Conformational Modulation of a Mobile Loop Controls Catalysis in the (βα)8-Barrel Enzyme of Histidine Biosynthesis HisF

The overall significance of loop motions for enzymatic activity is generally accepted. However, it has largely remained unclear whether and how such motions can control different steps of catalysis. We have studied this problem on the example of the mobile active site β1α1-loop (loop1) of the (βα)8-barrel enzyme HisF, which is the cyclase subunit of imidazole glycerol phosphate synthase. Loop1 variants containing single mutations of conserved amino acids showed drastically reduced rates for the turnover of the substrates N′-[(5′-phosphoribulosyl) formimino]-5-aminoimidazole-4-carboxamide ribonucleotide (PrFAR) and ammonia to the products imidazole glycerol phosphate (ImGP) and 5-aminoimidazole-4-carboxamide-ribotide (AICAR). A comprehensive mechanistic analysis including stopped-flow kinetics, X-ray crystallography, NMR spectroscopy, and molecular dynamics simulations detected three conformations of loop1 (open, detached, closed) whose populations differed between wild-type HisF and functionally affected loop1 variants. Transient stopped-flow kinetic experiments demonstrated that wt-HisF binds PrFAR by an induced-fit mechanism whereas catalytically impaired loop1 variants bind PrFAR by a simple two-state mechanism. Our findings suggest that PrFAR-induced formation of the closed conformation of loop1 brings active site residues in a productive orientation for chemical turnover, which we show to be the rate-limiting step of HisF catalysis. After the cyclase reaction, the closed loop conformation is destabilized, which favors the formation of detached and open conformations and hence facilitates the release of the products ImGP and AICAR. Our data demonstrate how different conformations of active site loops contribute to different catalytic steps, a finding that is presumably of broad relevance for the reaction mechanisms of (βα)8-barrel enzymes and beyond.

(A) Structural comparison between the loop1-closed conformation observed in the HisF/HisH complex (grey, PDB ID:7AC8 2 , chain E) in its ProFAR-bound state and the closed conformation obtained from the simulations of HisF in its PrFAR-bound state (tan).As can be seen from these structures, while we obtain loop1 conformations that have structural similarity to the closed conformation observed in the HisF/HisH complex, the loop is displaced compared to the conformation in the crystal structure.The bottom panels (B-C) show the results of interaction analysis performed using Key Interaction Networks 3 (KIN) on the (B) starting X-ray structure of HisF taken from the HisF/HisH complex, and (C) on closed conformations of loop1 from simulations of HisF in its PrFAR-bound states.The stick colors correspond to the different types of non-covalent interactions identified by KIN, where yellow represents salt bridges, red represents hydrophobic interactions, and blue represents H-bonding interactions.Interactions between loop1 and PrFAR were calculated using NCIplot 4 and are represented by the green surface of the electronic density.S3.S5.Table S3: Rate constants for the binding of PrFAR to HisF-CouA at 25°C derived from analysis of secondary plots.

S2. SUPPLEMENTAL TABLES
wt F38A KD1 6.6 ± 1.1 µM The values ± SE for constants KD1, kconf, k-conf were determined by fitting the observed rate constants in stopped-flow measurements (Figure S7E) to the hyperbolic equation , the values ± SE for the rate constants k1 and k-1 were determined by fitting the observed rate constants (Figure S7F) to linear equations 1 KD values for induced fit models were calculated using the equation KD = KD1/(1 + kconf/k-conf) 5 , SE for KD values were calculated according to the Gaussian law of error propagation. 2KD values for two-state binding models were calculated using the equation KD = k-1/k1, SE for KD values were calculated according to the Gaussian law of error propagation.The stopped-flow measurements from which the turnover rates were derived are shown in    PrFAR@O2 G82@N 2.0 3.0 5.0 10.0 0.0 10.0 PrFAR@O3 T104@N 2.0 3.0 5.0 10.0 0.0 10.0 PrFAR@C4 S101@OG 2.0 3.0 5.0 10.0 0.0 10.0 PrFAR@O9 D11@OD1 2.0 3.0 5.0 10.0 0.0 10.0 PrFAR@O11 S201@OG 2.0 3.0 5.0 10.0 0.0 10.0 PrFAR@O14 S225@H 2.0 3.0 5.0 10.0 0.0 10.0 PrFAR@O15 G203@N 2.0 3.0 5.0 10.0 0.0 10.0 a r 1 , r 2 , r 3 and r 4 values indicate the four positions used to define the distance restraints between the two atoms.rk 2 and rk 3 define the force constants for the left and right parabolas, respectively.Note that distance restraints are wall restraints that kick when the two atoms are 5.0 Å or greater apart.No restraints were placed between the PrFAR substrate and any loop1 residue.a r 1 , r 2 , r 3 and r 4 values indicate the four positions used to define the distance restraints between the two atoms.rk 2 and rk 3 define the force constants for the left and right parabolas, respectively.Note that distance restraints are harmonic restraints that kick when the two atoms are closer than 2.0 Å or greater than 6.0 Å.

Table S9. Distance restraints employed during all MD simulations to avoid interactions between neutralizing counterions and the
Figure S1: Crystal structures of HisF-G20P and HisF-F23A in comparison with wt-HisF.Alignment of the crystal structure of wt-HisF (PDB-code 1VH7 1 , red) with the structures of (A) HisF-G20P (PDB-code 8S8R, grey) and (B) HisF-F23A (PDB-code 8S8S, grey).The catalytic residues D11 and D130 are shown as blue sticks.Introduction of P20 leads to a slight distortion of the loop conformation.In the structure of variant HisF-F23A electron density for residues 20-24 (loop1) and residues 53-54 (loop2) is missing (indicated by dashed orange lines).

Figure
Figure S2: N-H S 2 order parameters computed from MD simulations of unliganded HisF variants.(A, B) Peaks in the calculated N-H S 2 order parameters of the wt-HisF are illustrated on the open crystal structure.The loop1 is highlighted in orange.The wt-HisF order parameters were compared with order parameters computed for (C) F38A, (D) F23A and (E) G20P.The loop1 S 2 value is the lowest in HisF-F23A, followed by HisF-F38A, wt-HisF and HisF-G20P, and agrees with the loop flexibility trend observed in NMR experiments.

Figure S3 :
Figure S3: Root mean square fluctuations (RMSF, Å) of the protein Cα-atoms during molecular dynamics simulations of unliganded and PrFAR-bound HisF.Shown in each panel are the absolute Cα-atom RMSF during simulations of wt-HisF (black), as well as the relative RMSF (ΔRMSF) between wt-HisF and each of the HisF-F38A (orange), HisF-F23A (purple), and HisF-G20P (green) variants.Shown here is data from simulations initiated from the (A) open unliganded, (B) closed PrFAR-bound, and (C) open PrFAR-bound states of each enzyme.Data were collected every 10 ps from 5 individual replicas of 1 μs length each.

Figure S4 :
Figure S4: Representative structures and interactions for the different closed conformations of wt-HisF from the X-ray structure and from simulations.(A)Structural comparison between the loop1-closed conformation observed in the HisF/HisH complex (grey, PDB ID:7AC8 2 , chain E) in its ProFAR-bound state and the closed conformation obtained from the simulations of HisF in its PrFAR-bound state (tan).As can be seen from these structures, while we obtain loop1 conformations that have structural similarity to the closed conformation observed in the HisF/HisH complex, the loop is displaced compared to the conformation in the crystal structure.The bottom panels (B-C) show the results of interaction analysis performed using Key Interaction Networks 3 (KIN) on the (B) starting X-ray structure of HisF taken from the HisF/HisH complex, and (C) on closed conformations of loop1 from simulations of HisF in its PrFAR-bound states.The stick colors correspond to the different types of non-covalent interactions identified by KIN, where yellow represents salt bridges, red represents hydrophobic interactions, and blue represents H-bonding interactions.Interactions between loop1 and PrFAR were calculated using NCIplot 4 and are represented by the green surface of the electronic density.

Figure S5 :
Figure S5: Comparison of Ramachandran plots of positions 19 and 20 in wt-HisF and HisF-G20P.Shown here are dihedral angles observed in residues K19 and G20/P20 of wt-HisF and the HisF-G20P variant overlayed over the Ramachandran plots that corresponds to the identity of the residue.Angles from the closed state simulations are plotted in black, while possible angles according to Ramachandran plot are shown in blue.Dihedral angles of the residues G20 and K19 for the closed crystal structure and a representative closed conformation are indicated by stars and green circles, respectively.This figure compares angles observed in (A) G20 of wt-HisF overlayed on glycine Ramachandran plot, (B) K19 in wt-HisF overlayed on general case Ramachandran plot, (C) P20 of HisF-G20P on the proline Ramachandran plot and (D) K19 in HisF-G20P on the pre-proline Ramachandran plot.

Figure S6 :
Figure S6: Spectral properties of wt-HisF-K132CouA.(A) SDS-PAGE gel after Coomassie staining and illumination with white light (left) and UV light (333 nm) (right).Lane 1: Molecular weight standard, lane 2: non-labelled wt-HisF, lane 3: wt-HisF-K132CouA.Under UV light the typical blue fluorescence of CouA becomes visible.(B) The absorbance spectrum of HisF-K132CouA (solid line) encompasses peaks at 330 and 370 nm, characteristic for CouA, in addition to the protein absorption maximum at 280 nm.The fluorescence emission spectrum (dashed line) shows a maximum at 450 nm, inherent to CouA, after excitation at 370 nm.Intensities of the spectra have been normalized to the maximal signal.

Figure S7 :
Figure S7: Kinetics of PrFAR binding to HisF-CouA monitored by stopped-flow measurements.(A) Time traces (coloured lines) recorded after mixing 0.05 µM wt-HisF with excess PrFAR at 25°C (Ex.367 nm, Em cut-off: 400 nm).Time traces were fit to double exponential equations (dotted black lines).(B) Time traces recorded after mixing 0.1 µM HisF-F38A with excess PrFAR.Time traces were fit to double exponential equations (dotted black lines) (C) Time traces recorded after mixing 0.05 µM HisF-F23A with excess PrFAR.Time traces were fit to single exponential equations (dotted black lines).(D) Time traces recorded after mixing 0.1 µM HisF-G20P with excess PrFAR.Time traces were fit to single exponential equations (dotted black lines).(E) kobs1-values associated with binding of PrFAR to wt-HisF and variant HisF-F38A were plotted vs. PrFAR concentration.The resulting plots show hyperbolic curves and were fitted to the equation kobs1 = k-conf + kconf *[PrFAR]/KD1 + [PrFAR]) according to an induced fit model.(F) kobs1-values associated with binding of PrFAR to variants HisF-F23A and HisF-G20P were plotted vs. PrFAR concentration.The resulting plots show linear dependencies and were fitted to linear equations (kobs = k-1 + k1 * [PrFAR]) according to a simple binding model.Numerical values obtained by curve fitting are listed in TableS3.

Figure S8 :
Figure S8: Kinetics of ImGP and AICAR binding to HisF-CouA monitored by stopped-flow measurements.(A) Time traces were recorded at 25°C after mixing equal volumes of wt-HisF and buffer (black), AICAR (red), ImGP (blue) or AICAR/ImGP (green).In the same way, the pre-formed binary complexes wt-HisF *ImGP (orange) and wt-HisF*AICAR (cyan) were mixed with the respective second reaction product AICAR or ImGP and the corresponding time curves were recorded.The formation of the binary complexes is completed within the dead-time of the stopped-flow instrument, while formation of the ternary complex HisFclosed*AICAR*ImGP is associated with a decrease in fluorescence with an observed rate constant kobs in the range of 50 s -1 .(B) Time traces obtained after mixing HisF-F23A with AICAR and ImGP.The formation of both, the binary and ternary complexes is completed within the instrument dead time.Specified concentrations refer to final concentrations in the observation cell.(C) Kinetic model for the interaction of the reaction products ImGP and AICAR with wt-HisF and HisF-F38A including a conformational change upon formation of the ternary complex.

Figure S9 :
Figure S9: Analysis of the binding reaction of AICAR and ImGP to wt-HisF-CouA.Kinetic traces monitoring the fluorescence changes at 25°C associated with binding of ImGP/AICAR to HisF are shown (Ex.367 nm, Em cut-off: 420 nm).A dataset of 16 time traces was recorded by mixing limiting concentrations of HisF or the binary complexes (HisF*ImGP and HisF*AICAR) with an excess of ligand.A kinetic model corresponding to the scheme in Figure S8C was fit to the stopped-flow transients, dashed lines represent the best fit after optimization of parameters in global fitting procedures.The fit parameters are summarized in Table S4.Concentrations denote final concentrations in the observation cell.(A) Time traces obtained after mixing HisF with AICAR.(B) Time traces obtained after mixing HisF with ImGP.(C) Time traces obtained after mixing the preformed HisF*ImGP complex with AICAR.(D) Time traces obtained after mixing the preformed HisF*AICAR complex with ImGP.(E) Time traces obtained after mixing HisF with combinations of AICAR/ImGP.

Figure S10 :
Figure S10: Analysis of the binding reaction of AICAR and ImGP to HisF-F38A-CouA.Kinetic traces monitoring the fluorescence changes at 25°C associated with binding of ImGP/AICAR to HisF-F38A are shown (Ex.367 nm, Em cut-off: 420 nm).A dataset of 16 time traces was recorded by mixing limiting concentrations of HisF-F38A or the binary complexes (HisF-F38A*ImGP and HisF-F38A*AICAR) with an excess of ligand.A kinetic model corresponding to the scheme in Figure S8C was fit to the stopped-flow transients, dashed lines represent the best fit after optimization of parameters in global fitting procedures.The fit parameters are summarized in Table S4.Concentrations denote final concentrations in the observation cell.(A) Time traces obtained after mixing HisF-F38A with AICAR.(B) Time traces obtained after mixing HisF-F38A with ImGP.(C) Time traces obtained after mixing the preformed HisF-F38A*ImGP complex with AICAR.(D) Time traces obtained after mixing the preformed HisF-F38A*AICAR complex with ImGP.(E) Time traces obtained after mixing HisF-F38A with combinations of AICAR/ImGP.

Figure S11 .
Figure S11.Multiple-and single turnover kinetics of HisF-F38A.(A) A representative transient is shown monitoring PrFAR turnover in multiple turnover mode at 25°C after mixing 0.1 µM HisF-F38A with 10.0 µM PrFAR in presence of ammonium acetate (turnover curve, blue line).A linear approximation of the steady-state phase (dashed line) yielded a turnover velocity of v = 0.188 µM s -1 .The control curve (light blue line) shows the progress of the reaction in absence of ammonium acetate.(B) Plot of the turnover velocity v vs. the respective PrFAR concentration in multiple turnover experiments.The data were fit to the Michaelis-Menten equation.(C) A representative transient is shown monitoring PrFAR turnover in single turnover mode after mixing an excess of HisF-F38A (20 µM) with 10 µM PrFAR in the presence of 100 mM ammonium acetate (turnover curve, red line).The turnover curve was fit with a single exponential decay function (dashed line,  =  *  −  *  + ).The control curve (orange line) shows the progress of the reaction in the absence of ammonium acetate.(D) Plot of the turnover rates, kobs, observed under single turnover conditions, vs. the respective PrFAR concentration.kcat-, KM-and kobs-values are summarized in TableS5.

Figure S12 :
Figure S12: Multiple-and single turnover kinetics of HisF-F23A.(A) A representative transient is shown monitoring PrFAR turnover in multiple turnover mode at 25°C after mixing 0.5 µM HisF-F23A with 10.0 µM PrFAR in presence of ammonium acetate (turnover curve, blue line).A linear approximation of the steady-state phase (20-60 s, dashed line) yielded a turnover velocity of v = 1.34 x 10 -3 µM s -1 .The control curve (light blue line) shows the progress of the reaction in absence of ammonium acetate.(B) Plot of the turnover velocity v vs. the respective PrFAR concentration in multiple turnover experiments.The data were fit to the Michaelis-Menten equation.(C) A representative transient is shown monitoring PrFAR turnover in single turnover mode after mixing an excess of HisF-F23A (20 µM) with 10 µM PrFAR in the presence of ammonium acetate (turnover curve, red line).The turnover curve was fit with a single exponential decay function (dashed line,  =  *  −  *  + ).The control curve (orange line) shows the progress of the reaction in the absence of ammonium acetate.(D) Plot of the turnover rates, kobs, observed under single turnover conditions, vs. the respective PrFAR concentration.kcat-, KM-and kobs-values are summarized in TableS5.

Figure S13 :
Figure S13: Multiple-and single turnover kinetics of HisF-G20P.(A) A representative transient is shown monitoring PrFAR turnover in multiple turnover mode at 25°C after mixing 1.5 µM HisF-G20P with 25.0 µM PrFAR in the presence of 100 mM ammonium acetate (turnover curve, blue line).A linear approximation of the steady-state phase (dashed line) yielded a turnover velocity of v = 2.1 x 10 -3 µM s -1 .The control curve (light blue line) shows the progress of the reaction in absence of ammonium acetate.(B) Plot of the turnover velocity v vs. the respective PrFAR concentration in multiple turnover experiments.The data were fit to the Michaelis-Menten equation.(C) A representative transient is shown monitoring PrFAR turnover in single turnover mode at 25°C after mixing an excess of HisF-G20P (20 µM) with 10 µM PrFAR in the presence of 100 mM ammonium acetate (turnover curve, red line).The turnover curve was fit with a single exponential decay function (dashed line,  =  *  −  *  + ).The control curve (orange line) shows the progress of the reaction in absence of ammonium acetate.(D) Plot of the turnover rates, kobs, observed under single turnover conditions, vs. the respective PrFAR concentration.kcat-, KM-and kobs-values are summarized in TableS5.

Figure S14 :
Figure S14: Assignment of the F23 signal in the ProFAR-bound state of wt-HisF.(Left) Overlay of 1 H 15 N-TROSY spectra of fully 15 N-labeled HisF in the presence of ProFAR (black) and of 15 N-labeled HisF, where the Phe and Tyr residues are unlabelled, in the presence of ProFAR (red).The signal of F23 is indicated.(Right) Overlay of the 1 H 15 N-TROSY spectra of fully 15 N-labeled HisF in the presence of ProFAR (black) and the 1 H 15 N-plane of an HNCO spectrum of 13 C, 15 N-labeled HisF, where the Asn are unlabelled, in the presence of ProFAR (red).The signal of F23 is indicated.It is missing from the HNCO spectrum due to the unlabelling of N22.As F23 is the only Phe or Tyr residue that succeeds an Asn residue the lack of this signal in both spectra of the selectively unlabelled HisF variants allows for an unambiguous assignment to F23.

Figure S15 :
Figure S15: Root mean square deviations (RMSD, Å) of the Cα-atoms during MD simulations of unliganded HisF, initiated from the open conformation of loop1.(A) wt-HisF, (B) HisF-F38A, (C) HisF-F23A and (D) HisF-G20P.Data was collected every 10 ps from 5 replicas of 1 s length each.The grey lines show the 5 individual runs, whilst the color solid line shows a rolling average RMSD from all 5 replicas for each system.

Figure S16 :
Figure S16: Root mean square deviations (RMSD, Å) of the Cα-atoms during MD simulations of PrFAR-bound HisF, initiated from the open conformation of loop1.(A) wt-HisF, (B) HisF-F38A, (C) HisF-F23A and (D) HisF-G20P.Data was collected every 10 ps from 5 replicas of 1 s length each.The grey lines show the 5 individual runs, whilst the color solid line shows a rolling average RMSD from all 5 replicas for each system.

Figure S17 :
Figure S17: Root mean square deviations (RMSD, Å) of the Cα-atoms during MD simulations of PrFAR-bound HisF, initiated from the closed conformation of loop1.(A) wt-HisF, (B) HisF-F38A, (C) HisF-F23A and (D) HisF-G20P.Data was collected every 10 ps from 5 replicas of 1 s length each.The grey lines show the 5 individual runs, whilst the color solid line shows a rolling average RMSD from all 5 replicas for each system.

Table S4 . Rate constants for the release of ImGP and AICAR from wt-HisF-CouA and HisF-F38A-CouA derived from global fitting analysis.
KD values were calculated from the rate constants obtained in the global fitting analysis using the equation KDx = k-x/kx, SE for KD values were calculated according to the Gaussian law of error propagation.

Table S7 . Non-standard force field parameters used to describe the substrate PrFAR in the molecular dynamics simulations. a
a All parameters were obtained using the General AMBER Force Field 2 (GAFF2) as described in the Supplemental Methods section.