4 Discussion

An important question in protein biology is how proteases recognize their substrates. The super-family of serine proteases, for example, contains peptidases with very diverse functions such as digestion, degradation, blood clotting, cellular and humoral immunity, fibrinolysis, fertilization, embryonic development, protein processing and tissue remodeling (Rawlings and Barrett, 1993). Additionally, the individual proteases have diverse functions. For instance, NE cleaves extracellular matrix proteins but it also has an important role in host defense. In neutrophils, NE is the key component in preventing the phagolysosomal escape of Shigella. NE specifically cleaves the Shigella virulence factors but it does not target proteins important for bacterial homeostasis or secreted proteins that are not associated with virulence. Thus, NE likely inhibits the interaction of Shigella with neutrophil cytoplasmic proteins. However, it is not known how NE recognizes its substrates. Moreover, it has remained unclear why other members of the chymotrypsin family of serine proteases do not target these virulence factors, although they are homologous to NE and are also in contact with the bacterial effectors in vivo. For example, CG, like NE, is part of the azurophilic granules of neutrophils. These two proteases are structurally almost identical, yet CG does not cleave Shigella virulence factors. In the present study, we approached the question of the specificity of NE for virulence factors from different angles. We first analyzed the substrate for a NE recognition motif. Secondly, we tried to identify the residues within NE that are crucial for the substrate specificity by a structure-function analysis of NE.

4.1  Do virulence factors contain a recognition motif for NE?

4.1.1  A recognition motif for NE was not detected in the primary sequence of IpaB

To test if NE recognizes a primary sequence motif in its substrate we analyzed twelve IpaB deletion mutants for their susceptibility to NE cleavage. These mutants were functional since they complemented a Shigella ipaB deletion strain in epithelial cell invasion and macrophage cytotoxicity when introduced on a plasmid (Guichon, et al., 2001). We found that NE cleaved all these mutant proteins, albeit with slightly different patterns. For example, the IpaB mutant harboring a deletion of the amino acids 207-216 showed a cleavage pattern that suggests a higher susceptibility to NE compared to wildtype IpaB. In contrast, an IpaB mutant harboring a deletion of the amino acids 217-226 seemed less susceptible to NE cleavage than wildtype IpaB. One explanation of these findings is that the relatively large amino acid deletions altered the folding of IpaB in a way that the accessibility of the NE cleavage sites was facilitated or reduced depending on the location of the deletion. However, this putative altered folding did not interfere with the function of the IpaB protein, since we restricted the analysis to testing functional IpaB mutants as mentioned above. By using only functional mutants we wanted to avoid scoring mis- or unfolded proteins. As a result, we were restricted to analyze a small part of the protein, covering only 21% of the 580 IpaB residues. It is therefore possible that a NE recognition motif in the primary amino acid sequence exists but was not identified using this approach. However, several reasons argue against the possibility of a primary sequence motif.

First, NE cleaves not only IpaB but also other virulence proteins, such as the Ipa proteins. IpaA and IpaC, and IcsA. Although these proteins are important for host cell manipulation, they belong to different protein families and hence are not homologous. Additionally, mass spectrometric analysis of NE derived cleavage products of IpaB, A, and C did not reveal a consensus sequence. Another argument against a primary sequence motif in the NE substrates results from a sequence comparison of the synthetic NE peptide substrate with the virulence factors. This peptide was used to measure NE activity. It consists of four amino acids (valine at the P1 position followed by proline and two alanines) linked to a chromophore. It was shown that NE cleaves peptide substrates with high efficiency if the substrates carry valine at the P1 and proline at the P2 position (Harper, et al., 1984;Marossy, et al., 1980;McRae, et al., 1980;Zimmerman and Ashe, 1977). A valine-proline sequence tag within the virulence factor could therefore serve as a NE recognition motif. However, we were unable to find such a tag in IpaB. Nevertheless, it might be worth to scan IpaB and other virulence factors for the occurrence of a “side chain profile” resembling the residues of the peptide substrate. Finally, one could assume that the number of residues after which NE can cleave is reduced in Shigella proteins not targeted by NE compared to virulence factors. NE prefers to cleave after valine but also after leucine or alanine (Harper, et al., 1984;Marossy, et al., 1980;Nakajima, et al., 1979;Powers, et al., 1977). However, the percentage of these residues does not significantly vary between the different protein groups. For example, alanine, valine and leucine constitute 17% of the primary sequence of IcsA, which is cleaved by NE, whereas they represent 26% in OmpA, which is not cleaved by NE.

The presence of discrete cleavage products of IpaB upon NE treatment suggests that IpaB was folded in the Shigella culture supernatant and that NE only attacked some of the potential cleavage sites. The question is if NE specifically recognizes these cleavage sites or if virulence factors in contrast to non-virulence factors display some of their valines on the surface of the folded protein. To address the latter question one would need to compare multiple structures of each protein group and search for exposed amino acid stretches containing valine. However, only few virulence factors have been crystallized to date, which would make this a challenging approach. Therefore, we tested if the secondary or higher order structures of IpaB are a prerequisite for cleavage by NE. In fact, NE cleaved denatured IpaB with the same cleavage pattern as native IpaB. Denatured IpaB was slightly more susceptible to NE than native IpaB as the 62 kDa band of IpaB was degraded to a greater extent than in native IpaB. This result indicates that secondary structures like coiled-coils do not serve as recognition motif for NE. Such structures have been proposed as common features for virulence factors (Pallen, et al., 1997). In addition, this also suggests that the cleavage of IpaB by NE does not require higher order structures beyond the secondary structure level. However, we cannot exclude the possibility that denatured IpaB refolded during the 1-2 h incubation time at 37⁰C in the NE cleavage experiment.

It is also possible that IpaB is generally not folded in the Shigella supernatant. Secreted effectors such as IpaB are released through a 2-3 nm wide pore within the needle-like structure of the type three secretion apparatus in Shigella (Blocker, et al., 1999). Folded IpaB as globular protein likely possesses a diameter of approximately 6 nm, which suggests that IpaB passages the needle in an unfolded state. Since IpgC, the chaperone of IpaB, remains within the cytoplasm of Shigella, one could assume that IpaB stays unfolded in the supernatant.

IpaB is thought to form an extracellular complex with IpaC after secretion across the bacterial membrane and to facilitate further effector delivery (Blocker, et al., 1999;Hayward and Koronakis, 1999). One could assume that the binding to IpaC is a prerequisite for folding of IpaB. Since we used supernatant from Shigella carrying ipaB on a high-copy plasmid, one could argue that overexpression of IpaB could outnumber the IpaC molecules leading to an accumulation of mis- or unfolded IpaB. However, it was shown that the amount of IpaB secreted by this Shigella strain is not significantly increased in comparison to the wildtype strain (Guichon, et al., 2001). Therefore it is rather unlikely that the IpaB analyzed in this study was different from IpaB secreted by a wildtype Shigella strain.

Taken together, we were unable to detect a recognition motif for NE in the primary or higher order structures of the NE substrate IpaB. Since NE has to recognize and bind the virulence factors before cleavage, the role of the substrate in this interaction remains to be elucidated. One approach could be to N-terminally sequence NE derived cleavage products of IpaB and map these sequences to the primary structure of IpaB. Subsequent analysis of the biochemical character of the amino acids C- and N-terminal to the scissile bond could yield a common pattern, for example a stretch of hydrophobic residues. The sequences of Shigella virulence factors and other enterobacteriacae could then be examined for this pattern.

Next, we addressed the question of localization of NE specificity by a functional analysis of wildtype and mutant NE. We hypothesized that a recognition motif for virulence proteins existed in NE and that it should be possible to mutate this motif without comprising the catalytic activity of the protease. More specifically, we speculated to identify a NE mutant that was still active towards its peptide substrate but would no longer recognize and cleave virulence factors like IpaB. For this purpose we expressed recombinant NE.

4.2.1  Recombinant expression of NE

In neutrophils, NE is synthesized as an inactive zymogen that requires the removal of amino acids at the N- and C-terminus for full activity. As a result, active mature NE starts with isoleucine instead of methionine. Since bacteria lack the proteases needed for the NE processing, we assumed that recombinant expression of the full-length protein in E. coli would not result in an active protein. Thus, we expressed recombinant mature NE that carried an additional methionine as first aminoterminal residue. However, this recombinant NE was not active, since it did not cleave the NE peptide substrate. Interestingly, it was mentioned that expression of the same mature NE in RBL-1 cells did also not result in an active protein (Li and Horwitz, 2001). Since expression of the full-length protein in RBL-1 cells does yield active NE [our results; (Li and Horwitz, 2001)], it is likely that the methionine with its bulky aromatic side chain interferes with the proteolytic function of the protein. But it is also possible that the N-terminal two-step processing of NE is necessary for correct post-translational modification or folding of the enzyme. Another reason why E.coli derived NE was not active could be the inability of prokaryotes to glycosylate proteins. NE carries two asparagine-linked side chains, which might be important for the activity of the enzyme (Sinha, et al., 1987;Watorek, et al., 1993). However, according to the structure of NE these sugar chains are located at the surface of the protein away from the active site (Bode, et al., 1989). Thus it is rather unlikely that they are involved in substrate recognition or important for enzymatic activity. Furthermore, the RBL-1 cells target NE to granules just as neutrophils do (Gullberg, et al., 1994). This may support folding of NE and is likely to prevent auto-degradation of the enzyme as it is less active in the low pH environment of the granules and when attached to a granular matrix (Avila and Convit, 1976). In summary, RBL-1 cells seem to provide optimal conditions to express a protease of the azurophilic granules of neutrophils.

Since the analysis of the substrate could not answer why and how NE specifically recognizes virulence factors, we approached this question by mutational analysis of the NE protein. NE belongs to a large family of serine proteases with sequence and structural similarity but with very different biological functions. CG does not degrade Shigella virulence proteins, although it is as abundant in the azurophilic granules as NE and belongs to the same subfamily of chymotrypsin-like serine proteases. Additionally, CG and NE are not only homologous, but their crystal structures are almost identical. Yet, specific differences should explain why NE but not CG targets virulence factors. Substrate specificities among this group of serine proteases are considered to depend on amino acid variations in the substrate-binding cleft (Perona and Craik, 1995). Therefore we analyzed the structure and amino acid composition of this cleft in NE and CG. We identified loci where single or multiple amino acids were strikingly different between the two enzymes. We assumed that replacement of these residues in NE by their structural CG counterparts or by the non-polar amino acid alanine would alter the NE specificity in a way that the NE mutants would not be able to target Shigella virulence proteins any more. All residues selected for mutation were either part of the NE substrate binding pockets or of surface loops containing residues of the different pockets. In general, these loops connect the walls of the pockets without necessarily contacting the substrate residues directly (Hedstrom, et al., 1992). Since loop sequences are often unique in individual proteases, they have been suggested to play a role in determining substrate specificity (Hedstrom, et al., 1992). The following chapters discuss the functional profiles of the generated NE mutants with regard to the location of the affected residues in the sequence and in the three-dimensional structure (see also chapter 3, figures 3.10 and 3.11 and for sequence alignment see appendix, table 5.1).

NE mutants 35-41 and 58A-61

The residue exchanges in the NE mutants LRGGHF 35-41 IQSPAGQSR and VANVNVR 58A-61 WGSNINV affected the S1-3’ pockets and the corresponding surface loops [figure 4.1 and (Bode, et al., 1989)]. However, these two NE mutants displayed different phenotypes.

	Fig. 4.1: Schematic representation of the NE residues 35-41 and 58A-61.
	The NE segments 35-41 (cyan) and 58A-61 (salmon) interact with the amino acids of substrate C-terminal to the scissile bond (P1’-P3’ in wheat). The catalytic residues aspartate, histidine and serine are shown in red. The residues discussed in the text are labeled using the three-letter code for amino acids. The picture was generated using Pymol and residues are depicted as sticks and cartoon (DeLano, 2002). The representation is based on the crystallization of NE with the inhibitor TOM (Bode, et al., 1986b).

In the NE mutant 35-41 the phenylalanine (F) at position 41 was exchanged by an arginine (R). Interestingly, this exchange did not affect the ability of the mutant to cleave the NE peptide substrate and Shigella virulence factor, although this residue had been suggested to influence specificity (Bode, et al., 1989). According to the structure of NE bound to an inhibitor, the phenylalanine interacts with the P1’ and the P3’ residue of this substrate [figure 4.1 and (Bode, et al., 1986b)]. In the mutant 35-41, arginine is at this position and its side chain is positively charged in contrast to the one of phenylalanine. However, arginine in the context of CG is flexible, which might allow some adaptation to bound substrate residues. This could explain why the mutant was still able to cleave the NE substrates (Hof, et al., 1996). In the rat mast cell protease II (RMCPII), another chymotrypsin-like serine protease, the residue segment 34-41 has been proposed to be important for substrate specificity (Perona and Craik, 1995;Perona and Craik, 1997). Yet, this suggestion was based on structural modeling but not on a functional analysis. In this study, we functionally show that the substrate specificity of NE is not encoded within this loop interacting with the P1’-P3’ residues. It is therefore tempting to argue that protease residues interacting with the Pn’ residues of the substrate are generally not important in determining NE specificity.

However, the mutant NE 58A-61 switched to a CG-like specificity for the peptide substrate and the exchanged residues do interact with the Pn’ residues of the substrate [figure 4.1 and (Bode, et al., 1989;Bode, et al., 1986b)]. This indicates that a protease-substrate interaction C-terminal of the scissile bond actually is important for substrate recognition. The NE mutant 58A-61 had lost the specificity towards the NE but had acquired the specificity towards the CG peptide substrate. Additionally, the NE mutant 58A-61 did not target the Shigella virulence factors IpaB and IcsA. But since its CG activity was low, we cannot make any valid conclusions regarding the specificity towards Shigella virulence factors.

The switch in specificity for the peptide substrate could have been caused by the exchange of valine at position 58D. This valine in NE is known to interact with the substrate residue P3’ [figure 4.1 and (Bode, et al., 1989)]. However, the peptide substrate used in our assays is only composed of a chromophore C-terminal of the scissile bond and it is unclear if the NE valine contacts the chromophore. Therefore its role in NE specificity requires a more detailed analysis, for example, by generating a NE mutant that only lacks this valine.

As mentioned above, the CG activity of this mutant towards the CG peptide substrate was low as its CG kinetics was almost indistinguishable from the kinetics of endogenous proteases. It is possible that the introduced amino acids only partially conferred CG specificity and that we would need to exchange an elongated stretch of residues in this locus to obtain a mutant with higher CG activity. Another explanation for the low CG activity could be that residues 58A and 58B are part of a β-sheet in NE and in close proximity to histidine 57, which is part of the catalytic triad [figure 4.1 and (Wei, et al., 1988)]. This β-sheet might be disrupted by the exchange, which in turn could negatively influence the stability of the triad and thus the activity of the protein. Finally, the low CG activity might be caused by improper glycosylation. CG contains one potential N-glycosylation site (Watorek, et al., 1993). This site is asparagine at position 60 and that is introduced into the mutant. It is possible that the mutant carries three asparagine-linked sugars, which interfere with the activity of protein.

Taken together, we showed that the segment 35-41 in NE is not important for NE specificity whereas the segment 58A-61 partly contributes to the specificity of NE.

Most residues that we found to be significantly different in NE and CG were part of the substrate specificity pocket S1. The residues of this pocket have been considered as prime determinants for the specificities of the different chymotrypsin-like serine proteases, because they are complementary to the preferred P1 residue (Berg JM, 2003;Steitz and Shulman, 1982). Multiple residues including phenylalanine at position 192 define the S1 pocket of NE (figure 4.2). According to (Bode, et al., 1989) the backbone of this phenylalanine contributes to the formation of the entrance of the S1 pocket and it is one the residues constricting the pocket towards the bottom. Importantly, the phenylalanine has been suggested to influence NE specificity (Bode, et al., 1989). We assumed that exchange of this residue by its CG counterpart or by alanine would affect the specificity of NE. Surprisingly, both mutants, F192A and F192K, did cleave the NE peptide substrate as well as the Shigella virulence factors IpaB and IcsA. It is intriguing that replacement of phenylalanine by lysine (K) did not impact NE specificity, since lysine, in contrast to phenylalanine, is highly positively charged and not aromatic.

Interestingly, the NE mutant F192A was also able to cleave the CG peptide substrate, albeit with lower activity than the NE peptide substrate. The CG peptide substrate carries phenylalanine with its bulky and aromatic side chain at the P1 position. It is therefore possible that the introduction of alanine with its short side chain opened the entrance of the S1 pocket and thus allowed larger side chains to enter. Yet, it remains unclear how the introduced alanine elicited a CG activity. In summary, since the NE mutants F192A and F192K retained NE specificity, the phenylalanine at position 192 does not seem to be a determinant for the specificity of NE, although it is part of the S1 pocket.

	Fig. 4.2: Schematic representation of the S1 pocket of NE.
	The residues phenylalanine at position 192 (lime) and alanine at position 213 (marine) form the entrance of the S1 pocket. The side chain of the substrates’ P1 residue valine (wheat) points towards the pocket. The side chains of valine at position 190 (grey) and 216 (green) restrict the accessibility to the bottom of the pocket. The catalytic residues aspartate, histidine and serine are shown in red. The remaining NE residues are shown as cartoon in grey without side chains or are not shown at all. The picture was generated using Pymol (DeLano, 2002) and is based on the crystallization of NE with the valine chloromethyl ketone inhibitor (Wei, et al., 1988).

Alanine at position 213 in NE is another residue defining the S1 pocket (figure 4.2). It is thought to contribute to the constriction of the NE S1 pocket at its entrance and had been suggested to influence specificity of NE (Bode, et al., 1989;Bode, et al., 1986b). Interestingly, we found that exchange of the alanine by its CG counterpart valine did not influence the NE specificity. The NE mutant A213V specifically cleaved the NE peptide substrate as well as the Shigella virulence factors IpaB and IcsA. However, its NE activity was decreased in comparison to the NE activity of the mutants F192A and F192K. Since the isopropyl side chain of valine is larger than the methyl side chain of alanine, the presence of valine instead of alanine could have narrowed the entrance of the pocket and thus affected the NE activity of the mutant negatively. However, this amino acid exchange did not interfere with the specificity of the mutant for the biological substrates of NE. Interestingly, a topologically close homolog of NE, the digestive chymotrypsin-like serine protease porcine pancreatic elastase (PPE), carries threonine at the position 213 and has been shown to preferably cleave after alanine (Powers, et al., 1977;Szabo, et al., 1980;Zimmerman and Ashe, 1977). In contrast to valine, threonine is a polar residue but both side chains are branched in a similar way and thus present a similar steric restriction at the entrance of the S1 pocket. Therefore it is possible that the NE mutant A213V would also exhibit a higher activity towards an NE peptide substrate carrying an alanine instead of a valine at the P1 position.

Taken together, we were not able to change the specificity of NE by mutating single residues in NE at positions 192 and 213, although they are part of the NE S1 pocket. These observations are in accordance with the finding of (Graf, et al., 1987), who analyzed the specificities of trypsin and chymotrypsin, two other prominent members of the chymotrypsin-like serine proteases. They showed that the specificity of trypsin could not be converted into a chymotrypsin-like specificity through exchange of a single S1 residue.

Phenylalanine at position 215 in NE is part of the S2 pocket and its side chain runs parallel to the side chain of the residue in the P4 position of the inhibitor [figure 4.3 and (Bode, et al., 1989)]. This phenylalanine was either replaced by its CG counterpart tyrosine (Y) or by alanine (A). We found that the NE mutant F215Y specifically cleaved the NE peptide substrate and the Shigella virulence factors IpaB and IcsA. One explanation why the substitution of phenylalanine by tyrosine did not affect the character of NE is that the side chains of phenylalanine and tyrosine are identical except for a hydroxyl group at the aromatic ring of tyrosine. Therefore the steric and biochemical features of tyrosine resemble the ones of phenylalanine and maintain the bowl-shaped and rather hydrophobic character of the NE S2 pocket (Wei, et al., 1988).

In contrast to tyrosine, the introduction of an alanine at position 215 did influence the activity and partially the specificity of the mutant enzyme towards the peptide substrate. F215A cleaved the NE peptide substrate but only with low activity since we could not measure the NE kinetics of the mutant. Furthermore, it acquired the ability to cleave the CG peptide substrate but this CG activity was also low. It is likely that alanine, whose side chain only consists of a methyl group, interfered with the character of the NE S2 pocket. This could have lead to an incorrect positioning of the P2 residue proline, which maybe influenced positioning of the P1 residue of the substrate or the binding of the peptide substrate in general. As mentioned above, the side chains of phenylalanine at position 215 and of the P4 amino acid of the substrate are in close contact. Therefore it is possible that the introduction of alanine at position 215 negatively affects the putative interaction of these side chains. Since proline and alanine are at the P2 and P4 positions in the NE and the CG peptide substrate, this might explain why both activities of the mutant NE were low. However, it does not explain why the NE mutant F215A was also able to cleave the CG peptide substrate. It is possible that the phenylalanine influences the S1 pocket in the native and soluble NE protein and therefore its exchange by alanine with its short side chain enlarged the S1 pocket allowing cleavage after a bulky residue. Another explanation could be that phenylalanine at position 215 influences the correct formation or function of the NE surface loop spanning the amino acids 163 to 181, because its side chain is in close proximity to the residues 177 and 180. This surface loop seemed important for substrate binding of NE in preliminary experiments (data not shown) and had been suggested to be crucial for specificity of trypsin (Perona and Craik, 1997).

	Fig. 4.3: Schematic representation of the S2 pocket of NE.
	The S2 pocket of NE is formed among other residues by phenylalanine at position 215 (magenta), histidine (His 57) and leucine (Leu 99) (Bode, et al., 1989). The residues of the substrate N-terminal to the scissile bond are shown in wheat (P1-P4). The side chain of phenylalanine at position 215 runs parallel to the substrate residue P4. Parts of the S1 pocket (Val 190 and Val 216) are shown for better orientation. The catalytic residues histidine and serine are depicted in red. All residues are presented as sticks The picture was generated using Pymol (DeLano, 2002) and is based on the crystallization of NE with the valine chloromethyl ketone inhibitor (Wei, et al., 1988).

Taken together, the substitution of phenylalanine at position 215 in NE with its CG counterpart tyrosine did not influence the NE specificity. In contrast, the substitution of that residue with alanine reduced the NE activity of the mutant and simultaneously allowed the cleavage of the CG peptide substrate. Since the NE as well as the CG activity of the NE mutant F215A was too low to measure kinetics, we are unable to draw any conclusions with regard to the activity against virulence factors.

The NE mutants VRGG 216-218 GKSS and VRGGCASGLY 216-224 GKSSGVP affected several NE specificity pockets as well as a surface loop, which is constituted by the residues 217-226 (figures 4.4 and 4.5). We could show that these two mutants lost the specificity for the NE peptide substrate but gained the ability to cleave the CG peptide substrate. Interestingly, the CG activity of the NE mutant 216-218 was approximately ten times higher than the CG activity of the NE mutant 216-224 per total protein concentration of the cell lysate. Since we did not purify the recombinant protein, it remains to be determined if this mutant actually presented a CG-like enzyme with a high substrate turnover rate or if the mutant protein was expressed to a higher level than NE mutant 216-224. Furthermore, NE mutant 216-224 did not target the Shigella virulence factors, whereas the NE mutant 216-218 showed marginal IpaB and IcsA degradation. Because of this residual NE specific activity, it is tempting to argue that the complete NE segment of amino acids 216-224 is crucial for the NE specificity for virulence factors. The segment 219-224 constitutes almost the complete surface loop that was suggested to influence specificity of chymotrypsin-like serine proteases (Perona and Craik, 1997). Additionally, the residues 223 and 224 were also proposed to influence NE specificity (Bode, et al., 1989). However, the NE mutant 216-218 did not cleave the Shigella virulence factors when its CG activity units were comparable to the ones of NE mutant 216-224. This strongly suggests that the four NE amino acids 216-218 are sufficient to encode for the NE specificity. It nevertheless might be interesting to generate a NE mutant in which only the residue segment 219-224 is exchanged and test its ability to cleave virulence proteins.

Assuming that the NE segment 216-218 does encode for the NE specificity, it is tempting to speculate that only one of the four residues determines this specificity. The two residues at position 217 and 218 in NE are glycines, which were replaced by two serines in the mutants. These residues possibly interact with the P5 position of the substrates [figure 4.4 and (Bode, et al., 1989)]. In both peptide substrates this position is composed of a succinyl group but it carries different modifications. The succinyl group in the CG peptide substrate contains a polar hydroxyl group whereas the methoxysuccinyl group in the NE peptide substrate harbors a non-polar methyl group at this position. It is possible that the introduced residues are able to interact with hydroxl- but not with methyl groups, which in turn influences specificity. Functional analysis of a NE mutant in which only the two glycines are replaced by serines should clarify the role of these residues in NE specificity. The residue 216B in NE is an arginine and its side chain points along the side chain of the P4 residue of the substrate [figure 4.4 and (Bode, et al., 1989;Bode, et al., 1986b)]. Since arginine was replaced by lysine that has similar biochemical features, it is rather unlikely that this exchange conferred the switch of specificity in the NE mutant 216-218. The most likely candidate for determining specificity is the residue at position 216. It had been speculated earlier that this residue might be important for specificity of trypsin (Perona and Craik, 1997).

In both NE mutants 216-218 and 216-224, valine at position 216 was substituted by its CG counterpart glycine. The side chains of both valine and glycine at this position reach into the S1 pockets of NE and CG, respectively. Thereby the residues define the size of the S1 pockets (figure 4.2 and see also chapter 1, figure 1.7). The side chain of glycine, which only consists of hydrogen, is smaller than the one of valine, resulting in an enlarged S1 pocket. Thereby it allows the bulky and aromatic side chain of phenylalanine, which is the preferred P1 residue in CG peptide substrates, to enter to the base of the pocket. In contrast, NE does not cleave peptide substrates with phenylalanine at the P1 position (Harper, et al., 1984), possibly because the valine sterically hinders the access of large side chains. In addition, the side chain of valine also contributes to the hydrophobic character of the NE S1 pocket by burying the acidic aspartate at position 226 (Navia, et al., 1989). Therefore, the replacement of valine by glycine could not only affect the geometrical but also the biochemical character of the S1 pocket in the mutant. Interestingly, trypsin also contains a glycine at position 216. It cleaves peptide bonds after arginine or lysine, whose side chains are the longest among the twenty common amino acids. Replacement of this glycine with alanine resulted in an almost complete loss of trypsin activity when measured using peptide substrates (Hedstrom, et al., 1992). Since the size of alanines’ side chain is in between the one of glycine and valine, it is possible that this trypsin mutant would be able to cleave the NE peptide substrate. Furthermore, it would be interesting to test if replacement of this glycine with valine would confer NE specificity to trypsin. However, it was stated that introduction of residues of the NE S1 site into trypsin failed to confer specificity towards elastase-specific substrates (Perona, et al., 1995). Eventually, the generation and analysis of a NE mutant in which only the valine at position 216 is replaced by glycine could prove if the specificity of NE for virulence factors is encoded in this residue. If this was the case, it would mean that the character of the S1 pocket determines substrate specificity. Consequently, it would indicate that in virulence factors, in contrast to non-virulence proteins, valines and other residues after which NE cleaves are more accessible than residues after which CG cleaves.

	Fig. 4.4: Schematic representation of the NE segment 216-224.
	The residues 216-224 (green) are part of a surface loop and interact with the residues of the inhibitor N-terminal to the scissile bond (P1-P5 shown in wheat). In addition to Val 216 parts of the S1 pocket (Val 190) are shown for better orientation. The catalytic residues aspartate, histidine and serine are depicted in red. All amino acids are presented as sticks. The representation was generated using Pymol (DeLano, 2002) is based on the crystallization of NE with the valine chloromethyl ketone inhibitor (Wei, et al., 1988)

Taken together, we demonstrated that the residue segment 216-224 is crucial for the NE specificity for virulence factors. Additionally, by inserting the analogous CG residues we were able to introduce CG specificity for peptide substrates. Thereby we functionally proved the hypothesized importance of the surface loop 217-225 in substrate specificity among chymotrypsin-like serine proteases (Perona and Craik, 1997).

Asparagine at position 98 in NE is part of a protruding loop segment that in CG comprises the residues 94-99 (Hof, et al., 1996). This residue was replaced by its structural counterpart leucine or by alanine, resulting in the two NE mutants N98A and N98L. Interestingly, the substitution of asparagine by leucine did not influence the NE specificity for the peptide substrate or the Shigella virulence factors. Leucine carries a nonpolar side chain, whereas the side chain of asparagine is polar. However, both side chains are branched in a similar way, which might explain why the exchange did not influence the NE specificity of the mutant. In contrast, replacement of asparagine with alanine led to a complete change of specificity. The NE mutant N98A did not cleave the NE peptide substrate and Shigella virulence factors but cleaved the CG peptide substrate. As mentioned above, the asparagine is part of a surface loop, and this loop was proposed via structural computing to influence specificity of an enteropeptidase of the chymotrypsin-like family (Perona and Craik, 1995). However, the asparagine was not suggested to influence specificity of NE. Furthermore, alanine is a nonpolar residue like leucine and the asparagine in NE does not directly contact the inhibitor (figure 4.5). It is therefore difficult to explain why the introduction of alanine had such a strong effect. Since the side chains of alanine and leucine differ in size, it is possible that the side chains’ steric character at this position is important for the interaction with the substrate N-terminal to the scissile bond and thus for the specificity of the enzyme. To understand the critical function of this NE residue at position 98 in substrate recognition, one would have to analyze the crystal structure of this mutant bound to IpaB.

Taken together, we proved asparagine at position 98 to be a determinant of NE specificity. The replacement of this single amino acid with alanine introduced CG specificity to the mutant enzyme. This is in contrast to the hypothesis of (Krem, et al., 1999). They postulate that the C-terminal part, starting at residue 189, encodes for the function of serine proteases. Additionally, it is not possible to confer chymotrypsin specificity into trypsin by replacement of a single residue. To achieve this transfer in specificity, one residue of the S1 pocket and two adjacent surface loops have to be exchanged (Hedstrom, et al., 1992).

	Fig. 4.5: Schematic representation of NE.
	The residues 217-225 form a surface loop (green). The valine at position 216 (green) is shown with its side chain. The asparagine at position 98 (yellow) does not directly contact the inhibitor (P1-P5, wheat) and it is located on the surface of the protein. The catalytic residues aspartate, histidine and serine are shown in red. The remaining NE residues are shown as cartoon in grey. The two barrels of β-sheets (arrows) and a single α-helix are shown. The presentation was generated using Pymol (DeLano, 2002) is based on the crystallization of NE with the valine chloromethyl ketone inhibitor (Wei, et al., 1988).

In the present study we identified the residues at position 98 and 216-224 to be critical for NE specificity. Mutation in these residues to alanine or the analogous CG residues changed the specificity CG-like. Furthermore, the NE mutants N98A, 216-218 and 216-224, as CG, did not degrade Shigella virulence factors, although they were able to cleave the CG peptide substrate. This finding contributes to our understanding of how proteases recognize their substrates, since it shows that single residues apart from the catalytic center can determine the interaction of the full-length enzyme with its substrate. We assume that introduction of these NE residues into CG would confer NE specificity to the CG mutants. To further underline the biological importance of our results, it would be very interesting to perform in and ex vivo experiments. It was shown that NE null mice are more susceptible to Gram-negative bacteria, whereas CG knockout mice are susceptible to infections with Gram-positive bacteria (Reeves, et al., 2002;Tkalcevic, et al., 2000). Therefore, mice expressing mutant instead of wildtype NE should also be more susceptible to infections with Gram-negative bacteria. Furthermore, isolated neutrophils from these mice should allow the escape of Shigella from the phagolysosome to the cytoplasm, as it was observed in neutrophils from NE null mice (Weinrauch, et al., 2002). Additionally, wildtype and transgenic mice expressing our identified NE mutants could be infected with various pathogens and this way the spectrum of NE specificity in vivo could be assessed. On the other hand, one could test if the three CG-like NE mutants do not only cleave the CG peptide substrate but are also able to target biological substrates of CG such as the human brahma protein (Biggs, et al., 2001).

Our results also present an important finding with regard to the evolution of proteases. The members of the large family of chymotrypsin-like serine proteases derived from a common ancestor and exist in pro- and eukaryotes. This parental enzyme is thought to possess a prototypical fold containing the catalytic machinery. However, the broad divergence in substrate specificity within this protease family is incompletely understood (Krem, et al., 1999). The variety in function is thought to have derived from mutation of residues that are not involved in the catalytic or structural function. Furthermore, it has been postulated that enzymes of this family that evolved early are less diverse and thus less stringent in their substrate P1 preference. In contrast, proteases that evolved late, such as NE and CG, are very stringent in their P1 substrate specificity leading to specific functions and thus individual ecological niches of the proteases (Perona and Craik, 1997). We show that these niches can be switched by mutation of a single residue that is not involved in the architecture of the S1 pocket. The interaction of the immune system and pathogens is characterized by a constant battle of detection and elimination on the one and evasion on the other side. Pathogens usually have a short life span, which facilitates the appearance of new variants that can be selected to escape the host immune system. The fact, that the specificity of NE can be converted to that of CG by a single mutation, even though their sequence identity is only 37%, could indicate that evolution lead to specific chymotrypsin-like serine proteases but that the ability to exchange functions was retained in this group of proteases. Thereby, the chances are increased that a new pathogen with novel virulence factors will encounter a mutant protease that can recognize and degrade the pathogenic factors.

In summary, this study has defined regions in NE distinct from the catalytic site that are important for recognition of virulence factors. In future studies, using crystal structures, X-ray scattering and High-throughput screening for substrates that are recognized by the mutant protein or peptide, we can begin to analyze whether these regions are important for binding to specific substrates and if conformational changes as a result of binding are important for enzyme specificity. These studies offer a potentially valuable tool to expedite biochemical studies and drug design. Since NE leads to aberrant lung epithelium destruction and mucous production in cystic fibrosis, for example (Birrer, et al., 1994;Bruce, et al., 1985;Schuster, et al., 1995) the results of this study could be useful for the design of a novel and precise NE inhibitor. On the other hand, the determination of the NE residues that are crucial for recognition of virulence factors could present a template for the design of new types of antimicrobials that block a bacterial infection by masking bacterial virulence factors.