A Viral Mechanism for Inhibition of the Cellular Phosphatase Calcineurin

See allHide authors and affiliations

Science  24 Jul 1998:
Vol. 281, Issue 5376, pp. 562-565
DOI: 10.1126/science.281.5376.562


The transcription factor NFAT (nuclear factor of activated T cells) controls the expression of many immunomodulatory proteins. African swine fever virus inhibits proinflammatory cytokine expression in infected macrophages, and a viral protein A238L was found to display the activity of the immunosuppressive drug cyclosporin A by inhibiting NFAT-regulated gene transcription in vivo. This it does by binding the catalytic subunit of calcineurin and inhibiting calcineurin phosphatase activity.

Viruses encode many proteins that interfere with host defense systems (1). Nucleotide sequence analysis of the African swine fever virus (ASFV) genome (2) identified genes encoding proteins that are potentially able to interfere with the host response to viral infection. These include A238L, which has sequence similarity with IκB, and prevents activation of nuclear factor kappa B (NF-κB)–dependent gene transcription (3). The similarity between A238L and IκB is limited to the central region of the protein, which contains three ankyrin-like repeats. The NH2- and COOH-terminal regions of A238L are unlike those of the cellular IκB proteins (4), indicating that A238L may function by means of a mechanism different from that of IκB.

To identify host proteins with which A238L interacts, we used the yeast two-hybrid system (5, 6) to screen a cDNA library from pig alveolar macrophages. Nine clones specifically interacted with A238L, including four containing cDNA encoding the entire porcine cyclophilin A (CypA) gene. Another four clones contained cDNAs encoding all but the first 30 to 40 NH2-terminal amino acid residues of the catalytic (A) subunit of the Ca2+-calmodulin–regulated cellular phosphatase calcineurin (CaN). The A238L homolog from the virulent Malawi (LIL20/1) ASFV isolate also interacted with CaN and CypA. CaN and CypA interacted with A238L specifically, with no binding to the unrelated Gal4 DNA binding domain fusions SNF1 (5) and CDK2 (7). The CaN-A238L interaction was also detected when the genes were fused to the alternative domains of Gal4. The immunosuppressive drug cyclosporin A (CsA) binds to CypA, and this complex binds to and inhibits the activity of CaN (8). We postulate that A238L might function as a protein analog of CsA and inhibit the activity of CaN, either alone or as an A238L-CypA complex.

To confirm the interaction between A238L and CaN, we tested for in vitro binding of these proteins (9) (Fig. 1). In vitro–translated A238L, IκB, or an irrelevant ASFV protein (l14L) was immunoprecipitated with antibodies to epitope tags [SV5 or hemagglutinin (HA)] fused to these genes. Purified CaN added before immunoprecipitation was coprecipitated with A238L but not with the other proteins; it was detected by an antiserum that recognizes both the CaN A and B subunits (Fig. 1B) or by antiserum to the B subunit [anti-CaN(B)] (Fig. 1D). With the anti-CaN(B) we detected binding of endogenous CaN to SV5-A238L (Fig. 1D). CaN coprecipitated with A238L in the absence of added CypA, but we cannot exclude a requirement for CypA in the reaction, given that small amounts of CypA are present in the in vitro translation mixes (10). The A238L-CaN interaction was unaffected by a high concentration of CsA (10 μM) (Fig. 1B), implying that A238L either interacts with CaN at a different site from the CsA-CypA complex or can displace CsA–CypA complexes.

Figure 1

Direct binding of A238L to CaN in vitro (9). In vitro–translated HA-A238L (lanes 2 through 5) or HA-I14L (lane 6) was mixed with CaN (2 μg; lanes 2, 3, 4, 6), CypA (2 μg; lanes 2, 3, 6), and CsA (10 μM; lane 2). Mixtures were incubated with 2 μg of HA-specific monoclonal antibody, and the immune complexes were collected on protein A–Sepharose beads before separation by SDS–polyacrylamide gel electrophoresis (PAGE). Purified CaN (100 ng, lane 1) was run in parallel. Immunoprecipitated proteins were detected by autoradiography (A) or by protein immunoblot analysis with polyclonal anti-CaN (B). In vitro–translated SV5-A238L (lanes 2 through 5) and SV5-IκB (lane 6) were mixed with CaN (2 μg; lanes 2, 3, 4, 6), CypA (2 μg; lanes 2, 3, 6), and CsA (10 μM; lane 2). Purified CaN (50 ng, lane 1) was run in parallel. Mixtures were incubated with 5 μg of SV5-specific antibody, and the immune complexes were collected and analyzed by autoradiography (C) or by protein immunoblot analysis with monoclonal antibody to CaN(B) (D).

We tested for the interaction between A238L and CaN under physiological conditions by analyzing proteins from extracts of ASFV-infected cells that coimmunoprecipitated with A238L (11). A recombinant ASFV was constructed (SV5-A238L) in which the A238L open reading frame (ORF) was tagged at the NH2-terminus with the SV5 epitope (12). Synthesis of the SV5-A238L protein was detected 4 hours after infection (Fig. 2A), as previously reported (13). CaN was coprecipitated with SV5-tagged A238L in SV5-A238L ASFV–infected cell extracts but not in the controls (Fig. 2B). Radiolabeled, SV5-tagged A238L or IκB was transiently expressed in BSC1 cells (Fig. 2A); CaN coprecipitated with SV5-A238L but not with SV5-IκB (Fig. 2B), showing that the A238L-CaN interaction is specific.

Figure 2

Coprecipitation of CaN and A238L from cells (11). Vero cells were uninfected (lane 1), infected with wild-type BA71V ASFV for 4 (lane 2) or 10 hours (lane 3), or infected with SV5-A238L ASFV for 4 (lane 4) or 10 hours (lane 5). BSC1 cells were infected with MVA-T7 and transfected with vector alone (lane 6), pT7–SV5–A238L (lane 7), or pT7–SV5-IκB (lane 8). Purified CaN (50 ng) was run in parallel (lane 9). Radiolabeled cell extracts were immunoprecipitated with monoclonal anti-SV5, immune complexes were separated by SDS-PAGE, and immunoprecipitated proteins were detected by autoradiography (A) or by protein immunoblot analysis with monoclonal antibody to CaN(B) (B).

To study the effect of A238L on CaN phosphatase activity, we constructed a recombinant ASFV (ΔA238L) in which the A238L-coding region was deleted (Fig. 3A) (12). Removal of the A238L gene was confirmed by Southern (DNA) blot analysis (Fig. 3B) (12). The growth characteristics of ΔA238L and wild-type ASFV BA71V in Vero cells were the same (10).

Figure 3

Construction of recombinant viruses and CaN, PP1, and PP2A phosphatase assays of cell extracts (12, 14). (A) Genome map of wild-type (wt) and ΔA238L ASFV DNA, showing location of Sal I sites. (B) Southern blot analysis of wt and ΔA238L ASFV genomic DNA digested with Sal I and probed with [32P]deoxadenosine triphosphate–labeled pIKGAL or the A238L gene fragment. CaN phosphatase assays of cell extracts prepared from porcine alveolar macrophages (C) or Vero cells (D). Cells were either uninfected or infected with wt or ΔA238L ASFV and were treated with CsA where indicated. (E) PP1 and PP2A assays of Vero cell extracts treated with CsA. Cells were either uninfected or infected with wt or ΔA238L ASFV and were treated with okadaic acid (OA) where indicated. (F) CaN phosphatase assays of SF21 insect cell extracts infected with wild-type ACNPV baculovirus or A238L–Bac, with or without CsA. Results show mean ± SEM of 32P released from a labeled peptide substrate. (G) Protein immunoblot analysis of wt or A238L–Bac–infected SF21 insect cell extracts determined with anti-A238L.

CaN phosphatase activity was assayed in extracts from primary porcine alveolar macrophages (Fig. 3C) and Vero cells (Fig. 3D) that were either uninfected or infected with wild-type or ΔA238L ASFV (14). The assays were specific for CaN; CsA reduced the phosphatase activity (Fig. 3, C and D). Macrophages and Vero cells infected with wild-type ASFV contained about half as much CaN activity as the uninfected cells. In macrophages, ASFV infection reduced CaN activity to the background level observed in the presence of CsA (Fig. 3C). In ΔA238L-infected cells, CaN activity was two to three times that of wild-type–infected cells, demonstrating that expression of A238L inhibits CaN activity. CaN activity was higher in ΔA238L-infected cells than in uninfected cells, which suggests that, although ASFV infection may increase CaN activity, A238L counteracts this effect. The amount of CaN in both cell extracts was the same (10). The activity of the other major Ser-Thr protein phosphatases (PP1 and PP2A) was similar in uninfected cells and in cells infected with wild-type ASFV or ΔA238L, indicating that A238L inhibits CaN specifically (Fig. 3E). CaN activity was also reduced about one-half (Fig. 3F) by A238L in insect cells that had been infected with a recombinant baculovirus expressing A238L (A238L-Bac) (12, 14) (Fig. 3G).

CaN is a ubiquitously expressed phosphatase with diverse functions. One substrate is the NFAT (nuclear factor of activated T cells) family of transcription factors, to which CaN directly binds (15). Inhibition of CaN activity by the CsA–CypA complex (or by the FK506–FKBP12 complex) prevents NFAT activation by inhibiting CaN-dependent dephosphorylation of the cytoplasmically located subunit of NFAT (16). This prevents its nuclear translocation as well as the transcription of NFAT-dependent genes, which include immunomodulatory cytokines.

CaN inhibition by A238L in ASFV-infected macrophages might prevent activation of an NFAT factor, thus preventing transcription of the immunomodulatory genes that depend on NFAT. Using a porcine macrophage cDNA library, we identified by polymerase chain reaction (PCR) and nucleotide sequence analysis an amplified fragment similar to NFATc (GenBank U08015), which displayed 86.1% nucleotide identity and 94.2% amino acid identity (Fig. 4A) (17). We also identified (by reverse transcriptase-PCR and nucleotide sequence analysis) an amplified fragment from porcine RS-2 cells similar to NFAT1 (Genbank U43341), displaying 91.1% nucleotide identity and 100% amino acid identity (Fig. 4A) (17).

Figure 4

Sequence comparison of porcine NFAT-like sequences and the effect of A238L on the expression of an NFAT-dependent luciferase reporter gene (17, 18). (A) Amino acid sequence comparison between human NFATc (amino acids 437 to 556; U08015), a porcine macrophage NFAT gene fragment (NFAT Mac; AF069996), and a porcine RS-2 cell NFAT gene fragment (NFAT RS-2; AF069995); only residues that differ from the NFATc sequence are shown. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr. (B) An NFAT-luc reporter plasmid was cotransfected with pCDNA3 or pCDNA3–A238L into RS-2 cells treated with the indicated compounds. By 24 hours after transfection, the cells were harvested and luciferase activity was assayed. (C) An AP-1 luc reporter plasmid was cotransfected with pCDNA3 or pCDNA3–A238L into RS-2 cells. All cells were treated with ionomycin and phorbol 12-myristate 13-acetate (PMA), and some were treated with CsA (as indicated). Cells were harvested and assayed for luciferase activity. Data represent mean ± SEM.

Cotransfection of a vector expressing A238L with an NFAT-dependent reporter gene cassette in RS-2 cells consistently (n = 5) reduced reporter gene expression to 50 to 60% of that of the vector only (pCDNA3) controls (Fig. 4B) (18). The A238L homolog from the LIL20/1 ASFV isolate also reduced NFAT-dependent reporter gene expression, whereas the vector expressing an irrelevant ASFV gene (l14L) had no effect (10). Reporter gene expression was undetected using a construct in which the NFAT-binding sites had been mutated (10, 18). Reduction of NFAT-driven gene transcription by A238L was specific; reporter gene expression from a construct containing AP-1–binding sites was unaffected by either expression of A238L or treatment with CsA (Fig. 4C) (18).

A238L also inhibits NF-κB–dependent gene transcription (3). However, inhibition of CaN activity cannot explain the reduction of NF-κB activation by A238L in cells treated with PMA alone (3). PMA-stimulated activation of NF-κB is not inhibited by CsA, which shows that this pathway is not CaN-dependent (10, 19). Therefore, A238L seems to have two functions: first, to bind to CaN and inhibit its phosphatase activity and thus CaN-dependent pathways; second, to inhibit NF-κB–dependent transcription by an unknown mechanism.

A238L may provide a versatile mechanism that enables ASFV to evade host defense systems by preventing transcription of immunomodulatory proteins, which is dependent on NFAT or NF-κB. Virus genes are thought to be captured from the host and to mimic the function of host genes; the implication is that cellular homologs of A238L exist.


View Abstract

Navigate This Article