Stability and folding of the tumour suppressor protein p161

Article No. jmbi.1998.2420 available online at http://www.idealibrary.com on

J. Mol. Biol. (1999) 285, 1869±1886

Stability and Folding of the Tumour Suppressor Protein p16 Kit S. Tang, Benjamin J. Guralnick, Wooi Koon Wang, Alan R. Fersht and Laura S. Itzhaki* Centre for Protein Engineering Medical Research Council Hills Road, Cambridge CB2 2QH, UK

The tumour suppressor p16 is a member of the INK4 family of inhibitors of the cyclin D-dependent kinases, CDK4 and CDK6, that are involved in the key growth control pathway of the eukaryotic cell cycle. The 156 amino acid residue protein is composed of four ankyrin repeats (a helix-turn-helix motif) that stack linearly as two four-helix bundles resulting in a non-globular, elongated molecule. The thermodynamic and kinetic properties of the folding of p16 are unusual. The protein has a H2 O , of 3.1 kcal molÿ1 at 25  C. very low free energy of unfolding,
GDÿN The rate-determining transition state of folding/unfolding is very compact (89 % as compact as the native state). The other unusual feature is the very rapid rate of unfolding in the absence of denaturant of 0.8 sÿ1 at 25  C. Thus, p16 has both thermodynamic and kinetic instability. These features may be essential for the regulatory function of the INK4 proteins and of other ankyrin-repeat-containing proteins that mediate a wide range of protein-protein interactions. The mechanisms of inactivation of p16 by eight cancer-associated mutations were dissected using a systematic method designed to probe the integrity of the secondary structure and the global fold. The structure and folding of p16 appear to be highly vulnerable to single point mutations, probably as a result of the protein's low stability. This vulnerability provides one explanation for the striking frequency of p16 mutations in tumours and in immortalised cell lines. # 1999 Academic Press

*Corresponding author

Keywords: p16; p16

INK4A

Introduction Progression through the eukaryotic cell cycle is regulated by the activities of the cyclin dependent kinase (CDK) family of proteins, a key control point being the G1 to S transition when the cells commit themselves to replicate DNA. Two cyclin D-dependent kinases, CDK6 and CDK4, become activated at this transition and phosphorylate the retinoblastoma gene product (pRb), which then allows the transcription factor E2F to activate several genes associated with S phase progression (for a review, see Weinberg, 1995). The INK4 proAbbreviations used: CD, circular dichroism: CDK, cyclin dependent kinase; DTT, dithiothreitol; DTE, dithioerythritol; EDTA, ethylenediaminetetraacetic acid; GST, glutathione-S-transferase; PPIase, peptidyl prolyl isomerase; SEC, size-exclusion chromatography; [urea], concentration of urea. E-mail address of the corresponding author: [email protected] 0022-2836/99/041869±18 $30.00/0

; ankyrin; tumour suppressor; protein folding

teins bind to and inhibit speci®cally CDK4 and CDK6, and thus negatively regulate cell cycle progression. p16INK4A (referred to subsequently as p16) is a member of the INK4 family and is mutated in many types of tumours (Pollock et al., 1996; Ragione et al., 1997; Shapiro et al., 1995; SmithSorensen & Hovig, 1996). Loss of p16 function is second only to inactivation of p53 as the most frequent event observed in human tumours. Many of the mutations in p16 are inactivating point mutations and they are spread throughout the sequence (Cordon-Cardo, 1995; Koh et al., 1995; Ranade et al., 1995; Yang et al., 1995). Serrano et al. (1993) and Kamb et al. (1994) ®rst identi®ed the gene for p16, and demonstrated that p16 can inhibit the activity of CDK4. There is considerable evidence for its tumour suppressor activity, the most compelling of which are the observations that mice lacking p16 develop tumours early (Serrano et al., 1996), and that germline mutations of p16 are found in patients with familial melanoma and # 1999 Academic Press

1870 co-segregate with predisposition to this disease (Hussussian et al., 1994; Ranade et al., 1995); p16 may also play a critical role in the senescence mechanism (Alcorta et al., 1996; Loughran et al., 1996; Serrano et al., 1996): p16 mRNA has a very low turnover (Hara et al., 1996), and expression of p16 in primary cell cultures is maximal as they approach the limit of cell divisions. Further, it has been shown that senescence breaks down following p16 inactivation (Serrano et al., 1996). The INK4 protein family includes, in addition to p16, p15INK4b (Hannon & Beach, 1994), p18INK4c (Guan et al., 1994), and p19INK4d (Chan et al., 1995); p15 and p16 have 80 % sequence identity whereas p18 and p19 are 40-50 % identical with p16 and with one another. The four proteins are biochemically and functionally indistinguishable with respect to their interactions with CDK4 and CDK6 in vitro, but only p16 has been linked directly to tumours. This suggests that the other INK4 proteins may have non-redundant and cellspeci®c functions (Sherr, 1996; Sherr & Roberts, 1995). The INK4 proteins have a series of 33 amino acid residue repeats named ``ankyrin'', that were originally found in the protein ankyrin (Lux et al., 1990). Subsequently, they have been identi®ed in many proteins of diverse function, including many nuclear cell cycle regulators (Bork, 1993; Greenwald & Rubin, 1992; Helps et al., 1995), and they occur in at least four consecutive copies in each protein. The ankyrin repeat regions of several proteins are involved in protein-protein interactions, and the diversity makes it unlikely that the motif has a highly specialised function. The ankyrin repeat consists of a helix-turn-helix motif, with successive repeats linked by a b-hairpin in the structures of p53-binding protein, 53BP2 (Gorina & Pavletich, 1996), p19 (Luh et al., 1997) and p18 (Venkataramani et al., 1998). The helices between two neighbouring repeats interact to form four-helix bundles, and the b-strands interact to form an antiparallel b-sheet, and the plane of the b-strand regions is approximately orthogonal with the axis of the helical segments. The b-sheet in 53BP2 forms along the entire length of the protein, whereas in p18 this is limited to inter-repeat structures; p18 and p19 both have an additional ankyrin repeat compared with p16 (Byeon et al., 1998), and p16 appears to be more conformationally ¯exible than p18 and p19. Speci®cally, the antiparallel b-hairpin structure was not detected in p16. The greatest sequence diversity between the INK4 proteins is in the b-hairpin region, and this region of the fourth repeat in 53BP2 interacts with p53 (Gorina & Pavletich, 1996). Unlike other CDK inhibitor proteins, p16 is able to bind to CDK4 in the absence and in the presence of cyclin D (Guan et al., 1994; Hirai et al., 1995; Parry et al., 1995; Serrano et al., 1993; Zindy et al., 1997). Thus, p16 has been thought to function by displacing the cyclin D subunit and/or by direct inhibition of the CDK catalytic site. Recent structures of p16 and p19 bound CDK6 (Russo et al.,

Stability and Folding of p16

1998) and of p19 bound to CDK6 (Brotherton et al., 1998) are consistent with these two mechanisms of action. The INK4 proteins bind next to the ATP binding site of the catalytic cleft, opposite the site where the activating cyclin subunit is expected to bind, and they cause structural changes that could indirectly prevent cyclin binding. They also distort the kinase catalytic cleft and thus interfere with ATP binding, and this explains the observation that they can inhibit the pre-assembled CDK-cyclin complexes. Studies have demonstrated that p16 is relatively unstable and suggested that its structure is highly ¯exible (Boice & Fairman, 1996; Byeon et al., 1998; Tevelev et al., 1996; Zhang & Peng, 1996). We present a detailed analysis of the thermodynamics and kinetics of folding of p16. The results may provide insights into the ankyrin repeat motif and its role in biology. We also describe a comprehensive assay for the structural and thermodynamic effects of point mutations of p16. Studies have shown that some cancer-associated mutations have reduced helical structure and increased propensity to aggregate (Byeon et al., 1998; Tevelev et al., 1996; Zhang & Peng, 1996). Our approach gives a detailed description of the conformational behaviour of the mutant proteins. The results provide an alternative point of view with which to discuss the behaviour of the INK4 proteins observed in vivo.

Results The original cDNA clone of p16 was N-terminally truncated by eight amino acid residues (Serrano et al., 1993), and later the full-length sequence of 156 residues was discovered (Hannon & Beach, 1994). The truncated form has the same activity as the full-length protein (Ranade et al., 1995; Serrano et al., 1993; Yang et al., 1995). We have expressed full-length p16 as a glutathione-Stransferase (GST) fusion protein and removed the GST tag, leaving a glycine and serine at the N terminus of p16. The thermodynamics and kinetics of folding of full-length p16 were measured and were identical with those of the truncated form (data not shown). This result is consistent with the structure of p16 which shows that the ®rst 13 residues of the full-length protein are unstructured and do not form part of an ankyrin repeat (Byeon et al., 1998). The truncated version could be expressed at signi®cantly higher levels (Tevelev et al., 1996), and therefore the studies below have used this form of the protein. Equilibrium stability of p16 The denaturation of p16 was monitored using several probes in order to determine whether any intermediate species are populated at equilibrium. Studies of p16 have used circular dichroism in the far-UV region to monitor loss of helical structure upon chemical denaturation (Boice & Fairman, 1996; Tevelev et al., 1996; Zhang & Peng, 1996).

Stability and Folding of p16

The urea-induced denaturation curve, measured by far-UV CD in 50 mM Tris-buffer (pH 7.5) at 25  C, is shown in Figure 1(a). The data can be ®tted to a two-state model (equation (3)), according to which only native and denatured states are populated, and there are no partly folded species. p16 unfolds reversibly as judged by complete regain of the native CD spectrum. The midpoint of denaturation is 1.87(0.01) M urea, the m-value is 1.7(0.1) kcal molÿ1 Mÿ1, and the free energy H2 O , is 3.1(0.1) kcal of unfolding in water,
GDÿN ÿ1 mol . Another simple conformational probe, frequently used to monitor tertiary structure, is ¯uorescence spectroscopy. p16 has two tryptophan residues, at positions 15 and 110. The ¯uorescence spectrum of p16 has a maximum intensity at 355 nm, indicating that the tryptophan side-chains are fully exposed to solvent, in agreement with other other studies (Boice & Fairman, 1996; Byeon et al., 1998; Tevelev et al., 1996; Zhang & Peng, 1996). However, unlike

Figure 1. (a) Urea-induced denaturation of p16 at 25  C monitored by (a) near-UV CD, in 50 mM Tris (pH 7.5) (open circles), and in 15 mM sodium phosphate buffer (pH 7.5), 350 mM sodium sulphate (®lled circles). The protein concentration is 5 mM. The cell pathlength was 0.1 cm. The data are ®tted to a two-state model. (b) Fluorescence in 50 mM Tris (pH 7.5). The excitation wavelength is 280 nm and the emission wavelength is 355 nm. The protein concentration is 3 mM. The data are ®tted to a two-state model.

1871 these studies, a decrease in intensity was observed when the protein was unfolded in 4 M urea; there was no shift in the wavelength maximum. The ¯uorescence at 355 nm is plotted as a function of [urea] in Figure 1(b). There is a co-operative decrease in ¯uorescence between 0 M and 4 M urea, and an approximately linear increase of similar magnitude above 4 M urea. Thus, the change in ¯uorescence intensity upon unfolding is so small that the change within the baseline is magni®ed. However, the data ®t to a two-state equation, and the midpoint of denaturation of 1.98(0.07) M and m-value of 1.7(0.3) kcal molÿ1 Mÿ1 are in excellent agreement with the CD experiment. Size-exclusion chromatography provides another probe of tertiary structure, since it measures the hydrodynamic volume of a protein. The elution volume of the native state of a protein should be larger than that of the denatured state and, if the equilibrium between the two is on a slower time scale than the time of chromatography, both species can be separated as individual elution peaks under the appropriate denaturant concentration (Corbett & Roche, 1984; Uversky, 1993). If the equilibrium is fast, only one peak is observed with an elution volume that is a weighted-mean value of that for the population of the folded and unfolded forms present at equilibrium. The latter behaviour was observed for p16. Size exclusion chromatography was performed in 15 mM sodium phosphate buffer at pH 7.5, containing 100 mM NaCl. p16 elutes as a single peak at a volume of 16.4 ml on an analytical S200 column, which corresponds to a molecular mass of 23 kDa. This is larger than the expected molecular mass of the truncated p16 (15.8 kDa), but this discrepancy can be explained by the non-globular, elongated shape of the protein. The plot of elution volume versus urea concentration is shown in Figure 2, together with a far-UV CD denaturation curve performed under identical buffer conditions. The elution volume decreases with increasing urea concentration in a linear fashion at low urea concentrations. Similar behaviour for other proteins has been interpreted as non-speci®c hydrophobic interaction of the protein with the support (Corbett & Roche, 1984; Sanz & Fersht, 1993). The size-exclusion chromatography data can be ®tted to a two-state equation (assuming a highly sloping pre-transition baseline), giving a midpoint of denaturation of 2.05(0.07) M, an m-value of 1.7(0.2) kcal molÿ1 ÿ1 2O Mÿ1, and
GH DÿN of 3.6(0.4) kcal mol . The free energy of unfolding is the same within error as that obtained by far-UV CD under the same ÿ1 2O The conditions (
GH DÿN ˆ 3.7(0.2) kcal mol ). increase in stability of approximately 0.5 kcal molÿ1 is not a result of the phosphate buffer that was used instead of Tris buffer, but rather was due to the presence of 100 mM NaCl. Thus, p16 is stabilised by salt. The coincidence of the secondary structure probe and the two tertiary structure

1872

Stability and Folding of p16

Figure 2. Urea-induced denaturation of p16 monitored by size-exclusion chromatography using a Superdex-200 column in the presence of 100 mM NaCl. Plot of elution volume versus [urea] (®lled circles), and comparison with ellipticity at 222 nm (measured under identical buffer condition; open circles). The protein concentration is 25 mM for the SEC and 5 mM for CD. The data are ®tted to a two-state model.

probes indicate that the protein unfolds at equilibrium in a two-state manner. Effect of added solutes on the stability of p16 A number of salt solutions were tested to determine whether they stabilise p16. Sodium chloride has a small but signi®cant stabilising effect at a concentration of 0.1 M, as described above. The same concentration of sodium sulphate has a greater stabilising effect (data not shown). The ureainduced denaturation curve is shown at a concentration of 0.35 M sodium sulphate (Figure 1). The m-value (1.8(0.1) kcal molÿ1 Mÿ1) is unchanged, but the midpoint of denaturation increased 2O to 2.9(0.02) M.
GH DÿN was calculated to be 5.3(0.2) kcal molÿ1.

The urea dependence of the rate constant of unfolding shows slight downward curvature over the wide range of urea concentrations from 3 M and 8 M, and the data ®t better to a second order polynomial of the type: ln ku ‰UŠ ˆ ln kuH2 O ‡ mku ‰UŠ ‡ mku ‰UŠ2

Unfolding kinetics Unfolding was accompanied by a large and rapid decrease in ¯uorescence. Under strong unfolding conditions outside the transition region, the kinetic trace could be ®tted to a single exponential function, representing a single unfolding phase (Figure 3(a)). The observed rate constant increased with increasing urea concentration outside the transition region (Figure 4(a)), according to the commonly observed equation: ln ku ‰UŠ ˆ ln kuH2 O ‡ myÿN ‰UŠ

Figure 3. Kinetics of unfolding of native p16 (in 50 mM Tris (pH 7.5)) in 4.8 M urea (®nal concentration). (a) Decrease in ¯uorescence, excitation at 280 nm and total emission above 320 nm; (b) loss of negative ellipticity at 222 nm, using a cell of pathlength 0.2 cm. The ellipticity of native protein (in 50 mM Tris (pH 7.5)) is indicated by a square. The protein concentration after mixing is 2 mM for the ¯uorescence and 8 mM for the CD. The data are ®tted to a single exponential function.

…1†

where kuH2 O is the rate constant of unfolding in water and m{ ÿ N is a constant that is proportional to the increase in solvent accessible surface area between the native state and the transition state for unfolding.

…2†

Similar non-linearity has been observed for a number of other proteins, and has been interpreted as a result of movement of the transition state for unfolding according to Hammond behaviour, and rigorously tested in barnase by mutation (Matouschek & Fersht, 1993; Matouschek et al., O 1994). The values of kH u 2 , mku and mku* determined using equation (2), are 0.8(0.1) sÿ1, 0.33(0.03) Mÿ1, and ÿ0.02(0.003) Mÿ2, respectively. Thus, the kinetic m-value, m{ ÿ N, is denaturant-dependent: mzÿN ˆ mku ÿ mku ‰UŠ Stopped-¯ow CD at a wavelength of 222 nm was also used to monitor unfolding (Figure 3(b)).

1873

Stability and Folding of p16

Figure 3(b), showing that the amplitude of the observed kinetics accounts for the entire ellipticity change expected for unfolding. Refolding kinetics The CD spectrum of p16 in 30 mM HCl shows that the protein is essentially unstructured in acid (data not shown). Refolding was initiated by mixing the acid-denatured protein with buffer to give ®nal conditions of 50 mM Tris (pH 7.5), and urea concentrations up to 2 M, and was accompanied by a large increase in ¯uorescence (Figure 5). Under strongly native conditions outside the transition region, the data could be ®tted to the sum of three exponentials, representing a fast-refolding

Figure 4. Urea dependence of unfolding (®lled symbols) and refolding (open symbols) kinetics measured by ¯uorescence. The major, fast-refolding phase is represented by circles, the slower refolding phase by squares, and the slowest-refolding phase by triangles. (a) Rate constants. The continuous curve is the rate constant calculated from equilibrium and kinetic unfolding data for a two-state model. (b) Relative amplitudes of the refolding phases, calculated as a percentage of the total ¯uorescence change observed at each urea concentration. (c) The ¯uorescence reading at long folding/ unfolding times.

The loss of negative ellipticity could be ®tted to a single exponential function with the same rate constant as that determined by ¯uorescence. The ellipticity of native protein is also plotted in

Figure 5. Kinetics of refolding of acid-denatured protein (in 30 mM HCl) into 50 mM Tris buffer (pH 7.5; ®nal conditions), monitored by (a) increase in ¯uorescence. Inset, slowest-refolding phase from 2-200 seconds. (b) Gain of negative ellipticity at 222 nm, using a cell of pathlength 0.2 cm. The ellipticity of denatured protein in 30 mM HCl is indicated by a square. The data are ®tted to a double exponential function, representing the major, fast-refolding phase and the minor, slowest-refolding phase. Inset, fast-refolding phase from 0-0.5 second. The ®nal protein concentration was 2 mM for the ¯uorescence experiment and 8 mM for the CD experiment.

1874 phase accounting for 60 % of the amplitude, a slow-refolding phase accounting for 5 % of the amplitude, and slowest-refolding phase accounting for 35 % of the amplitude (Figure 4(b)). The rate constants of the fast phase decreased with increasing urea concentration, as shown in Figure 4(a). The rate constants of the intermediate and slow phases exhibited a weaker denaturant dependence. There was no dependence of the rate constants and their relative amplitudes on the protein concentration in the range 0.5 to 8 mM, showing that the observed kinetics do not arise from oligomerisation or aggregation. Further, identical kinetics were observed when urea was added to the aciddenatured protein to a concentration of 2 M, or when refolding was initiated by dilution of denaturant starting from protein denatured in 3.5 M urea. When the baseline ¯uorescence at long folding/unfolding times is plotted versus [urea], it describes well the equilibrium unfolding transition (Figure 4(c)). The slow-refolding reaction is limited by proline isomerisation Multiple phases in refolding kinetics can arise from intermediates accumulating on a sequential pathway, or from parallel folding pathways. The latter may be a result of heterogeneity in the denatured state arising when proteins contain proline residues, because some of the unfolded molecules may have peptidyl-proline bonds in a non-native conformation, and isomerisation of these bonds can be the rate-limiting event in the folding of these molecules to the native state (Brandts et al., 1975; Nall, 1994; Schmid, 1992). The ratio of trans to cis isomers in the denatured state depends on the residue that precedes the proline (Dyson et al., 1988a,b; Grathwohl & WuÈthrich, 1976), but studies of other proteins have estimated it to be at least 5:1 (Jackson & Fersht, 1991a; Kato et al., 1981; Kiefhaber & Schmid, 1992a; Matouschek et al., 1990; Ridge et al., 1981). There are ten proline residues in p16 which, by analogy with its homologues, all adopt the trans conformation. Cyclophilin catalyses the cis-trans isomerism of many peptidyl-proline bonds (Lang et al., 1987; Lin et al., 1988). When acid-denatured p16 was refolded in the presence of an equimolar concentration of cyclophilin, the amplitudes of the three refolding phases were approximately the same as in the absence of cyclophilin. The rate constants of the fast and slow phases were unchanged, but the rate of the slowest phase was accelerated approximately ®vefold (Table 1). This suggests that this phase represents refolding of a denatured species that has a non-native peptidyl-proline bond, isomerisation of which to the native form is the ratelimiting step for folding. The weak denaturant dependence of the rate constant for this phase is consistent with this.

Stability and Folding of p16 Table 1. Catalysis of refolding of p16 by cyclophilin A

k1 (sÿ1) Percentage of amplitude (1) k2 (sÿ1) Percentage of amplitude (2) k3 (sÿ1) Percentage of amplitude (3)

No cyclophilin A

Cyclophilin A (1 mM)

33.4 60 2.6 4 0.07 36

33.3 63 1.7 8 0.3 29

All rate constants were measured for refolding of 1 mM (®nal) p16 from 30 mM HCl to ®nal buffer conditions of 50 mM Tris buffer (pH 7.5), at 25  C.

Double-jump experiments to detect slow isomerisation processes in the denatured state In a double-jump refolding experiment, native protein is unfolded under conditions in which the reaction is fast, and then transferred to refolding conditions before proline isomerisation can take place. Thus, it should be possible to identify kinetic complexity arising from a heterogeneous population of slowly interconverting unfolded species (Brandts et al., 1975; Nall et al., 1978; Schmid, 1986a; Schmid & Baldwin, 1978). p16 was unfolded by rapid mixing of native protein in water with an equal volume of 60 mM HCl. Unfolding was measured under these conditions and is rapid (approximately 200 sÿ1). After 250 ms, at which time the protein was >99 % unfolded, refolding was initiated by mixing with an equal volume of 100 mM Tris (pH 8.1), 2 mM EDTA, to give conditions identical with those for the simple folding experiment. The kinetics of refolding of transiently unfolded protein could be ®tted to a double exponential function, with rate constants and relative amplitudes that were the same as those of the fast and slow phases observed in refolding of equilibrium-unfolded protein. Protein that was allowed to unfold for 50 seconds refolded with kinetics identical with that of equilibrium-unfolded protein. This result provides further evidence that the slowest-refolding pathway is produced by a slow equilibration process (proline isomerisation) in the denatured state. The rate constant for this phase is somewhat faster than that expected for proline isomerisation in an unstructured peptide (Brandts et al., 1975). Acceleration of proline isomerisation by the formation of structure has been observed in model peptides and in some proteins (Drakenburg et al., 1972; Grathwohl & WuÈthrich, 1981; Jackson & Fersht, 1991b; Schmid, 1986b; Tan et al., 1997). It is likely, therefore, that there is some residual structure around the proline in the denatured state of the protein. The slow-refolding phase has a rate constant of approximately 2.5 sÿ1 and, like the slowest-refolidng phase, also exhibits only a weak denaturant dependence. Minor folding phases with similar time constants (1-10 sÿ1) have been observed for

1875

Stability and Folding of p16

several other proteins (Khorasanizadeh et al., 1996; Kragelund et al., 1995; Rousseau et al., 1998; Walkenhorst et al., 1997), and it is likely that proline isomerisation in a partly folded state gives rise to this phase. Double-jump experiments to detect nativelike intermediates The ¯uorescence-detected refolding reaction may correspond to the formation of a native-like intermediate that is spectroscopically indistinguishable from the native state, and that is converted to the native state in an undetected step. It is assumed that an intermediate state would have a lower activation barrier for unfolding than would the native state, and would therefore unfold faster. The following double-jump unfolding experiment allows formation of the native state, speci®cally, to be monitored (Kiefhaber, 1995; Schmid, 1983, 1986a). Acid-denatured protein was mixed with an equal volume of 100 mM Tris (pH 8.1), 2 mM EDTA. The ¯uorescence-detected fast-refolding reaction is complete within 200 ms and the slow reaction within 50 seconds. Therefore, refolding was allowed to proceed for times, tf, of 200 ms and 50 seconds before unfolding was initiated by mixing with an equal volume of 9.6 M urea. Irrespective of the refolding time, the unfolding kinetics could be ®tted to a single exponential function with a rate constant equal to that measured for unfolding of native protein. This result shows that the ¯uorescence-detected refolding reaction does monitor formation of the native state, and not formation of a partly folded intermediate. The amplitude of the unfolding reaction gives the fraction of native protein present at the time when refolding was interrupted. The amplitude of unfolding when tf ˆ 200 ms was approximately 60 % of the amplitude of unfolding when tf ˆ 50 seconds, in good agreement with the relative amplitudes of the fast and slow ¯uorescencedetected refolding phases. Thus, the kinetics observed by ¯uorescence monitor the proportions of native molecules folding in fast and slow reactions. Rate-limiting proline isomerisation events during refolding of p16 There are ten proline residues in p16, which by analogy with its homologues, all adopt the trans conformation in the folded protein. Proteins that have cis proline residues in the native state often fold slowly because the trans conformation is usually the major form in the denatured state and, therefore, the slow isomerisation steps are rate-limiting in the refolding of the majority of molecules. We have shown that the transiently denatured protein, that should have all proline residues in their native conformations, refolds at the same rate as the major, fast phase observed by ¯uorescence and CD for refolding of equilibrium-denatured protein.

Further, the fast phase was shown to represent the formation of the native state, and not the formation of a native-like intermediate. Therefore, the fastrefolding phase most likely corresponds to folding of the fraction of denatured protein that has all its proline residues in the native conformations. Two minor, slower-refolding phases could be detected. The slowest phase has a rate constant of 0.07 sÿ1, exhibits only a weak denaturant dependence, and is ef®ciently catalysed by PPIase (peptidyl prolyl isomerase). These properties are characteristic of a proline isomerisation reaction occurring in the denatured state. Identification and characterisation of a kinetic folding intermediate A two-state analysis was applied to the fast refolding phase that corresponds to folding of denatured protein that has all its proline residues in the native conformation. The method has been described in detail (Jackson & Fersht, 1991a; Matouschek et al., 1990). Figure 4(a) shows the comparison between kobs and k calculated according to two-state behaviour. The experimental rate constants of the fast-refolding reaction are in good agreement with the calculated rate constants at urea concentrations of greater than 1.5 M, but the experimental rate constants are slower than the calculated ones at lower urea concentrations. We can conclude, therefore, that the folding process of p16 does not ®t to a simple two-state model, but there is at least one intermediate state that accumulates in the refolding kinetics under strongly native conditions (Jackson & Fersht, 1991a; Matouschek et al., 1990). The intermediate is formed in a fast step, and the observed refolding rate constant is that for the formation of the native state from the intermediate: fast

slow kÿu

D ÿ! ÿN ÿ I ÿ! ku

In order to characterise the intermediate further, the refolding kinetics were monitored using the ellipticity at 222 nm (Figure 5(b)). The data could be ®tted to the sum of two exponential processes. The amplitude and rate constant of the major phase was 6.6(0.1) millidegrees and 30.7(0.4) sÿ1, respectively, for refolding into buffer alone. The amplitude and rate constant of the minor phase were 2.1(0.2) millidegrees and 0.14(0.01) sÿ1, respectively. The two rate constants are similar to those of the fastest and slowest-refolding phases observed by ¯uorescence. The intermediate-refolding phases that is observed by ¯uorescence could not be ®tted to the CD data. This may be because the percentage amplitude of this phase is very small and the CD data are too noisy to ®t to a triple exponential function. Measurements were also made at urea concentrations of 0.5 M, 1 M and

1876

Stability and Folding of p16

Figure 6. Far-UV CD spectra of mutant proteins. (a) D74N (), P81L (^) and D84N (!). (b) R24P (*), H98P (^), P114L (~) and V126D (‡). The spectrum of wild-type is included for comparison (*).

1.5 M, and the rate constants obtained by ®tting the data to the sum of two exponentials were in good agreement with the fast and slowest phases of ¯uorescence data. Extrapolation of the kinetic curves through the instrument dead-time of 7-8 ms gives an ellipticity at zero time of ÿ10.6(0.2) millidegrees, and a total amplitude for the reaction of 10.6(0.2) millidegrees. The ellipticity of denatured p16 in 30 mM HCl is ÿ8.2 millidegrees (Figure 5(b)), and the total ellipticity change on refolding is 13 millidegrees. Thus, any intermediate that may be formed in an initial, rapid step that is too fast to be detected in the stopped-¯ow apparatus would have only a small amount of helical structure, if any, given the experimental errors in the measurements. Structural model of p16 The structures of p16 (Byeon et al., 1998) and p18 (Venkataramani et al., 1998) have the same global fold, with similar Ca backbones. The only differences are ®rst, that p18 and p19 have an additional ankyrin repeat, and second, the antiparallel b-hairpin structure in the loops of p18 and p19 was not

Figure 7. Schematic representation (created using the programme MOLSCRIPT) of the model of p16, derived from the Ca co-ordinates of p18 that were kindly provided by R. Marmorstein, The Wistar Institute, USA. The positions of the mutations analysed in this study are shown in red.

observed in p16, possibly because of the conformational ¯exibility in these regions of p16. The coordinates of p16 have not been released, but a model of p16 was constructed using the p18 coordinates, kindly provided by R. Marmorstein (The Wistar Institute, USA). The model contains four ankyrin repeats. Each exists as a b-strand helixturn-helix b-strand. The overall structure gives an extended L-shaped molecule. The eight residues, mutation of which are presented here, were mapped to the model (Figure 7). Refolding assay Wild-type and mutant p16 were expressed in inclusion bodies in Escherichia coli, and the protein was therefore puri®ed under denaturing conditions and subsequently refolded by dialysis into a refold-

Stability and Folding of p16

1877

ing buffer, as described in Materials and Methods. We used a number of probes to assay refolding. Preparative size-exclusion chromatography (SEC) was used to detect high-order aggregates. No aggregation or oligomerisation was detected upon refolding of wild-type p16, although oligomerisation was observed in other studies (Boice & Fairman, 1996; Tevelev et al., 1996). Non-aggregated fractions of mutant proteins were collected and analysed further, as follows: analytical SEC and ¯uorescence spectroscopy were used to monitor the integrity of the tertiary structure. The Stoke's radius that is obtained from this experiment gives an estimate of the molecular dimensions of the protein. The secondary structure content was determined by measuring the CD spectrum in the far-UV region. The conformational properties were then probed further by studying the response to urea. Q50R The mutant Q50R is a germline familial melanoma mutation (Walker et al., 1995). It has not been functionally tested. Q50 is located in the ®rst fourresidue helix of the second ankyrin repeat (Figure 7), and the side-chain interacts with residues in the ®rst helix of the third repeat. The mutant protein aggregated completely upon refolding (data not shown). H98P, P114L and V126D H98 is located in the second helix of the third repeat, and its side-chain interacts with residues in the second helix of the fourth repeat. The mutation H98P was isolated from primary melanoma, and failed to arrest G1 growth or to inhibit cyclinCDK4 kinase activity (Koh et al., 1995). P114 is located in the ®rst helix of the fourth repeat, and its side-chain interacts with residues in the ®rst helix of the third repeat. The mutation P114L was identi®ed in melanoma cell line (Kamb et al., 1994) was unable to bind CDK4 or CDK6 (Harland et al., 1997). V126D, a germline mutation in the second helix of the fourth repeat, was detected in several unrelated families with familial melanoma (Hussussian et al., 1994). The side-chain interacts with the second helix of the third repeat. An eightfold excess of the protein was required for the inhibition of cyclin-CDK4 activity and completely failed to inhibit cyclin D1-CDK6 activity (Ranade et al., 1995), and it was unable to bind CDK4 and CDK6 at the higher temperature of 42  C (Parry & Peters, 1996). For all three mutants, approximately 50 % of the sample formed high-order aggregates upon refolding. Non-aggregated proteins were isolated and analysed further. All the mutants had disrupted helical structure, as shown by the reduced ellipticity minimum at 222 nm when compared with wild-type p16 (Figure 6(a)). This agrees with previous results (Tevelev et al., 1996; Zhang & Peng,

Figure 8. Urea-induced denaturation monitored by far-UV CD at 222 nm. The protein concentration was 5 mM in 15 mM sodium phosphate (pH 7.5), 100 mM NaCl and 1 mM DTE at 25 C. The wild-type data (*) are included for comparison. (a) The data for mutants R24P (*), H98P (^), P114L (~) and V126D (‡) are shown for clarity with a smoothing function. (b) The data for D74N (), P81L (^) and D84N (!) were ®tted to a two-state equation, from which the thermodynamic parameters in Table 1 were derived.

1996). Only weakly co-operative transitions were observed upon addition of urea (Figure 8(a)). Further, there was no change in the ¯uorescence spectra of the mutant proteins on addition of urea, indicating a lack of folded tertiary structure (data not shown). The elution volumes of the mutant proteins correspond with Stoke's radii that are more than 50 % greater than that of wild-type p16 (Table 2), and the volumes decreased in a non-co-operative manner with increasing [urea], indicating the absence of cooperative tertiary structure (Figure 9). The elution volumes of the mutants in the absence of urea decreased with increasing protein concentration, showing that the proteins were not monomeric. The sulfate ion stabilises some proteins. Wildtype p16 is stabilised by 2.2 kcal molÿ1 by the

1878

Stability and Folding of p16

Table 2. Conformational and thermodynamic parameters of p16 wild-type and mutants p16 variants Wt R24P Q50R D74N P81L D84N H98P P114L V126D

Extent of aggregationa

Relative Stoke's radiusb

Relative helical contentd

me (kcal molÿ1 Mÿ1)

[Urea]e50 % (M)

H2 O e
GDÿN (kcal molÿ1)

ÿ ‡ ‡‡ ÿ ÿ ÿ ‡ ‡ ‡

1 1.3 1 1.2 0.9 -c -c -c

1 0.7 1.1 1 1.2 0.6 0.7 0.6

2.0(0.1) 1.2(0.1) 1.1(0.3) 2.0(0.2) -

1.9(0.01) 1.4(0.05) 3.4(0.2) 1.8(0.03) -

3.7(0.1) 1.7(0.2) 3.7(0.2) 3.6(0.1) -

a

No aggregation (ÿ), partial aggregation (‡) and complete aggregation (‡‡) upon refolding. The Stoke's radius of the mutant relative to that of the native state of wild-type. Measurements were made at 25  C at a protein concentration of 25 mM in 15 mM sodium phosphate (pH 7.5), 1 mM DTT and 100 mM sodium chloride. c Values for these mutants are not given because they were dependent on protein concentration. d Comparative helical content relative to that of the native state of wild-type p16, calculated from the ellipticity minimum at /ewt 222 nm using the expression emutant 222 222. e Calculated using far-UV CD measurements at 222 nm at 25  C. The protein concentration was 5 mM in 15 mM sodium phosphate (pH 7.5), 1 mM DTT and 100 mM NaCl. b

presence of 350 mM sodium sulphate (1.8 kcal molÿ1 when compared with the buffer used for the mutant proteins, which contains 100 mM NaCl). The CD spectra of the three mutants in this buffer were unchanged in the presence of 350 mM sodium sulfate; however, analytical SEC revealed that the proteins formed high order aggregates under these conditions (data not shown). Proline residues are known helix-breakers, and therefore the substitution of H98 with a proline may disrupt the helical structure in p16. Proline

residues frequently initiate the a-helices in the INK4 proteins. Indeed, proline is completely conserved at the start of the ®rst helix of the fourth repeat (P114) and appears to be conserved for stability, since substitution of the proline with a leucine in p16 results in aggregation and misfolding. V126 in p16 appears also to be completely conserved for the same reason. R24P R24P is a germline familial melanoma mutation located in the turn of the ®rst repeat (Holland et al., 1995). It was able to bind CDK6 but not CDK4 (Harland et al., 1997). Some aggregation was observed for this mutant upon refolding. The monomeric fraction had a larger Stoke's radius than the wild-type (Figure 9), and a somewhat lower helical content (Figure 6(a)). The helical structure has a co-operative transition with increasing [urea] (Figure 8(a)). A co-operative transition was also observed by SEC. But even in 0 M urea, the mutant protein is partly unfolded at room temperature, and this would explain the larger Stoke's radius. P81L

Figure 9. Size-exclusion chromatography of wild-type p16 and mutants upon urea-induced denaturation measured using a Superdex 200 gel-®ltration column. The Kav value is the partition coef®cient relative to the void and bed volumes of the column, calculated as described in Materials and Methods. The protein concentration was 25 mM in 15 mM sodium phosphate (pH 7.5), 100 mM NaCl and 1 mM DTT at 25 C. The wild-type data are included for comparison. The data for wild-type (*), D74N (‡) and D84N (!) were ®tted to a two-state transition; the data for R24P (*), P81L (^), H98P (^), P114L (~) and V126D () are shown for clarity with a smoothing function.

P81L is a somatic mutation of the residue initiating the ®rst helix of the third repeat. Its side-chain interacts with residues in the short ®rst helix of the third repeat. It was identi®ed in a melanoma-prone pedigree and had a greatly reduced af®nity for CDK4 and CDK6 (Reymond & Brent, 1995). It has a similar helical content to the wild-type protein (Figure 6(b)), in agreement with the study by Zhang & Peng (1996). However, the Stoke's radius is greater than that of wild-type (Figure 9). The CD-monitored denaturation curve could be ®tted to a two-state equation (Figure 8(b)), with a midpoint of denaturation of 3.4(0.2) M which is greater than the corresponding value for wild-type (1.9 M), and an m-value of 1.1(0.3) kcal molÿ1

1879

Stability and Folding of p16

Mÿ1, which is smaller than that of wild-type (2.0 kcal molÿ1 Mÿ1). It was not possible to ®t the denaturation curve obtained by SEC to a two-state equation because of the strong urea-dependence of the baselines, but a transition could be observed and the midpoint of denaturation, approximated by visual inspection, agrees with that measured by CD. D74N D74N is a de novo acquired somatic mutation located in the loop between the second and third repeats. It forms part of the interface between p16 and CDK4. It was isolated independently from tumours of the esophagus and bladder (Mori et al., 1994), and failed in G1 growth arrest and cyclin D1/CDK4 kinase inhibition (Koh et al., 1995). No aggregation was observed and the Stoke's radius was similar to that of the wild-type. The study by Zhang & Peng (1996) indicated that this mutant had disrupted secondary structure, but according to our measurements the secondary structure content was similar to that of wild-type (Figure 6(b)). The m-value and midpoint of denaturation were lower than the wild-type values (Figure 8(b)), and the free energy of unfolding in water was calculated to be 1.7(0.2) kcal molÿ1. Protein destabilisation is, therefore, likely to be the cause of inactivation by the mutation D74N. Sodium sulfate (350 mM) had a stabilising effect. The urea concentration of the midpoint of denaturation was increased, and the m-value was also greater and was similar to that of the wild-type protein. Sodium sulfate stabilised the mutant protein by 2.6 kcal molÿ1, compared with the stabilisation of 1.8 kcal molÿ1 for wild-type (Figure 10). D84N Mutation of this residue, that is located at the p16-CDK4 interface, was found in esophageal squamous cell carcinomas (Mori et al., 1994). It

Figure 10. Stabilization of D74N in the presence of 350 mM sodium sulphate. The data were ®tted to a two-state transition.

effectively inhibited cyclin-CDK4 kinase activity, but was intermediate in its ability to cause G1 growth arrest (Koh et al., 1995). Both SEC (Figure 9) and CD (Figures 6(b) and 8(b)) show that the mutant has similar biophysical properties to those of the wild-type, in agreement with the study by Tsai and co-workers (Tevelev et al., 1996), except that D84N appears to have an increased helical content. As suggested, D84 is a surface residue and is probably involved in CDK4 binding (Byeon et al., 1998).

Discussion We have shown that the folding of full-length p16 has thermodynamic and kinetic properties identical with those of the truncated form (lacking the ®rst eight amino acid residues). We have used the truncated form for more detailed characterisation and for determination of the effects of cancerassociated mutations, since this form expresses at a higher yield.

Equilibrium stability of p16 The coincidence of the denaturation curves obtained by different conformational probes, namely CD and ¯uorescence spectroscopies and size-exclusion chromatography, indicates that p16 unfolds at equilibrium in a two-state manner. The protein has a low thermodynamic stability (3.1 kcal molÿ1 in Tris buffer (pH 7.5), at 25  C) compared with other small proteins, which tend to have stabilities in the range 5-15 kcal molÿ1. The m-value, however, lies within the range observed for other proteins of a similar size (Myers et al., 1995). The intrinsic helical propensity of the amino acid sequence of p16 was investigated using the programme AGADIR (Munoz & Serrano, 1995a,b). The sequences corresponding to the helices of repeats 1, 2 and 4, were predicted to have high helical propensities of approximately 20 %, and those of repeat 3 had much lower predicted propensities of approximately 5 %. Thus, the amino acid sequence of p16 has a reasonable helical propensity. Interestingly, synthetic peptides corresponding to repeat 3 bind CDK4 and CDK6, and inhibit pRB phosphorylation by CDK4/6 in vitro (Fahrñus et al., 1996). The ankyrin repeats in the INK4 proteins and in 53BP2 stack in a linear fashion to give an elongated, non-globular structure. Most of the van de Waals' interactions are between helical segments of neighbouring repeats, and the lack of longer-range interactions could result in a relatively unstable molecule. Another potential instability source is the presence of long loops that connect the helix-turn-helix motifs. These interact to form an antiparallel b-sheet structure in p18 and p19. These regions are unstructured in p16, however, which would be costly in entropic terms.

1880 Characteristics of the major refolding intermediate The minimal scheme for the refolding of p16 involves two steps. An intermediate species is formed very rapidly, and transformation of the intermediate to the native state occurs in a subsequent, rate-determining step. The rate constant of this is reaction 33 sÿ1 in buffer at 25  C. The stability of the intermediate relative to the denatured state can be estimated from the ratio of the observed refolding rate constant and the calculated refolding rate constant according to a twostate model. It is calculated to be 1.2 kcal molÿ1. The slope of the folding limb of the plot of the rate constants versus [urea] (Figure 4(a)) in the region where the intermediate is populated gives mI ÿ {, the increase in solvent accessible surface area between the intermediate and the rate determining transition state. This is calculated to be 1.2 Mÿ1 (0.7 kcal molÿ1 Mÿ1) and by comparison with the equilibrium m-value, mD ÿ N, of 1.7 kcal molÿ1 Mÿ1, the intermediate is estimated to be 55 % as compact as the native state. The transformation of the intermediate to the native state was also followed by ellipticity at 222 nm. The results indicate that the intermediate has only a small amount of helical structure, if any. The transition state for unfolding of p16 is very compact The shallow slope of the unfolding arm of the plot of the observed rate constant versus [urea] (Figure 4(a); m{ ÿ N ˆ 0.33 Mÿ1 or 0.19 kcal molÿ1 Mÿ1 at 0 M urea) indicates that the rate-limiting transition state for refolding/unfolding is very compact (89 % compared with the native state). This is more compact than the rate-limiting folding/unfolding transition states of most other small proteins, which are between 60 % and 75 % as compact as the native state. There is one exception to-date: the cold-shock family of proteins are all-b proteins that fold in a two-state manner (Perl et al., 1998; Schindler et al., 1995). These proteins have transition states that are between 80 and 90 % as compact as their native states. Since b-sheet structure involves a higher proportion of non-local interactions relative to local interactions than does a-helical structure, it has been suggested that the transition state of unfolding of all-b proteins would be reached at an earlier point in the reaction than that of all-helical proteins, because breaking only a few non-local interactions would be required to destabilise the structure. A weak correlation was shown between relative contact order (the average sequence separation between contacting residues in the native state normalised by the total sequence length) and the position of the transition state of folding (Plaxco et al., 1998). p16 can be considered an allhelical protein, since the b-sheet region that links the helical structure in the other INK4 proteins is

Stability and Folding of p16

¯exible and unstructured in p16. Indeed, the relative contact order of the p18 structure has a value of 6 (K. Wong, University of Cambridge, personal communication), which is very low in comparison with the values of 9 for some all-helical proteins to 20 for some all-b proteins. Yet, the transition state of unfolding of p16 is very close to the native state. p16 has a very low energy barrier for unfolding Unfolding of most small proteins is very slow in the absence of denaturant (again, with the exception of CspB, which unfolds with a rate constant of 10 sÿ1 in water; Schindler et al., 1995), and a high energy barrier may be important in order to prevent proteins from undergoing proteolysis or aggregation. p16, however, unfolds with a rate constant of 0.8 sÿ1 in the absence of denaturant, which is relatively fast. The ¯exibility of the protein, conferred by thermodynamic and kinetic instability, may be necessary for its regulatory function. It will be interesting to see whether these are characteristics of all the INK4 proteins, and of other proteins containing multiple ankyrin repeats. The ubiquitous motif is found in proteins of many different functions, and conformational ¯exibility may be needed in order to mediate a diverse range of protein-protein interactions. Other proteins that consist of multiple copies of short (30-40 amino acid residue) repeats have been identi®ed (Bignell et al., 1997; Das et al., 1998; Huber et al., 1997; Peifer et al., 1994). They, like the ankyrin-repeat structures, may stack in a linear fashion that, combined with the low stability, could provide a very large and ¯exible surface for interaction with other molecules. This suggests a very different class of molecular recognition processes from that involving the tight matching of complementary surfaces. Of interest in this regard, a recent study of p21Waf1, a member of an unrelated family of CDK inhibitors, revealed that the protein lacks stable secondary or tertiary structure in solution, but becomes more ordered on binding to CDK2 (Kriwacki et al., 1996). Contribution of individual amino acids to p16 folding and stability The mechanisms of p16 inactivation by the single point mutations investigated in this study include extensive aggregation, loss of folding ability leading to formation of partly folded or misfolded species, destabilisation of the native state and change in surface residue charge. Inactivation by destabilisation of the native state was also observed for a number of cancer-associated point mutations in the core domain of p53 (Bullock et al., 1997). A 20 amino acid residue peptide fragment of p16 has been shown to mimic the function of the full-length protein (Fahrñus et al., 1996), suggesting that only a small section is required for molecular recognition. But the cancer-associated mutations are spread all over the protein, and

1881

Stability and Folding of p16

those outside the peptide sequence must in¯uence the structure of the protein in such a way as to effect the conformation of this sequence. Our results are consistent with this, since most of the mutations have a global rather than local affect on the structure. The quantitative effects of the mutations on the secondary and tertiary structure of p16 are summarised in Table 2. The Stoke's radii of the mutants H98P, P114L and V126D were greater than that of the wild-type, and their radii increased with increasing protein concentration, indicating that the proteins were oligomeric. That of the mutant R24P was greater by almost 30 %. The proportion of high-order aggregated species was greatest for these mutants. This may be a result of increased exposure of hydrophobic surface in these partly folded mutant proteins. The ¯uorescence spectra of the mutants and the lack of any change on addition of urea, indicate that they do not have folded structure. No non-aggregated species could be detected for the mutant Q50R, suggesting that this mutation had an even more drastic effect on the stability of the native state. The mutation D74N had a less dramatic effect on the structure of the protein than did the other mutants. The protein had a similar helical content and Stoke's radius to the wild-type. However, the m-value obtained upon urea-induced denaturation was considerably smaller, and the protein is destabilised relative to wild-type by 2 kcal molÿ1. The free energy of unfolding in water is 1.7 kcal molÿ1, which means that only 90 % of molecules are folded even in the absence of denaturant. D84 is the starting residue of the functional p16 peptide. The mutant D84N has similar biophysical properties to wild-type. D74N and D84N are inactivating mutations because the residues participate in the hydrogen-bond network at the p16-CDK4 complex interface. One feature of ankyrin-repeat proteins has been predicted to be a central hydrophobic a-helix, which interacts with other repeats (Bork, 1993), and the ability to form a stable tertiary structure requires multiple copies of the repeats. Consistent with this, the most drastic mutations in p16 are those that disrupt interactions between consecutive repeats. It will be interesting to see if other ankyrin-repeat proteins, in particular the more distantly related INK4 proteins p18 and p19, are as sensitive to point mutations as is p16. The method described here will provide a comprehensive approach for determining whether additional repeats can give the proteins extra stability and to observe the effects of mutations on their folding and structure. Relationship between p16 folding and cellular immortalization Many of the p16 mutations in tumour cells are missense mutations resulting in single amino acid substitutions, rather than large genomic deletions

or truncations as observed, for example, for pRb. p16 is extremely vulnerable to single amino acid changes: mutations result in loosely packed species that tend to aggregate, which may be a consequence of the relatively low thermodynamic and kinetic stability of the wild-type protein. This vulnerability provides a possible explanation for the striking frequency of cancer-associated missense mutations identi®ed in p16, and provides further support for the important role of p16 in cellular immortalization. Implications for drug therapy The ``magic bullet'' approach requires a target that is common to all tumour cells. Accumulating evidence suggests that the frequency of alterations in the INK4 locus (p16 and p19ARF) in human cancers may be second only to that of p53, irrespective of patient age, sites and tumour types (Hainaut et al., 1997; Hall & Peters, 1996; Sherr, 1996). Therefore, p16-functionally compromised tumour cells are likely potential targets of future therapeutic candidates.

Materials and Methods Materials High-purity urea was obtained from Rose Chemicals Ltd. Pure cyclophilin A was kindly provided by N. Foster and Dr S. E. Jackson, Cambridge, UK. All other reagents were from Sigma and BDH. p16 cloning and mutagenesis A cDNA for (truncated) p16 was synthesised from HeLa cell mRNA using a ®rst strand cDNA synthesis kit (Pharmacia). This was then ampli®ed by PCR and cloned into the EcoRI and NdeI sites of pET21-a vector (Novagen). Correct clones were identi®ed by sequencing (Department of Biochemistry, University of Cambridge, and Oswel, University of Southampton). Full-length p16 was made by insertion in two steps using the QuickChange kit (Stratagene). This was then sub-cloned into a vector containing the GST and cloning sites from pGEX2T cloned into the pRSET A vector from Invitrogen. The p16 mutants were made using the QuickChange kit. Correct clones were identi®ed by DNA sequencing. p16 expression and purification The expression and puri®cation protocol of the truncated p16 was modi®ed from Tevelev et al., (1996). A ®ve hour culture of BL21(DE3) containing the p16 expression vector in 2TY medium with 0.5 mg mlÿ1 ampicillin was transferred to fresh medium. The cells were grown at 37  C to A600 ˆ 0.6-1.0, and induced with 0.5 mM IPTG for four to ®ve hours. Cells were harvested by centrifugation at 5000 g for ten minutes. The cell pellet was resuspended in buffer 1 (50 mM Tris buffer (pH 8.0), containing 25 % (w/v) sucrose, 1 mM EDTA, 0.5 mM PMSF) and lysed by sonication. After centrifugation at 15,000 rpm for 20 minutes, the pellet was resuspended in buffer 1 and sonicated again. Centrifugation was repeated and the pellet was resuspended in 20 mM

1882 Tris buffer (pH 8.0), containing 1 % (v/v) Triton X-100, 1 mM EDTA, 0.5 mM PMSF, centrifuged again and the pellet resuspended in 8 M urea, in 50 mM Tris buffer (pH 8.0), 50 mM DTT, 5 mM EDTA. The sample was loaded onto a DEAE fast-¯ow Sepharose (Pharmacia) column equilibrated in buffer 2 (4 M urea, 20 mM Tris buffer (pH 8.0), 5 mM DTT, 1 mM EDTA). p16 was eluted using a gradient of 0 M to 0.5 M NaCl. Fractions containing p16 were pooled and concentrated using an Amicon concentrator with a YM10 membrane. The sample was loaded onto a preparative Superdex-75 column (Pharmacia) equilibrated in buffer 3 (buffer 2 with 140 mM NaCl). Fractions containing p16 were pooled and refolded by dialysis at 4  C against buffer 3 without the urea, and dialysed further against 10 mM Tris (pH 8.0), 1 mM DTT, 5 mM EDTA. The refolded protein was then loaded onto a preparative Superdex-200 column in order to check for high-order aggregates that may have formed upon refolding, and eluted with buffer containing 10 mM Tris (pH 8.0), 1 mM DTT, 5 mM EDTA. Fractions containing monomeric p16 were pooled, dialysed against 10 mM Tris (pH 8.0), 1 mM DTT, 5 mM EDTA, and concentrated to 50-100 mM using an Amicon Centriprep 3 concentrator. The protein was ¯ash-frozen and stored at ÿ70  C. Fresh transformants of BL21(DE3) containing the fulllength p16 expression vector were transferred to 2TY medium containing 0.5 mg mlÿ1 ampicillin. The cells were grown at 26  C to A600 ˆ 0.6-1.0, and induced with 0.1 mM IPTG for 16-18 hours. Cells were harvested by centrifugation at 5000 g for 10 minutes. The cell pellet was resuspended in 50 mM PBS buffer containing 1 mM DTT and lysed by sonication. After centrifugation at 15,000 rpm for 20 minutes, p16 was puri®ed from the supernatant in a batchwise manner using glutathione Sepharose 4B resin (Pharmacia). p16 was cleaved from the resin with thrombin and further puri®ed by a preparative Superdex-75 column equilibrated with 50 mM Tris (pH 7.5), 100 mM NaCl, 1 mM DTT, 5 mM EDTA. p16 eluted as a monomer and fractions were pooled and dialysed against 10 mM Tris pH 8.0, 1 mM DTT, 5 mM EDTA, ¯ash-frozen and stored at ÿ70  C. Both truncated and full-length forms of p16 was pure as judged by SDS-PAGE chromatography and electrospray mass spectrometry, and the proteins ran as monomer on an analytical S200 gel ®ltration column (Pharmacia). Protein concentration was measured spectrophotometrically using an extinction coef®cient of e280 ˆ 13,340, calculated by the method by Gill & von Hippel (1989). Mutant (truncated) proteins were puri®ed as for the wild-type. Detection of aggregation/oligomerization of refolded mutant proteins The extent of protein aggregation/oligomerization was monitored by size-exclusion chromatography (SEC). The refolded protein was loaded onto a preparative Superdex 200 or Superdex 75 column (Pharmacia) and eluted with a buffer containing 10 mM Tris (pH 8.0), 1 mM DTT, 5 mM EDTA. Equilibrium CD measurements All equilibrium experiments were performed in the presence (1 mM) of the reducing agents DTT or DTE to keep the single cysteine residue in a reduced state. Solutions of urea were prepared using a Hamilton Micro-

Stability and Folding of p16 Lab M, and the protein stock was then added using a Gilson pipette. The samples were equilibrated at the required temperature for one hour before measurement. CD spectra were recorded on a Jasco J720 spectropolarimeter using a cell of pathlength 0.1 cm. The protein concentration was 5 mM. Spectra were acquired at a scan speed of 50 nm per minute with a 1 nm slit and four second response time, and two scans were averaged. The cell was thermostatted using a waterbath at 25  C and the temperature of each sample monitored before measurement using an Edale Instrument thermometer. A baseline correction was made with buffer. The data were ®tted to equation (3), which assumes a two-state model in which the ¯uorescence of the native and denatured state is dependent on urea concentration (Clarke & Fersht, 1993): eˆ

…aN ‡ bN ‰UŠ ‡ …a ‡ bD ‰UŠ† expfmDÿN ‰UŠ ÿ ‰UŠ50% †=RTg 1 ‡ expfmDÿN …‰UŠ ÿ ‰UŠ50% †=RTg …3†

e is the ellipticity at a given concentration of urea, [U], aN and aD are the intercepts, and bN and bD are the slopes of the baselines at low (N) and high (D) urea concentrations. The [U]50 % value is the concentration of urea at which half of the protein is denatured, R is the gas constant, and T is the temperature in K; mD ÿ N is a constant that is proportional to the increase in solvent accessible surface area between native and denatured states (Pace, 1975). The data were ®tted to this equation by non-linear least-squares analysis using the curve-®t option of the software Kaleidagraph (Abelbeck Software), which gives the values of [U]50 %, and m and their standard errors. The free energy of unfolding in the presence of urea,
G[U] D ÿ N, is related to the concentration of denaturant (Pace, 1986): H2 O
G‰UŠ DÿN ˆ
GDÿN ÿ mDÿN ‰UŠ

…4†

Therefore, the free energy of unfolding in the absence of 2O urea,
GH DÿN , can be calculated as: H2 O
GDÿN ˆ mDÿN ‰UŠ50%

…5†

Fluorescence spectroscopy Denaturation was also monitored by ¯uorescence using an Aminco-Bowman Series 2 luminescence spectro¯uorimeter with excitation at 280 nm (bandpass 4 nm), and emission scanned between 300 and 400 nm (bandpass 4 nm) at a rate of 60 nm per minute. Protein (100 ml) was added to 800 ml of buffer containing urea, and the sample equilibrated for one hour at 25  C before measurement. The protein concentration was 3 mM. The temperature was monitored by a thermocouple in the cell. Size-exclusion chromatography Changes in the hydrodynamic volume of p16 in the presence of urea were monitored by gel ®ltration chromatography using an analytical Superdex-200 column (Pharmacia), equilibrated with 15 mM phosphate buffer (pH 7.5), containing 1 mM DTT, 100 mM NaCl and the corresponding concentration of urea. The salt was used to screen ionic interactions with the support. Samples of p16 at 25 mM were equilibrated in the same buffer as present in the column, and allowed to equilibrate for ten minutes before they were loaded onto the column. The

1883

Stability and Folding of p16 chromatography was run at a ¯ow rate of 0.7 ml per minute. The column was calibrated using a series of molecular mass marker proteins according to the column instruction manual. The plot of elution volume versus [urea] was ®tted to the two-state model using equations (3) to (5).

was measured at four urea concentrations between 0 M and 1.5 M. The protein concentration was 8 mM after mixing. Data from between two and ®ve scans were averaged.

Determination of Stoke's radii

Acknowledgements

The partition coef®cient, Kav, is calculated from the elution volume of the sample, Vx , using the expression:

where s is slope and Kav,0 the Y-intercept, respectively, of a plot of Kav versus RS obtained using the low molecular mass calibration kit from Pharmacia.

We thank Dr Mark Bycroft for helpful discussion and for assistance with the cloning of p16, Dr C. M. Johnson for assistance with the stopped-¯ow CD experiments, and Dr Y. W. Chen for assistance with the p16 structural model. This work was supported by the Medical Research Council of the United Kingdom. K.S.T. holds a Croucher Foundation Scholarship, a UK Overseas Research Student Award and a Prince Philip Graduate Exhibition Award (Friends of Cambridge University in Hong Kong). W.K.W. is funded by a studentship from the University Putra Malaysia and the Ministry of Science and Technology (Malaysia), and L.S.I. holds a Beit Memorial Fellowship in Medical Research (UK).

Stopped-flow fluorescence

References

Kinetic experiments were performed using an Applied Photophysics SX-17MV stopped-¯ow instrument at 25  C. An excitation wavelength of 280 nm was used, and emission was monitored at wavelengths above 320 nm using a cut-off ®lter. Unfolding experiments were performed by mixing protein in 50 mM Tris (pH 7.5) with one or ®ve volumes of urea containing the same buffer. pH-jump refolding was performed as follows: the protein was denatured in 30 mM HCl. Refolding was initiated by rapid mixing with an equal volume of renaturing buffer (100 mM Tris (pH 8.1)) to give ®nal buffer conditions of 50 mM Tris (pH 7.5). Refolding by urea dilution was also performed as follows. The protein was denatured in 3.5 M urea, 50 mM Tris (pH 7.5). Refolding was initiated by rapid mixing with ®ve or ten volumes of renaturing buffer (50 mM Tris (pH 7.5)). The protein concentration after mixing was 2 mM for all experiments. DTT (1 mM) in the refolding buffer had no effect on the kinetics. The kinetics of refolding was also measured in the presence of cyclophilin A. Cyclophilin A at a concentration of 2 mM was added to the refolding buffer (without urea). Refolding was initiated by mixing pHdenatured p16 with this buffer to give ®nal concentrations of p16 and cyclophilin A of 1 mM. Data collected from at least three scans were averaged and ®tted using Kaleidagraph. Unfolding traces were ®tted to a single exponential function. Refolding traces were ®tted to a triple exponential function. A term was included to account for baseline instability.

Alcorta, D. A., Xiong, Y., Phelps, D., Hannon, G., Beach, D. & Barrett, J. C. (1996). Involvement of the cyclindependent kinase inhibitor p16 (ink4a) in replicative senescence of normal human ®broblasts. Proc. Natl Acad. Sci. USA, 93(24), 13742-13747. Bignell, G., Micklem, G., Stratton, M. R., Ashworth, A. & Wooster, R. (1997). The BRC repeats are conserved in mammalian BRCA2 proteins. Hum. Mol. Genet. 6, 53-58. Boice, J. A. & Fairman, R. (1996). Structural characterization of the tumor-suppressor p16, an ankyrin-like repeat protein. Protein Sci. 5, 1776-1784. Bork, P. (1993). Hundreds of ankyrin-like repeats in functionally diverse proteins: mobile modules that cross phyla horizontally? Proteins: Struct. Funct. Genet. 17, 363-374. Brandts, J. F., Halvorson, H. R. & Brennan, M. (1975). Consideration of the possibility that the slow step in protein denaturation reactions in due to cis-trans isomerism of proline residues. Biochemistry, 14, 4953-4963. Brotherton, D. H., Dhanaraj, V., Wick, S., Brizuela, L., Domaille, P. J., Volyanik, E., Xu, X., Parisini, E., Smith, B. O., Archer, S. J., Serrano, M., Brenner, S. L., Blundell, T. L. & Laue, E. D. (1998). Crystal structure of the complex of the cyclin D dependent kinase Cdk6 bound to the cell-cycle inhibitor p19INK4d. Nature, 395, 244-250. Bullock, A. N., Henckel, J., DeDecker, B. S., Johnson, C. M., Niklova, P. V., Proctor, M. R., Lane, D. P. & Fersht, A. R. (1997). Thermodynamic stability of wild-type and mutant p53 core domain. Proc. Natl Acad. Sci. USA, 94, 14338-14342. Byeon, I.-J. L., Li, J., Ericson, K., Selby, T. L., Tevelev, A., Kim, H.-J., O'Maille, P. & Tsai, M.-D. (1998). Tumor suppressor p16INK4A: determination of solution structure and analyses of its interaction with cyclin-dependent kinase 4. Mol. Cell, 1, 421-431. Chan, F., Zhang, J. K., Cheng, L., Shapiro, D. N. & Winoto, A. (1995). Identi®cation of human and mouse p19, a novel cdk4 and cdk6 inhibitor with homology to p16INK4. Mol. Cell. Biol. 15(5), 26822688.

Kav ˆ …Vx ÿ Vo †=…Vt ÿ Vo †

…6†

where Vo and Vt are the void volume and the total column volume, respectively. The Stoke's radius of the protein, RS, can be calculated from Kav using the following expression: Kav ˆ s: log RS ‡ Kav;0

…7†

Stopped-flow CD Measurements were made at 25  C on an Applied Photophysics SX-17MV stopped-¯ow apparatus ®tted with a circular dichroism detector. Ellipticity was monitored at 222 nm, and the cell pathlength was 0.2 cm. A baseline correction was made with buffer. Unfolding and refolding experiments were performed as described for the stopped-¯ow ¯uorescence. Unfolding was measured at four different urea concentrations between 2.4 M and 8 M, and refolding by pH-jump

1884 Clarke, J. & Fersht, A. R. (1993). Engineering disul®de bonds as probes of the folding pathway of barnaseincreasing the stability of proteins against the rate of denaturation. Biochemistry, 32, 43322-4329. Corbett, R. J. T. & Roche, R. S. (1984). Use of high-speed size-exclusion chromatography for the study of protein folding and stability. Biochemistry, 23, 18881894. Cordon-Cardo, C. (1995). Mutation of cell cycle regulators. Biological and clinical implications for human neoplasia. Am. J. Pathol. 147, 545-560. Das, A. K., Cohen, P. & Barford, D. (1998). The structure of the tetratricopeptide repeats of protein phosphatase 5: implications for TPR-mediated protein-protein interactions. EMBO J. 17, 1192-1199. Drakenburg, T., Dalqvist, K.-I. & Forsen, S. (1972). The barrier to internal rotation in amides. IV. N,NDimethylamides; substituent and solvent effects. J. Phys. Chem. 76, 2178-2183. Dyson, H. J., Rance, M., Houghten, R. A., Lerner, R. A. & Wright, P. E. (1988a). Folding of immunogenic peptide fragments of proteins in water solution. I. Sequence requirements for the formation of a reverse turn. J. Mol. Biol. 201, 161-200. Dyson, H. J., Rance, M., Houghten, R. A., Wright, P. E. & Lerner, R. A. (1988b). Folding of immunogenic peptide fragments of proteins in water solution. II. The nascent helix. J. Mol. Biol. 201, 201-217. Fahrñus, R., Paramio, J. M., Ball, K. L., Lain, S. & Lane, D. P. (1996). Inhibition of pRb phosphorylation and cell-cycle progression by a 20-residue peptide derived from p16CKDN2/INK4A. Curr. Biol. 6(1), 84-91. Gill, C. S. & von Hippel, P. H. (1989). Calculation of protein extinction coef®cients from amino acid sequence data. Anal. Biochem. 182, 319-326. Gorina, S. & Pavletich, N. P. (1996). Structure of the p53 tumor-suppressor bound to the ankyrin and SH3 domains of 53BP2. Science, 274, 1001-1005. Grathwohl, C. & WuÈthrich, K. (1976). The X-Pro peptide bond as an NMR probe for conformational studies of ¯exible linear peptides. Biopolymers, 15, 20252041. Grathwohl, C. & WuÈthrich, K. (1981). NMR studies of the rates of proline cis-trans isomerisation in oligopeptides. Biopolymers, 20, 2623-2633. Greenwald, I. & Rubin, G. M. (1992). Making a difference: the role of cell-cell interactions in estabishing separated identities for equivalent cells. Cell, 68, 271-281. Guan, K. L., Jenkins, C. W., Li, Y., Nichols, M. A., Wu, X. Y., Okeefe, C. L., Matera, A. G. & Xiong, Y. (1994). Growth suppression by p18, a p16INK4/mts1related and p15INK4b/mts2-related cdk6 inhibitor, correlates with wild-type pRb function. Genes Dev. 8, 2939-2952. Hainaut, P., Soussi, T., Shomer, B., Hollstein, M., Greenblatt, M., Hovig, E., Harris, C. C. & Montesano, R. (1997). Database of p53 gene somatic mutations in human tumors and cell lines: updated compilation and future prospects. Nucl. Acids Res. 25, 151-157. Hall, M. & Peters, G. (1996). Genetic alterations of cyclins, cyclin-dependent kinases, and cdk inhibitors in human cancer. Advan. Cancer Res. 68, 67-108. Hannon, G. J. & Beach, D. (1994). p15INK4b is a potential effector of TGF-b-induced cell- cycle arrest. Nature, 371, 257-261. Hara, E., Smith, R., Parry, D., Tahara, H., Steven, S. & Peters, G. (1996). Regulation of p16(CDKN2)

Stability and Folding of p16 expression and its implications for cell immortalization and senescence. Mol. Cell. Biol. 16(3), 859-867. Harland, M., Meloni, R., Gruis, N., Pinney, E., Brookes, S., Spurr, N. K., Frischauf, A.-M., Bataille, V., Peters, G., Cuzick, J., Selby, P., Bishop, D. T. & Bishop, J. N. (1997). Germline mutations of the CDKN2 gene in UK melanoma families. Hum. Mol. Genet. 6, 2061-2067. Helps, N. R., Barker, H. M., Elledge, S. J. & Cohen, P. T. (1995). Protein phosphatase 1 interacts with p53BP2, a protein which binds to the tumour suppressor p53. FEBS Letters, 377, 295-300. Hirai, H., Roussel, M. F., Kato, J. Y., Ashmun, R. A. & Sherr, C. J. (1995). Novel INK4 proteins, p19 and p18, are speci®c inhibitors of the cyclin D-dependent kinases CDK4 and CDK6. Mol. Cell. Biol. 15, 2672-2681. Holland, E. A., Beaton, S. C., Becker, T. M., Grulet, O. M. C., Peters, B. A., Rizos, H., Kefford, R. F., Mann, G. J. (1995). Analysis of the p16 gene, CDKN2, in 17 Australian melamoma kindreds. Oncogene, 11, 2289-2294. Huber, A. H., Nelson, W. J. & Weis, W. I. (1997). Threedimensional structure of the armadillo repeat region of beta-catenin. Cell, 90, 871-882. Hussussian, C. J., Struewing, J. P., Goldstein, A. M., Higgins, P., Ally, D. S., Sheahan, M. D., Clark, W. H., Tucker, M. A. & Dracopoli, N. C. (1994). Germline p16 mutations in familial melanoma. Nature Genet. 8(1), 15-21. Jackson, S. E. & Fersht, A. R. (1991a). Folding of Chymotrypsin Inhibitor 2. 1. Evidence for a two-state transition. Biochemistry, 30, 10428-10435. Jackson, S. E. & Fersht, A. R. (1991b). Folding of Chymotrypsin Inhibitor 2. 2. In¯uence of proline isomerization on the folding kinetics and thermodynamic characterization of the transition state of folding. Biochemistry, 30, 10436-10443. Kamb, A., Gruis, N. A., Weaverfeldhaus, J., Liu, Q. Y., Harshman, K., Tavtigian, S. V., Stockert, E., Day, R. S., Johnson, B. E. & Skolnick, M. H. (1994). A cell-cycle regulator potentially involved in genesis of many tumor types. Science, 264(5157), 436-440. Kato, S., Okamura, M., Shimamoto, N. & Utiyama, H. (1981). Spectral evidence for a rapidly formed structural intermediate in the refolding kinetics of hen egg-white lysozyme. Biochemistry, 20, 1080-1085. Khorasanizadeh, S., Peters, I. D. & Roder, H. (1996). Evidence for a three-state model of protein folding from kinetic analysis of ubiquitin variants with altered core residues. Nature Struct. Biol. 3, 193-205. Kiefhaber, T. (1995). Kinetic traps in lysozyme folding. Proc. Natl Acad. Sci. USA, 92, 9029-9033. Kiefhaber, T. & Schmid, F. X. (1992a). Kinetic coupling between protein folding and prolyl isomerization. I. Folding of ribonuclease A and ribonuclease T1. J. Mol. Biol. 224, 231-240. Koh, J., Enders, G. H., Dynlacht, B. D. & Harlow, E. (1995). Tumor-derived p16 alleles encoding proteins defective in cell- cycle inhibition. Nature, 375, 506510. Kragelund, B. B., Robinson, C. V., Knudsen, J., Dobson, C. M. & Poulsen, F. M. (1995). Folding of a 4-helix bundle - studies of acyl-coenzyme-A binding-protein. Biochemistry, 34, 7217-7224. Kriwacki, R. W., Hengst, L., Tennant, L., Reed, S. I. & Wright, P. E. (1996). Structural studies of p21Waf1/ Cip1/Sdi1 in the free and CDK2-bound state: con-

Stability and Folding of p16 formational disorder mediates binding diversity. Proc. Natl Acad. Sci. USA, 93, 11504-11509. Lang, K., Schmid, F. K. & Fischer, G. (1987). Catalysis of protein folding by prolyl isomerase. Nature, 329, 268-270. Lin, L. N., Hasumi, H. & Brandts, F. J. (1988). Catalysis of proline isomerization during protein folding reactions. Biochim. Biophys. Acta, 956, 256-266. Loughran, O., Malliri, A., Owens, D., Gallimore, P. H., Stanley, M. A., Ozanne, B., Frame, M. C. & Parkinson, E. K. (1996). Association of CDKN2A/ p16(INK4A) with human head and neck keratinocyte replicative senescence - relationship of dysfunction to immortality and neoplasia. Oncogene, 13, 561-568. Luh, F. Y., Archer, S. J., Domaille, P. J., Smith, B. O., Owen, D., Brotherton, D. H., Raine, A. R. C., Xu, X., Brizuela, L., Brenner, S. L. & Laue, E. D. (1997). Structure of the cyclin-dependent kinase inhibitor p19INK4d. Nature, 389, 999-1003. Lux, S. E., John, K. M. & Bennett, V. (1990). Analysis of cDNA for human erythrocyte ankyrin indicates a repeated structure with homology to tissue-differentiation and cell-cycle control proteins. Nature, 344, 36-42. Matouschek, A. & Fersht, A. R. (1993). Application of physical organic chemistry to engineered mutants of proteins: Hammond postulate behavior in the transition state of protein folding. Proc. Natl Acad. Sci. USA, 90, 7814-7818. Matouschek, A., Kellis, J. T., Jr, Serrano, L., Bycroft, M. & Fersht, A. R. (1990). Transient folding intermediates characterized by protein engineering. Nature, 346, 440-445. Matouschek, A., Matthews, J. M., Johnson, C. M. & Fersht, A. R. (1994). Extrapolation to water of kinetic and equilibrium data for the unfolding of barnase in urea solution. Protein Eng. 7, 1089-1095. Mori, T., Miura, K., Auki, T., Nishihira, T., Mori, S. & Nakamura, Y. (1994). Frequent somatic mutation of the MTS1/CDK4I (multiple tumour suppressor cyclin dependent kinase 4 inhibitor) gene in esophagael squamous-cell carcinoma. Cancer Res. 54, 33963397. Munoz, V. & Serrano, L. (1995a). Elucidating the folding problem of helical peptides using empirical parameters . 2. Helix macrodipole effects and rational modi®cation of the helical content of natural peptides. J. Mol. Biol. 245, 275-296. Munoz, V. & Serrano, L. (1995b). Elucidating the folding problem of helical peptides using empirical parameters. 3. Temperature and ph-dependence. J. Mol. Biol. 245, 297-308. Myers, J. K., Pace, C. N. & Scholtz, J. M. (1995). Denaturant m-values and heat capacity changes: relation to changes in accessible surface areas of protein unfolding. Protein Sci. 4, 2138-2148. Nall, B. T. (1994). Proline isomerization as a rate-limiting step. In Mechanisms of Protein Folding (Pain, R. H., ed.), pp. 80-103, Oxford University Press Inc., New York. Nall, B. T., Garel, J.-R. & Baldwin, R. L. (1978). Test of the extended two-state model for kinetic intermediates observed in the folding transition of ribonuclease A. J. Mol. Biol. 118, 317-330. Pace, C. N. (1975). The stability of globular proteins. CRC Crit. Rev. Biochem. 3, 1-43.

1885 Pace, C. N. (1986). Determination and analysis of urea and guanidium denaturation curves. Methods Enzymol. 131, 266-279. Parry, D. & Peters, G. (1996). Temperature-sensitive mutants of p16CDKN2 associated with familial melanoma. Mol. Cell. Biol. 16, 3844-3852. Parry, D., Bates, S., Mann, D. J. & Peters, G. (1995). Lack of cyclin d-cdk complexes in rb-negative cells correlates with high-levels of p16(ink4/mts1) tumor-suppressor gene-product. EMBO J. 14, 503-511. Peifer, M., Berg, S. & Reynolds, A. B. (1994). A repeating amino-acid motif shared by proteins with diverse cellular roles. Cell, 76, 789-791. Perl, D., Welker, C., Schindler, T., Schroder, K., Marahiel, M. A., Jaenicke, R. & Schmid, F. X. (1998). Conservation of rapid two-state folding in mesophilic, thermophilic and hyperthermophilic cold shock proteins. Nature Struct. Biol. 5, 229-235. Plaxco, K. W., Simons, K. T. & Baker, D. (1998). Contact order, transition state placement and the refolding rates of single domain proteins. J. Mol. Biol. 277, 985-994. Pollock, P. M., Pearson, J. V. & Hayward, N. K. (1996). Compilation of somatic mutations of the CDKN2 gene in human cancers: non-random distribution of base substitutions. Genes Chromosom. Cancer, 15, 7788. Ragione, F. D., Borrielo, A., Giordani, L. & Iolascon, A. (1997). Cell division cycle alterations in human malignancies. Cancer J. 10, 151-157. Ranade, K., Hussussian, C. J., Sikorski, R. S., Varmus, H. E., Goldstein, A. M., Tucker, M. A., Serrano, M., Hannon, G. J., Beach, D. & Dracopoli, N. C. (1995). Mutations associated with familial melanoma impair p16INK4 function. Nature Genet. 10, 114-116. Reymond, A. & Brent, R. (1995). p16 proteins from melanoma-prone families are de®cient in binding to CDK4. Oncogene, 11, 1173-1178. Ridge, J. A., Baldwin, R. L. & Labhardt, A. M. (1981). Nature of the fast and slow refolding reactions of iron(III) cytochrome-c. Biochemistry, 20, 1622-1630. Rousseau, F., Schymkowitz, J. W. H., SaÂnchez del Pino, M. & Itzhaki, L. S. (1998). Stability and folding of the cell cycle regulatory protein, p13suc1. J. Mol. Biol. 284, 503-519. Russo, A. A., Tong, L., Lee, J. O., Jeffrey, P. D. & Pavletich, N. P. (1998). Structural basis for inhibition of the cyclin-dependent kinase Cdk6 by the tumour suppressor p16INK4a. Nature, 395, 237-243. Sanz, J. M. & Fersht, A. R. (1993). Rationally designing the accumulation of a folding intermediate of barnase by protein engineering. Biochemistry, 32, 1358413592. Schindler, T., Herrler, M., Marahiel, M. A. & Schmid, F. X. (1995). Extremely rapid folding in the absence of intermediates. Nature Struct. Biol. 2, 663-673. Schmid, F. X. (1983). Mechanism of folding of ribonuclease A - slow refolding is a sequential reaction via structural intermediates. Biochemistry, 22, 4690-4696. Schmid, F. X. (1986a). Fast-folding and slow-folding forms of unfolded proteins. Methods Enzymol. 131, 70-82. Schmid, F. X. (1986b). Proline isomerization during refolding of ribonuclease-A is accelerated by the presence of folding intermediates. FEBS Letters, 31, 217-220. Schmid, F. X. (1992). Kinetics of unfolding and refolding of single-domain proteins. In Protein Folding

1886

Stability and Folding of p16

(Creighton, T. E., ed.), pp. 197-241, Freeman, New York. Schmid, F. X. & Baldwin, R. L. (1978). Acid catalysis of the formation of the slow-refolding species of RNaseA: evidence that the reaction is proline isomerisation. Proc. Natl. Acad. Sci. USA, 75, 4764-4768. Serrano, M., Hannon, G. J. & Beach, D. (1993). A new regulatory motif in cell cycle control causing speci®c inhibition of cyclin D-CDK4. Nature, 366, 704-707. Serrano, M., Lee, C.-W., Chin, L., Cordon-Cardo, C., Beach, D. & DePinho, R. A. (1996). Role of the INK4A locus in tumour suppression and cell mortality. Cell, 85, 27-37. Shapiro, G. I., Prk, J. E., Edwards, C. D., Mao, L., Merlo, A., Sidransky, D. E., Ewen, E. & Rollins, B. J. (1995). Mulatiple mechanisms of p16INK4A inactivation in non-small cell lung cancer cell lines. Cancer Res. 55, 6200-6209. Sherr, C. J. (1996). Cancer cell cycles. Science, 274, 16721677. Sherr, C. J. & Roberts, J. M. (1995). Inhibitors of mammalian G(1) cyclin dependent kinases. Genes Dev. 9, 1149-1163. Smith-Sorensen, B. & Hovig, E. (1996). CDK2A (p16INK4A) somatic and germline mutations. Hum. Mutat. 7, 294-303. Tan, Y.-J., Oliveberg, M., Otzen, D. E. & Fersht, A. R. (1997). The rate of isomerisation of peptidyl-proline bonds as a probe for interactions in the physiological denatured state of chymotrypsin inhibitor 2. J. Mol. Biol. 269, 611-622. Tevelev, A., Byeon, I., Selby, T., Ericson, K., Kim, H. J., Kraynov, V. & Tsai, M. D. (1996). Tumor-suppres-

sor p16INK4A - structural characterization of wildtype and mutant proteins by NMR and circulardichroism. Biochemistry, 35, 9475-9487. Uversky, V. N. (1993). Use of fast protein size exclusion liquid chromatography to study the unfolding of proteins which denature through the molten globule. Biochemistry, 32, 13288-13298. Venkataramani, R., Swaminathan, K. & Marmorstein, R. (1998). Crystal structure of the CDK4/6 inhibitory protein p18INK4c provides insights into ankyrin-like repeat structure/function and tumor-derived p16INK4 mutations. Nature Struct. Biol. 5, 74-81. Walkenhorst, W. F., Green, S. M. & Roder, H. (1997). Kinetic evidence for folding and unfolding intermediates in staphylococcal nuclease. Biochemistry, 36, 5795-5805. Walker, G. J., Hussussian, C. J., Flores, J. F., Glendening, J. M., Haluska, F. G., Dracopoli, N. C., Hayward, N. K. & Fountain, J. W. (1995). Mutations of the CDKN2/p16INK4 gene in Australian melanoma kindreds. Hum. Mol. Genet. 4, 1845-1852. Weinberg, R. A. (1995). The retinoblastoma protein and cell cycle control. Cell, 81, 323-330. Yang, R., Gombart, A. F., Serrano, M. & Koef¯er, H. P. (1995). Mutational effects on the p16INK4A tumor suppressor protein. Cancer Res. 55, 2503-2506. Zhang, B. & Peng, Z. Y. (1996). Defective folding of mutant p16INK4 proteins encoded by tumor-derived alleles. J. Biol. Chem. 271, 28734-28737. Zindy, F., Soares, H., Herzog, K. H., Morgan, J., Sherr, C. J. & Roussel, M. F. (1997). Expression of INK4 inhibitors of cyclin D-dependent kinases during mouse brain development. Cell Growth Diff. 8, 11391150.

Edited by J. Karn (Received 18 August 1998; received in revised form 19 November 1998; accepted 19 November 1998)