Substrate and inhibitor specificities differ between human cytosolic and
mitochondrial thioredoxin reductases: Implications for development
of specific inhibitors
Oliver Rackhama, Anne-Marie J. Shearwooda, Ross Thyera, Elyshia McNamaraa, Stefan M.K. Daviesa,
Bernard A. Callusb, Antonio Miranda-Vizuetec, Susan J. Berners-Priced, Qing Chenge,
Elias S.J. Arnére, Aleksandra Filipovskaa,⁎
aWestern Australian Institute for Medical Research, Centre for Medical Research, University of Western Australia, Perth, WA 6000, Australia
bSchool for Biomedical, Biomolecular and Chemical Sciences, University of Western Australia, Perth, WA 6000, Australia
cInstituto de Biomedicina de Sevilla, Hospital Universitario Virgen del Rocío/CSIC/Universidad de Sevilla, E-41013 Sevilla, Spain
dInstitute for Glycomics, Gold Coast Campus, Griffith University, Queensland, QLD 4222, Australia
eDivision of Biochemistry, Department of Medical Biochemistry and Biophysics, Karolinska Institutet, SE-171 77 Stockholm, Sweden
a b s t r a c ta r t i c l e i n f o
Received 16 August 2010
Revised 9 December 2010
Accepted 10 December 2010
Available online 21 December 2010
The cytosolic and mitochondrial thioredoxin reductases (TrxR1 and TrxR2) and thioredoxins (Trx1 and Trx2)
are key components of the mammalian thioredoxin system, which is important for antioxidant defense and
redox regulation of cell function. TrxR1 and TrxR2 are selenoproteins generally considered to have
comparable properties, but to be functionally separated by their different compartments. To compare their
properties we expressed recombinant human TrxR1 and TrxR2 and determined their substrate specificities
and inhibition by metal compounds. TrxR2 preferred its endogenous substrate Trx2 over Trx1, whereas
TrxR1 efficiently reduced both Trx1 and Trx2. TrxR2 displayed strikingly lower activity with dithioni-
trobenzoic acid (DTNB), lipoamide, and the quinone substrate juglone compared to TrxR1, and TrxR2 could
not reduce lipoic acid. However, Sec-deficient two-amino-acid-truncated TrxR2 was almost as efficient as
full-length TrxR2 in the reduction of DTNB. We found that the gold(I) compound auranofin efficiently
inhibited both full-length TrxR1 and TrxR2 and truncated TrxR2. In contrast, some newly synthesized gold(I)
compounds and cisplatin inhibited only full-length TrxR1 or TrxR2 and not truncated TrxR2. Surprisingly,
one gold(I) compound, [Au(d2pype)2]Cl, was a better inhibitor of TrxR1, whereas another, [(iPr2Im)2Au]Cl,
mainly inhibited TrxR2. These compounds also inhibited TrxR activity in the cytoplasm and mitochondria of
cells, but their cytotoxicity was not always dependent on the proapoptotic proteins Bax and Bak. In
conclusion, this study reveals significant differences between human TrxR1 and TrxR2 in substrate specificity
and metal compound inhibition in vitro and in cells, which may be exploited for development of specific
TrxR1- or TrxR2-targeting drugs.
© 2010 Elsevier Inc. All rights reserved.
The thioredoxin system consists of isoenzymes of thioredoxin
reductase (TrxR)1and thioredoxin (Trx) that are responsible for a
wide range of cellular functions, including redox regulation, antiox-
idant defense, and synthesis of deoxyribonucleotides . The major
cytosolic forms of TrxR and Trx are known as TrxR1 and Trx1 and
those present in mitochondria are known as TrxR2 and Trx2 ; the
four proteins are encoded by distinct genes that are all essential, as
their respective deletions are embryonically lethal in mice [2–4]. All
mammalian TrxR isoenzymes are selenoproteins with a redox-active
selenocysteine (Sec) residue in their active sites and belong to the
pyridine nucleotide-disulfide oxidoreductase family of proteins that
catalyze NADPH-dependent reduction of their native Trx substrates
[5,6]. They are homodimers containing a flavin adenine dinucleotide
(FAD) redox cofactor and a redox-active disulfide within a conserved
CVNVGC motif in one subunit, which interacts with a redox-active
selenenylsulfide/selenolthiol motif at the C-terminus of the other
subunit [6–9]. In addition to Trx, mammalian TrxR isoforms also
accept a range of low-molecular-weight disulfide-containing sub-
strates, including 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB), lipoic
acid, and lipoamide, as well as nondisulfide substrates such as
selenite, quinone compounds, and ascorbic acid [1,6,7,10,11]. This
Free Radical Biology & Medicine 50 (2011) 689–699
Abbreviations: DTNB, dithionitrobenzoic acid; DTT, dithiothreitol; TrxR, thioredoxin
reductase; TrxR1, cytosolic thioredoxin reductase; TrxR2, mitochondrial thioredoxin
reductase; Trx1, cytosolic thioredoxin; Trx2, mitochondrial thioredoxin; PAO, pheny-
larsine oxide; auranofin, 2,3,4,6-tetra-O-acetyl-1-thio-β-D-glucopyranosato-S-triethyl-
phosphine gold(I); [Au(d2pype)2]Cl, bis[1,2-bis(dipyridylphosphino)ethane] gold(I)
chloride; [Au(d2pypp)2]Cl, bis[1,3-bis(di-2-pyridylphosphino)propane] gold(I) chlo-
ride; [(iPr2Im)2Au]Cl, bis(1,3-diisopropylimidazol-2-ylidene) gold(I) chloride.
⁎ Corresponding author. Fax: +61 8 9224 0322.
E-mail address: firstname.lastname@example.org (A. Filipovska).
0891-5849/$ – see front matter © 2010 Elsevier Inc. All rights reserved.
Contents lists available at ScienceDirect
Free Radical Biology & Medicine
journal homepage: www.elsevier.com/locate/freeradbiomed
broad substrate specificity of mammalian TrxRs is usually attributed
to two unique features of the enzymes, the easily accessible C-
terminal selenenylsulfide/selenolthiol active center and the ability of
the N-terminal FAD/–CVNVGC redox center to directly reduce certain
small substrate compounds independent of the canonical Sec-
containing active site [12–15].
Most of the current knowledge about mammalian TrxR enzymes is
based upon extensive studies of the cytosolic TrxR1 purified from
bovine , rat , or human tissues [5,18,19], as well as mouse and
rat proteins expressed in recombinant form [7,15]. Although the gene
for mitochondrial TrxR2 has been cloned from a range of mammals
including rat, mouse, cow, and human and has been purified from rat
liver (reviewed in ), there is still limited knowledge about its
exact role in mitochondria. Much of what is known about the
biochemical properties of TrxR2 stems from the work of Bindoli and
Rigobello with co-workers studying the purified enzyme from rat 
and studies of semisynthetic variants produced by Hondal and co-
workers [15,21]. Moreover, the crystal structure of mouse TrxR2 has
been solved , as well as that of both rat [7,11] and human TrxR1
. Because of evident structural similarities between TrxR2 and
TrxR1, particularly regarding the redox-active centers and broad
substrate specificities of both enzymes, it is generally thought that
TrxR2 has redox-regulating and antioxidant functions in mitochon-
dria that are similar to those TrxR1 would have in the cytosol
[7,10,11]. A functional analogy to this would be found in Drosophila
melanogaster, in which the mitochondrial and cytosolic forms of TrxR
are encoded by the same gene . Although mammalian TrxR1 and
TrxR2 have some substrates in common, the different subcellular
localizations of the isoenzymes have probably given rise to some
differences, including specific protein substrates such as Trx2,
glutaredoxin 2, and cytochrome c for TrxR2 [24,25] and thiore-
doxin-related protein 14 for TrxR1 . The important question of
whether TrxR2 may have some distinct biochemical features differing
from those of TrxR1 has not yet been well addressed, but clues for
such differences can be found in the literature. For example, it was
recently reported that several TrxR-catalyzed substrate reactions
might be selenium-independent, which was found using TrxR2-
derived enzyme scaffolds , whereas such activity with similar
substrates was not found using a TrxR1-derived protein .
Furthermore, it was suggested recently that TrxR2 might prefer Trx1
as a substrate . However, the substrate specificities of human
TrxR1 and TrxR2, with their natural Trx1 and Trx2 substrates or with
other low-molecular-weight compounds, have not been investigated
previously in a direct side-by-side comparison using well-defined
enzyme preparations. Here we have compared the activities of pure
human recombinant TrxR1 and TrxR2 using their endogenous human
Trx1 and Trx2 substrates and small-molecule dithiol compounds and
found that human TrxR2 has different substrate affinities compared to
human TrxR1. We also compared the inhibition patterns of human
TrxR1 and TrxR2 with metal-based inhibitors and found pronounced
differences in inhibition between TrxR1 and TrxR2 using different
recently synthesized gold(I) compounds. These compounds inhibited
TrxR activity in cells and caused cell death via mitochondria, but by
different pathways, some of which required Bax and Bak, whereas
others did not.
Materials and methods
[(iPr2Im)2Au]Cl, [Au(d2pype)2]Cl, and [Au(d2pypp)2]Cl were
synthesized and used as reported elsewhere [29–31]. Platinol
(cisplatin) was from Mayne Pharm Pty Ltd. Native 4–16% polyacryl-
amide gels were from Invitrogen. 2′,5′-ADP Sepharose was from GE
Healthcare. All other reagents used in this study were from Sigma,
including the gold(I) compounds auranofin and aurothioglucose. Full-
length human TrxR1 was amplified from a testes cDNA library
(Clontech) by PCR with primers that introduced flanking NcoI and
BseAI sites. NcoI and BseAI digestionwas used to removethe rat TrxR1
insert along with the engineered SECIS from pET-TRSter  and
replace it with the hTrxR1 PCR fragment. The engineered SECIS from
pET-TRSter was reinserted subsequently into the human TrxR1-
encoding plasmid as a BseAI fragment. DNA sequencing confirmed
that the TrxR1 sequence was identical to the human TrxR1 isoform 2,
Entrez Gene reference sequence NM_182729.1, in which the second
amino acid, asparagine, was changed to aspartic acid owing to the
insertion of an NcoI restriction site. Human TrxR2 was amplified from
HeLa cell cDNA by PCR. The fragment corresponding to amino acids
37–623 of TrxR2 followed by an engineered bacterial SECIS  was
flanked by NcoI and PacI sites. The PCR product was digested with
NcoI and PacI and inserted into NcoI- and PacI-cut pETDuet-1
(Novagen) to make pET-hTrxR2. DNA sequencing showed that the
hTrxR2 sequence was identical to the Entrez Gene reference sequence
NM_006440.3. Trx1 was a kind gift from Professor A. Holmgren
(Karolinska Institutet, Stockholm, Sweden). Trx2 (amino acids 59–
166, NP_036605.2) was cloned into pGEX-4T-2 and expressed as a
GST fusion protein in Escherichia coli Rosetta2, purified using
glutathione–Sepharose (GE Healthcare), eluted by thrombin cleavage
according to the manufacturer's instructions, and further purified by
size-exclusion chromatography in PBS.
Expression and purification of human recombinant TrxR1 and TrxR2
Recombinant human TrxR1 or TrxR2 was expressed in either E. coli
ER2566 (New England Biolabs) or E. coli BL21(DE3) cells (New
England Biolabs), transformed with the corresponding pET-hTrxR1 or
pET-hTrxR2 plasmids, and in both cases the pSUABC plasmid 
according to the previously established protocol , in growth
medium containing 10 g NaCl, 10 g peptone, and 10 g yeast extract
per liter water, supplemented with 0.5 mM isopropyl-β-D-1-thioga-
lactopyranoside, 100 μg/ml L-cysteine, and5 μM sodium selenite. Cells
were harvested and lysed for 1 h using lysozyme, snap-frozen in
liquid nitrogen, and sonicated (six 15-s pulses at setting 20 using a
Sonifier 150; Branson). The initial purification on 2′,5′-ADP Sepharose
(GE Healthcare Life Sciences) and subsequent purification on PAO
Sepharose were performed as described previously . The full-
length enzymes were further purified by an ÄKTA-Explorer system
(GE Healthcare) using a Superdex 200 10/300 column (GE Health-
care) with a total bed volume of 24 ml. The two-amino-acid Sec-
deficient truncated human TrxR2 was deliberately expressed as such
in E. coli ER2566 cells by transformation with a plasmid lacking a
bacterial-type SECIS element and having a UAA replacing the original
UGA stop codon, thus expressing the enzyme without its last two
amino acids (resulting in a C-terminal Gly-Cys-COOH instead of Gly-
Cys-Sec-Gly-COOH). This enzyme was purified using 2′,5′-ADP
Sepharose, followed by gel filtration on an ÄKTA Purifier system
using a Superdex 200 10/300 column (GE Healthcare).
Mass spectrometry analyses were performed on a 4000 Q-TRAP
(AppliedBiosystems, Foster City, CA, USA) operating withan ion spray
voltage of 5500 V and an ion source gas 1 at 30 and scanning over a
mass range from 600 to 2000 m/z and detection in Q3 scanning mode.
Initial m/z peaks were deconvoluted with the Bayesian Protein
Reconstruct tool to provide the intact protein masses within the
Analyst 1.5.1 software (Applied Biosystems).
We used either 4–16% Bis–Tris native polyacrylamide gels or 10%
Tris–Glycine SDS denaturing polyacrylamide gels to resolve the
O. Rackham et al. / Free Radical Biology & Medicine 50 (2011) 689–699
enzymes for 1 h at 100 V. The gels were stained with Coomassie
Brilliant Blue R250 in 50% methanol and 7% acetic acid for 1 h and
destained for 3 h in 20% methanol and 7.5% acetic acid.
Total TrxR concentration was determined by absorbance mea-
surement at 463 nm (for oxidized FAD) and calculated using the
extinction coefficient for FAD of 11,300 M−1cm−1assuming one FAD
per subunit . The concentration of active enzyme was calculated
assuming two subunits per molecule for all TrxRs. Protein concentra-
tion determined by the absorbance measurement at 463 nm was used
to calculate the specific activities of the enzymes. Protein concentra-
tion was also determined by the bicinchoninic acid (BCA) assay using
bovine serum albumin (BSA) as a standard as described previously
. This was used to compare the protein concentration determined
by the absorbance at 463 nm and to determine protein concentration
of cell lysates.
TrxR activity measurements
The activity of TrxR was determined using the DTNB reduction
assay to calculate units and specific activity, as described previously
. The rate of DTNB reduction (0.0125–6.4 mM) by different TrxR
preparations (14.5–90 nM) in the presence of 150 μM NADPH in
50 mM Tris–HCl, 1 mM EDTA, pH 7.0 or pH 8.0, buffer was monitored
following an increase in absorbance at 412 nm in a final volume of
500 μl. The amount of product formed was determined from the
extinction coefficient 13,600 M−1cm−1of the thionitrobenzoate
(TNB−) anion, considering that 2 mol TNB−are formed per mol of
NADPH, with 1 unit corresponding to the oxidation of 1 μmol NADPH
The oxidation of NADPH by TrxR preparations using racemic (R,S)-
lipoic acid or lipoamide as substrate was measured as a decrease in
absorbance at 340 nm. A mixture of TrxR (14.5–90 nM) and 150 μM
NADPH in 50 mM Tris–HCl, 1 mM EDTA, pH 8.0, buffer was incubated
with either lipoic acid (0.4–6.4 mM) or lipoamide (0.5–10 mM) in a
total volume of 500 μl. Similarly, the oxidation of NADPH by TrxR
(14.5–90 nM) using the quinone substrate juglone (0.625–80 μM)
was measured at 340 nm. The decrease in absorbance was monitored
on a PerkinElmer Lambda 30 spectrophotometer and the rate of
NADPH consumption was determined using the extinction coefficient
The activity of TrxRs with Trxs as substrates was determined using
the coupled insulin assay according to published procedures . A
constant amount of TrxR (14.5–90 nM, final assay concentration) and
100 μM NADPH in 50 mM Tris–HCl, 1 mM EDTA, pH 7.0 or 8.0, buffer
was preincubated at room temperature for 5 min. To determine the
enzyme activity, the incubation mixture was then added to a cuvette
containing 0.16 mM insulin, 100 μM NADPH, and 0.45–14.4 μM either
Trx1 or Trx2 in a total volume of 500 μl. The decrease in absorbance
was recorded at 340 nm with a reference containingthe samemixture
but without the addition of Trx. TrxR activity was recorded as the
TrxR inhibition assay
Inhibition of recombinant TrxRs with increasing concentrations of
the metal compounds was measured by the DTNB assay adapted for
microtiter plates. TrxRs (100 nM) were reduced with 150 μM NADPH
in TE buffer (50 mM Tris–HCl, 1 mM EDTA, pH 8.0). Metal compounds
at varying concentrations were mixed with the reduced TrxRs in a
total volume of 50 μl and incubated for 30 min at room temperaturein
a microtiter plate. At the end of the incubation, 250 μl of TE buffer
containing150 μM NADPHand 2.5 mMDTNB wasadded. The NADPH-
dependent TrxR-catalyzed reduction of DTNB was monitored imme-
diately at 30 °C for 3 min and determined as the linear increase in
absorbance at 412 nm using a BioTek ELx808 absorbance microplate
reader. Data were expressed as a percentage of control TrxR activity
when not incubated with metal compounds.
Cell culture and cell death assay
Factor-dependent myeloid (FDM) mouse wild-type and Bax/
Bak−/−cell lines (a kind gift from Associate Professor Paul Ekert,
Murdoch Children's Research Institute and Royal Children's Hospital,
Melbourne, Australia)  were cultured at 37 °C under humidified
95% air/5% CO2 in Dulbecco's modified Eagle's medium without
phenol red, containing Earle's balanced salt solution and supple-
mented with 2 mM Glutamax, penicillin (100 U/ml), streptomycin
(100 μg/ml), 10% heat-inactivated fetal calf serum, and 0.5 ng/ml
interleukin-3. To test for toxicity, cells were grown to 90%
confluence and incubated for 24 h with their growth medium
containing increasing concentrations of the inhibitors. The cells
were collected (1000 g for 5 min), gently resuspended in 0.5 ml
binding buffer (10 mM Hepes/NaOH, pH 7.4, 150 mM NaCl, 2.5 mM
CaCl2, 1 mM MgCl2, 4% BSA) containing 10 μl propidium iodide
(30 μg/ml), and incubated for 15 min at room temperature in the
dark. Cell death was quantified using a Becton–Dickinson FACS Scan
flow cytometer. The data are expressed as a percentage of cells
grown in the absence of compounds.
Mitochondria were prepared from 3×106FDM wild-type and Bax/
Bak−/−cells treated with 50 μM each compound for 8 h in their
growth medium. Cells were sedimented (150 g for 5 min at 4 °C),
washed in solution A (100 mM sucrose, 1 mM EGTA, 20 mM Mops, pH
7.4), incubated on ice for 3 min in 100 μl solution B (100 mM sucrose,
1 mM EGTA, 20 mM Mops, 10 mM triethanolamine, 0.1 mg/ml
digitonin, pH 7.4), and disrupted using a 1-ml homogenizer. The
nuclei were sedimented (1000 g for 5 min at 4 °C), the pellet was
washed, and the combined supernatants were centrifuged (10,000 g
for 10 min at 4 °C). The mitochondrial pellet was washed once with
STE (250 mM sucrose, 5 mM Tris-HCl, 1 mM EGTA, pH 7.4) and
cytosolic and mitochondrial protein concentrations were determined
by the BCA assay using BSA as a standard. There was 90–95% and 1–3%
total citrate synthase and lactate dehydrogenase activity, respectively,
in the mitochondrial fractions.
TrxR activity in lysates
FDM wild-type and Bax/Bak−/−cells were incubated in their
growth medium with either 5 or 50 μM concentration of the metal
compounds auranofin, aurothioglucose, cisplatin, [Au(d2pype)2]Cl,
[(iPr2Im)2Au]Cl, and [Au(d2pypp)2]Cl. The cells were collected,
suspended in 100 μl cell extraction buffer (50 mM Tris, pH 7.6,
2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and 0.5% Igepal
CA-630), and lysed by three cycles of rapid freezing and thawing. The
cell lysates were clarified by centrifugation (16,000 g for 15 min at
4 °C) and the supernatant was used to measure total TrxR activity by
the end-point Trx-dependent insulin reduction method . Briefly,
5 μg of cell lysate, mitochondrial or cytosolic lysate, was incubated
with 20 μM recombinant human Trx1 or Trx2 (for mitochondrial
lysates) in the presence of 297 μM insulin, 1.3 mM NADPH, 85 mM
Hepes buffer, pH 7.6, and 13 mM EDTA for 40 min at 37 °C, in a total
volume of 50 μl. The reaction was stopped by the addition of 200 μl of
7.2 M guanidine–HCl in 0.2 M Tris–HCl, pH 8.0, containing 1 mM
O. Rackham et al. / Free Radical Biology & Medicine 50 (2011) 689–699
DTNB. The thioredoxin-dependent thiols of the reduced insulin were
determined by measuring the absorbance at 412 nm using a BioTek
ELx808 absorbance microplate reader, with the background absor-
bance reference for each sample containing the same components
except Trx1 or Trx2. Data were expressed as a percentage of TrxR
activity in cells treated with the metal compounds compared to
control, untreated cells. The cytosolic TrxR activity of wild-type and
Bax/Bak−/−cells was 146.8±9.1 and 117.7±2.3 nmol/min/mg,
respectively, and the mitochondrial TrxR activity was 73.4±3.4 and
61.3±2.3 nmol/min/mg, respectively.
Data were analyzed and Kmand kcatvalues were calculated using
Prism GraphPad 5.0. Statistical analyses were performed using a two-
tailed Student t test.
Identification of the molecular masses of human TrxR1, TrxR2, and the
truncated form of TrxR2
To enable this side-by-side comparison of TrxR1 and TrxR2, first
we expressed the human enzymes, as well as the truncated version of
TrxR2 (TrxR2Δ) missing the last two amino acids of the selenenyl-
sulfide/selenolthiol redox-active center, in E. coli as recombinant
proteins. This was enabled by the engineering of a bacterial-like SECIS
element in the open reading frame and by overexpressing the
bacterial SelA, SelB, and SelC genes, as described previously for
expression of rat TrxR1 . We purified all of the proteins to near
homogeneity following established protocols, which resulted in
several preparations and the overall yields and specific activities of
representative preparations are shown in Table 1. The increase in the
specific activities of the enzymes after PAO Sepharose purification
suggested that this purification step was necessary to enrich for the
full-length TrxRs, as was also found earlier . Analyzing the purity
and molecular weights of the different TrxR preparations by mass
spectrometry showed that the TrxR1 molecular mass was 54,769 Da
(Fig. 1A), TrxR2 was 53,117 Da (Fig. 1B), and the truncated hTrxR2
was 52,907 Da, confirming that it was missing the Sec and Gly amino
acids at its C-terminus (Fig. 1C). All recombinant enzymes had the
expected molecular weights and this analysis also confirmed the
purity of the preparations. The apparent complete absence of UGA-
truncated species in the full-length TrxR preparations indicated that
the PAO affinity purification step effectively separated these proteins
from truncated variants arising from premature (Sec-encoding) UGA
termination, which was surprising in view of previous findings that
some UGA-truncated subunits of the rat TrxR1 always copurified with
the full-length enzyme . One explanation could be that truncated
forms of thehuman isoenzymesare highly unstable,which alsowould
be in agreement with our observations that these proteins easily
precipitated upon storage (not shown). All preparations analyzed in
this study weretherefore used within10 days of preparation, whereas
for long-term use the enzymes were stored in 20% glycerol, which
increased their stability.
Human thioredoxin reductases form dimers and tetramers
We used native polyacrylamide gel electrophoresis to determine if
the human TrxR proteins differed in their tendency to form dimeric,
Thioredoxin reductase activities and protein amounts after 2′,5′-ADP Sepharose and
PAO Sepharose purification
2′,5′-ADP Sepharose PAO Sepharose
aThe activity from two independent preparations of TrxR1 and TrxR2 is presented as
(1) and (2).
bTrxR2Δ was purified only on 2′,5′-ADP Sepharose and, as it is missing the Sec
residue, its purification on PAO Sepharose was not applicable (N/A).
Fig. 1. Mass spectrometry analysis of human thioredoxin reductases. Samples from
PAO-purified (A) hTrxR1 and (B) hTrxR2 and (C) 2′,5′-ADP Sepharose-purified
hTrxR2Δ were prepared and analyzed by mass spectrometry as described under
Materials and methods. Deconvoluted zero-charged ions corresponding to the average
mass are shown. A peak at 54,769 Da for the full-length hTrxR1, a peak at 53,117 Da for
the full-length hTrxR2, and a peak at 52,907 Da for the truncated hTrxR2 correlate with
their respective theoretical masses.
O. Rackham et al. / Free Radical Biology & Medicine 50 (2011) 689–699
tetrameric, or oligomeric complexes. We found that human TrxR1
existed as both a dimer and a tetramer (Fig. 2, top), and in addition it
formed higher molecular weight complexes, as previously found for
the rat TrxR1 . The full-length TrxR2 and TrxR2Δ were
predominantly dimeric, and higher oligomers were virtually absent
(Fig. 2, top). To analyze whether disulfides, selenenylsulfides, or
diselenides were involved in the formation of these dimers or
tetramers, we treated the proteins with high concentrations of the
reducing agent dithiothreitol (DTT) before analyzing them on either
native or denaturing gels. We found that DTT resolved the majority of
the tetramers of all the TrxRs (Fig. 2, top). The higher molecular
weight complexes formed by the rat TrxR1 were, however, reduced
predominantly to both tetrameric and dimeric forms, indicating the
remaining tetramers were not disulfide- or selenenylsulfide-linked
dimersof theprotein, astheywerehighly resistantto DTT(Fig. 2, top).
In the denaturing gels (Fig. 2, bottom), monomers were mainly seen
in the absence of DTT for both TrxR1 and TrxR2, suggesting that the
earlier identified intersubunit disulfide in the dimer of mouse TrxR2
 is either not formed in human TrxR2 or rather labile. Importantly,
however, we also observed that some dimeric species of the different
TrxR preparations persisted in the denaturing gels even in the
presence of 100 mM DTT, suggesting that these were not only
disulfide- or selenenylsulfide-linked dimers (Fig. 2, bottom). The
exact natures of the various oligomeric forms of mammalian TrxR
are yet unknown, but these are regularly seen in mammalian cell
extracts and the propensity to form these species is apparently an
inherent feature of these proteins. Future investigations using tryptic
digests of the enzymes followed by mass spectrometry could
potentially identify residues involved in the formation of these
dimers and tetramers.
Activity of human TrxR isoenzymes using human Trx1 and
Trx2 as substrates
Previous studies have shown that different mammalian TrxR
isoenzymes can effectively use E. coli Trx, human Trx1, and rat Trx2
[16,17,28]. Here, we investigated the efficiency of human TrxR1 and
TrxR2 with their cognate substrates, human Trx1 and Trx2, at both pH
7 and pH 8 to mimic the pH of their respective subcellular
compartment microenvironments. We found that the Kmof TrxR1
was similar for both Trx1 and Trx2 at pH 7, whereas at pH 8 the Kmof
TrxR1 was lower for Trx1 compared to Trx2. As illustrated by the
catalytic efficiencies, however, TrxR1 may clearly use both Trx1 and
Trx2 as substrates (Table 2). In contrast, the Kmvalues of TrxR2 for
Trx2 were significantly lower than for Trx1 at both pH 7 and pH 8, and
the catalytic efficiency using Trx2 as substrate was about 10-fold
higher than using Trx1, thus showing that TrxR2 is clearly more
effective at using its endogenous substrate Trx2 compared to Trx1
(Table 2). Not surprisingly, we did not observe any reduction of either
Trx substrate using TrxR2Δ, indicating that the Sec residue of TrxR2 is
required for Trx reduction.
Activity of human TrxR isoenzymes using low-molecular-weight
Determining the activity with the model substrate DTNB, we found
that TrxR1, at both pH 7 and pH 8, had about 10-fold higher catalytic
efficiency and about twice the kcatfor this substrate compared to
TrxR2 (Table 3). However, we also found that the truncated variant,
TrxR2Δ, missing the Sec residue, still reduced DTNB almost as
efficiently as full-length TrxR2 (Table 3), in contrast to the two-
amino-acid-truncated rat TrxR1, which displayed only about 5–10%
activity compared to the full-length enzyme [32,37].
We also investigated whether the human TrxR isoenzymes could
reduce the low-molecular-weight disulfide compounds lipoamide
and lipoic acid, which are known substrates for rat TrxR1 and mouse
TrxR2 [15,38]. We found that the Kmof human TrxR1 using lipoamide
was lower than the corresponding Km of TrxR2 and the catalytic
efficiency was fivefold higher, showing that human TrxR1 reduces
lipoamide more effectively than TrxR2 (Table 3). TrxR2Δ reduced
lipoamide with comparable efficiency compared to the full-length
TrxR2. Surprisingly, we found that although TrxR2 reduced lipoamide
it could not reduce lipoic acid (Table 3), in contrast to the human
TrxR1 (Table 3), which can reduce both lipoic acid and lipoamide,
similar to the rat TrxR1 .
Activity of human TrxR isoenzymes with the quinone substrate juglone
We investigated whether the human enzymes could reduce the
quinone substrate juglone, which was previously found to be
efficiently reduced both by the selenolthiol motif and, in a Sec-
independent redox cycling manner, by rat TrxR1 . We found that
human TrxR1 could indeed reduce juglone and was significantly more
efficient than TrxR2, mainly because of a lower Kmthan that seen with
TrxR2 (Table 4). Highly efficient redox cycling with juglone has been
shown before for the truncated TrxR1 [39,40] and we found that
TrxR2Δ could also reduce juglone to an extent similar to that of full-
length TrxR2 (Table 4).
Fig. 2. Analysis of human thioredoxin reductases by native and denaturing
polyacrylamide gel electrophoresis. Recombinant full-length rTrxR1, hTrxR1, hTrxR2,
and truncated hTrxR2 were resolved on (top) 4–16% native gels or on (bottom)
denaturing 10% SDS Tris–glycine gels in the presence or absence of 100 mM DTT.
Kinetic analysis of human thioredoxin reductases using human thioredoxin 1 and
thioredoxin 2 as substrates
SubstratepH 7.0pH 8.0
aThe activity of TrxR2Δ using Trx1 and Trx2 as substrates was not detected (ND).
O. Rackham et al. / Free Radical Biology & Medicine 50 (2011) 689–699
Metal-based compounds have different inhibitory effects on human
TrxR1 and TrxR2
Gold(I) compounds such as auranofin and aurothioglucose have
been identified as highly effective inhibitors of selenoprotein
thioredoxin reductases in vitro, as well as in cells, in the case of
auranofin [41–43]. Therefore we analyzed whether these compounds
could inhibit human TrxR1 and TrxR2 to the same extent. Indeed, we
found that concentrations above 5 μM of either auranofin or
aurothioglucose effectively inhibited both of the full-length enzymes
(Figs. 3A and B). Furthermore, the DTNB reductase activity of the
TrxR2Δ enzyme was also inhibited by auranofin and aurothioglucose
at these concentrations, indicating that the N-terminal redox center of
this enzyme was also inhibited (Figs. 3A and B). Thus, auranofin and
aurothioglucose do not selectively inhibit the Sec-containing active-
site motif of human TrxRs. Notably, at lower concentrations (ranging
from 0.1 to 1 μM) auranofin and aurothioglucose were here found to
be more specific inhibitors for TrxR1 compared to TrxR2.
Cisplatin has long been used as a chemotherapeutic agent, based
on its ability to bind to DNA and lead to cell death . In addition,
cisplatin has also been shown to effectively inhibit rat and bovine
TrxR1 [7,45], and when TrxR1 was knocked down in human A549
cancer cells by siRNA these became more resistant to cisplatin .
Therefore, we analyzed the effects of cisplatin on the human
recombinant TrxR enzymes and found that the activities of both
were inhibited at similar levels by cisplatin using concentrations
above 10 μM (Fig. 3C). Inhibition of the TrxR2Δ DTNB reductase
activity using cisplatin was very limited, suggesting that the Sec-
containing motif was the prime target of cisplatin (Fig. 3C).
Bis-chelated gold(I) phosphine compounds have been shown to
pyridyl phosphine derivatives were selectively toxic to tumorigenic
cells but not nontumorigenic cells and could also inhibit the in vitro
activity of rat TrxR1 and total TrxR activity in cells [31,46]. Therefore
here we investigated whether the observed reduction in TrxR activity
in human cells was likely to have been mainly due to inhibition of
TrxR1, TrxR2, or both enzymes. We found that [Au(d2pype)2]Cl was a
potent inhibitor of both TrxR1 and TrxR2 (Fig. 3D). Although the [Au
(d2pype)2]Cl complex was a more effective inhibitor of TrxR1, the
inhibition of TrxR2 was probably selective for the Sec-containing
active site because the DTNB reductase activity of hTrxR2Δ was
unaffected at concentrations up to 50 μM (Fig. 3D). We also
investigated the inhibitory effects of the related gold complex [Au
(d2pypp)2]Cl on the enzymes and found that it was a potent inhibitor
of TrxR1 but, surprisingly, above 5 μM concentration this compound
was a significantly less potent inhibitor of TrxR2 (Fig. 3E). The
inhibition by this compound also seemed to be limited to the Sec
active site of TrxR2 because the activity of TrxR2Δ was not affected
Recently, we reported that a new series of gold(I) N-heterocyclic
carbene compounds exhibited anticancer activity [30,47], and the
gold complex [(iPr2Im)2Au]Cl was selectively reactive toward Sec
compared to Cys . Consequently, we also observed that this
compound specifically inhibited TrxR activity in cells but not the
activity of glutathione reductase . Here, surprisingly, we found
that at high concentrations [(iPr2Im)2Au]Cl inhibited TrxR2 but not
TrxR1 or the truncated TrxR2Δ (Fig. 3F). These findings collectively
showed that TrxR1 and TrxR2, although both probably inhibited by
targeting of their Sec residues, displayed markedly different rates of
inhibition between the pyridyl phosphine compounds, [Au(d2pype)2]
Cl and [Au(d2pypp)2]Cl, and the carbene compound [(iPr2Im)2Au]Cl,
suggesting that specific TrxR1- or TrxR2-targeting metal inhibitors
can be developed for potential specific clinical use.
Effects of metal-based TrxR inhibitors on cell viability in wild-type and
There is evidence in the literature to suggest that the molecular
mechanisms of antitumor activity of different classes of linear and
lipophilic gold(I) compounds may be different, although these classes
ofcompoundsall seemto causecell deathvia mitochondrialpathways
. Also, cisplatin has been suggested to involve mitochondrial
dysfunction at the onset or progression of cell death , although
additional cytosolic- or endoplasmic reticulum-related events of
oxidative stress may also be involved . Because all of these
metal-based compounds inhibit mammalian forms of TrxR, to
different extents (see above), we investigated whether their mech-
anisms of action at the cellular level will always depend upon Bax and
Bak and thus probably involve the opening of the Bax/Bak pore in the
mitochondrial outer membrane. To test this we used wild-type and
Bax/Bak−/−cell lines, of which the latter are resistant to apoptotic
stimuli involving the formation of the Bax/Bak pore . We treated
these cell lines with increasing concentrations of selected gold(I)
compounds or cisplatin and measured cell death by flow cytometry.
This showedthatat 2.5 μM auranofininduced cell deathdependenton
the Bax/Bak pore; however, at higher concentrations auranofin did
not require the pore to induce cell death (Fig. 4A). The [Au(d2pype)2]
Cl and [Au(d2pypp)2]Cl compounds induced cell death independent
of Bax and Bak (Figs. 4D and E). Aurothioglucose was not toxic to
eitherof the cell lines(Fig. 4B), evenafter a 72-h incubationwithup to
100 μM concentrations of aurothioglucose (data not shown). Very
high concentrations of up to 50 μM cisplatin after a 24-h incubation
were not toxic to the Bax/Bak−/−cells, whereas the wild-type cells
showed ~50% cell death at about 15 μM, suggesting that cisplatin
leads to cell death that requires Bax/Bak pore formation (Fig. 4C). The
Bax/Bak−/−cells also survived at concentrations up to 20 μM
[(iPr2Im)2Au]Cl, whereas this compound was highly toxic to the
wild-type cells, with approximately 90% cell death at 1 μM, suggesting
that the mechanism of cytotoxicity of [(iPr2Im)2Au]Cl also involved
the Bax/Bak pore, although very high concentrations could lead to cell
death independent of Bax/Bak (Fig. 4F).
Kinetic analysis of human thioredoxin reductases using low-molecular-weight disulfide
compounds DTNB, lipoamide, and lipoic acid as substrates
SubstratepH 7.0pH 8.0
aThe activity of TrxR2 and TrxR2Δ using lipoic acid as substrate was not detected
(ND). The values are averages from determinations using several different preparations
for each enzyme.
Kinetic analysis of human thioredoxin reductases using the quinone juglone as
O. Rackham et al. / Free Radical Biology & Medicine 50 (2011) 689–699
Effects of metal-based inhibitors on TrxR activities in wild-type and
We also investigated whether the various metal compounds
inhibited TrxR activity in cells. Although aurothioglucose effectively
inhibited the activity of the recombinant TrxR enzymes (Fig. 3), we
observed limited inhibition of TrxR activity in both cell lines (Fig. 5).
The lack of cytotoxicity (Fig. 4B) and low cellular TrxR inhibition
(Figs. 5A–D), even at high-micromolar concentrations of aurothioglu-
cose, suggest that poor cellular uptake may be an explanation for its
lack of effect in these cells. We found that both 5 and 50 μM
concentrations of the other five metal compounds effectively
inhibited total TrxR activity in both cell lines after 5 h (Fig. 5). The
inhibition of TrxR activity was greater after an 8-h incubation with the
five compounds, particularly with 50 μM concentration of the
compounds (Figs. 5C and D), indicating that the inhibition of TrxR
activity in cells is time and concentration dependent.
Next we incubated FDM cells with 50 μM metal compounds and
separated them into cytosolic and mitochondrial fractions to
investigate the effects of these compounds on the endogenous
cytosolic and mitochondrial TrxR activities. We found similar levels
of enzyme inhibition in both cell lines (Fig. 6). Auranofin was a
significantly more effective inhibitor of both enzymes compared to
cisplatin, whereas aurothioglucose had limited inhibitory effects,
most probably because of its poor cellular uptake. The lipophilic gold
compounds were more effective inhibitors of mitochondrial TrxR,
probably as a result of their accumulation inside mitochondria
[30,31,47]. These findings collectively suggest that the inhibition of
TrxRs may contribute to cytotoxicity and, furthermore, that the
observed cell death triggeredby metal-based compounds may involve
different mitochondrial pathways not always dependent on the
presence of Bax and Bak.
In this study we made a side-by-side comparison of substrate and
inhibition specificities of pure recombinant human TrxR2 and TrxR1
and found significant differences between the two human enzymes
that were previously not observed and are beyond those determined
solely by their different subcellular compartments. The divergent
sensitivities of TrxR1 and TrxR2 to different inhibitors should
underpin inhibitor-specific differences in cytotoxicity profiles of
metal-based anticancer drugs that target cellular TrxR as part of
their molecular mechanism of toxicity.
Although we used established protocols for recombinant expres-
sion of TrxRs , this is the first study comparing the two human
isoenzymes and the first to express full-length recombinant human
TrxR2. Earlier attempts to express recombinant human TrxR1 gave
much lower yields than expressing rat TrxR1, which was explained by
a higher frequency of rare codon usage in the human gene and a
higher instability of the human enzyme [49,50], which we also noted.
We were surprised by the near-complete enrichment of the full-
length enzymes after PAO Sepharose purification, considering that
earlier studies found rat TrxR1 to be purified as heterodimers of
truncated and full-length subunits still displaying a specific activity of
40 U/mg or more . This suggests that homodimeric full-length rat
TrxR1 would have significantly higher specific activity than 40 U/mg
, whereas we found that homogeneously full-length human TrxR1
Fig. 3. Inhibition of hTrxR1, hTrxR2, and hTrxR2Δ activity using six different metal compounds. In vitro inhibition of hTrxR activity with increasing concentrations of the metal
compounds was measured by the DTNB assay. Recombinant TrxRs (100 nM) were reduced with 150 μM NADPH in TE buffer (50 mM Tris–HCl, 1 mM EDTA, pH 8.0). Varying
concentrations of the metal compounds (A) auranofin, (B) aurothioglucose, (C) cisplatin, (D) [Au(d2pype)2]Cl, (E) [Au(d2pypp)2]Cl, and (F) [(iPr2Im)2Au]Cl were mixed with the
reduced TrxRs and incubated for 30 min at room temperature. The NADPH-dependent TrxR-catalyzed reduction of DTNB was monitored immediately after the addition of TE buffer
containing 150 μM NADPH and 2.5 mM DTNB at 30 °C for 3 min and determined as the linear increase in absorbance at 412 nm. Data are expressed as a percentage of control TrxR
activity when not incubated with metal compounds. Means±SEM from three independent experiments; *pb0.05 for TrxR2 compared to TrxR1 and †pb0.05 for TrxR2 compared to
TrxR2Δ, by a two-tailed Student t test.
O. Rackham et al. / Free Radical Biology & Medicine 50 (2011) 689–699
Fig. 4. Effects of metal compounds on factor-dependent myeloid (FDM) wild-type and Bax/Bak−/−cells. FDM wild-type and Bax/Bak−/−cell lines were grown to 90% confluence
and incubated for 24 h in their growth medium containing increasing concentrations of (A) auranofin, (B) aurothioglucose, (C) cisplatin, (D) [Au(d2pype)2]Cl, (E) [Au(d2pypp)2]Cl,
and (F) [(iPr2Im)2Au]Cl. To test for toxicity, cells were collected and gently resuspended in 0.5 ml binding buffer containing 10 μl propidium iodide (30 μg/ml) and cell death
was quantitated by flow cytometry. Data are expressed as means±SEM from three independent experiments. *pb0.05 for FDM Bax/Bak−/−cells compared to wild-type cells by a
two-tailed Student t test.
Fig. 5. Inhibition of total thioredoxin reductase activity in cells. Factor-dependent myeloid (FDM) wild-type and Bax/Bak−/−cell lines were incubated for 5 or 8 h in their growth
medium with (A and B) 5 μM or (C and D) 50 μM concentration of the metal compounds auranofin, aurothioglucose, cisplatin, [Au(d2pype)2]Cl, [(iPr2Im)2Au]Cl, and [Au(d2pypp)2]
Cl. The cells were collected, suspended in 100 μl cell extraction buffer (50 mM Tris, pH 7.6, 2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and 0.5% Igepal CA-630), and lysed by
three cycles of rapid freezing and thawing. Total cellular TrxR activity was measured by the end-point Trx-dependent insulin reduction method using 5 μg of the cell lysate that was
clarified by centrifugation (16,000 g for 5 min at 4 °C). Data are expressed as means±SEM from three independent experiments.
O. Rackham et al. / Free Radical Biology & Medicine 50 (2011) 689–699
and TrxR2 had specific activities of about 40 U/mg in the DTNB model
assay, which is the commonly denoted inherent activity of mamma-
lian TrxR .
Human TrxR2 had a clear preference for its native Trx2 substrate
and its catalytic efficiency was higher at pH 8, indicating that the
mitochondrial environment of TrxR2, where the pH is generally 1 unit
higher compared to the cytosol, has tailored its differences in affinity
and activity for its natural substrate Trx2. Importantly, the differences
in catalytic efficiency between the mitochondrial and the cytosolic
TrxRs as found here should have implications for their intracellular
functions. It is clear that TrxR1 can reduce the mitochondrial Trx2
equally as well as its native Trx1, as observed before for Trxs from
other species , but the question remains whether TrxR1 would
ever have the possibility of reducing Trx2 in vivo, considering the
different subcellular compartments of the two proteins.
Human TrxR1 clearly reduced DTNB and lipoamide more effi-
ciently than TrxR2, while the selenenylsulfide active center motif of
TrxR2 ; but in contrast to that study, we found that human TrxR2
could not reduce lipoic acid. Although lipoic acid is a substrate for
human TrxR1 we still found that its efficiency was two-fold lower
compared to that for lipoamide. Endogenous lipoic acid is a covalently
bound dithiol cofactor for the α-keto acid dehydrogenase enzyme
complexes, such as pyruvate dehydrogenase and α-ketoglutarate
dehydrogenase , and is negatively charged compared to the
neutral charge of lipoamide. This charge difference should play an
important role in substrate discrimination for human TrxR2 and in its
reduced efficiency of lipoic acid reduction, compared to the lipoamide
reduction by humanTrxR1. It may be thusthat in humanslipoic acid is
not reduced by TrxR2 and that other enzymes such as lipoamide
dehydrogenase and glutathione reductase may catalyze this reaction.
We foundthathuman TrxR1hada three-foldhigheraffinityfor the
quinone substrate juglone compared to TrxR2 although both TrxR2
and TrxR2Δ reduced juglone, suggesting that the N-terminal motif
alone can reduce this substrate. Juglone can be rapidly reduced by the
selenenylsulfide/selenolthiol active center of rat TrxR1, which can
result in the formation of a nucleophilic arylating juglone derivative
that can target selenocysteine and irreversibly inactivate this motif.
Nevertheless, juglone may continue to efficiently redox cycle directly
with the N-terminal CVNVGC/FAD motif of rat TrxR1 and thus show
high activity [7,39]. Our findings with the human enzymes are in line
with those earlier results.
Our kinetic data indicate that the affinity of TrxR2 for disulfide
substrates other than its endogenous Trx2 substrate is significantly
lower than the affinity of TrxR1 for these types of substrates. The
mitochondrial localization of TrxR2 probably contributes to the more
efficient use of endogenous mitochondrial-localized substrates such
as Trx2, Grx2, and cytochrome c [24,25] and there may also be other
mitochondria-specific substrates for which TrxR2 has high affinity yet
to be discovered. Our results suggest that the functions of the
mitochondrial TrxR2 and cytosolic TrxR1 are less overlapping than
generally believed, also when considering their native biochemical
and kinetic properties. The previously reported crystal structure of
mitochondrial TrxR2 indeed indicates that there are distinct differ-
ences in the positioning of residues around the redox-active centers of
TrxR2 compared to TrxR1, which may contribute to or be responsible
for the divergent reduction of different substrates as well as their
affinities . From our data we conclude that (i) TrxR1 has a broader
substrate specificity compared to TrxR2, (ii) TrxR2 has greater affinity
for its endogenous substrates that are different from those for TrxR1,
which may reflect other functional requirements within the mito-
chondrial compartment, and (iii) hTrxR2 is catalytically more efficient
at the higher endogenous pH of the mitochondrial matrix.
Several emerging anticancer therapies identify TrxR as a target for
drug development , as altered activities of the Trx system proteins
have been observed in several human diseases [12,52]. Many
therapeutically used compounds have been identified as TrxR
inhibitors [12,52–54], including gold(I) compounds that have also
shown early promise as anticancer drugs [30,31,43,53,54]. Auranofin
is probably the most effective inhibitor of mammalian TrxR found to
date [12,52,53] and here we show that at micromolar concentrations
it is an effective inhibitor of both the N-terminal dithiol and the C-
terminal selenolthiol redox centers of human TrxRs. This indicates
that auranofin binds nonselectively to both selenocysteine and
cysteine residues in these proteins, consistent with recent reports
. The finding that the related compound, aurothioglucose, also
inhibited all three enzymeswas in contrast to a recent findingthat it is
a better inhibitor of the N-terminal redox center compared to the C-
terminal selenenylsulfide/selenolthiol active center of the mouse
TrxR2 . These findings may reflect differences between the human
and the mouse TrxR2 enzymes. Interestingly, at nanomolar concen-
trations auranofin and aurothioglucose were more specific inhibitors
of TrxR1 compared to TrxR2 and less effective at inhibiting TrxR2Δ,
suggesting that the selenolthiol motif of TrxR is inhibited in
preferenceto thedithiolmotif. Also,the active-site microenvironment
around the C-terminal selenolthiol motif of the reduced enzyme must
fine-tune its susceptibility to different inhibitors, as illustrated by the
differences in inhibition between TrxR1 and TrxR2. Thus, the
tetrapeptide C-terminal active site of TrxR should not be regarded
only as an easily accessible Sec-presenting motif targeted indiscrim-
inately by electrophiles, which is a rather simplified view often found
in the literature. Cisplatin inhibited the activity of TrxR1 and TrxR2,
but not TrxR2Δ, suggesting that its mechanism of inhibition probably
involves coordination of platinum to the Sec-containing redox center.
This was suggested to generate selenium-compromised thioredoxin
reductase-derived apoptotic proteins (SecTRAPs) that may induce cell
Fig. 6. Inhibition of cytosolicand mitochondrial TrxR activity in cells. Factor-dependent myeloid (FDM) (A) wild-type and (B) Bax/Bak−/−cells were incubated for 8 h in their growth
medium with 50 μM concentration of the metal compounds auranofin, aurothioglucose, cisplatin, [Au(d2pype)2]Cl, [(iPr2Im)2Au]Cl, and [Au(d2pypp)2]Cl. The cells were separated
into mitochondrial and cytosolic fractions and TrxR activities were measured by the end-point Trx-dependent insulin reduction method using 5 μg of lysate that was clarified by
centrifugation (16,000 g for 5 min at 4 °C). Data are expressed as means±SEM from two independent experiments. *pb0.05 for mitochondrial TrxR compared to cytosolic TrxR by a
two-tailed Student t test.
O. Rackham et al. / Free Radical Biology & Medicine 50 (2011) 689–699
death by a gain of function . The bis-chelated Au(I) pyridyl
phosphine compounds were inspired by auranofin, but designed to
have lower thiol reactivities while improving the selectivity for
Sec . Consequently we found that the bis-chelated gold(I)
compounds [Au(d2pype)2]Cl and [Au(d2pypp)2]Cl were effective
inhibitors of the selenenylsulfide/selenolthiol active center of the
TrxRs, thus potentially also yielding cytotoxic SecTRAPs. However,
such proteins have not been shown to form from TrxR2 yet.
The cationic gold(I) compounds could inhibit both TrxR1 and
TrxR2 in vitro; however, because they are lipophilic and consequently
accumulate inside mitochondria as a result of the high mitochondrial
membrane potential (Δψm), we observed their specific inhibition of
TrxR2 in cells and conclude that previously observed cellular TrxR
inhibition  also probably involved inhibition of TrxR2. Surpris-
ingly, the [Au(d2pype)2]Cl and [Au(d2pypp)2]Cl compounds were
significantly better inhibitors for TrxR1 in vitro, whereas[(iPr2Im)2Au]
Cl inhibited only TrxR2. This indicates not only that there is a
difference in the inhibition mechanisms of these compounds, but also
that the active centers of the cytosolic and mitochondrial TrxR
isoenzymes may have different affinities for these compounds. We are
currently attempting to investigate the mechanism of binding of these
gold(I) compounds by cocrystallizing them with the purified TrxRs.
At low concentration auranofin led to Bax/Bak-dependent cell
death probably by preferentially reacting with selenoproteins such as
TrxR2, whose inhibition by auranofin has been shown to lead to
peroxiredoxin-3 oxidation and Bax/Bax-dependent apoptosis .
High concentrations of auranofin caused cell death independent of
Bax/Bak, most probably by binding both thiols and selenols, causing
changes in the mitochondrial thiol redox pool that have been shown
to cause apoptosis by inducing mitochondrial permeability transition
(MPT) . Therefore the mechanism of action of auranofin via the
Bax/Bak pore is concentration dependent and may vary between cell
types. Nevertheless, the high thiol reactivity of auranofin is likely to
limit its anticancer activity in vivo, as its toxicity to cultured cancer
cells was reduced 10-fold in the presence of serum proteins, where
the loss of activity was attributed to binding to extracellular protein
thiols . Although aurothioglucose was an effective inhibitor of
recombinant TrxRs, its inhibition of cellular TrxRs was limited and it
was not toxic to the two cell lines, most probably because polymeric
gold(I) thiolates do not readily enter cells and consequently have very
low cytotoxicity and lack anticancer activity . In contrast, cell
death induced by cisplatin was entirely dependent on the presence of
Bax and Bak, suggesting that TrxR inhibition and/or DNA binding
could lead to Bax/Bak-dependent cell death. Although the exact
mechanism of cisplatin-induced cell death needs to be investigated
further, the combined ability of cisplatin to form DNA lesions and
inhibit TrxR activity makes it an effective chemotherapeutic agent and
may contribute to its success as a clinically used anticancer drug.
The lipophilic cations [Au(d2pypp)2]Cl and, in particular, [Au
(d2pype)2]Cl were more specific inhibitors of TrxR1 and caused cell
death independent of the Bax and Bak proteins. We have previously
shown that these compounds cause cell death via mitochondria 
and this may be a result of their high lipophilicity, which can cause
generalized permeabilization of the mitochondrial membrane or
induce the MPTthat dissipates the Δψm. In addition, a small amount of
these compounds may remain in the cytoplasm, as we observed some
inhibition of cytosolic TrxR activity that may further contribute to the
onset of cell death independent of Bax and Bak. In contrast to these
compounds, the[(iPr2Im)2Au]Cl compound led to Bax/Bak-dependent
cell death at lower concentrations. We have shown previously that
[(iPr2Im)2Au]Cl is a moderately lipophilic cation that selectively
accumulates in the mitochondrial matrix, driven by the Δψm, and
here we showed that it is a TrxR2-specific inhibitor. We suggest that
the mechanism of action of this compound involves mitochondrial
uptake followed by specific inhibition of the TrxR2 that leads to
apoptosis via mitochondria, which requires the formation of the Bax/
Bak pore. However, at higher concentrations this compound can cause
cell death independent of the Bax/Bak pore, much like [Au(d2pype)2]
Cl and [Au(d2pypp)2]Cl, most likely as a result of its lipophilicity. An
alternative explanation may be that the compounds irreversibly
modify the Sec active site and thereby form SecTRAPs that induce cell
death. Cell death induced by different SecTRAPs may be caused by
different mechanisms that may not always require the Bax and Bak
proteins. The molecular mechanism by which the inhibition of TrxRs
leads to cell death and whether it requires Bax/Bak pore formation
must evidently be specifically studied and resolved for every
individual TrxR-targeting drug and for each specific cell type.
In conclusion, we have shown that the human cytosolic TrxR1 and
mitochondrial TrxR2 have different affinities for low-molecular-
weight disulfide substrates and metal-based inhibitors. In combina-
tion with cell death assays in Bax/Bak double-knockout cells our data
enable models to be proposed for the mechanisms of action of these
compounds thattakeintoaccountthe physiologicallyrelevantcellular
locations of TrxR1 and TrxR2, their compound-specific targeting
profiles, and the patterns of inhibition determined by different kinetic
properties of TrxR1 and TrxR2.
The work was supported by The Australian Research Council
(Future Fellowships FT0991008 to A.F. and FT0991113 to O.R. and
Discovery Grants DP0986318 and DP0878438), the Karolinska
Institutet,The SwedishResearchCouncil(Medicine), andTheSwedish
Cancer Society. Mass spectrometry analyses were performed in
facilities provided by the Lotterywest State Biomedical Facility–
Proteomics Node, WAIMR. We thank Associate Professor Paul Ekert
for kindly giving us the FDM cells. The gold pyridyl phosphine
compounds were synthesized by Coby Sutcliffe and Anthony Hum-
phreys in Professor Sue Berners-Price's research group and the
carbene compound was synthesized by James Hickey in Associate
Professor Murray Baker's and Professor Sue Berners-Price's groups.
The synthesis of these compounds was published previously and we
thank them for providing them for this study.
 Arner, E. S.; Holmgren, A. Physiological functions of thioredoxin and thioredoxin
reductase. Eur. J. Biochem. 267:6102–6109; 2000.
 Conrad, M.; Jakupoglu, C.; Moreno, S. G.; Lippl, S.; Banjac, A.; Schneider, M.; Beck,
H.; Hatzopoulos, A. K.; Just, U.; Sinowatz, F.; Schmahl, W.; Chien, K. R.; Wurst, W.;
Bornkamm, G. W.; Brielmeier, M. Essential role for mitochondrial thioredoxin
reductase in hematopoiesis, heart development, and heart function. Mol. Cell. Biol.
 Jakupoglu, C.; Przemeck, G. K.; Schneider, M.; Moreno, S. G.; Mayr, N.;
Hatzopoulos, A. K.; de Angelis, M. H.; Wurst, W.; Bornkamm, G. W.; Brielmeier,
M.; Conrad, M. Cytoplasmic thioredoxin reductase is essential for embryogenesis
but dispensable for cardiac development. Mol. Cell. Biol. 25:1980–1988; 2005.
 Nonn, L.; Williams, R. R.; Erickson, R. P.; Powis, G. The absence of mitochondrial
thioredoxin 2 causes massive apoptosis, exencephaly, and early embryonic
lethality in homozygous mice. Mol. Cell. Biol. 23:916–922; 2003.
 Gladyshev, V. N.; Jeang, K. T.; Stadtman, T. C. Selenocysteine, identified as the
penultimate C-terminal residue in human T-cell thioredoxin reductase, corre-
sponds to TGA in the human placental gene. Proc. Natl Acad. Sci. USA 93:
 Zhong, L.; Arner, E. S.; Holmgren, A. Structure and mechanism of mammalian
thioredoxin reductase: the active site is a redox-active selenolthiol/selenenylsul-
fide formed from the conserved cysteine-selenocysteine sequence. Proc. Natl Acad.
Sci. USA 97:5854–5859; 2000.
 Cheng, Q.; Sandalova, T.; Lindqvist, Y.; Arner, E. S. Crystal structure and catalysis of
the selenoprotein thioredoxin reductase 1. J. Biol. Chem. 284:3998–4008; 2009.
 Lee, S. R.; Bar-Noy, S.; Kwon, J.; Levine, R. L.; Stadtman, T. C.; Rhee, S. G.
Mammalian thioredoxin reductase: oxidation of the C-terminal cysteine/
selenocysteine active site forms a thioselenide, and replacement of selenium
with sulfur markedly reduces catalytic activity. Proc. Natl Acad. Sci. USA 97:
 Zhong, L.; Holmgren, A. Essential role of selenium in the catalytic activities of
mammalian thioredoxin reductase revealed by characterization of recombinant
enzymes with selenocysteine mutations. J. Biol. Chem. 275:18121–18128; 2000.
 Biterova, E. I.; Turanov, A. A.; Gladyshev, V. N.; Barycki, J. J. Crystal structures of
oxidized and reduced mitochondrial thioredoxin reductase provide molecular
O. Rackham et al. / Free Radical Biology & Medicine 50 (2011) 689–699
details of the reaction mechanism. Proc. Natl Acad. Sci. USA 102:15018–15023;
 Sandalova, T.; Zhong, L.; Lindqvist, Y.; Holmgren, A.; Schneider, G. Three-
dimensional structure of a mammalian thioredoxin reductase: implications for
mechanism and evolution of a selenocysteine-dependent enzyme. Proc. Natl Acad.
Sci. USA 98:9533–9538; 2001.
 Arner, E. S. Focus on mammalian thioredoxin reductases—important selenopro-
teins with versatile functions. Biochim. Biophys. Acta495–526; 2009.
 Fang, J.; Lu, J.; Holmgren, A. Thioredoxin reductase is irreversibly modified by
curcumin: a novel molecular mechanism for its anticancer activity. J. Biol. Chem.
 Hashemy, S. I.; Ungerstedt, J. S.; Zahedi Avval, F.; Holmgren, A. Motexafin
gadolinium, a tumor-selective drug targeting thioredoxin reductase and ribonu-
cleotide reductase. J. Biol. Chem. 281:10691–10697; 2006.
 Lothrop, A. P.; Ruggles, E. L.; Hondal, R. J. No selenium required: reactions
catalyzed by mammalian thioredoxin reductase that are independent of a
selenocysteine residue. Biochemistry 48:6213–6223; 2009.
 Holmgren, A. Bovine thioredoxin system: purification of thioredoxin reductase
from calf liver and thymus and studies of its function in disulfide reduction. J. Biol.
Chem. 252:4600–4606; 1977.
 Luthman, M.; Holmgren, A. Rat liver thioredoxin and thioredoxin reductase:
purification and characterization. Biochemistry 21:6628–6633; 1982.
 Gasdaska, P. Y.; Berggren, M. M.; Berry, M. J.; Powis, G. Cloning, sequencing and
functional expression of a novel human thioredoxin reductase. FEBS Lett. 442:
 Gromer, S.; Arscott, L. D.; Williams Jr., C. H.; Schirmer, R. H.; Becker, K. Human
placenta thioredoxin reductase: isolation of the selenoenzyme, steady-state
kinetics and inhibition by therapeutic gold compounds. J. Biol. Chem. 273:
 Rigobello, M. P.; Callegaro, M. T.; Barzon, E.; Benetti, M.; Bindoli, A. Purification of
mitochondrial thioredoxin reductase and its involvement in the redox regulation
of membrane permeability. Free Radic. Biol. Med. 24:370–376; 1998.
 Eckenroth, B.; Harris, K.; Turanov, A. A.; Gladyshev, V. N.; Raines, R. T.; Hondal, R. J.
Semisynthesis and characterization of mammalian thioredoxin reductase.
Biochemistry 45:5158–5170; 2006.
 Fritz-Wolf, K.; Urig, S.; Becker, K. The structure of human thioredoxin reductase 1
provides insights into C-terminal rearrangements during catalysis. J. Mol. Biol.
 Missirlis, F.; Ulschmid, J. K.; Hirosawa-Takamori, M.; Gronke, S.; Schafer, U.;
Becker, K.; Phillips, J. P.; Jackle, H. Mitochondrial and cytoplasmic thioredoxin
reductase variants encoded by a single Drosophila gene are both essential for
viability. J. Biol. Chem. 277:11521–11526; 2002.
 Johansson, C.; Lillig, C. H.; Holmgren, A. Human mitochondrial glutaredoxin
reduces S-glutathionylated proteins with high affinity accepting electrons from
either glutathione or thioredoxin reductase. J. Biol. Chem. 279:7537–7543; 2004.
 Nalvarte, I.; Damdimopoulos, A. E.; Spyrou, G. Human mitochondrial thioredoxin
reductase reduces cytochrome c and confers resistance to complex III inhibition.
Free Radic. Biol. Med. 36:1270–1278; 2004.
 Jeong, W.; Yoon, H. W.; Lee, S. R.; Rhee, S. G. Identification and characterization of
TRP14, a thioredoxin-related proteinof 14 kDa:new insights into the specificity of
thioredoxin function. J. Biol. Chem. 279:3142–3150; 2004.
 Rengby, O.; Cheng, Q.; Vahter, M.; Jornvall, H.; Arner, E. S. Highly active dimeric
and low-activity tetrameric forms of selenium-containing rat thioredoxin
reductase 1. Free Radic. Biol. Med. 46:893–904; 2009.
 Turanov, A. A.; Su, D.; Gladyshev, V. N. Characterization of alternative cytosolic
forms and cellular targets of mouse mitochondrial thioredoxin reductase. J. Biol.
Chem. 281:22953–22963; 2006.
 Bowen, R. J.; Navarro, M.; Shearwood, A. M.; Healy, P. C.; Skelton, B. W.; Filipovska,
A.; Berners-Price, S. J. 1 : 2 adducts of copper(I) halides with 1,2-bis(di-2-
pyridylphosphino)ethane: solid state and solution structural studies and
antitumour activity. Dalton Trans. 48:10861–10870; 2009.
 Hickey, J. L.; Ruhayel, R. A.; Barnard, P. J.; Baker, M. V.; Berners-Price, S. J.;
Filipovska, A. Mitochondria-targeted chemotherapeutics: the rational design
of gold(I) N-heterocyclic carbene complexes that are selectively toxic to cancer
cells and target protein selenols in preference to thiols. J. Am. Chem. Soc. 130:
 Rackham, O.; Nichols, S. J.; Leedman, P. J.; Berners-Price, S. J.; Filipovska, A. A gold
(I) phosphine complex selectively induces apoptosis in breast cancer cells:
implications for anticancer therapeutics targeted to mitochondria. Biochem.
Pharmacol. 74:992–1002; 2007.
 Arner, E. S.; Sarioglu, H.; Lottspeich, F.; Holmgren, A.; Bock, A. High-level
expression in Escherichia coli of selenocysteine-containing rat thioredoxin
reductase utilizing gene fusions with engineered bacterial-type SECIS elements
and co-expression with the selA, selB and selC genes. J. Mol. Biol. 292:1003–1016;
 Cheng, Q.; Stone-Elander, S.; Arner, E. S. Tagging recombinant proteins with a Sel-
tag for purification, labeling with electrophilic compounds or radiolabeling with
11C. Nat. Protoc. 1:604–613; 2006.
 Arner, E. S.; Holmgren, A. Measurement of thioredoxin and thioredoxin reductase.
Current Protocols in Toxicology. Wiley, New York,pp. 7.4.1–7.4.14; 2000.
 Ekert, P. G.; Jabbour, A. M.; Manoharan, A.; Heraud, J. E.; Yu, J.; Pakusch, M.;
Michalak, E. M.; Kelly, P. N.; Callus, B.; Kiefer, T.; Verhagen, A.; Silke, J.; Strasser, A.;
Borner, C.; Vaux, D. L. Cell death provoked by loss of interleukin-3 signaling is
independent of Bad, Bim, and PI3 kinase, but depends in part on Puma. Blood 108:
 Eriksson, S. E.; Prast-Nielsen, S.; Flaberg, E.; Szekely, L.; Arner, E. S. High levels of
thioredoxin reductase 1 modulate drug-specific cytotoxic efficacy. Free Radic. Biol.
Med. 47:1661–1671; 2009.
 Rengby, O.; Johansson, L.; Carlson, L. A.; Serini, E.; Vlamis-Gardikas, A.; Karsnas, P.;
Arner, E. S. Assessment of production conditions for efficient use of Escherichia
coli in high-yield heterologous recombinant selenoprotein synthesis. Appl.
Environ. Microbiol. 70:5159–5167; 2004.
 Arner, E. S.; Nordberg, J.; Holmgren, A. Efficient reduction of lipoamide and lipoic
acid by mammalian thioredoxin reductase. Biochem. Biophys. Res. Commun. 225:
 Cenas, N.; Nivinskas, H.; Anusevicius, Z.; Sarlauskas, J.; Lederer, F.; Arner, E. S.
Interactions of quinones with thioredoxin reductase: a challenge to the
antioxidant role of the mammalian selenoprotein. J. Biol. Chem. 279:2583–2592;
 Anestal, K.; Prast-Nielsen, S.; Cenas, N.; Arner, E. S. Cell death by SecTRAPs:
thioredoxin reductase as a prooxidant killer of cells. PLoS ONE 3:e1846; 2008.
 Cox, A. G.; Brown, K. K.; Arner, E. S.; Hampton, M. B. The thioredoxin reductase
inhibitor auranofin triggers apoptosis through a Bax/Bak-dependent process that
involves peroxiredoxin 3 oxidation. Biochem. Pharmacol. 76:1097–1109; 2008.
 Gandin, V.; Fernandes, A. P.; Rigobello, M. P.; Dani, B.; Sorrentino, F.; Tisato, F.;
Bjornstedt, M.; Bindoli, A.; Sturaro, A.; Rella, R.; Marzano, C. Cancer cell death
induced by phosphine gold(I) compounds targeting thioredoxin reductase.
Biochem. Pharmacol. 79:90–101; 2010.
 Rigobello, M. P.; Scutari, G.; Folda, A.; Bindoli, A. Mitochondrial thioredoxin
reductase inhibition by gold(I) compounds and concurrent stimulation of
permeability transition and release of cytochrome c. Biochem. Pharmacol. 67:
 Jung, Y.; Lippard, S. J. Direct cellular responses to platinum-induced DNA damage.
Chem. Rev. 107:1387–1407; 2007.
 Arner, E. S.; Nakamura, H.; Sasada, T.; Yodoi, J.; Holmgren, A.; Spyrou, G. Analysis
of the inhibition of mammalian thioredoxin, thioredoxin reductase, and
glutaredoxin by cis-diamminedichloroplatinum (II) and its major metabolite,
the glutathione–platinum complex. Free Radic. Biol. Med. 31:1170–1178; 2001.
 Berners-Price, S. J.; Filipovska, A. The design of gold-based, mitochondria-targeted
chemotherapeutics. Aust. J. Chem. 61:661–668; 2008.
 Jellicoe, M. M.; Nichols, S. J.; Callus, B. A.; Baker, M. V.; Barnard, P. J.; Berners-Price,
S. J.; Whelan, J.; Yeoh, G. C.; Filipovska, A. Bioenergetic differences selectively
sensitize tumorigenic liver progenitor cells to a new gold(I) compound.
Carcinogenesis 29:1124–1133; 2008.
 Mandic, A.; Hansson, J.; Linder, S.; Shoshan, M. C. Cisplatin induces endoplasmic
reticulum stress and nucleus-independent apoptotic signaling. J. Biol. Chem. 278:
 Arner, E. S. Recombinant expression of mammalian selenocysteine-containing
thioredoxin reductase and other selenoproteins in Escherichia coli. Meth. Enzymol.
 Bar-Noy, S.; Gorlatov, S. N.; Stadtman, T. C. Overexpression of wild type and SeCys/
Cys mutant of human thioredoxin reductase in E. coli: the role of selenocysteine in
the catalytic activity. Free Radic. Biol. Med. 30:51–61; 2001.
catalytic agent associated with pyruvate dehydrogenase. Science 114:93–94; 1951.
 Gromer, S.; Urig, S.; Becker, K. The thioredoxin system—from science to clinic.
Med. Res. Rev. 24:40–89; 2004.
 Bindoli, A.; Rigobello, M. P.; Scutari, G.; Gabbiani, C.; Casini, A.; Messori, L.
Thioredoxin reductase: a target for gold compounds acting as potential anticancer
drugs. Coord. Chem. Rev. 253:1692–1707; 2009.
 Engman, L.; McNaughton, M.; Gajewska, M.; Kumar, S.; Birmingham, A.; Powis, G.
Thioredoxin reductase and cancer cell growth inhibited by organogold(III)
compounds. Anticancer Drugs 17:539–544; 2006.
 Mirabelli, C. K.; Johnson, R. K.; Hill, D. T.; Faucette, L. F.; Girard, G. R.; Kuo, G. Y.;
Sung, C.; Crooke, S. T. Correlation of the in vitro cytotoxic and in vivo antitumor
activities of gold(I) coordination complexes. J. Med. Chem. 29:218; 1986.
O. Rackham et al. / Free Radical Biology & Medicine 50 (2011) 689–699