Domains G1, G2 and the rod-like segment were much more stable against proteolysis. Kinetic analysis indicated a fast cleavage of several different sites in the ...
Eur. J. Biochem. 217, 877-884 (1993) 0 FEBS 1993
Sites of nidogen cleavage by proteases involved in tissue homeostasis and remodelling Ulrike MAYER', Karlheinz MAN", Rupert TIMPL' and Gillian MURPHY' ' Max-Planck-Institut fur Biochemie, Martinsried, Germany Strangeways Research Laboratory, Cambridge, England (Received June 8, 1993) - EJB 93 085512
The cleavage of recombinant mouse nidogen in its native form was examined with granulestored proteases (leucocyte elastase, mast-cell chymase), blood proteases (thrombin, plasmin, kallikrein), matrix metalloproteinases (stromelysin, matrilysin, collagenases) and, for comparison, with trypsin and the endoproteinase Glu-C. More than 50 major cleavage sites were identified by Edman degradation of several large fragments and smaller peptides. The data show an almost exclusive localization of protease-sensitive sites to the flexible segment, connecting the N-terminal globular domains G1 and G2, and within the C-terminal, laminin-binding domain G3. Domains G1, G2 and the rod-like segment were much more stable against proteolysis. Kinetic analysis indicated a fast cleavage of several different sites in the link region followed by destruction of G3 but this was to some extent variable depending on the particular protease. Leucocyte elastase was identified as the most active protease in the cleavage of nidogen whilst stromelysin, matrilysin, plasmin and kallikrein were of distinctly lower activity. No cleavage could be detected with interstitial collagenase and gelatinase A. The peptide analyses also allowed the location of two disulfide bridges within the G3 domain. Complex formation between nidogen and laminin fragments caused some protection against cleavage by thrombin, leucocyte elastase and stromelysin particularly in domain G3. The data indicate a relatively uniform cleavage pattern of nidogen which may be relevant in the context of proteidligand-binding activities associated with domains G2 and G3. The proteolytic processes involved in remodelling and the cellular penetration of basement membranes could therefore be essential for the modulation of the mediator function of nidogen.
Nidogen is a ubiquitous basement-membrane protein and is considered to play a crucial role in the supramolecular organization of these unique extracellular matrices (Aumailley et al., 1993). This was particularly indicated from the binding properties of nidogen which can mediate the formation of ternary complexes between laminin, collagen IV and heparan-sul€ate proteoglycan (Fox et a]., 1991 ; Battaglia et al., 1992). Structural studies and cDNA sequencing have shown that mouse and human nidogen consist of a single, approximately 1200-residue polypeptide chain (Paulsson et al., 1986; Mann et al., 1989; Nagayoshi et al., 1989), and that nidogen and entactin, a protein obtained from a cellculture matrix (Carlin et al., 1981; Durkin et al., 1988), are identical. Recent studies with recombinant mouse nidogen have in addition shown that the protein is composed of three globular domains, G1, G2 and G3 (Fox et al., 1991). The Nterminal domains G1 and G2 are connected by a flexible link region and domains G2 and G3 are connected by a rigid rod composed of five epidermal-growth-factor (EGF)-like repeats. It was further demonstrated that domain G3 binds to laminin (Mann et al., 1988; Fox et al., 1991) and domain Correspondence to R. Timpl, Max-Planck-Institut fur Biochemie, D-82152 Martinsried, Germany Fax: +49 89 8578 2422. Abbreviations. EGF, epidermal growth factor; MMP, matrix
metalloproteinase.
G2 binds to collagen IV and the proteoglycan core protein (Battaglia et al., 1992; Reinhardt et a]., 1993). This separation of binding sites is apparently responsible for the simultaneous interaction of nidogen with different basement-membrane constituents. No potential function has so far been assigned to domain G1 except that it might bind calcium (Aumailley et al., 1993). The folding of nidogen into several domains has also become obvious in previous studies with tissue-derived nidogen which consistently showed the generation of 130-kDa and 100-kDa fragments during extraction and purification (Dziadek et al., 1985; Paulsson et al., 1986). Sequence studies of these fragments demonstrated the release of domain GI and cleavage in front of domain G2 by unknown endogenous proteases (Mann et a]., 1988, 1989). Further studies with laniinin-nidogen complex and nidogen purified from tissues showed a similar and more extensive cleavage with various proteases including thrombin, leucocyte elastase and hemorrhagic metalloproteinases (Mann et al., 1988; Aumailley et al., 1989; Bruch et al., 1989; Baramova et al., 1991). However, nidogen used in these studies was exposed to denaturing agents to dissociate it from laminin which may have changed its native conformation. That this change in conformation had in fact occurred was shown by a collapse of domains GI and G2 and a higher protease sensitivity of tissue-derived nidogen when compared to recombinant nidogen (Fox et al.,
878 1991). However, nidogen bound in tissues also exhibits a high protease sensitivity as indicated by the fast loss of immunoreactivity after exposure to proteases (Dziadek et al., 1988) and during the invasive phase of malignant melanoma (Schmoeckel et al., 1989). Nidogen degradation could therefore be a major step in the disruption of basement membranes. The versatile binding potential and high protease sensitivity of nidogen may together play an important role in the modulation of basement-membrane functions under normal and pathological conditions. For a structural analysis of these possibilities we have now examined the action of eleven different proteases on recombinant nidogen which obviously represents the most native form of the protein (Fox et al., 1991). Most of the proteases used have a prominent role in tissue invasion of cells, the coagulation cascade or, in the case of the matrix metalloproteinases (Woessner, 1991 ; Matrisian et al., 1992), in the remodelling of extracellular matrices. The latter proteases have been shown to cleave a large cartilage proteoglycan (aggrecan), elastin and several collagen types in a restricted fashion (Gadher et al., 1988; Senior et al., 1991 ; Wu et al., 1991 ; Hughes et al., 1991 ; Murphy et al., 1991; Flannery et al., 1992; Fosang et al., 1992). In this study, we demonstrate that these and other proteases, but not collagenases, attack nidogen primarily in the link region between domain G1 and G2 and within domain G3 which emphasizes that a similar restricted degradation may occur in situ.
MATERIALS AND METHODS
I
130
30
,
130
I I
I
165-~ 95 100 -
t
80
I
&
I 1
I
I
200 360
1
I
I
I
I
630
890
1217
Fig. 1. Correlation of the three-domain structure of nidogen with the size of major fragments produced by proteolysis. The model indicates that the globular domains G1 and G2 are connected by a flexible link and G2 and G3 are connected by a rigid rod (Fox et al., 1991). The approximate borders of these domains within the nidogen sequence are denoted underneath by position numbers. The positions of major large fragments are shown by horizontal lines with numbers indicating their apparent molecular mass (ma)as determined by electrophoresis. A certain overlap occurs in the link region due to the different proteases used. Previous studies with recombinant G1 and G2 fragments indicate apparent molecular masses of 30 kDa and 40 kDa, respectively (Reinhardt et al., 1993).
was activated with 1 mM (final concentration) 4-aminophenylmercuric acetate (Sigma) for 90 min at 25 "C. Matrilysin (MMP-7) was activated with 2 mM 4-aminophenylmercuric acetate for 60 min at 37 "C. All proteolytic digestions were performed at 37°C for 0.5-24 h.
Sources of proteins and proteinases Recombinant mouse nidogen was purified from the culture medium of stably transfected human cell clones (Fox et al., 1991). Nidogen was either mixed with laminin fragment P1 (Fox et al., 1991) or with the recombinant laminin fragment B2III3-5 (Mayer et al., 1993) to prepare stable noncovalent complexes which were purified by molecular sieve chromatography. Human leucocyte elastase was kindly provided by Dr H. P. Nick (Ciba-Geigy, Basel) and mast-cell chymase was provided by Dr T. Vartio (University of Helsinki). Trypsin treated with N-a-tosyl-L-phenylalanine chloromethane (Worthington), the endoproteinase Glu-C (ICN Immuno Biologicals), plasmin (Kabi AB), thrombin (Boehringer Mannheim) and kallikrein (Calbiochem) were obtained from commercial sources. Recombinant matrix metalloproteinases (MMP) were expressed from NSO myeloma cells and purified as described previously (Murphy et al., 1991).
Proteolytic cleavage Nidogen was dissolved in 50mM Tris/HCl, pH 7.4, 0.1 M NaC1, 2 mM CaC1, at a concentration of 0.2 mg/ml for cleavage on an analytical scale, or at 0.5-1 mg/ml in the same buffer containing 2 mM N-ethylmaleimide for cleavage on a preparative scale. All proteases were used at an enzyme/ substrate ratio of 1 : 100 with the exception of leucocyte elastase which was used at 1:lOOO. Stromelysin (MMP-3) and interstitial collagenase (MMP-1) were activated prior to incubation with the substrate with 5 pg/ml (final concentration) trypsin for 30 min at 25°C. The activation was stopped with 50 pg/ml (final concentration) soybean trypsin inhibitor (Sigma) for 30 min at 25°C. 72-kDa gelatinase A (MMP-2)
Chromatography and Edman degradation Digests prepared from 0.5-2 mg nidogen were subjected to chromatography using a Superose 12 column (HR 16/50 or HR 10/30, Pharmacia) and aliquots of different pools were hydrolyzed (6 M HCI, llO"C, 16 h) to determine the peptide content on a LC 5001 (Biotronik) amino acid analyzer. Small peptides were further purified by reverse-phase chromatography (Mayer et al., 1991). Large fragments were separated by SDS electrophoresis in polyacrylamide gradient gels ( 5 15%) and transfered to Immobilon membranes (Matsudaira, 1987). N-terminal sequences of these fragments (100200 pmol) were determined using gas phase sequencer models 473A and 470A (Applied Biosystems) following the manufacturer's instructions.
RESULTS Recombinant nidogen (150 kDa) was recently shown to consist of three globular domains (G1 -G3) which are connected by either a long flexible segment or a rod-like structure (Fig. 1). Since evidence was provided that this structure corresponds to the most native form of the protein obtained so far (Fox et al., 1991), recombinant nidogen was used as a substrate for a comprehensive analysis of its protease-sensitive sites. Cleavage with all proteases was initially examined in kinetic studies (0.5-24 h) using electrophoresis and Nterminal sequencing for the identification of large fragments. This showed a rather uniform cleavage in the link and G3 domain. Therefore, digests obtained with some proteases were separated into large and small peptides by molecular sieve chromatography as is shown for leucocyte elastase in
879
A
B
0.3 -
- 94
(349) VFSYNT + (349) V F S Y M +
- 67
(349) VFSYNT +
E
(1)
K
g cu
-43
LNXQ+
- 30
0.2-
- 20
- 14
W
0
K
N d P 1 2 3
(d
e2
0.1-
-
3 4 5 6 , -
1 2 ,
0-
3Yo
I
a
52%
I
I
12
I
8%
5%
28%
I
16
3%
I
I
20
Effluent volume (ml) Fig. 2. Separation of a leucocyte-elastase digest of nidogen using Superose. (A) The digest was obtained after 2 h and the pools (1-6), collected after chromatography, are denoted by horizontal bars. The relative recovery (%) of peptide material in each pool is indicated. (B) Electrophoresis of the major pools 1-3 in comparison to the complete digest (P) and non-digested nidogen (Nd). The molecular masses of calibrating proteins are indicated ( m a ) . N-terminal sequences and their first position in the nidogen sequence (numbers in parentheses) are shown for several of the large fragments.
Fig. 2. The small peptides were separated by reverse-phase chromatography (data not shown) and used to identify more cleavage sites after sequence analysis. The general and special aspects of the cleavage by individual proteases will be described.
Leucocyte elastase and mast cell chymase Leucocyte elastase was by far the most active protease to cleave nidogen. When used at an enzyme/substrate ratio of 1 : 1000 a stable digestion pattern was observed after approximately a 2-h treatment with no intact nidogen left. The first two major fragments that appeared had a mass approximately of 40 kDa and 100 kDa and the latter was subsequently converted to a 80-kDa band. Edman degradation showed that the 40-kDa band originated from the N-terminus and that the 100-kDa and 80-kDa bands possessed the same N-terminal sequence starting just in front of domain G2 (Fig. 2). This indicates a substantial loss of domain G3 upon conversion of the 100-kDa to the 80-kDa component (Fig. 1). Further cleavage of the 80-kDa fragment was very slow but results in a weak 50-kDa band that possessed the same N-terminus (Fig. 2) and could thus correspond to domain G2. A digest possessing the 80-kDa and 100-kDa bands in approximately equal amounts was separated on Superose 12 into two major peaks that contained primarly the S O - k D d 100-kDa (pool 2) and 40-kDa (pool 3) components (Fig. 2). Approximately 16% of the digest was recovered as smaller fragments (pools 4-6) which were further separated by reverse-phase chromatography. Sequence analysis of the major peaks identified six more cleavage sites all located in the G3 domain (Table 1). For the digestion with mast-cell chymase we used a higher enzyme/substrate ratio of 1 : 100 that was also applied in subsequent studies with all the other proteases. Complete digestion of nidogen was achieved after 4 h and yielded ma-
jor fragments of 100 kDa and 65 kDa and a less abundant 80-kDa band (data not shown). The 65-kDa fragment was identified as the N-terminal fragment and apparently contains the G1 domain and most of the link region (Fig. 1). Both larger fragments started with the same sequence SYNTG demonstrating cleavage at position 350/351 just shortly after the major cleavage site of leucocyte elastase. The 80-kDa band did not increase very much at the expense of the 100kDa band even after longer incubations (24 h) indicating that the degradation of the G3 domain is rather slow.
Thrombin, plasmin and kallikrein The digestion of nidogen with thrombin resulted in almost complete cleavage after 8 h to generate a fragment pattern which appeared to be stable even after 32 h treatment (Fig. 3). Major fragments were a 100-110-kDa double band and a 45-kDa band that stained only faintly on the gel. However, both components were released in approximately equal proportions as indicated from similar yields in Edman degradation after blotting the bands. This observation is also in agreement with stoichiometric yields of both fragments when separated using a molecular sieve from a digest of tissuederived nidogen (Mann et al., 1988). Sequence analysis identified the faint 45-kDa band as the N-terminal fragment. The upper 110-kDa band apparently arose by a single cleavage in the link region at position 300/301 as both fragments account for the whole size of nidogen. Further thrombin degradation of nidogen was shown by sequence analysis of the lower 100-kDa band indicating the presence of a fragment similar to the 110-kDa band, a slight shortening at the N-terminal site by cleavage at position 321/ 322 and a cleavage within a disulfide loop of domain G3 at position 1037/1038 (Fig. 3). All three sequences seem to be present in approximately equal proportions as estimated from their comparable yields after Edman degradation. A more
880 Table 1. Peptide bonds of nidogen cleaved by various proteases. The proteases used were chymase (Ch), endoproteinase Glu-C (EG), kallikrein (K), leucocyte elastase (LE), matrilysin (M), plasmin (P), stromelysin (S), thrombin (Th) and trypsin (T). The cleavage sites were identified using the cDNA-derived sequence of mouse nidogen (Mann et al., 1989) and correlated to the domain structure (Fox et al., 1991; Fig. 1). Nidogen domain
Peptide bond (position)
Protease
G1
E-L (5-6) E-L (34-35) D-L (82-83) N-V (91-92)
EG EG S S
K-S (203-204) N-L (220-221) K-S (223-224) E-L (236-237) K-G (244-245) G-L (275-276) R-I (300-301) R-S (321-322) R-F (329-330) E-T (345-346) V-V (348-349) F-S (350-351) S-Y (351-352)
T,P M T EG T M T,Th P,K,Th,T T EG LE Ch S
P-I (896-897) E-R (925-926) R-L (926-927) E-A (938-939) 1-1 (950-951) I-G (951-952) R-A (974-975) S-L (976-977) T-I (984-985) G-I (995-996) R-T (1003-1 004) R-I (1016-1017) E-V (1016-1017) A-K (1018-1019) K-M (1019-1020) R-V (1026-1027) V-L (1027-1028) R-G (1037-1038) G-I (1038-1039) K-I (10.59-1060) E-T (1061-1062) S-H (1063-1064) R-I (1071-1072) G-L (1083-1084) D-A (1087-1088) R-A (1102-1103) E-C (1104-1105) K-T (1139-1140) S-V (1142-1143) K-E (1152-1153) E-M (1153-1154)
S EG T EG LE LE T,Th S,M S M T T EG LE M T LE Th S T EG LE T M EG T EG T S T EG
Link
G3
extensive thrombin cleavage at position 1037/1038, yielding fragment Th22, has been reported for tissue-derived nidogen as well as the release of substantial amounts of two smaller peptides (Thl, Th2) that are located at the N-terminal site of this cleavage (Mann et al., 1988). A similar analysis with the thrombin digest of recombinant nidogen yielded, however, only small amounts of a peptide with the N-terminal sequence of Thl and no evidence could be obtained for the
Fig. 3. The kinetics of thrombin cleavage of nidogen analyzed by electrophoresis. The N-terminal sequences of the double band at approximately 100 kDa were analyzed separately after blotting and showed a single sequence for the upper band but three sequences in approximately equal proportions for the lower band. The calibration was as described in Fig. 2 and the molecular masses are indicated (ma).
Kallikrein
Plasmin
Fig.4. Examples of an incomplete cleavage of nidogen by the plasma proteases kallikrein and plasmin over a period of 24 h. Undigested nidogen is present at t = 0 h. The calibration with marker proteins (kDa) and the N-terminal sequences of several large fragments, with their position numbers indicated in parentheses, are indicated.
presence of Th2. This indicates that thrombin cleavage within domain G3 is slowed down in recombinant nidogen when compared to tissue-derived nidogen. Kallikrein and plasmin were less efficient in cleaving nidogen than thrombin since approximately 50% undigested material remained after a 24-h treatment (Fig. 4). The digestion pattern was relatively simple for kallikrein showing a major N-terminal fragment of 65 kDa and a C-terminal fragment of 95 kDa starting at position 322. Further cleavage to a 30-kDa N-terminal fragment and to a 80-kDa band occurred only at low rates. This suggests a single major cleavage site in the link region and only little cleavage within the G3 domain. Plasmin released mainly a 130-kDa and a 30kDa N-terminal fragment. The larger component was released by cleavage at position 2031204 indicating that this is the initial event in plasmin attack. Minor bands of approximately 115 kDa and 100 kDa were, however, observed and their N-terminal sequences indicate either partial degradation of the link region (cleavage at position 321/322) or of domain G3 since the same N-terminal sequence was found for the 130-kDa and 115-kDa components (Fig. 4). The latter observation is also supported by the presence of the N-terminal sequence in substoichiornetric amounts in the 130-kDa com-
881 Stromelysin
(1)
Matrilysin
LNRQ +
(352) YNTGSQQT + (352) YNTGSQQT (352) YNTGS (1) LNXQELFP +
3
Fig. 5. Cleavage of nidogen by matrix metalloproteinases. The identification of various fragments by N-terminal sequences and size was as described in the previous figure legends. The uppermost band in each lane is undigested nidogen. Several unidentified minor bands could not be sequenced.
ponent. This indicates two alternative mechanisms of cleavage. The major route starts at position 2031204 while a slower reaction may simultaneously occur in domain G3.
Matrix metalloproteinases Since these proteases are important for tissue remodelling (Woessner, 1991; Matrisian, 1992) we have examined nidogen cleavage with stromelysin-1 (MMP-3), matrilysin (MMP-7), interstitial collagenase (MMP-1) and 72-kDa gelatinase A (MMP-2). Collagenase and gelatinase produced no or only negligible cleavage products over a period of 24 h, whilst stromelysin and matrilysin produced partial degradation into several fragments. Sequencing of the stromelysinderived bands demonstrated two N-terminal fragments of approximately 130 kDa and 70 kDa and a 100-kDa band and a doublet band of 80 kDa which were generated by cleavage in front of domain G2 (Fig. 5). Since these bands were of similar density this observation indicates two equivalent routes of cleavage, one starting in the G3 domain, the other by releasing domain G1 and the link (Fig. 1). The 80-kDa bands apparently represent the final product of both cleavage routes and consist mainly of domain G2 and rod. The stromelysin digest was also separated by molecular sieve chromatography which yielded approximately 12% of the total material in the form of small peptides. Sequence analysis of seven major peptides demonstrated five cleavage sites in domain G3 and two more in the center of domain G1 (Table 1). The cleavage pattern with matrilysin was to some extent different to that with stromelysin, but again showed two Nterminal fragments (130 kDa and 45 kDa) and a 100-kDa fragment, resulting from a cleavage in the center of the link region (Fig. 5). Two weaker bands (115 kDa and 70 kDa) were also observed but could not be sequenced. This also indicates that cleavage may start at each end of nidogen. Additional cleavage sites were determined from sequence analysis of small peptides that accounted for 9% of the material in a 24-h digest. This identified a further matrilysinsensitive site in the link and four more major cleavage sites in domain G3 (Table 1). From this analysis also became apparent that, with one exception, none of the major cleavage sites of stromelysin and matrilysin were identical.
Trypsin and endoproteinase Glu-C Trypsin and endoproteinase Glu-C were chosen because of their high activity and restricted cleavage specificity and were used for comparative purposes. Digestion with trypsin was similar to that with leucocyte elastase and resulted initially in a 35-kDa N-terminal fragment and a 100-kDa fragment starting within the link region (Fig. 6). The 100-kDa component is, however, rapidly converted to a 80-kDa band by release of a short N-terminal peptide and substantial cleavage within domain G3. Digestion with endoproteinase Glu-C was somewhat slower and apparently started by a single cleavage in front of domain G2 to release a 67-kDa Nterminal and a 95-kDa C-terminal fragment (Fig. 6). This is followed by conversion of the 95-kDa into an 80-kDa fragment by exclusive degradation of the G3 domain. Analysis of the N-terminal fragment after 8 h digestion, which was purified by molecular sieve chromatography, indicated a progressive shortening from the N-terminus as shown by the appearance of two new sequences. This demonstrated cleavage at positions 516 and 34/35 of nidogen (Table 1). Molecular sieve chromatography of both digests showed similar profiles as for leucocyte elastase (Fig. 2), with pools of smaller peptides that accounted for approximately 20% of the whole digest. These pools were used to purify and sequence the major small peptides. This demonstrated that trypsin cleaved at least four more peptide bonds in the link region besides those identified with the large fragments (Table 1, Fig. 6). Since the size of the initial large fragments only accounted for approximately 90% of the mass of nidogen (approximately 135 kDa) this very likely indicates simultaneous cleavage of all these vulnerable sites. Ten more trypsin cleavage sites were identified within domain G3 (Table 1) which are responsible for the major loss of mass during the 100-kDa fragment to 80-kDa fragment conversion (Fig. 6). Endoproteinase Glu-C, was with seven cleavage sites within domain G3, equally efficient as trypsin (Table 1). Only one additional cleavage site was, however, observed in the link region in agreement with the long persistence of a large N-terminal fragment (67 kDa) which should contain domain G1 and a large portion of the link region.
Localization of two disulfide bridges in domain G3 Domain G3 consists of a C-terminal EGF-like repeat with six cysteines and an approximately 280-residue segment with four cysteines (Mann et al., 1989). The bridging pattern of these four cysteines (cysteines 1-4) was determined by the analysis of several purified peptides demonstrating the connections Cysl-4 and Cys2-3 (Table 2). This included the analysis of a 20-kDa component after electrophoresis and blotting (peptide EG-6) and of several smaller fragments which were obtained with either leucocyte elastase, trypsin or endoproteinase Glu-C and purified by reverse-phase chromatography. All these four fragments showed a clear double sequence with approximately equimolar yields which started in front of two different cysteines. Each disulfide bridge was predicted by two peptides which were obtained with different proteases and are apparently of different size. Together this excludes the possibility that two different but non-disulfidelinked peptides were accidentally copurified. An identical connection pattern was predicted from the proteolytic cleavage of tissue-derived nidogen (Reinhardt, 1992).
882 Trypsin
Endoproteinase Glu-C
Fig. 6. Electrophoretic analysis of the time-dependentcleavage of nidogen by trypsin and endoproteinase Glu-C. The starting material is indicated at t = 0 and the molecular masses of standards are indicated (kDa). N-terminal sequences, including their starting position numbers, are shown for all large fragments identified by staining. Table 2. Localization of two disulfide bridges in nidogen domain G3 by N-terminal sequencing of double peptides obtained with trypsin (T), leucocyte elastase (LE) or endoproteinase Glu-C (EG). Unidentified residues (X) and. unidentified residues which correspond to cysteine are indicated (X). Cysteine bridge (position)
Peptide
N-terminal sequences (starting position)
CYSI/CYS~ (957, 1175)
EG-6
( 939) XXAFLXIPAK (1154) XDXFXPKKQ ( 951) IGLAFDCVD (1171) ALSQCPQP
LE4/5 Cys2ICys3 (1094, 1105)
T4l8 EG4/3
(1072) ILAQDNLGLPNGLTFDA (I 103) AEXLNPAQPGR (1oss) AFSXQLXXVDA (1i 05) XLNPAQPGRRXV
Protection of domain G3 against proteolysis by binding to laminin fragments Domain G3 was previously identified as possessing the single binding site for laminin (Mann et al., 1988; Fox et al., 1991) which raised the interesting question whether formation of the laminin-nidogen complex prevents some proteolytic cleavages. We examined this possibility with a stable non-covalent complex between recombinant nidogen and laminin fragment P1 (approximately 200 kDa) using initially thrombin because of its limited attack on the G3 domain. Electrophoretic analysis of a 24-h digest showed a fragment pattern of N-terminal and C-terminal fragments indistinguishable from that observed with non-complexed nidogen (Fig. 3). Sequence analysis of the 110-kDd100-kDa doublet band, however, revealed cleavage at positions 3001301 and 3211322 but not at position 103711038 clearly demonstrating protection of this G3 site. Further studies were performed with leucocyte elastase and stromelysin using nidogen complexed to the small recombinant laminin fragment B2III3-5 that consists of three EGF-like repeats from the P1 structure (Mayer et al., 1993). Complex formation prevented the significant conversion of the 100-kDa to the 80-kDa band but did not influence the release of the N-terminal40-kDa fragment by leucocyte elastase (Fig. 7). An even stronger effect was observed on the digestion pattern of stromelysin. This included a considerable decrease in the N-terminal fragments of 130 kDa and 70 kDa
Stromelysin
- + +
24 24 0
Leucocyte Elastase
+
4 8 2 4 4 8 2 4
h
Fig. 7. The protection of nidogen proteolysis by complex formation with the recombinant laminin fragment B2III3-5. The noncovalent complex nidogen-B21113-5 (+) and nidogen alone (-), for comparison, were cleaved with stromelysin or leucocyte elastase for the indicated periods of time. The single lane (0)shows a control incubation (24 h) of the complex in the absence of any exogenous proteases. The calibration with marker proteins (ma)allows identification of protein bands generated by leucocyte elastase (Fig. 2) and stromelysin (Fig. 5). The band at approximately 30 kDa in the complex corresponds to B2III3-5 which is resistant against proteolysis (Mayer et al., 1993).
but also a reduced production of the 100-kDa fragment which is generated by a single cleavage in the link region (Fig. 7). In addition, the 100-kDa to 80-kDa fragment conversion was impaired. These data demonstrate a strong protection of the G3 domain by bound B2III3-5 but also, as in the case of stromelysin, some reduced cleavage in more remote regions such as the flexible link.
DISCUSSION Examination of eleven proteases showed that, except for interstitial collagenase and gelatinase, distinct cleavage of native nidogen was restricted to a set of large fragments. This demonstrated that the flexible link and domain G3 present the most vulnerable sites for proteolytic digestion (Fig. I). The kinetic data indicated a prefered start either in the link region, i.e. by leucocyte elastase, trypsin, endoproteinase Glu-C, or a comparable attack in both link and G3 by plasmin, stromelysin and matrilysin. The fragment patterns correspond with some variations to the size of nidogen frag-
883
1 -
~ G2___
collagen IV proteoglycan
C
U
+
T T
4 Ttypsin
4 t
7
7
EG
rod
N-giycosylation
larninin
1
TTT
U L T I
Thrombin
?
Plasmin
i
LeuCOCVe Elastase
Kallikrein
T
Chyrnase
T
Matrilysin
1 Strornelysin
0-glycosylation
Fig. 8. Correlation of the linear nidogen sequence and the location of protein binding and glycosylation sites with various protease-sensitivesites (A-C). Cleavage sites are with endoproteinase Glu-C (EG) and trypsin (A), with thrombin, kallikrein, plasmin, chymase and leucocyte elastase (B) and with matrilysin and stromelysin (C). This scheme is incomplete since not all the cleavage sites have been determined. In addition, the top scheme shows the localization of two disulfide bridges (Table 2) in domain G3 which were determined by sequence analysis of cross-linked proteolytic fragments.
ments Nd-130, Nd-100 and Nd-80 that were generated during purification of nidogen from EHS tumor (Dziadek et al., 1985; Paulsson et al., 1986) and to fragments obtained by digesting tissue-derived nidogen with trypsin or pancreatic elastase (Mann et al., 1988). However, in the latter study no N-terminal fragments could be obtained, except with thrombin, and the final digestion product was a 40-kDa fragment comprising the rod-like segment. This indicates that exposure of tissue nidogen to 2 - 6 M guanidine increases protease sensitivity, as has been also shown for recombinant nidogen (Fox et al., 1991). This was recently confirmed by the study of Sires et al. (1993) who showed degradation of nidogen that was exposed to denaturing and reducing agents by matrilysin, collagenase and gelatinase B. However, the matrilysin cleavage pattern was considerably different to that observed in this study and included vulnerable bonds in domain G1 and the rod-like segment. A map of cleavage sites within the link region and domain G3 demonstrates that they possess sensitive peptide bonds distributed over almost their entire length (Fig. 8) and suggests that the link is composed of approximately 150 residues. Actually, the flexible link was first predicted from the location of cleavage sites by endogenous proteases (subdomain Ib; Mann et a'., 1989) and subsequently by electronmicroscopical evidence for a non-visible connecting element between domains GI and G3 (Fox et al., 1991). The average length of the connection (5 nm) is, however, much to short to accomodate 150 residues in a random coil. This suggests a distinct folding pattern for the link which cannot be properly visuaIized by negative staining or rotary shadowing. In this context, it is of interest that some proteases (thrombin, plasmin, kallikrein, stromelysin) produce obviously only one or two cleavages in the link region while others like trypsin exhibit a broader action. The high protease sensitivity of globular domain G3 is more surprising since it contains, at the C-terminal end, an EGF-like repeat with six cysteines (Durkin et al., 1988; Mann et al., 1989) and four more cysteines in the larger N-ternunal part whose bridging patterns were determined in the present study for recombinant nidogen (Fig. 8). However, many cleavages were observed be-
tween these two disulfide bridges and a few at the N-terminal site of the first cysteine residue. The latter are located in a predicted subdomain IIIa (Mann et al., 1989) which could represent another cysteine-free connecting segment of approximately 50 residues. The two other globular domains G1 and G2 were remarkably resistant to proteolysis. Domain G2 also contains an EGF-like repeat starting at position 360, which determines the border to the flexible link (Mann et al., 1989) and a further disulfide bridge that connects the repeat to the C-terminal end of the domain (Reinhardt, 1992). Domain GI, however, contains no disulfide bridges but is also highly protease resistant except for short N-terminal regions cleaved only by endoproteinase Glu-C and stl-omelysin (Table 1). Denatmation of G1 apparently causes irreversible changes in its conformation and generates sensitivity towards trypsin and pancreatic elastase (Mann et al., 1988). Both globular domains show no sequence similarity to other protein motifs (Durkin et al., 1988; Mann et al., 1989; Nagayoshi et al., 1989) which makes it an interesting task to determine their three-dimensional structures. One explanation for the protease resistance of domains G1 and G2 could be the presence of single N-glycosylation sites (Fig. 8) which are fully occupied in mouse nidogen by complex type oligosaccharides (Fujiwara et al., 1993). These sites are, however, not conserved in human nidogen (Nagayoshi et al., 1989) and the role of N-glycosylation needs, therefore, to be further examined. Mouse nidogen contains, in addition, approximately seven O-glycosidically linked oligosaccharides (Paulsson et al., 1986; Fujiwara et al., 1993) that are located in the link region (positions 271-320) and in subdomain IIIa of the G3 domain (positions 892 and 905). These regions are readily cleaved by several proteases (Fig. 8) and occupation by oligosaccharides seems therefore not to have a protective effect. The peptide bonds cleaved in nidogen (Table 1) showed consistently for trypsin, thrombin, plasmin and kallikrein, as expected, an arginine or lysine at the P1 site. The metalloproteinase-cleavage sites N-terminal to hydrophobic residues were also typical of these enzymes both in vitru and in vivo. Endoproteinase Glu-C cleaved at the C-terminal site of glutamic acid and in a single case for aspartic acid corresponding to the established specificity (Houmard and Drapeau, 1972). Leucocyte elastase cleaves mainly between hydrophobic residues and the pattern for nidogen, including a cleaved S-H bond, was very similar to that for the insulin B chain determined previously (Blow, 1977). Simi I arly, mast-cell chymase with a chymotrypsin-like specificity cleaved a single F-S bond of nidogen as has been observed for bradykinin (Trong et al., 1987). The peptide bonds cleaved by endogenous proteolysis of tissue nidogen were Lys-Ser (position 203/204) in Nd-130, Asn-Val (287/288) in Nd-100 and PheSer (350-351) in Nd-80 (Paulsson et al., 1986; Mann et al., 1989). The first and third peptide bond correspond to cleavage sites of plasmin and mast-cell chymase, respectively. This suggests that these proteases may become released or activated and generate the nidogen fragments during extraction of the EHS tumor. Whether they exhibit a similar role and cleavage specificity during physiological or pathological situations in situ, however, remains to be demonstrated. Plasmin, stromelysin and matrilysin are the proteinases most likely to be involved in nidogen turnover during normal remodelling in vivo, since they are generated by resident connective tissue cells. However, during inflammatory situations the action of elastase, thrombin, kallikrein and chyrnase may
884 be significant. Other proteinases, including the lysosomal cysteine proteinases, may also function extracellularly at such times, as well as metalloproteinases and plasmin produced by infiltrating haematopoietic cells. Our data on diverse cleavage specificities could therefore provide guidance to identify the participation of such proteinases in various situations in vivo. It was previously shown that laminin binds by a single high-affinity site in its B2 chain to domain G3 of nidogen (Mann et al., 1988; Gerl et al., 1991; Fox et al., 1991). This site is located within a single EGF-like motif of laminin (Mayer et al., 1993). As shown in this study, combination of nidogen with the motif present in the recombinant 171-residue fragment B2III3-5 leads to an extraordinary stabilization of domain G3 against proteolysis. Whether this occurs by steric hindrance and/or a conformational change remains to be analyzed. The extensive degradation of the cross-linked complex with endoproteinase Glu-C was shown to cause the slow release of B2III3-5 bound to an approximately 80-residue nidogen fragment which originated from a segment between the first two cysteine residues of domain G3 (Mayer et al., 1993). However, binding of B2III3-5 also reduces leucocyte elastase and stromelysin cleavage of G3 in front of the first cysteine residue (Fig. S), indicating that this region is also protected. The protection of more remote sites, such as sites in the link, may also occur. Since this could indicate allosteric modulation of nidogen by laminin binding this point will be worth further examination. Nidogen probably has a mediator role in the formation of basement membranes since it promotes the in vitro formation of ternary complexes including laminin, collagen IV and heparan-sulfate proteoglycan (Fox et al., 1991; Battaglia et al., 1992; Aumailley et al., 1993). Proteolytic release of G3 in situ could, therefore, reduce the potential participation of laminin in these complexes. This would, however, not interfere with the binding of collagen IV and proteoglycan which can apparently be mediated by different sites in domain G2 (Battaglia et al., 1992; Reinhardt et al., 1993). The possible functional consequences of the release of domain G1 by cleavage in the flexible link are so far not clear. Preliminary data (Mayer, U. and Timpl, R., unpublished results) indicate that G1 is involved in the limited self-assembly of nidogen to 'nest-like' oligomers. The biological relevance of this process has so far not been explored but it could potentiate the postulated mediator function of nidogen by polyvalent interactions. These possibilities would also explain the rapid release of nidogen from tissues by endogenous and exogenous proteases under pathological conditions (Dziadek et al., 1985, 1988; Schmoeckel et al., 1989).
Bararnova, E. N., Shannon, J. D., Fox, J. W. & Bjamason, J. B. (1991) Biomed. Biochim. Acta SO, 763-768. Battaglia, C., Mayer, U., Aumailley, M. & Tirnpl, R. (1992) Eul: J. Biochem. 208, 359-366. Blow, A. M. J. (1977) Biochem. J. 161, 13-16. Bruch, M., Landwehr, R. & Engel, J. (1989) Eur: J. Biochem. 185, 271 -279. Carlin, B., Jaffe, R., Bender, B. & Chung, A. E. (1981) J. Biol. Chem. 256, 5209-5214. Durkin, M. E., Chakravarti, S., Bartos, B. B., Liu, S.-H., Friedrnan, R. L. & Chung, A. E. (1988) J. Cell Biol. 107, 2749-2756. Dziadek, M., Paulsson, M. & Timpl, R. (1985) EMBO J. 4, 25132518. Dziadek, M., Clements, R., Mitrangas, K., Reiter, H. & Fowler, K. (1988) Eul: J. Binchem. 172, 219-225. Flannery, C. R., Lark, M. W. & Sandy, J. D. (1992) J. Biol. Chem. 267,1008-1014. Fosang, A. J., Neame, P. J., Last, K., Hardingham, T. E., Murphy, G. & Hamilton, J. A. (1992) J. Biol. Chem. 267, 19470-19474. Fox, J. W., Mayer, U., Nischt, R.. Aumailley, M., Reinhardt, D., Wiedemann, H., Mann, K., Tirnpl, R., Krieg, T., Engel, J. & Chu, M.-L. (1991) EMBO J. 10, 3137-3146. Fujiwara, S., Shinkdi, H., Mann, K. & Timpl, R. (1993) Matrix 13, 215 -222. Gadher, S. J., Eyre, D. R., Duance, V. C., Wotton, S. F., Heck, L. W., Schmid, T. M. & Woolley, D. E. (1988) Eur: J. Biochem. 175, 1-7. Gerl, M., Mann, K., Aumailley, M. & Tirnpl, R. (1991) Eur. J. Biochem. 202, 167-174. Hournard, J. & Drapeau, G. R. (1972) Proc. Nut1 Acad. Sci USA 69, 3506 -3.509. Hughes, C., Murphy, G. & Hardingham, T. E. (1991) Biochenz. J. 279. 733-739. Mann, K., Deutzmann, R. & Timpl, R. (1988) Eur: J. Biochem. 178, 71 -80. Mann, K., Deutzmann, R., Aumailley, M., Tirnpl, R., kdimondi, L., Yamada, Y., Pan, T., Conway, D. & Chu, M.-L. (1989) EM30 J. 8, 65-72. Matrisian, L. M. (1992) Bioessays .14, 4.55-463. Matsudaira, P. (1987) J. Biol. Chem. 262, 10035-10038. Mayer, U., Aurnailley, M., Mann, K., Timpl, R. & Engel, J. (1991) Eur: J. Biochem. 198,141-150. Mayer, U., Nischt, R., Poschl, E., Mann, K., Fukuda, K., Gerl, M., Yamada, Y. & Timpl, R. (1993) EM30 J. 12, 1879-1885. Murphy, G., Cockett, M. I., Ward, R. V. & Docherty, A. J. P. (1991) Biochem. J. 277, 277-279. Nagayoshi, T., Sanbom, D., Hickok, N. J., Olsen D. R., Fazio, M. J., Chu, M.-L., Knowlton, R., Mann, K., Deutzmann, R., Timpl, R. & Uitto, J. (1989) DNA (NY) 8, 581-594. Paulsson, M., Deutzmann, R., Dziadek, M., Nowack, H., Tirnpl, R., Weber, S. & Engel, J. (1986) Eur: J. Biochem. 156, 467-478. Reinhardt, D. (1992) Ph. D. thesis, University of Munich. Reinhardt, D., Mann, K., Nischt, R., Fox, J. W., Chu, M.-L., Krieg, T. & Timpl, R. (1993) J. Biol. Chem. 268, 10881-10887. Schmoeckel, C., Stolz, W., Sakai, L. Y., Burgeson, R. E., Timpl, We are grateful to the excellent technical assistance of Heidi R. & Krieg, T. (1989) J. Invest. Dermatol. 92, 663-668. Alberti, Vera van Delden, Christa Wendt and Wolfgang Strasshofer Senior, R. M., Griffin, G. L., Fliszar, C. J., Shapiro, S. D., Goldberg, and to financial support by the Deutsche ForschunRsgemeinschu~ G. I. & Welgus, H. G. (1991) J. Biol. Chem. 266, 7870-7875. (project Ti 95/7-2). Sires, U. I., Griffin, G . L., Broekelman, T. J., Mecham, R. P., Murphy, G., Chung, A. E., Welgus, H. G. & Senior, R. M. (1993) J. Biol. Chem. 268, 2069-2074. REFERENCES Trong, H. L., Neurdth, H. & Woodbury, R. G. (1987) Proc. Nut1 Aurnailley, M., Wiedemann, H., Mann, K. &L Timpl, R. (1989) Eur: Acad. Sci USA 84, 364-367. J. Biochem. 184, 241 -248. Woessner, J. F. (1991) FASEB J. 5, 214552154, Aumdilley, M., Battaglia, C., Mayer, U., Reinhardt, D., Nischt, R., Wu, J.-J., Lark, M. W., Chuh, L. E. & Eyre, D. E. (1991) J. Bid. Timpl, R. & Fox, J. W. (1993) Kidney lnt. 43, 7- 12. Chem. 266, 5625-5628.
