Characterization and site-specific mutagenesis of the calcium ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry and Molecular Biology, Inc. Vol. 264, No. 6, Issue of February 25, pp. 3470-3477,1989

Printed in U.S.A.

Characterization and Site-specific Protein Oncomodulin Produced by

Mutagenesis of the Calcium-binding Recombinant Bacteria*

(Received for publication, September 16, 1988)

John P. MacManus, Cindy M. L. Hutnik, Brian D. SykesS, Arthur G. Szabo, Thomas C. Williams§, and Denis Banvillell From the Diuision of Biological Sciences, National Research Council, Ottawa, Ontario, Canada KIA OR6, the $Medical Research Council Group in Protein Structure and Function and the Department of Biochemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2H7, the $Department of Pharmacology, Medical University of South Carolina, Charleston, South Carolina 29425-2251, and the (Biotechnology Research Institute, National Research Council, Montreal, Quebec, Canada H4P 2R2

A bacterial expression system for the parvalbumin- like calcium-binding protein oncomodulin has been constructed. This system can yield 50-fold more oncomodulin than the richest known mammalian source, the rat Morris hepatoma 5123. The bacterially produced protein folded correctly as monitored by UV, fluorescence, and ‘H nuclear magnetic resonance spectroscopy and is immunologically identical to rat hepatoma oncomodulin. A calcium-specific conformational change is observed in the bacterial oncomodulin similar to that of the hepatoma protein. A modification of the putative calcium-specific CD loop by site-directed mutagenesis, which changed Asp-59 to Glu-59, elimi- nates calcium-specific effects. In contrast to the native molecule, the mutant Glu-59 now exhibits a magnesium-induced conformational change when monitored by UV difference or fluorescence excitation spectroscopy. The availability of large amounts of bacterially produced oncomodulin combined with the ability to modify at will the metal-binding ligands should now allow dissection of the unusual pairing in oncomodulin of one calcium-specific calmodulin-like site with one calcium/magnesium parvalbumin-like site.

A family of calcium-binding proteins genetically related to calmodulin has been recognized which have in common re- peated domains of approximately 30 residues comprising a he1ix:calcium-binding 1oop:helix motif (1-4). This family in- cludes proteins with four such domains such as calmodulin and troponin C, which are calcium-dependent modulators capable of binding to enzymes (4-6). Other members are smaller proteins with three domains such as parvalbumin or two domains such as calbindin 9K. However, these smaller proteins have not been shown to bind to other target proteins (7-9). In addition, the two functional loops on parvalbumin and the two COOH-terminal loops on troponin C are capable of binding magnesium. It is not known what confers calcium specificity or calcium-dependent protein-protein interaction to these members of the family.

Oncomodulin is a member of this family, sharing 50% identity of amino acid sequence with rat parvalbumin (10). It is unique in being expressed during prenatal development in the placenta but not in the fetus (10-12). In addition, it is noteworthy in not having been detected in normal adult

* This is National Research Council of Canada Paper 29886. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

human or rodent tissues. However, following neoplastic transformation by viruses or chemicals, oncomodulin is found in a majority of tumors (10, 12). This apparent de rwuo activation of oncomodulin expression in the adult is in contrast to other calcium-binding proteins whose concentrations in a particular tissue are merely increased above normal following neoplastic transformation (13-15).

The function of oncomodulin during normal development and its role in carcinogenesis are unknown. Despite its simi- larity to parvalbumin in primary structure, oncomodulin is also noteworthy in apparently having a calcium-specific site, positive cooperativity of binding between the two sites, and the capability of activating some enzymes such as phosphodiesterase or calcineurin in a calcium-dependent manner (16- 18). Since oncomodulin is relatively small, studies to define what delineates a calcium-specific site from a calcium-magnesium site or what confers the ability to interact with other proteins should be easier using oncomodulin than the larger four-domain proteins such as calmodulin or troponin C.

The major limiting factor in studying such matters has been the limited supply of oncomodulin. The richest known source of the protein remains the transplantable hepatomas in which this protein was first detected (19). However, maintenance of these tumors is restricted to the specific Buffalo strain of rat and is expensive in animal care, time, and effort. We report herein the establishment of a bacterial expression system for oncomodulin which permits the isolation of this protein in large quantities, thereby removing the previous impediment to investigation of its unique properties.

MATERIALS AND METHODS

Construction of Expression Plasmids-A bacterial vector for expression of oncomodulin was constructed from the coding sequence of an oncomodulin cDNA joined to the Tac promoter (20, 21) (Tac promoter Genblock, Pharmacia LKB Biotechnology Inc.) inserted in the plasmid pGEM-1 (Promega Biotec). The oncomodulin coding sequence was obtained from pONCO-4, a plasmid containing the entire oncomodulin sequence as well as 73 nucleotides of the 5’-noncoding sequence and 253 nucleotides of the 3’-noncoding sequence of the oncomodulin messenger RNA (20). All synthetic oligonucleotides were made by the phosphoramidite method using an Applied Biosys- tems 380A synthesizer. The strategy used is described in Fig. L4. The construction was designed so that the Shine-Dalgarno polymerase- binding sequences were immediately upstream of the ATG start codon of the oncomodulin coding sequence. A new unique ClaI restriction site was also created by this ligation, which proved useful in screening for the desired recombinant. The resulting nucleotide sequence of the junction between the Tac promoter and the oncomodulin coding sequence in the new plasmid was determined from the SP6 promoter region, which is upstream of the Tac promoter, by the dideoxy chain termination method (Fig. 1B) (22).

Oncomodulin Expression and Purification-The plasmid pGEM-

3470

Bacterially Expressed Oncomodulin 3471

B. cLn I

110

FIG. 1. A, the Tac promoter was introduced into pGEM-1 as a HindIII-BamHI fragment to yield pGEM-TAC (steps 1-3). The oncomodulin coding sequence was obtained from pONCO-4 (20) after digestion with HindIII followed by treatment with Ba131 (which deleted the 5"noncoding sequence and part of the coding region) and finally digestion with DraI (steps 5 and 6). The blunt-ended fragments were cloned into pGEM-TAC which had been prepared by digestion with BamHI and the resulting ends filled with the large fragment of DNA polymerase I (Klenow) (step 4). A subclone in which a RamHI restriction site had been regenerated was chosen (step 7) and the junction sequenced. This plasmid was then linearized with BamHI, and the ends were rendered blunt by treatment with mung bean nuclease (step 8). A double-stranded DNA fragment of 15 base pairs (step 9) composed of two complementary synthetic oligonucleotides was then introduced between the Tac promoter and the oncomodulin sequence to restore the entire coding region (step IO). In addition, this generated a new unique ChI restriction site and a Shine-Dalgarno sequence, AGGA, four nucleotides upstream of the start codon. B, DNA sequence of the junction region showing part of the Tac promoter with its two ribosome-binding sites ( S / D ) located immediately upstream of the ATG start codon. The sequence of one of the two complementary synthetic oligonucleotides inserted between the Tac promoter and the oncomodulin coding sequence is shown in smaller characters. T>

G> TAC-ONCO was used successfully to transform Escherichia coli JM101, JM103, DH5, or GW5889, the latter being Ion- (i.e. free of endogenous Ion) protease. The transformants were screened for oncomodulin production. The isolation of oncomodulin from bacteria was based on procedures used for calmodulin (23, 24). Bacteria were harvested from 2 liters of culture in L-broth by centrifugation and the pellet (approximately 5-7 g) resuspended in 20 ml of 2.4 M sucrose, 40 mM Tris-HCI, 10 mM EDTA, pH 8.0. The suspension was incu- bated on ice for 30 min. 80 ml of 50 mM Hepes'-HCI, 100 mM KCI, 1 mM EDTA, 1 mM dithiothreitol, 100 pg/ml lysozyme were then added and the bacteria lysed at 4 "C overnight. The lysed bacteria were centrifuged at 40,000 X g for 30 min in a Beckman LS65 ultracentri- fuge (Ti-60 rotor). The clear supernatant was heated rapidly in a

boiling water bath, held between 65 and 80 'C for 5 min, and cooled immediately on ice. The denatured proteins were removed by centrifugation at 40,000 X g for 30 min. The oncomodulin in the resulting supernatant was isolated by ammonium sulfate precipitation followed by sequential ion exchange and gel filtration as described previously for extracts of tumor tissue (25, 26).

Oncomodulin Mutagenesis-The fragment of pGEM-TAC-ONCO composed of the sequence from the HindIII to the XmaI site (which was originally in the polylinker of pGEM but now is 230 nucleotides downstream of the TAA in Fig. L4, 10) was subcloned into the polylinker region of pTZ19R (27) (Pharmacia) to produce pTZ19- ONCO. Single-stranded DNA was obtained from the resulting recombinant after infection of a bacterial culture with helper phage M13K07 (Pharmacia) using the procedure of Zoller and Smith (28). A synthetic oligonucleotide (21-mer) containing the desired mutation (ex. for Glu-59, TGGATACCTCGAGGGAGATGAG) was used to prime the synthesis of the second DNA strand before transformation of com- petent E. coli JM103. A new XhoI site was also created which was useful in screening. The kinase-labeled oligonucleotide was also used to screen the resulting transformants, and the nucleotide sequence of the choice candidates was determined by the dideoxy chain termination method (22) (Fig. 2).

Characterization of Recombinant Oncomodulin-Ultraviolet spectra were obtained with a Beckman DU8 spectrophotometer using oncomodulin at 5 mg/ml in a buffer consisting of 10 mM sodium cacodylate, 150 mM KCI, 1 mM dithiothreitol, pH 7.0, which had been passed over the cation exchange resin Chelex 100 (Bio-Rad). The resulting working buffer had a residual calcium concentration of less than 0.002 mM estimated by atomic absorption spectrometry (Pye Unicam SP191).

The fluorescence spectra were obtained with an SLM 8000C spec- trofluorimeter equipped with a Neslab Instruments, Inc. Endocal RTE-5DD circulating bath. The spectra were corrected for contri- butions of the blank and normalized for comparison. The fluorescence (Lx = 280 nm) of a solution of oncomodulin in the cacodylate buffer described above was measured in 0.5-cm quartz cuvettes at 20 'C (absorbance readings at 280 nm were approximately 0.05). The excitation and emission band passes were both 4 nm. Anisotropy measurements were made using Glan-Thompson polarizers. The stock of metals was prepared and used as described (29).

400-MHz 'H NMR spectra of oncomodulin were recorded on a

G A T C G A T C

*G

*C

' The abbreviations used are: Hepes, 4-(2-hydroxyethyl)-l-pipera- zineethanesulfonic acid; HPLC, high pressure liquid chromatography; SDS, sodium dodecyl sulfate; CD, the helix C-loop-helix D region of oncomodulin; EF, the helix E-loop-helix F region of oncomodulin.

FIG. 2. Proof of the correct plasmid sequence for expression of oncomodulin mutant Glu-59. The DNA sequence of the mutant Glu-59 and the native Asp-59 is shown where the native sequence for the 59th amino acid has been altered from GAT to GAG.

3472 Bacterially Expressed Oncomodulin

- kD 92 -

I- - N a

68-

45 -

n-

21 -

14-

FIG. 3. SDS-polyacrylamide gel electrophoresis is shown from various stages of purification of oncomodulin from E. coli JM103 transformed with pGEM-TAC-ONCO. SN, supernatant; hSN, heated supernatant; AMs, ammonium sulfate.

Varian XL-400 spectrometer operating in the conventional pulsed Fourier transform mode. NMR spectral parameters were: sweep width, 4,500 Hz; acquisition time, 1.5 s; preincubation delay, 1.5 s; pulse width, 17 ps (90"); and the number of acquisitions was 10,000. Sample conditions were: 1.25 mM oncomodulin in 150 mM KCI, pH 6.8,60 "C in 90% H20,10% D20. Presaturation (1 s, decoupling gated off during acquisition) was used to suppress the Hz0 resonance.

The antigenic comparison of the recombinant oncomodulin with the native protein from hepatoma was made using an immunoradiometric assay (31).

The tryptic peptides of recombinant oncomodulin were separated by reverse-phase HPLC and their amino acid composition and sequence obtained as described (32).

RESULTS

Expression-The construction of the expression system is shown in Fig. U, and the sequence of the TAC junction shows it is within the required number of bases of the initiation ATG for correct expression. A similar system has been used for the expression of calmodulin (24). However, the addition of isopropyl 8-D-thiogalactopyranoside (1 mM) had no effect on oncomodulin expression in E. coli JM103. Expression in other strains of E. coli (i.e. JM101, DH5) gave no useful improvement nor did expression in the endogenous protease- free (lon-) strain GW5889 (not shown).

Purification-When the extracts of recombinant bacteria were subjected to SDS gel electrophoresis, a band could be seen at about 14 kDa where the 11.7-kDa oncomodulin runs anomalously (Fig. 3). Precipitation of proteins from the extract by heat (65-80 "C) and by ammonium sulfate (65% of saturation) removed many contaminants and greatly enriched the visible putative oncomodulin. This mixture was then subjected to ion exchange chromatography, and the sole calcium-binding peak was pooled (Fig. 44). This again led to

FRACTION

- I00

-so f 0,

-60 n

g-40 $

8 - 2 0 7 -F

-0 A

0 . 2 i , e ! I 5 r OO 20 40 €0 80

O L ;s &3 a0 ;z FRACTION 050 FWCrrm

FIG. 4. Purification of oncomodulin from E. coli JM103. A shows DEAE-Sephacel (0.9 X 30 cm) ion exchange chromatography in 20 mM imidazole, 20 mM NaCI, pH 6.5, of the proteins remaining in solution after treatment of the extract with 65% saturated ammonium sulfate. Fractions from the salt gradient containing calcium- binding activity by Chelex assay (40) were pooled and subjected to (23) gel filtration on Sephadex G-75 (1.5 X 90 cm) in 20 mM imidazole, 150 mM NaCI, 50 mM KCI, 1 mM MgCI2, pH 7.0, to remove a small amount of large molecular mass contaminants. The major protein peak was symmetrical for absorbance at 280 nm and for calciurn- binding activity, and all fractions showed one band on SDS gels (inset).

further enrichment of SDS gels (Fig. 3, DEAE lane). The pooled fractions from DEAE were concentrated and subjected to gel filtration (Fig. 4B). Again, a single peak of calcium- binding activity was observed which appeared homogenous by SDS gel electrophoresis (Fig. 4, inset). Table I shows the outcome of purification from a typical batch of 7 g of packed bacteria with a yield of 50 mg of protein, which amounted to approximately 7% of cellular protein. The material at this stage was also shown to consist of one peak by reverse-phase HPLC (32) which co-eluted with hepatoma oncomodulin (not shown).

Characterization-The UV spectrum of this bacterial oncomodulin was identical to that of the hepatoma protein (Fig. 5) with both the absorption bands due to Phe and Tyr being evident.

A comparison of the primary structure of the oncomodulin from both the bacterial and hepatoma sources was initiated by generation of tryptic peptide maps by HPLC (Fig. 6) and amino acid analysis of each peptide (Table 11). The bacterial tryptic peptides r01-rO9 co-eluted with and had identical composition to the hepatoma counterparts which spanned residues 29-108 of oncomodulin. The peptides r010-r012 eluted at approximately the same time as the NHz-terminal hepatoma peptide and had a similar amino acid composition. When these three peptides were sequenced, 1010 gave the sequence Met-Ser-Ile-Thr-Asp-Ile-Leu-Ser-Ala-Glu-Asp-Ile- Ala-Ala-Ala-Leu-Gln-Glu-Cys-Gln-Asp-Pro-Asp-Thr-Phe- Gln-Pro-Gln-Lys,whichareresidues 1-28ofoncomodulin (32); r o l l gave Met-Ser-Ile-Thr ... Gln-Pro-Gln-Lys, and r012 appeared inexplicably identical to r011. This analysis led to the conclusion that the bacterial oncomodulin was identical to the hepatoma protein in all 108 residues but with 50-60% of


TABLE I Purification of oncomodulin from 7 g, wet weight, of E. coli JM103

Volume Protein Recovery Oncomodulin"

ml d m 1 total mg % d m 1 Homogenate 100.00 6.50 650.00 100.00 600.00 100,000 x g S N b 75.00 5.80 435.00 67.00 590.00 SN after heat 75.00 1.50 113.00 17.00 585.00 Dialyzed 65% AMS' SN 300.00 0.26 78.00 12.00 220.00 DEAE to load Sephadex G-50 2.50 18.00 45.00 6.90 SeDhadex G-50 25.00 58.00d

Measured by immunoradiometric assay (31). SN, supernatant.

Freeze-dried. e AMs, ammonium sulfate.

u O220 250 300 350

nrn

FIG. 5. UV spectra of the bacterially expressed (br) oncomodulin showing identity with the material from rat hepatoma.

MINUTES

FIG. 6. Tryptic peptide map from reverse-phase HPLC. A, bacterially expressed (br) oncomodulin is compared with that of B, which is material from rat hepatoma. In B the inclusive span of amino acid sequence positions of each peptide is indicated. Each peptide in A was subjected to amino acid analysis (Table 11).

the expressed protein still retaining the initiation amino acid N-formyl-Met.

Since amino acid analysis cannot clearly show the presence of N-formyl-Met, the presence of this amino acid was confirmed by 'H NMR whereby the singlet resonance of an additional Met-CHa was clearly visible at 2.01 ppm (Fig. 7). Further, one of the two methyl resonances near 2.09 ppm, due to the methyl protons of the N-acetyl group of oncomodulin, is absent in bacterial oncomodulin. The assignment in Fig. 7 is reversed from that of Ref. 30 on the basis of this bacterial oncomodulin spectrum. The integrated area under the new N-formyl-Met resonance at 2.01 ppm is 47% of that of the resonance of Met-38, which is very close to the expected

proportion from the peptide analysis (see above). The gross tertiary structure of the bacterial oncomodulin

was qualitatively determined by 'H NMR spectroscopic mon- itoring of upfield methyl resonances (Fig. 8) with some as- signable to Val-106 (0.13-0.15 ppm) and Ile-97 (0.49-0.53 ppm) (30). This showed that the hydrophobic core of the bacterial oncomodulin was essentially identical to the hepatoma protein. In addition, the most downfield shifted amide- NH resonances (Fig. 9) are essentially identical in the two proteins, indicating very similar secondary structures and hydrogen-bonding patterns in both bacterial and hepatoma proteins.

Finally, the bacterial oncomodulin was shown to be immunologically identical to the hepatoma protein when a two-site immunoradiometric assay was used (Fig. 10).

Mutant Oncomodulin Expression-The conclusion arising from previous studies on hepatoma oncomodulin was that the calcium-specific conformational change over and above that induced by magnesium was caused by calcium binding to the CD loop (29). In other words, unlike the CD loop of parvalbumin, the CD loop of oncomodulin apparently did not bind appreciable magnesium. Comparison of the sequences of CD loops of both parvalbumin and oncomodulin shows several differences in the metal-binding ligands. One of particular note is the -X position, an Asp-59 in oncomodulin, but a Glu-59 in parvalbumin. This residue has been pinpointed in lanthanide-substitution experiments as being potentially im- portant in causing differential binding of these metals to the CD uersus the EF loops (30). This residue was therefore chosen as one of the first to modify by site-specific mutagenesis.

The construction of the expression system for mutant oncomodulin was based on the pGEM-TAC-ONCO in Fig. lA, and the selected mutant plasmid was shown to have the desired base change by DNA sequencing (Fig. 2). This protein, Glu-59, was purified and characterized as described above for the native protein. The yield was only about one-third to one- half that of the native protein for unknown reasons. The tryptic peptide map of Glu-59 had only one peptide different from that of the native molecule shown in Fig. 6. Amino acid analysis of the peptide equivalent to r 0 8 showed 1 less Asp and 1 more Glu (not shown).

The UV difference spectrum of aporecombinant native oncomodulin uersus either the magnesium- or calcium-loaded form was obtained (Fig. 11). As seen previously with the hepatoma protein (29), magnesium had little effect, but calcium induced a significant conformational change in the bacterially expressed oncomodulin (Fig. 1lA). There was little of significance in the difference spectrum of the calcium forms of this protein of native sequence (i.e. Asp-59) compared with the mutant Glu-59 (Fig. 11B). However, the Glu-59 protein

3474

n IN-formyl Met

Bacterially Expressed Oncomodulin

TABLE I1 Amino acid compositions of tryptic peptides of bacterially expressed oncomodulin

Parentheses show residues/mol of equivalent peptide from hepatoma oncomodulin. Peptide r01 r 0 2 r 0 3 r 0 4 r05 r 0 6 r 0 7 r 0 8 r09

Asx 0.96 (1) 1.05 (1) 5.34 (5) 4.82 (5) 1.16 (1) Thr 1.34 (2) 0.84 (1) Ser 0.86 (1) 0.95 (1) 1.89 (2) 1.05 (1) 1.74 (2) 1.06 (1) 0.86 (1) Glx 1.15 (1) 3.15 (3) 1.78 (1) 0.99 (1) 1.05 (1) 1.85 (2) 3.20 (3) Pro GlY 2.04 (2) 1.12 (1) 2.19 (2) 1.15 (1) Ala 1.11 (1) 1.35 (1) 2.02 (2) 1.05 (1) CYS Val 1.32 (1) 0.97 (1) Met 0.82 (1) 0.72 (1) 0.62 (1) Ile 0.98 (1) 1.05 (1) 0.98 (1) Leu 1.13 (1) 1.08 (1) 1.11 (1) 1.06 (1) 2.29 (2)

P he 0.9 (1) 1.05 (1) 1.11 (1) 1.98 (2) 1.12 (1) 1.11 (1) His 0.99 (1) LYS 1.19 (1) 0.86 (1) 1.05 (1) 0.96 (1) 1.06 (1) 0.93 (1) Arg 1.02 (1) 0.92 (1) Deduced sequence 70-75 76-83 38-44 45-48 84-96 65-69 29-37 49-64 97-108

Tyr 0.82 (1) 0.71 (1)

)'Ut:, Hepatoma Oncomodulin

2 1 2 0 1 . 9 1 8 PPM

FIG. 7. 400-MHz 'H NMR spectra of oncomodulins showing the resonances attributable to the methionine residues and the initiation N-formyl-Met in the bacterially expressed (br) oncomodulin. Assignments for closely spaced methyl resonances of the N-acetyl (N-Ac) group and Met-38 are reversed from those of Ref. 30 based on the bacterially expressed oncomodulin.

now had a much larger response to magnesium than did the native Asp-59 protein (Fig. 1lC).

Fluorescence-Binding of magnesium to the Glu-59 oncomodulin mutant resulted in changes in the fluorescence properties when compared with the native protein. On binding magnesium, the fluorescence anisotropy of the Glu-59 mutant increased significantly relative to that of the apoprotein (Ta- ble 111). In the case of the native protein, binding of magnesium did not significantly alter the anisotropy value.

The corrected fluorescence excitation spectra (Aern = 306 nm; normalized at 280 nm) of the apo-, magnesium, and calcium forms of native and Glu-59 mutant proteins also indicated increased sensitivity of the Glu-59 oncomodulin to magnesium. As reported for the hepatoma protein (29), the excitation spectra of the apo- and magnesium forms of the bacterial oncomodulin were essentially superimposable, while the calcium form displayed a considerable increase in inten- sity and an altered shape at the blue edge (Fig. 12). In contrast, the excitation spectrum of the magnesium form of the Glu-59

Met 38 liN-Ac AMet IO5

\ FIG. 8. 400-MHz 'H NMR spectra of the upfield methyl

resonances of oncomodulins showing that the bacterially (br) expressed oncomodulin assumes the hydrophobic core char- acteristic of the hepatoma protein (see also Ref. 30).

mutant was no longer superimposable upon the apo- form, but exhibited a blue edge shape change intermediate between the apo- and calcium forms of this mutant protein (Fig. 13).

DISCUSSION

The best known source for oncomodulin has been the rat Morris hepatomas (10, 12, 19) which yield 0.1 mg of protein/ g, wet weight. The bacteria described here yield at least 5 mg of protein/g of bacteria. The amount of oncomodulin expressed (approximately 7% of total protein) does not appear to depend on the strain of E. coli (JM101, JM103, DH5) or whether the strain is lon- and low in endogenous protease activity (GW5889). Neither does the addition of isopropyl /3- D-thiogalactopyranoside further increase the amount of oncomodulin obtained despite being controlled by the Tac promoter.

The oncomodulin synthesized by bacteria is the intact molecule and appears to fold correctly as shown by comparison with the hepatoma molecule using UV spectroscopy and 'H NMR of core or p-sheet side chains. In addition, accurate folding is evidenced by complete immunological identity of


1, br Oncomodulin

I1 Oncomodulin

1 0 6 1 0 4 1 0 2 1 0 0 9 8 9 6 9 4 9 2 9 O P P M

FIG. 9. 400-MHz 'H NMR spectra of the downfield amide- NH resonances of oncomodulins showing that the bacterially expressed (br) oncomodulin has @-sheet contacts between the metal-binding loops similar to the protein from rat hepatoma.

; A I b A 3 & I d 0 ng/rnl

FIG. 10. Antigenic cross-reactivity of the bacterially expressed (br) oncomodulin is identical to that of the protein from rat hepatoma. A two-site immunoradiometric assay was used with solid phase rabbit anti-oncomodulin IgG in conjunction with iodinated goat anti-oncomodulin IgG (31).

FIG. 11. UV difference spectra of bacterially expressed native oncomodulin (Asp-59) and a mutant (GZu-59). The protein was stripped of bound metals by acid precipitation (29). The difference is plotted between the apoprotein and a parallel sample to which either calcium (2.5 mM final) or magnesium (5.0 mM final) has been added. A, the response of the native Asp-59 protein; B, a comparison of the response to the addition of calcium to Asp-59 protein or to the mutant Glu-59; C, demonstration of a response to magnesium of the mutant Glu-59 in comparison with the native Asp-59 protein.

TABLE I11 Measurements of steady-state fluorescence anisotropy

Samples were measured in 10 mM cacodylate, 150 mM KCl, 1 mM dithiothreitol, pH 7. Magnesium and calcium were added to a 2.5 mM final concentration. Samples were excited at 280 nm and emission measured at 305 nm.

Recombinant protein Anisotropy

Aponative (Asp-59) 0.151 Mg form 0.154 Ca form 0.190

Apomutant (Glu-59) 0.116 Mg form 0.127 Ca form 0.187

f Native: Asp 59

nm

FIG. 12. Fluorescence excitation spectra of aponative Asp- 59 bacterially expressed oncomodulin and its response to the addition of either magnesium or calcium. A significant calcium- specific conformational change was observed. The concentration of protein was .30-40 p M (Aex = 280 nm; A ~ N , approximately 0.05 in 0.5- cm quartz cuvettes). Spectra were normalized at 280 nm.

nm

FIG. 13. Fluorescence excitation spectra of apomutant Glu- 59 and its response to the addition of either magnesium or calcium. In contrast with the native Asp-59 protein (Fig. 12), a definite conformational response to magnesium can now be observed. Conditions are similar to those in legend of Fig. 12.

the protein from both sources. The subtle differences between the NMR spectra are attributed to the differences in the NHz termini of the molecules from the bacterial and tumor sources. The hepatoma oncomodulin commences with N-acetyl-Ser- Ile (32), but the bacterial oncomodulin appears to be an approximately 1:l mixture with termini N-formyl-Met-Ser- Ile and NHz-Ser-Ile. This cannot easily be explained because for bacterial methionine aminopeptidase, oncomodulin has one of the most readily cleavable sequences, Met-Ser (33). The retention of 50% uncleaved initiation Met is not consid-

3476 Bacterially Expressed Oncomodulin

ered to be due to the large amount of a single protein being processed (7% of total protein) because some other oncomodulin mutants are expressed a t much lower levels (0.5% of protein) but still retain 50% of oncomodulin with N-formyl- Met.' This mixture of NH2 termini in recombinant oncomodulin is in contrast with recombinant calbindin 9K, where 100% of protein retains the Met (34) or calmodulin, which has it entirely cleaved off (23). The presence of this NH2- terminal Met has no observable effect on the tertiary structure of recombinant calbindin 9K (34) or on the function of recombinant troponin C (35).

Parvalbumins contain two metal-binding sites per molecule which both bind calcium and magnesium competitively (9). The unusual pairing of one calcium-specific site with one calcium/magnesium site in oncomodulin was originally suggested from UV, circular dichroic, and fluorescence spectroscopic studies (29). It was later confirmed by studies using lanthanide fluorescence (36) or lanthanide exchange (30). The reason for the specificity of the calcium sites in calmodulin/ troponin C uersw the calcium/magnesium sites of troponin C/parvalbumin is not known. Calcium-binding constants of 107-10s have been observed within the family of metal-binding proteins, but there is no obvious correlation with the nature or the position of the octahedral chelating ligands.

In oncomodulin the CD site is the one most different from parvalbumin and was suggested to be the calcium-specific site (29). Binding of magnesium to the native protein (presumably at the EF site) induces a conformational change in the protein. When calcium binds to the CD loop an additional specific conformational change in the protein structure is observed which is not observable in the presence of excess magnesium. This was clearly shown in the excitation spectra of the native protein (e.g. Fig. 12). In troponin C the ligand in the -X position is Asp, and in oncomodulin it is also Asp, whereas in most parvalbumins and in the E F site of oncomodulin it is a Glu (4, 32). This smaller side chain is thpught to favor the bulkier calcium ion (ionic ra$us 0.99 A) compared with magnesium (ionic radius 0.65 A), thereby conferring calcium specificity (30). In an attempt to determine whether this Asp- 59 is involved in calcium specificity, we mutated it toward parvalbumin by changing it to a Glu. The excitation spectra of the Glu-59 oncomodulin show that magnesium is now capable of inducing a conformational change in this mutant which it did not do in the native protein. This would result if magnesium were now binding to the CD loop. That this single amino acid change undoubtedly allowed the oncomodulin CD site to respond to magnesium is seen also by UV difference (Fig. 11) and fluorescence anisotropy measurements (Table 111). It is not suggested that Asp-59 alone confers calcium specificity. There are other differences between the CD sites of oncomodulin and parvalbumin. Modification of Gly-56 to Ala, Leu-58 to Ile, Gly-60 to Glu, and Glu-62 to Asp should prove informative and hopefully should give some further insight into the nature of calcium specificity. I t is also realized that the flanking helices do contribute to metal-binding affin- ity.

The construction of this bacterial expression system for oncomodulin should relieve the supply problem which has blocked inquiry into its function. This first attempt using site-directed mutagenesis to modify the CD site of oncomodulin in the direction of parvalbumin has established the feasibility of the approach. This may be only a partial answer especially in light of evidence that the coordination of magnesium differs from that of calcium in troponin C (37). It has

J. P. MacManus, C. M. L. Hutnik, B. D. Sykes, A. G. Szabo, T. C. Williams, and D. Banville, unpublished observations.

been presumed here that the CD and EF sites are the only places where magnesium can bind. The presence of additional putative magnesium-binding sites in the NH2-terminal flanking helices of each site in calmodulin has been suggested (38). However, clusters of Asp or Glu which are suggested to bind this additional magnesium do not occur in oncomodulin. Additional information may come from the three-dimensional structure of oncomodulin which will be available soon (39) although crystallization and analysis of the magnesium form may be necessary to answer fully the question of magnesium chelation. The availability of this expression system will also allow studies of the interaction domain of oncomodulin with enzymes such as cyclic-AMP phosphodiesterase or calcineurin (16-18).

Acknowledgments-We thank Dr. M. Yaguchi and D. Watson (Division of Biological Sciences, National Research Council of Can- ada, Ottawa) for amino acid analysis and sequencing of peptides.

REFERENCES

1. Baba, M. L., Goodman, M., Berger-Cohn, J., Demaille, J. G. &

2. Kretsinger, R. H. (1980) CRC Crit. Reu. Biochem. 8, 119-174 3. Van Eldik, L. J., Zendequi, J . G., Marshak, D. R. & Watterson,

D. M. (1982) Znt. Reu. Cytol. 77, 1-61 4. Grand, R. J. A. (1985) in Metalloproteins. Part 2: Metal Proteins

with Non-redox Roles (Harrison, P. M., ed) pp. 65-122, Mac- millan Press Ltd., Hampshire, England

5. Klee, C. B., Crouch, T. H. & Richman, P. G. (1980) Annu. Reu. Biochem. 49,489-515

6. Potter, J. D. & Johnson, J. D. (1982) in Calcium and Cell Function (Cheung, W. Y., ed) Vol. 11, pp. 145-173, Academic Press, New York

7. Demaille, J. G. (1982)in Calcium and Cell Function (Cheung, W. Y., ed) Vol. 11, pp. 111-144, Academic Press, New York

8. Heizmann, C. W. (1984) Experientia 440,910-921 9. Wnuk, W., Cox, J. A. & Stein, E. A. (1982) in Calcium and Cell

Function (Cheung, W. Y., ed) Vol. 11, pp. 243-278, Academic Press, New York

10. MacManus, J. P. & Whitfield, J. F. (1983) in Calcium and Cell Function (Cheung, W. Y., ed) Vol. IV, pp. 411-440, Academic Press, New York

Matsuda, G. (1984) Mol. Biol. Euol. 1, 442-455

11. Brewer, L. M. & MacManus, J. P. (1985) Deu. Biol. 112, 49-58 12. MacManus, J. P., Brewer, L. M. & Gillen, M. F. (1987) in The

Role of Calcium in Biological Systems (Anghileri, L. J., ed) Vol. IV, pp. 1-19, CRC Press, Inc., Boca Raton, FL

13. Watterson, D. M., Van Eldik, L. J., Smith, R. E. & Vanaman, T. C. (1976) Proc. Natl. Acad. Sci. U. S. A. 73, 2711-2715

14. Dorin, J. R., Novak, M., Hill, R. E., Brock, D. J. H., Secher, D. S. & van Heyningen, V. (1987) Nature 326,614-617

15. Heizmann, C. W. & Berchtold, M. W. (1987) Cell Calcium 8, 1- 41

16. MacManus, J . P. (1981) F E B S Lett. 126, 245-249 17. Mutus, B., Karuppiah, N., Sharma, R. K. & MacManus, J. P.

(1985) Biochem. Biophys. Res. Commun. 131,500-506 18. Mutus, B., Palmer, E. J. & MacManus, J. P. (1988) Biochemistry

27,5615-5622 19. MacManus, J. P. & Brewer, L. M. (1987) Methods Enzymol. 139,

156-168 20. Gillen, J. F., Banville, D., Rutledge, R. G., Narang, S., Seligy, V.

L., Whitfield, J . F. & MacManus, J. P. (1987) J. Biol. Chem.

21. Russell, D. R. & Bennett, G. N. (1982) Gene (Amst.) 20,231-243 22. Sanger, F. G., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl.

Acad. Sci. U. S. A. 74,5463-5467 23:Roberts, D. M., Crea, R., Malecha, M., Alvarado-Urbina, G.,

Chiarello, R. H. & Watterson, D. M. (1985) Biochemistry 2 4 ,

24. Putkey, J. A,, Slaughter, G. R. & Means, A. R. (1985) J. Biol.

25. MacManus, J. P. (1980) Biochim. Biophys. Acta 621,296-304 26. Durkin, J . P., Brewer, L. M. & MacManus, J. P. (1983) Cancer

27. Mead, D. A., Szczesna-Skorupa, E. & Kemper, B. (1986) Protein

262,5308-5312

5090-5098

Chem. 260,4704-4712

Res. 43,5390-5394

Eng. 1 , 67-74


28. Zoller, M. J. & Smith, M. (1983) Methods Enzymol. 100, 468-

29. MacManus, J. P., Szabo, A. G. &Williams, R. E. (1984) Biochem.

30. Williams, T. C., Corson, D. C., Sykes, B. D. & MacManus, J. P.

31. Brewer, L. M. & MacManus, J. P. (1987) Placenta 8, 351-363 32. MacManus, J. P., Watson, D. C. & Yaguchi, M. (1983) Eur. J.

Biochem. 136 , 9-17 33. Ben-Bassat, A., Bauer, K., Chang, S.-Y., Myambo, K., Boosman,

A. & Chang, S . (1987) J. Bacteriol. 169, 751-757 34. Brodin, P., Grundstrom, T., Hofmann, T., Drakenberg, T., Thu-

lin, E. & Forsen, S. (1986) Biochemistry 25 , 5371-5377

500

J. 220,261-268

(1987) J. Biol. Chem. 262,6248-6256

35. Reinach, F. C. & Karlsson, R. (1988) J. Biol. Chem. 263 , 2371- 2376

36. Henzl, M. T., Hapak, R. C. & Birnbaum, E. R. (1986) Biochim. Bwphys. Acta 872,16-23

37. Satysur, K., Rao, S. T., Pyzalska, D., Drendel, W., Greaser, M. & Sundaralingam, M. (1987) in Calcium Binding Proteins in Health and Disease (Norman A. W., Vanaman, T. C. & Means, A. R., eds) pp. 385-387, Academic Press, New York

38. COX, J. A. (1988) Biochem. J. 249,621-629 39. Przybylska, M., Ahmed, F. R., Birnbaum, G . I. & Rose, D. R.

40. Wasserman, R. H., Corradino, R. A. & Taylor, A. N. (1968) J. (1988) J. Mol. Biol. 199,393-394

Biol. Chem. 243,3978-3986

Characterization and site-specific mutagenesis of the calcium ...

Documents

Transcript of Characterization and site-specific mutagenesis of the calcium ...