An effective plasma membrane proteomics approach for small tissue samples

Advancing the quest for new drug targets demands the development of innovative plasma membrane proteome research strategies applicable to small, functionally defined tissue samples. Biotinylation of acute tissue slices and streptavidin pull-down followed by shotgun proteomics allowed the selective extraction and identification of >1,600 proteins of which >60% are associated with the plasma membrane, including (G-protein coupled) receptors, ion channels and transporters, and this from mm3-scale tissue.

. Workflow for plasma membrane proteomic analysis of small tissue samples. (a) Dissect organ of interest, like mouse brain, in artificial cerebrospinal fluid (aCSF). (b) Make slices, allow recovery, label with EZ-Link Sulfo-NHS-SS-Biotin. After quenching, dissect region of interest like the visual cortex (c, red) and mechanically homogenize (d). (e) Separate the plasma membrane protein (PMP) enriched fraction (P) from the rest of the proteome (wash-through, W) by streptavidin pull-down. Panel (f) illustrates SDS-PAGE for P and W. After digestion (g) analyse the protein samples and annotate (h). c adapted from 30 . Scale bar: 1 mm.
Scientific RepoRts | 5:10917 | DOi: 10.1038/srep10917 all 5 wash-through fractions. For each of these protein samples we calculated the relative proportion of protein quantity in its PMP enriched fraction to the initial total protein content, that is the sum of the wash-through and PMP enriched fraction. The percentage of proteins in the PMP enriched fraction from each of the extracts ranged between 6.0 and 7.2%.
The clear dissimilarity in band pattern between a PMP enriched fraction and wash-through of one and the same ASBA extract (Fig. 1f), is indicative of a clear difference between the proteins retained on the Streptavidin agarose resin versus those in the eluent. This prompted us to identify the proteins present in the two fractions of each of the 5 brain samples using shotgun proteomics (Table 1 and Supplementary Data 1 and 2). An intermediate step of tube-gel digestion on 25 μ g proteins per fraction improved solubilisation and digestion efficiency of membrane proteins 21 , and facilitated the removal of detergents prior to mass spectrometric analysis of 1 μ g samples (Fig. 1g,h). Next, we used IPA to categorize each identified protein present in the 5 PMP enriched fractions into their subcellular compartment, as an extra validation of the capability of the ASBA method to truly enrich PMPs from a proteomic sample. The percentage of proteins categorised as PMPs by IPA in the 5 separate PMP enriched fractions ranged between 26.8 and 28.8%, illustrating enrichment in and reproducibility of our workflow (Table 1).
For a more detailed analysis of the plasma membrane proteome, we then merged all identifications of the 5 PMP enriched fractions together into 1 protein list. We also merged the identification lists of the 5 wash-through fractions. This resulted in respectively 1,698 and 2,872 discrete proteins that were identified in that PMP enriched fraction and wash-through fraction (Table 1). Of these 1,698 proteins identified in this PMP enriched fraction, IPA successfully annotated 1,625 proteins. Out of these 1,625, 417 proteins or 25.7% were classified as PMP (Table 1 and 2). Because IPA only provides one subcellular localization per protein and does not consider the additional cellular compartments in which a protein can occur 15 , secondary annotations were also checked in IPA and DAVID, in combination with an intensive literature search 6 . As such, a large number of proteins (372) could be additionally assigned to potentially reside in association with (peripheral proteins) or even be fully embedded within the plasma membrane (integral proteins), leading to a total of 789 or 48.6% of PMPs in the PMP enriched fraction ( Table 2, Supplementary Table 1, Fig. 1h). Of note, this additional analysis classified an even larger subset of PMPs as ion channel, transmembrane receptor, transporter or G-protein coupled receptor (Table 2  and Supplementary Table 1). Together with these 789 annotated PMPs, another 8 proteins located at the cell surface (0.5%), 163 at the cellular membrane (10.0%) and 51 proteins in the extracellular space (3.1%) (Supplementary Table 2), this accounts for 1,011 or 62.2% proteins in the PMP enriched fraction that have been reported to reside in or near the cell surface (Fig. 1h). The remaining 37.8% proteins in the PMP enriched fraction might be the result of co-purification of large intracellular complexes, with the biotinylated proteins still associated to the plasma membrane or with the readily releasable vesicle pools. These proteins deserve attention in future research to either confirm or exclude their capacity to potentially reside at the PM. In sum, our yields are in agreement with or even higher than recent PMP enrichment studies based on aqueous two-phase affinity partitioning of a much larger tissue sample, a complete rat or mouse cerebellum 13,22 , or on biotinylation and affinity purification of cell surface proteins of cultured mouse cortical neurons 14 .

Continued
The efficiency of the presented workflow for the enrichment of PMPs was further validated using the DAVID web tool. This tool allows visualization of the specific protein enrichment in the PMP enriched and wash-through fractions by a gene ontology enrichment analysis on all protein identifications of each fraction relative to a background. The background was built by merging all proteins identified in the PMP enriched with those from the wash-through fraction into one background data set. We used the functional annotation charts of the DAVID web tool based on cellular component ontology and visualized the results in ReViGO treemaps (Supplementary Fig. 1 and 2). The treemap adapted from ReViGO for the PMP enriched fraction (Supplementary Fig. 1) is summarized in Table 3, illustrating an enrichment of proteins associated to the cell surface specific for neuronal cells in the clusters 'plasma membrane' , 'plasma membrane part' , 'ion channel complex' , 'intrinsic' and 'integral component of PM' , and the clusters 'synapse' , 'synapse part' , 'neuron projection' , 'dendritic spine' , and 'cell junction' . The treemap adapted from ReViGO resulting from our wash-through data set ( Supplementary Fig. 2) summarized in Table 4 shows clusters of proteins associated with different intracellular organelles, especially with  mitochondrial function and the 'respiratory chain' . This reflects the high energy demand and oxygen consumption of neurons, and thus the high metabolic rate of the tissue under study [23][24][25][26][27] . Other clusters contain proteins with a role at the envelope, within the endomembrane system, and within the membrane-enclosed lumen. Importantly, no clear cluster was suggested for cell surface-associated proteins for the wash-through fraction.
In conclusion, in this report we present the new combination of a procedure for the specific extraction of cell surface-associated proteins including PMPs originating from mm 3 -scale tissue derived from acute tissue slice preparations, with proteomic analysis. This reproducible and efficient enrichment methodology is undoubtedly applicable to many different tissue types and can significantly contribute to future differential plasma membrane proteomics research in many fields of application ranging from neuroscience, cancer, and immunology, to stem cell research. MgSO 4 , 2 mM CaCl 2 , 1.2 mM KH 2 PO 4 , 25.6 mM NaHCO 3 , and 10 mM D-glucose; pH 7.4) saturated with 95% O 2 and 5% CO 2 . Subsequently, each brain was separated in half along the longitudinal fissure, and the left hemisphere was cut into 300 μ m-thick coronal slices using a Vibratome (HM 650 V, Prosan). A thickness of tissue slices between 300 and 400 μ m is essential to biotinylate an amount of functionally healthy and intact cells that will outnumber the sliced ones 28,29 . Four slices between Bregma level -2.70 and -4.16 of each hemisphere were placed in an incubation chamber (65-0076/BSC-PC, Harvard Apparatus by Warner Instruments) filled with aCSF and provided with a continuous flow of 95% O 2 and 5% CO 2 for 90 min in order to recover.
Next, the sections were put on ice and supplemented with CO 2 and O 2 throughout the whole biotinylation procedure. They were washed twice in ice-cold aCSF, incubated for 45 min with EZ-Link Sulfo-NHS-SS-Biotin (0.5 mg ml −1 in aCSF; Thermo Scientific) and washed twice with ice-cold aCSF complemented by 100 μ M lysine (Sigma) to block the remaining reactive Sulfo-NHS-SS-biotin. It has already been demonstrated that these tissue slices stay viable with the cells intact during the complete process. Incubation on ice during biotin labelling will limit protein internalization to a minimum in order to create a snapshot of the cell surface proteome. The Sulfo-NHS-SS-biotin can reach all cell layers Scientific RepoRts | 5:10917 | DOi: 10.1038/srep10917 of a slice up to 350 μ m-thick after incubation with minimum 100 μ M for at least 45 min. Because of the intact cell membranes and the membrane impermeability of the biotin label, cell surface-exposed proteins will be biotinylated and labelling of intracellular proteins will be minimal 16 , as substantiated with our approach and data set.
The sections were washed and kept in ice-cold, saturated aCSF until dissection of the region of interest under a binocular microscope (ASZ30E; Bausch & Lomb). The borders of the visual cortex were determined based on the stereotaxic mouse brain atlas 30 . For each brain, the dissected visual cortex tissue from the four slices was collected in 100 μ L lysis buffer (1% Triton X-100, 0.1% sodium dodecyl sulphate (SDS), 1 mM ethylenediaminetetraacetic acid (EDTA), 50 mM NaCl, 20 mM Tris; pH 7.5) and 4 μ l of a mix of protease inhibitors (Roche). After mechanical homogenization and centrifugation (10,000g, 5 min, 4 °C) the supernatant was collected and stored at -80 °C. The total protein concentration of the 5 biotin-labelled samples was determined according to the Qubit TM Quantification Platform (Invitrogen) using a Qubit TM fluorometer (Invitrogen, Merelbeke, Belgium).

Isolation of plasma membrane proteins.
Biotin-labelled proteins were separated from the rest of the proteome by a protocol based on biotin's affinity for streptavidin. For this purpose, 150 μ l of at room temperature (RT) calibrated Streptavidin agarose resin (Thermo Scientific) was loaded onto a Pierce ® Spin Cups -Cellulose Acetate Filter (Thermo Scientific). After centrifugation, the column was centrifuged (500 g, 1 min) and washed four times by adding 700 μ l of phosphate-buffered saline (PBS; 0.1 M H 3 PO 4 , 0.15 M NaCl; pH 7.2) and centrifugation (500 g, 1 min). The 5 biotinylated samples, each containing ± 500 μ g of proteins extracted from ± 1 mm 3 tissue, were loaded onto 5 prepared columns and all were shaken for 15 min at RT. After centrifugation (500 g, 1 min), the columns were washed three times with 700 μ l PBS. All of the non-biotinylated proteins were eluted during the wash steps with the vast majority of proteins eluted in the first wash-through. Next, 5 μ l 2% SDS, 45 μ l 200 mM Triethylammonium bicarbonate (TEAB; Fluka Analytical), 45 μ l MilliQ (Merck Millipore) and 5 μ l 200 mM Tris (2-carboxyethyl) phosphine (TCEP; Thermo Scientific) were added to each column and they were incubated for 1 h at 55 °C for denaturation and reduction. Samples were alkylated for 30 min in the dark in 5 μ l 375 mM iodoacetamide (IAA; Thermo Scientific), centrifuged (500 g, 1 min) and the 5 obtained eluents were collected. Subsequently, 25 μ l 200 mM TEAB and 25 μ l MilliQ were added to each column, they were centrifuged (500 g, 1 min), each eluent added to the corresponding previous one and these 5 combined samples, being the plasma membrane protein (PMP) enriched fractions, were stored at -20 °C, together with the 5 first wash-through fractions.
Protein quantity determination. The total protein concentration of the first wash-throughs and the PMP enriched fractions, were determined according to the Qubit TM Quantification Platform (Invitrogen) as described above. For each sample, the relative proportion of protein quantity in the PMP enriched fraction was calculated to the total protein content of the initial sample, the wash-through and PMP enriched fraction combined. For the calculation of these percentages, original Qubit protein quantitation measurements were used and renumbered according to the total volume of each fraction.
Gel-electrophoresis and total protein stain. To 1 μ g protein of each of the wash-throughs and 1 μ g protein of each of the PMP enriched fractions, 5 μ l of XT Sample Buffer solution (4x; Bio-Rad) and 1 μ l of XT Reducing Agent (20x; Bio-Rad) was added. The protein samples were denatured and reduced for 10 min on 70 °C. Proteins were separated on a Criterion TM Precast Gel XT 4-12% Bis-Tris (Bio-Rad) in the Criterion TM Cell (Bio-Rad) system and 4 μ L of the Spectra TM Multicolor High range protein ladder (Thermo Scientific) was used as molecular weight standard. Next, we performed a total protein stain with Serva Purple (SERVA) according to manufacturer's instructions. After the staining, the gel was scanned with the Bio-Rad ChemiDoc TM MP Imaging System. Proteomic analysis. For the 5 wash-throughs and the 5 PMP enriched fractions, 25 μ g was diluted in MilliQ to a total volume of 100 μ L. The samples were subsequently transformed into tube-gels by adding 25 μ L of Acrylamide/Bis solution (40% 29:1; Bio-Rad), 1 μ l SDS (10%), 0.5 μ l APS (ammonium persulfate, 10%; Sigma) and 0.1 μ l Temed (N, N, N' , N' -Tetramethylethylenediamine; Fluka BioChemika) and incubation of 30 min at RT. Peptides were extracted from these tube-gels by in-gel trypsin digestion. The tube gels were cut in pieces of approximately 1 mm 3 . The pieces were washed twice with 50 μ l MilliQ, followed by 3 × 50 μ l acetonitrile. After three cycles of hydration with acetonitrile and rehydration with 100 mM ammonium bicarbonate, the gel pieces were vacuum dried in a vacuum concentrator. To start the enzymatic digestion, 25 μ l of a solution containing 5 ng μ l −1 trypsin (Promega), 50 mM ammonium bicarbonate and 5 mM calciumchloride was added to each gel piece and placed on 37 °C overnight. The next day, the tryptic peptides were extracted using 50 mM ammonium bicarbonate followed by an extraction with 50% acetonitrile and 5% formic acid. This step was repeated twice. Afterwards, the pooled extracts were vacuum dried and the peptides were stored at -20 °C. The equivalent of 1 μ g of total protein was loaded and analysed by nanoLC-mass spectrometry. Liquid chromatography mass spectrometric analysis was performed on a Waters nanoAquity LC system connected to a Thermo Scientific LTQ Velos Orbitrap mass spectrometer. The equivalent of 2 μ g of total protein of the digested sample was dissolved in 20 μ l of 2% acetonitrile in HPLC-grade water. 10 μ l of the sample was loaded onto the Scientific RepoRts | 5:10917 | DOi: 10.1038/srep10917 trapping column (Pepmap C18 300 μ m x 20 mm, Dionex) with an isocratic flow of 2% acetonitrile in water with 0.1% formic acid at a flow rate of 5 μ l min −1 . After 2 min, the column-switching valve was switched, placing the pre-column online with the analytical capillary column, a Pepmap C18, 3 μ m 75 μ m x 150 mm nano column (Dionex). Separation was conducted using a linear gradient from 2% acetonitril in water, 0.1% formic acid to 40% acetonitril in water, 0.1% formic acid in 160 min. The flow rate was set at 400 nl min −1 . The LTQ Orbitrap Velos (Thermo Scientific) was set up in a data dependent MS/MS mode where a full scan spectrum (350-5,000 m/z, resolution 60,000) was followed by a maximum of ten CID tandem mass spectrum (100 to 2,000 m/z). Peptide ions were selected as the twenty most intense peaks of the MS1 scan. Collision induced dissociation (CID) scans were acquired in the LTQ iontrap part of the mass spectrometer. The normalized collision energy used was 35% in CID. We applied a dynamic exclusion list of 45 s. Data analysis. Proteome Discoverer 1.3.0.339 (Thermo Scientific) was used as a workflow manager to handle the data. The tandem MS data were searched using both SEQUEST and Mascot (Matrix science) against the Swissprot database (v 09/2012, 538010 sequences) for Mus musculus taxonomy. All tandem mass spectra in the range of 300 Da to 8,000 Da were interpreted.
Monoisotopic peak assignment, charge state determination, co-isolation interference, and delta mass calculation between the measured and theoretical monoisotopic masses were determined by Proteome Discoverer. The precursor mass tolerance was set at 5 ppm, while fragment mass tolerance was set to 0.5 Da. A maximum of two missed cleavages by trypsine was allowed for. A static modification of 57.021 Da on cysteine was defined to allow for carbamidomethylation. Further, a dynamic modification of 15.9955 Da was introduced to account for possible oxidation of methionine. The use of average precursor masses and average fragment masses was prohibited. Only first ranked PSMs were considered for further analysis. The false discovery rate is controlled by a target-decoy approach on a reversed database. Peptide spectrum matches were found significant at an FDR of 5%.
Protein discoverer 1.3.0.339 was used to combine the significant peptide annotation from Mascot and SEQUEST in a parsimonious protein list.
Protein grouping follows the rule of parsimony. Essentially only the minimal list of proteins that can explain all the peptides in the data set is reported.
Ingenuity Pathway Analyis (IPA ® , QIAGEN Redwood City, www.qiagen.com/ingenuity) and Database of Annotation, Visualization and Integrated Discovery (DAVID, version 6.7) were used for cellular component assignment.
The enrichment of proteins in the PMP enriched fraction was investigated by analysing all identifications within this fraction relative to a background composed of all proteins identified in the wash-through and PMP enriched fraction together with DAVID 31,32 . DAVID summarized the cellular component ontology identified via a functional annotation chart and calculated a Fisher exact test for each ontology as a measure for enrichment within the fraction. Similarly, information about the enrichment of proteins in the wash-through fraction was retrieved. The GO-terms and corresponding p-values with Benjamini correction were subsequently submitted to ReViGO, a web server that Reduces and Visualizes long lists of Gene Ontology terms 33 , and visualized in treemaps that cluster ontologies with high semantic similarity. The size of these cluster representatives, which are joined in different superclusters each indicated by 1 colour, is proportional to the p-values derived by ReViGO. The treemap figures were then adapted by changing the colours of the superclusters.