Truncated thioredoxin is a mitogenic cytokine for resting human peripheral blood mononuclear cells and is present in human plasma.

Human thioredoxin (Trx) catalyzes intracellular disulfide reductions but has also co-cytokine activity with interleukins after leaderless secretion. A recombinant truncated form of thioredoxin with the 80 N-terminal residues (Trx80) was purified to homogeneity. We discovered that Trx80 by itself is a potent mitogenic cytokine stimulating growth of resting human peripheral blood mononuclear cells. No effect was seen by Trx, but Trx80 at 50-100 nm induced cell proliferation of human peripheral blood mononuclear cells in serum-free synthetic medium, measured as [(3)H]thymidine incorporation after 72 h, with a maximum effect being comparable with that of 5 units/ml of interleukin-2. Trx80 lacked redox activity, but CD spectra suggested a secondary structure similar to Trx. Reduced Trx80 had an M(r) of 25,000, indicating that it is a dimer in solution. We also developed two different sandwich enzyme-linked immunosorbent assays that distinguish between full-length Trx and Trx80 and determined plasma levels of Trx and Trx80 in blood donors. The levels of Trx80 varied from 2 to 175 ng/ml; in comparison levels of Trx varied from 16 to 55 ng/ml without correlation to Trx80. In conclusion, the naturally occurring Trx80 is a novel mitogenic cytokine for normal resting human blood mononuclear cells.

Human thioredoxin (Trx) catalyzes intracellular disulfide reductions but has also co-cytokine activity with interleukins after leaderless secretion. A recombinant truncated form of thioredoxin with the 80 N-terminal residues (Trx80) was purified to homogeneity. We discovered that Trx80 by itself is a potent mitogenic cytokine stimulating growth of resting human peripheral blood mononuclear cells. No effect was seen by Trx, but Trx80 at 50 -100 nM induced cell proliferation of human peripheral blood mononuclear cells in serum-free synthetic medium, measured as [ 3 H]thymidine incorporation after 72 h, with a maximum effect being comparable with that of 5 units/ml of interleukin-2. Trx80 lacked redox activity, but CD spectra suggested a secondary structure similar to Trx. Reduced Trx80 had an M r of 25,000, indicating that it is a dimer in solution. We also developed two different sandwich enzyme-linked immunosorbent assays that distinguish between full-length Trx and Trx80 and determined plasma levels of Trx and Trx80 in blood donors. The levels of Trx80 varied from 2 to 175 ng/ml; in comparison levels of Trx varied from 16 to 55 ng/ml without correlation to Trx80. In conclusion, the naturally occurring Trx80 is a novel mitogenic cytokine for normal resting human blood mononuclear cells.
Thioredoxins (12 kDa) are ubiquitous enzymes that catalyze thiol-disulfide exchange reactions via two Cys residues in the conserved active site sequence -Trp-Cys-Gly-Pro-Cys-(1-3). The three-dimensional structure of thioredoxins (the thioredoxin fold) is characterized by a central five-stranded sheet surrounded by four ␣-helices (4 -7). The active site disulfide in the oxidized Trx is reduced to a dithiol by NADPH and TrxR 1 (8,9). Human T-lymphotropic virus-1 transformed human T cells produce a factor previously named adult T cell leukemiaderived factor, which is identical to human Trx (10). Trx is also secreted from several other cell types including activated nor-mal B lymphocytes, B cell lines from B-type chronic lymphocytic leukemia, liver cells, fibroblasts, and T lymphocytes (11,12). Among its regulatory functions, Trx is involved in redox control of DNA binding of transcription factors like NF-B and AP-1 (13)(14)(15), the latter by a direct association with redox factor 1 (16). Trx up-regulates IL-2 receptors in leukemic T cells (17). Moreover, Trx promotes cell growth of several malignant cell types, e.g. liver cell lines (PLC/PRF/5), B cell lines (BL41), or lymphoblastoid cell lines (CRAG8, CRB 95, and 1G8) (18 -20) and prevents apoptosis via direct binding interaction of the reduced form with apoptosis signal-regulating kinase 1 (21). Chemokine activity of Trx has also recently been discovered in vivo and in vitro, and this function is not mediated via known chemokine receptors (22). An active site mutant with the cysteine residues substituted to serines is not active as a chemokine, suggesting that the redox function of Trx is required for its chemokine activity (22).
By using monoclonal antibodies to Trx and Trx80, endogenous Trx80 was localized on the plasma membrane of the monocyte and macrophage cell lines THP-1 and U937, whereas Trx was found on the cell surface of many different cells (29). Both Trx and Trx80 have been reported to be secreted from cytotrophoblasts (30). It was also reported that Trx80 as a fusion protein increased HIV production in HIV-infected macrophages, whereas Trx showed inhibitory effects, and when Trx was added to HIV-infected macrophages cleavage of the protein to a truncated form was seen (31), although these results have not been confirmed.
In this study we have prepared large quantities of recombinant homogenous Trx80 and characterized the molecule. We also report the discovery of a unique mitogenic cytokine activity of Trx80 and have developed methods to measure the protein in plasma samples.

EXPERIMENTAL PROCEDURES
Materials-AIM V cell culture medium and L-glutamine were purchased from Life Technologies, Inc.; primers were from Scandinavian Gene Synthesis AB; isopropyl thiogalactoside was from Saveen Biotech AB; NcoI was from Promega; Ficoll-Paque, Sephadex G-50 and G-75, and RPC  structed on the rationale that the lysine-rich region of amino acids 80 -84 in full-length thioredoxin (KKGQK) could be a favorable site for a cleavage event to take place. Two primers were synthesized (5Ј-ATT  CTA AGG AAA ACC ATG GTG AAA CAG-3Ј and 5Ј-CAC CCA CCC  ATG GTC ACT ACT TAA AAA ACT GG-3Ј) introducing an NcoI cleavage site at the ATG and an NcoI site and stop codon at the C terminus. A polymerase chain reaction was run with pACA/Trx (9) as template using 27 cycles of 94°C for 1 min, 50°C for 1 min, and 72°C for 1 min. A product of approximately 240 base pairs was purified. Escherichia coli BL-21 (DE3) cells transformed with the pET 3d-Trx80 plasmid were grown in LB medium containing 50 g/ml ampicillin at 37°C to an optical density of 0.5 at 595 nm and then induced with 0.5 mM isopropyl thiogalactoside. The induced cells were harvested by centrifugation after 4 h of induction. The bacterial pellet was dissolved in 5 ml of TE buffer with 1 mM phenylmethylsulfonyl fluoride for 1 gram of bacterial pellet and lyzed using a combination of lysozyme and sonication. Then 10 mM DTT, 2 mM MgCl 2 , and 50 g/ml DNase1 were added, and the cell lysate was centrifuged at 13,000 rpm for 30 min. Trx80 was found to be present in inclusion bodies. The pellet fraction was therefore washed with 0.5 M urea, incubated on ice for 15 min, centrifuged at 13,000 rpm for 30 min, and subsequently dissolved in 5 M urea. After this incubation the protein was soluble and was upon centrifugation at 13,000 rpm for 30 min dialyzed extensively against TE buffer. Subsequently the protein was applied to a DE52 column pre-equilibrated with TE buffer and was eluted with a linear NaCl gradient. The protein was eluted around 125 mM NaCl and apparently pure when analyzed on a silverstained 20% homogenous SDS-polyacrylamide gel. Nevertheless it was reduced by excess DTT and applied to a reverse-phase column (RPC 18, fast protein liquid chromatography) pre-equilibrated with 0.1% NH 4 HCO 3 and eluted with 0.1% NH 4 HCO 3 and 50% acetonitrile. To concentrate the protein, the acetonitrile was removed by evaporation, and the buffer was changed to TE by repeated additions using a Diaflo equipment and a 3-kDa cut-off membrane. This pure Trx80 was then aliquoted and stored at Ϫ20°C. The protein was fully characterized as pure Trx80 using N-terminal sequencing and electrospray ionization mass spectrometry (Protein Analysis Center, Karolinska Institutet). Trx, calf thymus TrxR, and Trx84 were purified as described previously (9,29,32). A Coomassie-stained 20% homogenous SDS-polyacrylamide gel with Trx80 and Trx using the Amersham Pharmacia Biotech phastgel system is shown in Fig. 1.
Estimating Concentrations of Pure Trx80 and Trx and Free SH Groups-A theoretical molar extinction coefficient for Trx80 was calculated to 8383 M Ϫ1 cm Ϫ1 (33). An experimental molar extinction coefficient was determined by dissolving 5 mg of dry protein powder in 10 mM Tris, 1 mM EDTA, pH 7.5, and A 280 nm was recorded after appropriate dilutions. This gave an experimental value of 8250 M Ϫ1 cm Ϫ1 . This was used to calculate concentrations of pure Trx80. Concentrations for Trx was measured at 280 nm and calculated with molar extinction coefficient 8250 M Ϫ1 cm Ϫ1 (9). Free SH groups were determined at A 412 nm using 1 mM 5,5Ј-dithio-bis(2-nitrobenzoic acid) and a molar extinction coefficient of 13 600 M Ϫ1 cm Ϫ1 (9).
CD Spectra-CD spectra of Trx80 was recorded at 25°C in TE buffer and was compared with the CD spectra of full-length Trx. The CD spectra analyses were performed in a cell path length of 0.2 mm on an Aviv spectrophotometer.
Gel Filtration-Gel filtration experiments were carried out on Sephadex G-50 and G-75 columns. The Sephadex columns were pre-equilibrated with TE buffer containing 1 mM DTT. Both Trx and Trx80 were reduced with 10 mM DTT prior to applying the proteins on the columns. The Sephadex G-75 column had a volume of 131 ml, the sample volume was 2.6 ml, and the flow rate was 5 ml/cm 2 /h. The concentrations of protein ranged from 1.3 to 2.6 mg/ml.
Enzymatic Assays-The NADPH consumption assay was carried out at 20°C in a total volume of 500 l with 200 M NADPH in TE buffer. The reaction cuvette contained 10 M oxidized Trx80 with three free SH groups, and the reaction was initiated by adding 10 g of calf thymus TrxR (32) to both the reference and the reaction cuvette, and the A 340 nm was recorded. This was followed by addition of 1 M Trx to both the reference and the reaction cuvette, and NADPH consumption was followed at 340 nm using a Zeiss PM Q3 spectrophotometer equipped with an automatic zero control of the reference cuvette and a recorder.
Cell Experiments-Human PBMC were prepared by standard Ficoll-Paque centrifugation of heparinized blood from healthy donors (Blood Bank, Karolinska Hospital). The cells at a concentration of 1 ϫ 10 6 cells/ml were suspended in serum-free AIM V medium supplemented with 2 mM L-glutamine, penicillin (100 units/ml), and streptomycin (100 g/ml) for all cell experiments except the experiment where the increase in cell number was investigated. In this experiment PBMC were at a concentration of 0.5 ϫ 10 6 cells/ml in RPMI medium supplemented with 10% fetal calf serum, 2 mM L-glutamine, penicillin (100 units/ml), and streptomycin (100 g/ml). Before addition to cell cultures, Trx and Trx80 were reduced with 10 mM of DTT in 37°C for 60 min. To remove any possible contamination with endotoxin, the protein solutions were passed over a column with polymyxin B-linked agarose according to manufacturer's recommendation. Subsequently DTT was removed by chromatography at ϩ4°C over a PD 10 column pre-equilibrated with degassed ice-cold PBS. The number of free SH groups was determined, and all Trx and Trx80 used in this study had at least 2.8 free SH groups per molecule. The protein solutions were finally, prior to cell culture experiments, filtered with spin-X filters to ensure sterility. All [ 3 H]thymidine assays were performed in 96-well flat bottom plates at 1 ϫ 10 6 cells/ml in a final volume of 200 l with all cultures incubated in triplicate at 37°C and 5% CO 2 in humidified cell incubators. For determination of [ 3 H]thymidine uptake, each well was pulsed with 0.5 Ci of 5-[ 3 H]thymidine for 9 h prior to harvesting onto fiberglass filters using a multi-channel cell harvester (Tomtec, Wallac, Sweden). Filters were counted with a MicroBeta Plus Scintillation counter (Tomtec, Wallac). All data reported are mean cpm of triplicate cultures. Catabolism of [ 3 H]thymidine to [ 3 H]dihydrothymine was determined as described (34 -36) and was found to be negligible in all treatments, showing that [ 3 H]thymidine uptake thereby was a trustworthy measure of DNA replication (34 -36). As controls to show that the cell growth effect seen was caused by Trx80 and not from any contaminant, two sets of experiments were performed. First, pure polyclonal goat anti-Trx80 antibodies were kept at 0.83 mg/ml in 0.2 M citrate buffer, pH 6.5, and coupled to cyanogen bromide-activated Sepharose 4B, to give a 1.5-ml column with 8 nmol of coupled antibody. Over this column 1.5 ml of Trx80 solution with 1.5 nmol of Trx80 was passed. After that 1.5 ml of PBS was added to column, and the flow through was collected; this was sterile-filtered and added to cells. In a second control experiment, the endotoxin inhibitor, polymyxin B sulfate, was added to cell cultures treated with Trx80 or lipopolysaccaride, and the [ 3 H]thymidine incorporation was determined. Cell death was measured by trypan blue exclusion and did not exceed 10% in any experiment.
The cell experiments where optical density of cell cultures were used to determine the cell number was performed as described earlier (37). Briefly, 200 l of cell suspension was seeded in 96-well plates with 3 M DTT. 2.5 g/ml phytohemoglutine A, 1 M Trx80, or 1 M Trx84 were added in triplicate to the cell suspension. Cells were grown for 6 days at 37°C and 5% CO 2 in humidified cell incubators, and the optical density at 650 nm was measured during the incubation time using a micro plate reader (Thermo max , Molecular Devices). Data were analyzed using the accompanying SOFTmax version 2.31 software. As a blank only medium and DTT was used, and as a control only 3 M DTT was used. The final concentration of DTT was 3 M in all cultures.
Specific Sandwich ELISA for Full-length Trx and Trx80 -Specific monoclonal mouse anti-Trx (clone 2G11) and anti-Trx80 (clone 7D11) antibodies were obtained as described previously (29). Standard samples of Trx and Trx80 were aliquoted at 100 g/ml in PBS with 0.5% bovine serum albumin and kept at Ϫ70°C. Each aliquot was discarded after being thawed once. 96-well plates were coated with 50 l of 10 g/ml anti-Trx or anti-Trx80 antibody in PBS and incubated at ϩ4°C overnight. Subsequently the mixture was discarded, and 200 l of incubation buffer (0.5% bovine serum albumin, 0.05% Tween 20, 0.02% NaN 3 in PBS) was added and incubated 2 h in room temperature to block unspecific protein binding sites. Plates were washed four times with washing buffer (0.05% Tween 20 in 0.9% saline). Standard dilutions of Trx or Trx80, and samples were prepared in incubation buffer, and 50 l of standards or samples were added in duplicates and incubated at ϩ4°C overnight. Thereafter plates were washed four times with washing buffer and biotinylated goat anti-Trx polyclonal antibody was added at 2 g/ml in 50 l and incubated 2 h in room temperature. Subsequently plates were washed four times with washing buffer, and 50 l of alkaline phosphatase-conjugated streptavidin diluted 1:1000 in incubation buffer was added and incubated 1 h in room temperature. Plates were then washed six times in washing buffer and 50 l of 1 mg/ml p-nitrophenyl phosphate dissolved in 10% diethanolamine, 0.02% NaN 3 , 0.5 mM MgCl 2 , pH 9.8, was added to each well, and absorbance was recorded at 405 nm by a microplate reader (Thermo max , Molecular Devices). Data were analyzed using the accompanying SOFTmax version 2.31 software.
Sample Preparation and Hemoglobin Measurements-Blood was collected anonymously by vein puncture in heparinized tubes from blood donors (n ϭ 12) at Karolinska Hospital. Samples were subsequently centrifuged at 2000 ϫ g for 20 min at room temperature and plasma was collected and kept at Ϫ20°C. Degree of hemolysis in the plasma samples was determined using a commercial plasma hemoglobin kit from Sigma following the manufacturer's instructions. The hemoglobin levels in these samples varied from 29 to 67 g/ml, corresponding to 0.02-0.05% hemolysis, assuming a normal total blood hemoglobin level of 140 g/liter. The contribution of released intracellular full-length Trx because of this degree of hemolysis to the total determined extracellular levels given in Table III was then calculated following the method of Nakamura et al. (38) and found to be 2.7-6.3 ng/ml. Crude extracts of human placenta tissue were prepared by homogenization in 2 ml of TE buffer/gram of fresh placenta tissue using a Polytron tissue disruptor. The supernatant after centrifugation at 10,000 ϫ g for 10 min at ϩ4°C was used for the Trx determinations.

RESULTS
Purification of Trx80 and Structure Analysis-The novel procedure to produce Trx80 developed in this study yielded pure protein (Fig. 1A) free from LPS and suitable for testing in cell cultures. Because it was produced as inclusion bodies and refolded from 5 M urea, we established that the refolding had succeeded. Therefore, CD spectra of Trx80 and Trx were recorded and compared; Fig. 1B shows that Trx and Trx80 had similar spectra with characteristic minima at 208 -212 and 221-223 nm. The mean residue ellipticities of these signals were, however, lower for Trx80, which also showed a small negative band around 187 nm, whereas Trx only showed a positive shoulder at the same wavelength. The loss of the C-terminal 25 residues corresponds to loss of one helix (residues 93-105 of Trx) and one strand (residues 84 -92 of Trx) (5). The CD spectra of Trx80 found here clearly established that truncated Trx80 has a substantial secondary structure and are compatible with loss of the C-terminal helix strand motif.
When reduced Trx80 was applied to a G-50 Sephadex gel filtration column and chromatographed under reducing conditions, the protein eluted with the void volume and ahead of Trx (data not shown). However, Trx80 eluted at approximately 60% of the column volume on a calibrated Sephadex G-75 column, showing an apparent M r of 25,000 or double the size of Trx, which has a molecular weight of 12,000 (2) (Fig. 2). The anomalous elution of Trx80 is compatible of a dimer in solution.
Enzymatic Activity of Trx80 -Previous results showed that Trx80 as a fusion protein, in contrast to Trx, did not catalyze the reduction of insulin disulfides by DTT (28). Our Trx80 was not a substrate for TrxR (Fig. 3). It also did not act as an inhibitor for reduction of oxidized Trx by TrxR (data not shown). We also tried to complement Trx80 with the C-terminal peptide of 24 residues of full-length Trx to examine whether this could restore any redox activity by noncovalent reconstitution (39). However, this was not the case (data not shown). Trx80 did not have any protein-disulfide isomerase-like activity in refolding of RNase when tested in an assay previously used to assess such activity (data not shown) (40). When Trx was added to a mixture containing TrxR, NADPH, and oxidized Trx80 with three free SH groups, Trx80 was reduced by Trx (Fig. 3). This showed that Trx and Trx80 interact and that Trx can reduce disulfides in oxidized Trx80, whereas Trx80 itself lacks redox activity in all systems analyzed above.
Effects of Trx80/Trx on PBMC-Knowing that Trx80 was homogeneous and devoid of detectable disulfide oxidoreductase activity, we wanted to analyze cytokine-like effects of the protein, and we chose normal human PBMC for this analysis. PBMC proliferation in RPMI medium with 10% fetal calf serum measured by optical density showed that 2.5 g/ml phytohemaglutine A increased the cell number by 60% after 4 days in culture compared with 3 M DTT alone. Trx80 at 1 M indeed showed a stimulation of cell proliferation, with a significantly better effect than that of Trx84 (35% compared with 20% increase in cell number), whereas Trx alone showed no effect compared with DTT alone (data not shown). This finding made us continue with Trx80 for further analyses, and we switched to a serum-free synthetic AIM V medium with no fetal calf serum added, because Trx is known to be present in fetal calf serum (41). We confirmed the increase in cell number by an increase in [ 3 H]thymidine incorporation after stimulation of PBMC with Trx80 using concentrations as low as 10 nM. The maximum thymidine incorporation was seen after 72 and 96 h of treatment. With 100 nM Trx80 there was a 10 -15-fold increase in thymidine incorporation as compared with untreated cells. The effect of Trx80 increased dose-dependently up to 50 -100 nM, after which the effect on cell proliferation was the same or even lower at higher Trx80. The stimulation of thymidine incorporation by 100 nM Trx80 was at the same level as that given by 5 units/ml of IL-2. In great contrast, Trx lacked PBMC-stimulating effects at concentrations up to 1 M. These results are summarized in Fig. 4. There were no synergy effects between IL-2 and Trx80/Trx or between Trx80 and Trx in stimulating cell proliferation (data not shown). To be certain that the PBMC stimulatory effects seen with Trx80 were due to a real effect of the molecule and not dependent on contaminating endotoxins, we probed the effects using polymyxin B sulfate. When cells were incubated with 1 g/ml of polymyxin B and 10 ng/ml of LPS, the growth stimulatory effects of LPS were totally abolished. When 100 nM of Trx80 was incubated with 1 g/ml of polymyxin B, however, no change in thymidine incorporation compared with treatment with Trx80 alone was seen (Fig. 5). This result demonstrated that endotoxin contamination in the Trx80 preparation was not giving the effects seen. To further ascertain that the effects were due to Trx80, we prepared a column with covalently linked polyclonal antibodies toward Trx80. The Trx80 preparation was passed over the antibody column from which the flow through subsequently was collected and added to PBMC cultures. This almost entirely abolished the cell stimulatory effect (Fig. 6).
These results showing that Trx80 but not Trx is a potent mitogenic cytokine for resting human PBMC make the question whether Trx80 can be detected in plasma interesting. To probe this question, two separate ELISA systems were developed, specific for either Trx or Trx80. These ELISA systems were based on monoclonal antibodies to Trx or Trx80 bound to plastic surface and a biotinylated polyclonal goat antibody to detect bound Trx or Trx80 (sandwich ELISA).
Sandwich ELISA for Human Trx-The ELISA for Trx revealed a reproducible titration curve for human Trx in the range 0.5-200 ng/ml, and no cross-reactivity with Trx80 was seen (Fig. 7). The detection limit, determined as three times the standard deviation above the blank, was approximately 0.5 ng/ml (Fig. 7). The intra-assay coefficient of variation was 4.5% (n ϭ 7) and 10.4% (n ϭ 7) in two different placenta extracts. The inter-assay coefficient of variation was 4.7% (n ϭ 4) and 12.2% (n ϭ 4) in two different plasma samples. The amount of Trx in plasma extracts measured by sandwich ELISA, and activity in an enzymatic thioredoxin reductase-coupled insulin disulfide reduction assay was compared and found to correlate well. Trx levels in the placenta extract was 0.26 g/mg protein measured by activity and 0.23 g/mg measured by ELISA. When amounts of Trx in placenta extracts were determined before and after addition of known amounts of Trx the recovery of Trx varied between 106 and 113%. These results are summarized in Table I. Finally a placenta extract was diluted stepwise. The dilution curve and the standard curve were parallel (Fig. 7).
Sandwich ELISA for Human Trx80 -The detection limit for the corresponding Trx80 sandwich ELISA was approximately 1 ng/ml (Fig. 8). The intra-assay coefficient of variation was 4.2% (n ϭ 7) and 4.6% (n ϭ 8) in two different plasma samples. The inter-assay coefficient of variation was 6.4% (n ϭ 4) and 7.5% (n ϭ 4) in two different plasma samples. Addition of known Trx80 to a plasma sample with known Trx80 concentration displayed a recovery between 104 and 112%. These results are summarized in Table II. Finally a plasma sample was diluted stepwise, and the dilution curve was compared with the standard curve, showing parallell curves (Fig. 8). The cross-reactivity with human full-length Trx was 3-8% (data not shown).
Levels of Trx and Trx80 in Human Plasma-Knowing that the sandwich ELISA systems were reliable and specific, we then measured Trx and Trx80 levels in human plasma. The levels of Trx in plasma varied between 16 and 55 ng/ml, and the median value was 29 ng/ml (n ϭ 12). In contrast the values for Trx80 varied significantly between different subjects (2-175 ng/ml). The median value for Trx80 was, however, lower then the same for Trx, 20 ng/ml. Interestingly, there was no correlation between Trx and Trx80 levels. These results are summarized in Table III. The hemolysis in the plasma samples measured as above was found to be negligible in all samples. DISCUSSION The effects of full-length Trx on cell proliferation as a cocytokine described in earlier reports have consistently been based on experiment with virus-transformed cells, lymphoblastoid cell lines, or established cell lines of other descent but not with normal resting cells (18,20). In this study we show that Trx80, but not full-length Trx, is capable of inducing prolifer-  a Plasma samples from two different donors were measured in duplicates seven times in the same ELISA plate. The standard deviation of the seven samples were divided with the mean to get the intra-assay coefficient of variation (CV). b Plasma sample from two different donors were measured in duplicates four times in different ELISA plates. The standard deviation of the four samples were divided with the mean to get the inter-assay coefficient of variation (CV).
c Placenta extracts were measured for Trx content with sandwich ELISA for Trx and by activity with 5,5Ј-dithio-bis(2-nitrobenzoic acid) coupled insulin assay. Total protein content in placenta extracts were measured by absorbance and calculated using the formulas 1.55(A 280 Ϫ A 310 ) Ϫ 0.76(A 260 Ϫ A 310 ) ϭ mg protein/ml and (A 230 ϫ 183) Ϫ (A 260 ϫ 76) ϭ g protein/ml. d To a diluted placenta sample with 2.6 ng/ml Trx, known amounts of Trx were added, and the levels of Trx after addition were measured. The measured value was then divided with the value that was expected (known value ϩ the amount of added Trx) and gave the recovery in percentages.
ation of normal resting human PBMC. It has previously been shown that PBMC fail to enter S phase when deprived of thiols (42), but this is unlikely to be the mechanism, because Trx failed to show the effects on PBMC seen with Trx80. In the earlier studies of truncated thioredoxin by Silberstein and coworkers (25,28,31), the protein preparation contained the maltose-binding fusion protein with which Trx80 was overex-pressed, and the studies were limited to eosinophils and macrophages, but no mitogenic effects were reported. In the present study, we have succeeded in purifying the reduced form of Trx80 to homogeneity using a modified procedure. Also, based upon the control experiments with polymyxin B sulfate and the anti-Trx80 column, there is no doubt that Trx80 is a potent mitogenic cytokine for PBMC. The MP-6 factor from a CD4 ϩ T cell clone that contained both full-length Trx and truncated forms, was observed as a co-cytokine with mitogenic effects only on a specific CLL B-cell line in the presence of other cytokines (18). This is the first time it has been shown that Trx80 exists in human plasma and that Trx80, by itself, is capable of inducing cell proliferation in the mixture of untransformed resting monocytes, B cells, and T cells constituting PBMC.
The fact that Trx80 did not act as a substrate for TrxR is consistent with the C-terminal part of Trx being crucial for its interaction with TrxR. This part harbors the conserved Gly 92 equivalent of E. coli Trx forming part of the flat and hydrophobic surface, which should be involved in binding Trx to other protein molecules (1,7). The C terminus is also important for the active site cysteines, because the deletion of the C-terminal strand-helix motif abolished the ability to catalyze DTT reduction of insulin disulfides (43). With CD spectroscopy we have ascertained that the Trx80 protein is folded. From gel filtration we present evidence that Trx80 differs from Trx in structure and that Trx80 is a dimer. The removal of the C-terminal helix and strand upon truncation could be envisioned to expose a hydrophobic surface in Trx80. We therefore suggest that dimerization could involve interactions via a hydrophobic patch exposed upon truncation of Trx. Clearly the biophysical and enzymatic properties of Trx80 and full-length Trx are very different, i.e. disulfide reducing activity, monomer versus dimer, substrate for TrxR, and it is probable that cytokine-like effects of the two proteins occur via different mechanisms.
HIV-infected individuals with low CD 4 counts have elevated Trx levels in plasma (38), and it has been suggested that Trx can be cleaved by certain cells, for example, HIV-infected macrophages to a truncated molecule (31). It is therefore probable that some of the effects on cell proliferation seen with higher concentrations of Trx actually are due to a cleavage of Trx to Trx80 by macrophages with subsequent stimulation of cell proliferation by Trx80. Furthermore, the ability of Trx to reduce Trx80 suggest that the two molecules can interact on  a Plasma samples from two different donors were measured in duplicates seven or eight times in the same ELISA plate. The standard deviation of the seven samples were divided with the mean to get the intra-assay coefficient of variation (CV). b Plasma sample from two different donors were measured in duplicates four times in different ELISA plates. The standard deviation of the four samples were divided with the mean to get the inter-assay coefficient of variation (CV).
c To a plasma sample with 6.6 ng/ml Trx80, known amounts of Trx80 were added, and the levels of Trx80 after addition were measured. The measured value was then divided with the value that was expected (known value ϩ the amount of added Trx) and gave the recovery in percentages. b Measured with sandwich ELISA for Trx80, presented as mean of duplicate values with standard deviations. About 5% (3-8%) of the measured Trx levels contributes to the given Trx80 levels because of the cross-reactivity of the Trx80 ELISA as described under "Experimental Procedures," resulting in a corrected Trx80 range of 1-171 ng/ml. the cell surface known to bind Trx80 (29). Protein-disulfide isomerase is also present at the cell surface (44), and also this protein could possibly interact with Trx and/or Trx80 as a mechanism in mediating their activation of cells. The data presented here also imply a new immunomodulatory function for full-length Trx, namely being a reservoir for a growth factor (Trx80) that can be activated by proteolytic cleavage to activate lymphocytes or as previously known to enhance the cytotoxicity of eosinophils (23-25, 28, 45). Our results imply a mechanism of function for the truncated protein other than via the classic thiol-disulfide oxidoreductase activity of full-length Trx.
The plasma levels of Trx and Trx80 do not correlate. It is also clear that the levels of Trx80 between different individuals vary more than the levels of Trx. The levels of Trx80 obtained and presented in this study are not corrected for the crossreactivity with full-length Trx in the sandwich ELISA for Trx80. This cross-reactivity is 5%, and therefore, it was crucial to use both ELISA systems when Trx80 or Trx should be measured. If the Trx80 levels in the individuals tested are correlated for cross-reactivity to full-length Trx, the levels are between 1 and 171 ng/ml. Although these plasma samples were taken from apparently healthy blood donors, these individuals were not screened for auto-immune states and diseases such as allergy and asthma. Possibly in these diseases, or other immunomodulating conditions, up-regulation of Trx80 now should activate resting cells in the PBMC, which is crucial to the function of the immune system. The highest plasma levels of Trx80 here determined correspond to approximately 20 nM, and local concentrations at cell-cell interactions could be much higher. This shows that the concentrations of Trx80 where we found effects in the PBMC cultures are well within the physiological range.
From further unpublished studies that we have conducted, it is evident that the primary target cell for Trx80 in the PBMC population is the monocyte. 2 We have purified monocytes, B cells, and T cells from PBMC and shown that when a pure population of monocytes are stimulated with Trx80, they proliferate and express several surface antigens. When pure T cells or B cells or a mixture of these two cell types were treated with Trx80, no stimulating effects were seen; however, in PBMC cultures Trx80 induces a cascade of cell-cell interactions and Th1 cytokines expression that finally activates T cells to proliferate. 2 When PBMC was cultured for 3 days in the presence of Trx80, Trx, or IL-2 and analyzed by flow cytometry using anti-CD19 and anti-CD3 antibodies to analyze the proportion of B and T cells, respectively, the number of T cells in unstimulated PBMC and in PBMC cultured with Trx was 60 -65%. In contrast, in PBMC cultures stimulated with IL-2 or Trx80, the number of CD3 ϩ T cells had increased to 70 -80% of the total cell population. Moreover, if the same number of purified monocytes or the original PBMC were stimulated with Trx80, and their thymidine incorporation was compared, the radioactivity in the PBMC was approximately 10-fold higher compared with the monocytes, indicating that cells other than the monocytes contributed to the major part of the thymidine incorporation seen in PBMC cultures. 2 We therefore state that the primary target cell for Trx80 is the monocyte and that the activation of monocytes by Trx80 leads to an activation of T cells but not of B cells. Obviously, further studies have to be conducted to delineate the mitogenic effects of Trx80 to clarify whether Trx80 operates via a specific receptor, to identify a protease that in vivo generates the cytokine from full-length Trx and to investigate the biological functions of Trx80.