Correlation of HX with crystal structures of HRV14 capsid proteins
Correlation of HX with capsid protein structure may be illustrated by Figure 5, where the amino acid sequences of VP1–4 are presented along with their secondary structures. The rectangles represent identified peptic fragments, with the color codes indicating their HX rates. Fragments with deuterium levels >80% in 12 sec are colored red, indicating very fast exchange; those in the remaining fragments with >70% deuterium in 180 sec are colored pink, indicating fast exchange. Slow exchange regions, colored green, are defined as regions with <30% deuterium in 2.5 h. Very slow exchange regions, colored blue, are those with <20% deuterium in 30 h. Peptides that did not fit into any of the above categories were colored yellow, indicating exchange with intermediate rates.
HX results for VP1 (Fig. 5A) show clearly that highly dynamic regions undergo rapid exchange. For instance, exchange in peptide segments 1–6, 9–12, 82–98, 142– 146, and 277–289 was rapid, indicating a relatively large degree of structural flexibility, consistent with the crystal structure of HRV14 capsid showing that these peptides are mostly located in the extended loops. In particular, one fast exchanging peptide segment including residues 82–98 is composed of the BC loop located in the protrusion areas of the exterior surface of the capsid and contains immunogenic sites, suggesting structural flexibility of antibody binding sites. In addition, the N-terminal residues 1–6 that were not observed in the crystal structure were found exchanging with fast rates, consistent with high disorder. Limited proteolysis of HRV14 with trypsin showed fast cleavage of VP1 peptides 1–8 and 1–13, suggesting transient exposure of the N terminus to the exterior surface of virus (Lewis et al. 1998). This highly dynamic nature of the VP1 N terminus was clearly consistent with our HX results. The C terminus of VP1 (segment 277–289) was also found with fast exchange rates, suggesting structural freedom, which agreed with the crystal structure model, indicating that this segment was a flexible loop exposed to the exterior surface of the virus.
In contrast with the fast exchange regions, VP1 segments of 110–124 and 163–177 (Fig. 5A, blue) were found with <20% deuterium, even after 30 h, indicating very slow HX. Two α-helices, αA and αB, were located in these regions, which essentially had little deuterium exchanged in. The slow exchange nature suggests that the two α-helices are stable, rigid structures in the capsid. Slow exchange was also found in the segments including residues 99–102, 178–200, and 219–225, most of which are β-strands (CFGH), consistent with general stability of the β-barrel fold in VP1. Among these strands, βD, βF, and βG were found having slower exchange rates than βC, βE, and βH, indicating some structural or dynamic difference of these β-strands. The former strands (DFG) make up the bottom of the β-barrel pocket, where the antiviral drugs bind, while the latter β-strands (CEH) are located close to the pocket entrance. The exchange results suggested relatively more flexibility for the β-strands at the entrance of the pocket than those at the bottom. The higher exchange rates may also result from more solvent exposure of the outside strands. This structural feature allows various antiviral agents to enter the same binding pocket with less structural hindrance. Intermediate exchange was mostly found in the regions that were composed primarily of the loops 13–60 and 71–81 and the loops between the β-strands that form the β-barrel fold. Our exchange results indicate partial flexibility in the VP1 β-barrel, which is important for HRV capsid to meet the need of structural rearrangement upon binding with antiviral agents or pocket factors (Smith et al. 1986; Arnold and Rossmann 1990).
The general features of the VP2 exchange kinetics (Fig. 5B) were similar with those of VP1. The disordered N terminus (residues 1–8, shown in pink) was found to exchange rapidly. Other fast-exchange segments included the C-terminal segment 252–262, which is on the exterior surface of the capsid. About 85%of amide linkages in the segment 156–168 were completely deuterated in the shortest time period (12 sec), indicating very fast exchange (red) and much structural flexibility. This region is located in a long exterior puff loop of VP2 containing immunogenic sites of NIm-II. The majority of the β-barrel fold in VP2 (residues 80–92, 98–112, 199–211, and 221–238) was found to have slow exchange (in green). Slow exchange regions also included residues 16–27 that form two β-strands (A1 and A2), consistent with their role in stabilizing the capsid. Strands β-A1 and -A2 run antiparallel to form the protein–protein interface of capsid pentamers. As was found for the two helices in VP1, the helix αB exhibited extremely slow exchange, indicating high stability for this region.
As shown in Figure 5C, VP3 was observed having fast exchange rates at exposed loop regions, including residues 53–66, 74–84, and 222–236, at the C terminus (in red and pink). Although some regions of the β-barrel were not detected (uncolored regions), it is largely a stable fold with slow exchange, particularly in the segment 117–129 and 211–221 (in green or blue), similar to VP1 and VP2. Exchange in residues 97–103 in the helix αA was very slow, a common feature of all three VPs. However, compared with VP1 and VP2, the N terminus of VP3 (segment 1–7, in yellow) had much slower exchange, indicating more structural order. This result is consistent with the crystal structure of HRV14 capsid, which shows a β-strand at the N terminus of VP3 (residues 1–7), as well as five strands from five subunits of VP3 forming a unique β-cylinder structure to stabilize the capsid around the fivefold axis (Arnold and Rossmann 1990). This β-cylindrical structure is conserved in many rhinoviruses and plays a critical role in viral disassembly.
VP4 of HRV14 was shown by X-ray diffraction to be an extended polypeptide lining the interior surface of the capsid (Rossmann et al. 1985). With little secondary structure, VP4 is expected to have fast exchange rates at most of its solvent accessible amide linkages, assuming that there is no significant hindrance for D2O to penetrate inside the virus. The present HXMS results for VP4 support this prediction. As shown clearly in Figures 4D and 5D, most fragments of VP4, including the N and C termini, had deuterium levels of >70%–90% in only 2.5 min, which is an indication of fast exchange. The overall fast exchange nature of internal protein VP4 indicates that the interior of the virus is solvent accessible, possibly due to the dynamic nature of the viral capsid. A recent finding that alkylating agents can permeate the protein capsid of a naked virus without affecting the particle integrity also supports this idea (Broo et al. 2001).
In contrast with fast exchange in most segments of VP4, deuterium exchange was found slower in two regions. One is segment 46–61, indicated by the gradually increased deuterium level from 27% in 12 sec to 88% in 30 h (Fig. 4D). This slower exchange may be explained by the presence of a short α-helix (residues 49–54) in this segment and hydrogen bonds formed between VP4 residues 53, 55, and 57, and residues from VP1 and VP2 (Arnold and Rossmann 1990). Slower exchange was also found in segment 27–31, which may indicate presence of unreported interchain hydrogen bonds or other unknown factors that decrease the exchange rates. The general consistency of our HX results with the fine structural features of HRV14 capsid proteins found in crystallographic studies demonstrates the sensitivity of HX rates as a probe for protein structure and dynamics.
Correlation of HX with B-factors and solvent accessibility
Thermal parameters, or B-factors, of individual atoms in protein determined by crystallography reflect atomic vibration or disorder and may be used to assess protein dynamics in solution. Since hydrogen atoms are usually invisible in protein crystal structures, deuterium levels in the HRV14 capsid are plotted with B-factors of amide nitrogens as well as with their solvent accessible areas to see whether there is general agreement between HX and crystallographic models of HRV14 capsid (Fig. 6). Results presented in Figure 6, A–C, show general correlation between deuterium levels in VP1 and the solvent accessibility of the amide nitrogens and their B-factors. It was found that higher deuterium levels generally corresponded with higher solvent accessible areas and B-factors. For instance, the amide nitrogens in VP1 segment 82–98 that contain the BC loop have an average B-factor of 28.9 Å2 and an average solvent accessible area of 3.8 Å2, which are higher than those in most other fragments. Accordingly, a high deuterium level (∼72% in 12 sec) was found in this segment, consistent with its higher structural flexibility and solvent accessibility. Meanwhile, the fragment 110–118 containing the αA helix has much lower average B-factors (11.7 Å2) and smaller average solvent accessible areas (0.6 Å2) for amide nitrogens, consistent with very slow exchange (∼6% in 12 sec). Note that a few regions seem not to have good correlation, such as N-terminal fragments including residues 35–43, 44–50, 53–60, and 71–81. These segments have relatively low B-factors (10–15 Å2) of amide nitrogens (Fig. 6). However, ∼40% of their amide linkages were found deuterated in 12 sec, indicating intermediate exchange rates. This rather high level of exchange may be due to the high solvent accessibility of some of these amide linkages.
Protein dynamics at HRV14 capsid protein–protein interfaces
The HRV14 capsid is composed of 240 protein subunits associated together by extensive protein–protein interactions. Examining HX rates at protein–protein interfaces in the viral capsid led to information on the structure and dynamics in these regions, which are of critical importance in understanding capsid assembly and disassembly.
A cartoon diagram of the crystal structure of a protomer of HRV14 composed of VP1–4 is presented in Figure 7, A and B. Note that different color schemes are used in these figures. In Figure 7A, VP1–4 are colored blue, green, yellow, and red, respectively, illustrating the different locations of the four capsid proteins in the protomer. The color coding in Figure 7B indicates different exchange rates in various regions of the capsid protomer. Three α-helices, αA from VP1, αB from VP2, and αA from VP3, were found clustering in the protomer interface to form a triangle, with each helix at an apex of the triangle. As shown in Figure 7B, the three helices, colored blue, had very slow exchange rates (with <15% deuterium level in 30 h, as discussed previously), indicating high stability of the protomeric protein–protein interface. It is proposed that these stable helical structures may play an important role in the formation of the 6S protomer, an early product of HRV assembly.
A view of the interior surface of the viral capsid pentamer is presented in Figure 7, C and D. Similarly, different color codes are used indicating different locations of VP1–4 (Fig. 7C) and exchange kinetics in different regions of viral capsid (Fig. 7D). As shown in Figure 7C, five copies of VP1 are located in the center of pentamer circling around the fivefold axis. The N-terminal arm of VP1 extends out under VP3, with the N terminus winding back toward the center of the pentamer. The N-terminal residues 1–16 are disordered and not shown. Next to VP1 are VP2 and VP3, which form the five edges of the pentamer. Note that the N-terminal arms of VP3 curve through VP1 and cluster around the fivefold axis forming a β-tube structure. As the least folded polypeptide among the four proteins, VP4 extends under VP1–3 with its C terminus heading away from the fivefold axis. The N terminus of VP4, including residues 1–28, is disordered in the crystal structure of HRV14 and not shown.
It is clearly shown in Figure 7D that the capsid pentamer has quite different HX rates in different regions, suggesting unique capsid dynamics in these regions. Slow exchange regions (deuterium level <30% in 2.5 h, colored green) included the fringes of the pentamer, where VP2 subunits are located, indicating that these regions have generally stable conformations. Very slow exchange regions were identified mostly in the protomer interface (in blue), as observed in Figure 7B, where residues form stable and rigid structures (the three α-helices) to stabilize the whole capsid. Very slow exchange was also observed in a few β-strands in VP2 at the edges of the pentamer where the pentamer contacts another pentamer, indicating the highly stable nature of such contacts.
The pentamer interface around the fivefold axis (Fig. 7E) is composed of five copies of the N terminus of VP3 (in yellow), VP1 loops (flexible, in yellow and pink), and the VP4 N terminus (flexible or disordered, in pink; residues 1–28 missing from the crystal structure). These regions exhibited intermediate and fast exchange, indicated by their color codes. Although the pentamer interface is stabilized by the β-cylinder structure, finding intermediate exchange rates indicates that this region is partially flexible. In addition, the N terminus and surface loops of VP1, and the N-terminal arms of VP4 exhibited much faster exchange (in pink and red), suggesting more flexibility than the β-cylinder structure. These results are consistent with the proposed uncoating events occurring at this interface (Kolatkar et al. 1999), in which cellular receptor binding with VP1 triggers a hinge-type movement of VP1 away from the fivefold axis, thus opening a channel for the release of VP4 and RNA release. The partial flexibility at the protein–protein interface around the fivefold axis allows such channel opening. More recent studies of HRV uncoating by cryoelectron microscopy suggested different dynamics for the N terminus of VP3 (Hewat et al. 2002; Xing et al. 2003; Hewat and Blaas 2004). An unconcerted movement of the VP3 N terminus was proposed for HRV14 to allow at least one β-cylinder channel to open at a time for RNA release (Hewat and Blaas 2004). Comparing exchange rates in the N-terminal region of VP3 in different HRVs and monitoring the changes of exchange profiles during uncoating may provide additional information to understand the role of capsid dynamics in HRV uncoating.
In conclusion, HXMS was shown as a powerful technique in characterizing the structure and dynamics of HRV14 capsid. The viral capsid was labeled intact with deuterium at neutral pH for a large range of exposure times, which allows for detection of a wide range of amide HX rates in the capsid. Local exchange profiles for each of the four capsid proteins were established through protease fragmentation of HRV14 capsid at minimal exchange conditions. The exchange results obtained from the in-solution experiments were generally consistent with the crystal structure of HRV14 capsid and could be correlated with structural parameters such as B-factors and solvent accessibility. In addition, dynamics in the protein–protein interfaces of the capsid protomer and pentamer were probed based on exchange results. The three α-helices of VP1–3 clustering at the protomeric interface were found in extremely slow exchange, suggesting high stability of this interface. Strong pentamer–pentamer interactions were indicated by slow exchange in the regions of pentamer edges. In addition, it was noted that the interface around the fivefold axis exhibited partial flexibility, consistent with its role in viral disassembly. As the first HX study of HRV, these results give additional information regarding viral capsid dynamics and may provide further insights for the relationship between the structure and function of the HRV14 capsid. Our results also suggest the potential applications of HXMS in studying structural changes in the HRV capsid induced by a wide range of events, such as binding with cellular receptors or antiviral agents. In such cases, HX rates in the native capsid would be compared with exchange rates in the capsid bound with receptors or drug molecules.