Development of Herceptin resistance in breast cancer cells

Abstract

Background

Herceptin, a humanized antibody to HER-2, is now utilized in the clinic for metastatic breast cancer treatment. The response rate for HER-2+ patients is only 30% and little is known as to mechanisms of resistance. The mechanism of Herceptin action is also unknown but has been related to cell cycle inhibition.

Methods

The effects of Herceptin and other antibody treatments were determined by cell counting and cell cycle analysis. HER-2 and p27 expression levels were analyzed by flow cytometry and levels of activated AKT were compared by Western blot analysis. Cellular HER-2 and p27 expression was measured by immunofluorescence.

Results

Herceptin treatment of BT-474 cells results in inhibition of cell growth and arrest in the G₁ phase. The efficacy of growth arrest was not directly correlated to the binding affinity of antibodies to Her-2. Our laboratory has developed cell lines that are resistant to Herceptin treatment. In resistant cell lines, binding of antibodies is not hindered. However, Herceptin has completely lost the ability to inhibit cell proliferation. Yet, the mouse isotype 4D5 maintains significant inhibitory activity upon Herceptin-resistant clones.

Conclusions

Herceptin binds effectively to Her-2 on the cell surface of Herceptin-resistant cell lines and the level of Her-2 expression on the cell surface is not downregulated. Herceptin resistance is not due to downregulation of levels of AKT protein expression, although, phosphorylation of AKT is enhanced in resistant lines and could have a role in resistance. Resistance appears to correlate with the loss of nuclear expression of the cyclin-dependent kinase inhibitor, p27, as defined by immunofluorescence and flow cytometry studies and cdk-2 binding studies. © 2004 Wiley-Liss, Inc.

The detection of unique overexpression of human epidermal growth factor receptor 2 (HER-2) in some breast cancer tumors has led to the development of a specific treatment that is tumor selective, effective at extending life expectancy, and nontoxic with taxol treatment in the patient with metastatic breast cancer (1). Genentech (South San Francisco, CA) has developed a humanized monoclonal antibody to HER-2, Herceptin, which is now approved for the treatment of patients with metastatic breast cancer (2). The patient's tumor must demonstrate an amplified copy number for the HER-2 oncogene and/or overexpress the HER-2 oncoprotein in order for the patient to be a candidate for Herceptin treatment. Clinical research protocols have also demonstrated a synergistic response with chemotherapy when Herceptin is added to various regimens (2). As a single agent, Herceptin has limited effectiveness (3) and it is very expensive. It is also known to have specific side effects on the heart when used in concert with Adriamycin (4). Patients that show a response to Herceptin relapse after a short period of time, despite continued treatment. It would be interesting to know how tumors develop resistance to Herceptin, so that these alterations might be reversed or prevented. Knowing sites of resistance might also provide novel strategies for alternative treatments.

From a biochemical point of view, Herceptin is known to have a high affinity for binding to HER-2 (i.e., Kd = 0.5 nM) and can reach concentrations of over 100 nM in serum when given to patients with breast cancer (5). Since this is a humanized monoclonal antibody, the half-life is around two weeks. Given that a patient's tumor cells must demonstrate high expression of the HER-2 protein to be a candidate for Herceptin treatment, it is only logical that the Herceptin–HER-2 complex will be formed on the surface of the tumor cells. It is surprising that while overexpressing cell lines are sensitive to Herceptin, greater than 70% of patients whose tumors also overexpress HER-2 do not respond to Herceptin treatment (3). Understanding mechanisms of Herceptin resistance in vitro may help elucidate mechanisms of both acquired Herceptin resistance and innate Herceptin resistance.

In this article, we demonstrate Herceptin resistance developed in cell lines previously sensitive to Herceptin (6). In addition, we address the mechanism of drug resistance given the currently defined pathways for Herceptin action.

MATERIALS AND METHODS

Cell Lines and HER-2 Antibodies

The human breast cancer cell line, BT-474, was obtained from the American Type Tissue Culture Collection (Manassas, VA) and maintained in RPMI 1640 (Carlsbad, CA) with penicillin/streptomycin, 5% L-glutamine, and 10% FBS. Cells were maintained at 37°C with 5% carbon dioxide. Herceptin was obtained from the Wake Forest University Hospital pharmacy, and two other antibodies to HER-2 (4D5 and 7C2) were a gift from Genentech.

HER-2 Expression

The expression of HER-2 was determined by flow cytometry using indirect fluorescent staining methods. The cells were harvested by trypsin and EDTA, and then washed with PBS containing 1% BSA (PBS-A). In the primary reaction, the cells were treated at 4°C for 30 minutes with either no antibody, or 10 μg/ml of the various antibodies to HER-2. After washing the cells with PBS-A to remove the nonbound primary antibody, the cells were treated with a 1:50 dilution of either goat anti-human IgG-PE labeled or goat anti-mouse IgG-FITC (Jackson Immuno Research Laboratory, West Grove, PA), labeled depending on the primary antibody used. After washing with PBS-A, the cells were brought up in PBS-A and flow cytometry was performed using a BD-FACStar^Plus instrument (Palo Alto, CA). The cell population was gated on forward scatter and side scatter and the relative fluorescence was determined by excitation at 488 nm, and 520 nm or 540 nm bandpass filters were used to measure either FITC or PE fluorescence, respectively. The percent of cells with specific staining and the intensity of staining were determined by the Becton Dickinson Cell Quest program (Palo Alto, CA). In competition studies, mixtures of primary mouse antibodies and/or human antibodies to HER-2 were used in the primary reaction in order to determine relative displacement affinities for these antibodies (see details for these studies). In the studies for HER-2 expression on cells as a function of Herceptin treatment, Ab-5 (Oncogene Research Products, Cambridge, MA) was used to measure HER-2 expression levels and it does not compete with the epitope of Herceptin.

Measurement of Cell Cycle

BT-474 cells were harvested after the various treatments and then resuspended in DNA staining solution. This solution contained 50 μg/ml of PI, 0.6% NP-40, and 37 μg/ml of RNAase in a 3.6 mM citrate buffer (pH 7.4). At least 20,000 cells were analyzed for DNA content per cell, after correction for debris and doublets. The cell cycle analysis from the DNA histogram was determined by Modfit (Modfit Verity Software House, Topsham, ME), as described previously (7).

Western Blot Analysis for AKT/phospho-AKT

Western blot analysis was performed using methods previously described (8). Using a lysis buffer (0.5% NP40, 150 mM NaCl, 20 mM Tris-HCl, 5 mM EGTA, 200 μM Na₃VO₄, 1 mM PMSF, and 2 μg/ml of leupeptin and aprotinin), proteins were extracted. After loading approximately 50 μg of protein per lane, the amount of proteins was analyzed using SDS-PAGE after transferring to nitrocellulose. The proteins (AKT and phospho-AKT [pAKT]) were detected using a 1:1000 dilution of either mouse anti-human AKT (#9272) or mouse anti-human pAKT (#9275), obtained from Cell Signaling Technology (Beverly, MA), and followed by goat anti-mouse IgG labeled with horseradish peroxidase (Jackson Immuno Research Laboratory, West Grove, PA). The peroxidase activity was then assayed using chemiluminescence, and quantified using the program NIH Image (Bethesda, MD).

Immunofluorescence Assays for p27 Expression

Cells were harvested and plated into 35-mm cell culture dishes. They were then treated with either nothing or 10 μg/ml of Herceptin for various periods of time. At the end of the treatment, the media was removed and the cells were washed with PBS. The cells were then fixed with 3.7% PBS-formalin for 10 min and then they were washed with PBS. The cells were treated with 0.1% saponin and 1% BSA for 15 min to make the cells permeable. The cells were then treated with either no antibody or a 1:250 dilution of mouse anti-human p27 (Transduction Laboratories Lexington, KY) for 1 h. After a wash step, the cells were treated with affinity-purified rhodamine-conjugated goat anti-mouse IgG (Jackson Immuno Research Laboratory, West Grove, PA) for 45 min. The cells were then washed again and analyzed for p27 expression and localization using a Zeiss Axioplan2 fluorescence microscope with an Axiocam camera (Thornburg, NY).

Nuclear p27 Analysis

The harvested cells were washed and treated with 1 ml of cold nuclear isolation buffer (0.32 M sucrose, 5mM MgCl₂, 10 mM Hepes, 1% triton x-100, pH 7.4) for 10 min. The resulting nuclei were then labeled for p27 expression as described above, except the nuclear isolation buffer was used instead of the 0.1% saponin and 1% BSA and the cells were kept at 4°C during the whole time of labeling.

Immunoprecipitation and Immunoblot Analysis for p27

BT-474 cells and the respective resistant clones were treated with and without 10 μg/ml of Herceptin for three days. The cells were harvested and lysed in NP40 lysis buffer containing 0.5% NP40, 150 mM NaCl, 20 mM Tris-HCl (pH 8.0), 5 mM EGTA, 1 mM PMSF, and 10 μg/ml of leupeptin and aprotinin. The lysed extract (2 mg of protein) was then treated with 5 μg of rabbit anti-human cdk-2 (sc-163, Santa Cruz, CA) overnight and then immunoprecipitated with 20 μl of protein A agarose (Oncogene Research Products, Boston, MA) for 2 h. After washing four times with PBS, the precipitate was dissolved in SDS-PAGE buffer and analyzed by gel electrophoresis. The expression of p27 was determined by using mouse anti-human p27 (Transduction Laboratories Lexington, KY) and followed by goat anti-mouse IgG labeled with horse radish peroxidase (Jackson Immuno Research Laboratory, West Grove, PA). The peroxidase activity was then assayed using chemiluminescence and quantified using the program NIH Image. Analysis of p27 by gel electrophoresis in the lysate was also performed to determine the expression of the p27 levels prior to immunoprecipitation.

RESULTS

Our previous work (9) and that of others (8, 10, 11) have indicated that cells that overexpress HER-2 are growth-inhibited by antibodies directed against this oncoprotein. In order to understand how cells become resistant to these antibodies, BT-474 cells were treated for two weeks with 10 μg/ml of Herceptin. After two weeks, almost all of the cells were destroyed except for a few colonies. Six colonies were cloned, mechanically separated, and replaced in media containing 10 μg/ml of Herceptin. These cells were then defined as resistant, since they grew in the presence of Herceptin. Two of the cell lines have been continued in this media with 10 μg/ml of Herceptin for over 1.5 years, except for the reversal studies. The remaining groups have been frozen down for future studies after initial evaluations. In all cases, the resistant populations demonstrate no detrimental effect due to long-term exposure to Herceptin, based on cell count, cell cycle analysis, or apoptotic activity.

The growth of wild-type BT-474 cells was strongly inhibited by Herceptin, but the two resistant cell lines grew at the same rate as the wild type (Figure 1, top panel). This growth inhibition was at the G1/S interphase as defined by DNA flow cytometry analysis. Under the same conditions of 10 μg/ml of Herceptin for four days of treatment, the two resistant clones demonstrated no G1/S blockage (Figure 1, bottom panel). The amount in S phase was 33.4%, 5.7%, 36.2%, and 37.3% for parental BT-474 not treated, parental BT-474 treated, resistant clone 5 treated, and resistant clone 6 treated, respectively. These data have been reproduced several times over the course of the development of the resistance (data not shown).

Figure 1. Herceptin's effect on cell growth in BT-474 cells and in two BT-474 Herceptin-resistant clones. Top panel: Growth curves were performed for BT-474 or BT-474 Clone 5 and Clone 6 treated with 10 μg/ml of Herceptin (H). The resistant clones were developed from Herceptin treatment and these growth curves were performed in the presence of 10 μg/ml of Herceptin. Each bar going from left to right represents a timepoint in hours on the growth curves: at 0, 21, 48, 76, 172, 194, and 217 h. Bottom panel: Cell-cycle analyses were performed on BT-474, BT-474+Herceptin, and two BT-474 Herceptin-resistant clones treated with Herceptin. The cells were treated for four days with and without 10 μg/ml of Herceptin and then harvested and analyzed for cell cycle analysis using flow cytometry. The S% activity for BT-474 alone, (A), BT-474+Herceptin (B), Clone 5+Herceptin (C), and Clone 6+Herceptin (D) were 33.4%, 5.7%, 36.2%, and 37.3%, respectively.

The inhibitory effect of different antibodies to HER-2 was next tested on parental BT-474 cells using cell cycle inhibition as the end point. Cells were treated with Herceptin, 7C2, a mouse monoclonal antibody to a different epitope, and 4D5, the mouse monoclonal antibody from which Herceptin was derived. At 10 μg/ml for 24 or 48 h of treatment, the potency of inhibition was 4D5 > Herceptin > 7C2, with mean inhibitions of 77%, 22%, and 0%, respectively. If 7C2 was added to Herceptin, the mean inhibition was increased to 62% compared to Herceptin alone. If 4D5 was added with Herceptin, the mean inhibition was decreased to 36% compared to the 4D5 alone. These data would suggest that there are complex interactions of these antibodies with the HER-2 oncoprotein, which results in perturbation of cell cycle inhibition. In attempting to better understand these interactions, antibody competition assays were performed as defined in the methods. The results indicate that 7C2 does not compete with either Herceptin or 4D5 binding to HER-2. There is a competition between Herceptin and 4D5, but 4D5 has a higher relative affinity constant than Herceptin. This can be observed by the fact that it takes higher concentrations of Herceptin to displace 4D5 compared to 4D5 replacing Herceptin.

In Table 1, a dose–response curve for 4D5 inhibitory activity for three days was performed on both the wild-type and resistant clones. These data would suggest that at 0.1 μg/ml, the resistant clones are less sensitive to 4D5 than the wild type. At concentrations of 1 μg/ml and 10 μg/ml, both parental BT-474 cells and resistant clones are growth-inhibited by 4D5. However, previous work (Fig. 1) has shown that 10 μg/ml of Herceptin has growth inhibition effects only on parental BT-474 cells and not on the resistant clones. Since Herceptin and 4D5 might be interacting with the HER-2 in a slightly different manner, it is possible that resistance to Herceptin would not predispose the cells to be completely resistant to 4D5 at the higher concentrations.

Table 1. Cross Reactivity of 4D5 in the Parental and Herceptin Resistant Clones as Defined by S Phase Inhibition

4D5 Treatmenta	S phase activity (% inhibition)
	BT-474	Clone no. 5	Clone no. 6
a Treatment was for three days.
Control	31.3% (0%)	24.0% (0%)	27.1% (0%)
0.1 ug/ml	18.9% (40%)	20.1% (16%)	25.4% (6%)
1 ug/ml	4.6% (85%)	10.7 (55%)	16.2% (40%)
10 ug/ml	5.3% (83%)	7.9% (67%)	7.9% (61%)

The mechanism of HER-2 overexpression resulting in the increased growth of tumors has been studied for many years. Most investigators would propose that the first step is dimerization of HER-2 with itself or with other members of the HER family. The second step is signaling through the PI3-kinase/AKT pathway to effect the localization of the cyclin-dependent kinase inhibitor, P27. AKT phosphorylation of p27 is known to inhibit the translocation of p27 into the nucleus, and therefore prevent biological inhibition of proliferation by p27. Treatment of HER-2–positive cells with Herceptin has been shown to increase both phosphorylation and nuclear exclusion of p27 (8).

A first step in attempting to understand the resistance to Herceptin was to determine the concentration of HER-2 on the cell surface in parental cells and in the resistant populations. Our approach for determining the site of Herceptin resistance is observing different signaling alterations when Herceptin is added to both the wild-type cells and the resistant clones. In Figure 2, it is clear that the addition of Herceptin to BT-474 cells does not affect the overall protein expression levels of HER-2 in either the parent cell line or in four of the six resistant clones that were developed. This was determined by flow cytometry analysis of HER-2 expression using a primary antibody that did not cross react with the Herceptin epitope. In these studies, the mean fluorescence intensities were all similar. This was confirmed by immunofluorescence analysis demonstrating an intense membrane staining in all of the cells. Western blot analysis also did not show any significant change in HER-2 expression in either the parental cells or in clones 5 and 6 when 10 μg/ml of Herceptin was added (data not shown).

Figure 2. Measurement of HER-2 expression in Herceptin-sensitive and Herceptin-resistant BT-474 cells. A: The typical forward scatter side scatter dot plot profile for all of the cells; used to gate out debris. BT-474 cells were grown in no Herceptin (dotted line) or in 10 μg/ml of Herceptin (solid line) for 24 h (B) and the resistant cells were grown in continuous 10 μg/ml of Herceptin (C–F). The cells were harvested after 24 h and the expression of surface HER-2 was determined by flow cytometry using an antibody (primary antibody) that did not compete with Herceptin (see Methods). B–F represent either nonspecific staining (fluorescence uptake with only the secondary fluorescent antibody) or specific staining (fluorescence uptake with both the primary and secondary fluorescent antibody) for HER-2 expression in either the BT-474 cells or for different Herceptin-resistant clones. The mean channel fluorescence was similar in all of the cells that were either treated or not treated with Herceptin, indicating similar levels of overexpressed cell surface HER-2.

Downstream from HER-2 expression is the activation of AKT via the pI3 kinase pathway that alters p27. When Herceptin or 4D5 is added to the BT-474 cells, this activation step is blocked (Fig. 3). In a similar manner, the addition of Herceptin to the two resistant clones also results in a blockage of the activation of AKT. It should be noted that the AKT levels are similar between the wild-type and the resistant clones, while the pAKT nontreated levels in the resistant clones are much higher than that observed in the wild type. It is clear that Herceptin does downregulate pAKT levels in both the parental cells and the resistant clones.

Figure 3. Downregulation of pAKT by Herceptin in the Herceptin-sensitive and Herceptin-resistant cell lines. BT-474 cells (either parental cells [WT] or resistant clones were grown in the absence or in the presence of either 10 μg/ml of Herceptin (H) or 4D5. The cells were then harvested, lysed, and analyzed for either AKT expression or pAKT expression, using Western blot analysis (see Methods). The relative intensities are given for each of the analyses as defined by the NIH Image program.

The next step in the proposed major pathway is the phosphorylation of p27, in which the protein remains in the cytoplasm and does not block the cell cycle through inhibition of the cyclin dependent kinase system (cdk2). Our data show that treating the parental cells with Herceptin results in an increase in nuclear p27 by immunofluorescence from 8% to 32%. This increase does not occur or occurs to a much lower extent in the two Herceptin-resistant clones (Fig. 4). Flow cytometry analysis of nuclear p27 expression in isolated nuclei from BT474 treated with and without Herceptin is presented in Figure 5. In Figure 5A–C, the mean fluorescence units of wild-type BT-474 for background staining is 8.5, specific p27 staining in BT-474 cells is 67.8, and specific p27 staining in BT-474 cells treated with Herceptin is 200. The specific p27 staining in clone 5 and clone 6 under continuous treatment of Herceptin was 63.8 and 135.9, respectively. The levels of p27 expression in the resistant clones are similar to the wild-type BT-474 cells not treated with Herceptin.

Figure 4. Nuclear and cytoplasmic expression of p27 in Herceptin-sensitive and Herceptin-resistant BT-474 cells. Cells were harvested and plated in the presence or absence of 10 μg/ml of Herceptin. After 48 h, the plated cells were analyzed for p27 expression using immunofluorescence. The upper row represents phase contrast and the lower row represents p27 expression in the cells. The percent nuclei stained to the number of cells analyzed was a summation of different areas from three different experiments.

Figure 5. Nuclear p27 expression in Herceptin-sensitive and Herceptin-resistant BT-474 cells. Nuclei were isolated and stained for p27 expression and analyzed by flow cytometry. A: Minus primary p27 antibody (nonspecific staining). Treatment with both p27 and the secondary fluorescent antibodies. B,C: BT-474 cells minus or plus treatment with 10 μg/ml of Herceptin for seven days, respectively. D,E: Clones 5 and 6 with 10 μg/ml of Herceptin for seven days. Resistant clones show lower levels of p27 when treated with Herceptin (D,E) than do parental cells treated with Herceptin (C).

To further determine if nuclear p27 expression is altered in the Herceptin-resistant clones compared to the parental cells, the total lysates of resistant clones and parental cells were treated with and without Herceptin for three days and were then immunoprecipitated with anti-cdk-2 antibody. The highest levels of p27 associated with cdk-2 were found in Herceptin-treated parental cell lines. All resistant clones, whether treated with Herceptin or not, had low levels of p27 associated with cdk-2 (Fig. 6). The levels of actin expression were similar, indicating no loading artifacts

Figure 6. Association of p27 with cdk2 by immunoprecipitations of cell lysates. The parental cells or resistant clones were treated with and without 10 μg/ml of Herceptin for three days. The cells were harvested and immunoprecipitated with rabbit anti-cdk-2 antibody followed by protein A agarose. The washed pellet was dissolved in SDS PAGE buffer and analyzed for p27 expression. Lanes 1–6: BT-474, BT-474+Herceptin, Clone 5, Clone 5+Herceptin, Clone 6, Clone 6+Herceptin, respectively. The relative intensities for the cells and the treatments given for each of the analyses are as defined by the program NIH Image. Resistant clones show less p27 expression associated with cdk-2 than the parental cells do. Bottom row shows actin expression.

DISCUSSION

This work has demonstrated that BT-474 cells can be made resistant to Herceptin treatment by continued treatment with the drug. The resistant cells continue to grow in the presence of Herceptin and have a high proliferative activity based on DNA analysis using flow cytometry. Using different antibodies to HER-2 results in different degrees of cell cycle arrest. In our comparative studies, the order of effectiveness was 4D5 > Herceptin > 7C2. This is in agreement with previous work with these antibodies (10), but what was interesting was the agonistic effect of 7C2 when added with Herceptin, and the antagonistic effect of Herceptin when added to 4D5. These data would suggest that antibodies to different sites on HER-2 might provide additional effectiveness in breast cancer treatment. However, addition of Herceptin with 4D5 impedes the growth inhibition of 4D5. This is consistent with the higher binding affinity of Herceptin and its more modest biological effect. 4D5 growth–response curves for either the parental cells or the Herceptin-resistant clones have indicated that the parental cell line is most sensitive to 4D5. Herceptin-resistant clones are less sensitive to 4D5 growth inhibition, but are inhibited at higher concentrations of 4D5. At identical concentrations of Herceptin, no growth inhibition was detected in these resistant clones. These data would indicate that these antigen–antibody interactions are very complex and agrees with the results of Spiridon et al. (13), who have demonstrated that multiple antibodies at different epitope sites results in additive to synergistic effects on inhibition of cell proliferation. Work by Agus et al. (14) has demonstrated that another antibody, 4C2, can inhibit proliferation by blocking dimerization of HER-2 family members. Treating patients with multiple antibodies might be more beneficial than single agents (14).

Our data and that of others (8, 12, 15) would indicate that cells that overexpress HER-2 proliferate faster because p27, an inhibitor of cdk2 and of cell proliferation, remains in the cytoplasm and does not translocate to the nucleus. This loss of translocation of p27 is because AKT can phosphorylate the threonine 157, which is in the nuclear localization domain of p27 and, thus, destroys the sequence necessary for translocation (15). Herceptin downregulates AKT activity and the resulting threonine 157 is not phosphorylated. This lack of phosphorylation allows for p27 to enter the nucleus (Fig. 4). Once in the nucleus, p27 can inhibit cdk2 activity and the cells remain in the G1/G0 phase of the cell cycle. Our studies have analyzed the various steps from Herceptin binding to HER-2 to translocation of p27. Our data suggest that HER-2 expression is not altered by Herceptin treatment (Fig. 2) and activation of AKT is downregulated in both the parental cells and the two resistant clones (Fig. 3). We propose that Herceptin resistance resides in the lack of p27 translocation to the nucleus, as demonstrated in Figures 4 and 5. In cdk2 immunoprecipitation of p27 expression, there is less association of p27 with cdk2 in the resistant clones compared to the parental cells. The addition of Herceptin markedly increases the association of p27 expression with cdk2 in the parental cells, but not in resistant clones (Fig. 6). One concern is defective nuclear import of p27 in the resistant clones in the absence of AKT activity. There are several potential reasons that still need to be explored. First, the AKT activity might be diminished, but not eliminated. Measuring the AKT kinase activity between the resistant and parental cells could test this hypothesis. Secondly, there could be a mutation of the p27 that affects the translocation, binding to cdk-2, or stability of the nuclear p27. Thirdly, there could be alterations in the transport at the nuclear membrane level. Finally, there could be an increase of p27 degradation in the nucleus of the resistant clones compared to the parental cells. All of these alterations are currently under investigation.

Other groups have also obtained cells that are resistant to Herceptin. Dugger et al. (16) developed a model system of spontaneous mammary tumor growth in HER-2–overexpressing transgenic animals. The resulting tumors were treated with Herceptin and resistant tumors were obtained. They found that mutations in the epitope sequence of HER-2 where Herceptin binds was one mechanism of resistance. The cells still overexpressed HER-2, but failed to bind Herceptin. Chan et al. (17) characterized Herceptin-resistant cells from BT-474 cells, similar to our methods. They observed similar results to ours and found that the resistant clones were now more sensitive to tamoxifen treatment. Finally, Dr. Arteaga's group has also studied Herceptin resistance by continuous growth of BT-474 cells in athymic nude mice under long treatment. Their results indicate that the resistant clones are also resistant under in vitro conditions, and that they have not lost HER-2 expression (personal communication, Dr. Carlos Arteaga, Vanderbilt University). It is clear that there are many pathways to develop resistance to surface antibody treatments. Recent data would suggest that measurement of nuclear to cytoplasmic p27 expression could predict the prognosis of the patient (15, 18, 19). It would be of interest to understand if nuclear exclusion of p27 expression is an indicator of Herceptin resistance.