Anal Bioanal Chem (2013) 405:18751884

DOI 10.1007/s00216-012-6294-y

ORIGINAL PAPER

Selenium speciation in paired serum and cerebrospinal

fluid samples
Nikolay Solovyev & Achim Berthele & Bernhard Michalke

Received: 21 May 2012 / Revised: 18 July 2012 / Accepted: 20 July 2012 / Published online: 7 August 2012
# Springer-Verlag 2012

Abstract Se speciation was performed in 24 individual

paired serum and cerebrospinal fluid (CSF) samples from
neurologically healthy persons. Strong anion exchange
(SAX) separation, coupled to inductively coupled plasma
dynamic reaction cellmass spectrometry (ICP-DRC-MS),
was employed. Species identification was done by standard
matched retention time, standard addition and by size exclusion chromatography followed from SAX (2-D SECSAX-ICP-DRC-MS) and by SAX followed from CE-ICPDRC-MS (2-D SAX-CE-ICP-DRC-MS). Limit of detection
(LoD, 3standard deviation (SD) of noise) was in the range
of 0.0260.031 g/L for all investigated species and thus
was set uniformly to 0.032 g/L. Quality control for total Se
determination was performed by analysing control materials
human serum and urine, where determined values met
target values. Several Se species were found in both sample
types having following median values (sequence: serum/
CSF, each in g Se/L): total Se, 58.39/0.86; selenoprotein
P (SePP), 5.19/0.47; Se-methionine (SeM), 0.23/<LoD; glutathione peroxidase (GPx), 4.2/0.036; thioredoxinreductase
Published in the topical collection Metallomics with guest editors Uwe
Karst and Michael Sperling.
N. Solovyev
Institute of Toxicology, FMBA,
St. Petersburg 192019, Russia
A. Berthele
Department of Neurology, Klinikum rechts der Isar,
Technische Universitt Mnchen,
81675 Munich, Germany
B. Michalke (*)
Research Unit Analytical BioGeoChemistry,
Helmholtz Center MunichGerman
Research Center for Environmental Health,
Ingolstdter Landstr. 1,
85764 Neuherberg, Germany
e-mail: bernhard.michalke@helmholtz-muenchen.de

(TrxR), 1.64/0.035; Se IV, 12.25/0.046; Se-human serum

albumin (Se-HSA), 18.03/0.068. Other Se species, such as
Se-cystine (SeC), Se VI and up to four non-identified compounds were monitored (if ever) only in very few samples
usually close to LoD. Therefore, their median values were
<LoD. Linear relationships based on median values provide
information about Se-species passage across neural barriers
(NB): SePP-serum is significantly correlated to total Se-serum
when the latter was >65 g/L; however, SePP-CSF appeared
independent of SePP-serum. For Se-HSA-serum versus (vs.)
Se-HSA-CSF, a weak linear relationship was found (r2 0
0.1722). On the contrary, for anti-oxidative Se-enzymes,
higher r2 values were calculated: GPx-serum vs. GPx-CSF,
r2 00.3837; TrxR-serum vs. TrxR-CSF, r2 00.6293. Q-Se-species
values (0 ratios of CSF-Se-species/serum-Se-species) were compared with the Q-Alb value (HSA-CSF/HSA-serum 0clinical
index of NB integrity) for deeper information about NB
passage of Se species. The Q-Se-HSA value (3.8103) was
in accordance to the molecular mass dependent restriction at
NB (Q-Alb at 5.25103). Increased Q values were seen for
TrxR (21.3 103) and GPx (8.3 103) which are not
(completely) explained by molecular size. For these two
anti-oxidative Se-enzymes (GPx, TrxR), we hypothesize
that there might be either a facilitated diffusion across NB
or they might be additionally synthesized in the brain.
Keywords Selenium speciation . Cerebrospinal fluid .
Serum . Thioredoxin reductase . Glutathione peroxidase

Introduction
Selenium is an essential trace element for humans. Its beneficial role for human health is generally accepted and
includes protection against oxidative stress, prevention of
heart diseases and cancer or optimal immune responses

1876

[16]. Innate and adaptive immunity are impaired in Sedeficient individuals. Increased Se intake boosted Ca2+ flux
and downstream cell signaling in CD4+ T cells, which
influenced their activation, proliferation and differentiation
[7]. The biological effects of Se are exerted mainly through
its incorporation into selenoproteins. There, Se is incorporated as SeC where it has been shown to inhibit NF-kappa B
activation, which leads to reduced expression of NF-kappa
B target genes, including inflammatory cytokines [8]. Thyroid metabolism, too, is dependent on Se as deiodinases are
selenoproteins. Some selenoproteins, such as GPx and
TrxR, have been functionally characterized as antioxidant
enzymes, serving to mitigate damage caused by reactive
oxygen species (ROS) and to regulate redox tone [9]. Both
GPx and TrxR are present in brain [10]. In the central
nervous system, oxidative stress has been implicated in
pathophysiology of neurodegenerative diseases, such as
Alzheimers (AD) and Parkinsons disease (PD) and even
stroke [1012]. Fahn and Cohen [13] published data indicating oxidative stress specifically in the substantia nigra of
PD patients. Brain Se levels are primarily maintained by
selenoprotein P (SePP) directly expressed in neuronal tissue
[14], which in turn depends on proper Se supply across the
bloodbrain barrier (BBB). Cerebrospinal fluid (CSF) is
mainly an excretion of the choroid plexus in the brain
ventricles. It plays an important role in the homeostasis
and metabolism of the central nervous system. Since CSF
has close molecular exchange with the extracellular space of
brain parenchyma, the composition of CSF and extracellular
brain fluids are similar and a misbalance, a depletion of
elements or change of element species in the brain is likely
to be reflected in CSF, which is summarized by Siverling
[15]: Cerebrospinal fluid is a vital matrix for understanding
the events ongoing in the brain as the medium bathes the
brain, yet does not come into contact with other organs or
fluid compartments under normal physiological conditions,
thereby producing highly dependable information
concerning brain pathology. Therefore, Se depletion or
changes in Se speciation in neurological diseases can be
monitored in the CSF. Total Se concentration in CSF was
described to be low in healthy control subjects (approximately 1.24 g/L) in recent papers which adhere to sufficient quality control (QC) measures [1618].
The transport mode of Se species to the brain across NB
is not completely understood. In previous work, we have
shown that the total Se concentration in CSF is independent
from serum Se concentration and a strictly controlled, regulated pathway across NB was assumed [17]. Se species in
CSF were investigated where four Se compounds were
identified but several remained unknown [18]. A promising
concept to understand, in which form Se is crossing the NB,
is the analysis of paired serum (before NB) and CSF samples (behind NB). No Se speciation data of such paired

N. Solovyev et al.

samples are published to the best of our knowledge, except

for SePP. In mice, SePP was found to be transcribed in brain
independently from liver-borne SePP [14].
In the present paper, the analysis of Se species in paired
serum and CSF of 24 individuals (a 2 replicates) was
performed with SAX-ICP-DRC-MS. The performance for
species identification was 2D by SEC-SAX- and SAX-CEICP-DRC-MS. The DRC mode of the mass spectrometer
was chosen for removing interferences on Se isotopes. A
previously developed SAX method from Xu et al. [19] was
modified and used for separation of SePP, SeM, GPx, SeC,
TrxR, Se-HSA, selenite and selenate.

Experimental
Chemicals and reagents
Chemicals and reagents used throughout this work were of
suprapure grade. The chemical list consists of the following:
certified selenium and rhodium stock standards (1,000 mg/L,
CPI, Santa Rosa, USA); selenite, selenate, SeM, SeC, TrxR
(EC 1.8.1.9.), GPx (EC 232-749-6), human serum albumin
(HSA) and TRIS buffer (Sigma-Aldrich, Deisenhofen, Germany); protein standards for mass calibration of a size exclusion column were -globuline (150 kDa), arginase (107 kDa),
glutathione peroxidise (86 kDa) transferrin (78 kDa), albumin
(66.5 kDa), -lactoglobuline (36.5 kDa), lysozyme (14.3 kDa),
metallothionein (7 kDa), each from Sigma-Aldrich; ammonium acetate and acetic acid (Merck, Darmstadt, Germany); Arliq
and methane (99.999 % purity, Air Liquide, Grbenzell,
Germany).
Samples
Serum and CSF sample pairs, drawn at the Department of
Neurology of the Technische Universitt Mnchen, were
obtained from 24 patients with unspecific neurological complaints, like headache, dizziness and various sensory symptoms. CSF and serum samples were collected and handled as
described previously [18]. In short terms, CSF was collected
from each individual by standardized lumbar puncture and
serum was obtained from blood drawn from the cubital vein
directly after the spinal tab. Thus CSF and serum are referred to
as paired samples. In the case of unremarkable CSF test
results, CSF and serum samples were considered to origin
from neurologically healthy individuals. After patients consented to the use of their samples for scientific investigations,
the previously aliquoted, frozen-stored samples were subjected
to total Se determination by FI-ICP-DRC-MS [17] and subsequently to Se speciation. The samples were thawed at 4 C in
the refrigerator, vortexed (and only for serum samples: diluted
1/5 with Milli-Q water) and injected to SAX-ICP-DRC-MS.

Selenium speciation in paired serum

Selenite and selenate stock solutions were prepared at a

concentration of 1,000 mg Se/L by dissolving in Milli-Q
water (18.2 M cm, Milli-Q system, Millipore, Bedford,
MA, USA). HSA was prepared accordingly at a concentration of 1,000 mg/L. Preparation of an Se-HSA was done by
mixing 10 mg Se/L selenite to this stock solution and
incubation for at least 14 days. Working standards of Se
species were prepared daily from their stock standard solutions by appropriate dilution with Milli-Q water.
HPLC parameters
Strong anion exchange (SAX)
SAX separation was based on the method of Xu et al. [19],
but slightly modified, to ensure baseline separation of close
eluting peaks mainly in the middle of the chromatographic
gradient [18]. A Knauer 1100 Smartline inert Series gradient
HPLC system was connected to an anion exchange column
ProPac SAX-10 (2502 mm ID) from Dionex (Idstein,
Germany). The sample volume was 100 L. The mobile
phases were: eluent A, 10 mM TrisHAc buffer, pH 8.0; and
eluent B, A+500 mM ammonium acetate, pH 8.0. Gradient
elution was as follows: 03 min 100 % eluent A; 310 min
10060 % eluent A (040 % eluent B); 1023 min 6045 %
eluent A (4055 % eluent B); 2326 min 4543 % eluent A
(5557 % eluent B); 2628 min 430 % eluent A (57
100 % eluent B); 2852 min 100 % eluent B, 5260 min
100 % eluent A. The flow rate was 0.20 mL min1. The
column effluent was mixed with 1 g/L Rh (final concentration) as internal standard (total flow rate: 0.25 mL min1)
and directed to ICP-MS.
Size exclusion chromatography (SEC)
SEC was used to prepare mass range-characterized SEC
fractions for a 2D approach for species identification. The
Knauer 1100 Smartline inert HPLC system was connected
to a Biobasic 300mesh column (3007.8 mm ID, Fischer
Scientific, separation range ca. 8003 kDa). TrisHAc
(10 mM, pH 7.4) +250 mM NH4Ac was used as the eluent
at a flow rate of 0.75 mL min1. Mass calibration was
performed using protein standards of defined molecular
mass. Retention times related to molecular masses followed
the equation ln(Da)0263RT+13.597 (r00.9916). For
subsequent analysis by the regular SAX-ICP-MS method,
the SEC column effluent was fractionated by a Fraction
Collector 100 (Pharmacia, Freiburg, Germany). Fractions
were frozen at 20 C and freeze dried (Heraeus-Christ,
type Beta, Osterode, Germany). The freeze-dried, sizecharacterized fractions were dissolved in 300 L Milli-Q
water and kept frozen again until SAX-ICP-DRC-MS
measurements.

1877

Capillary electrophoresis (CE)-ICP-DRC-MS

A Biofocus 3000 capillary electrophoresis system (BioRad, Munich, Germany), equipped with an uncoated capillary (CS-Chromatographie Service GmbH, Langerwehe,
Germany) 120 cm50 m ID was used for CE-ICP-DRCMS analysis. Hyphenation is detailed in [20]. Before each
run, the capillary was purged with NH4-acetate/acetic acid
buffer, 10 mM, pH 3.0 (70 s, 8 bar). Pressurized sample
injection was performed for 2 s, followed from 1 s buffer
injection. The separation voltage was +25 kV. A sheath flow
(diluted running buffer 1/25) around outlet electrode and
capillary end at nebulizer provided the electrical connection.
Preparation of SePP from human serum as a standard
SePP is not commercially available. However, several
papers describe a selective purification from human plasma
using affinity chromatography [21]. This method is used
also by Jitauru et al. for Se-speciation investigations [22].
We slightly modified the elution program according to the
column instruction sheet. Briefly, 200 L serum was
injected on a Heparin-affinity column (Amersham, GE
Healthcare Europe GmbH, Munich, Germany) and chromatographed by a linear 12-min lasting gradient of A050 mM
Tris, 10 mM NH4-acetate/acetic acid, pH 6.0 to buffer B0A
but 800 mM NH4-acetate, pH 8.5, and remaining at 100 % B
for further 8 min. SePP was eluted at a flow rate of
1.0 mL min1 and collected under 280 nm monitoring. Se
was subsequently determined in fractions with FI-ICPDRC-MS [17]. The SePP fraction was preconcentrated by
freeze drying and was re-dissolved in 1 mL of 10 mM Tris
HCl buffer, pH 7.2. The resulting SePP solution was immediately frozen and stored at 20 C until use. For verification, an aliquot of this affinity chromatographic SePP
fraction was analysed additionally on the thoroughly mass
calibrated SEC column, where it eluted at RT calculated for
60 kDa, which fits to literature data [23].
Inductively coupled plasma mass spectrometry
Table 1 shows the experimental settings chosen for ICPDRC-MS after optimization.
Total Se determination and peak quantification from
chromatograms
Flow injection analysis ICP-DRC-MS was applied for total
Se determination according to our previously published
method [17]. In short terms, a Knauer 1100 Smartline inert
Series binary HPLC system equipped with vacuum degasser
and an electronic valve with a 25-L injection loop (Perkin
Elmer, Rodgau-Jgesheim, Germany) was directly coupled

1878

N. Solovyev et al.

Table 1 The instrumental settings for ICP-DRC-MS

Instrument
Plasma conditions
RF power (W)
Plasma gas flow (L/min)
Auxiliary gas flow (L/min)
Nebulizer gas flow (L/min)
Nebulizer (optimal flow rate
according to provider)
Mass spectrometer settings
Dwell time (ms)
Sweeps per reading
Readings per replicate
Autolens
Ions montored
Reaction gas
Reaction gas flow rate (mL/min)
Rejection parameter, q
Rejection parameter, a

Perkin Elmer NexIon DRC

1,250
15
1.05
0.98 daily optimized
Meinhard low flow (300 L/min)

300
6
1,000
On
78
Se, 80Se,
CH4
0.6
0.6
0

103

Rh

to ICP-DRC-MS. The flow rate was 1 mL min1 of 50 %

eluent A and B, each from SAX separation. CSF samples
were diluted 1:1.1 by adding 1 g/L Rh (final conc.) as
internal standard.
Peak quantification from FI-quantification and from
chromatograms was done by comparing peak areas with
peak area calibration curves from FI-ICP-DRC-MS. Standard addition method was used for standard retention time
matched identification of Se species and as QC means in
quantification.
Data processing of FI-ICP-(DRC)-MS
Rh and Se data files were exported from the NexIon software and processed with the Knauer HPLC software Clarity for peak area integration. For each sample (or standard),
a quotient of Se-peak area to Rh peak-area was calculated
and taken as the result corrected for the internal standard
(Rh). These values were used for the calibration curve
(standards) or for calculating the concentration according
to the calibration curve (samples).
Quality control experiments
Total Se determination Quality control for total Se determination was performed by analysing control materials
human serum and urine from Recipe, Munich. Control materials were reconstituted as indicated on flask
labels and the resulting solutions were diluted 1/50
(serum) or 1/10 (urine) with Milli-Q water before
measurements.

Mass balances during SAX-ICP-DRC-MS Defined amounts

of single Se species (related to Se: solutions prepared at
10 g/L) of those Se species which were regularly above
LoD, were injected to the SAX-ICP-DRC-MS system and
peak Se concentrations were quantified. These Se concentrations were related to the injected Se amounts (0100 %)
for recovery calculation. Analogous, serum or CSF samples
were quantified for total Se (0100 %) before injection. The
sum of eluted and quantified peaks was subsequently related
to this total Se concentration.
Species identification Additionally to standard retention time
match and standard addition, species identification was performed by two 2D approaches: (a) mass range defined SEC
fractions were analysed with the regular SAX-ICP-DRC-MS
method and (b) fractions from SAX separation were analysed
by CE-ICP-DRC-MS. Species identification was regarded as
OK when species matched the standard compounds after
serial analysis with both chromatography/electrophoretic
techniques (match in first and second technique).

Results and discussion

Quality control results
Checking total Se determination by analysis of urine and serum
control materials resulted in Se concentrations of 613 g/L for
serum or 243 g/L for urine. The manufacturers target mean
values of 62 g/L for serum and 23 g/L for urine were found.
Mass balances during SAX-ICP-DRC-MS were calculated as described above. Recoveries were 10611 % for
serum, 969 % for CSF, 10610 % for GPx, 978 % for
Se IV, 1028 % for TrxR, 898 % for Se-HSA, and 84
13 % for SePP. The lower recovery for SePP is explained by
stability problems accompanied by an increasing Se VI
signal, both already observed previously [18].
Species identification using two 2D approaches
A) Figure 1 shows chromatograms of two SEC fractions
from serum (mass range 6644 kDa, containing SePP,
Se-HSA and (traces of) TrxR or from CSF (mass range
9580 kDa, containing GPx). These SEC fractions were
analysed by SAX-ICP-DRC-MS, i.e. serially first characterized by SEC and second by SAX. In the 6644 kDa
SEC fraction, SePP (3.5 min), Se-HSA (23.5 and
37.5 min, see below) and traces of TrxR (19.5 min) are
eluting at their specific SAX retention times. Thus, these
proteins appeared in the correct size fraction and at
their specific retention times in SAX. Analogous, in the
9580 kDa fraction from CSF, human GPx (86 kDa) was
collected and appeared subsequently at its typical

Selenium speciation in paired serum

1879

SEC fraction from serum: MW 66 kDa - 44 kDa, containing:

SePP, Se-HSA (1+2), [traces of TrxR]

80Se/103Rh

(2)

0
0

10

20

30

40

50

min
0.9

SEC fraction from CSF: MW 95 kDa - 80 kDa, containing: GPx

grey line: GPx standard
5

80Se/103Rh

0.7

0.5

0.3
0

10

15

20

25

30

35

40

45

min

Fig. 1 Species identification by a 2D approach: mass rangecharacterized SEC fractions from serum or CSF were subsequently
analysed by the SAX-ICP-DRC-MS method. Retention time matches
were found in SEC (correct mass fraction) and in SAX (correct RT).
Peak assignment (top): 10SePP, (2)0traces of TrxR, 30Se-HSA-1, 40
Se-HSA-2. (Bottom): 50GPx, compared to GPx standard

retention time in SAX. This human GPx fraction also coeluted with a GPx standard, commercially prepared from
bovine erythrocytes. Therefore, the latter was used as RT
standard for GPx from our human samples.
B) For further strengthening, the peak identification SAX
fractions were analysed with CE-ICP-DRC-MS, too. As
an example, Fig. 2 shows electropherograms of two
fractions of regular SAX separation from serum, collected at typical RT of GPx or of TrxR and Se IV. Observed
peaks match the migration times of respective standard
compounds. By both approaches, we found a match in
both used techniques (SEC-SAX or SAX-CE). Species
identification therefore was regarded as sufficient.
Results on Se species in serum and CSF
According to our previous experience from [18], where
several peaks were eluting close to each other and

consequently identification with standard-matched method

was partly difficult and time-consuming, we changed the
elution gradient and flow rate to values given in the Experimental section. This resulted in a flatter gradient in the
mid-chromatogram and peaks appeared with more distance
to each other. With this optimized method, the subsequent
analysis of Se standard mixtures, serum and CSF samples
were performed. Figure 3 shows examples of chromatograms from CSF and serum.
Notably, Se-HSA showed two elution peaks, the first
with a narrow peak shape at 25.4 min (where it had been
expected from previous experiments) and surprisingly a
second broader elution at 37.2 min. This observation was
made first with Se-HSA single standards but also in samples
(CSF and serum) with and without standard addition, i.e. the
two peak signals were seen in all sample types in both, UVand ICP-MS detection. Therefore, the sum of both peaks
was considered as total Se-HSA. A double peak for HSA
was shown also by Xu et al. (there Fig. 1) [19]. The method
of these authors was taken as a basis for our Se speciation
method. Aside from identified compounds, such as SePP,
SeM, GPx, SeC, TrxR, Se IV, Se-HSA and Se VI also
unknown peaks (U0unknowns) were monitored in some
of the samples, where no standards matched the retention
times or met peaks after standard addition. Table 2 shows
the Se concentrations of total Se, the median and concentration range (minimum, maximum) of Se species from 24
paired serum and CSF samples.
For both, total Se from serum and CSF, the values were
lower than found in our previous studies from 2009 and
2011 [17,18], but are in agreement with other findings e.g.
from reference [14]. The presence of SePP in both sample
types was to be expected as SePP had been found in serum
and CSF previously. In this study, SePP concentration in
serum is less compared to findings of other authors [19,22].
In turn, Se IV in part showed considerable amounts, which
could point to stability problems of SePP, observed already
previously [18]. In CSF, however, the median of 0.474 g/L
Se from SePP corresponds roughly with 15 g/L SePP
(whole protein) from a former, preliminary approach
(Schweizer and co-workers (Ulrich Schweizer, J. Khrle,
B. Michalke, unpublished results).
Scharpf et al. [14] found SePP-CSF independent from
SePP-serum, locally expressed in the brain. Further, SePP
seems to play an important role in neuronal survival by
protection against reactive oxygen species (ROS) [24]. The
presence of TrxR fits well to the supposed protective action
against ROS [25]. The primary defense line against oxidative stress is based on superoxide dismutase activity, which
however, is resulting in H2O2 production. Subsequently,
peroxides are eliminated by the activity of the selenoenzymes TrxR and GPx, both being known to be expressed
in brain of animals [10]. Aside from TrxR, this time we also

1880
SAX fraction (17 - 19 min) containing GPx
3000

2500

GPx standard
Se, cps

2000

1500

80

Fig. 2 Species identification by

another 2D approach: Fractions
of SAX separation from serum
are analysed by CE-ICP-DRCMS. Top: The SAX fraction at
GPx-retention time shows also
matched peak in CE-ICP-DRCMS analysis. Bottom: The SAX
fraction at TrxR and Se IVretention time shows matched
peaks in CE-ICP-DRC-MS
analysis. Retention time
matches thus are found for SAX
(correct RT/fraction) and CE
(correct migration time). For
improved visibility standard
electropherograms (grey) are
plotted with an off-set

N. Solovyev et al.

1000

500

SAX fraction
0
0

10

15

20

min

SAX-fraction (19-21 min) containing TrxR and Se IV

3500

Se IV
3000

2500

80

Se, cps

TrxR
2000

1500

Se IV and TrxR standard

1000

500

SAX fraction
0
0

10

15

20

min

found GPx. This was an improvement compared to our investigations in 2011, where we had observed stability problems
for GPx [18]. The presence of GPx is in accordance with Arnr
et al. [25], who had determined GPx in brain. Pyne-Geithman
et al. [26] analysed GPx in CSF, too. In serum, GPx was
already determined from several authors previously in a range
of ca. 1018 g/L [2,19]. The GPx Se concentration found in
this study strives this range at the lower level (cf. Table 2).
The finding of Se-HSA in CSF was also not surprising.
HSA is not synthesized in the brain, but crosses NB to a certain
extent. Actually, the HSA quotient of CSF/serum is the standard mean to evaluate the intactness of the barrier. Se-HSA is
generally accepted for serum [e.g. 19,27,28]. Therefore SeHSA, like HSA, is transported in small amounts across NB
into CSF. Jitaru et al. [22] presented Se-speciation in serum
where Se-HSA amounted to 19 g/L. This is in accordance to
the median value from this study being 18.4 g/L.
To understand in which form selenium is crossing the NB,
Se species median values from serum and CSF were taken
from Table 2 and relationships were calculated. First, this was
done for Se-species-serum vs. Se-total-serum to elucidate which

of the Se-compounds mainly account for Se-total concentration

in serum. Relationships were evaluated according to their
linear regression coefficients, where r2 was assigned to:
<0.10no relationship, independent; 0.10.250low influence;
0.250.40some influence; 0.40.50clear influence; 0.50.60
high influence; >0.60leading influence.
In serum, it became apparent that Se-total concentration vs. SeHSA (r2 00.011), or vs. Se IV (r2 00.0099), or vs. Se VI (r2 0
0.0088) and vs. GPx (r2 00.0265) did not show linear correlation. The influence of those Se species on total Se concentration
in serum seemed to be low. Contrary, this was different for
SePP: a leading influence was seen for SePP-serum on total Se2
serum (r 00.9332), however, only as long as total Se was >65 g/
L in serum. Below this concentration, no clear relationship could
be confirmed (r2 00.0229). This is shown in Fig. 4.
The strong relationship between SePP-serum and total Seserum is known in general, since SePP is known as main Se
species in serum [19]).
When elucidating Se-species-serum vs. Se-species-CSF, the
independence of total Se-CSF from total Se-serum found previously (r2 00.0001, [17]) was confirmed (this time, r2 0

Selenium speciation in paired serum

1881

Fig. 3 Examples of
chromatograms from serum and
CSF samples are seen. In total,
paired samples from 24
individuals (a 2 replicates)
were analysed. Replicate
measurements resulted in equal
chromatograms

SePP

1.2

CSF

0.2

Se VI

Se-HSA1

Se-HSA2

Se IV
GPx
U

SeC

0.4

TrxR

0.6

(SeM)

Se/Rh

0.8

0
0

10

20

30

40

50

SePP

min
7

serum

Se VI

Se-HSA2

Se IV

SeC

GPx

(SeM)

TrxR

Se/Rh

Se-HSA1

0
0

10

20

30

40

50

min

0.0002). Similarly, SePP-CSF was independent of SePP-serum

(r2 00.0365). This result fits well to the local SePP expression in brain being independent of SePP-serum, reported by
Scharpf et al. [14].

As for SePP, no linear relationship

CSF could be found for several of
compounds, except however for GPx,
For these latter compounds, positive

between serum and

the remaining SeSe-HSA and TrxR.
linear relationships

Table 2 Concentrations of Se species in serum and CSF

Serum
Se, g/L
Mean total
59.67
CSF, Se, g/L
Mean total
0.958

Min
Max
Median
Min
Max
Median

total Se

SeP

SeM

U1

U2

U3

U4

GPx

SeC

TRxR

Se(IV)

HSA

Se(VI)

36.22
91.60
58.39
0.274
2.371
0.861

1.55
50.60
5.19
0.040
1.213
0.474

<LoD
19.59
0.23
<LoD
0.116
<LoD

<LoD
7.40
<LoD
<LoD
0.352
<LoD

<LoD
8.73
<LoD
<LoD
0.562
<LoD

<LoD
9.00
<LoD
<LoD
<LoD
<LoD

<LoD
22.10
6.34
<LoD
<LoD
<LoD

<LoD
9.56
4.27
<LoD
0.334
0.036

<LoD
19.33
<LoD
<LoD
0.035
<LoD

<LoD
32.11
1.64
<LoD
0.296
0.035

<LoD
24.68
12.25
<LoD
0.534
0.046

2.31
29.85
18.03
<LoD
0.647
0.068

<LoD
2.62
<LoD
<LoD
0.750
<LoD

Ranges are given as minimum and maximum concentrations; additionally, the median values are shown. LoD, calculated as 3SD of noise, was in
the range of 0.0260.031 g/L for all investigated species and thus was set uniformly to 0.032 g/L

1882

N. Solovyev et al.

Fig. 4 The relationship of

SePP from serum to total Se
concentration in serum is
shown: A leading influence is
seen for SePP-serum on total Se2
serum (r 00.9332), as long as
total Se is >65 g/L. Below this
concentration, no clear relationship could be confirmed
(r2 00.0229)

were observed. Still only small r2 values were calculated for

Se-HSA-serum vs. Se-total-CSF (r2 00.1650) or vs. Se-HSA-CSF
GPx grouped means: CSF vs. serum
0.14

0.10

y = 0.0091x + 0.0044
R2 = 0.7348

0.08
0.06

Q-Se-species

0.04

0,025
21.34 x 10-3

0.00
0.00

2.00

4.00
6.00
8.00
GPx in serum, g Se/L

10.00

12.00

TrxR grouped means: CSF vs. serum

TrxR in CSF, g Se/L

0.35
0.3
0.25

y = 0.0188x - 0.0013
R2 = 0.8193

0.2
0.15
0.1

Q [median (CSF) / median (serum)]

0.02

0,02

0,015

0,01

8.31 x 10-3

5.25 x 10-3
3.82 x 10-3

0,005

3.85 x 10-3

0.05
0
0

4 5 6 7 8 9 10 11 12 13 14
TrxR in serum, g Se/L

Fig. 5 xy plots are shown for GPx and TrxR, where grouped-mean
values (00.5, 0.51, 11.5g/L) from serum-borne Se species are
related to the means of the respective CSF-borne Se species. This
allowed showing the relationships of GPx or TrxR between serum
and CSF more clearly. r 2 values are 0.7348 (GPx) and 0.8193
(TrxR). Slopes are 0.0091 (GPx) or 0.0188 (TrxR), which demonstrate
a moderate influence of GPx-serum on GPx-CSF, but a more profound
influence of TrxR-serum on TrxR-CSF

measure for
NB integrity

GPx in CSF, g Se/L

0.12

(r2 00.1722). This can be explained by the regular NB function based on barrier integrity, which is usually expressed by
Q-Alb, which is the quotient of HSA-CSF/HSA-serum. Q-Alb
values from our samples (median, 5.25103) were in the
normal range for healthy adults (age-dependent range, 3
103 to 8103, [29]) and thus, only very little amount of
Se-HSA (or HSA in general) could pass NB. Protein transfer
from blood to CSF is restricted by the laws of diffusion and
consequently influenced by molecular size [30].

0
Q-GPx

Q-TrxR

Q-Se (IV)

Q-SeHSA

Q-Alb

Fig. 6 The Q values (ratios of Se species: CSF/serum) of median

values from Se species are shown for those Se compounds which had
median values >LoD. These values are compared to Q-Alb, which is the
clinical evaluation parameter for NB integrity. Further information
about the passage of Se species across NB is gained. Since SePP is
accepted to be expressed in brain independently from serum, it is not
considered here

Selenium speciation in paired serum

In contrast to Se carriers, such as SePP or Se-HSA, the

linear relationships of Se enzymes protecting against oxidative stress were stronger expressed: a clear linear relationship could be confirmed for GPx -serum vs. GPx -CSF
(r2 00.3837). GPx together with TrxR-serum are known as
typical Se enzymes protecting against oxidative stress
[15,16,25,26]. Interestingly, TrxR also showed a considerable linear relationship between serum and CSF: TrxR
revealed a correlation coefficient at a comparable level as
GPx when related to Se-total-CSF (r2 00.3836), but r2 was
significantly higher, reaching a value of r2 00.6293, when
being related to TrxR-CSF. This confirmed a leading influence of TrxR-serum on TrxR-CSF.
However, the r2 values determined for GPx and TrxR
appeared lower than expected from the xy plots which
showed apparently clear relationships. This was suspected
to be due to high variance of those values which were close
to LoD. For compensating variances of values close to LoD
but maintaining slope and equation, an additional evaluation
model was used, where grouped-mean values (00.5, 0.51,
11.5g/L) from serum-borne Se species were related to
the means of the respective CSF-borne Se species. This
resulted in xy plots showing the same relationships clearer
with r2 values of even 0.7348 (GPx) and 0.8193 (TrxR).
Slopes were the same as without grouped-mean calculations
and demonstrated an influence of GPx-serum on GPx-CSF, and
again a leading influence of TrxR-serum on TrxR-CSF. Figure 5
shows the y-x-plots of grouped-mean values relationships
for GPx and TrxR.
The concentration ratios (CSF/serum) were calculated similar to Q-Alb for those Se species which were regularly detectable in both sample types, i.e. where median values were >LoD,
to get further information about their passage across NB. Since
SePP is accepted to be expressed in brain independently from
serum, this Se species was not considered here.
The ratios are shown in Fig. 6 and compared to Q-Alb.
With an intact neural barrier (Q-Alb in normal range, as in
our paired samples) protein transfer from blood to CSF is
restricted by the laws of diffusion and consequently influenced by molecular size [30]. Therefore, HSA, having a
molecular size of 66 kDa, is crossing NB only in small extent
with a Q-Alb of 5.25103. However, the increased ratios of
TrxR (55 kDa) and GPx (86 kDa) are not explained by a
molecular size dependence, as GPx is even larger than SeHSA and TrxR ratio is increased much more than expected
regarding the somewhat smaller molecular mass. This is clearly seen when calculating molecular mass dependence at NB
from Q values of various proteins and applying this value to a
molecular mass of 55 kDa (TrxR): From molecular masses (of
proteins) and Q values provided by Reiber et al. [29] a simple
calibration of Q vs. ln(molecular mass) is calculated, resulting
in the linear equation: y00.2649x+5.5736. Based on this
equation the Q value for 55 kDa (TrxR) is 5.91103, which

1883

is close to Q-Alb, but significantly lower than the actually

measured Q-TrxR of 21.34103.
Thus, we hypothesize, that there could be a facilitated
diffusion across NB for the anti-oxidative Se-enzymes GPx
and TrxR or that they might be additionally expressed in brain.
Conclusion
For the investigation of selenium species distribution on
both sides of NB, 24 individual paired serum and cerebrospinal fluid samples from 24 neurologically healthy persons
were analysed. SAX-ICP-DRC-MS was optimized and used
for analysis. Species identification was proofed by use of
two 2D approaches. Linear regression coefficients were
calculated to evaluate relationships between species concentrations present in serum and CSF, as well as between
species concentration and total selenium content.
In serum, a leading influence on total Se-serum was seen
only for SePP-serum when total Se was >65 g/L. The
previously described independence of total Se-CSF on total
Se-serum and of SePP-CSF on SePP-serum was confirmed.
Linear relationships of Se compounds between serum
and CSF could be found only for GPx, Se-HSA and TrxR
with varying r2 values. Q-Alb values from our samples were
in the normal range for adults. However, the increased Q
values of TrxR and GPx are not completely explained by a
molecular size dependence, as GPx is even larger than SeHSA and TrxR ratio is increased much more than expected
regarding the somewhat smaller molecular mass. Thus, we
suggest that there may be a facilitated diffusion across NB
for GPx and TrxR or that they might be additionally
expressed in brain.
Acknowledgments The authors thank Katharina Fernsebner for
reading the manuscript.

References
1. Morenoa P, Quijano MA, Gutirrez AM, Prez-Conde MC,
Cmara C (2004) Study of selenium species distribution in biological tissues by size exclusion and ion exchange chromatography
inductively coupled plasmamass spectrometry. Anal Chim Acta
524:315327
2. Ganther HE (1999) Selenium metabolism, selenoproteins and
mechanisms of cancer prevention: complexities with thioredoxin
reductase. Carcinogenesis 20:16571666
3. Rayman MP (2000) The importance of selenium to human health.
Lancet 356:233241
4. El-Bayoumy K (2001) The protective role of selenium on genetic
damage and on cancer. Mutat Res 475:123139
5. Birringer M, Pilawa S, Floh L (2002) Trends in selenium biochemistry. Nat Prod Rep 19:693718
6. Rayman MP (2009) Selenoproteins and human health: insights from
epidemiological data. Biochim Biophys Acta 1790:15331540

1884
7. Hoffmann FW, Hashimoto AC, Shafer LA, Dow S, Berry MJ,
Hoffmann PR (2010) Dietary selenium modulates activation and
differentiation of CD4+ T cells in mice through a mechanism
involving cellular free thiols. J Nutr 140:11551161
8. Pahl HL (1999) Activators and target genes of Rel/NF-kappaB
transcription factors. Oncogene 18(49):68536866
9. Reeves MA, Hoffmann PR (2009) The human selenoproteome:
recent insights into functions and regulation. Cell Mol Life Sci 66
(15):24572478
10. Schweizer U, Bruer AU, Khrle J, Nitsch R, Savaskan NE (2004)
Selenium and brain function: a poorly recognized liaison. Brain
Res Rev 45:164178
11. Andersen JK (2004) Oxidative stress in neurodegeneration: cause
or consequence? Nat Med 10(Suppl):S18S25
12. Aguilar V, Jimenez-Jimenez FJ, Molina JA, Meseguer I, MateosVega CJ, Gonzales-Munoz MJ, de Bustos F, Gomez-Escalonilla C,
Orti-Pareia M, Zurdo M, Martinez-Para MC (1998) Cerebrospinal
fluid selenium and chromium levels in patients with Parkinsons
disease. J Neural Transm 105:12451251
13. Fahn S, Cohen G (1992) The oxidative stress hypothesis in Parkinsons disease: evidence supporting it. Ann Neurol 32:802812
14. Scharpf M, Schweizer U, Arzberger T, Roggendorf W, Schomburg
L, Khrle J (2007) Neuronal and ependymal expression of selenoprotein P in human brain. J Neural Transm 114:877884
15. Siverling JM (2009) Master of Science in Analytical Chemistry,
Cerebral Vasospasm post Subarachnoid Hemmorhage: an exploration of selenospecies and Se quantification in cerebrospinal fluid.
Department of Chemistry of the College of Arts and Sciences,
University of Cincinnati
16. Gerhardsson L, Lundh T, Minthon L, Londos E (2008) Metal
concentrations in plasma and cerebrospinal fluid in patients with
Alzheimers disease. Dement Geriatr Cogn Disord 25:508515
17. Michalke B, Grill P, Berthele J (2009) A method for low volume
and low Se concentration samples and application to paired cerebrospinal fluid and serum samples. J Trace Elem Med Biol
23:243250
18. Michalke B, Berthele A (2011) Contribution to selenium speciation in cerebrospinal fluid samples. J Anal Atom Spectrom
26:165170
19. Xu M, Yang LM, Wang QQ (2008) Quantification of seleniumtagged proteins in human plasma using species-unspecific isotope

N. Solovyev et al.
dilution ICP-DRC-qMS coupled on-line with anion exchange
chromatography. J Anal At Spectrom 23:15451549
20. Michalke B (2004) Manganese speciation using capillary
electrophoresis-ICP-mass spectrometry. J Chromatogr A 1050:6976
21. Shigeta K, Sato K, Furuta N (2007) Determination of selenoprotein
P in submicrolitre samples of human plasma using micro-affinity
chromatography coupled with low flowICP-MS. J Anal At Spectrom 22:911916
22. Jitaru P, Prete M, Cozzi G, Turetta C, Cairns W, Seraglia R, Tralfi
P, Cescon P (2008) Speciation analysis of selenoproteins in human
serum by solid-phase extraction and affinity HPLC hyphenated to
ICP-quadrupole MSJ. Anal Atom Spectrom 23(3):402406
23. Ballihaut G, Kilpatrick LE, Kilpatrick EL, Clay Davis W (2012)
Multiple forms of selenoprotein P in a candidate human plasma
standard reference material. Metallomics 4:533538
24. Yan J, Barrett JN (1998) Purification from bovine serum of a
survival-promoting factor for cultured central neurons and its
identification as selenoprotein-P. J Neurosci 18(21):86828691
25. Arnr ES, Holmgren A (2000) Physiological functions of thioredoxin
and thioredoxin reductase. Eur J Biochem 267(20):61026109
26. Pyne-Geithman GJ, Caudell DN, Prakash P, Clark JF (2009) Glutathione peroxidase and subarachnoid hemorrhage: implications
for the role of oxidative stress in cerebral vasospasm. Neurol Res
31(2):195199
27. Eberle B, Haas HJ (1993) Purification of selenoprotein Ph from
human plasma. J Trace Elem Electrolytes Health Dis 7(4):217221
28. Shigeta K, Matsumura K, Suzuki Y, Shinohara A, Furuta N (2008)
Distribution and dynamic pathway of selenium species in
selenium-deficient mice injected with 82Se-enriched selenite. Anal
Sci 24:11171122
29. Reiber H, Thompson EJ, Grimsley G, Bernardi G, Adam P,
Almeida SM, Fredman P, Keir G, Lammers M, Liblau R,
Menna-Barreto M, S MJ, Seres E, Sindic CJM, Teelken A,
Trendelenburg C, Trojano M, van Antwerpen MP, Verbeek MM
(2003) Quality assurance for CSF protein analysis: international
consensus by an internet-based group discussion. Clin Chem Lab
Med 41:331337
30. Hhmer AF, Biringer RG, Amato H, Fonteh AN, Harrington MG
(2006) Protein analysis in human cerebrospinal fluid: physiological aspects, current progress and future challenges. Dis Markers 22
(12):326