A Pilot Study of Ex Vivo Human Prefrontal RNA Transcriptomics in Parkinson’s Disease

Abstract

Parkinson’s disease (PD) can dramatically change cortical neurophysiology. The molecular basis for PD-related cortical changes is unclear because gene expression data are usually derived from postmortem tissue collected at the end of a complex disease and they profoundly change in the minutes after death. Here, we studied cortical changes in tissue from the prefrontal cortex of living PD patients undergoing deep-brain stimulation implantation surgery. We examined 780 genes using the NanoString nCounter platform and found that 40 genes were differentially expressed between PD (n = 12) and essential tremor (ET; n = 9) patients. One of these 40 genes, STAT1, correlated with intraoperative 4-Hz rhythms and intraoperative performance of an oddball reaction-time task. Using a pre-designed custom panel of 780 targets, we compared these intraoperative data with those from a separate cohort of fresh-frozen tissue from the same frontal region in postmortem human PD donors (n = 6) and age-matched neurotypical controls (n = 6). This cohort revealed 279 differentially expressed genes. Fifteen of the 40 intraoperative PD-specific genes overlapped with postmortem PD-specific genes, including CALB2 and FOXP2. Transcriptomic analyses identified pathway changes in PD that had not been previously observed in postmortem cases. These molecular signatures of cortical function and dysfunction may help us better understand cognitive and neuropsychiatric aspects of PD.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10571-023-01334-8.

Introduction

Parkinson’s disease (PD) is the second most common neurodegenerative disorder of aging and results in motor and cognitive impairments (Jankovic 2008; Narayanan and Albin 2022). PD involves forebrain aggregation of the protein alpha-synuclein (Singleton et al. 2003; Spillantini et al. 1998) and decreased midbrain dopamine (Hornykiewicz 1998; Narayanan et al. 2013), as well as profound basal ganglia dysfunction (Obeso et al. 2000). A key pathological feature of PD is Lewy bodies, which consist of aggregated forms of alpha-synuclein. Alpha-synuclein aggregates can be found widely throughout the brain including in the cortex (Braak et al. 2001, 2006). Cortical dysfunction in PD patients is thought to contribute to cognitive and neuropsychiatric symptoms (Aarsland and Kurz 2010; de Hemptinne et al. 2013; Narayanan et al. 2013; Williams-Gray et al. 2009).

Past work has focused on protein-level changes in dopamine or synuclein (Braak et al. 2006, 2005; Deutch 1993). However, changes in cortical gene expression might contribute, respond, or drive adaptation to cortical changes in PD. Studies of postmortem tissue have revealed PD-associated genes such as FOXO1 (Dumitriu et al. 2012). Other studies have found PD-associated genes that are specifically related to dementia, including PSMB4, OXR, SLIT2, NSAP, FGF9, and KIF21A (Stamper et al. 2008). There are two primary caveats with postmortem genetic analyses: (1) variation in genetic profile data from tissue samples collected at the end of a patient’s life could be related to multiple factors such as advanced progression of PD or age, and (2) the postmortem interval between death and brain removal might profoundly affect cortical expression profiles. Because of these limitations, it is difficult to collect accurate gene expression data, and it is unclear how PD changes cortical transcriptomics.

We addressed how PD changes cortical transcriptomics from frontal cortical tissue sampled from living, awake PD patients during deep-brain stimulation (DBS) implantation surgery. Tissue was also collected from a non-PD but demographically similar cohort of patients with essential tremor (ET), also undergoing DBS implantation surgery. To complement tissue analyses, we explored how differentially expressed cortical PD genes are linked to neurophysiology and behavior via intraoperative local field potential (LFP) recording during a novelty detection task in the same patients. Finally, we compared our intraoperative gene expression profiles with similar profiles from postmortem autopsy samples from brains of PD patients and neurotypical control donors. Although our pilot exploratory study had small sample sizes, our data provide preliminary insight into cortical function and dysfunction in PD.

Materials and Methods

Collection and Processing of Human Brain Tissue

Brain tissue samples came from 33 individuals in two cohorts: (1) 21 intraoperative specimens from patients undergoing DBS implantation surgery (12 PD patients off medication for surgery, 9 ET) and (2) 12 postmortem donors (6 PD, 6 neurotypical controls). Age and disease duration did not differ across groups (Table 1). All specimens were collected at the University of Iowa. All procedures were conducted under IRB-approved protocols (#201601722 [intraoperative], #202011287 [postmortem]) and performed after obtaining written informed consent from the participant or the family, respectively.

Table 1.

Subject characteristics

Case	Cohort	Age (y)	Sex	Disease group	Disease duration (y)
O1	Intraoperative	73	F	ET	12
O2	Intraoperative	78	M	ET	11
O3	Intraoperative	67	F	ET	25
O4	Intraoperative	77	M	ET	6
O5	Intraoperative	71	F	ET	74
O6	Intraoperative	71	F	ET	11
O7	Intraoperative	70	F	ET	12
O8	Intraoperative	79	F	ET	11
O9	Intraoperative	61	M	ET	15
O10	Intraoperative	73	F	PD	11
O11	Intraoperative	69	M	PD	2
O12	Intraoperative	70	M	PD	3
O13	Intraoperative	71	M	PD	7
O14	Intraoperative	69	M	PD	7
O15	Intraoperative	71	F	PD	11
O16	Intraoperative	67	M	PD	2
O17	Intraoperative	67	M	PD	7
O18	Intraoperative	65	M	PD	4
O19	Intraoperative	66	M	PD	10
O20	Intraoperative	62	M	PD	3
O21	Intraoperative	63	M	PD	6
	Mean	69.5	8F, 13M	9 ET, 12 PD	11.9
A1	Postmortem	84	M	PD	5
A2	Postmortem	74	M	PD	10
A3	Postmortem	83	M	PD	14
A4	Postmortem	69	M	PD	10
A5	Postmortem	70	M	PD	9
A6	Postmortem	66	M	PD	15
A7	Postmortem	72	M	Control	–
A8	Postmortem	62	M	Control	–
A9	Postmortem	74	M	Control	–
A10	Postmortem	75	M	Control	–
A11	Postmortem	74	M	Control	–
A12	Postmortem	69	M	Control	–
	Mean	72.7 (p = 0.12)	0F, 12M	6 PD, 6 Control	10.5 (p = 0.83)

Intraoperative Tissue Collection

Intraoperative tissue was collected during previously scheduled DBS implantation surgery. All PD patients were off dopaminergic medication for the surgery. Since DBS electrode insertion requires a cortical incision, no additional cortical tissue was disrupted for the patients in this study. A standard coronal burr hole placement for usual clinical DBS trajectory and targeting was utilized. These were ~ 14-mm diameter holes, typically located over the middle frontal gyrus or rarely located over the lateral aspect of the superior frontal gyrus (i.e., Brodmann areas 8, 9/46). After dural opening, a corticotomy at the DBS insertion point was made bilaterally under direct vision to avoid visible blood vessels. Through this opening, ~ 2-mm cupped biopsy forceps (~ the same diameter as a DBS insertion cannula) were used to grasp a small biopsy (~ 5 mg) from directly beneath the pial surface (Fig. 1). The collected tissue was immediately passed off the sterile field and frozen in the operating room using dry ice. Samples were collected separately from the left and right hemispheres and pooled together for RNA extraction and downstream analysis. Due to the brevity of tissue collection, no meaningful additional time was added to the surgery. One to three 2-mm diameter stainless steel DBS implantation cannulas were then passed through the same cortical opening, and surgery commenced in the usual manner. No immediate or delayed complications from the tissue collection procedure were observed.

Fig. 1.

Biopsy technique. Intraoperative photographs from deep-brain stimulation (DBS) implantation surgery showing example left prefrontal burr hole location and magnified view of biopsy site (inset). Through a standard 14-mm diameter burr hole and the site of cortical incision for DBS cannula placement (asterisk), ~ 2-mm cupped biopsy forceps were used to grasp cortical samples under direct vision

To compare gene expression profiles from our intraoperative tissue specimens with prior data, we collected postmortem samples from the superior/middle frontal gyrus (BA 8, 9/46), the same frontal region targeted in DBS patients. All postmortem frontal cortex tissue was obtained from the Iowa NeuroBank Core (https://medicine.uiowa.edu/inbc/), with informed consent forms signed by the next of kin as required by state law. The postmortem cases were PD patients or neurotypical control patients available for study without PD or ET.

Gene Expression Analysis Using NanoString nCounter Human Neuropathology Panels

Because of the challenge of intraoperative tissue, comparisons between intraoperative and postmortem cohorts, and small samples, we chose the NanoString nCounter Human Neuropathology Panels. Gene expression profiles of intraoperative tissue collected during DBS implantation surgery and postmortem frontal tissue were generated using identical NanoString nCounter Human Neuropathology Panels (NanoString Technologies, Inc., Seattle, WA). Human tissue samples were submitted to the Iowa NeuroBank Core in the Iowa Neuroscience Institute. Total RNA (100 ng per sample) was extracted using the RNeasy Plus Mini Kit (Qiagen, Germantown, MD) following the manufacturer’s protocol. The quality and concentration of RNA samples were quantified using NanoDrop (Thermo Scientific, Waltham, MA), Fragment Analyzer (Agilent, Clara, CA), and Bioanalyzer (Agilent) systems in the Iowa Neuroscience Institute and in the Iowa Institute of Human Genetics, Genomics Division. Samples with a Bioanalyzer RNA integrity number (an index of RNA quality control) of > 3 were analyzed with the NanoString nCounter Human Neuropathology Panel and Custom CodeSet (C8574, C9676) assays, designed in consultation with the NanoString bioinformatics team. All intraoperative mRNA assays were performed on the NanoString nCounter SPRINT Profiler for image acquisition and data processing, without amplification or generation of cDNA (Geiss et al. 2008). Due to instrumentation updates, the postmortem assays were performed using the NanoString nCounter MAX system, according to the manufacturer’s instructions. Gene expression profiles for a total of 780 genes were screened using target-specific biotinylated probes and barcoded reporter probes in the NanoString nCounter neurodegenerative cartridge (Supplementary Tables 1 and 2).

NanoString Analysis: Differentially Expressed Genes, Databases, and Biological Processes

Reporter Code Count (RCC) and Reporter Library Files (RLF) generated by the nCounter SPRINT Profiler were imported into NanoString nSolver 4.0 Advanced Analysis software, version 2.0.134. The manufacturer’s recommended default parameters were used for imaging, quality control flagging, binding density, positive control linearity, and positive control limit of detection. Each sample in the intraoperative and postmortem cohorts was first normalized to the geometric mean determined by the six External RNA Controls Consortium (ERCC) synthetic RNA Spike-In Controls designated as positive controls (PosA to PosF), with default flagging of normalization factors (< 0.3 and > 3). These ERCC RNA Spike-In Controls were used both within each experiment and across all the nCounter experiments to create a standard baseline measurement of RNA samples across various samples and batches. We then applied normalization using the geometric mean of custom housekeeping genes, with default flagging of normalization factors (< 0.1 and > 10). Thirteen stably expressed housekeeping genes were selected for the intraoperative samples: CCDC127; UBE2D2; ASB7; LARS; MTO1; CSNK2A2; TADA2B; SUPT7L; CNOT10; AARS; FAM104A; CYC1; and RPL13. The geometric mean of the two most stable housekeeping genes, CYC1 and UBE2D2 were selected for normalization in the postmortem cohort. Normalized gene expression datasets were assessed using fold changes, p values, and enriched pathways.

To validate the differentially expressed genes identified in our study and to gain further insight into the function of the most dysregulated target genes, we compared the molecular signatures of human PD found in this study with published resources. These included the Comparative Toxicogenomics Database (CTD) through Harmonizome (https://maayanlab.cloud/Harmonizome/) Davis et al. 2022; Rouillard et al. 2016) and Gene4PD (Li et al. 2021). To identify the molecular function, cellular component, human disease phenotype, and protein–protein interaction network of PD signatures in the intraoperative and postmortem cohorts, all differentially expressed genes were subjected to Gene Ontology (GO) pathway analysis using the PANTHER classification system (http://www.pantherdb.org/) and the results highlight the dysregulated biological process in both intraoperative (Supplementary Table 5) and postmortem cohorts (Supplementary Table 6).

Genotype–Phenotype Correlation Analysis

We leveraged DBS implantation surgery to obtain one index of cognitive function and explore correlation between gene expression and behavior for the same intraoperative cohort. We used an auditory oddball reaction-time task to index cognitive function during surgery as in our prior work (Cole et al. 2022). Oddball reaction-time tasks measure detection of and reaction to infrequently occurring novel stimuli in contrast to frequently occurring standard stimuli. This task is ideal for intraoperative neurophysiology because (1) it uses an auditory domain that does not require a monitor for visual stimuli, (2) the task is relatively simple, and (3) several trials can be collected in a few minutes.

Our auditory oddball reaction-time task consisted of trials in which one of three possible sounds was presented: a standard tone that occurred frequently (500 Hz, 50% of trials); a target sound that occurred infrequently (the word “Go,” 30% of trials; this trial type was used to maintain attention); or a novel sound (unique birdcall, 20% of trials). Patients were instructed to press with their left hand (typically thumb) if they heard the word “Go” and to press with their right hand/thumb if they heard anything else (Cole et al. 2022).

All stimuli were presented at an appropriate volume through either earbuds or an external speaker to be heard clearly over the noise of the operating room. The task and stimuli were presented using PsychToolbox-3 in Matlab2018B (MathWorks, Natick, MA) on a Windows laptop, and responses were captured with a Kinesis pedal (Kinesis Computer Ergonomics, Bothell, WA) that patients held in their hands. The pedals served as large response buttons and made it easier for patients to respond, even if they were not able to see their hands or the response device. Due to its size and weight, the response device was also more resistant to tremor than a typical computer keyboard or mouse, which is especially helpful as patients are off PD medications during surgery.

Patients practiced the oddball reaction-time task with 30 trials (15 standard trials, 9 “Go” trials, and 6 novel trials) preoperatively and again on the morning of surgery. The intraoperative experiment was conducted after the clinically necessary microelectrode recording electrodes were confirmed to be in the subthalamic nucleus, based on clinical recordings. We focused our post hoc analyses on how quickly patients responded to novel tones (reaction time).

Scalp encephalograph EEG was recorded simultaneously in the operating room during the oddball reaction-time task using an EEG electrode (Cadwell Industries, Inc, Kennewick, WA) placed on the forehead at the hairline, which approximated Fz. A reference electrode was placed on the left mastoid. EEG recordings were recorded at 24 kHz and later downsampled to 500 Hz. To adjust for the noisy recording environment of the operating room, pre-processing included filtering the signal below 50 Hz with a low-pass filter, followed by a 1-Hz high-pass filter. To analyze the EEG signals, we first epoched the data (− 500 to 1000 ms relative to each stimulus onset). We analyzed standard trials only, as there were too few novel trials to provide stable power estimates. We then calculated time–frequency power for each trial. We computed spectral measures by multiplying the fast Fourier transformed (FFT) power spectrum of single-trial EEG data with the FFT power spectrum of a set of complex Morlet wavelets (defined as a Gaussian-windowed complex sine wave (e^i2πtfe^{−t∧2/(2xσ∧2)}, where t = time and f = frequency). Wavelets increased from 1 to 50 Hz in 50 logarithmically spaced steps, which defined the width of each frequency band, increasing from 3 to 10 cycles between 1 and 50 Hz and taking the inverse FFT. This process results in estimates of instantaneous power (the squared magnitude of the analytic signal) for each trial. Power for each trial was normalized by converting to a decibel (dB) scale (10*log10 (power_t/power_baseline). The baseline for each frequency was calculated by averaging power for each frequency from − 300 to − 200 ms prior to cue onset. A 100-ms duration is often used as an effective baseline since pixel-wise time–frequency data points have already been resolved over smoothed temporal and frequency dimensions with the wavelets. We took an a priori approach focused on the theta band as a region of interest (ROI; 4–7 Hz, 200–300 ms). This ROI has been strongly justified based on extensive past work in cognitive control tasks, EEG, and PD (Cole et al. 2022; Singh et al. 2021). We conducted Pearson correlation tests between expression levels of genes found to be differentially expressed in the intraoperative tissue study and both behavioral performance (reaction time on novel trials) and time–frequency power on standard trials.

Statistical Analyses

Statistical significance for nCounter assays was determined using an unadjusted p < 0.05 and the Benjamini–Hochberg (BH) procedure to control the FDR at 0.05 (Supplementary Table 3). Data are displayed as means ± SEM and p values of ≤ 0.05 (*), ≤ 0.01 (**), and ≤ 0.001 (***) were considered statistically significant. Pearson correlations between gene expression and reaction time and theta power were run in R and were adjusted for multiple inference using the Holm method via the function rcorr.adjust.

Results

Prefrontal Differentially Expressed Transcripts from Intraoperative and Postmortem Cases

We sought to compare expression landscapes of 780 genes between intraoperative and postmortem samples. To determine differential expression, all RNA expression was normalized to selected housekeeping genes for the intraoperative and postmortem cohorts. NanoString nCounter analyses revealed that the selected housekeeping genes were highly stable within the intraoperative (n = 21 cases; Fig. 2a) and postmortem (n = 12 cases; Fig. 2b) cohorts.

Fig. 2.

The expression of 780 genes in intraoperative and postmortem prefrontal cohorts. The variance of each gene was plotted in log-scaled normalized data against its mean, and highly variable genes were labeled with their gene name. Selected housekeeping genes are color-coded here (blue) and show stability in the intraoperative (a) and postmortem cohorts (b)

Transcripts were considered significantly PD-specific in each cohort if their expression had an unadjusted p < 0.05 (Student’s t test). Forty (11 upregulated, 29 downregulated) significantly differentially expressed genes were identified in the intraoperative cases, and 279 (276 upregulated, 3 downregulated) significantly differentially expressed genes were identified in the postmortem cases. Statistics and p values of all detected genes are in Supplementary Table 3. The pathway scoring plots from Nanostring nCounter analysis software were in the Supplementary Figs. 1 and 2 for both of the intraoperative and postmortem cohorts, including cell structure integrity, neuroinflammation and metabolism related genes within the human nCounter neuropathology panel.

PD-Specific Differential Expression in the Prefrontal Cortex

We were interested in examining gene expression differences specific to PD by comparing samples from PD patients to non-PD controls. In the intraoperative cohort, this involved comparisons with ET control patients also undergoing DBS. In the postmortem cohort, this involved comparisons with neurotypical controls without a PD diagnosis. The differentially expressed genes included LRRK2 (downregulated) in the intraoperative cohort, in addition to SNCA and MAPT (upregulated) in the postmortem cohort. A Venn diagram comparing the list of differentially expressed genes (Fig. 3a) shows the overlapping and unique differentially expressed genes between the intraoperative and postmortem comparisons. Overlapping differentially expressed genes between the intraoperative and postmortem comparisons demonstrated clear differences between PD and controls (Fig. 3b). GNPTAB (N-acetylglucosamine-1-phosphate transferase subunits alpha and beta), FOXP2 (forkhead box P2), CDK5R1 (cyclin dependent kinase 5 regulatory subunit 1), and NRXN1 (neurexin 1) were commonly downregulated in PD in both comparisons.

Fig. 3.

Differentially expressed genes in PD. a Venn diagram comparing differentially expressed genes specific to PD between intraoperative and postmortem cohorts. Hierarchically clustered heat map of overlapped genes in PD from intraoperative (blue circle) and postmortem (pink circle) cohorts. b Upregulated genes are shown in red, and downregulated genes are shown in green for each cohort. Log2 fold change (FC) of gene expression changes between groups on a logarithmic scale to base 2

To examine the extent to which the 40 intraoperative differentially expressed genes correlated with behavioral performance, we used an oddball reaction-time task (Fig. 4a) and found that higher STAT1 (signal transducer and activator of transcription 1) expression was associated with slower reaction time on oddball trials (r = 0.75, p = 0.02). Interestingly higher IDE (insulin degrading enzyme) expression was associated with faster reaction time (r = − 0.68, p = 0.04). We also examined how these 40 genes correlated with 4-Hz rhythms (Fig. 4b). Higher STAT1 expression was associated with less theta power 300–400 ms after the standard cue (r = − 0.72, p = 0.03). IDE was not significantly related to theta power, but higher IL10RA (interleukin 10 receptor subunit alpha) expression was correlated with less theta power (r = − 0.70, p = 0.04).

Fig. 4.

Combined assessments of mRNAs, behavioral phenotypes, and prefrontal 4-Hz activity. Scatterplots showing statistically significant correlations between (a) gene expression and reaction time on novel oddball trials for STAT1 (upper panel) and IDE (lower panel). Correlation was also identified between gene expression and theta power (b) for STAT1 (upper panel) and IL10RA (lower panel)

Comparison of Prefrontal PD-Specific Expression with Other Studies

Using the 40 genes identified in the intraoperative cohort and the 279 genes identified in the postmortem cohort, we found that the majority of differentially expressed genes were unique from the comparison datasets of other studies (Supplementary Table 4). Common differentially expressed genes were reported across three postmortem studies (current postmortem, CTD (Davis et al. 2022) and Gene4PD (Li et al. 2021); Fig. 5) including SNCA, MAPK1, MAPK3, ATF4, SIRT1, RELA, CASP7, TGFB1, NFKBIA, and GSTP1. Our results showed CALB2, CCND1, BCL2 and ITP3 were expressed in our data and in other publicly available postmortem PD datasets (Fig. 5). Interestingly, FOXP2 downregulation was found in our intraoperative cohort and a recent intraoperative study based on living PD and cancer patients, (Benoit et al. 2020) but showed upregulation in our postmortem cohort.

Fig. 5.

Comparative analysis of current and previously published PD studies and databases. Venn diagram shows the number of PD-related genes collected from the intraoperative (OR; orange oval) and postmortem (green) cohorts in the current study and two publicly available postmortem datasets, CTD (blue) and Gene4PD (yellow)

Functional enrichment analysis revealed enrichment of PD genes associated with calcium channels (gene ontology (GO): 0099604, FDR 7.14E-05) in the intraoperative cohort and with protein binding (GO: 0044877, FDR 1.44E-22) in the postmortem cohort (Supplementary Table 5), reflecting the role of calcium binding proteins (e.g., CALB2) in the progression of PD.

Discussion

Our goal was to begin to characterize the molecular basis of cortical dysfunction in PD from living cortical tissue. This study was based on both intraoperative tissue samples from patients undergoing DBS implantation surgery and postmortem specimens, and therefore is an advance from prior postmortem-based studies. We found 40 differentially expressed genes in the intraoperative cohort with PD, including 15 that overlapped with our postmortem cohort. Some PD-related genes were downregulated in the intraoperative cohort but upregulated in the postmortem cohort. As suggested by stage-focused RNA-Seq transcriptomic studies in PD and AD (Benoit et al. 2020; Mathys et al. 2019), we found that very few differentially expressed genes had the same directional changes in expression across cohorts. Whether these directional alterations are predictive of disease states is an important question. However, overlapping alterations in PD-related genes appear to be an important common mechanism in calcium signaling and neuronal survival. Overlapped results of these upregulated genes in the postmortem cohort might imply a compensatory response to impaired signaling pathways. Data from this pilot study provide insight into the cortical molecular changes in PD.

CALB2 is consistently identified in pre-existing database cohorts of PD and is associated with calcium influx, neuronal excitability, neurotransmitter release, and pathogenesis of PD (Li et al. 2019; Ma et al. 2014; Mura et al. 2000; Surmeier 2007). In addition to colocalizing with α-synuclein, CALB2 is dysregulated in both in vivo and in vitro models of PD (Gries et al. 2021; Mura et al. 2000). Interestingly, given that FOXP2 is implicated in speech production and behavioral flexibility (Co et al. 2020; Fisher and Scharff 2009), we found that FOXP2 is consistently dysregulated in the intraoperative and postmortem cohorts of this study and a previous intraoperative study (Benoit et al. 2020). CCND1 and BCL2 were also identified in intraoperative, postmortem, and pre-existing datasets of PD. These results provide insight into cortical physiology of PD. Since up to 80% of patients with PD are expected to suffer from cognitive symptoms, including memory deficits, there is an increasing need to study and understand the importance of mixed neuropathology for developing both Alzheimer’s disease (AD) and Lewy body dementia in PD.

We identified several AD and related dementia and PD genes that were altered in the postmortem cases with PD. These included upregulated transcripts associated with α-synuclein, Amyloid-beta (Aβ) and tau proteins, such as SNCA, SNCAIP, APP, APOE, and MAPT in our postmortem cohort with PD. Alterations in these common neurodegeneration genes provide further support for the intersection between PD and AD neuropathology. This pilot study can serve as a hypothesis-generating experiment for follow-up mechanistic, biomarker, and therapeutic studies. Our study also complements existing large-scale PD RNA sequencing datasets that have two separate cohorts associated with PD (Benoit et al. 2020; Li et al. 2021; Rouillard et al. 2016).

Our pilot study has several limitations. First, our postmortem and intraoperative PD sample sizes constrained our analyses, although we note that this is, to our knowledge, the largest intraoperative cortical tissue sample in PD. We found interesting correlations between the oddball reaction-time task, theta power, and the STAT1 gene, which is a neuroimmune gene that may play a role in brain injury and PD (Qin et al. 2016; Takagi et al. 2002), although this analysis only involved 8 patients, and future work needs to replicate these analyses in independent cohorts to establish the reliability and robustness of these findings. A second limitation is that tissue collection conditions between the intraoperative and postmortem cohorts were quite different. These distinct conditions and small sample sizes may have affected our sensitivity to detect differentially expressed genes. We chose NanoString nCounter assay given these tissue limitations because these assays can easily accommodate extremely low RNA yields from the small intraoperative surgical tissue harvested in the living patients, and directly count individual RNA transcripts. The NanoString nCounter assay also does not require further enzymatic reactions, amplification, and cDNA conversion. To compare with suboptimal RNA samples from the human postmortem brain samples, nCounter assays can use very small probe pairs to measure RNA transcripts from partial degraded samples directly and does not require additional steps that may be needed for whole transcriptomics RNA sequencing. Due to the small sample sizes, the 800-gene panel of NanoString gave us effective statistical power to evaluate both technical performance and major biological pathways.

Third, our sample sizes did not permit rigorous examination of potential effects of covariates, such as sex and age. We had more female ET cases and more male PD in the intraoperative cohort due to due to the nature of ET and PD and our subject pools, which may have influenced our results. Future studies will include sufficient numbers of patients to balance sex differences between groups. To obtain spatial and temporal profiles of candidate genes at a higher resolution, methods such as in situ hybridization or cutting-edge spatial transcriptomics technologies will be of value in future work. Although our intraoperative biopsies were done prior to DBS implantation, in future studies it will be of value to identify the effects of DBS on PD molecular signatures in a separate cohort.

In summary, we present novel evidence of cortical transcriptomics of PD. Data from this approach might help illuminate cortical function and dysfunction in PD and may contribute to our understanding of cognitive and neuropsychiatric symptoms of PD.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

DBS

Deep brain stimulation

PD

Parkinson’s disease

RCC

Reporter Code Count

RLF

Reporter Library File

Author Contributions

LL, JG, and NN conceived the study, designed, and supervised the experiment, and interpreted results. LL, JG, and RC carried out the intraoperative tissue collection and transportation. LL prepared RNA samples, performed the NanoString nCounter expression profile studies, and provided human postmortem brain tissue and data interpretation. RC, JG, and NN performed the genotype–phenotype correlation studies. LL and NN wrote the manuscript. All authors discussed and commented on the manuscript.

Funding

NINDS P20NS123151-01.

Data Availability

Data are available at https://narayanan.lab.uiowa.edu/article/datasets and shared via NIH GEO at https://www.ncbi.nlm.nih.gov/geo/ with the accession number: GSE227990.

Declarations

Competing interests

All authors have no financial interests.