Báo cáo khoa học: Effects of replacing active site residues in a cold-active alkaline phosphatase with those found in its mesophilic counterpart from Escherichia coli

Effect of heat on secondary structure stability of the wild-type Vibrio AP and the H116D, W274K and H116D ⁄ W274K mutants.. coli enzyme also had the lowest Ki for inorganic phosphate as

alkaline phosphatase with those found in its mesophilic counterpart from Escherichia coli

Katrı´n Gudjo´nsdo´ttir and Bjarni A´ sgeirsson

Department of Biochemistry, Science Institute, University of Iceland, Reykjavik, Iceland

Alkaline phosphatase (AP; EC 3.1.3.1) is a nonspecific

catalyst for the hydrolysis or transesterification of

phosphoryl esters, and usually functions best in an

alkaline environment [1] AP from Escherichia coli has

been most extensively studied [2], and its

three-dimen-sional structure was the first to be determined [3,4]

Various mutants of the E coli enzyme have been made

[5–14], and crystal structures for many of these are

known To date, structures of APs from only three

other species have been solved, namely human

placen-tal AP [15], shrimp AP [16], and AP from an Antarctic

bacterium [17] Sequence comparisons of APs from a

variety of species combined with structural information

suggest that functionally important domains are

mostly conserved [3] The residues that directly react with the substrate, Ser102 and Arg166, are conserved

in all cases In addition, there are three metal-binding sites, M1, M2, and M3, containing two Zn2+ (occupy-ing M1 and M2) and one Mg2+(in M3), in the active site of the E coli enzyme [3] All three metal ions par-ticipate in the reaction mechanism [18] Amino acid residues that interact with the zinc ions in the M1 and M2 sites of APs are conserved in all known sequences However, variations occur at amino acids Asp153 and Lys328 near the Mg2+-binding site (M3) of the E coli

AP (Fig 1A) The only change observed at posi-tion 153 in other variants is from Asp to His, whereas position 328 is commonly changed from Lys to either

Keywords

cold adaptation; metalloenzyme;

mutagenesis; protein stability; psychrophilic

Correspondence

B A ´ sgeirsson, Science Institute, University

of Iceland, Dunhaga 3, 107 Reykjavik,

Iceland

Fax: +354 552 89 11

Tel: +354 525 48 00

E-mail: bjarni@raunvis.hi.is

(Received 26 September 2007, revised

3 November 2007, accepted 7 November

2007)

doi:10.1111/j.1742-4658.2007.06182.x

Alkaline phosphatase (AP) from a North Atlantic marine Vibrio bacterium was previously characterized as being kinetically cold-adapted It is still unknown whether its characteristics originate locally in the active site or are linked to more general structural factors There are three metal-binding sites in the active site of APs, and all three metal ions participate in cataly-sis The amino acid residues that bind the two zinc ions most commonly present are conserved in all known APs In contrast, two of the residues that bind the third metal ion (numbered 153 and 328 in Escherichia coli AP) are different in various APs This may explain their different catalytic efficiencies, as the Mg2+ most often present there is important for both structural stability and the reaction mechanism We have mutated these key residues to the corresponding residues in E coli AP to obtain the dou-ble mutant Asp116⁄ Lys274, and both single mutants All these mutants displayed reduced substrate affinity and lower overall reaction rates The Lys274 and Asp116⁄ Lys274 mutants also displayed an increase in global heat stability, which may be due to the formation of a stabilizing salt bridge Overall, the results show that a single amino acid substitution in the active site is sufficient to alter the structural stability of the cold-active VibrioAP both locally and globally, and this influences kinetic properties

Abbreviations

AP, alkaline phosphatase; DG u , free energy of unfolding; pNPP, p-nitrophenyl phosphate; T m , melting temperature; T 50% , temperature required for the enzymes to lose half of the initial activity in 30 min.

Trp or His Asp153 in the E coli enzyme ligates the

Mg2+ and the substrate’s phosphoryl group through water molecules Lys328 has also been shown to be important for binding of phosphate in the active site

of the E coli enzyme through a water-mediated link [9,19] Furthermore, these two amino acids, Asp153 and Lys328, are connected by a salt bridge in the

E coli enzyme [3,9] It has also been shown that the type of amino acid at position 328 has an effect on which metal occupies the M3 site [10,11] and which metals the enzyme can use for catalysis [14,20] Experi-ments have shown that the presence of Mg2+in M3 in

E coli AP is important for full catalytic activity and for stability of the enzyme [9,21,22]

Information about the reaction mechanism of APs comes mostly from studies on the E coli enzyme, but

it is believed to be generally the same for all APs The reaction starts with a nucleophilic attack of a Ser102 alkoxide on the phosphate group of the monoester substrate, forming the covalent phospho-enzyme intermediate The second step is the hydrolysis

of this phosphoseryl intermediate to the noncovalent enzyme–phosphate complex [23] In the presence of a phosphate acceptor, such as Tris or diethanolamine, the enzyme also displays transphosphorylating activity,

an organic nucleophile replacing water [24] The rate-determining step of the E coli enzyme reaction is pH-dependent At acidic pH, the hydrolysis of the covalent phosphoseryl intermediate is the slowest step, but at basic pH, the dissociation of the phosphate from the noncovalent enzyme–phosphate complex becomes rate-determining [25,26] The type of amino acid at posi-tion 328 in the E coli AP has been shown to affect the affinity for inorganic phosphate [11,19] and thus the rate-determining step The E coli AP mutants K328H, K328A and K328C have been made and characterized [27,28] All of these mutants had an altered rate-deter-mining step as compared to the wild-type enzyme That is, at pH 8.0, the hydrolysis of the covalent phosphoseryl intermediate was the slowest step The mutants also had lower affinity for inorganic phos-phate

Psychrophilic organisms have adapted to low-tem-perature environments by making enzymes with bet-ter catalytic efficiency at low temperatures (higher

kcat or kcat⁄ Km) The common features of such cold-active enzymes as compared to their mesophilic counterparts are believed to be the consequence of enhanced structural flexibility, often leading to decreased thermostability [29–32] It is not clear whether increased flexibility is brought about by local weakening of stabilizing bonds around the active site, or by more general structural factors, and

A

B

Fig 1 (A) Active site of E coli AP and (B) active site of Vibrio

AP The residues near the Mg2+ ion in the third metal-binding

site of these two APs are shown The only variations between

the two species are seen in positions 328 and 153 (E coli

num-bering) Lys328 and Asp153 are Trp274 and His116 in the Vibrio

AP In the E coli enzyme, a salt bridge connects Lys328 and

Asp153 Lys328 has been shown to be important for phosphate

binding, and the type of amino acid in this position has been

shown to affect what metal occupies the M3 site and which

metals the enzyme can use for catalysis Residues that ligate

the two Zn 2+ are conserved in all APs and are not shown The

position of the substrate phosphate leaving group is shown, as

well as the Arg166 ⁄ 129 residue near the substrate-binding site.

The figures were generated using the PYMOL viewer The E coli

model was (A) is made using the Protein Data Bank file 1ALK.

The Vibrio model is a homology model made with SWISS -

PDB-VIEWER

it may be attained differently within different enzyme

families [32]

An AP was previously isolated from a costal water

Vibriobacterial strain that was found to be a

kinetical-ly cold-adapted 55 kDa monomer [33] Most other

known APs are dimers composed of identical subunits,

but other apparent monomeric APs have been

reported Examples include AP from Vibrio cholera

[34], the heat-labile APs from a Shewanella sp [35],

and the AP from an Antarctic bacterium [36] In

com-parison to the E coli AP, the Vibrio AP had higher

catalytic efficiency as well as considerably less

struc-tural stability, being sensitive to room temperature

The Asp153⁄ Lys328 pair in E coli AP, important for

substrate binding and the identity of the metal in M3

[11], is His116⁄ Trp274 in Vibrio AP (Fig 1B) This

may explain their different catalytic efficiencies We

have mutated Trp274 in Vibrio AP to Lys and His116

to Asp We have also made the double mutant

Asp116⁄ Lys274, which is an analog of the E coli

enzyme with regard to amino acids in the active site

Our aim was to determine whether these amino acids

take part in the cold adaptation of the Vibrio AP, and

we found they clearly affect both stability and activity

in a reciprocal manner

Results

Heat stability of the active site conformation

The rate of heat inactivation was determined at several

temperatures for wild-type Vibrio AP and the three

mutants The results for 35C and 45 C are shown in

Fig 2 Rate constants were determined and used to

obtain t1 ⁄ 2values (Table 1) Furthermore, the

tempera-ture required for the enzymes to lose half of the initial

activity in 30 min (T50%) was also calculated, and the

melting temperature (Tm) values were obtained by CD,

as described in the next section (Table 2) The

wild-type Vibrio AP was very sensitive to heat inactivation,

confirming our previous results [33] Thus, the t1⁄ 2 at

25C was only 94 min (Table 1), and T50% was

40.1C (Table 2) In contrast, all three mutants were

stable at 25C for at least 30 min (Table 1) At 55 C,

the double mutant H116D⁄ W274K lost half of its

ini-tial activity at 5 min, but was more stable than either

the W274K mutant (t1⁄ 2 of 1.2 min) or the H116D

mutant (t1 ⁄ 2 of 0.2 min) Interestingly, the H116D

mutant was slightly less heat-stable than the wild-type

in terms of the T50% values (38.5C versus 40.1 C),

whereas both the W274K and H116D⁄ W274K mutants

were more heat-tolerant than the wild-type enzyme

(Table 2)

Effect of heat on secondary structure The effect of heat on the secondary structures of the wild-type and mutant Vibrio APs was measured by CD

at several temperatures Figure 3 shows CD spectra of the wild-type enzyme at temperatures from 0C to

90C Changes in secondary structures during heating from 15C to 90 C were further monitored at 222 nm for the wild-type and all the mutants (Fig 4) From these data, Tm values were determined (Table 2) The double mutant had the highest Tm value, at 58.6C The wild-type had a Tmof 56.5C, slightly higher than that of the H116D mutant, which had a Tmof 55.0C The value for the W274K mutant was 57.2C

–2 –1.5 –1 –0.5 0

–2 –1.5 –1 –0.5 0

35 °C

Time (min)

45 °C

5

Fig 2 Heat inactivation of wild-type Vibrio AP (black squares), the H116D mutant (open squares), the W274K mutant (black circles) and the H116D ⁄ W274K double mutant (open circles) at 35 C (upper panel) or 45 C (lower panel) Samples were incubated at the indicated temperatures, and residual activity was measured at intervals of a few minutes using the standard enzyme assay.

m ⁄ m 0 = relative activity.

Urea denaturation of wild-type Vibrio AP and free energy of unfolding

The effect of urea on wild-type Vibrio AP activity and tertiary structure was determined by p-nitrophenyl phosphate (pNPP) hydrolysis and fluorescence mea-surements (Fig 5) The results showed that loss of activity occurred at urea concentrations below 1 m, before any change was detected in the global protein structure as monitored by tryptophan fluorescence The global structure was unfolded at urea concentra-tions in the range 1–3 m ([urea]1⁄ 2= 1.7 m)

Free energy of unfolding (DGu) was calculated from these data using the linear extrapolation method [37]

DGu for unfolding (measured by fluorescence) was 3.2 kcalÆmol)1, but it was 2.0 kcalÆmol)1for the inacti-vation transition

Zn2+content and release from active site

of wild-type Vibrio AP The metal content of wild-type Vibrio AP was analyzed

by inductively coupled plasma MS and atomic absorp-tion Only Zn was consistently found in the protein peak in significant amounts by MS after desalting by column chromatography Very insignificant amounts of tin, cadmium, nickel, vanadium and cobalt were incon-sistently seen coeluting with the protein peak in the var-ious batches of samples that were analyzed The molar ratio of metals in a monomeric Vibrio AP determined

Table 2 Heat stability of the Vibiro AP mutants T m values were

determined by measuring CD at 222 nm with continuous heating

from 15 C to 90 C at 1 CÆmin)1 T50%values were determined

by following inactivation at various temperatures, and refer to the

temperature at which the enzyme loses 50% of initial activity at

30 min (n = 3).

H116D ⁄ W274K mutant 58.6 ± 0.2 47.0 ± 0.2

Wild-type Vibrio AP 56.5 ± 0.2 40.1 ± 0.2

–15

–10

–5

0

0 °C

20 °C

40 °C

50 °C

60 °C

80 °C

λ (nm)

Fig 3 Heat-induced unfolding of wild-type Vibrio AP secondary

structures monitored by CD measurements The enzyme was heated

from 0 C to 90 C at a rate of 1 CÆmin)1, and CD spectra from 200

to 250 nm were taken every 10 C Selected samples are shown.

The sample buffer was 50 m M Mops and 1 m M MgSO4(pH 8.0).

Table 1 Heat inactivation of the Vibrio AP variants at different

temperatures Samples were incubated at the indicated

tempera-tures, and residual activity was measured at intervals of a few

minutes using a standard enzyme assay The t 1⁄ 2 values were

cal-culated using the relationship t 1⁄ 2 = ln(2) ⁄ k, where k is the rate

constant of inactivation at the appropiate temperature (n = 3) ND,

no detectable activity loss over 30 min –, activity loss too fast to

measure.

T (C)

Wild-type,

t1⁄ 2(min)

W274K,

t1⁄ 2(min)

H116D,

t1⁄ 2(min)

H116D ⁄ W274K,

t1⁄ 2(min)

0 0.2 0.4 0.6 0.8 1

Wt H116D W274K H116D/W274K

Fig 4 Effect of heat on secondary structure stability of the wild-type Vibrio AP and the H116D, W274K and H116D ⁄ W274K mutants The enzymes were heated from 15 C to 90 C at a rate 1 CÆmin)1, and ellipticity at 222 nm was monitored and plotted at 0.1 C inter-vals Ellipticity at 222 nm has been normalized to fraction of unfolded protein The protein concentration was 0.06 mg ⁄ mL The sample buffer was 50 m M Mops and 1 m M MgSO 4 (pH 8.0).

by atomic absorption spectrometry was determined as

1.9 ± 0.6 (n = 12) for zinc and 4.1 ± 1.7 (n = 6) for

magnesium No cobalt or copper was detected Thus,

wild-type Vibrio AP most likely has two Zn2+and one

Mg2+in the active site in addition to three Mg2+that

bind elsewhere Evidence for additional Mg2+-binding

sites in APs, possibly with a functional role, is emerging

from known crystal structures [17]

Zn2+release from the active site of wild-type Vibrio

AP was monitored while the enzyme was inactivated by

either heat or urea treatment The chelator

4-(2-pyridyl-azo)-resorcinol was used to detect free Zn2+ At 37C,

the enzyme lost activity quite quickly, but no Zn2+

release was detectable over a period of 4 h At 60C,

the enzyme released two equivalents of Zn2+ within

30 min At that temperature, inactivation of the enzyme

was spontaneous Similar results were obtained with

urea denaturation experiments No Zn2+ release was

observed at urea concentrations below 1 m, whereas at

that concentration all activity had been lost (data not

shown) The results suggest that Zn2+release from the

active site coincides with total protein structure

unfold-ing but that the activity is lost before that event

Effect of Mg2+on active site stability of wild-type

Vibrio AP

The effect of Mg2+concentration on wild-type Vibrio

AP enzyme activity was investigated Enzyme samples

were incubated at 18C either with no Mg2+ in the solution or with 10 mm Mg2+ added (Fig 6) With

10 mm Mg2+, the enzyme activity was completely sta-ble over 160 min However, with no Mg2+in solution, the activity was lost quite quickly and had dropped to 20% of the initial activity after 160 min At higher temperatures, the inactivation was very fast

Kinetic properties The catalytic efficiencies of wild-type Vibrio AP and the three mutants were determined both under hydrolysing and under transphosphorylating condi-tions (Tables 3 and 4, respectively) Kinetic constants for the E coli AP were also measured for comparison Under conditions in which the phosphoseryl intermedi-ate is hydrolyzed (Table 3), the wild-type Vibrio AP had the highest catalytic efficiency (kcat⁄ Km), 10-fold higher than that of the E coli enzyme All enzyme activities were measured at their optimal pH under hydrolyzing conditions (pH 8.0 for the E coli enzyme and pH 9.8 for the Vibrio enzymes) All the mutants had much reduced catalytic efficiencies as compared to the wild-type Vibrio enzyme The kcat for the W274K mutant was seven times lower than that for the wild-type (changing from 1580 s)1 to 220 s)1) The Km of that mutant was also increased from 0.08 mm for the wild-type enzyme to 0.19 mm Overall, the catalytic efficiency of the W274K mutant was 16.5 times lower

0

20

40

60

80

100

340 345 350 355

Fig 5 Effect of urea on activity (open squares) and tryptophan

flu-orescence (black squares) of wild-type Vibrio AP The protein

con-centration was 0.01 mg⁄ mL Samples were incubated for 4 h at

15 C in 25 m M Mops and 1 m M MgSO4(pH 8.0) and different

con-centrations of urea before measurement of each spectrum

Fluo-rescence is given as maximum emission wavelength (kmax) The

excitation wavelength was 290 nm, and fluorescence was scanned

from 310 to 400 nm at 15 C Residual activity was measured

using the standard enzyme assay.

0

20 40 60 80 100

Time (min)

Fig 6 Effect of Mg 2+ on wild-type Vibrio AP active site stability Wild-type Vibrio AP in 10 m M Mg 2+ solution (black squares) and wild-type Vibrio AP in a solution without added Mg 2+ (open squares) The background solution was 20 m M Tris and 15% ethyl-ene glycol (pH 8.0) Samples were incubated at 18 C Residual activity was measured using the standard enzyme assay.

than that of the wild-type The H116D and the

H116D⁄ W274K mutants also had much reduced

turn-over rates, and for both of these enzymes, substrate

binding was greatly impaired (15–18 times greater Km)

The wild-type enzyme was a 30-fold better catalyst

(kcat⁄ Km) than these two mutants under hydrolyzing

conditions Although the E coli enzyme had a much

lower turnover rate than the three Vibrio mutants, it

was overall a slightly better catalyst, due to tighter

substrate binding The E coli enzyme also had the

lowest Ki for inorganic phosphate as compared to the

wild-type Vibrio AP, indicative of higher binding

affin-ity for the competitive inhibitor Binding affinities of

the H116D and H116D⁄ W274K mutants were

espe-cially low for the same reason Under

transphosphory-lating conditions, the enzymes were compared at

pH 8.0 (the pH optimum for the E coli enzyme) The

Vibrio wild-type AP was still the best catalyst of the

five enzymes, although it was not at its optimum pH,

with 10-fold better catalytic efficiency than the E coli

AP Under these conditions, the Vibrio mutants

showed insignificantly lower turnover rates as

com-pared to the wild-type, except for the H116D variant

However, reduced substrate binding of all three

mutants made them much less efficient catalysts For

the wild-type Vibrio AP, the turnover rate did not

increase much when going from hydrolyzing to

trans-phosphorylating conditions However, the three Vibrio

AP mutants, especially the W274K variant, seemed to

be much more dependent on the organic phosphate

acceptor in attaining high turnover rates, an indication that breakdown of the phosphoseryl intermediate is the rate-limiting step in the cold-active AP variants

Discussion

Stability of wild-type Vibrio AP and the mutants The active site area of wild-type Vibrio AP was very heat-sensitive At a temperature as low as 25C, the t1⁄ 2

of the enzyme’s activity was only 94 min At that tem-perature, no activity loss was detected for the three mutants produced here H116D began to lose activity rapidly at 35C, but the W274K and the H116D ⁄ W274K mutants were stable up to 40C However, the heat stability of APs from E coli and other mesophilic bacteria is even greater [1] The E coli enzyme is espe-cially heat-tolerant, with a t1⁄ 2of 8 min at 90C [1], and

AP from the mesophilic Bacillus subtilis has a t1⁄ 2 of

28 min at 65C Only one AP has been characterized that is more heat-labile than the Vibrio AP, the AP from the Antarctic bacterium HK47 [36], with a t1⁄ 2of only

2 min at 40C as compared to 4 min for the Vibrio AP

at the same temperature Other heat-labile and cold-active APs displayed t1⁄ 2values of 15 min at 45C [38] and 3–30 min at 55–65C [35,39,40] From these find-ings, it may be concluded that the Vibrio AP is a good example of a cold-adapted AP

The effect of heat on activity is often used as an indi-cation of the total heat stability of enzyme structures

Table 3 Kinetic constants for the native Vibrio AP and its mutants as compared to E coli AP under hydrolyzing conditions Conditions were

50 m M Mops, 50 m M CAPS, and 1 m M MgSO 4 , using pNPP as substrate at the optimum pH for each enzyme (pH 8.0 for the E coli AP and

pH 9.8 for the Vibrio variants) Each number is the average of at least three independent experiments.

kcat(s)1) Km(m M )

kcat⁄ K m (s)1Æm M )1)

Ki(for Pi) (m M )

Table 4 Kinetic constants for the native Vibrio AP and its mutants as compared to E coli AP under transphosphorylating conditions Condi-tions were 1.0 M Tris, 1 m M MgSO4and pH 8.0 with pNPP used as substrate Each number is the average of at least three independent experiments.

k cat (s)1) K m (m M )

kcat⁄ K m (s)1Æm M )1) K

i (for P i ) (m M )

For the wild-type Vibrio AP, however, loss of activity

and loss of secondary structures did not coincide

Activity was lost at much lower temperatures than the

changes in secondary structures observed by CD Urea

denaturation of the wild-type enzyme gave similar

results (Fig 5), indicating local vulnerability in the

active site area Structural stability (Tm obtained by

CD) was increased for the W274K and the

H116D⁄ W274K mutants by 0.7 C and 2.1 C,

respec-tively, as compared to the wild-type enzyme

Interest-ingly, however, the H116D variant showed decreased

structural stability, and Tm was lowered by 1.5C A

correlation was observed whereby for the two more

stable variants (W274K and H116D⁄ W274K), the

more robust global structure brought increased

stabil-ity to the active site area, whereas for the H116D

vari-ant, both factors were reversed as compared with the

wild-type structure We may conclude that the active

site of the cold-active Vibrio AP has evolved to be

especially flexible and consequently more

heat-intoler-ant than the whole structure of the enzyme A possible

explanation for the increased heat stability of the

W274K variant might stem from the fact that lysine is

able to form ionic bonds with an adjacent negatively

charged amino acid and forms more stabilizing bonds

in the active site than tryptophan does, perhaps for

steric reasons Tryptophan has some preference for

binding sites where large conformational changes

occur, causing hydrophobic–hydrophobic, aromatic–

aromatic and hydrophobic–polar residue pair

interac-tions [41] Inspection of the homology model revealed

several possibilities for bonding with W274 These

include Q18, H286, H316, and E317 Furthermore,

bond networks mediated by water molecules could

extend the crosslinking possibilities Further

specula-tion regarding individual interacspecula-tions is not warranted

until a precise structure has been solved Our

experi-mental results showed that the E coli AP mimic

H116D⁄ W274K was indeed the most stable of the

mutants made here, with an increased Tmof 2.1C as

compared to the wild-type Vibrio AP In the E coli

AP, a salt bridge exists between Asp153 and Lys328

[3] (corresponding to amino acids 116 and 274 in the

VibrioAP) It may be that such a salt bridge has also

formed in the H116D⁄ W274K mutant and gives the

enzyme increased stability, as was observed However,

it was a long way from equalling the reported Tm of

E coliAP of up to 97.0C [1,42] The 2.1 C increase

in Tmfor the H116D⁄ W274K mutant is clearly a result

of the favorable combination of the two single

muta-tions, as the H116D mutant had a reduction in Tm of

1.5C, and the W274K mutant showed an increase in

Tm of 0.7C The absence of an ionic bond involving

amino acids 116 and 274 in the wild-type Vibrio AP may thus represent a part of the enzyme’s cold adapta-tion

Catalytic efficiency The catalytic efficiency of wild-type Vibrio AP is 10 times greater than that of the E coli enzyme, both under hydrolyzing and under transphosphorylating conditions Kmwas higher for all the mutants, and the turnover number was reduced Furthermore, all three mutants studied here had higher Kivalues for the com-petitive inhibitor inorganic phosphate, indicating reduced binding affinity If the rate-determining step of the Vibrio AP reaction is the release of the phospho-rous product from the noncovalent enzyme–phosphate complex, as found for the E coli AP [25,26], lower affinity for inorganic phosphate should lead to an increase in the catalytic rate (kcat) in the mutants Such

an increase was not seen, which points to a different rate-determining step for the Vibrio AP as compared

to the E coli enzyme Deacylation would be the most likely candidate, perhaps involving a conformational change [23,43] The conformational change needed for the deacylation step may be more easily achieved in cold-active enzymes such as the Vibrio AP, as they are believed to have more flexible structures than their mesophilic counterparts

The Km of the W274K mutant was increased three-fold as compared to the wild-type enzyme under both assay conditions examined (Tables 3 and 4) In the

E coli enzyme, the equivalent Lys328 makes a water-mediated connection to inorganic phosphate [3] that the Lys in the Vibrio AP W274K mutant apparently does not make, because it showed reduced affinity both for the substrate and for inorganic phosphate as compared to the wild-type enzyme The kinetic con-stants of a W274A mutant (data not shown) were nearly identical to those of the W274K mutant, sup-porting this conclusion This is in contrast to the posi-tion of Lys328 in the D153H mutant of E coli AP [9], which is an analog of our W274K mutant with respect

to residues 153 and 328 Affinity for the substrate and inorganic phosphate was increased in the D153H mutant as compared to the wild-type E coli AP, possi-bly due to loss of the salt bridge between the Asp nor-mally in position 153 and Lys328 When no ionic connection was present, it was suggested that the Lys would change position and connect directly to the phosphorous group, leading to tighter binding of both substrate and inhibitor Although Lys is probably posi-tively charged in the Vibrio AP active site environ-ment, and could therefore make a connection to

inorganic phosphate, as may be the case in the D153H

E colimutant, it does not appear to do so It is

possi-ble that it points away from the substrate and forms

an ionic connection to some negatively charged amino

acid elsewhere Results similar to ours were obtained

for the H317K mutant of human placental AP [44],

which is also an analog of our W274K mutant In that

case, substrate affinity was reduced in an H317K

mutant, suggesting a worse connection of a Lys in this

position to the phosphorous moiety of the substrate,

leading to a conclusion similar to ours: namely, that

the position of Trp274 in Vibrio AP does not allow it

to make stabilizing interactions with bound phosphate

groups of equivalent strength to that which produces

the low Km of the E coli AP even when it is changed

to a Lys

As compared to wild-type Vibrio AP, the H116D

single mutant had highly reduced affinity for the

sub-strate and the inhibitor inorganic phosphate, as judged

by Km and Ki values (Tables 3 and 4) This may be a

result of the negative charge of the Asp repelling the

negative charges of both substrate and inhibitor The

K328W mutant of E coli AP is an analog of our

Vibrio AP H116D mutant with respect to those two

sequence positions That enzyme also showed reduced

substrate affinity as compared to wild-type E coli AP

[14], as did AP from Saccharomyces cerevisiae, which

also has Asp in the corresponding position [45] All

these results show that the presence of an Asp residue

in this position (residue 153 according to E coli

num-bering) without a Lys residue in the position

corre-sponding to residue 328 in E coli causes less affinity

for both substrate and the inhibitor, inorganic

phos-phate

Finally, the H116D⁄ W274K Vibrio AP double

mutant is the active site analog of the wild-type E coli

AP, where there is an ionic connection between Asp153

and Lys328 [3] The H116D⁄ W274K Vibrio AP mutant

needed very high concentrations of substrate or

inhibi-tor as compared with the wild-type or the single

mutants A possible explanation for the observed

char-acteristics could be a combination of two things:

repel-lent forces between the negative charge on Asp116 and

negative charges on substrate and inhibitor, and the

reduced connection of Lys274 to the substrate and

inhibitor as compared to Trp274, due to lack of the

expected salt bridge and poorer positioning Heat

stability measurements point to additional ionic

con-nections being formed involving Asp116 and Lys274

(Figs 2 and 4, and Tables 1 and 2) Given that it is

unlikely, due to distance, that the Lys–Asp salt bridge

is formed in the H116D⁄ W274K mutant, the reduced

substrate⁄ inhibitor affinities of this enzyme may be

explained by altered electrostatics and possibly increased flexibility of the active site A structural con-striction resulting from new crosslinking that impedes the chemical step(s) may be postulated to explain lower turnover rates of this mutant as compared to the wild-type Vibrio AP, making its catalytic and stability prop-erties change in the direction of the E coli counterpart without fully reaching the properties of the latter

Experimental procedures

Materials

Salts and general chemicals were obtained either from Sigma (St Louis, MO, USA) or Merck (Darmstadt, Ger-many) l-Histidyldiazobenzylphosphonic acid on agarose, p-nitrophenyl phosphate (pNPP), ampicillin, SDS, E coli

AP (P-5931), lysozyme and deoxyribonuclease were also obtained from Sigma PhastSystem 8–25% polyacrylamide gels, SDS buffer strips, Coomassie Brilliant Blue for Phast-System and MonoQ ion exchange columns were purchased from Amersham Pharmacia-Biotech (Uppsala, Sweden) Tryptone and yeast extract were purchased from Difco Laboratories (Detroit, MI, USA) Restriction endonucleases Apa1, Dpn1, HindIII and PstI, as well as DNA T4 ligase, DNA size standards, SDS protein standards, and LB agar, were obtained from Fermentas (Burlington, Canada) Aga-rose was purchased from FMC Bioproducts (Rockland,

ME, USA) Plasmid pBluescript II SK (+) and the Quik-Change mutagenesis kit were obtained from Stratagene (La Jolla, CA, USA) Oligonucleotide primers were obtained from T-A-G Copenhagen (Copenhagen, Denmark) Chemi-cals for sequencing were obtained from Applied Biosystems (Foster City, CA, USA)

Gene cloning

The Vibrio AP gene was cloned from a pUC18 plasmid that had been previously inserted [46] into the vector pBlue-script KS II (+)

Construction of mutants

Mutants of Vibrio AP were constructed with a QuikChange mutagenesis kit, according to the manual from the manufacturer (Stratagene) PCR was performed in a Gene-Amp-PCR system 2700 from Applied Biosystems Pfu poly-merase (Stratagene or Fermentas) was used for the reactions Chemically competent E coli TOP10 cells were transformed with the reaction product Colonies were cul-tured and plasmid DNA was isolated from the cells using the Qiaprep Spin Miniprep Kit from Qiagen Those plas-mids were sequenced, and one of the plasplas-mids containing the desired mutation was used for the transformation of

competent E coli cells of strain LMG194 (Invitrogen) that

lack native AP [F–, DlacX74, galE, thi, rpsL, DphoA,

(PvuII), Dara714, leu::Tn10]

Expression

Expression was performed in 4 L portions (divided into

nine 1 L flasks) LMG194 cells containing the plasmid

encoding the Vibrio AP enzyme or mutants were cultured

in LB medium supplemented with 100 lgÆmL)1 ampicillin

Cells were grown at 18C with shaking (220 cyclesÆmin)1)

Cell density was monitored by measuring A600 Cells were

further handled for protein extraction 10–12 h after they

had reached stationary phase

Extraction of protein from cells

Cultured cells were centrifuged for 10 min at 6000 g using a

Sorvall RC5C centrifuge The precipitated cells were then

redissolved in 1 : 10 of the original volume in 0.01%

Tri-ton X-100, 0.5 mgÆmL)1 lysozyme, 20 mm Tris, 10 mm

MgCl2 (pH 8.0) The solution was left to stand for 2–3 h,

and then frozen at – 20C After freezing, the solution was

thawed, DNase was added to a final concentration of

0.05 lgÆmL)1, and the resulting solution was left to stand

for 30 min at room temperature The solution was then

centrifuged at 10 000 g for 15 min at 4C Active AP was

now in a clear solution

Protein purification

The solution containing active Vibrio AP was purified on

an l-histidyldiazobenzylphosphonic acid agarose column at

4C as previously described [33] The enzyme was further

purified and concentrated on a MonoQ ion exchange

col-umn connected to an FPLC apparatus (Amersham

Pharma-cia-Biotech) maintained at 18–20C The enzyme was

eluted with a 0–0.7 m NaCl gradient

Enzymatic assay

Enzyme activity was routinely measured with 5 mm pNPP

in 1.0 m diethanolamine buffer, containing 1.0 mm MgSO4,

at pH 9.8 and 25C Reactions were initiated by the

addition of enzyme, and the release of p-nitrophenol

(e = 18.5 m)1Æcm)1) was monitored at 405 nm

Protein determination

Protein concentration was estimated by measuring

absorbance at 280 nm and using a calculated absorbance

coefficient of 0.96 cm2Æmg)1 (e = 55 810 m)1Æcm)1) for

the wild-type and H116D mutant, and 0.87 cm2Æmg)1

(e = 50 310 m)1Æcm)1) for the W274K mutant and the

double mutant [47] The protein concentration for E coli

AP enzyme solutions was estimated by measuring absor-bance at 278 nm and using the absorabsor-bance coefficient 0.71 cm2Æmg)1[48]

Kinetics

Determination of kinetic constants was performed at seven different substrate concentrations in the range 0.04– 0.8 mm Kinetic constants were determined both under hydrolyzing and transphosphorylating conditions at 25C The transphosphorylating conditions were 1 mm Tris and

1 mm MgSO4 (pH 8.0), and the hydrolyzing conditions were 50 mm Tris and 50 mm Caps (pH 9.8 or pH 8.0) Kinetic constants were calculated using the Lineweaver– Burk transformation of the Michaelis–Menten equation

Kcatvalues were calculated from Vmaxvalues using molecu-lar masses of 58 kDa for the Vibrio phosphatases and

94 kDa for the E coli phosphatase [49] For determination

of Kifor inorganic phosphate, Kmapparentfor catalysis in the presence of 1.0 and 2.5 mm phosphate was measured, and

Kiwas calculated using Kmapparent= Km(1 + [I]⁄ Ki)

Temperature stability measurements

For determination of active site thermal stability, enzyme samples were incubated at different temperatures in 20 mm Tris and 10 mm MgSO4 (pH 8.0), and activity was mea-sured after different incubation times by the standard pro-tocol A J-810 CD spectrometer from Jasco (Tokyo, Japan) was used to obtain CD spectra from 190 to 260 nm and determine Tmcurves for the Vibrio APs Experiments were carried out in 50 mm Mops and 1 mm MgSO4 (pH 8.0) For the Tm experiments, the CD signal at 222 nm was monitored as the temperature increased by 1CÆmin)1 The protein concentration was in the range 0.05–0.1 mgÆmL)1

Urea denaturing experiments

Enzyme samples were incubated for 4 h at 15C in 25 mm Mops and 1 mm MgSO4(pH 8.0) solutions containing dif-ferent concentrations of urea prior to measurements obtained by dilution from a 9 m stock solution Residual enzyme activity was measured using the standard assay The intrinsic fluorescence of each sample was measured using a Fluoromax instrument and analyzed using Datamax software (Jobin Yvon) The excitation wavelength was

290 nm, and fluorescence was scanned from 310 to 400 nm

at 15C An average of three scans was taken For each sample, the maximum emission wavelength (kmax) was determined The protein concentration was 0.009– 0.011 mgÆmL)1 Background spectra were determined at each urea concentration and subtracted General data han-dling followed well-established procedures [37,50]

Metal ion analysis

Immediately before mass analysis, the samples were run at

0.6 mLÆmin)1and room temperature through a HiTrap

des-alting column in an Agilent (Morges, Switzerland) 1200

Series apparatus, using 20 mm NH4H2PO4 (pH 8) as a

mobile phase Monitored masses on an Agilent 7500ce

inductively coupled plasma MS instrument were Mg, V,

Mn, Co, Ni, Zn (66), Zn (68), Mo, Cd, and Sn For atomic

absorption analysis, samples were dialyzed with three

changes over a 2 day period in Spectropor 2 membrane

tubing at 4C using a 1 : 250 excess of Chelex-treated

20 mm Tris (pH 8.0) buffer The concentration of protein

for analysis was 0.25–0.4 mgÆmL)1, and samples were

acidi-fied with HCl prior to injection A Varian (Crawley,

England) Spectr220 FS atomic absorption spectrometer was

employed to measure Zn, Mg, Co, and Cu

Acknowledgements

The Icelandic Research Fund and the University of

Iceland Research Fund supported this work

finan-cially We thank Dr Ernst Schmeisser for the mass

analysis of metals, and Dr Sigridur Jonsdottir for

assistance with atomic absorption analysis Professor

Olafur S Andresson assisted with early gene-cloning

work

References

1 McComb RB, Bowers GN & Posen S (1979) Alkaline

Phosphatase Plenum Press, New York, NY

2 Coleman JE (1992) Structure and mechanism of

alka-line phosphatase Annu Rev Biophys Biomol Struct 21,

441–483

3 Kim EE & Wyckoff HW (1991) Reaction mechanism of

alkaline phosphatase based on crystal structures J Mol

Biol 218, 449–469

4 Sowadski JM, Handschumacher MD, Murthy HMK,

Foster BA & Wyckoff HW (1985) Refined structure of

alkaline phosphatase from Escherichia coli at 2.8A˚

reso-lution J Mol Biol 186, 417–433

5 Butler-Ransohoff JE, Rokita SA, Kendall DA, Banzon

JA, Carano KS, Kaiser ET & Matlin AR (1992)

Active-site mutagenesis of E coli alkaline phosphatase –

replacement of serine-102 with nonnucleophilic amino

acids J Org Chem 57, 142–145

6 Chaidaroglou A & Kantrowitz ER (1989) Alteration of

aspartate 101 in the active site of Escherichia coli

alka-line phosphatase enhances the catalytic activity Protein

Eng 3, 127–132

7 Dealwis CG, Brennan C, Christianson K, Mandecki W

& Abadzapatero C (1995) Crystallographic analysis of

reversible metal binding observed in a mutant

(Asp153->Gly) of Escherichia coli alkaline phospha-tase Biochemistry 34, 13967–13973

8 Dealwis CG, Chen LQ, Brennan C, Mandecki W & Abadzapatero C (1995) 3-D structure of the D153G mutant of Escherichia coli alkaline phosphatase: an enzyme with weaker magnesium binding and increased catalytic activity Protein Eng 8, 865–871

9 Janeway CML, Xu X, Murphy JE, Chaidaroglou A & Kantrowitz ER (1993) Magnesium in the active site of Escherichia colialkaline phosphatase is important for both structural stabilization and catalysis Biochemistry

32, 1601–1609

10 Murphy JE, Xu X & Kantrowitz ER (1993) Conversion

of a magnesium binding site into a zinc binding site by

a single amino acid substitution in Escherichia coli alka-line phosphatase J Biol Chem 268, 21497–21500

11 Murphy JE, Tibbitts TT & Kantrowitz ER (1995) Mutations at positions 153 and 328 in Escherichia coli alkaline phosphatase provide insight towards the struc-ture and function of mammalian and yeast alkaline phosphatases J Mol Biol 253, 604–617

12 Tibbitts TT, Xu X & Kantrowitz ER (1994) Kinetics and crystal structure of a mutant Escherichia coli alka-line phosphatase (Asp-369->Asn): a mechanism involv-ing one zinc per active site Protein Sci 3, 2005–2014

13 Tibbitts TT, Murphy JE & Kantrowitz ER (1996) Kinetic and structural consequences of replacing the aspartate bridge by asparagine in the catalytic metal triad of Escherichia coli alkaline phosphatase J Mol Biol 257, 700–715

14 Wojciechowski CL & Kantrowitz ER (2002) Altering of the metal specificity of Escherichia coli alkaline phos-phatase J Biol Chem 277, 50476–50481

15 Le Du MH, Stigbrand T, Taussig MJ, Me´nez A &

Stur-a EA (2001) CrystStur-al structure of Stur-alkStur-aline phosphStur-atStur-ase from human placenta at 1.8 A˚ resolution Implication for a substrate specificity J Biol Chem 276, 9158–9165

16 de Backer M, McSweeney S, Rasmussen HB, Riise BW, Lindley P & Hough E (2002) The 1.9 A˚ crystal struc-ture of heat-labile shrimp alkaline phosphatase J Mol Biol 318, 1265–1274

17 Wang E, Koutsioulis D, Leiros HK, Andersen OA, Bouriotis V, Hough E & Heikinheimo P (2007) Crystal structure of alkaline phosphatase from the Antarctic bacterium TAB5 J Mol Biol 366, 1318–1331

18 Stec B, Holtz KM & Kantrowitz ER (2000) A revised mechanism for the alkaline phosphatase reaction involv-ing three metal ions J Mol Biol 299, 1323–1331

19 Xu X & Kantrowitz ER (1991) A water-mediated salt link in the catalytic site of Escherichia coli alkaline phosphatase may influence activity Biochemistry 30, 7789–7796

20 Wang J, Stieglitz KA & Kantrowitz ER (2005) Metal specificity is correlated with two crucial active site