We have previously employed a classical fractionation approach for the purification of synaptic membranes and cytosol from CA3-CA1 hippocampal synapses 19. The starting material was obtained from dissociated cultures, which were prepared from neonatal rat brain (P4-P5) after microdissection of the CA3-CA1 region. In the present study the same source of synaptic material was used to obtain pure synaptosomes and generate 2DE maps of this subpopulation of hippocampal synapses. To evaluate the extent of enrichment in synaptic proteins and the degree of contamination with glia and myelin proteins, subcellular fractions were analyzed by standard electron microscopy (EM) and by western blotting (see Additional file 1). These two methods showed that the degree of sample pureness was indeed much greater in protein extracts from cultures than in extracts from CA3-CA1 hippocampal tissue (19 and unpublished data).

In figure 1 is presented the 2D reference map of synaptosomes purified from CA3-CA1 cultured neurons. Around 1000 spots were detected in the size range 15–150 kDa with isoelectric point (pI) values comprised between 4 and 9. The general appearance of these 2D maps, the molecular weight (MW) and pI distributions, the number and the relative intensity of individual protein spots were all reproducible in different experiments (See below; n = 12, number of gels used for this study to. This number includes n = 9 analytical and preparative gels, and n = 3 gels used for Western blot analysis; the number of spots detected in the best analytical silver stained gels which were used for differential expression analysis is n = 1154 ± 231, mean ± sem; n = 3 gels.)

To obtain a set of reference points to efficiently put in frame and compare different gels, we selected individual protein spots and subjected them to trypsin digestion and MALDI-TOF MS analysis. These spots were chosen among those that apparently gave the most characteristic features of synaptosomes maps so that the whole map area was well represented (n = 105 spots; see methods for details). The list of reference proteins here used is presented in table 1. A number of specialized pre-synaptic molecules, ion channels and transporters, adhesion and cytoskeletal elements molecules, components of various metabolic pathways, mitochondria components, chaperones, proteosomal subunits, receptors and signaling molecules could be identified among these reference points (table 1). In parallel experiments, 2D gels were blotted and immunolabelled with specific antibodies to localize additional references based on well known synaptic markers (n = 3; see method for details; table 1). Despite the finding that sometime 2D gels could present slightly distorted migration pattern, with variation in the spatial distribution of protein spots and MW – pI markers, the reference spots could always be recognized and allowed a straightforward comparison of gels from different cell culture preparations (n = 9). Based on this reproducibility we can then conclude that 2D maps of synaptosomes from cultures can be reliably obtained.

To evaluate differences between cultured synapses and those that can be purified from mature brain tissue we obtained 2D protein profiles of CA3-CA1 hippocampal tissue synaptomes (n = 3; Figure 2). To allow a more straightforward comparison of both set of maps, the same reference spots selected for cultured synapses were searched in tissue maps and subjected to MALDI-TOF analysis. We could recognize and sequence 73 of the original reference points while 8 additional reference points were added for a total of 81 reference points for the tissue map (see list in table 1). Using common reference spots we could compare the two sets of maps. Both the macroscopic appearance and the number of spots were found to be reasonably conserved (number of spots in CA3-CA1 hippocampal tissue n = 1406 ± 378, mean ± sem; n = 3) (Figure 2, panel A). This is somehow expected based on the common origin of these samples and on the equivalent developmental age and degree of maturation of synapses (P4-P5 rats plus 14 days of development in vitro for cultured synapses; P21 rats for brain tissue). Despite this general similarity, when these 2D maps were compared in greater details, some clear differences in protein profiles and in relative protein abundances emerged, with only ~50% of spots that were found to have similar expression levels in both sets of gels. The latter finding cannot be accounted by neither a difference in total protein content (proteins loads were comparable) nor by a decreased in the signal to noise ratio in gels of cultured synapses. The distribution of spot intensities and gel background were indeed always well separated from each other and the results were found to similar in both groups (Figure 2, panel B). This excludes a simple detection problem as a source of discrepancy. In figure 2 (panel C) are presented some magnified regions of these 2D maps to illustrate some cases of proteins whose expression differed between cultures and tissue up to exemplars well below the threshold for detection in either one of the two data sets. Some of the proteins listed in table 1 are likely to be protein contaminants from non neuronal sources. Based on their volumes in 2D maps, these protein contaminants were clearly more prominent in extracts from brain tissue, an expected finding according to the abundance of non synaptic structures seen in tissue synaptosomes preparations by EM analysis (unpublished data; see discussion).

In parallel experiments we obtained two-dimensional gel electrophoresis maps of crude cytosolic extracts from CA3-CA1 hippocampal cultures. Because of the very low density of glial cells and because of the very large extent of neuronal dendrites in these cultures, most of this material is likely to derive from cytosol and endoplasmic reticulum of dendrites. Across different experiments the general appearance and the number and the distribution of protein spots from cytosolic extracts was found to be reproducible (n = 12, number of gels used including n = 9 analytical and preparative gels, and n = 3 gels used for Western blot analysis; the number of spots detected in the best analytical silver stained gels which were used for differential expression analysis is n = 1099 ± 236, mean ± sem; n = 3 gel). In figure 3 (panel A) two exemplar maps of cytosolic and synaptic extracts of CA3-CA1 hippocampal cultures are compared. Already at the visual level, clear differences in the spatial distribution of spots can be appreciated, indicating that we are dealing with two largely different protein sets. To better compare the spatial distribution of protein spots in these two maps, the detected protein spots were color coded and spatially superimposed (Figure 3, panel C). The digital superimposition shows the enrichment in low pI proteins for synaptosomes and in low MW proteins for cytosolic fractions. As indicated in Figure 3 (panel B), these results cannot be attributed to different signal to noise ratios because in the two sets of gels distributions of spot intensities were similar and well separated from background noise.

When comparing synaptic and crude cytosolic maps of CA3-CA1 cultures, we searched for the same protein spots used as markers of the synaptosomal maps. We could recognize and identify by MALDI-TOF MS 89 of the original reference points while 16 novel additional reference points, more easily appreciated in cytosolic maps, were sequenced for a total of 105 reference points (Figure 4) (see list in table 1). The identified proteins belong to different functional groups including: soluble components of synapses (synapsin II), metabolic and mithocondria components, chaperones, elements of the ubiquitin-proteosomal system, cytoskeletal elements, signaling or regulatory molecules (table 1). As expected very few trans-membrane proteins were present except for some components of the endoplasmic reticulum and of the endosomal compartment. Several proteins listed in table 1 were present in both synaptic and cytosolic fractions. On the contrary, many cases of proteins, with clear differences in relative expression levels, including some which were selectively absent in either one set of maps were clearly identified (Figure 4, panel B; additional file 2 and 3). In particular proteins belonging to metabolism and protein fate (synthesis, folding, modification and destination) functional groups were specifically expressed or the expression level was increased in the crude cytosolic fraction (additional file 2). On the contrary, synaptic components were mostly detected only in the synaptosomal fraction (additional file 2). These results suggest that many of these elements might be steadily required in one or both cellular districts or might shuttle between cytosol and synapses.

Postnatal CA3-CA1 hippocampal cultures were prepared from P4-5 neonatal rats essentially as previously described 16. Briefly, the hippocampal regions, CA3-CA1, were removed from rat brains and neurons recovered by enzymatic digestion (trypsin type XI, 10 mg/ml, DNAse I type IV, 0.5 mg/ml) followed by mechanical dissociation. Cells were cultured in Minimal Essential Medium, glucose 0.6%, glutamine 1 mM, NaHCO₃ 2.4 g/l, bovine transferrin 100 mg/ml, insulin 25 mg/ml, fetal calf serum 5% and plated at a density of 50,000 per dish onto petri dishes (Falcon, 30 mm) coated with polyornithine and Matrigel (Collaborative Research). Cultures were maintained at 37°C in a 95% air/5% CO₂ humidified incubator. The culture media was replaced every 3 to 4 days. From the second day in culture, the media was supplemented with cytosine-D-arabinofuranoside (5 μM). Synaptosomes were prepared from hippocampal neurons after 14 days in culture. The culture media and the fetal calf serum were obtained from GIBCO (Italy), all other chemicals were obtained through Sigma (Italy).

Synaptosomes were obtained from control CA3-CA1 hippocampal cultures and tissue (P21 rats), the latter after surgical microdissection of the CA3-CA1 region. For culture experiments, Petri dishes (3 cm diameter) were briefly washed with a phosphate buffered saline (4°C) and then neuronal cells were detached in 0.32 M sucrose (100 μl/Petri dish) by gentle scraping in the presence of a protease inhibitor cocktail. Neurons were homogenized using a small capacity (2 ml) Teflon/glass homogenizer in a final volume of 500 μl (5 Petri dishes for each cycle) at 4 C. After all Petri dishes were processed, the total homogenate was transferred into small centrifuge tubes (Beckman Ultraclear 344057 Centrifuge tubes) and centrifuged at 1200 g/10 × min (4°C) using a fixed angle rotor (Beckman TL100 Ultracentrifuge). The resultant supernatant was then centrifuged again at 12.500 g/20 × min (4°C) to obtain a mitochondria- and synaptosome-enriched pellet and the crude cytosolic fraction (containing also microsomes). Similar methods were used to prepare the synaptosomal and crude cytosolic fractions from CA3-CA1 hippocampal tissue utilized for the generation of 2D maps employed in differential analysis. In some experiments with brain tissue, to set up the method for protein extraction, synaptosomes were further purified using multiple differential gravity centrifugation 29. At the end of the procedure, synaptosomal and crude cytosolic fractions were collected and quickly frozen in small aliquots. The amounts of proteins were quantified by Bradford assay (Bio-Rad, Hercules, CA). On average, a petri dish (3 cm diameter; ~50.000 cells) yielded ~100 μg of total proteins and after subfractionation ~2–3 μg of synaptosomal proteins. Before electrophoresis, synaptosomes were resuspended in lysis buffer. Various lysis buffers were tested: a) 8 M urea, 4% CHAPS, 100 μl resuspension volume; b) 2% SDS, 100 l resuspension volume; c) 1% SDS in 100 mM PBS, 100 μl resuspension volume; d) 10% Zwittergent 3–10 (Calbiochem, Milano), 100 μl resuspension volume; e) 10% Zwittergent 3–10, 1 ml resuspension volume. In most experiments the lysis buffer (e) (10% Zwittergent 3–10, 1 ml resuspension volume) was used because it yielded the best protein display. Synaptosomal and cytosolic proteins were then precipitated by cold acetone (80% final concentration; 2 hr, -20°C) followed by centrifugation at 8000 g for 30 min (4°C). Protein pellets were then washed twice with acetone 80%, dried under a flow of nitrogen and then resuspended in 350 μl of 6 M urea, 2 M thiourea, 4% CHAPS, 2% IPG-buffer 3–10 NL (GE Healthcare, Milan, Italy), 65 mM DTT, 0.05% Bromophenol blue. At this stage sample protein concentration was re-determined.

Proteins were separated by 2D gel electrophoresis. Protein samples were applied to 18 cm IPGstrips pH 3–10 NL (GE Healthcare) by in-gel rehydration (1 hr at 0 V followed by 8 hrs at 30 V; 20°C). Analytical 2D gels were loaded with 100 μg of proteins while preparative gels, for mass spectrometry identification of protein spots, were loaded with 1 mg of proteins (equivalent to 50–500 culture Petri dishes/gel). Focusing was performed with an IPGphore system (GE Healthcare) at 50 μA max per IPG strip with a gradient voltage (8000 V max) for a total of 65 kVh. Strips were equilibrated for 18 min in 0.5 mM pH 6.8 Tris-HCl buffer containing 6 M urea, 30% glycerol, 2% SDS, 0.05% Bromophenol blue and 2% DTT, then for 5 min in the same buffer containin 2.5% iodacetamide in place of DTT. The strips were then transferred onto 9–16% gradient acrylamide SDS-PAGE gels (20 × 20 cm, × 1.5 mm) for the second dimension separation (20 mA per gel at 14°C). Protein spots in analytical and preparative gels were visualized by highly sensitive silver staining 30 or MS compatible silver stain 31 respectively. Apparent Mr was estimated by comparison with values obtained from molecular weight reference markers (Precision, Bio-Rad). pI values were calibrated as described in the GE Healthcare guide lines for estimation of pI values for 18 cm IPG pH 3–10NL. Most of the reagents here used were from Sigma (St. Louis, MO). Zwittergent 3–10 was obtained from Calbiochem.

Proteins from total homogenate and synaptosomial fraction resolved on 10% acrylamide SDS-PAGE and some of the preparative 2DE gels were electrotransferred to nitrocellulose sheets (tris-glycine methanol buffer, 70 mA, overnight at 4°C) and immunolabelled with some specific antibodies against well known synaptic markers (the same nitrocellulose sheets were probed several times with different synaptic antibodies). The antibodies used in these experiments were: anti glutamate receptors GluR1 (rabbit policlonal made in house, 1:500); anti-P38 (Synaptic System, 1:500): anti-synaptotagmin-1 (rabbit policlonal against the lumenal domain, made in house, 1:1000); anti-VAMP-2, anti-synaptophysin, anti-syntaxin1A, anti-rab3A and anti-Chromogranin A (Synaptic system,1:500); anti-sec8 (Stressgene), anti-βActin and anti-βtubulin (Sigma, 1:1000). Antibodies reactivity was revealed by using appropriate HRP-conjugated secondary antibodies followed by chemiluminescence reaction and film exposure. By Ponceau red staining of nitrocellulose membranes and comparison with protein spots visualized in the silver-stained gels, antibodies reactivity was used as landmarks to precisely put in frame the reference 2D map.

Stained gels were scanned at high resolution using a Personal SI Laser densitometer (Molecular Dynamics, Sunnyvale, CA). The resulting 2D patterns were analyzed using Progenesis PG240 v2006 software (Nonlinear dynamics, Newcastle, UK). Statistical analysis was performed using GraphPad Prism 4, V4.03 software (GraphPad Inc., SanDiego, CA). The statistical analysis of protein expression changes (evaluated as normalized spot volume) was performed by using unpaired Student's t-test with two-tailed p value. In all analysis, p < 0.05 was considered to be statistically significant. To apply a more stringent and reliable statistical criteria to the analysis of expression changes, we considered only changes affecting those spots that could be reliably identified in all gels belonging to the same group. This criterion was applied only after automatic spot detection which was run together in all gels according to a combined warping and matching tool available in the PG240 software. Furthermore, only statistically significant differences showing a variation above 1.5-folds were considered as relevant. To assess the signal to noise ratio of the single experiment, quality related information was extracted from gel images. In particular, the distribution of the grey intensity levels of the pixels belonging to the background, as excluded from the areas segmented as protein spots, was calculated as well the intensity distribution for the spot peaks. The juxtaposition of the two density functions, estimated from the occurrence frequencies of the intensity levels, provides a qualitative measurement of how the signal emerges from the image background. For a comparison of the spatial distribution of the detected spots in gel images corresponding to different cellular compartments, a two colors map was adopted, where positions are considered in terms of pI and MW, with a resolution of 0.1 pI and 1 kDa. The histograms of the relative abundance of detected spots along each dimension, for both the compared channels were obtained from the linearized map.

Protein spots were visualized by staining preparative gels with MS compatible silver stain 32. Spots of interest were excised from gels, reduced, alkylated and digested overnight with bovine trypsin (Roche), as previously described 31. After tryptic digestion of the fractionated/resolved proteins, the resulting peptides mixtures were analyzed using MALDI-TOF MS.

One μL aliquots of the supernatant were used for MS analysis using the dried droplet technique and CHCA as matrix. MS spectra were obtained on a MALDI-TOF Voyager-DE STR (Applied Biosystem) mass spectrometer. Alternatively, when no clear identification was obtained, acidic and basic peptide extraction from gel pieces after tryptic digestion was performed (with ammonium bicarbonate 50 mM and 50% acetonitrile or 10% Formic acid and 50% acetonitrile). The resulting peptide mixtures subjected to a single desalting/concentration step before MS analysis over Zip- TipC18 (Millipore Corporation, Bedford, MA, USA). Spectra were internally calibrated using trypsin autolysis products and processed via Data Explorer software. Proteins were identified by searching a comprehensive non-redundant protein database using both ProFound and Mascot programs 3334 for cross validation of protein identifications. One missed cleavage per peptide was allowed, and an initial mass tolerance of 50 ppm was used.