372 sera of 148 melanoma patients and sera of 119 control patients were screened for Raf specific antibody responses by an ELISA assay using a secondary antibody directed against IgG, IgA and IgM. Serial dilutions of the sera were analysed for their OD405
and titers were attributed as described using a cut off OD405
of 0.275 (fig. and ). Positive sera reacted against B-Raf or B-Raf V599E, but not against an unrelated antigen (fig. ). Specificity was further confirmed by coating the ELISA plate using different concentration of the antigen (fig. ) and WESTERN Blot analysis (additional figure 1 [see additional file 1
]). In some positive sera analysed for IgG antibodies, very weak B-Raf specific titers could be detected (fig. ), but most sera were negative for specific IgG antibodies (data not shown). Some positive sera were additionally tested for specificity for the mutated epitope using a competition ELISA by coating B-Raf or B-Raf V599E and competing Raf specific human antibodies by titrating in unbound B-Raf or B-Raf V599E (fig. ). In all cases tested, competition was achieved in every combination at corresponding antigen concentrations, whereas a control serum with non-specific reactivity (fig. ) showed no competition as expected. Taken together these results demonstrate the presence of Raf specific antibodies. In most cases, titers against B-Raf and B-Raf V599E were consistently higher compared to titers for C-Raf (fig. ); only some weakly positive sera showed a response dominated by C-Raf antibodies and in a minority of positive sera the specific response was equilibrated. Even though antigen preparations were not checked for biological activity and correct refolding and therefore a direct comparison between C-Raf and B-Raf specific antibody levels based on these assays is difficult, the data suggest that the Raf specific antibody response in patients is mainly directed against B-Raf. The moderate cross reactivity against C-Raf but not against an unrelated antigen might be explained by the high conformational and sequence similarity between Raf family members. In contrast, the differences observed between the reactivity against B-Raf or B-Raf V599E are weak (fig. ) and suggest that B-Raf and B-Raf V599E are recognized to a similar extend. This notion is further strengthened by the results of the competition ELISA which failed to detect any differences in the pattern of competition for a given serum regardless of the antigen combination used (figure ). Therefore we conclude that the antibodies can not discriminate between the naïve or the mutated form of B-Raf.
Figure 1 Raf specific antibody response in different patients' sera. Analysis of 2 representative positive (42) and negative (12) sera using a secondary antibody detecting human IgG, IgA and IgM is depicted in A. Sera no. 42 and 12 were tested for specificity (more ...)
Figure 2 Competition ELISA to test for specificity for mutated B-Raf. ELISA was performed using 2 positive sera from melanoma patients (A, B) and 1 positive sera from the control group (C) as well as a human serum with known background reactivity (D; serum from (more ...)
Comparison between Raf specific antibody responses in sera from melanoma patients and control patients reveals apparent differences (figure ). Raf specific responses are significantly different from background values for melanoma patients and all Raf variants tested (two tailed Mann-Whitney U test). In contrast, the difference observed within the control group is not significant (Mann-Whitney U test). However, reactivity against C-Raf is rather weak in all sera tested and only 5 of 271 tested sera from melanoma patients reach C-Raf specific titers of 1:300. A comparison between the percentage of patients with B-Raf specific antibody levels above a given titer reveals that sera derived from melanoma patients show consistently higher values compared to the control group (fig. ). In contrast, the frequency of C-Raf specific responses are only marginally higher in the melanoma population and not detectable at titers above 1:300. If positive response is defined as sera with titers of 1:300 or higher, 2.5% of the control group is positive for B-Raf and 1.68% for B-Raf V599E antibodies (fig. ). However, 5.41% (P = 0.12, Fisher's exact test) and 8.86% (P = 0.028, Fisher's exact test) of sera derived from melanoma patients were positive for B-Raf V599E and B-Raf respectively (fig. ). Using this cut off, no control patients could be identified with C-Raf specific antibodies, and only 1.85% (P = 0.329, Fisher's exact test) of the melanoma patients had detectable C-Raf specific antibodies (fig. ). These results suggest that melanoma is associated with higher rates of patients with detectable B-Raf specific antibody responses. The difference between the values obtained for B-Raf V599E and B-Raf specific antibodies, which is expressed in lower scores for B-Raf V599E, is most probably due to differences in the antigen preparation and is less likely due to different specificity as in most cases the scores against B-Raf V599E are in the same range as the scores against B-Raf (see figure ), whereas the C-Raf response consistently shows much lower titers in positive cases and a different shape of the curve (figure ). Taken together, these data strongly suggest that 8.9% of melanoma patients have a B-Raf specific antibody response and 2.5% of the control patients. At this stage, it is far too early to foresee the consequences of this finding for diagnostics or therapy of melanoma. All melanoma patients were in a very advanced, rapidly progressing stage with high tumor loads (stage IV) and consequently the presence of antibodies was not correlated with survival (data not shown). However, at least some patients had low but detectable levels of B-Raf specific IgG antibodies which is in line with the independent finding of B-Raf V599E specific T-cells in melanoma patients [18
Figure 3 Comparison of B-Raf V599E, B-Raf and C-Raf specific antibody responses between control patients or melanoma patients. The comparison has been performed between 119 control sera with 371 sera of melanoma patients tested for B-Raf V599E and 272 sera of (more ...)
Figure 4 Fraction of sera positive at different antibody levels. In A, the percentage of B-Raf positive sera of control patients () or melanoma patients (■) is plotted as a function of the cut off titer, whereas sera are termed positive if the (more ...)
Another interesting point to note is the small, although not significant, number of control patients with detectable levels of Raf specific antibodies. At least two out of 119 control patients showed a very strong B-Raf specific response in the same range as we have observed for positive melanoma patients (figure ). Due to the design of the study, no information is available for this patient group. Serum was sampled during diagnostic procedures, therefore other underlying diseases as a cause for Raf specific antibodies cannot be excluded, including autoimmune diseases or neoplastic diseases of different origin. Furthermore, as the control group has been sampled in the dermatology department, this group might contain patients with high numbers or irregular melanocytic nevi or even unrecognised melanoma.
For different extracellular tumor antigens, including gangliosides, in melanoma, positive associations between the induction of IgM and IgG antibodies and survival time have been reported [21
] and the protective effect was attributed to antibody dependent cellular toxicity [22
]. Therefore, the induction of Raf antibodies might be a favourable goal for the design of melanoma vaccines. However it is questionable whether humoral responses against an intracellular antigen like Raf will have an effect. In this respect it is interesting to ask why responses against intracellular antigens like Raf or survivin [23
] can be observed at all. The most straightforward explanation, which is particularly likely for advanced stage IV melanoma patients, relies on the fact that the partial necrosis of big tumor masses allows the crosspresentation of intracellular proteins. The longitudinal analysis of our patients, e.g. the one depicted in figure , illustrates, that the occurrence of humoral immune responses was correlated rather to tumor burden than therapeutic measures, since the conversion occurred during the period when no therapy was applied.
As oncogenic mutation of B-Raf and even transformation by other oncogenic events is frequently accompanied by Raf overexpression, the induction of an antibody response is not necessarily due to the presence of the mutation. Even though a polyclonal IgM response is very unlikely to be specific for a point mutation, this notion would explain the lack of specificity for the mutational epitope. However, it is also possible that a cellular immune response prior to the induction of Raf specific antibodies has occurred. Such a T-cell response directed against Raf or other tumor antigens, would also result in the lysis of tumor cells and the liberation of intracellular antigens. It is important to mention that the lysis leading to the formation of antibodies against intracellular antigens can be caused by T-cells other than the already observed B-Raf V599E specific CD4+ or CD8+ T-cells. This question still remains to be clarified as up to now only advanced stage patients have been examined. It is obvious that at this stage the occurrence of antibodies is not correlated with prognosis and that it is very unlikely that the antibody response itself will have an influence on the disease. However, as it is rather unlikely that the frequency of a B-Raf specific T-cell response exceeds the frequency of antigen responses, our data can help to set the upper limit for the expected frequency of B-Raf or B-Raf V599E specific T-cells as 8.5% which allows the rational design of search strategies especially for Raf specific CD8+ T-cells.
Even though the detection of B-Raf specific humoral and T-cell responses suggest that B-Raf/B-Raf V599E is immunogenic and that a response could be induced in at least a part of the patients, this notion does not allow conclusions on its suitability as a target for immunotherapy per se. In theory, an optimal tumor antigen is exclusively expressed in the tumor tissue and essential for tumor cell growth and survival to avoid the emergence of escape mutants or antigenic loss, whereas the immunological attack of the tumor cell is independent of the function of the antigen [17
]. To date, many TAA used for the immunotherapy of melanoma including MAGE [24
] or tyrosinase provide no obvious advantage to the tumor cell and therefore there is no pressure to retain the antigen. In contrast, mutated B-Raf has a high prevalence in melanoma and the consequences of its activation, including induction of proliferation and block of apoptosis are well known. However, it is still a matter of debate whether B-Raf activity mainly drives melanoma initiation or whether its function is also relevant for later stages. The first conclusion is supported by studies demonstrating that the mutations already occur in melanocytic nevi at frequencies comparable to late stage melanoma [8
], are not correlated with clinical outcome [25
] and finally the Raf inhibitor Bay 43-9006 seems to have only moderate efficiency in advanced melanoma as reported by Ahmed et al. during the ASCO meeting 2004 [26
]. However, other studies correlated B-Raf mutations with progression rather then initiation [9
] and we and others suggested a role for B-Raf as a negative prognostic factor in metastatic melanoma [25
]. Despite these conflicting data on the clinical relevance of mutated B-Raf at late stages, the high prevalence would already justify to evaluate B-Raf/B-Raf V599E as a target for immunotherapy, as this prevalence is in the same range as for the TAA currently used without any evident advantage for tumor cell growth.