NK-lysin, a disulfide-containing effector peptide of T-lymphocytes, is reduced and inactivated by human thioredoxin reductase. Implication for a protective mechanism against NK-lysin cytotoxicity.

The cytotoxic and antibacterial polypeptide NK-lysin has a molecular mass of approximately 9 kDa and contains three disulfide bonds. The activity was highly dependent on intact disulfides, because the bactericidal effect on Escherichia coli and the cytolytic effect on human 3B6 lymphocytes was inhibited when NK-lysin was treated with dithiothreitol prior to incubation with the cells. NK-lysin was a direct substrate for human or calf thymus thioredoxin reductase and preincubation of the peptide with mammalian thioredoxin reductase, and NADPH abolished its antibacterial and cytolytic activities. The addition of human thioredoxin further enhanced the inhibitory effect of thioredoxin reductase and NADPH. In contrast, e. coli thioredoxin reductase showed no direct disulfide reductase activity with NK-lysin in agreement with previous data showing large differences in structure and substrate specificity between the mammalian and E. coli enzymes. NK-lysin is the first identified macromolecular disulfide substrate for human thioredoxin reductase apart from human thioredoxin. When 3B6 cells were incubated with NADPH, thioredoxin, and thioredoxin reductase prior to addition of NK-lysin, cytotoxicity was markedly reduced. These data suggest that thioredoxin reductase inactivates NK-lysin and provides a mechanism by which the cytotoxic activity of NK-lysin is regulated.

The cytotoxic and antibacterial polypeptide NK-lysin has a molecular mass of approximately 9 kDa and contains three disulfide bonds. The activity was highly dependent on intact disulfides, because the bactericidal effect on Escherichia coli and the cytolytic effect on human 3B6 lymphocytes was inhibited when NK-lysin was treated with dithiothreitol prior to incubation with the cells. NK-lysin was a direct substrate for human or calf thymus thioredoxin reductase and preincubation of the peptide with mammalian thioredoxin reductase, and NADPH abolished its antibacterial and cytolytic activities. The addition of human thioredoxin further enhanced the inhibitory effect of thioredoxin reductase and NADPH. In contrast, E. coli thioredoxin reductase showed no direct disulfide reductase activity with NKlysin in agreement with previous data showing large differences in structure and substrate specificity between the mammalian and E. coli enzymes. NK-lysin is the first identified macromolecular disulfide substrate for human thioredoxin reductase apart from human thioredoxin. When 3B6 cells were incubated with NADPH, thioredoxin, and thioredoxin reductase prior to addition of NK-lysin, cytotoxicity was markedly reduced. These data suggest that thioredoxin reductase inactivates NK-lysin and provides a mechanism by which the cytotoxic activity of NK-lysin is regulated.
Peptide antibiotics are considered to be an important part of the innate immunity of animals (1)(2)(3). A large number of antimicrobial peptides have been identified that are used under different physiological conditions. Many animals have circulating phagocytic cells containing antibiotic peptides stored in granula that participate in the destruction of engulfed bacteria (4). The mucosal surfaces of the intestine, lung, and tongue have epithelial cells that produce antibiotic peptides (5)(6)(7)(8)(9) that are likely to act extracellularly and locally protect the host from microbial invasion. In insects, antibiotic peptides are released in the hemolymph and are potent host defense effector molecules. Animal peptides that have evolved to kill engulfed microbes inside phagocytic vacuoles may be cytotoxic also to the host as exemplified by mammalian defensin and bactenecin (10,11). On the other hand peptide antibiotics that are delivered to the circulating system like insect cecropins (12) and insect defensins (13) are not harmful to the producing organism. Although most of these peptides were discovered on the basis of the antimicrobial potency, other properties such as promotion of wound healing (14,15), stimulation of monocyte chemotaxis (16), and effects on intracellular signal transduction pathways (17,18) have been reported. Recently, a novel antimicrobial peptide, NK-lysin, was isolated and shown to be present in CD2 ϩ , CD4 ϩ , and CD8 ϩ cells, suggesting a function as effector peptide in T-lymphocytes (19). NK-lysin is a cyst(e)ine-containing peptide that in addition to its antimicrobial activity is also cytotoxic to certain tumor cells but does not lyse red blood cells. The six cysteines in NK-lysin form three disulfide bonds that are well characterized (19).
Thioredoxin is a small protein (12 kDa) present in all prokaryotic and eukaryotic cells (20). In its oxidized form (Trx-S 2 ), 1 it contains a redox active disulfide with the sequence Cys-Gly-Pro-Cys localized in a protrusion of the known threedimensional structure (21,22). Trx-S 2 is reduced by the flavoenzyme thioredoxin reductase (20) and NADPH (the thioredoxin system), and Trx-(SH) 2 operates as a general protein disulfide reductase (Reactions 1 and 2): Trx-(SH) 2 ϩ protein-S 2 9 | = Trx-S 2 ϩ protein-(SH) 2 REACTION 2 The flow of electrons in catalysis by thioredoxin reductase is from NADPH to FAD and then to the active site disulfide, which forms a dithiol that reduces the disulfide of oxidized thioredoxin (23). Escherichia coli TR has a molecular mass of 70 000, with two identical subunits and known three-dimensional structure (24). Mammalian thioredoxin reductases are larger proteins (M r 116,000) also with two identical subunits but a wider substrate specificity as compared with the E. coli enzyme (25,26).
The thioredoxin system has been implicated in a large variety of physiological processes (20). It serves, for example, as a hydrogen donor for ribonucleotide reductase (27) and a regulator of enzymes by thiol redox control (28). Trx also modulates the activity of transcription factors such as NF-B (29), AP-1 (30), and steroid receptors (31,32). More recently several cyto-kines or cytokine-like factors such as adult T cell leukemiaderived factor (33), 3B6-interleukin-1 (34), T-hybridoma-derived (MP-6) B cell stimulatory factor (35), and early pregnancy factor (36) have been reported to be identical to thioredoxin. Trx is thus secreted by activated human lymphocytes (37). These results show that Trx has cytokine-like extracellular activities.
In contrast to thioredoxin, which can act as a general disulfide oxidoreductase, thioredoxin reductase has only few known substrates. Beside small molecules like vitamin K, alloxan, and selenite (26,38,39), only Trx and proteins that contain thioredoxin domains (protein disulfide isomerase, CaBP1, and CaBP2) are known substrates for thioredoxin reductase (40,41). In addition to being located in the cytosol, thioredoxin reductase has been reported to be present on the plasma membranes as well as on the surface of some cancer cells and could be responsible for the reductive activation of extracellular Trx (42). Thus the Trx system may play an important role in modulating the activity of extracellularly circulating peptides and proteins. The Trx system has previously been shown to prevent toxic effects of venom neurotoxin (43) and cytolysis by tumor necrosis factor (44), but the mechanism of protection is not clear.
In this report, we show that the antibacterial and cytotoxic activities of NK-lysin are dependent on intact disulfide bonds. NK-lysin kills human 3B6 lymphocytes, and this lytic effect may be inhibited by pretreatment of the peptide with thioredoxin reductase or the complete Trx system.

MATERIALS AND METHODS
Recombinant human thioredoxin was prepared as described by Ren et al. (45); E. coli thioredoxin and E. coli TR were prepared as described (20). Mammalian thioredoxin reductase was prepared from human placenta or calf thymus essentially as described by Luthman and Holmgren (26). Yeast glutaredoxin reductase was from Sigma. RPMI medium was from Flow Laboratories (Ayshire, UK), and fetal calf serum, glutamine, and antibiotics were from Life Technologies, Inc. The cyto96 cytotoxic kit was from Promega. The columns used for HPLC analyses were: C18 column (Vydac 4.6 ϫ 250 mm, The Separation Group, Hesperia, CA) and C8 column (Sephasil 2.1 ϫ 10 mm, Pharmacia Biotech Inc.). All other chemicals were from Sigma.
Assay of Antibacterial Activity-Thin plates (1 mm thick) were poured with LB broth plus Medium E (46), 1% agarose, and about 6 ϫ 10 4 cells/ml of E. coli D21 (47). Small wells (diameter, 3 mm) were punched in the plates, and 3-l test samples were applied in the wells. After overnight incubation at 30°C, the diameter of inhibition zones were recorded with a magnification lens with an internal mm scale. The activity was monitored as inhibition zones, and units were read from a standard plot of zone diameter against the log of the amounts of cecropin A.
Assay of Cytotoxic Activity-The cell lines used were 3B6 and 1G8, Eppstein-Barr virus-infected B cell lines originating from the lymphoblastoid cell line 721/84.5 (48), and U937, a human histocytic lymphoma cell line. They were maintained suspended in RPMI 1640 culture medium supplemented with 10% heat-inactivated fetal calf serum, glutamine, and antibiotics in 5% CO 2 atmosphere. For experiments, the cells were washed, viable cells were counted after trypan blue staining, and the cell density was adjusted to 2 ϫ 10 5 cells/ml in RPMI 1640, 2% fetal calf serum. Incubations were carried out in a 96-well flat microtiter plate at 37°C using 100 l of target cell suspension and 10 -20 l of peptide sample diluted in medium (test) or medium only (control). Routinely, duplicate or triplicate determinations were made, and the specific lysis was calculated after 4-h incubations, measuring absorbance at 490 nm using the cyto96 cytotoxic kit. Total lysis was determined by adding lysis buffer to the cells. Specific lysis was calculated as (test A490 Ϫ control A490)/(total A490 Ϫ control A490).
Reduction of NK-lysin-Reactions were carried out in a final volume of 10 -40 l in 50 mM Tris-HCl buffer, pH 7.5, 1 mM EDTA, and 0.25 mM NADPH. All enzymes used were added and kept on ice prior to the addition of NK-lysin (final concentration, 20 M) and subsequent incubation for 20 min at 37°C. The samples were then cooled on ice, and 3-l samples were taken for antibacterial assay (or 15-l samples for cytotoxic assays).

Effect of DTT on NK-lysin Antibacterial Activity-
The anti-E. coli activity of NK-lysin is apparently dependent on intact S-S bonds. Fig. 1 shows the effect of 20 min of treatment at 37°C of NK-lysin with various concentrations of DTT prior to incubation with bacteria. NK-lysin activity is dose-dependently inhibited by DTT, and at a 15-fold molar excess of DTT over half-cystines, a complete inactivation of anti-E. coli activity was observed. DTT only did not kill the cells at concentrations up to 5 mM, the highest concentration tested. The importance of intact S-S is in line with findings that the pore-forming activity of amoebapore A, a structurally similar peptide to NK-lysin, is also inactivated after DTT treatment (49).
Reduction of NK-lysin by Thioredoxin Reductase-Mammalian thioredoxin reductases from either human placenta (HP-TR) or from calf thymus gave similar results, both dose-dependently inactivating NK-lysin with a half-maximal inhibition of anti-E. coli activity at approximately 50 nM TR, a 400:1 peptide to enzyme ration on a molar basis (Fig. 2A). The inactivation was not species-specific because both enzymes gave similar results.
In contrast to the mammalian enzyme, E. coli TR does not inactivate NK-lysin. Even higher concentrations of E. coli TR (up to 0.5 M) did not show any effect. When NK-lysin was incubated with reduced human Trx or E. coli Trx alone at concentrations up to 5 M, 100% of its anti-E. coli activity was preserved.
We then tested if the complete thioredoxin system was a more efficient inactivating system. The concentrations of the HP-TR and the E. coli TR were kept constant at 50 nM while increasing concentrations of the homologous thioredoxins were used. As shown in Fig. 2B, human Trx dose-dependently enhanced the inactivation of NK-lysin anti-E. coli activity. E. coli Trx also inactivated NK-lysin when combined with E. coli TR but not as efficiently as the mammalian homologue. NK-lysin incubation with the Trx system was also performed in the presence of 0.1 mM phenylmethylsulfonyl fluoride, 1 M leupeptin, and 1 M pepstatin with no change in the results (data not shown), demonstrating that the reduction in activity is not caused by enzymatic degradation.
Protective Effect of Thioredoxin Reductase against NK-lysin Cytotoxicity-As shown in Fig. 3, NK-lysin is a potent lytic agent of 3B6 cells, with approximately 50% of the cells lysed at 20 g/ml (2.2 M) after 4 h of incubation. NK-lysin also lysed the 1G8 cells as well as the transformed U937 cells at similar concentrations (data not shown). The lysis proceeds with a rapid initial phase and is clearly detected after 10 min. After that the lytic effect continues more slowly. The reason for this is not clear, but one explanation may be the heterogeneity of the cell population with some cells being more sensitive for NK-lysin than others.
When NK-lysin was preincubated with 2.5 mM DTT or 220 nM HP-TR or with the complete thioredoxin system (2.8 M human Trx/60 nM HP-TR) and 0.25 mM NADPH prior to addition to 3B6 cells, more than 95% of the lytic effect was blocked (Fig. 4). If NK-lysin was incubated with 5 M human Trx alone, there was no inactivating effect.
Likewise, if 3B6 cells were incubated for 15 min with a fixed concentration of TR (25 nM) and increasing concentrations of human Trx prior to the addition of NK-lysin, there was a dose-dependent inhibition of the cytolytic activity (Fig. 5). This protective effect was not seen when the cells were incubated with human Trx alone. At present, the limited amount of TR available did not permit experiments with preincubation of cells with higher concentrations of TR only.
HPLC Analyses of the TR-treated NK-lysin-To demonstrate that NK-lysin is reduced by thioredoxin reductase, 9 nmol NK-lysin were treated with 100 nM HP-TR for 45 min under identical conditions as described above. The ϪSH groups of the cysteines were labeled with 4-vinyl pyridine (2.85 mol) for 2 h at room temperature, and the products were acidified and separated by reverse-phase HPLC on an analytical C18 column. The resulting chromatogram showed two major peaks, a and b, representing more than 70% of eluted material according to their absorbance at 214 nM (Fig. 6B). These two peaks were collected, concentrated, and subsequently digested with endoproteinase Asp-N, which cleaves NK-lysin at positions between all cysteines except Cys 70 and Cys 76 (Fig. 6A) (19). The resulting fragments were separated by reverse-phase HPLC using an analytical C8 column (simultaneous detection at 214, 280, and 254 nm), and their identity was determined by either N-terminal sequencing or total amino acid composition. This analysis showed that in peak a all cysteines were alky-lated, and in peak b all Cys 35 and Cys 45 were alkylated, whereas no alkylation was found in Cys 4 and Cys 7 . This indicates that the Cys 35 -Cys 45 disulfide bond is more amenable to reduction by TR than Cys 4 -Cys 76 or Cys 7 -Cys 70 . NK-lysin not treated with TR and incubated in parallel with 4-vinyl pyridine did not change its elution mobility and gave no change in absorption at 280 nm, suggesting no interference of 4-vinyl pyridine with the disulfide bonds.

DISCUSSION
Many effector molecules contribute to host immunity, and one group that may play an important role is the group of antibiotic peptides. At present, a large number of peptide sequences (or cDNA sequences coding for peptides) with antibiotic activity are known (3). On a chemical basis, these peptides may be divided into groups based on structural similarities, e.g. (i) linear peptides devoid of cysteines, often forming amphipatic helices, (ii) linear peptides with a high proportion of certain residues like proline and arginine, (iii) loop-forming peptides with one disulfide bond, and (iv) peptides with one or more internal disulfide bonds.
A 78-residue peptide, NK-lysin, belonging to group iv was recently purified from pig intestine and shown to have a disulfide pattern different from other members in this group like the ␣and ␤-defensins (19). Immunostaining with a polyclonal antibody against NKL co-stained with monoclonal markers for CD2 ϩ , CD4 ϩ , or CD8 ϩ , suggesting its presence in cytolytic T-lymphocytes (often localized in granules) (19). Most, but not all, peptide antibiotics are believed to interact with the lipid bilayer and express their activity through membrane destruction. Defensins are postulated to form ␤-sheet structures that dimerize (50) and interact with the lipid bilayer that may eventually form multimeric pores (51). The structure and mode of action is not known for NKL, but the structural similarities of NKL to the saposin-like family of peptides suggest that membrane interaction is likely to occur (52). Cysteine-rich peptides like snake toxins and amoebapore A are usually stable on exposure to acid or high temperature but loose all activity upon disulfide reduction (49,53). On the other hand reduction of disulfide bonds in guinea pig defensin does not reduce its biological activities (54). NK-lysin also resists 100°C for 10 min but is inactivated by DTT. Thus it was of interest to investigate if the thioredoxin system could modulate the activity of NK-lysin because NK-lysin is potentially an effector peptide of pig cytotoxic T lymphocytes. Trx has been shown to have many extracellular activities, some playing a role as mediators in the immune system or as regulators of cell growth (35,48,55) in an autocrine manner. Trx may also prevent venom neurotoxin (43) and tumor necrosis factor (44) cytotoxic effects. In this report we show that thioredoxin reductase alone or the complete Trx system can inhibit the antibacterial and cytotoxic effects of NK-lysin.
The Trx system has a broad substrate specificity and can reduce disulfides in a number of proteins and peptides (56) and also inhibits the anti-E. coli activity of NK-lysin, most probably by reduction of disulfides. Thioredoxin reductase has a narrower substrate specificity than Trx, and so far only a limited number of substrates, all proteins containing thioredoxin domains, are identified (40,41). Interestingly, NK-lysin is inactivated by mammalian TR in the absence of Trx and represents a new substrate for TR. E. coli TR is more substrate-specific than mammalian TR (56) and does not affect the activity of NKL (this might be a positive factor for the anti-E. coli activity of NKL). Another protein with thiol-disulfide reductase activity present in mammalian cells is glutaredoxin, which is reduced by glutathione and glutathione reductase, the glutaredoxin system (57). Both human and E. coli glutaredoxin efficiently inactivate NK-lysin when combined with yeast glutaredoxin reductase, glutathione, and NADPH. 2 The complete system was needed, and no single component showed any effect.
NK-lysin is cytolytic for the mouse tumor cell line YAC-1 (19) and also kills virus-infected B-lymphocytes (3B6) as well as other cells (1G8 and U937) at similar half-maximal concentrations (2-5 M). In contrast with defensin (58), the lysis proceeds rapidly and is clearly detected within 10 min compared with a lag time of approximately 4 h for defensin. Also, NKL lytic activity is not efficiently quenched by serum protein (Ref. 19 and this paper) as is the case with defensin (58). At present, red blood cells are the only example of mammalian cells not lysed by NKL. The reason for this is not clear. TR appears not to specifically reduce a particular disulfide bond in NKL because all cysteines could be modified by 4-vinyl pyridine. However Cys 35 -Cys 45 was preferentially reduced because no Cys 35 -Cys 45 bond could be found in the partially reduced NK-lysin.
Thioredoxin reductase has been reported to be associated with plasma membranes and in some cells to be present on the surface. Thus, the amount of thioredoxin reductase in different cell types may influence the sensitivity of certain cells to NKlysin. Thioredoxin reductase as well as thioredoxin vary during the cell cycle, reaching a maximum at the S phase and are decreased when cells enter G 2 /M. 3 These variations during the cell cycle, and the heterogeneity of the cell population may explain why NKL lysis proceeds with a fast initial rate (Fig. 3). Immunohistochemical localization showed that TR and Trx and colocalized in most cells. Nevertheless, there are cells, e.g. spermatocytes and Sertoli cells, that are positive for thioredoxin reductase but negative for thioredoxin (59). These observations suggest that thioredoxin reductase has more functions than reducing thioredoxin.
Thus, we describe a novel mechanism by which the cytotoxic effect of NK-lysin can be inhibited. The direct action of thioredoxin reductase alone or of the thioredoxin system can inactivate NK-lysin and could play a role in modulating NK-lysin cytotoxicity. However, other mechanisms such as dilution or quenching by protein absorption may also reduce the toxic effect of NK-lysin. A, schematic drawing of the S-S bonds in NK-lysin. B, NK-lysin was reduced by 100 nM TR and 0.25 mM NADPH, and the SH group of the cysteines was reacted with 4-vinyl pyridine and analyzed by reversephase HPLC on a C18 column.