material 1). The heatmap illustrates the expression profile of DEPs in each sample (Figure 6A). Clustering of experimental groups is shown in the Principal Component Analysis (PCA) graph (Figure 6B). Finally, the distribution of proteins according to the up or downregulation profile is depicted in Volcano plot (Figure 6C).

Functional enrichment analysis was conducted to identify molecular pathways associated with the DEPs in Townes HbSS retinas. Enrichment analysis of downregulated proteins identified 76 molecular pathways, mainly related to phototransduction, energetic metabolism (citric acid cycle and oxidative phosphorylation), and neurodegenerative processes (Figure 7A). For the upregulated proteins, 21 molecular pathways are associated with neurodegenerative processes, glycolysis/gluconeogenesis, and HIF-1 (Hypoxia-Inducible Factor 1) signaling pathway, as shown in Figure 7B.

The following analysis was based on pathways related to downregulated DEPs. Protein–protein interaction networks were constructed based on proteins present in each biologically relevant molecular pathway. The pathways selected from the downregulated DEPs were: Phototransduction (Cnga1, Pde6b, Pde6g, Slc24a1, Pde6a, Grk1); Citrate cycle (Aco2, Idh3b, Pcx, Sucla2, Ogdhl); and Oxidative phosphorylation (Ndufb11, Atp5a1, Atp6v1e1, Cox6c, Atp5o, Atp6v1g2, Atp6v1g1, Ndufa7, Uqcrfs1, Uqcrc2) (Figures 8A,B). Additionally, to obtain a more comprehensive view of the molecular profiles of selected targets, differential gene expression analysis was performed. We selected the Slc24a1 and Cnga1 genes due to their biological function and relevance in the phototransduction cascade. Both genes are downregulated in Townes HbSS retinas compared to C57BL/6J, in accordance with the protein expression profile observed in the proteomic analysis (Figure 8C).