Cloning and Characterization of a Gene Expressed during Terminal Differentiation That Encodes a Novel Inhibitor of Growth*

We report here the cloning and initial characterization of a novel growth-related gene (EEG-1) that is located on the short arm of chromosome 12. Two spliced transcripts were cloned from human bone marrow and human erythroid progenitor cells: EEG-1L containing a 4350-nucleotide open reading frame encoding a putative protein of 1077 amino acids including a C1q-like globular domain, and an alternatively spliced transcript lacking exon 5 (EEG-1S) encodes a significantly smaller coding region and no C1q-like domain. Quantitative PCR revealed expression of both EEG-1 transcripts in all analyzed tissues. Plasmids encoding green fluorescent protein-tagged genes (GFP-EEG-1) were transfected into Chinese hamster ovary cells for localization and functional assays. In contrast to the diffuse cellular localization of the GFP control, GFP-EEG-1L was detected throughout the cytoplasm and excluded from the nucleus, and GFP-EEG-1S co-localized with aggregated mitochondria. Transfection of both isoforms was associated with significantly increased levels of apoptosis. Stable transfection assays additionally demonstrated decreased growth in those cells expressing EEG-1 at higher levels. Quantitative PCR analyses of mRNA obtained from differentiating erythroid cells from blood donors were performed to determine the transcriptional pattern of EEG-1 during erythropoiesis. EEG-1 expression was highly regulated with increased expression at the stage of differentiation associated with the onset of global nuclear condensation and reduced cell proliferation. We propose that the regulated expression of EEG-1 is involved in the orchestrated regulation of growth that occurs as erythroblasts shift from a highly proliferative state toward their terminal phase of differentiation.

Controlled growth of erythroid tissues represents a precisely regulated balance of proliferation and differentiation that can be examined in vivo and in vitro. Developmentally, relatively quiescent stem cells become committed erythroid progenitors in the bone marrow of adults, and those rapidly proliferating progenitors subsequently undergo terminal differentiation into circulating erythrocytes. The processes of proliferation, differentiation, and apoptosis are balanced to produce enough mature erythrocytes to satisfy the oxygen demands of the host. Acute and chronic tissue hypoxia results in additional eryth-roid stress responses (1). The erythroid lineage is one of the most prolific developmental pathways of adult human life with ϳ10 11 mature erythrocytes produced weekly. In addition to normal erythropoiesis, dysregulation of those mechanisms that control erythroid growth may result in anemia, polycythemia, or erythroleukemia in humans (2). Hence, the well defined relationship between hypoxia and erythroid cell production as well as the association between dysregulated erythroid growth and disease make examination of erythroid growth and differentiation both interesting and relevant.
Erythropoiesis is primarily regulated by the hormone erythropoietin released from the kidneys in response to hypoxia. Erythropoietin is thought to permit or promote the survival of erythroid progenitors, and erythroblasts undergo apoptosis in its absence (3). In addition to erythropoietin, stem cell factor may be involved in a proliferation response to acute erythroid stress (4). At the opposite end of the spectrum, recovery from erythroid stress may be modulated by paracrine mechanisms involving the Fas ligand (5). Hence, several environmental signal transduction cascades exist to regulate the absolute number of circulating erythrocytes generated by each progenitor cell. The switch from committed progenitor cell proliferation toward terminal erythroid differentiation also depends upon regulation of cell cycle kinases and their inhibitors (6). Erythroid proliferation and differentiation are controlled globally by expression of erythroid-specific genes as well as more generic factors. Indeed, gene disruption studies have demonstrated that expression of specific transcription factors like GATA-1 (7), as well as those involved in more general developmental regulators like rbtn2, are required for effective erythropoiesis to occur (8).
In addition to the expression of several genes already known to be involved in the control of erythroid proliferation, differentiation, and apoptosis, we hypothesized that novel growthregulating genes may be identified by gene profiling of developing erythroid cells. We described previously (9) a large number of growth-related genes that are up-regulated in response to the hormone erythropoietin. Those gene profiles also include numerous genes known to be generally involved in tissue differentiation and apoptosis. Here we report the identification, cloning, and initial characterization of a novel growth-regulating gene called EEG-1 expressed during terminal erythroid maturation.

MATERIALS AND METHODS
EEG-1 cDNA Cloning-Informatic analysis of transcription profiles of human erythroid cells (hembase.niddk.nih.gov/) was used to identify the gene studied here. Experimentally, forward 5Ј-TGCTAAAACTTG-AGGCTGAGAA-3Ј and reverse 5Ј-AGTAGCTGAAGGCGGTTGTGA-C-3Ј primers were used to amplify template cDNA from a human bone marrow library by PCR amplification in a GeneAmp PCR System 9700 (PerkinElmer Life Sciences) with the following conditions: one cycle at 94°C for 1 min; 30 cycles at 60°C for 30 s, 72°C for 3 min, and 94°C * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ To whom correspondence should be addressed: Laboratory of Chemical Biology, Bldg. 10, Rm. 9B17, National Institutes of Health, Bethesda, MD 20892. Tel.: 301-402-2373; Fax: 301-402-0101; E-mail: jm7f@nih.gov. for 30 s; and 72°C for 10 min. The amplified products were subcloned into the pCR R 4-TOPO R vector (Invitrogen) and sequenced. Amplification of cDNA 3Ј-and 5Ј-ends was then performed using a human bone marrow cDNA library to define the transcriptional boundaries of the gene. Full-length cDNA was then amplified by reverse transcription (RT) 1 -PCR with cDNA prepared from human erythroid progenitor cells, using the forward primer derived from 5Ј-end cDNA sequence and reverse primer derived from 3Ј-end sequence. The PCR product was purified on an agarose gel and subcloned into pCR R 4-TOPO R vector (Invitrogen). The nucleotide sequences were determined using an ABI prism dye terminator cycle sequencing kit (PerkinElmer Life Sciences) and an ABI Prism 310 genetic analyzer. Chromosomal location, intronexon boundaries, and homology scores were assigned by comparing the cDNA sequence with those deposited in GenBank TM using BLAST programs (www.ncbi.nlm.nih.gov/).
RT-PCR Analyses-To assess the relative expression level of EEG-1L and EEG-1S transcripts in human tissues, the multiple tissue cDNA panel I (K1420-1) and human immune system MTC TM panel (K1426-1) from Clontech were used as templates of real time PCR. Quantitative real time PCR was performed by an ABI Prism 7700 sequence detection system (PerkinElmer Life Sciences, Applied Biosystems) using genespecific forward and reverse primers and dual-labeled fluorogenic internal probe. Dual-fluorescent nonextendable probes labeled with 6-carboxyfluorescein (FAM) at the 5Ј-end and with 6-carboxytetramethylrhodamine (TAMRA) at the 3Ј-end were used for detection. The primers and probes for EEG-1L (forward 5Ј-CTGAGTGTTGAAGACCAGATGGAG-3Ј and reverse 5Ј-CGTTCCTACCACTGCTTTCTCACTA-3Ј, probe 5Ј 6FAM-AGTCATCCTTGTACTTTTGGGACCTTTTGGAAG-TAMRA 3Ј) and EEG-1S (forward 5Ј-TTTTCAAAACTGACCTGCCCTGA-3Ј and reverse 5Ј-GGATACTTTCAAAATAGCCTGAGTTCA-3Ј, probe 5Ј 6FAM-AGAA-ATGAAAGTCTGAGACAAACACTTGAAGGA-TAMRA 3Ј) were designed using the PRIMER EXPRESS software (PerkinElmer Life Sciences). Amplifications were performed at 95°C for 10 min and amplified for 40 cycles of 15 s at 95°C and 60 s at 60°C. Serial dilutions of plasmid DNA were amplified in each experiment for calculation of the mRNA expression levels of the target genes according to the manufacturer's protocol. For RT-PCR, the normalized cDNA were amplified by PCR denaturation at 94°C for 1 min; 30 cycles at 94°C for 30 s, 60°C for 30 s, 72°C for 1 min 30 s and extension at 72°C for 5 min using sense (5Ј-TGCTAAAACTTGAGGCTGAG-3Ј) and antisense (5Ј-GT-TCAGCAATTTAGACAGTA-3Ј) primer pairs located in exon 4 and exon 6, respectively. For analyses of erythroblasts, total cellular RNAs from human erythroid progenitor cells were isolated using the TRIzol Reagent (Invitrogen) according to the manufacturer's instructions. The first strand cDNA was synthesized by Superscript II reverse-transcriptase (Invitrogen) with oligo(dT) 12-18 primer from 1 g of total RNA. The amplification was carried out in 50 l of reaction mixture containing 2 l of the first strand cDNA products, 10 M each of the sense and antisense primer, and 5 units of TaqDNA polymerase (Invitrogen) using a similar PCR condition described above. In addition to EEG-1 transcripts, other growth-related genes were examined by RT-PCR from the same cDNA templates.
Transfection with EEG-1 Expression Vectors-To insert the fulllength EEG-1 gene to pEGFP-C1 expression vector (Clontech), a forward primer containing the added XhoI site at the upstream of the start codon (5Ј-CGCTCGAGGGATGGAAGTACAAGTATCT-3Ј) and a reverse primer including the EcoRI site at downstream of stop codon (5Ј-CG-GAATTCTGTACTGACTTTCAATCTTG-3Ј) were used. The PCR products were digested with the XhoI/EcoRI and inserted in-frame into the XhoI and EcoRI site of pEGFP-C1 mammalian expression vector (Clontech). All of the constructs were confirmed by DNA sequencing.
Chinese hamster ovary (CHO) cells were used for all transfection experiments and maintained in Ham's F-12 medium supplemented with 10% fetal bovine serum. Six-well plates were seeded with 1.6 ϫ 10 6 cells per well in 2 ml of medium. Cells were transiently transfected the following day with 1.5 g of each pEGFP vector, GFP-EEG-1L and GFP-EEG-1S, using PolyFect transfection reagent (Qiagen) according to the manufacturer's instructions. For stable transfections, the cells were maintained in culture medium containing G418 (GIBCO)(0.8 mg/ ml) for 2 weeks.
Flow Cytometry and Confocal Microscopy-Flow cytometric analyses were performed using an EPICS ELITE ESP flow cytometer (Beckman Coulter), and 10,000 cells were analyzed from each sample. Experiments were performed in triplicate with similar results. To determine the expression level of the EEG-1L or EEG-1S, the cultured monolayers were washed and trypsinized to generate cell suspensions. GFP expression was quantitated by calculating the percentage of cells with negative, low, medium, and bright GFP intensities. GFP-expressing populations were identified by fluorescence at levels greater than 2 S.D. above the control population. Fluorescence microscopy was also performed after transfection for comparison. A Zeiss LSM 410 confocal microscope equipped with an external krypton/argon laser was used for localization studies. Image processing was performed using IPLabs (Scanalytics) and Adobe PhotoShop 5.0. Mitochondrial staining was demonstrated by incubation with 50 nM MitoTracker Red CMXRos (Molecular Probes) according to the manufacturer's protocol.
Detection of Apoptosis-Apoptosis studies were performed using two methods. First, dual-color flow cytometry was used to quantitate annexin V-PE staining of GFP-expressing cells 24 h after transfection using an Apoptosis Detection Kit I (BD Pharmingen). For confirmation TUNEL staining was performed with an In Situ Cell Death Detection Kit, TMR Red (Roche Applied Science). The percentages of apoptotic cells among the transfected population were calculated by flow cytometry analyses of Ն5000 transfected cells (annexin V-PE) versus microscopic analyses of Ն100 cells (TUNEL) in triplicate experiments. All statistical analyses were performed by Student's t test.

RESULTS
Identification and Cloning of Human EEG-1 cDNA-As part of a systematic effort to identify and characterize novel genes expressed during erythropoiesis, we identified two erythroid expressed sequence tags (EST). As shown in Fig. 1, those EST (AX57C03 and CL39A09) are homologous to regions in close proximity on chromosome 12, so we speculated that they might encode the same gene. Cloned and sequenced cDNA fragments from bone marrow and erythroid cell libraries were used to generate a 4.3-kb full-length cDNA fragment. Sequencing analysis demonstrated the presence of two EEG-1 cDNAs, which we term EEG-1L and EEG-1S.
EEG-1L and EEG-1S were encoded by a 45-kb region of chromosome 12p11.21. Comparison of cDNA and genomic sequences suggested that human EEG-1 gene consists of at least 18 exons (Fig. 1B). All of the intron-exon junctions exhibit consensus sequences of eukaryotic splice junctions G(T/A)G rule (Table I). A comparison of the EEG-1L and EEG-1S revealed that they were generated as splice variants of the same gene. EEG-1L lacks the 147-bp fragment corresponding to exon 13, whereas EEG-1S lacks the 83-bp fragment corresponding to exon 5 (Fig. 1C). Beginning with an ATG initiation codon located in exon 1, EEG-1L contains an open reading frame encoding a protein of 1077 amino acids. In contrast, the loss of exon 5 in EEG-1S results in a shifted reading frame that introduces a stop codon (TGA) within exon 6. The predicted protein is truncated by 279 amino acids in its C-terminal protein (Fig. 1C). A Kozak consensus sequence is present at the designated ATG, and the putative polyadenylation signal sequence AATAAA is found 26 bp upstream of the poly(A)-tail. No larger cDNA clones were generated by further attempts to amplify the cDNA 3Ј-and 5Ј-ends. Comparisons of an EEG-1 cDNA sequence to the GenBank TM data base using the BLAST program revealed significant homology at protein or nucleotide levels among all the organisms examined except Caenorhabditis elegans. The EEG-1 cDNA sequence was submitted to the GenBank TM with accession number AY074491. Analysis of protein structure with the PROSITE data base revealed that the EEG-1L contains a C-terminal gC1q-like domain with significant homology to the gC1q-like domain of a number of proteins (10). Other predicted motifs include several putative posttranslational phosphorylation motifs, one tyrosine-kinase phosphorylation site, one cAMP-and cGMP-dependent protein kinase phosphorylation site, several casein kinase II and pro- 1 The abbreviations used are: RT-PCR, reverse transcription-PCR; CHO, Chinese hamster ovary; GFP, green fluorescent protein; TAMRA, 6-carboxytetramethylrhodamine; TUNEL, terminal dUTP nick-end labeling; ESTs, expressed sequence tags; G3PDH, glyceraldehyde-3-phosphate dehydrogenase; FAM, 6-carboxyfluorescein. tein kinase C phosphorylation sites, and the N-terminal nuclear localization sequence.
Tissue Distribution of EEG-1-RT-PCR was performed to determine the tissue distribution of EEG-1 isoforms due to low level signals detected by Northern analysis. Quantitative PCR and an RT-PCR assay with primers spanning exon 5 were designed to identify the relative abundance of the splice variant identified in EEG-1S among a range of tissues (Fig. 2). As shown, both EEG-1L-and EEG-1S-derived products (320-versus 235-bp products) were detected in all the tissues tested at generally low levels (Ͻ400 copies/ng mRNA). Quantitative PCR provided a comparison of the transcriptional expression levels of EEG-1L and EEG-1S among several tissues. The highest level of EEG-1L was detected in the brain (117 Ϯ 4 molecules/ng mRNA), and EEG-1S was most abundant in the spleen (376 Ϯ 36 molecules/ng mRNA).
Intracellular Localization and Functional Analysis of EEG-1-To gain insight into the subcellular localization and function of the two EEG-1 isoforms, the EEG-1L and EEG-1S transcripts were fused to the C terminus of a pEGFP expres-sion vector plasmid, and the fusion constructs were transfected into CHO cells. The cellular localization of the GFP-tagged proteins seen in a majority of cells (experiments in triplicate) are shown by the representative images in Fig. 3. Expression of the control GFP vector demonstrated the expected distribution in both the cytoplasm and nucleus. Transfection of the GFPtagged EEG-1L protein exhibited a diffuse pattern throughout the cytoplasm, but exclusion from the nucleus was noted. The pattern of GFP-tagged EEG-1S fluorescence was also distinct. Whereas low levels of GFP-EEG-1S were detectable in the cytoplasm, more intense fluorescence was seen in the perinuclear regions of those cells. Co-staining with a mitochondrialocalized dye (MitoTraker) demonstrated that the perinuclear GFP-EEG-1S fluorescence was due to mitochondrial localization of that protein. Compared with the expected punctate staining pattern of mitochondria seen in GFP (control) and GFP-EEG-1L transfections, the GFP-EEG-1S containing mitochondria appeared to be aggregated.
In addition, the EEG-1-transfected cells were cultured for 2 weeks in the presence of G418 to further correlate EEG-1 expression and cell growth. After that period, the culture plates were washed and examined by fluorescent microscopy. Unlike the clear GFP signals detected among the control GFP colonies after 2 weeks, dim or absent fluorescence was observed by microscopy in EEG-1L-or EEG-1S-transfected colonies. The colonies were also examined by flow cytometry to quantitate differences in the level of transgene expression. As shown in Fig. 4, expression of the control and tagged constructs was present at all levels when measured 24 h after transfection. After 2 weeks, almost no cells demonstrated high level expres-sion in the EEG-1L or EEG-1S. This negative correlation between EEG-1 expression and growth was also present among populations expressing GFP at medium signal intensities in repeated experiments. As shown in Fig. 4, 17% of the control cells, compared with only 2% of tagged EEG-1L-transfected pools and 0% of the EEG-1S-transfected pools, possessed medium level fluorescence. These data demonstrate a growth disadvantage associated with the expression of tagged EEG-1L and EEG-1S.
The loss of high level EEG-1 expression in Fig. 4 led us to explore further the possibility that EEG-1 gene expression may be involved in the regulation of cell growth or death. Thus, we examined the percentage of apoptotic cells in those transfected  populations compared with the control vector (Fig. 5). Two methods (annexin-V binding and TUNEL) were used to analyze the apoptotic features of the transfected populations. Annexin V binds to apoptotic cells due to exposure of phosphatidylserine translocation from the inner to the outer layer of the plasma membrane at an early stage of apoptosis (11). At 24 h posttransfection, the percentage of annexin V-positive cells was significantly increased in EEG-1-transfected cells compared with cells transfected with the empty vector control (GFP control, 20 Ϯ 9%; GFP-EEG-1L 51 Ϯ 7%, p Ͻ 0.05; GFP-EEG-1S, 59 Ϯ 2%, p Ͻ 0.05). For confirmation, we also measured apo-ptotic cell death 24 -72 h after transfection by TUNEL assay. This assay detects cellular endonuclease-mediated ordered DNA fragmentation, a later event in apoptotic cell death (11). As shown in Fig. 5B, the TUNEL results confirmed the association between EEG-1 expression and apoptosis. A significant increase in TUNEL-positive cells was observed over the 72 h after transfection with GFP-tagged EEG-1L or EEG-1S. The percentage of TUNEL positive cells did not significantly change after transfection of the control GFP plasmid.
Since cell growth appears to be inhibited by EEG-1 gene expression, we next determined the relative level of EEG-1 gene expression during erythropoiesis. For this, CD34 ϩ cells from normal blood donors were cultured for 14 days for semisynchronous development of mature erythroblasts over that period (12). To determine the expression pattern of EEG-1 during erythroid differentiation of CD34 ϩ cells, quantitative RT-PCR was performed with mRNA from cells collected every 48 h over the 2-week culture period. As shown in Fig. 6, EEG-1 expression is highly regulated during erythroid development. Both isoforms were present at low levels during the initial culture period as the CD34 ϩ population underwent rapid proliferation. The transcription levels of both EEG-1 isoforms then rapidly increased and reach the maximal level on day 8 (984 Ϯ 96, p ϭ 0.001). After day 10, their expression was down-regulated. On all days, the EEG-1S isoform was detected at higher levels. Combined with the low level of expression detected in unfractionated bone marrow, this highly regulated pattern suggests EEG-1 exerts its growth-related effects among cellular subsets rather than throughout hematopoietic development. In this culture system, days 7-10 represent a distinct erythroid transition from a highly proliferative phase toward terminal erythroid maturation marked by a loss of cell proliferation, nuclear condensation, shrinkage of cell size, and high level expression of hemoglobin associated with terminal erythroid maturation (12). Fig. 6 also demonstrates a similar pattern of regulated expression during erythropoiesis for other growth regulatory genes including those associated with cell cycle regulation and the anti-apoptosis gene BCL-2. The EEG-1 expression pattern most closely matched those genes involved with the regulation of the cell cycle (P21 and CDK2). Similar to EEG-1, the expression of the cyclin-dependent kinase inhibitor P21 gene reached peak levels on days 8 -10. Increases were detected in the gene expression level of the CDK2 transcript slightly earlier during the culture period. Interestingly, increased levels of the anti-apoptotic BCL-2 mRNA occurred earlier during the 1st week of the culture period. No significant changes were observed in control G3PDH transcript. DISCUSSION Erythropoiesis entails a precisely controlled developmental process that integrates the growth-related themes of cell proliferation, differentiation, and apoptosis with erythroid-specific gene expression. Hence, we predicted previously that gene expression profiles from highly purified and developmentally staged erythroid cells provide a robust source of information regarding elements involved in the regulation and control of erythroid cell growth. In this study, we identified a previously uncharacterized growth-related gene that we call EEG-1. The gene is expressed from a 45-kb locus on human chromosome 12p11.21, and two EEG-1 splicing variants present in the erythroid lineage were studied. Informatic analysis of the EEG-1L open reading frame identified this gene as a new member of a family of genes containing a C1q globular domain (10). The family of genes containing a C1q globular domain includes the complement protein C1q itself (13), proteins associated with animal hibernation (14), and precerebellin (15). However, experimental evidence is required to determine whether this and other genetic motifs in the EEG-1 transcripts are functional in the translated protein. Of note, a cellular receptor for C1q globular proteins is located in mitochondria and elsewhere in cells (16).
In order to learn more about distribution and possible function of EEG-1, we transfected plasmids encoding GFP fused to the EEG-1L and EEG-1S isoforms into CHO cells. Transient transfections with plasmids expressing the proteins fused to green fluorescent markers, the cellular localization of the tagged EEG-1L, and EEG-1S proteins were visualized. Unlike the control transfections with GFP alone, neither EEG-1 product was accumulated in the nucleus, despite the presence of a nuclear localization motif. Previous studies have demonstrated that those signals are not always predictive of nuclear uptake (17). The tagged EEG-1L was distributed throughout to the cytoplasmic compartment. In contrast to tagged EEG-1L, the EEG-1S signals were co-localized with aggregated mitochondria. Although expression of both isoforms was associated with apoptosis, the cellular localization of EEG-1S suggests this isoform may be involved in apoptotic pathways involving mitochondria. Several other apoptosis-related genes including the BCL-2 group of proteins exert their effects in mitochondria (18).
Among developing erythroid cells, transcription of the gene appears to be regulated with a detectable rise and fall 8 -10 days after exposure of CD34 ϩ cells from peripheral blood to the hormone erythropoietin. Based upon previous studies of morphology and cell cycling during erythropoiesis, this period correlates with a major transition from the proerythroblast to basophilic normoblast stage of erythroid differentiation and the transition to lower levels of cell cycling (12). This cellular transition also marks the onset of nuclear condensation and a more rapid accumulation of hemoglobin. The pattern of increased EEG-1 on days 8 -10 correlated most closely with genes involved in the regulation of the cell cycle including the cyclindependent kinase inhibitor P21. Increased expression of P21 has been associated previously with induced differentiation and withdrawal from the cell cycle in hematopoietic cells (19). P21 up-regulation also occurs when cells undergo terminal differentiation after stem cell factor is withdrawn from proliferating erythroblasts in culture (20). These patterns suggest that increased expression of the apoptosis-associated, growth inhibiting EEG-1 gene during erythropoiesis is involved in the developmental transition from rapid proliferation toward terminal differentiation. Co-regulation of pro-apoptotic genes with those involved in withdrawal from the cell cycle has been associated previously with the tumor-suppressive effects of p53 (21). Of note, the infrequency of erythroleukemia among more differentiated erythroblasts suggests that co-regulation of cell cycle inhibitory and pro-apoptotic genes like EEG-1 may serve a cancer-protective role during the final stages of erythropoiesis. The concept that pro-apoptotic factors play a role in erythroid differentiation is supported by the recent finding that caspase expression is important (22). Interestingly, during the final stages of maturation, erythroblasts possess several features usually attributed to apoptosis including a reduction in size, withdrawal from the cell cycle, and nuclear condensation. Since EEG-1 expression is not restricted to erythroid cells, this gene may serve a more general role in cellular differentiation. Perhaps regulated expression of growth-related genes like EEG-1 are important developmentally for cells to undergo terminal differentiation rather than apoptosis in association with a withdrawal from the cell cycle.