Clinical Chemistry 44: 2264-2269, 1998;
(Clinical Chemistry. 1998;44:2264-2269.)
© 1998 American Association for Clinical Chemistry, Inc.
|
Molecular Diagnostics and Genetics |
Electrospray ionization mass analysis of normal and genetic variants of human serum albumin
Stephen O. Brennan
 |
Abstract
|
|---|
Both normal albumin (Al A) and genetically modified forms were isolated
from six heterozygous subjects. Albumins from each individual were
analyzed by electrospray ionization mass spectrometry (ESI MS), and the
mass was compared with that predicted from the protein sequence. In all
cases, the Al A was heterogeneous, with components of mass (± SE)
66 463 ± 4, 66 586 ± 3, and 66 718 ± 5 Da. Each
genetic variant showed similar heterogeneity. The mass increase in Al
Casebrook (2214 Da) was very close to that predicted (2205 Da) from
protein and carbohydrate sequence analysis, whereas the increase in Al
Redhill (2378 Da) was close to that expected (2392 Da) for an
Arg-albumin with a disialylated N-linked biantennary oligosaccharide
and an Ala
Thr mutation. The circulating proalbumins, Christchurch
and Blenheim, had mass increases of 748 and 756 Da, respectively, over
Al A; in excellent agreement with theoretical values of 744 and 756.
Clear shifts in mass were also detected for the point substitutions
177CysPhe (44 Da), 1AspVal (20 Da), and Arg-albumin (160 Da).
|
Introduction
|
|---|
Human serum albumin, the most abundant plasma protein, consists of
a single 585-residue nonglycosylated polypeptide chain that is
synthesized in the liver as preproalbumin (1). The
prepeptide is cotranslationally cleaved in the endoplasmic reticulum,
and the propeptide (RGVFRR-) is removed in Golgi/vesicles before
secretion from the hepatocyte (2). By virtue of its
concentration, albumin regulates plasma osmotic pressure; however, it
functions primarily as a transport protein and reversibly binds a wide
array of ligands, such as fatty acids, steroids, bilirubin, tryptophan,
and copper (1)(3).
During its 18-day circulatory half-life, the mature molecule is
subjected to a barrage of additional covalent modifications, including
proteolytic cleavage, oxidation, and deamidation, and the covalent
attachment of a variety of endogenous (e.g., glucose, cysteine, and
nitric oxide) and exogenous (e.g., acetylsalicylic acid and penicillin)
compounds (1)(4). In addition to these aging
modifications, genetic variation can superimpose an additional level of
heterogeneity on the population of molecules.
Albumin is probably the most commonly measured serum protein, and
genetic variation, or bisalbuminemia, is usually detected by
electrophoresis of plasma collected for diagnostic reasons.
Bisalbuminemia, however, may also be acquired as a result of
proteolytic cleavage associated with pancreatitis or increased
bilirubin binding in severe liver disease. Some 50 different genetic
variants have been characterized by protein and/or DNA sequence
analysis (5), and these have provided unique insights into
the intracellular processing of albumin and its ligand binding sites.
Indeed, all the variants examined here (Table 1
) have defects that
affect their hepatic processing. Both proalbumin Christchurch (-1RQ)
and proalbumin Blenheim (1DV) have mutations that allow the
proprotein to evade convertase cleavage in the Golgi (6),
whereas albumin Kaikoura (-2RC) has a cryptic signal peptidase site
that favors cleavage after residue -2 instead of after residue -7,
leading to the export of Arg-albumin into the plasma (7).
Albumins Casebrook (494DN) and Redhill (320AT) have mutations
that introduce Asn-Glu-Thr and Asn-Tyr-Thr sequences that act as
acceptors for N-linked oligosaccharides (8)(9).
In addition, albumin Redhill has a -1 Arg residue, the result of an
additional -2RC mutation. The final variant examined, albumin Hawkes
Bay (177CF), has a defect that abolishes one of its disulfide bonds,
greatly decreasing its plasma concentration (10).
Electrospray ionization mass spectrometry (ESI MS) is being applied
increasingly to the molecular analysis of complex proteins. Impressive
results have been obtained for human hemoglobin which, because of the
comparatively small size of its component chains (16 126 and 15 876
Da) and lack of posttranslational modifications, produces clean spectra
with high ion currents. Indeed, a number of genetic variants of
hemoglobin have been detected and characterized using ESI MS
(11)(12).
The traditional methods of mutation detection rely on identifying a
functional or physical change in a protein; electrophoresis,
isoelectric focusing, and reversed-phase HPLC permit the segregation of
gene products on the basis of charge, conformation, and polarity. The
accurate determination of protein molecular mass by ESI MS provides not
only a novel means of mutation analysis but may indicate the type of
change involved (e.g., amino acid substitution, truncation/elongation,
or possible carbohydrate attachment). Where a number of polypeptide
chains make up a particular protein, such as in the case of hemoglobin
(13) or fibrinogen (14), then MS can often
identify which chain has the molecular lesion
(13)(14). Protein fragmentation followed by mass
analysis of, for example, tryptic or CNBr peptides, can then be used to
rapidly pinpoint a mutation site, and this can be zeroed in on by
peptide sequencing. Alternately, targeted regions of a gene could then
be amplified by polymerase chain reaction and subjected to direct
DNA sequence analysis (15).
The theoretical average isotopic mass of disulfide-linked albumin A is
66 438 Da. Using ESI MS, Loo et al. (16) noted the
heterogeneity of albumin from a number of different species and
reported the molecular mass of human as 66 605 Da. Subsequently, Yang
et al. (17) reported a value of 66 745 Da for human serum
albumin, whereas a value of 66 440 Da was determined for (a more
homogeneous) recombinant human albumin derived from yeast expression
(18).
In this investigation, we isolated both normal and variant forms of
albumin from the plasma of six heterozygous subjects with different
genetic mutations and compare measured mass changes with those expected
on the basis of the aberrant protein sequence. This confirms that ESI
MS can be used to detect and characterize both genetic and circulatory
modifications of albumin.
|
Materials and Methods
|
|---|
Electrophoretic variants of albumin were purified from the plasma
of six heterozygous subjects who gave informed consent and under
protocols approved by Christchurch Hospital. Briefly, 1–10 mL of
plasma was dialyzed against 16 mmol/L sodium acetate buffer, pH 5.2,
and applied to a DEAE-Sephadex A-50 column (1.6 x 16 cm)
equilibrated in the same buffer (6). The column was eluted
with a linear gradient of 300 mL of this buffer and 300 mL of 16 mmol/L
sodium acetate buffer, pH 4.3. A flow rate of 1 mL/min was used, and
the column effluent was monitored at 280 nm. The normal and variant
albumin peaks were collected, concentrated to 2–3 mL in a 50-mL Amicon
cell, diluted to 15 mL with water, and reconcentrated. This latter
dialysis step was repeated an additional two times before the protein
was freeze-dried and stored at -20 °C. Some albumins had been
stored in this way for up to 5 years. Analysis by agarose gel
electrophoresis demonstrated that each isolated albumin component was
95% pure. All the mutant albumins used in this investigation had
been characterized earlier by protein and/or DNA sequence analysis
(6)(7)(8)(9)(10) and, in one case, carbohydrate analysis
(19).
Both the normal and variant albumin components from each plasma were
redissolved at between 1.5 and 2.5 g/L in 500 mL/L acetonitrile–1 mL/L
formic acid and analyzed by ESI MS on a VG Platform quadrupole
instrument (Micromass) operating in positive ion mode (20).
Ten to twenty microliters of each protein was injected into the source
at a flow rate of 5 µL/min. The probe was charged at +3000 V, and the
source maintained at 60 °C. The mass range 1000–2000 m/z
was scanned every 2.5 s, and a cone voltage ramp of 40–60 V was
applied over this range. Between 60 and 100 scans were collected and
averaged for each analysis, using high and low and high mass resolution
parameters of 14. Calibration was made over this same m/z
range (voltage ramp, 50–80 volts), using the charge series generated
by human 
-globin. For this a 100 g/L red cell lysate calibrator was
diluted to 0.5 g/L in 1 mL/L HCOOH–500 mL/L acetonitrile
(11), and ions from m/z 1009.43 to 1891.80 were
used for the calibration. Data were acquired and processed using
Mass-Lynx software and transformed onto a true molecular mass scale
using maximum entropy (Max-Ent) software supplied with the instrument.
Max-Ent deconvolution of the hemoglobin calibrator yielded molecular
masses of 15 126 and 15 867 Da for the and ß chains,
respectively. Molecular masses are based on the following atomic
weights: C, 12.011; H, 1.00794; N, 14.00674; O, 15.9994; P, 30.97376;
and S, 32.006.
For S-carboxymethylation, albumin (10 g/L) was first reduced for
16 h with 0.1 mol/L dithiothreitol in 0.1 mol/L Tris-HCl, pH 8.0,
8 mol/L urea. After reduction, the albumin was reacted for 45 min with
a 2.5 mol/mol excess of iodoacetic acid (pH 8.0) over total thiol
groups and then dialyzed against water and freeze-dried.
|
Results
|
|---|
The essential requirement of any MS analysis is that the analyte
exists as a gas phase ion. In protein solutions, ionization is easily
achieved by raising or lowering the pH, and vaporization is effected by
rapid (heat- and nebulization-assisted) evaporation as the protein
solution enters the atmospheric pressure ion source. In
electrospray spectra there is typically little ion fragmentation,
and what is seen is a series of molecular ions, each differing from its
neighbor by one charge.
With native albumin in 500 mL/L acetonitrile–1 mL/L formic acid the
charge envelope usually extended from +60 to +32. Portions of the ion
spectra of normal albumin (Al A) from two of the subjects is shown in
Fig. 1
. The ions depicted extend from [M+51H] to [M+41H] and have
a charge to mass ratio extending from 1306.0 to 1624.6 m/z.
What is immediately apparent is that each m/z peak is
reproducibly heterogeneous and is composed of at least three or four
different components. As in the two cases depicted, the Al A gene
product from the other four subjects all showed very similar
heterogeneity, although the ratio of the individual components varied.
All variant ion spectra displayed similar heterogeneity, although their
precise m/z positions varied in accordance with their
specific mutations.
View larger version (35K):
[in this window]
[in a new window]
|
Figure 1. ESI mass spectra of purified Al A from individuals
heterozygous for proalbumin Christchurch (lower spectrum)
and proalbumin Blenheim (upper spectrum).
For clarity, only a portion of the ion spectra (from [M+51] to
[M+41]) are shown. Raw data were background-subtracted and smoothed
using the Savitzky function supplied with Mass-Lynx.
|
|
Ion spectra spanning an m/z range of 1100–1900 were
transformed onto a true mass scale using Max-Ent software; two of the
transformed Al A spectra are shown (Fig. 2
, a and f). Data from these and the Al A components from the
other four subjects are summarized in Table 1
. The heterogeneity that was detected in the raw data is
reflected in these results. The three principal components had mean
molecular masses (±SE) of 66 463 ± 4; 66 586 ± 3; and
66 718 ± 5 Da, respectively. The peak at 66 463 Da is in good
agreement with the predicted mass (66 438 Da) of Al A. The other two
components are heavier by 123 ± 2 and 255 ± 5 Da,
respectively, probably reflecting the attachment of cysteine (119 Da)
and palmitic acid (256 Da). Similar mass differences were apparent
between the components in the spectra of isolated genetic variants
(Fig. 2
, b–e, g, and h) where mean mass increases of 120 ± 4 and
254 ± 6 Da were observed. The presence of these different
isoforms within the purified albumin peaks did not, however, compromise
the ability to discern mass changes associated with specific mutations.
This heterogeneity was greatly diminished after reduction and
S-carboxymethylation, where transformed spectra indicated a mass of
68 524 Da, in good agreement with a predicted value of 68 504 Da
(Fig. 3
).
View larger version (29K):
[in this window]
[in a new window]
|
Figure 2. Transformed ESI spectra of albumin variants.
(a–c) The three albumin components from Blenheim plasma, Al
A, albumin Blenheim, and proalbumin Blenheim, respectively;
(d) purified proalbumin Christchurch; (e)
purified Arg-albumin; (f and g) Al A and albumin
Casebrook from Casebrook plasma; (h) albumin Redhill.
Subtracted raw data were transformed using Max-Ent software.
|
|
View larger version (11K):
[in this window]
[in a new window]
|
Figure 3. Transformed ESI mass spectrum of
S-carboxymethylalbumin A.
The expected mass after carboxymethylation is 68 504. Note that the
heterogeneity is substantially reduced, and signals at +119 and +255 Da
are removed. Conditions were as for native albumin except that raw data
were scanned between 700 and 1500 m/z.
|
|
To determine the effect of mutation on molecular mass, each individual
variant was analyzed immediately after its corresponding Al A gene
product (i.e., within the same batch). To ascertain the error
associated with these comparisons, nine repeat injections were made of
Al A from a single individual. In this case the dominant component had
a molecular mass of 66 448 Da with an SE of ±2 Da. All sequence
changes associated with the variants, together with predicted and
calculated mass changes, are summarized in Table 1
. For simplicity the
putative Al-Cys peaks are used as the basis of comparison in the
ensuing discussion. Predictably the 1AspVal substitution in albumin
Blenheim would cause a 16-Da decrease in the mass in the mature
protein, whereas the expected increase in the resulting proalbumin
Blenheim would be 756 Da. These values are in excellent agreement with
the measured changes of -20 and +756 Da, respectively, for these two
species. The 756-Da change is clearly indicative of a major change in
protein structure. However, given the within-batch SE of ±3 Da
associated with this difference, the 20-Da decrease can only suggest
the possibility of a point mutation. The mutation in this case,
however, can be readily confirmed electrophoretically because of the
charge change associated with the AspVal substitution. The other two
N-terminal extended forms, Arg-albumin and proalbumin Christchurch, had
measured increases of 160 and 748 Da, respectively, over their normal
gene products, compared with expected increases of 156 and 744 Da.
The introduction of new Asn-Glu-Thr and
Asn-Tyr-Thr sequences in albumins Casebrook and Redhill
cause the incorporation of an N-linked oligosaccharide side chain into
each of these proteins. The structure of the side chain has been
determined as
-NAcGlc-NAcGlc-Man-(Man-NAcGlc-Gal-NAcNeu)2 for
albumin Casebrook (19); however, the structure of the side
chain has not been investigated in the case of albumin Redhill.
With the 494AspAsn mutation (-1 Da) and oligosaccharide (2206 Da),
the expected mass increase in albumin Casebrook is 2205 Da, in
excellent agreement with a measured increase of 2214 Da. The observed
increase in mass in albumin Redhill was 2378 Da, whereas the increase
associated with the -1 Arg (156 Da), the 318AlaThr substitution (30
Da), and the attachment of a similar oligosaccharide (2206 Da) is 2392
Da. The closeness of these values indicates for the first time that
albumin Redhill has a fully sialylated biantennary oligosaccharide side
chain.
Albumin Hawkes Bay is a low concentration variant with aberrant
disulfide bridging resulting from a substitution of 177CysPhe. The
direct mass increase of this mutation is 44 Da, with an additional
increase of 1 Da arising from the formation of the unpaired -SH group;
this value (45 Da) was again very close to the measured value of 44 Da.
|
Discussion
|
|---|
In previous studies, researchers using ESI MS analysis of
commercially available human serum albumin noted the heterogeneity of
its ion spectra and reported transformed molecular masses of 66 605
(16) and 66 754 Da (17). It is clear from this
investigation that this heterogeneity can be resolved, and individual
isoforms with masses of 66 463 ± 8; 66 586 ± 7; and
66 718 ± 11 Da were identified for Al A. The lower value of
66 463 Da is very close to the theoretical value of 66 438 Da for
virgin native albumin; indeed, Geisow et al. (18) determined
a value of 66 440 Da for recombinant human albumin expressed in yeast
and isolated directly from culture supernatants. Albumin, however,
undergoes a barrage of modifications during its average 26-day voyage
through the circulation. Undoubtedly some minor modified forms might
have been purified out during ion-exchange chromatography, but three
principal components were resolved.
The mean mass differences between resolved isoforms of Al A were
123 ± 4 and 255 ± 12 Da, and similar isoforms were present
in the variants, with mean differences of 120 ± 10 and 255
± 16 Da. The 123- and 120-Da differences most probably represent the
covalent attachment of cysteine (119 Da) to the free thiol at position
34 of the normal and variant albumin, respectively, because it is well
established that this thiol is blocked in >33% of the circulating
molecules. In support of this, the area of this peak was seen to
decreased substantially after incubation of the albumin for 4 h in
2 mol/L dithiothreitol (not shown), and the peak disappeared entirely
after reduction and carboxymethylation (Fig. 3
).
The species with an increased mass of 255 ± 14 Da most probably
represents noncovalently bound palmitic acid (256 Da), some of which
seems to remain bound despite the presence of 500 mL/L acetonitrile.
Although covalent palmitylation is not excluded, the expected mass
increase would be 238 and not 256 Da. Furthermore, if esterification
was involved it would be expected to survive the effects of reduction
in 8 mol/L urea and S-carboxymethylation. This was not the case; after
this treatment the albumin was converted to a more homogeneous peak
with no indication of a +255 Da signal.
Because in ESI MS there is a trade-off between resolution and
sensitivity, it was important to use relatively high protein
concentrations (1.5–2.5 g/L) and low flow rates (5 µL/min) so that
between 60 and 100 m/z scans could be collected and averaged
for each analysis. The sequential analysis of Al A and its companion
variant also facilitated more accurate determination of mass
differences between the two allelic products.
In the case of albumin Casebrook, the mass increase of 2214 Da confirms
a fully sialylated biantennary oligosaccharide. The same 494AspAsn
mutation was recently identified in Italy, and ESI MS analysis of the
isolated glycopeptide CNBr 6 indicated that ~10% of the molecules
lacked one of their sialic acid residues (21). These two
findings are not in conflict because the extent of protein sialylation
can vary greatly between individuals. For example, the amount of fully
sialylated fibrinogen Bß chain was found to vary from 10% to 60% in
six healthy individuals (20). The mass increase in Albumin
Redhill (2378 Da) indicates for the first time that it too has a fully
sialylated biantennary oligosaccharide in addition to its two
mutations.
The results here establish the utility of ESI MS in the direct analysis
of underivatized genetic variants of serum albumin and other large
proteins. The technique allows the rapid identification of genetic
variation, posttranslational modifications, and circulatory changes.
|
Acknowledgments
|
|---|
This investigation was supported by the Canterbury
Medical Research Foundation, Lottery Health, the Health Research
Council of New Zealand, and the University of Otago.
|
Footnotes
|
|---|
Molecular Pathology Laboratory, Canterbury Health Laboratories, Christchurch Hospital, Christchurch, New Zealand. Fax 64-3-364-0750; e-mail SOB{at}chmeds.ac.nz.
|
References
|
|---|
-
Peters T Jr. All about albumin: biochemistry
genetics and medical applications, San Diego, CA: Academic Press,
1996:431.
-
Brennan SO, Peach RJ. The processing of human proinsulin and chicken proalbumin by rat hepatic secretory vesicles suggests a single convertase specific for X Y Arg Arg or Arg X Y Arg sequences. J Biol Chem 1991;266:21504-21508.
[Abstract/Free Full Text]
-
Kragh-Hansen U. Molecular aspects of ligand binding to serum albumin. Pharmacol Rev 1981;33:17-53.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Kashiba-Iwatsuki K, Miyamoto M, Inoue M. Effect of nitric oxide on the ligand binding activity of albumin. Arch Biochem Biophys 1997;345:237-242.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Madison J, Galliano M, Watkins S, Minchiotti L, Porta F, Rossi A, et al. Genetic variants of serum albumin in Italy: point mutants and a carboxyl-terminal variant. Proc Natl Acad Sci U S A 1994;91:6476-6480.
[Abstract/Free Full Text]
-
Brennan SO, Peach RJ, Boswell DR. Novel human proalbumin variant with intact dibasic sequence facilitates identification of its converting enzyme. Biochim Biophys Acta 1989;993:48-50.
[Medline]
[Order article via Infotrieve]
-
Brennan SO, Arai K, Madiso J, Laurell C-B, Galliano M, Watkins S, et al. Hypermutability of CpG dinucleotides in the propeptide sequence of the human albumin gene. Proc Natl Acad Sci U S A 1990;87:3909-3913.
[Abstract/Free Full Text]
-
Peach RJ, Brennan SO. Structural characterization of a glycoprotein variant of human albumin: Al Casebrook 494AspAsn. Biochim Biophys Acta 1991;1097:49-54.
[Medline]
[Order article via Infotrieve]
-
Brennan SO, Myles T, Peach RJ, Donaldson D, George PM. Albumin Redhill (-1 Arg; 320AlaThr): a glycoprotein variant of human albumin whose precursor has an aberrant signal peptide cleavage site. Proc Natl Acad Sci U S A 1990;87:26-30.
[Abstract/Free Full Text]
-
Brennan SO, Fellowes AP. Albumin Hawkes Bay (177CysPhe): a variant of human serum albumin caused by loss of a sulphydryl group at 177. Biochim Biophys Acta 1993;1182:46-50.
[Medline]
[Order article via Infotrieve]
-
Witkowska HE, Bitsch F, Shackleton CHL. Expediting rare hemoglobin variant identification by combination HPLC/electrospray mass spectrometry. Hemoglobin 1993;17:227-247.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Shackleton CHL, Witkowska HE. Characterizing abnormal hemoglobins by MS. Anal Chem 1996;68:29A-33A.
[Medline]
[Order article via Infotrieve]
-
Brennan SO, Mathews JRD. Hb Auckland (87HisAsn): a new mutation of the proximal histidine identified by electrospray mass spectrometry. Hemoglobin 1997;21:393-403.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Fellowes AP, Brennan SO, Ridgway HJ, Heaton DC, George PM. Electrospray ionisation mass spectrometry identification of fibrinogen Banks peninsula (
280TyrCys): a new variant with defective polymerisation. Br J Haematol 1998;101:24-31.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Chua EKM, Brennan SO, George PM. Albumin Church Bay: 560LysGlu: a new mutation detected by electrospray ionisation mass spectrometry. Biochim Biophys Acta 1998;1382:305-310.
[Medline]
[Order article via Infotrieve]
-
Loo AJ, Edmonds CG, Smith RD. Tandem mass spectrometry of very large molecules: serum albumin sequence information from multiply charged ions formed by electrospray ionisation. Anal Chem 1991;63:2488-2499.
[Medline]
[Order article via Infotrieve]
-
Yang YZ, Asawamahasakda W, Meshnick SR. Alkylation of human albumin by the antimalarial artemisin. Biochem Pharmacol 1993;46:336-339.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Geisow MJ, Harris R, Dodsworth N, Green B, Hutton T. Characterisation and quality assurance of recombinant human serum albumin. Villafranca JJ eds. Techniques in protein chemistry 1991:567-572 Academic Press San Diego, CA. .
-
Haynes PA, Batley M, Peach RJ, Brennan SO, Redmond JN. Characterisation of oligosaccharides from a glycoprotein variant of human serum albumin (Al Casebrook) using high performance anion exchange chromatography. J Chromatogr 1992;581:187-193.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Brennan SO. Electrospray ionisation analysis of human fibrinogen. Thromb Haemost 1997;78:1055-1058.
[Web of Science][Medline]
[Order article via Infotrieve]
-
Minchiotti L, Watkins S, Madison J, Putnam FW, Kraig-Hansen U, et al. Structural characterisation of four genetic variants of human serum albumin associated with alloalbuminemia in Italy. Eur J Biochem 1997;247:476-482.
[Web of Science][Medline]
[Order article via Infotrieve]
Top
Abstract
Introduction
