- Email: [email protected]
Ultrastructure and assembly of segmental long spacing collagen studied by atomic force microscopy
PERGAMON
Micron 32 (2001) 355–361 www.elsevier.com/locate/micron
Ultrastructure and assembly of segmental long spacing collagen studied by atomic force microscopy M.F. Paige, M.C. Goh* Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario, Canada M5S 3H6
Abstract The in vitro formation of segmental long spacing (SLS) collagen as induced by the addition of ATP to acidified Type I collagen solutions has been examined with the atomic force microscope (AFM). AFM images obtained suggest that the assembly proceeds in a stepwise manner, through an intermediate stage of oligomers, which then associate laterally to form the so-called “SLS crystallites”. Attempts to induce SLS formation by the addition of other polyanionic species to monomeric collagen solutions met with mixed success; ATP-g-S and GTP produced SLS crystallites, whereas inorganic phosphate and other polyanionic dyes did not. This indicates that the formation of SLS cannot simply be attributed to the negation of positive charges believed to be located on the end of the collagen monomer, but rather it is a complex function of the structure and charge of both the collagen monomer and polyanion. 䉷 2000 Elsevier Science Ltd. All rights reserved. Keywords: Collagen; Atomic force microscope; SLS collagen; Assembly; Mechanism
1. Introduction Aggregation of type I collagen typically leads to the formation of fibrils with a characteristic banding pattern of about 67 nm period (Nimni, 1988). The in vitro assembly of these native type fibrils has been examined using a variety of techniques and is reviewed in a number of papers (Veis and George, 1994; Kadler et al., 1996; Holmes, et al., this issue). The effects of solvent conditions such as concentration, temperature and ionic strength on the final morphology of the products and on the process as a whole have also been investigated (Trelstad et al., 1976; Goh et al., 1997). The conditions under which the assembly proceeds can have a drastic effect on the kinetics and the final structure of the aggregate. While most collagen assembly products are fibrillar, the addition of dilute, acidified solutions of adenosine triphosphate (ATP) to monomeric collagen gives rise to a final collagen aggregate that is completely different in morphology, one which is block-like rather than fibrillar. This was first noted by Schmitt et al. (1953), who named the aggregates segmental-long-spacing (SLS) collagen or SLS crystallites. Electron microscopy (EM) measurements show that these SLS collagen aggregates have approximately the same length as the monomer units (⬃280 nm) and have a welldefined internal banding pattern perpendicular to their * Corresponding author. Tel.: ⫹1-416-978-6254; fax: ⫹1-416-978-6254. E-mail address: [email protected] (M.C. Goh).
lengths. The proposed model for the structure of SLS collagen is thus of monomers aligned in register, as if in a crystalline lattice, and the observed banding pattern was interpreted to correspond with the distribution of amino acid residues along the collagen polypeptide. Kuhn (1982) showed that there is an excellent correlation between this banding pattern and the location of the positively charged regions in the known sequence of the monomer. Given this model for the structure of the SLS aggregate, the main interest in it and its in vitro formation is its utility to the overall study of collagen monomers (Kuhn, 1982): SLS crystallites have been utilized as a means of measuring lengths of various collagen monomers; for inferring the amino acid composition of different collagen types (Timpl et al., 1978); and for characterization of various collagen precursors (Layman et al., 1971). In vivo, structures similar to SLS aggregates have been reported and investigated in association with the intracellular translocation of procollagen molecules (Bruns et al., 1979). Efforts have been made towards understanding the nature of the association between ATP and the collagen monomers in SLS collagen. To reduce possible effects of heavy metal staining on the banding patterns of the aggregates, Kobayashi (1992) studied the structure of the aggregates using the incorporated ATP for visualization, instead of traditional EM staining techniques. The EM band patterns obtained in this manner were well correlated with the distribution of charged basic residues in the monomer. A strong band
0968-4328/00/$ - see front matter 䉷 2000 Elsevier Science Ltd. All rights reserved. PII: S0968-432 8(00)00038-X
356
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361
initial reversible aggregation of monomers until a specific size (the ‘critical nucleus’) is reached. Beyond the critical nucleus, further growth is achieved by the addition of the monomers to the critical nucleus. The atomic force microscope (AFM), which is capable of close to atomic resolution, has previously been shown a useful tool in the examination of collagen ultrastructure and assembly (Gale et al., 1995; Goh et al., 1997; Paige et al., 1998). In this paper, we utilize the AFM to investigate the formation of SLS collagen by visualization of the assembly products at early stages of the process. The images we obtain suggests that the process proceeds, at least in part, in a step-wise fashion. We also investigate the role of ATP in comparison with other negatively charged ions in inducing SLS formation.
2. Materials and methods 2.1. In vitro SLS collagen assembly
Fig. 1. (a) and (b): AFM images of SLS collagen aggregates formed via addition of ATP to acidified collagen monomers, taken in deflection mode. Product was diluted by 100 × prior to imaging. For (a), image size is 15 × 15 mm 2; for (b), image size is 5 × 5 mm 2.
near one end corresponded with the presence of a distinct positively charged region near the N-terminus of the monomer, as shown by the calculations of Veis and George (1994) based upon collagen’s known amino acid sequence. These results led to the suggestion that the ATP molecules were located within the aggregate and acted as bridging units between the basic residues of adjacent monomers. It has thus been suggested that the role of ATP in SLS formation is to inhibit the electrostatic repulsion between similarly charged regions of monomers aligned in register, which are preventing the crystallization. While it is generally accepted that SLS aggregates are composed of laterally aligned monomers, their mechanism of aggregation has received minimal attention. The proposed ordered structure, and the term “SLS crystallite”, suggests that they form in a manner similar to that of typical crystals. In this case, there is nucleation step, which is an
Type I calf-skin collagen (Sigma) was dissolved over ice in 0.05% acetic acid, with occasional sonicating to facilitate the breakdown of collagen aggregates. The mixture was centrifuged at 10,000 rpm for 60 min at 4⬚C. After centrifugation, the supernatant was filtered through 0.45 mm Millipore filters (Sigma) and mixed with the appropriate polyanion solution. Solutions of the polyanions were prepared by dissolving the solid (ATP, ADP, GTP, Amaranth, Cibacron 3GA from Sigma; ATP-g-S from Calbiochem) in 0.05% acetic acid. Collagen and polyanion solutions were combined at room temperature to yield mixtures with final collagen concentrations typically of ⬃0.5 mg/ml, polyanion concentrations of ⬃2 mg/ml and a final pH of 3.5. The mixtures were allowed to equilibrate overnight at room temperature. A series of dilutions ranging from 10–1000 fold were prepared by adding aliquots of the reaction mixture to an appropriate volume of Millipore water. The diluted samples were deposited onto freshly cleaved sheets of mica in 20 ml aliquots and allowed to dry for an hour prior to imaging in the AFM.
2.2. Atomic force microscopy Samples were imaged with a Nanoscope III instrument (Digital Instruments, Santa Barbara, CA), typically using a square pyramidal silicon nitride tip of nominal spring constant ⬃0.58 N/m. Tips produced by electron beam deposition were also used in attempts to improve resolution. Images were taken in contact mode, in air. Images obtained were consistently reproducible, showing minimal perturbation from the probe tip. Both height and force mode images were taken simultaneously, with height measurements being taken from height mode images only.
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361
Fig. 2. (a)–(c): Sectional analysis of SLS aggregate. Part (a) shows longitudinal section of SLS aggregate. Part (b) shows diameter measurement. Part (c) shows width measurement.
3. Results and discussion 3.1. Structural studies When monomeric collagen solutions were combined with ATP, ATP-g-S or GTP according to the preparation described above, white, turbid solutions were obtained. Analysis of the resulting mixtures by means of the AFM revealed an abundance of SLS aggregates. A typical AFM image of a 100-fold diluted sample obtained from an ATPcollagen preparation is shown in Fig. 1a and b. The AFM images of SLS aggregates were consistent with the EM results described by Schmitt et al. (1953) in terms of their overall structure. The aggregates appeared rectangular, with a height of 30–50 nm above the flat mica substrate. The lateral dimensions are ⬃360 nm by 200–300 nm, with a much larger spread in the latter value than in the former. Note that because AFM images are a convolution of the
357
sample and probe tip geometry (Keller, 1991; Markiewicz and Goh, 1994), the lateral dimensions are an overestimate of the actual ones. Sectional analysis of a typical aggregate is shown in Fig. 2. We shall ascribe the dimension with less variance (⬃360 nm) to correspond with the long axis of the monomer (plus tip geometry), in accordance with previous model of monomers arranged in register. Measurements taken along this collagen monomer axis will be referred to as the longitudinal section, while perpendicular measurements (shown in Fig. 2c) will be called the cross section. It should be noted that the smallest dimension of the block-like aggregate (30– 50 nm, shown in Fig. 2b) is always that above the substrate. This is probably the result of a positive interaction between the aggregate and the mica substrate, causing the aggregates to always lie flat in order to maximize the area of contact with the substrate. The various sections clearly indicate that the aggregates are not cylindrically symmetric. A closer examination of Fig. 2 shows that the longitudinal section is almost rectangular with a steep slope, while the cross section shows a less steep slope, with an obvious rounding at the apex. These are consistent with the previous assignment: the sharp slope in the longitudinal section can be ascribed to the ends of the aligned monomers, which would terminate more abruptly. If one were to view the SLS aggregate as a small crystal, faceting would be expected, instead of the observed rounded cross section. This could be taken as an argument against a completely crystalline packing within the SLS aggregate. However, there is an alternative possibility that this rounding is caused by the interaction between SLS and the mica substrate which is much too strong, and enough to deform a weakly crystalline aggregate. We are unable to resolve this issue at this point. With the use of electron beam deposited tips of high aspect ratio, finer structural features on the surface of an aggregate can be resolved. The longitudinal-section, shown in Fig. 2a, indicates the presence of a series of minor grooves on the surface of the aggregate. These features generally consisted of a higher ridge on one end of the aggregate, which is ⬃2.5 nm in height above the average aggregate surface. A series of smaller ripples along the section are also present, the resolution of which is possibly limited by the finite width and aspect ratio of the AFM tip. The presence of these ultrastructural features are in accord with the results from EM (Kuhn, 1982), and are presumably caused by the alignment of corresponding amino acids within the monomers. Unfortunately, we were unable to obtain better resolution in order to make a quantitative comparison; AFM probes with much smaller tip radii, such as those constructed from carbon nanotubes (Dai et al., 1996) may be useful for this characterization. 3.2. Assembly via intermediates By changing the concentration of collagen and ATP, the
358
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361
Fig. 3. AFM images of SLS aggregates in intermediate stage of formation. (a) Height mode image (scan size: 6 × 6 mm 2); (b) Amplitude mode image (scan size: 6 × 6 mm 2); (c) Zoom in of (a); image size (3 × 3 mm 2); and (d) Zoom in of (b), image size (3 × 3 mm 2).
kinetics of the assembly process can be slowed down enough so we can quench the system at various stages before fully developed SLS crystallites are formed. While we are not yet able to follow the assembly in situ, we can derive information about the process by examination of various assembly products that are observed on many different trials. Examples of images of these aggregates are shown in Fig. 3, where we can see a variety of structures, ranging from objects that appear fibrous to mature SLS crystallites. We can classify the patterns of structures we observed into a few distinct types, and divide the apparent stages of assembly into four different steps, each labelled by a letter (a)–(d). We do not claim that these patterns correspond to distinct and specific stages of the assembly; however, they do provide a clue in inferring how the assembly proceeds. The first pattern of structure that emerges is indicated by the (a), and shows an abundance of fibrous structures
dispersed over a relatively wide length scale. These fibrous structures have the same approximate length of a monomer, but a larger cross-section. Measurements of over 30 such entities yielded typical diameters of 2.5 nm, in contrast with ⬃1 nm for the monomer, indicating that they are oligomers composed of at least several monomers in register. Even at this early stage in the assembly, there appears to be some long-range ordering of the oligomers present; while they are distinct, independent entities, a definite lateral alignment amongst them is evident. In pattern (b), we see that a set of oligomers that are closer to one another and are beginning to merge, creating markedly larger bodies. Regions of the substrate between these larger bodies can still be resolved, however, indicating that the assembly process is still incomplete. In (c), the larger structures have fused and packed together to the point that the substrate can no longer be seen. Sectional analysis of structures, at this stage, show bumps and ridges which are not
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361
359
Fig. 4. Illustration of proposed assembly mechanism for SLS aggregates. Assembly involves the formation of stable oligomer intermediates, followed by merging of these intermediates to form the final aggregates.
apparent in the fully formed SLS aggregate. Presumably, the formation of a mature SLS involves further packing of the aggregate constiutents until only the smallest of ultrastructural features can be distinguished. Pattern (d) corresponds to what is essentially a fully mature SLS aggregate, having dimensions and ultrastructure typified by that shown in Fig. 2. While we are not able to follow the actual evolution of a single SLS aggregate, the existence of distinct patterns indicate that the formation of SLS crystallites occurs, at least to some extent, in a stepwise manner. The repeated occurrence of distinct features indicate that there are stable intermediates in the process, instead of a pure nucleation and growth, wherein we would expect to see is a wide distribution of sizes of the block-like SLS aggregate and no other pattern. We postulate that SLS formation occurs in the following fashion: in an as-of-yet unobserved step, monomers combine to form fibrous oligomers. These oligomers are stable intermediates, which then associate laterally. They are then compacted, and ultimately fuse to form the final crystallite. This is illustrated schematically in Fig. 4, with the labeled axes on the final aggregate corresponding to the sections shown in Fig. 2.
3.3. Role of ions in the assembly The role of ATP in the assembly process has been previously speculated upon, and the general view is that its negative charges serve to neutralize the positive charges on the collagen monomers, which serve to counteract aggregation (Veis and George, 1994). However, our attempts to induce SLS aggregation by the addition of a myriad of anionic reagents, including inorganic phosphate (in the form of KH2PO4) and ADP, met with uniform failure, indicating that simple charge neutralization is not the sole requirement. On the other hand, the addition of GTP to monomeric collagen produced aggregates whose features are indistinguishable from those obtained with ATP. These results with GTP, ATP and ADP may suggest that the aggregation is activated by the hydrolysis of the phosphate bond, similar to the case for the ATP-assisted assembly of actin filaments, which involves an irreversible hydrolysis of a phosphate group on a G-actin-associated ATP complex, followed by a reversible exchange of ADP against ATP on a monomeric G-actin unit (Wegner and Engel, 1975). In order to examine this possibility, we used ATP-g-S, which is a non-hydrolyzable ATP analogue. This
360
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361
Kobayashi (Kobayashi, 1992) has suggested that the role of the polyanion is to act as a bridging unit between basic residues on the laterally aligned monomer. The results we obtained suggest that not only are three negatively charged groups needed, but that they have to be arranged in a particular way with respect to one another. It is also possible that the ability of the polyanion to participate in hydrogen bonding may play a significant role as well, since both adenosine and guanosine functionalities are well known for this ability.
4. Summary
Fig. 5. (a)–(c): Chemical structures of polyanionic dyes and ATP; (a) Amaranth; (b) Cibacron blue 3GA; and (c) Adenosine 5 0 triphophosphate (ATP).
preparation produced SLS aggregates with no observable structural difference from those produced by the addition of ATP or GTP, indicating that the role of the polyanion is not to provide assistance with the energetics of the assembly by means of exothermic degradation. The fact that aggregation occurs with either ATP, ATP-g-S, or GTP, but not with ADP, indicates that assembly requires the presence of three adjacent negatively charged groups, but that it is unaffected by small structural alterations in the nitrogenous base of these polyanions. To further investigate the role of charged polyanions in the assembly, we examined the effect of two dyes, Cibacron blue 3GA and Amaranth (structures are shown along with that of ATP in Fig. 5a–c). Both of these dyes possess three negatively charged groups. In both cases, SLS aggregation was not induced, although heavy, fibrous precipitate of indeterminate structures were formed. This suggests that while a high local concentration of negative charge can yield aggregation, SLS formation is by no means inevitable; the amount and type of charge is clearly an important prerequisite for SLS formation, but the manner in which it is distributed within a polyanion appears to also play a crucial role.
SLS collagen aggregates have been formed in vitro and studied by means of the AFM. Measurements of the aggregate yield typical longitudinal and cross sections of and 360 and 200–300 nm, respectively, which are consistent with results obtained from electron microscopy. Electron beamdeposited AFM tips have revealed small ultrastructural features on the surface of the aggregate, but the small size of these features have made them difficult to resolve. Images of SLS aggregates in the early stages of the formation process have been obtained. The aggregates appear to form by the lateral addition of collagen oligomers which are ⬃2.5 nm in height, followed by the packing and fusing together of these intermediates. This suggests that the assembly of SLS collagen proceed by means of a stepwise mechanism, with the oligomers being stable intermediates in the process. Additionally, it was found that SLS aggregation could be induced by addition of ATP-g-S and GTP but not with various other polyanions. The structural details of the supporting polyanion, and perhaps its ability to accept and donate hydrogen bonds, appear to be important factors dictating SLS aggregate formation.
Acknowledgements This work was supported by grants from the Natural Science and Engineering Research Council of Canada, Photonics Research Ontario and the ACS-PRF.
References Bruns, R., Hulmes, D., Therrien, S., Gross, J., 1979. Procollagen segmentlong-spacing crystallites: their role in collagen fibrillogenesis. Proceedings of the National Academy of Science of the USA 69, 313–317. Dai, H., Hafner, J., Rinzler, A., Colbert, D., Smalley, R., 1996. Nanotubes as nanoprobes in scanning probe microscopy. Nature 384, 147–150. Gale, M.A., Pollanen, M.S., Markiewicz, P., Goh, M.C., 1995. Sequential assembly of collagen revealed by atomic force microscopy. Biophysical Journal 68, 2124–2128. Goh, M.C., Paige, M.F., Gale, M.A., Yadegari, I., Edirisinghe, M., Strzelczyk, J., 1997. Fibril formation in collagen. Physica A 239, 95–102. Kadler, K., Holmes, D., Trotter, J., Chapman, J., 1996. Collagen fibril formation. Biochemical Journal 316, 1–11.
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361 Keller, D., 1991. Reconstruction of STM and AFM images distorted by finite-size tips. Surface Science 253, 353–364. Kobayashi, K., 1992. Electron microscopic visualization of collagen aggregates without chemical staining. Journal of Electron Microscopy 41, 235–241. Kuhn, K., 1982. Segment-long-spacing crystallites, a powerful tool in collagen research. Collagen Related Research 2, 61–80. Layman, D., McGoodwi, E., Martin, G., 1971. The nature of the collagen synthesized by cultured human fibroblasts. Proceedings of the National Academy of Science of the USA 68, 454–458. Markiewicz, P., Goh, M., 1994. Atomic force microscopy probe tip visualization and improvement of images using a simple deconvolution procedure. Langmuir 10, 5–9. Nimni, M., 1988. Collagen, CRC Press, Boca Raton, CA. Paige, M.F., Rainey, J.K., Goh, M.C., 1998. Fibrous long spacing collagen
361
ultrastructure elucidated by atomic force microscopy. Biophysical Journal 74, 3211–3216. Schmitt, F., Gross, J., Highberger, J., 1953. A new particle type in certain connective tissue extracts. Proceedings of the National Academy of Science of the USA 39, 459–470. Timpl, R., Martin, G., Bruckner, P., Wick, G., Wiedemann, H., 1978. Nature of the collagenous protein in a tumor basement membrane. European Journal of Biochemistry 84, 43–52. Trelstad, R., Hayashi, K., Gross, J., 1976. Collagen fibrillogenesis: Intermediate aggregates and suprafibrillar order. Proceedings of the National Academy of Science of the USA 73, 4027–4031. Veis, A., George, A., 1994. Fundamentals of Interstitial Collagen Selfassembly, Academic Press, San Diego, CA. Wegner, A., Engel, J., 1975. Kinetics of the cooperative association of actin to actin filaments. Biophysical Chemistry 3, 215–225.
Micron 32 (2001) 355–361 www.elsevier.com/locate/micron
Ultrastructure and assembly of segmental long spacing collagen studied by atomic force microscopy M.F. Paige, M.C. Goh* Department of Chemistry, University of Toronto, 80 St. George Street, Toronto, Ontario, Canada M5S 3H6
Abstract The in vitro formation of segmental long spacing (SLS) collagen as induced by the addition of ATP to acidified Type I collagen solutions has been examined with the atomic force microscope (AFM). AFM images obtained suggest that the assembly proceeds in a stepwise manner, through an intermediate stage of oligomers, which then associate laterally to form the so-called “SLS crystallites”. Attempts to induce SLS formation by the addition of other polyanionic species to monomeric collagen solutions met with mixed success; ATP-g-S and GTP produced SLS crystallites, whereas inorganic phosphate and other polyanionic dyes did not. This indicates that the formation of SLS cannot simply be attributed to the negation of positive charges believed to be located on the end of the collagen monomer, but rather it is a complex function of the structure and charge of both the collagen monomer and polyanion. 䉷 2000 Elsevier Science Ltd. All rights reserved. Keywords: Collagen; Atomic force microscope; SLS collagen; Assembly; Mechanism
1. Introduction Aggregation of type I collagen typically leads to the formation of fibrils with a characteristic banding pattern of about 67 nm period (Nimni, 1988). The in vitro assembly of these native type fibrils has been examined using a variety of techniques and is reviewed in a number of papers (Veis and George, 1994; Kadler et al., 1996; Holmes, et al., this issue). The effects of solvent conditions such as concentration, temperature and ionic strength on the final morphology of the products and on the process as a whole have also been investigated (Trelstad et al., 1976; Goh et al., 1997). The conditions under which the assembly proceeds can have a drastic effect on the kinetics and the final structure of the aggregate. While most collagen assembly products are fibrillar, the addition of dilute, acidified solutions of adenosine triphosphate (ATP) to monomeric collagen gives rise to a final collagen aggregate that is completely different in morphology, one which is block-like rather than fibrillar. This was first noted by Schmitt et al. (1953), who named the aggregates segmental-long-spacing (SLS) collagen or SLS crystallites. Electron microscopy (EM) measurements show that these SLS collagen aggregates have approximately the same length as the monomer units (⬃280 nm) and have a welldefined internal banding pattern perpendicular to their * Corresponding author. Tel.: ⫹1-416-978-6254; fax: ⫹1-416-978-6254. E-mail address: [email protected] (M.C. Goh).
lengths. The proposed model for the structure of SLS collagen is thus of monomers aligned in register, as if in a crystalline lattice, and the observed banding pattern was interpreted to correspond with the distribution of amino acid residues along the collagen polypeptide. Kuhn (1982) showed that there is an excellent correlation between this banding pattern and the location of the positively charged regions in the known sequence of the monomer. Given this model for the structure of the SLS aggregate, the main interest in it and its in vitro formation is its utility to the overall study of collagen monomers (Kuhn, 1982): SLS crystallites have been utilized as a means of measuring lengths of various collagen monomers; for inferring the amino acid composition of different collagen types (Timpl et al., 1978); and for characterization of various collagen precursors (Layman et al., 1971). In vivo, structures similar to SLS aggregates have been reported and investigated in association with the intracellular translocation of procollagen molecules (Bruns et al., 1979). Efforts have been made towards understanding the nature of the association between ATP and the collagen monomers in SLS collagen. To reduce possible effects of heavy metal staining on the banding patterns of the aggregates, Kobayashi (1992) studied the structure of the aggregates using the incorporated ATP for visualization, instead of traditional EM staining techniques. The EM band patterns obtained in this manner were well correlated with the distribution of charged basic residues in the monomer. A strong band
0968-4328/00/$ - see front matter 䉷 2000 Elsevier Science Ltd. All rights reserved. PII: S0968-432 8(00)00038-X
356
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361
initial reversible aggregation of monomers until a specific size (the ‘critical nucleus’) is reached. Beyond the critical nucleus, further growth is achieved by the addition of the monomers to the critical nucleus. The atomic force microscope (AFM), which is capable of close to atomic resolution, has previously been shown a useful tool in the examination of collagen ultrastructure and assembly (Gale et al., 1995; Goh et al., 1997; Paige et al., 1998). In this paper, we utilize the AFM to investigate the formation of SLS collagen by visualization of the assembly products at early stages of the process. The images we obtain suggests that the process proceeds, at least in part, in a step-wise fashion. We also investigate the role of ATP in comparison with other negatively charged ions in inducing SLS formation.
2. Materials and methods 2.1. In vitro SLS collagen assembly
Fig. 1. (a) and (b): AFM images of SLS collagen aggregates formed via addition of ATP to acidified collagen monomers, taken in deflection mode. Product was diluted by 100 × prior to imaging. For (a), image size is 15 × 15 mm 2; for (b), image size is 5 × 5 mm 2.
near one end corresponded with the presence of a distinct positively charged region near the N-terminus of the monomer, as shown by the calculations of Veis and George (1994) based upon collagen’s known amino acid sequence. These results led to the suggestion that the ATP molecules were located within the aggregate and acted as bridging units between the basic residues of adjacent monomers. It has thus been suggested that the role of ATP in SLS formation is to inhibit the electrostatic repulsion between similarly charged regions of monomers aligned in register, which are preventing the crystallization. While it is generally accepted that SLS aggregates are composed of laterally aligned monomers, their mechanism of aggregation has received minimal attention. The proposed ordered structure, and the term “SLS crystallite”, suggests that they form in a manner similar to that of typical crystals. In this case, there is nucleation step, which is an
Type I calf-skin collagen (Sigma) was dissolved over ice in 0.05% acetic acid, with occasional sonicating to facilitate the breakdown of collagen aggregates. The mixture was centrifuged at 10,000 rpm for 60 min at 4⬚C. After centrifugation, the supernatant was filtered through 0.45 mm Millipore filters (Sigma) and mixed with the appropriate polyanion solution. Solutions of the polyanions were prepared by dissolving the solid (ATP, ADP, GTP, Amaranth, Cibacron 3GA from Sigma; ATP-g-S from Calbiochem) in 0.05% acetic acid. Collagen and polyanion solutions were combined at room temperature to yield mixtures with final collagen concentrations typically of ⬃0.5 mg/ml, polyanion concentrations of ⬃2 mg/ml and a final pH of 3.5. The mixtures were allowed to equilibrate overnight at room temperature. A series of dilutions ranging from 10–1000 fold were prepared by adding aliquots of the reaction mixture to an appropriate volume of Millipore water. The diluted samples were deposited onto freshly cleaved sheets of mica in 20 ml aliquots and allowed to dry for an hour prior to imaging in the AFM.
2.2. Atomic force microscopy Samples were imaged with a Nanoscope III instrument (Digital Instruments, Santa Barbara, CA), typically using a square pyramidal silicon nitride tip of nominal spring constant ⬃0.58 N/m. Tips produced by electron beam deposition were also used in attempts to improve resolution. Images were taken in contact mode, in air. Images obtained were consistently reproducible, showing minimal perturbation from the probe tip. Both height and force mode images were taken simultaneously, with height measurements being taken from height mode images only.
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361
Fig. 2. (a)–(c): Sectional analysis of SLS aggregate. Part (a) shows longitudinal section of SLS aggregate. Part (b) shows diameter measurement. Part (c) shows width measurement.
3. Results and discussion 3.1. Structural studies When monomeric collagen solutions were combined with ATP, ATP-g-S or GTP according to the preparation described above, white, turbid solutions were obtained. Analysis of the resulting mixtures by means of the AFM revealed an abundance of SLS aggregates. A typical AFM image of a 100-fold diluted sample obtained from an ATPcollagen preparation is shown in Fig. 1a and b. The AFM images of SLS aggregates were consistent with the EM results described by Schmitt et al. (1953) in terms of their overall structure. The aggregates appeared rectangular, with a height of 30–50 nm above the flat mica substrate. The lateral dimensions are ⬃360 nm by 200–300 nm, with a much larger spread in the latter value than in the former. Note that because AFM images are a convolution of the
357
sample and probe tip geometry (Keller, 1991; Markiewicz and Goh, 1994), the lateral dimensions are an overestimate of the actual ones. Sectional analysis of a typical aggregate is shown in Fig. 2. We shall ascribe the dimension with less variance (⬃360 nm) to correspond with the long axis of the monomer (plus tip geometry), in accordance with previous model of monomers arranged in register. Measurements taken along this collagen monomer axis will be referred to as the longitudinal section, while perpendicular measurements (shown in Fig. 2c) will be called the cross section. It should be noted that the smallest dimension of the block-like aggregate (30– 50 nm, shown in Fig. 2b) is always that above the substrate. This is probably the result of a positive interaction between the aggregate and the mica substrate, causing the aggregates to always lie flat in order to maximize the area of contact with the substrate. The various sections clearly indicate that the aggregates are not cylindrically symmetric. A closer examination of Fig. 2 shows that the longitudinal section is almost rectangular with a steep slope, while the cross section shows a less steep slope, with an obvious rounding at the apex. These are consistent with the previous assignment: the sharp slope in the longitudinal section can be ascribed to the ends of the aligned monomers, which would terminate more abruptly. If one were to view the SLS aggregate as a small crystal, faceting would be expected, instead of the observed rounded cross section. This could be taken as an argument against a completely crystalline packing within the SLS aggregate. However, there is an alternative possibility that this rounding is caused by the interaction between SLS and the mica substrate which is much too strong, and enough to deform a weakly crystalline aggregate. We are unable to resolve this issue at this point. With the use of electron beam deposited tips of high aspect ratio, finer structural features on the surface of an aggregate can be resolved. The longitudinal-section, shown in Fig. 2a, indicates the presence of a series of minor grooves on the surface of the aggregate. These features generally consisted of a higher ridge on one end of the aggregate, which is ⬃2.5 nm in height above the average aggregate surface. A series of smaller ripples along the section are also present, the resolution of which is possibly limited by the finite width and aspect ratio of the AFM tip. The presence of these ultrastructural features are in accord with the results from EM (Kuhn, 1982), and are presumably caused by the alignment of corresponding amino acids within the monomers. Unfortunately, we were unable to obtain better resolution in order to make a quantitative comparison; AFM probes with much smaller tip radii, such as those constructed from carbon nanotubes (Dai et al., 1996) may be useful for this characterization. 3.2. Assembly via intermediates By changing the concentration of collagen and ATP, the
358
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361
Fig. 3. AFM images of SLS aggregates in intermediate stage of formation. (a) Height mode image (scan size: 6 × 6 mm 2); (b) Amplitude mode image (scan size: 6 × 6 mm 2); (c) Zoom in of (a); image size (3 × 3 mm 2); and (d) Zoom in of (b), image size (3 × 3 mm 2).
kinetics of the assembly process can be slowed down enough so we can quench the system at various stages before fully developed SLS crystallites are formed. While we are not yet able to follow the assembly in situ, we can derive information about the process by examination of various assembly products that are observed on many different trials. Examples of images of these aggregates are shown in Fig. 3, where we can see a variety of structures, ranging from objects that appear fibrous to mature SLS crystallites. We can classify the patterns of structures we observed into a few distinct types, and divide the apparent stages of assembly into four different steps, each labelled by a letter (a)–(d). We do not claim that these patterns correspond to distinct and specific stages of the assembly; however, they do provide a clue in inferring how the assembly proceeds. The first pattern of structure that emerges is indicated by the (a), and shows an abundance of fibrous structures
dispersed over a relatively wide length scale. These fibrous structures have the same approximate length of a monomer, but a larger cross-section. Measurements of over 30 such entities yielded typical diameters of 2.5 nm, in contrast with ⬃1 nm for the monomer, indicating that they are oligomers composed of at least several monomers in register. Even at this early stage in the assembly, there appears to be some long-range ordering of the oligomers present; while they are distinct, independent entities, a definite lateral alignment amongst them is evident. In pattern (b), we see that a set of oligomers that are closer to one another and are beginning to merge, creating markedly larger bodies. Regions of the substrate between these larger bodies can still be resolved, however, indicating that the assembly process is still incomplete. In (c), the larger structures have fused and packed together to the point that the substrate can no longer be seen. Sectional analysis of structures, at this stage, show bumps and ridges which are not
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361
359
Fig. 4. Illustration of proposed assembly mechanism for SLS aggregates. Assembly involves the formation of stable oligomer intermediates, followed by merging of these intermediates to form the final aggregates.
apparent in the fully formed SLS aggregate. Presumably, the formation of a mature SLS involves further packing of the aggregate constiutents until only the smallest of ultrastructural features can be distinguished. Pattern (d) corresponds to what is essentially a fully mature SLS aggregate, having dimensions and ultrastructure typified by that shown in Fig. 2. While we are not able to follow the actual evolution of a single SLS aggregate, the existence of distinct patterns indicate that the formation of SLS crystallites occurs, at least to some extent, in a stepwise manner. The repeated occurrence of distinct features indicate that there are stable intermediates in the process, instead of a pure nucleation and growth, wherein we would expect to see is a wide distribution of sizes of the block-like SLS aggregate and no other pattern. We postulate that SLS formation occurs in the following fashion: in an as-of-yet unobserved step, monomers combine to form fibrous oligomers. These oligomers are stable intermediates, which then associate laterally. They are then compacted, and ultimately fuse to form the final crystallite. This is illustrated schematically in Fig. 4, with the labeled axes on the final aggregate corresponding to the sections shown in Fig. 2.
3.3. Role of ions in the assembly The role of ATP in the assembly process has been previously speculated upon, and the general view is that its negative charges serve to neutralize the positive charges on the collagen monomers, which serve to counteract aggregation (Veis and George, 1994). However, our attempts to induce SLS aggregation by the addition of a myriad of anionic reagents, including inorganic phosphate (in the form of KH2PO4) and ADP, met with uniform failure, indicating that simple charge neutralization is not the sole requirement. On the other hand, the addition of GTP to monomeric collagen produced aggregates whose features are indistinguishable from those obtained with ATP. These results with GTP, ATP and ADP may suggest that the aggregation is activated by the hydrolysis of the phosphate bond, similar to the case for the ATP-assisted assembly of actin filaments, which involves an irreversible hydrolysis of a phosphate group on a G-actin-associated ATP complex, followed by a reversible exchange of ADP against ATP on a monomeric G-actin unit (Wegner and Engel, 1975). In order to examine this possibility, we used ATP-g-S, which is a non-hydrolyzable ATP analogue. This
360
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361
Kobayashi (Kobayashi, 1992) has suggested that the role of the polyanion is to act as a bridging unit between basic residues on the laterally aligned monomer. The results we obtained suggest that not only are three negatively charged groups needed, but that they have to be arranged in a particular way with respect to one another. It is also possible that the ability of the polyanion to participate in hydrogen bonding may play a significant role as well, since both adenosine and guanosine functionalities are well known for this ability.
4. Summary
Fig. 5. (a)–(c): Chemical structures of polyanionic dyes and ATP; (a) Amaranth; (b) Cibacron blue 3GA; and (c) Adenosine 5 0 triphophosphate (ATP).
preparation produced SLS aggregates with no observable structural difference from those produced by the addition of ATP or GTP, indicating that the role of the polyanion is not to provide assistance with the energetics of the assembly by means of exothermic degradation. The fact that aggregation occurs with either ATP, ATP-g-S, or GTP, but not with ADP, indicates that assembly requires the presence of three adjacent negatively charged groups, but that it is unaffected by small structural alterations in the nitrogenous base of these polyanions. To further investigate the role of charged polyanions in the assembly, we examined the effect of two dyes, Cibacron blue 3GA and Amaranth (structures are shown along with that of ATP in Fig. 5a–c). Both of these dyes possess three negatively charged groups. In both cases, SLS aggregation was not induced, although heavy, fibrous precipitate of indeterminate structures were formed. This suggests that while a high local concentration of negative charge can yield aggregation, SLS formation is by no means inevitable; the amount and type of charge is clearly an important prerequisite for SLS formation, but the manner in which it is distributed within a polyanion appears to also play a crucial role.
SLS collagen aggregates have been formed in vitro and studied by means of the AFM. Measurements of the aggregate yield typical longitudinal and cross sections of and 360 and 200–300 nm, respectively, which are consistent with results obtained from electron microscopy. Electron beamdeposited AFM tips have revealed small ultrastructural features on the surface of the aggregate, but the small size of these features have made them difficult to resolve. Images of SLS aggregates in the early stages of the formation process have been obtained. The aggregates appear to form by the lateral addition of collagen oligomers which are ⬃2.5 nm in height, followed by the packing and fusing together of these intermediates. This suggests that the assembly of SLS collagen proceed by means of a stepwise mechanism, with the oligomers being stable intermediates in the process. Additionally, it was found that SLS aggregation could be induced by addition of ATP-g-S and GTP but not with various other polyanions. The structural details of the supporting polyanion, and perhaps its ability to accept and donate hydrogen bonds, appear to be important factors dictating SLS aggregate formation.
Acknowledgements This work was supported by grants from the Natural Science and Engineering Research Council of Canada, Photonics Research Ontario and the ACS-PRF.
References Bruns, R., Hulmes, D., Therrien, S., Gross, J., 1979. Procollagen segmentlong-spacing crystallites: their role in collagen fibrillogenesis. Proceedings of the National Academy of Science of the USA 69, 313–317. Dai, H., Hafner, J., Rinzler, A., Colbert, D., Smalley, R., 1996. Nanotubes as nanoprobes in scanning probe microscopy. Nature 384, 147–150. Gale, M.A., Pollanen, M.S., Markiewicz, P., Goh, M.C., 1995. Sequential assembly of collagen revealed by atomic force microscopy. Biophysical Journal 68, 2124–2128. Goh, M.C., Paige, M.F., Gale, M.A., Yadegari, I., Edirisinghe, M., Strzelczyk, J., 1997. Fibril formation in collagen. Physica A 239, 95–102. Kadler, K., Holmes, D., Trotter, J., Chapman, J., 1996. Collagen fibril formation. Biochemical Journal 316, 1–11.
M.F. Paige, M.C. Goh / Micron 32 (2001) 355–361 Keller, D., 1991. Reconstruction of STM and AFM images distorted by finite-size tips. Surface Science 253, 353–364. Kobayashi, K., 1992. Electron microscopic visualization of collagen aggregates without chemical staining. Journal of Electron Microscopy 41, 235–241. Kuhn, K., 1982. Segment-long-spacing crystallites, a powerful tool in collagen research. Collagen Related Research 2, 61–80. Layman, D., McGoodwi, E., Martin, G., 1971. The nature of the collagen synthesized by cultured human fibroblasts. Proceedings of the National Academy of Science of the USA 68, 454–458. Markiewicz, P., Goh, M., 1994. Atomic force microscopy probe tip visualization and improvement of images using a simple deconvolution procedure. Langmuir 10, 5–9. Nimni, M., 1988. Collagen, CRC Press, Boca Raton, CA. Paige, M.F., Rainey, J.K., Goh, M.C., 1998. Fibrous long spacing collagen
361
ultrastructure elucidated by atomic force microscopy. Biophysical Journal 74, 3211–3216. Schmitt, F., Gross, J., Highberger, J., 1953. A new particle type in certain connective tissue extracts. Proceedings of the National Academy of Science of the USA 39, 459–470. Timpl, R., Martin, G., Bruckner, P., Wick, G., Wiedemann, H., 1978. Nature of the collagenous protein in a tumor basement membrane. European Journal of Biochemistry 84, 43–52. Trelstad, R., Hayashi, K., Gross, J., 1976. Collagen fibrillogenesis: Intermediate aggregates and suprafibrillar order. Proceedings of the National Academy of Science of the USA 73, 4027–4031. Veis, A., George, A., 1994. Fundamentals of Interstitial Collagen Selfassembly, Academic Press, San Diego, CA. Wegner, A., Engel, J., 1975. Kinetics of the cooperative association of actin to actin filaments. Biophysical Chemistry 3, 215–225.