Abstract
Oral mucosal barrier injury (mucositis) is a frequent, painful, serious, dose-limiting toxicity associated with many anti-neoplastic drugs and radiation to the head and neck. Results of recent studies suggest that mucositis is the result of a complex series of interactive biological events that take place in the submucosa and epithelium. The nuclear transcription factor NF-κB has been implicated in the control of a broad range of biological responses, the activation of a large number of specific cellular genes, and the determination of the fate of cells exposed to ionizing radiation and anti-neoplastic drugs. Of particular importance to mucositis is the fact that NF-κB regulates key elements in the apparent sequence that leads to normal tissue toxicity. Not the least of these is the effect that NF-κB activation has on apoptosis. In particular, a paradox exists between the potential pro-apoptotic effect NF-κB exerts on normal cells, and the anti-apoptotic and cytoprotective effect it causes in tumor cells. This paper provides a review of the structure and function of NF-κB and speculates how its apparent enigmatic effect on normal and tumor cells may occur.
(I) Introduction
Oral mucositis is one of the most significant non-hematologic toxicities associated with drug and radiation therapy for cancer. Lesions associated with mucositis result in pain, decreased quality of life, increased length of hospitalization, higher risk of local and systemic infection, and modification of anti-neoplastic treatment regimens (Sonis et al., 2001). The fact that mucositis risk restricts the use of optimum drug combinations and radiation schedules has a negative impact on the use of the best possible cancer therapy.
The importance of mucositis in the overall scheme of cancer management is now well-recognized and has resulted in interest in the development of effective forms of intervention. While it seems probable that some form(s) of treatment will be forthcoming in the next few years, a “magic bullet” seems unlikely, given the biologic complexity of mucosal barrier injury. The identification and description of the molecular and cellular underpinnings of mucositis have provided a variety of potential therapeutic targets. In addition, it appears that mucositis may provide an excellent model for other forms of mucosal injury.
A few years ago, we hypothesized that the development of mucositis represented a complex tissue response that involved mucosal endothelium and connective tissue, in addition to epithelium (Sonis, 1998). Results of studies by our group with oral mucositis, and others with intestinal mucositis, support this concept (Sonis et al., 2000; Paris et al., 2001). We also proposed that pro-inflammatory cytokines were likely to be an important mechanistic component in mucositis pathogenesis. The findings of increased plasma levels of tumor necrosis factor-alpha (TNF-α) and interleukin-6 in chemotherapy-treated patients with non-hematologic toxicities (Hall et al., 1995; Sleijfer et al., 1998), and the concordant expression of genes associated with TNF-α and local cell-surface expression of IL-1β were consistent with this concept (Sonis et al., 2000).
Our initial hypothesis was largely based on the “downstream” events that followed mucosal exposure to radiation or chemotherapy. In investigators’ attempts to paddle “upstream” and identify the initiating biologic events that result in mucosal barrier injury, a biologic paradox has emerged that suggests contrasts in the ways that normal and tumor cells respond to pivotal pharmacological and biologic challenges. In particular, differences in the consequences of activation of the transcription factor NF-κB may effect tumor cell resistance to cytotoxic drugs, and normal cell susceptibility to injury and death.
The objectives of this paper are to provide the reader with a review of the structure and function of NF-κB, to describe its role in disease, apoptosis, and differentiation and proliferation, and to analyze the possible biological paradox created by the purported effects of NF-κB on tumor cell survival.
(II) General Description of NF-κB
The inducible transcription factor, NF-κB, was discovered in 1986 (Sen and Baltimore, 1986) and has been extensively studied since. NF-κB is the collective name for a group of dimeric ubiquitous transcription factors that comprise the Rel family (Chen and Ghosh, 1999). The NF-κB/Rel family has five members: NF-κB1 (p50/p105), NF-κB2 (p52/p100), p65 (Rel A), Rel 3, and cRel (Huxford et al., 1999). These factors have been implicated in the control of a broad range of biological responses, the activation of a large number of specific cellular genes, and the determination of the fate of cells exposed to extracellular stimuli such as cytokines, ionizing radiation, and chemotherapeutic drugs (Pahl, 1999). It is the observation that NF-κB activation can be either pro-apoptotic or anti-apoptotic that suggests that it might be a key factor in determining the fate of normal mucosal tissue that has been exposed to cytotoxic cancer therapy. It should be pointed out early in this review that, like many other areas of science, new data emerge almost weekly and with them a changing landscape in our attempts to define the complexities of cellular injury. Consequently, the best that this paper can do is to present a snapshot of the status of the field at the moment and to hypothesize how conflicting bits of data best fit to explain a clinical observation.
NF-κB most typically resides as the heterodimer of the p65/RelA and p50 or p52 subunits in the cytoplasm of cells. The p65/p50 heterodimer is considered to be the classic form (Tak and Firestein, 2001) which, in its inactive state, is bound to the members of the IκB (Inhibitor kappa B) class of proteins (Delhase and Karin, 1999). Five members of the IκB class of proteins have been described (IκB-α, Iκ-β, IκB-å, IκB-γ, and BCL3) (Foo and Nolan, 1999). With the exception of BCL3, all reside in the cytoplasm and inhibit NF-κB activation. BCL3 is found in the nucleus, where it reportedly enhances NF-κB activation. Upon stimulation from any of a number of extracellular stressors, the dimer is sequentially phosphorylated and “ubiquitinated” by activation of IKK (IκB kinase) and ubiquitin ligase (Makarov, 2000). The “ubiquitinated” form is then degraded by the 26S proteosome (Makarov, 2000; Tanaka et al., 2001). No longer encumbered by complexed Iκ-B, NF-κB goes from the cytoplasm to the nucleus, where it may activate a broad range of genes that could have a significant function in mucosal injury (Fig. 1).
(III) Activation and Target Genes of NF-κB
The number and range of agents that may activate NF-κB are potential indicators of the central role that NF-κB plays in the response to physiologic or iatrogenic stress. A recent review describes 196 known inducers of NF-κB (Pahl, 1999). Among activating agents are bacteria and their products, viruses, including members of the herpes family, a long list of cytokines, oxidative stress (Barnes and Karin, 1997; Pahl, 1999; Makarov, 2000), and, interestingly, free-radical scavengers (Murley et al., 2001). Of the cytokines, TNF-α and IL-1β are probably among the most important (Cao et al., 1999; Yamamoto and Gaynor, 2001). With respect to cancer therapy, ionizing radiation is a potent and consistent activator of NF-κB, as are many commonly used anti-neoplastic agents (Table 1) (Brach et al., 1991; Das and White, 1997; Hellin et al., 1998; Lee et al., 1998; Pahl, 1999; Gurujeyalakshmi et al., 2000; Weldon et al., 2001). Of these, many are stomatotoxic (Wardley et al., 2000; Polee et al., 2001). Furthermore, the impact of concomitant regimens in which radiation and drug therapy are given together or the use of multi-agent protocols has not been evaluated with respect to impact on NF-κB activation.
Equally extensive are the target genes of NF-κB, which also number close to 200 (Pahl, 1999). Many are potentially important in mucosal injury and disease. Among them are genes associated with the expression of many cytokines, chemokines, and immunoglobulins (Table 2). These include virtually all of the pro-inflammatory cytokines, cytokines involved in angiogenesis, and growth factors. Also on the list are immunoreceptors, cell adhesion molecules (including those of potential significance in oral mucosal disease), acute phase proteins, and stress response genes. As will be discussed below, NF-κB targets a variety of genes associated with apoptosis and early response genes, including p53 (Benoit et al., 2000; Bours et al., 2000).
While many factors, including IL-1 and TNF-α, have the ability to cause IκB degradation and NF-κB translocation to the nucleus, they may not, alone, result in NF-κB transcription. Rather, in some cases, co-activators may be required. The necessity of a co-activator to up-regulate a particular gene provides a mechanism for secondary control of the process (Kasibhatla et al., 1998; Sheppard et al., 1999). Such co-activators are controlled independently, yet may act in synergy to up-regulate target genes. Since it is clear that it is physiologically impossible for every target gene to be up-regulated each time NF-κB is induced, the presence of co-activators may serve as a selection mechanism and regulator of such activity and add a level of discretion to NF-κB gene induction. This mechanism might be of particular importance in determining the lack of a consistent response of different cell types to the same stressor.
(IV) NF-κB as a Regulator of Cell Growth, Proliferation, and Apoptosis
Of all of its potential functional ramifications, one of the most significant is the ability of NF-κB to mediate and affect apoptosis (Lin et al., 1999; Bours et al., 2000). Apoptotic cell death is among the most important mechanisms in maintaining tissue homeostasis. While activation-induced cell death is an important regulator of immune cell viability, lack of apoptosis has the potential to lead to hyperplasia and carcinogenesis (Foo and Nolan, 1999; Makarov, 2001). On the other hand, therapy-directed apoptosis, as is the goal with anti-neoplastic therapy, is the critical event in determining tumor outcomes following chemotherapy or radiation. While it is clear that NF-κB plays a significant role in deciding a cell’s fate, it seems that NF-κB can play both sides of the fence, being pro-apoptotic in some situations and anti-apoptotic in others (Baichwal and Baeuerle, 1997; Lin et al., 1999; Bours et al., 2000; Kaltschmidt et al., 2000). How NF-κB effects apoptosis depends on the cell type, the activating stress, and co-factors.
Although the mechanism by which NF-κB regulates apoptosis has yet to be fully resolved, Barkett and Gilmore (1999) have proposed three general models: First, NF-κB could directly effect apoptosis-controlling genes; second, NF-κB could effect the cell cycle in such a way as to sensitize or desensitize the cell to apoptotic signals; or, third, NF-κB could interact with cellular proteins that themselves effect cell survival.
The tumor suppressor p53 has been implicated in cell-cycle arrest and apoptosis, and its inhibition appears to play a major role in the development of many tumors. Ryan and his colleagues observed that induction of p53 leads to activation of NF-κB in cultured cells, and that correlated with the ability of p53 to induce apoptosis (Ryan et al., 2000). Furthermore, they found that the loss of NF-κB eliminated the p53-mediated apoptotic response. Consequently, they concluded that NF-κB may be an important and necessary co-factor in p53-mediated tumor cell death. Interestingly, induction of p53 by NF-κB is well-known.
A role for NF-κB in effecting cell survival was first suggested by the observation that several genes targeted by the transcription factor favored apoptosis (Table 2). Among these were fasl, TNF-α, and c-myc (Chen et al., 2001). In addition, activation of these genes was associated with known apoptotic agents, including some anti-cancer drugs and radiation. Virally induced apoptosis of cultured hepatocytes was blocked when cells were treated with NF-κB decoys (Marianneau et al., 1997). Similar observations of the pro-apoptotic effects of NF-κB activation have been reported in response to neuronal injury in both in vitro and in vivo systems. Grilli et al.(1996) found that aspirin and its metabolite sodium salicylate protected rat neuronal cultures and hippocampal slices from glutamate-induced neurotoxicity by inhibiting NF-κB. In an in vivo model for Huntington’s disease, Nakai et al.(2000) found that activation of NF-κB contributed to the induction of p53 and c-myc and led to apoptosis.
Studies of the potential role of NF-κB in Alzheimer’s disease are illustrative of the complexity of its activity relative to apoptosis. For example, it has been reported that the immunoreactivity of NF-κB in neuritic plaques from Alzheimer’s patients decreases with the progression of the disease. Amyloid β protein, which is thought to cause neurodegeneration in Alzheimer’s disease, appears to be a potent activator of NF-κB. One interpretation of these findings is that NF-κB activation is an initiating event in neuritic plaque formation and neuronal apoptosis (Yamamoto and Gaynor, 2001). In contrast, Kaltschmidt et al.(1999) conclude that the reduction of NF-κB activity in more advanced plaques suggests that its role is cytoprotective (anti-apoptotic) and that its loss favors tissue destruction.
Other studies also suggest an anti-apoptotic role for NF-κB. In 1995, Beg et al. observed that mice devoid of RelA died an embryonic death as a consequence of hepatocyte apoptosis. Similarly, IKK-β gene-deficient animals died as embryos from the same cause. In 1998, Taub reported that activation of NF-κB enhanced liver regeneration following partial hepatectomy by blocking apoptosis. Furthermore, activation of NF-κB by known hepatic carcinogens, such as Phenobarbital, inhibits apoptosis through a mechanism in which stabilization of IκBα occurs.
TNF-α is one of the most consistent inducers of apoptosis in mammalian cells and a persistent activator of NF-κB (Wang et al., 1996; Yamamoto and Gaynor, 2001). The ability of NF-κB activation to block TNF-α-induced apoptosis has been studied extensively and has been demonstrated in a wide range of cells, including fibrosarcoma, keratinocytes, endothelial cells, myeloid cells, chronic lymphocytic leukemia, lymphoid cell lines, cutaneous T-cell lymphoma, hepatocyte cell lines, melanoma cells, prostate cancer cells, epithelial cells, glomerular mesangial cells, pancreatic cancer cells, Ewing tumor cells, and head and neck squamous cell carcinoma (Aggarwal, 2000).
That NF-κB can up-regulate the expression of anti-apoptotic genes is based on the observations that (1) up-regulation of the target gene by NF-κB in response to an apoptotic-producing agent results in cell survival, (2) that elimination or inhibition of the gene eliminates this protective effect, and (3) that ectopic expression of the gene also protects the cell. Among genes that are activated by NF-κB and thought to be protective are cellular inhibitors of apoptosis (cIAP)-1 and cIAP-2, and adaptor molecules like the TNF-receptor-associated factors TRAF1 and TRAF2 (Speiser et al., 1997; Pahl, 1999). The mechanism(s) by which TNF-receptor-associated factors affect NF-κB-mediated apoptotic effects is still speculative (Yamamoto and Gaynor, 2001). One hypothesis is that overexpression of superoxide dismutase, induced by TNF and NF-κB, results in resistance to TNF-induced apoptosis (Manna et al., 1998).
Some members of the Bcl-2 gene family that have anti-apoptotic activity are up-regulated by NF-κB. Among these are A1/Bfl-1, Bcl-XL, and Nrl3. Using a fibrosarcoma cell line, Wang et al.(1999) demonstrated that NF-κB activation resulted in an anti-apoptotic endpoint following the suppression of the mitochondral release of cytochrome c through the activation of the A1/Bfl-1 gene. Zong et al.(1999) reported that NF-κB-dependent up-regulation of the same Bcl-2 homologue Bfl-1/A1 conferred resistance to TNF-induced apoptosis.
There appears to be increasing evidence to suggest that NF-κB also has a role in cell proliferation and differentiation. Activation of NF-κB has been shown to be required for cell cycle progression and differentiation. Using a myoblast cell line, Guttridge et al.(1999) found that cells lacking NF-κB activity demonstrated both cell cycle and proliferation defects. Upon analysis, these investigators identified cyclin D1 as the target for NF-κB, and that NF-κB was required for the progression of cells from the G(1) to the S phase of the cell cycle.
That NF-κB may have a role in differentiation has been explored in skin. Takeda et al.(1999) disrupted the gene coding for IκB kinase-alpha by gene-targeting in mice. While these animals died perinatally, the authors observed that the skin epidermal cells were highly proliferative and poorly differentiated, suggesting that NF-κB is critical for controlling the cell cycle and forcing cells to terminal differentiation.
(V) The Role of NF-κB in Disease
While the activation of NF-κB seems to be a critical step in the genesis and chronicity of several diseases, its absence or inability to function may also be significant. As noted above, the multifunctional aspects of NF-κB help balance and regulate the extent and consequences of inflammatory responses, proliferation, immune function, and apoptosis. A molecule with so many functions is consequently critical in providing a nidus of control for a variety of simultaneous and independent functions. The absence of such a factor, or a modification in its ability to perform normally, is likely to have significant clinical consequences.
(A) Inflammatory diseases
Because NF-κB has such a wide range of target genes that are associated with inflammatory processes, it has been implicated in the pathogenesis of several diseases (Baldwin, 2001a). Genes associated with the activation of pro-inflammatory cytokines, molecules that are chemoattactrants for leukocytes (chemokines), enzymatic mediators of inflammation, immune receptors, and adhesion molecules are all potentially regulated by NF-κB (Pahl, 1999; Makarov, 2000). Thus, it is possible that a single transcription factor can act to create a coordinated inflammatory response involving multiple cell and tissue types. For example, NF-κB has been implicated in the etiopathology of atherosclerosis. Following injury to the endothelium and smooth muscle of vessel walls, it has been proposed that activation of NF-κB, because it is anti-apoptotic, stimulates endothelial proliferation and results in hyperplasia and consequent atherosclerosis (Collins and Cybulsky, 2001).
NF-κB may play a role in arthritis, both as a consequence of its role as an inflammatory regulator and because of its ability to function as an immunologic modifier (Makarov, 2001; Yamasaki et al., 2001). Activation of NF-κB by synovial TNF-α (Fujisawa et al., 1996) results in local joint inflammation (Makarov, 2001; Tak and Firestein, 2001). Animal studies have shown that synovial NF-κB binding precedes the development of experimental collagen-induced arthritis. Similarly, NF-κB also appears to play a pivotal role in arthritis induced by streptococcal cell wall products (Miagkov et al., 1998; Palombella et al., 1998). Makarov notes a broader involvement of NF-κB in arthritis that results from its ability to affect T-helper 1 responses, apoptosis and proliferation of synovial cells, and the differentiation and activity of osteoclasts (Makarov, 2001). Furthermore, while the mechanism of action of drugs typically used to treat arthritis has focused on their ability to inhibit prostaglandin synthesis, aspirin and some of the NSAIDs have been shown to inhibit NF-κB activation effectively (Tak and Firestein, 2001).
Using a rat model for crescentic glomerulonephritis, Tomita et al.(2000) recently demonstrated a potential role for NF-κB in the development of glomerulonephritis. They hypothesized that since the expression of glomular pro-inflammatory cytokines (IL-1β and TNF-α) was a target for NF-κB transcription, it seemed likely that the transcription factor could play a significant role in the pathogenesis of the disease. To test their hypothesis, they primed animals for glomerulonephritis by the injection of anti-glomerular basement membrane antibody. They then perfused test kidneys with a NF-κB blocker. Whereas control, sham-treated animals developed clinical and histologic signs of severe glomerulonephritis, animals treated with the NF-κB blocker demonstrated substantially less disease.
The role of NF-κB in two other inflammatory diseases is noteworthy. Biopsy specimens from the intestinal tissue of patients with active inflammatory bowel disease (IBD and Crohn’s disease) demonstrated the presence of activated NF-κB (Schottelius and Baldwin, 1999). This finding correlated well with the increased cytokine activity that accompanies clinical exacerbations of IBD. Other epithelial processes have also been associated with NF-κB. The tissue changes noted in psoriasis may be the result of both the proliferative and inflammatory consequences of NF-κB activation (Robert and Kupper, 1999). Since several oral diseases, such as lichen planus, demonstrate a similar combination of hyperplastic and inflammatory changes, it is not unreasonable to consider NF-κB as a potential participant in their pathogenesis.
(B) NF-κB and diabetes
NF-κB may have a role in the etiology of diabetes mellitus. In a study using the non-obese mouse model for spontaneous insulin-dependent diabetes mellitus (Type 1 or juvenile onset diabetes), Hayashi and Faustman (2000) noted a specific genetically based proteasome defect. As noted above, proteasome 26 plays a key role in NF-κB activation. Consequently, animals with this defect were unable to produce or activate NF-κB. The authors concluded that the inability to activate NF-κB was likely to be a key element in the development and pathology of Type 1 diabetes.
(VI) NF-κB and Cancer
At least three roles have been proposed for NF-κB in relation to cancer: oncogenesis, tumor progression, and the response of tumors to therapy.
A function for NF-κB in oncogenesis is suggested by both circumstantial and direct evidence (Rayet and Gelinas, 1999; Baldwin, 2001b). Genes that encode for c-Rel, NF-κB2, and Bcl-3 proteins are located within the regions of the genome that are involved in re-arrangements and amplifications. Chromosomal translocation of these genes have been associated with B-cell chronic lymphocytic leukemia, non-Hodgkin’s lymphoma, and Hodgkin’s lymphoma. Second, NF-κB can be activated by several viral transforming proteins. Finally, and perhaps the most compelling, are the observations that blocking NF-κB results in tumorigenesis. The oncogenic fusion protein, Bcr-Abl, is a chimeric oncoprotein associated with acute and chronic lymphocytic leukemia. The tumorigenicity of the protein is blocked in nude mice by the super-repressor IκBα that effectively prevents NF-κB activation (Reuther et al., 1998).
As discussed below, one of the most interesting aspects of NF-kB as it relates to cancer is its potential role in ensuring tumor growth and survival.
(A) NF-κB and cancer therapy
The breadth of NF-κB’s activity, particularly with respect to its function as a significant gatekeeper for apoptosis, has led to intense interest in the potential role of NF-κB in tumor proliferation, drug and radiation resistance, and as a therapeutic target. That NF-κB can be activated by a wide range of commonly used chemotherapeutic agents is well-established. The anthracyclines (daunorubicin and doxorubicin) (Arlt et al., 2001; Maestre et al., 2001), the plant alkaloids (vinblastine and vincristine) (Das and White, 1997), paclitaxel (Das and White, 1997), cisplatin (Kato et al., 2000), melphalan (Donepudi et al., 2001), bleomycin (Weldon et al., 2001), and etoposide (Arlt et al., 2001)—all are reported to activate NF-κB. In contrast, taxol selectively inhibits phorbol-ester-induced NF-κB activation (Spencer et al., 1999). And surprisingly, the work of Aota et al.(2000) found that 5-fluorouracil (5-FU), an anti-neoplastic agent frequently used to treat head and neck and colorectal cancers, suppresses NF-κB by mediating the up-regulation of IκB-α expression. Ionizing radiation, such as is used in the treatment of many solid tumors, is a potent NF-κB activator, probably through a pathway that involves reactive oxygen intermediates that cause the degradation of IκB-α (Lee et al., 1998).
Since most anti-neoplastic agents kill cancer cells via apoptosis, the consequence of NF-κB activation may have significant clinical ramifications. If NF-κB activation leads to a pro-apoptotic outcome, then its drug- or radiation-induced activation should proceed to increased tumor cell death and, simultaneously, to increased toxicity of normal bystander cells such as those of the gastrointestinal mucosa, including the mouth. On the other hand, if activation of NF-κB results in an anti-apoptotic endpoint, its activation by chemotherapeutic agents or radiation could be cytoprotective for the tumor. In fact, there is a great deal of evidence to suggest that this may be the case.
The results of a series of studies support the notion that activation of NF-κB plays a critical role in protecting cancer cells from chemotherapy- or radiation-induced apoptosis. Hellin et al.(1998) reported that daunomycin, a drug that is commonly used to treat solid tumors and leukemia, exerts its effect by activation of p53, and that the p53-activating signal was partially regulated by NF-κB. They found that inhibition of NF-κB activation by overexpression of a stable unresponsive IκBα mutant impaired p53 expression, which could have a subsequent impact on daunomycin cytotoxicity (Hellin et al., 1998). Cheng and colleagues (2000) found that NF-κB-defective cells were exquisitely sensitive to etoposide, adriamycin, and cisplatin compared with cells of a similar type in which NF-κB function was intact. The mechanism to which this effect was ascribed was associated with an anti-apoptotic activity attributed to NF-κB that was mediated through the differential activation of death antagonist genes of the Bcl-2 family (Bfl-1 and Bcl-x). Weldon et al.(2001) demonstrated that inhibition of the NF-κB pathway in breast cancer cells resistant to chemotherapeutic agents enhanced their susceptibility to paclitaxel and doxorubicin. Arlt et al.(2001) had a similar experience with human pancreatic carcinoma cells. They found that treatment with various NF-κB inhibitors or stabilization of IκB-α through transfection with the IκBα super-repressor strongly enhanced the apoptotic effects of etoposide (VP-16) and daunorubicin on cells that had been previously resistant to these agents (Arlt et al., 2001).
Since control or elimination of a cytoprotective role for NF-κB could have significant clinical ramifications, the mechanisms by which the transcription factors control this function is under intense scrutiny. NF-κB-mediated activation of TRAF and IAP (inhibitor of apoptosis protein) suppresses the activation of caspase 8, the apical caspase in the TNF signaling cascade (Wang et al., 1999). The latter investigators demonstrated that the activation of NF-κB hindered the mitochondral release of cytochrome c, which is pro-apoptotic. This suppression occurred as the result of a Bcl-2 family intermediary, Al/Bfl-1. With respect to tumor cell chemotherapy resistance, they further observed that activation of Al resulted in resistance to etoposide-induced apoptosis by inhibition of both cytochrome c release and activation of caspase 3, the effector caspase leading to cell kill.
However, the complexity of NF-κB function relative to tumors is illustrated by the findings of Aota et al.(2000), who studied the effect of 5-FU, a drug that is highly toxic to the oral and gastrointestinal mucosa. As noted above, 5-FU is commonly used for the treatment of head and neck and colorectal cancers, and Aota and co-workers examined its effect on human salivary gland tumor cells. In stark contrast to the vast number of reports indicating that the majority of chemotherapeutic drugs activated NF-κB and its anti-apoptotic and cytoprotective effects, they found that 5-FU suppressed NF-κB in salivary gland tumor cells by up-regulating IκB-α expression (Aota et al., 2000). Furthermore, they reported that 5-FU-induced apoptosis occurred as a consequence of the suppression of NF-κB, which led to inactivation of TRAF-2 and c-IAP, both anti-apoptotic proteins, and subsequent increased activity of caspases-8 and -3. The same group (Azuma et al., 2001) studied the mechanism by which 5-FU-mediated suppression of NF-κB occurred and found that 5-FU suppressed IκB kinase (IKK) activity, thereby inhibiting NF-κB activation. Their findings have particular clinical relevance in that they suggest that NF-κB responds differently to different anti-neoplastic agents. This creates a level of mechanistic complexity in the now-common situation in which patients are treated with multi-agent or concomitant therapy. Furthermore, while a mechanism for 5-FU-mediated normal tissue damage is readily explained by this model, the similar effect of NF-κB-activating cancer drugs becomes an enigma.
A role for NF-κB in the induction of multidrug resistance (MDR) has also been recently demonstrated (Um et al., 2001). MDR is a phenomenon in which cancer cells develop resistance to treatment by drugs of different structures and functions. The mechanism(s) by which MDR develops have not been fully elucidated, but since NF-κB had been associated with a cell survival response in tumor cells, Um et al.(2001) recently investigated NF-κB to determine if it was associated with an MDR phenotype. They found that constitutive levels and activity of NF-κB were higher in MDR cells than in cells that were drug-sensitive. Furthermore, the DNA damage sensor and the double-strand break-repair protein, Ku, was positively correlated with NF-κB activity in MDR cells. This finding is of potential clinical significance, since double-strand DNA breaks are induced by radiation and many forms of chemotherapy. Etoposide activated both NF-κB and Ku, and a proteasome inhibitor, essential for NF-κB activation, inhibited them. They concluded that this co-activity could provide a mechanism for MDR.
(B) NF-κB and mucosal toxicity in cancer patients
Cancer-therapy-induced cell death has been attributed to apoptosis. This is true of both tumor cells and normal cells. For the most part, tumor response to anti-neoplastic agents and radiation is dose-related. Unfortunately, neither cancer drugs nor radiation has the ability to discriminate between normal and tumor cells. Consequently, anti-neoplastic therapy elicits marked mucosal injury along the gastrointestinal tract, but most dramatically in the mouth.
The onset and severity of mucositis are associated with many activators of NF-κB. In addition to the anti-neoplastic agent, earlier studies have shown a relationship between the levels of the pro-inflammatory cytokines TNF-α and IL-1β and non-hematologic toxicities (Hall et al., 1995; Sleijfer et al., 1998). Examination of mucosal levels of these proteins in tissue from irradiated animals demonstrated a correlation between cytokine levels and tissue injury (Sonis et al., 2000). Additional studies demonstrated that mucosal ulceration was followed by bacterial colonization. Subsequently, the release of bacterial cell wall products into the submucosa resulted in the stimulation of additional cytokine production and the amplification of tissue injury. Early p53 expression followed the exposure of normal oral mucosa to a radiation challenge (Sonis et al., 2000). It appears that the generation of reactive oxygen species is an important precipitating event leading to mucositis, since topical application of anti-oxidants, such as N-acetyl cysteine, attenuates experimental mucositis (Rosenthal, 2001). Finally, the finding that inhibition of cytokine production and reduction in bacterial load favorably affects the severity and duration of mucositis (Loury et al., 1999; Sonis et al., 2000) implicates both in its pathogenesis.
Given the prevalence of known activators, it seems highly likely that NF-κB activation occurs in the normal oral mucosa of treated cancer patients. Herein lies a paradox. As noted above, with the exception of 5-FU, NF-κB activation induced by anti-neoplastic agents and radiation, and even TNF-α, is thought to elicit an anti-apoptotic response and be cytoprotective of tumor cells. Yet, in the same patients, massive apoptosis of normal mucosa results in significant injury. How can the presumed activation of NF-κB have such conflicting outcomes in the same patient?
There are at least five possible alternatives to explain the seemingly contradictory effects of NF-κB activation on normal and cancer cells. First, cells may behave differently in response to NF-κB activation. There are compelling data to suggest that activation of NF-κB can be either pro-apoptotic or anti-apoptotic, depending on the target cell (see above). It is possible that NF-κB activation in normal mucosal tissue results in apoptotic signals, while the opposite is true in tumor cells. Given the multi-signal nature of the pro- or anti-apoptotic message, normal and abnormal cells could vary in the scope of up-regulated genes or involvement of co-activators. For example, activation of NF-κB results in up-regulation of Bcl-2 (Fig. 2). Among this class of proteins are those which suppress cell death (Bcl-xL, Mcl-1, Bcl-w, and A-1) and death agonists that promote cell death (Bax, Bak, Bik, Bid) (Huang, 2000). Chemotherapeutic drugs trigger high expression of Bcl-2 via NF-κB. It may be that the Bcl-2 genes activated in normal cells are those associated with cell death, whereas those expressed in cancer cells are cytoprotective. Such an outcome would be consistent with the shifting of balance between bcl-2 family members as a mechanism to produce different apoptotic signals (Bartke et al., 2001).
Different pathways might mediate the responses of normal and tumor cells to chemotherapy and radiation. For example, activation of TNFR1 may result in conflicting apoptotic signals, depending on the subsequent pathway followed (Cheng et al., 2000). Whereas a TNFR1-induced pro-apoptotic signal may occur via pre-formed pro-death caspases, the opposite result may be seen when activation of NF-κB results in the up-regulation of survival genes. Since TNF-α seems to be an early and significant component of mucosal damage, its presence may serve to amplify the destructive effects of radiation or chemotherapy in normal cells if TNF-α causes the activation of proteolytic caspases rather than NF-κB, or if the balance of cell death signals is stronger than cell survival signals. In contrast, if the opposite balance takes place in cancer cells, chemoresistance could be the result.
As noted by Kaltschmidt et al.(2000), the consequences of NF-κB activation represent the culmination of actions and interactions of multiple factors. The deferential way in which NF-κB affects these downstream effectors might provide another way in which its effects on normal and cancer cells differ. Using human colon carcinoma cells, Hellin et al.(1998) found that daunomycin is a potent inducer of both NF-κB and p53. p53 plays a critical role in determining the fate of injured cells, and p53 levels are elevated in cells exposed to DNA-damaging agents such as radiation or chemotherapy. Its activation may result in a pause in cell growth to permit repair, or it may initiate apoptosis. Hellin et al.(1998) found that p53 activation was at least partially regulated by a signal from NF-κB. This finding is relevant to injury, since a subsequent study reported that p53 mediated activation of the mitochondral pathway of apoptosis (Bartke et al., 2001). However, this same study concluded that there was no essential role for NF-κB in p53-induced cell death, and that p53 activation interfered with NF-κB activation. Nonetheless, the way normal or cancer cells process and react to simultaneous activation of p53 and NF-κB might potentially explain differences in response to the same stresses. For example, conflicting data exist as to the effect that the anti-apoptotic NF-κB pathway can have on p53-induced cell death. Whereas Shao et al.(2000) demonstrated that inhibition of NF-κB resulted in p53 induction of cell death, Ryan et al.(2000) found the opposite in that loss of NF-κB activity eliminated p53-mediated cell death. P53 is only one of many genes up-regulated by NF-κB. In addition, the ability of co-activators to work synergistically or antagonistically in normal or tumor cells could markedly affect the way both cell types respond to NF-κB activation.
It also seems that the specific activator may play a role in determining whether NF-κB is pro-apoptotic or anti-apoptotic. Reactive oxygen species (ROS), such as hydrogen peroxide, are generated in tissue following radiation, and these are viewed as important drivers of tissue damage (Riley, 1994). ROS are also cited as effective activators of NF-κB. TNF-α is also well-known as a classic and potent activator of NF-κB. Nonetheless, in a recent study, Kaltschmidt and her co-workers compared the apoptotic (pro- or anti-) behavior of HeLa cells in response to different inducers (Kaltschmidt et al., 2000). They found that inhibition of NF-κB rendered the cells more susceptible to TNF-α-induced cell killing, but protected them from hydrogen-peroxide-induced apoptosis. The authors hypothesized that the signals elicited by the respective death inducers determined whether NF-κB activation was associated with apoptosis or survival. Most importantly, they concluded that NF-κB does not act alone. Rather, the combination of NF-κB activation coupled with other undefined signals influenced the apoptotic outcome. In the case of the cancer patient, one must consider that not only might normal tissue NF-κB activation respond differently to chemotherapy or radiation than cancer tissue, but also that the chemotherapeutic agent, its route of administration, the use of concomitant radiation, and every other treatment permutation might affect the strength of the activation signal and the subsequent biologic endpoints associated with NF-κB activation. Furthermore, activation of each cell type might result in the recruitment of different synergistic or antagonist signals that affect whether NF-κB causes cell death of normal cells and cytoprotection of tumor cells.
The latter implies that normal cells may be more resistant to NF-κB activation than tumor cells. The dose-response relationship between tumor cell destruction and anti-neoplastic therapy and toxicity might support this thought. If NF-κB is in fact cytoprotective to tumor cells, tumor cells must be destroyed by chemotherapy or radiation through an NF-κB-independent pathway (such as the ceramide pathway) in which the apoptotic effect must be sufficient to overcome the anti-apoptotic influence of NF-κB. On the other hand, if there is constitutive NF-κB activity in normal cells that maintains the homeostasis of normal mucosa, one would expect that an anti-apoptotic effect from chemotherapy- or radiation-induced NF-κB should result in mucosal hyperplasia and not cell destruction. Interestingly, one of the first clinical phases of mucositis is the appearance of islands of hyperkeratosis. However, this is then followed by cell death.
The oral environment and, particularly, the indigenous bacteria play a role in the severity and course of mucositis. Xerostomia and myelosuppression result in an increased bacterial load and inhibit patients’ ability to eliminate pathogens effectively. Reducing the bacterial load by the application of antimicrobial agents may favorably effect mucositis (Epstein and Schubert, 1999). It is quite likely that bacteria colonizing ulcerative lesions of mucositis exacerbate the condition through the impact of their cell wall products on NF-κB activation and function. The particular importance of this bacterially mediated effect is that it permits a sustained mechanism for NF-κB activation that starts long after the initial challenge of radiation or chemotherapy. The implication is that bacteria-mediated NF-κB activation results in a tissue outcome that is unfavorable to normal tissue. The particular mechanism(s) by which this occurs is unresolved. However, local tissue defense mechanisms are probably activated (Grassmé et al., 2001).
Among naturally occurring antibacterial agents, defensins and protegrins are of particular interest (Mizukawa et al., 1999; Bellm et al., 2000). Both are peptides found in normal tissue. Defensins are derived from epithelium and have been identified in the oral mucosa (Weinberg et al., 1998). Early trials of porcine leukocyte protegrins have suggested a favorable impact on mucositis. Recently, the antibacterial peptide, PR39, isolated from porcine intestine was shown to bind reversibly to the α7 subunit of the 26S proteasome and thereby block degradation of the NF-κB inhibitor (Gao et al., 2000). Consequently, PR39 effectively inhibits NF-κB gene expression. If defensins have similar activity, their response to the bacterial challenge that follows mucositis could provide a mechanism by which NF-κB could continue to have an impact on the clinical course of the condition beyond simply reducing bacterial load.
Clearly, the potential role of NF-κB in cancer-treatment-related mucosal toxicity has yet to be fully defined. The vast scope of the potential impact of NF-κB activation suggests that it may be a gatekeeper for subsequent biological and clinical events. The paradox of conflicting outcomes of NF-κB activation has been suggested in other areas (DeMeester et al., 2001), but nowhere is it of potentially more significance than in the dichotomous response that is seen when normal and cancer cells react to radiation or chemotherapy. Without doubt, studies of NF-κB’s role in stomatotoxicity will provide therapeutic targets for intervention, as well as serve as a model for other forms of mucosal injury.
(VII) Directions for Future Research
It appears that NF-κB has the potential to play a regulatory role in a wide variety of disease processes. The scope and diversity of both activators of the transcription factor and its target genes are already immense and still require additional definition. How NF-κB exerts its selective biological influence remains to be studied. While it is clear that NF-κB co-activators play a modulating role, the mechanism by which they work and the triggers for their activity are not resolved. Perhaps most intriguing is the ability of NF-κB to cause the expression of genes which have diametrically conflicting activities. Identifying the differential signals that activate NF-κB, defining how they are processed, and establishing why NF-κB activation in one cell type results in a totally different biological outcome than that noted in a different cell type are critical to understanding how this transcription factor ultimately exerts its control of biological events. Clearly, knowledge of each of these milestones represents the development of interventional opportunities.
Signaling pathway for NF-κB activation. Extracellular inducers including reactive oxygen species (ROS), cytokines, viral and bacterial products, ionizing radiation, and cancer chemotherapy drugs activate cytoplasmic kinases (IKK) and lead to the phosphorylation of IκB. The phosphorylated complex is recognized by ubiquitin ligase leading to the attachment of ubiquitin chains. The ubiquinated (Ub) IκB is degraded by the 26S proteosome which permits NF-κB to enter the nucleus (after Makarov, 2000).
NF-κB-mediated up-regulation of different Bcl-2 genes. One possible way to explain the dichotomous responses of normal cells and tumor cells to anti-neoplastic activation of NF-κB might be that different Bcl-2 genes are up-regulated. In normal cells, those genes (Bax, Bak, Bik, Bicl) are associated with apoptosis, while those in tumor cells (Bcl-X2, Mcl-1, Bcl-w, A-1, Bfl-1) might be cytoprotective.