Tau protein quantification in skin biopsies differentiates tauopathies from alpha-synucleinopathies

Abstract

Abnormal accumulation of microtubule-associated protein tau (τ) is a characteristic feature of atypical parkinsonisms with tauopathies, such as progressive supranuclear palsy and corticobasal degeneration. However, pathological τ has also been observed in α-synucleinopathies like Parkinson’s disease and multiple system atrophy. Based on the involvement of the peripheral nervous system in several neurodegenerative diseases, we characterized and compared τ expression in skin biopsies of patients clinically diagnosed with Parkinson’s disease, multiple system atrophy, progressive supranuclear palsy and corticobasal degeneration and in healthy control subjects.

In all groups, τ protein was detected along both somatosensory and autonomic nerve fibres in the epidermis and dermis by immunofluorescence. We found by western blot the presence of mainly two different bands at 55 and 70 kDa, co-migrating with 0N4R/1N3R and 2N4R isoforms, respectively. At the RNA level, the main transcript variants were 2N and 4R, and both were more expressed in progressive supranuclear palsy/corticobasal degeneration by real-time PCR. Enzyme-linked immunosorbent assay demonstrated significantly higher levels of total τ protein in skin lysates of progressive supranuclear palsy/corticobasal degeneration compared to the other groups. Multivariate regression analysis and receiver operating characteristics curve analysis of τ amount at both sites showed a clinical association with tauopathies diagnosis and high diagnostic value for progressive supranuclear palsy/corticobasal degeneration versus Parkinson’s disease (sensitivity 90%, specificity 69%) and progressive supranuclear palsy/corticobasal degeneration versus multiple system atrophy (sensitivity 90%, specificity 86%). τ protein increase correlated with cognitive impairment in progressive supranuclear palsy/corticobasal degeneration.

This study is a comprehensive characterization of τ in the human cutaneous peripheral nervous system in physiological and pathological conditions. The differential expression of τ, both at transcript and protein levels, suggests that skin biopsy, an easily accessible and minimally invasive exam, can help in discriminating among different neurodegenerative diseases.

tau protein, skin biopsy, Parkinson’s disease, tauopathies, biomarkers

For the podcast associated with this article, please visit https://academic.oup.com/brain/pages/podcast

See Dugger et al. (https://doi.org/10.1093/brain/awac281) for a scientific commentary on this article.

Introduction

Tau (τ) is a microtubule-associated protein coded by the MAPT gene.¹ In the adult human CNS, post-transcriptional modifications of τ pre-mRNA can produce six different mature mRNA molecules: 0N3R, 0N4R, 1N3R, 1N4R, 2N3R and 2N4R.¹ Such isoforms result from the combination of alternative splicing of exons 2 and 3, generating the amino-terminal inserts 0N, 1N and 2N, and alternative splicing of exon 10 coding for three or four carboxy-terminal repeat domains named, respectively, 3R and 4R.¹ τ is also subjected to several post-translational modifications,^2,3 such as phosphorylation,⁴ acetylation⁵ and ubiquitination,⁶ which strongly regulate its function. Mutations in the MAPT gene or abnormalities in post-translational regulatory mechanisms can induce τ aggregations and toxicity; for instance, hyperphosphorylation in different sites of τ seems to precede and enhance aggregation.^7–9 Although the presence of mutated and/or aggregated τ in the brain is the hallmark of a family of neurodegenerative diseases called tauopathies,¹⁰ such as Alzheimer’s disease, progressive supranuclear palsy (PSP) and corticobasal degeneration (CBD), pathological τ has also been observed in synucleinopathies at later stages like Parkinson’s disease¹¹ and multiple system atrophy (MSA).^12,13

Previous studies have demonstrated the presence of hyperphosphorylated τ in tauopathies patients’ peripheral tissues like olfactory¹⁴ and oral¹⁵ epithelium and colon specimens,¹⁶ boosting the possibility of using peripheral tissues not only for research but also as a diagnostic tool. Due to its easy, minimally invasive accessibility and its great content in nerve fibres, skin is a candidate for biomarker discovery in neurodegenerative disorders.¹⁷ Indeed, skin biopsy has already been exploited to detect α-synuclein, the pathological hallmark of Parkinson’s disease, in patients with synucleinopathies^18–20 and to demonstrate and quantify a small fibre neuropathy in Parkinson’s disease by intraepidermal nerve fibre density (IENFD) measurement.^20–22

Nevertheless, few studies have explored the expression of τ in peripheral tissues and skin^23–26; in particular, a comprehensive investigation of τ in human skin in physiological and pathologic conditions is lacking. Thus, the aims of this study were to: (i) assess the presence of total and phosphorylated forms of τ in skin biopsy from healthy controls (HC) and patients clinically diagnosed with Parkinson’s disease, MSA and PSP/CBD; (ii) quantitatively analyse differences in τ expression among groups; and (iii) evaluate the diagnostic capacity of cutaneous τ for Parkinson’s disease, MSA and PSP/CBD.

Materials and methods

Subject recruitment

Patients were consecutively recruited from the movement disorders outpatient clinic at NSI Lugano from July 2015 to October 2021. HC were recruited among hospital staff and patients’ partners as part of the NSIPD001 study.²⁰

Inclusion criteria for Parkinson’s disease were: a definite clinical diagnosis according to the UK Brain Bank diagnostic criteria,²⁷ no family history and no significant cognitive impairment or dysautonomic symptoms in the history. The inclusion criteria for MSA,²⁸ PSP²⁹ and CBD³⁰ were based on published diagnostic criteria. Exclusion criteria were significant comorbidities, diabetes, renal failure, thyroid pathology, vitamin B12 deficiency, HIV infection, syphilis, coagulopathy, acute and chronic inflammatory diseases and chemotherapy.

This study was performed in line with the principles of the Declaration of Helsinki. Subjects were included according to the study protocol, approved by the Cantonal Ethics Committee. All enrolled subjects gave written informed consent to the study.

Clinical assessments

Parkinson’s disease, MSA, PSP and CBD patients underwent a standard clinical evaluation: disease gravity was assessed by the Hoehn and Yahr scale³¹ and the Movement Disorder Society-Unified Parkinson’s Disease Rating Scale³² (MDS-UPDRS; scale I—clinical evaluation of mental state, behaviour and mood; scale II—self-assessment by the patient of specific daily activities; scale III—clinical evaluation of motor skills); cognitive profile by the Mini-Mental State Evaluation (MMSE)³³ and Montreal Cognitive Assessment (MoCA)³⁴ scales; mood disorder by the Beck Depression Inventory-II (BDI-II) scale³⁵; autonomic dysfunction by the Composite Autonomic Symptom Score 31 (COMPASS-31, OH: orthostatic hypotension, VM: vasomotor, SM: sudomotor, GI: gastrointestinal, BL: bladder, PM: pupillomotor)³⁶; rapid eye movement sleep behavior disorder (RBD) by the RBD screening questionnaire³⁷; olfactory function by the Sniffin’ Sticks Smell test (Burghart Messtechnik GmbH). Levodopa equivalent daily dose (LEDD) was calculated.³⁸

Skin biopsy

On the more clinically affected side, a double 3 mm-diameter punch skin biopsy was performed in the neck at the C8 dermatomal level (cervical) and in the distal leg 10 cm above the lateral malleolus (ankle).^20,39 For each anatomical site, one biopsy was fixed for immunofluorescence (IF) studies^20,22,39 and one biopsy was stored at −80°C until further analysis by western blot, enzyme-linked immunosorbent assay (ELISA) or PCR assays.

Immunofluorescence

Immunofluorescence assays were performed as previously described.^20,22,39 Briefly, skin biopsies were fixed overnight at 4°C in paraformaldehyde–lysine–periodate 2% fixative. The day after, they were frozen and cut with a cryotome to obtain 50-μm thin tissue sections for free-floating IF analysis. Three non-consecutive sections per biopsy were incubated o.n. at room temperature (RT) with a panel of primary antibodies (Table 1). The day after, sections were washed twice in Tris-buffered saline (ThermoFisher Scientific) for 10 min at RT and incubated with the appropriate fluorescence-tagged secondary antibody (Table 1) for 90 min at RT. Cell nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich, 1:5000). Sections were viewed and analysed with a Nikon confocal microscope (×40× magnification, successive frames of 2 µm increments on a Z-stack plan) using NIS Elements 4.11.01 imaging software.

Table 1

List of antibodies

Name	Type	Dilution			Specificity	Company
		IF	Wb	ELISA
Primary
Tau	Rb pAb	1:1000	1:200	1:500	Unphosphorylated and phosphorylated at Ser262 human Tau	Abcam
Tau13	Ms mAb	1:1000	1:500	–	Unphosphorylated human Tau, aa 2-18	Santa Cruz Biotechnology
HT7	Ms mAb	1:1000	–	–	Unphosphorylated human Tau, aa 159-163	ThermoFisher Scientific
Tau5	Ms mAb	1:1000	–	–	Unphosphorylated human Tau, aa 218–225	Abcam
pS404	Rb pAb	1:500	1:500	–	Tau phosphorylated at Ser404	ThermoFisher Scientific
pS262	Rb pAb	1:500	1:500	–	Tau phosphorylated at Ser262	ThermoFisher Scientific
pThr212	Rb pAb	1:500	1:500	–	Tau phosphorylated at Thr212	ThermoFisher Scientific
pSer396	Ms mAb	1:500	1:500	–	Tau phosphorylated at Ser396	ThermoFisher Scientific
PGP9.5	Rb pAb	1:1000	–	–	Axonal marker	Abcam
βTubIII	Ms mAb	1:500	–	–	Axonal marker	Sigma Aldrich
TH	Ms mAb	1:200	–	–	Adrenergic nerve fibres	Millipore
VIP	Rb pAb	1:2000	–	–	Cholinergic nerve fibres	Immunostar
Secondary
Alexa-Fluor 488	G × Ms	1:400	–	–	AffiniPure goat anti-mouse IgG (H + L)	Jackson ImmunoResearch
Alexa-Fluor 488	G × Rb	1:400	–	–	Highly cross-adsorbed goat anti-rabbit IgG (H + L)	ThermoFisher Scientific
Alexa-Fluor 594	G × Ms	1:400	–	–	Highly cross-adsorbed goat anti-mouse IgG (H + L)	ThermoFisher Scientific
Alexa-Fluor 594	G × Rb	1:400	–	–	AffiniPure goat anti-rabbit IgG (H + L)	Jackson ImmunoResearch
IRDye® 680RD	G × Ms	–	1:5000	–	IRDye® 680RD goat anti-mouse IgG	LI-COR Biosciences
IRDye® 800CW	G × Rb	–	1:5000	–	IRDye® 800CW goat anti-rabbit IgG	LI-COR Biosciences
Anti-rabbit HRP	G × Rb	–	–	1:5000	Goat anti-rabbit IgG (H + L)-HRP conjugate	Bio-Rad

Characteristics of primary and secondary antibodies used in this study: name, type, dilution for IF, western blot (Wb) and ELISA assay, specificity and company. - = no antibody used; G = goat; MsmAb = mouse monoclonal antibody; RbpAb = rabbit polyclonal antibody; TH = tyrosine hydroxylase VIP = vasoactive intestine peptide.

Intraepidermal nerve fibre density

Intraepidermal nerve fibre density was assessed as a measure of small fibre neuropathy.^40–43 At least three tissue sections per localization per patient were stained with the primary antibody against protein gene product 9.5 (PGP9.5; Abcam, 1:1000). PGP9.5-positive nerve fibres crossing the dermal–epidermal junction were counted. The length of the section was measured using NIS Elements 4.11.01 imaging software (Tool: manual measure). IENFD was obtained by dividing the number of fibres by the length of the section and expressed as the ‘number of fibres/mm’. IENFD was determined at both ankle and cervical sites, and total IENFD was calculated as the mean of nerve densities at both localizations.²²

Western blot

Skin biopsies from HC (n = 20) and Parkinson’s disease (n = 26), MSA (n = 9) and PSP/CBD (n = 10) patients were homogenized in 1 ml of radioimmunoprecipitation assay buffer (RIPA; Sigma Aldrich) supplied with protease inhibitor (Sigma Aldrich, 1:10) and used for western blot and ELISA assays. After centrifugation at 13 000 rpm for 3 min, supernatants were analysed. Protein quantification was determined with the QuantiPro BCA assay kit (Sigma Aldrich). Proteins were separated by SDS-PAGE on a 12% acrylamide gel and transferred onto polyvinylidene fluoride (PVDF) membranes, which were subsequently incubated with a panel of primary antibodies (Table 1) overnight at 4°C. After three washings in TBS–Tween 0.1%, incubation with secondary antibody for 90 min was performed at RT (Table 1). The membranes were analysed with the Odyssey CLx system (Li-cor Biosciences). Phosphorylated τ quantification in skin extracts was performed by an expert, blind to the clinical diagnosis, with IS Image studio software Ver 5.0. The intensity of the two bands at 70 and 55 kDa was normalized to the β-actin signal. The sum of the two bands was calculated for each patient and each localization.

Enzyme-linked immunosorbent assay

Homogenized samples were diluted at 1:20 in PBS 1× (ThermoFisher Scientific). Fifty microlitres of diluted sample per well were put in a coated 96-well plate (Nunc MaxiSorp, ThermoFisher Scientific) overnight at 4°C. Subsequently, samples were washed three times with PBS–Tween 0.05% and then incubated with: blocking buffer (PBS–Tween 0.05%, bovine serum albumin 1%) for 1 h at 37°C; primary antibody (unphosphorylated and phosphorylated at Ser262 human Tau, 1:500; Table 1) for 1 h at 37°C; secondary antibody [goat anti-rabbit horse radish peroxidase (HRP); Biorad, 1.5000] for 1 h at 37°C. Fifty microliters of TMB-substrate (ThermoFisher Scientific) were added to each well, and absorbance (450 nm) was measured immediately after adding the stop solution (phosphoric acid 2 M) using an Infinite M-PLEX TECAN instrument. τ concentration was normalized to the sample protein concentration (QuantiPro BCA assay kit).

RNA extraction and reverse transcription

Skin biopsies from HC (n = 4) and Parkinson’s disease (n = 5), MSA (n = 5) and PSP (n = 4)/CBD (n = 1) patients were homogenized in 1 ml of TRIzol (ThermoFisher Scientific). A smaller cohort was analysed based on the tissue availability in the bio-bank of the NSIPD001 study (demographic and clinical characteristics are summarized in Supplementary Table 1). As a positive control for PCR assays, we analysed a little cortical sample from a patient without a neurodegenerative disorder who underwent a craniotomy for diagnostic purposes after obtaining informed consent. Two hundred and fifty microliters of chloroform (Sigma Aldrich) were added to the homogenate, which was subsequently centrifuged at 10 000 rpm for 5 min. The aqueous phase was transferred to a new sterile tube containing 550 μl of isopropanol (Sigma Aldrich). Samples were kept at RT for 5 min, followed by 1 h at  80°C. Next, samples were centrifuged at 14 000 rpm for 20 min; the supernatant was discharged and substituted with 500 μl of 75% ethanol (Sigma Aldrich). Ethanol was removed after centrifugation at 9500 rpm for 5 min and pellets were resuspended in 20 μl of pure water. The total RNA quantification of samples was determined using a Nanodrop 200 Spectrophotometer (ThermoFisher Scientific). Eight hundred nanograms of total RNA were reverse-transcribed into cDNA with GoScript™ Reverse Transcription System (Promega Corporation), following the manufacturer’s protocol. The resulting cDNA was stored at − 20°C until use.

PCR assay

PCR was performed using Dream-Taq DNA polymerase (ThermoFisher Scientific) following manufacturer protocol; the annealing temperature was set at 60°C and 35 reaction cycles were performed. To examine the alternate splicing of exons 2 and 3 at the N-terminal domain of τ, the cDNA was amplified using the primers for exon 1 (F 5′-ACACGGACGCTGGCCTGAAA-3′) and exon 4 (R 5′-TCACGTGACCAGCAGCTTCGTCTT-3′). To evaluate the splicing of exon 10 in the microtubule-binding domain repeat region, the cDNA was amplified using the primers for exon 9 (F 5′-CAGTGGTCCGTACTCCACCCAA-3′) and exon 11 (R 5′-TGGTTTATGATGGATGTTGCCTAATGAG-3′). GAPDHwas used as a housekeeping gene and amplified with the primers: F 5′-TGCACCACCAACTGCTTAGC-3′ and R 5′-GGCATGGACTGTGGTCATGAG-3′. Primers were designed with the primer design tool PrimerBlast (NCBI).

The resulting PCR products were separated by 2.5% Tris–acetate–EDTA agarose gel electrophoresis. Obtainable bands were a 2N amplicon of 255 base pairs (bp), 1N and 0N of, respectively, 168 and 81 bp in length; 4R and 3R of, respectively, 317 and 224 bp. The GAPDH amplicon was 87 bp.

Real-time PCR

Real-time PCR was performed using SsoAdvanced™ Universal SYBR Green Supermix (Biorad), following the manufacturer’s instructions in triplicate. MAPT was amplified with two different pairs of primers published on Primerbank: F 5′-CCAAGTGTGGCTCATTAGGCA-3′, R 5′-CCAATCTTCGACTGGACTCTGT-3′ and F 5′-GAGTCCAGTCGAAGATTGGGT-3′, R 5′-GGCGAGTCTACCATGTCGATG-3′.

The 2N splicing variants of the MAPT gene were amplified using the primers for exon 3 (F 5′-TGACAGCACCCTTAGTGGATGA-3′) and exon 4 (R 5′-TCACGTGACCAGCAGCTTCGTCTT-3′), while the 4R splicing variants were amplified using the primers for exon 9 (F 5′-CAGTGGTCCGTACTCCACCCAA-3′) and a primer spanning exons 9 and 10 (R 5′-AGCTTCTTATTAATTATCTGCACCTTC-3′). Primers were designed with the PrimerBlast tool (NCBI). The RPL27 gene was amplified as endogenous controls for normalization (F 5′-TGGTAGGGCCGGGTGGTTGC-3′; R 5′-ACTTTGCGGGGGTAGCGGTC-3′). The PCR reaction was carried out on CFX-Connect real-time PCR Detection System (Biorad): τ cDNA expression was calculated using ΔΔCt methods as previously reported.⁴⁴ Real-time PCR was performed in triplicate, pooling together samples of subjects of the same group. Sample pools were obtained by loading the same amount of cDNA from every sample, and the same quantity of cDNA pools per group was loaded for real-time PCR amplification. Values were normalized to the mean of τ cDNA expression of HC at the ankle and cervical sites.

Statistical analysis

Statistical analysis was performed using IBM SPSS Statistics 26.0. The distribution of variables was assessed by the Kolmogorov–Smirnov test. One-way ANOVA with a post hoc Bonferroni’s test for multiple comparisons was used for normally distributed variables (age), expressed as mean ± SD. The Kruskal–Wallis test was used for non-normally distributed variables (disease duration, Hoehn and Yahr, MDS-UPDRS, BDI-II, MMSE, MoCA, COMPASS-31, Olfactory test, RBD, LEDD, tau concentration, IENFD), expressed as medians and interquartile range (IQR). χ² or Fisher’s exact tests were used for categorical variables (sex), defined as a percentage (%). Univariate and multivariate logistic regression was used to test compound τ for association with disease status (PSP/CBD versus all other groups and PSP/CBD versus MSA/Parkinson’s disease), with odds ratio (OR) and 95% confidence interval (CI) calculation. Receiver operating characteristics (ROC) curves analysis was used to evaluate the area under the curve (AUC) and compare selected variables’ diagnostic performances. Younden index (J = Sensitivity + Specificity –1) was calculated to determine the cut-off with greater accuracy. For linear discriminant analysis, canonical components 1 and 2 were calculated from weighted linear combinations of variables to maximize separation between the four groups (HC, Parkinson’s disease, MSA, PSP/CBD); in the plot, each patient is represented by a point, the centroid indicates the mean of canonical 1 and canonical 2 for each diagnosis. Correlations were evaluated by Pearson’s R test and regression curve analysis; correlations were considered strong for R between |1.0| and |0.5|, moderate between |0.5| and |0.3|, weak between |0.3 and |0.1|. A P-value <0.05 was considered significant.

Data availability

The raw data that support the findings of this article are available on request to the corresponding author.

Results

Patients

Thirty-one patients with idiopathic Parkinson’s disease, 14 with probable MSA, 15 PSP/CBD (11 probable PSP, four possible CBD) and 24 age-matched HC for the Parkinson’s disease group were enrolled. The demographic characteristics and clinical assessments of the study groups are summarized in Table 2. Sex ratio did not differ across all groups, while disease duration did not differ across patients. PSP/CBD subjects were significantly older than HC. MSA and PSP/CBD patients had a more severe disease burden than Parkinson’s disease, as measured by H&Y and MDS-UPDRS scales and a higher cognitive impairment as measured by MMSE. PSP/CBD subjects were more depressed than Parkinson’s disease as measured by the BDI-II scale and showed a more severe cognitive impairment than both Parkinson’s disease and MSA as assessed by MoCA. Finally, MSA patients had greater autonomic impairment in genitourinary function than Parkinson’s disease and PSP/CBD, as measured by the COMPASS-31 questionnaire (BL subdomain). No significant difference in LEDD was detected among groups.

Table 2

Demographic data and clinical scores

	HC (n = 24)	Parkinson’s disease (n = 31)	MSA (n = 14)	PSP/CBD (n = 15)	Overall P-value	Pairwise comparisons
						Parkinson’s disease versus HC	MSA versus HC	PSP/CBD versus HC	Parkinson’s disease versus MSA	PSP/CBD versus Parkinson’s disease	PSP/CBD versus MSA
Age (years)	62 ± 8	67 ± 12	67 ± 8	75 ± 6	0.011	0.090	0.105	0.000	0.868	0.094	0.065
Sex (ref. male)	12 (50.0%)	17 (54.8%)	4 (28.6%)	7 (46.7%)	0.433	–	–	–	–	–	–
Disease duration (years)	–	5.0 [3.0–9.0]	5.0 [1.8–7.3]	4.0 [3.0–5.0]	0.305	–	–	–	–	–	–
Hoehn and Yahr	–	2.0 [2.0–3.0]	4.0 [3.0–5.0]	4.0 [3.0–4.0]	0.000	–	–	–	0.000	0.000	0.377
MDS-UPDRS-I	–	4.0 [2.0–6.0]	8.0 [6.5–10.5]	12.0 [6.5–22.0]	0.006	–	–	–	0.016	0.011	0.310
MDS-UPDRS-II	–	6.0 [5.0–9.8]	15.0 [8.5–21.0]	17.0 [13.0–27.5]	0.005	–	–	–	0.050	0.002	0.421
MDS-UPDRS-III	–	17.0 [12.0–23.3]	32.5 [23.0–46.5]	31.5 [22.5–41.5]	0.000	–	–	–	0.003	0.000	0.798
MDS-UPDRS-Total	–	28.5 [20.3–37.8]	53.0 [31.5–67.0]	53.0 [39.3–97.5]	0.004	–	–	–	0.050	0.001	0.792
COMPASS-31 OH	–	0.0 [0.0–16.0]	0.0 [0.0–21.0]	0.0 [0.0–20.0]	0.864	–	–	–	–	–	–
COMPASS-31 VM	–	0.0 [0.0–0.0]	0.0 [0.0–0.0]	0.0 [0.0–0.0]	0.222	–	–	–	–	–	–
COMPASS-31 SM	–	0.0 [0.0–4.2]	2.1 [0.0–4.2]	0.0 [0.0–4.2]	0.434	–	–	–	–	–	–
COMPASS-31 GI	–	0.9 [0.0–6.3]	3.7 [0.0–5.8]	0.0 [0.0–6.3]	0.697	–	–	–	–	–	–
COMPASS-31 BL	–	0.0 [0.0–0.0]	0.5 [0.0–2.5]	0.0 [0.0–0.0]	0.003	–	–	–	0.002	0.795	0.050
COMPASS-31 PM	–	0.0 [0.0–0.0]	0.0 [0.0–0.0]	0.0 [0.0–0.0]	1.000	–	–	–	–	–	–
COMPASS-31 Total	–	6.3 [0.8–24.1]	12.3 [5.5–27.5]	11.9 [4.2–28.0]	0.675	–	–	–	–	–	–
BDI-II	–	6.0 [3.0–8.5]	8.5 [3.7–13.0]	14.0 [8.0–18.0]	0.006	–	–	–	0.238	0.001	0.068
MMSE	–	30.0 [29.0–30.0]	28.0 [25.5–29.0]	26.0 [22.5–28.0]	0.000	–	–	–	0.004	0.000	0.081
MoCA	–	27.0 [24.0–29.0]	26.0 [21.5–27.7]	20.0 [17.0–24.0]	0.001	–	–	–	0.227	0.000	0.011
Olfactory test	–	7.0 [5.0–9.0]	9.0 [9.0–10.8]	7.0 [4.3–9.8]	0.067	–	–	–	–	–	–
RBD questionnaire	–	3.0 [1.0–5.3]	3.5 [1.0–5.2]	2.5 [0.0–5.0]	0.643	–	–	–	–	–	–
LEDD (mg)	–	495.5 [288.0–775.0]	437.5 [174.8–500.0]	250.0 [100.0–451.0]	0.167	–	–	–	–	–	–
IENFD total (fibres/mm)	14.9 [11.5–16.6]	10.9 [7.3–14.8]	11.8 [8.4–13.8]	13.5 [7.8–17.3]	0.044	0.014	0.013	0.214	0.985	0.475	0.497
IENFD cervical (fibres/mm)	20.6 [14.7–24.8]	14.3 [10.8–21.5]	18.1 [11.4–21.4]	19.9 [12.4–24.8]	0.288	–	–	–	–	–	–
IENFD ankle (fibres/mm)	9.1 [7.5–10.6]	5.7 [4.1–9.7]	6.2 [2.9–8.1]	6.7 [4.0–9.7]	0.029	0.012	0.010	0.082	0.781	0.741	0.604

Clinical characteristics of patients. Variables are reported as mean ± SD, median [IQR], absolute number (per cent), as appropriate. P < 0.05 was considered significant and is shown in bold.

τ Protein is expressed in epidermal and dermal nerve fibres

The presence of τ protein in skin biopsies of HC at both anatomical sites was assessed by IF assays with several primary antibodies against different epitopes of τ (Table 1 and Fig. 1A): the N-terminal region (Tau13), the proline-rich domain (HT7 and Tau5) and the tubulin-binding region (Tau). They all showed clear immunoreactivity along cutaneous nerve fibres (Fig. 1B) counterstained with the panaxonal markers PGP 9.5 or beta tubulin class III (βTubIII). τ Positive nerve fibres were found in the epidermis, dermis and autonomic structures such as sweat glands, sebaceous glands and muscle arrector pili. Indeed, the colocalization of τ with a specific cholinergic marker (vasoactive intestine peptide) and adrenergic marker (tyrosine hydroxylase) confirmed the presence of τ in autonomic nerve fibres as well (Fig. 1C).

Figure 1

τ protein in healthy skin. (A) Schematic representation of τ protein four domains and binding sites of antibodies used for IF. (B) Confocal microscopy images of skin sections from HC stained with axonal markers (PGP9.5 and βTubIII) shown in red and full-length τ markers (Tau13, HT7, Tau5 and Tau) shown in green. For each antibody is shown the staining at the epidermis and dermis, sweat gland (SG), sebaceous gland (SbG) and muscle arrector pili (MAP). Right: Enlargement of intraepidermal small fibres (1) and autonomic fibres in SG (2), SbG (3) and MAP (4) showing the colocalization between Tau and βTubIII in yellow. (C) IF of skin sections of HC showing cholinergic (vasoactive intestine peptide, VIP) and adrenergic innervation (tyrosine hydroxylase, TH) in red and Tau (unphosphorylated and phosphorylated at Ser262 human Tau) in green in SG (5) and SbG (6). Right: Enlargement showing the colocalization between Tau and VIP (top) and TH (bottom) in yellow. Magnification ×40, scale bar = 50 µm. Images are representative of at least three independent experiments on eight subjects.

τ protein was detected in skin biopsies from ankle and cervical sites in Parkinson’s disease, MSA and PSP/CBD (Fig. 2). Overall, qualitative IF analysis showed high levels of τ in the PSP/CBD group, especially at the cervical site and in both intraepidermal somatosensory fibres and autonomic fibres surrounding sweat glands, sebaceous glands and muscle arrector pili. Finally, we tested several antibodies against different phosphorylated epitopes of τ by IF (Table 1): Threonine212 (pThr212), Serine262 (pSer262), Serine396 (pSer396), Serine404 (pSer404). However, we did not detect a positive signal in the skin by IF.

Figure 2

τ Protein in skin in pathological groups. Confocal microscopy images of skin sections from patients with Parkinson’s disease, MSA and PSP/CBD at the ankle and cervical sites. Colocalization between unphosphorylated and phosphorylated at Ser262 human Tau and βTubIII (yellow). For each group, images of epidermis and dermis, sweat gland (SG), sebaceous gland (SbG) and muscle arrector pili (MAP) are shown. Magnification ×40, scale bar = 50 µm. Images are representative of at least three independent experiments in 18 subjects (Parkinson’s disease  = 10, MSA = 3, PSP/CBD = 5).

Two τ isoforms and phosphorylated τ are present in skin

To analyse τ isoforms expressed in skin lysates at the ankle and cervical sites, samples (10 HC, 10 Parkinson’s disease, 10 PSP/CBD and nine MSA patients) were tested with Tau and Tau13 antibodies (Table 1, Fig. 3A and B) and compared with human recombinant tau ladder (Sigma-Aldrich, 1:5000). In skin lysate, two major bands were displayed with both antibodies: one at 55 kDa, which co-migrated with 0N4R/1N3R isoform, and one at 70 kDa corresponding to the 2N4R isoform. Big-tau, the high molecular weight isoform of τ identified in peripheral nerves, was not detected in the skin.e

Figure 3

Skin τ isoforms and phosphorylated τ. (A) Schematic representation of τ isoforms obtained by alternative splicing of exons 2, 3 and 10 (green) and 4a (purple) in the human nervous system. (B) Representative image of western blot analysis of skin lysates from HC, Parkinson’s disease, MSA and PSP/CBD subjects at the ankle and cervical sites, stained with unphosphorylated and phosphorylated at Ser262 human Tau. Images are representative of at least five independent experiments. (C) PCR analysis of skin lysates from ankle and cervical sites. Images are representative of at least five independent experiments. (D) Real-time PCR analysis of MAPT, 2N variant (E) and 4R variant (F) of skin lysates at ankle and cervical sites in pathological groups compared to the healthy human brain. The amount of RNA was normalized to the expression of the housekeeper RPL27. For each group, the mean and SD are reported, and each point represents an experiment. (G) Schematic representation of τ protein four domains and binding sites of antibodies for phosphorylated τ. (H–K) Representative image of Wb analysis of skin lysates from HC, Parkinson’s disease, MSA and PSP/CBD subjects at the ankle and cervical locations, stained with antibodies against phosphorylated τ. MW = molecular weight.

In all groups pThr212, pSer262 and pSer404 antibodies detected both bands at 55 and 70 kDa at the ankle and cervical sites (Fig. 3F, G and I), while pSer396 antibody detected mainly the band at 70 kDa (Fig. 3H). In all subjects, phosphorylated τ resulted more abundant at the cervical site by semiquantitative analysis of bands intensity (P = 0.001), while no significant difference was found between groups. As a negative control, antibodies for phosphorylated forms of τ did not mark the unphosphorylated isoforms of human recombinant τ (Tau ladder, Fig. 3F–I).

2N and 4R mRNAs are more expressed in PSP/CBD

The analysis of the alternative splicing by PCR showed the presence of the 2N variant in skin samples and a lower amount of 1N, while the 0N was never observed. Regarding the splicing at the C-terminal, we observed mainly the 4R and, in some cases, a fainter 3R isoform mainly in HC and Parkinson’s disease. Therefore, according to PCR results, the main transcript variants were 2N and 4R (Fig. 3C).

The quantification by real-time PCR of MAPT gene in skin showed, as expected, a reduced expression compared to the brain and a trend in reduced expression in cervical skin compared with the ankle in Parkinson’s disease but no significant difference among and within groups (Fig. 3D). Because we observed mainly the 2N and 4R by PCR assay, we performed the real-time PCR considering primers specific for these isoforms (Fig. 3E and F). The 2N transcript was higher at the cervical site than the ankle site, and while Parkinson’s disease and MSA showed similar levels to HC, the PSP/CBD group displayed a major increase of the 2N variant at both anatomical sites compared with the other groups and also compared to the brain control. The 4R transcript showed a similar level at both anatomical regions and, similarly to 2N, an increase in the PSP/CBD group.

Skin total τ concentration is a biomarker for PSP/CBD diagnosis

The amount of total τ protein was quantified by ELISA assay using Tau antibody, which targets unphosphorylated and phosphorylated τ at Ser262 τ. Skin lysates from 65 subjects were analysed (HC = 20, Parkinson’s disease = 26, MSA = 9, PSP/CBD = 10). Considering the two anatomical sites separately, the PSP/CBD group showed significantly more τ than Parkinson’s disease and MSA in cervical skin (Fig. 4A and Supplementary Table 2). No differences were observed between the two anatomical sites within groups (Fig. 4A). A higher amount of τ considering both locations (compound τ) was found in PSP/CBD than in all other groups (Fig. 4A). Univariate logistic regression analysis (Table 3) showed a significant association of compound τ concentration and PSP/CBD diagnosis versus all other groups [OR 1.94 (CI: 1.18–3.18), P = 0.008] and versus synucleinopathies (MSA/Parkinson’s disease) [OR 2.09 (CI: 1.15–3.82), P = 0.01]. This association was still significant after adjusting for age, gender and LEDD in a multivariate model [OR 7.4 (CI: 1.27–43.950), P = 0.02, Table 3].

Figure 4

Diagnostic value of skin τ protein concentration. (A) Skin τ concentration measured by ELISA assay at both anatomical sites and at each anatomical region in HC and pathological groups. Kruskal–Wallis analysis is shown, data are expressed as median and IQRs. P-values < 0.05 were considered significant (*<0.05, **<0.01). (B) ROC curves analysis of τ concentration identifying the best cut-off for each parameter. The referral line is reported in grey. For each comparison, the table provides the asymptotic significance, AUC with 95% CI, sensitivity and specificity for τ expression at the ankle, cervical and both sites. P-values < 0.05 were considered significant. (C) The canonical plot stratifies patients according to the clinical diagnosis. The model was built considering τ concentration and IENFD at the ankle and cervical sites. The plot axes canonical 1 and canonical 2 were calculated from weighted linear combinations of variables to maximize separation between the four groups. Each subject is represented by a point. HC versus patients: accuracy 73.8%, leave-one-out validation 69.2%. PSP/CBD versus Parkinson’s disease: accuracy 77.8%, leave-one-out validation 72.2%. PSP/CBD versus MSA: accuracy 84.2%, leave-one-out validation 68.4%.

Table 3

τ concentration and likelihood of tauopathies diagnosis

		Univariate		Multivariate
		OR (95% CI)	P-value	OR (95% CI)	P-value
PSP/CBD versus Parkinson’s disease/MSA	Compound τ	2.09 (1.15–3.82)	0.015	7.49 (1.27–43.95)	0.025
	Age	1.09 (0.99–1.20)	0.050	–	0.111
	Sex	–	0.525	–	0.057
	LEDD	–	0.090	–	0.481

Univariate and multivariate logistic regression analyses were performed to assess the OR and 95% CI for τ protein at both sites (compound), age, sex and LEDD, between PSP/CBD group and Parkinson’s disease/MSA. An OR >1 indicates an increased likelihood of PSP/CBD diagnosis; P-values < 0.05 were considered significant and are shown in bold.

These results were confirmed by ROC curve analysis; compound τ concentration allowed the identification of PSP/CBD group versus Parkinson’s disease (P = 0.025, AUC = 0.812) and versus MSA (P = 0.004, AUC = 0.900), while for PSP/CBD versus HC, the best performance was obtained by τ amount at ankle site (P = 0.017, AUC = 0.774; Fig. 4B and Supplementary Table 3).

Skin τ concentration and IENFD stratify subjects according to clinical diagnosis

IENFD was assessed as a measure of small fibre neuropathy and a marker of neurodegeneration. While Parkinson’s disease and MSA showed a significant reduction of total and ankle IENFD compared with HC (Table 2), PSP/CBD patients did not present skin denervation compared to HC.

IENFD contributes to the stratification of all groups simultaneously; an linear discriminant analysis model based on τ protein expression and IENFD at both sites allowed the separation of subjects according to their clinical diagnosis with 53.8% accuracy (Fig. 4C, leave-one-out validation 47.7%). In particular, HC could be separated from patients (Parkinson’s disease, MSA, PSP/CBD) with 73.8% accuracy (leave-one-out validation 69.2%) and PSP/CBD from synucleinopathies with 82.2% accuracy (leave-one-out validation 77.8%). Of interest, PSP/CBD were discriminated from Parkinson’s disease and MSA with 77.8% and 84.2% accuracy, respectively (leave-one-out validation 72.2% and 68.4%, respectively).

Skin τ concentration correlates with clinical scales

In HC, compound and cervical τ concentration inversely correlated with age, showing a decrement of the protein with ageing (Fig. 5A and B). In MSA, compound concentration directly correlated with COMPASS-31 OH and COMPASS-31 BL, while ankle τ directly correlated with COMPASS-31 OH and cervical τ with COMPASS-31 BL (Fig. 5C–F). In PSP/CBD, compound and ankle τ inversely correlated with MoCA (Fig. 5G and H). No correlations were observed in the Parkinson’s disease group.

Figure 5

τ protein concentration and clinical scores correlations. Correlations between τ protein concentration and age in HC (blue) and between τ protein concentration and clinical scales in MSA (orange) and PSP/CBD (green). For each correlation, the P-value (P), Pearson’s correlation coefficient (R), regression line (black) and 95% CI (dashed line) are reported.

Discussion

This study demonstrated that τ protein is highly expressed in human epidermal somatosensory nerve fibres and cholinergic/adrenergic nerves surrounding dermal autonomic structures, such as sweat glands, sebaceous glands and muscle arrector pili. In particular, we demonstrated an increased amount of τ both at transcript and protein levels in skin biopsies of patients with PSP/CBD compared with healthy subjects and patients with Parkinson’s disease and MSA. This is of relevance because it highlights that accumulation of τ occurs in the PNS as well as in the CNS, and it can be detected in an easily accessible tissue such as skin.

A previous study in abdominal skin biopsy of patients with tauopathies using HT7 antibody against full-length τ demonstrated τ immunoreactivity only in autonomic fibres surrounding sweat glands.²³ The discrepancy is probably explained by technical issues: formalin-fixed and 4-μm-thick paraffin-embedded sections were used in the previous study, while we used PLP 2% fixative and 50-μm sections. Indeed, formalin fixation reduces the integrity of peripheral antigen retrieval and is not suitable for the study of intraepidermal small nerve fibres.⁴⁵ Besides, thicker tissue sections are required for adequate skin nerve fibres sampling because a standard paraffin-embedded tissue section provides only a fraction of the tissue volume obtained with a 20–50-mm frozen tissue section.⁴⁶ In addition, thin tissue sections, by disrupting the nerve fibre structure, reduce the ability to visualize intraneural proteins such as τ.⁴⁵

In the skin, we detected two bands by western blot at 55 and 70 kDa, which is different from the human adult brain, which presents six τ isoforms.⁴⁷ The 55 kDa isoform was also observed in abdominal²⁴ and cervical skin²⁶ and in colon mucosa samples¹⁶ and co-migrated with the 0N4R-1N3R isoforms expressed in the adult human brain, similarly to what has been demonstrated for τ in the gastro-enteric system.¹⁶ Indeed, skin and gastro-enteric tissues comprise small nerve fibres encompassing amyelinated somatosensory fibres (epidermis) and thinly myelinated Aδ (dermis and autonomic nervous system), which share anatomical and physiological characteristics with the dopaminergic and cholinergic system, prone to synucleinopathies and tauopathies, respectively.¹⁷ The 70 kDa band was not previously observed in the skin; however, it was confirmed by a panel of different antibodies, so that it is unlikely unspecific. Consistently with previous evidence,^16,24 we did not detect the peripheral τ isoform named ‘Big-tau’ at 110 kDa⁴⁸ in the skin. Big-tau is generated by the incorporation of exon 4a that doubles the N-terminal domain and allows a greater stabilization and spacing between the microtubules in large diameter peripheral nerves, such as sciatic nerves and central process projecting to the dorsal horn of the spinal cord.⁴⁹ Thus, Big-tau is likely absent in skin innervation where small fibres are predominant.

The transcript analysis of τ revealed the 2N and 4R variants are the most frequent in the skin; we observed mainly the 4R in all groups and the 3R in HC and Parkinson’s disease. This is of interest because we know that in PSP and CBD, the abnormal τ aggregates are composed primarily of 4R isoforms and that in neurodegenerative disease, the ratio of 3R/4R is altered.⁵⁰ Of note, quantitative analysis of isoform transcripts showed a major increase of 2N and 4R mRNA in the skin of PSP/CBD compared to the other groups and the healthy brain sample. In physiological conditions, the 2N represents only the 10% of τ isoforms in the brain⁵⁰ and displays a higher propensity for somatodendritic localization; thus, this increase in the skin axonal compartment looks abnormal.⁵¹ Another study has demonstrated that protein interactors of 2N τ isoforms are highly enriched in disease-related pathways.⁵² In line with the increased level of the transcript, we also demonstrated an increased amount of total τ protein in the skin of PSP/CBD, and it is known that the expression level of substrates can amplify the pathogenic seed and contribute to the selected vulnerability of different neuronal populations in misfolded protein-associated neurodegenerative diseases.^53,54 Further, recent evidence suggests that τ can assemble into fibrils and filaments before post-translational modification, such as phosphorylation and ubiquitination.⁵⁵ Beyond confirming the utility of skin biopsy analysis for diagnostic purposes, these results may have relevance in terms of mechanisms of disease in tauopathies.

We demonstrated phosphorylated τ species in the skin in both HC and patients with neurodegenerative disorders by western blot and not by IF, likely due to the lower sensitivity of IF. In all groups, phosphorylated τ was more abundant at the cervical site. The presence of phosphorylated τ in healthy conditions is consistent with an increasing body of evidence suggesting how τ protein can be physiologically regulated by phosphorylation/dephosphorylation processes.⁴ Further, phosphorylation at Ser262, Ser396 and Ser404 were already demonstrated in non-pathological conditions.^4,56 Moreover, it has been observed that τ protein freshly isolated from surgical brain tissues presents a more highly phosphorylated state than normal adult human τ from autoptic studies due to the intense dephosphorylation activity of the post-mortem period and that adult human brain τ is phosphorylated at many of the same sites as Alzheimer’s disease paired helical filaments.⁵⁷ The finding of phosphorylated τ in synucleinopathies is not unexpected. In a previous work on post-mortem Parkinson’s disease striatum, it has been shown that τ was hyperphosphorylated at Ser262 among other epitopes,¹¹ and a more recent study showed that τ was hyperphosphorylated at 10 epitopes in Parkinson’s disease striatum, sharing 50% overlap with Alzheimer’s disease.⁵⁸ Previous studies showed greater AT8 (pS202/T205) immunoreactivity in skin biopsies of patients with Alzheimer’s disease than HC and with PSP than Parkinson’s disease, while no differences were detected with PHF-1 (pSer396/Ser404).^25,26 We did not find a specific pattern of phosphorylation in the skin of PSP/CBD patients; this is possibly explained by methodological issues, as increased phosphorylated sites are usually detected in insoluble aggregates of τ derived from brain of patients, while we analysed the soluble fraction only.

We found a significantly higher concentration of τ measured by ELISA in the skin of PSP/CBD patients compared with the other groups. Increased τ concentration in skin predicted the PSP/CBD diagnosis after adjusting for age, gender and LEDD, and ROC curve analysis demonstrated an optimal performance, especially at the cervical site, in discriminating PSP/CBD from synucleinopathies, Parkinson’s disease and MSA. This is of clinical relevance as discriminating PSP/CBD versus Parkinson’s disease and versus MSA is often a challenging task also for movement disorders experts. Furtherly, τ protein concentration in skin correlated with cognitive impairment in PSP/CBD.

Between the two anatomical sites, cervical skin biopsy showed the largest differences among groups and the highest diagnostic performance. The diagnostic potential of cervical skin biopsy in differentiating Parkinson’s disease and atypical parkinsonism was also demonstrated by measuring α-synuclein oligomers by proximity ligation assay.²² The cervical area possibly bears the advantage of being anatomically closer to the CNS and brainstem and less exposed to confounders such as peripheral neuropathies that are more frequent in the ageing population. These data, taken together, push towards a deeper and combined evaluation of τ and α-synuclein as well as their pathological forms in cervical skin as a promising diagnostic tool to differentiate misfolding protein-related neurodegenerative disorders. The increased expression of τ in skin biopsy of patients with tauopathies is also relevant because it sheds light on possible pathological events occurring in the PNS. In supporting a potential role of τ in causing peripheral neurodegeneration, a recent work showed that overexpressing human non-mutant τ in mice is associated with peripheral neuropathy with somatofugal degeneration.⁵⁹ In fact, τ plays an important role in stabilizing neuronal microtubules and regulating axonal transport as well as mitochondria homeostasis,¹⁷ and an overexpression/dysfunction may cause axonal degeneration. On the other hand, in this study, we did not find a clear intraepidermal nerve fibre reduction in PSP/CBD, while a small fibre neuropathy was detected in Parkinson’s disease and MSA. Thus, it is conceivable that α-synuclein is associated with a dying-back degeneration while τ may be involved in more centrifugal-like neurotoxicity.

The limitations of this study are the relatively small number of subjects evaluated and the lack of a post-mortem pathological confirmation of the clinical diagnosis. Besides, we analysed only a few of the numerous phosphorylated epitopes of τ, and it would be appropriate to also investigate other post-translation modifications such as acetylation or ubiquitination dysregulation that may have a significant role in pathology.^5,6 It would be also of interest to analyse insoluble τ aggregates in the skin of patients with PSP/CBD and synucleinopathies. Further, we used a custom ELISA assay, and for reproducibility in future studies, it would be recommended that a commercially available ELISA for Tau be utilized. However, the strengths of the study are the evaluation of a prospective cohort with a group of healthy, age-matched subjects and the exclusion of other concomitant systemic pathologies and possible confounders like dopaminergic treatment.

In conclusion, this comprehensive study accurately characterized the presence of τ protein in skin biopsies of multiple neurodegenerative disorders. It raises new questions about the pathogenesis of tauopathies, but it also shows how skin biopsy is a robust research tool in the field of diagnostic biomarkers for neurodegenerative diseases.

Acknowledgements

The authors are very grateful to all the patients and their relatives who participated in this study. They are grateful to Mrs Nicole Vago and Mrs Mara Kuster, CTU nurses, for their valuable work on the clinical database.

Funding

The research leading to these results received funding from FIDINAM 05.2020, Swiss Parkinson and Jacques und Gloria Gossweiler Stiftung.

Competing interests

The authors report no competing interests.

Supplementary material

Supplementary material is available at Brain online.

Abbreviations

 
• CBD

  corticobasal degeneration

 
• ELISA

  enzyme-linked immunosorbent assay

 
• HC

  healthy controls

 
• IF

  immunofluorescence

 
• IENFD

  intraepidermal nerve fibre density

 
• LEDD

  Levodopa equivalent daily dose

 
• MSA

  multiple system atrophy

 
• PSP

  progressive supranuclear palsy