MIP-2A Is a Novel Target of an Anilinoquinazoline Derivative for Inhibition of Tumour Cell Proliferation

We recently identified a novel anilinoquinazoline derivative, Q15, as a potent apoptosis inducer in a panel of human cancer cell lines and determined that Q15 targets hCAP-G2, a subunit of condensin II complex, leading to abnormal cell division. However, whether the defect in normal cell division directly results in cell death remains unclear. Here, we used an mRNA display method on a microfluidic chip to search for other Q15-binding proteins. We identified an additional Q15-binding protein, MIP-2A (MBP-1 interacting protein-2A), which has been reported to interact with MBP-1, a repressor of the c-Myc promoter. Our results indicate that Q15 inhibits the interaction between MIP-2A and MBP-1 as well as the expression of c-Myc protein, thereby inducing cell death. This study suggests that the simultaneous targeting of hCAP-G2 and MIP-2A is a promising strategy for the development of antitumor drugs as a treatment for intractable tumours.


Introduction
Globally, tumour formation is the leading cause of death in developed countries and the second most common cause of death in developing countries [1]. Despite improvements in the treatment of various common tumours, such as the combined use of chemotherapy and chemoradiation, colorectal tumours, lung tumours and multiple myeloma (a hematopoietic tumour) remain particularly intractable. Thus, the development of potent drugs is required to treat such intractable tumours.
We recently used compound screening to identify an anilinoquinazoline derivative, Q15 (Fig. 1A), as a novel compound showing potent antitumor activity [2]. Additionally, using an mRNA display method [3][4][5], we identified hCAP-G2, a subunit of condensin II as a Q15-binding protein and confirmed that Q15 binds to the condensin II complex; accordingly, hCAP-G2 may affect chromosomal segregation in mitosis, leading to abnormal cell division and cell death [2]. However, it remains controversial whether only abnormal cell division caused by inhibition of condensin II leads to cell death.
In this present study, we aimed to search for additional targets of Q15 by using an mRNA display method in a microfluidic system, which is highly efficient for the selection of drug-binding proteins [6]. The use of the microfluidic chip also reduces the false-negative rate due to its low background. This feature made it possible to find another Q15-binding protein that was not detected in the previous system.
Consequently, we identified MIP-2A (MBP-1 interacting protein-2A) as a Q15-binding partner. MIP-2A has been identified as a MBP-1 binding protein using a yeast two-hybrid system. Prior to this, MBP-1 had been reported as a transcriptional repressor of c-Myc, binding to a TATA-box of the c-myc P2 promoter. Overexpression of exogenous MBP-1 leads to reduced c-Myc expression and cell death [7]. As c-Myc is a proto-oncogene product that plays a major role in the control of cell proliferation, MBP-1 exerts a regulatory effect on cell growth through regulation of c-Myc expression. Interaction of MIP-2A with MBP-1 inhibits the c-Myc repressor activity of MBP-1 [8]. We further confirmed that Q15 inhibits the interaction between MIP-2A and MBP-1 and thereby induces cell death by repressing the expression of c-Myc. Q15 may induce cell death by targeting both condensin II and MIP-2A.

Results
In vitro selection of Q15-binding proteins by mRNA display We performed an mRNA display experiment to identify Q15binding proteins (Fig. 1B). We first prepared a cDNA library derived from human multiple myeloma KMS34 cells, which are highly sensitive to Q15. From the cDNA library, we prepared mRNA-protein conjugates followed by affinity selection on biotinylated Q15 (Fig. 1C) immobilised on a microfluidic chip. After four rounds of selection, we analysed 18 clones. Among the candidates, we found that only MIP-2A 1-66 bound to Q15 (Fig. 1D), while the other candidates did not. Also, we focused on the protein MIP-2A because it has been reported to be involved in apoptosis. Furthermore, the putative Q15-binding region MIP-2A   (Fig. 1E), determined using the mRNA display method, is essential for interacting with myc-binding protein 1 (MBP-1) [7][8][9][10][11][12], which represses the transcription of c-Myc (Fig. 1F). Therefore, we hypothesised that Q15 inhibits the interaction between MIP-2A and MBP-1, thereby down-regulating the expression of c-Myc and leading to tumor cell death.

Q15 inhibits the interaction between MIP-2A and MBP-1
To verify whether MIP-2A binds directly to Q15, we performed a kinetic analysis to evaluate the interaction between MIP-2A and Q15. We prepared MIP-2A recombinant protein in an E. coli expression system. The soluble fraction was treated with nickel affinity resin and then purified by gel-filtration chromatography on a Superdex 75 column ( Fig. 2A, left). We then performed surface plasmon resonance analysis in which biotinylated Q15 was immobilised on an SA sensor chip. As a result, we found that MIP-2A binds to Q15 with a K D value of 3.8610 27 M ( Fig. 2A,  right).
We subsequently examined whether Q15 inhibits the interaction between MIP-2A and MBP-1, an interactor of MIP-2A as described above. We performed an in vitro assay of T7-MIP-2A binding to GST-MBP-1 immobilised on glutathione sepharose beads in the presence of 0-10 mM free Q15. Western blot analysis of each bead fraction showed that 1 mM Q15 was sufficient to disrupt the complex, and T7-MIP-2A was bound to GST-MBP-1 Step 1: A cDNA library derived from KMS34 cells was transcribed and then ligated with a PEG-Puro spacer.
Step 2: The resulting mRNA was in vitro translated to form a library of protein-mRNA conjugates.
Step 3: The library was injected into a microfluidic chip on which Q15 was immobilised, and unbound molecules were washed away.
Step 4: The bound molecules were eluted, and their mRNA portion was amplified by RT-PCR. The resulting DNA was used for the next round of selection and analysed by cloning and sequencing. (C) Chemical structure of biotinylated Q15. (D) The selected T7-MIP-2A 1-66 was generated by in vitro translation and used in a pull-down assay with biotinylaed Q15 immobilized on streptavidin beads. Each fraction was separated by gel electrophoresis using 4-12% Bis-Tris Gel, followed by western blot analysis using an antibody against T7 tag.

Q15 increases the intranuclear localisation of MBP-1
MBP-1 mainly localises in nucleus and binds to the c-Myc P2 promoter to repress its transcription [9]. However, nuclear entry and the transcriptional activity of MBP-1 are prevented by MIP-2A [12]. Because Q15 inhibits the interaction between MIP-2A and MBP-1, as shown in Fig. 2B, the intranuclear localisation of MBP-1 should be increased by Q15. To evaluate this effect, we examined the cellular localisation of MBP-1 using immunofluorescence staining. HeLa cells expressing FLAG-MBP-1 with or without HA-MIP-2A were treated with 0 or 5 mM Q15 for 24 h followed by staining with anti-FLAG and anti-HA antibodies ( Fig. 3A

Q15 down-regulates the expression of c-Myc
MBP-1 is reported to be a repressor of c-Myc [9], and thus the enhanced intranuclear localisation of MBP-1 should lead to the down-regulation of c-Myc. To evaluate this effect, we examined whether Q15 affects the expression of c-Myc at the mRNA or protein level. We first analysed the mRNA level of c-Myc in HeLa cells treated with 0 or 5 mM Q15 for 24 h. RT-PCR analyses showed that when cells were treated with Q15, the c-Myc mRNA level decreased (Fig. 4A). To assess the expression level of c-Myc protein, HeLa cells were treated with 0250 mM Q15 for 24 h followed by western blotting using anti-PARP or anti-c-Myc antibodies (Fig. 4B). As we have previously reported, the cleavage of PARP was detected by treatment with Q15 [2]. Furthermore, the c-Myc protein level also decreased in a Q15 concentrationdependent manner and correlated with the extent of PARP cleavage. These results suggest that the increase of the intranuclear localisation of MBP-1 caused by Q15 results in the downregulation of c-Myc at the both mRNA and protein level.    (Fig. 5A). Knockdown of MIP-2A also resulted in cell death (Fig. 5B). When c-Myc expression was suppressed with siRNA in HeLa cells, PARP was cleaved, as in the case of treatment with Q15 (Fig. 5C), suggesting that a decrease of c-Myc leads to apoptosis. Moreover, the cell viability decreased to less than 50% (Fig. 5D). These results suggest that inhibition of MIP-2A results in a decrease of c-Myc expression, leading to cell death.

Discussion
In our recent study, Q15 was identified as a novel anilinoquinazoline derivative that represses the proliferation of several types of tumour cell lines, including intractable tumours [2]. Q15 showed a more potent activity than gefitinib [13], a wellestablished anilinoquinazoline derivative with a potent activity against non-small-cell lung tumours. Furthermore, we also reported that Q15 interacts with hCAP-G2, a subunit of the condensin II complex, and thereby induces a structural aberration of the chromosome in mitosis, leading to abnormal cell division [2]. However, whether such a defect in normal cell division is directly linked to apoptosis remains to be determined. For example, it is plausible that Q15 may target other proteins for the induction of apoptosis of tumour cells.
In this study, we performed a selection of Q15-binding proteins using an mRNA display method on a microfluidic chip. This approach has several advantages over selection on beads, including a lower non-specific binding and higher efficiency of enrichment [6], and was used to identify a drug-binding protein in a previous study [14]. Using this mRNA display selection system, we identified MIP-2A as an additional Q15-binding protein.
Unexpectedly, hCAP-G2, which was identified in 6 of 100 clones in the previous selection, was not obtained, possibly because we analysed a smaller number of clones in this present study.
Several studies have reported that MIP-2A interacts with MBP-1, which acts as a repressor of the c-Myc promoter and thereby serves as an apoptosis inducer [8,9]. We found that Q15 inhibits the interaction between MIP-2A and MBP-1, leading to a decrease in the c-Myc expression level. It has also been reported that depression of the proto-oncoprotein c-Myc by knockdown results in apoptosis [9]. However, in several cell lines, variations in the c-Myc protein levels do not affect the cell viability [15]. In our study, we confirmed that the knockdown of c-Myc in HeLa cells leads to apoptosis. Although cell viability decreased to below 50% after the knockdown of c-Myc, only a small percentage of PARP was cleaved. We think these results may indicate that inhibition of c-Myc leads to not only apoptosis but also lower cell proliferation. Moreover, cleavage of PARP was detected in the c-Myc knockdown trial, it was not detected in the case of knockdown of MIP-2A (data not shown), possibly because of insufficient repression of c-Myc due to the low knockdown efficiency of MIP-2A.
As shown in Figure S1, we found a negative correlation (correlation coefficient = 20.62) between expression level of c-Myc and sensitivity to Q15 in several cell lines we previously tested [2].  [16][17][18][19]. We speculate that Q15 may show potent activity against other cell lines that are addicted to c-Myc.  By selecting Q15-binding proteins using an mRNA display method, we identified MIP-2A 1-66 as a putative Q15-binding region. Based on previous studies using X-ray crystallographic analysis [10,11], we determined that MIP-2A 1-66 corresponds to a part of the region responsible for interacting with other proteins including MBP-1. In addition to MBP-1, MIP-2A has also been reported to bind to and modulate the transcriptional activity of the orphan nuclear receptor SF-1 and homeodomain protein PITX1 [12]. Although Q15 may prevent the interaction of MIP-2A with these proteins as well as with MBP-1, the relationship between SF-1 and PITX1 and cell growth remains elusive.
As shown in Fig. 2B, double bands of GST-MBP-1 were detected after in vitro translation, suggesting that post-translational modification occurs. Previous findings by Sedoris et al. suggested that post-translational modification of MBP-1 affects its DNAbinding activity [20]. Although such modification, if indeed it occurs, may affect the interaction between MBP-1 and MIP-2A, further study will be needed to examine this issue.
In conclusion, we determined that Q15 binds not only to hCAP-G2 but also MIP-2A. Our data indicate that the inhibition of the interaction between MIP-2A and MBP-1 leads to a decrease in the expression of c-Myc protein, resulting in lower cell viability. Further analyses such as ChIP to examine directly whether binding of MBP-1 to c-myc P2 promoter is affected by Q15 would provide further support for our hypothesis. Either induction of abnormal cell division via the inhibition of hCAP-G2 or suppression of c-Myc via the inhibition of MIP-2A may be sufficient for the direct induction of cell death. Regardless of mechanism, Q15 clearly targets both hCAP-G2 and MIP-2A, resulting in the inhibition of tumour cell growth. Therefore, the simultaneous targeting of these proteins may be a promising strategy for the treatment of intractable tumours.

mRNA display selection for Q15-binding proteins
The affinity selection of target proteins of Q15 was performed as previously described [14] with modification of the bait drug.

Preparation of plasmids
All primers used in this study are listed in Table 1 The cDNA of MBP-1 was amplified from total RNA of KMS34 cells by RT-PCR using HA-MBP1-f and MBP1-FLAG-His6-polyA-stop-r or BamHI-MBP1-f and MBP1-HindIII-r, and the PCR products were cloned into plasmid vector pCR3.3-TOPO or pCMV-tag2A (Stratagene, Santa Clara, CA, USA), respectively. From the resulting plasmid, the MBP-1 coding DNA was amplified by PCR using MBP1-f and MBP1-FLAG-His6-polyAstop-r. The PCR product was mixed with a GST-coding DNA fragment amplified by PCR using O29-GST-f and GST-MBP1-r. The mixture was used as a template for overlap-extension PCR and was subsequently subcloned into pCR3.3-TOPO.

Surface plasmon resonance analysis
The recombinant MIP-2A protein was prepared as follows. Escherichia coli strain BL21 (DE3) codon+ was transformed with pET15b-T7-MIP2A-FLAG-His66. The cells were grown in LB at 37uC. When the OD 600 reached 0.5-0.6, 1 mM IPTG was added, and the cells were incubated for an additional 6 h. The culture was centrifuged at 20,000 g, for 5 min at 4uC. The pellet was lysed using lysis buffer (50 mM Tris-HCl, pH 7.6, 200 mM NaCl) containing protease inhibitor cocktail (Sigma, St. Louis, MO, USA) and homogenised by sonication. The homogenate was centrifuged at 6,000 g for 20 min at 4uC, and the supernatant was collected as the soluble fraction. The soluble fraction was mixed with Cosmogel His-Accept (Nacalai Tesque, Kyoto, Japan) on a rotator for 2 h at 4uC. After removal of the supernatant, the beads were washed with lysis buffer containing 20 mM imidazole, and the protein was then eluted with 300 mM imidazole. The resulting eluates were separated by gel-filtration chromatography using a Superdex 75 column (GE Healthcare, Waukesha, WI, USA) with AKTA (GE Healthcare).
Binding kinetics were determined by surface plasmon resonance (SPR) analysis using a Biacore 3000 system (GE Healthcare). All experiments were performed at 25uC using TBS buffer (20 mM Tris-HCl, pH 7.5, 138 mM NaCl). Biotinylated Q15 was immobilised onto the SA sensor chip (GE Healthcare). The measurements were performed using 392.6 resonance units of the ligand and a flow rate of 20 ml/min. To determine the dissociation constants, four different concentrations of purified MIP-2A were injected. The injection period for association was 300 s. After each measurement, the chip surface was regenerated with 15 ml of Glycine 2.0 (GE Healthcare). The binding data were analysed with the steady-state affinity model using BIAevaluation software version 4.1 (GE Healthcare).