The nuclear factor kappa B (NF-κB): A potential therapeutic target for estrogen receptor negative breast cancers

Inhibition of Tumor Growth by Go6976.

The inhibitory influence of Go6976 on tumor growth was studied by administering the drug within 48 h of implantation of CSMLO cells when no palpable tumors were detected. Periodical measurement of tumor volume (32) revealed a substantial inhibition of tumor growth in Go6976-treated mice in comparison to the untreated controls (DMSO-treated group, five animals in each group, Fig. 1). These results demonstrate that Go6976 inhibited the growth of CSMLO-induced tumors when administered within a short period after implantation of the cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Inhibition of growth of tumors by Go6976. Tumors were generated in female A-J mice by implantation of CSMLO cells (10⁶) on day 0 (indicated by the arrow). Go6976 (0.5 ml of 0.05 mM solution in 0.1% DMSO) was administered 48 h later to a group of five animals, twice a week locally at the site of implantation of cells (♦). Control group (five animals) received 0.5 ml of 0.1% DMSO similarly (□). Growth of tumors was monitored by measurements of tumor volume at regular intervals (32).

Regression of Tumor Growth by Go6976.

Go6976 not only inhibited tumor growth, but also regression of the full-grown tumors. Tumors in all of the five animals treated with Go6976 21 days after implantation of CSMLO cells regressed sharply (Fig. 2 A and B, animal 5). The second group of tumor-bearing mice (five animals) received nothing (Fig. 2 A and B, animal 3) and the third group (five animals) received the solvent DMSO (Fig. 2 A and B, animal 4). The tumor volumes in the second and the third group of animals continued to increase (Fig. 2 A and B, animals 3 and 4). Mice without tumor treated either with DMSO or with Go6976 (Fig. 2B, animals 1 and 2) did not show any apparent physical defects as judged by body weight and agility, and these parameters were not different from the animals without any treatments (Fig. 2, animal 6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Regression of tumors by Go6976. (A) CSMLO cells were implanted on day 0 (indicated by the arrow) as for Fig. 1. Go6976 was administered on day 21 (indicated by the arrow), locally under the tumors of the five animals with comparatively larger tumors (■). Another five animals received the same volume of 0.1% DMSO (♦) on the same day, and the control group of five animals received nothing (□). (B) One representative of five animals from each group. Animals 1 and 2 are DMSO- and Go6976-treated non-tumor-bearing animals, respectively. Animal 3 is tumor-bearing and without treatment. Animals 4 and 5 represent tumor-bearing groups treated with DMSO and Go6976, respectively. Animal 6 is an A-J mouse that received nothing.

Inhibition of PMA-Induced Activation of NF-κB by Go6976 in CSMLO Cells.

The previously reported NF-κB inhibitory effect of Go6976 in human breast cancer (7) and Jurkat T lymphocytic (33) cell lines is confirmed also in CSMLO cells. CSMLO cells are ER− as determined by Western blot analysis. A low basal level activity of NF-κB was detected in nuclear extracts from CSMLO cells as determined by EMSA (Fig. 3A, lane 1), which was stimulated by PMA (20 ng/ml; lane 2). Activation could be detected at 2–4 h of PMA treatment, was maximal at 6 h, and remained the same for 24 h (data not shown). By simultaneous treatment with Go6976 for 48 h, the PMA-induced elevation of NF-κB activation was inhibited in a concentration-dependent manner (lanes 3–7). The growth of cells under culture conditions was not affected by this short-term drug treatment with 10 μM Go6976 (7). The retarded radioactive DNA–protein complex was supershifted with either anti-p50 (Fig. 3B, lanes 3 and 4) or anti-p65 (lanes 5 and 6) antibody, but not with anti-cRel antibody (data not shown), thus characterizing the active NF-κB in CSMLO cells as a p50/p65 heterodimer.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Active NF-κB complex in CSMLO cells: Stimulation by PMA and inhibition by Go6976. (A) Nuclear extracts (5 μg protein) from control and PMA-treated (20 ng/ml for 18 h) CSMLO cells were incubated in a standard EMSA reaction mixture containing [γ-³²P]-labeled double-stranded oligonucleotide plus Go6976 at the indicated μM concentrations for 48 h and subjected to nondenaturing PAGE (7, 31). The autoradiographic signals of the retarded NF-κB–[³²P]DNA complex is indicated by the upper arrow and the free [γ-³²P]-labeled probe by the lower arrow. (B) The NF-κB–[³²P]DNA complex was characterized by supershift assay with anti-p50 (lanes 3 and 4) or p65 (B, lanes 5 and 6) antibodies. Nuclear extracts were incubated with specific antibodies for 15 min at room temperature, followed by incubation for an additional 30 min in the presence of [γ-³²P] double-stranded NF-κB oligonucleotide, and subjected to EMSA as described (7, 31). The supershifted complexes are indicated by the upper arrow.

Fig. 4 shows histological changes in tumor, liver, and lung tissues from untreated (Left) and Go6976-treated animals (Right). Tumor tissues from untreated and DMSO-treated (data not shown) animals demonstrated characteristic tumor cell phenotypes, such as enlarged nuclei and increased mitotic index (indicated by arrowheads). In contrast, in Go6976-treated animals extensive necrosis (indicated by N) of the tumor tissues was detected with few residual tumor cells (indicated by T). Similar histological examination of liver and lung tissue (Fig. 4) sections from tumor bearing animals (Fig. 2B, animals 3, 4, and 5) or non-tumor-bearing animals (data not shown) did not show any microscopically detectable pathological changes. Because CSMLO cells are characterized as noninvasive, these observations were expected (11, 12). Examination of residual tumor tissues in treated animal at higher magnification (100×) revealed a large number of pycnotic cells with nuclear clumping suggestive of DNA fragmentation and apoptosis (indicated by stars in Right), whereas this was not observed in tumor tissues from untreated animals (Left). These results demonstrated that localized Go6976 treatment selectively killed the target tumor cells at a concentration not toxic to the vital organs, liver, and lung. These results (Figs. 1 and 2) and prior data (7) suggest that Go6976 is a selective killer of ER− mouse mammary epithelial tumor cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Histology of tissues from untreated and Go6976-treated tumor-bearing animals. Tumor growth and treatment conditions are the same as described for Fig. 1. Tumor, liver, and lung tissues from untreated and treated tumor bearing animals were dissected 11 days after the initiation of treatment and 32 days after implantation of the cells. Tissues were processed for hematoxylin/eosin (H&E) staining, examined under a light microscope, and photographed at the indicated magnifications. Arrows show mitotic cells in untreated tumor tissue. Residual tumor (T) and necrotic cells (N) of the treated tumor are shown. Stars in the treated block indicate pycnotic cells with apparent fragmentation and clumping of nuclear DNA. Treated and untreated liver and lung tissues did not show significant microscopically detectable damages and were not different from liver and lung tissues of normal mice without tumors (not shown).

Selective Inhibition of NF-κB Activation by Stable Expression of Dominant-Negative Ikk-β (dnIKKβ) Mutants.

Our second experimental approach for examining the role of NF-κB in the CSMLO cell-induced tumorigenesis was to selectively block activation of the NF-κB–IκB complex and thereby retain it in its inactive state in the cytoplasm. This was accomplished by the stable expression of a dominant-negative mutant of one of the subunits of Ikk. Because Ikk-β is more potent in activating NF-κB than is Ikk-α, we focused our initial studies on dnIkkβ-expressing transfectants and compared them with parental CSMLO cells and transfectants expressing the vector control plasmid. The dnIkkβ-expressing transfectants grew comparatively slower than the parent CSMLO cells.

Immunofluorescent Detection of Flag-Tagged dnIkkβ in Transfectants.

The dnIkkβ transfectants were characterized by identifying the Flag protein tagged to it by an immunofluorescence technique using Flag-specific antibody. Fig. 5A demonstrates the immunofluorescence signal of one dnIkkβ transfectant (Upper, iii) that represents the exogenously introduced Flag-tagged dominant-negative mutant dnIkkβ protein in the cytoplasm. A G418 resistant transfectant with empty vector plasmid did not show this fluorescence (Lower, vi), suggesting that the positive signal observed in this transfectant is specific for Flag-tagged dnIkkβ protein.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Inhibition of NF-κB activation and down stream events by stable expression of dominant-negative IκB-kinase β (dnIkkβ). (A Upper) dnIkkβ-Expressing transfectant cells by light microscopy (i), 4′,6-diamidino-2-phenylindole (DAPI)-stained nuclei (ii), and immunofluorescence with anti-FLAG antibody in the presence of the secondary antibody (iii). The positive signals show dnIkkβ-conjugated FLAG protein in the cytoplasm. (Lower) Processed vector-control plasmid-transfected CSMLO cells in which no FLAG protein could be detected (vi). (B Upper) Active NF-κB was determined by its [γ-³²P]DNA binding activity by EMSA, in three dnIkkβ-expressing stable transfectants (dnIkkβ1-1, dnIkkβ1-3, and dnIkk-β1-5) and (Lower) in parent CSMLO cells and vector control plasmid expressing transfectant (vect1-5). (C Upper) The level of ccD1 in the same three-dnIkkβ-expressing CSMLO and vector control transfected cells, as measured by Western blot analysis. (Lower) Actin analyzed similarly by immunodetection with anti-β-actin antibody that serves as a loading control.

The Level of Active NF-κB in dnIkkβ-Expressing Transfectants.

The functional state of NF-κB in the dnIkkβ-expressing transfectants was established by measuring its DNA-binding activity with EMSA. The basal level of active NF-κB was not further stimulated by PMA in three different clones of dnIkkβ-expressing transfectants as judged by its unaltered [³²P]DNA binding activity (Fig. 5B Upper). In contrast, the basal level of active NF-κB was strongly elevated by PMA in parent CSMLO cells and in transfectants with vector control plasmid vect-1–5 (Fig. 5B Lower). These results demonstrated that PMA activation above the basal level of NF-κB is blocked by externally introduced dnIkkβ, thereby establishing the experimental goal of selective inhibition of NF-κB activation by specifically targeting the Ikk with dnIkkβ.

The Level of ccD1 in dnIkkβ-Expressing Transfectants.

Up-regulation of the ccD1 is a downstream consequence of NF-κB activation (7, 34). Similar to the results obtained in human breast cancer cells (7), Go6976 also blocked transactivation of ccD1 in CSMLO cells transfected with dnIkkβ. The elevated level of ccD1 following PMA treatment measures activation of the NF-κB–IκB complex. Fig. 5C demonstrates that ccD1 was not altered by PMA treatment of the three clones of dnIkkβ-expressing transfectants (5C, lanes 1–6) in which NF-κB activation was effectively blocked (5B, Upper), whereas ccD1 level was substantially elevated by PMA in parental and transfectants carrying the control plasmid vect-1-5 (5C, lanes 7–12). These levels of ccD1 in CSMLO cells and transfectants correlates well with activation of NF-κB–IκB complex (Fig. 5B). The basal level of ccD1 in the dnIkkβ-expressing transfectants was lower than that in the parent and vector-control-expressing cells. This may be a reflection of the short half-life of ccD1. Thus, ability to up-regulate downstream ccD1 gene expression following PMA treatment in the parent CSMLO and in transfectant expressing vector control plasmid, and the inability to do the same in dnIkkβ-expressing transfectants can be correlated to the level of activation of inactive NF-κB–IκB complex in these cells.

Tumorigenic Potential of CSMLO Cell-Expressing dnIKKβ Mutants.

The consequence of blocked NF-κB activation in dnIkkβ-expressing transfectants on tumor growth was examined in the mouse tumor model. Fig. 6 shows volumes of tumors induced by CSMLO cells (five animals), two transfectants with vector control plasmid (six animals, three per clone), and two transfectants expressing dnIkkβ (six animals, three per clone). The CSMLO cells and transfectants carrying the vector control plasmid generated tumors similarly. One dnIkkβ-expressing clone (dnIkkβ1-1) did not form tumor in any of the three animals even after 32 days. The other dnIkkβ-expressing clone (dnIkkβ1-3) formed tumors at a substantially reduced rate and of smaller sizes compared with those formed by the CSMLO or the vector control plasmid expressing transfectants (Fig. 6). The loss or decreased tumorigenic potential of the dnIkkβ-expressing transfectants, compared with parent and the vector controls was in agreement with the reduced level of NF-κB activation and cellular level of ccD1. All of these results, especially those with dnIkkβ expression that selectively blocked NF-κβ–IκB activation and ccD1 transactivation, strongly suggest a role of NF-κB in tumorigenesis in ER− mammary epithelial cells.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Loss of tumorigenic potential of CSMLO cells by the stable expression of pdnIkkβ in A-J mice. Tumors were generated in female A-J mice by implanting either CSMLO cells (five animals), or two vector control plasmid transfected clones (vect1-3 and vect1-5, three animals per clone) or two pdnIkkβ-expressing clones (pdnIkkβ1-1 and pdnIkkβ-1-3, three animals per clone). Average with standard deviations of CSMLO-administered animals (♦), vector-control-administered animals (■, six animals), and one of the two pdnIkkβ-expressing-clones (pdnIkkβ1-3)-administered animals (▵, three animals) are plotted. The other pdnIkkβ1-1 clone did not form any tumor even 32 days after implantation of the cells in any one of the three animals (ρ).