Expression and regulation of mouse SERDIN1, a highly conserved cardiac-specific leucine-rich repeat protein

Authors


Abstract

Despite recent progress, the precise mechanisms responsible for vertebrate cardiac development are still enigmatic. Better understanding of cardiac biology and disease necessitates identification and analysis of a full spectrum of regulatory and structural proteins specific to the developing heart. By performing an in silico screen, we identified a cardiac-specific gene we named Serdin1. The Serdin1 gene is conserved, and the message is restricted to the heart in several vertebrate species, thus implicating Serdin1 as an important gene in cardiac development. In situ hybridization confirmed that the Serdin1 message is cardiac-specific in mice as early as embryonic day 8.5. Antibody staining demonstrated predominantly nuclear staining in immortalized cardiac cell lines (P19 and HL-1) and proliferating cultured cardiomyocytes, whereas in vivo SERDIN1 localizes to I bands of the sarcomere. Seven kilobases of the upstream regulatory sequence of Serdin1 is sufficient for cardiac-specific expression. Computer analysis revealed an 80-bp homologous region between the mouse and the human Serdin genes that contains GATA, SRF, and MEF sites. Cardiac specificity and localization patterns suggest that SERDIN1 is intimately integrated with the molecular pathways controlling cardiogenesis in vertebrates. Developmental Dynamics 233:540–552, 2005. © 2005 Wiley-Liss, Inc.

INTRODUCTION

The complexity of cardiac development is reflected by the high incidence of congenital heart disease in humans (5–8 of 1,000 births; Bower and Ramsay, 1994; Wren and O'Sullivan, 2001; McConnell and Elixson, 2002). Multiple cellular interactions and morphogenic cues are required to form a primitive embryonic heart and to reshape it into the mature four-chamber organ for timely adaptation to its central function (Mably and Liew, 1996; Olson and Schneider, 2003). The identification of novel cardiac-specific proteins and the examination of their function is essential to improve our incomplete knowledge of cardiac-specific gene expression and human cardiac development (Small and Krieg, 2004).

The conventional approach for identifying novel important genes in mammals relies mostly on either positional cloning or the isolation of homologous genes followed by expression analysis and reverse genetics. Although positional cloning is nonbiased, the reverse genetics (gene targeting) is still rather cumbersome in mice, so it is normally reserved for genes with well-characterized functional domains. We reasoned that several cardiac-specific proteins with important roles in heart morphogenesis could have eluded us so far because they lack typical functional motifs used in most screens. In this project, we concentrated on cardiac specificity as our major objective (rather than the presence of recognizable functional domains). We took advantage of the recent advances in the quality and quantity of publicly available computer databases containing randomly sequenced expressed sequence tags (ESTs; e.g., Marra et al., 1999). This in silico approach has been used successfully to identify tissue-restricted genes, including cardiac-specific genes (Wang et al., 2001). By performing the screen, we identified a cardiac-specific gene we called Serdin1 that encodes a novel leucine-rich repeat (LRR) protein. LRRs have been found in proteins with diverse functions. In LRR-containing proteins the number of repeats, their lengths, as well as the LRR consensus sequences are also quite diverse. LRR-containing proteins include hormone and tyrosine kinase receptors, cell-adhesion molecules, bacterial virulence factors, enzymes, and extracellular matrix-binding glycoproteins (for review, see Enkhbayar et al., 2004). In situ hybridization and Northern analysis confirmed that the Serdin1 message is cardiac-specific in mice. Antibody staining demonstrated predominantly a nuclear localization of SERDIN1 in differentiated beating P19 cells, an HL-1 cardiomyocyte cell line, and freshly isolated mouse embryonic cardiomyocytes. In the adult heart, the protein localizes to I bands of the sarcomere. Seven kilobases of the upstream regulatory sequence of mouse Serdin1 is sufficient for cardiac-specific expression in transgenic mice. A single short (∼80 bp) region of homology with the human gene is located approximately 3 kb upstream of the Serdin1 start site. This homology is likely to be significant because its sequence contains several binding sites for cardiac-restricted transcription factors. Hence, in addition to the protein sequence, the cardiac-specific regulation of the Serdin1 gene is likely to be conserved between mice and humans. In summary, cardiac specificity and localization patterns suggest that Serdin1 is intimately integrated with the molecular pathways controlling cardiac development and function in vertebrates.

RESULTS

In Silico Isolation of Mouse Serdin

To isolate novel cardiac-specific transcriptional regulators, we performed an in silico screen for previously uncharacterized ESTs that are highly enriched in murine cardiac-specific libraries. We used a RIKEN full-length enriched 13-day embryo heart library (NCBI dbEST Library ID.5466). This library contains 8,308 clones. We selected RNA tags that were present in this library and appeared in other cardiac libraries, but were absent elsewhere. By using this approach, we identified several such tags that clearly encoded proteins. Because we were mostly interested in the identification of nuclear regulatory factors, we also screened for the presence of a nuclear localization signal in the candidate proteins.

Expression of AK084466.1 EST appeared to be highly specific to mouse heart libraries. All mouse EST tags (23 tags as of 4/27/03) corresponding to AK084466.1 have been isolated exclusively from various murine cardiac libraries (UniGene cluster 37320). The conceptual translation of this EST is a 274 amino acid sequence that consists mostly of LRRs and contains a putative nuclear localization motif (PRxARR; Cokol et al., 2000). Using this protein sequence as a query against translated EST database TBLASN, we identified ESTs in human, pig, rat, chicken, and frog that are likely to be the products of orthologous genes in these species. Similarly, all of these ESTs were isolated from cardiac libraries. Although we did not come across fish ESTs, an ortholog of AK084466.1 is present in the Fugu genome (Fig. 1). Based on such high specificity of expression, we named this gene Serdin1 (from Russian “serdtse,” heart).

Figure 1.

The alignment of deduced amino acid sequences encoded by available expressed sequence tags (ESTs) from mouse, rat (XP_235162), human (XP_370698), pig (AW430936), chicken (BU360047), frog (BG578625), and fish (CAAB01000017); part of the human and all of the fugu amino acid sequence are derived from conceptual translations of the genomic clones rather then ESTs. Sequences were aligned by using the program PILEUP (GCG, version 8.0) and shaded by using the program BOXSHADE. The darker shading represents identity at a given residue, whereas light shading represents amino acid similarity between the proteins.

Analysis of Serdin RNA Expression

To validate cardiac-specific expression of Serdin1 RNA, we performed in situ hybridization analysis. We generated a Serdin1 cDNA fragment by reverse transcriptase-polymerase chain reaction (RT-PCR) from embryonic day (E) 13.5 heart cDNA, cloned it under the T7 promoter, and used it as a probe for whole-mount in situ analysis. We could not detect any expression in E7.5–E8.0 embryos (data not shown); however, Serdin1 RNA appears in the mouse heart at E8.5 and continues to be strongly expressed there at later stages (Fig. 2A–C and data not shown). In the heart, Serdin1 expression is localized to the myocardium (Fig. 2D). Our subsequent analysis of RNA from mouse adult tissues on a multiple-tissue Northern blot confirmed that expression of Serdin1 is also restricted to the adult heart tissue (Fig. 3A, lane1). In accordance with EST database information, Northern blot analysis identified a major RNA band at ∼1.35 kb. Importantly, it is absent from skeletal muscle (Fig. 3A, lane 6). This message encodes a protein of a predicted size of ∼31 kDa. Re-probing the blot for a ubiquitously expressed gene (glyceraldehyde-3-phosphate dehydrogenase) demonstrated the presence of RNA in every lane (Fig. 3B). The high specificity of Serdin1 expression to heart was also confirmed by a sensitive RT-PCR analysis (Fig. 3C).

Figure 2.

A–C: Whole-mount RNA in situ hybridization analysis of Serdin1 expression. Serdin1 transcript expression is confined to the developing mouse heart (arrows) in embryonic day (E) 8.5 (A), E9.5 (B), and E12.5 (C) embryos. D: A stained E12.5 embryo was sectioned to demonstrate the expression in the myocardium (m) and not endocardium or epicardium (arrows). Original magnification, ×100. Scale bar = 50 μm in D.

Figure 3.

A: Northern blot analysis of Serdin1 expression in mouse adult tissues. BD-Clontech multiple-tissue Northern (MTN) blot membrane was hybridized with a 32P-labeled 1.3-kb EcoRI fragment of pCR-TOPO-TA-Serdin1. The position of the Serdin1 mRNA band (∼1.3 kb) is indicated by an arrow in A; the identity of the higher migrating bands is currently unknown. B: The blot was rehybridized with EcoRI SerdinR probe (data not shown) followed by a 3-Gapdh probe. ht, heart; br, brain; sp, spleen; lg, lung; li, liver; sk, skeletal muscle; kd, kidney; ts, testis RNA. C: Semi-quantitative reverse transcriptase-polymerase chain reaction (RT-PCR) analysis of Serdin1 expression. cDNA was isolated from adult mouse lung (lg), spleen (sp), liver (li), brain (br), kidney (kd), heart (ht), and skeletal muscle (sk) and subjected to RT-PCR with Serdin1 primers (top) or Hprt primers (bottom). Lane M is the 1-kb DNA marker.

A computer BLAST search using the Serdin1 gene sequence as a queue revealed a close homologue of Serdin1 in the database (Fig. 4A). We termed this gene Serdin-Related, or SerdinR. We could not detect SerdinR expression by Northern blot analysis (data not shown); RT-PCR analysis demonstrated that SerdinR RNA is expressed in several tissues, including heart (Fig. 4B). Similar to Serdin1, the deduced amino acid sequence of SerdinR also contains a putative nuclear localization signal (PRxxARR; Cokol et al., 2000).

Figure 4.

A: The amino acid alignment of the deduced protein sequence of mouse Serdin1 and SerdinR (XM_204937) genes. B: Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) analysis of SerdinR expression. cDNA was isolated from adult mouse lung (lg), spleen (sp), liver (li), brain (br), kidney (kd), heart (ht), and skeletal muscle (sk) and subjected to RT-PCR with SerdinR primers (top) or HPRT primers (bottom). Lane M is the 1-kb DNA marker.

Generation of an Anti-SERDIN1 Antibody

To extend our analysis of Serdin1 expression to the protein level, we generated an affinity-purified anti-SERDIN1 antibody. We confirmed its specificity by transiently transfecting pCMV-FLAG-Serdin1 into HEK 293 cells: the 293 cells do not express the Serdin1 message (data not shown). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis established that the anti-SERDIN1 antibody specifically recognizes a ∼32-kD protein in the pCMV-FLAG-Serdin1–transfected cell extract but not in the control 293 cell extracts (Fig. 5, compare lanes 4 and 2). The antibody also recognizes a single band at ∼32 kDa that is present in the heart (Fig. 5, lanes 3 and 5), but not in the liver protein extract (Fig. 5, lane 1). To further establish the antibody specificity, we performed antibody blocking experiments. The addition of recombinant SERDIN1 protein to the primary antibody incubation step completely blocks the detection of the 32 kDa band (Fig. 6, arrowhead), while it does not interfere with the ability of an unrelated anti-troponin I antibody to recognize its cognate antigen (Fig. 6, arrow). We conclude that the anti-SERDIN1 antibody is specific to SERDIN1.

Figure 5.

Endogenous and ectopic SERDIN1 expression detected by the rabbit anti-SERDIN1 polyclonal antibody. Western blotting analysis was performed with whole cell extracts from the adult mouse liver (lane 1), untransfected 293 control (lane 2), 293 cells transfected with pCMV-FLAG-SERDIN (lane 4), and the adult heart (lanes 3 and 5). Equal amounts of protein extracts (60 μg) were loaded (lanes 1–4); the protein extract in lane 5 was prepared by direct boiling in the sodium dodecyl sulfate loading buffer. The positions of the molecular weight markers are shown on the right; the position of the band corresponding to SERDIN1 is shown (arrow). The doublet in the transfected lane (lane 4) most likely represents the FLAG-tagged and untagged version of the SERDIN1 protein.

Figure 6.

Antigen blocking experiment. Rabbit anti-SERDIN1 or control (mouse anti-troponin I) antibodies were preincubated with increasing concentrations of recombinant GST-SERDIN1 protein and used to probe membrane strips containing cardiac extracts. Lane 1, no ligand; lane 2, 5 μg; lane 3, 50 μg; lane 4, 100 μg; lane 5, 200 μg of recombinant SERDIN1. The positions of the bands corresponding to SERDIN1 (arrowhead) and troponin I (arrow) are shown.

Immunohistochemical Analysis of Serdin1 Expression

By using an affinity-purified anti-SERDIN1 antibody, we examined the expression of SERDIN1 by immunofluorescence microscopy. SERDIN1 protein is heart-specific in E12.5 embryos, and the expression overlaps with staining for cardiac troponin T (cTnT), a marker specific to cardiac muscle (Fig. 7A). Specificity of SERDIN1 expression in cardiac muscle was also confirmed by costaining of sections of embryonic and adult mouse heart for the endothelial cell marker platelet endothelial cell adhesion molecule-1 (PECAM-1) that shows no overlap of its expression with that of SERDIN1 (Fig. 7B,C). This confirms that expression of Serdin1 is specific to cardiac muscle (see also Fig. 2).

Figure 7.

Immunofluorescence analysis of SERDIN1 expression in frozen sections of embryonic (A,B) and adult mouse heart (C). A: SERDIN1 (green; a) colocalizes with cardiac muscle marker cardiac troponin T (cTnT, red; b). B: SERDIN1 (red; a, c, d, f) does not colocalize with endothelial marker PECAM-1 (green; b, c, e, f) in embryonic heart sections. C: SERDIN1 (red; a, c, d, f, g, i) is not expressed in the epicardium (e, arrowhead) or the endocardium (h, arrows) of adult mouse heart stained for platelet endothelial cell adhesion molecule-1 (PECAM-1, green; b,c, e,f, h,i). Nuclei are counterstained with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI, blue). Scale bars = 200 μm in A (applies to Aa–c), 200 μm in Ba (applies to Ba–c), 50 μm in Bd (applies to Bd–f), 200 μm Cc (applies to Ca–c), and 50 μm in Cf (applies to Cd–f), in Ci (applies to Cg–i).

Based on the presence of a putative nuclear localization signal, we anticipated detecting SERDIN1 in the nucleus; however, SERDIN1 did not appear to localize to the nucleus in cardiac sections (Fig. 7). Because protein localization often provides clues to its molecular function, we attempted to localize transfected SERDIN1-FLAG in HEK293 and NIH 3T3 cells with anti-SERDIN1 or anti-FLAG antibodies by immunofluorescence microscopy. We found diffuse localization throughout transfected HEK293 and NIH 3T3 cells using either antibody (data shown for NIH 3T3 cells in Fig. 8A–D). Although correct nuclear localization of ectopically expressed proteins can often be achieved in transient transfection assays, proper localization of this highly cardiac-restricted protein may require a cell-specific signal(s). Therefore, we switched to P19 cells (Paquin et al., 2002). Serdin1 message is absent in undifferentiated P19 cells (data not shown). Dimethyl sulfoxide (DMSO) treatment leads to differentiation and spontaneous contraction of P19 with concomitant expression of cardiac-specific makers (Gata4, CARP(Ankrd1), and cardiac troponin I), including Serdin1 as ascertained by RT-PCR analysis (data not shown). In P19 cells, SERDIN1 accumulates in the nuclear compartment (compare Fig. 8E and G) and colocalizes with transcription factor GATA4 (nuclear, Fig. 8F) rather than with cTnT (cytoplasmic, compare Fig. 8H and I). Although in the majority of differentiated P19 cells the protein is nuclear, some cells demonstrate a more diffuse localization (data not shown).

Figure 8.

Immunofluorescence analysis of SERDIN1expression in cultured mammalian cells. A–D: NIH 3T3 cells were transiently transfected with SERDIN1 expression plasmid and stained with anti-SERDIN1 antibody (A,B) or anti-FLAG antibody (C,D) followed by an rat anti-rabbit Alexa Fluor 488 antibody (green) or goat anti-mouse Alexa Fluor 488 (green). Nuclei were counterstained with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI, blue). A–D: Cells were visualized either through a single filter to detect SERDIN1 only (A,C) or through a triple filter to detect both (B,D). B,D: The SERDIN1 staining is detected throughout the cell. E–G: Differentiated (beating) P19 cells were costained with SERDIN1 (rabbit) and GATA4 (goat) antibody followed by secondary chicken anti-rabbit (green, Santa Cruz) and donkey anti-goat Alexa 555(red, Alexa). E–G: SERDIN1 (E) and GATA4 (F) are coexpressed and localized to the nuclei (G) in these cells. H–J: Differentiated (beating) P19 cells were costained with SERDIN1 (rabbit) and cTrT (mouse) antibody followed by secondary chicken anti-rabbit (green) and goat anti-mouse Alexa Fluor 555 (red, Alexa). To clearly observe cytoplasmic staining, a group of well-separated differentiated cells is shown. SERDIN1 (H) is localized to the nuclei (J), while cTnT is cytoplasmic (I). K–M: HL-1 cardiomyocytes were stained and photographed as described above to contrast cytoplasmic expression of cTrT (K,M; red) with nuclear expression of SERDIN1 (L; green). N–S: Embryonic cardiomyocytes were stained and photographed as described above. N–P,R,S: In cTrT-positive cells (red; N,P), SERDIN1 (green; O,R) is nuclear (DAPI; P,S). Q: As in P19 cells, SERDIN1 colocalizes with GATA4 (red) to the nuclei.

DMSO-induced differentiation of the pluripotent P19 cells leads to the appearance of beating cardiomyocytes; however, the whole population remains relatively heterogeneous with respect to cardiomyocyte differentiation (Boheler et al., 2002). To extend our analysis to a homogeneous cell population, we examined SERDIN1 expression in HL-1 cardiomyocytes and embryonic cardiomyocytes in culture. HL-1 cells are derived from the AT-1 mouse atrial cardiomyocyte tumor lineage and can be repeatedly passaged in culture without losing its cardiac-specific phenotype (Claycomb et al., 1998; White et al., 2004). It has been proposed that HL-1 cells have an ultrastructure that is similar to a mitotic embryonic myocyte, which is less differentiated and less organized than that of mature cardiomyocytes (Claycomb et al., 1998; White et al., 2004). As in differentiated P19 cells, SERDIN1 is localized in the nucleus in HL-1 cells (Fig. 8K–M). Additionally, in freshly isolated beating E12.5–E13.5 embryonic cardiomyocytes, we also observe nuclear localization of the SERDIN1 protein; SERDIN1 colocalizes with GATA4 and not cTnT (Fig. 8N–S).

To better localize SERDIN1 in adult cardiac tissue, we performed deconvolution analysis. In frozen sections of adult rat heart, endogenous SERDIN1 was detected as a clear, narrow doublet in the I band (flanking the Z line) of sarcomeres (Fig. 9a). Its distinctive “dot-like” pattern suggests that it might be membrane-associated. To better define the region where SERDIN1 is localized, double-labeling experiments with antibodies generated against SERDIN1 together with antibodies generated against the N2A I band epitope of sarcomeric titin were performed (Fig. 9a–c). This experiment demonstrated that, although staining for SERDIN1 partially overlaps with N2A staining, its staining pattern appears to be situated further out in the I band, away from the Z line.

Figure 9.

a,b: Confocal immunofluorescence analysis of frozen sections of adult rat muscle stained for SERDIN1 (a) and an I band epitope of titin, N2A (b) as described (Centner et al., 2000). c: Comparison of the merged image (see also inset) reveals that SERDIN1 (green) is localized within the I band at a further distance from the Z line (that is, closer to the center of the sarcomere) than titin N2A (red). Scale bar = 10 μm in c (applies to a–c).

To confirm that the same protein is being recognized in the adult heart tissue and in the nuclei of cultured cells, we compared protein extracts isolated from cultured embryonic cardiomyocytes and adult hearts by Western blot analysis. The presence of a single band at the same mobility in both extracts strongly suggested that the anti-SERDIN1 antibody recognizes the same protein (Fig. 10). To explain the differential localization patterns detected for SERDIN1, we reasoned that SERDIN1 is nuclear only in actively proliferating cultured cells. Indeed, in long-term cultures of embryonic cardiomyocytes (∼2 weeks after isolation and plating) a striated pattern of SERDIN1 staining was observed (Fig. 11a,b).

Figure 10.

SERDIN1 expression in cultured embryonic cardiomyocytes. Western blotting analysis was performed with the whole cell extracts from the adult mouse heart (lane 1), 293 control (lane 2), adult mouse liver (lane 3), and embryonic mouse cardiomyocytes cultured for 48 hr (lane 4). The positions of the molecular weight markers are shown on the right; the position of the band corresponding to SERDIN1 is indicated (arrow).

Figure 11.

a,b: Isolated mouse fetal cardiomyocytes after 2 weeks in culture demonstrate a clear striated staining pattern (arrows) of SERDIN1 staining. c: No signal is detected with the preimmune antibody. Scale bars = 10 μm in a–c.

Transgenic Analysis of Serdin1 Expression in Mice

Analysis of Serdin's immediate genomic neighbors indicates that the nearest 5′ gene is located 20 kb upstream of Serdin1. To determine the cis-regulatory elements of the SERDIN1 gene that are responsible for its cardiac-specific expression, we isolated the genomic sequences corresponding to the Serdin1 gene from mouse bacterial artificial chromosomes (BACs). We generated several transgenic vectors with Serdin1 regulatory sequences driving bacterial β-galactosidase (lacZ) expression, Serdin1-lacZ. Transgenic constructs were injected into fertilized eggs, and F0 transgenic embryos were collected and analyzed at E12.5. A 7.3-kb fragment of the upstream genomic sequence of Serdin1 was sufficient to confer cardiac-specific expression to the lacZ reporter at E12.5 in several F0 embryos (Fig. 12A,B). In the F1 newborn animals expression of Serdin1 is also cardiac-restricted (Fig. 12C).

Figure 12.

A–C: Seven kilobases of the Serdin1 5′-regulatory region is sufficient for cardiac-specific SERDIN1 expression. A,B: A right lateral view of the representative F0 transgenic embryonic day (E) 12.5 embryo is shown in A and the transverse section of another embryo is photographed with Nomarski optics in B (original magnification, ×40). Expression of lacZ in these embryos recapitulates endogenous expression (compare with Fig. 2). C: Cardiac-specific expression in F1 newborn animals is shown; notice the absence of expression in the lung. D: A sequence alignment of the highly homologous stretch of mouse and human SERDIN1 regulatory region. The darker shading represents identity at a given nucleotide, whereas light shading shows purine conservation. The binding sites for GATA (−), SRF(=), and MEF-2 (*) are underlined. Scale bar = 200 μm in B. LA, left atrium; RA, right atrium; LV, left ventricle; RV, right ventricle; H, heart.

High conservation of SERDIN1 (Fig. 1) between mouse and human prompted us to compare the regulatory sequences of the corresponding genes within this 7.3-kb region. Computer analysis identified only one, highly conserved, region of homology at ∼3 kb from the Serdin1 RNA starting site containing the binding sites for GATA, SRF, and MEF transcriptional factors (Fig. 12D).

DISCUSSION

In this report, we describe the cloning and expression analysis of a novel LRR protein, SERDIN1. With the exception of leucine repeats and a putative nuclear localization signal, SERDIN1 lacks any obvious motifs that provide clues to its cellular role. Clearly, the presence of the LRRs per se does not allow assignment of any particular function to a protein. It has been hypothesized by many investigators that LRR modules serve as bridges between molecules involved in diverse cellular processes (e.g., signal transduction, cell adhesion, RNA processing, plant and mammalian immune regulation, and so on; Bell et al., 2003; Conley, 2003; Sawada et al., 2003; Bottcher et al., 2004; Dievart and Clark, 2004). We demonstrated that the Serdin1 gene is conserved and the message appears to be highly (if not exclusively) restricted to the heart in several vertebrate species. By in situ hybridization and Northern analysis, we demonstrated that the Serdin1 message is myocardial-specific in mouse embryos as early as E8.5; immunofluorescent analysis also confirmed that SERDIN1 expression is specific to cardiac muscle.

Despite the presence of a putative nuclear localization motif (PRxARR) SERDIN1 does not localize to the nucleus in heterologous cells. In contrast, antibody staining of endogenous protein reveals predominantly nuclear localization of SERDIN1 in differentiated beating P19 cells, HL-1 immortalized cardiomyocytes, and freshly isolated embryonic cardiomyocytes. Finally, in adult rat heart muscle SERDIN1 localizes to the I bands, flanking the Z disk. The reasons for this differential localization of the protein are unknown, although the nuclear localization of SERDIN1 clearly correlates with proliferative activity of cultured cells. The degree of maturation of the cardiac muscle cells could also cause differential localization of proteins. Growing fibers of skeletal muscle have more active nuclei than fully mature fibers (Newlands et al., 1998); it is possible that, in appropriately matured and terminally differentiated cardiac cells, SERDIN1 is transported out of the nucleus.

Although nuclear localization of SERDIN1 in cultured embryonic cadiomyocytes and immortalized cells (P19 and HL-1) could be a result of the in vitro differentiation or culture in monolayer (see Schaub et al., 1997, for review; Armstrong et al., 2000), it could also suggest that, under certain physiological conditions (e.g., mechanical stress), SERDIN1 may translocate and have a function in the nucleus in vivo. We hypothesize that localization of SERDIN1 to the nucleus in the proliferating cardiomyocytes and to the I band in the sarcomere in the adult cells hints at its potential role as a signal transduction and/or transcriptional regulatory molecule. In this regard, due to the asymmetric nature of the giant sarcomeric titin filaments, as opposed to the thin filaments that are relatively uniform down the length of the filament, it is tempting to speculate that SERDIN1 associates (directly or indirectly) with the I band region of titin. Recent evidence suggests that titin as well as some other I-band titin-associated proteins may have acquired a specialized function as strain sensors of the sarcomere (for review, see Miller et al., 2004). The I band region of titin contains its unique PEVK (rich in proline, glutamate, valine, and lysine) and N2B elements, which are the major source of elasticity and restorative stretch in cardiac muscle (Linke et al., 1999; Cazorla et al., 2000). Part of the stretch sensing/signaling will ultimately require the transfer of information to the nucleus and our data point to SERDIN1 as a candidate molecule for this relay process.

With the notable exception of SerdinR, database analysis of human and mouse genomes did not identify any other members of the Serdin family. Based on the RT-PCR and EST analysis SerdinR is broadly expressed in mouse tissues albeit at significantly lower levels compared with cardiac-specific Serdin1.

To understand the basis of cardiac-specific expression of Serdin1, we performed analysis of its regulatory region in transgenic mice. Seven kilobases of the upstream regulatory sequence of Serdin1 are sufficient for cardiac-specific expression. BLAST comparison demonstrated that mouse and human genes share a short 80-bp stretch of highly homologous sequence that is located approximately 3 kb from the start site of Serdin1 in these species. This region is very similar to a recently reported “heart element” capable of driving pan-myocardial–specific transcription from the Xenopus MLC2 promoter in mouse embryos (Latinkic et al., 2004). The MLC2 “heart element” contains a CarG box flanked by two GATA-binding sites, an arrangement we find in the putative Serdin1 regulatory sequence (Fig. 12D). We are currently testing whether the deletion of this region results in abrogation of cardiac-specific expression. The high degree of homology between the corresponding regulatory regions in mice and humans strongly suggests that the cardiac-specific regulation of the Serdin1 gene is likely to be conserved in humans as well. In summary, cardiac specificity and localization patterns suggest that Serdin1 is intimately integrated with the molecular pathways controlling cardiac development and function in vertebrates.

EXPERIMENTAL PROCEDURES

In Silico Screen for Cardiac-Specific EST Clones

Approximately 16,000 ESTs from the RIKEN full-length enriched, 13 days embryo heart library (dbEST Library ID.5466; http://ncbi.nlm.nih.gov/UniGene/library.cgi?ORG=Mm&LID=5466) were screened against other EST libraries using the BLAST search engine. ESTs enriched in cardiac libraries were selected for further analysis.

Cloning of Mouse Serdin1 and SerdinR

cDNAs.

Total RNA was isolated from various adult tissues using TRIzol reagent (Invitrogen), according to manufacturer's protocol. RNA from embryonic tissues was isolated using Quiagen RNeasy columns. All RNA samples were treated with DNAse I (Promega), phenol extracted, and ethanol precipitated. E13.5 heart and E13.5 brain cDNA were generated using SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen). Serdin1 cDNA was amplified from the embryonic mouse heart cDNA library and cloned into pCR TOPO TA as described above using primers LRP1-5 5′-CAC TCC CTG ACA GAG TTG GTG-3′ and LRP1-3 5′-GGT AGA GTT GTC CCT TTC AGG-3′. SerdinR was amplified from the E13.5 brain cDNA library using primers LRP2-5 5′-CCT GGA TCC AGA GCT GGA AGG AAG-3′ and LRP2-3 5′-TCA GGT GCC CAA TCC TGG AGG-3′. The PCR reactions were performed in a PCR Engine (MJ Research) using the following conditions: 94°C for 2 min followed by 30 cycles of 94°C for 30 sec, 56°C for 30 sec, and 72°C for 2 min, followed by an extension at 72°C for 10 min.

Whole-Mount In Situ Hybridization

pCR TOPO-serdin plasmid was used to generate digoxigenin-labeled RNA probes for whole-mount in situ hybridization. Digoxigenin nucleotide mix, T7 polymerase/buffer, and RNAse inhibitor were from Roche, DNAse I was from Promega. Embryos were isolated from timed-pregnant Swiss Webster females and embryonic sample preparation and in situ hybridizations were performed as previously described (Tevosian et al., 1999). Whole-mount embryos were dehydrated, embedded in paraffin, and sectioned at 10 μm. The embryos and sections were photographed by using Olympus microscopes and a Magnafire digital camera. Images were assembled and labeled using Adobe Photoshop and Corel DRAW software.

RT-PCR Analysis

Embryonic P19 carcinoma cell (ATCC, CRL-11825) were differentiated into cardiomyocytes by induction with DMSO as previously described (Paquin et al., 2002), and RNA was isolated with Trizol on days 0, 5, 7, and 9 of DMSO treatment. The following primers were used as follows: Serdin1, LRP1-5 5′-GGT AGA GTT GTC CCT TTC AGG-3′ and LRP1-3 5′-CAC TCC CTG ACA GAG TTG GTG-3′; SerdinR, LRP2-RT1 5′-CTT GCA GAA GCT GTA TGT GAG C, LRP2-RT2-3′ and 5′-GTC GAG GTC TAG GAT GTG CAG-3′; Cardiac troponin I, CTIleft 5′-CTC CTC TGC CAA CTA CCG AG-3′, CTIright 5′-CGG CAT AAG TCC TGA AGC TC-3′; Gata4, GATA4Lt 5′-CTG TCA TCT CAC TAT GGG CA-3′ and GATA4Rt 5′-CCA AGT CCG AGC AGG AAT TT-3′; and Carp, CARPL1 5′-TGC CAG TTG TAG AGA AAT TGG TGT-3′, CARPR1 5′-ATC TCT TGT AGG CAT TCT CCT TGG-3′. RT-PCR analysis was performed essentially as described above using 24–29 cycles of amplification. All reactions included a no-RT control.

Expression in Mammalian Cells

To generate the Serdin1 mammalian expression vector, pCR-TOPO-serdin was re-amplified using SRD–EcoRI 5′-GAA TTC ATG GGA AAC ACC ATC CGG-3′ and SRD–BamHI 5′-CCT GGA TCCAGA GCT GGA AGG AAG-3′ primers and re-cloned in pCRTOPO-TA. The EcoRI –BamHI fragment was excised from the resulting vector and cloned into pFLAG-CMV-5a (Sigma). The pFLAG-CMV-serdin plasmid was introduced into HEK 293 cells by CaCl2 transfection (Tevosian et al., 1999) or into mouse NIH 3T3 cells by electroporation. For electroporation, ∼ 6 × 106 cells were trypsinized, washed with phosphate buffered saline (PBS), and re-suspended in 800 μl of PBS and electroporated at 250 V and 975 μF in a Gene Pulser (Bio-Rad) and plated at a density of 5 × 105 cells/well using six-well plates (Fisher Scientific).

Anti-Serdin Antibody Generation

To generate the rabbit anti-serdin polyclonal antibody, we transferred EcoRI fragment from pTOPO-TA into pGEX-3X (Pharmacia). The resulting plasmid was transformed into DH5αcells, and protein expression was induced in 2 L of LB broth with IPTG (125 μg/l). The fusion protein was isolated by batch precipitation using GST-Sepharose (Pharmacia) and eluted with glutathione (Sigma). The purity of the protein preparation was confirmed by SDS-PAGE gel analysis followed by staining with Coomassie blue. The rabbit anti-SERDIN1 polyclonal antibody was generated by Animal Pharm Services, Inc. (San Francisco, CA) and affinity-purified using the Amminolink Immobilization protocol (Pierce); its specificity was confirmed by Western blot analysis and an antigen blocking assay (see below).

Western Blot Analysis

The animals used for all the experiments were of 129/C57Bl mixed background. Total protein extract was prepared from mouse tissues (heart and liver), from freshly isolated embryonic cardiomyocytes, from 293 cells that were transiently transfected with pFLAG-CM-5a Serdin and from untransfected 293 cells by SDS lysis followed by boiling. The samples were centrifuged and loaded on a 12% SDS-PAGE gel along with BenchMark Protein Ladder (Invitrogen) followed by a semidry transfer (Owl) onto a PVDF (Millipore) membrane. The incubation with the preimmune or anti-Serdin antibody was followed by signal detection using horseradish peroxidase–conjugated anti-rabbit secondary antibody and the ECL system (Amersham). For the antigen blocking experiments, 5 μg of the anti-SERDIN1 antibody was incubated with increasing amounts (5–200 μg) of recombinant SERDIN1 protein in 10 ml of PBS–3% Carnation milk for 10 min; this mixture was applied to strips of membrane containing identical amounts of the cardiac extract. To confirm the specificity of this experiment, the strips were re-probed with anti-troponin I antibody (US Biological) plus recombinant SERDIN1 under the same conditions.

Northern Blot Analysis

Northern blot analysis was performed as previously described (Tevosian et al., 1999). PCR-TOPO Serdin1 and pCR-TOPO SerdinR were digested with EcoRI to release the corresponding gene fragments. The DNA was labeled with Random Primer kit (Roche) and purified using Sephadex G-25 columns (Roche) to generate the radioactive probes. A Mouse Multiple Tissue Northern Blot (BD Biosciences) was probed sequentially with both Serdin probes. The gel was re-probed with a radioactive Gapdh (Ambion) probe to ensure the presence of RNA in every lane. The blots were washed and exposed to Biomax film (Kodak) with intensifying screen for 3 hr (Gapdh), overnight (Serdin1) or 48 hr (SerdinR).

Immunohistochemistry

Immunofluorescent analysis of HL-1 cells (a kind gift of Dr. William Claycomb: Claycomb et al., 1998), P19 cells, and embryonic cardiomyocytes was performed essentially as previously described (Tevosian et al., 1999), except that 0.1% saponin (Malinckrodt) was used instead of Triton X-100. Briefly, cells grown on chamber slides (Lab-Tek) were washed with PBS, fixed with 4% paraformaldehyde or cold methanol, washed twice in PBS, and blocked for 30 min in PBS, 5% bovine serum albumin (BSA), and 0.1% saponin. Next, the cells were incubated in primary antibodies diluted in PBS, 1%BSA, and 0.1% saponin. After 1-hr incubation, the cells were washed twice with PBS and incubated with secondary antibodies diluted in the PBS/1%BSA/0.1% saponin for 1 hr. Cells were washed and mounted in medium with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI; Vector Laboratories). The following antibody combinations were used: affinity purified rabbit anti-SERDIN1 (100 μg/ml; 1:100 dilution) followed by goat anti-rabbit Alexa Fluor 488-conjugated antibodies (Alexa), goat anti-Gata4 antibodies (SantaCruz, 1:150 dilution) followed by donkey anti-goat Alexa Fluor 555-conjugated antibodies (Alexa), and mouse anti-cardiac Troponin T (cTnT) antibodies (US Biological, 1:300 dilution) followed by goat anti-mouse Alexa Fluor 555-conjugated antibodies (Alexa). All secondary (conjugated) antibodies were diluted 1:500. Mouse cardiomyocytes were isolated from E12.5 mouse embryos according to a previous protocol (Kelly et al., 2003). Briefly, hearts (both atria and ventricles) were isolated from three to six E12.5 embryos in ice-cold Ca2+-free PBS (Invitrogen). After a 20-min incubation, PBS was replaced by ice-cold 0.25% Trypsin–ethylenediaminetetraacetic acid (Gibco) and incubated for 30 min at 37°C with gentle shaking. The cells were spun, re-suspended in DMEM with 10% fetal bovine serum, and plated onto Lab-Tek chamber slides (Lab-Tek). The medium was replaced after 18 hr. The immunofluorescent analysis of cardiomyocytes was performed after 48 hr or 12 days as described above. Images of cells were photographed by using Olympus BX-51 microscope and a Magnafire digital camera. Images were assembled and labeled as described above.

The immunofluorescence analysis of mouse embryonic tissues and adult hearts was performed essentially as described above with 10-μm frozen sections of E12.5 embryos and adult mouse hearts. The anti–PECAM-1 antibody (BD Pharmingen) was used at a 1:150 dilution and detected by goat anti-rat Alexa Fluor 488-conjugated antibody (Alexa).

For immunofluorescence analysis of adult rat heart, 10-μm frozen sections were fixed in 3% paraformaldehyde/PBS for 15 min, washed in PBS, permeabilized in 0.2% Triton X-100/PBS for 15 min, and blocked in 2% BSA/1% donkey serum/PBS for 30 min. Sections were then incubated with rabbit anti-serdin antibodies (1:100 dilution) followed by Cy2-conjugated goat anti-rabbit IgG (1:600). Sections were costained using anti-titin N2A antibodies (10 μg/ml: generously provided by Dr. Siegfried Labeit: Centner et al., 2000) and Texas Red-conjugated donkey anti-chicken IgG (1:100). Identical results were obtained when the order of the primary antibodies was switched. “Secondary-alone” controls, as well as staining for SERDIN on sections of rat skeletal muscle demonstrated negligible fluorescence. Secondary antibodies were purchased from Jackson ImmunoResearch Laboratories. Sections were analyzed on an Olympus inverted immunofluorescence microscope (model IX70). Photomicrographs were recorded as digital images (with Z-series containing 0.15-μm sections) using a CCD camera (model CH350, Photometrics) and deconvolved using DeltaVision software (Applied Precision).

Generation of Transgenic Mice

The BAC DNA containing Serdin1 gene was obtained from Research Genetics and amplified using standard techniques. A 6.7-kb NheI-NheI fragment containing the 5′ region of Serdin1 was isolated from BAC and cloned into pLitmus-38. The transgenic constructs were generated in a pBSK KSII vector (Stratagene) that was modified to incorporate SfiI sites on both flanks of the MCS sequence to yield pBSK-SfiI. Briefly, a 618-bp fragment of Serdin1 containing the ATG codon was generated by PCR using serdin BAC as a template and serdin-specific primers SRD BN 5′-CGG GAT CCC CAT GGG AAA CGG GCT GCA AGG-3′ and SRD-618 5′-CAC AAG GAG CAG TTG AAC GTA-3′. The PCR fragment was digested with NcoI–BamHI and cloned into pLimus-38. The β-galactosidase gene followed by SV40 polyadenylation site was cloned from pSDKlacZpA (provided by Dr. J. Rossant) into NcoI–BamHI of the resulting construct. The SalI–BamHI fusion fragment was re-cloned into pBSKII SfiI. Finally the NheI–XhoI fragment containing the 6.7 kb of serdin was incorporated in NheI–XhoI sites. Transgenic serdin-lacZ fusion fragment was isolated by SfiI digest, gel purified, and eluted by QIAquick gel extraction (Qiagen). Transgenic animals were generated at Dartmouth College Transgenic facility by pronuclear injection into B6129PF1/j fertilized eggs. Foster mothers were killed, and F0 embryos were collected and analyzed at E12.5 (F0) by an X-gal staining assay essentially as previously described (Tevosian et al., 2000). Stable transgenic lines were obtained using the same construct as above, with F1 animals obtained by crossing to C57/129 mouse strain. Organs from the neonatal F1 animals were dissected and also analyzed by X-gal staining.

Acknowledgements

This work was partially supported by the HHMI Grant to Dartmouth Medical School under the Biomedical Research Program for Medical Schools. S.G.T. and C.C.G. were funded by the NIH.

Ancillary