Properdin, the positive regulator of complement, is highly C-mannosylated.

Properdin is the positive regulator of the alternative pathway of complement activation. The 53-kDa protein is essentially composed of six thrombospondin type 1 repeats, all of which contain the WXXW motif, the recognition sequence for C-mannosylation. C-Mannosylation is a post-translational modification of tryptophan residues in which, in contrast to the well known N- and O-glycosylation, the carbohydrate is attached via a C-C bond to C-2 of the indole moiety of tryptophan. C-Mannosylation was first found in human RNase 2 and interleukin-12. The terminal complement proteins C6-C9 also carry this modification as part of their thrombospondin type 1 repeats. We studied the C-mannosylation pattern of human properdin by mass spectrometry and Edman degradation. Properdin contains 20 tryptophans of which 17 are part of a WXXW motif. Fourteen tryptophans were found to be modified 100%. This is the first example of a protein in which the majority of tryptophan residues occurs in the C-mannosylated form. These results show that C-mannosylated proteins occur at several steps along the complement activation cascade. Therefore, this system would be ideal to investigate the function of C-mannosylation.

Glycosylation is one of the most abundant and widespread post-translational modifications of proteins. In the common cases of N-and O-glycosylation the glycan is attached via an amide or a hydroxyl group of an amino acid side chain to the protein. In human RNase 2 a fundamentally different type of glycosylation was discovered (1)(2)(3). An ␣-mannosyl residue was found to be linked via a C-C bond to the C-2 of the indole ring of tryptophan (Fig. 1). It was shown that this modification is enzyme-catalyzed and that dolichylphosphate mannose is the sugar donor (4). In RNase 2 the recognition signal for the transferase was determined to be WXXW (or less efficiently WXXF), in which the first tryptophan becomes modified (5). Meanwhile a total of 22 tryptophans in 7 proteins have been shown to be C-mannosylated, i.e. RNase 2 (1), interleukin-12 (6), and terminal complement proteins C6, C7, C8␣ and ␤, and C9 (7). The terminal complement proteins showed a more complex pattern of C-mannosylated tryptophans than RNase 2 and interleukin-12. They contain as a part of their thrombospondin repeats (TSRs) 1 WXXWXXW motifs, in which both of the first two tryptophans and, surprisingly, also the last one can be C-mannosylated. In addition, in TSRs of C6 and C7, tryptophans were found to be modified although they were not even part of a WXXW motif (7).
The complement system is an innate first line host defense mechanism against microbes. Its activation pathways are composed of three proteolytic cascades, which merge at the cleavage step of (inactive) C3, converting it into active C3b ( Fig. 2; reviewed in Ref. 8). Properdin (or factor P) is a positive regulator of the complement system. It stabilizes the C3 convertase (C3bBb) in the feedback loop of the alternative pathway ( Fig.  2), protecting it from rapid inactivation (9 -11). In addition, the C5 convertase (C3bBbC3b), which converts C5 into C5b, can also bind properdin. This eventually leads to the formation of the membrane attack complex, the actual lytic moiety of complement that will kill the attacked microbe. The importance of properdin is demonstrated in properdin-deficient individuals. These patients have a higher susceptibility to meningococcal infections by Neisseria, leading to fulminant meningitis with mortality rates as high as 75% (reviewed in Ref. 12).
Mature properdin monomer is a 53-kDa protein that occurs in plasma at a specific ratio of dimers, trimers, and tetramers of 26:54:20 (13) and a concentration of 15-25 mg/liter. It consists of six TSRs, which are enclosed by N-and C-terminal parts (14) that show no homology to other proteins. Five of the TSRs contain a WXXWXXWXXC motif, but in the fourth repeat the last tryptophan has been replaced by a valine. One Nglycosylation site is found in the sixth TSR, which is not essential for activity of the protein (15,16).
Finding C-mannosylation in the TSRs of the terminal complement proteins prompted the question of whether properdin, being composed mainly of TSRs, is also C-mannosylated. In this paper we show that properdin is the first protein in which the majority of tryptophans occurs in the C-mannosylated form. This, together with its clearly defined biological function, makes properdin an ideal protein for functional studies on C-mannosylation.
Protein Chemistry-Properdin was reduced and aminoethylated according to Ref. 17. In brief, 50 g of properdin were dissolved in 50 l of 0.5 M Tris-HCl, 6 M guanidine HCl, 10 mM EDTA and reduced with 0.56 mol of dithiothreitol for 4 h at room temperature under argon. After adding 1 ⁄10 volume of methanol, the solution was heated at 50°C. 3.1 l of 4.5 M N-(iodoethyl)trifluoroacetamide dissolved in methanol were added twice and incubated at 50°C for 60 min and 90 min, respectively. Removal of the trifluoroacetyl protection group was achieved by adding 40 mol of acetic acid and incubating for 60 min at * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
37°C. The protein was dialyzed against 50 mM Tris-HCl, 1 M guanidine HCl, pH 9.0 containing 20% methanol, followed by the same buffer without methanol. Digestion with endoproteinase Lys-C was performed at 37°C at an enzyme to substrate ratio of 1:40. After 20 h a fresh portion of protease was added, and the incubation was continued for another 20 h. Digests were fractionated by C 8 reversed phase LC-ESIMS as described (18). Peptides containing an aminoethylated cysteine followed by a proline were not cleaved under these conditions. These peptides were further digested in the same buffer without guanidine HCl for 6 h at 37°C with 500 ng of Lys-C. Fractionation by high pressure liquid chromatography at pH 6.0 was achieved as described (7).
Nano ESIMSMS and solid-phase Edman degradation were performed according to Refs. 19 and 20. The elution position of PTH-(C-2-Man-)Trp was established by comparison to standard runs of residues 5-10 of human RNase 2 [FT(C-2-Man-)WAQW] (1) directly before or after analysis of each of the properdin peptides as described (7).
Peptides have been numbered as they occur in the sequence of the mature protein and labeled with the prefix "K" to indicate that they were generated by endoproteinase Lys-C.

RESULTS
To obtain peptides from properdin that were suitable for analysis, we took advantage of the spacing of the cysteine residues of its TSRs by aminoethylation and endoproteinase Lys-C digestion. This yielded peptides ranging from 12 to 21 residues in length. Fractionation of an endoproteinase Lys-C digest of reduced and aminoethylated properdin was achieved by reversed phase LC-ESIMS (Fig. 3). The mass spectrometry data were extracted for the masses of the expected peptides containing tryptophan(s) without and with one, two, or three hexosyl residue(s), which corresponds to an additional mass of 162 Da per hexosyl residue. The fractions were analyzed by ESIMSMS and Edman degradation.
Peptide K52 from the Sixth TSR (TSR 6)-A peptide with a mass of 1928 Da was found to elute at 35.1 min, corresponding to peptide K52 (residues 353-364) with all three tryptophans C-mannosylated. In the same fraction a second peptide with a mass of 1149 Da was observed, which was identified by ESIMSMS as peptide K5 (GLLGGGVSVEDCae). ESIMSMS of the doubly charged ion of peptide K52 with an m/z value of 965 gave the spectrum shown in Fig. 4. Three times the parent ion [M ϩ 2H] 2ϩ showed the loss of 120 Da (corresponding to an m/z value of 60 in the case of a doubly charged ion; indicated with dotted lines in Fig. 4). This loss of 120 Da is characteristic for aromatic C-glycosides (21,22) and has been instrumental in determining the presence of (C-2-Man-)Trp in several proteins (1,6,7). Moreover, multiple water losses can be seen (labeled with #), which together with the 120-Da loss, form the fingerprint for a C-linked hexosyl residue. A nearly complete series of y ions together with several b ions confirmed the identity of peptide K52. For all y ions and for some b ions containing a C-mannosylated tryptophan, at least one loss of 120 Da (indicated with dashed lines), often together with the multiple water losses, was observed. This has been illustrated in Fig. 4. Edman degradation of this peptide confirmed that all tryptophans were C-mannosylated. Cycles 3, 6, and 9 of peptide K52 showed a PTH-derivative that comigrated with (C-2-Man-)Trp from RNase 2. No signal for the PTH-derivative of unmodified tryptophan was observed. The C-mannosylation pattern of this peptide is another example of the modification of the last tryptophan in a WXXWXXC context, as it was first found in the terminal complement proteins (7). These results have been summarized in Table I.
Peptide K10 from TSR 1-At 40.3 min a peptide with a mass of 2099 Da was found that was 324 Da heavier than expected from the cDNA sequence for residues 49 -66 (14, 23), corresponding to a modification with two hexosyl moieties. The pure peptide was directly analyzed by ESIMSMS and Edman degradation. The tandem mass spectrum of the doubly charged peptide twice showed a loss of 120 Da from the parent ion. Furthermore, a nearly complete series of b ions was observed, and all of the b ions containing a C-mannosylated tryptophan showed the 120-Da loss (Table I). Interestingly, the spectrum also showed a series of a ions, again with all possible 120-Da losses (data not shown). This experiment allowed us to localize the modified tryptophans to position 56 and 59 (Table I). The assignment was confirmed by Edman degradation.
Peptide K17-18 from TSR 2-At 34.8 min a peptide with a mass of 2352 Da was found. This mass fitted to peptide K17-18 (residues 106 -121), with an N-terminally aminoethylated cysteine and all three tryptophans C-mannosylated. The series of b and y ions observed in the ESIMSMS spectrum, the triple loss of 120 Da from the parent ion, and Edman degradation of the  (Table I).
Peptide K26 -27 from TSR 3-A peptide with a mass of 2753 Da was found at 37.7 min that corresponded to the combination peptide K26 -27 (residues 158 -178) with two hexosyl residues. Apparently, endoproteinase Lys-C had not cleaved the (aminoethyl)Cys-Pro bond at position 163-164. ESIMSMS analysis showed two 120-Da losses from the parent ion. A series of b and y ions confirmed the identity of the peptide and localized the two modified tryptophans to positions 169 and 172 and one unmodified tryptophan to position 175. Edman degradation showed the same pattern of modified and unmodified tryptophans (Table I).
Peptide K37 from TSR 4 -In the case of peptide K37, endoproteinase Lys-C in the presence of guanidine HCl did not cleave the (aminoethyl)Cys-Pro bonds at positions 227-228 and 242-243. This resulted in the combination peptide K36 -38 (eluting at 36.6 min), which was too large for analysis by ESIMSMS and Edman degradation. To obtain peptide K37, the peptide K36 -38 was subjected to a Lys-C digest, but now without guanidine HCl. Its mass of 1834 Da matched that of peptide K37 (residues 228 -242) modified with two hexosyl residues. The peptide was purified to apparent homogeneity by LC-ESIMS using a pH 6 buffer system (data not shown). The ESIMSMS spectrum showed two 120-Da losses from the parent ion and a complete y ion series with nearly all possible (C-2-Man-)Trp-related 120-Da losses. Edman degradation confirmed that both tryptophans are C-mannosylated (Table I).
Peptide K43 from TSR 5-The sequence of interest was found in two different Lys-C cleavage products: the fully cleaved peptide K43 (at 36.9 min) and the combination peptide K42-43 (at 38.3 min). In both cases the masses were 324 Da heavier than predicted from the cDNA sequence, suggesting the presence of two hexosyl residues (Table I). The ESIMSMS spectra of both peptides showed two characteristic 120-Da losses from the parent ions. Both spectra allowed the assignment of the C-mannosylated tryptophans to position 294 and 297, with Trp-291 unmodified. The ESIMSMS spectrum of K43 showed a complete y ion series and a nearly complete b ion series, both with several secondary C-linked sugar losses. Edman degradation of K43 confirmed the modification pattern (Table I). DISCUSSION The results presented in this paper (Fig. 5) show that 14 of 20 tryptophans in properdin are stoichiometrically C-mannosylated. These tryptophans were identified by tandem mass spectrometry experiments and confirmed by Edman degradation (Table I) between different hexoses but enable the establishment of the presence of a C-C bond between the hexosyl residue and tryptophan by observing the 120-Da loss, typical for aromatic Cglycosides (21,22). Therefore, it was verified by Edman degradation that all PTH-derivatives of modified tryptophans comigrated with authentic PTH-(C-2-Man-)Trp (Table I). For three proteins, RNase 2 (2, 3), interleukin-12 (6), and C9 (7), NMR always identified the hexosyl group to be ␣-mannose, C-glycosidically linked to the C-2 of the indole moiety. Importantly, the NMR data showed that in C9 this is true for both tryptophans in the WXXW motif (7). Up to now no C-␣-hexosylated tryptophan derivatives other than the mannosylated one have been detected; nor are synthetic compounds available. The recent synthesis of (C-2-Man-)Trp should also allow the production of other hexosyl derivatives, which would provide suitable controls (24,25).
So far seven polypeptides have been shown to contain a total of 22 C-mannosylated tryptophans. Properdin increases this number substantially, further supporting the notion that Cmannosylation is not a rare post-translational modification (7). In the case of RNase 2 (4, 5) and interleukin-12 (6) it has been well established that WXXW is the recognition signal for Cmannosylation, where only the first tryptophan becomes modified. In properdin, only the C-mannosylation of TSR 3 follows this rule (Fig. 5). In the other TSRs, tryptophans occur that are not followed by a second tryptophan (or another aromatic residue) at position ϩ3. Most of the time a cysteine is found at this position; however, in TSR 4 a valine is present. C-Mannosylation of tryptophan residues in such a context has also been observed in the terminal components of complement. This led to the hypothesis that in addition to the WXXW motif, another signal must exist in these proteins (7). At present, it is not   known what constitutes such a signal. In vitro experiments using synthetic substrates strongly suggest that the additional signal is not formed by the residue at the ϩ3 position (7). If this signal were solely encoded in the primary structure, the signal would lie outside the amino acid sequence used in these studies, and a second kind of C-mannosyltransferase would be involved (7). Alternatively, the signal could be formed by a three-dimensional signal patch, as has been described for substrates of UDP-GlcNAc lysosomal enzyme:GlcNAc-1-phosphotransferase (26). In TSRs 1 and 5 of properdin the first tryptophan is not C-mannosylated, indicating that negative signals must exist. These could be either neighboring amino acids or three-dimensional constraints.
The high degree of C-mannosylation of properdin poses the question of degradation of (C-2-Man-)Trp. On the one hand, pathways could exist that degrade (C-2-Man-)Trp to intermediates suitable for entering metabolic pathways. On the other hand, (C-2-Man-)Trp could be excreted, either as part of a peptide or as free (C-2-Man-)Trp. The latter pathway is compatible with the recent finding of (C-2-Man-)Trp in human urine (27). Considering the number of C-mannosylated tryptophans in properdin and the terminal complement proteins (7), and the concentrations of these proteins in plasma, the amounts of (C-2-Man-)Trp found in urine could readily be explained by turnover of only 5% of these proteins per day.
Properdin stabilizes the C3 convertase by binding to its components C3b and Bb (Fig. 2). This inhibits the deactivation of the C3 convertase by factors H and I (11). In this way properdin is not only essential for the activation of the alternative pathway but also for the amplification of the classical and mannose-binding lectin pathways. Although there is no three-dimensional structure of properdin or a TSR module available, it seems very likely that the 14 C-linked mannoses, as hydrophilic moieties, will be exposed at the surface of properdin and therefore could influence the binding to C3b and Bb. Furthermore, it is known that many pathogenic microbes carry mannose-binding receptors on their surfaces (56). If properdin were to interact with these receptors through the C-linked mannoses and thereby get deposited on the surface of an invading microorganism, it could form a focus for attack through any of the three pathways of complement. Its stabilization of the C3 activation feedback loop would ensure local production of C3b, C5b, and the membrane attack complex. The same could be true for a possible interaction between the C-linked mannoses of properdin and the multivalent mannose-binding lectin from serum, which has been shown to bind a variety of microorganisms (57). Further studies are needed, however, as it is not known whether the (C-2-Man-)Trp residues in properdin are able to bind to lectins of bacterial or mammalian origin.
In conclusion, it has been shown here that human properdin is a highly C-mannosylated protein. The fact that, in addition to the terminal components C6-C9, the positive regulator properdin (with its well defined biological role) is C-mannosylated makes complement an ideal system to investigate the function of this post-translational modification.