|Home | About | Journals | Submit | Contact Us | Français|
Aromatic residues are relatively rare within the collagen triple-helix, but they appear to play a specialized role in higher order structure and function. The role of aromatic amino acids in the self-assembly of triple-helical peptides was investigated in terms of the kinetics of self-association, the nature of aggregated species formed, and the ability of these species to activate platelet aggregation. The presence of aromatic residues on both ends of a type IV collagen model peptide is observed to greatly accelerate the kinetics of self-association, decreasing the lag time and leading to insoluble, well defined linear fibrils as well as small soluble aggregates. Both macroscopic visible aggregates and small multi-molecular complexes in solution are capable of inducing platelet aggregation through the glycoprotein VI receptor on platelets. Proline-aromatic CHπ interactions are often observed within globular proteins and in protein complexes, and examination of molecular packing in the crystal structure of the integrin binding collagen peptide shows Phe interacts with Pro/Hyp in a neighboring triple-helical molecule. An intermolecular interaction between aromatic amino acids and imino acids within the triple-helix is also supported by the observed inhibitory effect of isolated Phe amino acids on the self-association of (Pro-Hyp-Gly)10. Given the high fraction of Pro and Hyp residues on the surface of collagen molecules, it is likely that imino acid-aromatic CHπ interactions are important in formation of higher order structure. It is suggested that the catalysis of type I collagen fibrillogenesis by non-helical telopeptides is due to specific intermolecular CHπ interactions between aromatic residues in the telopeptides and Pro/Hyp residues within the triple-helix.
The amino acid sequence of collagen is uniquely related to its structure and function. The collagen triple-helix is composed of three polypeptide chains, each in an extended polyproline II-like helix which are supercoiled about a common axis (1–3). The primary structure of the collagen molecule has a requirement for Gly as every third residue, (Gly-Xaa-Yaa)n, generated by the close packing of the three chains near the helix axis. The Gly residues are all buried in the center of the triple-helix, while residues in the Xaa and Yaa positions are exposed to solvent. These Xaa and Yaa residues modulate triple-helix stability (4) and determine molecular self-association to higher order structures. A high content of the imino acids Pro and hydroxyproline (Hyp, O) in the Xaa and Yaa positions stabilize the extended polyproline II helix of individual chains, while Hyp confers additional stability through stereoelectronic effects (5) and hydration (2). In addition to these general amino acid features of collagens, the distribution of charged and hydrophobic residues in the Xaa and Yaa positions in fibril forming collagens has been correlated with the staggered arrangement of molecules leading to periodic fibrils (6, 7). Aromatic residues are relatively rare within the collagen triple-helix, but they appear to play a specialized role in collagen structure and function (6,8–11).
The most well characterized collagens form axially periodic fibrils and the most abundant fibrillar collagen is type I, which constitutes the structural backbone of tendon, skin, bone, blood vessels, and cornea (12). The process of fibrillogenesis is well characterized for type I collagen, and the rate of fibrillogenesis increases with increasing temperature, showing a maximum just below the collagen molecular melting temperature (13). Short non-helical peptides, known as telopeptides, flank the long central triple-helix in fibrillar collagens. These telopeptides are unusually rich in Tyr and Phe and have been shown to catalyze the process of fibril formation (9–11). It was noted some years ago that Phe residues within type I collagen are aligned in the gap region of the fibril and it was proposed that Phe-Phe interactions confer rigidity and stability to this more flexible region of the fibril (6, 14)
Collagen model peptides have proved capable of self-assembly into higher order structures and offer an approach to relate amino acid sequence with the self-association of triple-helical molecules (15–23). The process of self-assembly of the model peptide (Pro-Hyp-Gly)10 was found to have many similarities to collagen fibril formation, in terms of its dependence on temperature, pH, and solvent. But the (Pro-Hyp-Gly)10 self-association has a much higher critical concentration than collagen and did not form the highly ordered periodic structures seen for collagen (17). Inclusion of a selected distribution of charged residues in a collagen peptide led to formation of large banded fibrils (21), while inclusion of hydrophobic sequences led to fibrillar structures, some with supercoiled and branching features like those found in basement membrane networks (23). These results suggest triple-helical peptides may have common interactions involving hydration and hydroxyproline (Hyp, O) that promote non-specific self-association, but that charged and hydrophobic interactions confer specificity and a high affinity for self-association. Recently, addition of aromatic residues to the ends of triple-helical peptides has been shown to accelerate the self-association process and to increase the order and size of the final aggregates (15, 16).
Triple-helical peptides have also proved valuable in probing collagen interactions leading to biological activity (8). The information gained on peptide self-association to higher order structure may be used to probe the role of aggregates of triple-helical molecules in biological processes. For example, a non-aggregated triple-helical form of the peptide, e.g. (Pro-Hyp-Gly)10 is not sufficient for activation of platelet aggregation, while a disulfide cross-linked higher molecular form of the same peptide, designated CRP (collagen reactive peptide) is a strong inducer of platelet aggregation (24). CRP binds and activates the same glycoprotein VI (GPVI) tyrosine kinase receptor that responds to collagen, triggering a signal cascade that results in platelet aggregation (25). Recent peptide studies from the Farndale laboratory showed that at least two sequential Gly-Pro-Hyp sequences are necessary to activate platelets and that peptides containing specific type III collagen sequences can bind to platelets and induce aggregation (25, 26). Large fibrils formed from non-cross-linked (Pro-Hyp-Gly)10 peptides capped by aromatic residues on both ends were observed to be good activators of this process (15, 16). The precise nature of the triple-helical peptide aggregate required to induce platelet activation is not well defined.
Here, investigation of the role of aromatic amino acids in the self-assembly of triple-helical peptides is extended to characterize the kinetics of self-association, the nature of aggregated species, and the ability of different species to induce platelet activation. The presence of aromatic residues on both ends of a type IV collagen model peptide is observed to result in marked acceleration of aggregation, and decreased lag time, leading to insoluble, well defined linear fibrils as well as small soluble aggregates. Both macroscopic visible aggregates and small soluble multi-molecular assemblies are capable of inducing platelet aggregation through the glycoprotein VI receptor on platelets. CHπ interactions between imino acids within the triple-helix and aromatic residues are suggested by the inhibitory effect of isolated Phe amino acids on the self-association of (Pro-Hyp-Gly)10 and by an analysis of a high resolution crystal structure of a collagen peptide containing Phe. The relevance of these studies on triple-helical model peptides for collagen fibrillogenesis is considered.
A typical sequence from the α5 chain of type IV collagen (residues 491–499) including Hyp at every Yaa position and several hydrophobic residues was selected as described previously (23). The peptides were synthesized by Tufts University Core facility (Boston, MA): (POG)3QOGLOGLOG(POG)4, denoted as T4; (GPO)3GQOGLOGLO(GPO)4GY, denoted as T4Y; and F(GPO)3GQOGLOGLO(GPO)4GY, denoted as FT4Y. Peptides were purified on a C-18 column using a reverse-phase HPLC system (Shimadzu) and the purity was ensured by mass spectrometry using MALDI-TOF (DE-PRO mass spectrometer). The peptide (Pro-Hyp-Gly)10 was obtained from Peptides International (Louisville, Kentucky, USA). For peptides having Tyr residues, peptide concentration was determined using an extinction coefficient of ε280 nm = 13,980M −1cm1 (27). Concentrations of the peptides without Tyr were measured by monitoring the absorbance at 214 nm using the (ε 214 = 2200 cm−1·M−1 per peptide bond). Buffers used included 20 mM PBS buffer (10 mM NaH2PO4, 10 mM Na2HPO4 and 150 mM NaCl) for pH 7; acetate buffer (20 mM with 150 mM NaCl) for pH 3. Amino acids L-phenylalanine, L-leucine, L-alanine, L-histidine, L-tyrosine were purchased from Sigma.
Turbidity curves monitoring the process of self-assembly collagen peptides were obtained using the optical density at 313 nm as a function of time on a Beckman DU 640 spectrophotometer with a Peltier temperature controller. A peptide solution of 600 µL was kept in a 5 mm cell sealed to avoid evaporation and then subjected to the desired constant temperature.
Circular dichroism (CD) measurements were carried out using an Aviv Model 62DS spectrophotometer (Aviv Biomedical, Inc.). Prior to CD measurements each sample was incubated at 4°C for 2–3 days to allow the formation of triple helix. The characteristic triple-helix CD maximum at 225 nm was used to monitor thermal transitions (28).
Electron microscopy was carried out on negatively stained samples of peptide aggregates to visualize the morphology of the higher order structures. A small aliquot of the sample was placed on a 400-mesh carbon coated copper grid, air dried and stained with uranyl acetate for 10 sec. Specimens were examined by transmission electron microscope using a Phillips 420 instrument.
DLS measurements were performed using a DynaPro Titan (Wyatt Technology Corp., Santa Barbara, CA) equipped with a temperature controller using a 12 µl quartz cuvette. All samples were centrifuged and filtered through 0.1µm Whatman Anotop filters before measurements. To obtain the hydrodynamic radii (Rh), the intensity autocorrelation functions were analyzed by Dynamics software (Wyatt Technology Corp.). For data analysis, a viscosity value of η20 °C = 1.019 centipoise was used for PBS.
Differential scanning calorimetry (DSC) measurements were performed on a Nano-DSC II, Model 6100 scanning calorimeter from Calorimetry Sciences Corp. All DSC profiles were obtained at a scan rate of 1°C/min and each curve was baseline subtracted before the data analysis. Prior to all measurements, peptide solutions were dialyzed. ΔHcal values were obtained by integrating the excess heat capacity curve.
The crystal structure of the GFOGER collagen model peptide, sequence GPOGPOGFOGERGPOGPOGPO, which binds to the α2β1 integrin (PDB entry 1Q7D) (29) was analyzed to characterize molecular packing around Phe 9. Structures in the unit cell were generated using the ‘Build Crystallographic Symmetry’ function of DeepView v4.0 (www.expasy.org/spdbv), using the C2221 space group.
Apyrase (type VII), Indomethacin, P2Y1 antagonist MRS 2179 and P2Y12 antagonist AR-C69931 were obtained from Sigma (St. Louis, MO). Convulxin was purchased from Centerchem (Norwalk, CT). Primary antibody Syk (4D10) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA) and phospho-Syk Tyr525/526 from Cell Signaling Technologies (Beverly, MA). LI-COR Odyssey blocking buffer and secondary antibodies goat anti-rabbit IRDye 800CW and goat anti-mouse IRDye 680 were purchased from LI-COR Biosciences (Lincoln, NE). Whatman protran nitrocellulose transfer membrane was obtained from Fisher Scientific (Pittsburgh, PA).
Whole blood was drawn from healthy, consenting human volunteers into tubes containing one-sixth volume of ACD (2.5g sodium citrate, 1.5g citric acid and 2g glucose in 100 ml of deionized water) and centrifuged at 230 g for 20 min at room temperature to obtain PRP (platelet-rich plasma). The PRP was then centrifuged at 980Xg for 10 min at room temperature to pellet the platelets. Platelets were resuspended in Tyrode’s buffer (138 mM NaCl, 2.7 mM KCl, 1mM MgCl2, 3 mM NaH2PO4, 5 mM glucose, 10 mM Hepes at pH 7.4) and containing 0.2 units/ml of apyrase. Cells were counted using a Coulter Z1 Particle Counter and the concentration of cells was adjusted to 2 × 108 platelets/ml. All experiments using washed platelets were performed in the absence of extracellular calcium.
Aggregation of 0.5 ml of washed platelets was analyzed using a P.I.C.A. lumi aggregometer (Chrono-log). Platelets were activated with varying concentrations of collagen peptides or 100ng of convulxin. To examine the effect of peptides in their associated forms, the peptides (Pro-Hyp-Gly)10 (7mg/ml), T4 (7mg/ml), T4Y(7mg/ml), and FT4Y (3mg/ml) were incubated at their respective optimal temperatures to promote self-association. After turbidity reached the plateau phase, aggregated peptide samples were incubated in a platelet activation assay at 37°C. The dose dependent platelet activation by FT4Y peptide (3 mg/ml, PBS) was measured by adding different volumes (1µl to 35 µl) of the turbid peptide sample of FT4Y to 500µl of platelet suspension and measuring aggregation using light transmission while stirring (900 rev/min) at 37°C. Aggregation tracings are representative of results obtained from three separate experiments on three different donors. Experiments are also carried out in presence of P2Y1 receptor antagonists MRS 2179, and P2Y12 receptor agonist AR-C69931, and indomethacin which were added just prior to the addition of agonist.
For western blot analysis, platelets were stimulated for 30 seconds under stirring conditions with varying concentrations of FT4Y peptide or 100ng of convulxin as a control. The reaction was stopped by the addition of 6.6N perchloric acid and the resulting acid precipitate was collected and chilled on ice. The pellets were centrifuged at 13,000Xg for 10 minutes, followed by rinsing and subsequent resuspension in 0.5 ml of deionized water. The protein was again pelleted by centrifugation at 13,000Xg for 10 minutes. The protein pellets were solubilized in sample buffer containing 2M Tris, 10% SDS, (v/v) glycerol, 0.5% bromophenol blue, and 100mM DTT then boiled for 10 minutes. Bovine skin collagen was also used as a control.
Samples were subjected to SDS-PAGE on 10% polyacrylamide gels. Proteins were transferred to Whatman Protran nitrocellulose membrane, blocked with Odyssey blocking buffer for 1 hour and incubated overnight at 4°C with primary antibody anti-Syk (4D10) and anti-phospho-Syk525/526 (1:1000 in Odyssey blocking buffer) with gentle agitation. After 4 washes for 5 minutes each with PBS-T, the membranes were probed with Li-Cor Odyssey goat anti-rabbit IRDye 800CW and goat anti-mouse IRDye 680 (1:1000 in PBS-T) for 1 hour at room temperature(23°C). After 4 washes for 5 minutes each with PBS-T and 1 wash in deionized water, membranes were examined under Li-Cor Odyssey infrared Imaging System (Lincoln, NE).
The effect of having aromatic residues on none, one or both ends of a collagen model peptide was investigated in terms of kinetics of self-association, the aggregated species formed, and the ability of all species to activate platelet aggregation. A collagen model peptide denoted as T4 contains a central hydrophobic sequence GPOGQOGLOGLOGPO from the α5 chain of type IV collagen (residues 491–499) flanked by (Gly-Pro-Hyp) triplets. The peptide was previously studied with no aromatic residues and with a C-terminal Tyr residue (23), and here these are compared with a new construct capped by an N-terminal Phe residue as well as a C-terminal Tyr, designated as peptide FT4Y (Table 1). CD spectroscopy indicates that peptide FT4Y forms a triple helix with a thermal stability of Tm= 48 °C which is higher than seen for the homologous peptide with only a C-terminal Tyr (Tm=44 °C) or no aromatic residues (Tm=41°C). Differential scanning calorimetry (DSC) of FT4Y at 1mg/ml shows a single thermal transition at a higher temperature, 55.5°C, consistent with the faster heating rate for the DSC (1°C/min) compared with CD (0.1°C/min) under non-equilibrium conditions (28). The FT4Y calorimetric enthalpy is 284 kJ·mol−1, compared to 287 kJ·mol−1 and 259 kJ·mol−1 for T4Y and T4 respectively at 1mg/ml, suggesting similar hydrogen bonding and hydration (Table 1).
The presence of a single tyrosine residue at the C-terminal end of the triple helix was previously shown to greatly accelerate peptide self-association (23), with t1/2 decreasing from ~ 24 hr for T4 to ~ 3 minutes for T4Y (c = 7mg/ml, ~2°C below Tm). The additional N-terminal Phe in FT4Y increased the rate of self-assembly so much that aggregates were formed before the sample could be monitored by turbidity at c=7mg/ml (Table 1). When the FT4Y concentration was reduced to 3mg/ml, lag, growth and plateau phases were observed (Figure 1a), while neither T4Y nor T4 show any aggregation at this concentration within 48 hr. The self-association of FT4Y (c=3mg/ml) was temperature dependent, with an optimum near T=46°C (Figure 1b), consistent with previous observations that the optimal rate of peptide self-association is several degrees below its Tm value (17, 23). Centrifugation of the suspension indicates a critical concentration of ~1.4 mg/ml for FT4Y, a value lower than seen for T4Y (3.4mg/ml) and T4 (5.9mg/ml) (Table 1). Under aggregating conditions (3mg/ml, PBS pH 7), the DSC profile of FT4Y shows two distinct peaks (Figure 1c). The first transition at 55.5°C corresponds to melting of triple helical molecules into unfolded chains, while the second smaller transition at 88.7°C corresponds to a loss of turbidity and is likely to reflect dissociation of higher order structures formed during the DSC scan (17).
To visualize the structures formed as result of self-association, negatively stained samples of FT4Y were observed by electron microscopy (Figure 2). The micrographs show linear fibrillar structures with diameters ranging from 20–40 nm. In some cases it appears that two fibrillar units are twisted around each other. There is no indication of axial banding. These structures looked similar to the higher order structures reported previously for T4Y and T4 (23).
Dynamic Light Scattering (DLS) studies were carried out on peptides T4, T4Y and FT4Y to determine translational diffusion constants and characterize homogeneity and aggregation (Figure 3). At low temperature (c=7mg/ml), peptide T4 shows only one peak with a hydrodynamic radius Rh~2nm, a value close to that reported previously for a single trimer molecular species of (Pro-Hyp-Gly)10 and other triple-helical peptides of similar length (17) (Table 1). The presence of an aromatic residue on one or both ends of the T4 peptide is observed to lead to formation of soluble multimolecular species. The DLS profiles of peptides T4Y and FT4Y at low temperature (c=1mg/ml and 3mg/ml) show a higher molecular weight peak with Rh ~32nm in addition to the single triple-helical molecular species with Rh ~2.1 and ~2.5nm respectively. Increasing temperature leads to a decrease in the intensity of the 2.5nm peak and corresponding increase in the intensity of the ~32nm peak, until the temperature reaches ~46°C where turbidity and visible aggregation prevent collection of DLS data (Figure 3e). When the 46°C aggregated sample is cooled back to 4°C, the visible aggregates disappear and become completely soluble, while the trimer species (Rh~2.5nm) and the larger molecular species (Rh~32nm) are present again, indicating reversibility of the association process.
The peptides (Pro-Hyp-Gly)10, T4, T4Y and FT4Y in soluble and aggregated states were tested for their ability to induce platelet aggregation. There was no platelet activation by either the soluble form of (Pro-Hyp-Gly)10 or by its temperature induced clump-like aggregated structure. The fibrous forms of peptides T4, T4Y and FT4Y and the soluble forms of FT4Y and T4Y which contained multi-molecular species did induce activation of platelet aggregation, displaying activation similar to that of a convulxin control and bovine skin collagen (data not shown). In contrast, the soluble form of peptide T4 with only trimer species did not induce activation. The effect of FT4Y aggregates on platelet functional responses was characterized in detail. The visible aggregates of FT4Y activated platelets in a concentration-dependent manner. At low concentrations (3µg/ml) FT4Y peptide induced platelet shape change, while increasing concentrations caused full platelet aggregation (Figure 4). In the presence of the thromboxane inhibitor indomethacin, and ADP receptor antagonists ARC-69931MX, and MRS 2179 (30), platelets failed to aggregate in response to FT4Y, indicating a dependence of the process on positive feedback from thromboxane and secreted ADP. The ability of FT4Y to activate GPVI pathways was supported by the concentration dependent phosphorylation of Syk on tyrosine residues 525/526 (31, 32) (Figure 4b).
To further investigate the interactions between aromatic residues and triple-helices, the effect of isolated Phe residues on the aggregation process of collagen model peptides was characterized. Monitoring the self-association of (Pro-Hyp-Gly)10 (c=7mg/ml; 58°C) by turbidity indicates the addition of Phe (molar ratio of peptide:Phe = 1:60) leads to a significantly longer lag time, a decreased rate of growth and a decreased magnitude of the maximal height in the plateau phase (Figure 5a). Addition of isolated Leu or Ala amino acids show no effect on self-assembly of (Pro-Hyp-Gly)10 at same concentrations, indicating the importance of the aromatic nature of the amino acid (Figure 5a). The presence of Phe also led to a substantial delay in self-association for peptides T4Y (c=7mg/ml; 44°C) (Figure 5b), and for FT4Y (c=3mg/ml; 46°C) (data not shown).
Since Phe interfered with the self-assembly of (Pro-Hyp-Gly)10 and other triple-helical peptides, the possibility of direct binding of the isolated amino acid Phe to (Pro-Hyp-Gly)10 was investigated. Isothermal titration calorimetry experiments were not successful due to strong signals from Phe residues in solution, which could result from aromatic stacking or solvation. Differential scanning calorimetry (DSC) measurements of (Pro-Hyp-Gly)10 (PBS, pH 7, c=1mg/ml) showed a slight destabilization in the presence of high concentrations of Phe (molar ratio of peptide:Phe = 1:200) (Figure 5c).
The interactions of Phe residues in triple-helical molecules were explored using the high resolution structure of the collagen model peptide that binds α2β1 integrin (pdb 1Q7D) (29). This peptide sequence contains one Phe residue at position 9 in the middle of the triple-helix (GPOGPOGFOGERGPOGPOGPO) and the crystal packing shows a distorted hexagonal arrangement of triple-helices with an average axis-to-axis distance of 14Å. The Phe residue, in the Xaa position of the triple-helix, does not interact with any residues within the same molecule, but examination of Phe environments in the unit cell shows a significant number of intermolecular contacts (Figure 6a). CHπ interactions between Phe and imino acids are observed between an antiparallel pair of triple helices (Figure 6b). The Cδ of Hyp 16 together with the Cα and the Cβ of Pro 15 participate in a network of CHπ hydrogen bonds with Phe 9 across a crystal packing interface. The distance and angles between the Phe and the atoms of Pro 15 Cβ and Pro 15 Cα satisfy the criteria defined by Brandl et al (2001)(33) for CHπ hydrogen bonds (dCβR < 4.5 Å; θCδ-HR between 120° – 180°), while the Hyp 16 Cδ interaction falls slightly outside the angle constraint. At the parallel crystal packing interface between a pair of triple helices in the GFOGER peptide unit cell, Phe 9 on one triple-helix molecule is in the vicinity of an equivalent Phe on another molecule, but the aromatic ring center-to-center distance of 7.6 Å precludes any possibility for π-π stacking or other commonly observed aromatic-aromatic interaction motifs (34). In this parallel pair of triple-helices, the Phe does make an interchain edge-to-edge contact with Hyp 10, but this does not satisfy the geometric criteria for a CHπ hydrogen bond (33).
The studies reported here provide new information on interactions of aromatic amino acids in the context of collagen model peptides, showing that aromatic residues play a role in accelerating the aggregation of triple-helical peptides and affecting their biological activity. The addition of aromatic residues on both ends of a collagen model peptide greatly accelerated the association to higher order structures as well as affecting morphology. Recent studies by Cejas et al. (15,16) suggested these effects are mediated largely through end to end aromatic-aromatic interactions, but it is also likely that aromatic residues in one molecule participate in CHπ interactions with imino acids within the triple-helix region of another molecule. The probability of such aromatic-Pro CHπ interactions is supported by the extensive literature on high resolution globular protein structures, the analysis of the molecular packing in the crystal structure of a triple-helical peptide structure containing a Phe residue, and experimental evidence presented here.
There are detailed analyses of protein structures documenting interactions of aromatic amino acids with Pro residues in proteins, peptides and complexes (33, 35, 36). The Cα and Cδ carbons of the Pro ring are adjacent to the backbone amide, making them more acidic and potent hydrogen bond donors. Protons donated by Pro can interact with electron rich aromatic acceptors such as Phe, Tyr and Trp, leading to CHπ interactions. These CHπ interactions are weaker than those involving strong electron withdrawing donors such as nitrogen or oxygen, but have been shown to contribute 0.5–1.0 kcal/mol to the stability and can play a role in protein folding and function (33, 35, 36). Hydroxyproline residues would be expected to exhibit similar stabilizing interactions. Although the frequency of these interactions observed in globular proteins increases when the Pro and aromatic residues are close in sequence, there are also well documented examples in complexes where the Pro is in one polypeptide chain and the aromatic amino acid in another. The high content of both Pro and Hyp on the surface of the collagen triple-helix would favor their interactions with available Phe residues in adjacent molecules.
The crystal packing of the collagen integrin binding peptide containing a GFOGER sequence (29) provides visualization of molecular details of CHπ intermolecular interactions between an aromatic amino acid in one molecule and imino acids in a neighboring molecule within a triple-helix peptide context. Although it is not possible to draw general conclusions from a single crystal structure, such favorable intermolecular CHπ interactions should be considered when collagen molecules interact with themselves or other proteins containing aromatic residues.
The experimental studies reported here are consistent with a role for CHπ interactions in the self-association of triple-helical peptides. The increased rate of aggregation and fibril formation of a model triple-helical peptide T4 when there is a Tyr on one end suggests that Tyr may be interacting with Pro/Hyp residues within neighboring triple-helical molecules. Since there are many Pro and Hyp residues in similar sequences within the T4 peptide triple-helix domain, favorable CHπ interactions between aromatic residues in the terminus and Pro/Hyp residues in the triple-helix domain of another molecule would be expected to be non-specific, so the generation of fibrils with no axial periodicity is not surprising (Figure 7a). Such non-specific CHπ interactions would still favor nucleation leading to the observed significant reduction in the lag time of self-assembly and lower critical concentration when one aromatic residue terminates a peptide T4Y. The dramatic acceleration of the rate of self-association when there are aromatic residues on both ends FT4Y suggests a synergistic effect of π-π interactions involving aromatic residues on the peptide ends and the CHπ interactions between a terminal aromatic residue and imino acids within the neighboring triple-helix (Figure 7b). The interactions between aromatic residues at the ends of different molecules could lead to long linear fibrils, while non-specific CHπ interactions could promote lateral aggregation.
The presence of CHπ interactions between imino acids in the triple-helix and aromatic residues is supported by the ability of isolated Phe residues, but not hydrophobic residues, to inhibit the self-association of (Pro-Hyp-Gly)10. A strong interaction between Phe and the triple-helical form of the peptide would be expected to lead to an increased stability, but in fact a small destabilization is observed by DSC when Phe is added to (Pro-Hyp-Gly)10. It is possible that Phe is binding to the unfolded as well as the triple-helical state, with somewhat stronger binding to the unfolded peptide. Alternatively, Phe binding could lead to dehydration which would be expected to lower triple-helix stability. It is worth noting that earlier studies suggest the presence of Tyr as a terminal residue can drive the self-assembly process only if the sequence has the intrinsic ability for aggregation process (23).
The addition of aromatic residues to the T4 peptide modulated molecular association into higher order structures which influenced their ability to induce platelet aggregation. The ability of soluble and insoluble forms of the peptides FT4Y and T4Y peptides to activate platelets in a dose dependent manner appears to be related to the presence of a soluble species that is of a larger hydrodynamic radius (Rh ~32nm) than the single molecule (Rh~2.5nm). Collagen mediated platelet aggregation is activated through the GPVI receptor, triggering Syk phosphorylation on tyrosine residues 525/526 which has been shown to be important for its kinase activity (31, 32). This process is also dependent on positive feedback from thromboxane and secreted ADP. The ability of the FT4Y peptide higher molecular weight forms to lead to Syk phosphorylation and inhibition by thrombaxane inhibitors and ADP receptor antagonists suggests it is acting through the GPVI mediated pathway. These results suggests even a small aggregate of triple-helical molecules may suffice to interact with GPVI, while a single trimer is not sufficient. The structure of the associated species is important since the aggregated form of (Pro-Hyp-Gly)10, which has a clump-like rather than fibrous appearance has no activity.
The intermolecular aromatic interactions studied in peptides may be relevant to collagen self-association to higher order structures in tissues. There are few aromatic residues within the (Gly-Xaa-Yaa)n domain of type I collagen (1.1 %), and these are almost exclusively Phe residues, which are found more than 50% of the time in Gly-Phe-Hyp triplets. Fraser et al. (6, 14) reported that a set of Phe residues in type I collagen are concentrated in a short segment in the gap region of D periodic fibrils and suggested this may constitute a stable site within the otherwise flexible gap region. Examination of the sequences of the three major fibril forming collagens, types I, II and III, shows that almost all of the Phe residues within the triple-helix as well as the aromatic residues within the telopeptides align with a 234 residues periodicity within the collagen molecule in the overlap as well as the gap region, leading to the formation of 3–4 bands of aromatic amino acids across D=670Å periodic fibrils (Figure S1 of the Supporting Information). The D-periodic distribution of Phe residues in fibrillar collagens could lead to π-π interactions between Phe residues in adjacent molecules within the fibril, but the high frequency of Gly-Phe-Hyp triplets and the proximity of Pro to Phe residues suggests an alternative stabilizing mechanism involving intermolecular Phe-imino contacts such as seen in the crystal packing of the integrin binding protein. The intermolecular distances and packing symmetry in the integrin binding protein crystal structure resembles collagen molecular packing in fibrils in rat tail tendon as measured by fiber diffraction (2, 6), suggesting that molecular interactions identified in the peptide crystal structure may be biologically relevant.
Fibril forming collagens contain non-helical terminal telopeptides which are rich in aromatic amino acids, and the Phe and Tyr residues in telopeptides are highly conserved. Many reports have shown that the telopeptides catalyze collagen fibril formation (9–11). Synthetic peptides that contain telopeptide sequences inhibit collagen fibrillogenesis, and the aromatic residues in the α2(I) C-telopeptide were shown to be essential for this inhibitory effect (11). Prockop and Fertala et al. (1998) suggest the telopeptides nucleate self-association by binding to a specific region of the central triple-helix and that the aromatic residues are essential for this binding. By analogy to the studies on triple-helical peptides, we suggest that favorable CHπ interactions between aromatic residues in the telopeptides and Pro/Hyp residues in the triple-helix domain of another molecule could catalyze self-association and fibril formation. In the collagen case, the aromatic residues must interact with a specific site in the adjacent molecule to promote a staggered periodic molecular arrangement, as proposed by Prockop and Fertala (1998). Such CHπ interactions would favor nucleation leading to the observed significant reduction in the lag time of self-assembly and lower critical concentration for acid extracted collagen when the telopeptides are present compared with pepsin extracted collagen. The specific nature of the telopeptide interaction with the collagen triple-helix would also lead to a much lower critical concentration for collagen fibril formation compared with peptide self-association. Consistent with this hypothesis is our observation that addition of Phe (but not Leu or Ala) inhibited fibril formation of pepsin extracted collagen (pers. obs.). Less consistent effects were seen for salt extracted collagen which still retains telopeptides.
The sequence incorporated into the T4 model peptide comes from type IV collagen, where there is a high percentage of Phe residues present largely in Gly-Phe-Hyp triplets as well as in interruptions between the (Gly-Xaa-Yaa)n sequences. It is possible that aromatic-imino acid interactions play a role in the self-association of type IV collagen to the network structure found in basement membranes. Studies also indicate that basement membrane collagen can activate platelet aggregation, and the ability of these type IV model peptides to induce platelet activation could be of physiological significance (37, 38).
Interactions between aromatic residues and Hyp and/or Pro should be considered as factors in promoting self-association of triple-helices, in addition to the previously identified roles of the hydration network, hydrogen bonding, and Hyp-mediated interactions (17, 39–41). An understanding of the role of aromatic-imino acid interactions in the self-association of triple-helical molecules may provide a tool to understand the higher order structure requirement of collagen activated biological processes such as platelet activation as well as for designing biomaterials and tissue engineering scaffolds.
We thank Dr. John Ramsaw for helpful discussions. SI was supported by the RISE (Research In Science and Engineering) at Rutgers/UMDNJ summer program, jointly sponsored by the Graduate School of Biomedical Sciences at RWJMS and the Rutgers Graduate School-New Brunswick
This work was supported by NIH grant GM60048 (BB) and NIH grant HL81321 (SPK).
Schematic diagram of the amino acid sequences of the three major fibril forming collagens, type I, II, and III, with a 67nm=234 residue stagger, showing the alignment of aromatic residues and proximity of imino acids (Figure S1). This material is available free of charge via the Internet at http://pubs.acs.org.