Response to Letter:
Detection of enterovirus in the islet cells of patients with type 1 diabetes: what do we
learn from immunohistochemistry? by Hansson et al.
Sarah J Richardson1, Pia Leete1, Shalinee Dhayal1, Mark A Russell1, Maarit Oikarinen2, Jutta E
Laiho2, Emma Svedin3, Katharina Lind3, Therese Rosenling4, Nora Chapman5, Adrian J Bone6,
Alan K Foulis7, Gun Frisk4, Malin Flodstrom-Tullberg3, Didier Hober8, Heikki Hyoty2,9, Alberto
Pugliese10 and Noel G Morgan1.
1
University of Exeter Medical School, Devon, UK
of Virology, Medical School, University of Tampere, Tampere, Finland
3 Department of Medicine HS, The Center for Infectious Medicine, Karolinska Institutet,
Stockholm, Sweden.
4 Division of Immunology, Genetics and Pathology, Uppsala University, Uppsala, Sweden
5Department of Pathology and Microbiology, University of Nebraska Medical Centre,
Omaha, Nebraska, USA.
6 School of Pharmacy and Biomolecular Sciences, University of Brighton, Moulescoomb,
Brighton, UK
7 GG&C Pathology Department, Southern General Hospital, Glasgow, UK.
8 University Lille 2, CHRU Lille Laboratory of Virology EA3610, France.
9 Fimlab Laboratories, Pirkanmaa Hospital District, Tampere, Finland
10Diabetes Research Institute, University of Miami Miller School of Medicine, Miami, Florida,
USA
2 Department
Word count: 1135
Correspondence to:
Dr Sarah Richardson or Prof Noel G Morgan
Institute of Biomedical and Clinical Sciences
University of Exeter Medical School
RILD Building, Barrack Road
Exeter EX2 5DW
Email: S.Richardson@exeter.ac.uk; n.g.morgan@exeter.ac.uk
We are grateful to Hansson et al [1] for their endorsement of the conclusions reached
following our recent evaluation of the fidelity of immunolabelling achieved with clone 5D8/1
(an antibody raised against an enteroviral capsid protein, VP1) in human pancreas [2]. In
particular, we welcome their support for our view that, when used under optimal
conditions, the labelling achieved with this antibody is likely to retain its specificity for viral
antigens. Nevertheless, we would differ in that we do not consider that a one size fits all
approach to immunolabelling should be adopted when using this antibody. This is because
optimal immunolabelling is not determined solely by the features of the antibody but also
by the quality and preservation of the tissue under study. In our experience, when using
tissue that has been recovered and processed according to recently adopted Standard
Operating Procedures, such as those used within the JDRF nPOD programme [3], then a high
dilution (e.g. 1:2000) of clone 5D8/1 can be employed to achieve optimal, selective,
immunolabelling of viral antigen. However, for those historical collections in which tissues
were recovered at autopsy and then fixed according to non-standardised methods
(including the important tissue collection used in several of our earlier publications [4-6])
the use of high dilutions of clone 5D8/1 is not appropriate. Indeed, in such samples, a
dilution of 1:2000 of clone 5D8/1 fails to detect immunopositivity in proven Coxsackievirusinfected tissues, even after vigorous antigen retrieval. By contrast, a dilution of 1:500 stains
the Coxsackievirus-infected tissues very clearly, whereas controls remain immunonegative.
Thus, we contend that it is not appropriate to attempt to define a single labelling protocol or
antibody dilution. Rather, it is more critical to ensure that appropriate optimisation is
achieved for each set of tissue samples independently (including by the use of validated
positive and negative controls). This has always been our approach and allows us to retain
full confidence in the immunolabelling we reported previously with clone 5D8/1 in human
pancreas [e.g. 4-6].
A second important factor concerns the chemical configuration of the antibody itself. In
various studies cited in their letter, Hansson et al note that high (and, by implication, suboptimal) concentrations of antibody were used. However, it is important to emphasise that
in some of these experiments, the antiserum had been chemically modified prior to its use
(for example, by biotinylation [7,8]) and that the dilutions were then dictated by the
modifications employed. As such, they illustrate again, that it is not feasible to define a
single dilution or a unique set of incubation conditions that apply in all circumstances.
In drawing more general conclusions about the validity and outcomes of
immunohistochemistry, Hansson et al [1] also make reference to the immunolabelling
achieved with two further antisera used in our recent study; those raised against ATP5B and
creatine kinase B respectively. They argue that the labelling achieved with these reagents
may have been sub-optimal because we did not detect “the general expression of these
proteins within all cells in the human pancreas” [1]. This would be an important conclusion
if the underlying premise was true. Unfortunately, it is not. In contrast to their notion that
CKB may be both mitochondrial [9] and ubiquitously expressed [1] this enzyme is known to
adopt a predominantly cytosolic localisation within cells and to display a restricted tissue
distribution [10]. Importantly for the current debate, while we can readily detect CKB in
human heart and stomach by immunohistochemistry, it is not present in the majority of
cells of the pancreas. This concurs fully with data obtained in the mouse [10] and with the
distribution of CKB defined in the Human Protein Atlas (www.proteinatlas.org) using two
independent antibodies. ATP5B, on the other hand, is a mitochondrial component
expressed in a wider range of cells and, accordingly, we show ubiquitous immunolabelling of
this protein within both the endocrine and exocrine compartments in sections of pancreas.
By contrast, we find that clone 5D8/1 does not label ATP5B under any conditions studied
and we were unable to detect any co-localisation of ATP5B and VP1 (labelled with clone
5D8/1) in pancreas. Thus, we consider that our ability to detect the expression of CKB and
ATP5B according to their expected cellular and tissue distribution, bears testimony to the
rigour and validity of the optimisation procedures employed.
Finally, it is important to comment on a further, significant, finding reported in our previous
work [4], that immunopositive labelling was detected when probing the pancreases of
either normal adults (13% of cases) or patients with type 2 diabetes (40% of cases) with
clone 5D8/1. A critical point is that the proportion of immunopositive islet cells was
extremely low in both of these populations (much less than that seen in children with type 1
diabetes) although occasional immunopositive islet cells were present. This might be taken
to imply that non-viral antigens were detected more readily by clone 5D8/1 in the islets of
adults than in children. However, since enteroviruses circulate widely in the population and
have a relatively high degree of tropism for the islet beta-cells, an alternative explanation is
that they can establish (persistent?) infections which are then detected more frequently in
the pancreases of individuals at increasing age. Hence, we suspect that it is not the presence
or absence of virus per se within a beta-cell which is of most significance for the onset of
diabetes (either type 1 or type 2). Rather, it may be the response of the cell to an on-going
viral infection that matters. Conceivably, for type 1 diabetes, it is those children who
respond most vigorously to an early beta-cell enteroviral infection (including by secretion of
interferons and subsequent hyper-expression of islet class I MHC) who are at greatest risk of
triggering islet autoimmunity, while those who respond minimally (or not at all) are
relatively protected. By contrast, in adults predisposed to type 2 diabetes, the presence of
an enteroviral infection within some beta-cells may exacerbate an already defective insulin
secretory response in these cells. This would then be manifest as an apparent increase in
the frequency of viral detection among patients with type 2 diabetes compared to other
adults. In either case, we would not wish to dismiss the detection of immunopositivity
simply as evidence of some random loss of specificity of the immunolabelling technique.
We conclude by emphasising one further point. Namely, that we concur fully with Hansson
et al [1] in calling for additional studies to verify the presence of enterovirus in pancreas. We
accept that irrespective of the perceived degree of fidelity shown by any given antibody, it is
not sufficient to rely solely on a single immunohistochemical reagent when reaching
conclusions about the molecular target. Additional confirmation is required. This may come
in various forms but, for enteroviral infections of beta-cells in human type 1 diabetes, the
isolation and characterisation of viral RNA remains a key priority.
References
[1]
Hansson SF, Korsgren S, Ponten F, Korsgren O (2013) Detection of enterovirus in the
islet cells of patients with type 1 diabetes: what do we learn from
immunohistochemistry? Diabetologia (In press).
[2]
Richardson SJ, Leete P, Dhayal S, Russell MA, Oikarinen M, Laiho JE, Svedin E, Lind K,
Rosenling T, Chapman N, Bone AJ; The nPOD-V Consortium, Foulis AK, Frisk G,
Flodstrom-Tullberg M, Hober D, Hyoty H, Morgan NG. (2013) Evaluation of the
fidelity of immunolabelling obtained with clone 5D8/1, a monoclonal antibody
directed against the enteroviral capsid protein, VP1, in human pancreas.
Diabetologia. In press.
[3]
Campbell-Thompson M, Wasserfall C, Kaddis J, Albanese-O'Neill A, Staeva T,
Nierras C, Moraski J, Rowe P, Gianani R, Eisenbarth G, Crawford J, Schatz D,
Pugliese A, Atkinson M. (2012) Network for Pancreatic Organ Donors with Diabetes
(nPOD): developing a tissue biobank for type 1 diabetes. Diabetes Metab Res Rev. 28,
608-17.
[4]
Richardson SJ, Willcox A, Bone AJ, Foulis AK, Morgan NG. (2009) The prevalence of
enteroviral capsid protein vp1 immunostaining in pancreatic islets in human type 1
diabetes. Diabetologia. 52, 1143-1151.
[5]
Willcox A, Richardson SJ, Bone AJ, Foulis AK, Morgan NG. (2011)
Immunohistochemical analysis of the relationship between islet cell proliferation and
the production of the enteroviral capsid protein, VP1, in the islets of patients with
recent-onset type 1 diabetes. Diabetologia. 54, 2417-2420.
[6]
Richardson SJ, Leete P, Bone AJ, Foulis AK, Morgan NG. (2013) Expression of the
enteroviral capsid protein VP1 in the islet cells of patients with type 1 diabetes is
associated with induction of protein kinase R and downregulation of Mcl-1.
Diabetologia. 56, 185-193.
[7]
Flodström M, Maday A, Balakrishna D, Cleary MM, Yoshimura A, Sarvetnick N. (2002)
Target cell defense prevents the development of diabetes after viral infection. Nat
Immunol. 3, 373-382.
[8]
Flodström-Tullberg M, Hultcrantz M, Stotland A, Maday A, Tsai D, Fine C, Williams B,
Silverman R, Sarvetnick N. RNase L and double-stranded RNA-dependent protein
kinase exert complementary roles in islet cell defense during coxsackievirus infection.
J Immunol. 174, 1171-1177.
[9]
Hansson SF, Korsgren S, Ponten F, Korsgren O (2013) Enteroviruses and the
pathogenesis of type 1 diabetes revisited: cross-reactivity of enterovirus capsid
protein (VP1) antibodies with human mitochondrial proteins. J Pathol. 229, 719-728.
[10]
Sistermans EA, de Kok YJM, Peters W, Ginsel LA, Jap PHK, Wieringa B (1995) Tissueand cell-specific distribution of creatine kinase B: a new highly specific monoclonal
antibody for use in immunohistochemistry. Cell Tissue Res. 280, 435-446.