A novel rat CVB1-VP1 monoclonal antibody 3A6 detects a broad range of enteroviruses

1 SCIENTIFIC REpORTS | (2018) 8:33 | DOI:10.1038/s41598-017-18495-4 www.nature.com/scientificreports A novel rat CVB1-VP1 monoclonal antibody 3A6 detects a broad range of enteroviruses Niila V. V. Saarinen1, Jutta E. Laiho1, Sarah J. Richardson 2, Marie Zeissler2, Virginia M. Stone3, Varpu Marjomäki5, Tino Kantoluoto5, Marc S. Horwitz6, Amirbabak Sioofy-Khojine1, Anni Honkimaa1, Minna M. Hankaniemi1, Malin Flodström-Tullberg 3, Heikki Hyöty1,4, Vesa P. Hytönen 1,4 & Olli H. Laitinen1 Enteroviruses (EVs) are common RNA viruses that cause diseases ranging from rash to paralytic poliomyelitis. For example, EV-A and EV-C viruses cause hand-foot and mouth disease and EV-B viruses cause encephalitis and myocarditis, which can result in severe morbidity and mortality. While new vaccines and treatments for EVs are under development, methods for studying and diagnosing EV infections are still limited and therefore new diagnostic tools are required. Our aim was to produce and characterize new antibodies that work in multiple applications and detect EVs in tissues and in vitro. Rats were immunized with Coxsackievirus B1 capsid protein VP1 and hybridomas were produced. Hybridoma clones were selected based on their reactivity in different immunoassays. The most promising clone, 3A6, was characterized and it performed well in multiple techniques including ELISA, immunoelectron microscopy, immunocytoand histochemistry and in Western blotting, detecting EVs in infected cells and tissues. It recognized several EV-Bs and also the EV-C representative Poliovirus 3, making it a broad-spectrum EV specific antibody. The 3A6 rat monoclonal antibody can help to overcome some of the challenges faced with commonly used EV antibodies: it enables simultaneous use of mouse-derived antibodies in double staining and it is useful in murine models. Enteroviruses (EVs) belong to the family of Picornaviridae and are small non-enveloped viruses with a single-stranded positive-sense RNA genome. There are thirteen enterovirus species, seven of which affect humans. Currently, nearly 300 human EV serotypes/genotypes, including Rhinoviruses have been identified (http://www.picornaviridae.com/enterovirus/enterovirus.htm, May 30th 2017). The majority of EV infections cause only mild common cold-type symptoms, or are asymptomatic, however, in some cases they can trigger more severe acute and chronic diseases such as myocarditis, chronic cardiomyopathy, hand, foot and mouth disease, paralysis, encephalitis and meningitis1,2. In some clinical forms the main contributing EVs that cause a particular disease are well known. Three poliovirus serotypes cause poliomyelitis and Coxsackievirus B (CVB) serotypes are the main cause of acute viral myocarditis1. Correspondingly, EV71 and EV68 have recently been found to cause severe neural symptoms3,4. However, in the case of other diseases like type 1 diabetes (T1D), where a link between EVs and the onset of the disease has been proposed and several candidate serotypes have been identified5–8 the association has not been fully established. Therefore, it is essential to develop tools and techniques that are capable of recognizing multiple EV serotypes for detecting the range of EVs that could potentially contribute to these different diseases. The EV capsid is an icosahedron that consists of repeats of four structural proteins, VP1-4. VP1, VP2, and VP3 are located on the outer surface of the capsid, whereas VP4 resides on the inner capsid surface9. The EV VP1 protein displays the greatest sequence variation between serotypes, but it also contains a conserved immunodominant epitope10,11. Therefore, it is an attractive candidate for the development of diagnostic tools to both detect 1Faculty of Medicine and Life Sciences, University of Tampere, Tampere, Finland. 2University of Exeter Medical School, Exeter, Devon, UK. 3Department of Medicine HS, Karolinska Institutet, Stockholm, Sweden. 4Fimlab Laboratories, Pirkanmaa Hospital District, Tampere, Finland. 5Department of Biological and Environmental Science/ Nanoscience center, University of Jyväskylä, Jyväskylä, Finland. 6Department of Microbiology and Immunology, Life Sciences Institute, University of British Columbia, Vancouver, Canada. Correspondence and requests for materials should be addressed to O.H.L. (email: [email protected]) Received: 21 July 2017 Accepted: 12 December 2017 Published: xx xx xxxx OPEN www.nature.com/scientificreports/ 2 SCIENTIFIC REpORTS | (2018) 8:33 | DOI:10.1038/s41598-017-18495-4 EVs and differentiate between EV groups. There are a number of different EV antibodies available, which vary in their species and serotype specificity, such as Cox mAB 31A2 that binds to certain CVB types12 (Laiho et al. submitted Virchows Archiv 2017) and others that react more widely with different EV types, such as the monoclonal antibody 5D8/113 by Dako Cytomations, Denmark (referred hereafter as 5D8/1) and the 9D5 (Millipore); the last two are widely used in EV research14. These antibodies each have their own strengths and shortcomings with regards to sensitivity and specificity12,14,15. For example, the cross-reactivity of 5D8/1 to smooth muscle tissue is well documented12,15, which may make the interpretation of results more challenging in tissues like heart and intestine. The Cox mAB 31A2, which was raised against the VP1 of CVB3, recognizes CVB1 and CVB312 (Laiho et al., submitted), but not other EV serotypes. This narrow recognition pattern of 31A2 needs to be considered when it is used in clinical diagnostics. Mice are the most commonly used animal model for studying EV infections and pathology16,17 and due to the fact that most monoclonal antibodies against EVs are based on mouse hybridomas, their use in histology can be problematic due to high background caused by endogenous antibodies (mouse-on-mouse unspecific binding)18. Therefore, there is a need for novel non-mouse EV antibodies. To address this, we generated a new rat monoclonal EV antibody, clone 3A6, that solves the host-species issue and that can help to overcome some of the current challenges associated with other EV-antibodies in research and diagnostics. Objectives Our aim was to develop a novel monoclonal antibody that recognizes an EV group-specific epitope located in the N-terminus of VP1 protein and that could be used in multiple applications to study EVs. Rats were chosen for the generation of the hybridoma clones to address the challenges faced with mouse-derived monoclonal antibodies which are currently widely utilized in EV research. Results 3A6 recognizes CVB viruses and shows reactivity towards VP1 N-terminus in ELISA. All 14 generated monoclonal antibodies were strongly positive for CVB1 full-length VP1, but binding to heat inactivated viruses was detectable only with clones 3A6 and 12A4 (Fig.1a). Based on the preliminary screening of the clones using ELISA and immunocytochemistry (data not shown), five clones were selected for further studies: two staining/binding positively to the N-terminus of CVB3 VP1 (clones 3A6 and 12A4) and three binding other parts of VP1 (4D12, 7C1, 9B9) (Fig.1). For the general structure of VP1 protein and its parts, see Fig.2a and b. Interestingly, none of the clones were strongly binding the VP1 C-terminus (Fig.1b).Out of these clones, 3A6 showed the highest potential as an antibody with broad reactivity against EVs as it gave the strongest signal when purified viruses were used as antigens (Fig.1a). Therefore, we focused our further efforts in characterizing 3A6. Figure 1. Reactivity of 14 different monoclonal CVB1-VP1 antibody clones against CVB1-3 viruses and CVB1 VP1 as well as mapping of their binding region in ELISA. (a) Screening of hybridoma supernatants against purified CVB1-3 viruses and CVB1 VP1 protein. Virus samples were heat-inactivated before coating. (b) Screening of hybridoma supernatants against CVB1 VP1, and the Nand C-terminal fragments of CVB3 VP1. Samples were run as duplicates, with 250 ng/well of antigen. The absorbance of blocked wells was subtracted from the sample absorbance values before plotting the data. Error bars show standard deviation of duplicates. In (b) N and C refers to terminal GST fusion constructs. Signal intensity differences between (a) and (b) are due to differing incubation times with substrate. www.nature.com/scientificreports/ 3 SCIENTIFIC REpORTS | (2018) 8:33 | DOI:10.1038/s41598-017-18495-4 The 3A6 binding site on the N-terminus of VP1 overlaps with the enterovirus group specific epitope. As we identified that the epitope of 3A6 resides in the N-terminal region of VP1 (Fig.1) and as it is known that the 5D8/1 epitope also locates in the N-terminal part of VP110, we examined whether 3A6 binds to the same or to a unique epitope. To study this, two peptides (Fig.2c) previously used to characterize the epitope bound by 5D8/119 were utilized in an Immunofluorescent assay (IFA). We used CVB3 VLP as a positive control for binding20. Peptide IFAs confirmed that both the 3A6 and the 5D8/1 antibodies recognize the VP1 peptide RPTNSESIPALTAAE (Fig.3a–d), however neither antibody recognized the overlapping downstream peptide PALTAVETGATNPLV (Fig.3c and d), suggesting that the 3A6 epitope overlaps with that of 5D8/1, and if they are not the same, the epitope bound by 3A6 may be located closer to the N-terminus of VP1. We also studied whether either antibody could bind to peptide epitopes of CKB or ATP5B (Fig.3c and d), two mitochondrial proteins proposed to cross-react with 5D8/115,19. We observed a weak binding of 5D8/1 to the CKB peptide (Fig.3c), whereas 3A6 showed no reactivity towards these two peptides (Fig.3d). 3A6 recognizes CVB1-6 in Western blotting with similar intensity. In order to compare how 3A6 and 5D8/1 antibodies detect different EVs in Western blotting, EV-infected cell lysates and concentrated CVB viruses were used. Both antibodies detected CVB serotypes 1-6, Echovirus 6, Echovirus 30 and Coxsackievirus A9 in infected cell lysates, but neither of them stained Poliovirus 3 (PV3). PV3 VP1 has a lysine at position 40, where in EV-B there is an uncharged amino acid, usually glutamine (Supplementary FigureS1). This might explain thelack of binding in WB, where the epitope is linearized when compared to IHCP and IF. Neither antibody gave unspecific binding to non-infected cells or proteins in extracts of infected cells (Fig.4: lanes 1–2, 3–6 and 8–13). Unlike 5D8/1, 3A6 showed less variation in signal intensity between the different CVB serotypes (Fig.4: lanes 8–13) particularly when comparing their relative binding to CVB5 and CVB6 (Fig.4; lanes 12 and 13 in panels b and c). 3A6 recognizes EV-infected cells in immunocytochemistry (ICC). The ability of 3A6 and 5D8/1 to recognize different EV serotypes was further tested in chromogenic ICC and in IFA by two independent laboratories (Tampere, Finland and Exeter, UK) using an array of EV-infected cells (Table1). Both antibodies recognized the EV-B viruses well, but only 5D8/1 showed some reactivity towards viruses from EV-A species. In addition, both antibodies recognized PV3, which belongs to EV-C. Neither of the antibodies bound the non-infected control cell lines (A549, GMK, Vero, RD and HeLa), or the Adenovirus or Human parechovirus 1 (HPeV1) infected cells (Table1). A slight variation was observed in the staining intensity between IF and IHC-P (Supplementary FigureS2) however, the overall recognition spectrum was similar in both assays. The 3A6 also recognized all tested CVB1 (N = 30) strains in the CVB1 cell microarray21. To test the suitability of 3A6 to recognize both acutely-infected and chronically-infected cells in non-paraffin embedded PFA-fixed coverslips, samples with acutely CVB1-infected PANC-1 cells and chronically CVB1-infected 1.1B4 cells (unpublishedmodelthat is based on previously published data22) were utilized. The Figure 2. Design of the VP1 constructs used in antibody generation and ELISA screening. (a) CVB3 VP1 structure (PDB id 4GB3) presented as cartoon model showing the N-terminal segment (light blue), the core region (green) and the C-terminal part of the protein (dark blue). (b) Schematic illustrations of full length CVB1 VP1 and truncated CVB3 VP1 expression constructs used for immunization and for immunoassays. (c) Recombinant protein constructs and peptides used in the antibody generation (full-length CVB1 VP1) and in ELISA assays (all constructs except Core region). www.nature.com/scientificreports/ 4 SCIENTIFIC REpORTS | (2018) 8:33 | DOI:10.1038/s41598-017-18495-4 coverslips were co-stained with 3A6 and dsRNA antibody23 J2 to visualize the localization of viral protein and dsRNA in the same cell. Both antibodies worked well under the conditions tested. 3A6 staining was distributed throughout the cell cytosol whereas J2 had punctate pattern (Fig.5, Supplementary FigureS3). In addition, dual staining with 3A6 and 5D8/1 in CVB1-infected Vero cells showed that the two antibodies co-localized (Supplementary FigureS4). The relative sensitivity of 3A6 in other EV detection methods was tested with IFA and IHC-P staining using the previously published limited dilution series cell microarray (CMA)24, which involves dilutions of A549 cells infected with CVB1. In both assays the antibody detected virus infections in cells diluted to 10−3, reflecting a sensitivity similar to that as previously obtained with several polyclonal EV antibodies24. However the 3A6 showed lower sensitivity when compared to 5D8/1 in IHC-P (10−6)24. 3A6 detects acute CVB infections in tissue samples by immunohistochemistry. To study the applicability of 3A6 in IHC, we tested it in different mouse and human tissue samples. In EV-infected mouse pancreas, 3A6 showed a clear reactivity towards CVB1 and CVB3 (Fig.6), and CVB4 (not shown), with no background staining in non-infected mouse pancreata (Fig.6c and d). As shown previously25 the CVBs preferentially target the exocrine tissue in the mouse pancreas. Human pancreas and spleen samples from the PanFin network26 with unknown EV status were stained with 3A6, 5D8/1 and rat isotype antibody control in consecutive sections. These samples showed no EV positivity with these antibodies and the rat isotype control was also negative. Furthermore, heart tissue from a confirmed CVB infection was stained with 3A6 and 5D8/1: clear staining was observed with both antibodies in the infected tissue, but not in non-infected control heart (Fig.7). Figure 3. Binding of 3A6 and 5D8/1 antibodies to different antigens in immunofluorescence assay. In (a) the binding of 5D8/1 (grey bars) and 3A6 (black bars) to pVP1 peptide RPTNSESIPALTAAE (500 ng/ well) and CVB3 VLP (250 ng/well) was compared. (b) A dose response curve of 5D8/1(black line) and 3A6 (dotted line) binding to VP1. (c) The binding of 5D8/1 and 3A6 (d) to peptides derived from VP1 (VP1 – RPTNSESIPALTAAE; VP1 Chia – PALTAVETGATNPLV), CKB, ATP5B and whole virus capsid representing CVB3 VLP. www.nature.com/scientificreports/ 5 SCIENTIFIC REpORTS | (2018) 8:33 | DOI:10.1038/s41598-017-18495-4 3A6 labels VP1-proteins in cryo-immuno EM. Cryo-immuno-EM was used to study the VP1 containing cellular compartments in CVB1 infected exocrine pancreatic cells using both 3A6 and 5-D8/1 antibodies. Mouse pancreas tissue showed clear signs of infection 3 days after intraperitoneal inoculation with CVB1 (105 PFU/ mouse) (Fig.8). The presence of virus in the pancreas was also confirmed by PCR, which showed high levels of CVB1 (data not shown). At low magnification, morphological changes due to infection were obvious: apoptotic nuclei, swollen endoplasmic reticulum, emergence of novel membranous structures in the cytoplasm. These Figure 4. Comparison of 3A6 and 5D8/1 enterovirus detection profiles in Western blotting. (a) Sample total proteins visualized with the Stain-Free technology. (b) Immunoblot stained with 5D8/1 (1:3000) and (c) with 3A6 (1:1000). VP1 proteins are ~30 kDa, depending on the EV serotype. Sample order from left to right: lysates from non-infected GMK and Vero cells (8 µg/lane; lanes 1 and 2), CVA9, Echovirus 6, Echovirus 30 and Poliovirus 3 (Sabin) infected cell lysates (8 µg/lane; lanes 3–6), ladder (lane 7) and concentrated and quantified CVB1–6 (0,66 µg) (lanes 8–13). Virus Cell line Antibody Species Serotype Strain Rat anti-VP1 mAb Mouse anti-VP1 mAb Clone 3A6 Clone 5D8/1 EV B CVB1 ATCC GMK ++ +++ EV B CVB2 ATCC GMK ++ +++ EV B CVB3 ATCC GMK ++ +++ EV B CVB4 ATCC GMK ++ +++ EV B CVB5 ATCC GMK +++ +++ EV B CVB6 ATCC GMK ++ ++ EV B Echo3 PB-E3DiT23 GMK + + EV B Echo4 ATCC GMK ++ + EV B Echo6 ATCC GMK +++ +++ EV B Echo9 ATCC GMK ++ ++ EV B Echo11 ATCC GMK +++ + EV B Echo30 ATCC A549 +++ +++ EV B CVA9 ATCC GMK ++ +++ EV B CVA9 PB-CVA9V59 RD ++ ++ EV A CVA2 PB-CVA2V38 RD − + EV A CVA4 PB-CVA4V36 RD − + EV A CVA5 PB-CVA5V43 RD − ++ EV A CVA6 PB-CVA6V303V RD — + EV A CVA10 PB-CVA10V2530 RD — + EV A CVA16 PB-CVA16V55 RD — + EV A CVA16 ATCC Vero — — EV A EV71 PB-EV71Hus GMK — — EV C PV3 Sabin GMK ++ ++ Adeno C VR846 HeLa − − HPeV HPeV1 ATCC A549 − − Table 1. 3A6 mainly recognizes EVs of the B species. A cell microarray including 20 different EVs and two non-EV viruses (Adenovirus C and Human Parechovirus 1) was immunostained with 3A6 in two independent laboratories to verify the recognition spectrum of the antibody. The results of chromogenic immunostaining by 3A6 are shown in comparison to those obtained with 5D8/121,24 are also shown. Recognition scale: +++ = strong, ++ = moderate, + = weak, − = negative. www.nature.com/scientificreports/ 6 SCIENTIFIC REpORTS | (2018) 8:33 | DOI:10.1038/s41598-017-18495-4 tubulovesicular structures, which are characteristic for infected cells27, were antibody positive (3A6 and 5D8/1) and not detected in non-infected cells. 3A6 does not neutralize CVB1. No neutralizing activity of CVB1 was observed with any of the 14 culture hybridoma supernatants. Also, no neutralization activity was seen in the total antisera of the rats immunized with VP1. Figure 5. 3A6 recognizes enterovirus in persistently infected 1.1B4 cells. Confocal images of PFA-fixed chronically CVB1-infected 1.1B4 cells (a–d) and non-infected 1.1B4 control cells (e) double stained with 3A6 (red) and dsRNA antibody J2 (green). Merged images of the double stained infectedand non-infected cells are visualized in (d and e), respectively. www.nature.com/scientificreports/ 7 SCIENTIFIC REpORTS | (2018) 8:33 | DOI:10.1038/s41598-017-18495-4 Discussion EV infections are common, especially in children, and their clinical manifestations range from subclinical to severe or even fatal outcomes. Most EV infections are mild, but tools are required for detecting EVs from clinical samples, including blood, cerebral spinal fluid, stool and in histologic sections in more serious infections. As multiple EV types are linked to diseases like viral myocarditis, encephalitis and type 1 diabetes1,28, reagents that are capable of detecting a broad range of different EV types with both high specificity and sensitivity are needed. Recently, the broad-spectrum EV antibody 5D8/1 from Dako has been regularly used in both clinical and research settings. This antibody is a mouse monoclonal that recognizes an epitope within the capsid protein VP111,19. 5D8/1 mainly identifies EV-B viruses, including CVBs and echoviruses, but also many species A EVs and poliovirus 3 of the C species. However, in non-optimal conditions 5D8/1 has been reported to cross-react with some cellular proteins such as CKB and ATP5B12,15. As such, there is an ongoing effort to produce better antibodies with superior selectivity and specificity due to the limit in the number of specific antibodies that recognize a broad-range of EVs. With this aim, the new broad-reactive rat monoclonal antibody 3A6 was generated. 3A6 detects all CVBs, along with other EV-Bs (Table1 and Fig.4). In addition to broad-reactivity, we aimed to produce an antibody that will further support the findings made with other EV antibodies, and improve detection in different experimental models, including the mouse. The 3A6 antibody was successfully validated for use in different methods including Western blotting, peptide IFA, immuno-TEM, IHC-p and IFA. IHC-P and IFA methods were tested for paraffin and PFA-fixed samples of different origins. 3A6 worked well under the acute infection settings, both in vitro and in vivo derived models as well as with a persistent infection in vitro model. As a rat antibody, the 3A6 will be especially advantageous in mouse models, since no cross-reactivity to mouse cellular proteins was observed. We also tested 3A6 alongside 5D8/1 in human tissue samples, some of which had been previously confirmed positive for EVs1. In human samples 3A6 did not show non-specific binding in non-infected tissues and showed a comparable signal to 5D8/1 in infected tissues (Fig.7). 5D8/1 showed reactivity to smooth muscle that was not seen with 3A6 (not shown). We also observed that antigen retrieval at pH 9 using Tris-buffer resulted in non-specific binding of 3A6 to red blood cells in some cases, but this could be avoided by using citrate buffer at pH 6. We showed that 3A6 and 5D8/1 co-stain the same cellular areas (Supplementary FigureS4), which renders 3A6 as a valid diagnostic tool replicating the results achieved with 5D8/1. For instance: double staining with 3A6 and 5D8/1 tissues in IFA could help validate incoherent findings as true positives or negatives for example in sections with smooth muscle tissue. Figure 6. 3A6 recognizes CVBs in infected mouse pancreata. CVB1and CVB3-infected mouse pancreata were stained with 3A6 by IHC-P and IF (a and b), respectively. Positive signal from the virus recognition is indicated by brown (a) and red colors (b). The non-infected mouse pancreata (c) and (d) were used as negative controls. (a,c and b,d) have the same magnification. Blue color in (b) and (d) is DAPI, which stains the cell nuclei. www.nature.com/scientificreports/ 8 SCIENTIFIC REpORTS | (2018) 8:33 | DOI:10.1038/s41598-017-18495-4 There are relatively few studies showing the high-detail morphological changes occurring in EV infections27. Here, we show several changes related to infection and replication of EVs in the cytoplasm of acinar pancreatic cells, including nuclear apoptosis, endoplasmic reticulum swelling and novel membranous structures in the cytoplasm containing CVB1 VP1 evidenced with either 3A6 or 5D8/1. Importantly, we also show that 3A6 can be used in combination with the dsRNA mouse antibody J2 to simultaneously visualize the localization of viral protein and replicating RNA within infected cells (Fig.5 and Supplementary FigureS3). These results together demonstrate that 3A6 can be used for studying the intracellular localization of the viruses. In Western blotting, the 3A6 recognizes the six CVB serotypes with similar sensitivity, whereas CVB6 detection with 5D8/1 could have been missed, with a shorter exposure time (Fig.4b and c). Thus, the conditions in some assays, such as Western blotting, seem to favor 3A6. Additionally, 3A6 has already proven to be very useful in our own quality controls when culturing and purifying enteroviruses29 and their VP1 proteins (Saarinen, unpublished). Although 5D8/1 is a broadly reactive antibody, it cannot neutralize CVB1 infections in vitro13. Also 9D5 lacks neutralizing capacity, which is understandable since its epitope, just like that in 5D8/1, is not exposed in an intact virus particle14. Similarly, none of our 14 individual monoclonal antibodies targeting VP1, or polyclonal sera from VP1-immunized rats, could neutralize CVBs. This indicates that recombinant proteins which do not represent the Figure 7. Comparison of 3A6 and 5D8/1 detection of virus in neonate heart tissue with IHC-P. Representative images of Clone 5D8/1 (a,c and e) and 3A6 (1:700) (b,d and f) staining in control heart (a and b) and heart from neonates with either a lethal culture-confirmed CVB2 infection (c and d) or CVB4 infection (e and f). Scale bar 20 μm. www.nature.com/scientificreports/ 9 SCIENTIFIC REpORTS | (2018) 8:33 | DOI:10.1038/s41598-017-18495-4 native EV capsid may not be efficient in the generation of antibodies capable of blocking viral infection. The above observations question the rationale of VP1-based subunit vaccines for preventing EV infections28. As the number of broadly-reactive and serotype-specific EV antibodies expands, the construction of a panel of antibodies that overlap in their ability to bind different EVs the detection of group and serotype specific EV infections will improve. Maccari et al.14 suggested that simultaneous use of 5D8/1 and 9D5 allows for the identification of enterovirus groups based on their differences in their ability to bind to different EVs. Adding 3A6 to this panel, which does not bind to EV-A viruses would make such a test more accurate. Importantly, binding differences could be utilized to construct a multiplexing application based on biosensors to identify a wide range of EV serotypes from liquid samples. Figure 8. Immunolabeling of CVB1 infected mouse tissue in cryo-immuno EM with 3A6 and 5D8/1. Sections from CVB1 infected (a–d) and non-infected (e,f) mouse pancreas exocrine tissue acinar cells three days post infection: labelled with antibodies 5D8/1 (panels a, b and e) and 3A6 (panels c, d and f). Infected tissues (a,c) show gross changes during infection, and panels on the right(b,d) show details of cytoplasmic membranous structures positive for CVB1 VP1. Dots indicated by arrows are protein A gold labeled antibodies. Areas of novel membranous structures formed during infection are indicated by asterisks (a,c).