Von Willebrand factor A1 domain affinity for GPIbα and stability are differentially regulated by its O-glycosylated N-linker and C-linker

Hemostasis in the arterial circulation is mediated by binding of the A1 domain of the ultralong protein von Willebrand factor to GPIbα on platelets to form a platelet plug. A1 is activated by tensile force on VWF concatemers imparted by hydrodynamic drag force. The A1 core is protected from force-induced unfolding by a long-range disulfide that links cysteines near its N and C-termini. The O-glycosylated linkers between A1 and its neighboring domains, which transmit tensile force to A1, are reported to regulate A1 activation for binding to GPIb, but the mechanism is controversial and incompletely defined. Here, we study how these linkers, and their polypeptide and O-glycan moieties, regulate A1 affinity by measuring affinity, kinetics, thermodynamics, hydrogen deuterium exchange (HDX), and unfolding by temperature and urea. The N-linker lowers A1 affinity 40-fold with a stronger contribution from its O-glycan than polypeptide moiety. The N-linker also decreases HDX in specific regions of A1 and increases thermal stability and the energy gap between its native state and an intermediate state, which is observed in urea-induced unfolding. The C-linker also decreases affinity of A1 for GPIbα, but in contrast to the N-linker, has no significant effect on HDX or A1 stability. Among different models for A1 activation, our data are consistent with the model that the intermediate state has high affinity for GPIbα, which is induced by tensile force physiologically and regulated allosterically by the N-linker. Impact Statement Both the polypeptide and attached O-glycans N-terminal to the A1 domain in von Willebrand factor lower its affinity for its ligand GPIbα on platelets, its stability, and structural dynamics and decrease population of a high-affinity, intermediate state in unfolding. Key points Ligand-binding affinity of A1 and the stability of its native state is regulated by an N-terminal interdomain, O-glycosylated linker

• Ligand-binding affinity of A1 and the stability of its native state is regulated by an N-terminal inter- von Willebrand factor to GPIbα on platelets to form a platelet plug. A1 is activated by tensile force on VWF 27 concatemers imparted by hydrodynamic drag force. The A1 core is protected from force-induced unfolding by a 28 long-range disulfide that links cysteines near its N and C-termini. The O-glycosylated linkers between A1 and its 29 neighboring domains, which transmit tensile force to A1, are reported to regulate A1 activation for binding to 30 GPIb, but the mechanism is controversial and incompletely defined. Here, we study how these linkers, and their 31 polypeptide and O-glycan moieties, regulate A1 affinity by measuring affinity, kinetics, thermodynamics, 32 hydrogen deuterium exchange (HDX), and unfolding by temperature and urea. The N-linker lowers A1 affinity 33 40-fold with a stronger contribution from its O-glycan than polypeptide moiety. The N-linker also decreases HDX 34 in specific regions of A1 and increases thermal stability and the energy gap between its native state and an 35 intermediate state, which is observed in urea-induced unfolding. The C-linker also decreases affinity of A1 for 36 GPIbα, but in contrast to the N-linker, has no significant effect on HDX or A1 stability. Among different models 37 for A1 activation, our data are consistent with the model that the intermediate state has high affinity for GPIbα, 38 which is induced by tensile force physiologically and regulated allosterically by the N-linker.

40
The ultra-long length of the blood plasma protein 41 von Willebrand factor (VWF) enables its activation by Monomers are connected head-to-head and tail-to-tail. B) When exposed to elongational flow and tethered (black triangle) on a vessel wall, VWF concatemers extend to a linear shape. At higher elongational flow, which exerts higher mechanical tension along the length of the concatemer (highest at the tether point and zero at the downstream end) A1 undergoes transition from a low to a high affinity state as a consequence of tensile forces transmitted to it through its linkers. The disulfide bond between Cys1272 and Cys1458 is indicated with red S-S. C) A1 protein constructs studied here that differ in length of linkers and were expressed in E coli (not glycosylated) or in mammalian cells (O-glycosylated) (lollipops). The long-range disulfide bond is schematized in red. D) A1 constructs were subjected to SDS-PAGE and stained with Coomassie Blue.

84
Here, to understand how binding of the VWF A1 domain to platelet GPIbα is regulated by its N-and C-85 terminal mucin-like linkers, we have made fundamental measurements of the effects of both the glycan and 86 polypeptide moieties of these linkers on stability, thermodynamics, and ligand-binding affinity and kinetics of A1.

87
Mucin-like regions in proteins have unique characteristics distinct from both intrinsically disordered polypeptide 88 segments and folded domains. The bulky O-linked glycans which include sialic acid that are attached to threonine 89 and serine residues in mucins are highly solvated, repel one another, and make mucins extended, with an average 90 length of ~ 2 Å/residue (Clemetson, 1983;Woollett et al., 1985). Electron microscopy showed that D3 and A1 in aggregates. Global fits at all analyte concentrations to a single on and off-rate for each A1 construct were good 117 (Fig. 2). In contrast, when gel filtration was omitted, data could not be fit to a single on and off-rate. Differences 118 among all seven A1 constructs in 150 mM NaCl showed that the linkers were of great importance in regulating affinity by 2.5-fold. Glycosylation of the N-linker was also more important than the C-linker; the affinity of 124 A1+OGly+N was 10-fold lower than A1+N while the affinity of A1+OGly+C was only 2-fold lower than A1+C.

125
O-glycosylated linkers consistently lowered on-rates (Fig. 2B). Overall, the results showed that both the 126 polypeptide linker moiety and the O-glycan moiety of A1 linkers contributed to lowering A1 affinity for GPIbα, 127 that the N-linker was more important than the C-linker in lowering affinity, and that the combined effect was very 128 large at 50-fold.

129
To test electrostatic contribution to binding affinity and on-rates, binding of A1 short and the three

136
To independently test affinity, and obtain binding thermodynamics, we used isothermal titration 137 calorimetry (ITC). Although the GPIbα protein used in ITC lacked the avitag and its biotinylation, KD 138 measurements in BLI and ITC were on average within 1.3-fold of one another and showed the same trends ( Fig.   139 3). A1 short showed a 39-fold increase in affinity compared to A1+OGly+ N+C, the N-linker was more important 140 than the C-linker, and the O-glycan moiety was more important than the polypeptide moiety in regulating affinity.

141
Binding of all A1 protein constructs to GPIbα releases heats that showed the reaction was endothermic, i.e., 142 entropically driven (Fig. 3B). The thermodynamics of A1-GPIbα binding showed entropy-enthalpy 143 compensation: the higher the entropic contribution to binding, the higher the enthalpic cost, with all seven 144 constructs following the same rank order for the entropic and enthalpic terms. Thus, the linkers affected both the 145 entropic and enthalpic components of GPIbα binding.        Supplemental Figs. 1-5). HDX as a function of secondary structure and sequence position trended similarly for all exchange for all 108 peptides at all time points, from 10 s to 4 h. Over the A1 sequence from N to C-terminal, the 184 least exchange was seen for the β1-strand, the α1-helix, the β2 and β3-strands, the α3-helix, and the α6-helix, 185 which all had at least one peptide with less than 20% deuteration at 4 h. These slowly exchanging secondary 186 structural elements neighbor one another and the GPIbα binding site on A1 (Fig. 5D).
187 Figure 5. The N-linker decreases A1 dynamics measured by HDX. A) Relative deuterium exchange at all timepoints for A1 short as % of the available amide backbone H atoms in each peptide, colored according to the key. Figure Supplements  1 and 2 show data for all constructs. B) Difference (Δ) in HDX at all time-points of A1 short minus HDX for the other three constructs as indicated. In A) and B), residues with VWD type 2B mutations are shown below sequence numbers as red circles. C) Full HDX kinetics for selected peptides. Figure Supplements 3, 4, and 5 show data for all peptides. D) Structure of A1 bound to GPIbα (PDB 1SQ0). A1 is colored rainbow from N (blue) to C-terminus (red). Residues with VWD type 2B mutations are shown as Cα atom spheres. Labeled residues are shown in stick. GPIbα is shown in silver, from the βfinger to the β-switch. β2 loop and the β3-α2 loop. These loops locate near to GPIbα (Fig. 5B-D). We also observed reduced exchange 191 in the α2-α3 loop, but the most meaningful difference was only seen at the 4 h timepoint. The α1-β2 loop is 192 nearby the long-range disulfide and in a region with multiple gain of function VWD type 2B mutations (Fig. 5D).

193
The β3-α2 loop is near to GPIbα and has four basic residues but they are not close enough for optimal interaction 194 with acidic residues in the GPIbα leucine-rich repeats.

195
The HDX data showed no significant HDX differences between A1 short and A1+OGly+C. Further, the 196 magnitude of HDX differences between A1 short and both A1+OGly+N+C and A1+OGly+N was very similar.

197
However, while A1+OGly+N+C and A1+OGly+N were similar to one another in HDX, they both exchanged less 198 deuterium than A1 short and A1+OGly+C, indicating that both A1 short and A1OGly+C were more dynamic than 199 A1+OGly+N+C and A1+OGly+N. These HDX findings resembled the measurements of the stability of the 200 native states relative to the intermediate states (Figure 4), which showed that A1+OGly+N+C and A1+OGly+N

201
were similarly stable and more stable than A1 short and A1+OGly+C, which were also similar in stabiliy.

203
Among domains in proteins, the A1 domain of VWF is highly unusual in being separated from

250
We are unaware of previous thermodynamic measurements of A1 and GPIbα binding. ITC results show 251 that the interaction of GPIbα with all seven A1 constructs was endothermic with a large entropic term that ranged 252 from -8.2 to -14.7 kcal/mol at 22°C. Among all seven A1 constructs, the enthalpy and entropy terms were highly 253 correlated, so that their rank orders were identical. Entropy-driven binding of proteins is usually attributed to the 254 increase in water entropy when waters are released from hydrophobic binding interfaces (Richards, 1977).

255
Our HDX studies revealed that the O-glycosylated N-linker has a significant effect in stabilizing the α1-256 β2, β3-α2, and the α2-α3 loops, as shown by less deuterium incorporation. Previous HDX studies have shown that 257 destabilizing A1 by reducing and alkylating its long-range disulfide, introducing VWD Type 2B mutations, or

282
Our results on the effects of the O-linked N and C-linkers on affinity and stability differ. The C-linker 283 decreased A1 affinity for GPIbα in two different measures, BLI and ITC, but had no effect on A1 stability as 284 measured by HDX, melting temperature, and the relative stability of the A1 N and I states in urea denaturation 285 experiments. These results suggest that the C-linker decreases affinity by direct mechanisms such as by steric or 286 electrostatic repulsion. O-glycans are decorated with sialic acid, which would repel negatively charged GPIbα. In 287 contrast to the C-linker, the N-linker not only decreased affinity but also increased A1 stability as shown by 288 significantly decreased HDX, increased melting temperature, and increased stability of the N state relative to the I 289 state. The ~2 kcal/mol difference in ΔG between the N and I state for A1+Ogly+N compared to A1 short corresponds to a ~30-fold difference in population of the I state. Thus, if the I state was the high affinity state, 291 much of the 40-fold difference in affinity between the I and N states could be accounted for by the shift in 292 equilibrium toward the native state caused by the N-linker, and the remainder could be caused by steric and 293 electrostatic repulsion of GPIbα. In VWF concatemers, the ΔG required for activation by mechanical tension of 294 the high affinity state was measured as 1.9 kcal/mol (Fu et al., 2017), considerably lower than the difference in 295 energy between the I and N states measured for A1+Ogly+N+C here of 4.7 kcal/mol. These differences suggest 296 that the directional nature of energy input by mechanical tension allows for more efficient activation of A1 than 297 urea denaturation and may induce a distinct and less disordered intermediate state.

309
The association of the N-linker but not the C-linker with A1 stability is consistent with structural analysis 310 of A1 complexes with GPIbα that demonstrate that their association strains the N-terminal portion of A1 311 (reviewed in (Blenner et al., 2014)). The main site of GPIbα association is at the β-switch in its C-terminal cap.

312
The flexible β-finger in the N-terminal cap of GPIbα associates with A1 over a smaller region, including the α1-313 β2 loop, with few specific contacts that vary among GPIbα-complexes (Fig 5D). Strain in A1 upon binding GPIbα causes shifts in some residues in the α1-β2 loop and Cys-1272 away from GPIbα. Furthermore, the hydrogen is conserved in all structures of the isolated A1 domain, is lost in all A1-GPIbα complex structures. This loss including the α1-β2 loop, which showed increased HDX dynamics in the absence of the N-linker. Crystal

325
In summary, our results show that the A1 N-linker, but not the C-linker, increases the stability of the A1

332
Glycosylated proteins. Human VWF A1 domain constructs as shown in Figure 1C  Storage and re-purification. All proteins were stored at -80°C and subjected to a second round of size 367 exclusion chromatography on the day of measurements. This was essential to prevent protein aggregates from 368 contributing to affinity and kinetics measurements, i.e., to obtain data that can be reliably fitted with a 1:1 binding 369 model.   409 ,+Exp (-

437
The peptides were identified using PLGS 3.0.1 software and the HDX MS data were processed using DynamX