THE JOURNAL OF BIOLOGICAL Vol. 266, No. 29, Issue of ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 266, No. 29, Issue of October 15, pp. 19688-19696,1991 Printed in V.S.A. Cloning,

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 266, No. 29, Issue of October 15, pp. 19688-19696,1991 Printed in V.S.A.

Cloning, Bacterial Expression, Purification, and Characterization of the Cytoplasmic Domain of Rat LAR, a Receptor-like Protein Tyrosine Phosphatase*

(Received for publication, February 26, 1991)

David A. Pot, Terry A. Woodford, Eumorphia Remboutsika, Randy S . Haun, and Jack E. Dixon$$ From the Department of Biochemistry and the Walther Cancer Institute, Purdue University, West Lafayette, Indiana 47907

This report describes the cloning and characteriza- tion of rat leukocyte common antigen-related protein (rLAR), a receptor-like protein tyrosine phosphatase (PTPase). The recombinant cytoplasmic PTPase do- main was expressed at high levels in bacteria and purified to homogeneity. Kinetic properties of the PTPase were examined along with potential modula- tors of PTPase activity. Several sulfhydryl-directed reagents were effective inhibitors, and a surprising distinction between iodoacetate and iodoacetamide was observed. The latter compound was an extremely poor inhibitor when compared to iodoacetate, suggesting that iodoacetate may interact selectively with a posi- tive charge at or near the active site of the enzyme. Site-directed mutants were made at 4 highly conserved cysteine residues found at positions 1434,1522,1723, and 1813 within the protein. The Cys-1522ISer mu- tation resulted in a 99% loss of enzymatic activity of the pure protein. This observation is consistent with greater than 99% of the PTPase activity being found in the first domain of the PTPase and demonstrates the critical importance of this cysteine residue in catalysis. The recombinant C1522S mutant phosphatase could also be phosphorylated in vitro by protein kinase C and ~ 4 3 ‘ “ ~ ~ tyrosine kinase. When pure recombinant PTPase was mixed with “P-labeled tyrosine substrate and then rapidly denatured, a 32P-labeled enzyme in- termediate could be trapped and visualized by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The catalytically inactive C1522S mutant did not form the phosphoenzyme intermediate.

Tyrosine phosphorylation plays an important role in the regulation of cell proliferation, differentiation, and transfor- mation (Hunter and Cooper, 1985). Over the past year, a

* This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant 18024 from the National Insti- tutes of Health (to J. E. D.) and the Walther Cancer Institute, Indianapolis, IN. This is journal paper 12885 from the Purdue Uni- versity Agriculture Experiment Station. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

been submitted to the GenBank”/EMBL and PIR Data Banks with The nucleotide and protein sequences reported in this paper have

accession number(s) M60103 and A33154. $ Recipient of a Joe Dawson Graduate Fellowship from the Walther

Cancer Institute. § To whom correspondence should be addressed.

family of protein tyrosine phosphatases (PTPases)’ which catalyze the dephosphorylation of phosphotyrosine residues in proteins have been characterized at the cDNA level (re- viewed by Tonks and Charbonneau, 1989; Hunter, 1989; Alex- ander, 1990). Structurally the PTPases fall into two groups: molecules of approximately 50 kDa which contain a single copy of the tyrosine phosphatase domain (Cool et al., 1989; Chernoff et al., 1990; Guan et al., 1990); and receptor-like phosphatases of greater than 100 kDa that have diverse extracellular sequences and, with one exception, a duplicated intracellular phosphatase domain (Krueger et al., 1990). The biological substrates of these molecules are unknown, but the presence of two subclasses of phosphatases suggests that their subcellular localization may play an important role in their function, and this in turn may limit their substrate specificity. No ligands that bind the extracellular domains of the recep- tor-like PTPases have been identified.

The first receptor-like phosphatases were cloned with no knowledge of their dephosphorylation activity. Lymphoblastic cell surface molecules synonymously termed CD45, LCA, or Ly5 were isolated on the basis of their immunologic signifi- cance (reviewed by Thomas, 1989). Human LAR was cloned in an attempt to find molecules similar to leukocyte common antigen (LCA) (Streuli et al., 1988). Not until the partial amino acid sequence of soluble placental protein tyrosine phosphatase 1B was determined (Charbonneau et al., 1988) did the cytoplasmic domains of these two molecules suggest a definite catalytic function. Many other receptor-like PTPase cDNAs were subsequently isolated and sequenced based upon their homology with the placental PTPase. At present, the family includes 11 unique receptor-like molecules, some of which have been cloned from a number of species (Alexander, 1990). A widely distributed receptor-like PTPase LRP has been cloned from both mouse and human tissues. This cDNA encodes a protein with a relatively short extracellular domain that is probably glycosylated in COS cells (Sap et al., 1990). Variable cDNAs suggest that alternative splicing takes place in the first tyrosine phosphatase domain of LRP (Matthews et al., 1990). The gene for LRP has been localized on chro- mosomes 20 and 2 in humans (Kaplan et al., 1990; Jirik et al., 1990) and mice (Sap et al., 1990), respectively. Other human receptor-like PTPases with varying similarity to CD45 and LAR have also been cloned (Nishi et al., 1990; Krueger et al., 1990). One of these, hPTPP, is unique in that it contains only

The abbreviations used are: PTPase, protein tyrosine phospha- tase; rLAR, rat leukocyte common antigen-related protein; kb, kilo- base; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electro- phoresis; MES, 2-(N-morpholino)ethanesulfonic acid;pNPP,p-nitro- phenyl phosphate; AMP-PNP, adenyl-5”yl imidodiphosphate; pFSBA, 5’-p-fluorosulfonylbenzoyladenosine; LRP, leukocyte com- mon antigen-related peptide.

19688

Rat LAR PTPase: Expression, Purificati~n, and Characterization 19689

one cytoplasmic phosphatase (Krueger et d., 1990). Evidence for receptor-like PTPase molecules in Drosophila (Streuli et ai., 1989) and in the p r o t ~ h o r ~ t e Styela p ~ ~ u ~ u ( ~ a t t ~ e w s et ut , 1990) documents the long evolutionary history of the rbceptor-like PTPases. The primary structures of several receptor-like PTPases are known, however, the only struc- ture/~nct ion studies of re~eptor -~ke PTPases that have been performed used crude Escherichia coli extracts containing bacterially expressed PTPases (Streuli et al., 1990). This paper describes the characterization of pure rat LAR cyto- plasmic domain PTPase expressed in bacteria. Kinetic studies with various modulators of wild type and mutant proteins have enabled some of the properties of this important mole- cule to be determined.

EXPERIMENTAL PROCEDURES

~ater ia~-Res t r ic t~on endonucleases and modifying enzymes were purchased from New England Biolabs, US . Biochemicals, or Boeh- ringer Mannheim. pBluescript KS' was obtained from Stratagene. Oligonucleotide primers were obtained from the Purdue University M a c ~ m o l e c u l ~ Structure Laboratory. DE52 resin and P81 phospho- cellulose paper were from Whatmann, phenyl-Sepharose CL-4B was from Sigma, and Centricon 30 microconcentrators were from Amicon. Human angiotensin I and RR-SRC were obtained from Peninsula Laboratories, ~ 4 3 " ~ and Raytide were obtained from Oncogene Sci- ence. Bovine myelin basic protein and casein, as well as modulators, were obtained from Sigma. Affi-Gel 10 was purchased from Bio-Rad. Problot membrane was obtained from Applied Biosystems.

I s o ~ t ~ n of cDHA Clones-A rat hypothalamic cDNA library (pro- vided by M. Brownstein, National Institutes of Health), was plated at a density of 30,000 colonies/l50-mm plate. Colonies were trans- ferred to nitrocellulose filters according to the method of Grunstein and Hogness (1975). A total of 5 4 0 , ~ O recombinants were initially screened using a degenerate 32P-labeled oligonucleotide probe, d[TTCTG(T/G)TCCCA(A/G/C/T)ACCAT(T/C)TCCCA(G/A)AA]. Hybridization was carried out as previously described (Guan et al., 1990). Two clones were related to each other by several common restriction enzyme patterns. Each clone was sequenced following subcloning of the pCD chimeric BamHI fragments into pBluescript KS'.

Site-directed Mutagenesis-Site-directed mutagenesis of rLAR was carried out using the Amersham o l i g o n u c l e o t i d e - ~ r ~ ~ d in vitro mutagenesis system. The following mutants were prepared C1434S, C1522S, C1723S, C1813S. Specific nucleotide changes were confirmed by DNA sequencing.

Expression of rLAR C y ~ o p ~ m i c ~ o m a ~ n in E. coZ~-pBluescript clones containing the mutant or wild type cDNAs were digested with XmnI and BamHI to yield a cDN4 fragment encoding the cyto- plasmic domain of rLAR which included its natural stop codon. This fragment was d i r ~ i o n a l l y subcloned into the bacterial expression vector pT7-7 (provided by Stan Tabor, Harvard University; Tabor, 1990) which had been digested with EcoR1, filled in by Klenow enzyme, and cut with BamHI. The resulting plasmid, pT7 cyto-rLAR, is depicted in Fig. 3A. The EcoRI site was fo~uitously regenerated by ligation to the XmnI blunt end.

Expression of the cytoplasmic domain of the recombinant rat LAR protein was achieved using E. coli cell strain BL21(DE3) according to the methods of Studier and Moffat (Studier and Moffat, 1986). The cytoplasmic portion of LAR, which encompasses amino acid residues 1264-1881, was expressed as a fusion protein containing an addition of 4 amino acids on the amino terminus derived from the pT7-7 vector. The calculated mass of the resultingprotein was 71,392 daltons. Bacterial cultures were grown to an Am of 0.8. Expression was induced by the addition of isopropyl-1-thio-P-D-galactopyrano- side to a final concentration of 0.5 mM. Cells were harvested by centrifugation, resuspended in imidazole buffer (20 mM imidazole, pH 7.2, 0.1 mM EDTA, 0.2% 2-mercaptoethanol, 0.1 mM benzami- dine), and frozen at -20 "C.

Purification of rLAR Cytoplasmic Domuin-All steps of the purifi- cation were performed at 4 "C. Frozen bacterial pellets (5 g wet weight) were lysed by vigorous mixing with 0.1-mm glass beads in imidazole buffer containing 10 mM benzamidine, 50 pg/ml NCY-p- tosyl-L-lysine chloromethyl ketone, 0.5 pg/ml each of pepstatin, leu- peptin, chymostatin, and antipain, and 1 mM phenylmethylsulfonyl

fluoride. After centrifugation at 100,000 X g the supernatant was loaded onto a DE52 column (2.3 X 4.5 cm) equilibrated in imidazole buffer. The column was washed with 10 column volumes of imidazole buffer after which a 200-ml linear salt gradient from 0.0 to 0.2 M NaCl was applied. PTPase activity eluted at 0.05 M NaCI.

Pooled fractions containing PTPase activity from the DE52 c01- umn were added to 40 ml of phenyl-Sepharose CL-4B equilibrated in imidazole buffer and gently mixed for 15 min. The resin was then washed with 200 mi of imidazole buffer to remove contaminants. PTPase activity was eluted with 80 ml of imidazole buffer containing 50% glycerol. This eluent was then concentrated by reapplication to a 1-ml DE52 column. The bound protein was removed with 2 mi of imidazole buffer containing 0.2 M NaCI. This was further concen- trated to a volume of less than 0.3 ml using a Centricon 30 microcon- centrator before applying it to a Sephadex G-100 column (1.5 X 45 cm) equilibrated in imidazole buffer. The column was eluted at a flow rate of 0.5 mI/min. Fractions containing phosphatase activity were concentrated on another 1-ml DE52 column equilibrated in imidazole buffer. The final purified rLAR cytoplasmic domain was eluted from DE52 column with 2 ml of imidazole buffer containing 0.2 M NaCl and 30% glycerol. The enzyme was stable for several months a t 4 "C in this solution.

Protein sequencing of the 35-kDa contaminant was done by an Applied Biosystems model 470A gas-phase sequenator using 1 nmol of protein transferred to Problot membrane. The sequence deter-

phate dehydrogenase) was TIKVGINGFGRIGRIVFRAAQKR. Pro- mined for the 35-kDa protein contaminant (glyceraldehyde-3-phos-

tein concentration was measured by the method of Bradford (1976) using bovine serum albumin as a standard.

Determination of Phosphatase Activity-The ability of wild type and mutant rLAR protein to catalyze the hydrolysis of p-nitrophenyl phosphate @NPP) to p-nitrophenolate was measured in a 200-4 phosphatase assay mixture c o n ~ i n i n g 40 mhd MES, pH 5.0, 1.6 mM dithiothreitol, and 25 m~ pNPP. The reaction was terminated after 10 min at 30 "C by the addition of 1 ml of 0.2 N NaOH. Absorbance at 410 nm was determined and a molar extinction coefficient of 1.78 X IO4 M" cm-' was utilized to calculate the concentration of the p- nitrophenolate ion produced in the reaction. For modulator studies, preincubation of the modulator with the phosphatase was done for 10 min at room temperature prior to assay.

Phosphotyrosine-specific phosphatase activity was measured using 32P-labeled Raytide or human angiotensin 1. These substrates were phosphorylated by p4Tdf tyrosine kinase to specific activities of 4 X lo8 and 2 X lo5 cpm/nmol, respectively. Casein and histone H2a were phosphorylated on serine and threonine by the catalytic subunit of CAMP protein kinase type I to specific activities of 5 X lo5 and 6 X lo5 cpm/nmol, respectively. For the kinetic studies of human angio- tensin I, chemically synthesized unlabeled phosphotyrosine angioten- sin I (a gift of Drs. M. Cushman and R. Geahlen, Dept. of Medicinal Chemistry, Purdue University) was mixed with [3zPJphosphoangi- otensin (DRVpYIHPFHL) to provide a substrate with a specific activity of 3.8 X lo5 cpm/nmol. This allowed for kinetic determina- tions using substrate concentrations above and below the K,,, with peptide quantitatively phosphorylated on tyrosine. PTPase activity of rLAR was measured in a 5O-pl assay mixture containing 40 mM MES, pH 6.0, 20 mM dithiothreitol, and 0.2 p~ 32P-labeled tyrosine phosphorylated angiotensin I or Raytide. After incubation for 5 min at 30 "C, the reaction was terminated by the addition of 750 p1 of activated charcoal solution (0.9 M HCl, 90 mM sodium pyrophosphate, 2 mM NaH2PQ4, 4% (w/v) Norit A) (Streuli et al., 1989). The mixture was centrifuged for 2 min at room temperature and the radioactivity in 400 pl of supernatant was measured.

Preparation of rLAR-specific Polyclonat Antibody-Polyclonal an- tibody was raised against the bacterially expressed cytoplasmic region of rLAR by three subcutaneous injections of 100 pg of SDS-PAGE purified rLAR into a New Zealand White rabbit. Antibodies directed against nonrecombinant bacterial proteins were adsorbed out by

der of BL21(DE3) cells expressing the pT7-7 vector with no insert mixing ammonium sulfate-fractionated serum with an acetone pow-

(Harlow and Lane, 1988). The antibody was affinity purified by adsorption to purified recombinant rLAR protein conjugated to Affi- Gel 10. Immunoblot analysis was used to confirm that the purified immune serum reacted exclusively with the recombinant rLAR pro- tein (data not shown),

Co~paru t~ve Q ~ ~ t ~ t a ~ ~ o ~ of Various rLAR Proteins-The relative amounts of recombinant wild type and mutant rLAR proteins was quantitated using Western blot analysis and densitometry scanning. Following electrophoretic separation of the six variants on 10% SDS-

Rat LAR PTPase: Expression, Purification, and Characterization 19691

Sequence identity at the cDNA level was 90 and 78% in the coding and noncoding regions, respectively. This high degree of identity between the deduced amino acid sequence of this rat brain cDNA clone and the human LAR protein supported the idea that the respective genes encoded the same protein in two different species. Fig. 1C shows the structure of rLAR (rat leukocyte common antigen-related protein) with respect to other members of the PTPase family.

Comparison of rLAR with other tyrosine phosphatases revealed several interesting features. The LAR family of receptor-like PTPases contains the most highly conserved second domain sequences flanking cysteine 1813 of any of the receptor-like family of PTPases (Fig. 2.4). This VHCSAGVGRT sequence in the first domain of the receptor- like PTPases includes the catalytically essential cysteine. rLAR and many other double domain phosphatases also con- tain the amino acid sequence GXGXXGX17_,,Ac (where Ac is an acidic residue) downstream from cysteine 1522 of the first domain. This motif is similar to the nucleotide binding fold present in protein kinases and dehydrogenases (Ross- mann et al., 1975; Hanks et al., 1988; Sternberg and Taylor,

1984) and is less conserved in the second domain of the receptor-like PTPases. Interestingly, rLAR phosphatase ac- tivity was insensitive to the nonhydrolyzable ATP analogues, AMP-PNP and pFSBA, which bind the nucleotide-binding sites of various kinases and dehydrogenases, a t concentrations up to 1 mM (data not shown).

Comparison of nonreceptor with receptor-like protein ty- rosine phosphatases revealed that only nonreceptor-like PTPases contained the sequence FKVRES upstream from the conserved GXGXXG motif and near the active site cys- teine residue (Fig. 2B). The FKVRES sequence is similar to the FLVRES sequence that is the most highly conserved region of the SH2 domain of the Src tyrosine kinase family. This domain binds phosphotyrosine containing proteins (An- derson et al., 1990; Moran et al., 1990; Mayer et al., 1991). The FLVRES sequence in c-src may be associated with cell- specific oncogenesis (Hirai and Varmus, 1990a, 1990b). As in the case of the conserved glycine motif, the significance of the FKVRES sequence remains to be elucidated.

Expression and Purification of I A R PTPase Cytophwnic Domain-The activity of the cytoplasmic portion of rLAR was examined following its expression and isolation from E.

A

rlAR h l U I hPTP6 hPTPo rCD45 hPTP,9 hPTP7

dlAR r P T P l 8

dPTP

hlAR rlAR

hPTP6 hPTPa rCD45 hPTP,9 hPTPy

dlAR r P T P l 8

dPTP

B

hlAR rlAR

hPTP6 hPTPo rCD45 hPTPB hPTPy

d U R r P T P l 8

dPTP

hPTPl

LTCELL r P T P l

1 5 2 0 ursdmah

1 5 4 0 . . * 1560

pT7 cyto ILAR

HI 1 8 1 0 - .. *

1830 1 8 5 0 I 4

Met ala a r g Ile hlr M pro ser M l y a ssp Q u I

1500 1520 1540

A P T P F L A F L R R V K T C N - T P ; I G M L K E L K K V K A C N - D P H L L ~ K L R R R V N A F S - Y SLPVETEYRKAAYAK-

HEAPFLOELRRCRALT- Y A L P V E T I V R R S S A A R - Y S L P V A V R K T A O A K -

H E H G I I K E I R P I N S V Y -

21 I 1 0 7 69 46

29

18

I5

FIG. 2. Alignment of PTPase sequences surrounding the probable active site. A, alignment of first and second domains of receptor-like PTPases. Asterisks indicate the position of the nucleo- tide binding fold motif. Black highlights indicate greater than 80% identity, grey highlights indicate 60430% identity. Numbering corre- sponds to the sequence of rLAR. B, position of the FKVRES sequence with respect to the first domain of the receptor-like PTPases and the single PTPase domain of the nonreceptor-like PTPases. Asterisks indicate the position of the nucleotide binding fold motif. Numbering corresponds to the sequence of rLAR. Ten or more identities of 13 are highlighted with grey. References: hLAR, Streuli et af., 1988, hPTPb, -0, -7, Krueger et al., 1990; rCD45, Thomas et af., 1985; hPTPB, Kaplan et af., 1990; rPTP18, Guan and Dixon, 1990; dLAR, dPTP, Streuli et of., 1989.

1 2 3 4

FIG. 3. Production and purification of rLAR cytoplasmic domain. A, schematic representation of bacterial expression plasmid pT7 cyto-rLAR. Important restriction endonuclease cleavage sites used for cloning are indicated. The coding region of the cI)NA is indicated by the internal circular arrow. The numbers 1-4 indicate the position of the mutated cysteine residues. The sequence around the start of translation is shown indicating the extent of the fusion protein. R, Coomassie Blue-stained 12”; SDS-polyacrylamide gel showing stages in the purification of rLAH. The arrow between lanm 2 and 3 indicates the position of glyceraldehyde-%phosphate dehy- drogenase. I a n e . 9 1-4 contain 6, 4. 3, and 1.25 pg of total protein, respectively.

19692 Rat LAR PTPase: Expression, P u r ~ ~ ~ c a t ~ n , and C ~ r ~ t e ~ ~ z a t i o ~ TABLE I

Purification of rLAR Protein Total

concentration protein min/ml" Of lJNPP/ Total %::$ Yield Purification

mg % -fold

Crude extract 75 1.16 87 0.65 49 0.56 (100) DE52

(1)

Phenyl-Sepharose 73 0.016 1.17 0.58 42.1 36 85.9 64.3 Sephadex G-100 2.0 0.25 0.5 16.2 32.3 64.6 65.9 115

60 0.26 15.6 1.04 62.1 4.0 127 7.14

a Phosphatase activity was measured with pNPP as a substrate, as described under "Experimental Procedures."

coli extracts. A 2.3-kb fragment of the rLAR cDNA encoding the entire cytoplasmic region of the protein was subcloned behind the bacteriophage T7 promoter in the bacterial pT7-7 plasmid (Fig. 3A). Following isopropyl-l-thio-,8-D-g~actopy- ranoside induction, a high level of a soluble 71-kDa protein representing the cytoplasmic portion of rLAR was observed (Fig. 3B, lane 1).

To reduce the complexity of interfering contaminating phosphatase activities while analyzing the recombinant rLAR, it was necessary to purify the bacterially expressed 71- kDa protein to homogeneity. This purification was achieved in three steps involving anion exchange, hydrophobic inter- action, and gel permeation chromato~aphy. Following DE52 anion exchange c h r o m a ~ ~ a p h y , the amount of phosphatase activity in the extract was increased, possibly by the removal of a phosphatase inhibitory activity (Table If. The recombi- nant rLAR protein was eluted from the column in low salt (0.05 M NaCI), resulting in a 7-fold purification.

The second purification step involved chromatography using phenyl-Sepharose CL-4B. rLAR remained bound to the matrix even under low ionic strength, a condition under which most proteins are eluted from hy~ophobic columns (Ken- nedy, 1990). This high affinity of the PTPase for the phenyl- based matrix suggested that its binding might be taking place through an interaction at the active site of the enzyme. This is supported by the observation that 100 mM pNPP specifi- cally released the protein from the resin (data not shown). As indicated in Table I and Fig. 3B, lanes 2 and 3, phenyl- Sepharose chromatography was an effective separation step giving a further 9-fold purification. One fraction from the phenyl-Sepharose column contained a single 35-kDa protein (Fig. 38, lane 2). Since the cytoplasmic portion of the rLAR can be subdivided into two 35-kDa phosphatase domains, this 35-kDa protein could have represented a single catalytic do- main of rLAR generated by proteolysis. Protein sequencing identified this polypeptide as E. coli glyceraldehyde-3-phos- phate dehydrogenase (Branlant and Branlant, 1985)) a protein that can account for up to 20% of soluble bacterial protein.

The third p~ification step of rLAR involved gel exclusion c h r o m a ~ ~ a p h y (Fig. 3, lane 4) . This resulted in a final overall purification of 11Bfoid. The purified enzyme could be stored at a protein concentration greater than 0.1 mg/ml in imidazole buffer containing 30% glycerol at 4 "C without loss of enzyme activity over a period of several months. Yields of 5 mg of pure protein have been routinely produced from 1 liter of bacterial culture (5 g of wet pellet).

Phosphatase Activity of Purified rLAR Cytoplasmic Domain Was Specific for ~ y F o s i ~ P ~ s p ~ ~ p ~ ~ e s - ~ r o s i n e - s p ~ i ~ c dephosphorylation was demonstrated by assaying purified enzyme with various radiolabeled substrates. Tyrosine phos- phorylated proteins and peptides that served as substrates included human angiotensin I, Raytide, RR-SRC, myelin basic protein, and casein. Autophosphorylated rat liver insulin receptor ,8 subunit was also partially dephosphorylated by

TABLE I1 Kinetic constants of pNPP and human angiotensin I

Substrate K m ka t kt/Kl

irM S" s-l&f"

pNPP 420 6.1 1.4 x lop DRVpYIH~FHLn 118 3.1 2.6 X lo4

a Human phosphotyrosine angiotensin I.

rLAR (data not shown). Casein and histone phosphorylated on serine or threonine were not substrates for the enzyme. Kinetic constants for the substratespNPP and human angio- tensin I are shown in Table 11. While both compounds serve as substrates, the synthetic phosphoan~otensin I (DRVp- YIHPFHL) has a more favorable binding constant. The kcat constants for the two substrates differ by less that a factor of 2, having values of 6.1 and 3.1 8-l for pNPP and angiotensin I, respectively.

Modulators of rLAR Protein srosine Phosphatase Activ- ity-pH versus rate profiles of rLAR were determined using pNPP and 32P-labeled tyrosine phosphorylated human angio- tensin I as substrates. The pH optima for dephosphorylation of these substrates were 5.0 and 6.0, respectively.

Compounds known to modulate phosphatase activity were also evaluated for their effects on rLAR (Table 111). Vanadate, at a concentration of 100 PM, gave 50% inhibition (ICSO). This value was similar to that observed for CD45 and an order of magnitude higher than that needed to inhibit PTP1.' Phen- ylarsine oxide has been shown to inhibit CD45 with an IC50 between 5 and 10 @M (Garcia-Morales et al., 1990). rLAR was not effected by phenylarsine oxide, even at 10-fold the con- centration used to inhibit CD45, both at pH 5.0 and 7.0. Similarly rLAR phosphatase activity was not inhibited by zinc cation at pH 5.0 and 7.0, another difference of rLAR from other PTPase activities.

Characterization of protein tyrosine phosphatases have shown that thiol reducing agents are necessary for enzymatic activity (Jones et al., 1989; Tonks et al., 1988; Swarup and Subrahmanyam, 1989). This suggests that cysteines play im- portant roles in enzyme structure or function. Three irrevers- ible s u l ~ y ~ l - d i r e ~ e d inhibitors were tested for their impact on the tyrosine phosphatase activity of rLAR. At a concen- tration of 1 mM, N-ethylmaleimide, p4hydroxymer- curi)benzoate, and iodoacetate reduced enzyme activity by 95, 98, and 99%, respectively.

An interesting differential effect was seen with iodoacetate and iodoacetamide. The negatively charged iodoacetate com- pletely inhibited phosphatase activity of rLAR at 1 mM, whereas iodoacetamide had no inh ib i to~ effect on the enzyme at a similar concentration. Since the major difference between these two compounds was the presence of a negative charge, these results suggested that iodoacetate may interact specifi- cally with a positive charge at or near the active site of the

* K. Guan, personal communication.

Rat LAR PTPase: Expression, Purification, and Characterization 19693

TABLE I11 Influence of various agents on PTPase activity of rLAR

Modulator Concentration Original activity" %

Na vanadate 1000 p M 10 100 pM 46 10 pM 74 1 PM 87

NH4 molybdate 10 p M 21 Heparin 50 d m 1 93 4:l Poly(Glu/Tyr) 5 PM 76 NaF 5 mM 97

50 mM 88 Na tartrate 5 mM 91 Tetramisole 1 mM 82 Znz+ acetate 0.1 mM 90

1 mM 87 CaC12 1 mM 90 M$' acetate 1 mM 98 Phenylarsine oxide 100 pM 104

10 pM 98 1 PM 88

Spermine I mM 70 Spermidine 1 mM 80 EDTA 5 mM 71 N-Ethylmaleimide 1 mM 5 p-(Hydroxymercuri)benzoate 1 mM 2 Iodoacetate 1 mM 1 Iodoacetamide 1 mM 95

PTPase activity was measured with pNPP as described under "Experimental Procedures." Activities are expressed as a percentage of that measured under control conditions in the absence of other additions. All results represent the mean of at least duplicate reac- tions.

100

x 5 a0 - 6

E 6o

40 0 5 0" 20

0

4 5 6 p~ 7 a 9

4 5 p.. e FIG. 4. Effects of pH on rLAR. A, pH profile of rLAR enzyme

activity using both pNPP and human angiotensin I (hAG-1) as substrates. B, effect of hydrogen ion concentration on iodoacetate and iodoacetamide inhibition. 10 mM concentration of inhibitor was used.

enzyme. The importance of the negative charge in iodoacetate on PTPase inhibition was further examined as a function of pH. As shown in Fig. 4B, the extent of iodoacetate inhibition remained high until the pH of the solution fell below 5.0. Incubation of the enzyme with 10 mM pNPP prevented the inhibition by iodoacetate. Sodium sulfate at 10 mM also protected the enzyme from inhibition by iodoacetate, whereas

the same concentration of sodium phosphate allowed no pro- tection. This was not a cation effect since the presence of Na2S04 and MgS04 gave similar results. The pH uersus inhibition profile with iodoacetamide was strikingly different from that noted for iodoacetate (Fig. 4B). Inhibition by iodo- acetamide increased when the pH of the reaction was above 7.0.

The Majority of Tyrosine Phosphatase Activity Was Cata- lyzed by the First Domain-The number of cloned PTPases from which deduced amino acid sequences are available has allowed for sequence alignment and identification of con- served residues which might be essential for maintaining structure and function. Two conserved cysteine residues are present in each phosphatase domain of the receptor-like PTPases. In rLAR these amino acids are at positions 1434 and 1522 in domain I and 1723 and 1813 in domain 11. Site- directed mutagenesis was used to introduce serine into these positions in the protein. The mutants were designated m l through m4, according to their location in the primary se- quence, with m l being most amino-terminal (Figs. 1C and 3A). Each of these mutant proteins were expressed in the bacterial expression system previously described and the re- spective proteins purified. As seen from Table IV, mutation of all but m2 yielded enzymes with activities between 33 and 111% of that observed with wild type rLAR. A mutant enzyme with an alteration at the second cysteine of the first domain (m2) still retained 0.09% of wild type tyrosine phosphatase activity. To determine whether this low activity level in m2 originated from domain 11, a double mutant termed m24 was constructed with serine substitution at both the second and fourth conserved cysteines. Mutation of both these cysteines reduced phosphatase activity to only 0.04% of wild type rLAR activity (Table IV). Although a small amount of residual activity may have been present in domain 11, greater than 99% of the phosphatase activity of rLAR was associated with domain I.

rLAR Could Be Phosphorylated in Vitro by Protein Kinase C and ~ 4 3 " ~ ' Tyrosine Kinase-Many possible sites of phos- phorylation exist within the cytoplasmic domain of rLAR. To determine if this phosphatase was a substrate for serine/ threonine or tyrosine kinases in vitro, phosphorylation of rLAR by protein kinase C, ~ 4 3 " ~ ' tyrosine kinase, the cata- lytic subunit of CAMP protein kinase (type I), and casein kinase 11 was examined. rLAR did not appear to be a substrate for the latter two serine/threonine protein kinases (data not shown). However, rLAR was phosphorylated by protein ki- nase C in a time-dependent manner (Fig. 5A). This produced no observed effect on rLAR phosphatase activity. Catalyti- cally inactive C1522/S rLAR was also phosphorylated by

TABLE IV Relative activities of mutant rLAR proteins

Raytide pNPP Specific Relative Specific Relative activitp activityb activity' activityb

Wild type 9320 100 75.5 100 ml 3130 33.5 30.8 40.8 m2 8.72 0.09 0.057 0.08 m3 10400 111 110 146 m4 4950 53.1 29.9 39.6 m24 3.27 0.04 0.018 0.02

"Raytide specific activity is expressed as picomoles of Raytide

* Relative activity is expressed as percent of wild type rLAR activ-

'pNPP specific activity is expressed as micromoles ofp-nitrophen-

dephophorylated min" mg of rLAR".

ity.

olate produced min" mg of rLAR".

19694

A kD

271

107 69.3

45.8

28.7

B

15'

".

Rat LAR PTPase: Expression, Purification, and Characterization

PKC 5' 10' 15' 20'

WILD TYPE C1522/s

+ 1ouM NaVk 30' 45' ' 15' 30' 45' ' , 5' 10' 15'

"LAR -00

"p43 .- . .

FIG. 5. Phosphorylation of rLAR. 2.5 pg of purified rLAR was incubated as described under "Experimental Procedures" with 20 ng of protein kinase C or 1 unit of p43'.& in a 7O-pl reaction mixture. At the designated times, 15 1 1 of the reaction was removed, mixed with SDS gel sample buffer, and loaded onto a 10% polyacrylamide gel. A, autoradiograph displaying a time course of rLAR phosphoryl- ation by protein kinase C. Coomassie Blue-stained SDS-PAGE gel molecular weight markers and rLAR used in the assay are shown on the left. B, autoradiograph displaying time-dependent phosphoryla- tion of rLAR by ~ 4 3 " ~ tyrosine kinase. C1522/S represents the catalytically inactive PTPase. The purity of rLAR is shown on a Coomassie Blue-stained SDS-PAGE gel at the left. The relative positions of rLAR and ~ 4 3 " ~ are indicated.

~ 4 3 " ~ . Phosphorylation of the active wild type enzyme was more difficult to demonstrate due to the trans- or autode- phosphorylating ability of the phosphatase (Fig. 5B) . The presence of 10 p~ sodium vanadate in the reaction mixture allowed limited phosphorylated wild type rLAR to be detected, however, ~ 4 3 " ' ~ ' tyrosine kinase was also partially inhibited by this concentration of vanadate (as seen by a reduction in its autophosphorylation state). Phosphorylation of rLAR using a 1000-fold molar excess of ATP and multiple additions of active kinase resulted in phosphorylation stoichiometries of 0.6 and 0.4 mol of phosphate/mol of catalytically inactive rLAR for protein kinase C and ~ 4 3 " ~ kinase, respectively.

A Phosphocysteine Reaction Intermediate Could Be Trapped by Rapid Denaturation of the Substrate-Enzyme Complex- Evidence from this laboratory suggested that a cysteine resi- due within the active site of the nonreceptor PTPl appears to form a thiol-phosphate enzyme intermediate which could be trapped by rapid denaturation and then visualized by SDS- PAGE.3 Experiments were undertaken to determine if the receptor-like PTPases also formed a phosphoenzyme inter- mediate which could be similarly trapped. Purified rLAR was incubated for 10 s with [32P]phosphotyrosine-labeled angio- tensin I, rapidly denatured with SDS, and subjected to gel electrophoresis. A radioactive band of 71 kDa was seen follow- ing autoradiography (Fig. 6). The intensity of this band was dependent upon the concentration of rLAR protein in the reaction, consistent with the formation of an enzyme-sub- strate complex (Fig. 6, lanes 24). Furthermore, no band was detected when the enzyme was heat denatured or preincubated with iodoacetate (lanes 5 and 6). The specificity of this labeling was further demonstrated by use of the purified

a Guan, K., and Dixon, J. E. (1991) J. Bwl. Chern., in press.

wt rLAR m2 rLAR "

Y m

F t-4 v,

M C

s t-4

B - 2 m

ai, C 0 s

J Q) 0 m -0 0 0 .- ai, C 0 v, t-4

M C v, m

00 C

5

1 2 3 4 5 6 7 8

FIG. 6. Trapping of phosphoenzyme intermediate. Autora- diography of 10% SDS-PAGE gel. Molecular weight of the resulting band is 71 kDa. The amounts of rLAR protein added to each reaction are indicated above each line. wt rLAR indicates "wild type" rLAR m2 rLAR indicates C1522S rLAR mutant.

catalytically inactive m2 rLAR protein which was incapable of forming a radioactive phosphointermediate.

DISCUSSION

The cloning and characterization of a large family of protein tyrosine phosphatases has allowed an "active site motif' for these enzymes to be defined. The amino acid sequence com- posing part of the active site motif is centered around the second highly conserved cysteine of the PTPase domain and contains the core sequence VHCSAGVGRTG. Within this core sequence is a nucleotide binding fold pattern seen in dehydrogenases, kinases, and other nucleotide-binding pro- teins (Rossmann et al., 1975; Hanks et al., 1988; Sternberg and Taylor, 1984). This sequence is present in all receptor and nonreceptor-like PTPases. Modification of the first gly- cine in the sequence to any other residue eliminates hLAR activity (Streuli et al., 1990). The presence of a nucleotide binding motif in PTPases suggests that structural features within this region of the PTPases are likely to be similar to structures which have been determined for the dehydrogen- ases and kinases.

Another similarity between the nonreceptor-like FTPases and cytoplasmic protein tyrosine kinases is their shared FKVRES sequence. Recent studies directed at determining the molecular basis of interaction between tyrosine kinases and their cytoplasmic substrates have focused on the SH2 domain found in the Src protein. Peptides encompassing only this domain are able to specifically bind tyrosine-phosphoryl- ated receptors but not nonphosphorylated receptors (Ander- son et al., 1990; Moran et al., 1990; Mayer et al., 1991). One of the most conserved regions of the SH2 domain is the sequence FLVRES which has been implicated in the deter- mination of cell-specific oncogenesis patterns of c-src (Hirai and Varmus, 1990a, 1990b). The function of a related se- quence in the nonreceptor-like PTPases is unknown. How- ever, i ts possible role in binding of tyrosine-phosphorylated proteins or in other aspects of protein-protein recognition should not be overlooked.

This paper describes the methodologies for the first purifi- cation of a bacterially expressed receptor-like protein tyrosine phosphatase cytoplasmic domain. Substantial quantities of enzymatically active rLAR were isolated by use of three rapid and simple chromatographic steps. The high affinity of rLAR for the phenyl-Sepharose resin, and the fact that it was eluted by substrate from the resin, indicated that it may be function- ing as an affinity purification step which interacts through the active site of the enzyme. This suggests that this resin

Rat LAR PTPme: Expression, Purification, and Characterization 19695

could also be useful for the purification of other PTPases. rLAR specifically dephosphorylated phosphotyrosine con-

taining proteins and not proteins phosphorylated on serine or threonine. The pH optima for rLAR, determined with pNPP and human angiotensin I, were similar to PTPase 1A and 1B (Tonks et al., 1988) showing that the behavior of this recom- binant enzyme was not unlike those from natural tissue sources. The inhibitory effects of vanadate and molybdate on the PTPase activity were similar to those for other PTPases (Tonks et al., 1988, 1990; Jones et al., 1989; Swarup and Subrahmanyam, 1989), while phenylarsine oxide specifically inhibits CD45 but has no effect on rLAR (Garcia-Morales et al., 1990). Other phosphatase inhibitors such as tetramisole and fluoride had no effect on rLAR activity.

The differential effects of iodoacetate and iodoacetamide on rLAR activity suggest that iodoacetate may interact with a positive ligand at or near the active site of the enzyme. This may take place through direct ionic interaction or through hydrogen bonding, as has been shown in the three-dimen- sional structure of periplasmic phosphate-binding protein of bacteria (Luecke and Quiocho, 1990). The reduced inhibition by iodoacetate at acidic pH could be explained by the proton- ation of the carboxyl group of iodoacetate (pK, = 3.12). Increased inhibition by iodoacetamide at a more basic pH may be due to deprotonation of residues on the enzyme which are important in catalysis. Sequence comparison of rLAR with other cloned PTPases has highlighted 6 conserved argi- nines and lysines that may participate in substrate binding. Iodoacetate could interact with one or more of these residues.

The presence of tandem PTPase domains in receptor-like PTPases suggested that both domains could have phosphatase activity. Comparison with other phosphatases revealed that rLAR retains the most similar amino acid sequence between the two domains of any of the receptor-like molecules. This raised the possibility that rLAR may contain phosphatase activity in its carboxyl-terminal domain. Results showed that greater than 99% of the phosphatase activity of rLAR could be attributed to the first phosphatase domain. This agrees with previous studies using crude bacterial extracts containing the cytoplasmic domains of hLAR (Streuli et al., 1990). The residual activity present in the protein after mutation of Cys- 1522 may be associated with the second PTPase domain since mutagenesis of Cys-1522 and -1813 lowers the apparent phos- phatase activity of the recombinant protein. Participation of the substituted serine in the reaction mechanism may also be possible, since the reciprocal functional substitution of a cysteine for the essential serine of alkaline phosphatase has been observed (Ghosh et al., 1986). Alternatively one might argue that the extremely low levels of PTPase activity in the second domain may result from the absence of appropriate substrates to monitor the PTPase activity. The second do- main of rLAR may also function to modulate the first domain activity, as has been suggested from studies using crude prep- arations of hLAR (Streuli et al., 1990).

The ability of both serine/threonine and tyrosine kinases to phosphorylate rLAR in vitro suggests that the PTPase may be regulated in uiuo by these or similar enzymes. Many regu- latory phosphorylation sites have already been demonstrated in protein kinases and serine/threonine phosphatases (re- viewed by Hunter and Cooper, 1985; Cohen, 1989). The ability of rLAR to undergo auto- or transdephosphorylation on phos- photyrosine residue(s) parallels the autophosphorylation abil- ity of tyrosine kinases (Hunter and Cooper, 1985).

The important role that sulfhydryl residues play in rLAR catalysis is evident from the inhibition data obtained for iodoacetate, N-ethylmaleimide, and mercurial reagents. Ad-

ditional insights into the catalytic mechanism of rLAR were obtained by trapping a radioactive phosphoenzyme interme- diate. These results suggest that a cysteine-phosphate inter- mediate may be formed during hydrolysis of phosphotyrosine containing proteins by the receptor-like PTPases. This ob- servation would distinguish the PTPases from alkaline (Schwartz and Lipmann, 1961) and certain acid phosphatases (Van Etten, 1982) which appear to proceed through serine and histidine intermediates. We have recently demonstrated the occurrence of an enzyme-phosphate intermediate using the recombinant nonreceptor-like PTPase from rat brain: The phosphoenzyme intermediate has a pH stability profile which suggests it forms a cysteine-phosphate intermediate. The intermediate is also rapidly inactivated by Ip and Brz, properties which again distinguish it from histidine and serine-phosphate intermediates? Collectively the similarity in primary structure of receptor and nonreceptor-like PTPases, the importance of sulfhydryl residues in catalysis, and our ability to trap a phosphoenzyme intermediate with rLAR suggests, but does not directly demonstrate, that cysteine 1522 is forming a thiol phosphate intermediate with rLAR. It is likely that PTPl and other PTPases also proceed through a similar phosphocysteine intermediate.

Acknowledgments-We would like to acknowledge Dr. Jan Rhodes for aid in affinity purification of the antibodies and Jim Clemens and Mary Woenker for Problot protein sequencing. Dr. K. H. Kim kindly provided the purified rat insulin receptor. Purified protein kinase C was kindly provided by Dr. C. L. Ashendel. The catalytic subunits of CAMP protein kinase type I and casein kinase I1 were kindly provided by Drs. J . Corbin and N. Osheroff, respectively, Dept. of Molecular Physiology and Biophysics, Vanderbilt University.

REFERENCES

Alexander, D. R. (1990) New. Biol. 2 , 1049-1062 Anderson, D., Koch, C. A., Grey, L., Ellis, C., Moran, M. F., and

Bradford, M. M. (1976) Anal. Biochem. 72,248-254 Branlant, G., and Branlant, C. (1985) Eur. J. Biochem. 150,61-66 Charbonneau, H., Tonks, N. K., Walsh, K. A., and Fischer, E. H.

Chernoff, J., Schievella, A. R., Jost, C. A., Erikson, R. L., and Neel,

Cohen, P. (1989) Annu. Reu. Biochem. 58,453-508 Cool, D. E., Tonks, N. K., Charbonneau, H., Walsh, K. A., Fischer,

E. H., and Krebs, E. G. (1989) Proc. Natl. Acad. Sci. U. S. A. 8 6 ,

Garcia-Morales, P., Minami, Y., Luong, E., Klausner, R. D., and

9259 Samelson, L. E. (1990) Proc. Natl. Acad. Sci. U. S. A. 8 7 , 9255-

Ghosh, S. S., Bock, S. C., Rokita, S. E., and Kaiser, E. T. (1986) Science 231, 145-148

Grunstein, M., and Hogness, D. S. (1975) Proc. Natl. Acad. Sci. U. S.

Guan, K., and Dixon, J. E. (1990) Science 249,553-556 Guan, K., Haun, R. S., Watson, S. J., Geahlen, R. L., and Dixon, J.

E. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 1501-1505 Hanks, S. K., Quinn, A. M., and Hunter, T. (1988) Science 241,42-

52 Harlow, E., and Lane, D. (1988) Antibodies, A Laboratory Manual,

pp. 632-633, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

Hirai, H., and Varmus, H. E. (1990a) Proc. Natl. Acad. Sci. U. S. A.

Hirai, H., and Varmus, H. E. (1990b) Genes & Deu. 4 , 2342-2352 Hunter, T. (1989) Cell 58,1013-1016 Hunter, T., and Cooper, J. A. (1985) Annu. Reu. Biochem. 5 4 , 897-

Jirik, F. R., Janzen, N. M., Melhado, I. G., and Harder, K. W. (1990)

Jones, S. W., Erikson, R. L., Ingebritsen, V. M., and Ingebritsen, T.

Kaplan, R., Morse, B., Huebner, K., Croce, C., Howk, R., Ravera, M.,

Pawson, T. (1990) Science 250,979-982

(1988) Proc. Natl. Acad. Sci. U. S. A. 85, 7182-7186

B. G. (1990) Proc. Natl. Acad. Sci. U. S. A. 87, 2735-2739

5257-5261

A. 72,3961-3965

87,8592-8596

930

FEBS Lett. 273,239-242

S. (1989) J. Biol. Chem. 264,7747-7753

19696 Rat LAR PTPase: Expression, Purification, and Characterization

Ricca, G., Jaye, M., and Schlessinger, J. (1990) Proc. Natl. Acad. Sci. U. S. A. 8 7 , 7000-7004

Kennedy, R. M. (1990) Methods Enzymol. 182,339-343 Krueger, N. X., Streuli, M., and Saito, H. (1990) EMBO J. 9 , 3241-

Luecke, H., and Quiocho, F. A. (1990) Nature 347 , 402-406 Matthews, R. J., Cahir, E. D., and Thomas, M. L. (1990) Proc. Natl.

Acad. Sci. U. S. A. 87,4444-4448 Mayer, B. J., Jackson, P. K., and Baltimore, D. (1991) Proc. Natl.

Acad. Sci. U. S. A. 88,627-631 Moran, M. F., Koch, C. A., Anderson, D., Ellis, C., England, L.,

Martin, G. S., and Pawson, T. (1990) Proc. Natl. Acad. Sci. U. S.

Nishi, M., Ohagi, S., and Steiner, D. F. (1990) F E B S k t t . 2 7 1 , 178- 180

Rossmann, M. G., Liljas, A., Brand&, C., and Banaszak, L. J. (1975) in The Enzymes (Boyer, P. D., ed) Vol. 11, pp. 61-102, Academic Press, New York

Sap, J., D’Eustachio, P., Givol, D., and Schlessinger, J. (1990) Proc. Natl. Acad. Sci. U. S. A. 87,6112-6116

Schwartz, J. H., and Lipmann, F. (1961) Proc. Natl. Acad. Sci. U. S.

Sternberg, M. J. E., and Taylor, W. R. (1984) FEBS Lett. 175,387-

3252

A. 87,8622-8626

A., 47 , 1996-2005

392

Streuli, M., Krueger, N. X., Hall, L. R., Schlossman, S. F., and Saito,

Streuli, M., Krueger, N. X., Tsai, A. Y. M., and Saito, H. (1989) Proc.

Streuli, M., Krueger, N. X., Thai, T., Tang, M., and Saito, H. (1990)

Studier, F. W., and Moffat, B. A. (1986) J. Mol. Bwl. 189 , 113-130 Swarup, G., and Subrahmanyam, G. (1989) J. Biol. Chem. 264,7801-

7808 Tabor, S. (1990) in Current Protocols in Molecular Biology (Ausubel,

F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A., and Struhl, K., eds) Vol. 2, pp. 16.2.1-11, John Wiley and Sons, New York

H. (1988) J. Exp. Med. 168 , 1523-1530

Natl. Acad. Sci. U. S. A. 86,8698-8702

EMBO J. 9,2399-2407

Thomas, M. L. (1989) Annu. Reu. Immuml. 7 , 339-369 Thomas, M. L., Barclay, A. N., Gagnon, J., and Williams, A. F. (1985)

Tonks, N. K., and Charbonneau, H. (1989) Trends Biochern. Sci. 14 ,

Tonks, N. K., Diltz, C. D., and Fischer, E. H. (1988) J. Biol. Chem.

Tonks, N. K., Diltz, C. D., and Fischer, E. H. (1990) J. Bwl. Chem.

Van Etten, R. L. (1982) Ann. N. Y. Acad. Sci. 390,27-51

Cell 41 , 83-93

497-500

263,6731-6737

265,10674-10680