PREP1 (aka PKNOX1, PBX-knotted homeobox 1) was identified by Chen et al. (1997) in their effort to characterize genes mapping on the human chromosome 21. At the same time, PREP1 was identified as one of the components of the human transcription factor complex UEF3, urokinase enhancer factor 3. PREP1 was found to be a homeodomain-containing DNA-binding protein belonging to the TALE (three aminoacid loop extension) superclass characterized by the MEINOX domain, most closely related to TGIF (TGF β–induced factor homeobox) and MEIS1 (Meis homeobox 1). ChIP-seq analysis shows that in vivo PREP1 recognizes the decameric TGAXTGACAG consensus, both alone and as dimer with PBX1 (Penkov et al. 2013). PREP1 heterodimerizes with PBX through the amino-terminal MEIS-A and MEIS-B motifs present in the MEINOX domain (Berthelsen et al. 1998); PREP1 binds PBX1 in the cytoplasm and is transported to the nucleus where it prevents PBX nuclear export (Berthelsen et al. 1999) (hence the acronym, for Pbx regulating protein 1). Upon binding to the specific DNA, the PREP1-PBX1 dimer undergoes a conformational change, becoming more compact (Mathiasen et al. 2016).
Gene and Protein Structure
The human PREP1 gene (gene ID 5316, GenBank) maps on chromosome 21 (21q22.3) and covers 59,421 bp, from nucleotide 42,974,510 to 43,033,931, on the direct strand. The cDNA coding for PREP1 contains a 126-bp-long 5′-UTR (5′-untranslated region), a 1308-bp-long open reading frame, and a 3′-UTR region of more than 1,000 bp. The coding region encodes a protein of 436 amino acids, migrating in SDS-PAGE at 64 kDa (Berthelsen et al. 1998). Murine Prep1 maps on chromosome 17 and covers 42,940 bp on the direct strand. The mRNA coding for the murine Prep1 is 4,093 nucleotides long. The protein has the same length and the same characteristics as human PREP1 with only 21 aminoacidic substitutions, 10 of which are conservatives (Ferretti et al. 1999).
Human PREP2 gene maps on chromosome 11 (11q24) and covers 268,730 bp, from nucleotide 124,539,770 to 124,808,495, on the direct strand. The full-length protein is 461 amino acids long (Fognani et al. 2002). Murine Prep2 maps on chromosome 9. Three transcripts (3.8, 1.6, and 0.8 kb long) were identified; the relative accumulation of the longer ones differs among organs, whereas the 0.8 kb form is detected only in testis. Murine Prep2 protein is 462 residues long (Haller et al. 2002). Alternative splicing gives rise to multiple isoforms, including a 25-kDa variant due to retention of intron 4, that lacks the C-terminal half of the protein including the homeodomain (Haller et al. 2004).
The PREP family of proteins (including Prep1 and Prep2 in vertebrates) belongs to the MEINOX class of the TALE superclass of DNA-binding proteins. Homeoproteins belonging to this superclass contain a divergent homeodomain (HD) with a three aminoacid loop extension (TALE, P-Y-P) between the first and second α-helix.
PREP1 has a poor affinity for DNA, but the affinity is drastically increased upon dimerization with PBX1.
The binding to DNA of PREP1-PBX1 homeodomain dimers is cooperative. Sequences close to but outside the homeodomain are also involved and are required for high affinity DNA binding (Zucchelli et al. 2017). Moreover, binding to DNA induces a conformational change in the protein (Fig. 1b).
Prep1 is expressed ubiquitously but the levels are modulated in different organs and tissues. In early zebrafish and mouse embryos, Prep1 is expressed in all organs. In the adult mouse, the highest levels were observed in testis, brain, and thymus (Ferretti et al. 2006). In the hematopoietic system, Prep1 is expressed in the stem cell compartment with levels decreasing upon cell differentiation (Di Rosa et al. 2007). Interestingly, PREP1 levels are usually lower in human tumor samples than in the normal tissue (Longobardi et al. 2010).
Human Prep2 is highly expressed in heart, brain, skeletal muscle, and ovary (Fognani et al. 2002). Testis-specific transcripts have been detected in both human (7–8 kb) and mouse (0.8 kb) tissues (Fognani et al. 2002; Haller et al. 2002).
Despite its ubiquitous expression, PREP1 binds to a rather small core of genes in all cells and most of the binding sites are cell-type specific (Penkov et al. 2013; Laurent et al. 2015; Dardaei et al. 2015).
PREP proteins are able to bind the conserved sequence TGAXTGACAG and at lower frequency the same sequence bound by the HOX-PBX complexes, TGXXATXT (Penkov et al. 2013). ChIP-seq and RNA-seq data show that many genes are bound and regulated (Penkov et al. 2013; Dardaei et al. 2015; Laurent et al. 2015), including somatostatin (Goudet et al. 1999), HoxB2 (Ferretti et al. 2000), Pax6 (Mikkola et al. 2001), and FSH β-subunit (Bailey et al. 2004).
The interaction between PBX and PREP proteins is important for their translocation to the nucleus. In its homeodomain, PBX proteins contain two nuclear localization signals (NLS) that are not present in PREP. Therefore, the interaction with PBX1 allows nuclear translocation of PREP1 (Berthelsen et al. 1999). PBX also has a nuclear export signal (NES) in the PBC-A domain that favors its exit from the nucleus. The interaction with PREP (or MEIS) masks this signal, allowing nuclear retention of the dimeric complex (Berthelsen et al. 1999). Cytoplasmic Prep2 localization is dependent on Crm-1-mediated nuclear export and association with the actin and microtubule cytoskeleton (Haller et al. 2004).
The interaction between PBX and PREP (or MEIS) is compatible with that of HOX and PBX. Indeed, PREP and PBX recognize each other through their N-terminal regions (PBC-A and B domains in Pbx and MEIS-A/MEIS-B in PREP), while the interaction of HOX proteins with PBX involves the homeodomains. Therefore, ternary complexes between PBX, HOX, and PREP can be formed and are able to bind DNA (Berthelsen et al. 1998). This increases the specificity of Hox proteins in DNA consensus sequences recognition, including both TGAXTGACAG and TGATNNAT. For the binding of a ternary complex to DNA, both consensus sequences are needed. The ChIP-seq analyses have confirmed the binding of Prep, Pbx, and HoxB1 to some of Hox genes, as well as that of Meis, Pbx1, and HoxB1 (Penkov et al. 2013).
PREP1-PBX1 complexes interact also with non-HOX proteins, cooperating in the activation of different promoters. For example, Pbx1 and Prep1 activate the somatostatin promoter when coexpressed with the pancreatic homeodomain transcription factor Pdx1. Also in this context, the cooperative binding of two regulatory elements (UE-A by the Pbx1/Prep1 dimer and TSEI by Pdx1) of the promoter is necessary for full activation (Goudet et al. 1999).
Prep1 and Pbx1 form trimeric complexes also with nonhomeodomain proteins. Indeed, Pbx1 and Prep1 have been identified as Smad partners in a trimeric complex involved in the regulation of FSH β gene by activin (Bailey et al. 2004).
Besides HOX and PBX, other PREP1 interactors have been discovered such as MYBBP1A (p160), which may function as an inhibitor of its transcriptional activity (Díaz et al. 2007). As MYBBP1A is a nucleolar protein, it can interact with PREP1 only when the nuclear and nucleolar membranes are dissolved, i.e., at the start of mitosis.
Prep1-null mouse embryos die at e7.5 before gastrulation because epiblast cells undergo p53-dependent apoptosis (Fernandez-Diaz et al. 2010). Mouse embryos carrying a hypomorphic Prep1 i/i mutation (expressing 2% mRNA, 3–7% protein) show general organ hypoplasia and, in 75% of cases, die at about e17 with major alterations in hematopoiesis, angiogenesis, and eye development (Ferretti et al. 2006; Di Rosa et al. 2007). Minor Prep1 gene expression differences have a strong impact on embryonic development. Indeed, double heterozygous Prep1 i/- embryos that express half of Prep1 mRNA as Prep1 i/i (about 1%) display an intermediate phenotype with embryonic lethality around E12.5 (Rowan et al. 2010). Remarkably, homozygous Prep1 i/i hypomorphic mice surviving embryonic lethality are prone to develop tumors within 20 months of age, and Prep1 heterozygosity accelerates Myc-dependent lymphomagenesis. These data indicate that Prep1 is a tumor suppressor. Indeed, the majority of human tumors shows no or little PREP1 (Longobardi et al. 2010).
Prep1 i/i MEFs or PREP1-downregulated human fibroblasts show a much higher level of γ-H2Ax or activated ATM foci, demonstrating ongoing DNA repair, and a much greater response to γ-irradiation. This therefore suggests that in the absence of Prep1 cells accumulate DNA damage that needs to be repaired. Indeed, Prep1i/i MEFs are easily transformed by a single oncogene that can bypass oncogene-induced replicative stress and senescence (Iotti et al. 2011; Dardaei et al. 2014).
A loss-of-function mutation for Prep2 has not been described.
Overexpression Related Phenotypes
As expected, since PREP1 is a tumor suppressor, the overexpression in human or mouse cancer cells strongly inhibits growth rate in vitro and in vivo (Dardaei et al. 2014; Risolino et al. 2014). At the molecular level, the overexpression of PREP1 in MEFs drastically reduces the level of the MEIS1 oncogene and its downstream effects. Indeed, the interaction of MEIS1 with the oncogenes DDX3X and DDX5 is abolished and the spectrum of genes bound by MEIS1 is strongly reduced. In particular, it appears that the oncogenicity due to overexpression of MEIS1 is due to the binding to low affinity sites and decrease of DNA selectivity, since under these conditions MEIS1 binds genes normally bound by the Jun/Fos family of oncogenes. The reduction of Meis1 by Prep1 overexpression also prevents the binding to these genes, hence inhibiting tumorigenicity (Dardaei et al. 2014, 2015).
Overexpression of PREP1 in human non-small cells lung cancer (NSCLC) A549 cells induces an epithelial-to-mesenchymal transition (EMT). This is due to the hijacking of the TGF-β pathway, with decreased cadherin and increased vimentin expression, activation of the Jun-Fra1 pathway, and increased exoproteases production and migration rate. The first step is the binding of PREP1 to a Smad3 enhancer that later induces the Jun-Fra1 pathway. In vivo, overexpression of PREP1 in the A549 cells on one side decreases their growth rate in nude mice in agreement with the tumor-suppressing activity. On the other hand, however, a fraction of the cells become able to metastasize in vivo. Accordingly, the immunohistochemical comparison between human primary tumors and NSCLC-generated brain metastases shows that while most primary tumors express little or no PREP1, 30/30 brain metastases express normal or elevated PREP1 levels. Indeed, the Kaplan-Meyer survival curves for human stage-1 NSCLC patients show that the expression/overexpression of PREP1 is a strong negative prognostic marker, patients expressing PREP1 presenting an over threefold higher risk of recurrence (Risolino et al. 2014).
PREP proteins have been extensively studied as critical mediators of the activity of other homeodomain transcription factors (in particular HOX and PBX). At least for Prep1, a loss-of-function model has recently demonstrated that Prep1 is the only TALE-class protein essential at very early stages of embryonic development (Fernandez-Diaz et al. 2010). This strongly suggests that this protein has other, not yet identified, cellular functions.
Pleiotropy: Prep1 has a pivotal role in several embryonically unrelated cell types, consistent with its ubiquitous expression, suggesting a role in the biology of stem cells of different tissues.
Gene-dosage dependence: Small differences in its expression have great impact on cellular and tissue homeostasis, suggesting that Prep1 levels must be tightly regulated.
Furthermore, recent data demonstrate that (unlike the other TALE proteins) PREP1 functions as a tumor suppressor possibly by protecting DNA from damage, prompting a deeper investigation of its mechanisms of action. In addition, Prep1 and Meis1 compete in tumorigenesis, and Prep1 inhibits Meis1 oncogenic activity. This is both based on the Prep1 control of Meis1 level and on the change in DNA selectivity of Prep1 and Meis1 when they are overexpressed.
- Ferretti E, Marshall H, Pöpperl H, Maconochie M, Krumlauf R, Blasi F. A complex site including both Pbx-Hox and Prep-Meis-responsive elements and binding a retinoic acid-inducible ternary Hoxb1-Pbx-Prep1 complex is required for HOXB2 rhombomere 4 expression. Development. 2000;127:155–66.PubMedPubMedCentralGoogle Scholar
- Ferretti E, Villaescusa JC, Di Rosa P, Fernandez-Diaz LC, Longobardi E, Mazzieri R, et al. Hypomorphic mutation of the TALE gene Prep1 (pKnox1) causes a major reduction of Pbx and Meis proteins and a pleiotropic embryonic phenotype. Mol Cell Biol. 2006;26:5650–62.PubMedPubMedCentralCrossRefGoogle Scholar
- Longobardi E, Iotti G, Di Rosa P, Mejetta S, Bianchi F, Fernandez-Diaz LC, et al. Spontaneous tumor development and acceleration of EmMyc lymphomas in mice haploinsufficient for the homeodomain transcription factor gene Prep1 (pKnox1) indicates a tumor suppressor function. Mol Oncol. 2010;4:226–34.CrossRefGoogle Scholar
- Penkov D, Mateos San Martin D, Fernandez-Diaz LC, Rosselló CA, Torroja C, Sánchez-Cabo F, Warnatz HJ, Yaspo ML, Gabrieli A, Tkachuk V, Brendolan A, Blasi F, Torres M. Analysis of the in vivo DNA-binding profile and function of TALE homeoproteins reveals their specialization and differential interactions with Hox genes and proteins. Cell Rep. 2013;3:1321–33.PubMedCrossRefGoogle Scholar
- Risolino M, Mandia N, Iavarone F, Dardaei L, Longobardi E, Fernandez S, Talotta F, Bianchi F, Pisati F, Spaggiari L, Harter PN, Mittelbronn M, Schulte D, Incoronato MR, Di Fiore PP, Blasi F, Verde P. The transcription factor PREP1 induces EMT and metastasis by controlling the TGF-beta-SMAD3 pathway in non-small cell lung adenocarcinoma. PNAS Plus. 2014;111:E3775–84.CrossRefGoogle Scholar
- Zucchelli C, Ferrari E, Blasi F, Musco G, Bruckmann C. New insights into cooperative binding of homeodomain transcription factors PREP1 and PBX1 to DNA. Sci Rep Nat. 2017;7:40665.Google Scholar