Expression of the Infectious Salmon Anemia Virus Receptor on Atlantic Salmon Endothelial Cells Correlates with the Cell Tropism of the Virus

Abstract

Infectious salmon anemia (ISA) is a World Organization for Animal Health (OIE)-listed disease of farmed Atlantic salmon, characterized by slowly developing anemia and circulatory disturbances. The disease is caused by ISA virus (ISAV) in the Orthomyxoviridae family; hence, it is related to influenza. Here we explore the pathogenesis of ISA by focusing on virus tropism, receptor tissue distribution, and pathological changes in experimentally and naturally infected Atlantic salmon. Using immunohistochemistry on ISAV-infected Atlantic salmon tissues with antibody to viral nucleoprotein, endotheliotropism was demonstrated. Endothelial cells lining the circulatory system were found to be infected, seemingly noncytolytic, and without vasculitis. No virus could be found in necrotic parenchymal cells. From endothelium, the virus budded apically and adsorbed to red blood cells (RBCs). No infection or replication within RBCs was detected, but hemophagocytosis was observed, possibly contributing to the severe anemia in fish with this disease. Similarly to what has been done in studies of influenza, we examined the pattern of virus attachment by using ISAV as a probe. Here we detected the preferred receptor of ISAV, 4-O-acetylated sialic acid (Neu4,5Ac₂). To our knowledge, this is the first report demonstrating the in situ distribution of this sialic acid derivate. The pattern of virus attachment mirrored closely the distribution of infection, showing that the virus receptor is important for cell tropism, as well as for adsorption to RBCs.

INTRODUCTION

Infectious salmon anemia (ISA) is a significant disease of farmed Atlantic salmon (Salmo salar L.). It was first identified in Norway in 1984, and since then, disease outbreaks have occurred globally. The disease is listed by the World Organization for Animal Health (OIE) and is caused by a negative-sense, segmented single-stranded RNA (ssRNA) virus in the Orthomyxoviridae family (37). Like other orthomyxoviruses, ISA virus (ISAV) has two glycoproteins embedded in the viral envelope, forming surface spikes. These proteins are the hemagglutinin esterase protein (HE) with combined hemagglutinating (receptor binding) and receptor-destroying enzyme (RDE) activity and the fusion protein (F) with fusion activity (1, 11, 13). Similarly to influenza C virus, ISAV attaches to O-acetylated sialic acids on target cells. However, while the influenza C virus receptor is N-acetyl-9-O-acetylneuraminic acid (Neu5,9Ac₂) (34), ISAV attaches to N-acetyl-4-O-acetylneuraminic acid (Neu4,5Ac₂) (20). Furthermore, the ISAV RDE is a sialate-4-O esterase (20). Although ISAV proteins have basic functions similar to those of other orthomyxoviruses, there are no significant nucleotide or amino acid sequence similarities (6). Two distinct pathotypes of ISAV have been identified, one highly pathogenic type causing classical ISA and one low-pathogenic type (ISAV-HPR0), which is associated with subclinical infection in gills (5, 29, 32). A hallmark of the highly pathogenic ISAV strains is differential deletions in the HE stalk region immediately upstream of the transmembrane region (7, 34).

ISA is a relatively novel disease with poorly understood pathogenesis. Initially, descriptions focused on the severe anemia and the findings of hemorrhagic liver necrosis. Other common findings are increased hemophagocytosis, ascites, petechia, and congestion in multiple organs suggesting circulatory failure (10, 47, 48). The first cases of ISA on the east coast of Canada were named “hemorrhagic kidney syndrome,” as the main pathological findings were dominated by changes in the kidney, including hemorrhages, tubular necrosis, and circulatory changes (4). Thus, the amount and location of pathological changes may vary, but anemia and general circulatory disturbances are constant features. Several studies focusing on virus detection have demonstrated virus in endothelial cells by electron microscopy (EM), immune detection, and in situ hybridization (ISH) (12, 18, 19, 27). The virus has also been demonstrated in leukocytes by EM and ISH (27, 35), though the role of these cells in disease development has not been thoroughly explored.

The best-studied viruses within the Orthomyxoviridae family, the low-pathogenic avian influenza viruses, mainly cause mild, localized infections of the upper respiratory tract in mammalian and avian species. Highly pathogenic avian influenza (HPAI) viruses, on the other hand, cause serious infections of the lower respiratory tract, including systemic infection in poultry and some mammalian species, including humans, cats, and ferrets (14, 28, 37, 40, 42). Endotheliotropism appears to be a general phenomenon in the systemic influenza virus infections, and hemorrhages and edema indicate an affliction of the vascular system (24, 40, 45, 52).

Attachment to its host cell is the first step in the replication cycle of viruses. Thus, the interaction between the viral attachment protein and its cellular receptor is among the critical molecular determinants that regulate host, tissue, and cell tropism. The viral attachment protein for orthomyxoviruses, the hemagglutinin (HA), attaches to sialic acid-containing receptors. Using virus proteins as probes in a method called virus histochemistry, several studies have shown the importance of the distribution of virus receptor for the influenza virus cell tropism (14, 23, 30, 50, 51).

The purpose of the present study was to characterize and explain the tissue and cell tropism of highly pathogenic ISAV. Here we demonstrate that the virus is endotheliotropic and causes a generalized infection of the vascular system. We also demonstrate virus attachment to erythrocyte (RBC) surfaces. Using virus histochemistry, we show that this tropism is determined by the distribution of Neu4,5Ac (4-O-acetylated sialic acids), the preferred receptor for the virus (20), which is located on the luminal surface of endothelial cells.

MATERIALS AND METHODS

Virus and cells.

The Norwegian ISAV isolate Glesvaer/2/90 (8) was used throughout this study. Cultures of ASK cells (9) grown at 20°C in Leibovitz L-15 medium (L-15) supplemented with 10% fetal bovine serum, glutamine (4 mM), and gentamicin (50 μg ml⁻¹) were used for propagation. The cells were incubated at 15°C after inoculation with virus. Unless otherwise stated, cells with the fluid overlay removed were inoculated with virus at a multiplicity of infection of 0.1 in serum-free L-15. Infectivity titrations were done by endpoint titration in 96-well culture plates as previously described (13).

Preparation of ISAV HE antigen.

Infected cell cultures in 75-cm² tissue culture flasks were harvested following 3 to 5 days of incubation, i.e., at the time when cytopathic effect (CPE) was apparent but cells had not detached. The cells were washed twice in ice-cold L-15 and scraped off using cell scrapers. The HE protein was then extracted using Tween 80 and ether (36). Briefly, Tween 80 was added under constant stirring to a final concentration of 0.125% (vol/vol) at +4°C and stirred for 5 min. Diethyl ether (0.5 vol) was then added to the preparation, further incubated for 15 min at room temperature, and centrifuged at 1,000 × g for 15 min. The water phase containing the viral proteins was collected and stored at −80°C. The hemagglutinating activity (HA units [HAU]) was determined by hemagglutination in microtiter plates using Atlantic salmon erythrocytes as previously described (13). Control antigen was prepared from confluent noninfected ASK cells.

Experimental infection and fish sampling.

Two experiments using intraperitoneal (i.p.) injection and one experiment using immersion (bath) challenge were included in this study. Experiments 1 and 2 were performed at the VESO Vikan research station (Namsos, Norway) using Atlantic salmon presmolts (40 g) from a hatchery at the station with no history of ISA. The fish were held in freshwater tanks at 12°C. For experiment 1, the fish were challenged by i.p. injection of 5 x 10³ 50% tissue culture infective doses (TCID₅₀) of ISAV. Samples of fish were taken 11 days postinfection (p.i.). Based on a parallel experiment, this is when virus load is expected to be at its maximum and 4 to 5 days before peak mortality. Peripheral blood was collected in heparin, followed by autopsy. Tissues from heart, liver, kidney, anterior kidney, gill, spleen, pseudobranch, eye, brain, esophagus, pylori, hind gut, skin, and muscle were either collected in 10% buffered formalin or snap-frozen in liquid nitrogen. For experiment 2, the fish were challenged as in experiment 1 and sampled at days 2, 4, 7, 10, 14, 18, and 22 p.i. Peripheral blood was collected in heparin, and tissues from heart, liver, kidney, gill, spleen, and gut were collected in 10% buffered formalin. At days 2, 10, and 14 p.i., leukocytes were isolated from blood, anterior kidney, and spleen, and cytospots were prepared (see below). Experiment 3 was performed at the Norwegian Veterinary Institute using Atlantic salmon presmolts (20 g) in freshwater tanks at 8°C. The low temperature slows down the development of infection. The fish were challenged by immersion (bath) for 3 h with 10⁴ TCID₅₀ per ml. Fish with no clinical signs of disease were sampled at 8 h, 24 h, 48 h, 8 days, and 20 days p.i., and tissues from heart, liver, kidney, gill, spleen, and gut were collected in 10% buffered formalin. Control tissues were sampled before the start in all experiments, and fish were anesthetized with methane tricaine sulfonate (MS222; Sigma; 0.1 mg/ml) before handling. Archival formalin-fixed paraffin-embedded tissues collected form moribund Atlantic salmon during verified ISA field outbreaks were from routine diagnostic samples at the Norwegian Veterinary Institute.

Cytocentrifuge preparations of leukocytes.

The leukocyte fractions were obtained by using a discontinuous Percoll gradient (GE Healthcare) as previously described (3). The cell concentration was adjusted to 10⁶ cells ml⁻¹ in phosphate-buffered saline (PBS). Cytospot preparations were made by centrifugation of 100 μl of cell suspension at 1,500 rpm for 5 min in a cytocentrifuge (Shandon Cytospin 2; Pittsburgh, PA) followed by air drying for 2 h. The preparations were stored at −80°C until assayed.

IHC.

ISAV immunohistochemistry (IHC) was performed as previously described (21). Briefly, sections of formalin-fixed paraffin-embedded tissue were prepared on poly-l-lysine-coated slides, dewaxed, and subjected to microwave oven treatment. Rabbit antibody to recombinant ISAV NP (1) was used for virus detection. The Vectastain ABC-AP kit (Vectastain anti-rabbit Ig ABC-AP kit; AK 5001; Vector Laboratories, Inc.) was used for detection of bound antibody employing Fast Red (1 mg ml⁻¹) and naphthol AS-MX phosphate (0.2 mg ml⁻¹) with 1 mM levamisole in 0.1 M Tris-buffered saline (TBS) (pH 8.2) as the substrate.

Virus histochemistry.

Detection of the ISAV receptor was performed on formalin-fixed paraffin-embedded tissues from Atlantic salmon with no known history of ISA. Sections were dewaxed in xylene and hydrated in graded alcohol, followed by treatment with peroxidase block (EnVision+ System-HPR K4007; Dako) for 5 min. Nonspecific blocking was performed by incubation with 5% bovine serum albumin (BSA) in Tris-buffered saline (TBS; pH 7.6) for 20 min. Sections were then incubated with ISAV HE antigen (100 HAU ml⁻¹) in TBS for 1 h, followed by detection of bound antigen with a monoclonal antibody (MAb) to ISAV HE (12) and a horseradish peroxidase (HRP)-conjugated anti-mouse Ig amplified detection system (EnVision+) with 3,3′-diaminobenzidine (DAB) substrate. To test for the binding specificity of the method, the following controls were included. (a) The sections were pretreated with 0.1 N NaOH for 30 min for saponification. This procedure is well established and results in de-O-acetylation of sialic acids on glycoproteins and glycolipids (41). (b) The ISAV HE antigen was incubated for 1 h before application to the sections, with horse or guinea pig serum, either of which is known to be rich in Neu4,5Ac₂ and thus blocks the putative binding site. Neither of these sera is known to contain significant amounts of 9-O-acetylated sialic acids (20, 22, 44). Such treatment has previously been demonstrated to block the ISAV receptor binding (20). The HE antigen was also treated with rat, pig, and bovine serum known to contain predominantly 9-O-acetylated sialic acids and not significant amounts of 4-O-acetylated sialic acids. (c) To further demonstrate the specificity of the inhibition by horse and guinea pig serum, control preparations of the sera were made by saponification of O-acetyl esters, i.e., the sera were incubated with 0.1 N NaOH for 30 min at room temperature (RT) followed by neutralization with 1 N HCl as previously described (20). (d) The ISAV HE antigen was replaced with antigen prepared from uninfected ASK cells. All tested samples were negative after applying this preparation.

The indirect immunofluorescent antibody test (IFAT) was performed as previously described (12). Briefly, cytospot preparations, cryosections, or blood smears were fixed in either cold acetone or 10% formalin for 10 min. A monoclonal antibody to ISAV HE (12), a rabbit antibody to recombinant ISAV NP (1), or both for the double staining experiments were used as primary antibodies. Alexa Fluor 488-conjugated anti-mouse IgG and/or Alexa Fluor 594-conjugated anti-rabbit IgG (Molecular Probes) was used for detection of bound antibody. Immunofluorescent virus histochemistry was performed by incubating sections with ISAV HE antigen (100 HAU ml⁻¹) for 1 h, followed by detection of bound antigen using a monoclonal antibody to ISAV HE and Alexa Fluor 488-conjugated anti-mouse IgG (Molecular Probes). Sections were mounted in SlowFade Gold (Molecular Probes) and were examined with a Zeiss LSM 710 confocal laser scanning microscope (Carl Zeiss, Germany) using a 63× oil immersion objective.

RESULTS

Endothelial cells are the main target cells of ISAV infection in Atlantic salmon.

To determine the cell tropism, tissues from experimentally infected Atlantic salmon (experiment 1) were examined at 11 days p.i. by immunohistochemistry (IHC). Based on a parallel experiment, this is when virus load is expected to be at its maximum and 4 to 5 days before peak mortality. IHC revealed strong positive ISAV staining confined to the endothelial layer of all types of vessels in all organs, including capillaries, veins, and arteries, and also in the endocardium, in endothelial scavenger cells in the kidney, in sinusoidal endothelium in the liver, and in pillar cells of the gill lamella (Fig. 1; see also Fig. S1 in the supplemental material). This shows that endothelial cells are the main target of ISAV infection. Dissemination from endothelial cells to adjacent parenchymal cells was not found, and no evidence of vasculitis or hemorrhagic necrosis was observed in the sections investigated. The location of stained ISAV in the endothelial cells was sometimes confined only to the nucleus but most frequently present in the cytoplasm. In contrast to the strong ISAV staining, only minor pathological changes were observed at autopsy and on histological examination. Findings included sparse ascites, small hemorrhages in skin and eye, a small increase in the number of melanomacrophages, blood congestion/stasis in anterior kidney, moderate hemophagocytosis in spleen, and eosinophilic concrements in kidney tubules. ISAV infections of Atlantic salmon have previously been linked to anemic conditions as shown by low hematocrit (43). However, the link between the anemic state and ISAV pathogenesis remains unresolved. To further determine if cells other than endothelial cells play a role in ISAV tropism, RBC preparations were examined at day 11 postinfection (p.i.) by IFAT. While IHC revealed a weak, diffuse staining of RBCs in tissue sections, IFAT labeling of blood smears with anti-ISAV HE and anti-ISAV NP antibodies revealed abundant virus antigen associated with the surface of most RBCs. No expression of viral proteins was detected in either the cytoplasm or the nucleus of RBCs (Fig. 2). It appears that ISAV produced by vessel endothelium attaches to the RBC surface but does not infect or replicate within the RBCs.

Fig 1.

Immunohistochemical detection of ISAV on tissue from experimentally infected Atlantic salmon. Infected cells (red) are present in capillaries and blood vessels. (a) Heart with ISAV in endocardial and endothelial cells. Co, compact layer; Sp, spongiosa layer. (b) Kidney with ISAV in vessel endothelium, glomerular capillaries, and peritubular endothelial cells. T, tubuli; H, hematopoietic tissue. (c) Liver with ISAV in endothelium of vessels and sinusoidal (Si) walls. CV, central vein. (d) Gills with ISAV in endothelial cells, including pillar cells (arrows). F, filament; L, lamella.

Fig 2.

IFAT detection of ISAV in RBC preparation from experimentally infected Atlantic salmon analyzed by confocal microscopy. Virus is attached to the cell surface. No expression of viral proteins in the cytoplasm or nucleus of RBCs indicates no replication of ISAV in RBCs. (a) Antibody to nucleoprotein (NP) demonstrates a widespread attachment of ISAV to RBCs. (b) Higher magnification of panel a. (c) Antibody to hemagglutinin esterase protein (HE). (d) Merged image shows partial colocalization of NP and HE.

ISAV infection is detected in leukocytes and gill epithelium.

To further assess the infection and tissue tropism of ISAV, samples were taken sequentially during the experimental infections. Tissue samples from experiments 2 and 3 and cytospin preparations from experiment 2 were examined by IHC. The cytospin preparations showed approximately 0.1% of the leukocytes positive for ISAV antigen at day 2 p.i., with no positive staining observed in endothelial cells in tissue sections at this time point. Lymphocyte depletion was observed in experiment 2 by differential counting (data not shown), which is in accordance with a recent report demonstrating tissue depletion of CD8-positive lymphocytes (21). Interestingly, lymphocyte depletion has been observed with influenza virus infection in domestic poultry (38) and in mice (15) and infection of leukocytes has also been observed in mice.

ISAV antigen was observed in endothelial cells in tissue sections at day 4 p.i. in experiment 2 and at day 8 p.i. in experiment 3. There was also indication of epithelial infection at 8 days p.i. in experiment 3; however, only a very few positive gill epithelial cells were found in each fish sampled (0 to 3 cells in each fish). No infection was observed in other cells in either experiment or in earlier samplings. This strengthens the suggestion that endothelial cells are the main target cells during ISAV infection in both immersion (bath) and i.p. challenges. However, our results indicate that there may be a primary replication stage in gill epithelial cells and in leukocytes.

Necrotic cells in liver and kidney lesions of severely anemic, moribund fish appear noninfected by IHC.

Since results from experimental infections do not always correlate with natural infections, we next compared results from our experimental infections with typical ISAV field outbreaks. Tissue samples were collected from ISAV-infected, moribund fish with severe anemia, low hematocrit (below 10), circulatory disturbances, ascites, and small hemorrhages in skin and eye. The samples were examined by histopathological evaluation and ISAV IHC staining. All organs investigated contained large amounts of ISAV confined to endothelial cells of the vasculature, as described above in experimentally infected fish. However, in the hemorrhagic, coagulative necrosis in liver and kidney, which are lesions considered archetypical of ISA (43), no ISAV was detected by IHC, except in endothelial cells in adjacent intact tissue. The liver necrosis in ISAV-infected Atlantic salmon is usually distributed in a multifocal to zonal pattern (43). In situ hemadsorption was observed in some cases, as was in situ RBC agglutination (Fig. 3; see also Fig. S2 in the supplemental material). However, we were not able to detect any thrombus formation in infected vessels by martius, scarlet, blue (MSB) staining. Increased hemophagocytosis was observed in spleen and anterior kidney (bone marrow analogue in fish) and in several cases also in other organs where hemophagocytosis is not normally seen. This finding may be explained by the attachment of virus to RBCs as observed by IFAT.

Fig 3.

Immunohistochemical detection of ISAV on tissue from naturally infected Atlantic salmon. (a and b) Infected endothelial cells (red) were detected adjacent to hemorrhagic liver necrosis (HCN) (CV, central vein) (a) and tubular necrosis (TN) in kidney (b). (c) HE-stained spleen with hemophagocytosis (arrow). (d) Heart with hemadsorption (arrows) on IHC-labeled infected endothelial cells.

ISAV pattern of virus attachment largely explains endotheliotropism in Atlantic salmon.

We have observed that endothelial cells and RBCs are the main virus target cells during ISAV infection (Fig. 1; see also Fig. S1 in the supplemental material). Our next question was whether the presence of the ISAV receptor correlates with the cell tropism seen in Atlantic salmon. Applying a novel virus histochemistry method, we determined cell-specific expression of ISAV receptors in tissue sections from non-ISAV-infected Atlantic salmon.

The ISAV receptor was mainly found on endothelial cells and on RBCs (Fig. 4; see also Fig. S3 in the supplemental material). The ISAV receptor was demonstrated in capillaries, veins, and arteries in all organs, including specialized endothelial cells such as pillar cells in the gills, glomerular capillary and scavenger endothelial cells in kidney, and sinusoidal endothelial cells in liver. Importantly, the ISAV receptor was also detected on epithelial cells in gill and in the newly described gill interbranchial lymphoid tissue (26), on luminal epithelial cells in hind gut, and on basal keratinocytes in epidermis. Next, we performed virus histochemistry on ISAV-infected Atlantic salmon. We did not find any changes in the distribution, suggesting that infected endothelium was intact. Nor did we find any indications that the receptor was more or less abundant in tissues from ISAV-infected fish, indicating that the receptors are neither up- nor downregulated as a result of infection.

Fig 4.

Virus histochemical detection of the 4-O-acetylated ISAV sialic acid receptor in tissue from healthy Atlantic salmon using ISAV preparations as a probe. Labeled cells (brown) were present in all organs. (a) Heart with receptor on endocardial cells, RBCs, and endothelial cells in capillaries, arteries, and veins. Co, compactum; Sp, spongiosum. (b) Kidney with receptor on glomerular (G) and peritubular endothelial cells. T, tubuli; H, hematopoietic tissue. (c) Liver with receptor on endothelium of vessels and sinusoidal (Si) walls. CV, central vein. (d) Gills with receptor on epithelial cells and on endothelial cells, including pillar cells (P). L, lamella.

The preferred ISAV cellular receptor is 4-O-acetylated sialic acids.

We have previously presented results suggesting that Neu4,5Ac₂ is the preferred receptor for ISAV (20). By performing a series of control experiments, we here further substantiated this conclusion. First, we pretreated the sections with mild alkali treatment, which is a well-established method for de-O-acetylation of sialic acids (41). This completely abolished the virus histochemistry reaction, thus demonstrating that the virus preparation binds to O-acetylated sialic acids. Second, we incubated the ISAV HE antigen with serum from either horse or guinea pig, either of which predominantly contains 4-O-acetylated sialic acids, or serum from rat, swine, or cattle, which predominantly contains 9-O-acetylated sialic acids, before applying them in virus histochemistry. Native horse and guinea pig serum blocked the reaction while no inhibition was observed with native rat, pig, or cattle serum, suggesting specificity for 4-O-acetylated sialic acids. Third, preparations of horse and guinea pig serum that were alkali treated and saponified before incubation with ISAV HE antigen did not inhibit the reaction. The combined data of this work and those presented by Hellebø et al. (20), who also demonstrated that both hemagglutination with horse erythrocytes and virus binding to horse serum were significantly inhibited following treatment with recombinant rat coronavirus 4-O-acetylesterase, strongly suggest that ISAV HE binds to 4-O-acetylated sialic acids.

ISAV buds from the luminal surface of endothelial cells.

We have demonstrated that ISAV mainly targets endothelial cells, but the mechanism by which the virus is released from host cells is unknown. To gain a better understanding of ISAV tropism, we performed IFAT of ISAV-infected tissues with an anti-ISAV HE monoclonal antibody (MAb). We found that the HE protein was targeted to the luminal surface of endothelial cells in vessels of infected Atlantic salmon. Moreover, virus histochemistry revealed an abundance of ISAV receptor on the luminal surface of endothelial cells (Fig. 5). These results support luminal budding of ISAV. Furthermore, the fact that we have not observed any viral dissemination from endothelial cells to the surrounding parenchymal tissue strengthens this conclusion.

Fig 5.

IFAT staining of heart tissues analyzed by confocal microscopy. (a) Staining with anti-ISAV HE MAb demonstrating a greater presence of HE at the luminal side (arrows) of the endothelial cells. (b) Virus histochemical staining of uninfected heart shows greater presence of ISAV receptor at the luminal side (arrow) of endothelial cells. L, luminal; N, nucleus.

DISCUSSION

In recent years, several studies have examined patterns of virus attachment by virus histochemistry to investigate influenza virus cell tropism and pathogenesis (23, 30, 50, 51). Here we applied a similar approach combined with IHC to explore the tropism and pathogenesis of ISAV infection in Atlantic salmon. Our results demonstrate that endothelial cells and RBCs are the major target cells of ISAV and that this tropism was largely determined by the pattern of virus attachment. To our knowledge, this is also the first report demonstrating data that are consistent with the in situ tissue distribution of 4-O-acetylated sialic acids.

ISA is characterized by severe anemia and circulatory disturbances (43). In accordance with this, our study revealed abundant viral replication in vascular endothelial cells only, seemingly as a noncytolytic event, without inciting vasculitis. Further, we found indications of apical release of virus into the bloodstream, accumulation of virus on the surface of RBCs, and extensive hemophagocytosis. This may contribute to the severe, progressive anemia of ISA. The endotheliotropism is in keeping with previous observations (12, 18, 19, 27), including demonstration of apical virus budding from endothelial cells by electron microscopy (27). In our study, we show that this pattern of infection closely mirrors the distribution of the virus receptor on all types of endothelial cells as well as on RBCs. For ISAV as for other orthomyxoviruses, the interaction of the hemagglutinin with its receptor on the host cell surface is necessary for infection. Thus, the receptor distribution is an important determinant for the pathogenesis and tissue tropism during the systemic phase of the infection. With regard to the early phase of infection established by immersion trial, we could find only a very few infected gill epithelial cells. Gill epithelial infection could be sufficient to establish the systemic infection that appeared similar regardless of immersion or i.p. challenge. However, the level of gill epithelial infection was low, and this opens the possibility that other infection routes are important and/or that we have not sampled intensively enough to include a possibly rapid and transient epithelial infection. Our recent studies of ISAV interaction with primary gill cell cultures (S. C. Weli, M. Aamelfot, O. B. Dale, E. O. Koppang, and K. Falk, submitted for publication) revealed the potential of ISAV replication in gill epithelium and point to the gill as a possible site of entry and primary replication as previously suggested (33).

During the systemic infection, focal necrosis in the liver or kidney is commonly found; however, we were not able to show either the virus receptor or virus infection of hepatocytes or interstitial kidney cells. It appears that cells lacking a measurable number of receptors, such as parenchymal cells, cannot be infected by ISAV. The necrosis could be the result of local circulatory failure compounded by a general hypoxia due to the severe anemia, rather than a consequence of direct virus-cell interaction.

The most important cells supporting viral replication were endothelial cells forming the inner surface of blood vessels. Such cells play a key role in regulation of blood pressure, homeostasis, and antithrombogenicity (46). Early endothelial infection and viremia are key events during systemic influenza virus infection in several birds (25, 49) and have also been demonstrated in cats (40, 42). Stressed endothelium alters its permeability function and can express molecules which may promote recruitment of inflammatory cells. However, in this study, even at terminal stages of disease, the endothelium seemed to be morphologically intact with limited cellular inflammatory reactions, nor was apoptosis induced. Similar findings have been reported for other endotheliotropic viruses such as the filoviruses (i.e., Ebola and Marburg viruses), where infection of endothelial cells does not disrupt the architecture of vascular endothelium (17). Virus infection of leukocytes in cytospin preparations suggests that these cells also may play some role in replication and dissemination of the virus.

Colocalization of ISAV receptor and ISAV on RBCs was demonstrated with significant amounts of virus seen attached to the RBC surface. The RBCs are thus major viral targets. However, contrary to earlier reports suggesting virus replication within RBCs, we were not able to demonstrate expression of virus proteins by IFAT in either RBC cytoplasm or nucleus. The anemia appears linked to hemophagocytosis, as this can be directly observed, and clear plasma and no jaundice indicate that hemolysis is not prominent. Possibly, the presence of virus on the RBC surface is recognized as foreign by innate immunity mechanisms. Virus binding to the RBC can also trigger changes in membrane conformation (39), which may lead to premature RBC destruction and anemia. Interestingly, hemophagocytic syndrome is frequently observed during systemic, fatal human influenza A virus infections (2, 28).

From the perspective of virology, fish constitute an important biological transition, including large-scale species diversification, the origin of the adaptive immune system, and the role as predecessors of the terrestrial vertebrates. Several virus families prevalent in mammalian hosts seem to have coevolved with bony fishes, including paramyxoviruses and orthomyxoviruses, which were essentially absent from all organisms that would represent predecessors of bony fish (53). In general, fish viruses have low amino acid identities with their mammalian counterparts, and this is also the case with ISAV. However, it appears that some key functional activities characteristic of orthomyxoviruses have been retained through evolution, including binding to sialic acid cellular receptors; RDE activity; fusion activity; the three polymerases constituting, together with the nucleoprotein, the RNP complex; and an NS1 analogue downregulating type I interferon (IFN) promoter activity (16, 31). Also, there are similarities between ISA pathogenesis in Atlantic salmon and the systemic phase of influenza virus infection in domestic poultry (14, 52) and cats (40). All are endotheliotropic, generalized infections with polarized virus budding, resulting in hemorrhages and circulatory disturbances. In addition, differential deletions of the ISAV hemagglutinin-esterase stem region are found in highly pathogenic ISAV as opposed to the low-pathogenic HPR0 type (5, 29, 32). Similar deletions in the functionally equivalent neuraminidase stem are also found in avian influenza viruses causing systemic infection, during adaptation from wild aquatic birds to domestic poultry (54), and seemingly favor replication and enhanced pathogenesis in chickens. Another common feature is the apparent imbalance between receptor binding and release, demonstrated by the lack of elution of the hemagglutination reaction with chicken and salmon erythrocytes, respectively (13, 54).

In conclusion, we have demonstrated the receptor specificity and cell tropism of the highly pathogenic ISAV in Atlantic salmon and correlated this with the pathology.