This study provides the first evidence of distinct p-tau immunoreactive structures in the peripheral nerves of tauopathies, with significantly greater involvement in PSP compared to other tauopathies (Table 1). All studied PSP cases exhibited tau accumulation in cranial or spinal nerves, particularly in cranial nerves III, IX/X, XII, and anterior roots. Additionally, the tau biosensor assay demonstrated, for the first time, that pathological tau not only in the CNS but also in the peripheral nerves possess seeding capacity.

While our understanding of tau aggregation in peripheral nerve tissues is still limited in histopathological studies,^11-13 it has been well documented in α-synucleinopathies.⁵ In PD/LBD, α-synuclein predominantly accumulates in the axons,³² and in multiple system atrophy (MSA), it accumulates in Schwann cells.²¹ Furthermore, animal studies have suggested that α-synuclein may accumulate in the PNS before affecting the corresponding nuclei in the CNS.^33,34 In addition, these models have been shown to result in associated peripheral neurological symptoms.^35,36

PNS-tau in PSP exhibited 4R-tau with an immunoreactivity profile for different anti-tau antibodies resembling the neurofibrillary tangles observed in the CNS. PNS-tau in CBD displayed morphological features with fine diffuse/granular structures, resembling the p-tau deposition pattern seen in neurons.^1,37 Thus, tau lesions in the PNS reflect well with the morphology of tau pathology in the CNS. On the other hand, the frequency of PNS-tau was significantly different between PSP and CBD.

The underlying mechanisms for the differences in why PNS-tau is observed only in 4R-tauopathies such as PSP and CBD but not AD, CTE or Pick’s disease and why PNS-tau is more prevalent and prominent in PSP are unclear. Analogously, the distinct neuronal vulnerability patterns among tauopathies have also not yet been completely elucidated. However, a recent study suggests that tauopathies can be distinguished based on the structure of tau filaments.³ We can speculate that, similarly to CNS-tau, differences in the conformational properties of misfolded tau and predominance of tau isoforms, as seen also in PSP and CBD,³ may be related to the susceptibility of tau lesions to peripheral nerves, namely, the pathological tau in PSP may be more prone to invade PNS. Indeed, we additionally evaluated some cases of Pick’s disease (3R-tauopathy) and one further case with the 4-repeat tauopathy, LNT.¹⁴ Interestingly, we found no PNS-tau lesions in the Pick’s disease cases (Supplementary Table 2), while p-tau-positive deposits were present in the PNS in LNT (Supplementary Fig. 1). In the latter case, the severity of tau pathology in the PNS resembled PSP rather than CBD cases. Interestingly, it has been reported that LNT is hierarchically closer to PSP than CBD in terms of tau filaments³ and CNS histopathology.¹⁴ Furthermore, it may also be necessary to investigate the relationship between PNS-tau and one of the special tau isoforms that are abundantly expressed in the PNS, the so-called ‘big tau’.^38-40 Interestingly, big tau has been reported as 4R-tau.^41,42

This study shows that tau accumulation is predominantly in axons and not in the myelin or Schwann cells. This result differs from α-synuclein; Schwann cells are involved in MSA,²¹ while in LBD the accumulation in the PNS occurs predominantly in axons with minimal involvement of the Schwann cells.³² Therefore, neurodegenerative proteins show distinct selective vulnerability of different components of the PNS.

Importantly, in PSP, the severity of p-tau lesions in the cranial and spinal nerves generally correlated with that in the corresponding nuclei; however, focal p-tau deposits were present in some nerves without involvement in the corresponding nuclei. On the other hand, in CBD, we observed tau pathology in lower motor neurons without or only with mild involvement of anterior spinal roots. These observations suggest that (i) propagation of disease-associated tau into the periphery depends on the type of fibrillar tau; or (ii) that p-tau lesions in peripheral nerves are not necessarily secondary to the corresponding nuclei. We cannot exclude the possibility that tau accumulation starts in the periphery in some cases, although the lack of fibrillar tau pathology in peripheral organs does not support this concept.^12,13

Interestingly, PNS-tau was observed predominantly in peripheral motor nerves in PSP. A recent study reported that PSP cases frequently show TDP-43 pathology in lower (38%) and upper (19%) motor neurons.³⁰ Although in our collection we could not find a similar frequency (upper motor neuron: 6.7%; lower motor neuron: there were only a few neurites and positive glia), together with further studies describing involvement of lower motor neurons in PSP,¹¹ these observations emphasize the vulnerability of the motor system in 4R tauopathies. Indeed, another 4R tauopathy, globular glial tauopathy,^43,44 is a disorder where this is more prominent. Importantly, globular glial tauopathies show overlap with PSP, exemplified by the type II subtype, and also the structure of tau filaments shows similar features.³ When comparing PSP cases with and without TDP-43 pathology, no remarkable differences in the extent or degree of PNS-tau pathology were found. One of the two cases also showed LATE-neuropathologic change. However, the previous study examining TDP-43 pathology in PSP documented that the presence or absence of spinal cord TDP-43 pathology was not significantly associated with LATE stage and that extra-limbic TDP-43 pathology differs from the spreading manner of LATE.^20,30 Altogether we concluded that no association between PNS-tau and TDP-43 pathology can be confirmed in our study.

We also examined the extent of tau pathology in the PNS within each PSP case. Although cranial nerves I (olfactory bulb) and II (optic chiasm and nerve) are part of the CNS, they were included in the analysis. Strikingly, all PSP cases showed extensive tau accumulation in multiple cranial nerves, including the olfactory bulb and optic nerve, and spinal anterior/posterior roots from cervical to lumbar cord levels. Notably, even cases with relatively short disease durations (PSP Cases 2 and 6, with 3 years duration of illness), exhibited p-tau lesions in multiple nerves supporting the notion that PNS-tau lesions may occur relatively early in the disease progression of PSP. Tau accumulation in the olfactory bulb is often associated with tauopathy, including PSP and particularly AD.⁴⁵ Although the number of cases available for evaluation was small, we assessed tau accumulation in the optic nerve and found it to be present in all examined PSP cases (3/3). This is compatible with findings of the involvement of the visual system in AD and PSP⁴⁶ and with a recent study that reported tau accumulation in the retina as a diagnostic biomarker for tauopathies.⁴⁷

Although, motor nerves were predominantly affected, involvement of the IX/Xth (91%) and Vth (100%) cranial nerve and the spinal posterior roots (60%) indicate that the sensory nervous system is also involved in the disease process of PSP. These observations are important for the potential of skin biopsy as a biomarker also for tauopathies. Importantly, we show that tau in the spinal roots has seeding ability. Previous studies have predominantly focused on the involvement of the PNS in α-synucleinopathies. PD frequently presents with non-motor symptoms, including sensory and autonomic disturbances, and pathologically the aggregation of pathological proteins in peripheral nerves is a usual feature in α-synucleinopathies.^4,5 Recently, RT-QuIC seeding activity of α-synuclein in the skin has been shown to have potential diagnostic value in synucleinopathies.^6,7

Despite the paucity of observations on tau pathology in nerves, recent studies have begun to capture involvement of the PNS in tauopathies. Indeed, autonomic and sensory disturbances, with electrophysiological and pathological features have been reported in PSP.^8-10 Histopathologically, PSP patients displayed a length-dependent loss of sensory and autonomic nerve fibres in the skin biopsy samples.⁹ Interestingly, the P301L mutant tau transgenic animal model exhibits motor deficits with severe peripheral nerve involvement, such as axonal spheroids in the anterior spinal roots, axonal degeneration in peripheral nerves and neurogenic skeletal muscle atrophy,⁴⁸ supporting the notion of peripheral involvement in tau-related diseases. However, the findings related to the expression of p-tau immunoreactivity in peripheral nerves are still limited in the previous studies.^5,11-13 In the spinal ganglia, no NFTs were detected in eight patients with PSP and 20 patients with AD.⁵ Stejskalova et al.¹¹ reported p-tau-immunoreactivity (AT8) in the anterior and posterior roots in the PSP cases. Dugger et al.¹² examined abdominal skin, submandibular gland and sigmoid colon in the PSP cases. Phospho-tau-immunoreactivity (AT8) of those nerve elements was very rare. Recently, in the peripheral nerves of skin, increased tau mRNA expression and total tau protein in the PSP cases has been reported,¹³ suggesting its potential as a biomarker.^6,7 However, even in that study, immunostaining (immunofluorescence) did not demonstrate positivity for phosphorylated tau.¹³ This may have been due to a small number of peripheral nerve tissues and a few tau depositions in the skin. Our immunohistological analysis did not reveal clear tau aggregation in peripheral nerves in cases of AD and CTE, while there is a report of tau expression in peripheral tissues in AD.¹² The possibility that peripheral tissues in tauopathies other than PSP could also be used as biomarkers for diagnostic purposes merits further studies.

Regarding clinical features, not surprisingly we observed a significantly higher frequency of eye movement disorders and bulbar symptoms in PSP cases compared with other tauopathies. Interestingly, some PSP cases with PNS-tau lesions in the anterior spinal roots were also accompanied by lower-motor neuron signs. However, cranial nerve palsy is not a classical feature of PSP and was not detected in our cohort. Therefore, based on the limited retrospective clinical data in this study, we cannot conclude that these symptoms were truly due to peripheral nerve impairment. While prospective studies can identify PNS-related symptoms, the message of finding PNS-tau pathology is more to support the notion that detection of misfolded tau in the PNS can be a target for biomarker development, such as implemented for the α-synucleinopathies, for example using skin biopsies.

This study has several limitations. First, peripheral neurological symptoms were evaluated retrospectively. A larger comprehensive and prospective continuous assessment of peripheral neuropathy may be necessary to accurately evaluate its presence in tauopathies. Second, the samples we used did not allow assessment of a mild reduction and degeneration of peripheral nerve fibres. Prominent loss of nerve fibres, such as in motor neuron disease or hereditary neuropathies, was not seen in the PSP cases examined. However, the presence of previously reported peripheral nerve loss and symptoms, abnormalities in nerve conduction velocity,^9,10 increased expression of tau protein in the PNS,¹³ and the presence of large tau inclusions and p-tau deposits filling many nerve fibres in this study suggest that accumulated tau in the peripheral nerve fibres is likely to cause some dysfunction of the peripheral nerves. Third, skin and peripheral ganglia were not evaluated. Evaluation of skin is necessary for future biomarker studies. Since the amount of pathological tau might be too low to be consistently detected by immunohistochemistry, seeding amplification assays (RT-QuIC) might be a more feasible approach.

In conclusion, in this study, we showed for the first time that tau pathology with distinct p-tau immunoreactive fibrillar structures is evident in the cranial nerves and spinal nerve roots in PSP and to a far lesser extent in CBD. The morphological characteristics of PNS-tau resemble those in the CNS and show selectivity for PSP. Furthermore, we showed for the first time that tau in peripheral nerves has seeding ability using tau biosensor cells. Tau pathology in PNS is a previously unappreciated feature of tauopathies that could serve as a potential diagnostic biomarker in the future.