A naturally occurring HER2 carboxy-terminal fragment promotes mammary tumor growth and metastasis

Abstract

HER2 is a tyrosine kinase receptor causally involved in cancer. A subgroup of breast cancer patients with particularly poor clinical outcomes expresses a heterogeneous collection of HER2 carboxy-terminal fragments (CTFs). However, since the CTFs lack the extracellular domain that drives dimerization and subsequent activation of full-length HER2, they are in principle expected to be inactive. Here we show that at low expression levels one of these fragments, 611-CTF, activated multiple signaling pathways because of its unanticipated ability to constitutively homodimerize. A transcriptomic analysis revealed that 611-CTF specifically controlled the expression of genes that we found to be correlated with poor prognosis in breast cancer. Among the 611-CTF-regulated genes were several that have previously been linked to metastasis, including those for MET, EPHA2, matrix metalloproteinase 1, interleukin 11, angiopoietin-like 4, and different integrins. It is thought that transgenic mice overexpressing HER2 in the mammary glands develop tumors only after acquisition of activating mutations in the transgene. In contrast, we show that expression of 611-CTF led to development of aggressive and invasive mammary tumors without the need for mutations. These results demonstrate that 611-CTF is a potent oncogene capable of promoting mammary tumor progression and metastasis.

FIG. 1.

Generation and characterization of cellular models expressing individual CTFs of HER2. (A) Schematic showing the primary sequence of the juxtamembrane regions of HER2, the alpha-secretase cleavage sites (α), the transmembrane domain (TM), a putative gamma-secretase cleavage site (γ), the NLS, the N-glycosylated Asn-629, and the position of amino acid residues 611, 648, 676, and 687 (corresponding to the N termini of the HER2 CTFs). The schematics below the sequence represent cDNA constructs used for expression of the different CTFs. (B) Schematic depicting the different HER2 CTFs generated by alternative initiation of translation (left) and proteolytic processing (right). (C) Expression from the cDNA constructs shown in panel A. MCF7 Tet-Off clones stably transfected with the empty vector (-) or with the vector containing the cDNA of HER2 or 611-, 648-, 676-, or 687-CTF under the control of a Tet/Dox-responsive element were kept with or without doxycycline (Dox) for 24 h, lysed, and analyzed by Western blotting with an antibody against the cytoplasmic domain of HER2. (D) The MCF7 clones stably transfected with vector (-), HER2, or 611-, 648-, 676-, or 687-CTF, as in panel C, were analyzed 24 h after induction of expression with a confocal microscope by indirect immunofluorescence with an antibody against the cytoplasmic domain of HER2. The bar in the first photo from the left represents 30 μm.

FIG. 2.

Characterization of CTFs from BT474 cells and breast cancer samples. (A) BT474 cells were treated with APMA, 1,10-phenanthroline, and/or control solvent, as indicated. Cell lysates were analyzed by Western blotting with an antibody against the cytoplasmic domain of HER2. (B) Lysates of BT474 cells treated with APMA and membrane fraction from tumor sample 108 were incubated overnight with or without N-glycosidase F, followed by Western blot analysis with an antibody against the cytoplasmic domain of HER2. (C) Tissue samples of human mammary tumors were fractionated, and equal amounts of total soluble (S) and membrane (M) fractions were analyzed by Western blotting with an antibody against the cytoplasmic domain of HER2 or, as a control, with an antibody against the cytosolic protein LDH.

FIG. 3.

Activation of signal transduction pathways in cells expressing the different CTFs. (A) The MCF7 clones stably transfected with vector (-), HER2, or 611-, 648-, 676-, or 687-CTF, as in Fig. 1C, were washed with media with or without doxycycline (Dox), cultured for 24 h, lysed, and analyzed by Western blotting with the indicated antibodies. Lanes marked with an arrow were used for quantification. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (B) Western blot signals from the expressed constructs and the indicated phosphoproteins in the three independent experiments described for panel A (and in Fig. S6 in the supplemental material) at different time points were quantified. The level of each phosphoprotein was divided by the level of expressed HER2 or CTF in the same sample, followed by normalization to the level of phosphoprotein in cells transfected with empty vector. The averages are presented as bars ± standard deviations.

FIG. 4.

Transcriptomic analysis with HER2 and 611-CTF. (A) Venn diagram showing the number and overlap of genes regulated after 15 or 60 h of expression of 611-CTF or HER2 in MCF7. The identities and n-fold induction of in total 624 genes in the seven different groups are listed in Table SI in the supplemental material. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. (B) Cells of the MCF7 clones stably transfected with HER2 or 611-CTF were washed with media with or without doxycycline (Dox), cultured for 15, 30, 60, or 90 h, lysed, and analyzed by Western blotting with the indicated antibodies. (C) Unsupervised hierarchical clustering of 395 primary breast tumors from two independent publicly available studies, on the basis of expression levels of the 76-gene signature of 611-CTF (see Fig. S11 in the supplemental material). *, patients that died from breast cancer; **, mean survival time (years). (D) Kaplan-Meier survival plots of the two patient groups classified in panel C.

FIG. 5.

Active 611-CTF complexes are maintained by disulfide bonds. (A) The MCF7 clones stably transfected with vector (-), HER2, or 611-, 648-, 676-, or 687-CTF, as in Fig. 1C, were washed with media without doxycycline, cultured for 24 h in the presence or absence of 1 mM lapatinib, lysed, fractionated by SDS-PAGE in the presence or absence of beta-mercaptoethanol, and analyzed by Western blotting with the indicated antibodies. Arrow, dimers. (B) Schematic showing the primary sequence of the extracellular juxtamembrane region of HER2 with the transmembrane domain (TM), the position of amino acids 611 and 648, and the five juxtamembrane cysteines marked. The schematics below the sequence show the different cysteine substitutions inserted in the cDNA of 611-CTF. (C) MCF7 cells were transiently transfected with the empty vector (-), the vector containing the cDNA of HER2, or wild-type 648- or 611-CTF or CTF-611 with one (C-S), three (3C-A), or five (5C-A) cysteines substituted. After 48 h, the cells were lysed and analyzed as in panel A. (D) The same lysates as in panel C with beta-mercaptoethanol were analyzed by Western blotting with the indicated antibodies. The specific signals of three independent experiments were quantified as in Fig. 3B. Statistical analysis of the normalized ratios using Student's t test showed statistically significant differences for P-Erk1/2 and P-Akt (S473) between cells expressing 611/5C-A and wild-type 611-CTF (*, P < 0.01).

FIG. 6.

Expression of 611-CTF in mouse mammary glands results in mild abnormal alveolar development. (A) Total protein lysates from mammary glands of 7-week-old TG mice were analyzed by Western blotting with an antibody against the cytoplasmic domain of HER2 (left). The average expression, normalized to HER2 in wild-type (WT) mice, in four independent Western blot analyses is presented as bars ± standard deviations (right). (B) Carmine-stained whole mounts (left), ×25 magnifications of the same whole mounts (middle), and hematoxylin and eosin-stained sections of mammary glands (right; magnification, ×400) from 7-week-old animals.

FIG. 7.

TG 611-CTF mice develop mammary adenocarcinomas more aggressive than those in HER2-expressing mice. (A) Representative TG HER2 (35-week-old) and 611-CTF (25-week-old) mice. (B) Numbers of tumors that had developed in 40-week-old mice. Animals with tumors in every gland were counted as 10. The averages are presented as bars ± standard deviations. (C) Tumor-free survival curves of TG HER2 (n = 32), 611-CTF (n = 7), and 687-CTF (n = 26) mice. (D) Individual tumors from the indicated TG animals were measured at different time points. Each point represents the average ± standard deviation of ten individual tumors.