The present protocol describes and compares the procedure to perform a full-region or sub-region of interest analysis of sagittal mouse brain sections to quantify amyloid-beta load in the APP/PS1 transgenic mouse model of Alzheimer’s disease.
Extracellular accumulation of amyloid-beta (Aβ) plaques is one of the major pathological hallmarks of Alzheimer’s disease (AD), and is the target of the only FDA-approved disease-modifying treatment for AD. Accordingly, the use of transgenic mouse models that overexpress the amyloid precursor protein and thereby accumulate cerebral Aβ plaques are widely used to model human AD in mice. Therefore, immunoassays, including enzyme-linked immunosorbent assay (ELISA) and immunostaining, commonly measure the Aβ load in brain tissues derived from AD transgenic mice. Though the methods for Aβ detection and quantification have been well established and documented, the impact of the size of the region of interest selected in the brain tissue on Aβ load measurements following immunostaining has not been reported. Therefore, the current protocol aimed to compare the Aβ load measurements across the full- and sub-regions of interest using an image analysis software. The steps involved in brain tissue preparation, free-floating brain section immunostaining, imaging, and quantification of Aβ load in full- versus sub-regions of interest are described using brain sections derived from 13-month-old APP/PS1 double transgenic male mice. The current protocol and the results provide valuable information about the impact of the size of the region of interest on Aβ-positive area quantification, and show a strong correlation between the Aβ-positive area obtained using the full- and sub-regions of interest analyses for brain sections derived from 13-month-old male APP/PS1 mice that show widespread Aβ deposition.
Alzheimer's disease (AD), the sixth leading cause of death in the United States, continues to be a public health threat, with an estimated 6.2 million Americans living with AD. This is expected to reach 13.8 million by 20601. To date, symptomatic management through medications such as cholinesterase inhibitors and memantine is the primary course of treatment2. AD is characterized by neuropathologic manifestations such as extracellular deposition of amyloid-beta (Aβ) plaques and intracellular hyperphosphorylated tau accumulation in the form of neurofibrillary tangles3,4. Formed by an endoproteolytic cleavage of the amyloid precursor protein (APP) via beta- and gamma-secretase, Aβ aggregates to form oligomers and fibrils, leading to neurotoxic effects5. Aβ has been hypothesized to serve a primary pathological role since the 1980s, and is the therapeutic target of the only FDA-approved disease-modifying therapy for AD6. As a result, transgenic AD mouse models harboring mutations in genes resulting in robust cerebral Aβ accumulation have been widely used for preclinical AD research since the early 1990s7.
Detection of Aβ species in these AD transgenic mouse brains is generally done using two immunoassays: enzyme-linked immunosorbent assay (ELISA) and immunostaining. The former assay enables quantitative determination of different Aβ species and is less time-consuming compared to immunostaining, which requires several sequential tissue processing and imaging steps, including tissue sectioning, immunostaining, imaging, and quantification8. Further, the results obtained following immunostaining are semi-quantitative8. However, the ability to spatially localize Aβ makes immunostaining an attractive approach for Aβ detection in brain tissues8.
While using Aβ immunostaining, several different quantification paradigms have been employed by different research groups. For example, some research groups quantify the Aβ load in the entire region of interest (cortex or hippocampus), while others quantify the Aβ load in a specified sub-region of interest (a part of the cortex or the hippocampus)9,10,11. Though the methods for Aβ detection and quantification have been well established and documented, the impact of the size of the region of interest on Aβ load measurements following immunostaining has not been reported. Therefore, the current protocol aimed to compare the Aβ load measurements across the full- and sub-regions of interest using an image analysis software, ImageJ.
The current study used 13-month-old APP/PS1 double transgenic male mice, which express a chimeric mouse/human APP and a mutant presenilin 1, to model early-onset of AD12. Aβ deposits start developing by 6-7 months of age, and abundant Aβ accumulation is observed both in the cortex and the hippocampus of these mice by 9-10 months of age12. The transgenic amyloid peptides and holoprotein can be detected by 6E10-immunostaining13, making it a desirable animal model for the present protocol. The procedure covered herein includes brain tissue preparation, immunostaining of free-floating sections, imaging, and quantification of Aβ load in full- versus sub-regions of interest. The analysis shows a strong correlation between the full- and sub-regional quantification, indicating robust agreement between these two methods in the brain tissue sections derived from 13-month-old APP/PS1 male mice that show abundant Aβ deposits.
All animal experiments were conducted in compliance with the University Laboratory Animal Resources under protocols approved by the University of California, Irvine, Institutional Animal Care and Use Committee. The experiments were performed with male B6C3-Tg(APPswe, PSEN1dE9)85Dbo/Mmjax (APP/PS1) mice (13-month-old, n = 35). The mice were obtained from commercial sources (see Table of Materials).
1. Brain tissue preparation
2. Immunofluorescence
3. Imaging
4. Full-region of interest analysis
NOTE: The present work's two regions of interest are the hippocampus and the iso-cortex. Full-region of interest analysis represents the analysis of the entire iso-cortex (referred to as the cortex going forward) or the hippocampus in the imaged brain tissue section.
5. Sub-region of interest analysis
NOTE: Sub-region of interest analysis represents the analysis of a part of the cortex or the hippocampus in the imaged brain tissue section.
Here, two different methods are compared to quantify the 6E10-positive area in the hippocampus and the cortex of mouse brain tissues. The two methods are the full-region and sub-region of interest analyses (Figure 1). The full-region of interest analysis, as the name suggests, involves outlining the entire region of interest (in this case, either the iso-cortex or the hippocampus) to determine the 6E10-positive area (Figure 1A,B). The sub-region of interest analysis involves selecting a pre-defined region within the region of interest to determine the 6E10-positive area (Figure 1C,D). The stepwise ImageJ protocol for the two methods is shown in Figure 2, Figure 3, and Figure 4.
This study used three readers; two independent readers performed the sub-region of interest analysis, and the third reader performed the full-region of interest analysis. As seen in Figure 5A,B, there was a strong significant positive correlation (p < 0.0001) between the 6E10-positive area reported by the two readers performing the sub-region of interest analysis (Pearson correlation coefficient r = 0.97 for the cortex and r = 0.96 for the hippocampus). The 6E10-positive areas reported by the two readers for the sub-region of interest analysis were averaged, and the averaged sub-region of interest 6E10-positive area shared a strong significant positive correlation (p < 0.0001) with the 6E10-positive area obtained using the full-region of interest analysis for both the cortex (Pearson correlation coefficient r = 0.96; Figure 5C) and the hippocampus (Pearson correlation coefficient r = 0.95; Figure 5D). The mean cortical- and hippocampal-6E10-positive area obtained by the full-region and sub-region of interest analyses were comparable with no significant difference, confirming the agreement between the two methods (Figure 5E). Further, the insoluble Aβ1-42 was measured in whole-brain homogenates in a subset of mice and the cortical (Figure 5F) and hippocampal (Figure 5G) 6E10-positive area determined by the full-region of interest analysis was significantly (p < 0.01) correlated with insoluble Aβ1-42 load using ELISA (see Table of Materials).
Figure 1: Full- versus sub-regions of interest selection. Representative images showing the full iso-cortex (cortex) and hippocampus outlined for the full-region of interest analysis in (A) and (B), respectively. Representative images show the selection of sub-region/s of the cortex and hippocampus for the sub-region of interest analysis in (C) and (D), respectively. Scale bar = 200 µm. Please click here to view a larger version of this figure.
Figure 2: Protocol for the full-region of interest 6E10-positive area quantification. Image analysis steps showing the original image (A), the image after brightness/contrast adjustment (B), selection of the area of interest (C), clearing (D), threshold adjustment (E), and the final image ready for analysis (F). The numbers in the figure designate the step numbers in the protocol. Scale bar = 200 µm. Please click here to view a larger version of this figure.
Figure 3: Protocol for the sub-region of interest 6E10-positive area quantification. Image analysis steps showing the original image (A), the image after brightness/contrast adjustment (B), selection of the area of interest (C), duplication of the image of the region of interest (D), changing the image to 8-Bit and inverting the image (E), threshold adjustment (F), and final image ready for analysis (G). The numbers in the figure designate the step numbers in the protocol. Scale bar = 200 µm. Please click here to view a larger version of this figure.
Figure 4: Protocol for the region of interest rotation. Image analysis steps showing the selection of the area of interest and rotation of the selection box to fit the tissue curvature (A), the image of the region of interest after duplication (B), and the image after clearing the outside (non-region of interest) area (C). The numbers in the figure designate the step number in the protocol. Scale bar = 200 μm. Please click here to view a larger version of this figure.
Figure 5: Correlation between the full- and sub-regions of interest analyses. Scatter plots show the correlation between the 6E10-positive area by the two independent readers performing the sub-region of interest analysis for the cortex (A) and the hippocampus (B). A strong positive correlation is observed between the 6E10-positive area resulting from the sub-region of interest analysis and the full-region of interest analysis for both the cortex (C) and the hippocampus (D). There is no statistically significant difference in the mean 6E10-positive area by the full-region of interest and the sub-region of interest analyses in the cortex and the hippocampus (E). A significant correlation is observed between whole-brain homogenate insoluble Aβ1-42 measurements using ELISA and the full-region of interest analysis for both the cortex (F) and the hippocampus (G). Data were analyzed using the Pearson correlation coefficient, r, in (A-D) and (F-G), and using two-way repeated-measures ANOVA in (E) using a graphing and statistics software. Data are presented as mean ± standard error of the mean (SEM) of n = 35 mice in (E), and a two-tailed p < 0.05 was considered statistically significant. Please click here to view a larger version of this figure.
The protocol described herein outlines the procedure for hemi-brain preparation for sagittal sectioning, immunofluorescent staining of Aβ deposits using the 6E10 antibody on free-floating sections, imaging of the Aβ-stained brain sections followed by quantification of the Aβ deposits in the cortex and the hippocampus of mouse brain tissue using an image analysis software. While there are published protocols to quantify Aβ load in brain tissue sections8,10, this protocol describes the steps involved in the quantification of Aβ load in the entire iso-cortex (referred to as cortex) and hippocampus in comparison with the Aβ load in a sub-region of interest in the cortex and the hippocampus, when this may be desired. The correlation between the full- versus sub-regions of interest analyses is also provided.
There are several critical steps in the protocol. First, the protocol described is for 20 µm thick brain tissue sections subjected to free-floating immunostaining, which results in optimal antibody penetration within the tissue section18. The free-floating technique may require the tissue sections to be manually transferred between the different solutions during the immunofluorescent staining, and careful handling of the tissue sections throughout the procedure. This is especially crucial when the tissue sections are immersed in the 70% formic acid solution for antigen retrieval in the current protocol, increasing tissue fragility for thin sections. Alternate approaches to the described protocol include using thicker tissue sections (e.g., 30-40 µm) or using tissue sections that are directly mounted onto positively-charged slides before the immunofluorescent staining. Second, the protocol described herein uses a fluorescently-labeled 6E10 antibody. Besides using the fluorescently-labeled 6E10 antibody, non-fluorescent 6E10 antibodies (e.g., horseradish peroxidase-conjugated 6E10 antibody) can also be used to detect Aβ load in brain tissue sections, and the current protocol can be adapted to quantify Aβ-positive immunochemical stains in the brain tissue sections as described previously8. Third, the accuracy of the results for Aβ load quantification will depend on the appropriate threshold selection in the analysis software, which is dependent on the tissue background and signal intensity. Threshold selection must be performed by the end-user such that only Aβ-positive stains are selected for quantification. End-user intervention is required to optimize the specific threshold that can be applied to all the images to assure the accuracy of the threshold setting. Fourth, since the sub-region of interest analysis requires selecting a small region of interest in the tissue section, two independent readers were used for this analysis. To maintain independence and blinding during data collection, all the images were number coded; the image analysis sequence was randomized between the readers such that the different readers analyzed different images at any given time, and the data was submitted at the end of each week. Due to the increased likelihood of inter-reader variability in the region of interest selection in the sub-region of interest analysis, the readers were trained using several sample images to optimize region selection in the cortex and the hippocampus before beginning the data collection. This training is crucial to reduce inter-reader variability, and as can be seen (Figure 5A,B), the 6E10-positive area reported by both the readers shows a strong relative agreement in the current study.
The current protocol and the results provide valuable information about the impact of the region of interest size on Aβ-positive area quantification. A larger region of interest is expected to represent the tissue more than a smaller region of interest. Therefore, sampling a larger tissue is desirable to accurately quantify the Aβ load in tissues. However, in the case of homogenous Aβ load distribution within the tissue, a smaller sampling region is generally considered a good representation of the larger tissue under analysis. The current study results confirm this, and the Aβ load in the entire cortex and the hippocampus was a strong correlate of Aβ load in a selected sub-region of the cortex and the hippocampus (Figure 5C,D). To further confirm the agreement between the full- and sub-regions of interest analyses, the mean 6E10-positive area in the cortex and the hippocampus were compared, and no difference between the two methods (Figure 5E) was found. This confirms that either of these methods (full- or sub-region analysis) yield comparable Aβ load measurements.
The current protocol has some limitations. The two methods (full-region versus sub-region analysis) may not always be used interchangeably. The choice of using the full- or sub-region analysis will depend on the regional distribution of Aβ within the tissue, which is impacted by the age, sex, and strain of the AD mouse model. At 13 months, the Aβ load is distributed throughout the cortex and hippocampus of the APP/PS1 mice. However, at 6 months, Aβ deposits are limited to the cortex, and minimal deposits are observed in the hippocampus12. Under such conditions, the full-region of interest analysis may be the desired approach to increase the tissue sampling area and thereby the Aβ signal. On the other hand, sub-region analysis may be the method of choice when Aβ load in a specific brain region is of interest, (e.g., the somatosensory cortex). Additionally, at 13 months, the APP/PS1 male mice demonstrate intense 6E10-positive stains, and the immunofluorescent staining results in an excellent signal with a very low background, making the current protocol very suitable for quantification under the given conditions. It is unclear if this quantification method can be applied successfully to less intense staining, and future work will be required to answer this question. The immunofluorescent and image quantification method presented herein detects all forms of Aβ, including the precursor form13. As a result, if there is an interest in the detection of a specific Aβ specie (e.g., Aβ1-40 or Aβ1-42), antibodies specific to these Aβ isoforms can be used. Therefore, though the 6E10 immunofluorescent staining and detection method correlated with the measurements of Aβ1-42 measurements in whole-brain homogenates using ELISA (Figure 5F,G), the correlation was only modest. This can be attributed to the measurement of only Aβ1-42 using the ELISA and detection of all Aβ species using the 6E10 immunostaining. The current study uses three readers to assess the agreement and correlation between the full- and sub-regional analyses. Having additional readers may improve the robustness of the study and can further validate the agreements between the two methods presented here. Further, we use the Pearson correlation as a measure of agreement, which is widely used among other methods to describe the agreement between continuous variables19. However, one limitation of using the Pearson correlation to determine agreement is that the two methods used herein may provide related results, but one method may result in overall higher values than the other due to systematic bias. Therefore, Pearson correlation is a good measure of relative agreement19. To increase the robustness of the protocol, additional methods to confirm the absolute agreement, such as comparing the mean 6E10-positive area by the two methods (Figure 5E), can be used19. Taken together, the current protocol compares the Aβ load detected by immunofluorescent staining and analyzes the full- and sub-regions of interest in brain tissue sections. The results show a strong correlation between these two methods for brain tissue sections derived from 13-month-old APP/PS1 male mice that show abundant Aβ deposits.
The authors have nothing to disclose.
Research reported in this publication was supported by the National Institute of Aging of the National Institutes of Health under award numbers R01AG062840 (to RKS) and R01AG072896 (to RKS). The content is solely the authors' responsibility and does not necessarily represent the official views of the National Institutes of Health. Approximately $200k (100%) of federal funds supported this project. We would also like to thank Dr. Joshua Yang for his assistance with manuscript editing.
15 mL conical tubes | ThermoFisher Scientific, MA, USA | 339650 | https://www.thermofisher.com/order/catalog/product/339650 |
24-well plates | Fisher Scientific, NH, USA | FB012929 | https://www.fishersci.com/shop/products/jet-biofil-surface-treated-steriletissue-culture-plates-3/FB012929 |
Amyloid beta 42 human ELISA kit | ThermoFisher Scientific, MA, USA | KHB3441 | https://www.thermofisher.com/elisa/product/Amyloid-beta-42-Human-ELISA-Kit/KHB3441 |
Aqueous mounting media | Vector laboratories, CA, USA | H-5501-60 | https://vectorlabs.com/products/mounting/vectamount-aq-aqueous-mounting-medium |
Bovine serum albumin | Sigmaaldrich, MO, USA | A2153-50G | https://www.sigmaaldrich.com/US/en/product/sigma/a2153?gclid=CjwKCAjw9aiIBhA1EiwAJ_G TSiZ9B3YGz3RvSpWqCH4CPW78 Dj4WlgxmzPs631z5IHmy5XLV TdC_jBoC9zQQAvD_BwE |
BZ-X710 Keyence all-in-one fluorescence microscope | Keyence, IL, USA | BZ-X710 | https://www.keyence.com/products/microscope/fluorescence-microscope/bz-x700/models/bz-x710/ |
Clear nail poilsh | User preference | NA | None |
Cryostat | Leica Biosystems, IL, USA | Leica CM1860 Cryostat | https://www.leicabiosystems.com/us/histology-equipment/cryostats/leica-cm1860/ |
Formic acid | Sigmaaldrich, MO, USA | F0507-500ML | https://www.sigmaaldrich.com/US/en/product/sigald/f0507?gclid=CjwKCAjw9aiIBhA1EiwAJ_G TSheH6JMGnla50C3Ag0cLzXE8 BObxvDApl0udjYAPZmBGe7a 8PRUv1RoCt34QAvD_BwE |
Glass coverslips | VWR, PA, USA | 48393-081 | https://us.vwr.com/store/product/4645817/vwr-micro-cover-glasses-rectangular |
GraphPad Prism | GraphPad Software, CA, USA | Version 8 | https://www.graphpad.com/scientific-software/prism/ |
ImageJ 1.51k | National Institutes of Health, MD, USA | Version 1.53e | https://imagej.nih.gov/ij/download.html |
Mice | Jackson Laboratories, ME, USA | 034829-JAX | https://www.mmrrc.org/catalog/sds.php?mmrrc_id=34829 |
Paraformaldehyde | Sigmaaldrich, MO, USA | P6148-500G | https://www.sigmaaldrich.com/US/en/product/sial/p6148?gclid=CjwKCAjw9aiIBhA1EiwAJ_G TShtLb9Ax9MmRyrFn6Rfmmg 1l52_5XZFXOeXT24ik8Lkw GH7fvlDoHBoChzYQAvD_BwE |
Phenytoin/pentobarbital based anesthetic (Euthasol) | Patterson Veterinary, MA, USA | 07-805-9296 | https://www.pattersonvet.com/Supplies/ProductFamilyDetails/PIF_32818 |
Phosphate-buffered saline | Fisher Scientific, NH, USA | BP661-50 | https://www.fishersci.com/shop/products/pbs-1x-powder-concentrate-white-granular-powder-fisher-bioreagents-2/BP66150 |
Plus (+) microscope slides | Ted Pella, Inc., CA, USA | 260100 | https://www.tedpella.com/histo_html/slides.htm#260384 |
Primary antibody (6E10) | Biolegend, CA, USA | 803013 | https://www.biolegend.com/en-us/products/alexa-fluor-488-anti-beta-amyloid-1-16-antibody-10833 |
Sucrose | Sigmaaldrich, MO, USA | 47289 | https://www.sigmaaldrich.com/US/en/product/supelco/47289?gclid=CjwKCAjw9aiIBhA1EiwAJ_ GTSuwlymWL_PUl2KIMHymi GLOWluZdPjf3pRcjMEjQD siItfWiG-C2-RoCxyoQAvD_BwE |
Triton X 100 | Sigmaaldrich, MO, USA | T8787-100ML | https://www.sigmaaldrich.com/US/en/product/sigma/t8787?context=product |