Electrophoresis 1996, 17, 932-937

Alessandra Bossi' Pier Giorgio Righetti' Carlo Visco' Umberto Breme' Maurizio Mauriello' Barbara Valsasina' Gaetano Orsini' Elisabeth Wenisch"

'University of Milano, L.I.T.A., Segrate, Italy 'Pharmacia & Upjohn, Nerviano, Italy

Investigation on minor degraded derivatives of the recombinant hirudin variant HM2 from Hirudinaria manillensis isolated by isoelectric focusing in multicompartment electrolyzers On isoelectric focusing in immobilized pH gradients (IPG) a preparation of recombinant hirudin from Hirudinaria manillensis, purified to homogeneity, was found to still contain a total of 5% minor components: three with higher p l values (pls 4.10, 4.25 and 4.31), one with a lower p l value (pZ3.98) as com- pared with the main form (pl 4.03). Multicompartment electrolyzers with iso- electric membranes and micropreparative IPG gel slabs allowed the recovery of pure fractions of such minor components, which were further characterized by electrospray mass spectra, limited proteolysis, and sequence analysis. All four minor isoforms were found to be cleavage products of the parent, full- length hirudin molecule (molecular mass 6797 Da), as follows: the p l 4.31 (5032 Da) had lost sixteen amino acids from the N-terminus, the pl4.25 (6212 Da) lacked five amino acids from the C-terminus, the p l 4.10 (2980 Da) was a cleavage product at residue Cys", and the pZ3.98 (6610 Da) lacked the dipep- tide Val-Ser at the N-terminus. Combining the extreme resolving power of IPGs with the high accuracy of mass spectra was found to be an attractive stra- tegy in decoding post-synthetic modifications often encountered in r-DNA pro- teins.

1 Introduction

Hirudin, a polypeptide first isolated from salivary glands of the blood sucking leech Hirudo medicinalis 111, is the most potent and selective inhibitor of the serine pro- tease thrombin. In blood circulation, thrombin inhibi- tion prevents the thrombin-catalyzed cleavage of fibrin- ogen to fibrin and the subsequent fibrin clot formation. By now, more than twenty hirudin isoforms, showing a high degree of similarity, have been isolated from leech- es belonging to the Hirudinidae family [2]. Hirudin mole- cules, consisting of a 63-66 amino acid polypeptide chain with six cysteine residues, forming three intramolecular disulfide bridges [3], can be divided into two regions: an N-terminal globular domain tightly packed by the three disulfide bonds and a disordered C-terminal tail with a high content of acidic amino acid residues. The resolu- tion of the crystal stucture of thrombin/hirudin com- plexes showed a bivalent binding mode, with the hirudin N-terminal core bound near to the active site of thrombin and the negatively charged C-terminal tail interacting with the fibrinogen binding site of thrombin [4].

The availability of large amounts of hirudins produced by recombinant DNA technologies [5-81 made it pos- sible to investigate the effect of thrombin inhibition in coagulation disorders and the potential clinical applica- tion of these molecules under different pathological con- ditions, for example, in the treatment of arterial and venous thrombosis [9, 101. Recombinant proteins for therapeutic applications require an extensive characteri- zation of purity and the identification of protein micro- heterogeneities such as deamidation, oxidation, or partial

Correspondence: Dr. Gaetano Orsini, Pharmacia & Upjohn, via Gio- vanni XXIII, 1-20014 Nerviano, Italy (Tel: +39-331-583806; Fax: +39-331-58375s)

Keywords: Isoelectric focusing / Immobilized pH gradients / Hirudin / Proteolysis I Mass spectra

proteolysis, which can occur during the biosynthesis within the producing cells or during the downstream purification process [l l] . We studied a problem of pro- tein microheterogeneity during the production in Esche- richia coli of a novel recombinant hirudin variant, denominated HM2, from the Asian leech Hirudinaria manillensis [5]. As reported in this paper, the application of analytical and preparative immobilized pH gradient techniques made it possible to identify and isolate a number of minor degraded derivatives of HM2, which were then characterized by a combination of column chromatography, mass spectrometry, sequence analysis and peptide mapping.

2 Materials and methods

2.1 Materials

Recombinant hirudin variant HM2 from Hirudinaria manillensis was biosynthesized in Escherichia coli by ex- pressing a periplasmic construction containing the HM2 synthetic gene and purified as previously reported [5].

N-Tosyl-L-phenylalanine chloromethyl ketone (TPCK)- treated trypsin was from Fluka (Buchs, Switzerland), tri- fluoroacetic acid (TFA) was from ABI (Foster City, CA, USA). Other chemicals were reagent grade from Carlo Erba (Milan, Italy). Horse myoglobin and Ponceau S were from Sigma (St. Louis, MO, USA). Acrylamide, N,N-methylenebisacrylamide, TEMED and persulfate were from Bio-Rad (Hercules, CA, USA). Micropure fil- tering devices (0.22 pm pore size) were from Amicon (Beverly, MA, USA). The Immobiline chemicals, with pK values of 3.1, 4.6, 6.2 and 9.3, were from Pharmacia Bio- tech (Uppsala, Sweden). Immobilon P blotting mem- branes were from Millipore (Waltham, MA, USA).

* On leave of absence from the University of Forestry and Agriculture, Vienna, Austria

0 VCH Verlagsgesellschaft mbH, 69451 Weinheim, 1996 0173-0835/96/0505-0932 $10.00+.25/0

Electrophoresis 1995, 17, 932-937 Characterization of hirudin derivatives 933

2.2 RP-HPLC

RP-HPLC was performed on a Hewlett-Packard 1090M liquid chromatographic apparatus with a Vydac 218TP C18 column, 4.5 X 250 mm (The Separation Group, Hes- peria, CA, USA). Mobile phases A and B were H,O and acetonitrile, containing 0.1 and 0.078 O/o TFA, respectively. A linear gradient from 18 to 26% of phase B was applied in 46 min at a flow rate of 0.45 mL/min. The elution pro- file was monitored with a Hewlett-Packard 1040A diode array detector at the wavelength of 215 nm.

2.3 Electrospray mass spectrometry

Molecular mass measurements were performed on an HP 5989 A MS-Engine single quadrupole instrument equipped with an HP 59987 A electrospray interface (Hewlett Packard, Wilmington, DE, USA). Samples, as recovered from the multicompartmen-t electrolyzer, or from micropreparative IPG gels, were diluted with 50% methyl alcohol-H,O containing 1% acetic acid and in- jected into the ion source at a flow rate of 2-5 pL/min. The electrospray potential was approximately 6 KV; the quadrupole mass analyzer was set to scan over mass-to- charge ratios (m/z) from 1000 to 1700, at 2 s per scan, for a total time of 10-12 s. The sum of data acquired over this time constituted the final spectrum. Molecular masses were calculated from several multiply-charged ions within a coherent series. Calibrations were per- formed with horse skeletal muscle myoglobin.

2.4 Limited proteolysis

Limited trypsin proteolysis was carried out as previously reported [12]. Briefly, protein samples in 50 mM ammo- nium bicarbonate, pH 7.8, buffer were incubated with trypsin at an enzyme/substrate ratio of approximately 1:20 by mass, at room temperature. Proteolysis was stopped after 5 min by acidification with TFA and pep- tide fragments were purified by micropreparative RP- HPLC on a Vydac C-18 column (250 X 4.6 mm) eluted with a linear gradient from 5 to 75% mobile phase B, where phase A was 0.1% TFA in water and phase B was acetonitrile-water-TFA (95:5:0.07).

2.5 Sequence analysis

N-terminal sequence analysis was performed by auto- mated Edman degradation using a pulsed liquid-phase sequencer model 477 A with an online analyzer model 120 A (Applied Biosystems, Foster City, CA, USA) for the detection of phenylthiohydantion derivatives of amino acids. Standard manufacturer’s procedures and programs were used with minor modifications.

2.6 IPG

IEF in IPG was performed in 5%T, 4%C matrices in the pH 3.6-4.5 and pH 3.5-4.3 intervals. The recipes for these IPG ranges were from Righetti [13]. The gels, after polymerization, were washed, dried and reswollen in 20% glycerol. Run: 2 h at 400 V, followed by 12 h at

2000 V, 10°C. Since hirudin is only poorly stained with Coomassie Brilliant Blue R-250, at the end of the run the gels were blotted onto PVDF membranes, followed by staining with Ponceau S. Hirudin and its minor iso- forms appeared as intense red/pinkish zones on a white background. The membranes had to be photographed within a few hours, preferably wet (in the reflectance mode), because the color tended to fade with time.

2.7 Preparative purification in multicompartment electrolyzers

Purification of hirudin from its minor isoforms was car- ried out with the IsoPRIME apparatus (Hoefer, San Francisco, CA, USA), operated according to Righetti et al. [14, 151. This equipment consists of a multichamber electrolyzer to be assembled with isoelectfic buffering membranes. Four isoelectric membranes were prepared with plvalues 3.0, 4.0, 4.19, and 5.0. The membrane com- position is reported in Table 1. Calculation of the amounts of buffering and titrant ions was performed with the program of Giaffreda et al. [I61 (available from Hoefer). All membranes were cast as a 1O0/oT, 4%c matrix in the form of disks of 4.7 cm diameter and a thickness of ca. 1 mm. Note that the membranes are sup- ported by glass fiber filters (see [I41 for a detailed de- scription of their properties). After washing the mem- branes three times in distilled water, the multicompart- ment apparatus was assembled and 7 mL (10 mg) total of hirudin were loaded (in a single chamber, at the anodic side). Only the contents from the anolyte and catholyte chambers were recycled from 200 mL reser- voirs. No reservoirs were connected to the other electro- lyzer chambers where the various isoforms were sup- posed to collect, since the amount of protein available to be processed was minute. Purification of the minor iso- forms of hirudin was continued overnight at 600 V. The anolyte was 10 mM acetic acid (pH 2.88; conductivity, 85.5 pmhos) and the catholyte 50 mM isoelectric His (pH 7.47; conductivity, 67.4 pmhos). No circulating coolant was utilized and joule heat was dissipated in the cold room (7°C). Under the above conditions, the tem- perature rise of the liquid in the electrolyzer, at steady- state, was only 1°C.

Table 1. Composition of the isoelectric membranes used in the IsoPRIME apparatus

Immobiline Membrane Membrane Membrane Membrane DI 5.00 DI 4.19 DI 4.00 DI 3.00

~~~

pK 3.6 10.8 mM 11.1 mM 11.2 mM 11.6 mM

pK 10.3 20 mM 12.2 mM 10.4 mM 0.8 mM pK 4.6 11.8 mM 9.8 mM 9.4 mM 6.9 mM

3 Results

Recombinant hirudin variant HM2 was expressed in Escherichia coli as a primary translational product con- sisting of an OmpA leader peptide-HM2 fusion protein, which was exported into the bacterial periplasmic com- partment and processed by endogenous leader peptidase I [17], to release the mature form of HM2 of 64 amino acid residues with three disulfide bonds (Fig. 1). A simple purification protocol with two column chromato-

934 A Bossi rr a1

graphies gave a preparation of HM2 more than 95% homogeneous, as demonstrated by RP-HPLC [ 5 ] . Anal- ytical IPG-IEF on slab gels (Fig. 2B, lane I), showed the major band of HM2 at p l 4.03 along with four minor bands migrating as both more basic (PI 4.10, 4.25 and 4.31) and more acidic proteins (pl 3.98). Separation of HM2 by preparative IPG-IEF, carried out in a multicom- partment electrolyzer, assembled with defined mem- branes having PIS of 3.0, 4.0, 4.19 and 5.0, enabled us to isolate, in a single fraction, the two more basic forms (Fig. 2B, lane 4), which were then separated by RP-HPLC. The net charge differences among HM2 and the other contaminants were too small to allow for an efficient separation in the electrolyzer; however, micro- preparative IPG-IEF carried out on slab gels in the p l range between 3.5 and 4.3 clearly separated HM2 from the closely migrating contaminants (Fig. 2A). Gel por- tions corresponding to each protein band were scraped away from the supporting plastic film and the proteins were eluted with an acetonitrile-water mixture by using a 0.22 pm filtering device [18]. The isolation of the four contaminant forms allowed their identification by a com- bination of RP-HPLC, mass spectrometry, peptide map- ping, and limited proteolysis experiments (Table 2).

The major contaminant was the more acidic compound e (PI 3.98), which represented approximately 3% of the main protein HM2 while each of the other three impuri- ties were present at concentrations of 0.5% or lower (Fig. 3, upper panel). Amino acids identified in the first cycles

1 15 Val Ser Tyr Thr Asp CJs Thr Glu Ser Gly Gln Asn Tyr Cbs Leu

16 30 Cks Val Gly Ser Asn Val Cys Gly Glu Gly Lys Asn Cbs Gln Leu

45

60

31 a Ser Ser Ser Gly Asn Gln Cys Val His Gly Glu Gly Thr Pro Lys

46 Pro Lys Ser Gln Thr Glu Gly Asp Phe Glu Glu Ile Pro Asp Glu

61 64 b Asp Ile Leu Asn

A

A C

i A

A

a: fragment 17-64; b: fragment 1-59; c: fragment 38-64; e: fragment 3-64.

Figure 1. Primary sequence of HM2. Recombinant hirudin variant HM2 is constituted by 64 amino acid polypeptidic chain with disulfide bonds between C ~ s ~ - C y s ' ~ , Cys'6~cy"* and C y ~ * ~ - C y s ~ ~ . The posi- tions of proteolytic cleavages are indicated with black arrows and the resulting fragments are reported under the sequence.

PI A

Electrophoresis 1996, 17, 932-937

B

I 2 3 4 Figure 2. IEF on IPG of recombinant hirudin variant HM2. Analytical separation of HM2 ( p l 4.03) in the p l range 3.60-4.50, showed four minor contaminants (B, lane 1). Control of preparative separation on the multicompartment electrolyzer (B, lanes 2, 3 and 4, respectively, corresponding to chambers 1, 2 and 3) allowed us to separate the two more basic compounds a (~24.31) and b (pl4.2.5) while microprepara- tive slab gel, in the p l range 3.50-4.30, made it possible to separate the other two contaminants, c (pf 4.10) and e (pf 3.98).

of N-terminal sequence analysis of compound e were as follows: Tyr3-Thr-Asp-, indicating the loss of the first two residues from the HM2 molecule while molecular mass measurement of compound e gave a relative molecular mass of 6610 Da, corresponding to the calculated value of a polypeptidic chain from residue 3 up to the last C-terminal asparagine residue at position 64 of HM2. N-terminal sequence analysis of compound a (PI 4.31) started with residues Val'7-Gly-Ser-Asn-Val-, Val being in position 17 of the HM2 chain. Trypsin-specific cleavage at position 47 [12], followed by RP-HPLC separation of the two resulting peptides, allowed us to collect the C-terminal fragment whose sequence was found to exactly correspond to position 48-64. Finally, mass measurements of compound a gave a value of 5032 Da, corresponding to the calculated mass of the peptide chain 17-64 of HM2.

Compound b ( P I 4.25) had the correct sequence starting with the N-terminal residues of HM2: Val'-Ser-Tyr-Asp-; however, the experimental mass value was 6212 Da (585 Da less than HM2 mass value) indicating the cleavage of

Table 2. Characterization of recombinant HM2 and of its proteolytically degraded derivatives

Compound Net charge RP-HPLC retention time ES-MSQ PI min Da

Full length HM2 (1-64) 4.03 37.05 Compound a (17-64) 4.31 48.09 Compound b (1-59) 4.25 22.42 Compound c (38-64) 4.10 30.84 Compound e (3-64) 3.98 35.44

6797 5032 6212 2980 6610

a) The reported experimental relative molecular mass values were within t l Da of the calculated mass values.

Electrophoresis 1995, 17, 932-931

b c

Characterization of hirudin derivatives 935

a

r; 60

m Q) 40 0 c m

b

m Q) 40 0 E m

0 v)

20

0

i 0 i0 i0 40 i 0 i 0

Time (min)

. I

Ill I a b e

C

10 20 30 40 50 60

Time (min)

Figure 3. Analytical RP-HPLC of hirudin variant HM2 and its degraded derivatives. Elution profile of an HM2 preparation (upper panel) showed the main peak of HM2 (peak d, approximately 95% of total peak areas) and its degraded contaminants (peaks b, c, e and a). The degraded derivatives, after separation by isolectric focusing on immobilized pH gradients, were analyzed by RP-HPLC as shown in the lower panel.

the last five carboxy-terminal residues. Selective trypsin cleavage at position 47 followed by complete sequencing of the resulting C-terminal fragment gave the sequence from residue 48 to 59, confirming that compound b cor- responded to peptide chain 1-59 of HM2 (Fig. 4). The N-terminal sequence of compound c (PI 4.10) was Val3*- His-Gly-Gln-, Val being at position 38 of the HM2 chain. ES-MS analysis gave a value of 2980 Da, corresponding to peptide chain 38-64 of HM2.

4 Discussion

4.1 On the postsynthetic fate of recombinant hirudin

Heterologous recombinant proteins produced in Esche- richia coli can undergo intracellular proteolysis by action of cytoplasmic proteinases. A common way to avoid this degradation is to insert the sequence encoding the pro- tein of interest behind a signal sequence which makes the translated construct competent for secretion into the bacterial periplasmic space. These signal or leader sequences are able to translocate the protein precursor through the inner membrane and afterwards they are cleaved off by membrane-bound leader peptidases [ 191. The strategy of periplasmic export was applied to pro- duce, in Escherichia coli, the novel hirudin variant HM2 from Hirudinaria manillensis. High levels of HM2 were

64

E S 0 N N

t8 8 32 r: t8

0 u)

a

e 2

0

1 4 7 L 147

L

48-64

_.i I

5 15 35

Time (min) Figure 4. Trypsin limited proteolysis of HM2 and compound b. Selec- tive trypsin cleavage of HM2 (profile B) gave fragments 1-47 and 48-64 while selective cleavage of compound b gave the same frag- ment 1-47 and the new fragment 48-59. The composition of peptidic fragments was confirmed by N-terminal sequence analysis.

achieved using an expressionlsecretion vector containing an oligonucleotide sequence coding for OmpA leader peptide [20] fused to a synthetic gene coding for HM2 [S]. The final preparation of mature HM2 was more than 95 O/o homogeneous after purification by two ion- exchange chromatographic steps, while the remaining 5 O/o was constituted by four differently degraded HM2- derived peptidic chains. These by-products were identi- fied and isolated by isolectric focusing techniques and fully characterized according to their amino acid sequence and molecular mass; in addition, selective trypsin cleavage at the residue was applied to con- firm the structural data [12].

It is known that E. coli possesses a number of proteolytic enzymes not only in the cytoplasm but also distributed in inner and outer membranes and in the periplasmic compartment although the physiological functions and the substrate specificty of most of these proteases are still unknown [21]. The main contaminant of HM2 was compound e (see Table 2 and Fig. l), lacking the first two amino acid residues and probably derived from the action of the same leader peptidase I, which correctly processed the primary translational construct OmpA leader peptide-HM2. It has been reported that, in Esche- richia coli, leader-peptide-fused pre-proteins are proc- essed according to the so-called “-3, -1 rule” since the

936 A BOW pr a / Elecrrophoresry 1996, 17, 932-937

C-terminal portion of leader peptide requires amino acids with short, neutral side chains (such as Ala, Gly, Ser and Thr) at position -1 and -3, with position -1/+l representing the cleavage site [22]. If we consider the application of this rule on the sequence of OmpA leader peptide-HM2: - Val-Ala-3-Gln-Ala '-Va1+'-Sert2-Tyr+'- the correct cleavage will give HM2, starting with the Val'' residue. However, leader peptidase also seemed able to act on the secondary cleavage site Ser-Tyr, originating the compound e starting with Tyf3, since non-bulky neu- tral residues Ser+' and Ala ' are in position -1 and -3, as requested by the rule.

Compound b, corresponding to peptidic chain 1-59 of HM2, was originated by a cleavage at the disordered and flexible C-terminal tail of the molecule which can easily interact with proteases. Surprisingly, both compounds a (17-64 HM2) and c (38-64 HM2) derived from cleavage on residues located in the two P-sheets of the tightly packed N-terminal domain of HM2, which are expected to be resistant to proteolysis [12]. However, the fused proteins are probably translocated in unfolded conforma- tion to the periplasmic space and they can be partially attacked by protease before reassuming the folded final conformation [23]. Since an in vitro study demonstrated that unfolded hirudins, under optimized conditions, can assume the correct native conformation within 30 s [24], it can be speculated that periplasmic exported, unfolded HM2 is susceptible to proteases attack for only a short time before assuming the native compact three-dimen- sional structure. Note also that compounds a and c resulted from cleavage on the same Cys-Val dipeptide, respectively, at positions C~s '~ -Va l '~ and C ~ s ~ ~ - V a l ~ * ; this finding suggests that the Cys-Val motif could be a new putative specific cleavage site of a periplasmic protease.

4.2 On interfacing IPGs with mass spectra

Mass spectrometry has become an important methodol- ogy for characterizing peptides and proteins, in particular when combined with other techniques, as for example gel electrophoresis and amino acid sequence analysis [25]. The coupling of SDS-PAGE separation of proteins with mass spectrometry techniques has already been de- scribed. In this approach, gel-separated proteins are first transferred onto a blotting membrane and the blotted protein of interest is analyzed by matrix-assisted laser- desorption time-of-flight mass spectrometry (MALDI-

TOF-MS) [26, 271. In another study, proteins were extracted from agarose gel slices by a "freeze-squeeze" technique before MALDI-TOF-MS [28]. However, in these applications, essentially the same information, namely the molecular mass of the protein, was provided by two different methods. We recently reported the sequential application of IEF-IPG followed by electro- spray mass spectrometry (ES-MS) [18]. This approach had not yet been described, and it is of particular interest since it allows the identification and characteri- zation of protein variants according to both their net charge (PO and their molecular mass. Thus, it is a true 2-D technique, giving the most accurate values for p l and mass. From a technical point of view, samples for ES-MS analysis should be salt-free, as is usually the case with IEF-IPG separations. In fact, IPG is the only tech-

nique allowing recovery of proteins that are simulta- neously isoelectric and isoionic [29]. This also explains why this technique could not be applied to conventional IEF in carrier ampholyte (CA) buffers. CAs are a multi- tude of amphoteric oligoamino, oligocarboxylic buffers, generated by a random synthetic process, coupling oli- goamines with acrylic acid [30]. As proteins are retrieved from a CA-IEF gel, they might be purified from other proteins, but inevitably they are heavily contaminated by the soluble CA buffers, giving a tremendous signal in the ES-MS technique. There is an extra bonus in inter- facing the IPG technique with ES-MS: in quite a number of cases, it is not even necessary to stain the protein band to detect it. Several proteins produce a strong refractive index gradient in the focused zone, so that they are already visible to the naked eye at concen- trations of 5-10 pg/band. Thus, the protein can be detected and the zone excised without resorting to staining techniques. This does not apply, of course, to CA-IEF, where the carrier ampholyte buffers, once focused, generate a continuum of refractive zones covering the entire gel surface from anode to cathode.

The 2-D technique IPG-MS has also been applied by us for solving a unique problem with a recombinant, trun- cated interleukin 6 (IL-6) preparation [31]. This purified sample also contained, in addition to small amounts of lower p l species, a higher p l component, whose structure was most elusive. Only MS data finally helped us to solve the riddle: a mass increment of 32 Da in this higher p l IL-6, coupled to IEF in presence of dithio- threitol, allowed us to conclude that we had a unique protein modification: a trisulfide derivative. The present data, in which MS is also utilized to obtain a final answer on the product, allowed the discovery of a number of partial proteolytic cleavage products, gene- rating both higher and lower p l components, contami- nating the main hirudin band. Thus, it would appear that interfacing multi-separation techniques (conventional IEF, IPGs, SDS-PAGE, chromatographic steps) with MS has become a key technique in investigating macromo- lecular modifications, as also recently pointed out by an in-depth review on this topic [32].

4.3 On the generation of higher pZ species from a parent molecule

The present data also highlight one important mech- anism of post-synthetic protein modification leading to higher p l forms, often unexplained. Whereas a variety of modifications leading to lower p l species are well known and fully documented in the literature, no mechanisms leading to generation of higher p l forms from a parent macromolecule have been advanced so far. It is now evident, from our data, that at least one clear mechanism can be identified: proteolytic cleavage. Let us review our results: (i) Compound a (17-64 amino acids), p l 4.31. This polypeptide has lost a stretch of 16 amino acids from the N-terminus, containing two acidic residues (Asp and Glu). Its p l has thus increased to 4.31 (pl of full length hirudin: 4.03); the A p l being +0.14/residue. (ii) Compound b (fragment 1-59), pZ4.25. This polypep- tide has lost a pentapeptide at the C-terminus, con- taining two acidic residues (Asp and Glu). Its pl has thus

Electrophoresis 1995, 17, 932-931 Characterization of hirudin derivatives 937

increased to 4.25, the A p l being +O.ll/residue. (iii) Compound c (fragment 38-64), PI 4.10. This polypeptide has lost a stretch of 37 amino acids from the N-terminus, containing 1 Asp, 2 Glu and 1 Lys. Here the total A p l i s minute: +0.07. However, it must be stressed that, at pH = pl, it takes more than two Glu residues to neutralize one Lys, since the former is only partially ionized. (iv) Compound e (fragment 3-64), PI 3.98. This molecule has lost only a neutral dipeptide at the N-terminus (Val- Ser). Thus, the negative A PI can only be ascribed to minute pK variations of some charged residues in the three-dimensional native structure. Thus our data sug- gest that whenever an r-DNA protein presents several more basic components, in addition to the usual, more acidic forms, one should immediately look at proteolytic cleavage as the potential villain. The fact that the abso- lute A PIS or the A pZ/residue differ from one cleavage product to the next should not be surprising: we are looking here at “native” structures, still possessing a folded configuration, and not at randomized, fully dena- tured coils as typically existing in 8 M urea and 8-mercaptoethanol. Thus, pK (and PI) variations due to the folded structure should be expected.

4.4 Conclusions

In conclusion, this work confirms that periplasmic export can be an efficient strategy to produce recom- binant proteins provided that they can quickly refold after translocation, even if minor proteolytic degrada- tions can be expected to occur through the action of endogenous membrane and periplasmic proteases. An additional measure to minimize proteolytic degradation of the protein of interest could be the use of protease- deficient Escherichia coli strains [33]. Finally, the precise structural identification and physico-chemical characteri- zation of all minor degraded contaminants of recom- binant hirudin variant HM2 would make it possible to improve the purification protocol and the quality of the final preparation.

PGR gratefully acknowledges support from ASI (Roma, Italy) and from Radius in Biotechnology (ESA, Paris). E . K thanks the European Space Agency for a fellowship, ena- bling her Yo carry out this work at the University of Milano.

Received December 23, 1995

5 References

[ l ] Markwardt, F., Hoppe-Seyler’s Z . Physiol. Chem. 1957, 308,

[2] Dodt, J., Angew. Chem. Znt. Ed. Engl. 1995. 34, 867-880. 131 Dodt, J., Seemiiller, V., Maschler, R., Fritz, H., Biol. Chem. Hoppe-

[4] Rydel, T. J., Tulinsky, A,, Bode, W., Huber, R., J. Mol. Biol. 1991,

[5] Scacheri, E., Nitti, G., Valsasina, B., Orsini, G., Visco, C., Ferrera, M., Sawyer, R. T., Sarmientos, P., Eur. J. Biochem. 1993, 214, 295-304.

[6] Dodt, J., Schnitz, T., Schafer, T., Bergmann, T., FEBS Lett. 1986, 202, 373-377.

[7] Rihel-Bellon, N., Carvallo, D., Acker, M., Van Dorsselaer, A,, Mar- quet, M., Loison, G., Lemoine, y., Brown, S. W., Courtney, M., Roitsch, C., Biochemistry 1989, 28, 2941-2949.

[8] Heim, J., Takabayashi, K., Meyhack, B., Marki, W., Pohlig, G., Eur. J. Biochem. 1994, 226, 3541-3543.

[9] Topol, E. J., Plow, E. F., Thromb. Haemostasis 1993, 70, 94-98. [lo] Marki, W., Wallis, R. B., Thromb. Haemostasis 1990, 64, 344-348. Ill] Anicetti, V. R., Keyt, B. A,, Hancock, W. S., Trends Biotechnol.

[12] Vindigni, A,, De Filippis, V., Zanotti, G., Visco, C., Orsini, G.,

[13] Righetti, P. G., Immobilized p H Gradients: Theory and Methodol-

[14] Righetti, P. G., Wenisch, E., Faupel, M., J. Chromatogr. 1989, 475,

[15] Righetti, P. G., Wenisch, E., Jungbauer, A,, Katinger, H., Faupel,

[16] Giaffreda, E., Tonani, C., Righetti, P. G., J. Chromatogr. 1993, 630,

[17] Black, M . T., Munn, G. R., Allsop, E., Biochem. J. 1992, 282,

[18] Breme, U . , Breton, J., Visco, C., Orsini, G., Righetti, P. G., Electro-

[19] Debadov, D. V., Mol. Biol. 1994, 28, 328-334. [20] de Taxis du Poet, P., Scacheri, E., Benatti, L., Nitti, G., Valsasina,

[21] Maurizi, M. R., Experientia 1992, 48, 178-201. [22] Miiller, M., Experientia 1992, 48, 118-129. [23] Randall, L. L., Hardy, S. J. S., Cell 1989, 46, 921-928. [24] Chang, J. Y., Biochem. J. 1994, 300, 643-650. 1251 Patterson, S. D., Anal. Biochem. 1992, 221, 1-15. [26] Eckerskorn, C., Strupat, K., Karas, M., Hillenkamp, F., Lottspeich,

[27] Klarskov, K., Roepstorff, P., Biol. Mass. Spectrom. 1993, 22,

[28] Dunphy, J. C., Busch, K. L., Hettich, R. L., Buchanan, M.V., Anal.

[29] Righetti, P. G., Isoelectric Focusing: Theory, Methodology and Appli-

[30] Righetti, P. G., Pagani, M., Gianazza, E., J. Chromatogr. 1975, 109,

[31] Breton, J., Avanzi, N., Valsasina, B., Sgarella, L., La Fiura, A., Breme, U., Orsini, G., Wenisch, E., Righetti, P. G., J. Chrornatogr.

[32] Patterson, S. D., Aebersold, R., Electrophoresis 1995, 16,

[33] Meerman, H. J., Georgiou, G., Biotechnology 1994, 12, 1107-1110.

147-156.

Seyler 1985, 366, 379-385.

221, 583-601.

1989, 7, 342-348.

Fontana, A,, Eur. J . Biochem. 1994, 226, 323-333.

ogy, Elsevier, Amsterdam 1990.

293-309.

M., J. Chromatogr. 1990, 500, 681-696.

3 13-327.

539-543.

phoresis 1995, 16, 1381-1384.

B., Sarmientos, P., Blood Coagul. Fibrinol. 1991, 2, 113-120.

F., Electrophoresis 1992, 13, 664-665.

433-440.

Chem. 1993, 65, 1329-1335.

cations, Elsevier, Amsterdam 1983.

341-356.

A 1995, 709, 135-146.

1791-1814.