Proc. Natl. Acad. Sci. USA
Vol. 76, No. 12, pp. 6226-6230, December 1979 Biochemistry
In vitro assembly of intermediate filaments from baby hamster kidney (BHK-21) cells (10-nm filaments/protein polymerization/decamin)
ROBERT V. ZACKROFF AND ROBERT D. GOLDMAN Department of Biological Sciences, Carnegie-Mellon University, 4400 Fifth Avenue, Pittsburgh, Pennsylvania 15213
Communicated by Edward A. Adelberg, September 17, 1979
ABSTRACT Intermediate filaments (IF) from baby hamster kidney (BHK-21) cells can be disassembled at low ionic strength and reassembled upon addition of salt. Turbidimetric analyses show that reassembled IF exhibit the light scattering properties of long rods under physiological conditions (5 mM Na+/K+ phosphate, pH 7.2/170 mM NaCI at 21°C). IF weight concentration, determined by centrifugation, is directly proportional to the optical density at 300 nm. Thus, turbidity can be used as a quantitative assay for IF assembly. Turbidimetric and centrifugation analyses both indicate that IF assembly exhibits a critical protein concentration of 0.05-0.15 mg/ml. Above the critical concentration, IF weight concentration at steady-state is a linear function of the total protein concentration. Negative stain observations at early stages of the assembly process suggest lateral association of protofilaments to form short IF. This lateral association is accompanied by a rapid turbidity increase which is then followed by IF elongation and 4 slower turbidity increase to plateau. Further purification of IF by low/highNaCI-induced cycles of disassembly/reassembly results in retention of 54- and 55-kilodalton (decamin) polypeptides. These results constitute a quantitative description of in vitro reassembly of IF from homogeneous cultures of nonkeratinizing cells and establish conditions for further studies on the regulation of IF assembly.
Intermediate (n-10 nm) filaments (IF), actin-containing microfilaments, and microtubules are the three major proteinaceous fiber systems comprising the cytoskeleton and cytomusculature of mammalian cells. Studies of in vitro assembly of actin filaments (1) and microtubules (2) have contributed greatly to an understanding of how these cytoplasmic fibers function in vivo (3). IF have been much less extensively characterized, and, with the exception of epidermal keratin (4), quantitative procedures for studying IF reassembly in vitro have not been reported. Many of the difficulties surrounding IF studies have stemmed from the relative resistance of IF to solubilization. In most cases, solubilization has been achieved only after treatment at extremes of pH (5, 6) or in the presence of protein denaturing agents (4, 7-10). Nevertheless, there is evidence that IF undergo assembly, disassembly, and reorganizational changes during the cell cycle (11). The means by which these processes are regulated remain unknown. IF obtained from different cells and tissues appear to differ in their immunological properties, molecular weight, and number of subunits. Despite these differences, IF from different cell types appear to be structurally and chemically related, and all of those characterized appear to be highly a-helical proteins of the k-m-e-f class (12). It remains to be demonstrated which similarities and differences are associated with IF structural proteins and which may be due to contaminating or accessory proteins. Thus, in vitro systems would be useful in characterThe publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate
izing the assembly and structure of IF. This laboratory has developed procedures for the rapid purification of IF from cultures of mammalian cells (13, 14). These IF can be disassembled at low ionic strength into protofilaments that reassociate upon addition of salt to form IF structurally identical to those found in vio (13, 15). In this paper, we report the establishment of conditions for quantitatively studying the in vitro reassembly of IF from baby hamster kidney (BHK-21) cells. The initial characterization of the assembly process is also presented. MATERIALS AND METHODS BHK-21/C13 cells were grown as described (16) and IF were isolated by a modification of the procedure of Starger et al. (13). Confluent roller bottles were rinsed with three 15-ml changes of phosphate-buffered saline without Ca2+ or Mg2+ (6 mM Na+/K+ phosphate, pH 7.4/0.171 M NaCl/3 mM KCl). Lysing solution (10 ml per bottle) was added and the cell remains were detached by rolling the bottles at room temperature. The lysing solution consisted of 0.6 M KC1, 1% Triton X-100, 10 mM MgCl2, 0.5 mM phenylmethylsulfonyl fluoride (PhMeSO2F), 0.5 mg of p-tosyl-L-arginine methylester HC1 per ml in Ca2+/Mg2+-free phosphate-buffered saline. The cell remains were further disrupted with one or two strokes of a motordriven Teflon pestle at low speed in a tight-fitting glass homogenizer. DNase I (Sigma) was added to a final concentration of 0.5 mg/ml and the solution was incubated for 5 min on ice. The solution was then centrifuged at 1600 X g for 5 min at 40C. The resulting IF pellet was washed three times by centrifugation for 3 min at 1600 X g (4°C) in Ca2+/Mg2+-free phosphate-buffered saline containing 5 mM EDTA and 0.5 mM PhMeSO2F. This was followed by two washes in 5 mM Na+/K+ phosphate/0. 1 mM PhMeSO2F at pH 7.2-7.6. This pellet was resuspended in a small volume of the last buffer (approximate protein concentration, 10 mg/ml) and the resulting suspension was dialyzed against 200 vol of that buffer for 12-18 hr at 18°C. The solution was then centrifuged at 22,000 rpm for 30 min at 15°-18°C (Beckman type 65 rotor). The supernatant was removed from below a white lipidlike layer with a syringe and further clarified at 55,000 rpm for 1 hr at 15°-18°C. The resulting supernatant contained 20-50% of the protein present prior to clarification; this protein was in the form of soluble protofilaments (see Fig. 4 Upper Left). An additional 0.1 mM PhMeSO2F was added to these solutions, which were used immediately for all turbidity experiments. For some experiments in which the protein was further purified by cycles of disassembly/reassembly, the isolation procedure was altered slightly to help reduce proteolysis. In vitro reassembly of IF was initiated by addition of either Abbreviations: PhMeSO2F, phenylmethylsulfonyl fluoride; IF, intermediate filaments.
this fact. 6226
Biochemistry: Zackroff and Goldman
Proc. Natl. Acad. Sci. USA 76 (1979)
4 vol of ice-cold fixation medium (1.5% glutaraldehyde/5 mM Na+/K+ phosphate, pH 7.6). Fixed samples were placed on formvar and carbon-coated copper grids, rinsed with one or two drops of distilled water, stained for 20-30 sec with 3% uranyl acetate, blotted, and air dried. Grids were photographed by using a Philips 201C electron microscope. Sodium dodecyl sulfate/polyacrylamide gel electrophoresis was performed on 7.5% acrylamide slab gels with 4.5% stacking gels according to the procedure of Laemmli (17). Protein concentrations were determined by the procedure of Itzhaki and Gill (18) with lyophilized BHK-21 IF protein as a standard.
To1 0 -1
2.7 2.6 log X FIG. 1. Wavelength exponent of reassembled IF. Assembly was performed in 0.17 M NaCl/5.0 mM Na+/K+ phosphate, pH 7.2/0.2 mM PhMeSO2F. The wavelength exponent (= slope) was determined by linear regression. Protein concentration, 0.45 mg/ml.
NaCl or potassium phosphate at 210C to final concentrations of 0.17 M or 0.095 M, respectively. Polymer turbidity was monitored with a Zeiss PM-6 spectrophotometer. Turbidity was corrected for absorbance due to protein by using blanks in which no salt, which induces assembly, was added. Samples were prepared for negative staining by dilution into I I I
RESULTS AND DISCUSSION In order to characterize assembly, we sought a method by which the weight concentration of reassembled IF could be rapidly and reliably measured. Although procedures such as viscometry and flow birefringence may be -applicable, turbidimetric analyses possess distinct advantages in that they do not physically perturb the system and can be performed without use of specialized equipment. Turbidity has been used successfully to measure the weight concentration of long rodlike polymers such as microtubules (19), fibrin (20), and keratin (4). The turbidity, r, of a solution of macromolecules is a function of X-w, where X is the wavelength of incident light and w is called the wavelength exponent. Thus: T X:w-w (20). The wavelength exponent is a complicated function of particle size and shape (21). For long rodlike polymers of small radius and of diameter sufficiently greater than the wavelength of light employed, w = 3 and the turbidity is expected to be a linear function of polymer weight concentration (22). Deviations of w from the theoretically derived value of 3 may be due to the presence of particles of other shapes or to solution nonideality (21). The wavelength exponent of reassembled IF was found to be strongly dependent upon the pH and ionic strength of the reassembly buffer (unpublished results). At 0.17 M Nab or K+ (pH 7.2), a wavelength exponent very close to the theoretical value of 3.0 was obtained (Fig. 1). Under these conditions, the turbidity of reassembled IF was a linear function of the polymer weight concentration when assembly was carried out at relatively low (below 1.5 mg/ml) protein concentrations (Fig. 2). Therefore, these conditions were used for turbidimetric analysis of the assembly reaction. Having established conditions under which turbidity reliably measures IF weight concentration, we correlated the kinetics of turbidity development during IF reassembly with the ultrastructure of assembly intermediates. The turbidity increased instantaneously after addition of salt. This initial turbidity burst
400 600 Pelleted protein, gg/ml
FIG. 2. Polymer weight concentration dependence of turbidity. Assembly was induced in samples at various protein concentrations in 5 mM Na+/K+ phosphate/0.2 mM PhMeSO2F by addition of NaCl (0, &) or potassium phosphate (3) to final monovalent ion concentrations of 0.17 M (pH 7.2). Each set of symbols represents experiments performed with different preparations of purified protein. Polymer weight concentration (shown on the abscissa) was determined from the amount of protein pelleted by centrifugation at 38,000 rpm for 45 min (Beckman type 65 rotor).
Time, min FIG. 3. Kinetics of IF turbidity development. Assembly was initiated in 5.0 mM Na+/K+ phosphate/0.2 mM PhMeSO2F by addition of NaCl to a final concentration of 0.17 M (pH 7.2). The initial portion of the curve was too rapid to monitor and is shown as a dashed line. Protein concentration, 1.1 mg/ml.
Biochemistry: Zackroff and Goldman
Proc. Natl. Acad. Sci. USA 76 (1979)
FIG. 4. Ultrastructure of IF reassembly. Conditions were identical to those in Fig. 7. (Upper Left) Protofilaments prior to addition of salt. (Upper Right) Short IF, 1 min after addition of NaCl to 0.17 M. (Lower) Long IF, 30 min after addition of salt. Samples were fixed in 1.5% glutaraldehyde prior to negative staining. (X45,000.)
was followed by a much slower rise to plateau (Fig. 3). The biphasic turbidity curve suggested the occurrence of two stages in IF assembly. Examination of negatively stained preparations also supported the idea of a two-step process. Before addition of salt, protofilaments of less than one-half the diameter of IF were present (Fig. 4 Upper Left). Within 1 min after addition of salt, short IF were observed (Fig. 4 Upper Right). Therefore, the appearance of short IF was correlated with the initial rise in turbidity. After 30 min, IF elongation took place (Fig. 4 Lower), which was concomitant with the slow phase of the increase in turbidity. Because turbidity due to long rods is a linear function of the
fiber mass-to-length ratio (20), it appears likely that the initial turbidity burst is due to the rapid diameter increase seen by electron microscopy. These data strongly suggest a rapid lateral association of protofilaments to form short IF, followed by their slower elongation. It is noteworthy that the wavelength exponent of samples containing only protofilaments (i.e., before addition of salt) is close to 3, confirming the negative stain images that indicate that the assembling subunits are also rodshaped (Fig. 4 Upper Left). The rapid formation of short IF, which appear to serve as nuclei for subsequent elongation, is consistent with a nucleation-condensation model of polymer assembly as formalized
Zackroff and Goldman I
Proc. Natl. Acad. Sci. USA 76 (1979)
1.2 E I
0~ 0.8 -mm,
-I I 1.2 Total protein, mg/ml
FIG. 7. Purification of IF by disassembly/reassembly. IF were isolated as described in Materials and Methods, with the exception that 1 mM PhMeSO2F was present in all buffers used in the isolation and purification scheme described below. Freshly isolated IF (lane a) were dialyzed for 4 hr at 40C against 5 mM K phosphate (pH 7.2) (disassembly buffer) and centrifuged for 1 hr at 55,000 rpm at 40C. The pellet (lane b) was discarded, and the supernatant (lane c) was made 0.095 M in K phosphate (pH 7.2) and incubated for 1 hr at 210C to reassembly IF. Reassembled IF were pelleted by centrifugation at 38,000 rpm for 45 min at 21'C. The supernatant was discarded and the reassembled IF pellet (lane e) was resuspended in disassembly buffer and incubated at 41C for 2 hr. The preparation was then clarified by centrifugation at 55,000 rpm for 1 hr at 40C. The clarified supernatant (lane e) was made 0.095 M in K phosphate (pH 7.2) and incubated for 1 hr at 21IC to reassembly IF. The pellet from the above clarification (lane d) was discarded. The twice-in vitro-reassembled IF were then pelleted by centrifugation at 38,000 rpm for 45 min at 21'C (lane g). Sodium dodecyl sulfate/polyacrylamide gel electrophoresis was performed with approximately 15 sg of protein applied per lane. All centrifugations were performed in a Beckman type 65 rotor. The arrows indicate the major 54- and 55-kilodalton (decamin) polypeptides.
slope of such a plot is equal to the fraction of protein that is inactive in the assembly reaction, and the vertical intercept is equal to the true critical concentration corrected for this inactive component (24). The slope shown in Fig. 5 is equal to 0.2, indicating that 80% of the protein participates in the assembly reaction. The ability to reversibly reassemble IF enabled us to further
by Oosawa and Kasai (23). In such a reaction, one would predict: (i) the existence of a critical subunit protein concentration which must be exceeded before stable nuclei can support elongation, and (ii) above this critical concentration, steadystate polymer weight concentration should be a linear function of the total protein concentration. The IF weight concentration was found to vary linearly with total protein concentration, and a critical concentration between 0.05 and 0.15 mg/ml was obtained by using ultracentrifugation (Fig. 5) or turbidimetric (Fig. 6) assays. In Fig. 5, the unpolymerized protein concentration has also been plotted vs. total protein concentration. The
FIG. 5. Ultracentrifugation analysis of protein concentration dependence of IF assembly. IF were reassembled at various protein concentrations in 0.17 M NaCl/5 mM Na+/K+ phosphate, pH 7.2/0.2 mM PhMeSO2F. The weight concentration of IF (pelleted protein) (0) and the concentration of unpolymerized (supernatant) protein (A) were determined by centrifugation at 38,000 rpm for 45 min (Beckman type 65 rotor).
FIG. 6. Turbidimetric analysis of the protein concentration de-
pendence of IF assembly. IF were reassembled at various protein concentrations, and the plateau turbidity was plotted as a function of total protein concentration. Conditions were identical to those in Fig. 5.
purify them by cycles of low/high-salt-induced disassembly/ reassembly (Fig. 7). We accomplished this by assembling IF at high ionic strength, pelleting them, resuspending the pellets in low-ionic-strength buffer to depolymerize the IF, and clarifying the resulting suspensions to remove insoluble material. This procedure is analogous to that now in standard use for purifying microtubule protein (25). During such cycling 7080% of the solubilized protein was generally recovered in the pellet of reassembled IF. The fraction of protein recovered after each subsequent solubilization and clarification was more variable and ranged between 20 and 50%. Thus, the fraction of isolated IF protein recovered as two-cycled IF ranged between 2 and 10%. The use of longer incubation times in the solubilization steps favored greater recovery of protein in the supernatants after clarification. However, this also caused increased proteolysis. Conditions were therefore chosen in which proteolysis was minimized, but sufficient protein could still be recovered for further purification cycles.
Biochemistry: Zackroff and Goldman
Minor polypeptides, in addition to the major 54- and 55kilodalton (decamin) bands, were present in freshly isolated IF (Fig. 7). However, fewer minor polypeptides were retained along with decamin after two disassembly/reassembly cycles. These included high molecular weight protein (>250 kilodaltons) and some lower molecular weight material (Fig. 7). The lower molecular weight bands were not always present in fresh preparations, and they increased during storage in the absence of inhibitors of proteolysis. Therefore, these bands appeared to be largely, if not completely, the result of proteolysis of decamin or the high molecular weight material. The high molecular weight material appeared to be associated with IF through at least two purification cycles. However, intact high molecular weight proteins were not required for assembly because IF without these polypeptides were obtained after cycling at lower PhMeSO2F concentrations or longer solubilization times, during which more extreme proteolysis occurred (not shown). In summary, we have established conditions for quantitative analysis of in vitro assembly of IF from BHK-21 cells. Evidence was presented that IF assembly is a two-step, nucleation-condensation phenomenon, involving lateral association of protofilaments to form short IF, followed by their elongation. In addition, we have employed the ability to assemble and disassemble IF in vitro to demonstrate the association of specific polypeptides with IF. These studies provide a starting point from which cellular factors involved in IF assembly, regulation, and function can be systematically studied in a purified in vitro system. This work was supported by a grant from the National Science Foundation. R.V.Z. is a Fellow in Cancer Research supported by Grant DRG-316-F of the Damon Runyon-Walter Winchell Cancer Fund. 1. Oosawa, F. & Asakura, A. (1975) Thermodynamics of the Polymerization of Protein, eds. Hbrecker, F., Kaplan, N. O., Mumur, J. & Scheraga, H. (Academic, New York). 2. Weisenberg, R. C. (1972) Science 177, 1104 1105.
Proc. Natl. Acad. Sci. USA 76 (1979) 3. Hatano, S. & Owaribe, K. (1976) Cell Motility, Cold Spring Harbor Conferences on Cell Proliferation, eds. Goldman, R., Pollard, T. & Rosenbaum, J. (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY), Vol. 3, Book A, pp. 499-511. 4. Steinert, P. M., Idler, W. W. & Zimmerman, S. B. (1976). J. Mol. Biol. 108,547-567. 5. Small, J. V. & Sobieszek, A. (1977) J. Cell Sci. 23, 243-268. 6. Matolsty, A. (1975) J. Invest. Dermatol. 65, 127-142. 7. Davison, P. F. & Winslow, B. (1974) J. Neurobiol. 5, 119-133. 8. Cooke, P. (1976) J. Cell Biol. 68, 539-556. 9. Izant, J. G. & Lazarides, E. (1977) Proc. Natl. Acad. Sci. USA 74, 1450-1454. 10. Culbertson, V. & Freedberg, I. (1977) Biochim. Biophys. Acta 490, 178-191. 11. Goldman, R., Berg, G., Bushnell, A., Chang, C., Dickerman, L., Hopkins, N., Miller, M. L., Pollack, R. & Wang, E. (1973) in Locomotion of Tissue Cells, Ciba Foundation Symposia, eds. Porter, R. & Fitzsimons, D. W. (Assoc. Sci. Publ., New York). Vol. 14, pp. 83-107. 12. Goldman, R. D., Milsted, A., Schloss, J. A., Starger, J. & Yerna, M.-J. (1979) Annu. Rev. Physiol. 41, 703-722. 13. Starger, J., Brown, W., Goldman, A. & Goldman, R. (1978) J. Cell Biol. 78, 93-109. 14. Starger, J. & Goldman, R. (1977) Proc. Natl. Acad. Sci. USA 74, 2422-2426. 15. Steinert, R. M., Zimmerman, S. B., Starger, J. M. & Goldman, R. D. (1978) Proc. Natl. Acad. Sci. USA 75,6098-6101. 16. Goldman, R. (1971) J. Cell Biol. 51, 752-762. 17. Laemmli, U. K. (1970) Nature (London) 227,680-685. 18. Itzhaki, R. & Gill, D. (1964) Anal. Biochem. 9,401-410. 19. Gaskin, F., Cantor, C. R. & Shelanski, M. L. (1974) J. Mol. Biol. 84,739-758. 20. Carr, M. E. & Hermans, J. (1978) Macromolecules 11, 46-50. 21. Carmerini-Otero, R. D. & Day, L. A. (1978) Biopolymers 17, 2241-2249. 22. Berne, B. J. (1974) J. Mol. Biol. 89, 755-758. 23. Oosawa, F. & Kasai, M. (1962) J. Mol. Biol. 4,10-21. 24. Borisy, G. G., Marcum, M. M., Olmsted, J. B. & Johnson, K. A. (1975) Ann. N.Y. Acad. Sci. 253, 107-132. 25. Shelanski, M. L., Gaskin, F. & Cantor, C. R. (1973) Proc. Natl. Acad. Sci. USA 70,765-768.