Structure and Proposed Mechanism for the pH-Sensing ... · Structure and Proposed Mechanism for the pH-Sensing Helicobacter pylori ... green and red. (B) Overlay of TlpBpp (cyan)

Structure, Volume 20 Supplemental Information

Structure and Proposed Mechanism

for the pH-Sensing Helicobacter pylori

Chemoreceptor TlpB

Emily Goers Sweeney, J. Nathan Henderson, John Goers, Christopher Wreden, Kevin G. Hicks, Jeneva K. Foster, Raghuveer Parthasarathy, S. James Remington, and Karen Guillemin Inventory of Supplemental Information Figure S1, related to Figure 1. Stero view of TlpBpp monomer with urea bound and overlays of TlpBpp with putative MCPs from Vibrio cholerae (3C8C), Vibrio parahaemolyticus (2QHK) and Photoactive Yellow Protein (1OTD). Figure S2, related to Figure 2. Overview of the urea binding assay and results, urea binding affinity to TlpBpp, simulated unfolding curves, and reversible stabilization of TlpBpp by urea. Figure S3, related to Figure 3. Crystallographic studies show urea-like molecules bind TlpBpp in the urea binding site. Movie S1, related to Figure 4. WT H. pylori form a barrier in response to 100mM HCl treatment in the barrier assay. Figure S4, related to Figure 4. TlpB mutants form barriers to AI-2 and the acid barrier assay forms a pH gradient. Supplementary Experimental Procedures and Discussion: Experimental Discussion S1. Structural comparisons of TlpBpp with receptors containing PAS domains Experimental Procedure S1. Crystallization, derivatization and diffraction data collection Experimental Procedure S2. Data reduction and structure determination Experimental Procedure S3. Reassessment of Protein Data Bank entry 2QHK Experimental Procedure S4. Determining an apparent binding affinity of urea to TlpBpp using equilibrium dialysis. Experimental Procedure S5. Numerical simulation of coupled binding/thermal denaturation equilibria

Supplementary Information

Figure S1, related to Figure 1. Stero view of TlpBpp monomer with urea bound

and overlays of TlpBpp with putative MCPs from Vibrio cholerae (3C8C), Vibrio

parahaemolyticus (2QHK) and Photoactive Yellow Protein (1OTD). (A) Stero

view of TlpBpp monomer. Blue is the N-terminus and red is the C-terminus. Urea

is the molecule in blue, green and red. (B) Overlay of TlpBpp (cyan) and 3C8C

(green) with urea and alanine as spheres. (C) Overlay of TlpBpp (cyan) with

2QHK (green) and urea and pyruvate as sticks. (D) Ball and stick representation

of the presumed pyruvate binding site of 2QHK (C), showing hydrogen bonds to

critical protein side chains as dashed lines. The magenta contour cage

represents the (Fo-Fc) difference electron density map, calculated after removal

of pyruvate (cyan) from the atomic model and contoured at a level of +2.0sigma.

The right panel shows the Fo-Fc difference electron map for the original 2QHK

model, containing a putative glycerol ligand in the binding site. The difference

density map is contoured at +/- 3sigma. Red contours indicate a strong (>

3sigma) negative feature at the 3-hydroxyl of modeled glycerol. Blue contours

indicate a strong (> 3sigma) positive feature corresponding to the methyl group of

pyruvate, the preferred alternative interpretation for the bound ligand. Lys131 and

His79 in 2QHK correspond to Lys166 and Asp114 in TlpB. (E) Cartoon diagram

showing a structural overlay of the PAS domain of TlpB (green) with Photoactive

Yellow Protein (yellow). Urea (cyan, red and blue) and cinnamic acid (gray and

red) are shown in ball and stick representation. Oxygen atoms are shown as red

spheres, and nitrogen as blue.

Figure S2, related to Figure 2. Overview of the urea binding assay and results,

urea binding affinity to TlpBpp, simulated unfolding curves, and reversible

stabilization of TlpBpp by urea. (A) An example of the urea standard used in the

urea colorimetric assays to determine percent mole urea bound per mole of

protein. Inset includes absorbance values for urea-like molecules (100 µM),

controls for the specificity of the assay, and urea (80 µM). (B) Results of the urea

colorimetric assay. (C) Results for identifying an apparent disassociation

constant using equilibrium dialysis. TlpBpp-formamide (“apo-TlpBpp”) was dialyzed

against increasing concentrations of urea and the fraction of TlpBpp bound by

urea was plotted. (D) CD thermal melts of 1µM TlpBpp with 0-5000µM urea added

exogenously. All traces were normalized to -29 mdeg at temperature = 0°C.

Panels E, F, and G show simulated unfolding curves for the coupled equilibria of

equation (1), as a function of temperature and ligand concentration. Panel E

shows the consequences of simulated weak binding (Kd = 1 mM), panel F,

simulated moderate binding (Kd = 1 μM) and panel G tight binding (Kd=1 nM).

See text of Supplemental Experimental Procedures for a detailed discussion. (H)

CD re-folding experiment with 0mM urea (green), 1mM urea (blue) and 2mM

urea (red) added to TlpBpp at pH 4. Percent of maximum fold refers to the CD

signal (220nm) of TlpBpp with 5mM urea at 2 °C. (I) CD thermal melt of TlpBpp

with 5mM urea added prior to melt (solid line) and TlpBpp partially melted, then

re-folded with 5mM urea from part G (dashed line) at pH 4.

Figure S3, related to Figure 3. Crystallographic studies show urea-like

molecules bind TlpBpp in the urea binding site. Stick diagrams of the urea binding

sites in TlpBpp crystal structures with urea bound, acetamide bound, formamide

bound, or hydroxyurea bound. The small molecules are represented as sticks

and electron density (pink cages) in the models. Notice the urea-like small

molecules have several conformations, whereas urea has one. The electron

density observed in each case for the three urea-like ligands could not be

modeled by a single conformation, and hence, was modeled as a superposition

of two copies of the ligand in different orientations, each at 50% occupancy.

Movie S1, related to Figure 4. WT H. pylori form a barrier of moving bacteria in

response to 100mM HCl treatment (treatment introduced from the right) in the

barrier assay. The movie was taken 10min after adding HCl treatment. The arrow

in the first few frames indicates the “barrier” region where bacteria (white specs)

are highly motile.

Figure S4, related to Figure 4. TlpB mutants form barriers to AI-2 and the acid

barrier assay forms a pH gradient. (A) Images of barriers of moving bacteria form

in response to AI-2 treatment. (B) To determine if a pH gradient is formed,

bromophenol blue was added throughout the neutral media under the coverslip

and throughout the acid treatment (100mM HCl, pH~2) entering from the right.

The image was taken on the same time scale as all the other barrier assays (3-

10min after addition of treatment). Barriers of bacteria generally form near the

region where the pH shifts from low to high (at the yellow to blue interface).

Supplementary Experimental Procedures and Discussion

Experimental Discussion S1. Structural Comparison of TlpBpp with Receptors

Containing PAS Domains

The crystal structures of the periplasmic domains of two putative methyl-

accepting chemotaxis proteins (MCPs) of unknown function have been deposited

in the Protein Data Bank (PDB) but appear to be otherwise unpublished. Both

models contain PAS domains bound to a tentatively identified small molecule.

Entry 3C8C from the human pathogen Vibrio cholerae Mcp_N contains tandem

PAS domains (residues 123-203 and 220-286), with overall organization very

similar to that proposed by Glekas et al. (Glekas et al., 2010) for the Bacillus

subtilis Mcp_B (see Discussion) (Figure S1B). Residues 123-203 of 3C8C

constitute a PAS domain and is proposed to bind alanine, whose binding site is

equivalent to that of bound urea in TlpB (Figure S1B).

Entry 2QHK describes the periplasmic domain of a hypothetical MCP derived

from the pathogen Vibrio parahaemolyticus and contains a single PAS domain

(residues 64-151), with glycerol assigned as a ligand (Figure S1C). 2QHK is

remarkably similar in overall organization to H. pylori TlpBpp. Overall 108 of the

~150 alpha-carbons in the 2QHK protomer were found to superimpose onto

TlpBpp to ~1.6Å rms, so the overall backbone folds are essentially identical. The

PAS domains align with1.3Å rms deviation for 65 -carbons, which also aligns

critical residues in the ligand binding sites. As described in the Supplementary

Experimental Procedure S3, the deposited data had to be reinterpreted and

instead of glycerol, pyruvate is suggested to occupy the binding site.

Remarkably, two residues that we have identified as the key urea ligands in TlpB

(Lys166 and Asp114) correspond structurally to Lys131 and His79 in 2QHK, both

of which interact with the carboxylate moiety of pyruvate (Figure S1D).

Presuming the pyruvate to be anionic as bound, Lys131 ensures charge

neutrality of the binding site. We thus predict that 2QHK is a chemotaxis receptor

with pyruvate as a cofactor or a chemoeffector.

The Photoactive Yellow Protein (PYP) was the first member of the PAS

superfamily to have its structure determined (Borgstahl et al., 1995). PYP is a

photoreceptor found in a number of marine bacteria and is responsible for

negative phototactic response to blue light. It contains a covalently bound

cinnamic acid chromophore. An overlay of residues 29-124 of PYP (PDB entry

1OTD) with residues 99-185 of TlpBpp results in 36 alpha-carbons superimposing

to 3.8 Å rms (Figure S1E). The β-sheet regions of the two domains are closely

similar but the corresponding helical regions are somewhat displaced. However,

the overlay shows that the chromophore binding site in PYP is topologically

equivalent to the urea binding site in TlpB.

Experimental Procedure S1. Crystallization, derivatization and diffraction data

collection

Diffraction-quality crystals of TlpBpp were obtained at 4C in a wide range of

conditions by using the hanging-drop vapor diffusion technique. The best crystals

grew from two conditions: 20-26% PEG 3350 (Hampton Research), 0.1 M buffer

(Bis-Tris, sodium acetate or PIPES) at pH 4.6 to 6.5 and 0.05-0.2 M ammonium

sulfate or 1.7-2.3 M ammonium sulfate and 0.1 M buffer (Bis-Tris, sodium acetate

or PIPES) at pH 4.6 to 6.5. Each 4 µL crystallization drop contained a one to one

mixture of protein to reservoir solution. Native crystals were soaked for 30

minutes at 4C in mother liquor containing 20% glycerol prior to flash-freezing.

The heavy atom derivative was obtain from a 30 minute soak with mother liquor +

15 % glycerol + 0.3 M potassium iodide. After flash-freezing, diffraction data were

collected at 100 K on some crystals using either an in house Rigaku rotating

anode generator with a Raxis 4 detector or crystals were stored under liquid

nitrogen prior to data collection at the Advanced Light Source (Berkeley, CA)

beamline 5.0.3 using the ADSC-Q315 detector at a wavelength of 0.97 Å and a

temperature of 100 K.

Experimental Procedure S2: Data reduction and structure determination

Home source and synchrotron data were processed using the HKL2000 suite

(Otwinowski and Minor, 1997). Initial phases were obtained and modified with

SHELXC/D/E using the SIRAS method (Schneider and Sheldrick, 2002;

Sheldrick, 2002). Seventeen heavy atom sites were found, two of which were

strongly occupied. ARP/wARP was then used to build a partial model into the

experimentally phased electron density maps (Morris et al., 2003). Automatic

model building resulted in a partial model consisting of 285 out of the 358 amino

acids. The phases from ARP/wARP were extended to 1.87 Å using REFMAC 5

(Murshudov et al., 1997). The final model was completed manually by using

COOT (Emsley and Cowtan, 2004). Water molecules were placed using COOT

with the combined criteria of a peak of greater than 2.5 in the (Fo – Fc)

difference map, or 1.0 in the (2Fo – Fc) map and reasonable intermolecular

interactions. This model was used as the starting point for further refinement

once higher resolution synchrotron data were collected (Table S1).

Experimental Procedure S3: Reassessment of Protein Data Bank entry 2QHK

The X-ray diffraction data for the putative MCP 2QHK are available, so we

subjected the deposited model to additional crystallographic refinement at 1.9 Å

resolution and checked the final electron density map. Strong positive and

negative features in an (Fo-Fc) difference electron density map indicate that

glycerol is not an appropriate model for the bound ligand. The crystallographic

data were reinterpreted and pyruvate was assigned, based on the size and

shape of the electron density feature and the nature of the side chains forming

the binding site. . In particular, a strong positive peak appeared for the methyl

group of pyruvate, and a strong negative feature indicated that the 3’ OH of

glycerol should be deleted (see Figure S1D). In further support of this

assignment, the methyl group of the presumed pyruvate ligand fits into a small,

strictly hydrophobic cavity lined by Phe118, Ile102 and Leu105, while the

carboxylate interacts with charged Lys131 and neutral His79. Pyruvate is

sandwiched between Phe68 and Trp120 (Figure S1D) with the latter probably

acting as a “swinging gate” allowing access to the binding site.

Experimental Procedure S4. Determining an apparent binding affinity of urea to

TlpBpp using equilibrium dialysis.

The apparent TlpBpp affinity for urea was determined by equilibrium dialysis and

the in vitro urea chemical assay. TlpBpp bound with the urea analog formamide

was used in lieu of insoluble apo-TlpBpp in all binding assays. TlpBpp-formamide

was generated from TlpBpp-urea by dialysis against 250 mM formamide, 0.30 M

NaCl, 10 mM Kphos pH 7.0 at 4oC for 48 hours followed by extensive dialysis

against 0.30 M NaCl, 10 mM Kphos pH 7.0 at 4oC to remove free formamide.

The resultant TlpBpp-formamide had 0.080 mol urea/mol TlpBpp. Volumes of 0.30

mL TlpBpp-formamide (at 5.0mg/mL, 2.4 x 10-4 M) were then dialyzed against

0.55 L of 0.1 – 100 �M urea in 0.30 M NaCl, 10 mM Kphos pH 7.0 at 4oC for 72

hrs. TlpBpp protein in the samples was measured by absorbance at 280 nm (∑ =

2.52 x 104 M-1cm-1) and bound urea determined by the in vitro urea binding assay

after correcting for free urea. A plot of urea bound TlpBpp (Fraction Bound)

verses total urea was fitted with a curve to determine the dissociation constant

(Stein et al., 2001).

Experimental Procedure S5: Numerical simulation of coupled binding/thermal

denaturation equilibria

The specific binding of a ligand always stabilizes the bound form of the

protein (relative to the ligand free form) toward denaturation, and therefore, the

protein complex becomes more resistant to thermally induced denaturation

(Brandts and Lin, 1990; Waldron and Murphy, 2003). For reversible thermally

induced protein unfolding, the coupled equilibria set forth in Equation 1 have

been shown to successfully model a very diverse array of protein-ligand systems,

with examples ranging from P22 bacteriophage Arc repressor binding to DNA

(Bowie and Sauer, 1989) to the binding of benzene within cavity-containing

mutants of T4 lysozyme (Eriksson et al., 1991) . The model assumes that the

ligand does not bind to the unfolded form of the protein.

As described in the manuscript, and experimentally observed for TlpB,

increasing the concentration of ligand (L) increases the apparent denaturation

temperature Tm of the protein-ligand complex (FL), even for the case that the

concentration of ligand [L] is orders of magnitude higher than (an assumed) Kd.

It is counterintuitive that Eqn 1 is applicable to cases where ligand binding

is too tight to measure by experimental methods such as equilibrium dialysis, i.e.

with Kd < 10-10 M (Brandts and Lin, 1990), and it is instructive to model the

behavior of a typical system numerically. Celej and Fidelio (Celej et al., 2005)

give an excellent pedagogical overview and present examples of the

consequences of Eqn. 1. Their approach was used to simulate the behavior of

the TlpB-urea complex. Although Eqn. 1 assumes a monomeric protein, Eqn. 1 is

also applicable to a dimer that unfolds non-cooperatively (Bowie and Sauer,

1989). Here, we show that Eqn. 1 models the observed thermal unfolding

behavior of TlpB for a range of assumed Kd values ranging from 10-9 M to 10-3 M

We define in the usual way:

[ ][ ] /[ ]dK L F FL  

( ) [ ] /[ ]uK T U F  

( ) ln ( )u uG T RT K T  

An empirical expression for the free energy of unfolding of a “typical” protein is

mm m

m m

T - T T( ) 2.92N + 0.058N(T - 333)( )- 0.058N(T - T(1 - ln )) ( / )

T TuG T kJ mole  

Where T is the temperature in Kelvin, Tm = the melting temperature for the

ligand free form of the protein and N is the number of amino acids. The above

empirical formula for ( )uG T was derived by Rees and Robertson (Rees and

Robertson, 2001) from a large number of published measurements on various

protein-ligand systems.

Assuming mass balance, total protein and ligand concentrations F0 and L0

respectively, at any given temperature T this system of equations can be solved

exactly for the quantities of interest, namely the concentrations of unfolded

protein [U(T)], free protein [F(T)] and the complex [FL(T)] (Celej et al., 2005). A

short program was written to solve the system for a range of temperatures,

assumed Kd values, assumed total concentrations of protein and ligand. As a

model for TlpB, Tm was assumed to be 300 K and the N, the number of amino

acids, was set to 200. Total protein concentration was assumed to be 2 μM,

which is typical for spectroscopic experiments. The results are plotted and

presented in Figure S2, panels E-G.

Clearly, the thermal stabilization induced by ligand binding is not

saturating and the effect continues to be significant at ligand contributions orders

of magnitude higher than Kd. Of particular interest is the bimodal transition

observed in Figure S3, panel D for Kd = 1 nM, [protein]=2 μM, [ligand]=1 μM. This

corresponds to half saturation so that unbound protein unfolds at the assumed

Tm = 300 K, whereas the bound protein unfolds at ~320 K. Both transitions are

further stabilized by increasing ligand, until the protein becomes effectively

saturated. This behavior is not what was experimentally observed for TlpB

(Figure S2D), as the higher temperature transition is not affected by increasing

urea concentrations. We suggest that the first unfolding transition of TlpB is the

PAS domain while the second represents the unfolding of the four helix bundle. 

Supplementary References

Borgstahl, G.E., Williams, D.R., and Getzoff, E.D. (1995). 1.4 A structure of

photoactive yellow protein, a cytosolic photoreceptor: unusual fold, active site,

and chromophore. Biochemistry 34, 6278-6287.

Bowie, J.U., and Sauer, R.T. (1989). Equilibrium Dissociation and Unfolding of

the Arc Repressor Dimer. Biochemistry 28, 7139-7143.

Brandts, J.F., and Lin, L.N. (1990). Study of strong to ultratight protein

interactions using differential scanning calorimetry. Biochemistry 1990, 6927-

6940.

Celej, M.S., Fidelio, G.D., and Dassie, S.A. (2005). Protein Unfolding Coupled to

Ligand Binding: Differential Scanning Calorimetry Simulation Approach. J Chem

Ed 82, 87-92.

Emsley, P., and Cowtan, K. (2004). Coot: model-building tools for molecular

graphics. Acta Crystallogr D Biol Crystallogr 60, 2126-2132.

Eriksson, A.E., Baase, W.A., Wozniak, J.A., and Matthews, B.W. (1991). A

cavity-containing mutant of T4 lysozyme is stabilized by buried benzene. Nature

355, 371-373.

Glekas, G.D., Foster, R.M., Cates, J.R., Estrella, J.A., Wawrzyniak, M.J., Rao,

C.V., and Ordal, G.W. (2010). A PAS domain binds asparagine in the chemotaxis

receptor McpB in Bacillus subtilis. J Biol Chem 285, 1870-1878.

Morris, R.J., Perrakis, A., and Lamzin, V.S. (2003). ARP/wARP and automatic

interpretation of protein electron density maps. Methods Enzymol 374, 229-244.

Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of

macromolecular structures by the maximum-likelihood method. Acta Crystallogr

D Biol Crystallogr 53, 240-255.

Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction data

collected in oscillation mode. In Methods in Enzymology, Charles W. Carter, Jr.,

ed. (Academic Press), pp. 307-326.

Rees, D.C., and Robertson, A.D. (2001). Some thermodynamic implications for

the thermostability of proteins. Protein Science 10, 1187-1194.

Schneider, T.R., and Sheldrick, G.M. (2002). Substructure solution with

SHELXD. Acta Crystallographica 58, 1772-1779.

Sheldrick, G.M. (2002). Macromolecular phasing with SHELXE. Macromolecular

Crystallography 217, 644-650.

Stein, R.A., Wilkinson, J.C., Guyer, C.A., and Staros, J.V. (2001). An analytical

approach to the measurement of equilibrium binding constants: application to

EGF binding to EGF receptors in intact cells measured by flow cytometry.

Biochemistry 40, 6142-6154.

Waldron, T.T., and Murphy, K.P. (2003). Stabilization of Proteins by Ligand

Binding: Application to Drug Screening and Determination of Unfolding

Energetics. Biochemistry 42, 5058-5064.