TF was collected using Schirmer test strips from 56 patients. Ten of these were healthy control patients without evidence of neurodegenerative disease, and 29 were patients with probable or clinically established PD (mean Hoehn and Yahr stage 2.2 +/− 1). In addition, we included 7 patients with either possible or probable MSA, of which 4 were of the cerebellar and 3 of the Parkinsonian subtype, and 10 patients with either possible or probable PSP were included. All patients were well matched regarding age, sex, concomitant eye diseases, or medication (Table 1). Crucially, the use of systemic medications with anticholinergic effects was evenly distributed between the groups. Disease duration, however, was significantly shorter in both MSA and PSP, compared to PD (mean 8.3 +/− 0.7 years in PD, 1.9 +/− 0.9 years in MSA, and 2.7 +/− 1.6 years in PSP, pANOVA = 0.005, pPosthoc-PSP = 0.02, pPosthoc-MSA = 0.03). Importantly, wetting length (WL) of the Schirmer test strips was significantly shorter in the PD but not the PSP and MSA group compared to the control group when correcting for age in a multiple linear regression model (37 +/− 22 mm/5 min in control, 21 +/− 17 mm/5 min in PD, 20 +/− 21 mm/5 min in MSA, 18 +/− 9 mm/5 min in PSP) (p = 0.02, estimate − 10.73, 95% confidence interval [− 19.39, − 2.07]).

Table 1 Characteristics of the study populationmiRNA Concentration Correlates with Wetting Length

We next isolated miRNA from the TF eluate and quantified its concentration. No significant difference between the groups was observed (Table 1). Correlating the miRNA concentration to clinical parameters, a significant and strong correlation with WL was observed in the overall cohort (RP = 0.68, p < 0.001, 95% CI [0.50, 0.80]) (Fig. 1b). This effect was persistently observed in each of the individual patient groups, as well (control RP = 0.68, p = 0.029, 95% CI [0.10, 0.92], PD RP = 0.58, p < 0.001, 95% CI [0.27, 0.78], MSA RP = 0.91, p = 0.004, 95% CI [0.50, 0.99], PSP RP = 0.74, p = 0.013, 95% CI [0.22, 0.94]) (Fig. 1c). Age at sampling and miRNA concentration did not show any significant correlation in the overall cohort (Fig. 1d). However, a strong negative correlation was found in the control group (R =  − 0.76, p = 0.011, 95% CI [− 0.94, − 0.24]). In contrast, no trend was observed for PD, MSA, and PSP (Fig. 1e). Lastly, we evaluated the relationship of disease duration and miRNA concentration. Combining all disease groups, PD, MSA, and PSP, no significant correlation was observed. Exploring the effect in the subgroups, no correlation was seen for PD and PSP, whereas a strong positive correlation was seen for MSA (R = 0.84, p = 0.038, 95% CI [0.08, 0.98]) (Supp. Fig. S1).

PCR-Profiling of miRNAs in the Tear Fluid of Patients with PD and aPD

After isolation of miRNAs, we quantified the miRNA levels of the pooled control, PD, MSA, and PSP samples using an RT-qPCR-based miRNA profiling kit. Of all 1113 quantifiable miRNAs, 286 were found in all groups, whereas 244 miRNAs were not found in any group. Unsupervised hierarchical clustering of the expression status of all quantified miRNAs revealed a clustering of the PD and PSP groups, as well as the MSA and control groups (Fig. 2a). Next, we aimed to identify miRNAs that were uniquely identified in each of the disease groups. For this, all miRNAs exclusively amplified with high certainty in the respective groups or exclusively amplified in all other groups, but not the disease group itself, were analyzed using an UpSetR plot (Fig. 2b). Fifty-five miRNAs were exclusively amplified with high certainty in the PD group, whereas 4 miRNAs were exclusively amplified with high certainty in all groups but the PD group. In the MSA group, 14 miRNAs were exclusively amplified with high certainty, and 41 miRNAs were exclusively amplified with high certainty in all other groups except for the MSA group. 35 miRNAs were exclusively amplified with high certainty in the PSP group, whereas 27 miRNAs were exclusively amplified with high certainty in all groups but the PSP group (a detailed listing of all miRNAs in the intersections is available in Supplementary Table S1). The annotated names from the profiling kit were converted to the current annotation in miRBase (v22), and only currently annotated miRNAs were used for further analysis.

Fig. 2

Expression analysis of the miRNA RT-qPCR screen. a Heatmap using unsupervised hierarchical clustering showing expression data for all 1113 quantified miRNAs. miRNAs were assigned to different groups of probability of expression: miRNAs with both technical replicates showing a CT value < 40 and an SD < 5 were considered to be “amplified with high certainty” (dark red), miRNAs not amplified in both replicates were classified as “not amplified” (grey); all else were considered “amplified with low certainty” (light red). b UpSetR plot showing the intersections of miRNAs between the different groups. miRNAs present in the intersects are amplified in both technical replicates with a cycle threshold of < 40 and a standard deviation of < 5. Intersections representing miRNAs exclusively amplified in each disease group are filled in color. In contrast, intersections representing miRNAs found in all but the disease group are outlined in color (blue = PD, orange = MSA, red = PSP). c Stacked bar graph representing the number of miRNAs in each disease intersection (consisting of the exclusively and exclusively not amplified miRNAs). The smaller bar to the right of the grey bar representing all miRNAs in the intersection shows the number and detailed annotation of miRNAs described for the respective disease in literature previously. PD, Parkinson’s disease; MSA, multiple system atrophy; PSP, progressive supranuclear palsy

Presence of TF miRNA in Other Biofluids

To understand whether the miRNAs identified have been previously described in other biomaterials or are potentially specific for tear fluid, we compared our findings to data obtained in other biofluids. A detailed listing of all literature used for this comparison can be found in Table 2 (for a detailed description of methodological aspects of the search, see the “Literature search” section). While approximately one-third of the miRNAs did not show any differential expression in the respective disease groups in other biofluids before, some miRNAs have been shown to be differentially regulated (Fig. 2c). For the PD intersections, 56 of the 59 miRNAs identified as being either exclusively present or absent in PD with the profiling kit were annotated in the current version of miRBase (v22). Of those, 40 were not previously described as significantly altered in PD. 12 were previously described in blood, hsa-miR-542-5p was previously described in CSF, and hsa-miR-516a-5p in brain tissue. Hsa-miR-95-3p and hsa-miR-374a-5p have been identified to be altered in CSF and brain tissue, while the latter was also described in blood. 53 of the 55 miRNAs found in the MSA intersections are currently annotated. Of these, 46 have not been shown to be altered in MSA before. Hsa-miR-130a-3p, hsa-miR-29c-3p, hsa-miR-92a-1-5p, and hsa-miR-93-5p have been depicted differently expressed in blood, while hsa-miR-1203 and hsa-miR-1909-5p were discriminative in brain tissue. Hsa-miR-24–1-5p was shown to be altered in blood and CSF. Lastly, while we identified 62 miRNAs in the PSP intersects, only 60 of them were annotated in miRBase. Of these, 56 have not been described as altered in other biofluids. Hsa-miR-425-5p and hsa-miR-99b-5p were previously shown to be altered in blood. Hsa-miR-423-5p was identified to be changed in CSF and hsa-miR-132-3p in brain tissue. Taken together, most previously described miRNAs have been shown in blood, while only two and four have been identified in CSF and brain tissue, respectively. Only three miRNAs, two of them belonging to the group of miRNAs identified in PD, have been described in more than one biomaterial. Most miRNAs, namely 16, have been described in PD, while only seven and four have been identified in MSA and PSP, respectively.

Table 2 MicroRNAs in the intersections of the cohort

To further explore the potential origin of all not previously described miRNAs, we utilized a tissue expression database approach using the DIANA-miTED online tool. Interestingly, mostly (35 out of 40 for PD, 35 out of 46 in MSA, and 52 out of 57 in PSP) have been identified in brain tissue before. For the remaining miRNAs, we employed literature research querying for the terms “tear fluid” or “tear” or “lacrimal”, which did not yield any results.

Overrepresentation Analysis of the Unique Intersections of miRNAs

To better inform about the potential function of the miRNAs identified in TF, we performed an exploratory overrepresentation analysis (ORA) using the DIANA miRPath online tool (v4). miRNAs that were exclusively identified in the respective disease groups were analyzed separately from those exclusively absent in one of the respective groups (Fig. 3). Negative enrichment ratio values were assigned to exclusively absent miRNAs and positive values to exclusively present miRNAs. Semantic similarity analysis (using REVIGO) was then performed to reduce redundancy and summarize the complete list of terms.

Fig. 3

Overrepresentation analysis of the miRNAs in the PD, MSA and PSP intersections. The top 5 terms sorted by enrichment ratio with FDR < 0.01 and at least 5 miRNAs with targets in the terms are depicted. For each disease group (PD = a–d (blue box), MSA = e–h (orange box), PSP = i–l (red box), GO MF, GO BP, GO CC, and KEGG are shown (from left to right, top to bottom). Enrichment ratio (indicated below each graph) was calculated by dividing the number of enriched genes in the term by the total number of genes in the term. Negative values were assigned to terms derived from the analysis of the intersection of miRNAs exclusively not found in the disease group, positive values to those terms derived from the analysis of the intersection of miRNAs exclusively found in the disease group. Sizes of the dots represent target genes in the term, and color represents the FDR. PD, Parkinson’s disease; MSA, multiple system atrophy; PSP, progressive supranuclear palsy; GO, Gene Ontology; MF, Molecular Function; BP, Biological Process; CC, Cellular Component; KEGG, Kyoto Encyclopedia of Genes and Genomes pathways

After filtering, the GO term analysis for the PD intersections revealed only significantly enriched terms pointing to the mitogen-activated protein kinase (MAPK) pathway in the terms annotated to the molecular function (MF) category, supported by the REVIGO analysis, which summarized the enriched terms under protein kinase activity (Fig. 3a, Supplementary Fig. S2, Supplementary Data S2). The MAPK pathway has been associated with cell proliferation, differentiation, and survival. Exploring the REVIGO analysis in detail, several terms belonging to in-utero embryonic development are listed in the biological process (BP) category, e.g., nervous system development, axon guidance, and generation of neurons (Supplementary Fig. S2, Supplementary Data S2). Additionally, several other terms are summarized under the terms regulation of cell population proliferation and apoptotic process (Supplementary Fig. S2, Supplementary Data S2).

For the miRNAs exclusively found in MSA, several apoptosis-related terms were enriched in the BP category, such as negative regulation of cell growth and negative regulation of extrinsic apoptotic signaling pathway (Fig. 3f, h), which was also complemented by the REVIGO analysis (Supplementary Fig. S3, Supplementary Data S2). Similar to the analysis in the PD group, heart development was another overarching term in the semantic similarity analysis of these BP terms, which contained many terms belonging to developmental pathways such as nervous system development (Supplementary Fig. S3, Supplementary Data S2).

Lastly, we analyzed the annotated terms of the ORA of the PSP miRNA intersections. Here, looking into the MF terms, neurotrophin TrkA receptor binding a term belonging to a family of terms encompassing many neurotrophic pathways associated with neuronal development was one of the most enriched ones among the MF terms enriched in the exclusively absent miRNAs (Fig. 3i). Interestingly, some terms associated with immune system-related functions were among the top terms enriched in the exclusively present miRNAs, such as MHC class II protein complex binding in the MF terms and cellular response to interleukin-7 in the BP terms (Fig. 3j). Semantic similarity analysis revealed apoptotic pathways summarized under the term of apoptotic process in the BP category of the exclusively present miRNAs (Supplementary Fig. S4, Supplementary Data S2). Interestingly, microtubule-based process was also among the terms enriched in this category (Supplementary Fig. S4, Supplementary Data S2).