Seeding Activity of Skin Misfolded Tau as a Biomarker for Tauopathies

Background Tauopathies are a group of age-related neurodegenerative diseases characterized by the accumulation of pathologically phosphorylated tau protein in the brain, leading to prion-like propagation and aggregation. They include Alzheimer’s disease (AD), progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), and Pick’s disease (PiD). Currently, reliable diagnostic biomarkers that directly reflect the capability of propagation and spreading of misfolded tau aggregates in peripheral tissues and body fluids are lacking. Methods We utilized the seed-amplification assay (SAA) employing ultrasensitive real-time quaking-induced conversion (RT-QuIC) to assess the prion-like seeding activity of pathological tau in the skin of cadavers with neuropathologically confirmed tauopathies, including AD, PSP, CBD, and PiD, compared to normal controls. Results We found that the skin prion-SAA demonstrated a significantly higher sensitivity (75–80%) and specificity (95–100%) for detecting tauopathy, depending on the tau substrates used. Moreover, increased tau-seeding activity was also observed in biopsy skin samples from living AD and PSP patients examined. Analysis of the end products of skin-tau SAA confirmed that the increased seeding activity was accompanied by the formation of tau aggregates with different physicochemical properties related to two different tau substrates used. Conclusions Overall, our study provides proof-of-concept that the skin tau-SAA can differentiate tauopathies from normal controls, suggesting that the seeding activity of misfolded tau in the skin could serve as a diagnostic biomarker for tauopathies.


INTRODUCTION
The deposition of disease-associated tau aggregates in the brain is a characteristic feature of tauopathies, including Alzheimer's disease (AD), progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), and Pick's disease (PiD) [1].These conditions share a common pathogenesis involving the seeding and propagation of misfolded and phosphorylated isoforms of the tau protein in the brain, reminiscent of the infectious prion protein (PrP Sc or prion) in prion diseases (PrD) [2].The human brain expresses six tau isoforms, resulting from the combination of three or four microtubule-binding repeats (3R or 4R tau) and 0-2 N-terminal inserts (0N, 1N, or 2N tau) [3].Different tauopathies exhibit variations in the composition of these tau isoforms, with AD displaying both 3R and 4R isoforms, PiD primarily containing 3R isoforms, and PSP and CBD characterized by the accumulation of 4R tau assemblies.These differences in tau isoform composition are proposed to be associated with the presence of distinct strains of neurotoxic tau aggregate conformers [4].Currently, the de nitive diagnosis of tauopathies relies on the availability of brain tissue obtained via biopsy or autopsy for the detection of tau pathology and phosphorylated tau.However, these methods are highly invasive or often too late in the disease process.Recent advancements in brain molecular imaging and highly sensitive immunoassays of phosphorylated and total tau (p-tau and t-tau) in plasma and cerebrospinal uid (CSF) have enabled early and reliable diagnosis of AD in living patients [5], but they have limitations such as invasiveness or high cost.Newly developed single molecular immunoassay (Simoa) is highly sensitive and speci c in detection of the blood biomarkers, but Simoa is unaffordable to the majority of patients, especially those in the developing countries.
RT-QuIC has also been reported to detect tau-SA in the brain tissues of deceased individuals with tauopathies [16][17][18][19][20].However, there have not been any reports on the application of tau-SAA in the body uids and peripheral tissues of patients with AD to date although there was a report about tau RT-QuIC detection of CSF from PSP and CBD [19].
In the current study, we investigated tau seeding activity in autopsy scalp skin samples from individuals with neuropathologically con rmed AD, PSP, CBD, PiD, and normal controls using RT-QuIC with truncated human tau fragments as substrates.We found that misfolded tau from AD and other tauopathies like PSP and CBD, but not from normal controls and PiD, selectively seeded the 4RCF substrate (equivalent to the human 4R tau fragment, K18CFh) [16,20] with a sensitivity of 80.5% and a speci city of 95.4%.Similarly, the 3RCF substrate (equivalent to the human 3R tau fragment, K19CFh) [18,20]showed a sensitivity of 77% and a speci city of 92%.Biopsied skin samples from individuals with AD, PSP, and controls also exhibited high diagnostic e cacy.Furthermore, the study demonstrated that skin-derived misfolded tau aggregates from AD, containing a mix of 4R and 3R tau, and from PSP and CBD, with solely 4R tau, could be successfully ampli ed using either 4RCF or 3RCF substrates.In contrast, skin misfolded tau aggregates from PiD, exclusively composed of 3R tau, showed ampli cation only with the 3RCF substrate, not with 4RCF.This study represents the rst application of SAA methodology to both post-mortem and biopsied cutaneous specimens and highlights the potential of skin tau-SAA as a novel biomarker for the diagnosis of tauopathies.

Phosphorylated-tau is detectable in skin samples of participants with tauopathies
We rst determined whether the pathologically phosphorylated tau is detectable in the skin samples of participants with tauopathies by western blotting.Skin homogenates from deceased individuals with AD (n = 5), PSP (n = 4), CBD (n = 4), PiD (n = 4) and NC (n = 4) were examined by western blotting with antibodies directed against phosphorylated tau (Anti-pT231 and Anti-pS396).P-tau231 and p-tau396 epitopes were previously identi ed from a systematic p-tau antibody screening in differential detection with AD and control brains [21].
Both Anti-pT231 (Fig. S1A) and Anti-pS396 (Fig. S1b) antibodies revealed protein bands migrating at between approximately 25 and 100 kDa in all AD, 1 of 4 PSP, 2 of 4 CBD, and 1 of 4 PiD skin samples (Fig. S1).These bands are expected to represent truncated and full-length monomers as well as oligomers of phosphorylated tau.
In contrast, normal controls mainly displayed a band migrating at about 100 kDa without low molecular weight bands.The levels of phosphorylated tau in the skin of AD participants were signi cantly elevated compared to those of other tauopathies and normal controls (Fig. S1D, 1E), as evidenced by the quantitative analysis after normalized each sample by corresponding ß-actin migrating at around 50 kDa as a loading control (Fig. S1C).To determine the speci city of the protein bands and exclude the possibility that the bands might be from the nonspeci c reaction of the secondary antibody, we probed the blots with the secondary antibody only in the absence of the primary antibody (Fig. S2).The absence of detectable bands following a long ECL exposure (90 minutes) suggested that the bands identi ed by phospho-tau antibodies were indeed speci c.The results revealed that the levels of the skin phosphorylated tau were signi cantly greater in AD than in other tauopathies and controls.
Moreover, to verify that the bands identi ed are indeed tau proteins, we employed two distinct anti-tau antibodies: tau 46 (Invitrogen, MA), which binds to tau residues spanning 404 to 411 (Fig. S3A), and 43D (Biolegend, CA), targeting tau residues 1 to 100 (Fig. S3B).These antibodies, designed to detect total tau in samples, speci cally recognized the protein bands in the same collection of skin samples.Notably, the pattern of these bands differed from those identi ed by phosphorylated tau antibodies, further substantiating that the tau proteins we detected are indeed speci c.
Skin tau-seeding activity can speci cally differentiate AD and non-AD tauopathies from normal controls and PiD cases using 4RCF as the substrate In contrast to PrP in PrD and α-synuclein in PD, the tau molecule in the human brain exhibits 6 isoforms, of which there are 3 isoforms with 4 microtubule-binding repeats (4R tau) and 3 with three repeats (3R tau), resulting from alternative mRNA splicing.In order to nd an appropriate tau substrate for the seed-ampli cation assay (SAA) of skin tau (sTau-SAA) with RT-QuIC, based on our recent nding with autopsy brain tissues from cadavers with AD and other non-AD tauopathies [20], we rst examined 4 types of tau substrates including 2 full-length tau isoforms (2N3R/2N4R) and 2 truncated tau fragments consisting of 4RCF (cysteine-free, equivalent to K18CFh)/3RCF (equivalent to K19CFh) using autopsy skin samples from AD cadavers.4RCF represents the aggregation-prone core sequences of the 4R tau isoforms [20,21].Correspondingly, 3RCF fragment represents the aggregation-prone core sequences of the 3R tau isoforms.The only difference between 4RCF and 3RCF is that 3RCF lacks R2 region.Compared to the full-length tau, the truncated tau substrates-based RT-QuIC revealed higher endpoint uorescence readings (exceeding 100,000 RFU) and shorter lag phases (less than 20 hours) (Fig. S4).
Then we compared tau-SA of brain and skin samples from AD and non-AD controls using two truncated tau fragments as substrates.AD skin samples exhibited a slightly extended lag phase compared to AD brain samples when assessed with both 4RCF (Fig. S6A) and 3RCF (Fig. S6B) tau substrates.Furthermore, both brain and skin samples from non-AD controls demonstrated signi cantly weaker ThT uorescence intensity compared to the AD samples when analyzed with the two types of tau substrates.As a result, we used the two truncated tau fragments as the substrates for the following examinations in our study.
Using RT-QuIC with 4RCF truncated tau as the substrate, we analyzed autopsy skin samples from AD (n = 46), CBD (n = 5), PSP (n = 33), PiD (n = 6), and NC (n = 43).4RCF-based RT-QuIC of skin tau-seeding activity (sTau-SA) exhibited that CBD cases had the highest ThT uorescence intensity, followed by AD, PSP, and PiD (Fig. 1A, B).We also compared lag phases of sTau-SA in each type of tauopathy, the time period between the beginning of the RT-QuIC measurement and the time point at which the curves re ecting the ThT uorescence started to increase in their ThT uorescence.Consistent with the endpoint ThT values, CBD displayed the shortest lag phase, while PiD was the longest (Fig. 1C).Our 4RCF-based RT-QuIC assay of the skin samples from AD and non-AD tauopathies yielded a sensitivity of 80.49% and a speci city of 95.35%.To quantitate the sTau-SA of each type of tauopathy, we conducted endpoint titration of RT-QuIC assay of each skin sample from different tauopathies.The x-axis represents SD50 that referred to the 50% seeding activity.PSP samples exhibited the highest sTau-SA, followed by AD samples.In contrast, PiD skin samples displayed markedly lower tau-SA, as shown in Fig. 1D.This observed variation is likely attributable to the incompatibility between seed and substrate, particularly noting that PiDderived misfolded tau is uniquely comprised of 3R tau, absent of 4R tau components.AUC analysis revealed an area value of 0.88 based on the comparison between AD and normal controls and 0.79 based on the comparison between tauopathies and normal controls (Fig. 1E, F).
The sTau-SA of AD and all tauopathies can be speci cally differentiated from that of normal controls using 3RCF as the substrate We next used 3RCF (K19)-based RT-QuIC assays to examine the same samples detected in 4RCF studies.The sTau-SA was signi cantly higher in tauopathies than in non-tauopathies, of which PiD was the highest, followed by AD, PSP, and CBD (Fig. 2A, B).We also compared their lag phases, which were inversely proportional to those of 4R tau: PiD exhibited the shortest lag phase, while CBD showed the longest lag phase (Fig. 2C).PiD and PSP demonstrated the highest seeding dose while CBD had the lowest one (Fig. 2D).In addition, the 3RCF-based skin tau RT-QuIC assay generated a sensitivity of 75% and a speci city of 100%, which was similar to that of 4RCFbased RT-QuIC in general.The AUC values (0.85 vs 0.79) of 3RCF-based RT-QuIC assay of skin tau were similar to those detected with 4RCF-based RT-QuIC shown above (Fig. 2E, F).Notably, in contrast to the 4RCF-based sTau-SA, 3RCF-based sTau-SA from PiD was the highest, in addition to differentiating other tauopathies from the normal controls.
The sTau-SA is signi cantly elevated over an increase in the Braak staging Based on Aβ/tau-pathology and Aβ/tau-positron emission tomography (PET) scan, the accumulation of Aβ/tau leading to clinical AD is a continuum process.To determine whether the skin tau-SA can re ect the severity or Braak staging in the AD brain, next we associated skin tau-SA with the Braak staging in autopsy brain tissues examined.Notably, when sTau-SA with both 4RCF and 3RCF served as a function of Braak staging, there was a clear association observed: the sTau-SA was signi cantly elevated upon an increase in the Braak staging (Fig. 3).Speci cally, with the exception of stages I and II, the overall ThT intensity for 4RCF-based skin tau-SA was increased following advancing of the Braak stages, although no signi cant difference was noted between stages V and VI (Fig. 3A).A similar trend was observed for 3RCF, where the overall ThT intensity increased with the progression of Braak stage (Fig. 3B), yet no signi cant difference was found between stages V and VI or between stages III and IV.
We also explored whether variables such as age, gender, and post-mortem interval (PMI) could in uence Tau-SA.
The sTau-SA is signi cantly higher in PD and dementia with Lewy bodies than in multiple system atrophy and normal controls but it is still lower than that in AD Accumulation of tau aggregates in the brain has been observed in some of cases with synucleinopathies including PD, dementia with Lewy bodies (DLB), and multiple system atrophy (MSA) (22)(23)(24)(25)(26). Next, we further explored whether tau-SA can be detected in the skin of synucleinopathies by our sTau-SAA.Autopsy skin samples from AD (n = 21), PD (n = 10), MSA (n = 6), DLB (n = 6), and NC (n = 17) were examined by 4RCF-based tau-SAA.
The ThT endpoint uorescence intensity of sTau-SAA was dramatically higher in AD than in synucleinopathies, whereas the skin-tau uorescence intensity was also signi cantly increased in synucleinopathies except for MSA than in the control group (Fig. 4), consistent with the previous observations that some of cases with synucleinopathies can have tau-pathology [10,19,37,44,48].To determine the presence of tau pathology in skin samples from individuals with synucleinopathies, we performed western blotting on representative skin samples from PD (n = 7), DLB (n = 5), and MSA cases (n = 5).These samples were analyzed using anti-phospho-tau antibodies Anti-pT231 (Fig. S7A) and Anti-pS396 (Fig. S7B).We observed that 4 out of the 5 DLB samples exhibited positive bands in the range of 25-50 kDa.Notably, these 4 samples also had coexisting AD pathology as indicated in Table 1.In MSA samples, positive bands were also detected migrating at 25-50 kDa on the gels, although none of these cases had co-morbidities with con rmed tauopathies.However, tau tangles and plaques were detectable in their brain tissues (as shown in Table 1).Regarding the PD skin samples, 4 out of 7 cases displayed smeared bands that were indicative of positive phospho-tau detected by the two antibodies.Biopsy sTau-SA is signi cantly higher in tauopathies than in normal controls We then used the above 4RCF-or 3RCF-based SAA (4RCF-or 3RCF-SAA) to examine biopsied skin samples from AD (n = 16), PSP (n = 8) and NCs (n = 10).Skin 4RCF-SA was signi cantly higher in AD than in normal controls [83091 ± 57912 (mean ± SD) vs 20490 ± 9307, p = 0.0026 < 0.005]; skin 4RCF-SA was also signi cantly greater in PSP than in normal controls (73360 ± 49625 vs 20490 ± 9307, p = 0.0043 < 0.005) (Fig. 5A).Similar to skin 4RCF-SA, 3RCF-SA was signi cantly higher in AD than in NCs [97511 ± 54115 vs 30090 ± 13657, p = 0.0008 < 0.001] and greater in PSP than in NCs (60449 ± 20492 vs 30090 ± 13657, p = 0.0017 < 0.005) (Fig. 5B).There were no signi cant differences in sTau-SA between AD and PSP cases with both substrates (Fig. 5).This observation implied the potential for sTau-SA to serve as an antemortem diagnostic biomarker to differentiate tauopathies from normal controls.The individual clinical data are listed in Tables 2 and 3. To determine whether ThT uorescence levels re ecting sTau-SA represent the formation of skin tau-seeded aggregates, we correlated the end-point ThT uorescence levels with the dot-blot intensity of tau aggregates captured by a lter-trap assay (FTA) (Fig. 6).After obtaining the end point ThT uorescence levels of sTau-SA of 4 cases each from AD, PSP, CBD, PiD and NC with 4RCF or 3RCF as the substrate (Fig. 6A, B), we then ran FTA with their corresponding end products, followed by probing the dot-blots with anti-4R tau antibody (RD4) (Fig. 6C) and anti-3R antibody (RD3) (Fig. 6D).The semiquantitative densitometric scanning of protein dot intensity on the dotblots revealed that similar to ThT uorescence levels, the intensity of tau aggregates captured by FTA on the blots was signi cantly higher in tauopathies than in normal controls by both RD4 (Fig. 6E) and RD3 antibodies (Fig. 6F).Correlation analyses demonstrated that the intensity of the trapped aggregates from the end products correlated positively with the ThT uorescence levels (r = 0.86 for 4R, r = 0.68 for 3R) (Fig. 6G, H).

Table 2 Demographic and clinical features of biopsied cases in different groups
Transmission electron microscopy of 4RCF-or 3RCF-based RT-QuIC end products displays the proto bril-like structures While our FTA apparently was able to detect captured tau aggregates, to further determine the morphology of the ampli ed skin tau aggregates we performed transmission electron microscopy (TEM) of the RT-QuIC end-products of skin misfolded tau from 3 cases each of AD, PSP, CBD, PiD and normal controls (NC) with either 4RCF or 3RCF as the substrate (Fig. 7).For 4RCF, TEM revealed that except for PiD and NC (Fig. 7E), other end-products had a small number of proto bril-like structures.For 3RCF, while NC showed only oligomer-like structure (Fig. 7F), proto brils were detectable in the end-product of tau-RT-QuIC of cases with all tauopathies by TEM (Fig. 7G through 7J).
The end products of 4RCF-and 3RCF-SAA exhibit different patterns of resistance to proteinase K digestion The pattern of protein aggregates from the skin tau RT-QuIC end products to proteinase K (PK) digestion has been widely believed to re ect the conformational properties of misfolded proteins examined.Since the small fragments of 3RCF tau lower than 7 kDa were more di cult to digest by PK, we next performed the titration of varied PK concentrations ranging from 0, 1.25 µg/mL, 2.5 µg/mL, 3.75 µg/mL, 5 µg/mL, 7.5 µg/mL, to 10 µg/mL for the 4R (Fig. 8A) and 0, 1.25 µg/mL, 5 µg/mL, 10 µg/mL, 12.5 µg/mL and 25 µg/ml for the 3R tau (Fig. 8B) SAA end products of skin tau from AD and non-AD subjects.Without PK treatment, the end products of 4RCF-based RT-QuIC of skin tau from non-AD exhibited 3 protein bands by the RD4 tau antibody, migrating at approximately 25-26 kDa, 12-14 kDa, and 7-10 kDa (Fig. 8A).The short exposure of our blots showed that the 7-10 kDa bands actually consisted of 7 kDa and 10 kDa proteins.As a result, the above four bands could represent the trimer and dimer of a 7 kDa band as well as the monomers of a full-length 4RCF (~ 12 kDa) and a truncated 4RCF (~ 7-10 kDa), respectively.Of them, the two lower monomeric bands were predominant, whereas the top dimeric and trimeric tau bands were underrepresented, accounting for less than 1-2% of total tau (Fig. 8A, C).In contrast, the end product from AD skin samples without PK-treatment exhibited an additional band migrating at approximately 48-50 kDa in addition to the four bands found in the end product of non-AD skin described above.This band could be oligomers of full-length or truncated tau molecules.In addition, the intensity of truncated trimer and dimer of tau migrating at 24-27 kDa was signi cantly increased compared to that of non-AD end product.
Upon PK-treatment, for the non-AD end products, the intensity of the two lower monomeric tau bands was signi cantly decreased while they were still detected at PK of 10 µg/mL (Fig. 8A, C).The intensity of the tau band migrating at ~ 25-27 kDa was increased rst up to PK of 2.5 µg/mL and then decreased, until became undetectable at PK of 10 µg/mL.The band migrating at ~ 12 kDa seemed to be completely PK-sensitive and no band was detectable even at the lowest PK concentration at 1.25 µg/mL.In contrast, after PK-treatment the end product of the RT-QuIC with AD skin samples showed the decreased intensity of the tau band migrating at ~ 25-27 kDa but generated additional smaller band migrating at about 24 kDa (Fig. 8A, D, red arrow).This band was most likely derived from the truncation of the 25-27 kDa band since it was generated and increased over the increase in the PK concentration.In contrast with the non-AD end products, AD end product also exhibited an additional band between 7 kDa and 10 kDa bands migrating at about 8 kDa in the PK-treated AD skin end products (marked with the red arrow in Fig. 8A).The intensity of this band was similar to that of 7 kDa and 10 kDa bands and showed no changes upon the increase in the PK concentration (Fig. 8A, D).
Regarding the end product of sTau-SAA using 3RCF as the substrate, without PK-treatment, the AD and non-AD samples all mainly exhibited 3 bands migrating at about 7 kDa, 10-11 kDa, and 22-23 kDa on the gel (Fig. 8B).
According to the sequence of the 3RCF molecule, the molecular weight of the monomeric 3RCF should be 10.5 kDa.Therefore, the 7 kDa band could be a truncated fragment of 3FCF while 22-23 kDa band could be a dimer of 3RCF.Since we got high intensity of the low molecular weight bands (~ 7 kDa), we increased the PK concentration to 25 µg/mL for 3R and decreased the loading amounts of samples (Fig. 8B, E).The intensity of the 3 bands from non-AD end products all decreased while there was a faint band emerging, migrating at approximately 5 kDa over the increase in PK concentrations.The intensity of the 3 bands from AD skin tau RT-QuIC end products was also all decreased while there were two additional bands emerging, migrating at approximately 16-18 kDa and 5-6 kDa over the increase in PK concentrations (Fig. 8B, F, red arrows).The monomers of truncated 4RCF (at ~ 12 kDa) and 3RCF (at ~ 10-11 kDa) exhibited no resistance to PK treatment.However, the dimers at higher molecular weights demonstrated increased resistance to PK, with the exception of the 4R negative end product dimers, which were digested at PK concentrations exceeding 7.5 µg/mL.Additionally, the low molecular weight bands below the monomers were more resistant to being digested within the chosen PK concentration range, especially with 4RCF.
Conformational-stability assay of skin tau aggregates ampli ed by tau-SAA We treated 4RCF (Fig. S9A) and 3RCF (Fig. S9B) tau-SAA end products with GdnHCl ranging from 0 to 3.2 M, followed by PK digestion at 10 µg/mL and quantitative analyses of GdnHCl/PK-resistant protein intensity of each treated sample.This approach is grounded on the principle that subtle differences in protein structure can be ascertained by assessing conformational stability when the protein is exposed to a denaturant such as GdnHCl at appropriate concentration ranges (23).In the absence of GdnHCl and PK, 3 tau bands migrating at 48 kDa, 25 kDa and 7-12 kDa were observed for 4RCF-based RT-QuIC end products, while 2 bands migrating at 22-23 kDa and 5-10 kDa were detected in 3RCF-based RT-QuIC end products.In contrast, both AD skin 4RCF-/3RCF-based tau RT-QuIC end products were found to have multiple or smear bands above 30 kDa (Fig. S9A, B).After GdnHCl and PK-treatment, there were virtually no PK-resistant tau bands detectable from both non-AD 4RCF/3RCF-based RT-QuIC end products.In contrast, positive skin tau RT-QuIC from AD participants with either 4RCF or 3RCF as the substrate showed PK-resistant tau fragments, especially for bands migrating at 25 kDa or lower for low concentration of GdnHCl (Fig. S9).Notably, there was an additional partially PK-resistant tau fragment migrating between 7 kDa and 5 kDa bands for 3RCF-based RT-QuIC end products, which was not detectable in the 4RCFbased skin tau RT-QuIC end products (Fig. S9).The GdnHCl concentrations required to make half of the tau end product sensitive to PK, referred to as GdnHCl1/2, were 2.6 M for positive 4R tau and 2.3 M for positive 3R tau for bands migrating at 22-25 kDa and 10 kDa, indicating that the positive 4R tau end product was approximately 1.13-fold more stable than the 3R tau end product.But, the 3RCF-based RT-QuIC end products from AD cases generated stable 7 kDa band (Fig. S9B, F).

DISCUSSION
The demand for early and precise biomarkers in Alzheimer's disease (AD) clinical practice has remained unmet.These biomarkers are essential not only for diagnosing and predicting the disease but also for facilitating patient participation in clinical trials and monitoring the effectiveness of therapeutic interventions [5].With the recent positive outcomes from the Clarity AD trial [27] and the critical juncture in developing new strategies to prevent and slow down AD progression [28], the urgency for such biomarkers has intensi ed.
The recently developed AD-speci c biomarkers utilized in clinical research have signi cantly enhanced our ability to diagnose and monitor AD pathology in living patients.These biomarkers have also provided valuable insights into the accumulation and spread of misfolded Aβ and tau aggregates in AD [29].In both the initial 2018 and updated 2023 A/T/N research frameworks by the National Institute on Aging and the Alzheimer's Association (NIA-AA), the detection of misfolded proteins, including Aβ and tau, through brain molecular imaging and body uid analysis, has been central [30,31].A recent study assessing various brain imaging modalities in monitoring cognition and predicting cognitive decline has highlighted the signi cance of neocortical tau pathology as a key factor in cognitive decline over time, with tau-PET showing superior prognostic value compared to other neuroimaging measures [32].An important advancement in the updated 2023 research framework is the inclusion of recently developed plasma biomarkers, supplementing the previously relied-upon biomarkers from cerebrospinal uid (CSF) and brain imaging [30,31].However, despite these developments, there remain limitations associated with brain imaging, CSF, and Simoa-based plasma biomarkers, as outlined in the 2023 update [31].It is also uncertain whether current biomarkers cover all relevant neuropathologies and fully re ect all aspects of pathogenesis.Moreover, the current list of AD biomarkers in the updated framework lacks markers capable of re ecting the pathological functions of misfolded proteins, such as the seeding activity of pathogenic tau or Aβ, as seen in the αSyn-SAA for PD.As a result, there is an ongoing quest for new minimally invasive or non-invasive biomarkers that can directly capture the unique pathogenic features of neurotoxic misfolded proteins.
Utilizing ultrasensitive RT-QuIC and/or PMCA techniques, our prior investigations have successfully identi ed minute quantities of misfolded proteins in skin samples obtained from individuals with PrD and PD by detecting their seeding activity, a prion-like characteristic of misfolded proteins.For example, we have illustrated that the seeding activity of PrPSc and pathogenic αSyn can be discerned in patients with Creutzfeldt-Jakob Disease (CJD), prion-infected rodents, and individuals affected by PD or other synucleinopathies, respectively [33][34][35][36][37][38]8].
These ndings have been corroborated by independent research groups [39][40][41][42][43]. Motivated by these discoveries, we have expanded our inquiry to encompass the most prevalent neurodegenerative condition, AD, and other tauopathies.
Our recent study has unveiled several groundbreaking discoveries.First, autopsy skin tissue from individuals with AD exhibited signi cantly higher levels of phosphorylated tau proteins detected by western blotting compared to non-AD tauopathies and normal controls.Moreover, the gel pro le of the detected phosphorylated tau differed between AD and control groups, including other non-AD tauopathies and normal subjects.Second, similar to tau levels in AD brains [16,19,20], autopsy skin tau-SA detected by RT-QuIC was markedly higher in tauopathies than in normal controls, suggesting the potential use of sTau-SA as a novel diagnostic biomarker for tauopathies.Third, sTau-SA was detectable in cases with PD and DLB but not in MSA, although their seeding activity was signi cantly lower than that of AD and higher than normal controls.Fourth, autopsy skin tau aggregates from all tauopathies could seed the 3RCF substrate, whereas the 4RCF substrate could be seeded by skin tau aggregates from AD, PSP, and CBD but not from PiD. Fifth, the levels of skin tau-SA appeared to be associated with the progression of Braak staging observed in the brain.Sixth, biopsy skin tissues from individuals with AD and PSP showed signi cantly higher levels of tau-SA compared to normal controls, implying the potential of sTau-SA as a diagnostic biomarker for living tauopathy patients.Seventh, analysis of RT-QuIC end products revealed that sTau-SAA with AD skin samples formed tau oligomers and aggregates, con rmed by FTA, PK-treatment, conformational-stability assays, and TEM.ThT uorescence intensity at the endpoint of the reaction correlated well with tau aggregate dot intensity by FTA.Eighth, PK-treatment of skin tau RT-QuIC end products in AD exhibited greater amounts of PK-resistant tau fragments than in normal controls.Finally, conformational-stability assays showed that skin tau could seed different strains in 4RCF and 3RCF substrates.These ndings raise several important implications regarding the role of skin tau in the diagnosis and pathogenesis of tauopathies.
Studies have demonstrated an increase in tau gene expression in the skin of aging males [44].Additionally, pathological tau deposits have been identi ed in peripheral organs such as the aorta, liver, spleen, and stomach of individuals with AD, but not in controls [45].Notably, a distinct tau isoform known as big tau has been observed in peripheral tissues of both rodents and humans [46][47][48][49].Phosphorylated tau has been detected in the skin of both AD participants and normal controls through various methods including immunohistochemistry, western blotting, and MALDI-MSI [26, [50][51][52].Notably, most of the above studies revealed that the skin tau gel pro le is different from that of brain tissues.Moreover, there have been inconsistencies in the gel pro les of skin tau observed in different studies [26,51,52].For example, while some studies reported multiple tau bands in AD skin samples, others observed a single band or different tau band patterns [26,48,[51][52][53].Our own western blotting analysis revealed a distinctive skin tau gel pro le with multiple bands in AD and other non-AD tauopathies, differing from previous observations [26 , 51, 52].These discrepancies could be attributed to variations in antibodies used, experimental conditions, and sample processing methods.This underscores the importance of standardizing experimental procedures and validating results across studies to better understand the role of skin tau in neurodegenerative diseases.
In addition to the previously reported bands at ~ 70 kDa, 55 kDa, and 45 kDa, our study revealed several additional bands migrating at approximately 110 kDa, ~ 25 kDa, and a high molecular weight band (HMWB) migrating above 110 kDa.The 110 kDa band may correspond to the big-tau isoform, as its molecular weight aligns with previous descriptions [46][47][48][49].This band was consistently detected in all cases examined, including normal controls, except for one case with PiD pathology.Notably, the HMWB was present in cases from all groups except those with AD pathology.Furthermore, our analysis highlighted a distinct gel pro le of tau bands in AD participants compared to other tauopathies and normal controls, with signi cantly elevated levels of all tau bands in AD.These ndings suggest that skin tissue provides valuable insights into the levels and patterns of tau molecules, making it a promising specimen for investigating tau's role in the pathogenesis of various tauopathies and for developing differential diagnostic tools to distinguish AD from other tauopathies and control subjects.
Our study is the rst to demonstrate that tau extracted from both autopsy and biopsy skin samples of individuals with AD and other non-AD tauopathies exhibits signi cantly higher seeding activity compared to normal controls.This suggests that sTau-SA could serve as a novel diagnostic biomarker for tauopathies.Our 4RCF-or 3RCFbased RT-QuIC assay of autopsy skin samples from AD and non-AD tauopathies achieved a sensitivity of 75-80% and a speci city of 95-100%, respectively.Tau-SA was markedly elevated in biopsy skin samples from tauopathy patients compared to controls.Furthermore, our skin tau-based RT-QuIC assay holds promise for detecting the comorbidity of AD and PD.Both skin tau-or αSyn-SA can be detected depending on the substrate used [35-38, 54-56, current study].Therefore, if a case exhibits both tau-and αSyn-pathology in the brain, our RT-QuIC assay is likely to detect both tau-and αSyn-SA in the skin, indicating comorbidity of AD and PD.However, reliable diagnosis of comorbidity requires ensuring that it does not result from cross-seeding between tau and αSyn, as tau-seeds can trigger αSyn aggregation and vice versa.Moreover, our ndings regarding sTau-SA patterns shed light on the heterogeneity of tauopathies.Skin tau from PiD, characterized by 3R-dominated tau aggregates in the brain, seeded the 3RCF but not the 4RCF substrate in the RT-QuIC assay.In contrast, skin tau from AD, which features a mixture of 3R/4R tau, and from PSP/CBD, characterized by 4R-dominated tau aggregates in the brain, seeded both 4RCF and 3RCF substrates.These observations underscore the complexity of tauopathies and highlight the potential of skin tau seeding activity as a diagnostic tool to differentiate between different tauopathy subtypes.
Our investigation of skin-tau RT-QuIC end products con rmed that the increased tau-SA was accompanied by the formation of tau aggregates, as evidenced by our FTA, PK treatment, and conformational stability assays.TEM revealed the presence of oligomers and proto brils, rather than mature brils, in the 3RCF-or 4RCF-based RT-QuIC end products of skin tau from AD cases, consistent with ndings in the existing literature [57,58].Our previous study demonstrated that recombinant 4RCF fragments can spontaneously form mature brils in vitro within approximately two weeks [20].Furthermore, analysis of PK-digested skin tau RT-QuIC end products showed the generation of additional PK-resistant tau fragments in positive RT-QuIC end products.Importantly, 3RCF-and 4RCF-based RT-QuIC end products generated different PK-resistant fragments, suggesting structural differences between the two.It will be intriguing to investigate whether the end products of tau-SAA with brain and skin samples yield similar or different structural and physicochemical features in future studies.This exploration could provide valuable insights into the pathogenic mechanisms underlying tau aggregation in both the central nervous system and peripheral tissues.

Conclusions
In summary, our research highlights the promise of utilizing skin samples as a minimally invasive and readily accessible means for diagnosing tauopathies.Detecting phosphorylated tau and tau seeding activity in skin samples has the potential to advance novel diagnostic methods for these conditions.It is imperative to conduct additional studies involving larger patient groups and re ne detection techniques to fully ascertain the clinical value of a skin-based diagnostic approach for distinguishing between various tauopathies.METHODS containing C322S mutation) was rst PCR-ampli ed of 3R repeats sequence from 2N3R tau plasmid and cloned into the same expression vector using Nde I and Xho I restriction sites, followed by site-directed mutagenesis at Cys322 site to Serine using QuikChange Site-directed mutagenesis kit (Agilent, Santa Clara, CA).4RCF construct (four microtubule-binding repeats and cysteine-free construct containing C291S and C322S mutations) was rst PCR-ampli ed of 4R repeats sequence from 2N4R tau plasmid and cloned into the same expression vector using Nde I and Xho I restriction sites, followed by site-directed mutagenesis at Cys291 and Cys322 sites using QuikChange Site-directed mutagenesis kit.All constructs were designed with a his6-tag at their carboxy-termini to facilitate protein puri cation and were veri ed by DNA sequencing.
Engineered tau fragments 3RCF and 4RCF expression and puri cation Recombinant 3RCF and 4RCF was prepared as previously described [20].In brief, plasmids encoding human tau engineered constructs 3RCF and 4RCF were transformed into BL21-DE3 E. coli cells.Overnight starter cultures of BL21-DE3 E. coli cells transformed with recombinant tau plasmids were inoculated into multi-liter LB broth at 1:50 dilution and 100 mg/mL ampicillin.Cultures were incubated at 37°C, shaking until OD600 reached between 0.5 and 0.6.Tau expression was induced using 1 mM IPTG and continued to grow for an additional 4 hours.BL21-DE3 cells containing expressed tau were pelleted and resuspended in 50 mM NaH2PO4, pH 8.0 and 300 mM NaCl (sonication lysis buffer) at a concentration of 20 mL/L of culture preparation and sonicated at 60% power in ten 30-second intervals over 10 minutes.Cell lysates were centrifuged and supernatant containing the protein was applied to Ni-NTA column equilibrated with sonication lysis buffer.The columns were washed with 40-50 times of bed volumes of column buffer (sonication lysis buffer) followed by washing buffer (50 mM NaH2PO4, pH 8, 300 mM NaCl, and 20 mM imidazole).Recombinant protein was then eluted using elution buffer (50 mM NaH2PO4, pH 8, 300 mM NaCl, and 200 mM imidazole).Fractions were tested for protein concentration using 5 μL of protein sample mixed with 10 μL Coomassie Protein Assay reagent (ThermoFisher Scienti c).Pooled fractions were concentrated to 4 mL using 10 kDa molecular weight cut-off spin columns (Millipore) and ltered using 0.22 μm low-binding Durapore PVDF membrane lters (Millipore).3RCF and 4RCF tau proteins were further puri ed by FPLC using size exclusion Superdex-75 and Superdex-200 columns (GE Healthcare) in 1 x PNE buffer (25 mM PIPES, 150 mM NaCl and 1 mM EDTA, pH 7.0).Final 3RCF and 4RCF proteins were over 90% purity as evaluated by SDS-PAGE.Protein concentrations were quanti ed by BCA protein assays (ThermoFisher Scienti c).

Skin tissue preparation
Skin samples of approximately 30-100 mg in weight and 3-5 mm x 3-5 mm in size, primarily contained epidermis and dermis were collected as previously and prepared described [35,37].Brie y, skin tissues were homogenized at a 10% (w/v) concentration in a lysis buffer containing 2 mM CaCl2 and 0.25% (w/v) collagenase A (Roche) in Tris-Buffered Saline (TBS).The samples were incubated in a shaker at 37°C for 4 hours, shaking at 500 rpm, followed by homogenization using a Mini-BeadBeater (BioSpec, Laboratory Supply Network, Inc., Atkinson, NH).

RT-QuIC Analysis
The RT-QuIC assay was modi ed as previously described with a slight modi cation [16,[18][19][20]60].In brief, the reaction mix for skin tau was prepared with 10 mM HEPES, pH 7.4, 200 mM NaCl, 10 µM ThT, and 10 µM either 4RCF or 3RCF tau substrate.In a 96-well plate (Nunc), 98 µL aliquots of the reaction mix were added to each well, followed by seeding with 2 µL of diluted skin homogenate (1:200 from 5% homogenate supernatant prepared by centrifugation at 3,000 g for 10 min at 4°C) in 10 mM HEPES, 1 x N2 supplement (Gibco), 1 x PBS and centrifuged at 5,000 g for 5 min at 4 °C.The plate was sealed with a plate sealer lm (Nalgene Nunc International) and then incubated at 37°C in a BMG FLUOstar Omega plate reader.The incubation involved cycles of 1 min of orbital shaking followed by a 15-min of rest for the speci ed duration.ThT uorescence measurements from bottom read (450 ± 10 nm excitation and 480 ± 10 nm emission) were recorded every 45 min.Each sample dilution contained 4 replicate reactions.The average ThT uorescence values per sample were calculated using data from all four replicate wells, regardless of whether they crossed the threshold de ned by ROC.A sample was considered positive if at least 2 of 4 replicate wells exceeded this threshold.
To quantify tau-SA detected by RT-QuIC, end-point dilution titrations were employed to determine the estimates of the sample dilution that generated positive reactions in 50% of the replicate reactions as the 50% seeding dose or SD50 (usually 2 out of 4 replicates) [15].

Conformational stability immunoassay
The conformational stability immunoassay of RT-QuIC end products was conducted as previously described with a minor modi cation [61].Brie y, 20 µL aliquots of end products were mixed with 20 µL of GdnHCl stock solution, resulting in nal GdnHCl concentrations ranging from 0 to 3.0 M.After incubating at room temperature for 1.5 hours, samples were precipitated with a 5-fold volume excess of pre-chilled methanol overnight at -20°C.

Western blotting
The samples prepared as described above were separated using 15% Tris-HCl Criterion pre-cast gels (Bio-Rad) in SDS-PAGE.Proteins from the gels were transferred onto Immobilon-P polyvinylidene uoride (PVDF, Millipore) membranes for 90 minutes at 70 V.To probe with the anti-tau antibodies (RD3, RD4, pT231, or pS396), the membranes were incubated overnight at 4°C with a 1:1,000-1:4,000 dilution of the primary antibodies.After incubation with a 1:4,000-1:5,000 dilution of horseradish peroxidase-conjugated sheep anti-mouse IgG, tau bands were visualized on Kodak lm using ECL Plus as instructed by the manufacturer.Densitometric analysis was used to measure the intensity of tau protein bands, which were quanti ed with UN-SCAN-IT Graph Digitizer software (Silk Scienti c, Inc., Orem, Utah).

Filter-trap assay
The lter-trap assay was used to determine the RT-QuIC reaction mixtures with increased ThT uorescence formed aggregates and to evaluate their sizes as described previously [62].In brief, the end products of RT-QuIC were mixed with washing buffer containing 2% SDS, 10 mM Tris-HCl, pH 8.0 and 150 mM NaCl for an hour at room temperature.Following incubation, the samples were ltered through a cellulose acetate membrane (Advantec MFS, Dublin, CA).After ltering, the membrane was rinsed with washing buffer to remove unbound proteins and subsequently blocked with 5% BSA in 0.1% Tween-20 in 1 x PBS for an hour.The membrane was then probed with RD3 and RD4 antibodies, followed by incubation with sheep anti-mouse secondary antibody.The proteins on the membrane were visualized using ECL Plus, and the resulting signal was captured using a chemiluminescent imaging system with X-ray/automatic lm processor.Densitometric analysis was used to measure the intensity of tau protein dots for the quantitative analysis as mentioned above.

Transmission electron microscopy
Transmission electron microscopy (TEM) images were collected as previously described [20,63].Brie y, the skin tau-SAA end product samples at a concentration of 10 µM were maintained in a frozen state until ready for TEM analysis.Before imaging, 2 µL of the sample was applied to a 200 mesh formvar-carbon coated grid and allowed to sit for 5 minutes.The excess sample was gently removed using lter paper.A thorough examination of each grid was conducted to qualitatively assess the presence of oligomers or brils.Representative images were taken from 15-20 distinct locations on each grid.The TEM studies were conducted using a JEOL-1400 transmission electron microscope (JOEL United States, Inc., Peabody, MA) at an operating voltage of 120 kV.

Statistical analysis
Experimental data were analyzed using Student's t-test for comparing two groups.McNemar's test was employed to assess marginal homogeneity and differences in agreement.For comparisons between PD versus CBD and PSP, where the sample size allowed, we conducted a paired area under the ROC curve (AUC) analysis to evaluate signi cant differences in AUC values.Tests adopted a two-sided type II error level of 0.05.

Supplementary Files
Pick's disease (PiD) Prion disease (PrD) Proteinase K (PK) Protein misfolding cyclic ampli cation assay (PMCA) Progressive supranuclear palsy (PSP) Real-time quaking-induced conversion (RT-QuIC) Seed-ampli cation assay (SAA) Seeding activity (SA) Sheep anti-mouse (SVM) Single molecular immunoassay (Simoa) Transmission electron microscopy (TEM) Declarations developed, performed, and interpreted the analyses of RT-QuIC end products of the skin samples.LW and BX designed, developed and performed preparation of human recombinant 4RCF and 3RCF proteins and identi ed site-speci c phospho-tau epitopes.SG, RL and VD provided biopsy skin samples and related clinical data.WZ and WQZ wrote the rst version of the paper.All authors critically reviewed, revised, and approved the nal version of the manuscript.

Figures Figure 1
Figures

Figure 7 Transmission
Figure 7

Table 1
Demographic and neuropathological features of autopsied cases in different groups

Table 1
Demographic and neuropathological features of autopsied cases in different groups

Table 2
Demographic and clinical features of biopsied cases in different groups

Table 3
Demographic and clinical features of individual biopsied cases in different groups * N/A: Not applicable.*N/A: Not applicable.ThT uorescence levels of sTau-SAA end-point correlate with dot-blot intensity of captured tau aggregate of the RT-QuIC end products