Auditory Brain Atlas


Next Generation Sequencing
(Total RNA: RNAseq)



Sequencing-driven transcriptome profiling of the central auditory pathways
The goal of these studies is to build gene expression libraries of individual central auditory structures that can be used to guide targeted neuroanatomical and neurophysiological studies (e.g., in situ hybridization, immunohistochemistry, optogenetics, pharmacology).

Two types of resources are being developed:

(1) Normative baseline libraries
  • Baseline libraries of coding and noncoding mRNA expression in structures of the central auditory pathway of normal hearing adult and developing animals.
  • Targeted analyses of functional groups of genes (e.g., receptors, ion channels) to create normal profiles by brain area / cell type

(2) Experimental animal libraries
  • Sequencing and tissue assay libraries for experimental subjects (e.g., acquired hearing loss; disease models)


Sequencing and Tissue Library Workflow


Stacks Image 1458


Outline of Workflow

Samples collected from target areas in the central auditory pathway
(auditory cortex, medial geniculate, inferior colliculus, cochlear nucleus)

Total RNAseq (Illumina HiSeq 2500 platform)
(paired-end, 30M reads)(FASTQ datafile format)

Analysis
(targeted functional analyses)(selection of genes for tissue assays)

Tissue Assays
(in situ hybridization: single colorimetric, multiplexed fluorescence)
(immunohistochemistry: single and multi fluorescence)

Imaging
(image acquisition: high resolution scanning of tissue sections)
(brightfield, fluorescence, confocal)

Auditory Brain Atlas
(incorporation of image sets into atlas)




COMPLETED STUDIES
1) Auditory forebrain maturation during postnatal development (normal hearing mice)
Hackett TA, Guo Y, Clause AR, Hackett NJ, Garbett K, Zhang P, Polley DB (2015) Transcriptional maturation of the mouse auditory forebrain.
BMC Genomics, Aug 14, 16(1):606. doi: 10.1186/s12864-015-1709-8. PMID: 26271746.

Mouse: C57BL/6J
Postnatal ages: P7, P14, P21, adult (before and after onset of hearing)
Sex: 3 male + 3 female per age group
Brain areas sampled: primary auditory cortex (A1), medial geniculate body (MGB), inferior colliculus (IC), cochlear nucleus (CN)
[ Note: current sequencing limited to A1 and MGB ]
Method: Next generation sequencing of total RNA (Illumina platform, paired-end 50, minimum 30 million reads).
Analyses: hierarchical clustering and differential expression by age, brain region and sex; gene set enrichment (GSEA) and pathways analyses.

Resources available from this study (scroll down for figures and other data):
  • All raw data files, differential expression by single genes for all comparisons, GSEA of 51 gene ontology categories and 111 gene families selected for relevance to brain development, neurotransmission and plasticity
  • Profiles of ~4700 genes in 237 gene families were plotted for on-demand inspection of the developmental trajectories for each gene.
  • A simple Look-Up tool (Excel format) was also developed that generate profile plots form manual entry of up to 25 genes at a time.
Publication info:
  • Follow this link to download the published paper and supplementary materials.

STUDIES IN PROGRESS
1) Macaque monkey (normal hearing and noise-induced hearing loss)
Species: Adult rhesus (M. mulatta)
Sex: 4 male + 4 female
Brain areas sampled: primary and secondary areas of auditory cortex, dorsal and ventral divisions of medial geniculate body (MGB), central nucleus of the inferior colliculus (IC), medial superior olive (MSO), ventral cochlear nucleus (VCN), dorsal cochlear nucleus (DCN)



Stacks Image 2239
Hierarchical Clustering Analysis (all genes)
Hierarchical clustering analyses of RNAseq data from the mouse auditory forebrain (A1, MG) are summarized in the figure above for all genes (heatmap truncated to save space here). Gene expression changed significantly with age in both regions. P7 samples clustered together first, then by brain region. From P14 to adult, samples were separated into two large clusters by regions, and then by age within each regional cluster. Thus, gene expression in A1 and MG was more similar prior to P14, but regional differences dominated clustering thereafter. This implies that expression patterns became more regionally distinct with maturation. Gender played no significant role in gene expression in A1 or MG regions. (Note: heat map truncated).

Stacks Image 2473
Differential expression: MG vs A1
(A) The total numbers of differentially expressed genes between A1 and MG are plotted for each postnatal age. (B) Overlapping differential expression in A1 and MG. The Venn diagram depicts the total numbers of genes that were differentially expressed (MG vs A1) at only one postnatal age, and the numbers that were commonly expressed in all age group combinations.

Stacks Image 2481
Differential expression: between age groups
The total numbers of differentially-expressed genes are plotted for each of the six possible comparisons. Comparisons with P7 yielded the largest numbers of differentially expressed genes, and totals declined with increasing age.

Stacks Image 2303
Expression trend analysis
The number of genes with increasing, decreasing, static or other maturational trajectories is plotted for A1 and MG. Also plotted are the numbers of genes with these profiles in both A1 and MG.

Gene Set Enrichment Analyses (GSEA)

Description: GSEA enables identification of genes and genesets that are highly enriched in a sample. We applied GSEA to 111 gene families and 51 gene ontology (GO) categories. The results are too extensive to present here, but the top genesets are listed in the Tables below, using a false discovery rate (FDR) of q<0.25 as a cut off. The overall expression trajectory (trend) for each gene-set (up- or down-regulation) is also indicated. These analyses enabled efficient screening of the entire transcriptome for categorical trends of interest (e.g., plasticity) Within each category, note that expression trajectory and level of some genes may or may not match the overall GSEA categorization. This possibility was offset by constructing gene-wise plots for each family or GO category (see below).
Stacks Image 2364
GSEA of selected gene ontology categories
Geneset enrichment analysis (GSEA) of a subset of gene ontology (GO) categories in the Mouse Genome Informatics (MGI) database. From the 51 categories listed in Table S24, 20 reached the FDR q-value cutoff of 0.25. Ten are listed below. For each geneset, the number of genes in the group (size), FDR q-value, and direction is given. Categories with upward (UP) and downward (DOWN) maturational trajectories were grouped separately in the Table.

Stacks Image 2372
GSEA of selected gene familes
Geneset enrichment analysis (GSEA) of a subset of gene families in the HUGO database. From 111 categories listed in Table S22, 27 families reached the FDR q-value cutoff of 0.25. Ten are listed below. For each gene family, the number of genes in the group (size), FDR q-value, and direction of expression (up, down) are listed. Categories with upward (UP) and downward (DOWN) maturational trajectories were grouped separately in the Table.


Gene families database of the developing auditory forebrain.
Perhaps the most powerful resource derived from this study was construction of a gene families database. Expression levels of individual genes were plotted by gene family as a function of brain region (A1, MG) and postnatal age (P7, P14, P21, adult). The analysis included ~4,700 genes in 237 gene families in 20 major functional categories. from the Gene Families database maintained by the HUGO Gene Nomenclature Committee at the European Bioinformatics Institute (http://www.genenames.org) (HUGO).

The most recent version of the entire
Gene Families Database for the mouse auditory forebrain is contained in a single Excel file that can be downloaded by following this link. It is highly recommended to view the READ ME tab before beginning to explore the database.

Note: permission to use these data in other publications may be requested by contacting the corresponding author: troy.a.hackett@vanderbilt.edu

Examples from the gene families database that have been profiled are illustrated in the figures below:
(1) Glutamate and GABA receptors
(2) Other neurotransmitter/neuromodulator receptors
(3) Potassium channels family
(4) Critical periods (opening and closing)
(5) Extracellular matrix (formation and maintenance)


Look-up tool
To facilitate screening and extraction of profiles from the database, a Look-Up tool was developed (link below). The tool automatically plots the maturational profiles and correlation matrices for any single gene or list of genes (up to 25 at a time). It also generates a listing of the normalized counts for all samples for extraction for other purposes. Excel format.

Link to
Look-up tool

Instructions are given in the Excel file.

Note:
permission to use these data in other publications may be requested by contacting the corresponding author: troy.a.hackett@vanderbilt.edu



Glutamate and GABA Receptors

Stacks Image 1434
Gene expression profiles of glutamate and GABA receptor families. For each gene, mean normalized read counts and % of maximum read counts are plotted by postnatal age. The overall age-related trajectory in expression is indicated by arrows (up, down, none). Significance was based on the log2 fold-change when observed by both DEseq2 and EdgeR (*p<0.05; **p<0.01), or by just one of those methods (~p<0.05).

This example illustrates how plotting the RNAseq data by gene families can be used to screen for genes of interest in a given brain region -- in this case, A1 and MG as a function of postnatal age.


Other Receptors Involved in Neurotransmission and Neuromodulation

Stacks Image 1681
Gene expression profiles of neurotransmitter receptor families. Six receptor families are profiles (adenosine, noradrenaline, dopamine, acetylcholine, serotonin, glycine). For each gene, mean normalized read counts and % of maximum read counts are plotted by postnatal age. The overall age-related trajectory in expression is indicated by arrows (up, down, none). Significance was based on the log2 fold-change when observed by both DEseq2 and EdgeR (*p<0.05; **p<0.01), or by just one of those methods (~p<0.05).


Potassium Channels

Stacks Image 1708
Gene expression profiles of the potassium channel family. For each gene, mean normalized read counts and % of maximum read counts are plotted by postnatal age. The overall age-related trajectory in expression is indicated by arrows (up, down, none). Significance was based on the log2 fold-change when observed by both DEseq2 and EdgeR (*p<0.05; **p<0.01), or by just one of those methods (~p<0.05).


Genes related to opening and closing of critical periods

Stacks Image 1558
Gene expression profiles of a custom gene ontology category involved in critical period and developmental plasticity. 91 genes spanning multiple families are profiled. Genes for neurotransmitter and neuromodulator receptors were omitted (refer to Figs. 3 - 4). For each gene, mean normalized read counts and % of maximum read counts are plotted by postnatal age. The overall age-related trajectory in expression is indicated by arrows (up, down, none). Significance was based on the log2 fold-change when observed by both DEseq2 and EdgeR (*p<0.05; **p<0.01), or by just one of those methods (~p<0.05).


Genes involved in construction and composition of the extracellular matrix

Stacks Image 1590
Gene expression profiles of gene families involved in extracellular matrix composition. Structural genesets include selected chondroitin sulfate (CSPG) and heparin sulfate (HSPG) proteoglycans, collagens, cadherins, and contactins. Ligands and receptors include selected neural cell adhesion molecules (NCAM), ephrins and their receptors, fibroblast growth factors (FGFs), laminins, and several mixed families. For each gene, mean normalized read counts and % of maximum read counts are plotted by postnatal age. The overall age-related trajectory in expression is indicated by arrows (up, down, none). Significance was based on the log2 fold-change when observed by both DEseq2 and EdgeR (*p<0.05; **p<0.01), or by just one of those methods (~p<0.05).

Stacks Image 1505
Tissue-Level Assays (FISH, FIHC) Example:

Rationale:
RNAseq provides a reference for conducting tissue level assays of gene of interest. Tissue-level assays are essential for relating the sequencing data to the underlying architecture. Two examples illustrate the importance of these assays for interpreting RNAseq data:
(1)
High read count: if the read-count for a given gene is high in A1, it could be that many neurons express this gene at moderate to high levels, or that particular subpopulations expresses the gene at very high levels.
(2)
Low read count: if the read-count is low, it may be that many neurons express the gene at very low levels, or that a small subpopulation expresses the gene are moderate to high levels.

Application:
For systems-level questions about the structure of brain circuits, tissue-level assays are the most efficient and informative means of bringing the RNAseq data to life (e.g., in neurons, glia, local and long-range circuits). In addition to assays of single genes or proteins, multi fluorescent techniques enable patterns of co-expression and co-localization among multiple genes to be revealed.

Example (refer to figure above):
Using RNAseq as a guide, we are currently conducting tissue-level assays on selected groups of genes and proteins related to neurotransmission. In addition to standard histological stains (e.g., Nissl, myelin), we use multifluorescence in situ hybridization (FISH) and immunohistochemistry (FIHC) assays to characterize expression in all brain regions at all ages. In the example figure above (panels A & B), FISH was used to profile neurons that were enriched for genes in 4 neurotransmitter receptor families in A1 and MG of mice (vesicular glutamate transporters: VGluT1, VGluT2; vesicular GABA transporter, VGAT; and neuronal nuclear protein: NeuN).

(Panel C illustrates the method used to harvest samples from frozen tissue during sectioning).