pure.qub.ac.uk€¦ · web viewcdnas. an overnight culture grown in luria–bertani media...

61
The metastability of human UDP-galactose 4ʹ-epimerase (GALE) is increased by variants associated with type III galactosemia but decreased by substrate and cofactor binding. Angel L. Pey 1* , Esperanza Padín-Gonzalez 1 , Noel Mesa-Torres 1 and David J. Timson 2* . 1 Department of Physical Chemistry, Faculty of Sciences, University of Granada, Av. Fuentenueva s/n, 18071, Spain. 2 School of Biological Sciences, Queen´s University Belfast, Medical Biology Centre, 97 Lisburn Road, Belfast BT9 7BL, UK. * correspondence to: [email protected] (A.L.P.) and [email protected] (D.J.T.) 1

Upload: others

Post on 05-Apr-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

The metastability of human UDP-galactose 4ʹ-epimerase (GALE) is increased by

variants associated with type III galactosemia but decreased by substrate and

cofactor binding.

Angel L. Pey1*, Esperanza Padín-Gonzalez1, Noel Mesa-Torres1 and David J. Timson2*.

1 Department of Physical Chemistry, Faculty of Sciences, University of Granada, Av.

Fuentenueva s/n, 18071, Spain.

2 School of Biological Sciences, Queen´s University Belfast, Medical Biology Centre,

97 Lisburn Road, Belfast BT9 7BL, UK.

* correspondence to: [email protected] (A.L.P.) and [email protected] (D.J.T.)

1

Page 2: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Abstract

Type III galactosemia is an inherited disease caused by mutations which affect

the activity of UDP-galactose 4ʹ-epimerase (GALE). We evaluated the impact of four

disease-associated variants (p.N34S, p.G90E, p.V94M and p.K161N) on the

conformational stability and dynamics of GALE. Thermal denaturation studies showed

that wild-type GALE denatures at temperatures close to physiological, and disease-

associated mutations often reduce GALE’s thermal stability. This denaturation is under

kinetic control and results partly from dimer dissociation. The natural ligands, NAD+

and UDP-glucose, stabilize GALE. Proteolysis studies showed that the natural ligands

and disease-associated variations affect local dynamics in the N-terminal region of

GALE. Proteolysis kinetics followed a two-step irreversible model in which the intact

protein is cleaved at Ala38 forming a long-lived intermediate in the first step. NAD+

reduces the rate of the first step, increasing the amount of undigested protein whereas

UDP-glucose reduces the rate of the second step, increasing accumulation of the

intermediate. Disease-associated variants affect these rates and the amounts of protein in

each state. Our results also suggest communication between domains in GALE. We

hypothesize that, in vivo, concentrations of natural ligands modulate GALE stability and

that it should be possible to discover compounds which mimic the stabilising effects of

the natural ligands overcoming mutation-induced destabilization.

Keywords.- Protein conformational stability; Protein dynamics; Protein aggregation;

Proteolysis; Type III galactosemia; Ligand binding.

2

Page 3: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Introduction

UDP-galactose 4ʹ-epimerase (GALE; EC 5.1.3.2) catalyses the interconversion

of UDP-glucose (UDP-glc) and UDP-galactose (UDP-gal), a reaction which is required

for galactose metabolism [1]. Defects in GALE activity due to mutations in the

corresponding gene cause the inherited metabolic disease type III galactosemia (OMIM

#230350). To date, 22 disease-associated variants of the protein have been described in

the literature [2]. The symptoms of type III galactosemia are very varied. In the mildest

forms of the disease, altered blood chemistry is observed and no interventions are

recommended. In contrast, the most severe forms of the disease result in progressive

damage to key organs including the kidneys, liver and brain. In such cases, reduction in

dietary galactose intake is required. Currently this is the only available therapy for the

disease. However, while this slows down the development of symptoms, it does not

prevent or reverse them [3].

Human GALE functions as a dimer of two identical, 38 kDa subunits [4].

Sequence and structural analysis showed that GALE is a member of the short-chain

dehydrogenase/reductase (SDR) family of enzymes [5]. Each subunit contains one

active site with a tightly bound NAD+ cofactor which plays a key role in the catalytic

mechanism of the enzyme [6] . This cofactor transiently oxidises the C4-OH group on

the sugar moiety of UDP-galactose. Rotation of the sugar moiety, followed by its

reduction from the opposite face reverses the stereochemistry at C4 and produces the

product UDP-glucose [7]. Many GALE enzymes, including the human one, can also

catalyse the interconversion of UDP-N-acetylgalactosamine and UDP-N-

acetylglucosamine [8]. These compounds are precursors in the synthesis of

glycoproteins and glycolipids and disturbances in this metabolism are likely to

contribute to molecular pathology [9].

3

Page 4: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Many of the disease-causing mutations affect GALE catalytic properties (e.g.

p.G90E, p.V94M and p.K161N) and there is some degree of correlation between the

degree of impairment of the turnover number (kcat) and the severity of the disease

phenotype. In some cases, disease-associated variations in GALE affect its

conformational stability (e.g. p.N34S, p.G90E and p.K161N) or reduce the affinity for

the NAD+ cofactor [10-13]. Thus, disease-associated variants may result in altered

stability, reduced catalytic activity and/or cofactor binding. However, binding of the

substrate (UDP-gal), the product (UDP-glc) or its cofactor (NAD+) may enhance the

protein’s conformational stability [10, 12, 14]. Nevertheless, the effects of sequence

alterations on local and global GALE stability and the modulation by ligand binding on

the pathogenesis of type III galactosemia are not well explored or understood.

Here, we provide new insight on the conformation, stability, dynamics and

ligand binding to WT GALE and four disease-associated variants (p.N34S, p.G90E,

p.V94M and p.K161N), using a combination of spectroscopic analyses, thermal

denaturation and kinetics of proteolysis. These variants were chosen to represent a

diverse spectrum of GALE variants. p.N34S (c.101A>G; rs121908046)1 has near wild-

type kinetics when assayed in excess NAD+, but has a much lower affinity for this

cofactor than the wild-type [10, 16]. Similarly, p.K161N (c.483G>T; no rs number) also

has reduced affinity for the cofactor, but has very low activity even in the presence of

excess NAD+ [12]. p.G90E (c.269G>A; rs28940882) is highly unstable towards limited

proteolysis studies and has a turnover number (kcat) reduced approximately 800-fold

compared to wild-type [11]. p.V94M (c.280G>A; rs121908047) is the variant most

commonly found in severely affected patients [10, 11]. It is also highly catalytically

impaired (30-fold reduction in kcat) [10, 11]. Interestingly, it has been reported that this

1 “p.” refers to the protein sequence and “c.” to the DNA coding sequence [15]. “rs” stands for “reference SNP” and identifies these single nucleotide polymorphisms in databases such as NCBI SNP and Ensembl.

4

Page 5: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

variant is slightly more stable towards limited proteolysis than the wild-type [11]. A

recent computational study predicted little change in the global flexibility of p.V94M

compared to the wild-type [13]. Of particular significance are our findings on the

stabilizing effects towards partial denaturation and proteolysis upon ligand binding;

these suggest that long-range transmission of binding effects between domains leading

to significant changes in protein local flexibility and dynamics, and in dimer stability.

Therefore, our results provide a deep understanding of the mutational effects on protein

stability and dynamics in type III galactosemia which are important, first for unravelling

the fundamental links between sequence changes and disease, and second for the design

of small molecules to stabilise the disease-associated variants (i.e. pharmacological

chaperones) [17-19].

5

Page 6: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Materials and methods

Protein expression and purification

Human GALE proteins were expressed in, and purified from, E. coli essentially

as described previously [10]. E. coli BL21(DE3) cells were transformed with plasmids

containing WT and mutant GALE cDNAs. An overnight culture grown in Luria–Bertani

media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–

Bertani media supplemented with 0.1 mg/ml ampicillin. The cells were grown at 37 ºC

to OD600 of 0.6 and then induced with 0.5 mM isopropyl--D-thiogalactopiranoside

(IPTG) for 8 h at 25 ºC. Cells were then harvested and frozen at -80 ºC for 16 h. Cells

were resuspended in binding buffer (20 mM Na-phosphate, 300 mM NaCl, 20 mM

imidazole pH 7.4 and COMPLETE EDTA-FREE protease inhibitor cocktail from

Roche) and disrupted by sonication. The supernatants obtained after ultracentrifugation

(70000 g, 30 min, 4 ºC) were loaded onto immobilized metal affinity chromatography

columns (GE Healthcare), washed with binding buffer (50 bed volumes) and eluted

using binding buffer supplemented with 250-500 mM imidazole. These eluates were

loaded onto a Superdex 200 prep grade column (GE Healthcare) running in 20 mM

Hepes-OH 200 mM NaCl pH 7.4 and calibrated using the following standards: blue

dextran (void volume), thryroglobulin (669 kDa), alcohol dehydrogenase (141 kDa),

bovine serum albumin (66 kDa), ovalbumin (45 kDa), carbonic anhydrase (29 kDa),

cytochrome C (12.3 kDa) and acetone (total volume). Those fractions corresponding to

GALE dimers (retention time of 88.00.2 ml for the five variants in two independent

purifications) were pooled, concentrated, flash frozen in liquid nitrogen and stored at -

80 ºC. Protein concentration was measured spectrophotometrically using ε280 of 46215

M-1·cm-1, which is based on the primary sequence of GALE and using the procedure of

Pace and coworkers [20].

6

Page 7: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Spectroscopic and dynamic light scattering studies

Absorption spectra were measured in an Agilent 8453 diode-array

spectrophotomer, using 40 M GALE in monomer and 3 mm pathlength quartz

cuvettes. Emission fluorescence spectra were acquired in Cary Eclipse

spectrofluorimeter using 3 mm pathlength quartz cuvettes and 4 M GALE in

monomer, with an excitation wavelength of 280 or 295 nm, and slits for emission and

excitation of 5 nm and a scan rate of 200 nm/min. Circular dichroism was measured in a

Jasco J-710 spectropolarimeter, using 4 M (far UV) or 20 M (near UV-visible)

GALE, and scan rates of 50 (far UV) and 100 (near UV-visible) nm/min. Dynamic light

scattering (DLS) was carried out in a DynaPro MSX instrument (Wyatt) using 1.5 mm

path length cuvettes and 20 µM protein monomer at 25 ºC. 25 spectra were acquired for

each DLS analysis in three independent replicates, averaged and used to determine the

hydrodynamic radius and polydispersity using the average autocorrelation function and

assuming a spherical shape. These experiments were performed in 20 mM HEPES-OH

200 mM NaCl pH 7.4 at 25 ºC, except far UV CD which were acquired in 20 mM K-

phosphate 200 mM KCl pH 7.4, and blanks in the absence of protein were routinely

measured and subtracted. Secondary structure content was determined using the K2D3

algorithm [21] (available online http://www. ogic.ca/projects/k2d3).

Differential scanning calorimetry (DSC)

DSC experiments were performed on a capillary VP-DSC differential scanning

calorimeter (GE Healthcare) with a cell volume of 0.135 mL. Thermal scans were

performed at a rate of 3°C·min-1 in a temperature range of 2−80 °C using 10 μM GALE

(in monomer) in 20 mM HEPES-OH, 200 mM NaCl pH 7.4, unless otherwise indicated.

In some experiments, NAD+ or UDP-glucose was added to a final concentration of 1

mM (unless otherwise indicated), and their concentration was measured using an ε259

7

Page 8: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

of 16900 M-1·cm-1(NAD+) and ε262 of 10000 M-1·cm-1(UDP-glucose). To estimate the

apparent Tm and denaturation enthalpies (H), we applied a simple two-state irreversible

denaturation model as described [22-24].

Isothermal titration calorimetry (ITC)

ITC experiments were performed on an ITC200 titration microcalorimeter (GE

Healthcare) with an operating cell volume of 206 L. In the calorimetric cells, we

placed a solution of 25 M GALE (monomer) in 20 mM HEPES-OH 200 mM NaCl

pH 7.4, while in the titrating syringe we put 500M of ligand (NAD+ or UDP-

glucose) in 20 mM Hepes-OH 200 mM NaCl pH 7.4. Experiments were performed at

25 ºC by performing 1 × 0.5L plus 30 × 1.2L (NAD+) or 1 × 0.5L plus 60 × 0.6

L (UDP-glucose) injections upon continuous stirring at 1000 rpm. Binding heats were

integrated and corrected for dilution heats. Binding isotherms were analyzed using a

single type of independent and identical binding sites model found in the software

provided by the manufacturer. Briefly, the heat (Q) evolved from the non-ligated GALE

species to a given saturation fraction is expressed by the following equation:

Q=n M T Δ H V 0

2 (1+ LT

n M T+

1n K a M T

−√(1+LT

nM T+

1n Ka M T )

2

−4 LT

n M T ) where n is the binding stoichiometry per GALE monomer, MT is the GALE

concentration in monomer, H is the binding enthalpy, V0 is the cell volume, LT is the

total ligand concentration and Ka is the association ligand binding constant. This

equation can be used to determined to total heat evolved after the ith injection [Q(i)].

The corresponding expression for the heat evolved [Q(i)] between two consecutive

injections i-1 and i is provided by the following equation:

8

Page 9: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

ΔQ ( i )=Q ( i )+ΔV i

V 0[ Q (i )+Q(i−1)

2 ]−Q(i−1)

Where Q(i) and Q(i-1) are the total heats evolved after i and i-1 injections, respectively,

and Vi is the volume of injection i. From this equation, the binding parameters (n, H

and Ka) are obtained by Marquadt methods and iteration.

Proteolysis by thermolysin

Thermolysin from Bacillus thermoproteolyticus rokko was purchased from

Sigma-Aldrich, buffer exchanged to 20 mM HEPES-OH, 200 mM NaCl pH 7.4 and

stored at -80 ºC (its protein concentration was measured using 280=66086 M-1·cm-1). For

SDS-PAGE analyses, thermolysin and GALE enzymes were customarily incubated for

10 min at 25 ºC in 20 mM HEPES-OH, 200 mM NaCl pH 7.4, in the absence or

presence of 1 mM NAD+ or UDP-glucose, and then mixed to a final concentration of 1

M and 20 M, respectively, in a final volume of 200 L. The time-course of

proteolysis was monitored upon withdrawal of 20 L of the reaction mixture, mixed

with 5 L of EDTA 100 mM pH 8 and denaturated in Laemmli’s buffer. Samples were

analyzed in 12% acrylamide gels and the bands were scanned and their intensities

integrated using the ImageJ software (http://rsbweb.nih.gov/ij/).

To identify the primary cleavage site in GALE, thermolysin and p.G90E GALE

enzymes were incubated for 10 min at 25 ºC in 20 mM HEPES-OH, 200 mM NaCl pH

7.4, and then mixed to a final concentration of 1 M and 10 M, respectively, in a final

volume of 1 mL. The reaction was allowed to proceed for 5 min at 25 ºC, quenched

adding 250 L of EDTA 100 mM pH 8. This mixture was buffer exchanged to water

using VIVAspin 500 filters (10 kDa cut-off) and then concentrated to 200 L and

splitted in two aliquots: i) 100 L was denatured in Laemmli´s buffer and run in a 12%

9

Page 10: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

SDS-PAGE followed by electrofransference to a PVDF membrane. The membrane was

stained with Coomasie blue G-250 and the band corresponding to the 35 kDa cleaved

form were cut, destained and equilibrated in water for N-terminal sequencing by the

Edman´s method (performed at the service of Protein Chemistry, Centro de

Investigaciones Biológicas, Madrid, Spain). ii) 100 L was submitted for High

performance liquid chromatography/electrospray ionization mass spectrometry

(HPLC/ESI-MS) analyses. HPLC/ESI-MS was performed in a Acquity UPLC system

(Waters), using a gradient of water/formic acid (0.1%) and acetonitrile/formic acid

(0.1%) in a Acquity UPLC® BEH300 C4 column (2.1x50 mm; Waters) coupled to a Q-

TOF Synapt62 HDMS (Waters) (performed at the high-resolution mass spectrometry

unit, Centro de Instrumentacion Cientifica, Universidad de Granada).

Kinetic models for irreversible denaturation/proteolysis of GALE enzymes in the

absence or presence of a ligand

We considered a simple kinetic model to simulate the effect of ligands on the

rate of denaturation/proteolysis of GALE in which ligand-bound (GALE-L) and ligand

free (GALE) undergo irreversible denaturation/proteolysis with first-order rate

constants kF-L and kF respectively to yield the final state F:

GALE−L↔GALE+L

kE-L kE

F+ L F

In the presence of ligand the kinetic stability of GALE will shift from the stability of the

ligand-free protein (GALE) to the stability of the ligand-bound protein (GALE-L), and

the overall rate of denaturation will be determined by the rate constants of the ligand-

10

Page 11: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

free (kE) and ligand-bound protein (kE-L) to undergo denaturation or proteolysis as well

as the concentration of ligand-free and ligand-bound protein, according to this rate law:

d [GALE ]total

dt=−(k¿¿ E−L[GALE−L]+kE [GALE ])¿

For illustration, in these simulations the value of kE is arbitrarily fixed to 10 min-1, while

kE-L values used are 1, 0.1 or 0.01 min-1. Therefore, the ratio of kE/ kE-L but not the

absolute values of these constants, will determine the dependence of the kinetic

stabilization exerted by the ligands as a function of total ligand concentration.

The fraction of GALE protein as free and ligated protein as a function of free ligand

concentration ([Ligand]) are estimated from a binding polynomial considering one site

per GALE monomer, a Kd for the ligand of 1 M (Ka=106 M-1) and total GALE

concentration of 10 M in monomer. The binding polynomial (P) is given by:

P=1+Ka·[Ligand]. The protein concentration in each state is given by:

[ GALE ]= 1P

[ GALE−L ]=K a ·[ Ligand ]

P

Results

The effect of disease-causing mutations on GALE conformation and ligand binding

Figure 1A shows the location of the altered residues in p.N34S, p.G90E,

p.V94M and p.K161N on the crystal structure of GALE in the presence of NADH and

UDP-glucose (PDB:1EK6; [25]). N34, G90 and K161 cluster around the NAD+ binding

site, while V94 is next to the UDP-glucose binding site. Interestingly, p.G90E, p.V94M

11

Page 12: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

and p.K161N cause a dramatic decrease in catalytic efficiency, from 40-fold

(p.V94M) to 1000-2000-fold (p.G90E and p.K161N) [10, 12].

We have expressed WT and the four disease-causing mutants in E. coli, and the

purified proteins were analyzed by spectroscopic methods under native conditions. We

first noticed that the UV-visible absorption spectra of WT and p.V94M differed from

the other GALE enzymes (Figure 1B, upper panel). These spectral changes are

compatible with the presence of cofactor/substrates bound to these two variants and not

released along the purification process. In the case of WT GALE, intrinsic fluorescence

spectra also support the presence of bound cofactor (Figure 1B, lower panel), but in all

cases, the fluorescence spectra show a maximum at 325 nm, consistent with similar

tertiary structures. The overall secondary structure of GALE enzymes was evaluated by

far-UV CD spectroscopy using the K2D3 algorithm, showing similar content for all

variants (Figure 1C; The average for the five GALE enzymes -helical and -sheet

content were 31.31.2% and 22.71.2%, respectively). The tertiary structure of the

GALE enzyme was further investigated by Near-UV CD spectroscopy (Figure 1D),

showing similar signals in the aromatic region for all enzymes (250-300 nm) and some

signals in the 300-400 nm region in WT and p.V94M which may arise from bound

substrate and/or cofactor. Finally, we studied the hydrodynamic behavior of GALE

enzymes by dynamic light scattering (Figure 1E). WT GALE displayed a hydrodynamic

radius of 3.80.1 nm, consistent with a molecular size of 765 kDa (i.e, a dimer), and

very similar to the size estimated along GALE purification by SEC (~70 kDa). None of

the variant GALE enzymes showed significant changes in the hydrodynamic size, and

the high monodispersity found for all them suggested that the GALE dimer is the main

species under native conditions.

We have evaluated the binding affinity of GALE enzymes for NAD+ and UDP-

12

Page 13: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

glucose by isothermal titration calorimetry (see Figure 2A and 2C for representative

titrations). As isolated, WT GALE binds both ligands sub-stoichiometrically, suggesting

that the binding sites are already partially filled with these (or equivalent) ligands, in

agreement with the spectroscopic evidence shown in Figure 1B. Regarding GALE

variants, p.V94M and p.K161N show partially occupied binding sites (n0.4), while

ligand binding to p.N34S and p.G90E is almost stoichiometric (n0.8; Table 1). Thus,

the Kd values determined using these GALE enzymes must be considered as apparent

values, which enable comparison of the variants with a given ligand and both ligands

for a given enzyme. The binding affinity of WT for both substrates is moderate, with Kd

values around 1 M and 8 M, for NAD+ and UDP-glucose, respectively The disease-

associated variants display different effects on the binding of ligands (Table 1). All

mutants show similar affinity for UDP-glucose to WT GALE, with the exception of

p.K161N, which binds this ligand with about 10-fold higher affinity. Regarding NAD+

binding, the most significant changes are observed for p.K161N, which decreases the

binding affinity by 10-fold, in agreement with previous evidence [12], and p.G90E,

which does not show a noticeable binding signal (which could be explained with a

markedly decreased binding affinity and/or binding enthalpy). The low response of this

mutant to NAD+ regarding thermal stability and proteolysis kinetics further support a

low binding affinity in this mutant (see below, Figures 4 and 5).

The presence of partially saturated binding sites in some GALE enzymes as purified

prompted us to attempt removal of these pre-bound ligands by dilution of the samples

to favour ligand release by dialysis, SEC and dilution-concentration cycles (see

Supplementary Information for a detailed description of these methods and associated

results). The success in removing bound ligands was tested by direct ITC titrations with

NAD+ and UDP-glucose (see Table S1). Unfortunately, none of these procedures

13

Page 14: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

restored the full binding capacity of GALE enzymes, even though in some cases a slight

improvement is observed. The apparent Kd values obtained with or without using these

procedures were comparable, and support our conclusions regarding the mutational

effects on the apparent binding affinity for NAD+ and UDP-glucose (Table 1 and S1).

We must note all these procedures led to significant protein loss due to aggregation, and

SEC analyses suggested that dimer-monomer equilibrium is shifted towards the

monomer at very low protein concentrations (see Supplementary Information), which

implies that further dilution of GALE enzymes to facilitate ligand binding is leading to

irreversible denaturation of the enzyme due to the instability of monomeric GALE.

Thermal denaturation studies

We studied the thermal stability of GALE enzymes by differential scanning

calorimetry (DSC; Figure 3). Thermal scans of WT GALE showed an apparent single

transition with a Tm44 ºC and a denaturation enthalpy of 88 kcal·mol-1 (Figure 3A

and Table 2). In GALE variants, this apparent single transition seems to split into two

well resolved transitions, since the sum of their denaturation enthalpies (5410 and

368 kcal·mol-1, for the low and high temperature transitions; see Table 2) agree well

with the single transition observed for WT GALE. These results suggest that the single

transition in WT GALE may be composed of two overlapping transitions (with Tm

values of about 44 °C and 51 °C, see Figure 3A-C). The low temperature transition in

GALE mutants and the single transition in WT GALE are highly irreversible, scan-rate

and protein-concentration dependent (Figure 3B-G), indicating that this transition in

denaturation of GALE is under kinetic control and involves dimer dissociation [26]. The

high temperature transition may result from further denaturation of partially folded

monomers. p.N34S, p.G90E and p.K161N seem to decrease the Tm of the low

temperature transition by 4-7 ºC compared to WT GALE, while p.V94M shows little or

14

Page 15: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

no effect (Figure 3A and Table 2). Regarding the high temperature transition, only

p.K161N shows an effect; there is a 5 ºC increase in the Tm of this transition (Figure

3A). We must note that denaturation enthalpies are known to scale with the protein size

and the degree of denaturation upon heating [27]. For a protein of GALE’s size (354

residues, including the his-tag), a denaturation enthalpy of about 168 kcal·mol-1 is

expected at the Tm of WT GALE which is almost double the experimental value. This

suggests that the thermally denatured GALE proteins retain a significant amount of

residual structure. The presence of residual structure in the thermally denatured state of

GALE is supported by comparison of far-UV CD and fluorescence spectra at low (25

°C) and high (60 °C, at which both denaturation transitions have fully developed)

temperatures and in the presence of 6M guanidium hydrochloride at low temperature

(Figure S1).

Addition of a large excess of either GALE cofactor (NAD+) or substrate (UDP-

glucose) has remarkable effects on the denaturation of GALE enzymes (Figure 4).

Addition of NAD+ increased the Tm of the low temperature transition, leading to an

apparent single peak for WT and p.V94M (with a Tm of 50 ºC), while in p.N34S and

p.K161N two peaks were still well resolved, but the low Tm transition was shifted

upwards by 5-6 ºC. In the case of p.G90E, NAD+ up-shifts the low Tm transition by only

1 ºC, supporting the strong defect in this enzyme for NAD+ binding shown by ITC

(Figure 2 and Table 1). Addition of UDP-glucose always led to a two-peak profile, with

an up-shift in the low temperature transition from 2-3 ºC (WT and p.V94M) to 6-8 ºC

(p.N34S, p.G90E and p.K161N) and no clear effect on the high-temperature transitions.

Thus, the stabilizing effect of NAD+ and UDP-glucose on these GALE enzymes

correlates well with their corresponding ligand binding affinities. Since denaturation of

GALE is under kinetic control and involves dimer dissociation, these ligand effects

15

Page 16: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

must translate at least into kinetic stabilization of the GALE dimer in the presence of

these natural ligands.

Proteolysis of GALE enzymes

Proteolysis have been proven to be an insightful tool to evaluate mutational

effects on protein folding, stability and dynamics (e.g. [10, 28]). We have measured the

proteolysis kinetics of GALE enzymes by thermolysin under native conditions (at 25

ºC, well below the denaturation temperature shown by DSC; Figure 5A). Proteolysis

rates linearly depended on the protease concentration (Figure 5B-C), with a 5.8-fold

increase in the proteolysis rate constant from 0.2 to 1 M protease, demonstrating that,

under these conditions, the proteolysis step is rate-limiting [28, 29]. All the disease-

associated variants show enhanced sensitivity to proteolysis, with half-lives lower than

that of WT GALE, from 1.4-fold (p.K161N), 2.3-fold (p.V94M) to 15-fold (p.N34S and

p.G90E) (Figure 5C-G and Table 3). Interestingly, these results correlate well with the

Tm values for the low-temperature transition obtained from DSC (Figure S2). Since the

proteolysis step is rate-limiting, an interesting possibility is that thermal denaturation

and proteolysis experiments are correlated because native state local

flexibility/dynamics and stability towards partial unfolding are linked.

In the presence of NAD+, WT GALE is degraded 1.6-fold slower (Figure 6C and

Table 3). All the variants show stabilization upon NAD+ binding, ranging from 1.8-fold

(p.G90E) to 18-fold (p.N34S) (Table 3 and Figure 6D-G). In the presence of UDP-

glucose, WT GALE is degraded 1.9-fold slower (Figure 6C and Table 3). All the

mutants show stabilization upon UDP-glucose binding, ranging from 2.0-fold (p.V94M)

to 10-13-fold (p.K161N and p.N34S) (Table 3 and Figure 6-D). The lower stabilizing

effect of these ligands on WT GALE compared to some mutants might be explained the

16

Page 17: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

partial saturation of these binding sites in the WT protein in the absence of exogenously

added NAD+ or UDP-glucose (Table 1).

Beyond the effects on the sensitivity of GALE native state towards proteolysis,

disease-associated variants and ligands have significant effects on the partial proteolysis

pattern of GALE (see Figure 5A and 6). Proteolysis of WT GALE showed the

accumulation of a 35 kDa band in the absence of ligand, which comigrates with the

thermolysin but its intensity is stronger and time-dependent, supporting the

accumulation of a partially proteolyzed form of GALE (Figure 5A and 6A).

Interestingly, the presence of NAD+ decreased the intensity of this band to values

corresponding to those of thermolysin (Figure 6B). In the presence of UDP-glucose, this

partially cleaved state is populated to a larger extent (Figure 5A and 6C), suggesting

that UDP-glucose binding stabilizes this partially cleaved state towards proteolytic

attack. All these results can be qualitatively explained by using a simple two-step

mechanism as the following:

where proteolysis of native GALE (N) leads to the formation of a partially proteolyzed

state of 35 kDa (I, determined by a rate constant k1), and this state is further cleaved to

low molecular mass fragments (P, with a rate constant k2). In the absence of ligands, k2

~k1, leading to low population of the intermediate I state in WT GALE (Figure 6A). In

the presence of NAD+, the N state is kinetically stabilized to a larger extent than the I

state, leading to very low population of I (i.e. k2 >k1) (Figure 6B). In the presence of

UDP-glucose, the I state is stabilized to a larger extent than the N state (i.e. k2 <k1),

leading to accumulation of I state (indeed, the sum of N and I over two hours of

proteolysis almost equal the initial load of GALE WT; Figure 6C). This simple model

17

Page 18: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

also explains the behavior found for variant GALE enzymes. For instance, p.N34S and

p.G90E strongly destabilize the N state towards proteolysis (i.e. k2 <k1) and show a

larger accumulation of the I state (Figure 6D and 6G). Addition of NAD+ stabilizes

largely p.N34S but not p.G90E, thus causing accumulation of the I state only for

p.G90E (Figure 6E and 6H). Consistently, addition of UDP-glucose stabilized the N

state of both p.N34S and p.G90E, thus leading to accumulation of the I state for both

GALE enzymes (Figure 6F and 6I) (i.e. k2 <k1). Although this model is likely to present

a simplified picture of GALE proteolysis kinetics, it nevertheless explains the

accumulation of the I state due to effects on the stability of the N state (e.g. by

mutations or NAD+ binding) or the stabilization of the I state (e.g. by UDP-glucose

binding).

To identify the primary cleavage site of GALE by thermolysin, we used GALE

p.G90E and a 5 min digestion time, in order to minimize proteolysis of secondary

cleavage sites and to form a significant amount of the 35 kDa cleavage product (about

20% of initial GALE). HPLC/ESI-MS analyses of this proteolysis mixture provided two

main forms, with a mass of 39116.4 Da (corresponding to intact GALE; theoretical

mass 39104.5) and 34329 Da (corresponding to the 35 kDa cleavage product

identified by SDS-PAGE). Among the 105 theoretical cleavage sites in GALE, these

results are consistent with cleavage between Ala38 and Phe39 cleavage, which would

release a fragment with a theoretical mass of 34300 Da. N-terminal sequencing of this

cleavage product showed the sequence Phe-Arg-Gly-Gly, confirming the cleavage

between Ala38 and Phe39. The primary cleavage site is located in a highly solvent

exposed loop (residues 34-45) at the N-terminal domain (Figure 7A and B). This loop

seems to be quite flexible based on the comparatively high B-factors determined from

the crystal structure (Figure 7C). Thus, it is likely that thermolysin cleaves at this site in

18

Page 19: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

the native state without requiring a local or global unfolding event. Thus, the sensitivity

of the native state towards proteolysis and the mutational- and ligand-effects must be

explained by changes in native state conformational dynamics.

Kinetic modeling supports that pre-bound ligands do not account for the

mutational and ligand effects on thermal stability and proteolysis sensitivity

In the previous section, we have shown that disease-causing GALE mutations affect

protein thermal stability and sensitivity to proteolysis, while NAD+ and UDP-glucose

binding stabilize GALE enzymes. However, WT GALE and to lesser extent p.V94M

and p.K161N, have significant amounts of pre-bound ligands as purified, and our

attempts to remove them were unsuccessful. Interestingly, the dependence of thermal

stability and proteolysis of purified GALE enzymes over a wide range of ligand

concentration show that the addition of stoichiometric amounts of ligand only slightly

affect thermal stability and sensitivity to proteolysis (Figure S3; note that the changes in

Tm exponentially translate into kinetic stabilization), while an large excess of ligand

stabilize to a larger extent. To test whether pre-bound ligands might explain the

differences in stability between GALE variants, we have evaluated the effect of ligands

(with the same binding affinity as NAD+) on the rate of denaturation/proteolysis by

kinetic model simulations.

To do so, we have evaluated the fraction of GALE monomer found in a

unligated- and ligated-state as a function of total ligand concentration (Kd=1 M; Figure

8A). Then, we determined the initial rate of denaturation/proteolysis using a model in

which ligated and unligated GALE undergo the irreversible process with different rate

constants kE-L and kE, respectively, and using different ratios between these two rate

constants (Figure 8B), observing two clearly different regimes: i) at total ligand

19

Page 20: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

concentrations equal or lower than the monomer concentration, there is a sharp decrease

in the fraction of unligated species as total ligand concentration is raised, leading to a

several-fold kinetic stabilization. In this regime, the extent of kinetic stabilization

isquite insensitive to the ratio of the ratio kE/kE-L (Figure 8C); ii) at higher total ligand

concentrations, further kinetic stabilization is observed as the total ligand concentration

is raised. In this regime, a small change in the saturation fraction (for instance, from

0.99 to 0.999) causes a large kinetic stabilization (up to a 10-fold increase at very high

kE/kE-L ratios ) due to the decrease in the fraction of kinetically sensitive unligated-

species (from 0.01 to 0.001, in this particular example). Therefore the dependence of the

kinetic stabilization on total ligand concentration clearly depends on the kE/kE-L value at

very high ligand concentrations (see the different saturation behaviour in Figure 8C).

Then, we considered saturation values consistent with the prebound-ligand

fraction determined for WT (0.8), p.V94M and p.K161N (0.5) and p.N34S and p.G90E

(0.2) (Figure 8B). The corresponding initial rates towards irreversible denaturation for

these saturation fractions are 2.590.08 (WT), 1.700.02 (p.V94M and p.K161N) and

1.240.01 (p.N34S and p.G90E) fold lower than the rate of the corresponding unligated

species (means.d. using three different kE/kE-L ratios, showing the rate is independent

of this ratio at the low ligand concentration regime). These analyses have two important

implications: i) prebound-ligands might exert up to a three-fold stabilization in WT vs.

GALE variants, which is much lower than fifteen-fold difference obtained

experimentally for p.N34S and p.G90E (Table 1). Thus, the pre-bound ligands can only

explain a small fraction of the higher stability of WT towards proteolysis (and possibly

thermal denaturation as judged by Tm(1) values) without added ligands; ii) the higher

stabilizing effect of added ligands towards proteolysis (and thermal stability) in some

GALE variants might imply different values of the ratio kE/kE-L from those of WT

20

Page 21: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

GALE, since these experiments are performed with 1 mM ligand (the high ligand

concentration regime) and these ratios determine the saturation behaviour of the kinetic

stabilization vs. total free ligand concentration (Figure 8C). Thus, these simulations may

also explain how the different ligand concentration dependence of proteolysis rates

constants and Tm values for thermal denaturation between GALE variants (Figure S3)

together with changes in ligand binding affinities.

Discussion

The role of protein stability, conformational dynamics and ligand binding have

been investigated here for WT GALE and four variants associated with type III

galactosemia. Our results show that GALE dimer is marginally stable and partially

denatures at temperatures close to physiological. p.N34S, p.G90E and p.K161N may

further destabilize GALE dimer towards partial denaturation, thus rendering a

kinetically unstable protein at physiological temperature. Interestingly, p.N34S and

p.G90E are also much more sensitive to proteolysis than WT GALE, indicating greatly

altered local conformational dynamics in the N-terminal region of these two variants. In

contrast, p.V94M shows similar resistance to proteolysis compared to WT. This is

consistent with molecular dynamics (MD) simulations of this variant which predicted

little change in the global flexibility of the protein [13] and previous experimental

studies [10, 30]. In all cases, substrate and/or cofactor binding modulate the sensitivity

of GALE enzymes towards thermal denaturation and proteolysis, suggesting that

alterations in dimer stability and local dynamics by disease-causing mutants can be

efficiently modulated by binding to either natural or pharmacological ligands. These

effects were comparable between WT and p.V94M, again consistent with MD work

which predicted little or no change in binding affinity of this variant for the substrate

and the cofactor [13]. The results are also consistent with previous biochemical work

21

Page 22: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

which showed that p.G90E is one of the most unstable of the currently known variants

[10]. This suggests that this variant could be associated with very severe forms of the

disease; however, to date, it has only been found in a heterozygous patient [31]. In a

diploid Saccharomyces cerevisiae model deleted for both copies of the yeast GALE

(GAL10) and heterozygous for human WT and p.G90E, GALE activity was

approximately 50% of that detected in strains homozygous for WT human GALE [11].

This suggests that the G90E allele is recessive to the wild-type one, possibly due to a

combination of largely impaired catalytic function, conformational instability and

altered local conformational dynamics.

Our proteolysis analyses provide an insight into the mutational and ligand effects

on GALE protein dynamics. p.N34S and p.G90E locally alter flexibility or dynamics in

the N-terminal domain most likely due to repulsive/destabilizing interactions with the

loop 34-45. In the crystal structure of human GALE (PDB:1EK6), Asn-34 lies adjacent

to the adenine moiety of NAD+ [25]. Asn-34 is predicted to hydrogen bond with NAD+

and is close in sequence to other residues (i.e. Asp33 and Asn37) which interact with

the cofactor [19, 25] which may explain the 4-fold decreased affinity for NAD+ (Table

1). In general, serine residues tend increase the local backbone flexibility of the

polypeptide chain and it would be expected that increased flexibility in this region

would also weaken these protein-cofactor interactions [32]. G90 is located close to N34

(<1.5 Å) in the structure of human GALE and adjacent to the phosphate groups in

NAD+ [25]. All together, these structural analyses may explain the very low affinity of

p.G90E for NAD+ (indeed, no binding signal was detected by ITC; Figure 2A and B).

The replacement of glycine by the much larger glutamate side chain will require

considerable rearrangement and, probably, destabilisation and distortion of the local

structure, especially of the sequence connecting -sheet 4 and helix-4 where G90 is

22

Page 23: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

located. This may be exacerbated by repulsive interactions between the negatively

charged side chain and the phosphate moieties in NAD+. Thus, both p.N34S and p.G90E

are predicted to result in local changes to the structure and flexibility of the protein. This

may also explain why these two mutations cause accumulation of the partially cleaved

state I (lacking the N-terminal 38 residues) due to selective destabilization of the native

state, probably because the effects of these two mutations are weakened or eliminated

when the N-terminal 38 residues are removed (Figure 6A, D and G).

NAD+ and UDP-glucose binding increase the resistance of the native state

towards proteolysis and this effect is mutant-selective, corresponding fairly well with

the affinity of their native state for these ligands (Tables 1 and 3), which also supports

the proposition that binding of a ligand (UDP-glucose) far from the cleavage site have

long-range effects that propagate between domains in the native state. Interestingly, the

two ligands affect differently the accumulation of the partially cleaved state (Figure 6),

which can be reasonably explained based on the structure of GALE in the presence of

substrate and cofactor [25]. The cofactor binds to the N-terminal domain and establishes

hydrogen bonds with residues close to the 34-45 loop (such as Asp33 and Asn37 and

Lys161). However, upon proteolytic attack on this primary site, most of these

favourable interactions are removed, thus leading to the release of the bound cofactor or

substantially decreased binding affinity for it, but not affecting the stability of the

cleavage intermediate I state. Accordingly, cleavage of the loop 34-45 might not greatly

affect binding of UDP-glucose which binds to the C-terminal domain, and thus,

resulting in the stabilizing effects of UDP-glucose on the cleavage kinetics of the

intermediate and its accumulation along the proteolysis reaction (Figure 6C, F and I).

To our knowledge, we provide the first evidence for long-range communication

between domains in GALE structure (upon mutation and ligand binding), which allow

23

Page 24: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

further understanding on the conformational consequences of disease-associated

variants and might be therapeutically exploited to correct mutation-induced

destabilization in type III galactosemia.

We must note that GALE enzymes as purified seem to have partially occupied

binding sites for NAD+ and UDP-glucose (this work and [12]), especially for WT,

p.V94M and p.K161N enzymes (Table 1). Attempts to remove by different procedure

were unsuccessful, probably due to a shift in the dimer-monomer equilibrium towards

monomers causing GALE destabilization and aggregation (see Supplementary

Information and Table S1). The presence of pre-bound ligands will influence thermal

stability and proteolysis results beyond the effects of mutations. However, we provide

experimental (Figure S3) and theoretical (Figure 8) evidence strongly supporting a

minor role of pre-bound ligands in the mutational and NAD+ and UDP-glucose effects

described here. This is a logical consequence of the kinetic control of thermal

denaturation and proteolysis and the comparatively low affinity for the ligands (see [22,

26, 33]). In conclusion, the differences observed in stability for GALE variants must

reflect to a large extent true effects of the disease-causing mutations.

Taken together, our data strongly support the conclusion that the sequence

alteration results in destabilization and altered conformational dynamics of GALE dimer

at physiological conditions (temperature and pH) which are linked to loss of enzyme

function in type III galactosemia. We also demonstrate that ligand and cofactor binding

trigger large changes in global dimer stability and local conformational dynamics. We

hypothesise that these changes might modulate GALE intracellular protein turnover and

functionality, as shown for other conformational diseases [17, 34-36]. These results

further support the previous proposal for the use of ligands, such as cofactor analogues

or pharmacological chaperones, as potential approaches to treat type III galactosemia

24

Page 25: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

[19]. Critically, our finding that the natural ligands can stabilise different states of the

protein supports the idea that pharmacological ligands which reduce the rate of the N→I

transition are likely to be effective in rescuing GALE activity. Furthermore, the

approaches taken here suggest methods by which such compounds could be identified

through high-throughput screening. After initial identification of protein binders, these

hits could be used in a limited proteolysis assay similar to that used in this work in order

to identify the subset that reduces the rate constant of the first transition.

Acknowledgements.- We thank Dr. Jose Manuel Sanchez-Ruiz for support. This work

was supported by grants from MINECO (BIO2012-34937 and CSD2009-00088), Junta

de Andalucia (P11-CTS-07187), The Royal Society (2004/R1) and FEDER Funds. A.L.P.

is supported by a Ramón y Cajal research contract from MINECO (RyC-2009-04147).

N.M-T. is supported by a FPI predoctoral fellowship from MINECO.

Abbreviations: GALE: UDP-galactose 4´-epimerase; UDP-glc: UDP-glucose;

HPLC/ESI-MS: high performance liquid chromatography/electrospray ionization mass

spectrometry; MD: molecular dynamics; CD.- circular dichroism; DSC.- differential

scanning calorimetry.

25

Page 26: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

References

[1] H.M. Holden, I. Rayment, J.B. Thoden, J Biol Chem 278 (2003) 43885-43888.[2] D.J. Timson, IUBMB Life 58 (2006) 83-89.[3] J.L. Fridovich-Keil, J.H. Walter, in: D. Valle, A. Beaudet, B. Vogelstein, K.W. Kinzler, S.E. Antonarakis, A. Ballabio (Eds.), The Online Metabolic and Molecular Bases of Inherited Diseases, McGraw-Hill, New York, 2008.[4] T.J. McCorvie, D.J. Timson, in: N. Taniguchi, K. Honke, M. Fukuda, H. Narimatsu, Y. Yamaguchi, T. Angata (Eds.), Handbook of Glycosyltransferases and Related Genes, Springer, 2014.[5] K.L. Kavanagh, H. Jornvall, B. Persson, U. Oppermann, Cell Mol Life Sci 65 (2008) 3895-3906.[6] J. Axelrod, H.M. Kalckar, E.S. Maxwell, J.L. Strominger, J Biol Chem 224 (1957) 79-90.[7] U.S. Maitra, H. Ankel, Proc Natl Acad Sci U S A 68 (1971) 2660-2663.[8] F. Piller, M.H. Hanlon, R.L. Hill, J Biol Chem 258 (1983) 10774-10778.[9] J.M. Daenzer, R.D. Sanders, D. Hang, J.L. Fridovich-Keil, PLoS Genet 8 (2012) e1002721.[10] D.J. Timson, FEBS J 272 (2005) 6170-6177.[11] T.M. Wohlers, N.C. Christacos, M.T. Harreman, J.L. Fridovich-Keil, Am J Hum Genet 64 (1999) 462-470.[12] T.J. McCorvie, Y. Liu, A. Frazer, T.J. Gleason, J.L. Fridovich-Keil, D.J. Timson, Biochim Biophys Acta 1822 (2012) 1516-1526.[13] D.J. Timson, S. Lindert, Gene 526 (2013) 318-324.[14] J.S. Chhay, C.A. Vargas, T.J. McCorvie, J.L. Fridovich-Keil, D.J. Timson, J Inherit Metab Dis 31 (2008) 108-116.[15] S.E. Antonarakis, Hum Mutat 11 (1998) 1-3.[16] B.B. Quimby, A. Alano, S. Almashanu, A.M. DeSandro, T.M. Cowan, J.L. Fridovich-Keil, Am J Hum Genet 61 (1997) 590-598.[17] A.L. Pey, Amino Acids 45 (2013) 1331-1341.[18] A.L. Pey, M. Ying, N. Cremades, A. Velazquez-Campoy, T. Scherer, B. Thony, J. Sancho, A. Martinez, J Clin Invest 118 (2008) 2858-2867.[19] T.J. McCorvie, D.J. Timson, Gene 524 (2013) 95-104.[20] C.N. Pace, F. Vajdos, L. Fee, G. Grimsley, T. Gray, Protein Sci 4 (1995) 2411-2423.[21] C. Louis-Jeune, M.A. Andrade-Navarro, C. Perez-Iratxeta, Proteins 80 (2012) 374-381.[22] A.L. Pey, T. Majtan, J.M. Sanchez-Ruiz, J.P. Kraus, Biochem J 449 (2013) 109-121.[23] D. Rodriguez-Larrea, S. Minning, T.V. Borchert, J.M. Sanchez-Ruiz, J Mol Biol 360 (2006) 715-724.[24] J.M. Sanchez-Ruiz, J.L. Lopez-Lacomba, P.L. Mateo, M. Vilanova, M.A. Serra, F.X. Aviles, Eur J Biochem 176 (1988) 225-230.[25] J.B. Thoden, T.M. Wohlers, J.L. Fridovich-Keil, H.M. Holden, Biochemistry 39 (2000) 5691-5701.[26] J.M. Sanchez-Ruiz, Biophys J 61 (1992) 921-935.[27] A.D. Robertson, K.P. Murphy, Chem Rev 97 (1997) 1251-1268.[28] A.L. Pey, Biochim Biophys Acta 1834 (2013) 2502-2511.

26

Page 27: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

[29] C. Park, S. Marqusee, J Mol Biol 343 (2004) 1467-1476.[30] T.M. Wohlers, J.L. Fridovich-Keil, J Inherit Metab Dis 23 (2000) 713-729.[31] P. Maceratesi, N. Daude, B. Dallapiccola, G. Novelli, R. Allen, Y. Okano, J. Reichardt, Mol Genet Metab 63 (1998) 26-30.[32] J.L. Goodman, M.D. Pagel, M.J. Stone, J Mol Biol 295 (2000) 963-978.[33] F. Conejero-Lara, P.L. Mateo, F.X. Aviles, J.M. Sanchez-Ruiz, Biochemistry 30 (1991) 2067-2072.[34] A. Martinez, A.C. Calvo, K. Teigen, A.L. Pey, Prog Mol Biol Transl Sci 83 (2008) 89-134.[35] J.V. Rodrigues, B.J. Henriques, T.G. Lucas, C.M. Gomes, Curr Top Med Chem 12 (2012) 2546-2559.[36] S. Santos-Sierra, J. Kirchmair, A.M. Perna, D. Reiss, K. Kemter, W. Roschinger, H. Glossmann, S.W. Gersting, A.C. Muntau, G. Wolber, F.B. Lagler, Hum Mol Genet 21 (2012) 1877-1887.[37] A. Shrake, J.A. Rupley, J Mol Biol 79 (1973) 351-371.[38] C. Chothia, J Mol Biol 105 (1976) 1-12.[39] Collaborative Computational project number 4, Acta Crystallogr D Biol Crystallogr 50 (1994) 760-763.

27

Page 28: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Table 1. Binding properties of NAD+ and UDP-glucose to GALE enzymes

determined by ITC. Data are best-fit values and their corresponding fitting errors.

GALE

enzyme

NAD+ UDP-glucose

n Kd

(M)

H

(kcal/mol)

n Kd

(M)

H

(kcal/mol)

WT 0.2

30.01

1.00.1 -9.60.4 0.2

00.05

7.91.1 -17.45.2

p.N34S 0.7

80.01

3.80.2 -11.70.3 0.6

60.02

3.90.4 -21.00.9

p.G90E Not detected 0.8

30.03

9.30.6 -14.90.7

p.V94M 0.3

60.01

0.3

50.01

-11.60.1 0.4

40.08

10.31.6 -8.91.9

p.K161N 0.4

20.03

121 -10.91.0 0.4

90.01

0.7

50.06

-22.30.3

28

Page 29: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Table 2. Apparent Tm values and denaturation enthalpies (H) for thermal

denaturation of GALE enzymes. Experiments performed at 3oC·min-1 and 10 M

protein in monomer. Parameters are obtained from fittings to a two-state irreversible

model and errors are those from fittings.

GALE

enzyme

Parameter

Tm1 (°C) H1 (kcal·mol-1) Tm2 (°C) H2 (kcal·mol-1)

WT 44.10.1 883 Not applicable

p.N34S 39.60.1 451 49.60.1 422

p.G90E 38.80.1 501 52.00.1 301

p.V94M 42.30.1 542 49.10.1 456

p.K161N 37.00.1 681 55.80.1 291

29

Page 30: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Table 3. Half-lives for proteolysis of native GALE enzymes in the absence or

presence of ligands (1 mM). The concentration of GALE enzymes and thermolysin

was 20 M and 1 M, respectively. Experiments were performed at 25 ºC.

GALE enzyme Half-life (min)

No ligand NAD+ UDP-glucose

WT 595 947 11411

p.N34S 3.90.3 726 515

p.G90E 4.00.2 7.30.3 211

p.V94M 264 13917 537

p.K161N 8.02.0 334 796

30

Page 31: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Figure 1. Mutational effects on GALE conformation. A) Location of the altered

residues in the crystal structure of GALE (PDB code: 1EK6; [25]) indicated as red

sticks. UDP-glucose is shown in black and NADH in magenta. GALE monomers are

displayed in different colors (green and cyan). B) UV-visible absorption spectra (lower

panel; inset show a magnified view of the absorption spectra between 300 and 400 nm)

and intrinsic fluorescence spectra (upper panel, excitation at 280 nm; inset: excitation at

295 nm) at 4 M protein in GALE monomer; C) Far-UV CD spectra at 4 M; D)

Near-UV CD at 20 M; E) DLS at 20 M.

31

Page 32: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Figure 2. Ligand binding to GALE enzymes by ITC. Thermograms and binding

isotherms for the binding of NAD+ (panel A and B) and UDP-glucose (panel C and D)

to GALE enzymes. Symbols in panels B and D are as follows: WT (circles), p.N34S

(up-triangles), p.G90E (down-triangles), p.V94M (squares), p.K161N (diamonds).

32

Page 33: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Figure 3. Thermal denaturation of GALE enzymes by differential scanning

calorimetry (DSC). A) Denaturation profiles of WT and variant GALE enzymes at 3

ºC·min-1 and 10 M protein (monomer equivalent). B and E) Reversibility tests on WT

(B) and p.N34S (E) GALE. Protein samples (10 M protein monomer equivalent)

were heated up to 57 ºC or 47-48 ºC, cooled down to 2 ºC and rescanned. A scan rate of

3 ºC/min was used. C and F) scan rate dependence of thermal denaturation of WT (C)

and p.N34S (F) GALE (10 M protein monomer equivalent); Scan rates were 3 °C min-

1 (closed circles), 2 °C ·min-1 (open down triangles) or 1 °C·min-1 (closed squares); D

and G) protein concentration dependence of thermal denaturation of WT (D) and

p.N34S (G) GALE (3 ºC/min scan rate). Protein concentrations were 20 M (closed

circles), 10 M (open down triangles) and 5 M (closed squares).

33

Page 34: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

34

Page 35: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Figure 4. Effect of ligands (NAD+ or UDP-glucose, 1 mM) on the thermal

denaturation of GALE variants by DSC. Protein concentration was 10 M, and

experiments were performed in 20 mM HEPES-OH, 200 mM NaCl pH 7.4 at a scan

rate of 3 ºC·min-1.

35

Page 36: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Figure 5. Kinetics of proteolysis of GALE enzymes by thermolysin. A)

Representative SDS-PAGE gels for WT (upper panel) and p.G90E (lower panel). B-G)

Kinetics of proteolysis of GALE enzymes in the absence (closed circles) or presence of

1 mM ligands (NAD+, open circles; UDP-glucose, closed down-triangles). GALE

enzymes were at 20 M and thermolysin at 1 M (panels C-G) or 0.2 M (panel B).

Experiments were performed at 25 ºC. Lines in panels B-G are fits to a single

exponential function.

36

Page 37: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

37

Page 38: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Figure 6. Time dependent population of the full-length GALE (closed circles), the

35 kDa band (open circles) and sum of them (open stars) from SDS-PAGE

analyses. The horizontal dashed line shows the fraction that corresponds to the

contribution of thermolysin to the 35 kDa. Fractions were determined by

normalization using the band intensity of full-length GALE without thermolysin (t=0

min).

38

Page 39: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Figure 7. Conformation of the primary cleavage site of GALE enzymes by

thermolysin (loop 34-45). A) Structure of GALE dimer (PDB code: 1EK6)

highlighting the solvent exposed loop 34-45 and the primary cleavage site between

Ala38 and Phe39 (in black); B) Side-chain accessibility of the loop 34-45 and adjacent

residues calculated using the Shrake-Rupley algorithm [37] with a radius of 1.4 Å for

the solvent probe and the Chothia set for the protein atoms [38]. The asterisks indicate

glycine (no side-chain, i.e. 0% accessibility). C) Average-B factors calculated from the

GALE crystal structure (PDB code: 1EK6) using the program Baverage from CCP4

Suite [39].

39

Page 40: pure.qub.ac.uk€¦ · Web viewcDNAs. An overnight culture grown in Luria–Bertani media supplemented with 0.1 mg/ml ampicillin was diluted 1:20 into 1 L of Luria–Bertani media

Figure 8. Kinetic modeling of GALE stability in the absence or presence of bound

ligands. A) Dependence of the concentration of unligated (GALE) and ligated-GALE

(GALE-L) on total ligand concentration; B) Dependence of the initial rate for

irreversible denaturation/proteolysis as a function of total ligand concentration using a

rate constant kE=10 min-1and different values of kE-L (from 1 to 0.01 min-1). The symbols

indicate the initial rates considering saturation fractions of 0.8 (WT), 0.5 (p.V94M and

p.K161N) and 0.2 (p.N34S and p.G90E) derived from pre-bound ligand values

estimated from ITC titrations; C) Dependence of kinetic stabilization (rate with

ligand/rate without ligand) on total ligand concentration using a rate constant kE=10

min-1 and different values of kE-L (from 1 to 0.01 min-1). All calculation are performed

using 10 M GALE monomer and Kd of 1 M (similar to NAD+).

40