Eur. J . Biochem. 185,271 -279 (1989) 0 FEBS 1989

Dissection of laminin by cathepsin G into its long-arm and short-arm structures and localization of regions involved in calcium dependent stabilization and self-association Marcel BRUCH, Ruth LANDWEHR and Jiirgen ENGEL Abteilung Biophysikalische Chemie, Biozentrum der Universitat, Basel

(Received June 12, 1989) - EJB 89 0743

Native laminin - nidogen complex isolated from mouse Engelbreth-Holm-Swarm tumor was treated with purified cathepsin G or leucocyte elastase, two neutral serine proteases which play a role in inflammatory processes accompanied by degradation of basement membranes. Both enzymes were found to be more active than porcine pancreatic elastase. In the absence of Ca2 +, laminin fragments produced by leucocyte elastase resembled those formed by the pancreatic enzyme but at physiological concentrations of CaZf cleavage by cathepsin G was much more selective. Initially laminin (900 kDa) was cleaved at two major sites only with similar rates leading to three fragments. Fragment C1-4 (about 550 kDa) comprises the intact three short arms of the molecule and fragment C8 - 9 (about 350 kDa) contains the entire triple-coiled region by which its three chains are assembled and the major part of the terminal globular domain of the long arm. The remaining C-terminal region of this domain was recovered as fragment C3 of about 50 kDa. Stabilization against proteolytic attack was restricted to the region of fragment C1-4 and only this fragment exhibited strong Ca2+ dependent self-association similar to that of intact laminin or of its complex with nidogen. The associative properties of fragment C1-4 were dramatically diminished upon removal of the tip of one of the short arms comprising fragment 4. In addition, this provides a clear assignment of the important laminin function to a distinct domain in one of its short arms. The new fragment C8 - 9 may be employed for exploring the properties and possible functions of the upper long-arm region which so far has not been available as a fragment.

Laminin, a major glycoprotein component of basement membranes [ l , 21, has attracted much interest because of its biological functions [3] which include cell attachment [4 - 61, promotion of neurite outgrowth [7], stimulation of cell growth [8], self assembly [9] and interaction with other components of the extracellular matrix [lo]. The protein (m = 900 kDa) consists of three polypeptide chains A (m = about 440 kDa), B1 and B2 (about 220 kDa each) which are arranged in a structure of an elongated cross with three short arms (35 nm) and one long arm (76 nm) [ l l - 131. It is a typical mosaic protein [14] which contains a large number of different struc- tural and functional domains. Laminin forms a Stable 1:l complex with nidogen, also referred to as entactin [I , 2, 151.

Fragmentation by limited proteolysis [13, 151 is instrumen- tal for the elucidation of its domain organization and assign- ment of distinct functions to individual domains. Various enzymes are employed and a number of well defined fragments originating from the short-arm structures and from the long arm are available for structural and functional studies. The most useful and widely used fragments 1, 3 , 4 and 8 (for their localization in laminin see Fig. 7) are obtained after limited proteolysis by pancreatic elastase or pepsin [13, 161. These enzymes, however, completely degrade some regions of the molecule into small peptides. There is a need for proteases which cleave at more restricted sites of laminin and for frag-

Correspondence to J. Engel, Abteilung Biophysikalische Chemie, Biozentrum, Klingelbergstrasse 70, CH-4056 Basel, Switzerland

Enzymes. Cathepsin G (EC 3.4.21.20); leucocyte elastase (EC 3.4.21.37); pancreatic elastase (EC 3.4.21.36).

mentation procedures by which susceptible regions can also be retained in fragments. Proteolytic fragmentation of laminin is also of interest in view of the importance of the degradation of basement membranes during inflammatory processes [17, 181 and tumor invasion [18, 191. In the present work, we studied the action of two physiologically important proteases, cathepsin G and leucocyte elastase, on laminin. These pro- teases are released by polymorphonuclear leucocytes during inflammation and tissue remodelling [20, 211. Since calcium is important for stabilization [22] and self assembly [9, 23, 241 of laminin, its influence on the limit digests and on the association of the fragments formed was also studied.

MATERlALS AND METHODS

Materials

Laminin - nidogen complexes were prepared from the mouse Engelbreth-Holm-Swarm tumor as previously de- scribed [24, 251. Cathepsin G and elastase were purified from human polymorphonuclear leucocytes as described [26]. Frag- ment E8, prepared according [24], was a generous gift of 1. Hunter. Laminin uncomplexed with nidogen and the affinity purified antibody against intact nidogen were kindly provided by D. Aeschlimann. Human prothrombin was purified from human plasma [27] and thrombin was formed from prothrom- bin as previously described [28]. Porcine pancreatic elastase was purchased from Serva (Heidelberg). Phenylmethylsul- fonyl fluoride was purchased from Fluka (Buchs).

272

Enzymatic digestion

Proteolytic digestion of the laminin - nidogen complex (1.6 mg/ml) by cathepsin G in the presence or absence of calcium was performed at 37 "C at an enzyme/substrate mass ratio of 1 : 300 in 20 mM Tris/HCl pH 8.0 containing 150 mM NaCl. Digestion of laminin - nidogen complexes by leucocyte elastase or thrombin was performed at 37°C at an enzyme/ substrate mass ratio of 1 : 100. At given time intervals, aliquots (55 pg) of the digestion mixture were taken and the activity of cathepsin G was stopped by addition of an excess of phenylmethylsulfonyl fluoride (1 mM final concentration). The digestion products were identified by sodium dodecyl sulphate/polyacrylamide gel electrophoresis (SDS/PAGE) [29] on 3 - 15% gradient gels without prior reduction.

Preparation of the long arm fragment ojlaminin

Laminin - nidogen complex (7 mg, 1.05 rngiml) in 20 mM Tris/HCl pH 7.4 containing 150 mM NaC1, 5 mM Ca2+ were incubated for 6 h at 18 "C at an enzyme/substrate mass ratio of 1 : 300. The reaction was stopped by addition of an excess of phenylmethylsulfonyl fluoride (1 mM final concentration) and diluted 1 :2 with 20 mM Tris/HCl pH 7.4. The reaction mixture was submitted to fast protein liquid chromatography (FPLC, Pharmacia, Uppsala, Sweden) on a Mono Q HR 5/5 (Pharmacia) ion exchange column equilibrated in 20 mM Tris/HCl pH 7.4. The polypeptide fragments were eluted by application of a linear gradient of 0-0.5 M NaCl in 20 mM Tris/HCl pH 7.4. The separation and identification of the different fragments were monitored by SDSjPAGE and elec- tron microscopy. Fractions containing the long-arm fragment were pooled, concentrated by ultrafiltration on PM-10 Amicon membranes and subjected to gel filtration on a Superose 6 column (Pharmacia) equilibrated in 20 mM Tris/ HC1 pH 7.4 containing 100 mM NaCI. This step removed low-molecular-mass contaminants.

Circular dichroism and transition profiles

Circular-dichroic spectra were recorded on a Cary 61 spectropolarimeter calibrated with d-10-camphorsulfonic acid. Thermostatted quartz cells of 0.01 -0.1 cm path length were used. The molecular ellipticity [O] (expressed in deg cm2/ dmol) was calculated on the basis of a mean residue molecular mass of 110 Da. Thermal transition curves were recorded at fixed wavelength (222 nm) by raising the temperature at the rate of 18"C/h using a temperature programmer. The esti- mation of cc-helix content was determinded according to the procedure of Chen [30].

Electron microscopy

The rotary shadowing technique was used as described previously [31]. Protein samples were dissolved in 0.2 M am- monium bicarbonate, pH 7.9 and, after addition of an equal volume of glycerol, sprayed onto freshly cleaved mica disks. Negative staining followed a previously used procedure [31].

Analytical ultrucentriJigation

Analytical ultracentrifugation was performed in a Spinco model E analytical ultracentrifuge (Beckman Instruments) equipped with a photoelectric scanner. The sample (fragment C8 - 9) was 0.33 mg/ml in 20 mM phosphate pH 7.4 contain-

ing 100mM NaCl. The absorbance A was measured at 280 nm. Sedimentation velocity runs were performed at 52000 rpm at 20 "C with an epon double-sector cell in an AnD rotor. Sedimentation coefficients were corrected to water at 20 "C. Sedimentation equilibrium runs were performed at 7100 rpm in an AnF rotor. The molecular mass was evaluated from In A versus r2 plots where r is the distance from the rotor center. The frictional coefficientfwas calculated in the usual manner. The axial ratios a/b of prolate ellipsoids of revolution, as hydrodynamic equivalents, were estimated from the ffmin

ratios to Perrin's tables [32].

RESULTS

Fragmentation of laminin - nidogeit complex and laminin by cathepsin G in the presence and absence of calcium

Native laminin was employed for the digestion studies in the form of its complex with nidogen in order to secure native protein (Fig. 1A) [25]. The 150-kDa protein nidogen and its 100-kDa fragment were well separated from fragments of laminin. Their identification was confirmed in control exper- iments (Fig. IB) where laminin uncomplexed with nidogen was proteolyzed under identical conditions. Comparison of the two fragmentation patterns clearly assigned the presence of the 100-kDa band to the 100-kDa fragment of nidogen [33] (Fig. 1A). This was also confirmed by immunoblotting after electrophoresis with an affinity purified antibody against in- tact nidogen. Only the 100-kDa fragment of nidogen and intact nidogen were detectable.

Three laminin fragments designated Cl', C4 and C3 were produced by cathepsin G (Fig. 1) in the presence of Ca2+ or EDTA but at different rates. They closely resemble in size the well known fragments El', E4 and E3 produced by porcine elastase [13, 15, 24, 341. Fragment E4 makes up the outer portion of the short arm [13] and El ' was shown to represent most of the remaining short-arm structures [15,24]. Fragment E3 originates from the C-terminal region of the globular struc- ture at the tip of the long arm [13, 351. For the localization of these fragments in the laminin molecule see Fig. 7 below. The close resemblance of the cathepsin-G-produced fragments with the corresponding elastase fragments was confirmed by electron microscopy after isolation (not shown).

We focussed our interest on two new large fragments of laminin produced by cathepsin G, designated C1- 4 and C8 - 9 in Fig. 1. C1- 4 was only transiently detectable in the presence of Ca2+ and was subsequently converted to Cl' with simultanous formation of fragment C4 (Fig. 1). In the absence of Ca2+ fragments C1' and C4 are formed directly without intermediates. The data suggests that fragment C1- 4 com- prises the intact short-arm structures including the fragment 4 region. The interpretation is compatible with its large appar- ent molecular mass of 550 kDa and is confirmed by electron microscopy (see below). Fragment C1-4 appears simulta- neously with C8 - 9 and C3 in the initial period of digestion (5-15 min, Fig. 1) at which no other fragments of laminin are apparent. Fragment C8 -9 appears as two closely spaced bands in positions of 170 and 190 kDa. It is significantly larger than fragment E8 (140 + 80 kDa) produced by elastase [25,35]. Only at prolonged digestion periods is fragment C8 - 9 converted to a fragment (C8) corresponding in size to E8 (not shown). Since E8 can be identified as a fragment comprising about half of the long arm [25, 351, the data suggested that fragment C8 - 9 is a novel large fragment of the long arm.

273

+ Ca ++

A

L-

N-

-m -Cl' -

-=- -CL - -c3 -

+ EDTA

- 150

- 91

- 67

- L3

-3

0 1 5 10 15 30 60 1 5 10 l5 3 0 6 0 Reaction time(min 1

B L- --9oo -a

-Cl -

-=- -CL - -c3 -

- 150

- 67

- L3

- 30

0 1 5 1 5 3 0 6 0 1 5 15 30 60 Reaction time (min 1

Fig. 1. Time course ofjragmentation o j the laminin -nidogen complex and laminin by cathepsin G. Laminin -nidogen complex [A] or laminin [B] were digested at an enzyme/protein mass ratio of 1 :300 in the presence of 5 mM CaZ+ [ + C a + + ] or 5 mM EDTA [ + EDTA] at 37°C for the times indicated. Samples were analyzed by SDSjPAGE under non-reducing conditions. Untreated complex was analyzed as a rcference (lane 0). L = laminin, N = nidogen, N-100 = 100-kDa fragment of nidogen; C1-4, Cl', C8-9, C4 and C3 are cathepsin derived fragments of laminin. Molecular masses and positions of marker proteins (laminin, 900 kDa; nidogen 150 kDa, phosphorylase b, 11 7 kDa; bovine serum albumin, 67 kDa; chicken ovalbumin, 43 kDa; carbonic anhydrase, 30 kDa) are indicated at the right margin

Similar to porcine pancreatic elastase [15, 221, leucocyte elastase cleaves the laminin - nidogen complex directly into fragments El ' and ES without significant amounts of larger intermediates (not shown). At the same enzyme/substrate mass ratio 1 : 100 and at 37"C, leucocyte elastase was found to cleave laminin at a 5-10-fold higher rate than porcine elastase. Since selective cleavage of laminin by thrombin was reported [36], some experiments were also performed with this enzyme. Laminin - nidogen was treated by thrombin at an enzyme/substrate mass ratio of 1 : 100 at 37°C. No fragmen- tation of laminin was detectable, whereas nidogen was frag- mented into its 100-kDa fragment within the shortest time of digestion, i.e. 15 min [15, 371.

Stabilization of short-arm structures by Ca''

As described above, Ca2 + stabilizes the short-arm struc- tures against proteolytic attack. For further quantification,

the cathepsin G digestions were carried out at various Ca2+ concentrations and the fraction of fragment C1-4 which was observed after 30 min was determined (Fig. 2). Half- maximum stabilization of fragment C1- 4 against further fragmentation into fragments C1' and C4 was observed at about 90 FM Ca2+.

Selective aggregation of short-arm structures by Ca2 ' and separation from non-associated jragments of the long arm

Laminin - nidogen complex was treated by cathepsin G at 37°C as described above. After 30 min, the reaction was stopped by inhibition of the enzyme and aggregated material was separated from non-associated fragments by centrifuga- tion at low speeds. According to analysis by SDSjPAGE (Fig. 3), fragments CS - 9 and C3 are quantitatively retained in the supernatant which, however, contains only about 20% of the amount of fragment C1-4 which is initially present in

274

0 10 100 1000 cat+ I U M )

Fig. 2. Dependence ofthe resistance ojfragment CI - 4 to cathepsin G digestion as a function of CaZ+ concentration. Digestions were carried out as described in the legend of Fig. 1 for 30 min and the amount of fragment C1-4 was determined by densitometry of the correspond- ing band (see Fig. 1). Fractions of the maximal amounts were calcu- lated by normalization of these values with the maximum values observed at 5 mM Ca2+

c I-L - c 1' -

C 8-9 -

R M SN P

Fig. 3. Identification of associated and non-associated fragments by SDSIPAGE. Laminin - nidogen complex was digested by cathepsin G as described in the legend of Fig. 1 . The reaction was stopped by addition of phenylmcthylsulfonyl fluoride after 30 min and aggre- gated material was pelleted by centrifugation (15 min at 10000 g). Samples of the reaction mixture before separation (RM), the super- natant (SN) and the pellet (P) were analyzed by SDSjPAGE under non-reducing conditions. Before application, the pellet was solubilized by heating to 70°C in the SDSjPACE sample buffer con- taining 1 % SDS

the digestion mixture. This was also demonstrated by electron microscopic inspection of the supernatant (Fig. 4 a). Almost homogenous fields of particles comprising the long arm of laminin as well as small globules reminiscent of fragment 3 are observed. Very little contamination by fragment C1-4 is visible.

Even before centrifugation, only a few non-associated short-arm structures are visible (Fig. 4 b). These all reveal three well preserved short arms with globular domains at their tips and in their middle regions. One of the arms consistently reveals a third globular domain (arrows in Fig. 4b) located

near the center of the cross. These particles are interpreted as intact short-arm structures preserved in the form of fragment C1- 4. Most of this fragment is pelleted at the relatively low centrifugal field employed (1 0 000 g) indicating its high degree of association in the presence of 5 mM Ca2+. Interestingly the 100-kDa fragment of nidogen is also selectively pelleted, indicating that it is highly associated with fragment C1-4. Electron micrographs of the pelleted material (not shown) confirm the presence of large complexes of entangled short- arm structures but fail to reveal details because of the high degree of aggregation. Attempts to redissolve the pellet under non-denaturing conditions by dilute acetic acid and buffers containing 10 mM EDTA were not successful.

In similar experiments with cathepsin G digests in the presence of 5 mM EDTA (for conditions see Fig. 1) no pellet was formed. Also no insoluble aggregates were formed when Ca2+ was added (2.5 mM excess over EDTA, 3772) to such digests which contained fragment C1' instead of C1-4 (see Fig. 1). In control experiments reported findings [24] were confirmed: untreated laminin - nidogen complex did not ag- gregate in the presence of 5 mM EDTA but a pellet was formed in the presence of 5 mM Ca2 + .

Isolation and chain composition of fragment C8 - 9

Laminin - nidogen complex was digested with cathepsin G i n the presence of 5 mM Ca2+ at 18 "C for 6 h. The resulting gel pattern is similar to that observed after 30 min at 37°C (Fig. 1A). Purification of the native fragment was achieved by a combination of ion-exchange and gel-permeation chromatography (see Materials and methods). Fragment C8 - 9 is eluted from the Mono Q column by an NaCl gradient at 0.25 M salt. Fragments C1-4 and C1' elute in a broad profile over 0.35 -0.45 M NaCl. Electron microscopy reveals monomeric C8-9 (Fig. 6) but highly aggregated C1-4 frag- ments (not shown). The native C8 -9 fragment elutes with a single profile in gel permeation chromatography and dis- sociates into two bands of 170 kDa and 190 kDa on SDS/ PAGE (Fig. 5A). Under reducing conditions the faster migrat- ing band disappears and instead two bands of 90 and 110 kDa are observed. This is more clearly demonstrated by two- dimensional gel electrophoresis in which the 190-kDa chain can be identified as the A-chain portion in fragment C8 - 9 and the 170-kDa chain as the disulfide-linked dimer of the corresponding B-chain segments. No clear assignment of the 90-kDa and 110-kDa bands to segments of the B1 and B2 chains can be made. The B2 chain of fragment E8 has a lower mobility than the B1 chain (Fig. 5A) suggesting that the same relative mobilities may hold true for the B-chain segments of fragment C8 - 9.

Shape and conformation of the long-arm fragment C8 - 9

Already in whole cathepsin G digests and after removal of short-arm structures by centrifugation (Fig. 4), fragment C8 - 9 could be recognized by electron microscopy as a frag- ment of the long arm of laminin consisting of a long flexible rod and a globular terminal domain. For quantification, the purified fragment was visualized by rotary shadowing and negative staining (Fig. 6) and its dimensions were evaluated. For the length of the arm identical values ( I = 72 f 6 nm) are obtained from negatively stained and shadowed particles. Furthermore the same value is found when dimensions are evaluated for C8 - 9 seen in cathepsin G digests before purifi-

275

Fig. 4. Rotary shadowing electron microscopy of a mixture offragmenis C8-9, C3 and C1-4. (a) The supernatant of the cathcpsin G digest (30 min at 37°C in the presence of CaZ+) after centrifugation, which contained fragments C8-9 and C3 (sec Fig. 3), was inspected without further fractionation. (b) Selected particles of short-arm structures (fragment C1 -4) observed in the same cathepsin G digest after centrifugation. The bar corresponds to 50 nm

-ME A

-L

B +ME C8-9 B1B2

C8-9 A 7 1

E 8 A -

E 8 B1-B2-

-N E 8 A -

C8-9 A -

C8-9 E l - C8-9 62-

E8 C8-9 LN E8 C8-9 LN

Fig. 5. Polypeptide composition of isolated fragment C8-9. 3-22% gradient SDSjPAGE (A) was performed with (+ ME) or without (-ME) prior reduction by 2-mercaptoethanol. Laminin-nidogen complex and fragment E8 of laminin were used as references. Their band patterns were identified as follows: L = laminin, N = nidogen, ESA, E8B1 and E8B2 = segments of the A, B1 and B2 chains contained in fragment E8, E8BZ-B2 = disulfide linked dimer of the B-chain segments. Fragment (28-9 was analyzed by two-dimensional SDS/PAGE [B] under non-reducing conditions in the first (vertical) direction and under reducing conditions in the second (horizontal) direction. Designations of chains are by analogy with fragment E8

27 6

Fig. 6. Electron micrographs of the long-arm fragment C8-9 and the short-armfragment Cl -4 . Isolated and purified fragment C8-9 was visualized after rotary shadowing (a) or negative staining (b). Typical fields are shown from which the dimensions and position of kinks were evaluated. Selected negatively stained particles are shown at higher magnification (c) for better visualization of the substructure of the terminal globe. The bar corresponds to 50 nm

cation. For negatively stained particles the diameter of the arrowhead-shaped terminal globular domain were also evalu- ated. Its largest diameter is 8.3 1.6 nm. For many particles three globular substructures of 4 & 0.6 nm diameter each are resolved (Fig. 6c). The average diameter of the rod-like region is 3 1 nm. Often kinks of variable angles are visible in the rod-like regions of fragment C8 - 9. Kinks occur everywhere along the arm with a certain frequency giving rise to a flexible appearance of the arms but a strong clustering is observed at

a distance of 32 _+ 4 nm from the junction of the terminal globular domain and the rod-like region. A model of fragment C8 - 9 which is based on electron microscopic data is shown in Fig. 7 and the position of the prominent kink is indicated by a star. The electron microscopically observed size and shape of fragment C8-9 was confirmed by solution data. Equilibrium ultracentrifugation of the native fragment yields a molecular mass of 360 & 40 kDa. The fragment sediments with a single profile at s20,w = 7.2 S. The high frictional ratio

277

1-L

C 8-9

0 3 n m -

h n m q

t

i--J

3 Fig. I . Structural model of laminin with localizutions and dimensions of frugments C1- 4 , C8 - Y and C3. The domain model of laminin is based on biochemical, physical and sequence information (for reviews see [I, 311). Designations of domains by roman numbers i s according to [12]. Cys-rich rod domains in the short arms are designated by symbols S and the triple-coiled-coil region (domain 1-11) of the long arm by parallel straight lines. In the B1 chain the cr-helical coiled-coil domains are interrupted by a small Cys-rich domain a. Disulfide bridges are indicated by thick bars. The primary cleavage sites of cathepsin G are marked by arrows. Regions of the molecule corre- sponding to fragments 1-4, 8 -9, 8, and 3 are indicated. A model of fragment C8-9 was drawn with the dimensions and the position of a flexible site (star) as determined by electron microscopy

flJ, = 2.4 which follows from these values confirms the very asymmetric shape of fragment C8 - 9.

CD spectra of fragment C8 - 9 (not shown) resembled those observed for fragment E8 [35] but a higher negative amplitude ([O],,, = - 19900 deg cm2/dmol) indicated an a- helix content of 73% for C8-9 as compared to 45% for E8. The higher a-helix content corresponds to the larger rod-like coiled-coil region in fragment C8 - 9. The increase in length however does not lead to an increase of thermal stability. This was indicated by thermal melting profiles as monitored by circular dichroism at 220 nm which closely resembled those previously reported for fragment E8 [35]. Protein sequencing of the C8 - 9 A yielded no unique N-terminal sequence in- dicating the presence of several cathepsin-G-susceptible sites in the region indicated in Fig. 7.

Shape and dimension of fragment CI - 4 So far it was not possible to isolate native fragment C1-4

from the cathepsin G digests because of its insolubility. We

therefore had to rely on a relatively small number of non- associated C1- 4 particles seen in electron micrographs of the digestion mixture (Fig. 4). These show a high degree of preservation of the entire short-arm structures of laminin in striking contrast to fragments of this region of the molecule prepared by other proteases 113, 241. Clearly the globular regions and the full length of all three short arms are preserved (Fig. 4). In agreement with sequence data which indicated different numbers of Cys-rich domains and a third globular domain in the short arm formed by the A chain (Fig. 7), the three arms appear with different lengths. At the longest arm, consistently three instead of two globular domains are visible. These were positioned at 12, 33 and 48 nm from the site of connection of the three arms. The total length of the arm with three globular domains was 48 A 4 nm and that of the two shorter arms was 34 f 4 nm. All distances were measured between the center of the cross and the center of the terminal globe. The values compare well with the published average length of all three short arms of 36 6 nm 1111.

DISCUSSION

In the initial phase of the reaction in the presence of Ca2+ cathepsin G dissects laminin essentially into two halves : the intact short-arm structures C1-4 and a long-arm fragment C8 - 9. The latter comprises the entire triple-coiled coil do- main but is truncated at the C-terminus by removal of the small fragment C3. The two cleavage sites, one near the con- nection between the three short arms and the other somewhere between subdomains G3 and G4 (arrows in Fig. 7), define the most susceptible sites in the laminin molecule when its native structure is stabilized by calcium at physiological concen- tration. Leucocyte elastase and porcine pancreatic elastase [13,22] apparently cleave at the same sites but, less selectively, at other sites of the molecule also. Cathepsin G and leucocyte elastase, which are both present in the azurophile granules of polymorphonuclear leucocytes 121 J, exhibit a higher activity than porcine elastase from pancreas. As revealed from the time course of proteolytic digestion, this latter enzyme was 5 - 10-fold less active than leucocyte elastase. Neutral serine proteases released from polymorphonuclear leucocytes are involved in localized breakdown of extracellular matrix and cell attachments which may constitute a possible step in the mechanism of transition from carcinoma in situ to invasive carcinoma [18, 191.

We did not observe degradation of laminin by thrombin even at an enzyme/substrate mass ratio of 1 : 100. This agrees with observations of another laboratory (R. Timpl, personal communication) but not with the published work of Rao et al. [36] in which a large fragment similar to C1 -4 was ob- served after thrombin treatment. The fragment patterns after cathepsin G treatment obtained by the same group [38] are difficult to compare with those of the present work because they were monitored after reduction and interpreted in terms of an earlier laminin model. The truncation of short arms which was observed by these authors may be explained by lack of calcium in the digestion mixture but we have no expla- nation for the observed loss of the long arm.

Calcium protects the short arm structures of laminin against the proteolytic attack by cathepsin G. Only in its absence are the short arms fragmented and fragment C4, comprising the outer region of one of the arms (Fig. 7), is released. T h s finding substantiates earlier studies [22] in which larger fragments were obtained from the central part

278

of laminin in the presence of Ca2+ than in its absence. These studies were performed with porcine pancreatic elastase and no assignment of the stabilization to a specific region in the short-arm structures was possible. However, the concentra- tion dependence observed for the stabilization against pancre- atic elastase [22] matches that observed for cathepsin G within a factor of 2. It may therefore be concluded that both enzymes probe for a stabilizing effect in the region of fragment 4.

The fragment 4 region in laminin is not only involved in calcium-dependent stabilization of laminin but also in calcium-dependent self-assembly. This follows from the high aggregation tendency of intact fragment C1- 4 as compared to fragment C1' which lacks this region. Calcium-induced associatian of fragment C l - 4 and of intact laminin - nidogen complex leads to large aggregates which can be quan- titatively pelleted by low-speed centrifugation, whereas frag- ment C1' stays in the supernatant under the same conditions. Previous experiments [22] with a similar fragment El ' pre- pared by pancreatic elastase have shown that aggregates formed by such fragments are comparatively small and cannot be detected by turbidity development. The importance of the role of the fragment 4 region is further emphasized by unpub- lished demonstration of binding of isolated fragment 4 to laminin and strong inhibition of laminin by addition of this fragment in soluble form (J. Schittny and P. Yurchenco, per- sonal communication). Already in the work of Yurchenco et al. [9] it was concluded that laminin self-assembly proceeds mainly by interactions between the tips of the short arms. It may now be specified that the fragment 4 region is involved and is responsible for the Ca2+ dependence of the process. Binding of calcium to the laminin - nidogen complex, frag- ment El ' and to fragment E8 of the long arm has been mea- sured directly [24]. It was noticed that the number of Ca2+ ions which can bind to the fragments only accounts for a relatively small fraction of Ca2+ bound to the intact complex. The situation may be further complicated by the presence of nidogen which binds to the short arm structures of laminin. Its C-terminal fragments remain tightly associated with frag- ments of this region [I 51. Accordingly the 100-kDa fragment of nidogen coprecipitated with fragment C1-4. Attempts to isolate fragment C1- 4 and purify it from nidogen fragments have been unsuccessful so far. Possible association of nidogen fragments also renders an interpretation of electron micro- graphs of fragment C1-4 difficult. It is clearly apparent that one of the three short arms is longer than the other two. This arm must therefore be assigned to the one formed by the A chain which, according to sequence data, contains additional Cys-rich repeats and a third globular domain not present in the B chains (Fig. 7). In some cases the third globular domain is visible 12 nm from the center of the cross.

Clearly fragment C8 - 9 does not associate. It stays mono- meric under conditions in which fragment C1-4, laminin or the laminin - nidogen complex are highly aggregated. It shares this property with fragment E8 which comprises the lower half of the long arm. Fragment C1-4 has lost the fragment 3 region comprising subdomains G4 and G5 (Fig. 7). The link region between domains G3 and G4 is one of the two proteolytically most susceptible sites in laminin and there is evidence (J. Engel, Ch. Fauser and R. Timpl, unpublished results) that subdomains G4 and G5 are also missing in frag- ment E8. This may explain the apparent discrepancy between initial observation of the terminal globe at the long arm being involved in self-assembly [23] and the failure to demonstrate such interactions with fragments C8 - 9 (present study) and E8 [22]. The molecular mass of fragment C8 - 9 as determined

by sedimentation equilibrium (360 & 40 kDa) is significantly higher than the polypeptide molecular mass of 260 kDa calcu- lated from sequence data [12], assuming cleavage sites as indi- cated in Fig. 7. The difference may be explained by a high degree of glycosylation. Laminin contains 13% (by mass) carbohydrates present mainly as N-linked oligosaccharides [39]. About 70% of its potential N-glycosylation sites are located in the long arm [12, 341. Assuming equal occupation of all sites, glycosylation of fragment C8 -9 would account for an additional molecular mass of 80 kDa, in good agree- ment with the experimental value. An interesting structural feature of fragment C8 - 9 is the flexible site detected in the middle part of its triple-coiled-coil domain. This region corre- sponds to a susceptible site cleaved by many proteases includ- ing porcine pancreatic elastase and is responsible for the cre- ation of the well known fragment E8 by this enzyme [35]. No kink or other structural detail was detectable in the region of the domain LX (Fig. 7) suggested by sequence data [12]. Since fragment C8 - 9 contains the entire triple-coiled-coil domain by which the three chains of laminin are held together it is currently used for studies of the specificity of chain assembly (I. Hunter, T. Schulthess, M. Bruch, K. Beck and J. Engel, unpublished results). The new fragment may also be used in exploring the properties and functions of the upper-long-arm region, which has so far not been available as a fragment.

We thank Charlotte Fauser for performing the electron mi- croscopy and Ariel Lustig for analytical ultracentrifugation. We also thank Drs K . Beck, M. Paulsson and R. Timpl for constructive criticism and discussions. Financial support by the Swiss National Science Foundation (grant 31 -9088.87) is gratefully acknowledged. Protein sequencing of the C8-9 A was kindly performed by Dr R. Deutzmann, University of Regensburg.

REFERENCES

1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14. 15.

16.

Martin, G. R. & Timpl, R. (1987) Annu. Rev. Sell Bid. 3, 57-

Martin, G. R., Timpl, R. & Kiihn, K . (1988) Adv. Protein Chem.

Timpl, R . & Dziadek, M. (1986) Int. Rev. Exp. Pathol. 2Y, 1 -

Aumailley, M., Nurcombe, V., Edgar, D., Paulsson, M. & Timpl,

Nurcombe, V., Aumailley, M., Timpl, R. & Edgar, D. (1 989) Eur.

Terranova, V. P., Rohrbach, D. H. & Martin, G. R. (1980) Cell

Edgar, D., Timpl, R. & Thoenen, H. (1984) EMBO J . 3, 1463- 3468.

Panayotou, G., End. P., Aumailley, M., Timpl, R. & Engel, J . (1989) Cell56, 93-101.

Yurchenco, P. D., Tsilibary, E. C., Charonis, A. S. & Furthmayr, H. (1985) J . Biol. Chem. 260, 7636-7644.

Yurchenco, P. D., Tsilibary, E. C., Charonis, A. S . & Furthmayr, H. (1986) 1. Histochem. Cytochem. 34, 93-102.

Engel, J., Odermatt, E., Engel, A,, Madri, J . A,, Furthmayr, H., Rohde, H. and Timpl, R. (1981) J . Mol. Biol. 150, 97- 120.

Sasaki, M., Kleinmann, PI. K., Huber, H., Deutzmann, K. & Yamada, Y. (1988) J . Biol. Chem. 263, 16536-16544.

Ott, U., Odermatt, E., Engel, J., Furthmayr, H. & Timpl, R . (1982) Eur. J. Biochem. 123, 63-72.

Doolittle, R. F. (1985) Trends Biochem. Sci. 10, 233-237. Mann, K., Deutzmann, R. & Timpl, R. (1988) Eur. J . Biochem.

Timpl, R., Paulsson, M., Dziadek, M. & Fujiwara, S. (1987)

85.

39, 1 - 50.

112.

R. (1987) J . Biol. Chem. 262, 11 532 - 1 1 538.

J . Biochem. 180, 9 - 14.

22,719-726.

178, 71 - 80.

Methods Enzymol. 145, 363 -391.

279

17. Russo, R. G., Liotta, L. A., Thorgeirsson, U., Brundage, R. &

18. Tryggvason, K., Hoyhtya, M. & Salo, T. (1987) Biochim. Biopys.

19. Mullins. D. E. & Rohrlich (1983) Biochim. Biuphys. Acta 6Y5,

20. Janoff, A. (1985) Annu. Rev. Med. 36, 207-219. 21. Haveman, K. & Gramse, M. (1984) Adv. Exp. Med. 167, 1-20. 22. Paulsson, M., Saladin, K. & Landwehr, R. (1988) Eur. J . Biochem.

23. Charonis, A. S., Tsilibary, E. C., Saku, T. & Furthmayr, H. (1 986)

24. Paulsson, M. (1988) J . Biol. Chern. 263, 5425 - 5430. 25. Paulsson, M., Aumailley, M., Deutzmann, R., Timpl, R., Beck,

K. & Engel, J. (1987) Eur. J . Biochem. 166, 11 -19. 26. Baugh, R. J. & Travis, J. (1976) Biochemistry 15, 836-842. 27. Miletich, J. P., Broze, G. J . & Majerus, R. W. (1980) Anal.

28. Rosenberg, R. D. & Damus, P. S. (1973) J . Biol. Chem. 248,

29. Laemmli, U. K. (1970) Nature 227,680-6685.

Schiffmann, E. (1981) J . Cell Biol. 91, 459-467.

Acta 907, 191 -217.

177 -214.

177,477-481.

J . Cell Biol. 103, 1689-1697.

Biochern. 105, 304-310.

6490-6505.

30. Chen, Y. H., Jang, Y. T. & Martinez, H. M. (1972) Biochemistry

31. Engel, J. & Furthmayr, H. (1987) Methods Enzymol145, 3-78. 32. Tanford, C. (1961) in Physical chemistry of macromolecules, John

Wiley, New York, London. 33. Paulsson, M., Deutzmann, R., Dziadek, M., Nowak, H. & Timpl,

R. (1986) Eur. J . Biochem. 156,467-478. 34. Deutzmann, R., Huber, J . Schmetz, K. A,, Oberbiumer, I. &

Hartl, L. (1988) Eur. J . Biochem. 177, 35-45. 35. Paulsson, M., Deutzmann, R., Timpl, R., Dalzoppo, D.,

Odermatt, E. & Engel, J. (1985) EMBO J . 4 , 309-316. 36. Rao, C. N., Margulies, I. M. K., Tralka, T. S. Terranova, V. P.,

Madri, J. A. & Liotta, L. A. (1982) J . Biol. Chern. 257, 9740- 9744.

37. Dziadek, M., Clements, R., Mitrangas, K., Reiter, H. & Fowler, K. (1988) Eur. J . Biochem. 172, 219-225.

38. Rao, C. N., Margulies, M. K., Goldfarb, R. H., Madri, J. A., Woodley, D. T. & Liotta, L. A. (1982) Arch. Biochem. Biophys.

39. Fujiwara, S., Shinkai, H., Deutzmann, R., Paulsson, M. & Timpl,

11,4120-4131.

21 9, 65 - 70.

R. (1988) Biochem. J . 252,453-461.