Peters Plus Syndrome Is a New Congenital Disorder of Glycosylation and Involves Defective O-Glycosylation of Thrombospondin Type 1 Repeats*

Peters Plus syndrome is an autosomal recessive disorder characterized by anterior eye chamber defects, disproportionate short stature, developmental delay, and cleft lip and/or palate. It is caused by splice site mutations in what was thought to be a β1,3-galactosyltransferase-like gene (B3GALTL). Recently, we and others found this gene to encode a β1,3-glucosyltransferase involved in the synthesis of the disaccharide Glc-β1,3-Fuc-O-that occurs on thrombospondin type 1 repeats of many biologically important proteins. No functional tests have been performed to date on the presumed glycosylation defect in Peters Plus syndrome. We have established a sensitive immunopurification-mass spectrometry method, using multiple reaction monitoring, to analyze O-fucosyl glycans. It was used to compare the reporter protein properdin from Peters Plus patients with that from control heterozygous relatives. In properdin from patients, we could not detect the Glc-β1,3-Fuc-O-disaccharide, and we only found Fuc-O-at all four O-fucosylation sites. In contrast, properdin from heterozygous relatives and a healthy volunteer carried the Glc-β1,3-Fuc-O-disaccharide. These data firmly establish Peters Plus syndrome as a new congenital disorder of glycosylation.

Glycosylation is the most common and diverse way by which protein molecules are modified. The biological importance of this co-and post-translational modification is becoming increasingly clear from the rapidly rising number of congenital disorders of glycosylation (CDG) 2 (1,2). Recently, Lesnik Oberstein et al. (3) discovered that Peters Plus syndrome (MIM 261540) is caused by truncating mutations in a ␤1,3-galactosyl-transferase-like gene (B3GALTL) that was originally identified by Heinonen et al. (4). The characteristic features of this disorder include anterior eye chamber defects, disproportionate short stature, developmental delay, and cleft lip and/or palate (5). Peters Plus syndrome was classified as a putative CDG (2), because it is not known whether the mutations affect a specific enzymatic activity or cause the formation of a defective glycan structure.
In the meantime, we and others (6,7) found that B3GALTL does not in fact code for a galactosyltransferase but rather for a glucosyltransferase, ␤3Glc-T, that is involved in the synthesis of the unusual disaccharide Glc-␤1,3-Fuc-O-. This small O-linked glycan has been found in the TSRs of thrombospondin-1 (8), properdin, f-spondin (9), ADAMTS-like 1 (10), and ADAMTS-13 (11). TSRs are independent folding modules of ϳ60 amino acids in length that contain three disulfide bonds and a core consisting of alternating tryptophan and arginine residues. The human genome encodes ϳ100 TSR-containing proteins that perform a variety of important biological functions, including regulation of the coagulation system by proteolysis, inhibition of angiogenesis, and cell and axon guidance (12).
The biosynthesis of Glc-␤1,3-Fuc-Ois initiated by the protein O-fucosyltransferase 2 (POFUT2), which attaches the fucosyl residue to a serine or threonine in the consensus sequence CX 2-3 (S/T)CX 2 G (8,11). Subsequently, ␤3Glc-T transfers the glucose onto TSR-fucose. A number of studies have addressed the importance of O-fucosylation. We have recently demonstrated that an enzymatically inactive POFUT2 gene in Caenorhabditis elegans results in abnormal distal tip cell migration and shape of the gonad (13). Both POFUT2 and ␤3Glc-T are localized in the endoplasmic reticulum (7,14). Abolishing O-fucosylation causes a diminished secretion of ADAMTSlike 1 and ADAMTS-13 in cell culture experiments (10,11). Therefore, it has been proposed that O-fucosylation plays a role in the quality control of folding in the early part of the secretory pathway, possibly by tagging properly folded protein.
An alternative O-fucosylation pathway operates on proteins that contain epidermal growth factor-like repeats (15). The first step is catalyzed by POFUT1, which is followed by extension of the fucose by Fringe, a ␤1,3-N-acetylglucosaminyltransferase. Further elongation by ␤4-galactosyltransferase 1 and a sialyltransferase yields the tetrasaccharide NeuAc␣2,6Gal␤1,4GlcNAc␤1,3Fuc-␣1-O- (16,17). It is important to note that the enzymes of the two pathways are specific for either TSRs or epidermal growth factorlike repeats and do not cross-react (6,7,18).
Traditionally, electrophoretic analysis of plasma proteins, like transferrin and apoC-III, has played an important role in the diagnosis of CDG. More recently, it has become apparent that the use of mass spectrometry (MS) provides the needed detailed information to unravel the complexity of some of these disorders (19,20). To investigate the putative glycosylation defect in Peters Plus syndrome, a suitable protein that contains TSRs is required. We have developed a sensitive immunopurification-MS method that detects the O-fucosyl glycans in the reporter protein properdin, the positive regulator of complement. Comparison of patient and control samples shows that in patients that carry the exon 8-skipping mutation in the gene encoding ␤3Glc-T, the glucosyl residue is missing at all four properdin O-fucosylation sites.

EXPERIMENTAL PROCEDURES
Materials-Monoclonal anti-properdin antibody HYB039-06 and goat anti-human properdin were purchased from the Antibody Shop, Gentofte, Denmark. MagnaBind carboxylated derivatized beads were from Pierce. Properdin was obtained from Complement Technology, Inc., Tyler, TX, and its concentration was determined by measuring the absorbance at 280 nm, using the extinction coefficient provided by the manufacturer (A 280 1% ϭ 17.8). Trypsin was from Promega.
Patients-Blood was drawn and collected in a coagulation tube and serum obtained by centrifugation from Peters Plus patients, their parents, and a healthy control. The samples are from seven individuals (numbered 1-7, see Table 1). Individual 1 is a healthy control. Individuals 4 and 5 are two brothers with Peters Plus syndrome that are compound heterozygous for the c.660 ϩ 1G3 A exon 8 donor splice site mutation in the B3GALTL on the paternal allele and a ϳ1.5-Mb 13q12.3q13.1, B3GALTL including deletion on the maternal allele. Their healthy parents are individuals 2 and 3 and are heterozygous for the respective mutations in their sons. Individuals 6 and 7 have Peters Plus syndrome and are not related to each other or the other patients. Both are homozygous for the c.660 ϩ 1G3 A exon 8 donor splice-site mutation. The project was approved by the ethical committee of the Leiden University Medical Center, and written and oral informed consent were given by each individual or his/her legal guardian.
Properdin ELISA-Nunc Maxisorp 96-well microtiter plates were coated overnight at 4°C with 100 l per well of 2 g/ml monoclonal antibody HYB039-06 in bicarbonate buffer (50 mM Na 2 CO 3 , 117 mM NaHCO 3 , pH 9.6, 10 mM NaN 3 ). At this point, and between all other additions of protein, the plates were washed three times with TBS-T (10 mM Tris, 150 mM NaCl, pH 7.5, 0.05% Tween 20). Unbound sites were blocked with 200 l of phosphate-buffered saline, containing 1% bovine serum albumin (buffer A) for 1 h at 37°C. One hundred microliters of serum samples diluted 1:500, 1:1000, and 1:5000 in buffer A or 100 l of standard human properdin (0.6 -80 ng/ml buffer A) were added in duplicate for 2.5 h at 37°C. The plates were incubated for 1 h at 37°C with 100 l of goat anti-human properdin diluted 1:5000 in buffer A. Donkey F(abЈ) 2 anti-goat IgG-peroxidase (100 l; Jackson ImmunoResearch) diluted 1:15,000 in TBS-T was added for 1 h at 37°C. After the last wash, the plates were rinsed with TBS and developed with 100 l of FAST o-phenylenediamine dihydrochloride (Sigma). The reaction was stopped with 50 l of 1 N H 2 SO 4 and the absorbance read at 490 nm.
Immunopurification of Properdin-Monoclonal anti-properdin antibodies were coupled to MagnaBind carboxylated derivatized beads using 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide at a concentration of 2.7 mg/ml, following the manufacturer's instructions. Two hundred microliters of serum were dialyzed overnight at 4°C against phosphate-buffered saline and incubated with 10 l of washed antibody beads overnight on a roller shaker in LoBind Eppendorf tubes. The beads were washed with 1 ml of phosphate-buffered saline, Peptide Mapping and Mass Spectrometry-Preliminary experiments revealed that in-gel tryptic digests of properdin did not yield the relevant glycopeptides (data not shown), which necessitated an in-solution digestion approach. Purified properdin (1 g) was precipitated with chloroform/methanol (21), reduced, carboxymethylated (22), and digested with 300 ng of trypsin at 37°C for 2 h, followed by another 200 ng for 2.5 h in a total volume of 42 l. The digest was diluted 5-fold with 0.1% trifluoroacetic acid, containing 2% CH 3 CN, and 5 l (ϳ400 fmol) were injected onto a reversed phase column for liquid chromatography-mass spectrometry (LC-MS) analysis in the information-dependent acquisition mode. Electrospray ionization LC-MSMS was performed using a Magic C18 HPLC column (100 m ϫ 10 cm; Spectronex) in an 1100 Nano-HPLC system (Agilent Technologies) that was connected to a 4000 Q Trap (Applied Biosystems). The peptides were loaded onto a peptide captrap (Michrom BioResources) at a flow rate of 10 l/min. They were eluted at a flow rate of 300 nl/min with a linear gradient of 3.6 -38% CH 3 CN in 0.1% formic acid/H 2 O in 40 min. Information-dependent acquisition analyses were done according to the manufacturer's recommendations, i.e. 1 enhanced multicharge, 4 enhanced resolutions, and 4 enhanced product ion scans, were repeatedly cycled, and precursors were excluded for 60 s after their second occurrence. Individual MSMS spectra, containing sequence information for a single peptide, were compared by the program Mascot (23), using a data base that only contained the properdin sequence (SwissProt entry number P27918). The modifications Fuc-O-and Glc-Fuc-O-of Ser and Thr were defined as neutral losses of 146.1 and 308.1 Da, respectively, and the C-mannosyl on tryptophans as a stable modification (ϩ162.1 Da). The MS instrument was operated under the following conditions: ion spray voltage, 3500; gas1, 25; declustering potential, 80 V; Q1 and Q3,low resolution (500 at m/z 1000); and dwell time, 60 ms. Amino acids and tryptic peptides have been numbered as they occur in the sequence defined in SwissProt entry number P27918.
For quantitative comparison of a specific glycoform of peptide T7 between different properdin samples, we injected 0.5 l of the 5-fold diluted tryptic digest (ϳ40 fmol) for LC-MSMS in the multiple reaction monitoring (MRM) mode. The CID tan-dem mass spectra of the MMFG and MMF0 forms (see Fig. 1C for the nomenclature) of peptide T7 are indistinguishable (supplemental Table S1). This results from the loss of the fucosyl glycan, because of the gas phase instability of the sugar-peptide linkage. In an MRM LC-MSMS experiment, they can, however, be distinguished by the 162-Da difference in mass of their precursor ions. The y9 (m/z ϭ 1037.5) and y13 (m/z ϭ 1480.6) product ions from the triply charged precursors at m/z 1044.8 and 990.8 were recorded to detect the MMFG and MMF0 glycoforms, respectively. To distinguish the two isobaric forms MMF0 and M0FG, the formation of the [M ϩ 2H-Fuc] 2ϩ quasi-molecular ion from the same precursor was monitored under low collision energy conditions in the same LC-MS experiment. To allow for a quantitative comparison of specific glycoforms between properdin samples, we also measured transitions that are specific for two nonglycosylated peptides (T24 and T28). The intensity of the latter was used to normalize all other transitions for the amount of analyzed protein.

Serum Levels of Properdin in Patients with Peters Plus Syndrome Are Slightly
Reduced-For the analysis of the structural consequences of mutations in the gene encoding ␤3Glc-T in patients with Peters Plus syndrome, we chose properdin, the positive regulator of the complement system (24). This protein is well suited as a reporter for this disorder, because it is relatively abundant, can easily be purified, and the positions of the four Glc-␤1,3-Fuc-O-disaccharides have been determined (9,25). To examine whether its level in patients with Peters Plus syndrome had significantly changed, we performed an ELISA on serum from four individuals that are homozygous or compound heterozygous for a mutation in the gene encoding ␤3Glc-T. The concentration of properdin in the heterozygous individuals 2 and 3 ( Table 1) was very similar to that of a control individual (Fig. 1A). In contrast, we observed a 1.3-1.7-fold   (Table 1) by immunoadsorption. SDS-PAGE showed that the procedure yields properdin of sufficient purity for further analysis (Fig. 1B). It contained as a major contaminant a 75-kDa protein that was found by mass spectrometric analysis to be histidine-rich glycoprotein (data not shown), as well as a minor amount of an unidentified protein with a mass of 40 kDa that we did not further analyze. Both proteins, but not properdin, were also isolated when nonspecific mouse IgG was used in a control experiment. This indicates that they do not interact with properdin but rather with the immunoaffinity beads (data not shown).
TSRs contain, in addition to the Glc-␤1,3-Fuc-O-disaccharide, C-linked mannosyl residues on most of their tryptophans. Because the two types of modification are not always complete, we chose to analyze the major glycoforms of the four O-fuco-sylated tryptic peptides (9, 25), indicated in Fig. 1C, by LC-MSMS.
The MMFG form of peptide T7 from the controls (individuals 1-3; Table 1) had a mass of 3129.6 Da (mono-isotopic; Table 2), and its CID tandem mass spectrum (supplemental Table S1) was in agreement with its previously determined structure (9). This form of peptide T7 was not observed in properdin from the four patients. Instead we found a strong signal for a glycopeptide with a 162-Da lower mass (2967.3 Da; Table 2), which corresponds to the MMF0 glycoform. Theoretically, this mass difference could also result from the lack of a C-mannosyl residue. This could be excluded, however, because the b7 and b8 ions that show the presence of a C-mannose on Trp-83 and -86 were present in the CID tandem mass spectrum (Fig. 2). It is important to note that none of the observed fragment ions carried the O-fucosyl glycan. Its loss in tandem MS experiments in a quadrupole-based instrument is well documented (8,26) and is a result from the gas phase instability of this sugar-peptide linkage. However, the identity of the O-fucosyl glycan can be deduced from the neutral loss observed in a CID experiment at low collision energy. The MMF0 form of the peptides from the patients (as well as the minor component with the same structure from the controls) underwent a 146-Da loss (the mass of a fucosyl residue), whereas the MMFG and M0FG glycoforms from healthy controls lost 308 Da, corresponding to the residue of the disaccharide. Thus, peptide T7 from patients is only substituted with the Fuc-O-monosaccharide.
The analysis of the other O-fucosylation sites was performed in a very similar way. Peptides T11, T17 ϩ 18, and T23 were identified in the chromatogram, based on their masses and CID tandem mass spectra (Table 2 and supplemental Tables S2-S4). In properdin from control individuals, we detected the MMMFG form of peptide T11, the MMFG form of T23, as well  Table 1). The triply charged precursor at m/z ϭ 990.7, whose position is indicated by an arrow, was fragmented during an LC-MSMS experiment. Note that all fragment ions lack the Fuc-O-monosaccharide, because of the instability of the carbohydrate-peptide linkage (dashed arrow) (8,26). W # , C-mannosyltryptophan. The neutral loss of 120 Da, characteristic for the C-linked mannose, is indicated by 120 and 60, for singly and doubly charged ions, respectively (30). Water losses are indicated with an asterisk. as the MM0FG and MMMFG forms of peptide T17 ϩ 18 (Table  2), in agreement with our previous findings (9). For peptides T11 and T23, the observed neutral loss of 308 Da provided further evidence for the Glc-␤1,3-Fuc-Odisaccharide. In comparison, in properdin from patients we did not identify any of the glucosylated glycoforms, but instead we observed the corresponding peptides with a 162-Da lower mass (Table 2). Peptides T11 and T23 underwent a 146-Da neutral loss ( Table 2), demonstrating that they carried the Fuc-O-monosaccharide. For the larger peptide T17 ϩ 18, we could not establish conditions for neutral loss experiments to distinguish between the lack of glucose or C-mannose without fragmenting the polypeptide backbone too much. Therefore, we relied on the different chromatographic behavior of C-and O-glycosylated peptides to assign the glycoforms. In agreement with our previous experience, we consistently observed in the LC-MSMS experiments with peptide T7 that the presence of a C-mannosyl residue caused a much larger decrease in retention time than an O-linked hexose (see also below and Fig.  3, A and B). This allowed us to identify the 162-Da lighter version of peptide T17 ϩ 18 that had an elution time very similar to that of the MMMFG form, as MMMF0, and the well separated but later eluting isobaric one as MM0FG. Consistent with this, the form observed in patients that was still another 162 Da lighter (MM0F0) eluted close to MM0FG. Taken together, the mass spectrometry data on the O-fucosyl glycan-carrying pro-perdin peptides from the patients with Peters Plus syndrome showed that glucosylation is lacking at all four sites.

Detection of O-Fucosylation Defects by Tandem LC-MS in the
MRM Mode-For routine screening of the structure of O-fucosyl glycans in patient samples and for the general analysis of the effect of mutations in the gene encoding ␤3Glc-T, a simple method would be desirable that would allow relative quantification. A sensitive way to achieve this, without the availability of modification-specific antibodies, is nano-LC-MSMS in the MRM mode. This method records the formation of one or more specific product ions from a precursor with a specified m/z value, so-called "transitions" (27). The duration of each measurement is short, so that multiple transitions can be monitored in series, allowing the detection of the relevant forms of an O-fucosylated peptide during chromatography. A representative pair of MRM chromatograms obtained for peptide T7, showing the results for a parent (individual 2) and that of one of his sons with Peters Plus syndrome (individual 5), is depicted in Fig. 3. As expected from the results above, the MMFG form was readily detected in properdin from the parent (Fig. 3A), in addition to the two minor forms MMF0 and M0FG ( Fig. 3B; note the considerable shorter elution time of MMF0). Although isobaric, the latter two were distinguished from each other based on the difference in the fucosyl glycans lost in the collision process. For MMF0 we monitored the formation of the [M ϩ 2H-Fuc] 2ϩ fragment ion at m/z 1412.0 and for M0FG the [M ϩ 2H-Fuc-Glc] 2ϩ fragment ion at m/z 1331.0. In addition, the assignments were confirmed by inspection of the complete CID tandem mass spectra. Strikingly, the MMFG form was undetectable in properdin from the son (arrow in Fig. 3C), and instead we observed an 18-fold increased signal for the MMF0 form (Fig. 3D). Similarly, the M0FG form had disappeared (arrow in Fig. 3D), which was paralleled by the appearance of the M0F0 form (data not shown). The lack of detection of the MMFG form in the son's properdin did not result from an insufficient amount of analyzed protein, as can be concluded from the intensity of the MRM traces for the nonglycosylated peptides T24 and T28 (Fig.  3, A and C). Peptide T7 from the other individuals was analyzed in the same way, and the data, normalized for the amount of analyzed protein, have been summarized in Fig. 3E. In all cases the MMFG (and M0FG) form was present in control individuals but was undetectable in the samples of the patients. The MRM approach was also applied to peptide T23, which yielded the same result, i.e. complete lack of glucosylation in the four patients (data not shown). The reliability of the MRM approach is evident from the good quantitative agreement between the results obtained by monitoring two different product ions produced from the precursor of peptide T7. For example, the relative increase in MMF0 of the patient (individual 5) compared with the control (individual 2), calculated from the y9 and y13 ions, was 17-and 19-fold, respectively. Furthermore, the relative amount of a particular glycoform was very consistent within a particular group of individuals, as evidenced by the small error in the normalized intensity of the MMF0 form calculated from the y9 data of the four patients, i.e. 75 Ϯ 2 (means Ϯ S.E.).

DISCUSSION
The results presented here are the first to provide functional evidence for the type of glycosylation defect caused by biallelic truncating mutations in the gene encoding ␤3Glc-T, and which are likely to lead to Peter Plus syndrome. In patients that were either homozygous for the donor splice site mutation c.660 ϩ 1G3 A, or carried this mutation on the paternal chromosome and a deletion on the maternal one, a complete loss of glucosylation of O-linked fucose in the TSRs of properdin was observed. Although we have not carried out a complete analysis of the limit of detection, we estimate from dilution experiments with peptide T7 that glucosylation would have been detectable at 5% of the level in control individuals.
Properdin can be synthesized in monocytes, neutrophils, and peripheral blood T cells (28), suggesting that the splice site mutation causes a major change in the activity of ␤3Glc-T in one or more of these cell types. Our data do not establish whether the lack of glucosylation of properdin itself actually contributes to the congenital defects seen in Peters Plus syndrome. Properdin stabilizes the C3 convertase, creating a positive feedback loop at the core of the alternative pathway of complement. Its importance is evident from patients that are properdin-deficient, who have a higher susceptibility to meningococcal infections with mortality rates as high as 75% (reviewed in Ref. 28). Recurrent bacterial infections of this type have not been reported to occur in Peters Plus patients. Thus, properdin lacking the glucose modification, even at somewhat reduced serum concentrations, apparently functions sufficiently well to sustain the alternative pathway of complement. It is of interest to note that decreased levels of serum glycoproteins, e.g. coagulation factors and immunoglobulins, have been observed for many CDGs that involve N-glycosylation (29,30). However, in patients with Peters Plus syndrome N-glycosylation of serum proteins appears not to be affected (3).
Recently, it has been found that the lack of multiple O-fucosylation in the TSRs of ADAMTSL-1 and ADAMTS-13 causes a strong decrease in the secretion of these proteins in cell culture (10,11). Obviously, the glucosyl residue is also missing in that case, but it remained unclear whether that by itself was causing reduced secretion. Our data showed only a modest decrease in secreted properdin in the patients with Peters Plus syndrome (Fig. 1), suggesting that in the entire organism the glucosyl residue is not crucial for secretion of this protein. We cannot, however, exclude that some unknown mechanism compensates for the lack of this sugar residue.
The availability of an assay to monitor the glycosylation status of the reporter proteins transferrin and apoIII-C has been very helpful in the clinical diagnosis of type I and II CDG (1,2). However, these proteins do not contain TSRs and do not undergo O-fucosylation. Here we have established that MS in the MRM mode of peptide T7 of properdin can be used to monitor the glycosylation defect in Peters Plus syndrome. The method is simple, sensitive, and highly reproducible, and each of its steps can be automated with available technology. Furthermore, the assay is not only useful to monitor the consequences of other mutations that cause Peters Plus syndrome but also to investigate types of CDG for which the molecular defect has not been elucidated. The method also allows the detection of changes in C-mannosylation, which occurs on tryptophans close to the O-fucosylation sites (9).
The data presented here firmly establish Peters Plus syndrome as a congenital disorder of glycosylation, and the results constitute the basis for further molecular studies, focusing on members of the TSR-containing protein family. Furthermore, the results demonstrate the usefulness of MS in the MRM mode to detect changes in post-translational modifications, even when no modification-specific antibodies are available.