Identification of residues within the African swine fever virus DP71L protein required for dephosphorylation of translation initiation factor eIF2  and inhibiting activation of proapoptotic CHOP

The African swine fever virus DP71L protein recruits protein phosphatase 1 (PP1) to dephosphorylate the translation initiation factor 2  (eIF2  and avoid shut-off of global protein synthesis and downstream activation of the pro-apoptotic factor CHOP. Residues V16 and F18A were critical for binding of DP71L to PP1. Mutation of this PP1 binding motif or deletion of residues between 52 and 66 reduced the ability of DP71L to cause dephosphorylation of eIF2  and inhibit CHOP induction. The residues LSAVL, between 57 and 61, were also required. PP1 was co-precipitated with wild type DP71L and the mutant lacking residues 52- 66 or the LSAVL motif, but not with the PP1 binding motif mutant. The residues in the LSAVL motif play a critical role in DP71L function but do not interfere with binding to PP1. Instead we propose these residues are important for DP71L binding to eIF2  .

5 nuclear localisation of CHOP. Initially mutant genes were constructed that had the predicted PP1 binding site mutated (V 16 E, F 18 L, Figure 1 B). In addition deletions were made of C-terminal sequences. This region is similar in location, relative to the predicted PP1 binding site, to the putative eIF2binding domain of ICP34.5 (residues 233-248 ICP34.5, Figure 1 A). Following transfection of plasmids expressing these mutant DP71L proteins into tunicamycin-treated or untreated cells, expression of CHOP was tested by confocal microscopy (Fig. 2) or by Western blotting of cell extracts (Fig. 3). A summary of results showing the level of CHOP induced following transfection of plasmids expressing different DP71L mutants into cells is shown in Table 1. Expression of DP71L protein was detected to varying levels following transfection of all tested plasmids tested (data not shown and Figure 3). No expression was detected from plasmids that encoded DP71L proteins with the C-terminal 10 or 20 amino acids deleted (data not shown). To better define the regions within the residues 52-66 that are required to inhibit CHOP activation deletions were made of residues 52 to 61 (-20 to -10 from the C-terminus ) and 57 to 66 (-15 to -5 from the C-terminus) (see Fig. 1). A reduction in the efficiency of CHOP inhibition from 98% in cells expressing wild type DP71L to 23% and 18% in cells expressing DP71L Δ 52-61 and DP71L Δ 57-66 was observed respectively (Table 1). These mutants lack a common motif (LSAVL) between residues 57 to 61. A mutant, DP71L Δ LSAVL, lacking these residues also had a reduced ability to 6 inhibit the accumulation of CHOP within the nucleus when compared to wild type DP71L (39% compared to 98% CHOP inhibition) ( Table 1).
The two leucine residues within the LSAVL motif are most important for DP71L inhibition of CHOP nuclear localisation.
Each residue within the LSAVL motif was replaced with an alanine. Mutating residues S 58 and V 60 had no effect on the wild type function of DP71L as CHOP was detected in the nucleus of transfected cells at similarly high levels as in cells expressing wild type DP71L (Table 1). Mutation of the two leucine residues to alanine reduced the ability of DP71L to inhibit CHOP nuclear localisation (Table   1). Of these two leucine residues mutation of L 57 had the greater effect, reducing the percentage of CHOP inhibition to 31%, compared to 57% for mutant DP71L L 61 A, compared to 98% for wild type DP71L (Table 1). Together, these results suggest that the PP1 binding domain and the motif LSAVL are critical for function, and of the LSAVL motif the two leucine residues are the most important.
DP71L proteins with these domains mutated retain some ability to inhibit CHOP induction suggesting additional residues may be important for activity or that the substitutions did not completely inactivate the protein. Mutants with both the mutation V 16 E, F 18 L and deletion of LSAVL domain were generated (V 16 E, F 18 L LSAVL) and tested for their ability to inhibit CHOP activation. Mutant DP71L L 57, 61 A has both the leucine residues within the LSAVL motif mutated to alanine, whilst mutant DP71L V 16 E, F 18 L, L 57, 61 A lacks both leucine residues and has mutation V 16 E, F 18 L. Finally, mutant DP71L V 16 E, F 18 L, Δ LSAVL has the mutation V 16 E, F 18 L, and deletion of the LSAVL motif.
These mutants had reduced ability (20 to 31% compared to 98%) to inhibit CHOP nuclear localisation compared to the wild type DP71L (see Table 1). This suggests that although the V 16 ,F 18 residues and LSAVL sequence have important roles, there are additional residues required for the ability of DP71L to inhibit CHOP induction. The inhibition of nuclear localisation of CHOP by wild type DP71L in cells undergoing ER stress is predicted to be a downstream effect of the DP71L mediated recruitment of PP1 to dephosphorylate eIF2α. Therefore, we tested for levels of dephosphorylated eIF2α in cells transfected with plasmids expressing mutant DP71L proteins.
Plasmids expressing the selected DP71L mutants were transfected into cells which were treated with tunicamycin, or left untreated. Western blots of cell lysates were probed with antibodies that recognised total or phosphorylated eIF2α (38 kDa), CHOP (30 kDa), the HA epitope-tagged DP71L protein and the loading control γ tubulin (51 kDa) (see Figure 3). The levels of expression of wild type DP71L and the V 16 E,F 18 L mutant was consistently higher than that of other mutants. The mean relative ratio of phosphorylated eIF2α to total eIF2α and of CHOP to total eIF2α was determined relative to control Vero cells (see Fig 3 panel

B) from three independent experiments.
In all untreated cells CHOP was not detected but was induced by tunicamycin treatment. In control cells, as expected, tunicamycin treatment increased the level of eIF2α phosphorylation compared to untreated cells. In contrast in cells expressing wild type DP71L the band corresponding to phosphorylated eIF2α was not detected in either untreated or tunicamycin-treated cells ( Fig. 3 lanes 2 and 9) although the total level of eIF2α remained stable (compare Fig. 3 lanes 1 and 8). In tunicamycin-treated cells a band corresponding to the CHOP protein was detected, although reduced in amount relative to that observed in the untransfected control cells, with a mean relative ratio of 0.3 compared to 1. In lysates from cells expressing DP71L which had the V 16 E, F 18 L mutation a higher level of eIF2α phosphorylation was detected in both untreated and tunicamycin-treated cells with relative ratios of 1.3 and 3.4 respectively compared to 1 in control Vero cells (compare Fig 3 panel A lanes 3 and 10, and panel B). In these lysates CHOP was strongly induced following tunicamycin treatment (Fig 3 lane 10). In lysates from cells expressing DP71L 52-66 phosphorylated eIF2α was detected in both resting and tunicamycin-treated cells, indicating that this mutant no longer caused 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 8 dephosphorylation of eIF2α. Furthermore, the CHOP protein was expressed upon stimulation with tunicamycin (Fig 3 lane 11).
In cells expressing DP71L with mutations in V 16 E, F 18 L and the LSAVL motif the following results were obtained. In lysates of cells expressing DP71L with a deletion LSAVL, this deletion in combination with V 16 E, F 18 L, Δ LSAVL, and V 16 E, F 18 L, L 57, 61 A phosphorylated eIF2α was detected in untreated cells, and increased following tunicamycin treatment (compare Fig 3 lanes 5-7 and 12-14). The CHOP protein was also detected in tunicamycin-treated cells expressing DP71L mutants with the LSAVL sequence deleted (ΔLSAVL), with V 16 E, F 18 L, mutated in combination with the LSAVL sequence mutated (V 16 E, F 18 L, L 57, 61 A) and with the mutated in combination with the LSAVL sequence deleted (V 16 E, F 18 L, ΔLSAVL).
Interestingly, lysates from cells expressing DP71L 52-66 had the highest relative ratio between phosphorylated and total eIF2α in the untreated cell lysates (Fig 3 A lane 4, 1.8 to 1) and was also higher in the tunicamycin-treated cells (Fig 3 A lane 11, 3.6 to 2.3) compared to other mutants tested.
This mutant form of DP71L may still bind PP1 and thus sequester PP1 preventing its interaction with cellular factors which control the level of eIF2α phosphorylation. This could explain the observed increase in the level of phosphorylated eIF2α in cells expressing this DP71L mutant.
The data confirmed that mutation of the V 16 , F 18 residues or LSAVL motif in DP71L resulted in the loss of wild type DP71L ability to cause dephosphorylation of eIF2α and inhibit induction of CHOP following tunicamycin treatment.

DP71L mutants with mutations in residues V 16 , F 18 do not co-precipitate with PP1
Our previous results using the yeast three hybrid system indicated that DP71L binds to PP1 and this complex interacts with eIF2 (Zhang et al., 2010). A prediction from this is that mutation of the PP1 binding domain in DP71L would reduce the interaction between these proteins and the failure to recruit PP1 to eIF2 would explain the failure to inhibit eIF2phosphorylation. Levels of the wild type and mutant DP71L detected in total lysates detected by Western blotting varied (Fig 3). The wild type DP71L, V 16 E, F 18 L, and this mutation combined with a deletion of the LSAVL sequence (V 16 E,  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 10 enhanced translation of both cap-dependent and independent translation by over 450% of the pcDNA3 control for firefly luciferase cap dependent translation (panel A), and just under 300% for renilla luciferase cap independent translation (panel B). In contrast, all of the DP71L mutants tested abolished this enhancement, as the levels of reporter expression were similar to the plasmid control.
By performing a one way ANOVA in GraphPad Prism with multiple comparisons test against wild type DP71L, it was established that each of the mutants significantly reduced the translation enhancer effect of DP71L (P = <0.0001).
Plasmids expressing mutants DP71L Δ52-66 and DP71L Δ LSAVL consistently displayed a reduced induction of reporter activity compared to the pcDNA3 control plasmid, and this reached statistical significance for the DP71L mutant Δ LSAVL, (P value of <0.05 see Fig 6, blue asterisk). Possibly because these DP71L mutants still bind PP1 they may sequester PP1 from cellular factors such as CReP, reducing de-phosphorylation of eIF2α, and so decreasing translation below basal conditions.

Discussion
Shut-off of host protein synthesis is a major limitation to viral replication, and as such many viruses, including ASFV, have evolved mechanisms to evade or limit this response. The DP71L protein acts avoid the shut-off of global protein synthesis. This may be induced by either the double-stranded RNA activated protein kinase PKR or the protein kinase-like ER resident kinase (PERK), which is activated as part of the unfolded protein response (UPR). The UPR also feeds into the innate immune response through the activation of pro-inflammatory cytokines and NF-κB. This occurs via the IRE1/XBP1 and PERK pathways, as for example, IL-6 and IL-8 are targets of XBP1, whilst the shut off of protein synthesis by PERK leads to an imbalance in the ratio of IκB to NF-κB, leading to NF-κB activation (Deng et al., 2004;Gargalovic et al., 2006;Kaneko et al., 2003;Urano et al., 2000) . In previous studies ASFV infection of Vero cells was shown to activate the ATF6 branch (ATF6 activates ER chaperones and XBP1) of the UPR but not PERK or the inositol-requiring enzyme   1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60  61  62  63  64  65 11 (IRE1) (Galindo et al., 2012;Netherton et al., 2004). This evidence suggests that ASFV infection does not cause PERK activation and thus DP71L function may primarily function to counteract PKRactivated phosphorylation of eIF2.
In this study we identified critical residues which are required for function of DP71L to reduce phosphorylation of translation initiation factor eIF2and to inhibit the downstream effects including induction of the pro-apoptotic CHOP protein in response to stress induced by tunicamycin. We confirmed that the residues V 16 , F 18 within a predicted PP1 binding site (VRF) in DP71L were required for these functions. We showed that wild type DP71L co-precipitated with PP1 whereas DP71L with a mutation V 16 E, F 18 L did not. Thus we conclude that these residues are essential for PP1 binding and function of DP71L.
We expected to identify a domain in DP71L required for binding to eIF2 and investigated whether sequences downstream from the PP1 binding domain were required for DP71L function. Mutation of the sequence LSAVL between residues 57 and 61 reduced DP71L ability to cause dephosphorylation of eIF2and inhibit CHOP induction. Within this LSAVL sequence the two leucine residues were most critical. DP71L mutants of the LSAVL sequence retained the ability to co-precipitate PP1 suggesting that the LSAVL sequence has a critical functional role other than PP1 binding. We failed to detect co-precipitation of eIF2with DP71L and PP1, possibly due to a weak or transitory interaction (data not shown). Therefore we were unable to confirm that this sequence is involved in binding of DP71L to eIF2although consider this likely.
Studies with ICP34.5 and GADD34 proteins have also investigated domains critical for function. In one study (Li et al., 2011) the eIF2α binding domain of ICP34.5 was mapped to residues 233-248; and in GADD34 the sequence between residues 578-597, Rx[Gnl]x1-

Figure 6. Wild type, but not mutant, DP71L acts as a translation enhancer
Vero cells were co-transfected with equal amounts of the bi-cistronic reporter plasmid pIRES FF luc/Ren luc and pcDNA3, wild type or mutant DP71L as indicated. 24 hours post-transfection cells were lysed and reporter activity assessed using the Dual-Luciferase Reporter Assay kit (Promega).
The firefly (A) or renilla (B) reporter activity of control cells transfected with pcDNA3 was set at 100% and wild type or mutant activity expressed as a percentage relative to pcDNA3. Experiments were performed in triplicate three times. Error bars represent the standard deviation. Statistical analysis was carried out in GraphPad Prism using a one way ANOVA with multiple comparisons test.
Asterisks represent a significant difference in value between WT DP71L and the mutants tested (* = P value of <0.5, **** = P value of <0.0001)