| NF1 tumor suppressor in epidermal differentiation and growth - implications for wound epithelialization and psoriasis | ||
|---|---|---|
| Prev | Chapter 6. Discussion | Next |
The results obtained with cultured keratinocytes demonstrated a specific change in the subcellular localization of NF1 protein during cell differentiation. Specifically, NF1 protein showed co-localization with cytokeratin 5/14 filaments shortly after the induction of keratinocyte differentiation. This co-localization was transient, and the cells later became negative for NF1 protein. The association of NF1 protein with cytokeratin filaments was temporally parallel to the re-organization of the cytoskeleton and the formation of cell-cell adhesions (Hennings et al. 1980, O"Keefe et al. 1987, Eckert et al. 1997). Cytokeratin-positive filaments are known to undergo specific re-organization after the calcium switch, and desmoplakin is simultaneously transported from the perinuclear area to the cell surface along cytokeratin filaments (Jones & Goldman 1985, Pasdar & Nelson 1988b, Pasdar & Nelson 1988a). Co-localization of NF1 protein and intermediate filaments reappeared after the cells had migrated to the second layer on top of the original monolayer. The cells in the second layer also showed cytoskeletal re-organization and formation of new cell-cell adhesions.
CSK buffer treatment was used to study the affinity of the adhesion between NF1 protein and intermediate filaments. NF1 protein seemed to form a high-affinity bond with cytokeratin filaments since the adhesion remained after CSK treatment. NF1 protein formed a timely and transient high-affinity bond with cytokeratin filaments after calcium-induced keratinocyte differentiation. Consequently, one can speculate on the role of the NF1 protein in the re-organization of cytokeratin filaments or in the formation of desmosomes by affecting the transport of desmoplakin to the cell surface. Another possibility is that this phenomenon could be part of the regulation of Ras signaling by withdrawing the NF1 protein from the cell membrane. Interestingly, previous studies have shown that the p53 tumor suppressor associates with the actin cytoskeleton in response to increased calcium concentrations (Metcalfe et al. 1999).
The NF1 protein was also shown to be associated with cytokeratin filaments in skin in vivo. Immunoelectron microscopy showed NF1 protein to be in close association with cytokeratin filaments in basal keratinocytes, but not in the more superficial layers of the epidermis. Furthermore, immunogold particles representing the NF1 protein were commonly found to locate a specific distance apart. In keratinocytes, the average minimum distance between any two immunogold particles was shown to be ~48nm, which is the length of an intermediate filament monomer (Alberts 2002). Interestingly, in fibroblasts, the minimum distance between any two immunogold particles was also ~48nm, suggesting that the NF1 protein could associate with intermediate filaments even in these cells. The location of immunogold particles in the nucleus or close to the cell membrane was also evaluated due to the previous suggestions that the NF1 protein could be located in the nucleus and near the plasma membrane (Malhotra & Ratner 1994, Hsueh et al. 2001, Li et al. 2001). Only a fraction of the immunogold particles were detected in close proximity to the plasma membrane, and no immunogold was found in the nucleus. The findings suggest that the NF1 protein mainly does not locate inside the plasma membrane or in the nucleus in keratinocytes or fibroblasts. Another possibility is that the processing of the samples through fixation could destroy the antigen epitopes, resulting in poor labeling.
Keratinocytes cultured from NF1 patients were used as a model for NF1 gene deficiency and analyzed for possible alterations in cell differentiation caused by this deficiency. NF1 keratinocytes displayed altered morphology and differentiation responses compared to control keratinocytes. Specifically, NF1 keratinocytes were larger in size and displayed greater variation in cell size. The alterations were more pronounced in differentiating keratinocytes, which showed the co-localization of NF1 protein and cytokeratin-positive filaments. This emphasizes that the NF1 gene has a role in cytoskeletal re-organization and cell contact formation. The cytoskeleton and cell adhesions have a central role in regulating cell size and shape (Braga 2000, Pedersen et al. 2001, Runswick et al. 2001, Goldmann 2002). Furthermore, keratinocytes cultured from NF1 patients were immunolabeled for NF1 protein. These cells displayed a lower number of NF1 protein-positive fingers representing the association between NF1 protein and cytokeratin filaments.
Additional studies are needed to shed further light on the association between the NF1 protein and cytokeratin filaments and its link with the function of the NF1 gene and the altered morphology of keratinocytes cultured from NF1 patients. Keratinocytes homozygous (-/-) for NF1 would be a novel approach to studying these phenomena. Keratinocytes cultured from NF1 patients have a heterozygous (+/-) genotype and are less informative for the characterization of the function of the NF1 gene. Homozygous keratinocytes for NF1 can be generated by a conditional knockout mouse, in which the two copies of the NF1 gene can be removed in cultured cells or in specific tissues. Another possibility to generate NF1 gene-deficient cells would be the RNA interference (RNAi) technology, which has recently been successfully used in gene expression silencing in cell cultures. RNAi uses double-stranded RNAs 21-25 nucleic acids long, which silence a specific gene expression with an identical nucleic acid sequence. Double-stranded RNAs can be generated by chemical synthesis or by DNA vector-based strategy (Hannon 2002). The DNA vector-based strategy allows gene silencing over extended periods of time and would be more suitable for the NF1 gene due to the long half-life of the NF1 protein.
Psoriatic epidermis is marked by hyperproliferation of keratinocytes and their abnormal maturation. Inflammation and inflammatory cytokines are thought to have an important role in the development of the psoriatic phenotype. However, it is still under debate whether psoriasis is a primary disorder of keratinocytes or whether the changes in keratinocytes are caused by the activation of the immune system. NF1 protein has previously been shown to be downregulated in psoriatic lesions (Peltonen et al. 1995). Furthermore, elevated levels of active Ras have been demonstrated in psoriatic lesions (Lin et al. 1999), and animal models for psoriasis display increased MAPK activity (Haase et al. 2001).
The results of the present study displayed downregulation of both the NF1 mRNA and protein in psoriatic lesions compared to apparently healthy psoriatic epidermis. The results showed that in the healthy psoriatic epidermis, NF1 protein was expressed in the whole epidermis and the most intense signal was seen basally. This finding differed slightly from previous studies made with frozen sections (Hermonen et al. 1995, Peltonen et al. 1995, Koivunen et al. 2002), which have displayed NF1 protein expression to be more restricted to the basal layer of the epidermis. This might be due to the differences in the epitope availability caused by different fixation protocol or by the differences in fluorescence or color reactions based detection systems. In the psoriatic lesion, the NF1 protein expression decreased compared to the healthy looking epidermis. Interestingly, NF1 mRNA expression in healthy psoriatic epidermis displayed different localization compared to the NF1 protein expression. The most intense signal for the NF1 mRNA was seen suprabasally compared to the NF1 protein expression which was more restricted to the basal layer. This could be caused by post-translational regulation of the NF1 protein half-life which is known to occur (Kaufmann et al. 1999a, Cichowski et al. 2003). Furthermore, NF1 mRNA and protein expressions are known not to be upregulate together (Kaufmann et al. 1999a). In the psoriatic lesions, NF1 mRNA expression decreased and no clear difference in the expression was evident between the different layers of the epidermis.
In vitro, keratinocytes cultured from psoriatic patients revealed decreased levels of NF1 mRNA and protein in both lesional and non-lesional cells. It was also noted that cultured psoriatic keratinocytes had more pronounced downregulation of NF1 mRNA compared to the NF1 protein. The downregulation of NF1 gene expression suggests that cultured psoriatic keratinocytes are able to retain their altered NF1 expression after removal of the cells from the presence of lymphocytes or fibroblasts. This favors the assumption of the primary role of keratinocytes in the pathogenesis of psoriasis. It should be emphasized that decreased NF1 gene expression could be a subclinical condition that occurs prior to the clinical manifestations. Previous studies have suggested that cultured psoriatic keratinocytes can maintain their altered phenotype and function (Baden et al. 1981, Baker et al. 1988, Szabo et al. 2002). In addition to the quantitative studies on NF1 gene expression in cultured psoriatic keratinocytes, subcellular localization analysis was done by using immunostaining and in situ hybridization. Immunostaining for NF1 protein in differentiating keratinocytes displayed less intense labeling on the intermediate filament network. Non-radioactive in situ hybridization methods with digoxigenin-labeled probes have made it possible to identify subcellular distribution of certain mRNAs (Yla-Outinen et al. 2002), and this method was used in the current study to identify the localization of NF1 mRNA. Localization of NF1 mRNA in differentiating cells displayed less marked re-distribution of NF1 mRNA in psoriatic cells compared to controls.
One can speculate about the role of the NF1 gene and the mechanism and effect of its downregulation in psoriasis. Increased amounts of active Ras could lead to increased consumption of NF1 mRNA/protein. Furthermore, it has been shown that NF1 mRNA is redistributed towards the cell-cell adhesion area via actin filaments (Yla-Outinen et al. 2002). Previous studies have demonstrated alterations in the cytoskeleton and cell adhesions in psoriasis (Jahn et al. 1988, Pellegrini et al. 1992, De Luca et al. 1994, Cavicchini et al. 1996, Magaudda et al. 1997). Alterations in cytoskeleton and cell adhesions could lead to improper transport and faster breakdown of NF1 mRNA, resulting in decreased levels of NF1 protein. NF1 deficiency was shown in the first part of the current study to result in an altered phenotype of keratinocytes. Whether NF1 gene deficiency has a role in the formation of psoriatic phenotype or whether it is just a consequence of the alterations in psoriatic keratinocytes needs to be further studied.
It has been suggested that trauma could play a part in the formation of the skin manifestations associated with neurofibromatosis, such as neurofibromas and CALMs (Friedman & Riccardi 1999, Karvonen et al. 2000). Furthermore, the NF1 gene is expressed in keratinocytes and fibroblasts, which are the most important elements in the process of wound healing (Malhotra & Ratner 1994, Hermonen et al. 1995, Yla-Outinen et al. 1998). One can hypothesize that the NF1 gene may have a role in wound healing, and that abnormal wound healing could result in skin manifestations associated with the disease. Previous studies have addressed the connection between wound healing and the NF1 gene. Specifically, NF1 gene expression is increased in fibroblasts of scars and cultured fibroblasts exposed to various growth factors operative during wound healing (Yla-Outinen et al. 1998). Mice heterozygous for the NF1 gene have shown defects in wound healing after a deep dermal wound (Atit et al. 1999).
The current study investigated, for the first time, experimental wound healing and NF1. Suction blisters were used to generate epidermal injuries, and their healing was followed. In normal epidermal wound healing, NF1 gene expression increased both in epidermal keratinocytes and in dermal fibroblasts in response to wounding, suggesting that the gene may have a role in the process. Consequently, wound healing efficiency was studied in NF1 patients. All the patients were first evaluated retrospectively for any abnormalities in wound healing. Clinical evaluation of all the scars on NF1 patients showed that two patients had neurofibromas rising directly from scars. This suggests that neurofibromas or CALMs mainly do not co-localize with scars. The suction blister method was used to study the efficiency of epidermal wound healing. In the early stages of wound healing, no significant differences between NF1 patients and healthy controls were found as determined by clinical evaluation, TEWL measurements or histological analysis. Potential long-term effects of wounding were studied clinically by re-evaluating the blister and biopsy areas three years after wounding. There were no long-term effects of the wounding.
In conclusion, epidermal wound healing and the healing of biopsy scars seem to be equally effective in NF1 patients as in controls. Furthermore, trauma caused by epidermal or deep skin injury does not appear to play any major role in the formation of skin manifestations associated with NF1.
The best characterized function of the NF1 gene is its function to act as a Ras-GAP and to catalyze the hydrolysis of active Ras-GTP into inactive Ras-GDP (Ballester et al. 1990, Martin et al. 1990, Xu et al. 1990a, Bollag & McCormick 1991, Basu et al. 1992, Bollag et al. 1996). The most important downstream target of Ras-GTP signaling is the Raf-MEK-ERK cascade, which leads to increased cell growth and proliferation in most tissues (Kolch 2000). It is still under debate whether the most important function of NF1 gene is to regulate Ras signaling, or whether it has some other functions. Previous studies have shown that NF1 deficiency/functional insufficiency may lead to increased Ras-MAPK activities in various tissues (Bollag et al. 1996, Sherman et al. 2000, Gitler et al. 2003). However, NF1 deficiency and Ras-GTP levels do not go together in all tissues (Johnson et al. 1994, Boddrich et al. 1995, Griesser et al. 1995, Sherman et al. 2000). Furthermore, NF1 gene impairment may lead to alterations in cells without changes in Ras-GTP levels (Jimbow et al. 1973, Ishida & Jimbow 1987, Griesser et al. 1995, Atit et al. 1999, Sherman et al. 2000), and the NF1 gene may carry out its function, such as the regulation of cell growth, independently of Ras signaling (Johnson et al. 1994, Li & White 1996, Guo et al. 1997, The et al. 1997).
The present study investigated the behavior and function of the NF1 gene in epidermal keratinocytes. The NF1 protein is known to function in epidermal keratinocytes because mice heterozygous for the NF1 gene display increased susceptibility to chemical carcinogens. However, it is not clear whether this susceptibility is caused by impaired GAP activity, although Ras signaling and NF1 gene impairment seemed to have a synergistic effect on it (Atit et al. 2000). The results of the current study favor the theory that the NF1 gene has functions other than acting as a GAP protein in keratinocytes. The NF1 protein specifically adhered to the intermediate filament network in differentiating cells rather than localized close to the plasma membrane both in vitro and in vivo. Active Ras is attached to the plasma membrane through a farnesyl residue, and the prerequisite for the GAP function of the NF1 protein is a close spatial relationship with the plasma membrane. Furthermore, NF1 patients did not display an increased number of cells with active MAPK in normal, hyperproliferative or migrative epidermis. In psoriasis, increased Ras activity of keratinocytes has previously been reported (Lin et al. 1999). Further studies are needed to investigate the link between decreased expression of the NF1 gene in psoriatic keratinocytes and elevated Ras-GTP levels.
Dermal tissues were also studied for the activity of the Ras-MAPK pathway. In fibroblasts, there was no difference in the number of cells active for MAPK between NF1 patients and healthy controls. This is in agreement with the previous results showing that fibroblasts cultured from NF1 patients have normal Ras-GTP levels (Sherman et al. 2000). Unexpectedly, an increased number of MAPK-active cells were seen in the periarterial area of NF1 patients. This correlated with the increased cell proliferation detected by KI-67 immunostaining. The periarteriolar area is located just underneath the endothelium and is mainly composed of smooth muscle cells or pericytes. It is likely that the active cells were smooth muscle cells because smooth muscle cell actin positive cells were located roughly on the same area as the positive cells, and NF1 patients displayed increased amount of smooth muscle cells around the arterioles. Previous studies have suggested that vasculopathy related to neurofibromatosis is caused by proliferative smooth muscle cells in the vessel media. Increased smooth muscle cell proliferation may result from failure of the endothelium to regulate smooth muscle cells or abnormal smooth muscle cell responses to endothelial signals (Greene et al. 1974, Lehrnbecher et al. 1994, Hamilton & Friedman 2000, Riccardi 2000). The NF1 gene has previously been identified in the endothelium and the smooth muscle layer of blood vessels (Norton et al. 1995). Furthermore, a recent study with conditional NF1 knockout mice has shown the importance of the NF1 gene as a downregulator of Ras-MAPK signaling in the endothelium (Gitler et al. 2003). It is interesting to note that HMG-CoA inhibitors (statins), which are used widely to lower high cholesterol and to treat atherosclerosis, do have other functions apart from inhibiting cholesterol synthesis. Furthermore, not all of the beneficial effects of statins can be explained by the cholesterol-lowering action. Statins also inhibit the synthesis of the farnesyl moiety, which is essential for the Ras protein function. It has been suggested that some of the beneficial effects of statins could be caused by Ras inhibition (Libby & Aikawa 2002). One can speculate about the role of the Ras-MAPK pathway in atherosclerosis and neurofibromatosis associated vasculopathy, which both share similar features, such as distribution of the lesions and increased smooth muscle cell proliferation of the vessel wall.
In conclusion, the present study supported the theory that the NF1 gene functions as Ras-GAP, but it also has other functions. The results suggest that NF1 deficiency results in increased Ras-MAPK signaling only in some tissues, such as smooth muscle cells, but not in keratinocytes or fibroblasts.