|
|
|
|||||||
|
|||||||||
(Received for publication, September 8,
1994; and in revised form, November 22, 1994) From the
Hydroxyproline-rich glycoproteins (HRGPs) occur in the
extracellular matrix of land plants and green algae. HRGPs contain from
2 to 95% of their dry weight as carbohydrate, predominantly as
oligoarabinosides and/or as heteropolysaccharides which are O-linked to the hydroxyproline residues. A glycosylation code
that determines the presence or absence and extent of arabinosylation
at each hydroxyproline residue is likely, as each HRGP has a unique
arabinosylation profile. Previously we noted a positive correlation
between the contiguity of hydroxyproline residues and the extent of
HRGP O-arabinosylation (Kieliszewski, M., deZacks, R., Leykam,
J. F., and Lamport, D. T. A.(1992) Plant Physiol. 98,
919-926); most arabinosylated hydroxyproline residues and the
longer arabinofuranoside chains occur in HRGPs where Hyp residues occur
as blocks of tetrahydroxyproline, while those with little or no
contiguous Hyp exhibit very little Hyp arabinosylation. In order to
test this Hyp contiguity hypothesis, we have for the first time
determined the arabinosylation site specifics of an HRGP, namely the
proline and hydroxyproline-rich glycoprotein (PHRGP) isolated from
Douglas fir (Pseudotsuga menziesii). Pronase digests of PHRGP
yielded a major peptide and three glycopeptides whose structures were
determined directly from the unfractionated, underivatized Pronase
digest by tandem mass spectrometry using collisionally induced
dissociation. We corroborated the peptide and glycopeptide structures
by Edman degradation, neutral sugar analyses, hydroxyproline
arabinoside profiles, and further mass spectrometric analyses after
purification of the major peptide and glycopeptides by a combination of
hydrophilic interaction and reverse phase column chromatography.
Consistent with the Hyp contiguity hypothesis, the structural analyses
indicate that while the sequence Ile-Pro-Pro-Hyp is never
arabinosylated and Lys-Pro-Hyp-Val-Hyp is only occasionally
monoarabinosylated at Hyp-5, the peptide containing contiguous Hyp,
Lys-Pro-Hyp-Hyp-Val, is always arabinosylated at Hyp-3, mainly by a
triarabinoside. We also obtained precise molecular masses for both
intact and anhydrous hydrogen fluoride-deglycosylated PHRGPs (73.113
and 53.834 kDa) via matrix-assisted laser desorption/ionization time of
flight mass spectrometry, representing the first HRGP to be analyzed by
this method. Hydroxyproline O-glycosylation is a posttranslational
modification unique to plants and Chlorophycean
algae(1, 2) . The hydroxyproline-rich glycoproteins
(HRGPs) ( Most commonly, hydroxyproline substituents are short
arabinofuranoside chains (degree of polymerization: 1-5
residues), usually isolated as hydroxyproline arabinosides
(Hyp-arabinosides); they occur in every type of HRGP examined thus far,
including the Ser-Hyp Recently, we discovered that Hyp-arabinosylation
is not random, but correlated with the contiguity of the Hyp
residues(21) . Thus, the tetrahydroxyproline blocks of the
Ser-Hyp Thus the Hyp
contiguity hypothesis approaches the problem of Hyp arabinosylation
coding by predicting that Hyp arabinosylation increases with Hyp
contiguity rather than with Hyp mole percent, and that noncontiguous
Hyp residues are rarely, if ever, arabinosylated. In order to test and
further refine this hypothesis, we set out to determine the
arabinosylation site specifics for one of the simpler cases of HRGP
glycosylation, notably the proline-rich HRGP (PHRGP) from Douglas fir (Pseudotsuga menziesii). Its Hyp residues are only lightly
arabinosylated, arabinogalactan polysaccharide is absent, and the
polypeptide backbone is largely made up of the simple tandem repeat:
Lys-Pro-Hyp-Val-Hyp-Val-Ile-Pro-Pro-Hyp-Val-Val-Lys-Pro-Hyp-Hyp-Val-Tyr (21) , with low Hyp contiguity. Unique problems arise,
however, in attempts to identify the precise sites of
Hyp-arabinosylation as these involve a base-stable linkage to a
MS/MS using
collisionally induced dissociation (CID) has become increasingly
popular for characterizing the posttranslational modifications of
proteins and
peptides(28, 29, 30, 31, 32, 33) ;
however, it also has drawbacks, as the more hydrophilic components of a
sample, such as glycopeptides, frequently are not detected due to a low
ion yield (32, 33, 34) . Furthermore,
glycopeptides preferentially cleave at O-glycosidic bonds,
while the peptide backbone remains intact(29, 33) ,
which precludes the precise identification of glycosylation sites when
more than one potential glycosylation site is present. Thus, the
sequencing of glycopeptides by MS/MS typically requires chemical
degradation or modification of the glycopeptide before analysis.
Recently, however, Medzihradszky et al.(29) reported
that MS/MS identified both the peptide sequence and carbohydrate
attachment site of a purified, underivatized glycopeptide containing a
single hexose residue. This demonstrated the potential of MS/MS for the
structural characterization of intact, underivatized glycopeptides,
which we have extended to the unfractionated Pronase digests of the
gymnosperm glycoprotein, PHRGP.
SDS-PAGE gave a molecular
weight of 97,400 for PHRGP(21) , while size exclusion
chromatography gave a molecular mass of about 669 kDa( In contrast
to these conventional methods, which have a mass accuracy of only
± 5-10% for the average globular protein(46) ,
MALDI-TOF MS is a straightforward, sensitive (to picomolar levels of
protein), and accurate (
Figure 1:
MALDI-TOF mass spectrometry of
glycosylated (a) and HF-deglycosylated (b) PHRGP. The
MALDI-TOF mass spectrum of glycosylated PHRGP (a) contained
broad peaks corresponding to the triply (M +
Interestingly, the size distribution of the
arabinooligosaccharides is skewed and seems non-random, as we estimate
from PHRGP Hyp-arabinoside profiles ((21) ; cf.Table 1, column 1) and the empirical formula that PHRGP
contains a single Hyp-Ara
Figure 2:
Experimental flow chart. We digested
glycosylated PHRGP with Pronase, then analyzed aliquots of the
unfractionated digest by electrospray ionization (ESMS/MS) and fast
atom bombardment tandem mass spectrometry (FABMS/MS). To corroborate
the results from MS analyses of the Pronase digest and determine the
percent arabinosylation of each glycopeptide, we also purified the
major peptide and glycopeptide components by a combination of
hydrophilic interaction (HILIC) and reverse-phase chromatography. The
purified peptide H2P1 and glycopeptides H3P2 and H4P1 were then
structurally characterized by Edman degradation, sugar analyses, and
mass spectrometry (ESMS and ESMS/MS).
Figure 3:
The PHRGP Pronase digest analyzed by
electrospray ionization mass spectrometry. We freeze-dried aliquots of
the digest to remove ammonium bicarbonate, dissolved the residue in the
appropriate matrix solution (see ``Materials and Methods''),
and then analyzed the solutions by both CF-FABMS (not shown) and ESIMS
(above). We selected the ions common to both spectra (i.e. ions at m/z 439, 701, 833, and 965) for further analysis
by CID and MS/MS.
Figure 4:
Nomenclature of fragment ions produced by
high energy CID analysis of molecular ion m/z 965,
Lys-Pro-[Ara
Figure 5:
MS/MS analyses and corresponding fragment
ion series of PHRGP Pronase glycopeptide,
Lys-Pro-[Ara]
Cleavage of the arabinosyl
rings occurred only in the FAB mass spectrometer (i.e. high
energy CID) giving rise to fragment ions at m/z 905, 875, 861,
729, and 597 which correspond to the The peptide sequence itself was established only by the high energy
CID spectrum, as the similar, but less complex low energy CID analysis (Fig. 5b) was ambiguous regarding the peptide sequence.
However, low energy CID produced fragment ions y
Figure 6:
MS/MS analyses of PHRGP Pronase
glycopeptide, Lys-Pro-Hyp-Val-[Ara] Hyp, M +H
Fragment ions corresponding to
the cleavage of the sugar ring occurred only in high energy CID (Fig. 6a). That is, the ions at m/z 611
([M + 1 - 90]
Consistent with
the Hyp-arabinoside profile of H3P2, the ES mass spectrum (not shown)
yielded three molecular ions, m/z 965, 833, and 701,
corresponding to the tri-, di-, and monoarabinosylated glycoforms,
respectively, which apparently cochromatographed on the HILIC and
reverse-phase columns. The low energy CID spectrum of H3P2 (not shown)
corroborated the structures that had been deduced by CID analyses of
the molecular ions m/z 965 and 833 obtained from the
unfractionated Pronase digest (cf.Fig. 3and Fig. 5). Thus, Lys-Pro-Hyp-HypVal is always glycosylated at
Hyp-3, and usually with a triarabinoside. Such consistent and specific
arabinosylation of dipeptidyl-Hyp is also in accordance with the Hyp
contiguity hypothesis.
Our recent structural work led us to suggest that
Hyp-glycosylation is not random, but follows simple rules, such as Hyp
contiguity(21) . To test the Hyp contiguity hypothesis, we
needed to sequence a Hyp-rich polypeptide and identify the glycosyl
substituents at each position. For reasons already stated, the
determination of HRGP arabinosylation site specifics is of great
interest, albeit a non-trivial task, as it requires the correct
assignment of mono-, di-, tri-, and tetra-arabinosides, or lack
thereof, to each Hyp residue in an HRGP sequence. We have virtually
achieved this, as the glycopeptides characterized here represent the
bulk of the PHRGP glycosylated sequences; we calculate from peptide and
sugar recoveries that glycopeptides H3P2 and H4P1 together contained
89% of the PHRGP arabinose residues (i.e.
Figure 7:
A correlation between Hyp contiguity and
Hyp-arabinosylation. The sequences Lys-Pro-Hyp-Val-Hyp (H4P1) and
Lys-Pro-Hyp-Hyp-Val (H3P2) are peptide structural, or positional,
isomers that differ in the arrangement of their Hyp residues. The
Hyp-arabinoside profile of H4P1 (Table 1) indicated that 10% of
its total Hyp residues were monoarabinosylated, while CID
indicated that arabinosylation occurred only on Hyp-5. Thus, Hyp-5 in
the repetitive sequence Lys-Pro-Hyp-Val-Hyp is monoarabinosylated 20%
of the time. The Hyp-arabinoside profile of H3P2 (Table 1)
indicated that half of the Hyp residues of the sequence
Lys-Pro-Hyp-Hyp-Val were arabinosylated predominantly with the
triarabinoside, while CID pinpointed Hyp-3 as the only glycosylation
site.
Figure 8:
Proposed structure of the PHRGP major
repetitive glycopeptide. Three peptide sequences, H4P1, H2P1, and H3P2 (underlined) and their glycoforms, comprise the bulk of the
PHRGP and occur as part of a larger 18-residue tandem repeat
characterized previously(21) . Featured here is the dominant
PHRGP glycomotif; however, some variation occurs in the chainlength of
the arabinoside (from 1 to 3 residues), and occasionally the single Hyp
located between the 2 valine residues (Hyp-5 of H4P1) is
monoarabinosylated.
Pronase cleavage of PHRGP yielded three major glycopeptides, which
corresponded to glycoforms of the peptide positional isomers,
Lys-Pro-Hyp-Val-Hyp and Lys-Pro-Hyp-Hyp-Val, thereby testing the Hyp
contiguity hypothesis. The results were consistent with the hypothesis
and showed that extensive arabinosylation occurred only for a
contiguous Hyp residue, while non-contiguous Hyp is arabinosylated only
occasionally, i.e. Hyp-5 of Lys-Pro-Hyp-Val-Hyp, or not at
all, as in Ile-Pro-Pro-Hyp. Why Hyp-3 but not Hyp-4 is the inevitable
arabinosylation site in H3P2, while Hyp-5 is only an occasional site in
H4P1, remains for future work. However, Fig. 8shows that the
Hyp-3 triarabinoside of H3P2 is distal rather than proximal to the
Val-Tyr-Lys motif, which for dicot extensins, is a putative
intermolecular cross-link site (14) and therefore far less
likely to sterically hinder cross-link formation than a Hyp-5
triarabinoside. Currently, however, there is no definitive evidence for
PHRGP cross-linking, either in vitro or in muro. Such precise Hyp-arabinosylation suggests a sequence-dependent,
rather than a conformation-dependent, enzymic mechanism, as previously
suggested for O-Thr/Ser glycosylation (52) and proline
hydroxylation(39) . Judging from the number of different
arabinosyl linkages in wall proteins (19, 53) and
polysaccharides(54) , an arsenal of arabinosyl transferases in
plant cells also includes sequence-specific glycosyl transferases. Remarkably, we captured most of the above structural information in
CID spectra of underivatized glycopeptides present in unfractionated
PHRGP Pronase digests. This was possible in part because the simple
repetitive PHRGP polypeptide backbone produced only a few molecular
ions, greatly simplifying MS/MS analyses. However, composition is also
a critical factor as the pyrrolidine rings of Hyp and Pro impose
conformational constraints which probably lower the dissociation energy
of nearby peptide bonds(55) . Thus, HRGPs seem to be uniquely
tailored for structural analysis by CID and tandem mass spectrometry.
Assuming other HRGPs also fragment readily at Hyp and Pro residues
while retaining their saccharide substituents, it may be possible to
determine the glycosylation site specifics of any extensin-HRGP family
member, including the highly arabinosylated
Ser-Hyp
Volume 270,
Number 6,
Issue of February 10, 1995 pp. 2541-2549
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS AND DISCUSSION
CONCLUSIONS
FOOTNOTES
ACKNOWLEDGEMENTS
REFERENCES
)so modified occur predominantly at the cell
surface, where they are implicated in diverse roles ranging from
network formation (3, 4) and disease resistance (5, 6) to cell differentiation and
morphogenesis(7, 8) . HRGPs are characteristically
extended rodlike
molecules(9, 10, 11, 12, 13) with highly repetitive peptide motifs extensively
posttranslationally modified by hydroxylation of
proline(3, 14) , O-glycosylation of serine
and hydroxyproline residues(15, 16) , and both intra-
and intermolecular cross-linking(4, 17) . Thus,
questions about HRGP function must deal not merely with the primary
amino acid sequence as deduced from clones, but also with the mature
glycoprotein whose three-dimensional structure, conformation, and
arrangement in the cell wall all depend on extensive posttranslational
modifications. Of these, hydroxyproline O-glycosylation
figures prominently, accounting for up to 95% of an HRGP's dry
weight (13, 14, 18, 19, 20, 21) ,
and ranges from the addition of a single arabinose or galactose residue
up to a 75-residue
arabinogalactan(2, 13, 18, 19) . 
-containing
extensins(1, 2, 10, 20) ,
arabinogalactan-proteins (AGPs)(2, 18, 22) ,
gum arabic glycoprotein (13) , repetitive proline-rich proteins
(RPRPs)(12, 14, 21) , and the solanaceous
lectins(23, 24) . Each HRGP possesses its own unique
Hyp-arabinoside profile based on the number of substituted Hyp residues
and the relative proportion of each oligoarabinoside chainlength.
Hence, four questions arise about the site specificity of
Hyp-arabinosylation in any given HRGP: what determines (a) the
total number of Hyp residues substituted, (b) the size of each
oligoarabinoside substituent, (c) the precise arrangement of
the different oligoarabinosides, and (d) the positions of the
Hyp-arabinosides relative to the larger arabinogalactan polysaccharide
substituents of the AGPs and gums? Elucidation of the rules for HRGP O-glycosylation will help enable us to predict mature
glycoprotein structure from genomic/cDNA sequences and help us
understand how glycosylation contributes to HRGP molecular topography
and, hence, to the possible roles of HRGPs in molecular recognition and
wall self-assembly.-containing extensins represent highly contiguous
Hyp that is also highly arabinosylated mostly with the larger
(3-4 residues) rather than the smaller (1-2 residues)
oligoarabinosides. RPRPs represent the other extreme, where Hyp
residues are largely interspersed with other residues, i.e. there is little contiguous Hyp (25) and correspondingly
little arabinosylated Hyp(14, 21) .-carbon(1, 2, 15) , which is unlike
glycosylation of seryl or threonyl residues, whose base-labile
-linkage permits -elimination of the carbohydrate with the
concomitant conversion of Ser or Thr to another amino
acid(26) . Nor can direct Edman degradation determine the site
specifics of arabinosylation, for unlike other glycosylated amino acid
residues, which can often be inferred from blank cycles(27) ,
the trifluoroacetic acid-phenylthiohydantoin cleavage step hydrolyzes
the labile arabinofuranosides removing the distinction between
arabinosylated and nonarabinosylated Hyp. These complications led us to
consider using tandem mass spectrometry (MS/MS) to elucidate the
arabinosylation site specifics of the PHRGP.
Preparation of PHRGP from Douglas Fir Suspension
Cultures
We prepared PHRGP from suspension-cultured cells of
Douglas fir as described previously(21) .Deglycosylation of PHRGP with Anhydrous Hydrogen Fluoride
(HF)
We deglycosylated 500 µg of PHRGP (henceforth dPHRGP)
with HF using methods described previously(24, 35) .Digestion of Glycosylated PHRGP with Pronase
PHRGP
(4-55 mg quantities) was digested overnight with Pronase (2%
sodium bicarbonate, pH 8, 10 mM CaCl
;
substrate:enzyme ratio, 100:1). Aliquots of the digest were
freeze-dried and then directly analyzed by electrospray ionization and
fast atom bombardment mass spectrometry.Hydrophilic Interaction Liquid Chromatography
(HILIC)
We fractionated PHRGP Pronase digests on a
polyhydroxyethyl A column (200, inner diameter, 
9.4 mm; 30 nm
pore, PolyLC) at 0.75 ml/min, using gradient elution with the mobile
phase solvents A (30 mM triethylamine/phosphoric acid buffer
(TEAP), pH 3.4, 40% (v/v) aqueous acetonitrile) and B (30 mM TEAP, pH 3.4). The gradient began at 100% A and increased to 50% B
in 40 min. The arabinosides were unstable during concentration in the
TEAP buffer; therefore, after elution of each glycopeptide from the
column, we stabilized the arabinosides by precipitating the phosphoric
acid with saturated Ba(OH) (cf.Table 1,
columns 1-3).
Peptide and Glycopeptide Purification by Reverse Phase
Liquid Chromatography (RPLC)
The HILIC-fractionated peptides and
glycopeptides were purified on a Hamilton polymeric reversed-phase
column (PRP-1; 4.1, inner diameter, 150 mm) using conditions
described previously(21) . Hyp-arabinoside profiles indicated
the arabinosides were stable in the 0.1% trifluoroacetic acid buffers
used for these experiments (data not shown).Amino Acid Composition and Sequence Analysis of the PHRGP
Peptides and Glycopeptides
We determined amino acid compositions
of the peptides and glycopeptides using methods described previously (21) . Peptides were sequenced at the Michigan State University
Biochemistry Department Macromolecular Facility on an Applied
Biosystems, Inc. 477A gas phase sequencer.Hydroxyproline Arabinoside and Neutral Sugar Analyses of
the PHRGP and of the PHRGP Peptides and
Glycopeptides
Hyp-arabinoside profiles and sugar compositions of
the PHRGP glycoprotein and glycopeptides were determined by methods
described previously (21, 36) .Electrospray Ionization Mass Spectrometry (ESIMS) of the
PHRGP Pronase Digest and of the Purified Glycopeptides
ESIMS
spectra were acquired on an API-III Biomolecular analyzer (Perkin-Elmer
Sciex Instruments) operated in the positive ion mode with an ion spray
voltage of 5000 V and orifice potential of 35 V. Solutions of peptides
or glycopeptides (1 µg/µl) in 50% aqueous acetonitrile
containing 0.1% formic acid and 2 mM ammonium formate were
introduced into the ES source at 2 µl/min using a Harvard 22
syringe infusion pump. The mass range was scanned from 100 to 1500
atomic mass units. Ten scans were collected and averaged. For ESIMS/MS
the appropriate protonated molecular ion ([M + H]) was selected in the first
quadrupole, then MS/MS spectra were obtained using CID with argon as
the collision gas (collision gas thickness 300
10/cm
) in the opened-structured quadrupole
collision cell. The fragment ions generated by CID were separated in
the third quadrupole. A total of 50-100 scans (100-1000
atomic mass units) were collected and averaged.Continuous Flow-Fast Atom Bombardment Mass Spectrometry
(CF-FABMS) of the PHRGP Pronase Digest
We acquired CF-FABMS and
CF-FABMS/MS spectra with a JEOL HX/HX110A tandem four-sector mass
spectrometer operated at 10 kV accelerating potential in combination
with a JEOL complement data system. Ions were produced by bombardment
with xenon using a JEOL FRIT-FAB ion source and a JEOL FAB gun operated
at 6 kV. MS1 was calibrated with CsI. MS1 spectra were acquired from
300 to 1500 atomic mass units (scan rate: m/z 1-6000/min). The filtering rate was 300 Hz, and the
resolution was 1000 m/
m. Samples were dissolved in a 1:1 mixture
of water-methanol (1 mg/ml) containing 5% trifluoroacetic acid and 4%
thioglycerol as the CF-FAB matrix. Samples were introduced into the
FRIT-FAB source using a fused silica capillary tube (60 µm
37 cm) at a flow rate of 2 µl/min. CID was performed in the third
field free region using helium as the collision gas at a pressure that
attenuated the primary ion beam by 75%. The collision cell was floated
at 8 kV. Fragment ions were detected by a JEOL MS-ADS11 variable
dispersion array detector(37) . A delay time of 100 ms was used
between magnetic and electric field steps to allow stabilization of the
magnetic field. MS 2 had an approximate resolution of 1000 m/m,
and was calibrated with a mixture of LiI, NaI, KI, RbI, and CsI.Molecular Weight Determinations of Glycosylated and
Deglycosylated PHRGP by Matrix-assisted Laser Desorption/Ionization
Time of Flight Mass Spectrometry (MALDI-TOF MS)
We determined
the molecular mass of the PHRGP and dPHRGP on a Hewlett Packard LDI
1700XP mass spectrometer operated at 30 kV accelerating voltage and a
pressure of 6
10 torr. The mass
spectrometer was calibrated with a mixture of equine heart cytochrome c (M
= 12,360), equine heart
myoglobin (M = 16,950) and bovine serum
albumin (M = 66,465) to give a mass
accuracy of ± 0.2%. Samples were desorbed/ionized from the probe
tip with a nitrogen laser ( = 337 nm) having a pulse width
of 3 ns and delivering approximately 8 µJ of energy/laser pulse.
Aqueous solutions of PHRGP and dPHRGP (10 mg/ml) were diluted 1:8 in
aqueous 90% methanol containing 100 mM sinnapinic
acid(38) . We vacuum-crystallized 1 µl of the sample/matrix
solution onto the probe and then recorded mass spectra over a m/z range of 1-100,000 using a deflector to attenuate the
abundance of ions below 10,000 m/z.
Molecular Masses of PHRGP and dPHRGP
Recently, we demonstrated that PHRGP is a hydroxyproline-rich
glycoprotein homologous with members of the extensin protein family in
that it contains variations of the highly repetitive pentameric motif,
Hyp/Pro-Hyp-Val-X-Lys, which characterizes the RPRPs and other
extensins(14, 21, 25) . The repetitive PHRGP
motifs form a longer 18-residue tandem repeat unit:
Pro-Hyp-Val-Hyp-Val-Ile-Pro-Pro-Hyp-Val-Val-Lys-Pro-Hyp-Hyp-Val-Tyr-Lys,
which is lightly glycosylated with short chains of arabinofuranosides O-linked to hydroxyproline.);
however, neither of these methods provide reliable estimates of HRGP
size, judging from visualization of HRGP monomers by electron
microscopy and polypeptide size as deduced from cDNA
clones(9, 13, 39, 40, 41) .
Indeed, HRGPs behave as if they are much larger than the equivalent
globular protein. Their asymmetric rodlike character arises from a high
pyrrolidine ring content and is further reinforced by glycosylation
(10, 13-16, 22-25, 42, 43). These conformational
constraints maintain the extended conformation, which characterizes all
HRGPs examined(10, 22, 39, 43, 44, 45) and
probably explains their anomalous behavior on gel
filtration(9, 13) . On the other hand, SDS-PAGE
probably overestimates PHRGP molecular weight because Lys-rich
polycations, such as HRGPs, reduce the overall negative charge due to
bound SDS. Glycosylation may further contribute to this effect by
sterically restricting the amount of peptide-bound SDS.
0.1-0.2%; (46) and (47) ) method for measuring molecular mass, as the measurements
are based on mass and charge. Here, we report that MALDI-TOF mass
spectra of glycosylated and HF-deglycosylated PHRGP gave molecular
masses of 73,186 ± 146 Da and 53,953 ± 108 Da,
respectively (Fig. 1). These estimates were significantly
smaller than those obtained by SDS-PAGE or gel filtration and imply
that other estimates of HRGP molecular size based exclusively on
SDS-PAGE or size exclusion chromatography (48) need to be
reevaluated.
H), doubly (M +
2H)
, and singly charged (M + H)
molecular ions at m/z 24538.5,
36415.1, and 73113.2, respectively. b, the spectrum of
deglycosylated PHRGP also showed three peaks at m/z 18044.2,
26945.8, and 53834.5, corresponding the the triply, doubly, and singly
charged species, respectively.
PHRGP Saccharide Content
The molecular mass of PHRGP was 19,233 Da greater than that
of HF-deglycosylated PHRGP (Fig. 1); thus sugar accounts for
26% of the PHRGP mass. Judging from neutral sugar
analyses(21) , this corresponds to 127 arabinose residues and
13 galactose residues in the average PHRGP molecule. Combining earlier
amino acid and carbohydrate analyses (21) and the MALDI-TOF
data, we calculated the following empirical formula for the intact
glycoprotein:
, 40 Hyp-Ara, 10
Hyp-Ara, and 14 Hyp-Ara, with 82 Hyp residues
nonarabinosylated. In order to determine if the arabinosylated Hyp
residues occurred randomly throughout the protein, or in accordance
with the Hyp contiguity hypothesis, we proteolytically degraded the
PHRGP into small glycopeptides and characterized the arabinosylation
sites biochemically and by MS/MS (see flow chart, Fig. 2).
MS Analyses of the PHRGP Unfractionated Pronase Digest
Pronase digestion of PHRGP yielded a relatively few major
peptides and glycopeptides as evidenced by analyses of the digests both
before and after chromatographic separation. Both ESIMS (Fig. 3)
and CF-FABMS mass spectra (not shown) of the unfractionated digest
contained only a few molecular ions; we chose those common to both
spectra, i.e. the signals at m/z 439, 701, 833, and
965, for further analysis by CID and tandem mass spectrometry.
Glycopeptide m/z 965
The high energy CID spectrum of protonated molecular ion m/z 965 contained fragment ions originating from cleavages of
the peptide backbone, O-glycosidic bonds, and the sugar rings
( Fig. 4and Fig. 5, a and c). Fragment
ions from the C terminus: y (Val), y (Hyp-Val),
yY
, (Hyp-Hyp-Val), yY (Pro-Hyp-Hyp-Val), and Y (Lys-Pro-Hyp-Hyp-Val), and
from the N terminus: b (Lys), b (Lys-Pro), and
a (Lys-Pro-[Ara]Hyp-Hyp), defined the
peptide sequence. Fragment ions Y and Y occurred in both the high and low energy spectra (Fig. 5, a and b) and arose from cleavage of the O-glycosidic bonds of the arabinoside side chain only,
corresponding to the mono- and diarabinosylated peptide, respectively.
Most importantly, the glycopeptide peptide backbone fragmented yet
retained its arabinosyl substituents; thus, we were able to determine
which Hyp residue was glycosylated. Although the sequence
Lys-Pro-Hyp-Hyp-Val contains two potential glycosylation sites,
apparently only Hyp-3 is glycosylated judging by the presence of fragment ions y (Hyp-Val),
bY (Lys-Pro-[Ara]Hyp),
yY ([Ara]Hyp-Hyp-Val),
yY ([Ara]Hyp-Hyp-Val),
yY (Pro-[Ara]Hyp-Hyp-Val), y ([Ara]Hyp-Hyp-Val), and a (Lys-Pro-[Ara]Hyp-Hyp), and from the absence of ions at m/z 627, 495, and 363, that would
correspond to the glycosylation of Hyp-4.
]Hyp-Hyp-Val. The nomenclature used
here to describe the PHRGP peptide and glycopeptide fragment ions
combines the Roepstorff and Fohlman system (49) for peptides
(as modified by Biemann; (50) ) with that proposed by Costello
and Vath (51) for glycoconjugates and oligosaccharides. a, fragment ions designated with lowercase letters (a, b
,
and y
) originate from cleavage of the
peptide backbone only, with a
and b
ions arising from the N terminus, and y
ions from the C terminus. Subscript
numbers indicate the residue at which cleavage occurred, numbered
upward from the respective terminus. a and b, uppercaseletters (X
and Y
) designate ions arising
from oligosaccharide fragmentation only (without cleavage of the
peptide backbone), with the charge retained on the
``reducing'' end (i.e. sugar fragments attached to
the peptide). Again, subscripts number the sugar residues from
the reducing end while the superscripts preceding X
ions (e.g.
X
)
define cleavages of carbon-carbon or carbon oxygen bonds within
arabinosyl ring(51) . yY
ions in a arise from fragmentation of both the peptide
backbone and the sugar side chain.
Hyp-Hyp-Val, M + H = 965. Both high energy CID (a) and low energy CID (b) of molecular ion m/z 965 gave similar spectra, indicating that the glycopeptide
contained a triarabinosyl chain at Hyp-3. However, only the high energy
spectrum contained ions y
(Val) and a (Lys-Pro-Hyp-Hyp + 3 Ara), which
distinguished Hyp rather than Val at position 4 of the sequence (a and c). High energy CID also produced X fragment
ions at m/z 905, 875, 861, 729, and 597, which originated from
cleavage within the arabinosyl rings. Because of its intensity, the
molecular ion (M + H =
965) was not included in the high energy spectrum. Fragment ions m/z 922 and 551 correspond to the molecular ion minus the
valine side-chain (-43 atomic mass units) and the deglycosylated
peptide minus water (Y
- 18 atomic mass
units), respectively. b, the low energy CID spectrum lacked
ions defining the sequence of the peptide C terminus, that is, the ions
here correspond to either peptide sequence, Lys-Pro-Hyp-Val-Hyp or
Lys-Pro-Hyp-Hyp-Val. Fragment ions y and yY indicate Hyp-3 is the
arabinosylation site. c, the fragment ion series arising from
high energy CID of molecular ion m/z 965. The masses
corresponding to the fragment ion series are listed either above (y, Y
, and y
Y
ions),
or below (b
, a
, and b
Y
ions) the glycopeptide. The peptide fragmentation site of
ion a
(m/z 198) is not shown in c.
X
, X
, X
, X
, X
ion series (Fig. 4b and Fig. 5a). (Hyp-Val)
and yY ([Ara]Hyp-Hyp-Val),
corroborating Hyp-3 as the glycosylation site.Glycopeptide m/z 833
The high energy CID spectrum (not shown) of protonated
molecular ion m/z 833 (cf.Fig. 3) contained
fragment ions originating predominantly from the cleavage of the amide
linkage, O-glycosidic bonds, and the sugar rings. The peptide
sequence was only partially defined by fragment ions b (Lys), b (Lys-Pro), y (Hyp-Val or
Val-Hyp), yY (Hyp-Hyp-Val or Hyp-Val-Hyp),
yY (Pro-Hyp-Hyp-Val or Pro-Hyp-Val-Hyp), and
Y (Lys-Pro-Hyp-Hyp-Val or Lys-Pro-Hyp-Val-Hyp), as we
observed no ion that indicated whether Hyp or Val occupied position 4
of the sequence. However, the peptide sequence is probably
Lys-Pro-Hyp-Hyp-Val, judging by analyses of the corresponding isolated
glycopeptide H3P2 (i.e. Lys-Pro-[Ara] Hyp-Hyp-Val), which is
discussed later. Fragment ions Y
([M + H - Ara]) and Y
,
([M + H -
2Ara]) indicated that the glycopeptide contained
2 glycosyl residues, while fragment ion y
(Hyp-Val/Val-Hyp), which contained no sugar, and ions y (i.e. [Ara]Hyp-Hyp-Val or
[Ara]Hyp-Val-Hyp), y (Pro-[Ara]Hyp-Hyp-Val or
Pro-[Ara]Hyp-Val-Hyp),
yY(Pro[Ara]Hyp-Hyp-Val or
Pro-[Ara]Hyp-Val-Hyp), and yY ([Ara] Hyp-Hyp-Val or [Ara]Hyp-Val Hyp),
which do contain sugar, suggest that Hyp-3 is the
glycosylation site. We also observed fragment ions arising from the
cleavage of the sugar rings at m/z 773, 743, 729, and 597,
which corresponded to the X
, X
, X
, X
series similar to that in Fig. 4. We did not select
molecular ion m/z 833 for ESIMS/MS analysis due to the low
intensity of molecular ion m/z 833 in the ESI mass spectrum of
the Pronase digest (Fig. 3).Glycopeptide m/z 701
Both high and low energy CID of the Pronase digest protonated
molecular ion m/z 701 determined the peptide sequence as
Lys-Pro-Hyp-Val-Hyp, as both mass spectra contained series of fragment
ions arising from both the C and N terminus (Fig. 6). Both mass
spectra also contained the internal fragment at m/z 310 (i.e. Pro-Hyp-Val), which indicated that Val rather than Hyp
occurred at position 4, and also the fragment ion y (Hyp-Ara), which points to Hyp-5 rather than Hyp-3 as the
arabinosylation site. This was corroborated by ions in the high energy
spectrum, b, b, and a, which
corresponded to the nonglycosylated sequences: Lys-Pro-Hyp and
Lys-Pro-Hyp-Val (Fig. 6a), with no evidence of ions of
fragments containing sugar on Hyp-3.
= 701. Both
high energy (a) and low energy (b) CID spectra of
molecular ion m/z 701 contained fragment ions which determined
its peptide sequence, Lys-Pro-Hyp-Val-Hyp, and monoarabinosylation site
at Hyp-5. The diagnostic ions for the peptide sequence were b
(m/z 438), and internal fragment ion
Pro-Hyp-Val at m/z 310, while fragment ion y determined Hyp-5 as the arabinosylation site. Because of its
intensity, the molecular ion (M + H = 701) was not included in the high energy spectrum. (c)
Fragment ions and masses which belong to the same ion series are
listed in rows above (y, yY, and Y ions), or below (a and b ions) the glycopeptide
sequence. The peptide fragmentation sites of the a ion series
is not shown in c.
) and 597
([M + 1 - 104]
)
correspond to the
X
and X
fragments originating from
cleavage of carbon-carbon and carbon-oxygen bonds of the glycosyl ring (cf. Fig. 4b).Peptide m/z 439
Low energy CID (not shown) of protonated molecular ion m/z 439 (cf. Fig. 3) produced fragment ions at m/z 183 (a), 211 (b), 229
(y), and 326 (y), which were consistent with
the partial sequence Ile-Pro-Pro-Hyp present in the PHRGP 18-residue
repeat reported previously(21) ; however, because Ile and Hyp
have identical molecular masses, the MS/MS-derived peptide sequence was
ambiguous. As such, we determined the sequence by Edman degradation of
the purified peptide, H2P1, described below.Purification and Characterization of PHRGP Peptides
and Glycopeptides Obtained by Pronase Digestion
MS/MS analyses of the PHRGP Pronase digest determined the
arabinoside chain lengths and glycosylation site of each glycopeptide
and established the complete sequences for two peptides. However,
because such MS analyses are not quantitative, we did not know if there
was arabinoside chain-length heterogeneity at the arabinosylation
sites, nor were all of the deduced peptide sequences completely
unambiguous. Therefore, we purified the individual peptides and
glycopeptides and determined the percent glycosylation by
hydroxyproline arabinoside profile analyses, and we corroborated the
peptide sequences by Edman degradation. We also analyzed the purified
peptides and glycopeptides by ESIMS and ESIMS/MS to confirm their
molecular weights and structures (cf. Fig. 2).Purification of (Glyco)peptides by HILIC and RPLC Followed by
Peptide and Glycopeptide Characterization
HILIC fractionation of the PHRGP Pronase digest (not shown)
yielded four major peaks, designated H1-H4, and a few minor
peaks. H1 was largely tyrosine (>90 mol %) released as the free
amino acid by Pronase. HILIC fractions H2, H3, and H4 eluted as single
peaks when fractionated by RPLC, while RPLC fractionation of the minor
HILIC peaks each gave several (i.e. 6 or 7) minor components
that we did not analyze further. Although direct concentration of the
glycopeptides in TEAP/30 mM phosphoric acid buffer led to
significant hydrolysis of the acid labile arabinosides, particularly of
Hyp-Ara, control experiments showed that the arabinosides
were stable after removal of the phosphoric acid by Ba(OH) precipitation (Table 1, columns 1-3).Peptide H2P1 (Ile-Pro-Pro-Hyp) Corresponds to Molecular Ion m/z
439 in the Pronase Digest
Peptide H2P1 (HILIC peak H2, PRP-1
peak 1) was obtained by RPLC fractionation of HILIC peak H2 (not shown; cf. Fig. 2) and sequenced by Edman degradation. H2P1
contained no sugar according to GLC analysis data, while ESIMS
confirmed its mass as 439 atomic mass units (not shown). Low energy CID
analysis of H2P1 (not shown) was consistent with CID of the molecular
ion m/z 439 present in the unfractionated Pronase digest.
Thus, molecular ion m/z 439 from the Pronase digest (cf.Fig. 3) and purified H2P1 both originate from the peptide
sequence Ile-Pro-Pro-Hyp, which contained a single possible
arabinosylation site, yet is never glycosylated: MS analyses of the
PHRGP digest (Fig. 3) showed no evidence for molecular ions
corresponding to mono-, di-, or triarabinosylated Ile-Pro-Pro-Hyp. This
result is consistent with the Hyp contiguity hypothesis, which predicts
that single, non-contiguous Hyp residues are rarely arabinosylated, if
at all.Glycopeptide H3P2,
Lys-Pro-[Ara
Reverse-phase purification of HILIC peak H3 (not shown)
gave one major glycopeptide, designated H3P2 (HILIC peak H3, PRP-1 peak
2), with the Edman sequence Lys-Pro-Hyp-Hyp-Val, which contains
contiguous Hyp residues. A quantitative Hyp-arabinoside profile showed
that approximately one-half of the H3P2 Hyp residues are glycosylated,
mainly with the triarabinoside, although small amounts of the mono- and
diarabinoside also occur (Table 1, column 4).]Hyp-Hyp-Val, Corresponds to
Molecular Ions m/z 965 and 833 in the Pronase
Digest
Glycopeptide H4P1, Lys-Pro-Hyp-Val-[Ara]Hyp,
Corresponds to Molecular Ion m/z 701 in the Pronase
Digest
Fractionation of HILIC peak H4 by RPLC (not shown)
yielded the major component, H4P1 (HILIC peak H4, PRP-1 peak 1), which
was shown by Edman degradation to have the sequence
Lys-Pro-Hyp-Val-Hyp. This peptide is a positional isomer of H3P2
(Lys-Pro-Hyp-Hyp-Val), differing only by the inversion of the Hyp-Val C
terminus. This inversion provides a rigorous test of the Hyp contiguity
hypothesis, as it permits us to compare the arabinosylation specifics
of two peptides of identical composition, but with a crucial
distinction between contiguous Hyp residues (H3P2) and non-contiguous
Hyp residues (H4P1). Again, data for H4P1 were consistent. A
quantitative Hyp-arabinoside analysis of H4P1 showed that 90% of its
Hyp residues were not glycosylated, with the remaining 10%
monoarabinosylated (Table 1, column 5), while the ES mass
spectrum of H4P1 (not shown) contained two molecular ions, m/z 569 and m/z 701, corresponding to the structures,
Lys-Pro-Hyp-Val-Hyp and the monoarabinosylated glycoform,
Lys-Pro-Hyp-Val-[Ara]Hyp, which co-chromatographed on the
HILIC and reversed-phase columns. Low energy CID analysis of H4P1 (not
shown) also confirmed the arabinosylation site as Hyp-5 in agreement
with the high and low energy CID mass spectra of ion m/z 701
obtained from the Pronase digest (cf.Fig. 3and Fig. 6). Thus, a simple inversion of the Hyp-Val C terminus
changes both the site and extent of arabinosylation from consistent
arabinosylation predominantly with a triarabinoside on Hyp-3 (H3P2) to
an occasional monoarabinosylation of Hyp-5.
113 of the 127
residues estimated from the MALDI-TOF mass spectra data). The remaining
arabinose, as well as galactose and the amino acids Ser, His, Arg, and
Thr, which are minor components of the PHRGP, apparently occurs in the
minor HILIC peptides which we did not characterize. Thus, the Douglas
fir PHRGP is not only the first HRGP to be weighed by mass
spectrometry, but also the first for which it has been possible to
define the peptide sequences surrounding the major arabinosylation
sites, pinpoint the precise Hyp residues which are arabinosylated, and
also determine both the frequency of arabinosylation and arabinoside
chain lengths at those sites ( Fig. 7and Fig. 8).
-containing extensins, as well as the AGPs and gums
which contain both arabinosides and polysaccharides O-linked
to Hyp. The Hyp contiguity hypothesis is, therefore, a step toward the
elucidation of more precise Hyp glycosylation codes, ultimately leading
to a complete description of the HRGP molecular topographies that may
be involved in molecular recognition, self-assembly, and morphogenesis
of the extracellular matrix.
))
We thank Drs. Linda Schnabelrauch, Barbara Burgers,
and Derek Lamport for helpful comments on the manuscript, and the
Suntory Institute (Osaka, Japan) for use of their JEOL HX/HX110A tandem
mass spectrometer.
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  J. M. Estevez, P. V. Fernandez, L. Kasulin, P. Dupree, and M. Ciancia Chemical and in situ characterization of macromolecular components of the cell walls from the green seaweed Codium fragile Glycobiology, March 1, 2009; 19(3): 212 - 228. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. S. Karnoup, V. Turkelson, and W.H. K. Anderson O-Linked glycosylation in maize-expressed human IgA1 Glycobiology, October 1, 2005; 15(10): 965 - 981. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Sun, J. Xu, J. Yang, M. J. Kieliszewski, and A. M. Showalter The Lysine-rich Arabinogalactan-protein Subfamily in Arabidopsis: Gene Expression, Glycoprotein Purification and Biochemical Characterization Plant Cell Physiol., June 1, 2005; 46(6): 975 - 984. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. A. Held, L. Tan, A. Kamyab, M. Hare, E. Shpak, and M. J. Kieliszewski Di-isodityrosine Is the Intermolecular Cross-link of Isodityrosine-rich Extensin Analogs Cross-linked in Vitro J. Biol. Chem., December 31, 2004; 279(53): 55474 - 55482. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Tan, F. Qiu, D. T. A. Lamport, and M. J. Kieliszewski Structure of a Hydroxyproline (Hyp)-Arabinogalactan Polysaccharide from Repetitive Ala-Hyp Expressed in Transgenic Nicotiana tabacum J. Biol. Chem., March 26, 2004; 279(13): 13156 - 13165. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Tan, J. F. Leykam, and M. J. Kieliszewski Glycosylation Motifs That Direct Arabinogalactan Addition to Arabinogalactan-Proteins Plant Physiology, July 1, 2003; 132(3): 1362 - 1369. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. G. Spiro Protein glycosylation: nature, distribution, enzymatic formation, and disease implications of glycopeptide bonds Glycobiology, April 1, 2002; 12(4): 43R - 56R. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. M. West, H. van der Wel, and E. A. Gaucher Complex glycosylation of Skp1 in Dictyostelium: implications for the modification of other eukaryotic cytoplasmic and nuclear proteins Glycobiology, February 1, 2002; 12(2): 17R - 27R. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Shpak, J. F. Leykam, and M. J. Kieliszewski Synthetic genes for glycoprotein design and the elucidation of hydroxyproline-O-glycosylation codes PNAS, December 21, 1999; 96(26): 14736 - 14741. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Shpak, E. Barbar, J. F. Leykam, and M. J. Kieliszewski Contiguous Hydroxyproline Residues Direct Hydroxyproline Arabinosylation in Nicotiana tabacum J. Biol. Chem., March 30, 2001; 276(14): 11272 - 11278. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| All ASBMB Journals | Molecular and Cellular Proteomics |
| Journal of Lipid Research | ASBMB Today |