Differential expression of cytoskeletal genes in the cochlear nucleus



The relationship between structure and function is clearly illustrated by emerging evidence demonstrating the role of the neuronal cytoskeleton in physiological processes. For example, alterations in axonal caliber, a feature of the cytoskeleton, have been shown to affect reflex arc latencies and are prominent features of several neuropathological disorders. Even in the nonpathologic situation, however, axonal diameter may be a crucial element for the normal function of specialized auditory neurons. To investigate this relationship, we used serial analysis of gene expression and microarray analyses to characterize the expression of cytoskeletal genes in the central auditory system. These data, confirmed by real-time RT-PCR, identified differential expression of intermediate neurofilament transcripts (i.e., Nefh, Nef3, and Nfl) among the subdivisions of the cochlear nucleus. In situ hybridization was used to identify specific classes of neurons within the cochlear nucleus expressing these neurofilament genes. Robust neurofilament expression was seen in bushy cells and cochlear nerve root neurons, suggesting an association between cytoskeletal structure and rapid conduction velocities. Gene expression data were also identified for other classes of cytoskeletal and structural genes important in neuronal function. These results may help to explain some causes of hearing loss in hereditary neuropathies and provide an anatomic basis for understanding normal neuronal function in the central auditory system. Anat Rec Part A, 2006. © 2006 Wiley-Liss, Inc.

The cochlear nucleus receives all primary afferent input from the auditory nerve and processes such information for delivery to higher brain stem regions. The nucleus is divided into subdivisions, each with dominant classes of neurons and with particular roles in signal processing (for review, see Cant and Benson,2003). The anterior ventral cochlear nucleus (AVCN) has large populations of spherical bushy cells that send timing information to the olivary complex for sound localization. The posterior ventral cochlear nucleus (PVCN) contains similar classes of neurons, the globular bushy cells, but also has many octopus and multipolar cells with unique electrophysiological properties (Oertel et al.,1990). The dorsal cochlear nucleus (DCN) shows very little similarity to the ventral divisions and has morphologically and physiologically distinct cell types (Ostapoff et al.,1994). This parsing of auditory processing roles and general segregation of cell types suggests that each subdivision can be distinguished on a structural and functional level. Furthermore, there may be genetic determinants of structure and function that can be identified by molecular biological methods and explain the various responses of cochlear nucleus neurons to auditory nerve stimuli.

Signal processing is most conspicuously a function of neurotransmitter and ion channel kinetics. However, a stronger role for the cytoskeleton in regulation of neuronal function has emerged (Fitzpatrick et al.,1998; Fuchs and Cleveland,1998; Al-Chalabi and Miller,2003; Lalonde and Strazielle,2003). The morphological features that the cytoskeleton imparts on neurons, such as axon caliber and dendritic spine shape, affect the physiological properties of electrical conduction and interactions with surrounding neurons (Hoffman et al.,1987; Kriz et al.,2000). Recent studies have demonstrated cytoskeletal gene differences in the hippocampus of behaviorally disparate animals and there are implications of cytoskeletal abnormalities in some psychiatric disorders (Feldker et al.,2003; Benitez-King et al.,2004). Additional evidence as to the importance of the cytoskeleton in neuronal function has been the identification of cytoskeletal gene mutations underlying some hereditary neuropathies and neurodegenerative disorders (Lupski,2000; De Jonghe et al.,2001; Lalonde and Strazielle,2003; Bruijn et al.,2004; Skvortsova et al.,2004; Mendonca et al.,2005).

This investigation hypothesizes that the cytoskeleton of cochlear nucleus neurons plays a role in determining morphological and physiological properties important for auditory processing. As such, this study uses high-throughput genetic methods to characterize normal cytoskeletal gene expression patterns in the three major subdivisions of the cochlear nucleus. These data were correlated with morphologically and physiologically distinct cell classes to help understand some of the molecular mechanisms that make auditory neurons unique. These data also provide a template for understanding changes in the central auditory system that occur with sensorineural hearing loss and in association with many neurological disorders.

Cytoskeletal Components

The cellular cytoskeleton is composed of microtubules, intermediate filaments, and microfilaments. Microtubules, the thickest of the cytoskeletal core structures, are principally involved in maintenance of cell body shape and the movement of organelles. Microtubules are composed of α- and β-tubulin molecules and interact with an assortment of microtubule-associated proteins (MAPs). The smallest cytoskeletal structures are the microfilaments, which primarily consist of actin fibers and lie just under the cell membrane anchoring and organizing membrane proteins. Intermediate in size between the microtubules and microfilaments are the aptly named intermediate filaments. The neuron-associated intermediate filaments include internexin, peripherin, and three forms of neurofilaments. The cytoskeletal classification can be organized using the gene ontology consortium hierarchy for cellular components (Fig. 1).

Figure 1.

Treemap representation of the cytoskeleton node of the gene ontology. This nested diagram demonstrates the components of the cytoskeleton and their subcategories. The intermediate filaments, which are most critical in neurons, represent only a small fraction of the overall cellular scaffold (yellow). Within this category are the neuron-specific neurofilaments.

Light-, Medium-, and Heavy-Chain Neurofilaments

Neurofilaments are members of the neuron-specific type IV intermediate filament group and named for the molecular masses of their proteins as determined by gel electrophoresis. The light chain is about 68 kDa and is encoded by the human NEFL and rat Nfl genes. The medium chain is 150 kDa and a product of the NEF3 gene in humans and Nef3 gene in rats. The heavy chain ranges between 190 and 210 kDa and is a product of the human NEFH and rat Nefh genes. Each protein consists of a head, rod, and tail. The rod regions are highly conserved and consist of an α-helical region, which intertwine to form coiled-coil dimers. In vivo, these dimers contain at least one Nfl coiled with either an Nef3 or an Nefh. These obligate heterodimers then align in antiparallel fashion (head-to-tail and tail-to-head) to form tetramers, which link to form protofilaments and then organize into the 10 nm neurofilaments.

The neurofilaments are the main components of the axonal cytoskeleton and are the principal determinants of axonal caliber. Axonal caliber is thought to be regulated by the phosphorylation of the neurofilaments, principally the heavy chain, which repels neighboring filaments and increases spacing between the fibers (Cleveland et al.,1991; Garcia et al.,2003). In addition, levels of medium- and heavy-chain neurofilaments and their specific interaction with the light chain appear to regulate axon thickness (Xu et al.,1996). Additional studies have shown that the stoichiometric relationship between the neurofilament subtypes is also an important determinant of structure with effects on dendritic arborization also noted (Kong et al.,1998). Axonal caliber is important for neuronal function as there is a positive correlation with conduction velocity (Hoffman et al.,1987; Kriz et al.,2000). In fact, mice with altered neurofilament levels had significant axonal atrophy and an almost 50% reduction in conduction velocity (Kriz et al.,2000).

Such findings demonstrate that the neurofilament cytoskeleton has a significant effect on neuronal transmission and may play an important role in establishing normal electrophysiological properties. This may be even more critical in sensory systems, such as the auditory system, which is dependent on rapid and high-fidelity responses to stimuli. Cytoskeletal protein abnormalities also underlie many neuropathological disorders, which have associated signs and symptoms of hearing loss and central auditory neuropathy.


This study employed two different methods of high-throughput transcriptome analysis to identify cytoskeletal genes and their differential expression among the divisions of the cochlear nucleus. We utilized serial analysis of gene expression (SAGE) due to its potential to detect any expressed transcript and for the strength of statistical comparisons among libraries of similar size. We also used microarrays (i.e., genechips) for their ease of use, consistency, and ability to provide quantitative measures of gene expression. The specific protocols for each of these methodologies have been well reviewed in the literature.


Six-week-old female Brown Norway rats (n = 40) were used for isolation of RNA from the cochlear nucleus. The rats were anesthetized with Nembutal (50 mg/kg i.p.), tested for loss of paw-pinch reflex, and decapitated. Brains were removed and the cochlear nuclei dissected off the surface of the brain stems on RNAse-free Petri dishes. The nuclei were grossly divided into the three dominant subdivisions: AVCN, PVCN, and DCN. Tissue from 40 rats (i.e., 80 cochlear nuclei) was pooled to minimize bias in RNA sampling due to slight variations in dissection of the cochlear nucleus subdivisions. RNA was extracted from the dissected tissue using the TRIzol protocol (Invitrogen, Carlsbad, CA). The RNA was analyzed by gel electrophoresis for integrity and spectrophotometry for purity and concentration.

SAGE Library Generation

A commercial kit is available for SAGE library generation (Invitrogen) and the general protocol and our methods are available in the existing literature (Velculescu et al.,1995; Halum et al.,2004). The foundation of SAGE is that a 10 base pair portion of an mRNA transcript located near the 3′ terminus, referred to as a SAGE tag, is sufficiently unique to be able to identify that gene exclusively. In practice, a SAGE tag can occasionally be found in more than one transcript and thus represent more than one gene. Also, a single gene may be large enough to be represented by more than one tag. Overall, however, the general principles have been confirmed and SAGE can provide a consistent gene expression profile of the tissue of interest (Velculescu et al.,1995; Lee et al.,2002). The final data consist of a list of SAGE tags and the frequency with which they have been found in the library. As tags are pulled at random from the library, the frequency is a relative measure of expression level. Tags are subsequently matched to known genes or transcribed sequences using a regularly updated public database (SAGEMap: http://www.ncbi.nlm.nih.gov/SAGE/). Theoretically, continuing to sequence a SAGE library will identify all the tags in that tissue and thus potentially every gene that is transcribed. We generated SAGE libraries from AVCN, PVCN, and DCN that contain about 34,000 total tags representing approximately 3,000 known genes in each region. Statistical analyses of differential expression between libraries were performed using eSAGE, which uses the statistic of Audic and Claverie (1997) and has been shown to be equivalent to the Fisher's exact test and z-test (Audic and Claverie,1997; Margulies and Innis,2000).

Microarray Experiments

Genechips differ from SAGE in basic principles but represent an economical and well-established means of probing the transcriptome. Microarray studies start with a finite set of genes of interest. These genes are represented by oligonucleotide probes, which are aligned and fixed to a glass slide. Microarray experiments detect the hybridization of tissue mRNA (specifically, cDNA made from the mRNA) to the probes on the slide by fluorescence to determine the presence and level of expression of a given transcript. For this study, microarray experiments were performed commercially (GenUs Biosystems, Chicago, IL). The CodeLink Rat Whole Genome Bioarray (GE Healthcare) with over 35,000 probes was used to analyze the expression in each of the cochlear nucleus subdivisions. Fluorescence levels were compared to local background to determine whether each hybridization represented a “good call” and thus a valid representation of expression. Hybridization intensity was normalized to the mean of the chip with GeneSpring software (Silicon Genetics, Redwood, CA). Each subdivision was run on two chips and the average normalized intensities utilized to determine expression levels for each probe.

Real-Time RT-PCR

Real-time reverse transcriptase-PCR (real-time RT-PCR) was used to determine the relative expression levels of the three neurofilament genes. The housekeeping gene hypoxanthine guanine phosphoribosyl transferase (HPRT) was used as a reference because of its consistent expression levels throughout the cochlear nucleus in both the SAGE and microarray experiments. RNA was treated with DNase I, Amp Grade (Invitrogen), and then reverse-transcribed using oligo-dT primers (Invitrogen) and the Omniscript Reverse Transcription kit (Qiagen, Valencia, CA).

Real-time PCR was performed using Taqman Gene Expression Assays (Applied Biosystems, Foster City, CA) for Nef3 (ABI: Rn00566763_m1), Nfl (ABI: Rn00582365_m1), Nefh (ABI: Rn00709325_m1), Peripherin (ABI: Rn00561807_m1), Vimentin (ABI: Rn00579738_m1), and HPRT (ABI: Rn01527838_g1). The reactions were run on the iCycler iQ Multicolor Real-Time Detection System (Bio Rad Laboratories, Hercules, CA). The thermal cycling conditions were as follows: 50°C hold for 2 min, 95°C hold for 10 min, followed by two-step PCR for 40 cycles of 95°C for 15 sec and 60°C for 1 min. Negative controls, in which reverse transcriptase was omitted from the reaction, were run for each sample and all samples were performed in triplicate. Threshold cycle (Ct) values of HPRT and the gene of interest were then used to calculate the relative mean normalized expression in each subdivision (Muller et al.,2002). Each experiment was run in triplicate and the relative fold increase was averaged for the three runs. The data were normalized to the subdivision with the lowest level to determine a relative fold increase in the other regions.

In Situ Hybridization

Brown Norway rats (n = 3 per probe) were perfused transcardially with ice-cold 4% paraformaldehyde in 0.1 M PBS, pH 7.2. Brain stems were dissected out, postfixed for 1 hr in the same fixative, and cryoprotected in 20% sucrose in RNase-free 0.1 M PBS until sectioning. Cryostat sections were taken transversely through the brainstem at 20 μm and collected on SuperfrostPlus slides (Fisher Scientific, Tustin, CA). The in situ hybridization protocol is well-documented and a brief description with probe-specific information is provided below.

RT-PCR was used to generate amplicons to be used as probes for in situ hybridization. Authenticity of the probes was confirmed by DNA sequencing and the probes were subcloned into pCR II-TOPO (Invitrogen). The following PCR primer sets were used: Nef3: forward 5′-AGAGCGCAAAGACTACCTGAAGA, reverse 5′-GGATATTGTGACTGAGGGCTGTC; Nfl: forward 5′-TCTGAAGGAGAAGCAGAAGAGGA, reverse 5′-ACA- TTGCCGTAGATCCTGAACTC; and NefH: forward 5′-CTCTGCCCAAGAGGAGATAACTG, reverse 5′-CCCTCTTCTGCCTCTTCTTCTTC.

Digoxigenin-labeled single-stranded cRNA probes were transcribed from linearized templates in 20 μl reaction mixtures: 1 μg linearized plasmid, 2 μl DIG RNA labeling mix (10 × concentration; Roche Applied Science, Penzberg, Germany), 1 μl RNasin (40 U/μl), 10 mM DTT, 4 μl of 5 × transcription buffer, and 20 units of Sp6 or T7 RNA polymerase (all from Promega, Madison, WI). Slide-mounted sections were sequentially treated and predigested with proteinase K, washed in DEPC ddH2O, rinsed in 0.1 triethanolamine (pH 8.0), rinsed in 2 × SSC (1 × SSC = 0.15 M NaCl/0.04 M NaCitrate, pH 7.2), dehydrated in graded ethanols, and air-dried. Slides were covered with 200 μl/slide of hybridization mixture and incubated at 55°C for 16 hr. Sections were exposed to RNase (100 μg/ml in 0.5 M NaCl, 10 mM Tris, 1 mM EDTA, pH 8.0) for 30 min at 37°C, washed in decreasing concentrations of SSC, with the final rinse done in 0.1 × SSC at 55°C for 30 min. Following a final wash in Tris-buffered saline, the slides were incubated with alkaline phosphatase-conjugated sheep antidigoxigenin antibody (Roche; 1:500 in TBS containing 0.3% Triton X 100 and 1% blocking reagent) for 16–24 hr at 4°C. Sections were washed and subsequently incubated in NBT/BCIP (Roche; NBT: nitroblue tetrazodium; BCIP: 5-bromo-4-chloro-3-indoyl-phosphate, 4 toluidine salt; buffer: 0.1 M Tris, 0.1 M NaCl, 0.05 M MgCl2, pH 9.5) for 16–24 hr. The color reaction was frequently checked and, upon reaching the desired intensity, slides were washed in distilled H2O, air-dried, and mounted with Gel/Mount (Biomeda, Foster City, CA). Negative control sections were treated with RNase (100 μg/ml) for 30 min at 37°C prior to the proteinase K treatment.


Data analysis focused specifically on those genes mapping to the cytoskeleton node of the cellular component ontology from the gene ontology consortium (www.geneontology.org). SAGE data were mapped onto the cytoskeleton gene ontology and visualized in a treemap to identify highly expressed cytoskeletal classes (for example, see Fig. 2). As expected in neuronal populations, the microtubule component was very highly represented and had many genes mapping to the kinesin complex. Among the microfilament cytoskeletal components, the dynein complex, myosins, and actin filaments had good representation. The intermediate filament class had genes found in all its subcategories, including the neurofilaments, keratins, and lamins. The microarray data were similarly analyzed and expression levels for the individual genes were determined (for treemap example, see Fig. 3). These analyses identified high levels of tubulin in the cochlear nuclei as well as elevated levels of Nef3 and Nfl.

Figure 2.

Treemap representation of SAGE data from the cochlear nucleus incorporated into the cytoskeleton node of the Gene Ontology. The treemap has been filtered to show the major categories of cytoskeletal protein and a list of all genes found by SAGE that fit this category (for details of gene names, see Table 1). The colors reflect the relative frequencies of the SAGE tags, with brighter green representing the most abundant. In the broad categories, it is apparent that although the intermediate filaments are a small component of the ontology, they are the most highly expressed class in these tissues (i.e., brighter green).

Figure 3.

Treemap representation of microarray data from the cochlear nucleus incorporated into the cytoskeleton node of the Gene Ontology. The treemap has been filtered to show the major categories of cytoskeletal protein and a list of all genes found by microarray that fit this category (for details on gene names, see Table 2). Similar to Figure 2, the brighter color represents higher hybridization intensity and the intermediate filament category is most highly represented. This treemap has been secondarily filtered to represent hybridization intensity as box size for the individual genes so that those below the median expression level do not interfere with the analysis. Probes for uncharacterized genes have been left in this view, in contrast to the gene list in Table 2, and these may present future targets for genetic analyses in the cochlear nucleus.

Table 1. Frequency of SAGE tags for cytoskeletal genes in the cochlear nucleus subdivisions
Cfl1cofilin 1619188
Snap25synaptosomal-associated protein784541
Csrp1cysteine and glycine-rich protein 1405651
Nefhneurofilament, heavy polypeptide546626
Myo5bmyosin 5B425440
Tmsb4xthymosin beta-4504326
Nef3neurofilament 3, medium573410
Gfapglial fibrillary acidic protein492118
Stmn3stathmin-like 340169
Actbactin, beta162021
Stmn4stathmin-like 4141724
Olfm1olfactomedin related ER localized protein81621
Nflneurofilament, light polypeptide22144
Maptmicrotubule-associated protein tau131410
Vamp2vesicle-associated membrane protein 2131113
Spna2alpha-spectrin 218134
Map1bmicrotubule-associated protein 1b1488
Mtap1amicrotubule-associated protein 1A12117
Fez1protein kinase C-binding protein Zeta11694
Inexainternexin, alpha1963
Tmsb10thymosin, beta 108514
Dncic2dynein, cytoplasmic, intermediate polypeptide 26710
Ptk2protein tyrosine kinase 23126
Dctn4dynactin 4676
Dncli2LIC-2 dynein light intermediate chain 53/55694
Vapavesicle-associated membrane protein, associated protein a5410
Cyln2cytoplasmic linker 24105
Klc1kinesin light chain 11133
Stmn1stathmin 1647
Bin1myc box dependent interacting protein 1636
Stx7syntaxin 71104
Gabarapgamma-aminobutyric acid receptor associated protein752
Stx5asyntaxin 5a435
Stmn2stathmin-like 2254
Cabp1calcium binding protein 1361
Kif1bkinesin family member 1B262
Tpm1tropomyosin 1, alpha522
Tubb5tubulin, beta 5333
Pindynein, cytoplasmic, light chain 1512
Cnn3calponin 3, acidic332
Dctn1dynactin 1143
Lmnalamin A142
Egfrepidermal growth factor receptor501
Myh6myosin heavy chain, polypeptide 6312
Nme1expressed in non-metastatic cells 1231
Stk39serine threonine kinase 39 (STE20/SPS1 homolog, yeast)240
Krt2-8keratin complex 2, basic, gene 8141
Olfm3olfactomedin 3114
Usp2ubiquitin specific protease 2132
Vapbvesicle-associated membrane protein, associated protein B and C132
Aurkbaurora kinase B033
Add1adducin 1, alpha320
Csnk1dcasein kinase 1, delta212
WaspipWiskott-Aldrich syndrome protein interacting protein212
Ap1g1adaptor protein complex AP-1, gamma 1 subunit122
Myh9myosin, heavy polypeptide 9104
Vdpvesicle docking protein, 115 kDa122
Stx4asyntaxin 4032
Gabarapl2GABA(A) receptor-associated protein like 2112
Bet1blocked early in transport 1 homolog (S.cerevisiae)013
Sh3kbp1SH3-domain kinase binding protein 1022
Pfn2profilin II300
Prph1peripherin 1300
Acta1actin alpha 1210
Als2cr3amyotrophic lateral sclerosis 2 chromosome region, candidate-3210
Coro1bcoronin, actin-binding protein, 1B111
Gosr2golgi SNAP receptor complex member 2120
Mtap6microtubule-associated protein 6120
Ptk2bprotein tyrosine kinase 2 beta111
Stx6syntaxin 6102
Gpr51G protein-coupled receptor 51012
Myolemyosin IE021
Sycp3synaptonemal complex protein 3030
Cspg6chondroitin sulfate proteoglycan 6101
Jak1Janus kinase 1110
Lamb2laminin, beta 2101
Tekt1tektin 1101
Tmod2tropomodulin 2110
Fez2fasciculation and elongation protein zeta 2 (zygin II)011
Gas7growth arrest specific 7002
Jak3Janus kinase 3002
Rhoip3Rho interacting protein 3011
Vamp3vesicle-associated membrane protein 3020
Hspb7heat shock 27kD protein family, member 7 (cardiovascular)100
Lmnb1lamin B1100
Tctex1t-complex testis expressed 1100
Actg2actin, gamma 2010
Add3adducin 3, gamma001
Arpc1bactin related protein 2/3 complex, subunit 1B010
Arhgap4Rho GTPase activating protein 4001
Cetn2centrin 2010
Clasp2CLIP-associating protein 2010
Cutl1cut (Drosophila)-like 1010
Kif3ckinesin family member 3C001
Mlc3fast myosin alkali light chain010
Myo1cmyosin IC010
Nup62nuclear pore glycoprotein 62010
Pacsin2protein kinase C and casein kinase substrate in neurons 2010
Pscd3pleckstrin homology, Sec7 and coiled/coil domains 3001
Stx3asyntaxin 3010
Sycp2synaptonemal complex protein 2001
Tmod1tropomodulin 1001
Vil2villin 2 (ezrin p81)001
Table 2. Normalized microarray hybridization intensities of cytoskeletal genes in the cochlear nucleus
Klc1kinesin light chain C mRNA281.75309.69204.81
Nef3neurofilament 3, medium152.95169.9347.30
Pindynein, cytoplasmic, light chain 1123.30128.84108.44
Nf1neurofilament, light polypeptide140.10151.0354.57
Fez1protein kinase C-binding protein Zeta1122.65121.6695.58
Olfm1olfactomedin related ER localized protein97.7499.5689.46
Fez1protein kinase C-binding protein Zeta1102.5097.4880.31
Mtap1amicrotubule-associated protein 1 A100.64105.3464.35
Pfn2profilin II104.9488.2073.50
Gfapglial fibrillary acidic protein84.82100.5062.74
Dncli2LIC-2 dynein light intermediate chain 53/5579.9875.7463.84
Cfl1cofilin 171.3574.6371.56
Spna2alpha-spectrin 269.0871.5753.54
Dctn1dynactin 167.9465.7057.93
Cnn3calponin 3, acidic62.1959.8164.98
Prph1peripherin 161.1562.7144.69
Vamp3vesicle-associated membrane protein 355.6659.3553.44
Dncic2dynein, cytoplasmic, intermediate polypeptide 257.1256.1749.42
Tubb5tubulin, beta 554.9055.6644.93
Stmn2stathmin-like 253.4150.7045.40
Mtap6microtubule-associated protein 652.6652.5040.65
Maptmicrotubule-associated protein tau48.7343.5639.02
Gpr51G protein-coupled receptor 5139.9136.1748.26
Clasp2CLIP-associating protein 241.7346.2235.70
Map1bmicrotubule associated protein IB, mRNA44.6546.8831.51
Stk39serine threonine kinase 3943.6144.2934.50
Bin1myc box dependent interacting protein 139.2846.0936.34
Stmn3stathmin-like 340.3031.6727.19
Trim3tripartite motif protein 334.3934.2627.17
Gabarapgamma-aminobutyric acid receptor associated protein32.9632.0029.42
Nme1expressed in non-metastatic cells 128.2428.4827.94
Olfm3olfactomedin 324.6524.2235.02
Errpepidermal growth factor receptor related protein, mRNA27.0728.1920.85
Stmn2stathmin-like 228.5420.9924.93
TMBr-3brain alpha-tropomyosin, mRNA, 3′ end25.1723.5723.57
Bet1blocked early in transport 1 homolog, S. cerevisiae21.5219.7819.89
Zrp3zygin-related protein type III, mRNA22.0319.0116.81
Vdpvesicle docking protein, 115 kDa19.4316.2318.03
Zrp3zygin-related protein type III, mRNA19.9916.5614.53
Actg2actin, gamma 215.6516.5716.31
Stx5asyntaxin 5a15.7816.4416.25
Pacsin2protein kinase C/casein kinase substrate in neurons 216.9815.4215.22
Stx4asyntaxin 415.7915.3513.58
Cnn3calponin 3, acidic14.3712.5215.30
Ptk2protein tyrosine kinase 211.6912.4412.33
Tctex1t-complex testis expressed 113.0111.3711.50
InexaInternexin, alpha13.589.757.02
Sh3kbp1SH3-domain kinase binding protein 110.5812.8010.01
Snx16sorting nexin 1612.129.7111.26
Vil2Rattus norvegicus mRNA for ezrin p818.008.8413.54
Vamp2vesicle-associated membrane protein 27.5910.1612.26
Coro1bcoronin, actin-binding protein, 1B10.889.449.18
Trim3tripartite motif protein 310.179.189.27
Ubp45deubiquitinating enzyme Ubp45, mRNA9.588.708.43
Tubb5tubulin, beta 58.969.208.53
Stk12serine/threonine kinase 123.933.1019.61
Usp2ubiquitin specific protease 29.267.948.20
Tekt1tektin 15.605.3814.01
Cspg6chondroitin sulfate proteoglycan 66.736.337.24
Olfm1olfactomedin related ER localized protein6.566.397.19
Cabp1calcium binding protein 17.636.035.58
Cyln2cytoplasmic linker
Lmnalamin A5.816.995.50
Apcadenomatosis polyposis coli5.845.506.15
Bin1myc box dependent interacting protein 15.545.685.28
Myl3myosin, light polypeptide 35.525.984.89
Lamb2laminin, beta 24.555.875.27
Stk39serine threonine kinase 395.025.115.01
Acta1actin alpha 14.655.365.08
Arcactivity regulated cytoskeletal-associated protein6.804.362.79
Myr8myosin heavy chain5.524.133.98
Jak3Janus kinase 33.715.553.81
Myr5Unconventional myosin from rat
Tmod1tropomodulin 14.823.283.77
Jak3Janus kinase 32.874.152.94
Mtap6microtubule-associated protein 63.442.632.61
Stx1asyntaxin 1a2.172.262.99
Myo5bmyosin 5B4.152.060.64
Vamp2bvesicle associated membrane protein 2B2.202.291.80
Dctn4dynactin 42.141.932.17
Tnni3troponin 1, type 30.700.664.57
Add3adducin 3, gamma1.591.391.64
Myo7amyosin VIIA1.321.211.89
Csnk1dcasein kinase 1, delta1.011.181.42
Tnnt2troponin T21.061.570.91
Sycp2synaptonemal complex protein 20.971.031.39
Kif3ckinesin family member 3C1.141.001.10
Pscd2pleckstrin homology, Sec7 and coiled/coil domains 21.160.841.05
Vapbvesicle-associated membrane protein, associated protein B1.051.060.88
Dlgap2discs, large, Drosophila, homolog-associated protein 21.120.950.89
Arpc1bactin related protein 2/3 complex, subunit 1B0.981.130.84
Spag4sperm antigen 40.640.641.49
Myl2myosin, light polypeptide 21.100.880.77
MefvMediterranean fever0.980.940.83
Krt8keratin 80.510.401.61

The transcripts of known cytoskeletal genes and their SAGE tag frequencies are listed in Table 1. There were 111 genes found by SAGE and the majority of these were identified in each cochlear nucleus subdivision. Only 29 of the 111 genes (26%) were expressed uniquely in a single subdivision; of these, 22 were represented by a single tag. Of the remaining seven transcripts found in a lone subdivision, none were represented by more than three tags. Three tags were found for peripherin 1 (Prph1) and profillin 2 (Pfn2), which were only found in the AVCN, and another three tags for synaptonemal complex protein 3 (Sycp3), which was found only in the PVCN. For those represented by two tags, the majority were in the DCN (i.e., growth arrest-specific 7: Gas7, Janus kinase 3: Jak3, and striatin: Strn), with the other being found only in the PVCN (vesicle-associated membrane protein 3: Vamp3). None of the genes found in only one subdivision were expressed in sufficient tag numbers to have statistically significant differences in expression.

The genes mapping to the cytoskeleton ontology in the microarray experiments were identified with their respective hybridization intensities (Table 2). Only those genes with average intensities greater than the normalized median in at least one sample were included, as probes with overall lower expression levels did not demonstrate reliable reproducibility. There were 105 microarray probes with cytoskeletal roles among the three cochlear nucleus subdivisions. A liberal comparison based on a twofold change in intensity with at least one transcript expressed higher than the median showed that only myosin 5b was expressed in the AVCN higher than the PVCN. No genes meeting these criteria were found with higher expression in the PVCN than the AVCN. There were five genes with higher expression in the AVCN than DCN and these included the light- and medium-chain neurofilaments, Nef3 and Nfl. The other three genes were myosin 5b, activity-regulated cytoskeletal-associated protein (Arc), and an uncharacterized clone. Nef3, Nfl, and myosin 5b were also expressed at similarly elevated levels in the PVCN as compared to the DCN. Several genes were found to be higher in the DCN than the AVCN, including serine/threonine kinase 12, tektin1, troponin 1 type 3, sperm antigen 4, and keratin 8. No additional genes meeting the above criteria for differential expression were found between the DCN and PVCN.

The intermediate filament class was specifically analyzed due to the functional implications of these structural proteins. There were 10 genes found by SAGE and 13 genes present on the microarray that fell into this cytoskeletal subcategory (Tables 3 and 4). The results of SAGE showed that only two, peripherin1 and lamin B1, were expressed in a single subdivision. Both of these were found exclusively in the AVCN and peripherin1 expression was above baseline levels and represented by three tags. A single tag for the intermediate filament gene vimentin was found in PVCN and DCN and not in AVCN. The remaining intermediate filament genes were represented by at least one tag in all subdivisions of the cochlear nucleus. In comparing subdivisions, the neurofilament genes Nef3, Nefh, and Nfl were all highly expressed in the AVCN and PVCN as compared to the DCN. The differences in frequency between each ventral division and the DCN for these tags were statistically significant (P < 0.05). Statistically significant differential expression was also seen for the abundance of internexin in the AVCN as compared to the two other subdivisions.

Table 3. Frequency of SAGE tags for intermediate filament genes
Nef3neurofilament 3, medium573410
Nefhneurofilament, heavy polypeptide546613
Gfapglial fibrillary acidic protein492118
Nflneurofilament, light polypeptide22144
Inexainternexin, alpha1963
Prph 1peripherin 1300
Krt2-8keratin complex 2, basic, gene 8141
Lmnalamin A142
Lmnb 1lamin B1100
Table 4. Microarray hybridization intensities for intermediate filament genes
Nef3neurofilament 3, medium152.95169.6347.30
Nflneurofilament, light polypeptide140.10151.0354.57
Gfapglial fibrillary acidic protein84.82100.5062.74
Prph1peripherin 161.1562.7144.69
InexaInternexin, alpha13.589.757.02
Lmnalamin A5.816.995.50
Dlgap2discs, large (Drosophila) homolog-associated protein 21.120.950.89
Lmnb1lamin B10.580.590.71
Krt8keratin 80.510.401.61
Bfsp1lens fiber cell beaded-filament structure protein0.490.640.43

Analysis of microarray data similarly noted an approximately threefold greater expression level of Nef3 and Nfl in the AVCN and PVCN as compared to the DCN. Nefh was not represented on the microarray and no expression levels were obtained. The other primary intermediate filament, internexin, was also not on the microarray, although the SAGE experiment showed differential expression of this gene as well. The microarray had representations for intermediate filament genes that were not found by SAGE. These included Bfsb1, Krt21, Krt9, Dlgap2, and Rsn. None of these had expression levels significantly greater than the median and were therefore not considered in further analyses.

The notable difference between the SAGE and microarray data was in their representative levels of peripherin1 and vimentin. SAGE identified relatively few tags for these transcripts but they both had robust hybridization levels in all three subdivisions in the microarray experiments. To investigate this further, real-time RT-PCR was performed using probes specific for vimentin and peripherin1. We inferred relative abundance by comparing the real-time PCR cycle number at the point the threshold crossed background (CT) to our housekeeping gene, HPRT, and to other genes we have studied. Both these transcripts appeared at cycle numbers relative to our housekeeping gene that were consistent with having moderate expression levels in all three subdivisions of the cochlear nucleus. Their actual levels of expression thus appear to be more accurately represented by the microarray hybridization intensities. Real-time RT-PCR showed that vimentin expression in the PVCN and DCN were between 1.5- and 1.8-fold higher than the AVCN. This result is consistent with SAGE, as tags for this gene were found only in the PVCN and DCN. Results consistent with SAGE were also found for peripherin1, which showed a 10.9-fold higher expression level in AVCN than in DCN and 5-fold higher expression compared to PVCN by real-time RT-PCR. Recall this transcript was represented by three tags in AVCN and no tags were found in the other regions.

Further characterization of the differential expression of the intermediate filaments focused on the three neurofilament genes (i.e., Nef3, Nefh, and Nfl) as these have been most frequently implicated in neuropathological processes and in normal neuronal function. In order to determine accurately their relative expression levels, real-time RT-PCR was performed on RNA extracted from each of the cochlear nucleus subdivisions. The data were normalized to the subdivision with the lowest level to determine a relative fold increase in the other divisions (Fig. 4). Nef3 was found to be least expressed in the DCN with the AVCN and PVCN showing 6.4- and 4.0-fold higher expression, respectively. Similarly, Nfl was between 4.4- and 4.7-fold more highly expressed in AVCN and PVCN than in DCN. Nefh was also found to have higher levels of transcript in the VCN than the DCN: 4.4-fold higher in the AVCN and 2.2-fold higher in the PVCN. Thus, all the neurofilament cytoskeletal genes were more highly expressed in the ventral subdivisions of the rat cochlear nucleus than in the DCN. There was no significant differential expression between the AVCN and PVCN.

Figure 4.

Graphic representation of the relative expression levels of the neurofilament genes in each of the cochlear nucleus subdivisions as determined by real-time RT-PCR. All the neurofilaments were more highly expressed in AVCN and PVCN than in DCN by at least twofold. As all expression levels are normalized to a standard, these data reflect differential gene expression by cochlear nucleus neurons rather than methodological issues of tissue sampling, cell number, or cell density. These data confirm the findings by SAGE and microarray and were further investigated using in situ hybridization (Figs. 5 and 6).

Figure 5.

Photomicrographs of in situ hybridization for Nef3 in the cochlear nucleus. An overview of the cochlear nucleus (A) shows darkly labeled cells in the PVCN and AVCN with little staining in the DCN. Higher-power views provide better localization. B shows the PVCN, in which Nef3 expressing cells are seen at the nerve root and dorsally near octopus cell domains. A more rostral cut through the PVCN (C) show positive cells in globular bushy cell regions. Further sections through the nucleus (D) show staining in the ventral regions of the AVCN in spherical bushy cell areas. This panel also demonstrates the relative paucity of staining in the DCN. E and F depict Nef3 expressing neurons in the AVCN and PVCN, respectively. These cells are fairly homogeneous, round to ovoid, and consistent with bushy cell appearances in traditional light microscopy. Some of the cells in the PVCN appear to have prominent axon hillocks, which would indicate large-caliber axons important for rapid conduction. Examination of the DCN (G) shows few strong staining cells and those that are present are in the middle and deep layers of the nucleus. These likely represent fusiform and giant cells. Scale bars = 200 μm (B–D); 50 μm (E–G).

Figure 6.

Photomicrographs of in situ hybridization for Nfl in the cochlear nucleus. A shows the PVCN in which Nfl is heavily expressed in neurons around the nerve root, ventral portion of the nucleus, and dorsomedially. These cell populations may represent nerve root neurons, globular bushy cells, and octopus cells, respectively. A section through the midportion of the ventral cochlear nucleus (B) shows positive cells diffusely distributed throughout the subdivision. These cells are polymorphic and may include several types of multipolar cells. C shows a higher-magnification view of some of the Nfl expressing neurons within the cochlear nerve root. These cells are thought to be important in startle reflexes and thus would have large axons allowing for rapid signal conduction. Cells in the ventral regions of the PVCN (D) and AVCN (E) that were strongly positive for Nfl had generally ovoid appearances with occasional prominent axon hillocks. These cells are likely globular bushy and spherical bushy cells, respectively. Other cells in the central core of the VCN had irregular shapes consistent with multipolar cells (F). These cell populations included both large and small neurons, suggesting more than one class of multipolar cell is expressing this neurofilament. Scale bars = 200 μm (A); 100 μm (B); 50 μm (C); 20 μm (D–F).

Localization of the transcripts for the neurofilament genes within each subdivision of the cochlear nucleus was investigated using in situ hybridization. The neurons expressing each of the neurofilament subtypes were analyzed with regards to location within each subdivision and the relative size and morphology of the cells. In general, Nef3, Nefh, and Nfl appeared to localize to similar, but restricted classes of cells and had notable expression in the ventral divisions. These findings are shown in Figures 5 to 7 and specifically addressed below.

Figure 7.

Photomicrographs of in situ hybridization for Nefh in the cochlear nucleus. An overview of the VCN shows Nefh expressing cells throughout the nucleus with a prominent heavily staining cluster of cells near the ventral-rostral pole (A). These latter cells may be spherical bushy cells while the remainder may include classes of multipolar cell. B and C show Nefh expressing cells in the DCN. Only a few cells demonstrated such gene expression and were predominantly localized to the deeper layers of the nucleus. These may thus represent one or more class of giant cell. D and E show the polymorphic cells expressing Nefh found throughout the ventral cochlear nucleus. This was in contrast to the more discrete expression of Nef3. The various sizes and shapes of these cells suggest they belong to the multipolar class. We postulate that there is a different stoichiometric ratio of the neurofilament subunits in these cells that distinguishes them morphologically and physiologically from bushy cells. Scale bars = 100 μm (A); 200 μm (B); 50 μm (C); 50 μm (D); 20 μm (E).

Strong hybridization with the Nef3 probe was observed in neurons of the AVCN and PVCN (Fig. 5). This staining was most pronounced in the rostral AVCN, the ventral PVCN, the dorsomedial PVCN, and around the nerve entry zone. These locations corresponded to regions known to have large populations of spherical bushy cell, globular bushy cell, octopus cell, and cochlear nerve root neurons, respectively. There were some populations of small cells in these ventral nuclei that also seemed to be expressing the Nef3 transcript and these were in the marginal cell regions. In contrast, the DCN showed sparse staining. The few positively staining cells were large with round to elongated shapes, located in the middle and deep layers of the nucleus. These neurons are discussed further below as similar staining patterns in the DCN were seen with Nefh.

The pattern of Nfl expression was similar to that observed with Nef3 (Fig. 6). Staining was seen in predominantly large cells of the AVCN and PVCN. The cells around the cochlear nerve root were most obvious due to their intensity of staining in all specimens. Large Nfl expressing neurons were also regularly seen in the PVCN near the nerve root in globular bushy cell regions and extending dorsomedially toward octopus cell domains. In the rostral AVCN, large round cells were readily apparent in the spherical bushy cell regions. Irregular and variably sized neurons within the mid-portions of the VCN were seen in tissues with higher background staining and based on location and morphology most likely represent the various classes of multipolar cells. Hybridization in the DCN was not observed with Nfl except in a few sections with very high background. In these sections, the majority of cells were small and diffusely distributed throughout the nucleus. These cells could not be definitively identified as having hybridized probe.

Nefh expression also showed similar patterns to Nef3 but variably sized and shaped neurons in the AVCN and PVCN were more readily apparent (Fig. 7). Their somatic features and distribution are consistent with the multipolar cell populations. As with the other neurofilament probes, expression was also pronounced in the nerve root, where globular bushy cells and octopus cells reside. In contrast to Nfl, but similar to that seen with Nef3, expression was seen in large cells in the DCN. The few positive cells found with the Nefh probe were in the deep layers of the DCN and did not seem to extend as superficial as some of the Nef3-positive cells.

Further analysis of DCN-positive neurons employed morphological measurements of maximal cell body length and width with calculations of a roundness factor (width:length ratio). This was performed to investigate whether these cells could be classified into categories of DCN neurons, as has been previously described (Kane et al.,1981; Ryugo and Willard,1985). Measurements of Nefh cells (n = 7) showed an average maximal diameter of 30.56 ± 4.93 μm (range, 26.41–40.46 μm) and an average roundness value of 0.65 (range, 0.44–0.84). The Nef3 neurons (n = 9) had a similar average maximal diameter of 27.42 ± 4.06 μm (range, 23.31–36.44 μm) with an average r-value of 0.71 (range, 0.56–0.83). The use of in situ hybridization stained tissue to perform such morphological measures has not been compared to the traditional histological preparations previously used in characterizing these cell types and these results should be interpreted with caution.


Cytoskeletal proteins provide the structural framework within neurons but also serve to define some of the electrophysiological properties of these cells. This study sought to provide an initial characterization of the expression patterns of specific cytoskeletal genes among the subdivisions of the cochlear nucleus. The patterns of expression were analyzed with respect to known morphological and physiological properties of the cochlear nucleus subdivisions and their neurons. These results may provide a molecular basis for further understanding normal and pathological auditory processing.

The principal finding in this investigation is that the neurofilament cytoskeletal genes appear to be highly expressed in the ventral subdivisions of the cochlear nucleus as compared to the DCN. These differential expression patterns were confirmed using real-time RT-PCR. The basis for this differential expression was further explored using in situ hybridization, which identified high levels of transcript expression in several neuronal populations of the AVCN and PVCN. This observation contrasted with the expression patterns in the DCN, where only a few large cells in the deeper layers demonstrated appreciable levels of staining. Further analysis of these data examined specific ventral cochlear nucleus neurons and their morphological and physiological properties that may correlate with high levels of neurofilament expression.

Nef3 and Nfl transcript appeared to colocalize throughout the ventral cochlear nucleus by in situ hybridization and showed similar patterns of differential expression in the SAGE, microarray, and real-time RT-PCR experiments. This colocalization is not unexpected as neurofilaments form as obligatory heteropolymers with Nfl promoting polymerization (Petzold,2005). The regions of colocalization for the Nef3 and Nfl neurofilament transcripts in the VCN included spherical and globular bushy cell domains, nerve root neurons, and potentially octopus cell regions.

The spherical and globular bushy cells of the AVCN and PVCN are considered important in functions of sound localization (Manis and Marx,1991; Wang et al.,1998; Cant and Benson,2003). These cells have been well studied as to the morphological and electrophysiological properties that may subserve this role. The most conspicuous specialized feature associated with the spherical bushy cell is the large auditory nerve synapse engulfing each neuron: the endbulb of Held (Ryugo and Fekete,1982; Wang et al.,1998). This large synapse ensures faithful transmission of peripheral auditory stimuli to the central nervous system. Globular bushy cells lack the endbulb-type contacts but receive a large number of auditory nerve terminals. These neurons project to the superior olivary complex (SOC), where interaural differences are further processed. Bushy cell axonal morphology has been studied as potentially contributing to the timing differences reaching the SOC (Smith et al.,1993; Beckius et al.,1999). In a study by Beckius et al. (1999), bushy cell axonal caliber was noted to be quite “substantial” near the main trunk and consistently about 2 μm in diameter along the midportion. Based on their electrophysiological model, this thickness would provide very high conduction rates, as is seen experimentally. They found little variability in axon caliber, suggesting that regulation of axon thickness to control transmission velocity was not likely and that a major factor in establishing interaural timing cues was fiber length (Beckius et al.,1999). Thus, holding conduction velocity at a high and consistent rate among all the bushy neurons projecting to the SOC would “standardize” the initial stimulation allowing for the anatomy of the projection, rather than individual cellular electrophysiology, to influence interaural timing cues. As axon caliber is directly related to conduction velocity, similarly sized axons would control for this variable. In addition, a large-caliber axon would maximize the conduction velocity, thus facilitating the rapid localization of the sound stimulus (Yagi et al.,1977; Kriz et al.,2000; Higashimori et al.,2005).

Morphologically, the spherical and globular bushy cells are characterized by having large axons that do not branch within the cochlear nucleus prior to leaving via the trapezoid body (Smith and Rhode,1987; Cant and Benson,2003). While both classes have prominent axons, globular bushy cell projections appear to have the largest axon calibers within the cochlear nucleus (Tolbert and Morest,1982; Tolbert et al.,1982; Spirou et al.,1990). A primary determinant of axonal caliber is the neurofilament cytoskeleton and thus, on anatomical terms, high levels of neurofilament transcript would be expected to support this scaffold. Further, the endings of the globular bushy cells in the medial superior olive (MSO) (i.e., the calyces of Held) are among the largest in the mammalian brain (Guinan and Li,1990; Smith et al.,1991). While transcript level cannot always be equated to levels of protein, simple maintenance of this extensive cytoskeleton may account for the high levels of neurofilament mRNA identified in these cells. The intermediate filament network, however, has been shown to be in a constant dynamic state and ongoing gene transcription would be expected to maintain the robust cytoskeleton in these large neurons (Vikstrom et al.,1992).

Another cochlear nucleus region found to express high levels of neurofilament transcript was in the mid to posterior PVCN, where octopus cells have been reported (Osen,1969; Kane,1973). This region may also contain globular bushy cells and definitive distinction between these cell types could not be obtained with this histochemical method. The octopus cells, named because of their prominent unidirectional dendritic projections, fire at the onset of an acoustic stimulus with great precision (Oertel et al.,1990,2000). They project to the ventral nucleus of the lateral lemniscus (VNLL), where they end in moderately sized endbulb-like synapses and to the superior paraolivary nucleus, where they terminate in large boutons (Schofield,1995). Large terminals arising from these axons may be indicative of a robust cytoskeleton and account for increased levels of neurofilament expression in the region of the PVCN, where these cell bodies reside.

Consistent neurofilament expression was also observed in the region of the cochlear root neurons. These cells have very short latencies and project to the reticular formation and areas important in startle and reflex control (Lee et al.,1996; Lopez et al.,1999; Scott et al.,1999). This class of neuron does not appear to process auditory information but rather functions as a warning system. The root neurons have remarkably large axons that, like the previously discussed bushy cells, do not branch until leaving the cochlear nucleus (Lopez et al.,1999). The lack of early branching and large-caliber axons may require specialized or augmented structural support by the neurofilaments and help account for the elevated expression levels identified in this study.

The cochlear root neurons and globular bushy cells colocalize in the ventral VCN, which represented the most heavily and consistently staining region for the neurofilament genes. These cells have similar light microscopic appearances consisting of an ovoid shape with eccentric nucleus but have been distinguished by morphometric measures (Merchan et al.,1988). Given the relative abundance of bushy cells, we considered that some of our more ventral populations may not represent such nerve root neurons. However, we consistently observed these cells in the body of the auditory nerve and oriented, as described by Merchan et al. (1988), with their long axes parallel to the nerve fibers (Figs. 5C and 6A and C). Measurements of these cells, in these preparations, showed those cells within the nerve fibers had maximum diameters of > 30 μm, while the ovoid cells at the entry zone had diameters between 25 and 30 μm. This dimorphism is consistent with previously reported measures of these two cell populations under light microscopy (Merchan et al.,1988). The highest overall neurofilament expression, by in situ hybridizaton, thus appears to localize to the two VCN cell populations with the largest reported axon calibers and functional dependence on rapid conduction velocity. As these cells are unique to the VCN, this would also account for the high levels of neurofilament gene expression found in this region by SAGE, microarray, and real-time RT-PCR.

Nefh expression patterns differed from that seen with Nef3 with the observation of staining in neurons of various morphologies distributed throughout the ventral cochlear nucleus. Given their location and appearance, these likely represent multipolar cells of the VCN. Multipolar cells are described in detail in this issue and likely consist of many subtypes based on anatomical and physiological criteria (Doucet and Ryugo,2006). They project within and between the cochlear nuclei and to other auditory brainstem centers without demonstrating the large, morphologically specialized endings seen with bushy cells. Multipolar axons are thin and can branch widely within the cochlear nucleus itself (Oertel et al.,1990; Doucet and Ryugo,1997; Palmer et al.,2003). The multipolar cells exhibit response properties that consist of very regular firing intervals in relation to the stimulation and are quite distinct from the patterns of response in bushy or octopus cells (for review, see Doucet and Ryugo,2006). It may be these anatomical and physiological differences between multipolar cells and those neurons previously discussed account for the apparent abundance of Nefh in the former as compared to Nef3.

These observed differences in neurofilament gene expression may represent a stoichiometric shift among the cell populations in the relative combinations of Nef3, Nfl, and Nefh forming the intermediate filament skeleton. This in turn may affect the structural anatomy of the axon and ultimately physiological response properties. Experimental overincorporation of Nefh chain into existing neurofilament networks results in a reduction of axonal caliber, which is seen pathologically and may also be critical for normal physiological mechanisms (Straube-West et al.,1996; Gotow,2000). Analysis of rat brain shows dynamic shifts in the relative ratios of Nefh to Nef3 and Nfl during development. Nefh appears later and in a lower relative level than Nef3 or Nfl (Scott et al.,1985). This observation is consistent with recent knockout genetic experiments for the individual neurofilaments, which have shown that Nef3 is more important than Nefh in determining axonal growth (Lariviere and Julien,2004). Thus, shifts in neurofilament ratios have morphological implications with an association between Nef3 expression and large axonal calibers. The contrasting correlate is that relatively high Nefh expression may be associated with thinner axons under normal physiological conditions. Thus, multipolar cells may have a unique proportion of Nefh, which accounts for the differential gene expression seen in these experiments. Such a stoichiometric shift in Nefh abundance may then underlie the morphological differences between multipolar and bushy cells and some of their physiologically distinct properties.

Expression of all the neurofilament genes was lower in the DCN and this finding was mirrored by the paucity of staining in this region by in situ hybridization. Nef3 and Nefh were seen in large cells of the middle and deep layers and morphometric analysis suggests they may represent a combination of fusiform and giant cells. Two cells had diameters greater than 35 μm, which places them in previously reported ranges for giant cells (Osen,1969; Brawer et al.,1974), but the majority were between 25 and 35 μm and consistent with fusiform cell sizes (Brawer et al.,1974). Width-to-length ratios were examined in an attempt to narrow down the cell identification but the range of ratios indicated that such cells may represent elongate, ovoid, and spherical classes of giant cell (Kane et al.,1981). The few cells with width:length ratios less than 0.5 may represent fusiform cells but their location in the deep DCN is also consistent with elongate giant cells (Kane et al.,1981). Measurements of greater numbers of cells and ultrastructural analyses would be necessary to better distinguish these cell forms.

Comparison of Neurofilament Expression to Other Brain Regions

The central auditory nuclei are part of a highly specialized system and may be expected to have unique gene expression patterns as compared to nonauditory centers. In addition, there may be differences between different brain stem auditory nuclei, as has been shown among the subdivisions of the cochlear nucleus. The use of SAGE for the investigation of gene expression allows comparison of transcript frequency to existing libraries without requiring simultaneous performance of experiments. We compared the frequencies for the principal SAGE tags used to assess levels of Nef3, Nefh, and Nfl in the cochlear nucleus to existing libraries for the normal rat SOC and hippocampus (Datson et al.,2001; Koehl et al.,2004). The tag frequencies have been normalized to counts per million to account for differences in library size (Table 5). Nfl is the basic neurofilament polymer and is well represented in all regions by two different SAGE tags without obvious differential expression to SOC or hippocampus. The majority of the differential expression of Nfl between DCN and the rest of the nucleus in this study appears to arise from one specific tag (i.e., GAAAAATAGT). Such a finding may indicate the presence of a splice variant of Nfl that is differentially expressed and may impart functionally distinct properties on these cells.

Table 5. Relative SAGE tag frequency for neurofilament genes in the rat brain
NeurofilamentSAGE TagAVCNPVCNDCNSOCHippocampus
  1. Tag counts are normalized to tags per million.


There was only a single principal tag for Nef3, which was found in high numbers in the AVCN and SOC in contrast to low levels in the DCN and hippocampus. Similarities between the SOC and AVCN would include the presence of large Endbulb synapses and there may be unique features of the neuronal targets of these endings that Nef3 imparts: potentially high-fidelity response and rapid transmission rates. The principal Nefh tag was high in AVCN and PVCN but relatively low in other regions. This, as discussed above, may reflect the contribution of multipolar cells to the pool of RNA transcript in the ventral nucleus and again suggests a unique role for Nefh in these cell types. Further investigation of such differential transcript expression throughout the auditory brainstem and on a cellular level may help characterize these genetic differences.

Intermediate Filament Mutations and Neuropathological Disease

The interdependence between structure and function is clearly illustrated by the number of neuropathological conditions associated with mutations and abnormalities of the intermediate filament cytoskeletal genes. These associations may be manifest as pathological accumulations of neurofilament proteins as markers of disease or by direct mutations of the cytoskeletal gene, which become the causative agent of disease. Pathological accumulations of neurofilaments are seen in many neurodegenerative disorders, including amyotrophic lateral sclerosis, Parkinson's disease, Alzheimer's disease, and various other neuropathies. These accumulations have traditionally been thought to be a by-product of neuronal dysfunction, but it appears that such cytoskeletal disorganization can have its own functional sequelae.

Amyotrophic lateral sclerosis (ALS) is an adult-onset neurodegenerative disorder leading to the loss of motor neurons in the spinal cord and brain. About 10% of these cases are associated with a congenital mutation in the SOD1 gene but the majority are sporadic without identifiable cause. A common feature in this disorder is an accumulation of neurofilaments in the soma and proximal axons. Even though there is neurofilament accumulation, studies have demonstrated up to a 60% reduction in Nfl expression in ALS motor neurons (Bergeron et al.,1994). In contrast, transgenic mice with overproduction of Nefh develop a progressive ALS-like neuropathy (Cote et al.,1993; Julien et al.,1995). In humans, there does indeed appear to be increased immunoreactivity for phosphorylated Nefh in ALS (Petzold,2005). These seemingly contradictory findings may reflect a stoichiometric imbalance in the normal neurofilament ratios, which may disrupt axonal transport and lead to heavy-chain accumulations near the site of translation. In addition to accumulations of Nefh, multiple reports have found deletions and insertions in the tail region in a small subset of sporadic ALS cases (Lariviere and Julien,2004). This region contains the specific amino acid repeats important for phosphorylation.

Increased staining for phosphorylated Nefh is also seen in multiple sclerosis and Alzheimer's disease (Petzold,2005). Such staining may reflect a response to injury by the primary underlying pathology, but there is some evidence that the axons with high Nefh staining are the ones that degenerate (Morrison et al.,1987). Similar patterns of staining have also been seen in diabetic neuropathy, glaucoma, and stroke.

Mutations in the light-chain neurofilament gene have also been associated with neuropathological diseases. Charcot-Marie-Tooth (CMT) is a progressive neurological disorder with two general phenotypes: a primary demyelinating form that results in conduction delays (CMT1) and a neuropathy with primary axonal loss that has normal conduction velocity for remaining axons but reduced spike amplitudes (CMT2). The latter is further divided into several forms and CMT2E has been linked to mutations in the Nfl gene (Mersiyanova et al.,2000; Zuchner et al.,2004). A CMT pedigree with Nfl mutation included a member with deafness but the relationship to the discovered mutation is not clear (Zuchner et al.,2004). Other severe forms of CMT, typically associated with PMP22 (peripheral myelin protein) mutations, have also been associated with sensorineural hearing loss (Sambuughin et al.,2003; Joo et al.,2004).

Cytoskeleton and Auditory Function

Many of the neurodegenerative disorders discussed have associations with auditory disturbance. For example, degeneration in the auditory system and/or auditory processing abnormalities have been described in Alzheimer's disease and ALS (Ohm and Braak,1989; Gil et al.,1995; Iliadou and Kaprinis,2003). Also, a subset of phenotypes of CMT is clearly associated with sensorineural hearing loss. Further, audiometric studies of patients with hereditary motor and sensory neuropathy (HMSN), which CMT is a form of, demonstrated abnormal central auditory function (Musiek et al.,1982; Alcin et al.,2000; Papadakis et al.,2003). A notable feature of some of these patients are discrimination deficits out of proportion to their pure tone averages, which may suggest a central processing disorder. Indeed, auditory brainstem response (ABR) testing showed an absence of the central wave responses suggesting that the pathology began proximal to the cochlear nerve and not in the peripheral cochlea. Overall, the association between auditory dysfunction and global neuropathies is not unexpected as normal auditory processing is highly dependent on timing and high-fidelity transmission of the initial stimulus. Given the association of many of the neuropathies with neurofilament abnormalities, the deficits in auditory function thus appear to arise from alterations in the cytoskeleton.

A neurofilament deficient quail (Quv) has provided a model for studying auditory function in cytoskeletal abnormalities. Auditory evoked potential testing was performed on Quv quails and, although electrocochleography remained normal, there was abnormal latency for the brain stem responses (Sheykholeslami et al.,2001b). Further electrophysiological studies showed that higher brainstem auditory pathways also had significant conduction delays. This study demonstrated normal peripheral function but central auditory pathology. Electron microscopic evaluation of the cochlear nerve showed normal myelin but reduced axonal diameters due to an absence of neurofilaments (Sheykholeslami et al.,2001a). The reduced fiber diameter correlated well with current understanding of the relationship between axon caliber and conduction.

These clinical and experimental data suggest that cytoskeletal genes function as more than just a scaffold and instead play significant roles in cellular electrophysiology. The current investigation into the cochlear nucleus demonstrates differential expression of cytoskeletal genes among the subdivisions of the cochlear nucleus, and between brain regions, which reflect the underlying neuroanatomical and electrophysiological properties. These findings help to establish a profile of normal cytoskeletal gene expression in the first central auditory center, which in turn can be used as a template for assessing changes in hearing that may occur with neuropathologic disease and other central processing disorders. Further investigations, such as the development of specific neurofilament knockout animals, can better characterize the functional deficits that may arise from abnormalities of the auditory neuron cytoskeleton.