The sizes determined for rrPar-4DLZ 44.5 Da difference between expected and observed and rrPar-4SAC 6.6 Da difference between expected and observed after accounting for15N labelling of t
Trang 1apoptosis response factor 4
David S Libich1,*, Martin Schwalbe1,*, Sachin Kate1, Hariprasad Venugopal1, Jolyon K Claridge1, Patrick J B Edwards1, Kaushik Dutta2 and Steven M Pascal1
1 Centre for Structural Biology, Institute of Fundamental Sciences and Department of Physics, Massey University, Palmerston North, New Zealand
2 New York Structural Biology Centre, NY, USA
Introduction
Prostate apoptosis response factor-4 (Par-4) is an
ubi-quitously expressed and evolutionary conserved protein
that was initially identified as a pro-apoptotic factor in
rat AT-3 androgen-independent prostate cancer cells
exposed to ionomycin [1,2] The identified
pro-apopto-tic and tumour-suppressive roles of Par-4 are
consid-ered to be its most important cellular functions and,
accordingly, Par-4 is downregulated in various cancers
[3] The anti-cancer strategy employed by Par-4 is
achieved by direct activation of the cell-death machinery
(e.g Fas⁄ FasL) [4] and inhibition of pro-survival fac-tors (e.g nuclear factor-kappa B) [5] Furthermore, ectopic over-expression of Par-4 can either directly induce apoptosis or sensitize cancer cells to apoptotic stimuli, dependent on cell type [6] Primarily a cyto-plasmic protein, translocation of Par-4 to the nucleus
is linked with the direct induction of apoptosis in cancer cells [3,7,8] Initially characterized in prostate cancer, Par-4 has also been demonstrated to function
in renal cell carcinomas [9], leukaemia [10] and
Keywords
circular dichroism; coiled-coil; intrinsically
disordered protein; prostate apoptosis
response factor 4; solution NMR
spectroscopy
Correspondence
D S Libich or S M Pascal, Institute of
Fundamental Sciences, Massey University,
Turitea Site, Private Bag 11222, Palmerston
North 4442, New Zealand
Fax: +64 6 350 5682
Tel: +64 6 356 9099
E-mails: d.s.libich@massey.ac.nz;
s.pascal@massey.ac.nz
*These authors contributed equally to this
work
(Received 24 March 2009, accepted 6 May
2009)
doi:10.1111/j.1742-4658.2009.07087.x
Prostate apoptosis response factor-4 (Par-4) is an ubiquitously expressed pro-apoptotic and tumour suppressive protein that can both activate cell-death mechanisms and inhibit pro-survival factors Par-4 contains a highly conserved coiled-coil region that serves as the primary recognition domain for a large number of binding partners Par-4 is also tightly regulated by the aforementioned binding partners and by post-translational modifica-tions Biophysical data obtained in the present study indicate that Par-4 primarily comprises an intrinsically disordered protein Bioinformatic analysis of the highly conserved Par-4 reveals low sequence complexity and enrichment in polar and charged amino acids The high proteolytic suscep-tibility and an increased hydrodynamic radius are consistent with a largely extended structure in solution Spectroscopic measurements using CD and NMR also reveal characteristic features of intrinsic disorder Under physio-logical conditions, the data obtained show that Par-4 self-associates via the C-terminal domain, forming a coiled-coil Interruption of self-association
by urea also resulted in loss of secondary structure These results are consistent with the stabilization of the coiled-coil motif through an intra-molecular association
Abbreviations
CREB, cAMP-responsive element-binding protein; DLS, dynamic light scattering; GST, glutathione S-transferase; HSQC, heteronuclear single quantum coherence; IDP, intrinsically disordered protein; IPTG, isopropyl thio-b- D -galactoside; LZ, leucine zipper; NLS, nuclear localization sequence; Par-4, prostate apoptosis response factor 4; PK, protein kinase; SAC, selective apoptosis of cancer cells.
Trang 2neuroblastomas [11], as well as endometrial [12],
pancreatic [13] and gastric [8] cancers
In addition to its role in cancer, Par-4 is thought to
assist in normal neuronal development by preventing
the hyper-proliferation of nerve tissues, in turn
con-trolling the number of neurones and glial cells in both
the peripheral and central nervous systems [14,15]
Par-4 is upregulated in several neurodegenerative
dis-eases, such as Alzheimer’s disease [16,17], Parkinson’s
disease [18], Huntington’s disease [19] and amyotrophic
lateral sclerosis [20] Par-4 is also reportedly involved
in immune response modulation [21], synaptic function
modulation [22] and apoptosis of neurones that have
received a traumatic insult [23]
The C-terminal quarter of Par-4 (Fig 1) is highly
conserved and shares some homology with the death
domains of other apoptotic proteins, such as Fas,
receptor-interacting protein, Fas-associated death
domain protein and tumour necrosis factor
receptor-associated death domain protein [24,25] This region
functions as the primary recognition and binding site
for various partners of Par-4, including Wilms’ tumour
1 [7], Akt1⁄ protein kinase (PK) B [26], atypical PKCs
(PKCs f and k⁄ i) [24], p62 [27], death-associated
pro-tein-like⁄ zipper interacting kinase [28], THAP [29],
Amida [30], E2F1 [31], D2 dopamine receptor [32],
b-site amyloid precursor protein cleaving enzyme 1
[17], apoptosis-antagonizing transcription factor [33]
and topoisomerase 1 [34] In addition, several binding
partners have been shown to interact at various sites
N-terminal to the aforementioned C-terminal segment,
including the androgen receptor [35], F-actin [36],
14-3-3 [26] and the SPRY domain-containing suppressor
of cytokine signalling box proteins 1, 2 and 4 [37]
Par-4 contains several conserved phosphorylation
sites that are modified by kinases, such as PKA, PKC,
casein kinase II and Akt1, adding a further level of
regulation of the function of Par-4 [38]
Phosphoryla-tion of an absolutely conserved threonine (rat T155,
human T163 or mouse T156; Fig 1) by PKA is
required for nuclear translocation [8] Phosphorylation
of a C-terminal serine residue (rat S249, human or
mouse S231; Fig 1) by Akt1 effectively inactivates
Par-4 by allowing the chaperone protein 14-3-3 to bind
and sequester it in the cytoplasm, even if it is
phosphorylated on T155 [26]
These multiple interactions coupled with a high
degree of sequence conservation and post-translational
modification suggest that the in vivo role(s) of Par-4
are highly temporally and spatially regulated
Simi-larly, the ubiquitous expression, post-translational
modifications and a plethora of binding partners are
characteristics common to many intrinsically
disor-dered proteins (IDPs) [39] In the present study, we demonstrate that residual structure exists in Par-4 because the measured hydrodynamic radius increased under denaturing conditions, suggesting that the ensemble becomes less compact CD and NMR indi-cate that Par-4 is primarily intrinsically disordered under physiological conditions and exists as an ensem-ble of fast-averaging (on the NMR time-scale) struc-tures Furthermore, Par-4 forms a stable coiled-coil through a self-association event mediated by the C-ter-minus The coiled-coil was probed using increasing concentrations of chaotropic agents and was found to
be very stable Using NMR, the segment of Par-4 not involved in the coiled-coil was shown to have spectral features that were similar to those of a C-terminal deletion mutant This is important because it suggests that Par-4 is able to bind more than one partner at a time and thus could function as a hub linking the functions of several proteins The coiled-coil region of Par-4 represents an important functional domain that
is an example of a gain of structure upon binding tran-sition, which is another important feature of IDPs [40]
Results
All sequence numbering is made with reference to rat Par-4, to reflect the recombinant rat (rrPar-4) constructs used in these studies Three constructs were created; 4FL (Par-4 full-length, residues 1–332), 4DLZ (deleted leucine zipper, residues 1–290) and rrPar-4SAC [selective apoptosis of cancer cells (SAC) domain construct, residues 137–195] (Fig 2A) The sequence identity expressed relative to rat Par-4 of mouse and human is 92% and 76%, respectively, whereas African clawed frog and zebra fish share 52% and 47% sequence identity with rat, respectively (Fig 1)
The nuclear localization sequences (NLS) 1 (residues 20–25) and 2 (residues 137–153) are strictly conserved
in all known Par-4 sequences (Fig 1) The SAC domain, which includes NLS2, is the minimum frag-ment of Par-4 that is absolutely required for apoptosis [6] and is completely conserved amongst mammals (Fig 1) Furthermore, there is a high degree of sequence conservation in the C-terminal quarter of
Par-4, which contains primarily a coiled-coil-like sequence (residues 254–332; Figs 1 and 2A) In particular, a leu-cine zipper (residues 292–330), which is a subset of the coiled-coil domain, is almost conserved in all known Par-4 sequences, suggesting a common functionality (Figs 1 and 2A) Relatively few Par-4 genes have been sequenced It has been suggested that the general pat-tern of sequence conservation shown in Fig 1 is likely
to be conserved across other mammalian sequences [1]
Trang 3Based on disembl analysis [41], the majority (> 70%)
of Par-4 is predicted to be disordered The putative
regions of order in Par-4, as indicated by grey bars in a
disemblplot (Fig 2B), align with or occur within func-tionally important regions of Par-4 (Fig 2A), namely NLS1, NLS2, SAC and the coiled-coil Secondary
Fig 1 Sequence alignment of the prostate apoptosis response factor 4 (Par-4) A BLASTP ⁄ CLUSTALW [102,103] alignment of sequences of Par-4 from various species: rat (Rattus norvegicus), mouse (Mus musculus), human (Homo sapiens), African clawed frog (Xenopus laevis) and zebra fish (Danio rero) The amino acids are coloured: red (nonpolar side chains: G, A, V, L, I, M, P, F and W), blue (polar side chains: S,
T, N, Q, Y and C) and green (polar, charged side chains: K, R, H, D and E) Symbols: residues in that column are identical in all sequences (*); substitutions are conservative (:); and substitutions are semi-conservative (.) The high degree of sequence conservation of Par-4 suggests functional significance and thus resistance to evolutionary pressure With reference to the numbering of rat Par-4, several seg-ments are of notable interest: two nuclear localization sequences [NLS1 (20–25) and 2 (137–153)], which are completely conserved among all known Par-4s, and the SAC domain (137–195), which is defined by being the absolute minimum fragment required for apoptosis and includes NLS2 [6] The C-terminal domain (254–332) is a coiled-coil (CC) motif that encompasses a LZ (292–330) as a subset Two important phosphorylation sites, T155 and S249, are denoted by red arrows.
Trang 4structure prediction using gor4 [42] shows that the
regions with the highest helical propensity also occur in
the aforementioned regions and align with the disembl
predicted ordered regions (Fig 2C) The hydrophobic
cluster analysis [43] of Fig 2D indicates that the most
hydrophobic regions align with the putative ordered and
predicted helical regions
A plot of mean net charge against mean
hydropho-bicity determined from a protein’s primary structure
may be used to classify it as folded or intrinsically
dis-ordered Plot space is divided by an empirically
deter-mined line (R¼ 2:785 H 1:151) based on an analysis
by Uversky et al [44] The three constructs used in this study are plotted in Fig 3A along with several ‘classi-cally folded’ proteins Here, rrPar-4FL, rrPar-4DLZ and rrPar-4SAC clearly fall into disordered space gen-erally characterized by low mean hydrophobicity and high net charge The construct representing the SAC domain (rrPar-4SAC), with 14 positively charged and
13 negatively charged residues but few hydrophobic residues, lies further in the disordered region
Figure 3B describes the sequence complexity of rrPar-4FL by comparison with the percent difference between the amino acid usage of a set of known IDPs
Fig 2 (A) A block diagram of the three constructs of rrPar-4 used in the present study Marked on each construct are the primary regions
of functional importance, including the nuclear localization sequences [NLS1 (20–25) and 2 (137–153), coloured green], the region necessary for SAC (137–195), the coiled-coil C-terminal domain (CC, 254-332, coloured red) and the LZ (292–330, shown with hatching) The rrPar-4DLZ construct lacks residues 291–332, which is approximately one-half of the coiled-coil and the entire leucine zipper The rrPar-4SAC con-struct represents residues 137–195 of Par-4, including NLS2 All three concon-structs used in the present study have an N-terminal GGS tag, a remnant from the cleavage of the purification tag, which is omitted here for simplicity (B) DISEMBL predicts regions of order ⁄ disorder in pro-teins using neural networks trained on multiple definitions of disorder [41] The dashed line in (B) represents a threshold value separating order and disorder (C) Secondary structure (a-helix only shown) prediction using GOR 4 [42] and (D) hydrophobic cluster analysis (HCA) [43], a visually enhanced representation of the primary sequence that highlights clustering of hydrophobic residues using symbols ( , T; , S; ¤, G;
w , P) and colours (red: P and acidic residues D, E, N, Q; blue: basic residues, H, K, R; green: hydrophobic residues, V, L, I, F, W, M, Y; black: all other residues, G, S, T, C, A) The grey bars indicate the predicted regions of order in (B) and, for comparison, are extended over (C) and (D).
Trang 5versus a set of folded proteins (black bars) Positive
values indicate a depletion, whereas negative bars
indi-cate an enrichment relative to folded proteins The
pat-tern of amino acid usage for rrPar-4FL (grey bars) is
in accordance with that generally observed for IDPs
[45,46], namely a depletion of order-promoting amino
acids (L, N, F, Y, I, W, C) and enrichment of
dis-order-promoting residues (S, Q, K, P, E) The amino
acid usage for rrPar-4DLZ and rrPar-4SAC follows a similar pattern (not shown)
As calculated (i.e from sequence) and experimen-tally determined [i.e from MS, Tricine-PAGE and dynamic light scattering (DLS)], the molecular weights for rrPar-4FL, rrPar-4DLZ and rrPar-4SAC are given
in Table 1 Because DLS measures the Stokes radius (RS) of a particle, the equation log(RS) = 0.357· log(MW) ) 0.204 was used to convert RS to MW for comparative purposes [47,48] Although this approxi-mate calculation does not take into account the shape
of the particle (i.e it assumes a sphere), the result is useful for illustrating the degree of extended structure
in the protein
The primary structure predicts MWs of 36.1, 31.1 and 7.0 kDa for rrPar-4FL, rrPar-4DLZ and rrPar-4SAC, respectively MALDI-TOF mass spectroscopy was used
to assess the purity and determine the sizes of the con-structs produced The sizes determined for rrPar-4DLZ (44.5 Da difference between expected and observed) and rrPar-4SAC (6.6 Da difference between expected and observed after accounting for15N labelling of the sample used for MS analysis) agree within error (approximately 0.1%) with the sizes predicted from sequence analysis (Table 1) MS revealed that the rrPar-4FL construct is approximately 0.2 kDa larger than expected
Relative mobility analysis of the electrophoretic pro-files of rrPar-4FL, rrPar-4DLZ and rrPar-4SAC using
a denaturing Tricine-PAGE system (see Experimental procedures) determined apparent molecular weights of 49.1, 41.5 and 12.4 kDa, respectively These sizes are significantly larger (36%, 33% and 77% larger for rrPar-4FL, rrPar-4DLZ and rrPar-4SAC, respectively) than the expected MWs determined from the primary structure or MS (Table 1)
The results of DLS experiments are shown in Table 2 and summarized in Table 1 The measured RS for rrPar-4FL was 189 A˚, which is much larger than expected for a monomeric random coil, suggesting a polymeric state for rrPar-4FL under these conditions
Fig 3 (A) Charge ⁄ hydrophobicity plot of rrPar-4FL (335 residues),
rrPar-4DLZ (293 residues), and rrPar-4SAC (61 residues) The
divid-ing line R ¼ 2:785 H 1:151 represents an empirically determined
divisor between intrinsically disordered (high charge, low
hydropho-bicity) and structured (low charge, high hydrophohydropho-bicity) space.
Proteins such as aprotinin [104], actin [105], ubiquitin [106] and 3C
protease [107] are plotted as examples of classically folded
proteins (B) Sequence complexity of rrPar-4FL (grey bars)
com-pared with the average amino acid distribution of IDPs (black bars)
relative to the average amino acid distribution of globular proteins.
The relative distributions were sampled from proteins (both IDPs
and folded) deposited in the Protein Data Bank Positive and
nega-tive values indicate an enrichment or depletion, respecnega-tively, of a
particular residue relative to globular proteins Residues marked
with an asterisk occur two-fold more or less frequently, on average,
in IDPs than in globular proteins [46].
Table 1 Hydrodynamic properties of rrPar-4 constructs using vari-ous biophysical techniques MW (kDa) and hydrodynamic radius (A ˚ ) are shown in the format MW (RS) for three constructs using four techniques RSand MW were calculated from the primary structure
in reference to a folded conformation using log(R S ) = 0.357 · log(MW) ) 0.204.
Construct Sequence
Method of analysis
rrPar-4FL 36.1 (26.5) 36.2 (26.5) 49.5 (29.6) 8899 (189) rrPar-4-DLZ 31.1 (25.1) 31.2 (25.1) 41.5 (27.8) 64.1 (32.5) rrPar-4 SAC 7.0 (14.8) 7.1 (14.8) 12.5 (18.1) 18.7 (20.9)
Trang 6For comparison, the RS (calculated) for rrPar-4FL as
either monomeric globular (i.e folded), molten globule,
pre-molten globule, extended chain or urea-denatured
states are given in Table 2 The experimentally
deter-mined RS for rrPar-4DLZ (32.5 A˚) and rrPar-4SAC
(20.9 A˚) are larger than the expected folded RS (25.1
and 14.8 A˚, respectively) but still smaller than the
cal-culated random coil RS for either protein (Table 2)
This suggests that these constructs exist in an unfolded
yet monomeric form under these conditions The
volume weighted distributions for 4FL,
rrPar-4DLZ and rrPar-4SAC are shown in the Supporting
information (Fig S1A) The relatively broad
distribu-tion of sizes recorded for all three proteins is consistent
with an ensemble of interconverting conformations
rather than one single conformation
Upon addition of 1 m urea, the measured RS for
both rrPar-4DLZ and rrPar-4SAC slightly increases
(Table 2; see also Fig S1B,C) Conversely, the
intro-duction of 1 m urea to rrPar-4FL decreases the
mea-sured RS from 189 to 78.4 A˚, yet it remains larger
than the calculated RS of a random coil protein
(Table 2; see also Fig S1A)
A classically folded protein is less susceptible to
proteolysis than an IDP upon equilateral exposure to
a protease such as trypsin because most of its cleavage
sites are protected by tertiary folding [49,50] The
results of a limited trypsin digest of 4FL,
rrPar-4DLZ, rrPar-4SAC and BSA are shown in Fig 4
After 15 min of exposure to trypsin rrPar-4DLZ was
more than 95% digested, rrPar-4FL and rrPar-4SAC
were over 80% digested, whereas BSA was only 10%
digested BSA was chosen for comparison because it
has a similar percentage of predicted cut sites to that
of the Par-4 constructs
Figure 5 shows the full range (5C steps from
5–75C) and a sub-set of four spectra (5, 25, 45 and
65C) of a temperature series recorded by CD
spec-troscopy for rrPar-4FL (Fig 5A,B), rrPar-4DLZ
(Fig 5C,D) and rrPar-4SAC (Fig 5E,F) Significant
a-helical character in rrPar-4FL is immediately evident
and remains stable up to 65C (Fig 5A,B) By
con-trast, the CD spectra for rrPar-4DLZ (Fig 5C,D) and rrPar-4SAC (Fig 5E,F) show a typical profile of IDPs with a deep transition at 200 nm [51]
Pairwise overlays of 1H-15N heteronuclear single quantum coherence (HSQC) spectra for rrPar-4FL, rrPar-4DLZ and rrPar-4SAC are shown in Fig 6 The spectra of all three proteins display the features that characterize disorder in proteins, namely sharp peaks and narrow 1H chemical shift dispersion [51,52] Chemical shift similarities indicate some structural similarity between rrPar-4FL and rrPar-4DLZ (Fig 6A) Fewer peaks share similar chemical shifts when comparing 4FL or 4DLZ with rrPar-4SAC (Fig 6B,C) Thus, the majority of residues in rrPar-4SAC experience a different local environment and possibly a different conformation than the SAC domain in the context of either the rrPar-4FL or rrPar-4DLZ constructs Only 160 of the 308 peaks expected (335 – N-terminal residue – 26 prolyl resi-dues) for rrPar-4FL and 152 of the 266 expected peaks (293 – N-terminal residue – 26 prolyl residues) for rrPar-4DLZ are readily picked Conversely, 58 peaks
Table 2 Comparison of experimental and theoretical values of Stoke’s radii (RS) Measured RSwas recorded in 10 m M Tris (pH 7.0), 20 m M
NaCl in the presence or absence of urea Calculated R S was obtained using the mean values from equations given in Uversky [47] for globu-lar (folded) (G), molten globule (MG), pre-molten globule (PMG), random coil (RC) and urea-denatured (U) states.
Measured RS(A ˚ ) Calculated RS(A ˚ )
Fig 4 Limited proteolysis of rrPar-4FL (filled circles), rrPar-4DLZ (open circles), rrPar-4SAC (filled triangles) and BSA (open triangles) The proteins were dissolved in 20 m M NaPO4(pH 7.5), 50 m M NaCl and exposed to trypsin in a 280 : 1 (w ⁄ w) ratio.
Trang 7of the expected 60 (62 – N-terminal residue – one
pro-lyl residue) were readily identifiable for rrPar-4SAC
with only two glycyl residues being unobservable
To assess the degree of a-helicity in the rrPar-4FL
C-terminus, a CD difference spectrum between
rrPar-4DLZ and rrPar-4FL (25C) is shown in Fig 7A This
spectrum indicates a well-defined coiled-coil type
struc-ture (Fig 7A) The two constructs differ in the
dele-tion of the leucine zipper (Fig 2A); thus, rrPar-4FL
forms a stable coiled-coil under these conditions and
the majority of the a-helical character observed in
rrPar-4FL (Fig 5A,B) may be attributed to this
struc-ture The melting temperatures (based on the reduction
of the 222 nm transition in CD spectra) for rrPar-4FL
are 75, 55 and 25C when dissolved in native buffer, native buffer + 1 m urea or native buffer + 6 m urea, respectively (Fig 7B) The results of the DLS experi-ments on rrPar-4FL under the same conditions are shown in Fig 7C As the concentration of urea is increased from 1 to 6 m, the effective RSfor rrPar-4FL
is reduced from 189 to 58.5 A˚ The latter value is very close to the predicted RSof a random coil of the same molecular weight (for comparison, see Table 2)
Discussion
The structure-defines-function paradigm of molecular biology is currently under scrutiny because many
Fig 5 Temperature dependence of the CD spectrum of (A, B) rrPar-4FL, (C, D) rrPar-4DLZ and (E, F) rrPar-4SAC Data for all constructs were recorded in 10 m M Tris (pH 7.0), 20 m M NaCl over a temperature range of 5–75 C Traces for each temperature recorded in the exper-iment are shown in (A, C, E) For clarity, four equally spaced temperatures from the sampled range are shown in (B, D, F).
Trang 8proteins have been identified that are functional
with-out the need for well-defined secondary or tertiary
structure [46,53,54] The occurrence of intrinsic
disor-der in proteomes is correlated to the complexity of the
cell; thus, eukaryotic proteins have a higher proportion
of disorder (35–51% of proteins with disordered
regions of 40 residues or longer) than proteins from
prokaryotes and archaea (6–33%) [55]
The prevalence of intrinsic disorder is higher in
proteins that are involved in cell signalling, cytoskeletal
organization and ribosomal or cancer-related processes
[56] Disorder in proteins that control these processes
appears to be of functional importance because these
events are often tightly controlled and highly dynamic
and often become deregulated in cancerous cells [57]
Many signalling proteins function in pathways
associ-ated with cancer For example, the well-characterized
IDP p53 functions as a transcription regulator during
the G1 cell cycle phase Critical mutations of p53 lead
to loss of its transcriptional control and thus lead to
inappropriate survival of damaged or mutated cells [58]
Sequence analysis of Par-4
Bioinformatic analysis of rrPar-4FL reveal
characteris-tic features of IDPs, including high net charge, low
mean hydrophobicity and low sequence complexity
[45,59] Relative to the amino acid usage observed in
folded proteins, 4FL, 4DLZ and rrPar-4SAC are depleted in order-promoting amino acids and enriched in disorder promoting residues (Fig 3B) The lack of hydrophobic residues inhibits the forma-tion of a hydrophobic core and thus the formaforma-tion of stable tertiary structure (Fig 2D) [46]
More than 70% of rrPar-4FL is predicted to be dis-ordered by disembl, providing a strong argument against the formation of stable global tertiary structure (Fig 2B) Hydrophobic cluster analysis is a method of displaying the primary structure such that the cluster-ing of hydrophobic residues and thus regions of possi-ble order become evident [43] The regions of greatest hydrophobic clustering in rrPar-4FL correlate well with the predicted regions of order (Fig 2B) and with the secondary structure predictions (Fig 2C) Although the majority of rrPar-4FL is predicted to be disordered, this does not preclude the formation of short regions of structure or larger but transient sec-ondary structure elements Indeed, gor4 predictions of a-helical structure (Fig 2C) coincide with the more ordered regions of rrPar-4FL and fall within the highly conserved segments of the protein (Fig 1) Thus, regions of rrPar-4FL may be capable of forming a-helices either independently or upon association with binding partners Furthermore, the predicted regions
of order in rrPar-4LZ occur within the functionally relevant regions, namely NLS1 and 2, SAC and the
Fig 6 Pairwise overlays of 1H-15N HSQC spectra of (A) rrPar-4FL (black contours) and rrPar-4DLZ (blue contours), (B) rrPar-4DLZ (blue contours) and rrPar-4SAC (red contours) and (C) rrPar-4FL (black contours) and rrPar-4SAC (red contours) The compositions of the samples were: rrPar-4FL, 0.48 m M in 10 m M Tris (pH 7.0), 20 m M NaCl, 5% D2O, 15 N-rrPar-4DLZ, 0.09 m M in 20 m M NaPO4(pH 7.5), 100 m M NaCl,
1 m M dithiothreitol, 5% D 2 O and15N-rrPar-4SAC, 0.34 m M in 10 m M Tris (pH 7.0), 20 m M NaCl, 5% D 2 O All spectra were recorded at 5 C and the processing parameters (see Experimental procedures) were identical for qualitative comparison.
Trang 9coiled-coil (Fig 2) raising the possibility that Par-4 function may be associated with structure stabilization
in these regions
Intrinsic disorder in proteins is often erroneously considered to be a featureless random coil, although proteins do not achieve a completely random confor-mation even in strongly denaturing conditions [60] A more accurate depiction is that IDPs exist as ensembles
of rapidly interchanging conformers that sample vary-ing regions of secondary structure space [46] IDPs can
be broadly categorized into three non-exclusive groups (i.e a single IDP may fall into more than one category): random coil, pre-molten globule or molten globule [61]
Because of the high percentage of rrPar-4FL that is predicted as disordered, a random coil-like classifica-tion of the ensemble would appear to be the most appropriate Similar to the structural ensemble described for activator for thyroid hormone and reti-noid receptors [62], in the absence of interacting part-ners, rrPar-4FL exists predominantly unfolded in solution The kinase-inducible transcriptional-activa-tion domain of cAMP-responsive element-binding pro-tein (CREB) has been shown to be an IDP that folds into an orthogonal a-helix structure upon association with CREB binding protein [63,64] The intrinsically disordered nature along with the CREB binding protein-induced helical regions could be accurately pre-dicted from its primary structure [53] Similarly, the primarily intrinsically disordered nature and potential helical regions of Par-4 are predicted here (Figs 2 and 3)
Par-4 displays aberrant electrophoretic mobility and is susceptible to proteolysis
Aberrant electrophoretic mobility in a denaturing PAGE system is a hallmark of IDPs because their unique amino acid composition reduces the amount
of sodium dodecyl sulphate that is able to bind [46,51,65] Aberrant mobility on PAGE gels of Par-4 and Par-4 constructs (i.e deletion mutants) has been demonstrated, although it is not known whether the effects of IDP amino acid composition were con-sidered [34,36] In the present study, slower than expected migration of the Par-4 constructs resulted
in apparent MWs that were 1.3- (rrPar-4FL and rrPar-4DLZ) to 1.8- (rrPar-4SAC) fold larger than that predicted from sequence or measured using MS (Table 1)
Limited proteolysis can be used to distinguish ordered and disordered proteins based on their relative sensitivity to cleavage by proteases such as trypsin
Fig 7 (A) Difference between the 25 C traces of rrPar-4FL and
rrPar-4DLZ (Fig 5) The difference spectrum is characteristic of a
well-defined coiled-coil displaying a h222⁄ h 208 ratio > 1 (B)
Temper-ature dependence of the molar elipticity measured at 222 nm for
rrPar-4FL in buffer (10 m M Tris, pH 7.0, 20 m M NaCl) only (filled
circles), buffer + 1 M urea (open diamonds) and buffer + 6 M urea
(open triangles) (C) Volume distribution of DLS measurements of
rrPar-4FL showing the apparent hydrodynamic radius of the
parti-cles: buffer (10 m M Tris, pH 7.0, 20 m M NaCl) only (white bars),
buffer + 1 M urea (grey bars) and buffer + 6 M urea (hatched bars).
The reduction of the apparent RSupon increasing urea
concentra-tion suggests the disrupconcentra-tion of a polymeric complex.
Trang 10[45,50,66] Although rrPar-4FL, rrPar-4DLZ and BSA
contain an approximately equal percentage of trypsin
cut sites, BSA is digested at a much slower rate
(Fig 4) This implies that a significant portion of the
conformational ensemble of the rrPar-4 proteins are
more exposed to the solvent than BSA and largely lack
protection by folded and stable tertiary structure
The hydrodynamic radius of Par-4 is larger than
that predicted by sequence analysis
The observable Stokes radius of a protein increases in
proportion to its degree of ‘unfoldedness’; thus, an
IDP will have an observable RS larger than a folded
globular protein of the same MW [48,67] Some
exam-ples of IDPs with large RSvalues relative to MW have
been summarized previously [68] In the present study,
the RS measured for rrPar-4FL, rrPar-4DLZ and
rrPar-4SAC correspond to MWs of 8.9· 103, 64.1 and
18.7 kDa, respectively, and are much larger (713, 129
and 141%) than what would be expected for a folded
globular protein of similar MW The MW estimations
shown in Table 1 are used as a point of reference to
illustrate that the degree of ‘unfoldedness’ of Par-4 is
very high, which is similar to that expected for a
coil-like as opposed to a pre-molten globule ensemble [48]
The extremely large RSobserved for rrPar-4FL relative
to the other two constructs clearly indicates a
poly-meric state, as discussed further below
IDPs such as CREB and p27Kip1 (i.e
cyclin-depen-dent kinase inhibitor) exist as structurally
intercon-verting populations that have been demonstrated to
retain a nascent secondary structure to varying
degrees under physiological conditions [69] The width
of the volume-weighted distributions for rrPar-4FL,
rrPar4DLZ and rrPar-4SAC (see Fig S1) is consistent
with this type of conformational exchange
Interest-ingly, although the addition of 1 m urea causes a
sub-tle but significant increase in RS for rrPar4DLZ and
rrPar-4SAC (Table 2), the width of the distributions
are largely unaltered Together, these observations
suggest that 1 m urea can disrupt some folding
elements and bring the conformation of the ensemble
closer to random coil, but conformational exchange
continues
Secondary structure of Par-4 assessed by CD and
NMR
The CD spectra for rrPar-4DLZ and rrPar-4SAC are
exemplary of IDPs with a deep transition at 200 nm
and a minor transition at 222 nm (Fig 5) [51] The
CD spectra of IDPs are often complicated by minor
contributions from secondary structure elements, such
as alpha or poly-proline type II helices [70] Decon-volution of the 25C spectra estimates 32%, 17% and 18% of combined regular and distorted a-helix for rrPar-4FL, rrPar-4DLZ and rrPar-4SAC, respectively [71] All three constructs remain relatively stable throughout the heating cycle because the 5–65C traces exhibit similar features Thus, in addition to the coiled-coil region of rrPar-4FL, other regions of these proteins may transiently populate a-helical or other secondary structures Interestingly, although the over-all temperature-induced changes are minor, an isodich-roic point at 210 nm is observed for rrPar-4SAC, which may be interpreted as a two-state confor-mational change (Fig 5C) This could be the result of secondary and⁄ or tertiary structure that is thermally disrupted
The atomic resolution of NMR makes it uniquely suited to assess the ‘orderedness’ of an IDP [51] Because most residues of an IDP are solvent exposed and inherently flexible, they share a similar chemical environment and, consequently, share similar NMR frequencies, resulting in significant overlap of reso-nances (particularly for 1H resonances) and popula-tion-weighted average chemical shifts [72] Mobility also results in sharp peaks as a result of increased T2 values [73] The spectra of rrPar-4FL, rrPar-4DLZ and rrPar-4SAC shown in Fig 6 are characteristic of IDPs, with ensemble averages, narrow peaks and poor chemi-cal shift dispersion
A total of 52% of residues for both rrPar-4FL and rrPar-4DLZ are not readily observable in an 1H-15N HSQC From the current data, it is impossible to determine whether the same residues (or residues from the same regions) are unobservable in these two con-structs Possible reasons for this feature include poor chemical shift dispersion and intermediate exchange [74] A detailed examination of the dynamics of the visible regions of the proteins (dependent on assign-ments) may help to elucidate the time scales of motion involved and thus more definitive statements could then be made about particular residues or regions of rrPar-4LZ and rrPar-4DLZ [75]
The spectrum of rrPar-4SAC (Fig 6) is much more complete than those recorded for rrPar-4FL and rrPar-4DLZ Nonetheless, a similar degree of disorder
is suggested by the peak shape and chemical shift dis-persion The size of the rrPar-4SAC (7 kDa) relative
to that of the other constructs (> 30 kDa) is likely to
be a contributing factor in the observance of these res-onances because fewer residues equates to less chance
of spectral overlap and a lower likelihood of slow to intermediate exchange