During embryonic development as well as during tissue regeneration and repair, cell proliferation and differentiation are highly regulated and well coordinated events that require de novo transcription of tissue-specific or cell-specific genes. Activation of target gene transcription is initiated by tissue- or cell-specific transcriptional factors and their chromatin remodeling coactivators in contact with the RNA polymerase II (RNAP II) initiation complex (Emerson, 2002). During the elongation phase, RNAP II encounters numerous structural obstacles causing RNAP II to either pause or terminate transcription prematurely. In mammalian cells, several factors have been identified that can enhance RNAP II processivity in vitro (Shilatifard, 1998; Conaway and Conaway, 1999). Among these factors that process the elongation activity are Eleven-nineteen Lysine-rich Leukemia (ELL) family of proteins (Thirman et al., 1994, 1997; Mitani et al., 1995; Shilatifard et al, 1996; Shilatifard et al., 1997; Miller et al., 2000).
The ELL gene was initially identified on chromosome 19p 13.1, which undergoes frequent translocation with the MLL gene on chromosome 11q23 in acute myeloid leukemia (AML; Thirman et al., 1994). ELL may play an essential role in regulating developmental processes as its expression is spatially and temporally restricted during embryogenesis (Thirman et al., 1997) and inactivation of murine Ell by gene targeting resulted in embryonic lethality around the time of gastrulation (Mitani et al., 2000). In Drosophila, loss-of-function mutations in dELL locus cause recessive lethality and segmentation defects in developing larva (Eissenberg et al., 2002). In culture, overexpression of ELL inhibits cell division and induces apoptosis (Johnstone et al., 2001). Remarkably, the growth-suppression and death-promoting activities of ELL can be assigned to discrete functional domains that possess elongation-enhancing activities and immortalizing functions, respectively (Kanda et al., 1998; DiMartino et al., 2000; Johnstone et al., 2001). This finding further suggests the existence of distinct interaction surfaces on ELL and through which ELL could be differentially regulated.
Previously, several ELL-interacting partners have been identified and have been shown to mask the inhibitory activity of ELL in transcription initiation and to stimulate RNA elongation (Schmidt et al., 1999; Kamura et al., 2001) or transcription (Simone et al., 2001, 2003). Human EAF1 and EAF2 are two of these ELL-associated factors that had been shown to enhance transcription (Simone et al., 2001, 2003). ELL and EAF1 colocalize with p80 coilin and may function as components of Cajal bodies, a nuclear compartment involved in snRNP biogenesis (Gall, 2000; Polak et al., 2003). Further studies showed that EAF-interacting domain in ELL is essential for the transformation activity of the MLL–ELL fusion protein (Luo et al., 2001; Simone et al., 2003). EAF1 or EAF2 could replace ELL in its ability to immortalize hematopoietic progenitor cells when it was fused to MLL (Luo et al., 2001; Simone et al., 2003). Together, these data suggest that EAFs may function as oncoproteins if recruited by the MLL/ELL complex. However, very little is known about the expression pattern of these ELL-interacting molecules during embryonic development and their normal cellular functions.
In this report, we describe the cloning and expression of a new member of ELL-associated factors, murine Eaf2. Sequence comparisons revealed a high degree of amino acid sequence conservation among rodent and human EAF2 proteins. Eaf2 is expressed preferentially in the central nervous system (CNS) and sensory and neuroendocrine organs and in sites where active epithelium–mesenchymal interactions are occurring. Together these results suggest that Eaf2 may be an important regulator of morphogenesis, cell growth, and differentiation.
Identification of cDNA Encoding EAF2
In performing a yeast two-hybrid screen to identify proteins that may potentially interact with murine Msx1, we isolated a novel cDNA clone. This cDNA produced a predicted open reading frame of 262 amino acids with a predicted molecular weight of 29.187 kDa (Fig. 1A). Charged residues constitute 27.86% of overall amino acid content and 17.6% of amino acids are composed of serine residues.
BLASTp search of the GenBank nonredundant sequence database identified two human sequences (GenBank accession nos. AAH14209 and AAF67627) that are virtually identical to that of the mouse Eaf2 (Fig. 1B). Although the carboxyl terminus showed significant divergence, these proteins share 79% amino acid identities. In addition, we identified a rat ortholog of Eaf2 (GenBank accession no. AAL12225) that has 91% amino acids in common with the mouse Eaf2 (Fig. 1B). Furthermore, we identified a related human gene, EAF1 (GenBank accession no. AAK58687), that was previously cloned and characterized and its mouse Eaf1 ortholog (GenBank accession nos. BAC32058 and BAC31806; Fig. 1C). Although the overall amino acid homology between the murine Eaf2 and human EAF1 is only 53%, ClusterW alignment of protein sequences between Eaf2 and EAF1 identified two regions of significant homology: one at the N-terminus spanning from amino acids 10 to 127 (79% amino acid identities) and another at the extreme C-terminus between amino acids 243 and 262 (80% amino acid identities; Fig. 1C). A unique proline-rich domain found in the EAF1 distinguishes itself from the Eaf2 members of the family (Fig. 1C).
BLASTp search also identified a prototype sequence for the mammalian Eafs in the Drosophila genome encoded by the CG11166 gene (GenBank accession no. AAF59266). We tentatively named it dEaf. ClusterW alignment showed high degree of amino acid sequence similarities (48%) between the N-terminus of dEaf and the N-termini of mammalian Eafs (Fig. 1D). In addition, two other domains on dEaf exhibited considerable homology with mammalian Eafs; these were the serine-rich domain and the 20 amino acids at the extreme carboxyl terminus (40% similarities and 45% similarities, respectively; data not shown).
Expression Profile of Eaf2 in CNS
To gain a better understanding of sites where Eaf2 may exert its function alone or together with ELL, we decided to determine the expression pattern of Eaf2 during embryonic development by performing in situ hybridization on cryosections of developing mouse embryos. At embryonic day (E) 10, Eaf2 transcripts were found mainly in the developing brain and spinal cord (data not shown). In the mid-gestation embryo from E12 to E15, Eaf2 was expressed at very high levels throughout the developing CNS, including the brain, spinal cord, and cranial and spinal ganglia (Fig. 2A,B,E). Of interest, intense hybridization signals were also observed in sensory organs and neuroendocrine structures. These structures included the lens and retina of the developing eye, the cochlea of the embryonic ear, the olfactory epithelium, and the pituitary (Figs. 2A–D,F, 3A). In the developing eye, Eaf2 transcripts were absent from the developing optic vesicle before the appearance of retinal ganglion cells (RGCs; data not shown). In the mouse, RGC maturation initiates around E12.5 and peaked by E14.5 (Young, 1985; Brown et al., 2001). Expression of Eaf2 coincided remarkably with the birth of RGCs, and intense hybridization signal was preferentially found in the RGC layer of the developing retina (Fig. 2C,D). Of interest, Eaf2 transcripts were absent from actively cycling retinal cells in the retinal marginal zone and low level Eaf2 gene expression was detected in dividing cells in the residual portion of the neuroretina, excluding RGCs (Fig. 2C,D). In the developing lens, Eaf2 transcripts were found exclusively in the primary lens fiber cells but were ab sent in the anterior lens epithelium (Fig. 2C). As anterior lens epithelial cells move into the equatorial zone of the lens and exit cell cycle, Eaf2 was preferentially transcribed by the lens equatorial epithelial cells (Fig. 2D).
Expression Profile of Eaf2 During Epithelium–Mesenchymal Induction
Furthermore, Eaf2 transcripts were detected in structures that undergo extensive branching morphogenesis. Extensive expression was observed in the epithelia of developing nephrons, in the bronchial epithelium of the embryonic lung, in the secretory epithelium of submandibular glands, and in the tubular epithelium of the epididymis, although a greatly reduced level of Eaf2 expression was detected in the mesenchyme of these organs (Fig. 3A–E). In the ectodermal invaginations of mammary buds and vibrissae follicles, Eaf2 transcripts were also found (Fig. 3F,G). Of interest, Eaf2 transcripts were also detected in the developing incisors and molars (Fig. 3H–J). Again, Eaf2 expression was more intense in the dental epithelium, although a reduced amount of Eaf2 transcripts was also found in the dental mesenchyme and dental papilla (Fig. 3I,J). Other sites of expression included intestine and bladder endothelium (Fig. 3A; data not shown). In the developing liver and heart, Eaf2 transcript was absent or below the sensitivity of our detection method (Figs. 2A, 3A).
Expression Profile of Eaf2 in Adult Tissues
To determine the expression pattern of Eaf2 in adult tissues, we performed Northern blot hybridization on poly(A+) RNAs or total RNAs isolated from mouse adult tissues. Several adult tissues that perform reproductive functions, including the ovary, uterus, mammary glands, and testis, were found to produce Eaf2 transcripts (Fig. 4A,B). Eaf2 transcripts were also detected in the adult brain, spleen, liver, lung, thymus, and kidney (Fig. 4A). Skeletal muscle and skin expressed the Eaf2 gene at very low levels (Fig. 4B) as indicated by the relative intensity of hybridization signal in comparison to the control glucose-6-phosphate dihydrogenase probe. Eaf2 transcripts were undetectable in the adult heart (Fig. 4A).
Three transcripts with different molecular weights were detected based on Northern blot analysis. A major and the most abundant transcript is approximately 1.4 kb in length. A second minor transcript that was detected only in the brain, lung, and ovary was approximately 2.6 kb long. A third minor RNA species was detected in the spleen sample and corresponds to a transcript size of approximately 1 kb. These minor RNA species are most likely products of splicing variants not due to cross-hybridization with related genes, such as Eaf1, because Eaf1 and Eaf2 shared only 35% similarities at the nucleic acid level and BLASTn search of the mouse genome database and the GenBank nonredundant nucleic acid database did not produce additional DNA sequences with significant homology (data not shown).
We have isolated and characterized murine Eaf2. It showed remarkable amino acid conservation with its human ortholog; 79% amino acids were identical and homology approaches 86% if conserved changes were taken into consideration. By using the BLASTp algorithm, we have also identified the Drosophila prototype of mammalian Eafs in addition to the rat Eaf2 and the mouse Eaf1. Together with previously identified human EAFs, these genes constitute a new gene family. Moreover, by performing sequence alignments between the Drosophila Eaf and mammalian Eafs, we identified three highly conserved protein domains. At least two of these conserved protein domains were shown previously to be essential for binding to ELL or activating transcription, respectively (Simone et al., 2001, 2003), suggesting possible functional conservation. Although no loss-of-function mutation has been reported, dEaf encoded by the DrosophilaCG11166 gene appeared to functionally interact with pannier (pnr), a GATA-type zinc-finger transcriptional factor. Induced overexpression of CG11166 by a promoter-specific insertion resulted in the enhancement of pnr mutant phenotypes (Pena-Rangel et al., 2002).
Of interest, Eaf2 expression appears to be tightly regulated during embryonic development. Its expression is spatially and temporally restricted. Eaf2 transcripts were detected in organs whose development require series of reciprocal tissue–tissue inductions between the epithelium and the underlying mesenchyme. These included the tooth, mammary placodes, vibrissae follicles, submandibular glands, lung, pancreas, and kidney. This preferential spatial distribution of Eaf2 transcripts in branching structures suggests that Eaf2 may be actively involved in their inductive processes.
Furthermore, Eaf2 may play a key role in regulating growth and differentiation of the nervous system as demonstrated by its extensive expression in the embryonic brain, spinal cord, and cranial and spinal ganglia. The expression of Eaf2 in the CNS during early stages of mouse embryogenesis is very similar to the reported expression profile of Ell (Thirman et al., 1997). In the sensory organs, such as olfactory epithelium and the sensory epithelium in the cochlea, high levels of Eaf2 expression were found, suggesting its possible function in facilitating the maturation process of these specialized epithelia. In the developing retina, the appearance of Eaf2 transcripts coincides with the birth of retinal ganglion cells (RGCs) and RNA in situ showed intense hybridization signal in the RGC layer. This finding suggests that Eaf2 may be an active player in regulating the differentiation process of RGCs perhaps by promoting or stabilizing the state of differentiation. Further studies are needed to define its role in controlling neurogenesis.
In the developing lens, Eaf2 transcripts were absent in the proliferating anterior lens epithelial cells but were present aplenty in the terminally differentiated primary lens fiber cells and also in nonproliferating lens fiber cells in the equatorial zone where lens epithelial cells withdraw from the cell cycle and terminally differentiate into secondary lens fiber cells (Menko, 2002). This spatially restricted pattern of Eaf2 expression in the developing lens provides a good indication that Eaf2 may play an important role in regulating the differentiation program of lens fiber cells or their withdraw from the cell cycle. The functional significance of these observations awaits corroborating results from genetic perturbation experiments.
Isolation of Eaf2 and Plasmid Constructions
Eaf2 was identified as a putative positive-interacting target for Msx1 while performing a yeast two-hybrid screening using a mouse two-hybrid cDNA library prepared from E14.5 embryos (Stratagene, La Jolla, CA). The plasmid that contained Eaf2 cDNA was isolated from yeast cells and used to transform Escherichia coli (XL-Blue). This resulted in the cloning of Eaf2. However, we were not able to demonstrate its interaction with Msx1 by immunoprecipitation.
Nonradioactive In Situ Hybridization
Embryos were fixed in 4% paraformaldehyde overnight. Fixed embryos were then transferred into 30% sucrose and followed by equilibration in 15% sucrose in OCT before embedment in OCT. Cryosections of 20 microns thick were collected onto Histoplus slides (Fisher Scientific). Sense and antisense RNA probes were generated by incorporating digoxigenin–UTP (Roche Biochemicals), according to the manufacturer's recommendation. Before hybridization, sections were treated with 25 μg/ml proteinase K and post-fixed in 0.2% glutaraldehyde and 4% paraformaldehyde. Hybridization was carried out at 65°C with a probe concentration of 1–2 μg/ml in 50% formamide, 5× standard saline citrate (SSC), 0.1% Tween-20, 0.1% CHAPS, 0.2 mg/ml yeast tRNA, 0.005M ethylenediaminetetraacetic acid, 50 μg/ml Haperin, and 1% blocking reagent (Roche Biochemicals). Posthybridization washes were carried out once in 2 × SSC/0.1% CHAPS at 65°C, once with 1 μg/ml RNAse A in 2× SSC at 37°C, once in 2× SSC/0.1% CHAPS at 65°C, and the final wash in 0.2× SSC/0.1% CHAPS at 65°C. Sections were then blocked with 20% heat-inactivated goat serum and incubated overnight with mouse anti-digoxigenin antibody that carries alkaline phosphatase conjugate at 4°C. After extensive washes with phosphate-buffered saline, sections were subjected to colorimetric development in the presence of nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) solution (Roche Biochemicals). Washed sections were then post-fixed in 4% paraformaldehyde, washed, and cover-slide mounted. Images were collected by using a Spot digital camera mounted on an Olympus BX-50 light microscope.
Northern Blot Analysis
A multiple-tissue Northern blot that contained 2 μg of poly(A) RNA from 10 mouse tissues was hybridized and washed according to the manufacturer's recommendations (Ambion, Austin, TX). The mice used for RNA isolation were of mixed sex, 8–10 weeks old. The embryonic RNA was extracted from E14.5 embryos. Additional Northern blot that contained approximately 20–50 μg of total RNA per lane was probed and washed as previously described (Liu et al., 1999). The 32P-labeled probe for hybridization was generated by random-priming using a purified 1-kb Eaf2 cDNA fragment as the template.
We thank Dr. Cheng-Ming Chuong and Dr. Ping Wu for providing help on in situ hybridization. The Eaf2 sequence has been deposited in the GenBank and can be accessed by using GenBank accession no. AY034479.