Self-aggregation of fibrillar collagens I and II involves lysine side chains

Micron 37 (2006) 640–647 www.elsevier.com/locate/micron

Self-aggregation of fibrillar collagens I and II involves lysine side chains Ruggero Tenni a, Myriam Sonaggere a, Manuela Viola a, Barbara Bartolini a, M. Enrica Tira a, Antonio Rossi a, Ester Orsini b, Alessandro Ruggeri b, Vittoria Ottani b,* b

a Dipartimento di Biochimica ‘‘A. Castellani’’, University of Pavia, via Taramelli 3b, 27100 Pavia, Italy Dipartimento di Scienze Anatomiche Umane e Fisiopatologia dell’Apparato Locomotore, University of Bologna, via Irnerio 48, 40126 Bologna, Italy

Received 11 November 2005; received in revised form 31 January 2006; accepted 31 January 2006

Abstract Several properties of fibrillar collagens depend on abundance and position of ionic amino acids. We recently demonstrated that N-methylation and N-acetylation of Lys/Hyl amino group did not significantly alter the thermal stability of the triple helical conformation and that the binding of modified collagens I and II to decorin is lost only on N-acetylation. The positive charge at physiological pH of Lys/Hyl side chains is preserved only by N-methylation. We report here the new aspect of the influence of the same modifications on collagen self-aggregation in neutral conditions. Three collagen preparations are very differently affected by N-methylation: acid-soluble type I collagen maintains the ability to form banded fibrils with 67-nm periodicity, whereas almost no structured aggregates were detected for pepsin-soluble type I collagen; pepsin-soluble type II collagen forms a very different supramolecular species, known as segment long spacing (SLS). N-acetylation blocks the formation of banded fibrils in neutral conditions (as did all other chemical modifications reported in the literature), demonstrating that the positive charge of Lys/Hyl amino groups is essential for self-aggregation. Kinetic measurements by turbidimetry showed a sizeable increase of absorbance only for the two Nmethylated samples forming specific supramolecular aggregates; however, the derivatization affects aggregation kinetics by increasing lag time and decreasing maximum slope of absorbance variation, and lowers aggregation competency. We discuss that the effects of N-methylation on selfaggregation are caused by fewer or weaker salt bridges and by decrease of hydrogen bonding potential and conclude that protonated Lys side chains are involved in the fibril formation process. # 2006 Elsevier Ltd. All rights reserved. Keywords: Collagen; Fibril formation; Fibril morphology; Protein modification; Lysine; Aggregates stability

1. Introduction Collagens are related proteins sharing localization in connective tissues and a characteristic conformation, the triple helix, assembled from specific polypeptide chains (alpha chains) with the Gly–X–Y repeat. Different classes of collagens are known, as a function of the molecular species and of the presence in the different supramolecular assemblies (Brodsky and Shah, 1995; van der Rest and Garrone, 1991). Fibrillar collagens (types I, II, III, V and XI) are formed by three a chains without any interruption of the Gly–X–Y repeat and contain abundant ionic amino acids, 25% of all residues

Abbreviations: ASC I and PSC I, acid-soluble and pepsin-soluble type I collagen; PSC II, pepsin-soluble type II collagen; N0, Nm and Na, unmodified, N-methylated and N-acetylated samples; FLS, fibrous long spacing; SLS, segment long spacing; TEM, transmission electron microscopy * Corresponding author. Tel.: +39 051 2091552; fax: +39 051 2091659. E-mail address: [email protected] (V. Ottani). 0968-4328/$ – see front matter # 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.micron.2006.01.011

in the X and Y position. Empirical findings evidence the preferential position X for Glu and position Y for Lys and Arg (Hofmann et al., 1980 and Swiss Prot); ionic amino acids are mostly in the triplets occurring with high frequency, i.e. in collagen typical tripeptide units. This constitutes a substantial deviation from a random sequence, as is the fact that clusters of both charged and uncharged typical tripeptides are present along the triple helical domain (for the a1(I) chain of collagen I, (Do¨lz and Heidemann, 1986)). These facts suggest that the abundant presence and the position of acidic and basic residues in triple helices and their aggregates finely tune collagen biochemistry and biology. Ionic residues modulate the stability of the triple helical conformation, as reported in recent studies on model peptides (Persikov et al., 2000, 2002, 2005). The presence of ionic residues in recognition sites along collagen molecules may play a role also in fibril formation and in intermolecular interactions with other connective tissue macromolecules or cell receptors. For example, the self-assembly of collagen I to fibrils involves

R. Tenni et al. / Micron 37 (2006) 640–647

641

the recognition of the a2(I) C-telopeptide with positive residue(s) in the triple helical region of the a1(I) chain (Prockop and Fertala, 1998); the amino group of some Lys/Hyl residues may be part of the binding sites for decorin (Tenni et al., 2002); several integrins recognize the sequence GFOGER present in some collagen types (Xu et al., 2000; Zhang et al., 2003). Worth of mention for fibrillar collagens is the fact that hydrophobic interactions and electrostatic interactions between cationic and anionic amino acids bring a substantial contribute to the quartered-stagger of molecules and periodicity in collagens fibrils and fibers (Hulmes et al., 1973). Chemical modifications of ionic residues are expected to influence collagen biophysical and biochemical properties. Several derivatizations have been described in the literature and modified collagens were useful tools to study collagen properties, such as: thermal stability of the triple helical conformation, maturation of collagen reducible cross-links, supramolecular assembly, susceptibility to proteolysis, interactions with chaperones or components of the extracellular matrix or cell receptors (Balleisen et al., 1976; Davis et al., 1975; Diamond et al., 1991; Fitzsimmons et al., 1988; Giudici et al., 2003; Gratzer et al., 2004; Koide et al., 2002; Rauterberg and Ku¨hn, 1968; Thomson et al., 2003; Wang et al., 1978; Wen Hu et al., 1996; Wilner et al., 1971). Several past derivatization protocols, however, involved rather harsh conditions, such as very high or very low pH, where artifacts, non specific modifications and denaturation may occur. In previous work, we have performed in mild conditions two different chemical derivatizations on fibrillar collagens, viz. Nmethylation and N-acetylation of Lys/Hyl side chains. The results showed that both derivatizations did not significantly alter the thermal stability of collagen triple helical conformation (Giudici et al., 2003), and that there was a diversified effect on the interaction with decorin, N-acetylation causing the loss of the binding and N-methylation preserving it (Tenni et al., 2002). Since ionic groups play a role in fibril formation, we considered worth to determine if fibrillar collagens derivatized with the same protocols maintain in vitro and in neutral conditions the ability to recognize self and form aggregates of definite and identifiable morphology. The results show that Nmethylated acid-soluble collagen I and pepsin-soluble collagen II maintain such an ability; this positive outcome is not shared by any other chemical derivatization described so far. In addition, N-methylation affects self-aggregation rate and determines the nature of fibrils.

collected by centrifugation, redissolved in 0.5 M acetic acid and subjected to a second NaCl precipitation. The final precipitate, collected by centrifugation, was dialyzed exhaustively against 0.1 M acetic acid and freeze-dried (obtaining ASC I). The skin residue after acetic acid extraction was suspended in 0.5 M acetic acid and treated twice at 4 8C with pepsin (0.1 mg/ml). After centrifugation, collagen in the supernatants was subjected to two NaCl precipitations, dialysis and freeze-drying as above (obtaining PSC I). Pepsinized type II collagen was purified from bovine nasal septum (Reese and Mayne, 1981). Briefly, the tissue was extracted at 4 8C for 24 h with 4 M guanidinium hydrochloride in Tris–HCl, pH 7.4, in the presence of protease inhibitors. The residue was washed with water and resuspended at 4 8C for 48 h in 0.5 M acetic acid containing 1 mg/ml pepsin and 0.2 M NaCl. The solubilized material was dialyzed against 0.9 M NaCl in 0.5 M acetic acid and the precipitate of type II collagen collected by centrifugation, dialyzed against 0.1 M acetic acid and freeze-dried (obtaining PSC II). Two different chemical modifications have been performed on collagen preparations, both modifying the e-amino group of Lys/Hyl side chains: N-methylation with formaldehyde in the presence of NaBH3CN as reducing agent transforms the primary amino group to a tertiary one (i.e., at physiological pH, R–NH3+ groups are converted to R–NH(CH3)2+); N-acetylation with sulfosuccinimidyl acetate creates an amido bond (i.e., at physiological pH, R–NH3+ groups are converted to R–NH– CO–CH3). Therefore, the positive charge is maintained after Nmethylation only. Both derivatizations were performed as described in a previous publication and were performed in mild conditions that do not denature the samples (Giudici et al., 2003). All collagenous samples were analyzed by means of a quantitative Hyp assay (Huszar et al., 1980) after acid hydrolysis in 6 M HCl at 106–108 8C for 22–24 h, by gel electrophoresis in denaturing conditions and by CD spectroscopy (the latter method being able to discriminate if the collagens samples are in native triple-helical conformation or instead are in random coil conformation). The degree of Lys/ Hyl modification was determined by a colorimetric method involving sodium trinitrobenzenesulfonate, essentially as described (Kakade and Liener, 1969), using Na-acetyl-L-lysine as standard.

2. Materials and methods

Underivatized and N-methylated collagen samples were dissolved at 1 mg/ml in 5 mM acetic acid at 4 8C. After degassing all solutions on ice for 5 min, aliquots of 100 ml were mixed with 400 ml of a 1.25  Na phosphate–NaCl buffer. After mixing, concentrations were 0.2 mg/ml collagen, 5– 10 mM phosphate, 65–130 mM NaCl; pH was 7.2–7.4. The ratio dihydrogen phosphate/monohydrogen phosphate was 1:11.5 for 5 mM total phosphate and 1:9 for 10 mM total phosphate. N-acetylated samples were insoluble in acetic acid and therefore were dissolved directly in the final solution containing phosphate buffer, NaCl and acetic acid, as above.

2.1. Collagens preparations and modifications Calf skin was used to sequentially extract acid-soluble type I collagen (Rossi et al., 1996) and pepsin-soluble type I collagen (Sykes, 1976). Briefly, delipidized and dried calf skin was extracted with 0.15 M NaCl, 50 mM Tris–HCl, pH 7.4, containing protease inhibitors. The residue was extracted twice with 0.5 M acetic acid. The material dissolved by the acidic solvent was precipitated with NaCl (2 M final concentration),

2.2. Fibril forming kinetics

642

R. Tenni et al. / Micron 37 (2006) 640–647

Aggregation kinetic was followed for some samples also at an acidic pH (pH 4.8). Underivatized and N-methylated collagen samples were dissolved overnight in 50 mM acetic acid at 0.8 mg/ml; after clarification by centrifugation, 1 vol of collagen solution was mixed with 1 vol of water and 2 vol of PBS 2; final concentrations were 0.2 mg collagen/ml in PBS 1 and 12.5 mM acetic acid. N-acetylated samples were dissolved at 1 mg/ml in PBS; after centrifugation as above, 1 vol of collagen solution was mixed with 2 vol of PBS 2, 0.75 vol of water and finally 1.25 vol of 50 mM acetic acid; final concentrations were the same as above. Fibril formation was determined with a spectrophotometer as the turbidimetric increase of absorbance at 313 nm and 32 8C immediately after degassing and mixing the appropriate solutions. Turbidimetric measurement of each derivatized sample was always determined in parallel with its unmodified parent collagen. At the end of the experiment, the mixture was spun down in a microcentrifuge (15 min, 12.000 rpm, 4 8C) and collagen in the supernatant was measured by a quantitative Hyp assay (Huszar et al., 1980); the percentage of sample competent to form insoluble aggregates can thus be calculated. 2.3. Electron microscopy Samples of collagen solutions were prepared at neutral pH as reported for kinetic studies; final concentration of collagen samples were 0.2 mg/ml in 5 mM sodium phosphate, 65 mM NaCl, pH 7.2–7.4. The samples were incubated at 32 8C for 300 min or overnight and then one drop of each collagen solution and buffer was transferred to formwar-coated 200mesh copper grids and the material was allowed to settle for 1 h. The buffer was then drained cautiously with filter paper. Negative staining was carried out by dripping 1% phosphotungstic acid (pH adjusted to 7.3 with NaOH). Most of the mixture was then gently blotted off with filter paper and the remainder left to dry without washing. The grids were then observed under a Siemens Elmiskop 101 Transmission Elettron Microscope and micrographs with an instrumental magnification of 52000 and 105000 were obtained. Morphometrical evaluation of fibril period was made on images showing the best band-pattern resolutions for underivatized ASC I and its Nmethylated derivative; 30 meaurements of fibril period were performed on both samples using a computerized image analysis system (QWin, Leika Microsystem Imaging Solution Ltd, Cambridge, UK 2000). The mean period values differences was analyzed for statistical significance. 3. Results 3.1. Characterization of unmodified and derivatized collagen samples Collagen samples were analyzed by SDS-PAGE and CD spectroscopy; the results have been published (Giudici et al., 2003; Tenni et al., 2002); in particular, CD spectroscopy showed that N-methylation and N-acetylation do not denature the samples.

The protocols of derivatizations were finalized to the highest modification degree, as a consequence of the high reagent/ substrate ratio and the reaction time. The percentage of derivatization of both N-methylation and N-acetylation of Lys/ Hyl side chains was found to be greater than 80% for most samples, because of the high reagent/substrate ratio and the long time of reaction. The percentage of modification will be specified below for each modified sample used. Probably, modifications occur mostly randomly along the collagen molecules. In no occasion we mixed modified samples with underivatized ones. The relative amount of N-monomethyl and N-dimethyl derivatives in N-methylated samples was not determined; however, the latter derivatives are preferentially formed because the second methylation step is faster than the first one (Jentoft and Dearborn, 1983). 3.2. Aggregation kinetics of N-methylated collagen samples The kinetic experiments of ‘‘fibrillogenesis’’ on chemically modified collagens were initially performed at 32 8C at a collagen concentration of 0.2 mg/ml in a buffer of physiological ionic strength (10 mM sodium phosphate, 130 mM NaCl, pH 7.4–7.5). A sample of N-methylated ASC I (with 88% modification) failed to aggregate in these conditions as did in other neutral buffers containing 10 mM phosphate and 100– 120 mM NaCl, pH 7.4–7.5. This approach allowed to determine on the unmodified collagen samples we have used that the kinetics of fibril formation for ASC I and PSC II was modulated by NaCl concentration, in accordance with previous reports (Williams et al., 1978; Wood and Keech, 1960): on increasing the NaCl concentration, there is an increase of lag time and, depending on the preparation (ASC I or PSC II), a more or less pronounced decrease of the highest slope of the turbidimetric curve. Also temperature variations in the range 28–32 8C influences fibril formation kinetics, the effect being greater at a higher NaCl concentration (data not shown). Phosphate is reported to produce well ordered fibrils but to be a potent inhibitor of collagen fibrillogenesis mainly above 30 or 10 mM (Kuznetsova et al., 1998; Pogany et al., 1994; Williams et al., 1978). On this basis and considering the effect of NaCl concentration, we halved the concentration of phosphate and NaCl to 5 mM and 65 mM, respectively, obtaining measurable turbidimetry curves also for N-methylated samples. All the subsequent experiments were therefore performed at these low phosphate and NaCl concentrations, unless otherwise noted. Turbidity measurements on unmodified and N-methylated ASC I, PSC I and PSC II produced the results shown in Fig. 1, panels A–D, where the percentage of derivatization and aggregation is reported for each sample. Worth of initial mention is the fact that all unmodified samples showed on incubation measurable turbidimetric curves, high percentage of aggregation competency and, as reported below, banded fibrils. After N-methylation, measurable turbidimetric curves were obtained also for N-methylated ASC I (Fig. 1A and B) and PSC II (Fig. 1D) showing that these preparations form aggregates large enough to cause increase of absorbance at 313 nm. The

R. Tenni et al. / Micron 37 (2006) 640–647

643

Fig. 1. Turbidity curves of N-methylated derivatives at 32 8C. ASC I, PSC I and PSC II were tested at neutral pH and 0.2 mg/ml collagen in 5 mM phosphate and 65 mM NaCl. Curves for the unmodified parent samples and the modified ones are indicated by N 0 and Nm, respectively. Data enclosed in square and round brackets are the percentage of aggregation and the percentage of modification, respectively.

kinetic parameters are however influenced by N-methylation, in particular an increase of lag time and a decrease of maximum slope, with a lower aggregation competency. Some variability occurs on repeated incubations of the same sample (see, e.g., Nmethylated ASC I with 61% modification reported in Fig. 1A and B), but the same kinetic effects are evident for both determinations. Furthermore, a different degree of derivatization for two different preparations of N-methylated ASC I has only a minor effect on kinetic parameters, as shown in Fig. 1B for two samples with 61% and 98% modification (the same result was obtained for N-methylated PSC II with 90% and 99% modification, data not shown). On the contrary, turbidity was barely detectable for the Nmethylated PSC I (Fig. 1C) notwithstanding a sizeable amount of the sample became insoluble during the incubation; these findings are in accordance with morphological determinations by TEM, reported below, where only N-methylated PSC I did not show structured aggregates. The fraction of the sample able to aggregate has a great variability among the different N-methylated collagen and has no evident relation with the final absorbance. This last aspect might not be surprising, since the absorbance in turbidimetric methods depends on different factors, such as the relative value of aggregate size versus wavelength, and probably also the nature of the aggregates Ancillary turbidimetric measurements were performed for ASC I at pH 4.8 (data not shown). Unmodified ASC I was able to aggregate, with a higher increase of absorbance than at neutral pH; on the contrary, N-methylation showed no variation

of absorbance, most of the collagen molecules remaining soluble after 300 min of incubation at 32 8C. 3.3. Aggregation kinetics of N-acetylated collagen samples N-acetylated samples were not soluble in 5–50 mM acetic acid. Like for succinyl collagen (Rauterberg and Ku¨hn, 1968), this behaviour is caused by loss on derivatization of the positive

Fig. 2. Turbidity curves of N-acetylated ASC I at 32 8C. Curves were obtained on a preparation of N-acetylated ASC I (Na) with 66% modification in the neutral pH, 5 mM phosphate and NaCl concentrations indicated for each curve in parenthesis. Note the vertical scale and the very low variation of absorbance, but in the presence of a sizeable percentage of sample forming aggregates (this percentage is in square brackets).

644

R. Tenni et al. / Micron 37 (2006) 640–647

Fig. 3. Acid-soluble type I collagen after 300-min incubation at 32 8C: (a) underivatized and (b) N-methylated samples. Bar: 100 nm.

charge of Lys/Hyl side chain, as detailed under Section 2.1, causing a decrease of the isoelectric pH. All turbidity measurements were performed after the dissolution of samples in a neutral solvent. The results showed that in all conditions N-acetylated samples had a very low increase of absorbance (Fig. 2), much lower than the parent collagen or the N-methylated counterparts, but a sizeable amount became insoluble during incubation. At acidic pH, Nacetylated samples showed a low but measurable turbidity flanked by a high percentage of insoluble material, due to the low solubility mentioned above (data not shown). 3.4. Electron microscopy results Electron microscopy images were obtained for the collagen preparations, after 300 min or overnight incubation at 32 8C in 5 mM sodium phosphate, 65 mM NaCl, neutral pH. Acid-soluble type I collagen (Fig. 3). After 300 min and overnight incubation, both the underivatized sample and the Nmethylated derivative formed banded fibrils with D-periodicity; morphometric measurements showed mean fibril period of 67.1  0.48 nm (underivatized) and 67.3  0.69 nm (N-methylated) ( p > 99%); the means t-student test showed no significant differences ( p > 99%) between them. A background of microfibrillar material in early stage of aggregation was also detected.

Fig. 4. Pepsin-soluble type I collagen after overnight incubation at 32 8C: (a) underivatized samples: background of unbanded microfibrillar material (arrows) and banded twisted collagen fibrils (asterisks). (b) N-methylated samples: background of non aggregated microfibrillar material (arrows) and collagen banded fibril (asterisk). Bar: 100 nm.

Pepsin-soluble type I collagen (Fig. 4). For the unmodified sample unbanded microfibrillar material was evident after 300 min incubation, whereas twisted collagen fibrils with typical D-period were detected only after overnight incubation. The N-methylated derivative showed some aggregates only after overnight incubation; in particular, very few collagen molecules formed fibrils with D-periodicity that were dispersed on a background of non-aggregated microfibrillar material lacking any preferential orientation. Pepsin-soluble type II collagen (Fig. 5). After overnight incubation, the underivatized preparation formed a large number of thick fibrils with heterogeneous diameters but regular D-periodicity, dispersed in a microfibrillar material. In some regions of fibril bundles, interfibrillar spaces were more evident when compared to the other samples. On the contrary, only microfibrillar material and no fibrils were detected after 300 min incubation of N-methylated type II collagen; after overnight incubation, banded aggregates, similar to SLS were detected in a background of nonarranged microfibrillar material (Chapman and Hulmes, 1984). Finally, only a microfibrillar background and no periodic and structured fibrils were detected in the N-acetylated samples of ASC I.

R. Tenni et al. / Micron 37 (2006) 640–647

Fig. 5. Pepsin-soluble type II collagen after overnight incubation at 32 8C: (a) underivatized samples: thick fibrils with heterogeneous diameters but regular Dperiodicity. (b) N-methylated samples: segment long spacing collagen (SLS). Bar: 100 nm.

4. Discussion Lys/Hyl e-amino groups of three preparations of fibrillar collagens were N-methylated or N-acetylated in mild conditions. The positive charge in neutral conditions of Lys/Hyl side chain is preserved on N-methylation but is lost on Nacetylation. Recent findings in our laboratory showed that N-methylation and N-acetylation do not significantly alter the thermal stability of the collagen triple helical conformation (Giudici et al., 2003). At a higher molecular level, the interaction of collagens I and II with decorin is lost on N-acetylation, whereas Nmethylation modulates the same interaction (Tenni et al., 2002); similar results have been obtained with fibromodulin (unpublished results). The present work deals with a different phenomenon potentially influenced by modifications of ionic groups, namely self-aggregation in vitro of fibrillar collagens and morphology of supramolecular assemblies. As parent and control samples, we used ASC I, PSC I and PSC II, only the first bearing telopeptides because the other two were subjected to pepsin treatment. Kinetic measurements by turbidimetry and morphological analysis of the aggregates by TEM demonstrated that all three preparations were able to self aggregate, giving rise to banded fibrils with classical periodicity (Figs. 3–5).

645

The results obtained on the same collagen samples but after chemical modification demonstrate that N-methylated ASC I and PSC II maintain the ability to self-aggregate also after Nmethylation, giving rise to definite and identifiable supramolecular species, viz. banded fibrils with 67-nm periodicity for ASC I and SLS for PSC II. On the contrary, N-acetylation of ASC I and PSC II hinders any aggregation. The positive charge of Lys/Hyl side chains is therefore essential for specific selfaggregation of collagen. N-methylation is thus the most interesting chemical modification; in particular, it is the first derivatization of a fibrillar collagen that allows the formation of 67-nm banded fibrils in neutral conditions. The other derivatizations reported in the literature (esterification of the carboxylic groups, deamidation of Asn/Gln, N-succinylation of the primary amino groups, deamination of Lys/Hyl or blockage of Arg) cause the loss of such ability (Balleisen et al., 1976; Rauterberg and Ku¨hn, 1968; Wang et al., 1978; Wen Hu et al., 1996). The experimental results show that N-methylation affects at least three aspects of collagen self-aggregation: the requirement of low concentration of phosphate and NaCl in the incubation buffer; the aggregation kinetics (both lag time and slope of turbidimetric curves); nature and morphology of the supramolecular formed species. These effects originate from the characteristics of N-methyl groups. N-methylation transforms the primary amino group of Lys/ Hyl side chains to a tertiary amine, i.e. R–NH3+ becomes R– NH(CH3)2+ at neutral pH (mostly N-dimethyl derivatives are formed because the second methylation step is faster than the first one (Jentoft and Dearborn, 1983)). The number of H atoms bound to the nitrogen atom thus decrease from three to one, directly causing an identical decrease of the maximum number of hydrogen bonds with water molecules (or other groups of the same or a different collagen molecule). In addition, the formation of salt bridges is hindered, and/or the strength of salt bridges is weaker, in the presence of two N-linked methyls. These losses of number/strength of chemical bonds cause thermodynamical destabilization of aggregates. Since fibril formation by type I collagen is endothermic, but entropy driven (Kadler et al., 1987), the thermodynamical destabilization causes an enthalpy change on fibril formation more positive for N-methylated samples than for the unmodified parent collagens. The kinetic consequences of N-methylation indicate that the rates of fibril nucleation and growth are affected by the thermodynamical destabilization of aggregates. The destabilization is however insufficient to block fibril formation for ASC I and PSC II, whereas it is large enough to block fibril formation for PSC I. It should be noted that the presence of telopeptides may play a role in fibril formation, as evidenced by the different behaviour of N-methylated ASC I and PSC I. Literature reports the involvement of telopeptides in one or more steps of fibrillogenesis (Gelman et al., 1979; Sato et al., 2000; Snowden and Swann, 1979), and a catalytic role for telopeptides was suggested (Kuznetsova and Leikin, 1999). However, also PSC II lacks telopeptides but its N-methylated counterpart is able to self-aggregate.

646

R. Tenni et al. / Micron 37 (2006) 640–647

The destabilization caused by N-methylation may also account for the requirement of low concentrations of phosphate and NaCl in the incubation buffer, one half with respect to more frequently used conditions of physiological ionic strength (e.g., 10 mM phosphate and 130 mM NaCl). Recent evidences show some facets of phosphate action: (a) collagen solubility increases with increasing concentrations of phosphate (specifically due to the divalent HPO42 ion); (b) a small contrasting effect (promoting aggregation) comes into evidence at low concentrations of phosphate; (c) 1–2 binding sites for divalent phosphate (and sulfate) were reported under physiological conditions per collagen molecule in fibers, bound anions appearing to form salt bridges within regions of high excess positive charge (Mertz and Leikin, 2004). Since the interaction of N-methylated collagen samples with phosphate may block ion-pairing of N-methyl groups with anionic groups of different collagen molecules (i.e., may decrease electrostatic interactions between different collagen molecules), the exclusion of at least some of the bound phosphate is required for fibril formation. The energetic cost for phosphate exclusion is lower when bulk concentration of phosphate is lower. N-methylation of the three collagen preparation has a differential effect on the nature and morphology of supramolecular species (as determined by TEM) formed in vitro in neutral conditions: ASC I gives rise only to well defined banded fibrils with 67-nm periodicity; almost no aggregates were detected for type I collagen lacking telopeptides (PSC I); a different polymorphic species, SLS, was found for PSC II. Literature reports that experimental conditions influence which polymorphic form will form in vitro: segment long spacing (SLS) form can be obtained at low pH in the presence of small polyanionic substances; fibrous long spacing (FLS) form at a low initial pH plus the presence of suitable large polyanionic agents and the removal of small ions (Chapman and Hulmes, 1984). The supramolecular species SLS we have found for N-methylated PSC II was obtained at neutral pH in the absence of any polyanionic molecules. In this case, SLS formation depends solely on molecular or structural characteristics of the (derivatized) collagen itself. Consequently, we may think that N-methylation alters molecular recognition and/or the relative stability of fibril nuclei differing for the organization of collagen molecules. Early aggregates that are more stable or form with a faster rate direct the following steps of self-aggregation towards banded fibrils or SLS (as for Nmethylated ASC I and PSC II, respectively). In conclusion, our results indicate that the protonated primary amine of Lys/Hyl side chains is involved in fibril formation, and that its positive charge at physiological pH is essential, as demonstrated by lack of self-agregation of Nacetylated samples. N-methylation allows collagen aggregation to fibrils of specific morphology, but also this modification influences several aspects of collagen biochemistry: molecular recognition, interaction with self influencing kinetics of fibril formation, stability and nature of aggregates, and a diversified effect on apparently similar collagen preparations; these consequences derive from the molecular characteristics of the N-dimethyl groups discussed above. Worthnoting, these

characteristics are relevant for interhelical interactions, whereas they do not significantly influence triple helical thermal stability, suggesting that they play a minor role in interchain interactions. Acknowledgements We are indebted to Dr. Sergey Leikin (NICHD, NIH, Bethesda, MD, USA) for discussions on the role of phosphate and the characteristics of modified groups. Thanks also to Drs. Antonella Forlino and Camilla Giudici for helpful suggestion and criticism, to Elena Campari and Gianfranco Filippini for technical assistance. This study was supported by grants from University of Pavia (FAR) and italian MIUR (project PRIN 2003055193 and 2004055533). References Balleisen, L., Marx, R., Ku¨hn, K., 1976. Platelet–collagen interaction. Haemostasis 5, 155–164. Brodsky, B., Shah, N.K., 1995. Protein motifs. FASEB J. 9, 1537–1546. Chapman, J.A., Hulmes, D.J.S., 1984. Electron microscopy of the collagen fibril. In: Ruggeri, A., Motta, P.M. (Eds.), Ultrastructure of the connective tissue matrix. M. Nijhoff Publ., Boston, pp. 1–33. Davis, N.R., Risen, O.M., Pringle, G.A., 1975. Stable, nonreducible cross-links of mature collagen. Biochemistry 14, 2031–2036. Diamond, A.M., Gorham, S.D., Etherington, D.J., Robertson, J.G., Light, N.D., 1991. The effect of modification on the susceptibility of collagen to proteolysis. Matrix 11, 321–329. Do¨lz, R., Heidemann, E., 1986. Influence of different tripeptides on the stability of the collagen triple helix. Biopolymers 25, 1069–1080. Fitzsimmons, C.M., Cockburn, C.G., Hornsey, V., Prowse, C.V., Barnes, M.J., 1988. The interaction of von Willebrand factor (vWf) with collagen: investigation of vWf-binding sites in the collagen molecule. Thromb. Haemost. 59, 186–192. Gelman, R.A., Poppke, D.C., Piez, K.A., 1979. Collagen fibril formation in vitro. The role of the nonhelical terminal regions. J. Biol. Chem. 254, 11741–11745. Giudici, C., Viola, M., Tira, M.E., Forlino, A., Tenni, R., 2003. Molecular stability of chemically modified collagen triple helices. FEBS Lett. 547, 170–176. Gratzer, P.F., Santerre, J.P., Lee, J.M., 2004. Modulation of collagen proteolysis by chemical modification of amino acid side-chains in acellularized arteries. Biomaterials 25, 2081–2094. Hofmann, H., Fietzek, P.P., Kuhn, K., 1980. Comparative analysis of the sequences of the three collagen chains a1(I), a2 and a1(III). J. Mol. Biol. 141, 293–314. Hulmes, D.J., Miller, A., Parry, D.A., Piez, K.A., Woodhead-Galloway, J., 1973. Analysis of the primary structure of collagen for the origins of molecular packing. J. Mol. Biol. 79, 137–148. Huszar, G., Maiocco, J., Naftolin, F., 1980. Monitoring of collagen and collagen fragments in chromatography of protein mixtures. Anal. Biochem. 105, 424–429. Jentoft, N., Dearborn, D.G., 1983. Protein labeling by reductive alkylation. Methods Enzymol. 91, 570–579. Kadler, K.E., Hojima, Y., Prockop, D.J., 1987. Assembly of collagen fibrils de novo by cleavage of the type I pC-collagen with procollagen C-proteinase. Assay of critical concentration demonstrates that collagen self-assembly is a classical example of an entropy-driven process. J. Biol. Chem. 262, 15696– 15701. Kakade, M.L., Liener, I.E., 1969. Determination of available lysine in proteins. Anal. Biochem. 27, 273–280. Koide, T., Takahara, Y., Asada, S., Nagata, K., 2002. Xaa–Arg–Gly triplets in the collagen triple helix are dominant binding sites for the molecular chaperone HSP47. J. Biol. Chem. 277, 6178–6182.

R. Tenni et al. / Micron 37 (2006) 640–647 Kuznetsova, N., Chi, S.L., Leikin, S., 1998. Sugars and polyols inhibit fibrillogenesis of type I collagen by disrupting hydrogen-bonded water bridges between the helices. Biochemistry 37, 11888–11895. Kuznetsova, N., Leikin, S., 1999. Does the triple helical domain of type I collagen encode molecular recognition and fiber assembly while telopeptides serve as catalytic domains? J. Biol. Chem. 274, 36083–36088. Mertz, E.L., Leikin, S., 2004. Interactions of inorganic phosphate and sulfate anions with collagen. Biochemistry 43, 14901–14912. Persikov, A.V., Ramshaw, J.A., Brodsky, B., 2000. Collagen model peptides: Sequence dependence of triple-helix stability. Biopolymers 55, 436– 450. Persikov, A.V., Ramshaw, J.A., Kirkpatrick, A., Brodsky, B., 2002. Peptide investigations of pairwise interactions in the collagen triple-helix. J. Mol. Biol. 316, 385–394. Persikov, A.V., Ramshaw, J.A., Kirkpatrick, A., Brodsky, B., 2005. Electrostatic interactions involving lysine make major contributions to collagen triplehelix stability. Biochemistry 44, 1414–1422. Pogany, G., Hernandez, D.J., Vogel, K.G., 1994. The in vitro interaction of proteoglycans with type I collagen is modulated by phosphate. Arch. Biochem. Biophys. 313, 102–111. Prockop, D.J., Fertala, A., 1998. Inhibition of the self-assembly of collagen I into fibrils with synthetic peptides. J. Biol. Chem. 273, 15598–15604. Rauterberg, J., Ku¨hn, K., 1968. The renaturation behaviour of modified collagen molecules. Hoppe Seylers Z. Physiol. Chem. 349, 611–622. Reese, C.A., Mayne, R., 1981. Minor collagens of chicken hyaline cartilage. Biochemistry 20, 5443–5448. Rossi, A., Zuccarello, L.V., Zanaboni, G., Monzani, E., Dyne, K.M., Cetta, G., Tenni, R., 1996. Type I collagen CNBr peptides: species and behavior in solution. Biochemistry 35, 6048–6057. Sato, K., Ebihara, T., Adachi, E., Kawashima, S., Hattori, S., Irie, S., 2000. Possible involvement of aminotelopeptide in self-assembly and thermal stability of collagen I as revealed by its removal with proteases. J. Biol. Chem. 275, 25870–25875.

647

Snowden, J.M., Swann, D.A., 1979. The formation and thermal stability of in vitro assembled fibrils from acid-soluble and pepsin-treated collagens. Biochim. Biophys. Acta 580, 372–381. Sykes, B.C., 1976. The separation of two soft-tissue collagens by covalent chromatography. FEBS Lett. 61, 180–185. Tenni, R., Viola, M., Welser, F., Sini, P., Giudici, C., Rossi, A., Tira, M.E., 2002. Interaction of decorin with CNBr peptides from collagens I and II. Eur. J. Biochem. 269, 1428–1437. Thomson, C.A., Tenni, R., Ananthanarayanan, V.S., 2003. Mapping Hsp47 binding site(s) using CNBr peptides derived from type I and type II collagen. Protein Sci. 12, 1792–1800. van der Rest, M., Garrone, R., 1991. Collagen family of proteins. FASEB J. 5, 2814–2823. Wang, C.L., Miyata, T., Weksler, B., Rubin, A.L., Stenzel, K.H., 1978. Collagen-induced platelet aggregation and release. Biochim. Biophys. Acta 544, 555–567. Wen Hu, X., Knight, D.P., Grant, R.A., 1996. The effect of deamination and/or blocking arginine residues on the molecular assembly of acid-extracted rat tail tendon collagen. Tissue Cell 28, 215–222. Williams, B.R., Gelman, R.A., Poppke, D.C., Piez, K.A., 1978. Collagen fibril formation. J. Biol. Chem. 253, 6578–6585. Wilner, G.D., Nossel, H.L., Procupez, T.L., 1971. Aggregation of platelets by collagen: polar active sites of insoluble human collagen. Am. J. Physiol. 220, 1074–1079. Wood, G.C., Keech, M.K., 1960. The formation of fibrils from collagen solutions. Biochem J. 75, 588–598. Xu, Y., Gurusiddappa, S., Rich, R.L., Owens, R.T., Keene, D.R., Mayne, R., Hook, A., Hook, M., 2000. Multiple binding sites in collagen type I for the integrins a1b1 and a2b1. J. Biol. Chem. 275, 38981–38989. Zhang, W.M., Kapyla, J., Puranen, J.S., Knight, C.G., Tiger, C.F., Pentikainen, O.T., Johnson, M.S., Farndale, R.W., Heino, J., Gullberg, D., 2003. a11b1 integrin recognizes the GFOGER sequence in interstitial collagens. J. Biol. Chem. 278, 7270–7277.