key: cord-265224-mwgccr4a authors: Delmas, Bernard; Gelfi, Jacqueline; L'Haridon, René; Vogel; Sjöström, Hans; Norén; Laude, Hubert title: Aminopeptidase N is a major receptor for the enteropathogenic coronavirus TGEV date: 1992 journal: Nature DOI: 10.1038/357417a0 sha: doc_id: 265224 cord_uid: mwgccr4a CORONAVIRUSES, like many animal viruses, are characterized by a restricted host range and tissue tropism(1). Transmissible gastroenteritis virus (TGEV), a major pathogen causing a fatal diarrhoea in newborn pig, replicates selectively in the differentiated enterocytes covering the villi of the small intestine(2). To investigate the molecular determinants of the infection, we characterized the surface molecule used by the virus for binding and entry into host cells. Here we report that aminopeptidase N, an ectoenzyme abundantly expressed at the apical membrane of the enterocytes, serves as a receptor for TGEV. Monoclonal antibodies were selected for their ability to block infection by TGEV of porcine cell lines. They recognized a brush-border membrane protein of M(r), 150K, which was identified as aminopeptidase N by ammo-terminal sequencing. Two lines of evidence supported the view that the peptidase itself acts as a receptor. First, virions bound specifically to aminopep-tidase N that was purified to homogeneity. Second, recombinant expression of aminopeptidase N conferred infectivity by TGEV to an otherwise non-permissive cell line. fast termination of PDE activation observed under similar conditions 20 . More recent work indicates that transducin GTPase can be faster under more physiologial conditions 11,21-23, but the mechanism of GTPase acceleration has remained unclear. The data of Fig. 1 show that PDE itself serves as a GTPase-activating factor. The maximal GTPase rate observed in this reconstitution study (-0.15 S-I) is still about lO-fold slower than the rate of the recovery from a photoresponse. But a more rapid rate (> 0.6 s -I) is observed in suspensions of disrupted rod outer segments l2 for the fast component of GTPase suppressed by micromolar concentrations of cGMP. Our study allows us to conclude that this faster GTPase is a property of that transducin which activates PDE, and thus the extent of PDE-dependent GTPase acceleration is higher in rod outer segment suspensions than in reconstituted membranes. More recent data (V. Y.A. et al., manuscript in preparation) shows that further concentration of rod outer segment suspensions (> 100 fLM rhodopsin) increases GTPase rates by at least twofold, close to the turn-off time of the photoresponse. The data shown in Fig. 2 indicate a feedback mechanism in retinal rods based on cGMP-dependent regulation of the lifetime of activated PDE. Such a mechanism might function during rod background adaptation, when the duration and light sensitivity of the photoresponse is diminished 24 ,25. A reasonable model is that background light depletes intracellular cGMP levels, causing dissociation of cGMP from the non-catalytic binding sites on PDE. This would accelerate the GTPase activity that terminates each PDE activation event, leading to a faster and/or smaller photoresponse. Such a mechanism might work in parallel with the known calcium feedback regulation of adaptation 4 • The regulation of GTP-binding protein GTPase activity by an effector described here, although not previously described for a heterotrimeric G protein, has been documented extensively for several other classes of GTP-binding proteins (for example ref. 26) . It is observed for elongation and initiation factors whose GTPase activity is enhanced by ribosomes. The class of small GTP-binding proteins including the product of proto-oncogene ras interact with GTPase-activating proteins (GAPs) that may also be their effectors 27 . The intrinsic GTPase of the heterotrimeric signal-transducing G proteins is considerably more rapid than that of the small GTP-binding proteins (for example refs 5-7), but still in several systems such as photoreception 2 ,3, 0lfaction 28 and muscarinic receptor-induced potassium channel regulation 29 it has seemed to be too slow to explain the NATURE· VOL 357 . 4 JUNE 1992 LETTERS TO NATURE rapid on-off cycle of the relevant effectors. Because acceleration has now been associated with the effector enzyme in the photoreceptor, it is relevant to search for similar mechanisms in other systems using heterotrimeric G proteins. 0 Aminopeptidase N is a major receptor for the enteropathogenic coronavirus TGEV acts as a receptor. First, virions bound specifically to aminopeptidase N that was purified to homogeneity. Second, recombinant expression of aminopeptidase N conferred infectivity by TGEV to an otherwise non-permissive cell line. To obtain monoclonal antibodies against the TGEV receptor, hybridomas were prepared from a mouse immunized with ST cells, a swine testis cell line highly susceptible to TGEV. Several of the resulting antibodies exhibited, in three different porcine cell systems, a blocking activity comparable to that of a high-titre neutralizing anti-TGEV antibody (Fig. 1 ). By contrast, no significant protection by the antibodies was observed after virus challenge in a feline cell system, or towards irrelevant viruses (group A bovine rotavirus or vesicular stomatitis virus; data not shown). Therefore, the selected antibodies had the characteris- tics expected for antibodies recognizing a major TG EV receptor. The monoclonal antibodies all recognized a polypeptide of relative molecular mass IS0,000 (1S0K) in ST cell extracts, together with a faint band interpreted as the mannose-rich intracellular precursor (Fig. 2 , lane O. When using solubilized brush-border membranes from pig small intestine, three major species of lS0K, 9SK and SOK were coimmunoprecipitated (Fig. 2, lane 4) . The first 30 amino acids of the 9SK species were detemined through N-terminal sequencing: NHz-Ala-Lys-Gly-Phe-Tyr-I1e-Ser-Lys-Ala-Leu-Gly-I1e-Leu-Gly-I1e-Leu-Leu-Gly-Val-Ala-Ala-Val-Ala-Thr-I1e-I1e-Ala-Leu-Ser-Val-COOH . This sequence was identical to the N-terminal sequence (minus the first Met) of porcine aminopeptidase N, deduced from the exon I nucleotide sequence 3 Material immunoprecipitated with G43 antibody from the Triton X-100 solubilized cell Iysates was analysed by western blotting as described in Further evidence that the anti-TGEV-receptor antibodies recognized aminopeptidase N was obtained by showing that (1) an antibody raised against rabbit aminopeptidase N 4 reacted with the same three polypeptides in brush-border membrane preparations (Fig. 2, lane 2) : 95K and 50K, corresponding to the B (amino) and C (carboxy) subunits of the pig aminopeptidase, and 150K, uncleaved aminopeptidase 5 ; (2) the immunoprecipitated material hydrolysed leucine p-nitroanilide, a chromogenic substrate specific for aminopeptidase (ref. 6 ; data not shown). Two experiments were designed to demonstrate any direct association between aminopeptidase N and the virus. First, soluble aminopeptidase N was centrifuged after incubation in the presence of virions (Fig. 3a) . Aminopeptidase N-specific bands were recovered with pelleted TGEV virions only. Second, when the aminopeptidase was incubated in the presence of adsorbed virions (Fig. 3b) , it bound to TGEV and not the other enteric viruses. In both assays, earlier incubation with an antibody against aminopeptidase N reduced the binding considerably. Because the two components were purified to homogeneity, it was concluded that the interaction between the aminopeptidase and TGEV occurs in the absence of any other cellular protein. The gene encoding aminopeptidase N ( APN) was expressed in non-permissive cells to see whether this would confer them with the capacity to bind TGEV. A pig intestine complementary DNA library was screened by use of a homologous DNA probe derived from the 5' end of APN gene. A full-length cDNA copy was . cloned and contained an open reading frame of 2,889 nucleotides encoding a polypeptide 79% identical to human aminopeptidase (data not shown). MDCK cell clones stably transformed with the porcine APN cDNA expressed a polypeptide of 150K which reacted with antibodies against aminopeptidase N (Fig. 4a) . The aminopeptidase activitl of the transfected clones was about 40-fold higher compared with non-transfected clones. On viral challenge, all of the three independent clones tested seemed to be fairly susceptible to TGEV infection, as proved by extensive destruction of the infected monolayers and synthesis of the viral structural polypeptides (Fig. 4b, c) . Earlier incubation with an antibody specific for aminopeptidase N prevented the appearance of viral cytopathic effect. These results show that aminopeptidase N was the only porcine protein necessary to confer susceptibility on canine kidney cells naturally resistant to TGEV. Moreover, the protease function of the molecule did not seem to be involved because it was blocked by bestatin, an inhibitor of aminopeptidase, without preventing the infection (Fig. 4b) . So far, defined receptors include molecules that belong to the immunoglobulin superfamily, such as CD4 for HIV 7 , ICAM-l for rhinovirus B , poliovirus receptor 9 and a carcinoembryonic antigen for murine hepatitis coronavirus 15 , and also an aminoacid transporter for murine leukaemia retroviruses l1 • Our study provides strong evidence that porcine aminopeptidase N serves as a receptor for an enveloped RNA virus, TGEV. This emphasizes the diversity of the membrane-bound proteins that viruses subvert for gaining entry into cells. Aminopeptidase N is a well documented ectoenzyme that binds to the membrane through an N-terminal segment 5 , 12,13. Human aminopeptidase N is identical to CDB, a surface antigen of many myeloid cells 14 , It is a zinc-binding protease that catalyses the removal of N-terminal, preferentially neutral residues from peptides, It is expressed in many tissues at different levels 15 , the highest activity being found in the small intestinal mucosa, where the aminopeptidase represents about 8% of the protein content of the apical membrane of the differentiated enterocytes, and in the kidney brush border. It is also expressed to a lesser extent in liver, lung and colon, where the virus does replicate, but without causing the specific histopathological damage seen in the small intestine 16 • In the intestine, the distibution of the receptor and the site of multiplication of TGEV are thus strikingly correlated. This argues for a pivotal role of aminopeptidase N/CD13 in determining the tissue tropism of TGEV, Investigating the nature of the virus interaction with aminopeptidase N could provide a rationale for the design of an antiviral strategy against TGEV and related infections. To develop a monoclonal antibody against the HCV-229E receptor, we produced hybridomas against deoxycholatesolubilized membrane proteins of two HCV-229E-susceptible human cell lines (WI38 lung fibroblasts and HL60 myeloid leukaemia cells). A monoclonal antibody designated RBS protected WI38 and RD human cell lines from HCV-229E-induced cytopathic effects and protected WI38 cells from virus infection ( Fig. la-c) . RBS pretreatment reduced the number of HCV-229E-infected WI-38 cells at 10 h post-infection by 96%, compared with cells pretreated with control mouse ascites. By contrast, RBS did not inhibit replication of HCV-OC43 in WI38 or RD cells, indicating that the receptor specificities of HCV-OC43 and HCV-229E are different. Susceptibility to HCV-229E infection in mouse-human somatic cell hybrids depends on a gene located on human chromosome 15 (ref. 6) . A promising candidate for the HCV-II To whom correspondence should be addressed. • This exopeptidase removes amino-terminal residues to complete the digestion of short peptides in the gut and helps break down neurotransmitter peptides in the brain 2 ,3,5,8. hAPN is identical to CD13, a glycoprotein identified on granulocytes, monocytes and their bone marrow progenitors 9 ,1O. Porcine aminopeptidase N is a receptor for transmissible gastroenteritis virus, a porcine corona virus in the same serogroup as HCV-229E (ref. 11). Because aminopeptidase from humans, pigs and other mammals are structurally similar 9 , 12-14, we investigated whether HCV-229E and RBS would bind specifically to hAPN and whether expression of hAPN by murine cells would make them susceptible to infection with HCV-229E. Murine NIH3T3 cells transfected with hAPN cDNA in a retroviral vector 9 (hAPN-3T3) and untransfected NIH3T3 cells were chaUenged with HCV-229E and HCV-OC43 to determine their susceptibility to virus infection. Although the control NIH3T3 cells were resistant to HCV-229E infection (Fig. Id) , the hAPN-transfected mouse cells were susceptible to infection with this virus (Fig. Ie) . By contrast, hAPN-3T3 cells were no more susceptible than NIH3T3 cells to infection with HCV-OC43 (data not shown). Thus, expression of hAPN confers HCV-229E susceptibility, but not HCV-OC43 susceptibility, on murine cells. We analysed binding of RBS to membrane preparations from hAPN-3T3 or parental NIH3T3 fibroblasts. The antibody bound to membranes of hAPN-3T3 but not to those of NIH3T3 cells (Fig. 2a) , indicating that RBS recognized hAPN. Similarly, HCV-229E virions bound more strongly to hAPN-3T3 membranes than to NIH3T3 membranes (Fig. 2b) , and RBS competi- 1 gen Gray·Keller for participating in the preliminary stage of this work We thank R. Christon for microvillar preparations, J. C. Huet for N-terminal sequencing acknowledges the support of the Commission of the European Communities. LK.V., H.5. and D.N. are members of the Biomembrane Research Center