INDOLE-3-CARBINOL (I3C) INHIBITS CYCLIN DEPENDENT KINASE-2 FUNCTION IN HUMAN BREAST CANCER CELLS BY REGULATING THE SIZE DISTRIBUTION, ASSOCIATED CYCLIN E FORMS AND SUBCELLULAR LOCALIZATION OF THE CDK2 PROTEIN COMPLEX

of specific

Indole-3-carbinol (I3C), a dietary compound found in cruciferous vegetables, induces a robust inhibition of CDK2 specific kinase activity as part of a G1 cell cycle arrest of human breast cancer cells. Treatment with I3C causes a significant shift in the size distribution of the CDK2 protein complex from an enzymatically active 90 kDa complex to a larger 200 kDa complex with significantly reduced kinase activity. Coimmunoprecipitations revealed an increased association of both a 50 kDa cyclin E and a 75 kDa cyclin E immunoreactive protein with the CDK2 protein complex under I3C-treated conditions, whereas the 90 kDa CDK2 protein complexes detected in proliferating control cells contain the lower molecular weight forms of cyclin E. I3C treatment caused no change in the level of CDK2 inhibitors (p21, p27) or in the inhibitory phosphorylation states of CDK2. The effects of I3C are specific for this indole and not a consequence of the cell cycle arrest because treatment of MCF-7 breast cancer cells with either the I3C dimerization product DIM or the anti-estrogen tamoxifen induced a G1 cell cycle arrest with no changes in the associated cyclin E or subcellular localization of the CDK2 protein complex. Taken together, our results have uncovered a unique effect of I3C on cell cycle control in which the inhibition of CDK2 kinase activity is accompanied by selective alterations in cyclin E composition, size distribution and subcellular localization of the CDK2 protein complex.
Considerable epidemiological evidence show that diets high in vegetable and fiber lead to low cancer risks and confer protection from various forms of cancers, including breast cancer (1,2). In particular, consumption of vegetables belonging to the Brassica genus, which includes broccoli, cabbage and cauliflower, have been reported to correlate with a decrease in mammary tumor incidence (3).
These epidemiological studies suggest the existence of naturally occurring compounds in dietary sources that represent a largely untapped source of potential chemotherapeutic molecules.
One such phytochemical is indole-3-carbinol (I3C) 1 , an autolysis product of gluccosinolates present in Brassica vegetables. Dietary exposure to I3C reduced tumor occurrence and decreased the multiplicity of spontaneous as well as carcinogeninduced mammary tumor formation in rodent model systems (4)(5)(6)(7)(8). Furthermore, diets high in cabbage, a good source of I3C, reduced pulmonary metastasis of mammary tumor implants in mice (9). Consistent with these studies, I3C tested positive as a chemopreventative agent in several short-term bioassays relevant to carcinogeninduced DNA damage, tumor initiation and promotion, and oxidative stress (10).
Regulated changes in the expression and/or activity of G1 cell cycle components control the growth of normal mammary epithelial cells, whereas the loss of normal cell cycle control in the G1 phase has been implicated in mammary tumor development and proliferation. Key targets of these pathways are specific sets of cyclin/cyclin dependent kinase (CDK) protein complexes that function at specific stages of the cell cycle (34,35). For example, many breast tumors show an aberrant expression and/or amplification of Cyclin D1 or Cyclin E, (36,37), which both interact with G1 phase CDKs. In fact, cyclin E expression has been shown to be the best diagnostic marker for breast cancer to date (38). CDK protein kinase activity is tightly regulated during passage through the cell cycle by the timely appearance and degradation of cyclin-CDK protein complexes, CDK subunit phosphorylation, and interaction with a variety of CDK inhibitors (39). In early G1, activation of CDK4/cyclin D and CDK6/cyclin D protein complexes initiate the phosphorylation of the retinoblastoma tumor suppressor protein (Rb), partially releasing the E2F family of transcription factors to permit the synthesis of cyclin E in mid to late G1. CDK2 kinase activity increases as a result of its interaction with its regulatory partner, cyclin E, which directly hyperphosphorylates Rb, thereby further releasing the E2F transcription factors that in turn initiate the transcription of S phase genes. Although CDK2 and cyclin E have been shown to be dispensable for murine development (40,41), these genes play a critical role in the transformation process of human breast cancer cells. For example, Cyclin E deficient cells have been shown to be resistant to oncogenic transformations (40) and overwhelming genetic, cellular, biochemical and clinical evidence correlate aberrant cyclin E expression with tumorigenesis and poor patient prognosis, particularly in breast cancer (42)(43)(44)(45)(46). Thus, cyclin E/CDK2 protein complexes represents a critical potential target for anti-proliferative molecules, such as the natural indoles.
We have demonstrated that I3C directly induces a G1 cell cycle arrest of human breast cancer cell lines through a pathway that is independent of estrogen receptor signaling and which targets specific G1-acting cell cycle components (21). Our previous studies have shown that I3C down-regulates CDK6 transcription by disrupting the functional interactions of the Sp1 transcription factor with an Sp1-Ets composite DNA element in the CDK6 promoter (19). I3C also causes a pronounced decrease in CDK2 specific enzymatic activity and inhibited phosphorylation of endogenous Rb proteins (20), although the precise mechanism underlying this process has not been characterized. The activity, accessibility and cellular utilization of CDK2/cyclin E can be regulated at multiple levels, any one of which could be potentially targeted by I3C treatment. For example, in addition to cyclin association, the activation of CDK2 also requires the dephosphorylation of Thr14 and Tyr15 by Cdc25A (47) and an activating phosphorylation event at Thr160 by CDK activating kinase (CAK) (48). Another mode of regulating CDK2 kinase activity is through the association with cyclin-dependent kinase inhibitors (CKI) such as p21, p27 and p57. Recent studies have also shown that the subcellular compartmentalization of either cyclin E or CDK2 greatly alters its enzymatic activity and accessibility to nuclear residing CAK, Cdc25A and known substrates of CDK2 (49)(50)(51). In this study we demonstrate for the first time that I3C selectively controls the size distribution of the CDK2/cyclin E protein complex with concomitant alterations in cyclin E interactions and subcellular localization that results in an inhibition of CDK2 enzymatic activity in human breast cancer cells. Flow Cytometry Analysis of DNA Content -MCF-7 cells were plated at 30% confluency on 100mm tissue culture plates and treated for the indicated time points with 100 uM I3C in complete media. An aliquot of harvested cells were hypotonically lysed in 0.5-1ml of DNA staining solution (0.5 mg/ml propidium iodide, 0.1% sodium citrate, 0.05% Triton X-100). Cell debris was filtered and DNA content was analyzed as described previously (21).

Materials -
Tissue Culture and Cell Lines -MCF-7 human breast cancer cell lines were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum, 2mM L-glutamine, 10 ug/ml insulin, 1.25 ml of 20,000 units/ml penicillin and streptomycin. Cells were maintained at subconfluency at 37 degrees Celsius in humidified air containing 5% CO2. I3C, DIM, tamoxifen and tryptophol were dissolved in DMSO (99.9% high performance liquid chromatography grade; Aldrich) at concentrations 1000-fold higher than the final concentrations used.
Proteins were electronically transferred to nitrocellulose membranes (Micron Separations, Inc., Westboro, MA) and blocked as previously described (21). Blots were subsequently incubated at room temperature for an hour with anti-CDK2 (sc 748), CDK4 (sc260), and cyclin E1 (sc 198) antibodies. Blots probed with anti-p-Tyr (sc 7020), anti-p21(sc 398) or anti-p27 (sc 528) antibodies were either incubated at room temperature for 2-3 hours or incubated overnight. All antibodies were used at a concentration of 1 ug/ml in TBST with the exception of anti-pTyr antibody, which was made in 1% Nonfat Dry Milk/TBST. All antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA).

Immunoprecipitation and CDK2 Kinase
Assays -After the indicated treatments, cells were lysed for 15 minutes at 4 degrees Celsius in CDK2 lysis buffer with protease and phosphatase inhibitors (50ug/ml PMSF, 10 ug/ml aprotinin, 5ug/ml leupeptin, 0.1% sodium fluoride, 10 ug/ml beta-glycerophosphate, and 0.1 mM sodium orthovanadate). Samples (500-800 ug of protein) were pre-cleared as described previously (21) followed by incubation with 0.5 ug of anti-CDK2 or anti-Cyclin E1antibodies (Santa Cruz Biotechnology, Santa Cruz, CA) for 2 hours. Thirty microliters of a 1:1 bead slurry was added to each sample and left on a rocking platform for 30 minutes. The beads were washed with CDK2 lysis buffer and twice with kinase buffer (50 mM Hepes (pH 7.3), 5 mM MnCl 2 , 10 mM MgCl 2 ) with protease and phosphatase inhibitors. Half of the immunoprecipitation was checked by Western blot analysis to confirm the efficiency of the immunoprecipitation. The remaining half of the immunoprecipitation was used to assess for kinase activity as described above.

I3C Directly Added to CDK2 IP/Kinase
Assay -Immunoprecipitation of CDK2 was carried out as described above followed by a preincubation of the CDK2 immunoprecipitated protein complex with the indicated concentrations of I3C for 15 minutes at 30 degrees Celsius. Half of the CDK2 immunoprecipitate was resolved on SDS-PAGE and Western blot analyzed for CDK2 while the remaining half was used to assess CDK2 kinase activity as described above. In a separate set of CDK2 immunoprecipitation, I3C was added concurrently with the [gamma-32 P]-ATP and Histone H1 followed by a 15-minute incubation at 30 degrees Celsius and assayed for activity.33 Purified GST-Cdc25A was bacterially synthesized, sonicated in extraction buffer (1X PBS, 5 mM DTT, 10 mM Tris/HCl, pH 8), and the lysates were clarified by centrifugation. GST-Cdc25A fusion proteins were then absorbed onto glutathione-Sepharose beads, washed 3 times and eluted with 20 mM reduced glutathione. Cells were treated with either I3C or DMSO for 48 hours followed by a CDK2 immunoprecipitation as described above.
Half of the CDK2 immunoprecipitation was used for Western blot analysis while the remaining half was used for the Cdc25A kinase assay; CDK2 immune complexes were either incubated with 2ug of purified GST-Cdc25A or GST-alone for 15 minutes at 30 degrees Celsius in phosphate buffer (50 mM Tris, pH 8, 150 mM NaCl, 2 mM DTT, 2.5 mM EDTA). Half of the CDK2 immunoprecipitate was analyzed for CDK2 protein levels and the remaining half was used to assess CDK2 kinase activity as described above.
Cells were either treated with 100 uM I3C, 30 uM DIM, 1 uM tamoxifen, 100 uM tryptophol, or DMSO (control) for 48hours. Each sample was lysed in CDK2 lysis buffer with phosphatase and protease inhibitors and cleared of cellular debris with by centrifugation at 15,000 X g. Cell lysates (800 ug) were then applied immediately onto a Kontes size exclusion column packed with Superose 12 beads and a total of 80 fractions were collected. Forty microliter of column lysate was taken for immunoblot analysis of CDK2 and 400ul were precipitated (15 ul of 1 ug/ml BSA, 1ml of acetone, 400 ul of column lysate, mixed and incubated at -70 for 3 hours or overnight, followed by a 10 minute spin at 15,000 x g at 4 degrees Celsius), resuspended in 20 ul of 2X protein loading buffer used for immunoblotting for cyclin E and p21.
For the column lysate immunoprecipitation/kinase assays, 2 consecutive gel filtration fractionations were conducted for each of the treatment groups. Column lysates were pooled, followed by the CDK2 immunoprecipitation/kinase assay described above.
Indirect Immunoflourescence -MCF-7 cells were plated on 8-well Lab-Tek Permanox slides (Nage Nunc International, Naperville, IL) at 20% confluency and treated for 48-72 hours with either DMSO, I3C, DIM, tamoxifen or tryptophol (1 ul/ml of 1000X stock). Cells were washed with phosphate-buffered saline (PBS), fixed with 3.7% formaldehyde/.01 % glutaraldehyde for 15 minutes, and permeabilized with cold 50/50 acetone for one minute. Cells were blocked for 5 minutes with PBS containing 4% goat serum (Jackson ImmunoResearch, West Grove, PA) followed by incubation with rabbit anti-CDK2 or mouse anti-cyclin E antibodies at a 1:150 dilution for 1.5 hours (CDK2) at 25 degrees Celsius or overnight for cyclin E at 4 degrees Celsius. After 5 washes with cold PBS, cells were blocked and incubated with anti-rabbit rhodamine-texas red conjugated secondary antibody (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA; diluted 1:300 in PBS) or anti-mouse FITCconjugated secondary antibody (1:300) for 30 minutes. Cells were washed, mounted with clear nail polish and visualized on a Nikon Optiphot fluorescence microscope. Images were captured using Adobe Photoshop and a Sony DKC-5000 digital camera.
Nonspecific fluorescence, as determined by incubation with secondary antibody alone was negligible. degrees Celsius (1000 rpm). Cell pellets were resuspended at 2x volume in buffer A and dounced 100 strokes with type B pestle. Cells were checked with trypan blue to ensure at least 80% of cells were lysed, followed by a 20-minute spin at 550 rpm. Supernatants were transferred to a new tube followed by 100,000 rpm centrifugation for 30 minutes in a ultracentrifuge to clear cytosolic extract of cellular debris. Nuclear pellets were resuspended in 1.5x cell volume with buffer B (20 mM HEPES, pH 7.4, 25% glycerol, 1.5 mM MgCl 2 , 0.4M NaCl, 0.2 mM EDTA. 0.5 mM DTT, 0.5 mM PMSF) and dounced 20 strokes with type B pestle. Extracts were nutated at 4 degrees Celsius for 30 minutes followed by a 17,500 rpm centrifugation. The supernatants were dialysized against 1000 x volume of cold buffer C (.02M HEPES, pH 7.9, 5% glycerol, 1.5 mM MgCl 2 , 0.1M KCl, 0.2 mM EDTA, 0.5 mM DTT, 0.5 mM PMSF), followed by a 17,500-rpm centrifugation. Extracts were normalized and analyzed by Western blot.
Quantification of Autoradiography -Autoradiographic exposures were scanned with a UMAX UC630 scanner, and band intensities were quantified using the NIH Image program. Autoradiographs from a minimum of three independent experiments were scanned per time point.

Time Course of I3C Inhibition of CDK2
Kinase Activity and G1 Cell Cycle Arrest -To determine the kinetics of the inhibition of CDK2 kinase activity by I3C, estrogen responsive human MCF-7 breast cancer cells were treated with or without 100 uM I3C over a 96-hour time course. This concentration of I3C was previously shown to be the optimal dose for the inhibition of cell growth without affecting cell viability (21). CDK2 protein complexes were immunoprecipitated from total cell lysates by using a CDK2 specific antibody, and half of each sample was assayed in vitro for kinase activity and the other half analyzed for CDK2 protein by Western blot analysis. CDK2 protein kinase activity was monitored by the ability of the immunoprecipitated CDK2/cyclin protein complexes to phosphorylate the COOH-terminal domain of Rb fused to GST. As shown in Figure  1A, electrophoretic analysis of the phosphorylated GST-Rb showed that the I3C suppression of CDK2 kinase activity can be observed in cells treated as early as 24 hours post-treatment, with near maximal inhibition observed by 72 hours in I3C. A nonspecific IgG was used in one set of immunoprecipitations (no IP lane) as a negative control to demonstrate the specificity of the CDK2 kinase assay. A similar inhibition of CDK2 kinase activity was observed in estrogen receptor negative breast cancer cells (data not shown). During the sample time course, the level of G1 phase MCF-7 cells was examined by flow cytometry of propidium iodide stained nuclei. As shown in Fig 1B, I3C treatment induces a significant increase in the number of G1 phase arrested cells compared to untreated cells, which correlates closely to the kinetics of I3C inhibition of CDK2 enzymatic activity.
Given the chemical nature of I3C, it is conceivable that this indole could act as a direct enzymatic inhibitor of CDK2, and thereby prevents the CAK phosphorylation of CDK2. To examine this possibility, a range of I3C concentrations was added to individual CDK2 immunoprecipitates either prior to or simultaneous with the addition of the histone H1 substrate and [gamma-32P]ATP. As shown in Fig. 1C, I3C has no significant effect on the in vitro CDK2 kinase activity. These results demonstrate that I3C does not inhibit CDK2 kinase activity through direct interactions with the CDK2 protein immunocomplex.
Western blot analyses of CDK2 immunoprecipitates show that I3C treatment did not alter the total level of CDK2 protein ( Figure  1A), indicating that I3C inhibits total cellular CDK2 specific enzymatic activity. CDK2 protein migrates as a doublet, and the faster migrating CDK2 band represents the CAK phosphorylated Thr-160 form of CDK2, which is necessary for enzymatic activity (48). The Western blots also revealed that in MCF-7 breast cancer cells, the CAK phosphorylated form of CDK2 is modestly reduced in samples immunoprecipitated from the 72 and 96 hour I3C-treated cells, but not at early time points. These results suggest that the events leading to the inhibition of CDK2 kinase activity by I3C precedes the decrease in the CAK phosphorylation of CDK2 protein.

I3C Does Not Affect the Inhibitory
Phosphorylation of CDK2 or Levels of Known CDK2 Inhibitors that Associate with the CDK2 Protein Complex -To address whether I3C had an effect on the inhibitory phosphorylation events, CDK2 immune complexes from control or I3Ctreated cells were first treated with either recombinant GST-alone or GST-Cdc25A and the kinase activity of the CDK2 immunocomplexes assayed in vitro using histone H1 as a substrate. As shown in Fig. 2A, exposure to active Cdc25A caused the same approximate 2-fold increase in CDK2 specific enzymatic activity in the immune complexes isolated from control and I3C-treated cells. Importantly, after Cdc25A treatment, the enzymatic activity of the CDK2 complexes immunoprecipitated from I3C-treated cells remained significantly reduced compared to the control samples. Thus, the I3C inactivation of the CDK2 enzymatic activity was most likely not caused by an increase in the negative phosphorylation events on residues Thr14/Tyr15 on CDK2. This was confirmed by Western blot analysis of Tyr-phosphorylated forms of CDK2 in a CDK2 complex immunoprecipitated from I3Ctreated or untreated cells. As shown in Fig 2B, I3C had no effect on the presence of the Tyr14 phosphorylated forms of CDK2.
To address other canonical mechanisms of regulating CDK2 kinase activity, the presence of known CDK inhibitors associated with the CDK2 protein complex in I3C-treated or untreated cells were analyzed by co-immunoprecipitation. MCF-7 human breast cancer cells were treated with either DMSO or I3C for 48 hours followed by a CDK2 immunoprecipitation and Western blot analysis of p21 and p27. As shown in figure 2B, the level of p21 and p27 associated with CDK2 did not change with I3C treatment despite drastic reductions in kinase activity as measured by the phosphorylation of Histone H1 (fig. 2B, lower panel). No detectable p57 was observed in the CDK2 immunoprecipitates from these cells (data not shown). The cyclin E associated with the CDK2 protein complex is characterized later in this study. We conclude that the I3C-induced inhibition of CDK2 kinase activity is not due to an increase in association with known CDK2 inhibitors or changes in inhibitory phosphorylation states of CDK2.
I3C Alters the Size Distribution of the CDK2 Protein Complex -As a first approach to determine if I3C alters the composition of the CDK2 protein complex, the sizes of the CDK2 protein complexes produced in breast cancer cells treated with or without I3C were biochemically characterized. Equal amounts of total cell lysates (800 ug) were fractionated on a Sepharose size exclusion gel filtration column. Individual gel filtration fractions were electrophoretically separated by SDS-PAGE, followed by Western blot analysis of CDK2, cyclin E and p21. As shown in figure 3A (upper panel), in proliferating MCF7 breast cancer cells, the CDK2 protein complex eluted with a native size of approximately 90 kDa. Strikingly, in I3C growth arrested cells, CDK2 eluted as part of a much larger 180-200 kDa protein complex (Fig. 3A,  lower panel). Analysis of the fractionation profile of cyclin E and p21 ( Fig. 3B and 3C), showed a similar result.

CDK2 Kinase Activity is Associated
Primarily with the 90 kDa CDK2 Protein Complex -To determine the CDK2 enzymatic activity in each of the CDK2 protein complexes, individual fractions from the gel filtration columns were immunoprecipitated with CDK2 antibodies, and the kinase activity of the immune complexes were monitored in vitro. As shown in figure 4A, most of CDK2 protein kinase activity is associated with the 90 kDa protein complex, as assayed by the in vitro phosphorylation of Histone H1. Despite an abundant amount of CDK2 protein in the 200 kDa protein complex formed in I3C-treated cells, the majority of the kinase activity was observed in fractions that correspond to the 90 kDa CDK2 protein complex (Fig. 4B). Consistent with the overall change in the fractionation of the CDK2 protein complex, I3C-treated cells produce a significantly reduced level of the enzymatically active 90 kDa CDK2 protein complex compared to the same fractions from control cells.
These results demonstrate that I3C induces the formation of a larger (200 kDa) CDK2 complex that is enzymatically inactive. Interestingly, cyclin E by guest on March 22, 2020 http://www.jbc.org/ Downloaded from association with CDK2 in the column lysates changed significantly with I3C treatment in that the 50 kDa cyclin E form is more abundant in the inactive, 200 kDa CDk2 protein complex observed with I3C treatment as compared to control. Furthermore, a 75 kDa cyclin E immunoreactive protein co-precipitated with CDK2 in fractions that correspond to the 200 kDa protein complex, suggesting that it can partially account for the 100 kDa shift in size of the CDK2 protein complex.

I3C Alters the Cyclin E Form that
Associate with the Larger CDK2 Protein Complex -Different forms of cyclin E can associate with CDK2 and confer different levels of CDK2 enzymatic activity.
Several lower molecular weight cyclin E forms are found in breast cancer tissues and are associated with higher CDK2 enzymatic activity (52). In contrast, the larger form of cyclin E (50 kDa) is found in both normal and tumorogenic tissues, and are associated with CDK2 protein complexes that display a reduced enzymatic activity as compared to the smaller forms of cyclin E (52). Co-immunoprecipitation was used to examine the nature of the cyclin E forms that associate with CDK2 protein complexes in growing and I3C growth arrested breast cancer cells. As shown in figure 5, the 35 kDa lower molecular weight form of cyclin E associated with the active CDK2 protein complex in untreated cells is drastically reduced after I3C treatment. Correspondingly, there is a significant increase in the level of both a 50 kDa form and an unusual 75 kDa cyclin E immunoreactive protein that is associated with the inactive CDK2 protein complex detected in I3C-treated cells. The precise function of this 75 kDa protein is not known, although because it is specifically recognized by cyclin E antibodies and binds to CDK2, we propose that it likely functions as a regulator of CDK2. In whole cell extracts, I3C has no effect on the total cellular level of this 75 kDa cyclin E immunoreactive protein (data not shown), suggesting that I3C stimulates the association of the CDK2 protein complex with this 75 kDa cyclin E immunoreactive protein. We propose that the increase in CDK2 association with the 50 kDa cyclin E and the 75 kDa cyclin E immunoreactive protein likely accounts for the approximate 100 kDa increase in size of the CDK2 protein complex observed in I3C-treated cells (figure 4). CDK2 kinase activity was assessed within the CDK2 immunoprecipitates from control or I3C-treated cells using histone H1 as the substrate. As shown in the bottom panel of figure 5A, the CDK2 associated kinase activity was significantly reduced in samples isolated from I3C-treated cells, which contained elevated levels of the 50 cyclin E and 75 kDa cyclin E immunoreactive protein associated with the CDK2 protein complex.
The anti-tumorigenic effects of I3C in vivo are partly due to the formation of 3,3diindolyelmethane (DIM), which is a natural dimerization product of I3C in acidic conditions (12). DIM has been shown to inhibit MCF-7 breast cancer cell growth and down-regulate CDK2 kinase activity through the transcriptional induction of the p21 CDK inhibitor (53). Furthermore, due to the fact that a fraction of I3C is converted into DIM in breast cancer cells (54), it is critical to evaluate whether the observed changes in the 50 kDa cyclin E and 75 kDa cyclin E immunoreactive protein that associate with CDK2 protein complexes are specific for the I3C signaling pathway, or a general effect of growth inhibition by indoles. Under conditions in which I3C and DIM induce a parallel G1 cell cycle arrest as monitored by their similar flow cytometry profiles (Fig. 5B), DIM did not alter the association of the different cyclin E forms with CDK2 as compared to control (Fig. 5A, DMSO vs DIM) despite a drastic reduction in CDK2 kinase activity. These results suggest that the association of the higher molecular weight forms of cyclin E with the CDK2 protein complex is an I3C-specific indole response.

The Formation of the 200 kDa CDK2 Protein Complex is an I3C-Specific Response -
To test whether the change in size distribution of the CDK2 protein complex is specific to the I3C response, breast cancer cells were treated for 48 hours with several agents known to induce a G1 cell cycle arrest; 100 uM I3C, 30 uM DIM, or 1 uM of the estrogen antagonist, tamoxifen. Control cells were either treated with 100 uM tryptophol, an indole with a structure similar to I3C but does not exhibit any anti-proliferative properties (19), or with the DMSO vehicle control. Total cell extracts were fractionated by gel filtration chromatography and individual fractions were by guest on March 22, 2020 http://www.jbc.org/ Downloaded from Western blot analyzed for either CDK2 or p21 protein. As shown in figure 6, under conditions in which I3C treatment caused a 100 kDa shift in size of the CDK2 protein complex, tamoxifen, DIM or tryptophol did not significantly alter the size distribution of CDK2.
Additionally, close inspection of the results revealed that the CDK2 protein complex is slightly larger in DIM and tamoxifen treated cells compared to the vehicle control cells (Fig. 6A and 6B). Treatment with DIM or tamoxifen induced a significant increase in the level of CDK2 protein complex-associated p21 (Fig. 6B), which likely accounts for this small increase in the overall size of the CDK2 protein complex. This result is consistent with previous studies showing that tamoxifen and DIM treatments increase the level of p21 associated with the CDK2 complex, thus rendering it inactive (55,56).
To determine the distribution of CDK2 enzymatic activity, CDK2 immunoprecipitated protein complex from column fractions were assayed in vitro using histone H1 as a substrate. As shown in figure 6C, CDK2 kinase activity is significantly reduced in I3C, DIM and tamoxifen treated cells compared to control cells (DMSO or tryptophol). The majority of CDK2 kinase activity fractionated as a 90 kDa CDK2 protein complex. The reduced level of CDK2 enzymatic activity after I3C treatment is likely due to the formation of the larger 200 kDa inactive complex, whereas, in DIM or tamoxifen treated cells, the reduction in CDK2 kinase activity correlated with the increase in p21 observed in fractions that correspond to the distribution of CDK2. These results also show that I3C acts through a pathway that is distinct from either DIM or tamoxifen in regulating CDK2 enzymatic activity.

I3C Alters the Subcellular Localization of CDK2
and Cyclin -Subcellular compartmentalization of CDKs and their regulatory cyclins play a critical role in controlling their activity and cell cycle progression (57). Several studies have shown that the localization of CDK2 is highly dependent on cyclin association (51,58,59).
We therefore tested whether I3C treatment had an effect on cyclin E and CDK2 localization using indirect immunofluorescence microscopy. Cells were treated with either DMSO or 100 uM I3C for 48 hours, permeablized and incubated with either CDK2 or cyclin E primary antibodies, followed by the appropriate secondary antibodies. Typical immunofluorescence staining is shown in figure 7A where CDK2 and cyclin E are localized to the nucleus in control cells and is redistributed to the cytoplasm with I3C treatment. Individual cells in each population were counted, and the number of cells containing nuclear staining of cyclin E or CDK2 versus the total number of cells present were assessed and statistically analyzed using the student's T-test. Under proliferative conditions (DMSO-treated), approximately 80% of the cells contained nuclear staining for CDK2. In contrast, I3C treatment caused a significant increase in the level CDK2 and cyclin E that were localized to the cytoplasm. An average of three independent experiments showed that there was a forty percent reduction in the level of nuclear CDK2 and cyclin E in I3Ctreated human breast cancer cells as compared to control cells (Fig. 7B).
In conjunction with the immunoflourescence studies, the subcellular fractionation of CDK2 and cyclin E was examined by Western blot analysis of cytoplasmic and nuclear fractions from I3C-treated and untreated cells. As shown in figure 7C, I3C treatment significantly increased the level of CDK2 and the CDK2-associated, 50 kDa cyclin E and the 75 kDa cyclin E immunoreactive protein in the cytoplasmic fractions with a corresponding decrease in their protein levels in the nuclear fractions as compared to control cells.
To further establish the I3C specificity of the redistribution of CDK2 and cyclin E localization, indirect immunofluorescence was used to examine the subcellular distribution of both proteins in cells treated with I3C, DIM, tamoxifen, tryptophol or the DMSO vehicle control. As shown in figure 8A, under conditions where I3C increases the level of cytoplasmic CDK2 and cyclin E subcellular localization, DIM or tamoxifen have no effect on the subcellular localization of CDK2 or cyclin E compared to the controls sets of cells (tryptophol or vehicle control treated cells). Figure 8B shows a 40% reduction in the number of cells containing nuclear localization of both cyclin E and CDK2 with I3C by guest on March 22, 2020 http://www.jbc.org/ Downloaded from treatment as compared to control cells, and no change was observed in the localization profile of cyclin E or CDK2 among cells that were treated with tamoxifen or DIM. As expected, tryptophol treated cells were indistinguishable from the vehicle control treated cells.
Taken together, these results demonstrate that that the change in subcellular localization and the increase in the size of the CDK2 protein complex are specific to the I3C growth inhibitory response.

DISCUSSION
I3C is a promising chemotherapeutic agent for treatment of human breast cancer, and other reproductive cancers, because of its potent growth inhibitory properties (18)(19)(20)(21)(22)(23)(60)(61)(62), and repression of invasion and migration of cultured breast cancer cell lines (24). Elucidation of the molecular mechanisms and cell cycle targets of the I3C anti-proliferative signaling pathway provides critical information that could potentially be exploited in the development of novel indolebased therapeutic strategies. We previously documented that I3C induces a G1 cell cycle arrest of human breast cancer cells that is accompanied by the dual inhibition CDK6 and CDK2 kinase activities (19). I3C and its natural dimerization product DIM (12) have been shown to inhibit CDK2 enzymatic activity, although the distinction in the mechanisms underlying the effects of these two indoles was unknown. We now establish a new and unique feature of the I3C signaling pathway by establishing that I3C induces a robust down-regulation of CDK2 kinase activity that is accompanied by selective alterations in the cyclin E composition, size distribution and subcellular localization of the CDK2 protein complex. Furthermore, we show that this mechanism of regulating CDK2 function is specific to the I3C anti-proliferative signaling pathway that is distinguishable from the actions of DIM and the anti-estrogen tamoxifen.
As summarized in figure 9, in growing breast cancer cells, a 90 kDa CDK2 protein complex that is enzymatically active resides in the nucleus, and includes both the p21 CDK inhibitor and the 35 kDa lower molecular weight form of cyclin E. After I3C treatment, a 200 kDa CDK2 protein complex is formed that is enzymatically inactive and is localized mostly to the cytoplasm. This larger CDK2 protein complex was shown to contain elevated levels of the 50 kDa form of cyclin E and a unique 75 kDa cyclin E immunoreactive protein, but not the lower molecular weight cyclin E form observed in proliferating cells. The striking increase in size of the CDK2 protein complex detected in I3C-treated cells has never been observed in other cell systems.
Treatment of MCF-7 human breast cancer cells with either DIM or the anti-estrogen tamoxifen induces a G1 cell cycle arrest and inhibits CDK2 enzymatic activity (20,53). In contrast to the effects of I3C, DIM or tamoxifen did not alter the interaction of the lower molecular weight forms of cyclin E with the CDK2 protein complex or regulate the nuclear localization of the CDK2 protein complex (summarized in Fig. 9). Thus, the formation of the larger CDK2 protein complex observed in I3C-treated cells is not an indirect consequence of the cells undergoing a G1 cell cycle arrest.
A key problem in using tamoxifen in the clinical management of breast cancer is that a majority of the cancers that do respond to this anti-estrogen eventually acquire resistance to prolong tamoxifen treatment (63). We propose that the previously reported synergistic inhibition of CDK2 kinase activity by combinatorial treatments with I3C and tamoxifen (20) is due to ability of these two anti-cancer agents to disrupt CDK2 function through different cellular pathways.
Because lower doses of tamoxifen could potentially be used in the presence of I3C, conceivably this indole could potentially be used to overcome the tamoxifen resistance or prolong the duration in which tamoxifen is effective in suppressing the growth of breast cancer cells. Similarly, of I3C and DIM combinatorial regiments could also be developed for treating breast cancer.
A novel feature of our results is that I3C selectively increases the size of the CDK2 protein complex, from 90 kDa to 200 kDa, that we propose is due to the association of the higher molecular weight 50 kDa cyclin E and 75 kDa cyclin E immunoreactive protein that associate with CDK2. In growing cells not treated with I3C, by guest on March 22, 2020 http://www.jbc.org/ Downloaded from the 90 kDa CDK2 protein complex contains the 35 kDa lower molecular weight form of cyclin E. The discovery that the lower molecular weight cyclin E forms reside in the nucleus and associate with enzymatically active CDK2 compared to the 50 kDa cyclin E form suggests a specific role of cyclin E in regulating CDK2 function. In this regard, tumorigenic tissues from aggressive breast tumors express elevated levels of the lower molecular weight forms of cyclin E as compared to normal breast epithelium (52,64,65). Based on this newly appreciated level of regulating CDK2 kinase activity by cyclin E, we propose that the inhibition CDK2 kinase activity by I3C is due to the replacement of the lower-molecular weight forms of cyclin E with the 50 kDa form and the 75 kDa cyclin E immunoreactive protein (summarized in Fig. 9). Consistent with this concept, enzymatic activity of the 200 kDa CDK2 protein complex formed after I3C treatment is significantly reduced compared to the high enzymatic activity of the 90 kDa CDK2 protein complex detected in breast cancer cells not treated with I3C. Studies are underway to functionally characterize the precise roles of the 50 kDa cyclin E and 75 kDa cyclin E immunoreactive protein in controlling CDK2 kinase activity and mediating the G1 cell cycle arrest.
Our results also eliminated several established modes of regulating CDK2 kinase activity by I3C. The direct addition of I3C to CDK2 immunoprecipitates demonstrated that I3C does not act as a small molecule enzymatic inhibitor of CDK2 kinase activity.
We also eliminated the possibility that I3C could be changing the phosphorylation states of CDK2 or the level of known CDK inhibitors (p21 and p27) that associate with the CDK2 protein complex. Although the inhibitory phosphorylation events are unaffected by I3C treatment, the level of CAKphosphorylated CDK2 is modestly reduced with prolonged I3C treatment. This effect could be due to the decreased association of the 35 kDa cyclin E, which has been shown to induce CAK phosphorylation of CDK2 (66). Because the down-regulation of CDK2 kinase activity by I3C preceded the reduction in the level of CAK phosphorylation of CDK2, it is unlikely that the primary mechanism by which I3C regulates CDK2 kinase activity occurs through CAK.
The redistribution of the CDK2 protein complex to the cytoplasm in I3C-treated cells represents an effective mode of growth regulation because it prevents access to a variety of substrates and regulatory proteins.
For example, key regulators of CDK2 (Cdc25A and CAK), as well as its key substrates (Rb, NPAT, BRCA1, and Cdc6), which play key roles in initiating S-phase transition, reside in the nucleus. Nucleocytoplasmic shuttling of CDK and cyclin E has been shown to be dynamically regulated, and the steady state accumulation of nuclear CDK2/cyclin E is due to a faster rate of nuclear import as compared to the rate of nuclear export (67). This process can be regulated by several extracellular signals in other systems.
For example, TGF-beta and vitamin D induce the cytoplasmic localization of CDK2 leading to a decrease in Rb phosphorylation and a G1 cell cycle arrest (51,68). I3C treatment increases the steady state cytoplasmic localization of both CDK2 and cyclin E, and it is therefore conceivable that I3C could either reduce the rate of nuclear import of CDK2/cyclin E protein complexes or increase the rate of nuclear export. Recent studies show that CDK2 localization is controlled by its association with cyclin E (58,59). Therefore it is tempting to consider that association of the higher molecular weight forms of cyclin E with CDK2, in particular the 75 kDa cyclin immunoreactive protein, under I3C-treated conditions, could sequester the CDK2/cyclin E protein complex in the cytoplasm. For example, replacement of the lower molecular weight cyclin E form with the higher molecular weight forms of cyclin E could potentially mask a critical nuclear localization signal or activate a nuclear export signal. Further studies are needed to determine the precise mechanistic relationship between the increase in size distribution of the CDK2 protein complex, and the I3C control of nucleocytoplasmic shuttling of the CDK2 protein complex with the overall G1 cell cycle arrest.  CDK2 protein complex was isolated by immunonprecipitation from I3C treated and untreated cells. Each immunoprecipitate was incubated with either 2ug of purified GST-Cdc25A or GST-alone followed by an in vitro kinase assay (upper panel). Half of each immunoprecipitate were analyzed for CDK2 protein levels (lower panel). The bar graph represents the quantification of the level of CDK2 kinase specific activity observed in the presence or absence of Cdc25A. (B) CDK2 immune complexes from I3C treated and untreated cells were Western blot analyzed for the indicated protein: CDK2, phosphotyrosine (p-Tyr) phosphorylated CDK2, p21, and p27. The other half of the immunoprecipitate was assayed for kinase activity (lower gel).   4. CDK2 kinase activity is associated with the 90 kDa CDK2 protein complex. MCF-7 cells were treated with (B) or without (A) 100 uM I3C and total cell lysates were fractionated by a size exclusion column as described in Materials and Methods. CDK2 was immunoprecipitated from column lysates, and assayed for CDK2 protein kinase activity (top gels in each set) or for the presence of CDK2 or cyclin E protein by Western blot analysis (bottom gels in each set).  5. Effects of I3C on the associated cyclin E forms with the CDK2 protein complex. (A) CDK2 was immunoprecipitated from MCF-7 cells treated for 72 hours with either DMSO, 100 uM I3C or 30 uM DIM. Each set of immunopreciptated CDK2 protein complex was either fractionated by SDS-PAGE and Western blot analysis for cyclin E (top three panel), or was assayed for CDK2 kinase activity using histone H1 as a substrate (32P-H1). (B) Flow cytometry analysis of cells that were treated with either DMSO, I3C or DIM for 72 hours as described in the Materials and Methods. Fig. 6. Formation of the 200 kDa CDK2 protein complex is specific to the I3C response. Cells were treated with 100 uM I3C, 30 uM DIM, 100 uM tryptophol (Tryp), or 1 uM tamoxifen (Tam), or were incubated with the DMSO vehicle control for 48 hours. Cell lysates were fractionated from gel filtration chromatography as described in the Materials and Methods section, and column fractions analyzed for the presence of either CDK2 (A) or p21 (B) by Western blot analysis. A third set of immunoprecipitated CDK2 protein complex from the fractionated cell lysates was assayed for CDK2 kinase activity as described in Fig. 4, and the level of phosphorylated histone H1 quanified by phosphoimaging (C).

Fig. 7. Effects of I3C on the Subcellular Localization of CDK2 and Cyclin E.
(A) MCF-7 cells were treated with or without 100 uM I3C for 48 hours followed by indirect immunoflourescence using CDK2 or cyclin E specific antibodies. (B) The percentage of cells containing nuclear residing CDK2 protein (top graph) or cyclin E protein (bottom graph) was quantified by evaluating approximately 1000 random cells (C) I3C treated and untreated cells were biochemically fractionated into nuclear and cytoplasmic fractions by differential centrifugation. CDK2 was isolated via an immunoprecipitation and resolved by SDS-PAGE and Western blot analyzed for CDK2, and cyclin E.   9. Summary of the differential regulation of CDK2 protein complex by I3C, DIM, and tamoxifen in breast cancer cells. In proliferating cells, CDK2 protein complexes are approximately 90 kDa in size, contain higher levels of the 35 kDa cyclin E, and reside predominantly in the nucleus in an enzymatically active state. I3C treatment induces a G1 cell cycle arrest and causes the formation of a significantly larger 200 kDa CDK2 protein complex that contains a 50 kDa cyclin E and a 75 kDa cyclin E immunoreactive protein. This I3C induced 200 kDa CDK2 protein complex is enzymatically inactive, localizes to the cytoplasm, and displays no apparent change in the level of associated p21. Treatment with DIM or tamoxifen induces a G1 cell cycle arrest of MCF-7 breast cancer cells, however, in contrast to I3C, neither agent has an affect on the associated forms of cyclin E with the CDK2 protein complex or its subcellular localization. Treatment with DIM or tamoxifen causes the formation of an inactive 110 kDa CDK2 protein complex that is due to the increased association of CDK inhibitors (p21) with CDK2