Molecular Cloning and Functional Characterization of a Cell-permeable Superoxide Dismutase Targeted to Lung Adenocarcinoma Cells

In clinical oncology, many trials with superoxide dismutase (SOD) have failed to demonstrate antitumor ability and in many cases even caused deleterious effects because of low tumor-targeting ability. In the current research, the Nostoc commune Fe-SOD coding sequence was amplified from genomic DNA. In addition, the single chain variable fragment (ScFv) was constructed from the cDNA of an LC-1 hybridoma cell line secreting anti-lung adenocarcinoma monoclonal antibody. After modification, the SOD and ScFv were fused and co-expressed, and the resulting fusion protein produced SOD and LC-1 antibody activity. Tracing SOD-ScFv by fluorescein isothiocyanate and superoxide anions (\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{O}_{2}^{{\bar{{\cdot}}}}\) \end{document}) in SPC-A-1 cells showed that the fusion protein could recognize and enter SPC-A-1 cells to eliminate \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{O}_{2}^{{\bar{{\cdot}}}}\) \end{document}. The lower oxidative stress resulting from the decrease in cellular \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{O}_{2}^{{\bar{{\cdot}}}}\) \end{document} delayed the cell cycle at G1 and significantly slowed SPC-A-1 cell growth in association with the dephosphorylation of the serine-threonine protein kinase Akt and expression of p27kip1. The tumor-targeting fusion protein resulting from this research overcomes two disadvantages of SODs previously used in the clinical setting, the inability to target tumor cells or permeate the cell membrane. These findings lay the groundwork for development of an efficient antitumor drug targeted by the ScFv.

tumor formation (1)(2)(3). SOD is supposed to be the main cellular superoxide anion-degrading enzyme, inhibiting tumors (4 -6); however, ROS are essential to the existence and development of a normal cell (7)(8). For this reason, attempts to treat ROS-associated diseases with nonselective SOD have failed and may have caused deleterious effects in normal cells as well as in tumor cells. Using genetic techniques, Oyanagui (9) fused SOD with a polypeptide that could recognize heparin sulfates on vascular endothelial cells; the resultant HB-SOD could locate vascular endothelial cells by itself. In addition to potential nonspecific cell targeting, SOD has another critical disadvantage in clinical application; its large molecular weight interferes with its ability to permeate the cell membrane, which leads to a low cellular effect. Currently, a liposome is used to enclose the SOD, and the resultant liposome-SOD can permeate the cell membrane or tissue (10 -12). But thus far, modified SODs still either cannot distinguish the target cells (tumor cells) or cannot permeate the tumor cell membrane.
The single chain variable fragment (ScFv) antibody not only has a much lower immune binding than does monoclonal antibody (mAb) but also has a low molecular weight; thus, it can more easily reach intracellular targets. ScFv from the LC-1 hybridoma cell line can react at a high frequency with lung adenocarcinoma, lung squamous carcinoma, and large cell and small cell lung cancer, but not with normal and embryonic tissues (13). When this ScFv recognized and bound antigens on SPC-A-1 cells, the antigen-antibody complexes were internalized via the receptor-mediated endocytic pathway and permeated the cell membrane (14). This characteristic implies that using LC-1 ScFv to aid SOD targeting can be an effective anti-lung cancer drug because it overcomes both clinical SOD disadvantages, it will target only tumor cells and will be taken up intracellularly. In the current research, the Nostoc commune Fe-SOD and LC-1 ScFv were cloned and co-expressed, with the idea that the resultant fusion protein would possess dual activity. In theory, LC-1 ScFv would be able to introduce SOD into SPC-A-1 cells specifically and permeate the cell membrane via the receptor-mediated endocytic pathway. Once inside the cell, it would eliminate intracellular O 2 . , change the cellular redox state, and subsequently induce a pathway related to a low cellular O 2 . level, inhibiting tumor cell proliferation.
Cell Lines and Culture Conditions-Total RNA of the LC-1 hybridoma was kindly provided by Dr. Chen Liang of Yale University. N. commune stain CHEN, pET-22b (ϩ) vector, pUCm-T vector, Escherichia coli (DH5␣: supE44⌬lacU169; 80 lacZ⌬M15, hsdR17 recA1 endA1 gurA96 thi-1 relA1), E. coli BL21 (DE3: hsdS gal; cIts857, ind1 Sam7 nin5 lacUV5-T7), SPC-A-1 lung adenocarcinoma cell line, AGS gastric cancer cell line, and the HepG2 liver carcinoma cell line are available in our institute. N. commune was cultivated in BG11 culture medium. E. coli DH5␣ and E. coli BL21 (DE3) were maintained in LB culture medium. SPC-A-1 cells, AGS cells, and HepG2 cells were cultured in RPMI 1640 medium, which contained 10% fetal calf serum. Tumor cells were maintained in a CO 2 cell culture incubator (Forma Scientific, 3154 CO 2 Water-Jacketed Incubator) at 37°C under 5% CO 2 . To prepare for every experiment, 90 l of cells were grown in a 96-well plate until 70 -80% confluent. All experiments were repeated at least three times. Data shown in the graphs for various parameters represent the mean Ϯ S.E. Differences between means were considered significant when p Ͻ 0.05.
Construction of the pET-SOD Expression Vector System-A 0.2-g N. commune sample was taken, and its genomic DNA was isolated using the method of Kim et al. (15). Fe-SOD was amplified from the genomic DNA, primered with degenerate primer, SOD-forward (5Ј-CTAGCC-ATGGCATTTGTACAGC-3Ј, including the NcoI site) and SOD-back (5Ј-CCGCTCGAGAGCTTT(G/A)GC(C/A)AAGTT(T/C)TC-3Ј, including the Xhol I site), and cloned into pET-22b (ϩ) to be pET-SOD.
Expression, Purification, and Immunoblot Analyses of SOD, ScFv, and SOD-ScFv-The pET-SOD was induced in the presence of 0.5 mM IPTG and 300 M [Fe 3ϩ ] in 1 liter of LB culture medium at 37°C for 3 h. Cells were harvested by centrifugation at 5000 ϫ g and lysed in 50 ml of TE (20 mM Tris-HCl and 10 mM EDTA, pH 8.0). The lysate was disrupted using sonication (800 W, 20 min). The achieved sedimentation (SOD) after sonication was washed with wash buffer I (1 mM EDTA, 5% glycerol, 0.5% Triton X-100) and buffer II (wash buffer I with 2 M carbamide). Then the sedimented material was dissolved in lysis buffer (50 mM Tris-HCl, 0.3% sodium lauryl sulfate, 0.1 mM dithiothreitol) overnight. The lysate was poured into a column with 5 ml of nickel-nitrilotriacetic acid Superflow Resin, and the packed resin was washed with buffer consisting of imidazole gradient concentrations (0 -500 mM). A 280 was measured. The denatured protein was renatured in lysis buffer (excluding sodium lauryl sulfate) that included 100 M [Fe 3ϩ ], at 4°C. The SOD-ScFv fusion protein was expressed and purified from pET-SOD-ScFv as described above. The ScFv was expressed and purified from pET-ScFv as described above, except that there was no [Fe 3ϩ ] in the LB culture medium or the renaturing buffer. Natural Fe-SOD protein was purified from N. commune directly as follows: 500 g of N. commune was lysed in 1 liter of PBS (including 0.5 mM EDTA, pH 7.6) and disrupted using sonication (1 200 W, 60 min). Then the crude materials (Fe-SOD) were deposited in 45-85% (NH 4 ) 2 SO 4 for 6 h. The crude Fe-SOD was dissolved again and purified by DEAE-Sepharose fast flow and Sephadex G-100, respectively.
The expression of pET-SOD and pET-SOD-ScFv was subjected to Western blot analysis (the anti-SOD antiserum was raised against natural N. commune Fe-SOD in mice). The SOD activity of purified SOD and SOD-ScFv was assayed using the pyrogallol auto-oxidation method (17). The antibody activity of purified ScFv and SOD-ScFv was assayed by competitive enzyme-linked immunosorbent assay as follows: 0.01 g/l LC-1 mAb and the purified ScFv (or SOD-ScFv) were used to bind SPC-A-1 cells competitively (AGS cells and HepG2 cells served as controls); goat anti-mouse-labeled anti-IgM secondary antibody was used to bind to the LC-1 antibody. A 490 was measured.

Determination of the Cell Membrane Permeation Ability of SOD-ScFv by Fluorescein Isothiocyanate (FITC) Fluorescence Tracing and
Immunoblot Analyses-To trace the permeation process of SOD-ScFv through the SPC-A-1 cell membrane, SOD, ScFv, and SOD-ScFv were labeled with FITC essentially according to Lindsay et al. (18). A 0.1 M sample was dissolved in 10 ml of 100 mM carbonate buffer, pH 9.0. To each sample was added 10 l of 1% w/v FITC in dimethyl sulfoxide, and labeling was performed at room temperature overnight. Each labeled sample was dialyzed at 4°C for 72 h against pH 8.0 PBS. The samples were additionally separated by a Sephadex G-25 column. The fluorescence intensity of sample was measured by a fluorescence spectrophotometer (Hitachi, 850, Japan).
The prepared SPC-A-1 cells were washed twice with PBS and incubated in RPMI 1640 medium containing 0.2 M SOD-ScFv-FITC for 20, 40, and 60 min at 37°C. Following three washes to remove unbound protein, intracellular FITC fluorescence was observed by LCSM (LSM 510, Zeiss, Germany). The intracellular stability of SOD-ScFv was estimated by Western blotting as follows. After SPC-A-1 cells were treated with 0.2 M SOD-ScFv for 1 h, cells were washed and changed with a fresh culture medium. Then, cells were further incubated for 0, 6, 12, and 24 h followed by Western blot analysis with anti-SOD antiserum.
Determination of Targeted Antitumor Ability of SOD-ScFv-A total of 90 l of SPC-A-1 cells were grown in a 96-well plate until 30% confluent. Then, 10 l of SOD-ScFv of the working concentrations (0.2-2 M) were added for 1 h, and cells were washed and changed with a fresh culture medium for 48 h. Cells were counted using a hemocytometer (AGS cells and HepG2 cells served as negative controls); a cycle of cells (treated as described above) from a 6-well plate was analyzed by flow cytometry (FAC sort, Becton Dickinson) according to Mahadevan et al. (19).  (22).

Analysis of Mitochondrial Transmembrane Potential (⌬ m ) by
Fluorescence Spectrophotometer-Mitochondrial membrane potential (⌬ m ) in SPC-A-1 cells was measured using two different dyes, rhodamine-123 and CM-H 2 TMRos. Rhodamine-123, a conventional, cationic voltagesensitive probe that reversibly accumulates in mitochondria, was used in the present study to detect changes in ⌬ m . Cells were incubated with 1 M rhodamine-123 at 37°C for 30 min after 1 h of drug incubation. After three washings, the fluorescence intensity of the sample was analyzed by fluorescence spectrophotometer. CM-H 2 TMRos is a new fluorescent dye developed recently that is more photostable and stable in mitochondria (23). It was also used to measure the ⌬ m as described above, excepting cells that were incubated with 500 nM CM-H 2 TMRos.
Cellular Oxidative Stress and Reducing Power Assay by Flow Cytometry-The oxidative stress of SPC-A-1 cells was determined by detection of cellular ROS with DCFH-DC (24). Briefly, after incubation with 0.2 M SOD-ScFv in fresh culture medium for 1 h, cells were washed and incubated with 5 M DCF-DA for 20 min at 37°C. Cells were then washed three times with probe-free nitrate buffer. The fluorescence intensity of the sample was measured by flow cytometry. The cellular reduced glutathione (GSH) level was measured as an index to reflect the cellular reducing power in SPC-A-1 cells; GSH is oxidized to oxidized glutathione under the stress of superoxide anions or hydrogen peroxide. The GSH level was determined by staining cells with 500 M NDA (25) as described above, then cells were washed three times, and fluorescence intensity was measured by flow cytometry.
Reduced Oxidative Stress Effect on SPC-A-1 Cells through Akt/p27 kip1 Pathway-A total of 1.8 ml of SPC-A-1 cells were grown in a 6-well plate until 80% confluent, then treated with SOD-ScFv at 37°C for 1 h (PBS served as control). After treatment, 10 6 cells were disrupted, and then cellular phosphorylated Akt (p-Akt) levels and insulin-like growth factor-I (IGF-I) levels were analyzed by Western blotting.
To analyze the expression levels of cyclin-dependent kinase inhibitor p27 kip1 , 1.8 ml of SPC-A-1 cells were grown in 6-well plate until 30% confluent, 200 l of SOD-ScFv were added for 1 h (PBS served as control), and cells were washed and changed with a fresh culture medium for 24 h. A total of 10 6 cells were disrupted, and the cellular p27 kip1 level was analyzed by Western blotting.

RESULTS
Based on data for other similar Fe-SODs of prokaryotic algae in Gen-Bank TM , we assumed that the Fe-SOD coding region in N. commune genome was continuous. Based on these Fe-SOD sequences, the degenerate primer SOD-forward/SOD-back was designed, and a fragment (GenBank TM accession AY830114) of 603 bp was achieved, whose deduced protein was in accordance with the molecular weight of natural Fe-SOD, which was purified from N. commune (22 kDa). These findings implied that this 603-bp coding sequence was the N. commune Fe-SOD CDS. This conclusion was supported by Western blotting analysis of expressed Fe-SOD or SOD-ScFv against natural Fe-SOD antiserum ( Fig. 2A, lanes 3 and 4). BLAST search results from GenBank TM showed that this CDS had about 85% nucleotide similarity with Nostoc linckia Fe-SOD. It was found that when the cells were induced by IPTG, additional FeCl 3 could improve the yield of the expected protein (26). The expressed SOD, ScFv, and SOD-ScFv proteins were composed of ϳ78, 65, and 21%, respectively, of the total bacterial protein in the pET-SOD, pET-ScFv, and pET-SOD-ScFv expression vector systems, respectively (data not shown). The SOD activity assay result indicated that the specific activity of expressed SOD and SOD-ScFv was 1100 units/mg and 700 units/mg, respectively, whereas that of natural N. commune Fe-SOD was 2700 units/mg. The competitive enzyme-linked immunosorbent assay result of ScFv showed that the inhibitory rate of 0.1 g/l ScFv to 0.01 g/l LC-1 antibody was 71.0%, whereas that of ScFv-SOD was 61.2%. But neither showed the ability to bind to AGS cells or HepG2 cells (data not shown).
The ability of SOD to permeate the cell membrane is a key point in clinical applications of SOD. Based on the specific spatial structure of SOD, ScFv, or SOD-ScFv, we dyed each with FITC (27) to determine their ability to permeate the cell membrane and their cellular distribution after permeation. After incubation with SOD-ScFv-FITC for 20 min, the SPC-A-1 cells began to show green fluorescence (Fig. 3E). From Fig.  3I, the FITC localized inside the cell but did not reside on the cell membrane; what is more, it was distributed equally in the cells (Fig. 3, H and  I). The ScFv-FITC control exhibited the same pattern (Fig. 3C, some data not shown). However, the SOD-FITC control showed no obvious fluorescence (Fig. 3B), implying that the SOD barely permeated the cell membrane by itself. The intracellular stability of SOD-ScFv in SPC-A-1 cells was shown in Fig. 2B. After permeation, the SOD-ScFv was degraded gradually; however, it could persist in cells for up to 24 h.
The targeted antitumor ability of SOD-ScFv is shown in Fig. 4. Cells treated with 0.2 M SOD-ScFv showed a 46.6% inhibition rate; this increased significantly in a SOD-ScFv dose-dependent manner (Fig.  4A). In addition, SOD-ScFv did not show the same inhibition in AGS cells or HepG2 cells (Fig. 4B), which implied the specificity of SOD-ScFv for SPC-A-1 cells. To determine how SOD-ScFv inhibited growth of SPC-A-1 cells, a cycle of treated cells was performed (Fig. 4C). It was clear that there was a change in the distribution of cells in G 1 phase as a result of cellular overload of SOD. In the SOD-ScFv model, 60.4% of the cells were in G 1 phase, whereas only 50.4% of the SOD-treated cells, 51.7% of the ScFv-treated cells, and 47.4% of PBS-treated cells were in G 1 . A considerable increase in the G 1 fraction of SOD-ScFv model cells suggested that the SOD-induced inhibition in cell growth could be associated with the delay at G 1 .
The cellular O 2 . analysis result is shown in Fig. 5A. The oxidation rate of DHE in SPC-A-1 cells decreased significantly after pretreatment with SOD-ScFv. However, pretreatment of SOD or ScFv produced an oxidation rate not significantly different from PBS control, which showed the low permeation ability of SOD or the invalidation of ScFv in O 2 . elimination. Interestingly, it was found that the cellular calcium level did not respond to SOD-ScFv in 1 h, whereas cellular calcium levels increased rapidly and distinctly under stimulation with 50 M H 2 O 2 (data not shown). It is well known that H 2 O 2 can produce a high cellular oxidative stress.
The main source of ROS in most cells is the mitochondria (28). We confirmed this in SPC-A-1 cells according to the perfect superposition of cellular DHE fluorescence and Rh123 fluorescence, which localized to the area of origin of O 2 . and the mitochondrial area, respectively (Fig.   5B). One parameter that is altered because of ROS production is ⌬ m . Therefore, the effect of a lower O 2 . level on ⌬ m was determined using Rh123. Fig. 6 shows    Based on the fact that SOD-ScFv can eliminate O 2 . in SPC-A-1 cells, we determined the cellular redox state aroused by SOD-ScFv after 1 h by detecting cellular ROS and GSH levels. As Fig. 7 shows, it is notable that SOD-ScFv resulted in significant decreases in cellular ROS fluorescence intensity (0.54-fold, p Ͻ 0.05) and increases in GSH fluorescence intensity (5.4-fold, p Ͻ 0.05), implying that cellular oxidative stress can be regulated by O 2 . levels in SPC-A-1 cells. Increases in GSH with decreased oxidative stress have been demonstrated in previous studies (31)(32).
To understand the mechanism of how SOD-ScFv retarded the cell cycle, we examined the effects of reduced levels of intracellular ROS on some signal-transduction pathways. As shown in Fig. 8, it was found that SOD-ScFv dose dependently inhibited Akt phosphorylation (Thr-308) in a short time (1 h); in the subsequent 24 h, the expression of cyclin-dependent kinase inhibitors p27 kip1 (a downstream protein of Akt) was up-regulated in SPC-A-1 cells. This outcome indicated that the Akt/p27 kip1 pathway was activated under the reduced level of intracellular ROS in SPC-A-1 cells. However, the mechanism of Akt/p27 kip1 pathway activation by ROS remains unclear.

DISCUSSION
We predicted the tertiary structures of N. commune Fe-SOD by ExPASy and found that the four Fe 3ϩ -bound sites in the SOD active center, His-28, His-80, and His-166 and Asp-162 were close (a water molecule or a hydroxyl ion is the fifth bound site of Fe 3ϩ ) (33). Also, the C terminus amino acid residues of SOD were farther from the enzymatic active center than the N terminus; thus, the ScFv was added to the C terminus of SOD with link-1 to weaken the spatial interference between SOD and ScFv. The dual activity of SOD-ScFv indicated that SOD and ScFv were well separated by link-1.
High intracellular ROS levels and low SOD activity were found in many tumor cells. Based on these findings in M14 melanoma cells,  Pervaiz et al. (34) suggested that an intracellular environment with high ROS concentrations provided tumor cells with a survival advantage over their normal counterparts. It was suggested that ROS acts as an essential intracellular secondary messenger in the regulation of protein kinase activity and relative gene expression and in modulating cell growth, apoptosis, differentiation, and transformation (35)(36). Shi et al. (37) transfected the Mn-SOD gene into HepG2 liver carcinoma cells, eliminating intracellular ROS to inhibit cell proliferation, and established the Akt signaling pathway. It was reported that Akt would dephosphorylate at Thr-308 at a reduced level of intracellular ROS (37,38). However, phosphorylation at Thr-308 was necessary for activation of Akt and the subsequent regulation of many biological responses, such as p27 kip1 (39). In SPC-A-1 cells, we discovered that cellular SOD induced Akt dephosphorylation on Thr-308 (inactive form), in correlation with decreased oxidative stress. Furthermore, SOD-induced Akt inactivation subsequently up-regulated p27 kip1 transcription, retarding the cell cycle at G 1 . The cell cycle delay phenomenon from SOD in tumor cells was observed recently, but the mechanism was still unclear (40). Here for the first time, the Akt/p27 kip1 pathway induced by reduced oxidative stress was established. From our SPC-A-1 cancer cell model, we speculate that the intracellular environment with high SOD concentration is the rea-son for up-regulation of p27 kip1 , and the Akt dephosphorylation on Thr-308 is the hinge on which this relationship depends. Usually, down-regulation of p27 kip1 and phosphorylation of Akt are associated with stimulation of growth factors like IGF-I in cancers (41). However, the IGF-I expression level was unvaried in the SOD-ScFvinduced dephosphorylation of Akt in our research (data not shown). Lang et al. (42) found that various free radical scavengers could not change the level of IGF-binding protein-I (IGFBP-I) in HepG2 cells, neither could it attenuate the stimulation-induced increase in IGFBP-I. IGFBP-I was known as the IGF-I-binding protein that could inhibit anabolic actions of IGF-I. It seemed that the IGF-I pathway was not involved in the response of reduced cellular oxidative stress in some cancer cells.
Usually, the growth rate of a cancer cell population can be inhibited by either altering the rate of cell proliferation or the rate of cell death. In this study, we found, using two different fluorescent dyes, that SPC-A-1 cells pretreated with SOD-ScFv increased ⌬ m . This finding implied that the decreased growth rate caused by SOD-ScFv did not arise from cell apoptosis, which ubiquitously existed under high oxidative stress. It is well known that cellular high oxidative stress will lead to loss of ⌬ m , influx of extracellular Ca 2ϩ , activation of the mitochondrial permeability transition (which can be activated by cellular Ca 2ϩ ), and release of cytochrome c to cytoplasm, followed by apoptosis (43)(44)(45). Interestingly, the cellular Ca 2ϩ level, which responded to oxidative stress, was unvaried in this research (46). Also, it seems that the signal pathways under reduced oxidative stress are different from those under high oxidative stress in SPC-A-1 cells; we failed to detect the distinct difference in the expression level of Bcl-2/Bax-2, or in the distribution of cytochrome c (data not shown) (47,48). Gel electrophoresis analysis for DNA fragmentation excluded the possibility of apoptosis as a cause of the decreased growth rate of SOD-loaded cells (data not shown). It was consistent with Mn-SOD contributing to the overall redox environment of the cell (49).
After being treated with SOD-ScFv-FITC, SPC-A-1 cells fluoresced in 20 min. Taking this result together with the Western blotting results, we concluded that SOD-ScFv possessed the ability to permeate the cell membrane. Based on the results presented in Fig. 5B, we suggest that the O 2 . originated from mitochondria and localizes around its origin because of its short half-life. Luckily, the fact that our SOD-ScFv distributed equally in the cytoplasm provided the chance for SOD to reach the area of the mitochondria (Fig. 3) and execute its role in 24 h. Although SOD-ScFv was degraded gradually by an unknown mechanism, it can persist in cells up to 24 h, implying that some SOD-ScFv could escape from the endocytic pathway. It has been reported that retinoic acid After treatment with SOD-ScFv for 1 h, cells were disrupted, and cellular Thr-308-phosphorylated Akt (p-Akt) was analyzed by Western blotting. B, dose response of p27 kip1 by SOD-ScFv in SPC-A-1 cells. After treatment with SOD-ScFv for 1 h, the cells were washed and changed with a fresh culture medium for 24 h and then cells were disrupted, and the cellular p27 kip1 level was analyzed by Western blotting.
could change the Golgi reticulum to disturb the endocytic pathway (50).
Our future work will focus on the task of improving the antitumor effect of SOD-ScFv by exploring drug synergies (e.g. with retinoic acid). By the dismutation of escaped SOD-ScFv, the O 2 . in cells was greatly reduced.
As a strong oxidizing agent, O 2 . can react with compounds capable of donating H ϩ , then produce other types of ROS, such OH Ϫ , ONOO Ϫ (51). On the other hand, the reduced O 2 . certainly will lead to reduced oxidative stress, with a decrease in ROS levels and an increase in GSH levels. In this intracellular environment, the phosphorylation of Akt and expression of p27 kip1 varied and influenced progression through the cell cycle; thus, proliferation was inhibited (while AGS cells and HepG2 cells did not behave as SPC-A-1 cells in proliferation after SOD-ScFv treatment). Based on the above-described results and the mechanism of LC-1 ScFv acting on its antigen (14), we suggest an SOD tumor-targeting hypothesis. The ScFv will specifically guide SOD to the SPC-A-1 cell surface and bind its antigens. Then the antigen-antibody complexes will be internalized via the receptor-mediated endocytic pathway and diffuse into cytoplasm, eliminating the O 2 . around the mitochondria, leading to mitochondrial hyperpolarization. Last, reduced intracellular oxidative stress inhibits the growth of SPC-A-1 cells though a ROS-involved Akt/p27 kip1 pathway. Because ROS plays complicated roles in modulation of chemical reactions in the cell (7,52), many clinical trials with antioxidants fail to kill pathologic cells and in many cases even reveal an acceleration of the death of normal cells. The CHAOS study (54,55) and the ␤-carotene cancer prevention study (53) exemplified such failures. Hence, antioxidant therapy should be designed carefully, taking into consideration drug dose and its effect on normal cells. Thus, what course to use for antioxidants in clinical applications? Perhaps our ScFv tumor-targeting approach will provide the key.