Introduction

Human breast cancer is a heterogeneous disease comprised of three major subgroups, each encompassing unique molecular signatures, prognoses, and responses to therapies1. HER2 is a member of the epidermal growth factor receptor (EGFR) family of receptor tyrosine kinases, which includes EGFR, HER2, HER3, and HER42,3. Homo- or hetero-dimerization of these receptors results in the phosphorylation of residues in the intracellular domain and the consequent recruitment of adapter molecules responsible for the initiation of several signaling pathways involved in cell proliferation and survival. HER2-amplified/overexpressed breast cancer is one of the aforementioned subgroups. It accounts for approximately 20% - 30% of all breast cancer cases, and it is characterized by an aggressive phenotype and poor overall survival4,5,6. Despite therapeutic advances brought forth by anti-HER2 agents including monoclonal antibodies such as trastuzumab (Herceptin) and small molecule tyrosine kinase inhibitors, patients with advanced HER2-positive breast cancer frequently experience disease progression and/or recurrence7,8,9.

One of the hallmarks of cancer is dysregulated proliferation10. Unsurprisingly, molecular characterization of human tumors reveals that key cell cycle regulators are frequently dysregulated11. During cell cycle progression, cyclin-dependent kinases (CDKs) and cyclins constitute the central regulatory apparatus. In mammalian cells, kinase subunits (CDK4, CDK6, CDK2, and CDC2) are expressed alongside cyclins (cyclin D, E, A, and B) sequentially as the cells progress from G1 through mitosis. CDK4 and CDK6 form complexes with one of several D-type cyclins and function early in G1 phase, probably in response to growth factors. CDK2 forms complexes with cyclins E or A and functions in the G1/S phase transition and S phase DNA replication12,13,14. A major molecular consequence of HER2 up-regulation is the increased expression of G1-to-S cell cycle regulatory proteins cyclins D and E, which leads to aberrant cell proliferation7,12,13,14,15,16. High cyclin E expression is a marker that correlates strongly with poor outcome in patients with breast cancer. Since Cyclin E amplification/overexpression leads to trastuzumab resistance, disrupting Cyclin E expression could have therapeutic importance for HER2-positive breast cancers17,18.

Taspase1 was originally purified as the protease that cleaves MLL (the Mixed-Lineage Leukemia protein; also known as MLL1) for proper regulation of HOX gene expression19,20. Other genetically and biochemically proven Taspase1 substrates include MLL2 (also known as MLL4), TFIIAα-β, ALFα-β (TFIIA-Like Factor) and Drosophila HCF-1 (Host Cell Factor 1)20,21,22,23,24. Interestingly, all confirmed Taspase1 substrates are nuclear transcription factors that play important roles in gene regulation. Taspase1 encodes a highly conserved 50 kDa α-β proenzyme, which undergoes intramolecular autoproteolysis, producing the mature α28/β22 heterodimeric enzyme that displays an overall α/β/β/α structure20,25. A complete genetic knockout of Taspase1 in mice resulted in profound early postnatal lethality and the few surviving Taspase1−/− mice universally exhibited small body sizes and homeotic transformations at the axial skeleton22. Taspaes1−/− mouse embryonic fibroblasts (MEFs) displayed cell cycle progression defects with downregulation of cyclins E, A, and B and upregulation of CDKIs (cyclin-dependent kinase inhibitors) p16, p21 and p2722. Importantly, Taspase1−/− MEFs were resistant to oncogenic transformation in vitro26. In MEFs, Taspase1 cleaves MLL that interacts with E2Fs, core transcription factors of the mammalian cell cycle, to activate select Cyclin genes22,27,28. How Taspase1 regulates CDKIs, however, remains unclear. Importantly, Taspase1 shows a high level of expression in most human cancer cell lines22, and knockdown of Taspase1 in many cancer cell lines impairs cancer cell proliferation and even sensitizes brain cancer and melanoma cells to anoikis26.

Encouragingly, despite the fact that Taspase1 plays an important role in mammalian embryogenesis, acute genetic deletion of Taspase1 in adult mice does not confer discernible toxicities on the mice, which suggests a wide therapeutic index for Taspase1 inhibition in adult cancer patients29. Moreover, pharmacological inhibition of Taspase1 has been attempted29,30, and a primitive small molecular Taspase1 inhibitor (TASPIN) showed effects on U251 brain tumor xenografts and HER2-driven mouse breast cancers29. The latter findings prompted us to hypothesize that Taspase1 could play a critical role in HER2-positive breast cancer and that Taspase1 inhibitors may be developed as a safe treatment option for Taspase1-dependent cancers.

Here, we report the data from preclinical experiments that we conducted by constructing genetically well-defined mouse models, demonstrating that Taspase1 ablation blocks MMTV-neu-driven breast cancer initiation in vivo. We further pinpoint that MLL is the key Taspase1 substrate whose cleavage is required for MMTV-neu-induced tumor formation. The cleavage of MLL by Taspase1 enables HER2/neu-induced overexpression of Cyclins E and A, presenting an essential in vivo genetic network conferring breast tumorigenesis.

Results

Taspase1 deficiency disrupts the expression of cyclins and proliferation of HER2+ breast cancer cells

To determine whether Taspase1 is required for HER2-positive breast cancer cell proliferation, we conducted genetic knockdown experiments in two HER2-overexpressing breast cancer cell lines, BT474 and HCC1419. Taspase1 deficiency significantly reduced the cell number in both cell lines (Figure 1A). Cell death assay confirmed that there is no significant difference in cell death between the Taspase1 knockdown cells and the control in either cell line (Figure 1B). On the other hand, cell cycle analysis showed that Taspase1 knockdown significantly decreased the S phase population in both cell lines (Figure 1C). These data suggest that Taspase1 regulates HER2-positive breast cancer cell proliferation through promoting cell cycle progression.

Figure 1
figure 1

Taspase1 deficiency disrupts the proliferation of HER2-positive breast cancer cells. (A) Proliferation of Taspase1 knockdown BT474 and HCC1419 cells. 1 × 105 scramble-control (sh-scr) or Taspase1 (sh-T1) knockdown cells were seeded in triplicate wells and counted at day 4. Data presented are mean ± SD of three independent experiments. *P< 0.001. (B) Apoptosis of Taspase1 knockdown BT474 and HCC1419 cells. Cell death of scramble-control (sh-scr) or Taspase1 (sh-T1) knockdown cells was assessed by FACS analysis, following Annexin V staining. Data presented mean ± SD of four independent experiments. (C) Cell cycle profiles of Taspase1 knockdown BT474 and HCC1419 cells. The indicated scramble-control (sh-scr) or Taspase1 (sh-T1) knockdown cells were stained with propidium iodide (PI) and analyzed by FACS. Data presented are mean ± SD of four independent experiments. *P< 0.005. (D) Western blots of cyclins, CDKIs, and Taspase1 on scramble-control (sh-scr) or Taspase1 (sh-T1) knockdown BT474 and HCC1419 cells. β-Actin serves as loading control. (E) Soft agar assays using Taspase1 knockdown BT474 cells expressing either empty vector (sh-T1-Con) or knockdown resistant Taspase1 (sh-T1-RT1). Scale bar, 200 μm. Data presented are the mean ± SD. ***P< 0.0005. (F) Western blots of cyclin E2, cyclin A, Taspase1, and β-Actin on scramble-control knockdown (sh-scr), Taspase1 knockdown (sh-T1), Taspase1 knockdown expressing empty vector (sh-T1-Con), and Taspase1 knockdown expressing knockdown resistant Taspase1 (sh-T1-RT1) BT474 cells. (G) qRT-PCR analyses of scramble-control knockdown (sh-scr), Taspase1 knockdown (sh-T1), Taspase1 knockdown expressing empty vector (sh-T1-Con), and Taspase1 knockdown expressing knockdown resistant Taspase1 (sh-T1-RT1) BT474 cells. Expression was normalized to human 18S rRNA. Data presented are mean ± SD of three independent experiments. *P< 0.05; **P< 0.005; ***P< 0.0005.

Full size image

We next investigated the underlying mechanisms by which Taspase1 regulates cell division. Key regulators of the mammalian cell cycle machinery include E2Fs, Rbs, cyclins, cyclin-dependent kinases (CDKs), and CDK inhibitors (CDKIs), which form complex positive and negative epistatic regulatory loops to ensure accurate cell cycle progression. In MEFs, following the cleavage by Taspase1, MLLN320/C180 targets to Cyclins E and A promoters through interaction with E2Fs to methylate histone H3 at K4, thereby transactivating Cyclins E and A for cell proliferation22,28. To gain mechanistic insight into how Taspase1 regulates HER2-positive breast cancer cell proliferation, we examined the expression of several key cell cycle regulators. Western blot analyses of Taspase1-knockdown BT474 and HCC1419 cells revealed a significant decrease in cyclins E2 and A, but not D1 (Figure 1D), consistent with our prior results obtained in MEFs22. Altogether, these results indicate that in HER2-positive breast cancer cells Taspase1 assures the proper accumulation of cyclins E and A for proliferation.

The ability of cancer cells to form colonies on soft agar is a stringent in vitro surrogate of in vivo tumorigenicity. Soft agar assays assess the capacity of tumor cells to not only proliferate but also resist anoikis under three-dimensional culture conditions that imitate the in vivo tumor growth environment. We determined the degree to which Taspase1 is required for the colony formation capability of HER2-positive breast cancer cells on soft agar. Knockdown of Taspase1 (sh-T1) in BT474 cells severely compromised their ability to grow as colonies on soft agar (Figure 1E). To validate the specific requirement of Taspase1 for cancer cell growth on soft agar, we engineered a sh-T1 knockdown resistant version of Taspase1 (RT1). Retroviral delivery of RT1 rescued the ability of Taspase1-knockdown (sh-T1) BT474 cells to form colonies (Figure 1E). Western blot analysis confirmed the successful knockdown of Taspase1 in BT474 cells by sh-T1 and the resulting reduced protein levels of cyclins E2 and A (Figure 1F), and the resistance of RT1 to sh-T1 and the restoration of the protein levels of cyclins E2 and A in RT1-reconstituted sh-T1 BT474 cells (Figure 1F). We further interrogated the mechanisms by which Taspase1 sustains cyclins levels. Quantitative real-time PCR (qRT-PCR) assays revealed that the mRNA levels of Cyclins E1, E2, and A were reduced in sh-T1 BT474 cells and were restored to baseline in RT1 sh-T1 BT474 cells (Figure 1G). Similar results were obtained utilizing HER2-transformed MCF10A cells (Supplementary information, Figure S1A and S1B). Altogether, these results indicate that Taspase1 controls the cell division cycle of HER2-positive breast cancer cells largely by conferring proper transcription of the Cyclins E and A genes upon aberrant receptor tyrosine kinase signaling.

Deletion of Taspase1 in mouse mammary glands blocks MMTV-neu-driven breast cancer formation

To determine whether Taspase1 is required for breast tumorigenesis in vivo, we generated MMTV-neu;MMTV-cre;Tasp1F/− mice by employing the widely adapted MMTV-neu mouse model15,31. MMTV-neu;MMTV-cre;Tasp1+/+ female mice were generated as positive controls and monitored for breast cancer formation. All of our MMTV-neu;MMTV-cre;Tasp1+/+ virgin female mice (n = 30) developed breast tumors between 30 and 50 weeks of age (Figure 2A, 2B), as did their counterparts in previously published studies15,31. In stark contrast, 26 of 30 MMTV-neu;MMTV-cre;Tasp1F/− females remained breast cancer free at 60 weeks of age (Figure 2A). Notably, western blot analysis of 12 week-old mammary glands detected similar levels of HER2/neu protein expression in both MMTV-neu;MMTV-cre;Tasp1+/+ and MMTV-neu;MMTV-cre;Tasp1F/− female virgin mice (Supplementary information, Figure S2). Interestingly, Taspase1 ablation did not block the tumor initiation in MMTV-Wnt;MMTV-cre;Tasp1F/− female mice, which suggests the specificity of Taspase1 function in HER2-driven breast cancer (Figure 2C-2E). Altogether, these in vivo data in conjunction with our in vitro assays unequivocally establish an essential role for Taspase1 in HER2/neu-induced breast tumorigenesis in mice and probably in humans.

Figure 2
figure 2

Deletion of Taspase1 in the mammary gland protects mice from developing MMTV-neu-driven breast cancer. (A) Kaplan-Meier curve of breast cancer incidence of MMTV-neu;MMTV-cre;Tasp1F/− female mice. MMTV-neu;MMTV-cre;Tasp1+/+ female mice were used as control. n = 30 mice per genotype. P< 0.0001. (B) H&E staining of mouse mammary tumors for the indicated genotypes. Scale bar, 100 μm. (C) Kaplan-Meier curve of breast cancer incidence of MMTV-Wnt;MMTV-cre;Tasp1F/− female mice. MMTV-Wnt;MMTV-cre;Tasp1+/+ female mice were used as control. n = 30 mice per genotype. P= 0.5728. (D) H&E staining of mouse mammary tumors for the indicated genotypes. Scale bar, 100 μm. (E) Western blot analyses of Taspase1 using female mammary glands of the indicated genotypes. NS indicates non-specific band, serving as control for equal loading.

Full size image

Genetic deletion of Taspase1 impacts neither normal development nor physiological proliferation of mouse mammary glands

We next investigated the mechanisms by which loss of Taspase1 in the mammary gland protects MMTV-neu female mice from developing breast cancer by determining whether Taspase1 ablation simply disrupts mammary gland development and/or response to physiological proliferation signals such as pregnancy. To this end, we compared whole mounts of dissected, carmine-stained mammary glands from MMTV-cre;Tasp1+/+ and MMTV-cre;Tasp1F/− female mice, prepared at 6 weeks of age, day 13 of pregnancy, and day 1 of lactation. No macroscopic differences were detected between MMTV-cre;Tasp1+/+ and MMTV-cre;Tasp1F/− mice at any of these pre-specified developmental and physiological states (Figure 3 and Supplementary information, Figure S3). Thus, ablation of Taspase1 in mouse mammary tissues using MMTV-cre has no effect on normal development or physiological proliferation of mammary glands. Of note, the importance of Taspase1 in breast tumorigenesis (Figure 2A) and its dispensability in normal mammary gland physiology (Figure 3 and Supplementary information, Figure S3) are reminiscent of those reported with Cyclin D1 knockout mice15.

Figure 3
figure 3

Genetic deletion of Taspase1 does not disrupt mammary gland development or its proliferative response during gestation. Whole mounts of dissected, carmine-stained female mammary glands from mice of the indicated genotypes, age, and gestation status. n = 3-4 mice per each genotypes. Representative images are presented. Scale bar, 1mm.

Full size image

Taspase1 deficiency prevents tumor formation of MMTV-neu mammary glands

We analyzed mammary glands of transgenic mice to further dissect the mechanisms by which Taspase1 enables MMTV-neu- induced breast tumorigenesis. Whole mounts were performed using mammary glands of 12-week-old wild-type, MMTV-neu;MMTV-cre;Tasp1+/+, and MMTV-neu;MMTV-cre;Tasp1F/− female mice to evaluate glandular structures. In comparison to those of wild-type control mice, mammary glands of MMTV-neu;MMTV-cre;Tasp1+/+ mice displayed increased densities in ducts and end buds (Figure 4A), similar to what was previously documented for this mouse strain31. Interestingly, this overproliferation phenotype induced by the MMTV-neu transgene was not observed in the mammary glands of MMTV-neu;MMTV-cre;Tasp1F/− mice (Figure 4A). Immunohistochemistry for histone H3 serine 10 phosphorylation (pH3S10) which is a cell proliferation marker, showed that mammary glands of 12-week-old MMTV-neu;MMTV-cre;Tasp1+/+ mice had greater than 10-fold more pH3S10 positive cells than those of Wild-type and MMTV-neu;MMTV-cre;Tasp1F/− mice (Figure 4B). Altogether, these genetic data indicate that Taspase1 is required for the MMTV-neu-driven aberrant gland proliferation and thus breast carcinogenesis in mice.

Figure 4
figure 4

Loss of Taspase1 disrupts the overexpression of cyclin E and the cancer formation in MMTV-neu mouse mammary glands. (A) Whole mounts of dissected, carmine-stained female mammary glands of the indicated genotypes at 12 weeks of age. n = 5-15 mice per genotype. Representative images are presented. Scale bar, 1mm. (B) pH3S10 staining of female mammary glands of the indicated genotypes at 12 weeks of age. n = 8 mammary glands per group. Representative images are presented. Red arrow head indicates positive cells. Scale bar, 100 μm. *P< 0.0001. (C) Whole mounts of female mammary glands of the indicated genotypes at 20 weeks of age. n = 5-15 mice per group. Representative images are presented. Scale bar, 1mm. (D) Western blot analyses of cyclins, CDKIs, and Taspase1 using female mammary glands and tumor extracts of the indicated genotypes. β-Actin serves as loading control.

Full size image

MMTV-neu mice can develop multiple microscopic mammary gland tumors at as early as 14 weeks of age31. Accordingly, we examined mammary glands of 20-week-old wild-type, MMTV-neu;MMTV-cre;Tasp1+/+, and MMTV-neu;MMTV-cre;Tasp1F/− female virgin mice by whole mounts. The mammary glands of MMTV-neu;MMTV-cre;Tasp1+/+ mice showed significant glandular proliferation and developed tumor foci of various sizes, whereas those of MMTV-neu;MMTV-cre;Tasp1F/− mice displayed neither overproliferation nor tumor foci, which are macroscopically indistinguishable from wild-type (Figure 4C and Supplementary information, Figure S4). Western blot analysis of 10-week-old wild-type, MMTV-neu;MMTV-cre;Tasp1+/+, and MMTV-neu;MMTV-cre;Tasp1F/− mammary glands for cyclins E, A, and D, and CDKIs p16 and p27 detected consistent overexpression of cyclin E in MMTV-neu;MMTV-cre;Tasp1+/+ mice (Figure 4D). A higher level of Taspase1 expression was also detected in these MMTV-neu;MMTV-cre;Tasp1+/+ mammary glands, which is consistent with our prior observation that Taspase1 is commonly overexpressed in cancer cells26. Moreover, examination of MMTV-neu;MMTV-cre;Tasp1+/+ breast tumors revealed that cyclins E, A, D, and Taspase1 were all expressed at high abundance (Figure 4D). The successful deletion of Taspase1 was observed in the MMTV-neu;MMTV-cre;Tasp1F/− mammary glands and remarkably, MMTV-neu was no longer able to induce cyclin E overexpression in these glands (Figure 4D). These data support the notion that overexpression of cyclin E and Taspase1 occurs early in MMTV-neu-driven breast tumorigenesis and precedes the overexpression of cyclins A and D.

Taspase1-mediated proteolytic cleavage of MLL is required for MMTV-neu-driven breast cancer formation but dispensable for mammary gland development

Our biochemical and genetic data thus far suggest that Taspase1 enables HER2/neu-driven breast tumorigenesis by permitting transcriptional activation of Cyclin E (Figures 1,2,3,4). Since Taspase1 cleaves nuclear factors MLL, MLL2, TFIIA, and ALF to regulate transcriptional programs, we asked the cleavage of which substrate(s) by Taspase1 is necessary for MMTV-neu-induced breast carcinogenesis. Of note, our prior biochemical studies in MEFs revealed that Taspase1 cleaves MLL to transactivate Cyclin E22. However, MLL plays complex and context-dependent roles in cell cycle control. It can either positively or negatively regulate cell proliferation. For example, contrary to the positive regulation of cell proliferation that is more commonly observed, MLL was shown to negatively regulate pancreatic neuroendocrine cell proliferation. In this context MLL interacts with the tumor suppressor Menin to activate p27, a CDKI for cell cycle inhibition and thus prevent aberrant neoplastic proliferation32. Furthermore, an independently generated non-cleavable MLL mouse model showed no overt proliferation defects in MEFs33. Hence, it is of significance to address the role of MLL cleavage, if any, in MMTV-neu-driven breast tumorigenesis using genetic models. Our data suggested that specific ablation of Taspase1-mediated MLL cleavage could suppress MMTV-neu-driven breast cancer formation. We accordingly generated MMTV-neu;MLLnc/nc transgenic mice that harbor homozygous Taspase1 non-cleavable (nc) alleles of MLL. Unlike Taspase1−/− mice, which died prematurely, MLLnc/nc mice were viable and fertile34. Like MMTV-neu;MMTV-cre;Tasp1F/− mice (Figure 2A), most MMTV-neu;MLLnc/nc females (27 of 30) were, remarkably, free of breast cancer at 60 weeks of age (Figure 5A). We then determined whether MLLnc/nc females display any mammary gland defects. Whole mounts of MLLnc/nc female mammary glands at 6 weeks of age, day 13 of pregnancy, and day 1 of lactation did not reveal any abnormalities (Figure 5B). Altogether, these results indicate that Taspase1-mediated MLL cleavage plays a critical role in MMTV-neu-induced breast carcinogenesis in vivo.

Figure 5
figure 5

The cleavage of MLL by Taspase1 is required for MMTV-neu-driven breast cancer formation but dispensable for mammary gland development. (A) Kaplan-Meier curve of breast cancer incidence of MMTV-neu;MLLnc/nc mice. The same positive control MMTV-neu;MMTV-cre;Tasp1+/+ female mice (Figure 3) were plotted as control. n = 30 mice per group. P< 0.0001. (B) Whole mounts of female mammary glands of MLLnc/nc mice at the indicated age and gestation status. n = 3-4 mice per group. Representative images are presented. Scale bar, 1 mm.

Full size image

Non-cleavable MLL disrupts MMTV-neu-induced aberrant proliferation and cyclin E overexpression in mammary glands

We next investigated the molecular basis underlying MLLnc/nc females' resistance to MMTV-neu-induced breast tumorigenesis by comparing the mammary glands of MMTV-neu;MLLnc/nc female virgin mice to those of wild-type and MMTV-neu;MMTV-cre;Tasp1+/+ mice at 12 and 20 weeks of age. Consistently, the increased duct and end bud density induced by the MMTV-neu transgene was not observed in mammary glands of MMTV-neu;MLLnc/nc mice (Figure 6A). We subsequently performed immunohistochemistry analysis for histone H3 serine 10 phosphorylation (pH3S10) on the mammary glands of 12-week-old MMTV-neu;MLLnc/nc mice. Like MMTV-neu;MMTV-cre;Tasp1F/− and wild-type mice (Figure 4B), MMTV-neu;MLLnc/nc mice had 10-fold fewer pH3S10+ cells in their mammary glands than MMTV-neu;MMTV-cre;Tasp1+/+ mice (Figure 6B). Finally, we examined the expression of cyclins E, A, D, and CDKIs p16 and p27 in the mammary glands of 10-week-old MMTV-neu;MLLnc/nc mice. Non-cleavage of MLL suppressed MMTV-neu-induced cyclin E accumulation (Figure 6C), similar to what was observed in MMTV-neu;MMTV-cre;Tasp1F/− mice (Figure 4D). Overall, these genetic and biochemical results demonstrate that Taspase1 and thus mature MLLN320/C180, generated upon Taspase1-mediated cleavage of the precursor MLL500, are required for MMTV-neu-driven breast carcinogenesis, which involves transcriptional activation of Cyclin E in the mammary glands.

Figure 6
figure 6

Non-cleavage of MLL disrupts tumor formation and induction of cyclin E in MMTV-neu mouse mammary glands. (A) Whole mounts of female mammary glands of the indicated genotypes at 12 and 20 weeks of age. n = 5-15 mice per group. Representative images are presented. Scale bar, 1 mm. (B) pH3S10 staining of female mammary glands of the indicated genotypes at 12 weeks of age. n = 8 mammary glands per group. Representative images are presented. Red arrow head indicates positive cells. Scale bar, 100 μm. *P< 0.0001. (C) Western blot analyses of cyclins, CDKIs, and Taspase1 using female mammary glands and tumors of the indicated genotypes. Wt #1 and MMTV-neu;MMTV-cre;Tasp1+/+ tumor #2 extracts (as Figure 4D) serve as reference for relative abundance. β-Actin serves as loading control.

Full size image

Discussion

The cloning of Taspase1 (threonine aspartase) founded a novel class of endopeptidases that employ the NH2-terminal threonine of the mature β subunit to cleave protein substrates after P1 aspartate20. In this study, using genetically well-defined mouse breast cancer models we show that Taspase1 is required for MMTV-neu-driven mammary tumorigenesis, which represents the first in vivo study demonstrating that Taspase1 ablation suppresses tumor initiation.

The mixed-lineage leukemia(MLL) gene encodes an epigenetic transcriptional regulator belonging to the trithorax group family35. MLL is a confirmed substrate of Taspase122. In its best-known developmental role, MLL maintains proper expression of Hox genes and thus coordinates the segmental body plan of vertebrates36,37. Mice deficient for MLL (MLL+/− and MLL−/−) accordingly display homeotic defects in their axial skeleton37. In addition to patterning body axis, MLL also regulates hematopoiesis, cell cycle, and cancer cell invasion24,37,38,39,40,41,42. MLL possesses histone H3 lysine 4 (H3K4) methyl transferase (HMT) activity43,44,45,46. MLL-catalyzed H3K4 trimethylation (H3K4me3) activates transcription, leading to orchestrated upregulation of key developmental, cell cycle, and cancer cell invasion genes such as Hox genes, Cyclins E and A, and MMP1 and MMP3, respectively22,24,28,45,46. Notably, the activity of MLL can be modulated by post-translational modifications, such as phosphorylation, ubiquitination, and proteolysis20,41,42,47. The 500-kDa precursor MLL (MLL500) undergoes Taspase1-mediated proteolytic cleavage, which gives rise to the mature MLLN320/C180 heterodimer20,25 that binds to Cyclins E and A promoters22,28. MLL forms complexes with E2Fs to methylate H3K4 at promoters, and thereby transactivates Cyclins E and A22,28. In the absence of Taspase1, MLL exists as MLL500, a noncleaved precursor with reduced HMT activity20,22, and is unable to fully activate the expression of Cyclins E, A and B22. The in vivo significance of MLL cleavage has nonetheless been questioned on the grounds that whereas Taspase1−/− mice exhibit diverse developmental defects, MLLnc/nc mice are born at Mendelian ratio, fertile, and grossly normal24,33. Our genetic study here unequivocally establishes the in vivo significance of the Taspase1-MLL-cyclin E pathway in carcinogenesis by demonstrating that MMTV-neu;MLLnc/nc mice are protected from MMTV-neu-driven breast cancer formation. Taspase1-mediated cleavage of MLL is therefore required for HER2/neu-induced tumorigenesis. However, whether cleavages of other Taspase1 substrates, such as MLL2, have roles in tumorigenesis remains to be determined.

HER2-amplified/overexpressed breast cancer is characterized by an aggressive phenotype and poor overall survival. Although the application of anti-HER2 therapy has improved the clinical outcome of HER2-positive breast cancers, primary and secondary resistance constitute major obstacles to the further success of such treatment strategy. As cyclin E amplification/overexpression in HER2-positive human breast cancers results in decreased sensitivity to the anti-HER2 agent trastuzumab, high levels of cyclin E in human HER2-positive breast cancer may be predictive of resistance to anti-HER2 therapy7,16,17. Pharmacological inhibition of cyclin E expression may therefore benefit HER2-positive breast cancer patients by delaying disease progression and/or preventing recurrence while they are receiving anti-HER2 therapy. A series of studies including this current report clearly demonstrate that Taspase1 cleaves MLL to promote cyclin E expression and this Taspase1-MLL-cyclin E axis is required for HER2/neu induced breast tumorigenesis22,26. Therefore, administering Taspase1 inhibitors in conjunction with anti-HER2 agents could produce therapeutic benefits for HER2-positive breast cancer patients. As proteases are drug targets, small-molecule inhibitors of Taspase1 may be developed for cancer therapy.

Lines of evidence indicate the active participation of Taspase1 in tumorigenesis, and thus support the development of small molecule inhibitors of Taspase1 for potential cancer therapy. However, caution should be exercised while exploiting Taspase1 inhibition as a therapeutic means in treating human subjects. First and foremost, the severe perinatal lethality resulting from the embryonic loss of Taspase1 suggests that inactivation of Taspase1 by genetic or pharmacological means is inadvisable in pregnant females and children in order to avoid potential developmental sequelae22. Nevertheless, inactivation of Taspase1 in fully developed adult mammals appears to be well-tolerated29. Cancer commonly hijacks key developmental pathways during tumorigenesis and thus frequently exhibits unique properties such as stem cell-like and dedifferentiated states48, which may underlie the preferential therapeutic benefit conferred by targeting Taspase1 to treat cancers.

Primitive Taspase1 inhibitors (TASPINs) were designed and discovered, lending support to developing highly effective, specific Taspase1 inhibitors for cancer therapy29. Interestingly, although mammary gland–specific knockout of Taspase1 disrupts MMTV-neu–driven breast tumorigenesis, the same genetic deletion of Taspase1 in MMTV-wnt;MMTV-cre;Tasp1F/− mice did not deter MMTV-wnt–driven breast carcinogenesis (Figure 2C). These data highlight the various mechanisms underlying individual tumorigenesis and the importance of selecting responsive cancers that might benefit from the treatment with TASPINs. Although it is beyond the scope of this study, it is of our interests to discover biomarkers that are predictive of Taspase1 addiction. Based on our unpublished data, Taspase1 has a very long protein half-life and is likely under multilayers of regulations in addition to transcription. Nevertheless, we analyzed the data of Taspase1 expression in HER2-positive and -negative tumors from breast TCGA dataset, and did not observe significant differences in Taspase1 expression (Supplementary information, Figure S5). Since Taspase1 is important in many aspects of cancer biology, Taspase1 inhibitors likely will benefit patients with different cancer types. Further studies with regard to the involvement of Taspase1 in various oncogenic pathways and the pathogenesis of subtypes of cancer could guide selection of cancer patients who would benefit from the inhibition of Taspase1.

Materials and Methods

Animal studies

All animal work was performed in accordance with MSKCC guidelines and IACUC approval. Mice were monitored for tumors by palpation twice a week. Tumor free Kaplan-Meier survival was calculated using MedCalc analysis software.

Cell culture, knockdown, proliferation, cell cycle, cell death, and western blot assays

BT-474 and HCC1419 cell lines were obtained from American Type Culture Collection and cultivated for no more than 2 months after each frozen aliquot was thawed. Amphotropic retrovirus carrying Taspase1 specific knockdown hairpin was generated as described26,29. To assay cell proliferation, 1 × 105 cells were seeded onto each well of a 6-well plate and counted 4 days later. Cell cycle and cell death analyses were performed as described26. For western blot, cells and tissues were lysed in standard RIPA buffer. The anti-Taspase1 rabbit polyclonal antibody is as described22,26. Antibodies for cyclin E2 (4132, Cell Signaling), cyclin A (C4710, Sigma), p21 (sc-397, Santa Cruz Biotechnology), p27(sc-528, Santa Cruz Biotechnology), cyclin D1(sc-450, Santa Cruz Biotechnology), p16(554079, BD Pharmingen), and ErbB2(OP-15, Calbiochem) were purchased from indicated companies. Antibodies were detected using the enhanced chemiluminescence method (Western Lightning, PerkinElmer). Immunoblot signals were acquired with the LAS-3000 Imaging system (FujiFilm) and were analyzed with ImageJ software.

Knockdown resistant Taspase1

Knockdown resistant Taspase1 (RT1) was generated by site-directed mutagenesis to create synonymous mutations at amino acids 387 to 392 (GGA AAG GCG AAA ACG CAT) of Taspase1. The cDNA was inserted into an MSCV-neo expression construct.

Soft agar assay

1 × 105 cells were seeded onto a 6 cm dish containing a top layer of 0.3% noble agar and a bottom layer of 0.6% noble agar base. Cells were fed with media every 3 days. After 3 weeks, colonies with diameter larger than 200 μm were scored. Three independent triplicate experiments were performed.

Quantitative RT-PCR

mRNA was isolated using TRIzol (Invitrogen) and further purified using the RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol. Reverse transcription reactions were carried out as described with Superscript II (Invitrogen) and random decamer primers (Ambion)20,22,34. Quantitative RT-PCR was performed using TaqMan 2× buffer and an ABI Prism 7300 sequence detection system (Applied Biosystems). For cyclin E1 and E2 reactions, the TaqMan Hs01026536 and Hs00180319 probes were used, and for cyclin A, SYBR Green master mix and the following primers were used: CAA AGC ACC ACA GCA TGC ACA AC and GAT TTA GTG TCT CTG GTG GGT TGA GG. All reactions were normalized against 18s rRNA using an 18s rRNA TaqMan probe (Applied Biosystems).

Mammary gland whole-mount assays

Mouse mammary glands were surgically dissected, spread onto a glass slide, and fixed in a 1:3:6 mixture of glacial acetic acid/chloroform/100% ethanol. Following hydration, they were stained overnight in 0.2% carmine and 0.5% ALK(SO4)2; dehydrated in graded solutions of ethanol, cleared in xylenes; and mounted with Permount.

Immunohistochemistry

Tissue samples were fixed in 4% paraformaldehyde and embedded in paraffin. Sections of 5-μm thickness were prepared. pH3S10 was detected by immunohistochemistry using the antibody from Millipore (06-570).

Statistical analysis

Student's t-test was performed to compare means between two groups. Data were expressed as the mean ± SD or ± sem as indicated.

Taspase1 gene expression in human breast cancer

To determine whether Taspase1 (Tasp1) expression differs in HER2 positive and HER2 negative human breast cancer, normalized RNA sequencing (RNA-Seq) data produced by The Cancer Genome Atlas (TCGA)49 were downloaded from Broad GDAC Firehose. PAM50 subtype classifications were available for 500 of the 526 primary breast invasive carcinoma tumor samples TCGA subjected to mRNA expression profiling using the Illumina HiSeq 2000 RNA Sequencing Version 2 platform. The subtype classifications were obtained through cBioPortal for Cancer Genomics, and the 26 samples lacking classifications were discarded. The Tasp1 mRNA expression values of the remaining samples formed a dataset that was partitioned by HER2 status to form HER2 positive and HER2 negative datasets. Two-tailed Mann-Whitney tests were performed to compare the means of the HER2 positive and HER2 negative datasets.