Studies of Transforming Growth Factor Alpha in Normal and Abnormal Growth

Home , Betacellulin, Epidermal growth factor, Epigen, Growth factor, IGFBP2, TGF alpha

Linköping University medical dissertations, No. 1025

Anna-Lotta Hallbeck M. D.

Department of Clinical and Experimental Medicine, Faculty of Health Sciences, Linköpings Universitet, SE-581 85 Linköping, Sweden

Linköping 2007

ISBN 978-91-85895-61-8 ISSN 0345-0082 Series: Linköping University medical dissertations, No. 1025

2 PAPERS

This thesis is based on studies presented in the following four papers, referred to in the text by their roman numerals.

I. Hallbeck, A-L., Walz, T.M. and Wasteson, Å. (2001). Interleukin-6 enhances transforming growth factor-alpha mRNA expression in macrophage-like human monocytoid (U-937-1) cells. Bioscience Reports; Vol. 21(3):325-339. II. Hallbeck, A-L., Lirvall, M., Briheim, K. and Wasteson, Å. Single physiologic dose UVB radiation increases EGFR expression in cultured normal melanocytes. Manu- script. III. Hallbeck, A-L., Walz, T.M., Briheim, K. and Wasteson, Å. (2005). TGF-alpha and ErbB2 production in synovial joint tissue: increased expression in arthritic joints. Scandinavian Journal of Rheumatology; 34(3):204-211. IV. Hallbeck, A-L., Walz, T.M., Wasteson, Å. and Lindström, A. Cooperation of endogenous HER-family members in synovial sarcoma cells: stimulation of HER- 2/ErbB2 is dependent on EGF-R activation. Submitted.

3 CONTENTS PAPERS ...... 3 CONTENTS ...... 4 ABSTRACT ...... 5 ABBREVIATIONS ...... 7 FOREWORD ...... 9 INTRODUCTION ...... 10 GENERAL ASPECTS ON GROWTH REGULATION ...... 10 INFLAMMATION AND REPAIR ...... 13 TUMORS AND METASTASIS ...... 16 BIOLOGICAL CLASSIFICATION OF GROWTH FACTORS ...... 17 THE HER-FAMILY OF LIGANDS AND RECEPTORS ...... 18 THE SYNOVIAL JOINT ...... 27 SYNOVIAL SARCOMAS ...... 36 AIMS ...... 38 COMMENTS ON MATERIALS AND METHODS ...... 40 BRIEF SUMMARY ...... 40 IN VITRO-MODELS ...... 41 PATIENT SAMPLES ...... 44 CELL GROWTH ...... 45 RNA-ANALYSIS METHODS ...... 46 IMMUNOFLUORESCENT LABELING AND FLOW CYTOMETRY ...... 49 IMMUNOHISTOCHEMISTRY ...... 50 PROTEIN ANALYSIS ...... 51 MICROSCOPIC EVALUATION ...... 53 STATISTICAL METHODS ...... 54 RESULTS AND DISCUSSION ...... 55 IL-6 EFFECT ON TGF-ALPHA MRNA IN MONOCYTOID CELLS ...... 55 EFFECT OF UVB-IRRADIATION ON HER-1 EXPRESSION IN MELANOCYTES .. 56 TGF-ALPHA MAY HAVE A ROLE IN SYNOVIAL JOINT HOMEOSTASIS ...... 59 TGF-ALPHA/HER-FAMILY IN 4SS SYNOVIAL SARCOMA CELLS ...... 62 CONCLUSIONS ...... 65 ACKNOWLEDGEMENTS ...... 66 REFERENCES ...... 68

4 ABSTRACT Regulation of growth is of fundamental importance for development of the organism and to maintain health. The induction of cell proliferation and matrix production are influenced by several different signaling systems, most importantly by growth factors. The human HER-family of growth factor ligands and receptors is one of the most studied and, at present, one of the most complex including 4 tyrosine kinase receptors and at least 11 different ligands cooperating in the transfer of signals. The HER-family growth responses are also influenced by other intercellular and extracellular signals, including matrix components, cytokines and hormones mediating e.g. inflammation. HER-1 (EGFR) is one of the best known and most extensively studied growth factor receptors. TGF-alpha is possibly the most potent HER-1 ligand and influences wound healing, epidermal maintenance, gastro-intestinal function, lactation, pulmonary function and more. Several studies have shown important regulatory functions for some inflammatory cytokines on TGF- alpha production in white blood cells. HER-1 is widespread in epithelial cells but also in mesenchymal cells such as fibroblasts, osteogenic and chondrogen- ic cells. Consequently, many tumors arising from these cell types express HER-family members and often show TGF-alpha and/or HER activation. In- deed, mammary cancer development has been shown when over-expressing both TGF-alpha and HER-2 in mouse mammary cells in vivo. In recent years the first HER-1 and HER-2 inhibitors have come into clinical practice for treatment of breast cancer, lung cancer and gastrointestinal cancers, sometimes with great success. However, more knowledge is needed concerning the inflammatory regulation of HER-family expression including where and how the ligands and receptors cooperate. Therefore we were interested in studying the role of TGF-alpha in normal and abnormal growth. First we showed that the acute inflammatory cytokine IL-6 regulates TGF-alpha expression in U-937-1 monocytoid cells. Se- condly, we detected a possible long-term enhancing influence of single-dose UVR on HER-1 expression in normal human melanocytes. We continued thirdly by revealing TGF-alpha production concomitant with HER-2 in normal human synovia and release of soluble TGF-alpha into the synovial fluid. Both TGF-alpha and HER-2 production were significantly increased in inflammatory joint conditions, e.g. RA. Fourthly, we demonstrated expression of TGF-alpha, HER-1 and HER-2 in synovial sarcoma cells in culture; the observed HER-2 phosphorylation was dependent on ligand induced HER-1 activation. 5 The presented results indicate that TGF-alpha expression can be enhanced by acute inflammatory cytokine IL-6, possibly contributing to growth- stimulatory effects assigned to IL-6 itself. The acute effects of UVR on melanocytes mediate upregulated steady-state expression of HER-1, constituting a potential target for locally produced TGF-alpha that may induce melanocyte proliferation. TGF-alpha and HER-2 seem to have a role in the maintenance of synovial joint tissues. Upregulation of TGF-alpha and HER-2 in inflammatory joint conditions, e.g. RA, represents a novel mechanism for synovial proliferation contributing to joint deterioration. TGF-alpha, HER-1 and HER-2 may have a role in synovial sarcoma proliferation; further investigation is needed to evaluate HER-family inhibitors as a possible treatment alternative in this type of cancer.

6 ABBREVIATIONS

4SS 4SS synovial sarcoma cell line Ag1523 normal human fibroblast cell line CNS central nervous system COOH-terminal carboxy-terminal, C-terminal DAB diaminobenzidine tetrahydrochloride Da, kDa, MDa Dalton, kilo-, mega-Dalton for molecular weight DNA deoxyribonucleic acid EDTA ethylene-diamine-tetra-acetic acid EGF epidermal growth factor EGFR epidermal growth factor receptor ELISA enzyme-linked immunosorbent assay FISH fluorescent in situ-hybridization GM-CSF granulocyte-macrophage colony stimulating factor HA hyaluronic acid HB-EGF heparin-binding EGF-like growth factor HER human epidermal growth factor-receptor HRG heregulin(s) IHC immunohistochemistry IL interleukin

IP3 inositol 1,4,5-tris-phosphate ISH in situ-hybridization LDL low-density lipoprotein MAPK mitogen-associated protein kinase MESH medical subject headings mRNA messenger ribonucleic acid NF-κB nuclear factor-kappa-beta (transcription factor) NRG neuregulin(s) N-terminal NH2-terminal/amino-terminal PMA phorbol-12-myristate-13-acetate, phorbolester PTB phosphotyrosine-binding (domain)

7 RA rheumatoid arthritis rER ribosomal endoplasmic reticulum RT-PCR reverse transcriptase-polymerase chain reaction SDS-PAGE sodium dodechyl sulfate polyacrylamide gelelectropho- resis SE standard error SEM standard error of the mean SF synovial fluid SH2 src-homology 2 (domain) SHP-1/ SH-PTP-1 src homology phosphatase-1 SKBR3 breast cancer cell line SLE systemic lupus erythematosis SP-A surfactant protein-A SP-D surfactant protein-D Src tyrosine-protein kinase proto-oncogene STAT signal transducers and activators of transcription SYT-SSX translocation event between the SYT gene on chromosome 18 and one of 3 SSX genes (SSX1, SSX2 and SSX4) on chromosome X TACE tumor necrosis factor-alpha converting enzyme TBS Tris-buffered saline TGF-alpha transforming growth factor-alpha TGF-beta transforming growth factor-beta U-937-1 human histiocytic lymphoma cell-line, monocytoid cells UDPGD uridine diphosphoglucose dehydrogenase UVA ultra-violet radiation type A, 320 to 400 nm UVB ultra-violet radiation type B, 280 to 320 nm UVC ultra-violet radiation type C, 200 to 280 nm UVR ultraviolet-radiation Vit-D3 1,α25-dihydroxycholecalciferol

8 FOREWORD The focus of this thesis will be transforming growth factor-alpha (TGF- alpha) and its effects on human epidermal growth factor-receptor (HER, EGFR) family of receptors, their detection and possible function in synovial joints and constituent cell types during homeostasis and inflammation, as well as results achieved in cultured cells used as in vitro-models for synovial tissue cell types. It also includes studies on effects of ultraviolet-radiation (UVR) on HER-1 regulation in melanocytes illustrating acute effects of cell damage and inflammation on the expression of this receptor. The presentation begins with a rather extensive introduction aimed as a review of the HER-family and TGF-alpha in the biological contexts, tissues and cell types relating to our aims and experimental settings. Some of the in- formation is not directly included in the final discussion or in the four papers presented, but places our results in a larger perspective. It will hopefully also function as a comprehensive and understandable reference material for the less initiated reader.

9 INTRODUCTION General aspects on growth regulation The adequate regulation of growth and cell division is of fundamental importance in all biological organisms. Many of the medical disorders, diseases and complications we try to cure or alleviate are due to inadequate growth; too much, too little or bad timing. In many cases the dysregulation of growth is caused by mutations, dele- tions, translocations or multiplications in the DNA, e.g. tumors and inherited disorders. The cause of dysregulation can be found on the inside (endogenous), e.g. mainly inherited mutations, or come from outside the individual (exogenous). Important mechanisms include: • mutational defects in the gene coding for a growth stimulating protein (ligand) and/or its receptor that disturb their normal functions by dis- rupting, decreasing, increasing or broadening the signal transduction pathway • defects in DNA binding proteins that regulate transcription of receptor and/or ligand that result in failed, decreased or increased production of receptor and/or ligand with consequent deficient downstream signaling • decreased production of direct growth inhibitors or increased production of direct growth promoting proteins that shift the balance in regulatory loops The following text will present the human context in which growth factors play their part in sickness and in health, mainly concentrated on the human epidermal growth factor-receptor (HER, EGFR) family and transforming growth factor-alpha (TGF-alpha), one of the HER-1 ligands. Examples are therefore chosen to enlighten the role of this group of proteins; however, the mechanisms described are in most cases generally applicable in the growth biology of cells and tissues.

Development There are genetic “guidelines” for the development of each organ and tissue type in an organism. Numerous cooperative pathways of proteins signal to their target cells to speed up or slow-down, or even die by apoptosis, in response to environmental change within or outside the organism. The embryo- logical development is extremely dependent on correct timing and tuning. There are many known examples of growth factors and/or their receptors causing developmental defects, a few exemplified below. However, many rela-

10 tions remain to be discovered since most mechanisms are not clearly unders- tood. There is also a redundancy in function exemplified by knock-out studies in mice where e.g. a single knock-out of TGF-alpha, amphiregulin and HER-1, respectively, goes almost unnoticed and the triple knock-out is compatible with survival with some intestinal and mammary abnormalities1-3. Cleft lip and/or cleft palate have been directly linked to an endogenous congenital dysregulation of growth involving transforming growth factor-alpha (TGF-alpha) and its receptor, HER-1. The details are not known, but are likely dependent on both “to little” and “bad timing” in conjunction with other dysregulations4. Children affected by thalidomide (Neurosedyn) are tragic examples of an exogenous substance causing dysregulation of developmental growth. They were born with varying defects all involving the development of peripheral limbs, resulting in rudimentary arms and legs. Recent hypotheses suggest that thalidomide interferes with DNA binding, probably through increased production of free radicals that decreases the binding of transcription factor NF- κB to its promoter region/s. This blocks the normal induction of certain growth factors, e.g. “too little” fibroblast growth factor-10, resulting in distur- bance of a growth factor-loop that would have promoted limb outgrowth 5. During pregnancy the hormonal network prepares the mammary glands for production of milk. One important change is the proliferation of prolac- tin-producing cells in the anterior part of the pituitary gland that will maintain milk production during the nursing period. Increased estrogen levels in the hypothalamus induce local production of TGF-alpha which is necessary for the multiplication of prolactinous cells, reviewed by Denef 6. The total weight of the pituitary increases due to the increase in cell number and this change remains throughout the life of a mono- or multiparous woman.

Homeostasis The genetic program has developed during millions of years to an increasingly higher complexity to create and maintain a balance between growth and survival versus decline and controlled cell death, e.g. apoptosis. The main purpose is to keep the fully grown organism in optimal shape and condition. This is an important part of the overall so called homeostasis, a term which in its widest sense also includes the regulation of nourishment, disposal of waste products, oxygen supply, water balance, possibility of propagation of the species etc. (fig.1).

Fig.1. Overview of mechanisms influencing growth homeostasis. In a healthy organism growth and loss counterbalances to equilibrium, small influences from “normal” damage are quickly counteracted and balance is maintained. The weights beneath represents dysre- gulatory influences that will shift the balance resulting in deteriorating health due to exag- gerated abnormal growth or waste and loss of tissue and functions.

The epidermis as a model for continuous renewal The human skin is constantly renewed to keep an optimal barrier pro- tecting the body from external threats like ultraviolet-radiation (UVR) from the sun, bacteria, fungi, viruses, dehydration, mechanical damage, heat, cold, chemical substances etc. The protective layer, the epidermis, is maintained through keratinocyte stem cells that divide (proliferate), differentiate (mature) and finally die by apoptosis (controlled cell death) on their path moving outwards to the outermost surface of the skin, the keratin layer. The keratin constitutes remnants of dead cells, easily visible a few days after slight sunburn when the number of dead keratinocytes increase due to the damaging effect of UVR 7-9. However, the dead cells are continuously replaced due to the effect of mainly HER-1-stimulating ligands (TGF-alpha, EGF, amphiregulin and epiregulin) on keratinocyte stem cells, reviewed by Jost et al. 10 (fig. 1). Other tissues where the respective epithelium is continuously regenerated are the intestine, the airways and the urinary tract. Continuous remode- ling and renewal also take place in non-epithelial tissues, e.g. skeletal bone

12 and the hematopoietic system in the bone marrow. These processes are all promoted by growth stimulatory substances, carefully counter-balanced by the modulating effects of anti-proliferative mechanisms (fig. 1).

Ultra-violet radiation (UVR) Sunlight is the greatest source of human UVR exposure, affecting humans as well as all other creatures and plants on this planet. The spectrum of UVR wavelengths that reach the earth’s surface are 10% UVB (280 to 320 nm) and 90% UVA (320 to 400 nm); the highly energetic UVC (200 to 280 nm) is almost completely blocked out by the ozone layer 9. Frequent low-dose UVB that accumulates to a large dose over time correlates to the development of non-melanoma skin cancers, while intermittent high-dose exposures and sunburns seem to increase the risk of developing melanoma 7,11. The tumori- genic effect of UVR is due to DNA damage, mainly mutagenic effects of reactive oxygen species generated by UVA or direct DNA damage by UVB. The UVR also causes immunosuppression and increases local production of growth factors. UVB is directly absorbed by the DNA where it causes characteristic UV mutations resulting from the incorrect repair of cyclobutane pyrimidine dimers and pyrimidine (6-4) pyrimidone photoproducts 7,12. The normal epidermal substitution rate accelerates by exposure to UVR as a direct effect of increased production of cytokines and growth factors in the epidermis. There are also UVR-induced effects on melanocytes, the cell type responsible for pigmentation through its production of melanin. These cells are intermingled with the keratinocytes and increase both their distribution and production of melanin in the epidermis to improve the protective shield against UVR 9. Also, melanocytes proliferate in response to growth factors induced by sunburn; the details are not known but may include TGF- alpha and HER-1. Treatment with anti-HER-1 in different types of cancer is frequently associated with cutaneous toxicities (45-100% of patients) which underline the importance of HER-1 signaling in epidermal homeostasis 13. Inflammation and repair Tissue repair and healing is an important assignment for the less differentiated cells in a tissue, prompted by damage and host defense in the form of white blood cells and tissue based cell members of the immune system. All of these cell types communicate by various intercellular signaling peptides and proteins including cytokines and growth factors. They are small proteins with the role of signaling to other cells (paracrine) or to the producing cell itself (autocrine) in order to induce host response, repair and to reconstitute homeostasis. To accomplish this, the appropriate receptors need to be expressed 13 by the target cell. Often a few specific receptors are continuously present on the cell and once triggered by ligands the ensuing effect will be upregulation of various other receptors and proteins important for host response and repair. Another important family of cell stimulating factors that may induce growth and proliferation are hormones, but these are beyond the scope of this thesis (fig. 2).

Fig. 2 Overview of signal transduction pathways and the influence of hormones and different intercellular signaling peptides and proteins through cell surface receptors. The different pathways are activated by different stimuli but blend into each other, modulating the cellular response. Picture courtesy of Wikipedia Commons.

When repairing tissue damage, a variety of cell signaling pathways can be activated. The selection depends on several factors, most importantly the extent of damage and if the damaged area is clean or also involves pathogens or foreign bodies. The inflammatory response and tissue repair are processes closely related, almost always performing parallel tasks achieved through in- termingling signaling pathways (fig. 1 and 2). Key players are various types of white blood cells and platelets performing specific tasks in the defense against 14 potential invaders, removing remnants of dead tissue and stimulating healthy cells in the area to proliferate and heal the damage. TGF-alpha has been detected in monocytes, macrophages and granulocytes as well as in wound fluid, and its production is increased by inflammatory cytokines 14-19. Macrophages are residents in most tissues, guarding the homeostasis and responding fast when stimulated by bacterial or viral antigens, cell debris and various cytokines. They are also, together with certain tissue specific cells, antigen presenting cells guiding the immune defense towards invaders, e.g. bacteria etc.

Epidermal healing When healing an external wound the epidermal skin cells follow essentially the same program as when replacing worn-out dead keratinocytes. How- ever, after acute damage this process is commenced by keratinocyte growth factor and other factors belonging to the fibroblast growth factor family after which the standard program is activated20-23. This exemplifies the induced growth process, where replacement was not a scheduled repetitive activity but needed to restore homeostasis. TGF-alpha and EGF have been shown in vitro to promote the radial growth of keratinocytes helping to cover a surface, e.g. an epidermal wound 24. TGF-alpha was more effective than EGF in stimulating epidermal regeneration after burns 25. Many cell types are too highly differentiated to multiply, like nerve cells and cardiomyocytes residing in the G0 cell phase (fig. 1). The tissue response to damage will mainly consist of fibrosis, e.g. healing by proliferation of connective tissue cells that also are induced to produce matrix components (scar tissue). In healing of cutaneous wounds it is important to attenuate the cell division and, most importantly, terminate the matrix production in time to avoid damage caused by excessive scarring, e.g. keloids, or movement limiting scarring, e.g. after deep burns. Recent investigations into keloidal fibroblasts have shown over expression of TGF-beta1 as an important, but not independent, factor for abnormal scarring 26. The process of keloid formation is multi factorial and also includes over expression of vascular endothelial growth factor, TGF-beta2, platelet-derived growth factor receptor alpha, down regulation of apoptosis related genes and inherited predisposal. Data indicate that keloidal fibroblasts are unable to respond to, and/or activate, negative feedback signals needed to terminate the wound healing process 27. For healing to occur properly there is a definite need for vascular blood supply. A decreased blood flow in a fully developed organ can itself cause damage to its target cells, or the whole organ, depending on the lack of oxygen. The relative anoxia and accumulation of waste metabolites will generate

15 a tissue response, including inflammation, due to damage and necrosis (death) of cells. The response involves production of angiogenetic growth factors such as vascular endothelial growth factor to stimulate sprouting of new vessels, angiogenesis or neo-angiogenesis. This process will, in time, restitute the supply of oxygen and nourishment as well as clearing of toxic waste substances. HER-1 has been shown to mediate angiogenic responses, especially when stimulated by TGF-alpha 28,29. Furthermore, HER-2 seems to promote angiogenesis by increasing vascular endothelial growth factor receptor inde- pendently of hypoxia inducible factor 1 (HIF-1) and this effect can be blocked by the HER-2 inhibitor trastuzumab (Herceptin®) 30-32. Tumors and metastasis The process of tumor development is obviously, overall, an imbalance in the growth homeostasis due to genetic reprogramming and/or damage. The normal growth control mechanisms are overcome by genetic mutations tip- ping the balance towards growth, and limiting or completely blocking apoptotic mechanisms (controlled cell death) (fig. 1). Tumors ubiquitously exhibit mutations that inactivate the p53-family of proteins, resulting in proliferation of cells due to continuation of the cell cycle instead of differentiation or apoptotic clearance of damaged cells 33. The genetic changes also result in de- differentiation or unresponsiveness to differentiating signals resulting in an immature, more stem-cell like cell, with increased ability to proliferate. Tumor cells can often take advantage of genetic programs with the purpose of maintaining homeostasis, e.g. like the angiogenetic response to hypoxia or production of growth factors for maintenance. This ability is of special importance in the metastatic process. Normal connective tissue cells intermingled with the tumor cells show DNA alterations and can be induced to feed the tumor with growth factors promoting its survival at the new location 34,35. The hypothesis of self supporting tumor cells producing polypeptide growth factors for autocrine stimulation was formulated on the basis of disco- vering transforming growth factors (alpha and beta) able to promote transformation of fibroblasts as well as various human tumor cell lines 36-39. There are many examples where growth factors and/or their respective receptors are over expressed in tumors or pre malignant cells, as well as related models where over expression has been shown to increase the ligand promoted proliferation and promoting tumor development, e.g. pancreatic duct cells, colonic mucosa, hepatocellular cancer, mammary cancer 40-44. Also, receptors can be constitutively active, e.g. feeding the signal to proliferate into the tumor cell

16 machinery without activating ligand. Often tumor cells exhibit continuous activity in intracellular signaling peptides like Ras, Erk and mitogen activated protein kinases (MAPK) without an obvious link to receptor tyrosine kinases (RTKs). Growth factors that activate specific RTKs then augment the proliferative and anti apoptotic signal by feeding into the same intracellular signaling pathways (fig. 2). Examples of cooperation between TGF-alpha and neu/ErbB2/HER-2 in tumor development have elegantly been shown in female transgenic mice where over expression of each factor alone induced focal mammary cancer after long latency; this process was significantly faster when both TGF-alpha and HER-2 were over expressed in the same animal. Furthermore, evidence suggests that the process was independent of HER-1/HER-2 dimerization 45. Biological classification of growth factors

Intercellular Signaling Peptides and Proteins The intercellular signaling involves hundreds of signaling molecules of which a certain blend will work in concert on a cell to effectuate a specific biological response. Each factor, peptide and protein has been structurally, genetically and functionally classified, to our present knowledge, and thereby categorized and placed in a biological hierarchy by the US National Library of Medicine. This searchable bibliographic system is published in the Medical Subject Headings (MESH) database that also may function as a thesaurus with strict definitions explaining the contents of each heading, subheading and search term. Both cytokines and growth factors are classified as ‘Intercellular Signal- ing Peptides and Proteins’, by the definition: ‘Regulatory proteins and peptides that are signaling molecules involved in the process of paracrine com- munication. They are generally considered factors that are expressed by one cell and are responded to by receptors on another nearby cell. They are distinguished from hormones in that their actions are local rather than distal.’46. (According to the author’s opinion, this definition should, for completeness, also include autocrine signaling.) Intercellular Signaling Peptides and Proteins include growth factors, cytokines, semaphorins, parathyroid hormone related protein, kinins and endo- thelins. The group also contains Wnt proteins, hedgehog proteins and eph- rins, factors mainly important during embryonic and fetal development but recently also found in various tumors (fig. 2). In the MESH database growth factors are presented as a distinct group of intercellular signaling peptides and

17 proteins, but many authors include them as a subgroup of cytokines in the scientific literature.

Cytokines The definition of cytokines is: “Non-antibody proteins, secreted by inflammatory leukocytes and some non-leukocytic cells that act as intercellular mediators. They differ from classical hormones in that they are produced by a number of tissue or cell types rather than by specialized glands. They generally act locally in a paracrine or autocrine rather than endocrine manner.”47. Cytokines include several groups of signaling proteins with broad functions, also including growth: interleukins and related factors, hematopoietic cell growth factors, chemokines, hepatocyte growth factor, transforming growth factor-beta family, interferons, lymphokines, monokines, tumor necrosis factors, leukemia inhibitory factor, oncostatin M and osteopontin 47.

Growth factors The functionally related group ‘growth factors’ include growth promoting peptides and proteins that bind to receptor tyrosine kinases and with the main purpose of stimulating cell survival, cell division, tissue growth and production of matrix components. Consequently, most growth factors have a wide and shifting distribution in various cell types, leukocytes being an important but minor source. From a structural point of view growth factors may be seen as members of distinct families: angiogenic proteins including vascular endothelial growth factors, endothelial growth factors, fibroblast growth factors including keratinocyte growth factor, nerve growth factors, platelet-derived growth factors, somatomedins (insulin-like growth factors, IGF), transforming growth factor beta family (TGF-beta) and the HER-family including TGF-alpha, EGF and the neuregulins/heregulins. The HER-family of ligands and receptors

General aspects The prototype for all growth factors was named after its ability to stimulate proliferation of skin epithelial cells, hence, epidermal growth factor (EGF). The discovery of EGF was published in 1962 as the ability of mouse salivary gland extract to promote precocious tooth eruption and eyelid open- ing in newborn pups; the active component of the extract was isolated and shown to be a 53 amino acid peptide 48. Since then a whole family of ligands with functional and structural similarities to EGF has been characterized and 18 now includes seven different human ligands that bind to the EGF-receptor (ErbB1, HER-1). These are: EGF, TGF-alpha, amphiregulin, heparin-binding EGF-like growth factor (HB-EGF), betacellulin, epiregulin and epigen (fig. 3). Additionally, there are several different EGF-like viral proteins known to bind to and activate HER-1; these are not further discussed in this thesis 49. A second group of related ligands are heregulins/neuregulins; they do not bind to HER-1 but to other members of the complementary human epidermal growth factor receptor family (HER-family) of totally four receptors, further described below (fig. 5). Besides being efficient HER-1 ligands, HB- EGF and betacellulin also bind to HER-4. Epiregulin is pan-specific, e.g. binds to HER-1, HER-3 and HER-4, 50. The HER-1 ligands all share a consensus sequence known as the EGF motif which consists of six conserved cysteine residues forming three disulfide bonds resulting in a globular structure crucial for HER-1 affinity. They are all produced as transmembrane proteins that are inserted into the membrane before the active soluble ligands are cleaved off by proteases located at the cell surface (fig. 3). To a varying extent, the HER-1 ligands can also function as juxtacrine uncleaved ligands, in some cases supported by accessory molecules like the tetraspanin CD9 forming surface complex with HB-EGF. A special feature of amphiregulin and HB-EGF are their amino terminal heparin binding domains that, in the case of HB-EGF, have been shown to increase its mitogenic activity when associating with heparan sulfate proteog- lycans (fig. 3). Furthermore, activation of HER-4 by HB-EGF in uterine tissue is dependent on CD44 interaction. CD44v3 contains a heparan sulphate proteoglycan binding site and has been shown to interact with HB-EGF, reviewed in Harris et. al., 2003 50. In polarized epithelial cells in vitro, e.g. cells derived from epithelium in the colonic mucosa, both amphiregulin and TGF-alpha have been shown to localize to the basolateral surface where the majority of HER-1s can be found. Furthermore, the proteolytic activity releasing mature soluble TGF-alpha and amphiregulin is also more active here, resulting in difficulties when performing immunodetection. EGF is located at both surfaces, but more easily immunodetected at the apical surface where the proteolytic activity is low 50. Knock-out experiments have been performed for EGF, TGF-alpha and amphiregulin without producing any deleterious changes in phenotype. Even cross-breeded triple knock-outs were viable and essentially healthy besides abnormalities in mammary development and the small intestine 1-3. There seems to be a pronounced redundancy and overlapping functions for HER-1 ligands. 19

Fig. 3 The human HER-1 family of ligands: epidermal growth factor (EGF), TGF-alpha (TGF−alpha), amphiregulin (AR), heparin-binding EGF-like growth factor (HB-EGF), betacellulin (BTC), epiregulin (EPR) and epigen. All of these ligands are produced as membrane inserted pro-factors; black boxes depict the transmembrane domains. Filled ovals delineate the EGF-domains while the unfilled ovals show heparin sulfate binding domains. Arrows indicate cleavage sites where cell surface protease activities release soluble ligands. Adapted from Harris et al. 50.

Transforming growth factor alpha Human transforming growth factor alpha (TGF-alpha) was originally discovered in medium collected from fibroblasts transformed by simian sarcoma virus. It was named after its ability to transform fibroblasts in a reversible manner. It was later revealed that this activity was in part conferred by a distinct molecule designated transforming growth factor beta (TGF-beta). These two factors are only related by name and original discovery. TGF-beta is structurally different, activates a distinct family of receptors and confers mostly effects distinct from TGF-alpha/HER-1 that are sometimes also counteracting as reviewed by Lee et. al. 49. TGF-alpha has a widespread distribution in normal cells and tissues: skin, endocrine organs, breast, urinary organs, respirato- ry system, hematopoietic cells, CNS, bone and muscle. This is also reflected in its involvement and presence in a wide range of neoplasms, for comprehensive reviews see Lee et. al. and Junier 49, 51.

20 The TGF-alpha gene encompasses 80 kilobases of genomic DNA and has been mapped to chromosome 2p13. It consists of six exons (sizes: exon 1, 40 base pairs; exon 2, 57 base pairs; exon 3, 118 base pairs; exon 4, 150 base pairs; exon 5, 110 base pairs) transcribed and spliced to a 4.5 kb mRNA 4. The expression of the gene is under the influence of several Sp1-like binding sites, two neighboring transcriptional promoters where the distal part is activated by Ras-induced AP-2 and the more proximal by TEF-1, p53 binding to a portion of the proximal promoter, as well as an estrogen responsive element (ERE) 52-56. The human TGF-alpha protein exists in two active forms; the full length 159 amino acid pro-TGF-alpha produced as a transmembrane cleavable protein, and the 50 amino acid soluble s-TGF-alpha which is released into the extracellular compartment by cleavage of transmembrane pro-TGF-alpha (fig. 4). The membrane associated release of s-TGF-alpha is effected by tumor necrosis factor alpha converting enzyme (TACE), also termed metalloproteinase 17 (ADAM17), acting between the extracellular alanine-valine amino acids close to the cell membrane releasing an HER-1 binding 6 kDa molecule 50 (fig. 4). Larger forms of s-TGF-alpha have been found, containing N- and O- linked glycosylation 57. The enzymatic release of s-TGF-alpha is regulated, sensitive to increased intracellular Ca2+ levels and induced by activation of protein kinase C 49, 50. The vast amount of functional activities reported for TGF-alpha is attri- buted to its binding to HER-1 and consecutive HER-1/HER-family transacti- vation. Although both the soluble and full length TGF-alpha is known to activate HER-1, TACE was required for the activation of HER-1 by TGF-alpha in mammary tumors 58. The cytoplasmic tail of transmembrane full length TGF-alpha is important for the intracellular routing, containing two different binding sites for proteins that, respectively, mediate efficient surface delivery and basolateral sorting 50. Furthermore, transmembrane pro-TGF-alpha has been shown to associate with two proteins, p86 and p106, in Chinese ham- ster ovary cells transfected with the complete TGF-alfa gene 59. Efforts to completely identify these proteins were not successful. However, the 106 kDa protein was exposed at the cell surface and was tyrosine-phosphorylated as determined by immunoblotting of cross-linked pro-TGF-alpha immunoprecipitates. Moreover, when the complete TGF-alpha/p106/p86 kinase complex was cross-linked and then immunoprecipitated it was shown to also contain phosphorylation on serine and threonine substrates. The kinase activity was dependent on the association of intracellular p86 to the carboxy-terminal

21 (COOH-terminal) 6 amino acids, requiring the presence of two palmitoylated cysteines, at position 153 and 154 at the innermost region of TGF-alpha 60 (fig. 4).

Fig. 4 Transforming growth factor-alpha depicted as pro-TGF-alpha inserted in the cell membrane, also illustrating the cleaved-off N-terminal sequence. Arrows indicate cleavage- sites where s-TGF-alpha is released through the activity of TACE/ADAM-17. Filled circles show cysteines, bonded by di-sulfides in the active TGF-alpha protein. The most C- terminal cysteines are 153 and 154, shown to be palmitoylated and possibly mediating intracellular signals from pro-TGF-alpha under circumstances still unknown. For references, see text. As previously described in this thesis, it seems clear that TGF-alpha and EGF share both receptor and effects, but also that their respective potency varies and probably also the ability to induce secondary dimerization 28, 61-63. Some results indicate that TGF-alpha is a more potent HER-1-ligand than

22 EGF, e.g. in stimulating bone resorption and in inhibiting bone formation, in angiogenesis and in epidermal wound healing 28, 64-65.

The HER family of tyrosine kinase receptors The human epidermal growth factor-receptor (HER) family includes HER-1, HER-2, HER-3 and HER-4 as we know it today. They are all tyrosine kinase receptors with an overall sequence homology of 40-50% and a common structure of 4 extra-cellular domains (I-IV) including 2 ligand binding domains (I and III), one trans-membrane domain, one juxta-membrane domain, one kinase domain and one C-terminal domain (fig. 5). HER-1 is synthesized as a 1210-amino acid precursor which is inserted into the cell membrane as an 1186-residue protein after cleavage of the N- terminal sequence. The mature receptor monomer has a molecular weight of 170 kDa which includes extensive N-linked glycosylation required for translocation and function of the protein and constituting more than 20% of the molecular mass 66. Both HER-1 and HER-4 behave as fully functional signaling membrane tyrosine-kinase receptors activated by ligand binding. HER-2 is only activated by heterodimerization transferring the tyrosine-kinase activity from an active partner; no HER-2 binding ligand has yet been found. The trans-activation of HER-2 is probably an important regulatory step in cellular signal processing following ligand activation of HER-1, HER-3 and HER-4. HER-3 has no in- trinsic kinase activity, but is capable of modulating or transferring signal activity from its preferred heterodimerization partners, HER-1, HER-2 and HER-4 67-69 (fig. 5). The intracellular signaling pathways of HER-1 have been extensively studied and HER-1 has been used as a prototype for signaling through growth factor receptor tyrosine kinases. Recently, crystal structures of HER-1 with and without EGF or TGF-alpha have revealed that the HER-1 ligands are located at opposite sides of the receptor dimer which is not the case among the other 18 small subgroups of related receptor tyrosine kinases. Upon ligand binding the receptor molecule undergoes a conformational change due to bridging of regions II to I by the ligand. The receptor partly unfolds, thereby unmasking a beta-hairpin loop that before ligand binding was hidden inside the monomer. -

Fig. 5 The HER-family of receptors, their respective ligand specificities and possible dimer formation. The ligand binding preferences are delineated by arrows (upper part). Activa- tion of one specific receptor may result in hetero-dimerization with a receptor monomer not activated by ligand; possible hetero-dimers are indicated by arrows below the figure. Cysteine clusters involved in ligand binding are shown as black ovals and trans-membrane domains as black boxes. Intracellular tyrosine kinase domains are depicted as grey boxes. For references, see text.

This “dimerization loop” can be found in all HER-variants, is situated in region II and redirected outwards on opposite sides of the ligand where it medi ates receptor-receptor interaction. The orphan HER-2/HER-2 receptor constitutively presents its dimerization loop in the active conformation, without ligand binding; thereby it is constantly prepared for hetero-dimerization 69 (fig. 5).

24 The formation of a ligand bound HER-1 receptor dimer is a require- ment for receptor activation, as for other receptor tyrosine kinases. It greatly increases the enzymatic activity of the intracellular tyrosine kinase domain of the receptor resulting in phosphorylation on five C-terminal tyrosine residues. They serve as docking sites for intracellular signaling molecules containing so called Src homology 2 (SH2) or phospho-tyrosine-binding (PTB) domains. Other proteins are activated by release from HER-1 upon ligand binding, resulting in activation or translocation to other parts of the cell, i.e. zinc-binding protein ZPR-1 and Signal Transducers and Activators of Transcription protein (STAT) transcription factors. The C-terminal contains phosphorylation sites that vary between hetero-dimer partners and therefore results in recruitment of different signaling proteins depending on which heterodimer is formed. This greatly increases the diversity of signals originating from a set of HER-1 homo- and heterodimers. This extensive intracellular cross-talk results in simultaneous activation of several, well-known and important pathways: the Ras/MAPK/ERK pathway mediating growth and proliferative signals; proteins involved in cytoske- letal reorganization (FAK, dynamin); MEKK1 linking to JNK pathway and cadherin, all involved in cell-cell adhesion; src-family of cytosolic tyrosine kinases mediating proliferation and transformation; direct STAT activation independent of JAK-kinase linking to cytokine pathways; phospholipid metabolism through phospholipase Cγ, phosphatidylinositol-3-kinase (PI3-K), and phospholipase D, generating IP3, phophatidylinositol-3-phosphate (PIP3) and a raise in intracellular Ca2+-levels contributing to proliferation, survival, adhesion and migration (fig. 2). The activated HER-1 migrates from the cholesterol-rich caveolae/raft compartment of the cell membrane to clathrin-coated pits that are internalized. The rate of internalization and whether the receptor complex is broken down in the lysosomes or recycled as monomers after release of ligand deter- mines the strength and duration of the intracellular signals. Hetero- dimerization with HER-2 reduces the rate of HER-1 degradation and HER-2 is also the preferred dimer partner in cells expressing both HER-1 and HER-2 66. Studies concerning the nature of HER-family signaling through tyrosine residues have shown surprising differences concerning binding affinity of SH2- and PTB-domains to phosphorylated tyrosines on the respective receptors. HER-1 and HER-3 revealed only two highly promiscuous high affinity binding sites each that may serve as multifunctional docking sites for various SH2- and PTB-containing proteins while HER-2 exhibited many such sites. At

25 low concentrations of SH2- and PTB-proteins the majority of phosphotyrosines on HER-1 and HER-3 were more selective and presumably more specialized. However, under circumstances allowing low affinity binding, i.e. increasing the number of available receptors or ligands, many of the HER-1 and HER-2 phosphotyrosines turned more collective. This is one possible explana- tion to the oncogenic potential of increased expression of HER-1 and HER-2, respectively and cooperatively. Furthermore, HER-2 activation through either HER-1 or HER-3 further expands the range of intracellular signaling proteins that are recruited through the respective ligand activated receptor 70. The HER-1 is activated by ligand binding, the resulting dimerization of two monomers leading to phosphorylation of various tyrosine-residues in the carboxy-terminal domain. The affinity of EGF to the dimerized HER-1 is approximately 100 times greater than to an HER-1 monomer 71. Tyrosine- phosphorylated growth factor receptors are rapidly dephosphorylated by specific enzymes, so called protein-tyrosine phosphatases. One of these proteins, Src homology phosphatase-1 (SHP-1, SH-PTP-1) binds rapidly to the phosphorylated HER-1 tyrosine pY1173, dephosphorylates it and hence attenuates the MAPK stimulation that is transmitted by pY1173 72. In vitro experiments on normal human dermal fibroblasts have shown that HER-1 induced proliferation is decreased by cellular aging, i.e. after a certain number of cell divi- sions HER-1 signaling is gradually attenuated. This is in part due to the increased activity of SHP-1 creating an increasingly hampered MAPK activation. Hence the proliferation rate declines with time and number of passages 73. Down-regulation and degradation of HER-family members, as well as other activated receptors (hepatocyte growth factor receptor, platelet-derived growth factor receptor, colony stimulating factor-1 receptor), is dependent on growth factor stimulated ubiquitination of the receptor. This is mediated by binding of Cbl, an E3 ubiquitin ligase, to specific phosphorylated tyrosines on the respective receptor. Ubiquitinylated receptor-Cbl complexes are endocy- tosed by clathrin-coated pits leading to stepwise intracellular transporting and processing towards final degradation in the lysosomes. Binding of Cbl to activated HER-2 is prevented by heterodimerization with other HER-family members while binding of antibody, e.g. trastuzumab, increases the Cbl-HER- 2 association. Furthermore, it has been shown in vitro that over expression of TGF-alpha in human breast cancer cells (SKBR3) reduces the rate of HER-2 internalization and degradation 74.

26 Heregulins/Neuregulins Heregulins (HRGs), also called neuregulins (NRGs), constitute a family of low-affinity ligands binding to HER-3 and HER-4. Heregulins have been identified in many isoforms, all sharing a three-dimensional structure designated HRG egf domain, alpha or beta. Although the amino acid sequence of HRG egf differs from that of EGF itself, there are striking similarities concerning the three-dimensional structure conferring affinity to their specific receptors. Heregulin 1 and 2 function as HER-3 and HER-4 ligands while heregulin 3 and 4 have affinity for HER-4 alone (fig. 5). The presence of HER-2 concomitant with HER-3 or HER-4 confers formation of a heteromeric receptor complex with high affinity for heregulins 67, 68, 75-77. The synovial joint Synovial joints are functional units with the primary role of movement. The defined parts of these units consist of several diverse tissue types, all of mesenchymal origin but each highly differentiated to serve their respective specific purposes. The bones articulating in a synovial joint are held together by ligaments and the joint capsule; the bones meet and articulate at their respective cartilage covered endings. The minute space between the cartilage surfaces is occupied by a thin film of synovial fluid (SF) serving to lubricate the joint, eliminate friction during movement, and working in synergy with the cartilage to create a “liquid cushion” that protects the skeleton from the forces of weight, movement and gravitation. Since the hyaline cartilage at joint surfaces has no vascular supply it is extremely dependent on the SF for nourishment and maintenance. The internal surface of the joint capsule and ligaments is covered by the synovium which produces the SF, but also has other functions that contribute to the maintenance of a healthy synovial joint. The basic knowledge on synovial joint tissues and rheumatoid arthritis presented below can, where not specifically cited, be found in the comprehensive standard reference literature 78.

The synovium The synovial lining layer, lamina synovialis intima, is normally only a few cells deep, 2–5 layers, sometimes with patches where the lining layer is absent. The subsynovium, lamina synovialis subintima, is a supportive layer that varies in composition in different regions of the joint. It consists of loose areolar connective tissue, adipose tissue, or both, as well as dense fibrous connective tissue when covering a ligament or the joint capsule.

27 In a human knee the synovia is approximately up to 50 μm thick; however, its thickness increases greatly during inflammatory synovitis as in rheumatoid arthritis (RA). There is no basement membrane delimiting the synovia from the underlying tissue, as can be found beneath epithelial layers. Never- theless, it can easily be distinguished from the underlying connective tissue because of the decreased cellularity and the change in the extra-cellular matrix composition that occur 20 – 50 μm from the joint cavity.

The synovial lining cells Suggestions have been made that the specific properties of the synovial lining cells might be induced by the shear stresses affecting these cells by the more or less continuous movement of the adjacent SF. Indeed, synovial lining layers similar to ´synovium proper´ are found in bursal lining, teno- synovium, the regenerated synovium after synovectomy as well as in artificial articulations – either inflicted by badly healed traumatic fractures or delibe- rately produced in experimental models 79. Scanning electron microscopy of the synovial surface in rabbit knee reveals a discontinuous layer of synovial cells covering only 70 - 80 % of the underlying interstitial matrix. The synovial cells are highly active with cellular processes covering the surface of the extracellular matrix, as well as extending into it. Notably, the synoviocytes are separated from their neighboring synovial cells by spaces of extracellular matrix 1 – 2 μm wide. There are also numerous vesicular protrusions and lamellipodia extending into the SF.

Synoviocyte type A Type A synoviocytes are bone marrow-derived cells of the monocytic li- neage that are believed to reach the joint during embryonic development. In RA, as well as in other diseases causing inflammatory synovitis, additional monocytes are attracted to the synovium by a complex mixture of cytokine and chemokine influences. There are differences in phenotype between monocytes/macrophages in general and the type A synoviocyte, but the physiological and molecular basis of this specialization remains largely unknown. The type A cell is a very competent phagocyte, capable of engulfing macromolecules such as HA and ferritin as well as foreign material encountered in the SF. When studying the ultra-structure of these cells they were found to extend phagocytic lamellipodia and to contain numerous vesicles, as would an activated macrophage during phagocytosis. The phenotypic specifications of the type A cells are several and are listed below in comparison to type B synoviocytes, monocytes and macrophages (table 1).

28 Synoviocyte type B The second cell type found in the synovial lining layer is the fibroblast- like type B synoviocyte, often designated the ‘proper’ synoviocyte. Although having functions and morphological characteristics of a fibroblast, it is highly specialized with a phenotype distinct from the sub-synovial fibroblast (table 1).

Phenotype Type A Macro- Mono- Type B Fibro- SF Ref:s cells phage cyte cells blasts

Phagocytic Highly Highly When Yes No activated Phagocytic Yes Yes When No No lamellipodia activated Abundant Yes Yes Yes/No No No Vesicles Lamellar Yes No No Yes No Yes 80, 81 bodies SP-A No? Yes No Yes 80, 81 SP-D Yes? Yes 82, 83 CD68 High High Low Low Low Non-specific Yes No 79 esterase Matrix ? No No Yes Yes secretion Abundant Yes rER UDPGD Yes (4x) Yes (1x) 79 Prolyl 4- No Yes Yes 84 hydroxylase Hyaluronan Yes synthase CD44 Yes Yes VCAM-1, Yes No? 79 CD105 Table 1. Phenotypic characteristics of cell types present in synovial joint tissues; also listing related findings in synovial fluid (SF).

Especially the presence of lamellar bodies, containing bilayers of saturated phospholipids and an electron-dense core of surfactant proteins, distinguish type B synoviocytes from other types of fibroblasts 81,85,86.

The synovial fluid The composition of synovial fluid (SF) reflects the demands of nourishment, lubrication and transport of waste products from the joint cartilage. The bulk fluid is filtrated through the sub-synovium, in and out of vascular and lymphatic vessels, and essentially constitutes extracellular fluid with certain molecules added from the cells lining the joint space. The synoviocytes add hyaluronic acid (hyaluronan, HA) by active secretion into the SF. The HA content is normally 2-3 mg/l SF 87 and increases the viscosity of the fluid; it also “anchors” a semi-fluid proteoglycan layer to CD44 on the surface of synovial cells and chondrocytes. One important feature among the specific components in SF, is the relatively high level of saturated surface active phospholipids produced by the type B synoviocytes, accumu- lated in lamellar bodies and secreted by exocytosis 80. The presence of surfactant proteins, SP-A and SP-D, facilitate the lamination of articular surface by phospholipids resulting in an almost friction-less movement of the joint 88 89. There are findings from other organs, especially lungs, indicating that surfactant proteins also have immune regulatory functions. Specifically, they function as so called opsonins, e.g. labeling pathogens to facilitate phagocytosis 90.

Nutrition and transport – vascular system To supply the highly active lining cells and the joint cartilage, the SF is filled with nutrients, water and electrolytes supplied through an abundant capillary network situated in the synovial matrix below the lining surface. Numerous small post-capillary venules extend into the synovial lining layer to drain waste metabolites and electrolytes away from the joint fluid and the lining cells. The arterioles, larger venules and veins do not extend into the lining layer and about half of the capillaries are fenestrated on the side of the vessel facing the lining layer and joint cavity. Capillary fenestrations are extremely thin parts of the endothelial cells, in fact only membrane delimited, allowing rapid transfer of water and small molecules, e.g. nutrients, electrolytes and non-protein bound drugs. The fenestrae are highly non-permissive to plasma proteins ensuring no accidental leakage of functional proteins into the SF. Diffusion of plasma components would disturb the delicate liquid homeostasis in the joint fluid that is needed for friction-free movement and the shock absorbing function of the cartilage. 30 The subsynovium contains a network of fine lymphatic vessels that drain excess fluid from the joint cavity maintaining a subathmospheric pressure in the joint compartment. The lymphatics are also the only passage out of the joint for macromolecules not small enough to diffuse into the synovial capillaries. Alternatively, the macromolecules are phagocytosed by the type A-cells or macrophages in the synovium. Since the lymphatics are separated from the SF by the whole synovia, the cellularity and the extracellular matrix composition of the synovium influence the ease by which water and macromolecules reach the lymphatic system.

Surfactant proteins As mentioned above, type B synoviocytes can be distinguished from the subintimal fibroblasts, as well as from several other types of fibroblasts, by the presence of lamellar bodies containing bilayers of saturated phospholipids and an electron-dense core of surfactant proteins 81,85. Phospholipids bilayers released into the SF help to minimize friction during joint movement 88,89,91. Surfactant proteins are collagen-like lectins and are known to have immunoregulatory functions in the lung, e.g. increased phagocytosis of various pathogens and regulation of immune cell function; these effects have not yet been investigated in synovial joints 90,92. Both surfactant protein A (SP-A) and D (SP-D) have been detected in SF, in animal models as well as in humans 83,85. There are indications that the concentrations of SP-A, SP-D and phospholipids are increased in SF during rheumatoid arthritis 82. EGF has been shown to stimulate SP-A synthesis in human fetal explants 93. Furthermore, antisense inhibition of HER-1 was shown to effectively decrease both mRNA and protein expression of SP-A in human fetal lung tissue 94. The possible link between TGF-alpha/HER-1 and SP-A in the synovial compartment awaits further investigation.

CD44 isoforms CD44 is the principal cell surface receptor for hyaluronic acid (HA) and is normally widely distributed in different cell types and tissues of the human body 95, for comprehensive reviews on CD44 see Naor et al. 96,97. It is a cell surface glycoprotein generated in different isoforms made up of two constant regions of five and four exons separated by a selection of 9 variant exons in human (10 variant exons in mice). The standard CD44, designated CD44s, in human consists of 341 a.a. residues generated by direct splicing of constant exon 5 to exon 16, leaving out all variant exons in between. Although alternative splicing of the CD44 gene transcript could produce hundreds of different CD44 isoforms, as yet only a few dozen have been identified in vivo.

31 The HA receptor function of CD44 was initially discovered as a lymphocyte aggregation and homing activity 98. Later, CD44 was recognized as a receptor activity in several different systems resulting in a diverse nomenclature related to the various binding capacities that was discovered; Pgp-1: a poly- morphic integral membrane glycoprotein; ECMRIII: playing a role in matrix adhesion; Hermes antigen: contributing to lymphocyte activation and lymph node homing and RHAMM: the receptor for hyaluronic acid-mediated movement. CD44 is now known to be involved in many vital cell activities that depend on cell-to-cell contact or interaction between cells and extracellular matrix, e.g. presentation of cytokines and enzymes to surrounding tissues and cell surface signal transmission into the cell. The diversity of target tissues and functions is partly dependent on the generation of variant isoforms and partly on post-translational modification by glycosylation and attachment of glycosaminoglycans. CD44s is ubiquitously expressed on mesenchymal cells, such as synoviocytes and subsynovial fibroblasts, and in all types of hematopoietic cells. It is the primary receptor for HA in chondrocytes where it ties a huge complex of aggrecan proteoglycan to the surface of chondrocytes by binding a single fila- ment of HA which, in turn, binds a link protein associated to often more than 50 aggrecan proteoglycan. This retention of large hydrated HA/PG/link protein can be visualized as a gel-like pericellular matrix surrounding chondrocytes in vitro 99. Analysis of the nucleotide sequence of CD44 cDNA from human, baboon and mouse predict a 37 kDa polypeptide with homology to cartilage link proteins in a phylogenetically conserved amino-terminal domain 95. CD44 has alternative ligands, such as collagen, fibronectin, fibrinogen, laminin, chondroitin sulfate, osteopontin, mucosal vascular addressin, sergly- cin/gp600, L-selectin, E-selectin and the MHC class II invariant chain. Ligand binding to CD44 is variable and depends on the state of activation: active CD44 constitutively binds HA while inducible CD44 binds HA only weakly, or not at all, unless activated by cytokines, growth factors, monoclonal antibodies or phorbol ester. Additionally, there is an inactive CD44 isotype that neither binds HA in its naïve form, nor can be induced to bind HA by any known agent. The CD44 variants (CD44v) detected so far have mainly been found in epithelial cells, keratinocytes, activated leukocytes and different tumor cells. In vivo experiments on mice using collagen-induced arthritis (CIA) as a model for RA have shown that blocking of CD44 by treatment with anti-CD44 antibody partially inhibits the disease and alleviates paw swelling 100. More recent

32 results show that variant CD44 isoforms, mostly CD44v3-v10, are found in RA SF cells and in normal keratinocytes (CD44v3-v10 only) 101. Other results indicate that the alternative splicing forming CD44v might be influenced by pro-inflammatory cytokines such as IL-1 alpha. Furthermore, it has been shown that the CD44v3-v10 binds heparin- binding growth factors like fibroblast growth factor-2; this event is dependent on heparan sulfate attached to the v3 exon and can be abolished by pre- incubation with heparin and treatment with heparinase. The v3-HS fibroblast growth factor-2 complex can efficiently interact with soluble fibroblast growth factor receptor-1, inducing intracellular signal transduction events leading to an increase in cell proliferation. A similar function has been shown concerning CD44v3 and its interaction with HB-EGF 50. EGF stimulation of mouse fibroblasts expressing wild-type HER-1 induced a both time- and dose-dependent upregulation of CD44, mRNA as well as the 95 kDa major CD44 protein, and increased cell attachment to HA- coated plates. The effect was almost completely abolished by pre-treatment with CD44 antibody immediately prior to cell attachment-assay 102. Further- more, similar results have been achieved when treating astrocytoma cells with EGF, thus showing increased CD44s mRNA and protein expression as well as increased invasion of HA-containing Matrigel™. This effect was prevented by pretreatment with an inhibitor of tyrosine phosphorylation 103. Recently, it has been shown that CD44 co-immunoprecipitates with HER-1 and HER-2 in glioma cell lines 104. Reversely, HER-2 has been found to co-immunoprecipitate with CD44 in ovarian cancer cells and to mediate HA-synthesis promoting CD44-dependent growth and migration 105.

The synovial interstitial matrix Three classes of structural polymers comprise synovial interstitial matrix; the collagen scaffolding, the extra-fibrillar glycosaminoglycans and the structural glycoproteins. The roles of normal synovial interstitial matrix are several 106: • Provides a mechanical, elastic support and attachment for the lining capillaries and for the synovial lining cells • Probably influences lining cell adhesion, proliferation and behavior, especially during embryogenesis • Provides hydraulic resistance and thereby prevents rapid drainage of SF out of the joint cavity

33 • Modifies the rate of clearance of large macromolecules like HA from the joint cavity • Possibly traps or contains antigens that contribute to the inflammatory activation in rheumatoid arthritis “We know more about what is present than about how much.” “Com- parative studies of synovial interstitial matrix in different joints seem lacking at present.”106

Synovial collagens Synovium is rich in collagen which forms both striated fibrils of period 67 nm and microfibrils. The collagen forms distinct patterns related to the depth of the location in the synovium and subsynovium. The outermost 2 – 3 μm of the lining layer contains a fine network of randomly oriented microfibrils or microfilaments with a diameter of 9–10 nm composed of type VI collagen. These type VI microfibrils promote synoviocyte adhesion in culture, bind to HA and fibronectin, and are likely to be the skeleton that maintains the structural integrity of the lining layer SIM. Type VI collagen has been shown to be resistant to the metalloproteinases released by rheumatoid synovium. Notably, when examining RA-synovium it is often possible to distinguish the outermost layer of type A cells, possibly because its collagenous matrix is not degraded by the RA-induced metalloproteinases. Scattered amongst the type VI microfibrils are striated type III collagen fibrils of period 67 nm. Deeper into the tissue, in the subsynovium, the striated fibrils are thicker and composed of both type III and type I collagen. Type III mRNA is present in synovial lining cells; by contrast, cultured ‘synovial’ cells from rheumatoid arthritis patients secrete type I and type III collagen in a ratio of 5:1, 107, the first indication that these cells are not lining cells but rather subsynovial fibroblasts. Type IV collagen has been found in the basement membrane of synovial blood vessels, and possibly surrounding synoviocytes, together with laminin, especially in aged human synovium. There are many known modulators of collagen synthesis in rheumatoid synoviocytes, e.g. increased production by TGF-beta, and decreased by IL-1, tumor necrosis factor-alpha and IFN- gamma. The potential role of TGF-alpha/HER in synovial matrix production has not yet been investigated.

The synovium during synovitis The classical features of established RA includes hypertrophy of the synovial lining layer and cellular infiltration of macrophages, plasma cells, den-

34 dritic cells, neutrophils and lymphocytes in the synovial sublining layer. The hypertrophy of the synovium involves an increase in both the macrophage-like type A synoviocyte and the fibroblast-like type B synoviocyte, although the relative increase in macrophage-like cells appears greater. Furthermore, there is formation of secondary lymphoid organs in the form of follicles and peri- vascular aggregates in the RA synovium. Finally, characteristic changes in the form of increased synovial vascularity can be seen in RA, but also in other types of synovitis such as psoriatic arthritis and pelvospondylitis. However, the vascular phenotype in non-RA synovitis seems to reflect a more intense neo- angiogenesis with tortuous vessels and higher amounts of certain protein and glycoconjugate expressions related to angiogenesis, whereas the vessels are more straight and less dense in RA 108. The distinction between synovium and subsynovium is blurred in RA due to the heavily infiltrated and extremely thickened synovial lining. The increase in type A synoviocytes is most likely dependent on the recruitment of differentiating monocytes from the circulation, reflecting the ongoing inflammatory process 109. It has later also been suggested that bone marrow stem cells of mesenchymal origin could migrate into the synovium during stages of active RA and start to differentiate and/or proliferate as fibroblast-like (type B) synoviocytes in response to locally produced cytokines 110. In fact, there is little evidence for true hyperplasia in RA synovium since cell division seems to be rare and there is an open question to whether there is a decreased ´normal apoptosis´ of synoviocytes. In a comparative study the proportion of apoptotic cells in synovial tissue during inflammatory arthritis, i.e. RA, psoriatic arthritis and reactive arthritis, was increased as compared to control tissues 108. There seems to be a relative hypo-perfusion in end-stage rheumatoid arthritis and there is a debate to whether the vascularity increases or not during the disease process. Irrespective of possible angiogenic influences, the synovial metabolic rate is elevated with extensive secretion of matrix-degrading enzymes (matrix metalloproteinases, cysteine proteases etc.) concomitant with a highly increased cellularity. Taken together, this multiplies the demands of perfusion and probably contributes to the increased apoptosis. The expression of CD44 on synovial fibroblasts seems to be important for the invasion and/or degradation of cartilage since cartilaginous matrix degradation in vitro by fibroblasts from RA-patients is efficiently inhibited by pre-treatment with CD44 antibody 111. Furthermore, the production of HA is increased in human fibroblasts when stimulated by IL-1 beta or tumor necro- 112 sis factor-alpha, first published by Sampson et al. . During inflammation

35 the ratio of fragmented HA (less than 500 kDa) to high-molecular-mass HA (more than 1 MDa) increases by activation of hyaluronidase and reactive oxygen-derived free radicals. This results in a more efficient stimulation of cell- surface CD44, in turn leading to expression of NF-κB, nitric oxide synthase and other pro-inflammatory stimuli. It also leads to stimulation of angiogenesis through the low-molecular-mass HA. Synovial sarcomas Soft tissue sarcomas are derived from mesenchymal cells and are relatively rare malignant diseases with an incidence of 2-3/100.000 per year. Synovial sarcomas account for only 7 - 10 % of all malignant mesenchymal tumors. This subgroup has long been regarded and treated as a high-grade sarcoma which histologically has been separated into biphasic and monophasic variants, the latter less distinct and more difficult to separate from other spindle and round cell sarcomas. However, during the last few years modern principles of diagnosis have been developed including detection of the characteristic t(X; 18)(p11.2; q11.2) translocation and molecular diagnosis of the resulting SYT-SSX (1,2 or rarely 4) fusion transcript. HER-1 membrane expression has been found in a two-case study of synovial sarcomas, also cDNA expression profiling has shown clustering of HER- 1 with the SSX genes in monophasic synovial sarcoma specimens. HER-2 over expression has been demonstrated in 4/4 biphasic and 3/9 monophasic synovial sarcomas as compared to levels of HER-2 expression in spindle cell sarcomas, including malignant fibrous histiocytoma using cDNA array analysis. Immunohistochemical detection and fluorescent in situ-hybridization (FISH) analysis revealed strong immunoreactivity for HER-2 in all biphasic tumors and in approximately 30% of monophasic tumors without any HER-2 gene amplification, whereas malignant fibrous histiocytoma (25 cases) and fibrosar- coma (6 cases) were completely negative. The HER-2 expression was confined to epitheloid components and there were areas in both types of synovial sarcomas that were not associated to SYT-SSX fusion type 113.

4SS synovial sarcoma cells One of several existing synovial sarcoma cell lines is the 4SS cell, also designated 4T in early publications 98. This cell line originated from an ex- cised fibroxanthomatous lesion in a finger of a 44-year-old male Caucasian 114. It appeared as a giant cell tumor of tendon sheath origin and was thus described as a synovioma. The cell line, 4T/4SS, was established from the cells that attached to a plastic Petri dish after seeding small pieces of tumor tissue in Eagle’s minimal essential medium (MEM) with penicillin, streptomycin 36 and 10% postnatal calf serum. The cytology of the 4T/4SS cells were described as ‘polygonal with pleomorphic nuclei and prominent nucleoli’ with a proportion of the cells having 2 or more nuclei. The cells grew in a pavement growth pattern and their basic mitotic index was not influenced by crowded growth conditions. Chromosomal analysis was microscopically performed and mostly showed counts between 72 – 85, e g both triploid and tetraploid cells but also a minor proportion with diploidic counts. In a later study the chromosome numbers varied from 44 to 134 with a modal number of 68, 45% of the cells containing 66 to 70 chromosomes115. One characteristic of the 4SS cell line is the ability to produce substantial amounts of HA98.

37 AIMS

The overall objective was to study links between inflammation and abnormal growth. Interesting aspects on this subject had been emerging concerning growth stimulatory effects of cytokines, and reciprocal activation between cytokines and growth factors was an interesting hypothesis. The general knowledge and literature on this subject vastly increased during the project influencing the direction we took in our studies and the questions we asked.

Our group had previously shown increased expression of TGF-alpha in differentiating monoblastic human U-937-1 cells as well as in human promyelocytic leukemia HL-60 cells. Also, cytokines like granulocyte-macrophage colony-stimulating factor (GM-CSF), IL-3 and IL-5 had been shown to upregulate TGF-alpha expression in eosinophils. Could inflammatory cytokines upregulate TGF-alpha in monocyte/macrophage type cells? Suggestions had been made that IL-6 stimulation resulted in growth promoting activity. IL-6 is an activating and differentiating cytokine for blood monocytes and had been shown to induce differentiation of U-937-1 cells. Could IL-6 induce further upregulation and/or production of TGF-alpha expression in differentiating U-937-1 cells used as a model system for monocyte differentiation? Synovial hyperplasia includes infiltrating monocytes, macrophages and monocyte-derived synoviocytes and numerous leukocytes are found in inflammatory SF. There is also production of IL-6 and GM-CSF, among other cytokines, that influence or induce abnormal growth during synovitis. The joint compartment includes cell types expressing HER-1. Is TGF-alpha found in the SF of RA-patients? What about other inflammatory joint conditions? If TGF-alpha is found, by what cell types is it produced? What cell types carry receptors for TGF-alpha in the synovia? The factors important for synovial maintenance are largely unknown. There are both monocyte-related and fibroblast-type synoviocytes in the synovial membrane; also, resident macrophages are found in the subsynovium. Is TGF-alpha present in normal SF and synovia? If so, where is it produced and where are the receptors found? If TGF-alpha is produced in synovia and influences synovial membrane homeostasis, is TGF-alpha and the HER-family involved in synovial sarcoma? Many possible links have been shown between hyaluronic acid (HA), CD44

38 and HER-family. The 4SS synovial sarcoma cells were known to produce high amounts of HA. Could these cells be used to study the TGF- alpha/HER/CD44/HA relationship in vitro? Another type of project in our group dealt with the effect of UVR on HER-1 in epidermis. There was a debate to whether melanoma and melanocytes expressed HER-1. Studies made in our group had shown effects of UVB interpreted as increased mobility of HER-1 and platelet-derived growth factor- receptors in UV-irradiated fibroblasts and keratinocytes. We wanted to clarify this matter and also study possible effects of physiological UVB-doses on HER-1 expression in normal human melanocytes. The underlying question: is there a link between UVB, HER-1 expression and abnormal growth signaling potentially promoting melanoma development?

39 COMMENTS ON MATERIALS AND METHODS Brief summary The project has required several different methods with varying applications. Some were already established in the lab and modified when needed to fit the specific purpose, others were set up and evaluated for the project described in this thesis. The methods used were: • Cell culture of four different cell models: U-937-1 human histiocytic lymphoma (Paper I), normal human epidermal melanocytes (paper II), 4SS synovial sarcoma cells and Ag1523 human normal fibroblasts (Paper IV). • Differentiation and stimulation of cells in vitro using retinoic acid, 1,α25-dihydroxycholecalciferol (Vit-D3), interleukin-6 (IL-6), phorbol-12- myristate-13-acetate (PMA), calcium ionophore (A23187)(Paper I) and TGF-alpha and EGF (Paper IV). • UVB-irradiation of cultured melanocytes (Paper II). • Proliferation analysis by 3H-thymidine incorporation (Paper IV). • DNA and cell cycle-analysis (Paper IV). • Collection and preparation of in vivo-samples of synovial fluid and human synovial biopsies (Paper III). • Indirect and direct immunofluorescent labeling analyzed by flow cytometry (Papers I, II, IV). • TGF-alpha protein analysis in conditioned medium, serum, plasma and synovial fluid by TGF-alpha enzyme-linked immunosorbent assay (ELI- SA) (Papers I, III, IV). • RNA extraction and analysis by Northern-blot and/or reverse transcriptase-polymerase chain reaction (RT-PCR) (Papers I-III). • In situ-hybridization (Paper III). • Immunohistochemistry (Paper III). • Hyaluronic acid-analysis by radioimmunoassay and ELISA (Papers III- IV). • Isolation and differential counting of leukocytes (in conjunction with Paper III). • Receptor tyrosine kinase analysis by Western blot (Paper IV).

40 The detailed protocols can be found in the respective papers. The purpose of the following presentation is to discuss special features, advantages and limitations of the methods used. In vitro-models

U-937-1 human histiocytic lymphoma cell line The original cells were a kind gift from Dr. K. Nilsson, Dep. of Patholo- gy, Uppsala University. They were maintained in continuous exponential growth in RPMI-1640 medium supplemented with 10 % heat-inactivated fetal calf serum, glutamine, penicillin and streptomycin. The proliferating undifferentiated cells grew in suspension with a doubling capacity of 36-48 hours. The U-937-1 cells seemed less eager to respond to treatment after 20 – 25 passages. To avoid attenuating results new cell cultures were started after 15-20 passages; these were taken from frozen batches previously cultured 2-5 passages in our lab. Since U-937-1 cells resemble monoblasts we were careful to use LPS-free reagents to avoid unwanted activation of the cells. Harvest was always performed using ice-cold buffers to “freeze” the cellular state and avoid mechanical activation. Differentiation of the U-937-1 cells was achieved by treatment with phorbol-12-myristate-13-acetate (PMA), all-trans retinoic acid and 1,α25- dihydroxycholecalciferol (Vit-D3), respectively. The respective differentiation phenotypes have been described by Öberg et al to resemble secretory macrophages (PMA-differentiated), phagocytic macrophage (VitD3-differentiated) and the antibody-dependent cytotoxic-cell phenotype (retinoic acid) 116. In our hands the cells differentiated to morphologic phenotypes in accordance with previous publications 116-118. All differentiation procedures resulted in cell adhesion to the culture flask and, in the case of PMA, also to neighboring cells forming spheroids of activated and differentiating cells. Therefore all cells were harvested for RNA isolation, both adherent and free floating. However, the immunofluorescent labeling was only possible to perform on cells collected from the culture medium since the flow cytometry analysis depends on single cell suspension.

4SS synovial sarcoma cell line The cells and the original biopsies were anatomically, morphologically and histologically judged to have synovial origin 114. In a later study the 4SS cell line was determined to be homogenous and uncontaminated, exhibiting chromosome numbers ranging from 44 to 134 with a modal number of 68, 45% of the cells containing 66 to 70 chromosomes 115. 41 The cells were maintained in continuous exponential growth in RPMI- 1640 medium supplemented with 10 % heat-inactivated fetal calf serum, 2 mM glutamine, 50 U/ml penicillin and 50 μg /ml streptomycin. Cell passage was performed approximately once every week with a seeding density of normally 0.02 x 106 cells/cm2 resulting in exponential growth with a doubling time of 48 h. The growth rate declined when the cells became confluent, but they continued to proliferate in piling layers (see below). The 4SS cells were almost growth-arrested when applying serum-free culture with RPMI-1640 with 0.1% bovine serum albumin, 50 U penicillin /ml, 50 μg streptomycin /ml and 2 mM glutamine. To ensure healthy, adherent non-confluent cultures we decided to allow complete medium for 24 hours after passage before changing to starving conditions. We had no difficulties restarting in vitro-culture from the samples kindly provided by Dr. B. Westermark, Dep. of Pathology, Uppsala University. However, due to the many years of cryopreservation (freezing incubation in liquid nitrogen) we wanted to verify the integrity of the cell line, e.g. that it was stable and unchanged by freezing and thawing. DNA and cell cycle- analysis were performed on 4SS cells harvested at subconfluent conditions and in the exponential growth phase. The results were concordant to previous reports. The cell phenotype and proliferation rate were also in agreement with previous descriptions, with morphologic similarities to primary cultures of synovial fibroblasts established from synovial membrane-derived multi-potent mesenchymal cells 81,114,115. Shortly, there are two types of growth patterns. A few hours after seeding 95% of the cells are attached to the culture dish or flask and the majority of cells have developed a few elongated dendritic processes reaching out to neighboring cells. This represents the ‘oligodendritic’ phenotype (Paper IV). A minority of the adherent cells develop a ‘polydendritic’ phenotype with many stellate cell processes connecting the flat, round cell to it’s neighbours (Paper IV). During culture we found a >95 % survival, with a few dying cells gathering in the medium supernatant, intermingled with numerous healthy, round, unattached cells possibly moving to another location. These viable cells found in the medium could be seeded in new culture flasks or dishes to establish new adherent cultures. When reaching confluence the cells started to build a new layer with the appearance of a honeycomb, not a simple dense mass growth pattern. In our study (Paper IV) we show that despite their somewhat mixed growth pattern, 4SS cells behave essentially as a homogenous cell population when analyzed by flow cytometry. When subculturing confluent, layered cultures the cells were pooled and seeded subconfluent

42 with reproductive results concerning growth rate and surface expressions. The scattergram of cells from subconfluent and confluent cultures were similar and histograms showed one peak. The production of hyaluronic acid (HA) is prominent in cultured 4SS cells 98. During subculture we found a high viscosity layer, likely containing HA, still attached to the adherent cells when the culture medium was completely removed. This layer consumed trypsin activity when performing subculture, resulting in prolonged trypsin treatment to release >95 % of cells into suspension. To shorten the trypsin exposure time for 4SS cells we applied a brief rinse using 0.25% trypsin before passage as a pre-treatment. To assure that the high viscosity layer did not skew the analysis of HA in the culture medium, we analyzed both the conditioned medium and the first trypsin- containing supernatant yielded when performing subculture. However, neither this increased viscosity, nor the rich HA-content seemed to hamper harvest, immunofluorescent labeling, cell lysing for protein analysis or ELISA performed on conditioned media.

Ag1523 human fibroblasts The Ag1523 normal human fibroblasts were purchased from American Type Culture Collection (Manassas, VA, USA) and cultured parallel to the 4SS cells. The Ag1523 cells were kept in continuous growth in Eagle’s Mini- mum Essential Medium (EMEM, Gibco, Invitrogen Life Technologies) supplemented with 10% FCS, 50 U/ml penicillin, 50 μg /ml streptomycin and 2 mM glutamine. The Ag1523 cultures were passaged 1:2 approximately once every 7 – 10 days, when the cells reached confluence and a decreasing number of mitoses were observed. The cells were incubated at 37°C in humidified air with 5% CO2 and the experiments were performed on cells in the logarithmic growth phase at passage 17 - 20. The fibroblasts have a lower proliferation rate than 4SS cells and show density-dependent inhibition of growth. The Ag1523 cells were used as control cells and have been extensively characterized by others. Passage of cells was performed by treatment with 0.25% trypsin (Gibco, Invitrogen) at room temperature. Starving conditions were applied by seeding the respective culture as described and changing to FCS-free medium (EMEM with 0.1% bovine serum albumin, 50 U/ml penicillin, 50 μg /ml streptomycin and 2 mM glutamine) 20 - 24 hours after passage.

Melanocytes Normal human epidermal melanocytes were commercially obtained from Promo Cell (Heidelberg, Germany). The cells originated from foreskin 43 of a 6-year old Caucasian male and were delivered as a cryo-preserved cell suspension of 0.5 x 106 cells/ml. The cultures were maintained in serum-free melanocyte growth medium, as described (Paper II). Previously we had successfully been using primary cultures of melanocytes in our group (manuscript by Lirvall et. al.). However, the type of experiments performed in the current study required the availability of cells without months of delay and also a greater number than in previous studies. Furthermore, we considered the possibility of exact repetition of experiments, possibly involving iterated delivery, greater with commercial cells. The melanocytes were cultured as described (Paper II) and serially sub- cultured to expand the population. The experiments were performed at passage 10 – 20. Several plates were needed for each parameter in an experiment to achieve sufficient number of cells; the melanocytes are large and stop proliferating at confluence. Patient samples All patient samples were collected after informed consent from the do- nors and the project and procedures were approved by the local Ethical Committee at Linköping University hospital. Patients with RA had a long- standing disease (>3 yr duration) and fulfilled the 1987 American College of Rheumatology criteria for RA. They had various pharmacological treatment regimens set up on clinical indications and these were not interrupted by this study. Synovial tissue biopsies and synovial fluid (SF) were collected by the op- erating orthopedic surgeon from patients with RA and osteoarthritis undergoing knee joint surgery. Separate SF samples waere collected from patients with RA, psoriatic arthropathy, pelvospondylitis, reactive arthritis, multifocal aseptic arthritis and systemic lupus erythematosis (SLE) treated by needle joint puncture with SF aspiration in conjunction with intra-articular cortisone injections. The non-inflammatory control samples of SF and synovial biopsies were collected by the arthroscopist from patients aged 20 – 37 years undergoing explorative arthroscopy as part of the investigation of various non- inflammatory joint conditions (control), including post-traumatic controls and post-traumatic arthralgia. SF was aspirated immediately after penetration into the joint cavity, before any surgical or pharmacological treatment procedures were initiated, and put in sterile tubes for consecutive processing. Biop- sies of synovial tissue (2 - 4 x 2 - 4 mm) were taken from visually inspected synovium and immediately put in 4% paraformaldehyde in PBS, pH 7.4. Pa- tients were accepted as controls only when the arthroscopic inspection of the

44 synovium was performed without any visible signs of inflammation. One of the control samples was discarded after immunohistochemical evaluation due to infiltrating inflammatory cells found in the synovium. The osteoarthritis synovial tissue was sampled at the bordering line between cartilage and synovia, while RA samples were taken from synovium showing visually assessed inflammatory characteristics, including thickening. SF samples were centrifuged at 2000 x g for 10 minutes at room temperature to pellet debris and potential cells residing in the SF. Especially the control samples were highly viscous consequently demanding high impact centrifugation to yield possible cells residing in the SF. However, control samples were almost devoid of cells. Inflammatory SF’s were often less viscous and with increased numbers of cells, mostly infiltrating white blood cells. The supernatants were collected and split in two; 500 pg/ml human recombinant TGF-alpha (Peprotech) was carefully resuspended in one sample after which both samples were aliquoted and frozen at –20ºC.

Cells collected from SF Cells found in the SF were washed twice with cold phosphate-buffered saline (PBS, pH 7.4), counted and spun onto standard preparation slides using a Cytospin centrifuge (Shandon). When the cell yield was sufficient, the SF cells were also prepared for isolation of total RNA by treatment with ice- cold 4M guanidine-isothiocyanate and immediately frozen at -70°C until further processing. Cytospin slides of cells collected from SF were stained by hematoxylin/eosin and differentially counted for blood cell types. Some slides were also subjected to immunohistochemistry (IHC) using mouse-anti-human CD68 (KP-1, DAKO) to verify the number of macrophages estimated by differential counting. Cell growth Cell proliferation analysis was performed by counting cells during passage and by measuring incorporation of tritiated thymidine (3H-thymidine) into the cells. Thymidine is used for DNA synthesis and is incorporated into the DNA of multiplying cells; 3H-labeling includes a stable radioactive isotope into the DNA. The labeled cells are washed, dried and dissolved into a scintil- lation fluid and the emission of beta-particles is measured in a liquid scintilla- tion counter. For this purpose, 0.31 x 106 cells were seeded in 35 mm culture dishes with 2 ml fresh complete medium (see above) to yield a density of 0.03 x 106 cells/cm2. The medium was changed to 3H-thymidine supplemented medium after 6 – 24 hours, depending on experiment, and the cells were allowed to incubate for another 24 – 48 hours before termination. 45 The culture density of 4SS cells had important effects on DNA synthesis which declined with increasing cell density. The incorporation of 3H- thymidine during 48 hours showed a negative correlation to cell seeding density with a nearly logarithmic curve shape (Paper IV). To ascertain that the effect was due to increasing cell contacts and not lack of nutrients we performed these experiments at three different occasions. The cells were seeded at varying densities and each plate was supplemented with the same volume of complete medium to assure an excess of the nutritional needs. DNA and cell cycle-analyses were performed on 4SS cells harvested at subconfluent conditions and in the exponential growth phase. The results were concordant with previous reports showing predominantly aneuploid cells (76.4%) with a minority (23.6%) of cells with diploid chromatin after 48 hours of incubation in RPMI-1640 supplemented with glutamine and 10% FCS. 44.4% of the cells are in either G2- or S-phase showing a high capacity of cell division at 24 h after passage. This capacity declined with time, and increasing cell density, to 27.2 % in G2/S-phase at 72 h after passage. RNA-analysis methods To determine gene expression the alternatives are several. We have used three different methods depending on what cells and/or tissues we wished to analyze and the issues we needed to resolve. Naturally, the possibility to dis- criminate cell types and tissues, before or after RNA-analysis, is a prerequisite that needs to be considered before choosing method. In all cases one of the main problems is how to guard the integrity of the RNA through the whole analysis. Another main issue is to choose a method with sufficient sensitivity to detect relevant levels of gene expression, but also with specificity high enough to ensure that one analyzes the specific RNA in question. For that purpose, the non-specific background labeling should be as low as possible.

In situ-hybridization This is the method of choice when you want to pinpoint a specific gene expression to a cell type in a mixed population tissue without destroying the tissue morphology. In conjunction with immunohistochemical labeling of the protein in question, either on the same or consecutive sections, a positive result will be robust and hardly questionable. We successfully chose this combination for analyzing TGF-alpha in synovial biopsies. The in situ-hybridization was carried out essentially as described in the manual "Non-radioactive In Situ Hybridization" (Boeringer Mannheim), with

46 some modifications, using digoxigenin (DIG) labeling and colorimetric detection with NBT/BCIP. For transcription we used a digoxigenin-RNA (DIG- RNA) labeling kit (Boeringer Mannheim) producing full length probes from a 603 kb TGF-alpha cDNA. The non-radioactive labeling of probes accelerates the analysis as compared to classical radioactive probes; however, it is more difficult to perform concomitant immunohistochemistry. Also, it is probably not as sensitive as radioactive analysis, but proved sensitive enough for our purpose. The distinct morphology of the synovia made the analysis straight forward and no tissue staining was needed. We also could compare directly with the next consecutive section that was immunolabeled in combination with tissue staining (hematoxylin). The finished sections were mounted with an aqueous mounting solution and analyzed in bright-field microscopy. The in situ-hybridization was always performed with anti-sense and sense TGF-alpha riboprobes in parallel. Endothelium and vascular smooth muscle in the synovial tissue sections were used as internal positive controls for TGF- alpha mRNA expression. Sections of human skin tissue, treated as described for synovial biopsies, were used as external positive controls for TGF-alpha mRNA expression.

RNA-isolation Total RNA was purified from cells according to the original method by Chomzcynski and Sacchi, the principle method on which most of the commercially available RNA-purifying kits are based 119. The cells, adherent on a culture plate/flask or pelleted, were disrupted in ice-cold 4 M guanidinium thiocyanate, 0.3 M 2-mercaptoethanol and 25 mM sodium citrate. This treatment dissolves the cell membranes and protein complexes but keep the integrity of RNA and DNA. Total cellular RNA, including ribosomal, messenger RNA (mRNA) and hetero-nuclear RNA (hnRNA, newly transcribed non-spliced RNA) was extracted by centrifugation with 0.2 M sodium acetate pH 4.1, phenol-chloroform (5:1) pH 4.7 and precipitated from the resulting aqueous phase with isopropanol (-20°C). The RNA was extracted a second time as described above and precipitated with 5 M NaCl (1:10) in 95% ethanol. After centrifugation the resulting RNA-pellet was washed twice with ethanol, dissolved in RNAse-free water, quantified and evaluated by spectrophotometry and frozen (-70°C) for later PCR and/or Northern blot-analysis. When performing spectrophotometry the absorbance of UV-light at wavelengths 280/260 was > 0.6 and upwards to maximum 0.88, where =1 would be the optimal quote measured on a clean RNA-only sample.

47 The RNA was not further purified since we were interested in performing Northern Blot and the amount of total RNA was limited in some cases. Also, the detection of specific RNA in the following methods worked fine without isolating mRNA including the RT-PCR. The mean yield of total-RNA was approximately 100 μg/10 x 106 U-937-1 cells. The yield from SF cells was limited and samples containing <1 μg was only analyzed by RT-PCR (see below).

Northern Blot This can be the method of choice when the question is quantification, comparing samples, e.g. kinetic studies, and at the same time ensuring an approximate size measurement and separation of the specific RNA. It will also give a visual confirmation of the integrity of the RNA-samples. The main drawbacks are the large amount of samples required and that the method is time-consuming. The Northern Blot experiments were performed essentially according to a standard protocol described in Paper I 120. The RNA samples were electrophoresed in 1.5 % agarose-gels using bromophenol-blue to visualize the rRNA-bands used as size and separations markers. The gels were photographed in UV-light before blotted onto nylon membranes and the RNA was fixed to the membranes by UV-crosslinking. TGF-alpha (4.5 kb), HER-1 (6 kb), glyceraldehyde phosphodehydrogenase (1.3 kb) and beta-actin (1.8 kb) were detected by radioactive probes, using alpha-32P-dCTP labeled oligoprobes (Multiprime labeling kit, Amersham, UK) made from a 925-bp human TGF- alpha cDNA fragment 121, HER-1 cDNA, and cDNA for the house-keeping genes glyceraldehyde phosphodehydrogenase (GAPDH) and beta-actin as controls. The hybridized membranes were exposed for 6 – 24 h on a BAS-MS Im- aging plate system (Fujifilm), scanned by a PhosphoImager (BAS 1500, Fuji- film) and analyzed using the Image Gauge software (Fujifilm). The relative amount of electrophoresed total RNA/lane transferred onto the membrane was calculated by evaluation of the intensity of the beta-actin and GAPDH- bands, respectively, after hybridization. The intensity of the bands corres- ponding to TGF-alpha and HER-1 mRNA was measured and corrected for the calculated differences in total RNA.

RT-PCR Total cellular RNA was isolated from SF cells as described resulting in a limited number of samples with high-quality RNA suitable for TGF-alpha mRNA analysis (n=6). RT-PCR analysis was started with a standard reverse transcription procedure, it was then performed using a step-down PCR proto- 48 col fitted for our purpose. Step-down means starting the PCR-cycling with a higher stringency, using a higher melting temperature, making sure that only the desired mRNA is amplified. After a few cycles the so called melting temperature can be lowered to increase the amount of primer binding to target without loosing specificity since the amount of desired DNA copies is already increased. This speeds up the amplification process. The protocol used was briefly as follows: DNAse-treated total cellular RNA was linearized for 10 minutes at 70°C and reverse transcribed using either AMV reverse transcriptase or Superscript II (both from Promega Corpo- ration, USA). The resulting cDNA library was analyzed for TGF-alpha gene expression using forward primer (5´-GTC TGC CAT TCT GGG TAC GT- 3´) and reverse primer (5´-TCT GGG CTC TTC AGA CCA CT-3´) (GIB- CO BRL, Life Technologies) applying a step-down PCR protocol with totally 35 amplification cycles. The amplified DNA was denatured and electrophoresed in 3% TAE-agarose gels (3% agarose, 40 mM TrisAcetate, 1 mM ethylene-diamine-tetra-acetic acid (EDTA) and 1 x 10-3 mg/ml ethidium bromide). The gels were photographed and evaluated by UV-light examination. Immunofluorescent labeling and flow cytometry The detection of membrane-bound molecules on live cells has been made possible by the invention of stable fluorescent molecules possible to use as conjugates to i.e. antibodies. When activated by light these molecules emit photons at a specific wave-length and light-sensitive detectors are used to cap- ture this emission. The detectors can be part of a microscope for single cell analysis or a flow cytometer for collecting data on thousands or millions of cells in a few minutes. Data is captured, collected and analyzed using specialized computer software. Antibodies directed towards a molecule of interest, e.g. TGF-alpha, can be labeled by coupling fluorescent molecules to the Fc-part, the “tail”, of the antibody. Either the primary antibody is conjugated with a fluorochrome; this is known as direct immunofluorescent labeling. Alternatively, indirect immunofluorescent labeling is performed. First the TGF-alpha antibody is used for primary labeling of cells and secondly a fluorochrome-conjugated reporter- antibody recognizing the specific Fc-part of anti-TGF-alpha is allowed to incubate with the cells. The results are close to momentary images of TGF-alpha molecules present on the cells and can also give an approximate estimation to how many there are. When studying U-937-1 cells (Paper I) we established immunofluorescent cytometry in the lab using both direct and indirect labeling to detect IL-6

49 receptor subunits, gp130 and IL-6-R alpha subunit/p80, as well as TGF-alpha. The fluorescence intensity of the cells was measured by a FACS Calibur (Per- kin Elmer) and analyzed using the Cell Quest software. We developed a protocol starting with blocking of unspecific antibody- affinity sites on the cells by incubation with human immunoglobulins at room temperature for 5 minutes. This was immediately followed by primary antibody labeling at ice-cold (<4°C) conditions for 45 minutes. The following washes and secondary labeling were also performed ice-cold including the final fixation procedure. The fixation was accomplished by suspending the labeled cells in 4 % paraformaldehyde and directly adding ice-cold PBS to yield a resulting concentration of 0.5 % paraformaldehyde in PBS. Since the labeling of TGF-alpha was significant but relatively low in U-937-1 cells, we were careful testing the specificity with concomitant controls including non-labeled (autofluorescent) cells, isotype-negative (non-binding irrelevant primary antibody) and blocked anti-TGF-alpha, i.e. anti-TGF-alpha preincubated with human recombinant TGF-alpha protein to saturate all TGF-alpha binding sites. The autofluorescent and isotype-negative controls were thereafter used in each experiment. The same principles, apart from the blocked control, were applied when titrating immunofluorescent labeling of HER-1, HER-3, CD14, CD36, CD44, CD68 and VCAM. In the case of HER-2, HER-4 and CD68 the respective antibodies were directed towards intracellular sites of the molecule (HER-2 and HER-4) or the whole protein was located intracellularly (CD68). For this purpose the DAKO Intra-stain kit was successfully used according to the manufacturer’s description, concomitantly labeling primary target protein, autofluorescent cells and isotype-negative controls (Paper IV). Immunohistochemistry Immunohistochemistry (IHC) was performed with primary monoclonal mouse antibodies, polyclonal goat antibodies and polyclonal rabbit antibodies in parallel with the appropriate isotype-negative controls provided from the same manufacturers. The secondary linker antibodies used were rabbit-anti- mouse immunoglobulins, rabbit-anti-goat immunoglobulins and swine-anti- rabbit immunoglobulins detected by mouse peroxidase-anti-peroxidase complex (PAP), rabbit-PAP and goat-PAP with 3,3´-diaminobenzidine tetrahydrochloride (DAB) as substrate. Alternatively, the detection of primary antibody binding was performed using the LSAB+ kit (DAKO A/S, Copenhagen, Denmark) based on alkaline phophatase and Fast Red substrate.

50 Antigen retrieval was performed on freshly dewaxed, rehydrated slides by microwave treatment for 10 minutes at 100°C in 0.01 M citrate-buffer (pH 6.0). All IHC procedures from here were performed at room temperature. Slides were allowed to cool in the citrate-buffer and were then treated with 0.2 % Triton X-100 in Tris-buffered saline (TBS). The rinsed slides were blocked with 10 % swine serum in Tris-buffered saline (TBS) for 10 minutes, where after the IHC was performed using the LSAB+ kit according to the manufacturer´s description. In some cases detection was accomplished by a standard procedure with a PAP-conjugated antibody-complex, using DAB for visualization. In the latter case, potential endogenous peroxidase activity on the slides was blocked by treatment with 10 % hydrogen peroxide prior to immunolabeling. Slides were counterstained with Mayer´s hematoxylin (Merck, Germany), rinsed with tap water and mounted with cover slips. Negative control experiments were performed by using the appropriate isotype-negative primary antibody and by omitting the primary antibody. The endothelium and vascular smooth muscle were used as internal positive controls for TGF-alpha protein immunodetection in the synovial tissue sections. Protein analysis

Protein isolation Whole protein extracts were prepared from 4SS and Ag1523 cells using direct lysis of whole cells in the culture petri dishes. The culture medium was thoroughly removed; the adherent cells were washed twice with ice-cold TBS and then kept on ice through the whole lysis procedure. The cells were lysed in ice-cold RIPA-buffer (65 mM Tris-Base, 154 mM NaCl, 1% NP-40, 0.25% Na-deoxycholate, 1 mM EDTA, pH 7.4), containing freshly added 1 mM NaF, 0.5 mM orthovanadate and protease-inhibitor cocktail (20 μl/1 ml RIPA) to protect the proteins from degradation. The following preparation was performed according to a standard method. Special care was taken when scraping the lysates from the culture dishes to collect the protein lysate in two sessions, the second time gathering remnant proteins by washing the plates with ice- cold TBS. After incubation on ice for 20 minutes, the samples were carefully homogenized by pulling the lysates 10 times through a 22G needle. After repetitive centrifugations the supernatants were clear and the protein extracts were frozen and stored (-70°C) for later Western blot analysis. Measurement of protein content was performed by spectrophotometry using the Bradford method and always after 1 freeze-thaw cycle to improve the accuracy. Approx- imate protein yield was 200-400 μg/5x106 4SS cells and the samples were aliquoted and frozen again until performing Western blot. 51 Western blot analysis The prerequisite of this method is to yield good quality protein lysates collected from a clean system, preferably one cell type and free of microbes and other pathogens. The results are highly useful since the protein in question is both immunodetected and also size-separated. Comparisons are easily made between cell types and between treated/untreated cells. There are different types of electrophoresis and gels depending on the question asked, but SDS-PAGE is a common standard method that has been made easier, less time-consuming and highly replicative by commercially available precast gels of high quality. We used SDS-PAGE and Western Blotting for studying effects of culture conditions, EGF- and TGF-alpha stimulation, and blocking of HER-receptors and/or CD44 on the production and phosphorylation of HER-1 and HER-2 in 4SS and Ag1523 cells. Whole cell lysates were prepared as described above and processed for SDS-PAGE by thawing and measuring the protein content in triplets using the Bradford method. Equal amounts of cell lysate for each lane, 15 – 30 micrograms depending on gel size, were prepared for SDS- PAGE and immediately loaded onto Criterion 5% Tris-HCl gels (Bio-Rad La- boratories) and electrophoresed at 200 V for 50 – 60 minutes. Bio-Rad 2- colour Protein Standard was used as size marker. After electrophoresis, the gels/samples were carefully transferred onto PVDF membranes, rinsed in TBS and kept in cold TBS, 0.1% Tween, (4°C) until immunodetection. Immunoblotting was performed essentially according to standard protocols and at the recommended dilutions in the manufacturer´s guidelines supplied with the respective antibodies (see paper IV). The immunolabeled bands were visualized using enhanced chemoluminescence (Amersham Bios- cience, UK). Phospho-specific immunolabeling was always performed on fresh, previously unlabeled membranes followed by a stripping procedure with 25-30 minutes incubation in strip-buffer (62.5 mM Tris-Base, 10% SDS, pH 6.7) at 55°C ended by extensive washing in 10-12 changes of deionized clean distilled H2O to remove the antibody. A second immunolabeling was then performed using the respective general receptor-specific antibody to as- sess the total amount of protein. All PVDF-membranes were secondly or thirdly relabeled using beta-actin antibody to verify equal loading of protein in each well.

TGF-alpha ELISA Conditioned medium collected from cell cultures and SF supernatants were analyzed for s-TGF-alpha protein using a highly specific quantitative anti-

52 TGF-alpha enzyme-linked immunosorbent assay (ELISA) kit (Oncogene Science, Manhasset, NY, USA) according to the manufacturer´s instructions. This sandwich ELISA detects both native and denatured TGF-alpha via two different antibodies at a detection level of less than 10 pg/ml TGF-alpha protein. The SF samples were processed for TGF-alpha analysis by thawing and prediluting the samples 1:2, 1:5, 1:10 and 1:25 with Assay buffer (included in the TGF-alpha ELISA-kit, Oncogene Science) containing protease inhibitors and bovine serum albumin (BSA) . In parallel, we also analyzed SF taken from the same original SF samples loaded with 500 pg/ml hrTGF-alpha protein immediately before freezing storage, yielding a TGF-alpha recovery factor for each original SF sample. The final results were calculated by dividing the measured result by the hrTGF-alpha protein (500 pg/ml) recovery factor for each sample [recovery factor = (measured TGF-alpha in loaded sample – measured TGF-alpha in neat sample)/500 pg/ml]. The recovery factor was estimated by analyzing samples loaded with addi- tive hrTGF-alpha protein (500 pg/ml) at the time of collection, in parallel with crude samples and then calculating the recovery as [(loaded*dilution factor - crude*dilution factor)/500]. Microscopic evaluation All sections of synovial tissues subjected to either IHC or ISH were evaluated by qualitative analysis of staining in bright-field microscopy using 40 x, 100 x and 400 x magnification. The sections were systematically examined by assessors who were blinded to the identity and clinical details of the patients. The ISH sections were examined by two assessors (ALH and KB), whereas the immunostained sections were examined by one assessor (ALH). The results were scored in arbitrary units (0, (+), +, ++ and +++) and subdivided for synovial lining layer, subsynovial tissue, infiltrating inflammatory cells, lymphoid aggregates/nodules, as well as endothelium and vascular smooth muscle (vascular wall). The individual results of the different assessors were in agreement on both localisation of specific positive staining as well as on the intra-section difference in staining intensity between types of cells and/or tissue. Differen- tial counting of cells in SF was performed on cytospin-slides (as described above) that were stained and counted in the clinical routine lab.

Statistical methods Generally, the statistical methods used to compare cell culture results were paired and unpaired t-tests and the histograms were presented with standard error (SE) and standard error of the mean (SEM) as indicated in the figures. One asterisk indicates statistical significance with p<0.05, two aste- risks indicate p<0.01. Flow cytometry analyses were performed using the Kolmogorov-Smirnov statistics in the Cell Quest software to compare histograms and thereby verify- ing the statistical difference between the distributions of specific immunofluorescent labeling as compared to isotype negative control. The median fluorescence intensity was used to calculate differences in labeling intensity as an approximate measure of increased or decreased number of target molecules on/in the labeled cells. Statistical comparisons of the respective fluorescence intensity, after subtraction of the measured fluorescence intensity for sample- matched autofluorescent controls, were made using unpaired two-tailed t-test postulating unequal distributions. TGF-alpha protein results in RA SF were compared with controls as well as with SF from osteoarthritis and other inflammatory conditions, respectively, using one-tailed unpaired T-test for two groups with unequal variances. This choice was made on the basis of ‘0’pg/ml being the lowest result, and not uncommon among controls. Also, the number of RA-samples were three times that of the controls, consequently an unpaired test with unequal va- riance was applied. The results were presented as histograms with standard error of the mean as error bars. Correlation analyses were performed using linear regression and the Pearson coefficient. Statistical analyses were performed using Microsoft Excel statistics and Statistica 6.0 (Statsoft).

54 RESULTS AND DISCUSSION IL-6 effect on TGF-alpha mRNA in monocytoid cells The inflammatory cytokine influence on TGF-alpha production had previously been studied in eosinophils and macrophages using GM-CSF, IL-3 and IL-5 15,122,123. These factors all belong to the same subfamily of interleukins; they depend on the same heterodimerization subunit for intracellular signal transduction, the common beta receptor (βc), and they are important mediators of allergic inflammation and hematopoiesis induced by infection 124. We were interested in studying a more universal and acute influence on TGF-alpha expression, and especially in monocyte/macrophage-type cells. IL-6 is a multifunctional cytokine sharing interleukin heteromeric receptor subunit, gp130, with IL-11, oncostatin M type I and II (leukaemia inhibitory factor, LIF) and ciliary neurotrophic factor. IL-6 is known to activate monocytes, as well as other cell types, in acute inflammation. It is found in wound fluid and is expressed during inflammatory conditions such as RA, chronic inflammatory bowel disease and peritonitis; it is also involved in cancer 125. We used U-937-1 cells treated with phorbolester (PMA), retinoic acid and dihydroxycholecalciferol (VitD3), respectively, to imitate the development of mature, macrophage-like cells differentiated from blood monocytes by various functional stimuli 116,117. We could locate membrane-bound TGF-alpha by indirect immunofluorescent labeling in all three differentiations as well as in undifferentiated U-937-1 cells (Paper I). As previously shown, the levels of TGF-alpha mRNA varied depending on differentiating agent, showing the highest expression in cells treated with VitD3, i.e. the phagocyte-like phenotype 116,118. IL-6 has been shown to induce differentiation in peripheral blood monocytes, alleviating their extravasation and participation in host defense and repair. Undifferentiated U-937-1 cells were known to produce both IL-6 and IL-6-receptors, but only PMA-differentiated cells secreted IL-6 while retinoic 117 acid- and VitD3-differentiated cells did not . The IL-6-R had been detected in all types of U-937-1 cultures, but with a lower expression in the differentiated cells 117. We could show by immunofluorescent labeling that undifferentiated and differentiated U-937-1 cells had both IL-6-R-alpha subunit and the gp130 signaling subunit on the cell surface (described in text in Paper I). However, only the PMA-differentiated cells showed a functional response in the form of a statistically significant 170% increase in TGF-alpha mRNA (p<0.05) after treatment with human recombinant IL-6 (hr-IL-6) for 6 hours, and this increase had not faded after 12 hours (Paper I). The PMA- 55 differentiated cells also released the highest levels of s-TGF-alpha of all differentiations tested (Paper I). This is not unexpected but rather strengthens previous findings indicating that the PMA-cells resemble secretory macrophages producing and releasing various cytokines and growth factors 116. There were previous results assigning a growth-promoting effect to IL-6, e.g. in SF, mediating growth of synovial fibroblasts in vitro 126. The growth- promoting effect had also been shown in other systems and had been linked to the downstream signaling activation of MAPK-pathway, although the efforts to directly link IL-6 alone to these effectors were not convincing 127-129. The IL-6-induced increase of TGF-alpha mRNA in “macrophage- differentiated” U-937-1 revealed a candidate growth factor for mediating the growth-promoting effect of IL-6. It also indicated functional IL-6 receptor complexes on PMA-differentiated U-937-1 cells. Despite these encouraging results we could not show an increase in TGF-alpha protein production by treatment with IL-6, as measured by TGF- alpha release and TGF-alpha surface immunolabeling after 6 hours of incubation. Possibly the stimulation procedure used to collect conditioned medium samples increased the basal TACE-activity in our cells due to the increased cell density which also limited the incubation to 6 hours. We have no data on TGF-alpha release after 12 hours. Furthermore, the PMA-differentiated cells were probably already performing high cleavage activity by PMA-induced activation of TACE. Nevertheless, there was no visible increase in surface TGF- alpha immunolabeling suggesting that the increase in TGF-alpha mRNA did not result in any immediate increase in protein production. Effect of UVB-irradiation on HER-1 expression in melanocytes There is a rapidly increasing amount of knowledge concerning the cellular effects of ultraviolet radiation (UVR) on human skin. UVR causes activation of apoptotic and proliferative pathways, cytokine production inducing inflammatory responses, and DNA damage. We were interested in studying the effects on HER-1 in normal melanocytes after UVR. Keratinocytes and fibroblasts have been more intensively studied than melanocytes; they are more accessible and continuously growing in culture using principally standard conditions and medium. In vitro cultures of cell lines established from melanomas are also usually a rewarding and relatively simple task. Normal human melanocytes, however, are slowly replicating and have special demands concerning culture medium and growth factor additives that make this culture procedure a delicate and expensive task. It is difficult to es- 56 tablish primary melanocyte cultures, mainly dependent on restricted access to suitable fresh tissue samples resulting in a small number of purified viable cells. Expanding the primary cultures to reach the number of cells needed for a few experiments takes several weeks of sub-culturing. There is a limit to the number of passages possible to perform without ending up with phase III-cells 130 when also the phenotype has probably changed and is no longer represent- ative for melanocytes. We cultured commercially available normal human melanocytes isolated from Caucasian foreskin samples. The knowledge concerning effects of UVR on melanocyte expression of growth factor ligands and receptors was limited, although it was known that melanocytes differentiate and also proliferate in response to UVR 7. Keratino- cytes produce TGF-alpha and HER-1, they can also produce TGF-beta, amphiregulin, platelet-derived growth factor, basic fibroblast growth factor, nerve growth factor, vascular endothelial growth factor , stem cell factor, leukemia inhibitory factor, endothelin-1 and melanocortins 9,131. Cytokine production by keratinocytes includes IL-1-alpha and -beta, IL-6, IL-8, GM-CSF and tumor necrosis factor-alpha, some of which are known to induce TGF-alpha expression 15,123,127,131. UVB causes DNA damage, generates oxygen radicals and activates keratinocytes to increase their production of paracrine factors influencing melanogenesis and melanocyte survival 9. UVA radiation can augment UVB-induced inflammation by increased production of IL-1-alpha which in turn further increases cytokine and growth factor production 131. TGF-alpha and HER-1 expression had been detected in melanocytic lesions, both benign and dysplastic, as well as in melanocytes and melanomas 132-135. However, there was a debate to whether normal melanocytes actually produced HER-1, potentially linked to the density of melanocyte cultures. We set out to determine this and also to study the influence of UVB on HER-1 in normal human melanocytes in culture. We could show that subconfluent cultured normal melanocytes express low steady-state levels of HER-1 mRNA. This was supported by a study published concomitant with our ongoing work 136. After exposure to a single physiologic dose of UVB, low (not likely to induce sunburn) or high (likely to induce sunburn), we could show a biphasic change of the HER-1 mRNA expression, beginning with an initial decrease starting at 4 hours post-UVB and with a nadir at a significant decrease 12 hours after UVB-exposure. Concomi- tant analysis of HER-1 surface immunofluorescence revealed an immediate decrease in surface HER-1, followed by a slow reconstitution and a slight increase at 12 hours post-UVB. At 24 hours after UVB-exposure the HER-1 immunofluorescence had returned to control levels (Paper II).

57 Our results seem to be in agreement with the model of early and late responses to sunburn and previous results showing growth arrest of UV- irradiated melanocytes due to activation of p53 and p16 signaling pathways. The keratinocyte response to UVR is often a continuation of these pathways leading into apoptosis. However, the DNA repair systems are normally highly effective in melanocytes and the UV-induced growth arrest gives a longer cell cycle which allows for efficient repair and continued survival 137. The UV- irradiated melanocytes also differentiate, redistribute melanin and upgrade their melanin production system, then the cells are ready to proliferate. As judged by the timing of the early sunburn response, the initial inflammation appears within hours and starts to fade within 24 hours. The timing of the initial down-regulation of HER-1 in our experiments fits the sunburn inflammatory response effectuated by cytokines within hours; the growth arrest should include a down-regulation of HER-1 to reduce potentially proliferative stimuli by keratinocyte-produced TGF-alpha 138. Pre- sumably the paracrine TGF-alpha production increases after UVR exposure due to the expression of certain TGF-alpha inducing cytokines, e.g. IL-6 and GM-CSF. The subsequent upregulation of HER-1 may indicate a possible role for HER-1 in melanocyte survival and/or proliferation after repair. This hypothesis is supported by the detection of TGF-alpha in both benign and dysplastic melanocytic lesions 133. In vitro cultures of normal melanocytes had been shown to increase secretion and surface expression of TGF-alpha within 12 hours after single-dose UVR, with a complete induction within 24 hours 134. Furthermore, stimulation of melanocytes in vitro with TGF-alpha resulted in rapid phosphorylation of HER-1 which increased with prolonged stimulation, however without obvious proliferation 135. In vivo studies on mice engineered to over express TGF-alpha resulted in development of melanocytic lesions 139. We did not measure TGF-alpha protein expression in our cultures, but it is possible that the upregulation of HER-1 mRNA results in increased HER-1 protein that is rapidly internalized due to TGF-alpha stimulation and therefore is not available for immunofluorescent detection. A recent study performed on normal human melanocytes, provided from the same source as in our experiments, showed rapid down-regulation of HER-1 upon EGF stimulation 136. Other recent publications have shown the presence of HER-1, HER-2, HER-3 and HER-4 in bipolar non-confluent melanocyte cultures 140, and of HER-2, HER-3 as well as heregulin (HRG) in cultured melanocytes 141. They also showed migratory effects of heregulin-beta-1, a HER-3/HER-4 ligand, on

58 these cells. Unfortunately the effects of TGF-alpha stimulation on migration and proliferation were inconclusive due to the massive doses used (100 ng/ml) 140. The optimal effect of TGF-alpha stimulation is likely in the area of 500-1000 pg/ml whereas high doses may down-regulate HER-1 expression and shut down proliferative pathways. It seems as if the in vitro-results are extremely dependent on the model system and stimuli chosen and the challenge is to understand the synergistic effects of the HER-system in relevant models for UVR-induced melanocyte proliferation and transformation. TGF-alpha may have a role in synovial joint homeostasis Synovial joint inflammation, synovitis, in active rheumatoid arthritis (RA) involves synovial hyperplasia and an invasive behavior of the so called pannus cells at the synovia/cartilage transition zone. The bulky pannus formation constitutes proliferating synovial cells and macrophages, as well as other inflammatory cells, and contributes to cartilage breakdown. The joint compartment contains different HER-1 expressing cell types and tissues and EGF has been detected in synovial fluid (SF); moreover, normal SF contained a chondrocyte differentiating activity generating an equivalent chondrogenesis to that achieved by treatment with EGF and platelet-derived growth factor 142- 145. There were no published data on the presence of TGF-alpha in SF or joint compartment. We were interested in studying TGF-alpha expression in vivo in a condition characterized by inflammation and hyperproliferation involving monocytes, macrophages and monocytoid cells. We hypothesized that s- TGF-alpha would be present in SF during synovitis and that it could be produced by macrophages infiltrating the synovia. We started the project by analyzing SF samples from RA-patients as well as non-inflammatory controls and could detect varying amounts of TGF-alpha in the majority of samples using a TGF-alpha ELISA (Paper III). This was a new finding and there seemed to be a correlation between inflammation and higher levels of TGF-alpha, sometimes extremely high (>1700 pg/ml). We also found higher levels of TGF-alpha in SF from other types of synovitis than in control samples (Paper III). There are increased numbers of macrophages in synovitis and especially in pannus formation during active RA 78. Our previous results concerning IL-6 regulation of TGF-alpha expression (Paper I), others showing growth promoting effects of IL-6, and further evidence of increased TGF-alpha expression by GM-CSF treatment, all pointed towards a possible inflammatory upregulation of TGF-alpha in synovitis where these cytokines are known to be important mediators 15,109,123,126-129.

59 SF samples containing inflammatory cells were processed for differential counting and for TGF-alpha RT-PCR. The relative numbers of inflammatory cell types were in agreement with the literature 78. TGF-alpha mRNA, analyzed from crude SF cell samples, was detected in all patients tested (6/6). The amount of TGF-alpha measured in SF did not correlate to the relative abun- dance of either cell type indicating that the main producer of TGF-alpha was probably the synovium; the potential complementary production from e.g. macrophages and neutrophils was less important. We continued by collecting synovial biopsies from predominantly RA- patients undergoing synovectomy as a treatment for aggressive synovitis that was not responding to regular treatment. As control samples we received small pieces of macroscopically healthy synovia from patients undergoing explorative knee joint arthroscopy due to non-inflammatory conditions, most often post-traumatic or investigative. Concomitantly we collected new samples of SF from all these patients. Again the same pattern of TGF-alpha protein levels was found as in previous samples. Immunohistochemistry and in situ-hybridization performed on synovial biopsies demonstrated that TGF-alpha is produced in synovia, and predominantly in the outermost layer consisting of the monocytoid synoviocyte type A. TGF-alpha positive staining was more widespread and partly more intense in the thickened RA-synovia. We also performed CD68 immunostaining known to label phagocytes, especially macrophages, and synoviocyte type A. The patterns of TGF-alpha and CD68 immunostaining were almost completely overlapping revealing that the synovial hyperplasia in RA consists of both type A and type B synoviocytes as well as some infiltrating strongly CD68-positive macrophages. The CD68-positive macrophages were positive for TGF-alpha immunostaining irrespective of whether the macrophages were in the synovium or in the subsynovium intermingled with fibroblasts and blood vessels. HER-1 immunohistochemistry showed only a patchy expression of HER- 1 that was sometimes very faint. However, the HER-2 staining was prominent in synovium, subsynovium and also in the connective tissue in areas of dense vascularization. The cellular pattern was typical for that of a membrane-bound molecule present at the cell surface. The expression of HER-2 in control samples was surprising. Previously, HER-2 had been detected in rheumatoid synovia and not in synovia from patients with osteoarthritis or ligament injury likely to resemble our control material. Furthermore, primary cultures of synovial fibroblasts established from the same rheumatoid synovia samples were growth-inhibited by herceptin (HER-2 inhibitor) and by genistein (tyrosine-

60 kinase inhibitor), respectively 146 suggesting a growth-dependency on HER-2 activation. Concomitant expression of TGF-alpha and HER-2 has been shown to promote abnormal growth and also, with time, development of malignan- cies 45,147,148. When writing and submitting Paper III, new data were published showing HER-2 expression in synovial sarcomas 149,150. Some results suggest that TGF-alpha might interact with HER-2, others indicate a cooperation between TGF-alpha/HER-4/HER-2 independent of HER-1 74. We have not analyzed HER-4 in synovia, but cooperation between TGF-alpha and HER-2 would open new possibilities to explain the role of TGF-alpha in normal, abnormal and malignant synovial growth. There are many studies indicating functional links between the HER- system with respective ligands and CD44/HA linking primarily HER-2 and HER-1 to metastasis and matrix production (see background). The physical link described between HER-2 and CD44 is highly interesting in joint homeostasis where HA and TGF-alpha, the receptor activators, both are found in SF and seem to be induced by inflammatory cytokines. Co- immunoprecipitation of CD44/HER-2 has been found in several cancers but the nature of their cooperation is unknown, although they both phosphory- late the signaling intermediate Zap-70 in the downstream signaling cascade. While the synovial lining cells continuously produce HA for joint homeostasis, they also produce surfactant proteins A and D (SP-A, SP-D, see background). It is possible that TGF-alpha is important for the joint homeostasis e.g. through a negative influence on surfactant production; regulatory effects of TGF-alpha and HER-1 stimulation has been shown for lung surfactant production 82,93,94,151. The enzymatic activity releasing soluble TGF-alpha from membrane- bound precursors consists of TACE/ADAM17 49,152present in monocyte-like U-937-1 cells as well as in activated human elutriated monocytes 153. Specific inhibition of pro-TGF-alpha processing by certain peptide-hydroxamate metalloproteinase inhibitors is concomitant with blocked shedding of tumor necrosis factor-alpha, tumor necrosis factor-receptor 55, tumor necrosis factor- receptor 75, macrophage colony-stimulating factor (M-CSF), stem cell factor and IL-6 receptors 153. The potential benefit on the rheumatic inflammation by decreased shedding of tumor necrosis factor-alpha 109 makes the enzymatic mechanism highly interesting as a pharmaceutical target. Therefore it is of importance to study the physiological function of soluble TGF-alpha in the synovial joint. If the role of TGF-alpha is cartilage and/or synovial maintenance, there is a possibility of long-term complications, such as deterioration of synovium and/or cartilage, when treating patients with substances that would

61 decrease the amount of soluble TGF-alpha in SF. On the other hand, the effect of such a treatment could also be beneficial by reducing synovial hyperplasia, e.g. limiting a process contributing to cartilage breakdown and fibrosis, due to decreased availability of this molecule. The estrogen-responsive element in the TGF-alpha promoter region has been shown to induce a 2.5 – 3 fold increase in TGF-alpha transcription when treating hormone-sensitive breast cancer cells with estrogen in vitro 53. Patients receiving aromatase inhibitors as treatment for breast cancer with estrogen receptor-positive cells have greatly reduced levels of estrogen and often experience pain and stiffness in synovial joints. There are numerous ob- servations linking menopause to aggravated symptoms from joint disorders such as osteoarthritis and RA. It is tempting to relate low estrogen-levels to a possible down-regulation of TGF-alpha transcription causing a decreased synovial activity. This may possibly result in lack of surfactant proteins and/or other factors in SF important for joint homeostasis. TGF-alpha/HER-family in 4SS synovial sarcoma cells 4SS synovial sarcoma cells were used in order to study the TGF- alpha/HER-family signaling in synovial cells. These cells were known to have synovial origin and to produce hyaluronic acid. We could verify that the integrity of the cells and the culture characteristics were in agreement with previous results. The 4SS cells had not previously been phenotypically characterized which became our first goal (Paper IV). We could show that the cells continuously proliferated when cultured in serum-containing growth medium. In serum-free growth the cells continued to survive and proliferated slowly over 48 hours, but the long-term effect of only autonomous stimuli was not investigated. The growth pattern was very similar to that of synovial fibroblast primary cultures and there was some growth inhibition when the cells reached confluence 81. This resulted in the previously described honey-comb appearance where the confluent cultures continued proliferating by new discontinuous and piling layers of cells (fig. 1 in Paper IV). Furthermore, the 4SS cells produced considerable amounts of HA and some of this was attached to the cells as a viscous layer visible after removing the culture medium. We also characterized the 4SS phenotype by immunofluorescent labeling of certain surface markers and members of the HER-family (Paper IV). The presence of CD44 and small amounts of CD68 confirms the synovial origin together with the production of HA, and the described growth pattern. That the cells were devoid of CD36 and CD14, also acccounting for the low CD68

62 levels, suggests primarily a resemblance to synoviocyte type B (see table 1 in background). The expression of HER-family members was examined and revealed prominent production of HER-1, HER-2 and HER-4 as determined by surface immunofluorescence labeling (Paper IV). However, we could not detect HER- 3 by this method. Furthermore, TGF-alpha was detected both as a membrane- bound molecule and as s-TGF-alpha released into the medium measured by a highly sensitive and specific TGF-alpha ELISA (Paper IV). These results were all in agreement with our results concerning the presence of TGF-alpha, HER- 1 and HER-2 in synovial membrane in vivo as well as the s-TGF-alpha present in SF (Paper III). It is possible that TGF-alpha at the cell surface may stimulate neighboring cells in a paracrine manner 154. Although the 4SS cells are derived from a malignant tumor they seem to retain many of the characteristics specific for the original tissue. This includes adhesive growth and density- dependent inhibition of growth often lost during tumorigenesis. Our findings prompted us to examine the effects of TGF-alpha and EGF on the phosphorylation of HER-1 and HER-2, applying serum-free culture conditions to enhance the respective effects. To avoid distorsion of the results by endogenous growth factor production we first seeded the appropriate number of cells in complete medium for 24 hours allowing for adhesion and establishment of a non-confluent growing culture. We then changed to serum-free culture medium with the addition of the respective stimuli and harvested the cultures at different time end-points for protein analysis by SDS- PAGE and Western blot. The results showed an increase in HER-1 phosphorylation by both EGF and TGF-alpha, but a more prominent increase in HER-2 phosphorylation. Moreover, by adding blocking antibodies towards HER-1, TGF-alpha, HER-2, CD44 and HER-2+CD44 in the culture medium we could determine that phosphorylation of HER-2 was dependent on HER-1 phosphorylation. This was in agreement with many previous studies showing that TGF-alpha and EGF are HER-1 specific ligands (see background). We also suggest that although HER-4 seems to be present on 4SS cells, there is no autonomous production of heregulins or other HER-4-stimulating ligands since HER-2 phosphorylation was strongly inhibited by anti-HER-1 pre- treatment (Paper IV). The effect of anti-CD44 was principally inconclusive. However, a slight difference in HER-2 phosphorylation was observed comparing the effects of TGF-alpha and EGF stimulation, respectively, concomitant with anti-CD44. This suggests that TGF-alpha is more potent in inducing HER-2 phosphorylation in 4SS cells. Possibly this is related to a persistence of ligand-receptor ac-

63 tivation and slower down-regulation associated with TGF-alpha mediated stimulation of HER-2 74. Wild-type and alternatively spliced TGF-alpha precursors have been shown to have different affinity for the different HER-variants suggesting an alternative activation or signal transfer involving HER-2 and HER-4 61. The TGF-alpha produced in 4SS cells and in vivo synovia has not been further characterized and the presence of HER-4 in synovia has not yet been analyzed. The 4SS cell line is an interesting model system for studies on endogenous HER-family and TGF-alpha expression. The previous results showing intracellular signal transfer through the COOH-terminal part of membrane- associated pro-TGF-alpha were achieved using genetically transformed cells devoid of endogenous pro-TGF-alpha 60. The 4SS cell line may present a more complete set of intracellular signaling intermediates transferring a signal generated in the TGF-alpha COOH-terminal part, and therefore may also reveal more about its possible effects. These cells exhibit at least 3 of 4 HER-family members as well as CD44 presenting opportunities to study the CD44/HER- 2/HER-4 interactions. The 4SS cell line could serve as a model for synoviocytes when studying the role of the HER-family in synovial proliferation and production of SF components. Also, along with previous publications we have shown the importance of studying HER-family receptors and ligands in synovial sarcoma tumors. HER-2 expression may participate in the development of this type of cancer since the SYT-SSX1 translocation has been associated with increased HER-2 production 113,149,150.

64 CONCLUSIONS The expression of TGF-alpha in undifferentiated as well as differentiated monocytoid U-937-1 cells is detectable both as membrane-bound and soluble TGF-alpha. This finding may reflect a general occurrence of TGF-alpha in monocyte/macrophage type cells. PMA-differentiated U-937-1 cells resembling secretory macrophages have an increased release of TGF-alpha, and have functional IL-6 receptors mediating upregulation of TGF-alpha mRNA. This suggests a possible increase in TGF-alpha production in monocytes/macrophages in acute inflammation. Short exposure to physiologic UVB-doses influences mRNA expression and surface detection of HER-1 in normal human melanocytes in a biphasic pattern. The final upregulation of steady-state HER-1 mRNA and reconstitution of surface HER-1 possibly reflects an increased turnover of HER-1. This would suggest a resulting long-term increase in HER-1 activation in these cells. The distribution of TGF-alpha mRNA in synovia according to in situ- hybridization analysis indicates that TGF-alpha is mainly produced by synoviocyte type A. The presence of TGF-alpha and HER-2 protein in normal synovia and the increased detection of both in hyperplastic inflammatory synovia implicates a role for TGF-alpha/HER-family in synovial membrane maintenance and in synovitis. Infiltrating macrophages in synovitis produce excess amounts of TGF- alpha mRNA and protein. This may contribute to the synovial hyperplasia. The 4SS synovial sarcoma cell line retains many synoviocyte characteristics and therefore may function as a model for synoviocytes, including studies of TGF-alpha/HER-family and their possible relations to CD44/HA. The demonstrated HER-2 phosphorylation in 4SS cells is dependent on primary HER-1 activation. Consequently, HER-2 activation is achieved by secondary dimerization with ligand-bound HER-1. CD44 may participate in enforcing HER-2 phosphorylation in 4SS cells, in agreement with analogous results in other cell systems.

65 ACKNOWLEDGEMENTS

The work presented in this thesis was initiated at the Department of Cellbiology, which later became part of the Department of Biomedicine and Surgery, and now I finally finished this project at the Department of Clinical and Experimental Medicine, still working in the same lab at “Cellbiologen”! Many persons made this thesis possible and some of you have been especially important and will hereby receive special thanks: Professor Åke Wasteson, my supervisor, scientific tutor and mentor. Un- fortunately I have not been the only one in the need of your extra-ordinary qualities, but you have put a true meaning into the expression quality time! The sessions with you have given new insights into my “insignificant” results. You encouraged me to control my model systems, and taught me the need to know all about my starting material. Also, you were (almost) always positive to all academic and semi-political side-occupations that stole me away from the lab from time to time. Docent Thomas Walz, assisting tutor, senior oncology specialist and head of the Oncology clinic. This is all your fault! You lured me into the lab for the first time when I was in my second year of medical school and got me started at the lab bench. Your good spirit, energy and enthusiasm was, and still is, contagious. Our paths take similar directions and keep crossing in fruitful ways. Associate professor Annelie Lindström, assisting tutor and true friend. The addition of your positive, encouraging and lab-oriented tutorship half- way through my thesis work had perfect timing, and you are always available in one way or another. I really enjoy both the scientific and the trivial discus- sions and your company. Ms. Kristina Briheim and Ms. Kerstin Willander for never-failing prac- tical and emotional support, lab-work can be both boring and extremely fru- strating. Your technical skills and support have been highly appreciated, as well as your friendship. Kristina also made important contributions by performing the in situ-hybridizations and working in parallel with me in the cell culture room and in the lab. Dr Margareta Lirvall, a long time fellow PhD-student. Our fruitful cooperation in the lab resulted in Paper II. You have impressed me by your persistence and determination to “have it all”.

66 Dr Avni Adiu, an especially nice person to have around and with a healthy relationship to science. Devoted clinician - I will turn to you when my wrinkles become too deep! Fellow PhD-students Cilla, Emelie and Maria for good companionship and great help, especially Emelie, in delivering recommendations on protein analysis, immunoblotting conditions and generosity in preparing buffers when my time was consumed by clinical work. All of you at level 13, especially Anita and Anita, who made, and still make, this place such a helpful and friendly environment. Ms. Lena Ceder- gren and her colleagues for endlessly supplying utensils and RNAse-free material with a smile when visiting your haven at level 10. Through the years I have gratefully received contributions from many friends and colleagues at “Cellbiologen”, other research units at the Faculty of Health Sciences and at Linköping University hospital. None mentioned, none forgotten. Friends and relatives putting up with rare telephone calls and even more rare visits when life outside my little family became all science and medicine. Ylva, Robin and Joachim; Johanna, Tommy and Hugo; Niklas, Cecilia, Agnes, Simon and Elsa; you are always there and have shown great patience, support and understanding for this seemingly endless project. The tireless support from my dear parents-in-law, Sven-Eric and Gittan, for always being there for us, when Oscar and Theodor needed more time than Martin and I could muster and when we needed to “get away from it all”. You are truly generous and great persons. My dear parents for providing me with the qualities and self-esteem needed to reach my goals and to finish this thesis. You are inspiring in very different ways, complementing one another. My little family, I am so grateful for my good life with you. Martin, my love and lifetime companion; thank you for everything, now the future looks bright. Oscar and Theodor, our wonderful sons that bring so much joy and happiness: Nu är “den himla boken” klar! This work was supported by grants from Landstingets i Östergötland kommitté för medicinsk forskning och utveckling (ÖLL), Forskningsrådet i Sydöstra Sverige (FORSS nr 2000-122), Medicinska Forskningsrådet (MFR, anslag nr 4486), Lions forskningsfond vid Medicinska fakulteten i Linköping samt Polysackaridforskning i Uppsala AB.

67 REFERENCES 1. Luetteke, N.C., et al. Tgf-Alpha Deficiency Results in Hair Follicle and Eye Ab- normalities in Targeted and Waved-1 Mice. Cell 73, 263-278 (1993). 2. Mann, G.B., et al. Mice with a Null Mutation of the Tgf-Alpha Gene Have Ab- normal Skin Architecture, Wavy Hair, and Curly Whiskers and Often Develop Corneal Inflammation. Cell 73, 249-261 (1993). 3. Troyer, K.L., et al. Growth retardation, duodenal lesions, and aberrant ileum architecture in triple null mice lacking EGF, amphiregulin, and TGF-alpha. Gastro- enterology 121, 68-78 (2001). 4. Vieira, A.R. Association between the transforming growth factor alpha gene and nonsyndromic oral clefts: a HuGE review. Am J Epidemiol 163, 790-810 (2006). 5. Hansen, J.M. & Harris, C. A novel hypothesis for thalidomide-induced limb tera- togenesis: redox misregulation of the NF-kappaB pathway. Antioxid Redox Signal 6, 1-14 (2004). 6. Denef, C. Paracrine control of lactotrope proliferation and differentiation. Trends in Endocrinology and Metabolism 14, 188-195 (2003). 7. Gilchrest, B.A., Eller, M.S., Geller, A.C. & Yaar, M. The pathogenesis of melanoma induced by ultraviolet radiation. N Engl J Med 340, 1341-1348 (1999). 8. Zhuang, L., Wang, B. & Sauder, D.N. Molecular mechanism of ultraviolet- induced keratinocyte apoptosis. J Interferon Cytokine Res 20, 445-454 (2000). 9. Kadekaro, A.L., et al. Cutaneous photobiology. The melanocyte vs. the sun: who will win the final round? Pigment Cell Res 16, 434-447 (2003). 10. Jost, M., Kari, C. & Rodeck, U. The EGF receptor - an essential regulator of multiple epidermal functions. Eur J Dermatol 10, 505-510 (2000). 11. Atillasoy, E.S., et al. UVB induces atypical melanocytic lesions and melanoma in human skin. Am J Pathol 152, 1179-1186 (1998). 12. Daya-Grosjean, L., Dumaz, N. & Sarasin, A. The specificity of p53 mutation spectra in sunlight induced human cancers. J Photochem Photobiol B 28, 115-124 (1995). 13. Monti, M. & Motta, S. Clinical management of cutaneous toxicity of anti-EGFR agents. Int J Biol Markers 22, S53-61 (2007). 14. Walz, T.M., Malm, C. & Wasteson, A. Expression of the transforming growth factor alpha protooncogene in differentiating human promyelocytic leukemia (HL-60) cells. Cancer Research 53, 191-196 (1993). 15. Walz, T.M., Nishikawa, B.K., Malm, C., Briheim, K. & Wasteson, A. Transform- ing growth factor alpha expression in normal human blood eosinophils: differential regulation by granulocyte-macrophage colony-stimulating factor and interleukin-3. Leukemia 8, 612-619 (1994). 16. Sunderkotter, C., Steinbrink, K., Goebeler, M., Bhardwaj, R. & Sorg, C. Macro- phages and angiogenesis. J Leukoc Biol 55, 410-422 (1994). 17. Madtes, D.K., et al. Induction of transforming growth factor-alpha in activated human alveolar macrophages. Cell 53, 285-293 (1988).

68 18. Calafat, J., et al. Human monocytes and neutrophils store transforming growth factor-alpha in a subpopulation of cytoplasmic granules. Blood 90, 1255-1266 (1997). 19. Rappolee, D.A., Mark, D., Banda, M.J. & Werb, Z. Wound macrophages express TGF-alpha and other growth factors in vivo: analysis by mRNA phenotyping. Science 241, 708-712 (1988). 20. auf demKeller, U., Krampert, M., Kumin, A., Braun, S. & Werner, S. Keratino- cyte growth factor: effects on keratinocytes and mechanisms of action. Eur J Cell Biol 83, 607-612 (2004). 21. Finch, P.W. & Rubin, J.S. Keratinocyte growth factor/fibroblast growth factor 7, a homeostatic factor with therapeutic potential for epithelial protection and repair. Adv Cancer Res 91, 69-136 (2004). 22. Werner, S., Krieg, T. & Smola, H. Keratinocyte-Fibroblast Interactions in Wound Healing. J Invest Dermatol 127, 998-1008. 23. Grose, R. & Werner, S. Wound-healing studies in transgenic and knockout mice. Mol Biotechnol 28, 147-166 (2004). 24. Barrandon, Y. & Green, H. Cell migration is essential for sustained growth of keratinocyte colonies: The roles of transforming growth factor-[alpha] and epidermal growth factor. Cell 50, 1131-1137 (1987). 25. Schultz, G.S., et al. Epithelial wound healing enhanced by transforming growth factor-alpha and vaccinia growth factor. Science 235, 350-352 (1987). 26. Campaner, A.B., et al. Upregulation of TGF-[beta]1 Expression May Be Neces- sary but Is Not Sufficient for Excessive Scarring. J Invest Dermatol 126, 1168- 1176 (2006). 27. Robles, D.T. & Berg, D. Abnormal wound healing: keloids. Clinics in dermatol- ogy 25, 26-32 (2006). 28. Schreiber, A.B., Winkler, M.E. & Derynck, R. Transforming growth factor-alpha: a more potent angiogenic mediator than epidermal growth factor. Science 232, 1250-1253 (1986). 29. Grotendorst, G.R., Soma, Y., Takehara, K. & Charette, M. EGF and TGF-alpha are potent chemoattractants for endothelial cells and EGF-like peptides are present at sites of tissue regeneration. J Cell Physiol 139, 617-623 (1989). 30. Wen, X.F., et al. HER2 signaling modulates the equilibrium between pro- and antiangiogenic factors via distinct pathways: implications for HER2-targeted antibody therapy. Oncogene 25, 6986-6996 (2006). 31. Izumi, Y., Xu, L., di Tomaso, E., Fukumura, D. & Jain, R.K. Tumour biology: Herceptin acts as an anti-angiogenic cocktail. Nature 416, 279-280 (2002). 32. Yen, L., et al. Differential Regulation of Tumor Angiogenesis by Distinct ErbB Homo- and Heterodimers. Mol. Biol. Cell 13, 4029-4044 (2002). 33. Stiewe, T. The p53 family in differentiation and tumorigenesis. Nat Rev Cancer 7, 165-167 (2007). 34. Mueller, M.M. & Fusenig, N.E. Friends or foes - bipolar effects of the tumour stroma in cancer. Nat Rev Cancer 4, 839-849 (2004).

69 35. Pelham, R.J., et al. Identification of alterations in DNA copy number in host stromal cells during tumor progression. Proc Natl Acad Sci U S A 103, 19848- 19853 (2006). 36. Roberts, A.B., et al. Transforming growth factors: isolation of polypeptides from virally and chemically transformed cells by acid/ethanol extraction. Proc Natl Acad Sci U S A 77, 3494-3498 (1980). 37. Sporn, M.B. & Todaro, G.J. Autocrine secretion and malignant transformation of cells. N Engl J Med 303, 878-880 (1980). 38. Todaro, G.J., Fryling, C. & De Larco, J.E. Transforming growth factors produced by certain human tumor cells: polypeptides that interact with epidermal growth factor receptors. Proc Natl Acad Sci U S A 77, 5258-5262 (1980). 39. Todaro, G.J. & De Larco, J.E. Growth factors produced by sarcoma virus- transformed cells. Cancer Res 38, 4147-4154 (1978). 40. Sandgren, E.P., et al. Transforming growth factor alpha dramatically enhances oncogene-induced carcinogenesis in transgenic mouse pancreas and liver. Mol Cell Biol 13, 320-330 (1993). 41. Babyatsky, M.W., Rossiter, G. & Podolsky, D.K. Expression of transforming growth factor alpha and beta in colonic mucosa in inflammatory bowel disease. Gastroenterology 110, 975-984 (1996). 42. Wagner, M., Luhrs, H., Kloppel, G., Adler, G. & Schmid, R.M. Malignant transformation of duct-like cells originating from acini in transforming growth factor transgenic mice. Gastroenterology 115, 1254-1262 (1998). 43. Thorgeirsson, S.S., Factor, V.M. & Snyderwine, E.G. Transgenic mouse models in carcinogenesis research and testing. Toxicology Letters 112-113, 553-555 (2000). 44. Rose-Hellekant, T.A. & Sandgren, E.P. Transforming growth factor alpha- and c- myc-induced mammary carcinogenesis in transgenic mice. Oncogene 19, 1092- 1096 (2000). 45. Muller, W.J., et al. Synergistic interaction of the Neu proto-oncogene product and transforming growth factor alpha in the mammary epithelium of transgenic mice. Mol Cell Biol 16, 5726-5736 (1996). 46. National Library of Medicine. Intercellular Signaling Peptides and Proteins. in In PubMed: MESH database (National Library of Medicine, NIH, Bethesda, MD, USA, 2003). 47. National Library of Medicine. Cytokine. in In PubMed: MESH database (Nation- al Library of Medicine, NIH, Bethesda, MD, USA, 1991). 48. Cohen, S. Alfred P. Sloan Jr. Award. The epidermal growth factor (EGF). Cancer 51, 1787-1791 (1983). 49. Lee, D.C., Fenton, S.E., Berkowitz, E.A. & Hissong, M.A. Transforming growth factor alpha: expression, regulation, and biological activities. Pharmacological Reviews 47, 51-85 (1995). 50. Harris, R.C., Chung, E. & Coffey, R.J. EGF receptor ligands. Exp Cell Res 284, 2-13 (2003). 51. Junier, M.P. What role(s) for TGFalpha in the central nervous system? Prog Neu- robiol 62, 443-473 (2000).

70 52. Chen, X., Azizkhan, J.C. & Lee, D.C. The binding of transcription factor Sp1 to multiple sites is required for maximal expression from the rat transforming growth factor alpha promoter. Oncogene 7, 1805-1815 (1992). 53. El-Ashry, D., Chrysogelos, S.A., Lippman, M.E. & Kern, F.G. Estrogen induction of TGF-alpha is mediated by an estrogen response element composed of two imperfect palindromes. J Steroid Biochem Mol Biol 59, 261-269 (1996). 54. Shin, T.H., Paterson, A.J. & Kudlow, J.E. p53 stimulates transcription from the human transforming growth factor alpha promoter: a potential growth-stimulatory role for p53. Mol Cell Biol 15, 4694-4701 (1995). 55. Wang, D., Shin, T.H. & Kudlow, J.E. Transcription factor AP-2 controls transcription of the human transforming growth factor-alpha gene. J Biol Chem 272, 14244-14250 (1997). 56. Wang, D. & Kudlow, J.E. Purification and characterization of TEF1, a transcription factor that controls the human transforming growth factor-alpha promoter. Biochim Biophys Acta 1449, 50-62 (1999). 57. Luetteke, N.C., et al. Characterization of high molecular weight transforming growth factor alpha produced by rat hepatocellular carcinoma cells. Biochemistry 27, 6487-6494 (1988). 58. Borrell-Pages, M., Rojo, F., Albanell, J., Baselga, J. & Arribas, J. TACE is required for the activation of the EGFR by TGF-alpha in tumors. The EMBO journal 22, 1114-1124 (2003). 59. Shum, L., Reeves, S.A., Kuo, A.C., Fromer, E.S. & Derynck, R. Association of the transmembrane TGF-alpha precursor with a protein kinase complex. Journal of Cell Biology 125, 903-916 (1994). 60. Shum, L., Turck, C.W. & Derynck, R. Cysteines 153 and 154 of transmembrane transforming growth factor-alpha are palmitoylated and mediate cytoplasmic protein association. Journal of Biological Chemistry 271, 28502-28508 (1996). 61. Xu, X., Kelleher, K.F., Liao, J., Creek, K.E. & Pirisi, L. Unique carboxyl- terminal sequences of wild type and alternatively spliced variant forms of transforming growth factor-alpha precursors mediate specific interactions with ErbB4 and ErbB2. Oncogene 19, 3172-3181 (2000). 62. Chalazonitis, A., Kessler, J.A., Twardzik, D.R. & Morrison, R.S. Transforming growth factor alpha, but not epidermal growth factor, promotes the survival of sensory neurons in vitro. Journal of Neuroscience 12, 583-594 (1992). 63. Playford, R.J., et al. Comparison of the effects of transforming growth factor alpha and epidermal growth factor on gastrointestinal proliferation and hormone release. Digestion 57, 362-367 (1996). 64. Ibbotson, K.J., et al. Stimulation of bone resorption in vitro by synthetic transforming growth factor-alpha. Science 228, 1007-1009 (1985). 65. Stern, P.H., et al. Human transforming growth factor-alpha stimulates bone resorption in vitro. J Clin Invest 76, 2016-2019 (1985). 66. Jorissen, R.N., et al. Epidermal growth factor receptor: mechanisms of activation and signalling. Exp Cell Res 284, 31-53 (2003). 67. Yarden, Y. & Sliwkowski, M.X. Untangling the ErbB signalling network. Nat Rev Mol Cell Biol 2, 127-137 (2001).

71 68. Citri, A. & Yarden, Y. EGF-ERBB signalling: towards the systems level. Nat Rev Mol Cell Biol 7, 505-516 (2006). 69. Bublil, E.M. & Yarden, Y. The EGF receptor family: spearheading a merger of signaling and therapeutics. Current Opinion in Cell Biology 19, 124-134 (2007). 70. Jones, R.B., Gordus, A., Krall, J.A. & MacBeath, G. A quantitative protein interaction network for the ErbB receptors using protein microarrays. Nature 439, 168-174 (2006). 71. Jorissen, R.N., Treutlein, H.R., Epa, V.C. & Burgess, A.W. Modeling the epidermal growth factor -- epidermal growth factor receptor l2 domain interaction: implications for the ligand binding process. J Biomol Struct Dyn 19, 961-972 (2002). 72. Keilhack, H., et al. Phosphotyrosine 1173 mediates binding of the protein- tyrosine phosphatase SHP-1 to the epidermal growth factor receptor and attenuation of receptor signaling. J Biol Chem 273, 24839-24846 (1998). 73. Tran, K.T., Rusu, S.D., Satish, L. & Wells, A. Aging-related attenuation of EGF receptor signaling is mediated in part by increased protein tyrosine phosphatase activity. Exp Cell Res 289, 359-367 (2003). 74. Valabrega, G., et al. TGFalpha expression impairs Trastuzumab-induced HER2 downregulation. Oncogene 24, 3002-3010 (2005). 75. Citri, A., Skaria, K.B. & Yarden, Y. The deaf and the dumb: the biology of ErbB- 2 and ErbB-3. Exp Cell Res 284, 54-65 (2003). 76. Carpenter, G. ErbB-4: mechanism of action and biology. Exp Cell Res 284, 66-77 (2003). 77. Jones, J.T., et al. Binding interaction of the heregulinbeta egf domain with ErbB3 and ErbB4 receptors assessed by alanine scanning mutagenesis. J Biol Chem 273, 11667-11674 (1998). 78. Kelley´s Textbook of Rheumatology, (W.B. Saunders Company, Philadelphia, 2001). 79. Edwards, J.C. The nature and origins of synovium: experimental approaches to the study of synoviocyte differentiation. J Anat 184, 493-501 (1994). 80. Dobbie, J.W., et al. Lamellar body secretion: ultrastructural analysis of an unex- plored function of synoviocytes. Br J Rheumatol 34, 13-23 (1995). 81. Vandenabeele, F., et al. Morphological and immunocytochemical characterization of cultured fibroblast-like cells derived from adult human synovial membrane. Arch Histol Cytol 66, 145-153 (2003). 82. Kankavi, O. Increased expression of surfactant protein A and D in rheumatoid arthritic synovial fluid (RASF). Croat Med J 47, 155-161 (2006). 83. Kankavi, O. & Roberts, M.S. Detection of surfactant protein A (SP-A) and surfactant protein D (SP-D) in equine synovial fluid with immunoblotting. Can J Vet Res 68, 146-149 (2004). 84. Schwachula, A., Riemann, D., Kehlen, A. & Langner, J. Characterization of the immunophenotype and functional properties of fibroblast-like synoviocytes in comparison to skin fibroblasts and umbilical vein endothelial cells. Immunobiolo- gy 190, 67-92 (1994).

72 85. Dobbie, J.W., et al. Lamellar bodies in synoviocytes, mesothelium and specific epithelia as possible site of auto-antigen in rheumatoid disease. Br J Rheumatol 33, 508-519 (1994). 86. Riemann, D., et al. Caveolae/lipid rafts in fibroblast-like synoviocytes: ectopepti- dase-rich membrane microdomains. Biochem J 354, 47-55 (2001). 87. Engstrom-Laurent, A. Hyaluronan in joint disease. J Intern Med 242, 57-60 (1997). 88. Schwarz, I.M. & Hills, B.A. Synovial surfactant: lamellar bodies in type B synoviocytes and proteolipid in synovial fluid and the articular lining. Br J Rheuma- tol 35, 821-827 (1996). 89. Hills, B.A. & Butler, B.D. Surfactants identified in synovial fluid and their ability to act as boundary lubricants. Ann Rheum Dis 43, 641-648 (1984). 90. Bourbon, J.R. & Chailley-Heu, B. Surfactant proteins in the digestive tract, me- sentery, and other organs: evolutionary significance. Comp Biochem Physiol A Mol Integr Physiol 129, 151-161 (2001). 91. Hills, B.A. Surface-active phospholipid: a Pandora's box of clinical applications. Part II. Barrier and lubricating properties. Intern Med J 32, 242-251 (2002). 92. Wright, J.R. Immunoregulatory functions of surfactant proteins. Nat Rev Immunol 5, 58-68 (2005). 93. Whitsett, J.A., Weaver, T.E., Lieberman, M.A., Clark, J.C. & Daugherty, C. Dif- ferential effects of epidermal growth factor and transforming growth factor-beta on synthesis of Mr = 35,000 surfactant-associated protein in fetal lung. J Biol Chem 262, 7908-7913 (1987). 94. Klein, J.M., McCarthy, T.A., Dagle, J.M. & Snyder, J.M. Antisense inhibition of epidermal growth factor receptor decreases expression of human surfactant protein A. Am J Respir Cell Mol Biol 22, 676-684 (2000). 95. Aruffo, A., Stamenkovic, I., Melnick, M., Underhill, C.B. & Seed, B. CD44 is the principal cell surface receptor for hyaluronate. Cell 61, 1303-1313 (1990). 96. Naor, D. & Nedvetzki, S. CD44 in rheumatoid arthritis. Arthritis Research & Therapy 5, 105-115 (2003). 97. Naor, D., Nedvetzki, S., Golan, I., Melnik, L. & Faitelson, Y. CD44 in cancer. Critical Reviews in Clinical Laboratory Sciences 39, 527-579 (2002). 98. Wasteson, A., Westermark, B., Lindahl, U. & Ponten, J. Aggregation of feline lymphoma cells by hyaluronic acid. Int J Cancer 12, 169-178. (1973). 99. Knudson, W. & Loeser, R.F. CD44 and integrin matrix receptors participate in cartilage homeostasis. Cellular and Molecular Life Sciences (CMLS) 59, 36-44 (2002). 100. Nedvetzki, S., et al. CD44 involvement in experimental collagen-induced arthritis (CIA). J Autoimmun 13, 39-47 (1999). 101. Nedvetzki, S., et al. A mutation in a CD44 variant of inflammatory cells enhances the mitogenic interaction of FGF with its receptor. Journal of Clinical Investiga- tion 111, 1211-1220 (2003). 102. Zhang, M., Singh, R.K., Wang, M.H., Wells, A. & Siegal, G.P. Epidermal growth factor modulates cell attachment to hyaluronic acid by the cell surface glycoprotein CD44. Clin Exp Metastasis 14, 268-276 (1996). 73 103. Monaghan, M., et al. Epidermal growth factor up-regulates CD44-dependent astrocytoma invasion in vitro. J Pathol 192, 519-525 (2000). 104. Tsatas, D., Kanagasundaram, V., Kaye, A. & Novak, U. EGF receptor modifies cellular responses to hyaluronan in glioblastoma cell lines. J Clin Neurosci 9, 282-288 (2002). 105. Bourguignon, L.Y., Gilad, E. & Peyrollier, K. Heregulin-mediated ErbB2-ERK signaling activates hyaluronan synthases leading to CD44-dependent ovarian tumor cell growth and migration. J Biol Chem (2007). 106. Levick, J.R., Price, F.M. & Mason, R.M. Synovial matrix-Synovial fluid system of joints. in Extracellular matrix: Tissue function, Vol. 1 (ed. Comper, W.D.) 328 - 377 (Harwood Academic Publishers GmbH, Amsterdam, 1996). 107. Vuorio, E. Rheumatoid disease in cultured human synovial cells. A biochemical study on glycosaminoglycans, proteins and plasma membranes of synovial fibroblasts in culture. Scand J Clin Lab Invest Suppl, 1-72 (1977). 108. Hitchon, C.A. & El-Gabalawy, H.S. The histopathology of early synovitis. Clini- cal and Experimental Rheumatology 21, S28-S36 (2003). 109. Firestein, G. Etiology and Pathogenesis of Rheumatoid Arthritis. in Kelley´s Textbook of Rheumatology, Vol. 1 (eds. Ruddy, S., Harris, E. & Sledge, C.) 921- 966 (W.B. Saunders Company, Philadelphia, 2001). 110. Sen, M., et al. Expression and function of wingless and frizzled homologs in rheumatoid arthritis. Proc Natl Acad Sci U S A 97, 2791-2796 (2000). 111. Neidhart, M., Gay, R.E. & Gay, S. Anti-interleukin-1 and anti-CD44 interven- tions producing significant inhibition of cartilage destruction in an in vitro model of cartilage invasion by rheumatoid arthritis synovial fibroblasts. Arthritis Rheum 43, 1719-1728 (2000). 112. Sampson, P.M., Rochester, C.L., Freundlich, B. & Elias, J.A. Cytokine regulation of human lung fibroblast hyaluronan (hyaluronic acid) production. Evidence for cytokine-regulated hyaluronan (hyaluronic acid) degradation and human lung fibroblast-derived hyaluronidase. J Clin Invest 90, 1492-1503 (1992). 113. Allander, S.V., et al. Expression profiling of synovial sarcoma by cDNA microarrays: association of ERBB2, IGFBP2, and ELF3 with epithelial differentiation. Am J Pathol 161, 1587-1595 (2002). 114. Ponten, J. & Saksela, E. Two established in vitro cell lines from human mesenchymal tumours. Int J Cancer 2, 434-447 (1967). 115. Phan Thu, A., Zajac, V., Kuliffay, P. & Popovic, M. Karyological and isoenzyme characterization of established human sarcoma cell lines. Neoplasma 27, 557-566 (1980). 116. Oberg, F., Botling, J. & Nilsson, K. Macrophages and the cytokine network. Transplant Proc 25, 2044-2047 (1993). 117. Oberg, F. & Nilsson, K. Differentiation and activation associated expression of IL-6 and IL-6 receptors in U-937 monocytic cells: relationship to the expression of CD14. Growth factors (Chur, Switzerland) 7, 85-96 (1992). 118. Walz, T.M., et al. Production of transforming growth factor alpha by human leukemia cells (HL-60 and U-937) during monocytic differentiation. Leukemia 9, 671-676 (1995).

74 119. Chomczynski, P. & Sacchi, N. Single-step method of RNA isolation by acid guanidinium thiocyanate- phenol-chloroform extraction. Anal Biochem 162, 156-159 (1987). 120. Hallbeck, A.-L., Walz, T.M. & Wasteson, A. Interleukin-6 enhances Transform- ing Growth Factor-alpha mRNA expression in macrophage-like human monocytoid (U-937-1) cells. Bioscience Reports 21, 325-339 (2001). 121. Derynck, R., Roberts, A.B., Winkler, M.E., Chen, E.Y. & Goeddel, D.V. Human transforming growth factor-alpha: precursor structure and expression in E. coli. Cell 38, 287-297 (1984). 122. Zhu, J.-Q., et al. Recombinant human granulocyte macrophage colony- stimulating factor (rhGM-CSF) induces human macrophage production of transforming growth factor-alpha. Cell Mol Biol 37, 413-419 (1991). 123. Brach, M.A., Sott, C., Kiehntopf, M. & Herrmann, F. Expression of the transforming growth factor-alpha gene by human eosinophils is regulated by interleukin-3, interleukin-5, and granulocyte-macrophage colony-stimulating factor. Eur J Immunol 24, 646-650 (1994). 124. Shearer, W.T., Rosenwasser, L.J., Bochner, B.S., Martinez-Moczygemba, M. & Huston, D.P. Biology of common [beta] receptor-signaling cytokines: IL-3, IL-5, and GM-CSF. Journal of Allergy and Clinical Immunology 112, 653-665 (2003). 125. Rose-John, S., Scheller, J., Elson, G. & Jones, S.A. Interleukin-6 biology is coor- dinated by membrane-bound and soluble receptors: role in inflammation and cancer. J Leukoc Biol 80, 227-236 (2006). 126. Monier, S., et al. Growth factor activity of IL-6 in the synovial fluid of patients with rheumatoid arthritis. Clin Exp Rheumatol 12, 595-602 (1994). 127. Sawamura, D., et al. Induction of keratinocyte proliferation and lymphocytic infiltration by in vivo introduction of the IL-6 gene into keratinocytes and possibility of keratinocyte gene therapy for inflammatory skin diseases using IL-6 mutant genes. J Immunol 161, 5633-5639 (1998). 128. Schafer, K.H., Mestres, P., Marz, P. & Rose-John, S. The IL-6/sIL-6R fusion protein hyper-IL-6 promotes neurite outgrowth and neuron survival in cultured enter- ic neurons. J Interferon Cytokine Res 19, 527-532 (1999). 129. Ogata, A., et al. IL-6 triggers cell growth via the Ras-dependent mitogen- activated protein kinase cascade. J Immunol 159, 2212-2221 (1997). 130. Paulsson, Y., et al. Growth factors induce early pre-replicative changes in senes- cent human fibroblasts. The EMBO journal 5, 2157-2162 (1986). 131. Kondo, S. The roles of keratinocyte-derived cytokines in the epidermis and their possible responses to UVA-irradiation. J Investig Dermatol Symp Proc 4, 177- 183 (1999). 132. Mattei, S., et al. Expression of cytokine/growth factors and their receptors in human melanoma and melanocytes. Int J Cancer 56, 853-857 (1994). 133. Nanney, L.B., Coffey, R.J., Jr. & Ellis, D.L. Expression and distribution of transforming growth factor-alpha within melanocytic lesions. J Invest Dermatol 103, 707-714 (1994). 134. Ellem, K.A., Cullinan, M., Baumann, K.C. & Dunstan, A. UVR induction of TGF alpha: a possible autocrine mechanism for the epidermal melanocytic response

75 and for promotion of epidermal carcinogenesis. Carcinogenesis 9, 797-801 (1988). 135. Gordon-Thomson, C., Mason, R.S. & Moore, G.P. Regulation of epidermal growth factor receptor expression in human melanocytes. Exp Dermatol 10, 321- 328 (2001). 136. Mirmohammadsadegh, A., et al. Constitutive expression of epidermal growth factor receptors on normal human melanocytes. J Invest Dermatol 125, 392-394 (2005). 137. de Leeuw, S.M., et al. The UV action spectra for the clone-forming ability of cultured human melanocytes and keratinocytes. Photochem Photobiol 59, 430-436 (1994). 138. Coffey, R.J., Jr., et al. Production and auto-induction of transforming growth factor-alpha in human keratinocytes. Nature 328, 817-820 (1987). 139. Sutton, R., Gordon-Thomson, C., Cree, I.A., Mason, R.S. & Moore, G.P. Tyr- TGFalpha transgenic mice develop ocular melanocytic lesions. Melanoma Res 12, 435-439 (2002). 140. Gordon-Thomson, C., Jones, J., Mason, R.S. & Moore, G.P. ErbB receptors mediate both migratory and proliferative activities in human melanocytes and melanoma cells. Melanoma Res 15, 21-28 (2005). 141. Stove, C., et al. The heregulin/human epidermal growth factor receptor as a new growth factor system in melanoma with multiple ways of deregulation. J Invest Dermatol 121, 802-812 (2003). 142. Hiraoka, K., Sasaguri, Y., Komiya, S., Inoue, A. & Morimatsu, M. Cell proliferation-related production of matrix metalloproteinases 1 (tissue collagenase) and 3 (stromelysin) by cultured human rheumatoid synovial fibroblasts. Biochem Int 27, 1083-1091 (1992). 143. Shiozawa, S., et al. Human epidermal growth factor for the stratification of synovial lining layer and neovascularisation in rheumatoid arthritis. Annals of the Rheumatic Diseases 48, 820-828 (1989). 144. Kusada, J., et al. Immuno-reactive human epidermal growth factor (h-EGF) in rheumatoid synovial fluids. Nippon Seikeigeka Gakkai Zasshi 67, 859-865 (1993). 145. Skoog, V., Widenfalk, B., Ohlsen, L. & Wasteson, A. The effect of growth factors and synovial fluid on chondrogenesis in perichondrium. Scand J Plast Re- constr Surg Hand Surg 24, 89-95 (1990). 146. Satoh, K., et al. Involvement of ErbB-2 in rheumatoid synovial cell growth. Arth- ritis Rheum 44, 260-265 (2001). 147. Sandgren, E.P., Luetteke, N.C., Palmiter, R.D., Brinster, R.L. & Lee, D.C. Over- expression of TGF alpha in transgenic mice: induction of epithelial hyperplasia, pancreatic metaplasia, and carcinoma of the breast. Cell 61, 1121-1135 (1990). 148. Hamburger, A.W., Fernandes, A., Murakami, M. & Gerwin, B.I. The role of transforming growth factor alpha production and ErbB-2 overexpression in induction of tumorigenicity of lung epithelial cells. Br J Cancer 77, 1066-1071 (1998). 149. Thomas, D.G., et al. Expression of receptor tyrosine kinases epidermal growth factor receptor and HER-2/neu in synovial sarcoma. Cancer 103, 830-838 (2005).

76 150. Olsen, R.J., et al. C-erb-B2 (HER2/neu) expression in synovial sarcoma of the head and neck. Head Neck 27, 883-892 (2005). 151. Ikegami, M., et al. TGF-{alpha} perturbs surfactant homeostasis in vivo. Am J Physiol Lung Cell Mol Physiol 289, L34-43 (2005). 152. Bosenberg, M.W., Pandiella, A. & Massague, J. Activated release of membrane- anchored TGF-alpha in the absence of cytosol. J Cell Biol 122, 95-101 (1993). 153. Gallea-Robache, S., et al. A metalloproteinase inhibitor blocks the shedding of soluble cytokine receptors and processing of transmembrane cytokine precursors in human monocytic cells. Cytokine 9, 340-346 (1997). 154. Wong, S.T., et al. The TGF-alpha precursor expressed on the cell surface binds to the EGF receptor on adjacent cells, leading to signal transduction. Cell 56, 495- 506 (1989).