Role of Amino Acid Sequences Flanking Dibasic Cleavage Sites in Precursor Proteolytic Processing

Eur. J. Biochem. 227, 707-714 (1995) 0 FEBS 1995

Role of amino acid sequences flanking dibasic cleavage sites in precursor proteolytic processing The importance of the first residue C-terminal of the cleavage site Mohamed RHOLAM', Noureddine BRAKCH', Doris GERMAIN', David Y. THOMAS', Christine FAHY I , Hamadi BOUSSETTA', Guy BOILEAU3 and Paul COHEN' Biochimie des Signaux RCgulateurs Cellulaires et MolCculaires, UniversitC Pierre et Marie Curie, Unit6 de Recherches AssociCe 1682 au CNRS, Paris, France Eukaryotic Genetics Group, National Research Council of Canada Biotechnology Institute, MontrCal, Canada DCpartement de Biochimie, UniversitC de MontrCal, MontrCal, Canada (Received 28 October/9 December 1994) - EJB 94 1654/4

The amino acid sequences flanking 352 dibasic moieties contained in 83 prohormones and pro-proteins listed in a database were examined. Frequency calculations on the occurrence of given residues at positions P, to P: allowed us to delineate a number of features which might be in part responsible for the in vivo discrimination between cleaved and uncleaved dibasic sites. These include the following: amino acids at these positions were characterized by a large variability in composition and properties; no major contribution of a given precursor subsite to endoprotease specificity was observed; some amino acid residues appeared to occupy preferentially certain precursor subsites (for instance, Met in P, and P,, Asp and Ala in P;, Pro in P,, Gly in P, and P: etc.) whereas some others appeared to be excluded. Most amino acid residues occupying the P: position in these precursor cleavage sites were tolerated. But the P-carbon branched side chain residues (Thr, Val, Leu, Ile) and Pro, Cys, Met and Trp were either totally excluded or poorly represented, suggesting that they might be unfavourable to cleavage. The biological relevance of these observations to the efficacy of dibasic cleavage by model propeptide convertases was in vitro tested using both pro-ocytocin convertase and Kex2 protease action on a series of proocytocin related synthetic substrates reproducing the Pro7+Leul5 sequence of the precursor in which the Ala13 residue (Pi in the LysArg-Ala motif) was replaced by various amino acid residues. A good correlation was obtained on this model system indicating that Pi residue of precursor dibasic processing sites is an important feature and may play the role of anchoring motif to S: convertase subsite. We tentatively propose that the present database, and the corresponding model, may be used for further investigation of dibasic endoproteolytic processing of propeptides and pro-proteins. Keywords. Pro-ocytocin convertase ; Kex2 endoprotease ; p-carbon branched amino acids ; precursor sequence database.

Limited proteolysis of protein and peptide precursors by selective cleavage of peptide bonds is a general mechanism of bioactivation in biosynthetic processes. Although various sites of endoproteolysis have been described (reviewed in [l -4]), it appears that the most frequent processing loci consist in basic residues organized as singlets, doublets or higher order arrangements [5-61. There is now clear evidence that, in the case of peptide hormone precursors, as well as other proproteins like pro-receptors, pro-viral proteins and pro-enzymes, the most frequently occurring arrangements are doublets of the LysArg or ArgArg types [7]. From the original observation of extended forms of peptide hormones bearing on their N-terminus or Cterminus extra basic amino acid residues, was derived the concept of trypsin-like activation of precursors [8]. At the present Correspondence to P. Cohen, Biochimie des Signaux RCgulateurs Cellulaires et MolCculaires, UniversitC Pierre et Marie Curie, Unit6 de Recherches AssociCe 1682 au CNRS, 96 Bd Raspail, F-75006 Paris, France Abbreviations. PL subsite, the nth residue C-terminal of the cleavage site; Pn, the nth residue N-terminal of the cleavage site. Enzyme. Pro-ocytocin couvertase :magnolysin (EC 3.4.24.62).

time, a family of subtilisin-like, serine proteases homologous to the KEX2gene product of Saccharomyces cerevisiae [9] and to the furin gene [lo-121 have been identified as serious candidates in processing prohormones in vivo at basic amino acid pairs [13, 141. However, a number of endoproteases related to other enzyme families were reported and they were also proposed as putative processing enzymes [15 -231. Whereas the question of their exact relevance to in vivo mechanisms remains open, little is known, presently, about the detailed mechanism of processing enzyme action and about their substrate requirements. In contrast, recombinant DNA techniques have allowed the determination of the sequences of a great number of precursors precisely in the peptide hormone family. The selective pressure exerted on the evolution of these macromolecular precursors has led to the high conservation of basic residues as cleavage sites. However, examination of these sequences clearly indicated that whereas a number of the potential basic sites are indeed cleaved in vivo a significant percentage of these loci remain unprocessed (reviewed in [l-41). Moreover, differential processing in a cell-specific or tissue-specific manner which occurs in a number of cases suggested a discrimi-

708

Rholam et al. (Eul: J. Biochem. 22TI

nation of these sites. This might be, at least in part, due to the contribution of pleptide precursor conformation [7, 241. In an effort toward an understanding of the structural parameters which govern substrate recognition by the processing enzyme(s), several authors have attempted to define the amino acid sequences and/or higher order structures which may participate in the definition of the processing-enzyme-binding sites [18, 25-30]. Previous analysis conducted on 20 prohormones had led to the conclusion that these cleavage loci which are indeed processed in vivo may belong to exposed and flexible regions of the precursors situated in, or immediately next to, p-turns [7] or alternatively larger loops [31]. Existence of these structures was supported by preliminary observations [25, 261. They were characterized by one-dimensional and two-dimensional NMR spectroscopy in the case of the processing sequences around the LysllArgl2 doublet of the common precursor for ocytocin and neurophysin [27, 281 and around the dibasic sites of human proinsulin [32]. Although both the functional role of these structures in the substrate-binding site for processing enzyme recognition and their implicatiion in processing of many prohormones were only recently shovvn [29, 30, 331, a series of observations using site-directed mutagenesis of various prohormones had also underlined the particular importance of some amino acid residues other than the dibasic ones in specifying correct processing at the cleavage sites 125, 30, 34-39]. Since dibasic sites represent the large majority of cleavage motifs present in pro-peptides and pro-proteins, we have now used an extended database including the sequence of 83 protein and peptide precursors to analyze the occurrence of those amino acid residues which flanked 352 doublets of basic residues. The results of this analysis allow us to identify some amino acids at given positions around the dibasic cleavage sites which might constitute key structural features for proteolytic processing at these loci. Moreover they demonstrated that the key role of first (P:) residue C-terminal of the cleavage site is a common feature to all studied precursors. The potential of these statistical data was illustrated by an experimental evaluation of the role of those residues which occupy position Pi in precursor substrates and may preferentially interact with the enzyme subsite binding the P{ substrate amino acid side chain (S:) of processing endoprotease(s).

MATERIAL AND1 METHODS Peptide synthesis. The peptides listed in Table 5 were synthesized by the solid-phase method [40] using a NPS 4000 semi-automated multisynthesizer (Neosystem). They have been numbered in accordance with previous publications from this laboratory [26, 291. Peptide purification and analysis were routinely conducted using a set of analytical techniques including HPLC, amino acid 'composition, N-terminal sequencing and fast atomic bombardment mass spectrometry as in Nicolas et al. [41]. Enzyme assay. Each of the above described peptides were tested using purified preparations of pro-ocytocin convertase isolated from bovinle neurohypophysis as described in Plevrakis et al. [16] and of the recombinant Kex2 enzyme produced in baculovirus-infected cells as described elsewhere [42]. The standard assay was as follows. The peptide substrates in the range 100-2000 pM were incubated with an aliquot of purified enzyme ( 3 - 1.6 pg protein) in a final volume of 50 p1 buffer at pH 7.0. In the case of pro-ocytocin convertase, the incubations were made in 0.1 M ammonium acetate for 5 h whereas for the Kex2 enzyme, 0.1 W[ Hepes/Tris was used at different incubation times according to the peptide used.

At the end of each incubation period, the reaction was stopped with 0.1 M HC1 (10 pl) and the entire sample was subjected to HPLC analysis on a C,, column (ultrabase; SFCCShandon) eluted with a gradient from 5% to 40% or from 15% to 40% acetonitrile, 0.05% trifluoroacetic acid in 30 min at a flow rate of 0.5 mumin. Each of the fragments generated by the protease were identified by reference to standards and/or by amino acid composition using a Waters Picotag station. Controls were conducted to show that under these reaction conditions the enzyme activity was stable. Each assay was run at least in triplicate and the mean value was evaluated (SD < 5 %). K,,, and V,,,, of different peptide substrates (Table 5) were determined from initial velocity measurements plotted versus various substrate concentrations using the Lineweaver-Burk representation.

RESULTS Constitution of the database. To identify structural requirements in the precursor substrates that might define endoprotease specificity, a database containing the amino acid sequences around 352 potential processing sites from 83 peptide hormone and protein precursors was established. All calculations were made on this set of precursors by considering only the dibasic cleavage sites which were selected on the following basis: To avoid unsuitable repetition of identical sequences, only a unique precursor sequence corresponding to a single animal species was introduced in the bank and further considered in the calculations. When a polyfunctional precursor contained in its primary structure more than one copy of the given bioactive segment flanked by either an identical or homologous cleavage sequence pattern, only one of these motifs was taken into account in the statistical calculations. To emphasize the occurrence of given amino acid species in the protein segments flanking the dibasic potential cleavage sites, residues from position -4 (P,) from the dibasic to +4 (P:) were considered in this analysis. These positions correspond to the minimum substrate length required for cleavage by the pro-ocytocin convertase 116, 26) and for accessibility of cleavage sites [27-291. Based on the available data in the literature relative to precursor processing, the total number of dibasic motifs were grouped into 172 canonical sites known to be processed in vivo (49 %), whereas the remaining 180 stretches (51 %) were known to remain unaffected during proteolytic processing and were recovered intact either in precursor connecting fragments or in the mature peptide and protein sequences. Analysis of amino acid residues around the dibasic cleavage sites. Table 1 shows the occurrence [N,] of each given residue of type j at each of the respective positions i around the cleaved and uncleaved dibasic residues, respectively. Although none of the eight subsites had an absolute requirement for a given type of amino acid, one could observe however a difference between the two classes of sites, i.e. cleaved (cl) versus uncleaved (cn). Indeed, the observed mean values [(N,)"/(N,)'"] indicated that the majority of amino acids around the cleavage sites presented a significant variability in their mean distribution (Table 1). According to these observations, the parameter [R,],,,,, (= [f,],,/ [fJcn), defined as the ratio of the frequencies of a given residue i at position j in the cleaved site population ([f&) to that observed for the uncleaved ones ([f,Icn), was used to characterize differences and similarities which might exist between the two types of dibasic cleavage sites.

709

Rholam et al. ( E m J. Biochern. 227)

Table 1. Number of amino acids present at given P or P' positions around 172 cleaved and 180 uncleaved dibasic sites in a series of 83 protein and peptide precursors. (N,) is the mean abundance observed for the eight subsites in populations of cleaved (cl) and uncleaved (nc) sites. The values obtained for the uncleaved sites are indicated in parentheses. Amino acids

Number at subsite and position P6 -4

Phe TYr TrP His Ala Val Leu Ile Thr Ser Met CYS GlY Pro Asn Gln ASP Glu '4% LYS

p5

-3

Mean values

P4 -2

P3 -1

p: +l

P; +2

p: +3

22 (6) 7 (5) 3 (1) 1 (5) 4 (16) 12 (13) 13 (10) 7 (11) 3 (7) 5 (14) 10 (4) 0 (5) 10 (10) 13 (7) 6 (10) 9 (5) 3 (9) 14 (18) 23 (13) 7 (11)

4 (7) 4 (3) 1 (5) 3 (3) 10 (26) 11 (13) 18 (19) 2 (10) 7 (11) 13 (11) 8 (1) 0 (6) 43 (19) 5 (10) 3 (10) 8 (5) 15 (6) 17 (16) -

13 (7) 22 (11) 0 (3) 10 (7) 14 ( 5 ) 3 (12) 3 (19) 4 (12) 0 (7) 28 (17) 1 (5) 2 (5) 10 (19) 3 (15) 8 (7) 10 (13) 18 (4) 23 (12) -

6 (6) 3 (5) 1 (4) 7 (5) 27 (7) 8 (12) 20 (18) 2 (9) 6 (6) 18 (13) 12 (4) 3 (4) 20 (10) 8 (12) 4 (6) 2 (4) 7 (13) 9 (14) 3 (16) 6 (12)

2 (6) 3 (4) 1 (7) 3 (4) 4 (13) 5 (8) 5 (15) 4 (5) 5 (5) 14 (12) 4 (3) 5 (8) 19 (11) 17 (8) 5 (7) 10 (9) 15 (12) 30 (24) 14 (11) 5 (8)

p: +4

(NuY'/(Nc,F

1.61 1.11 0.58 1.23 0.78 0.63 0.74 0.53 0.71 1.30 2.00 0.32 1.50 1.26 0.65 1.09 1.36 1.11 1.16 0.69

Table 2. Frequency ratio of amino acids classified under respective families (cleaved or uncleaved). The amino acids were classified as follows. Hydrophobic, C, M, F, Y, W, H, A, I, V, L, T ; hydrophylic, S, Q, N, E, D, R, K; turn former, C, S, G, P, N, D; turn breaker, M, F, W, H, A, I, V, L; polar, Y, W, H, S, G, Q, N, E, D, R, K; non-polar, C, M, F, A, I, V, L, T, P; large, M, F, Y, W, H, I, V, L, Q, E, R, K ; small, C, A, T, S, G, P, N, D; branched, I, V, L, T; tiny, A, S, G. cl, cleaved; nc, uncleaved. Amino acid properties

Frequency ratio of amino acids at subsite and position P6 -4

P5 -3

P4 -2

P, -1

p: +I

P; +2

p: +3

P:,

Hydrophobicity Hydrophilichydrophobic (cl) Hydrophilichydrophobic (nc)

0.89 0.84

1.47 0.74

0.82 0.96

0.82 0.47

1.21 0.57

0.52 0.98

2.32 1.06

1.06 0.90

Reverse turns Formerbreaker (cl) Formerbreaker (nc)

1.36 0.99

1.67 0.83

0.64 0.91

1.53 0.81

1.65 1.14

0.75 0.95

3.04 1.10

1.25 1.13

Polarity Polarhon-polar (cl) Polarhon-polar (nc)

1.46 1.22

2.07 1.02

1.05 1.28

1.65 0.76

3.00 1.07

0.87 1.31

2.37 1.54

1.36 1.34

Size SmalUlarge (cl) Smallllarge (nc)

0.95 0.75

0.85 0.80

0.34 0.76

1.26 1.22

0.93 0.78

1.18 0.65

0.95 0.73

0.98 1.02

Other properties Branchedtiny (cl) Branchedtiny (nc)

0.84 1.17

0.78 1.29

1.84 1.02

0.58 0.93

0.19 1.22

0.55 1.50

0.51 0.92

0.50 0.81

The first representation of this analysis was conducted following different classifications of amino acid residues based upon their properties [43]. According to data shown in Tables 2 and 3, the patterns of differences and similarities in the positions flanking the cleaved and uncleaved dibasic residues could be summarized as follows: When the subsites of each sub-population were analyzed on the basis of different amino acid properties (Table 2), the distri-

+4

bution of residues showed large variations at the following positions : hydrophilic residues predominated at position Pi (dibasic cleaved sites) whereas the hydrophobic ones were more represented at position P; (cleaved sites) and in subsites P3 and P: (uncleaved sites) ; the dibasic cleaved sites were characterized by a high proportion of P-turn-forming amino acids in the majority of subsites; the polar amino acids, which were generally represented in the vicinity of both sites, presented however high

710

Rholarn et al. (Eu,:J. Biochem. 227)

Table 3. Frequency ratio of amino acids classified under respective families (cleaved and uncleaved). cl, cleaved; nc, uncleaved. Amino acid properties

Frequency ratio of amino acids at subsite and position P6 -4

P5 -3

P4

p3

-2

-1

P: +1 ~~

p; +2

P; +3

Pi +4

~~~

Hydrophobicity Hydrophilic (cVnc) Hydrophobic (clhc)

0.89 0.84

1.27 0.63

0.88 1.03

1.22 0.69

1.72 0.81

0.66 1.24

1.20 0.55

1.02 0.87

breaker (cVnc)

1.30 0.94

1.27 0.63

0.80 1.14

1.36 0.72

1.03 0.72

1.05 1.34

1.33 0.48

1.10 1.00

Polarity Polar (clhc) Non-polar (cVnc)

1.08 0.90

1.33 0.66

0.91 1.11

1.44 0.67

1.45 0.52

0.82 1.23

1.16 0.75

1.01 0.99

1.14 0.89

1.03 0.97

0.59 1.31

1.01 0.98

1.10 0.92

1.37 0.76

1.16 0.89

0.98 1.02

0.80 1.13

0.60 0.99

0.89 0.50

0.76 1.23

0.21 1.33

0.84 2.27

0.60 1.08

0.52 0.97

Reverse turns Former (cVnc)

Size Small (cVnc)

Large (cl/nc) Other properties Branched (cVnc) Tiny (cUnc)

percentages for the dibasic cleaved sites at positions P,, P: and Pi; the branched residues were significantly less abundant in the cleaved sites and particularly from position P, to P:. When the observed abundance of residues classified on the basis of their properties in the cleaved loci was compared with that obtained for the uncleaved ones (Table 3), it could be observed that the major difference between the two classes of sites appeared to be due to the number of, hydrophobic residues at positions P,, P, and P: and hydrophilic ones at position P:, pturn breaker amino acids at positions P, and Pi, non-polar residues at positions P, P, and Pi, small residues at position PA, branched residues in the majority of subsites and the small sidechain ones at positions P4 and Pi. The second representation of this analysis was achieved by calculating, at each position i, the parameter [R,],/,,for the 20 natural amino acids. Analysis of the resulting numbers (Table 4) allowed some conclusions to be drawn on the processing sites compared to the uncleaved ones. Only a few residues were found to be excluded at certain positions; this was the case of Cys in positions P, and P, as well as Thr and Trp in position P:. Whereas residues Val, Cys and Lys were found generally to be poorly represented in all P and P’ positions, amino acids like Ala, Leu, Ile, Thr and Asn were found only in one or two subsites of the cleaved site sub-population. Some amino acid residues were less represented on the N-terminal side of the dibasic sites but occurred more frequently on the C-terminus at all positions (residue Ser) or in a few number of subsites as for residues Phe, Tyr, Ala, Ser, Asp and Glu. Noticeably, Ala was quite poorly represented on the N-terminal side of the dibasic with highest scores at positions P: and Pi. Residues Gly and Pro, which were represented in the majority of subsites, rated relatively higher scores on the N-terminus than on the C-terminal side of the dibasic sites. A few high scores ([R,],/,,>2) were obtained in all subsites for certain amino acid residues (Table 4). Although thig analysis might potentially provide a basis for discrimination between in vivo cleaved and uncleaved dibasic sites, it is noteworthy that peptide hormone and neuropeptide precursors require a series of post-translational modifications (i.e. proteolysis, glycosylation, amidation, sulfatation etc.) for activation (reviewed in [I, 31). Therefore, since the amino acid residues at all subsites were characterized by a broad range of

properties (Tables 2 -4), the pattern of frequencies observed in positions flanking those dibasic residues might represent global structural requirements for those modifications. Moreover, certain types of amino acids at given positions might also participate in different biological functions. For instance, residue Gly, which scored its highest frequency at position P, (Table 4), was shown to be both implicated in C-terminal amidation of several peptide hormones [3, 441 and in the organization of reverse-turn structures which constitute recognition signals for many biological processes [45-481 including the dbasic cleavage sites of prohormones [27-301. Consequently, the validity of our model was in vitro tested by analysing the possible role of amino acid at P: in proteolytic processing of the common precursor for ocytocin and neurophysin at its LysArg cleavage site [16, 26, 491. Specificity of the Pi amino acid in pro-ocytocin :neurophysin cleavage site. Since dibasic processing most frequently occurs by selective cleavage of the peptide bond linking the basic residue on the C-terminus of the dibasic motif to the next amino acid (P:), the distribution of amino acids obtained at this position was particularly analyzed. Indeed, the ratios displayed in Table 4 allowed us to discriminate between three sub-populations of P: residues. The first one including Trp, Pro, Cys, Met, Gly and the family of branched side chain amino acids; i.e. Val, Leu, Ile and Thr, was very seldom represented (scores ranged from 0 to 0.6). The second one including Gln, Asn, His and Ser represented residues with average scores. The third one corresponded to those residues with highest scores (22), i.e. Phe, Glu, Tyr, Ala and Asp. Although it was not possible to deduce general rules from this broad distibution, consideration of the side-chain structure of excluded amino acids (Tables 2 and 3) indicated that amino acids at position P: shared the common feature of being both unbranched and, generally, polar. Based upon these considerations, the specificity of this P: residue in endoprotease recognition was analyzed by using different [Pro7, Leu151 substrate analogs of pro-ocytocin: neurophysin in which the Ala13 residue (subsite Pi) was replaced by a given residue pertaining to each one of the three previously defined sub-families. Accordingly, 13 [Pro7-Leul5] analogs were synthesized (Table 5 ) and their ability to be cleaved was tested by using two distinct proteases selective for paired basic

711

Rholam et al. (Eul: J. Biochern. 227)

Table 4. Ratio of amino acid frequencies in precursors (cleavedhncleaved). Amino acids

Ratio of amino acid frequencies at subsite and position

Phe

0.30 0.20 1.40 3.40 0.45 0.70 0.55 1.85 1.20 1.10 4.70 0.15 1.85 2.95 1.40 1.40 0.65 0.40 0.90 0.90

TYr TrP His Ala Val Leu Ile Thr Ser Met CYS

GlY Pro Asn Gln ASP Glu Arg LYS

0.80 1.05 0.70 1.65 0.25 0.35 0.70 0.70 0.60 1.30 0.85 0.35 1.85 1.55 0.50 0.65 2.00 1.35 2.45 0.65

3.85 1.45 3.15 0.20 0.25 0.95 1.35 0.65 0.45 0.35 2.60 0.00 1.05 1.95 0.65 1.90 0.35 0.80 1.85 0.65

0.60 1.40 0.20 1.05 0.40 0.95 1 .oo

0.20 0.65 1.25 8.35 0.00 2.35 0.50 0.30 1.65 2.60 1.10 -

amino acid residues. Firstly, the pro-ocytocin: neurophysin metallo-endoprotease [16] which was shown to be able to perform a selective cleavage of the LysArgl2-Alal3 bond within either the hemisynthetic precursor [26] or in a series of synthetic proocytocin :neurophysin peptides [29]. Secondly, the serine protease, product of the KEX2 gene in the yeast, Saccharomyces cerevisiae [9], which represents the prototype of the subtilisin family of processing enzymes [4]. The values of K, and V,,,, (Table 5), obtained for these peptide analogs, allowed us to deduce the following observations. The derivatives bearing one of the P-carbon-branched-sidechain amino acids (Val, Leu, Ile and Thr) at position 13 were poorly recognized by the two enzymes. The occurrence of those excluded amino acids was shown to be very low in keeping with the enzyme behavior (Table 4). The other derivatives, modified at Pi position, were recognized as enzyme substrates and were cleaved with relative efficacies irrespective of their frequency scores (Table 5). Although the relative values of K, and V,, varied over the same range for the two proteases, some specific comparisons indicated however that the behavior of substrates toward the two dibasic endoproteases exhibited certain differences. For example, whereas all peptide substrates were cleaved with similar values of cleavage efficiency by pro-ocytocin :neurophysin convertase, peptides bearing either Ser, Gln, His, Ala or Asp in position 13 were the best substrates hydrolyzed by the Kex2 endoprotease. Comparisons of peptide XXXXV (Aspl3) with peptide XXXXIX (Asnl3) and peptide XXXXXI (Gln13) with peptide XXXXIX (Asn13) revealed significant differences in their kinetic constants. This observation delineates the importance of both the side chain and the charge of Pi residue in the interaction with the Kex2 endoprotease. It is noteworthy that Glu occupies the P; position in the natural substrate of this protease [9]. The peptides bearing nitrogen-containing aromatic side chain residues (Trp and His) were recognized with a good affinity but, the similarity in kinetic parameters for peptides bearing Tyr or Phe in position 13 demonstrated the greater importance of planar aromatic character over polarity in enzyme recognition.

1.95 2.10 0.00 1.50 2.95 0.25 0.15 0.35 0.00 1.70 0.20 0.40 0.55 0.20 1.20 0.80 4.70 2.00 -

1.05 0.65 0.25 1.45 4.05 0.70 1.15 0.25 1.05 1.45 3.15 0.80 2.10 0.70 0.70 0.50 0.55 0.65 0.20 0.50

0.35 0.80 0.15 0.80 0.30 0.65 0.35 0.85 1.05 1.20 1.40 0.65 1.80 2.20 0.75 1.15 1.30 1.40 1.35 0.65

3.25 0.50 3.15 0.45 0.65 0.45 1.05 0.15 0.60 1.30 1.55 0.25 1.20 2.10 0.50 1.05 1.60 1.55 0.35 0.65

Since the tested amino acids do not represent a homologous series, we attempted to correlate the steady-state kinetic constants of peptide substrates with some parameters which describe the physical properties of amino acids which replaced the natural Pi residue (Alal3) of pro-ocytocin :neurophysin. Using different reported scales of hydrophobicity of amino acids [50],we observed significant correlations with lnVmaxof peptide XXV and its analogs ( r = 0.5 -0.75) in the case of cleavage by Kex2 endoprotease. Moreover, an apparent correlation was also observed between K,,, or V, and side-chain volume of Pi residues ( r = 0.45). All amino acids were not tested, consequently the role of this parameter in substrate-enzyme interactions cannot be excluded. When similar plots were drawn for pro-ocytocin :neurophysin convertase, no apparent relationship was observed between the kinetic constants of modified substates and the physical properties of residues at subsite P:. We also assessed whether the kinetic parameters, V,, and K,, might be themselves correlated. Indeed, whereas a poor correlation was observed for Kex2 protease, the plot of InV,,,,, versus lnK, was found to be linear for pro-ocytocin: neurophysin convertase (Y = 0.92). This observation reflects the dependence of both substrate binding and catalysis on properties of P: residue side chains in the case of pro-ocytocin :neurophysin convertase. Consequently, only in the case of Kex2 protease was a correlation observed between the values of lnV,,,,IK, and both InV,,,, ( r = 0.80) and In& ( r = 0.65). Since the S: subsite of the two dibasic endoproteases can accommodate a number of amino acid residues with different side chains, the potential role of other subsites in endoprotease recognition was also tested. Among the substrates cleaved by the two proteases, the substrate with Gly at P: position was recognized with the lower affinity by both enzymes (if K,,, = K,,). Therefore, the [Pro7, Leu151 analogs with Gly at P; and P: positions were synthesized and tested. The data, displayed in Table 5 , revealed the following. Like the natural substrate (peptide XXV), those modified peptides were recognized and cleaved by the pro-ocytocin :neurophysin convertase; for the Kex2 endoprotease, both modified peptides were in contrast characterized by a low affinity and a low efficiency. In particular, substitution of

Table 5. Kinetic parameters for pro-ocytocin:neurophysin-related-peptidecleavage. n. d., cleavage not detectable. Peptide substrate

Sequence of peptide substrates

XXXXVI XXXXVII XXXXVIII XXXXIX XXXXX XXXXXI XXXXXII XXXXXIII XXXXXIV XXXXXV XXXXXVI

P-L-G-G-K-R-A-V-L P-L-G-G-K-R-W-V-L P-L-G-G-K-R-IJ-V-L P-L-G-G-K-R-EJ-V-L P-L-G-G-K-R-Y-V-L P-L-G-G-K-R-F-V-L P-L-G-G-K-R-N-V-L P-L-G-G-K-R-S-V-L P-L-G-G-K-R-Q-V-L P-L-G-G-K-R-G-V-L P-L-G-G-K-R-I-V-L P-L-G-G-K-R-y-V-L P-L-G-C-K-R-L-V-L P-L-G-G-K-R-1-V-L

XXXXXII XXXXXVII XXXXXVIII

P-L-G-G-K-R-G-V-L P-L-G-G-K-R-A-G-L P-L-G-G-K-R-A-V-G

XXV XXXXIV

xxxxv

Pro-ocytocin: neurophysin convertase

Kex2 endoprotease

K,

V,,

Vm,JKm

K,

V,,

VmaJKm

PM 150 125 300 305 505 625 655 665 760 920

pmoVh

pmol . h-'

PM

pmolk

pmol . h-' . pM-'

n.d. n.d. n.d.

720 275 805 1130 2545 1835 2185 1620 2845 2500 n.d. n.d. n.d. n.d.

4.8 2.2 2.7 3.7 5.0 2.9 3.3 2.4 3.7 2.7 n.d. n.d. n.d. n.d.

n.d. n.d. n.d. n.d.

45 000 12 000 40 000 40 000 8 000 7 000 22 000 85 000 28 000 10 000 n.d. n.d. n.d. n.d.

161 20 156 114 76 46 47 303 193 17 n.d. n.d. n.d. n.d.

2.95 0.00 4.70 1.50 2.10 1.95 1.20 1.70 0.80 0.55 0.35 0.25 0.15 0.00

920 280 380

2500 1295 1375

2.7 4.5 3.5

590 1425 1470

10 000 5 000 20 000

17 3 14

0.55 2.10 13 0

n.d.

. pM-'

280 460 256 230 105 152 470 280 145 590

Occurence of amino acid residues

h

h TL

Rholam et al. (Eur J. Biochem. 227)

Val by Gly at P; position caused a 50-fold decrease in cleavage efficiency of peptide (Table 5).

DISCUSSION One of the most important post-translational processing events involved in the biosynthesis of peptide hormones and neuropeptides is the proteolytic cleavage of their precursors at basic amino acid residues [I, 3, 51. Arranged frequently as doublets, these moieties were predicted 17, 24, 311, and shown [27, 28, 331 to be situated in the immediate vicinity of reverseturn structures. Both the integrity [16, 511 and the accessibility [28-30, 32, 331 of basic doublets were also shown to play a critical role in these processes. However, a series of observations using site-directed mutagenesis of various prohormones had underlined the particular importance of other amino acid residues around the scissile bond in processing [25, 30, 34, 36-39]. Therefore, the aim of the present studies was an attempt to define other parameters which might also participate in the specificity of processing endoproteases. Whereas many types of analysis were applied to the amino acid sequences surrounding 352 potential dibasic endoprotease cleavage sites, no strict conservation of specific residues could be observed around those paired-amino-acid processing sites. However, some amino acids were preferentially represented (Pro or Gly) or not found (Ile or Cys) at certain subsites. In vitro analysis of the correlation which might exit between the occurrence of amino acid residues at P: position and the potential biological relevance of these residues in OT/Np precursor processing [I 61 confirmed this tendency. Indeed, the kinetic data, obtained with two different dibasic endoproteases, did not allow us to establish a simple relationship between the frequency of occurrence for each P: residue and prohormone proteolysis. Despite this lack of strict specificity for a given amino acid residue in P:, this in vitro study revealed however some interesting features relative to the specificity of these endoproteases. Firstly, both endoproteases did not tolerate, at their subsite S:, the p-carbon branched amino acid residues in accord with the low occurrence frequencies observed in the database for these excluded amino acids. Similar results were obtained by site-directed mutagenesis on some of these excluded amino acids occupying the Pi position at the cleavage sites of pro-albumin [35], pro-renin [37] and pro-factor IX [39] in different cell systems. Secondly, the other [Pro’l, Leu151 analogs, which were best hydrolyzed by the Kex2 enzyme, were cleaved by both dibasic proteases and comparable kinetic values were obtained. Taken together these observations suggested that these precursor-processing endoproteases probably share a common mechanism. They may also explain the fact that proteases homologous to the KEX2 gene product 111-12, 52-54] as well as endoproteases related to other enzyme families [20, 22, 231 such as pro-ocytocin:neurophysin convertase [I@, could cleave a wide variety of precursors [55 -601 or peptide substrates mimicking different prohormone cleavage sites [29]. Thirdly, although all naturally occuring amino acid residues were not tested at position Pi, the kinetic data delineated substantial differences in properties of those dibasic endoproteases. In the case of pro-ocytocin :neurophysin convertase, it can be observed that no relationship was seen between the kinetic constants obtained for modified substrates and the physical properties of residues at subsite Pi and, those kinetic parameters were themselves correlated. Since the specificity of proteases involves consideration of substrates with respect to their amino acid sequences and the nature of their folding and accessibility, these observations might reflect the fact that pro-ocytocin :neurophysin convertase could cleave all

71 3

substrates, given that amino acids surrounding the basic doublets could be accommodated in its binding pocket. The behavior of [Pro7, Leu151 peptides, modified at P: and P: positions (this report) or those in which the sequence Pro7-Leu-Gly-GlylO (subsites P, to P4) was mutated [29], support this hypothesis. In contrast, the kinetic data obtained with the Kex2 endoprotease, indicated that in this case some structural constraints were involved in substrate-protease interactions. Based on these considerations, the pattern of frequencies observed in the positions flanking the dibasic residues could be interpreted as follows. Firstly, since the amino acid sequences flanking the dibasic residues are characterized by a large variability in both their nature and their composition, this supports the concept that the specificity of processing proteases is not solely governed by preferential amino acid residues but dictated by both the stereochemistry and flexibility of polypeptide substrates [7,28-30,61, 621. Secondly, since no major contribution of a particular subsite to the endoprotease specificity was observed, this suggested that recognition of prohormone substrates by their endoproteases probably occurred through multiple anchoring on both sides of the scissile bond instead of a few strong interactions. This implies that changes in one, or in a few, binding subsites of the precursors may only alter but not necessarily abolish the endoprotease activity [25, 26, 29, 30, 391. In summary, these data underline the overall value of the calculated occurrence frequencies of amino acids to provide a database for further investigation of the structural requirements for proteolytic-site recognition by processing enzymes. Although an heuristic pattern could not emerge from the available data, they provided a workable basis for the definition of the specificity of various proteolytic enzyme families, such as those related to the KEX2 and furin genes [ l l , 12, 52-54]. Therefore, future studies could be conducted on those enzymes in order to identify additional P or P’ sites which show relationships of kinetic constants versus side-chain properties of amino acid residues. With this information, coupled to calculated frequencies of amino acids at different subsites, it might be possible to predict the effects of all permutations of multiple mutations within the substrates with respect to their sensitivity to endoprotease action. This work was supported in part by funds from the Minisdre de I’Enseignernent Supkrieur et de la Recherche (contrat Biennal from the Direction d la Recherche et aux Etudes Doctorales), the Centre National de la Recherche Scient$que (URA 1682) by a doctoral fellowship to N. B. from the Association Frangaise contre les Myopathies and by a postdoctoral fellowship to H. B. from Rhone-Poulenc-Rorcr (Programme BioAsenir). We thank Drs Guy HervB and John Vickrey for critical reading of the manuscript.

REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9.

10. 11.

Cohen, P. (1988) Biochimie 69, 87-89. Sossin, W., Fischer, J. & Scheller, R. (1989) Neuron 2, 1407-1417. Darby, N. & Smyth, D. (1990) Biosci. Rep. 10, 1-13 Steiner, D., Smeekens, S., Ohagi, S. & Chan, S. J. (1992) J. Bid. Chern. 267, 23435-23438. Bourdais, J. & Cohen, P. (1991) Ann. Endocrinol. 52, 339-347. Devi, L. (1991) FEBS Lett. 280, 189-194. Rholam, M., Nicolas, P. & Cohen, P. (1986) FEBS Lett. 207, 1-6. Docherty, K. & Steiner, D. (1982) Annu. Rev. Physiol. 44, 625638. Fuller, R., Brake, A. & Thorner, J. (1989) Proc. Natl Acad. Sci. USA 86, 1434-1438. Wise, R., Ban; P., l e f e r , M., Brake, A. & Kaufman, R. (1990) Proc. Nut1 Acad. Sci. USA 87, 9378-9382. Zollinger, L., Racine, C., Crine, P., Boileau, G., Germain, D., Thomas, D. & Gossard, F. (1990) Biochem. Cell. Biol. 68, 635640.

714

Rholam et al. (Eur J. Biochem. 227)

12. Seidah, N., Gaspar, L., Mion, P., Marcinkiewicz, M., Mbikay, M. & Chrktien, M. (1990) DNA Cell Biol. 9, 415-424. 13. Thomas, L., Leduc, R., Thome, B., Smeekens, S., Steiner, D. & Thomas, G. (1991) Proc. Natl Acad. Sci. USA 88, 5291-5301. 14. Wasley, L., Rehemtulla, A,, Bristol, J. & Kaufman, R. (1993) J. Biol. Chem. ;!68, 8458-8465. 15. Gluschankof, P., Gomez, S., Morel, A. & Cohen, P. (1987) J. Biol. Chem. 262, 9615-9620. 16. Plevrakis, I., Clamagirand, C., Crkminon, C., Brakch, N., Rholam, M. & Cohen, P. (1989) Biochenzistiy 28, 2705-2710. 17. Bourdais, J., Pierotti, A,, Boussetta, H., Barre, N., Devilliers, G. & Cohen, P. (1991) J. Biol. Chem. 266, 23386-23391. 18. Resnick, N., M,aloy, W., Guy, R. & Zasloff, M. (1991) Cell 66, 541-554. 19. Kawabata, S. &. Davie, E. (1992) J. Biol. Chem. 267, 1033110336. 20. Blache, P., Kervran, A,, Dufour, M., Martinez, J., Le-Nguyen, D., Lotersztajn, S., Pavoine, C., Pecker, F. & Bataille, D. (1990) J. Biol. Chem. 2155,21 514-21 519. 21. Rhodes, Ch., Thome, B., Lincoln, B., Nielsen, E., Hutton, J. & Thomas, G. (1993) J. Biol. Chem. 268, 4267-4275. 22. Azaryan, A,, Wong, M., Friedman, T., Crawley, N., Estivariz, F., Chen, H. & Peng Loh, Y. (1993) J. Biol. Chem. 268, 1196811975. 23. Chesneau, V., Pierotti, A,, BarrC, N., CrCminon, C., Tougart, C. & Cohen, P. (1994) J. Biol. Chem. 269, 2056-2061. 24. Geisow, M. & Smyth, D. (1980) in The enzymology of post-translational mod@cation of proteins (Freedman, R. B. & Hawkm, H. C., eds) vol. 1: pp. 259-287, Academic Press, London. 25. Gomez, S., Boileau, G., Zollinger, L., Nault, C., Rholam, M. & Cohen, P. (1989) EMBO J. 8, 2911 -2916. 26. Brakch, N., Boussetta, H., Rholam, M. & Cohen, P. (1989) J. Biol. Chem. 264, 15912-15916. 27. Rholam, M., Cohen, P., Brakch, N., Scatturin, A., Paolillo, L. & Di Bello, C. (1990) Biochem. Biophys. Res. Commun. 168, 10661073. 28. Paolillo, L., Simonetti, M., Brakch, N., D’Auria, G., Saviano, M., Dettin, M., Rholam, M., Scatturin, A,, Di Bello, C. & Cohen, P. (1992) EMBO J. 7, 2399-2405. 29. Brakch, N., Rholam, M., Boussetta, H. & Cohen, P. (1993) Biochemistry 32, 4925-4930. 30. Brakch, N., Boileau, G., Simonetti, M., Nault, C., Joseph-Bravo, P., Rholam, M. & Cohen, P. (1993) EUKJ. Biochem. 216, 39-47. 31. Bek, E. & Berry, R. (1990) Biochemistry 29, 178-183. 32. Weiss, M., Frank, B., Khait, I., Pekar, A., Heiney, R., Shoelson, S. & Neuringer, L. (1990) Biochemistry 29, 8389-8341. 33. Rayne, R. & O’Shea, M. (1993) EUKJ. Biochem. 217, 905-911. 34. Brakch, N., Rholam, M., Nault, C., Boileau, G. & Cohen, P. (1991) FEBS Lett. 282, 363-367. 35. Oda, K., Misumi, Y., Sohda, M., Takami, N., Sakaki, Y. & Ikehara, Y. (1991) Biochem. Biophys. Res. Commun. 175, 690-696.

36. Oda, K., Ikeda, M, Tsuji, E., Sohda, M., Takami, N., Misumi, Y. & Ikehara, Y. (1991) Biochem. Biophys. Res. Commun. 179,11811186. 37. Nagahama, M., Nakayama, K. & Murakami, K. (1991) Eul: J. Biochem. 197, 135-140. 38. Mark, M., Lokker, N., Zioncheck, T., Luis, E. & Godowski, P. (1992) J. Biol. Chem. 267,26166-26171. 39. Bristol, J., Furie, B. & Furie, B. (1993) J. Biol. Chem. 268, 75777584. 40. Memfield, R. (1963) J. Am. Chem. Soc. 85,2149-2154. 41. Nicolas, P., Delfour, A., Boussetta, H., Morel, A,, Rholam, M. & Cohen, P. (1986) Biochem. Biophys. Rex Commun. 140, 565573. 42. Germain, D., Dumas, F., Vemet, T., Bourbonnais, Y., Thomas, D. & Boileau, G. (1992) FEBS Lett. 299, 283-286. 43. King, R. & Sternberg, M. (1990) J. Mol. Biol. 216, 441-457. 44. Eipper, B. & Mains, R. (1990) Trends Endocrinol. Metab. I , 388394. 45. Ball, J. & Alewood, P. (1990) J. Mol. Recognit. 3, 55-64. 46. Collawn, J., Strangel, M., Kuhn, L., Esekogwu, V., Jing, S., Trowbridge, I. & Tainer, J. (1990) Cell 63, 1061-1072. 47. Eberle, W., Sander, C., Klaus, W., Schmidt, B., von Figura, K. & Peters, C. (1991) Cell 67, 1203-1209. 48. Laczko, I., Hollosi, M., Urge, L., Ugen, K., Weiner, D., Mantsch, H., Thurin, J. & Otvos, L. (1992) Biochemistry 31, 4282-4288. 49. Camier, M., Benveniste, D., Barre, N., Brakch, N. & Cohen, P. (1991) Mol. Cell. Endocrinol. 77, 141-147. 50. Comette, J., Cease, K., Margalit, H., Spouge, J., Berzofsky, J. & DeLisi, C. (1987) J. Mol. Biol. 195, 659-685. 51. Mc Cune, J., Rabin, L., Feinberg, M., Lieberman, M., Kosek, J., Reyes, G. & Weissman, 1. (1988) Cell 53, 55-67. 52. Brenner, C. & Fuller, R. (1992) Proc. Nut1 Acad. Sci. USA 89,922926. 53. Hatsuzawa, K., Nagahama, M., Tanahashi, S., Takada, K., Murakami, K. & Nakayama, K. (1992) J. Biol. Chem. 265, 22075-22078. 54. Zhou, Y. & Linberg, I. (1993) J. Biol. Chem. 268, 5615-5623. 55. Thomas, L., Leduc, R., Thome, B., Smeekens, S., Steiner, D. & Thomas, G. (1991) Proc. Natl Acad. Sci. USA 88, 5291-5301. 56. Nakayama, K., Hosaka, M., Hatsuzawa, K. & Murakami, K. (1991) J. Biochem. (Tokyo) 109, 803-806. 57. Hallenberger, S., Bosch, V., Angliker, H., Shaw, E., Klenk, H.-D. & Garten, W. (1992) Nature 360, 358-361. 58. Wulff, B., Johansen, T., Dalboge, H., O’Hare, M. & Schwartz, T. (1993) J. Bid. Chem. 268, 13327-13335. 59. Breslin, M., Linberg, I., Benjannet, S., Mathis, J., Lazure, CI. & Seidah, N. (1993) J. Biol. Chem. 268, 27084-27093. 60. Nakashima, H., Nishikawa, K. & Ooi, T. (1986) J. Biochem. (Tokyo) 99, 153-162. 61. Graf, L. & Hoolosi, M. (1980) Biochem. Biophys. Res. Commun. 93, 1089-1093. 62. Fontana, A., Fassina, G., Vita, C., Dalzoppo, D., Zamai, M. & Zombonin, M. (1986) Biochemistry 25,1847-1851.

Supplernentuly Material. Role of amino acid sequences flanking dibasic cleavage sites in precursor proteolytic processing. The importance of the first residue C-terminal of the cleavage site, the P: subsite. Table S1. Amino acid sequences around basic doublets in hormone and peptide precursors. This information is available, upon request, from the Editorial Office. A total of 17 pages is available.