Loss of adipocyte-type fatty acid binding protein and other protein biomarkers is associated with progression of human bladder transitional cell carcinomas.

Multifocal recurrent papillary tumors provide a unique model system to study the molecular mechanisms underlying the steps involved in transitional cell carcinoma progression and offer a valuable source of material to search for biomarkers that may form the basis for diagnosis, prognosis, and treatment. We have examined the protein expression profiles of normal bladder urothelium and of 63 transitional cell carcinomas of various histopathological grades and T stages using high-resolution, two-dimensional gel electrophoresis, microsequencing, mass spectrometry, and a two-dimensional gel protein database approach for polypeptide identification (http://biobase.dk/cgi-bin/celis). In general, the results revealed a striking similarity between the overall qualitative expression patterns of papillary tumors of all grades, as well as of papillary and solid tumors of grade III. With few exceptions, tumors of grades I-III expressed, albeit at different levels, all of the keratins (7, 8, 13, 17, 18, 19, and 20) found in the normal urothelium. Grade IV tumors lacked or expressed reduced levels of keratin 13 but most resembled low-grade tumors. One invasive grade IV tumor, however, expressed a fibroblast-like protein phenotype. Four proteins that were expressed by normal urothelium and were lost at various stages of progression were identified as glutathione S-transferase mu, prostaglandin dehydrogenase (PGDH), a fatty acid binding protein with homology to the adipocyte isoform (A-FABP), and keratin 13. The percentage of tumors expressing A-FABP was very high in low-grade lesions but decreased drastically (P = 0.0006) in grade III and IV neoplasms. In addition, low-grade tumors contained more A-FABP than their high-grade counterparts. The stage of the disease was also statistically (P = 0.0269) related to the presence or absence of A-FABP in grade III tumors. Similar analysis of glutathione S-transferase mu and PGDH showed a statistically significant decrease of these proteins in high-grade (grades III and IV) tumors (P = 0.0026 and P = 0.0044, respectively). Only PGDH showed a suggestive correlation (P = 0.0775) with the stage of the disease in grade III tumors. Keratin 13 showed a drastic decrease in grade IV tumors. In addition to identifying biomarkers that may have prognostic value, our studies have suggested that A-FABP is an important component of the pathway(s) leading to bladder cancer development.

The master two-dimensional gel database of human AMA cell proteins: towards linking protein and genome sequence and mapping information (update 1991).

The master two-dimensional gel database of human AMA cells currently lists 3801 cellular and secreted proteins, of which 371 cellular polypeptides (306 IEF; 65 NEPHGE) were added to the master images during the last 10 months. These include: (i) very basic and acidic proteins that do not focus under normal running conditions and (ii) low-abundant proteins that can only be detected after prolonged gel exposure. Annotation categories updated in this version include "protein name", "antibody against protein", "cellular localization", and "microsequenced proteins". New entries include "human autoantigens" and "cDNAs". For convenience we have included an alphabetical list of all known proteins recorded in this database. In the long run, the main goal of this database is to link protein and DNA sequencing and mapping information (Human Genome Program) and to provide an integrated picture of the expression levels and properties of the thousands of proteins that orchestrate various cellular functions both under physiological and abnormal conditions.

A two-dimensional gel protein database of noncultured total normal human epidermal keratinocytes: identification of proteins strongly up-regulated in psoriatic epidermis.

A two-dimensional (2-D) gel database of proteins from noncultured total normal human epidermal keratinocytes has been established. A total of 1449 [35S]methionine labelled proteins (1112 isoelectric focusing, 337 nonequilibrium pH gradient electrophoresis) were resolved and recorded using computer assisted (PDQ-SCAN and PDQUEST software) 2-D gel electrophoresis. By matching the protein patterns of total keratinocytes and transformed human amnion cells (master database; Celis et al., Leukemia 1988, 2, 561-602) as well as by 2-D immunoblotting and microsequencing of keratinocyte proteins, it was possible to identify 72 polypeptides in the keratinocyte database. The database also includes data on polypeptides that are synthesized at a higher level by keratinocytes enriched in basal cells, and on six secreted proteins which are produced, albeit at a reduced rate, by normal keratinocytes and that are strongly up-regulated in psoriatic epidermis (Celis et al., FEBS Letters, in press).

Protein-electroblotting and -microsequencing strategies in generating protein data bases from two-dimensional gels.

Coomassie blue-stained, heat-dried, and computer-imaged two-dimensional gels used to develop comprehensive human protein data bases served as the protein source to generate partial amino acid sequences. The protein spots were collected from multiple gels, rehydrated, concentrated by stacking into a new gel, electroblotted onto inert membranes, and in situ-digested with trypsin. Peptides eluting from the membranes were separated by HPLC and sequenced. Using this procedure, it was possible to generate partial sequences from 13 human proteins recorded in the amnion cell protein data base. Eight of these sequences matched those of proteins stored in data bases, demonstrating that a systematic analysis of proteins by computerized two-dimensional gel electrophoresis can be directly linked to protein microsequencing methods. The latter technique offers a unique opportunity to link information contained in protein data bases derived from the analysis of two-dimensional gels with forthcoming DNA sequence data on the human genome.

Computerized, comprehensive databases of cellular and secreted proteins from normal human embryonic lung MRC-5 fibroblasts: identification of transformation and/or proliferation sensitive proteins.

Databases of protein information from human embryonal lung fibroblasts (MRC-5) have been established using computer analyzed two-dimensional gel electrophoresis. One thousand four hundred and eighty-two cellular proteins (1060 with isoelectric focusing and 422 with nonequilibrium pH gradient electrophoresis, in the first dimension) ranging in molecular mass between 8 and 234 kDa were separated and numbered. Information entered in the database (in most cases for major proteins) includes: protein name, HeLa protein catalog number, mouse protein catalog number, proteins matched in transformed human epithelial amnion cells (AMA) and peripheral blood mononuclear cells (PBMC), transformation and/or proliferation sensitive proteins, synthesis in quiescent cells, cell cycle regulated proteins, mitochondrial and heat shock proteins, cytoskeletal proteins and proteins whose synthesis is affected by interferons. Additional information entered for a few transformation-sensitive proteins that have been selected for future studies includes levels of synthesis and amounts in fetal human tissues. A total of four hundred and seventy-six [35S]methionine labeled polypeptides (258 isoelectric focusing; 218, nonequilibrium pH gradient electrophoresis) secreted by MRC-5 fibroblasts were separated and recorded (J. E. Celis et al., Leukemia 1987, 1, 707-717). Information entered in this database includes molecular weight and transformation sensitive proteins. These databases, as well as those of epithelial and lymphoid cell proteins (J. E. Celis et al., Leukemia 1988, 9, 561-601), represent the initial stages of a systematic effort to establish comprehensive databases of human protein information. In the long run, these databases are expected to offer a useful framework in which to focus the human genome sequencing effort.

Comprehensive, human cellular protein databases and their implication for the study of genome organization and function.

Comprehensive, computerized databases of cellular protein information derived from the analysis of two-dimensional gels, together with recently developed techniques to microsequence proteins offer a new dimension to the study of genome organization and function. In particular, human protein databases provide an ideal framework in which to focus the human genome sequencing effort.

Towards establishing comprehensive databases of cellular proteins from transformed human epithelial amnion cells (AMA) and normal peripheral blood mononuclear cells.

Databases of protein information derived from the analysis of two-dimensional gels have been established from transformed human amnion cells (AMA) and peripheral blood mononuclear cells (PBMCs). A total of 1781 [35S]methionine-labeled AMA proteins (1274 IEF, 537 NEPHGE) and a total of 1311 proteins from PBMC (948 IEF, 363 NEPHGE) were resolved and recorded using computerized (PDQ-SCAN and PDQUEST softwares) two-dimensional gel electrophoresis. AMA and PBMC proteins (total, 454: 301 IEF, 153 NEPHGE) were matched both manually and by the computer. Information entered in the AMA database (in most cases for some major proteins) includes: molecular weight, protein name, HeLa protein catalogue number, mouse protein catalogue number, nuclear proteins, phosphorylated proteins, distribution of proteins in Triton X-100 supernatants and cytoskeletons, proliferation- and transformation-sensitive proteins, cell cycle-specific proteins, mitochondrial proteins, proteins matched in normal human embryonal lung MRC-5 fibroblasts and PBMC cells, heat shock proteins, proteins affected by interferons, cytoskeletal proteins, and the presence of antibody against protein in human sera. Additional information has been entered for the cell cycle-regulated and DNA replication protein cyclin (PCNA). Information entered in the PBMC database includes molecular weight and potential markers for sorted populations of lymphocyte subtypes. For those proteins that have been matched to AMA proteins, information contained in some entries may be transferred from the AMA database.

Major proteins in normal human lymphocyte subpopulations separated by fluorescence-activated cell sorting and analyzed by two-dimensional gel electrophoresis.

We have compared the overall patterns of protein synthesis of normal human lymphocyte subpopulations taken from five volunteers using high resolution two-dimensional gel electrophoresis. The lymphocytes were isolated using density gradient centrifugation, labeled with subtype-specific MoAbs, and separated to a high degree of homogeneity by FACS into CD4+ helper T cells, CD8+ suppressor T cells, CD20+ B cells, and N901 (NHK-1)+ NK cells. The four lymphocyte subpopulations were labeled with [35S]methionine for 14 hr, solubilized in lysis buffer, and analyzed by two-dimensional gel electrophoresis (IEF). Of about 1000 proteins resolved in each case, most were found to be common to all subpopulations. However, eight putative markers for B1+ (proteins 5525, Mr = 63,700; 5621, Mr = 63,700; 8311, Mr = 36,900; 2202, Mr = 36,300; 6121, Mr = 30,300; 106, Mr = 29,300; 5009, Mr = 23,000; 8012, Mr = 11,600) and one for N901+ (protein 8129, Mr = 30,400) were identified. In contrast, no major protein markers were found that could differentiate T4+ and T8+ cells from each other or from B cells and NK cells. With the exception of two B1+ markers (proteins 5525 and 5621), lower but variable levels of the other markers were observed in all cell types. All the putative protein markers have been identified in the protein database of human peripheral blood mononuclear cells (PBMCs) (see accompanying article by Celis et al.). Comparison of the overall patterns of protein synthesis of the unsorted PBMCs with those of the four subpopulations showed that the synthesis of some major PBMC proteins decreased substantially in the sorted subsets. These proteins are most likely not of monocyte origin, as these cells constituted only about 15% of the total PBMCs. Also, the inhibition does not seem to be due to the addition of the single MoAbs or to cell cycle differences. Taken together, the data provide a background for further studies of protein profiles in normal (resting or activated) and malignant hematopoietic cells.

Progressin: a novel proliferation-sensitive and cell cycle-regulated human protein whose rate of synthesis increases at or near the G1/S transition border of the cell cycle.

A novel proliferation-sensitive and cell cycle-specific basic protein, termed progressin (Mr = 33,000), has been identified in proliferating human cells of epithelial, fibroblast and lymphoid origin. Progressin is synthesized almost exclusively during the S-phase of transformed human amnion cells (AMA). Increased synthesis of this protein is first detected late in G1, at or near the G1/S transition border, reaches a maximum in mid to late S-phase, and declines thereafter. Contrary to histones, progressin synthesis is not coupled to DNA replication. As expected for an S-phase-specific protein, no detectable synthesis of progressin was observed in non-proliferating human MRC-5 fibroblasts and epidermal basal keratinocytes. Elevated, but variable levels of this protein were observed in proliferating normal fibroblasts and transformed cells of fibroblast, epithelial and lymphoid origin. Taken together the above observations suggest that progressin may be a component of the common pathway leading to DNA replication and cell division.

Secreted proteins from normal and SV40 transformed human MRC-5 fibroblasts: toward establishing a database of human secreted proteins.

Analysis by means of computerized two-dimensional gel electrophoresis (NEPHGE, IEF) of the [35S]-methionine labeled proteins secreted by normal human MRC-5 fibroblasts revealed 476 polypeptides (258 acidic and 218 basic), many of which appeared as charge trains due to modification. Similar analysis of the proteins secreted by SV40 transformed MRC-5 fibroblasts (MRC-5 V2) showed a striking decrease in the levels of many of these proteins as well as the appearance (or increased synthesis) of 47 polypeptides that were either absent or present in very low amounts in normal cells. Of the major secreted polypeptides whose relative proportion decreased dramatically in the MRC-5 V2 cells, 15 were found to be abundant components of other normal (nontransformed) fibroblasts (W138, Xeroderma pigmentosum cell lines). Low levels of these radioactively labeled polypeptides were observed in transformed human cell lines of fibroblast (W138, SV40, HT1080), epithelial (HeLa, transformed amnion cells (AMA), A431, A459) and myeloid (HL-60) origin. No major secreted polypeptide from MRC-5 V2 cells was synthesized exclusively by the transformed cell lines.

Identification of two human phosphoproteins (dividin and IEF 59dl) that are first detected late in G1 near the G1/S transition border of the cell cycle.

Two-dimensional gel electrophoretic analysis (NEPHGE, IEF) of the [32P]-orthophosphate-labeled proteins synthesized throughout the cell cycle of transformed human amnion cells (AMA) revealed two phosphoproteins (dividin, Mr = 54,000, pl = 8.4; IEF 59dl, Mr = 27,000, pl = 5.7) that are present mainly in S-phase cells. These proteins are first detected at the end of G1, near the G1/S transition border, and their levels reach a maximum late in S-phase. Together with the previously identified nuclear protein cyclin, these phosphoproteins are likely candidates for proteins that may play a role in the regulation of the onset of DNA synthesis and cell division.