Access the full text.
Sign up today, get DeepDyve free for 14 days.
doi:10.1093/jmcb/mjy002 Journal of Molecular Cell Biology (2018), 10(3), 195–204 j 195 Published online March 2, 2018 Article MINAR1 is a Notch2-binding protein that inhibits angiogenesis and breast cancer growth 1,† 1,† 2 1 1 Rachel Xi-Yeen Ho , Rosana D. Meyer , Kevin B. Chandler , Esma Ersoy , Michael Park , Philip 1 1 1 2 1, A. Bondzie , Nima Rahimi , Huihong Xu , Catherine E. Costello , and Nader Rahimi Department of Pathology, School of Medicine, Boston University Medical Campus, Boston, MA 02118, USA Department of Biochemistry and Center for Biomedical Mass Spectrometry, School of Medicine, Boston University Medical Campus, Boston, MA 02118, USA These authors contributed equally to this work. * Correspondence to: Nader Rahimi, E-mail: email@example.com Edited by Haian Fu Intrinsically disordered proteins (IDPs)/intrinsically unstructured proteins are characterized by the lack of ﬁxed or stable tertiary structure, and are increasingly recognized as an important class of proteins with major roles in signal transduction and transcrip- tional regulation. In this study, we report the identiﬁcation and functional characterization of a previously uncharacterized protein (UPF0258/KIAA1024), major intrinsically disordered Notch2-associated receptor 1 (MINAR1). While MINAR1 carries a single transmembrane domain and a short cytoplasmic domain, it has a large extracellular domain that shares no similarity with known protein sequences. Uncharacteristically, MINAR1 is a highly IDP with nearly 70% of its amino acids sequences unstructured. We demonstrate that MINAR1 physically interacts with Notch2 and its binding to Notch2 increases its stability and function. MINAR1 is widely expressed in various tissues including the epithelial cells of the breast and endothelial cells of blood vessels. MINAR1 plays a negative role in angiogenesis as it inhibits angiogenesis in cell culture and in mouse matrigel plug and zebraﬁsh angio- genesis models. Furthermore, while MINAR1 is highly expressed in the normal human breast, its expression is signiﬁcantly down- regulated in advanced human breast cancer and its re-expression in breast cancer cells inhibited tumor growth. Our study demonstrates that MINAR1 is an IDP that negatively regulates angiogenesis and growth of breast cancer cells. Keywords: intrinsically disordered proteins, Notch2, angiogenesis, breast cancer, MINAR1 Introduction cascade and dissociate quickly. These unique properties are crit- Intrinsically disordered proteins (IDPs) are increasingly recog- ically important for highly complex processes of signal transduc- nized as an important class of proteins with major biological func- tion and transcriptional regulation (Gsponer and Babu, 2009). tions. It was estimated that nearly 30% of proteins in eukaryotic Angiogenesis, the formation of blood vessels, is a multistep cells lack a unique 3D structure (Gsponer and Babu, 2009), and highly organized cellular process, which is largely initiated underscoring their ubiquitous functional importance in the biol- by the function of an intrinsically discorded transcription factor, ogy of eukaryotic cells. Due to their intrinsically disordered hypoxia-inducible transcription factor1α (HIF1α)(Dames et al., nature, IDPs are dynamic and versatile. They interact with other 2002; De Guzman et al., 2005). HIF1α activates transcription of proteins that are involved in key regulatory cellular events such critical proteins such as vascular endothelial growth factor as signal transduction and transcriptional regulation (Wright and (VEGF), which is a key culprit of pathological and physiological Dyson, 2015; Chavali et al., 2017). It has been proposed that dis- angiogenesis (Rahimi, 2006a, b, 2012). Through its intrinsically ordered sequences are essential for the function of transcrip- disordered carboxyl-terminal activation domain (CAD), HIF1α tional activators and copious other cell signaling proteins (Dyson associates with transcriptional coactivator protein (CREB)-bind- and Wright, 2005). As IDPs have greater ﬂexibility to interact with ing protein (CBP)/p300 to activate transcription of VEGF (Dames other proteins compared with the globular proteins, they rapidly et al., 2002). Initiation and maturation of angiogenesis is gov- form complexes with other proteins, and thus initiate a signaling erned by a myriad of cell surface receptors such as VEGF recep- tors (VEGFR-1 and VEGFR-2), Tie and Notch receptors. Received September 5, 2017. Revised January 4, 2018. Accepted January 5, 2018. The Notch pathway is a highly evolutionary conserved inter- © The Author (2018). Published by Oxford University Press on behalf of Journal of Molecular Cell Biology, IBCB, SIBS, CAS. All rights reserved. cellular signaling system and is activated by the interaction of Downloaded from https://academic.oup.com/jmcb/article-abstract/10/3/195/4796960 by Ed 'DeepDyve' Gillespie user on 26 June 2018 196 j Ho et al. transmembrane ligands, the Delta (Delta-like1 [Dll-1], Dll3, and Williams et al., 2006). Notch receptors also play major roles in Dll4) families and Jagged (Jagged1 and Jagged2) with Notch human cancers. They can function as either an oncogene or a receptors (Notch1−4), which are usually expressed on the sur- tumor suppressor in human cancers. For example, Notch2 face of neighboring cells. Ligand binding induces cleavage of expression in breast cancer is associated with better survival, the Notch receptors and subsequent nuclear translocation of and high expression of Notch2 is correlated with well- the Notch intracellular domains (NICDs). In the nucleus, NICD differentiated breast tumors (Parr et al., 2004). binds to multiple DNA-binding proteins (Lu and Lux, 1996; Here, we report the identiﬁcation and functional characteriza- Andersson et al., 2011; Chillakuri et al., 2012). Notch signaling tion of a previously unknown protein, MINAR1. We demonstrate plays a delicate role in angiogenesis. In the initial stages of that MINAR1 is as an IDP, and have identiﬁed Notch2 as a angiogenesis, Notch activation is generally suppressed to allow MINAR1-binding protein. MINAR1 binds to Notch2 and gains sta- endothelial cells to proliferate in response to VEGF stimulation, bility and function. MINAR1 activity inhibits proliferation of and its expression is subsequently upregulated when endothe- breast cancer cells and angiogenesis and its expression is down- lial cells stop proliferating and the vessels begin to stabilize regulated in human advanced breast cancer. (Henderson et al., 2001; Taylor et al., 2002). Considering its overall function in angiogenesis, the Notch signaling pathway Results plays a negative role in angiogenesis. Interfering with the Notch Identiﬁcation of major intrinsically disordered Notch2-binding pathway in a mouse resulted in the development of vascular receptor 1 tumors and lethal hemorrhage (Liu et al., 2011) and its activity We have recently identiﬁed multiple previously uncharacter- in cell culture inhibited the angiogenic functions of endothelial ized cell surface receptors through an in silico analysis of cells such as capillary tube formation, migration, and prolifer- human genome (Rahimi et al., 2012; Arafa et al., 2015; Wang ation (Leong et al., 2002; Itoh et al., 2004; Noseda et al., 2004; et al., 2016). Major intrinsically disordered Notch2-binding Figure 1 Identiﬁcation of MINAR1 as an IDP. (A) The amino acid sequence of human MINAR1. Amino acids boxed in the red correspond to disordered segments of MINAR1. The data were generated using online DISOPRED3 (Disorder Prediction) program. (B) The graph is the representation of the data shown in A.(C) The Kyte–Doolittle hydrophobicity score of MINAR1. The graph was generated using online pro- gram http://web.expasy.org/protscale/.(D) PAE cells ectopically expressing empty vector (EV) or MINAR1 were lysed with lysis buffer con- taining Triton X-100. Whole-cell lysates (WCL) were centrifuged and Triton X-100 soluble fraction was separated from the insoluble fraction. The remaining insoluble fraction was further solubilized with 1% SDS. Both fractions were then blotted on cellulose acetate ﬁlter via a dot blot apparatus. The same cell groups were also homogenized in PBS plus 1% SDS and similarly blotted on cellulose acetate ﬁlter. The cellu- lose acetate ﬁlter was blotted for the presence of MINAR1 using anti-MINAR1 antibody (the MINAR1/PAE cell lysate group loaded in dupli- cate). (E) Western blot of the cell lysates from EV/PAE and MINAR1/PAE. Data in D and E are representative of at least three independent experiments. Downloaded from https://academic.oup.com/jmcb/article-abstract/10/3/195/4796960 by Ed 'DeepDyve' Gillespie user on 26 June 2018 MINAR1 regulates angiogenesis and breast tumor growth j 197 receptor 1 (MINAR1), which is encoded by a previously unchar- disorder-promoting residues (Pro, Arg, Gly, Gln, Ser, Glu, Lys, and acterized gene KIAA1024 located in chromosome 15 (GRCh38. Ala) (Campen et al., 2008). Consistent with these characteristics, p7), was among these putative cell surface receptors. MINAR1 MINAR1 is considerably low in the order-promoting amino acids consists of 916 amino acids with a predicted molecular weight and high in the disorder-promoting amino acids (Supplementary of 103 kDa, and is composed of a large extracellular domain with Figure S2). Furthermore, our analysis showed that majority of multiple potential glycosylation sites, and a single transmembrane MINAR1 amino acids in Kyte–Doolittle scale are at the negative domain followed by a short intracellular domain (Supplementary range (Figure 1C). Analysis of MINAR1 aminoacidsequences via Figure S1A). MINAR1 is highly conserved among the species ran- the online disorder prediction program DISOPRED3 (http://bioinf. ging from human to rodents and Xenopus (Supplementary cs.ucl.ac.uk/psipred/?disopred=1) revealed that nearly 70%of Figure S1B) and the sequence similarity of the human and mouse MINAR1 is intrinsically disordered (Figure 1A and B). Similarly, MINAR1 is 96.5%(Supplementary Figure S1C), suggesting a other programs such as DISOclust (McGufﬁn, 2008; Roche et al., potentially evolutionarily conserved function for MINAR1. 2011), Protein Homology/analogY Recognition Engine V 2.0 (Phyre2) Uniquely, MINAR1 appeared to be a highly IDP, as its instabil- (data not shown), and Metadisorder, which calculates disorder con- ity index was 58.24 (Guruprasad et al., 1990; Wilkins et al., sensus from the results generated by 13 major different disorder 1999). Generally, proteins with higher stability have an instabil- programs such as DisEMBL, DISOPRED2, DISpro, Globplot, iPDA, ity index of less than 40, whereas unstable proteins have value IUPred, Pdisorder, Poodle-l, PrDOS, and Spritz, predicated MINAR1 above 40 (Guruprasad et al., 1990). IDPs commonly have a low- as a highly IDP (Supplementary Figure S3). er content of the order-promoting amino acids such as Cys, Trp, Intrinsically disordered or unstructured proteins tend to Tyr, Phe Ile, Leu, Val, and Asn, whereas they are enriched in the aggregate and are typically insoluble (Hazeki et al., 2000). Figure 2 MINAR1 is a cell surface receptor and is widely expressed in human organs and tissues. (A) HEK-293 cells expressing MINAR1-Myc were ﬁxed in 4% PFA and stained with anti-MYC or anti-MINAR1 antibody. Pictures were taken under immunoﬂuorescence microscope (40×). Expression of MINAR1 in HEK-293 cells is shown. (B) HEK-293 cells expressing GFP vector alone (pAcGFP1-C3) or GFP-MINAR1 were viewed under immunoﬂuorescence microscope and pictures were taken (40×). Expression of MINAR1 in HEK-293 cells is shown. (C) Human cornea, breast, colon, and cytoskeletal tissues were subjected to immunohistochemistry analysis using anti-MINAR1 antibody. Slides were scanned and corresponding pictures are shown (40×). (D) Proteins extracted from various human organs were subjected to western blot analysis using anti-MINAR1 antibody or anti-tubulin as a loading control. Downloaded from https://academic.oup.com/jmcb/article-abstract/10/3/195/4796960 by Ed 'DeepDyve' Gillespie user on 26 June 2018 198 j Ho et al. Therefore, we examined the tendency of aggregation of MINAR1 MINAR1 with a chemical chaperone 4-phenylbutyrate (4BPA) via the cellulose acetate ﬁlter trap assay, which is commonly increased the expression of MINAR1 (Supplementary Figure S4A) and used to study protein aggregation (Scherzinger et al., 1997). decreased its sensitivity to partial trypsin digestion (Supplementary Initially, we lysed PAE cells ectopically expressing MINAR1 with Figure S4B), demonstrating that in the presence of 4BPA, MINAR1 a lysis buffer containing Triton X-100 and SDS and soluble and gains a partial secondary structure, and thus exhibits a reduced sus- insoluble protein fractions were subjected to cellulose acetate ﬁl- ceptibility to degradation by partial trypsin digestion. ter trap assay. MINAR1 was detected only in the Triton X-100/SDS insoluble fraction (Figure 1D). However, when cells were homoge- MINAR1 is a cell surface receptor that is expressed in various nized and whole-cell homogenate was applied to the cellulose human organs and tissues acetate ﬁlter trap assay, MINAR1 was retained on the cellulose MINAR1 was predicated to be a cell surface protein (Figure 1). acetate ﬁlter (Figure 1D), indicating that MINAR1 forms protein Therefore, we asked whether MINAR1 is expressed as a cell sur- aggregates due to its intrinsically disordered characteristic. face protein. When expressed in HEK-293 cells, Myc-tagged Intrinsically discorded proteins usually have short half-lives and MINAR1 (MINAR1-Myc) was detected as a cell surface protein, are generally susceptible to degradation due to misfolding or as both anti-MINAR1 and anti-Myc antibodies located MINAR1 at lack of overall structure, which leaves them more accessible to the membranous region in HEK-293 cells (Figure 2A). Similarly, proteases. Treatment of HEK-293 cells endogenously expressing ectopic expression of GFP-tagged MINAR1 also showed that Figure 3 MINAR1 inhibits in vitro and in vivo angiogenesis. (A) PAE cells ectopically expressing empty vector (EV) or MINAR1 were subjected to in vitro angiogenesis assay and pictures were taken after 24 h. Pictures were taken and three randomly selected pictures from each group were selected and quantiﬁed using Image J program. (B) GPP-PAE cells expressing empty vector (EV) or MINAR1 were subjected to in vivo matrigel assay. Cells were mixed with growth factor-reduced matrigel and sub-dermally injected into mouse (three mice/group). After 7 days, the matrigel plugs were removed and were viewed under ﬂuorescent microscope after cryo ﬁxing and processing. The repre- sentative pictures are shown (40×). Expression of MINAR1 in PAE cells is shown. (C) HUVECs were transduced with an empty vector, pGIPZ, or two different MINAR1 shRNA. After 48 h, cells were subjected to in vitro angiogenesis assay. Cells were viewed under a ﬂuorescent micro- scope and pictures were taken. Quantiﬁcation of capillary tube formation was made by Image J program. Knockdown of MINAR1 is also shown. (D) Fli-eGFP transgenic ﬁsh embryos were injected with MINAR1 or LacZ mRNA at the 1-cell stage or 2-cell stage. The embryos were examined at 50 h post-fertilization (50 hpf) and representative immunoﬂuorescence images are shown. Quantiﬁcation of intersegmental ves- sels (ISV) and the dorsal lateral anastomosing vessel (DLAV) of 10 ﬁsh per group by ImageJ are shown in bar graph. Error bars represent SD. P = 0.007 for 10 ng MINAR1 mRNA compared with LacZ control. Western blot analysis of MINAR1 expression from tissue lysates of microinjected ﬁsh is performed with anti-MINAR1 antibody and protein loading control GFP. Downloaded from https://academic.oup.com/jmcb/article-abstract/10/3/195/4796960 by Ed 'DeepDyve' Gillespie user on 26 June 2018 MINAR1 regulates angiogenesis and breast tumor growth j 199 MINAR1 is expressed at the cell surface (Figure 2B). Furthermore, MINAR1 inhibits angiogenesis immunohistochemistry staining of human corneal epithelial cells Considering its expression in the endothelial cells, we sought showed that MINAR1 is expressed as a membrane-bound protein to examine possible biological activities of MINAR1 in angiogen- (Figure 2C). MINAR1 was also expressed in human colon, cyto- esis. To this end, we examined the effect of overexpression of skeletal muscle, breast, and endothelial cells of blood vessels MINAR1 in porcine aortic endothelial (PAE) cells. Overexpression (Figure 2C). Additionally, western blot analysis demonstrated of MINAR1 in PAE cells inhibited capillary tube formation in an that MINAR1 is highly expressed in heart, brain, liver, lung, skel- in vitro matrigel assay (Figure 3A). To corroborate the inhibitory etal muscle, and thymus (Figure 2D). Colon and placenta tissue effect of MINAR1 in angiogenesis further, we subjected PAE cells cell lysates also were positive for MINAR1, although at signiﬁ- to a mouse matrigel plug angiogenesis assay. To visualize the cantly lower levels (Figure 2D). The apparent molecular weight of tube formation of PAE cells, cells were also engineered to MINAR1 in human brain tissue was lower, which may be due to express GFP. Similar to in vitro matrigel assay, PAE cells expres- protein degradation. Furthermore, we used publically available sing MINAR1 did not form capillary tubes; instead, they grew in human genome data (http://biogps.org) to examine the expres- clusters (Figure 3B). Next, we silenced expression of MINAR1 in sion proﬁle of MINAR1. MINAR1 mRNA is widely present in the human umbilical vein endothelial cells (HUVECs) by two different majority of human organs; however, the highest of level of shRNAi and assessed their capillary tube formation. The knock- MINAR1 mRNA was present in heart, liver, skeletal muscle, car- down of MINAR1 by shRNA increased capillary tube formation diac myocytes, and adrenal cortex (Supplementary Figure S5). (Figure 3C). To further illustrate the in vivo function of MINAR1 in These results indicate that MINAR1 is a cell surface protein and angiogenesis, we tested the role MINAR1 in zebraﬁsh angiogen- widely expressed in human organs and tissues. esis. Microinjection of in vitro-translated human MINAR1 mRNA Figure 4 Expression of MINAR1 is downregulated in human advanced breast tumors and its expression in breast cancer cells inhibits tumor growth. (A) Cell lysates from human breast cancer cell lines, ZR-75-1,MDA-MB-231, and T47D, were prepared and subjected to western blot ana- lysis using anti-MINAR1 antibody or anti-GAPDH antibody for loading control. (B)MINAR1 expression in human invasive ductal carcinoma breast cancer with different tumor grades. MINAR1 expression was scored for staining strength as 0 (no staining), 1 (weak staining), 2 (moderate stain- ing), or 3 (strong staining) following a visual inspection of the cytoplasmic immunostaining for MINAR1. There is an inverse relation between MINAR1 expression and tumor grade (P < 0.05). (C) A tissue microarray consisting of 72 invasive ductal carcinoma tissues (16 cases of normal, 11 cases of grade 1, 32 cases of grade 2,and 12 cases of grade 3)for MINAR1 was carried out. Two pathologists assessed MINAR1 expression and tumor grades. The graph shows that MINAR1 expression is downregulated in advanced breast cancer (P < 0.05). (D)MDA-MB-231 cells expressing empty vector or MINAR1 were subjected to MTT assay. Expression of MINAR1 in MDA-MB-231 cells is shown. Downloaded from https://academic.oup.com/jmcb/article-abstract/10/3/195/4796960 by Ed 'DeepDyve' Gillespie user on 26 June 2018 200 j Ho et al. into 1-cell-stage zebraﬁsh embryos signiﬁcantly disrupted inter- breast cancer cell lines, including MDA-MB-231,T47D, and ZR-75-1 segmental vessels (ISV) and dorsal longitudinal anastomotic ves- cells for the expression of MINAR1 in western blot analysis. MINAR1 sel (DLAV) formation at 50 hpf (Figure 3D). Microinjection of LacZ was hardly detectable in these breast cancer cell lines (Figure 4A). mRNA, which was used as a control, had no effect on the blood This observation promoted us to examine the expression of vessel formation compared with control embryos with no injection MINAR1 in human in human breast cancer. IHC analysis showed (Figure 3D). Quantiﬁcation of the blood vessel formation (ISV and that MINAR1 is highly expressed in normal/benign human DLAV) in response to microinjection of human MINAR1 mRNA is breast (Figure 4B). However, the expression of MINAR1 was sig- shown (Figure 3D). Western blot analysis was performed to con- niﬁcantly reduced in poorly differentiated invasive ductal carcin- ﬁrm the overexpression of MINAR1 in microinjected ﬁshes oma (grade 3) of the breast (4 out 4 cases) (Figure 4B). (Figure 3D). Altogether, the data obtained from cell culture, Expression of MINAR1 in both well-differentiated (grade 1) and mouse, and zebraﬁsh angiogenesis assays all demonstrated that moderately differentiated (grade 2) breast carcinomas was also MINAR1 negatively regulates angiogenesis. moderately reduced (Figure 4B). To conﬁrm our observation, we stained a tissue microarray consisting of 72 invasive breast car- MINAR1 expression is downregulated in human advanced cinoma tissues (16 cases of normal, 11 cases of grade 1, 32 breast cancer cases of grade 2, and 12 cases of grade 3) for MINAR1 and two In addition to its expression in endothelial cells, MINAR1 is pathologists assessed MINAR1 expression and tumor grades. expressed in human epithelial cells of breast, colon, and other tis- The result showed that MINAR1 expression has an inverse rela- sues (Figure 2C). Therefore, we examined the possible role of tion with tumor grade (Figure 4C). The higher the tumor grade, MINAR1 in human breast cancer. We initially examined several the lower was the MINAR1 expression (Figure 4C). Figure 5 Identiﬁcation of Notch2 as a binding partner of MINAR1.(A) PAE cells expressing empty vector or Myc-tagged MINAR1 were lysed, immunoprecipitated with anti-Myc antibody, resolved in SDS-PAGE. The corresponding band was cut and subjected to proteolytic digestion followed by LC-MS/MS analysis of the tryptic peptides. Shown here is the MS2 higher energy collisional dissociation (HCD) spectrum corre- sponding to Notch2 peptide ‘MNDGTTPLILAAR’, labeled with assignments for the detected N-terminal b-ion (red) and C-terminal y-ion (blue) fragments. (B) HEK-293 cells expressing FLAG-tagged MINAR1 alone or transfected with Notch2 were lysed and immunoprecipitated with anti-FLAG antibody followed by immunoblotting with anti-Notch2 antibody. (C) Cell lysates from HUVECs were immunoprecipitated with anti- MINAR1 antibody or control IgG followed by immunoblotting with anti-Notch2 antibody. (D) HUVECs were transduced with two different MINAR1 shRNAs. After 48 h, cells were lysed and subjected to immunoprecipitation using anti-MINAR1 antibody, followed by immunoblot- ting with anti-Notch2 antibody. (E) PAE cells expressing MINAR1 were subjected to immunoﬂuorescence staining for MINAR1 and Notch2. (F) Whole-cell lysates (WCL) from HEK-293 cells expressing MINAR1 alone or co-expressing MINAR1 with Notch2 were subjected to a ﬁlter trap assay as in Figure 1 and immunoblotted with anti-MINAR1 antibody. WCL of HEK-293 cells expressing MINAR1 alone or co-expressing MINAR1 with Notch2 were subjected to western blot analysis and blotted for MINAR1, Notch2, and the loading control tubulin. Downloaded from https://academic.oup.com/jmcb/article-abstract/10/3/195/4796960 by Ed 'DeepDyve' Gillespie user on 26 June 2018 MINAR1 regulates angiogenesis and breast tumor growth j 201 Given its observed downregulation in breast cancer, we ecto- abolished the co-immunoprecipitation of Notch2 with MINAR1 pically expressed MINAR1 in MDA-MB-231 cells and assessed its (Figure 5D). To demonstrate the Notch2 and MINAR1 co-localization effect in the proliferation of MDA-MB-231 cells. Proliferation of in cells, we co-stained HUVECs with anti-MINAR1 and anti-Notch2 MDA-MB-231 cells expressing MINAR1 was signiﬁcantly inhib- antibodies. The result showed that MINAR1 was co-localized with ited as measured by MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphe- Notch2 (Figure 5E). These data further demonstrate that MINAR1 is nyltetrazolium bromide] assay (Figure 4D). Taken together, the a cell surface protein. To further corroborate the direct physical data suggest that MINAR1 expression is downregulated in binding of MINAR1 with Notch2, we generated GST fusion MINAR1 human breast cancer and its re-expression in breast cancer, encompassing the extracellular domain of MINAR1 (GST-E-MINAR1) MDA-MB-231 cells inhibits proliferation. without transmembrane and cytoplasmic domains (Supplementary Figure S6A and B). The puriﬁed GST-MINAR1 was used to examine MINAR1 binds to Notch2 and gains stability and function whether GST-E-MINAR1 could interact with Notch2-GFP ectopically The unstructured or disordered segments in proteins often expressed in HEK-293 cells in an in vitro GST pull-down assay. interact with other proteins and, in some cases, these regions The result showed that GST-E-MINAR1 binds to Notch2-GFP form complexes with other proteins to gain secondary structure (Supplementary Figure S6C), indicating that MINAR1 directly for functionality (Wright and Dyson, 1999; Tompa, 2002; interacts with Notch2. Uversky, 2002). Given the intrinsically disordered nature of Next, we examined whether Notch2 is required for stability of MINAR1, we hypothesized that MINAR1 could associate with MINAR1. Accordingly, we expressed MINAR1 alone or with other membrane-bound receptors or soluble proteins to gain Notch2 in HEK-293 cells. Cell homogenates were prepared and structure and function. To identify possible MINAR1-binding pro- subjected to the cellulose acetate ﬁlter trap assay. The result teins, we puriﬁed MINAR1 via immunoprecipitation from porcine showed that co-expression of MINAR1 with Notch2 signiﬁcantly aortic endothelial (PAE) cells ectopically expressing MINAR1 and reduced aggregation of MINAR1, as was evident by the decrease analyzed the immunoprecipitated proteins by liquid chromatog- in retention of MINAR1 on the cellulose acetate ﬁlter (Figure 5F). raphy–tandem mass spectrometry (LC-MS/MS). LC-MS/MS ana- Western blot analysis of soluble MINAR1 and Notch2 expression is lysis identiﬁed Notch2 as a putative MINAR1-binding protein shown (Figure 5F). To assess whether the apparent effect of (Figure 5A). We conﬁrmed the direct binding of MINAR1 with Notch2 in the stabilization of MINAR1 also affects the biological Notch2 in HEK-293 cells by co-transfection of MINAR1-FLAG and activity of MINAR1, we used a well-characterized recombinant full-length Notch2. Cell lysates were immunoprecipitated with monoclonal Notch2 blocking antibody, which binds to the extra- anti-FLAG antibody followed by immunoblotting with anti- cellular region of Notch2 and inhibits its interaction with other Notch2 antibody (Figure 5B). Moreover, we conﬁrmed the bind- proteins (Falk et al., 2012). Treatment of PAE cells, which express ing of MINAR1 with Notch2 in human umbilical primary vein a very low amount of MINAR1, with the Notch2 blocking antibody endothelial cells (HUVEC), which endogenously express both (NOTCH2-B9) had no apparent effect on their capillary tube forma- MINAR1 and Notch2. Cell lysates derived from HUVECs were immu- tion (Figure 6A). However, treatment of PAE cells expressing noprecipitated with anti-MINAR1 antibody or with a control anti- MINAR1 with NOTCH2-B9 antibody signiﬁcantly reversed the body, followed by immunoblotting with anti-Notch2 antibody. inhibitory effect of MINAR1 on the capillary tube formation of PAE Notch2 was co-immunoprecipitated with MINAR1 (Figure 5C). cells (Figure 6A). Taken together, we have identiﬁed Notch2 as a Furthermore, knockdown of MINAR1 by shRNA in HUVECs MINAR1-binding protein, which increases the stability of MINAR1. Figure 6 Blocking Notch2 antibody inhibits the effect of MINAR1 in tube formation of endothelial cells. PAE cells expressing empty vector or MINAR1 were subjected to a matrigel capillary tube formation assay treated with conditioned medium containing blocking anti-Notch2 anti- body or control medium. Pictures were taken after overnight and quantiﬁed via Image J/angiogenesis assay. Downloaded from https://academic.oup.com/jmcb/article-abstract/10/3/195/4796960 by Ed 'DeepDyve' Gillespie user on 26 June 2018 202 j Ho et al. Discussion of the Notch pathway in angiogenesis. Notch signaling is a well- Due to their ability to interact with proteins involved in key known negative regulator of angiogenesis, as it inhibits angiogen- cellular processes such as signal transduction and transcrip- esis (Leong et al., 2002; Itoh et al., 2004; Noseda et al., 2004; tional regulation, IDPs/intrinsically unstructured proteins (IUPs) Williams et al., 2006). Notch2 has also been suggested to negatively are considered a ‘protein–protein interaction hub’ (Wright and regulate breast cancer growth (Parr et al., 2004). However, further Dyson, 2015). Here we describe MINAR1, a previously uncharac- studies are needed to establish the precise role of MINAR1 and its terized protein, as an IDP/IUP. Remarkably, MINAR1 is a highly function, as an important participant in the Notch2 pathway. IDP with nearly 70% of its amino acid sequence predicted to contain no speciﬁc 3D structure or folding. Intrinsically disor- Materials and methods dered sequences in proteins are suggested to extend the inter- Plasmids, shRNAs, and antibodies action surface of proteins, providing ﬂexibility and adaptability MINAR1/UPF0258 protein KIAA1024 cDNA (accession# by allowing a given region of a protein to bind to a range of dis- BC160103 similar to accession# Q9UPX6) was purchased from tinct protein partners (Dosztanyi et al., 2006; Singh et al., 2007; Open biosystems/Dharmacon. MINAR1/KIAA1024 shRNAs cloned Cino et al., 2013). Such protein–protein interactions provide into pGIPZ lentivirus vector (three individual shRNA clones) were gain of 3D folding for IDP/IUPs that leads to stability and func- purchased from Dharmacon, Inc.: V3LHS_381948 (clone# tion (Berlow et al., 2015; Wright and Dyson, 2015). 200306615), V3LHS_411743 (clone# 200253412), and We have identiﬁed Notch2 as a key MINAR1-binding protein. V3LHS_411744 (clone# 200299719). The MINAR1 cDNA was sub- We propose that the Notch2 association with MINAR1 increases sequently cloned into multiple expression vectors including MINAR1 stability and function, as co-expression of Notch2 with pAcGFP1-C3, which generates GFP-tagged MINAR1,mammalian MINAR1 reduced aggregation of MINAR1 as determined by a ﬁl- and amphibian expression vector, pCS2+ vector with no tag, and ter trap assay, consequently resulting in the increased detection pCDNA3.1 myc, which generates MINAR1-myc. MINAR1 was also of stable and solubilized MINAR by western blot. Moreover, cloned into retroviral vector pLNCX2. All the constructs were Notch2 increased the level of endogenous MINAR1 in HEK-293 sequencedand thecompletesequenceofMINAR1 was veriﬁed. cells, likely due to binding to and inducing stability of MINAR1. Rabbit polyclonal anti-MINAR1 antibody was generated against a 21- This gain of stability of MINAR1, as demonstrated by 4PBA treat- amino acid long peptide (KDGFLVEQVFSPHPYPASLKA) corresponding ment in HEK-293 cells and subsequent trypsin digestion, likely to the extracellulardomainofMINAR1. The speciﬁcity of the antibody increased its half-life by reducing its susceptibility to degrad- was validated in cells ectopically expressing MINAR1 or in cells ation. In a recent genome-wide data analysis on protein–protein endogenously expressing MINAR1. The immunoreactivity of anti- interaction networks, it was demonstrated that IDP/IUPs gener- MINAR1 antibody was blocked by pre-incubation of anti-MINAR1 anti- ate more protein network activity than is the typical proteins body with the corresponding peptide, which conﬁrmed its speciﬁcity with the conserved 3D structure (Demarest et al., 2002; (data not shown). Anti-Nocth2 antibody was purchased from Cell Dosztanyi et al., 2006), a distinct feature that puts IDP/IUPs in Signaling. Notch2 blocking antibody (NOTCH2-B9) was purchased an advantageous position in the context of protein–protein from Addgene (Falk et al., 2012). Full-length Notch2-GFP construct interaction, signal transduction, and biological functions. Our was kindly provided by Dr Carmela Abraham (Boston University). data point toward a critical role for MINAR1 in angiogenesis, as Full-length Notch2 cDNA with no tag was kindly provided by Dr it regulates key angiogenic processes such as cell survival and Raphael Kopan (University of Cincinnati College of Medicine). capillary tube formation of endothelial cells. Interestingly, well- characterized IDP/IUPs such as p53, Mdm2,p300, BRCA1,or Cell lines XPA have been associated with various human cancers. Our Porcine aortic endothelial cells (PAE) were kindly provided by observation that MINAR1 expression is downregulated in Dr Carl-Henrik Heldin (Ludwig Cancer Research, University of advanced human breast cancer highlights the potential func- Uppsala, Sweden). Human embryonic kidney epithelial cells tional role of MINAR1 in regulating breast cancer progression. (HEK-293) were kindly provided by Dr Vipul Chitalia (Boston Highly coordinated processes of capillary tube formation of University). PAE and HEK-293 were grown in DMEM medium endothelial cells constitute the most critical aspect of proper supplemented with 10% FBS plus antibiotics. Human umbilical vessel formation in vivo. Our data demonstrate a signiﬁcant role vascular endothelial cells (HUVECs) were purchased from Cell for MINAR1 in angiogenesis. Overexpression of MINAR1 in endo- technologies (Frederick) and grown in the endothelial cell thelial cells inhibited capillary tube formation, whereas silencing medium. Breast carcinoma cell lines, ZR75-1, MDA-MB-231, and of its expression increased capillary tube formation in cell cul- T47D were grown in RPMI Medium 1640 supplemented with ture. MINAR1 also inhibited angiogenesis of endothelial cells in 10% FBS plus antibiotics. Retroviruses were produced in 293- a mouse matrigel plug assay and vessel formation in zebraﬁsh, GPG packaging cells as described (Rahimi et al., 2000). underscoring its functional importance in angiogenesis. The Lentiviruses were produced in 293T cells. underlying mechanism of MINAR1 function in angiogenesis likely requires Notch2, which binds strongly to Notch2 and appears to Immunoprecipitation and western blotting be necessary for stabilization of MINAR1. Remarkably, the bio- Cells were prepared and lysed as described (Hartsough et al., logical effect of MINAR1 in endothelial cells overlaps with that 2013). Brieﬂy, cells were washed twice with H/S buffer (25 mM Downloaded from https://academic.oup.com/jmcb/article-abstract/10/3/195/4796960 by Ed 'DeepDyve' Gillespie user on 26 June 2018 MINAR1 regulates angiogenesis and breast tumor growth j 203 HEPES, pH 7.4, 150 mM NaCl, and 2 mM Na VO ) and lysed in were selected for MS/MS. MS spectra were collected from m/z 3 4 lysis buffer (10 mM Tris-HCl, 10% glycerol, pH 7.4, 5 mM EDTA, 285–1700, and MS2 spectra were collected from m/z 50–3000. 50 mM NaCl, 50 mM NaF, 1% Triton X-100, 1 mM phenylmethyl- The source gas temperature was set to 225°C, and the ﬂow was sulfonyl ﬂuoride [PMSF], 2 mM Na VO , and 20 μg/ml aproti- set at 13 L/min, with a capillary voltage of 2100 V. Precursors 3 4 nin). Whole-cell lysates were subjected to immunoprecipitation ≥5000 counts and charge states ≥2 were selected for fragmen- or were directly subjected to western blotting analysis as indi- tation, and the collision energy was set according to the cated in the ﬁgure legends. equation y= mx + b, with y being the collision energy, slope m = 3.6, x representing the charge state, and the offset b = −4.8 In vitro GST pull-down assay for charge states 3 and 4. For charge state 2 peptides, slope m To generate the glutathione S-transferase (GST) fusion extra- = 3.1, and the offset b = 1. Spectra were collected in centroid cellular domain of MINAR1 (GST-E-MINAR1), the extracellular mode. The error for all peaks with signal-to-noise (S-to-N) ratio of domain of MINAR1 was PCR ampliﬁed and cloned into pGEX-2T >10 is within 5 ppm. MS/MS data were searched using a local vector. The puriﬁed GST-MINAR1 protein was subsequently copy of Mascot (www.matrixscience.com) using Uniprot database used for GST pull-down assay, which was performed as of human proteins. Parameters require a minimum of 2 peptides described (Rahimi et al., 2000). matching per protein with minimum probabilities of 95%atthe peptide level. Zebraﬁsh angiogenesis assay MINAR1 mRNA was prepared in vitro. Brieﬂy, MINAR1 cloned Filter trap assay into pCS2 vector were linearized with Not I restriction enzyme, Cells expressing MINAR1 were prepared and homogenized in treated with Proteinase K (Sigma) and extracted with phenol. phosphate buffered saline (PBS) as described (Scherzinger Linearized plasmid DNA (1 μg/μl) in RNAse-free water was used et al., 1997). Brieﬂy, cells were washed twice in cold PBS and for in vitro capped mRNA synthesis using the mMessage collected in ﬁlter trap buffer (PBS, 1 mM phenylmethylsulfonyl mMachine® SP6 kit (Ambion) according to manufacturer’s ﬂuoride, 20 mg/ml aprotinin) and brieﬂy sonicated using instructions and the RNA subsequently was used to injected Thermo-Fisher Sonicator Dismembrator Model 1000. Proteins into zebraﬁsh embryos. Fli-eGFP-transgenic adult male and were solubilized by adding 1% SDS to the cell homogenates female zebraﬁsh (Danio rerio) were housed in 14:19-h light-dark and these cell homogenates were subsequently blotted on a cel- cycle at a temperature of (26.5°C) and a pH of (7.0–7.4)ina lulose acetate ﬁlter via a dot blot apparatus. The homogenates controlled multi-tank recirculating water system. A glass capil- were subjected to western blot analysis and blotted for MINAR1. lary needle attached to a Femtojet injector (Eppendorf) was used for injecting RNA (10 or 5 ng/μlin ∼10 pl) into 1-cell or 4-Phenylbutyric acid treatment and partial trypsin digest 2-cell-stage embryos. The embryos were grown at 28°Cfor 3 days. HEK-293 cells were treated with 10 mM 4-phenylbutyric acid The embryos were examined after 50 hpf using an immunoﬂuores- (4PBA) for 2 h and lysed in PBS (Wang et al., 2016). Cells desig- cence microscope. The images of ﬁsh under same setting were nated for partial trypsin digest were homogenized by sonication obtained for 10 ﬁsh per group at every experiment and analyzed and centrifuged to obtain supernatant. Trypsin (20 ng) was for the length of the vessels using Image J software. added to cell lysates (50 μl) and incubated at room temperature (26.5°C) for 0, 15, and 30 min. Trypsin digestion was terminated Liquid chromatography–tandem mass spectrometry (LC-MS/MS) by adding sample buffer (5×), followed by incubation in the MINAR1 was immunoprecipitated with anti-Myc antibody from heat block at 95°C for 5 min. The samples were resolved on PAE cells ectopically expressing MINAR1. The immunoprecipi- 10% SDS-PAGE and analyzed by western blot using anti-MINAR tated proteins were subjected SDS-PAGE, then gel bands were antibody. excised, followed by trypsin digestion (at 37°C overnight in 50 mM ammonium bicarbonate). Peptides were separated and In vitro capillary tube formation assay analyzed using a 6550 Q-TOF MS with a 1200 series nanoﬂow PAE cells expressing pMSCV empty vector and MINAR (2 × 10 HPLC-Chip ESI source with a HPLC-Chip consisting of a 360 nl cells per well, triplicate wells per group) were seeded in each trapping column and a 150 mM × 75 μM analytical column, well of a 24-well plate coated with 200 μl growth factor-reduced both with Polaris C18-A 3 μm material (all from Agilent Corp.). Matrigel and allowed to adhere for 1 h. After 24 h, images of After injection of the sample onto the trapping column, the col- capillary tube formation were captured using a Zeiss microscope umn was washed at a rate of 2 μl/min with 2% acetonitrile and camera and Lumenera INFINITY ANALYZE Software. The length of 0.1% formic acid in water for 4 min. Peptides were then sepa- capillary tube branching was analyzed with the Angiogenesis rated on the analytical column at a ﬂow rate of 0.3 μl/min using Analyzer via ImageJ software. a gradient from 2%to 40% acetonitrile with 0.1% formic acid over a period of 25 min. The 6550 Q-TOF mass spectrometer Ethical Approval and Consent to participate was operated in positive mode using the high-resolution, The Boston University the Institutional Animal Care and Use extended dynamic range (2 GHz) setting. The instrument was Committee (IACUC) approved the use of mouse and zebraﬁsh in operated in data-dependent mode; the 20 most abundant ions this study. Boston University Medical Campus Institutional Downloaded from https://academic.oup.com/jmcb/article-abstract/10/3/195/4796960 by Ed 'DeepDyve' Gillespie user on 26 June 2018 204 j Ho et al. Henderson, A.M., Wang, S.J., Taylor, A.C., et al. (2001). The basic helix-loop- Review Board (IRB) approved the use of human tissues in this helix transcription factor HESR1 regulates endothelial cell tube formation. study, which is in accordance with the Declaration of Helsinki. J. Biol. Chem. 276, 6169–6176. Itoh, F., Itoh, S., Goumans, M.J., et al. (2004). Synergy and antagonism between Notch and BMP receptor signaling pathways in endothelial cells. Supplementary material EMBO J. 23, 541–551. Supplementary material is available at Journal of Molecular Leong, K.G., Hu, X., Li, L., et al. (2002). Activated Notch4 inhibits angiogen- Cell Biology online. esis: role of β1-integrin activation. Mol. Cell. Biol. 22, 2830–2841. Liu, Z., Turkoz, A., Jackson, E.N., et al. (2011). Notch1 loss of heterozygosity causes vascular tumors and lethal hemorrhage in mice. J. Clin. Invest. 121, Funding 800–808. This work was supported in part through grants from the NIH Lu, F.M., and Lux, S.E. (1996). Constitutively active human Notch1 binds to (R21CA191970 and R21CA193958 to N.R.) and P41GM104603 the transcription factor CBF1 and stimulates transcription through a pro- moter containing a CBF1-responsive element. Proc. Natl Acad. Sci. USA 93, (to C.E.C.). 5663–5667. McGufﬁn, L.J. (2008). Intrinsic disorder prediction from the analysis of mul- Conﬂict of interest: none declared. tiple protein fold recognition models. Bioinformatics 24, 1798–1804. Noseda, M., Chang, L., McLean, G., et al. (2004). Notch activation induces References endothelial cell cycle arrest and participates in contact inhibition: role of Andersson, E.R., Sandberg, R., and Lendahl, U. (2011). Notch signaling: sim- p21Cip1 repression. Mol. Cell. Biol. 24, 8813–8822. plicity in design, versatility in function. Development 138, 3593–3612. Parr, C., Watkins, G., and Jiang, W.G. (2004). The possible correlation of Arafa, E., Bondzie, P.A., Rezazadeh, K., et al. (2015). TMIGD1 is a novel adhe- Notch-1 and Notch-2 with clinical outcome and tumour clinicopathological sion molecule that protects epithelial cells from oxidative cell injury. Am. J. parameters in human breast cancer. Int. J. Mol. Med. 14, 779–786. Pathol. 185, 2757–2767. Rahimi, N. (2006a). Vascular endothelial growth factor receptors: molecular Berlow, R.B., Dyson, H.J., and Wright, P.E. (2015). Functional advantages of mechanisms of activation and therapeutic potentials. Exp. Eye Res. 83, dynamic protein disorder. FEBS Lett. 589, 2433–2440. 1005–1016. Campen, A., Williams, R.M., Brown, C.J., et al. (2008). TOP-IDP-scale: a new Rahimi, N. (2006b). VEGFR-1 and VEGFR-2: two non-identical twins with a amino acid scale measuring propensity for intrinsic disorder. Protein Pept. unique physiognomy. Front. Biosci. 11, 818–829. Lett. 15, 956–963. Rahimi, N. (2012). The ubiquitin-proteasome system meets angiogenesis. Chavali, S., Gunnarsson, A., and Babu, M.M. (2017). Intrinsically disordered Mol. Cancer Ther. 11, 538–548. proteins adaptively reorganize cellular matter during stress. Trends Rahimi, N., Dayanir, V., and Lashkari, K. (2000). Receptor chimeras indicate Biochem. Sci. 42, 410–412. that the vascular endothelial growth factor receptor-1 (VEGFR-1) modu- Chillakuri, C.R., Sheppard, D., Lea, S.M., et al. (2012). Notch receptor–ligand lates mitogenic activity of VEGFR-2 in endothelial cells. J. Biol. Chem. 275, binding and activation: insights from molecular studies. Semin. Cell Dev. 16986–16992. Biol. 23, 421–428. Rahimi, N., Rezazadeh, K., Mahoney, J.E., et al. (2012). Identiﬁcation of IGPR- Cino, E.A., Killoran, R.C., Karttunen, M., et al. (2013). Binding of disordered 1 as a novel adhesion molecule involved in angiogenesis. Mol. Biol. Cell proteins to a protein hub. Sci. Rep. 3, 2305. 23, 1646–1656. Dames, S.A., Martinez-Yamout, M., De Guzman, R.N., et al. (2002). Structural Roche, D.B., Buenavista, M.T., Tetchner, S.J., et al. (2011). The IntFOLD ser- basis for Hif-1α/CBP recognition in the cellular hypoxic response. Proc. ver: an integrated web resource for protein fold recognition, 3D model Natl Acad. Sci. USA 99, 5271–5276. quality assessment, intrinsic disorder prediction, domain prediction and De Guzman, R.N., Wojciak, J.M., Martinez-Yamout, M.A., et al. (2005). CBP/ ligand binding site prediction. Nucleic Acids Res. 39,W171–W176. p300 TAZ1 domain forms a structured scaffold for ligand binding. Scherzinger, E., Lurz, R., Turmaine, M., et al. (1997). Huntingtin-encoded Biochemistry 44, 490–497. polyglutamine expansions form amyloid-like protein aggregates in vitro Demarest, S.J., Martinez-Yamout, M., Chung, J., et al. (2002). Mutual syner- and in vivo. Cell 90, 549–558. gistic folding in recruitment of CBP/p300 by p160 nuclear receptor coacti- Singh, G.P., Ganapathi, M., and Dash, D. (2007). Role of intrinsic disorder in vators. Nature 415, 549–553. transient interactions of hub proteins. Proteins 66, 761–765. Dosztanyi, Z., Chen, J., Dunker, A.K., et al. (2006). Disorder and sequence Taylor, K.L., Henderson, A.M., and Hughes, C.C. (2002). Notch activation dur- repeats in hub proteins and their implications for network evolution. J. ing endothelial cell network formation in vitro targets the basic HLH tran- Proteome Res. 5, 2985–2995. scription factor HESR-1 and downregulates VEGFR-2/KDR expression. Dyson, H.J., and Wright, P.E. (2005). Intrinsically unstructured proteins and Microvasc. Res. 64, 372–383. their functions. Nat. Rev. Mol. Cell Biol. 6, 197–208. Tompa, P. (2002). Intrinsically unstructured proteins. Trends Biochem. Sci. Falk, R., Falk, A., Dyson, M.R., et al. (2012). Generation of anti-Notch anti- 27, 527–533. bodies and their application in blocking Notch signalling in neural stem Uversky, V.N. (2002). Cracking the folding code. Why do some proteins adopt cells. Methods 58, 69–78. partially folded conformations, whereas other don’t? FEBS Lett. 514, 181–183. Gsponer, J., and Babu, M.M. (2009). The rules of disorder or why disorder Wang,Y.H., Meyer, R.D.,Bondzie,P.A., et al.(2016). IGPR-1 is required for endo- rules. Prog. Biophys. Mol. Biol. 99, 94–103. thelial cell–cell adhesion and barrier function. J. Mol. Biol. 428, 5019–5033. Guruprasad,K., Reddy, B.V.,and Pandit, M.W.(1990). Correlation between stabil- Wilkins, M.R., Gasteiger, E., Bairoch, A., et al. (1999). Protein identiﬁcation ity of a protein and its dipeptide composition: a novel approach for predicting and analysis tools in the ExPASy server. Methods Mol. Biol. 112, 531–552. in vivo stability of a protein from its primary sequence. Protein Eng. 4, 155–161. Williams, C.K., Li, J.L., Murga, M., et al. (2006). Up-regulation of the Notch lig- Hartsough, E.J., Meyer, R.D., Chitalia, V., et al. (2013). Lysine methylation and Delta-like 4 inhibits VEGF-induced endothelial cell function. Blood promotes VEGFR-2 activation and angiogenesis. Sci. Signal. 6,ra104. 107, 931–939. Hazeki, N., Tukamoto, T., Goto, J., et al. (2000). Formic acid dissolves aggre- Wright,P.E., and Dyson,H.J.(1999). Intrinsically unstructured proteins: re- gates of an N-terminal huntingtin fragment containing an expanded poly- assessing the protein structure–function paradigm.J.Mol.Biol. 293, 321–331. glutamine tract: applying to quantiﬁcation of protein components of the Wright, P.E., and Dyson, H.J. (2015). Intrinsically disordered proteins in cellu- aggregates. Biochem. Biophys. Res. Commun. 277, 386–393. lar signalling and regulation. Nat. Rev. Mol. Cell Biol. 16, 18–29. Downloaded from https://academic.oup.com/jmcb/article-abstract/10/3/195/4796960 by Ed 'DeepDyve' Gillespie user on 26 June 2018
Journal of Molecular Cell Biology – Oxford University Press
Published: Mar 2, 2018
Access the full text.
Sign up today, get DeepDyve free for 14 days.