Proteomic profiling of neuromas reveals alterations in protein composition and local protein synthesis in hyper-excitable nerves

Neuroma formation

The method used for neuroma generation was adapted from that described in rats by Rivera et al. [8]. Ten C57BL/6 mice were used to generate neuromas; all mice were matched by age (8 weeks) and weight (30–35 g). Under deep anaesthesia with halothane (2–4% in pure O₂) and with sterile precautions, the saphenous nerve was exposed at the level of the mid-thigh, dissected free and tightly ligated with 8-0 silk. The nerve was cut distal to the ligature and the cut end inserted into a 5 mm long tube (0.45 mm internal diameter) to prevent lateral innervation of surrounding tissue. The tube was tied in place with the same piece of silk so that the cut end of the nerve was ~2 mm from the distal end of the tube. The distal end of the tube was left open and ~5 mm of the distal nerve stump was excised to prevent innervation. The incision was closed in layers. Neuromas were made in both saphenous nerves. The animals were housed in groups of two to four and inspected daily for infections or abnormal behaviour. The mice were given ad libitum access to water and food. After 3 days, 1 week, 2 weeks and 3 weeks, the silicone tube was removed, neuroma were cleared of connective and fatty tissues and dissected for 2D-DIGE experiments under a dissecting microscope. Four neuroma samples were generated from 2 mice in each time point after transection. At the same time, 4 control samples were generated from two control C57BL/6 mice without operation. Similar experiments were carried out on Sprague-Dawley rats. 32 rats were equally distributed in 4 groups (3 day, 1 week, 2 week and 3 week), one neuroma and one sham-operation control were generated from each rat. All rats were matched by age (3–4 months old) and weight (200–250 g). All animal care and experimental processes were performed in accordance with the United Kingdom Home Office Act 1986.

Chemicals

Protease-inhibitor tablets, Tris-base, urea, thiourea, ammonium persulfate, ammonium bicarbonate, iodoacetamide (IAA) and 3- [(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate) (CHAPS) were from Sigma (Steinheim, Germany). Protein assay kit was from Pierce (Rockford, USA). Immobilized pH gradient IPG strips and Pharmalyte were from GE Healthcare (Chalfont St. Giles, UK). Dithiothreitol (DTT) was from Melford Laboratories (Ipswich, UK). Trypsin (modified, sequence grade, porcine) was from Promega (Madison, WI, USA). 2,5-dihydroxybenzoic acid was from Bruker Daltonics (Leipzig, Germany). Ammonium sulphate and ortho-phosphoric acid (85%) were from Fluka (Steinheim, Germany). Methanol, ethanol and Coomassie brilliant blue G250 were from BDH (Lutterworth, UK). TFA was from Fisher (Loughborough, UK). NHS-Cy2 was purchased from GE Healthcare. NHS-Cy3/Cy5 were synthesized 'in-house' and stored as described [9].

Sample preparation

Pooled neuroma or control nerve samples in 1.5 ml centrifuge tubes were frozen in liquid nitrogen and crushed into powder with a plastic homogenization pestle. Samples were lysed in 2D-DIGE lysis buffer (2 M thiourea, 8 M urea, 4% (w/v) CHAPS, 10 mM Tris-HCl, pH 8.3) and quantified with a Bradford assay using BSA as a standard.

Fluorescence labelling with Cy-Dyes

Neuroma and control samples were quantified and labelled with NHS-Cy2, -Cy3 and -Cy5 as previously described [10]. Briefly, 50 μg of protein taken from pooled neuroma or control samples was minimally labelled with 400 pmol of Cy3 or Cy5 in triplicate. An equal pool of all samples was also prepared and labelled with Cy2 (also at 8 pmol/μg protein) as a standard run on all gels to aid in spot matching and cross-gel quantitative analysis. Labelling was performed on ice in the dark for 1 hour and reactions were quenched with a 20-fold molar excess of free lysine to dye for 10 minute, and then DTT added to 65 mM. In addition, two preparative samples were made using 300 μg of protein from control saphenous nerve labelled with Cy3 or Cy5. Labelled samples were mixed appropriately and carrier Ampholines/Pharmalyte (pH 3–10; 50:50(v/v)) added to a final concentration of 2%.

Two-dimensional gel electrophoresis (2-DE)

2-DE was performed as described [11–13] using 24 cm pH 3–10 NL IPG strips, and 12% SDS-PAGE gels bonded to low-fluorescence glass plates. Briefly, Cy3 and Cy5 labelled control, 3 day, 1 week, 2 week or 3 week neuroma samples and one Cy2 labelled internal standard sample were mixed appropriately and separated by first dimension isoelectric focusing and second dimension SDS-PAGE. Samples were run as technical triplicates with two preparative gels loaded with 300 μg of protein. Gels were run in Ettan 12 gel tanks at 2.2 mA per gel at 16°C until the dye front had run off the bottom. All steps were carried out in a dedicated clean room.

Image acquisition and biological variance analysis

Gels were scanned between glass plates using a Typhoon™ 9400 variable mode imager and ImageQuant software. The photomultiplier tube voltage was adjusted for each dye channel (Cy2, Cy3 and Cy5) for preliminary low-resolution scans to give maximum pixel values within 5–10% for each channel and below saturation, prior to the acquisition of 100 μm high-resolution images. Images were cropped and analysed using DeCyder™ V5.0. Comparisons of test spot volumes with corresponding standard spot volume were made and values were averaged across triplicates for each experimental condition. Statistical analysis was then performed to pick spots matching across all images, displaying a ≥1.75 average-fold increase or decrease in abundance between neuroma and control and with p values < 0.05 (Student's t-test). A 1.75-fold threshold was chosen to limit the substantial number of significant protein changes detected that would require MS-based protein identification, but without missing potentially interesting low-fold changes. It is important to note that only significant changes (p < 0.05) in protein expression were targeted for MS-based identification.

In-gel digestion and protein identification by mass spectrometry (MS)

Post-stained colloidal Coomassie blue (CCB) G-250 images were matched with corresponding fluorescent images using DyCyder software. Pick lists of coordinates were then created for spots of interest relative to a pair of reference markers fixed to the glass plates at casting. Spots were excised using an Ettan spot picker (GE Healthcare). In-gel tryptic digestion was performed as described [9, 10], and extracted peptides were resuspended in 6 μL of water. For peptide mass fingerprinting, 0.5 μL of each tryptic digest was mixed with 1 μL of matrix (saturated aqueous solution of 2,5-dihydroxybenzoic acid) and spotted onto a sample target plate and dried. Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectra were acquired using an externally calibrated Ultraflex TOF/TOF mass spectrometer (Bruker Daltonics) in the reflector mode. After internal calibration using trypsin autolysis peaks, prominent peaks in the mass range m/z 700–4000 were used to generate a peptide mass fingerprint which was searched against the updated mouse IPI database using Mascot version 2.1.3 (Matrix Sciences, London, UK). Identifications were accepted when a minimum of six peptide masses matched to a particular protein (mass error of ± 50 ppm allowing 1 missed cleavage), sequence coverage was >25%, MOWSE scores were higher than the threshold value (p = 0.05) and the predicted protein mass agreed with the gel-based mass. Nano-HPLC electrospray ionization (ESI) collision-induced dissociation (CID) MS/MS was performed on a Q-TOF mass spectrometer (Waters, Manchester, UK), coupled to an Ultimate system LC (Dionex) with a PepMap C18 75 μm inner diameter column at a flow rate of 300 nL/min. Spectra were processed using MassLynx software (Waters) and submitted to Mascot database search routines. Positive identifications were made when at least two peptide sequences matched an entry and MOWSE scores were above the significance threshold value (p = 0.05)

Immunoblotting

Since hydrophobic integral membrane proteins are difficult to resolve on 2D-gels, and previous data suggested that altered ion channel expression was correlated with pain [14–18], the expression of sodium channel Nav1.3, Nav1.8 and calcium channel α2δ-1 subunit were measured by immunoblotting. Briefly, extracts from rat neuromas separated by 1D-SDS-PAGE were electro-blotted onto Hybond-C Extra membrane (GE Healthcare). Membranes were blocked for 1 h with 5% w/v BSA in TBS (50 mM Tris HCl pH 8, 150 mM NaCl) plus 0.1% Tween-20 (TBS-T). Membranes were incubated for at least 2 h in primary antibody in TBS-T, washed in TBS-T (3 × 10 min) and probed with the appropriate horseradish peroxidase-coupled secondary antibody (GE Healthcare) for 1 h. After further washes in TBS-T (3 × 10 min), immunoprobed proteins were visualised using ECL (Perkin-Elmer Life Sciences, Waltham, MA, USA). Films were scanned on a Bio-Rad GS800 Densitometer. Polyclonal antibodies used were anti-Nav1.8 (1:2500 dilution; generated in-house), anti-Nav1.3 (1:2500 dilution; Sigma), anti-voltage calcium channel α2δ-1 subunit (1:500 dilution; a kind gift from Professor A Dolphin, UCL), anti-protein DJ-1 (1:1000 dilution; Santa Cruz Biotechnology, Calne, UK), anti-vimentin (1:1500 dilution; DAKO, Ely, UK) and anti-Na⁺/K⁺ATPase (1:1000 dilution; Santa Cruz Biotechnology). Monoclonal antibodies used were anti-actin beta (1:2000 dilution; Sigma) and anti-peroxiredoxin 2 (1:5000 dilution; Lab Frontier, Seoul, South Korea).

S³⁵methionine in vitro labelling

Mouse neuroma samples were generated as described. One week old neuromas from two mice were dissociated and washed 3 times with cold PBS then transferred to methionine-free DMEM medium for 30 min. S³⁵ methionine (GE Healthcare) was added to a final concentration of 100 mCi/mL and samples were incubated in a 5% CO₂ incubator at 37°C overnight. S³⁵-labelled samples were washed 3 times in cold PBS, homogenised and lysed in 2D lysis buffer. The S³⁵-labelled sample (50 μg total protein) was mixed with 200 μg of intact sciatic nerve and separated by 2-DE as described above. Gels were stained with CCB, dried, scanned by densitometry and then placed in a cassette with a phosphorimager screen (Kodak, Hemel Hempstead, UK) for 5 days. Screens were scanned with a Molecular Imager FX (Bio-Rad Laboratories, Hemel Hempstead, UK) and radio-labelled spots were picked from dried gels and identified by MS.

Protein expression profiling of control and neuroma samples

We used both proteomic and immunoblotting approaches to catalogue altered protein expression in hyper-excitable neuromas derived from the saphenous nerve. 2D gels lose many of the highly insoluble receptors and channels present at low abundance in cell membranes during the isoelectric focussing phase, so we also carried out immunoblotting to examine some proteins previously implicated in neuropathic pain. A 2D-DIGE analysis was first performed comparing 3 day, 1 week, 2 week and 3 week old neuromas with control uninjured saphenous nerve. Four samples from two animals were pooled from each condition, generating 160–210 μg of total protein and samples were analysed as technical triplicates using an internal standard on each gel consisting of an equal pool of all samples to aid in spot matching and quantification (Figure 1).

Approximately 1800 protein features were resolved on the gels and compared using DeCyder image analysis software (Figure 2). A Student's t-test was performed for every matched spot set, comparing the average and standard deviation of protein abundance across each condition. There were around 200 spots that were 1.75-fold differentially expressed (p ≤ 0.05) between the neuroma and control groups (see Table 1). For example, there were 218 spots detected with a ≥1.75-fold change between 2 and 3 week neuromas and control with 138 increased in neuroma and 80 decreased in neuroma versus control.

	3 day neuroma/control	1 week neuroma/control	2 week neuroma/control	3 week neuroma/control	2 and 3 week neuroma/control
Total number of differentially expressed spots (≥1.75-fold)	224	248	254	182	218
Increased	137	141	153	125	138
Decreased	87	109	101	57	80

Spots displaying a ≥1.75-fold increase or decrease in expression with a Student T-test P value of ≤ 0.05 (n = 3/4) and matching across all 8 analytical gels were included.

Identification of differentially expressed proteins

Validation by immunoblotting

Differential expression of several of the identified proteins was validated by immunoblotting, although samples from rats were used, due to insufficient material being obtained from the mice. The expression of protein DJ-1 increased with time of neuroma formation as shown from the 2D-DIGE/MS profiling of mouse samples (Figure 3). Similarly, the increased expression of peroxiredoxin 2 during neuroma formation was also confirmed, as was vimentin expression. Importantly, the immunoblotting results of rat neuromas were consistent with the 2D-DIGE analysis of mouse samples, showing that the observed changes can be translated across species and are likely to be functionally relevant in neuroma formation.

Ion channel protein expression level in rat neuroma

Various ion channels have been implicated in the pain response. However, due to their very hydrophobic nature and large masses, these proteins cannot be readily extracted from samples or resolved by 2-DE. Thus, sodium channels Nav 1.3 and Nav 1.8 and calcium channel α2δ-1 subunit expression were evaluated in individual rat neuromas and control nerves from the same animals by immunoblotting samples which had been lysed directly in boiling SDS-PAGE sample buffer. Loading was then normalised against Na⁺/K⁺-ATPase (Figure 4). The expression of Nav 1.3 appeared to be the same between 3 day neuroma and control, slightly increased in 1 week and 2 week neuromas and slightly decreased in 3 week neuromas. However, these differences were not significant when the loading was normalised to Na⁺/K⁺-ATPase (Figure 4A). Similarly, although neuroma-specific changes in expression were suggested for Nav 1.8 and calcium channel α2δ-1 subunit, these changes were not consistent when the protein loading was normalised against Na⁺/K⁺-ATPase (Figure 4B–C). We conclude that there are no significant changes in the expression of these proteins during neuroma formation.

Examination of local protein synthesis in neuromas

Neuromas and control saphenous nerves were incubated in vitro with S³⁵ methionine to detect de novo protein synthesis in the absence of the cell body/DRG. 2-DE was then used to separate and compare protein synthesis between the neuroma and control samples. Interestingly, most of the S³⁵-labelled protein spots detected were found in the neuroma sample and only a few were present in both sample types (Figure 5). S³⁵-labelled and colloidal Coomassie blue-stained spots were picked and digested for identification by MS. Vimentin, moesin, gamma-actin and neuron-specific protein neurofilament-L were among the proteins identified (Table 3). This demonstrates that gene-expression-independent translation of neuron-specific proteins occurs in neuromas. The presence of mRNA encoding a variety of proteins in DRG sensory neurons has been demonstrated by Willis et al [37]. In addition, protein changes in the central nervous system have been measured that are related to peripheral nerve damage [38]. Zheng et al provided evidence that DRG axons can synthesise proteins after crush [39], and these experiments confirm these findings. It is striking that the neuroma, defined as a useful model of the nerve terminal shows much higher levels of protein synthesis than isolated saphenous nerve. This is consistent with a significant role for local protein synthesis in regulation of terminal excitability in normal nerve endings.

Spot no	Protein name	Accession no.	MOWSE score	Est. mass	Est. pI	% Cov
53	Annexin A5	gi|74142393	172	35787	4.83	56
39	Vimentin	gi|2078001	127	51590	4.96	52
45	Actin, gamma 1 pro-peptide	gi|4501887	86	42108	5.31	33
11	Moesin	gi|70778915	68	67839	6.22	35
60	Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide, isoform C (14-3-3 zeta)	gi|148676868	61	29240	4.71	35
41	Vimentin	gi|31982755	57	53712	5.06	22
15	Protein disulfide isomerase associated 3	gi|112293264	52	57099	5.88	28
92	mTERF domain-containing protein 1, mitochondrial precursor	gi|81901619	46	47385	6.18	16
25	Unnamed protein product	gi|74210074	45	30222	9.34	26
38	Heat shock protein 1	gi|31981679	45	61089	5.67	37
12	Protein phosphatase 2, regulatory subunit B, delta isoform	gi|148685891	42	33815	6.18	9
50	Creatine kinase	gi|10946574	41	42971	5.4	24
47	Neurofilament-L	gi|200038	36	61542	4.63	11
62	Unnamed protein product	gi|26352624	40	15849	11.26	7
69	ORM1-like protein 1	gi|81174966	41	17401	9.56	37
73	S100 calcium binding protein G	gi|14789635	41	9150	4.69	57
8	Annexin A6	gi|31981302	42	76294	5.34	22

All proteins were identified by LC-ESI-MS/MS peptide sequencing. NCBI gene identifier (gi|) number, MOWSE score from Mascot searches using LC-MS/MS data, estimated molecular mass and pI, and the percentage coverage of the protein sequence where matching peptides were found are given for each protein.