Detection and localization of viral infection in the pancreas of patients with type 1 diabetes using short fluorescently-labelled oligonucleotide probes

Oncotarget1 www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget/ Oncotarget, Advance Publications 2017 Detection and localization of viral infection in the pancreas of patients with type 1 diabetes using short fluorescently-labelled oligonucleotide probes Niels Busse1,*, Federico Paroni1,*, Sarah J. Richardson2, Jutta E. Laiho3, Maarit Oikarinen3, Gun Frisk4, Heikki Hyöty3,5, Eelco de Koning6,7, Noel G. Morgan2 and Kathrin Maedler1 1 Islet Biology Laboratory, University of Bremen, Germany 2 Islet Biology Exeter, University of Exeter Medical School, UK 3 Department of Virology, School of Medicine, University of Tampere, Tampere, Finland 4 Department of Immunology, Genetics and Pathology, Uppsala University, Sweden 5 Fimlab Laboratories, Pirkanmaa Hospital District, Tampere, Finland 6 Department of Internal Medicine, Leiden University Medical Center, Leiden, The Netherlands 7 Hubrecht Institute/University Medical Center Utrecht, Utrecht, The Netherlands * Shared first authors Correspondence to: Federico Paroni, email: [email protected] Correspondence to: Kathrin Maedler, email: [email protected] Keywords: enteroviruses, type 1 diabetes, pancreas, islets, oligonucleotide probes, Pathology Section Received: November 25, 2016 Accepted: January 19, 2017 Published: January 29, 2017 ABSTRACT Enteroviruses, specifically of the Coxsackie B virus family, have been implicated in triggering islet autoimmunity and type 1 diabetes, but their presence in pancreata of patients with diabetes has not been fully confirmed. To detect the presence of very low copies of the virus genome in tissue samples from T1D patients, we designed a panel of fluorescently labeled oligonucleotide probes, each of 17-22 nucleotides in length with a unique sequence to specifically bind to the enteroviral genome of the picornaviridae family. With these probes enteroviral RNA was detected with high sensitivity and specificity in infected cells and tissues, including in FFPE pancreas sections from patients with T1D. Detection was not impeded by variations in sample processing and storage thereby overcoming the potential limitations of fragmented RNA. Co-staining of small RNA probes in parallel with classical immunstaining enabled virus detection in a cell-specific manner and more sensitively than by viral protein. INTRODUCTION Type 1 diabetes (T1D) is a chronic multifaceted disorder that results from selective autoimmunemediated destruction of the insulin producing β-cells. Environmental factors [1], together with genetic predisposition [2], interact cooperatively to initiate chronic islet autoimmunity [3]. Viruses have been proposed as possible initiators of islet autoimmunity and were first implicated as long ago as the nineteenth century although it was not until much later that a clear association was established between mumps and diabetes [4-6]. Improvements in molecular biology subsequently broadened the panel of viruses which are implicated in causing diabetes [7, 8] and the weight of evidence now suggests that coxsackieviruses [9] play a role. In support of this, a clear correlation between enterovirus infection and the onset of T1D was revealed in association studies [10] and via a comprehensive metaanalysis [11]. Coxsackieviruses belong to the Picornaviridiae, and are small positive-sense single stranded RNA viruses, which have been shown recently to induce a persistent, slowly-replicating infection in both myocardium and pancreas. This may result from alteration to the viral genome during the progress of infection including the generation of naturally occurring 5’-deletions [12-14]. Several direct (immunohistochemistry) and Oncotarget2 www.impactjournals.com/oncotarget indirect (serology, isolation of viruses from patients) approaches have confirmed the presence of enterovirus both in the circulation and in the islets of T1D patients [15-20]. Enteroviruses in the pancreas were detected by immunostaining for viral protein (VP1) [17, 21], which is highly expressed under acute viral infection and diminishes in persistent infection and may not be detected under circumstances where viral replication is compromised [22]. Also, non-specific interaction of the VP1 antibody with other cellular proteins has been reported [23, 24]. A well-characterized cohort of human pancreatic donor tissue has been established by nPOD (Network for Pancreatic Organ Donors with Diabetes) and is available, similar to other cohorts, mainly as formalin fixed, paraffin embedded samples (FFPE) [25]. This method ensures preservation of the samples for many years, but a limiting factor is the relatively poor RNA integrity often associated with FFPE preservation which means that analysis of tissue samples by PCR can be difficult [26]. Also, detection of the viral genome in the pancreas has been challenging and more sensitive and reliable methods are required. This has created pressure to develop alternative (accurate and equally sensitive) methods to allow the detection of RNA in single cells within FFPE tissue samples. An obvious candidate is in situ hybridization, where labeled oligonucleotide probes specifically pair with target nucleic acids via Watson-Crick base pairing and some success has been gained using long probes (~50-100nts) labeled with either radioisotopes or enzymes catalyzing chromogenic reactions. However, this approach also has limitations in the face of samples with poor RNA integrity or very low target RNA copy numbers. Thus, improved RNA FISH methods have been developed to overcome these hurdles. Enteroviral RNA was detected by a new generation of RNA probes (QuantiGene ® ViewRNA Assay), which depend on signal amplification [25]. This approach offers the advantage of signal amplification via the use of branched secondary probes but, theoretically, may be affected by several conditions. To create a docking site for the branched probes, two probes must sit adjacent to one-another on the target sequence, effectively lengthening the de facto short probes. However, RNA degradation can still affect the docking of probes and hence signal enhancement. Flexibility in choosing the binding site with the best thermodynamic characteristics may also be compromised. Because of such limitations to the detection of enteroviruses with high accuracy and sensitivity, we present here an adapted method to target single RNA molecules with short (~20nts) fluorescently labeled oligonucleotides in situ. These oligonucleotides anneal to common regions of the RNA genome of members of the coxsackievirus family. Such short singly labeled oligo RNA probes are resistant to RNAse and RNA detection is less affected by target RNA degradation, making these probes more versatile while retaining sensitivity and specificity. Because their binding to the target sequences occurs independently for each probe, this gives probe combinations the advantage that there is a degree of freedom in their positioning without risk of the loss of stringency, efficiency or specificity. To generate a distinct fluorescent signal above background noise, a sufficient amount of labeled probes must bind in close proximity [27-29], thereby ensuring high specificity and considerable flexibility of detection after hybridisation. The use of small contiguous RNA species also overcomes the potential limitation of fragmented target RNA. With our newly established protocol, we successfully detect viral RNA in both cell culture and FFPE tissue sections, in combination with classical immunostaining. Short probes were able to detect viral infection at lower viral loads than classical immunostaining and the method is comparable in sensitivity to that of semi-nested PCR [30]. RESULTS An established protocol for short RNA-oligoprobe labeling in FFPE tissue sections In order to develop a robust protocol for RNAoligoprobe labeling of FFPE pancreatic tissue sections, we initially tested a commercially available probe set targeting the housekeeping gene (GAPDH) in cultured HEK293 cells and in isolated FFPE human islets (Figure 1). Natural, as well as fixative-derived signal noise, is a major problem when employing fluorescence microscopy to detect probes targeting RNA molecules, especially when these are present in low abundance. Thus, while our test probes gave a very specific signal in cultured HEK293 cells (Figure 1A), probing of FFPE human islets generated high background noise both in the absence of probes (Figure 1B) and with the GAPDH probes (Figure 1C), when following the standard protocol. Reduction in the FFPE-derived background signal was achieved by the removal of any remaining paraffin wax crystals (see material and methods). This involved the use of a protocol in which xylene washes were undertaken at high temperatures prior to a step utilizing pepsin/HCl to separate proteins from nucleic acids. In addition, Sudan black was included to reduce the overall FFPE-derived background fluorescence. Using this modified protocol, single positive dots (representing hybridization to as few as one RNA molecule) could be detected within the Abbediffraction limit and were easily discriminated from any residual background noise (Figure 1D). Oncotarget3 www.impactjournals.com/oncotarget Figure 1: An established protocol for short RNA-oligoprobe labeling in FFPE tissue sections. A. GAPDH oligonucleotides were tested in the cell line HEK 293. Single fluorescent spots, each representing one single RNA molecule, are clearly visible. B. Normal deparaffinization procedure shows high background noise in FFPE islet sections even without probes (No Probes) and does not allow clear distinction of probe signal and background, when GAPDH oligonucleotides were added C.; Standard Protocol. Modification of the deparaffinization and post-hybridisation protocol leads to background reduction and increased signal intensity of GAPDH oligonucleotides D.; Modified Protocol. RNA Probes are labeled with Quasar 570 (red) and nuclei were stained with DAPI (blue), scale bar depicts 10µm. Oncotarget4 www.impactjournals.com/oncotarget RNA-oligonucleotide probe design and specificity Probe set CVB_1 included a wide range of group B coxsackieviruses; it consisted of a mixture of 40 short oligonucleotides, each comprising 17-22 nucleotides covering the whole viral genome (see material and methods and Figure 2A). This enabled us to target single RNA molecules. To detect a positive signal, it was determined that at least 17 of the probes in a given set must bind to their target sequence with only one mismatch allowed with respect to the stringency parameters [25]. To test the specificity of the probe set CVB_1, HEK293 cells were fixed and processed after either culture without viruses or following infection with coxsackieviruses CVB3 and CVB2 (Figure 2B) which share a sequence similarity of about 79%. Cells were infected for 2h at an MOI of 5 and the viruses allowed to replicate for 24h prior to fixation and analysis. Probes efficiently detected viral genomes within infected cells (Figure 2B, CVB3 and CVB2) whereas signal was absent from uninfected cells (Figure 2B, non-infected), confirming that the virus-specific probe set had no offtarget effects. The specificity of the probe set was further tested in HEK293 cells infected with cytomegalovirus (CMV), a DNA virus of the herpesviridae family (Figure 2B, CMV) or hepatitis A virus (HAV), a positive ssRNA virus of the picornaviridae family (Figure 2B, HAV). Following infection at an MOI of 5 and incubation for 4 days to ensure viral replication, there was no visible cytopathic effect. The presence of virus was confirmed by RT-PCR (Suppl.Figure 1A). CMV appeared not to be in an active phase of replication as no signal was generated by RTPCR of DNAse-treated samples. In neither HAV nor CMV infected cells were probe-specific hybridization spots detected, confirming the probe specificity. HAV shares partial sequence similarity with CVB3 (<45%) while CMV is a DNA virus; for each, the number of “on target” probes was below the detection limit of the assay. The likelihood that the signals detected in samples were non-specific was further excluded by staining CVB3 infected islets in the absence of the probe set and by staining uninfected cells with the probe set. In each case, the negative controls delivered no staining, whereas virally infected cells yielded positive signals (Suppl.Figure 1B). RNAse A treatment abolished the signal from infected cells and confirmed that the probe set is specific to viral RNA (Suppl.Figure 1B). We further tested the CVB_1 probe set on an array of cell lines previously generated for use with Quantigene® ViewRNA virus probes [21]. Green monkey kidney cells (GMK and Vero), the human cervix (HeLa), alveolar (A549) epithelial carcinomic and rhabdomyosarcoma (RD) muscle cells were infected with viruses from the enterovirus groups A and B or adenovirus (DNA virus) [21] (Table 1 and Suppl.Figure 2). In line with the results obtained with Quantigene® ViewRNA by Laiho et al. [21] the CVB_1 probe set yielded positive staining for viruses of both groups A and B, while it did not stain cells infected with adenovirus and human parechovirus 1 (HPeV1) (0/ 40 probes binding). Also, there was no binding to sequences from coxsackievirus A5 (11/40 probes theoretically match the virus sequence). Table 1: RNA oligonucleotide staining of different picornaviridae and control viruses Virus Strain Result Virus Strain Result EV71 PB-EV71Hus ++ Echo3 PB-E3DiT23 ++ CVB1 ATCC ++ Echo4 ATCC ++ CVB2 ATCC ++ Echo6 ATCC + CVB3 ATCC ++ Echo9 ATCC + CVB4 ATCC ++ Echo11 ATCC + CVB5 ATCC ++ Echo30 ATCC ++ CVB6 ATCC ++ PV3 Sabin + CVA2 PB-CVA2V38 ++ HPeV1 ATCC - CVA4 PB-CVA4V36 ++ Adenovirus C VR846 - CVA5 PB-CVA5V43 - A549 cells -- CVA6 PB-CVA6V303V + RD cells -- CVA9 ATCC ++ Vero cells -- CVA10 PB-CVA10V2530 ++ HeLa cells -- CVA16 PB-CVA16V55 ++ GMK cells -- Probe specificity was tested on a cell array (FFPE); different cell lines were spotted either as uninfected controls or infected with different viruses. The results obtained with RNA probes set CVB_1 are displayed. Representative images are shown in Fig. S2. Oncotarget5 www.impactjournals.com/oncotarget Figure 2: RNA-oligonucleotide probe design and specificity. A. Scheme of custom-designed oligonucleotide (CVB_1) annealing throughout the viral genome. B. Viral RNA probes were tested against non-infected, CVB3 and CVB2 (100% and 79% similarity to consensus sequence, respectively; positive control), CMV, (DNA virus; negative control) and HAV (<45% similarity to consensus sequence) infected HEK 293 cells. Cells were infected with an MOI of 5 and harvested after 24h (control, CVB3, CVB2) or 4 days (CMV, HAV) postinfection. RNA Probes are labeled with Quasar 570 (red) and nuclei were stained with DAPI (blue), scale bar depicts 10µm. Oncotarget6 www.impactjournals.com/oncotarget RNA-oligonucleotide probe sensitivity and consistency Next, we evaluated sensitivity of the CVB_1 probe set. Our RNA-FISH system was compared with two of the most widely used and well established techniques for virus detection: RT-PCR and immunohistochemistry using an antibody against the viral capsid protein 1 (VP1). RT-PCR is the most powerful and specific tool for RNA detection; but RNA accessibility and degradation are two important limiting factors. To compare the sensitivity of the RNA probes with analysis by PCR, the pancreatic line CM9 [31] and HEK293 cells were infected with CVB3. Since both RT-PCR and RNA-FISH can, in principle, detect the presence of a single RNA molecule, we used successive 10-fold serial dilutions of virus prior to infection of cells and to compare the sensitivity of the methods. Cells were plated in duplicates and the virus (CVB3; starting MOI of 100) was centrifugally inoculated at 16°C for 1h to synchronize the infection. After inoculation, any unbound viruses were removed by washing and the cells incubated at 37°C for an additional hour, to allow virus internalization and genome release from the capsid, before fixation or cell lysis for RNA extraction. Both RNA-FISH (Figure 3A, 3B and Suppl.Table 1) and RT-PCR (Figure 3C and Suppl.Table 2) were able to detect the presence of viral genomes, even at the highest dilution (10-8) (Figure 3B, 3C). Using RNA-FISH probes, single RNA molecules were detected and a plateau reached at a dilution of 10-3 virus by both visual counting and RT-PCR (Figure 3B-3C). The specificity of the RT-PCR results were confirmed by examination of dissociation curves, which were identical in all cases while samples from non-infected cells showed no signal (data not shown). We next compared the efficiency of the CVB_1 probe set with the widely used VP1 (clone 5-D8/1, Dako cytomations) antibody, using a cell array of the human alveolar basal epithelial cell line A549 infected with CVB1 for 2, 4 and 6 h to generate a population representing different stages of infection [27]. After infection, cells were serially diluted with uninfected A549 cells to achieve a range from undiluted to 10-8. The cells were then fixed and paraffin embedded. When employed at a dilution of 1:2000 (which ensures specificity and minimizes the possibility of false positives [23, 24]), the VP1 antibody yielded positive signals only at dilutions of 10 -1 or lower, whereas the RNA-FISH probes were able to detect viral RNA even at the highest dilution of 10-8 (Figure 3D, 3E). The design of the RNA-FISH system should circumvent the problem of RNA degradation since it employs multiple probes to detect the target RNA. RNA fragmentation frequently occurs in, for example, autopsy samples, where processing and storage under RNAse free conditions is unlikely. We therefore tested FFPE CVB1infected GMK cell sections, which had been infected and processed at the same time, but then cut and processed at different times (covering periods between 2012 and 2015) and stored at room temperature. The probes showed similar signals in all three samples regardless of the processing and storage time. Importantly, uninfected controls were negative (Suppl.Figure 3). Coupling RNA-FISH and immunohistochemistry Important advantages of the use of RNAFISH probes relate not only to their high sensitivity and specificity but also their ability to localize RNA molecules within specific cells of tissues such as the pancreas. We, therefore, investigated the localization and distribution of the signal emanating from the viral probes within the pancreas and compared this with detection by immunohistochemistry with the widely used VP1 antibody. Figure 4A shows a schematic representation of the expected profile of virus-staining: RNA probes (red) are not expected to anneal to the viral RNA while it is packaged within the capsid. However, once released within the cell, the probes should bind. Conversely, the VP1 antibody (green) should always bind to the capsid surface or, conceivably, to free VP1 which has not been incorporated into capsids. To visualize any differential labeling, human CM9 cells were infected with centrifugally inoculated CVB3 at an MOI of 1000. (Figure 4B). Anti-VP1 was detected at the plasma membrane and in the cytoplasm while RNA probes were localized exclusively in the cytoplasmic area (Figure 4B, middle panel and Suppl.Figure 4 for enhanced signals). As depicted at larger magnification, signals from the RNA molecules co-localized, or were in close proximity with those arising from the capsid protein, but the two were not superimposed. To verify the staining patterns, we also performed VP1 immunohistochemistry prior to cell permeabilization and subsequent RNA-FISH, so that the signals should not co-localize. As shown in Figure 4B (right panel), under these conditions the signals arising from each method of detection were clearly separate. Uninfected CM9 cells were used as controls and showed no nonspecific signal for either anti-VP1 or the RNAprobes (Figure 4B). Successful double staining of viral RNA and VP1 was confirmed in FFPE infected mouse spleen. Spleen sections from mice infected with CVB1 were stained with the CV_1 probe set to detect the viral genome and subsequently with anti-VP1 in an additional round of staining of the same section. Both techniques showed positive staining within the same regions of the samples (Suppl.Figure 5A). Double staining of FFPE CVB3 infected cultured rat INS1-E cells for insulin and virus probes showed colocalization of viral RNA within insulin positive INS1E cells (Figure 4C). Close proximity of viral RNA and insulin was also found in a Coxsackie infected neonatal Oncotarget7 www.impactjournals.com/oncotarget Figure 3: RNA-oligonucleotide probe sensitivity. A. CM 9 and HEK 293 were infected with a dilution series (MOI 102-10-8) of CVB3 and stained with custom-designed oligonucleotides (CVB_1). Representative images of CM9 and HEK293 cells infected with either the highest or lowest dilution of CVB3 of the series are shown. White arrows highlight single viral spots. B. Ten single images were acquired for each dilution and single fluorescent spots were manually counted. Results for HEK 293 (diamonds) and CM 9 (squares) are displayed as single spots per cell in logarithmic scale. In total, viral particles were counted in 4470 HEK293 cells and CM9 cells. C. Viral RNA from a parallel experiment was extracted and analyzed by PCR; MOI of 100 (102) was set as 100%. D. Viral RNA (red) and VP1 (1/2000; green) staining on a CVB1-dilution array of FFPE infected GMK cells mixed with uninfected cells. RNA Probes were labeled with Quasar 570 (red) and nuclei were stained with DAPI (blue), scale bar depicts 10µm. E. Summary of the viral RNA and VP1 signals obtained from a CVB1-dilution array of FFPE infected GMK cells. Oncotarget8 www.impactjournals.com/oncotarget Figure 4: Coupling RNA-FISH and Immunohistochemistry. A. Theoretical scheme of viral RNA and VP1 co-staining. Initially, labeled oligonucleotides cannot bind viral RNA within the capsid, but only when the virus is released. On the other hand, VP1 antibody can bind to the capsid surface. Over time as more RNA is released, more probes can anneal to their target sequence. When a sufficient amount of oligonucleotides is bound, green (VP1) and red (RNA) signal are visible in close proximity. B. CM9 cells were infected with CVB3 (MOI 1000) for 1h at 16°C, fixed and probed for viral RNA and VP1. VP1 (green) was found in close proximity with viral RNA (red), when capsid and RNA were present in the cytosol area (B; middle panel). When VP1 staining was performed before cell permeabilization no colocalization of viral RNA/VP1 was detected (B; right panel). No cross-reactivity was found with cellular proteins or RNAs (B; left panel). C. Co-staining and localization of viral RNA (red) and insulin (green) in CVB3-infected INS-1E cells and HgCl2 fixed paraffin embedded CVB-infected neonatal pancreas. Samples were stained with RNA FISH probes first, analyzed and then stained for insulin. The non-granular insulin staining in this slide appeared also by classical single insulin immunohistochemistry and is caused by the tissue condition. Nuclei were visualized by DAPI staining (blue); scale bar depicts 10µm. Oncotarget9 www.impactjournals.com/oncotarget mouse pancreas (Figure 4C), confirming the established double-staining protocol also in HgCl 2 fixed and paraffin embedded cells. To demonstrate efficacy for virus staining in various tissues, we stained a section from a coxsackieinfected neonatal heart and found cells displaying a strong signal corresponding to viral RNA (Suppl.Figure 5B). Viral RNA localization in the pancreas in T1D Having established the validity of our approach, we then used an enterovirus genome alignment (Suppl. Figure 6) to design two additional probe sets (CVB_2 and CVB_3; Suppl.Tables 4,5), which would complement the first set and allow detection of the entire range of group B enteroviruses. A scheme of the binding positions of the newly designed probe sets is shown in Figure 5A. We then used a combination of the three probe sets for a blinded analysis of autopsy pancreatic tissue recovered from nondiabetic and T1D patients having remaining residual β-cells from a UK cohort [31, 32]. As shown in Table 2, we were able to detect the presence of enteroviruses in 7 of 8 pancreas samples from T1D patients and in 2 of 8 nondiabetic controls. While viral RNA was found within the insulin-positive islet area in 6 of 8 T1D patients (Table 2 and Figure 5B), 5 of 8 pancreata also yielded positive signals for enteroviral RNA in the exocrine area. A single T1D patient from the UK collection had viral RNA exclusively in the exocrine area of the pancreas and not in the islets (although in this patient, only 1 fragmented and 2 normal islets were found throughout the whole section). Similarly, this distribution of viral RNA was also observed in a pancreas transplant biopsy from a patient with T1D who had developed recurrent disease from the nPOD-Transplantation cohort [32] (Figure 5C). In one nondiabetic control, we also found viral RNA in islets. Comparison of our results with anti-VP1 staining on the same samples performed separately in Exeter, 66% concordance was achieved. Thus, while virus was detected using viral probes in 6 out of 8 T1D pancreata within the insulin-positive islet area, VP1 staining was present in only 4/8 pancreata. DISCUSSION In this study we present a robust method to identify and localize enteroviral RNA in FFPE tissue, which was established for the analysis of viral infection in pancreata from patients with T1D. The use of RNA probes allowed the discrimination of virally-encoded RNA from virus replication loci with high fidelity and sensitivity. Adapting a method from Raj et al. [28, 29], we created a set of short fluorescently labeled oligonucleotide probes that anneal to common regions of the coxsackievirus family in situ to target single RNA molecules. Probe sets are easily developed using the online Stellaris® RNA FISH Probe Designer. With three different sets of probes we were able to localize viral RNA in all tested tissue including FFPE tissues from patients with proven coxsackievirus infection and in the pancreas of 7 Table 2: RNA oligonucleotide staining of human pancreases Sample Group ID Age Sex Fixation Viral RNA +* Islet viral RNA + VP1+ T1D E560 42 F BF ++++ + VP1+ T1D 11746 6M HgCl2++ + VP1+ T1D E375 11 F FS + + VP1+ T1D E554A 7MBouin - § + VP1 – T1D E428 5M BF ++ + VP1 – T1D E514 23 M BF - + VP1 – T1D E235 6M HgCl2- - § VP1 – T1D 8869 8M HgCl2+ - § Non-diabetic control 8579 7 - HgCl2- - Non-diabetic control 12054 7 - HgCl2- - Non-diabetic control 330/71 47 M BF - - Non-diabetic control 21/89 4F BF ++ - Non-diabetic control 274/91 6M BF - - Non-diabetic control 191/67 25 M BF - - Non-diabetic control 315/89 9M BF + + Non-diabetic control 540/91 11 M BF - - Summary of the comparative study of pancreata from T1D and non-diabetic control patients. Sections were first stained for viral RNA, with a combination RNA probe sets CVB_1, CVB_2 and CVB_3, followed by insulin staining. - = less than 10 fully-infected cells, + = 10 to 30 fully-infected cells, ++ = 30 to 100 fully-infected cells, +++ = 100 to 200 fully-infected cells, ++++ = more than 200 fully-infected cells in the exocrine. §E554A showed ten fully-infected cells, E235 contained just 5 islets on the section and 8869 just 2 full and 1 fragmented islets (very small sections). Oncotarget16 www.impactjournals.com/oncotarget M, Mall G and Kandolf R. Ongoing enterovirus-induced myocarditis is associated with persistent heart muscle infection: quantitative analysis of virus replication, tissue damage, and inflammation. Proc Natl Acad Sci U S A. 1992; 89(1):314-318. 23. Richardson SJ, Leete P, Dhayal S, Russell MA, Oikarinen M, Laiho JE, Svedin E, Lind K, Rosenling T, Chapman N, Bone AJ, n PODVC, Foulis AK, et al. 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. 2014; 57(2):392-401. 24. Hansson SF, Korsgren S, Ponten F and Korsgren O. Enteroviruses and the pathogenesis of type 1 diabetes revisited: cross-reactivity of enterovirus capsid protein (VP1) antibodies with human mitochondrial proteins. J Pathol. 2013; 229(5):719-728. 25. Laiho JE, Oikarinen S, Oikarinen M, Larsson PG, Stone VM, Hober D, Oberste S, Flodstrom-Tullberg M, Isola J and Hyoty H. Application of bioinformatics in probe design enables detection of enteroviruses on different taxonomic levels by advanced in situ hybridization technology. J Clin Virol. 2015; 69:165-171. 26. Skog O, Ingvast S and Korsgren O. Evaluation of RTPCR and immunohistochemistry as tools for detection of enterovirus in the human pancreas and islets of Langerhans. J Clin Virol. 2014; 61(2):242-247. 27. Femino AM, Fay FS, Fogarty K and Singer RH. Visualization of single RNA transcripts in situ. Science. 1998; 280(5363):585-590. 28. Raj A and Tyagi S. Detection of individual endogenous RNA transcripts in situ using multiple singly labeled probes. Methods Enzymol. 2010; 472:365-386. 29. Raj A, van den Bogaard P, Rifkin SA, van Oudenaarden A and Tyagi S. Imaging individual mRNA molecules using multiple singly labeled probes. Nat Methods. 2008; 5(10):877-879. 30. Laiho JE, Oikarinen M, Richardson SJ, Frisk G, Nyalwidhe J, Burch TC, Morris MA, Oikarinen S, Pugliese A, Dotta F, Campbell-Thompson M, Nadler J, Morgan NG, Hyoty H and Group Jn-V. Relative sensitivity of immunohistochemistry, multiple reaction monitoring mass spectrometry, in situ hybridization and PCR to detect Coxsackievirus B1 in A549 cells. J Clin Virol. 2016; 77:2128. 31. Gragnoli C. The CM cell line derived from liver metastasis of malignant human insulinoma is not a valid beta cell model for in vitro studies. Journal of cellular physiology. 2008; 216(2):569-570. 32. Pugliese A, Vendrame F, Reijonen H, Atkinson MA, Campbell-Thompson M and Burke GW. New insight on human type 1 diabetes biology: nPOD and nPODtransplantation. Curr Diab Rep. 2014; 14(10):530. 33. Dotta F, Galleri L, Sebastiani G and Vendrame F. Virus Infections: Lessons from Pancreas Histology. Curr Diabetes Rep. 2010; 10(5):357-361. 34. Morgan NG and Richardson SJ. Enteroviruses as causative agents in type 1 diabetes: loose ends or lost cause? Trends Endocrinol Metab. 2014; 25(12):611-619. 35. KOCH R. Ueber bakteriologische Forschung. . Verhandlungen des X Internationalen Medichinischen Congresses, Berlin 1890, August Hirshwald, Berlin, Germany, 10th International Congress of Medicine, Berlin, Germany 1. 1891. 36. Chapman NM and Kim KS. Persistent coxsackievirus infection: enterovirus persistence in chronic myocarditis and dilated cardiomyopathy. Curr Top Microbiol Immunol. 2008; 323:275-292. 37. Viegas MS, Martins TC, Seco F and do Carmo A. An improved and cost-effective methodology for the reduction of autofluorescence in direct immunofluorescence studies on formalin-fixed paraffin-embedded tissues. Eur J Histochem. 2007; 51(1):59-66. 38. Erben T, Ossig R, Naim HY and Schnekenburger J. What to do with high autofluorescence background in pancreatic tissues - an efficient Sudan black B quenching method for specific immunofluorescence labelling. Histopathology. 2016; 69(3):406-422. 39. Van de Winkle M, Maes E and Pipeleers D. Islet cell analysis and purification by light scatter and autofluorescence. Biochem Biophys Res Commun. 1982; 107(2):525-532. 40. Willcox A, Richardson SJ, Bone AJ, Foulis AK and Morgan NG. Analysis of islet inflammation in human type 1 diabetes. Clin Exp Immunol. 2009; 155(2):173-181. 41. Foulis AK, Liddle CN, Farquharson MA, Richmond JA and Weir RS. The histopathology of the pancreas in type 1 (insulin-dependent) diabetes mellitus: a 25-year review of deaths in patients under 20 years of age in the United Kingdom. Diabetologia. 1986; 29(5):267-274. 42. Ardestani A, Paroni F, Azizi Z, Kaur S, Khobragade V, Yuan T, Frogne T, Tao W, Oberholzer J, Pattou F, Kerr Conte J and Maedler K. MST1 is a key regulator of beta cell apoptosis and dysfunction in diabetes. Nat Med. 2014; 20(4):385-397. 43. Brack K, Frings W, Dotzauer A and Vallbracht A. A cytopathogenic, apoptosis-inducing variant of hepatitis A virus. J Virol. 1998; 72(4):3370-3376. 44. Abhat A. Low temperature latent heat thermal energy storage: heat storage materials. Solar energy. 1983; 30(4):313-332. 45. Olert J, Wiedorn KH, Goldmann T, Kuhl H, Mehraein Y, Scherthan H, Niketeghad F, Vollmer E, Muller AM and Muller-Navia J. HOPE fixation: a novel fixing method and paraffin-embedding technique for human soft tissues. Pathol Res Pract. 2001; 197(12):823-826. Oncotarget17 www.impactjournals.com/oncotarget 46. Rutledge RG and Stewart D. Assessing the performance capabilities of LRE-based assays for absolute quantitative real-time PCR. PLoS One. 2010; 5(3):e9731. 47. Ausubel FM, Brent R and Kingston RE. Short Protocols in Molecular Biology. John Wiley & Sons. 2003. 48. Kearse M, Moir R, Wilson A, Stones-Havas S, Cheung M, Sturrock S, Buxton S, Cooper A, Markowitz S, Duran C, Thierer T, Ashton B, Meintjes P and Drummond A. Geneious Basic: an integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics. 2012; 28(12):1647-1649.