The linker between the D3 and A1 domains of vWF suppresses
A1-GPIb catch bonds by site speci c binding to the A1 domain
Alexander Tischer1, Miguel A. Cruz2, Matthew Auton1 
June 1, 2013
1 Division of Hematology, Departments of Internal Medicine and Biochemistry and Molecular Biology, Mayo Clinic,
Rochester, MN
2 Cardiovascular Sciences & Thrombosis Research Section of the Department of Medicine, Baylor College of
Medicine, Houston, Texas, USA
 To whom correspondence is addressed, auton.matthew@mayo.edu.
Author Contributions: Thermodynamic and  ow chamber studies were designed and performed by
Alexander Tischer and Matthew Auton. All authors contributed to the writing of the paper.
Running Title: Conformational Inhibition of VWF-platelet adhesion.
Keywords: von Willebrand Factor, parallel plate  ow chamber, Catch Bonds, Pause Time, Thermodynamics, Irreversible
Denaturation, Peptide Binding Circular Dichroism, Fluorescence, Di erential Scanning Calorimetry
Abbreviations Used: vWF, von Willebrand Factor; vWD, von Willebrand Disease; DSC, Di erential Scanning
Calorimetry; CD, Circular Dichroism; FL, Fluorescence; RIPA, ristocetin-induced platelet agglutination; SIPA, shear-induced
platelet agglutination
Abstract
Platelet attachment to von Willebrand factor (vWF) requires the interaction between the platelet
GP1b and exposed vWF-A1 domains. Structural insights into the mechanism of the A1-GP1b
interaction have been limited to an N-terminally truncated A1 domain that lacks residues Q1238 - E1260
that make up the linker between the D3 and A1 domains of VWF. We have demonstrated that removal
of these residues destabilizes quaternary interactions in the A1A2A3 tri-domain and contributes to
platelet activation under high shear [1]. In this study we demonstrate that removal of these residues
from the single A1 domain enhances platelet pause times on immobilized A1 under rheological shear. A
rigorous comparison between the truncated A1-1261 and full length A1-1238 domains demonstrates a
kinetic stabilization of the A1 domain induced by these N-terminal residues as evident in the enthalpy
of the unfolding transition. This stabilization occurs through site and sequence speci c binding of the
N-terminal peptide to A1. Binding of free N-terminal peptide to A1-1261 has an a nity
KD = 46 6M and this binding although free to dissociate is su cient to suppress the platelet pause
times to levels comparable to A1-1238 under shear stress. Our results support a dual
structure/function role for this linker region involving a conformational equilibria that maintains
quaternary A domain associations in the inactive state of vWF at low shear and an intra-A1-domain
conformation that regulates the strength of platelet GP1b -vWF A1 domain associations in the active
state of vWF at high shear.1 Introduction
The multimeric plasma glycoprotein, von Willebrand factor (vWF), is secreted from vascular endothelial cells
into the blood and subendothelium as long  laments of covalently coupled monomeric units, each containing
a conformationally regulated hook for platelet attachment. These monomeric units each contain a series of
domains arranged in the order, D’-D3-A1-A2-A3-D4-B1-B2-B3-C1-C2-CK [2, 3]. Within this sequence, the
A1 domain functions as the hook for capturing platelets through interaction with platelet glycoprotein 1b
and contributes to the arrest of bleeding under the vascular shear stress of blood  ow [4].
Early studies of the domain structure of vWF isolated a tryptic fragment between residues V1212 - K1491
that retained functional binding to platelet GP1b [5, 6]. In a variety of studies designed to assess platelet
adhesion, heparin and collagen binding, this sequence was engineered recombinantly by several investigators
with various lengths of the N and C termini outside of the major disul de loop that comprises the A1 domain
[7, 8, 9, 10]. Eventually, optimal lengths of these N and C termini were obtained in a construct spanning
residues Q1238 - P1471 that eliminated disul de linked aggregates and improved expression, puri cation and
solubility of the recombinant A1 domain as a monomeric species in solution [11]. For the better part of
the last two decades, this sequence has been demonstrated to retain the regulatory mechanism for platelet
adhesion in response to the shear stress of blood  ow. It has been demonstrated to form high strength bonds
with platelet GP1b [12, 13], shear dependent platelet rolling on A1 domain coated surfaces [14, 15] and force
dependent single molecule catch bonds with platelet GP1b [15]. Furthermore, it has provided internally
consistent relationships between the interdependence of conformational thermodynamics with shear and force
dependent binding of platelet GP1b based on clinically identi ed mutations that cause opposite functional
phenotypes in von Willebrand disease (vWD) [16, 17].
Early e orts to obtain a crystal structure of this domain were unsuccessful until chymotrypsin was used
to cleave o residues Q1238 - Y1258 and di ractable crystals were obtained [18, 19]. While this truncated
domain spanning the structurally resolved residues D1261 - P1471 can bind to GP1b , platelet adhesion to
it at high shear rates was not as e cient as the domain containing the N-terminal  anking region which
showed similar adhesive activity as native VWF [20, 21]. Subsequent crystal structures of the truncated
A1 domain and its co-crystal complex with GP1b with and without type 2 vWD mutations show that
the structures are comparatively similar with the backbone RSMD less than 1 A over much of the sequence
within the disulphide loop (Figure 1) [22, 19, 23, 24]. Consequently, structural insights into the mechanisms
of type 2 von Willebrand Disease have been inadequate.
It has been known since the late 1980’s that 15 residue peptides spanning the region between L1232
- D1261 inhibit greater than 50% of the binding of vWF to platelet GP1b. Furthermore, these peptides
can completely inhibit the interaction between vWF and two monoclonal antibodies (NMC-4 and RG-46)
known to block ristocetin and botrocetin-induced binding of vWF to GP1b [25, 26]. Co-crystal structures
of A1 in complex with NMC-4 moAb show that NMC-4 binds A1 at the  4  5 helix-loop-helix [27]. Given
the location of the structurally resolved N-terminal residues beginning at D1261, these data suggest that
inhibition of the A1:NMC-4 interaction by the sequence Q1238 - E1260 would involve a signi cant wrapping
of the N-terminal sequence around the A1 domain. This type of interaction could result in a shielding e ect
that inhibits binding between A1 and GP1b as has been demonstrated using vWF A domain fragments
containing the N-terminal D’D3 domains [28]. What is clear is that this N-terminal sequence, Q1238 - E1260,
is highly dynamic and its inhibition of crystal growth has precluded an accurate structure-based mechanism
of GP1b binding because of the inability to resolve its conformation in complex with the A1 domain.
Our recent studies have shown that truncation of this N-terminal sequence destabilizes domain asso-
ciations within the A1A2A3 tridomain resulting in a moderate increase in avidity for platelet GP1b and
 brinogen-dependent platelet activation at high shear [1]. Similarly, moAb 1C1E7 directed at the N-terminal
sequence induces a vWF functional phenotype similar to type 2B VWD with enhanced RIPA, SIPA, and
1evidence of platelet activation via increased intracellular Ca2+ [29, 30]. Here we show using  ow chamber
studies that this N-terminal  anking sequence suppresses the catch bond by decreasing the pause time of
platelet interactions with surface immobilized A1 domain. Furthermore, we provide a thermodynamic com-
parison of the stability of the A1 domain with and without its N-terminal  anking region and demonstrate
that A1 binds speci cally to the N-terminal peptide in solution. The binding of the free N-terminal peptide
by A1 lacking this sequence also supresses the catch bond.
2 Results.
2.1 Platelet adhesion to immobilized A1 under shear  ow.
To understand the e ect of the N-terminal peptide on the function of the A1-domain of vWF, the protein was
expressed with (A1-1238) and without the N-terminal peptide (A1-1261). We used a  ow chamber to measure
the dynamics of platelet interactions with A1-1238 and A1-1261. Each domain variant was immobilized on
glass slides with a modi ed Cu2+ chelating chemistry that enabled surface capture via the N-terminally fused
histidine tag. This method ensured reproducible capture of the proteins eliminating possible structural issues
associated with nonspeci c immobilization on plastic or glass. The slides were completely inert to platelets.
1mL of citrated whole blood followed by Tris bu ered saline was perfused at low shear to enable platelets to
attach to the surface immobilized domains. After red cells were perfused away with bu er, the translocation
of the remaining attached platelets over the surfaces was recorded at each shear rate and the velocities and
pause times calculated from the coordinate data of each platelet on the surface as described in the methods.
Figure 2 shows the pause time data as a function of shear rate along with the number of platelet tracks
analyzed from each movie. A1-1261 has longer pause times than A1-1238 indicating that the presence of the
covalently linked N-terminal sequence decreases the time for which platelets remain attached. In addition,
the translocation velocities were increased for A1-1238 relative to A1-1261. To determine if this N-terminal
sequence could inhibit the catch bond when free in solution, we incubated surface immobilized A1-1261
with excess free peptide and added the free peptide to 1mL of whole blood and to our perfusion bu er and
performed the assay. The result was a signi cant decrease in pause times to the levels of that obtained for
A1-1238 and an increase in translocation velocities. Addition of free N-terminal peptide to A1-1238 or a
peptide with the same amino acid composition, but a scrambled sequence, to either A1-1238 or A1-1261 did
not signi cantly change the pause times or translocation velocities observed in the absence of peptide.
2.2 Structural characterization
CD spectra obtained for A1-1261 and for A1-1238 are shown in Figure 3A. Both proteins show a spectrum
dominated by relatively high  -helical content as indicated by the two minima at  = 222nm and 208nm.
Comparatively, the ellipticity of A1-1261 is slightly reduced relative to A1-1238, but a calculation of the
secondary structure contributions for the two proteins (Supplement Table 1) shows similar  -helical and
parallel and anti-parallel  -sheet structure [31]. By contrast, the spectrum of the N-terminal peptide is
dominated by the contribution of random coil-structure and spectrally equivalent to a peptide containing a
scrambled sequence but identical amino acid composition.
2.3 Urea-induced unfolding
A1-1261 and A1-1238 were chemically unfolded with urea in order to compare the obtained thermodynamic
parameters. The unfolding was monitored at 25C by CD at  = 222nm, Figure 3B. For both proteins,
unfolding through an intermediate state was observed, as has been previously reported [17]. The thermody-
namic parameters are illustrated in the insets of Figure 3B and listed in the Supplement Table 2. Relative to
2A1-1261, A1-1238 has a marginal loss in unfolding cooperativity (+0:55 0:38 kJ=mol=M) of the transition
and the midpoint, cm, of A1-1238 is increased by +0:27  0:04 M . These two contributions to the  G0
cancel each other so that the N-terminal peptide is neither stabilizing nor destabilizing the conformational
transition.
2.4 Temperature-induced unfolding
A1-1261 and A1-1238 were thermally unfolded using intrinsic protein tyrosine and tryptophan  uorescence
with an excitation  = 280nm and emission  = 359nm and CD monitored at  = 222nm. The observed
thermal scans shown in Figure 4 demonstrate a strong dependency of the primary unfolding transition on
the rate at which the temperature is changed indicating that the unfolding is kinetically controlled. As
the scan rate increases, the midpoint of the transition (apparent TM ) also increases. Closer inspection
of the thermal transition in Figure 4 shows that while both spectroscopic methods capture the primary
unfolding event between 40 C and 60 C,  uorescence captures a second transition at high temperature
in A1-1238 (Figure 4B) which is also scan rate dependent. This high temperature transition is absent in
A1-1261 suggesting that the removal of the N-terminal sequence further exposes  uorophors in the protein
(Figure 4A). Thermal scans of the peptide alone did not reveal any evidence of a conformational transition
by either CD or  uorescence indicating that this high temperature transition results from the exposure of
the tyrosine in the N-terminal peptide when covalently linked to the A1 domain.
Thermal transitions were analyzed in terms of the observed midpoint of the transition, apparent TM ,
and also by a two-state irreversible model or a three-state irreversible model as described in the Supplement
and the calculated parameters are summarized in Figure 5. Panel A demonstrates the increasing apparent
TM as a function of scan rate for all methods employed. If the protein unfolding was at equilibrium and
reversible, this apparent TM would remain constant at all scan rates. Extrapolation of the TM to a scan rate
of 0 C=min provides an indication of where reversible unfolding would occur under equilibrium conditions;
50 C for A1-1238 and 48 C for A1-1261 from both CD and  uorescence data. For the second transition
observed by  uorescence that occurs with A1-1238, the TM extrapolates to  73 C in the limit of zero scan
rate. Analysis by the two-state and three-state irreversible models yields an (R2 > 0:99) for both  uorescence
and CD and a transition temperature, T  , that is independent of the temperature scan rate. In addition,
the enthalpies of the transitions are independent of scan rate within experimental error (Panel B ). Panels
C & D show the resulting average T  and  HzT  the standard deviation obtained from the  tting of
the primary thermal transitions in Figure 4 for CD and  uorescence and a total average for both methods.
Comparing the transition temperatures for the primary unfolding transition shows that both A1-1261 and
A1-1238 have an identical T  within experimental error. A1-1261 (T  = 64:5  0:9C) is not signi cantly
di erent than A1-1238 (T  = 63:9  0:6C). However, the unfolding enthalpy was statistically larger for
A1-1238 (250  15 kJ=mol) than for A1-1261 (204  10 kJ=mol). The average T  and  HzT observed in
the second  uorescence transition of A1-1238 is 89:1 2:2C and 360 14 kJ=mol.
Thermal unfolding at 2C=min by DSC also shows that the T  is not signi cantly di erent between A1-
1261 and A1-1231, but the enthalpy of unfolding is increased for A1-1238 (Figure 6). The values obtained
via DSC were not quantitatively comparable to those obtained from CD and  uorescence as a result of the
aggregation of both proteins at the high temperatures and concentrations ( 50M) required to obtain
reliable signal/noise in the calorimetric power compensation. It should also be noted that all thermal
transitions remained irreversible in the presence of 1M arginine, an excipient known to solubilize proteins,
and a scan rate dependency was also observed in the presence of 1M urea con rming that thermal unfolding
is intrinsically irreversible and not an artifact of any nonspeci c aggregation.
32.5 Interaction of A1-1261 with the N-terminal peptide
Since the thermal denaturation of A1 revealed that the covalently linked N-terminal peptide increases the
enthalpy of the unfolding transition of the A1 domain, it was of interest whether this property could be used
to detect binding of the peptide in solution to A1-1261. This would imply a speci c interaction between
this N-terminal sequence and the A1 domain when covalently linked in sequence. The only way to examine
such a stabilization using spectroscopy was the measurement of thermal unfolding via  uorescence. Since
the N-terminal peptide also contains a single tyrosine residue, the excitation wavelength was changed to
 = 295nm in order to measure tryptophan  uorescence and the e ect of the peptide on the unfolding of
the A1-domain was observed indirectly.
Thermal scans of 2uM A1-1261 were performed as a function of increasing amounts of the N-terminal
peptide at 2 C=min. Figure 7A shows that the primary transition shifts from an apparent TM ’ 53 C in
the absence of peptide to ’ 55:5 C in the presence of saturating concentrations of peptide. The second
transition becomes more prominent as the peptide concentration increases indicating that the peptide is
binding to A1-1261. Thermal scans were also done in the presence of a peptide with the same amino acid
composition, but a scrambled sequence. The inset of Figure 7A compares the raw  uorescence thermal
scans in the presence of 300 M scrambled and N-terminal peptides. Unfolding in the presence of excess
scrambled peptide was equivalent to unfolding in the absence of either peptide. Figure 7B shows the resulting
 HzT and T
 obtained from these transitions as a function of the total concentration of N-terminal peptide.
Whereas the T  was constant within  1:0C, the enthalpy of the transition changed sigmoidally on a log
scale from  197kJ=mol to  242kJ=mol indicating that the unfolding enthalpy is sensitive to the binding of
the N-terminal peptide. Fitting the enthalpy data to a simple single site binding model yields a dissociation
constant of  46 6M . Also the binding of the N-terminal peptide to A1-1261 was sequence speci c since
the scrambled peptide did not change the unfolding enthalpy at saturating concentrations. The N-terminal
peptide did not change the unfolding enthalpy of A1-1238 suggesting that the binding is site speci c and the
covalently linked peptide occupies the binding site so that free N-terminal peptide cannot bind. In addition,
the binding of the N-terminal peptide is fully reversible as excessive equilibrium dialysis of A1-1261 following
incubation with 200 M peptide for 1 hour resulted in a thermal scan with a  HzT equivalent A1-1261 in
the absence of peptide.
The average T  and  HzT observed for the peptide-induced second  uorescence transition of A1-1261
are 90:1  8:5C and 142  37:1 kJ=mol. This second transition yields a comparable T  value to A1-1238
(89:1 2:2C), although is less well de ned in cooperativity with a signi cant drop in  HzT from 360 14
kJ=mol when covalently linked to 142  37:1 kJ=mol when free to dissociate from the A1 domain over the
full range of peptide concentrations studied.
3 Discussion.
The results presented demonstrate the following important properties of the A1 domain.
1) The covalently linked N-terminal peptide suppresses the A1-GPIb catch bond resulting in decreased
pause times and faster translocation velocities. Speci c binding of the N-terminal peptide free in solution
to A1-1261 results in diminished pause times and enhanced velocities that are comparable to those obtained
with A1-1238.
2) The N-terminal peptide, free in solution, is unstructured with > 80% of the sequence having  and
random coil content. In its native sequential context, this 23 residue peptide does not signi cantly alter the
overall secondary structure content of the A1 domain. Therefore, this peptide is likely unstructured in the
natural sequential context relative to the A1 domain.
3) Reversible urea denaturation illustrates that the stability of both A1-1261 and A1-1238 have an
4equivalent thermodynamic stability at 25 C despite small changes in the cm and the m-value. Irreversible
thermal denaturation illustrates that both A1-1261 and A1-1238 have equivalent transition temperatures
(T  ) that de ne where the rate of unfolding is unity. Interestingly, the enthalpy of unfolding is larger
for A1-1238 than A1-1261 which indicates that the rate of unfolding of 1261 to the intermediate state at
temperatures lower than the T  will be faster than A1-1238.
4) Although urea denaturation of A1 is reversible, thermal denaturation is irreversible and kinetically
controlled. Kinetic irreversibility is evident by the signi cant scan rate dependence of the apparent melting
temperature. Analysis of CD and  uorescence transitions using an irreversible two-state model de ned
by a  rst order rate constant yield transition temperatures that are scan rate independent. Furthermore,
the observed thermal transitions represent the native to intermediate conformational change observed in
urea denaturation. This is evident by the signi cant amount of CD ellipticity remaining in the thermally
denatured state and the presence of a second high temperature transition from the intermediate state to
denatured observed by  uorescence when excited at 280nm. The fact that this high temperature transition
is not observed with A1-1261 indicates that the N-terminal peptide is involved in this I ! D transition. In
addition, extrapolation of the  rst apparent Tm of A1-1238 to zero scan rate (equilibrium) gives 49:8 1:2C.
This is in agreement with the Tm = 51:1 0:5C obtained from the reversible N  I urea-temperature phase
diagram determined previously [17]. The apparent Tm of the second transition of A1-1238 is also comparable
to that obtained from the I  D urea-temperature phase diagram although the error associated with that
transition midpoint is larger [17].
5) In solution, the N-terminal peptide binds speci cally to the truncated A1-1261. This binding is
observed in the unfolding transition enthalpy which changes sigmoidally with respect to the natural log of
the peptide concentrations from the enthalpy associated with A1-1261 unfolding to an enthalpy characteristic
of A1-1238 unfolding. Scrambling the peptide sequence did not have any change in the unfolding properties
of A1-1261 and the N-terminal peptide of correct sequence did not bind to A1-1238. These results indicate
site and sequence speci c binding of the peptide to A1-1261 at a site that is normally occupied by the
covalently linked N-terminal peptide in A1-1238.
In von Willebrand Disease, the e ect of inherited type 2B and 2M mutations on the N  I conforma-
tional equilibrium is now established for serveral mutations [17]. These subtypes of mutations in the A1
domain cause clinically opposite phenotypes resulting in bleeding. From the perspective of reversible urea
denaturation at variable temperature, it has been shown that type 2M mutations shift the equilibrium in
favor of the native state and decrease the a nity of A1 for platelet GP1b. Conversely, type 2B mutations
shift the equilibrium in the opposite direction toward the intermediate state and increase the a nity of A1
for platelet GP1b. Here we have studied the irreversible thermal stability of A1 in the absence of urea
and have demonstrated that truncating the N-terminal linker peptide from the A1 domain decreases the
enthalpy of the N ! I transition without altering the transition temperature. This will necessarily increase
the rate of unfolding from the native to intermediate state at temperatures below the transition temperature.
However, the thermal stability of A1-1261 in solution can be restored to that of A1-1238 through speci c
binding of the free N-terminal peptide. The rate dependent e ect of this intramolecular interaction between
the A1 domain and its N-terminal linker establishes the possibility of an on/o conformational switch that is
a ected by vWD mutations and by rheological shear stress on vWF multimers. In the crystal structure, the
most frequently identi ed type 2B mutations cluster in the loops and turns near the interface between the
lower surface of the domain and the N-terminus. In addition to their destabilization of the globular domain
structure, part of their e ect on vWF conformation could be to favor dissociation of this N-terminal linker
resulting in the uncoupling of A and possibly D domains within VWF which would enhance interactions
between A1 and platelet GP1b. We have observed that a monospeci c antibody (A108), directed against
the sequence D1444 - E1452 within the  6 helix of the A1 domain, has a greater reactivity to the truncated
51261-A1A2A3 tridomain relative to the longer 1238-A1A2A3 tridomain. It also has a greater reactivity
to puri ed plasma vWF in the presence of 0:5 mg=mL ristocetin [1]. Therefore, it is plausible that this
structural region of the A1 domain plays a role for the binding of the N-terminal linker.
A corollary to the e ect of vWD mutations on A1 domain conformational equilibria is their e ect on
the shear stress dependent platelet rolling velocities on immobilized A1 and force dependent single bond
lifetimes between vWF A1 and platelet GP1b . These properties are proportional to the strength of the
interaction. We previously demonstrated on a single molecule scale that this interaction becomes stronger
and increases the lifetime of the protein complex up to a critical threshold force above which the bond
weakens and the lifetime decreases [15]. This force sensing binding process referred to as "catch-bonding"
regulates platelet detachment, rolling velocities and overall adhesiveness of vWF as a function of shear
stress. Type 2B mutations shift this threshold to lower forces due to the low stability of the native state.
Consequently, at lower shear, attached platelets persist due to the higher bond strength and increase the risk
for microthrombi. The stabilization of the native state caused by type 2M mutations shifts this threshold
to higher forces and weakens the bond strength at physiological shear stress thereby decreasing the adhesive
capacity of vWF for platelet GP1b [16].
Here we show that the platelet pause time to immobilized A1 has a similar behavior to the single
molecule bond lifetime measurements with a sudden increase in platelet pause time at a critical shear
threshold  1500s1 [15]. Truncation of the N-terminal linker from the A1 domain enhances the pause times
at low shear and shear rates that de ne the catch bond regime. However, addition of saturating amounts of
free N-terminal peptide to the surface immobilized A1-1261 restores the pause times to those observed for
A1-1238. Considering both of these observations, the fact that the binding of the free peptide in solution to
A1 restores both the kinetic thermal stability and the catch-bond properties of the truncated domain to that
of the full length A1 domain demonstrates that the binding of this N-terminal linker is speci c and critical
for modulating the rheological dependence platelet adhesion to VWF.
Our previous studies on the A1A2A3 tridomain fragment of vWF have demonstrated that similar to gain
of function mutations [32], truncation of this sequence also destabilizes the quaternary association of these
domains resulting in e ective inhibition of RIPA and  brinogen-dependent platelet activation under high
shear [1]. These studies con rmed a structural role for this N-terminal sequence to maintain inter-domain
associations and keep the tridomain in a thermodynamically stable binding incompetent conformation. The
observations reported here highlight another role of the N-terminal linker as a force sensor for the regulation
of platelet adhesion to the A1 domain in the open active state of vWF.
In conclusion, the rotational and elongational forces on plasma vWF and the tensile forces on vascular
wall tethered ULvWF multimers that occur in the presence of rheological shear likely activate vWF in a
two step process that  rst dissociates the A domains and subsequently modulates the kinetic stability of
the A1 domain and its a nity for platelet GP1b. These processes appear to be regulated by an on/o 
conformational equilibria between the A1 domain and its N-terminal  anking sequence although this may
be a relatively small contribution to the overall process of conformational activation of full length VWF in
vivo.
4 Materials & Methods
4.1 Proteins
Recombinant human vWF A1-1238 (amino acids Q1238 - P1471) and its N-terminal truncated variant A1-1261
(amino acids D1261 - P1471) were expressed in E. coli M-15 cells as fusion proteins containing a N-terminal
6xHis-Tag using BamHI and HindIII restriction sites in the Qiagen pQE-9 plasmid vector [33, 34]. Proteins
were puri ed from inclusion bodies by solubilization in 6 M GuHCl followed by refolding by excessive dilution
6into 4L TBS-T and isolated by a nity chromatography using Ni2+-chelated Sepharose followed by a second
puri cation via Heparin-Sepharose. The purity of the recombinant proteins was con rmed via reducing SDS-
PAGE and also by size exclusion chromatography using a Phenomenex SEC S3000 column on a BioCAD
Sprint perfusion chromatography system. Proteins were stored at 0 C in 150mM NaCl, 25mM Tris HCl,
pH = 7:4 and dialyzed overnight against a temperature stable bu er-mixture of 10mM sodium acetate,
10mM Na2HPO4, 10mM Glycine, 150mM NaCl, 1mM EDTA, pH = 8 before all measurements.
The N-terminal peptide (QEPGGLVVPPTDAPVSPTTLYVE) corresponding to residues Q1238 - E1260
and the scrambled peptide (QLPTGVLGEPSDAVPTVYEVTPPG) were synthesized at the Mayo Clinic
Proteomics Core Lab with a free N-terminal amine and C-terminal acid, checked for purity via reversed
phase HPLC and their correct masses of 2365:2Da and 2430:25Da were veri ed via electrospray ionization
mass spectrometry.
Protein and peptide concentrations were quanti ed using a Shimadzu UV2101PC spectrophotometer
with the Pace method [36] from absorption at  = 280nm minus twice the absorption at  = 333nm
for correction of light scattering. Extinction coe cients for A1-1238 ( = 15350 L=mol=cm) and A1-1261
( = 14235 L=mol=cm) and the N-terminal peptide ( = 1215 L=mol=cm) were calculated from the number
of tyrosines (A1-1238 = 8, A1-1261 = 7, N-terminal peptide = 1) and a single tryptophan.
4.2 Determination of platelet pause times.
A Glycotech1 rectangular parallel-plate  ow chamber was mounted to slides with a chelated Cu2+ surface
obtained from Microsurfaces Inc.2 to which the A1-1261 and A1-1238 domains in TBS-T were immobilized
by the 6xHis-Tag. 1mL of citrated whole blood obtained from informed consent of healthy donors with
approval from the Baylor College of Medicine institutional review board followed by bu er was perfused
through the chamber at 300s1 for A1-1261 and 700s1 for A1-1238 shear rate to attach platelets to the
surface immobilized proteins using a syringe driven variable  ow rate pump. The shear was increased
incrementally and 1min movies were recorded at 25 frames=sec in phase contrast with 2x2 pixel binning on
a Zeiss Axiocam Mrm camera attached to a Zeiss Axio Observer D1 inverted microscope. Tracking analysis
was performed using Mediacybernetics ImagePro Premier3. The resulting X-Y coordinate data for each
platelet track was di erentiated into instantaneous velocities using a Savitzky-Golay algorithm [37] with
a  ve data point window size and a second order polynomial implemented into a Mathematica4 notebook
written in our lab. These platelet tracks were retained for statistical analysis if the platelet was present for
at least 1sec and travelled a total distance greater than 1m. Distance travelled in pixels was converted to
 m given the camera pixel size = 6:45m2, pixel binning and microscope magni cation = 400x. Pause times
were determined by the amount of time (sec) a platelets velocity was = 0  0:13 m=s, within the noise.
Pause times for all platelet tracks are reported as the mean of the medians. Examples of this analysis for
individual platelets and for multiple platelets are provided in the Supporting Information.
4.3 Protein denaturation
Urea and thermal unfolding of A1-1238 and A1-1261 was monitored by CD and  uorescence spectroscopy
and DSC using an Aviv Biomedical Model 420C CD spectrometer, a Horiba Jobin-Yvon Fluorolog 3 spec-
tro uorimeter equipped with a Wavelength Electronics Model LF1-3751 temperature controller and a TA
Instruments NanoDSC.
Isothermal urea induced unfolding of A1-1261 and A1-1238 at 25C was monitored via CD at  = 222nm
1http://www.glycotech.com/apparatus/parallel.html
2http://www.proteinslides.com/histag.html
3http://www.mediacy.com/index.aspx?page=IP Premier
4http://www.wolfram.com/mathematica/
7using a quartz cuvette with a path length of 1mm and a protein concentration of 10M after an overnight
equilibration. CD-signal was averaged for 10min, corrected for the corresponding CD-signal of the bu er
and converted into mean molar ellipticities per amino acid residue ( MRW ).
Prior to all spectroscopic thermal scans, protein samples were equilibrated at 10C for 15min to obtain
a stable baseline. CD thermal scans between 10 and 95 C were recorded at  = 222nm using a protein
concentration of 1 or 2M in a 1cm quartz cell under moderate stirring. Scan rates were 0:5, 1:0, 1:5 and
2:0C=min. Integration time for each data point was 20s with a 1nm bandwidth. Fluorescence thermal scans
between 10 and 95 C were recorded at an emission  = 359nm after excitation  = 280nm or  = 295nm
using 2M protein in a 1cm quartz cell with moderate stirring. Scan rates were 0:4 or 0:6, 0:9, 1:6 and
2:0C=min. At each temperature, relative  uorescence intensity was collected for 4s and averaged. Thermal
scans in the presence of N-terminal peptide at 2:0C=min were preceded by 1h incubations at 20 C with
0:02 300 M peptide.
DSC was performed at 3atm pressure. Protein samples (40 to 50 M) and bu ers were degassed under
moderate stirring prior to use. The calorimeter was equilibrated overnight with bu er in the sample and ref-
erence cell to obtain a stable baseline and the protein sample was loaded during an equilibration step between
scans. One measurement per second was recorded at 2:0 C=min. DSC-traces were background corrected
with the following irreversible scan that was used as a baseline. The molar heat capacity was calculated
from the calorimetric power compensation (J=s) using the relation, CP (kJ=mol=K) = Power(J=s) =
[(( C=min))=60  V (mL)  c(mol=L)]. The excess molar heat capacity, hCP i (kJ=mol=K) was obtained
by subtracting a polynomial baseline  t to the pre- and post-transition regions of the heat capacity traces
using a Mathematica4 notebook written in our laboratory.
Isothermal urea unfolding was analyzed using a three state reversible model (N  I  D) as previously
described [17, 35]. All thermal unfolding transitions were analyzed according to the irreversible models
provided in the Supporting Information. [38, 39, 40, 41].
5 Acknowlegements
This work was supported by National Heart Lung and Blood Institute of the National Institutes of Health
HL109109 (M.A.) and the Mary R. Gibson Foundation (M.A.C. and M.A.).
References
[1] Auton M, Sowa KE, Behymer M, Cruz MA (2012) The N-terminal  anking region of A1 domain in Von
Willebrand Factor stabilizes the structure of the A1A2A3 complex and modulates platelet activation
under shear stress. The Journal of biological chemistry 287:1457914585.
[2] Bonthron DT, Handin RI, Kaufman RJ, Wasley LC, Orr EC, Mitsock LM, Ewenstein B, Loscalzo J,
Ginsburg D, Orkin SH (1986) Structure of pre-pro-von Willebrand factor and its expression in heterol-
ogous cells. Nature 324:270273.
[3] Sadler JE (1998) Biochemistry and genetics of von Willebrand factor. Annu. Rev. Biochem. 67:395424.
[4] Savage B, Saldivar E, Ruggeri ZM (1996) Initiation of platelet adhesion by arrest onto  brinogen or
translocation on von Willebrand factor. Cell 84:289297.
4http://www.wolfram.com/mathematica/
8[5] Fujimura Y, Titani K, Holland LZ, Russell SR, Roberts JR, Elder JH, Ruggeri ZM, Zimmerman TS
(1986) von Willebrand factor. A reduced and alkylated 52/48-kDa fragment beginning at amino acid
residue 449 contains the domain interacting with platelet glycoprotein Ib. J Biol Chem 261:381385.
[6] Fujimura Y, Titani K, Holland LZ, Roberts JR, Kostel P, Ruggeri ZM, Zimmerman TS (1987) A
heparin-binding domain of human von Willebrand factor. Characterization and localization to a tryptic
fragment extending from amino acid residue Val-449 to Lys-728. J Biol Chem 262:17341739.
[7] Pietu G, Meulien P, Cherel G, Diaz J, Baruch D, Courtney M, Meyer D (1989) Production in Es-
cherichia coli of a biologically active subfragment of von Willebrand factor corresponding to the platelet
glycoprotein Ib, collagen and heparin binding domains. Biochem. Biophys. Res. Commun. 164:13391347.
[8] Sugimoto M, Ricca G, Hrinda ME, Schreiber AB, Searfoss GH, Bottini E, Ruggeri ZM (1991) Func-
tional modulation of the isolated glycoprotein Ib binding domain of von Willebrand factor expressed in
Escherichia coli. Biochemistry 30:52025209.
[9] Gralnick HR, Williams S, McKeown L, Kramer W, Krutzsch H, Gorecki M, Pinet A, Gar nkel LI
(1992) A monomeric von Willebrand factor fragment, Leu-504{Lys-728, inhibits von Willebrand factor
interaction with glycoprotein Ib-IX. Proc. Natl. Acad. Sci. U.S.A. 89:78807884.
[10] Azuma H, Dent JA, Sugimoto M, Ruggeri ZM, Ware J (1991) Independent assembly and secretion of
a dimeric adhesive domain of von Willebrand factor containing the glycoprotein Ib-binding site. J Biol
Chem 266:1234212347.
[11] Cruz MA, Handin RI, Wise RJ (1993) The interaction of the von Willebrand factor-A1 domain with
platelet glycoprotein Ib/IX. The role of glycosylation and disul de bonding in a monomeric recombinant
A1 domain protein. J Biol Chem 268:2123821245.
[12] Arya M, Anvari B, Romo GM, Cruz MA, Dong JF, McIntire LV, Moake JL, Lopez JA (2002) Ultralarge
multimers of von Willebrand factor form spontaneous high-strength bonds with the platelet glycoprotein
Ib-IX complex: studies using optical tweezers. Blood 99:39713977.
[13] Arya M, Kolomeisky AB, Romo GM, Cruz MA, Lopez JA, Anvari B (2005) Dynamic force spectroscopy
of glycoprotein Ib-IX and von Willebrand Factor. Biophys. J. 88:43914401.
[14] Coburn LA, Damaraju VS, Dozic S, Eskin SG, Cruz MA, McIntire LV (2011) GPIb-vWF rolling under
shear stress shows di erences between type 2B and 2M von Willebrand disease. Biophys. J. 100:304312.
[15] Yago T, Lou J, Wu T, Yang J, Miner JJ, Coburn L, Lopez JA, Cruz MA, Dong JF, McIntire LV, et
al. (2008) Platelet glycoprotein Ibalpha forms catch bonds with human WT vWF but not with type 2B
von Willebrand disease vWF. J. Clin. Invest. 118:31953207.
[16] Auton M, Zhu C, Cruz MA (2010) The Mechanism of VWF-Mediated Platelet GPIbalpha Binding.
Biophys. J. 99:11921201.
[17] Auton M, Sedlak E, Marek J, Wu T, Zhu C, Cruz MA (2009) Changes in thermodynamic stability of
von willebrand factor di erentially a ect the force-dependent binding to platelet GPIbalpha. Biophys.
J. 97:618627.
[18] Emsley J, Cruz M, Handin R, Liddington R (1998) Crystal structure of the von Willebrand Factor A1
domain and implications for the binding of platelet glycoprotein Ib. J Biol Chem 273:1039610401.
9[19] Fukuda K, Doggett TA, Bankston LA, Cruz MA, Diacovo TG, Liddington RC (2002) Structural basis
of von Willebrand factor activation by the snake toxin botrocetin. Structure 10:943950.
[20] Kim J, Zhang CZ, Zhang X, Springer TA (2010) A mechanically stabilized receptor-ligand  ex-bond
important in the vasculature. Nature 466:992995.
[21] Miyata S, Ruggeri ZM (1999) Distinct structural attributes regulating von Willebrand factor A1 domain
interaction with platelet glycoprotein Ibalpha under  ow. J Biol Chem 274:65866593.
[22] Celikel R, Ruggeri ZM, Varughese KI (2000) von Willebrand factor conformation and adhesive function
is modulated by an internalized water molecule. Nat Struct Biol 7:881884.
[23] Huizinga EG, Tsuji S, Romijn RA, Schiphorst ME, de Groot PG, Sixma JJ, Gros P (2002) Structures
of glycoprotein Ibalpha and its complex with von Willebrand factor A1 domain. Science 297:11761179.
[24] Dumas JJ, Kumar R, McDonagh T, Sullivan F, Stahl ML, Somers WS, Mosyak L (2004) Crystal
structure of the wild-type von Willebrand factor A1-glycoprotein Ibalpha complex reveals conformation
di erences with a complex bearing von Willebrand disease mutations. J Biol Chem 279:2332723334.
[25] Mohri H, Fujimura Y, Shima M, Yoshioka A, Houghten RA, Ruggeri ZM, Zimmerman TS (1988) Struc-
ture of the von Willebrand factor domain interacting with glycoprotein Ib. J Biol Chem 263:1790117904.
[26] Fujimura Y, Usami Y, Titani K, Niinomi K, Nishio K, Takase T, Yoshioka A, Fukui H (1991) Studies
on anti-von Willebrand factor (vWF) monoclonal antibody NMC-4, which inhibits both ristocetin- and
botrocetin-induced vWF binding to platelet glycoprotein Ib. Blood 77:113120.
[27] Celikel R, Varughese KI, Madhusudan, Yoshioka A, Ware J, Ruggeri ZM (1998) Crystal structure of
the von Willebrand factor A1 domain in complex with the function blocking NMC-4 Fab. Nat Struct
Biol 5:189194.
[28] Ulrichts H et al. (2006) Shielding of the A1 domain by the D’D3 domains of von Willebrand factor
modulates its interaction with platelet glycoprotein Ib-IX-V. J Biol Chem 281:4699-4707.
[29] Tornai I, Arnout J, Deckmyn H, Peerlinck K, Vermylen J (1993) A monoclonal antibody recognizes
a von Willebrand factor domain within the amino-terminal portion of the subunit that modulates the
function of the glycoprotein IB- and IIB/IIIA-binding domains. J. Clin. Invest. 91:273282.
[30] Ulrichts H, Harsfalvi J, Bene L, Matko J, Vermylen J, Ajzenberg N, Baruch D, Deckmyn H, Tornai
I (2004) A monoclonal antibody directed against human von Willebrand factor induces type 2Blike
alterations. Journal of Thrombosis and Haemostasis 2:16221628.
[31] B ohm G, Muhr R, Jaenicke R (1992) Quantitative analysis of protein far UV circular dichroism spectra
by neural networks. Protein Eng. 5:191195.
[32] Auton M, Sowa KE, Smith SM, Sedlak E, Vijayan KV, Cruz MA (2010) Destabilization of the A1
domain in von Willebrand factor dissociates the A1A2A3 tri-domain and provokes spontaneous binding
to glycoprotein Ibalpha and platelet activation under shear stress. J Biol Chem 285:2283122839.
[33] Morales LD, Martin C, Cruz MA (2006) The interaction of von Willebrand factor-A1 domain with
collagen: mutation G1324S (type 2M von Willebrand disease) impairs the conformational change in A1
domain induced by collagen. J Thromb Haemost 4:417425.
[34] Cruz MA, Diacovo TG, Emsley J, Liddington R, Handin RI (2000) Mapping the glycoprotein Ib-binding
site in the von willebrand factor A1 domain. J Biol Chem 275:1909819105.
10[35] Auton M, Cruz MA, Moake J (2007) Conformational Stability and Domain Unfolding of the Von
Willebrand Factor A Domains. J. Mol. Biol. 366:9861000.
[36] Pace CN, Vajdos F, Fee L, Grimsley G, Gray T (1995) How to measure and predict the molar absorption
coe cient of a protein. Protein Sci. 4:24112423.
[37] Savitzky A, Golay MJE (1964) Smoothing and Di erentiation of Data by Simpli es Least Squares
Procedures. Anal Chem 34:16271639.
[38] Atkins PW Chapter 25: The rates of chemical reactions. In: Physical Chemistry. 5 ed. New York: W.
H. Freeman and Company; 1994. pp. 861897.
[39] Lyubarev AE, Kurganov BI (1998) Modeling of irreversible thermal protein denaturation at varying
temperature. I. The model involving two consecutive irreversible steps. Biochemistry (Mosc) 63:434440.
[40] Kurganov BI, Lyubarev AE, Sanchez-Ruiz JM, Shnyrov VL (1997) Analysis of di erential scanning
calorimetry data for proteins. Criteria of validity of one-step mechanism of irreversible protein denatu-
ration. Biophys. Chem. 69:125135.
[41] Rainville ED, Bedient PE Chapter 2 - Equations of order one. In: Elementary Di erential Equations. 7
ed. New York: Macmillan Publishing Company; pp. 1746.
116 Figures and Figure Legends
Figure 1: A - Structural alignment of WT A1 (1auq, black), type 2B I1309V A1 (1ijk, dark grey) and
type 2B R1306Q A1 extracted from the co-crystal complex with GP1b (1m10, light grey). B - RMSD of
all backbone atoms. Top: R1306Q - WT. Bottom: I1309V - WT. Di raction resolutions are 2:3A for WT,
3:1A for R1306Q and 1:8A for I1309V. Horizontal line indicates residues within the disul de bond.
Figure 2: Platelet pause times and translocation velocities (Inset) for A1-1261 (black circles), A1-1238
(open circles) and A1-1261 in presence of 66  M N-terminal peptide (black squares). The number of
platelets analyzed at each shear rate is in the lower panel. Data are representative of three independent
experiments from di erent donors. Both proteins were immobilized at a concentration of 5M for  2h and
300 M of N-terminal peptide were added to immobilized A1-1261 followed by 1h incubation. N-terminal
peptide was added to 1mL whole blood and bu er to a  nal concentration of 66 M prior to  ow.
Figure 3: A - Far-UV CD spectra at 20 C of A1-1261 (black circles), A1-1238 (open circles) and Inset:
the N-terminal peptide (black squares) and of the scrambled peptide (open squares). B - Urea denaturation
at 25C of A1-1261 (black circles) and of A1-1238 (open circles) via CD at  = 222nm. Insets are the  G0,
m-value and cm for the native to intermediate transition. A1-1261 (black bars) and A1-1238 (white bars).
Figure 4: Scan rate dependency of normalized thermal unfolding transitions by  uorescence ( = 359nm)
of A1-1261 (A) and A1-1238 (B) with excitation  = 280nm. Scan rates were 2:0C=min (black circles),
1:6C=min (open circles), 0:9 C=min (black squares) and 0:4C=min for A1-1261 and 0:6C=min for
A1-1238 (both with open squares). Insets show representative thermal scans and corresponding baselines
before data normalization. Scan rate dependency of thermal unfolding transitions by CD at  = 222nm for
A1-1261 (C ) and A1-1238 (D). Scan rates were 2:0C=min (black circles), 1:5 C=min (open circles),
1:0C=min (black squares) and 0:5 C=min (open squares).
Figure 5: The parameters obtained from  tting the thermal transitions in Figure 4 are reported for each
applied method. A - Scan rate dependence of the T  - values ( lled symbols and black average and
regression lines) and apparent TM - values (open symbols and gray average and regression lines). B - Scan
rate dependence of the enthalpy,  HzT . Lines are solid for A1-1261 and dashed for A1-1238. Symbols are:
FL, A1-1261 (circles); FL, A1-1238 (squares - dotted squares for the second transition); CD, A1-1261
(triangles); CD, A1-1238 (diamonds). Reported in C & D are the average T  and  HzT values  standard
deviation for each method. A1-1261 (black bars); A1-1238 (white bars).
Figure 6: Excess molar heat capacity by DSC at 2:0C=min for A1-1261 (black circles) and A1-1238
(open circles). Insets are the transition temperature, T  , and enthalpy,  HzT , of the native to
intermediate transition. A1-1261 (black bars) and A1-1238 (white bars).
Figure 7: A - Normalized thermal transitions of tryptophan  uorescence ( = 359nm) of 2M A1-1261
with excitation  = 295nm as a function of increasing concentrations of free N-terminal peptide.
Concentrations are 0M (black circles), 0:02M (open circles), 0:2M (black squares), 50M (open
squares), 300 M (black triangles) of the N-terminal peptide and 300 M of the scrambled peptide(open
triangles). The inset shows transitions in presence of 300 M N-terminal peptide (closed triangles) and in
presence of 300M scrambled peptide (open triangles). B - Determination of the binding a nity of the
N-terminal peptide to A1-1261 from  HzT . Symbols represent A1-1261 + N-terminal peptide (black
circles), A1-1261 + scrambled peptide (open circles; average - long dashed line) and A1 1238 + N-terminal
peptide (black squares; average - short dashed line). The lower panel shows T  values for all analzyed
thermalscans. Symbols and lines are homolog to the upper panel.
12A
N
C
Figure 1: A - Structural alignment of WT A1 (1auq, black), type 2B I1309V A1 (1ijk, dark grey) and type
2B R1306Q A1 extracted from the co-crystal complex with GP1b (1m10, light grey). B - RMSD of all
backbone atoms. Top: R1306Q - WT. Bottom: I1309V - WT. Di raction resolutions are 2:3A for WT, 3:1A
for R1306Q and 1:8A for I1309V. Horizontal line indicates residues within the disul de bond.
Figure 2: Platelet pause times and translocation velocities (Inset) for A1-1261 (black circles), A1-1238 (open
circles) and A1-1261 in presence of 66  M N-terminal peptide (black squares). The number of platelets
analyzed at each shear rate is in the lower panel. Data are representative of three independent experiments
from di erent donors. Both proteins were immobilized at a concentration of 5 M for  2h and 300M of
N-terminal peptide were added to immobilized A1-1261 followed by 1h incubation. N-terminal peptide was
added to 1mL whole blood and bu er to a  nal concentration of 66 M prior to  ow.
13Figure 3: A - Far-UV CD spectra at 20C of A1-1261 (black circles), A1-1238 (open circles) and Inset: the
N-terminal peptide (black squares) and of the scrambled peptide (open squares). B - Urea denaturation at
25 C of A1-1261 (black circles) and of A1-1238 (open circles) via CD at  = 222nm. Insets are the  G0,
m-value and cm for the native to intermediate transition. A1-1261 (black bars) and A1-1238 (white bars).
14Figure 4: Scan rate dependency of normalized thermal unfolding transitions by  uorescence ( = 359nm)
of A1-1261 (A) and A1-1238 (B) with excitation  = 280nm. Scan rates were 2:0C=min (black circles),
1:6C=min (open circles), 0:9 C=min (black squares) and 0:4C=min for A1-1261 and 0:6C=min for A1-
1238 (both with open squares). Insets show representative thermal scans and corresponding baselines before
data normalization. Scan rate dependency of thermal unfolding transitions by CD at  = 222nm for A1-
1261 (C ) and A1-1238 (D). Scan rates were 2:0 C=min (black circles), 1:5C=min (open circles), 1:0 C=min
(black squares) and 0:5 C=min (open squares).
15Figure 5: The parameters obtained from  tting the thermal transitions in Figure 4 are reported for each
applied method. A - Scan rate dependence of the T  - values ( lled symbols and black average and regression
lines) and apparent TM - values (open symbols and gray average and regression lines). B - Scan rate
dependence of the enthalpy,  HzT . Lines are solid for A1-1261 and dashed for A1-1238. Symbols are: FL,
A1-1261 (circles); FL, A1-1238 (squares - dotted squares for the second transition); CD, A1-1261 (triangles);
CD, A1-1238 (diamonds). Reported in C & D are the average T  and  HzT values  standard deviation
for each method. A1-1261 (black bars); A1-1238 (white bars).
16Figure 6: Excess molar heat capacity by DSC at 2:0C=min for A1-1261 (black circles) and A1-1238 (open
circles). Insets are the transition temperature, T  , and enthalpy,  HzT , of the native to intermediate
transition. A1-1261 (black bars) and A1-1238 (white bars).
Figure 7: A - Normalized thermal transitions of tryptophan  uorescence ( = 359nm) of 2M A1-1261 with
excitation  = 295nm as a function of increasing concentrations of free N-terminal peptide. Concentrations
are 0 M (black circles), 0:02 M (open circles), 0:2 M (black squares), 50 M (open squares), 300M (black
triangles) of the N-terminal peptide and 300M of the scrambled peptide(open triangles). The inset shows
transitions in presence of 300 M N-terminal peptide (closed triangles) and in presence of 300M scrambled
peptide (open triangles). B - Determination of the binding a nity of the N-terminal peptide to A1-1261
from  HzT . Symbols represent A1-1261 + N-terminal peptide (black circles), A1-1261 + scrambled peptide
(open circles; average - long dashed line) and A1 1238 + N-terminal peptide (black squares; average - short
dashed line). The lower panel shows T  values for all analzyed thermalscans. Symbols and lines are homolog
to the upper panel.
17