Statistics from Altmetric.com
Editor,—What is one to make of the immunohistochemical study of Iizuka et al (
)? Before addressing the possible scientific implications of their findings, it is worth clarifying a few points with respect to our own research. Iizukaet al state that “the measles hypothesis is based on the theory that measles antibody recognises measles virus itself and the measles virus antigen is uniquely present in Crohn's disease.” Is this the authors' hypothesis? It is certainly not ours. That the measles antibody used in our studies (not that used by the authors in the present study) recognises the measles virus is not in doubt. The appropriate question is “does the antibody detect human antigenic epitopes in addition to measles virus N-protein?” Specificity studies in our own laboratory, including application to mixed preparations of measles virus infected and uninfected cells, application to cell lines infected with other viruses, and tissue studies incorporating in situ hybridisation (ISH) and combined ISH-reverse transcription-polymerase chain reaction (RT-PCR)-ISH for measles virus N-gene in serial tissue sections of infected human tissues suggest that the antibody, when used appropriately, is specific for measles virus.
Our own hypothesis relates not to the unique presence of measles in Crohn's disease but rather to the specific localisation of the virus to the hallmark lesions of this disease—granulomas and secondary lymphoid reactions—that are themselves a likely response to persistent and potentially causative antigen(s). While the focus of our studies has been the exclusive presence of viral antigen in these foci in Crohn's disease, Iizuka et al have assiduously avoided these pathological structures altogether. Instead, they have identified a pattern of non-specific staining in inflamed mucosal biopsies that has never been observed in our own laboratory using the measles virus monoclonal antibody. We have particular anxieties about their statement regarding the cellular localisation of the positive signal. The authors state that, “As Wakefield and colleagues reported, positive cells comprised macrophages, lymphocytes, vascular endothelial cells, fibroblasts, and neutrophils”. This is clearly misleading. We described positive staining in “endothelium, macrophage-like cells, and occasional lymphocytes in foci of granulomatous inflammation”.1 It was not seen elsewhere in tissue sections, either in the presence or absence of inflammation. Iizuka et al have abbreviated this specifically worded statement, removing any reference to granulomatous inflammation and substituting their own findings of signal in fibroblasts and neutrophils. Moreover, the signal that they identified was predominantly cytoplasmic. The signal localisation that we reported was exclusively nuclear, except in rare syncytial giant cells where it was seen additionally in cytoplasm (fig 1). We are also concerned by the implication that we would consider a weakly positive signal to be negative thus potentially biasing the results. Signal strength in Iizuka's studies was evidently on a continuum. In contrast, we found that it could be readily dichotomised into present (discrete and punctate; fig 1) or absent.
A crucial experiment that has been omitted from their study is application of their antibody to primate cell lines either infected or stimulated with something other than measles virus. The indication from the observations in gut tissues is that expression of the epitope recognised by their antibody is a function of inflammation. Is this elicited by specific or non-specific processes?
It is worth emphasising the differences between the study of Iizuka et al and our own. They used a different antibody on tissues processed in an entirely different way (periodate/lysine/4% paraformaldehyde fixed, cryostat sections versus 10% neutral buffered formalin fixed, paraffin sections). They identified markedly different signal characteristics, both in the lineage of positive cells and the subcellular distribution (cytoplasmic versus nuclear). Most importantly, they excluded from analysis the specific foci that were identified as exclusively positive in our studies.
Despite its limitations, this work may contribute to our understanding of why measles has been linked with inflammatory bowel disease (IBD). It is our hypothesis that atypical exposure to the measles virus in early life increases the risk of subsequent IBD among genetically susceptible individuals. Parallels can be drawn from the rare neurological disease subacute sclerosing panencephalitis (SSPE), a delayed sequelae to persistent measles virus infection with a long natural history where the chronic disease manifests several years after the acute measles infection. In SSPE, several atypical characteristics of acute measles infection increase the risk of disease. This is likely to be because elements of the host immune response to the virus are established at the time of acute measles infection and atypical infection may alter the characteristics of this response, increasing the risk of inappropriate immune reactivity.
Measles (or at least components of measles viral material) may persist at very low copy number concentrations, making it difficult to detect using conventional RT-PCR, even with hybrid capture.2Indeed, it has been argued that such retention of viral material may be important in maintaining functional immunity.3 Therefore, two important questions are: (1) What patterns of acute infection increase the risk of inappropriate immune programming and subsequent IBD? (2) What are the likely mechanisms that may result in IBD? The answer to the first question is beginning to be answered by epidemiological studies that identify complex patterns of exposure to measles virus. This includes a close temporal relationship of measles with another paramyxovirus infection taking place during the first five or six years of life.4
It is possible that the work of Iizuka et almay provide a clue to a potential mechanism linking atypical measles infection with IBD. This group reported the presence in inflamed tissue of “measles related antigen” and speculated that this is a human protein as yet unidentified. An atypical measles infection and viral persistence could increase the risk of inappropriate immune reactivity to “measles related antigen”; therefore, molecular mimicry could be one mechanism to explain chronic inflammation in IBD patients. The presence of this protein in other inflammatory disorders of the bowel may indicate that it is produced as part on the inflammatory cascade. This may shed some light on why the onset of IBD appears to be triggered by transient enteric infections. If gut inflammation results in expression of “measles related antigen” in an individual who has been primed by an earlier atypical measles infection, this could trigger an ongoing immune response. Clearly this is highly speculative but if this group has identified a human protein that shares a high degree of homology with measles virus N-protein, it may be a useful tool to investigate the relationship between atypical exposure to measles virus and IBD.
Editor,—We thank Dr Wakefield and colleagues for their interest in our work and we acknowledge that we had initially pursued the same hypothesis that they proposed. However, we have come to a different conclusion through a series of studies1-1-1-3revealing that the antimeasles monoclonal antibody (Seralab, Crawley Down, Sussex, UK) that Wakefield et al used in their study1-4 and in fig 1 above recognises a host antigen as well as measles virus. Furthermore, we proved this cross reaction by identifying a clone with this antibody in a lambda-phage expression library constructed from a typical Crohn's disease patient, by subsequently obtaining a monoclonal antibody to this clone (which we found to be unrelated to measles virus), and finally, by demonstrating that this monoclonal antibody (4F12) reacted with both measles virus and the antigen used for immunisation (the host protein).
Further to our observation that no part of the measles virus genome was detected by reverse transcription-polymerase chain reaction,1-1 our reasons for concluding that what was stained with 4F12 (and the monoclonal antibody that we purchased from Seralab) was not the measles antigen but the host protein unrelated to the measles virus were given in detail in our paper (
) and we see no need to reiterate them here. One final word with regard to the comment of Wakefield et al on our description of molecular mimicry as a possible mechanism for pathogenesis, let us be clear that our report should not be interpreted as support for the hypothesis of measles virus or measles vaccination triggering Crohn's disease.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.