Here, we investigate the kinetics of in vitro processing of nsp7-11 using native mass spectrometry (MS). Native MS is a highly sensitive method allowing simultaneous monitoring of cleavage products, making it very versatile to study dynamic processes and heterogeneous samples33,34. In addition, native MS preserves protein-protein interactions, enabling detection of protein complexes formed by the processing products35. By acquiring mass spectra at different time points, we monitored increasing and decreasing nsps, intermediate products and protein complexes simultaneously. Based on the change in signal ratio, we established an approach to extract kinetic rate constants k in a multi-cleavage reaction. Thereby, we determined k of Mpro at CS7/8, CS8/9, CS9/10 and CS10/11 for human pathogenic SARS-CoV-2, SARS-CoV, HCoV-229E and MERS-CoV. In additional experiments, we studied processed and unprocessed polyprotein interaction with the SARS-CoV-2 methyltransferase nsp16, identifying critical and non-critical cleavage sites and a chimeric complex between SARS-CoV-2 and MERS-CoV proteins. Analysis of primary sequences, structural models, and experimentally determined rate constants revealed molecular mechanisms of CoV processing.
To investigate SARS-CoV-2 polyprotein processing, we expressed recombinant nsp7-11 polyprotein and cleaved it with SARS-CoV-2 M. Initial experiments used substrates with N- or C-terminal His-tags (nsp7-11N and nsp7-11C, respectively). However, the potential effects of these tags on M cleavage activity could not be excluded. Therefore, we also employed tag-free nsp7-11 substrates using an N-terminal His-SUMO-tag strategy to compare the four hCoVs. This tag was removed using SUMO-specific protease ULP-1, generating nsp7-11 with native termini (Fig. 1A).
We used native MS to analyze SARS-CoV-2 M cleavage reactions with nsp7-11 substrates from four hCoVs: SARS-CoV-2, SARS-CoV, HCoV-229E and MERS-CoV. Native MS confirmed that all four conserved cleavage sites (CS7/8, CS8/9, CS9/10 and CS10/11) were addressed in each of them. Despite the wide mass range spanning from the smallest product nsp11 (1325.654 Da ± 0.001 Da) to the heterotetrameric complex of nsp7 + 8 (62,244 Da ± 2 Da), we obtained precise masses from tagged and untagged substrates for monomeric nsps, intermediates and polyproteins (Figs. 1C and S1, Supplementary Table S1-S5). The processing reaction was additionally verified by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) (Figs. 1B and S2) as done in previous studies.
For detailed investigation of M mediated polyprotein processing, we followed two approaches to extract fast and slow cleavage rates (Fig. 1D). For fast rates, we used a continuous monitoring approach, where the ongoing enzyme reaction was performed "in-capillary" at 27 °C. For slow rates at 0 °C (on ice), we used a discontinuous approach, with processing reactions performed in a test tube and sampled discontinuously over time. Using a Python-based script, we extracted rate constants k of CS7/8, CS8/9, CS9/10 and CS10/11. Finally, we investigated how processing influences CoV core enzyme complex formation with recombinantly produced SARS-CoV-2 nsp16 and the cleaved as well as uncleaved polyprotein nsp7-11 (Fig. 1E).
Native MS enables monitoring of reaction products over time, though capturing sufficient data points from rapid reactions presents a challenge. To overcome this limitation and observe early M-mediated processing reactions, we implemented continuous "in-capillary" analysis using nsp7-11C and nsp7-11N substrate polyproteins. The capillary housing temperature was 27 °C, which approximates physiologically relevant viral propagation temperatures. This approach also ensures identical reaction start points, which is essential for extracting high-quality kinetic data. We monitored a decrease of intensity of the polyproteins, nsp7-11C (60,950 Da ± 4 Da) or nsp7-11N (61,085 Da ± 1 Da), and an increase of the cleavage intermediates nsp7-10 (dominant intermediate and plateauing in nsp7-11N after 5-10 min), nsp7-9, nsp7-8, nsp9-11, nsp9-10, and nsp10-11 within the first 30 min of the reaction (Figs. 2A and S3, Supplementary Table S1). This enabled us to monitor the fastest cleavage reaction of CS10/11 in detail (Fig. 2B).
To determine the rate constant k for the CS10/11 cleavage reaction, we developed a custom data processing approach. Signal intensities (area under the curve, AUC) of substrate and intermediate products were plotted over time. Signals of substrate and observed intermediates containing the intact cleavage site (nsp9-11, nsp10-11) were summed and plotted logarithmically. The initial linear slope shows that first order kinetics can be applied and hence k at CS10/11 at 27 °C is calculated from the slope of the linear region (Fig. 2C). Due to the low intensities below 1% of nsp9-11 and nsp10-11, the rate constant is dominated by the decrease of nsp7-11C or nsp7-11N (Figs. 2A and S3). Therefore, in the first 30 min, the main conversion is the cleavage of nsp11 with CS10/11 in both substrates. However, CS10/11 in nsp7-11 N (k = 0.240 min ± 0.005 min) was cleaved three times faster than in nsp7-11 C (k = 0.081 min ± 0.001 min), indicating that the His-tag decreases cleavage site efficiency in the latter. Linearity of the slope for CS10/11 breaks down when signal intensities decreased to 10% of their initial value, which occurred after 15 min for nsp7-11N and ~ 30 min for nsp7-11C. This deviation from linearity likely resulted from increased substrate depletion at 27 °C, whereas at 0 °C, similar substrate depletion would only be expected after longer incubation times of several hours.
The rate constants for the other cleavage sites (CS9/10, CS8/9, CS7/8) could not be determined accordingly because their corresponding products remained of low staggering intensity (14-48-fold slower than CS10/11). However, nsp11 or the His-tag do not influence either the order of processing or the conversion rates of the other CS (Supplementary Fig. S4). While this continuous processing approach effectively captures early reaction kinetics, it faces limitations for extended monitoring most likely due to limited protein stability at 27 °C and acidification within the capillary over time. Determining k for the other cleavage sites required experiments spanning longer time periods, which could only be conducted at a lower temperature. Therefore, we followed up with the discontinuous approach.
A more complete picture emerges using such a discontinuous approach. To further avoid any influence of the His-tag, untagged nsp7-11 with native termini was employed. We monitored M-mediated processing of nsp7-11 substrates from four hCoVs. Reactions were conducted on ice to ensure protein stability, and native mass spectra were acquired in triplicate at discrete time points between 30 min and 500 min. For SARS-CoV-2 nsp7-11 processing, we monitored a dynamic succession of cleavage products. Expectedly, the intensity of the cleavage intermediate nsp7-10 peaks first. It quickly gives rise to subsequent cleavage intermediates nsp7-8, nsp7-9, and nsp9-10. At later time points, when the substrate is depleted also lower populated species like nsp9-11 become visible (Fig. 3A). We validated the reaction after 24 h, showing mainly monomeric nsps, indicative of nearly complete cleavage (Supplementary Fig. S5). Both tagged and untagged substrates produced corresponding products and revealed similar cleavage order (Supplementary Fig. S4). However, all kinetic parameters were determined exclusively from experiments using untagged substrates to avoid any potential tag-induced artifacts.
To extract kinetic rate constants, we again applied our custom data processing to simplify the multiple cleavage reactions to first-order kinetics. Peak areas of intermediate species were summed based on their intact cleavage sites (Fig. 3B). These summed values were fitted with an exponential decay (Fig. 3C). Plotting the y-axis on a logarithmic scale allows verification of first-order kinetics through linear fits (Fig. 4A). Data of the other three hCoVs (Supplementary Figs. S5-S8) was processed (Supplementary Fig. S9) and fitted accordingly (Fig. 4).
During the experiment, we checked with collision-induced dissociation whether peaks originated from the intermediate product nsp7-8 or nsp7 + 8 heterodimer. The SARS-CoV-2 precursor ion m/z 3110 did not dissociate into nsp7 and nsp8, meaning < 1% product ion signal intensity compared to precursor ion intensity originates from the heterodimer. Thus, nsp7 and nsp8 are still too low abundant for complexation and only become significantly populated between 6 h and 24 h.
In SARS-CoV, the dominant products next to nsp7-10 were nsp7-9 and nsp7-8 early in the reaction, nsp9-11 also becomes visible later on (Supplementary Figs. S6 and S9). The observed relative intensities of intermediate cleavage products are largely similar to SARS-CoV-2 nsp7-11, suggesting a similar processing pattern from C- to N-terminus, consistent with previous studies. This similarity is reflected in the comparable order of rate constants, although SARS-CoV showed slower cleavage at CS7/8 and faster cleavage at CS9/10 compared to SARS-CoV-2 (Fig. 4). In HCoV-229E, the observed dominant early intermediate was nsp7-9. While nsp7-8 and nsp10-11 are observed at later time points, they never make up a relevant share of the intensity (Supplementary Figs. S7 and S9). While some early observed products from HCoV-229E nsp7-11 resemble those observed in SARS-CoV and -2, nsp7-10 is essentially absent. It is tempting to state that CS10/11 is hence not addressed first. However, the lack of populated nsp10-11 in the early phases of the reaction rather suggests that CS9/10 and CS10/11 are processed at similar rates, which is corroborated by the linear fits (Fig. 4). MERS-CoV exhibited the most distinct intermediate distribution, with nsp7-8 and nsp9-11 emerging as dominant species from the onset throughout the reaction over 500 min (Supplementary Figs. S8 and S9). This unique intermediate pattern effectively results in MERS-CoV nsp7-11 being processed 'in half' at CS8/9. The data suggests early and efficient cleavage at the CS8/9 site, while CS9/10, CS10/11 and CS7/8 all showed similarly retarded cleavage rates, a pattern distinct from the other three hCoVs (Fig. 4). Indeed, the rate constant for CS8/9 cleavage in MERS-CoV (k) is approximately twice as fast as any cleavage site rate constant in the other hCoVs. The other MERS-CoV cleavage sites are not processed slowest among hCoVs tested, as the rate constants of cleavage sites CS7/8 (k) in the other hCoVs are two to thirty times slower.
Native MS revealed distinct processing patterns across the four hCoVs. SARS-CoV and SARS-CoV-2 identified CS10/11 as the dominant early cleavage site, while HCoV-229E and MERS-CoV exhibited different patterns. Despite identical sequences at CS7/8 in both SARS species, their rate constant differed by an order of magnitude, suggesting structural rather than sequence effects on cleavage efficiency. The core residues P2 and P1 (L and Q) are conserved within CS7/8, 8/9 and 9/10 across all species, yet different rates were observed, particularly at CS8/9 between HCoV-229E and MERS-CoV, indicating that flanking sequences or structure influence processing. For CS9/10, where the P4 to P3' positions are identical across all species, the varying cleavage rates likely result from structural differences. CS10/11 showed the greatest sequence variability, especially at P2, with MERS-CoV containing proline and HCoV-229E containing isoleucine, potentially explaining their slower kinetics. These results reveal that C- to N-terminal processing of nsp7-11 is not conserved among SARS-CoV, SARS-CoV-2, HCoV-229E, and MERS-CoV, though delayed CS7/8 cleavage appears to be a common feature. The non-essential nature of fast CS10/11 cleavage raises the question whether uncleaved intermediates can still function as cofactors in complex formation.
Formation of the RTC requires processing, but whether the RTC incorporates exclusively mature nsps or also immature processing intermediates remains unknown. The functional RTC requires association of several proteins, including nsp10 and nsp16. We hypothesize that nsp16 + 10 complex formation similarly depends on polyprotein processing, specifically the cleavage of CS9/10 and, to a lesser extent, CS10/11 to release nsp10 from the polyprotein. To test this hypothesis, we performed protein-protein interactions using native MS (Fig. 5 and S10). Initially, we tested binding of uncleaved SARS-CoV-2 or MERS-CoV nsp7-11 (59674 Da ± 3 Da and 59658 Da ± 4 Da, respectively) and SARS-CoV-2 nsp16 (33323.27 Da ± 0.14 Da). For SARS-CoV-2 nsp7-11, predominant signal intensities originated from nsp16 monomer, nsp7-11 monomers and dimers, and low intensities for nsp7-11 + nsp16 complex (~ 2%). Increased levels (< 5%) of SARS-CoV-2 nsp16 complexed with MERS-CoV nsp7-11 were observed despite being a chimeric complex. The complexes were validated using collision-induced dissociation (Supplementary Fig. S11), which notably revealed the Zn binding of nsp10.
We then initiated processing of nsp7-11 by adding M and incubating with nsp16 overnight. Native mass spectra were distinct for SARS-CoV-2 and MERS-CoV nsp7-11, although both showed high levels of complexation between SARS-CoV-2 nsp16 and nsp10, suggesting specific binding. A heterodimeric complex containing mature nsp10 (nsp16 + 10, 48,236 Da ± 1 Da) was apparent in SARS-CoV-2 (Supplementary Table S6). Although k in MERS-CoV would suggest complete processing overnight, we observed more than 10% nsp10-11 intermediates and more than 40% heterodimeric nsp16 + 10-11 as protein complex. Strikingly, no nsp16 with mature nsp10 was observed, indicating that nsp16 binds to nsp7-11 or nsp10-containing intermediates, which potentially protects CS10/11 from further cleavage in the complex. Given the moderate sequence similarity between MERS-CoV and SARS-CoV-2 (70% for nsp10 and 80% for nsp16), the formation of chimeric nsp16 + 10 complexes represents an intriguing finding.
To mimic the viral ratio of pp1a to pp1ab, we tested increased proportions of cleaved SARS-CoV-2 nsp7-11 to nsp16, observing similarly increased proportions of nsp16 + 10 complex formation (Supplementary Fig. S12). These experiments yielded a complex dissociation constant K of 8 µM ± 1 µM. In comparison, titration measurements of purified recombinant nsp10 and nsp16 yielded a lower K of 1.4 µM. The higher K value observed here may result from the complex mixture of polyprotein cleavage products in our experimental system, which could lead to signal suppression for the complex. In summary, nsp16 showed weak binding to immature nsp10 within the polyprotein but strong binding to mostly mature nsp10. Our results indicate complex formation requires N-terminal CS9/10 cleavage but not necessarily C-terminal CS10/11 cleavage. Available crystal structures of nsp16/10 cannot explain this cleavage site preference, as both nsp10 termini are distant from the nsp16 binding site (Fig. 5C). We conclude that while complete processing of nsp7-11 is not essential, it greatly enhances CoV methyltransferase nsp16 + 10 complex formation.