Impact of heat-induced retrieval methods on silver staining for detecting senile plaques of Alzheimer’s disease in human brain tissue sections

Abstract

Background

Silver staining continues to serve as a critical technique for histological detection of pathological deposits, especially in the context of Alzheimer’s disease (AD). In clinical research, paraffin-embedded sections are often stored for extended periods, and tissues may remain in formaldehyde for prolonged durations, both of which can lead to false-negative results. The objective of this study was to systematically evaluate whether heat-induced retrieval can enhance the efficacy and sensitivity of classical silver staining in these suboptimally preserved brain tissues.

Methods

This study investigated the impact of heat-induced retrieval methods on silver staining efficacy in brain tissue. Three types of samples were examined: standard FFPE sections, tissues immersed in formaldehyde for 48 months, and paraffin‑embedded sections (initially fixed for 48 h in formalin, sections coated with paraffin and stored on slides) stored for seven years. Retrieval methods were performed using citrate-citric acid buffer, Tris-EDTA buffer, or proteinase K. Silver staining was conducted using three classical methods: Bielschowsky, periodic acid-methenamine silver (PAM), and Gallyas; untreated sections were used as controls.

Results

Results showed that both citrate-citric acid and Tris-EDTA significantly increased the signal-to-noise ratio of senile plaques in Bielschowsky staining. This improvement facilitated better the detection of amyloid plaque details and density, with Tris-EDTA yielding higher sensitivity in long-stored paraffin sections. Proteinase K digestion did not enhance Bielschowsky staining results. None of the retrieval methods improved signal-to-noise ratio for senile plaques or neurofibrillary tangles with PAM or Gallyas.

Conclusions

These findings indicate that heat-induced retrieval methods with citrate or Tris-EDTA buffer, enabling more efficient silver ion reduction at specific targets in Bielschowsky. This work provides an experimental basis for optimizing silver staining protocols, which helps enhance diagnostic accuracy in neuropathological studies of AD and related disorders.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13000-026-01799-3.

Introduction

In modern neuropathology, accurate diagnosis and mechanistic insight into neurodegenerative diseases depend on the ability to recognize and characterize pathological changes at both cellular and tissue levels [1]. A wide range of staining techniques is routinely used in clinical practice, providing essential morphological and molecular information for evaluating neuropathological alterations [2]. These methods are crucial for visualizing hallmark proteinopathies in Alzheimer’s disease (AD) and other neurodegenerative disorders [3, 4].

Among these, silver impregnation techniques rank among the most historically significant and widely used tools for this purpose. The Bielschowsky, periodic acid-silver methenamine (PAM), and Gallyas methods are particularly valuable for detecting pathological aggregates, including senile plaques (SPs), neurofibrillary tangles (NFTs), and Pick bodies [5–7]. The high contrast and detailed morphological resolution have made them a cornerstone of neuropathological practice, supporting precise staging of disease progression, exemplified by the Braak staging system for AD pathology [8].

However, classical silver staining can suffer from variability in staining intensity, background artifacts, and reduced sensitivity, particularly in suboptimally preserved tissues. Numerous modifications have been proposed to improve diagnostic performance [9], yet important limitations remain unresolved. Prolonged paraffin storage or extended formalin fixation in clinical and research samples may reduce assay sensitivity and lead to false-negative results [10–14], warranting systematic further investigation.

Formalin fixation and paraffin embedding (FFPE) has been the gold standard for tissue preservation in pathology for over a century, offering excellent morphological detail and enabling long-term archival storage. This process, while effective for morphology, induces methylene bridge formation between protein amino groups, producing extensive cross-links that can mask epitopes and alter protein tertiary structure [15] and mask target sites among proteins and other molecules within the tissue [13]. Consequently, both extended formalin fixation and the prolonged storage of paraffin sections have been linked to a progressive decline in biomolecule detectability and epitope immunoreactivity [10, 12]. In addition, prolonged storage of paraffin sections or extended formalin fixation has been linked to progressive loss of biomolecule detectability, likely due to subtle chemical modification over time [12]. While formalin induces covalent cross-linking, paraffin wax itself is chemically inert and does not cross-link tissue molecules; instead, the mandatory embedding steps (alcohol dehydration, organic solvent clearing, and high-temperature infiltration) severely impact brain sections by inducing aggressive lipid extraction and non-covalent protein precipitation [16–18], which collapses the lipid-rich architecture and sterically masks histochemical targets. Previous studies have shown that prolonged storage of paraffin sections leads to loss of epitope immunoreactivity [10, 13] and other molecules in tissues. This manifests as diminished staining intensity and increased background noise, potentially leading to diagnostic inaccuracies or underestimation of pathological burden, thereby hampering downstream analyses. Since its introduction by Shi et al. [19], epitope retrieval has transformed histochemical staining and become a standard pretreatment in immunohistochemistry for FFPE tissues [20]. It is hypothesized that heat- or enzyme-based treatment in a suitable buffer renatures protein conformations and unmasks hidden epitopes, likely by breaking methylene bonds formed during fixation [21], thereby enhancing the detection of numerous cellular proteins [22]. Silver staining operates by reducing silver ions bound to target structures into metallic silver particles [23, 24], generating visible signals under microscopic examination. When epitopes or chemical groups required for silver ion binding are masked by formalin cross-links or degraded during storage, binding efficiency declines, weakening subsequent reduction reactions and compromising staining quality.

Despite its extensive application in immunohistochemistry, the interaction between retrieval methods and the complex chemistry of different silver staining methods remains poorly understood. Moreover, the effects of retrieval may vary across techniques, as Gallyas, Bielschowsky, and PAM methods employ distinct chemical reactions and display differing sensitivities [7]. Bielschowsky staining primarily relies on the binding of silver ions to specific amino acid side chains (e.g., histidine, tyrosine) in proteins [25]. Conversely, PAM staining depends on the oxidation of carbohydrates to aldehydes [23], and the Gallyas method is a physical development process driven by nucleation [26]; both mechanisms are less reliant on native protein conformations. Although these silver staining methods are strictly chemical processes, this fundamental divergence in their targets led us to hypothesize that heat-induced retrieval could improve silver ion accessibility in the protein-dependent Bielschowsky method, while remaining ineffective for PAM and Gallyas staining. Therefore, this study aims to determine whether retrieval principles can be effectively applied to metallic impregnation histochemistry. To this end, we systematically evaluated the influence of three different retrieval protocols on three classical silver staining methods under controlled conditions with varying formaldehyde fixation times and paraffin storage durations. The overarching goal was to establish whether retrieval methods could mitigate the adverse effects of compromised tissue preservation on silver staining, thereby improving the reliability and diagnostic value of these essential techniques in further neuropathology.

Materials and methods

Tissue preparation

Temporal lobe tissue obtained from nine autopsied brains, fixed by immersion in 10% formalin, was used for this study. The autopsy brain tissue was obtained from the National Health and Disease Human Brain Tissue Resource Center (Anhui Medical University, China). Permission was obtained for the use of the tissue and clinical information for research purposes. The study protocols involving human brain tissue samples were approved by the Ethics Committee of Anhui Medical University (No. 83244549). Neuropathological examination confirmed that all nine cases met the conventional criteria for a diagnosis of fully developed AD. All samples were assessed by a neuropathologist for Braak staging, and staging was completed for every case included in the study. Human brain tissue samples were obtained from nine different donors diagnosed with Alzheimer’s disease (Braak stage 5), comprising five males and four females (age range: 79–95 years; post-mortem delay: 230–490 min, Supplementary Table 1). The study employed an independent group design, with the nine cases divided into three experimental groups (n = 3 per group) according to their fixation and storage history: (1) Standard FFPE group: Tissues fixed in 10% formalin for 48 h and then embedded in paraffin. This group serves as the contemporary control against the suboptimally preserved groups. (2) Long-term Formalin group: Authentic archival tissues that had been fixed in 10% formaldehyde for 48 months prior to processing. (3) Long-term Stored Section group: Archival paraffin sections (initially fixed for 48 h in formalin, sections coated with paraffin and stored on slides) that had been stored in a dedicated slide storage cabinet at room temperature (20–25 °C) under dry conditions for seven years. All long-term conditions were derived from real-world archival cases and were not laboratory-simulated. Prior to fixation and processing, all brain tissues were meticulously sectioned and trimmed to fit within standard embedding cassettes, which have internal dimensions of 40 mm × 30 mm × 7 mm. Paraffin-embedded brain tissue blocks were sectioned into adjacent 5 μm thick slices. All sections were processed by deparaffinized in two changes of xylene (10 min each), followed by two changes of 100% ethanol (10 min each). Sections were then treated sequentially with a graded series of ethanol solutions (95%, 90%, 80%, and 70%; 5 min each) and rinsed in distilled water for 2 min.

Retrieval methods

All reagents used in this experiment were from Macklin, unless otherwise specified. The reagents used for retrieval in this study are listed in Table 1, according to previous study [21]. After dewaxing and hydration, a subset of the deparaffinized tissue sections were stained directly without retrieval as the control group. The remaining sections underwent retrieval using one of three different methods: (1) Citrate-citric acid method: sections were immersed in Citrate-citric acid buffer (pH = 6.0) and subjected to high-pressure heat-induced retrieval for 22 min in epitope retriever (62700-20, Aptum, UK), then cooled to room temperature; (2) Tris-EDTA method: sections were immersed in Tris-EDTA buffer (pH = 9.0) and subjected to high-pressure heat-induced retrieval for 22 min in epitope retriever, then cooling; and (3) Proteinase K method: sections incubated with Proteinase K solution (20 µg/mL) for 3 min at room temperature.

Table 1.

Reagents used for retrieval

Buffer	Formulation
0.1 M Citrate buffer	- 21.01 g Citric acid monohydrate - 1 L distilled water
0.1 M Sodium citrate buffer	- 29.41 g tri-sodium citrate (dihydrate) - 1 L distilled water
Citrate-citric acid buffer, pH 6.0	- 7 mL 0.1 M Citrate buffer - 43 mL 0.1 M Sodium citrate buffer - 50 mL distilled water
1 mM EDTA buffer, pH 8.0	- 0.37 g EDTA - 1 L distilled water - Adjust pH to 8.0 using NaOH
Tris-EDTA buffer, pH 9.0	- 1.21 g Tris - 1 L of 1 mM EDTA buffer - Adjust pH to 9.0 using NaOH
20 × proteinase K solution (400 µg/mL, pH 8.0)	Dissolve 4 mg proteinase K in 10 mL of TE buffer, pH 8.0
Proteinase K (20 µg/mL working solution)	- 1 mL 20 × proteinase K solution - 19 mL TE buffer, pH 8.0

Silver staining

Bielschowsky method

Processed sections were immersed in a 20% aqueous silver nitrate (AgNO₃) solution at 37 °C for 20 min in the dark, then rinsed in distilled water for 10 min. Subsequently, sections were immersed in a 20% ammoniacal silver nitrate solution at room temperature in the dark for 20 min, then rinsed in a dilute aqueous ammonium hydroxide (NH₄OH) solution. Slides were placed in a developer solution formulated with 37–40% formaldehyde (20 mL), citric acid (0.5 g), concentrated nitric acid (HNO₃, 40 µL), and distilled water to a final volume of 100 mL. Prior to use, this developer was mixed with an equal volume of ammonium hydroxide (NH₄OH), with continuous observation of staining until completion. Specifically, the staining endpoint was defined by the clear visualization of target structures (e.g., NFTs, SPs, and neuropil threads) in dark brown or black, reaching a peak contrast against a light-yellow to brownish-yellow background. Once this predefined histological contrast was achieved, the reaction was immediately terminated to prevent over-staining. This approach ensures maximal signal-to-noise ratio (SNR) across different sample groups. After rinsing in distilled water, slides were immersed in 5% sodium thiosulfate solution for 5 min, then rinsed in distilled water for 5 min.

PAM method

This study used a commercial kit: Baso, Guangdong, China, Fungal Stain (Periodic-acid silver methenamine Method), Catalog No. BA4244B, Lot number C240301. Processed sections were immersed in 5% orthoperiodic acid at room temperature for 15 min, then rinsed in distilled water for 10 min. The slides were placed in a mixture of borax solution and methenamine silver reagent at 62 °C for 30– 60 min, with continuous observation of staining until completion. Finally, slides were immersed in 5% sodium thiosulfate solution for 5 min, and rinsed in distilled water for 5 min.

Gallyas method

Processed sections were immersed in 5% orthoperiodic acid for 30 min at room temperature, then rinsed in distilled water for 10 min. Slides were immersed in alkaline silver iodate solution (including 0.035% silver nitrate, 10% potassium iodide, 4% sodium hydroxide) at room temperature for 30 min, and rinsed in 0.5% acetic acid for 5 min. Slides were subsequently immersed in a physical developer solution (0.1% silver nitrate, 0.5% tungsto-silicic acid, 0.1% ammonium nitrate, 2.5% sodium carbonate, 0.03– 0.2% formalin), with continuous monitoring until staining was complete. Finally, slides were immersed in 5% sodium thiosulfate solution for 5 min, and rinsed in distilled water for 5 min. Detailed silver staining protocols are provided in Table 2.

Table 2.

Protocols for different silver-staining methods

	Modified Bielschowsky method for paraffin sections	Gallyas method	Periodic-acid silver methenamine method
Pretreatment	-	5% Orthoperiodic acid (RT, 30 min)	5% Orthoperiodic acid (RT, 15 min)
Washing	-	DW	DW
Silver solution	20% silver nitrate (37 °C, 20 min in the dark)	Alkaline silver iodate (RT, 30 min) 0.035% silver nitrate 10% potassium iodide 4% sodium hydroxide	Developer in mixture of Borax Solution and Methenamine Silver Reagent (62 °C,30–60 min)
Washing	DW	0.5% acetic acid 5 min
Silver solution	20%a ammoniacal silver (RT, 20 min in the dark)	Physical developerd (RT, 5–10 min) 0.1% silver nitrate 0.5% tungsto-silicic acid 0.1% ammonium nitrate 2.5% sodium carbonate 0.03–0.2% formalin
Washing	DW + ammoniab
Chemical developer	Developerc in 20% ammoniacal silver

DW distilled water, RT room temperature

^a Ammonium hydroxide was added to 20% silver nitrate drop by drop until the precipitates turn clear. Additional two drops of ammonium hydroxide were recommended

^b Add three drops of ammonium hydroxide to DW

^c Add three drops of the developer (formalin 20 mL, DW 100 mL, concentrated nitric acid 40 µL, citric acid 0.5 g) to the 20% ammoniacal silver used as silver solution II

^d In laboratory, equal amount of stock solutions A (0.5 g sodium carbonate-anhydrous,100 mL DW) and B (0.2 g silver nitrate, 1 g tungsto-silicic acid, 0.2 g ammonium nitrate, formalin 0.73 mL, 100 mL DW) are mixed immediately before use

To eliminate inter-experimental bias, all sections within each comparison group were processed on the same day. To mitigate selection bias at the operational level, the continuous serial sections obtained from each individual tissue block were randomly allocated to the subsequent experimental subgroups prior to the staining procedures.

All deparaffinized tissue sections were assigned to the following experimental groups: Group A (Control): Sections were stained without retrieval, using the Bielschowsky, PAM, or Gallyas methods. Group B: Sections were treated with one of the following retrieval methods: (a) citrate-citric acid buffer (pH 6.0), (b) Tris-EDTA buffer (pH 9.0), or (c) proteinase K solution (20 µg/mL). After retrieval, all sections in this group were stained using the Bielschowsky method. Group C: Sections underwent retrieval using (a) citrate-citric acid buffer (pH 6.0), (b) Tris-EDTA buffer (pH 9.0), or (c) proteinase K solution (20 µg/mL). Following retrieval, all sections in this group were stained using the PAM method. Group D: Sections underwent retrieval using (a) citrate-citric acid buffer (pH 6.0), (b) Tris EDTA buffer (pH 9.0), or (c) proteinase K solution (20 µg/mL). Following retrieval, all sections in this group were stained using the Gallyas method. After completion of the staining procedures, sections were rehydrated in ascending concentrations of ethanol (30%, 50%, 70%, and 96%) for 5 min each, cleared in two changes of xylene for 5 min each, and mounted with coverslips. The detailed schematic diagram of the study design is provided in Fig. 1.

Fig. 1.

Schematic representation of study design

Immunohistochemistry

To confirm that the SPs detected by the localized silver ions corresponded to Aβ deposits, immunohistochemistry for Aβ was performed on adjacent sections. For Immunohistochemistry, 5-µm sections were deparaffinized and incubated in 0.5% PBST for 1 h, then treated in Citrate-citric acid buffer for 22 min in epitope retriever. Subsequently, sections were blocked with 5% normal donkey serum (017-000-121, Jackson Immuno Research, USA) in 0.5% PBST at 37 C for 1 h. Sections were then incubated with mouse monoclonal Anti-Beta (β)-Amyloid antibody (A3981, Sigma-Aldrich, USA, 1:250) for 1 h at 37°C, followed by overnight incubation at 4°C. Subsequently, sections were incubated with biotinylated horse anti-mouse IgG (BA-2000, Vector Laboratories, USA, 1:200) and an avidin-biotin peroxidase complex (PK-4000, Vector Laboratories, USA, 1:200). Sections were developed with 0.05% 3,3’-diaminobenzidine (Sigma-Aldrich, USA) as chromogen. All sections were counterstained with hematoxylin, then rehydrated with ethanol at ascending concentrations (30, 50, 70, and 96%) for 5 min each, cleared in two changes of xylene for 5 min each, and mounted with coverslips.

Measurements and data analysis

All experiments and quantitative analyses were performed on rigorously matched adjacent sections. To ensure consistency and comparability across different treatment groups, all images were captured specifically within gray matter regions. For every subgroup (i.e., direct staining without retrieval or pre-treatment with citrate-citric acid buffer, Tris-EDTA buffer, or proteinase K solution) across the various silver staining methods, serial sections were utilized to identify and photograph identical microscopic fields. Critical imaging parameters, including exposure time, light intensity, and gain, were maintained strictly constant within each staining batch to ensure standardized quantification.

Quantification of SNR

After being mounted with coverslips, all sections were imaged using an optical microscope (BX53, Olympus, Japan) at a 20× magnification. To evaluate the impact of different retrieval protocols on pathological visualization, target structures were classified into two categories based on their staining markers: SPs and NFTs. Specifically, SPs were identified and analyzed using Bielschowsky silver staining and PAM staining, while NFTs were targeted and quantified using Gallyas silver staining. The staining quality was quantified by the SNR using ImageJ software. For each subgroup, target structures were identified within the selected microscopic fields until exactly 6–7 independent lesions (either 6–7 SPs for Bielschowsky/PAM or 6–7 NFTs for Gallyas) were sampled across each case, resulting in a standardized dataset. The SNR was determined by manually delineating the precise boundary of each pathological structure to measure its mean grayscale value, which was then divided by the mean grayscale value of its immediately adjacent background. This ratio provided a normalized index of staining contrast relative to the surrounding tissue. Additionally, we obtained 100× images of the identical senile plaque from adjacent sections processed under different conditions.

Calculation of SPs density

The density of SPs was quantified exclusively for the Bielschowsky staining method. For the assessment of SPs density, regions of interest were captured using an optical microscope (BX53, Olympus, Japan) at a 10× magnification. For each image, the average plaque density per standardized field (defined as a microscopic field of fixed area covering 1200 × 600 μm) was calculated by manually counting immunoreactive plaques within such fields. A systematically randomized sampling strategy was applied to select 10 microscopic fields per biological sample, accumulating exactly 30 independent field-density values per subgroup. The average plaque density per standardized field was calculated for each image. To properly evaluate the efficacy of different treatments, these absolute density data were subsequently normalized to the baseline values of the corresponding un-retrieved control group, generating relative density ratios.

Statistical analysis

All morphometric and intensity data were processed using ImageJ (version 1.54f) and GraphPad Prism 10. To ensure the appropriate use of parametric tests, before performing any group comparisons, the distribution of all datasets was formally assessed using the D’Agostino-Pearson omnibus normality test to evaluate both normality and lognormality. For datasets with normal distribution (p > 0.05), unpaired t-tests were used for comparisons between two groups, and one-way ANOVA was employed for comparisons involving more than two groups. For datasets that did not follow a normal distribution, non-parametric alternatives were used. Statistical significance was set at *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Results

Silver staining in standard FFPE sections

Three distinct silver staining techniques—Bielschowsky, PAM, and Gallyas—were employed to visualize neuropathological features in standard FFPE brain sections. Bielschowsky staining revealed obvious SPs and NFTs (Fig. 2a– d, a’– d’), which appeared brownish‑yellow coloring against a light‑yellow background. PAM staining highlighted numerous SPs (Fig. 2e– h, e’– h’). Gallyas staining was particularly effective in visualizing numerous NFTs, which exhibited a characteristic yellow-black coloration (Fig. 2i– l, i’– l’). Notably, heat-induced retrieval with citrate-citric acid or Tris-EDTA markedly enhanced SP staining in Bielschowsky‑stained sections (Fig. 2a– d, a’– d’). This enhancement was accompanied by reduced background staining, improving overall contrast and clarity.

Fig. 2.

The effect of retrieval methods on SNR of SPs/NFT in standard FFPE sections. Deparaffinized sections were treated with Citrate-citric acid buffer (b, f, j), Tris-EDTA buffer (c, g, k), Proteinase K solution (d, h, l). Sections were subsequently stained with Bielschowsky (a– d), PAM (e– h), Gallyas (i– l) methods. The sections stained without retrieval were shown as a control (a, e, i). Bar graphs demonstrating the SNR of SPs of Bielschowsky (m), PAM (n), Gallyas (o). Panels below (a’ – l’) show corresponding magnified views of the regions indicated by white dashed boxes. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar: 200 μm (a– l), 100 μm (a’– l’)

To systematically assess their effects on silver staining, the SNR of SPs and NFTs was calculated as the grayscale intensity of the target structure divided by that of the background. In standard FFPE sections, only citrate-citric acid and Tris-EDTA improved the SNR of SPs in Bielschowsky staining, from 1.24 ± 0.10 in the control group to 1.74 ± 0.28 and 1.80 ± 0.17, respectively, whereas proteinase K treatment had no such effect (1.24 ± 0.08) (Fig. 2m). In contrast, no retrieval method increased the SNR in PAM staining for SPs (control: 2.18 ± 0.42; citrate-citric acid: 1.95 ± 0.21; Tris-EDTA: 2.16 ± 0.27; proteinase K: 2.06 ± 0.31) or in Gallyas staining for NFTs (control: 3.08 ± 0.50; citrate-citric acid: 3.12 ± 0.41; Tris-EDTA: 3.16 ± 0.47; proteinase K: 3.34 ± 0.47) (Fig. 2n, o). All p-values in this paper are uniformly listed in Supplementary Table 2.

Silver staining in 48-months formalin-fixed tissues

In tissues immersed in formaldehyde for 48 months, Bielschowsky, PAM, and Gallyas silver staining were performed to assess neuropathological changes. Bielschowsky staining revealed abundant brownish‑yellow SPs and NFTs against a light‑yellow background (Fig. 3a– d, a’– d’). PAM staining predominantly highlighted SPs (Fig. 3e– h, e’– h’), whereas Gallyas staining optimally visualized yellow-black NFTs (Fig. 3i– l, i’– l’). In 48-month formalin-fixed tissues, citrate-citric acid (1.63 ± 0.23) and Tris-EDTA (1.47 ± 0.17) increased the SNR of SPs in Bielschowsky staining compared to the control (1.28 ± 0.13), whereas proteinase K showed no improvement (1.25 ± 0.14) (Fig. 3m). In contrast, no clear increase was observed in PAM staining for SPs (control: 2.95 ± 0.38; citrate-citric acid: 2.64 ± 0.34; Tris-EDTA: 2.85 ± 0.41; proteinase K: 2.82 ± 0.52) or in Gallyas staining for NFTs (control: 3.17 ± 0.41; citrate-citric acid: 3.14 ± 0.33; Tris-EDTA: 2.87 ± 0.36; proteinase K: 3.07 ± 0.48) (Fig. 3n, o). Thus, heat-induced retrieval with citrate-citric acid or Tris-EDTA markedly improved Bielschowsky staining, yielding darker SPs and a cleaner background, but produced no appreciable visual improvement in PAM or Gallyas staining.

Fig. 3.

The effect of retrieval methods on SNR of SPs/NFT in 48-months formalin-fixed tissues. Deparaffinized sections were treated with Citrate-citric acid buffer (b, f, j), Tris-EDTA buffer (c, g, k), Proteinase K solution (d, h, l). Sections were subsequently stained with Bielschowsky (a– d), PAM (e– h), Gallyas (i– l) methods. The sections stained without retrieval were shown as a control (a, e, i). Bar graphs demonstrating the SNR of SPs of Bielschowsky (m), PAM (n), Gallyas (o). Panels below (a’– l’) show corresponding magnified views of the regions indicated by white dashed boxes. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar, 200 μm (a– l), 100 μm (a’– l’)

In 48‑month formalin‑fixed tissues, retrieval effects paralleled those in standard FFPE samples, with only citrate-citric acid and Tris-EDTA increasing the SNR. Based on these findings, heat-induced retrieval effects in Bielschowsky staining were further analyzed. To compare SNR across standard FFPE and 48‑month formalin‑fixed groups, data were reorganized and presented (Fig. 4a– c) according to different retrieval methods. SNRs for Bielschowsky staining are shown for citrate-citric acid (Fig. 4a), Tris-EDTA (Fig. 4b), and proteinase K (Fig. 4c) to compared the effect between standard FFPE and 48-m-fixed groups. Furthermore, SNR improvement was calculated by subtracting values of unretrieved sections from those of each retrieval group. After 48 months of fixation, SNR improvement was markedly lower than standard FFPE tissues for both citrate-citric acid (48-month: 0.28 ± 0.14, standard FFPE: 0.40 ± 0.18; Fig. 4d) and Tris-EDTA (48-month: 0.15 ± 0.13, standard FFPE: 0.46 ± 0.15; Fig. 4e). Consistent with previous results, proteinase K had no significant effect on SNR in either 48-month formalin-fixed (-0.02 ± 0.09) or standard FFPE tissues (0.01 ± 0.07) (Fig. 4f).

Fig. 4.

Comparison of the retrieval effects on the SNR of SPs between standard FFPE and 48-month formalin-fixed tissues in Bielschowsky staining. The data from both standard FFPE and 48-M Formalin-fixed were reorganized for presentation: Citrate-citric acid buffer (a), Tris-EDTA buffer (b), Proteinase K solution (c). The ΔSNR of SPs were calculated by subtracting the SNR of SPs without retrieval from that after retrieval in Bielschowsky staining method along with Citrate-citric acid buffer (d), Tris-EDTA buffer (e), Proteinase K solution (f) treatment respectively. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Silver-staining in 7-years-stored paraffin sections

In paraffin sections stored at room temperature (20– 25°C) for 7 years, heat-induced retrieval with citrate-citric acid or Tris-EDTA markedly improved Bielschowsky staining (Fig. 5a– d, a’– d’), intensifying SP staining and reducing background. In contrast, retrieval produced negligible visual changes in PAM and Gallyas staining (Fig. 5e– l, e’– l’). In 7-Years-Stored paraffin sections, retrieval effects resembled those in standard FFPE samples (Fig. 5m– o). Specifically, heat-induced retrieval with citrate-citric acid (2.07 ± 0.33) and Tris-EDTA (2.40 ± 0.24) increased the SNR of SPs in Bielschowsky staining compared to the control (1.47 ± 0.15), whereas proteinase K showed little effect (1.59 ± 0.23) (Fig. 5m). By contrast, no clear SNR increase was observed in PAM staining for SPs (control: 2.75 ± 0.83; citrate-citric acid: 2.79 ± 0.72; Tris-EDTA: 3.01 ± 0.69; proteinase K: 2.25 ± 0.56) or in Gallyas staining for NFTs (control: 3.17 ± 0.51; citrate-citric acid: 2.94 ± 0.40; Tris-EDTA: 3.45 ± 0.29; proteinase K: 3.43 ± 0.48) (Fig. 5n, o). To illustrate Bielschowsky SNR changes, data from standard FFPE and 7‑year‑stored sections were reorganized and presented (Fig. 6a– c). SNRs for Bielschowsky staining are shown for citrate-citric acid (Fig. 6a), Tris-EDTA (Fig. 6b), and proteinase K (Fig. 6c).

Fig. 5.

The effect of retrieval methods on SNR of SPs/NFT in 7-years-stored paraffin sections. Deparaffinized sections were treated with Citrate-citric acid buffer (b, f, j), Tris-EDTA buffer (c, g, k), Proteinase K solution (d, h, l). Sections were subsequently stained with Bielschowsky (a– d), PAM (e– h), Gallyas (i– l) methods. The sections stained without retrieval were shown as a control (a, e, i). Bar graphs demonstrating the SNR of SPs of Bielschowsky (m), PAM (n), Gallyas (o). Panels below (a’– l’) show corresponding magnified views of the regions indicated by white dashed boxes. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar, 200 μm (a– l), 100 μm (a’– l’)

Fig. 6.

Comparison of the retrieval effects on the SNR of SPs between standard FFPE and 7-Years-Stored Paraffin Sections in Bielschowsky staining. The data from both standard FFPE and 7-Years-Stored Paraffin Sections were reorganized for presentation: Citrate-citric acid buffer (a), Tris-EDTA buffer (b), Proteinase K solution (c). The ΔSNR of SPs were calculated by subtracting the SNR of SPs without retrieval from that after retrieval in Bielschowsky staining method along with Citrate-citric acid buffer (d), Tris-EDTA buffer (e), Proteinase K solution (f) treatment respectively. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

After citrate-citric acid treatment, there was no significant difference in SNR improvement between 7-year-stored Sect.  (0.40 ± 0.19) and standard FFPE Sect.  (0.40 ± 0.18) (Fig. 6d). Tris-EDTA treatment yielded greater SNR improvement in 7-year-stored Sect.  (0.64 ± 0.23) than in standard FFPE Sect.  (0.46 ± 0.15) (Fig. 6e). Furthermore, Proteinase K produced a slight but significant SNR increase in 7-year-stored Sect.  (0.08 ± 0.12) compared with standard FFPE Sect.  (0.01 ± 0.07) (Fig. 6f).

Effects of retrieval on SPs microstructure

To investigate whether retrieval enhances the structural details of SPs in Bielschowsky staining, individual SPs were analyzed in adjacent tissue sections. Following heat-induced retrieval treatment with Citrate-citric acid buffer or Tris-EDTA buffer, both neuritic plaques (Fig. 7a– d) and diffuse plaques (Fig. 7e– h) revealed more detailed morphological features. It was observed that the SPs treated with Citrate-citric acid buffer or Tris-EDTA buffer exhibited superior visual clarity and more intense, deeper staining (Fig. 7b, c, f, g). Furthermore, these treated plaques demonstrated sharply defined borders against the background, in contrast to the ill-defined edges observed in specimens subjected to no heat-induced retrieval or protease K treatment (Fig. 7a, d, e, h). These results demonstrate that heat-induced retrieval improves the sensitivity of silver staining for differentiating distinct types of SPs in human brain tissues.

Fig. 7.

SPs stained with Bielschowsky viewed under 100× magnification. Sections without retrieval served as controls (a, e). Sections treated with citrate-citric acid buffer (b, f), Tris-EDTA buffer (c, g), or proteinase K solution (d, h) were presented. Scale bar, 50 μm

Effect of retrieval on SPs density

To assess the impact of retrieval on the density of SPs, low-magnification microscopy was employed to capture identical regions in adjacent sections, enabling quantitative analysis of SP numbers. SPs, absent in control sections, were revealed following heat-induced retrieval. (Fig. 8a– l). SPs per square millimeter were quantified for each retrieval treatment, and their identity verified by Aβ immunohistochemistry (Supplementary Fig. S1). Both the citrate-citric acid and Tris-EDTA methods consistently increased SP density (Fig. 8m, n). In contrast, Proteinase K showed a statistically significant but less consistent effect compared with the heat-induced retrieval methods (Fig. 8o).

Fig. 8.

Effect of retrieval methods on the density of SPs. Deparaffinized sections were treated with Citrate-citric acid buffer (b, f, j), Tris-EDTA buffer (c, g, k) or Proteinase K solution (d, h, l). Sections stained without retrieval served as controls (a, e, i). White arrows denote SPs absent in controls but visible following retrieval. Quantified SP density, normalized to control, is shown for citrate-citric acid buffer (m), Tris-EDTA buffer (n), Proteinase K solution (o). Values on the y-axis are normalized to the control group (Norm. to Con). The x-axis represents the specific retrieval treatments applied. Panels below (a’ – l’) show corresponding magnified views of the regions indicated by white dashed boxes. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar: 200 μm (a– l), 100 μm (a’– l’)

Discussion

Here, we conducted a systematic evaluation of retrieval methods and their influence on the performance of Bielschowsky, PAM, and Gallyas silver staining in formalin-fixed, paraffin-embedded human brain tissue. In Bielschowsky method, heat-induced retrieval using either citrate-citric acid buffer or Tris-EDTA buffer enhanced the SNR of SPs. Heat-induced retrieval also increased the sensitivity of Bielschowsky staining in differentiating distinct SPs subtypes, revealing finer morphological details, and markedly raised the density of detectable SPs. In contrast, proteinase K treatment produced no improvement in Bielschowsky staining. In PAM and Gallyas methods, none of the three retrieval approaches improved the SNR of SPs or NFTs. Notably, 48-month formalin-fixed samples exhibited a smaller increase in SNR of SPs after heat-induced retrieval compared with standard FFPE samples, indicating that heat-induced retrieval efficacy diminishes with prolonged formaldehyde fixation. In contrast, long-term stored paraffin sections responded more strongly to Tris-EDTA buffer, exhibiting a greater SNR enhancement after treatment compared to standard FFPE samples (Fig. 6e).

Potential mechanisms of retrieval effects in Bielschowsky silver staining

Principle of the Bielschowsky staining

Bielschowsky staining is particularly sensitive to NFTs and SPs. According to previous studies, silver nitrate impregnation in Bielschowsky staining does not occur randomly; rather, it results from precipitation of metallic silver onto protein substrates [23], particularly onto electron-rich groups [24]. At the molecular level, this affinity is attributed to interactions between silver ions and electron-rich functional groups— such as sulfhydryl, carboxyl, and amine groups— present in neuronal proteins like neurofilaments and tubulin [27].

Effect of formalin fixation time on retrieval efficacy

FFPE tissue preparation is widely employed as the standard method for specimen preservation in pathology. Formaldehyde fixation preserves tissue morphology but also introduces methylol groups and forms stable methylene bridges between adjacent protein chains, creating a cross-linked network that sterically hinders access to argyrophilic sites [28]. Seminal studies have demonstrated that retrieval techniques can restore staining intensity in previously masked tissues, suggesting that multiple synergistic mechanisms may be involved in reversing the effects of formaldehyde fixation [18] (Fig. 9a). Removal of methylol (-CH₂OH) groups and hydrolysis of methylene bridges formed during formaldehyde fixation [29] are thought to occur upon hot water exposure, reversing some of the chemical modifications induced by fixation [30]. The released peptide chains are likely to undergo conformational changes, such as denaturation or coagulation, rapidly at 100 °C, this is hypothesized to thereby expose previously inaccessible epitope determinants [20], including electron-rich reductive groups (argyrophilic sites) on neuronal proteins [31] (Fig. 9b). Thus, in tissue subjected to prolonged formalin fixation, we speculate that heat-induced retrieval with citrate-citric acid buffer or Tris-EDTA buffer enables the reduction reaction in Bielschowsky silver staining to proceed more specifically and efficiently at target sites (Fig. 9c), thereby enhancing the detectability of pathological hallmarks such as SPs. Consistent with this mechanism, the retrieval-induced improvement in Bielschowsky staining was smaller in 48-month formalin-fixed tissues than in standard FFPE sections. This finding suggests that prolonged formalin fixation produces more persistent masking and cross-linking effects, which can only be partially reversed by heat-induced retrieval. Accordingly, although citrate-citric acid and Tris-EDTA treatment still improved SP detection, the magnitude of SNR gain was reduced in long-fixed tissues, as reflected by the lower ΔSNR values. By contrast, proteinase K did not produce a significant improvement.

Fig. 9.

Proposed hypothetical mechanism of heat-induced retrieval method for Bielschowsky silver staining. Argyrophilic sites can be obscured by formaldehyde as a result of prolonged formalin fixation or long-term storage of paraffin-embedded tissue sections (a). Heat-induced retrieval exposed previously inaccessible epitope argyrophilic sites (b), thereby enabling the reduction reaction in Bielschowsky silver staining to proceed more efficiently at the target sites (c)

Effect of sections storage time on retrieval efficacy

A previous study [32] reported that sections can be stored refrigerated for up to 2 months with or without paraffin coating, and for up to 4 months when coated with paraffin. Meanwhile, storage in a desiccator is the most effective way to reduce immunoreactivity loss. Paraffin-coated sections are technically more difficult to store and exhibit the greatest loss of immunoreactivity, whereas refrigeration at 4 °C provides intermediate preservation [33]. Meanwhile, humidity and temperature also significantly influence epitope loss in FFPE tissues [34]. Prolonged storage of paraffin sections leads to alterations in epitope immunoreactivity, with preservation or loss dependent on both storage duration and temperature (Fig. 9a). The exact cause of epitope degradation remains unclear, but potential contributing factors include pre-analytical variables, oxidation, humidity and temperature [35]. One study concluded that humidity and elevated temperatures are the primary causes of epitope loss in stored paraffin sections, while oxidation is not [34]. In our study, paraffin sections were stored at room temperature (20–25 °C) for 7 years. Long‑term room temperature storage markedly reduced epitope immunoreactivity. Heat-induced retrieval reversed cross-linking or conformational changes in epitopes caused by humidity and temperature, thereby re-exposing masked epitopes (Fig. 9b). This increases the number of accessible targets for Bielschowsky silver staining and ultimately improves the SNR of SPs (Fig. 9c).

Limitations of the protease method in Bielschowsky staining

Proteolytic enzymes can produce a similar effect by cleaving certain peptide linkages, thereby increasing the flexibility of polypeptide chains [36]. However, proteolytic treatment indiscriminately cleaves peptide bonds, disrupting key structural elements that maintain the three-dimensional conformation of proteins [16, 37]. Consequently, rather than enhancing staining by restoring conformational epitopes as heat-induced retrieval does [38, 39], because Proteinase K non-specifically destroys the protein backbone rather than precisely breaking formaldehyde-induced cross-links to unmask hidden targets, this enzymatic approach inherently fails to improve the SNR in silver staining. In contrast, although Proteinase K treatment significantly affected SPs density, its effect was less consistent than that of citrate-citric acid buffer and Tris-EDTA. This variability may reflect the nonspecific proteolytic digestion of plaque-associated protein structures, which differs mechanistically from the more reproducible unmasking effect of heat-induced retrieval.

Potential mechanisms of retrieval ineffectiveness in Gallyas and PAM staining

The Gallyas method is commonly applied for visualizing NFTs, whereas the PAM method is used for detecting SPs [40, 41]. The Gallyas method relies on the capacity of myelin to bind colloidal silver particles, which are subsequently enlarged to visible dimensions by a specialized physical developer [41, 42]. The PAM method involves oxidation of stable 1,2‑glycol groups in tissue carbohydrates by periodic acid, producing reactive aldehyde groups [16]. These aldehyde groups then act as reducing agents in the methenamine-silver solution, precipitating metallic silver precisely at the sites of the original glycol groups [43, 44]. Given that neither the Gallyas nor PAM method relies on exposure of protein‑based chemical groups, retrieval appears unlikely to enhance their staining effectiveness.

Several limitations should be acknowledged, including the relatively small sample size which limits statistical power and generalizability, rendering these preliminary findings dependent on validation in larger cohorts. Variable postmortem delays among specimens may act as preanalytical confounders affecting tissue preservation and staining quality, while technical imaging variables could introduce measurement biases despite standardized procedures. Furthermore, without direct biochemical or structural evidence to pinpoint the specifically unmasked argyrophilic sites, our proposed molecular mechanism for heat-induced retrieval remains largely hypothetical. Future morphological assessments would benefit from combining traditional semi-quantitative ordinal scales with objective, automated digital image analysis to minimize inter-observer variability and mitigate these imaging biases. Overall, addressing these limitations through larger-scale studies, tightly controlled preanalytical variables, and direct mechanistic validations will be necessary to confirm and extend our findings.

Conclusions

This study systematically evaluated the influence of heat-induced retrieval techniques on tissues subjected to prolonged formalin fixation and long-term storage, using three silver staining methods—Bielschowsky, PAM, and Gallyas. For the first time, we demonstrate that heat-induced retrieval, employing either citrate-citric acid or Tris-EDTA buffers, markedly enhances the SNR, detectability of SPs in Bielschowsky staining, with Tris-EDTA showing higher effectiveness in long-stored sections. These improvements are attributed to partial reversal of conformational masking of proteins induced by formaldehyde cross-linking or prolonged storage, thereby expanding the applicability of silver staining in molecular neuropathological research. The results further clarify that Bielschowsky staining is more responsive to retrieval than PAM and Gallyas methods, supporting our central hypothesis that retrieval primarily improves silver ion accessibility in protein-dependent staining. Although further validations addressing inter-individual variability and larger sample sizes are warranted, collectively, this work improves the practicality and reliability of traditional silver staining in modern biomedical research, and offers technical support for large-scale retrospective pathological analyses and the validation of disease biomarkers in AD and related disorders. Ultimately, this methodology establishes a highly reliable morphological foundation for future multi-center quantitative evaluations and the discovery of novel therapeutic targets.

Abbreviations

AD

Alzheimer’s disease

PAM

Periodic acid-silver methenamine

SPs

Senile plaques

NFTs

Neurofibrillary tangles

FFPE

Formalin fixation and paraffin embedding

SNR

Signal-to-noise ratio

Authors’ contributions

Z.Z., X.W. and C.W. developed the concept and designed the study. Z.Z., S.Z. performed experiments and data acquisition. S.W., K.C., W.Y., Z.L. conducted data analysis and statistical evaluation. J.Z. contributed to data interpretation and provided critical reagents. X.W., C.W. supervised the project. All authors participated in drafting the manuscript, discussed the results, and approved the final version.

Funding

This work was supported by the STI2030-Major Projects (2022ZD0205202), the Anhui Province University Scientific Research Project (KJ2021A0212, 2022AH051169), the Anhui Medical University Research Funding (2021xkj008).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Approval was obtained from the ethics committee of the Ethics Committee of Anhui Medical University (No. 83244549).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.