Discrimination of fibroblast subtypes by multivariate analysis of gene expression

Immune interactions in hepatic fibrosis

Liver cirrhosis is caused by iterative cycles of tissue injury, inflammation, and repair. Although most causes of acute hepatitis resolve without scarring, chronic hepatitis is associated with persistent inflammation and matrix remodeling, which leads to fibrosis and, eventually, cirrhosis. The mechanisms that govern wound healing involve interactions between the innate and adaptive immune systems and stromal cells within a microenvironment composed of cytokines, growth factors, and modified matricellular proteins. The immune system plays a central role in the regulation of fibrosis, tissue repair, and recovery that is vital for the maintenance of tissue homeostasis. Chronic inflammation and fibrosis are inextricably linked and the cellular interactions between immune effector cells, local fibroblasts, and tissue macrophages at sites of scar formation determine the outcome of liver injury and the development of scarring.

Multipotent mesenchymal stromal cells obtained from diverse human tissues share functional properties and gene-expression profile with CD146+ perivascular cells and fibroblasts

OBJECTIVE

The relationship of multipotent mesenchymal stromal cells (MSC) with pericytes and fibroblasts has not been established thus far, although they share many markers of primitive marrow stromal cells and the osteogenic, adipogenic, and chondrogenic differentiation potentials.

MATERIALS AND METHODS

We compared MSCs from adult or fetal tissues, MSC differentiated in vitro, fibroblasts and cultures of retinal pericytes obtained either by separation with anti-CD146 or adhesion. The characterizations included morphological, immunophenotypic, gene-expression profile, and differentiation potential.

RESULTS

Osteogenic, adipocytic, and chondrocytic differentiation was demonstrated for MSC, retinal perivascular cells, and fibroblasts. Cell morphology and the phenotypes defined by 22 markers were very similar. Analysis of the global gene expression obtained by serial analysis of gene expression for 17 libraries and by reverse transcription polymerase chain reaction of 39 selected genes from 31 different cell cultures, revealed similarities among MSC, retinal perivascular cells, and hepatic stellate cells. Despite this overall similarity, there was a heterogeneous expression of genes related to angiogenesis, in MSC derived from veins, artery, perivascular cells, and fibroblasts. Evaluation of typical pericyte and MSC transcripts, such as NG2, CD146, CD271, and CD140B on CD146 selected perivascular cells and MSC by real-time polymerase chain reaction confirm the relationship between these two cell types. Furthermore, the inverse correlation between fibroblast-specific protein-1 and CD146 transcripts observed on pericytes, MSC, and fibroblasts highlight their potential use as markers of this differentiation pathway.

CONCLUSION

Our results indicate that human MSC and pericytes are similar cells located in the wall of the vasculature, where they function as cell sources for repair and tissue maintenance, whereas fibroblasts are more differentiated cells with more restricted differentiation potential.

In vitro reconstruction of a tissue-engineered endothelialized bladder from a single porcine biopsy

OBJECTIVE

Augmentation of the urinary bladder using a tissue-engineered approach with autologous cells is a very promising technique. To prevent risks of necrosis after transplantation, the graft vascularization process could be markedly enhanced by incorporation of autologous endothelial cells in the tissue-engineered organ. The purpose of this study was to develop a separation technique to extract four bladder cell types from the same biopsy, and to prepare an endothelialized reconstructed bladder.

MATERIALS AND METHODS

Fibroblasts, smooth muscle cells (SMC), urothelial cells (UC) and endothelial cells (EC) were extracted from a small porcine bladder biopsy. The SMC, fibroblasts and EC were seeded on the top of the sponge and cultured for 10 days. Then, the UC were seeded on top of these cells for 15 additional days to produce a three-dimensional bladder wall.

RESULTS

The UC and EC extracts from a single porcine biopsy were 97.2+/-0.6% keratin 8/18-positive and 97.7+/-0.3% PECAM-1-positive pure cells, respectively, as assessed by flow cytometry. The SMC could not be dissociated from fibroblasts, and were present as 37+/-0.5% desmin-positive cells. UC differentiated into a urothelium characterized by umbrella cells and a laminin-positive basal membrane. The EC reorganized in the matrix to form PECAM-1-positive capillary-like tubes.

CONCLUSION

This new model of tissue-engineered bladder has the main advantages of being at least 2mm thick, autologous, and able to promote the formation of capillary-like tubes. It could be a promising alternative to the use of gastrointestinal segments to improve bladder capacity.

The human H19 gene is frequently overexpressed in myometrium and stroma during pathological endometrial proliferative events

We studied the patterns of H19 expression in normal, hyperplastic and neoplastic human uterine tissues. H19 RNAs were detected by an in situ hybridisation technique (ISH). In both normal and pathological conditions, H19 was expressed in stromal and myometrial cells, but never in epithelial cells. 34/48 carcinomas overexpressed H19 compared with the expression in normal tissues. This high expression was frequently observed in the vicinity of malignant epithelial cells. This suggests that the level of H19 RNA synthesis could be the result of epithelium/stroma interactions. We also demonstrated that several cancerous or immortalised breast epithelial cells release factors into the culture medium, which in turn stimulate H19 expression in stromal cells. The level of H19 expression, estimated by ISH, was not significantly correlated with histological type when all types were considered together (P = 0.108), but was highly correlated to one type of cancer, i.e. carcinomas with an epidermoid component (P = 0.0015). The level of H19 expression was also strongly correlated with tumour invasion of the reproductive organs (P = 0.006) and significantly correlated with neoplastic cell invasion of the myometrium (P = 0.048). In conclusion, our results indicate that H19 overexpression is correlated with the progression of the disease and we propose that this frequent overexpression of the gene in the myometrium and in stroma is a reaction to pathological cell proliferation.

Hunting the primary: novel strategies for defining the origin of tumours

In 1995, two methods of genome-wide expression profiling were first described: expression microarrays and serial analysis of gene expression (SAGE). In the subsequent 10 years, many hundreds of papers have been published describing the application of these technologies to a wide spectrum of biological and clinical questions. Common to all of this research is a basic process of data gathering and analysis. The techniques and statistical and bio-informatic tools involved in this process are reviewed. The processes of class discovery (using clustering and self-organizing maps), class prediction (weighted voting, k nearest neighbour, support vector machines, and artificial neural networks), target identification (fold change, discriminant analysis, and principal component analysis), and target validation (RT-PCR and tissue microarrays) are described. Finally, the diagnostic problem of adenocarcinomas that present as metastases of unknown origin is reviewed, and it is demonstrated how integration of expression profiling techniques promises to throw new light on this important clinical challenge.

A profound alteration of blood TCRB repertoire allows prediction of cerebral malaria

Cerebral malaria (CM) is one of the severe complications of Plasmodium infection. In murine models of CM, Talphabeta cells have been implicated in the neuropathogenesis. To obtain insights into the TCRB repertoire during CM, we used high throughput CDR3 spectratyping and set up new methods and software tools to analyze data. We compared PBL and spleen repertoires of mice infected with Plasmodium berghei ANKA that developed CM (CM(+)) or not (CM(-)) to evidence modifications of the TCRB repertoire associated with neuropathology. Using distinct statistical multivariate methods, the PBL repertoires of CM(+) mice were found to be specifically altered. This alteration is partly due to recurrently expanded T cell clones. Strikingly, alteration of the PBL repertoire can be used to distinguish between CM(+) and CM(-). This study provides the first ex vivo demonstration of modifications of Talphabeta cell compartment during CM. Finally, our original approach for deciphering lymphocyte repertoires can be transposed to various pathological conditions.

A visual-quantitative analysis of fibroblastic stromagenesis in breast cancer progression

One fundamental difference between normal and transformed cells is the way they interact with their immediate environment. Exploring this difference is crucial for understanding the pathobiology of cancer progression. Benign epithelial tumors are constrained by a surrounding stroma consisting, among other cells, of fibroblasts embedded within fibrillar three-dimensional matrices. However, at a critical point in tumor progression, tumor cells become altered and overcome the barrier, inducing changes in the stroma, which promote, rather than impede, tumor progression. Inherited or acquired genetic aberrations affecting mammary gland epithelia are usually blamed for promoting neoplasia in individuals at "high risk" for breast cancer. However, in addition to these epithelial aberrations certain individuals possess permissive breast stroma. The occurrence of this permissive stroma results in a predisposition for cancer initiation or progression. Here we review stromagenic stages, experimental 3D systems, and discuss digital imaging analyses suitable for uncovering the mechanisms behind fibroblastic breast stromagenesis.

Human fibroblasts from normal and malignant breast tissue grown in vitro show a distinct senescence profile and telomerase activity

The telomerase activity and the senescence profile of cultured breast fibroblasts from normal human interstitial and malignant stromal tissue were studied in comparison with their proliferation and differentiation pattern. Fibroblasts were grown either in the presence or absence of a conditioned medium (CM) obtained from cultures of the oestrogen receptor-positive breast cancer MCF-7 cell line. At different passages (from the 2nd up to the 48th), fibroblasts were examined for the telomerase activity by the Telomerase Repeats Amplification Protocol (TRAP) assay, for proliferation profile by Ki-67 antigen expression, and the myofibroblast or smooth muscle cell-like differentiation pattern by immunofluorescence with monoclonal antibodies specific for smooth muscle markers. Serial passages of fibroblasts from normal or tumour breast reveal that the relationship between the levels of telomerase activity and phenotypic/proliferation profile changes with cell subcultivation in a different manner in the two cell populations. The fibroblasts from normal tissue completed 12 passages in a CM-independent way prior to senescence whereas fibroblasts from tumour stroma senescence were attained after 48 passages. These cells showed a marked decrease of telomerase activity, growth rate and smooth muscle alpha-actin expressing myofibroblasts after the 32nd passage. CM treatment of this fibroblast population induces a decline in the myofibroblast content, which precedes the changes in telomerase activity. Passaged fibroblasts from normal breast tissue can be converted to myofibroblasts upon CM treatment whereas those from tumour stroma were CM-insensitive. Taken together our data suggest that a heterogeneous fibroblast population with different life span is activated/recruited in the breast interstitium and poses the problem of a unique activation/recruitment of fibroblasts in neoplastic conditions.

Discriminant analysis to evaluate clustering of gene expression data

In this work we present a procedure that combines classical statistical methods to assess the confidence of gene clusters identified by hierarchical clustering of expression data. This approach was applied to a publicly released Drosophila metamorphosis data set [White et al., Science 286 (1999) 2179-2184]. We have been able to produce reliable classifications of gene groups and genes within the groups by applying unsupervised (cluster analysis), dimension reduction (principal component analysis) and supervised methods (linear discriminant analysis) in a sequential form. This procedure provides a means to select relevant information from microarray data, reducing the number of genes and clusters that require further biological analysis.

Identification of novel ethanol-sensitive genes by expression profiling

Chronic exposure to ethanol or other addicting drugs causes long-lasting, deleterious behavioral responses, such as tolerance, dependence, sensitization, and addiction. Changes in brain gene expression are thought to be a critical component of these behavioral adaptations. Our laboratory and others have utilized cultured neuronal cells as model systems for studying gene regulation by ethanol. Recently, the use of non-biased, high-throughput approaches to studying gene expression has allowed identification of gene regulation "patterns," rather than single genes responding to ethanol. This review will discuss how expression-profiling approaches can be used to identify functional changes occurring in neural cells with chronic exposure to ethanol.

Phenotypic and genetic alterations in mammary stroma: implications for tumour progression

In addition to the well documented role of cytokines in mediating tissue-level interactions, it is now clear that matrix macromolecules fulfil a complementary regulatory function. Data highlighted in the present review extend the repertoire of matrix signalling mechanisms, (1) introducing the concept of 'matrikines', these defined as proteinase-generated fragments of matrix macromolecules that display cryptic bioactivities not manifested by the native, full-length form of the molecule, and (2) indicating that a previously identified motogenic factor (migration stimulating factor [MSF]) produced by foetal and cancer patient fibroblasts is a genetically generated truncated isoform of fibronectin, which displays bioactivities cryptic in all previously identified fibronectin isoforms. These observations are discussed in the context of the contribution of a 'foetal-like' stroma to the progression of breast cancer.

Sage transcript profiles in cultured human fetal fibroblasts, WI-38

Serial analysis of gene expression (SAGE) allows us to analyze the profile of genes expressed in human lung fibroblast cell line WI-38. Manual sequencing of approximately 10,000 transcripts derived from WI-38 cells revealed 1025 genes expression among of which 431 tags matched GenBank entries, whereas 594 corresponded to novel previously undescribed transcripts. 7SL RNA (7L7) 5'-truncated pseudogene and the gene with no match (tag: TCCCTAGCT) were found the most abundant among all the analyzed genes (10.2 and 4.6% respectively). The expression pattern of the genes coding for cytoskeletal proteins, extracellular matrix components (ECM) and cytokines has been obtained. The results here demonstrate the unique gene expression pattern in human WI-38 fibroblast cell line and support a possibility of useful application of SAGE method for differential expression analysis.

Lumican and decorin are differentially expressed in human breast carcinoma

Previous studies have shown that lumican is expressed and increased in the stroma of breast tumours. Lumican expression has now been examined relative to other members of the small leucine-rich proteoglycan gene family in normal and neoplastic breast tissues, to begin to determine its role in breast tumour progression. Western blot study showed that lumican protein is highly abundant relative to decorin, while biglycan and fibromodulin are only detected occasionally in breast tissues (n=15 cases). Further analysis of lumican and decorin expression performed in matched normal and tumour tissues by in situ hybridization showed that both mRNAs were expressed by similar fibroblast-like cells adjacent to epithelium. However, lumican mRNA expression was significantly increased in tumours (n=34, p<0.0001), while decorin mRNA was decreased (p=0.0002) in neoplastic relative to adjacent normal stroma. This was accompanied by a significant increase in lumican protein (n=12, p=0.0122), but not decorin. Further evidence of altered lumican expression in breast cancer was manifested by discordance between lumican mRNA and protein localization in some regions of tumours but not in adjacent morphologically normal tissues. It is concluded that lumican is the most abundant of these proteoglycans in breast tumours and that lumican and decorin are inversely regulated in association with breast tumourigenesis.

Classification of human ovarian tumors using multivariate data analysis of polypeptide expression patterns

Large amounts of data on quantitative gene expression are generated by procedures such as 2-DE analysis of proteins or cDNA microarrays. Quantitative molecular variation may potentially be used for the development of methods for the classification of tumors. We used here the statistical concepts of principal components analysis (PCA) and partial least square analysis (PLS) in an attempt to type ovarian tumors. Using a set of 170 polypeptides, 22 tumors were used to establish a model ("learning set") for classification into 3 groups (benign/borderline/malignant). Eighteen tumors were then used to test the model. Six of 8 carcinomas and 3 of 4 borderline tumors were correctly classified. Two of 6 benign lesions were correctly classified, 3 were classified as borderline and 1 as carcinoma. We conclude that it may be possible to classify tumors according to their constitutive protein expression profile using multivariate analysis, thus making classification by artificial intelligence a future possibility.

Knowledge discovery in gene-expression-microarray data: mining the information output of the genome

A key aspect of the genomics revolution is the transformation of large amounts of biological information into an electronic format, leading to an information-based approach to biomedical problems. Large-scale RNA assays and gene-expression-microarray studies, in particular, represent the second wave of the genomics revolution, providing gene-expression data that complement gene-sequence data and help our understanding of the molecular basis of health and disease. They are being applied at several stages in the drug-development process and could ultimately have broad applications in disease diagnosis and patient prognosis.

From expressed sequence tags to 'epigenomics': an understanding of disease processes

Expressed sequence tags (ESTs) are at the forefront of technological change that is sweeping the biomedical research community. ESTs provide a high throughput means for identifying gene transcripts and monitoring complex gene expression patterns. EST-based technologies coupled with sophisticated computer analysis tools enable the informational content and output of the genome to be accessed and evaluated on a scale immensely larger than previously possible. EST-based technologies are being used to understand disease processes and to find better disease treatments, and will allow biology to move from single gene to multigene, or even more complex epigenetic, explanations for disease.