From 2004, HUVEC.com (www.huvec.com) shared a public database relative to human umbilical vein endothelial cells (HUVECs) proteome as assessed by the classical peptide mass fingerprinting approach combining two-dimensional gel electrophoresis (2-DGE) and Matrix Assisted Laser Desorption-Ionization Mass Spectrometry (MALDI-TOF-MS) [1]. More than 160 identifications were obtained corresponding to the major Coomassie-stained proteins separated under standard 2-DGE conditions [2,3]. Although encountering a good audience with more than 100,000 visits till September 2012, HUVEC.com database now appears as notably insufficient especially because of being restricted to a relatively low number of major, mainly soluble endothelial proteins. In the goal to further enrich HUVEC.com, we used Fourier transformed ion cyclotron resonance-mass spectrometry (FTICR-MS) applied to an unstained 2-DGE gel cut in 221 equal rectangles to avoid the relatively poor sensitivity and the spot overlapping inherent to 2-DGE with classical staining [4,5]. This study also allowed for identifying some protein species in HUVECs, such as heat-shock proteins and proteins from the cytoskeleton.

Two identical gels were prepared: the first gel (or control gel) was stained successively with CCB then with silver nitrate (not shown); the second gel was cut in 221 equal rectangles without staining, the resulting grid pattern being matched against the stained control gel (Figure 1). After in-gel trypsin digestion of each rectangle, their respective protein content was analyzed using nanoLC ESI-FTICR-MS/MS. Using stringent thresholds filters, i.e. at least two different peptide sequences with Xcorr and DeltaCn of 2.5 and 0.1 respectively [13], 521 distinct proteins were unambiguously identified (alphabetical list of identified proteins available at www.huvec.com). Furthermore, each identified protein could be individually visualized on a grid 2-D gel according to a “score imaging” deduced from rectangle locali-

zation (xand y-axis) and corresponding consensus scores from Sequest^TM database search (z-axis).

This approach permitted relatively accurate protein location due to rectangle dimensions (~6.0 × 10.5 mm). Furthermore, in absence of any interferences related to the coloration process, it allowed protein detection with a high level of sensitivity. For example, while 5 spots were detected using silver nitrate in rectangle #118, 22 proteins were identified using our approach corroborating the ability of FTICR-MS for the identification of proteins in very low amounts.

It has been shown that Sequest^TM peptide hits and by extension the associated identification consensus scores resulting from LC-MS analysis could be used for labelfree relative protein quantification [14]. Thus, although not giving a relative quantification between different proteins, the score imaging could provide a pattern of the relative abundance of a given protein in each area of the gel. To illustrate this point, we focused on the 27 kDa mammalian heat shock protein (HSP27) whose score imaging data are presented in Figure 2. Using a linear scale for Sequest^TM consensus score, three major spots corresponding to phosphorylation isoforms were visible at pI 5.2, 5.6 and 6.0 (Figure 2(a) and three-dimensional Figure 2(b)) in agreement with published proteomic data relative to HSP27 phosphorylation isoforms [2,15]. When switching to a logarithmic scale (Figure 2(c)), HSP27 was shown to be detected all over the area between pI 5.1 and 6.3.

In the field of spot overlapping and protein background, since each protein identification was considered individually, the corresponding score imaging could not theoretically be “contaminated” by other proteins. Nevertheless, the example of rectangle #110, in which only two major actin isoforms have been detected in this wellknown “overcrowded” 2-D gel area, strongly suggests that ion suppression effects [16] could have arisen in some rectangles notably where particularly abundant proteins were present. Further, when considering individually an abundant protein such as actin, tubulin or vimentin, it appears that the high sensitivity of the method could allow its detection/identification in highly numerous adjacent rectangles (covering up to ~ 75% of the gel area for actin) probably dealing with 1) insufficient isoelectric focusing (horizontal streaks) and insufficient SDS protein loading (vertical streaks) during 2-DGE [17] and 2), with protein complexes and fragments (vertical streaks; isolated “spots”) or protein isoforms (+/− horizontal streaks). Concerning isoforms, it should nevertheless be noted that the presented method, by sequencing specific peptides, was able to identify various complex and intricate isoforms as illustrated in Figure 3 for some tubulin β isoforms.

In biological terms, among the 521 proteins identified, FTICR-MS allowed to unveil a lot of proteins not yet known in HUVECs nor in other endothelial cells (ECs), and for many of them at low cellular concentration. As shown in Figure 4, 122 proteins (23%) were previously identified in HUVECs, 127 (24%) were identified only in other ECs, 206 proteins (40%) were not previously identified in HUVECs nor in other ECs and 66 (13%) possessed ambiguous or uncharacterized biological function. Furthermore, the identified proteins could be sorted into eight general categories i.e. sugar metabolism (37 proteins, 7%), mitochondria and lipid metabolism (30 proteins, 6%), amino acid and vitamin metabolism (41 proteins, 8%), cytoskeleton and organelles (160 proteins, 30%), replication and protein biosynthesis (129 proteins, 25%), detoxification and anti-oxidant defenses (25 proteins, 5%), membrane receptors, channels and transduction (67 proteins, 13%) and lastly, proliferation, cancer and apoptosis category (32 proteins, 6%).

The presented 2-DGE/FTICR-MS-based method constitutes an original, sensitive, and semi-quantitative alternative to classical 2-DGE staining for the establishment of protein databases. When applied to HUVECs, i.e. the most popular endothelial cell model in humans, it allowed to unambiguously identify and further localized on a 2-D gel 521 endothelial proteins representing to date the most protein-rich and informative database for HUVECs. The grid 2-D gel with links to identified proteins and related score imaging, as well as the alphabetical list of identified proteins (also linked with score imaging), are freely available at www.huvec.com.

Financial support from the TGE FT-ICR for conducting the research is gratefully acknowledged.