The deubiquitinating activity of Middle East respiratory syndrome coronavirus papain-like protease delays the innate immune response and enhances virulence in a mouse model

Middle East respiratory syndrome coronavirus (MERS-CoV) continues to cause zoonotic infections and serious disease, primarily in the Arabian Peninsula, due to repeated spill-over from dromedary camels and subsequent nosocomial transmission. Approved MERS vaccines for use in animals or humans are not currently available. MERS-CoV replication requires the virus-encoded papain-like protease (PLpro) to cleave multiple sites in the viral replicase polyproteins, thereby releasing functional non-structural proteins. Additionally, PLpro is a deubiquitinating enzyme (DUB) that can remove ubiquitin(-like) moieties from substrates, presumably to counteract host antiviral responses. In previous work, we determined the crystal structure of MERS-CoV PLpro in complex with ubiquitin, facilitating the design of PLpro mutations that impair DUB activity without affecting viral polyprotein cleavage. Here, we introduced these DUB-inactivating mutations into the viral genome and examined their impact on MERS-CoV infection both in cell culture and in a lethal mouse model. Although overall replication of DUB-negative and wild-type (wt) recombinant MERS-CoV was comparable in multiple cell lines, infection with DUB-negative virus markedly increased mRNA levels for interferon (IFN)-β and IFN-stimulated genes. Moreover, compared to a wt virus infection, the survival rate was significantly increased when DUB-negative MERS-CoV was used to infect transgenic mice expressing a human MERS-CoV receptor. Interestingly, DUB-negative and wt MERS-CoV replicated to the same titers in lungs of infected mice, but the DUB-negative virus was cleared faster, likely due to the observed accelerated and better-regulated innate immune responses, in contrast to delayed and subsequently excessive responses in wt virus-infected mice. This study provides the first direct evidence that the DUB activity of a coronaviral protease contributes to innate immune evasion and can profoundly enhance virulence in an animal model. Thus, reduction or removal of the innate immune-suppressive DUB activity of PLpros is a promising strategy for coronavirus attenuation in the context of rational vaccine development. Author Summary Although zoonotic coronaviruses such as Middle East respiratory coronavirus (MERS-CoV) have pandemic potential, therapeutics and vaccines that counteract this public health threat are not currently available. Coronaviruses typically employ multiple strategies to evade the host’s innate immune response, which may enhance clinical disease and/or reduce the efficacy of modified live vaccines. The MERS-CoV-encoded papain-like protease (PLpro) is not only crucial for the expression of functional replicase proteins, but has also been postulated to antagonize ubiquitination-dependent steps during the activation of the innate immune response. Here, we report the generation of engineered MERS-CoVs mutants in which PLpro’s deubiquitinating (DUB) activity was specifically disrupted without affecting virus viability. In this manner, we could demonstrate that the DUB activity of PLpro suppresses the interferon response in MERS-CoV-infected cells. Strikingly, in the lungs of mice infected with DUB-negative MERS-CoV, innate immune responses were induced at an earlier stage of infection than in wt virus-infected mice. This group also showed a clearly increased survival, indicating that the DUB activity is an important MERS-CoV virulence factor. This proof-of-concept study establishes that the engineering of DUB-negative coronaviruses, which elicit a more effective immune response in the host, is a viable strategy for vaccine development.

PL pro was equipped with substitutions that reduced the DUB activity but did not noticeably affect 160 polyprotein cleavage (65). This is important because PL pro plays a crucial role in virus replication 161 during the liberation of nsp1 to nsp4 from pp1a and pp1ab. In earlier research, the impact of such

274
Homogenized samples were centrifuged at 9,500 x g for 5 min and supernatants were collected.

275
Thereafter, rMERS-CoV titers were determined by plaque assay on HuH7 cells as described before.

280
To preserve the integrity of RNA, lungs were soaked in RNAlater (Thermo Fischer Scientific) directly 281 after harvesting, and stored at -80°C. For RNA extraction, lungs were transferred to a gentleMACS M 282 Tube containing 2 ml of RA1 buffer (Machery-Nagel) supplemented with 1% β-mercaptoethanol.

284
Subsequently, samples were centrifuged at 9,500 x g for 5 min and supernatants were aliquoted.

285
RNA was then isolated following the standard protocol of the NucleoSpin RNA II kit (Machery-Nagel).

286
RT-qPCRs and their analysis were performed as described above but with the following deviation. To 287 measure mRNA levels of CCL5, IL-6, and IFIT2, a triplex qPCR with TaqMan probes was performed 288 with HotStarTaq Master mix (Qiagen) whereas all other targets were amplified using gene-specific 289 primer sets and iTaq SYBR Green Supermix (BioRad). All primers and probe sequences can be 290 requested.

291
The obtained cDNA was also used to verify that the V1691R substitution in MERS-CoV PL pro 292 was retained after several rounds of virus replication in mice. PCR and sequencing to determine the 293 PL pro -coding sequence was executed as described above.

309
Virus neutralization tests were performed twice, and the mean of the two measurements was used. wt rMERS-CoV on both cell lines (Fig 1). Plaques of mutants T1653R and V1674S were comparable in 345 size to those of the wt control. However, when combined with the V1691R substitution in double or 346 triple mutants, plaque sizes were more variable and on average slightly decreased (Fig 1).  (Fig 2A and B). Also, the other mutants yielded titers comparable to wt virus 24 h p.i. when 363 MRC5 cells were infected with MOI 1 (Fig 2C). Notably, rMERS-CoV peak titers were approximately 364  (Fig 2D-G). Essentially similar results were obtained for all other DUB-negative mutants 368 tested (Fig 2D-G wt rMERS-CoV (Fig 3A-C). The double mutant showed a more pronounced increase while the single 385 T1653R mutation did not affect the mRNA levels of the qPCR targets at all (Fig 3A-C). Also, cells 386 infected with viruses carrying the V1674S substitution induced a very low response comparable to 387 that against wt virus (Fig 3A-C). Wild-type and mutant viral genomic RNA levels were similar in 18 388 infected cells (Fig 3D). In multicycle infections of MRC5 cells, V1691R rMERS-CoV induced higher 389 levels of IFN-β and ISGs mRNA than wt virus (IFN-β ~3-fold, IFIT2 ~4-fold, and viperin ~8-fold; Fig 3E-390 G). This was even more pronounced with the double mutant, while addition of the V1674S 391 substitution unexpectedly reduced IFN-β and ISG responses relative to the double mutant (Fig 3E-G).

392
The additive effect of T1653R implies that this substitution, when combined with V1691R, is further 393 reducing Ub-binding. PL pro mutants T1653R and V1674S generated responses comparable to those 394 against wt virus, and virus genome levels were similar (Fig 3E-H). Interestingly, MERS-CoV genome 395 levels were lower in cells infected with V1691R, with the double, and with the triple mutant, 396 compared to wt virus in these multicycle infections (Fig 3H)

412
We aimed to investigate V1691R rMERS-CoV infection of these transgenic mice with respect 413 to its lethality and the innate immune responses in the lung. To this end, mice were infected intranasally with 1x10 5 PFU of either wt or DUB-negative rMERS-CoV, whereas a third group of mice 415 was mock-infected. Each group contained a total of 16 mice of which 12 animals per group were 416 euthanized at early time points. The remaining four animals from each of these three groups were 417 monitored for a period of 14 days to assess their survival and measure their body weight on a daily 418 basis. All four mice infected with wt rMERS-CoV died or had to be euthanized after reaching a 419 humane endpoint by day 7 or 8 p.i. (Fig 4A). In contrast, only one of the four mice infected with 420 DUB-negative rMERS-CoV died at day 8 p.i., while the other three mice survived for the full 14-day 421 period (Fig 4A). Even with the small number of animals used for this experiment, this difference in 422 survival between mice infected with wt or DUB-negative rMERS-CoV is statistically significant (log-423 rank test p value = 0.028). Both wt and DUB-negative rMERS-CoV-infected mice lost weight over the 424 course of infection (Fig 4B), but in all three surviving mice infected with DUB-negative rMERS-CoV 425 this weight loss was reversed from day 10 p.i. forward (Fig 4C). The results clearly suggested that the 426 virulence of the DUB-negative virus was reduced compared to wt virus. progeny titers were approximately 1x10 6 PFU per g of lung tissue during the first four days after 433 infection (Fig 5). The DUB-negative virus reached similar virus titers as the wt control during the first 434 two days p.i., indicating that this virus also replicates well in mice. Unexpectedly, however, it proved 435 impossible to recover virus from the lungs of three of the mice infected with DUB-negative rMERS-436 CoV (twice in the day 1 p.i. group, and once in the day 2 p.i. group; Fig 5). Consequently, it remains 437 unclear whether these three animals had been successfully inoculated. All DUB-negative rMERS-CoV-438 infected mice that were euthanized at day 4 p.i. were virus-positive, and their lung titers were 439 significantly lower than in wt virus-infected mice at that time point (Fig 5), suggesting that the DUB-negative virus is cleared faster from the lungs than wt rMERS-CoV. Virus titers were also determined 441 in mice that died from the infection at day 7-8 p.i. and in survivors of the infection at day 14. By day 442 7-8 p.i., the rMERS-CoV wt titers were around 1x10 4 PFU per g of lung tissue, signifying that virus 443 titers decreased over time (Fig 5). The virus titer of the single mouse that died of the DUB-negative 444 virus infection was 2x10 3 PFU per g of lung tissue (Fig 5). Virus could not be recovered from the lungs 445 of any of the mice that had survived the infection with DUB-negative rMERS-CoV after 14 days (Fig   446  5).

447
In order to verify whether the DUB-inactivating substitution in the MERS-CoV mutant was 448 stably maintained during the animal experiment, sequencing of the PL pro -coding region of the 449 genome was performed. To this end, lung tissue was homogenized, RNA was isolated, and the PL pro -450 coding sequence was amplified by RT-PCR and sequenced. In wt virus-infected mice, the consensus 451 sequence of the PCR product was found to be identical to the sequence of the BAC-based cDNA 452 clone from which the rMERS-CoV had been launched (data not shown). RT-PCR and sequencing were 453 successful using RNA isolated from all mice infected with DUB-negative rMERS-CoV from which also 454 DUB-negative rMERS-CoV was recovered (Fig 5), and revealed the presence of the V1691R 455 substitution in PL pro without additional substitutions in PL pro (data not shown). No virus-specific RT-456 PCR product could be obtained from the lungs of mice that were found negative for infectious virus 457 at day 1 or 14 p.i. (data not shown; Fig 5). For one mouse inoculated with DUB-negative rMERS-CoV, 458 although virus could not be isolated at 2 days p.i. (Fig 5) IFN-λ), ISGs (ISG15 and IFIT2), the cytokine TNFα that is controlled by NF-κB activation, the pro-478 inflammatory cytokine IL-6, and the chemokine CCL5 (Fig 6) (76).

479
After the first day of infection, the mRNA levels for all targets were similarly low in wt and 480 DUB-negative rMERS-CoV-infected mice (Fig 6). At 2 days p.i., transcript levels in wt rMERS-CoV-481 infected mice were comparable to those in mock-infected mice while a clear increase was measured 482 in DUB-negative rMERS-CoV-infected mice (Fig 6). For ease of visualization, the mean fold increase 483 for each target is presented in Fig 6,  although not statistically significantly, higher than in the mock and wt virus-infected mice (Fig 6A   490 and B). ISG15 and IFIT2 mRNA levels were found to be slightly elevated in wt rMERS-CoV-infected mice relative to mock-infected mice, while an even more pronounced increase was seen in DUB-492 negative rMERS-CoV-infected mice (Fig 6C and D).

503
By day 4 p.i., a very different innate immune response in the mice was seen compared to the 504 response at 2 days p.i. (Fig 6). In the DUB-negative virus group, the increase of innate immune 505 transcripts was slightly lower relative to 2 days p.i. (Fig 6), and the variation in mRNA levels between 506 the four mice was smaller in this group at 4 days than at 2 days p.i. (S1 Fig, compare  showed, however, a very strong increase in transcript levels for all tested targets between 2 and 4 510 days p.i. Especially the mRNA levels of IFN-β, IFN-λ, and IL-6 were remarkably upregulated (Fig 6A,   both PL pro functionalities rely on the same enzymatic active site, which cannot be inactivated without 541 killing virus replication. Therefore, in our earlier research, we explored an alternative approach to 542 selectively inactivate the DUB activity of MERS-CoV PL pro . Based on the available crystal structure of 543 the PL pro -Ub complex, we disrupted the enzyme's Ub-binding site to minimize its affinity for Ub (65).

544
Based on assays carried out in expression systems, it was concluded that the DUB activity of PL pro 545 could indeed be inactivated almost completely without affecting the proteolytic activity needed for 546 viral polyprotein cleavage (65). Subsequently, our study suggested that the DUB activity of MERS-

547
CoV PL pro may indeed be involved in suppressing innate immune responses, as DUB-negative PL pro 548 was no longer able to downregulate IFN-β promoter activity. We have now engineered mutant DUB-549 negative viruses, thereby convincingly extending these observations to MERS-CoV infected-cells, in 550 culture and in vivo. We characterized the replication and the innate immune responses these viruses 551 induce in comparison to wt virus.

552
Interestingly, some of the DUB-negative viruses, specifically those containing the V1691R 553 substitution, induced higher levels of IFN-β and ISG mRNAs than wt virus in MRC5 cells, confirming a 554 role of PL pro in innate immune suppression, now demonstrated during infection (Fig 3). However, this 555 upregulation of the type I IFN response did not reduce the progeny titers of DUB-negative rMERS-556 CoV in MRC5 cells (Fig 2). We did observe that the plaque size of rMERS-CoV containing the V1691R 557 substitution was slightly smaller when compared to wt virus (Fig 1), suggesting that the mutant virus 558 was somehow affected after all. Also, in multi-cycle infections, MERS-CoV genome levels were 25 559 reduced upon infection with DUB-negative MERS-CoVs carrying the V1691R substitution (Fig 3H), CoV with a M1748R substitution (using SARS-CoV pp1a/pp1ab amino acid numbering) (81).

580
Compared to the wt control, replication of this mutant virus was slightly delayed in cell lines that can 581 mount an innate immune response, and it induced higher IFN-β mRNA levels (81). We here took the 582 characterization of DUB-negative CoV mutants to the next level, by testing a DUB-negative MERS-

583
CoV mutant in an animal model for the first time. inoculation, MERS-CoV replication was observed in lungs and wt rMERS-CoV titers were stable over 588 the first four days of infection, showing 1x10 6 PFU per g of lung tissue (Fig 5). Although this is lower 589 than the previously reported peak titer at 2 days p.i. of 6x10 7 PFU per g of lung tissue (73) infection was lethal also in our hands, and all mice died at 7-8 days after infection (Fig 4A). Likely, 597 the mice that succumbed to the infection also had MERS-CoV infection of the brain, however, the 598 experimental circumstances did not allow us to take samples to support this assumption. Strikingly, 599 infection with the DUB-negative virus resulted in a significantly increased survival compared to the 600 animals infected with the wt virus (Fig 4A). The DUB-negative virus reached the same titers in the 601 lungs as the wt virus by 2 days p.i., but these had dropped significantly by 4 days p.i. while they 602 stayed high in the wt virus-infected mice (Fig 5). In future research, it will be interesting to 603 determine whether histopathology is reduced in mice infected with DUB-negative rMERS-CoV to 604 confirm that earlier virus clearance and reduced tissue damage correlates with increased survival, 605 relative to in wt virus-infected mice.

606
Messenger RNA levels for IFNs (IFN-β and IFN-λ), ISGs (ISG15 and IFIT2), and pro-607 inflammatory cytokines (IL-6 and TNFα) started to rise in lungs of DUB-negative rMERS-CoV-infected 608 mice at 2 days p.i. (Fig 6). Although there was no statistically significant difference compared to wt 609 virus-infected mice, the trend of faster upregulation of these mRNAs was quite clear. Moreover, the 27 610 fact that the levels for all these transcripts increased simultaneously strengthens the conclusion that 611 the DUB-negative rMERS-CoV induced an accelerated innate immune response compared to the wt 612 virus, again confirming the role of the PL pro DUB activity in suppression of innate immune responses, 613 now also in the context of a mouse model. The transcript levels of IFNs and cytokines slightly 614 decreased in DUB-negative rMERS-CoV-infected mice at 4 days p.i., but in wt rMERS-CoV-infected 615 mice the mRNA levels were extremely upregulated (Fig 6). We therefore hypothesize that the 616 difference in survival rate is likely due to an accelerated, and after induction properly down-617 regulated, innate immune response in DUB-negative virus-infected mice, which is likely more 618 effectively clearing the virus and preventing tissue damage, protecting these mice from lethality 619 (Figs 4-6). In contrast, the response in wt virus-infected mice is first suppressed and then at later 620 time points it is extremely upregulated (Fig 6).

709
The mean per group is presented and the dashed line represents the average that was set to 1 for 710 the mock-infected mice. Statistical significance between wt and DUB-negative MERS-CoV-infected 711 mice is shown (* p < 0.05).