| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Department of Biochemistry, University of Zurich, Zurich, Switzerland;
Institute of Laboratory Animal Research, University of Zurich, Zurich, Switzerland; and
Neurotune AG, Schlieren, Switzerland
2Correspondence: University of Zurich, Department of Biochemistry, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland. E-mail: peter.sonderegger{at}bioc.uzh.ch
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Key Words: serine peptidase • extracellular matrix • cognitive function • nervous system
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
. Neurotrypsin is predominantly expressed in the neurons of the cerebral cortex, the hippocampus, and the amygdala (2
, 3)
. By immuno-electronmicroscopy, it was localized in the presynaptic membrane and the presynaptic active zone of both excitatory and inhibitory synapses (1)
. The 4 bp deletion in the seventh exon of the individuals suffering from neurotrypsin-dependent mental retardation results in a complete functional inactivation, as it causes the translation of a truncated protein without catalytic domain (1)
. The pathophysiological phenotype and the age of onset of the disease indicate that neurotrypsin does not play a critical role for early neural development and for the formation of synapses. Rather, neurotrypsin may be crucial for the reorganization of synapses and neuronal circuits occurring in later stages of development and for adult synaptic plasticity, which are required to establish and maintain higher cognitive functions.
The proteoglycan agrin exists in several isoforms that are generated by alternative splicing (4)
. These encode both secreted extracellular matrix proteins and type II transmembrane proteins with a very short N-terminal cytoplasmic segment. The secreted variant is the predominant form expressed by motoneurons. It is transported down the motor nerve and released from the nerve ending into the synaptic cleft of the neuromuscular junction (NMJ), where it becomes an essential component of the synaptic basal lamina. Motoneuron-derived agrin bears an insert of 8, 11, or 19 (8+11) amino acids at the B/z splice site located in the C-terminal LG3 domain and exhibits synapse-promoting activity, as demonstrated in vitro by the aggregation of acetylcholine receptors on cultured myotubes (5)
. Recent observations indicate an essential synapse-protecting function for motoneuron-derived agrin (6
, 7
, 8)
. The transmembrane isoform of agrin predominates in the brain, particularly in neurons, which are typically not surrounded by the extracellular matrix. The molecular mechanisms of agrin function in the central nervous system (CNS) appear to be fundamentally distinct from those found at the NMJ. In the CNS, the C-terminal LG3 domain of agrin binds to and inhibits the 
3-subtype of the Na+/K+-ATPase, thereby evoking depolarization at CNS synapses (9)
. Other studies indicate that clustering of agrin by antibodies (10)
and the overexpression of agrin in neurons (11)
stimulate the formation of filopodia, which in turn provide the basis for the formation of new synapses (12
, 13)
.
To analyze the function of neurotrypsin at the NMJ, we generated transgenic mice overexpressing neurotrypsin in motoneurons and found a neuromuscular phenotype that strongly resembled the phenotype found in the diaphragm of agrin-deficient mice (Bolliger, M. et al., unpublished observations). Four days after the full induction of the transgene, the majority of the previously established NMJs in the endplate band had disappeared and the motor nerves had grown out beyond the margins of the endplate band, forming numerous immature NMJs along their trajectory. The striking similarity of the abnormal innervation patterns resulting from neurotrypsin overexpression and agrin deficiency, together with the coincident expression of neurotrypsin and agrin in motoneurons, suggested agrin as a target of neurotrypsin. Here, we report on the identification of two homologous neurotrypsin-dependent cleavage sites in agrin. Agrin cleavage by neurotrypsin generates fragments of 90 and 22 kDa and separates the synapse-regulating activity of agrin from its N-terminal moiety. Agrin cleavage is not found in neurotrypsin-deficient mice. To explore the interaction of agrin and neurotrypsin in detail, we produced and purified recombinant full-length neurotrypsin and developed activity assays for each of the cleavage sites. Mutagenesis of the amino acids flanking the scissile bonds of both cleavage sites indicated a high substrate selectivity of neurotrypsin vs. agrin, supporting a role of the neurotrypsin-dependent cleavage of agrin for the establishment and the maintenance of cognitive functions.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Cleavage of agrin by coexpression with neurotrypsin in HeLa cells
For expression of neurotrypsin, the coding region of full-length human neurotrypsin (CAA04816) was inserted into the pcDNA3.1 vector (Invitrogen, Carlsbad, CA, USA). Catalytically inactive neurotrypsin (S/A) was generated by mutating Ser825 to Ala by means of overlap extension polymerase chain reaction (PCR). For expression of membrane-bound full-length agrin, the coding sequence ranging from Met1 to Pro1959 of rat agrin (P25304, splice variant y4z8) was inserted into pcDNA3.1. HeLa cells were cultured to 60–80% confluency in 12-well culture plates (Corning, Schiphol, The Netherlands) with Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS; Biochrom AG, Berlin, Germany). Polyethylenimine (PEI) was used as transfection reagent as described previously (14)
. The transfection mixture was removed 4–6 h after transfection by being washed once with phosphate-buffered saline (PBS). After 48 h in DMEM without FCS, the medium was harvested and the cells were lysed in 150 mM NaCl, 0.5 mM EDTA, 10% glycerol, and 1% Triton-X-100 in 20 mM Tris-HCl, pH 7.4. For the analysis of cleavage products, supernatants and cell lysates were separated on a 4–12% NuPAGE gel (Invitrogen) and analyzed by immunodetection.
Determination of the neurotrypsin-dependent cleavage sites of agrin
Neurotrypsin was coexpressed in HEK293T cells with either a transmembrane full-length form of rat agrin (splice variant y4z8) or the C-terminal fragment of agrin comprising the LG2, EG4, and LG3 domain using PEI transfection (14)
. After 3 days, the supernatants were harvested. For the purification of the 90 kDa fragment, 1l supernatant of cells transfected with full-length agrin was loaded onto a heparin Sepharose CL-6B column (GE Healthcare, Uppsala, Sweden) equilibrated with 150 mM NaCl in 20 mM Tris-HCl, pH 7.5. Proteins were eluted in a gradient from 150 mM to 1 M NaCl in 20 mM Tris-HCl, pH 7.5. The 22 kDa fragment was purified via its C-terminal StrepTag from 24 ml medium of cells transfected with the C-terminal fragment of agrin with a StrepTactin column (IBA, Göttingen, Germany) according to the manufacturer’s recommendation. N-terminal sequencing was carried out as detailed below.
Peptide analysis
For liquid chromatography electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS) analysis, the protein samples were separated on 4–12% NuPAGE gels, electrotransferred onto a polyvinylidene difluoride (PVDF) membrane, and stained with Coomassie brilliant blue G250 (Sigma-Aldrich). The analysis was performed on a Q-Tof Ultima API (Waters, Milford, MA, USA) at the Functional Genomics Center Zurich. For N-terminal sequencing by Edman degradation, the peptide samples were separated on 4–12% NuPAGE gels, electrotransferred onto a PVDF membrane and analyzed on a Procise 492 cLC Sequencer (Applied Biosystems, Foster City, CA, USA) at the Functional Genomics Center Zurich.
In vitro activity assay for neurotrypsin
To assess its proteolytic capacity, neurotrypsin (0.0375 µM) was incubated with either -agrin (0.6 µM) or ß-agrin (1 µM) for 1 h at 37°C in a buffer containing 150 mM NaCl, 5 mM CaCl2, and 0.1% PEG6000 in 20 mM MOPS, pH 7.0. The products of -agrin were separated on 12% NuPAGE gels and the ß-agrin products on 4–12% NuPAGE gels. The gels were stained with SYPRO Ruby (Molecular Probes, Eugene, OR, USA) and quantified by fluorescent densitometry using the imaging system LAS-3000 and the AIDA software (Fujifilm, Tokyo, Japan).
Assessment of the substrate preference of neurotrypsin with mutated agrin substrates
Mutations of the amino acids flanking the scissile bonds of the - and ß-cleavage sites of the agrin substrates were introduced by overlap extension PCR. Mutated - and ß-agrin forms were produced and purified as described in the Supplemental Materials. For activity measurements, neurotrypsin (0.0375 µM) was incubated with ß-agrin (0.3 µM) or -agrin (0.6 µM). At time points between 0 and 360 min for ß-agrin and 0 and 120 min for -agrin, samples were taken and the reaction was stopped with sample buffer (10% SDS, 30% glycerol, 0.6 M DTT, 100 mM EDTA, and 0.35 M Tris, pH 6.8). The cleavage products were quantified. Initial velocities were determined by linear regression in the early phase of the progress curve. Pseudo first-order rate constants were determined by fitting the integrated Michaelis-Menten equation to the whole progress curve of product release. Calculations were performed with the SigmaPlot 9.0 software.
Docking of agrin cleavage sites into the active site of neurotrypsin
The sequence of human neurotrypsin (CAA04816) was retrieved from the SwissProt database (15)
. The catalytic domain of human neurotrypsin, spanning over 251 residues, was taken as a query to search the pdb database using PSI-Blast (16)
to find templates for homology modeling. Based on preference for high resolution X-ray structures, gapless backbones, and high sequence identity/similarity, human factor XI (1XXD, chain A), sheep ß-acrosin (1FIW), and human urokinase-type plasminogen activator (uPA; 1GJ7) were chosen. Sequence-structure alignment and homology model were obtained using the Prime module (17)
of the Schrödinger program suite (LLC) with default parameters. The hexapeptides from P4 to P2' of the - and ß-cleavage sites of human agrin were generated and placed into the active site of neurotrypsin using the Schrödinger program suite. The initial placement was obtained by superimposing the cleavage sites on the ligand (3BP) of the crystal structure of human factor VIIa (1WTG), which has been superimposed on the homology model of neurotrypsin. 3BP has a Gln at P2 and, therefore, geometrically resembles the - and ß-cleavage sites of agrin with a Glu at P2. The initial poses of the - and ß-cleavage sequences were minimized within the active site of neurotrypsin with a convergence criterion of the gradient of 0.0418 kcal/mol Å using the OPLS_2005 force field with a constant dielectric electrostatic treatment (dielectric constant=1). A nonbonded cutoff distance of 12 Å was applied. Residues at a distance >4.5 Å around the peptides were kept fixed.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
A). The same signal was also found in lysates of cells cotransfected with inactive neurotrypsin. When wild-type neurotrypsin was cotransfected with agrin, no agrin signal was found in the cell lysate. However, new bands of 110, 90, and 22 kDa were found in the culture supernatant. None of these agrin fragments were found in the supernatants of cotransfections with the inactive form of neurotrypsin. Together, these results demonstrated that proteolytically active neurotrypsin cleaved agrin, resulting in the release of three C-terminal fragments into the culture supernatant.
|
Neurotrypsin-dependent cleavage of agrin occurs in vivo and is not found in neurotrypsin-deficient mice
We investigated different tissues from wild-type and neurotrypsin-deficient mice for the presence of agrin fragments. As shown in Fig. 1B
, strong immunoreactivity for full-length agrin was detected in brain, kidney, and lung of wild-type mice. In these tissues, we also found the 90 kDa cleavage product of agrin. The 22 kDa cleavage product was readily detectable in brain and kidney, but it was absent in the lung. In neurotrypsin-deficient mice, no agrin immunoreactivity was detectable at 22 or 90 kDa. These results demonstrate that agrin cleavage by neurotrypsin occurs in vivo, and, furthermore, the absence of cleavage products in neurotrypsin-deficient mice indicates that agrin cleavage strictly depends on neurotrypsin.
The two neurotrypsin-dependent cleavage sites of agrin are homologous and evolutionarily conserved
To determine the neurotrypsin-dependent cleavage sites of agrin, the 90 and 22 kDa fragments were generated in HEK293T cells cotransfected with the cDNAs encoding agrin and neurotrypsin. The fragments were purified chromatographically and subjected to N-terminal sequencing. We determined ASXYNSPLGXXSDGK for the 90 kDa fragment and SVGDLETLAF for the 22 kDa fragment. Therefore, one cleavage site is located between the first S/T-segment and the SEA domain, C terminally of Arg995 in the sequence PIER
ASCY (Fig. 2
A). The second cleavage site was localized between the fourth EGF-like and the LG3 domain, C terminally of Lys1754 in the sequence LVEKSVGD (Fig. 2A
). Amino acid numbers refer to membrane-anchored rat agrin (P25304; splice variant x4y0). We termed the scissile bond between R995 and A996 as the -cleavage site and the scissile bond between K1754 and S1755 as the ß-cleavage site. An alignment of the two cleavage sequences of rat agrin with corresponding segments of various vertebrate species showed a stringent conservation of the amino acids flanking the cleavage sites (Fig. 2B
) (18)
. The site had a strictly conserved Arg at P1 and a Glu at P2. The ß site had a strictly conserved Lys at P1 and a Glu at P2. The residues at the P3 position were more variable, while predominantly apolar residues were found at P4 of both sites. On the C-terminal side of the scissile bond, the P1' and P2' residues are well conserved at each site, with predominantly an Ala-Ser sequence at the and a Ser-Ala sequence at the ß site. Well-conserved but distinct residues were found at P3' and P4' of the two sites. In summary, the consensus sequences of the and ß site exhibit an identical profile, with a strict conservation of the P1 and P2 residue, a variable occupation of the P3 position, and a high degree of conservation at P4, P1', P2', and P3'. The gross profile of the individual consensus sequences of the and ß sites is maintained when a combined consensus sequence of the two cleavage sites is generated (Fig. 2C
), confirming the homology between the individual cleavage sequences at the and ß site.
|
Expression and purification of neurotrypsin
To study the catalytic activity of neurotrypsin, a cDNA construct encoding human full-length neurotrypsin (corresponding to amino acids 1–875), was generated and constitutively expressed in J558L cells. For the purification of neurotrypsin, a four-step procedure was established, consisting of an initial concentration by cross-flow filtration, followed by heparin affinity, hydrophobic interaction, and immobilized metal affinity chromatography (IMAC) (see Supplemental Materials). The SDS-PAGE analysis of the final product is shown in Fig. 3
. On the silver stained gel run under reducing conditions (Fig. 3B
), three major bands were found at 95, 63, and 32 kDa. Three additional bands were found at 105, 83, and 54 kDa. Analyses by LC-ESI-MS/MS and N-terminal sequencing identified all bands as either full-length neurotrypsin or proteolytic fragments of it (Fig. 3A, B
; See Supplemental Fig. S1 and Table SIV). Band 6 corresponds to the protease domain, generated by cleavage between Arg630 and Ile631 of the sequence HRRQKRIIGG. Proteolytic cleavage at this position has been predicted as the zymogen activation site of neurotrypsin (3)
based on homology with other serine proteases. The relative molecular mass (Mr) of 32 kDa on SDS-PAGE and the calculated peptide mass of 27,488 Da allow for an N-linked oligosaccharide of 4.5 kDa at the predicted glycosylation site of the protease domain (Asn683). Bands 1–3 represent the single-chain zymogenic form of neurotrypsin, and bands 4 and 5 reflect the N-terminal noncatalytic fragment of the activated two-chain form of neurotrypsin. Band 1, with a Mr of 105 kDa, corresponds to the full-length, single-chain form of neurotrypsin, starting with Phe21, as previously proposed (3)
. Two predicted sites for N-linked glycosylation (Asn26 and Asn683) could explain the difference between the observed Mr of 105 kDa and the calculated peptide mass of 97,009 Da. Band 2, with an Mr of 95 kDa and starting with Ala71, corresponds to a slightly shortened single-chain form of neurotrypsin. The difference of 10 kDa in the apparent Mr is explained by the size of the released peptide (calculated peptide mass of 5,813 Da) and a putative associated N-linked oligosaccharide of 
4 kDa at Asn26. We could not obtain the N-terminal sequence of band 3, but its identity as a fragment of neurotrypsin was unequivocally confirmed by MS/MS. With a Mr of 83 kDa and the inclusion of the peptide SPPAS, it is most likely a nonglycosyated form of the single-chain variant starting at Ser122 and ending with the C-terminal Leu875 (calculated mass of 83,540 Da). Band 4, with 63 kDa and starting with Ala71, is the major N-terminal fragment generated by cleavage at the activation site (calculated mass of 61,660 Da). Band 5, with 54 kDa and starting with Ser122, reflects a shorter variant of the N-terminal noncatalytic fragment generated by cleavage at the activation site (calculated mass of 56,070 Da).
|
These results characterize the purified neurotrypsin as a mixture of the activated two-chain form (bands 4–6), as generated by proteolytic cleavage at the predicted activation site after Arg630 (Arg15 in chymotrypsin numbering) (19)
and the zymogenic single-chain form (bands 1–3). Two additional proteolytic cleavages in the N-terminal region of neurotrypsin result in a staggered N terminus and explain the observed complexity of the band pattern.
To measure the ratio between single- and two-chain neurotrypsin, we used a protease domain-specific antibody for Western blotting (WB-G87; Fig. 3B
). This antibody yields equivalent immunoreactivity signals for the isolated band of 32 kDa representing the two-chain form of neurotrypsin (band 6) and the catalytic domains associated with the single-chain forms of neurotrypsin (bands 1–3). Quantification revealed that 75% of the purified neurotrypsin was in the two-chain form.
Analyses of purified neurotrypsin by nonreducing SDS-PAGE indicated that activation cleavage does not result in dissociation of the catalytic domain from the noncatalytic part (Fig. 3C
). Both silver staining and Western blotting indicated that a predominant fraction of neurotrypsin at apparent Mr between 90 and 75 kDa remained covalently linked by a disulfide bond, most likely between Cys619 and Cys750 (Cys1 and Cys122 in chymotrypsin numbering), as suggested by homology of the activation peptide of neurotrypsin with that of tissue-type plasminogen activator (tPA), chymotrypsin, and elastase 2 (20)
.
Agrin cleavage at both sites is due to direct proteolytic activity of neurotrypsin
To investigate each cleavage site individually, we generated and purified two soluble fragments of rat agrin, each containing only one cleavage site: -agrin contained the site and ß-agrin contained the ß site (Fig. 4
A, B; see Supplemental Materials). The substrates were incubated with purified neurotrypsin (see Supplemental Materials) for 1 h, and the products were analyzed by SDS-PAGE and SYPRO Ruby staining. The C-terminal cleavage product of -agrin run at 130 kDa, while the highly glycosylated N terminus smeared between 150–350 kDa (Fig. 4C
). The cleavage products of ß-agrin run at 22 and 24 kDa (Fig. 4D
). N-terminal sequencing of the bands confirmed the cleavage sites previously found in the cell culture assay. The C-terminal cleavage product of the -agrin substrate migrating at 130 kDa had the N-terminal sequence ASXYNS, while the N-terminal sequence of the 22 kDa cleavage product of ß-agrin was SVGDLE. In summary, the in vitro digestion experiments with purified recombinant neurotrypsin and agrin fragments showed that cleavage of agrin at both the - and the ß-cleavage sites was due to direct proteolytic activity of neurotrypsin.
|
We also tested chromogenic and fluorogenic peptide substrates based on the amino acid sequence of - and ß-agrin. No substantial activity of neurotrypsin vs. these substrates was found (see Supplemental Materials).
Enzymological characterization of neurotrypsin
The enzymatic activity of neurotrypsin was characterized in an in vitro assay with ß-agrin as substrate. As shown in Fig. 4E
, neurotrypsin activity was restricted to pH values between 5.5 and 10.0, with maximal activity between 7.0 and 8.5. Calcium was found to be essential for neurotrypsin activity. The half-maximal substrate turnover was reached at concentrations of 0.5 mM (Fig. 4F
). At a physiological calcium concentration of 2 mM, neurotrypsin activity was 80% and maximal activity was found above 5 mM calcium. Little sensitivity to NaCl was found from 50 to 200 mM (see Supplemental Fig. S2), but at 500 mM the catalytic activity dropped significantly. Similar results for the activity dependence on pH, calcium, and NaCl were obtained for -agrin (data not shown). Based on these results, we chose the following conditions for in vitro assays of neurotrypsin: 150 mM NaCl, 5 mM CaCl2, and 0.1% PEG 6000 in 20 mM MOPS, pH 7.0.
To determine the Michaelis constant Km, the neurotrypsin-dependent cleavage rate of ß-agrin was measured with substrate concentrations between 5 and 100 µM (Fig. 4H
). Substrate concentrations >100 µM did not allow product quantifications on gel. Because initial velocities vs. agrin concentration were linear in the studied range, the Km value for ß-agrin must be considerably >100 µM.
A pseudo first-order rate constant for the cleavage of ß-agrin of 5 x 10–4 s–1 was determined by fitting the integrated Michaelis-Menten equation to the experimental data (Fig. 4G
). The pseudo first-order rate constant k obtained in this way corresponds to the V/Km ratio, where V represents the limiting rate. Dividing this rate constant by the titrated enzyme concentration yields the kinetic constant kcat/Km, defining the catalytic efficiency. Unfortunately, an active site titrant for neurotrypsin was not available. However, we estimated that 75% (37.5 nM) of our neurotrypsin preparation was in the activated two-chain form, while 25% was in the zymogenic single-chain form (Fig. 3C
). Under this assumption, the catalytic efficiency kcat/Km was 1.3 x 104 M–1 · s –1 and comparable to that of other trypsin-like serine proteases with a regulatory function. For example, factor IXa, a crucial regulator of blood coagulation, cleaves and thereby activates factor X with a catalytic efficiency of 1.5 x 104 M–1 · s–1 (21)
.
Among common broadband inhibitors of trypsin-like serine proteases, only 3,4-dichloroisocoumarin efficiently inhibited neurotrypsin. Benzamidine inhibited neurotrypsin only weakly, and PMSF, AEBSF, TLCK, aprotinin, and leupeptin were not inhibitory at all (see Supplemental Materials). Furthermore, selective inhibitors of trypsin, thrombin, uPA, and plasmin were not inhibitory for neurotrypsin (Supplemental Materials). Together, these results indicate a highly selective substrate-binding pocket of neurotrypsin.
Neurotrypsin has an extended substrate binding pocket and exhibits a unique preference for glutamic acid in the P2 position
To assess the substrate selectivity of neurotrypsin, we generated single mutations of the amino acids flanking the scissile bonds of the - and ß-cleavage sites. The amino acids at positions P4 to P3' in ß-agrin and from P5 to P1 in -agrin were substituted by alanine. In addition, the P1 Lys at the ß-cleavage site was mutated to Arg and the Arg at the -cleavage site to Lys. Furthermore, the P3 to P2' positions of the ß site were replaced by the corresponding amino acids of the site. The substrate turnover of each mutant of ß-agrin was quantified, and pseudo first-order rate constants were determined by fitting the integrated Michaelis-Menten equation to our data. The rate constants of the ß-agrin mutants are presented in comparison to the rate constant of wild-type ß-agrin (Fig. 5
A). No substrate cleavage was observed when Lys at the position P1 was replaced by Ala, and a Glu to Ala mutation at P2 reduced the catalytic activity 50-fold. Mutations of P3 and P4 reduced the substrate turnover to 41 and 23%, respectively. The P1 mutation to the Arg and the replacement of the ß site by the recognition sequence reduced the proteolytic activity to 45 and 68%, respectively. Interestingly, substitution from Ser to Ala at P1' of the ß site reduced the cleavage activity to 38%, although an Ala is found at P1' of the site. Mutations at P2' and P3' showed a minor effect with an activity decrease to 97 and 84%, respectively.
|
We also performed double and triple mutations in the P region of the ß site. Double mutations to Ala resulted in a complete resistance to neurotrypsin for P2 + P3 and P2 + P4, while a residual activity of 11% was found for P3 + P4 (Fig. 5B
). The triple mutation P2 + P3 + P4 resulted in a complete resistance to neurotrypsin. Thus, the effects of suboptimal occupancy at multiple subsites were cumulative.
It was not possible to determine the rate constants of neurotrypsin for the much larger and highly glycosylated -agrin mutants, due to interference of the accumulating product bands on the gels at later time points. However, tests with ß-agrin as substrate demonstrated that the initial velocity measurement was equivalent with rate constant determination for assessing substrate preference (see Supplemental Fig. S3). Therefore, we used initial velocity measurements for -agrin-based substrate mutants. As shown in Fig. 5C
, -agrin with an Ala mutation at position P1 was not cleavable. An Ala replacing Glu at the P2 site resulted in a residual activity of 5%. Ala mutations at P3, P4, and P5 of -agrin decreased neurotrypsin activity to 87, 36, and 41%, respectively. Substitution of Arg by Lys at the P1 site decreased neurotrypsin activity to 70%. These results indicated virtually identical kinetic effects of P-site mutations at both cleavage sites.
In addition, the influence of different amino acids at the crucial P2 site was examined for the ß-cleavage site. As shown in Fig. 5D
, substituting Glu by Gln, Asp, and Leu decreased the catalytic efficiency to 20% (23–18%) of the wild-type activity. No cleavage was detectable for the Glu to Lys and the Glu to Ala mutations. The results emphasized the high preference of neurotrypsin for Glu at P2.
Spatial modeling of neurotrypsin and substrate docking reveals an intimate engagement of the P2 glutamic acid side chain with the strongly basic S2 subsite
To obtain a three-dimensional structural correlate of the substrate specificity of neurotrypsin for a Glu at P2, we generated a homology model based on the crystal structures of human factor XI, human urokinase-type plasminogen activator, and sheep ß-acrosin. Hexapeptides extending from P4 to P2' of the - and ß-cleavage sites of agrin were placed into the active site of neurotrypsin, and the initial poses were adjusted based on standard minimization criteria. As demonstrated in Fig. 6
A, C (24)
, the side chain of the P2 Glu of both cleavage sequences forms a salt bridge with the side chain amino group of the Lys60 and with the guanidino group of Arg60A of neurotrypsin forming the roof of the S2 subsite. The effect of the substitution of the P2 Glu by an Asp is shown for the -cleavage site in Fig. 6B
and for the ß-cleavage site in Fig. 6D
. The side chain of the P3 residue is predicted to point toward the solvent consistent with the fact that mutating this position to Ala has only a minor effect on cleavage efficiency. The overall binding mode of both cleavage sites is very similar. The Glu to Asp mutation at the P2 position resulted in the loss of the salt bridge with the side chain amino group of Lys60, while a close contact with the imidazol of His57 of the catalytic triad might unfavorably interfere with catalysis. The calculated Coulombic energy term of the Asp mutation was 16.7 kJ/mol lower than the one obtained with the wild-type Glu at P2. This observation is in accordance with previous studies reporting that artificial substrates with an Asp at position P2 were cleaved very inefficiently by tPA (22)
. Similarly, the "nonspecific" protease trypsin also was least efficient in cleaving substrates with an Asp at P2. In fact, based on structural information obtained from the complex of trypsin and bovine pancreatic trypsin inhibitor (23)
, it was argued that Asp at P2 may be within 2.6–3 Å from His57 of the catalytic triad and therefore may interfere with its function during catalysis.
|
Neurotrypsin-dependent cleavage of agrin is most prominent during late stages of synaptogenesis
We determined the temporal expression pattern of neurotrypsin and the neurotrypsin-dependent agrin fragments in mouse brain samples from embryonic day 16 through the age of 2 years. As shown in Fig. 7
, neurotrypsin expression in the brain can already be observed at prenatal stages. Peak expression levels were found between E16 and P10. During later developmental stages, expression declined toward lower levels in early adulthood. Relatively low, but clearly detectable, expression of neurotrypsin was maintained throughout adult life. The temporal course of agrin cleavage closely resembled the expression pattern of neurotrypsin, with peak levels also between E16 and P10. Lower, but well-detectable levels of the 90 kDa fragment, were found at all adult stages (see also Supplemental Fig. S4). At ages of 60 days and older, no 22 kDa fragment was detected. However, the persistence of the 90 kDa fragment indicated that neurotrypsin-dependent cleavage of agrin persists at both cleavage sites throughout life.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
Neurotrypsin-dependent cleavage of agrin is sequence specific
The two neurotrypsin-dependent cleavage sites of agrin are homologous and highly conserved in evolution. As the most striking characteristic, beyond the strictly conserved basic amino acids at P1, both sites exhibited a strictly conserved Glu at the P2 position. Glu substitution by Ala almost abolished cleavage at both sites. Substitution of Glu at the P2 position of the ß-cleavage site by Gln, Asp, Lys, and Leu reduced cleavage to 23, 18, 3, and 23%, respectively. Thus, the order of preference for S2 subsite occupancy in neurotrypsin is Glu >>> Gln = Leu > Asp >>> Lys = Ala.
The strong preference of neurotrypsin for the two conserved cleavage sequences of agrin is further supported by the fact that in addition to the dominant P1 and P2 residues at least four additional positions (P5 through P3, and P1') have to interact with the substrate-binding pocket of neurotrypsin. Although the individual effects of suboptimal subsite occupancy at the S5–S3 and the S1' subsites were not dramatic, their cumulative effects were considerable and may represent a substantial barrier for nonselective hydrolysis. Thus, the striking preference for Glu at P2, together with the contributions of the adjacent subsites, singles out the unique substrate binding pocket of neurotrypsin and strongly argues for the selectivity for agrin as substrate.
The selectivity of neurotrypsin for agrin may be further enhanced by exosite interactions. It is widely accepted that the specificity of serine proteases for macromolecular substrates is not only achieved by the specific nature of the substrate binding pocket but also by interactions of remote exosites (25
, 26)
. Exosite interaction resulting in the assembly of a protease with its macromolecular substrate and other accessory macromolecules (cofactors) results in a marked enhancement of the rate of substrate cleavage. Because of their specific nature, exosite interactions also enhance the specificity of the enzymatic reaction, as, for example, shown for thrombin that has an active site with a relatively wide spectrum of acceptable substrates (27
, 28)
. A strong argument for a role of exosite interaction(s) in the enhancement of neurotrypsin specificity (and activity) toward agrin can be derived from our observations: short peptides comprising only the N-terminal amino acid sequence of the - and ß-cleavage sites, including both substrates with p-nitroanilide leaving groups, as well as peptides bearing P and P' subsites on both sides of the scissile bond, were not or only very inefficiently cleaved by neurotrypsin. In summary, selectivity mechanisms at the level of the amino acids flanking the scissile bond and specific exosite interactions may cooperate to focus proteolytic activity of neurotrypsin toward agrin and to effectively curtail nonspecific proteolysis. In accordance with the highly selective nature of proteolytic activity of neurotrypsin, we were so far not able to identify other proteolytic target proteins of neurotrypsin despite extensive efforts using proteomic and other approaches.
Role of a deficient neurotrypsin-agrin pathway in the pathogenesis of mental retardation
Neurotrypsin-dependent cleavage of agrin is particularly abundant during neural development. Expression levels of neurotrypsin and agrin, as well as the neurotrypsin-dependent fragments of agrin, were highest during fetal and postnatal stages of neural development. However, neurotrypsin, agrin, and the 90 kDa fragment of agrin, as an indicator of neurotrypsin-dependent cleavage of agrin, were found at lower levels throughout adult life. Therefore, the temporal pattern suggests a role of neurotrypsin-dependent cleavage of agrin in a function occurring with highest activity during neuronal development but persisting at lower activity in the adult brain. In particular, a role of neurotrypsin-dependent cleavage of agrin in developmental and adult synaptogenesis would be well compatible with this expression pattern and with the observation of mental retardation in humans lacking neurotrypsin (1)
.
Mental retardation results from deficiencies in a large number of higher brain functions generally summarized as cognitive functions (29)
. The currently available data about monogenetic defects resulting in mental retardation reveal a relatively high proportion of genes that encode proteins involved in synaptic functions. Prominent representatives of these contribute to presynaptic release of neurotransmitter and recycling of neurotransmitter vesicles, e.g., IL1RAPL1, RabGDI1, and FACL4, or the regulation of cytoskeletal dynamics and vesicular trafficking in dendritic spines, e.g., OPHN1, ARHGEF6, and PAK3 (30
, 31)
. Both presynaptic vesicle release and dynamic functions of spines are elements of the adaptive response of synapses to altered functional needs that is summarized under the term synaptic plasticity and is thought to play a major role in the implementation of cognitive functions (32
33
34)
. Functions of agrin, the possibly unique substrate of neurotrypsin, are in good accordance with a role as a regulator of synaptic structure, function, and plasticity. Currently best characterized is the motoneuron-derived form of agrin, which serves as a prosynaptic (synapse-organizing, synapse-differentiating, and synapse-protecting) agent at the neuromuscular junction (8
, 35)
. The region of agrin that is responsible for these activities is located in the C-terminal moiety, specifically in the third laminin G domain (5)
. The role of agrin in the CNS has not been worked out with the same detail, although multiple interactions with extracellular matrix and cell surface molecules, including NCAM and dystroglycan, have been reported (4)
. Recent studies (9)
indicated a role of agrin in the regulation of presynaptic excitability via binding to and inhibiting the 3 type of Na+/K+-ATPase. Other recent reports (10
, 11)
indicated a role of agrin in the regulation of filopodia in hippocampal and other neurons. Dendritic filopodia are thin and usually long membraneous protrusions that are found along dendrites and that may be critically involved in synapse formation in the developing and adult nervous system by forming preliminary contacts with axonal boutons or shafts that have the potential to mature into a persistent synapse within days (12
, 13)
.
Although the neurotrypsin-dependent processing seems to terminate the function of agrin at the neuromuscular junction, the role of agrin cleavage by neurotrypsin in the CNS is currently not understood. Cleavage at both sites separates the synapse-regulating activity of agrin from the N-terminal moiety. In our cell culture experiments, cleavage of agrin released both fragments into the culture supernatant. If solubilization also occurs in vivo, cleavage by neurotrypsin may reduce the concentration of the active 22 kDa fragment on the neuronal surface, reduce its interaction with the receptor, and thus have an inactivating effect. However, an activating function of agrin processing is also conceivable and worthwhile of experimental investigation.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
3 Na+-K+-ATPase. The highly selective processing of agrin at CNS synapses by neurotrypsin may be key to our understanding of the pathology of mental retardation associated with loss of neurotrypsin function in humans. Based on the suggested functions of agrin in the control of presynaptic excitability and the generation of new synapses in the CNS, it is conceivable that a dysregulation of agrin processing due to a deficiency in neurotrypsin may result in mental retardation. Therefore, our results identify a defect in the proteolytic neurotrypsin-agrin pathway as a possible pathogenetic mechanism resulting in mental retardation in humans; these findings may shed light on a novel molecular mechanism involved in cognitive functions.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication April 13, 2007. Accepted for publication May 24, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. G. W. Hilgenberg, B. Pham, M. Ortega, S. Walid, T. Kemmerly, D. K. O'Dowd, and M. A. Smith Agrin Regulation of {alpha}3 Sodium-Potassium ATPase Activity Modulates Cardiac Myocyte Contraction J. Biol. Chem., June 19, 2009; 284(25): 16956 - 16965. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Stephan, J. M. Mateos, S. V. Kozlov, P. Cinelli, A. D. Kistler, S. Hettwer, T. Rulicke, P. Streit, B. Kunz, and P. Sonderegger Neurotrypsin cleaves agrin locally at the synapse FASEB J, June 1, 2008; 22(6): 1861 - 1873. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Frischknecht, A. Fejtova, M. Viesti, A. Stephan, and P. Sonderegger Activity-Induced Synaptic Capture and Exocytosis of the Neuronal Serine Protease Neurotrypsin J. Neurosci., February 13, 2008; 28(7): 1568 - 1579. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
