Vous êtes sur la page 1sur 21

Domain Movements upon Activation of

Phenylalanine Hydroxylase Characterized by


Crystallography and Chromatography-Coupled
Small-Angle X-ray Scattering
Steve P. Meisburger, Alexander B. Taylor, Crystal A. Khan, Shengnan Zhang,
Paul F. Fitzpatrick,, and Nozomi Ando,
Department of Chemistry, Princeton University, Princeton, New Jersey 08544, United
States.
Department of Biochemistry, University of Texas Health Science Center, San Antonio,
Texas 78229, United States.
To whom correspondence should be addressed.
Phone: 210-567-8264 E-mail: fitzpatrickp@uthscsa.edu
Phone: 609-258-6513 E-mail: nozomi.ando@princeton.edu

Supplementary Table S1

Supplementary Figures S1-S18

Supplementary References

S1

Table S1: Crystallographic data collection and refinement statistics.


Data collection
PDB Code
Space group
Cell dimensions
a, b, c ()
, , ()
Wavelength ()
Resolution ()*
Rsym
Rpim
CC(1/2)
Mean (I/I)
Completeness (%)
Redundancy
2
Wilson B-factor ( )
Refinement
Resolution ()
No. of reflections
Rwork /Rfree
Chain
Residues Modeled

PheH24 R270K

wt-PheH

5EGQ
P21 21 21

5FGJ
P21 21 21

90.1, 96.5, 202.6


90, 90, 90
0.9791
47.16-2.50 (2.64-2.50)
0.096 (0.623)
0.067 (0.424)
0.77
10.1 (2.0)
99.0 (99.7)
3.7 (3.8)
38.6

96.8, 102.7, 203.0


90, 90, 90
0.9791
102.69-3.60 (3.79-3.60)
0.388 (1.287)
0.144 (0.472)
0.59
6.5 (2.2)
100 (100)
8.2 (8.3)
76.0

47.16-2.5
61,113
0.211 / 0.256
A
B
32-136; 32-135;
143-450 143-449

101.48-3.6
24,126
0.269 / 0.296
A
B
20-136; 21-136;
140-450 140-449

C
D
35-83; 32-136;
87-136; 143-449
143-449

No. of atoms
Protein
13,332
Ligands
30 (5 SO2
4 )
Solvent
176
B -factors (2 )
Protein
43.6
Ligands
56.9
Solvent
40.6
R.m.s. deviations
Bond lengths ()
0.007
Bond angles ()
1.164
MolProbity statistics
Ramachandran plot
94.89, 4.86, 0.25
favored, allowed, outliers (%)
All-atom clash score
8.99
Rotamer outliers (%)
2.48
C deviations
5
*Values in parentheses are for the highest-resolution shell.

S2

13,807
5 (4 Fe3+ , 1 Mg2+ )
0
76.8
65.6
0.002
0.566
95.90, 3.86, 0.24
5.82
0.07
0

C
20-135;
143-450

D
22-136;
143-449

Figure S1: Asymmetry in tetrameric assemblies of aromatic amino acid hydroxylases. (A) The
previous structure of human PheH lacking the regulatory domains (2PAH) forms a tetramer via twofold symmetry about the axis shown. 1 (B) The crystal structures of PheH determined in this study
display a slight V-shaped asymmetry between the two dimers (blue/yellow dimer and orange/green
dimer), with an angle estimated by rotating residues 1-420 of chain a to coincide with those of chain
c. In addition, the central coiled coil formed by the tetramerization domains is rotated 15 . The
asymmetries can be attributed to a hinge-like motion within the tetramerization domain, visualized
on the right by aligning the alpha helixes of each chain (residues 420-449). (C) The crystal structure
of tetrameric tyrosine hydroxylase lacking the N-terminal domains (1TOH) has one monomer in the
asymmetric unit, and the tetrameric assembly is generated by the crystallographic two-fold axes. 2
S3

Figure S2: Alignment of truncated N-terminal PheH structure with those determined in this study.
(A) The crystal structure of truncated rat PheH (PheHN-term , residues 1-429, PDB 2PHM) 3 . The
asymmetric unit contains a single chain, shown in a cartoon representation with the regulatory
domain in red (residues 1-117), the catalytic domain in blue (residues 118-410), and the truncated
tetramerization domain in yellow (residues 411-429). The C-terminal alpha helix, missing in this
structure, is required for tetrameric assembly. (B) The orientation of the regulatory and catalytic
domains in the PheHN-term structure is also observed in the structures determined in this study. The
C atoms of chain a in wt-PheH (gray ribbon) were aligned with those of PheHN-term (ribbon colored
as in (A)) using PyMol (The PyMOL Molecular Graphics System, Version 1.7.2 Schrdinger, LLC.).
The root mean square deviation (RMSD) for the C atoms of residues 20-424 is 0.616 . Chains
b-d show similar alignments with C RMSDs of 0.617, 0.612, and 0.570 , respectively. Chains a-d
of PheH24 R270K align with C RMSDs of 0.582, 0.619, 0.770, and 0.559 . Although one chain
was present in the asymmetric unit of PheHN-term , the physiologically-relevent dimer is generated by
crystallographic symmetry (not shown) 3 . When the crystallographic dimer of PheHN-term is aligned
with the corresponding dimers in wt-PheH we obtain C RMSDs of 0.632 (a/d dimer) and 0.825
(b/c dimer). Similarly, dimers in PheH24 R270K align with C RMSDs of 0.768 (a/d dimer)
and 0.748 (b/c dimer).

S4

Figure S3: Comparison of ACT domain interactions in PheH and in prephenate dehydratase (PDT).
(A) The ACT domains of PDT from Chlorobium tepidum (residues 196-302) form a side-by-side homodimer that binds two molecules of L-Phe at the dimer interface 4 , similar to the interactions made
by the isolated ACT domains of PheH in the presence of L-Phe 5 . (B) In the inactive conformation
of tetrameric PheH, the ACT domain in each monomer (yellow, residues 34-111) interacts with
the catalytic domain of the adjacent monomer (blue). The interaction between the ACT domain
and catalytic domain involves conserved residues 6 in 2, 1, and adjacent loops that, in the PDT
structure, lie at the interface of the ACT domain dimer and form the binding pocket for L-Phe.
In PheH, dimerization of the ACT domains leads to the extension of the 2 strand to form an
antiparallel sheet across the dimer interface 4,5 .

S5

Figure S4: Structural consequences of the R270K mutation in PheH in the absence of bound LPhe. (A) The C atoms of wt-PheH (light blue) and PheH24 R270K (green) tetramers align with an
RMSD of 1.1 . (B) Stereo view of the active site residues in wt-PheH (light blue) and PheH24 R270K
(green), aligned using PyMol. The PKU-associated R270K mutation has previously been shown
weaken the binding of L-Phe in the active site by over two orders of magnitude; this was attributed
to the loss of the interaction of Arg270 with the substrate carboxylate moiety. 7 Close inspection of
the two structures shows that the effect of this mutation is localized to the loop containing Thr278
near the L-Phe binding site in the active site. (C) The substitution of Arg270 with Lys270 not only
eliminates ionic interactions that support binding of L-Phe in the active site, but also introduces
a new hydrogen bond between the Lys270 NZ and the main chain carbonyl O of Tyr277, which in
turn appears to favor an alternate conformation of the Thr278 loop. In this alternate conformation,
the Thr278 side chain flips towards the L-Phe binding site. Although the carbonyl O of Thr278
remains within hydrogen-bonding distance of the L-Phe amide N in the PheH24 R270K structure
the Thr278 side chain is placed within clashing distance of the L-Phe carboxylate.

S6

Figure S5: Stereo views of the active site residues in wt-PheH (A) and PheH24 R270K (B) superimposed on a 2mFo-DFc electron density map contoured at a 1.0 sigma level. In PheH24 R270K
Fe was lost during crystallization. (C) The wt-PheH C backbone superimposed on the composite
omit map calculated using PHENIX 8 and contoured at a 1.0 sigma level.

S7

Figure S6: SAXS reveals [L-Phe]-dependent changes in conformation and oligomeric state of PheH
in solution. (A) SAXS profiles of 25 M (concentration expressed as monomers) wt-PheH were
collected with varying concentrations of L-Phe, shown in the legend. Changes in the SAXS profiles
are observed at mid-q for [L-Phe] below 1 mM, indicative of a conformational change. An increase in
the low-q scattering is apparent for SAXS profiles at greater than 1 mM L-Phe (profiles multiplied
by 10 for clarity), suggesting that high concentrations of L-Phe favor the formation of high-mass
species. (B) The molecular weight (MW) was determined by Porod analysis of the profiles in (A)
with an expected accuracy of 10%. The apparent MW is larger than the 207 kDa expected for
tetrameric PheH, and increases rapidly above 1 mM L-Phe.

S8

Figure S7: SVD of L-Phe titration data. (A) To observe subtle changes in the scattering as a
function of [L-Phe] between 0 and 1 mM, a matrix A was constructed from the difference profiles
weighted by the experimental error. (B) SVD identifies two major components, the first (N = 1)
1
describes the scattering in the mid-q region (0.05 0.20 ) that changes systematically with [LPhe]. All other components (N > 1) describe experimental error (non-systematic aggregation) and
noise.

S9

Figure S8: Stability of buffer blanks in SEC-SAXS. The first and last scattering profiles collected
in each SEC-SAXS data set were compared to test for potential build-up of material on the Xray windows during the experiment or drift in the normalization that can affect the quality of
background subtraction. In both data sets (wt-PheH with and without 1 mM L-Phe in the running
buffer), scattering profiles overlay (top). Their differences are equal to zero within the experimental
noise (bottom).

S10

Figure S9: Rg and I(0) across the elution peak of wt-PheH with and without L-Phe. Although
only a single elution peak is observed in each case, the Rg values determined by Guinier analysis
vary, suggesting the presence of multiple species. While the low signal-to-noise of the profiles at the
beginning of the run complicates Guinier analysis, it clear that the Rg is significantly higher than
40 (dotted line).

S11

Figure S10: Conventional SVD of SEC-SAXS data from wt-PheH in 0 mM L-Phe. Three significant
singular values are observed (top panel). Although the corresponding right singular vectors (columns
of V ) have shapes that are reminiscent of elution peaks, we find that there are sign changes within
the curves. Moreover, when the scattering basis set is recovered by multiplying the columns of U
with the experimental error, we find that there are non-physical sign changes within the curves.
Because SVD produces orthogonal singular vectors, they cannot represent physical states, such as
elution peaks and scattering intensities, which must be positive numbers.

S12

Figure S11: Conventional SVD of SEC-SAXS data from wt-PheH in 1 mM L-Phe. Again, three
significant singular values are observed (top panel). Although the corresponding right singular
vectors (columns of V ) have shapes that are reminiscent of elution peaks, we find that there are sign
changes within the curves. Moreover, when the scattering basis set is recovered by multiplying the
columns of U with the experimental error, we find that there are non-physical sign changes within
the curves. Because SVD produces orthogonal singular vectors, they cannot represent physical
states, such as elution peaks and scattering intensities, which must be positive numbers.

S13

Start

Rotate Basis

Buffer-subtracted
SEC-SAXS profiles and
experimental errors

Start

Ij (q), Ij (q)
Prepare intensity data
and error matrices from
column vectors in order
of elution

Dij = Ij (qi )
Eij = Ij (qi )

Aij = Dij /

1
m

Pm

j=1

Eij

Cjk = Vjk

Mjk =

A is error-weighted so
each q-bin contributes
variance of 1

(f )

(r)

1 pk j pk
0 otherwise

Set numerical tolerance of


iterative refinement

Singular Value
Decomposition (SVD)

plot: singular
vectors and values

C0 = (G+ A)

Forward/reverse Evolving
Factor Analysis (EFA) is
used to find where peaks
begin/end from changes
in the rank of A as
columns are
added/removed

Normalize columns by their


peak areas

P 0
0 :=
0
Cjk
Cjk
/ ( h Chk
Mhk )
=

plot: evolving factors


List inflection points of
the forward/reverse
evolving factors in
increasing order

j,k

Compute absolute change in C


to check convergence




0
Cjk Cjk

< tol?

Yes

Calculate
coefficients


+
T
B = (M C)

Scattering
profiles of
the pure
components

No

Rotate Basis

Jk (qi ) = (D B)ik

C := C0

Concentration peak
profiles, scattering
profiles, and intensity
errors for each pure
component

Jk (qi ) =

End

p
((E E) (B B))ik

Propagate
uncertainty

End

Vectors and Scalars

Operators

Matrices

number of SAXS profiles collected


index of profiles: j = 1, 2, . . . , m
number of qvalues per profile
index of qvalues: i = 1, 2, . . . l
number of significant singular values
index of components: k = 1, 2, . . . , ns
absolute change of quantity being optimized
minimum change to stop optimization
vector of ns forward EFA inflection points
vector of ns reverse EFA inflection points

Symbols

New concentration matrix is


projection of the data onto the
basis set

Forward and
Backward EFA

m
j
l
i
ns
k

tol
(f )
p
p(r)

Compute basis set from


Moore-Penrose pseudoinverse of
masked concentration matrix


+
T
G = A (M C)

Graphically identify the


number of significant
singular values

user input: ns

C, Jk (q), Jk (q)

Create indicator matrix.


Mjk = 1 if profile number j falls
in window k

tol

SVD of A
A = U S VT

user input:
p(f ) , p(r)

Initialize C to the first ns


columns of V (from SVD of A)

Start/Stop

Input

D
E
A
U
S
V
C, C0
M
G
B

(l m) intensity data
(l m) standard errors
(l m) error-weighted intensitites
(l l) left singular vectors
(l m) singular values (diagonal)
(m m) right singular vectors
(m ns ) relative concentrations
(m ns ) binary indicator of peak ranges
(l ns ) rotated intensity basis vectors
(m ns ) intensity basis coefficients
Output

Assign

:=
AB
A+
AT

Routine

Assignment
Hadamard product
(A B)ij = Aij Bij
Moore-Penrose
inverse of A
Transpose of A

Decision

Figure S12: Adaptation of Evolving Factor Analysis (EFA) method 9,10 to separate overlapping
peaks in size exclusion chromatography-coupled small-angle X-ray scattering (SEC-SAXS) data.
All steps shown were implemented in MATLAB (The MathWorks, Inc., Natick, MA).
S14

Figure S13: Determination of peak ranges by EFA for wt-PheH in 0 mM L-Phe. EFA analysis
was performed twice: once in the forward direction (solid lines) and again in the reverse direction
(dashed). The inflection points (circles) show where there are abrupt changes in the number of
singular values as the size of the data matrix is varied. We find that the three significant components
identified by conventional SVD (Fig. S10) are found in three sets of images, corresponding to frame
numbers 147-200, 166-327, and 320-372. Only the first two occur within the main elution peak of
wt-PheH. The third occurs after the elution peak and is likely to represent a buffer component. After
peak ranges are determined by EFA, the singular vectors identified by SVD (Fig. S10) are rotated
(i.e. re-mapped) in a model-independent way to generate physical elution peaks and scattering
profiles.

S15

Figure S14: Determination of peak ranges by EFA for wt-PheH in 1 mM L-Phe. EFA analysis
was performed twice: once in the forward direction (solid lines) and again in the reverse direction
(dashed). The inflection points (circles) show where there are abrupt changes in the number of
singular values as the size of the data matrix is varied. We find that the three significant components
identified by conventional SVD (Fig. S11) are found in three sets of images, corresponding to frame
numbers 143-192, 163-358, and 320-410. Only the first two occur within the main elution peak of wtPheH . The third occurs after the elution peak and is likely to represent a buffer component. After
peak ranges are determined by EFA, the singular vectors identified by SVD (Fig. S11) are rotated
(i.e. re-mapped) in a model-independent way to generate physical elution peaks and scattering
profiles.

S16

Figure S15: Comparison of the wt-PheH crystal structure with X-ray scattering data at 0 mM LPhe. (A) The EFA-separated scattering profile of tetrameric wt-PheH in 0 mM L-Phe (blue points)
is well-described by a predicted curve from the wt-PheH crystal structure calculated in the program
CRYSOL 11 (solid lines), except at mid-q where the crystal structure prediction falls below the data.
Red and black lines correspond to CRYSOL calculations with and without modeling residues that
were missing in the crystal structure, respectively (see Fig. S16). The inclusion of these residues has
a relatively minor effect on the predicted scattering. (B) The pair distance distribution functions,
P (R), corresponding to each of the profiles on the left, differ in width, consistent with the differences
in the scattering at mid-q. The experimental curve suggests that wt-PheH is slightly expanded in
solution relative to the conformation depicted in the crystal structure.

S17

Figure S16: Effect of disordered residues on SAXS-based modeling of PheH conformation. Residues
1-18 and 136-143 that were missing from the crystal structure of wt-PheH were added using Modeller 12 . In the inset chain model, the fixed residues are drawn as gray ribbons, while modeled
residues are shown with different colors for each conformation. The scattering profiles predicted
for each model using CRYSOL 11 are shown with the same colors as their corresponding model in
the inset. For these models, the radius of gyration (RG ) varies by 1, as shown in the inset
histogram. Importantly, these calculations show that the particular conformation adopted by the
disordered residues does not greatly affect the shape of the scattering profile.

S18

Figure S17: Sequence alignment of the ACT domains of mammalian phenylalanine hydroxylases
(PheH) and bacterial prephenate dehydratases (PDT) performed using UCSF Chimera 13 . Percent
conservation is shown by the gray bars above the sequences. Secondary structure assignments for
Chlorobaculum tepidum PDT are based on the crystal structure where L-Phe is bound to its allosteric
site in the ACT domain (PDB 2QMX) 4 . For rat PheH, secondary structure assignments are shown
for the inactive conformation in the absence of L-Phe. When L-Phe is bound, the 2 strand has
been shown to elongate at the N-terminal end, becoming more like that of PDT 5 . Residues that
form the binding pocket for L-Phe in PDT and PheH 4,14 are shaded in gray. Filled circles indicate
residues that exhibit NOESY-HSQC cross-peaks between 15 N-labeled amide protons of rat PheH
regulatory domains and protons of L-Phe 5 .

S19

Figure S18: Regulatory L-Phe binding sites are conserved between mammalian PheH (PDB 5FII,
residues 34-112, drawn in green and yellow) 14 and bacterial PDT (PDB 2QMX, residues 196-302,
drawn in blue). 4 (A) The ACT domains of PDT from Chlorobium tepidum form homodimers that
bind two molecules of L-Phe at a pair of symmetrical sites 4 . In PheH, a similar binding mechanism
for L-Phe was first shown by NMR of the isolated regulatory domains, 5 and subsequently confirmed
by X-ray crystallography. 14 (B) In each L-Phe binding site of PheH (PDT), residues Leu48, Leu62,
Ser67, Phe79, and Tyr77 (Leu211, Leu225, Ser230, Tyr240, and Phe242) create a hydrophobic
pocket for the aromatic ring. Hydrogen bonding interactions are proposed between the -amino
group of L-Phe and the side-chains of Glu43 and Asn61 (Asn206 and Asp224), and between the
carboxylate of L-Phe and the main-chain amides of Gly46, Ala47, Leu48, and Asn61 (Gly209,
Ser210, Leu211 and Asp224). 4,14

S20

Supplementary References
(1) Fusetti, F.; Erlandsen, H.; Flatmark, T.; Stevens, R. C. J. Biol. Chem. 1998, 273, 16962
16967.
(2) Goodwill, K. E.; Sabatier, C.; Marks, C.; Raag, R.; Fitzpatrick, P. F.; Stevens, R. C. Nat.
Struct. Mol. Biol. 1997, 4, 578585.
(3) Kobe, B.; Jennings, I. G.; House, C. M.; Michell, B. J.; Goodwill, K. E.; Santarsiero, B. D.;
Stevens, R. C.; Cotton, R. G. H.; Kemp, B. E. Nat. Struct. Mol. Biol. 1999, 6, 442448.
(4) Tan, K.; Li, H.; Zhang, R.; Gu, M.; Clancy, S. T.; Joachimiak, A. J. Struct. Biol. 2008, 162,
94107.
(5) Zhang, S.; Fitzpatrick, P. F. J. Biol. Chem. 2016, 291, 74187425.
(6) Gjetting, T.; Petersen, M.; Guldberg, P.; Gttler, F. Am. J. Hum. Genet. 2001, 68, 1353
1360.
(7) Roberts, K. M.; Khan, C. A.; Hinck, C. S.; Fitzpatrick, P. F. Biochemistry 2014, 53, 7846
7853.
(8) Adams, P. D.; Afonine, P. V.; Bunkczi, G.; Chen, V. B.; Davis, I. W.; Echols, N.; Headd, J. J.;
Hung, L.-W.; Kapral, G. J.; Grosse-Kunstleve, R. W.; McCoy, A. J.; Moriarty, N. W.;
Oeffner, R.; Read, R. J.; Richardson, D. C.; Richardson, J. S.; Terwilliger, T. C.; Zwart, P. H.
Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010, 66, 213221.
(9) Maeder, M.; Neuhold, Y.-M. Practical Data Analysis in Chemistry; Data handling in science
and technology 26; Elsevier: Boston, 2007.
(10) Maeder, M. Anal. Chem. 1987, 59, 527530.
(11) Svergun, D.; Barberato, C.; Koch, M. H. J. J. Appl. Crystallogr. 1995, 28, 768773.
(12) Fiser, A.; Do, R. K.; Sali, A. Protein Sci. 2000, 9, 17531773.
(13) Meng, E. C.; Pettersen, E. F.; Couch, G. S.; Huang, C. C.; Ferrin, T. E. BMC Bioinformatics
2006, 7, 339.
(14) Patel, D.; Kopec, J.; Fitzpatrick, F.; McCorvie, T. J.; Yue, W. W. Sci. Rep. 2016, 6, 23748.

S21

Vous aimerez peut-être aussi