Determinants of duck Tembusu virus NS2A/2B polyprotein procession attenuated viral replication and proliferation in vitro

Duck Tembusu virus (DTMUV), a mosquito-borne Flavivirus, has caused serious economic losses for the Chinese poultry industry. The genome is translated into a polyprotein that is cleaved to mature protein by host and viral proteases in the host cell, and this proteolytic process is important for the viral life cycle. However, the cleavage mechanism of DTMUV polyprotein is still unclear. In this study, we identified that several amino acids (P1-R, P1′-G, P2-R, P3-T, and P4-V) were vital for NS2A/2B cleavage. Meanwhile, both NS2A and NS2B were essential in cis for polyprotein NS2A/2B intramolecular cleavage. Subsequently, a DTMUV replicon and an infectious clone showed that the P1 site is essential to viral replication, while a mutation in P1′ could boost viral RNA replication. Furthermore, a recombinant virus with P1 and P1′ site mutations named rDTMUV-NS2A/2B-P1P1′(AA) was rescued from transfected BHK21 cells. The maximum viral titers and viral genome copies of rDTMUV-NS2A/2B-P1P1′(AA) were much lower than those of rDTMUV-WT both in the intracellular and extracellular samples of transfected and infected BHK21 cells. Taken together, the NS2A/2B cleavage sites processed by the NS2B3 protease are vital for DTMUV proliferation and virulence.


Scientific Reports
| (2020) 10:12423 | https://doi.org/10.1038/s41598-020-68271-0 www.nature.com/scientificreports/ As mentioned before, the substrate of flavivirus NS2B3pro possesses a similar characteristic, where two sites (P1 and P2) before the junction are often occupied by basic amino acids and the site behind the junction (P1′) is often occupied by a short-chain amino acid 15 , such as Ala, Gly and Trp (Fig. 1B). In addition, previous reports have shown that the conformation of the substrate is very important for cleavage 16 . Not only the amino acids located on both sides of the junction but also the proteins spanning the cleavage sites are important determinants of processing. However, the cleavage of the DTMUV polyprotein by NS2B3 has not been described.
In our study, we aimed to clarify the proteinase cleavage sites of NS2A and NS2B, as well as determine amino acid requirements in trans for NS2B3 processing. Moreover, the role of the intramolecular proteinase cleavage sites between NS2A (25 kDa) and NS2B (14 kDa) in the proliferation of duck Tembusu virus was studied. We found that the amino acids near the cleavage sites showed differential effects on NS2A/2B cleavage. Moreover, both the NS2A and NS2B proteins were required in cis for NS2A/2B proteolytic processing, while NS2Aα intermolecular proteinase cleavage was NS2B independent. Subsequently, the effects of the cleavage sites on viral RNA replication and proliferation were studied in detail by using a DTMUV replicon and an infectious clone. Our study shed slight on the correlation between NS2A/2B polyprotein processing and virus virulence.

DTMUV NS2A/2B can be cleaved by NS2B3. A previous study demonstrated that DENV NS2A/2B
can be cleaved by DENV NS2B3 17 . However, there is no study has proven that DTMUV NS2A/2B can be cleaved by its NS2B3, so in our study, we demonstrated that firstly. After the plasmid pCAGGS-Myc-NS2A/2B-Flag was constructed successfully, we verified the expression of NS2A/2B in DEFs at different plasmid concentrations, and the protein was detected by Western blot, as shown in Fig. 1C. Then, pCAGGS-Myc-NS2A/2B-Flag and different volume of pCAGGS-NS2B3-His were cotransfected to DEFs, and the cells were harvested at 24 h post transfection and detected by western blot. DTMUV NS2A/2B could be cleaved even at a low dose of NS2B3, Both NS2A and NS2B are in cis needed for NS2A/2B cleavage, and DTMUV NS2B3pro can cleave NS2A. To determine whether NS2A or NS2B is needed in cis for its cleavage, we constructed plasmids expressing NS2A-20aa-GST, EGFP-20aa-NS2B, and EGFP-20aa-GST ( Fig. 2A). Each of the plasmids was cotransfected with NS2B3pro to verify its cleavage by NS2B3pro. The results showed that none of the recombinant proteins could be cleaved by NS2B3pro, excluding NS2A-20aa-GST (Fig. 2B). It has been shown that YFV NS2B3 can cleave NS2A at the sequence QK/T to form NS2Aα 23 . To determine whether NS2A is needed in cis for NS2A/2B cleavage, we performed the following study. An NS2A eukaryotic expression plasmid was constructed and named Myc-NS2A. Surprisingly, we found that NS2A could be cleaved by NS2B3 when using Myc-NS2A as a positive control. As shown in Fig. 2C, a protein band smaller than NS2A was detected, and was assumed to be NS2Aɑ. Moreover, a protein band larger than GST (GST-Flag as positive control) was detected and was assumed to be NS2Aβ (Fig. 2D). This further indicates that NS2A-20aa-GST can be cleaved by NS2B3 at the NS2Aα cleavage sites but not at the site between NS2A and NS2B. Overall, DTMUV NS2B3pro can cleave NS2A/2B into NS2Aα, NS2A, and NS2B, while both NS2A and NS2B were needed in cis for NS2A/2B cleavage.
The P1 and P1′ amino acid sites of NS2A/2B strongly affect DTMUV replication. To verifying whether the P1 and P1′ sites are important to the viral replication, DTMUV replicons with mutated P1 and/or P1′ sites were constructed (Fig. 5A). BHK21 cells were transfected with CQW1-Rep-Rluc-WT, CQW1-Rep-Rluc-GDD/AAA(the active sites of RNA-dependent RNA polymerase were mutated to A which cause lethal influence on replicon and used as negative control), CQW1-Rep-Rluc-NS2A/2B-P1(A), CQW1-Rep-Rluc-NS2A/2B-P1′(A), or CQW1-Rep-Rluc-NS2A/2B-P1P1′(AA) and the cells were harvested at 36 h, 48 h, and 60 h post transfection. Finally, the RLuc expression from the replicons was measured. As shown in Fig. 5B, when Arg at the P1 site was substituted by Ala, there plication of the DTMUV mutated replicon was largely blocked. When both the P1 and P1′ sites were substituted by Ala, there plication of the DTMUV mutated replicon as represented by luciferase activity showed a high Rluc level at early times post transfection, but it did not continue to increase. However, it is interesting to mention that the replication of DTMUV-P1′(A) replicon was even better than that of the DTMUV-WT replicon.
The P1 and P1′ amino acid sites of NS2A/2B are important for DTMUV proliferation and virulence. To verifying whether the P1 and P1′ sites are important to the viral proliferation and virulence, infection clone was used in experiment. A schematic of the experimental setup and data collection is shown in Fig. 6A. Based on our replicon data, we found that the replication of CQW1-Rep-Rluc-NS2A/2B-P1P1′(AA) was not inhibited by the mutations and was even better than that of CQW1-Rep-Rluc-WT 36 h post transfection (Fig. 5B). Therefore, we wanted to determine the effects of the mutations on the virulence of a recombinant virus. P1 and P1′ site mutations were introduced into an infectious recombinant DTMUV, which was rescued from BHK21 cells and was named rDTMUV-NS2AB-P1P1′(AA). Interestingly, the CPE and virus infection of transfected BHK21 were delayed as observed by indirect immunofluorescence using a mouse anti-DMTUV polyclonal antibody as the primary antibody ( www.nature.com/scientificreports/ rDTMUV-NS2A/2B-P1P1′(AA) was smaller than that of rDTMUV-WT. Then, growth curves and genome copy numbers of both intracellular and extracellular rDTMUV-WT and rDTMUV-NS2AB-P1P1′(AA) were measured, as shown in Fig. 6E and F. We found that the copy number of rDTMUV-NS2A/2B-P1P1′(AA) was only slightly lower than that of rDTMUV-WT; however, the viral titer of rDTMUV-NS2A/2B-P1P1′(AA) was much lower than that of rDTMUV-WT. To better understand this discrepancy, we performed one more round of virus infection, and infected BHK21 cells with rDTMUV-NS2A/2B-P1P1′(AA). The same trends in growth curve and genome copy number were observed (Fig. 7A, B). www.nature.com/scientificreports/ TCID 50 , all rDTMUV-WT-infected duck embryos died within 5 days, while no rDTMUV-NS2A/2B-P1P1′(AA)infected duck embryos died within 5 days (Fig. 7C). However, 80% rDTMUV-NS2A/2B-P1P1′(AA)-infected duck embryos died within 10 days (Fig. 7C). In an experiment with a lower virus challenge (10 TCID 50 ), 40% rDTMUV-WT infected duck embryos died within 8 days, while 20% rDTMUV-NS2A/2B-P1P1′(AA)-infected duck embryos died within 8 days (Fig. 7C), there was significantly difference of two group by statistical analysis.

Discussion
NS2A and NS2B participate in the construction of the replication complex and play an essential role in virus replication. NS2A is a membrane protein with 8 transmembrane domains located on the ER that can assist in the reconstruction of the ER 18 . NS2B is also a transmembrane protein, and the junction between NS2B and NS3 is self-cleaved by the viral protease, whereby NS2B binding to NS3 causes conformational changes in NS3, which converts NS3 into the more active and stable NS2B3 proteinase 19,20 . It has been reported that flavivirus NS2B3pro cleaves the viral NS2A/2B, but the key element of the proteolytic process between NS2A and NS2B has barely been discussed. Duck Tembusu virus is a newly emerging flavivirus. In this study, we mutated the NS2A/2B P1 and P1′ sites to alanine and the proteolytic process was decrease by this substitute, the products of the proteolysis of NS2A/2B(P1P1′-AA) seems less than other groups, which suggested that the double mutation can weak proteolytic process, but it isn't be abolished. But there are two sites can be processed by NS2B3pro, NS2Aα could been so formed. Accordingly, cleavage of NS2A/2B results in NS2Aα, NS2Aβ, NS2A(due to inner incomplete cleavage) and NS2B protein.  www.nature.com/scientificreports/ In other reports, two fluorophores were linked with the cleaved amino acids to detect cleavage efficiency; however, the studies did not show whether the first protein or the second protein is needed in cis for its cleavage 21,22 . In most studies, the efficiency of the proteolytic process has been detected by using different extrinsic fluorophores to mimic NS polyprotein cleavage by NS2B3; however, this assay does not explain the real relationship between the substrate NS protein and the protease NS2B3. In our study, NS2A or NS2B were substituted with EGFP or GST protein, respectively, forming different polyprotein combinations such as EGFP-20aa-GST, EGFP-20aa-NS2B, and NS2A-20aa-GST. We found that EGFP-20aa-GST and EGFP-20aa-NS2B could not be cleaved by NS2B3, while NS2A-20aa-NS2B and NS2A-20aa-GST could. Further study confirmed that NS2A could be intermolecularly cleaved by NS2B3 regardless of whether NS2A was linked with NS2B or GST. Collectively, these results demonstrate that both NS2A and NS2B are in cis needed for NS2A/2B intramolecular protease cleavage, while NS2A could be intermolecularly processed by NS2B3.
It has been reported that NS2A can be intermolecularly cleaved by NS2B3 of YFV and DENV to form NS2Aα 14,[23][24][25] . This is similar to what we found here that DTMUV NS2A could be intermolecularly processed by NS2B3 to form NS2Aα and NS2Aβ, and NS2Aα/β could be cleaved by NS2B3pro in trans. According to previous reports, this cleavage site is important for YFV virus particle assembly and the production of infectious www.nature.com/scientificreports/ particles 14 . In the flavivirus life cycle, NS2A/2B cleavage produces NS2A, and NS2A first plays a role in viral genome replication. Then, NS2A is cleaved into NS2Aα, which is important for progeny virus assembly. Therefore, the NS2Aα/β cleavage should take place after NS2A/2B cleavage. However, in our study, we found that NS2Aα/β could be cleaved by NS2B3 regardless of a prior NS2A/2B cleavage. This result indicates that NS2Aα/β cleavage is independent of NS2A/2B cleavage the NS2Aα/β cleavage can occurs before NS2A/2B if the NS2A/2B cleavage site was modified. By comparing the amino acid sequence of DTMUV and the reported sequence of YFV, we speculate that the NS2Aα/β cleavage site in DMTUV is KK 194 ↓G 195 , two basic amino acids and a short chain amino acid; however, the experimentally confirmed sites of NS2Aα/β in DTMUV still need to be verified. Additionally, the function of NS2Aα in the DTMUV life cycle should be studied. The P1 to P10 sites and the P1′ site alanine mutations showed different effects on NS2A/2B cleavage. To understand their effects on viral proliferation, we chose to mutate the P1 and/or P1′ sites (RG/AG, RG/RA and RG/AA) of a DTMUV replicon to best maintainNS2A/2B function and structure. We found that when the production of mature NS2A and NS2B decreased, viral genome replication decreased. Interestingly, our data showed that CQW1-Rep-Rluc-NS2A/2B-P1′(A) replicated better than CQW1-Rep-Rluc-NS2A/2B-WT, while the production of NS2A and NS2B decreased when P1′was mutated to Ala. Therefore, we hypothesize that the P1′ site of NS2A/2B possesses unknown functions that can improve viral RNA replication. www.nature.com/scientificreports/ Because P1 mutations had a detrimental effect on the replicon, we chose to mutate the P1 and P1′ sites of a DTMUV infectious clone. Consistent with previous results, rDTMUV-NS2AB-P1P1′(AA) was attenuated both in vitro and in vivo, showing lower viral genome copies and titers, as well as a lower mortality in duck embryos. In the IFA assay, we found that rDTMUV-NS2AB-P1P1′(AA) could only infect a portion of the cells throughout the experiment, which suggested that the mutation affected their plication and any other step of viral life cycle. However, we do not know the exact mechanism by which this mutation affects virus attachment, assembly or release.
In summary, by mutating the cleavage sites, we found that 5 amino acids (P1-R, P1′-G, P2-R, P3-T, and P4-V) are vital for NS2A/2B cleavage. In addition, both NS2A and NS2B were essential in cis for polyprotein NS2A/2B intramolecular cleavage, while NS2Aα intermolecular proteinase cleavage was NS2B independent. Subsequently, our data on a DTMUV replicon and an infectious clone showed that the P1 site is essential to DTMUV genome RNA replication, while mutation of P1′ could boost viral replication. Furthermore, a recombinant virus with P1 and P1′ site mutations, named rDTMUV-NS2A/2B-P1P1′(AA), was rescued from transfected BHK21 cells. rDTMUV-NS2A/2B-P1P1′(AA) showed CPE from 5 days and continued to increase up to 8 days post transfection, while rDTMUV-WT showed CPE from 3 days and peak at 5 days post transfection. The maximum viral titers and viral genome copies of rDTMUV-NS2A/2B-P1P1′(AA) were much lower than those of rDTMUV-WT both in the intracellular and extracellular samples of infected BHK21. The virulence of rDTMUV-NS2A/2B-P1P1′(AA) was attenuated in 9-day-old duck embryonated eggs. Taken together, the above data indicated that the mutation of the intramolecular protease cleavage sites between NA2A/2B attenuates DTMUV both in BHK21 and in duck embryonated eggs. The NS2A/2B cleavage sites processed by the NS2B3 protease are vital for DTMUV proliferation and virulence.

Mutated replicon and infection clone construction. The mutant plasmids were produced by a Fast
Mutagenesis System (TransGen Biotech, Beijing, China). The DTMUV replicon was provided by our lab. Three different mutated replicons were constructed and named CQW1-Rep-Rluc-WT-NS2A/2B-P1(A), CQW1-Rep-Rluc-WT-NS2A/2B-P1′(A), and CQW1-Rep-Rluc-WT-NS2A/2B-P1P1′(AA) (Fig. 4A). Then, the plasmids were used to transfect BHK21 cells and the cells were harvested at different times post transfection. The replication of the replicon was measured with a Renilla luciferase reporter located upstream of the replicon. The plasmid pACYC FL-TMUV containing the full-length cDNA of DMTUV was provided by our lab 27 . Then, the P1 and P1′ sites were mutated to Ala by using a Fast Mutagenesis System (TransGen Biotech, Beijing, China), and the mutated plasmid was named pACYC FL-TMUV-P1P1′(AA).
Renilla luciferase activity assay. BHK21 cells transfected with WT or mutant replicon plasmids were lysed using cell lysis buffer (Promega, USA) and harvested at various time points. A Renillaluciferase assay system (Promega) and a Varioskan Flash luminometer (Thermo Scientific, USA) were used to detect the Rluc activity according to the manufacturer's instructions.
In vitro RNA transcription and transfection. The plasmids pACYC FL-TMUV and pACYC FL-TMUV-P1P1′(AA) were linearized by the restriction enzymes NotII or SmaII and purified by using the TaKaRaMiniBEST DNA Fragment Purification Kit (Takara, Japan). Then, the linearized DNA was transcribed to RNA with mMES-SAGE mMACHINE T7 Transcription Kit (Ambion, USA). For transfection, BHK21 cells were plated on 6-well plates, and after culture at 37 °C for 16 -NS2A/2B-F  CAT CAT TTT GGC AAA GAA TTC GCC ACC ATG GAG CAG AAA CTC ATC TCT GAA GAG GAT CTG TTT  CAA GGG GGT GGC ATG G   pCAGGS-Myc-NS2A/2B-R  TTG GCA GAG GGA AAA AGA TCT CTA CTT ATC GTC GTC ATC CTT GTA ATC TCG TTG TTT TGC CTT  AGT   pCAGGS-EGFP-F  CAT CAT TTT GGC AAA GAA TTC GCC ACC ATG GAG CAG AAA CTC ATC TCT GAA GAG GAT CTG ATG  GTG AGC AAG GGC GAG GAG   pCAGGS-GST-Flag-R  TTG GCA GAG GGA AAA AGA TCT CTA CTT ATC GTC GTC ATC CTT GTA ATC TTT TGG AGG ATG GTC  GCC  www.nature.com/scientificreports/ Virus titration and plaque assay. Virus titers were measured by the median tissue culture infectious dose 50 (TCID 50 ) method on BHK21 cells. The virus was serially diluted tenfold in DMEM, and each dilution was distributed to 8 wells of a 96-well plate. Then, the culture was incubated at 37 °C with 5% CO 2 for 7 days. Next, the cells were detected by microscopy, and the viral titers were calculated according to the Karber method. For the plaque assay, BHK21 cells were cultured in DMEM in a 6-well plate for 16 h, and then the medium was removed and the cells were washed with PBS three times. Virus samples were diluted and added to the cells with the same TCID 50, and then the samples were incubated for 1.5 h and swirled every 15 min. After the incubation, the virus sample was removed, and the cells were washed with PBS 3 times. Finally, 2 ml of 0.75% methyl cellulose overlay containing 2% FBS and 1% penicillin/streptomycin was added to each well. At 3 days post infection, the medium was removed, and the cells were washed 3 times with PBS and fixed with 4% paraformaldehyde for 20 min. After the fixative was removed and the cells were washed with PBS, the cells were stained with 1% crystal violet and washed carefully. Then, plaques could be observed in the plate.
Virulence of rDTMUV in duck embryos. All duck embryos were purchased from the Waterfowl Breeding Center of Sichuan Agriculture University and randomly divided into 3 groups. Five 9-day-old embryo eggs per group were infected with a 100-μL dilution of rDTMUV-WT or rDTMUV-NS2A/2B-P1P1′(AA) by allantoic cavity inoculation with 1,000 or 10 TCID 50 , and DMEM was used as a negative control. Then, the embryos were incubated at 37 °C for 10 days and checked daily with an egg candler. If the embryos had stopped moving and did not have clear blood vessels, the embryos were regarded as dead.
Ethics statement. This study was approved by the Committee of Experiment Operational Guidelines and