- Email: [email protected]
The solution structure of apo-iron regulatory protein 1
Gene 524 (2013) 341–346
Contents lists available at SciVerse ScienceDirect
Gene journal homepage: www.elsevier.com/locate/gene
The solution structure of apo-iron regulatory protein 1 O'Neil Shand, Karl Volz ⁎ Department of Microbiology and Immunology, University of Illinois at Chicago, Chicago, IL 60612, USA
a r t i c l e
i n f o
Article history: Accepted 25 March 2013 Available online 13 April 2013 Keywords: Iron homeostasis Iron regulatory protein Bifunctionality RNA-binding Iron–sulfur cluster
a b s t r a c t Iron is a cofactor for many proteins that are involved in essential metabolic processes. However, iron must be strictly regulated because it can react with oxygen to generate cytotoxic reactive oxygen intermediates. Iron regulatory protein 1 (IRP1) is a bi-functional protein that can act either as a post-transcriptional regulator of mRNAs containing iron responsive elements, or as a [4Fe–4S] cluster-containing cytosolic aconitase. Previous X-ray crystallography results show that IRP1 is in an open L-shape conformation when bound to IRE-RNAs, and in a globular conformation when it binds an iron–sulfur cluster. The structure of apo-IRP1 and the mechanism by which it transforms to either functional state is unknown. Therefore, small angle X-ray scattering was used to determine the low resolution solution structure of apo-IRP1 and to characterize its biophysical properties. These results show that apo-IRP1 has a radius of gyration (Rg) of 33.6 ± 0.3 Å, and a Dmax of 118 ± 2 Å. The ab initio and rigid-body modeling results show that apo-IRP1 is in an open conformation in solution, and the ensemble optimization results show that the molecules stay narrowly distributed about a Rg of 33–34 Å. The open apo-IRP1 conformation seems optimal for subsequent conversion to either functional end state: RNA-binding, or cytosolic aconitase. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Iron is an essential element utilized by most organisms. However, iron levels must be strictly regulated in order to prevent deleterious effects of iron insufficiency or iron overload. In metazoans, iron regulatory protein 1 (IRP1) is a bi-functional protein that maintains iron homeostasis by integrating post-transcriptional regulation of iron metabolism genes with the cellular iron status (Anderson et al., 2012). Elucidations of the crystal structures of IRP1 in complex with iron responsive element (IRE) RNA (Walden et al., 2006) and in the form of cytosolic (c-) aconitase (Dupuy et al., 2006) were major steps forward in determining the structural principles responsible for IRP1 bi-functionality. Although much progress has been made in understanding the two functional roles of IRP1, little is known about the structure of the apo-protein and how it acquires either the IRE-RNA or the [4Fe–4S] cluster. Prior to this work, apo-IRP1 was characterized by several biochemical and biophysical methods. Proteolysis studies showed that apo-IRP1 was more susceptible to degradation than c-aconitase (Schalinske et al., 1997). Neutron scattering experiments were the first to determine the radii of gyration (Rgs) of the different states of IRP1, and demonstrate that apo-IRP1 has a Rg larger than the protein
Abbreviations: DLS, dynamic light scattering; DTT, dithiothreitol; IRE, iron responsive element; IRP, iron regulatory protein; Rg, radius of gyration; SAXS, small-angle X-ray scattering. ⁎ Corresponding author. Tel.: +1 3129962314; fax: +1 3123554535. E-mail addresses: [email protected] (O. Shand), [email protected] (K. Volz). 0378-1119/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.gene.2013.03.112
when complexed with IRE-RNA (Brazzolotto et al., 2002). Analytical ultracentrifugation results were consistent with the neutron scattering findings, showing a decrease in the hydrodynamic (Stokes) radius of IRP1 when binding IRE (Yikilmaz et al., 2005). Together, these experiments suggested that apo-IRP1 has an elongated shape. In this study we analyze small angle X-ray scattering (SAXS) data using ab initio calculations to determine the molecular shape of apo-IRP1 in solution, rigid body modeling to assess any differences between its solution conformation and previously determined crystal structures, and the ensemble optimization method (EOM) to investigate its conformational flexibility. 2. Results 2.1. SAXS, DLS, Rgs, and pair distribution functions SAXS profiles of apo-IRP1 were measured at protein concentrations of 0.54, 1.09, and 2.18 mg/ml at a scattering range (q) of 0.065–0.356 Å −1 (Fig. 1a). The Rg was determined from solution scattering profiles at a q × Rg limit of 1.3 using the Guinier approximation (Fig. 1b). At concentrations of 0.54 and 1.09 mg/ml, scattering intensities were linear within the Guinier region, but the 2.18 mg/ml (21.6 μM) samples deviated from linearity, so were not used in further analysis. Subsequent dynamic light scattering (DLS) results showed that apo-IRP1 was monomeric and monodisperse up to the concentration of 1.09 mg/ml (Supplemental Fig. 1). The average Rg was 33.6 ± 0.3 Å (Table 1). This value is much greater than the 28.7 Å calculated from the crystal structure of c-aconitase (PDB ID:
342
O. Shand, K. Volz / Gene 524 (2013) 341–346
Fig. 1. SAXS results. (a) Small-angle X-ray scattering profiles of apo-IRP1 at concentrations of 2.18 mg/mL ○ (green), 1.09 mg/mL □ (orange), and 0.54 mg/mL ◊ (brown). (b) Guinier plots from the solution scattering data of apo-IRP1 at 2.18 mg/mL, 1.09 mg/mL, and 0.54 mg/mL.
2B3X), but similar to the 33.4 Å calculated from the protein component of the IRP1:IRE-RNA complex (PDB ID: 3SNP). The protein molecular mass (Mm) determined directly from the SAXS data by the internal method of Fischer et al. (2010) was 99.2 ± 3.9 kDa, which agrees well with the published value of 100.8 kDa (Selezneva et al., 2006). When the solution scattering curve of apo-IRP1 is compared to the simulated scattering curve of cytosolic aconitase, differences appear at q values of ~ 0.08 Å −1 and greater (Fig. 2a). The overall goodness of fit (χ) between the apo-IRP1 and cytosolic aconitase curves is 6.08, strongly indicating that the curves arise from molecules with different shapes. In contrast, comparison of the apo-IRP1 scattering curve to the simulated scattering curve of the protein component of the IRP1:IRE complex shows that the scattering curves do not diverge from one another until a q value of approximately 0.13 Å −1 (Fig. 2b). The overall χ value between apo-IRP1 and IRP1:IRE complex is 1.34 suggesting that the two molecules have similar structures.
Table 1 Biophysical parameters of IRP1. Rg and Dmax values for apo-IRP1, IRP1 as bound to IRE-RNA, and c-aconitase. Values for RNA-bound IRP1 and c-aconitase were calculated from crystal structures 3SNP and 2B3X, respectively. Protein
RG (Å)
Dmax (Å)
c-Aconitase (PDB ID: 2B3X) IRP1 (PDB ID: 3SNP, chain A) Apo-IRP1
28.7 33.4 33.6 ± 0.3
92 108 118 ± 2
Fig. 2. Analysis of SAXS data. The experimental SAXS data for apo-IRP1 compared with theoretical curves of IRP1 in the (a) c-aconitase conformation and (b) RNA-binding conformation using the program CRYSOL. (c) Pair distribution functions calculated from the experimental data for apo-IRP1 (red), the crystal structure of IRP1 when complexed with IRE-RNA (green), and the crystal structure of c-aconitase minus the iron–sulfur cluster (black).
Pair distribution functions were calculated for apo-IRP1 from the experimental scattering data, and compared with the theoretical curves for IRE-bound IRP1 and cluster-free c-aconitase using the program CRYSOL (Svergun et al., 1995) (Fig. 2c). The pair distribution curve for apo-IRP1 is very similar to that for IRP1 in complex with IRE, and clearly distinct from that for c-aconitase. Apo-IRP1 has the largest Dmax of 118 Å, whereas Dmax of the IRE-bound form of IRP1 is next at 108 Å, and c-aconitase has the smallest Dmax of 92 Å (Table 1). 2.2. Ab initio structures of apo-IRP1 Thirty independent ab initio models of apo-IRP1 were generated from the SAXS data using the program GASBOR (Svergun et al.,
O. Shand, K. Volz / Gene 524 (2013) 341–346
2001). Two models (8 and 17) had normalized spatial discrepancy (NSD) values greater than the 2σ threshold, so were not used in the final composite (see Materials and methods). The remaining models had an average pair-wise NSD of 1.2, indicating good overall agreement. The molecular envelope of the average of the ab-initio dummy-atom models is shown in Fig. 3a. Although the envelope is asymmetric, the enantiomers give the same NSD value (1.02) when fit with the IRP1 model, so the choice of hand was resolved by visual comparison to the all-atom IRP1 model built from the crystal structures (Supplemental Fig. 2).
343
(Fig. 4). A well-defined minimum showed that the models with Rgs near ~ 33 Å agreed best with the experimental data. 2.5. Ensemble optimization of apo-IRP1 SAXS data
Rigid-body modeling was performed using the program SASREF (Petoukhov & Svergun, 2005). The IRP1 molecule was divided into three rigid bodies joined by two hinges (Supplemental Movie). The scattering curve for the final model had a χ value of 1.2, indicating acceptable fit with the experimental data. Consistent with the ab initio modeling results above, the rigid-body model is in an open conformation. Cross validation by superposition of the ab initio model with the rigid body model yielded an NSD value of 0.94.
The ensemble optimization method (EOM) (Bernadó et al., 2007) was used to determine whether apo-IRP1 had a distribution of conformations in solution. A series of fifty models of apo-IRP1 was generated by hinged domain rotations, ranging from the closed conformation of c-aconitase, through the RNA-bound conformation, to hyper-extended conformations (Supplemental Movie). The set of structures that best fit the experimental scattering data was selected using the EOM. The results show that the majority of the apo-IRP1 models are narrowly distributed within an Rg range of 32.8–33.9 Å, with a minor fraction between 31.2 and 32.1 Å (Fig. 5). Analysis of multiple scattering curves of apo-IRP1 scattering data using the ensemble optimization method consistently revealed the presence of the major peak. The presence of the smaller peak was variable; no peak was detected in several calculations. Comparison of the final ensemble scattering curve to the experimental curve gave an acceptable χ value of 1.35 (Fig. 6).
2.4. Conformational analysis of apo-IRP1
3. Discussion
Fifty theoretical models of full-length apo-IRP1 with the two moveable domains in varying degrees of flexion about their hinges were generated by combining the crystal structures of c-aconitase (PDB ID: 2B3X) and RNA-bound IRP1 (PDB ID: 3SNP) using RigiMOL (DeLano, 2002). The two domains' movements were synchronized, or open/closed to the same extent (Supplemental Movie). The scattering curve for each model was compared with the SAXS data, and the χ values for each comparison were plotted as a function of Rg
3.1. Apo-IRP1 is in an open conformation
2.3. Rigid-body models of apo-IRP1
Ab initio and rigid-body modeling analysis of the SAXS data clearly demonstrate that apo-IRP1 exists in an open conformation (Fig. 3). The two hinged domains of apo-IRP1 are at angular and spatial separations indistinguishable from IRP1 when binding IRE-RNA. The modelindependent density envelope from the ab initio calculations accommodates 90% of the protein component of the IRP1:IRE complex without
Fig. 3. Comparison of the three forms of IRP1. (a) The molecular envelope for the ab-initio dummy-atom model of apo-IRP1, with the rigid body model of apo-IRP1 fit into the envelope. The NSD for the ab-initio models is 1.2. (b) The molecular envelope of IRP1 calculated from the RNA-bound IRP1 crystal structure (PDB ID: 3SNP). (c) The envelope of IRP1 calculated from the crystal structure of c-aconitase (PDB ID: 2B3X).
344
O. Shand, K. Volz / Gene 524 (2013) 341–346
Fig. 4. Comparison of scattering data of apo-IRP1 to hypothetical rigid body models of apo-IRP1. Plot of goodness of fit (χ) values for the solution scattering data of apo-IRP1 compared to the 50 rigid-body models of apo-IRP1 generated by linear interpolation of the RNA-bound and c-aconitase conformations of IRP1. The minimum χ corresponds to an Rg of 33.2 Å. The small-dashed line represents the theoretically calculated Rg of c-aconitase (28.7 Å) and the long-dashed line represents the theoretical Rg of the protein component of the IRP1:IRE complex (33.4 Å).
adjustments. These results also give a practical upper limit on the dimensions of the native apo-IRP1 molecule. Based on domain rotations alone, apo-IRP1 has a theoretical maximum Rg of ~36 Å. Further widening makes the χ values progressively worse, and the Rg increments diminishingly small (Fig. 4). Larger values of Rg could be achieved only through translation of domains 3 and 4 away from the core domain. The ab-initio and rigid-body models of apo-IRP1 indicate that the cluster-ligating/RNA-binding surfaces are exposed to solvent. Exposure of these functionally important surfaces enables IRP1 to rapidly respond to the physiological state of the cell. In cells containing low levels of iron, open apo-IRP1 is in an optimal preset conformation to allow for immediate interaction with the IREs. The shift from apoto the RNA-binding form apparently would require minimal rotations of domains 3 and 4 in order to establish the high affinity protein-RNA interactions. Alternately, in cells with sufficient iron, the cysteine residues would be ready to accept an iron–sulfur cluster prior to the large conformational change necessary for closure of domains 3 and 4. 3.2. The open conformation of apo-IRP1 is static The EOM is useful for detecting an ensemble of available conformations in solution, especially for highly flexible molecules that dynamically sample many different conformations (Bernadó et al., 2007; Bernadó, 2010). Assuming that the dynamics of apo-IRP1 is restricted
Fig. 6. Scattering curve for experimental apo-IRP1 and ensemble of solutions. The scattering curve of apo-IRP1 obtained from experimental data (solid) compared with the scattering curve of the ensemble of structures (dashed). Quantitative agreement between the two curves (χ) is 1.35.
to the trajectory set by the two-hinge model (see Materials and methods), the EOM analysis of the SAXS results shows that the molecules have one static conformation, with an Rg narrowly distributed around 33.5 Å (Fig. 5). No closed conformations were detected. A much smaller peak, representing less than 4% of the ensemble, appeared at the intermediate Rg value of 31.5 Å. The EOM results are validated by comparison of the scattering curves with the solution scattering data (Fig. 6). The χ value from this analysis is 1.35, indicating a good match between the EOM results and the experimental scattering data. The curves' deviations at the highest scattering angles are negligible due to the dominance of the low angle scattering data in this method. 3.3. Apo-IRP1 is monomeric in the conditions studied A number of investigations point toward dimerization as important in aconitase structure and function (Tang et al., 2005; Tsuchiya et al., 2008; Tsuchiya et al., 2009), including an analytical ultracentrifugation study which indicated that the apo form of IRP1 has a monomer–dimer equilibrium in the μM range (Yikilmaz et al., 2005). The X-ray- and light-scattering results here do not show any evidence of IRP1 dimerization. Higher-order aggregation was observed at protein concentrations of ~ 20 μM, and those conditions were excluded from the final data analysis. In all our experiments, samples were prepared with 1 mM fresh DTT. Although IRP1 has normal cysteine content (nine cysteines in 889 residues), the proposed open conformation of apo-IRP1 would make the majority of the cysteine residues (six of nine) more exposed to solvent (Supplemental Fig. 3). In the previous analytical ultracentrifugation experiments (5), DTT was not included. Under non-reducing conditions apo-IRP1 dimers may arise from intermolecular disulfide bond formation. The early neutron scattering experiments with IRP1 showed minor aggregation in spite of 1 mM DTT (Brazzolotto et al., 2002), but as the authors noted, it may have been caused by the higher concentration of protein (5 mg/ml, or 50 μM), or D2O (62%), or both. 3.4. The open apo-IRP1 conformation facilitates subsequent conversions
Fig. 5. Ensemble optimization method analysis of apo-IRP1. This distribution of Rg values from EOM analysis of apo-IRP1 scattering data. A major peak is centered about an Rg value of 33–34 Å and a minor peak is center around 31.5 Å. The dotted lines represent the Rg values for c-aconitase (28.7 Å) and apo-IRP1 (33.6 Å).
Multidomain proteins take advantage of flexible architecture to optimize ligand binding and catalytic efficiency. As proteins evolve for functional purposes, so IRP1 has developed mechanisms for adapting to the demanding requirement of binding two structurally unrelated ligands: IRE RNA, or [4Fe–4S] cluster. The open conformation of apo-IRP1 appears most suitable for binding IRE-RNAs. In this conformation, the RNA-binding surfaces
O. Shand, K. Volz / Gene 524 (2013) 341–346
are already positioned to complement the rough shape of the stem loop, so there are no slow, large-scale transformational barriers to overcome. Any remaining protein adjustments would be on the faster local level, for capturing the dynamically exposed nucleotides of the IRE (McCallum & Pardi, 2003; Showalter et al., 2005). These include rearranging the residues of the 430s and 530s to bind the terminal loop, and opening the pocket in domain 4 to accept the bulged C8. While the crystal structures of the IRP1:IRE complexes (Walden et al., 2006; Walden et al., 2012) might give the impression of rigidly interlocked protein and RNA molecules, combining them with the SAXS results shows that minor rearrangements may be all that are necessary for complex formation. However, explaining the situation may require a more complicated model. Crowding theory argues that in the high-density interior of the cell, flexible macromolecules favor closed, globular conformations (Zhou et al., 2008; Dong et al., 2010). So if IRP1 is more closed under crowded conditions, different mechanisms may be at play in the protein:RNA association. This can be investigated through SAXS experiments on IRP1 and IREs with the appropriate crowding agents (Kilburn et al., 2010). But after formation of the IRP1:IRE complex, these excluded volume effects should only reinforce the effectively irreversible (picomolar) binding strength. Understanding the conversion of apo-IRP1 to c-aconitase is even more challenging because iron sulfur cluster biogenesis is a complex, assisted process involving multiple steps (Sharma et al., 2010). The attending cluster assembly proteins would conceivably need wide access to the active site. Thus the open form of apo-IRP1 seems the best conformation for the initial steps of cluster assembly and insertion. There are few structures of complexes in cluster biogenesis (Shi et al., 2010; Marinoni et al., 2012), and to our knowledge, none that involve IRP1 and the components of the cytosolic assembly system. The sequence of events immediately following cluster insertion in IRP1 can only be speculated: whether there is an open intermediate of assembled c-aconitase after release, or how it closes is not known. It is generally accepted that the holo-forms of iron–sulfur proteins are more stable than their apo-counterparts, so one would expect spontaneous closure for the sake of solvent exclusion and stability. Aconitase is an ancient enzyme (Beinert & Kennedy, 1993). The phylogenetic relationships in the aconitase superfamily argue that IRP1 was an enzyme long before it acquired RNA-repressor functionality (Gruer et al., 1997; Baughn & Malamy, 2002). IRP1 and most other aconitase homologs retained the basic four-domain organization (Gruer et al., 1997; Makarova & Koonin, 2003), but some unusual domain-sorting events occurred during evolution of the superfamily. Odd variations in the placement of domain four—such as a permuted sequence order (e.g., bacterial aconitases B), or a separately coded gene product (e.g., isopropylmalate dehydratases)—support the idea that positional lability of domain four has functional importance. This was recognized early in the structural analyses of aconitases (Lauble & Stout, 1995), and raises the question whether the apo forms of other aconitases (the IRP-aconitase A group, and the closely related m-aconitases) are in open conformations. Since the closest homologs of IRPs (bacterial aconitase-A and plant aconitases) are from organisms not known to have functioning IRP:IRE systems (Piccinelli & Samuelsson, 2007; Arnaud et al., 2007), accessibility for the sake of cluster insertion in IRP1 evolutionarily preceded IRE binding. It has been observed that bacterial aconitases interact with polynucleotides (Tang & Guest, 1999; Alén & Sonenshein, 2002). Using IRP1 as the structural example, high-affinity nucleic acid binding would require domains three and four to be widely separated. Although it is an exciting possibility that these IRP homologs also offer clamshell-like access for purposes of cluster assembly and RNA binding, there is as yet no definitive evidence that their apo forms exist in an open conformation and tightly bind RNA stem-loops. Additional SAXS experiments with other apo-aconitases may provide clearer answers regarding conformational variability in the aconitase superfamily.
345
4. Materials and methods 4.1. Protein purification and sample preparation Hexa-His tagged rabbit IRP1 was over-expressed in Saccharomyces cerevisiae strain BJ5465 (Swenson & Walden, 1994). Cells were grown overnight in YPAD media and then inoculated in SD-URA media to an OD600 of 0.05. Once cells reached logarithmic growth phase they were inoculated to an OD600 of 0.1 in SD media containing 2% raffinose and grown overnight. Protein expression was induced by adding galactose to a final concentration of 2%, and cells were grown for 6–9 h before harvesting. Cells were harvested by centrifugation and washed once with sterile water. All purification steps were performed aerobically. Cells were lysed by manual shaking with 0.04–0.06 mm diameter glass beads, and the lysate was centrifuged at 14,000 g for 1 h at 4 °C. The supernatant was filtered, passed through a column containing Ni-NTA resin (Qiagen), and washed with 500 mM NaCl in 20 mM sodium phosphate, pH 6.3. Protein was eluted from the column with 500 mM NaCl and 300 mM imidazole in 20 mM sodium phosphate, pH 6.3. The eluate was dialyzed overnight against 5 mM NaCl and 100 mM Tris, pH 8.0. Subsequently, the dialysate was loaded on a WPQUAT anion exchange column and eluted with a linear 5–500 mM gradient of NaCl. Fractions containing ≥90% pure IRP1 were pooled and dialyzed overnight against 100 mM NaCl, 1 mM DTT, 100 mM Tris, pH 7.4, and 10% glycerol. The extinction coefficient of IRP1 was calculated from the IRP1 protein sequence using the program ProtParm (Gasteiger et al., 2005). Protein concentration was determined by measuring absorbance at 280 nm and using the extinction coefficient IRP1 of 89,500 M−1·cm −1. Samples were checked for monodispersity using DLS with a Malvern Zetasizer Nano-S. Samples and matched dialysates were stored at −80 °C until needed. 4.2. SAXS experiments All scattering experiments were conducted at the BioCAT beamline 18-ID of the Advanced Photon Source, Argonne National Laboratory (Fischetti et al., 2004). Scattering data were collected at a q range between 0.006 and 0.356 Å−1. Samples were loaded in a quartz capillary flow cell and exposed to 12 keV X-rays for 1 s at 5 second intervals twenty times. Scattering data were measured using a CCD detector positioned 2422 mm from the sample. Radiation damage was monitored by analysis of the scattering data and by SDS-PAGE. Data reduction was performed using Igor-Pro (WaveMetrics Inc., Lake Oswego, OR, USA) equipped with custom SAXS analysis tools designed for the BIO-CAT beamline. 4.3. Ab initio modeling The program GASBOR (Svergun et al., 2001) was used to construct 30 dummy-residue models from the scattering curves of apo-IRP1. The DAMAVER (Volkov & Svergun, 2003) suite of programs was used to select a set of closely corresponding models, superimpose them (Kozin & Svergun, 2001), and generate a final composite model. Criteria for excluding models were done by calculating the average and variance of NSD values among all models. Models with NSD values 2σ greater than the average were rejected. An envelope representation of the final bead model was prepared with the program Situs (Wriggers & Chacón, 2001). 4.4. Rigid-body modeling The IRP1 molecule (PDB ID: 3SNP, molecule A) was divided into 3 parts: a static core (domains 1 and 2), and two flexing domains (domains 3 and domain 4). Rigid-body modeling was performed by two separate methods using the programs RigiMOL (DeLano, 2002) and SASREF (Petoukhov & Svergun, 2005). A total of 11 connectivity
346
O. Shand, K. Volz / Gene 524 (2013) 341–346
restraints were applied to the rigid body system (see Supplemental Tables 1 and 2). These restraints were designed to retain the continuity of the polypeptide backbone and establish connections between inter-domain atom pairs whose relative distances were predicted to not change upon rotation of domains. 4.5. Generation of hypothetical intermediate structures Thirty equally-spaced models between the two functional states of IRP1 were generated by RigiMOL (DeLano, 2002) using the crystal structures of c-aconitase (PDB ID: 2B3X) and RNA-bound IRP1 (PDB ID: 3SNP). Twenty additional models with domains 3 and 4 flexing along their same hinged trajectories but extending beyond RNA-bound IRP1 were added to explore hyper-extended conformations (Supplemental Movie). The scattering curves of the fifty model structures were compared to the experimental scattering data using the program CRYSOL (Svergun et al., 1995) (Fig. 4). 4.6. Restrained ensemble optimization method The ensemble optimization method (EOM) utilizes the fact that the scattering curve of a flexible multi-domain protein may arise from multiple conformations of the protein in solution. The EOM uses a genetic algorithm to determine a subset of conformations representative of the solution scattering data (Bernadó et al., 2007). Attempts were made to generate a larger set of conformers using the EOM software package; however, the majority of conformers generated were not structurally plausible. As described in the previous section, a pool of fifty structures of IRP1 on the trajectory containing the c-aconitase and RNA-bound conformations was analyzed. Although the structures represent only a fraction of total possible apo-IRP1 conformations, the set of 50 conformers cover the entire range of possible Rg values for IRP1. Conflict of interest There is no conflict of interest. Acknowledgments This work was supported by NIH grant GM-71504 to KV. We thank the staff at BioCAT for their assistance in the collection and preliminary processing of the SAXS data. Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.gene.2013.03.112. References Alén, C., Sonenshein, A.L., 2002. Bacillus subtilis aconitase is an RNA-binding protein. Proc. Natl. Acad. Sci. U. S. A. 96, 10412–10417. Anderson, C.P., Shen, M., Eisenstein, R.S., Leibold, E.A., 2012. Mammalian iron metabolism and its control by iron regulatory proteins. Biochim. Biophys. Acta 1823, 1468–1483. Arnaud, N., et al., 2007. The iron-responsive element (IRE)/iron-regulatory protein 1 (IRP1)-cytosolic aconitase iron-regulatory switch does not operate in plants. Biochem. J. 405, 523–531. Baughn, A.D., Malamy, M.H., 2002. A mitochondrial-like aconitase in the bacterium Bacteroides fragilis: implications for the evolution of the mitochondrial Krebs cycle. Proc. Natl. Acad. Sci. U. S. A. 99, 4662–4667. Beinert, H., Kennedy, M.C., 1993. Aconitase: a two-faced protein: enzyme and ironregulatory factor. FASEB J. 7, 1442–1449. Bernadó, P., 2010. Effects of interdomain dynamics on the structure determination of modular proteins by small-angle scattering. Eur. Biophys. J. 39, 769–780. Bernadó, P., Mylonas, E., Petoukhov, M.V., Blackledge, M., Svergun, D.I., 2007. Structural characterization of flexible proteins using small-angle X-ray scattering. J. Am. Chem. Soc. 129, 5656–5664.
Brazzolotto, X., Timmins, P., Dupont, Y., Moulis, J.-M., 2002. Structural changes associated with switching activities of human iron regulatory protein 1. J. Biol. Chem. 277, 11995–12000. DeLano, W.L., 2002. The PyMOL Molecular Graphics System. DeLano Scientific, San Carlos, CA. Dong, H., Qin, S., Zhou, H.-X., 2010. Effects of macromolecular crowding on protein conformational changes. PLoS Comput. Biol. 6, e1000833. Dupuy, J., Volbeda, A., Carpentier, P., Darnault, C., Moulis, J.-M., Fontecilla-Camps, J.C., 2006. Crystal structure of human iron regulatory protein 1 as c-aconitase. Structure 14, 129–139. Fischer, H., de Oliveira Neto, M., Napolitano, H.B., Polikarpov, I., Craievich, A.F., 2010. Determination of the molecular weight of proteins in solution from a single small-angle X-ray scattering measurement on a relative scale. J. Appl. Cryst. 43, 101–109. Fischetti, R., et al., 2004. The BioCAT undulator beamline 18ID: a facility for biological non-crystalline diffraction and X-ray absorption spectroscopy at the advanced photon source. J. Synchrotron Radiat. 11, 399–405. Gasteiger, E., et al., 2005. Protein identification and analysis tools on the ExPASy server. The Proteomics Protols Handbook. 571–607. Gruer, M.J., Artymiuk, P.J., Guest, J.R., 1997. The aconitase family: three structural variations on a common theme. Trends Biochem. Sci. 22, 3–6. Kilburn, D., Roh, J.H., Guo, L., Briber, R.M., Woodson, S.A., 2010. Molecular crowding stabilizes folded RNA structure by the excluded volume effect. J. Am. Chem. Soc. 132, 8690–8696. Kozin, M.B., Svergun, D.I., 2001. Automated matching of high- and low-resolution structural models. J. Appl. Crystallogr. 34, 33–41. Lauble, H., Stout, C.D., 1995. Steric and conformational features of the aconitase mechanism. Proteins 22, 1–11. Makarova, K.S., Koonin, E.V., 2003. Filling a gap in the central metabolism of archaea: prediction of a novel aconitase by comparative-genomic analysis. FEMS Microbiol. Lett. 27, 17–23. Marinoni, E.N., et al., 2012. (IscS-IscU)2 complex structures provide insights into Fe2S2 biogenesis and transfer. Angew. Chem. Int. Ed. 51, 5439–5442. McCallum, S.A., Pardi, A., 2003. Refined solution structure of the iron-responsive element RNA using residual dipolar couplings. J. Mol. Biol. 326, 1037–1050. Petoukhov, M.V., Svergun, D.I., 2005. Global rigid body modeling of macromolecular complexes against small-angle scattering data. Biophys. J. 89, 1237–1250. Piccinelli, P., Samuelsson, T., 2007. Evolution of the iron-responsive element. RNA 13, 952–966. Schalinske, K.L., Anderson, S.A., Tuazon, P.T., Chen, O.S., Kennedy, M.C., Eisenstein, R.S., 1997. The iron–sulfur cluster of iron regulatory protein 1 modulates the accessibility of RNA binding and phosphorylation sites. Biochemistry 36, 3950–3958. Selezneva, A.I., Cavigiolio, G., Theil, E.C., Walden, W.E., Volz, K., 2006. Crystallization and preliminary X-ray diffraction analysis of iron regulatory protein-1 in complex with ferritin IRE RNA. Acta Crystallogr. F62, 249–252. Sharma, A.K., Pallesen, L.J., Spang, R.J., Walden, W.E., 2010. Cytosolic iron–sulfur cluster assembly (CIA) system: factors, mechanism, and relevance to cellular iron regulation. J. Biol. Chem. 285, 26745–26751. Shi, R., et al., 2010. Structural basis for Fe–S cluster assembly and tRNA thiolation mediated by IscS protein–protein interactions. PLoS Biol. 8, e1000354. Showalter, S.A., Baker, N.A., Tang, C., Hall, K.B., 2005. Iron responsive element RNA flexibility described by NMR and isotropic reorientational eigenmode dynamics. J. Biomol. NMR 32, 179–193. Svergun, D.I., Barberato, C., Koch, M.H.J., 1995. CRYSOL: a program to evaluate X-ray solution scattering of biological macromolecules from atomic coordinates. J. Appl. Cryst. 28, 768–773. Svergun, D.I., Petoukhov, M.V., Koch, M.H.J., 2001. Determination of domain structure of proteins from X-ray solution scattering. Biophys. J. 80, 2946–2953. Swenson, G.R., Walden, W.E., 1994. Localization of an RNA binding element of the iron responsive element binding protein within a proteolytic fragment containing iron coordination ligands. Nucleic Acids Res. 22, 2627–2633. Tang, Y., Guest, J.R., 1999. Direct evidence for mRNA binding and post-transcriptional regulation by Escherichia coli aconitases. Microbiology 145, 3069–3079. Tang, Y., Guest, J.R., Artymiuk, P.J., Green, J., 2005. Switching aconitase B between catalytic and regulatory modes involves iron-dependent dimer formation. Mol. Microbiol. 56, 1149–1158. Tsuchiya, D., Shimizu, N., Tomita, M., 2008. Versatile architecture of a bacterial aconitase B and its catalytic performance in the sequential reaction coupled with isocitrate dehydrogenase. Biochim. Biophys. Acta 1784, 1847–1856. Tsuchiya, D., Shimizu, N., Tomita, M., 2009. Cooperativity of two active sites in bacterial homodimeric aconitases. Biochim. Biophys. Res. Comm. 379, 485–488. Volkov, V.V., Svergun, D.I., 2003. Uniqueness of ab initio shape determination in smallangle scattering. J. Appl. Crystallogr. 36, 860–864. Walden, W.E., et al., 2006. Structure of dual function iron regulatory protein 1 complexed with ferritin IRE-RNA. Science 314, 1903–1909. Walden, W.E., Selezneva, A., Volz, K., 2012. Accommodating variety in iron responsive elements: crystal structure of transferrin receptor 1 B IRE bound to iron regulatory protein 1. FEBS Lett. 586, 32–35. Wriggers, W., Chacón, P., 2001. Using Situs for the registration of protein structures with low-resolution bead models from X-ray solution scattering. J. Appl. Crystallogr. 34, 773–776. Yikilmaz, E., Rouault, T.A., Schuck, P., 2005. Self-association and ligand-induced conformational changes of iron regulatory proteins 1 and 2. Biochemistry 44, 8470–8478. Zhou, H.-X., Rivas, G.N., Minton, A.P., 2008. Macromolecular crowding and confinement: biochemical, biophysical, and potential physiological consequences. Annu. Rev. Biophys. 37, 375–397.
Contents lists available at SciVerse ScienceDirect
Gene journal homepage: www.elsevier.com/locate/gene
The solution structure of apo-iron regulatory protein 1 O'Neil Shand, Karl Volz ⁎ Department of Microbiology and Immunology, University of Illinois at Chicago, Chicago, IL 60612, USA
a r t i c l e
i n f o
Article history: Accepted 25 March 2013 Available online 13 April 2013 Keywords: Iron homeostasis Iron regulatory protein Bifunctionality RNA-binding Iron–sulfur cluster
a b s t r a c t Iron is a cofactor for many proteins that are involved in essential metabolic processes. However, iron must be strictly regulated because it can react with oxygen to generate cytotoxic reactive oxygen intermediates. Iron regulatory protein 1 (IRP1) is a bi-functional protein that can act either as a post-transcriptional regulator of mRNAs containing iron responsive elements, or as a [4Fe–4S] cluster-containing cytosolic aconitase. Previous X-ray crystallography results show that IRP1 is in an open L-shape conformation when bound to IRE-RNAs, and in a globular conformation when it binds an iron–sulfur cluster. The structure of apo-IRP1 and the mechanism by which it transforms to either functional state is unknown. Therefore, small angle X-ray scattering was used to determine the low resolution solution structure of apo-IRP1 and to characterize its biophysical properties. These results show that apo-IRP1 has a radius of gyration (Rg) of 33.6 ± 0.3 Å, and a Dmax of 118 ± 2 Å. The ab initio and rigid-body modeling results show that apo-IRP1 is in an open conformation in solution, and the ensemble optimization results show that the molecules stay narrowly distributed about a Rg of 33–34 Å. The open apo-IRP1 conformation seems optimal for subsequent conversion to either functional end state: RNA-binding, or cytosolic aconitase. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Iron is an essential element utilized by most organisms. However, iron levels must be strictly regulated in order to prevent deleterious effects of iron insufficiency or iron overload. In metazoans, iron regulatory protein 1 (IRP1) is a bi-functional protein that maintains iron homeostasis by integrating post-transcriptional regulation of iron metabolism genes with the cellular iron status (Anderson et al., 2012). Elucidations of the crystal structures of IRP1 in complex with iron responsive element (IRE) RNA (Walden et al., 2006) and in the form of cytosolic (c-) aconitase (Dupuy et al., 2006) were major steps forward in determining the structural principles responsible for IRP1 bi-functionality. Although much progress has been made in understanding the two functional roles of IRP1, little is known about the structure of the apo-protein and how it acquires either the IRE-RNA or the [4Fe–4S] cluster. Prior to this work, apo-IRP1 was characterized by several biochemical and biophysical methods. Proteolysis studies showed that apo-IRP1 was more susceptible to degradation than c-aconitase (Schalinske et al., 1997). Neutron scattering experiments were the first to determine the radii of gyration (Rgs) of the different states of IRP1, and demonstrate that apo-IRP1 has a Rg larger than the protein
Abbreviations: DLS, dynamic light scattering; DTT, dithiothreitol; IRE, iron responsive element; IRP, iron regulatory protein; Rg, radius of gyration; SAXS, small-angle X-ray scattering. ⁎ Corresponding author. Tel.: +1 3129962314; fax: +1 3123554535. E-mail addresses: [email protected] (O. Shand), [email protected] (K. Volz). 0378-1119/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.gene.2013.03.112
when complexed with IRE-RNA (Brazzolotto et al., 2002). Analytical ultracentrifugation results were consistent with the neutron scattering findings, showing a decrease in the hydrodynamic (Stokes) radius of IRP1 when binding IRE (Yikilmaz et al., 2005). Together, these experiments suggested that apo-IRP1 has an elongated shape. In this study we analyze small angle X-ray scattering (SAXS) data using ab initio calculations to determine the molecular shape of apo-IRP1 in solution, rigid body modeling to assess any differences between its solution conformation and previously determined crystal structures, and the ensemble optimization method (EOM) to investigate its conformational flexibility. 2. Results 2.1. SAXS, DLS, Rgs, and pair distribution functions SAXS profiles of apo-IRP1 were measured at protein concentrations of 0.54, 1.09, and 2.18 mg/ml at a scattering range (q) of 0.065–0.356 Å −1 (Fig. 1a). The Rg was determined from solution scattering profiles at a q × Rg limit of 1.3 using the Guinier approximation (Fig. 1b). At concentrations of 0.54 and 1.09 mg/ml, scattering intensities were linear within the Guinier region, but the 2.18 mg/ml (21.6 μM) samples deviated from linearity, so were not used in further analysis. Subsequent dynamic light scattering (DLS) results showed that apo-IRP1 was monomeric and monodisperse up to the concentration of 1.09 mg/ml (Supplemental Fig. 1). The average Rg was 33.6 ± 0.3 Å (Table 1). This value is much greater than the 28.7 Å calculated from the crystal structure of c-aconitase (PDB ID:
342
O. Shand, K. Volz / Gene 524 (2013) 341–346
Fig. 1. SAXS results. (a) Small-angle X-ray scattering profiles of apo-IRP1 at concentrations of 2.18 mg/mL ○ (green), 1.09 mg/mL □ (orange), and 0.54 mg/mL ◊ (brown). (b) Guinier plots from the solution scattering data of apo-IRP1 at 2.18 mg/mL, 1.09 mg/mL, and 0.54 mg/mL.
2B3X), but similar to the 33.4 Å calculated from the protein component of the IRP1:IRE-RNA complex (PDB ID: 3SNP). The protein molecular mass (Mm) determined directly from the SAXS data by the internal method of Fischer et al. (2010) was 99.2 ± 3.9 kDa, which agrees well with the published value of 100.8 kDa (Selezneva et al., 2006). When the solution scattering curve of apo-IRP1 is compared to the simulated scattering curve of cytosolic aconitase, differences appear at q values of ~ 0.08 Å −1 and greater (Fig. 2a). The overall goodness of fit (χ) between the apo-IRP1 and cytosolic aconitase curves is 6.08, strongly indicating that the curves arise from molecules with different shapes. In contrast, comparison of the apo-IRP1 scattering curve to the simulated scattering curve of the protein component of the IRP1:IRE complex shows that the scattering curves do not diverge from one another until a q value of approximately 0.13 Å −1 (Fig. 2b). The overall χ value between apo-IRP1 and IRP1:IRE complex is 1.34 suggesting that the two molecules have similar structures.
Table 1 Biophysical parameters of IRP1. Rg and Dmax values for apo-IRP1, IRP1 as bound to IRE-RNA, and c-aconitase. Values for RNA-bound IRP1 and c-aconitase were calculated from crystal structures 3SNP and 2B3X, respectively. Protein
RG (Å)
Dmax (Å)
c-Aconitase (PDB ID: 2B3X) IRP1 (PDB ID: 3SNP, chain A) Apo-IRP1
28.7 33.4 33.6 ± 0.3
92 108 118 ± 2
Fig. 2. Analysis of SAXS data. The experimental SAXS data for apo-IRP1 compared with theoretical curves of IRP1 in the (a) c-aconitase conformation and (b) RNA-binding conformation using the program CRYSOL. (c) Pair distribution functions calculated from the experimental data for apo-IRP1 (red), the crystal structure of IRP1 when complexed with IRE-RNA (green), and the crystal structure of c-aconitase minus the iron–sulfur cluster (black).
Pair distribution functions were calculated for apo-IRP1 from the experimental scattering data, and compared with the theoretical curves for IRE-bound IRP1 and cluster-free c-aconitase using the program CRYSOL (Svergun et al., 1995) (Fig. 2c). The pair distribution curve for apo-IRP1 is very similar to that for IRP1 in complex with IRE, and clearly distinct from that for c-aconitase. Apo-IRP1 has the largest Dmax of 118 Å, whereas Dmax of the IRE-bound form of IRP1 is next at 108 Å, and c-aconitase has the smallest Dmax of 92 Å (Table 1). 2.2. Ab initio structures of apo-IRP1 Thirty independent ab initio models of apo-IRP1 were generated from the SAXS data using the program GASBOR (Svergun et al.,
O. Shand, K. Volz / Gene 524 (2013) 341–346
2001). Two models (8 and 17) had normalized spatial discrepancy (NSD) values greater than the 2σ threshold, so were not used in the final composite (see Materials and methods). The remaining models had an average pair-wise NSD of 1.2, indicating good overall agreement. The molecular envelope of the average of the ab-initio dummy-atom models is shown in Fig. 3a. Although the envelope is asymmetric, the enantiomers give the same NSD value (1.02) when fit with the IRP1 model, so the choice of hand was resolved by visual comparison to the all-atom IRP1 model built from the crystal structures (Supplemental Fig. 2).
343
(Fig. 4). A well-defined minimum showed that the models with Rgs near ~ 33 Å agreed best with the experimental data. 2.5. Ensemble optimization of apo-IRP1 SAXS data
Rigid-body modeling was performed using the program SASREF (Petoukhov & Svergun, 2005). The IRP1 molecule was divided into three rigid bodies joined by two hinges (Supplemental Movie). The scattering curve for the final model had a χ value of 1.2, indicating acceptable fit with the experimental data. Consistent with the ab initio modeling results above, the rigid-body model is in an open conformation. Cross validation by superposition of the ab initio model with the rigid body model yielded an NSD value of 0.94.
The ensemble optimization method (EOM) (Bernadó et al., 2007) was used to determine whether apo-IRP1 had a distribution of conformations in solution. A series of fifty models of apo-IRP1 was generated by hinged domain rotations, ranging from the closed conformation of c-aconitase, through the RNA-bound conformation, to hyper-extended conformations (Supplemental Movie). The set of structures that best fit the experimental scattering data was selected using the EOM. The results show that the majority of the apo-IRP1 models are narrowly distributed within an Rg range of 32.8–33.9 Å, with a minor fraction between 31.2 and 32.1 Å (Fig. 5). Analysis of multiple scattering curves of apo-IRP1 scattering data using the ensemble optimization method consistently revealed the presence of the major peak. The presence of the smaller peak was variable; no peak was detected in several calculations. Comparison of the final ensemble scattering curve to the experimental curve gave an acceptable χ value of 1.35 (Fig. 6).
2.4. Conformational analysis of apo-IRP1
3. Discussion
Fifty theoretical models of full-length apo-IRP1 with the two moveable domains in varying degrees of flexion about their hinges were generated by combining the crystal structures of c-aconitase (PDB ID: 2B3X) and RNA-bound IRP1 (PDB ID: 3SNP) using RigiMOL (DeLano, 2002). The two domains' movements were synchronized, or open/closed to the same extent (Supplemental Movie). The scattering curve for each model was compared with the SAXS data, and the χ values for each comparison were plotted as a function of Rg
3.1. Apo-IRP1 is in an open conformation
2.3. Rigid-body models of apo-IRP1
Ab initio and rigid-body modeling analysis of the SAXS data clearly demonstrate that apo-IRP1 exists in an open conformation (Fig. 3). The two hinged domains of apo-IRP1 are at angular and spatial separations indistinguishable from IRP1 when binding IRE-RNA. The modelindependent density envelope from the ab initio calculations accommodates 90% of the protein component of the IRP1:IRE complex without
Fig. 3. Comparison of the three forms of IRP1. (a) The molecular envelope for the ab-initio dummy-atom model of apo-IRP1, with the rigid body model of apo-IRP1 fit into the envelope. The NSD for the ab-initio models is 1.2. (b) The molecular envelope of IRP1 calculated from the RNA-bound IRP1 crystal structure (PDB ID: 3SNP). (c) The envelope of IRP1 calculated from the crystal structure of c-aconitase (PDB ID: 2B3X).
344
O. Shand, K. Volz / Gene 524 (2013) 341–346
Fig. 4. Comparison of scattering data of apo-IRP1 to hypothetical rigid body models of apo-IRP1. Plot of goodness of fit (χ) values for the solution scattering data of apo-IRP1 compared to the 50 rigid-body models of apo-IRP1 generated by linear interpolation of the RNA-bound and c-aconitase conformations of IRP1. The minimum χ corresponds to an Rg of 33.2 Å. The small-dashed line represents the theoretically calculated Rg of c-aconitase (28.7 Å) and the long-dashed line represents the theoretical Rg of the protein component of the IRP1:IRE complex (33.4 Å).
adjustments. These results also give a practical upper limit on the dimensions of the native apo-IRP1 molecule. Based on domain rotations alone, apo-IRP1 has a theoretical maximum Rg of ~36 Å. Further widening makes the χ values progressively worse, and the Rg increments diminishingly small (Fig. 4). Larger values of Rg could be achieved only through translation of domains 3 and 4 away from the core domain. The ab-initio and rigid-body models of apo-IRP1 indicate that the cluster-ligating/RNA-binding surfaces are exposed to solvent. Exposure of these functionally important surfaces enables IRP1 to rapidly respond to the physiological state of the cell. In cells containing low levels of iron, open apo-IRP1 is in an optimal preset conformation to allow for immediate interaction with the IREs. The shift from apoto the RNA-binding form apparently would require minimal rotations of domains 3 and 4 in order to establish the high affinity protein-RNA interactions. Alternately, in cells with sufficient iron, the cysteine residues would be ready to accept an iron–sulfur cluster prior to the large conformational change necessary for closure of domains 3 and 4. 3.2. The open conformation of apo-IRP1 is static The EOM is useful for detecting an ensemble of available conformations in solution, especially for highly flexible molecules that dynamically sample many different conformations (Bernadó et al., 2007; Bernadó, 2010). Assuming that the dynamics of apo-IRP1 is restricted
Fig. 6. Scattering curve for experimental apo-IRP1 and ensemble of solutions. The scattering curve of apo-IRP1 obtained from experimental data (solid) compared with the scattering curve of the ensemble of structures (dashed). Quantitative agreement between the two curves (χ) is 1.35.
to the trajectory set by the two-hinge model (see Materials and methods), the EOM analysis of the SAXS results shows that the molecules have one static conformation, with an Rg narrowly distributed around 33.5 Å (Fig. 5). No closed conformations were detected. A much smaller peak, representing less than 4% of the ensemble, appeared at the intermediate Rg value of 31.5 Å. The EOM results are validated by comparison of the scattering curves with the solution scattering data (Fig. 6). The χ value from this analysis is 1.35, indicating a good match between the EOM results and the experimental scattering data. The curves' deviations at the highest scattering angles are negligible due to the dominance of the low angle scattering data in this method. 3.3. Apo-IRP1 is monomeric in the conditions studied A number of investigations point toward dimerization as important in aconitase structure and function (Tang et al., 2005; Tsuchiya et al., 2008; Tsuchiya et al., 2009), including an analytical ultracentrifugation study which indicated that the apo form of IRP1 has a monomer–dimer equilibrium in the μM range (Yikilmaz et al., 2005). The X-ray- and light-scattering results here do not show any evidence of IRP1 dimerization. Higher-order aggregation was observed at protein concentrations of ~ 20 μM, and those conditions were excluded from the final data analysis. In all our experiments, samples were prepared with 1 mM fresh DTT. Although IRP1 has normal cysteine content (nine cysteines in 889 residues), the proposed open conformation of apo-IRP1 would make the majority of the cysteine residues (six of nine) more exposed to solvent (Supplemental Fig. 3). In the previous analytical ultracentrifugation experiments (5), DTT was not included. Under non-reducing conditions apo-IRP1 dimers may arise from intermolecular disulfide bond formation. The early neutron scattering experiments with IRP1 showed minor aggregation in spite of 1 mM DTT (Brazzolotto et al., 2002), but as the authors noted, it may have been caused by the higher concentration of protein (5 mg/ml, or 50 μM), or D2O (62%), or both. 3.4. The open apo-IRP1 conformation facilitates subsequent conversions
Fig. 5. Ensemble optimization method analysis of apo-IRP1. This distribution of Rg values from EOM analysis of apo-IRP1 scattering data. A major peak is centered about an Rg value of 33–34 Å and a minor peak is center around 31.5 Å. The dotted lines represent the Rg values for c-aconitase (28.7 Å) and apo-IRP1 (33.6 Å).
Multidomain proteins take advantage of flexible architecture to optimize ligand binding and catalytic efficiency. As proteins evolve for functional purposes, so IRP1 has developed mechanisms for adapting to the demanding requirement of binding two structurally unrelated ligands: IRE RNA, or [4Fe–4S] cluster. The open conformation of apo-IRP1 appears most suitable for binding IRE-RNAs. In this conformation, the RNA-binding surfaces
O. Shand, K. Volz / Gene 524 (2013) 341–346
are already positioned to complement the rough shape of the stem loop, so there are no slow, large-scale transformational barriers to overcome. Any remaining protein adjustments would be on the faster local level, for capturing the dynamically exposed nucleotides of the IRE (McCallum & Pardi, 2003; Showalter et al., 2005). These include rearranging the residues of the 430s and 530s to bind the terminal loop, and opening the pocket in domain 4 to accept the bulged C8. While the crystal structures of the IRP1:IRE complexes (Walden et al., 2006; Walden et al., 2012) might give the impression of rigidly interlocked protein and RNA molecules, combining them with the SAXS results shows that minor rearrangements may be all that are necessary for complex formation. However, explaining the situation may require a more complicated model. Crowding theory argues that in the high-density interior of the cell, flexible macromolecules favor closed, globular conformations (Zhou et al., 2008; Dong et al., 2010). So if IRP1 is more closed under crowded conditions, different mechanisms may be at play in the protein:RNA association. This can be investigated through SAXS experiments on IRP1 and IREs with the appropriate crowding agents (Kilburn et al., 2010). But after formation of the IRP1:IRE complex, these excluded volume effects should only reinforce the effectively irreversible (picomolar) binding strength. Understanding the conversion of apo-IRP1 to c-aconitase is even more challenging because iron sulfur cluster biogenesis is a complex, assisted process involving multiple steps (Sharma et al., 2010). The attending cluster assembly proteins would conceivably need wide access to the active site. Thus the open form of apo-IRP1 seems the best conformation for the initial steps of cluster assembly and insertion. There are few structures of complexes in cluster biogenesis (Shi et al., 2010; Marinoni et al., 2012), and to our knowledge, none that involve IRP1 and the components of the cytosolic assembly system. The sequence of events immediately following cluster insertion in IRP1 can only be speculated: whether there is an open intermediate of assembled c-aconitase after release, or how it closes is not known. It is generally accepted that the holo-forms of iron–sulfur proteins are more stable than their apo-counterparts, so one would expect spontaneous closure for the sake of solvent exclusion and stability. Aconitase is an ancient enzyme (Beinert & Kennedy, 1993). The phylogenetic relationships in the aconitase superfamily argue that IRP1 was an enzyme long before it acquired RNA-repressor functionality (Gruer et al., 1997; Baughn & Malamy, 2002). IRP1 and most other aconitase homologs retained the basic four-domain organization (Gruer et al., 1997; Makarova & Koonin, 2003), but some unusual domain-sorting events occurred during evolution of the superfamily. Odd variations in the placement of domain four—such as a permuted sequence order (e.g., bacterial aconitases B), or a separately coded gene product (e.g., isopropylmalate dehydratases)—support the idea that positional lability of domain four has functional importance. This was recognized early in the structural analyses of aconitases (Lauble & Stout, 1995), and raises the question whether the apo forms of other aconitases (the IRP-aconitase A group, and the closely related m-aconitases) are in open conformations. Since the closest homologs of IRPs (bacterial aconitase-A and plant aconitases) are from organisms not known to have functioning IRP:IRE systems (Piccinelli & Samuelsson, 2007; Arnaud et al., 2007), accessibility for the sake of cluster insertion in IRP1 evolutionarily preceded IRE binding. It has been observed that bacterial aconitases interact with polynucleotides (Tang & Guest, 1999; Alén & Sonenshein, 2002). Using IRP1 as the structural example, high-affinity nucleic acid binding would require domains three and four to be widely separated. Although it is an exciting possibility that these IRP homologs also offer clamshell-like access for purposes of cluster assembly and RNA binding, there is as yet no definitive evidence that their apo forms exist in an open conformation and tightly bind RNA stem-loops. Additional SAXS experiments with other apo-aconitases may provide clearer answers regarding conformational variability in the aconitase superfamily.
345
4. Materials and methods 4.1. Protein purification and sample preparation Hexa-His tagged rabbit IRP1 was over-expressed in Saccharomyces cerevisiae strain BJ5465 (Swenson & Walden, 1994). Cells were grown overnight in YPAD media and then inoculated in SD-URA media to an OD600 of 0.05. Once cells reached logarithmic growth phase they were inoculated to an OD600 of 0.1 in SD media containing 2% raffinose and grown overnight. Protein expression was induced by adding galactose to a final concentration of 2%, and cells were grown for 6–9 h before harvesting. Cells were harvested by centrifugation and washed once with sterile water. All purification steps were performed aerobically. Cells were lysed by manual shaking with 0.04–0.06 mm diameter glass beads, and the lysate was centrifuged at 14,000 g for 1 h at 4 °C. The supernatant was filtered, passed through a column containing Ni-NTA resin (Qiagen), and washed with 500 mM NaCl in 20 mM sodium phosphate, pH 6.3. Protein was eluted from the column with 500 mM NaCl and 300 mM imidazole in 20 mM sodium phosphate, pH 6.3. The eluate was dialyzed overnight against 5 mM NaCl and 100 mM Tris, pH 8.0. Subsequently, the dialysate was loaded on a WPQUAT anion exchange column and eluted with a linear 5–500 mM gradient of NaCl. Fractions containing ≥90% pure IRP1 were pooled and dialyzed overnight against 100 mM NaCl, 1 mM DTT, 100 mM Tris, pH 7.4, and 10% glycerol. The extinction coefficient of IRP1 was calculated from the IRP1 protein sequence using the program ProtParm (Gasteiger et al., 2005). Protein concentration was determined by measuring absorbance at 280 nm and using the extinction coefficient IRP1 of 89,500 M−1·cm −1. Samples were checked for monodispersity using DLS with a Malvern Zetasizer Nano-S. Samples and matched dialysates were stored at −80 °C until needed. 4.2. SAXS experiments All scattering experiments were conducted at the BioCAT beamline 18-ID of the Advanced Photon Source, Argonne National Laboratory (Fischetti et al., 2004). Scattering data were collected at a q range between 0.006 and 0.356 Å−1. Samples were loaded in a quartz capillary flow cell and exposed to 12 keV X-rays for 1 s at 5 second intervals twenty times. Scattering data were measured using a CCD detector positioned 2422 mm from the sample. Radiation damage was monitored by analysis of the scattering data and by SDS-PAGE. Data reduction was performed using Igor-Pro (WaveMetrics Inc., Lake Oswego, OR, USA) equipped with custom SAXS analysis tools designed for the BIO-CAT beamline. 4.3. Ab initio modeling The program GASBOR (Svergun et al., 2001) was used to construct 30 dummy-residue models from the scattering curves of apo-IRP1. The DAMAVER (Volkov & Svergun, 2003) suite of programs was used to select a set of closely corresponding models, superimpose them (Kozin & Svergun, 2001), and generate a final composite model. Criteria for excluding models were done by calculating the average and variance of NSD values among all models. Models with NSD values 2σ greater than the average were rejected. An envelope representation of the final bead model was prepared with the program Situs (Wriggers & Chacón, 2001). 4.4. Rigid-body modeling The IRP1 molecule (PDB ID: 3SNP, molecule A) was divided into 3 parts: a static core (domains 1 and 2), and two flexing domains (domains 3 and domain 4). Rigid-body modeling was performed by two separate methods using the programs RigiMOL (DeLano, 2002) and SASREF (Petoukhov & Svergun, 2005). A total of 11 connectivity
346
O. Shand, K. Volz / Gene 524 (2013) 341–346
restraints were applied to the rigid body system (see Supplemental Tables 1 and 2). These restraints were designed to retain the continuity of the polypeptide backbone and establish connections between inter-domain atom pairs whose relative distances were predicted to not change upon rotation of domains. 4.5. Generation of hypothetical intermediate structures Thirty equally-spaced models between the two functional states of IRP1 were generated by RigiMOL (DeLano, 2002) using the crystal structures of c-aconitase (PDB ID: 2B3X) and RNA-bound IRP1 (PDB ID: 3SNP). Twenty additional models with domains 3 and 4 flexing along their same hinged trajectories but extending beyond RNA-bound IRP1 were added to explore hyper-extended conformations (Supplemental Movie). The scattering curves of the fifty model structures were compared to the experimental scattering data using the program CRYSOL (Svergun et al., 1995) (Fig. 4). 4.6. Restrained ensemble optimization method The ensemble optimization method (EOM) utilizes the fact that the scattering curve of a flexible multi-domain protein may arise from multiple conformations of the protein in solution. The EOM uses a genetic algorithm to determine a subset of conformations representative of the solution scattering data (Bernadó et al., 2007). Attempts were made to generate a larger set of conformers using the EOM software package; however, the majority of conformers generated were not structurally plausible. As described in the previous section, a pool of fifty structures of IRP1 on the trajectory containing the c-aconitase and RNA-bound conformations was analyzed. Although the structures represent only a fraction of total possible apo-IRP1 conformations, the set of 50 conformers cover the entire range of possible Rg values for IRP1. Conflict of interest There is no conflict of interest. Acknowledgments This work was supported by NIH grant GM-71504 to KV. We thank the staff at BioCAT for their assistance in the collection and preliminary processing of the SAXS data. Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.gene.2013.03.112. References Alén, C., Sonenshein, A.L., 2002. Bacillus subtilis aconitase is an RNA-binding protein. Proc. Natl. Acad. Sci. U. S. A. 96, 10412–10417. Anderson, C.P., Shen, M., Eisenstein, R.S., Leibold, E.A., 2012. Mammalian iron metabolism and its control by iron regulatory proteins. Biochim. Biophys. Acta 1823, 1468–1483. Arnaud, N., et al., 2007. The iron-responsive element (IRE)/iron-regulatory protein 1 (IRP1)-cytosolic aconitase iron-regulatory switch does not operate in plants. Biochem. J. 405, 523–531. Baughn, A.D., Malamy, M.H., 2002. A mitochondrial-like aconitase in the bacterium Bacteroides fragilis: implications for the evolution of the mitochondrial Krebs cycle. Proc. Natl. Acad. Sci. U. S. A. 99, 4662–4667. Beinert, H., Kennedy, M.C., 1993. Aconitase: a two-faced protein: enzyme and ironregulatory factor. FASEB J. 7, 1442–1449. Bernadó, P., 2010. Effects of interdomain dynamics on the structure determination of modular proteins by small-angle scattering. Eur. Biophys. J. 39, 769–780. Bernadó, P., Mylonas, E., Petoukhov, M.V., Blackledge, M., Svergun, D.I., 2007. Structural characterization of flexible proteins using small-angle X-ray scattering. J. Am. Chem. Soc. 129, 5656–5664.
Brazzolotto, X., Timmins, P., Dupont, Y., Moulis, J.-M., 2002. Structural changes associated with switching activities of human iron regulatory protein 1. J. Biol. Chem. 277, 11995–12000. DeLano, W.L., 2002. The PyMOL Molecular Graphics System. DeLano Scientific, San Carlos, CA. Dong, H., Qin, S., Zhou, H.-X., 2010. Effects of macromolecular crowding on protein conformational changes. PLoS Comput. Biol. 6, e1000833. Dupuy, J., Volbeda, A., Carpentier, P., Darnault, C., Moulis, J.-M., Fontecilla-Camps, J.C., 2006. Crystal structure of human iron regulatory protein 1 as c-aconitase. Structure 14, 129–139. Fischer, H., de Oliveira Neto, M., Napolitano, H.B., Polikarpov, I., Craievich, A.F., 2010. Determination of the molecular weight of proteins in solution from a single small-angle X-ray scattering measurement on a relative scale. J. Appl. Cryst. 43, 101–109. Fischetti, R., et al., 2004. The BioCAT undulator beamline 18ID: a facility for biological non-crystalline diffraction and X-ray absorption spectroscopy at the advanced photon source. J. Synchrotron Radiat. 11, 399–405. Gasteiger, E., et al., 2005. Protein identification and analysis tools on the ExPASy server. The Proteomics Protols Handbook. 571–607. Gruer, M.J., Artymiuk, P.J., Guest, J.R., 1997. The aconitase family: three structural variations on a common theme. Trends Biochem. Sci. 22, 3–6. Kilburn, D., Roh, J.H., Guo, L., Briber, R.M., Woodson, S.A., 2010. Molecular crowding stabilizes folded RNA structure by the excluded volume effect. J. Am. Chem. Soc. 132, 8690–8696. Kozin, M.B., Svergun, D.I., 2001. Automated matching of high- and low-resolution structural models. J. Appl. Crystallogr. 34, 33–41. Lauble, H., Stout, C.D., 1995. Steric and conformational features of the aconitase mechanism. Proteins 22, 1–11. Makarova, K.S., Koonin, E.V., 2003. Filling a gap in the central metabolism of archaea: prediction of a novel aconitase by comparative-genomic analysis. FEMS Microbiol. Lett. 27, 17–23. Marinoni, E.N., et al., 2012. (IscS-IscU)2 complex structures provide insights into Fe2S2 biogenesis and transfer. Angew. Chem. Int. Ed. 51, 5439–5442. McCallum, S.A., Pardi, A., 2003. Refined solution structure of the iron-responsive element RNA using residual dipolar couplings. J. Mol. Biol. 326, 1037–1050. Petoukhov, M.V., Svergun, D.I., 2005. Global rigid body modeling of macromolecular complexes against small-angle scattering data. Biophys. J. 89, 1237–1250. Piccinelli, P., Samuelsson, T., 2007. Evolution of the iron-responsive element. RNA 13, 952–966. Schalinske, K.L., Anderson, S.A., Tuazon, P.T., Chen, O.S., Kennedy, M.C., Eisenstein, R.S., 1997. The iron–sulfur cluster of iron regulatory protein 1 modulates the accessibility of RNA binding and phosphorylation sites. Biochemistry 36, 3950–3958. Selezneva, A.I., Cavigiolio, G., Theil, E.C., Walden, W.E., Volz, K., 2006. Crystallization and preliminary X-ray diffraction analysis of iron regulatory protein-1 in complex with ferritin IRE RNA. Acta Crystallogr. F62, 249–252. Sharma, A.K., Pallesen, L.J., Spang, R.J., Walden, W.E., 2010. Cytosolic iron–sulfur cluster assembly (CIA) system: factors, mechanism, and relevance to cellular iron regulation. J. Biol. Chem. 285, 26745–26751. Shi, R., et al., 2010. Structural basis for Fe–S cluster assembly and tRNA thiolation mediated by IscS protein–protein interactions. PLoS Biol. 8, e1000354. Showalter, S.A., Baker, N.A., Tang, C., Hall, K.B., 2005. Iron responsive element RNA flexibility described by NMR and isotropic reorientational eigenmode dynamics. J. Biomol. NMR 32, 179–193. Svergun, D.I., Barberato, C., Koch, M.H.J., 1995. CRYSOL: a program to evaluate X-ray solution scattering of biological macromolecules from atomic coordinates. J. Appl. Cryst. 28, 768–773. Svergun, D.I., Petoukhov, M.V., Koch, M.H.J., 2001. Determination of domain structure of proteins from X-ray solution scattering. Biophys. J. 80, 2946–2953. Swenson, G.R., Walden, W.E., 1994. Localization of an RNA binding element of the iron responsive element binding protein within a proteolytic fragment containing iron coordination ligands. Nucleic Acids Res. 22, 2627–2633. Tang, Y., Guest, J.R., 1999. Direct evidence for mRNA binding and post-transcriptional regulation by Escherichia coli aconitases. Microbiology 145, 3069–3079. Tang, Y., Guest, J.R., Artymiuk, P.J., Green, J., 2005. Switching aconitase B between catalytic and regulatory modes involves iron-dependent dimer formation. Mol. Microbiol. 56, 1149–1158. Tsuchiya, D., Shimizu, N., Tomita, M., 2008. Versatile architecture of a bacterial aconitase B and its catalytic performance in the sequential reaction coupled with isocitrate dehydrogenase. Biochim. Biophys. Acta 1784, 1847–1856. Tsuchiya, D., Shimizu, N., Tomita, M., 2009. Cooperativity of two active sites in bacterial homodimeric aconitases. Biochim. Biophys. Res. Comm. 379, 485–488. Volkov, V.V., Svergun, D.I., 2003. Uniqueness of ab initio shape determination in smallangle scattering. J. Appl. Crystallogr. 36, 860–864. Walden, W.E., et al., 2006. Structure of dual function iron regulatory protein 1 complexed with ferritin IRE-RNA. Science 314, 1903–1909. Walden, W.E., Selezneva, A., Volz, K., 2012. Accommodating variety in iron responsive elements: crystal structure of transferrin receptor 1 B IRE bound to iron regulatory protein 1. FEBS Lett. 586, 32–35. Wriggers, W., Chacón, P., 2001. Using Situs for the registration of protein structures with low-resolution bead models from X-ray solution scattering. J. Appl. Crystallogr. 34, 773–776. Yikilmaz, E., Rouault, T.A., Schuck, P., 2005. Self-association and ligand-induced conformational changes of iron regulatory proteins 1 and 2. Biochemistry 44, 8470–8478. Zhou, H.-X., Rivas, G.N., Minton, A.P., 2008. Macromolecular crowding and confinement: biochemical, biophysical, and potential physiological consequences. Annu. Rev. Biophys. 37, 375–397.