Proc. Natl. Acad. Sci. USA Vol. 86, pp. 8247-8251, November 1989

Biochemistry

Extent of N-terminal methionine excision from Escherichia coli proteins is governed by the side-chain length of the penultimate amino acid (protein maturation/methionyl-aminopeptidase/hybrid protein expression)

PH.-HERVt HIREL, JEAN-MARIE SCHMITTER, PHILIPPE DESSEN, GuY FAYAT, AND SYLVAIN BLANQUET Laboratoire de Biochimie, Unite Associde 240 Centre National de la Recherche Scientifique, Ecole Polytechnique, 91128 Palaiseau Cedex, France

Communicated by Marianne Grunberg-Manago, July 24, 1989

In a significant fraction of the Escherichia coli ABSTRACT cytosolic proteins, the N-terminal methionine residue incorporated during the translation initiation step is excised. The N-terminal methionine excision is catalyzed by methionylaminopeptidase (MAP). Previous studies have suggested that the action of this enzyme could depend mainly on the nature of the second amino acid residue in the polypeptide chain. In this study, to achieve a systematic analysis of the specificity of MAP action, each of the 20 amino acids was introduced at the penultimate position of methionyl-tRNA synthetase of E. coli and the extent of in vivo methionine excision was measured. To facilitate variant protein purification and N-terminal sequence determination, an expression shuttle vector based on protein fusion with (3-galactosidase was used. From our results, methionine excision catalyzed by MAP is shown to obey the following rule: the catalytic efficiency of MAP, and therefore the extent of cleavage, decreases in parallel with the increasing of the maximal side-chain length of the amino acid in the penultimate position. This molecular model accounts for the rate of N-terminal methionine excision in E. coli, as deduced from the analysis of 100 protein N-terminal sequences.

Experiments conducted in vitro on oligopeptides with purified MAP (8), in a eukaryotic cell-free expression system (12) or in vivo in yeast (11, 13), suggest that the MAP specificity mainly depends on the nature of the second amino acid of the polypeptide. To systematically investigate this hypothesis in vivo, we have submitted to MAP action a set of 20 protein species differing only by the nature of the penultimate amino acid residue. This was done by engineering and expressing in E. coli 20 mutant genes of methionyl-tRNA synthetase (EC 6.1.1.10). To facilitate the purification of the variant proteins, an expression shuttle vector based on protein fusion with 8-galactosidase (EC 3.2.1.23) was used

Incorporation of a methionine residue at the N terminus of each nascent polypeptide makes part of the universal translation initiation signal, used by prokaryotes as well as eukaryotes. In prokaryotes, the methionyl moiety carried by the initiator tRNA is N-formylated prior to its incorporation. However, soluble proteins retaining a formylated N terminus do not represent a measurable fraction of total proteins in Escherichia coli (1, 2). Moreover, in a cytosolic extract of E. coli, only 40% of the polypeptidic chains retain an N-terminal methionine. Instead, about 50% display alanine, serine, or threonine at their N termini (3). These observations are accounted for by early posttranslational modifications of the polypeptides. The formyl group and methionine residue are removed sequentially, the deformylation step being more tightly coupled to the translation process than the methionine excision (4). The occurrence of two separate activities was established by the purification of the E. coli deformylase enzyme (2, 5-7) and, very recently, by the cloning of the methionyl-aminopeptidase (MAP) gene from both E. coli and Salmonella typhimurium (8, 9). Previous studies (2, 10) and, more recently, a survey (11) of published protein sequences have attempted to find out a rule for the conditional in vivo excision of the initiator methionine residue. However, the specificity of the removal of this residue by the E. coli MAP could never be systematically described, neither in vitro nor in vivo, nor had its biological relevance ever been substantiated.

(Pharmacia Gene Assembler) and purified by ion-exchange chromatography [50 x 5 mm Mono Q column (Pharmacia)]. Mutagenesis was carried out according to the ung-dut method (15). E. coli strain RZ1032 was used for production of uridylated single-strand DNA, and strain JM1O1Tr (recA derivative of JM101) was used for mutant clone selection. Mutagenesis template M13mp19amIVX1 was constructed by ligating the Sca I-HindIII insert fragment of plasmid pX2 (16) to HincII- and HindIII-linearized M13mpl9amIV phage DNA. Mutant clones were screened by differential hybridization after adsorption of phage DNA on nitrocellulose (17). Shuttle Vector Construction. To make unique the EcoRI site located in front of the metG gene carried by the methionyl-tRNA synthetase-p-galactosidase hybrid (MRSH) expression vector pNav9 (14), the Acc I-Nar I region (carrying the second EcoRI site) was replaced by the Acc I-Asu II fragment from pMC1403 (18). All fragments were purified by size-exclusion chromatography (19) prior to ligation. The resulting plasmid was called pNavlO. To facilitate reconstitution of the 19 variant genes, an EcoRI-HindIII adaptor was ligated with plasmid pNavlO linearized by the same enzymes. JM101Tr cells transformed by the resulting shuttle vector pNavll yielded pale blue colonies on 5-bromo-4-chloro3-indolyl f-D-galactoside (X-Gal) plates (Fig. 1). For reconstruction of vectors carrying the 19 variants of the hybrid gene, the corresponding EcoRI-HindIII insert fragments prepared from the 19 mutagenized templates M13mp19V1

(14).

N-terminal sequence analyses of each of the 20 variants led us to propose a molecular enzymatic model accounting for the experimental results. This model is confirmed by a parallel computer analysis of the hundreds of known Nterminal sequences of soluble E. coli proteins.

MATERIALS AND METHODS Mutagenesis. Mutagenic oligonucleotides were synthesized

Abbreviations: MAP, methionyl-aminopeptidase; MRSH, methionyl-tRNA synthetase-f3-galactosidase hybrid; X-Gal, 5-bromo-

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

4-chloro-3-indolyl 8-D-galactoside; PTH, phenylthiohydantoin; NBRF, National Biomedical Research Foundation. 8247

8248

Biochemistry: Hirel et al.

were ligated to purified EcoRI-HindIII-linearized pNavll plasmid. Clones harboring the reconstituted hybrid gene were identified as blue colonies on X-Gal plates. The corresponding plasmids were named pMRSH-X, in which X is the one-letter code for the second residue. Chimaeric Protein Purification. MRSH variants (designated as MRSH-X, in which X stands for the amino acid specified by the second codon) were purified by affinity chromatography on a 2-ml p-aminophenyl f3-D-1-thiogalactopyranoside-agarose column (20), as previously described (14). Concentrations of MRSHs in crude cell extracts were calculated from 83-galactosidase activity measurements. It was assumed that the specific activity of each of the 19 variants was identical to that of the homogeneous wild type, MRSH-T (2.125 /imol of o-nitrophenyl f3-D-galactopyranoside hydrolyzed per sec per mg of MRSH-T at 280C). Volumes of crude bacterial extracts applied on the affinity chromatography column were adjusted to isolate 10 nmol of each MRSH species. Concentrations of pure MRSHs were derived from absorbance measurements at 280 nm, using an extinction coefficient of 1.78 mlhmg-1 cm-l (14). N-Terminal Sequencing. Protein samples (300 pmol) dissolved in 1 mM ammonium bicarbonate, pH 8.2, were analyzed for 10 cycles by means of an Applied Biosystems model 470A sequencer. Neat glass fiber filters were used for sample loading (without Polybrene pretreatment). Released phenylthiohydantoin (PTH) derivatives of amino acids were identified on line, after separation on an isocratic reversephase liquid chromatography system described previously (21). An IBM PC AT computer connected to a Nelson Analytical interface (Cupertino, CA) was used for data acquisition and processing. The N terminus of the MRSH-C variant was first sequenced in its underivatized form, then the presence of cysteine was ascertained after derivatization by 4-vinylpyridine (22). Calculation of the Extent of Methionine Removal. Direct measurements of methionine excision could not be effected for several MRSH samples which were contaminated by variable amounts of an unknown protein starting with the sequence Met-Glu-Lys-Lys-Tyr-Ile-Val-Val-Leu-Asp-. Thus, the proportion of processed versus nonprocessed species of MRSHs was calculated by measuring the concentration of the PTH derivative of a particular residue at two consecutive cycles for which the level of contamination was found to be reproducibly low. Valine occurring in the sequences of MRSHs as (Met)-Xaa-Gln-Val-Ala- was chosen for this purpose. In cases where methionine excision was negligible-i.e., MRSH-H, -Q, -E, -F, -K, -Y, -W, and -R-the value of methionine excision was set as zero, and the mean value of valine carryover (cl) from cycle 4 to cycle 5 was determined as

Val(5) Val(4) where Val(n) stands for the amount of valine PTH released at sequence cycle n. For other MRSHs, the percentage of excised methionine (p) was then calculated from the expression Val(3) = Val(4) + Val(3) - wVal(3) RESULTS Construction of a metG-lacZ Hybrid Gene. The proteins used as in vivo substrates of the MAP activity were expressed

Proc. Natl. Acad. Sci. USA 86 (1989)

from a hybrid gene composed of the first 623 codons of inetG, fused to the 3' part (1016 codons) of lacZ. When the wild-type metG sequence was used, the resulting chimaeric protein retained both the methionyl-tRNA synthetase and the 18galactosidase activities. The protein could be purified in a single step, by affinity chromatography specific for the 13galactosidase moiety (20). Because both chimaeric protein domains are able to fold independently (14), this purification process was expected to be insensitive to any point modification introduced in the synthetase domain. Twenty hybrid gene variants were constructed by systematic substitution of the second codon, using oligonucleotide site-directed mutagenesis. As schematized in Fig. 1, the second codon transition was performed by using as a template a recombinant M13 phage (M13mp19X1) carrying the first 277 codons of metG. In all cases, the second codon was replaced by a codon exhibiting a usage as close as possible to that of the wild-type ACT (threonine) codon (3. 1%, as deduced from sequences retrieved from the National Biomedical Research Foundation (NBRF) (February 1985 version) and GenBank (release 32). The hybrid gene variants were then reconstituted by inserting the EcoRI-HindIII fragments, derived from the 20 engineered M13mp19X1 variants, between the same restriction sites on the plasmidic shuttle vector pNavll. Transformed JM1O1Tr cells carrying the plasmids (pMRSHs) with the reconstituted hybrid genes were scored as blue colonies on rich medium containing the X-Gal chromogenic substrate. Expression of the Chimaeric Proteins. Strain JM1O1Tr was used to produce the 20 hybrid proteins from the pMRSH plasmids. Expression of the proteins was followed by measuring 3-galactosidase activity in crude extracts from cells harvested in the stationary phase (after overnight culture). It HindIII

EcoRI

EcoRI AccI

BamHI l

lN pNavZ9 - - -u 1;11

'lacZ

metG'

t

replaced by a AccI-AsuII fragment from pMC1403 to remove the EcoRI site

HindIII

EcoRI

NarI

BamH I

pNavlO

mM-

repi acement

of the

EcoR 1-HindIII fragment by a EcioRI-HindIII

Jrl

adaptator

EcoRI HindIII BamB[I

pNavl 1

ziI1 .~w.

EcoRI

/

Hin;,dIII

THR

ATG ACT GGT GCT GAA ATG TAC TCT GTT AAC CTG

Gly Ala Glu Met Tyr Ser

Val Asn Leu

A

.

5'-terminal parts of the 19 metG-LacZ hybrid genes were reconstituted by insertion of Ithe mutagenized fragments M13 mpl9xl CGT CCG TTC

Arg Pro Phe

AAA

Lys Trp Gln Cys Asp Ile

TGG CAG TGC GAC ATC

FIG. 1. Construction of the 20 hybrid gene variants. The three schematical maps in the upper part of the figure summarize the main steps of the pNav1l shuttle vector construction. In the lower part of the figure, the relevant portion of M13mp19X1 is shown together with the second codon transitions.

Biochemistry: Hirel et al.

Proc. Natl. Acad. Sci. USA 86 (1989)

verified that this procedure ensured optimal MRSH-T hybrid production. Hybrid concentrations in the extracts were calculated by assuming that the specific activity of the /3-galactosidase moiety was not affected by the composition of the N terminus of the methionyl-tRNA synthetase moiety. After purification on the affinity column, the recovered amounts of each of the 20 hybrid proteins were in close agreement with the values expected from the f3-galactosidase activity measurements in the crude extracts, thus justifying the above assumption. Yields were reproducibly around 75%, the purity of hybrid proteins being more than 95%. Purified chimaeric proteins were assayed for their activity in the tRNAMet aminoacylation reaction. In 18 cases (Fig. 2), the second amino acid substitution and/or the N-terminal methionine excision did not significantly affect the catalytic activity of the synthetase moiety-i.e., activity values did not differ by a factor greater than 2, as compared with the wild-type hybrid protein (MRSH-T hybrid). In contrast, the

4

was

-* Arg substitution led to a 75% decrease of the tRNA aminoacylation activity, and the Thr -* Cys transition abol-

Thr

ished this activity, without change of the specific activity of the p-galactosidase moiety. The introduction of an additional sulfhydryl function might have resulted in the formation of an illicit S-S bridge, leading to unfunctional folding of the polypeptide. In contrast, the production of the various hybrid proteins appeared highly sensitive to the second codon substitution. According to the second codon efficiency rule of Looman et al. (23) established with the lacZ gene, the translation initiation rate of a gene strongly depends on the nature of its second codon. As shown in Fig. 3, a plot of the values of hybrid production as a function of the introduced amino acid fits in well with the profile predictable from Looman's rule. However, in the cases of MRSH-A, -K, and -R, the productions were significantly lower than the levels predicted from this rule. Such differences may reflect a relative instability of these hybrids, since, in this study, we measure the accumulation of hybrids rather than their rate of expression. Extent of Methionine Excision Within the Chimaeric Proteins. The N-terminal sequences of the 20 hybrid variants were analyzed by means of automated gas-phase Edman degradation. First, in the cases of MRSH-H, -Q, -E, -F, -K, -Y, -W, and -R, quantitative measurement of released PTHs showed essentially methionine at the first sequence cycle. The amino acid specified by the second codon represented less than 1% of the total released methionine. Thus, these hybrids were classified as nonprocessed-i.e. their excision value was set to zero (cf. Table 1). In all other cases, initiator methionine was removed to variable extents. Direct measurement of PTHs present at the first cycle could not be reliably applied to all cases for the following 200

>%200 0

co

._

0

0.

C

100

I

A.

Ion

.2-

0c o- =

100

N-

IL. u O IjGGV NH A PS T V C ND L I HOQE F M K YW R

E-

S%

.O

g o

Q

C

I

FIG. 2. Production of the 20 MRSH variants. Hybrid concentrations in crude E. coli extracts were deduced from .8-galactosidase activity measurements, taking a constant value of 2.125 ,umol of o-nitrophenol produced per sec per mg of hybrid protein. Concentrations are expressed as pmol of hybrid per absorbance unit at 280 nm (black bars). Specific tRNAMCI aminoacylation activities, normalized to the activity of MRSH-T, were calculated by using the corresponding hybrid concentrations.

8249

3 c

0 U)

2

0. x

0) 0)

o

1711

1

11

G A P S T V C N D L

11

-I

HO E FM K YW R

FIG. 3. Influence of the second codon on the expression of the 20 hybrid proteins. The accumulation of the 20 MRSHs during overnight culture (-) is compared to the relative rate of initiation (o), as defined by Looman et al. (23). Production measurements and efficiency values are normalized to those of the variants carrying the threonine codon ACT in second position.

(i) Hybrid protein samples must be homogeneous. (ii) The method requires a precise knowledge of the specific yields of coupling of phenylisothiocyanate and of cleavage of the peptide bond for each of the 20 amino acids in the Met-Xaa-Glnand Xaa-Glnsequence environments. Our set of 20 MRSHs did not meet the first criterion. Indeed, a 10-fold variation has been observed between the expression of the various hybrid proteins, thus resulting in variable levels (0-5%) of contaminants. Consequently, a calculation method, described in Materials and Methods, was developed. It is based on the fact that two sequences with a one-residue shift are observed simultaneously when the initial methionine is not fully processed. Accordingly, the ratio of the PTH signal of a given residue at cycles n and n + 1, after correction for the carryover effect characteristic of the Edman chemistry, reflects the relative amount of processed and nonprocessed sequences. In MRSHs [N-terminal sequence (Met)-Xaa-Gln-ValAla-], the Val residue appeared particularly suitable to apply this calculation method, because in all MRSH preparations the contamination in its corresponding sequence cycle was low and constant. Calculated values of methionine excision are reported in Table 1. They agree within experimental error (±5%) with the values directly measured by the abundance of PTHs at cycle 1 in the case of the MRSH samples void of contamination by other proteins. The calculation method using the determination of Val concentration was also applied to derive the value of processing of the MRSH-M species. Interestingly, values in Table 1 indicate that excision of the N-terminal methionine is not a hit-or-miss process. In addition, there is no correlation between MRSH expression and the extent of methionine removal. For example, levels of expression of the serine and tryptophan variants were roughly identical, while their respective extents of Nterminal processing were 84% and 0%, respectively. Thus, the extent of methionine excision is likely to reflect the catalytic efficiency of MAP, as specified by the penultimate N-terminal residue of its substrate. The results obtained show that the extent of excision decreases when the size of the introduced second amino acid increases. Several physical parameters may characterize the size of an amino acid: accessible surface area, side-chain volume, gyration radius, etc. (9, 24-27). As shown in Fig. 4, the extent of N-terminal methionine excision could be satisfyingly related to the maximal distance (length parameter) between the a-carbon and the more distal nonhydrogen atom of the penultimate residue. reasons:

...

...

Biochemistry: Hirel et al.

8250

Proc. Natl. Acad. Sci. USA 86 (1989)

Table 1. Initiator methionine processing N-terminal methionine Max. Methionine in E. coli proteinst side-chain processing,t Removed Nonremoved length,* A t 0.00 97.1 2 0 1.51 95.8 30 0 2.40 88.2 ± 3.0 (2) 5 0 2.41 84.0 19 0 2.54 89.7 ± 4.5 (4) 7 1 2.55 83.7 0 1 2.83 71.0 ± 4.7 (4) 2 0

Amino acid in position 2 Gly Ala Pro Ser Thr Val Cys

3.68 16.4 Asn 0 2 3.74 Asp 16.1 0 2 Leu 3.90 16.3 0 2 lie 3.91 18.4 ± 3.5 (3) 0 5 His 4.64 0.0 0.0 4.93 0 5 Gln 7 Glu 4.97 0.0 0 Phe 5.10 0.0 0 1 Met 5.46 0.0 6.37 0.0 to 4.0 (5) 0 6 Lys 1 6.43 0.0 0 Tyr 0.0 6.64 Trp 7.40 0.0 0 3 Arg The extent of methionine processing determined for the 20 MRSHs was compared to the frequency of N-terminal methionine removal for 101 E. coli protein sequences. *Maximal distance between the a-carbon and the most distal nonhydrogen atom of the side chain. tValues of methionine excision determined from the 20 MRSH variants; results are mean ± SD and the number of determinations is given between parentheses. tNumber of cases encountered in the data bases of the NBRF and GenBank. Occurrence or nonoccurrence of methionine excision was established by comparing the DNA-encoded sequence with the corresponding mature protein sequence.

N-Terminal Methionine Excision Within E. coli Proteins. To validate our above definition of the MAP specificity, a systematic survey of all known N-terminal sequences of cytoplasmic E. coli proteins was undertaken. For this purpose, the DNA-encoded sequences (GenBank release 32) were compared to the corresponding mature protein sequences (NBRF release 4). The results from a set of 101 proteins for which the desired information was available are summarized in Table 1. They clearly show a correlation between methionine removal and the occurrence of Gly, Ala, Ser, Cys, Pro, or Thr in second position of the immature form of the protein. It should be noted that we could not find any correlation between the methionine excision and any other

A

G

I01

PST,V 8'

0

1

c

0

05

6' 0

4' 0_ 2'

0-

0 0

~~M,K,Y,W

NDLI

,,~

2

3

,9,

4

5

6

7

8

maximum side chain length (A) FIG. 4. Correlation between the extent of methionine processing and the maximum side-chain length of the penultimate amino acid.

parameter such as the secondary structure of the first 50 N-terminal residues, the conservation of a sequence at the level of either DNA or amino acid sequence, a peculiar codon usage, or the functional nature (hydrophobicity, acidic or basic character, .) of the N-terminal amino acids. Thus, the conclusion of the present survey of any E. coli soluble protein was qualitatively in agreement with the model study performed on the 20 variants of a constant MRSH substrate. .

DISCUSSION Through a systematic and quantitative analysis of N-terminal methionine excision, we may conclude that the E. coli MAP enzyme obeys the following rule: the catalytic efficiency of MAP-i.e., the extent of cleavage of a given substrateincreases when the length parameter of the penultimate amino acid decreases. Previously reported data on the maturation of endogeneous proteins expressed in E. coli are consistent with this rule (cf. Table 1). In addition, according to previous systematic studies, the rule appears valid also in yeast (11, 13) and is likely to apply to higher eukaryotic cells (11, 12). The present study enlarges the validity of the rule to cases of immature E. coli sequences such as Met-Met-, Met-Trp-, and Met-His-, for which no data were available in updated releases of NBRF and GenBank (versions 18 and 57, respectively). These sequences are not processed by the E. coli MAP, in agreement with the observations made in the yeast context (11, 13). The length parameter rule also applies to a Met-Prosequence. In other words, the catalytic efficiency of MAP is insensitive to this peculiar imide bond. Indeed, the extent of methionine removal from MRSH-P was found in close agreement with the value predicted from the rule (cf. Table 1). However, Ben-Bassat et al. (8) observed in vitro that a proline in third position inhibited MAP action, even when the second residue was such that the cleavage should be favored. To verify this point, a MRSH variant with a Met-Ala-ProN-terminal sequence was designed and expressed in vivo. The extent of methionine removal in this variant (result not shown) confirmed the inhibitory role of proline at position 3: excision dropped to 62% in the case of the Met-Ala-Pro-Valsequence, which may be compared with 91% in the case of Met-Ala-Gln-Val- (cf. Table 1). Furthermore, at least 6.5% of the processed Met-Ala-Pro-Val- hybrid protein had also lost the second Ala residue. This inhibition may reflect a reduced accessibility of the N-terminal dipeptide to MAP, due to a more rigid conformation of the peptidic backbone. It is reasonable to imagine that the active center of MAP comprises one subsite specific for the N-terminal nonacylated methionine and a second site, able to accommodate the penultimate amino acid, provided that its maximal side-chain length does not exceed 4 A. The occupation of this second subsite would be essential to trigger the peptide bond hydrolysis by the enzyme. Consequently, each of the 20 possible N-terminal dipeptide sequences must be recognized and processed by MAP with a characteristic catalytic efficiency. In this context, the extent of methionine cleavage for a given protein is expected to reflect a stationary state within the cell, resulting from (i) the translation initiation rate, (ii) the MAP intracellular concentration, and (iii) the catalytic efficiency specified by the N-terminal dipeptide sequence. This should remain valid whether the processing occurs posttranslationally or cotranslationally. In the second case, the stationary state would result from (i) the saturation level of ribosomes by MAP, (ii) the catalytic constants specific for the N terminus, and (iii) the residence time of the N-terminal extremity of the nascent polypeptide on the ribosome-MAP complex.

Biochemistry: Hirel et al. The occurrence of a stationary state is consistent with the following observations. (i) The extent of processing of a particular protein varies upon cell growth conditions (28, 29). (ii) N-terminal methionine excision of 83-galactosidase and human tumor necrosis factor a in E. coli is strongly stimulated by the addition of 1 mM manganese ion to the growth medium (30). To date, there is no clear evidence arguing in favor of a posttranslational methionine removal, rather than in favor of a cotranslational one. In vitro experiments have indicated that MAP is capable of removing the N-terminal methionine from folded proteins as well as from peptides, with a specificity in agreement with the maximal length rule (8). In vivo, MAP overproduction in E. coli has been observed to increase the extent of maturation of two exotic protein species [interleukin 2 and ricin A (8)]. In interpreting the above results, we have to take into account that, in the E. coli cell, the number of MAP molecules per ribosome is limiting. Their ratio is roughly equal to 0.1, as calculated from data published in refs. 8 and 31. Thus, a cotranslational MAP action should be improved by an increase of the enzyme cellular concentration. The incorporation of a methionine at the N-terminal position of proteins results from the usage of an AUG codon as the universal translation initiation signal. Now we shall discuss why a specialized enzymatic system has been retained by evolution to conditionally remove this residue. From our results, it is likely that methionine excision is not required to reveal the activity of a protein that is otherwise inactive. On the other hand, it is plausible that protein stability is specified by its N-terminal sequence. A rule, called the N-end rule, has been recently established in yeast. According to this rule, the lifetime of a protein, not beginning with a methionine, decreases (from 20 hr to 2 min) as the size of the N-terminal residue increases (32). Interestingly, this rule, which governs protein lifetime, is the reciprocal of the rule specifying MAP catalytic efficiency. Consequently, any protein having a second residue determining a short lifetime when exposed in N-terminal position should be a poor substrate of MAP and, thus, should retain its initiator methionine. However, as shown in Table 1, some middle-sized residues (Asn, Asp, Leu, and Ile) are submitted to significant N-terminal processing. Therefore, the degree of MAP cellular activity might be crucial to determine the lifetime of those proteins harboring such middle-sized penultimate residues. Beyond a plausible role of MAP on protein turnover in the cell, we may consider an additional reason to the occurrence of such a specific enzymatic system. In E. coli, the number of methionine molecules sequestered at the N terminus of unmatured polypeptides is 2 x 105 molecules per cell (calculated from ref. 31 and from this study). This value is about twice as big as the free methionine pool in the cytosol of cells grown in the absence of added methionine (105 molecules per cell, calculated from ref. 33). Therefore, we suggest that MAP action may be important in the recycling of a significant part of the cellular methionine pool, otherwise unusefully sequestered at the N termini of proteins. In this context, the occurrence of a MAP activity would present the double

Proc. Natl. Acad. Sci. USA 86 (1989)

8251

advantage of economizing an essential amino acid in animal cells and delaying the commitment of the methionine regulon in starved microorganisms. This work was supported by Grant U.A. 240 from the Centre National de la Recherche Scientifique and the Decision d'Aide 87-C0392 from the Ministdre de la Recherche et de l'Enseignement Supdrieur. Ph.-H.H. was supported by a fellowship from Rh6nePoulenc. 1. Marcker, K. & Sanger, F. (1964) J. Mol. Biol. 8, 835-840. 2. Adams, J. M. (1967) Ph.D. Thesis (Harvard Univ., Cambridge, MA). 3. Waller, J.-P. (1963) J. Mol. Biol. 7, 483-496. 4. Pine, M. J. (1969) Biochim. Biophys. Acta 174, 359-372. 5. Adams, J. M. (1968) J. Mol. Biol. 33, 571-589. 6. Fry, K. T. & Lamborg, M. R. (1967) J. Mol. Biol. 28, 423-433. 7. Livingston, D. M. & Leder, P. (1969) Biochemistry 8,435-443. 8. Ben-Bassat, A., Bauer, K., Chang, S.-Y., Myambo, K., Boosman, A. & Chang, S. (1987) J. Bacteriol. 169, 751-757. 9. Miller, C. G., Strauch, K. L., Kukral, A. M., Miller, J. L., Wingfield, P. T., Mazzei, G. J., Werlen, R. C., Graber, P. & Movva, N. R. (1987) Proc. Natl. Acad. Sci. USA 84, 27182722. 10. Vogt, V. M. (1970) J. Biol. Chem. 245, 4760-4769. 11. Tsunasawa, S., Stewart, J. W. & Sherman, F. (1985) J. Biol. Chem. 260, 5382-5391. 12. Boissel, J. P., Kasper, T. J. & Bunn, H. F. (1988) J. Biol. Chem. 263, 8443-8449. 13. Huang, S., Elliott, R. C., Liu, P.-S., Koduri, R. K., Weickmann, J. L., Lee, J.-H., Blair, L. C., Ghosh-Dastidar, P., Bradshaw, R. A., Bryan, K. M., Einarson, B., Kendall, R. L., Kolacs, K. H. & Saito, K. (1987) Biochemistry 26, 8242-8246. 14. Hirel, Ph.-H., Leveque, F., Mellot, P., Dardel, F., Panvert, M., Mechulam, Y. & Fayat, G. (1988) Biochimie 70, 773-782. 15. Kunkel, T. H. (1985) Proc. NatI. Acad. Sci. USA 82, 488-492. 16. Dardel, F., Fayat, G. & Blanquet, S. (1984) J. Bacteriol. 160, 1115-1122. 17. Zoller, M. J. & Smith, M. (1983) Methods Enzymol. 100, 468-500. 18. Casadaban, M. J., Chou, J. & Cohen, S. N. (1980) J. Bacteriol. 143, 971-980. 19. Schmitter, J. M., Mechulam, Y., Fayat, G. & Anselme, M. (1986) J. Chromatogr. 378, 462-466. 20. Ullmann, A. (1984) Gene 29, 27-31. 21. Beauvallet, C., Hountondji, C. & Schmitter, J. M. (1988) J. Chromatogr. 438, 347-357. 22. Andrews, P. C. & Dixon, J. E. (1987) Anal. Biochem. 161, 524-528. 23. Looman, A. C., Bodlaender, J., Comstock, L. J., Eaton, D., Jhurani, P., De Boer, H. & Van Knippenberg, P. H. (1987) EMBO J. 6, 2489-2492. 24. Chothia, C. (1984) Annu. Rev. Biochem. 53, 537-572. 25. Rose, G. D., Geselowitz, A. R., Lesser, G., Lee, R. H. & Zehfus, M. H. (1985) Science 229, 834-838. 26. Shrake, A. & Rupley, J. A. (1973) J. Mol. Biol. 79, 351-371. 27. Levitt, M. (1976) J. Mol. Biol. 104, 59-107. 28. Yamada, T., Kato, K., Kawahara, K. & Nishimura, 0. (1986) Biochem. Biophys. Res. Commun. 135, 837-843. 29. Latta, M. (1988) Thesis (Institut National de la Recherche

Agronomique, Paris-Grignon). 30. BASF AG (1987) F.R.G., Patent DE 3532134 Al. 31. Dennis, P. P. & Bremer, H. (1974) J. Bacteriol. 119, 270-281. 32. Bachmair, A., Finley, D. & Varshavsky, A. (1986) Science 234, 179-186. 33. Kadner, R. J. (1974) J. Bacteriol. 117, 232-241.