Limits...
Structural and dynamic insights into the energetics of activation loop rearrangement in FGFR1 kinase.

Klein T, Vajpai N, Phillips JJ, Davies G, Holdgate GA, Phillips C, Tucker JA, Norman RA, Scott AD, Higazi DR, Lowe D, Thompson GS, Breeze AL - Nat Commun (2015)

Bottom Line: Recent inhibitor-bound structures have unexpectedly revealed FGFR1 for the first time in a 'DFG-out' state.Our detailed structural and biophysical insights identify contributions from altered dynamics in distal elements, including the αH helix, towards the outstanding stability of the DFG-out complex with the inhibitor ponatinib.We conclude that the αC-β4 loop and 'molecular brake' regions together impose a high energy barrier for this conformational rearrangement, and that this may have significance for maintaining autoinhibition in the non-phosphorylated basal state of FGFR1.

View Article: PubMed Central - PubMed

Affiliation: Discovery Sciences, AstraZeneca R&D, Alderley Park, Macclesfield SK10 4TG, UK.

ABSTRACT
Protein tyrosine kinases differ widely in their propensity to undergo rearrangements of the N-terminal Asp-Phe-Gly (DFG) motif of the activation loop, with some, including FGFR1 kinase, appearing refractory to this so-called 'DFG flip'. Recent inhibitor-bound structures have unexpectedly revealed FGFR1 for the first time in a 'DFG-out' state. Here we use conformationally selective inhibitors as chemical probes for interrogation of the structural and dynamic features that appear to govern the DFG flip in FGFR1. Our detailed structural and biophysical insights identify contributions from altered dynamics in distal elements, including the αH helix, towards the outstanding stability of the DFG-out complex with the inhibitor ponatinib. We conclude that the αC-β4 loop and 'molecular brake' regions together impose a high energy barrier for this conformational rearrangement, and that this may have significance for maintaining autoinhibition in the non-phosphorylated basal state of FGFR1.

No MeSH data available.


NMR analysis of structural and dynamic perturbations to FGFR1 kinase onbinding of type I and type II inhibitors.(a) Backbone amide chemical shift perturbation (CSP) analysis onligand binding. Weighted CSPs were calculated asΔδave=(Δδ2(N)/50+Δδ2(H)/2)1/2between unbound and PDA complex (top right panel), and between unbound andponatinib complex (bottom right panel). The CSPs >0.25 for the twocomplexes are mapped on the X-ray crystal structure of unbound FGFR1(PDB-code: 1FGK). Solid bars represent regions of β-strandsecondary structure, open bars regions of α-helical secondarystructure. Selected regions of overlayed1H-15N TROSY-HSQC plots of representativeamino acids in the αC-β4 loop and D735 in the distalαH helix are shown in small panels (left, bottom). The contourplots are colour coded as follows: unbound (black); PDA bound (blue);ponatinib bound (red). Arrows of the corresponding color connect the sameresidue in different spectra. (b) Analysis of chemical exchangecontributions to transverse relaxation rates (R2,ex)measured for ligand-free (top), PDA-bound (middle) and ponatinib-bound(bottom) FGFR1 kinases at static fields of 600 MHz (black),800 MHz (red) and 950 MHz (blue circles), reflectingmotions on time scales >100 μs.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4525181&req=5

f3: NMR analysis of structural and dynamic perturbations to FGFR1 kinase onbinding of type I and type II inhibitors.(a) Backbone amide chemical shift perturbation (CSP) analysis onligand binding. Weighted CSPs were calculated asΔδave=(Δδ2(N)/50+Δδ2(H)/2)1/2between unbound and PDA complex (top right panel), and between unbound andponatinib complex (bottom right panel). The CSPs >0.25 for the twocomplexes are mapped on the X-ray crystal structure of unbound FGFR1(PDB-code: 1FGK). Solid bars represent regions of β-strandsecondary structure, open bars regions of α-helical secondarystructure. Selected regions of overlayed1H-15N TROSY-HSQC plots of representativeamino acids in the αC-β4 loop and D735 in the distalαH helix are shown in small panels (left, bottom). The contourplots are colour coded as follows: unbound (black); PDA bound (blue);ponatinib bound (red). Arrows of the corresponding color connect the sameresidue in different spectra. (b) Analysis of chemical exchangecontributions to transverse relaxation rates (R2,ex)measured for ligand-free (top), PDA-bound (middle) and ponatinib-bound(bottom) FGFR1 kinases at static fields of 600 MHz (black),800 MHz (red) and 950 MHz (blue circles), reflectingmotions on time scales >100 μs.

Mentions: To gain insights into the dynamic origins of slow access to the DFG-out state inFGFR1, we employed both NMR spectroscopy and HDX-MS. We have previously reportedNMR resonance assignments for FGFR1 kinase domain in the ligand-free state22. Titration of either PDA or ponatinib into samples of15N-labelled FGFR1 kinase resulted in amide chemical shiftperturbations (CSPs) in the slow-exchange regime that were completely saturatedat 1:1 molar stoichiometry, typical of high-affinity binding in the nanomolarKD range (Fig. 3a). Unlike for theunbound22 and the PDA-complex states of FGFR1, the first sixresidues (Asp641–Arg646) of the A-loop were observable in the1H-15N TROSY-HSQC spectrum of theFGFR1–ponatinib complex, indicative of altered A-loop dynamics in theponatinib complex compared with the unbound or PDA-bound kinase. Comparison of1H-15N TROSY-HSQC spectra of both PDA andponatinib complexes with unbound FGFR1 showed large amide chemical shift changesfor many residues. Mapping of these perturbations onto the crystal structure ofFGFR1 (PDB-code: 1FGK) shows that most of them are localized in thecatalytically important and structurally conserved regions surrounding theactive site (Fig. 3a). Significant CSPs were observed forAla564 in the hinge region of both complexes, due to direct hydrogen-bondinteractions with a ring nitrogen of the inhibitor; that seen in the ponatinibcomplex is substantially larger and may reflect a stronger hydrogen bond. Forthe PDA complex, CSPs were detected only for residues in the region of theP-loop, the N-terminus of the αC helix, the hinge region residues, andAla640, which are all in close promixity to the inhibitor (Fig.3a, upper right panel). Interestingly, ponatinib binding revealedboth local and distal changes (Fig. 3a, lower rightpanel). Local CSPs were observed in the P-loop, αC helix and hingeregions, and for Ile620 in the catalytic loop all of which participate in directinteractions with the inhibitor. The backbone amide nitrogen of Asp641 (of theDFG motif) also engages in a direct hydrogen-bond interaction with the amidecarbonyl oxygen of ponatinib, which is likely to dominate the observed CSP forthis residue, along with the change in the φ torsion angle associatedwith the DFG flip (Fig. 3a, lower right panel; Fig. 1c). Notably, substantial CSPs were also observed inthe αC-β4 loop around Ile544, and for Asp735 in theαH helix. These are all spatially distant from the active site; thus,the observed chemical shift changes (Fig. 3a, smallpanels) must be a result of structural or dynamic changes propagated through aninteraction network. The CSPs in the αC-β4 loop region are ofparticular interest, since these amides are likely to be highly sensitivereporters on changes in conformation or dynamics associated with movements ofthe αC helix35. By analogy with other kinases, thehydrophobic spine network363738 of FGFR1 is expected to bedisrupted on the reorientation of Phe642 in the inactive DFG-out state, whichmay be reflected in perturbations seen in the chemical shifts of the residuesneighbouring His621 in the catalytic loop. Direct contacts with the terminalmethylpiperazinyl group of ponatinib from residues including Ile620 and His621are also likely to contribute to the observed CSPs. Such perturbations are notseen for the PDA-bound state (which is assumed to populate predominantly theDFG-in conformation in solution). The large chemical shift change we observe forAsp735 in the ponatinib-bound complex is surprising, as Asp735 is situated inhelix αH, which is rather remote from the active site. The upfieldshift of the backbone amide resonance might reflect subtly altered hydrogenbonding and may report on perturbed dynamics in the αH helix asopposed to gross conformational change (vide infra), since the meanstructures from X-ray crystallography are essentially superimposable in thisregion. Further insights into the underlying dynamics of FGFR1 in the threestates were obtained from measurements of contributions from chemical exchangeeffects to the 15N transverse relaxation rates of backboneamides, R2,ex. Using data acquired at three different magneticfield strengths for unbound, PDA-bound and ponatinib-bound FGFR1, we observeparticularly large field-dependent chemical exchange contributions to the15N linewidth (attributable to dynamics on time scales longerthan ∼100 μs) for the ponatinib complex in the P-loop,compared with smaller but still significant effects for unbound FGFR1, and amarked suppression of millisecond time-scale P-loop dynamics in the PDA complex(Fig. 3b); this correlates with the additional P-loopprotein–ligand contacts that we observe in crystal structures ofPDA-bound FGFR1, but also suggests that DFG-out binding of ponatinib isaccompanied by loosening of restraining forces on P-loop conformation. However,in contrast to the enhanced P-loop dynamics, slow time-scale motions aremarkedly suppressed in the αC-helix of the ponatinib complex comparedwith either ligand-free or PDA-bound states. The R2,ex datafurther show the presence of significant slow time-scale motion in theαH helix region of the DFG-out ponatinib complex around Asp735, inagreement with CSP data.


Structural and dynamic insights into the energetics of activation loop rearrangement in FGFR1 kinase.

Klein T, Vajpai N, Phillips JJ, Davies G, Holdgate GA, Phillips C, Tucker JA, Norman RA, Scott AD, Higazi DR, Lowe D, Thompson GS, Breeze AL - Nat Commun (2015)

NMR analysis of structural and dynamic perturbations to FGFR1 kinase onbinding of type I and type II inhibitors.(a) Backbone amide chemical shift perturbation (CSP) analysis onligand binding. Weighted CSPs were calculated asΔδave=(Δδ2(N)/50+Δδ2(H)/2)1/2between unbound and PDA complex (top right panel), and between unbound andponatinib complex (bottom right panel). The CSPs >0.25 for the twocomplexes are mapped on the X-ray crystal structure of unbound FGFR1(PDB-code: 1FGK). Solid bars represent regions of β-strandsecondary structure, open bars regions of α-helical secondarystructure. Selected regions of overlayed1H-15N TROSY-HSQC plots of representativeamino acids in the αC-β4 loop and D735 in the distalαH helix are shown in small panels (left, bottom). The contourplots are colour coded as follows: unbound (black); PDA bound (blue);ponatinib bound (red). Arrows of the corresponding color connect the sameresidue in different spectra. (b) Analysis of chemical exchangecontributions to transverse relaxation rates (R2,ex)measured for ligand-free (top), PDA-bound (middle) and ponatinib-bound(bottom) FGFR1 kinases at static fields of 600 MHz (black),800 MHz (red) and 950 MHz (blue circles), reflectingmotions on time scales >100 μs.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4525181&req=5

f3: NMR analysis of structural and dynamic perturbations to FGFR1 kinase onbinding of type I and type II inhibitors.(a) Backbone amide chemical shift perturbation (CSP) analysis onligand binding. Weighted CSPs were calculated asΔδave=(Δδ2(N)/50+Δδ2(H)/2)1/2between unbound and PDA complex (top right panel), and between unbound andponatinib complex (bottom right panel). The CSPs >0.25 for the twocomplexes are mapped on the X-ray crystal structure of unbound FGFR1(PDB-code: 1FGK). Solid bars represent regions of β-strandsecondary structure, open bars regions of α-helical secondarystructure. Selected regions of overlayed1H-15N TROSY-HSQC plots of representativeamino acids in the αC-β4 loop and D735 in the distalαH helix are shown in small panels (left, bottom). The contourplots are colour coded as follows: unbound (black); PDA bound (blue);ponatinib bound (red). Arrows of the corresponding color connect the sameresidue in different spectra. (b) Analysis of chemical exchangecontributions to transverse relaxation rates (R2,ex)measured for ligand-free (top), PDA-bound (middle) and ponatinib-bound(bottom) FGFR1 kinases at static fields of 600 MHz (black),800 MHz (red) and 950 MHz (blue circles), reflectingmotions on time scales >100 μs.
Mentions: To gain insights into the dynamic origins of slow access to the DFG-out state inFGFR1, we employed both NMR spectroscopy and HDX-MS. We have previously reportedNMR resonance assignments for FGFR1 kinase domain in the ligand-free state22. Titration of either PDA or ponatinib into samples of15N-labelled FGFR1 kinase resulted in amide chemical shiftperturbations (CSPs) in the slow-exchange regime that were completely saturatedat 1:1 molar stoichiometry, typical of high-affinity binding in the nanomolarKD range (Fig. 3a). Unlike for theunbound22 and the PDA-complex states of FGFR1, the first sixresidues (Asp641–Arg646) of the A-loop were observable in the1H-15N TROSY-HSQC spectrum of theFGFR1–ponatinib complex, indicative of altered A-loop dynamics in theponatinib complex compared with the unbound or PDA-bound kinase. Comparison of1H-15N TROSY-HSQC spectra of both PDA andponatinib complexes with unbound FGFR1 showed large amide chemical shift changesfor many residues. Mapping of these perturbations onto the crystal structure ofFGFR1 (PDB-code: 1FGK) shows that most of them are localized in thecatalytically important and structurally conserved regions surrounding theactive site (Fig. 3a). Significant CSPs were observed forAla564 in the hinge region of both complexes, due to direct hydrogen-bondinteractions with a ring nitrogen of the inhibitor; that seen in the ponatinibcomplex is substantially larger and may reflect a stronger hydrogen bond. Forthe PDA complex, CSPs were detected only for residues in the region of theP-loop, the N-terminus of the αC helix, the hinge region residues, andAla640, which are all in close promixity to the inhibitor (Fig.3a, upper right panel). Interestingly, ponatinib binding revealedboth local and distal changes (Fig. 3a, lower rightpanel). Local CSPs were observed in the P-loop, αC helix and hingeregions, and for Ile620 in the catalytic loop all of which participate in directinteractions with the inhibitor. The backbone amide nitrogen of Asp641 (of theDFG motif) also engages in a direct hydrogen-bond interaction with the amidecarbonyl oxygen of ponatinib, which is likely to dominate the observed CSP forthis residue, along with the change in the φ torsion angle associatedwith the DFG flip (Fig. 3a, lower right panel; Fig. 1c). Notably, substantial CSPs were also observed inthe αC-β4 loop around Ile544, and for Asp735 in theαH helix. These are all spatially distant from the active site; thus,the observed chemical shift changes (Fig. 3a, smallpanels) must be a result of structural or dynamic changes propagated through aninteraction network. The CSPs in the αC-β4 loop region are ofparticular interest, since these amides are likely to be highly sensitivereporters on changes in conformation or dynamics associated with movements ofthe αC helix35. By analogy with other kinases, thehydrophobic spine network363738 of FGFR1 is expected to bedisrupted on the reorientation of Phe642 in the inactive DFG-out state, whichmay be reflected in perturbations seen in the chemical shifts of the residuesneighbouring His621 in the catalytic loop. Direct contacts with the terminalmethylpiperazinyl group of ponatinib from residues including Ile620 and His621are also likely to contribute to the observed CSPs. Such perturbations are notseen for the PDA-bound state (which is assumed to populate predominantly theDFG-in conformation in solution). The large chemical shift change we observe forAsp735 in the ponatinib-bound complex is surprising, as Asp735 is situated inhelix αH, which is rather remote from the active site. The upfieldshift of the backbone amide resonance might reflect subtly altered hydrogenbonding and may report on perturbed dynamics in the αH helix asopposed to gross conformational change (vide infra), since the meanstructures from X-ray crystallography are essentially superimposable in thisregion. Further insights into the underlying dynamics of FGFR1 in the threestates were obtained from measurements of contributions from chemical exchangeeffects to the 15N transverse relaxation rates of backboneamides, R2,ex. Using data acquired at three different magneticfield strengths for unbound, PDA-bound and ponatinib-bound FGFR1, we observeparticularly large field-dependent chemical exchange contributions to the15N linewidth (attributable to dynamics on time scales longerthan ∼100 μs) for the ponatinib complex in the P-loop,compared with smaller but still significant effects for unbound FGFR1, and amarked suppression of millisecond time-scale P-loop dynamics in the PDA complex(Fig. 3b); this correlates with the additional P-loopprotein–ligand contacts that we observe in crystal structures ofPDA-bound FGFR1, but also suggests that DFG-out binding of ponatinib isaccompanied by loosening of restraining forces on P-loop conformation. However,in contrast to the enhanced P-loop dynamics, slow time-scale motions aremarkedly suppressed in the αC-helix of the ponatinib complex comparedwith either ligand-free or PDA-bound states. The R2,ex datafurther show the presence of significant slow time-scale motion in theαH helix region of the DFG-out ponatinib complex around Asp735, inagreement with CSP data.

Bottom Line: Recent inhibitor-bound structures have unexpectedly revealed FGFR1 for the first time in a 'DFG-out' state.Our detailed structural and biophysical insights identify contributions from altered dynamics in distal elements, including the αH helix, towards the outstanding stability of the DFG-out complex with the inhibitor ponatinib.We conclude that the αC-β4 loop and 'molecular brake' regions together impose a high energy barrier for this conformational rearrangement, and that this may have significance for maintaining autoinhibition in the non-phosphorylated basal state of FGFR1.

View Article: PubMed Central - PubMed

Affiliation: Discovery Sciences, AstraZeneca R&D, Alderley Park, Macclesfield SK10 4TG, UK.

ABSTRACT
Protein tyrosine kinases differ widely in their propensity to undergo rearrangements of the N-terminal Asp-Phe-Gly (DFG) motif of the activation loop, with some, including FGFR1 kinase, appearing refractory to this so-called 'DFG flip'. Recent inhibitor-bound structures have unexpectedly revealed FGFR1 for the first time in a 'DFG-out' state. Here we use conformationally selective inhibitors as chemical probes for interrogation of the structural and dynamic features that appear to govern the DFG flip in FGFR1. Our detailed structural and biophysical insights identify contributions from altered dynamics in distal elements, including the αH helix, towards the outstanding stability of the DFG-out complex with the inhibitor ponatinib. We conclude that the αC-β4 loop and 'molecular brake' regions together impose a high energy barrier for this conformational rearrangement, and that this may have significance for maintaining autoinhibition in the non-phosphorylated basal state of FGFR1.

No MeSH data available.