Mohorko, Nina BS; Repovš, Grega PhD; Popović, Mara MD, PhD; Kovacs, Gabor G. MD, PhD; Bresjanac, Mara MD, PhD
Numerous markers that bind to the pathological protein aggregates are being developed to understand the conformational pathology of the protein hallmarks of neurodegenerative diseases. Successful searches may yield valuable diagnostic tests and potential therapeutic options. The pathogenesis and progression of tauopathies (a major group of neurodegenerative diseases) are closely related to fibrillation and aggregation of the microtubule-associated protein tau (1). In a healthy adult brain, there are 6 tau isoforms generated by the alternative splicing of pre-mRNA (2). They are usually classified into 2 functionally different groups according to the number of microtubule-binding domains (MBDs) they contain. Those with 3 MBDs are called 3-repeat tau (3R-tau) and those with 4 MBDs are called 4-repeat tau (4R-tau) (3). Normally, 3R-tau and 4R-tau are quantitatively equally represented, but changes in this ratio are present in several neurodegenerative disorders and might contribute to neurodegeneration (4).
Tau binding to microtubules and plasma membrane is determined by its phosphorylation, that is, the higher the phosphorylation, the weaker the binding (5). Aberrant hyperphosphorylation of tau results in an increase of the concentration of the unbound phospho-tau in the cytosol that leads to its aggregation and the formation of pathological fibrillar tau inclusions (FTIs) (6, 7). Fibrillar tau inclusions are initially intracellular, but cell death can result in their extracellular localization and accompanying changes in their morphology and molecular characteristics (8, 9).
Three main types of tau filaments have been observed in different FTIs: paired helical filaments (PHFs), straight filaments (SFs), and random coiled filaments (10). Different tauopathies are characterized by different types of tau filaments in the FTI sand by a different combination of tau isoforms with specific phosphorylation patterns. Tauopathies are thus divided into 5 different classes according to these characteristics (11), with most disorders falling into Classes 1 to 3. Alzheimer disease (AD) is the main representative of Class 1 disorders, with all 6 tau isoforms accumulating in the form of PHFs and SFs. Progressive supranuclear palsy (PSP) is representative of Class 2 disorders, with FTIs composed of predominantly 4R-tau that preferentially accumulates in the form of SFs. According to most authors, Pick disease (PiD) is the only representative of Class 3 disorders, in which only 3R-tau accumulates in the form of SFs and random coiled filaments (10-12).
Definite diagnosis of tauopathies is possible only by postmortem examination and is based on the anatomical distribution and morphological characteristics of the FTIs (13). They are detected and characterized using immunohistochemistry and/or silver staining methods (14, 15). Thioflavine S (ThS) is also used as an auxiliary detection method (16). Although new quick and reliable procedures for fluorescent detection of β-amyloid (Aβ) and prion amyloid have been reported (17-20), there have not yet been reports of their systematic evaluation in labeling FTIs in different tauopathies. In view of the relevance of tau pathology to the progression of dementing illnesses, the detection and reliable identification of FTIs by these small probes would offer new possibilities for in vivo diagnosis and therapy.
Curcumin (CCM) is a natural fluorescent compound from a rhizome of Curcuma longa Lynn., which gives curry its typical yellow color. Curcumin, or diferuloylmethane (IUPAC name, [1E,6E]-1,7-bis[4-hydroxy-3-methoxyphenyl]-1,6-heptadiene-3,5-dione), is a symmetrical molecule with 2 phenol groups connected by 2 α,β-unsaturated carbonyl groups that form a diketone. This structure determines the many biological functions of CCM, such as its antioxidant and anti-inflammatory activities as well as its fluorescent properties (6). In aqueous solutions, CCM absorbs light at around 420 nm and emits fluorescence at about 530 nm, both with low intensities (21). However, the fluorescence is enhanced with the decrease in polarity of the medium (21). Curcumin also undergoes a blue shift when it is bound to hydrophobic cavities inside a protein such as bovine serum albumin (22). These properties make CCM a candidate for a molecular imaging probe in fluorescent microscopy of protein aggregates. Moreover, it has been shown that CCM binds to aggregates of Aβ in vivo and reduces the formation and abundance of Aβ plaques (19, 23). Curcumin also binds to prion protein aggregates and inhibits PrPSc accumulation in vitro (20, 24). Curcumin seems to have similar effects on α-synuclein (25). These reports suggest a diagnostic and possibly therapeutic potential for CCM.
To our knowledge, analysis of CCM binding to tau aggregates has not been previously investigated. Thus, we examined CCM labeling of characteristic pathological neuronal tau deposits in brain sections obtained postmortem from confirmed typical cases of 3 representative tauopathies: AD, PSP, and PiD. We aimed to determine how CCM detection of FTIs compared with standard methods for visualization of tau pathology. To that end, we compared CCM labeling of brain tissue sections with an established fluorescent detection method with ThS and with Gallyas silver staining method (16), which is in routine use for neuropathological diagnosis of neurofibrillary tangles (NFTs) (26). In addition, we analyzed CCM labeling of immunochemically identified tau pathology to determine characteristics of CCM labeling of FTIs relative to the standard antibody against pathological hyperphosphorylated tau (AT8) (14). Finally, we aimed to determine whether CCM labeling of FTIs is influenced by their 3R-tau or 4R-tau isoform composition.
MATERIALS AND METHODS
Five-micrometer-thick deparaffinized sections of paraformaldehyde-fixed paraffin-embedded human cerebral samples from patients with previously confirmed AD (n = 6), PSP (n = 6; with verified 4R-tau only FTIs), and PiD (n = 6; with verified 3R-tau only FTIs) were obtained from the archive of the Institute of Pathology, Faculty of Medicine, University of Ljubljana and from the Medical University of Vienna, Vienna, Austria. The following brain regions were studied: hippocampus and neocortex (in AD and PiD) and midbrain and basal ganglia (in PSP). In addition, similarly processed brain tissue sections from a patient with vascular hyalinosis were used as a negative control, that is, nonfibrillar protein deposition. Brain sections from a patient with cerebral amyloid angiopathy were used as a positive control.
Sections from all regions of interest in all the brains were initially analyzed to assess AT8 and CCM labeling, patterns and distribution of hyperphosphorylated tau pathology, and the overlap of FTI labeling using both detection methods. Based on this analysis, subsequent colocalization analysis focused on neuronal fibrillar tau in the most representative portions of hippocampus in AD and PiD and mesencephalon in PSP.
Sequential labeling of the same sections with different techniques allowed direct comparison of these techniques for detection of the same pathological structures that are characteristic of tauopathies. Care was taken to ensure that the sequential techniques did not interfere with one another based on a prior pilot study. In addition, it was ascertained that every obtained signal was derived solely from the currently applied label. For this purpose, sections were thoroughly destained, and effects of destaining were verified by photography of the same loci before the subsequent labeling. For colocalization analysis of pairs of different labeling procedures, the recorded autofluorescence and trace fluorescence after every destaining was subtracted from the signal of the relevant subsequent labeling procedure. Colocalization analysis was performed on these subtracted images, thus eliminating not only fluorescence cross contamination but also autofluorescent noise. The figures display the original unsubtracted photomicrographs. All photomicrographs were taken with Nikon DXM2000 camera (Nikon, Tokyo, Japan) attached to Olympus AX-81 fluorescent microscope (Olympus, Tokyo, Japan).
Hematoxylin and Eosin Staining
Standard hematoxylin and eosin (H&E) staining was performed on all sections. In case of tauopathies, loci that seemed to contain FTIs were identified by a neuropathologist (Mara Popović) in H&E-stained sections in brightfield and photographed. Representative pathological features in controls were documented in the same way. The coordinates of the loci were recorded with the images. A minimum of 10 loci per section was photographed. The H&E was then thoroughly destained by incubating the sections in 1% HCl (Merck, Darmstadt, Germany) in 70% ethanol (EtOH; Merck) overnight. Subsequent labeling, destaining, image analysis, and presentation of the results were done in a blind fashion by non-neuropathologists (Nina Mohorko, Mara Bresjanac, Grega Repovš).
The sections were permeabilized with 0.3% Triton X-100 (Sigma-Aldrich, St Louis, MO) in Milli-Q water before incubation with CCM. Curcumin from Curcuma longa (Sigma) was first dissolved in 100% EtOH to make a 1-mmol/L stock, which was further diluted with Milli-Q water to a 10-μmol/L working solution. The sections were incubated in the working solution for 10 minutes, rinsed, differentiated in 70% EtOH for 1 minute, rinsed, and coverslipped with glycerol (Kemika, Zagreb, Croatia). Photomicrographs of CCM-labeled sites were taken using filter with excitation/emission wavelengths of 420 to 440/475 nm. Curcumin was destained by incubating the sections in 98% EtOH overnight. We verified that no CCM was detectable in the sections and photographed the recorded loci using filter with excitation/emission wavelengths of 460 to 495/510 nm as evidence of this.
To compare CCM labeling with that using ThS (Dako, Carpinteria, CA), the sections were further incubated in 0.05% ThS (wt/vol) in 50% EtOH for 8 minutes. The same loci as before were photographed using filter with excitation/emission wavelengths of 460 to 495/510 nm. Because of its resilience, ThS was not destained, but the same loci were photographed using the filter with excitation/emission wavelengths of 530 to 550/590 nm used for subsequent visualization of immunofluorescence. As illustrated by comparison of the third and fourth columns of Figure 1, the ThS signal did not bleed into this wavelength band.
After the sequential fluorescent labeling of the AD, PSP, and PiD samples, the samples were then divided into 3 groups for immunofluorescent (IF) visualization of tau.
To assess whether the photographed sites contained pathological FTIs, we performed IF with the standard monoclonal antibody AT8 (14) (1:100; Pierce Endogen, Thermo Fischer Scientific, Rockford, IL) on the first series of sections. Next, to determine whether CCM differentiated between 3R-tau- and 4R-tau-composed FTIs, monoclonal antibodies against 3R-tau (RD3 27, 1:150; a generous gift of Dr Rohan de Silva, University College London, London, UK) and 4R-tau (ET3 , 1:10; a generous gift of Dr Peter Davies, Albert Einstein College of Medicine, Manhasset, NY) were used on the second and third series of tissue sections from all tauopathies, respectively. Briefly, these sections were pretreated by 10-minute pressure cooking in sodium citrate buffer, pH 6.0. All sections were then rinsed in buffer before a 10% normal goat serum was applied in buffer, pH 7.2, for 30 minutes to block nonspecific binding of the secondary antibody. Incubation with the primary antibodies was performed overnight at 4°C. Finally, the sections were incubated in Alexa 546-conjugated goat anti-mouse secondary antibody (1:750; Molecular Probes, Invitrogen Co, Carlsbad, CA). The same loci as in previous labeling runs were photographed using filter with excitation/emission wavelengths of 530 to 550/590 nm.
Gallyas Silver Stain
Because the original Braak system of AD staging was based on Gallyas silver staining (1) and many neuropathologic laboratories use silver staining techniques to diagnose tauopathies, Gallyas silver staining was performed as the last step to visualize FTIs in the 3 diseases. This allowed direct comparison of individual tauopathy-characteristic FTI labeling by CCM, ThS, and Gallyas silver staining.
Briefly, the sections were placed in 5% periodic acid (Merck) for 5 minutes, rinsed in dH2O, placed in silver iodide for 1 minute, washed 2 × 5 minutes in 0.5% acetic acid (Merck), rinsed in dH2O, and immersed in developer until they were visibly pale brown-gray. Development was stopped in 0.5% acetic acid. Sections were then rinsed in dH2O, incubated in 0.1% gold (III) chloride for 5 minutes (Merck and Kemika), rinsed in dH2O, incubated in 1% sodium thiosulfate (Kemika), rinsed in dH2O, dehydrated, and coverslipped for viewing.
Image Processing, Colocalization Analysis, and Statistics
Digital photomicrographs of the selected loci were further processed and analyzed with custom-made software, mImage (29). The following steps were performed on each image before computation of colocalization coefficients. First, the fluorescent signal was filtered so that only the strongest color channel from each staining remained (CCM, blue; ThS, green; AT8, ET3, RD3 IF, red). Second, to remove noise, improve signal-to-noise ratio, and attenuate errors of spatial coregistration of images, the images were smoothed with a 2-dimensional Gaussian kernel with full width at half maximum of 3 pixels. Third, the recorded autofluorescence and trace fluorescence after every destaining were subtracted from the image of the relevant subsequent labeling procedure. Fourth, the image was thresholded so that background noise and remaining autofluorescence were removed (set to 0) and only signal attributed to the label used remained. This step was performed linearly over the whole image, so that no data were lost and no artifacts were created (30).
The colocalization of signal in pairs of images obtained by different labeling procedures was expressed using 2 colocalization coefficients (M). Each of the coefficients proposed by Manders et al (31) reflects the proportion of overall florescence signal intensity in 1 image that is colocalized with above-threshold signal in the other image. The colocalization coefficient of Labels A with B is therefore computed as:
where in(A) is the signal intensity of pixel n in image A. The value of cn(B) is 1 if the signal of pixel n in image B is above threshold and 0 otherwise. In an analogous manner, colocalization coefficient of Label B with A is computed as:
Equation (Uncited)Image Tools
Both coefficients were computed for each pair of images of the same object obtained using different labeling procedures.
Equation (Uncited)Image Tools
To visualize the results, colocalization scatter plots were created in which every data point represented an observed structure. The position along the y axis represented the value of M(A|B), and the position along the y axis represented the value of M(B|A) for the observed structure. These scatter plots enable visualization of different colocalization scenarios. As an example, 4 general cases can be illustrated. If Labels A and B overlapped to a large extent, the structure would be represented in the upper right corner. If A labeled the same but smaller part of a given structure than B, that structure would be represented by a data point in the upper left corner. If A labeled the structure extensively and B labeled only a part of it, that structure would be located in the lower right corner. If A and B labeled different parts of the structure and their signals did not overlap, the structure would be represented in the lower left corner. In case that at least one of the probes did not label any part of the structure, one of the coefficients could not be computed and the FTIs could not be represented in the scatter plot.
Explained differently, if Label B represented a reference label, then M(B|A) reflected the sensitivity of Label A, whereas M(A|B) reflected its specificity. A data point located in the upper right corner of the scatter plot would therefore represent a case of high sensitivity and high specificity of Label A, lower right corner a case of high sensitivity with low specificity of Label A, and the upper left the reverse.
CCM Detection of FTIs is Concordant With ThS Labeling
Overall, CCM and ThS labeled the same structures identified in H&E as likely FTIs (Figs. 1 and 2). A graph displaying the relationship between the 2 colocalization coefficients M(CCM|ThS) and M(ThS|CCM) for all labeled FTIs across the tauopathies analyzed revealed that CCM and ThS signals colocalized extensively; the CCM signal covered a slightly smaller area than ThS (Fig. 2A). Strong correlation was found between signal area size (r = 0.72; Fig. 2B) as well as signal intensity (r = 0.73; Fig. 2C) for CCM and ThS labeling of the same FTIs across all tauopathies.
CCM Labeled FTIs in AD and PSP but not in PiD
Comparing CCM with AT8, the standard antibody for detection of hyperphosphorylated tau, showed that both labeled H&E-identified FTIs in AD (Figs. 1B, D) and in PSP (Figs. 1F, H), whereas in PiD, neither cortical nor hippocampal Pick bodies were visualized by CCM fluorescence (Fig. 1J) but displayed AT8 IF (Fig. 1L). The detailed relationship of CCM labeling and AT8 IF is evident in the scatter plots showing the colocalization coefficients M(CCM|AT8) and M(AT8|CCM) for the identified FTIs in each of these tauopathies (Fig. 3).
The CCM and AT8 IF signal colocalization coefficient scatter plot for AD reveals 3 FTI clusters. The largest cluster in the upper right quadrant represents FTIs that displayed high colocalization of CCM fluorescence and AT8 IF (Fig. 3A). There was a smaller cluster in the upper left quadrant, reflecting those FTIs that labeled extensively with AT8 but only partially with CCM. A third large cluster of FTIs centered in the lower left corner and spread along the abscissa. This cluster represents FTIs robustly labeled with CCM but only partially with AT8 overlapping with CCM to a variable degree. Representative examples of the AD FTIs belonging to these 3 clusters can be seen in Figures 1A to D, as they were followed from their initial detection in H&E-stained sections through CCM, ThS fluorescence, and subsequent AT8 IFs.
Gallyas silver staining was performed last and displayed qualitatively the same pattern of labeling as CCM. Based on differential labeling by CCM, AT8, and Gallyas silver staining, 3 patterns of pathological tau deposits could be distinguished in AD. First, the basophilic apparently intracytoplasmic material without pronounced fibrillar morphology was often only partially or weakly labeled with CCM and Gallyas but was readily visualized by AT8 (empty arrow facing left in Figs. 4A-D). Second, the FTIs with discernible basophilic fibrillar morphology were extensively labeled by all 3 techniques (short white arrows facing right in Figs. 4A-D). Third, the extracellular FTIs with loose fibrillar morphology displayed strong CCM fluorescence and intense Gallyas silver staining but no or only punctate IFs to AT8 (thin white arrows facing right in Figs. 4A-D). These 3 labeling patterns are readily discernible in the CCM-AT8 colocalization scatter plot (Fig. 3A). The present analysis focused on NFTs in AD, but neuropil threads were visualized by CCM as well as ThS and Gallyas methods (not shown).
Another type of FTI, characteristic for PSP (i.e. mesencephalic globose NFTs), showed the same pattern of labeling as in AD. Most basophilic intracellular FTIs labeled well with CCM, ThS, AT8 (short white arrow facing right in Figs. 1E-H), and Gallyas (not shown). A large proportion of similar but pale H&E structures displayed intense and extensive labeling by CCM, ThS, and Gallyas, but only minimal or no AT8 IF (thin white arrows facing right in Figs. 1E-H). Only a few cells in PSP were AT8 positive, but not extensively CCM positive. This pattern is also evident in the CCM-AT8 colocalization scatter plot (Fig. 3B).
Pick bodies and Pick cells were not visualized by CCM fluorescence in any of the PiD cases in either the cortex or the hippocampus, despite extensive AT8 immunolabeling of abundant typical pathology (Pick cells and bodies; Figs. 1J, L, 3C). Data points in the colocalization analysis scatter plots do not represent CCM labeling of PiD-specific FTIs. Upon inspection, they were found to reveal scattered pixels of suprathreshold signal in CCM fluorescence that colocalized with AT8 signal. Neither ThS nor Gallyas silver staining detected FTIs in PiD.
With respect to other hyperphosphorylated tau pathology, we did not detect suprathreshold fluorescence of CCM in dystrophic neurites of neuritic AD plaques (although the plaques of all types were visualized by CCM) or in tufted astrocytes and coiled bodies (both types of pathology abundantly present) in the PSP cases (not shown).
CCM Labeling Differed Between the FTIs Composed of Different Tau Isoforms
The FTIs in AD had both 3R-tau and 4R-tau, whereas only 4R-tau was detected in PSP and only 3R-tau in PiD. Analysis of tau isoform composition of the CCM-positive structures in AD revealed that CCM colocalized well with 3R-tau IF. In some cases, it labeled only a small area, as illustrated by clusters of the analyzed FTIs in the upper right and upper left quadrants in Figure 5C, respectively. However, there was a group of outlying FTIs in which CCM and RD3 labeling showed no overlap (Figs. 5A-C, lower left corner). Numerous ET3-tau-positive FTIs in AD displayed considerable colocalization with CCM positivity (upper right quadrant in Fig. 5F), whereas another cluster of FTIs was CCM positive but only sparsely ET3 positive, overlapping with CCM to a variable degree (bottom left quadrant in Fig. 5F).
In PSP, a group of FTIs displayed extensive colocalization of CCM and ET3 (Fig. 5I). Another tight cluster revealed no overlap of the 2 signals (lower left corner in Fig. 5I), whereas a dispersed group of FTIs in the lower right quadrant represented a population of FTIs in which CCM labeling was colocalized with but only partly overlapped with the ET3 signal because of the smaller signal area covered by ET3.
Of the 32 FTIs identified in the PiD brain section used for this analysis, only 10 displayed pixels with suprathreshold CCM signal and are represented in the scatter plot. In only 3 of these data points did CCM label the same area as RD3, although to a much smaller extent (Fig. 5L). This labeling pattern was found also when ThS and Gallyas staining were applied to PiD. None of the 3R-tau-positive FTIs in PiD were labeled with either ThS or Gallyas (not shown); no 4R-tau IF was detected in PiD samples (not shown).
Curcumin (like ThS and Gallyas silver staining) labeled the vessel walls in the cerebral amyloid angiopathy case used as a positive control but did not label vessel walls in the case of vascular hyalinosis used as a negative control (not shown).
Our sequential labeling and image analysis shows that CCM reliably and reproducibly labeled most pathological neuronal fibrillar tau deposits in postmortem brain sections of AD and PSP, the representative diseases of Classes 1 and 2 tauopathies. Curcumin did not label Pick bodies in PiD, in which the FTIs consist solely of the 3R-tau isoform. Curcumin labeling was also found to be in good agreement with ThS and Gallyas staining, thereby potentially providing a nontoxic alternative to ThS for FTI detection.
Curcumin was consistent in detecting AT8-positive tau fibrils. However, it is possible that by using AT8 as the reference, we actually underestimated the specificity of CCM for FTIs because of loss of AT8 immunoreactivity (IR) in advanced stages of FTI evolution. A discrepancy between AT8-IR and fluorescent probe labeling similar to the one between AT8-IR and CCM fluorescence in the present study was previously reported for thiazine red (32). The maturation of FTIs through the disease progression has been reported in both AD (9,32-34) and PSP (8). In the first stage of FTI maturation (most often referred to as pretangles or pre-FTIs), pathological accumulation of hyperphosphorylated tau begins in the cytoplasm. At this stage, hyperphosphorylated tau is still predominantly nonfibrillar. The intermediate stage is represented by mature FTIs or NFTs, in which the hyperphosphorylated tau is aggregated in intracellular, usually tightly packed, basophilic fibrils. So-called ghost tangles, that is, extracellular loose tau fibrils displaying reduced basophilia, are the purported final stage of FTI development when the cell has been destroyed. Thus, during the progress of FTI maturation, tau fibrils are being formed, accumulate, and may be continuously modified. The AT8 immunohistochemistry and Gallyas-Braak staining of different stages of FTI in PSP were studied by Jin et al (8). They suggested that the inability of immunohistochemistry to detect ghost tangles was caused by the loss of specific epitopes through chemical modification of fibrils during the progression of the disease. Changes of epitopes during the development of FTIs were also observed in AD (9, 33).
These stages may have been reflected in the present study. First, basophilic material reflecting the form of the cell but without an observable fibrillar structure in H&E staining (i.e. tentative pretangles) was AT8 positive, but it was not extensively labeled by either CCM or ThS. Second, mature FTIs were observed in H&E staining as basophilic with a pronounced fibrillar structure with or without a visible nucleus. They were labeled by all of the detection methods used in the study. Third, the extracellular ghost FTIs, recognizable in H&E staining by their loose fibrillar structure, were consistently labeled with CCM and ThS but were often not visualized with AT8. These FTIs would therefore be considered false-positives because we used AT8 as a reference. However, they were consistently visible in ThS fluorescence and also in Gallyas silver staining, indicating that CCM was indeed labeling ghost FTIs.
Our results differ somewhat from those of Augustinack et al (33) who reported progressively increasing IR of extraneuronal ghost tangles with AT8. Our findings seem to be more in agreement with Bobinski et al who analyzed NFT formation and maturation in detail and saw a diverse picture of ghost tangle IR with the specific anti-PHF antibody, ranging from strongly immunoreactive compact clusters to very weakly or non-immunoreactive loosely arranged fibrils (9). The latter would likely be missed in the absence of another reference method for FTI detection (e.g. Gallyas silver staining), which may explain why Augustinack et al (33) described a "progressive increase in AT8-IR with tangle maturation" but no decline in AT8-IR with their extracellular modification at later stages when they display fibril dispersion and pronounced reduction in AT8-IR. We selected loci for analysis based on H&E staining, in which most of the ghost tangles were not immediately visible to the nonpathologist observers in our study (Nina Mohorko, Grega Repovš, Mara Bresjanac) and were occasionally recognized in retrospect based on strong CCM, ThS, and Gallyas labeling and fibrillar morphology. This enabled us to analyze a large number of ghost tangles in a nonbiased fashion. In this context, CCM demonstrated consistent reliability as a detection tool for pronouncedly filamentous mature FTIs and ghost tangles.
Detecting FTIs in tissue sections also presents a challenge because of the structural differences between FTIs among different tauopathies; the FTIs differ both in tau isoform composition and in type of tau filament present (10, 11). Different silver staining procedures can differentiate among the FTIs from different tauopathies. For example, Gallyas staining labels FTIs in AD and PSP but not in PiD (35, 36), whereas the Campbell-Switzer method labels FTIs in AD and PiD but not in PSP (35, 36). The authors of the silver staining studies suggested that Gallyas silver staining might preferentially label 4R-tau deposits, whereas the Campbell-Switzer method would be 3R-tau selective (36).
We observed a similar phenomenon using CCM, which reliably and reproducibly labeled FTIs in AD and PSP brain sections but not Pick bodies or Pick cells in PiD brain sections. This was in full agreement with the ThS and Gallyas silver labeling of the same structures. To explore whether the differential CCM labeling reflects different isoform compositions, we used a colocalization analysis using 2 isoform-specific antibodies. The 4R-tau IR colocalized well with CCM and ThS signal in both AD and PSP, even displaying similar overall signal colocalization patterns (Figs. 5F, I). The 3R-tau-positive Pick bodies in PiD were CCM and ThS negative. However, the 3R-tau-immunoreactive FTIs from AD were CCM positive and ThS positive, calling for an explanation of the discrepancy between 3R-tau overlap with CCM in AD and its absence in PiD. One possible explanation is that the spatial resolution of our colocalization analysis did not allow us to distinguish the 2 isoforms within the same FTI to allow for identification of the exact molecular target of CCM labeling. Thus, if CCM actually labeled 4R-tau within AD FTIs, the close proximity of 3R-tau and 4R-tau epitopes could yield a colocalization outcome suggesting CCM labeling of 3R-tau. Another possibility is that the ultrastructural characteristics rather than the isoform composition provide the basis for differential CCM (as well as ThS and Gallyas) labeling of FTIs.
Ultrastructural characteristics of various FTIs depend on the conformation of the tau molecules stacked in the filaments that form them. This issue of ultrastructural rather than isoform composition dependence of labeling has been difficult to address and awaits experimental clarification. Although there are reports of β structure-rich cores of PHFs (37), other authors claim that PHFs actually have an α-helix-rich core (38). In addition to PHFs, tau also forms SFs and random coiled filaments. In contrast to AD NFTs, with their characteristic PHF ultrastructure, Pick bodies seem to be composed of randomly distributed SFs and of PHF-like filaments with long periodicity (39, 40), both of which may lack the appropriate ultrastructural motif to bind CCM. An earlier study found this a plausible explanation for a similar result obtained in PiD with another molecular imaging probe, FDDNP (41). This view is supported by the work of Uchihara et al (35, 36), who found that Gallyas silver staining does not label in PiD, although it labels FTIs in AD and PSP. The results of the colocalization analysis in the present study implicitly favor the structural difference explanation for discrepant 3R-tau labeling in AD and PiD. Namely, if the presence of CCM-RD3 colocalization in AD was caused by the spatial overlap of 3R-tau and 4R-tau isoforms and if CCM actually only labeled 4R-tau components of these FTIs, the pattern of CCM-RD3 signal colocalization would be nearly indistinguishable from the one obtained in CCM-ET3 signal colocalization analysis of the adjacent AD brain sections; however, this was not the case. Indeed, CCM-RD3 colocalization in AD differed from the CCM-ET3 colocalization pattern not only by the absence of 1 type of FTI (bottom left quadrant cluster in Fig. 5F), but also by the presence of another (upper left quadrant cluster in Fig. 5F), suggesting that CCM indeed labels the 3R-tau isoform-containing structures in AD. Additional in vitro experiments are needed to determine the ultrastructural requirements for CCM binding to tau fibrils of variable isoform composition.
A possible structure-related explanation for differential labeling of FTIs by CCM could hinge on structural requirements for CCM fluorescence, which include regular periodicity and optimal spatial distribution of its binding sites on fibrils. It had been shown that CCM fluorescence increases when it binds to the prion protein amyloid fibrils in an ordered arrangement (20). An objective expression of the underlying structural order of protein deposits in tissue sections is their birefringence enhanced by Congo red staining. Pick bodies are composed of filaments mixed with amorphous material and lying in disarray. This is believed to be the reason that Pick bodies do not display birefringence in polarized optics, despite their filamentous structure (41). Thus, a correlation analysis of CCM labeling and Congo red-enhanced birefringence of the FTIs could provide the missing information.
The finding that CCM labels FTIs in AD and PSP suggests that CCM has the potential for diagnosing tauopathies in vivo. Radiolabeled CCM analogues have already been synthesized and examined for Aβ plaque positron emission tomography imaging in AD (42). Apart from favorable brain penetration caused by its lipophilic nature (23), CCM and its analogues might make ideal positron emission tomography probes, particularly because it has been a component of the human diet for thousands of years and is well tolerated. Finally, it has been suggested that CCM-rich diet tentatively improves cognitive performance in elderly Asians (43). In that context, it may be relevant that CCM has been shown to interfere with Aβ (19, 23), PrPSc (24), and α-synuclein (25) aggregation. Based on labeling characteristics, CCM binding to AD and PSP FTIs might have a potential to also directly interfere with tau accumulation. Studies to address this possibility may well be worthwhile.
The authors thank Mateja Drolec Novak and Marija Zupančič for excellent technical assistance, and especially Dr Lojze M. Smid for valuable contribution toward the pilot study protocol.
1. Braak H, Braak E. Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 1991;82:239-59
2. Goedert M, Spillantini M, Potier M, et al. Cloning and sequencing of the cDNA encoding an isoform of microtubule-associated protein tau containing four tandem repeats: Differential expression of tau protein mRNAs in human brain. EMBO J 1989;8:393-99
3. Bunker J, Wilson L, Jordan M, et al. Modulation of microtubule dynamics by tau in living cells: Implications for development and neurodegeneration. Mol Biol Cell 2004;15:2720-28
4. Levy S, Leboeuf A, Massie M, et al. Three- and four-repeat tau regulate the dynamic instability of two distinct microtubule subpopulations in qualitatively different manners. Implications for neurodegeneration. J Biol Chem 2005;280:13520-28
5. Maas T, Eidenmüller J, Brandt R. Interaction of tau with the neural membrane cortex is regulated by phosphorylation at sites that are modified in paired helical filaments. J Biol Chem 2000;275:15733-40
6. Hatcher H, Planalp R, Cho J, et al. Curcumin: From ancient medicine to current clinical trials. Cell Mol Life Sci 2008;65:1631-52
7. Hernández F, Avila J. Tauopathies. Cell Mol Life Sci 2007;64:2219-33
8. Jin C, Katayama S, Hiji M, et al. Relationship between neuronal loss and tangle formation in neurons and oligodendroglia in progressive supranuclear palsy. Neuropathology 2006;26:50-56
9. Bobinski M, Wegiel J, Tarnawski M, et al. Duration of neurofibrillary changes in the hippocampal pyramidal neurons. Brain Res 1998;799:156-58
10. Delacourte A. Tauopathies: Recent insights into old diseases. Folia Neuropathol 2005;43:244-57
11. Sergeant N, Delacourte A, Buée L. Tau protein as a differential biomarker of tauopathies. Biochim Biophys Acta 2005;1739:179-97
12. Cairns NJ, Bigio EH, Mackenzie IR, et al. Consortium for frontotemporal lobar degeneration. neuropathologic diagnostic and nosologic criteria for frontotemporal lobar degeneration: Consensus of the consortium for frontotemporal lobar degeneration. Acta Neuropathol 2007;114:5-22
13. Boeve B. A review of the non-Alzheimer dementias. J Clin Psychiatry 2006;67:1985-2001
14. Alafuzoff I, Arzberger T, Al-Sarraj S, et al. Staging of neurofibrillary pathology in Alzheimer's disease: A study of the BrainNet Europe Consortium. Brain Pathol 2008;18:484-96
15. Braak H, Alafuzoff I, Arzberger T, et al. Staging of Alzheimer disease-associated neurofibrillary pathology using paraffin sections and immunocytochemistry. Acta Neuropathol 2006;112:389-404
16. Sun A, Nguyen X, Bing G. Comparative analysis of an improved thioflavin-s stain, Gallyas silver stain, and immunohistochemistry for neurofibrillary tangle demonstration on the same sections. J Histochem Cytochem 2002;50:463-72
17. Agdeppa E, Kepe V, Petrič A, et al. In vitro detection of (S)-naproxen and ibuprofen binding to plaques in the Alzheimer's brain using the positron emission tomography molecular imaging probe 2-(1-[6-[(2-[(18)F]fluoroethyl)(methyl)amino]-2-naphthyl]ethylidene)malononitrile. Neuroscience 2003;117:723-30
18. Bresjanac M, Smid L, Vovko T, et al. Molecular-imaging probe 2-(1-[6-[(2-fluoroethyl)(methyl) amino]-2-naphthyl]ethylidene)malononitrile labels prion plaques in vitro. J Neurosci 2003;23:8029-33
19. Garcia-Alloza M, Borrelli L, Rozkalne A, et al. Curcumin labels amyloid pathology in vivo, disrupts existing plaques, and partially restores distorted neurites in an Alzheimer mouse model. J Neurochem 2007;102:1095-104
20. Hafner-Bratkovic I, Gaspersic J, Smid L, et al. Curcumin binds to the alpha-helical intermediate and to the amyloid form of prion protein-a new mechanism for the inhibition of PrP(Sc) accumulation. J Neurochem 2008;104:1553-64
21. Wang F, Wu X, Liu S, et al. The sensitive fluorometric method for the determination of curcumin using the enhancement of mixed micelle. J Fluoresc 2006;16:53-59
22. Barik A, Priyadarsini K, Mohan H. Photophysical studies on binding of curcumin to bovine serum albumins. Photochem Photobiol 2003;77:597-603
23. Yang F, Lim G, Begum A, et al. Curcumin inhibits formation of amyloid beta oligomers and fibrils, binds plaques, and reduces amyloid in vivo. J Biol Chem 2005;280:5892-901
24. Caughey B, Raymond L, Raymond G, et al. Inhibition of protease-resistant prion protein accumulation in vitro by curcumin. J Virol 2003;77:5499-502
25. Pandey N, Strider J, Nolan W, et al. Curcumin inhibits aggregation of alpha-synuclein. Acta Neuropathol 2008;115:479-89
26. Uchihara T. Silver diagnosis in neuropathology: Principles, practice and revised interpretation. Acta Neuropathol 2007;113:483-99
27. de Silva R, Lashley T, Gibb G, et al. Pathological inclusion bodies in tauopathies contain distinct complements of tau with three or four microtubule-binding repeat domains as demonstrated by new specific monoclonal antibodies. Neuropathol Appl Neurobiol 2003;29:288-302
28. Hogg M, Grujic Z, Baker M, et al. The L266V tau mutation is associated with frontotemporal dementia and Pick-like 3R and 4R tauopathy. Acta Neuropathol 2003;106:323-36
29. Kepe V, Ghetti B, Farlow MR, et al. PET of brain prion amyloid in Gerstmann-Straussler-Sheinker disease. Brain Pathology June 9, 2009 [epub ahead of print]
30. Rossner M, Yamada K. What's in a picture? The temptation of image manipulation. J Cell Biol 2004;166:11-15
31. Manders E, Verbeek F, Aten J. Measurement of co-localization of objects in dual-colour confocal images. J Microscopy 1993;169:375-82
32. Uchihara T, Nakamura A, Yamazaki M, et al. Different conformation of neuronal tau deposits distinguished by double immunofluorescence with AT8 and thiazin red combined with Gallyas method. Acta Neuropathol 2001;102:462-66
33. Augustinack J, Schneider A, Mandelkow E, et al. Specific tau phosphorylation sites correlate with severity of neuronal cytopathology in Alzheimer's disease. Acta Neuropathol 2002;103:26-35
34. Schmidt M, Gur R, Gur R, et al. Intraneuronal and extracellular neurofibrillary tangles exhibit mutually exclusive cytoskeletal antigens. Ann Neurol 1988;23:184-89
35. Uchihara T, Shibuya K, Nakamura A, et al. Silver stains distinguish tau-positive structures in corticobasal degeneration/progressive supranuclear palsy and in Alzheimer's disease-comparison between Gallyas and Campbell-Switzer methods. Acta Neuropathol 2005;109:299-305
36. Uchihara T, Tsuchiya K, Nakamura A, et al. Silver staining profiles distinguish Pick bodies from neurofibrillary tangles of Alzheimer type: Comparison between Gallyas and Campbell-Switzer methods. Acta Neuropathol 2005;109:483-89
37. von Bergen M, Barghorn S, Biernat J, et al. Tau aggregation is driven by a transition from random coil to beta sheet structure. Biochim Biophys Acta 2005;1739:158-66
38. Sadqi M, Hernández F, Pan U, et al. Alpha-helix structure in Alzheimer's disease aggregates of tau-protein. Biochemistry 2002;41:7150-55
39. Kato S, Nakamura H. Presence of two different fibril subtypes in the Pick body: An immunoelectron microscopic study. Acta Neuropathol 1990;81:125-29
40. King M, Ghoshal N, Wall J, et al. Structural analysis of Pick's disease-derived and in vitro-assembled tau filaments. Am J Pathol 2001;158:1481-90
41. Smid L, Vovko T, Popovic M, et al. The 2,6-disubstituted naphthalene derivative FDDNP labeling reliably predicts Congo red birefringence of protein deposits in brain sections of selected human neurodegenerative diseases. Brain Pathol 2006;16:124-30
42. Ryu E, Choe Y, Lee K, et al. Curcumin and dehydrozingerone derivatives: Synthesis, radiolabeling, and evaluation for beta-amyloid plaque imaging. J Med Chem 2006;49:6111-19
43. Ng T, Chiam P, Lee T, et al. Curry consumption and cognitive function in the elderly. Am J Epidemiol 2006;164:898-906