ELSEVIER Biochimica et Biophysica Acta 1247 (1995) 260-264

BB. Biochi~Pic~a et Biophysica AEta

Identification of tyrosine 108 in coffee bean a-galactosidase as an essential residue for the enzyme activity

Alex Zhu, Zhong-Kun Wang, Jack Goldstein * Lindsley F. Kimball Research Institute of The New York Blood Center, 310 East 67 Street, New York, NY 10021, USA

Received 16 June 1994; revised 27 September 1994; accepted 28 October 1994

Abstract

The cDNA for coffee bean a-galactosidase (a-GaD has been cloned and expressed in a baculovirus expression system. An early study of coconut a-Gal by chemical modification suggested that one tyrosine residue is at or near the active site. In order to identify such a critical residue, we replaced two tyrosine residues (positions 108 and 158) with phenylalanine by site-directed mutagenesis. The mutated DNA strands, as well as the wild-type ones, were subcloned into pVL vector and transformed into Sf9 insect cells for intracellular expression. The replacement of Tyr-158 with phenylalanine resulted in a mutant a-Gal (Y158F) which retained approx. 88% of the activity of wild-type enzyme. However, the substitution of Tyr-108 by phenylalanine (Y108F) almost abolished the enzymatic activity (1.8% of wild-type activity). The Vm~/K m value for the mutant Y108F was 0.027, which was over a 1000-fold lower than that of wild-type a-Gal. Our data suggest that Tyr-108 is critical for the enzymatic activity of a-Gal.

Keywords: a-Galactosidase; Lysosomal enzyme; Enzyme activity; (Coffee bean)

1. Introduction

a-Gal obtained from coffee bean cleaves the terminal a-galactose residues from oligosaccharide chains on group B red cells, thus generating O-type red cells [1]. The enzyme-converted red cells have been successfully tested in clinical studies of blood transfusion [2]. The cDNA encoding coffee bean a-Gal has recently been cloned and sequenced [3]. The deduced amino-acid sequence exhibits significant homologies with a-Gal from various sources, including guar (80%), yeast (58%) and human (59%). Moreover, a -Gal shows homology with a-N-acetylgalac- tosaminidase ( a - N A G A , EC 3.2.1.50) from chicken and human, supporting the hypothesis that these two function- ally specific glycosidases might have evolved from a common ancestral gene [4,5]. Ishii et al. [6] studied a-Gal activity by using chimeric proteins comprising human a-Gal and a -NAGA. Their data suggest that two regions encoded by exons 1, 2 and 6 contribute to the cleavage of

Abbreviations: ot-Gal, ot-galactosidase; a-NAG/t, ot-N-acetylgalac- tosaminidase (EC 3.2.1.50); wt, wild type; Sf9 cell, Spodoptera frugiperda cell; CRIM, cross-reacting immunologic material.

* Corresponding author. Fax: + 1 (212) 8790243.

0167-4838/95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0167-4838(94)00228-2

a-linked terminal galactose residues. However, little is known about amino-acid residues directly involved in the catalytic mechanism. An early study of coconut a-Gal using chemical modification procedures suggested the presence of two carboxyl groups, a tryptophan and a tyrosine, at or near the active site of the enzyme [7]. We reason that if a -Gal and a N A G A share high sequence homology, it is possible that these two enzymes also share a similar catalytic mechanism. Thus the residues critical to a-Gal are likely to be important to a - N A G A as well. Since we localized two tyrosine residues which are con- served among the cDNAs encoding these two enzymes, as determined by sequence alignment, we decided to mutate these two tyrosine residues (108 and 158) in coffee bean a-Gal. Here, we present data t o s u g g e s t that tyrosine at position 108 is critical for catalytic activity.

2. Materials and methods

2.1. Strains and plasmids

Escherichia coli I N V a F ' or JM109 was used for stan- dard cloning. Spodoptera frugiperda (Sf9) cells were host cells for baculovirus transfection and protein expression.

A. Zhu et al. / Biochimica et Biophysica Acta 1247 (1995) 260-264 261

pCRII vector (Invitrogen) was used for direct cloning of PCR amplified products and pVL-1393 vector (PharMin- gen) used for expressing a-Gal cDNA in Sf9 cells.

2.2. Plasmid constructions

Standard cloning techniques were applied [8]. The start- ing plasmid was pCR-BZ, which contains the coding re- gion of coffee bean a-Gal cDNA in a pCRII vector, as reported previously [3]. In order to introduce a BgllI site upstream of translation initiation site, the oligonucleotide P1 was made, in which the 5' end contains a BgllI recognition site and the 3' end sequence matches nt 90-108 of a-Gal cDNA (initiation site ATG is at nucleotides 103-105). The a-Gal coding sequence was PCR-amplified by using two primers, P1 and M13 reverse primer (M13R) which sits on pCRI vector adjacent to the 3' end of the coding sequence. PCR conditions were as follows: denatu- ration (1 min, 94°C) annealing (2 min, 50°C) and extension (3 min, 72°C). The resultant PCR product was digested with BgllI and EcoRI and cloned into the same sites of a baculovirus expression vector pVL-1393, generating the plasmid pVL-BZ. The PCR-amplified sequence was veri- fied by an automated DNA sequencer.

2.3. Site-directed mutagenesis

The oligonucleotides used for site-directed mutagenesis are shown in Table 1. The underlined single nucleotide in the oligonucleotides indic.ates the induced mutation. Tak- ing into consideration thal~ Taq polymerase attaches adeno- sine at the 3' end of the PCR-amplified fragment, all mutagenic oligonucleotides were designed in such a way that the first 5' nucleotide of the primer follows a T-re- sidue in the same strand of template sequence. The site-di- rected mutagenesis was carried out according to the PCR- overlap extension method described by Ho et al. [9]. Plasmids PCR-BZ and pVL-BZ were used as templates and M13R and BV-L as primers. BV-L resides on the pVL vector adjacent to the 5' end of a-Gal cDNA and ml3R on the pCRII vector adjacent to the 3' end of the cDNA. The generation of Y108F mutant by the overlap extension method was diagrammed in Fig. 1 as an example. The mutated fragment was then subcloned into BgllI and EcoRI sites of the baculovirus expression vector pVL-1939, generating pVL-BZ (Y108F). Automated DNA sequencing was performed to confirlrt the desired mutation and fidelity of the remainder of the sequence. The same strategy was used to make plasmid p~W_,-BZ(Y158F).

MI

l PcR

~, Primers: MI and T7

[ PCR I Primers: BV-L and T7

Bglll EcoRI

I ! I

Fig. 1. Strategy of generating site-directed mutagenesis by the overlap extension method. In the first PCR, the plasmid pCR-BZ was used as a template, and MI and T7 as primers. The PeR product was gel-purified and used as a template in conjunction with pVL-BZ in the second PCR (primers BV-L and T7). The products was digested with BglII and EcoRI, and cloned into the same sites of the expression vector pVL-1392.

2.4. Sf9 cell transfection

Recombinant baculovirus containing the normal or mu- tant a-Gal cDNA were produced by homologous recom- bination according to the procedure recommended by PharMingen. In brief, 2 /zg of the plasmid containing wt or mutated a-Gal cDNAs were co-transfected into Sf9 cells with the baculogold DNA, a lethal deletion of the virus DNA. After 4 days at 27°C, the supernatants from the infected cells were collected and added to fresh Sf9 cells for viral amplification (2- 106 cells/1 ml of super- natant). The viral amplification was repeated two more times (2-106 cells/0.1 ml of supernatant) in order to obtain a high titer recombinant viral stock.

2.5. Enzyme expression and activity

Sf9 cells ( 2 .10 6) infected with recombinant bac- ulovirus (3rd round viral amplification) were harvested and resuspended in 0.2 ml of medium. The cell lysate was prepared by the freeze-thaw method and then stored at -20°C until use. The protein expression in Sf9 cells was examined by a Western blot using the antibody raised against purified coffee bean a-Gal. The amount of ex- pressed a-Gal was estimated as cross-reacting immuno- logic material (CRIM) by comparing it to the known amount of purified a-Gal on the same Western blot [10,12].

Table 1 Oligonucleotides for site-directed mutagenesis

Designation Amino-acid substitution Nuclcotide substitution Oligonucleotides

Y108F Tyr-108 ~ Phe TAC ~ Trc ml: 5'-GGA ATI" TI'C TCTGATGCTG Y158F Tyr-158 --, Phe TAT ~ 111 m2: 5'-GAT TGG AAA CCTITCCI'I" G

Olignnucleotides were made on ABI model 391 DNA synthesizer; ml is the coding sequence, m2 is the non-coding sequence.

262 A. Zhu et al. /Biochimica et Biophysica Acta 1247 (1995) 260-264

Extracts of Sf9 cells infected with wild-type virus gave no detectable signal on Western blotting. The a-Gal activity was measured by incubating 10 /xl of Sf9 cell lysate with 100/zl of 1.25 mM PNP-a-Gal in citrate-phosphate buffer (pH 6.5). After incubation at 37°C for an hour the reaction was terminated by adding 1 ml of 0.2 M borate buffer (pH 9.8) and absorbance at 405 nm was measured. The specific activity of a-Gal expressed in Sf9 cells was defined as /xmol of substrate digested per hour per mg of the enzyme (CRIM) at 37°C under the assay conditions described above. Kinetic parameters were calculated from initial velocities in the 0.1-2 mM range of substrate concentra- tion for a-Gal (wt and Y158F), and in 2-16 mM range for a-Gal (Y108F). The computer program Enzpack 3 (Bio- soft) was applied to generate kinetic parameters.

kDa 1 2 3 4 5

100-

55- 43 - 36-

29-

.ql

3. Results

3.1. Expression of wt and mutant a-Gals in Sf9 cells

Significant homology at the amino-acid level has been indicated among various a-Gal and a-NAGA. It has been suggested these two functionally distinctive exoglycosi- dases may have evolved from a common ancestral gene [4,5]. Furthermore, it is also possible that they share a similar catalytic mechanism and have similar amino-acid residues at the active site, although experimental evidence has yet to be provided. Selective amino-acid replacements of coffee bean a-Gal by site-directed mutagenesis should provide insight into critical residues involved in the active site. Two residues, Tyr-108 and Tyr-158, chosen for muta- tion in this study are conserved in all a-Gal and a-NAGA cDNAs that have been cloned so far [3,4].

Site-directed mutagenesis (Fig. 1, Table 1), subcloning and expression in a baculovirus system were performed as described in Section 2. In order to evaluate the expression level in Sf9 cells, a Western blot was carried out with the antibody raised against the purified coffee bean a-Gal (Fig. 2). 10 /xl of cell lysates containing a-Gal Y108F, Y158F or wild type were loaded in lane 1 through lane 3. Lanes 4 and 5 were control cell lysates without and with the addition of 30 ng purified enzyme, respectively. The pattern and level of expression among the three a-Gal

Fig. 2. Western blot analysis of wt and mutant a-Gal enzymes expressed in Sf9 cells. 10 /~1 of cell lysate was subjected to 0.1% SDS-12% PAGE and proteins were then transferred to a nitrocellulose membrane. The membrane was incubated with rabbit antisera raised against purified coffee bean a-Gal (1:500 dilution) and then alkaline phosphatase-con- jugated goat anti-rabbit IgG (1:7500 dilution) according to the manufac- turer's procedure (Promega). Lane 1, mutant Y108F; lane 2, mutant Y158F; lane 3, wild type; lanes 4 and 5, control cell lysate without and with the addition of purified ot-Gal (30 ng), respectively.

constructs were very similar, one predominant band mi- grated to the same position as the control enzyme in lane 5. However, a minor band with slightly larger molecular weight was visualized in lane 1 through lane 3. It is possible that this represents expressed protein with an uncleaved signal sequence or some form of post-transla- tional modification. The same Western blot pattern was observed when the enzyme was expressed in tunicamycin- treated Sf9 cells (unpublished data), suggesting that the band with a slightly larger kDa than native a-Gal was not caused by over-glycosylation. The gel image was scanned and analyzed by the computer programs, Adobe Photoshop and SigmaScan. Expressed wild-type and mutant enzymes (the predominant band in lanes 1 through 3) were esti- mated to be approx. 150 ng (CRIM) per lane or 15 ng/ /zl of cell lysate by comparing it to the purified a-Gal in lane 5 (30 ng of ot-Gal). As expected the specific activity of

Table 2 Kinetic constants of wild-type and mutant a-Gal enzymes

Specific activity % of wt activity Vma ~ (/.Lmol/min per mg) K m (mM) Vmax//gm (ml/min per mg) (/.Lmol/min per mg)

wild type 4.49 100% 13.4 0.443 30.2 Y158F 3.96 88% 10.1 0.527 19.2 Y108F 0.08 1.8% 5.66 209 0.027

Sf9 insect cells were transfected with plasmids containing wild-type and mutant a-Gal enzymes. The cells were grown at 27°C for 3 days. After three virus amplifications, cell lysates were prepared and activities were measured according to Section 2. The protein concentration of cell lysates was measured by Bio-Rad protein agent Kit. Apparent Michaelis constants (K=) and maximal velocities (Vm~ ,) were generated by the computer program Enzpack 3 based on the method of Lineweaver-Burk, and are expressed in mM and btmole/min per nag of expressed protein (CRIM), respectively.

A. Zhu et al. / Biochimica et Biophysica Acta 1247 (1995) 260-264 263

50

40

30 0 E 3

0 20

X v

'F 10 o <

(a)

-~ • ¢ ¢ • I

0 E 0

X

>

o ,<

25

(b)

20

15

10

5

0

• • •

¢ ¢ ¢ ¢ ¢

I I I I I I I I I I I

0 1 2 3 4 5 6 0 1 2 3 4

Incubation time (hour, at 37°C) Incubation time (hour, at 25°C) Fig. 3. The heat stability of wt and mutant ot-Gals at 37°C and 25°C. Panel A, cell lysates were incubated at 37°C for different lengths of time. At each time point, 10/.d cell lysate was taken for the activity assay with 100/ t l 1 mM PNP-a-Gal as substrate (pH 6.5) at 37~C for 1 h. The reaction was stopped by adding 1 ml of borate (pH 9.8), and the activity was then measured at 405 nm. The same experiments were done at 25°C (panel B). ( • ) Wild type, ( • ) mutant Y158F, and ( 0 ) mutant Y108F.

expressed wt a-Gal is indistinguishable from that of the enzyme purified from coffee bean.

3.2. Characterization of expressed wt and mutant a-Gal enzymes

Since their expression ]Levels in Sf9 cells were similar as suggested by Western blot and image analysis, it was possible to directly compare the activities of mutant and wt a-Gals in the cell lysates. The activity assay was carried out according to the procedure described in Section 2. Furthermore, apparent Michaelis constant (K m) and maxi- mal velocity (Vma x) for wt and mutant a-Gals were deter- mined according to the method of Lineweaver-Burk. The values for Vma X and K m were generated by a computer program Enzpack. The results are summarized in Table 2. While replacement of Tyr-158 by Phe (Y158F) has only a limited impact on the activity (88%), less than 2% activity was observed with the mutant Y108F. The effect, however, is more pronounced on the Vmax/Km value. Vmax/K m for 108F dropped over a 1000-fold compared to the wild-type enzyme, whereas that for Y158F remained over 60%.

A heat stability study was performed with the mutant and wt t~-Gal enzymes (Fig. 3). The enzyme-containing cell lysates were preincubated at 37°C (panel A) or 25°C (panel B) prior to measuring the enzyme activity. In order to measure the residual activity from the mutant Y108F, its activity assay was carried out in a 24 h incubation. Almost 100% of the activities of Y108F, Y158F and wt a-Gal enzymes remained after incubation at 25°C for up to 4 h,

whereas their activities decreased gradually while incu- bated at 37°C. Although only minimal activity was de- tectable from the mutant Y108F, its pattern was similar to the other two samples.

30

25

~o

5

- " •

I ! I I I

4 5 6 7 8 9 10

pH Fig. 4. Effects of pH on the activity of the wild-type and mutant a-Gal enzymes. Standard enzyme assays were curried out in 100 /xl of 1 mM PNP-a-Gal in 0.2 M sodium phosphate buffer with pH ranging from 4.7 to 9.0. The reactions were stopped by adding 1 ml of borate (pH 9.8), and the activity was then measured at 405 nm. ( • ) Wild-type t~-Gal, (&) mutant Y158F, and ( 0 ) mutant Y108F.

264 A. Zhu et al. / Biochimica et Biophysica Acta 1247 (1995) 260-264

In order to compare the effect of pH on the activities of mutant and wt a-Gal enzymes, activity was measured according to the method described in Section 2, except that 0.2 M phosphate buffer with pH ranging from 4.5 to 9 was used. As shown in Fig. 4, the mutant Y158F demonstrated very similar pH profiles as the wild-type. The activity peaked at about pH 6, and dropped sharply after pH 6.5. The same pH profile was observed with the a-Gal purified from coffee bean (data not shown) indicating its indistin- guishable nature with the recombinant a-Gal. The mutant Y108F was basically inactive throughout the whole pH range.

4. Discuss ion

Although a-Gals from different sources demonstrate significant sequence homology, they have distinctive phys- ical and chemical properties. For instance, glycosylation of coffee bean a-Gal has not been detected (unpublished data), while human a-Gal is a glycoprotein [13]. Coffee bean a-Gal shows higher Vraax/Km values toward oligo- saccharide substrates such as raffinose (450-fold), gal a- 1,3-Gal (180-fold) and galactomannan (30-fold) than that of a-Gal isolated from Ehrl ich ascites tumor cells [14]. Nevertheless, the high degree of sequence homology among different a-Gals suggests that certain common residues are involved in the active site. Based on a study on chemical modification of coconut a-Gal, Mathew and Balsubrama- niam [7,11] suggested that there are two ionizable groups with p K a values of 3.8 and 6.5 involved in binding and catalysis. They further predicted that the group with a pKa of 6.5 is a perturbed carboxylic group in the protonated form due to the hydrophobic environment produced by the presence of a tryptophan and a tyrosine residue in its vicinity. The availability of coffee bean a-Gal cDNA made it possible to study its enzymatic mechanism by means of site-directed mutagenesis.

Wild-type a-Gal expressed in Sf9 cells demonstrates similar characteristics as a-Gal purified from coffee bean in size, pH profile, heat stability and specific activity. The expression in tunicamycin-treated Sf9 cells did not affect the migration of wt a-Gal on SDS-PAGE, suggesting that there is little glycosylation, if any, of the enzyme. There- fore, it is reasonable to assume that the data obtained from site-directed mutagenesis will be applicable to wild type coffee bean enzyme.

As shown in Table 2, the replacement of Tyr-108 of coffee bean a-Gal with Phe (Y108F) almost abolished catalytic activity, suggesting that residue Tyr-108 is critical for a-Gal activity. Furthermore, this mutation caused a dramatic increase in K m value, implying its involvement in substrate binding. This result supports Mathew et al. 's data: chemical modification of one tyrosine residue in coconut a-Gal inhibited enzyme activity by 84%. Since Tyr-108 is also conserved in aNAGA, it seems reasonable

to assume that this residue is also important for its enzy- matic activity. Thus it may be that these two families of enzymes share a similar catalytic mechanism, although further studies are required to support this hypothesis. On the other hand, the replacement of Tyr-158 with Phe (Y158F) had only limited impact on a-Gal activity even though its Vmax/K m value dropped to about 60% of wt. In addition, this mutant demonstrated similar patterns for heat stability and pH profile as wt a-Gal. Based on these observations, we suggest that residue Tyr-158 is not criti- cal for the enzymatic activity under our test conditions. However it is possible that the residue might be important for the enzyme in a fashion which was not revealed by studies conducted here.

As shown in Fig. 3A, the enzymatic activity of both wt and mutant Y158F decreased gradually by incubating at 37°C. One possible reason might be the activation of endogenous proteinases present in Sf9 cell extracts at 37°C. Recently, we have expressed and secreted a-Gal by using a signal peptide sequence from a viral envelope glycoprotein. The enzyme secreted in the baculovirus cul- ture medium retained 100% of activity even after 5 h of incubation at 37°C (unpublished data).

In conclusion, the data presented here indicate that Tyr-108 of coffee bean a-Gal is critical for its activity. Further analysis by site-directed mutagenesis is required in order to identify the other important residues involved in the catalytic mechanism of coffee bean a-Gal.

References

[1] Goldstein, J., Siviglia, G., Hurst, R., Lenny, L. and Reich, L. (1982) Science 215, 168-170.

[2] Lenny, L.L., Hurst, R., Goldstein, J. and Galbraith, R.A. (1994) Transfusion 34, 209-214.

[3] Zhu, A. and Goldstein, J. (1994) Gene 140, 227-231. [4] Zhu, A. and Goldstein, J. (1993) Gene 137, 309-314. [5] Wang, A.M., Bishop, D.F. and Desnick, R.J. (1990) J. Biol. Chem.

265, 21859-21866. [6] Ishii, S., Kase, R., Sakuraba, H., Fujita, S., Sugimoto, M., Tomita,

K., Semba, T. and Syzuki, Y. (1994) Biochim. Biophys. Acta 1204, 265-270.

[7] Mathew, C.D. and Balasubramaniam, K. (1987) Indian J. Biochem. Biophys. 24, $29-$32,

[8] Sambrook, J., Fritsch, E.F. and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual, 2nd Edn., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

[9] Ho, S.N., Hunt, H.D., Horton, R.M. Pullen, J.K. and Pease, L.R. (1989) Gene 77 (1), 51-59.

[10] Grace, M.E., Graves, P.N., Smith, F.I. and Grabowski, G.A. (1990) J. Biol. Chem. 265 (12), 6827-6835.

[11] Mathew, C.D. and Balasubramaniam, IC (1986) Phytochemistry 25 (11), 2439-2443.

[12] Carson, D.A., Goldblum, R. and Seegmiller, J.E. (1977) J. Immunol. 118, 270-273.

[13] Murali, R., Ioannou, Y.A., Desnick, R.J. and BurneR, R.M. (1994) J. Mol. Biol. 239, 578-580.

[14] Yagi, F., Eckhardt, A.E. and Goldstein, I.J. (1990) Arch. Biochem. Biophys. 280 (1), 61-67.