Supplementary Table S1. Gene lists for gene set enrichment analysis Supplementary Table S2. Abbreviations of cortical subfields
Figure captions
Figure 1. Flow chart of the study. (A) In situ hybridization. Traumatic brain injury (TBI) was induced by lateral fluid-percussion injury (FPI) (Kharatishvili et al., 2006).
(B) Generation of cortical unfolded maps for quantification of PV-positive interneuron numbers. Rats were perfused for histology at 1-month post-TBI. Coronal sections were stained for PV, NeuN, and thionin. Computer-generated plots were generated from PV preparations to assess the density and location of PV neurons in the lesioned cortex.
The plots were superimposed onto the cortical unfolded maps generated from thionin- stained sections. (C) Previous mRNA sequencing data of the perilesional cortical tissue were used for gene set enrichment analysis (GSEA) to detect alterations in PV- related gene expression. Three regulated genes were validated by quantitative reverse transcription polymerase chain reaction (qRT-PCR). Abbreviations: Fluid percussion injury, FPI; Gene set enrichment analysis, GSEA; neuronal nuclei, NeuN; Parvalbumin, PV.
Figure 2. In situ hybridization of Pvalb mRNA at 6 h, 24 h, 10 d, and 4 months post- TBI as compared with controls. (A) A digitized in situ hybridization film of a representative control case. Measured regions of interests (ROIs) are laid over the right hemisphere. The representative cortical areas for each ROI are also listed here.
(B) Representative digitized film showing a section from a rat at 10 d post-TBI. Intensity of in situ hybridization signal of each ROI in (C) deep layers (IV-VI) and (D) superficial
(I-III) layers of the cortex at different time-points. In general, gene expression was more reduced ventral than dorsal to the lesion core. At 6 h post-TBI, gene expression was decreased in the superficial and deep layers of the auditory areas (ROI3-4; 14%-44%
remaining as compared to controls, ROI3; superficial U9,6=2, p<0.01, deep U9,6=10, p<0.05, ROI4 superficial and deep; U9,6=0, p<0.001) and somatosensory areas (ROI5;
24%-29%, superficial U9,6=0, p<0.01, deep U9,6=2 p<0.01). At 24 h post-TBI, Pvalb gene expression was still reduced in the auditory areas (ROI2; 44%, superficial U9,5=6, p<0.01, ROI3-4; 3-26%, ROI3; superficial and deep U9,5=0, p<0.01, ROI4; superficial U9,5=1, p<0.01, deep U9,5=2, p<0.01). Instead, the intensity of Pvalb gene expression was increased in the superficial layers of the ectorhinal area (ROI1; 188%, U9,5=6, p<0.05). At 10 d post-TBI, gene expression was reduced in auditory areas (ROI2; 13- 17%, superficial U9,4=0, p<0.01, deep U9,4=1, p<0.01, ROI3-4; 0-23%, ROI3; superficial and deep U9,4=0, p<0.01, ROI4; superficial U9,4=0, p<0.01, deep U9,4=1, p<0.01), together with somatosensory and parietal association areas (ROI5-9; 21%-88%, ROI5;
superficial U9,4=4, p<0.05, ROI6; superficial U9,4=0, p<0.05, ROI7; superficial U9,4=4, p<0.05, deep U9,4=3, p<0.05, ROI8; superficial U9,4=0, p<0.01, ROI9; superficial U9,4=4, p<0.05) and retrosplenial dysgranular cortex (ROI11; 59%, deep U9,4=2, p<0.05). The level of Pvalb gene expression recovered partly within 4 months post- TBI. Scale bar equals 1.5 mm in panels A-B. Abbreviations: TBI, traumatic brain injury;
ROI, region of interest; *, p<0.05; **, p<0.01; ***, p<0.001 at 6 h post-TBI; #, p<0.05;
##, p<0.01 at 24 h post-TBI; ¤, p<0.05; ¤¤, p<0.01 at 10 d post-TBI; &, p<0.05; &&, p<0.01 at 4 months post-TBI.
Figure 3. Unfolded maps of the 6 TBI cases (A-F), showing the cortical lesion location (black dashed line) and extent and severity of reduction (% of that in controls) in the density of parvalbumin (PV) immunoreactive (ir) interneurons (shades of red and blue) at 1-month after lateral fluid-percussion injury (FPI). The Y-axis shows the rostrocaudal level from bregma (mm). Rat #1 had a large lesion that extended from bregma -2.3 mm to -8.0 mm. Patchy loss of PV-ir cells was centered on the somatosensory and visual cortices. In the rat #4, the lesion extended from -1.6 mm to –7.3 mm. Interneuron loss centered on the somatosensory and visual cortices and also to auditory and association cortices. In rat #5, a large lesion extended between bregma levels -2.5 mm to -8.3 mm. PV-ir neuronal loss was most robust in the auditory, entorhinal, somatosensory, and visual cortexes. Rat #6 had a lesion between -2.6 mm to -7.4 mm.
Interneuron loss was centered on the auditory and entorhinal cortexes. Some patchy interneuron loss was also seen in the somatosensory, audiovisual, and visual cortexes.
Rat #7 had a lesion between -3.3 mm and -7.0 mm. Loss of PV-ir cells was centered on the auditory and somatosensory cortexes. Rat #13 had a lesion from bregma -0.9 mm to -8.9 mm and loss of PV-ir cells was centered on the somatosensory and visual cortexes.
Figure 4. Parvalbumin (PV) and neuronal nuclei (NeuN) immunoreactivity in rat #4 at 1 month after lateral fluid-percussion injury (FPI). (A) A computer-generated plot (-3.3 mm from bregma), demonstrating the distribution of PV-immunoreactive (ir) neurons in the cortex (each red dot is a single PV-ir interneuron). In the plot, numbers (1-3) refer to the cytoarchitectonic areas from which the higher-magnification photomicrographs in panels B1-3 were obtained (1 S1BF, 2 LPtA, 3 MPtA). (C) A photomicrograph of a thionin-stained section with different cortical layers marked (I- VI). NeuN-immunostained panels D1-3 match with areas in panels A1-3. Note the severe loss of PV-ir cells. Scale bar equals 100 µm (all panels).
Figure 5. Parvalbumin (PV) and neuronal nuclei (NeuN) immunoreactivity in rat #7 at 1 month after lateral fluid-percussion injury (FPI). (A) A computer-generated plot (-1.3 mm from bregma), demonstrating the distribution of PV-ir neurons in the cortex (each red dot is a single PV-ir interneuron). In the plot, numbers (1-3) refer to the cytoarchitectonic areas from which the higher-magnification photomicrographs in panels B1-3 were obtained (1 S1ULp, 2 S1BF, 3 S1HL). (C) A photomicrograph of a thionin-stained section with different cortical layers marked (I-VI). NeuN- immunostained panels D1-3 match with areas in panels A1-3. Scale bar equals 100 µm (all panels).
Figure 6. Parvalbumin (PV) and neuronal nuclei (NeuN) immunoreactivity in rat #13 at 1 month after lateral fluid-percussion injury (FPI). (A) A computer-generated plot (- 1.3 mm from bregma), demonstrating the distribution of PV-ir neurons in the cortex (each red dot is a single PV-ir interneuron). In the plot, numbers (1-3) refer to the cytoarchitectonic area, from which the higher-magnification photomicrographs in panels B1-3 were obtained (1 S1ULp, 2 S1BF, 3 S1HL). (C) A photomicrograph of a thionin-stained section with different cortical layers marked (I-VI). NeuN-
immunostained panels D1-3 match with areas in panels A1-3. Scale bar equals 100 µm (all panels)
Figure 7. Parvalbumin (PV) immunoreactivity in the ipsilateral cerebral cortex at 1 month after lateral fluid-percussion injury (FPI) (rat #4). (A) A computer-generated plot demonstrating the distribution of PV-ir neurons in the ipsilateral cortex (bregma -3.3 mm). Each dot is a single PB-positive neuron. The box shows the region from which the higher magnification photomicrographs in panels B (PV immunopositive basket- like terminals, arrows) and C (thionin stained adjacent section) were obtained. (D) Thionin counter-stained PV section of rat #4 with severe loss of PV-ir interneurons.
Note the numerous basket-like axonal terminals surrounding thionin-stained somata (black arrows). Scale bar equals 50 µm (all panels).
Figure 8. Double-labeling of lectin from Wisteria floribunda (WFA, A.) and Fluoro-Jade B (FJB, B.). In the panels, cells that were positive to WFA and FJB are marked with black arrows in panel A and red arrows in panel B. White arrow points to a cell immunoreactive to WFA but not to FJB. Scale bar represents 2 µm (both panels).
Figure 9. Correlation-clustering co-analysis of the parvalbumin-related gene network at 3 months post-TBI. A strong positive correlation was found between the expression of Pvalb and Gabrd, Cck, and Slc32a1 genes (cluster 1). Another cluster (cluster 2) with a high positive correlation with Pvalb included Sst, Npy, Slc3a1, Gabra1, Gabrg2, and Gad1. Abbreviations: TBI, traumatic brain injury.
Figure 10. Downregulation of parvalbumin-related gene network and GABAA receptor subunits at 3 months post-TBI. (A-C) qRT-PCR validation indicated that expression of Gabrg2 (log2FC=-0.21, U6,7=6, p˂0.05) and Gabrd (log2FC=-0.36, U6,7=5, p<0.05) was downregulated in the perilesional cortex. We also found a strong trend towards downregulation of Gabra1 (log2FC=-0.23, U6,7=12, p˃0.05). (D) total of 45% (49/109) of the genes on the PV gene list were enrichened in the GSEA analysis (ES=-0.56, FDR<0.05). Abbreviations: RT, reverse transcriptase; qPCR, quantitative polymerase chain reaction; TBI, traumatic brain injury. Statistical significances: *, p<0.05 (Mann- Whitney U).
Table 1. Antibodies used in immunohistochemistry
ANTIBODY ID ANTIBODY NAME TARGET ANTIGEN
CATALOG NUMBER
CLONALITY HOST ORGANISM
DILUTION AB_10000343 Monoclonal
anti parvalbumin antibody
Parvalbumin Swant, 235 monoclonal antibody
mouse 1:15 000
AB_2298772 Anti-NeuN, clone A60 antibody
NeuN clone A60
Millipore, MAB377
monoclonal antibody
mouse 1:4 000
AB_2620171 Wisteria floribunda agglutinin,
biotin conjugate
N-acetyl-D- galactosamin e
Sigma- Aldrich, L-1516
N/A N/A 1: 4 000
Table 2. Distribution of parvalbumin-immunoreactive (PV-ir) neuron loss in various cortical subfields at 1 month after lateral fluid-percussion injury.
The cortical subfields analyzed are shown in the left and the case numbers on the top. Grey shading indicates the location of the lesion core in each of the six cases. The number of remaining PV-ir neurons in each cortical subfield in each case is shown as a percentage of that in sham-operated experimental controls. Green (no patchy microlesions), yellow (mild PV-ir neuronal loss), orange (moderate PV-ir neuronal loss) or red (severe PV-ir neuronal loss) colors indicate the increasing severity of PV-ir neuronal loss. Data was collected from the 2D- unfolded maps presented in Figures 3A-F. Abbreviations: PV, parvalbumin; TBI, traumatic brain injury.
Abbreviations of cortical subfields are listed in Supplementary table S2.
#1 #4 #5 #6 #7 #13
Au1 0-80 % 0-90 % 0-80 % 51-90 % 41-90 % 0-70 %
AuD 0-30 % 0-40 % 0-60% 0-90 % 0-90 % 0-90 %
AuV 41-80 % 41-90 % 0-70 % 61-90 % 41-90 % 0-70 % No patchy microlesions
TeA 21-60 % 51-90 % 0-90 % 61-90 % 61-90 % 0-70 % Mild PV-ir neuronal loss
DI 81-90 % 51-90 % 71-90 % 51-60 % 51-71 % 41-90 % Moderate PV-ir neuronal loss
Ect 61-80% 71-90 % 51-90 % 61-90 % 71-90 % 0 -20 % Severe PV-ir neuronal loss
GI 91-100 % 91-100 % 71-80 % 91-100 % 81-90 % 41-90 % Lesion core
LPtA 51-80 % 0-90 % 91-100 % 91-100 % 71-90 % 31-90 % M1 81-90 % 51-90 % 81-90 % 91-100 % 81-90 % 51-80 % M2 71-80 % 81-90 % 81-90 % 91-100 % 91-100 % 61-70 % MPtA 71-90 % 40-60 % 91-100 % 91-100 % 81-90 % 91-100 % PRh 61-80 % 91-100 % 41-90 % 61-90 % 71-90 % 11-90 % PtPD 81-90 % 21-90 % 91-100 % 81-90 % 71-80 % 11-80 % PtPR 0-80 % 41-90 % 61-70 % 81-90 % 41-90 % 0-50 % RSD 61-90 % 31-60 % 91-100 % 91-100 % 71-90 % 91-100 %
S1 0-50 % 0-90 % 0-90 % 0-30 % 21-90 % 0-50 %
S1BF 71-90 % 31-90 % 81-90 % 51-90 % 51-90 % 41-90 % S1DZ 81-90 % 31-90 % 81-90 % 91-100 % 71-80 % 71-80 % S1FL 81-90 % 71-80 % 91-100 % 91-100 % 71-90 % 61-90 % S1HL 81-90 % 51-90 % 81-90 % 91-100 % 71-90 % 51-90 % S1Sh 91-100 % 81-90 % 71-90 % 91-100 % 71-90 % 41-60 % S1Tr 51-80 % 31-90 % 71-80 % 91-100 % 81-90 % 91-100 % S1ULp 81-90 % 71-90 % 81-90 % 51-90 % 61-80 % 51-90 % S2 81-90 % 71-90 % 41-90 % 71-90% 61-90 % 71-90 % V1 71-80 % 0-60 % 81-90 % 81-90 % 21-60 % 61-80 % V1B 41-80 % 31-60 % 71-90 % 91-100 % 91-100 % 21-90 % V1M 41-90 % 11-80 % 71-90 % 71-90 % 91-100 % 41-90 %
V2L 0-80 % 0-90% 0-90 % 81-90 % 41-90 % 0-90 %
V2ML 41-90 % 0-60 % 71-80 % 71-90 % 41-90 % 71-90 % V2MM 61-90 % 31-60 % 71-90 % 71-80 % 81-90 % 71-90 %
Table 3. Gene set enrichment analysis of parvalbumin-related genes.
GENE RANK IN GENE LIST RANK METRIC SCORE RUNNING ES LOG2 FOLD CHANGE NOMINAL P-VALUE
Bdnf 14682 -7699 -0,142 -0,937 2,11E-06
Chrm2 14588 -7603 -0,204 -0,863 1,48E-05
Sstr3 14838 -7856 -0,013 -0,808 3,47E-09
Chrm4 14755 -7773 -0,060 -0,715 1,73E-07
Cckbr 14702 -7720 -0,108 -0,714 1,07E-06
Kcnj6 14508 -7522 -0,284 -0,680 5,53E-05
Sstr4 13731 -6742 -0,442 -0,668 4,81E-03
Gabrd 14811 -7829 -0,046 -0,661 1,61E-08
Kcnj3 14664 -7681 -0,158 -0,656 3,18E-06
Gad1 14910 -7928 0,000 -0,654 2,23E-16
Glrb 14275 -7287 -0,319 -0,640 3,28E-04
Tacr3 14561 -7576 -0,236 -0,636 2,52E-05
Gad2 14480 -7494 -0,299 -0,629 7,07E-05
Slc32a1 14637 -7653 -0,173 -0,619 5,61E-06
Chrna4 14582 -7597 -0,221 -0,609 1,75E-05
Npy5r 13695 -6705 -0,454 -0,606 5,44E-03
Kcnj9 14837 -7855 -0,030 -0,596 3,73E-09
Ache 14618 -7634 -0,189 -0,592 8,72E-06
Gabrg2 14530 -7545 -0,251 -0,584 3,89E-05
Th 12409 -5407 -0,517 -0,575 8,01E-02
Gabra1 14130 -7141 -0,406 -0,555 8,53E-04
Sstr2 13864 -6875 -0,435 -0,538 2,86E-03
Erbb4 13170 -6173 -0,491 -0,496 2,35E-02
Npy 14739 -7757 -0,076 -0,495 3,51E-07
Chrm3 14524 -7539 -0,268 -0,483 4,49E-05
Ocm2 11997 -4988 -0,548 -0,458 1,26E-01
Sstr1 14158 -7170 -0,392 -0,457 7,68E-04
Gabrb3 14202 -7214 -0,346 -0,434 5,44E-04
Chrm1 13998 -7009 -0,413 -0,425 1,73E-03
Gabra5 12079 -5072 -0,530 -0,410 7,16E-04
Ppp1r9a 14470 -7484 -0,315 -0,395 7,82E-05
Gabbr2 14711 -7729 -0,092 -0,393 8,10E-07
App 14687 -7704 -0,125 -0,383 1,98E-06
Gng4 12070 -5063 -0,541 -0,383 1,18E-01
Rbfox3 12917 -5917 -0,488 -0,378 3,91E-02
Nrg3 13374 -6380 -0,462 -0,332 1,42E-02
Sst 13359 -6365 -0,490 -0,300 1,46E-02
Gphn 13444 -6451 -0,453 -0,299 1,18E-02
Gnai1 13362 -6368 -0,476 -0,286 1,46E-02
Gng3 12488 -5487 -0,510 -0,268 7,26E-02
Cck 12504 -5503 -0,486 -0,265 7,08E-02
Npy1r 12174 -5170 -0,525 -0,260 1,06E-01
Gabra4 14170 -7182 -0,376 -0,254 1,01E-01
Htt 12494 -5493 -0,498 -0,242 7,22E-02
Gabbr1 12621 -5620 -0,482 -0,237 6,18E-02
Gabrb2 14227 -7239 -0,332 -0,235 1,17E-01
Gnb1 14194 -7206 -0,362 -0,229 5,67E-04
Slc6a1 13903 -6914 -0,422 -0,227 2,54E-03
Gabra3 12238 -5234 -0,518 -0,084 7,22E-01
The enriched genes on parvalbumin-related gene list based on GSEA of RNA-sequencing data (Lipponen et al., 2016) are shown in the left. The genes which contributed to the leading-edge subset within the gene set are presented in the table, meaning that only the genes that significantly contributed to the enrichment result were included. From the GSEA results, rank in gene list, rank in metric score, running ES, LOG2 fold change and nominal p-value are shown for each gene.
Abbreviations: ES, Enrichment score; GSEA, Gene set enrichment analysis.