Deep brain stimulation induces rapidly reversible transcript changes in Parkinson's leucocytes

Abstract Subthalamic deep brain stimulation (DBS) reversibly modulates Parkinson's disease (PD) motor symptoms, providing an unusual opportunity to compare leucocyte transcripts in the same individuals before and after neurosurgery and 1 hr after stimulus cessation (ON- and OFF-stimulus). Here, we report DBS-induced reversibility and OFF-stimulus restoration in 12 of 16 molecular functions and 3 of 4 biological processes shown in exon microarrays to be differentially expressed between PD patients and controls, post-DBS from pre-DBS and OFF from ON states. Intriguingly, 6 of 18 inflammation and immune-related functions exhibited reversibility, and the extent of stimulus-induced changes correlated with the neurological DBS efficacy, suggesting mechanistic implications. A minimal list of 29 transcripts that changed in all three comparisons between states discriminated pre-surgery and OFF states from post-surgery and controls. Six of these transcripts were found to be able to distinguish between PD patients and both healthy controls and patients with other neurological diseases in a previously published whole blood 3’ array data study of early PD patients. Our findings support the future use of this approach for identifying targets for therapeutic intervention and assessing the efficacy of current and new treatments in this and other neurological diseases.

. Hierarchical clustering (HCL) classifies PD patients post-STN on stimulation from controls accurately -Page2 References - Page6 Table S1a. Clinical parameters of the participating patients -Page7 Table S1b. .RNA integrity numbers and microarray hybridization batch number -Page7 Table S2. Clinical reflection of DRT medical therapy and DBS neurosurgery -Page7 Table S3. Quantitative real time PCR (qRT) primers -Page8 Table S4a -Differentially expressed genes detected in PD patients compared to HC based on exon array median gene level summaries iterative permutation T-test analysis.
-Page8 Table S4b. Differentially expressed genes detected in PD patients post STN-DBS on electrical stimulation compared to pre STN-DBS based on exon array median gene level summaries T-test analysis.
-Page9 Table S4c. Differentially expressed genes detected in PD patients post STN-DBS following one hour off electrical stimulation compared to post STN-DBS on electrical stimulation based on exon array median gene level summaries T-test analysis (patients were on identical medication dosage).
-Page11 Table S4d. Differentially expressed genes detected in PD patients post STN-DBS on electrical stimulation compared to HC based on exon array median gene level summaries T-test analysis.
-Page15 Table S5a. Post-hoc functional enrichment analysis of the genes detected as changed in patients compared to HC.
-Page19 Table S5b. Post-hoc functional enrichment annotation analysis of the genes detected as changed post STN-DBS on electrical stimulation compared to pre STN-DBS state.
-Page19 Table S5c. Post-hoc functional enrichment annotation analysis of the genes detected as changed post STN-DBS off electrical stimulation (sustained for 1 hour) compared to on stimulation state (patients were on identical medication dosage in both states).
-Page20 Table S5d. Post-hoc functional enrichment annotation analysis of the genes detected as changed post STN-DBS on electrical stimulation compared to HC.
-Page25 Table S6c. GO KS and Fishex Molecular Function (MF) ad-hoc analysis of PD patients pre-DBS compared to post STN-DBS on stimulation.
-Page29 Table S6d. GO KS and Fishex BP ad-hoc analysis of PD patients pre-DBS compared to post STN-DBS on stimulation.
-Page32 Table S6e. GO KS and Fishex MF ad-hoc analysis of PD patients pre-DBS compared to post STN-DBS on stimulation.
-Page36 Table S6f. GO      We predicted that the discriminative power of those genes that exhibited exon-level changes in advanced PD patients before and after surgery and in stim-OFF conditions will also stand for early PD patients. We compared the list of 29 signature identified genes (shown in Fig. 5A) to our previous results from the other data set (4). Of note, the previous dataset was derived from 3', rather than exon arrays, and would therefore represent a partial picture of the disease-induced changes. Nevertheless, 6 out of the currently selected 29 transcripts were also changed between early PD to HC (ANOVA p<0.01) in the independent cohort data, creating a minimalistic molecular signature (SF 7). Thus, we compared our current findings to our previous analysis [5] of a published dataset of PD blood cells in an independent larger cohort of 47 early PD patients, 21 healthy controls (HC) and 39 patients with other neurological diseases (NC) [6]. The 6 gene signature was intriguingly composed of transcripts involved with stress and inflammation, RNA metabolism and signaling, and ubiquitination, transcription and epigenetics control.
The 6 shared genes were the transcription factor NR2F1 (also termed COUP-TF1) [1]; PCBP2, a poly (rC) binding protein that affects RNA turnover [2]; PTPN1, a PD-associated protein tyrosine phosphatase; which was recently reported by others as correlated to PD through genome-wide SNP association study [3]. HNRPDL, a member of the hnRNP family which controls exon inclusions [4], PJA1, which is abundantly expressed in the brain where it is involved in protein ubiquitination [7]; and TRAM1, and endoplasmic reticulum membrane glycoprotein which is involved early in protein translocation [8]. Of note, TRAM1 induces interferonbeta [9] through coupling with Toll-like receptor 4 (TLR4) that mediates bacterial infection responses [10],[11] Unsupervised HCL, on these 6 transcripts alone, clustered correctly all PD pre-DBS from HC (Fig. 6A), and differentiated between most of the PD pre-DBS from the post-surgery state (Fig. 6B). HCL on the 3' datasets of those 6 genes using the larger PD cohort separated the early PD patients from HC (Jaccard distance, Fig. 6C), with the exception of 4 healthy controls who segregated with the PD patients (8.5% false positives). Also, two PD patients segregated with the controls, constituting 2/47 (4.25%) false negatives. Furthermore, those 6 transcripts separated between PD patients to those with other neurological diseases (NC) (Fig. 6D) yet more rigorously (7.14 and 0% false positive and false negative cases).
(a) Six-Transcript Signature classifies early PD patients from healthy controls and other neurological diseases. Shown is HCL based on the 6 signature genes detected as differentially expressed between all of the currently tested clinical states and which were also identified as changed in a larger cohort of early PD patients, at first diagnosis (starred in Fig. 5). (A) HCL discriminated advanced PD (n=7) from healthy control (n=6) subjects of the current study. (B) A less rigorous discrimination of advanced PD from stim-ON state (n=7) of the current study observed. (C) In the independent leukocytes data set (4), HCL based on the 3' arrays expression values of the above triply modified genes (using the Jaccard distance) clustered correctly early PD patient samples (n=47) as distinct from HC (n=21) samples. (D) HCL classification of early PD patients as distinct from patients with other neurological diseases (NC) (n=30) (see (4) for neurological patients description). PTPN1, PCBP2 and PJA1 exhibited the same directional change in both the 3' and exon array data sets.   Age, red and white blood counts pre-and post-STN stimulation, body mass index (BMI), disease duration in years and average blood pressure pre-STN stimulation are given for the patients participants.

Page8
The table lists measured clinical parameters and post-to pre-DBS gene expression personal differences. The clinical parameters are: Unified Parkinson's Disease Rating Scale (UPDRS) motor scale (section III), disease duration since first diagnosis (in years), relative UPDRS-III score change post-compared to pre-DBS, absolute UPDRS-III change post-compared to pre-DBS and levodopa equivalent dosage (LDE). The differential expression parameters are the average and standard deviation of the normal curve fitted to the histogram of per gene fold change per patient (personal fold change computation) of the differentially expressed genes detected in all the patients post-compared to pre-DBS (n=465 genes). The UPDRS-III score of participating patients (n=7) differs between six different states: pre-DBS on med, pre-DBS off med, post-DBS on med on stim, post-DBS off med on stim, post-DBS on med off stim and post-DBS off med off stim. The measured score post-DBS off med/stim is available only where patients agreed to be examined and conducted apart from the blood test (in a different day) by a neurologist. Generally, pre-DBS (on med) clinical motor state is better than off med (i.e lower UPDRS) and best under post-DBS on stim, on med; deteriorating fast upon 1 hour OFF electrical stimulus. The relative UPDRS change pre-on med compared to post-DBS on med, on stim (pre compared to post-ON) was calculated as: ((UPDRS PRE on med-UPDRS POST on med on stim)/UPDRS PRE on med)*100. The absolute UPDRS change post-compared to pre-DBS is (UPDRS PRE on med)-(UPDRS POST DBS on med on stim). All patients and all control subjects were males. Patients varied in disease duration since first diagnosed as PD patients.  Table S5a. Post-hoc functional enrichment analysis of the genes detected as changed in patients compared to HC.
Analysis conducted on the Gene Ontology (GO) FAT (narrowed) database of Biological Process (BP) sub-type with DAVID EASE program(*,**) on exon array detected genes as changed pre STN-DBS as compared with healthy control (HC) subjects. Homo sapiens genome served as background list. Terms enrichment scores are given. The total list of genes included 118 Homo sapiens genes (identified by DAVID database out of 173 detected as differentially expressed). Total population hits: 13,528. Note that the functional clustering enrichment results presented in figure 1 are of cluster specific functional analysis (based on the HCL clusters) rather than the whole group of genes. Analysis conducted on the Gene Ontology (GO) FAT (narrowed) database of Biological Process (BP) sub-type with DAVID EASE program(*,**) on exon array detected genes as changed post STN-DBS aon stimulation as compared with pre-STN-DBS state of the same patients. Homo sapiens genome served as background list. Terms enrichment scores are given. The total list of genes included 309 Homo sapiens genes (identified by DAVID database out of 465 detected as differentially expressed). Total population hits: 13,528. Note that the functional clustering enrichment results presented in figure 1 are of cluster specific functional analysis (based on the HCL clusters) rather than the whole group of genes. Last column: Fold Enrichment score.  Table S5c.   Table S5d. Post-hoc functional enrichment annotation analysis of the genes detected as changed post STN-DBS on electrical stimulation compared to HC.
Results of post-hoc functional Gene Ontology (GO) Biological Process (BP) enrichment analysis (conducted with DAVID EASE program*,**) on exon array detected genes as changed post STN-DBS on stimulation compared to HC. BP GO FAT terms enrichment scores are given. BP GO FAT terms enrichment scores are given. Analysis conducted on N=302 DAVID identified Homo sapiens genes (out of 321 genes detected as differentially expressed) on Homo sapiens genome background . Total population hits: 13,528. Note that the functional clustering enrichment results presented in figure 1 are of cluster specific functional analysis (based on the HCL clusters) rather than the whole group of genes. Last column: Fold Enrichment score.