SARS-CoV-2 SUD2 and Nsp5 Conspire to Boost Apoptosis of Respiratory Epithelial Cells via an Augmented Interaction with the G-Quadruplex of BclII

ABSTRACT The molecular mechanisms underlying how SUD2 recruits other proteins of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) to exert its G-quadruplex (G4)-dependent pathogenic function is unknown. Herein, Nsp5 was singled out as a binding partner of the SUD2-N+M domains (SUD2core) with high affinity, through the surface located crossing these two domains. Biochemical and fluorescent assays demonstrated that this complex also formed in the nucleus of living host cells. Moreover, the SUD2core-Nsp5 complex displayed significantly enhanced selective binding affinity for the G4 structure in the BclII promoter than did SUD2core alone. This increased stability exhibited by the tertiary complex was rationalized by AlphaFold2 and molecular dynamics analysis. In line with these molecular interactions, downregulation of BclII and subsequent augmented apoptosis of respiratory cells were both observed. These results provide novel information and a new avenue to explore therapeutic strategies targeting SARS-CoV-2.

Previously, most studies on SARS-CoV-2 proteins focused on the spike and RNA-dependent RNA polymerase proteins, which mediate entry and replication of viruses (6,7). The pathogenic proteins leading to the severe syndrome remain a separate puzzle (8,9). Historically, the initially discovered strains of human coronaviruses were identified as only mild pathogens (10)(11)(12) until the emergence of SARS-CoV in 2003 (13,14). Subsequent research revealed a sequence exclusively present in the genome of SARS-CoV encoding a special domain located at the N terminus of nonstructural protein 3 (Nsp3) of SARS-CoV (amino acid residues 389 to 652), which was originally named the SARS unique domain (referred to as SUD1 in this study) and was found to be responsible for the enhanced viral pathogenicity (15,16). One of the most intriguing characteristics of SUD1 was its ability to interact with a special noncanonical DNA structure, the G-quadruplex (G4), which might account for its severe pathogenicity (17)(18)(19).
The Nsp3 of SARS-CoV-2 also contains a SUD portion (referred to as SUD2 in this study) in Nsp3, which shares about 75% similarity in amino acid residues with SUD1 (20). The high homology of the two prompted us to investigate how the interactions of SUD2 with G4 are related to its pathogenicity. Our initial experiment indicated that some critical G4s did indeed interact with SUD2 with diverse binding affinities in vitro, and a similar phenomenon has been reported by another group as well (20). However, the exact mode of interaction (SUD2-G4) that leads to the severe pathogenicity of SARS-CoV-2 is still unclear.
Importantly, most viral proteins exert their functions as part of multiprotein complexes, particularly for many DNA binding proteins (21,22). The whole genome of SARS-CoV-2 can encode 27 different proteins, including Nsp1 to -16, structural proteins (S, E, M, and N), and accessory proteins (23,24). Since SUD2 is a part of Nsp3, we further investigated if SUD2 interacted with other Nsps, whether such interactions influenced the G4 binding property of SUD2, and how such interactions affected the host cell's fate, which is still an untouched theme.
In this study, using a yeast two-hybrid (Y2H) system (25)(26)(27), Nsp5, previously named main proteinase (Mpro) or 3-chymotrypsin-like proteinase (3CL pro ), was singled out among the Nsps as the only previously unknown partner of interaction for SUD2-N1M (named SUD2 core in this study) (28,29). In host human respiratory cells, both SUD2 core and Nsp5 were localized to the cytosol and nucleus, and their interaction in living human cells was confirmed by coimmunoprecipitation (co-IP) and bimolecular fluorescence complementary (BiFC) assays (30,31). Furthermore, a G4 pulldown assay indicated that Nsp5 strengthened the binding affinity of SUD2 core with the B cell lymphoma 2 (BclII) promoter G4 sequence (Bcl2G4). The atomic interaction between SUD2 core , Nsp5, and Bcl2G4 was simulated theoretically to infer their crucial structural features. This interaction downregulated the expression of BclII, which in turn led to augmented apoptosis of transfected host cells. Since the induced apoptosis of infected respiratory cells is a hallmark of severe COVID-19 illness (32)(33)(34)(35), these results provide a possible explanation for how SUD2 core , together with its partner Nsp5, lead to severe disease.

RESULTS
Screening of SUD2 binding partners in Nsp proteins of SARS-CoV-2. SUD2 (from K412 to S743 of SARS-CoV-2 Nsp3) consists of three subdomains: macrodomains II and III and the domain preceding Ubl2 and PL2 pro (DPUP) (30,36), which are also named SUD-N (K412 to E548), SUD-M (I549 to S675), and SUD-C (S676 to S743), respectively (Fig. 1A). In order to investigate the direct interactions of SUD2 with other Nsps of SARS-CoV-2, 14 Nsps were fused to the activation domain (AD Leu ) to act as 14 distinct preys, and SUD2 core (SUD2-N1M [K412 to S675]), due to its in vitro stability, was used as the bait fused to the DNA binding domain (BD Trp ) in the Y2H system (Fig. 1B). Through phenotypic analysis of positive clones growing on synthetically defined (SD) medium lacking histidine, the direct interactions could be efficiently determined, as SUD2 core recruits the directly interacting Nsp, thus activating the reporter gene histidine synthetase 3 (HIS3) (37)(38)(39). Successfully transfected yeast clones containing the 14 different AD-Nsps and BD-SUD2 core were selected from SD media lacking tryptophan and leucine (SD/-Trp/-Leu [DDO]) (Fig. 1C, upper panel). From the chosen clones, only the one encoding Nsp5 grew well in the SD medium lacking tryptophan, leucine, and histidine (SD/-Trp/-Leu/-His [TDO]) (Fig. 1C, lower panel), indicating that only Nsp5 could directly interact with SUD2 core . In this clone, the expression of both SUD2 core and Nsp5 were verified by appropriate antibodies (see Fig. S1A and B in the supplemental material). The negative-control clones could not activate HIS3 (Fig. S1C).
In order to further examine the contact domains of the SUD2-Nsp5 interaction, the full-length SUD2 (N1M1C) and four truncated versions were generated and scrutinized by the same process. SUD2-N1M1C, SUD2-N1M, SUD2-N, and SUD2-M grew well on TDO media, but SUD2-C did not, suggesting that the full-length and three N-and M-containing truncations interacted with Nsp5 and that SUD2-C did not (Fig. S1D). Next, to quantify the interaction strength of the different truncations, interaction-dependent LacZ reporter activities were tested (39). SUD2-N1M showed a significantly enhanced induction effect that was ;2.1-fold higher than the that with the full length of SUD2. Of note, SUD2-N and SUD2-M individually displayed reduced reporter activities (47% and 86%, respectively) compared with SUD2-N1M, and SUD2-C alone lacked this interaction (Fig. 1D). Apart from the enzymatic assay, a high physical affinity (equilibrium disassociation constant [K D ] of 37.4 nM) between SUD2 core and Nsp5 was quantitatively determined by biolayer interferometry (BLI) assay, in which the kinetic stability of the SUD2 core -Nsp5 complex was also indicated via the flat dissociation slopes ( Fig. 1E and F).
Consequently, Nsp5 has been clearly distinguished as the only Nsp component of SARS-CoV-2 to interact with SUD2 core . Moreover, the middle sequence between the N and M domains of SUD2 seems to constitute a major determinant of the Nsp5 interaction, while SUD-C does not interact with Nsp5. According to these results, SUD2 core was chosen for the subsequent studies.
SUD2 core and Nsp5 form a complex in respiratory epithelial cells. After the direct interaction between SUD2 core and Nsp5 had been established in yeast cells, the presence of this interaction in normal human bronchial epithelial (NHBE) cells, the first step in SARS-CoV-2 infection (40,41), was addressed by co-IP assays. The successful expression of green fluorescent protein (GFP), SUD2 core -FLAG, and Nsp5-GFP was confirmed with input samples (Fig. 1G, left). During anti-FLAG antibody pulldown, Nsp5-GFP was detected with anti-GFP antibody. Conversely, SUD2 core -FLAG was detected while using anti-GFP as immunoprecipitant (Fig. 1G, right). Nonspecific binding was not detected in the control experiment with GFP empty vector. These results demonstrated that SUD2 core and Nsp5 genuinely formed a complex in respiratory epithelial cells.
Regarding intracellular localization, only isolated Nsp3-N (the N terminus of Nsp3) and full-length Nsp5 have been previously reported in HEp-2 cells (42). The subcellular localization of the SUD2 core -Nsp5 complex in respiratory epithelial cells is as yet unknown. To address this problem, GFP-labeled SUD2 core and Nsp5 were separately expressed in NHBE and H1299 cells (Fig. S2A). Through confocal laser scanning microscope (CLSM) visualization, both SUD2 core and Nsp5 were found in the cytosol and nucleus of these cells (Fig. S2B).
To validate that the SUD2 core -Nsp5 complex also formed in cells, BiFC assays were performed (30,31). The VENUS N-segment fused to SUD2 core (SUD2 core -VENUS N ) and the VENUS C-segment fused to Nsp5 (Nsp5-VENUS C ) were cotransfected into NHBE and H1299 cells ( Fig. 1H and Fig. S2C). When detected at 515 nm, cells that contained both SUD2 core -VENUS N and Nsp5-VENUS C proteins yielded a strong yellow fluorescent protein (YFP) signal in both the cytosol and nucleus (Fig. 1I). The negative controls, VENUS N 1VENUS C , SUD2 core -VENUS N 1VENUS C , and Nsp5-VENUS C 1VENUS N , all failed to display any signal (Fig. 1I). Together, these experiments provided evidence of intracellular SUD2 core -Nsp5 complex formation in respiratory epithelial cells and showed both nuclear and cytoplasmic subcellular localization.
The G4 structure in the promoter of BclII has a negative regulatory effect. Since the SUD2 core -Nsp5 complex is found in the nucleus, an intriguing question is whether the SUD2 core -Nsp5 complex can also interact with G4 DNA structures, such as in the cases of individual SUD1 (18,20,43,44), and what the biological consequence of this interaction is. It has been estimated that there are 300,000 sequences in the human genome with the potential to form the G4 structure, and many of them have regulatory effects on gene expression when located in the promoter regions (45,46). In this study, we chose BclII as the target gene, because its promoter contains a G4 formation sequence and its downregulation is associated with the increased apoptosis of respiratory cells in patients with severe COVID-19 (35).
A 23-nucleotide (nt) sequence (referred as Bcl2G4 in this study) (Table S1) in the P1 promoter of BclII, about 1,700 nt upstream of the transcriptional start site, has been reported to have the capability to form a G4 structure in vitro (47)(48)(49). However, the relationship between the G4 structure of this sequence and the reduced expression of BclII in cells was unknown. Thus, we examined the formation of Bcl2G4 and its effect on the expression of BclII in respiratory epithelial cells.
In the in vitro validation, a G4 structure with a predominantly unimolecular parallel topological characteristic formed by Bcl2G4 in K 1 buffer (a necessary cationic condition) was indicated by the circular dichroism (CD) spectrum ( Fig. 2A and B), which displayed a characteristic positive peak around 264 nm and a negative peak close to 240 nm (50). This was further confirmed in native polyacrylamide gel electrophoresis (PAGE) with either Cy5-labeled or anti-G4 antibody (BG4) immunoblotting (Fig. 2C). MYCG4, a well-known G4 structure, was used as a positive control (20). In the intracellular validation, Cy5-labeled Bcl2G4 DNAs were transfected into NHBE and H1299 cells and detected at 649-nm excitation by CLSM. G4 formation in living cells was then visualized by the BG4 antibody immunofluorescence method (51). The overlap of Cy5 and BG4 signals for the wild-type Bcl2G4 DNA in living respiratory epithelial cells, appearing as yellow regions, decisively confirmed that Bcl2G4 can form the G4 structure intracellularly ( Fig. 2D and E). In contrast, no G4 structure formation was observed for Bcl2G4Mut DNA, in which the corresponding guanines were replaced by adenines in the quadrilateral planes.
Next, the regulatory effect on BclII expression by Bcl2G4 was investigated. The core region of the BclII promoter containing Bcl2G4 was constructed in the luciferase (LUC) expression system (Fig. 2F). A dual-luciferase reporter gene activity assay showed that the relative activities of firefly luciferase of pBcl2G4-WT in all three respiratory epithelial cells were decreased compared to that in the pBcl2G4-Mut cells, demonstrating that the Bcl2G4 structure suppressed the transcriptional activity of BclII promoter in host cells (Fig. 2G, H, and I). Based on these experiments, Bcl2G4 does indeed act as a regulator of BclII transcription.
SUD2 core tightly binds with Bcl2G4 DNA, and Nsp5 enhances this interaction. After Bcl2G4 formation and its role in living cells had been established, its interaction with SUD2 was investigated. Recombinant SUD2 core was expressed in Escherichia coli and purified by ion affinity and gel filtration chromatography ( Fig. 3A and Fig. S3A to C). An electrophoretic mobility shift assay (EMSA) demonstrated that, under the G4 formation conditions, SUD2 core bound tightly to Bcl2G4 DNA, displaying a lagging band close to the positive pole. In contrast, single-stranded DNA in the non-K 1 buffer barely bound to SUD2 core (Fig. 3A). Additionally, a G4 pulldown assay was also performed (Fig. 3B) to investigate the interactions (20). After incubation with 59-biotin-labeled Bcl2G4WT and then enrichment with streptavidin-coated magnetic beads, SUD2 core was detected by immunoblotting with anti-His antibodies, but the Bcl2G4Mut sequence displayed much lower band intensity (Fig. 3B). Furthermore, when another G4-forming sequence, MYCG4, was used as a positive control, a consistent result was observed ( Fig. 3B) (20).
Quantitatively, in the BLI assay, SUD2 core and Bcl2G4 DNA was found to exhibited a high binding affinity (K D = 85 nM) (Fig. 3C). By comparison, non-G4 structures in the absence of K 1 of the Bcl2G4 sequence showed a 14-fold-higher K D , 1.22 mM, illustrating the necessity of the G4 structure that mediates the interaction (data not shown). Intracellularly, a chromatin immunoprecipitation quantitative real-time PCR (ChIP-qPCR) assay was employed to verify the SUD2 core -Bcl2G4 interaction in NHBE, H1299, and A549 living cells. Constitutive SUD2 core -FLAG-expressing cell lines were obtained through lentivirus infection (Fig. S3E). In all tested cells, after immunoprecipitation with FLAG antibody and subsequent qPCR, a much higher enrichment of the Bcl2G4 . The amount of retained His-tagged SUD2 core with G4 DNA was detected with anti-His antibody. A 10% dilution of His-SUD2 core was used as input. The mutant forms of G4 are named with "Mut." (C) Binding kinetics of SUD2 core with Bcl2G4 measured by BLI assays. The biotin-labeled G4s were immobilized on the surface of streptavidin (SA) biosensors as indicated in the upper scheme. (D) ChIP-qPCR assays for the confirmation of SUD2 core -G4 interaction in different cells that constitutively expressed the SUD2 core domain. The genomic DNA templates were immunoprecipitated by FLAG antibody or IgG antibody (negative control). (E) Purified Nsp5 and SUD2 core detected by CBB staining. (F) Amount of retained His-tagged SUD2 core with G4s oligonucleotides was detected with anti-His antibody in a G4 pulldown assay. SUD2 core -Mut represents for K578A, S582A, and R586A mutants of Nsp3, which abolished the SUD2 core -Bcl2G4 interaction based on structural docking. (G) Relative abundance of retained SUD2 core corresponding to that in panel F. Data represent the means 6 SEM of triplicate biological results. Statistical significance was calculated by unpaired Student's t test. ***, P , 0.001. SUD2-Nsp5 Regulate BclII-Mediated Apoptosis Responses mBio region was detected, compared with the IgG negative controls (Fig. 3D). Therefore, intracellular complex formation of SUD2 core with Bcl2G4 was confirmed. The effect of Nsp5 on this SUD2 core -Bcl2G4 interaction was then inspected. Following a similar preparation procedure as for His-SUD2 core , glutathione S-transferase (GST)-Nsp5 was obtained ( Fig. 3E and Fig. S3D). As can be seen from G4 pulldown assay, the addition of Nsp5 had no impact without G4 formation (in the absence of K 1 ) (Fig. 3F). In contrast, in the presence of K 1 , the amount of SUD2 core binding with Bcl2G4 was significantly elevated 6.3-fold higher than without Nsp5 ( Fig. 3F and G). Furthermore, in order to examine the effect of Nsp5 on the binding specificity of SUD2 core , two other G4-forming sequences (VEGFR2G4 and KRASG4) (Table S1) in the promoters of vascular endothelial growth factor receptor 2 (VEGFR2) and Kirsten Rat sarcoma viral oncogene homolog (KRAS) were employed in comparison experiments (52,53) (Fig. S4A). Under G4 formation conditions, SUD2 core interacted preferentially with Bcl2G4 and not with VEGFR2G4 or KRASG4 ( Fig. S4B and C). Upon the addition of Nsp5, the selective binding behavior of SUD2 core was not altered.
These results demonstrated that Nsp5 enhanced the interaction between SUD2 core and Bcl2G4, and the tertiary complex of SUD2 core -Nsp5-Bcl2G4 was more stable than that of the binary SUD2 core -Bcl2G4 alone.
Inhibition of BclII expression by SUD2 core is augmented by Nsp5 in a G4-dependent manner. As Bcl2G4 downregulated BclII expression and Nsp5 enhanced the intrinsic binding affinity between SUD2 core and Bcl2G4, qPCR was performed to evaluate whether BclII expression was altered in the presence of SUD2 core versus SUD2 core -Nsp5 in different respiratory epithelial cells. SUD2 core alone reduced the expression of BclII to 53%, 52%, and 56% in NHBE, H1299, and A549 cells, respectively, compared with non-SUD2 core -transfected cells ( Fig. 4A to C), confirming the negative regulatory role of SUD2 core on BclII expression. Next, the influence of the SUD2 core -Nsp5 complex on the expression of BclII was examined. While Nsp5 had a minimal effect on BclII expression, cotransfection of SUD2 core and Nsp5 strongly reduced the BclII mRNA level to 21%, 22%, and 38% in NHBE, H1299, and A549 cells, respectively, nearly doubling the effects of SUD2 core alone ( Fig. 4A to C). Consistent with the low binding affinity of SUD2 core for VEGFR2G4 and KRASG4, no significant differences were observed for VEGFR2 and KRAS expression in any of the three cell lines (data not shown). Thus, we showed that SUD2 core selectively reduced the expression of BclII in respiratory epithelial cells, and more importantly, SUD2 core -Nsp5 showed enhanced downregulation of BclII expression compared to SUD2 core alone.
To confirm that this regulatory effect is G4 dependent, a dual-luciferase reporter gene assay was performed (Fig. 4D). The results clearly showed that the relative LUC activity of the BclII-WT promoter (which formed the G4 structure in NHBE and H1299 cells) was significantly decreased in the presence of SUD2 core , but not Nsp5 alone, and reached the lowest activity in the presence of both SUD2 core and Nsp5, compared with nontreated cells (Fig. 4E and F). In addition, this suppressive effect was lost in the BclII-Mut promoter, which could not form G4 structure, confirming that the downregulation of target gene expression by SUD2 core and the SUD2 core -Nsp5 complex is dependent on G4 structure formation.
Cohort studies indicated that decreased BclII expression resulted in the increased death of respiratory cells associated with severe COVID-19 in patients, due to the antiapoptotic role of BclII (33,35). Since SUD2 core has been found to negatively regulate the expression of BclII, to further investigate if this interferes with the expression of other BclII-mediated apoptosis-related genes (Fig. 4H), we performed RNA sequencing (RNA-seq) in NHBE cells in the presence and absence of SUD2 core (Fig. 4G). In addition to inhibiting the expression of BclII, SUD2 core negatively regulated the BclII-mediated antiapoptotic cascade. Specifically, expression of proapoptotic elements, such as BclII antagonist/killer 1 (Bak1), apoptotic peptidase activating factor 1 (Apaf1), and Caspase 3/8 were increased, but antiapoptotic AKT serine/threonine kinase 2 (Akt2) was decreased (48, 54, 55). These differential expressions were further confirmed by qPCR (Fig. 4I to L). SUD2 core and Nsp5 cooperatively promote apoptosis in epithelial cells. After confirming the inhibitory effect of SUD2 core on BclII expression and that the SUD2 core -Nsp5 complex enhanced this effect, we anticipated an overall proapoptotic biological consequence. To confirm this, the apoptosis rate of respiratory cells was measured by flow cytometry (FCM) after Annexin and propidium iodide (PI) staining. In transient expression systems, the cells underwent minimal apoptosis in the absence of SUD2 core and Nsp5. The apoptosis rates were increased up to 2.6-fold and 9.9-fold in the presence of SUD2 core FLAG in NHBE and H1299 cells, respectively, compared with that with FLAG empty vector (EV FLAG ) (Fig. 5A, B, G, and J), whereas Nsp5 myc alone had little effect on apoptosis rate, confirming the determinant role of SUD2 core and the auxiliary role of Nsp5 (Fig. 5A, B, G, and J). In accordance with their enhanced inhibitory effect on BclII expression, the apoptosis rate of NHBE cells in the presence of both SUD2 core FLAG and Nsp5 myc reached 56.5%, 1.5-fold higher than that of cells transfected with SUD2 core alone.
Similarly, in the SUD2 core FLAG stable expression cell lines generated by lentiviral processing, higher apoptosis rates (20.03% and 27.83%) were also observed in NHBE and H1299 cells, compared with that with lentiviral EV FLAG (6.7% and 6.04%) (Fig. 5C, D, H, and K). Moreover, the transient expression of Nsp5 myc in the stable SUD2 core FLAGexpressing cells robustly elevated the apoptosis rate, about 2.38-fold and 1.62-fold higher than that of non-Nsp5-treated NHBE and H1299 cells (Fig. 5C, D, H, and K). Immunoblotting confirmed that SUD2 core decreased BclII protein levels to about 41% of that with mock control (Fig. 5E and F). After cotransfection of SUD2 core and Nsp5, the BclII protein levels decreased dramatically to only 12% and 17% compared with the control NHBE and H1299 cells ( Fig. 5E and F). Consistently, the cells transfected with SUD2 core and Nsp5 exhibited the lowest proliferation rate compared with both mock control and individually transfected cells ( Fig. 5I and L). These results demonstrate that in line with negative regulation on BclII, SUD2 core -Nsp5 can significantly promote apoptosis in respiratory epithelial cells.
Structural features of the SUD2 core -Bcl2G4 versus SUD2 core -Nsp5-Bcl2G4 complex. Since atomic interacting features lay the foundation of the structure-propertyfunction relationship axis of the biological molecules and their complexes, which can also provide crucial information for developing potential intervening measures, after confirmation of the interaction and biological function of SUD2 core -Nsp5 with Bcl2G4, we further investigated the precise interacting surface of the complex and the conformation changes from monomer to ternary complex. Despite many tries, both the monomeric SUD2 core and corresponding complexes were refractory to forming suitable crystals for X-ray diffraction analysis. Thus, the final structural information was separately obtained through AlphaFold modeling and use of ZDOCK, followed by molecular dynamics (MD) optimization (Fig. 6A).
The overall structure of the SUD2 core monomer was quite similar to SUD1 core (PDB code 2W2G) (Fig. S5A), giving strong structural support for their common G4 DNA binding capacity. As a whole, SUD2 core is more conformationally unstable than are the individual SUD2-N or SUD2-M subdomains, due to the highly flexible linker between N and M domains, indicated by calculation of the root mean square deviation (RMSD) (Fig. S5B).
In the case of the binary SUD2 core -Bcl2G4 complex, a positively charged groove at the tip of the M domain of SUD2 was geometrically and electronically optimal for surface interaction with the negatively charged surface of Bcl2G4 (PDB code 2F8U), in which K578, S582, and R586 of SUD2 would form hydrogen bonds with the phosphodiester backbone of dG22 (2.89 Å), dG21 (2.72 Å), and dG17 (2.49 Å) of Bcl2G4 DNA, respectively (Fig. 6B). These three amino acids also displayed low free energy in the presence of Bcl2G4 (Fig. S6A to D). Based on this information, we expressed and purified the mutation of SUD2 core -Mut (K578A, S582A, and R586A) and found that the mutant protein domain did significantly reduce the binding affinity with Bcl2G4 DNA (Fig. 3E and F).
A well-fitted heterodimeric structure of the SUD2 core -Nsp5 complex was obtained by using the AlphaFold2 multimer algorithm, in which two complementary surfaces crossing the N and M domains of SUD2 mediate the crucial interactions with Nsp5: (i) a groove in the SUD2-N domain bound to a wedge of Nsp5 containing the amino acid residues from A191 to T198 of Nsp5 was identified; (ii) another binding pocket was determined to occur between a wedge of SUD2-M containing amino acid residues Q547, L557, V657, and Y668 and a groove of Nsp5 ( Fig. 6C and Fig. S7B to D). Therefore, in this complexation manner, Nsp5 could stabilize the lowest-energy conformation of SUD2 core (closed status exhibiting 94°between N and M domain) and prevent the conformational exchange (Fig. S5C). Consistently, in MD analysis, the SUD2 core -Nsp5 complex was more stable than the SUD2 core monomer (Fig. S5B and S7A). Together, the structural modeling results mesh very well with the affinity experiment with truncated proteins described above (Fig. 1D).
For the tertiary SUD2 core -Nsp5-Bcl2G4 complex, Bcl2G4 was calculated to interact well in the original pocket, albeit with altered interaction numbers and positions of SUD2 core : T577, K578, S609, and K610 formed hydrogen bonds with the phosphodiester backbone of dA13, dT16, dA10, and dG11 of Bcl2G4 DNA, respectively (Fig. 6D). In addition to these SUD2 core -Bcl2G4 interactions, there was an added interaction between the dG18 arm of Bcl2G4 and Arg188 of Nsp5 (Fig. 6D). The space-filling electrostatic surface charge distribution also showed that a positively charged groove on the SUD2 core -Nsp5 complex when bound to a negatively charged Bcl2G4 resulted in a more stable insertion of G4 into SUD2 core (Fig. 6D). The free energy analysis indicated a much lower averaged total free energy of SUD2 core -Nsp5-Bcl2G4 (;90 kcal/mol) compared with SUD2 core -Bcl2G4 (;40 kcal/mol) ( Fig. S6B and S8A to D), which could explain the experimental observation that Nsp5 enhanced the binding of SUD2 core with Bcl2G4 ( Fig. 3F and G).
Therefore, the geometric complementarity and intermolecular and electrostatic interactions together account for the enhanced stability of the tertiary complex of SUD2 core -Nsp5-Bcl2G4, compared to those for the binary complex of SUD2 core -Bcl2G4.

DISCUSSION
The discovery of the SUD1 sequence traces back to the SARS epidemic in 2003 (15). Later, the SUD1 domain of SARS-CoV-2 Nsp3 was reported to bind with oligo G nucleotides, especially G4 structural DNA, which may result in the severe pathogenicity of SARS-CoV (17,18,44). In the COVID-19 pandemic, the similar SUD2 sequence was again identified in the genome of SARS-CoV-2 (30). Recently, Lavigne et al. biophysically confirmed that the monomeric SUD2 core of SARS-CoV-2, just like SUD1, can also bind with some G4s structures in vitro (20). However, two aspects surrounding SUD2 core need further investigations. First, whether its interaction with DNA G4 structure exists in host cells remains to be confirmed, and the biological consequences in the host cells of this kind of interaction were not fully examined. Second, in addition to individual components, protein-protein interactions (PPI) mediated by embedded domains are of fundamental importance for understanding the disease mechanism and development of efficient intervention options (56)(57)(58). In the case of COVID-19, the elucidation of the interactions between viral and host proteins has led to the identification of some aspects of the infectious mechanism and potential effective therapies, including antibody and small-molecule agents (59). Apart from viral-host PPI, the interplay between the viral proteins themselves also play crucial roles in the life cycle of SARS-CoV-2. For instance, Nsp7 and Nsp8 help Nsp12 to make new copies of the RNA genome for assembling new viruses (60). However, how SUD2 core cooperates with other Nsp's to exert its pathological effect was unknown. Surrounding these two aspects, in this study new mechanistic and structural information regarding such interactions in vitro and in cells was provided, not just for SUD2 core monomer but also for its viral protein complex. We used a Y2H system to screen SUD2 core binding partners in Nsp's of SARS-CoV-2; Nsp5 was identified as the sole partner of SUD2 core , and their binding affinity (K D = 37.4 nM) was quantitatively measured in vitro. The intracellular interaction between SUD2 core and Nsp5 was confirmed and detected in the nucleus of respiratory epithelial cells, suggesting that SUD2 core and Nsp5 indeed formed a complex in host cells. Moreover, we validated that SUD2 core tightly binds with Bcl2G4 DNA and Nsp5 enhances this interaction. Consequently, this resulted in reduced BclII expression and an enhanced apoptosis rate of respiratory epithelial cells. The structural features of this complex, including detailed residue interactions and the more stable conformation of this protein-G4 DNA tertiary complex, have been rationalized through theoretical modeling and MD analysis.
Regarding the binding target, sequencing data and bioinformatic studies have revealed the prevalence of many G4 motifs at gene regulatory regions in the human genome (45,46). Despite the guanine tetrad serving as the common inner core unit of G4 quadruplexes, distinct topologies and local loop structures have been found for many different G4 sequences. This G4 structural diversity controls their interactions with other components and the resulting gene regulation effects. Though a certain extent selective binding of SUD2 core with G4 sequences has been observed (i.e., Bcl2G4 versus VEGFR2G4 and KRASG4), considering the huge number of G4 sequences existing at different loci of the human genome (which cannot be exhaustively investigated at one time), the interaction of SUD2 core with G4s in other gene regulatory regions cannot be excluded. For instance, as mentioned above, SUD2 core can also bind to MYCG4 (20). In line with this deduction, from GO and KEGG enrichment analysis of comparative RNA-seq data, SUD2 core also regulates the expression of genes involved in other biological processes and pathways, such as DNA binding, ISG15-protein conjugation, herpes simplex virus 1 infection, and interleukin-17 signaling pathway (Fig. S9). Regarding the complexity of transcription regulation, among these alterations induced by SUD2 core , further investigation is needed to determine which ones are the direct consequence of SUD2 core -Bcl2G4 interaction.
The molecular modeling and docking features of the SUD2 core -Bcl2G4 complex predicted in this study are different from those in a previous report (20). Here, a positively charged groove at the top of the M domain of SUD2 interacted with a negatively charged surface of Bcl2G4, in which hydrogen bonds formed between K578, S582, and R586 of SUD2 and the phosphodiester backbone of Bcl2G4 were detected. In comparison, in a previous report, Bcl2G4 docked at the SUD-N-SUD-M interface and comprised residues K592 and E595 (20). Two reasons might account for this discrepancy. First, in Lavigne's study a dimeric SUD2-NM was employed for the docking calculation, and we chose the monomeric unit, considering our screening, biophysical, and biochemical experiments. Second, different algorithms of computational performance were employed. In our study, the conformation search algorithm was performed by the fast Fourier transform in the ZDOCK program, which mainly aims at the docking of protein-protein or protein-nucleic acid complexes (61). In the previous study, the conformation search algorithm used was a stochastic global optimization with the Autodock-vina program, which is more commonly used for small-molecule-protein complexes (62). To verify our calculated result, the substitution of K578A, S582A, and R586A of SUD2 was generated for the BLI assay, which indicated that these three amino acids did play important roles in binding with Bcl2G4. Surely, the possibility of multiple G4 binding sites of SUD2 core may exist, and the roles of the K592 and E595 residues predicted in the previous study for G4 binding deserve further investigation.
Considering SUD2 as an individual functional domain, in addition to participating in G4 DNA binding, it was also characterized to enhance viral RNA translation (20,30). As a part of the largest nonstructural protein, the tracing back to and investigating the biological function of full-length Nsp3 is an important matter from a holistic aspect (30).
Unfortunately, due to technical limitations, we could not obtain the full length of Nsp3 or transfect it into host cells (63,64). Note that other previous studies also failed to express full-length Nsp3, and thus only the Nsp3-N was investigated. In a previous report, it was found that unlike discrete, punctate nuclear localizations found for Nsp1, Nsp5, Nsp9, and Nsp13, Nsp3-N was both cytoplasmic and rather diffusely found across the nucleus, indicating that Nsp3 may be involved in DNA transcription and genomic regulation (42). Our subcellular localization result for SUD2 core was consistent with Nsp3-N. In addition, we have found a nuclear localization sequence (KKAGGTTEMLAKALRKV) in the SUD2 domain (and hence also in Nsp3) (65); therefore, it can be inferred that the full Nsp3 may also have the capability to enter the nucleus. This deduction needs further experimental validation, and how the SUD2 core domain will perform in the milieu of full-length Nsp3 protein awaits further investigation.
In addition to the SUD2 domain, Nsp3 contains multiple modular protein domains, and these portions have also been investigated separately and display different roles during viral infection in host cells. For instance, the Mac1 domain removes ADP ribosylation posttranslational modifications (66), and the PL pro domain antagonizes MDA5mediated type I interferon signaling to suppress the initial immune response (67).
For the other complex components, apart from being the partner of SUD2 core , Nsp5 has also been previously identified as a main protease in the SARS-CoV-2 genome and is essential for cleaving the replicase polypeptides (pp1a and pp1ab) and the replication and transcription of SARS-CoV-2. Therefore, further research is needed to address how the interaction between SUD2 and Nsp5 influences the function of Nsp5 in this role. Despite these limitations, our current study provides new information that the SUD2 core -Nsp5-Bcl2G4 interaction leads to increased apoptosis of respiratory epithelial cells, which has been identified as a hallmark of severe COVID-19 disease. Based on these results, new therapies targeting SUD2, Nsp5, destabilizing G4, or their mutual interfaces can be envisioned.

MATERIALS AND METHODS
Plasmid construction. For the Y2H assay, the coding sequences (CDS) of all Nsps of SARS-CoV-2, except Nsp3, were synthesized by Tsingke commercial company, then cloned into the pGADT7 vector (Clontech catalog number 630442) by using NdeI and BamHI. The CDS of SUD2 core was cloned into pGBKT7 vector (Clontech catalog number 630443) by using EcoRI and SalI. For co-IP, RNA-seq, ChIP-qPCR, and apoptosis assays, SUD2 core was generated into a pcDNA3.1-3FLAG vector (Thermo Fisher Scientific catalog number V79020) and packed with lentiviruses by PackGene Co. The pcDNA3.1-SUD2 core -GFP plasmid was constructed by using EcoRI and BamHI, and pcDNA3.1-Nsp5-GFP was constructed by using BamHI and EcoRI for subcellular localization. For BiFC experiments, SUD2 core was fused to pBiFC-VN173 vector (Addgene catalog number 22010) containing the YFP N fragment, and Nsp5 was fused to pBiFC-VC155 vector (Addgene catalog number 22011), respectively. For the co-IP assay, SUD2 core and Nsp5 were fused into pcDNA3.1-SUD2 core -3FLAG and pcDNA3.1-Nsp5-GFP, respectively. For expression of recombinant SUD2 and Nsp5 in E. coli, CDS of SUD2 core was introduced into pET-28a(1) expression vector (Novagen catalog number 69866-3) by using BamHI and SalI and fused with His tag. Nsp5 was introduced into pGEX6P-1 expression vector (GE catalog number 28-9546-48) by using BamHI and EcoRI, which fused with GST tag. For the firefly luciferase assay, the DNA region located 2,000 bp upstream of the translational start site of the BclII gene, including the G4 sequences (pBcl2-WT or pBcl2-Mut), was introduced into the pNLCoI1 (luc2-P2A-NlucP/Hygro) vector (Promega catalog number N1461) to analyze the expression of BclII in human respiratory cells.
Cell culture and transfection. Normal human bronchial epithelial cells (NHBE) were purchased from Lonza, and human lung epithelial cells (A549 and H1299) were purchased from Shanghai Cell Bank, Chinese Academy of Sciences. NHBE cells were cultured in Dulbecco's modified Eagle's medium, and A549 and H1299 cells were cultured in RPMI 1640, containing 10% fetal bovine serum, 100 mg of streptomycin/mL, and 100 U of penicillin/mL. The cells were maintained in a chamber at 37°C with 5% CO 2 .
For cell transfections, the cells were cultured after 10 generations and then plated in 35-mm confocal dishes at about 200,000 cells per dish. One microgram amounts of plasmids of SUD2 core , Nsp5, or SUD2 core plus Nsp5 were added to the cells with Lipofectamine 3000 reagent according to the manufacturer's instructions.
Yeast two-hybrid assay. Y2H was performed following the instructions for the Yeastmaker yeast transformation system 2 (Clontech manual PT1172-1). Generally, AD-Nsps and BD-SUD2 core were cotransformed into the yeast strain Y2HGold using the polyethylene glycol-lithium acetate method (manual PT1172-1). All of the clones grew well on SD medium minus leucine and tryptophan (-LW). The positive clones were identified by the ability to grow on SD medium minus leucine, tryptophan, and histidine SUD2-Nsp5 Regulate BclII-Mediated Apoptosis Responses mBio (-LWH). The photographs were taken after 3 days. The primers that were used in this assay are listed in Table S1. For b-galactosidase activity assays, Y187 cells were grown to mid-log phase (optical density at 600 nm [OD 600 ] of 0.5). The cell pellets were resuspended in Z buffer (60 mM Na 2 HPO 4 Á7H 2 O, 40 mM NaH 2 PO 4 ÁH 2 O, 10 mM KCl). In order to permeabilize the cells, suspensions were subjected to 3 freeze and thaw cycles before adding o-nitrophenyl-b-D-galactopyranoside (ONPG). After incubation, the OD 420 of the samples was measured in a spectrophotometer, and b-galactosidase units were calculated according to the following equation: (1,000 Â OD 420 )/(t Â V Â OD 600 ), where t is the time (in minutes) to the appearance of yellow color after adding the ONPG, and V is the volume (in milliliters) of cell culture used.
Co-IP assay. The pcDNA3.1-Nsp5-GFP, pcDNA3.1-GFP empty vector, and pcDNA3.1-SUD2 core -3FLAG plasmids were transfected into human epithelial cells for the co-IP assay. First, Dynabeads-protein G was incubated with anti-FLAG antibody or anti-GFP antibody at 4°C for 1 h. After that, the total proteins of different cells were extracted and then put on ice with IP buffer (20 mM HEPES [pH 7.5], 200 mM NaCl, 10 mM MgCl 2 , 0.1% Triton X-100, 1 mM EDTA, 10% glycerol, 10 mM protease inhibitor cocktail, and 2 mM phenylmethylsulfonyl fluoride). The supernatants were incubated with antibody-protein G complexes at 4°C for 4 h. The beads were washed 5 times with IP buffer. Then, the samples were loaded on SDS-PAGE and detected with anti-FLAG antibody or anti-GFP antibody.
Subcellular localization and BiFC assay. For subcellular localization, the SUD2 core -GFP or Nsp5-GFP plasmids were transfected into NHBE cells, and the GFP signal was investigated after 24 h. Hoechst staining indicated the nuclei of epithelial cells. Photos were captured by a confocal fluorescence microscope (Stellaris; Leica, Germany). For the BiFC assay, enhanced YFP signaling (VENUS) was captured using the same instrument.
Protein expression and purification. The sequenced plasmids of pET28a (1)-SUD2 core , SUD2 core -Mut, and pGEX6P-1-Nsp5 were transformed into the E. coli expression strain BL21(DE3). For the protein expression procedure, 0.5 mM isopropyl-b-D-thiogalactoside was added to fresh E. coli cells and cultured overnight at 16°C. After sonification and centrifugation, cell pellets were collected in binding buffer (50 mM Tris-HCl [pH 8.0], 300 mM NaCl, and 5 mM imidazole), His-SUD2 core was eluted by 200 mM imidazole in a His-labeled nickel column, and Nsp5-GST proteins were eluted by 10 mM reduced glutathione in a GST-agarose resin. The eluted proteins were then added into an Amicon Ultra-15 ultrafiltration tube (Merck catalog number UFC901096) for concentration and desalting. Separation and purification were performed by HITRAP Q-affinity chromatography and Superdex 200 Increase 3.2/300 gel filtration chromatography (Cytiva). SDS-PAGE was used to detect the above proteins. The charge properties and polymerization degree of the proteins were uniform and the purity reached 95% for further study.
CD spectroscopy. The single-strand Bcl2G4, VEGFR2G4, and KRASG4 DNAs were synthesized by Sango Biotechnology in high-performance liquid chromatography level. The oligonucleotides were folded into G4s by the following procedure: 95°C heating in 50 mM Tris-HCl (pH 8.0), 100 mM KCl for 5 min, then slowly cooled to room temperature. The linear DNA was annealed without K 1 buffer. The oligonucleotides were then diluted to 1 mM, and 1.5 mL of the reaction solution was mixed in a 1-cm optical-path-length quartz colorimetric dish and then measured directly. The measurements recorded CD values between 220 nm and 300 nm at room temperature. Each CD spectrum curve represented the average of 10 measurements and was eventually smoothed by a Savitsky-Golay filter, using the CD instrument (Chirascan Plus, Applied Photophysics, UK). If an annealed oligonucleotide chain showed a trough at 240 nm and a peak at 260 nm, it could form a positively parallel G4 structure.
Biolayer interferometry assay. Binding kinetic analyses of SUD2 core -Nsp5 and SUD2 core -Bcl2G4 were performed by BLI at ForteBio Octet (Sartorius, Germany). For detection of the SUD2 core -Nsp5 interaction, the purified His-tagged SUD2 core was immobilized onto anti-His antibody biosensors (sensors were hydrated for 10 min before use) in a BLI buffer (50 mM Tris-HCl [pH 8.0], 100 mM KCl, 0.05% Tween 20, and 5% [vol/vol] glycerol). It was then incubated with Nsp5 protein as follows: baseline for 60 s, loading for 90 s, association for 180 s, and dissociation for 300 s. For detection of the SUD2 core -Bcl2G4 interaction, the G4 structure of biotin-labeled Bcl2G4 was immobilized onto the SA biosensors in the BLI buffer. Then, it was incubated with SUD2 core as follows: baseline for 60 s, loading for 90 s, association for 90 s, and dissociation for 150 s. The reference sensors without His-tagged SUD2 core or biotin-labeled G4 DNA served as background controls. Curves were fitted to a 1:1 interaction model, and the K D value was calculated using Octet data analysis studio software (Sartorius, Germany).
Luciferase reporter activity assay. The dual luciferase reporter assay was performed according to the manufacturer's instructions (Promega TM058). In brief, 80 mL of Dual-Glo reagent was added to each cell culture medium to lyse cells for 10 min, then the firefly luminescence was measured in a luminometer (Victor Nivo Alpha S, PerkinElmer, USA). After that, 80 mL of Dual-Glo Stop & Glo reagent was added to measure the Renilla luminescence. Relative response ratios were calculated from the normalized ratios.
EMSA. Purified SUD2 core (200 ng) was incubated with 10 ng Cy5-labeled G4 probe in EMSA buffer (20 mM  , and 0.05% [wt/vol] salmon sperm DNA) for 20 min at 4°C. A 10% nondenaturing polyacrylamide gel was prepared for electrophoresis. The FujiFilm Starion fla-9000 system was used to detect the fluorescent signal after electrophoresis, and the excitation light and emission light were 646 and 664 nm, respectively. The mobility of the probe was calculated.
G4 pulldown assay. G4 pulldown assays were performed as described previously (20). In brief, 59-biotin labeled DNAs (Table S1) were folded into G4s as follows: 95°C heating in 50 mM Tris-HCl (pH 8.0), 100 mM KCl for 5 min, then slow cooling to room temperature. Streptavidin magnetic beads were equilibrated with G4s in binding buffer (50 mM Tris-HCl [pH 8.0], 100 mM KCl, 0.1 mM EDTA, 1 mM DTT, and 0.05% Tween) for 1 h. Then, the purified SUD2 core was incubated with the G4-streptavidin magnetic bead complexes for 4 h at 4°C. The beads were washed 5 times before the samples were loaded for SDS-PAGE, and detection was with anti-His antibody (1:5,000; Genscript catalog number A00186).
Atomic modeling analysis. For three-dimensional (3D) structure predictions of proteins, the structures of SUD2 core and SUD2 core in complex with Nsp5 were predicted using the AlphaFold2 and AlphaFold2-multimer programs, respectively. The conformation with the highest score was used for subsequent molecular docking, dynamic simulation, or analysis.
For molecular docking, the structure of Bcl2G4 was obtained from the PDB database (PDB ID 2F8U). Subsequently, Bcl2G4 was docked to the just-predicted SUD2 core and SUD2 core -Nsp5 complexes using the ZDOCK online server (https://zdock.umassmed.edu/) to predict their binding mode.
For molecular dynamics simulation, GROMACS package (version 2021.03) was applied to run conventional MD simulations to investigate the changes in conformation of SUD2 alone and binding modes of SUD2 core -G4 and SUD2 core -Nsp5-G4 complexes. The force fields amber14sb and OL15 ff were employed to parameterize protein and G4, respectively. The TIP3P was used for the waters. The protein or protein-G4 complex were solvated in an octahedral water box, and then the charge of the system was neutralized by adding 0.150 M chloride and sodium ions. First, the steepest descent minimization method was used to minimize the energy of the system by 50,000 steps. In the next step, we restricted the position of heavy atoms to run both constant number of atoms, volume and temperature (NVT) equilibration and constant number of atoms, pressure and temperature (NPT) equilibration by 50,000 steps. The system temperature was maintained at 300 K, and the system pressure was maintained at 10 5 pa. Upon completion of the two equilibration phases, the system was considered well-equilibrated at the desired temperature and pressure. A 100-ns unrestrained simulation was carried out. Every 10 ps, the energy and coordinate system of the trajectory was saved. In the simulation trajectory, ChimeraX and PyMOL were used to map interaction patterns and animate kinetic trajectories. To study the conformational changes of SUD2 core , we performed PCA processing of molecular dynamics trajectories using the covar, anaeig, and sham commands that come with GROMACS to describe the magnitude of various conformational free energies of macromolecules.
Free energy calculations and residue decomposition. The MM-GBSA method has been widely adopted in the estimation of binding free energy in drug research. In our work, the MM-GBSA calculation was performed using the gmx_MMPBSA, a tool of GROMACS for MM-PB (GB) SA calculations. To understand the binding of protein and G4 at the molecular level, we used gmx_MMPBSA to decompose the free energy of binding to the contribution of each residue to the free energy of binding.
ChIP-qPCR assay. About 5 Â 10 7 well-grown cells were cross-linked by formaldehyde to a final concentration of 0.75% and were rotated gently at room temperature for 10 min. Glycine (125 mM) was added for 5 min at room temperature, and then cells were rinsed twice with 10 mL cold phosphate-buffered saline (PBS) and added to 5 mL of cold PBS. Cells were thoroughly scraped into a 50-mL tube and then centrifuged for 5 min at 4°C, 1,000 Â g. The pellets were resuspended in ChIP lysis buffer (50 mM HEPES-KOH [pH 7.5], 140 mM NaCl, 1 mM EDTA [pH 8.0], 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, and 1Â protease inhibitors) and incubated for 10 min on ice.
The samples were then sonicated using a BioruptorPico (Diagenode, Belgium) for 10 min with the following parameters: duty factor, 8%; intensity peak power, 120; cycles per burst, 200; bath temperature, 4°C. The DNAs were sheared to an average fragment size of 200 to 1,000 bp. After sonication, cell debris was pelleted by centrifugation for 10 min, 4°C, 8,000 Â g. Supernatants of chromatin were incubated with elution buffer (1% SDS and 100 mM NaHCO 3 ) plus NaCl and RNase A while shaking at 65°C overnight. Proteinase K (20 mg/mL) was added and incubated while shaking at 60°C for 1 h. DNAs were purified by using a PCR purification kit (QIAamp DNA Micro kit, Qiagen catalog number 56304). A 25-mg amount of purified DNA was diluted 1:10 for each sample (one sample for anti-FLAG antibody and one sample for the beads-only control) with RIPA buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 2 mM EDTA [pH 8], 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, and 1Â protease inhibitors). Fifty microliters of chromatin was removed to serve as input sample. Then, anti-FLAG antibody was added to all samples except the beadsonly control and rotated at 4°C for 1 h. Equal volumes of washed protein G beads, single-stranded herring sperm DNA, and BSA were added to a final concentration of 75 ng/mL of beads. BSA was also added to a final concentration of 0.1 mg/mL of beads. A 60-mL aliquot of blocked protein A/G beads was added to all samples and immunoprecipitated overnight with rotation at 4°C. Then, the immunoprecipitated samples were centrifuged for 1 min, 2,000 Â g, and the supernatants were removed.
RNA-sequencing. Total RNA was isolated and purified using TRIzol reagent, and the RNA integrity was assessed by using the Bioanalyzer 2100 system (Agilent, CA, USA) with a RNA integrity number (RIN) number of .7.0. Poly(A) RNA was purified from 1 mg of total RNA using Dynabeads Oligo(dT) 25-61005 (Thermo Fisher Scientific) for two rounds of purification. Then, the poly(A) RNA was fragmented into small pieces using the Magnesium RNA fragmentation module (NEB catalog number e6150) under 94°C for 5 to 7 min. Then, the cleaved RNA fragments were reverse-transcribed to create the cDNA by using SuperScript II reverse transcriptase (Thermo Fisher Scientific catalog number 1896649). These were then used to synthesize U-labeled second-stranded DNA with E. coli DNA polymerase I (NEB catalog number m0209), RNase H (NEB catalog number m0297), and dUTP solution (Thermo Fisher Scientific catalog number R0133). Single or dual index adapters were ligated to the fragments. After heat-labile UDG enzyme (NEB catalog number m0280) treatment of the U-labeled second-stranded DNAs, the ligated products were amplified with PCR as follows: initial denaturation at 95°C for 3 min; 8 cycles of denaturation at 98°C for 15 s, annealing at 60°C for 15 sec, and extension at 72°C for 30 sec; and then final extension at 72°C for 5 min. The average insert size for the final cDNA library was 300 6 50 bp. Next, 2 Â 150-bp paired-end sequencing (PE150) was performed on an Illumina Novaseq 6000 system (LC-Bio Technology Co., Ltd., Hangzhou, China) following the vendor's recommended protocol.
After removing the lower-quality bases and undetermined bases, HISAT2 software was used (hisat2-2.0.4) to map reads to the genome. The mapped reads of each sample were assembled using StringTie with default parameters. All transcriptomes from all samples were merged to reconstruct a comprehensive transcriptome using gffcompare software. After the final transcriptome was generated, StringTie and ballgown were used to estimate the expression levels of all transcripts and determine expression levels for mRNAs by calculating the fragments per kilo base per million mapped reads (FPKM). The differentially expressed mRNAs were selected with a fold change of .2 or fold change of ,0.5 and a P value of ,0.05 by R package edgeR or DESeq2, and then GO enrichment and KEGG enrichment were performed to analyze the differentially expressed mRNAs.
Quantitative real-time PCR analysis of gene expression. Total RNA was extracted from 10th-generation cells using TRIzol reagent (Thermo Fisher Scientific catalog number 15596026), and reverse transcription was performed using the HiScript II 1st Strand cDNA synthesis kit (Vazyme catalog number R212-01). All of the qPCR experiments for the analysis of differential gene expression were performed using the QuantStudio 3D instrument and the reagent PowerUp SYBR green (Applied Biosystems). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an internal control. All of the phenotypic assays were performed with three technical replicates and three independent biological repetitions with the same results. The primers that were used in this assay are listed in Table S1.
Apoptosis rate measurement. Different plasmids containing SUD2 core and Nsp5 were transfected into NHBE cells. Trypsin-digested cells (%1 Â 10 5 ) cells were washed with PBS 3 times and were centrifuged for 1 min, at 2,000 Â g. The pelleted cells were resuspended with staining buffer (4 mL Annexin Venhanced GFP and 4 mL PI; CellorLab catalog number CX005L) and were incubated at room temperature for 10 min without light. The apoptosis rates of different cells were then immediately detected by flow cytometry, which showed green fluorescence in apoptotic cells, red and green fluorescence in dead cells, and almost no fluorescence in living cells.
Cell proliferation calculation. Cell suspensions were inoculated into 96-well plates. The culture plates were placed in an incubator for preculture for 24 h. Ten-microliter aliquots of CCK-8 solutions were added to each well and incubated for 1 h. The absorbance at 450 nm was measured with a microplate reader (Victor Nivo Alpha S, PerkinElmer, USA). The intervals of testing were 0, 1, 2, 3, 4, and 5 days.
Quantification and statistical analysis. The number of experiments and replicates are indicated in individual figure legends. Data were processed and visualized using Origin2021. All quantified data are represented as means 6 standard errors of the means (SEM), as indicated, and quantification details are available in the figure legends. Western blotting band intensities were quantified using Image J.
Data availability. The data reported in this paper will be shared by the corresponding author upon request. Any source file required to reanalyze the AlphaFold modeling structure reported in this paper is available from the corresponding author upon request.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.